Author Correction: Identification of specialized pro-resolving mediator clusters from healthy adults after intravenous low-dose endotoxin and omega-3 supplementation: a methodological validation

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

studies suggest that intake of docosahexaenoic acid (DHA), eicosapentaenoic acid (EPA), and other ω-3 polyunsaturated fatty acids (PUFA) may provide tissue protection 4 . The ω-3 PUFA are precursors for specialized pro-resolving mediators (SPM) that include resolvins, protectins and maresins that are produced in the resolution phase of acute inflammation. By definition, each SPM pathway stimulates resolution of inflammation and infections by limiting the exposure to pathogens and collateral damage from tissue-destructive neutrophils. SPMs enhance innate host defense responses that include macrophage phagocytosis of apoptotic neutrophils and microbes 3 . These proresolving mechanisms include limiting neutrophil infiltration and stimulation of macrophage mediated uptake of apoptotic neutrophils, cellular debris and microbes. Each action is stimulated at pico to nanomolar ranges of SPM, requiring stereospecific biosynthesis (reviewed in ref. 3 ). Identification and profiling of SPMs has recently been operationalized with liquid chromatography-mass spectrometry (LC-MS) based approaches by multiple laboratories [5][6][7] . This has enabled elucidation of specific functional SPM clusters in several human tissues and fluids, including blood 5-7 , placenta 8 , and emotional tears 9 .

Results
LM-SPM were profiled from serum in Study A and plasma in Study B using LC-MS/MS (Fig. 2 Supplementary Fig. 1). The duration and dose of ω-3 supplementation implemented in the two studies was selected based on results from earlier studies where supplementation reduced triglycerides and increased incorporation of both EPA and DHA into erythrocytes (i.e., ω-3 index 11,12 ). Additional criteria for the individual studies were as follows: In Study A, supplementation was designed to reflect a range of dietarily achievable doses of EPA and DHA <2 grams/day, and the duration of supplementation was longer (~5 months) to permit incorporation of EPA and DHA into erythrocyte membranes 11 . In Study B, we implemented a crossover design with higher dosing of 3.4 g/d EPA and DHA and shorter duration (8-12 weeks) of supplementation ( Fig. 1 illustration). Both studies resulted in increases in erythrocyte content of membrane EPA and DHA 11,12 (and unpublished data below for Study B) and Table 1. Recently we found that coagulation activates production of a cluster of SPMs, including RvD1, resolvin D5 (7S, 17S-dihydroxy-docosa-4Z, 8E, 10Z, 13Z, 15E, 19Z-hexaenoic acid; RvD5), RvE1, maresin 1 (7R, 14S-dihydroxy-docosa-4Z, 8E, 10E, 12Z, 16Z, 19Z-hexaenoic; MaR1), and lipoxin B 4 (5S, 14R, 15S-trihydroxy-eicosa-6E, 8Z, 10E, 12E-tetraenoic acid; LXB 4 ) 13 . We therefore assessed the impact of omega-3 supplementation followed by endotoxin challenge on LM-SPM profiles obtained from serum (generated by coagulation ex vivo) as well as changes in plasma that reflect in vivo biosynthesis.
Serum LM-SPM in Study A. Using metabololipidomics focusing on cyclooxygenase and lipoxygenase pathways and products, we identified LM-SPM from each of the bioactive mediator metabolomes derived from DHA, EPA, and arachidonic acid (AA) . In serum LM-SPMs from Study A: Both laboratories identified a cluster of SPMs in human serum that included RvE1, RvD1, AT-LXA 4 , 17-HDHA, and 18-HEPE (Figs 3 and 4; Supplementary Tables 2 and 3). Each of these mediators has potent pro-resolving and anti-inflammatory actions 3,14-16 and were identified using MS-MS diagnostic ions as well as comparison with synthetic standards. All LM-SPMs and biosynthetic pathway markers were identified in accordance with published criteria 5,9 that included matching retention time and at least six characteristic and diagnostic fragment ions. Principal component analysis indicated that serum SPMs associated with ω-3 supplementation as well as prostaglandins and thromboxane, which associated with placebo throughout the endotoxin challenge time course (Fig. 5A,B). Total amounts of the SPM cluster consisting of RvD1, RvE1, AT-LXA 4 , 17-HDHA, and 18-HEPE increased 229% in human subjects given ω-3 vs. subjects with placebo ( Fig. 5C) at 120 hours post endotoxin challenge. To establish independent alignment in these results, a second aliquot of these coded samples was analyzed with metabololipidomics (operated and optimized as in Fig. 2) via a second laboratory. This second analysis identified the same SPM cluster, which was increased 169% in subjects receiving ω-3 supplementation vs. subjects with placebos.
Time course analysis indicated that serum prostaglandin and thromboxane concentrations in serum were highest at early time points, 0-8 hours post endotoxin challenge (Fig. 6A,B and Supplementary Tables 2 and 3). In this subset of study participants (n = 3/group), TXB 2 peaked at 2 hours with placebo and was statistically significantly reduced with ω-3 supplementation (Fig. 6A). C-reactive protein (CRP) concomitantly peaked at 24 hours and was significantly reduced in the ω-3 group vs. placebo group (  Tables 2 and 3). This SPM cluster in the ω-3 group initially decreased at early time points (2 h) and increased at 24 hours post endotoxin challenge (Fig. 7A). These results demonstrate that ω-3 PUFA supplementation increases a cluster of specific  pro-resolving mediators, namely RvD1, RvE1, AT-LXA 4 , 17-HDHA, and 18-HEPE in human serum. Thus, ω-3 PUFA supplementation increased the potential for blood cells to produce SPMs after endotoxin challenge, assuming that leukocytes and platelets are the main source of SPM in serum.
In the present study, distinct LM-SPM were quantifiable in a volume of 1 mL of human plasma or serum with ω-3 supplementation and endotoxin challenge (Figs 3 and 4, Supplementary Tables 2-5), even in the presence of substantial matrix suppression associated with these biological samples ( Supplementary Fig. 1) and the limits of detection with the workup and instruments used. Specifically, these include RvD1, RvE1, AT-LXA 4 , LXB 4 , 17-HDHA, and 18-HEPE. Both 17-HDHA and 18-HEPE, which are bioactive SPM 14,15,17 as well as precursors to resolvins, were consistently identified in quantifiable concentrations by both laboratories in the present study, and are thus strong candidates for plasma and serum clinical biomarkers of SPM and resolution of inflammation. RvD1 was identified (Supplementary Tables 4 and 5) in human plasma following endotoxin challenge, while other D-series resolvins (RvD2 -RvD6) that are present in human blisters 18 , lymph 5 and breast milk 19,20 were not quantifiable in these blood specimens from healthy individuals (Supplementary Tables 2 and 3). Thus, to optimize for future studies, ideally >1 mL plasma and serum from healthy subjects would improve identification of LM-SPM. Since SPMs including those identified in the serum and plasma clusters in the present study are increased via coagulation 13 , a dose response of lipid mediator production between the two (i.e. plasma vs. serum) could not be evaluated using the two study protocols and will require further investigations.
Interlaboratory results acquired in the present study were in agreement in the identification of clusters of SPMs in human serum and plasma that increased following ω-3 supplementation. Both laboratories employed essentially identical instrumentation and data acquisition parameters, albeit non-statistically significant differences in quantification between the two labs ( Fig. 8B) appear to reflect differences in sample extraction and data processing methods. Hence, it is imperative that standardization of LC-MS/MS metabololipidomic platforms across separate laboratories is validated for further improvement of quantitation of LM and SPM. Variations in identification and quantification of specific lipids by many separate laboratories in the field 21 likely reflect lab-dependent sample workup including lipid mediator solid-phase extraction procedures, liquid chromatography solvent systems and lack of deuterium-labeled internal standards for quantitation, as well as instrument specifications with respect to mass spectrometry collision energy and ionization parameters.

Discussion
In the present study, we provide evidence for the in vivo production of SPMs in humans with ω-3 supplementation and endotoxin challenge. These results are consistent with the many animal and human studies that have utilized LC-MS/MS analysis for SPM identification and quantification. Along these lines, in animal models of disease and inflammation, resolvins and other members of the SPM superfamily are identified and present at concentrations that are biologically active (reviewed in refs 3,9,13 ). For example, increasing ω-3 PUFA via transgenic overexpression of fat-1 in mice enhances the formation of RvE1, resolvin D3 (4S, 11R, 17S-trihydroxy-docosa-5Z, 7E, 9E, 13Z, 15E, 19Z hexaenoic acid; RvD3), and protectin D1 (10R, 17S-dihydroxy-docosa-4Z, 7Z, 11E, 13E, 15Z, 19Z-hexaenoic acid, PD1; also known as neuroprotectin D1 [NPD1]), reduces inflammation, and enhances protection against tissue injury in colitis 22 . During intestinal ischemia/reperfusion in mice, production of LXA 4 and 18-HEPE is increased and blockade of LXA 4 signaling to its receptor (ALX) mitigates its anti-inflammatory and pro-resolving actions 23,24 . In sterile murine peritonitis, RvD1, RvD2, RvD5, PD1, MaR1, and LXA 4 are produced during the resolution phase 25 , whereas, in infectious murine peritonitis with pathogenic E. coli, RvD1, RvD5, and PD1 are produced and lower antibiotic requirements for bacterial clearance 26 . In non-human primates, baboons infected with S. pneumoniae display diminished plasma levels of lipoxins and E-series resolvins, which are increased with therapeutic low doses of carbon monoxide 27 . In chronic inflammatory disease models, RvD1, RvD2, RvD3, and RvD4 are present in self-resolving murine arthritis, while RvD3 is reduced in delayed-resolving arthritis and reduces paw joint clinical scores, leukocytes, eicosanoids, and edema 28 . Additionally, AT-RvD1 and its precursor, 17R-HDHA, reduce inflammatory pain in an adjuvant-induced arthritis model 29,30 . Recently, in a murine model of Alzheimer's disease, AT-LXA 4 and RvE1 were endogenously produced via sphingosine kinase 1 (SphK1)-dependent acetylation of cyclooxygenase (COX)-2 and reduced disease pathology via enhancement of microglial phagocytosis 31 . Thus, the identification of 17R and 15S epimers of resolvins and lipoxins (triggered by aspirin or statins) in humans and animals can also be attributed to endogenous mechanisms of epimer biosynthesis in addition to those triggered by drugs (i.e. 17R-, 18R-, and 15R-SPM epimers) 3 .
SPMs exert potent pro-resolving and anti-inflammatory actions (at pM-nM range) on human leukocytes 3 . RvD1, RvD2, and LXA 4 , each at 1 nM concentration, stimulate shape change and stop neutrophil chemotaxis toward IL-8 gradients 41 . By definition, resolvins, protectins, maresins, and lipoxins are pro-resolving mediators because they reduce pro-inflammatory stimuli and stimulate human macrophage efferocytosis of apoptotic neutrophils and microbial clearance 13,[45][46][47][48] . In the present experiments, we have obtained evidence for increased SPMs in humans following ω-3 PUFA supplementation, which supports the theory that they are produced in human tissue at concentrations that are biologically active. These concentrations (at or above 100 pM) are produced in human peripheral blood and affect the functions of both neutrophils and monocytes (at the single cell level as determined by CyTOF mass cytometry), as well as increasing phagocytosis and killing of pathogenic E. coli 13 . Biomarker concentrations of ω-3 PUFA are associated with reduced incidence of fatal coronary heart disease 4,49,50 , and it has recently been established that ω-3 PUFA supplementation at doses up to 10 g/day (EPA and DHA) does not increase the risk of bleeding or affect other clinically meaningful coagulation parameters 51 . It is also noteworthy that both 17-HDHA and 18-HEPE are biosynthetic intermediates in human leukocyte SPM production 3 , and each is also reported to carry potent bioactions of their own. Namely, 18-HEPE displays vascular actions 15 and 17-HDHA has demonstrated potent reduction of arthritic pain 17 and reduces viral H1N1 infections by enhancing antibody mediated immune responses 14 . Local organ production of SPM, e.g. in human breast milk 28 , tears 9 , and muscle tissue 33 , may be the source of some blood plasma SPMs. On the other hand, local inactivation of SPM may contribute to the absence or diminished levels of select SPM in plasma. For example, at sites of inflammation, as in inflammatory exudates, leukocytes can convert SPM to further metabolites that carry diminished bioactivity via dehydrogenation and omega-oxidation, e.g. 15-oxo-LXA 4 , 13,14-dihydro-LXA 4 , 13,14-dihydro-15-oxo-LXA 4 52 , 22-hydroxy-PD1 53 , 17-oxo-RvD1 54 , 18-oxo-RvE1 55 , 14-oxo-MaR1 and 22-OH-MaR1 56 . For each of the SPMs that undergo rapid, local enzymatic conversion and inactivation, specific mimetic analogs have been introduced as potential therapeutic agonists of resolution 3,57 . Thus, the potential physiologic significance of circulating SPM remains of interest.
In the present report with LPS challenge of the subject, we identified specific clusters of both LM (PG and LT) and SPMs in human plasma (RvE1, RvD1, LXB 4 , 17-HDHA, and 18-HEPE) and serum (RvE1, RvD1, AT-LXA 4 , 17-HDHA, and 18-HEPE) that were specifically increased with ω-3 PUFA supplementation. Differences in composition between serum and plasma SPM clusters indicated the presence of AT-LXA 4 in serum with LXB 4 in plasma. The presence of RvE1, RvD1, and LXB 4 in plasma may be the result of LPS activation of platelets and leukocytes since these mediators can also be produced via platelet-leukocyte transcellular biosynthesis during coagulation 13 . For example, transcellular biosynthesis with human platelets and leukocytes produces lipoxins 58 . Serum formation during coagulation in the present study was preceded by endotoxin challenge, which might alter the SPM cluster profile, so these may reflect increases in further metabolism of LXB 4 (e.g., either 20-OH-LXB 4 or dehydrogenation to 5-oxo-LXB 4 ) to products not targeted in the present approach.
In the case of AT-LXA 4 , this natural lipoxin epimer (15R-LXA 4 ), in addition to its biosynthesis via aspirin acetylation of COX-2, is also biosynthesized via COX-2 nitrosylation 59,60 and/or via the recently discovered SphK1-dependent COX-2 acetylation which increases biosynthesis of R containing epimers of lipoxins and resolvins 31 . Thus, the appearance of AT-LXA 4 in serum as potential other AT-epimers of SPM could reflect aspirin ingestion or any of these endogenous mechanisms (note: subjects abstained from taking aspirin per study protocol). These results require further investigation to assess changes in specificity of LM production and metabolism following LPS challenge and coagulation.
The origins of plasma resolvins, lipoxins, 17-HDHA, and 18-HEPE remain to be identified. It is possible that they originate in organs that produce SPM such as bone marrow, adipose tissue, or sites of local coagulation in vivo that then appear in plasma. These local sites of LM biosynthesis and the potential physiologic functions of plasma SPM warrant further study. Nonetheless, the present results provide evidence for potential counter-regulation of inflammation in human circulation and host defense 13 via endogenous production Although the number of healthy human subjects in the current study was small, these results were cross validated between two separate laboratories using blinded, coded samples and indicate that it is possible to align LC-MS/MS identification and profiling of LM-SPMs between different laboratory settings for larger studies. By optimization of instrument parameters in separate laboratories, sample preparation and rigorous workup procedures, this LM-SPM profiling approach is useful for directly assessing the impact of drugs, nutrition and disease phenotypes in these mediator pathways and potent bioactive products such as the resolvins and other SPMs. Our results provide cross validation and identification of SPM as well as evidence for temporal production of specific pro-resolving mediators with ω-3 PUFA supplementation. These results support an immunoresolvent role of SPM in inflammation resolution in humans challenged with endotoxin that now warrants further investigations with other natural host responses.

Materials and Methods
Human ω-3 PUFA supplementation and low-dose endotoxin challenge. Both studies A and B were conducted and annually approved by the Pennsylvania State University Institutional Review Board, and all participants provided written informed consent. All experiments were performed in accordance with relevant guidelines and regulations. Study A was approved by the FDA and registered at clinicaltrials.gov with the number NCT01078909. Study B was approved by the FDA and registered at clinicaltrials.gov with the number NCT01813110.
In study A, healthy volunteers (n = 6) were randomly assigned to placebo capsules containing 0 mg EPA and DHA (n = 3) or to ω-3 supplementation (n = 3; EPA and DHA esterified in the triglyceride form; Nordic Naturals). Participants in the ω-3 supplementation group received either 900 mg/d with 550 mg as EPA and 350 mg as DHA (n = 1) or 1800 mg/d with 1100 mg as EPA and 700 mg as DHA (n = 2) for 5 months prior to low-dose intravenous endotoxin injection (0.6 ng LPS/kg body weight). Participants were required to fast for 12 hours prior to LPS administration. Blood samples were obtained at 9 different time points (0 hr, 1,2,4,8,24,48,72, and 120 hours post LPS administration; see Fig. 1 illustration).
In Study B, healthy men (n = 3) were supplemented for 8 weeks with 3.4 grams/day EPA and DHA (in the form of four capsules each containing 460 mg of EPA-ethyl ester and 380 mg of DHA-ethyl ester; supplied by Pronova BioPharma) and olive oil for placebo that did not contain either EPA or DHA, in random order, with an 8-week washout period in between supplementation periods. The low-dose endotoxin challenge procedure was the same as that used in Study A, except that the 120-hour blood draw was replaced by a blood sample obtained at 168 hours post LPS administration.
Lipid mediator metabololipidomics. Human plasma or serum samples were analyzed by two independent labs (Boston, MA and Detroit, MI) following similar protocols and shared internal standards. In both labs, the samples (approximately 1 mL) were thawed on ice and supplemented with 500 pg each of deuterated (d)8-5-hydroxy-eicosatetraenoic acid (HETE), d5-RvD 2 , d5-LXA 4 , d4-LTB 4 , d4-prostaglandin E 2 (9-oxo-11α, 15S-dihydroxy-prosta-5Z, 13E-dien-1-oic acid; PGE 2 ) (Cayman Chemical Company) in methanol before further processing. From here on, the two labs followed their own procedures for extraction and LC-MS analysis. In lab 1 (Boston, MA), four volumes of ice-cold LC-MS grade methanol was added to each sample and placed on ice for 45 minutes in the dark to allow for protein precipitation, followed by a centrifugation step (3,000 rpm, 10 min, 4 °C). Supernatants were collected from each sample, and solid phase extraction was carried out according to optimized and reported methods 9 . Methyl formate fractions were then analyzed by liquid chromatography-tandem mass spectrometry system, Qtrap 5500 (AB Sciex) equipped with a Shimadzu LC-20AD HPLC (Tokyo, Japan). The column implemented on this system was a Poroshell 120 EC-18 column (100 mm × 4.6 mm × 2.7 μm; Agilent Technologies, Santa Clara, CA, USA), housed in a column oven regulated at 50 °C, and lipid mediators (LMs) were eluted in a gradient of methanol/water/acetic acid from 55:45:0.01 (v/v/v) to 98:2:0.01 at 0.5 mL/min flow rate. Targeted multiple reaction monitoring (MRM) and EPI were utilized in order to quantify the mediator levels, with MS/MS matching to at least 6 diagnostic and signature ion fragments per molecule. A final analytic quantitation and recovery was performed using the deuterium labeled internal standards, and a LM-SPM profile was produced for each donor. All materials and methods, beginning with sample preparation and finishing with a LM-SPM profile, were completed by two independent labs simultaneously at separate locations.
In lab 2, (Detroit, MI), LC-MS grade methanol was added to the internal standard supplemented samples to a final concentration of 15%. The samples were sonicated in a bath sonicator for 2 min and left on ice for 1 h in dark. The samples were applied to pre-conditioned C18 solid phase extraction cartridges (StrataX C18, 30 mg, Phenomenex, conditioned with 2 ml methanol followed by 2 ml water containing 15% methanol), washed with 2 ml 15% methanol in water followed by 2 ml hexane, and dried under vacuum. The cartridges were eluted directly into HPLC autosampler vials with 1 ml methanol containing 0.1% formic acid. The eluates were evaporated to dryness under a gentle stream of nitrogen while maintaining the external temperature at 25 °C. The dried residue was immediately reconstituted in methanol, vials flushed with nitrogen, capped, and stored at −80 °C until analysis. At the time of LC-MS analysis, the samples were thawed to room temperature, and equal volume of 25 mM aqueous ammonium acetate was added, vortex mixed, and loaded in the autosampler maintained at 15 °C. HPLC is performed on a Prominence XR system (Shimadzu) using Luna C18 (3 µ, 2.1 × 150 mm) column. The mobile phase consists of a gradient between A: methanol-water-acetonitrile (10:85:5 v/v) and B: methanol-water-acetonitrile (90:5:5 v/v), both containing 0.1% ammonium acetate. The gradient program with respect to the composition of B is as follows: 0-1 min, 50%; 1-8 min, 50-80%; 8-15 min, 80-95%; and 15-17 min, 95%. The flow rate is 0.2 ml/min. The HPLC eluate is directly introduced to ESI source of QTRAP5500 mass analyzer (SCIEX) in the negative ion mode with the following conditions: Curtain gas, GS1, and GS2: 35 psi, Temperature: 600 °C, Ion Spray Voltage: −2500 V, Collision gas: low, Declustering Potential: −60 V, and Entrance Potential: −7 V. The eluate is monitored by Multiple Reaction Monitoring (MRM) method to detect unique molecular ion -daughter ion combinations for each of the transitions listed in Supplementary Data, Table 1. The MRM is scheduled to monitor each transition for 120 s around the established retention time for each lipid mediator. Optimized Collisional Energies (18-35 eV) and Collision Cell Exit Potentials (7-10 V) are used for each MRM transition. Mass spectra for each detected lipid mediator were recorded using the Enhanced Product Ion (EPI) feature to verify the identity of the detected peak in addition to MRM transition and retention time match with the standard. The data are collected using Analyst 1.6.2 software and the MRM transition chromatograms are quantitated by MultiQuant software (both from SCIEX). The internal standard signals in each chromatogram are used for normalization for recovery as well as relative quantitation of each analyte. For complete chemical names of lipid mediators in the current study, see ref. 9 .
Statistical analysis. Groups were compared with Student's t-test (two groups) using Prism version 6 (GraphPad, La Jolla, CA USA). The criterion for statistical significance was p < 0.05. Principal component analysis (PCA) was performed using SIMCA 13.0.3 software (MKS Data Analytics Solutions, Umeå, Sweden).

Data Availability
LC-MS/MS data for lipid mediator retention time and fragmentation matching are available at: http://serhanlab. bwh.harvard.edu.