Specific lipid mediator signatures of human phagocytes: microparticles stimulate macrophage efferocytosis and pro-resolving mediators

Inflammation is the organisms' response to local injury in vascularized tissues programmed to traffic leukocytes and plasma delivery to an injured site or point of bacterial invasion,¹  this protective response when uncontrolled in humans is associated with many widely occurring diseases. These include cardiovascular, metabolic, and the classic inflammatory diseases (ie, arthritis and periodontal disease) along with cancers.²  Nonresolving inflammation is now widely acknowledged as a major driver in most of these diseases.³  Resolution of inflammation and dissipation of the local chemical messengers involved in mounting the innate response were thought to be passively diluted with time at the site, hence stopping further leukocyte recruitment and resolving the exudate or battlefield of inflammation.^4,5 

Results from this laboratory indicate that resolution of self-limited inflammatory exudates is a biochemically active process that involves the local and temporal biosynthesis of a new genus of specialized pro-resolving mediators (SPMs) with their novel functions mapped employing resolution indices.^5-8  SPMs encompass several families of structurally and chemically distinct mediators. These chemical mediator families include lipoxins biosynthesized from arachidonic acid, E-series resolvins (Rv) from eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA)-derived D-series resolvins, protectins, and maresins. Each potent bioactive member of these families shares a defining action in resolving local inflammation. By definition, they each limit further neutrophil (PMN) recruitment to the site of injury and/or microbial invasion and enhance macrophage uptake of cellular debris and apoptotic PMN to bring about tissue homeostasis.²  Along with these defining properties, specific SPMs carry out more specialized tasks within programmed resolution; hence, the scope of their individual actions are nonoverlapping and evoked via specific cell surface receptors that are G-protein–coupled receptors.²  A systems approach led to the identification of novel bioactive structures coined resolvins and protectins in murine inflammatory exudates and isolated human cells based on liquid chromatography tandem mass spectrometry (LC-MS-MS)–based lipid mediator lipidomics and tandem assessment of their functions in anti-inflammation and pro-resolution.^6,9  The complete stereochemistry and total organic synthesis of several key resolvins, protectins, as well as their aspirin-triggered forms are established.¹⁰  These include resolvin D1 (7S, 8R, 17S-trihydroxy-4Z, 9E, 11E, 13Z, 15E, 19Z-docosahexaenoic acid), resolvin D2 (7S, 16R, 17S-trihydroxy-4Z, 8E, 10Z, 12E, 14E, 19Z-DHA), 17R-HDHA (17R-hydroxy-4Z,7Z,10Z,13Z,15E,19Z-DHA), neuroprotectin D1/NPD1 (10R, 17S-dihydroxy-4Z, 7Z, 11E, 13E, 15Z, 19Z-DHA), resolvin E1 (5S, 12R, 18R-trihydroxy-6Z, 8E, 10E, 14Z, 16E-eicosapentaenoic acid), and most recently maresin 1 (7R, 14S-dihydroxy-4Z,8E,10E,12Z,16Z,19Z-DHA).¹¹  In addition to confirming the original structural assignments and potent anti-inflammatory and pro-resolving actions in vivo of resolvins, lipoxins, and maresins,¹⁰  new studies from others demonstrate their potent actions in experimental colitis,¹²  arthritis,¹³  arthritic pain,¹⁴  ocular diseases,¹⁵  resolving obesity,¹⁶  and diabetes.¹⁷  Importantly, synthetic SPMs permitted their identification in other biologic sources, including, for example, human inflammatory responses,^18,19  human serum,²⁰  fish,²¹  and other marine organisms²²  and the invertebrate phyla.¹¹ 

The initial identification of resolvins and protectins in resolving exudates implicated that multiple cell types are involved in their biosynthesis given the dynamic process ongoing within evolving inflammatory exudates.^5,6  With authenticated SPMs and pro-inflammatory lipid mediator (LM), we can now carry out targeted LM metabololipidomics via profiling monitoring > 50 individual mediators and pathway markers from these autacoid pathways in initiation and resolution of inflammation.²³  Microparticles (MPs) in human synovium can influence the course of inflammation.²⁴  Because evolving self-limited inflammatory exudates produce functional MPs that also signal to stimulate resolution of inflammation in mice,²⁵  in this report we systematically profiled LM and SPM produced by individual human cell types and MP involved in initiation and gauge their contributions to resolution of inflammatory responses. Employing a targeted LM metabololipidomic approach, we found that apoptotic PMNs possess a pro-resolving LM profile and their uptake stimulated SPM biosynthesis in macrophages, a process that is also regulated by PMN MPs and transcellular biosynthesis.

Materials include the following: lipopolysaccharide (LPS)–serotype 055B:B5, Ficoll-Histopaque 1077–1 (Sigma-Aldrich); hr-GM-CSF, hr-IFN-γ, hr-IL-4, hr-M-CSF (R&D Systems), phenol red free RPMI 1640 (Invitrogen); pooled human Serum (Lonza); human monocyte isolation kit (StemCell Technologies); mouse anti–human CD80-FITC, anti-CD163–PE, CD206-FITC, CD68-FITC, CD80-PE, CD54-APC, HLA-DR–PE/Cy5.5 (BioLegend); carboxyfluorescein diacetate succinimidyl ester (CFDA; Invitrogen); LC grade solvents (Fisher Scientific); Luna C18 columns (Phenomenex); C18 SPE columns (Waters); fluorescently conjugated zymosan A (Invitrogen); arachidonic acid (AA), EPA, and DHA were purchased from Cayman Chemical along with synthetic standards for LC-MS/MS quantification and deuterated internal standards d₈-5S-HETE, d₄-leukotriene B₄ (LTB₄), d_5-lipoxin A₄ (LXA₄) and d₄-prostaglandin E₂ (PGE₂), d₈-AA, d₅-EPA, and d₅-DHA. Fresh human leukocytes were prepared from deidentified peripheral blood of healthy volunteers obtained by venipuncture in accordance with Partners Human Research Committee Protocol number 1999P001297. All participants gave written informed consent. This study was conducted in accordance with the Declaration of Helsinki.

Human PBMCs from deidentified healthy human volunteers from the Children's Hospital Boston blood bank were isolated by density-gradient Ficoll-Histopaque isolation¹¹  and monocytes purified using monocyte isolation kit (StemCell Technologies) yielding a 93%-97% CD14⁺ population. These were then cultured 7 days in RPMI 1640 (10% human serum). Macrophages were differentiated using 20 ng/mL GM-CSF for 7 days. Macrophage preparations were then incubated with fluorescently labeled anti-CD3 and anti-CD19 antibodies (lymphocyte markers). These macrophage incubations were > 95% CD3-CD19 negative (see supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). M1 macrophages were obtained by incubating isolated monocytes 7 days with 20 ng/mL GM-CSF, and stimulating for 24 hours with 20 ng/mL IFN-γ and 1 ng/mL LPS. M2 macrophages were obtained by incubating monocytes with 20 ng/mL M-CSF 7 days and stimulating for 48 hours with 20 ng/mL IL-4.²⁶  Phenotypic linage of the human monocyte-derived macrophages was assessed by flow cytometry using fluorescently conjugated antibodies in accordance with published criteria.²⁶ 

Macrophage subtypes were obtained as described in “Macrophage cultures.” Human PMNs were isolated by density-gradient Ficoll-Histopaque isolation; these were then cultured overnight at 5 × 10⁶ cells/mL in Dulbecco PBS (DPBS) to obtain apoptotic neutrophils (> 90% AnxAV⁺; supplemental Figure 2). In some cases, the cells were then counted and stained using CFDA.¹¹ 

Labeled apoptotic PMNs were added (3:1 ratio, PMN to macrophages) to differentiated macrophages in the presence of vehicle or MPs. Incubations (60 minutes, 37°C) were stopped after 1 hour. Subsequently, the cells were washed with PBS to remove nonphagocytosed PMN and extracellular fluorescence quenched using trypan blue (1:50 dilution). The number of phagocytosed cells (ie, intracellular) was determined using a SpectraMax M3 plate reader (Molecular Devices) monitoring fluorescence emission at 525 nm. To quantify the number of phagocytosed PMN, calibration curves were obtained using fluorescently labeled apoptotic PMNs that gave linear curves with an r² value > 0.9. In separate experiments, apoptotic PMNs were incubated with macrophages with or without MPs for 0 minutes or 60 minutes (37°C).

Human neutrophil MPs were prepared from peripheral blood PMN.²⁵  Briefly, neutrophils were prepared following density separation by layering on Ficoll-Histopaque 1077-1. Cells were centrifuged (300g, 30 minutes, 4°C) and red blood cells lysed by hypotonic lyses. PMN purity was assessed by flow cytometry after cell staining with fluorescently conjugated antibodies to CD3, CD14, CD16, CD11b with > 99% of the cells negative for CD14 and CD3. This suggests little to no monocytic or lymphocytic contamination (see supplemental Figure 2). Neutrophils were suspended in DPBS (2 × 10⁷ cells/mL), stimulated with formyl-methionyl-leucyl-phenylalanine (1μM; 20 minutes, 37°C). Cells were removed by 2 successive centrifugations (3000g; 10 minutes, 4°C), before pelleting MPs by ultracentrifugation (100 000g; 1 hour, 4°C). PMN MPs were characterized according to their forward and side scatter parameters and found to be > 85% AnxA V⁺ and > 95% CD66b⁺ (Figure 1), in accordance with published criteria.²⁵  MPs then were suspended in DPBS^+/+ and stained using fluorescently labeled anti-CD66b (specific marker for neutrophil MPs). Subsequently, MP numbers in 100 μL buffer were assessed as a function of CD66b positive events using flow cytometry.

To assess the impact of MPs on macrophage LM biosynthesis, macrophages (2.5 × 10⁵ cells/2 mL) were incubated with or without 5 × 10⁵ MPs (15 minutes, 37°C, pH 7.45). Incubations were stopped by adding ice-cold MeOH followed by extraction and LM profiles identified by LM metabololipidomics (see “Sample extraction and lipid mediator metabololipidomics”). In designated incubations, macrophages were incubated with pertussis toxin (1 μg/mL, 37°C, overnight) or cholera toxin (1 μg/mL, 37°C, 2 hours) before the addition of MPs to the incubations (15 minutes, 37°C, pH 7.45).

All samples for LC-MS/MS analysis were extracted using SPE columns as described previously.²³  Briefly, columns were equilibrated with 1 column volume of methanol and 2 volumes ddH₂O. Before extraction, 500 pg of deuterium-labeled internal standards d₈-5S-HETE, d₄LTB₄, d₅LXA₄, and d₄PGE₂ were added to facilitate quantification of sample recovery.

Sample supernatants were diluted with 10 volumes of ddH₂O, acidified (pH ∼ 3.5) and immediately loaded onto the SPE column. After loading, columns were washed with 1 volume of neutral ddH₂O and hexane. Samples were eluted with 6 mL methyl formate and taken to dryness using Speedvac or nitrogen stream. Samples were suspended in methanol/water for LC-MS/MS. The LC-UV-MS/MS system, QTrap 5500 (ABSciex) was equipped with an Agilent HP1100 binary pump and diode-array detector. An Agilent Eclipse Plus C18 column (50 mm × 4.6 mm × 1.8 μm or 100 mm × 4.6 mm × 1.8 μm) was used with a gradient of methanol/water/acetic acid of 60:40:0.01 (volume/volume/volume) to 100:0:0.01 at 0.5-mL/min flow rate. To monitor and quantify the levels of the various LMs, a multiple reaction monitoring (MRM) method was developed with signature ion fragments for each molecule. Identification was conducted using published criteria²³  with at least 6 diagnostic ions. Calibration curves were obtained using synthetic and authentic LM mixtures (they included d₈-5S-HETE, d₄-LTB₄, d₅-LXA₄, d₄-PGE₂, RvD1, RvD2, RvD5, PD1, MaR1, RvE1, RvE2, LXA₄, LXB₄, LXA₅, PGE₂, PGD₂, PGF_2α, thromboxane B₂ (TXB₂), PGE₃, PGF_3α, TXB₃, LTB₄, 17-HDHA, 14-HDHA, 7-HDHA, 4-HDHA, 18-HEPE, 15-HEPE, 12-HEPE, 5-HEPE, 15-HETE, 12-HETE, and 5-HETE) at 12.5, 25, 50, 100 pg. Linear calibration curves for each were obtained with r² values in the range 0.98-0.99. Quantification was carried out based on peak area of the MRM transition and the linear calibration curve for each compound. In case synthetic or biogenic standard for a given product was not available (ie, LXB₅, PGD₃), calibration curves for products with similar chromatographic behaviors (ie, tri-, di-, or mono-HETEs) to the analyte of interest were used.

Human PMNs isolated as described in “PMN MPs” were either placed directly in methanol, suspended in DPBS (5 × 10⁶ cell/mL), and stimulated with 0.1 mg zymosan or incubated overnight at 37°C, and apoptosis was assessed using propidium iodide and annexin (AnxA) V staining (see supplemental Figure 2). Incubations were stopped with 2 volumes of cold methanol. In designated experiments, PMNs were plated in 12-well (2 mL/well) with or without AA, EPA, and DHA (each at 100 ng) and incubated for the indicated time intervals. The incubations were stopped with 2 volumes of cold methanol and held at −80°C before solid phase extraction.

PMNs were incubated with 100 ng each of d₈-AA, d₅-EPA, or d₅-DHA for 5 minutes before addition of formyl-methionyl-leucyl-phenylalanine (1μM, 30 minutes, 37°C). The cell suspensions were centrifuged and MPs were obtained as described in “PMN MPs.” Next, apoptotic PMNs were incubated with d₈-AA, d₅-EPA, or d₅-DHA (100 ng each) for 30 minutes (37°C). To assess the contribution of MPs to transcellular biosynthesis of LM, deuterium-labeled MPs were coincubated with unlabeled PMNs and macrophages (1 hour, 37°C). Incubations were then extracted, products identified and quantitated using LC-MS/MS. Similarly, contribution(s) of apoptotic PMNs were determined using deuterium-labeled precursors in apoptotic PMNs incubated with unlabeled macrophages (2.5 × 10⁵ cells/2 mL) and MPs (5 × 10⁵ MP/incubation; 30 minutes, 37°C, pH 7.45).

Results are mean ± SEM. Multiple-group comparisons were made using either paired Student t test or 1-way ANOVA followed by Dunnett or Bonferroni posthoc analysis. The P values < .05 were considered significant.

Because PMN MPs exert anti-inflammatory and pro-resolving actions in vivo,²⁵  we assessed their ability to regulate macrophage uptake of apoptotic cells (efferocytosis).²⁷  PMN MPs were obtained from activated PMNs (Figure 1A) and incubated with macrophages before addition of fluorescent apoptotic PMNs (Figure 1), dose-dependently increased macrophage-associated fluorescence (Figure 1B). These results indicate that MPs regulate efferocytosis of apoptotic cells.

Next, we assessed endogenous pro- and anti-inflammatory LM biosynthesis by human macrophages (supplemental Figure 1B) during uptake of apoptotic PMNs using targeted LC-MS/MS metabololipidomics, emphasizing functional lipidomics (see supplemental Table 1 and references within). We identified mediators from both lipoxygenase and cyclooxygenase pathways, including RvD1, MaR1, PGE₂, and PGF_2α, (Figure 2A). All were identified in accordance with published criteria,²³  which included matching retention times, fragmentation patterns, and at least 6 characteristic and diagnostic ions for each as illustrated with results obtained for RvD1 (eg, 375 = M-H; 357 = M-H-H₂O; 339 = M-H-2H₂O; 331 = M-H-CO₂; 313 = M-H-H₂O-CO₂; 295 = M-H-2H₂O-CO₂; 287 = 303-H₂O; 241 = 277-H₂O), RvD2, LXB₄, and 5,15-diHETE (Figure 2B and insets for diagnostic ions).

LM quantification was achieved by MRM of signature ion pairs Q1 (parent ion) to Q3 (characteristic daughter ion; Figure 2C). Macrophage efferocytosis of apoptotic PMNs increased SPM biosynthesis, primarily RvD1 (18 ± 10 vs 127 ± 14 pg/2.5 × 10⁵ cells), RvD2 (28 ± 8 vs 124 ± 12 pg/2.5 × 10⁵ cells), and LXB₄ (19 ± 2 vs 44 ± 6 pg/2.5 × 10⁵ cells). This was accompanied by increases in prostanoids, including PGE₂ (7 ± 1 vs 44 ± 12 pg/2.5 × 10⁵ cells; Figure 2C)_. Having found that MPs stimulate efferocytosis, we next assessed their actions on SPM biosynthesis during apoptotic PMN uptake. MPs stimulated macrophage biosynthesis of RvD2 (219 ± 44 vs 124 ± 12 pg/2.5 × 10⁵ cells), LXB₄ (84 ± 15 vs 44 ± 6 pg/2.5 × 10⁵ cells), and RvE2 (478 ± 57 vs 199 ± 26 pg/2.5 × 10⁵ cells), to a greater extent than apoptotic PMN, while reducing PGF_2α (259 ± 29 vs 459 ± 54 pg/2.5 × 10⁵ cells) and TXB₂ (328 ± 21 vs 456 ± 62 pg/2.5 × 10⁵ cells; Figure 2C)_.

Temporal SPM biosynthesis contributes to regulating the onset of resolution programs.²  Therefore, we next assessed the cumulative levels for each LM family identified. During human macrophage efferocytosis, there was an increase in DHA-derived (Figure 2D) and AA-derived SPM (Figure 2E) biosynthesis, along with an augmentation in prostanoid production (Figure 2F). MPs further enhanced apoptotic PMN-stimulated SPM biosynthesis (Figure 2D-E) while reducing prostanoid production (Figure 2F). To rule out the possibility that MPs carried mediators in addition to their known ability to carry SPM precursors,²⁵  we profiled the MPs (supplemental Figure 3). MPs did not carry mature SPMs. Hence, we profiled the amounts of LM associated with apoptotic PMNs and MPs in the absence of efferocytosis (supplemental Figure 4). Although specific SPMs and LMs were demonstrable, their values were baseline compared with levels produced via macrophages during efferocytosis.

We next investigated whether PMN MPs were able to stimulate macrophage LM biosynthesis. Incubation of PMN MPs with macrophages led to the biosynthesis of lipoxygenase- and cyclooxygenase-derived LM. These included RvD5, PD1, and PGE₂ in accordance with criteria²³  that included matching retention times (supplemental Figure 5A) and at least 6 diagnostic ions as illustrated for MaR1, PD1, RvE2, and RvE1 (supplemental Figure 5B). These results are first to demonstrate each of these pro-resolving mediators in human macrophage incubations and their interrelationships.

For example, we observed that PMN MPs led to an increase in MaR1 (13 ± 4 vs 21 ± 2 pg/2.5 × 10⁵ macrophages), PD1 (7 ± 1 vs 28 ± 5 pg/2.5 × 10⁵ macrophages), LXB₄ (78 ± 4 vs 248 ± 26 pg/2.5 × 10⁵ macrophages), and RvE1 (14 ± 8 vs 72 ± 9 pg/2.5 × 10⁵ macrophages) production (supplemental Figure 5C). In addition, we found that PMN MPs stimulated macrophage prostanoids, in particular PGE₂ (18 ± 2 vs 40 ± 6 pg/2.5 × 10⁵ macrophages) and TXB₂ (28 ± 7 vs 63 ± 10 pg/2.5 × 10⁵ macrophages). Cumulative LM metabololipidomics for each LM family identified demonstrated that PMN MPs stimulated biosynthesis of DHA- (supplemental Figure 5D), AA- (supplemental Figure E), and EPA- (supplemental Figure 5F) derived SPM along with prostanoids (supplemental Figure 5G). Of interest, incubation of macrophages with G-protein inhibitors (pertussis toxin and cholera toxin) reduced SPM biosynthesis without altering prostanoid levels (supplemental Figure 5C-G). Together, these results demonstrate that MPs selectively stimulate macrophage SPM production in a G-protein–coupled receptor-dependent manner.

Because apoptotic PMNs stimulate macrophage SPMs during efferocytosis, we investigated mediator profiles of apoptotic PMNs compared with zymosan-stimulated PMNs. Using targeted LM metabololipidomics, we identified a profile signature of LM that included RvD1, LXB₄, PGD₂, and PGE₂ (Figure 3A). Identification was achieved as illustrated for RvE2, PD1 (also known as NPD1/PD1 when produced in neural systems⁷ ), RvD2, and LTB₄ (Figure 3B).

MRM LM quantification of signature ion pairs demonstrated that AA-derived LMs represented ∼ 85% of the apoptotic PMN AA, EPA, and DHA identified LM metabolome. In these cells, the DHA-derived mediators represented ∼ 6%, whereas EPA-derived mediators amounted to ∼ 11% of the targeted LM. On the other hand, in zymosan-stimulated PMNs, the AA-derived LM constituted ∼ 81% of the targeted AA, EPA, and DHA LM metabolomes (Figure 3A). Color scales represent the magnitude of LM within chromatographic clusters within each profile (see Figure 3 legend).

The comparison of endogenous LM identified in apoptotic versus zymosan-activated PMNs revealed that RvD1 (42 ± 12 vs 5 ± 2 pg/5 × 10⁶ cells) and LXB₄ (142 ± 37 vs 14 ± 7 pg/5 × 10⁶ cells) biosynthesis was reduced in zymosan-stimulated cells. Similarly, there was a down-regulation in a number of cyclooxygenase-derived LMs in zymosan-stimulated PMNs, including PGD₂ (123 ± 29 vs 82 ± 15 pg/5 × 10⁶ cells), PGE₂ (390 ± 75 vs 82 ± 16 pg/5 × 10⁶ cells), and PGF_2α (127 ± 40 vs 52 ± 18 pg/5 × 10⁶ cells). Production of the chemoattractant LTB₄ (including its 20-OH P450 metabolite) was elevated in zymosan-stimulated cells as opposed to apoptotic PMNs (Figure 3C-D). For intact PMNs without exposure to agonists, the levels for these mediators were close to or below limits (Figure 3C).

To gauge the effector functions endowed on PMNs by their individual LM profiles, we assessed the cumulative LM amounts for the distinct LM families identified using LM metabololipidomics. We found that apoptotic PMNs produced significantly higher SPMs (Figure 3E-G) and prostanoids (Figure 3H) than zymosan or intact PMNs. Assessment of AA, EPA, and DHA-derived monohydroxy products did not reveal significant differences between apoptotic, zymosan-stimulated, or intact PMNs (data not shown).

To appreciate the sequence of events involved in LM production during human PMN apoptosis, we next identified LM present at distinct stages of PMN progression to apoptosis determined by AnxAV staining (supplemental Figure 6A). Because, during inflammation, edema formation supplies the inflammatory response with extracellular substrate,²⁸  we assessed the time course of LM biosynthesis by apoptotic cells in the presence of extracellular substrate. MRM quantification of identified LM demonstrated a significant increase in SPM biosynthesis by apoptotic cells, including D- and E-series resolvins that reached maximum within 30 minutes (supplemental Figure 6B-D). Prostanoids were also elevated (supplemental Figure 6E), whereas LTB₄ (including its 20-OH metabolite) occurred during the initial phase and subsequently declined, both in the presence and absence of substrate (supplemental Figure 6F). Although we cannot rule out that a small proportion of potentially nonapoptotic PMNs that might have been present could have contributed to LM profiles in incubations with apoptotic PMNs, it is more than likely that at the 18-hour interval LM profiles reflect the capacity of apoptotic PMNs to biosynthesize specific LMs.

Next, we investigated the potential contribution of transcellular biosynthesis, in particular for SPM, during macrophage efferocytosis (Figure 4). For this purpose, we used deuterium-labeled (d) precursors, which allowed for the identification of mediators from substrate/precursors derived from either apoptotic PMNs or MPs. LM derived from d-substrate had a higher mass to charge ratio (m/z) than those biosynthesized from endogenous substrate. Specifically, LM produced from d₈-AA were 8 atomic mass units (amu) and those from d₅-EPA and d₅-DHA were 5 amu greater than LMs biosynthesized from endogenous precursors.

We first fortified MPs and apoptotic PMN with d-labeled precursors and assessed levels of d-labeled LM, essential fatty acids (AA, EPA, and DHA), and biosynthetic precursors (including 18-HEPE, 15-HEPE, 15-HETE, 17-HDHA, and 14-HDHA). We identified LM precursors, including d₅-DHA, d₅-17-HDHA, and d₈-AA (supplemental Figure 7A), along with a select number of mediators, including d₅-RvD5 and d₅-RvE2 in accordance with published criteria.²³  MRM quantitation of the identified LM demonstrated that the levels of these d-labeled mediators within MPs, and apoptotic PMNs were near limits of detection (< 5 pg/6 × 10⁵ MPs and < 1 pg/1 × 10⁶ PMNs). We also identified d-labeled biosynthetic precursors that included d₅-17-HDHA (16 ± 3 pg/5 × 10⁵ MPs and 18 ± 1 pg/7.5 × 10⁵ apoptotic PMNs) and d-labeled essential fatty acids that included d₅-DHA (423 ± 35 pg/5 × 10⁵ MPs and 459 ± 133 pg/7.5 × 10⁵ apoptotic PMNs; supplemental Figure 7B). These results suggest that, although there was a significant enrichment of d-labeled precursors in both MPs and apoptotic PMN, only a minor proportion was further converted to bioactive LM in the absence of macrophages.

We next added either d-labeled MPs or d-labeled apoptotic PMN to macrophages assessing LM transcellular biosynthesis during macrophage efferocytosis. Apoptotic PMN uptake by macrophages led to the biosynthesis of d₅-RvD2 and d₅-RvD5 from the D-series resolvins, d₅-PD1 from the protectins, from the lipoxins d₈-LXB₄ and d₅-RvE2 from the E-series resolvins, whereas d₈-PGD₂ and d₈-PGF_2α were the only prostanoids identified (Figure 4A) and d₅-RvD5, d₅-RvE2, and d₈-LXB₄ (Figure 4B) where the atomic mass of the parent molecules was increased by either 5 or 8 amu, respectively.

MRM quantification of d-LM biosynthesized during macrophage efferocytosis suggested that MPs acted as a nexus for d₅-RvD2, d₅-RvD5, (Figure 4C-D) d₈-LXA₄ (Figure 4F), and d₈-PGE₂ and d₈-PGF_2α (Figure 4G) production. d₅-MaR1 and d₈-LXB₄ biosynthesis was primarily sustained by precursors derived from within apoptotic PMN. d₅-PD1 (Figure 4D) and d₅-RvE2 (Figure 4E) biosynthesis were equally reliant on precursors from apoptotic cells and MPs. These results demonstrate that transcellular biosynthesis contributes to LM production during macrophage efferocytosis, and both apoptotic PMN and MPs contribute to and sustain a nidus for LM biosynthesis.

We next sought to assess the LM profiles of distinct human macrophage subtypes. Following LM metabolomics of M1 and M2 macrophages, which were prepared from human peripheral blood monocytes (supplemental Figure 1), we identified 35 distinct LMs that included RvD1, PD1, MaR1, LXA₅, LXB₅, PGD₂, and PGE₂ (Figure 5A). These mediators were identified using their diagnostic fragmentation patterns, matching a minimum of 6 characteristic fragments as shown for RvD2, PD1, RvE2, and LXA₅ (Figure 5B).

LM metabololipidomics of M1 macrophage AA, EPA, and DHA identified LM metabolome we found that AA-derived LM amounted to ∼ 48% of LM identified. On the other hand, LM metabololipidomics of M2 macrophages demonstrated that SPM represented ∼ 50% of the identified AA, EPA, and DHA metabolome, consisting of D-series resolvins, protectins, and maresins (∼ 11%), lipoxins (∼ 15%), and E-series resolvins and lipoxins (∼ 24%; Figure 5A).

Using MRM, we compared the endogenous biosynthesis of individual LM between the 2 macrophage subtypes. In these experiments, we found that biosynthesis of distinct SPM was elevated in M2 compared with M1 macrophage; these included RvD5 (196 ± 26 vs 43 ± 5 pg/2.5 × 10⁵ cells), MaR1 (299 ± 8 vs 45 ± 8 pg/2.5 × 10⁵ cells), PD1 (3442 ± 206 vs 1339 ± 206 pg/2.5 × 10⁵ cells), LXA₄ (977 ± 173 vs 675 ± 167 pg/2.5 × 10⁵ cells), LXB₄ (11 750 ± 724 vs 7560 ± 659 pg/2.5 × 10⁵ cells), LXB₅ (4370 ± 397 vs 2551 ± 859 pg/2.5 × 10⁵ cells), and RvE2 (20 680 ± 4910 vs 10 910 ± 4232 pg/2.5 × 10⁵ cells). Conversely, production of cyclooxygenase derived LM was elevated in M1 macrophages; these included PGE₂ (2285 ± 435 vs 1573 ± 490 pg/2.5 × 10⁵ cells), PGF_2α(11 325 ± 428 vs 2108 ± 256 pg/2.5 × 10⁵ cells), TXB₂ (550 ± 135 vs 222 ± 73 pg/2.5 × 10⁵ cells), and the EPA-derived prostanoids (Figure 5C).

Having found that different human macrophage subtypes possess distinct LM profiles, we next compared the cumulative amounts of different LM families identified in M1 and M2 macrophages to determine effector functions that these endow on the individual macrophage subtypes. SPM biosynthesis was higher in M2 compared with M1 macrophages (Figure 5D-F), whereas production of prostanoids (Figure 5G-H) and leukotrienes LTB₄ and LTB₅ (Figure 5I-J) was elevated in M1 macrophages compared with M2.

Because apoptotic PMNs stimulate LM biosynthesis in macrophages, we investigated whether this finding held for different macrophage subtypes. Assessment of LM biosynthesis after apoptotic cell efferocytosis demonstrated that apoptotic PMNs uptake gave SPMs in M1 macrophages (supplemental Figure 8A-C) while reducing prostanoids (supplemental Figure 8D-E) and leukotrienes (supplemental Figure 8F,G). Apoptotic PMNs with M2 macrophages reduced overall LM production, including SPMs (supplemental Figure 8A,C). These results indicate that human M1 and M2 macrophages, as defined,²⁶  possess distinct endogenous LM signature profiles.

In the present report using targeted LM metabololipidomics, we determined the endogenous LM profiles of human phagocytes with distinct stimuli. Efferocytosis was found to stimulate SPM, a process regulated by PMN MPs and dependent, at least in part, on precursors obtained from both apoptotic cells and MPs. Apoptotic cells also displayed a pro-resolving LM signature profile, indicating that these specific LM profiles are key determinants of their distinct effector functions from those of zymosan-stimulated PMNs. This distinction was also found for macrophage subtypes (known as M1 and M2²⁶ ) involved with initiation and resolution of inflammation.

PMNs are the first leukocyte responders to an inflammatory insult. On recruitment from vasculature, they orchestrate clearance of the ensuing insult and stimulate either propagation or termination of inflammatory responses via temporal release of soluble mediators.²⁹  SPMs are a genus of lipid mediators that include lipoxins, resolvins, protectins, and maresins; by definition, each shares potent anti-inflammatory and pro-resolving actions demonstrable within the picogram to nanogram range. SPMs were initially identified in resolution of self-limited inflammation where they are formed in vivo at levels commensurate with their activity. Specific SPM members (eg, RvD1) also carry additional actions, such as pain regulation and control of local chemokines.^12-15  In human skin blisters (a PMN-driven response), biosynthesis of 15-epi-LXA₄ and LXA₄ and up-regulation of ALX/FPR2 (LXA₄ receptor) are key in mediating the anti-inflammatory and pro-resolving actions of low-dose aspirin.¹⁹  Therefore, in the present studies, we systematically investigated LM profiles of human apoptotic PMNs and activated freshly isolated PMNs to determine their contributions, as well as PMN MPs.

A number of arachidonic acid-derived lipid mediators produced via both lipoxygenase and cyclooxygenase pathways, such as the classic prostaglandins and leukotrienes, are implicated in the onset and perpetuation of inflammation.³⁰  Leukotriene B₄, a potent pro-inflammatory LM, was the most abundant LM from both ionophore- and zymosan-activated PMN (Figure 6A), reaching levels within its bioactive range (supplemental Table 1, which contains complete stereochemical assignments and names for each specific SPM and LM, and references within), underscoring the pro-inflammatory phenotype elicited by these stimuli with intact human PMN from peripheral blood. Human PMN stimulation also up-regulates expression of cyclooxygenase-2³¹  and, consequently, PGE₂ and PGD₂, which were identified in both. These mediators also possess pro-inflammatory properties that include enhancing leukocyte recruitment and vascular permeability (supplemental Table 1). In addition, both prostaglandins stimulate lipid mediator class switching and up-regulation of SPM at sites of inflammation.²  In the present studies, apoptotic PMN displayed elevated PG biosynthesis, suggesting that during acute inflammation these cells can initiate LM class switching and the onset of resolution. SPM biosynthesis, in particular that of RvE2 and LXB₄, was also a distinct signature of apoptotic cells. This emphasizes the pro-resolving phenotype of apoptotic PMN because both RvE2 and LXB₄ exert potent actions dampening leukocyte recruitment and reducing vascular leakage (supplemental Table 1). LM metabololipidomics with intact PMN showed that lipoxins, as verified by the presence of LXB₄, occurred in the absence of further stimulation, suggesting that SPM biosynthesis can occur in peripheral blood PMN. These findings are in accord with those from patients with abdominal aortic aneurism, where elevated plasma SPM levels are associated with better postoperative outcomes.³² 

The acute inflammatory response in humans is a dynamic process where cell trafficking to ideally contain microbial challenge resolves to homeostasis, a protective stance.^1,2  In models of self-resolving sterile inflammation, LTB₄ levels peak in the initiation phase, reaching ∼ 3 ng/mouse exudate.^7,33  Prostaglandins, including PGE₂, peak later in these systems coinciding with the onset of resolution, the PGE₂ levels reaching ∼ 6.5 ng pg/mouse exudate.⁷  SPM biosynthesis, including RvE2, peaks during the resolution phase with levels for this mediator reaching ∼ 2.5 pg/mouse exudate.³⁴  Similarly, in murine models of live infections, PGE₂ levels peak with the onset of resolution, reaching ∼ 1773 pg/mouse exudate.³⁵  Hence, the present findings with isolated human PMN staged in vitro (ie, on activation, progression to apoptosis, and during macrophage efferocytosis) follow the in vivo events and levels of LM in self-contained exudates. For example, the primary mediator produced by zymosan-activated human PMN in vitro was LTB₄ ∼ 31.5 pg/1 × 10⁶ PMN (Figure 3), levels that are comparable with those reported in vivo.^7,33  With apoptotic human PMN, we found elevated biosynthesis of prostaglandins and SPMs that included PGE₂ ∼ 62 pg/1 × 10⁶ PMN and RvE2 ∼ 33.2 pg/1 × 10⁶ PMN (Figure 3). These LM levels are consistent with those reported in vivo in mice during the resolution phase of acute inflammation within contained inflammatory exudates.^7,33,35  The present results are the first to demonstrate this dynamic shift with human leukocytes in vitro and their interrelationships in the LM profile that resemble distinct exudate stages observed for self-limited inflammatory responses in vivo. This is of general relevance to inflammatory diseases in humans because it is sometimes difficult to obtain temporal changes in LM profiles during disease states and their resolution.³⁶  Nonetheless, LXA₄ levels increase during resolution of glomerulonephritis,³⁷  and PD1 levels are diminished in exhaled breath condensates from asthmatics.³⁸ 

Systematic mapping of cellular trafficking in a murine model of self-limited acute inflammation demonstrated that mononuclear cell accumulation at the site of inflammation occurs as part of the events that lead to the onset of resolution.^1,2,7  During this phase, macrophages accelerate resolution by clearance of apoptotic cells and cellular debris in a nonphlogistic manner in a process referred to as efferocytosis.²⁷  Herein we determined the endogenous LM profiles produced by human macrophages during efferocytosis. These experiments demonstrated an increase in SPM, specifically LXB₄ and RvD1. RvD1 possesses potent actions: at concentrations as low as 10nM, it reduces pro-inflammatory cytokines from peritoneal macrophages and stimulates up-regulation of M2-associated markers.¹⁶  Cellular activation leads to release of MPs that elicit pro-inflammatory³⁹  as well as protective actions in vivo.²⁵  Specific subtypes of PMN MPs, carrying SPM precursors, accumulate in self-limited exudates preceding onset of resolution.²⁵  Herein we found that the addition of PMN-derived MPs to human macrophages increased SPM biosynthesis in a G-protein–coupled receptor-dependent manner. This enhances macrophage efferocytosis of apoptotic PMNs that in turn led to a further increase in macrophage LM production, in particular SPMs. These include RvD2, LXB₄, and RvE2, which each reached concentrations > 100nM. RvD2, at doses as low as 100nM, reduces pro-inflammatory cytokine production in murine bone marrow macrophages stimulated by LPS and reverses TNF-α–induced leukocyte-endothelial interactions at 1nM. In addition, LXB₄ and RvE2 are potent tissue protective SPMs (supplemental Table 1). These findings underscore the importance of efferocytosis in resolution and macrophage SPM biosynthesis during efferocytosis, a process regulated by PMN MPs that can contribute to resolution in humans.

LM biosynthesis involves multiple enzyme-regulated steps that dictate the specific stereochemistry of these local mediators. In some instances, one cell type does not possess all the enzymes necessary for LM biosynthesis.⁴⁰  Therefore, in a process coined as transcellular biosynthesis, intermediates are donated to a second cell type that is equipped with the necessary enzymes for their conversion to bioactive mediators.⁴⁰  SPM production, including RvD5, LXB₄, and RvE2, during efferocytosis was found to rely, at least in part, on substrate/intermediates (using deuterium label) that can be supplied by both apoptotic PMN and MPs, thus demonstrating that this biosynthetic route is also responsible for SPM production, a key step in stimulating resolution.

Distinct macrophage subtypes exert diverse effector functions in propagation and resolution of the inflammatory response.⁴¹  In obesity, for example, amelioration in insulin resistance and adipose tissue inflammation is associated with a macrophage phenotypic switch from M1- to M2-like phenotypes.¹⁶  Using targeted LM metabololipidomics, we assessed LM profiles in M1 and M2 demonstrating herein that M1 macrophages, characterized by their higher expression of CD68, CD80, MHC class II and CD54,²⁶  eicosanoid biosynthesis, including TXB₂ and LTB_4, were elevated. RvD2 biosynthesis, which activates potent antibacterial actions (supplemental Table 1), was also increased in line with the concept that M1-macrophages contribute to protection against bacterial infection.⁴²  Hence, lower levels of individual SPM were obtained for M2 compared with M1 macrophages, perhaps reflecting the higher predisposition of certain mediators (eg, RvE1) to undergo further metabolism. This together with higher expression of inactivating enzymes, such as eicosanoid oxidoreductase/15-prostaglandin dehydrogenase that can further convert LM, may give the apparent lower levels of specific LM by M2 macrophages. It is noteworthy that the SPMs are potent mediators that are bioactive in the low nanomolar to picomolar range. Moreover, recent results demonstrate that D-series resolvins, specifically RvD1 and RvD2, are substrate for further metabolism by dehydrogenation, with RvD1 being more rapidly further metabolized than RvD2.⁴³  SPM biosynthesis, in particular PD1, was higher in M2 macrophages consistent with the finding that IL-4 induces 15-LOX expression in human monocytes.⁴⁴  Similarly MaR1, recently shown to exert potent pro-resolving and tissue regenerative actions,¹¹  production was also elevated in line with the tissue regenerative properties associated with this macrophage subtype.⁴² 

Transcriptomic analysis of murine macrophages during resolution of acute inflammation suggest that at the onset of resolution there is a switch in macrophage phenotype from M1-like to displaying characteristics of both M1 and M2.⁴⁵  Resolving macrophages possess 15-LOX type 1, which initiates LX and D-series resolvin biosynthesis,^6,45  also expressed in the eye,⁴⁶  and initiated PD1 by cardiac stem cells.⁴⁷  Because we found that apoptotic PMN stimulate SPM in human macrophages, we assessed the impact of apoptotic cell uptake by lineage-defined macrophages on their LM profile. Uptake of apoptotic cells by M1 cells led to a decrease in prostanoids and leukotrienes, including LTB₄ and LTB₅ (supplemental Table 1) with a concomitant enhanced SPM biosynthesis to give a pro-resolving LM profile or signature.

It is now well established that cardiovascular diseases are at least in part driven by inflammation.⁴¹  ω-3 fatty acids are now also appreciated for their ability to exert a positive bearing on several intermediate determinants of cardiovascular risk.⁴⁸  One mechanism involves RvE1, demonstrated to be protective in murine cardiovascular disease.⁴⁹  In addition, in mice, eosinophils contribute to resolution, producing both LXA₄ and PD1.⁵⁰  Our present findings indicate that human phagocytes, apoptotic PMN, and pro-resolving MPs produce these LM, which can contribute to resolution of inflammation and thus may contribute the beneficial actions noted for ω-3 fatty acids⁴⁸  via the local biosynthesis of potent mediators.

In conclusion, we systematically assessed the main human phagocyte components of an acute inflammatory response (ie, PMN, MPs, and macrophages), using LM metabololipidomics and defined surface markers. The present results translate to defined human cell types, in vivo findings in resolving exudates in murine systems.^5,6  In addition, we demonstrate that apoptotic PMN possess a pro-resolving LM signature profile that activates macrophage SPM during efferocytosis. Both apoptotic PMN and MPs are a nexus for SPM biosynthesis by macrophages when performing efferocytosis. A potential caveat to the present findings is the lack of cell cycle synchronicity during PMN apoptosis in vitro. Nonetheless, this is a widely used procedure to obtain apoptotic PMN in the literature, and our findings demonstrate the dynamic biosynthetic range of these apoptotic PMN that produce both pro-inflammatory and pro-resolving mediators. We also establish that distinct macrophage subtypes produce specific LM signature profiles, where M1 cells display a pro-inflammatory, whereas M2 cells display a pro-resolving LM profile, which are modulated on uptake of apoptotic PMN. Taken together, these findings demonstrate that distinct PMN and macrophage phenotypic subpopulations possess signature LM profiles that endow them with specific effector functions modulated during the dynamic process of inflammation and its timely resolution.

The authors thank Mary Halm Small for expert assistance in manuscript preparation, Dr Nan Chiang (Center for Experimental Therapeutics and Reperfusion Injury, Brigham and Women's Hospital/Harvard Medical School) for helpful discussions, and Dr Sungwhan Oh (Center for Experimental Therapeutics and Reperfusion Injury, Brigham and Women's Hospital/Harvard Medical School) for initial setup of the LC-MS/MS system I.

This work was supported in part by the National Institutes of Health (grants R01GM038765 and P01GM095467).

National Institutes of Health

Contribution: J.D. designed and carried out experiments, analyzed data, and contributed to manuscript and figure preparations; and C.N.S. carried out overall experimental design, conceived the research plan, and contributed to manuscript and figure preparations.

Conflict-of-interest disclosure: C.N.S. is an inventor on patents (resolvins) assigned to BWH and licensed to Resolvyx Pharmaceuticals. C.N.S. is a scientific founder of Resolvyx Pharmaceuticals and owns equity in the company. The interests of C.N.S. were reviewed and are managed by the Brigham and Women's Hospital and Partners HealthCare in accordance with their conflict-of-interest policies. J.D. declares no competing financial interests.

Correspondence: Charles N. Serhan, Harvard Institutes of Medicine, 77 Avenue Louis Pasteur, HIM 829, Boston, MA 02115; e-mail: cnserhan@zeus.bwh.harvard.edu.