Resolution of inflammation is essential. Although supplementation of ω-3 fatty acids is widely used, their availability at sites of inflammation is not known. To this end, a multidisciplinary approach was taken to determine the relationship of circulating ω-3 to inflammatory exudates and the generation of resolution signals. In this study, we monitored resolvin precursors in evolving exudates, which initially paralleled increases in edema and infiltrating neutrophils. We also prepared novel microfluidic chambers to capture neutrophils from a drop of blood within minutes that permitted single-cell monitoring. In these, docosahexaenoic acid-derived resolvin D1 rapidly stopped neutrophil migration, whereas precursor docosahexaenoic acid did not. In second organ injury via ischemia-reperfusion, resolvin metabolically stable analogues were potent organ protectors reducing neutrophils. Together, these results indicate that circulating ω-3 fatty acids rapidly appear in inflammatory sites that require conversion to resolvins that control excessive neutrophil infiltration, protect organs, and foster resolution.
Mounting evidence indicates that the resolution of acute inflammation is a highly coordinated and biochemically active process that was once thought to be a passive event (for recent review, see Ref. 1). During inflammatory responses to infections, neutrophils first migrate into tissues to participate in host defense, and then, if successful, these tissues return to their homeostatic functions (2). Tissue level resolution programs are actively initiated, and the number of tissue-associated polymorphonuclear neutrophils (PMN)4 is reduced (1). Thus, it is possible that excessive inflammatory responses and their progression to chronic inflammation can result from a local failure to resolve (1). The active processes by which acute inflammation is resolved to tissue homeostasis are of considerable interest because many widely occurring human diseases are associated with chronic inflammation. These include arthritis (3) and periodontal disease (4) as well as diseases that are recognized relatively recently to have aberrant inflammation as a component of the disease, including asthma, cardiovascular disease, and Alzheimer’s disease (5, 6, 7).
The process of resolution is actively controlled in part by formation of newly described endogenous chemical mediators that as local autacoids stimulate proresolving mechanisms (for recent reviews, see Refs. 1 and 8). These proresolving mediators are derived from essential fatty acids that include lipoxins from arachidonic acid (C20:4) and resolvins and protectins from eicosapentaenoic acid (EPA; C20:5) and docosahexaenoic acid (DHA; C22:6) (8) that are biosynthesized in inflammatory exudates during spontaneous resolution. When specific resolvins and/or protectins are administered in vivo, each stimulates multicellular tissue-level responses geared to bring tissues back to homeostasis in a process coined cellular and molecular catabasis (9).
The impact of ω-3 fatty acids (EPA and DHA) themselves has been evaluated in numerous clinical studies (10, 11, 12, 13, 14). These reports and many others emphasize the potential benefits of dietary supplementation with ω-3 fatty acids. Along these lines, i.v. administration of ω-3 fatty acids leads to clinical improvements in patients with rheumatoid arthritis (12). In healthy subjects, ω-3 fatty acids reduce LPS-induced fever as well as inflammatory responses (13). Soy nuts, enriched with ω-3 fatty acids, improve systolic blood pressure and low-density lipoprotein cholesterol levels in hypertensive postmenopausal women (14). In children at risk of type 1 diabetes, ω-3 fatty acids reduce risk (15). DHA is widely appreciated for its neurotrophic and neuroprotection roles that require esterification into phospholipids (16, 17). DHA tissue levels also appear to be critical because in diseases such as cystic fibrosis, the DHA tissue stores appear to be depleted (18). As noted above, ω-3 fatty acids are precursors to potent families of mediators, namely resolvins and protectins that are biosynthesized in spontaneously resolving exudates and are stereoselective in their formation and actions (1). Recently, the original structural elucidation of resolvin D1 (RvD1) (19) was confirmed by total organic synthesis, and its complete stereochemistry was established as 7S,8R,17S-trihydroxy-4Z,9E,11E,13Z,15E,19Z-docosahexaenoic acid (20). Thus, the relationship between circulating EPA and DHA and inflammation resolution is of interest and a focus of this report.
The ω-3 fatty acids have been measured in humans in many studies and show a wide range. In healthy subjects, EPA levels range from ∼0.5 to 2.8% of total fatty acids and DHA ranges from ∼1.3 to 5.0% (21, 22, 23, 24, 25). For the purpose of direct comparisons, see Supplementary Table I (online and references within).5 This wide range of values for humans appears to reflect diet (25). Because ω-3 fatty acids are now recognized as biosynthetic precursors to potent local mediators, their relationship between circulation and exudate formation may be relevant in the resolution of acute inflammation. Given the importance of this question, a multidisciplinary effort was undertaken to address the availability of ω-3 fatty acids to acute inflammatory exudates and to assess the direct actions on cells and tissues compared with their potent bioactive product(s), namely a resolvin.
In this study, using a multipronged and multidisciplinary approach, we report the following: 1) that both labeled EPA and DHA from circulation rapidly appear at sites of evolving inflammatory exudates in vivo; 2) in microfluidic chambers, RvD1 and not equal concentrations of its precursor directly act on single human PMN; and 3) metabolically stable resolvin mimetics have potent organ-protective actions in murine ischemia-reperfusion injury. These findings emphasize the importance of circulating levels of ω-3 fatty acids and their conversion by exudates to resolvins that directly regulate both tissue inflammation and organ injury.
Materials and Methods
All animals used in the present study were male FVB mice (Charles River Laboratories) that were 6–8 wk old (weighing 20–25 g). They were maintained in a temperature and light-controlled environment, and had unlimited access to food (laboratory standard rodent diet 5001 (Lab Diet)), containing 1.5% EPA, 1.9% DHA of total fatty acids, and tap water. Experiments were performed in accordance with the Harvard Medical School Standing Committee on Animals guidelines for animal care (Protocol 02570).
RvD1 and 17-(R/S)-methyl RvD1 were each synthesized by total organic synthesis for these experiments from starting materials of known chirality in enantiomerically and geometrically pure form (N. Petasis, J. Uddin, and C. Serhan, manuscript in submission) by the Organic Synthesis Core, National Institutes of Health P50-DE-016191. Resolvin E1 (RvE1) and the 19-p-fluorophenoxy-RvE1 methyl ester were prepared in an enantiomerically and geometrically pure form, as described previously (26) (N. Petasis, R. Yang, G. Bernasconi, J. Uddin, and C. Serhan, manuscript in submission). Physical and spectroscopic properties matching biogenic and synthetic RvD1 and RvE1were as reported (20, 27). The integrity of each synthetic resolvin and their analogues was assured by monitoring the physical properties with LC-UV-tandem mass spectrometry just before evaluating their biological activities.
Male FVB mice were anesthetized with isoflurane (Hospira). Both 1 μg of 4Z,7Z,10Z,13Z,16Z,19Z-docosahexaenoic-21,21,22,22,22-d5 acid (d5-DHA; Cayman Chemical) and 1 μg of 5Z,8Z,11Z,14Z,17Z-eicosapentaenoic-19,19,20,20,20-d5 acid (d5-EPA; Cayman Chemical) in 100 μl of 2% ethanol (v/v) in saline or vehicle alone were administered by bolus tail vein injection. Approximately 5 min later, peritonitis was initiated with i.p. administration of 1 mg of zymosan A (Sigma Aldrich). For control experiments, 1 ml of saline was injected i.p. At indicated time intervals, mice were euthanized with an overdose of isoflurane, and peritoneal exudates were collected by lavage with 5 ml of Dulbecco’s PBS without either Ca2+ or Mg2+. Exudate cells and supernatants were separated by centrifugation (10 min, 1000 rpm, 4°C), aliquots of supernatants were collected, and 4 vol of cold acetone were added to precipitate proteins. Serum albumin and total protein levels were determined (28).
For tracer experiments, mice fasted overnight (∼18 h) before 500 nCi of [1-14C]DHA (docosahexaenoic acid 4,7,10,13,16,19-[1-14C]; American Radiolabeled Chemicals) were injected via the tail vein. At indicated time intervals, exudates were collected (vide supra) and the radioactivity was determined using a scintillation counter (multipurpose scintillation counter LS6500; Beckman Coulter).
Differential counts and FACS analysis
Aliquots of lavage cells were assessed for total and differential leukocyte counts via light microscopy to identify individual cell types (i.e., neutrophil, monocyte, etc.). For flow cytometry analysis, aliquots of 0.5 × 106 cells were stained with 0.25 μg of FITC-conjugated anti-mouse F4/80, 0.1 μg of PE-conjugated anti-mouse Ly-6G, and 0.1 μg of PerCP-Cy5.5-conjugated anti-mouse CD11b, or 0.1 μg of PE-conjugated rat IgG2c,κ isotype control as a background stain. Cells were washed and analyzed using FACSort (BD Biosciences) (29).
Mass spectrometric analysis of deuterium-labeled EPA and DHA
Upon collection, 2 vol of cold methanol containing internal standard (500 ng/sample, trans-parinaric acid; Molecular Probes) were added to exudates, which were stored at −20°C for 1 h. After centrifugation, samples were diluted and applied to C18 cartridge columns (Extract-Clean EV SPE; Alltech Associates). Lipids were extracted using hexane and methyl formate, and each fraction was collected and its unesterified fatty acids converted to corresponding methyl ester using diazomethane treatment (30). The esters were suspended in hexane, and then injected in gas chromatography-mass spectrometry (GC-MS). Electron-impact GC-MS was conducted using an HP 6890 GC system with HP5973 Mass Selective Detector (Hewlett-Packard) equipped with an HP-5MS capillary column (0.25 mm ID × 30 m, 0.25 μm; Agilent) operating at a mass range from m/z 70 to 800. The ionization voltage was 70 eV, and the ion source temperature was 230°C. Chromatography was conducted using a column temperature maintained at 150°C for 2 min; programmed to increase 10°C/min up to 230°C, 5°C/min up to 280°C; and then maintained at 280°C for 10 min with a helium flow rate of 1.0 ml/min. EPA and DHA were separated, and the area beneath each chromatographic peak of their deuterium labels was obtained by integration using a Chemstation integrator Version D.02.00.275 (Agilent). For an internal standard, parinaric acid was selected because it exists in plants and its spectrum does not overlap with those of endogenous fatty acids for diagnostic ions. We selected the ω-3-derived ion +CH(CH=CH)CH2(CH=CH)CD2CD3 = m/z 113 for both d5-EPA and d5-DHA methyl ester and M+ (290) for methyl parinarate. Identification was conducted by examining and retention times as in Fig. 1 (d5-EPA-methyl ester: 11.7 min, methyl-parinarate; 13.1 min, d5-DHA; 13.9 min). Calibration curves were obtained for each, as follows: d5-EPA, y = 0.0016x + 0.0313 (r2 = 0.9954, 50–500 ng/ml); d5-DHA, y = 0.0015x − 0.0524 (r2 = 0.9809, 50–500 ng/ml).
Ischemia-reperfusion-induced second-organ injury
Mice were anesthetized by i.p. injection of pentobarbital (80 mg kg−1, Nembutal sodium solution NDC 0074-3778-04). Hind limb ischemia was initiated using tourniquets consisting of a rubber band placed on each hind limb (as in Ref. 31). Mice were subjected to hind limb ischemia for 1 h, after which the tourniquets were removed to initiate reperfusion. Resolvins and their analogues were each administered at 1 μg/mouse (i.e., DHA, RvD1, RvD1 methyl ester, 17-(R/S)-methyl-RvD1 methyl ester, RvE1, or 19-p-fluorophenoxy-RvE1 methyl ester) in vehicle (5 μl of ethanol in 120 μl of sterile saline) and compared with vehicle alone. They were administered i.v. to the tail vein ∼5 min before the start of the reperfusion period. For postreperfusion challenge, at 60 and 90 min after tourniquet release, mice were administered 100 ng of 17-(R/S)-methyl-RvD1-methyl ester (prepared as in Ref. 20) or vehicle alone i.v. At the end of this reperfusion period (2 h), the mice were euthanized with an overdose of anesthetic and the lungs were quickly harvested, frozen in liquid nitrogen, and stored at −80°C. The right lungs were homogenized from individual mice and centrifuged, and the tissue levels of myeloperoxidase (MPO) in the resulting supernatants were determined using a mouse MPO ELISA (Hycult Biotechnology; Cell Sciences).
Microfluidic chambers and leukocyte motility
A new microfluidic chamber was designed for testing the actions of putative novel lipid mediators on the chemotactic function(s) of neutrophils. These analyses were performed immediately after neutrophils were separated from components of whole blood within the chamber (vide infra), which routinely averaged less than 5 min, and lipid mediator (test compounds) adsorption to the chamber surfaces was prevented through the use of polyethylene glycol (PEG) coatings. The microfluidic chamber with microstructured valves was fabricated in elastomeric materials using standard microfabrication technologies (32). Briefly, two layers of poly(dimethylsiloxane) (PDMS), molded on two silicon wafers, were bonded together and on top of a glass slide. A network of channels was formed between the glass and the first layer, and control channels for the microstructured valves were formed between the first and second layers of PDMS. For details of the fabrication process, see Supplementary Materials.5
After assembly, different sections of the chamber were isolated through the actuation of the valves, and each section was chemically modified for specific functionalities. The surfaces of the network channels in the gradient generator section were modified with PEG to prevent absorption of the lipid mediators to the PDMS surfaces (see Results; Fig. 3 illustration). Briefly, the PDMS surface of the gradient generation networks was treated with 5% (v/v) 3-(trimethoxysilyl)-propyl acrylate (92%; Aldrich) in acetone and followed by rinsing the chamber with acetone. Next, PEG diacrylate (Aldrich) with 1% photo initiator (2,2-dimethoxy-2-phenyl-acetophenone; Aldrich) was introduced to the chamber. After 15 min, the chamber was washed with 5% H2O in acetone and with pure acetone subsequently, and then dried with air. After this surface modification procedure, the chambers were stored in the refrigerator for up to 1 wk. The surfaces of the chemotaxis chambers, where neutrophils were captured from whole blood, were modified by physical absorption of P-selectin (10 ng/ml for 30 min; R&D Systems), followed by blocking the entire channel network with 2% human serum albumin (Sigma-Aldrich) in HBSS (Sigma-Aldrich) immediately before use. The separation between the different sections of the chamber during the surface modification was achieved by keeping the valves between the main channels and the gradient generators closed.
For neutrophil chemotaxis experiments, the chambers were purged with HBSS containing 0.1% human serum albumin, with care to remove all bubbles. Four syringes, two with buffer, one with IL-8 chemoattractant (R&D Systems), and one with the lipid mediator, i.e., RvD1 or other related compounds tested in HBSS were placed in syringes (1 ml), with a syringe pump set for 0.1 μl/min. The flow was allowed to stabilize for 3 min, with the microscale valves between the chemotaxis chamber and the two gradient generators closed (see Fig. 3). Two steady-state gradients of IL-8 (0–10 nM) were formed in the two gradient generators, one alone and one with an overlaying uniform concentration of the lipid mediator, and were ready to be delivered to the chemotaxis chambers. Approximately 5–10 μl of capillary blood was collected from healthy volunteers by finger prick using a BD genie lancet (BD Biosciences) using Protocol 1999-P-001297 approved by the Partners Human Research Committee. The whole blood was then quickly mixed with heparin (10 μl) in a syringe tip (30G; Small Parts). After opening the cell inlet valve, the suspension was slowly pushed through the tubing (Tygon; Small Parts) into the microfluidic chamber. The blood stayed in the main channel for 3 min, and then the valve was opened for the two chemokine gradient generators. The flow removed the majority of RBC and other cells that were not tethered with the chamber-attached P-selectin, thus allowing direct observation of the neutrophils captured on the surface of the chamber. After 10–15 min, the gradient was switched to the gradient containing the chemokine alone or either native DHA or RvD1. The migration of neutrophils in the chemokine gradient and their individual response(s) to addition of RvD1 or DHA were recorded with a video and/or digital camera mounted on the microscope, and cell migration was analyzed using the cell tracking function in Metamorph (Molecular Devices). At least a dozen cells per condition were tracked and analyzed for displacement in the direction of the gradient and along the flow.
Results from both in vitro and in vivo experiments were analyzed by Student’s t test with p values ≤0.05 taken as statistically significant. The migration of neutrophils in the microfluidic chambers is presented as average and SEM displacement from the original position at the time of the gradient switching.
From circulation to inflammatory exudates
To determine whether circulating ω-3 fatty acids in their unesterified form are available to evolving inflammatory exudates, we administered i.v. deuterium-labeled ω-3 fatty acids, e.g., d5-EPA and d5-DHA, to mice simultaneously experiencing localized inflammatory responses, namely peritonitis. The magnitude of the inflammatory insult was selected to be a self-limited spontaneous resolving peritonitis (33) to monitor the potential presence of deuterium-labeled fatty acids in the inflammatory exudates (Fig. 1, A and B). Both deuterium-labeled EPA and DHA were identified within the local inflammatory exudates within 2 h. As shown in Fig. 1 C (upper panel), both d5-EPA and d5-DHA rapidly appeared in exudates during the time course of initiation of inflammation. The levels of both unesterified d5-EPA and d5-DHA gradually declined at 24 h. In this spontaneous resolution system, maximum inflammation as defined by maximum PMN infiltrates within the exudates was at ∼4 h, and the resolution phase ranged between 12 and 24 h.
As depicted in Fig. 1,C (middle panel), PMN infiltration into the exudates was monitored. By definition, these exudates contain serum proteins (34) that were determined within the peritoneal exudates throughout the time course (Fig. 1,C, lower panel; cf Ref. 35). Of interest, the time course of both d5-EPA and d5-DHA paralleled the initial appearance of edema and increases in protein levels in these inflammatory exudates. Their presence also coincided with PMN infiltration (Fig. 1,C). At 48 h, both d5-EPA and d5-DHA levels were significantly greater than at 24 h. It is noteworthy that, without inflammation (i.e., peritonitis), neither d5-EPA nor d5-DHA appeared in peritoneal lavages (Fig. 1). Thus, EPA and DHA, the precursors for resolvins and protectins, rapidly appeared within these developing inflammatory exudates from peripheral circulation and are delivered to exudates during both the initiation and resolution phases.
Next, we used a second approach to verify that DHA, i.e., a representative ω-3 fatty acid, indeed appears in exudates directly from peripheral circulation (Fig. 2). To this end, radiolabeled ω-3 fatty acid tracer 14C-DHA was administered i.v., and we monitored recovery of 14C from DHA and the total protein amount in murine exudates. To reduce potential and immediate influences of diet, these experiments were performed with mice that fasted overnight before receiving i.v. 14C-DHA. Exudates were collected at 1, 2, 4, and 12 h. At 1 h, 14C label had already reached maximal exudate levels, and protein levels appeared to parallel the 14C radioactivity profile. These results suggest that 14C-DHA and its products rapidly appear coincident with increases in protein levels and increases in PMN infiltration within the exudate site of inflammation. Thus, they confirm that tracer levels of 14C-DHA and its initial transformation products rapidly appeared within the exudates during early initiation of the inflammatory response in vivo (Fig. 2). Of note, these tracer levels of DHA did not reduce leukocyte trafficking as monitored by FACS analysis of the exudates (see Supplement5 Fig. 2).
Single-cell monitoring of real-time responses demonstrates the direct impact of RvD1, but not its precursor DHA
The original isolation and structure elucidation of RvD1 demonstrated its presence and potent anti-inflammatory and proresolution actions in spontaneous resolving exudates and disease models in vivo (19, 20). In this study, we questioned whether RvD1 or its precursor DHA has direct actions with human neutrophils, and specifically whether RvD1 can alter the directed movements of single cells along chemotactic gradients. This is a central point because excessive infiltration of PMN and release of proinflammatory mediators is a well-appreciated key event in causing tissue damage and uncontrolled inflammation. To address this, we used a 1-μl microfluidic chamber that was engineered with microstructure membranes (see Fig. 3). The chamber permitted the isolation of individual PMNs from one drop of whole blood (see Materials and Methods and Fig. 3,C) following a simple finger prick. After isolation, the microstructured valves in the chamber (Fig. 3,B) were used to control the timing of exposure of the captured PMNs to the two distinct chemotactic gradients (Fig. 3,A) (see Materials and Methods). Images of the PMN before and during exposure to the two gradients were continuously monitored using a camera and recorded (Fig. 4 A).
The PMN trafficked along the IL-8 gradient (see Supplement5 Fig. 3) and displayed the typical shape change and morphology of PMN during chemotaxis in a linear gradient (cf Ref. 36 and Fig. 4,A, left panel). Between 0 and 8 min, RvD1 at 10 nM was uniformly infused in the microfluidic chamber. Before exposure to RvD1, individual PMN movements and distances were proportional to time and totaled ∼30–40 μm. Almost immediately on exposure to RvD1, PMN dramatically changed shape (see Fig. 4,A, middle panel) and ceased directed chemotactic movements (Fig. 4, A and B, and see Supplementary Video online).5 Fig. 4,B shows the average displacement, demonstrating a highly reproducible stopping of PMN migration when exposed to RvD1 (n = 12). In sharp contrast, DHA, the biosynthetic precursor to RvD1, at equimolar concentrations did not stop PMN migration (Fig. 5). Thus, by tracing the displacement of single cells in the direction of an IL-8 chemotactic gradient before and after RvD1 was introduced into the chamber, we were able to record the direct actions of RvD1 on PMN. Thus, these results establish the ability of RvD1 to essentially completely stop PMN movements, almost immediately upon exposure, as well as the ability of RvD1 to stimulate rapid shape changes of PMN. These single-cell-recorded PMN responses were not shared by DHA, the RvD1 metabolic precursor in exudates, when introduced at equal molar concentrations in these microfluidic chambers.
Resolvins are organ protective in vivo
Ischemia followed by reperfusion is a well-appreciated pathophysiologic mechanism of organ injury and local tissue damage that can be initiated by excessively activated PMN. This system in mice (Fig. 6) models the second organ injury and tissue damage observed in humans following tourniquet release of vessels in surgery involving, for example, extremities (37, 38). The occlusion of blood vessels creates local ischemia and remote organ injury that has many features of rapid, acute inflammation and tissue damage (31). On release of the tourniquet occlusion, reflow is initiated and aberrantly activated PMN rapidly perfuse and infiltrate secondary organs, causing local damage (31). In this study, we evaluated RvD1 in a remote organ injury system, i.e., hind limb ischemia-reperfusion, to assess whether the ability of RvD1 to stop PMN migration in microchambers can in turn reduce PMN-mediated tissue and organ damage in vivo. Using the time line illustrated in Fig. 6 (upper panel), PMN accumulation in the lung was assessed by monitoring increases in both tissue histology and tissue MPO, a leukocyte marker enzyme. The extent of lung tissue injury is associated with PMN activation and organ infiltration (31).
In this system, we also directly compared the actions of RvD1 with its biosynthetic precursor DHA as well as with another resolvin, namely, RvE1, which is a potent product of EPA that is anti-inflammatory in both oral inflammation (39) and colitis (reviewed in Ref. 1). RvD1, but neither RvE1 nor DHA, was able to protect the lung tissues from excessive leukocyte infiltration (Fig. 6). The histology and reduction in leukocyte infiltration (data not shown) were confirmed by the reduction in the lung-associated MPO values. At 1 μg/mouse, RvD1 sharply reduced ∼50% leukocyte infiltration in the reperfusion tissues. In contrast, at equal doses, neither DHA nor native RvE1 gave significant reduction in PMN tissue infiltration. Because RvD1 undergoes local metabolic inactivation (20), as does RvE1 (26, 40), blocking of their respective metabolic sites of inactivation within each molecule de novo was undertaken using metabolically stable analog mimetics. To this end, both 17-(R/S)-methyl RvD1 and RvD1 were prepared by total organic synthesis for in vivo administration. Both RvD1 and the related analogues (structures shown in Fig. 7) sharply reduced MPO levels in lung tissues. Of interest, the metabolically stable analog of RvE1, namely 19-p-fluorophenoxy RvE1 at 1 μg/mouse, also significantly reduced leukocyte infiltration, whereas native RvE1 was unable to protect lung from PMN-induced damage in this system. It is noteworthy that mouse lung tissue enzymatically converts RvE1 to 18-oxo-RvE1 (26), which is devoid of bioactivity. The 19-p-fluorophenoxy RvE1 prevents this metabolic inactivation (Fig. 7), and as documented in this study, displayed potent protective actions dramatically reducing PMN infiltration and second organ injury of the lung. The actions of resolvins during reperfusion were investigated. Ninety minutes after tourniquet release, 100 ng of 17-(R/S)-methyl-RvD1 methyl ester was administered. The RvD1 analog lowered the increases in MPO levels in lung, whereas at 60 min after ischemia-reperfusion, administration of this RvD1 analog (100 ng i.v.) did not protect from second organ injury. Thus, by preventing the local metabolic inactivation of resolvins, their ability to reduce PMN infiltration into distal organs was sustained for an enhanced beneficial action when given before reflow and at 90 min after reflow compared with controls.
During the course of spontaneous resolution of acute inflammation, ω-3 fatty acids are precursors for the biosynthesis of newly uncovered anti-inflammatory and proresolving lipid mediators (1, 8). These new families of mediators, coined resolvins and protectins, were identified in vivo in resolving exudates and serve in molecular circuits that actively promote endogenous anti-inflammation and resolution of local inflammation (35). However, the mechanism of ω-3 fatty acid mobilization in vivo during inflammation resolution has not been addressed. In this present study, evidence for new mechanisms is presented that indicates that unesterified ω-3 fatty acids, also known as free fatty acids, rapidly appear within the inflammatory exudates moving directly from circulation into the site of inflammation. The movement of EPA and DHA parallels those of both albumin and trafficking leukocytes. Also, single-cell analyses of human PMN using a newly engineered microfluidics chamber provide direct evidence that DHA-derived RvD1 at nanomolar concentrations, and not its precursor DHA at equimolar levels, stops PMN chemotactic responses to spatial gradients of the chemokine IL-8.
Once formed, resolvins are active on target cells in their immediate milieu and are then inactivated by site-specific metabolism (26, 40). To enhance and prolong their actions, analogues of both RvD1 and RvE1 were prepared that delay their local inactivation in tissues, which proved to protect organs in vivo from ischemia-reperfusion injury by reducing neutrophil tissue infiltration. These results demonstrate for the first time that ω-3 levels in circulating blood are rapidly made available to sites of inflammation in vivo for their local use by exudates to generate potent bioactive local mediators, i.e., resolvins in situ. The resolvins in turn act directly on target cells to stimulate endogenous anti-inflammation and protect organs from excessive neutrophil infiltration and subsequent local tissue damage.
After ingestion, EPA and DHA are distributed throughout the human body (41). DHA is predominantly distributed in retina, sperm, cerebral cortex, spleen, and RBC, whereas EPA is found in muscle, liver, spleen, and RBC (42). For example, DHA is esterified in phospholipids of microglial cells in culture and, on activation of these cells, DHA is released from the phospholipids for enzymatic processing (43, 44). DHA is also, from recent results, the precursor for two separate families of mediators that are structurally distinct, namely D series resolvins and protectins. The protectins possess potent biological actions and a conjugated triene structure as distinguishing features (19). EPA is the precursor for E series resolvins that show potent actions in several complex disease models, including inflammatory bowel disease, periodontal diseases, and asthma (reviewed in Ref. 45). However, the availability of unesterified or free ω-3 fatty acids EPA and DHA for processing during inflammation resolution was of interest in the present studies and required a multidisciplinary approach and new tools to address these key points.
The level of total fatty acids in human blood is ∼343 mg/100 ml plasma (46). Based on this value and Supplementary Table I,5 between 48 and 490 mg of EPA and DHA exists in human blood as basal levels. The contribution of de novo ω-3 fatty acids to the total amount in healthy human subjects, however, appears to be quite low. The proportion of α-linolenic acid converted to EPA is most likely on the order of 0.20–8.0%, and the extent of conversion of α-linolenic acid to DHA is 0.05–4.0% (47, 48). Thus, humans require ω-3 fatty acid intake via diet and/or supplementation. Currently, the FDA states that the dietary intake of EPA and DHA should not exceed 3 g/day (49) because excess supplementation on the order of g/day was found to reduce the already low endogenous EPA and DHA biosynthesis (47).
Although DHA and EPA are widely believed to possess anti-inflammatory properties themselves, the specific mechanisms responsible for these actions are still evolving. The ω-3 fatty acids are generally thought to replace the sn-2 position in phospholipid stores that is usually the positional site of esterified ω-6 fatty acids such as arachidonic acid (41). It is well appreciated that, upon activation, cells release arachidonic acid from the sn-2 position of phospholipids via cytosolic phospholipase A2 for conversion to eicosanoids. For example, among the potent bioactive eicosanoids produced by leukocytes, the PGs and leukotrienes are broadly considered proinflammatory mediators (50). The sn-2 position of phospholipids that can become substituted with ω-3 fatty acids (DHA, EPA) is currently thought to simply compete for these enzymatic reactions, thus blocking the use of arachidonate and production of specific eicosanoids that are proinflammatory and prothrombotic mediators. This view is consistent with results from both cultured and isolated cells in vitro when ω-3 fatty acids are supplied to isolated cells (41). To address these points in a pathophysiologic setting in the present investigations, we determined the appearance of EPA and DHA at local sites of inflammation in exudates from circulation.
By monitoring both deuterium-labeled d5-EPA and d5-DHA levels from circulation as well as increases in protein levels within exudates, we found that they appear coincident in the forming exudates. Of interest, both d5-EPA and d5-DHA were identified in exudates within 1 h of initiation of inflammation and maintained levels up to 48 h. At 48 h, both d5-EPA and d5-DHA levels were significantly greater within the exudates than their levels at 24 h. The first peak of D5 fatty acids was directly delivered from the circulation. Meanwhile, the second peak at 48 h most likely reflects recirculation and phospholipase A2 (PLA2) expression in the resolution of inflammation. Moreover, cytosolic PLA2 and secretory PLA2 were highly expressed during the resolution phase (51). The second peak at 48 h could be from esterified d5-EPA and d5-DHA and released by PLA2 mechanisms. Because in human plasma the t1/2 of EPA is 67 h and that of DHA is 20 h (48), the clearance of EPA and DHA is relatively slow compared with xenobiotic small molecules. Taken together, the present results suggest that circulating ω-3 fatty acids are available for sites of acute inflammation, initially, directly from circulation. The presence of albumin at sites of inflammation, by definition, determines whether the inflammatory site is considered an inflammatory exudate or transudate (37). The main protein component in the inflammatory exudates generated in the current zymosan-initiated peritonitis is, indeed, serum albumin demonstrated by two-dimensional gel electrophoresis and proteomics (35). Albumin is a well-appreciated carrier protein of unesterified fatty acids and particularly DHA (52).
Hence, the present results indicate that circulating DHA and EPA are directly used by developing inflammatory exudates and do not require specialized mobilization from complex lipids or specific phospholipase activation to initially impact inflammation and its timely resolution. Both EPA and DHA appeared rapidly in exudates in their unesterified form coincident with edema generation and movement of circulating albumin and leukocyte trafficking into the evolving inflammatory exudates. Lundy et al. (53) also indicated that edema formation and time course of arachidonic acid in mouse peritonitis were parallel. These findings in mice suggest that EPA and DHA are directly mobilized for resolvin production from the circulation via albumin as the most abundant and likely main carrier into the inflammatory sites. In humans, 13C-DHA in phosphatidylcholine was rapidly hydrolyzed and available as a free fatty acid in plasma (54). Furthermore, after ingestion of supplement, nonesterified EPA is detected in plasma (55). Taken together, these findings imply that in humans the circulating levels of EPA and DHA do not require storage and subsequent release from complex lipid precursors to have an important contribution to controlling inflammation and its resolution.
Next, we questioned the relationship between circulating ω-3 fatty acids and their anti-inflammatory properties in vivo. Using a new microfluidic chamber approach to rapidly isolate human neutrophils directly from circulating whole blood via capture on P-selectin-coated surface, we assessed the direct actions of both precursor DHA and RvD1 on single neutrophil chemotaxis responses. Earlier procedures required time-consuming isolation of neutrophils from whole blood before in vitro analyses. These protocols involved several steps of centrifugation and RBC lysis that usually take several hours (56) to perform that could lead to changes in the responsiveness and characteristics of the isolated neutrophils. By contrast, the P-selectin-based capture of neutrophils from a single drop of whole blood (∼5–10 μl) is capable of performing the PMN separation in less than 5 min. This short time interval is ideal for assessing the activation and/or inhibition status of neutrophils from the blood of healthy donors as well as patients. The combination of separation and assessment of shape and migration responses in the same chamber is closely akin to in vivo scenarios in which neutrophils roll on the endothelial surfaces (37), stick to the endothelium in regions of higher selectin expression, and respond via chemotaxis in the gradient of a chemokine, e.g., IL-8, and migrate into tissues (see Ref. 37). The small amount of blood used in the microfluidic chamber is an additional advantage for performing these analyses with human PMN because they circumvent the need for venous phlebotomy and associated risks.
Another key feature of this microfluidic chamber system is the ability to record real-time changes in morphology of PMN upon exposure to chemokines, DHA, and lipid mediators, such as RvD1, as well as to track migration through switches. The fast gradient switches in the chamber allowed visual assessment and recording of the earliest events after exposure of cells to RvD1 or native DHA as well as precise measurement of these changes in migration direction and velocity. Currently, no other chemotaxis systems are available that allow this level of precision or time resolution. For instance, in Boyden chambers (57), one cannot directly observe the cells, the information obtained is indirect, and the system requires a large number of cells and separate controls. The Boyden chamber is an end point assay, and one would not be able to assess whether neutrophils migrated before or after the addition of the inhibitor, or by adding the inhibitor before cells start moving if the effect would extrapolate to chemotaxis. In the Dunn and Zigmond chambers (58, 59), although one could visualize the cells in real time, it is not possible to swiftly switch between cell incubation and exposure conditions as with the present microfluidics chambers. The engineered microstructured valves for switching between independent gradients allowed the same neutrophils to serve as positive controls for migration in a chemoattractant gradient and then probed with either RvD1 or its precursor native DHA. The preservation of chemoattractant gradient before and after the switch is important for avoiding neutrophil responses to sudden changes of chemoattractant gradient. Of interest, sudden decrements in the concentration of chemokines alone have the potential to stop neutrophil migration for 3–5 min (60). Thus, the present direct assessment of DHA with PMN indicates that DHA itself is not a potent bioactive stop signal for PMN, but rather requires exudate conversion to RvD1 to evoke its signaling properties on these cells. Hence, following their actions, local tissues inactivate resolvins and permit organs to return to homeostasis.
Along these lines, ischemia-reperfusion is an event of significant clinical importance. Reperfusion-related tissue injury occurs during surgical procedures, particularly those involving extremities, causing both local and remote organ injury as well as increasing costs associated with prolonged postoperative recovery (38). Briefly, when vessels are surgically clamped or occlude, stasis of blood at the occlusion site leads to local ischemia and the neutrophils in that blood become activated. Upon release of the occlusion, activated leukocytes give rise to second organ or remote organ injury. For example, when neutrophils from the occluded vessels of the hind limb reach the lung, they cause local tissue damage (31). In many respects, the aberrant activation of neutrophils in this scenario exhibits features similar to those in uncontrolled acute inflammation and neutrophil-mediated tissue damage (31, 37, 38). Given the clinical importance and pathophysiology of this type of organ injury, we investigated the direct actions of DHA, resolvins, and related stable analogues (i.e., directly comparing the actions of RvD1, its 17-(R/S)-methyl analog, RvE1, and its 19-p-fluorophenoxy analog) in ischemia-reperfusion second organ injury. At equivalent doses, DHA was not protective, whereas RvD1 and its analog, as well as the stable analog of RvE1, showed potent antileukocyte actions, each reducing infiltration into lung tissues.
Native RvE1 itself was not able to protect the lung at these low doses most likely because of local inactivation. Of interest, RvE1 itself is both anti-inflammatory and proresolving in several inflammatory disease models (45). Recently, RvE1 was also shown to have potent actions in preventing joint damage and cartilage destruction in collagen-induced rodent arthritis (61) and protects from inflammation-induced bone loss in periodontal disease (39). Both RvD1 and RvE1 undergo site-specific metabolic inactivation (26, 40). Thus, the RvD1 and RvE1 analogues that display potent organ protective actions, as demonstrated in this study (Fig. 7), may provide new approaches to reduce organ damage characterized by excessive PMN infiltration. PMN are now appreciated to play an important role in the pathogenesis of many lifelong chronic inflammatory diseases such as arthritis (62). Recurring bouts of unresolved local acute inflammation may underlie many chronic inflammatory diseases. Thus, new means of monitoring neutrophils from peripheral whole blood within minutes, as demonstrated in this study with these newly designed microfluidic chambers, may have implications in selecting appropriate therapies for acute and chronic inflammatory diseases.
In summation, we tracked the movements of labeled EPA and DHA from circulation to inflammatory exudates and recorded in real time the direct actions of RvD1 on human PMN migration using a newly engineered microfluidic chamber (32, 60). Using this approach, we visualized individual cellular responses with neutrophils obtained from a single drop of peripheral blood in less than 5 min. This new system provided a unique opportunity to investigate the actions of RvD1 at the single-cell level. As shown in the Supplementary Video (online),5 PMN moving along the chemotactic gradient stopped movement essentially immediately after exposure to RvD1, but not with equimolar amounts of its exudate precursor DHA.
We also demonstrate new organ protective actions of resolvins in a murine model of second organ lung injury following hind limb ischemia-reperfusion. These results emphasize the substantial contribution of unesterified ω-3 fatty acid levels in circulation and their rapid availability via edema formation in exudates to mount resolution and assist in the return of local tissues to homeostasis. Recent review of the evidence for beneficial effects of fatty acids in human inflammation indicates a poor relationship and predictive value suggesting specific dose-response relationships for dietary ω-3 fatty acids, and fatty acid profiles of immune cells differ between species (63). These are consistent with our results and suggest that circulating ω-3 DHA and EPA levels and their transfer by edema rather than phospholipid-esterified precursors are an initial direct source for exudate biosynthesis of resolvins, and thus should be taken into account in evaluating the impact of nutrition in humans. The new pathways uncovered in this study are likely to be relevant in maintaining health, as well as in the many diseases characterized by excessive uncontrolled inflammation.
We thank Dr. Birgitta Schmidt (Department of Pathology, Children’s Hospital, Boston, MA) for microscopic evaluation of tissue histology, Dr. Jasim Uddin (University of Southern California) for preparation of some of the synthetic resolvins and their analogues, and Mary Halm Small for expert assistance with manuscript preparation.
C.N.S. is inventor on patents assigned to The Brigham and Women’s Hospital on resolvins that are subjects of licensing agreements and consultant arrangements for C.N.S.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by National Institutes of Health Grant R37-GM38765 (to C.N.S.); the Specialized Center for Oral Inflammation and Resolution, P50-DE016191 (to C.N.S. and N.A.P.); and the BioMEMS Resource Center, P41-EB002503 (to M.T. and D.I.). K.K. is the recipient of a postdoctoral fellowship award from the Arthritis Foundation.
Abbreviations used in this paper: PMN, polymorphonuclear neutrophil; d5-DHA,4Z,7Z,10Z,13Z,16Z,19Z-docosahexaenoic-21,21,22,22,22-d5 acid; d5-EPA, 5Z,8Z,11Z,14Z,17Z-eicosapentaenoic-19,19,20,20,20-d5 acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; GC-MS, gas chromatography-mass spectrometry; MPO, myeloperoxidase; PDMS, poly(dimethylsiloxane); PEG, polyethylene glycol; PLA2, phospholipase A2; RvD1, resolvin D1; RvE1, resolvin E1.
The online version of this article contains supplemental material.