Modulation of Lipid and Protein Mediators of Inflammation by Cytosolic Phospholipase A₂α during Experimental Sepsis¹ Free

Sepsis is a disease of systemic inflammatory response associated with infection (1). Its occurrence is rising, and the incidence rate for severe cases (i.e., sepsis with organ dysfunction) lies between 50 and 100 cases per 100,000 individuals in industrialized nations (2, 3). Although a gradual decrease in mortality rate has been achieved in recent decades, it still exceeds 30% according to recent reports (3). The pathogenesis of sepsis is thought to include an initial hyperinflammatory reaction evoked by infectious pathogens, followed by a hypo- or dysregulated inflammatory state (4, 5, 6). TLR and their ligands, various cytokines, and coagulation factors have all been identified as key molecules in this process, but our knowledge of sepsis pathogenesis remains incomplete. Cecal ligation and puncture (CLP)⁴ is an experimental procedure commonly used to model sepsis (7, 8, 9). In CLP, primary polymicrobial infectious foci cause septicemia, septic shock, and mortality.

Arachidonic acid metabolites, or eicosanoids, are a class of lipid mediators involved in inflammatory as well as other physiological processes (10, 11). PGs and leukotrienes (LT) are well characterized both in metabolic pathways and in signaling pathways through their respective cell surface G protein-coupled receptors (12, 13, 14, 15). They are generated from a single molecule of C20 polyunsaturated fatty acid, arachidonic acid, through multistep reactions catalyzed by oxygenases (cyclooxygenases (COX) and lipoxygenases (LO)), and terminal synthases. The rate-limiting step for eicosanoid generation is the availability of arachidonic acid. Phospholipase A₂ (PLA₂, EC 3.1.1.4) hydrolyzes membrane phospholipid at the sn-2 position to liberate fatty acid (16, 17). Among more than 20 mammalian PLA₂s, cytosolic PLA₂α (cPLA₂α; official gene symbol, Pla2g4a) is unique in its substrate preference for arachidonoyl-phospholipid, and gene-targeted mice have demonstrated its critical contribution to eicosanoid production from cells and in disease models (18, 19, 20, 21, 22, 23). We have reported attenuated disease symptoms for cPLA₂α-deficient mice in diverse models, including anaphylaxis (20), lung injury (24, 25), pulmonary embolism (26), arthritis (27), inflammatory bone resorption (28), experimental encephalomyelitis (29), and intestinal polyposis (30). We have also described an enhanced diabetic phenotype for cPLA₂α-deficient mice in NOD (31), indicating that cPLA₂α does not always promote disease.

In this study, we investigated the phenotype of cPLA₂α-deficient mice in the CLP model of sepsis. We observed suppressed accumulation of IL-6 and CCL2 in the peritoneum and serum, and a smaller decrease in body weight after surgery, as symptoms of attenuated disease. Interestingly, these milder responses did not result in a survival advantage for cPLA₂α-deficient mice. This contrasts to a recent report that 5-LO-deficient mice are protected from mortality in the CLP model (32). We conducted multiplex analyses of lipid mediators and cytokines in an attempt to obtain mechanistic insights into the disease phenotype.

C57BL/6J mice were from CLEA Japan. Male cPLA₂α-deficient (cPLA₂α^−/−) mice (20) had more than 12 generations of backcrosses to the C57BL/6J strain. cPLA₂α^−/− mice were obtained by crossing cPLA₂α heterozygotes (cPLA₂α^+/−). Mouse genotype was determined by PCR, as described (20). When comparisons by cPLA₂α genotype were attempted, wild-type (cPLA₂α^+/+) mice as well as cPLA₂α^−/− mice were prepared from cPLA₂α^+/− intercrosses. Commercially available C57BL/6J male mice (CLEA Japan) were used when comparisons among wild-type groups were intended. Animals were maintained in a light-dark cycle, with light from 7:00 a.m. to 8:00 p.m. at 22°C. Mice were fed with a standard laboratory diet and water ad libitum. All of the mice in this study were used under a protocol approved by the University of Tokyo Ethics Committee for Animal Experiments.

Male mice (20–30 g), anesthetized with sodium pentobarbital, 60 mg/kg, i.p., were given a 1- to 1.5-cm incision in the lower left abdominal wall. The cecum was exposed and ligated with a silk suture ∼5 mm below the ileocecal valve, and punctured twice with a 23-G needle. The cecum was placed in the abdomen, and the incision was closed with silk suture by two layers. Mice were returned to their cages with free access to food and water. Survival and body weight were monitored daily. As controls for the CLP-operated groups, sham and ligation groups were prepared; sham-operated mice did not receive ligation or puncture, and ligation-operated mice were cecum ligated, but not punctured.

Mice were anesthetized with sodium pentobarbital, 60 mg/kg, i.p., and placed in the supine position. Blood, collected from incisions made on femoral vessels, was either allowed to clot at room temperature for 1 h to obtain serum, or introduced into a heparinized tube (Mini Collect; Funakoshi) for plasma preparation. Serum was recovered by centrifugation at 1000 rpm for 15 min at 4°C. Plasma was obtained as the supernatant from spinning at 3000 × g for 10 min at room temperature, following the manufacturer’s instructions. These samples were stored at −80°C until use. Once bleeding had ceased, peritoneal lavage was conducted with 5 ml of ice-cold Dulbecco’s PBS without calcium chloride and magnesium chloride (D(−)-PBS) (Sigma-Aldrich) supplemented with 2 mM EDTA. Aliquoted peritoneal lavage fluid was stored at −80°C until use in mediator quantitations. Cell pellets were hemolyzed with ACK lysing buffer (Lonza) and resuspended to 1 ml with D(−)-PBS and 2 mM EDTA. The total cell number was counted with a hemocytometer, and the differential count was made with a Cytospin slide: Cytospin 3 (Thermo Shandon; 3 × 10⁴ cells, 300 rpm for 5 min), stained with Diff-Quik (International Reagents), and at least 300 cell counts in randomly selected areas.

The concentrations of inflammatory cytokines in serum and lavage samples were determined by a multiplex cytometric bead array (Mouse Inflammation CBA; BD Biosciences). This kit is designed to measure IL-6, IL-10, CCL2, IFN-γ, TNF-α, and IL-12 p70. Lipid mediators in peritoneal lavage fluid were quantitated by liquid chromatography-tandem mass spectrometry (LC-MS/MS), as described (33), allowing for the simultaneous monitoring of 14 mediators, as follows: 5 PGs (PGD₂, E₂, F₂α, I₂ (as 6-keto-PGF₁α), and thromboxane (Tx) A₂ (as TxB₂)), 3 LTs (LTB₄, C₄, and D₄), 5 hydroxy-eicosatetraenoic acids (HETEs; 5-, 8-, 11-, 12-, and 15-HETE), and platelet-activating factor (PAF).

After 1.5 h of surgery, mice were given 5 μg of carbaprostacyclin (cPGI₂; a stable analog of prostacyclin (PGI₂) with an agonist activity; Cayman Chemical) or 1 μg of PGE₂ (Cayman Chemical) by i.p. injection of 0.2 ml of saline (0.9% w/v NaCl) containing 5% (v/v) ethanol as vehicle.

Statistical calculations were performed with Prism-4 (GraphPad software). A value of p < 0.05 was considered to have statistical significance. The values for measurements were expressed as mean ± SEM.

Survival after CLP surgery was not altered by deletion of cPLA₂α (Fig. 1,A). With an impression that cPLA₂α^−/− mice had a better coat appearance during the first several days after surgery, body weight was monitored daily in a second experiment (Fig. 1,B). A 9% reduction in body weight was observed in both genotypes in the first 24 h after CLP surgery. This decrease was considered to be nonspecific because sham-operated mice showed equivalent transient body weight decreases during the same time period, followed by quick recovery to the original body weight within 1 wk (data not shown). Over the next 24 h, cPLA₂α^+/+ mice lost an additional 6% (total 15%) of their original body weight compared with the cPLA₂α^−/− loss of additional 3% (total 9%) in that period. These differences became larger for days 3 and 4, and were statistically significant, as follows: p < 0.0001, two-way ANOVA for comparisons of two curves, and p < 0.01 and p < 0.05 for days 3 and 4, respectively, by Bonferroni’s posthoc test. Despite the smaller body weight losses in cPLA₂α^−/− mice, this set of mice did not show a survival advantage either (Fig. 1 B, inset).

Previous studies of CLP-operated mice noted inflammatory cytokine accumulations within hours after surgery with impacts on mortality of CLP-operated mice (7, 9, 34). We chose a 5-h time point at which to measure inflammatory cytokine concentrations in sera using a multiplex cytometric bead array. IL-6 and CCL2 were the two most abundant mediators observed, followed by TNF-α (Fig. 2). The remaining three cytokines (IFN-γ, IL-10, and IL-12p70) were under the detection limits for the majority of serum samples, and were excluded from further analyses. IL-6 concentrations increased in a stepwise manner in cPLA₂α^+/+ mice, indicating that cecal ligation alone evoked inflammatory responses and that perforation of the cecum wall and subsequent leakage of its content further enhanced the response (Fig. 2,A). cPLA₂α^−/− mice showed similar IL-6 responses to ligation. Additional responses to cecal puncture appeared suppressed in the cPLA₂α^−/− group, resulting in a statistically significant difference by cPLA₂α genotype, as follows: 18.7 ± 3.6 ng/ml in the cPLA₂α^+/+ group (n = 20, mean ± SEM), and 11.5 ± 1.5 ng/ml in the cPLA₂α^−/− group (n = 15, mean ± SEM), p < 0.05, ANOVA, with Newman-Keuls’ multiple comparison test for pairwise comparison (Fig. 2,A). Upon cecal ligation, CCL2 concentrations increased to a similar extent in both genotypes of mice. Although cecal perforation in cPLA₂α^+/+ mice did not cause marked changes in serum CCL2 concentrations, cPLA₂α^−/− mice showed a decreasing trend (Fig. 2,B). TNF-α concentrations were mostly of the same patterns in both cPLA₂α^+/+ and cPLA₂α^−/− mice (Fig. 2 C). Cecal puncture did not induce any further increase in TNF-α concentrations compared with ligation groups for both genotypes.

To test whether the observed changes in inflammatory cytokine concentrations were restricted to serum samples, we examined peritoneal lavage fluid at the same time point as above, i.e., 5 h after surgery. Concentrations of peritoneal lavage fluid IL-6 showed almost the same pattern as that of serum IL-6 (Fig. 3,A). cPLA₂α^−/− mice responded in a similar manner to cecal ligation compared with cPLA₂α^+/+ mice. When the cecum was punctured, cPLA₂α^−/− mice stayed at the equivalent average concentrations of IL-6 (11.7 ± 1.8 and 12.0 ± 2.8 ng/ml for ligation and CLP groups, respectively, mean ± SEM, n = 8 both groups), in contrast to cPLA₂α^+/+ mice, whose IL-6 increased (10.8 ± 1.8 and 20.6 ± 3.4 ng/ml for ligation and CLP groups, mean ± SEM, n = 12 and 13, respectively,). CLP-operated cPLA₂α^+/+ mice contained a statistically significantly higher concentration of IL-6 compared with the ligation group of the same genotype and with CLP-operated cPLA₂α^−/− mice (p < 0.05, ANOVA with Newman-Keuls’ multiple comparison test for pairwise comparison). We did not detect differences between ligation groups. The pattern of lavage CCL2 concentrations mirrored that of lavage IL-6; statistical significance was limited to comparison of sham-operated and CLP-operated cPLA₂α^+/+ mice (Fig. 3 B). Concentrations of CCL2 in cPLA₂α^−/− peritoneal lavage fluid for ligation-operated and CLP-operated groups were close to those of the cPLA₂α^+/+ ligation group (2.9 ± 0.3, 2.7 ± 0.5, and 2.9 ± 0.4 ng/ml for ligation and CLP groups of cPLA₂α^−/− and ligation group of cPLA₂α^+/+, mean ± SEM, n = 8, 8, and 12, respectively). Only the CLP-operated cPLA₂α^+/+ group showed an elevated CCL2 concentration (4.4 ± 0.6 ng/ml, mean ± SEM, n = 13). TNF-α, IFN-γ, IL-10, and IL-12p70 were excluded from analyses because of rare occurrence of these molecules above the lower detection limit.

The dominant contribution of cPLA₂α to eicosanoid generation is well documented (18, 19). Using a LC-MS/MS device, we determined lipid mediator concentrations in the same peritoneal lavage fluid, as described above (33). Among the 14 mediators analyzed, PGI₂ (detected as 6-keto-PGF₁α, the stable natural derivative of PGI₂) and 12-HETE were the most abundant molecular species in CLP-operated cPLA₂α^+/+ mice, followed by PGE₂ (Fig. 3, C–E). LTB₄ was the major LT detected in peritoneal lavage (Fig. 3,F). Concentrations of 6-keto-PGF₁α for cPLA₂α^+/+ mice showed a stepwise pattern similar to those for serum and peritoneal lavage IL-6 concentrations (Figs. 2,A and 3, A and C). cPLA₂α^−/− mice generated 14–19% of the amount of 6-keto-PGF₁α of cPLA₂α^+/+ mice (Fig. 3,C). Significant differences were detected even for comparison of ligation groups (3.4 ± 0.3 and 0.6 ± 0.1 ng/ml for cPLA₂α^+/+ and cPLA₂α^−/− groups, mean ± SEM, n = 12 and 8, respectively, p < 0.01, ANOVA with Newman-Keuls’ multiple comparison test for pairwise comparison), as well as between CLP groups (5.2 ± 1.0 and 0.8 ± 0.1 ng/ml for cPLA₂α^+/+ and cPLA₂α^−/− groups, mean ± SEM, n = 12 and 8, respectively, p < 0.001, ANOVA with Newman-Keuls’ multiple comparison test for pairwise comparison). This differs from the observation for IL-6 and CCL2, in which no differences by genotypes were detected between ligation groups. Interestingly, 12-HETE showed a distinct pattern (Fig. 3,D). No significant differences were observed for any comparisons by genotype or surgical operation. PGE₂ behaved in a similar pattern as 6-keto-PGF₁α, i.e., a stepwise increase upon ligation and cecal puncture on cPLA₂α^+/+ mice, and a more than 80% decrease upon cPLA₂α gene disruption (Fig. 3,E). The absolute values of PGE₂ were smaller than those of 6-keto-PGF₁α by a factor of ∼1/6. LTB₄ was detected in abundance in cPLA₂α^+/+ mice when the cecum was ligated and punctured, but not when the cecum was only ligated. The concentrations of LTB₄ from cPLA₂α^−/− mice stayed near the lower limit of detection, 40 pg/ml, for all three groups (Fig. 3,F). Results for the remaining lipid mediators were shown as Table I.

In summary, cPLA₂α^−/− mice showed attenuated generation of lipid and protein inflammatory mediators, but no survival advantage compared with cPLA₂α^+/+ mice.

Having observed robust accumulations of both cytokines and lipid mediators in the peritoneum at 5 h after ligation and CLP surgery, we sought to assay the differential timelines of lipid and protein mediators. We examined 1.5-h samples from C57BL/6J male mice, seven mice for each surgical group. LC-MS/MS analysis of peritoneal lavage samples revealed accumulation of 6-keto-PGF₁α, 12-HETE, and PGE₂ to levels comparable to those seen 5 h after surgery (Fig. 4 and Table II). However, inflammatory cytokines were undetectable or under reliable ranges for quantitation for all six cytokines in either peritoneal lavage or serum samples. These findings indicate that lipid mediators are generated earlier than inflammatory cytokines in the peritoneum. LTB₄ rarely (one and two of seven mice for ligation-operated and CLP-operated groups gave reliable measurement, respectively) exceeded the quantitation limit (40 pg/ml) at the 1.5-h time point (Table II), suggesting that production of LTB₄ starts later than other eicosanoids.

The distinct pattern of peritoneal lavage 12-HETE compared with other mediators at 5 h after surgery prompted us to ask whether this pattern is preserved in plasma. The average 12-HETE concentrations were found to be 100–300 ng/ml, with no statistically significant differences (p = 0.29, ANOVA, n = 7 each group; Table III). Basal 12-HETE concentrations in plasma and in serum of wild-type mice were determined in parallel to be lower than 10 ng/ml and higher than 3.5 μg/ml, respectively (n = 3).

Some eicosanoids, such as LTs, are known for strong chemoattractivity. Although one possible mechanism underlying decreased inflammatory cytokine accumulations in cPLA₂α^−/− mouse is changes in infiltrating cell populations into the site of inflammation, neither cellularity, nor differential counts of peritoneal exudates collected at 5 h after surgery was found to be significantly different by either genotype or treatment (Table IV).

The results described above suggested a model in which lipid mediators modulate responses of the whole body directly and/or indirectly through cytokine production. The 5-LO-deficient mice were protected from CLP-induced death (32), and cPLA₂α^−/− mice could be regarded as mice with both 5-LO and COX pathway deficiency given their attenuated productions of both LTs and PGs. Complementation experiments were conducted where we treated CLP-operated mice i.p. with 5 μg of cPGI₂, a stable PGI₂ analog, or 1 μg of PGE₂, at 1.5 h after surgery. Smaller body weight losses for cPLA₂α^−/− mice were reproduced, as described earlier, after 2 days of surgery (Fig. 5, A and B). Among the same genotype groups, neither cPGI₂ treatment nor PGE₂ treatment had any effect on body weight changes for 2 wk (Fig. 5, A and B). We could not find significant effects on survival for any combination of genotypes and treatments (Fig. 5, C and D).

As described in Materials and Methods, peritoneal lavage was preceded by blood collection. Thus, each mouse used to generate Fig. 3 has corresponding data points in Fig. 2. Scattergrams were plotted to find correlations between the concentrations of major mediators (Fig. 6). Serum IL-6 and CCL2 showed a moderate linear correlation (R² = 0.440, all mice combined, n = 75) that appeared more profound for cPLA₂α^+/+ mice (R² = 0.529, cPLA₂α^+/+, n = 46, and R² = 0.234, cPLA₂α^−/−, n = 29; Fig. 6,A). The correlation was more evident for lavage IL-6 and CCL2 concentrations (R² = 0.866, all mice combined, n = 54), and this linear correlation was not affected by cPLA₂α genotype (R² = 0.862, cPLA₂α^+/+, n = 34, and R² = 0.875, cPLA₂α^−/−, n = 20; Fig. 6,B). Concentrations of 6-keto-PGF₁α and PGE₂ in peritoneal lavage had the strongest linear correlations, less so in cPLA₂α^−/− mice (R² = 0.881, cPLA₂α^+/+, n = 36, and R² = 0.171, cPLA₂α^−/−, n = 18; Fig. 6, C and D). When serum mediator concentrations were plotted against peritoneal lavage IL-6 concentrations (Fig. 6, E and F), they showed a different trend from those between serum (Fig. 6,A) and lavage (Fig. 6,B) IL-6 and CCL2. Correlations between IL-6 concentrations appeared to have a higher-order relationship rather than a linear one. Although bar graphs of lavage IL-6 and 6-keto-PGF₁α from cPLA₂α^+/+ mice shown in Fig. 3, A and C, had certain similarities, the plot for these two mediators turned out to have only a modest linear correlation (R² = 0.469, cPLA₂α^+/+, n = 36; Fig. 6 G). We could not find any apparent correlations for 12-HETE concentrations with any of the mediators quantitated in this study (data not shown).

Lipid mediators generated through cPLA₂α are undoubtedly involved in inflammatory and allergic responses. Previous studies demonstrated a marked reduction in disease in cPLA₂α^−/− mice for various inflammatory and allergic disease models, as follows: anaphylaxis (20), brain infarction (21), acute lung injury (24), lung fibrosis (25), arthritis (27), inflammatory bone resorption (28), and experimental encephalomyelitis (29). This strongly suggests that cPLA₂α is one of the key molecules in immunological pathophysiology. In this study of experimental sepsis, we demonstrated attenuated inflammatory responses in cPLA₂α^−/− mice. Accumulations of IL-6 and CCL2 in the peritoneum and serum were significantly suppressed, and smaller body weight losses were observed compared with cPLA₂α^+/+ mice. The amounts of PGs and LTB₄ in the peritoneum also showed marked decreases relevant to the lack of cPLA₂α activity. One might expect better prognosis of cPLA₂α^−/− mice for all of these circumstances; however, we could not find any differences in the survival curves by cPLA₂α genotype (Figs. 1–3).

The diverse eicosanoids include products with counteracting activity in many systems, including the following: 1) pro- and anti-aggregatory effects on platelets by TxA₂ and PGI₂ (26, 35); 2) sleep/awake regulation by PGD₂/PGE₂ (36); 3) LTs and PGs in asthma (11, 14, 37, 38, 39, 40, 41); and 4) PGI₂ and cysteinyl LTs in bleomycin-induced lung injury (25, 42, 43, 44, 45, 46, 47). Products of the 5-LO pathway are thought to have proinflammatory functions in murine CLP (32, 48). The 5-LO-deficient mice are protected against CLP-induced death, and cysteinyl LTs (the collective name for LTC₄, D₄, and E₄) were pharmacologically suggested to be key mediators of CLP lethality acting through effects on vascular permeability (32). Mice deficient in BLT1, the high-affinity receptor for LTB₄, were slightly protected against CLP-induced death (48). We noticed dramatic decreases in LTB₄ accumulation in the peritoneal lavage fluid of CLP-operated cPLA₂α^−/− mice (Fig. 3,F). Although the peritoneum did not exhibit marked accumulation of cysteinyl LT in CLP, it seems reasonable to presume decreased production of cysteinyl LTs in cPLA₂α^−/− mice as well, because we have documented marked decreases in cysteinyl LT accumulations in bronchoalveolar lavage fluid upon acute lung injury (24) and in bleomycin-induced lung fibrosis models (25). At a cellular level, multiple reports, including ours, have described diminished production of cysteinyl LTs from stimulated macrophages and mast cells (20, 21, 49, 50, 51). Hypothesizing a counteraction model in CLP, we suggest that favorable effects on survival, given decreased 5-LO product(s) in cPLA₂α^−/− mice, are cancelled out by lack of another eicosanoid(s) at the same time. LC-MS/MS analyses of peritoneal lavage fluid on 14 different lipid mediators indicated that two COX pathway products, 6-keto-PGF₁α, the stable metabolite of PGI₂, and PGE₂, were abundantly accumulated in the lavage fluid, and that more than 80% of their productions was mediated by cPLA₂α (Fig. 3, C and E). We attempted to demonstrate roles for these PGs by treating mice with cPGI₂, a stable PGI₂ analog, or PGE₂ 1.5 h after the surgery, but detected no differences in body weight changes or survival by the treatment (Fig. 5). We could not find any effects on cytokine production using C57BL/6J male mice treated with cPGI₂ or PGE₂ following ligation or CLP operation (data not shown). A previous pharmacological study with a selective inhibitor of COX-2 found improvement for early survival in murine endotoxemia, but not in the CLP model (52). Candidate fatty acid-derived lipid mediators on the protective side include arachidonic acid-derived lipoxins, and eicosapentaenoic acid- and docosahexaenoic acid-derived resolvins and protectins, which have anti-inflammatory functions (53, 54, 55, 56). Considering strong antimicrobial effector functions of LTB₄ (57), lack of LTB₄ production in cPLA₂α^−/− mice might have resulted in insufficient bacterial clearance masking the favorable aspects of decreased LT production. Further studies are needed to determine whether this counterregulatory model of lipid mediators holds for CLP and sepsis.

We can consider another counterregulatory model at the level of cytokines. Both IL-6 and CCL2 showed decreases for CLP-operated cPLA₂α^−/− mice compared with their genotype control group (Figs. 2 and 3). IL-6 concentration is generally believed to reflect the strength of inflammatory responses in CLP-operated mice. Higher concentrations of plasma IL-6 after 6 h of CLP surgery were shown to be indicative of early mortality in female BALB/c mice (34). In contrast, CCL2 has been demonstrated to have anti-inflammatory roles in CLP by the use of neutralizing Ab (58, 59, 60) and CCL2-deficient mice (61); mice with deficient or blocked CCL2 functions were more susceptible to CLP-induced mortality. Decreases in both IL-6 and CCL2 responses in CLP-operated cPLA₂α^−/− mice might have functionally cancelled each other out. The concentrations of IL-6 and CCL2 in ligation-operated mice did not appear to have any differences by cPLA₂α genotype. When cPLA₂α^−/− mice were CLP operated, they showed small changes in the average concentrations of these mediators compared with the ligation group in contrast to cPLA₂α^+/+ mice (Figs. 2 and 3). This suggests that IL-6 and CCL2 responses can be divided into two mechanisms, as follows: a cPLA₂α-independent ligation-induced mechanism, and a cPLA₂α-dependent puncture-induced mechanism. The latter might be explained by the ability of LTB₄ to transcriptionally activate both IL-6 and CCL2 expression, which was shown with cultured human peripheral blood monocytes (62, 63). Of interest is that LTB₄, IL-6, and CCL2 are detected from 5-h but not in the 1.5-h harvests, unlike certain PGs and 12-HETE, which were abundantly detected starting at the earlier time point. Specific detection of LTB₄ in the CLP-operated cPLA₂α^+/+ mice (Fig. 3,F) suggests the existence of a puncture-specific mechanism for lipid mediator generation. This might be related to specific cell population(s), although the current study failed to detect any differences in infiltrating inflammatory cell numbers (Table IV).

TNF-α is one of the critical mediators for CLP and sepsis (64). At cellular levels, both resident (31) and proteose peptone-elicited (data not shown) peritoneal macrophages produce more TNF-α by LPS stimulation compared with genotype controls. Lipid mediators suggested to be responsible for this modulation of TNF-α are PGE₂, and PGI₂ acting through respective EP2 or EP4, and IP G protein-coupled receptor on the cell surface (65, 66). This comprises an example of a negative regulatory role for lipid mediators on cytokine generation, which might have affected the CLP phenotype in cPLA₂α^−/− mice, although we could not detect any changes in the TNF-α production profile by cPLA₂α genotype in this study (Fig. 2 C).

The 12-HETE was unexpectedly revealed to be one of the most abundant eicosanoids in peritoneal lavage fluid (Fig. 3,D). Its concentration was in the same range as 6-keto-PGF₁α, and exceeded those of PGE₂ and LTB₄ by 5- to 10-fold. Notably, 12-HETE concentrations did not show a stepwise pattern upon ligation and puncture of cecum as observed for other mediators in peritoneal lavage fluid and plasma (Table III). Furthermore, cPLA₂α genotype did not affect 12-HETE accumulation. Murine serum contains μg/ml levels of 12-HETE, and we have previously reported that serum from cPLA₂α^−/− mice contained significantly less 12-HETE than from wild-type mice in two genetic backgrounds (C57BL/6J and C3H/HeN) (26). This suggests that 12-HETE production in CLP is regulated by a distinct cPLA₂α-independent mechanism(s) apart from platelet-derived 12-HETE accumulation in serum, which is mostly cPLA₂α dependent. The physiological role of 12-HETE, 12-LO, and 12/15-LO products is documented for platelet aggregation (67), cancer metastasis (68), atherosclerosis (69, 70), and neuronal plasticity (71), but is not known for sepsis and CLP. The functions of this abundant lipid mediator, as well as the PLA₂ molecule(s) responsible for its production, remain to be elucidated.

The lack of differences in infiltrating inflammatory cell populations by cPLA₂α genotype (Table IV) stands apart from previous observations of inflammatory disease model studies, in which marked decreases in infiltrating inflammatory cell numbers were associated with attenuated disease phenotypes (24, 25, 27, 29). We previously reported no significant differences in either cellularity or differential counts for casein-induced inflammatory exudate cells recovered from the peritoneal cavity of cPLA₂α^−/− mice compared with cPLA₂α^+/+ controls at 4–5 h after i.p. casein injection (72). The degree of eicosanoid contribution to inflammatory cell infiltration appears to differ depending on triggers and/or locations of inflammation.

Scattergram analysis (Fig. 6) is useful to unveil shared regulatory mechanisms for pairs of parameters (73). Linear correlations were best detected for two pairs of mediators: lavage IL-6 and CCL2 (Fig. 6,B), and 6-keto-PGF₁α and PGE₂ (Fig. 6,C). It is tempting to think that correlated mediators are produced by a shared set of cells using common transcription factors and/or metabolic enzymes. Identification of such putative cell types and determination of stimuli and signaling pathways that produce correlated amounts of mediators should greatly advance our knowledge. Serum IL-6 and CCL2 concentrations (Fig. 6,A) were less linearly correlated than those of peritoneal lavage fluid (Fig. 6,B), observed as smaller R² values for the former. Much less linear correlations were observed for lavage IL-6 concentrations and serum cytokine and peritoneal lavage 6-keto-PGF₁α concentrations (Fig. 6, E–G). These weaker correlations suggest function(s) of undefined modulating factors for production of these mediators.

The production of multiple lipid mediators in various cell and tissue types is regulated by cPLA₂α activity. One should not assume that cPLA₂α deficiency leads in only one direction, either attenuated or enhanced disease. The attenuated phenotypes of cPLA₂α-deficient mice do not simply indicate one-sided effects of lipid mediators, but rather the dominance of the prosymptomatic side of lipid mediators in the wild-type animals. One can find reports of adverse effects of cPLA₂α deficiency, as follows: intestinal ulcers (30), the diabetic model in NOD background (31), hypoxic pulmonary vasoconstriction (74), renal functions (75), and female reproduction (20, 21, 76). Not only mouse models, but an inherited human cPLA₂α deficiency, reported recently, was associated with impaired eicosanoid biosyntheis, small intestinal ulceration, and platelet dysfunction (77). We do not yet have a global view of eicosanoid functions, because of their multiplicity of active molecular species. Another complexity for lipid mediator functions arises from molecular diversity of mammalian PLA₂. Six gene loci were identified for the mammalian cPLA₂ (18, 19, 78, 79), and together with secretory PLA₂ (sPLA₂), calcium-independent PLA₂, and PAF acetyl hydrolase categories, they form a large PLA₂ family of more than 20 members (16, 17). As we noted for 12-HETE, cPLA₂α does not always contribute to the major part of eicosanoid generation. Is 12-HETE in CLP-operated mice an exception? Many sPLA₂s have been claimed to be involved in lipid mediator generation, but little compelling in vivo evidence has been provided to date (17). A recent study using group X sPLA₂ (official gene symbol, Pla2g10)-deficient mice demonstrated profound prodisease roles for this sPLA₂ in experimental asthma models (80). It seems unlikely that simple inhibition of one specific PLA₂ would be of significant benefit to septic patients. Indeed, clinical trials in sepsis targeting two PLA₂ molecules (group IIA sPLA₂ (official gene symbol, PLA2G2A) and secretory PAF acetylhydrolase (official gene symbol, PLA2G7)) failed (81, 82, 83). Still, we believe that lipid mediators are good therapeutic targets in sepsis for two reasons. First, they are a class of mediators distinct from cytokines and coagulation factors, their production having a distinct regulatory mechanism. Second, they function interactively and specifically with other classes of mediators. Involvement of lipid mediators in various aspects of sepsis pathophysiology is expected, and multiplex approaches such as those described in this study will be powerful tools for uncovering the underlying mechanisms.

We thank T. Yokomizo, S. Ishii, and M. Taniguchi for discussion, and C. L. Karp for discussion and critical reading of the manuscript.

The authors have no financial conflict of interest.