Activation of Cytosolic Phospholipase A2α in Resident Peritoneal Macrophages by Listeria monocytogenes Involves Listeriolysin O and TLR2*

Eicosanoid production by macrophages is an early response to microbial infection that promotes acute inflammation. The intracellular pathogen Listeria monocytogenes stimulates arachidonic acid release and eicosanoid production from resident mouse peritoneal macrophages through activation of group IVA cytosolic phospholipase A2 (cPLA2α). The ability of wild type L. monocytogenes (WTLM) to stimulate arachidonic acid release is partially dependent on the virulence factor listeriolysin O; however, WTLM and L. monocytogenes lacking listeriolysin O (ΔhlyLM) induce similar levels of cyclooxygenase 2. Arachidonic acid release requires activation of MAPKs by WTLM and ΔhlyLM. The attenuated release of arachidonic acid that is observed in TLR2-/- and MyD88-/- macrophages infected with WTLM and ΔhlyLM correlates with diminished MAPK activation. WTLM but not ΔhlyLM increases intracellular calcium, which is implicated in regulation of cPLA2α. Prostaglandin E2, prostaglandin I2, and leukotriene C4 are produced by cPLA2α+/+ but not cPLA2α-/- macrophages in response to WTLM and ΔhlyLM. Tumor necrosis factor (TNF)-α production is significantly lower in cPLA2α+/+ than in cPLA2α-/- macrophages infected with WTLM and ΔhlyLM. Treatment of infected cPLA2α+/+ macrophages with the cyclooxygenase inhibitor indomethacin increases TNFα production to the level produced by cPLA2α-/- macrophages implicating prostaglandins in TNFα down-regulation. Therefore activation of cPLA2α in macrophages may impact immune responses to L. monocytogenes.

Eicosanoid production is an early response to microbial infection that can regulate innate immunity (15,16). Eicosanoids are oxygenated metabolites of arachidonic acid that exert their potent biological effects by binding to G-proteincoupled receptors (17). Arachidonic acid is metabolized through the cyclooxygenase (COX) and lipoxygenase pathways for the production of prostaglandins and leukotrienes, respectively. Leukotrienes induce acute inflammatory responses such as increased vascular permeability and recruitment of granulocytes. Prostaglandins have pro-inflammatory effects by increasing vascular permeability but also exert immune-suppressive effects (18). Resident peritoneal macrophages have been used extensively as a model to study the regulation of arachidonic acid release and eicosanoid production; however, these responses have not been investigated thoroughly in the context of bacterial infection. In this study we demonstrate that nonopsonized L. monocytogenes activates group IVA cytosolic phospholipase A 2 (cPLA 2 ␣) in resident peritoneal macrophages resulting in arachidonic acid release and eicosanoid production. Optimal activation of cPLA 2 ␣ involves the bacterial virulence factor LLO and host toll-like receptor 2 (TLR2). Our results also demonstrate an important role for cPLA 2 ␣ activation in suppression of TNF␣ production by macrophages infected with L. monocytogenes, implicating eicosanoids in regulating immune responses to L. monocytogenes.

EXPERIMENTAL PROCEDURES
Materials-Pathogen-free ICR mice were obtained from Harlan Sprague-Dawley and used for all experiments unless otherwise specified. TLR2 and MyD88 knock-out mice (C57BL/6) were generated as described previously (19), and age-matched control C57BL/6 mice were obtained from Harlan Sprague-Dawley. cPLA 2 ␣ Ϫ/Ϫ mice were generated using 129 embryonic stem cells in a C57BL/6 strain as described previously (20). The mixed strain was backcrossed onto a Balb/c background and used after 10 generations. All mice were used between 5 and 10 weeks of age for macrophage isolation. [5,6,8,9,11,12,14, H]Arachidonic acid (specific activity 100 Ci/mmol) was from PerkinElmer Life Sciences. Fetal bovine serum (Gemini Bio-Products) was heat-inactivated at 56°C for 30 min before use. DMEM was from Cambrex BioScience Walkersville, Inc. Hanks' balanced salts solution was purchased from Invitrogen. Human serum albumin was obtained from Intergen. The cPLA 2 ␣ inhibitor, pyrrolidine-2, was generously provided by Dr. Michael Gelb (University of Washington, Seattle). Polyclonal antibody to cPLA 2 ␣ was raised as described (21). Antibodies to phosphorylated extracellular signal-regulated kinases (ERKs) and p38 were obtained from Cell Signaling Technology, Inc. Polyclonal antibody to murine COX2 was obtained from Cayman Chemical Co. Antibody to L. monocytogenes was obtained from Difco (BD Biosciences). Gentamicin and latrunculin A were from Sigma. U0126 and SB202190 were obtained from Calbiochem. Complete protease inhibitor tablets were obtained from Roche Diagnostics. Tryptic soy agar and brain heart infusion broth were purchased from Fluka Bio-Chemika and BD Biosciences, respectively. Cyto-Tox ONE homogeneous membrane integrity assay kit was obtained from Promega. Fura Red-AM was from Invitrogen.
Macrophage Isolation and Arachidonic Acid Release Assay-Resident mouse peritoneal macrophages were obtained by peritoneal lavage using 8 ml of DMEM containing 10% heat-inactivated fetal bovine serum, 100 g/ml streptomycin sulfate, 100 units/ml penicillin G, and 2 mM glutamine (supplemented DMEM) containing 10 units/ml heparin. Cells were plated at 5 ϫ 10 5 cells/well (48-well plate) and incubated for 2 h at 37°C in a humidified atmosphere of 7.5% CO 2 in air. Cells were washed twice with calcium-and magnesium-free Hanks' balanced salts solution to remove nonadherent cells followed by one wash with supplemented DMEM. Macrophages were cultured in supplemented DMEM containing [ 3 H]arachidonic acid (0.1 Ci/250 l/well) for 16 -18 h at 37°C. The cells were washed twice with antibiotic-free stimulation medium to remove unincorporated [ 3 H]arachidonic acid, and then infected with L. monocytogenes in stimulation medium. The culture medium was removed at the indicated times after infection, centrifuged, and the amount of radioactivity in the supernatant measured by scintillation counting. The cell-associated radioactivity was measured following solubilization of the monolayer with 0.1% Triton X-100. The amount of radioactivity released into the culture medium was expressed as percent of the total radioactivity incorporated (cell-associated plus medium).
Cell Cytotoxicity Assay-Resident mouse peritoneal macrophages were plated in 96-well plates (0.16 ϫ 10 6 cells/well) and cultured as described above. The effect of L. monocytogenes on macrophage viability was assayed using Cyto-Tox ONE homogeneous membrane integrity assay kit according to manufacturer's protocol using the LS55 spectrofluorometer (PerkinElmer Life Sciences). The macrophage culture medium was collected at various times after infection with L. monocytogenes and 50 l used to determine the percent lactate dehydrogenase (LDH) release.
L. monocytogenes Uptake Assay-Macrophages plated at 0.5 ϫ 10 6 cells/well (48-well plate) were cultured overnight in complete medium and then incubated with WTLM and ⌬hlyLM (m.o.i. 25) in stimulation medium for 1 h. Macrophages were washed twice and incubated in stimulation medium containing gentamicin (50 g/ml) for 30 min to kill extracellular bacteria. After washing twice with PBS, macrophages were lysed with 0.1% Triton X-100 in PBS and the intracellular L. monocytogenes plated on tryptic soy agar plates. The number of colony-forming units (CFU) was determined after incubation of plates for 22-24 h at 37°C. For comparing uptake of L. monocytogenes by peritoneal macrophages from cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ mice, a similar protocol was used except the macrophages were not incubated with gentamicin. After a 1-h incubation with WTLM, the macrophages were washed six times to remove extracellular L. monocytogenes following a protocol described previously (22).
Eicosanoid and Cytokine Analysis-Resident mouse peritoneal macrophages were plated as described for arachidonic acid release assays but incubated overnight without [ 3 H]arachidonic acid. Macrophages were incubated with either WTLM or ⌬hlyLM for 1 h, washed, and incubated in stimulation medium containing 50 g/ml gentamicin. The culture medium was harvested 3 h after infection, complete protease inhibitor tablet (1ϫ) added, and stored at Ϫ20°C. Eicosanoids in the culture medium were quantified by enzyme-linked immunosorbent assay (ELISA Tech, Aurora, CO). TNF␣ in the culture medium was quantified by enzyme-linked immunosorbent assay (ELISA Tech, Aurora, CO) and by Luminex assay, which gave similar results.
Western Blots-Macrophages were washed twice in ice-cold PBS and lysed in buffer containing 50 mM HEPES, pH 7.4, 150 mM sodium chloride, 10% glycerol, 1% Triton X-100, 1 mM EGTA, 1 mM EDTA, 200 M sodium vanadate, 10 mM tetrasodium pyrophosphate, 100 mM sodium fluoride, 300 nM p-nitrophenyl phosphate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin. After incubation on ice for 30 min, cell lysates were centrifuged at 15,000 rpm for 15 min, and protein concentration in the supernatant was determined by the bicinchoninic acid method. Lysates were boiled for 5 min after addition of Laemmli buffer, and then proteins (10 -30 g) were resolved on 10% SDS-polyacrylamide gels. After transfer, nitrocellulose membranes were incubated in blocking buffer (20 mM Tris-HCl buffer, pH 7.6, containing 137 mM NaCl, 0.05% Tween, and 5% nonfat milk), and then incubated overnight at 4°C with polyclonal antibodies to cPLA 2 ␣, COX2, phospho-ERKs, and monoclonal antibody to phospho-p38. The membranes were incubated with horseradish peroxidase-linked secondary antibodies (1:5000) in blocking buffer for 30 min at room temperature. The immunoreactive proteins were detected using the Amersham Biosciences ECL system.
Production of Recombinant Adenovirus and Microscopy-Macrophages (5 ϫ 10 5 ) were plated onto glass-bottomed Mat-Tek dishes in complete medium, incubated for 2 h, and then washed three times. Enhanced cyan fluorescent protein-cPLA 2 ␣ (ECFP-cPLA 2 ␣) was expressed in resident peritoneal macrophages using recombinant adenovirus (23). Human cPLA 2 ␣ was cloned into the ECFP vector (Clontech), and Ala-206 in ECFP was mutated to Lys to produce the monomeric form. ECFP-cPLA 2 ␣ was subcloned into pVQ CMV k-NpA vector, and virus (AdECFP-cPLA 2 ␣) was produced and purified by ViraQuest Inc. (North Liberty, IA). Macrophages were incubated in stimulation medium (150 l) containing AdECFP-cPLA 2 ␣ for 2 h. Supplemented DMEM (1.0 ml) was added, and the macrophages were incubated at 37°C in 7.5% CO 2 for 26 h. Macrophages were washed and incubated in stimulation medium with EGFP-WTLM (m.o.i. 25), and then fixed with 3% paraformaldehyde in PBS containing 3% sucrose for 15 min as described previously (23). In some experiments, paraformaldehyde-fixed macrophages infected with WTLM or ⌬hlyLM were probed with rabbit polyclonal antibody to L. monocytogenes followed by Texas Red-conjugated secondary antibody. Golgi was visualized using a rabbit polyclonal antibody to Giantin followed by Texas Red-conjugated secondary antibody. Cells were imaged on an inverted Zeiss 200M microscope with a 175-watt xenon lamp using 63ϫ oil immersion objective. Images were acquired using a CCD camera from Sensicam and data analyzed using Intelligent Imaging Innovations Inc. (3I) software.
Calcium Imaging-Macrophages (5 ϫ 10 5 ) plated onto glassbottomed MatTek dishes were cultured overnight, washed with phenol red-free DMEM containing 2 mM probenecid and 25 mM HEPES, pH 7.4 (imaging medium), and then incubated in imaging medium containing 5 M Fura Red-AM and 0.02% pluronic at room temperature. After 30 min, the macrophages were washed three times and incubated in imaging medium for an additional 30 min. L. monocytogenes strains were added, and live cell imaging was carried out at room temperature using an inverted Zeiss 200M microscope with a 40ϫ oil immersion objective and a Fura Red long pass dichroic mirror and emission filter set (Chroma) for the calcium indicator. Cells were illuminated at 403 and 490 nm (to detect calcium-bound and calcium-free form of the Fura Red indicator, respectively). Images were acquired using a CCD camera from Sensicam, and data were analyzed using Intelligent Imaging Innovations Inc. (3I) software.
Statistical Analysis-Statistics were calculated in GraphPad using paired t test to obtain two-tailed p values.

RESULTS
L. monocytogenes Stimulates cPLA 2 ␣-mediated Arachidonic Acid Release from Resident Peritoneal Macrophages-Unopsonized WTLM stimulated arachidonic acid release from resident mouse peritoneal macrophages at multiplicities of infection (m.o.i.) in the range of 5-25 (bacteria:macrophage) (Fig. 1A). An L. monocytogenes mutant strain lacking LLO (⌬hlyLM) induced ϳ60% less arachidonic acid release than WTLM at 60 min after infection. The role of L. monocytogenes PLCs was also investigated by testing the double mutant lacking PI-PLC and BR-PLC, and the triple mutant strain lacking LLO, PI-PLC, and BR-PLC. Results demonstrated that PLCs do not play a role in stimulating arachidonic acid release because the response to the double PLC mutant was identical to WTLM, and the response to the triple mutant was identical to ⌬hlyLM (data not shown). Arachidonic acid release induced by WTLM was inhibited by the cPLA 2 ␣ inhibitor pyrrolidine-2 ( Fig. 1B) and was reduced by 85% in cPLA 2 ␣ Ϫ/Ϫ macrophages infected with WTLM and was near background levels in cPLA 2 ␣ Ϫ/Ϫ macrophages infected with ⌬hlyLM (Fig. 1C). Thus nonopsonized L. monocytogenes stimulates cPLA 2 ␣-mediated arachidonic acid release from resident peritoneal macrophages that involves the virulence factor LLO.
A time course illustrated that the release of arachidonic acid increased continuously from 20 min to 3 h after WTLM infection ( Fig. 2A). In contrast, arachidonic acid release increased up to 60 min after ⌬hlyLM infection and then leveled off. The amount of arachidonic acid released 20 min after infection with WTLM and ⌬hlyLM was not significantly different, but by 40 min WTLM induced significantly higher levels of arachidonic acid release. Because L. monocytogenes proliferates in tissue culture medium, LLO released from growing WTLM over time may promote toxicity in resident peritoneal macrophages. As shown in Fig. 2B, ⌬hlyLM was not cytotoxic during the 3 h of infection. In contrast, WTLM at m.o.i. of 13 and 25 induced 38 and 80% release of LDH, respectively, by 3 h. Therefore, the release of arachidonic acid beyond 60 min by WTLM correlated with cytotoxicity. The ability of WTLM and ⌬hlyLM to stimulate arachidonic acid release was investigated under conditions in which growth of extracellular L. monocytogenes was inhibited with gentamicin. When macrophage cultures were washed 1 h after addition of WTLM, and then incubated in the presence of gentamicin for 2 h, cytotoxicity was prevented (data not shown). To determine the effect of blocking extracellular growth on arachidonic acid release, macrophages labeled with [ 3 H]arachidonic acid were infected (m.o.i. 25) with L. monocytogenes for 60 min, and the cultures were rinsed and then incubated in media containing gentamicin. The amount of arachidonic acid released into the culture medium was measured at 1, 2, and 3 h after adding gentamicin (Fig. 2C). During the initial 60 min infection (no gentamicin), the level of arachidonic acid released in response to WTLM and ⌬hlyLM was 9 and 3.5%, respectively, similar to the results shown in Fig. 2A. After removing the culture medium and adding fresh medium containing gentamicin, the release of arachidonic acid 60 min after adding gentamicin was 5-fold greater from macrophages infected with WTLM than those infected with ⌬hlyLM suggesting that cPLA 2 ␣ remained more active in macrophages infected with WTLM. There was little additional release of arachidonic acid from macrophages infected with WTLM at 120 and 180 min after adding gentamicin, unlike cultures without gentamicin (compare Fig. 2, A and C). The results suggest that extracellular LLO contributes to cPLA 2 ␣ activation at these later time points as shown in Fig. 2A.
Role of L. monocytogenes Internalization in Regulating cPLA 2 ␣ Activation-The increased cPLA 2 ␣ activation by WTLM could in part be due to its ability to escape into the cytosol and proliferate. Alternatively, LLO could play an extracellular role by inducing signals that contribute to cPLA 2 ␣ activation. This latter process would presumably not be dependent on WTLM internalization. Thus, experiments were carried out to determine whether internalization of WTLM and ⌬hlyLM is required for stimulation of arachidonic acid release. It has been reported that inhibiting phosphoinositide 3-kinase (PI 3-kinase) with wortmannin blocks internalization of L. monocytogenes into epithelial cells (24). As shown in Fig. 3A, wortmannin also inhibited internalization of WTLM and ⌬hlyLM by resident peritoneal macrophages. This inhibition was similar for WTLM and ⌬hlyLM (68 and 71%, respectively) by 60 min after infection. In control macrophages, not treated with wortmannin, there was greater uptake of ⌬hlyLM (2.7 ϫ 10 5 CFU/well) than WTLM (1.1 ϫ 10 5 CFU/ well). J774 macrophages have been reported to internalize ⌬hlyLM to a greater extent than WTLM (25). Wortmannin inhibited internalization to a greater extent than arachidonic acid release in response to WTLM suggesting that arachidonic acid release is in part stimulated extracellularly.
Because inhibition of PI 3-kinase could affect signaling pathways that regulate cPLA 2 ␣ activation, the effect of latrunculin A, which depolymerizes actin, on internalization of L. monocytogenes and arachidonic acid release was also determined (Fig.  3B). Latrunculin A at 0.1 and 1 M inhibited internalization of WTLM (47 and 87%, respectively) and ⌬hlyLM (59 and 85%, respectively) to a similar extent. As observed for experiments in Fig. 3A, there was greater uptake of ⌬hlyLM (2.9 ϫ 10 5 CFU/ well) than WTLM (1.3 ϫ 10 5 CFU/well) by control macrophages not treated with latrunculin A. There was a direct correlation between blocking internalization of ⌬hlyLM with latrunculin A and inhibition of arachidonic acid release, indi-  ). B, macrophages were incubated for 30 min with and without 5 M pyrrolidine-2 prior to adding WTLM. C, macrophages isolated from cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ mice (C57BL/6/129 mixed strain) were used. Similar results were obtained with macrophages isolated from cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ BALB/c mice. The amount of [ 3 H]arachidonic acid released into the medium from infected and uninfected (Uninf.) macrophages was measured and expressed as a percentage of the total incorporated radioactivity. The results are the averages of two independent experiments in triplicate Ϯ S.D. FEBRUARY 22, 2008 • VOLUME 283 • NUMBER 8 cating that internalization of ⌬hlyLM is required for cPLA 2 ␣ activation. However, there was significantly less inhibition of arachidonic acid release by latrunculin A (0.1 and 1.0 M) from macrophages infected with WTLM (22 and 62%, respectively) than ⌬hlyLM (76 and 87%, respectively). Therefore, cPLA 2 ␣ activation by ⌬hlyLM is dependent on internalization, whereas internalization-dependent and -independent pathways are involved in cPLA 2 ␣ activation by WTLM. The results suggest that LLO exerts an extracellular effect on macrophages leading to greater stimulation of cPLA 2 ␣.

cPLA 2 ␣ Activation and Eicosanoid Production in Macrophages
Activation of MAPKs by L. monocytogenes Regulates Arachidonic Acid Release-Activation of MAPKs regulates cPLA 2 ␣mediated arachidonic acid release from resident peritoneal macrophages in response to a variety of agonists (26). The ability of L. monocytogenes to activate MAPKs in resident peritoneal macrophages was investigated by Western blotting using phospho-specific antibodies to the activated forms of p38 and p42/p44 ERKs (Fig. 4A). WTLM and ⌬hlyLM activated p38 and ERKs to a similar extent, which occurs between 20 and 60 min and then diminishes by 120 min. The MEK1 inhibitor U0126 and the p38 inhibitor SB202190 significantly blocked arachidonic acid release implicating a role for these MAPKs in regulating cPLA 2 ␣ activation in response to L. monocytogenes (Fig. 4B). In control experiments, the MAPK inhibitors did not block internalization of L. monocytogenes (data not shown). cPLA 2 ␣ is phosphorylated by p38 and ERKs on Ser-505, which causes a decrease in electrophoretic mobility on SDSpolyacrylamide gels (27,28). As shown in Fig. 4, WTLM induced a cPLA 2 ␣ gel shift indicating phosphorylation on Ser-505. Preincubation of macrophages with either the MEK1 or p38 inhibitors prior to infection with L. monocytogenes had no effect on Ser-505 phosphorylation. Treatment of macrophages with both U0126 and SB202190 partially blocked the gel shift, suggesting that ERKs and p38 contribute to the phosphorylation of cPLA 2 ␣ in response to L. monocytogenes. However, the MAPK inhibitors significantly blocked arachidonic acid release under conditions that did not detectably inhibit cPLA 2 ␣ Ser-505 phosphorylation, suggesting that MAPKs have additional roles in regulating arachidonic acid release induced by L. monocytogenes.
Role of Calcium in Regulating cPLA 2 ␣ Activation-cPLA 2 ␣ is regulated by increases in concentrations of intracellular calcium ([Ca 2ϩ ] i ) that binds to the C2 domain and promotes its translocation from cytosol to membrane where it hydrolyzes arachidonic acid from phospholipid (29). To determine the role of extracellular calcium in regulating arachidonic acid release and internalization of L. monocytogenes, macrophages were incubated in medium containing EGTA. Chelating extracellular calcium significantly blocked both arachidonic acid release stimulated by WTLM and ⌬hlyLM (80 and 83%, respectively) and internalization (53 and 30%, respectively) (Fig. 5A). In control macrophages not treated with EGTA, the numbers of intracellular ⌬hlyLM (3.9 ϫ 10 5 CFU/well) were greater than WTLM (1.4 ϫ 10 5 CFU/well). Although chelating extracellular calcium with EGTA tended to block internalization of WTLM to a greater extent than ⌬hlyLM, this did not reach statistical significance. The results suggest that calcium has a dual role and regulates cPLA 2 ␣ activation and internalization of L. monocytogenes. The ability of WTLM and ⌬hlyLM to induce increases in [Ca 2ϩ ] i was compared (Fig. 5B). Live cell imaging of calcium mobilization over We previously reported that cPLA 2 ␣ translocates to the phagosome during internalization of zymosan by resident peritoneal macrophages (23). To determine whether this occurs during phagocytosis of L. monocytogenes, the subcellular localization of ECFP-cPLA 2 ␣ in macrophages infected with WTLM or ⌬hlyLM was monitored (Fig. 5C). In uninfected macrophages, ECFP-cPLA 2 ␣ exhibited diffuse cytoplasmic localization (Fig. 5C, panel A). Infection of macrophages with WTLM and ⌬hlyLM caused considerable cell spreading (Fig.   5C, panels B and C). In macrophages incubated with EGFPtagged WTLM, ECFP-cPLA 2 ␣ localized to a juxtanuclear region (Fig. 5C, panel B) indicative of Golgi, which was confirmed by showing co-localization with the Golgi marker Giantin (Fig. 5C, panels D-F). Localization of ECFP-cPLA 2 ␣ to membrane ruffles also occurred in infected macrophages. A similar pattern of ECFP-cPLA 2 ␣ localization was observed in macrophages infected with ⌬hlyLM (Fig. 5C, panel C). There was no evidence of ECFP-cPLA 2 ␣ fluorescence surrounding either WTLM or ⌬hlyLM, indicating that cPLA 2 ␣ does not translocate to the phagosome.
Arachidonic Acid Release Stimulated by L. monocytogenes Involves TLR2 and MyD88-MyD88-deficient mice are highly susceptible to L. monocytogenes infection; however, contradictory results have been reported for the involvement of TLR2 in controlling infection (30 -32). The role of TLR2 and MyD88 in regulating cPLA 2 ␣ activation was investigated by comparing the ability of L. monocytogenes to induce arachidonic acid   FEBRUARY 22, 2008 • VOLUME 283 • NUMBER 8 release from resident peritoneal macrophages isolated from wild type, TLR2, or MyD88 knock-out mice. Arachidonic acid release in response to WTLM and ⌬hlyLM was attenuated in TLR2 Ϫ/Ϫ macrophages by 55 and 90%, respectively (Fig. 6A).

cPLA 2 ␣ Activation and Eicosanoid Production in Macrophages
The extent of L. monocytogenes internalization did not differ in TLR2 ϩ/ϩ and TLR2 Ϫ/Ϫ macrophages (data not shown). As shown in Fig. 6B, arachidonic acid release was attenuated to a similar extent in MyD88 Ϫ/Ϫ macrophages as TLR2 Ϫ/Ϫ macrophages indicating that TLR2 is the principal MyD88-dependent receptor involved in regulating cPLA 2 ␣ activation in response to L. monocytogenes. The attenuated release of arachidonic acid from TLR2 Ϫ/Ϫ macrophages correlated with less activation of p38 and p42/p44 ERKs in response to WTLM and ⌬hlyLM (Fig.  7A). The defect in the activation of these MAPKs in TLR2 Ϫ/Ϫ macrophages was evident from 20 to 120 min after addition of L. monocytogenes (Fig. 7B).
COX2 Expression and Eicosanoid Production Induced by L. monocytogenes-WTLM and ⌬hlyLM induced the expression of COX2 to a similar extent with maximal up-regulation occurring 3 h after infection (m.o.i. 25) (Fig. 8A). Expression of COX2 induced by WTLM and ⌬hlyLM did not involve TLR2 because   similar levels of expression occurred in TLR2 ϩ/ϩ and TLR2 Ϫ/Ϫ macrophages (Fig. 8B). Levels of COX1 were unaffected in macrophages infected with L. monocytogenes and were similar in TLR2 ϩ/ϩ and TLR2 Ϫ/Ϫ macrophages.
L. monocytogenes stimulated the production of PGE 2 , PGI 2 , and LTC 4 by cPLA 2 ␣ ϩ/ϩ macrophages, but eicosanoid production was almost completely ablated in cPLA 2 ␣ Ϫ/Ϫ macrophages (Fig. 9A). The level of prostanoids produced by cPLA 2 ␣ ϩ/ϩ macrophages 3 h after infection was lower in macrophages infected with ⌬hlyLM than WTLM, although this did not reach statistical significance. However, WTLM stimulated significantly more LTC 4 production than ⌬hlyLM consistent with the ability of WTLM to increase [Ca 2ϩ ] i , which regulates 5-lipoxygenase activation. Thus some of the increased arachidonic acid released by WTLM is diverted through the 5-lipoxygenase pathway.
Eicosanoids have been reported to differentially affect bacterial internalization. PGE 2 suppresses phagocytosis of L. monocytogenes but leukotrienes enhance phagocytosis of Klebsiella pneumoniae by macrophages (22,33). cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ peritoneal macrophages were compared to determine whether the decreased production of diverse lipid mediators by cPLA 2 ␣ Ϫ/Ϫ macrophages affects internalization of WTLM. For these experiments, CFUs were determined after incubation of macrophages with WTLM for 60 min followed by extensive washing to remove extracellular bacteria, rather than using gentamicin, to compare our results with the study of Hutchison and Myers (22). To ensure that the washing procedure sufficiently removed extracellular WTLM, macrophages were treated with latrunculin A, which blocks WTLM internalization (Fig.  9B). Latrunculin A blocked WTLM uptake by 90% indicating that the CFU represent internalized bacteria and not extracellular bacteria bound to the cell surface. Using the washing technique, the number of WTLM internalized by cPLA 2 ␣ ϩ/ϩ macrophages (1.3 ϫ 10 6 CFU/well) was higher than obtained following gentamicin treatment. Treating cell cultures with gentamicin is a commonly used protocol to kill extracellular L. monocytogenes and is essential for preventing growth of extracellular bacteria during long term incubations. However, it has been reported that gentamicin can gain access to the phagosomal compartment and contribute to killing  . cPLA 2 ␣ is required for eicosanoid production by macrophages infected with L. monocytogenes. A, resident peritoneal macrophages isolated from cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ mice (BALB/c) were incubated with WTLM or ⌬hlyLM (m.o.i. 25) for 60 min, washed, and then incubated in fresh medium containing gentamicin. Levels of PGE 2 , 6-keto prostaglandin F 1 ␣ (6-keto PGF 1 ␣, the stable metabolite of PGI 2 ), and leukotriene C 4 (LTC 4 ) were measured in the culture medium collected 3 h after adding L. monocytogenes. The results are the averages Ϯ S.E. of five independent experiments for cPLA 2 ␣ ϩ/ϩ macrophages and averages Ϯ S.D. of two experiments for cPLA 2 ␣ Ϫ/Ϫ macrophages. There was significantly (*, p Ͻ 0.05) less LTC 4 production by cPLA 2 ␣ ϩ/ϩ macrophages infected with ⌬hlyLM than WTLM. B, resident peritoneal macrophages isolated from cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ mice were incubated with or without (none) PGE 2 (4 ng/ml), leukotriene C 4 (10 ng/ml), or latrunculin A (1 M) for 15 before addition of WTLM (m.o.i. 25). After incubation for 1 h, the macrophages were washed six times and lysed with 0.1% Triton X-100 for determination of CFU. The results are the average Ϯ S.E. of at least three independent experiments. There was significantly (*, p Ͻ 0.05) less CFU in untreated cPLA 2 ␣ Ϫ/Ϫ macrophages than cPLA 2 ␣ ϩ/ϩ macrophages and in PGE 2 -or LTC 4 -treated cPLA 2 ␣ Ϫ/Ϫ macrophages than similarly treated cPLA 2 ␣ ϩ/ϩ macrophages. There was significantly (**, p Ͻ 0.05) less CFU in cPLA 2 ␣ ϩ/ϩ macrophages treated with PGE 2 and more CFU in cPLA 2 ␣ ϩ/ϩ macrophages treated with LTC 4 compared with untreated cPLA 2 ␣ ϩ/ϩ macrophages (none). There was significantly (#, p Ͻ 0.05) less CFU in cPLA 2 ␣ Ϫ/Ϫ macrophages treated with PGE 2 than in untreated cPLA 2 ␣ Ϫ/Ϫ macrophages. Uninf, uninfected. FEBRUARY 22, 2008 • VOLUME 283 • NUMBER 8

JOURNAL OF BIOLOGICAL CHEMISTRY 4751
of intracellular L. monocytogenes, which may explain the higher CFU obtained with the washing protocol (34). A comparison of the number of internalized WTLM in cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ macrophages after incubation with bacteria for 60 min revealed that cPLA 2 ␣ Ϫ/Ϫ macrophages internalized significantly fewer WTLM (25% less CFU) than cPLA 2 ␣ ϩ/ϩ macrophages (Fig. 9B). The direct addition of PGE 2 , at a concentration in the range produced by cPLA 2 ␣ ϩ/ϩ macrophages, significantly blocked internalization of WTLM by cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ macrophages. In contrast, LTC 4 addition induced a small (15%) but significant increase in WTLM uptake by cPLA 2 ␣ ϩ/ϩ macrophages but not by cPLA 2 ␣ Ϫ/Ϫ macrophages. Thus the decreased uptake of WTLM by cPLA 2 ␣ Ϫ/Ϫ macrophages is not solely explained by the lack of LTC 4 production. cPLA 2 ␣-mediated Prostaglandin Production Down-regulates WTLM-stimulated TNF␣ Production-We investigated the contribution of COX1 and COX2 in mediating prostaglandin production in macrophages infected with WTLM. As shown in Fig. 10A, the COX2 inhibitor NS398 blocked PGE 2 and PGI 2 production by 72 and 55%, respectively, whereas the nonselective COX inhibitor indomethacin blocked prostaglandin production by over 90%. The results suggest a role for both COX2 and COX1 in prostaglandin production in response to L. monocytogenes.

DISCUSSION
The ability of L. monocytogenes to disseminate and evade host responses is in part because of its ability to survive and multiply within a variety of cell types, including tissue macrophages (2,44,45). The results of this study demonstrate that L. monocytogenes stimulates cPLA 2 ␣ activation and eicosanoid production in resident peritoneal macrophages and occurs in the absence of serum with un-opsonized bacteria. WTLM and ⌬hlyLM stimulate similar levels of arachidonic acid release at early times (20 min), but at later times (40 -60 min) WTLM stimulates to a greater extent than ⌬hlyLM. The results demonstrate that LLO augments cPLA 2 ␣ activation but is not essential. In contrast, the activation of host phospholipases C and D in J774 macrophages by WTLM is absolutely dependent on LLO (46). The increased arachidonic acid release by WTLM could in part be due to LLO-dependent escape from the primary phagosome and proliferation in the cytosol, which occurs within the time frame observed for enhancement of arachidonic acid release by WTLM. However, if LLO-dependent escape of WTLM from the primary vacuole or cell-to-cell spread contributed to cPLA 2 ␣ activation, we would have expected the double PLC mutant to be less effective at inducing arachidonic acid release because PLCs contribute to escape and spread (8 -11). However, the double L. monocytogenes mutant lacking PI-PLC and BR-PLC stimulated arachidonic acid release as effectively as WTLM. Experiments also demonstrated that L. monocytogenes mutants lacking the virulence factors p60, an autolysin that hydrolyzes peptidoglycan, and ActA, a protein important for actin-based L. monocytogenes motility, stimulated arachidonic acid release as effectively as WTLM (data not shown) (47)(48)(49).
Results using latrunculin A, which blocks internalization of WTLM to a greater extent than arachidonic acid release, suggest that LLO may act extracellularly to promote cPLA 2 ␣ activation. In contrast, arachidonic acid release induced by ⌬hlyLM is more dependent on internalization than for WTLM. The delay in the enhancement of arachidonic acid release by WTLM may be due to the time-dependent production and secretion of extracellular LLO to levels needed for augmenting arachidonic acid release. LLO may play a signaling role at the plasma membrane in addition to its role in promoting escape of L. monocytogenes from the primary phagosome (3). Recent studies have demonstrated that LLO exhibits cholesterol-dependent poreforming ability at physiological pH and may induce cell signaling by aggregating lipid rafts (50,51). It has been shown that L. monocytogenes-stimulated leukotriene production by human neutrophils is completely dependent on LLO. It occurs inde- FIGURE 10. Regulation of TNF␣ production by prostaglandins. A, resident peritoneal macrophages isolated from cPLA 2 ␣ ϩ/ϩ mice were incubated for 30 min with or without 20 M indomethacin or 10 M NS398 and then incubated with WTLM for 3 h as described above for Fig. 9A. Control macrophages not treated with inhibitors produced 9.9 ng/ml PGE 2 and 9.1 ng/ml 6-keto PGF 1 ␣. B, cPLA 2 ␣ ϩ/ϩ macrophages, incubated with or without 20 M indomethacin, and cPLA 2 ␣ Ϫ/Ϫ macrophages incubated without indomethacin, were treated as described in Fig. 9A. The culture media were collected 3 h after adding WTLM or ⌬hlyLM for TNF␣ analysis. Results are the averages of two independent experiments Ϯ S.D. (A) or the averages of three independent experiments Ϯ S.E. (B). cPLA 2 ␣ Ϫ/Ϫ macrophages and indomethacintreated cPLA 2 ␣ ϩ/ϩ macrophages produced significantly more (p Ͻ 0.05) TNF␣ production than untreated cPLA 2 ␣ ϩ/ϩ macrophages.
pendently of phagocytosis and by purified LLO (52). In macrophages, we find that both LLO-dependent and -independent mechanisms contribute to cPLA 2 ␣ activation. Heat-killed L. monocytogenes also stimulated arachidonic acid release (data not shown), although more weakly than ⌬hlyLM, suggesting that a cell wall component of L. monocytogenes contributes to cPLA 2 ␣ activation.
To mediate arachidonic acid release, cPLA 2 ␣ must translocate from the cytosol to the membrane to access substrate. This is an important regulatory step induced by elevations in [Ca 2ϩ ] i that binds to the cPLA 2 ␣ C2 domain increasing its affinity for membrane (29,53,54). However, in response to certain stimuli cPLA 2 ␣-mediated arachidonic acid release from macrophages occurs with no detectable increase in [Ca 2ϩ ] i , although resting levels of calcium are required (55)(56)(57)(58). Previous studies have shown that WTLM stimulates [Ca 2ϩ ] i increases in endothelial cells, epithelial cells, and J774 macrophages to a greater extent than ⌬hlyLM, and that LLO forms calcium-permeable pores in cells (25, 59 -62). Similarly, we found in resident peritoneal macrophages that WTLM but not ⌬hlyLM increases [Ca 2ϩ ] i , which may contribute to the ability of WTLM to stimulate greater arachidonic acid release. However, arachidonic acid release stimulated by ⌬hlyLM is blocked by chelating extracellular calcium. This suggests a role for resting [Ca 2ϩ ] i , which may be lowered by extracellular EGTA. It is also possible that ⌬hlyLM stimulates local, transient increases in [Ca 2ϩ ] i that we were not able to detect. During zymosan phagocytosis, cPLA 2 ␣ translocates to the forming phagosome, membrane ruffles, and Golgi (23). Similarly, cPLA 2 ␣ translocates to Golgi and membrane ruffles in macrophages infected with L. monocytogenes. However, we found no evidence that cPLA 2 ␣ translocates to the L. monocytogenes phagosome suggesting differences in the properties of the phagosome membrane.
Activation of cPLA 2 ␣ in resident peritoneal macrophages is partially dependent on engagement of TLR2 by L. monocytogenes. Arachidonic acid release is more dependent on TLR2 in response to ⌬hlyLM than to WTLM, consistent with a TLR2independent role for LLO in cPLA 2 ␣ activation in response to WTLM. Resident peritoneal macrophages lacking MyD88 exhibited a similar decrease in arachidonic acid release as TLR2 Ϫ/Ϫ macrophages, in response to WTLM and ⌬hlyLM. This suggests that TLR2 engagement is the principal MyD88dependent pathway involved in regulating cPLA 2 ␣ activation in response to L. monocytogenes. Our results demonstrate that TLR2 is required for optimal activation of ERKs and p38, which regulate activation of cPLA 2 ␣-mediated arachidonic acid release in response to L. monocytogenes. In contrast to reports using nonphagocytic cells, MAPKs are activated to a similar extent by WTLM and ⌬hlyLM in peritoneal macrophages, and are not required for L. monocytogenes internalization (63)(64)(65). MAPKs (ERKs and p38) phosphorylate cPLA 2 ␣ on Ser-505, which enhances activity of cPLA 2 ␣ and promotes arachidonic acid release (27,28,66). L. monocytogenes induces the stoichiometric phosphorylation of cPLA 2 ␣ on Ser-505 as evidenced by the complete shift in electrophoretic mobility on SDS-polyacrylamide gels. However, inhibition of either ERKs or p38 blocked L. monocytogenes-induced arachidonic acid release under conditions that did not prevent phosphorylation of cPLA 2 ␣ on Ser-505. These results suggest an additional role independent of Ser-505 phosphorylation for MAPKs in regulating cPLA 2 ␣ in response to L. monocytogenes, as observed in other cell models (26,67). Targeted disruption of MAPK-activated protein kinase, a substrate for p38 and ERKs, increases susceptibility of mice to L. monocytogenes infection (68). Our data suggest another role for MAPKs, promoting cPLA 2 ␣ activation, in regulating immune responses to L. monocytogenes.
We previously reported that engagement of TLR2 is not sufficient for cPLA 2 ␣ activation. The TLR2 ligand MALP2 poorly induces arachidonic acid release but acts synergistically with the dectin-1 agonist particulate ␤-glucan (69). Therefore it is likely that engagement of an unknown receptor by L. monocytogenes, on resident peritoneal macrophages, mediates internalization and induces signals for cPLA 2 ␣ activation cooperatively with TLR2. We found that internalization of WTLM and ⌬hlyLM is blocked by inhibition of PI 3-kinase and actin polymerization, and to a lesser extent by chelating extracellular calcium. Uptake of L. monocytogenes into nonphagocytic cells is mediated in part by internalin A and internalin B, which interact with E-cadherin and Met receptors on host cells, respectively (70,71). Internalin B binding to the tyrosine kinase receptor Met mediates internalization of L. monocytogenes through activation of PI 3-kinase, MAPK, and actin polymerization (3,24,65,70). Internalin B also exhibits divalent-cation dependent binding to C1q-R, the receptor for the complement component C1q that is expressed on many cells, including peritoneal macrophages (72,73). In J774 macrophages, internalin B mediates activation of Ras and PI 3-kinase, and internalin A mediates phagocytosis of L. monocytogenes (74,75). However, we found that mutant L. monocytogenes lacking both internalin A and internalin B stimulates arachidonic acid release in resident peritoneal macrophages as effectively as WTLM (data not shown) suggesting that the internalins are not involved in mediating cPLA 2 ␣ activation.
L. monocytogenes induces transcriptional responses in macrophages by interaction of cell wall components with pattern recognition receptors such as TLRs, and by MyD88-independent mechanisms that are triggered by escape into the cytosol (76). The cytosolic specific transcriptional response induced by WTLM but not ⌬hlyLM, in bone marrow-derived macrophages, involves activation of p38 (77). In contrast, our results demonstrate that WTLM and ⌬hlyLM both stimulate the TLR2dependent early activation of ERKs and p38 in resident peritoneal macrophages. Activation of MAPKs is not completely ablated in TLR2-or MyD88-deficient resident peritoneal macrophages implicating a role for another signaling pathway independent of cytosolic sensing. In addition, ⌬hlyLM induces similar levels of COX2 expression as WTLM by a TLR2-independent mechanism. The results demonstrate that the signals induced by L. monocytogenes for eicosanoid production, including acute activation of cPLA 2 ␣ and up-regulation of COX2, are not dependent on cytosolic sensing and involve TLR2-dependent and -independent pathways. The data implicate the presence of another signaling pathway triggered by interaction of L. monocytogenes with a receptor on macrophages that remains to be identified.
The production of eicosanoids by resident peritoneal macrophages infected with L. monocytogenes was dependent on cPLA 2 ␣. Therefore a comparison of macrophages from cPLA 2 ␣ knock-out and wild type mice provided a model system to determine whether production of eicosanoids influences macrophage responses to L. monocytogenes infection. WTLM induced a similar degree of cytotoxicity in cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ macrophages 24 h after infection despite incubation in medium containing gentamicin (data not shown). This suggests that cPLA 2 ␣ activation does not influence escape, intracellular growth, and killing of resident peritoneal macrophages by WTLM.
A comparison of the ability of cPLA 2 ␣ ϩ/ϩ and cPLA 2 ␣ Ϫ/Ϫ macrophages to internalize WTLM revealed a small (25%) but significant decrease in uptake of WTLM by cPLA 2 ␣ Ϫ/Ϫ . We found opposing effects of adding PGE 2 and LTC 4 on internalization of WTLM. PGE 2 blocked uptake of WTLM consistent with a previous report showing a decrease in phagocytosis of WTLM by peritoneal macrophages as determined by microscopic evaluation (22). The lack of PGE 2 production by cPLA 2 ␣ Ϫ/Ϫ would be expected to result in increased WTLM uptake, which was not observed. Addition of LTC 4 to cPLA 2 ␣ ϩ/ϩ macrophages slightly increased phagocytosis of WTLM, a trend similar to that observed for phagocytosis of K. pneumoniae by rat alveolar macrophages (33). However, LTC 4 addition did not reverse the defective ability of cPLA 2 ␣ Ϫ/Ϫ peritoneal macrophages to internalize WTLM. cPLA 2 ␣ initiates the production of numerous COX and lipoxygenase-derived lipid mediators that can differentially effect cellular responses depending on the profile of eicosanoid receptor expression. Therefore, it is unlikely that the decrease in phagocytosis of WTLM by cPLA 2 ␣ Ϫ/Ϫ macrophages can be ascribed to the loss of one particular lipid mediator.
We found that cPLA 2 ␣ activation influences TNF␣ production, a cytokine essential for combating L. monocytogenes infection (78 -80). cPLA 2 ␣ ϩ/ϩ macrophages produced significantly less TNF␣ than cPLA 2 ␣ Ϫ/Ϫ macrophages. The ability of indomethacin to enhance production of TNF␣ in cPLA 2 ␣ ϩ/ϩ macrophages to the level produced by cPLA 2 ␣ Ϫ/Ϫ macrophages in response to L. monocytogenes suggests that prostaglandins act in an autocrine fashion to suppress TNF␣ production. It has been reported that TNF␣ production by human and mouse macrophages, including Kupffer cells, is suppressed by exogenous addition of prostaglandins and enhanced by blocking endogenous prostaglandin production with cyclooxygenase inhibitors (35)(36)(37)(38)(39)(40)(41)(42)(43). We show that this regulatory program is initiated by the activation of cPLA 2 ␣ in response to L. monocytogenes infection. The receptors for PGE 2 (EP2 and EP4) and PGI 2 (IP) mediate increases in cAMP, which down-regulates TNF␣ production (35,42,81). It has been shown that treatment with indomethacin increases susceptibility of mice infected intraperitoneally with L. monocytogenes suggesting that prostaglandins or thromboxane A 2 has a protective role (82). In contrast, administering PGE 1 suppresses cellular immunity in mice infected intravenously with L. monocytogenes (83). It is difficult to reconcile the role of COX-derived metabolites from these studies because of the differences in route of infection and use of indomethacin, which blocks production of all prostanoids and thromboxane A 2 , versus using a single prostaglandin. However, they do point to a role for lipid mediators in regulat-ing immune response to L. monocytogenes. Prostaglandins may have multiple effects in regulating immunity. In addition to affecting TNF␣ levels, prostaglandins suppress production of a variety of chemokines. They also up-regulate IL-10 and IL-6 production (36, 84, 85) and will have differential effects that depend on the types of prostaglandin receptors expressed on target cells (86). In addition to regulating cytokine production, prostaglandins promote increases in vascular permeability and blood flow, which regulate the migration of effector cells from the blood to the site of infection (86). Our results demonstrate that regulation of cytokine production by prostaglandins occurs as a result of cPLA 2 ␣ activation by L. monocytogenes, suggesting that its early activation in tissue macrophages may have an impact on immune responses to microbial infection.