A New Immunomodulatory Role for Peroxisomes in Macrophages Activated by the TLR4 Ligand Lipopolysaccharide

Peroxisomes are eukaryotic membrane-bound subcellular organelles that compartmentalize several proteins or pathways for the metabolism of lipids and reactive oxygen species (ROS). A variety of bioactive lipid derivatives that cannot be degraded in the mitochondria is metabolized in peroxisomes (1). Peroxisomes are involved in cytoprotection by mediating intracellular lipid and ROS homeostasis (2) and, when compromised, are coupled with mitochondrial dysfunction (3, 4). The peroxisomal compartment and its metabolic capacity were suggested to be altered by bacterial compound endotoxin in the liver (5–7). However, the role of peroxisomes in immune homeostasis has only recently been begun to be understood. Mitochondrial antiviral signaling protein, an adaptor protein primarily localized on the mitochondrial membrane, is also localized on peroxisomes, and this phenomenon is crucial for mounting an effective antiviral response (8). Follow-up studies identified peroxisomes as an important subcellular signaling platform for the production of IFN-λ during viral or intracellular bacterial infection (9). Similarly, peroxisome-derived ether lipids were shown to be vital for the maturation of invariant NK T cells in the thymus (10). Additionally, our recent findings suggest that peroxisomal alterations in idiopathic pulmonary fibrosis lead to impaired TGF-β signaling and that the downregulation of peroxisomes in primary human lung fibroblasts causes a profibrotic phenotype in these cells (11).

The notion that peroxisomes are involved in the regulation of inflammation arises from three important functions attributed to the organelle. Peroxisomes are equipped with a series of antioxidant enzymes, such as catalase and peroxiredoxins, whose primary function is the neutralization of ROS generated during β-oxidation of lipids, but they may also play a role in maintaining cellular redox homeostasis (12). Indeed, various independent studies confirmed that deficiency of peroxisomes leads to increased oxidative stress (13). Peroxisomes were shown in vitro to metabolize leukotrienes (14) and PGs (15), which are crucial modulators of inflammation, and peroxisomes are the primary intracellular site for the production of docasohexaenoic acid (DHA) and eicosopentaenoic acid, which form the backbone for a series of resolution mediators of inflammation that includes resolvins, protectins, and maresins (12).

Despite their potential immunomodulatory properties, little is known about the specific role of peroxisomes in inflammatory responses in myeloid immune cells, such as monocytes, macrophages (MΦs), and dendritic cells. We chose to address this issue in MΦs, because they are unique in their capacity to initiate and resolve inflammation. MΦs are broadly divided into two main subclasses: M1 (classically activated) and M2 (alternatively activated). M1 MΦs exhibit a proinflammatory phenotype, whereas M2 MΦs exhibit an anti-inflammatory phenotype (16). In this article, we report that induction of peroxisomes in MΦs dampens the LPS-induced proinflammatory proteins cyclooxygenase (COX)-2, TNF-α, IL-6, and IL-12 without exhibiting the classical features of M2 MΦs. In contrast, disruption of peroxisomal function leads to a hyperinduction of COX-2 and TNF-α. This peroxisome-driven anti-inflammatory effect was found to be mediated via extrinsic factors released by MΦs upon LPS activation.

DMEM, RPMI 1640, and inactivated FBS were purchased from PAA Laboratories (Pasching, Austria), and polyvinylidene difluoride membranes were from Millipore (Darmstadt, Germany). LPS ultrapure (0111:B4), lipoteichoic acid (LTA), and polyinosinic:polycytidylic acid (poly I:C) were purchased from InvivoGen (San Diego, CA). PG (PGJ₂, PGD₂, PGE₂), leukotriene B4 (LTB₄), arachidonic acid (AA), DHA, and deuterated internal standards LTB₄-d4 and AA-d11 were purchased from Cayman Chemical (distributor: Biomol, Hamburg, Germany). Strata-X polymeric SPE sorbents (33 μm, 100 mg/3 ml tubes) were purchased from Phenomenex (Torrance, CA). All other chemicals were obtained from Sigma-Aldrich (St. Louis, MO) and Roche Applied Science (Penzberg, Germany), unless otherwise indicated.

Fifteen adult male mice (6–12 wk of age) on the C57BL/6J genetic background were obtained from Charles River (Sulzfeld, Germany). The mice were kept on a normal laboratory diet and water ad libitum and were housed in cages under standardized environmental conditions (12-h light/dark cycle, 23 ± 1°C and 55 ± 1% relative humidity) at the central animal facility of the Justus Liebig University Giessen. All experiments using laboratory mice were approved by the governmental ethics committee for animal welfare (Regierungspräsidium Giessen, Germany; permit number V 54-19 C 20/15 c GI 20/23).

RAW264.7 (RAW) cells were from the American Type Culture Collection, and primary mouse embryonic fibroblasts (MEFs) (17) were grown in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin, as described previously (18). Murine alveolar MΦ (AMs), peritoneal MΦs (PMs), and bone marrow–derived MΦs (BMDMs) were isolated as described previously (19). AMs were cultured and maintained in RPMI 1640 supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. Peritoneal and bone marrow–derived macrophages were cultured and maintained in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. Conditioned medium obtained from culture of L-929 fibroblasts in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin was used as a source of macrophage colony stimulating factors to differentiate bone marrow–derived cells to MΦs. Peripheral blood–derived human MΦs were obtained by Ficoll isolation of PBMCs from 50 ml of blood. Briefly, isolated PBMCs were plated onto bacterial petri dishes in RPMI 1640 + 5% AB serum (C.C.pro, Oberdola, Germany). The medium was replaced after 2 h to remove all nonadherent cells and was cultured for 5 d to differentiate into MΦs. The following chemicals were used at the indicated concentrations: antimycin A (5–10 μM), chloroquine (75 μM), cycloheximide (CHX; 100 μM), epoxomicin (1 μM), hydroxy urea (HU; 100 μM), l-sulforaphane (10 μM), 3-methyladenine (5 mM), MG132 (10 μM), NS-398 (20 μM), and S-flurbiprofen (100 μM).

All cell cultures were kept under air/CO₂ (19:1) at 100% humidity.

Cell culture supernatants were briefly centrifuged and used fresh or stored at −80°C until use. ELISA kits for TNF-α, IL-6, IL-10, and IL-12 were purchased from eBioscience (San Diego, CA) and used according to the manufacturer’s instructions.

Briefly, 1 × 10⁶ cells were blocked with Fc blocker for 15 min on ice and stained with 0.25 μg of anti-mouse CD86 or anti-mouse CD36 or the respective isotype controls (BioLegend, San Diego, CA) for 20 min before analysis by flow cytometry. For intracellular staining of PEX14p, the cells were fixed for 10 min in 2% paraformaldehyde, followed by permeabilization for 5 min using 0.2% Triton X-100. Cells were incubated for 30 min at 4°C with Ab against PEX14p (1:1200) in PBS containing 2% FCS+ 0.2% sodium azide. The cells were washed twice and incubated with donkey anti-rabbit IgG–Alexa Fluor 488 secondary Ab (Invitrogen, Carlsbad, CA) for 20 min, washed twice with 1× PBS, and analyzed using a BD FACSCanto II flow cytometer.

Cells were plated on coverslips in 12- or 24-well plates and subjected to an indirect immunofluorescence staining protocol, as previously described (18). Briefly, cells grown on coverslips were washed three times with PBS, fixed using 4% paraformaldehyde and 2% saccharose, and permeabilized using 1% glycine containing 0.2% Triton X-100. Nonspecific binding sites were blocked with 1% BSA in TBS containing 0.05% Tween for 1 h at room temperature (RT), after which the coverslips were incubated overnight at 4°C with the primary Abs against PEX14p (1:2000; a generous gift from D. Crane), total p65 (1:400), phospho–c-jun (1:600), and COX-2 (1:1000) (Cell Signaling Technology, Danvers, MA). Coverslips were washed and incubated with Alexa Fluor 488–conjugated secondary Ab at 1:800 dilutions (Invitrogen). Nuclei were counterstained with 1 μM Hoechst 33,342 for 5 min at RT and the coverslips were embedded in Mowiol 4-88 with N-propyl gallate as an antifading agent. All samples were examined with a Leica DMRS fluorescence microscope equipped with a DR4 camera or a confocal laser scanning microscope (Leica TCS SP5; Leica Mikrosystem Vertrieb, Wetzlar, Germany). All images were processed with Adobe Photoshop CS5 extended version.

Cell supernatants were collected from LPS-stimulated cultures of RAW cells, with or without pretreatment with 4-phenyl butyric acid (4-PBA), and stored at −80°C until further analysis. A total of 8.5 ml of each cell culture supernatant was spiked with deuterated internal standards (6 ng), and lipid species were extracted with a solid phase extraction method using solid phase extraction columns, according to the manufacturer’s protocol. Briefly, the columns were equilibrated with 3.5 ml of 100% methanol, followed by 3.5 ml of water. After sample extraction, the columns were washed with 3.5 ml of 10% methanol, and the lipids were eluted with 1.5 ml of methanol. The methanol was dried down under a stream of nitrogen gas, and the dried extract was reconstituted with 55 μl of 25% acetonitrile. Lipid species were quantified in cell supernatants by parallel reaction monitoring using a quadrupole orbital trapping mass spectrometer (Q Exactive; Thermo Scientific, Bremen, Germany) in high-resolution mode. Briefly, 10 μl of final reconstituted extract was injected into a Dionex UltiMate 3000 UHPLC (Thermo Scientific), and separation was performed using a reverse-phase Kinetex C18 2.6-μm column (100 × 2.1 mm, 100 Å) with a flow rate of 500 μl/min of binary solvent system consisting of mobile phase A (100% water with 0.1% formic acid) and mobile phase B (100% acetonitrile with 0.1% formic acid). The gradient was 27–35% B in 15 min, 35–60% in 2 min, and 60–80% B in 10 min. The mass spectrometer was operated at a mass resolution of 70,000 at m/z 200 with electrospray ionization in negative ion mode. The method used consisted of full-scan mass spectrometry (MS) and targeted tandem MS of the lipid precursor with defined mass and retention time. The transitions of lipid species were selected based on previously published data (20, 21), and fatty acids were quantified from full-scan MS. Standard calibration curves were constructed by plotting the area under the curve of transition of lipids by extracted ion chromatogram against the concentrations of serial dilutions of lipid species ranging from 10 pg to 2.56 ng or in case of fatty acids, 1–256 ng on column.

A total of 1 × 10⁵ cells was plated on coverslips in a 12-well plate. After overnight culture, cells were treated with antimycin A (5–10 μM) for 1 h, followed by staining with MitoTracker (150 nM; Molecular Probes), according to the manufacturer’s instructions. The cells were fixed with 4% paraformaldehyde, and the images were captured using a Leica DMRS fluorescence microscope equipped with a DR4 camera. Digital photographs were processed with Adobe Photoshop version 9.

Total RNA was isolated from cells using an RNeasy Kit (QIAGEN), and cDNA was synthesized by reverse transcription using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). One microgram of total RNA was used in each reverse transcription reaction. PCR analyses were performed with inventoried primer mixes for mouse Arg-1, Cox-2, Il-6, Il-10, iNos, and Tnf-α (Applied Biosystems). Amplification was performed using TaqMan Gene Expression Master Mix on a StepOnePlus Real-Time PCR System (Applied Biosystems), according to the manufacturer’s instructions. Thermal cycling was performed at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. The constitutively expressed gene Hprt was used as a control for normalization of cDNA levels. The ΔΔ threshold cycle method was used to semiquantify mRNA levels according to the manufacturer’s protocol.

A total of 8 × 10⁴ RAW cells was plated in a 12-well plate and cultured overnight. The next day, cells were transfected with 30 nM of control small interfering RNA (siRNA) (Ambion siRNA ID: AM4611), Pex14 siRNA (Ambion siRNA ID: s80240), or Mfp2 siRNA (Ambion siRNA ID: s67851) using ScreenFect A transfection reagent (Incella, Eggenstein-Leopoldshafen, Germany), according to the manufacturer’s instructions. The cells were incubated for an additional 48 h, after which the medium was replaced with medium containing LPS wherever indicated.

Thioflavin T (ThT) staining was performed as previously mentioned (22). Briefly 5 × 10⁴ cells were plated on coverslips in a 24-well plate and were left untreated or treated with 4-PBA (2.5–5 mM) for 72 h. Cells were washed with fresh medium and incubated with medium containing LPS (100 ng/ml) for 12 h. After incubation, the cells were fixed using 4% paraformaldehyde for 20 min, washed twice with PBS, and washed three times for 1 min each with double distilled H₂O. The cells were incubated with 5 μM ThT in double distilled H₂O for 10 min and washed again with double distilled H₂O (three times for 1 min). Coverslips were mounted on slides with Mowiol 4-88 containing N-propyl gallate as an antifading agent. Images were taken using a Leica DMRS fluorescence microscope equipped with a DR4 camera. Digital photographs were processed with Adobe Photoshop version 9.

Western blot analyses were performed as described previously (18). Briefly, cells were washed with PBS, lysed, and the protein concentrations were estimated using Bradford’s assay. Thirty micrograms of total protein was separated on 10% SDS–polyacrylamide gels and subsequently blotted onto polyvinylidene difluoride membranes. Membranes were blocked with TBS containing 5% skim milk, 50 mM Tris HCl (pH 7.6), 150 mM NaCl, and 0.1% Tween 20 for 1 h at RT. The blots were incubated overnight with primary Abs against COX-2 (1:1000; Cell Signaling Technology or Abcam, Cambridge, U.K.), Pex14P (1:14,000), multifunctional protein 2 (Mfp2; 1:1000; Abcam), heme oxygenase-1 (HO-1; 1:1000; Enzo Lifesciences, Loerach, Germany), β-actin (1:5000; Abcam), and GAPDH (1:10,000; Hytest, Turku, Finland). The secondary Ab mouse anti-rabbit IgG alkaline phosphatase or rabbit anti-mouse IgG (Sigma) was used at 1:30,000. Immun-Star alkaline phosphatase substrate (Bio-Rad, Hercules, CA) was used for detection, according to the manufacturer’s instructions. The bands were visualized by exposing the blots to Kodak BioMax MR Films or with a FluorChem (Alpha Innotec, San Leandro, CA) and quantified with ImageJ software.

All values are expressed as mean ± SEM. A Student t test or ANOVA was used to assess the difference between two groups. Differences were considered statistically significant when p < 0.05.

To elucidate the role of peroxisomes in the MΦ inflammatory response, we sought to compare MΦ activation by the prototypical proinflammatory stimulus LPS between control MΦs with normal abundance of peroxisomes and MΦs in which peroxisomes are induced. For this purpose, we used 4-PBA, a nonclassical peroxisome proliferator that was shown by us and other investigators to proliferate peroxisomes in cell cultures of mouse (23) and human (24) fibroblasts and is independent of the peroxisome proliferator activated receptor (PPAR)-α. Immunofluorescence studies revealed that treatment with 4-PBA for 72 h induced peroxisome proliferation, as detected by staining for the peroxisomal biogenesis protein PEX14p [which is an ideal marker for the abundance of peroxisomes (25)] in various MΦ cultures, such as the RAW murine MΦ cell line, murine primary AMs, and murine primary PMs (Fig. 1A–F). Unexpectedly, treatment with 4-PBA in primary BMDMs did not induce peroxisome proliferation (Fig. 1G, 1H). Similarly, Western blot analyses confirmed that PEX14p was induced after 4-PBA treatment in RAW MΦs (Supplemental Fig. 1A) but not in primary BMDMs (Supplemental Fig. 1B).

In summary, treatment with the peroxisome proliferator 4-PBA for 72 h induces peroxisome proliferation in murine primary AMs, primary PMs, and RAW MΦs but not in primary BMDMs.

Next, untreated control and 4-PBA–treated MΦs were stimulated with LPS, and the induction of proinflammatory proteins was analyzed by immunofluorescence, Western blot analysis, and ELISA. LPS stimulation in primary AMs led to a strong increase in the staining for COX-2, a typical LPS-induced proinflammatory protein (18) involved in the production of proinflammatory PGs (Fig. 2A, middle panel). In contrast, AMs pretreated with 4-PBA for 72 h showed a very weak staining for COX-2 after LPS stimulation (Fig. 2A, right panel). Similarly, Western blot analyses confirmed that the LPS-induced upregulation of COX-2 was significantly reduced in 4-PBA–treated-RAW MΦs and PMs (Fig. 2B, 2C, Supplemental Fig. 1C, 1D). Interestingly, treatment of RAW MΦs with 4-PBA for only 2 h prior to LPS stimulation did not block LPS-induced COX-2 (Fig. 2B). Because 4-PBA did not induce peroxisomes in primary BMDMs, these cells were used as controls to test the peroxisome-independent effects of this compound. Accordingly, treatment with 4-PBA did not decrease LPS-induced COX-2 induction in BMDMs (Fig. 2D, Supplemental Fig. 1E). Unexpectedly, treatment with 4-PBA at a higher concentration (5 mM) enhanced LPS-dependent upregulation of COX-2 in BMDMs (Fig. 2D, Supplemental Fig. 1E).

To further substantiate the role of peroxisomes in MΦs, two other peroxisome inducers, HU and L-sulforaphane (L-SF) (26), were applied. Pretreatment with HU or L-SF increased the protein abundance of PEX14p (Supplemental Fig. 2A, 2B) and subsequently blocked LPS-induced upregulation of COX-2 in RAW MΦs (Supplemental Fig. 2C, 2D).

Recently, the global anti-inflammatory effect of 4-PBA in a LPS-induced lung inflammation model was suggested to be mediated via inhibition of endoplasmic reticulum (ER) stress (27). However, the concentration of 4-PBA (2.5 mM) used in our experimental setup did not inhibit LPS-induced ER stress, as indicated by ThT (Supplemental Fig. 2E). In contrast, a higher concentration of 4-PBA (5 mM) inhibited LPS-induced ER stress (Supplemental Fig. 2E). If the anti-inflammatory property of 4-PBA was mediated via inhibition of ER stress, then treatment of BMDM with 4-PBA should not inhibit ER stress. However, ThT staining in BMDMs revealed that LPS stimulation increased the intensity of ThT staining, and it was blocked when the cells were pretreated with 5 mM 4-PBA (Supplemental Fig. 2F).

To investigate a more global role for peroxisomes in the inflammatory response in MΦs, we measured important inflammatory cytokines by ELISA. 4-PBA–treated AMs, PMs, and RAW MΦs secreted significantly lower amounts of the proinflammatory cytokines TNF-α, IL-6, and IL-12 after stimulation by LPS in comparison with the respective controls (Fig. 2E). In contrast, pretreatment of BMDMs with 4-PBA did not reduce, but rather stimulated, the expression of the proinflammatory cytokines secreted after LPS stimulation at higher concentrations of 4-PBA (Fig. 2F). Interestingly, ELISA measurements also revealed that the concentration of the anti-inflammatory cytokine IL-10 was significantly increased in 4-PBA–treated RAW MΦs stimulated with LPS (Fig. 2G). In contrast, 4-PBA treatment dose dependently reduced the LPS-induced secretion of the anti-inflammatory cytokine IL-10 in BMDMs (Fig. 2G). However, IL-10 was not detected by ELISA in PMs.

These results indicate that peroxisome induction in murine MΦs leads to suppression of the proinflammatory phenotype in these cells, which is independent of ER stress.

The inhibitory effects of peroxisome induction on LPS-dependent upregulation of proinflammatory proteins may suggest that peroxisomes affect the proinflammatory phenotype of MΦs. To substantiate this hypothesis, we tested the proinflammatory response in MΦs, in which specific peroxisomal functions were modulated. Using an siRNA-mediated approach, we targeted the peroxisomal biogenesis protein PEX14p, because earlier reports demonstrated that knockout of peroxisomal biogenesis proteins cause a lack of functional peroxisomes (3, 4, 28). Immunofluorescence studies revealed an efficient knockdown of PEX14p in RAW MΦs (Fig. 3A, 3B). Further, Western blot and ELISA analysis indicated that treatment with LPS in Pex14 siRNA–transfected RAW MΦs significantly increased the protein levels of COX-2 and TNF-α in comparison with control siRNA–transfected RAW MΦs stimulated with LPS (Fig. 3C, 3D, Supplemental Fig. 3A).

Peroxisome deficiency may be coupled with mitochondrial alterations, including effects on the respiratory chain (3). Thus, it is conceivable that the regulatory effect on COX-2 expression might be indirectly caused by disturbance of the mitochondrial respiratory chain. To test this hypothesis, RAW MΦs were treated with the complex III inhibitor antimycin A, and mitochondrial dysfunction was evaluated by staining with MitoTracker (Supplemental Fig. 3B), which requires an intact mitochondrial potential to be incorporated into the mitochondria. Indeed, in cells treated with antimycin A, MitoTracker staining shifted from a mitochondrial pattern to a nuclear pattern, indicating mitochondrial dysfunction (Supplemental Fig. 3B). Upon treatment with LPS, the COX-2 protein levels in antimycin A–treated cells were downregulated in comparison with control cells stimulated with LPS (Supplemental Fig. 3C). Thus, mitochondrial dysfunction per se may not induce upregulation of COX-2 protein in LPS-treated MΦs.

To determine a potential role for the peroxisomal β-oxidation machinery in the regulation of COX-2, we knocked down Mfp2, a gene encoding a central peroxisomal β-oxidation enzyme, in RAW MΦs (Fig. 3E, 3F). Similar to Pex14 siRNA–transfected cells, stimulation with LPS in Mfp2 siRNA–transfected RAW MΦs led to a pronounced increase in COX-2 and TNF-α protein levels in comparison with LPS-treated control siRNA–transfected RAW MΦs (Fig. 3G, 3H, Supplemental Fig. 3D). In summary, these results show that peroxisomes and their β-oxidation machinery negatively regulate COX-2 and TNF-α protein expression.

TLR4-mediated signaling to proinflammatory genes is tightly regulated at the transcriptional level via the transcription factors NF-κB and activator protein-1 (AP-1). Because LPS-dependent upregulation of proinflammatory proteins is suppressed in 4-PBA–treated MΦs, we hypothesized that induction of peroxisomes may affect the activation of these transcription factors. However, treatment with 4-PBA did not block LPS-induced activation of NF-κB (as determined by the nuclear translocation of the NF-κB subunit p65) (Fig. 4A–C) or AP-1 (as determined by phosphorylation of c-jun) (Fig. 4D–F). Accordingly, real-time PCR analysis revealed that LPS-induced mRNA expression of COX-2 and TNF-α were not reduced, but rather were increased in 4-PBA–treated RAW MΦs in comparison with controls (Fig. 4G). Moreover, real-time PCR analysis showed that LPS-dependent induction of COX-2 and TNF-α mRNA levels was comparable in Pex14 siRNA–, Mfp2 siRNA–, and control siRNA–transfected MΦs (Fig. 4H, 4I)

Because treatment with 4-PBA did not interfere with the transcriptional upregulation, we analyzed whether 4-PBA affects COX-2 protein stability. PMs were stimulated with LPS for 12 h and then treated with CHX for 2, 4, and 6 h, followed by Western blot analyses of COX-2 protein. In control PMs, COX-2 was not significantly reduced after up to 4 h of CHX treatment, and a reduction in COX-2 protein levels was only apparent after 6 h. In contrast, in 4-PBA–treated PMs, a reduction in COX-2 protein levels was evident as early as 2 h after CHX treatment and continued to decrease at 4 and 6 h (Fig. 5A, 5B). The calculated protein half-lives indicate that LPS-induced COX-2 protein has a much shorter half-life in 4-PBA–treated MΦs (Fig. 5B).

Because observations in RAW MΦs are similar to those in primary AMs and PMs (Figs. 1, 2), most of the following experiments were conducted in RAW MΦs (unless specified) to minimize the number of animals sacrificed.

To evaluate whether the observed protein degradation was mediated via a proteasomal or lysosomal pathway, we applied two chemical inhibitors to block these pathways. The concentration of the chemicals used was based on existing literature (29–32). However, pretreatment with proteasomal (MG132, epoxomicin) and lysosomal (3-methyladenine, chloroquine) inhibitors for 2 h did not block 4-PBA–mediated COX-2 protein degradation (Fig. 5C). Similar to our present findings, recent studies demonstrated a lysosome and proteasome–independent protein-degradation pathway for COX-2. This new COX-2–degradation pathway can be blocked by specific COX-2 enzyme inhibitors (33). To evaluate this possibility, 4-PBA–treated RAW MΦs were treated with the specific COX-2 inhibitor NS-398 and the general COX inhibitor S-flurbiprofen 2 h prior to the addition of LPS. As demonstrated in Fig. 5D and 5E, the blunted LPS induction of COX-2 protein was restored by these two inhibitors in 4-PBA–treated RAW MΦs.

In summary, LPS-induced COX-2 protein is destabilized in peroxisome-induced cells, which can be blocked by the inhibition of COX-2 enzyme activity but not by blocking classical protein-degradation pathways.

Because exposure of 4-PBA–treated MΦs to LPS led to release of the anti-inflammatory cytokine IL-10 into the medium, we specifically evaluated the effect of this cytokine on LPS-induced COX-2 expression. Pretreatment of RAW MΦs with recombinant mouse IL-10 for 24 h induced protein expression of the anti-inflammatory gene HO-1, as described in earlier reports (34), which was blocked when IL-10 was subjected to heat inactivation at 95°C for 30 min (Fig. 6A). However, pretreatment with IL-10 for 2 or 24 h did not block the upregulation of COX-2 by LPS (Fig. 6B). Alternatively, Cox-2 regulation in Mfp2-knockdown cells may implicate a role for peroxisomal β-oxidation machinery in this pathway. Hence, we asked whether a peroxisome-derived lipid mediator is involved in LPS-dependent regulation of COX-2. To test this, we collected conditioned medium from 4-PBA–treated RAW MΦs stimulated with LPS and added them to naive RAW MΦs with an additional dose of LPS (100 ng/ml). We chose conditioned medium obtained from RAW MΦs treated with 5 mM 4-PBA because this concentration gave the maximum increase for peroxisomal protein PEX14p in our experimental set-up (Supplemental Fig. 4A). Interestingly, conditioned medium from 4-PBA–treated RAW MΦs, but not from control RAW MΦs, reduced LPS induction of COX-2 in naive RAW MΦs (Fig. 6C). To further evaluate the nature of this secreted cell-extrinsic anti-inflammatory mediator(s), conditioned medium was subjected to heat inactivation; however, this did not block its ability to reduce the LPS induction of COX-2 in naive RAW MΦs (Fig. 6D). Further, conditioned medium from murine RAW MΦs was added to primary peripheral blood monocyte–derived human MΦs to test a potential species specificity of this effect. Conditioned medium from control RAW MΦs had minimal effects on the LPS induction of COX-2 in primary monocyte-derived human MΦs. In contrast, conditioned medium from 4-PBA–treated MΦs blunted the LPS-dependent induction of COX-2 protein in primary monocyte-derived human MΦs (Fig. 6E).

Next, the conditioned medium obtained from control RAW MΦs and 4-PBA–treated RAW MΦs stimulated with LPS were analyzed by liquid chromatography–tandem MS (LC-MS/MS) for the presence of lipid mediators. As expected, the levels of AA that are the substrate of COX-2 were higher in conditioned medium from 4-PBA–treated RAW MΦs (Fig. 6F). Interestingly, DHA, which is the precursor of many lipid resolution mediators, and PGJ₂, metabolites of which are anti-inflammatory, were also significantly higher in conditioned medium from 4-PBA–treated RAW MΦs in comparison with conditioned medium obtained from LPS-treated control RAW MΦs (Fig. 6F). The concentration of the proinflammatory LTB₄, but not PGE₂/D₂, was significantly reduced in the conditioned medium from 4-PBA–treated MΦs (Fig. 6F).

In summary, peroxisome-derived regulation of COX-2 may be mediated by a heat-stable extrinsic factor(s) (most likely lipid in nature) released into the cell supernatant but not by IL-10.

Next, we asked whether treatment of MΦs with 4-PBA might affect the polarization to M2 MΦs that could explain the anti-inflammatory nature of these MΦs. To this end, we determined levels of the M1 marker CD86 and the M2 marker CD36 by flow cytometry. To our surprise, the M1 marker CD86, but not the M2 marker CD36, was significantly higher in 4-PBA–treated MΦs in comparison with untreated control MΦs (Fig. 7A). Further, real-time PCR analysis revealed that the M1 marker inducible NO synthase was reduced, whereas IL-6 was induced, in 4-PBA–treated MΦs (Supplemental Fig. 4B). Similarly, expression of the M2 marker Arg-1 was significantly reduced and that of IL-10 was induced in 4-PBA–treated MΦs (Supplemental Fig. 4B). Stimulation with LPS induced expression of the M2 markers Arg-1 and IL-10 in 4-PBA–treated MΦs in comparison with control MΦs. In contrast, M1 markers inducible NO synthase and IL-6 did not change markedly (Supplemental Fig. 4B), indicating that 4-PBA treatment may not skew murine MΦs to the classical M2 phenotype that exhibits reduced gene expression of proinflammatory cytokines.

Next, we analyzed whether the peroxisome-dependent regulation of COX-2 may serve as a physiological negative-feedback mechanism in MΦs activated with LPS. To explore this possibility, BMDMs were treated with LPS for different periods of time. LPS stimulation induced PEX14p, as observed by Western blot, with a strong increase at 48 h (Fig. 7B, 7C). Strikingly, LPS induction of COX-2 protein was decreased after 48 and 72 h in comparison with 24 h (Fig. 7B). However, at the mRNA level, COX-2 was differentially regulated, with a maximum peak at 48 h (Fig. 7D). In conclusion, these findings show that upregulation of PEX14p correlates with the late-phase inhibition of COX-2 protein in LPS-activated MΦs. In conclusion, our findings indicate that induction of peroxisomes is a physiological anti-inflammatory response at late time points that potentially restores immune homeostasis.

Next, we also explored whether this peroxisome-dependent anti-inflammatory effect on MΦs was TLR4 specific. To address this we used two additional TLR ligands: LTA (a major component of Gram-positive bacteria) for TLR2 and poly I:C (mimicking double-stranded viral RNA) for TLR3. Treatment with LTA induced the protein expression of COX-2, which was blocked when cells were pretreated with 4-PBA for 72 h. In contrast, the minor induction of COX-2 upon poly I:C stimulation was not affected by pretreatment of MΦs with 4-PBA (Fig. 7E). To summarize, peroxisome induction blocks LPS- and LTA-induced COX-2 expression but not poly I:C–mediated COX-2 induction.

Finally, we also tried to address the cell specificity of the peroxisome-mediated anti-inflammatory response in a nonimmune cell type. For this purpose, we used primary MEFs. Treatment with 4-PBA at a concentration of 3.5 mM for 5 d induced peroxisomes (Supplemental Fig. 4C). In contrast to MΦs, MEFs had a constitutive expression of COX-2, and stimulation with LPS only caused a minor induction of COX-2 (Fig. 7F). However, treatment with 4-PBA did not block COX-2 expression in MEFs (Fig. 7F). To summarize, peroxisome-dependent anti-inflammatory regulation of COX-2 may not exist in MEFs.

Peroxisomes were initially described in 1954 as microbodies and were later identified in 1966 as organelles containing enzymes involved in H₂O₂ metabolism (35). In this study, we show that 4-PBA–mediated proliferation of peroxisomes leads to an attenuated LPS-mediated proinflammatory response in MΦs. In contrast, disturbing peroxisome function by knockdown of the peroxisome-specific genes Pex14 and Mfp2 causes an enhanced inflammatory response marked by an increase in the LPS-induced upregulation of proinflammatory proteins COX-2 and TNF-α, suggesting a direct immunomodulatory role for peroxisomes in MΦs.

4-PBA is considered a nonclassical peroxisome proliferator, because it mediates peroxisome proliferation independent of PPAR-α and does not belong to the group of classical PPAR activators (23). To induce peroxisome proliferation, the primary choice is the classical PPAR agonists; however, in MΦ cultures, PPAR agonists exhibit pleiotropic anti-inflammatory effects, such as inhibition of NF-κB and AP-1 signaling pathways that leads to the suppression of various proinflammatory genes (36). Thus, the use of a classical PPAR agonist would make it almost impossible to distinguish the specific effects of peroxisome proliferation from the described anti-inflammatory effects of PPAR activation. However, it is noteworthy that peroxisome proliferation by these agonists was never analyzed in connection with the anti-inflammatory effects observed. In this article, we report that 4-PBA is a potent inducer of peroxisomes in cell cultures of RAW MΦs and various mouse primary MΦs, such as AMs and PMs, but not primary BMDMs (Fig. 1). 4-PBA–mediated peroxisome proliferation subsequently downregulated the LPS-induced protein expression of the proinflammatory gene COX-2 and the proinflammatory cytokines TNF-α, IL-6, and IL-12. Similar results were observed with two other nonclassical peroxisome inducers: HU and L-SF. This conclusion is supported by the finding that 4-PBA treatment did not induce peroxisomes or provide an anti-inflammatory effect in BMDMs (Fig. 2). The importance of peroxisome proliferation to mediate the 4-PBA–dependent anti-inflammatory effect is also indicated by the observation that pretreatment with 4-PBA for 2 h, a period not sufficient for proliferation of peroxisomes, failed to abrogate LPS-induced COX-2 (Fig. 2). Additionally, a role for ER stress inhibition for these anti-inflammatory effects was ruled out, because the concentrations used in our experiments did not block ER stress (Supplemental Fig. 2). Furthermore, the anti-inflammatory effect of 4-PBA on COX-2 protein levels was not only TLR4 specific, it also was observed with the TLR2 ligand LTA but not with the TLR3 ligand poly I:C (Fig. 7E). TLR4 and TLR2 work as pathogen-associated molecular patterns for bacterial infection, whereas TLR3 is involved in a viral response. Because COX-2 was only weakly induced by the TLR3 ligand poly I:C, further studies need to be done to address whether there are specific differences in peroxisome-mediated anti-inflammatory responses upon stimulation with bacterial and viral agents.

The homeostasis of MΦs after proinflammatory activation is tightly controlled to prevent exacerbated pathological events. Activation of MΦs is followed by a late-phase upregulation of anti-inflammatory proteins, such as Nod2, IL-10, and TGF-β, which act as inflammation suppressors/MΦ deactivators in an autocrine and paracrine fashion (37). The regulatory checkpoints for deactivation include transcriptional repression (repressors of signaling cascades) (37), posttranscriptional regulation (mRNA stability) (38), and translational silencing (39). These regulatory checkpoints exist independent of the M1-M2 paradigm of MΦ activation. Because proliferation of peroxisomes does not skew the MΦs into a classical M2-like phenotype (Fig. 7A, Supplemental Fig. 4B), we propose that the upregulation of peroxisome proliferation may serve as an autoregulatory mechanism in MΦs to protect against uncontrolled activation. This hypothesis is supported by our findings that knockdown of peroxisomal proteins led to an induction of proinflammatory proteins, indicating the existence of an intrinsic peroxisome-mediated regulation of proinflammatory proteins (Fig. 3). Moreover, the finding that LPS-dependent upregulation of PEX14p coincides with the downregulation of COX-2 (Fig. 7B) also supports this conclusion. Hence, we speculate that in vivo peroxisomes may act as late-phase inflammation suppressors at the posttranslational level to self-regulate inflammatory MΦs. In line with this notion are the earlier findings that, in a mouse model of endotoxemia in vivo, peroxisomes are initially downregulated, followed by proliferation and increased abundance in the late phase of inflammation (40). Moreover, in the earlier report, accumulation of nonfunctional peroxisomes was shown to aggravate the proinflammatory condition (40).

COX-2, the inducible isoform of the PG synthase, is upregulated in inflammatory conditions and thought to have a major role in promoting inflammation via production of proinflammatory PGs (41). Peroxisome-dependent COX-2 regulation was mediated via a posttranslational mechanism independent of proteasomal and lysosomal protein degradation. These findings agree with previous reports showing that the COX-2 protein undergoes a distinct protein-degradation pathway in vitro as well as in vivo, which is dependent upon the substrate turnover (33, 42). Accordingly, LC-MS/MS analysis of conditioned medium from LPS-activated 4-PBA–treated MΦs revealed a dramatic increase in the levels of free AA, which may induce the substrate-dependent suicidal pathway of COX-2, as shown by other investigators. Additionally, LC-MS/MS analysis revealed increased levels of PGJ₂ and DHA in the conditioned medium from 4-PBA–treated MΦs. PG15d-PGJ₂ is anti-inflammatory in nature (43) and might be involved in the downregulation of proinflammatory proteins. Similarly, the increased levels of DHA that forms the backbone for lipid resolution mediators, such as resolvins, protectins, and maresins, may also have an impact on the anti-inflammatory nature of 4-PBA–treated MΦs. Indeed, treatment of MΦs with DHA was shown to downregulate the LPS induction of proinflammatory proteins (44, 45). Although the presence of peroxisome-derived cell extrinsic factors was confirmed, identification and individual application of these metabolites will be crucial to understand the exact nature of this phenomenon. Currently, we are characterizing the lipid metabolites produced in 4-PBA–treated MΦs. Although our findings indicate that peroxisome-mediated effects on COX-2 expression may be MΦ specific (Fig. 7F), further characterization of the lipid mediators is necessary to evaluate the cell-type specificity of this phenomenon in other immune and nonimmune cell types. Peroxisome-mediated immunomodulation of MΦ inflammatory responses may open up new options for therapeutic interventions in chronic inflammatory conditions. 4-PBA is already in clinical use for treating a range of disorders, and phase II clinical trials for the treatment of cystic fibrosis were performed. Additionally, 4-PBA was shown to block LPS-induced expression of the proinflammatory cytokines TNF-α and IL-6 in an in vivo mouse model of lung inflammation (27). The beneficial effects of blocking proinflammatory proteins, such as COX-2 and TNF-α in various inflammatory disorders have been well documented. Moreover, the broad release of lipid-resolution mediators and the potent anti-inflammatory cytokine IL-10 further impact the resolution of inflammation. However, the lack of a PPAR-independent monospecific inducer of peroxisomes remains a limiting factor in exploring the complete potential of these organelles.

In summary, we showed that peroxisomes provide an anti-inflammatory phenotype to MΦs and conclude that 4-PBA–mediated peroxisome proliferation may be beneficial for therapeutic interventions in chronic inflammatory disorders.

We thank Petra Hahn-Kohlberger, Andrea Textor (E.B.-V group), and Anette Sarti-Jacobi (S.I. group) for excellent technical assistance. We also thank Dr. Kashyap Krishnasamy (Department of Nephrology and Hypertension, Medical School Hannover) for valuable experimental discussions. Furthermore, we thank Prof. Dr. Denis I. Crane (School of Biomolecular and Physical Sciences, Griffith University, Brisbane, QLD, Australia) for providing the anti-PEX14 Ab and Privatdozent Dr. Christoph Rummel (Faculty of Veterinary Medicine, Justus Liebig University Giessen) for providing some of the lipid standards.

The authors have no financial conflicts of interest.