Mycobacterium tuberculosis–Induced Prostaglandin J₂ and 15-Deoxy-Prostaglandin J₂ Inhibit Inflammatory Signals in Human M1 Macrophages via a Negative Feedback Loop

Tuberculosis (TB), caused by Mycobacterium tuberculosis, is considered one of the major epidemics and the top contagious disease killer (1), with 1.6 million deaths in 2021 (2). Although most people are affected by latent TB infection, only 5–10% of infected humans develop active TB (3). Host macrophages are the first-line responders that phagocytose M. tuberculosis and serve as reservoir for the bacterium (4). To invade these host cells, M. tuberculosis modulates macrophage lipid metabolism, resulting in eicosanoid release, and diminishes inflammatory responses in M1-like macrophages (5). Macrophage infection by M. tuberculosis first favors M1 polarization, accompanied by upregulation of lipid mediator (LM) biosynthesis, followed by a resolution phase characterized by dampening the proinflammatory and antimicrobial responses of macrophages (6).

Eicosanoids, including PGs and leukotrienes, are bioactive LMs derived from arachidonic acid (AA) and function as key regulators during macrophage reactions to M. tuberculosis (7, 8). PGs are produced by consecutive actions of cyclooxygenase-1 (COX-1) and COX-2 and various PG synthases, where COX-derived PGH₂ is transformed to PGE₂, PGF₂α, PGI₂, TXs, and PGD₂ (9, 10); the latter yields the metabolite 15-deoxy-Δ^12,14PGD₂ (15d-PGD₂) and PGJ₂ that further decomposes to 15-deoxy-Δ^12,14PGJ₂ (15d-PGJ₂) (11) (Fig. 1A). Among the major bioactive PGs in humans, PGE₂ and PGD₂ possess either proinflammatory or anti-inflammatory functions depending on the phase of inflammation, the type of tissue, and the expression of respective receptor subtypes (e.g., EP1–4 for PGE₂ or DP1 and chemoattractant receptor homologue expressed on Th2 cells [CRTH2]/DP2 for PGD₂) (12–16). For example, PGE₂ is massively produced by inducible COX-2 and microsomal PGE₂ synthase-1 (mPGES-1) in monocytes and M1 macrophages (10) at the onset and progression of inflammation with proinflammatory features but acts anti-inflammatory at the later resolution phase (13, 14). PGE₂ promotes the defense against M. tuberculosis infection (17), e.g., by inhibiting M. tuberculosis replication in macrophages (18), and the PGE₂-EP4 axis acts as positive feedback loop for PGE₂ formation in M. tuberculosis–infected macrophages (19). Accordingly, deletion of COX-2 or EP4 caused dysfunctional phagocytosis ability of macrophages and increased expression of the proinflammatory cytokines TNF-α, IL-6, and IL-1β (20).

The role of PGD₂ in inflammation is controversial, with proinflammatory effects inducing immune cell accumulation but anti-inflammatory impact by inhibiting the recruitment of dendritic cells and neutrophils (21, 22). PGD₂-DP1 mediates immunosuppression in acute lung injury (21), but the impact of PGD₂ in M. tuberculosis infection is unknown. The J₂-PGs (e.g., PGJ₂ and its 15-deoxy metabolite 15d-PGJ₂) are dehydration products of PGD₂, characterized by a cyclopentenone ring with α,β-unsaturated carbonyl groups that form covalent Michael adducts with cysteines in proteins (23). They modulate inflammatory responses and other cellular events by interacting with peroxisome proliferator-activated receptor-γ (PPARγ) or the PGD₂ receptors DP1 and CRTH2/DP2 (11). J₂-PGs are considered as anti-inflammatory mediators but also induce pathological processes relevant to neurodegenerative disorders, including Alzheimer’s disease. It was shown that J₂-PGs mitigate proinflammatory cytokine production in macrophages, including IL-1β, IL-6, and IL-12 (24). With respect to host defense, 15d-PGJ₂ enhanced phagocytosis of alveolar macrophages via PPARγ (25, 26) but also suppressed phagocytosis, independent of PPARγ, in murine bone marrow–derived macrophages (27). Despite the extensive research on the (patho)physiological roles of J₂-PGs, their functions on M. tuberculosis–invaded macrophages remain to be determined.

In a previous study, we performed LM analysis in human M1-MDMs (monocyte-derived macrophages) exposed to M. tuberculosis (H37Rv strain)–conditioned medium (MTB-CM) that contains stimulatory factors that markedly elevated release of PGE₂ (28). In this study, we found that besides PGE₂, abundant PGJ₂ was generated by the COX-2 pathway in MTB-CM–stimulated human M1-MDMs. PGJ₂ pretreatment increased the survival rate of bacteria in THP-1 macrophages, likely attributed to its anti-inflammatory effects. Moreover, PGJ₂ suppressed upregulation of COX-2 expression and proinflammatory cytokine release but fostered expression of anti-inflammatory markers. These effects correlate with the activation status of the TGF-β–activated kinase 1 (TAK1)/NF-κB p65/p38 MAPK and ERK1/2 pathways, which seemingly are mediated by DP1 and CRTH2/DP2 receptors, but not via PPARγ.

M. tuberculosis strain H37Ra (25177; ATCC) and M. tuberculosis strain H37Rv (25618; ATCC) were cultured in 7H9 broth (BD Biosciences, San Jose, CA) with 0.2% glycerol (Sigma-Aldrich, Merck, Darmstadt, Germany), 10% Middlebrook oleic acid-albumin-dextrose-catalase enrichment (BD Biosciences), and 0.05% Tween 80 (Sigma-Aldrich). The cultures were incubated at 37°C for 1–2 wk with shaking to the midlogarithmic phase (OD at 600 nm was 0.3–0.8) before experiments. In all experiments involving mycobacterial infection, differentiated THP-1 macrophages (TIB-202; ATCC) were infected with H37Ra or H37Rv at a multiplicity of infection (MOI) of 10 (10:1 mycobacteria/macrophage). For THP-1 cell differentiation, human monocytic THP-1 cells were grown in RPMI 1640 (Sigma-Aldrich) supplemented with l-glutamine (2 mM) and 10% (v/v) heat-inactivated FCS (Biochrom/Merck, Berlin, Germany). The THP-1 cells were treated with 40 ng/ml PMA (Sigma-Aldrich) in 12-well plates at 37°C for 24 h for differentiation to macrophages. Before infection, the 7H9-cultured M. tuberculosis bacteria were washed once with PBS, resuspended in serum-free RPMI 1640 medium, and then sonicated for 5 min to obtain a single-cell suspension. The M. tuberculosis cells were not opsonized before infection.

M. tuberculosis strain H37Rv and M. bovis strain BCG (DSM 43990; German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany) were inoculated in Middlebrook 7H9 broth with 0.2% glycerol and 10% oleic acid-albumin-dextrose-catalase supplement (Merck, Germany). The cultures were incubated at 37°C for 2–3 mo without shaking. Bacterial cells were pelleted by centrifugation (4150 × g, 10 min) before the supernatant was collected and filtrated (0.2-µm filter). For control purposes (mock), noninoculated broth was sterile filtrated and used for experiments.

PGE₂ (Item No. 14010), PGD₂ (Item No. 12010), PGJ₂ (Item No. 18500), 15d-PGJ₂ (Item No. 18570), and the reagents GW9662 (Item No. 70785), MK-0524 (Item No. 10009835), BAY-u3405 (Item No. 10156), and AT-56 (Item No. 13160) were purchased from Cayman Chemicals/Biomol (Hamburg, Germany) and stored at −80°C under nitrogen atmosphere. The PGs were dissolved in methyl acetate (CAS No. 79209; Merck, Darmstadt, Germany), and other reagents were dissolved in DMSO and then further diluted in RPMI 1640 medium for cell treatments.

For isolation of human MDMs, leukocyte concentrates from freshly withdrawn peripheral blood from healthy human male and female volunteers (aged 18–65 y, without anti-inflammatory treatment for the last 10 d) were obtained from the Institute of Transfusion Medicine at the University Hospital Jena (Jena, Germany). All protocols for experiments involving human blood cells were approved by the ethical commission (approval no. 5050–01/17) of the Friedrich Schiller University Jena (Jena, Germany). All methods were performed in accordance to the relevant guidelines and regulations (ethical guidelines of the 1975 Declaration of Helsinki). PBMCs were separated by dextran (from Leuconostoc spp., m.w. ∼40,000; Sigma-Aldrich) for sedimentation, followed by centrifugation (2000 × g, 10 min, room temperature) on lymphocyte separation medium (Histopaque-1077; Sigma-Aldrich). PBMCs were washed twice with ice-cold PBS and counted and then seeded in cell culture flasks (Greiner Bio-One, Frickenhausen, Germany) for 1.5 h at 37°C and 5% CO₂ in PBS with Ca²⁺/Mg²⁺ for adherence of monocytes. For differentiation of monocytes to MDMs and polarization toward M1 and M2 phenotypes, published criteria were used (29). In brief, M1-MDMs were generated by incubating monocytes with 20 ng/ml GM-CSF (PeproTech, Hamburg, Germany) for 6 d in RPMI 1640 containing 10% (v/v) heat-inactivated FCS, 2 mM l-glutamine, 100 U/ml penicillin, and 100 µg/ml streptomycin (Sigma-Aldrich), followed by treatment with 100 ng/ml LPS and 20 ng/ml IFN-γ (PeproTech). M2-MDMs were generated by incubating monocytes with 20 ng/ml M-CSF (PeproTech) for 6 d, followed by treatment with 20 ng/ml IL-4 (PeproTech). Routinely, cells were polarized for 48 h unless stated otherwise.

Human M1-MDMs (2 × 10⁶/ml) were incubated with control (10% MB broth) or 10% MTB-CM for 24 h at 37°C or pretreated with vehicle (methyl acetate), 10 µM PGJ₂, or 10 µM 15d-PGJ₂ for 2 h, then incubated with control (10% MB broth) or 10% MTB-CM for 24 h at 37°C. LM formation was stopped by transferring cell culture supernatants (1 ml) into 2 ml ice-cold methanol.

The samples from the supernatant of incubated M1-MDMs were kept at −20°C for at least 60 min to allow protein precipitation. After centrifugation (1200 × g, 4°C, 10 min), 8 ml acidified H₂O (final pH 3.5) was added, and the samples were subjected to solid-phase cartridges (Sep-Pak Vac 6cc 500 mg/6 ml C18; Waters, Milford, MA). The columns had been equilibrated with 6 ml methanol and 2 ml H₂O before sample loading. After washing with 6 ml H₂O and subsequently with 6 ml n-hexane, the LM samples were eluted with 6 ml methyl formate. The eluate was brought to dryness using an evaporation system (TurboVap LV; Biotage, Uppsala, Sweden), and the remaining sample was resuspended in 100 μl methanol/water (50/50, v/v) for ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS-MS) automated injection and analysis. LMs were analyzed with an Acquity UPLC system (Waters, Milford, MA) and a QTRAP 5500 Mass Spectrometer (ABSciex, Darmstadt, Germany) equipped with a Turbo V Source and electrospray ionization. LMs were eluted using an ACQUITY UPLC BEH C18 column (1.7 μm, 2.1 mm × 100 mm; Waters, Eschborn, Germany) at 50°C with a flow rate of 0.3 ml/min and a mobile phase consisting of methanol-water-acetic acid of 42:58:0.01 (v/v/v) that was ramped to 86:14:0.01 (v/v/v) over 12.5 min and then to 98:2:0.01 (v/v/v) for 3 min (30). The QTRAP 5500 was operated in negative ionization mode using scheduled multiple reaction monitoring coupled with information-dependent acquisition. The scheduled multiple reaction monitoring window was 60 s, optimized LM parameters were adopted (30), and the curtain gas pressure was set to 35 ψ. The retention time and at least six diagnostic ions for each LM were confirmed by means of an external standard (Cayman Chemical/Biomol). Quantification was achieved by calibration curves for each LM. Linear calibration curves were obtained for each LM and gave r² values of ≥0.998 (for fatty acids ≥0.95). In addition, the limit of detection for each targeted LM was determined as previously reported (30).

Total cellular RNA was extracted using E.Z.N.A Total RNA Kit 1 (Omega Bio-tek, Norcross, GA), and the isolated RNA was reverse transcribed into cDNA with High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions. The cDNA was mixed with PerfeCTa SYBR Green SuperMix, ROX kit (Quantabio, Beverly, MA), and the real-time PCR was performed on a qTOWER^3G touch Instrument (Analytic Jena, Jena, Germany). The primers used for the real-time PCR are as follows: COX-2 (forward: 5′-TGCCTGATGATTGCCCGACT-3′; reverse: 5′-TGAAAGCTGGCCCTCGCTTA-3′); 5-lipoxygenase (5-LOX; forward: 5′-ACCCACCTTCTGCGAACACA-3′; reverse: 5′-GTGGCGTTGGCCTTGTCAAA-3′); mPGES-1 (forward: 5′-GGAACGACATGGAGACCATC-3′; reverse: 5′-GGAAGACCAGGAAGTGCATC-3′); 15-LOX-1 (forward: 5′-CTTCAAGCTTATAATTCCCCAC-3′; reverse: 5′-GATTCCTTCCACATACCGATAG-3′); L-PGDS (forward: 5′-ACCAGTGTGAGACCCGAACC-3′; reverse: 5′-CAGCGCGTACTGGTCGTAGT-3′); IL-6 (forward: 5′-TTCGGTACATCCTCGACGGC-3′; reverse: 5′-TCTGCCAGTGCCTCTTTGCT-3′); IL-1β (forward: 5′-TTCGAGGCACAAGGCACAAC-3′; reverse: 5′-TTCACTGGCGAGCTCAGGTA-3′); TNF-α (forward: 5′-ACTTTGGAGTGATCGGCCCC-3′; reverse: 5′-TGGGCTACAGGCTTGTCACT-3′); IFN-γ (forward: 5′-GGCTTTTCAGCTCTGCATCGT-3′; reverse: 5′-CGCTACATCTGAATGACCTGC-3′); IL-10 (forward: 5′-GGGCACCCAGTCTGAGAACA-3′; reverse: 5′-GACAAGGCTTGGCAACCCAG-3′); Arg-1 (forward: 5′-TACTGGCACACCAGTCGTGG-3′; reverse: 5′-TCCCCAGGGATGGGTTCACT-3′); CD163 (forward: 5′-CAGCGGCTTGCAGTTTCCTC-3′; reverse: 5′-TGGCCTCCTTTTCCATTCCAGA-3′); and β-actin (forward: 5′-ACAGAGCCTCGCCTTTGCC-3′; reverse: 5′-CCATCACGCCCTGGTGCC-3′).

Human M1-MDMs were seeded at a density of 1 × 10⁶ cells/well into 12-well culture plates. After experimental treatment, cells were lysed on ice with Seaman lysis buffer (31) with protease inhibitors (1 mM phenylmethanesulfonylfluoride [Sigma-Aldrich], 60 μg/ml soybean trypsin inhibitor [Merck], 10 μg/ml leupeptin [Sigma-Aldrich]). Samples were then centrifuged for 5 min at 4°C to remove cell debris. Protein concentrations were determined using DC Protein Assay Kit (Bio-Rad, Munich, Germany). Samples were then mixed with 5× Laemmli buffer (31) and heated to 95°C for 10 min. A total of 25–35 µg whole-cell extracts was then separated on 10% or 14% SDS-PAGE acrylamide gels and subsequently transferred to nitrocellulose membranes (Amersham Protran Supported 0.45 μm nitrocellulose; Cytiva, Freiburg, Germany). After blocking with 5% BSA (T844.3; Roth, Karlsruhe, Germany) in TBST, the proteins were detected by incubation with primary Abs: rabbit monoclonal anti–COX-2, 1:500 (12282S; Cell Signaling, Danvers, MA); mouse monoclonal anti–β-actin, 1:1000 (3700S; Cell Signaling); rabbit polyclonal anti–mPGES-1, 1:5000 (kindly provided by Dr. P.-J. Jakobsson, Karolinska Institute, Stockholm, Sweden); rabbit monoclonal anti-p38 MAPK, 1:1000 (8690S; Cell Signaling); rabbit polyclonal anti–phospho-p38 MAPK (Thr180/Tyr182), 1:1000 (9211S; Cell Signaling); rabbit monoclonal anti–NF-κB p65 (C22B4), 1:1000 (4764S; Cell Signaling); mouse monoclonal anti–phospho-NF-κB p65 (Ser536), 1:1000 (13346S; Cell Signaling); mouse monoclonal anti-phospho-ERK 1/2 (Thr202/Tyr204), 1:1000 (9106; Cell Signaling); rabbit monoclonal anti-ERK1/2, 1:1000 (4695; Cell Signaling); rabbit monoclonal anti–phospho-TAK1 (Thr184/187), 1:500 (4508; Cell Signaling); and rabbit monoclonal anti-TAK, 1:500 (5206; Cell Signaling). Immunoreactive bands were stained with IRDye 800CW goat anti-mouse IgG (H+L), 1:10,000 (926-32210; LI-COR Biosciences, Lincoln, NE); IRDye 800CW goat anti-rabbit IgG (H+L), 1:15,000 (926-32211; LI-COR Biosciences) and/or IRDye 680LT goat anti-mouse IgG (H+L), 1:40,000 (926-68020; LI-COR Biosciences); IRDye 680LT goat anti-rabbit IgG (H+L), 1:80,000 (926-68021; LI-COR Biosciences). Blots were washed with TBST three times after primary and secondary Ab and finally visualized by an Odyssey infrared imager (LI-COR Biosciences). Data from densitometric analysis were background corrected.

For the CFU counting, 5 × 10⁵ differentiated THP-1 macrophages were seeded into each well of a 12-well plate and infected with M. tuberculosis, as described previously (32). After 6 h of infection, extracellular bacteria were removed by washing twice with sterile PBS. Fresh prewarmed culture medium was added, and the cells were kept for 72 h. Extracellular bacteria were removed by washing twice with sterile PBS. The cells were lysed in 0.1% SDS gradient dilution in sterile PBS and then plated onto 7H10 agar plates (BD Biosciences). The plates were incubated at 37°C at 5% CO₂ for 2–3 wk, and the colonies were enumerated. The survival rate, represented as intracellular bacteria survival, was calculated as follows: survival rate = CFUs of infection (6 h postinfection)/CFUs of media (72 h postinfection) (33).

To evaluate cell viability, we seeded cells in 96-well plates at a density of 1 × 10⁵ per well. The cells were incubated with 1, 3, or 10 µM PGD₂, PGJ₂, or 15d-PGJ₂ for 24 h at 37°C and 5% CO₂. The same volume of methyl acetate was used as a negative vehicle control, and 1 µM staurosporine was used as a positive control. MTT (5 mg/ml, 20 μl; Sigma-Aldrich) solution was added; then the cells were incubated at 37°C at 5% CO₂ for 3 h. The formazan product was solubilized after cell lysis by adding SDS (10% in 20 mM HCl). The absorbance was measured at 570 nm using a Multiskan Spectrum microplate reader (Thermo Fisher Scientific).

To determine M1 and M2 surface marker expression, we seeded human M1-MDMs at a density of 2 × 10⁶ cells/well into six-well culture plates. After experimental treatment, cells were detached with PBS plus 0.5% BSA, 5 mM EDTA, and 0.1% sodium azide for 20 min at 37°C. To determine cell viability, we stained the cells using Zombie Aqua Fixable Viability Kit (BioLegend, San Diego, CA) for 5 min at room temperature. Nonspecific Ab binding was blocked by using mouse serum (10 min, 4°C). Then cells were stained by fluorochrome-labeled Ab mixtures (20 min, 4°C), including FITC mouse monoclonal anti-human CD14 (20 μl/test, clone M5E2, #555397; BD Biosciences), allophycocyanin-H7 mouse monoclonal anti-human CD80 (5 μl/test, clone L307.4, #561134; BD Biosciences), PE-Cy7 mouse monoclonal anti-human CD54 (5 μl/test, clone HA58, #353116; BioLegend), PE mouse monoclonal anti-human CD163 (20 μl/test, clone GHI/61, #556018; BD Biosciences), and allophycocyanin mouse monoclonal anti-human CD206 (20 μl/test, clone 19.2, #550889; BD Biosciences). After staining, MDMs were measured using BD LSR Fortessa cell analyzer (BD Biosciences), and data were analyzed using FlowJo X Software (BD Biosciences).

Data are expressed as means ± SEM. Statistical analysis was performed using GraphPad Prism 9 software (GraphPad Software). A two-tailed unpaired Student t test was used to evaluate the significance of single parameters between two groups. One-way ANOVA with Tukey’s multiple comparisons test was used to analyze more than two groups.

We previously showed that the proinflammatory genes COX-2 and mPGES-1 are increased in human MDMs postinfection with live M. tuberculosis connected to increased PGE₂ levels in MTB-CM–stimulated MDMs, and PGE₂ was also elevated in the plasma of TB patients (28). In this study, we confirmed increased expression of COX-2 and mPGES-1 on the transcription level in both M1-MDMs and M2-MDMs upon exposure to MTB-CM, with unexpected strong effects on mPGES-1 in M2-MDMs (Fig. 1B). Note that the increase of COX-2 mRNA transcripts (at 24 h) in M2-MDMs was more pronounced than in M1-MDM. In contrast, quantities of transcripts of other genes encoding enzymes that are typically involved in LM biosynthesis in macrophages, such as 5-LOX and 15-LOX-1, decreased in M1-MDM, and lipocalin-type PG synthase (L-PGDS) also in M2-MDMs, whereas 5-LOX and 15-LOX-1 transcripts were not affected in the latter phenotype (Fig. 1B). Similarly, protein expression of COX-2 strongly increased in M1-MDMs upon MTB-CM treatment in a temporal manner, peaking at 12 h with subsequent decline. Note that during culture of resting M1-MDMs after polarization in new medium (devoid of polarizing agents), the COX-2 protein levels continuously decreased, obvious at 24 and 48 h (Fig. 1C). Although M2-like alveolar macrophages express COX-2 (34), in the M2-MDMs that we used in this study, the COX-2 levels are very low, in agreement with previous reports (29, 35), but also in this M2 phenotype, MTB-CM elevated COX-2 levels peaking at 6 h with subsequent decline (Fig. 1C). Due to the low COX-2 levels in resting M2-MDMs, MTB-CM caused a relatively strong elevation (38-fold) of COX-2 at 6 h, but nevertheless the absolute protein levels in M1-MDMs (only 13-fold elevated but with already abundant basal COX-2 protein) were overall higher as compared with M2-MDMs, especially at 12, 24, and 48 h. Only minor effects of MTB-CM treatment were evident on mPGES-1 and L-PGDS protein expression (Fig. 1C). Stimulation with attenuated M. bovis bacillus Calmette-Guérin–conditioned medium (BCG-CM) also markedly induced COX-2 on the mRNA (Fig. 1B) and protein level (Fig. 1C) in M1-MDMs and M2-MDMs, again peaking after 4 h, but with minor efficiency as compared with MTB-CM.

To assess the effect of MTB-CM on LM signature profiles of MDMs, we performed comprehensive LM profiling of M1-MDMs and M2-MDMs treated with and without MTB-CM by metabololipidomics using UPLC-MS-MS (30). After exposure of the MDMs to MTB-CM for 24 h, all COX-derived LMs were increased with moderate to high orders of magnitudes, depending on the individual COX product and the MDM phenotype. Thus, we found up to almost 30-fold elevations of PGE₂ and PGJ₂ in M1-MDMs with minor increases (up to ∼10-fold) in M2-MDMs, the latter also producing overall less amounts of COX products as compared with M1-MDMs (Fig. 1D, Table I), in agreement with lower COX-2 protein levels. Note that PGD₂ is unstable and during the long incubation period of 24 h converts to 15d-PGD₂ and PGJ₂, followed by dehydration of PGJ₂ to 15d-PGJ₂ (Fig. 1A) (11). Of interest, the levels of 15d-PGD₂ and 15d-PGJ₂ were comparably low and only moderately increased (≤2-fold) in M1-MDMs and/or M2-MDMs (Table I, Fig. 1D). Formation of other LMs produced by various lipoxygenases, as well as fatty acid release, was rather decreased upon exposure to MTB-CM in both M1-MDMs and M2-MDMs (Table I). These findings suggest that MTB-CM stimulation of MDMs causes marked elevation of COX-derived products connected to an increased expression of COX-2, with striking upregulatory effects particularly on PGE₂ and PGJ₂ formation, implying bioactive functions of these PGs in macrophages infected by M. tuberculosis.

PGs have been reported to modulate phagocytosis (36) and intracellular clearance of bacteria (18). To study whether the anti-inflammatory PGs produced by macrophages upon exposure to M. tuberculosis–derived factors have functional consequences, we investigated their impact on phagocytosis and survival of virulent (H37Rv) and nonvirulent (H37Ra) M. tuberculosis strains within THP-1 macrophages (37). We first assessed cell viability of THP-1 macrophages incubated with H37Ra for 6 and 72 h by MTT assay, and we observed no cytotoxic effects on the macrophages (Supplemental Fig. 1). Exogenously added PGD₂, PGJ₂, and 15d-PGJ₂ (10 µM each) significantly impaired the phagocytic uptake of M. tuberculosis bacteria by the macrophages at the first 6 h postinfection (Fig. 2A, 2D). These lower phagocytosis rates of the bacteria caused by the J₂-PGs were accompanied by slightly elevated numbers of intracellular live bacteria in the macrophages after 72 h of infection, indicating that J₂-PGs rather promote bacterial growth (Fig. 2B, 2E). To determine whether the PGD₂ and the J₂-PGs altered the bacterial burden in the macrophages, we calculated the survival rate of the intracellular bacteria, and we found increased numbers of bacteria, suggesting that PGD₂, PGJ₂, and 15d-PGJ₂ decrease the bactericidal activity of the macrophages (Fig. 2C, 2F). Conclusively, the upregulated PGD₂, PGJ₂, and 15d-PGJ₂ upon M. tuberculosis infection in macrophages compromise the phagocytic activity of the macrophage and exacerbate bacterial burden within these host cells.

Eicosanoids can regulate both innate and adaptive immune responses in M. tuberculosis–infected macrophages (38). For example, 15d-PGJ₂ suppressed the release of proinflammatory molecules from monocytic cells or astrocytes, such as IL-1β, IL-6, NO, and IFN-γ, via a PPARγ-dependent or -independent mechanism (39, 40), which impacts anti–M. tuberculosis activities of macrophages (41). We supposed that the immunosuppressive effect of J₂-PGs may contribute to exacerbated mycobacterial burden in macrophages. Therefore, we determined the effects of the M. tuberculosis–induced J₂-PGs on MDM inflammatory functions in more detail. Because upregulation of COX-2 protein expression and related PG levels caused by MTB-CM stimulation was much more pronounced in M1-MDMs versus M2-MDMs, we proceeded with the inflammatory M1 phenotype. None of the three PGs compromised the viability of M1-MDMs within 24 h at concentrations up to 10 µM, assessed by MTT assays assessing the metabolic activity of the cells (Fig. 3A). Based on the tremendous elevation of PGs upon treatment of M1-MDMs with MTB-CM for 24 h versus mock-treated cells (where COX-2 levels vanished during 24-h culture; compare with Fig. 1C), it appeared reasonable that these PGs could affect the expression of the inducible COX-2 within a negative feedback loop in the M1-MDMs. In fact, pretreatment of M1-MDMs (1 h) with exogenously added PGD₂ (Supplemental Fig. 2A), PGJ₂, and 15d-PGJ₂ impaired the robust induction of COX-2 protein expression due to MTB-CM or BCG-CM at 24 h in a concentration-dependent manner (Fig. 3B). Importantly, no significant reduction of COX-2 protein expression was obvious when M1-MDMs were pretreated with PGE₂ (Supplemental Fig. 2B). Similar suppressive effects of the J2-PGs on COX-2 protein expression were evident when M1-MDMs were incubated with MTB-CM for 12 h, and also M2-MDMs were responsive to PGJ₂ and 15d-PGJ₂ in this respect (Supplemental Fig. 3A). The reduced COX-2 expression levels were accompanied by decreased formation of most COX-2 products (i.e., PGE₂, PGF₂α, and thromboxane B₂ [TXB₂]) in M1-MDMs treated with PGJ₂ and 15d-PGJ₂ (Fig. 3C, Table II). However, PGJ₂, but not 15d-PGJ₂, elevated the release of PGD₂, which is unexpected.

Next, we addressed the effects of PGJ₂ and 15d-PGJ₂ on the mRNA expression of proinflammatory/anti-inflammatory cytokines, as well as of COX-2, in MTB-CM–stimulated M1-MDMs after 12 and 24 h. The mRNA levels of COX-2 were strongly suppressed by both PGs at 12 h (Supplemental Fig. 3B) and at 24 h (Fig. 3D), in agreement with the data from the COX-2 protein analysis (Fig. 3B, Supplemental Fig. 3A). Moreover, the upregulated mRNA expression of the proinflammatory cytokines IL-1β, IL-6, and IFN-γ because of MTB-CM was significantly suppressed by PGJ₂ and 15d-PGJ₂ (10 µM, each), with more pronounced effects by 15d-PGJ₂ (Fig. 3D, Supplemental Fig. 3B). Of interest, the mRNA levels of TNF-α were further increased by PGJ₂, which is not readily understood, without significant modulation by 15d-PGJ₂. In contrast, PGJ₂ elevated the expression of the anti-inflammatory IL-10 and prevented the MTB-CM–induced downregulation of the anti-inflammatory marker Arg-1 with minor impact on CD163; 15d-PGJ₂, in contrast, was not effective in this respect (Fig. 3D). To confirm that MDM-derived PGJ₂ and 15-PGJ₂ contribute to suppression of COX-2, we used the L-PGDS inhibitor AT-56 to block endogenous PGD₂ (and thus PGJ₂ and 15d-PGJ₂) formation. As can be seen from Fig. 3E, blocking J2-PG formation increased the MTB-CM–induced COX-2 protein levels, seemingly because of ablation of the suppressive effects mediated by the endogenously formed J2-PGs.

We further studied how MTB-CM modulates the development of macrophage phenotypes during polarization and the impact of PGJ₂ on this process. M1 and M2 phenotypic surface markers were analyzed by flow cytometry. After stimulation by MTB-CM for 24 h, the expression of the M2 marker CD206 was increased, whereas the M1 markers CD54 and CD80, as well as another M2 marker (CD163), were not altered (Supplemental Fig. 4A). CD206 acts as a pattern recognition receptor and interacts with the M. tuberculosis lipoglycan mannose-capped lipoarabinomannan (42); incubation of MDMs with M. tuberculosis or mannose-capped lipoarabinomannan led to activation of CD206 (43). Interestingly, PGJ₂ reverted the elevated expression of CD206, but also CD80 and CD163 were downregulated, suggesting that PGJ₂ in general diminishes macrophage phenotype development (Supplemental Fig. 4A). Treatment of MDMs with PGJ₂ did not alter CD54, CD80, CD163, and CD206 in M1-MDMs and only slightly reduced CD80, CD163, and CD206 in M2-MDMs (Supplemental Fig. 4B). Together, the data support that elevated PGJ₂ dampens M. tuberculosis–induced inflammatory functions in macrophages, which culminates in exacerbation of mycobacterial burden.

Mycobacterium was shown to induce COX-2 expression in immune cells through TLR2-dependent signaling and activation of signaling pathways, including NF-κB, ERK1/2, and p38 MAPK (19, 44–46). In fact, 15d-PGJ₂ can counteract proinflammatory responses through interference with NF-κB (23, 47), p38 MAPK (48), and ERK1/2 (49). TAK1 is a central regulator of NF-κB and MAPK and might be triggered by M. tuberculosis (50). We thus hypothesized that TAK1/NF-κB/MAPK pathways are involved in the COX-2–suppressive effects of J₂-PGs in M1-MDM. The activation of TAK1, NF-κB p65, p38 MAPK, and ERK1/2 after treatment with PGJ₂ and 15d-PGJ₂ was assessed by monitoring the phosphorylation status of these proteins by Western blot. Phosphorylation of TAK1, NF-κB p65, p38 MAPK, and ERK-1 in M1-MDMs was increased 15 min after stimulation with MTB-CM, peaking at 30 min, without alteration of the amounts of the total proteins, respectively (Fig. 4A, 4B). Pretreatment of M1-MDMs with 10 µM PGJ₂ or 15d-PGJ₂ clearly reduced these stimulatory effects of MTB-CM with similar efficiencies (Fig. 4A, 4B). These data suggest that PGJ₂ and 15d-PGJ₂ may prevent MTB-CM–induced COX-2 expression in M1-MDM by interference with the TAK1/NF-κB/MAPK pathways.

Finally, we studied which receptors of PGJ₂ and 15d-PGJ₂ are involved in the prevention of MTB-CM–induced COX-2 expression with PPARγ, DP1, and CRTH2/DP2 as likely candidates. M1-MDMs were pretreated with the PPARγ antagonist GW9662, the DP1 antagonist MK0524, or the CRTH2/DP2 antagonist BAY u3405 (10 µM each). GW9662 failed to reverse the inhibitory effects of PGJ₂ and 15d-PGJ₂ (5 µM, each) on MTB-CM–induced COX-2 expression (Fig. 5A). However, MK0524 and BAY u3405 partially restored the strong COX-2 expression, thus reversing the inhibitory effects of PGJ₂ and 15d-PGJ₂ (Fig. 5B). These results suggest that the negative feedback loop on COX-2 expression and related PG formation by 15d-PGJ₂ or PGJ₂ depend, at least partially, on DP1 and/or CRTH2/DP2 receptors, whereas PPARγ is not involved in this respect.

In this study, we employed metabololipidomics for comprehensive LM profiling of human MDMs that were stimulated by secreted factors from M. tuberculosis using MTB-CM. Besides the expected massive generation of PGE₂ upon stimulation with MTB-CM, along with elevation of COX-2 expression (28), we surprisingly detected high amounts of PGJ₂ produced by M1-MDMs, with minor effects in M2-MDMs. Due to the superior response of M1-MDMs for PGJ₂ formation, we focused on this macrophage phenotype for functional and more detailed analysis without comparison with M2-MDMs, which is currently a limitation of the study and might be subject for future investigations. We found that PGJ₂ but also the related PGD₂ and 15d-PGJ₂, known to act via DP1 and CRTH2/DP2 receptors, dampened the phagocytic ability of macrophages infected by M. tuberculosis and promoted the intracellular survival of bacteria. Our data indicate that J₂-PGs induce a negative feedback loop via DP1 and/or CRTH2/DP2 receptors to counter-regulate the robust M. tuberculosis–induced upregulation of COX-2 expression and PG formation, along with impaired release of the inflammatory cytokines IL-1β, IL-6, and IFN-γ but elevated expression of anti-inflammatory markers IL-10, Arg-1, and CD163. This suggests a proresolving effect of these J₂-PGs in a proinflammatory environment, which may negatively impact the anti–M. tuberculosis ability of macrophages.

PGs are COX-derived LMs with mainly proinflammatory activities and accordingly, nonsteroidal anti-inflammatory drugs that inhibit COX enzymes are used to treat inflammatory diseases and pain (10, 51). M. tuberculosis infection of macrophages induced COX-2 expression and related PG formation (18, 28), where PGE₂ plays a critical role in inhibition of M. tuberculosis replication (18), and massive PGE₂ release was clearly confirmed in M1-MDMs and M2-MDMs in our study. It was shown before that COX-2–mediated PG production modulates macrophage immune activation and is involved in intracellular mycobactericidal activity (17). Virulent M. tuberculosis inhibits macrophage apoptosis and promotes necrosis by inhibiting PGE₂ production, an environment that favors M. tuberculosis growth (18, 52). Of interest, clinically approved drugs that augment PGE₂ levels prevented acute mortality of M. tuberculosis–infected mice (17). Together, the current data imply key roles for PGE₂ among PGs in the host macrophage defense against M. tuberculosis.

COX-2 and mPGES-1 typically dominate in M1- over M2-like macrophages with massive PGE₂ production by M1-like cells (29, 35). In line with our previous study (28), in this study, we found that MTB-CM strongly potentiated COX-2 expression and PGE₂ formation in human M1-MDM, but also in M2-MDM, which was not observed before. But also the unexpected strong upregulation of PGJ₂ in M1-MDM in response to MTB-CM was yet unknown and implies biological functions of this PG with respect of M. tuberculosis infection. Interestingly, in M1-MDMs, PGJ₂ was the second most abundant PG next to PGE₂, being detectable in much higher amounts than its precursor PGD₂. In M2-MDMs, however, the levels of MTB-CM–induced PGJ₂ were >20-fold lower versus those in M1-MDMs, despite PGD₂ being only 2-fold depressed, suggesting that the M1 phenotype may play a superior role over M2 macrophages in providing PGJ₂ for potentially anti-inflammatory functions.

J₂-PGs and their precursor PGD₂ stimulate anti-inflammatory activities (11, 16, 21, 23). However, the functions of PGD₂ and J₂-PGs in M. tuberculosis–infected macrophages are largely unexplored. Our data reveal that 15d-PGJ₂, PGJ₂, and PGD₂ compromise the macrophage phagocytotic efficiency connected to increased intracellular M. tuberculosis survival, with ranked potency in the given order. Similarly, J₂-PGs markedly inhibited the release of major proinflammatory cytokines (IL-6, IL-1β, and IFN-γ) along with elevated levels of anti-inflammatory markers (IL-10, Arg-1, and CD163) in M1-MDMs, thereby impairing the inflammatory status of these cells, which again may favor M. tuberculosis survival (41, 44). Most intriguingly, however, is our discovery of the J₂-PG–induced decrease of COX-2 expression and PG formation, implying a negative feedback loop. We propose that the rapid and strong induction of COX-2 expression and PG formation in MDMs caused by MTB-CM is self-limited because of the de novo–generated J₂-PGs that dampen these responses. Indeed, COX-2 protein expression peaked at 4–12 h after MTB-CM challenge but then declined up to 48 h possibly because of such negative feedback by elevated J₂-PGs. This anti-inflammatory feedback also applies to regulation of proinflammatory cytokines (downregulated) and anti-inflammatory markers (upregulated) in M1-MDMs and ensures limitation of inflammatory reactions.

Besides J₂-PGs, other anti-inflammatory LMs, such as specialized proresolving mediators (SPMs), exhibit specific but divergent functions in M. tuberculosis–infected macrophages. Lipoxin A₄, induced by virulent M. tuberculosis, promoted macrophage necrosis resulting in dissemination of bacteria (18). It was suggested that alterations of the phenotype from proinflammatory M1 that substantially produce PGE₂ to SPM-producing M2 phenotypes might be one of the reasons for the increased bacterial survival rate (6). In our present study, MTB-CM elevated the M2 marker CD206 during M1 polarization. In contrast, resolvins (e.g., RvD1) and maresin-1 exerted antimycobacterial effects in macrophages by exciting the production of antimicrobial peptides (53). SPMs are generally formed in greater amounts in M2-MDMs than M1-MDMs (29), but our results show that in contrast with J₂-PGs, the formation of SPMs is not increased upon M. tuberculosis infection in either MDM phenotype.

In the innate immune system, the macrophage is the main cell type to phagocytize, control, and recognize M. tuberculosis (54). PGD₂ production can be triggered by allergens during allergic inflammation, mainly in mast cells, after activation of DP1 and CRTH2/DP2, which results in the recruitment of immune cells (55). PGJ₂ and 15d-PGJ₂ activate both DP1 (56) and CRTH2/DH2 (55), which triggers potent anti-inflammatory activities (57–59). Previous studies showed that PGs, such as PGE₂ and PGD₂, impair phagocytosis of macrophages in obesity at low concentrations (i.e., 100 nM), via EP2 and DP1 pathways, respectively (60). Moreover, macrophages treated with the COX-2 inhibitor celecoxib or an EP2 antagonist showed higher M. tuberculosis phagocytosis versus untreated controls (61), and also depletion of DP1 augments the phagocytic capacity of macrophages (62). These findings support the hypothesis that PGJ₂ and 15d-PGJ₂ dampen macrophage phagocytosis of M. tuberculosis through DP1 and CRTH2/DH2 signaling. In fact, the results from the analysis of the involved receptors and signaling pathways indicate that DP1 and CRTH2/DP2, but not PPARγ, are operative as receptors of PGJ₂ mediating the suppressive effects on COX-2 expression. Thus, both the DP1 antagonist MK-0524 and the CRTH2/DP2 antagonist BAY-u3405 reversed the PGJ₂- and 15d-PGJ₂–induced downregulation of COX-2 expression, suggesting that DP1 and CRTH2/DP2 mediate the suppressive effects of J₂-PGs in MTB-CM–activated M1-MDMs. Nevertheless, 15d-PGJ₂, produced in macrophages, activated PPARγ, which resulted in inhibition of proinflammatory cytokines release and improved the survival rate of septic mice (63). In our hands, the PPARγ antagonist GW9662 could not restore the negative feedback loop of COX-2 expression caused by J₂-PGs, suggesting that PGJ₂ and 15d-PGJ₂ operate independent of PPARγ in MDMs.

The MTB-CM–induced COX-2 expression in M1-MDMs correlated with activation of TAK-1/NF-κB/p38 MAPK and ERK, and accordingly, the COX-2–repressive effects of PGJ₂ impaired the activation of these signaling proteins, suggesting their involvement. M. tuberculosis induced COX-2 expression in immune cells involving NF-κB, ERK1/2, and p38 MAPK signaling (19, 44–46). Moreover, 15d-PGJ₂ was proposed to counteract various proinflammatory responses in different cell types through interference with NF-κB (16, 23), MAPK p38 (48), and ERK1/2 (49). Finally, TAK1 was shown to regulate NF-κB and MAPK upon activation by M. tuberculosis (50). Therefore, induction of COX-2 by MTB-CM and its counterregulation by PGJ₂ both correlate with the activation states of TAK-1/NF-κB/p38 MAPK and ERK and provide a reasonable interrelation. In contrast, in the murine macrophage cell line RAW264.7, 15d-PGJ₂ directly inhibited NF-κB–dependent COX-2 expression through covalent modifications of critical cysteine residues in IκB kinase and in the DNA-binding domains of NF-κB subunits, again independent of PPARγ (47).

In conclusion, we here revealed robust elevation of J₂-PGs via inducible COX-2 in human M1-MDMs upon challenge by secreted factors of M. tuberculosis, which may function as endogenous suppressors of proinflammatory reactions in the macrophages. This negative feedback loop of J₂-PGs is seemingly mediated by DP1/CRTH2 receptors, independent of PPARγ, and correlates to the activation of TAK1/NF-κB p65/p38MAPK and ERK1/2 pathways that potentially trigger the suppression of COX-2 expression and proinflammatory cytokines but may elevate the levels of anti-inflammatory markers.

The authors have no financial conflicts of interest.

We thank Alrun Schumann und Anna König for expert technical assistance.