Deuterated and non-deuterated LM standards for ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS-MS) quantification were purchased from Cayman Chemical/Biomol GmbH (Hamburg, Germany). SASP and all reagents were obtained from Sigma-Aldrich (Taufkirchen, Germany) unless mentioned otherwise.

Plasma samples were collected from human subjects at the Shenzhen Third People’s Hospital. The experimental protocol was approved by the ethical committee of Shenzhen Third People’s Hospital. Diagnosis of active TB in these subjects was based on chest radiography, clinical symptoms, and microscopy for acid fast bacilli (AFB), sputum and/or BALF M.tb culture and the response to anti-TB chemotherapy. Healthy controls (HC) with normal chest radiographic outcome and without clinical history of TB were recruited. M.tb-specific interferon-γ release assays (IGRA) were used to differentiate individuals with latent TB infection (LTBI) from HC without an infection. The non-TB lung diseases group included patients with pneumonia (PN). Plasma samples were stored at −80°C. Unless otherwise indicated, the samples from patients were collected prior to initiation of antibiotic treatment.

Leukocyte concentrates from freshly withdrawn peripheral blood of male and female healthy adult volunteers were provided by the Institute of Transfusion Medicine at the University Hospital Jena (Jena, Germany). The experimental protocol was approved by the ethical committee of the University Hospital Jena. All methods were performed in accordance with the relevant guidelines and regulations. Polymorphonuclear leukocytes (PMNL), peripheral blood monocytic cells (PBMC) and platelets were separated using dextran sedimentation, followed by centrifugation on lymphocyte separation medium (Histopaque-1077) as reported elsewhere by us (18). For isolation of monocytes, PBMC were resuspended in RPMI 1640 containing 10% (v/v) heat-inactivated fetal calf serum (FCS), 100 U/ml penicillin and 100 µg/ml streptomycin in cell culture flasks (Greiner Bio-One, Frickenhausen, Germany) for 1.5 h at 37°C and 5% CO₂ for adherence of monocytes. For differentiation of the monocytes to MDM and further polarization towards an M1-like phenotype, published criteria were used (19). M1-MDM were generated by incubating monocytes with 20 ng/ml GM-CSF (PeproTech, Hamburg, Germany) for 6 d in RPMI 1640 supplemented with 10% FCS, 2 mmol/L glutamine (Biochrom/Merck, Berlin, Germany), and penicillin–streptomycin (Biochrom/Merck) to obtain M0_GM-CSFMDM (briefly “MDM”), followed by treatment with 100 ng/ml lipopolysaccharide (LPS) and 20 ng/ml IFNγ (PeproTech) to obtain pro-inflammatory M1-like MDM. Routinely, MDM were polarized for 48 h unless stated otherwise.

Live-dead reporter M.tb strain H37Ra constitutively expresses Emerald and a tetracycline inducible tgRFP. Bacteria were cultured in 0.5 mg/ml Hygromycin 7H9 media (BD Biosciences, San Jose, CA, USA) at 37°C in log phase, washed, sonicated, and passed through 0.45 μm Millex filter to obtain a single cell suspension. For experiments involving induction of tgRFP fluorescence expression in transcriptionally active bacteria, anhydrotetracycline (200 ng/ml) was added for 24 h.

MDM were infected with M.tb strain H37Ra at a MOI of 5. After 4 h incubation, non-internalized bacteria were washed away from MDM with PBS and incubation with RPMI 1640 medium. For colony-forming unit (CFU) assays, cells were incubated for 72 h and then lysed with 0.1% SDS. Cell lysates were diluted, plated on 7H11 Middlebrooks agar plates, and incubated at 37°C and 5% CO₂. CFU were counted 3 to 4 weeks later. For gene expression assay, cells were harvested at 24 h after infection and subjected to RNA isolation. For flow cytometry, cells were harvested at 72 h and were subjected to fluorescence-activated cell sorting (FACS) for further analysis. In some experiments, cells were pretreated with SASP for 1 h.

M.tb (H37Rv) was inoculated in Middlebrook 7H9 broth with glycerol (0.2%) and OADC supplement (Merck, Germany). The cultures were incubated at 37°C for 2–3 months 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 not inoculated broth was sterile filtrated and used in experiments.

For each MDM subgroup, 200 homogenous cells were pooled for RNA extraction, reverse transcribed and amplified with QIAseq FX Single Cell RNA Library Kit (Qiagen, Gemany). Libraries were built for the cDNA generated from each group and sequenced using Illumina X Ten platform with paired-end 150 bp read length. Reads were filtered for the ones with low quality and adapters were mapped to the human reference genome hg19 (GRCh37) with STAR 2.6 (20). Raw reads were counted for each uniquely mapped gene with featureCounts (21). With a negative binomial distribution-based model adopted by edgeR, the gene expression levels were normalized among replicates and between groups, and logarithmic transformation of normalized count per million (logCPM) for each gene was used for further comparative analysis. Gene expression was compared between groups using edgeR with quasi-likelihood F-tests (22). The significance level was preset as a multi-testing corrected False Discovery Rate (FDR) of 0.05. Sequencing data can be found at GenBank, accession number PRJNA817162.

Pro-inflammatory M1-like MDM (2×10⁶/ml) were incubated in PBS containing 1 mM CaCl₂. M.tb (H37Rv strain)-conditioned medium (MTB-CM, 1%) or the 19-kD antigen were added for 4 h at 37°C. Signs for apoptotic or cytotoxic effects of MTB-CM or p19 antigen (number and integrity of cells routinely counted after treatment with MTB-CM or 19-kD antigen by light microscopy and trypan blue staining) were not obvious. The supernatants were then transferred to 2 ml of ice-cold methanol containing 10 ml of deuterium-labeled internal standards (200 nM d8-5S-hydroxyeicosatetraenoic acid, d4-leukotriene B₄, d5-lipoxin A₄, d5-resolvin D2, d4-PGE₂ and 10 mM d8–arachidonic acid [AA]) to accomplish quantification and sample recovery. Sample preparation was conducted by adapting published criteria (23, 24). Briefly, samples were kept at -20°C for 60 min to allow protein precipitation. After centrifugation at 1200×g and 4°C for 10 min, 8 ml acidified water was added (final pH = 3.5), and the samples were subjected to solid phase extraction. Solid phase cartridges (Sep-Pak Vac6cc 500 mg/6 ml C18; Waters, Milford, MA) were first equilibrated with 6 ml methanol and 2 ml water prior to sample loading onto columns. After washing with 6 ml water and additional 6 ml n-hexane, LM were eluted with 6 ml methyl formate. Finally, the samples were brought to dryness using an evaporation system (TurboVap LV; Biotage, Uppsala, Sweden) and resuspended in 100 µl methanol–water (50/50, v/v) for UPLC–MS-MS automated injections. LM profiling was analyzed with an Acquity UPLC system (Waters) and a QTRAP 5500 Mass Spectrometer (AB Sciex, Darmstadt, Germany) equipped with a Turbo V Source and electrospray ionization. LM were eluted using an Acquity UPLC BEH C18 column (1.7 µm, 2.1 mm x 100 mm; Waters, Eschborn, Germany) at 50°C with a flowrate 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. 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 (i.e., collision energy, entrance potential, declustering potential, collision cell exits potential) were adopted (23, 24). The curtain gas pressure was set to 35 psi. The retention time and at least six diagnostic ions for each LM were confirmed by means of external standards (Cayman Chemical/Biomol GmbH, Hamburg, Germany). Quantification was achieved by calibration curves for each LM.

Cell lysates of M1-MDM, corresponding to 2 × 10⁶ cells, were separated on 10% polyacrylamide gels, and blotted onto nitrocellulose membranes (Amersham Protran Supported 0.45 μm nitrocellulose, GE Healthcare, Freiburg, Germany). The membranes were then incubated with the following primary antibodies: rabbit polyclonal anti-COX-1, 1:1000 (4841S, Cell Signaling); rabbit polyclonal anti-COX-2, 1:1000 (4842S, Cell Signaling); rabbit monoclonal anti-ERK1/2, 1:1000 (4695S, Cell Signaling); mouse monoclonal anti-phospho-ERK1/2 (Thr202/Tyr204), 1:1000 (9106, Cell Signaling); 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 (Ser536), 1:1000 (13346S, Cell Signaling); mouse monoclonal anti-β-actin,1:1000 (3700S, Cell Signaling); rabbit polyclonal anti-β-actin, 1:1000 (4967S, 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, LICOR Biosciences), and visualized by an Odyssey infrared imager (LI-COR Biosciences). Data from densitometric analysis were background-corrected.

Extraction of total RNA was performed with the RNAeasy Minikit (Qiagen). The cDNA was synthesized using an oligo-dT primer and SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA). Gene expression was assessed using SYBR Green-based real-time quantitative PCR as previously described (17). Data were calculated by the 2−ΔΔCT method using β-actin as house-keeping gene.

Results are given as mean ± S.E.M. of n observations, where n represents the number of experiments with different donors performed on different days as indicated. Analyses of data were performed using GraphPad Prism 8 software (San Diego, CA). Paired t-test was used for comparison of two groups. For multiple comparisons, ANOVA with Bonferroni or Dunnett´s post hoc tests were applied as indicated. The criterion for statistically significant is p < 0.05.

To characterize outcomes of individual M.tb-macrophage interactions, we infected M0_GM-CSFMDM (briefly “MDM”) with the M.tb H37Ra strain carrying an established fluorescence dilution plasmid that reports the bacterial status (live or dead) after internalization. This plasmid provides two different fluorescent protein markers: a constitutively expressed (Emerald, green) and an anhydrotetracycline-inducible (tag-RFP, red) fluorescent protein yielding three distinct outcomes: (1) infection of the MDM with intracellular survival of live M.tb, (2) infection resulting in intracellular dead bacteria, and (3) no infection. While MDM with live bacteria display both red and green fluorescence, MDM with dead bacteria display only green fluoresce, and exposed but uninfected MDM display no fluorescence ( Figure 1A ). Importantly, using these reporters it is possible to distinguish MDM that had been initially infected but cleared the bacterium (Emerald+, tagRFP–) from cells that had never been infected (Emerald-, tagRFP–) designated as bystanders. We used fluorescence-activated cell sorting (FACS) to reveal the diverse phenotypes as well as to isolate uninfected MDM and MDM infected with live or dead bacteria ( Figure 1B ). The gating strategy enabled the selective isolation of MDM subgroups confirmed by CFU assays ( Supplementary Figure S1A ).

We then subjected 200 isolated MDM of each subgroup to RNA-seq analysis. With the single-cell-like RNA-seq library preparing strategy, the gene expression of few subgroup cells could be well profiled. All of the subgroups in each repeat had more than or nearly 10 million reads that could be uniquely mapped to human genomes ( Supplementary Figure S2A ). The genes covered for each subgroup with > 1 count per million reads (CPM) were around 11,000 and quite even among subgroups or repeats.

MDM bystanders, MDM with live bacteria, and MDM with dead bacteria were compared with a fourth group that are non-exposed MDM, independently. 332 and 249 genes were detected with higher expression levels in the bystander subgroup and the live bacteria subgroup, while 36 and 37 genes showed decreased expression, respectively (FDR < 0.05, fold change > 3). MDM with dead bacteria, however, showed fewer genes expressed with significant differences to non-exposed cells, with only 33 genes being up-regulated and 11 genes down-regulated ( Figure 2A ). 23 genes in individual exposed MDM subgroups versus non-exposed ones showed always higher expression while 2 genes showed decreased levels in bacteria-exposed cells ( Figures 2A, B ).

We next compared the transcriptome profile of MDM exposed to intracellular live and dead bacteria. Totally, 137 genes were differentially expressed between the two subgroups, with a majority (125/137, 91.2%) that were up-regulated and only a small subset (12/137, 8.8%) that were down-regulated in MDM with live bacteria ( Figure 2C ). Many up-regulated genes in MDM with live bacteria are involved in immune response processes including the genes encoding COX-2 and mPGES-1 (PTGES) which are the key enzymes in PGE₂ biosynthesis ( Figure 2D ). qRT-PCR experiments confirmed higher expression levels of COX-2 and mPGES-1 in MDM with live bacteria ( Figures 2E, F ). These data suggest that secreted factors released from M.tb increase the expression pro-inflammatory genes such as COX-2 and mPGES-1 in MDM leading to formation of pro-inflammatory PGE₂ which might be one mechanism by which bacteria escape intracellular killing during the process of infection.

Based on the results from above, we hypothesized that elevated PGE₂ in the plasma of TB patients may correlate with the severity of the disease. We analyzed the PGE₂ plasma levels in patients with either latent TB infection (LTBI), pulmonary tuberculosis (TB) or in patients suffering from non-TB lung disease (non-TB) as well as in healthy donors (HD). The concentration of PGE₂ in the plasma of TB patients was significantly higher versus healthy cohorts or patients with non-TB lung diseases or latent TB infection ( Figure 3A ). Moreover, plasma PGE levels in active TB patients were assessed for 12 months after initiation of anti-TB chemotherapy. As shown in Figure 3B , plasma PGE₂ levels decreased significantly over the first three months of TB chemotherapy with consistent impairment up to 12 months.

It appeared reasonable that the elevated expression of COX-2 and mPGES-1 as well as the corresponding PGE₂ levels in MDM are caused by secreted factors released from M.tb. In fact, several pathogenic bacteria secrete exotoxins or other factors that act on host cells to provoke PGE₂ and related lipid mediator formation, such as S. aureus-derived α-hemolysin (25), lipoteichoic acid (26) or phenol-soluble modulins (27). Several M.tb secreted effector molecules are known to modify host cell pathways such as phagosome maturation, cell death, cytokine response, xenophagy, reactive oxygen species response via NADPH oxidase 2, nitric oxide response via NO synthase 2 and antigen presentation via MHC class I and class II molecules (28). Thus, we hypothesized that also for M.tb such secreted factors may be causative for the induction of COX-2. We thus studied if M.tb-conditioned medium (MTB-CM) would be capable of inducing COX product formation on the cellular level in vitro and also addressed other related LM branches such as 5-LOX- and 12/15-LOX-derived products in MDM and for comparison in relevant immune or blood cells. Thus, human M1-MDM and freshly isolated monocytes, PMNL, and platelets from human peripheral blood of healthy volunteers were stimulated with MTB-CM for 4 h and metabololipidomics using UPLC-MS-MS (23, 25) was performed to assess the respective LM signature profiles. Exposure of M1-MDM to MTB-CM caused strong formation of COX-derived prostanoids (PGD₂, PGE₂, PGF₂α, TXB₂, 11-HETE and 11-HEPE), most prominently PGE₂ ( Figure 4A ), but not upon exposure of PMNL, monocytes or platelets ( Figure 4B ; Supplementary Table 1 ). Nevertheless, in PMNL, a minor increase in the formation of 5-LOX-derived products (e.g., LTB₄, t-LTB₄ and 5-HETE) and other LOX-derived LM like 17-HDHA, 14-HDHA, 12-HETE, 12-HEPE or 4-HDHA was apparent ( Supplementary Table 1 ).

We found before that mice treated with SASP had significantly reduced M.tb bacterial load in the lung and other organs (17). We hypothesized that SASP-mediated decrease of M.tb bacterial loads could be due to interference of the drug with COX-2/mPGES-1 and concomitantly lowered PGE₂ formation. Thus, we pre-incubated the M1-MDM with 200 µM SASP prior to exposure to MTB-CM and measured LM formation after 4 h. MTT-based cytotoxicity assays indicated no significant decrease of MDM viability by treatment with 200 µM (data not shown), which is in line with result from our previous report (17). The MTB-CM-induced release of PGE₂ and of other COX-derived prostanoids into the supernatant of M1-MDM was efficiently suppressed by SASP ( Figures 4C, D ), while LM generated by LOX pathways were hardly affected ( Figure 4C ). Note that also the MTB-CM-induced release of pro-inflammatory cytokines such as TNF-a, IL-1β and IL-6 from M1-MDM was significantly reduced by SASP pretreatment ( Figure 4E ).

The 19-kD antigen was previously identified as a major virulence factor of M.tb (29) and it appeared possible that this antigen is, at least in part, causative for the MTB-CM-induced COX product formation in MDM. We thus exposed the M1-MDM to the isolated 19-kD antigen (2 µg/ml) and analyzed LM formation after 4 h. In line with the results obtained from M1-MDM stimulated by MTB-CM, high levels of COX products, with PGE₂ as most abundant LM, were evident in the supernatants of 19-kD antigen-activated M1-MDM, and this was clearly reduced when cells were pretreated with SASP ( Figures 5A, B ). In contrast, the levels of PGE₂ and other COX products were not elevated when M1-MDM were exposed to the early secretory antigenic target 6 (ESAT6) ( Supplementary Table 4 ), which is another important virulence factor of M.tb (30). Together, MTB-CM and the 19-kD antigen induce formation of PGE₂ and other COX products in M1-MDM, which is efficiently blocked by SASP, with minor effects on LOX-derived LM formation.

Our results prompted us to investigate if secreted factors from M.tb would increase the expression of PGE₂-biosynthetic enzymes in human M1-MDM and if this is counteracted by SASP. First, we found that in M1-MDM the expression of COX-2 protein was time-dependently induced by MTB-CM treatment, which peaked after 2 h and decreased again after 4 h ( Figures 6A, B ). Of interest, pretreatment of M1-MDM with SASP impaired COX-2 protein levels after 4 h without effects at earlier (2 h) time points. The expression of COX-1 protein slightly increased after exposure to MTB-CM whereas the level of mPGES-1 did not change. Pre-treatment with SASP did not influence the expression of either COX-1 or mPGES-1 ( Figures 6A, B ).

COX-2 expression in general is governed by different signaling routes and protein kinases including the MKK3/6-p38MAPK, MEK-ERK-1/2, and NF-κB pathways (31–34). We assumed that elevated COX-2 protein levels due to MTB-CM may involve some of these pathways that in turn would be affected by SASP. In M1-MDM exposed to MTB-CM for 30 min, the phosphorylation of NF-κB p65 was significantly increased, while treatment for 10 or 60 min did not alter NF-κB p65 phosphorylation. The phosphorylation of p38MAPK or ERK-1/2 did not change after exposure to MTB-CM. Pre-incubation of M1-MDM with SASP for 1 h and subsequent short-term treatment with MTB-CM for 30 min significantly decreased the phosphorylation of NF-κB p65 ( Figures 6C, D ). These data suggest that the elevated COX-2 expression due to MTB-CM could be conferred through activation of the NF-κB pathway and that SASP may mediate its suppressive effects by blocking NF-κB signaling.