This study was carried out in strict accordance with the guidelines of the Council of the European Union (Directive 86/609/EEC) regarding the protection of animals used for experimental and other scientific purposes. The experimental protocol was approved by the institutional animal care and use committee of the Aix-Marseille University (N° 87–848, October 19, 1987, Paris, modified by Decree 2001-464 and Decree 2001-131 relative to European Convention, EEC directive 86/609). All experiments were done in accordance with French and European guidelines for animal care. Blood was obtained from the Etablissement Français du Sang. As it is a commercial product, the Ethics Committee agreement is not required.

Six to eight weeks old females C57BL/6 (Charles River Laboratories) were maintained under pathogen-free conditions and fed sterile food and water ad libitum. All infections were performed under Animal Biosafety Level 3 facility.

Orentia tsutsugamushi strain Kato (CSUR R163) (gift from Dr. Jean-Louis Mège, URMITE, Marseille, France) was propagated in L929 cells (ATCC), as previously described (Barry et al., 2012). Briefly, highly infected cells were sonicated, or lysed using glass beads for O. tsutsugamushi, and centrifuged at 500 × g for 10 min to remove cell debris. Supernatants were collected and centrifuged at 10,000 × g for 10 min to pellet the bacteria. Live bacteria were quantified using infected cell-counting units. Coxiella burnetii organisms (RSA493 Nile Mile strain) (gift from Dr. Jean-Louis Mège, URMITE, Marseille, France) were obtained by passage in BALB/c mice (Charles River Laboratories) and culture in L929 cells. The concentration of organisms was determined by immunofluorescence using specific antibodies and/or PCR using known concentrations of bacterial DNA. The quantification of organisms was performed as previously described (Barry et al., 2012). Bacterial viability was assessed using the LIVE/DEAD Bac Light bacterial viability kit (Invitrogen). Brucella abortus strain 2308 and Brucella melitensis strain 16M (gift from Dr Ignacio Moriyon, University of Pamplona, Spain) were grown on TSA (Sigma-Aldrich) at 37°C for 7 days, as previously described (Pizarro-Cerdá et al., 1998). The Twist-Marseille (CNCM I-2202) strain of Tropheryma whipplei (gift from Dr. Jean-Louis Mège, URMITE, Marseille, France), a bacterium known to live in macrophages, was cultured in HEL cells and purified. Quantification of inocula was performed by measuring the percentage of infected cells by immunofluorescence, as previously described (Gorvel et al., 2010).

Peripheral blood mononuclear cells from buffy coats were recovered from the Ficoll-Hypaque (GE healthcare) interface after centrifugation at 700 × g for 20 min. Monocytes and T lymphocytes were isolated from PBMCs using magnetic beads coupled to antibodies specific for CD14 and CD3, respectively, as described by the manufacturer (Miltenyi Biotec). The purity of the monocyte and T-cell preparations was assessed by flow cytometry and was greater than 98%. Monocytes were then incubated in RPMI 1640 (Gibco) containing 20 mM HEPES, 2 mM glutamine, 10% fetal calf serum (FCS) (Invitrogen), 0.1 ng/mL IL-4 and 1 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF) (R&D Systems) for 7 days to obtain moDC.

Bone marrow-derived dendritic cells were prepared from 6 to 8 week-old females C57BL/6 mice as previously described (Salcedo et al., 2008). Briefly, tibias and femurs were flushed with RPMI-5%FCS (Eurobio)-50 μM 2β-mercaptoethanol (Sigma-Aldrich) to extract bone marrow.

In order to generate GM-CSF-BMDC, 3.10⁶ cells per well were seeded onto 6-well plates in medium supplemented with supernatant of J558L GM-CSF producing cell line (gift from Dr. Philippe Pierre, Centre d’Immunologie de Marseille-Luminy, Marseille, France). The medium was changed at day 2.5.

GM-CSF-DC were infected with B. abortus smooth virulent strain 2308 using a MOI of 30:1. Bacteria were centrifuged onto cells at 400 × g for 10 min at 4°C. Then, cells were incubated for 30 min at 37°C with 5% CO₂. After two washes with medium, cells were incubated for one hour with medium containing antibiotic [gentamicin (Sigma-Aldrich) at 50 μg/ml] in order to kill the extracellular bacteria. Thereafter, the antibiotic concentration was decreased at 10 μg/ml until the end of the infection time.

Brucella melitensis 16M with a chromosomal integration of DsRed was used in this study. B. melitensis were grown under shaking in TSB (Sigma-Aldrich) containing 25 μg/ml kanamycin (Sigma-Aldrich) at 200 rpm for 16 h at 37°C. Inocula were prepared by pelleting bacteria and by adjusting bacterial concentration in endotoxin free PBS (Gibco) to obtain the right concentration depending on the inoculation route. Oral infection was performed as previously described (von Bargen et al., 2014). Briefly, pelleted bacteria were re-suspended to previously heat-sterilized 5% milk and adjusted at respective MOI in 5% milk (Scharlau). For all the other inoculation routes, mice were anesthetized with Ketamine/Xylazine (Virbac). For intranasal infection, the bacterial inoculum containing 10⁵ CFU suspended in 35 μL of endotoxin free PBS (Gibco) was applied onto mice nostril via pipet. For intradermal infection, the bacterial inoculum containing 10⁴ CFU suspended in 10 μL of endotoxin free PBS was injected into the ear pinna. For conjunctival infection, the bacterial inoculum containing 10⁹ CFU suspended in 10 μL was applied onto conjunctival surfaces (5 μL per eye). In some experiments, the selective COX-2 inhibitor NS-398 (Cayman Chemicals) were administered by intraperitoneal injection (15 mg/kg) right after intradermal infection. A stock solution of NS-398 was prepared in DMSO (Sigma-Aldrich) and was diluted in PBS just prior to treatment to the right concentration in a 1:3 (DMSO:PBS) ratio for a total volume of 100 μL per mouse or vehicle control (DMSO:PBS alone).

Treatments were repeated daily for 7 days. At 8 and 29 days post-infection mice were sacrificed by cervical dislocation. Immediately after sacrifice, CLN (including mandibular, accessory mandibular, and superficial parotid lymph nodes referred here as CLN; Van den Broeck et al., 2006) and spleens were collected, weighted, and proceeded for bacterial count and RNA extraction.

Spleens and CLN were removed and homogenized in Triton X-100 (Sigma-Aldrich) 0.1%. Serial dilutions were performed in PBS, and plated in triplicate onto TSA agar. CFU were enumerated after 3-4 days of culturing at 37°C.

Monocyte derived dendritic cells (5.10⁶ cells per well) were plated in 6-well plates and stimulated or not with B. abortus (MOI of 300:1) for 6 hours, and total RNA was then extracted using the RNeasy minikit (Qiagen) and DNase treatment. This study utilized the 4X44k Human Whole Genome microarrays (Agilent Technologies), representing 44,000 probes. Reverse transcription, samples labeling, and hybridization were performed according to the protocols specified by the manufacturer (One-Color Microarray-Based Gene Expression Analysis). Three samples per experimental condition were included in the analysis. The slides were scanned at a 2-mm resolution with a G2505C DNA microarray scanner (Agilent Technologies).

Total RNAs from infected GM-CSF-BMDC were extracted using RNeasy Micro Kit (Qiagen) following manufacturer’s instructions. CLN were stabilized in RNAlater (Qiagen) immediately after sampling. Organs were homogenized in RLT buffer (Qiagen) and then extracted using RNeasy Micro Kit (Qiagen) following manufacturer’s instructions. cDNAs were obtained by using Quantitech Reverse Transcription Kit (Qiagen) following manufacturer’s instructions using 300 ng of RNA as a matrix. qPCR were conducted using a 7500 Fast-Real-time PCR (Applied Biosystem) with SYBER Green (Takara) following manufacturer’s instructions. HPRT was used as housekeeping gene to determine ΔCt. Fold change compared to base line expression in uninfected cells, or control mice was determined using 2^-ΔΔCt method where ΔΔCt = (Ct_target-Ct_HPRT)_infected-(Ct_target-Ct_HPRT)_non-infected as previously described (Papadopoulos et al., 2016). mRNAs whose expression level was twice as high compared to control were considered as significantly up-regulated. Primers used in this study are listed in Table 1.

Monocyte derived dendritic cells were stimulated by bacterial pathogens (MOI 50:1 for T. whipplei or 20:1 for others) or Escherichia coli LPS (Sigma-Aldrich) (1 μg/ml) for 16 h, suspended in 0.3 mL of cold methanol (MeOH). Fatty acid methyl esters (FAME) extraction was performed as previously described (Schutter and Dick, 2000; Ecker et al., 2012), 15 μL of MeOH/Deuterium-labeled eicosanoids (LTB4-d4 and 5-HETE-d8, 400 ng/ml) standard solution were added to 0.2 ml of homogenate. After centrifugation, supernatants were diluted in 10 mL of hydrochloric acid (HCl) 0.02 M and submitted to solid-phase extraction on C18 cartridge (Macherey Nagel). After complete loading, columns were washed, dried and lipid mediators were eluted in methyl formate (MeFor). Solvent was evaporated under N2 and samples were dissolved within 30 μL MeOH and stored at -80°C.

For AA quantification, 0.1 ml of homogenates were extracted in dichloromethane/methanol/water, in the presence of 2 μg glyceryl triheptadecanoate (Sigma-Aldrich) and hydrolyzed in potassium hydroxyde (KOH) (0.5 M in MeOH) at 50°C for 30 min. Lipids were transmethylated in 14% boron trifluoride MeOH solution (Sigma-Aldrich) and hexane at 80°C for 1 h. FAME were extracted with hexane, dried and dissolved in ethyl acetate.

Fatty acid methyl esters were analyzed by gas-liquid chromatography as previously described (Lepage and Roy, 1986) on a Clarus 600 Perkin Elmer system using Famewax RESTEK fused silica capillary columns (30 m × 0.32 mm i.d, 0.25 μm film thickness). Oven temperature was programmed from 110°C to 220°C at a rate of 2°C per min and the carrier gas was hydrogen (0.5 bar). The injector and the detector were at 225°C and 245°C, respectively. LC-MS/MS analysis was performed on UHPLC system (Agilent LC1290 Infinity) coupled to Agilent 6460 triple quadrupole MS (Agilent Technologies) equipped with electrospray ionization source. Separation was done at 40°C on a Zorbax SB-C18 column (2.1 mm, 50 mm, 1.8 μm) (Agilent Technologies). The mobile phases were, respectively water, ACN and FA (75/25/0.1) (A) and ACN, FA (100/0.1) (B). The linear gradient was as follows: 0% B at 0 min, 85% B at 8.5 min, 100% B at 9.5 min, 100% B at 10.5 min, and 0% B at 12 min. The flow rate was 0.35 mL/min. The autosampler was set at 5°C and the injection volume was 5 μL. Electrospray ionization was performed in negative ion mode with a spray voltage fixed at -3500 V. After optimization, source conditions were as follows: source temperature was 325°C, nebulizer gas (nitrogen) flow rate was 10 L/min, sheath gas temperature was 400°C and sheath gas (nitrogen) flow rate was 12 L/min. Analyses were acquired in multiple reaction mode (MRM) using nitrogen as collision gas. For each compound the conditions of separation: retention time in minute (RT), and of quantification were defined: specific Q1/Q3 transition (T) fragmentor (F), and collision energy (CE). Thus, optimized parameters were as follows: LTB4 (RT: 4.32 min, T: 335/195, F: 115 V; CE: 4 V), LTB4-d4 (RT: 4.31 min, T: 339/197, F: 120 V, CE: 6 V). Peak detection, integration and quantitative analysis were performed using Mass Hunter Quantitative analysis software (Agilent Technologies).

The results are expressed as mean ± SD and statistical significance was assessed using the unpaired, two-tailed Student’s t-test or by analysis of variance (ANOVA) followed by Tukey’s multiple comparison test provided by GraphPad Prism software when more than two conditions were analyzed simultaneously. p values over 0.05 were not considered as significant.

We first measured the AA concentration in B. abortus infected moDC culture supernatants at 16 h post-infection. In parallel, other intracellular bacteria (T. whipplei, C. burnetii, O. tsutsugamushi) were tested. Only B. abortus and O. tsutsugamushi induced a significant AA production compared to non-stimulated moDC (Figure 2A).

AA can be metabolized by the COX pathway (Harizi et al., 2008). Thus, to determine the impact of B. abortus on this pathway, we analyzed the transcriptional profile of several key genes involved in the PG and leukotriene pathways in infected cells (Figure 2B). Compared to unstimulated cells, infected human moDC strongly expressed PTGS2 (33 times higher) and in a lesser extend PTGER2, PTGER4 and PTGDS (1,7, 3,1, and 1,3 times higher, respectively) (Figure 2B, upper panel). On the contrary, none of the genes involved in the leukotriene pathway were up-regulated following Brucella infection (Figure 2B, lower panel).

We next investigated the expression of Ptgs2 in murine GM-CSF-derived DC at 4, 8, 24, and 48 h post-infection. We observed that as in human DC, murine Ptsg2 mRNA level was strongly up regulated. Ptsg2 expression was time-dependent, reaching a peak at 24 h post-infection (229.4-fold up regulated compared to non-infected cells) (Figure 2C).

These results show that B. abortus infection is capable of stimulating the AA metabolic pathway and downstream PG pathway both in human and mice DC.

We next investigated the involvement of the PTGS-2 pathway in vivo upon infection in mice. We have first compared three different Brucella strains (B. abortus 2308, B. melitensis 16M, and B. suis 1330) and their ability to colonize CLN at 8 and 15 days post-infection using the oral infection model we previously shown to induce a specific colonization of CLN (von Bargen et al., 2014). We avoided to use intraperitoneal inoculation since it is associated to systemic infection and recedes from Brucella physiological routes of infection or vaccination (Grilló et al., 2012). Although bacterial loads in CLN were very similar at any time point for all Brucella species (Supplementary Figure 1A), B. melitensis and B. suis induced a higher lymphadenopathy compared to B. abortus at 15 days post-infection suggesting an enhanced inflammatory response in the CLN (Supplementary Figure 1B). We then decided to continue this study with B. melitensis 16M.

WT C57BL/6 mice were inoculated with B. melitensis by oral administration (with 10⁹ bacteria per mouse) as previously described (von Bargen et al., 2014) or by intranasal (with 10⁵ bacteria per mice), conjunctival (10⁹ bacteria per mice) or intradermal (10⁴ bacteria per mice) routes. Mice were sacrificed at 8 and 29 days post-infection and the weight and bacterial load of CLN and spleens were measured (Figures 3A,B).

At 8 days post-infection, the intradermal route was the only route capable of inducing lymphadenopathy compared to mock-treated mice and to other inoculation methods (Figure 3A, Left panel). No splenomegaly was observed at 8 days post-infection compared to mock-treated mice (Figure 3A, Right Panel). These results correlated with bacterial loads. CFU numbers in CLN were higher in mice infected intradermally compared to the other infection routes (Figure 3B, Left panel). However, spleens were poorly colonized as observed in all infection conditions (Figure 3B, Right panel). At 29 days post-infection, all inoculation routes resulted in lymphadenopathy and induced splenomegaly (Figure 3A). The increase in CLN and spleen weight was not associated with an increase in bacterial load (Figure 3B). This suggests the establishment of an inflammatory response in these organs.

We next investigated the local immune response induced by Brucella infection in the spleens and CLN by extracting mRNA at 8 and 29 days post-infection. Results were analyzed as a fold increase compared to their mock counterparts. No significant response was observed in the spleen for any infection routes (Supplementary Figure 2). On the contrary, a clear inflammatory response was observed in the CLN of mice infected intradermally from 8 days post-infection onwards (Figure 3C). This response led to a strong up-regulation of Ifng, Gzmb and Ptgs2 mRNAs (64, 18.6, and 30.8-fold, respectively) (Figure 3C). At 29 days post-infection, Ifng, Gzmb, and Il6 mRNA levels strongly decreased whereas a strong up-regulation of Ptgs2 and in a lesser extent Nos2 and Ccl2 mRNA levels remained (Figure 3C). The inflammatory response was less pronounced using the other inoculation routes. Indeed, no significant changes between day 8 and 29 were observed in terms of mRNA expression levels except for conjunctival inoculation displaying an increase of Nos2 and Ccl2 mRNA levels (Fold change: 10.3 and 2.9, respectively) (Figure 3C).

In conclusion, the intradermal infection route is the most potent at inducing inflammatory genes expression and Ptgs2 seems to be a robust marker of Brucella intradermal infection compared to intranasal, conjunctival or oral infection routes. At 29 days post-infection following B. melitensis intradermal inoculation, the expression of Il6 was significantly higher compared to intranasal infection. Nos2 (compared to the oral and intranasal routes), Ptgs2 and Ccl2 genes (compared to all the other routes) were also increased (Figure 3C). Interestingly, no systemic response was observed in mice sera at any time independently from the inoculation route except for IFN-γ after intradermal infection. However, low but significant amounts of TNF-α were detected in the sera of mice infected intradermally at 8 days post-infection (Supplementary Figure 3).

Overall, these results strongly suggest that in the CLN after intradermal infection a robust inflammatory response characterized by an Ifng and a Ptgs2 signature takes place.

After having identified that the intradermal route was the more potent to induce Ptgs2 expression during B. melitensis infection, we evaluated in vivo the involvement of COX-2 during infection. WT C57BL/6 mice were challenged intradermally with B. melitensis (10⁴ bacteria per mice) and were given NS-398, a specific COX-2 inhibitor, or DMSO vehicle (mock control) by intraperitoneal administration daily for 7 days.

We first looked at CLN and spleen weights at 8 days post-infection. Independently of the treatment, infected mice presented a lymphadenopathy compared to uninfected mice. However, NS-398 treatment induced a decrease in CLN weight compared to non-treated infected mice (Figure 4A, Left panel). This effect was consistent with the anti-inflammatory properties of NS-398. No significant effect could be observed in the spleen (Figure 4A, Right panel). This response was accompanied by a decrease of CFU numbers in CLN in treated mice (Figure 4A, Left panel) and no significant response was observed in the spleen (Figure 4B, Right Panel).

The local inflammatory response in NS-398-treated mice was characterized by a strong decrease of Il10 and Nos2 expression and in a lesser extent of Ccr7 and Foxp3 compared to untreated mice. However, Ifng mRNA level strongly increased following NS-398 treatment. All the other genes tested did not exhibit any significant changes in their expression (Figure 4C).

Altogether, these results indicate that inhibition of COX-2 decreases bacterial load in CLN by acting on Il10 and Ifng expression levels.