TLR9 is required for MAPK/NF‐κB activation but does not cooperate with TLR2 or TLR6 to induce host resistance to Brucella abortus

Short abstract

TLR9 signaling through MAPK/NF‐κB pathways promotes proinflammatory responses, and helps control B. abortus infection in vivo independently of TLR2.

Abstract

Brucella abortus is a Gram‐negative intracellular bacterial pathogen that causes a zoonosis of worldwide occurrence, leading to undulant fever in humans and abortion in domestic animals. B. abortus is recognized by several pattern‐recognition receptors triggering pathways during the host innate immune response. Therefore, here, we determined the cooperative role of TLR9 with TLR2 or TLR6 receptors in sensing Brucella. Furthermore, we deciphered the host innate immune response against B. abortus or its DNA, emphasizing the role of TLR9‐MAPK/NF‐κB signaling pathways in the production of proinflammatory cytokines. TLR9 is required for the initial host control of B. abortus, but this TLR was dispensable after 6 wk of infection. The susceptibility of TLR9^−/−‐infected animals to Brucella paralleled with lower levels of IFN‐γ produced by mouse splenocytes stimulated with this pathogen compared with wild‐type cells. However, no apparent cooperative interplay was observed between TLR2–TLR9 or TLR6–TLR9 receptors to control infection. Moreover, B. abortus or its DNA induced activation of MAPK/NF‐κB pathways and production of IL‐12 and TNF‐α by macrophages partially dependent on TLR9 but completely dependent on MyD88. In addition, B. abortus‐derived CpG oligonucleotides required TLR9 to promote IL‐12 and TNF‐α production by macrophages. By confocal microscopy, we demonstrated that TLR9 redistributed and colocalized with lysosomal‐associated membrane protein‐1 upon Brucella infection. Thus, B. abortus induced TLR9 traffic, leading to cell signaling activation and IL‐12 and TNF‐α production. Although TLR9 recognized Brucella CpG motifs, our results suggest a new pathway of B. abortus DNA‐activating macrophages independent of TLR9.

Abbreviations

^−/−

deficient/knockout

BB

Brucella broth

BMDM

bone marrow‐derived macrophage

DC

dendritic cell

DHX9/36

aspartate‐glutamate‐any amino acid‐aspartate/histidine‐box helicase 9/36

ER

endoplasmic reticulum

HKBa

heat‐killed Brucella abortus

IRAK

IL‐1R‐associated kinase

KO

knockout

LAMP‐1

lysosomal‐associated membrane protein‐1

LCCM

L929 cell‐conditioned medium

MOI

multiplicity of infection

ODN

oligodeoxynucleotide

p

phosphorylation

PAMP

pathogens‐associated molecular pattern

PBT

Triton X‐100 in PBS

PIKfyve

phosphatidylinositol 3P 5‐kinase

UFMG

Federal University of Minas Gerais

Introduction

Microbial recognition is the first step to trigger an innate immune response against pathogens [1], which regularly display a molecular signature known as PAMPs. The most common PAMPs are LPS, peptidoglycan, bacterial lipoproteins, flagellin, and nucleic acids derived from viruses, bacteria, fungi, and protozoa [2]. The host innate immune system can interact with these PAMPs using a broad range of germline‐encoded pattern recognition receptors [3], which comprise, among others, the membrane‐bound TLRs, located at the cytoplasmic membrane or at the membrane encompassing endosomal vesicles [2]. Upon recognition of the cognate PAMP, TLRs (except TLR3) associate with MyD88, initiating cell signaling leading to the activation of the IκB kinase complex and MAPKs. These pathways result in the activation of NF‐κB [4] and AP‐1 [5] transcriptional factors, respectively, leading to proinflammatory cytokine production [6]. There are evidences demonstrating that multiple rather than single TLRs are required for immune response against most pathogens [7]. In that context, TLRs sensing multiple PAMPs trigger multiple pathways that work in cooperation generating additive or synergistic effects between among TLRs. For instance, mice genetically deficient for TLR2 or TLR9 exhibited only minor reductions in acute resistance to Mycobacterium tuberculosis infection. However, TLR2/9 double‐deficient mice displayed higher susceptibility to pathogen infection when compared with either of the single TLR‐deficient animals [8]. Additionally, treatment with single synthetic TLR2/6 and TLR9 agonists promotes only slight protection against bacterial pneumonia. However, simultaneous stimulation of these TLRs induces a high level of resistance against pneumonia [9, 10].

Brucella abortus is a facultative, intracellular Gram‐negative bacterium that causes brucellosis, a systemic infectious zoonotic disease [11]. In humans, this pathogen causes undulant fever, endocarditis, arthritis, and osteomyelitis. In cattle, B. abortus evokes infertility and abortion, resulting in drastic economical losses [12]. Macrophages are considered the main cells of B. abortus residence in the host; however, this bacterium also infects and multiplies in nonphagocytic cells [13]. Typically, the host reacts against B. abortus engaging a Th1 immune response with IFN‐γ production and CD8⁺ T cell activation [14, 15]. Nevertheless, a multiple set of host factors may contribute to a broad immune system reaction against B. abortus infection [16].

MyD88 and IRAK4‐deficient mice are reportedly unable to control B. abortus infection, and macrophages or DCs derived from these mice produce lower levels of proinflammatory cytokines in response to this pathogen [17, 18]. Moreover, MAPK and NF‐κB signaling pathways triggered by B. abortus are abrogated in macrophages genetically deficient for MyD88 or IRAK4 [18]. Indeed, corroborating the participation of MyD88 and IRAK4, B. abortus is recognized by various TLRs, triggering cell signaling pathways, resulting in proinflammatory cytokine production that interferes with pathogen behavior [16]. For instance, TLR2 and TLR6, 2 innate immune receptors displaying a high affinity to acylated lipopeptides [19], interact with B. abortus [20]. TLR2 activation by Brucella triggers pERK1/2 and pp38 and production of proinflammatory cytokines [20]. The usual mouse model of B. abortus infection through an intraperitoneal route demonstrated that TLR2 does not contribute to host control of brucellosis in vivo [21]. However, TLR2 showed a role controlling B. abortus infection 1 wk after intratracheal infection [22]. In addition, our group demonstrated that TLR6 is required for full production of proinflammatory cytokines and MAPK activation in DCs, resulting in resistance to B. abortus infection [23].

An additional innate immune receptor associated with B. abortus infection is TLR9 [24]. This receptor senses unmethylated CpG DNA motifs that are common in bacterial but unusual in mammalian genomes. In unstimulated cells, TLR9 is retained in the ER [25]. However, upon stimulation, TLR9 traffics to CpG DNA‐containing structures, initiating cellular signaling [26]. Our group has previously demonstrated that TLR9 is implicated at early phases of infection in host resistance to B. abortus infection in mice [17]. Moreover, DCs treated with bacterial DNA, extracted from HKBa or with HKBa itself, induced IL‐12 production in a TLR9‐dependent manner [24]. Furthermore, DNA methylation mitigates inflammatory cytokine production by splenocytes, indicating that the stimulatory property of DNA resides, at least partially, in unmethylated CpG motifs [24]. In the present study, we demonstrated that TLR9 activation is independent of cooperation with TLR2 or TLR6 during B. abortus infection. In addition, we investigated the role of TLR9 in a host innate immune response against B. abortus or its DNA, emphasizing the role of the TLR9‐MAPK/NF‐κB signaling pathway in the production of proinflammatory cytokines.

MATERIALS AND METHODS

Experimental animals

Wild‐type strain control mice (C57BL/6) were purchased from UFMG (Belo Horizonte, Brazil). Genetically deficient mice TLR2^−/−, TLR6^−/−, TLR9^−/−, and MyD88^−/− have been described previously [17]. Double‐KO TLR2/9^−/− and TLR6/9^−/− mice were generated by crossing the respective single‐KO mice. Animals from the F2 generation were genotyped using PCR‐based methods to identify the double‐KO mice. Mice were maintained at UFMG and used at 6–8 wk of age. All animal experiments were preapproved by the Institutional Animal Care and Use Committee of UFMG (CETEA #128/2014).

Bacterial strain

B. abortus strain S2308 was obtained from our own laboratory collection. Bacteria were grown in BB liquid medium (Difco, Detroit, MI, USA) at 37°C under continuous agitation. After 3 d of growth, the bacterial culture was centrifuged and the pellet resuspended in 0.15 M PBS, pH 7.4 (2.8 mM Na₂PO₄, 7.2 mM Na₂HPO₄, 0.14 M NaCl), containing 25% glycerol. Aliquots of these cultures were frozen at −80°C until use. To determine the bacterial concentration, aliquots were serially diluted and plated on BB medium containing 1.5% bacteriological agar. After incubation for 72 h at 37°C, bacterial numbers were determined by counting CFUs.

B. abortus infection

Fifteen mice from each group (C57BL/6, TLR2^−/−, TLR6^−/−, TLR9^−/−, TLR2/9^−/−, or TLR6/9^−/−) were infected intraperitoneally with 1 × 10⁶ CFU of virulent B. abortus strain S2308. After 1, 3, and 6 wk, 5 mice from each group were euthanized and their spleens harvested. The spleens were weighed independently before maceration in 10 ml saline (NaCl 0.9%). Following maceration, a portion of splenic extract was 10‐fold serially diluted and plated in duplicate on BB agar. After 3 d of incubation at 37°C, the number of CFU was determined. Results were expressed as the mean log CFU/g spleen of each group. The results are representative of 3 independent experiments. Alternatively, a second portion of splenic extract was prepared for culture to measure cytokine production.

Proinflammatory cytokine production in Brucella‐primed spleen cells

Spleen cells were harvested from 5 infected mice from each group (C57BL/6, TLR2^−/−, TLR6^−/−, TLR9^−/−, TLR2/9^−/−, or TLR6/9^−/−), 1 wk postinfection. After maceration of the organs, cells were treated with ammonium‐chloride‐potassium buffer (0.15 M NH₄Cl, 1.0 mM KHCO₃, 0.1 mM Na₂ EDTA, pH 7.2) to lyse RBCs. Then, cells were washed with saline and resuspended in RPMI 1640 (Gibco, Carlsbad, CA, USA), supplemented with 2 mM l‐glutamine, 25 mM HEPES, 10% heat‐inactivated FBS (Gibco), penicillin G sodium (100 U/ml), and streptomycin sulfate (100 µg/ml). To measure cytokine concentration by ELISA, 1 × 10⁶ spleen cells were plated/well in a 96‐well tissue‐culture dish. Splenocytes were stimulated with B. abortus S2308 at a MOI of 100:1, 1 µg/ml Escherichia coli LPS (Sigma‐Aldrich, St. Louis, MO, USA), or 5 µg/ml Con A (Sigma‐Aldrich). Unstimulated cells were used as the negative control. Spleen cells were incubated at 37°C in 5% CO₂, and culture supernatants were harvested 48 or 72 h after stimulation to measure TNF‐α/IL‐12 or IFN‐γ, respectively, by ELISA (DuoSet kit; R&D Systems, Minneapolis, MN, USA).

Purification of genomic DNA from B. abortus

To extract bacterial genomic DNA, a modification of the protocol described previously by our group was performed [27]. In brief, cells were resuspended in 10 mM Tris‐HCl, 1 mM EDTA, pH 8.0, buffer; heat killed at 60°C for 1 h; and incubated at 37°C for 1 h with 0.5% SDS and proteinase K (0.4 mg/ml). Bacterial debris, denatured proteins, and polysaccharides were removed by precipitation of the lysate with 5 M NaCl and cetyltrimethylammonium bromide‐NaCl solutions at 65°C for 10 min. DNA was extracted by a standard protocol with phenol‐chloroform‐isoamyl alcohol, precipitated with isopropanol, washed with 70% ethanol, and dried. Finally, the DNA‐containing pellet was reconstituted in 100 µl nuclease‐free water containing RNAse (50 µg/ml). The concentration and purity of the DNA were determined spectrophotometrically. Occasionally, extracted DNA was treated with DNAse I (Fermentas, Hanover, MD, USA), according to the manufacturer's instructions.

Mining CpG motifs from B. abortus genome

Genome mining was performed according to Bartholomeu et al. [28] using the complete sequenced genome of B. abortus available at the GenBank (Accession Numbers AE017223 and AE017224 for chromosomes I and II, respectively). In brief, a search for immunostimulatory CpG DNA motifs was conducted using the fuzznuc algorithm (EMBOSS package). The fuzznuc reverse option was turned on so that both DNA strands were searched. B‐Class‐like CpG sequences were found throughout searching for the presence of 2 CpG motifs having the general formula 5′‐YYGACGTN(1,4)GACGTY‐3′ or 5′‐GACGTN(1,4)GACTGY‐3′ motifs [29], where the numbers between parenthesis indicate the number of intervening nucleotides between each CpG motif. Three corresponding oligonucleotides containing CpG motifs found ( Table 1 ) were synthesized and tested for stimulation of BMDMs. Phosphorothioate oligonucleotides were used, as they are more resistant to endonuclease degradation. In addition, as a negative control, a nonstimulatory oligonucleotide containing a GpC sequence (5′‐TCCATGAGCTTCCTGAGCTT‐3′) instead of the canonical CpG was also synthesized.

Table 1.

Sequence and expected copy number of stimulatory mouse CpG motifs present in the B. abortus strain 2308 genome

Identification name	Sequence (5′–3′) a	Number of copies
CpG 1	TTGACGTTGACGTC	1
CpG 2	ATGACGTTTATGACGTC	1
CpG 3	AAGACGTTTGACGTT	1
CpG 4	TTGACGTTACAGACGTA	1
CpG 5	AAGACGTTCGACGTC	1
CpG 6	TTGACGTCGAAGACGTG	1
CpG 7	GAGACGTTCGGACGTA	1
CpG 8	GTGACGTGGATGACGTC	1
CpG 9	GCGACGTGGCGACGTT	1
CpG 10	GCGACGTGGACGTT	1
CpG 11	GCGACGTCAAGGACGTG	1
CpG 12	GAGACGTGGGCGACGTC	1

^a CG nucleotides belonging to the canonical CpG motifs are in bold.

Generation and in vitro stimulation of BMDMs

Macrophages were derived as described elsewhere [30]. In brief, bone marrow cells from C57BL/6, TLR9^−/−, and MyD88^−/− mice were extracted from the tibias and femurs. Cells were cultured in 24‐well plates (5 × 10⁵ cells/well) in DMEM (Gibco) containing 10% FBS (HyClone, Logan, UT, USA), 1% HEPES, and 10% LCCM. After 4 d, LCCM (100 μl/well) was added to the cells. At d 7, the medium was renewed. At d 10 of culture, medium was harvested, and 1 of the following was added: B. abortus (MOI 1:100); DNA extracted from B. abortus (2 μg/ml), treated or not with DNase; CpG ODN 1826 (1 µg/ml; InvivoGen, San Diego, CA, USA); E. coli LPS (1 µg/ml); Brucella‐derived CpG oligonucleotides (CpG 1–3; 1 µg/ml); negative control GpC ODN (1 µg/ml); or DMEM medium alone. Culture supernatants were harvested 24 h after stimulation, and IL‐12 and TNF‐α secretion was determined by ELISA (R&D Systems), according to the manufacturer's instructions. In the experiments with inhibitors, BMDMs were pretreated with vehicle or selective inhibitors for ERK1/2, JNK, p38, and NF‐κB (U0126, SP600125, SB203580, and BAY1170‐82, respectively; Cell Signaling Technology, Danvers, MA, USA) for 45 min. Then, cells were infected with B. abortus for 24 h, and harvested supernatants were used to measure cytokine production by ELISA.

Western blot analysis

BMDMs from C57BL/6, TLR9^−/− and MyD88^−/− mice were derived as described above. At d 10 of culture, cells were serum deprived for 16 h and then treated with B. abortus (MOI 1:1000), DNA extracted from B. abortus (2 μg/ml), CpG ODN 1826 (1 µg/ml), LPS (1 µg/ml), and medium alone for 30 min. After treatment, cells were washed with HBSS at room temperature and lysed [50 mM Tris‐HCl, pH 7.4, 150 mM NaCl, 50 mM NaF, 10 mM β‐glycerophosphate, 0.1 mM EDTA, 10% glycerol, 1% Triton X‐100, 1 mM sodium orthovanadate, and 1:100 protease inhibitor cocktail (Sigma‐Aldrich)]. Protein concentrations were determined by the bicinchoninic acid assays. Equal amounts of protein were loaded onto 12% SDS‐polyacrylamide gels and then transferred to nitrocellulose membranes (Amersham Biosciences, Uppsala, Sweden), according to standard techniques. Membranes were blocked 1 h at room temperature with blocking buffer (TBS containing 0.1% Tween‐20 and 5% nonfat dry milk) before incubation with rabbit mAb (pERK1/2 #4370, pp38 #4511, pJNK #4668, pNF‐κB p65 #3033, and β‐actin #4970; Cell Signaling Technology) overnight at 4°C. Subsequently, membranes were incubated with secondary IgG rabbit HRP‐conjugated #7074 antibody (1 h at room temperature), and protein bands were visualized using Luminol chemiluminescent HRP substrate (EMD Millipore, Billerica, MA, USA) in a Storm System 860 scanner (Amersham Biosciences). Densitometry analysis was performed using software Kodak version 1D 3.5.

Confocal microscopy

BMDMs were differentiated on 12 mm glass coverslips in 24‐well plates, as described before. Infection with the B. abortus S2308 strain was performed at a MOI of 1:100. Plates containing macrophages and bacteria were centrifuged at 1200 g for 5 min at 4°C. After 30 min and 2 and 4 h of infection, BMDMs were washed twice with PBS and fixed in 4% paraformaldehyde, pH 7.4, at 37°C for 15 min. Once fixed, cells were incubated with the primary antibody rat anti‐LAMP‐1 diluted 1:50 (Sigma‐Aldrich) and rabbit anti‐TLR9 diluted 1:100 (Abcam, Cambridge, MA, USA) in 0.3% PBT for 1 h at room temperature. Then, cells were washed 3 times with PBT and incubated for 30 min at room temperature with an anti‐rabbit secondary antibody conjugated to Alexa 488 (Invitrogen, Carlsbad, CA, USA) diluted 1:1000 and anti‐rat secondary antibody conjugated to Alexa 546 (Invitrogen) diluted 1:250 in PBT. The cells were washed twice with PBT and once with PBS and then mounted in the mounting medium ProLong Gold with DAPI (Invitrogen). Stained cells were examined by confocal microscopy (Nikon C2 confocal microscope). The colocalization between TLR9 and LAMP‐1 was performed using ImageJ software version 1.47n. Colocalization areas were highlighted in white dots.

Statistical analysis

The results were analyzed using GraphPad Prism 4 (GraphPad Software, La Jolla, CA, USA). Data were evaluated by ANOVA using P < 0.05 or P < 0.01, as specified in each figure legend.

RESULTS

TLR9 does not cooperate with TLR2 and TLR6 to control brucellosis in vivo

To investigate TLR9 cooperation with TLR2 or TLR6 in host defense against B. abortus infection in vivo, mice were infected intraperitoneally with 1 × 10⁶ CFU of B. abortus strain S2308. Furthermore, bacterial load was monitored in TLR9^−/−, TLR2/9^−/−, and TLR6/9^−/− mouse spleens by CFU counting at 1, 3, and 6 wk postinfection. Murine brucellosis was aggravated in TLR9^−/− mice at early stages of infection (1 and 3 wk) but not at a later phase of bacterial infection (6 wk; Fig. 1A and B ). Bacterial load recovery was higher in TLR9^−/− animals compared with wild‐type and TLR2^−/− mice. Additionally, Brucella CFU numbers observed in TLR2^−/− were similar to those determined in wild‐type mice. Indeed, TLR2 had no role in controlling B. abortus infection at all intervals studied. In addition, double‐KO TLR2/9^−/− mice showed a susceptibility profile similar to TLR9^−/−. Likewise, TLR6/9^−/− mice (Fig. 1B) showed a similar bacterial burden compared with TLR9^−/− mice, except at 6 wk after infection, when these double‐KO mice demonstrated higher susceptibility to B. abortus infection as a result of the lack of TLR6. These results corroborate that TLR2 is dispensable for control of B. abortus infection and indicate that there is no cooperative role between TLR2 and TLR9. Furthermore, TLR6 or TLR9 demonstrated a more important role during the immune response against B. abortus, however, with no apparent cooperation observed between these 2 receptors. Finally, by measuring the spleen weight of infected mice, we observed that single or double deficiency of TLRs did not affect splenomegaly induced by B. abortus (data not shown).

Figure 1.

TLR9 does not cooperate with TLR2 and TLR6 to control brucellosis in vivo. (A) C57BL/6, TLR2^−/−, TLR9^−/−, and TLR2/9^−/− mice and (B) C57BL/6, TLR6^−/−, TLR9^−/−, and TLR6/9^−/− animals were infected intraperitoneally with 1 × 10⁶ CFU of B. abortus S2308. Residual B. abortus CFU in the spleen was determined at 1, 3, and 6 wk postinfection (wpi). Data are expressed as means ± sd of 5 animals/time point. These results are representative of 3 independent experiments. *P < 0.05, 2‐way ANOVA, statistically significant difference in relation to wild‐type. #P < 0.05, 2‐way ANOVA, statistically significant difference among KO mice.

TLR9 and TLR6 are required for full IFN‐γ production by splenocytes

As immunity against B. abortus involves proinflammatory cytokine production, such as IFN‐γ and TNF‐α [14], we investigated the involvement of TLR2, TLR6, and TLR9 in the production of these cytokines during B. abortus infection. Hence, total spleen cells harvested from infected mice were stimulated with live bacteria and cytokine levels measured in the supernatants by ELISA. The production of IFN‐γ was lower in TLR6^−/−, TLR9^−/−, and TLR6/9^−/− mice when compared with wild‐type mice ( Fig. 2D ). However, although TLR9^−/− and TLR6/9^−/− spleen cells produced lower TNF‐α than wild‐type cells, TLR6^−/− displayed a similar TNF‐α production when compared with wild‐type (Fig. 2E). TLR9^−/− cells, producing less TNF‐α, might also be associated with reduced resistance to infection observed in these animals, as lack of TNF‐α has been related to increased susceptibility to murine brucellosis [31]. Moreover, TLR2^−/− mice showed no decrease in the levels of IFN‐γ and TNF‐α produced by splenocytes (Fig. 2A and B). Regarding TLR2/9^−/− mice, cytokine production was similar to TLR9^−/− mice and lower than TLR2^−/− and wild‐type cells. Although no cooperation was observed between TLR6 and TLR9, these 2 receptors are required for full IFN‐γ production by mouse splenocytes. As IL‐12 is an important cytokine involved in production of IFN‐γ by lymphocytes [32], we also measured this cytokine after stimulation with B. abortus (Fig. 2C and F). As expected, TLR2 was dispensable for IL‐12 production, and TLR6 and TLR9 were partially required for this cytokine production, correlating with the IFN‐γ production profile. Taken together, the reduced proinflammatory cytokine production paralleled higher susceptibility observed at 1 wk after B. abortus infection in TLR6^−/− and TLR9^−/− mice. Hence, the susceptibility of TLR9^−/−‐ and TLR6^−/−‐infected animals to Brucella may be associated with lower levels of IFN‐γ produced by spleen cells.

Figure 2.

TLR9 is required for full IFN‐γ and TNF‐α production by Brucella‐primed splenocytes. Splenocytes cultured (1 × 10⁶ cell/well) from 1 wk‐infected mice were stimulated in vitro with B. abortus S2308 strain (MOI 1:100), 5 μg/ml Con A, 1 μg/ml LPS, or medium as a negative control. Splenocyte supernatants were harvested 48 or 72 h after stimulation and measured by ELISA for TNF‐α (B and E), IL‐12 (C and F), or IFN‐γ (A and D). Data are expressed as means ± sd of 5 animals/group. These results are representative of 3 independent experiments. *P < 0.05, 2‐way ANOVA, statistically significant difference in relation to wild‐type. #P < 0.05, 2‐way ANOVA, statistically significant difference among KO mice.

TLR9 is partially required for IL‐12 and TNF‐α production by macrophages induced with B. abortus and its DNA

Murine macrophages play a prominent role mediating resistance or susceptibility to some intracellular pathogens [33]. Furthermore, macrophages are considered the main cells of Brucella residence in the host [13]. Therefore, we evaluated the role of TLR9 in macrophage activation and proinflammatory cytokine production upon activation by B. abortus or its DNA. Given that MyD88 is described as a pivotal molecule controlling B. abortus [17], we also compared the role of TLR9 with MyD88 during infection. Here, macrophages derived from bone marrow of TLR9^−/−, MyD88^−/−, and wild‐type mice were stimulated with B. abortus or its DNA, and the levels of IL‐12 and TNF‐α were measured in culture supernatants. Macrophages derived from TLR9^−/− mice stimulated with B. abortus or bacterial DNA showed reduced cytokine production compared with wild‐type cells ( Fig. 3A and B ). Additionally, the production of IL‐12 and TNF‐α by MyD88^−/−‐derived macrophages was below the level of detection. As expected, there was a drastic reduction of cytokine production in macrophages from both KO mice stimulated with CpG ODN 1826. In summary, we observed a MyD88‐dependent induction of IL‐12 and TNF‐α in macrophages induced by B. abortus or its DNA; however, this effect was only partially dependent on TLR9.

Figure 3.

Cytokine production in macrophages is partially dependent on TLR9 upon activation by B. abortus or its DNA. BMDMs from C57BL/6, TLR9^−/−, and MyD88^−/− mice were stimulated with B. abortus (MOI 1:100); B. abortus DNA (2 μg/ml), treated or not with DNase; CpG ODN 1826 (1 µg/ml); LPS (1 µg/ml); or medium alone. Culture supernatants were harvested 24 h after stimulation, and the TNF‐α (A) and IL‐12 (B) secretion was determined by ELISA. *P < 0.001, 2‐way ANOVA, statistically significant difference in relation to wild‐type.

TLR9 participates in the cell signaling pathway in macrophages

As B. abortus and its DNA induce macrophage production of proinflammatory cytokines, we evaluated the role of MAPK components and the NF‐κB transcription factor during cell activation. Macrophages from wild‐type mice were treated with selective inhibitors against ERK1/2 (U0126), JNK (SP600125), p38 (SB203580), and NF‐κB (BAY1170‐82) before stimulation with B. abortus. Harvested supernatants were used to assess IL‐12 and TNF‐α production ( Fig. 4A and B ). The treatment with the ERK1/2 inhibitor slightly reduced the production of both cytokines, whereas the inhibition of JNK or NF‐κB resulted in a drastic decrease in IL‐12 and TNF‐α levels. Moreover, the inhibition of p38 only affected the production of IL‐12, whereas the production of TNF‐α remained unaltered. In summary, MAPK and NF‐κB activation participates in inflammatory cytokine production following B. abortus infection.

Figure 4.

Cell signaling pathways in B. abortus‐infected macrophages. (A and B) BMDMs from C57BL/6 mice were pretreated for 45 min with selected inhibitors or vehicle, as indicated, and then infected with B. abortus for 24 h. TNF‐α (A) and IL‐12 (B) levels from harvested supernatant were measured by ELISA. *P < 0.05, 1‐way ANOVA; **P < 0.01, 1‐way ANOVA, statistically significant difference in relation to vehicle treatment. (C) BMDMs from C57BL/6, TLR9^−/−, and MyD88^−/− mice were stimulated for 30 min with B. abortus (MOI 1:1000); B. abortus DNA (2 μg/ml); CpG ODN 1826 (1 µg/ml); LPS (1 µg/ml); and medium alone. Cell lysates were then subjected to Western blot analysis with indicated antibodies. The amount of β‐actin was used as a loading control. Values below each band indicate the quantification of band intensities relative to β‐actin loading control.

Next, we determined the role of TLR9 in MAPK and NF‐κB signaling pathways to activate macrophages infected with B. abortus or treated with bacterial DNA (Fig. 4C). TLR9^−/− and MyD88^−/− macrophages showed reduced NF‐κB, p38, and JNK activation, as measured by detection of p65, p38, and p46/p54 JNK subunit phosphorylation, respectively, compared with wild‐type cells after B. abortus infection or treatment with its DNA. This cell signaling attenuation is more pronounced in MyD88^−/− than TLR9^−/− macrophages. Regarding ERK1/2 proteins of the MAPK family, TLR9 appeared not to participate in ERK1/2 activation, whereas MyD88 was fully required after B. abortus infection or its DNA treatment. As control, CpG ODN 1826 activation of MAPK and NF‐κB was dependent on TLR9 and MyD88. In addition, TLR9 was dispensable for LPS cell signaling stimulation, whereas MyD88 contributed partially to this effect. In summary, we concluded that TLR9 is partially required for MAPK and NF‐κB signaling pathways of macrophage activation by B. abortus or its DNA.

B. abortus infection redistributes TLR9 in BMDMs

The intracellular traffic of TLR9 initiates upon CpG stimulation [26]. This receptor moves from ER to a CpG‐containing endosome that colocalizes with LAMP‐1, a marker for late endosomes [34, 35]. Therefore, TLR9 colocalization with LAMP‐1 was investigated during B. abortus infection ( Fig. 5 ). TLR9 from BMDMs traffics to endosomes containing LAMP‐1 at 30 min after B. abortus infection. Then, colocalization between LAMP‐1 and TLR9 reaches the level observed in noninfected cells. Moreover, we observed that TLR9 molecules are spread throughout the cells after infection, suggesting that TLR9 molecules are translocated in infected BMDMs, probably to initiate a cellular signaling pathway leading to proinflammatory cytokine production.

Figure 5.

B. abortus infection promotes TLR9 colocalization with LAMP‐1‐positive compartments. BMDMs were infected with B. abortus (MOI 1:100) at 30 min and 2 and 4 h. Subsequently, macrophages were fixed and stained to detect host cell DNA (blue), TLR9 (green), and LAMP‐1 (red) by confocal microscopy. The colocalization between TLR9 and LAMP‐1 is highlighted (white dots). Original scale bar, 10 µM. NI, Noninfected.

B. abortus‐derived CpG oligonucleotides require TLR9 to promote IL‐12 and TNF‐α production by macrophages

Bacterial DNA, mainly unmethylated CpG motifs, is a canonical agonist of TLR9 [36]. As we observed here, cytokine production in macrophages, induced by B. abortus DNA, was partially dependent on TLR9. Therefore, we decided to identify the CpG oligonucleotides derived from B. abortus responsible for TLR9‐dependent macrophage activation. A search for Brucella genome immunostimulatory CpG DNA motifs was performed using bioinformatics tools (fuzznuc algorithm from the EMBOSS package). The fuzznuc reverse option was turned on to interrogate both DNA strands. The genomic mining for stimulatory CpG motifs retrieved 12 sequences listed in Table 1. The localization of these oligonucleotides in the B. abortus genome showed that CpG sequences are widely distributed in both DNA strands into chromosomes I and II ( Fig. 6 ). To evaluate the immunostimulatory properties of the selected oligonucleotides, 3 CpG motifs were synthesized and tested in macrophages for their ability to induce IL‐12 and TNF‐α production ( Fig. 7 ). We observed that CpG 1 is a weak agonist to induce IL‐12 and TNF‐α production in macrophages. However, CpG 2 and CpG 3 stimulated robust proinflammatory cytokine production, and this biologic effect was dependent on TLR9 and MyD88. As expected, treatment of macrophages with the oligonucleotide containing a GpC sequence instead of a canonical CpG, resulted in lower cytokine production. Taken together, TLR9 is required for full production of inflammatory cytokines by macrophages treated with B. abortus‐derived CpG oligonucleotides.

Figure 6.

Localization of putative mouse immunostimulatory CpG motifs present in the B. abortus strain 2308 genome. The CpG oligonucleotides are highlighted in black blocks inserted in DNA strands of chromosomes (Chr) I and II.

Figure 7.

Brucella‐derived CpG oligonucleotides require TLR9 to promote cytokine production by macrophages. BMDMs from C57BL/6, TLR9^−/−, and MyD88^−/−mice were stimulated with Brucella‐derived CpG oligonucleotides (CpG 1–3), CpG ODN 1826, negative control GpC ODN 2138, LPS (1 µg/ml), or medium alone. Culture supernatants were harvested 24 h after stimulation to determine by ELISA the TNF‐α (A) and IL‐12 (B) secretion. *P < 0.05, 2‐way ANOVA, statistically significant difference in relation to wild‐type. #P < 0.05, 1‐way ANOVA, statistically significant difference in wild‐type macrophages in relation to CpG3 treatment.

DISCUSSION

Pathogenic microorganisms display several strategies to subvert the immune system to establish a propitious environment to survive and replicate. However, host cells express several receptors/sensors that interact directly or indirectly with pathogens to control the infection. In that context, several molecules, such as nucleotide‐binding oligomerization domain‐like receptor family, pyrin domain containing 3 or absent in melanoma 2, besides TLRs, emerge as sentinel molecules in the host immune response against B. abortus infection [30]. In fact, as the pivotal role has been identified for MyD88 in host response against B. abortus infection [17], the participation of TLRs has been investigated extensively by our group. Despite initial data demonstrating the participation of TLR9 conferring host resistance to B. abortus, that study was conducted only 2 wk after infection [17]. Additionally, there has been no investigation to evaluate a potential cooperative role of TLR9 with other receptors during B. abortus infection in vivo. In the present study, we confirm that TLR9 is required for the initial host control of this pathogen, but this TLR was dispensable after 6 wk of infection in vivo (Fig. 1). Furthermore, no apparent cooperative interplay was observed between TLR2–TLR9 or TLR6–TLR9 receptors concerning bacterial load in infected mice. However, crosstalk between TLR2 and TLR9 in DCs stimulated with HKBa has been reported in vitro [37]. In such a scenario, TLR2 can drive the TLR9 response in DCs upon HKBa stimulation, leading to IL‐12 production [37]. Although TLR2–TLR9 cooperation occurs between these receptors in vitro, this phenomenon may have no effect on host susceptibility in vivo, at least in the experimental conditions performed here. Additionally, we did not observe any difference in IL‐12 production from TLR2^−/− splenocytes compared with wild‐type mice. The difference between our study and Zhang et al. [37] might be a result of the host cell type used (DCs vs. splenocytes) or because we performed experiments with live bacteria, and they used heat killed. TLR2 and TLR6 were shown to cooperate with increasing cell signaling through NF‐κB activation upon B. abortus stimulation in vitro [23]. However, the TLR2/6 double‐KO mice showed no increased susceptibility to B. abortus infection compared with single TLR6^−/− animals [23]. Obviously, the in vivo response is more complex, as many cell types and receptors respond to the pathogen.

Immunity to B. abortus infection is strongly correlated with the type 1 pattern of immune response, where IFN‐γ and TNF‐α are the key cytokines. Previous studies have demonstrated that IFN‐γ plays an essential role to control Brucella infection [14]. Indeed, IFN‐γ^−/− mice succumb to brucellosis before the third week postinfection [38]. Furthermore, TNF‐α production upon Brucella infection results in massive splenomegaly as a consequence of extensive recruitment of monocytes, DCs, and neutrophils. Furthermore, blockade of TNF‐α function with specific antibodies augments susceptibility to B. abortus [39, 40]. In that context, we investigated the production of IFN‐γ and TNF‐α by splenocytes from infected mice. The restimulation of these cells with B. abortus resulted in robust production of IFN‐γ and TNF‐α, partially dependent on TLR9. Regarding TLR6, this receptor contributed to IFN‐γ but was dispensable to full production of TNF‐α by splenocytes. However, no cooperative role was observed between TLR2–TLR9 and TLR6–TLR9 to induce IFN‐γ and TNF‐α in spleen cells. The partial requirement of TLR9 for IFN‐γ production was also described previously following HKBa treatment and measurement of IFN‐γ in serum [24]. Because of the essential role of IFN‐γ in B. abortus resistance, reduced synthesis of this cytokine likely correlates with enhanced susceptibility observed in TLR9^−/− mice.

Upon B. abortus recognition, macrophages are activated and produce proinflammatory cytokines to attempt the control of pathogen replication [16]. In the present study, the TLR9 pathway triggered by B. abortus or its DNA results in IL‐12 and TNF‐α production. However, the contribution of TLR9 is only partial, suggesting the involvement of other innate immune receptors to sense this pathogen and its DNA. The inhibition of common pathways that generate these cytokines reveals the role of cell signaling events triggered by Brucella. The treatment with selective inhibitors of JNK (SP600125) and NF‐κB (BAY1170‐82) drastically abrogated IL‐12 and TNF‐α production by macrophages stimulated with B. abortus. Moreover, p38 inhibition preferentially blocked IL‐12 production. Furthermore, the activation of these cell signaling molecules was affected by the lack of TLR9 or MyD88, as assessed by Western blot. Taken together, our results suggest that MAPK and NF‐κB signaling pathways activated through TLR9 by B. abortus lead to proinflammatory cytokine production. Furthermore, B. abortus genomic DNA induced cytokine production partially dependent on TLR9; however, MyD88 was fully required, suggesting an alternative pathway for B. abortus DNA to activate MAPK and NF‐κB signaling pathways. MyD88‐dependent DNA sensor other than TLR9 was identified during murine cytomegalovirus infection [41]. Subsequently, 2 proteins named DHX36 and DHX9 were identified as DNA sensors that recruit MyD88, acting as nucleotide receptors independent of TLR9 but MyD88 dependent [42]. Therefore, we need to determine whether these DNA sensors might be involved in detecting Brucella DNA.

To investigate Brucella DNA motifs sensed by TLR9, we performed a search for immunostimulatory CpGs in bacterial genome. Two oligonucleotides, termed here as CpG2 and CpG3, stimulated a robust production of IL‐12 and TNF‐α in macrophages. Moreover, this biologic effect, promoted by Brucella‐derived CpG oligonucleotides, was totally abrogated in animals that lack TLR9. These data corroborate initial findings demonstrating that Brucella DNA methylation reduces the stimulatory properties of this molecule in splenocytes [24]. This may reflect the nature of TLR9—recognizing CpG motifs that are not methylated—present in bacterial DNA and suggests that the recognition of genomic DNA by innate receptors (TLR9 independent but MyD88 dependent) was probably a result of non‐CpG motifs.

BMDMs infected with B. abortus changed the TLR9 localization pattern, resulting in TLR9 colocalizing with LAMP‐1‐positive endosomes. This colocalization occurs mostly at 30 min postinfection—the same time of MAPK and NF‐κB activation. Indeed, DCs that express a mutant form of UNC93B1, a membrane protein that delivers the nucleotide‐sensing TLRs from the ER to endolysosomes, display deficiency in proinflammatory cytokine production when stimulated with CpG [43]. Likewise, TLR9 trafficking is regulated by PIKfyve, and blockage of this enzyme activity reduces TLR9 and CpG colocalization to LAMP‐1‐positive compartments [35]. Moreover, PIKfyve inhibition results in impaired pJNK and pp38 MAPK in response to CpG [34]. Thus, B. abortus induction of endosome trafficking may be critical for the initiation of TLR9‐mediated signaling, resulting in proinflammatory cytokine production.

Altogether, TLR9 displays a prominent role during the initial phase of B. abortus infection. However, there is no cooperative interplay between this receptor and TLR2 or TLR6 concerning host resistance to this pathogen. Moreover, here, we identified Brucella CpG DNA motifs involved in activation of the host innate immune responses through TLR9. However, our data suggest a new pathway of B. abortus DNA‐activating macrophages, independent of TLR9 but MyD88 dependent. In addition, TLR9 initiated cellular traffic upon B. abortus infection in macrophages, and MAPK/NF‐κB signaling may be activated in this process, resulting in proinflammatory cytokine production. Further experiments may shed light on this intricate circuit by which the host immune senses Brucella‐derived nucleic acids.

AUTHORSHIP

M.T.G., P.C.C., and G.S.P. performed the experiments presented in the paper. M.T.G., D.C.B., G.S., and S.C.O. contributed to the design of experiments. M.T.G., G.S., and S.C.O. wrote the manuscript. S.C.O. was responsible for the study.

DISCLOSURES

The authors declare no conflicts of interest.