Brucella Cyclic β-1,2-Glucan Plays a Critical Role in the Induction of Splenomegaly in Mice

Brucella, the etiological agent of animal and human brucellosis, is a bacterium with the capacity to modulate the inflammatory response. Cyclic β-1,2-glucan (CβG) is a virulence factor key for the pathogenesis of Brucella as it is involved in the intracellular life cycle of the bacteria. Using comparative studies with different CβG mutants of Brucella, cgs (CβG synthase), cgt (CβG transporter) and cgm (CβG modifier), we have identified different roles for this polysaccharide in Brucella. While anionic CβG is required for bacterial growth in low osmolarity conditions, the sole requirement for a successful Brucella interaction with mammalian host is its transport to periplasmic space. Our results uncover a new role for CβG in promoting splenomegaly in mice. We showed that CβG-dependent spleen inflammation is the consequence of massive cell recruitment (monocytes, dendritics cells and neutrophils) due to the induction of pro-inflammatory cytokines such as IL-12 and TNF-α and also that the reduced splenomegaly response observed with the cgs mutant is not the consequence of changes in expression levels of the characterized Brucella PAMPs LPS, flagellin or OMP16/19. Complementation of cgs mutant with purified CβG increased significantly spleen inflammation response suggesting a direct role for this polysaccharide.


Introduction
Brucella is a bacterial pathogen that infects ruminants as the primary host and can be transmitted to humans by consumption of animal-derived contaminated products (e.g. unpasteurized dairy food) leading to brucellosis, a reticular endothelial disease, typically characterized by undulant fever [1]. Brucella, as the result of a longstanding association with the mammalian host, has evolved modified PAMPs (pathogen-associated molecular patterns molecules) such LPS and flagellin. These modified PAMPs limit host recognition by innate immune receptors (TLRs and NLRs) during the acute phase of the infectious process, reducing the induction of the inflammatory response known to be necessary for an efficient control of the infection [2][3][4]. In addition Brucella expresses virulence factors that actively interfere with the innate immune response such as the Brucella proteins Btp-A/TcpB and BtpB that down-regulate TLR2/TLR4 signaling [5][6][7]. Paradoxically, when the host is persistently infected at the chronic phase of brucellosis, inflammation becomes a dominant clinical sign that has been described to affect several organs producing symptoms such as arthritis, endocarditis, meningitis, epididymitis and splenomegaly [8].
To date few virulence mechanisms have been described in Brucella infection and different genomic studies have confirmed the absence of classical virulence factors like fimbriae, pilli, toxins or plasmids [9,10]. In our laboratory, we have identified and characterized a critical molecule for Brucella virulence: the cyclic b-1,2-glucan (CbG). CbG is composed by a family of cyclic polymers of 17 to 25 D-glucose molecules linked by b-1,2 linkages, synthesized by a membrane bound enzyme named Cgs (for cyclic glucan synthase) that utilizes UDP-glucose as the sugar donor [11]. Cgs initiates and elongates a linear chain of glucoses covalently bound to the Cgs which is subsequently cyclated and released to the bacterial cytoplasm [12]. A specific Brucella ABC-transporter Cgt (for Cyclic glucan transporter) translocates the CbG to the periplasmic space where they become chemically modified with the addition of succinyl groups, a reaction catalyzed by the membrane enzyme Cgm (for Cyclic glucan modifier) [13,14]. Although CbG is present within the periplasmic space, being this localization critical for hypo-osmotic adaptation, secretion of large amount of this polysaccharide to the supernatant by a noncharacterized mechanism has been described in Agrobacterium and Sinohizobium [15]. Interestingly, Brucella cgs mutant strain presents a defect in intracellular trafficking in epithelial cells that can be complemented by the addition of purified CbG (or Cyclodextrins) to tissue culture medium [16]. This observation suggests that also in Brucella, CbG must be secreted within the host cell to exert its role in virulence. The proposed mechanism of action for CbG in Brucella infection is the sequestration of cholesterol from intracellular host membranes leading to lipid raft disorganization and modulation of intracellular trafficking [16].
We have previously observed that mice infected with a cgs mutant had a reduced spleen inflammatory response even though they had a high number of replicating bacteria within this organ [17]. Since inflammation is the consequence of the host recognition of microbial PAMPs that ultimately lead to the induction of an inflammatory process, we hypothesized that either the periplasmic CbG may be stabilizing the expression of Brucella PAMPs such as LPS, flagellin, and OMPs or that CbG could be recognized by the innate immune receptors in the context of Brucella infection triggering inflammation.
It has been shown recently, using in vitro studies and purified CbG, that this molecule acts directly as an agonist of the innate immune system mediated by TLR4 in a MyD88/TRIF dependent fashion (and in a CD14 independent manner) [18]. Here we describe the role of Brucella CbG in splenomegaly and inflammation.

Ethics statement
The protocol of this study (reference number 10/2011) was approved by the Committee on the Ethics of Animal Experiments of the Universidad Nacional de San Martín, which also approved protocol development under the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

CbG purification from B. abortus
For preparative purposes, 6-liter of stationary-phase cultures of B. abortus S19 grown for 48 h at 37uC (250 rpm) were harvested by centrifugation at 10,0006g for 10 min. The pellets were extracted with ethanol (70% ethanol; 1 h at 37uC). The ethanolic extracts were centrifuged and the supernatants were dried in a speed-vac centrifuge and subjected to Bio-Gel P4 chromatography and HPLC on a C18 silica column for further purification as described previously [14]. Purity of CbG was confirmed by NMR analysis [14]. Thin-layer chromatography (TLC) was performed as previously describe [12].

Western blot analysis
To monitor the expression levels of outer membrane proteins and LPS in B. abortus strains, bacteria were grown in TSB and harvested at stationary phase. Equivalent bacterial pellets were resuspended in Laemmli buffer and samples were subjected to SDS-PAGE. Proteins were transferred onto nitrocellulose mem-branes using a semi-dry transfer procedure. Immunoblotting was performed using mouse monoclonal antibodies against Omp16 and Omp19 (kindly provided by Dr. Axel Cloeckaert) and mouse anti-O-polysaccharide monoclonal antibody M84 (kindly provided by Dr. K Nielsen). The correct O-antigen display on the membrane of B. abortus was confirmed by a Brucella phage sensitivity test (Tb s , Rc r ) and crystal violet staining [30].
In order to detect the expression of flagellin in B. abortus strains, bacteria were grown in 2YT medium [19] and harvested at early log phase (OD 600 0.2). Equivalent bacterial pellets were resuspended in Laemmli buffer and samples were subjected to 12% SDS-PAGE. Proteins were transferred as described above. Immunoblotting was performed using rabbit polyclonal antibodies against Brucella flagellin (kindly provided by Dr. J J Letteson).
Effect of the osmolarity on B. abortus growth All strains were grown in regular LB (170 mM NaCl) until stationary phase. Cultures were harvested by centrifugation (12,5006g for 5 min) and washed twice with PBS. Cells were suspended to an OD 600 of 0.9 and serially diluted in PBS, and 10 ml of the dilutions were spotted on LB agar or LB agar without the addition of NaCl. The plates were incubated at 37uC for 5 days before being read.

B. abortus virulence in mice
Virulence was determined by quantitating Brucella survival in mouse spleens at two weeks postinfection, as previously described [20]. Groups of 9-week-old female BALB/c mice were intraperitoneally (i.p) injected with different doses of B. abortus strains in 0.2 ml of sterile PBS. The animals were euthanized, and the spleens were removed, weighed, homogenized in PBS, serially diluted, and plated onto TSA with the appropriate antibiotics to determine the number of CFU per spleen.

Complementation of B. abortus cgs mutant with purified CbG
Mice were i.p inoculated with 1610 6 CFU of B. abortus S19 or 1610 6 CFU of its isogenic strain B. abortus cgs mutant. Sets of five mice inoculated with B. abortus cgs mutant were i.p injected with 15 mg of purified CbG during the first five days of infection. At two weeks postinfection, the animals were euthanized, and the spleens were removed, weighed, homogenized in PBS, serially diluted, and plated onto TSA with the appropriate antibiotics to determine the number of CFU per spleen.

Histological analysis of spleens infected with B. abortus
Group of mice were i.p infected with 1610 6 CFU of B. abortus S19 or its isogenic B. abortus cgs mutant strain. At two weeks postinfection spleens were excised, fixed with 10% neutral buffered formalin and paraffin embedded. Finally, four micrometers thick longitudinal sections of spleens were obtained and stained with hematoxylin and eosin to assess the degree of spleen inflammation.

Preparation of single-cell suspensions of spleens infected with B. abortus
Spleens were aseptically removed and single cell suspensions were prepared by gently teasing through a sterile stainless steel screen. Erythrocytes were lysed in red blood cell lysis buffer and cells were washed twice in PBS solution.
Determination of inflammatory cell recruitment in spleens infected with B. abortus BALB/c mice were infected with 1610 6 CFU of B. abortus S19 or its isogenic B. abortus cgs mutant strain. Spleens were obtained at two weeks postinfection and single cell suspensions were prepared. To assess recruitment of different cell subtypes after infection, splenocytes (2610 6   Determination of cytokines in bone-marrow derived macrophages (BMDM) infected with B. abortus strains Macrophages were derived from bone marrow of C57BL/6, Mal/tirap, TLR4, TLR6 and TLR9 KO mice as follows. Each femur and tibia was flushed with 5 ml of Hank's balanced salt solution (HBSS). The resulting cell suspension was centrifuged, and the cells were resuspended in Dulbecco's modified Eagle's medium (DMEM; Gibco, Grand Island, NY) containing 10% fetal bovine Serum (FBS; Gibco), 1% penicillin/streptomycin (100 mg/ mL) and 10% L929 cell-conditioned medium (LCCM) as a source of macrophage colony-stimulating factor (M-CSF). The cells were distributed in 24-well plates and incubated at 37uC in a 5% CO 2 atmosphere. Three days after seeding, another 0.1 ml of LCCM was added. On the seventh day, the medium was renewed. On the 10th day of culture, the cells were completely differentiated into macrophages [21]. The BMDM were infected with B. abortus S19 or its isogenic B. abortus cgs mutant strain, corresponding to a multiplicity of infection (MOI) of 100:1. After 60 min incubation with the bacteria, wells were washed three times with phosphatebuffered saline (PBS) and incubated with fresh medium containing 50 mg ml21 Gm and 100 mg ml21 streptomycin to kill noninternalized bacteria After 24 h postinfection, the level of IL-12 (p40) and TNFa in the supernatants of BMDM were measured by ELISA Duoset kit (R&D, Minneapolis, MN). At 4, 24 and 48 hours postinfection, infected C57BL/6 BMDM were washed three times with PBS and lysed with 500 ml 0.1% Triton X-100 (Sigma-Aldrich Co.). The intracellular CFU was determined by plating serial dilutions on TSA with the appropriated antibiotic.

Results and Discussion
Transport of Brucella CbG to periplasm is required for bacterial-host interaction As shown in Fig. 1A-a, Cgs, a 320 kDa membrane bound enzyme (the second largest protein in Brucella), is the enzyme responsible for the synthesis of cytoplasmic CbG, which is afterwards translocated to the periplasmic space by Cgt, a CbGspecific ABC transporter, and modified with succynil groups by the activity of Cgm (Fig. 1A-a). In order to determine if the different biosynthetic intermediate states of CbG play differential   roles in the bacteria the phenotype of three different B. abortus CbG mutants were compared: cgs [17], cgt [13] and cgm mutant [14]. As shown in Fig. 1A-b, deletion of cgs gene abolished the presence of CbG but mutations in cgt or cgm lead to the production of neutral CbG (Fig. 1A-c and d) with a different cellular localization being cytoplasmic in cgt mutant and periplasmic in the case of cgm. From the comparison between cgt and cgm it became evident that CbG plays at least two roles: adaptation to hypoosmotic growth conditions (Fig. 1B) and Brucella-host interaction (Fig. 1C). While the adaptation to hypo-osmotic conditions requires the presence of anionic periplasmic CbG (Fig. 1B), the sole requirement for host-pathogen interaction is the transport of CbG to the periplasmic space regardless of its anionic charge, as demonstrated by the cgm phenotype in intracellular replication and chronic infection in mice (Fig. 1C).
B. abortus mutants unable to produce CbG or to transport it to the periplasm elicit a reduced splenomegaly in mice As we reported, deletion of cgs gene either in the wild type strain B. abortus 2308 or in the vaccine strain B. abortus S19 (which is virulent for humans) reduced their ability to infect mice and hampered their efficiency to reach the intracellular replication niche in epithelial cells [16,17]. Although attenuation of virulence occurs in both Brucella backgrounds, in the S19 strain the phenotype is already evidenced after 4 weeks postinfection while in the 2308 background, attenuation is observed after 12 weeks postinfection [17]. Interestingly, at two weeks postinfection when the infection is still not resolved and the number of bacteria in the spleens are equivalent (in the case of B. abortus cgs08 mutant, Fig. 2A left panel) or slightly reduced (in the case of B. abortus cgs19 mutant, Fig. 2B left panel) compared to their respective wild type parental strains, the splenomegaly was significantly reduced in mice infected with both cgs isogenic mutants ( Fig. 2A and 2B, right panel). In an earlier report, Crawford et al. observed that splenomegaly elicited by Brucella melitensis infection (which peaks from two to three weeks postinfection) is dependent on the initial dose of infection rather than on the bacterial burden, concluding that very early events in the Brucella infection are the driving force controlling the severity of the inflammatory response in the spleen [22].
Since cgs mutants induced a reduced splenomegaly phenotype in both B. abortus 2308 and B. abortus S19 backgrounds, we decided to use the vaccine strain of B. abortus S19 for the next set of experiments as the wild type control because it elicited a significantly increased splenomegaly compared to the one induced by the B. abortus 2308 strain [23] allowing us to develop a more sensitive assay. Additionally, since B. abortus S19 presents an intracellular trafficking defect in epithelial cells [24], similar to the defect described for B. abortus cgs mutant strains, we reasoned that the use of S19 as our wild type control would reduce also the experimental variability due to differences in intracellular localization. Fig 2B (right panel) shows that the B. abortus S19 and B. abortus cgm19 mutant strains evoked a significantly increased inflammatory response in the spleens in comparison with the mutants B. abortus cgs19 and B. abortus cgt19, suggesting that the splenomegaly correlated with the presence of CbG within the periplasmic space. As mentioned above, CbG is likely secreted within the host cell and therefore periplasmic localization requirement might be potentially a prerequisite for its delivery outside the bacteria.

Splenomegaly is dependent on the initial dose of B. abortus infection
To study the impact of the initial dose of infection on the intensity of the splenomegaly, BALB/c mice were intraperitoneally infected with different doses of B. abortus S19 (10 3 , 10 6 and 10 9 CFU) (Fig. 3) and after two weeks postinfection spleens were removed, weighed, homogenized and the number of bacteria determined by serial dilution and plating to determine CFUs. Fig. 3 shows that, at two weeks postinfection although the numbers of replicating B. abortus S19 recovered from spleens were similar (about 10 6 -10 7 CFU per spleen) (Fig. 3A), the splenomegaly varied from negligible (about 0.1 grams) to a massive one (1 gram) (Fig. 3B) depending on the initial dose of infection, confirming previous observations in Brucella melitensis [22]. To estimate the impact of cgs phenotype on spleen inflammation, we infected mice with different doses of B. abortus cgs19 mutant strain and at two weeks postinfection, splenomegaly was determined. The results showed that, to achieve an equivalent degree of splenomegaly elicited by 1610 6 B. abortus S19, it was necessary to increase a thousand times the initial dose of the B. abortus cgs19 (1610 9 ) (Fig. 3B).  Reduced splenomegaly elicited by the cgs mutant is the consequence of a lesser degree of cell recruitment At two weeks postinfection spleens from Brucella infected mice were processed for histological analysis and the results are shown in Fig 4A. While spleens from wild type strain infected mice showed an increase in the global cellularity with a massive infiltration of the red and white pulp (Fig. 4A-c), spleens from cgs infected mice shown a reduced degree of cellular infiltration (Fig. 4A-b) similar to what was observed in spleens of non-infected animals ( Fig. 4A-a). Remarkably, at 4006 magnification, the presence of neutrophils and macrophages within the red pulp of B. abortus S19 infected mice was observed (Fig. 4A-i) indicating massive cell recruitment to the spleen. In addition, spleen white pulp of B. abortus S19 and B. abortus cgs19 infected mice presented reactive lymphocytes (larger and medium lymphocytes) likely due to antigen stimulation (Fig. 4A-e and f).
To further characterize the spleen cell population, antibodies against specific markers were used to identified and quantify neutrophils, monocytes, B cells, T cells and dendritic cells by flow cytometry. As shown in Fig 4B spleens from B. abortus S19 infected mice presented eight times more neutrophils, ten times more monocytes, three times more dendritic cells and two times more B cells than spleens infected with B. abortus cgs19 mutant strain. No difference was observed in T-cell recruitment (not shown).
One possible explanation for the reduced splenomegaly and cell recruitment associated with the lack of CbG biosynthesis or CbG transport to the periplasm could be that cgs/cgt strains have a differential expression of PAMPs that might lead to a less efficient engagement of the host innate immune receptors and consequently to a diminished inflammatory response.

B. abortus cgs mutant has normal expression of flagellin and Omps, displaying normal amounts of smooth LPS (S-LPS) on the membrane
It has been demonstrated that Agrobacterium and Sinorhizobium cgs mutants have a reduced expression of flagellin and a defect in flagella assembly that consequently leads to a non-motile phenotype [25] [26]. Since bacterial flagellins are powerful agonists of innate immune receptors, being recognized extracellularly by the surface receptor TLR-5 and intracellularlly by Nodlike receptors (such Ipaf or Naip5) [2], we explored if the Brucella cgs mutant has an altered expression of flagellin that might explain the diminished inflammatory response. In Alphaproteobacteria (including the Brucella genus) flagellin has a modification in the protein domain recognized by the TLR-5 innate immune receptor supporting the idea that Brucella flagellin has evolved to escape the host innate immune recognition [2]. In addition, in Brucella, flagellin expression is tightly controlled and only expressed under very strict culture conditions [19] and, because it has never been identified in any intracellular Brucella proteomic studies, this suggests that it is expressed poorly within the host cell [27,28]. In order to determine if the absence of periplasmic CbG affects flagellin expression, B. abortus S19 and its isogenic cgs, cgt and cgm mutants were grown in the conditional media to allow flagella assembly as described in Materials and Methods. Expression of Brucella flagellin was monitored by Western blot analysis, and the results, shown in Fig. S1A, indicates no changes in flagellin expression. Afterwards, we studied the expression of other critical Brucella PAMPs such as OMP16, OMP19 [29] or LPS [3,4] [31] and no differences in expression were observed associated to any B. abortus CbG mutant strains ( Fig. S1A and S1B).
Purified Brucella CbG partially complemented the splenomegaly defect in cgs-infected mice As was mentioned above, purified CbG is capable to restore the intracellular replication deficiency of cgs-mutant strain in HeLa cells [16] and this observation suggests that CbG must be secreted within the host cell to exert its role in virulence. To evaluate the direct role of Brucella CbG in the inflammatory response we designed a trans-complementation experiment using purified CbG. For this, we added the purified carbohydrate to the bacterial initial inoculum and during the following five days postinfection by i.p injection. Splenomegaly determined at two weeks postinfection showed an increase in spleen weight in the CbG complemented mice compared to the cgs mutant although not to the degree of the Brucella wild type strain (Fig. 5A). Although significant, trans-complementation with purified CbG elicits a moderate increase on spleen enlargement compared with cgs mutant strain, an effect that can be explained as the result of the intrinsic limitations of this experimental approach. For instance, since injected CbG is diluted within the peritoneal cavity is not possible to know the effective CbG concentration at the Brucella intracellular replicative niche. However, these results suggest that Brucella CbG plays a direct role in the induction of the inflammatory response in the spleen. It has been recently proposed that purified Brucella CbG may act as a novel class of adjuvant that can induce, in vitro, the activation of mouse and human dendritic cells, enhancing T cell responses and CD4+ T cells memory immune responses in a TLR4 dependent/CD14 independent fashion [18]. Interestingly, it was described recently that inflammation far from be an undesirable byproduct of the bacterial infection can be a process that pathogens can actively promote to create favorable conditions for its own advantage. Thus enterobacterial pathogens can outgrow commensal bacteria or promote host release of compounds of nutritional relevance mediated by the activity of its virulence mechanisms [32,33]. In that manner it is conceivable to speculate that CbG dependent splenomegaly might work as an advantage for host colonization in Brucella infection.

B. abortus cgs mutant strain elicits a reduced inflammatory response in BMDM
In previous studies, Zhan et al demonstrated that the macrophage-synthesized cytokines IL-12 and TNF-a are required for an efficient control of Brucella infection. In addition, it was shown that depletion of both pro-inflammatory cytokines by antibody treatment abolished the development of splenomegaly in animals infected with B. abortus S19 at two weeks postinfection [34]. To understand if the reduced spleen inflammation observed in mice infected with B. abortus cgs19 was due to a lower induction of IL-12 or TNF-a, we performed an in vitro infection experiment with naïve BMDM. Differently to the defective intracellular replication phenotype reported for B. abortus cgs mutant strains in HeLa cells [17], in BMDM B. abortus cgs19 mutant strain showed no defect in intracellular replication in comparison with its parental wild type strain (Fig. 6A). The same phenotype was also reported for cgs mutant strain for intracellular replication in dendritic cells [5]. As shown in Fig. 6B and 6C, wild type BMDM infected with B. abortus S19, secreted higher levels of IL-12 (Fig. 6B) and TNF-a (Fig. 6C) to the supernatant than cells infected with B. abortus cgs19 mutant strain, suggesting that CbG promotes the induction of proinflammatory cytokines from BMDM. To understand if this CbGdependent IL-12/TNF-a induction is dependent on Toll-like receptor (TLR) recognition, an in vitro experiment with BMDM from Mal/Tirap (the TLR2/TLR4 adapter protein), TLR4, TLR6 and TLR9 KO mice was performed ( Fig. 6B and C). As shown in Fig. 6B, CbG-dependent IL-12 induction was independent on the presence of TLR2, TLR4, TLR6 or TLR9 (Fig 6B) while CbG-dependent TNF-a induction was independent on the presence of TLR4 or TLR9 (Fig. 6C). In absence of TLR2 or TLR6, B. abortus S19 and its isogenic cgs mutant strain elicited similar levels of TNF-a (Fig. 6C). These results suggest that TLR2 and TLR6 are potentially involved in the TNF-a induction elicited by CbG. It is interesting to notice that TLR2 and TLR6 are able to interact to form a heterodimer which is responsible for bacterial deacylated lipoproteins recognition [35].
Taken together all these results suggest that the reduced splenomegaly observed in cgs mutant strain infected mice is a consequence of a lower induction of proinflammatory cytokines that lead to a lesser cell recruitment to this organ.

Concluding Remarks
In the present study we describe the role of the Brucella cyclic b-1,2-glucan in promoting spleen enlargement during bacterial infection. Splenomegaly was the result of massive cell recruitment, mediated by the induction of pro-inflammatory cytokines. Since mutants deficient in CbG biosynthesis in the soil bacteria Sinorhizobium and Agrobacterium have shown to have membrane alterations that lead to non-motile phenotypes and an increased sensitivity to dyes and detergents, we evaluated if the lowinflammation phenotype observed with the cgs/cgt mutants was due to changes in expression of membrane bound complexes with inflammatory activity. No differences in flagellin, OMPs or LPS expression were evident and results suggested that CbG per se is responsible for the splenomegaly observed. The molecular mechanism underlying CbG induced splenomegaly remains to be identified and further studies will be performed to characterize this process. Figure S1 Western blot analysis of flagellin, outer membrane proteins (Omps) (A) and LPS (B) in B. abortus CbG mutant strains. Immunoblotting was performed using: rabbit polyclonal antibodies against Brucella flagellin, monoclonal antibodies against Omp16 and Omp19; and Oantigen specific monoclonal antibody (M84). SDS-PAGE and Western blot were carried out as described in Materials and Methods. The same amount of total protein extracts were loaded into the gels. The estimated molecular weight of each protein is shown. (TIF)