|Home | About | Journals | Submit | Contact Us | Français|
Citrobacter koseri is a Gram-negative bacterium that can cause a highly aggressive form of neonatal meningitis, which often progresses to establish multifocal brain abscesses. Despite its tropism for the brain parenchyma, microglial responses to C. koseri have not yet been examined. Microglia use TLRs to recognize invading pathogens and elicit proinflammatory mediator expression important for infection containment. In this study, we investigated the importance of the LPS receptor TLR4 and MyD88, an adaptor molecule involved in the activation of the majority of TLRs in addition to the IL-1 and IL-18 receptors, for their roles in regulating microglial activation in response to C. koseri. Proinflammatory mediator release was significantly reduced in TLR4 mutant and MyD88 knockout microglia compared with wild-type cells following exposure to either live or heat-killed C. koseri, indicating a critical role for both TLR4- and MyD88-dependent pathways in microglial responses to this pathogen. However, residual proinflammatory mediator expression was still observed in TLR4 mutant and MyD88 KO microglia following C. koseri exposure, indicating a contribution of TLR4- and MyD88-independent pathway(s) for maximal pathogen recognition. Interestingly, C. koseri was capable of surviving intracellularly in both primary microglia and macrophages, suggesting that these cells may serve as a reservoir for the pathogen during CNS infections. These results demonstrate that microglia respond to C. koseri with the robust expression of proinflammatory molecules, which is dictated, in part, by TLR4- and MyD88-dependent signals.
In the United States, ~30– 40% of bacterial meningitis cases are caused by Gram-negative organisms (1, 2). One example of a Gram-negative pathogen that exhibits a remarkable degree of tropism for the brain is Citrobacter koseri (formerly C. diversus) (3–9). One intriguing feature of C. koseri meningitis is the propensity of organisms to disseminate into the CNS parenchyma and establish multifocal abscesses (6–9). Indeed, infants that become infected with C. koseri either by transmission via an infected mother during parturition or hospitalization experience an alarming frequency of brain abscess development (i.e., ~77%), which far exceeds the occurrence by any other meningitis-causing bacteria (7). A significant percentage of infants infected with C. koseri experience a high mortality rate (i.e., 30–50%) and among those that survive, ~75% experience long-term neurological deficits, including seizures, cognitive disabilities, and hearing loss (8). Therefore, although the incidence of C. koseri infections is rather low, the serious impact of this disease highlights the need to understand the pathogenic mechanisms induced by this organism. Although several studies have examined the pathogenesis of C. koseri-induced meningitis and brain abscess in animal models (10–15), to date, the responses of resident microglia to C. koseri have not yet been examined, despite the tropism of this organism for the CNS parenchyma.
Microglia are the resident mononuclear phagocyte population in the CNS parenchyma and represent an important component of the innate immune response against invading pathogens (16–19). Because of the high predilection of C. koseri for the brain, it is likely that resident microglia play a key role in sensing CNS colonization and mounting an initial antibacterial immune response before the recruitment of peripheral immune cells. Microglia are equipped with a repertoire of pattern recognition receptors (PRRs),3 including TLRs that play a pivotal role in detecting invariant motifs expressed by pathogens (known as pathogen-associated molecular patterns or PAMPs) (20, 21). Gram-negative bacteria such as C. koseri contain an abundance of LPS in their outer cell wall that exerts potent proinflammatory activity (22, 23). The LPS recognition complex is formed by a tripartite set of proteins, namely CD14, TLR4, and MD-2. The current consensus is that LPS binds to CD14, which subsequently triggers TLR4 activation and downstream signaling pathways, culminating in the release of a wide array of inflammatory mediators that participate in antibacterial immunity (20, 21, 24, 25). MyD88 is a pivotal downstream signaling adaptor molecule recruited by the TLR4 receptor complex that leads to proinflammatory gene expression by NF-κB and MAPK signaling pathways (25, 26). However, TLR4 can also transduce signals via MyD88-independent pathways through its association with other adaptors such as TIR domain-containing adaptor protein and TIR domain-containing adaptor inducing IFN-β that lead to the induction of IFN-inducible gene products (21, 25). Microglia express CD14, TLR4, and MyD88 that represent important players in the LPS-sensing machinery of these cells (27). Since C. koseri is a Gram-negative pathogen, we predicted that this organism would trigger TLR4-dependent pathways; however, a role for MyD88-dependent vs -independent signaling was uncertain. Because microglial responses to C. koseri, let alone the involvement of TLR4 and MyD88 in bacterial recognition has never been reported, one objective of the current study was to evaluate the functional importance of these molecules using primary microglia isolated from MyD88 knockout (KO) animals in addition to mice that possess a point mutation in the cytoplasmic tail of TLR4 (hereafter referred to as TLR4 mutant), resulting in its inability to signal.
Phagocytosis refers to the process whereby particulate material is engulfed by phagocytic cells via opsonic or nonopsonic pathways. Phagocytes, such as macrophages and neutrophils, utilize this mechanism to internalize and kill microbes via the action of reactive oxygen and nitrogen intermediates and degradative enzymes (28, 29). However, many pathogens, such as Mycobacterium tuberculosis (30) and Leishmania species (31), have evolved strategies to evade microbial killing mechanisms and survive intracellularly in phagocytes. It has been proposed that the infiltration of Citrobacter-harboring monocytes/macrophages into the brain ventricles and periventricular parenchyma is important in the pathogenesis of brain abscess formation by this organism (10, 14). Ancillary evidence to support this possibility was provided in studies where C. koseri intracellular survival and replication was demonstrated in the human monocyte cell line U937 (14). Although resident microglia are equipped with numerous phagocytic receptors (16, 32), to date, no studies have examined whether C. koseri is capable of intracellular survival and/or replication in this CNS phagocyte population. We envision that microglia may harbor viable C. koseri intracellularly and serve as a reservoir for bacterial survival and replication within the brain parenchyma, which may contribute to disease pathogenesis.
In the current study, we demonstrated that C. koseri is a potent stimulus of microglial activation typified by the synthesis of numerous proinflammatory mediators, including NO, TNF-α, IL-1β, CXCL2 (MIP-2), and CCL2 (MCP-1). The induction of these mediators by C. koseri was found to be mediated primarily via TLR4 and MyD88, because their production was significantly attenuated in primary microglia deficient for either molecule. However, a small residual response to C. koseri was still observed in TLR4 mutant and MyD88 KO microglia, suggesting a minor contribution for pathways independent of both molecules.
Analysis of intracellular survival by gentamicin protection assays revealed that although C. koseri replicated efficiently in a human monocyte cell line (U937), microglia did not support significant bacterial proliferation. However, C. koseri was capable of surviving intracellularly in primary microglia since organisms were not eradicated from cells during the 168 h time course evaluated in these studies. Collectively, these results suggest that microglia may play a dual role during brain colonization by C. koseri. On the one hand, C. koseri is a potent inducer of numerous proinflammatory molecules by primary microglia, which likely participate in the genesis of a protective antibacterial immune response. Conversely, microglia demonstrate poor intracellular killing of C. koseri and the persistence of bacteria within microglia may facilitate the spread of infection or maladaptive microglial responses. Further studies are needed to address these questions that may impact the pathogenesis of these devastating infections caused by C. koseri colonization of the CNS.
The C. koseri isolates used in this study (strains 4036 and 4277) were provided by Dr. J. G. Vallejo (Baylor College of Medicine, Houston, TX). Strain 4036 was originally isolated from the cerebrospinal fluid (CSF) of an infant with meningitis and multiple brain abscesses, whereas strain 4277 was recovered from a tracheostomy of a chronically ventilator-dependent child without signs or symptoms of respiratory infection (12). On the basis of the antibiotic sensitivity profile of both strains (performed by the Clinical Microbiology laboratory at Arkansas Children's Hospital, Little Rock, AR), isolates were propagated in the presence of ampicillin (20 μg/ml; MP Biomedicals) for positive selection of C. koseri. C. koseri glycerol stocks were streaked onto blood agar plates and incubated for 20 h one day before initiating broth cultures. C. koseri strains 4036 and 4277 were propagated in brain-heart infusion broth with constant agitation (250 rpm) and recovered at mid-log-phase by washing twice with ice-cold PBS (pH 7.4). The numbers of colony-forming units used for infection were initially estimated by OD620 readings before microglial infection and subsequently determined by plating out serial dilutions on blood agar plates. To prepare heat-killed C. koseri stocks, bacteria were incubated at 56°C for 1 h with frequent vortexing. The timing regimen for preparing C. koseri working stocks was held constant throughout these experiments to avoid potential confounds by changes in virulence factor expression that could occur at various growth phases.
Neonatal TLR4 mutant and TLR4 WT (C3H/HeJ and C3H/FeJ, respectively, The Jackson Laboratory) as well as MyD88 KO mice (provided by Dr. S. Akira, Osaka University, Japan) (33) were used to prepare primary microglial cultures as described previously (34, 35). Microglia isolated from C57BL/6 mice were used as wild-type (WT) controls for MyD88 KO cells since the latter had been backcrossed to C57BL/6 mice for a total of 10 generations. The C3H/HeJ strain harbors a missense mutation in the third exon of the TLR4 gene, rendering it incapable of signaling (36). C3H/FeJ mice are on the same genetic background as the TLR4 mutant strain and are used as WT controls for comparison. The purity of microglial cultures in these studies was routinely >98%.
C57BL/6 mice (5–7 wk old) were used to procure thioglycollate-elicited peritoneal macrophages. Briefly, each mouse received an i.p. injection of 2–3 ml of sterile 4% thioglycollate broth (Brewer's modified medium; Baltimore Biological Laboratory). On day 4 following thioglycollate injections, mice were euthanized using an overdose of inhaled isoflurane, and peritoneal macrophages were harvested by vigorous lavage of the perito-neal cavity with 10 ml of ice-cold sterile PBS. When necessary, any contaminating RBC were eliminated by hypotonic lysis (BD PharmLyse) according to the manufacturer's instructions. Elicited macrophages were washed extensively, resuspended in RPMI 1640 medium (Mediatech), supplemented with 10% FBS (HyClone), and seeded into 24-well tissue culture plates at 6 × 105 cells/well for gentamicin protection assays.
The human monocyte cell line U937 was provided by Dr. S. Barger (University of Arkansas for Medical Sciences, Little Rock, AR). Cells were cultivated in RPMI 1640 medium (Mediatech) supplemented with 2 mM l-glutamine, 1 mM sodium pyruvate, and 10% FBS. U937 cells were seeded in 24-well plates at 5 × 105 cells/well in medium containing 0.1 μg/ml PMA (Sigma-Aldrich) for 1 day before gentamicin protection as-says, to induce cell adherence and maturation. Incubation of U937 cells with PMA was discontinued 24 h before the initiation of gentamicin protection assays to avoid any potential activation artifacts.
Quantitation of TNF-α and IL-1β (mouse CytoSet; BioSource International), CXCL2 (mouse DuoSet; R&D Systems), and CCL2 (mouse OptEIA; BD Biosciences) levels in conditioned supernatants from micro-glia treated with C. koseri strains was performed using standard sandwich ELISA according to the manufacturer's instructions. The lower limit of sensitivity for these assays was 15.6 pg/ml.
Levels of nitrite, a stable end product of NO following its reaction with O2, were quantitated in conditioned supernatants of microglia treated with C. koseri using the Griess reagent (0.1% naphtyletylenediamine dihydrochloride, 1% sulfanilamide, and 2.5% phosphoric acid; all from Sigma-Aldrich). The absorbance at 550 nm was measured on a plate reader (Spectra Max 190; Molecular Devices), and nitrite concentrations were determined using a standard curve with sodium nitrite (NaNO2; Sigma-Aldrich; level of sensitivity 0.4 μM).
The effect of C. koseri on microglial viability was evaluated by the ability of mitochondria to convert the substrate (MTT) into formazan crystals as described previously (37).
To evaluate the ability of C. koseri clinical isolates to survive and/or replicate in microglia or macrophages, GPAs were performed. Gentamicin kills extracellular C. koseri, but because its ability to permeate the eukaryotic cell membrane is limited, intracellular organisms are protected from its bactericidal activity. Primary microglia and thioglycollate-elicited perito-neal macrophages from C57BL/6 mice and the human monocyte cell line U937 were plated into 24-well tissue culture plates in antibiotic-free medium for 24 h. Subsequently, cells were infected with live C. koseri strains at a multiplicity of infection (MOI) of 10 (10 bacteria to 1 microglia/ macrophage) for 1 h at 37°C. After the 1-h incubation, all wells were washed extensively with PBS to remove extracellular organisms and incubated with medium supplemented with 100 μg/ml gentamicin for one additional hour at 37°C. Finally, all wells were washed twice and received fresh medium supplemented with 20 μg/ml gentamicin (exceeding the minimal inhibitory concentration for both C. koseri strains; data not shown) and incubated at 37°C until the appropriate time point for analysis. The continued presence of gentamicin ensured effective killing of any residual extracellular bacteria. Microglia, macrophages, and U937 monocytes were lysed anywhere from 24 to 168 h after C. koseri infection with PBS containing 0.5% Triton X-100 (Fisher Scientific), whereupon cell lysates were serially diluted and plated onto blood agar plates to quantitate the numbers of viable intracellular bacteria. In some experiments, microglia and macrophages were pretreated with LPS (Escherichia coli O111:B4; List Biological Laboratories) for 6 h before live C. koseri exposure to determine whether the activation status of cells would impact bacterial survival and/or replication. Results are presented as the number of colony-forming units per well after normalization to the actual MOI of the infectious inoculum as determined on blood agar plates. This was required because OD620 readings can only provide an estimate of bacterial density on the day of infection. The number of viable bacteria in wells devoid of microglia or macrophages served as a control to confirm adequate killing of extracellular C. koseri by gentamicin. All assays were performed in triplicate with results repeated in a minimum of four independent experiments.
Protein extracts were prepared from primary microglia as previously described (38) and quantified using a standard protein assay (bicinchoninic acid protein assay reagent; Bio-Rad). CD14 and MyD88 expression was evaluated by Western blot using rat anti-mouse CD14 (BD Biosciences) and rabbit anti-mouse MyD88 (eBioscience) Abs. Blots were developed using the Immobilon Western substrate (Millipore) and visualized using an Alpha Innotech imager (Alpha Innotech). Blots were stripped and reprobed with a rabbit anti-actin polyclonal Ab (Sigma-Aldrich) for verification of equivalent protein loading.
Significant differences across experimental groups were determined with a one way ANOVA followed by the Holm-Sidak method for pairwise multiple comparisons using SigmaStat 3.0 (SPSS Science).
Microglia are the resident innate immune cells in the brain parenchyma (16) and produce numerous proinflammatory mediators upon exposure to bacterial pathogens, including but not limited to NO, TNF-α, IL-1β, CXCL2, and CCL2 (18, 39–44). Many of these proinflammatory molecules are important for inducing the influx and subsequent activation of peripheral immune cells into the infected CNS (45–47). Despite the fact that C. koseri is capable of colonizing the brain as evident by its propensity to establish meningitis and multi-focal brain abscesses (7, 8), the effector responses of primary micro-glia to this pathogen have not yet been described. Both C. koseri clinical isolates evaluated in this study induced microglial activation in a dose-dependent manner typified by NO, IL-1β, TNF-α, CXCL2, and CCL2 production within 24 h following bacterial exposure (Fig. 1). There were no significant differences between the stimulatory effects of the two C. koseri strains examined despite the divergent sites from which these organisms were originally recovered (CSF and trachea). Cell viability assays revealed that these C. koseri isolates were not toxic to primary microglia at any of the doses examined (Fig. 1F). Similar effects were observed with live C. koseri strains except the time course of microglial activation was accelerated and could not be examined past 6 h following C. koseri exposure due to overwhelming bacterial replication (Fig. 2).
Since C. koseri is a Gram-negative pathogen, we next examined the effects of bacterial stimulation on the expression of CD14 and MyD88, two molecules known to play a pivotal role in transducing activation signals in response to LPS (48). We also attempted to examine microglial TLR4 expression; however, we were unable to successfully detect TLR4 by Western blotting with any of the commercially available Abs tested. Low levels of CD14 were detected on unstimulated microglia; however, following treatment with either C. koseri strain receptor expression increased dramatically (Fig. 3A). In contrast, little change in MyD88, a central adaptor molecule for TLR signaling, was observed in response to bacterial treatment (Fig. 3B). These results demonstrate that microglia express a repertoire of molecules involved in mediating cell activation in response to Gram-negative pathogens.
Since C. koseri is a Gram-negative organism, it was envisioned that its LPS motifs would trigger microglial activation via TLR4. However, despite the tropism of this pathogen for the CNS parenchyma, no studies to date have examined the importance of TLR signals in eliciting microglial activation in response to C. koseri.It was also possible that alternative bacterial moieties, such as lipoproteins, flagellin, and bacterial DNA, may also contribute to microglial activation via engagement of TLR2 (49), TLR5 (50), and TLR9 (51), respectively. To assess the functional importance of TLR4 in eliciting inflammatory mediator release in response to C. koseri, we compared microglia isolated from TLR4 mutant and WT mice. The production of several molecules, including NO, IL-1β, TNF-α, CXCL2, and CCL2 was significantly attenuated in TLR4 mutant microglia compared with WT cells in response to heat-inactivated C. koseri (Fig. 4). Despite a pivotal role for TLR4 in C. koseri-dependent microglial activation, TLR4-independent effects were also evident since low but significant levels of inflammatory mediator production were still detected in TLR4 mutant microglia. The reduction in proinflammatory mediator expression in TLR4 mutant microglia following C. koseri exposure was not the result of cytotoxic effects or differences in cell density (Fig. 4F). Since the production of virulence factors released by viable bacteria may conceivably affect microglial activation via TLR4-dependent pathways, we also examined responses to viable C. koseri strains. The results still demonstrated a pivotal role for TLR4-dependent signaling in microglial responses to live C. koseri (Fig. 5). Collectively, these data suggest that TLR4-mediated pathway(s) are pivotal for C. koseri-dependent microglial activation and the generation of a robust proinflammatory response. However, the residual activation detected in TLR4 mutant microglia suggests that alternative receptors besides TLR4 are required to achieve maximal microglial activation.
MyD88 is the central adaptor molecule involved in the majority of TLR signal transduction pathways as well as the IL-1R and IL-18R. TLR4 engagement by LPS facilitates the activation of both MyD88-dependent and MyD88-independent signaling pathways, leading to the expression of several gene products, including inflammatory cytokines and chemokines (21). The relative importance of MyD88-dependent vs -independent mechanisms in microglial responses to C. koseri has not yet been examined and could reveal interesting insights into the role of alternative pathways of microglial activation. The expression of numerous proinflammatory mediators, including NO, IL-1β, TNF-α, CXCL2, and CCL2, was significantly attenuated in MyD88 KO microglia following C. koseri exposure compared with WT cells (Fig. 6). Despite the pivotal role of MyD88-dependent signals in regulating inflammatory mediator release, residual production of these molecules was still detected in MyD88 KO microglia, suggesting that MyD88-independent mechanisms also contribute to C. koseri recognition and cytokine signaling in microglia. The inability of MyD88 KO microglia to respond to C. koseri was not attributable to differences in cell number or bacterial cytotoxicity (Fig. 6F). Similar responses were obtained with MyD88 KO microglia in response to live C. koseri (Fig. 7). The fact that C. koseri-mediated microglial activation was more attenuated with the loss of MyD88-compared with TLR4-dependent signaling implicates additional MyD88-driven pathways for inducing maximal responses to this Gram-negative pathogen.
Several findings have suggested that intracellular survival of C. koseri in macrophages may play a role in primary colonization and pathogen dissemination in the CNS. First, macrophages harboring C. koseri were found infiltrating the periventricular parenchyma in an infant rat model of infection (10). Second, intracellular survival and replication of C. koseri has been demonstrated in the human monocyte cell line U937 (14). However, despite its tropism for the CNS parenchyma, no studies have examined whether C. koseri can persist in primary microglia or macrophages, or whether these cells types are permissive for intracellular replication of bacteria. To examine these possibilities, we performed gentamicin protection assays using primary mouse microglia and macrophages. Although neither cell type was capable of supporting C. koseri replication over the initial 72-h interval examined, microglia and macrophages did provide a permissive environment for bacterial survival as evident by the failure to completely clear viable intracellular organisms (Fig. 8, A and B). Although a slight increase in intracellular burdens was observed over time with strain 4277 in microglia, these differences did not achieve statistical significance (Fig. 8A). The number of surviving extracellular bacteria in the absence of microglia or macrophages was negligible, indicating effective killing of extracellular organisms by gentamicin (data not shown). Interestingly, the C. koseri isolate recovered from the CSF (strain 4036) demonstrated enhanced intracellular survival compared with the tracheal pathogen (strain 4277). The human U937 monocyte cell line supported intracellular replication of both C. koseri isolates, in agreement with previous reports (Fig. 8C) (14, 52).
In the setting of C. koseri CNS infection, resident microglia and macrophages likely encounter a complex milieu of molecules, both host- and bacterial-derived, that may subsequently modify their antibacterial responses. For example, cell wall components are shed during normal bacterial growth and as such, LPS would be available to exert its proinflammatory effects that could feed back in an autocrine/paracrine manner to influence the ability of microglia/macrophages to regulate bacterial growth and/or survival. To evaluate this possibility, primary microglia, macrophages, and the U937 cell line were treated with 100 ng/ml LPS 6 h before the addition of live C. koseri strains for GPA. This LPS dose and timing regimen is sufficient to elicit detectable proinflammatory mediator release (i.e., IL-1β, CXCL2) from cells. Unexpectedly, the intracellular survival and persistence of strain 4277 (tracheal isolate) was augmented in all three cell types over the 72-h interval examined (Fig. 8, C–E). Interestingly, with LPS pretreatment, the intracellular survival of strain 4036 (CSF isolate) was decreased in macrophages (Fig. 8E) but enhanced in microglia (Fig. 8D). These findings suggest that activated macrophages are more efficient than microglia in killing C. koseri strain 4036 upon phagocytosis.
To more fully define the duration of C. koseri intracellular survival in microglia, we extended our analysis out to 168 h postinfection to determine whether intracellular bacteria persist or decline over time. Interestingly, although the number of viable intracellular C. koseri eventually declined over the first 120 h following infection, bacterial burdens showed an upward trend at the latest time point examined (i.e., 168 h), which occurred with both C. koseri strains tested (Fig. 9). We also investigated whether the effect of LPS pre-treatment on intracellular C. koseri survival in microglia was dose-dependent. In general, we were unable to demonstrate any differential effects of LPS dose on the numbers of intracellular bacteria (Fig. 9). However, a 6 h LPS pretreatment period, regardless of dose, led to a dramatic increase in the number of phagocytized bacteria compared with non-stimulated microglia. This disparity in intracellular bacteria between LPS pretreated and nontreated microglia persisted throughout the 168 h time interval examined (Fig. 9).
We next examined the extent of proinflammatory mediator expression by microglia during intracellular C. koseri infection. The levels of CXCL2 and CCL2 were highest at 24 h following bacterial exposure, decreased dramatically from 48 to 120 h, and began to rise at 168 h following infection (Fig. 10, C and D, respectively), which coincided with the upward trend in viable intracellular bacteria (Fig. 9). In contrast to chemokine production, IL-1β and TNF-α were only detected during the first 72 h following C. koseri exposure and decreased dramatically in a time-dependent manner (Fig. 10, A and B, respectively). Overall, these results suggest that microglia may serve as a reservoir for C. koseri survival/replication for at least a 1-week period and that continued chemokine production may be triggered by an as of yet unidentified intracellular PRR.
Among the Gram-negative enteric bacilli responsible for meningitis in neonates, C. koseri is an infrequent but formidable etiological agent based on its propensity to disseminate and establish multifocal brain abscesses (7, 8). Several studies have focused on the pathogenesis of this devastating disease using an infant rat model of C. koseri infection. However, despite the tropism of C. koseri for the CNS, no studies to date have examined microglial responsiveness to this pathogen or the receptor signaling pathways leading to cell activation and subsequent proinflammatory mediator release. In the current report we demonstrate that TLR4- and MyD88-dependent pathways are critical for microglial activation in response to C. koseri and that microglia can serve as an intra-cellular reservoir for bacterial survival.
TLRs are type I integral membrane glycoproteins that facilitate pathogen recognition of motifs conserved across broad classes of microorganisms. Pertinent to C. koseri, we were interested in examining the functional importance of TLR4 based on its well-described role in LPS recognition. Interestingly, although microglia utilized TLR4 signaling to elicit the production of numerous proinflammatory mediators, including NO, IL-1β, TNF-α, CXCL2, and CCL2, evidence for a TLR4-independent recognition pathway was also apparent based on the residual expression of these molecules by TLR4 mutant microglia. One explanation that may account for this finding is the triggering of alternative TLR by antigenic motifs contained within C. koseri including lipoproteins, flagellin, and bacterial DNA that could engage TLR2, TLR5, and TLR9, respectively. Indeed, recent studies have identified TLR2 as an important TLR4-driven sensor of Gram-negative bacterial infection both in the CNS and the periphery (53, 54), and an earlier study has revealed an important role for the C. koseri flagellar molecule fliP in regulating IL-10 production and virulence (15). This scenario does appear plausible because the residual activation observed in MyD88 KO microglia in response to C. koseri was barely detectable compared with the significant levels of inflammatory mediators still produced by TLR4 mutant microglia. The low levels of inflammatory mediator production remaining in MyD88 KO microglia in response to C. koseri are similar to a previous study by our laboratory demonstrating residual responses to LPS (55). A recent report (56) provides added evidence to support our findings, where LPS stimulation was found to induce several chemokine and cytokine genes such as CCL3, CCL4, CXCL2, and TNF-α in MyD88-deficient cells as evaluated by microarray analysis. This report demonstrated that the TIR domain-containing adaptor inducing IFN-β-dependent pathway activated TNF-α production and secretion via a NF-κB-independent manner and suggested that secreted TNF-α acts in an autocrine manner to induce delayed NF-κB activation. Since depletion of IRF-3 was found to impair NF-κB activation, IRF-3 has been implicated as a mediator of TNF-α activation via a MyD88-independent pathway. A more recent study (57) has demonstrated an alternative pathway of TLR4 signaling in response to Gram-negative bacteria in human bladder epithelial cells involving cAMP. However, it is uncertain whether a similar mechanism exists in microglia, making this possibility highly speculative.
An alternative explanation to account for the residual proinflammatory mediator production in TLR4 mutant and MyD88 KO microglia is that C. koseri also engages additional PRR to trigger cell activation. This possibility is not unexpected given the essential nature of an effective CNS antibacterial response to contain infection and the fact that natural polymorphisms in TLR4 have been reported in the human population and are proposed to result in aberrant cytokine responses and increased susceptibility to Gram-negative infections (58). Some candidates include members of the intracellular NOD-like receptor family of proteins, scavenger receptors, and C-type lectin receptors, but these wait testing in future studies to assess their contributions toward microglial cytokine signaling in response to C. koseri. The former appears to represent a viable possibility based on our findings that chemokine production by microglia harboring intracellular C. koseri began to increase at 168 h following bacterial exposure, which coincided with the amplification of viable intracellular bacteria.
During CNS infections with C. koseri, macrophages appear to harbor a significant number of intracellular bacteria, suggesting that they may be exploited by the organism for intracellular replication and/or survival. Since microglia may also serve as an obvious target during C. koseri infection in the brain, we wanted to examine whether these cells would enable bacterial propagation or exert antimicrobial activity. We were able to confirm earlier findings by Townsend et al. (14) where the human monocyte cell line U937 was capable of supporting the intracellular replication of C. koseri. However, C. koseri did not multiply to a considerable extent in primary microglia during the initial 72 h following bacterial exposure, although the organism was capable of survival as evidenced by the persistence of viable intracellular bacteria. We next extended the time intervals following microglial C. koseri exposure out to 168 h to ascertain whether intracellular bacteria were eventually cleared or persisted. Interestingly, intracellular C. koseri burdens declined between 72 and 120 h after bacterial exposure, whereupon a rebound in viable bacteria was detected suggestive of ongoing replication. The reason for this delayed increase in intracellular C. koseri is uncertain but may be influenced by the lengthy infection period where maximal microglial antibacterial responses may have begun to wane. Importantly, this organism is capable of surviving within the phagolysosome (14), which is an important virulence determinant for other intracellular pathogens such as Leishmania (59) and Coxiella (60).
Next, we were interested in determining whether the intracellular persistence of C. koseri was unique to microglia by comparing responses with primary macrophages. Similar results were obtained with macrophages, suggesting that C. koseri does not co-opt either cell type to replicate to a significant extent; however, our results do indicate that the organism is capable of thwarting antibacterial effector responses to prolong intracellular survival. Currently, the reason(s) to account for the preferential replication of C. koseri in U937 monocytes but not microglia or macrophages are not known; however, several explanations can be considered. Differences in C. koseri isolates, immortalized cell lines vs primary cells, species of origin, or maturational state of the target cell may all impact whether microglia, macrophages, or monocytes can support C. koseri intracellular replication. For example, compared with more terminally differentiated cell types (i.e., microglia and macrophages) monocytes (as represented by U937 cells) signify a more immature state and possibly exert less antimicrobial activity, which might make them more permissive for intracellular bacterial replication. This is supported by earlier studies demonstrating that overexpression of inducible NO synthase resulted in enhanced bactericidal activity to Leishmania and Brucella in U937 monocytes (52, 61). Although microglial and macrophage antibacterial responses to C. koseri are not highly effective, as demonstrated by the failure to clear intracellular burdens by 168- and 72-h postinfection, respectively, their response is sufficient to limit bacterial replication. With regard to pathology, it is possible that during C. koseri sepsis, monocytes harboring intracellular bacteria serve as a “Trojan horse” to introduce organisms into the CSF and periventricular parenchyma following CSF extravasation. Once CNS infection is established, resident microglia and newly recruited macrophages can internalize C. koseri, but antimicrobial killing mechanisms are ineffective, permitting bacterial persistence. Therefore, an endogenous reservoir for C. koseri may be present within the CNS to facilitate the dissemination and persistence of infection.
One intriguing finding that surfaced during the course of these studies relates to the differential effects of cell activation on the ability of C. koseri to survive intracellularly in microglia vs macrophages. Specifically, preactivation of macrophages with LPS led to a more robust killing of the C. koseri CSF isolate, whereas this organism persisted in activated microglia. In contrast, the tracheal C. koseri strain was capable of intracellular replication in LPS activated microglia, whereas the bacteria survived in macrophages. Overall, it appeared that LPS treatment resulted in enhanced bacterial clearance/containment in macrophages compared with microglia, which is in agreement with the more potent bactericidal activity of the former. Importantly, the dose of LPS used to treat microglia/macrophages before C. koseri exposure (100 ng/ml) well exceeds that reported to induce LPS hyporesponsiveness or tolerance (62), suggesting that a tolerogenic mechanism is unlikely to be involved in the differential abilities of microglia and macrophages to respond to live C. koseri. This is supported by the fact that similar responses were observed with a lower concentration of LPS (i.e., 1 ng/ml). A final note that should be acknowledged is that we use the term “persistence” to reflect a relatively consistent number of intracellular bacteria over time. However, in reality this finding could also represent a net balance between bacterial killing and de novo replication, a possibility that was not examined in the current study.
C. koseri was a potent stimulus of microglial activation as typified by the ability of both live and heat-killed organisms to induce the expression of numerous inflammatory mediators, including NO, TNF-α, IL-1β, CXCL2, and CCL2. With regard to the potential roles of these factors in disease pathogenesis, NO is a potent antibacterial mediator that induces DNA damage, lipid peroxidation, and protein nitrosylation. The importance of NO scavenging from the bacterial perspective is apparent from the fact that many bacterial species encode for NO metabolizing enzymes (63, 64). TNF-α and IL-1β are important for activating the vascular endothelium of the blood-brain barrier as well as inducing the expression of antibacterial effector molecules such as NO and TLRs (35). Both CXCL2 and CCL2 are chemokines that are important for neutrophil and monocyte/macrophage/lymphocyte recruitment, respectively. Since C. koseri infection in the brain is typified by the presence of all these immune cell populations, it is likely that chemokines produced by activated microglia participate in cell recruitment into the infected CNS. By virtue of its ability to induce monocyte migration, CCL2 production by microglia could inadvertently enhance C. koseri colonization to the brain by recruiting additional cells harboring viable intracellular bacteria. Interestingly, both CXCL2 and CCL2 production were increased in microglia as the number of intracellular C. koseri began to increase slightly at 168 h following bacterial exposure, suggesting that chemokine induction may serve as a means to elicit additional monocytes/macrophages to continue a productive CNS infection.
In summary, the results presented in this study demonstrate that TLR4- and MyD88-dependent pathways play an essential role in signaling proinflammatory mediator release by microglia in response to C. koseri. The ability of bacteria to survive within microglia and infiltrating macrophages could represent a pathogenic mechanism to ensure C. koseri dissemination and persistence in the CNS.
We thank Dr. Shizuo Akira for providing the MyD88 KO mice and Gail Wagoner and Teresa Fritz for maintaining the TLR4 mutant and MyD88 KO mouse colonies.
1This work was supported by the National Institute of Neurological Disorders and Stroke, National Institutes of Health Grant R01 NS055385 (to T.K.) and the National Institute of Neurological Disorders and Stroke supported Core Facility at UAMS (P30 NS047546). S.L. was the recipient of a Graduate Student Research Funds award from the University of Arkansas for Medical Sciences Graduate School.
3Abbreviations used in this paper: PRR, pattern recognition receptor; CSF, cerebro-spinal fluid; GPA, gentamicin protection assay; KO, knockout; MOI, multiplicity of infection; WT, wild type.
The authors have no financial conflict of interest.