|Home | About | Journals | Submit | Contact Us | Français|
The gram-negative, facultative intracellular bacterium Francisella tularensis causes acute, lethal pneumonic disease following infection with only 10 CFU. The mechanisms used by the bacterium to accomplish this in humans are unknown. Here, we demonstrate that virulent, type A F. tularensis strain Schu S4 efficiently infects and replicates in human myeloid dendritic cells (DCs). Despite exponential replication over time, Schu S4 failed to stimulate transforming growth factor β, interleukin-10 (IL-10), IL-6, IL-1β, IL-12, tumor necrosis factor alpha, alpha interferon (IFN-α), and IFN-β throughout the course of infection. Schu S4 also suppressed the ability of directly infected DCs to respond to different Toll-like receptor agonists. Furthermore, we also observed functional inhibition of uninfected bystander cells. This inhibition was mediated, in part, by a heat-stable bacterial component. Lipopolysaccharide (LPS) from Schu S4 was present in Schu S4-conditioned medium. However, Schu S4 LPS was weakly inflammatory and failed to induce suppression of DCs at concentrations below 10 μg/ml, and depletion of Schu S4 LPS did not significantly alleviate the inhibitory effect of Schu S4-conditioned medium in uninfected human DCs. Together, these data show that type A F. tularensis interferes with the ability of a central cell type of the immune system, DCs, to alert the host of infection both intra- and extracellularly. This suggests that immune dysregulation by F. tularensis operates on a broader and more comprehensive scale than previously appreciated.
Francisella tularensis is a gram-negative, facultative intracellular bacterium and the causative agent of tularemia. Although the bacteria was identified nearly 100 years ago, the nature of its interaction with host cells and tissues has remained largely undefined until recent times. The unfortunate realization of the potential of F. tularensis as a biological weapon has resulted in a resurgence of research on tularemia. Much of our recent understanding of tularemia pathogenesis has come from manipulation of the mouse model of Francisella infections. While these murine studies have yielded many important advances in our understanding of tularemia pathogenesis, very little is understood about how Francisella, especially the most virulent type A strains such as Schu S4, interacts with human cells.
Dendritic cells (DCs) serve as a central cell type in the immune system, bridging the innate and adaptive immune response to effectively eradicate invading pathogens. Thus, it is unsurprising that many microbes have developed mechanisms by which to modulate these cells to evade host immune responses (4, 30). F. tularensis is no exception in its ability to modify DC responsiveness to infection. For example, in the mouse model, both attenuated and virulent strains of F. tularensis fail to induce the production of cytokines required for mobilization of key effector cells (7, 8, 14, 50). However, only virulent F. tularensis also prevents increased expression of key surface receptors required for antigen presentation in the lung in vivo (7). It is not known if virulent F. tularensis has the same effect on human DCs in vitro.
A few recently published studies have focused on the attenuated F. tularensis live vaccine strain, LVS, and F. tularensis subsp. novicida in human cells. However, both of these subspecies are attenuated in humans. Due to this attenuation, it is unsurprising that following infection with these strains, human DCs and monocytes are activated to produce proinflammatory cytokines and upregulate receptors associated with antigen presentation and that DCs can easily control bacterial replication when exposed to the appropriate stimulus following infection (22, 26, 27, 31). However, this does not necessarily reflect how human cells, in particular DCs, might behave when infected with a more formidable and virulent strain of F. tularensis.
Here, we provide data supporting our hypothesis that, unlike more attenuated Francisella species and despite exponential replication, type A F. tularensis fails to induce production of both proinflammatory and anti-inflammatory cytokines by human DCs. As observed in the murine model, this suppression of cytokine secretion mediated by F. tularensis was an active process requiring live organisms for optimal effect. Furthermore, we made the surprising observation that F. tularensis also inhibited responsiveness of uninfected bystander DCs. At high concentrations (100-fold greater than Escherichia coli lipopolysaccharide [LPS]), purified LPS from F. tularensis strain Schu S4 was capable of inducing minimal endotoxin tolerance to E. coli LPS. However, our data suggest that Schu S4 LPS present in tissue culture medium was not the primary mediator of the suppression observed in human DCs. Rather, an undefined, moderately heat-stable molecule shed or secreted by Schu S4 significantly inhibited human DC responsiveness to secondary stimuli. Furthermore, our evidence suggests that modulation of DCs by Schu S4 components in the extracellular space is different from that observed among infected cells. Together, these results correlate with our in vivo observation that Schu S4 potently suppresses cells critical for initiation of innate immunity. Data presented herein also represent an important step forward in understanding how virulent F. tularensis manipulates the human host and provides a novel mechanism by which this bacterium successfully suppresses host cells in a global sense.
F. tularensis strain Schu S4 was kindly provided by Jeannine Peterson (Centers for Disease Control, Fort Collins, CO). Green fluorescent protein (GFP)-expressing Schu S4 was generated as previously described (12). Schu S4 was cultured in modified Mueller-Hinton broth ([MMH] Mueller-Hinton broth supplemented with CaCl2, MgCl, 0.1% glucose, 0.025% ferric pyrophosphate and 2% Isovitalex) at 37°C with constant shaking overnight, aliquoted into 1-ml samples, frozen at −80°C, and thawed just prior to use, as previously described (7). Titers were determined in frozen stocks by enumerating viable bacteria from serial dilutions plated on MMH agar as previously described (9, 19). The number of viable bacteria in frozen stock vials varied less than 5% over a 10-month period.
Killed Schu S4 was generated by resuspending 1 ml of 2 × 109 CFU/ml Schu S4 in 1% paraformaldehyde. Following overnight incubation at 4°C, bacteria were washed three times in phosphate-buffered saline (PBS), resuspended in 1 ml of PBS, and stored at 4°C. Killed Schu S4 was detected via intracellular staining with AlexaFluor 647-labeled anti-F. tularensis LPS(anti-LPSFt) antibody (US Biological, Swampscott, MA). Isotype control antibody (R&D Systems) labeled with AlexaFluor 647 served as a negative control for the presence of killed Schu S4. Anti-LPSFt and isotype antibodies were labeled using an AlexaFluor 647 monoclonal labeling kit (Invitrogen) according to manufacturer's instructions.
Schu S4-conditioned medium (S4CM) was generated following incubation of 107 CFU of F. tularensis Schu S4 overnight in RPMI medium supplemented with 10% heat-inactivated fetal calf serum, 0.2 mM l-glutamine, 1 mM HEPES buffer, and 0.1 mM nonessential amino acids (all from Invitrogen) (complete RPMI [cRPMI] medium) at 37°C in 7% CO2. Cultures were collected and filter sterilized through a 0.2-μm-pore-size syringe filter. As indicated, some aliquots of S4CM were heated at 95°C for 1 h [S4CM(95C)]. F. tularensis Schu S4 LPS was generated as previously described (10, 15). Ultrapure E. coli LPS was purchased from InvivoGen (San Diego, CA).
Human myeloid DCs were differentiated from peripheral blood monocytes that had been subjected to apheresis as previously described (6, 42). Briefly, monocytes that had been subjected to apheresis were enriched using Ficoll-Paque (GE Healthcare). CD14+ progenitor cells were enriched via negative selection using a Dynabeads MyPure Monocytes Kit for untouched human cells, per the manufacturer's instructions (Invitrogen). Cells were resuspended at 5 × 105/ml in cRPMI medium supplemented with 100 ng/ml granulocyte-macrophage colony-stimulating factor and 20 ng/ml interleukin-4 (IL-4) (both from Peprotech), plated at 2 ml/well in six-well plates, and incubated at 37°C in 5% CO2. On day 2 of culture, 0.5 ml of medium was removed and replenished with 1 ml of cRPMI medium supplemented with 100% cytokines. All cells were used on day 4 of culture. The resulting differentiated DCs were >97% CD1a+/DC-specific intercellular adhesion molecule-grabbing nonintegrin (DC-SIGN)-positive and <1% CD14−. Human blood cells were collected from anonymous volunteers under a protocol reviewed and approved by the NIH Institutional Biosafety Committee.
DCs were infected at a multiplicity of infection (MOI) of 50 with F. tularensis strain Schu S4. DCs were removed from their original cultures and pelleted. The resulting DC-conditioned medium was reserved for replating of cells. DC numbers were adjusted to 2 × 107 to 3 × 107/ml in reserved DC medium. Cells treated with medium alone served as negative controls. Schu S4 was added, and cells were incubated at 37°C in 7% CO2 for 1 h, washed once, and then incubated with 50 μg/ml gentamicin (Invitrogen) for 45 min to kill extracellular bacteria. Then, cells were washed extensively and adjusted to 5 × 105 cells/ml in reserved DC medium and then plated at 1 ml/well in 24-well tissue culture plates. As indicated, 10 μg/ml of anti-IL-10, anti-pan transforming growth factor β (TGF-β), or the appropriate isotype control (all from R&D Systems) was added to each well immediately after infection of DCs. In additional experiments, 4 μg/ml gentamicin was added 12 h prior to stimulation of DCs with Toll-like receptor (TLR) agonists. This concentration has been previously shown to be the MIC for control of F. tularensis replication (3). In these experiments, immediately prior to the addition of TLR agonist, an aliquot of the medium was collected, serially diluted, and plated onto MMH agar to enumerate extracellular bacteria. At the indicated time points, ultrapure LPS (TLR4 agonist), zymosan (TLR2 and C-type lectin agonist), or tripalmitoyl-Cys-Ser-(Lys)4 (Pam3CSK4; a TLR2 agonist) (10 ng/ml, 10 μg/ml, and 1 μg/ml, respectively) (InvivoGen) was added to the appropriate wells. Cultures were incubated for 24 to 72 h at 37°C in 7% CO2. Supernatants were collected at each time point and stored at −80°C. To determine uptake and replication of bacteria, DCs were harvested from each well, pelleted at 1,200 rpm for 5 min, and incubated with 50 μg/ml gentamicin for 45 min at 37°C in 7% CO2. Cells were then washed extensively and lysed following incubation in water for 5 min. The last wash from each sample was plated onto MMH agar to confirm depletion of residual extracellular bacteria. There were no bacteria detected in these samples in each experiment tested. Cell lysates were serially diluted, plated onto MMH agar plates, and incubated at 37°C in 7% CO2 for approximately 48 h and individual colonies were enumerated. Replication of GFP-Schu S4 was also analyzed by flow cytometry (described below).
As indicated, uninfected DCs were resuspended in S4CM or S4CM(C95). DCs were incubated overnight, followed by addition of ultrapure LPS as described above. DCs were then stained for intracellular cytokines as described below.
Experimental and control groups were all performed in triplicate. The standard error of the mean (SEM) and statistical significance between treatment groups were determined by analysis of variance, followed by Tukey-Kramer's comparison of means.
Schu S4 LPS was depleted from S4CM using anti-LPSFtantibodies (US Biological). Briefly, S4CM was incubated with 5 μg/ml anti-LPSFtantibodies or anti-mouse immunoglobulin G2a (IgG2a) as an isotype control for 1 h at 4°C with constant rotation. To remove antigen/antibody complexes, magnetic beads bound with anti-mouse IgG antibodies (Invitrogen) were added to treated S4CM and allowed to incubate for 1 h at 4°C with constant rotation. Samples were then placed on a Dynal MPC-50 magnet (Invitrogen) to immobilize complexed beads, and the resulting supernatant was collected. Depletion of Schu S4 LPS from S4CM was monitored by Western blotting and enzyme-linked immunosorbent assays (ELISAs) (described below). The indicated samples were prepared per manufacturer's instructions and were first resolved on a 12% Bis-Tris sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel, followed by blotting onto polyvinylidene difluoride using XCell Mini Cell and XCell II Blot modules (Invitrogen). Blots were blocked with 5% nonfat milk in Tris-buffered saline plus Tween 20 (TBST), washed, and then incubated with anti-LPSFtantibody (US Biological) diluted 1:200 in TBST. Blots were washed and incubated with horseradish peroxidase (HRP)-conjugated anti-mouse IgG to detect primary antibody bound to the membrane. Blots were washed again and incubated with SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL) to detect HRP. HRP activity was detected using a Kodak 4000MM imager and Kodak MI software (Kodak).
Quantification of Schu S4 LPS in S4CM was performed by ELISA. Briefly, Schu S4 LPS was coated onto Immunlon 2HB plates in 0.1 mm sodium bicarbonate buffer overnight at 4°C. Plates were washed with PBS-0.2% Tween 20, and samples were added. A standard curve was generated by serially diluting purified Schu S4 LPS and adding it to the plate. Plates were incubated for 1 h at room temperature and washed as above. Mouse anti-Francisella LPS antibody (US Biological) was added, and plates were incubated for 1 h at room temperature. Plates were washed, incubated with HRP-conjugated donkey anti-mouse antibody (Jackson ImmunoResearch), and washed again. Bound antibody was detected following addition of 1-Step ABTS [2,2′-azinobis(3-ethylbenzthiazolinesulfonic acid)] (Pierce) and analysis of absorption at 450 nm using a Dynex MRX Revelation instrument and Revelation software (Dynex Technologies, Chantilly, VA).
GFP-Schu S4 and intracellular cytokines were detected by flow cytometry as previously described (21). At the indicated time points in each figure and 1 h after addition of ultrapure LPS (10 ng/ml), zymosan (10 μg/ml) or Pam3CSK4 (1 μg/ml) and 5 μg/ml brefeldin A (BFA; Invitrogen) was added to DC cultures. Cells were then incubated at 37°C in 7% CO2 for 5 h. After incubation, cells were washed once and resuspended in Flow Cytometry Staining buffer (fluorescence-activated cell sorter [FACS] buffer; eBioscience). Then, cells were fixed in 2% paraformaldehyde in PBS for 10 min at 37°C in 7% CO2 and washed twice more in Perm buffer (FACS buffer supplemented with 0.25% saponin [Sigma-Aldrich]). Next, cells were incubated at room temperature with anti-human tumor necrosis factor alpha (TNF-α) (phycoerythrin [PE]conjugated; clone Mab11), anti-human IL-12/IL-23p40/70 (PE coupled; clone C8.6) or mouse IgG1 (PE coupled) as an isotype control (all from eBioscience). Cells were incubated with cytokine or isotype control antibodies for 20 min at room temperature. Cells were washed twice in Perm buffer, fixed in 1% paraformaldehyde for 30 min, and then resuspended in FACS buffer at 4°C until analysis. Cells were acquired and analyzed using a CyFlow ML flow cytometer and FlowMax software (Partec).
Culture supernatants were assayed for the presence of TNF-α, IL-6, IL-10, IL-1β, IL-12/IL-23p40, TGF-β, alpha interferon (IFN-α), and IFN-β by ELISA using commercially available kits according to the manufacturer's instructions (R&D Systems and PBL Interferon Source).
DCs were infected with GFP-Schu S4 as described above and then adhered to a glass slide using a Shandon Cytospin 4. Cells were fixed in 3% paraformaldehyde for 30 min at 37°C in 5% CO2. Cells were washed with PBS and stained for cleaved caspase-3 using rabbit anti-cleaved caspase-3 antibody (Cell Signal Technology) and then AlexaFluor 568 anti-rabbit IgG, following the manufacturer's instructions exactly. Samples were visualized using a Carl Zeiss LSM 510 confocal scanning laser microscope for quantitative analysis and image acquisition. Confocal images of 1,024 by 1,024 pixels were acquired and assembled using Adobe Photoshop CS2 software (Adobe Systems, San Jose, CA).
Statistical differences between two groups were determined using an unpaired t test with the significance set at a P value of <0.05. For comparison between three or more groups, analysis was done by one-way analysis of variance followed by Tukey's multiple comparisons test with significance determined at a P value of <0.05.
As a facultative intracellular pathogen, F. tularensis Schu S4 can replicate in both the intra- and extracellular environments. We along with others have shown that more attenuated LVS replicates exponentially in both mouse and human DCs (5, 8). However, the efficiency of Schu S4 replication in human myeloid DCs has not been described. Since infection and intracellular replication are likely to be an integral element of Schu S4 pathogenesis, we first determined if human DCs were permissive to uptake and growth of Schu S4. Schu S4 was readily taken up by DCs and replicated intracellularly exponentially over time in human DCs (Fig. (Fig.1A).1A). Due to the low numbers of bacteria recovered from the earliest time point, it was unlikely that all DCs in the cultures were infected. To monitor the number of cells infected at any given time, we also infected DCs with GFP-expressing Schu S4 (GFP-Schu S4) and analyzed DCs by flow cytometry. Corresponding with recovery of low numbers of viable Schu S4 from DC lysates, only about 5% of DCs were positive for GFP-Schu S4 24 h after infection. In correlation with exponential intracellular replication observed in cell lysates, by 48 h 20 to 25% of cells were positive for GFP-Schu S4, and by 72 h after infection nearly all DCs (>80%) were positive for GFP-Schu S4 (Fig. 1B and C). GFP-Schu S4 detected by flow cytometry was primarily intracellular since the addition of trypan blue prior to analysis did not significantly change the percentage of infected cells (data not shown).
One DC function is antigen presentation to T cells. This function is concomitant with increased expression of several surface receptors including HLA-DR, HLA-ABC, CD86 and CD83, a process also known as phenotypic maturation (42). One evasion strategy employed by successful pathogens is to interrupt this phenotypic maturation. Thus, we evaluated changes in DC surface expression of several DC receptors following infection with Schu S4. Forty-eight hours after infection, Schu S4-infected DC cultures had significantly more (P < 0.01) HLA-DR, HLA-ABC, CD83, and CD86 on their cell surfaces than untreated controls (Fig. (Fig.2).2). However, this increase was significantly less than that observed with cells stimulated with E. coli LPS (Fig. (Fig.2).2). This increase in receptor expression reflects changes in the entire cell population (uninfected and infected cells). However, similarly increased expression of HLA-DR, CD83, and CD86 was observed among Schu S4-infected cells, while HLA-ABC expression was decreased in approximately 20% of these infected cells (data not shown).
In addition to phenotypic maturation, another primary function of DCs is secretion of cytokines to aid in alerting or regulating the immune response to invading pathogens. DCs recognize microbial products via an array of pathogen recognition receptors and secrete either pro- or anti-inflammatory cytokines following ligation of these receptors. Thus, we next determined if human DCs could respond to infection with Schu S4 via secretion of various cytokines previously shown to be crucial in developing effective immune responses against F. tularensis infections (13, 19).
Despite exponential replication and phenotypic maturation, we did not detect TNF-α, IL-12/IL-23p40, IL-6, or IL-1β at any time point (24, 48, or 72 h) after infection (Fig. (Fig.33 and data not shown). We along with others have observed induction of anti-inflammatory cytokines and type I IFN by F. tularensis following infection of antigen-presenting cells (7, 8, 26, 51). As expected, DCs responded to E. coli LPS by secreting IL-12, TNF-α, IL-1β, and TGF-β into culture supernatant (Fig. (Fig.3).3). However, we also failed to detect IL-10, TGF-β, prostaglandin E2, IFN-α, and IFN-β at any time point after infection of human DCs with Schu S4 (Fig. (Fig.33 and data not shown).
Given the striking lack of cytokine secretion in Schu S4-infected DCs, we next determined if Schu S4 was simply evading detection by human DCs or if it actively suppressed the ability of human DCs to respond to pathogen stimuli. At various time points after infection, we stimulated DCs with ultrapure E. coli LPS. DCs infected with Schu S4 failed to produce TNF-α or IL-12/IL-23p40/70 in response to E. coli LPS (Fig. 4A and B). A similar pattern of poor responsiveness was observed when infected cultures were exposed to zymosan and Pam3CSK4, suggesting that the inhibition among infected DCs was not limited to responsiveness to E. coli LPS but, rather, may be a more general phenomenon applicable to multiple pattern recognition pathways (Fig. (Fig.4B4B and data not shown).
To determine if live Schu S4 cells were required for the observed inhibition of responsiveness to E. coli LPS, we next examined the ability of paraformaldehyde-killed Schu S4 to induce or suppress production of cytokines by human DCs. Although dead Schu S4 did not elicit production of TNF-α or IL-12/IL-23p40/70, cells treated with killed Schu S4 were able to respond to LPS, as indicated by their ability to produce these cytokines in response to E. coli LPS (Fig. (Fig.5).5). This suggested that either metabolically active Schu S4, a molecule sensitive to paraformaldehyde degradation, or a molecule removed from bacterial preparations following washing of Schu S4 after fixation was responsible for suppression of DCs. Furthermore, the observation that killed Schu S4 was unable to elicit production of TNF-α and IL-12 from human DCs suggests that perhaps dying Schu S4 in vivo also fail to stimulate an inflammatory response.
One explanation for the lack of proinflammatory cytokines and poor responsiveness to secondary stimuli is that the cells are undergoing apoptosis. Since activated DCs can display phosphatidyl serine (which binds annexin) on their surfaces, we opted to assess apoptosis by monitoring cells for cleaved caspase-3. Cleaved caspase-3 is a highly specific and early indicator of apoptosis. We did not observe increased cleaved caspase-3 staining in Schu S4-infected DC cultures compared to uninfected controls (Fig. 6A to E). In fact, among Schu S4-infected cultures, the majority of DCs positive for cleaved caspase-3 were negative for Schu S4 (Fig. 6C and D). Furthermore, there were significantly more cleaved caspase-3-positive cells in uninfected controls than in Schu S4-infected DC cultures (Fig. (Fig.6E).6E). Thus, these data suggest that apoptosis of DCs is not the mechanism by which these cells fail to secrete proinflammatory cytokines and respond to microbial stimuli following Schu S4 infection.
Following infection with live Schu S4, we made the surprising observation that production of TNF-α and IL-12 among cells negative for F. tularensis in infected cultures was significantly less than that observed in uninfected cultures stimulated with E. coli LPS (Fig. 4A and B). The percentage of Schu S4-negative cells that failed to respond to E. coli LPS could not be accounted for in the GFP-Schu S4-positive population. It is possible that the GFP-Schu S4-negative cells harbor 1 to 2 bacteria and therefore appeared uninfected. However, the idea that the presence of such a small number of bacteria per cell exerted a significant effect on the ability of DCs to respond seemed unlikely. Thus, we reasoned that live Schu S4 may be mediating suppression of bystander DCs indirectly. Following infection and replication in DCs, Francisella infects new cells by lysing the primary cell and infecting the next nearest cell. Thus, Francisella spends at least a portion of its life cycle in the extracellular milieu. Viability of the bacterium as it spreads from cell to cell in the extracellular compartment is essential for successful reinfection of bystander cells since retention of high doses (10 to 50 μg/ml) of gentamicin in cell culture over time effectively inhibits infection of bystander cells. Francisella is also capable of limited replication in tissue culture medium. Thus, the suppression we observed in Schu S4-negative bystander cells may have been a result of the replication of bacteria in the extracellular space prior to reinfection of bystander cells. To determine the contribution of extracellular replication of Schu S4 on suppression of responsiveness to TLR agonist, we infected DCs as described in the Materials and Methods and added gentamicin at the reported MIC (4 μg/ml) approximately 15 h prior to analysis of responsiveness to E. coli LPS. Addition of gentamicin at this time point significantly reduced the number of viable bacteria in the medium of infected DC cultures (Fig. (Fig.5A).5A). As expected, reduction of viable bacteria in the extracellular compartment resulted in fewer infected DCs (Fig. (Fig.7A).7A). As previously shown, Schu S4 significantly inhibited production of both TNF-α and IL-12 in directly infected and uninfected bystander cells (P < 0.01) (Fig. (Fig.7B).7B). Presence of gentamicin did not reverse the suppression of TNF-α and IL-12 production among directly infected DCs (Fig. (Fig.7B).7B). However, limiting extracellular replication of Schu S4 with gentamicin restored the ability of uninfected bystander cells to produce TNF-α in response to E. coli LPS (Fig. (Fig.7B).7B). Interestingly, limiting extracellular replication of Schu S4 did not restore the ability of human DCs to produce IL-12 in response to E. coli LPS. Together, these data suggest that extracellular bacteria may play an active role in inhibiting TNF-α responses but that the suppression of IL-12 is not as tightly controlled by viable bacteria in the extracellular space.
IL-10 and TGF-β are two host cytokines previously associated with suppression of induction of proinflammatory responses (32, 39). Although we did not detect these cytokines in our Schu S4-infected DC cultures, it is possible that they were present in concentrations that were physiologically active but below the level of detection. Thus, to determine if IL-10 and TGF-β were contributing to poor DC responsiveness following infection with Schu S4, we added neutralizing anti-IL-10, anti-pan TGF-β antibodies, or the appropriate isotype control to Schu S4-infected DCs and assessed responsiveness to E. coli LPS. Neither addition of neutralizing anti-IL-10 nor anti-TGF-β antibodies restored the ability of Schu S4-infected human DCs to respond to E. coli LPS, as measured by production of TNF-α and IL-12 (Fig. (Fig.7).7). Thus, neither of these host components was responsible for the inability of DCs to respond to secondary stimuli.
The inability of DCs in the presence of live Schu S4 to respond to secondary stimuli may stem from molecules that originate from the host, the bacterium, or both of these elements. However, we failed to detect several of the host-derived cytokines, e.g., TGF-β, IL-10 and prostaglandin E2, that are typically associated with eliciting anti-inflammatory or suppressive responses following infection with Schu S4. Furthermore, addition of neutralizing antibody against TGF-β and IL-10 did not restore DC ability to respond to secondary stimuli (Fig. (Fig.7).7). Thus, we reasoned that Schu S4-derived products may inhibit DC responsiveness to E. coli LPS. Unlike attenuated F. tularensis LVS, Schu S4 replicates logarithmically over 48 h in cRPMI medium in the absence of host cells (C. M. Bosio and D. Crane, unpublished observations). We took advantage of this trait to assess the ability of Schu S4 components that might be shed or secreted during extracellular replication to interfere with DC function. Following overnight incubation of Schu S4 in cRPMI medium, the resulting S4CM was sterile filtered and in some cases heated for 1 h at 95°C [S4CM(95)] and then tested for its activity on human DCs. Although no IL-6 was detected in the supernatants, unlike intact bacterium, S4CM elicited very low concentrations of both TNF-α and IL-12/IL-23p40 (Fig. (Fig.8A)8A) into the culture medium. This stimulatory activity from S4CM was partially heat stable since cells incubated with S4CM(95C) secreted TNF-α and IL-12/IL-23p40 but at significantly lower concentrations than DCs incubated in S4CM (Fig. (Fig.8A).8A). However, despite this weak cytokine response, S4CM had a strong inhibitory effect on DC responsiveness to E. coli LPS. Approximately one-third fewer DCs preincubated with S4CM were able to produce TNF-α or IL-12 in response to E. coli LPS compared to untreated controls (Fig. 8B and C). Unlike the induction of primary cytokine secretion, the inhibitory effect of S4CM on DCs was not completely lost following heat treatment (Fig. 8B and C), suggesting that the Schu S4 components responsible for suppressing responses to E. coli LPS in DCs were primarily heat stable.
F. tularensis is a gram-negative bacterium and, thus, possesses LPS. Given the heat-stable nature of the suppressive factor in S4CM, we next determined if purified Schu S4 LPS may be inducing endotoxin tolerance and, thus, mediating nonresponsiveness to E. coli LPS in human DCs. DCs were incubated with purified Schu S4 LPS followed by exposure to E. coli LPS. Schu S4 LPS did not elicit production of intracellular IL-12 or TNF-α at any concentration tested (Fig. (Fig.9A).9A). Analysis of culture supernatants incubated with Schu S4 LPS overnight also failed to yield concentrations of TNF-α, IL-12, IL-6, or IL-10 above that observed in untreated controls (data not shown). Furthermore, only DCs exposed to high concentrations of Schu S4 LPS (10 μg/ml) were impaired in their ability to respond to E. coli LPS (Fig. (Fig.9B9B).
LPS was detected in S4CM (Fig. 10A). Although our data show that concentrations of purified Schu S4 LPS 1,000-fold higher than the concentration of E. coli LPS are required to induce tolerance, it was possible that F. tularensis LPS was, in part, mediating the tolerance we observed in S4CM. Thus, we next depleted S4CM of Schu S4 LPS. As described above, S4CM inhibited the ability of DCs to produce TNF-α and IL-12 in response to E. coli LPS (Fig. 10B). Similarly, depletion of Schu S4 LPS in S4CM failed to restore DC production of these two cytokines following exposure to E. coli LPS (Fig. 10B). Furthermore, we quantified the Schu S4 LPS present in S4CM and found that it was present at less than 1.5 μg/ml. This concentration is nearly 10-fold less than what was required to induce refractory responses in human DCs (Fig. (Fig.9B).9B). Therefore, it is unlikely that Schu S4 LPS alone present in S4CM was responsible for mediating nonresponsiveness of uninfected cells to unrelated LPS molecules. These data suggest that other heat-stable moieties may contribute to the poor responsiveness of Schu S4-negative cells observed in Schu S4-infected DC cultures.
DCs are a central component of the mammalian immune system and serve as both responders during innate immunity and a bridge to effective and long-lasting memory immune responses. Thus, it is unsurprising that many pathogens have developed mechanisms to modulate and disable DCs (1, 6, 33, 40, 45-48). One way in which pathogens interfere with host immune responses, including those mediated by DCs, is manipulation of normal host systems meant to limit hyperresponsiveness to microbial products, resulting in control of inflammation. For example, the temporary state of refractoriness to endotoxin and other microbial products following sepsis (referred to as “microbial tolerance”) tempers the inflammatory response and development of septic shock (as reviewed in reference 35). However, this state of tolerance also leaves the host susceptible in a hyporesponsive state and, thus, more susceptible to secondary infection (23, 25, 34, 49). It has been postulated that some pathogens may use induction of tolerance as a mechanism by which to evade the immune response to successfully infect and replicate in the host (28). The specific mechanism(s) of microbial tolerance is varied and ill defined and has been suggested to operate at multiple levels from alteration of receptor expression to impairment of cell signaling pathways.
Data presented herein suggest that virulent F. tularensis may utilize microbial tolerance as a mechanism to evade elicitation of proinflammatory cytokines in human DCs, ultimately dysregulating the immune response in favor of the bacterium. Using GFP-Schu S4, we also observed that cells negative for this bacteria responded poorly to secondary stimulation with E. coli LPS. In contrast to previous reports with more attenuated strains of F. tularensis, we demonstrate that direct infection of DCs was not required to successfully interfere with DC responsiveness to multiple TLR agonists. It is possible that these cells harbored small numbers of GFP-Schu S4 cells that were below the threshold of detection. However, due to the magnitude of nonresponsiveness in Schu S4-negative cells and the ability of sterile, S4CM to inhibit DC responsiveness, we do not believe that small (1 to 2 CFU) numbers of undetectable GFP-Schu S4 cells were the primary contributors to tolerance in Schu S4-negative cells. Rather, our data suggest that in addition to modulation of infected cells, Schu S4 secreted or shed molecules that induced a refractory state in uninfected bystander human DCs. The specific mechanism of immunosuppression may be different among bystander and directly infected cells. However, both result in poor DC responsiveness and may contribute to the overall pathogenesis of Schu S4-mediated disease. Our observations of DC suppression represent an important advance in understanding how pathogens might manipulate these critical host cells. As described, the induction of microbial tolerance in human cells following exposure to microbial products is a well-established phenomenon. However, in the case of traditional tolerance, the pathogen stimulates an initial wave of proinflammatory cytokines which may serve to alert the host to the invading pathogen. Our data with Schu S4 suggest that highly virulent pathogens may have developed mechanisms by which they can effectively suppress cell responsiveness in the complete absence of proinflammatory cytokine production, thus evading both the initial wave of host effector responses as well as tempering the ability of the host to mobilize additional defenses. Furthermore, little is understood about how human DCs are modulated by virulent pathogens. Thus, data presented herein provide an important concept concerning modulation of primary DCs by highly virulent pathogens and should be considered when analyzing the interaction of these cells with other infectious agents.
The suppression of bystander human DCs presented here correlates with our previous findings in the murine model of pneumonic tularemia. In those studies, inhalation of Schu S4 inhibited the recruitment of proinflammatory cells throughout the lung in response to LPS (7). Importantly, this occurred at a point when infection was restricted to the airways, suggesting that fulminant infection of the lung was not required to exert organ-wide tolerance to LPS. Although we did not define the mechanism of unresponsiveness in that study, it is possible that direct modulation and induction of tolerance by Francisella products contributes to the unresponsiveness. Indeed, it has been observed that humans experimentally infected with virulent F. tularensis develop in vivo tolerance to LPS from other gram-negative bacteria (24). The specific mechanism of tolerance in humans infected with Francisella was not defined at the time of that study. However, with the recognition of TLRs as key receptors for recognition of LPS and other pathogen-associated molecular products, it is possible that Francisella is directly modifying TLR-mediated activity. Inhibition of TLR signaling and the resulting cytokine production may occur at several levels including decreased expression of TLRs on the cell surface, inhibition of phosphorylation of specific signaling molecules required for induction of cytokine production, and induction of TLR-negative regulators such as Tollip, suppressor of cytokine signaling 1 (SOCS1) to SOCS3 and IRAK-M (as reviewed in reference 29). The direct effect of Francisella on each of these steps in TLR signaling is currently being examined in our laboratory. Thus, although the murine lung environment is quite different from that of in vitro cultured human DCs, it is tempting to speculate that identification of the specific mechanisms involved in the inhibition of human DCs may also be at work in the murine lung. Thus, the data presented herein may provide important new clues as to how virulent Francisella, and potentially other important human pulmonary pathogens, might be influencing both directly infected cells and uninfected bystanders.
Although a number of microbial products are capable of inducing modulating immune responses, the most well-studied molecule is LPS (as reviewed in references16 and 20). F. tularensis is a gram-negative pathogen and, thus, possesses LPS. While high concentrations of purified Schu S4 LPS could suppress human DC responsiveness, further depletion of LPS from S4CM failed to restore responsiveness of human DCs to unrelated TLR agonists (E. coli LPS) (Fig. (Fig.66 and and7).7). The poor stimulatory capacity of Schu S4 LPS is in agreement with previous reports examining immunogenicity and the tolerogenic capacity of LPS derived from the more attenuated F. tularensis strain LVS (2, 17). However, it is well established that optimal induction of signaling via E. coli LPS in mammalian cells requires both CD14 and LPS binding protein (LBP). Although E. coli LPS is capable of stimulating primary human DCs in the absence of these two proteins (Fig. (Fig.33 to to5,5, ,8,8, ,9,9, and and11),11), it is possible that CD14 and LBP are required for cellular activation or suppression mediated by Schu S4 LPS. Both CD14 (cell membrane bound and soluble) and LBP are present in abundance in vivo. Thus, while Francisella LPS does not play a predominant role in the induction of tolerance in host macrophages and DCs in vitro, it may exert a more profound effect in the in vivo setting. Furthermore, there may be other heat-stable Schu S4 components capable of mediating tolerance to multiple microbial TLR agonists in vitro and in vivo that were not identified in this study. Identification of these components is currently being pursued in our laboratory. This identification will enable us to develop novel therapeutics and vaccines against tularemia. Additionally, given the strong suppressive activity mediated by these molecules, they may also represent a useful class of anti-inflammatory agents capable of dampening destructive inflammatory responses present during other diseases.
In addition to microbial tolerance, there are other potential bacterial components and mechanisms that may contribute to the suppression of human DC activity. F. tularensis encodes a pathogenicity island (FPI) that is critical for survival and replication of the bacterium (36, 37, 43). Specifically, several of these genes have been shown to be important for F. tularensis escape from the phagosome, enabling replication of the bacterium in host cell cytoplasm (43). Interestingly, one gene and the resulting protein have been implicated in both phagosomal escape (resulting in replication of the bacterium) and induction of tolerance in murine and human macrophages. LVS bacteria lacking the iglC gene failed to inhibit responsiveness to E. coli LPS following infection of J774.1 macrophages (50). Furthermore, a spontaneous LVS mutant that failed to produce IglC induced both proinflammatory cytokines and was unable to suppress responsiveness of human macrophages to E. coli LPS (11). We detected IglC protein in S4CM. However, depletion of iglC from S4CM failed to restore responsiveness of human DCs to other TLR agonists (data not shown). This suggests that IglC alone is not responsible modulation of human DC.
The FPI also encodes a type VI secretion system. In other bacteria, type VI secretion systems are responsible for secretion of molecules that are capable of disabling the host immune response. For example, proteins secreted via the type VI secretion system described in Vibrio cholera are responsible for cytotoxicity in mammalian macrophages (38). Porphyromonas gingivalis utilizes type VI secretion to expel gingipain proteases. Interestingly, an important function of these proteases is the induction of LPS tolerance in macrophages (18, 41). Type VI secretion systems have also been identified in Yersinia and Burkholderia species (44, 52). While their presence is associated with virulence, the specific effectors and their mechanisms of action on a cellular level have not been defined. A similar phenomenon may be at work in Francisella. That is, in Francisella species, it is possible that while IglC may have a role in phagosomal escape, it is not the effector molecule mediating hyporesponsiveness. Rather, IglC may act as part of a type VI secretion apparatus that aides in the delivery of a yet to be defined effector molecule ultimately responsible for inhibiting host cell responsiveness to inflammatory stimuli. Additional studies examining the specific role of IglC and other proteins encoded in the FPI in the induction of Francisella-mediated tolerance are currently under way in our laboratory.
As a facultative intracellular pathogen, F. tularensis must contend with the host immune system from both inside and outside individual host cells. It is likely that the mechanisms by which F. tularensis accomplishes this are complex and involve a number of strategies. Here, we show for the first time that one way in which virulent F. tularensis interferes with host immunity is by suppressing responsiveness of both directly infected cells and uninfected bystander cells. Interestingly, in contrast to the production of an array of proinflammatory cytokines observed in cells exposed to other microbially derived TLR agonists such as LPS or peptidoglycan prior to the onset of unresponsiveness, the Schu S4 LPS and other components present in the culture medium were only weakly inflammatory. Thus, the profound suppression mediated by F. tularensis in DCs was not obtained at the cost of an initial burst of cytokines that might alert the immune system. Importantly, the broad suppression of DC responsiveness (in terms of cytokine secretion) is similar to that observed in the lungs of mice following aerosol infection with Schu S4 (7). Together, these data suggested that Schu S4 suppresses the host immune response in a global sense; that is, interference with host responsiveness is not restricted to infected cells but extends to uninfected cells that may aid in the resolution of infection. These data also have broad implications for how both human DCs and the murine pulmonary environment are manipulated by virulent pathogens and represent an important consensus of global dysregulation by a pathogen in murine and human cells.
We thank Paul Brett for guidance and assistance in preparation of the Schu S4 LPS. We also thank John Belisle and Brian Kelsall for their helpful discussion and criticism pertaining to this work. Also, we thank Steven Holland, Eugene Howerton, and the Department of Transfusion Medicine at NIH for their assistance in providing human monocytes.
This work was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Allergy and Infectious Diseases.
Editor: S. R. Blanke
Published ahead of print on 3 November 2008.