Serving the new masters – dendritic cells as hosts for stealth intracellular bacteria

Authors


For correspondence. E-mail fabrik@pmfhk.cz; Tel. +420 973 255 199; Fax +420 495 513 018.

Summary

Dendritic cells (DCs) serve as the primers of adaptive immunity, which is indispensable for the control of the majority of infections. Interestingly, some pathogenic intracellular bacteria can subvert DC function and gain the advantage of an ineffective host immune reaction. This scenario appears to be the case particularly with so-called stealth pathogens, which are the causative agents of several under-diagnosed chronic diseases. However, there is no consensus how less explored stealth bacteria like Coxiella, Brucella and Francisella cross-talk with DCs. Therefore, the aim of this review was to explore the issue and to summarize the current knowledge regarding the interaction of above mentioned pathogens with DCs as crucial hosts from an infection strategy view. Evidence indicates that infected DCs are not sufficiently activated, do not undergo maturation and do not produce expected proinflammatory cytokines. In some cases, the infected DCs even display immunosuppressive behaviour that may be directly linked to the induction of tolerogenicity favouring pathogen survival and persistence.

Dendritic cells meet intracellular bacteria

Professional phagocytes, including neutrophils, macrophages and dendritic cells (DCs), constantly inspect their surroundings and actively eliminate intruders by phagocytosis; hence forming the backbone of the innate immune response. However, the role of DCs is broader, as they are the most potent antigen-presenting cells (APCs) capable of naïve T-cell priming (Banchereau and Steinman, 1998; Banchereau et al., 2000; Joffre et al., 2009). An encounter by an immature DC with an antigen leads to DC activation and is connected with phenotypic change, decrease in phagocytosis and lymph-node migration, which are all part of the process of DC maturation (Banchereau and Steinman, 1998; Banchereau et al., 2000). The undisturbed realization of T-cell priming ensures the existence of several DC subtypes specializing on innate response, antigen transport, cytokine/chemokine secretion, presentation via MHC II or cross-presentation via MHC I (reviewed by Shortman and Naik, 2007; Villadangos and Schnorrer, 2007; Wu and Liu, 2007; Lopez-Bravo and Ardavin, 2008; Helft et al., 2010; Liu and Nussenzweig, 2010; del Rio et al., 2010; Schmid et al., 2010; Watowich and Liu, 2010). Therefore, the invasion of DCs is crucial infection strategy of intracellular bacteria targeting phagocytes as these hosts may trigger an effective Th1 immunity in response (Kapsenberg, 2003). Inevitably, successful pathogens have found specific ways to subvert DC function and even to turn their hosts into their servants. This is very well documented in the case of Mycobacterium tuberculosis (reviewed by Herrmann and Lagrange, 2005; Sinha et al., 2007; Jozefowski et al., 2008; Baena and Porcelli, 2009; Krishnan et al., 2010; Natarajan et al., 2011; Khan et al., 2012; Prendergast and Kirman, 2013). Mycobacterium, notoriously known for its persistence, supplies DC pathogen recognition receptors (PRRs) with cell-wall-derived ligands, which, however, activate signalling leading away from Th1. Notably, mannosylated lipoarabinomannan (ManLAM) engages C-type lectin dendritic cell-specific intercellular adhesion molecule-3-grabbing non-integrin (DC-SIGN), which in response induces expression of immunosuppressive IL-10 (Švajger et al., 2010). As a consequence, DC maturation and migration is suppressed, impairing the process of effective antigen presentation. Onset of the adaptive immune response toward tuberculosis is indeed delayed and insufficient for the eradication of bacteria (Urdahl et al., 2011). Alternatively, different intracellular pathogen Salmonella enterica, exploits DC functions mainly for transport (reviewed by Wick, 2002; 2007; 2011; Sundquist et al., 2004; Biedzka-Sarek and Skurnik, 2006; Tam et al., 2008; Bueno et al., 2010; 2012; Watson and Holden, 2010). Intestinal lumen-sampling CX3CR1+ DC subset provides trans-epithelial entry route for less invasive strains; however, the main target of Salmonella subversion are DCs residing under the epithelium. Although DCs with internalized Salmonella appear to be activated and migrate into draining mesenteric lymph nodes, presentation of pathogen antigens is not prominent. Infected DCs are moreover sensitive to Salmonella-induced cell death, by which the transported cargo may be released into the new environment. Collectively, it seems that Salmonella uses DC migration for spreading, at least in the initial phase of systemic infection, while avoids priming of effector T-cells.

In contrast to Mycobacterium and Salmonella little is known about the interaction of DCs with intracellular bacteria that do not typically receive clinical attention. This review focuses particularly on zoonotic Coxiella, Brucella and Francisella, which are often under-diagnosed, partially due to their ability to induce only idle immune response, yet can cause life-threatening complications (Cooper et al., 1999; Brouqui and Raoult, 2001; Marmion et al., 2005; Baud et al., 2009; Buzgan et al., 2010; Kaya et al., 2011; Wegdam-Blans et al., 2012). We feel there may be a connection between the outcome of the early DC interactions and the mild nature of the infection initial phase. By summarising the current knowledge, we intended to explore the interplay between DCs and Coxiella, Brucella and Francisella in the hope that this will help us to understand the way these pathogens fool the immune system.

Stealth pathogens invade DCs

Coxiella

Coxiella burnetii is Gram-negative intracellular pathogen, which in humans causes a mild and often asymptomatic flu-like illness called Q-fever (Maurin and Raoult, 1999; Madariaga et al., 2003). The disease may however develop into a chronic form, particularly in persons with pre-existing valvulopathy where it manifests as endocarditis (Parker et al., 2006).

Although recently introduced axenic medium supporting Coxiella growth (Omsland et al., 2009) opened new possibilities for species genetic engineering, the research progress of DC–Coxiella interaction lags behind those of other intracellular pathogens. Moreover, high infectivity combined with phase variation-dependent virulence restricts manipulation of Coxiella to BSL-3 laboratories. So-called Phase I strains are highly virulent and are obtained from infected animals or humans. After several passages of Phase I strains in cell culture or embryonated eggs, an attenuated Phase II phenotype with truncated lipopolysaccharide (LPS) can be obtained (Maurin and Raoult, 1999). Despite the different virulence, both phenotypes are able to proliferate inside human monocyte-derived DCs (HmDCs), occupying large acidic phagolysosome-like LAMP-1+ (Sauer et al., 2005), LAMP-3+ (Shannon et al., 2005b) vacuoles. However, the difference lies in the ability to induce DC maturation and cytokine production. Significant upregulation of CD80, CD86, CD83, CD40 and HLA-DR has been observed only by HmDCs infected with Phase II strains (Shannon et al., 2005b), and similarly, only those cells have been found to secrete IL-12p70 and TNF-α (Shannon et al., 2005a,b). Toll-like receptor (TLR) engagement is proposed, deduced from p38 MAPK phosphorylation (Shannon et al., 2005a,b), probably of TLR-2 origin (Zamboni et al., 2004). TLR-4 seems not to be involved (Shannon et al., 2005b) and in fact, lipid A moieties of both Phase I and Phase II LPS inhibit TLR-4 signalling (Zamboni et al., 2004). Nevertheless, the fact that three Phase I strains with chemically distinct LPS forms were unable to elicit IL-12p70 production from HmDC (Shannon et al., 2005a) suggests lipid A-independent functional differences between the Phase I and Phase II LPS, which are crucial for shielding of Coxiella pathogen-associated molecular patterns (PAMPs) (Shannon et al., 2005b; Wei et al., 2011). However, passive elusion is not the only hallmark of the pathogen lifestyle. Coxiella also expresses proteins with the potential to elicit IL-10 production in DCs (Wang et al., 2011; Wei et al., 2011), reminding the immunosuppressive infection strategy of Mycobacterium (see above).

Nevertheless, Phase I Coxiella-mediated proinflammatory DC activation has been shown to be possible. In one study, the co-cultivation of HmDCs with antibody-opsonized Phase I Coxiella led to the apparent maturation of the host cells, together with increased secretion of IL-1β, IL-6, IL-12p70 and TNF-α (though bacterial replication appeared to be preserved) (Shannon et al., 2009). Experiments with murine bone marrow-derived DCs (BMDCs) have confirmed that the DC activation is FcR-dependent. The similar role of antibody opsonization was observed in the case of S. enterica (Herrada et al., 2007). BMDCs infected by non-opsonized Salmonella were unable to induce T-cell activation; however this was restored by the internalization of IgG-coated Salmonella. The opsonization enhanced Salmonella lysosomal degradation, and probably also the antigen processing, in FcγRIII-dependent manner. The situation in the case of Coxiella is however more complex, as there are most likely more Ab receptors involved in the control of the infection and also the role of complement must be taken into account (Shannon et al., 2009). These results suggest a supportive role for antibodies during coxiellosis that justifies the pursuit for Coxiella antigenic structures. Therefore, as potential primers of adaptive response, DCs were exposed to several Coxiella proteins to assess their potential as a subunit vaccine (Wang et al., 2011; Wei et al., 2011; Xiong et al., 2012). Coxiella-outer membrane protein 1 (Com1) has been shown to be particularly effective in the stimulation of DCs to mature and secrete proinflammatory cytokines. Moreover, Com1-pulsed DCs also activated CD4+ (Wei et al., 2011) or both CD4+ and CD8+ T-cells (Wang et al., 2011; Xiong et al., 2012), which in turn increased the production of IFN-γ (Wang et al., 2011; Wei et al., 2011; Xiong et al., 2012) or IL-17 (Wei et al., 2011). Importantly, mice receiving Com1-pulsed BMDCs had lower bacterial loads after subsequent Coxiella infection (Wang et al., 2011; Wei et al., 2011; Xiong et al., 2012). Therefore, in contrast to the idle response of DCs toward virulent Coxiella, it appears that DCs may become crucial in the development of an efficient Th1-stimulating anti-Coxiella subunit vaccine (Xiong et al., 2012).

Brucella

Brucellosis is a zoonotic disease that causes substantial losses of livestock worldwide and presents a potential risk also to human health (Boschiroli et al., 2001). From the economic and public health points of view, Brucella melitensis, Brucella abortus and Brucella suis are the most dangerous members of the Brucellaceae responsible for the observed zoonosis (Martirosyan et al., 2011).

Brucella is able to invade bovine monocyte-derived DCs (Heller et al., 2012) and to proliferate inside murine BMDCs (Salcedo et al., 2008) and/or HmDCs (Billard et al., 2005; 2008; Zwerdling et al., 2008). The high infectivity of the bacteria is a consequence of its efficient intracellular parasitism strategy (Celli, 2006; Martirosyan et al., 2011). Compared with human monocyte-derived macrophages, B. suis invades HmDCs more efficiently (Billard et al., 2005). It appears this is partially due to the induced presence of veil-like structures on the surface of DC membrane that engulf the bacteria in a lipid raft-dependent way. Early after phagocytosis, about 80% of Brucella-containing vacuoles (BCV) are LAMP-1+ in BMDC model (Salcedo et al., 2008). During 24 h however LAMP-1 colocalization diminishes and BCV becomes replicative niche, acquiring endoplasmic reticulum-like (KDEL+, calnexin+) phenotype (Salcedo et al., 2008; Conde-Alvarez et al., 2012). The sensitivity of DCs to Brucella infection has biological relevance in vivo. Approximately 50% of intracellular B. abortus were internalized by CD11c+ cells in a mice intestinal infection model (Salcedo et al., 2008), and similarly, the majority of Brucella-infected cells in mice lungs were CD11c+ MHC II+ (Archambaud et al., 2010). However, despite high susceptibility, Brucella-infected DCs are not amenable to phenotypic maturation (Billard et al., 2007a,b; 2008; Salcedo et al., 2008; Archambaud et al., 2010; Surendran et al., 2011; Heller et al., 2012) nor to the production of the cytokines TNF-α and IL-12 in vitro (Billard et al., 2007a,b; Salcedo et al., 2008; Surendran et al., 2010), even in cases where the DCs appeared to be matured (Zwerdling et al., 2008). In fact, cell–cell interactions of γδ T-cells with B. abortus-infected DCs are needed to restore inflammatory DC behaviour (Ni et al., 2012). Direct contact of non-activated, but infected, HmDCs with Vγ9Vδ2 T-cells induces secretion of proinflammatory TNF-α and IFN-γ from T-cells, which in turn, at least partially help to engage CD80 and CD86 upregulation and IL-12 production in infected HmDCs. The similar cell–cell cross-talk was needed for activation of Mycobacterium-infected HmDCs (Meraviglia et al., 2010), albeit the interaction did not trigger Vγ9Vδ2 T-cell proinflammatory functions. The interrelationship of DCs and γδ T-cells illustrates complex interaction network between the cells of immune system but also demonstrates the elusiveness of Brucella for DCs. In line with the unresponsiveness, the migrational nature of DCs contributes to the dissemination of B. abortus to the spleen and liver in pulmonary infection models (Archambaud et al., 2010). Although this may boost immune response and promote the generation of iNOS+ and TNF-α-producing inflammatory CD11b+ DCs (Copin et al., 2007; 2012; Archambaud et al., 2010), results suggest that DCs facing Brucella alone (Ni et al., 2012) only poorly stimulate T-cells (Billard et al., 2007a,b; 2008). Virulent B. abortus has been shown to induce caspase-2-dependent apoptotic/necrotic cell death in BMDCs; however, this had only little or no effect on activation of bystander cells (Li and He, 2012). These results imply a balancing between pathogen proliferation and transport on one hand and host cell death-induced suppression of antigen presentation and facilitated pathogen dissemination on the other hand (Li and He, 2012), similarly to what is proposed for Salmonella (see above). Taken together, these studies indicate that the primary reason for DC functioning failure lies in the ability of Brucella to avoid the activation of host cells. Several Brucella virulence factors are likely involved in maintaining this low-profile infection. Mutant strains with deleted genes for Omp25 or for two-component regulatory system BvrR/S (Billard et al., 2007a; 2008) have been shown to induce HmDC maturation. Genomes of B. abortus and B. melitensis also contain a gene for a Toll/Interleukin-1 receptor homologue (Brucella-TIR-Protein1; Btp1) that actively interferes with TLR-2 signalling, DALIS formation and TNF-α and IL-12p40 production (Salcedo et al., 2008). Interestingly, infection by B. suis lacking the gene virB, which encodes a subunit of a type IV secretion system, does not lead to a significant upregulation of HmDC maturation markers, cytokine production or enhanced T-cell stimulation (Billard et al., 2008). However, these mutants are not able to proliferate inside the host cell most likely due to a failure to avoid lysosomal degradation (Salcedo et al., 2008). The fact that intracellular killing of Brucella does not provide enough stimuli for DC activation illustrates the pathogen masking ability that is embodied by the LPS of the bacteria.

Brucella LPS engages a classic TLR-4-dependent activation with minor or no role of TLR-2 (Campos et al., 2004; Barquero-Calvo et al., 2009). However, this event has very limited potential to induce proinflammatory responses in DCs (Barquero-Calvo et al., 2009; Surendran et al., 2012). This is caused largely by the absence, modification or inaccessibility of PAMP structures on the Brucella surface, which has been achieved over a long co-evolution with eukaryotic hosts (Barquero-Calvo et al., 2009). A B. abortus vaccine strain RB51 lacking the O-antigen (so-called rough phenotype) has been shown to stimulate the surface expression of BMDC maturation markers in vitro (Surendran et al., 2010; 2011; 2012). Identical results were obtained for engineered attenuated mutants with a deletion of gene for phosphomannomutase ManB (Billard et al., 2005; 2007b; 2008), which is involved in the biosynthesis of LPS. However, the nature of attenuation is more complex as B. abortus rough LPS alone does not differ significantly from LPS of wild-type strains in terms of induction of HmDC maturation (Billard et al., 2007b). Moreover, wild-type B. abortus LPS has been shown to display stronger adjuvant effects for the stimulation of DC–T-cell interactions when compared with rough LPS (Campos et al., 2004). The explanation for these contrasting findings most likely lies in the more heterogeneous acylation pattern of wild-type lipid A (Campos et al., 2004). The authors of a different study did not focus on Brucella O-antigen but rather on the oligosaccharide core (Conde-Alvarez et al., 2012). B. abortus with deleted gene for putative core mannosyltransferase WadC retained its smooth phenotype, but the mutant was unable to proliferate inside BMDCs and colocalized with a LAMP-1+ compartment. Moreover, infection by the mutant resulted in increased DALIS formation and IL-12p70 and TNF-α production signalizing BMDC activation (Conde-Alvarez et al., 2012). Therefore, whether it is O-antigen or core oligosaccharide, it appears that every portion of the Brucella LPS has a role in avoiding unnecessary DC attention.

Even though the structures of live Brucella are only poorly immunogenic, heat-killed Brucella (HKB) strains are able to promote DC maturation (Macedo et al., 2008; Salcedo et al., 2008; Zwerdling et al., 2008) and a proinflammatory cytokine production (Huang et al., 2003; 2005; Macedo et al., 2008; Salcedo et al., 2008; Zwerdling et al., 2008; Oliveira et al., 2011) in BMDCs or splenic DCs. Moreover, HKB induces Th1-shaping DC functions (Huang et al., 2001) and DC-dependent T-cell proliferation (Zwerdling et al., 2008). A possible explanation for the change compared with live bacteria would be exposure of Brucella PAMP structures that may trigger signalling of PRR. However, the identification of the particular PRRs involved could be problematic, even though the DC response to HKB is IRAK4 (Oliveira et al., 2011) and MyD88-dependent (Macedo et al., 2008). HKB-induced production of IL-12p40 appears to be TLR-2-dependent in HmDCs (Zwerdling et al., 2008) and in mice BMDCs in vitro (Macedo et al., 2008). However, DCs isolated from TLR-2−/− mice spleen are able to both mature (Macedo et al., 2008) and to secrete IL-12p40 at the same levels as DCs isolated from wild-type mice (Huang et al., 2003). Notably, TLR-9 sensing also contributes to IL-12p40 production in response to HKB in both BMDCs (Macedo et al., 2008) and splenic DCs (Huang et al., 2005). The involvement of TLR-9 is expected and suggests the availability of bacterial DNA for innate immune sensors, perhaps due to the penetration of Brucella envelopes (Huang et al., 2005).

In the search for more detailed identification of immunogenic Brucella components and a potential subunit vaccine, several proteins including lumazine synthase (Berguer et al., 2006; 2012), lipidated (Zwerdling et al., 2008) and unlipidated (Pasquevich et al., 2011) Omp19 and unlipidated Omp16 (Pasquevich et al., 2010) have been found to induce maturation and proinflammatory cytokine production in DCs. In addition, the outer membrane vesicles from a rough strain of B. melitensis were able to rapidly stimulate BMDCs to produce Th1-shaping cytokines (Avila-Calderon et al., 2012). Nevertheless, the immunogenic structures of Brucella are generally sufficiently masked to prevent the maturation and activation of DCs, thus granting the bacteria a mild host environment and sufficient time for intracellular proliferation.

Francisella

Francisella tularensis is a highly infectious Gram-negative coccobacillus responsible for tularaemia, a plague-like disease affecting mostly rodents and lagomorphs (Tarnvik and Berglund, 2003). Despite the fact that transmission to human occurs mostly accidentally, the possibility of intentional abuse is of major public concern as simple inhalation of a small number of bacteria is sufficient to cause severe disease in humans (Dennis et al., 2001; Oyston et al., 2004). Two Francisella subspecies are responsible for most cases of illness in humans: Francisella tularensis subsp. tularensis and the less virulent Francisella tularensis subsp. holarctica. The latter species was used as a background for the development of a live vaccine strain (LVS) that was shown to be attenuated in humans but retained virulence in mice and thus it is often used in model experiments. However, the reasons for the high infectivity of Francisella remain unclear, partially due to a lack of knowledge about the functions of several known virulence factors (Sjostedt, 2006).

Consistently with preserved LVS virulence in mice, the bacterium is able to enter and to multiply inside BMDCs in vitro (Bar-Haim et al., 2008). Trafficking of Francisella in BMDCs seems to be similar to situation in macrophages and by 4 h p.i. most of the bacteria escapes from phagosome to cytoplasm, where it replicates (Belhocine and Monack, 2012). Compared with macrophages however, BMDCs show slightly higher bacterial uptake efficiency and permissivity to bacterial proliferation (Bosio and Dow, 2005). Although LVS is attenuated in humans, it replicates inside HmDCs in vitro as well (Ben Nasr et al., 2006; Bauler et al., 2011; Schmitt et al., 2012). These findings confirm the not-straightforward nature of the LVS–DC interaction, which leads to obtaining contradictory results even in controlled in vitro studies. There is a relative agreement in DC capability to mature when confronted with LVS. DCs differentiated from both human and mice precursors undergo phenotypic maturation, which results in the upregulation of CD40, CD80, CD83, CD86, CCR7 or MHC II (Bosio and Dow, 2005; Ben Nasr et al., 2006; Katz et al., 2006; Bar-Haim et al., 2008) on the cell surface. However, discrepancies arise in regards to observed DC cytokine production. In response to LVS, BMDCs produced proinflammatory TNF-α (Katz et al., 2006) and the same cells also secreted IL-12p70 (Hong et al., 2007; Wickstrum et al., 2009). It is worth noting that the production of the heterodimeric form of IL-12 was hallmark only of BMDCs and not of bone marrow-derived macrophages. In different study however, LVS actually prevented TNF-α production in BMDCs co-cultivated with Escherichia coli LPS or zymosan and even induced secretion of anti-inflammatory TGF-β (Bosio and Dow, 2005). Similarly to murine counterparts, contrasting results are characteristic also for HmDCs which either produced (Bauler et al., 2011) or did not increase production (Schmitt et al., 2012) of TNF-α, IL-6 or IL-12p40 in response to LVS. This inconsistency would seem discouraging, but these data variations can at least be partially explained by the different settings of the in vitro infection experiments. First of all, bacterial behaviour varies with the Francisella growth medium used (Hazlett et al., 2008; Periasamy et al., 2011). LVS grown on brain heart infusion broth (BHI) or on bone marrow-derived macrophages does not induce proinflammatory cytokines from BMDCs, in contrast to Mueller–Hinton broth-grown bacteria (Periasamy et al., 2011). This difference resembles a phase variation-like phenomena that is likely induced by the different availability of nutrients in the growth media (Hazlett et al., 2008). Second, bacterial internalization strongly depends on the presence of complement in the medium. Phagocytosis of C3-opsonized Francisella is significantly more efficient and leads to different cytokine profiles, as was shown in the case of LVS-infected HmDCs (Ben Nasr et al., 2006). In this study, the levels of TNF-α and IL-12p70 produced by HmDCs responding to LVS did not exceed the detectable amounts in the absence of C3 in serum, but the cytokines concentrations were greatly enhanced by bacterial opsonization. Similar effects have also been observed for IL-1β, IL-6, IL-8 and IL-10 (Ben Nasr et al., 2006). Another crucial parameter of the experiment is the multiplicity of infection (MOI). Increased MOI leads to higher absolute numbers of internalized LVS (Ben Nasr et al., 2006), and it appears there is a threshold MOI under which the bacteria do not elicit the maturation of DCs and cytokine production (Katz et al., 2006). Lastly, there are several protocols for differentiation of DCs from their precursors, and deviations in cultivation procedures could result in the harvesting of functionally different cells. In one study, monocytes produced TNF-α, IL-1β, MIP-1α and MIP-1β upon infection with a virulent SchuS4 Francisella strain, but DCs differentiated from the same monocytes did not (Chase and Bosio, 2010). Therefore, the discrepancies observed in studies focused on LVS–DCs interactions could be the result of different experimental setups and care must be taken when designing the experimental workflow. Although these are general rules, the results obtained from in vitro DC infections by virulent SchuS4 Francisella strain appear to be more consistent when compared with LVS. HmDCs mature in terms of CD83, CD86, HLA-ABC and HLA-DR upregulation after contact with virulent Francisella, however, the expected proinflammatory cytokine response is not prominent (Chase et al., 2009; Bauler et al., 2011). A similar situation has been observed in murine BMDCs (Wickstrum et al., 2009) together with the promoted production of IL-10 and TGF-β (Periasamy et al., 2011). The suppression of cytokine production could be mediated by a thermostable substance released by Francisella into the environment (Chase et al., 2009); however, there may be another explanation. HmDCs infected by virulent Francisella produce high amounts of IFN-β, which, in these cells, reduces the secretion of IL-12p40 (Bauler et al., 2011). As IFN-β signalling helps to initiate inflammasome formation, it is possible that infected HmDCs alter their proinflammatory behaviour to aid pyroptosis, though the latter might never occur. Indeed, in contrast to avirulent strains (Belhocine and Monack, 2012), virulent Francisella is less potent (Wickstrum et al., 2009; Bauler et al., 2011) in triggering pyroptosis in DCs and thus, it can exploit the anti-inflammatory functions of IFN-β. Results of in vivo experiments are in agreement in noting that TNF-α and IL-12p40 production in the lungs is low after intranasal virulent Francisella administration (Bosio et al., 2007). Moreover, a relative decrease in the numbers of MHC II+ DCs in the mediastinal lymph nodes (MdLN) was observed together with induced maturation insensitivity (Bosio et al., 2007). The impairment of DC functions could redefine the roles that these cells fulfil in the context of the immune system when facing virulent or attenuated Francisella. In experiments with LVS introduced intranasally, approximately 95% of bacteria were internalized by CD11c+ cells during the first hour after infection (Bosio and Dow, 2005). After 24 h p.i., a very rapid increase in the numbers of respiratory tract DCs migrating from the lungs was observed in the MdLN (Bar-Haim et al., 2008). Importantly, approximately 38% of the arriving CD11b+ cells contained LVS, which indicates the role of APCs in the transport of bacteria. The depletion of APCs or inhibition of their migration capabilities prior to infection by LVS increases mice survival and reduces bacterial loads in organs (Bosio and Dow, 2005; Bar-Haim et al., 2008). Conversely, dissemination to lymphatic organs may result in priming of adaptive immunity in which LVS and fully virulent Francisella could differ. LVS induces maturation of DCs in the lungs during intranasal infection (Bosio and Dow, 2005). In vitro experiments with LVS-activated BMDCs revealed that CCL19-dependent chemotaxis relies on the secretion of the p40 subunit of IL-12 (Slight et al., 2011), which suggests that, in the case of LVS, DC migration and proinflammatory behaviour may be linked. It is therefore possible that LVS-infected DCs are able to provide enough signals for Th1 priming of naïve T-cells. However, DCs could have another function during LVS challenge. Although neutrophils are the main target for Francisella independently of the strain virulence by the 3rd day p.i., approximately 10% of LVS infected cells are still represented by DCs (Hall et al., 2008). Interestingly, over the same timeframe p.i., classic DCs, together with unconventional subtype natural killer dendritic cells, appear to play a very crucial role in the direct production of IFN-γ following intradermal LVS administration (De Pascalis et al., 2008). The role of DCs as IFN-γ producers in an LVS infection model was also highlighted in a study confirming the importance of the Th17 immunologic response in lungs (Lin et al., 2009). Atypically for an intracellular pathogen, early Th17 engagement helped to control the LVS infection, even though only IL-17A appeared to be involved. The co-cultivation of IL-17A and BMDCs pulsed with LVS increased the production of IL-12 compared with LVS-infected BMDCs alone. IL-17A was also able to elicit IL-12 and IFN-γ production in naïve BMDCs, which was also confirmed with lung CD11c+ cells ex vivo. The authors demonstrated that the position of Th17 needs to be reconsidered in the immunologic response to LVS, possibly as a Th1 enhancer, and DCs may be pre-T-cell IFN-γ producers (Lin et al., 2009). Virulent Francisella however forces DCs to secrete TGF-β and thus promotes their tolerogenic phenotype (Periasamy et al., 2011). Provided that priming by TGF-β-producing DCs leads to differentiation of naïve T-cells into Tregs, Francisella gains the advantage of an immunosuppressive environment favouring its survival. The correlation of TGF-β levels with the numbers of Fox3P+ cells in mice infected by virulent-like BHI-grown LVS (see above) indicated clear dependence (Periasamy et al., 2011). All these results suggest that although LVS maintains its virulence in mice, the engagement of DCs could lead to a more apparent and effective immunologic response compared with fully virulent strains where DCs could potentially play a role in the induction of tolerogenicity. In humans, LVS is attenuated and has been considered for vaccination purposes. In accordance with that, transcriptome analysis of peripheral blood mononuclear cells of human volunteers infected by LVS revealed the upregulation of genes considered as cell-specific activation markers for DCs and connected with antigen presentation (Fuller et al., 2007).

The same study also demonstrated an early increase in TLR mRNA levels, thus confirming the role of these PRRs for Francisella sensing and the initiation of a proper immune response (Fuller et al., 2007). Although DnaK protein promotes DC activation and maturation in a TLR-4-TRIF/MyD88-dependent way (Ashtekar et al., 2008), the TLR-4 receptor is generally not involved in the proinflammatory response of DCs to Francisella (Katz et al., 2006). These findings are in a good agreement with the poor response of DCs toward the LPS of Francisella (Ben Nasr et al., 2006; Chase et al., 2009). Instead, TLR-2 appears to be a DC prominent TLR involved in Francisella recognition. LVS-induced TNF-α secretion and NF-κB activation is mediated in a TLR-2/TLR-6 heterodimer-dependent manner (Katz et al., 2006), and similarly, production of the IL-12p40 (Slight et al., 2011) or IL-12p70 heterodimers relies on TLR-2 (Hong et al., 2007). It seems therefore surprising that TLR-2−/− mice showed higher amounts of IFN-γ in their lungs during pulmonary LVS infection (Malik et al., 2006). Furthermore, the production of immunosuppressive TGF-β in BMDCs infected by virulent SchuS4 (or virulent-like BHI-grown LVS) was TLR-2-dependent (Periasamy et al., 2011). Understanding of TLR-2-signalling events is clearly needed for the explanation of pro-/anti-inflammatory dichotomy of TLR-2 in Francisella infection. It is known that interplay between different host receptors may have unexpected consequences, as shown by DC-SIGN/TLR collaboration in IL-10 production in response to mycobacterial ManLAM (Gringhuis et al., 2009). Recently, interactions of human monocyte-derived macrophages with C3-opsonized SchuS4 revealed interesting cross-talk between CR3 complement receptor and TLR-2 (Dai et al., 2013). Although TLR-2 signalling activates p38 and ERKs, it also partially stimulates CR3-medaited phagocytosis of C3-opsonized SchuS4. CR3 signalling in turn however dampens MAPK/NF-κB activation through Lyn-PI3K/Akt-MKP-1-dependent process, resulting in suppression of TNF-α, IL-6 and IL-1β production. Note, that PI3K/Akt is downstream of both TLR-2 and CR3 and therefore slight deviations from initial conditions may possibly lead to different findings (Leander et al., 2012), showing the importance of correct interpretation of experimental data. Moreover, described receptor cross-talk is specific for virulent SchuS4 strain and probably limited to human phagocytes (Dai et al., 2013). Although TLR-2-proinflammatory signalling was shown to be suppressed by PIK3K/Akt also in murine macrophages infected by LVS, the production of proinflammatory cytokines was still prominent (Medina et al., 2010).

In summary, the outcome of the DC–Francisella relationship reflects the severity of illness. Highly virulent Francisella strains apparently do not provide sufficient stimuli for proinflammatory DC behaviour. With decreasing virulence of the strain, DC awareness towards the pathogen increases, as in the case of LVS. However, it must be stressed that ambiguous findings regarding LVS–DC interactions currently do not allow for clear conclusions. The complexity of the situation is reflected in the restriction of LVS use for human vaccination purposes due to the enigmatic attenuation background.

Conclusion

The above-mentioned studies have demonstrated that virulent strains of Coxiella, Brucella and Francisella are able to bypass DCs in the immune response (Fig. 1). Essentially, the pathogens shield their PAMPs from DC detection by employing LPS with weak proinflammatory characteristics and thus avoid unwanted attention. In addition to passive hiding, they also actively interfere with DC functions. Analogously to systemic Salmonella infection, Brucella and Francisella were shown to exploit DCs as disseminating cells. Virulent Coxiella and Francisella elicit production of immunosuppressive cytokines in DCs, which, as in the case of Mycobacterium, may contribute to bacterial persistence. Despite the progress however, many important pieces of DC interaction puzzle still missing: How does the signalling cross-talk of bacteria-engaged PRRs affect DC activation? How does the intracellular trafficking of these bacteria in DCs differ from the situation in macrophages? What is the role of γδ T-cell–DC interaction during infection in vivo? Is there any connection with early Th17 and what is the position of Th17 in the immune response against these bacteria? Disregard these questions, Coxiella, Brucella and Francisella subversion of DCs functions is an important aspect of the infection strategy, showing the interesting directions for the future host–pathogen research.

Figure 1.

Different response of dendritic cells (DCs) toward infection by either virulent (red-coloured names) or attenuated (green-coloured names) strains of Coxiella (C.b.), Brucella (B.sp.) and Francisella (F.t.). The most prominent PRRs involved in pathogen-induced DC activation are highlighted. Pathogen/strain-specific DC behaviour is described in terms of produced cytokines, upregulated maturation markers and triggered cell death programmes. Activated or non-activated/immunosuppressive nature of infected DCs is in general emphasized by red or green filling colour respectively, together with the potential impact of their state on the course of infection.

Acknowledgement

We thank Dr Marek Link for fruitful discussions and for critical revision of the text. The work was supported by Specific Research grant of Ministry of Education, Youth and Sports of Czech Republic (SV/FVZ201107) and by a long-term organization development plan 1011.

Ancillary