Dr G. Lombardi, Immunoregulation Laboratory, Department of Nephrology and Transplantation, 5th Floor Thomas Guy House, Guy's Hospital Campus, King's College London, Guy's, King's and St Thomas School of Medicine, London SE1 9RT, UK. Email: firstname.lastname@example.org Senior author: Giovanna Lombardi
Dendritic cells produce cytokines that regulate the class of the adaptive immune response. Microbial recognition is mediated, at least in part, by pattern recognition receptors such as Toll-like receptors, which influence dendritic cell maturation. In humans it is not yet clear how intact pathogens modulate the developing immune response. To address the effects of intact pathogens on the maturation and effector functions of human dendritic cells, we investigated their responses to a number of microbial pathogens. We studied a range of micro-organisms including Gram-negative bacteria (Escherichia coli and Salmonella enterica sv. typhimurium), Gram-positive cocci (Staphylococcus aureus) and atypical bacteria (Mycobacterium tuberculosis and Mycoplasma hominis) as well as the human protozoal parasite Trichomonas vaginalis. The micro-organisms were fixed in formaldehyde to prevent replication whilst preserving surface morphology. All the pathogens induced similar up-regulation of dendritic cell activation-associated cell surface markers but there was a profound difference in the patterns of cytokines produced by the stimulated dendritic cells. Some pathogens (E. coli, Salmonella enterica sv. typhimurium and S. aureus) induced interleukin-12 (IL-12), IL-10 and interferon-α whereas others (M. tuberculosis, Mycoplasma hominis and T. vaginalis) induced only IL-10. This differential effect was not altered by costimulation of the dendritic cells through CD40. These results support the notion that human dendritic cells are plastic in their response to microbial stimuli and that the nature of the pathogen dictates the response of the dendritic cell.
The immune system has evolved to protect individuals from potentially dangerous antigens, such as invading pathogens. The generation of an appropriate response is essential and higher animals have evolved complex recognition systems to ensure that relevant responses are induced.1,2 Dendritic cells (DCs) are potent, professional antigen-presenting cells (APCs) that recognize pathogens, either directly or indirectly, by sensing changes in the environment such as cell damage or inflammation.3,4 Following activation by appropriate ‘danger signals’, peripheral DCs begin to mature and migrate from the periphery to local lymph nodes where maturation is completed following interaction with the appropriate T cells.5,6 Tissue-resident immature DCs can identify pathogens directly by recognizing pathogen-associated molecular patterns (PAMPs) using highly conserved pattern recognition receptors (PRRs) such as the Toll-like receptors (TLRs).7 The responsiveness to a given PAMP is linked to the expression of its relevant TLR – for example TLR4 is necessary for the recognition of lipopolysaccharide (LPS)8 and TLR2 identifies peptidoglycan from Gram-positive bacteria and lipoprotein from different bacteria.9–14 The different TLRs interact to form homo- or heterodimers; TLR6, in combination with TLR2, recognizes zymosan and peptidoglycan15–17 and TLR9 and TLR5 have been shown to recognize bacterially derived CpG DNA motifs18 and flagellin,19 respectively. Following ligand binding, the receptors activate intracellular signalling pathways via a variety of mediators including MyD88 and NFκB.20 The net result of TLR activation is the expression of a variety of genes that have direct and indirect effects on the developing immune response.7,21
DC maturation involves enhanced expression of major histocompatibility complex (MHC) class I and class II molecules, increased formation of MHC–peptide complexes and their delivery to the cell surface, in combination with up-regulation of costimulatory molecules such as CD80, CD86 and CD40.22–25 It is now clear that the interaction between TLRs and PAMPs plays a key role in enhancing the release of cytokines, such as interleukin-10 (IL-10), IL-12 and type I interferons. Production of IL-12 and IL-10 by mature murine DCs can be elicited by many pathogens or their products.22,25–27 In the human system, Gram-negative bacteria, but not Gram-positive bacteria, prime DCs to produce the IL-12 family of cytokines.28–30 In contrast, Mycobacterium tuberculosis blocks DC maturation and induces IL-10 release by targeting DC-SIGN.31 DCs have also been shown to produce type I interferons in response to micro-organisms as well as in response to viral infection or following interaction with T cells.32–38 In mice, enhanced IL-12 or IL-10 production by DCs requires two signals, one from the encounter with the pathogen and the second from the interaction with T cells,39–41 suggesting that the role of the pathogen in mice is to ‘sensitize’ DCs such that they can be fully activated following encounter with T cells. Similar studies in human-derived DC have not been reported.
It is now well established that DCs are both the initiators and the regulators of the developing immune response.27,42 The signals that instruct naive T-cell development are not yet fully understood but the cytokine milieu plays a central role with IL-12 driving T helper type 1 (Th1) responses whereas IL-10 inhibits them, favouring Th2-type responses instead.22 Type I interferons have been shown to regulate T-cell differentiation, but their role has not yet been clearly resolved.43,44 Since DCs control the developing T-cell response, the pattern of cytokines that they release when activated is believed to play an important role in determining the ultimate nature of the T-cell response. However, it is not yet clear how cytokine production is regulated in DCs. One possible explanation is that there are different DC subsets that are preprogrammed to produce different cytokine profiles and selective activation of these different subsets dictates the fate of the developing immune response. An alternative model is that each individual DC is pluripotential and the nature of the maturation signal determines the pattern of the cytokines that are produced during subsequent encounters with T cells.4,26,45
To examine the effects of the binding of intact pathogens on human DCs, we studied three pathogens associated with acute sepsis, Escherichia coli, Salmonella enterica sv. typhimurium, Staphylococcus aureus and three that are more associated with chronic infections, Mycoplasma hominis, Mycobacterium tuberculosis and Trichomonas vaginalis. Here we show that monocyte-derived human DCs possess significant plasticity and that different pathogens induce the production of different cytokines by a single subset of DCs. These results suggest that the same subset of DC has the potential to have different effects dependent on the primary encounter with intact pathogens. Specifically our results suggest that the pathogens typically associated with acute infections and sepsis induce IL-12 and type I interferons, whereas those that are more often associated with chronic infections gave rise to IL-10-secretion only.
Materials and methods
Pathogens and their preparation
The micro-organisms used in this study were E. coli, Salmonella enterica sv. typhimurium (hereafter S. enterica sv. typh.), Staphylococcus aureus (hereafter Staph. aureus), Mycoplasma hominis, Mycobacterium tuberculosis and T. vaginalis.
Enteropathogenic E. coli (EPEC strain, characterized by the presence of eaeA and bfpA genes by polymerase chain reaction) and Staph. aureus were obtained, respectively, from patients affected by enteritis and acute cystitis. Escherichia coli, Staph. aureus and S. enterica sv. typh. (reference strain ATCC, 14028s) were cultured in Lauria Bertani (LB) broth at 37°. Mycoplasma hominis reference strain PG21 was cultured in BEA medium.46Mycobacterium tuberculosis reference strain H37RV, was cultured in liquid medium Middlebrook 7H9 medium and T. vaginalis strain SS-22, was obtained from a patient affected by acute trichomoniasis. After axenization and before conducting the experiments, these protozoa were tested by polymerase chain reaction to detect any bacterial presence using universal bacterial primers. The T. vaginalis were cultured by daily 1 : 16 passages in fresh Diamond's trypticase yeast maltose (TYM) medium with 10% fetal bovine serum. The absence of symbiont mycoplasmas was assessed by both cultural and molecular methods, as described elsewhere.47,48
The numbers of all bacteria were evaluated as CFU (colony-forming units) in specific solid media. The T. vaginalis were counted using a haemocytometric chamber. All micro-organisms were extensively washed in LPS-free phosphate-buffered saline (PBS) then fixed for 4 hr in 4% paraformaldehyde. After fixation, cells were washed in PBS.
Generation of DCs and pathogen stimulation
DCs were generated from peripheral blood mononuclear cells separated from buffy coats obtained from the National Blood Transfusion Centre (Colindale, UK) as previously described.32 DCs were cultured with RPMI-1640 (Invitrogen, Paisley, UK), supplemented with 10% fetal calf serum (FCS, SeraQ, Sussex, UK), penicillin, streptomycin and l-glutamine, which contained 10 ng/ml IL-4 (First Link, West Midlands, UK) and 20 ng/ml granulocyte–macrophage colony-stimulating factor (GM-CSF; kindly donated by Dr S. Brett, GlaxoSmithKline, Stevenage, UK). Purity of DC preparations was evaluated by fluorescence-activate cell sorter (FACS) analysis using anti-CD11c and anti-CD3 monoclonal antibodies (mAbs) and was always more than 95%. On day 5 of DC culture, the Gram-negative bacteria (E. coli, S. enterica sv. typh.), Gram-positive cocci (Staph. aureus), the atypical bacteria (Mycobacterium tuberculosis, Mycoplasma hominis) and the human protozoal parasite (T. vaginalis) were added at different doses and then incubated at 37°. The following pathogens, E. coli, S. enterica sv. typh., Staph. aureus and Mycoplasma hominis were used at between 105 and 108 cells/ml, T. vaginalis was used at the maximal concentration of 107 cells/ml to account for its larger size, and Mycobacterium tuberculosis was used at a concentration between 104 and 107 because this was considered to be more representative of the concentrations present during infection. DCs were also cultured alone (iDCs) or in the presence of LPS (E. coli strain 026:B6, Sigma, St Louis, MO) at a concentration of 50 ng/ml. DCs were used for FACS analysis 48 hr after the exposure to the different pathogens. Supernatants were harvested after 24 hr and 48 hr and were stored at −20° until analysis.
Cell surface staining
After 48 hr of incubation with pathogens, DCs were harvested, counted and plated at a concentration of 105 per staining condition. Mouse anti-human mAbs, directly conjugated to fluorescein isothiocyanate (FITC) or phycoerythrin (PE), specific for the following molecules were added: HLA-DR (Sigma), CD40 (Caltag, Burlingame, CA), CD83 (Caltag) and CD86 (Caltag). After a 1-hr incubation the cells were washed, fixed in 1% paraformaldehyde (PFA; Merck, Letterworth, UK) and analysed using a FACScalibur flow cytometer and cellquest software (Becton Dickinson, San Jose, CA).
CD40 ligation of DCs
DCs were cocultured with CD40L-expressing L cells (CD40L) either simultaneously, or 48 hr after pathogen addition. Supernatants were harvested after 24 hr or 48 hr from the beginning of the coculture and stored at −20°.
Enzyme-linked immunosorbent assay (ELISA)
Supernatants collected from DCs incubated with pathogens in the presence or in the absence of CD40L transfectants were analysed by ELISA. IL-12p70 and IL-10 were measured using a standard indirect sandwich ELISA. The following pairs of purified and biotinylated antibodies were used: anti-IL-10 and anti-IL-12 (BD PharMingen, San Diego, CA). Optical density (OD) was measured at 450 nm on an ELISA plate reader (Titertek Multiscan PLUS, Helsinki, Finland).
Anti-viral assay for type I interferon detection
The production of interferon-α (IFN-α) was measured using an antiviral assay as previously published.32 Briefly, duplicate, serial three-fold dilutions of supernatants from DCs cultured with pathogens were added to confluent cultures of A549. The cells were then incubated for 24 hr after which supernatants were discarded and replaced with murine encephalomyelitis (EMC) virus for 1 hr. The virus was then removed and cell viability was assessed by staining with 0·25% methyl violet (Sigma) in 2% PFA (Merck) and the OD was measured by spectrophotometry using an automated plate reader. The concentration of type I interferon was determined by comparison with a standard curve prepared from IFN-α2a (Roferon) at known concentrations and the antiviral activity of the original sample calculated.
Maturation of human DC by pathogens
To determine how different pathogens affect DC maturation, immature DCs were prepared from human monocytes (see Materials and methods section). At day 5 DCs were homogeneous with 100% expression of CD11c (data not shown). DCs were incubated for 48 hr in the presence of a variety of intact, paraformaldehyde-fixed pathogens. Surface expression of activation/maturation markers (HLA-DR, CD86, CD40 and CD83) was assessed by flow cytometric analysis using unstimulated immature DCs (iDCs) as a negative control and DCs stimulated by E. coli-derived LPS (LPS-DCs) as a positive control. We compared different doses of intact E. coli, S. enterica sv. typh., Staph. aureus, Mycoplasma hominis, Mycobacterium tuberculosis and T. vaginalis and Fig. 1 shows a representative result with Mycoplasma hominis. There was a dose-dependent increase in the expression of all of the activation/maturation markers studied.
Figure 2 shows the effects of a single, optimal concentration of all the pathogens tested on the expression of MHC class II, CD86, CD40 and CD83. All the pathogens induced changes in cell surface expression of DC markers. The data obtained from 18 experiments are summarized in Table 1. Intact pathogens induced maturation/activation of DCs, with some markers such as CD86 being more consistently up-regulated by most of the pathogens tested. However, T. vaginalis had little effect on the expression of all cell surface markers.
Table 1. Effects of different pathogenic stimuli on DC maturation
DCs were cultured with the different stimuli for 48 hr and stained for the markers indicated. Results are expressed as the proportion (and percentage) of experiments where the mean fluorescence intensity (MFI) of the marker was up-regulated by more than 10% as compared with the MFI of the marker on immature DCs.
Salmonella enterica sv. typhimurium
Production of IL-10 and IL-12 from human DCs exposed to different pathogens
To assess whether the different pathogens had differential effects on the cytokine profile of activated DCs, iDCs were cultured with various concentrations of the pathogens and cytokine production (IL-12 and IL-10) was measured in the supernatants by ELISA after 24 and 48 hr of culture. A dose-dependent increase in IL-10 secretion was induced by both E. coli and Mycoplasma hominis after 48 hr of culture. Only E. coli led to measurable IL-12 release, and this was only seen at the highest concentration of the organism. In contrast, LPS derived from E. coli led to the release of IL-10 only (Fig. 3). The data from 48 hr of culture are shown and similar cytokine levels were seen after 24 hr incubation with these pathogens (data not shown).
Figure 4 shows the cytokine profile in response to all the pathogens tested after 48 hr of coculture with DCs. Two different response patterns were observed –E. coli, S. enterica sv. typh. and Staph. aureus (all at 108 cells/ml), induced production of both IL-10 (> 90% of experiments) and IL-12. However, IL-12 was not produced in all experiments – from 12 separate experiments IL-12 was induced by E. coli in seven (58%), S. enterica sv. typh. in four (33%) and Staph. aureus in five (42%). LPS from E. coli induced IL-10 production by DCs in all the experiments; however, IL-12 release was detected in only two out of 14 experiments (14%). This was not a consequence of the dose of LPS used. In fact, while IL-10 was induced in a dose-dependent manner, IL-12 was undetectable (Fig. 5). Mycobacterium tuberculosis (107 cells/ml) promoted the release of IL-10 in the absence of any detectable IL-12 and both Mycoplamsa hominis (107 cells/ml) and T. vaginalis (106 cells/ml) induced IL-10 production alone, although the levels detected were always less than those seen with other pathogens (Fig. 4). Similar cytokine levels and profiles were seen after 24 hr of culture (data not shown). Hence, although all of the pathogens induced the maturation of DCs and IL-10 secretion, some pathogens did not promote the release of IL-12.
Production of type I interferons from human dendritic cells exposed to different pathogens
Type I interferons are potent antiviral cytokines that play a central role in the development of Th1 cells in humans.44 We therefore examined their production in supernatants from stimulated DCs. Figure 6 shows that interferon was detected in DCs cultured in the presence of E. coli, S. enterica sv. typh. and Staph. aureus which correlated with pathogens that also induced IL-12 production for one representative experiment. Those that did not stimulate IL-12 release –Mycoplasma hominis, T. vaginalis and Mycobacterium tuberculosis– did not induce IFN-αβ production when compared to unstimulated, immature DCs. To ensure that these pathogens did not induce the production of an IFN-α subtype that had limited antiviral activity (e.g. IFN-α1) supernatants from Mycobacterium tuberculosis-stimulated DCs were analysed with a commercial ELISA (PBL Laboratories, New Jersey, NJ) that detects all IFN-α subtypes. No significant IFN-α was detected in this assay (data not shown) confirming that paraformaldehyde-fixed Mycobacterium tuberculosis do not induce the production of IFN-α.
Again there was some heterogeneity in the response of different DC preparations from different donors and not all preparations induced type I interferon – for E. coli the interferon was detected in 10 of 12 experiments (83%), for S. enterica sv. typh., eight out of 13 experiments tested positively (62%) and Staph. aureus induced interferon secretion in eight out of 11 experiments (72·7%). As previously shown for IL-12 production, LPS induced IFN-α release in only two out of nine experiments at these time-points (22%).
Effects of CD40 ligation on production of different cytokines from pathogen-stimulated human DCs
In view of reports in murine models showing that DCs stimulated with bacteria do not release cytokines unless they are activated by contact with CD40L expressed on the surface of T cells, we examined cytokine production in our pathogen-stimulated DCs after ligation of CD40. To determine whether those donors who did not induce IL-12 when stimulated with E. coli could be induced to express this cytokine after ligation of CD40, three donors who were ‘poor IL-12 producers’ were simultaneously exposed to pathogens and CD40L-expressing transfectants. Results from a representative experiment are shown in Fig. 7(a). There was a marked increase in IL-10, with all pathogens tested and a slight up-regulation in IL-12 release with some pathogens (S. enterica sv. typh. and Staph. aureus) compared with immature DCs (iDCs). However, in the presence of the CD40L transfectants, the pathogens that failed to induce IL-12 secretion when used in isolation did not cause any increase in IL-12 release above the baseline (immature DCs). When CD40L transfectants were added to the DCs 48 hr after the addition of the pathogens (Fig. 7b) a similar phenomenon was seen with an increase in IL-10 production and a modest increase in IL-12 with some organisms. However, CD40L stimulation did not modify the clear difference between those pathogens that induce large amounts of IL-12 and those that did not do so, suggesting that ligation of CD40 amplifies the response to the pathogens but does not fundamentally modify the nature of the response.
We also examined the effects of CD40 ligation on the production of type I interferons. In a series of eight experiments, no significant differences were observed between CD40L-stimulated and non-stimulated DCs after they were exposed to the pathogens. Figure 8 shows a typical experiment when the pathogens were added simultaneously with the CD40L transfectants. Similar results were obtained when the pathogens were added before ligation of CD40 on DCs (data not shown).
We have examined the effects of different pathogens on the maturation and activation of human monocyte-derived DCs. Paraformaldehyde-fixed intact pathogens were used to enable us to study the role of the entire micro-organism and to evaluate the effects of microbiological structures without the influence of secreted metabolic products. Paraformaldehyde has been suggested to preserve most of the structural features on the organism concerned. Furthermore, the epitopes of interest are not normally obscured.49,50 An additional benefit of fixed pathogens is that variability induced by microbe replication and changes in the number of microbes during the course of the experiment are eliminated. All of the organisms tested had an effect on DC maturation. In addition, the pathogens had a wide range of effects on the cytokine profile of activated DCs. These differences are of physiological significance as the specific conditioning stimuli caused DCs to produce very different cytokine profiles. On the basis of these effects we can divide the pathogens studied into two groups – group 1 includes pathogens such as E. coli, S. enterica sv. typh. and Staph. aureus (which can induce DCs to release IL-12 and type I interferon in addition to IL-10) and group 2 includes Mycoplasma hominis, Mycobacterium tuberculosis and T. vaginalis (which induce DCs to produce IL-10 alone). In addition, Mycoplasma hominis and T. vaginalis seem to have a reduced ability to induce cytokine release by DCs. Notably, these two pathogens are structurally very different from the group 1 pathogens. For example, Mycoplasma hominis lacks a cell wall and T. vaginalis is a protozoa as opposed to a bacterium. It is noteworthy that the pathogens that induced both IL-12 and type I interferons are normally associated with acute, severe infections and sepsis whereas pathogens that induce IL-10 alone are usually associated with chronic indolent disease. Further experiments in animal models will be required before we can definitively conclude that DC induction of IL-12 and type I interferons is associated with acute sepsis while IL-10 release is associated with chronic infections. Nevertheless, our in vitro experiments indicate that this may be the case. Our experiments support the notion of DC plasticity and show, for the first time, that intact micro-organisms can condition DCs to release cytokines that direct the phenotype of T cells.
In general DCs respond to foreign particles by up-regulating both cell surface expression of a variety of maturation markers and releasing a wide range of cytokines.22,24,25,27 Some heterogeneity among the different DC preparations has been observed, mainly for the up-regulation of CD83 following the interaction with pathogens. The variability observed is a common finding in experiments involving human DC51–53 and presumably reflects genetic heterogeneity that is absent in experiments using inbred mice. We found a clear difference between the production of cytokines and the induction of maturation markers – all pathogens tested led to similar changes in DC phenotype, however, only some induced IL-12 and interferon production by DCs. Hence it seems likely that specific intracellular signalling pathways are responsible for these different responses. Type I interferons are usually regarded as antiviral cytokines but it is now clear that they play a role in a wide variety of host defence systems. We have previously shown that IFN-α is produced following the interaction between DCs and naive T cells, suggesting that these cytokines play a role during the initiation of the immune response and it has been shown that they are important for the development of a Th1 response.32,44 Studies with LPS and whole bacteria have shown that they too can induce production of this key initiator of immune responses.33–37 Previously published work using live Mycobacterium tuberculosis and human DCs37,54 suggest that in this experimental model, large amounts of IFN-α are produced, but in our assay using fixed bacteria this was not seen, suggesting that surface ligation of extracellular receptors may have effects that are different from those seen during intracellular replication.
Previous studies have examined the effects of isolated TLR ligands on the maturation/activation of DCs and have provided valuable information relating to the key events in early pathogen recognition.55 We have chosen to use intact pathogens that, presumably, activate a number of different TLRs and we believe that this experimental system mimics more closely the situation in vivo where DCs are exposed to micro-organisms rather than to their components in isolation. The micro-organisms were inactivated to avoid the presence of metabolites. It would be of interest to repeat the experiments presented in this paper with live pathogens. Our experiments indicate that there are significant differences in the response to isolated bacterial products and intact pathogens. For example, intact E. coli induced IL-12 and IFN-α production in addition to IL-10, whereas the purified LPS from E. coli induced predominantly IL-10 release in our experimental system, suggesting that LPS is not the only component involved in the effect seen with E. coli. We also found marked differences between the responses in human cells and reports from other groups in murine models. For example, LPS from E. coli has been shown to induce both IL-10 and IL-12 production in the murine models.56–60 In the human system, the production of IL-12 in response to LPS has been shown to be variable and is affected by the way that DCs are generated61 and by the amount of serum used in the culture media.62 In addition, the amount of IL-12 produced in response to LPS can be affected by the amount of IL-10 endogenously produced. As previously described, LPS induced only small amounts of IL-12 that were increased by neutralizing IL-10.63 Furthermore, it has been shown that bacterial interactions with murine DCs can have a very limited effect and CD40L engagement is required before significant cytokine release occurs.4,26,45 By contrast, in our human model we find that bacterial engagement per se can induce cytokine release although this is amplified by costimulation with CD40L. These data suggest that there are profound differences in the responses in humans and mice and extrapolations from murine studies with bacterial components to human responses with intact pathogens may not be straightforward.
It has been previously shown that there is an inverse relationship between IL-10 and IL-12 production by activated DCs.64–66 Although the autocrine effect of IL-10 in suppression of IL-12 production by DCs has been reported,63,66–68 it is not known whether this autocrine effect is involved in the inverse relationship reported by others. In our experimental system we found that both IL-10 and IL-12 can be produced simultaneously although we have been unable to determine whether the same DCs simultaneously produce both cytokines. It therefore seems likely that the reported ‘reciprocal’ arrangement between IL-10 and IL-12 is not as straightforward as is currently supposed and further work will be required to understand the nature of the relationship between these two cytokines.
It was noteworthy that some DC preparations failed to secrete both IL-12 and IFN-α in response to group 1 pathogens. It is not yet clear whether this failure was related to preparation artefacts, introduced during the separation and maturation of the intact DC, or whether some individuals are predisposed to produce different cytokines in response to the same pathogen. Studies to address this possibility and to determine the clinical relevance of this observation are under way.
Thus the host response to invading micro-organisms is orchestrated by the APCs that identify and present antigenic fragments to the lymphocytes in association with a cytokine milieu that dictates the subsequent course of the immune response. We found that the cytokine response of the DCs is profoundly influenced by the nature of the pathogen. Our experiments indicate that studies of TLR engagement in isolation may be poorly representative of the situation when entire organisms are studied and further work will be required to determine how the activated DC integrates the signals that arise when multiple TLRs are engaged by intact pathogens.
This work was financed at least in part by the Wellcome Trust.