Cord blood-derived mast cells
Pathogen-associated molecular pattern
In the present report we have analyzed whether human normal cord blood-derived mast cells (CBMC) could interact with bacterial products, especially lipopolysaccharide (LPS) from Escherichia coli and peptidoglycan (PGN) from Staphylococcus aureus, known as Toll-like receptor (TLR) 4 and TLR2 agonists, respectively. We found that both LPS and PGN induced significant release of not only tumor necrosis factor-α (TNF-α), but also IL-5, IL-10 and IL-13 by human mast cells (MC). We also established that the stimulation of CBMC with LPS or with PGN is mediated through interactions with TLR4 or with TLR2, respectively. Thus, our data indicate that activation of either TLR2 or TLR4 pathway may lead to a pro-Th2 immune response. However, the release of TNF-α induced by LPS, conversely to PGN, required the priming of CBMC by IL-4 and the presence of serum components, in particular soluble CD14. Of interest, stimulation by PGN, but not by LPS, induced release of histamine by human MC. Altogether, these findings provide the first evidence that human MC differentially respond towards bacterial components, and that their responses depend on TLR pathways and reveal human specificities in the pattern of cytokine production.
Mast cells (MC) are secretory cells strategically located at the host-environment interface, such as skin, lung and mucosal surfaces 1. MC have primarily been viewed for their deleterious effects in the pathogenesis of type I hypersensitivity reactions. Indeed, after allergic activation, they may immediately extrude granule-associated mediators, such as histamine, and generate lipid-derived mediators that induce immediate allergic inflammation 1. MC are crucial effectors and regulatory cells in Th2-dominated immune response since MC activation results not only in the release of inflammatory mediators such as histamine or cytokines such as TNF-α, but also Th2-associated cytokines such as IL-5, IL-13 and IL-10 2, which are important players in the pathogenesis of allergic reactions and host immune defense against bacteria.
Recent works have reported that MC could play a key role in immune responses against bacteria 3. Indeed, MC are activated in vitro by direct contact with various bacteria and then release a panel of mediators 3. In rodent models in vivo, MC were shown to promote the clearance of bacteria through the release of various mediators, oneof utmost importance being TNF-α, a potent chemoattractant of neutrophils 4.
While MC are activated by direct contact with bacteria, it has been postulated that bacteria-derived components might also affect MC functions. For instance, rodent MC are activated by LPS andconsecutively release IL-6, but not histamine 5. LPS, a major constituent of the outer membrane of Gram-negative bacteria, first interacts with the serum LPS-binding protein (LBP) 6. Thereafter, LBP facilitates the binding of LPS to CD14, which is expressed as a glycosylphosphatidylinositol (GPI)-anchored membrane receptor on monocytes and macrophages 7 and as a soluble form in serum (sCD14) 8. LPS is a member of the pathogen-associated molecular pattern (PAMP) family, which consists of specific elements sharing non-self structures recognized by host cells during infection 9. Another PAMP is peptidoglycan (PGN), a major cell-wall component of Gram-positive bacteria. Both of these structures induce major clinical manifestations of bacterial infections, essentially through the induction of the release of various mediators 10, 11.
Several bacterial components interact with immune cells through recognition by Toll-like receptors (TLR) 12. The TLR are type I transmembrane proteins, with an extracellular domain consisting of leucine-rich repeats and an intracellular domain with homology to the IL-1 receptor (IL-1R), called Toll/IL-1R (TIR) domain. Currently, ten TLR have been identified in human, among which, at least, TLR2 and TLR4 are involved in innate immunity 12, 13. In fact, TLR2 is the signaling receptor for a variety of microbial components, including PGN 14. Although certain species of LPS, derived from Leptospira interrogans15 or Porphyromonas gingivalis16, induce signaling via TLR2, they are structurally different from the typical enteric LPS from Escherichia coli or Salmonella Minnesota, which signals mainly through TLR4, as demonstrated using LPS-hyporesponsive strains of mice 17. Of note, very recent studies have demonstrated that rodent MC are activated by bacteria-derived components through the TLR-signaling pathways 18, 19. However, MC are heterogeneous in terms of morphology, receptor expression, mediator content and reactivity towards various stimulants, according to their localization in the body but also among animal species.
Because of the striking differences of MC reactivity and functions between animal species, it seemed of great interest to assess whether interactions between PAMP and TLR could also exist in the human MC system. Thus, we have analyzed the effects of bacterial compounds, PGN from Staphylococcus aureus and LPS from E. coli, described, respectively, as TLR2 and TLR4 agonists, on the activation of human normal MC. We demonstrate here for the first time that PGN and LPS activate human MC through TLR2- and TLR4-dependent pathways, respectively, leading to distinct secretory responses. Furthermore, we show that activation of either TLR2 or TLR4 pathway may lead to a pro-Th2 immune response.
2.1 Priming of cord blood-derived mast cells (CBMC) with IL-4 and serum components is necessary for the response of CBMC towards LPS, but not towards PGN
In a first set of experiments, we analyzed the effect of LPS or PGN on the time-dependent release of TNF-α by untreated CBMC. We observed that LPS (0.1 or 1 μg/ml) was unable to induce the release of TNF-α by CBMC in the absence of priming by IL-4 and serum, whatever timing of stimulation (Fig. 1A and data not shown). By contrast, PGN (5 or 10 μg/ml), applied for 6 h (an optimal time determined in preliminary experiments), induced a significant release of TNF-α by non-primed MC (Fig. 1B).
Because IL-4 was shown to prime human MC for their mediators' release 20, and since cellular responses towards LPS require serum components 6, and in particular CD14 7, 8, we performed further experiments after pretreatment of the cells with rhIL-4, in the presence or in the absence of serum, and then stimulated the differentiated cells with either LPS (0.1 and 1 μg/ml) or PGN (5 and 10 μg/ml) for 6 h. Of interest, we observed that priming with rhIL-4 or with serum was required for a significant LPS-induced release of TNF-α by CBMC (Fig. 1A). Conversely, we observed that, even if priming of CBMC with rhIL-4 or with FBS was not required for TNF-α release induced by PGN, IL-4-treatment induced a significant increase in the levels of TNF-α (Fig. 1B).
Since differentiation of CBMC with serum was mandatory for their response towards LPS, and because we observed that CD14 was not expressed on CBMC surface, as assessed by flow cytometry (data not shown), we hypothesized that this serum-dependent effect was related to the presence of sCD14 in the culture medium. Thus, using blocking anti-human CD14 mAb (clone MY4) directed against the functional domain of CD14, we observed a complete inhibition of the release of TNF-α by LPS-stimulated CBMC previously cultured with rhIL-4 and serum, as compared with untreated cells or cells treated by a non-blocking anti-CD14 mAb (Fig. 2).
2.2 Expression of TLR2 and TLR4 mRNA and proteins by human normal MC
Since we observed that priming of CBMC by rhIL-4 was necessary for their full response towards bacterial components, all the experiments performed to analyze the expression of TLR2 and TLR4 in CBMC, at the RNA as well as the protein level, were conducted on IL-4-primed MC. In parallel experiments, the monocytic THP-1 cell line was used as a positive control. In a first set of experiments, the synthesis of TLR2 and TLR4 mRNA was examined by reverse transcription (RT)-PCR in CBMC. Of note, PCR products showed positive bands of the expected size for TLR2 (394 bp) as well as for TLR4 (506 bp) in human MC (Fig. 3A).
To confirm that these specific mRNA give rise to the synthesis of the corresponding receptors, the expression of TLR2 and TLR4 was also evaluated by immunocytometry and Western blotting using specific anti-human TLR Ab. As shown in Fig. 3B, a constitutive and significant expression of TLR2 and TLR4 proteins was evidenced in CBMC. Whereas mastocytes constitutively expressed TLR2 and TLR4, they were barely detectable on the cell surface by flow cytometry (data not shown). This result is probably related to a low cell surface expression of these receptors and to the high fluorescence background of CBMC.
2.3 Activation of CBMC by LPS and PGN is mediated through interactions with TLR4 and TLR2, respectively
To assess the implication of TLR in the responses of CBMC towards LPS and PGN, we pre-incubated CBMC with blocking mAb directed against human TLR4 or human TLR2, used at optimal concentrations, prior to addition of the agonists. These experiments evidenced that the release of TNF-α by LPS-stimulated CBMC was totally abrogated by the addition of an anti-human TLR4 mAb, as compared with untreated MC or with MC treated with an isotype-control Ab (Fig. 4A). Of note, an anti-TLR2 mAb added before the stimulation of human MC with LPS had no effect on the release of TNF-α at the concentration tested (Fig. 4A). Conversely, the concentration of anti-TLR4 mAb that we used in blocking experiments with LPS had no effect on the release of TNF-α by PGN-stimulated CBMC (Fig. 4B). Finally, the release of TNF-α after stimulation of CBMC by PGN was only partially inhibited by the anti-human TLR2 mAb (approximately twofold decrease) at the molecular ratio PGN/ anti-TLR2 mAb of 1:2 (Fig. 4B).
2.4 PGN, but not LPS, induces histamine release by CBMC
We examined whether bacterial products could trigger the degranulation of human MC. For this purpose, we determined the release of histamine after 30 min of incubation of CBMC with PGN or with LPS. We performed our experiments on MC primed or not by rhIL-4 and/or in the presence or in the absence of FBS. Of note, while LPS, used at concentrations up to 1 μg/ml, was unable to induce histamine release by the various cell preparations, PGN induced a dose-dependent release of histamine by human MC, independently of the priming by rhIL-4 or of the presence of serum (Fig. 5 and data not shown). Besides, priming of human MC with rhIL-4 resulted in an approximately twofold increase in the release of histamine after PGN stimulation (Fig. 5).
2.5 Both LPS and PGN induce release of multiple cytokines by human CBMC
Since human MC are able to release not only TNF-α, but also a wide spectrum of cytokines, such as IL-5, IL-10 and IL-13, we measured the levels of these latter mediators in the supernatants of LPS- or of PGN-activated CBMC. These experiments were conducted within the previously described conditions for an optimal stimulation of CBMC, i.e. after pretreatment with IL-4 (20 ng/ml) during 5 days and in the presence of 2% of FBS. Interestingly, the results evidenced that both LPS or PGN, used at concentrations of 1 and 10 μg/ml, respectively, were able to induce the release of comparable and significant levels of IL-5, IL-10 and IL-13 by human MC (Fig. 6), indicating that these two distinct inflammatory stimuli were equally potent for inducing secretion of a pro-Th2 cytokine pattern.
We report here that CBMC constitutively express TLR4 and TLR2 mRNA and proteins. Whereas these receptors were detected by Western blot, these molecules are barely detectable by flow cytometry. Similarly, a low cell surface expression of these receptors has previously been observed on different resting human cell types 21–23. Nevertheless, our data show for the first time that these receptors are involved in the stimulation of human MC by their respective agonists, LPS and PGN. These results are in agreement with those previously obtained on mouse bone marrow-derived MC (BMMC) from TLR4- and TLR2-deficient mice 18, 19. However, while anti-TLR4 mAb completely blocked the release of TNF-α induced by LPS in our human model, the release of TNF-α induced by PGN was only partially inhibited by an anti-TLR2 mAb. Whereas this result confirms that PGN is an agonist of the TLR2 signaling pathway, we cannot exclude a selective recognition by heterodimers formed from TLR2 and other TLR. Indeed, Ozinsky et al. have described that PGN signaling involved cooperation between TLR2 and TLR6 in macrophages 24, while Takeuchi et al. have described that TLR6–/– macrophages significantly responded to PGN 25. Since we have recently noticed a differential expression pattern of TLR on human and murine MC (data not shown), the relevance of some other TLR signaling pathways in human MC activation will be examined in a separate work. Furthermore, the four currently cloned human PGN recognition proteins bind PGN, although the subsequent events following this binding are still unknown 26.
Besides, it appears that different factors can influence TLR-CBMC interactions. Indeed, we found that the priming of human MC by IL-4 was necessary for their release of TNF-α after stimulation by LPS. By contrast, this priming was not required for, but enhanced the release of TNF-α induced by PGN. As IL-4 up-regulated the expression of TLR4 on human peripheral B cells 27, we hypothesized that IL-4 might induce the same event on CBMC. However, IL-4 might also act on the expression of other molecules such as MD-2 28 or signaling molecules such as MyD88 29 or on other mechanisms, such as homo- or heterodimerization of TLR 24, all known to be involved in the activation of cells by PAMP.
Interestingly, interactions between LPS and human MC were also dependent on the presence of serum components, and particularly of sCD14. This is not surprising since LPS-dependent TLR4 signaling requires serum components in other cell types 8, 30, as in murine MC 18. Conversely, even if the activation of certain cell types by PGN seems to be mediated through interactions with mCD14 31, in our hands PGN-induced activation of CBMC was independent of the presence of sCD14.
Additionally, it appears that if LPS and PGN differentially interacted with human MC, they also induced differential activation according to their release of mediators. Indeed, we observed that the stimulation of CBMC by both agonists induced a significant release of TNF-α. However, it appears that high concentrations of LPS are required for the release of low levels of TNF-α by human MC via TLR4, as compared with macrophages. Our findings are consistent with those obtained by McCurdy et al. 18, using wild-type BMMC.
However, although we used the same commercial preparations of LPS, our results differed from those obtained by Supajatura et al. 19. These authors found approximately sixfold higher TNF-α levels (∼600 pg/ml/106 cells) after 3 h of stimulation of wild-type BMMC by 1 μg/ml of LPS, as compared with the ones that we and McCurdy et al. 18 measured after 6 h of stimulation (88±5 pg/ml/106 cells and 60 to 150 pg/ml/106 cells, respectively). The different responsiveness of BMMC to LPS in these two studies could be explained by the different sources of growth factors used. Indeed McCurdy et al. have used a medium containing only IL-3 to growth BMMC, while Supajatura et al. grew their BMMC in a medium that might contain also IL-4. Conversely, these two studies 18, 19 reported that BMMC released low levels of TNF-α (∼100 pg/ml/106 cells and 42±26 pg/ml/106 cells, respectively) after, respectively, 3 and 6 h of stimulation with high doses of PGN (100 μg/ml), as compared with our results (363±16 pg/ml/106 cells stimulated by 10 μg/ml of PGN during 6 h). However, these contrasting data once again underline the heterogeneity of MC responses towards the same stimulants in different animal species.
Besides, the finding that PGN, but not LPS, can induce the release of histamine by CBMC is in agreement with previous studies performed on BMMC showing that TLR2- but not TLR4-dependent stimulation of MC resulted in degranulation and Ca2+ mobilization 19. This is of particular interest, because histamine can modulate the production of cytokines by many cell types of the immune system 32, promoting for example the Th2 cell-type polarization by inhibiting the production of Th1-driving cytokines such as IL-12 or by favoring the production of Th2-related cytokines such as IL-10 33. In this study, we have demonstrated that both LPS and PGN induce production of Th2-cytokines such as IL-5, IL-10 and IL-13 from MC, in agreement with recent data showing the release of some of these cytokines after stimulation of murine cells 34.
In conclusion, our results show that human normal MC react towards various bacterial components, being thus able to participate in the immune responses accompanying bacterial infections in humans. Indeed, we demonstrated that CBMC are able to recognize PAMP through their interaction with TLR. Additionally, we showed that TLR4- but not TLR2-mediated activation of CBMC is dependent on environmental factors. Furthermore, distinct TLR pathways induced different patterns of responses in MC, and some of these immune responses are species-dependent. Finally, the fact that TLR2 agonists, but not TLR4 ones, induced the degranulation of human MC, opens a new area for investigations upon the downstream events occurring after activation of TLR on human MC.
4 Materials and methods
4.1 TLR agonists
PGN purified from S. aureus and LPS from E. coli, serotype O111:B4) (Sigma, St Louis, MO) were dissolved in endotoxin-free phosphate-buffered saline (ATGC Biotechnologies, Marne-la Vallée, France) and sonicated before use.
4.2 Generation of CBMC
CBMC were obtained by culture of CD34+ progenitors purified from normal cord blood (CB) as previously described 35. Briefly, heparinized CB, obtained after informedconsent of the patients, was centrifuged over a Ficoll separating solution (ATGC Biotechnologies). CD34+ cells, purified from light-density CB cells with a specific immunoaffinity system (Miltenyi Biotech, Paris, France), were seeded at 105 cells/ml in a complete serum-free medium (SFM) made of Iscove Modified Dulbecco's Medium (Bio Whittaker, Verviers, Belgium), L-glutamine (Sigma, St Louis, MO), penicillin, streptomycin (Life Technology, Cergy Pontoise, France), 1% bovine serum albumin (BSA) (Boehringer Mannheim, Grenoble, France), insulin (10 μg/ml; Sigma), transferrin (5 μg/ml; Sigma) and 100 ng/ml of recombinant human stem cell factor (Amgen Inc, Thousand Oaks, CA). At the beginning of the culture, rhIL-3 (1 ng/ml; Novartis, Basel, Switzerland), Flt-3 L (20 ng/ml; R&D Systems Europe Ltd., Abingdon, GB) and thrombopoietin (2 ng/ml; R&D Systems Europe Ltd.) were also added to induce expansion of hematopoietic progenitors. After 7 days, cells were replaced in complete SFM that was then renewed every 7 days. After 5 weeks in culture, >99% of the cells were identified as MC as previously described 35. In all experiments, unless otherwise noticed, these pure populations of CBMC were primed with rhIL-4 (Novartis) at 20 ng/ml during 5 days in complete SFM, in the presence or the absence of 2% FBS (Dutscher, Brumath, France) before use.
4.3 RT-PCR analysis of TLR2 and TLR4 mRNA expression by CBMC
Total RNA was isolated from 5×106 MC or 5×106 monocytic THP-1 cells used as a positive control, as previously described 35. One microgram of eachRNA sample was reverse-transcribed using the ProSTAR First-Strand RT-PCR kit (Stratagene, Amsterdam, The Netherlands). PCR reactions were performed in a final volume of 50 μl of a mixed solution containing 2.5 U of the SuperTaq DNA polymerase (ATGC Biotechnologies) and 10 pM from each of the following primers (Life Technology): β2-microglobulin (sense 5′-CAGTTCCACCCGCCTCAC; antisense 5′-CACATGTCTCGATCCCAG); TLR2 (sense 5′-GCCAAAGTCTTGATTGATTGG; antisense 5′-TTGAAGTTCTCCAGCTCCTG) and TLR4 (sense 5′-TGGATACGTTTCCTTATAAG; antisense 5′-GAAATGGAGGCACCCCTTC). Amplifications were performed exactly as previously described 35. The PCR products were analyzed by electrophoresis in a 1.2% agarose gel stained with ethidium bromide and were 600 bp (β2-microglobulin), 394 bp (TLR2) or 506 bp (TLR4).
4.4 Western blotting
Five millions CBMC or 5×106 THP-1 cells (positive control) were washed in PBS and then lysed in Laemmli buffer. After centrifugation, the supernatants containing the whole protein fraction were collected. Total proteins were separated using 10% SDS-PAGE and then transferred onto nitrocellulose membranes. The membranes were blocked during 30 min in blocking buffer (PBS, 0.05% Tween 20 and 5% non-fat dry milk), then probed overnight at 4°C with a goat anti-human TLR2 or with a rabbit anti-human TLR4 polyclonal Ab (TEBU, Le Perray-en-Yvelines, France), at the final concentration of 10 μg/ml. Finally, horseradish peroxidase (HRP)-conjugated donkey anti-goat Ab (TEBU) or HRP-conjugated goat anti-rabbit Ab (Pharmacia, Orsay, France) was added at the final concentration of 1 μg/ml during 1 h at room temperature. Specific bands at 89 kDa for TLR4 and at 93 kDa for TLR2 were visualized with an enhanced chemiluminescence ECL detection kit (Pierce, Bezons, France).
4.5 Analysis of the release of mediators by stimulated CBMC
Pure CBMC suspensions, primed or not with rhIL-4, in the presence or absence of 2% of FBS, were seeded at 106 cells/ml, then stimulated with LPS from E. coli serotype O111:B4(0.01 to 1 μg/ml) or with PGN from S. aureus (1 to 10 μg/ml). After 3, 6, 24 and 48 h of incubation, cell-free supernatants were collected, centrifuged and frozen until determination of TNF-α content by the use of an ELISA kit according to the manufacturer's instructions (Coulter, Villepinte, France).
For histamine-release experiments, CBMC, primed or not by rhIL-4, in the presence of FBS, were stimulated with 0.1 or 1 μg/ml of LPS, with 2 or 10 μg/ml of PGN, or with PMA (10–9 M; Sigma) plus ionomycin (10–6 M; Sigma). After 30 min of incubation, histamine content in cell-free supernatants and in cell pellets was evaluated as previously described 36. Histamine release was expressed, after subtraction of the spontaneous release, as the net percentage of the total intracellular content of histamine.
For IL-5-, IL-10- and IL-13-release experiments, CBMC were stimulated for 6 h in the presence of 1 μg/ml of LPS or of 10 μg/ml of PGN. The content of the various cytokines in the cell-free supernatants was then analyzed using specific ELISA kits (Coulter).
4.6 Flow cytometric analysis
Cells (5×105) were incubated with purified PE-conjugated mouse anti-human CD14 mAb (BD PharMingen, San Diego, CA), PE-conjugated mouse anti-human TLR4 (Clinisciences, Montrouge, France), PE-conjugated mouse anti-human TLR2 (Clinisciences) or PE-conjugated mouse IgG isotype control (BD PharMingen) during 1 h at 4°C in a buffer composed of PBS, 1% BSA, 0.2% sodium azide and 10% human serum AB (Bio Media, Boussens, France). After three washes, cells were resuspended in 0.4 ml of the buffer and analyzed on a FACScan flow cytometer (BD PharMingen).
4.7 Blocking experiments
Involvement of sCD14 in the interactions between CBMC and LPS was investigated by the addition of a blocking anti-CD14 mAb (clone MY4; Coulter) or of a non-blocking anti-CD14 mAb (clone MO2; Coulter) to the medium at a final concentration of 10 μg/ml, 1 h before the addition of stimulants. To assess whether PGN or LPS interactions with CBMC are mediated by TLR2 or TLR4, respectively,we treated cells with specific mouse mAb against human TLR2 (clone TL2.1, 20 μg/ml) or against human TLR4 (clone HTA 125, 10 μg/ml) (Clinisciences) or IgG2aκ isotype control (Sigma) during 1 h at room temperature prior to addition of the agonists. These concentrations of anti-human TLR2 mAb and of anti-human TLR4 mAb were chosen because they induced optimal inhibition of the corresponding activation in preliminary dose-dependent experiments. After 6 h of stimulation, supernatants were harvested and TNF-α content measured.
4.8 Statistical analysis
Data are presented as mean ± standard error (SE). Significant differences were determined with Fisher's protected least-significant difference test (p<0.05).
We would like to express our appreciation to Amgen Inc and to Novartis for their generous contributions. We would like to thank also René Lai-Kuen, Laurence Leriche and François Machavoine for expert technical assistance and Antoine Ribadeau Dumas for fruitful discussions. S. Varadaradjalou is a recipient of grants from The Fondation Marcel Bleustein-Blanchet pour la Vocation and from the Association Française pour les Initiatives de Recherche sur le Mastocyte et les Mastocytoses (AFIRMM).