Bordetella pertussis and B. parapertussis are the etiological agents of pertussis, yet the former has a higher incidence and is the cause of a more severe disease, in part due to pertussis toxin. To identify other factors contributing to the different pathogenicity of the two species, we analyzed the capacity of structurally different lipooligosaccharide (LOS) from B. pertussis and LPS from B. parapertussis to influence immune functions regulated by dendritic cells. Either B. pertussis LOS and B. parapertussis LPS triggered TLR4 signaling and induced phenotypic maturation and IL-10, IL-12p40, IL-23, IL-6, and IL-1β production in human monocyte-derived dendritic cells (MDDC). B. parapertussis LPS was a stronger inducer of all these activities as compared with B. pertussis LOS, with the notable exception of IL-1β, which was equally produced. Only B. parapertussis LPS was able to induce IL-27 expression. In addition, although MDDC activation induced by B. parapertussis LPS was greatly dependent on soluble CD14, B. pertussis LOS activity was CD14-independent. The analysis of the intracellular pathways showed that B. parapertussis LPS and B. pertussis LOS equally induced IκBα and p38 MAPK phosphorylation, but B. pertussis LOS triggered ERK1/2 phosphorylation more rapidly and at higher levels than B. parapertussis LPS. Furthermore, B. pertussis LOS was unable to induce MyD88-independent gene induction, which was instead activated by B. parapertussis LPS, witnessed by STAT1 phosphorylation and induction of the IFN-dependent genes, IFN regulatory factor-1 and IFN-inducible protein-10. These differences resulted in a divergent regulation of Th cell responses, B. pertussis LOS MDDC driving a predominant Th17 polarization. Overall, the data observed reflect the different structure of the two LPS and the higher Th17 response induced by B. pertussis LOS may contribute to the severity of pertussis in humans.
Pertussis or whooping cough is an acute respiratory disease caused by Bordetella pertussis or B. parapertussis, two strictly human pathogens (1). These two bacteria possess overlapping subsets of virulence factors, with the notable exception of pertussis toxin, which is not expressed by B. parapertussis (2). Whooping cough in neonates infected with B. pertussis develops huge bacterial loads and massive lymphocytosis, and may be lethal. In contrast, severe disease was never observed following B. parapertussis infection (3, 4, 5). Although pertussis toxin is considered to play a critical role for disease occurrence and severity (6, 7), it has also been suggested that pertussis toxin may not play a decisive role in causing the typical symptoms of whooping cough (8).
Another major difference between the two species is represented by the structure of their LPS. These molecules share a degree of common architecture, both having a penta-acylated lipid A and a branched-chain core oligosaccharide, yet also structural differences. B. pertussis lipooligosaccharide (LOS)4 lacks an O-side chain, having in its place a nonrepeating trisaccharide, whereas B. parapertussis LPS has an O-Ag structure consisting of a homopolymer of 2,3-dideoxy-2,3-di-N-acetylgalactosaminuronic acid (9, 10, 11).
LPS is a critical molecular pattern, exerting a major role in the host-bacteria relationship, particularly driving a number of activation processes in professional and nonprofessional APC such as macrophages and dendritic cells (DC). Thus, differences in LPS structures may fundamentally affect differences in pathogenesis (12).
To evaluate whether structural differences in LPS between the two species may in some way explain different pathogenicity, the impact of B. pertussis LOS and B. parapertussis LPS was studied on human monocyte-derived DC (MDDC), an ex vivo model that allows the evaluation of aspects of the regulation of immune response.
DC are professional APC distributed in lymphoid and nonlymphoid organs with the ability to translate innate into adaptive immunity (13). They express a wide array of receptors specialized in pathogen recognition, permitting detection and identification of pathogen-associated molecular patterns (14, 15). TLR4 is required for detection of LPS (15, 16), together with other proteins directly involved in ligand-binding such as MD-2, which mediates subsequent receptor activation, and CD14, which controls ligand presentation to the TLR4/MD-2 complex (17). The interactions between LPS and the TLR4 receptorial complex crucially influence the amplitude of cellular responses (12, 17).
We previously showed that B. pertussis LOS induces partially mature MDDC not expressing IL-12p70. Such committed MDDC are able to induce Th2 polarization in vitro (18). In this study, to propose a differential pathogenetic role of LPS molecules from B. parapertussis and B. pertussis, we adopted a strategy based on comparative analysis of the ability of B. pertussis LOS and B. parapertussis LPS to induce TLR4-mediated signaling and the functions that are specifically acquired by MDDC upon maturation. Escherichia coli LPS served as control. The consequent approach was the identification of MDDC intracellular pathways potentially helping in the transduction of activation signals. The last approach was the exploration of the regulatory role exerted by B. pertussis LOS and B. parapertussis LPS on MDDC in the polarization of T cells, which may have a crucial role in Bordetella pathogenicity.
Materials and Methods
Ultrapure E. coli LPS from O111:B4 strain, Pam2CSK4 (synthetic bacterial lipoprotein S-[2,3-bis(palmitoloxy)-(2RS)-propyl]-[cysteinyl-[S]-lysil-[S]-lysine x 3 CF3COOH), p44/42 MAPK (ERK1/2) inhibitor PD98059, and p38 MAPK inhibitor SB203580 were purchased from InvivoGen. Human recombinant GM-CSF and recombinant IL-4 were from R&D Systems. Culture supernatants (4%, v/v) of IL-4–62 cell line were also used as a source of IL-4 (19). Recombinant IL-2 was obtained from Roche. Fluorochrome-conjugated anti-human CD1a, CD14, CD38, CD80, CD83, and CD86 mAbs and purified anti-human IFN-γ-inducible protein (IP)-10 mAb were from BD Biosciences. Anti-human CD14-neutralizing mAb was from R&D Systems. Rabbit polyclonal IgG anti-phospho-STAT1 (Tyr701), anti-phospho-ERK1/2 (Thr202/Tyr204), anti-phospho-p38 MAPK (Thr180/Tyr182), and anti-phospho-IκBα (Ser32) were from Cell Signaling Technology. Mouse anti-β-tubulin and brefeldin A were from Sigma-Aldrich. Mouse anti-STAT1 was from Transduction Laboratories.
B. parapertussis and B. pertussis LPS molecule purification
B. pertussis LOS from BP338 strain and B. parapertussis LPS from CN8234 strain were purified and characterized as described (20). By SDS-PAGE characterization, purified B. pertussis LOS revealed distinct band A and band B, whereas B. parapertussis LPS revealed ladder-like bands showing that the B. parapertussis LPS has an O-Ag polysaccharide; the parallel gel that was stained to check protein contamination revealed no bands (20). Protein content checked by Bradford assay (Bio-Rad) was below detection limit.
Human epithelial kidney (HEK) 293 cells stably transfected with human TLR4, MD-2 and CD14 (HEK/TLR4) or with human TLR2 (HEK/TLR2) were purchased from InvivoGen. The HEK/TLR clones were grown in DMEM (Life Technologies), supplemented with heat-inactivated 10% LPS-screened FCS (Limulus amoebocyte lysate < 1 ng/ml; HyClone Laboratories) supplemented with 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 2 mM l-glutamine (all from HyClone Laboratories), and Normocin (100 μg/ml; InvivoGen). HEK/TLR4 culture medium was supplemented with 0.2 mM blasticidin (InvivoGen) and HygroGold (50 μg/ml; InvivoGen). HEK/TLR2 culture medium was supplemented with 0.6 mM G418 sulfate (InvivoGen).
Stable transfection with NF-κB-inducible reporter plasmid and TLR signaling assay
HEK/TLR cells were transfected with a plasmid encoding secreted alkaline phosphatase (SEAP, pNifty2-SEAP; InvivoGen) as previously described (18). HEK/TLR/pNifty2-SEAP cells were either untreated or treated for 16 h with B. pertussis LOS, B. parapertussis LPS, ultrapure E. coli LPS, or Pam2CSK4 at a dose range from 0.01 to 1 μg/ml. Supernatants were then incubated with QUANTI-Blue (InvivoGen), and SEAP activity was measured by reading OD at 655 nm with a 3550 UV Microplate Reader (Bio-Rad).
Purification and culture of MDDC
CD14 monocytes were purified and cultured as previously described (19), At day 7, MDDC were analyzed for CD1a and CD14 expression by use of a FACScan flow cytometer (BD Biosciences). MDDC were further cultured for 48 h in the presence of 1 μg/ml B. pertussis LOS, B. parapertussis LPS, or E. coli LPS to induce maturation. When indicated, MDDC were pretreated, using a predetermined optimal dose, with neutralizing anti-CD14 mAb (10 μg/ml), or ERK1/2 inhibitor PD98059 (100 μM) or p38 inhibitor SB203580 (20 μM) for 1 h and further stimulated.
Polarization of T lymphocytes
To evaluate the potential of MDDC to polarize T lymphocytes, experiments were performed using allogeneic T cells purified from PBMC by negative sorting with magnetic beads (pan-T cell kit; Miltenyi Biotec) (21). MDDC (0.5 × 105/ml) and T cells (0.5 × 106/ml) were cultured in complete medium in 24-well plates (Corning). On day 5, IL-2 (50 U/ml) was added. On day 12, supernatants were harvested for cytokine measurement.
MDDC were incubated with specific fluorochrome-conjugated mAbs for immunophenotypic analysis. Isotype-matched Abs (BD Biosciences) were used as negative control. Cells were analyzed with a FACScan (BD Biosciences). Fluorescence data are reported as a percentage of positive cells when treatment induces the expression of the marker in cells that were negative; median fluorescence intensity was used when treatment increased the expression of the marker in cells that were already positive.
Cytokine measurement by ELISA
MDDC culture supernatants were collected, and cytokines assayed by ELISA specific for IL-1β, IL-6, IL-10, IL-12p40, IL-12p70 (Quantikine; R&D Systems), and IL-23 (Bender MedSystems). The lower detection limits were 1.0 pg/ml for IL-1β, 0.7 pg/ml for IL-6, 3.9 pg/ml for IL-10, 15.0 pg/ml for IL-12p40, 5.0 pg/ml for IL-12p70, and 20.0 pg/ml for IL-23. OD was read at 450 nm with a 3550 UV Microplate Reader (Bio-Rad).
Cytokines in the supernatants from polarized T cells were assayed by ELISA specific for IFN-γ, IL-5, and IL-17 (Quantikine; R&D Systems). The lower detection limits were 8.0 pg/ml for IFN-γ, 3.0 pg/ml for IL-5, and 15.0 pg/ml for IL-17.
IP-10 intracellular staining
MDDC were incubated with different stimuli. After 1 h, brefeldin A, a compound that blocks proteins in the endoplasmic reticulum, was added (10 μg/ml) and cells further cultured for 5 h. MDDC were then fixed and permeabilized using Cytofix/Cytoperm and Perm/Wash protocols (BD Biosciences) and then stained with pretitrated anti-IP-10 mAb or an appropriate isotype control, followed by incubation with FITC-conjugated goat anti-mouse Ig (DakoCytomation). Cells were analyzed by FACScan (BD Biosciences).
mRNA cytokine expression by TaqMan real-time RT-PCR analysis
To measure cytokine mRNA expression, TaqMan RT-PCR analysis was used (Applied Biosystems). Total RNA was extracted from MDDC at different time points and reverse transcription was conducted as previously described (21, 22). PCR was performed, amplifying the target cDNA transcripts and the β-actin cDNA as endogenous control. Specific primers and probes were obtained from Applied Biosystems. mRNA transcript levels were expressed as fold increase compared with basal condition.
Western blot analysis
MDDC were starved by culturing overnight in culture medium supplemented with 1% LPS-screened FCS. Cells were then stimulated and lysed at the indicated time points as previously described (23). Proteins were separated by 12% SDS-PAGE and transferred onto a nitrocellulose membrane (GE-Healthcare) and immunoreactive proteins were detected by incubating blots (23). Specific phosphorylation levels were measured by Image Station 2000R (Kodak), and data were expressed in pixel intensity arbitrary units.
All data were recorded in a computerized database. Results are reported as mean ± SEM. Statistical analyses were conducted using the SPSS 13.0 software. Differences between mean values were assessed by Student’s t test. The statistical significance was set at p < 0.05.
B. parapertussis and B. pertussis LPS molecules differentially trigger TLR4 receptorial complex
To identify the ability of B. parapertussis LPS and B. pertussis LOS to engage TLR4 receptorial complex, we used HEK 293 cells expressing TLR4, MD-2 and CD14, and stably transfected with pNifty2-SEAP, a reporter gene under the control of a NF-κB-dependent promoter (18). Penta-acylated B. parapertussis LPS and B. pertussis LOS triggered TLR4/MD-2/CD14-dependent signaling at each dose tested. B. parapertussis LPS triggered signaling more efficiently than B. pertussis LOS, particularly at lower doses (Fig. 1). However, Bordetella LPS induced lower reporter gene expression compared with hexa-acylated E. coli LPS. To evaluate whether LPS from Bordetella may activate TLR2, as described for other LPS molecules (24), a similar experiment was performed using pNifty2-SEAP transfected HEK/TLR2 cells (18). Both types of LPS were unable to trigger TLR2-dependent signal transduction at any dose tested, whereas Pam2CSK4, a synthetic TLR2 ligand, efficiently induced the expression of the reporter gene (data not shown).
Overall this set of experiments documented that TLR4 receptorial complex is triggered at different intensity by structurally different LPS and underlined a hierarchy that correlated with the degree of lipid A acylation and the presence of a distal O-chain, E. coli LPS being a stronger stimulus compared with B. parapertussis LPS, which in turn proved more potent than B. pertussis LOS.
B. parapertussis and B. pertussis LPS molecules differentially promote MDDC maturation and function
The acquisition of a mature phenotype by MDDC in response to B. parapertussis LPS and B. pertussis LOS was studied. As shown in Fig. 2, upon stimulation, immature cells shifted to a mature phenotype, witnessed by enhanced expression of maturation (CD83, CD38) and costimulatory (CD80, CD86) molecules. The degree of phenotypic maturation was dependent on the stimulus used, showing the same hierarchy observed upon stimulation of TLR4 receptorial complex. Indeed, maximal maturation was induced by E. coli LPS, whereas B. parapertussis LPS was a more efficient inducer compared with B. pertussis LOS, achieving statistical significance relative to all markers studied.
B. parapertussis LPS- and B. pertussis LOS-induced cytokine production by MDDC was then analyzed (Fig. 3,A). Both Bordetella LPS were inducers of a set of immunoregulatory cytokines including IL-10, IL-12p40, IL-23, IL-6, and IL-1β. E. coli LPS was the only stimulus able to drive IL-12p70 secretion. In fact, neither B. pertussis LOS, as already described (18), nor B. parapertussis LPS were inducers of this cytokine. In addition, E. coli LPS was a more potent cytokine inducer than B. parapertussis LPS, which in turn promoted statistically significant higher levels of cytokine production compared with B. pertussis LOS, with the notable exception of IL-1β (Fig. 3 A).
Fig. 3,B showed gene expression analysis of EBV-induced gene (EBI)3 and p28 as subunits of heterodimeric IL-27, which is a cytokine belonging to the IL-12 family endowed with important immunoregulatory functions and implicated in the down-regulation of Th17 cells (25, 26). Data demonstrated a differential behavior of Bordetella LPS because B. parapertussis LPS induced EBI3 and p28 subunit transcription of IL-27, whereas B. pertussis LOS was able to induce only the EBI3 subunit (Fig. 3,B). E. coli LPS induced the expression of both IL-27 subunits (Fig. 3 B).
B. parapertussis and B. pertussis LPS molecules differentially trigger CD14-mediated signaling
It has been proposed that CD14 contributes to TLR4 signaling via binding to the LPS distal O-chain (27). Although MDDC are CD14-negative, soluble CD14 present in the serum added to the cell culture medium may take part into LPS signaling in this ex vivo model (28). To evaluate the involvement of CD14 in B. pertussis LOS- and B. parapertussis LPS-induced signaling, MDDC were stimulated in the presence of a neutralizing anti-CD14 mAb to prevent the interactions of LPS with soluble CD14. CD80, CD83, and IL-10 induction were then measured as a readout for MDDC maturation and activation. The results obtained showed that the activity of E. coli LPS and B. parapertussis LPS rely on soluble CD14 binding. Indeed, the addition of the anti-CD14 mAb significantly reduced the expression of CD80, CD83, and IL-10 release (Fig. 4). Remarkably, signals delivered by B. pertussis LOS were shown to be CD14-independent.
B. parapertussis and B. pertussis LPS molecules induce signaling pathways involved in MDDC functions
The intracellular signal transduction events triggered by LPS from Bordetella in MDDC were then investigated. It is known that exposure to E. coli LPS results in DC activation through the receptor complex TLR4/MD-2/CD14 and subsequent recruitment of Toll/IL-1R domain-containing adaptor molecules to the cytoplasmic domain of the receptor. These adaptors include MyD88 (29). Recruitment of the MyD88 adaptor initiates a MyD88-dependent pathway that culminates in the early activation of NF-κB, which plays an important role in DC maturation and function (30). In parallel, a MyD88-independent pathway results in a late-phase activation of NF-κB (29).
To evaluate the Bordetella LPS signaling, we first evaluated MyD88-dependent pathway by the analysis of NF-κB and the MAPK pathways induction. To follow the NF-κB activation, we focused on IκBα phosphorylation because this process is necessary to induce nuclear translocation of NF-κB. LPS from Bordetella promoted IκBα phosphorylation, notwithstanding E. coli LPS being the stronger inducer of IκBα-phosphorylated protein especially at later time points (Fig. 5 A).
Concerning MAPK family members, we measured p38 and p44/p42 (ERK1/2), which have been described as master regulators of DC maturation and inducers of cytokine transcription (31). B. pertussis LOS and B. parapertussis LPS both induced phosphorylation of p38 and ERK1/2. However, B. pertussis LOS showed ability to induce a more pronounced ERK1/2 phosphorylation, especially at early time points (Fig. 5 A).
This specific induction of signaling pathways was confirmed using inhibitors of ERK1/2 (PD98059) and p38 (SB203580), measuring CD80, CD83, and IL-10 induction as readout of cell maturation and activation, respectively (Fig. 5 B). When cells were treated with B. pertussis LOS, CD80 expression and IL-10 production were significantly reduced by both inhibitors, and remarkably, inhibition of ERK1/2 was significantly more potent compared with p38 inhibition. When B. parapertussis LPS and E. coli LPS were used, CD80 and IL-10 were significantly reduced by both inhibitors, with no statistically significant difference. CD83 expression was inhibited only by p38 inhibitor independently of the stimulus used. Overall these data confirmed a predominant role for ERK1/2-mediated signaling when MDDC are stimulated with B. pertussis LOS, whereas B. parapertussis LPS appears to act through the activation of both p38 and ERK1/2.
B. parapertussis LPS activates TLR4-dependent MyD88-independent gene induction
MyD88-independent signaling is responsible for the activation of intracellular pathways that lead to the induction of IFN-β and IFN-inducible genes (29, 32). To study this aspect in more depth, we analyzed by Western blot the phosphorylation of the transcription factor STAT1, pivotal in the regulation of this pathway (29). Fig. 6,A, shows that B. pertussis LOS induced only a weak STAT1 phosphorylation, with a modest increase with respect to untreated cells, whereas B. parapertussis LPS promoted substantial levels of phosphorylated protein. E. coli LPS stimulation induced high levels of phosphorylated STAT1, confirming other studies in human MDDC (32). The expression of the IFN-inducible gene IFN regulatory factor (IRF)-1 was then studied. As shown in Fig. 6,B, both B. parapertussis LPS and E. coli LPS stimulation induced high levels of IRF-1, and in contrast, B. pertussis LOS was unable to promote consistent mRNA transcription. IP-10 expression, which is a marker of MyD88-independent pathway (33), was then assessed (Fig. 6 C). B. parapertussis LPS increased the percentage of MDDC positive for IP-10 staining, reaching statistical significance in comparison to B. pertussis LOS and to untreated cells. B. pertussis LOS did not enhance the number of IP-10-positive cells with respect to untreated MDDC. E. coli LPS promoted high levels of IP-10-expressing cells.
Overall these data indicate that B. parapertussis LPS, differently from B. pertussis LOS, activates the MyD88-independent pathway in MDDC, although not to its full potential, as witnessed by intermediate phospho-STAT1 and IP-10 level induction.
Bordetella LPS-matured MDDC drive the expansion of polarized Th2/Th17 effectors
In a previous study we demonstrated that B. pertussis LOS-matured MDDC skewed Th cell polarization toward a Th2 phenotype (18). In this study, we have shown that both B. pertussis LOS and B. parapertussis LPS induce in MDDC IL-23, IL-6, and IL-1β (but not IL-12p70), a cytokine profile compatible with the induction of Th17 lymphocytes (34, 35, 36). To verify this point, polarization experiments were performed by coculturing purified allogeneic T lymphocytes with MDDC stimulated with the different LPS molecules (Fig. 7). Both B. pertussis LOS- and B. parapertussis LPS-treated MDDC induced statistically significant higher levels of IL-5 and IL-17 as compared with untreated and E. coli LPS-treated MDDC. However, IL-17 was induced to significantly greater amounts by B. pertussis LOS as compared with B. parapertussis LPS, suggesting a predominant Th17 polarization.
Noteworthy, MDDC maturated by B. parapertussis LPS were more prone to drive expansion of IFN-γ-producing Th1 cells, as shown by the IFN-γ levels measured in coculture supernatants, which were significantly higher compared with IFN-γ cells induced by untreated MDDC and intermediate compared with B. pertussis LOS- or E. coli LPS-treated MDDC. E. coli LPS stimulation drove the expansion of Th1 effectors producing high levels of IFN-γ and low levels of IL-5 and IL-17, confirming results reported in our study and other studies (18, 36).
Several earlier studies indicated that B. pertussis infection promotes a Th1 immune response, based largely on IFN-γ production (37, 38, 39). We have previously demonstrated that incubation of human MDDC with B. pertussis induced a Th1 predominant polarization and expression of IL-23, involved in Th17 cell expansion, in the absence of the Th1-inducing cytokine IL-12p70 (21, 22). Furthermore, evidence is accumulating to show that Bordetella infection might bias the host response toward a Th17 response (40, 41). In this study, novel data support this view.
Bordetella LPS triggered TLR4 signaling, but their signaling potency was reduced in comparison to E. coli LPS because we found that E. coli LPS, bearing a hexa-acylated lipid A, is a stronger activator of the TLR4 receptorial complex and inducer of DC responses compared with penta-acylated lipid A present in B. parapertussis LPS and B. pertussis LOS.
However, the number of acylated lipid A chains is not the only factor influencing the amplitude of LPS-induced response because B. parapertussis LPS was significantly more potent than B. pertussis LOS in determining the intensity of TLR4 receptor signaling. It was recently reported that the presence of a distal O-chain is required for LPS binding to CD14, and that the intensity of LPS activation parallels the presence of the polysaccharide portion and consequently triggers the induction of MyD88-independent signaling and maximal TLR4 response (12, 27). Because B. pertussis LOS lacks an O-side chain and B. parapertussis LPS has an O-Ag structure, to understand the structural basis for the differences observed, we inhibited the interactions that may have taken place with soluble CD14 molecules present in cell culture medium. The results were consistent with our study hypothesis; in fact, E. coli LPS and B. parapertussis LPS activities were dramatically inhibited by the presence of the neutralizing anti-CD14 mAb, whereas B. pertussis LOS stimulation was unaltered. Differences in CD14 usage by Bordetella LPS may explain the apparent contrast between our findings and the results published by Mann and colleagues (42), who have reported that B. pertussis LOS is a more efficacious stimulator of TLR4-transfected HEK 293 cells compared with B. parapertussis LPS. However, the cell line used by Mann et al. (42) did not express CD14, thus nullifying the contribution of B. parapertussis LPS distal O-chain to TLR4 signaling.
Differences in structure of B. pertussis LOS and B. parapertussis LPS are at the basis of quantitative and qualitative differences observed in intracellular pathways activated by Bordetella LPS. B. pertussis LOS triggered ERK1/2 phosphorylation more rapidly and at higher levels than B. parapertussis LPS, and a preferential role for ERK1/2 in B. pertussis LOS-mediated MDDC activation was confirmed by using specific MAPK inhibitors. Furthermore, B. pertussis LOS was unable to induce MyD88-independent gene induction, which was instead activated by B. parapertussis LPS, although at low levels. This pathway leads to the induction of IFN-β, which in turn, activates STAT1 phosphorylation, leading to the induction of several IFN-inducible genes, including IP-10 (32, 33). TLR4-induced type I IFN was shown to be involved in the expression of the p35 subunit of IL-12p70, probably through the induction of IRF family members such as IRF-1 and IRF-8 (32, 43). In this study, we showed that B. parapertussis LPS, although unable to induce IL-12p35 as verified by quantitative RT-PCR (data not shown) and inferred by the detection of IL-12p40 but not IL-12p70 dimeric protein, promoted STAT1 phosphorylation and the expression of IRF-1 and IP-10. These findings provide evidence in support of the activation of the TLR4-MyD88-independent pathway (29, 32, 33).
Differences in intracellular pathways triggered by Bordetella LPS in MDDC reflect the diverse ability of B. pertussis LOS and B. parapertussis LPS to induce MDDC activation. B. parapertussis LPS was a significantly more potent activator compared with B. pertussis LOS in all MDDC functions examined in this study, including up-regulation of maturation markers and expression of IL-10, IL-12p40, IL-23, and IL-6 cytokines. The notable exception was IL-1β, recently described as a pivotal cytokine in Th17 polarization (36), equally induced by B. pertussis LOS and B. parapertussis LPS. Another principal difference between B. pertussis LOS- and B. parapertussis LPS-induced MDDC activation was the ability of B. parapertussis LPS to promote the transcription of the IL-27 subunits, EBI3 and p28. The capacity of B. parapertussis LPS to induce IL-27p28 expression again supports its ability to activate the TLR4-MyD88-independent pathway, as reported by recent studies showing the involvement of the MyD88-independent pathway and IRF-1 in up-regulation IL-27p28 subunit (44, 45).
The different ability shown by B. pertussis LOS and B. parapertussis LPS to activate intracellular pathways and cytokine induction in MDDC might determine differences in the regulation of Th cell responses during infection by Bordetella. The induction and expansion of Th17 cells in humans has been described recently to be dependent on the production of IL-1β, IL-6, and IL-23, whereas IL-12p70 possesses an inhibitory activity. Furthermore, it was demonstrated that IL-1β is fundamental in Th17 differentiation, whereas IL-6 and IL-23 enhance Th17 cell expansion (36). MDDC treated with LPS from Bordetella produced high levels of IL-6, IL-1β, and IL-23 in the absence of IL-12p70. Of interest, B. pertussis LOS induced a relatively higher amount of IL-1β, giving support for a possible major involvement in Th17 differentiation. Indeed, we found that purified T cells cocultured with B. pertussis LOS-driven or B. parapertussis LPS-driven MDDC expressed IL-17. Interestingly, the capacity of B. parapertussis LPS to induce IL-27 expression in MDDC may be linked to the reduced IL-17 and enhanced IFN-γ levels measured in T cell B. parapertussis LPS-MDDC cocultures as compared with T cell B. pertussis LOS-MDDC cocultures. IL-27, initially characterized as a proinflammatory cytokine with Th1-inducing activity (25) was subsequently attributed important immunoregulatory functions in vivo, specifically down-regulating Th17 cells (26, 46, 47).
B. pertussis LOS- or B. parapertussis LPS-matured MDDC drove the expansion of IL-5-producing Th2 cells, as previously demonstrated for B. pertussis LOS (18). It is conceivable that Th2 polarization occurs by default, due to absence of IL-12p70 and production of IL-10. However, it is not possible to rule out that a specific Th2-polarizing signal is induced by Bordetella LPS, and further studies are needed to identify which surface molecules or cytokines are involved.
Overall these results suggest that B. pertussis infection biases the host immune system toward a Th17 response through the action exerted by B. pertussis LOS, in accordance with the recent hypothesis that Th17 may be one major cause of cough pathology in pertussis (41). These data are also in agreement with that obtained in mice, in which vaccination with a whole cell pertussis vaccine, known to contain LOS in high concentration, promotes the expansion of IL-17-producing T cells dependent on TLR4 signaling, as well as expansion of Th1 cells (48). It has also been shown that B. bronchiseptica induces a strong Th17 immune response both in vitro and in vivo (40). IL-17-producing T cells may determine in patients a sustained inflammation of the airways causing cough, as reported for other respiratory infections (49). In contrast, B. parapertussis LPS is a stronger activator of innate immune response compared with B. pertussis LOS in terms of DC maturation and cytokine production, and mediates a less robust Th17 polarization with enhanced levels of Th1 cells.
In conclusion, we provide in this study evidences that the different immunomodulatory properties are related to the diverse structures of Bordetella LPS molecules and may contribute to explaining differences in the natural history of diseases caused by B. pertussis and B. parapertussis; however, these data should be confirmed in infected patients. Furthermore, differences in LPS structures might not be the only factors involved in the differential balance of Th cell polarization in Bordetella infections; other virulence factors may indeed be relevant in vivo.
The editorial assistance of Adam Nixon is gratefully acknowledged. We thank Antonio Cassone for constructive discussions, support, and critical reading of the manuscript and to Alison Weiss for critical reading of the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Grants 5303 and 28C6 from Istituto Superiore di Sanità-National Institutes of Health (USA) Scientific Cooperation Agreement, by Grant 6ACF/6 from the Italian Ministry of Health, Istituto Superiore di Sanità, Sixth AIDS Project and l’Agenzia Italiana del Farmaco Project (to C.M.A.), and Contract LSHP-CT-2003-503240 from the Commission of the European Communities, Sixth Framework Program, “Mucosal Vaccines for Poverty-Related Diseases.”
Abbreviations used in this paper: LOS, lipooligosaccharide; IP, IFN-inducible protein; IRF, IFN regulatory factor; HEK, human epithelial kidney; EBI, EBV-induced gene; DC, dendritic cell; MDDC, monocyte-derived DC; SEAP, secreted alkaline phosphatase.