Introduction Bordetella pertussis is a Gram-negative bacterium that causes whooping cough (pertussis), a severe respiratory tract infection that kills almost 200,000 children annually worldwide. Whole cell vaccines (Pw) introduced in the 1950s significantly reduced the incidence of pertussis but were associated with side effects and were replaced by safer acellular pertussis vaccines (Pa) in most developed countries following successful clinical trials in the 1990s [1]–[3]. However the incidence of pertussis is increasing, especially in adolescents and adults [4], [5] and this may be related to suboptimal or waning immunity induced by Pa [6]. Despite recent progress, the mechanism of protective immunity induced by pertussis vaccines remains unclear. Analysis of serological responses in immunized children revealed a correlation between antibody response to the B. pertussis antigens, pertactin, pertussis toxin (PT) or fimbrae and Pa-induced protection [7]. Analysis of T cell responses in children demonstrated that Pa promote Th2-type responses, whereas Pw preferentially induce Th1 cells [8], [9]. Studies in mouse models have suggested that Th1 cells play a critical role in immunity induced by Pw or previous infection, whereas Th2 cells and antibody confer protection induced by Pa [10]–[13]. However it has also been reported that the superior long term protection induced by Pw in mice, when antibody responses had waned significantly, was associated with the induction of potent Th1 responses [14]. More recently it has been reported that Th17 cells also play a role in protection induced by natural infection or immunization with Pw [15]–[18], but their role in Pa-induced immunity has not been examined. Like most other licensed infectious disease vaccines, Pa are delivered to children using alum as the adjuvant. Traditionally it had been accepted that alum enhances immune responses to the antigens in a vaccine by facilitating retention of the antigen at the site of injection, thus promoting antibody responses and antigen uptake by antigen presenting cells for priming of T cell responses in the draining lymph nodes [19]. It also emerged that alum preferentially promoted Th2 cells, which are considered to be important for protection against parasites and extracellular bacteria by providing help for antibody production. More recently, it was demonstrated that alum functions as an adjuvant in mice by activating the Nlrp3 inflammasome [20], [21], involved in processing of IL-1β. It has also been reported that activation of caspase-1 and Nlrp3, although required for IL-1β production, were dispensable for alum-mediated Th2-associated antibody production [22]. However, the role of Th17 cells has not been addressed. We and others have shown that caspase-1-processed IL-1β plays a crucial role in the induction of Th17 cells that mediate autoimmunity [23]–[25]. Th17 cells are also required for protective immunity against infection, primarily fungi and extracellular bacteria, such as Klebsiella pneumonia, where IL-17 promotes recruitment of neutrophils [26]. The aim of this study was to examine the relative contribution of Th1 and Th17 cells in host immunity to B. pertussis, both in the clearance of a primary infection in naive mice and in response to vaccination and to use this information to help in the rational design of a more effective Pa. Our findings demonstrate both Th1 and Th17 cells contribute to clearance of a primary infection of mice with B. pertussis, and that IFN-γ has a critical role in adaptive immunity to B. pertussis induced by Pw. In contrast, an alum-adjuvanted Pa induced Th17 as well as Th2-type responses, but surprisingly we found that IL-17A played an essential role, while IL-4 was unnecessary for bacterial clearance. The induction of Th17 responses by Pa required activation of IL-1R-signalling in innate immune cells and protection was associated with cellular recruitment to the lungs after challenge with B. pertussis and activation of bacterial killing by neutrophils. Furthermore, the protective efficacy of experimental Pa could be enhanced to that of Pw by substituting alum with an adjuvant that induces Th1 cells. Results Th17 and Th1 cells mediate natural immunity to B. pertussis Previous infection with B. pertussis is effective in inducing protective immunity against subsequent infection and this has been associated with the induction of Th1 cells [13], [27]. Indeed, it has already been established that IFN-γ plays a critical role in clearance of a primary infection with B. pertussis [10], [11]. However there is also evidence that Th17 cells may be involved [16], [18]. Here we examined the relative role of T cell subtypes in host immunity to a primary infection with B. pertussis in naive mice and first concentrated on defining the role of IL-17. We found that infection of mice with B. pertussis was associated with induction of B. pertussis-specific Th17 cells. Antigen-specific IL-17A (Figure 1A) and IL-17F (Figure S1A) production was detected in lungs as early as 7 days post challenge and reached a peak after 3–4 weeks. Interestingly, B. pertussis filamentous hemagglutinin (FHA), which is considered to be the least important antigen in Pa from the perspective of antibody responses [7], was a major target for Th17 cells from infected mice (Figure S1B). In order to confirm these findings and to examine the cellular source of IL-17, we performed intracellular cytokine staining (ICS) and flow cytometry analysis on lung mononuclear cells ex vivo, without re-stimulation. We found significant increase in the frequency (Figure 1B, C) and absolute numbers (Figure S2) of IL-17A-producing CD4 T cells in the lung throughout the course of infection with B. pertussis. The earlier peak of IL-17A+CD4+ T cells (day 14) compared with antigen-specific IL-17A detected by ELISA (day 21), probably reflect the difference in the assay system, with the latter involving a re-stimulation in vitro and therefore including memory cells, while the ICS was a more direct ex vivo measure of activated effector Th17 cells. Taken together these data show that B. pertussis infection of mice induces significant numbers of B. pertussis-specific Th17 cells in the lungs. 10.1371/journal.ppat.1003264.g001 Figure 1 Th17 and Th1 cells mediate host immunity to B. pertussis in the respiratory tract of naive mice. (A–C) Naive C57BL/6 mice were exposed to an aerosol infection with B. pertussis and groups of 4 mice were sacrificed at the indicated time points. (A) Lung mononuclear cells were stimulated with heat-killed B. pertussis and after 3 days of culture IL-17A was quantified in supernatants by ELISA. (B–C) lung mononuclear cells were incubated with brefeldin-A for 1 h and intracellular cytokine staining for IL-17A, together with surface staining for CD4 was performed, followed by FACS analysis. Results are expressed as mean frequencies of IL-17A+CD4+ cells (B), with sample FACS plots (C) (D–E) C57BL/6 WT and IL-17A−/− mice were aerosol challenged with B. pertussis and groups of 4 mice were sacrificed at the indicated time points. CFU counts were performed on lung homogenates (D) ** p<0.01 IL-17A−/− versus WT. Neutrophil recruitment was determined by FACS analysis on lung lavage (E). (F) Spleen cells from IFN-γ−/− or WT mice that had cleared a respiratory infection with B. pertussis were stimulated in vitro with killed B. pertussis and IL-12 (Th1) or IL-1β and IL-23 (Th17) respectively. After 4 days of culture surviving cells were harvested and B. pertussis-specific Th1, Th17 or both (10×106) were transferred to naive mice, which were aerosol challenged with live B. pertussis 24 hours later. Naive mice that did not receive a cell transfer and mice injected with T cells from a naive mouse were used as controls. The course of infection was followed by performing CFU counts on the lungs at intervals after challenge. +p<0.05, +++ p<0.001 Th1+Th17 versus control; ** p<0.01, *** p<0.001 Th17 versus control. Results (except panel C) are mean values for 4 mice per group at each time point and each panel is representative of either 3 to 4 independent experiments. In order to examine the role of IL-17A in bacterial clearance, we compared the course of infection in IL-17A-defective (IL-17A−/−) and WT mice. IL-17A−/− mice had 100–1000 fold more CFU in the lungs at the later stages of infection with bacteria still detectable in the lungs up to week 6 (Figure 1D). The more severe infection in IL-17A−/− mice was associated with a significant reduction in CXCL1 (KC) production (Figure S3) and impaired neutrophil recruitment (Figure 1E) to the lungs post challenge. We used an adoptive cell transfer approach to examine the relative role of Th1 and Th17 cells in protective immunity to B. pertussis. We generated polarized B. pertussis-specific Th1 or Th17 cells (Figure S4) by culture of spleen cells from convalescent WT or IFN-γ−/− (to overcome the problems of reversion of Th17 cells to Th1) with antigen and IL-12 or IL-1β and IL-23 respectively. Transfer of either Th1 or Th17 cells alone before B. pertussis challenge reduced the CFU counts by about 10 fold over the course of infection (Figure 1F). Transfer of both populations together had a greater effect with CFU count significantly reduced by 50–100 fold compared to controls. In contrast transfer of naïve T cells from WT mice failed to confer protection to infected mice. These findings demonstrate that both Th1 and Th17 cells contribute to natural immunity induced by infection with B. pertussis in mice. Protective immunity induced by Pw is mediated largely by IFN-γ production Pw are more protective than Pa in mice [12], [13], [28], [29], which is even more pronounced when mice are challenged at an extended interval after immunization [14] and this has been attributed to the induction of Th1 cells by Pw [13]. Although a Connaught laboratories Pw only had an efficacy of 36 or 48% compared with 84 or 85% for 3 and 5-component Pa in the pertussis clinical trials carried out in Sweden and Italy in the 1990s [1], [2], most good Pw have efficacy of 93–96% in children [3], [30], [31] and a UK Pw was significantly more protective than the three-component Pa in a randomized controlled trial [32]. Here we examined the relative roles of IFN-γ and IL-17 in clearance of B. pertussis from the respiratory tract of mice immunized with Pw. We used a plain (without alum) Pw reference preparation. Although most recent Pw are absorbed to alum, plain Pw, such as the one manufactured by Wellcome laboratories, were routinely used until the 1980s in many European countries and had high efficacy against pertussis [30], [33]. Furthermore, we have found that plain Pw induce similar immune responses and protection against infection as alum-absorbed Pw [12] [and Mills, unpublished]. Here we found that protective immunity induced by Pw was significantly compromised in IFN-γ−/− mice, with 100–1000 fold more bacteria in the lungs compared with Pw-immunized WT mice 3, 7 and 10 days after aerosol challenge (Figure 2A). The CFU counts were also significantly higher in Pw-immunized IL-17A−/− compared with WT mice 3 days post B. pertussis aerosol challenge, but IL-17A−/− mice, like WT mice had cleared the infection by day 7. 10.1371/journal.ppat.1003264.g002 Figure 2 Protective immunity induced with Pw is mediated by IFN-γ and IL-17. WT, IFN-γ−/− or IL-17A−/− mice were immunized i.p. twice (0 and 28 days) with Pw. 14 days after the second immunization, mice were challenged by exposure to an aerosol of live B. pertussis. (A) The number of CFU in the lungs were quantified at intervals after challenge. (B) B. pertussis-specific cytokine production by spleen cells on day of challenge. *p<0.05, **p<0.01, ***p<0.001 IFN-γ−/− or IL-17A−/− versus WT. Results are mean values for 4 mice per group at each time point and each panel is representative of 2 independent experiments. Mice immunized with Pw developed strong Th1 responses with high concentrations of IFN-γ produced by spleen cells from Pw immunized WT and IL-17A−/− mice, which was undetectable in IFN-γ−/− mice (Figure 2B). B. pertussis-specific IL-17 was also induced by Pw and this was enhanced in IFN-γ−/− mice. B. pertussis-specific IL-13 was at background concentrations in spleen cells from Pw-immunized WT mice, but was induced at significant concentrations in IFN-γ−/− mice (Figure 2B). These findings demonstrate that Pw induce Th1 and Th17 cells and confer protective immunity in mice via IFN-γ induction, but that IL-17A also contributes, though less significantly. Protective immunity induced by Pa is dependent on IL-17A but not IL-4 or IFN-γ Having shown that protection induced by Pw is mediated largely by Th1 cells, we examined the mechanism of host immunity induced by immunization with a licensed alum-absorbed Pa. Immunization with Pa by either i.p. or i.m. routes conferred protection against B. pertussis infection (Figure S5A). We have previously reported that Pa selectively induced Th2-type responses whereas Pw promoted Th1 responses [12], [13]. Here we found that Pa also induced B. pertussis-specific IL-17A from CD4+ T cells (Figure S5B). We next examined the role of Th17 versus Th1 and Th2 cells in Pa-induced immunity. The bacterial clearance curves were almost identical for Pa-immunized WT and IL-4−/− or IFN-γ−/− mice (Figure 3A). In contrast, the rate of bacterial clearance was dramatically slower in IL-17A−/− mice, with 100 fold more bacteria on day 3 and significant bacteria in the lungs on day 10, when the WT mice had cleared the infection (Figure 3A). Pa still induced Th2 responses in IL-17A−/− mice, with B. pertussis-specific IL-13 similar to that in WT mice (Figure 3B). In contrast, B. pertussis-specific IL-13 production by spleen cells was close to background concentrations in IL-4−/− mice, whereas IL-17 was similar to that seen in Pa-immunized WT mice. FHA-specific IFN-γ was undetectable in Pa immunized mice and the low levels of IFN-γ detected in response to HKBp was not significantly different between WT, IL-17A−/− and IL-4−/− mice (Figure 3B). Collectively these findings demonstrate an essential role for IL-17A, but not for IL-4 or IFN-γ, in protective immunity induced by Pa in mice. 10.1371/journal.ppat.1003264.g003 Figure 3 Protective immunity induced with Pa is dependent on IL-17A but not IL-4. WT, IL-17A−/−, IL-4−/− or IFN-γ−/− mice were immunized i.p. twice (0 and 28 days) with Pa. 14 days after the second immunization, mice were challenged by exposure to an aerosol of live B. pertussis. (A) The number of CFU in the lungs were quantified at intervals after challenge. (B) B. pertussis-specific cytokine production by spleen cells on day of challenge. (C) B. pertussis-specific antibody in serum on the day of challenge. *p<0.05, **p<0.01, ***p<0.001 knockout versus WT. Results are mean values for 4 mice per group at each time point and each panel is representative of 2 independent experiments. An examination of antibody responses revealed that total IgG and IgG1 were significantly reduced in both IL-4−/− and IL-17A−/− mice (Figure 3C). IgG2a (Figure 3C) and IgG2c (data not shown) were significantly higher in IL-4−/− than WT mice, but similar in IL-17A−/− and WT mice. To examine the mechanism of immune protection mediated by IL-17A in the lungs, we investigated phagocytic cell influx upon B. pertussis challenge. There was a significant increase in the recruitment of both neutrophils and macrophages to the lungs after B. pertussis challenge in Pa-immunized mice compared to non-immunized mice, which peaked at day 7 post challenge (Figure 4A). Cellular recruitment to the lungs was similar in Pa-immunized WT and IL-4−/− mice. In contrast, the influx of neutrophils and macrophages was significantly reduced in Pa-immunized IL-17A−/− mice. This was associated with dramatically lower CXCL1 and CCL3 (MIP-1α) concentrations in the lungs of Pa-immunized IL-17A−/− compared with WT or IL-4−/− mice post challenge (Figure 4B). 10.1371/journal.ppat.1003264.g004 Figure 4 Induction of protective Th17 cells is associated with neutrophil recruitment and killing of B. pertussis. (A, B) WT, IL-4−/− and IL-17A−/− mice were immunized i.p. with Pa and challenged with B. pertussis as described in Figure 3. Recruitment of GR1+ neutrophils and F4/80+ macrophages in the lungs (A) and CXCL1 and CCL3 concentrations in lung homogenates (B) following aerosol challenge with live B. pertussis. p<0.05, **p<0.01, ***p<0.001 WT + Pa or IL-4−/− + Pa versus WT + PBS; +p<0.05, ++ p<0.01, +++ p<0.001 WT + Pa or IL-4−/− +Pa versus IL-17A−/− +Pa. (C) Effect of recombinant IL-17A, IL-17F or IFN-γ, in the presence of mouse serum from naive or immune mice (containing B. pertussis antibodies from Pa-immunized mice) on neutrophil-mediated killing of B. pertussis in vitro. *p<0.05, **p<0.01 versus control. Results in A and B are mean values for 4 mice per group at each time point and each panel is representative of 3 independent experiments. Results in C are mean values for triplicate assays and are representative of 3 experiments. We have previously reported that IL-17 can promote macrophage killing of B. pertussis [15]. Here we demonstrate that neutrophils were also capable of killing B. pertussis following opsonisation with normal mouse serum, with killing detected after 1–3 hours, and this was significantly enhanced by IL-17A or IFN-γ but not IL-17F (Figure 4C). Furthermore, killing was further, though not significantly, enhanced following addition of immune serum from Pa-immunized mice (Figure 4C). These findings suggest that Pa-induced IL-17A enhances chemokine production, which recruits macrophages and neutrophils to the lungs soon after challenge with B. pertussis and these cells mediate killing of B. pertussis. Contrary to the perceived wisdom, our study suggests that Th2 cells are unnecessary, and Th17 cells play a critical role in protective immunity induced with Pa. Pa promotes the induction of Th17 cells via IL-1 We examined the mechanism of Th17 cell induction with Pa, in particular the role of Nlrp3 and IL-1. It had previously been reported that alum functions as an adjuvant by activation of the Nlrp3 inflammasome [20], [21], although this has been questioned by others [22]. Furthermore, we had previously shown that IL-1β, induced via caspase-1 and Nlrp3, plays a critical role in IL-17-mediated pathology in autoimmune disease [23], [25], [34]. Here we found that Pa or alum induced significant concentrations of IL-1β from LPS-primed DC and this was significantly reduced following addition of a caspase-1 inhibitor (Figure S6A) or using DC from Nlrp3−/− mice (Figure S6B). Furthermore, significant concentrations of IL-1β were detected in draining lymph nodes 4 hours after injection of Pa (Figure S6C). These finding demonstrate that alum-absorbed Pa promotes IL-1β production by DC in vitro via activation of caspase-1 and Nlrp3. We also examined the role of Nlrp3 in the protective efficacy of Pa in vivo. Bacterial clearance was reduced though only significantly on day 5 post challenge in Pa-immunized Nlrp3−/− compared with WT mice (Figure S7A). Furthermore, IL-17A production determined by ELISA on B. pertussis antigen-stimulated spleen cells (Figure S7B) or by intracellular cytokine staining on CD4+ T cells (Figure S7C) was similar in Pa-immunized Nlrp3−/− and WT mice. Finally IL-1β production in the lungs of B. pertussis infected mice was not significantly different between Nlrp3−/− and WT mice (Figure S7D). In contrast to the rather limited attenuation of anti-B. pertussis immunity in Pa-immunized Nlrp3−/− mice, we found a dramatic reduction in the rate of bacterial clearance in Pa-immunized IL-1RI−/− mice, with 1000 fold more bacteria in the lungs when compared with Pa-immunized WT mice at 3 and 7 days post challenge (Figure 5A). Furthermore, WT mice had completely cleared the bacteria by day 10, where as there were significant numbers of bacteria in the lungs of IL-1RI−/− mice at this time point. These findings demonstrate that IL-1 is critical for protection, and its induction in vitro is dependent on caspase-1 and NLRP3, but in vivo NLRP3 appears to be dispensable, suggesting that NLPR3-independent IL-1 pathways may be involved. 10.1371/journal.ppat.1003264.g005 Figure 5 IL-1RI signalling is required for induction of Th17 responses and Pa-induced protection against B. pertussis. IL-1RI−/− and WT mice were immunized i.p. twice (0 and 28 days) with Pa. 14 days after the second immunization, mice were challenged by exposure to an aerosol of live B. pertussis. (A) The number of CFU in the lungs were quantified at intervals after challenge. (B, C) B. pertussis-specific cytokine production by spleen cells on the day of challenge (B) or B. pertussis-specific cytokine production by lung mononuclear cells 3, 7 and 10 days post challenge (C) was determined by ELISA. (D) B. pertussis-specific antibody in serum on the day of challenge (Co: control; KO: IL-1RI−/−). *
