Inefficient Toll-Like Receptor-4 Stimulation Enables Bordetella parapertussis to Avoid Host Immunity

The recognition of bacterial lipopolysaccharide (LPS) by host Toll-like receptor (TLR)4 is a crucial step in developing protective immunity against several gram negative bacterial pathogens. Bordetella bronchiseptica and B. pertussis stimulate robust TLR4 responses that are required to control the infection, but a close relative, B. parapertussis, poorly stimulates this receptor, and TLR4 deficiency does not affect its course of infection. This led us to hypothesize that inefficient TLR4 stimulation enables B. parapertussis to evade host immunity. In a mouse model of infection, B. parapertussis grew rapidly in the lungs, but no measurable increase in TLR4-mediated cytokine, chemokine, or leukocyte responses were observed over the first few days of infection. Delivery of a TLR4 stimulant in the inoculum resulted in a robust inflammatory response and a 10- to 100-fold reduction of B. parapertussis numbers. As we have previously shown, B. parapertussis grows efficiently during the first week of infection even in animals passively immunized with antibodies. We show that this evasion of antibody-mediated clearance is dependent on the lack of TLR4 stimulation by B. parapertussis as co-inoculation with a TLR4 agonist resulted in 10,000-fold lower B. parapertussis numbers on day 3 in antibody-treated wild type, but not TLR4-deficient, mice. Together, these results indicate that inefficient TLR4 stimulation by B. parapertussis enables it to avoid host immunity and grow to high numbers in the respiratory tract of naïve and immunized hosts.


Introduction
The ability of a pathogen to persist in its host for an extended period of time requires that it first evades rapid control and clearance by the innate immune response. Lipopolysaccharide (LPS), a major component of the outer membrane of gram negative bacteria, stimulates host Toll-like receptor (TLR)4 to initiate the production of pro-inflammatory cytokines and chemokines that recruit and activate leukocytes [1,2], which is important to protection against many bacterial pathogens [3,4,5,6,7,8]. Interestingly, LPS is not an invariant structure among gram negative bacteria. For example, Salmonella deacylates and palmitoylates lipid A in response to the host environment, allowing this bacterium to evade TLR4 responses [9,10]. Yersinia and Pseudomonas species also modulate their LPS structures, resulting in diminished TLR4 responses to infection [11,12,13,14]. These findings have led to the recent realization that bacteria can modulate pathogen associated molecular patterns, such as LPS, to optimize interactions with the host.
Bordetella bronchiseptica, B. pertussis and B. parapertussis are three very closely related species that make up the classical bordetellae. Bordetella bronchiseptica, infects a wide-range of mammals where it chronically colonizes the nasal cavity [15] and is often observed as a commensal [16,17]. Both B. pertussis and B. parapertussis are highly infectious pathogens that cause the acute disease whooping cough in humans [16]. Each of these human-adapted species has independently evolved from a B. bronchiseptica-like progenitor [18,19]. The comparative immunobiology of the bordetellae has shed light on some key differences among these bacteria. For example, the LPS structures of each of these bordetellae differs [20,21,22] which results in a wide-range of TLR4 responses and requirements [3]. The LPS of B. pertussis and B. bronchiseptica are very stimulatory of TLR4 and TLR4 is required for their clearance [3]. In contrast, the LPS of B. parapertussis LPS is much less stimulatory of TLR4 and TLR4-deficiency does not render mice more susceptible to B. parapertussis [3].
The clearance of these Bordetella species by antibodies also differs and appears to relate to their epidemiology [15]. B. bronchiseptica is rapidly cleared, three days post-inoculation, by adoptively transferred antibodies [15]. Previous studies have shown that this rapid antibody-mediated clearance is due to TLR4-dependent leukocyte recruitment [23]. B. bronchiseptica can persist for years within the nasal cavity of its host, where serum antibodies have no effect, and therefore a strong selection to avoid antibody-mediated clearance does not exist [15]. In contrast to B. bronchiseptica, B. pertussis and B. parapertussis avoid rapid antibody-mediated clearance for the first week of infection until a sufficient T-cell response is generated [15,24]. Therefore, while both B. pertussis and B. parapertussis are more closely related to B. bronchiseptica than they are to each other [18], they share the ability to resist rapid antibody-mediated clearance from the lower respiratory tract. The high prevalence of detectable antibodies to B. pertussis and B. parapertussis in human populations, either due to vaccination or previous infection, presents a strong selection for the ability to avoid antibody-mediated clearance, allowing for repeated infection of individuals [15].
While both of these human-adapted species avoid rapid antibodymediated clearance, they do so by distinct mechanisms. B. pertussis avoids rapid antibody-mediated clearance by inhibiting the TLR4dependent recruitment of leukocytes to the lungs via pertussis toxin (Ptx) [25,26]. A strain lacking Ptx (B. pertussisDPTx) is rapidly cleared from the lungs upon adoptive transfer of antibodies [25]. Since B. parapertussis lacks Ptx [27], this bacterium must avoid antibodymediated clearance in a Ptx-independent manner.
Since the rapid antibody-mediated clearance of B. bronchiseptica is dependent on TLR4 [23] and B. parapertussis is a weak stimulator of TLR4 [3], we hypothesized that the inefficient TLR4 stimulation by B. parapertussis allows it to avoid the robust inflammatory response required for rapid antibody-mediated clearance. Using a mouse model of infection, we determined that co-inoculation of B. parapertussis with a TLR4 stimulant led to enhanced pro-inflammatory cytokine production and leukocyte accumulation as well as more efficient control and rapid antibodymediated clearance of the bacteria. These results, observed in wild type but not TLR4-deficient animals, explain several characteristics of this important human pathogen and suggest interventions in the disease process. They also demonstrate how very closely related organisms can change complex structural components such as LPS to modulate stimulation of innate immune receptors to optimize their interactions with the host.

Materials and Methods
Bacterial strains and growth B. parapertussis strain 12822 was isolated from German clinical trials [28] and 12822G is a gentamicin-resistant derivative of 12822 [24]. B. bronchiseptica strain RB50 was originally isolated from a rabbit [29]. Bacteria were maintained on Bordet-Gengou agar (Difco) containing 10% defibrinated sheep blood (Hema Resources) and appropriate antibiotics. Liquid culture bacteria were grown at 37uC overnight on a roller drum to mid-log phase in Stainer-Scholte broth.
Inoculation of mice C57BL/6, C3H/HEOuJ (wild type), and C3H/HEJ (TLR4deficient) mice were obtained from Jackson Laboratories and bred in our Bordetella-free, specific pathogen-free facilities at The Pennsylvania State University. Bacteria grown overnight (to an optical density at 600 nm of approximately 0.3) in liquid culture were diluted in PBS to approximately 10 7 CFU/ml. 50 ml of the inoculum (5610 5 CFU) was pipetted on to the external nares of 4-6 week old mice that had been lightly sedated with 5% isoflurane in oxygen. For co-inoculations with B. parapertussis and B. bronchiseptica, both species were present at 10 7 CFU/ml in the inoculum and mice were inoculated as above (5610 5 CFU of each species in 50 ml). For co-inoculation with heat-killed B. bronchiseptica, bacteria were grown overnight to an optical density of 0.3 and heat-inactivated by incubating in a water bath at 75uC for 30 minutes. Bacteria in the inoculum were plated before and after heat-inactivation to quantify the number of bacteria present and ensure that the incubation killed B. bronchiseptica. Inocula were prepared so that they contained 10 7 CFU/ml of B. parapertussis and 10 9 CFU/ml of heat-killed B. bronchiseptica (5610 5 CFU B. parapertussis and 5610 7 CFU of heat-killed B. bronchiseptica in 50 ml per mouse). For co-inoculations with LPS, inocula were prepared containing 10 7 CFU/ml of B. parapertussis and 10 mg/ml of purified LPS from B. bronchiseptica, B. parapertussis, or E. coli (5610 5 CFU B. parapertussis and 500 ng LPS in 50 ml per mouse). All protocols were reviewed by the university's Institutional Animal Care and Use Committee and all animals were handled in accordance with institutional guidelines.

Adoptive transfer of serum antibodies
To generate convalescent phase (immune) serum, C57BL/6 mice were inoculated with 5610 5 CFU of B. parapertussis and allowed to convalesce for 28 days. By this time, these mice have generated high titers of B. parapertussis-specific antibodies [24]. Blood was then collected from these mice and the serum portion was isolated and stored at 280uC until use. 200 ml of immune serum was delivered by I.P. injection into mice immediately before inoculation. Serum from uninfected mice (naïve serum) was used as a control.

Quantification of bacteria, leukocytes, and cytokines in the lungs
To quantify bacterial numbers, the lungs were excised on day 0, 0.5, 1, 2, 3, 7, or 14 post-inoculation. Lungs were homogenized in 1 ml of PBS. The lung homogenate was then plated onto Bordet-Gengou agar plates at the appropriate dilutions and CFU were counted 4 days later for B. parapertussis and 2 days later for B. bronchiseptica. To quantify leukocytes, mice were infected for 0, 0.5, 1, 2, 3, 7, or 14 days, sacrificed, and bronchoalveolar lavage (BAL) fluid was collected. Red blood cells were lysed by treatment with ammonium chloride as previously described [30]. Leukocytes were counted on a hemocytometer to quantify total numbers of leukocytes in the BAL fluid. Aliquots of cells were stained with FITC-labeled anti-Ly-6G (BD Biosciences Phramingen), and the percentage of Ly-6G positive cells was multiplied by the total number of leukocytes to calculate the number of neutrophils. For the quantification of cytokines and chemokines in the lungs, wild type or TLR4-deficient mice were inoculated with B. parapertussis, B. bronchiseptica, or both species and sacrificed 2 hours or 1 day later. Lungs were homogenized in 1 mL of PBS and samples were run on ELISAs specific for TNFa, KC, MIP-1a, and/or IL-1b according to the manufacturer's protocols (R&D Systems, Minneapolis, Minnesota, USA).
In vitro growth curves of B. parapertussis and B. bronchiseptica and enumeration of co-inoculated samples Both bacteria (RB50 and 12822G) were grown overnight to an optical density of 0.3. They were then diluted in fresh Stainer-Scholte broth to 10 7 CFU/ml. The liquid cultures were then grown on a roller drum at 37uC and aliquots were plated at the indicated times on Bordet-Gengou agar plates with 20 mg/ml of streptomycin or gentamicin. The streptomycin plates, on which both species could have grown, were counted 2 days later, before B. parapertussis colonies became visible. The gentamicin plates, on which only B. parapertussis could grow, were counted 4 days later.

Statistical Analysis
The mean+/2SD (error bars) was determined for CFU, leukocytes, and cytokines. For experiments quantifying bacterial numbers, either three or four mice were used per group. For all other experiments, four mice were used per group. Two-tailed, unpaired Student's T-tests were used to determine statistical significance between groups. All experiments were performed at least twice with similar results and P-values,0.05 were taken to be statistically significant.

Reduction of B. parapertussis numbers correlates with an accumulation of leukocytes in the lungs
B. parapertussis grows rapidly over the first few days postinoculation but does not induce an early recruitment of neutrophils, which are known to be essential to eliminating this pathogen [3,24]. Therefore, we sought to determine if the eventual reduction of B. parapertussis numbers in the lungs correlates with a delayed accumulation of neutrophils. C57BL/6 mice were inoculated with B. parapertussis and sacrificed on days 0, 3, 7, or 14 post-inoculation to quantify the numbers of bacteria in the lungs. B. parapertussis numbers peaked at approximately 4610 6 CFU on day 3, but began to decline by day 7 and were reduced to 6610 3 CFU by day 14 post-inoculation (Fig. 1A). Groups of C57BL/6 mice were also sacrificed on days 0, 3, 7, or 14 postinoculation to quantify the numbers of leukocytes in the BAL fluid. Approximately 6610 4 leukocytes and less than 1610 4 neutrophils were recovered from the BAL fluid of uninfected mice. Leukocyte numbers had slightly, but significantly, increased by day 3 postinoculation (1.6610 5 leukocytes, 1610 5 neutrophils), and peaked on day 7 post-inoculation (6610 5 leukocytes, 2610 5 neutrophils), declining thereafter (Fig. 1B). Together, these data show that the time when B. parapertussis numbers began to decline in murine lungs correlated with peak numbers of neutrophils in the lungs.

B. parapertussis does not induce an early, TLR4-mediated leukocyte response
Although the induction of TLR4 responses is crucial to protection against other Bordetella species, B. parapertussis LPS does not efficiently stimulate these responses [3]. We addressed whether or not this pathogen induces any TLR4-dependent recruitment of leukocytes to the lungs over the course of infection. Wild type (C3H/HEOuJ) and TLR4-deficient (C3H/HEJ) mice were inoculated with B. parapertussis and sacrificed 0, 2 hours, 1, 3, 7, or 14 days later to quantify the numbers of bacteria in the lungs and leukocytes in the BAL fluid. As previously shown, similar bacterial numbers were observed in wild type and TLR4-deficient mice [3] (Fig. 2A). In the lungs of wild type mice, leukocyte numbers were highest on day 7 post-inoculation (,3610 5 cells), and the same was true for TLR4-deficient mice (,6610 5 cells) (Fig. 2B). Fewer than 10 5 neutrophils were found in the lungs of both wild type and TLR4-deficient mice over the first 3 days postinoculation but peaked on day 7 in both wild type (,2610 5 cells)  and TLR4-deficient (,5610 5 cells) mice (Fig. 2C). Interestingly, more leukocytes accumulated in the lungs of TLR4-deficient mice compared to wild type mice. Therefore, TLR4 signaling does not measurably enhance the recruitment of leukocytes or the control of B. parapertussis infection, but may affect anti-inflammatory signals in response to this bacterium.
TLR4-mediated cytokine and chemokine responses are not inhibited by B. parapertussis during infection of mice The lack of a measurable TLR4-mediated accumulation of leukocytes in response to B. parapertussis (Fig. 2) led us to assess whether or not B. parapertussis actively inhibits TLR4-mediated cytokine production. For these experiments, the effects of B. parapertussis on TLR4-mediated responses to a respiratory pathogen that is closely related and a potent stimulator of TLR4, B. bronchiseptica, were examined. Wild type and TLR4deficient mice were inoculated with B. parapertussis, B. bronchiseptica, or both bacteria and sacrificed 2 hours later. B. parapertussis did not induce significant levels of TNF-a, KC, or MIP-1a in wild type or TLR4-deficient mice relative to mock-infected controls ( Fig. 3A-C). Approximately 1000 pg of IL-1b was produced in the lungs of wild type mice in response to B. parapertussis, but this was not significantly different from the amount produced by TLR4deficient mice (Fig. 3D). B. bronchiseptica induced the production of approximately 3000 pg of TNF-a, 3500 pg of KC, 9000 pg of MIP-1a, and 2300 pg of IL-1b in the lungs of wild type mice, but much lower levels in TLR4-deficient mice (250, 125, 200, and 1200 pg respectively) (Fig. 3A-D). Similar to B. bronchiseptica, the lungs of wild type mice that were inoculated with both species contained approximately 3500 pg of TNF-a,4500 pg of KC, 8500 pg of MIP-1a, and 2400 pg of IL-1b, and this production was also dependent on TLR4 (Fig. 3A-D). Together, these data indicate that B. parapertussis does not stimulate TLR4 or inhibit the TLR4-mediated cytokine and chemokine responses to B. bronchiseptica infection.
Co-inoculation with B. bronchiseptica allows for more efficient control of B. parapertussis The robust, TLR4-mediated cytokine and chemokine responses to a co-inoculation with B. parapertussis and B. bronchiseptica led us to examine the effect of the co-inoculation on the accumulation of leukocytes and clearance of these bacteria. Wild type mice were inoculated with B. parapertussis, B. bronchiseptica, or both species and sacrificed 12 hours, 1, 2, or 3 days later to quantify neutrophils in the BAL fluid. Consistent with Figures 1 and 2, the BAL fluid of B. parapertussis-infected mice contained few neutrophils (,10 5 /ml of BAL fluid) over the first three days post-inoculation (Fig. 4A). In contrast, B. bronchiseptica induced the accumulation of approximately 1.5610 6 neutrophils/ml of BAL fluid over the first two days. This early recruitment of neutrophils to the lungs upon B. bronchiseptica infection is dependent on TLR4 [3]. The BAL fluid of co-inoculated mice also contained approximately 1.5610 6 neutrophils/ml for the first two days (Fig. 4A), indicating that B. parapertussis did not measurably inhibit the TLR4-mediated recruitment of neutrophils to the lungs in response to B. bronchiseptica infection.
Groups of C57BL/6 mice were then inoculated with B. parapertussis, B. bronchiseptica, or both bacteria and sacrificed 12 hours, 1, 2, or 3 days later to quantify bacterial numbers in the lungs. B. bronchiseptica numbers were not affected by a coinoculation with B. parapertussis (Fig. 4B). When inoculated alone, B. parapertussis numbers rose over the first three days, peaking at approximately 3610 6 CFU on day 3 post-inoculation. When coinoculated with B. bronchiseptica, however, B. parapertussis numbers began to decline after one day and were reduced to approximately 5610 4 CFU by day 3 post-inoculation, a 99% reduction from numbers of B. parapertussis alone (Fig. 4C). Together, these data indicate that a co-infection with B. bronchiseptica results in increased neutrophil recruitment and more efficient control of B. parapertussis.
To determine if B. parapertussis and B. bronchiseptica directly affected the growth of one another, they were grown together in liquid culture. B. bronchiseptica grew from approximately 10 7 CFU/ ml to 10 11 CFU/ml in 24 hours and its growth was not affected by a co-inoculation with B. parapertussis (data not shown). B. parapertussis, which grows slower than B. bronchiseptica, grew from approximately 10 7 CFU/ml to 10 10 CFU/ml in 24 hours and its growth rate was not affected by a co-inoculation with B. bronchiseptica (data not shown). Thus, B. parapertussis and B. bronchiseptica do not directly affect each other's growth, even when grown to high density in vitro.
Co-inoculation with B. bronchiseptica results in rapid antibody-mediated clearance of B. parapertussis B. bronchiseptica is cleared by antibodies within about three days via a TLR4-dependent mechanism [23]. In contrast, antibodies have no effect on the course of B. parapertussis infection during the first week but eliminate the infection during the second week [15,24] (Fig. 5A), after significant numbers of neutrophils have accumulated in the lungs. Thus, we hypothesized that the lack of an early TLR4-mediated neutrophil recruitment allows B. parapertussis to delay antibody-mediated clearance. To test this, we examined the effect of stimulating TLR4 responses on the rapid antibody-mediated clearance of B. parapertussis by inoculating mice with one species or both species and giving an I.P. injection of naïve serum or convalescent phase (immune) serum. Groups of mice were then sacrificed on day 3 or 7 postinoculation for the enumeration of bacteria in the lungs. While immune serum alone had no measurable effect on the numbers of B. parapertussis [15] (Fig. 5B), immune serum with a coinoculation of B. bronchiseptica rapidly reduced B. parapertussis numbers .99%, to approximately 100 CFU by day 3 and to undetectable levels by day 7 post-inoculation (Fig. 5B). These data indicate that a co-inoculation with B. bronchiseptica results in rapid antibody-mediated clearance of B. parapertussis. The coinoculation did not affect the ability of B. bronchiseptica to colonize the lungs of mice treated with naïve serum, but B. bronchiseptica numbers were approximately 500-fold lower in the lungs of mice treated with immune serum (Fig. 5C). This was likely due to B. parapertussis-induced antibodies being cross reactive with B. bronchiseptica antigens.

Rapid clearance of B. parapertussis upon co-inoculation with B. bronchiseptica is dependent on TLR4
We hypothesized that the protective effects of adding B. bronchiseptica to the B. parapertussis inoculum were due to the robust TLR4-mediated inflammatory response to B. bronchiseptica. To test this, groups of wild type and TLR4-deficient mice were inoculated with B. parapertussis alone or B. parapertussis with heat-killed B. bronchiseptica. Heat-killed B. bronchiseptica was used because live B. bronchiseptica is lethal to TLR4-deficient mice within approximately 3 days [3]. This also allowed us to address whether or not the effect on B. parapertussis numbers required live B. bronchiseptica, or if stimulation of the immune response by heat-inactivated compo- nents was sufficient to reduce bacterial numbers. The cytokine and leukocyte responses were measured 1 day post-inoculation with B. parapertussis alone versus B. parapertussis with heat-killed B. bronchiseptica. Inoculation with B. parapertussis did not induce levels of TNFa and KC in the BAL fluid of wild type or TLR4-deficient mice that were measurably different from mock-infected lungs (Fig. 6A-B). In contrast, co-inoculation with B. parapertussis and heat-killed B. bronchiseptica resulted in high levels of TLR4-dependent TNFa and KC production (approximately 550 and 300 pg respectively). When leukocyte numbers were examined, B. parapertussis alone did not induce significant levels of leukocyte accumulation (,1610 5 leukocytes, ,2610 4 neutrophils) relative to mock-infected mice (Fig. 6C-D). Co-inoculation with heatkilled B. bronchiseptica, however, resulted in the accumulation of 8610 5 leukocytes and 6610 5 neutrophils in the lungs of wild type mice by this time, while the lungs of TLR4-deficient mice contained only 3610 5 leukocytes and 4610 4 neutrophils (Fig. 6C-D). Thus, heat-killed B. bronchiseptica induced robust, TLR4mediated cytokine and leukocyte responses.
To address the effect on bacterial numbers, wild type and TLR4-deficient mice were then inoculated with B. parapertussis alone or B. parapertussis with heat-killed B. bronchiseptica. These mice were also given an I.P. injection of naïve serum or immune serum and sacrificed 3 days later. In wild type mice that were treated with naïve serum, co-inoculation with heat-killed B. bronchiseptica resulted in a 10-fold reduction of B. parapertussis numbers. In wild type mice that were treated with immune serum, the coinoculation resulted in B. parapertussis numbers being reduced to nearly undetectable levels within 3 days (Fig. 7A). In TLR4deficient mice, however, the co-inoculation had no effect on B. parapertussis numbers in either naïve serum treated or immune serum treated mice (Fig. 7A). Wild type and TLR4-deficient mice were then inoculated with B. parapertussis and purified LPS from B. bronchiseptica, E. coli, or B. parapertussis to determine if the addition of TLR4 stimulatory LPS was the key to rapid clearance of B. parapertussis. Co-inoculation with B. bronchiseptica LPS resulted in a 10,000-fold reduction in bacterial numbers in the lungs of wild type mice treated with immune serum, but did not affect bacterial numbers in TLR4-deficient mice (Fig. 7B). Similar results were observed when mice were co-inoculated with LPS from E. coli (Fig. 7C). In contrast, co-inoculation with purified LPS from B. parapertussis, a weak TLR4-stimulant [3], had no effect on B. parapertussis numbers in the lungs of wild type or TLR4-deficient  mice (data not shown). Thus, TLR4 was required for the enhanced clearance of B. parapertussis upon co-inoculation with strong TLR4 stimulants ( Fig. 7B-C). Combined, these data suggest that a lack of TLR4 stimulation enables B. parapertussis to avoid immune clearance and grow to higher numbers within the host.

Discussion
B. parapertussis is able to delay clearance by avoiding the induction of a robust innate immune response. Here, we show that its slow clearance from murine lungs correlates with the accumulation of neutrophils in these lungs, which is delayed in comparison to the neutrophil responses to other closely related bacteria (Mann, Harvill, unpublished data). We predicted that inefficient TLR4 stimulation by B. parapertussis LPS may result in the low level of neutrophil accumulation in response to infection over the first few days and may allow this pathogen to grow rapidly during this time, even in animals given a passive transfer of immune serum. In support of this, co-inoculation with a potent stimulator of TLR4 resulted in enhanced control and rapid antibody-mediated clearance of B. parapertussis from wild type, but not TLR4-deficient mice. This study provides an example of how expressing an LPS that is a poor stimulator of TLR4 can facilitate persistence of a gram negative bacterium within its host.
LPS modulation is often utilized by gram negative bacterial pathogens to optimize interactions with host immunity. For example, Yersinia pestis produces a TLR4-stimulatory LPS at 26uC, but an unstimulatory LPS at 37uC, the body temperature of its typical mammalian host [11,12,13]. Montminy et.al. genetically modified Y. pestis so that it would produce the stimulatory 26uC LPS at all times [31]. While wild type Y. pestis causes sepsis and mortality in a mouse model of infection, the expression of TLR4stimulatory LPS resulted in containment of the infection by the innate immune response and less efficient systemic spread of the infection [31]. Similarly, co-inoculation of B. parapertussis with a TLR4 agonist resulted in an attenuated course of infection (Fig. 4,  Fig. 7). The expression of LPS molecules that poorly stimulate TLR4 appears to hinder the generation of effective immunity against Y. pestis [31] and B. parapertussis, and may be a stealth strategy shared by other gram negative bacteria as well [32,33,34,35].
TLR4 stimulation by LPS results in a branched downstream signaling pathway consisting of a Mal/MyD88 branch and a TRIF/TRAM branch that leads to the production of several different pro-inflammatory cytokines [36]. However, each branch is crucial to the production of a different subset of cytokines. For example, TNF-a and CCL3 are MyD88-dependent cytokines while IFN-b is a TRIF-dependent cytokine [37,38]. Although our in vivo data presented above suggested that B. parapertussis does not induce measurable amounts of TLR4-mediated cytokine production or leukocyte recruitment (Fig.3, Fig. 6), the higher numbers of leukocytes in TLR4-deficient mice suggests that leukocyte responses to B. parapertussis infection may be limited by a TLR4dependent mechanism. B. parapertussis may upregulate the TRIF/ TRAM branch of TLR4 signaling, as this branch appears to play a role in endotoxin tolerance [39]. We have also recently observed that IL-10 dampens the inflammatory response to B. parapertussis in vivo and induces the production of IL-10 in vitro (Wolfe and Hester, unpublished data). Since IL-10 production can be induced by TLR4 stimulation [4], it is reasonable to suggest that the antiinflammatory effect of TLR4 in Figure 2 may be mediated by IL-10.
In contrast to B. bronchiseptica, passively transferred antibodies have no effect on colonization by the human pathogens B. parapertussis and B. pertussis over the first week of infection [15,24,25]. This is likely important to the success of these pathogens, as they are able to re-infect the same host multiple times despite a measurable antibody response [40]. Our lab previously showed that Ptx delays antibody-mediated clearance of B. pertussis by inhibiting the migration of neutrophils to the lungs [25]. Although B. parapertussis does not express Ptx, poor induction of TLR4 signaling appears to be an alternative method for limiting the neutrophil response and delaying antibody-mediated clearance. Limiting and/or inhibiting pro-inflammatory TLR4 stimulation may be crucial to the ability of B. parapertussis to remain endemic in human populations.
In addition to the inefficient stimulation of pro-inflammatory TLR4 responses [3], likely due to its lipid A structure, the Oantigen of B. parapertussis LPS also appears to allow it to avoid rapid clearance by antibodies induce by B. pertussis infection or vaccination [24]. O-antigen prevents the binding of B. pertussisinduced antibodies to the surface of B. parapertussis, allowing the latter to colonize hosts that had been previously immunized against the former. This provides an example of a single molecule, LPS, providing multiple, non-overlapping mechanisms to protect a bacterium against the effects of antibodies.
Despite excellent vaccine coverage, whooping cough has been re-emerging in vaccinated populations [41,42,43,44,45], but it is unclear what the relative roles of B. pertussis and B. parapertussis are in this resurgence [46]. Importantly, immunity induced by current vaccines protects against B. pertussis disease, but is largely ineffective against B. parapertussis disease [47,48,49,50,51]. The widespread use of vaccines appears to have resulted in a higher incidence of B. parapertussis as the causative agent of whooping cough in vaccinated individuals relative to unvaccinated individuals [49]. Current acellular vaccines induce a T cell response that is Th2-skewed [52]. Given that our data shows that the clearance of B. parapertussis by antibodies is enhanced by pro-inflammatory responses, a vaccine that generates a strong Th1-skewed response to B. parapertussis, as opposed to a partially cross-reactive Th2 type response, could potentially provide more efficient protection against this pathogen.