Lack of MyD88 Protects the Immunodeficient Host Against Fatal Lung Inflammation Triggered by the Opportunistic Bacteria Burkholderia cenocepacia¹

The increasing number of immunodeficient patients and the growing resistance of bacteria to commonly used antibiotics have made infections in the compromised host one of the most persistent medical issues. Burkholderia cenocepacia is a multidrug resistant Gram-negative bacillus that is widely distributed in the natural environment. This opportunistic pathogen is of particular concern for patients with cystic fibrosis (CF),⁴ in whom it can cause respiratory infection and has been associated with increased rates of morbidity and mortality (1, 2, 3). B. cenocepacia is also a well-known nosocomial pathogen among non-CF patient populations, including immunocompromised cancer patients and transplant recipients (4, 5). Spread of B. cenocepacia is likely facilitated by patient-to-patient airborne transmission (6, 7). However, the underlying mechanisms by which B. cenocepacia triggers a rapid health deterioration of the susceptible host have yet to be characterized.

TLRs represent a conserved family of innate immune recognition receptors that play key roles in detecting microbes, initiating innate immune responses, and linking innate and adaptive immunity (8, 9, 10, 11). The interaction of a microbial-associated molecular pattern with its specific TLR triggers the binding of the adapter molecule MyD88 to the intracellular carboxyterminus domain of the receptor (10). This is the first step of the signaling cascade that eventually activates NF-κB and other nuclear factors that regulate the expression of a large array of immune and inflammatory genes (10). Thus, TLRs appear to be directly involved in the fight of infections as illustrated by the prominent susceptibility of TLR- and MyD88-deficient mice to numerous pathogens (12).

Host defenses can be reduced by the underlying disease but also by a specific therapy such as chemotherapeutic drugs (13, 14). In regard to this latter aspect, vinblastine and other vinca alkaloids cause transient but severe immunosuppression (15). In this study, to gain insight into the role of TLR signaling in B. cenocepacia-induced pathogenesis in the immunocompromised host, we examined the responses of vinblastine-treated MyD88- and wild-type mice to infection by this opportunistic bacteria. Strikingly, we observed that immunosuppressed MyD88-deficient mice had an unexpected survival advantage. Furthermore, we demonstrated that a MyD88-dependent, TNF-α-mediated-inflammatory response induced by B. cenocepacia triggers fatal pneumonia and sepsis, which can be prevented by corticosteroids.

B. cenocepacia of the epidemic ET12 lineage (strain J2315) was provided by the Pasteur Institute microorganisms depository. Bacteria were grown on tryptic soy agar (TSA) at 33°C for 48 h. Single colonies removed from the plate were grown in 5 ml of tryptic soy broth at 33°C with shaking for 16–18 h, corresponding to midlog phase. Bacteria were harvested by centrifugation (3,000 × g for 15 min), resuspended in saline, and optical densities of the suspensions were adjusted to give the desired bacterial concentration; this later was verified by serial dilutions and plating on TSA.

TNFα^−/− and IL-1R^−/− were bred at the animal facility of Centre National de la Recherehe Scientifique. MyD88^−/− mice were obtained from S. Akira (Osaka University, Osaka, Japan). All mice were backcrossed at least eight times with C57BL/6 to ensure similar genetic backgrounds. Wild-type C57/BL6 mice were supplied by Centre d'Elevage R. Janvier and were used as control animals. Mice were cared for in accordance with Pasteur Institute guidelines in compliance with European animal welfare regulations. Mice were fed normal mouse chow and water ad libitum and were reared and housed under standard conditions with air filtration.

Chemotherapy-induced granulocytopenia was achieved by the i.v. administration of 5 mg/kg of the antineoplastic drug vinblastine (Cell Pharm) 66 h before infection. Under these conditions, the hematological profile of the mice after vinblastine and during the infection displayed total polymorphonuclear cell depletion from day 0 (the day of infection) until day 2 for both wild-type and MyD88^−/− mice. Then, the percentage of polymorphonuclear cells was 81.3 and 63 at day 3 post infection (p.i.) and 81 and 57.3 at day 4 p.i., in wild-type and MyD88^−/− animals, respectively (n = 3 to 5). Of note is that in a distinct investigation, we previously observed that under the vinblastine regimen, mouse survival and cytokine production were similar to those obtained when neutropenia was induced with the antigranulocyte mAb RB6-8C5 (16). The anti-inflammatory drug cortisone acetate (10 mg/mouse in saline; Sigma-Aldrich) was administered by the i.p. route. Mice were treated 2 days before infection, on the day of infection, and on day 2 after infection (17). Animals were infected intratracheally under general anesthesia achieved with a mixture of ketamine (40 mg/kg) and xylazine (8 mg/kg) administered via the i.m. route. A catheter (diameter, 0.86 mm) was inserted into the trachea via the oropharynx. Proper insertion was verified by checking the formation of mist due to expiration on a mirror placed in front of the external end. A 50-μl B. cenocepacia suspension (4 × 10⁷ cfu/mouse) was placed at the internal end of the catheter by introducing a micropipette with a sterile disposable tip for gel loading into the catheter. Mice were then immediately held upright to facilitate inhalation of bacteria and until normal breathing resumed. This protocol allows highly reproducible infection of the lungs, and 10 times more inoculum reaches the lungs via this route than via the intranasal route (18). Mice were then observed daily for signs of morbidity. Alternatively, mice were killed at different time points by i.p. injection of 300 mg/kg sodium pentobarbital, and 1 ml of heparinized blood was collected by the vena cava. After centrifugation at 300 × g, the resulting plasma was stored. Airways were washed twice with 1 ml saline, and the bronchoalveolar lavage (BAL) was collected to further determine cytokine concentrations using DuoSet ELISA kits (R&D Systems). Lungs were also washed free of blood by perfusing the heart and lungs with cold PBS. The lungs were homogenized in 1 ml of cold PBS. One hundred microliters of the homogenates were diluted and plated on TSA plates to determine the number of cfu of B. cenocepacia.

A commercial Ab-based protein array designed to detect 32 inflammatory mediators was also used according to the manufacturer’s instructions (RayBio Mouse Cytokine Array II; RayBiotech). Membrane arrays were hybridized with BAL fluids to compare wild-type and MyD88^−/− mice and were always processed simultaneously.

Whole lungs were fixed in 3.7% neutral buffered formaldehyde, embedded in paraffin, and cut into 5-μm sections. Sections were stained with H&E for tissue examination.

Survival of wild-type, MyD88^−/−, IL-1R^−/−, and TNF-α^−/− animals was compared using Kaplan-Meier analysis log-rank test. Inflammatory mediators levels and bacterial counts were expressed as the mean ± SEM. Differences between groups were assessed for statistical significance using the Kruskal-Wallis ANOVA test, followed by the Mann-Whitney U test. A value of p < 0.05 was considered statistically significant.

To dissect the role of the TLR adaptor MyD88 in B. cenocepacia pathogenesis, we characterized within the immunocompromized host the time-course of major dynamic parameters (i.e., animal mortality, pathogen load as well as secretion of critical mediators in two distinct compartments, the lung tissues and blood) in infected wild-type and MyD88^−/− animals pretreated with vinblastine. Fig. 1,A shows that all immunocompromized wild-type mice infected with B. cenocepacia at a dose of 4 × 10⁷ cfu per animal died within 7 days. Among the numerous signs of infection, piloerection, tachypnea, and anorexia were associated with a loss of weight (Fig. 1,B). Remarkably, we found that MyD88^−/−-infected mice had a higher survival rate than wild-type infected mice; i.e., ∼80 and 0%, respectively, at day 7 postinfection (p.i.; n = 11 and 15, respectively, p < 0.001; Fig. 1,A). Moreover, the aspect of MyD88^−/−-infected mice at day 2 p.i. was not different from that observed in noninfected animals (Fig. 1 B). These results clearly show that the absence of MyD88 undeniably protects mice from the lethal effects of B. cenocepacia in immunocompromized individuals.

As MyD88 is also known as an essential adaptor protein that integrates and transduces intracellular signals generated by IL-1R (19), we checked whether the secured phenotype of MyD88^−/− mice was due to a reduced impact of IL-1R-dependent signaling. By comparing the survival rate of IL-1R^−/− and wild-type animals infected under the same conditions as described above, we found that contrary to MyD88^−/− mice (Fig. 1,A), IL-1R^−/− animals were as susceptible as wild-type mice to B. cenocepacia fatal pneumonia (Fig. 1 C).

Immunosuppressed animals were killed at different intervals postinfection and BAL samples were collected to assess mediators content in the airspaces. We first analyzed kinetics of leukocyte influx into the bronchoalveolar spaces of either infected wild-type or MyD88^−/− mice. We found that polymorphonuclear neutrophils (PMN) and mononuclear cells recruitment in wild-type mice was detectable 24 h p.i. and increased at least until 96 h p.i. In MyD88^−/− mice, the kinetics of PMN influx was rather similar. In contrast, mononuclear cells recruitment was impaired when compared with the influx of those cells in wild-type animals (supplementary Fig. S1).⁵ Next, Fig. 2,A shows the time course of the expression of inflammatory chemokines and cytokines. In wild-type-infected mice, KC, TNF-α, MIP-2, and G-CSF concentrations peaked early at 6–24 h p.i. and decreased significantly thereafter. In regard to IL-6, its secretion was low compared with the other mediators and showed a two-wave shape, with the highest level at 6 h and a sustained expression until day 3 p.i. A very modest, delayed, secretion was observed for the anti-inflammatory cytokine IL-10 with a peak at 3 day p.i. Interestingly, the amount of the foregoing mediators were all much lower in MyD88^−/− than in wild-type BALs (Fig. 2,A). For instance, keratinocyte-derived cytokine (KC) secretion in MyD88^−/− animals was ∼5% of that observed in wild-type mice. Next, BAL fluids were analyzed by an inflammatory protein array to examine whether this distinct profile in wild-type and MyD88^−/− mice could be confirmed when considering expression at 24 h p.i. of additional mediators. The obtained data (Fig. 2,B) not only validated the previous ELISA findings but also revealed a similar reduced expression of MCP-1, MIP-1α, or sTNFR1α in MyD88^−/− mice relatively to wild-type ones. Finally, histological analysis of the lungs at 72 h p.i. indicated that in wild-type mice, B. cenocepacia induced a severe bronchopneumonia characterized by widespread patchy areas of inflammation that began as a bronchiolitis and alveolitis extending to the alveolar lining epithelium. Numerous PMN and alveolar macrophages were recruited at this time point and could be seen within the bronchioles and alveolae, mixed with necrotic alveolar epithelial cells and exsudation of RBC (Fig. 2,C). By contrast, in MyD88^−/− mice, the lesions were less severe despite the presence of few polymorphonuclear cells. The alveolar wall was not disrupted, only a few necrotic cells could be observed and no hemorrhage was apparent (Fig. 2 C). Altogether, these data suggest that lung lesions induced by B. cenocepacia infection in an immunocompromized host are strongly reduced in the absence of MyD88.

Next, we performed additional experiments to examine B. cenocepacia pathogenesis in immunocompetent, nonvinblastin-treated, wild-type and MyD88^−/− mice challenged by 4 × 10⁷ cfu/mouse. We observed no animal death under these experimental conditions (even using a higher inoculum, i.e., 10⁸ cfu/mouse; data not shown). Nevertheless, an inflammatory response was observed in the BALs of infected immunocompetent wild-type animals although its intensity and/or the duration were low. Thus, except for IL-6, which was increased (by 55%), the level of all other studied mediators were reduced in the BALs of immunocompetent wild-type mice in comparison with vinblastin-treated wild-type animals at 24 h postinfection, i.e., 100% for KC, ∼75% for TNF-α, 100% for MIP2α, and ∼48% for G-CSF. Moreover, this immune response was strongly reduced in the BALs of immunocompetent MyD88^−/− mice. As a single example, TNF-α content at 6 h p.i. in the BALs of immunocompetent wild-type and MyD88^−/− was 2447 ± 216 pg/ml and 554 ± 68 pg/ml, respectively (n = 4).

The previous results indicated that a potent inflammatory reaction occurs in the lungs of wild-type mice after B. cenocepacia infection and that this process is critically reduced in MyD88^−/− animals. Inflammatory signaling pathways during microbial infection have been interpreted in some cases as a protective response of the host, whereas in other cases pathogens can use these pathways to enhance their replication (20, 21). Thus, to investigate whether MyD88-dependent host response might regulate the amount of B. cenocepacia, immunosuppressed wild-type and MyD88^−/− mice were killed 6, 24, and 72 h after infection and bacterial numbers both in BAL fluids and in lung tissues were determined. Fig. 3 shows a similar elevated bacterial load at 6 h p.i in lung tissues from both wild-type and MyD88^−/− mice. Paradoxically, while lungs of wild-type animals had very significantly cleared B. cenocepacia as early as 24 h p.i., bacteria persisted at that time in MyD88^−/− mice with a load ∼10 times higher than in wild-type animals. However, for both groups of mice, lungs were almost cleared from the bacterial infection by 72 h p.i.

It is well established that patients with B. cenocepacia pneumonia are likely to develop the so-called “cepacia syndrome”, a multiple organ failure due to a septic shock, associated or not with bacteremia (22). In view of this consideration and to better understand the role of MyD88 signaling in the pathogenesis of B. cenocepacia in our immunodeficient host model, we assessed blood bacterial content as well as the expression of circulating inflammatory cytokines in infected wild-type and MyD88^−/− mice. Whatever the time point at which blood samples were collected (6 or 24 h p.i.), we could not recover growing B. cenocepacia in either group (data not shown). Next, we used a multiplex cytokine assay (assessing IFN-γ, KC, TNF-α, IL-1α, IL-5, IL-10, IL-17, and IL-12p40), as well as specific ELISA to determine whether inflammatory mediators could be present in plasma of both murine groups and whether their relative amount could be affected by MyD88 deficiency. Among all searched molecules, only IL-6 and KC were significantly detected in plasma harvested at 6 and 24 h p.i. in wild-type and MyD88^−/− mice. Their concentration was further precisely quantified by ELISA. Fig. 4 shows that in wild-type mice, a peak of IL-6 was observed at 6 h p.i. followed by a rapid decrease within 24 h toward the basal values measured in noninfected animals. In contrast, the systemic expression of KC was sustained with a 2-fold higher level at 24 h than at 6 h p.i. In regard to MyD88^−/− animals, concentrations of both KC and IL-6 were significantly reduced compared with wild-type samples, especially at 6 h p.i. (p < 0.01 and p < 0.001, respectively).

In view of the apparent deleterious MyD88-dependent acute inflammation triggered by B. cenocepacia in the immunocompromised host, we speculated that administration of an anti-inflammatory regimen such as corticosteroids (23) would limit the life-threatening consequences of this bacterial infection. To verify such hypothesis, cortisone acetate was administered i.p. to vinblastine-treated wild-type mice (n = 10). A second group of vinblastine-treated mice received saline under the same experimental conditions and served as controls (n = 6). Mice were then observed for 6 days to measure their survival. Fig. 5,A indicates a significant protective effect of corticosteroids. Indeed, while >80% of vinblastine plus saline-treated mice died by day 6 p.i., all animals receiving vinblastine plus cortisone acetate survived. In a separate set of experiments, mice were infected and treated exactly as above, and BAL fluids were collected at different time points p.i. to ascertain whether the administration of cortisone acetate was effective against the acute pulmonary inflammation triggered by B. cenocepacia. Consistent with the survival data, corticosteroid-treated immunocompromised and infected mice had a contained inflammatory response as illustrated by the highly significant decrease of KC, IL-6, TNF-α, MIP-2, and G-CSF concentrations compared with saline-treated mice (Fig. 5 B). Interestingly, only the anti-inflammatory cytokine IL-10 was moderately, but not significantly, increased (p > 0.05). In regard to leukocyte recruitment into the bronchoalveolar space of infected wild-type mice pretreated with vinblastine alone or with the combination of vinblastine and corticosteroids, the supplementary Fig. S2 indicates that leukocyte influx differed only moderately between the two groups of animals.

More importantly, using a specific gene-deficient mouse model, we demonstrated a critical role of TNF-α in mediating B. cenocepacia pathogenesis. Indeed, when wild-type and TNF-α^−/− mice were challenged by 4 × 10⁷ cfu/animal, the absence of TNF-α conferred a striking advantage (0 and 100% of animal death in TNF-α^−/− and wild-type, respectively, even at day 10 p.i.; Fig. 5 C).

Combating nosocomial disease associated to B. cenocepacia is still a major clinical challenge despite considerable advances in epidemiology, taxonomy, as well as in identification of virulence factors (24, 25). Therefore, it is of importance to gain insight into the immune response to B. cenocepacia infection to potentially identify new therapeutic strategies. Our knowledge of how the immune system recognizes pathogens has increased exponentially in recent years due to the discovery of the TLR family. Recently, we demonstrated using in vitro approaches that TLR5 and MyD88, but neither TLR2 nor TLR4, play a central role in the response of bronchial epithelial cells to B. cenocepacia (26). These findings implied that flagellin, the agonist of TLR5 was the key B. cenocepacia proinflammatory molecule. Our results were consistent with an in vivo study showing that flagellin-induced signaling was required for B. cenocepacia virulence in a mouse lung infection model (27). More generally, several clinical reports confirmed the contribution of TLRs to the pathophysiology of infectious diseases, and polymorphisms in TLR genes are associated with predisposition to severe infections (28, 29, 30).

In view of this major information, we anticipated at the start of our investigation that MyD88 expression would be a prerequisite for effective host defense in vivo against B. cenocepacia. Conversely, its absence would render MyD88^−/− mice more susceptible to B. cenocepacia pneumonia. Surprisingly, our study reveals that mice deficient in MyD88, even after immunosuppression by vinblastine, have an unexpected advantage. Thus, in comparison with wild-type mice, we found in MyD88^−/− animals a clearly reduced level of inflammatory mediators both in lung airspaces and in blood, and, most important, a paradoxical extended survival. Based on these findings and in view of our results describing the rescuing effect of corticosteroids in the pathogenesis of B. cenocepacia infection, we discuss below the fact that the enhanced resistance of B. cenocepacia-infected MyD88^−/− might be due to a lower MyD88-mediated release of inflammatory mediators, especially of TNF-α.

The abrogated response to TLR ligands by MyD88 knock-out mice has provided an invaluable tool for analyzing the critical role of MyD88-dependent signaling in host defense against infection (12). Multiple studies have demonstrated that MyD88 deficiency renders animals highly susceptible to microbial infection. As a few examples, one can mention the protective in vivo role of MyD88 in infections by the Gram-positive or negative bacteria Pseudomonas aeruginosa, Staphylococcus aureus, Chlamydia pneumoniae, or Mycobacterium tuberculosis, the fungus Candida albicans, the parasite Toxoplasma gondii, and the herpes simplex virus-2 (18, 31, 32, 33, 34, 35). Accordingly, our current demonstration that MyD88^−/− mice are protected from the lethal effects of B. cenocepacia revisits noticeably the paradigm. Moreover, our findings are especially remarkable because, in our experimental model, animals were immunocompromised while all other previous studies have tested the role of MyD88 in immunocompetent hosts.

There is growing evidence that the mediators that have a central role in the resolution of bacterial infections, are the same that causes many clinical signs related to these diseases. Thus, cytokines trigger a vigorous localized inflammatory response that is required for elimination of the particular pathogen. However, the magnitude of inflammatory cytokine expression must be tightly regulated and compartmentalized to prevent excessive tissue injury (36, 37, 38, 39). We show in the present study that B. cenocepacia infection leads to the synthesis of major inflammatory cytokines and chemokines in lung tissues. More importantly, our study establishes that MyD88 plays a major role in their production, the lack of this signaling molecule resulting in a significant decrease of KC, IL-6, TNF-α, MIP-2, and G-CSF synthesis. Interestingly, elevated cytokine levels have been implicated in both systemic and local inflammation and may play a role in “cepacia syndrome” (40). How lung inflammation triggers multiorgan dysfunction is unclear but neural and humoral mechanisms may be involved (41). In that regard, our study is the first to reveal the involvement of the proinflammatory cytokine TNF-α in the development of fatal pneumonia induced by B. cenocepacia. Moreover, a clear conclusion from our work is that a reduction in MyD88-mediated inflammation reduces the clinical manifestations of B. cenocepacia-induced pneumonia. It is thus reasonable to hypothesize that targeting MyD88 and/or TNF-α signaling pathway in B. cenocepacia-infected immunocompromised patients has great therapeutic potential, although significant complexities must be considered. Indeed, MyD88 and TNF-α fulfils important functions in innate immunity, cell death, and is intimately related to adaptive immunity, endocrine, circulatory, and nervous systems (42, 43).

Meanwhile, our study suggests that corticosteroids are a promising therapeutic approach to specifically dampen inflammation during B. cenocepacia pneumonia. This possibility is supported by previous works showing that corticosteroids spare or even enhance numerous lung innate immune responses while they concomitantly suppress inflammation (44). Incidentally, the present work sheds a new light on the limited number of clinical case reports that indicated improvement of critically sick CF and non-CF patients with B. cepacia syndrome, after use of corticosteroids (22, 45).

We are grateful to Micheline Lagranderie (Laboratoire Immunothérapie, Institut Pasteur) for the multiplex cytokine assay and for helpful discussions.

The authors have no financial conflict of interest.