ABSTRACT
Burkholderia dolosa caused an outbreak in the cystic fibrosis (CF) clinic at Boston Children's Hospital from 1998 to 2005 and led to the infection of over 40 patients, many of whom died due to complications from infection by this organism. To assess whether B. dolosa significantly contributes to disease or is recognized by the host immune response, mice were infected with a sequenced outbreak B. dolosa strain, AU0158, and responses were compared to those to the well-studied CF pathogen Pseudomonas aeruginosa. In parallel, mice were also infected with a polar flagellin mutant of B. dolosa to examine the role of flagella in B. dolosa lung colonization. The results showed a higher persistence in the host by B. dolosa strains, and yet, neutrophil recruitment and cytokine production were lower than those with P. aeruginosa. The ability of host immune cells to recognize B. dolosa was then assessed, B. dolosa induced a robust cytokine response in cultured cells, and this effect was dependent on the flagella only when bacteria were dead. Together, these results suggest that B. dolosa can be recognized by host cells in vitro but may avoid or suppress the host immune response in vivo through unknown mechanisms. B. dolosa was then compared to other Burkholderia species and found to induce similar levels of cytokine production despite being internalized by macrophages more than Burkholderia cenocepacia strains. These data suggest that B. dolosa AU0158 may act differently with host cells and is recognized differently by immune systems than are other Burkholderia strains or species.
INTRODUCTION
Cystic fibrosis (CF) has an estimated prevalence of 1/2,500 live births and is the result of mutations in the CFTR gene. This disorder manifests itself in an imbalance of chloride ions in airway surface liquid and leads to the accumulation of dehydrated mucus in the lungs and the gastrointestinal tract. The mucus is a nutrient-rich breeding ground for bacteria, resulting in acute and chronic lung infections in CF patients throughout the course of their lives. Patients with CF experience periodic bouts of lung function decline. Usually, following hospitalization and high-dose antibiotic treatments, a return of some lung function is observed, although the pulmonary function steadily declines over time. Common pathogens that cause acute lung exacerbations in CF patients are Pseudomonas aeruginosa, Staphylococcus aureus, Stenotrophomonas maltophilia, Achromobacter xylosoxidans, and members of the Burkholderia cepacia complex (BCC) (1).
Members of the BCC are capable of producing a necrotizing pneumonia in infected patients, characterized by a rapid decline in lung function (1 to 3 months) and mortality, often with bacteria detected in the bloodstream, a phenomenon often known as “cepacia syndrome” (2). Certain species within the BCC, including the species Burkholderia dolosa and Burkholderia cenocepacia, are capable of facile patient-to-patient transmission and have caused epidemics in CF clinics worldwide (3). The virulence factors known to be important to the BCC include intrinsic antimicrobial resistance (4–6), biofilm formation (7, 8), catalase and superoxide dismutase (9), invasion and intracellular survival within host cells (10, 11), lipopolysaccharide (LPS) (12, 13), quorum sensing (8, 14–18), siderophore production (19–22), and type III secretion systems (23–25). The role of the B. cenocepacia flagella has also been shown to be important for disease and cytokine production in murine lungs and Toll-like receptor 5 (TLR5)-dependent interactions with host cells (26, 27).
B. dolosa caused an outbreak in Boston Children's Hospital, Boston, MA, starting in 1998 arising from an index patient who came from Salt Lake City, UT, in 1992 (28). The index patient carried B. dolosa (then known as part of Pseudomonas cepacia) for at least 6 years and experienced no decrease in lung function associated with B. dolosa colonization. Additional CF patients contracted this particular strain of B. dolosa through patient-to-patient spread (3), and while some developed cepacia syndrome, others remained asymptomatic. This outbreak was unusual, as B. dolosa had been rarely isolated in clinical settings previously. The genome of one outbreak isolate, AU0158, was sequenced and found to have three circular chromosomes (similar to other BCC members) which result in a total genome size of 6.41 Mbp (29). Little research has been done on the mechanisms used by B. dolosa to cause pathogenicity or persist for extended periods inside the human lung. However, it is known that B. dolosa is comparable to other strains for invasion of host cells and in biofilm formation (7).
Members of the BCC can produce a robust immune response in several mouse strains (30–32) and can invade and replicate in cultured respiratory epithelial cells (7, 13), thus leading to the conclusion that BCC strains may be intracellular pathogens. More recent work in transplanted or autopsied CF lungs suggests that BCC strains, including B. dolosa and B. cenocepacia, can be found primarily associated within host cells, particularly macrophages, or in mucus (33). Previous studies have demonstrated that a well-studied CF pathogen, P. aeruginosa, colonizes the respiratory tract of CF patients and triggers an immune response (34–39). It was previously thought to primarily be an extracellular pathogen, which is supported in a histological study of CF lungs (33), though other work in eye models suggests that P. aeruginosa may be able to invade and survive inside host cells (40–44). Both B. cenocepacia, the best-studied of the BCC members, and P. aeruginosa have been shown to induce production of cytokines (i.e., tumor necrosis factor [TNF], interleukin-6 [IL-6], IL-8, and IL-1β) in murine alveolar macrophages in vitro (45–47). From these observations, we hypothesized that B. dolosa can colonize host cells in vivo and in vitro, leading to cytokine production, and that B. dolosa would likely interact with the host in a manner different from P. aeruginosa due to its intracellular life cycle in host cells.
To elucidate the host immune response to B. dolosa, bacterial clearance, host immune cell recruitment, and cytokine responses following infection in the mouse lung were examined. Additionally, the cytokine profile of cultured cells infected with B. dolosa was also tested to determine if in vitro and in vivo responses were similar. Last, the role of the B. dolosa polar flagellum in bacterial colonization and immune recognition was investigated, as this is a well-studied virulence factor that interacts with the host immune response through recognition by host Toll-like receptor 5 (TLR5). The result of these experiments may help explain why B. dolosa can persist in the lungs of CF patients for extended periods of time without being cleared by the host immune response.
RESULTS
Bacterial survival, neutrophil recruitment, and cytokine production in a mouse model of infection.When performing a pilot examination of B. dolosa AU0158 in C57BL/6 mice, it was observed that dosages up to 1010 CFU/mouse did not result in symptoms or lethality but did result in a persistent colonization for more than several days. In order to compare B. dolosa with the well-studied positive control in our study, P. aeruginosa, a dosage of ∼106 CFU/mouse was chosen for all bacterial strains to eliminate dosage-induced variability. This dosage was previously shown to be 10-fold lower than the average 50% lethal dose (LD50) for P. aeruginosa PAO1 in this mouse line (48). Groups of 5 C57BL/6 mice were intranasally inoculated with B. dolosa AU0158 wild type (WT), B. dolosa AU0158 ΔfliC, P. aeruginosa PAO1, and P. aeruginosa PAO1 ΔfliC strains each at an inoculum of ∼2.5 × 106 CFU/mouse.
At 1, 6, 24, and 144 h postinfection (hpi), mice were sacrificed and the lungs were processed by collecting bronchoalveolar lavage (BAL) fluid and homogenizing the lungs. Bacterial survival was assessed by plating 10-fold serial dilutions of the BAL fluid and lung homogenates on a standard lab medium. As shown in Fig. 1, all strains were found at similar levels in the BAL fluid and lung homogenates at the 1- and 6-h time points. However, at 24 hpi, P. aeruginosa-infected mice showed a marked decrease in bacterial levels compared to those infected with B. dolosa in both the BAL fluid and lung homogenates. At 144 hpi, B. dolosa wild-type strain AU0158 was undetectable in the BAL fluid (Fig. 1A) but was found at relatively high levels in the lung homogenates (Fig. 1B), while P. aeruginosa strains were not detected at this time point.
Bacterial persistence in the mouse lung over time. Replicate mice were inoculated with ∼5 × 106 CFU of each strain. Bacterial levels in the BAL fluid (A) or lung homogenates (B) were measured by serial dilution and plating at 1, 6, 24, and 144 h postinfection. The color of each bar corresponds to an individual strain: black, P. aeruginosa PAO1 wild type; dark gray, P. aeruginosa PAO1 ΔfliC; light gray, B. dolosa AU0158 wild type; white, B. dolosa AU0158 ΔfliC. Error bars represent 1 standard deviation of the data. Brackets correspond to significant differences between strains (**, P < 0.005) as calculated by t tests. One-way ANOVA indicated no significant differences.
To examine the recruitment of immune cells into the lungs in response to each bacterial strain, we analyzed host cells in the BAL fluid using flow cytometry. To quantify the number of neutrophils, fluorescent antibodies which recognized the surface proteins CD11b and Ly6G were used, as neutrophils have typically high levels of these surface markers (i.e., CD11bhigh Ly6Ghigh). P. aeruginosa strains recruited high levels of CD11bhigh Ly6Ghigh cells at 6 and 24 hpi (Fig. 2A). As expected, infection by the P. aeruginosa wild-type strain produced a robust response that was significantly higher than that induced by the isogenic ΔfliC mutant, most likely due to the lack of TLR5 binding and subsequent downstream signaling and cytokine production. The B. dolosa strains recruited a very low level of CD11bhigh Ly6Ghigh cells early in infection but slightly more at 24 h, and this level of recruitment was significantly less than the number of cells recruited by P. aeruginosa. Interestingly, we also used another surface marker antibody which recognizes the F4/80 surface protein and observed the recruitment of an unusual population of Ly6Ghigh F4/80high cells for P. aeruginosa strains at 6 hpi and B. dolosa strains at 6 and 24 hpi (Fig. 2B).
Recruitment of immune cells to the BAL fluid in response to bacterial strains. FACS analysis was used to quantify CD11b+ Ly6G+ cells (neutrophils) (A) and Ly6G+ F4/80+ cells (B) at 6 and 24 hpi. The color of each bar corresponds to an individual strain: black, P. aeruginosa PAO1 wild type; dark gray, P. aeruginosa PAO1 ΔfliC; light gray, B. dolosa AU0158 wild type; white, B. dolosa AU0158 ΔfliC. Overall one-way ANOVA indicates significantly different means for neutrophils at 6 (P < 0.0001) and 24 (P < 0.0001) hpi and Ly6G+ F4/80+ cells at 6 (P < 0.05) and 24 (P < 0.001) hpi. Asterisks atop brackets indicate significant differences calculated by one-way ANOVA with Tukey's multiple-comparison test (*, P < 0.05; ***, P < 0.001).
Because P. aeruginosa is known to elicit host secretion of several cytokines such as granulocyte-macrophage colony-stimulating factor (GM-CSF) and TNF, we determined whether B. dolosa also produced a robust proinflammatory cytokine response in the murine lung. P. aeruginosa colonization led to the production of high levels of GM-CSF, TNF, IL-12 p40, IL-1α, IL-1β, and IL-6 all between 6 and 24 hpi (Fig. 3), of which only TNF was significantly dependent on the flagellar status of P. aeruginosa, which agrees with previous studies (39, 49). Colonization by B. dolosa strains did not lead to a strong production of any tested cytokine but produced mild responses for gamma interferon (IFN-γ), IL-2, IL-12 p40, and IL-6. In fact, for most of the cytokines measured, the levels produced in mice exposed to B. dolosa were close to or below the limit of detection. This suggests that mice are not sensing or responding to B. dolosa in a similar manner as to P. aeruginosa.
Cytokine production in response to P. aeruginosa PAO1 and B. dolosa AU0158 wild-type and mutant strains. Cytokine production was measured for those involved in the Th1 proinflammatory pathway (A), Th2 allergy pathway (B), innate/acute immunity (C), and other pathways (D). Cytokine levels in the BAL fluid were measured at 6, 24, and 144 hpi for five replicates each, and the median value is represented. The color of each bar corresponds to an individual strain: black, P. aeruginosa PAO1 wild type; dark gray, P. aeruginosa PAO1 ΔfliC; light gray, B. dolosa AU0158 wild type; white, B. dolosa AU0158 ΔfliC. Dashed lines on graphs indicates minimum levels of detection. N.A. indicates that 3 or more replicate samples had no detectable level of cytokine. Asterisks next to cytokine titles indicate overall two-way ANOVA significance values based on the interaction between time and strain. Brackets above the bars indicate a significant difference between wild-type strains or between mutants and the respective wild type based on Bonferroni posttests (*, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001).
Role of FliC in murine colonization and cytokine production.We also tested the hypothesis that the host-B. dolosa interaction may be dependent on bacterial flagellin recognition by the host innate immune response (27, 50). We observed no significant difference between B. dolosa wild type and the fliC mutant strain in bacterial loads following colonization (Fig. 1), in the recruitment of neutrophils (Fig. 2), or in the production of most cytokines (Fig. 3). The exception to the latter was IL-12 p40, which depended on the flagellar status, with the ΔfliC mutant showing a 2.4-fold increase in IL-12 p40 production over the wild type only at 144 hpi. Additionally, IL-12 p40 cytokine levels from mice infected with B. dolosa strains were considerably lower than those observed for either the P. aeruginosa wild type or the ΔfliC mutant.
Response of cultured murine cells to B. dolosa and P. aeruginosa.To distinguish whether host cells can recognize B. dolosa, we used cultured murine macrophages (RAW 264.7 cells) to determine if cytokines were produced in the absence of the full repertoire of immune effectors present in vivo. Based on our in vivo mouse data, we hypothesized that cultured cells would not sense and respond to B. dolosa strains by producing cytokines. We tested our hypothesis by measuring the levels of two proinflammatory cytokines, TNF and MIP-2 (an IL-8 homolog), following exposure to P. aeruginosa and B. dolosa wild-type and ΔfliC mutant strains. Wild-type P. aeruginosa rapidly kills cultured cells at multiplicities of infection (MOIs) of 20 or more. At lower MOIs, P. aeruginosa wild-type and ΔfliC mutant strains induced a strong TNF and MIP-2 response (Fig. 4). Interestingly, RAW 264.7 cells produced more TNF in response to B. dolosa strains than to P. aeruginosa, and this difference was statistically significant. The B. dolosa wild type and the ΔfliC mutant did not induce these cytokines to different degrees. Together, these results suggest that cultured cells recognize and respond to B. dolosa strains, and this contradicts the results that we observed in our in vivo model.
Cytokine production in response to P. aeruginosa PAO1 and B. dolosa AU0158 wild-type and mutant strains in cultured murine macrophages. TNF and the murine IL-8 homolog MIP-2 proinflammatory cytokine were measured in RAW 264.7 cells after a 2-hour exposure. Overall one-way ANOVA indicated a significant difference (P < 0.0001) between strains for TNF but not MIP-2. Brackets above the bars indicate a significant difference between wild-type strains or between mutants and the respective wild type based on one-way ANOVA with Tukey multiple-comparison posttests (***, P < 0.001).
Response of cultured human cells to B. dolosa. B. dolosa is known to persist in the cystic fibrosis lung for years in some cases and sometimes without a significant decline in lung function, as observed in some patients during the epidemic occurring at Boston Children's Hospital from 1998 to 2005 (28). To determine if the cytokine profile of cultured human cells resembles those from the mouse cell studies, we exposed peripheral blood mononuclear cells (PBMCs) taken from the donated blood of healthy humans to the same panel of strains used in our prior experiments and examined the levels of the proinflammatory cytokines TNF, IL-8, and IL-1β produced. Live bacterial cells were used initially, and PBMCs were exposed for 24 h, after which enzyme-linked immunosorbent assays (ELISAs) were performed. Interestingly, the wild-type P. aeruginosa strain rapidly killed the PBMCs and thus could not be used for ELISAs (Fig. 5A); however, the P. aeruginosa ΔfliC mutant was relatively noncytotoxic, which is in accordance with published studies (51–53). When live B. dolosa strains were added to PBMCs, the eukaryotic cells produced high levels of TNF, IL-1β, and IL-8, and this production was not dependent on the B. dolosa flagellar status.
Cytokine production in response to P. aeruginosa PAO1 and B. dolosa AU0158 wild-type and mutant strains in peripheral blood mononuclear cells (PBMCs). TNF, IL-1β, and IL-8 proinflammatory cytokine production was measured in PBMCs after exposure to live cells (LC) (A) and heat-killed cells (HC) (B) at a multiplicity of infection of 50 bacteria per PBMC. N.A. indicates that 3 or more replicate samples had no detectable level of cytokine. Overall one-way ANOVA significance values are given by asterisks next to the y axis label. Brackets above the bars indicate a significant difference between wild-type strains or between mutants and the respective wild type based on one-way ANOVA with Tukey multiple-comparison posttests (*, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001).
We next used heat-treated cells to ensure that the PBMCs remained alive over the course of the experiment. PBMCs were treated with bacteria previously heated at 65°C for 30 min and cooled. Under these conditions, wild-type P. aeruginosa induced significantly different cytokine responses than did the wild-type B. dolosa strain for all three cytokines measured (Fig. 5B). In the case of TNF and IL-8, B. dolosa induced these cytokines to a higher degree than P. aeruginosa. B. dolosa induced a slightly smaller amount of IL-1β than did P. aeruginosa. The production of all the cytokines in response to heat-killed B. dolosa strains was significantly dependent on the flagellar status of the strain, as the B. dolosa AU0158 ΔfliC strain showed significantly higher secretion levels for all three cytokines than did wild-type B. dolosa.
Taken together, these results suggest that human primary cells readily sense and respond to intact, live B. dolosa strains and generally this recognition is not fliC dependent. These data further support the in vitro murine cell experiments (Fig. 4) that also show a robust cytokine response in single cell types. Because flagellar dependence relies on whether the bacterial cells are living, this may imply that live B. dolosa cells are involved in an active process to mitigate cytokine responses in vitro.
Comparison between B. dolosa AU0158 and other Burkholderia strains and species for macrophage internalization and cytokine production.To assess whether B. dolosa AU0158 is unique compared to other B. dolosa and BCC isolates, we conducted internalization assays on B. dolosa outbreak and nonoutbreak isolates collected from CF patients as well as 3 Burkholderia multivorans isolates (CF1, CF2, and CGD2), 4 B. cenocepacia isolates (H111, AU1054, HI2424, and MC0-3), and 1 isolate each from Burkholderia ambifaria (AMMD), Burkholderia lata (ATCC 17760), and Burkholderia vietnamiensis (G4). For comparison, we also examined the internalization by murine macrophage cells for a species closely related to the BCC, Burkholderia pseudomallei Bp82, as well as P. aeruginosa PAO1 and macrophages alone as controls. We observed very few significant differences between the BCC isolates in host cell internalization (Fig. 6A), suggesting that B. dolosa AU0158 wild type can invade host cells at a similar rate as many BCC isolates. However, we observed significantly higher internalization rates for 3 additional B. dolosa strains, R25-003 (∼4-fold higher), AU131412 (∼3-fold higher), and AU19373 (∼6-fold higher), compared to B. dolosa AU0158. R25-003 and AU19373 are isolates taken during the Boston Children's Hospital outbreak. The result from B. dolosa AU13412, a nonoutbreak B. dolosa strain obtained from a CF patient, is intriguing as it reveals that the elevated macrophage internalization rates observed in the Boston Children's Hospital outbreak isolates may not be restricted to this epidemic strain but may be a broader trait of clinical B. dolosa strains.
Burkholderia species internalization into RAW 264.7 murine macrophage cells and TNF production. B. dolosa AU0158 was compared to other Burkholderia strains and species and control treatments in RAW 264.7 murine macrophage infections by internalization assays (A) and the induction of TNF from macrophages (C). For both panels, strains are B. dolosa (black with white stripes indicating nonoutbreak strains), B. multivorans (dark gray), B. cenocepacia (medium gray), other Burkholderia species (light gray), or controls (white). N.D. indicates none detected. One-way ANOVA indicated an overall P value of <0.001 for data represented in panels A and C, and Dunnett's multiple-comparison tests were conducted for all tested strains against the B. dolosa AU0158 wild-type control strain. Data for each strain were aggregated by species and compared for internalization (B) or TNF induction (D) and analyzed using one-way ANOVA and Tukey's multiple-comparison test. Asterisks in all panels indicate P values of <0.05 (*), <0.01 (**), or <0.001 (***).
To further assess whether B. dolosa AU0158 differed from other BCC members, we quantitated the TNF levels from murine macrophage cells following a 6-h exposure to these same strains. Interestingly, there was a wide range of TNF levels induced by these strains. As shown in Fig. 6C, B. dolosa AU0158 induced a moderate level of TNF under these conditions, and while TNF production varied in other B. dolosa strains, only the lower production of TNF by macrophages infected with AU19373 was statistically significant. We also observed significant decreases in induction of TNF by B. multivorans CF1 (but not other B. multivorans strains), B. cenocepacia strains H111 and HI2424, B. vietnamiensis G4, B. pseudomallei Bp82, P. aeruginosa PAO1, and the no-bacterium control. We note that several of the strains that showed higher internalization than B. dolosa AU0158 showed a significantly lower TNF induction, possibly due to less recognition by extracellular host receptors, such as the Toll-like receptors.
Last, the data from all strains within a species were aggregated and analyzed to determine whether B. dolosa was more likely to invade macrophages or induce TNF production in these cells than were the other major BCC CF pathogens, B. multivorans and B. cenocepacia. For this, we tested only those species in which we utilized more than one isolate. We observed that B. dolosa is not significantly different from B. multivorans for internalization or TNF induction (Fig. 6B and D) but shows a significantly increased rate of internalization compared to B. cenocepacia isolates and the P. aeruginosa control. The only significant difference observed was the cytokine levels induced by B. dolosa and P. aeruginosa (for which we included in our assays a second isolate, C3719), in which B. dolosa induced a significantly higher production of TNF from murine macrophages than did P. aeruginosa under these conditions. We conclude from these data that B. dolosa species are better at invading cultured macrophages than B. cenocepacia but not B. multivorans; however, B. dolosa induces TNF production to levels similar to those for other BCC species.
DISCUSSION
The results reported in this study indicate that in an in vivo murine model of B. dolosa lung infection, there was a minimal proinflammatory cytokine response elicited, and this was in contrast to the ability of cultured murine and human cells to recognize this pathogen and secrete high levels of proinflammatory cytokines. We envision that the mechanism that B. dolosa uses to avoid detection could result (i) from evasion of the immune response through masking itself from host receptors or through altering the signaling pathways in host cells or (ii) via direct suppression of the immune response through the recruitment of cells that turn off the immune response.
Evasion of the immune system could be occurring through one or more mechanisms. It has been observed in other BCC members that glycosylation of flagella may contribute to the bacterium's ability to hide from the immune system, though this glycosylation did not prevent recognition by host cells completely (54). However, these studies were not performed using B. dolosa, and whether B. dolosa has similar glycosylation patterns is yet to be determined. It is also possible that B. dolosa may have changes in the flagellin amino acid sequence which reduce the affinity of this protein for TLR5 on host epithelial cells and macrophages. While B. dolosa AU0158 does have a number of changes at binding sites identified to be important in P. aeruginosa and Salmonella enterica, this cannot solely explain the lack of host immune response in vivo, as receptors other than TLR5 (e.g., TLR4 and NOD receptors) should recognize additional pathogen-associated molecular patterns (PAMPs) and be able to induce a cytokine response.
If evasion of the immune system either via masking PAMPs from host receptors or by altering the signaling pathways in host cells was the main mechanism by which B. dolosa was able to persist in the host, then the in vitro data from cultured cells should have recapitulated the results observed in the mouse (i.e., no cytokine production). However, we observed robust production of several cytokines, indicating that these cells can readily recognize B. dolosa. These in vitro experiments remove any interaction between B. dolosa and other host cell types (such as immune suppressor cells) since they are not present in the cultures. If direct suppression of the host immune response is the mechanism by which B. dolosa persists in the host, the data should have shown that B. dolosa is recognized by the host immune response in vitro, and in fact, that is what we observed. Interestingly, using heat-killed cells resulted in higher overall IL-8 production from PBMCs but did not result in higher overall levels of TNF or IL-1β produced, which may suggest that live cells are better able to evade some host responses but not all. This experiment also revealed significantly higher levels of all three tested cytokines for the B. dolosa flagellin mutant only when heat killed, which might suggest that live bacterial cells may be able to alter the flagella to prevent immune recognition but heat-killed cells may not be able.
B. dolosa is readily internalized by macrophages (Fig. 6), and this may also contribute to host immune evasion. In doing so, bacterial cells avoid recognition by the Toll-like receptors and thus the downstream cytokine production resulting from the activation of the subsequent pathways. However, intracellular mechanisms of pathogen detection through the RIG-I and Nod-like receptors should still be able to detect internalized bacterial pathogens (55–59) through RNA or peptidoglycan-related products. If B. dolosa can evade the immune response through internalization, it must have altered the composition of many putative targets for these receptors as well in order to result in a virtual lack of detection (55–59).
While it is clear that members of the BCC can induce immune responses in cultured mammalian cells (12, 54, 60–63) (Fig. 6), there is much less known regarding the host immune responses in vivo to this group of organisms. Ventura et al. showed that B. cenocepacia, the best-studied of the BCC members, can elicit strong TNF, IL-6, and IL-10 responses in a MyD88-dependent manner in C57BL/6 mice (30). Other groups have demonstrated that B. cenocepacia can induce strong production of numerous cytokines, including IL-1β, TNF, monocyte chemoattractant protein 1 (MCP-1), IL-17, IFN-γ, IL-4, and IL-10 in cultured PBMCs from CF patients (47) and in neutropenic C57BL/6 mice or CFTR−/− knockout mice (31). Similarly, others have shown that another commonly isolated BCC member, B. multivorans, is capable of producing robust cytokine responses in mice (64), suggesting that recognition of BCC members is not confined to only one BCC member. The closely related, non-BCC, select agent species B. pseudomallei also induces TNF, IFN-γ, and IL-1β in C57BL/6 mice after intranasal inoculation (65, 66). Taken together, these results suggest that for Burkholderia species, the production of cytokines is robust in vivo. Yet, this is in stark contrast to our results which show that B. dolosa elicits virtually none of these cytokines inside the murine host. It appears that B. dolosa may be recognized differently by the host, as the ability of B. dolosa to form biofilms and grow is similar to those of B. cenocepacia and B. multivorans strains in vitro (7).
The difference in murine immune system recognition of B. dolosa AU0158 from what has been observed for other Burkholderia species may be partially explained by our observation that B. dolosa strains are more readily taken up by macrophages than is B. cenocepacia (Fig. 6). If B. dolosa can more readily be internalized by or invade host cells, extracellular immune recognition mechanisms such as the TLRs will be less effective at detection of this pathogen. Presumably, internal receptors such as the Nod-like and RIG-I-like receptors would still function to detect B. dolosa as they do for the other Burkholderia species. However, B. multivorans was not significantly different from B. dolosa at internalization, and yet in vivo cytokine production is still observed with this species. It should also be noted that this study may have used bacterial strains not previously tested in other studies and also includes several B. dolosa isolates from the same epidemic strain, which may skew the interpretations for the whole species; thus, caution should be used when drawing conclusions from this collection of isolates, and further study of these species as a whole is warranted.
The bacterial flagellum has long been recognized as an important virulence factor for CF pathogens. The flagellum is necessary for pathogens to move from the nasal cavity to the lower lungs (67–72). Moreover, mutations that result in a lack of flagellar motility show reduced virulence in several model systems and strains (67, 73). Studies have shown that both P. aeruginosa and B. cenocepacia can be recognized by host cells through TLR5 (27, 34, 37–39, 45, 54, 74–79), which might indicate that the flagella of both of these species play a significant role in virulence and immune clearance of these organisms. The results presented in Fig. 2 and 3 in this study support this hypothesis for P. aeruginosa, as the wild-type and ΔfliC mutant strains exhibit significantly different neutrophil recruitment and cytokine responses. However, for B. dolosa, this hypothesis may not hold true, as the wild type and ΔfliC mutant exhibit few significant differences for any of the in vivo tests that we performed. Only one of the cytokine assays from BAL fluid, that for IL-12 p40, showed a dependence on fliC in which the flagellar mutant showed a stronger induction than the wild-type strain. This suggests that the flagella may be involved in masking wild-type cells from some immune cell recognition. IL-12 p40 is one of the subunits of IL-12, which is produced by some antigen-presenting cells to activate T cell differentiation into Th1 cells. The reason for this increase is still unclear, however, as other Th1 pathway cytokines were not similarly fliC dependent. It should be noted that if B. dolosa is actively suppressing the host immune response, then this could mask any putative virulence or immunogenicity of the B. dolosa flagellin; thus, whether the B. dolosa polar flagella play a role in immune recognition or virulence in the lung needs to be investigated further.
In this study, we discovered the recruitment of an unusual population of Ly6Ghigh F4/80high cells to high levels in the BAL fluid following infection by B. dolosa (Fig. 2B). The markers on these cells were consistent with published reports of myeloid-derived suppressor cells (MDSCs) (80, 81) which have strong immunosuppressive properties, most likely through inactivating T cell responses to stimuli (82–84). While further research needs to be done to verify the nature of these cells, recruitment of MDSCs to the lungs during chronic infections by other pathogens has been demonstrated, such as for Mycobacterium tuberculosis (85), P. aeruginosa (76), and the respiratory pathogenic fungi Aspergillus fumigatus and Candida albicans (86). Thus, recruitment of these cells may be very important in lung infections and tissue pathology. It should be noted that MDSCs have been traditionally associated with advanced, long-term infections rather than the acute infections used in this study and perhaps would be unlikely to be produced after only 6 or 24 hpi as we observe for both B. dolosa and P. aeruginosa strains. Thus, future research will focus on examining the nature and prevalence of these unique cells during B. dolosa infection and infection by other BCC members and examining the presence of these cells in CF sputum isolated from humans colonized by BCC members.
In summary, the outbreak at the CF clinic at Boston Children's Hospital was due to a strain of B. dolosa that seems to have some unique properties compared to other BCC members. This provided us with the opportunity to investigate some of the mechanisms potentially underlying the virulence of B. dolosa or its ability to be transmitted from patient to patient. These results suggest that B. dolosa may have evolved a means to avoid or suppress the host immune response in vivo and may explain why this organism can remain in the host lung for long periods without causing responses that damage host lung function but also allow for a serious infection to emerge when bacterial growth is inadequately controlled. Because of the preliminary nature of the experiments conducted in this study, further work will need to focus on whether evasion and/or suppression contributes to the survival of B. dolosa in the lung.
While there is still much to be learned regarding the lifestyle of B. dolosa in the host and the molecular interactions that dictate this relationship, the information from this study may influence future work with other BCC strains, which will need to be examined both in vivo and in vitro to gain a better understanding of this group of important pathogenic species.
MATERIALS AND METHODS
Bacterial cultures and human and murine cell lines. Escherichia coli DH5α and SM10λ were used for cloning procedures. Burkholderia and Pseudomonas strains used in this study are listed in Table 1. All bacterial species were maintained in lysogeny (Luria) broth (LB) Lennox formulation. Kanamycin (50 μg/ml or 75 μg/ml), gentamicin (15 μg/ml), tetracycline (10 μg/ml or 75 μg/ml), Irgasan (25 μg/ml), or sucrose (6% or 10%) was added as indicated. Human peripheral blood mononuclear cells (PBMCs) were isolated from whole blood obtained from anonymized healthy donors. The murine macrophage cell line RAW 264.7 was cultured in Dulbecco's modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS) containing 1× penicillin-streptomycin (Sigma-Aldrich) for routine maintenance.
Burkholderia and Pseudomonas strains used in this study
B. dolosa mutant generation.An in-frame deletion mutation in the B. dolosafliC gene was generated via splicing by overlap extension (87). Briefly, 700 bp upstream and 700 bp downstream of BDAG_00084 (sequences obtained at burkholderia.com [88]) were amplified by PCR. These templates were used in a second PCR as the templates to generate fused amplicons. The resultant amplicons were cut using XmaI and EcoRI restriction enzymes and ligated into the XmaI and EcoRI sites of pEXKm5 (89, 90), transformed into DH5α competent cells, and plated on LB agar supplemented with 50 μg/ml kanamycin. Plasmids from selected transformants were analyzed by Sanger sequencing. Because B. dolosa cells can acquire resistance to kanamycin at high frequency, the tet gene was additionally inserted into pEXKm5. To do this, the tet gene was amplified from mini-CTX plasmid (91) and then cut with the SpeI restriction enzyme and ligated into the SpeI sites of pEXKm5-fliCdel to create pEXKm5Tet-fliCdel. E. coli SM10λ was transformed with pEXKm5Tet-fliCdel and selected on LB agar supplemented with 50 μg/ml kanamycin plus 10 μg/ml tetracycline. Plasmids were then transferred into B. dolosa by conjugation and plated on LB agar supplemented with 50 μg/ml kanamycin and 75 μg/ml tetracycline. Insertion of the plasmid into the chromosome was verified by PCR. Counterselection was performed by plating the transconjugants on LB medium plus 10% sucrose. Likewise, the merodiploid P. aeruginosa PAO1 ΔfliC strain (M. Merighi, personal communication) was counterselected with 6% sucrose to remove the pEXG2 plasmid backbone and create a ΔfliC mutant in P. aeruginosa PAO1. Both B. dolosa ΔfliC and P. aeruginosa ΔfliC strains were verified by PCR and lack of motility in swimming assays.
Murine models.Six- to 7-week-old C57BL/6 female mice were purchased from Charles River. Mice were anesthetized by intraperitoneal injection of ketamine-xylazine, and 20 μl of bacterial inoculum was instilled intranasally into 20 mice for each strain as described previously (92). Briefly, the different bacterial inocula (wild-type [WT] B. dolosa and P. aeruginosa and both ΔfliC mutants) were prepared from dilutions of overnight cultures grown in LB to obtain a culture of 2.5 × 108 CFU/ml. Ten microliters of this culture was instilled into each naris, resulting in an inoculum of 5 × 106 CFU/mouse. P. aeruginosa and its ΔfliC mutant were chosen as positive controls because they have been well studied based on the available literature. Three mice were left uninoculated as negative controls. Mice were observed twice daily until day 6. There was no murine mortality associated with these dosages in our study or in previously conducted pilot studies. All mouse work was conducted in accordance with the policies of the Institutional Animal Care and Use Committee at Brigham and Women's Hospital according to approved protocols.
At 1, 6, 24, and 144 h postinfection (hpi), 5 mice for each bacterial strain were sacrificed, and the lungs were harvested, weighed, and macroscopically examined. A bronchoalveolar lavage (BAL) was performed with ice-cold phosphate-buffered saline (PBS) containing protease inhibitors, and lungs were then homogenized in 1 ml sterile PBS. The BAL fluid was examined for immediate count of immune cells, fluorescence-activated cell sorting (FACS) analysis, and serial dilutions for bacterial counts and then centrifuged. The supernatant was frozen for subsequent cytokine measurements. Homogenized lungs were serially diluted and plated on LB agar for viable bacterial counts. A similar pilot experiment was performed previously with wild-type B. dolosa AU0158 and P. aeruginosa PAO1, and the data for mouse survival and bacterial cell counts were reproducible with this study (Lawrence Rhein, personal communication and unpublished data).
Flow cytometry.BAL fluid samples were harvested from the infected mice at different time points, and erythrocytes were lysed using a mouse red cell lysis kit (R&D Systems). The remaining cells were blocked with anti-CD16/CD32 (BD Pharmingen) and 15% FBS-PBS for 30 min and stained with F4/80 (BD Pharmingen), CD11b (BD Pharmingen), and Ly6G (BD Pharmingen) for 30 min at 4°C. Samples were analyzed on the FACSCalibur cytometer (BD Biosciences, Singapore) against single-cell control stains which represent samples of cells either nonstained or stained only with single antibodies.
Cytokine assays.Cytokine measurements were performed on supernatants of centrifuged BAL fluid from each mouse separately by use of a Luminex magnetic assay with a mouse cytokine 15-plex panel (Life Technologies, Grand Island, NY). The following murine cytokines were evaluated: IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12 p40, IL-12 p70, IL-17, IL-33, TNF (previously known as TNF-α [93]), IFN-γ, MCP-1, and GM-CSF.
For in vitro murine-based ELISAs, RAW 264.7 murine macrophage cells were seeded into a 96-well plate and incubated overnight. Bacterial cells were grown overnight in LB medium, diluted 1:10 in fresh medium, and harvested at mid-log phase followed by washing and resuspension in DMEM. Bacterial cells were added at various multiplicities of infection (MOIs) in quadruplicate samples, centrifuged at 800 × g for 4 min, and incubated at 37°C and 5% CO2 for 6 h. At 6 hpi, cells were lysed, and TNF and MIP-2 levels were assessed by the Ready-Set-Go! TNF-α ELISA kit (eBioscience) and mouse MIP-2/CXCL2 ELISA kit (Boster), respectively, according to the manufacturer's instructions. Data shown represent cytokine production at MOIs of ∼10 for TNF and ∼2 for MIP-2 for the experiments shown in Fig. 4 and an MOI of ∼2 for TNF experiments shown in Fig. 6.
For ELISAs on PBMCs, the buffy coat was collected from a lymphocyte separation medium (Fisher Scientific, Fair Lawn, NJ) gradient and washed in PBS (10 mM, pH 7.2) (Gibco, Grand Island, NY), and the PBMCs were plated in RPMI 1640 (Gibco) containing 10% FBS (Atlanta Biologicals, Norcross, GA). PBMCs were incubated overnight at 37°C in 5% CO2. The following day, the PBMCs were stimulated with either live or heat-killed (30 min, 65°C) and washed bacteria at an MOI of 10 bacteria to 1 host cell for 20 h. IL-1β, IL-8, and TNF release into 20-hour cell-free supernatants was measured using eBioscience ELISA Ready-SET-Go! kits (San Diego, CA) according to the manufacturer's instructions.
Internalization assays.Bacterial internalization into RAW 264.7 murine macrophage cells was measured as described above in in vitro murine-based ELISAs except that bacteria were added to macrophage cells at an average MOI of ∼2 (B. dolosa strain, average MOI, 2.67; B. multivorans species, average MOI, 1.41; B. cenocepacia strain, average MOI, 2.70; other Burkholderia species, average MOI, 1.68) and were incubated for 2 h, the medium was removed, and 100 μl of DMEM containing an antibiotic cocktail (ceftazidime and kanamycin at 1 mg/ml each, 1× Pen-Strep [Sigma-Aldrich catalog number P4333]) was added to each well. This level of antibiotics is 5- to 250-fold higher than the MICs for many B. cenocepacia (94, 95, 119, 120), B. multivorans (96), or B. pseudomallei (97–99) strains. B. dolosa AU0158 has a MIC of <64 μg/ml for both ceftazidime and kanamycin (100; data not shown), and thus, the concentrations used in these experiments are also 15-fold higher than the MICs for this strain. The antibiotic concentrations of ceftazidime (for BCC) and kanamycin (for B. pseudomallei) used in this study are higher than what is routinely used in internalization/invasion assays (7, 101–112), although we note that amikacin is usually used in conjunction with ceftazidime for BCC invasion assays but is the less effective of the two antibiotics for B. cenocepacia, B. multivorans, and B. dolosa (100). Levels of killing for all strains used in this experiment were greater than 99% under these conditions. After 2 h at 37°C in 5% CO2, RAW 264.7 cells were checked by microscopy for lysis, and there was no observable lysis in the macrophage cells by any bacterial strain used. Macrophages were then lysed by removing the medium and replacing it with DMEM containing 0.5% Triton X-100 and were flushed with a pipette. Lysis of macrophage cells was confirmed by microscopy. Each well was serially diluted and plated by drip dilutions. Average inputs (technical replicates) were used to estimate the percent invaded for each biological output replicate. For species-level comparisons, internalization data and TNF data for B. dolosa (AU0158 wild type, R25-003, AU13412, AU19373, and AU0746), B. multivorans (CF1, CF2, CGD1, and CGD2), B. cenocepacia (J2315, K56-2, AU0154, HI2424, H111, and MC0-3), and P. aeruginosa (PAO1 wild type and C3719) were used.
Statistical analysis.All statistical analysis was performed using Prism software v.5.01 (GraphPad). Microbial counts were assessed using t tests and one-way analysis of variance (ANOVA) using Tukey's pairwise comparisons as noted. Neutrophil recruitment and ELISAs from cultured cells were tested for significance through one-way ANOVA using Tukey's range test for multiple pairwise comparisons. In vivo cytokine profiles were tested using two-way ANOVA with the Bonferroni posttest to compare profiles induced by different bacterial strains and account for differences in timing in production of cytokines. When values were beneath the limit of detection for any given sample, the limit of detection was used for statistical purposes only. For cellular counts and cytokine assays, wild-type strains (both B. dolosa and P. aeruginosa) were compared only to their respective flagellin mutants or to each other. Flagellin mutants of one species were not compared to the wild-type strain of the other species, as there were two variables (strain and treatment) that changed. For the internalization assays, one-way ANOVA using the Dunnett multiple-comparison test was performed using the B. dolosa AU0158 wild-type strain as a control, and species-level comparisons were analyzed similarly, but Tukey's pairwise comparison posttest was employed to compare all pairs of species using aggregated data.
ACKNOWLEDGMENTS
We thank Massimo Merighi for the donation of the P. aeruginosafliC merodiploid strain, John LiPuma for providing the B. dolosa isolates, and Herbert Schweizer for B. pseudomallei Bp82. We also thank Stephen Lory for his expertise during the conducting of this study.
This work was supported by the William Randolph Hearst Fund and the Société de Réanimation de Langue Française (D. Roux), the University of Louisville (RIG50895, M. Weatherholt), the National Institutes of Health (NEI EY022054, M. Gadjeva; DE019826 and DE017680, D. Scott), the Slifka Family Fund (C. Gerard), and the Cystic Fibrosis Foundation (C. Gerard; PRIEBE13I0, G. P. Priebe; YODERH10F0, D. R. Yoder-Himes).
FOOTNOTES
- Received 8 September 2016.
- Returned for modification 11 November 2016.
- Accepted 20 March 2017.
- Accepted manuscript posted online 27 March 2017.
- Copyright © 2017 American Society for Microbiology.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.↵
- 107.↵
- 108.↵
- 109.↵
- 110.↵
- 111.↵
- 112.↵
- 113.
- 114.
- 115.
- 116.
- 117.
- 118.
- 119.↵
- 120.↵