Differential Toll-Like Receptor-Signalling of Burkholderia pseudomallei Lipopolysaccharide in Murine and Human Models

The Gram-negative bacterium Burkholderia pseudomallei causes melioidosis and is a CDC category B bioterrorism agent. Toll-like receptor (TLR)-2 impairs host defense during pulmonary B.pseudomallei infection while TLR4 only has limited impact. We investigated the role of TLRs in B.pseudomallei-lipopolysaccharide (LPS) induced inflammation. Purified B.pseudomallei-LPS activated only TLR2-transfected-HEK-cells during short stimulation but both HEK-TLR2 and HEK-TLR4-cells after 24 h. In human blood, an additive effect of TLR2 on TLR4-mediated signalling induced by B.pseudomallei-LPS was observed. In contrast, murine peritoneal macrophages recognized B.pseudomallei-LPS solely through TLR4. Intranasal inoculation of B.pseudomallei-LPS showed that both TLR4-knockout(-/-) and TLR2x4-/-, but not TLR2-/- mice, displayed diminished cytokine responses and neutrophil influx compared to wild-type controls. These data suggest that B.pseudomallei-LPS signalling occurs solely through murine TLR4, while in human models TLR2 plays an additional role, highlighting important differences between specificity of human and murine models that may have important consequences for B.pseudomallei-LPS sensing by TLRs and subsequent susceptibility to melioidosis.


Introduction
The category B bioterrorism agent, as classified by the Center for Disease Control and Prevention, Burkholderia pseudomallei is a facultative intracellular Gram-negative bacterium and the causative agent of melioidosis [1][2][3]. Melioidosis, an important cause of sepsis in Southeast Asia and Northern Australia, is characterized by pneumonia and the formation of multiple abscesses and is associated with case fatality rates of up to 40% despite appropriate antibiotic treatment [1,2]. Among the multiple putative virulence factors that have been described for B.
Aldrich) (50mg/ml) digestion step for 6 hours at room temperature. The LPS was then treated by a modified hot phenol method. Briefly, the cell paste and 90% phenol (Sigma-Aldrich) were independently heated to 70°C before adding the phenol to the cell paste at a 1:1 ratio. The mixture was vigorously stirred by hand whilst maintaining 70°C. This mixture was dialysed against water until no phenol remained, after which it was lyophilised. A further round of Proteinase K, DNase I and RNase A digestions preluded a final ultracentrifugation step at 100,000 x g for 3 hours. Pellets were solubilised with distilled water before being dialysed into further distilled water. The final product was purified by SDS-PAGE on a 12.5% separating gel and visualised by silver staining.
Protein contamination in the LPS preparation was tested by silver staining, Coomassie blue staining (Bio-Rad, Hercules, CA) and BCA assay (Thermo Fisher Scientific, Rockford, IL). In order to further confirm the purity of the LPS, different concentrations of LPS-binding cationic antimicrobial polymyxin B (PMB) [25,26] were used in a stimulation assay using the murine alveolar macrophage cell line MH-S (American Type Culture Collection, Rockville, MD).

Isolation and mass spectrometric analysis of lipid A
Lipid A components of B.pseudomallei-LPS of strains 1026b and K96243 [27] were obtained by treatment of LPS as described [20]. Matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry (MALDI-TOF-MS) of the lipid A component of B.pseudomallei-LPS of strains 1026b and K96243 was performed in negative-ion mode on a Bruker UltraFlex MAL-DI-TOF/TOF instrument using 2,4,6-trihydroxyacetophenon as a matrix. Sample preparation and analysis were conducted as described [28].
In additional experiments, lipoprotein lipase (LPL) (Sigma-Aldrich) was added to HEKcells during LPS stimulation. The murine alveolar macrophage cell line MH-S was grown in RPMI 1640 medium supplemented with 10% FCS, 1% L-glutamine, 1% PenStrep and 0.05 mM 2-mercaptoethanol (Sigma-Aldrich) and seeded at 1.5 x10 6 /ml, followed by stimulation with the stimuli as described above in the presence or absence of LPS-binding polymyxin B (25-100 ug/ml; Sigma-Aldrich). Supernatants were harvested at 6 and 24 h and stored at -20°C until assayed for murine tumor necrosis factor (TNF)-α. For selected experiments whole blood derived from healthy volunteers was collected; for these studies inhibitory anti-human TLR2 (2500 ng/ml) and TLR4 (1000 ng/ml) antibodies were used (InvivoGen) as described [29]. Dose-response curves were performed prior to electing the appropriate dose of each stimulus and for the different cells used (data not shown).
Blood and peritoneal macrophages were harvested from WT, TLR2 -/and TLR4 -/mice. Blood was then transferred to V-bottom 96-well plates (Greiner Bio-One, Frickenhausen, Germany; 100 ul/well) and directly stimulated with 10 7 CFU/ml of heat-killed B. pseudomallei 1026b or its O-antigen lacking mutant SRM117 [9,11], LPS of B. pseudomallei 1026b (100 ng/ ml), LPS of E. coli (100 ng/ml), PAM3CSK4 (100 ng/ml) or RPMI 1640 for 6 h or 24 h. Peritoneal macrophages were harvested by 5 ml peritoneal lavage with sterile PBS. Cells were resuspended in RPMI 1640 enriched with 1% L-glutamine, 1% penicillin/streptomycin, and 10% FCS and seeded in a 96-well plate at a concentration of 0.5 x 10 6 /ml and. The next day the cells were washed with RPMI 1640 to remove non-adherent cells and incubated with the same stimuli as used for whole blood; a MOI of 50 of heat-killed B. pseudomallei 1026b and SRM117 was used. Dose-response curves for these stimuli were performed prior to these experiments (data not shown).

LPS-induced lung inflammation
Lung inflammation was induced in mice as described previously [31]. In short, mice were lightly anesthetized by inhalation of isoflurane, after which 50 μl of LPS of B.pseudomallei 1026b (200 ug/ml) dissolved in PBS was administered intranasally. This dose of 10 μg LPS intranasally was based on a previous dose-finding experiment (data not shown). After 6 h, mice were sacrificed by bleeding from the vena cava inferior. Bronchoalveolar lavage with two 0.5 ml aliquots of sterile saline was performed as described [31]. The 6 h time point was chosen as the ideal time to simultaneously assess cyto-and chemokine alterations and leukocyte recruitment [31]. BAL fluid (BALF) differential cell counts were performed on Giemsa-stained cytospin preparations. After centrifuging the BALF samples at 650x g for 10 minutes, supernatants were harvested and stored at -20°C for cytokine analysis.

Assays
Human IL-8 levels were measured in the supernatants of HEK-cell supernatant by ELISA (Biosource, Etten-Leur, the Netherlands). Release of TNF-α, interleukin-6 (IL-6) and cytokineinduced neutrophil chemo-attractant (KC) was measured in supernatant and BALF by ELISA (R&D Systems, Minneapolis, MN) in accordance with the manufacturer's recommendations. Cell counts in BALF were determined using a Beckman Coulter Counter (Miami, United States).

Statistical analysis
Data were analyzed using GraphPad Prism for Windows (version 5.01; GraphPad Software). Data were analyzed by Mann-Whitney U test. When > 2 groups were treated with the same stimuli, a Kruskall-Wallis test was performed first. Data are expressed as mean ± SEM values. Pvalues < 0.05 were considered statistically significant.

Ethics and safety statement
The Animal Care and Use of Committee of the University of Amsterdam approved all experiments, which were conducted according to national guidelines (DIX 102324/25). Mice were sacrificed under ketamine (Eurovet, Bladel, the Netherlands) and medetomine (Pfizer, Capelle, Netherlands) or isoflurane anesthesia, and all efforts were made to minimize suffering. In addition, the biosafety committee of the University of Amsterdam approved all in vitro experiments.

Extraction and purity of B.pseudomallei-LPS
To ensure that the LPS of B.pseudomallei 1026b was not degraded after extraction, 5 μg of LPS (0.5 mg/ml) was run on a 12.5% gel and silver-stained, which demonstrated a typical LPS pattern [32,33] (Fig 1A) and no protein contamination. In addition, the LPS was assessed for protein contamination by an enhanced BCA assay and Coomassie Blue staining ( Fig 1B) which were negative (data not shown). Even though B.pseudomallei itself is not or just slightly growth-inhibited by its presence [34], PMB was able to completely inhibit the inflammatory response induced by 50 μg/ml of purified B.pseudomallei-LPS in the murine alveolar cell line MH-S ( Fig 1C). Of note, PMB by itself did not have an effect on MH-S cell viability, neither did it have any effect of the inflammatory response caused by PAM3CSK4, a synthetic TLR2-ligand (data not shown). These results underscore that there are no potential immune stimulatory contaminating agents present in the LPS preparation used.
Since it is known that the structure of the lipid A component of LPS is of importance recognition by TLRs [35], the isolated lipid A components of B.pseudomallei strains 1026b and K96243 were structurally analyzed by MALDI TOF mass spectrometry. As expected, the analyses revealed a high degree of heterogeneity of these isolates, which gave spectra that were very similar (S1 Fig). We confirmed that the lipid A species in the LPS of both B. pseudomallei K96234 and B. pseudomallei 1026b were predominantly tetra-acylated with a small proportion that were penta-acylated (Table 1, S1 and S2 Figs), which is in line with previous studies [16]  In order to investigate via which TLR the LPS of B.pseudomallei signals, we first stimulated HEK-293 cells stably transfected with either human CD14-TLR2 or CD14-TLR4/MD2. In line with our previous experiments we confirmed that after 6 h of stimulation B.pseudomallei-LPS signals via TLR2 and not TLR4 in this cell system [21] (Fig 2A). However, after 24 hours of stimulation the LPS of B.pseudomallei clearly interacted with both TLR2 and TLR4 (Fig 2B).
To examine the signalling more carefully, we next performed the reverse experiment using human whole blood derived from healthy volunteers in combination with inhibitory antihuman TLR2 or TLR4 antibodies. After pre-incubation with an anti-TLR2 or anti-TLR4 antibody, we stimulated whole blood with B.pseudomallei-LPS (100 ng/ml) for 6 h. In concordance with the experiments with HEK-cells, these experiments demonstrated that B.pseudomallei-LPS signals mainly via TLR4 in a human in vitro system ( Fig 2C). Interestingly, anti-TLR2 antibody tended to reduce the inflammatory response, but this was not significant (p = 0.09). However, combining the anti-TLR2/4 antibodies did demonstrate an additional effect of TLR2 blockade. With regard to the inflammatory response induced by the whole bacterium, we did not observe any inhibitory effect on the inflammatory response induced by B.pseudomallei when these antibodies were used (data not shown). In all experiments, E.coli-LPS, used as positive control, was confirmed to signal only through TLR4. In addition, these results showed that the extent of cytokine responses induced by B.pseudomallei-LPS was comparable to those induced by E.coli-LPS.
TLR4 -but not TLR2-deficiency attenuates the murine cellular inflammatory response to B.pseudomallei-LPS After demonstrating the important role of both TLR2 and TLR4 in the signalling of B.pseudomallei-LPS in human cell models, we next examined the TLR-signalling in an ex vivo murine model. Whole blood and primary peritoneal macrophages harvested from WT, TLR2 -/and TLR4 -/mice were collected and stimulated with B.pseudomallei-LPS, heat-killed B.pseudomallei and its mutant SRM117 that lacks the O-antigen for 6 and 24 h. In this model, B.pseudomallei-LPS behaved in the same way as E. coli-LPS: TLR4 -/but not TLR2 -/murine whole blood ( Fig 3A) and peritoneal macrophages (Fig 3B) showed a completely abolished TNF-α response  upon stimulation. Interestingly, reduced TNF-α release was observed in whole blood derived from TLR4 -/but not TLR2 -/mice in response to both heat-killed B.pseudomallei and the SRM117 mutant ( Fig 3A). Surprisingly, stimulation of TLR2 -/peritoneal macrophages with heat-killed B.pseudomallei demonstrated an equally hampered response as TLR4 deficient macrophages (Fig 3B). An even more remarkable finding was that this TLR-2-mediated response was O-antigen-dependent since it differed from the response to stimulation with mutant strain SRM117. Despite being less virulent in animal models [9], B. pseudomallei mutant SRM117 induced a similar inflammatory response in WT blood and macrophages. In a separate set of experiments whole blood and peritoneal macrophages derived from TLR2/4 double knock-out mice were exposed to B.pseudomallei-LPS; however no additional effect of TLR2 or double TLR deficiency was observed in the inflammatory response (data not shown). Taken together, this set of experiments showed that-in sharp contrast to the human studies-B.pseudomallei-LPS interacts with TLR4 and not TLR2 in ex vivo murine models.

TLR4 acts as the main TLR receptor for neutrophil recruitment during acute murine lung inflammation induced by B.pseudomallei-LPS
We have previously shown that intranasal inoculation with B.pseudomallei causes significant and rapid inflammation and neutrophil, but not monocyte, recruitment toward the lung [36]. To evaluate the role of B.pseudomallei-LPS during pulmonary inflammation, we intranasally inoculated WT, TLR2 -/-, TLR4 -/and TLR2x4 -/mice with B.pseudomallei-LPS to examine polymorphonuclear leukocyte (PMN) influx. BALF was obtained 6 h after LPS administration since this time point is representative for both neutrophil influx and local cytokine/chemokine release. No differences in pulmonary neutrophil influx were seen between WT and TLR2 -/mice (Fig 4). However, when compared to WT mice, B.pseudomallei LPS-induced PMN influx was equally and strongly reduced in TLR4 -/and TLR2x4 -/mice (Fig 4).

B.pseudomallei-LPS induced pulmonary inflammation in mice is TLR4 dependent
Intranasal administration of B.pseudomallei-LPS resulted in a steep increase in the concentrations of TNF-α and IL-6 in BALF (Fig 5A and 5B). TLR4, but not TLR2, did profoundly influence this characteristic inflammatory response. B.pseudomallei-LPS induced TNF-α and IL-6 levels determined 6 h after administration were found to be significantly lower in TLR4 -/and TLR2x4 -/mice, but not in TLR2 -/mice when compared to WT mice (Fig 5A and 5B). Of note, TLR2 deficiency resulted in increased IL-6 levels when compared to WT; however no additional effect of TLR2 was seen in the TLR2x4 -/mice when compared to the TLR4 -/mice. In correlation with the strongly diminished neutrophil influx in TLR4 -/and TLR2x4 -/mice, an equally sharp decline in the production of the neutrophil attracting chemokine KC in BALF was seen (Fig 5C). These findings further confirm our conclusion that TLR4 acts as the main in vivo receptor for B.pseudomallei-LPS in mice.

Discussion
LPS recognition by the host is of crucial importance for the initiation of a swift host immune response to Gram-negative bacteria [37]. In the case of melioidosis, a main cause of community-acquired Gram-negative sepsis in Southeast Asia and Northern Australia however, essential insights on both the role of LPS as a putative virulence factor and its recognition by the innate immune system remain ill defined [1,2]. We now studied the recognition and contribution of TLR2 and TLR4 in B.pseudomallei-LPS induced inflammation, linking in vitro experiments using human and murine cells with mouse studies involving LPS-induced pulmonary inflammation. Using these models we found that the LPS of B.pseudomallei induces a strong inflammatory response that is comparable to the response elicited by E.coli-LPS, supporting previous data [38]. Moreover, we established that TLR4 is the main receptor for LPS of B.pseudomallei in murine in vitro and in vivo models. Remarkably, in human in vitro models TLR2 has an additional role in the recognition of B.pseudomallei-LPS. Taken together these results further characterize B.pseudomallei-LPS as a key driver of the innate immune response in melioidosis. In addition, these findings underscore important species differences in the  specificity of LPS-TLR interactions, which means caution should be applied when generalizations are made when extrapolating from murine disease models to humans [39].
TLRs and CD14 are key pattern-recognition receptors of the innate immune system that are capable of recognizing 'pathogen-associated molecular patterns' (PAMPs) and form a crucial link between innate and adaptive immunity. The key role of TLRs in the pathogenesis of melioidosis is underscored by the finding that mice deficient in myeloid-differentiation-primary-response-gene-88 (MyD88), which is the key signalling adaptor protein for all TLRs, except for TLR3, show a strongly accelerated lethality upon intranasal infection with B.pseudomallei [40]. TLR4 has been implicated as the canonical LPS-receptor for Gram-negative bacteria, recognizing the lipid A part of LPS [41]. Efficient LPS signalling requires both the LPSbinding protein (LBP) and the surface-anchored receptor CD14, which has demonstrated to be involved in the recognition of B.pseudomallei [42]-and the extracellular protein MD2. Furthermore, the glycosylation status of LPS influences its signalling; smooth LPS, which contains the complete O-antigen, requires CD14 for its detection, whereas the O-antigen lacking rough LPS and lipid A do not [43]. B.pseudomallei-LPS seems to be largely conserved across this species. LPS profiling of >700 B.pseudomallei isolates using SDS-PAGE showed that most isolates had a 'typical' ladder pattern of extracted LPS while a minority had an 'atypical' or rough pattern [44]. B.pseudomallei-LPS diversity may correlate with differential immunopathogenicity and virulence, such as biofilm formation among B.pseudomallei strains [32,45]. Observed differences in TLR-signalling of LPS might further be explained by different three-dimensional shapes of the lipid A part [46], which are due to a different disaccharide backbone, the presence or absence of carboxyl or phosphate groups, numbers and (non)symmetrical position of acylgroups [47]. Conical shaped lipid A such as that of E.coli signals via TLR4, while cylindrical shaped lipid A has been described to signal via TLR2 [46]. B.pseudomallei's lipid A is known to be penta-acylated, while the lipid A part of most other virulent Gram-negative bacteria often possesses 6 acyl groups, and contains a fatty acyl mutation (C 14:0 (2-OH) [16]. In our study we found similar acylation patterns for the lipid A part of both B. pseudomallei 1026b and K96243, making it less likely that TLR signalling of LPS differs between these B.pseudomallei strains. A role for TLR2 in the recognition of B.pseudomallei-LPS would not be unique: it is known that the LPS of Legionella pneumophila, Leptospira interrogans and Porphyromonas gingivalis can signal via TLR2 [8,18,19]. Remarkably, the TLR2-mediated response was O-antigen-dependent since it differed from the response to stimulation with the O antigen mutant strain SRM117. One could hypothesize that differences in the recognition of smooth LPS (which includes O-antigen) and rough LPS (which lacks O-antigen) may play a role herein. Another explanation might be that O-antigen deficiency leads to changes in the expression of other B. pseudomallei associated PAMPs and thus TLR ligands.
We now demonstrate that, depending on the model used, B.pseudomallei-LPS recognition occurs solely via TLR4 (murine models) or via a combination of TLR2 and -4 (human models. Our results underscore previous studies that have demonstrated important differences in TLRsignalling in humans and mice [48,49]. Moreover, in a study on Pseudomonas aeruginosa infection and cystic fibrosis (CF) it has been shown that P. aeruginosa adapts itself to the CF airway by synthesizing both hexa-and penta-acylated LPS. Human TLR4 can discriminate between these LPS-forms, whereas murine TLR4 cannot [50]. This is thought to be mediated by an 82-aa region of human TLR4 that varies highly across species [50]. Using mass spectrometry, we confirmed that lipid A of B.pseudomallei has a tetra-and penta-acylated saccharide backbone, which could be the reason the TLR4 response to B.pseudomallei-LPS is much stronger in mice than in humans. In this respect it should be mentioned that TLR4-region genetic variants in humans are associated with susceptibility to melioidosis [51]. Of note however, this effect could also be attributed to a non-LPS TLR4 effect, since this promiscuous receptor not only recognizes LPS but also recognizes numerous endogenous danger signals called 'dangerassociated molecular patterns' (DAMPs) [52].
Pneumonia is the most common manifestation of acute melioidosis [1,53]. Our study is, to our knowledge, the first to provide insights into the ability of B.pseudomallei-LPS to elicit broncho-alveolar inflammation. We demonstrated that B.pseudomallei-LPS elicits a profound inflammatory response, comparable to that induced by instillation of an equal dose of E.coli-LPS [31]. B.pseudomallei-LPS induced the release of TNF-α, IL-6 and KC into BALF together with a marked influx of neutrophils into the alveolar space. Together with our in vitro data that show that the extent of cytokine responses induced by B.pseudomallei-LPS was comparable to those induced by E.coli-LPS, these results challenge current views that B.pseudomallei-LPS is only weakly inflammatory, which is in line with recent reports [38]. This effect was dependent on TLR4. TLR2 deficiency did not play a role in murine LPS recognition: either TLR2 deficiency on its own or TLR2/TLR4 double deficiency did not significantly influence these inflammatory parameters.
Our study has several limitations. Potential contamination with TLR2-stimulating agents such as lipopeptides is a well-known problem during LPS purification [54]. The LPS of B.pseudomallei used in this study was obtained by a modified aequous hot-phenol re-extraction method and we have confirmed its integrity and purity by both silver and Coomassie staining and a BCA protein assay. To further test potential contamination, in supplementary experiments, we studied whether digestion of B.pseudomallei LPS with lipoprotein lipase (LPL) could abolish the response in HEK-TLR2 cells; however LPL treatment did not impact on TLR-signalling by B.pseudomallei-LPS (S3 Fig). No TLR2 dependent response was observed in the murine in vitro stimulation experiments. In addition, the LPS-binding antimicrobial PMB was able to completely inhibit the inflammatory response induced by the purified B.pseudomallei-LPS. It has been demonstrated that TLR signalling might differ in established cell lines and primary cells [55]; this can be of particular relevance for our observation of B.pseudomallei-LPS-TLR2 signalling when using a TLR overexpressing system of HEK-cells. However, when we performed the reverse experiment on human whole blood with the use of blocking TLR antibodies we still observed a significant effect of TLR2 on the signalling of B.pseudomallei-LPS. Lastly, it should be noted that TLR expression is known to vary considerably across mammalian species [49]. Murine and human inflammatory responses towards LPS stimulation can be markedly different [39]. Our study now further accentuates the implications of these findings.
In summary, we here demonstrate that LPS of B.pseudomallei is capable of inducing a strong inflammatory response in both human and murine models underscoring an important role for LPS as a virulence factor of B.pseudomallei. We conclude that the main receptor of B.pseudomallei-LPS is TLR4. However, in human ex vivo models there is an additional role of TLR2 in the signalling of B.pseudomallei-LPS. These data emphasize species differences in TLR-signalling in men and mice and clearly imply that extrapolation of murine data to humans should be carried out with great caution. Future research on these species discrepancies is therefore needed.
Supporting Information S1 Fig. Negative-ion mode MALDI-TOF mass spectra of lipid A components of B. pseudomallei 1026b (A) and K96243 (B). Based on peaks in these spectra and comparison to literature data [16] acylation patterns of lipid A were proposed ( Table 1). The spectra of the observed negative-ion peaks were similar for both strains. (m/z) = mass-to-charge ratio.