Bartonella quintana lipopolysaccharide (LPS): structure and characteristics of a potent TLR4 antagonist for in-vitro and in-vivo applications

The pattern recognition receptor TLR4 is well known as a crucial receptor during infection and inflammation. Several TLR4 antagonists have been reported to inhibit the function of TLR4. Both natural occurring antagonists, lipopolysaccharide (LPS) from Gram-negative bacteria as well as synthetic compounds based on the lipid A structure of LPS have been described as potent inhibitors of TLR4. Here, we have examined the characteristics of a natural TLR4 antagonist, isolated from Bartonella quintana bacterium by elucidating its chemical primary structure. We have found that this TLR4 antagonist is actually a lipooligosaccharide (LOS) instead of a LPS, and that it acts very effective, with a high inhibitory activity against triggering by the LPS-TLR4 system in the presence of a potent TLR4 agonist (E. coli LPS). Furthermore, we demonstrate that B. quintana LPS is not inactivated by polymyxin B, a classical cyclic cationic polypeptide antibiotic that bind the lipid A part of LPS, such as E. coli LPS. Using a murine LPS/D-galactosamine endotoxaemia model we showed that treatment with B. quintana LPS could improve the survival rate significantly. Since endogenous TLR4 ligands have been associated with several inflammatory- and immune-diseases, B. quintana LPS might be a novel therapeutic strategy for TLR4-driven pathologies.


Bartonella quintana LPS does not induce production of pro-or anti-inflammatory cytokines.
The first sets of experiments were designed to investigate whether exposure to B. quintana LPS results in the induction of pro-or anti-inflammatory cytokines by human PBMCs. As shown in Fig. 1, B. quintana LPS itself does not induce the production IL-1β , TNF-α , IL-6 or IL-8. In addition, exposure for 24 h with B. quintana LPS did not result in the production or release of IL-1Ra, or IL-10 by human primary PBMCs (data not shown). However, B. quintana LPS efficiently blocks production of IL-1β , TNF-α , IL-6, IL-8 or after stimulation of human PBMCs with E. coli LPS, indicating the potency of B. quintana LPS as TLR4 antagonist. In addition, we performed dose-response experiments to examine the IC50 of B. quintana LPS for the standard dose of 10 ml E. coli LPS. Figure 2A shows that already 20 ng of B. quintana LPS reduced the IL-6 production. At higher concentrations of Human PBMCs were isolated from healthy subjects, using a standard protocol. PBMCs were pre-incubated with 100 ng/ml B. quintana LPS for 2 h, and thereafter 10 ng/ml purified E. coli LPS was added as indicated in the graph. Cytokines were determined after 24 h of culture by ELISA, IL-1β (A), TNF-α (B), IL-6 (C) and IL-8 (D). PBMCs of 6 healthy donors were examined. *P < 0.001, two-sided Mann-Whitney U test.
B. quintana LPS, 5-fold or higher, a strong suppression of the IL-6 production was stated. Figure 2B demonstrated that B. quintana LPS revealed to have an IC50 of 37.04 ng/ml at a dose of 10 ng/ml ultra pure E. coli LPS. These data confirmed that LPS isolated of B. quintana is a very potent TLR4 antagonist at a low concentration 23 .

Prolonged blocking of TLR4 by B. quintana LPS.
In order to investigate the kinetics of the bindings capacity of B. quintana LPS to TLR4 and whether removing of the TLR4 inhibitor has an effect on blockade TLR4 function, we pre-incubated human PBMCs with B. quintana LPS for 1 hour. Two different approaches were investigated. The first approach was that B. quintana LPS was continuously present during the exposure to E. coli LPS and in the second approach we removed the B. quintana LPS by thorough washing (3 times). Thereafter the PBMCs were exposed to E. coli LPS and cells were incubated for additional 24 h, 48 h or 72 h hours. At each time point the cells were microscopically checked and the supernatant was collected to measure IL-1β , IL-6, IL-8, or TNF-α . Figure 3 shows that B. quintana LPS blocks the cytokine production by E. coli LPS at least for a period of 72 h. Cytokine production by human PBMCs is reduced for more than 90% over this exposure period, when the B. quintana LPS is present in the culture medium. The second approach in which B. quintana LPS was removed after 1 hour by repeated washing, identical effects on the neutralizing capacity of B. quintana LPS were seen (Fig. 4): Almost complete inhibition of the E. coli LPS-induced cytokine production after 72 h culture in the presence of 10 ng/ml of this classical TLR4 agonist. Thus, our data show that the blocking of TLR4 by B. quintana LPS is strong and stable for at least 72 hours.  Human PBMCs were isolated from healthy subjects, using a standard protocol. PBMCs were pre-incubated with a dose-range of B. quintana LPS (10-1000 ng/ml) for 2 h. Thereafter, 10 ng/ml E. coli LPS was added and the PBMCs were cultured for another 24 h. IL-6 was determined by using ELISA (A). (B) Percentage inhibition was calculated using the IL-6 concentration of E. coli LPS exposure as 100%. PBMCs of 6 subjects were used in this experiment. IC50 was 37.04 ng/ml. *P < 0.001, two-sided Mann-Whitney U test.

Rapidity of binding of
Scientific RepoRts | 6:34221 | DOI: 10.1038/srep34221 supernatants. Figure 5A shows that 15 minutes' pre-incubation is sufficient for 100 ng/ml of B. quintana LPS to block TLR4 receptor. The rapid binding of B. quintana LPS to TLR4 indicates that this TLR4 inhibitor is efficient. B. quintana LPS neutralizes TLR4 even in the presence of E. coli LPS. Since we noted that B. quintana LPS binds very rapidly to TLR4, we investigated the neutralizing capacity of B. quintana LPS when added together with E. coli LPS or even after addition of E. coli LPS to the culture medium. Figure 5B indicates that B. quintana LPS added together with E. coli LPS blocks the TLR4 receptor for at least 72 hours. IL-6 production due to 10 ng/ml E. coli LPS was completely suppressed by 10 times excess of B. quintana LPS. Remarkably, when B. quintana LPS was added 2 hours after the PBMC were exposed to E. coli LPS, we still noted a strong suppression of the IL-6 production. Figure 5C reveals that even a dose of 100 ng/ml B. quintana LPS (10 times excess) blocks TLR4 for 72 hours, in the presence of the TLR agonist.

Polymyxin B does not inactivate the antagonistic effect of B. quintana LPS.
It is well known that polymyxin B binds to LPS from several Gram-negative microorganisms and neutralizes the activity. First, we compared B. quintana LPS with polymyxin B to analyze the difference in the neutralizing capacity. Figure 6A,B showed that B. quintana LPS is far more potent to inhibit E. coli LPS mediated TNF-α or IL-6 production by human PBMCs. A concentration of 100 ng/ml B. quintana LPS was equally potent as 10 μ g/ml polymyxin B to block 1 or 10 ng/ml of E. coli LPS. Thereafter, we investigated whether polymyxin B was able to bind and inactivate B. quintana LPS. Figure 6C demonstrated that after pre-incubation of B. quintana LPS with 10 μ g/ml polymyxin B for 2 hours, the inhibitory capacity of B. quintana LPS was still very high. As control, polymyxin B neutralized the E. coli LPS as expected.

B. quintana LPS showed efficacy in E. coli LPS-induced murine model of endotoxaemia.
To explore whether B. quintana LPS can be used for in vivo studies to neutralize TLR4, we administered B. quintana LPS in an endotoxemia model. B. quintana LPS was injected 30 minutes before a sub-lethal dose of E. coli LPS was injected in combination with D-galactosamine. One single injection of B. quintana LPS revealed to be protective as can be seen in Fig. 7A,B. In contrast to the LPS/D-galactosamine group (30% survival after 10 days),  Figure 8 showed the predicted structure of B. quintana LPS and it reveals that the TLR4 antagonist is actually a lipooligosaccharide (LOS).
To determine the aggregate structure of B. quintana LPS, small-angle X-ray scattering (SAXS) at the Hamburg synchrotron source PETRA was applied. For this, LPS at a concentration of 1 mg/50 μ l was analyzed at two temperatures 20 and 40 °C (Fig. 9). The scattering patterns are indicative of a main maximum at d = 6.67 and 6.29 nm for 20 and 40 °C, respectively, and further reflections each at d/2, d/3, and d/5, which can be assigned to a multi-lamellar aggregate structure of the LPS dispersion.

Discussion
Here we described the in-vitro and in-vivo characteristics of the natural TLR4 antagonist B. quintana LPS. LPS of B. quintana appears to be a very potent, rapidly binding TLR4 blocker of a potent TLR4 agonist (E. coli LPS). In addition, the blockade of TLR4 is prolonged: at least 72 h after exposure to human PBMCs the effect persists. Since TLR4 activation is associated with many inflammatory and autoimmune diseases, B. quintana LPS might be considered a new therapeutic strategy for TLR4-driven pathology.
Bartonella quintana is an emerging Gram-negative pathogen, which may cause endocarditis, cerebral abscess and bacillary angiomatosis usually with the absence of septic shock in humans. Nowadays, the B. quintana infection can be found in homeless people, mainly due to body lice 26 . It has been reported in the past that LPS, isolated . It is very likely that this crude preparation of B. quintana LPS activates human PBMCs in a TLR2-dependent pathway since peritoneal macrophages obtained from TLR2ko mice did not respond crude B. quintana LPS in vitro, in contrast to wild type mice (data not shown).
Here we described the structure of B. quintana LPS for the first time. It revealed that B. quintana LPS has 5 fatty acid tails, two of C12, two of C16 and one very long C26 (Fig. 8). It has been shown previously that LPS structures with 4 fatty acid chains are endotoxically inactive 28 . However, B. quintana LPS consists of 5 fatty acids and still is acts as a very potent TLR4 antagonist. This is in line with several other reports demonstrating that LPS originated from Gram-negative bacteria, such as Bradyrhizobium elkanii consists of 5 fatty acids tails and reveals to have antagonistic properties 29,30 . Studies using small angle X-ray scattering (SAXS) technology indicated that this particular LPS has a multilamellar structure. The data for the aggregate structure of LPS from B. quintana, Figure 5. Kinetics of the B. quintana LPS to block the TLR4. Human PBMCs were isolated from healthy subjects, using a standard protocol. (A) PBMCs were pre-incubated with 100 ng/ml or 1000 ng/ml B. quintana LPS for different times before E. coli LPS was added to the culture medium. After 1 hour, 45, 30 and 15 minutes PBMC were exposed to 10 ng/ml E. coli LPS for 24 h. (B) B. quintana LPS was added together with E. coli LPS, thereafter the PBMCs were cultured for additional 24 h, 48 h or 72 h. C, B. quintana LPS was added in a range before (− 2 h) and after (+ 2 h) the cells were exposed to 10 ng/ml E. coli LPS. IL-6 was determined by using ELISA. PBMCs of 4 subjects were used in this experiment. *P < 0.001, two-sided Mann-Whitney U test.  31,32 . In this kind of aggregate structure, the binding epitopes in LPS to the TLR4 receptor necessary for cell signaling are hidden, in contrast to the situation for bioactive LPS with its cubic aggregate structure 31 . However, since the cell activation is a membrane step and also a multilamellar LPS can incorporate into the immune cell membrane, cell receptors such as TLR4 may be blocked by them in this way inhibiting the cell signaling via bioactive LPS. A further observation in accordance to the chemical analysis described here should be mentioned: The periodicities in the range of 6.3 to 6.7 nm as shown in Fig. 9 are characteristic also for multi-lamellar structures of LPS from rough mutant Re and/or Rd from Salmonella minnesota, which in a previous report 33 were found to result from the addition of divalent cations such as Mg 2+ or at low water content. The final structure of the TLR4 antagonist revealed that the particular molecule is a lipooligosaccharide (LOS) and not a classical lipopolysaccharide (LPS).
Many TLR4 antagonists are based on LPS or lipid A structures obtained from non-pathogenic bacteria such as Rhodobacter capsulatus and Rhodobacter sphaeroides 34 . Compounds like E5531 (analogue of R. capsulatus lipid A) or Eritoran/E5564 (based on R. sphaeroides lipid A) were developed for the treatment of sepsis. In line with our results, Eritoran is significantly protective in animal models of sepsis 35 . In general, the TLR4 antagonists based on lipid A binds to MD-2 and thereby prevents binding of the agonist to the MD-2/TLR4 complex. This was interpreted to be due to the multilamellar aggregate structure of these antagonists which do not represent a disturbance of the membrane architecture at the site of the receptors, in contrast to the behavior of the non-lamellar aggregate structures of hexaacylated agonistic LPS 36 . Our data show that B. quintana LPS has a similar mode of action as E5564. Recently, potent low molecular inhibitors of TLR4 have been reported that interfere with the TLR4-MD2 complex formation 37 .
Polymyxin B is an antibiotic primarily used for resistant Gram-negative infections and it is derived from the bacterium Bacillus polymyxa. It has a bactericidal action against almost all Gram-negative bacilli and polymyxin binds to the cell membrane and alters its structure, making it more permeable, resulting in death of the microbe. Polymyxin B is well known for its LPS neutralizing capacity in vitro. This can be correlated with the observation, that polymyxin B converts the aggregate structure of agonistic LPS into a multilamellar form 38 . In the case of B. quintana LPS, its aggregate structure is already multilamellar, and is not changed furthermore by polymyxin B Furthermore, the fluidization observed when PMB interacts with hexaacylated endotoxins is absent 32,38 .
Here we demonstrated that B. quintana LPS binds very rapidly to the TLR4 complex, within 15 minutes the B. quintana LPS prevents activation of TLR4. Even 2 h after E. coli LPS was added to the cell cultures, B. quintana LPS was able to prevent cytokine production. This delayed antagonistic effect was only reported for one other natural TLR4 antagonist, isolated from the cyanobacterium Oscillatoria planktothrix FP1 39 . It was demonstrated that the LPS from this cyanobacterium could block DC maturation and activation even 6 h after E. coli LPS was added. In line with our report, the cyanobacterium LPS was able to prevent LPS/D-galactosamine induced lethal shock. Although, the dose needed for protection was much higher (750 μ g per mouse) than we showed in this current report (100 μ g per mouse), indicating the potency of B. quintana LPS.   Activation of TLR4 has been associated with many inflammatory diseases and infectious complications 40,41 . Therefore, many efforts have been taken to develop or identity potent inhibitors of TLR4 for in-vivo applications. Apart from lipid A-derived structures or small molecules that interfere with MD-2/TLR4 formation 42 , antibodies have been developed. However, it seems that anti-TLR4 antibodies do not bind only TLR4, but via the Fc portion also to Fcγ Rs. This dual action of these anti-TLR4 antibodies may be of importance to target inflammatory cells that express both receptors 43 .
Apart from infectious agents that can trigger TLR4 signaling, several endogenous TLR4 ligands have been described in the recent years. Many damage-associated products (DAMP) and inflammatory mediators have been linked to TLR4 for their pro-inflammatory behavior. Most of these TLR4 ligands are released after cells or tissues have been activated or damaged. A few examples of these endogenous TLR4 ligands are HMBG1, S100A7/8, fibronectin extra domain A and fetuin 7,[44][45][46][47] . Potent inhibitors of TLR4 that interfere with both microbial TLR4 ligands as well as endogenous TLR4 ligands binding to TLR4 will have significant therapeutic value. Since the most TLR4 inhibitors are based on the disruption of the TLR4/MD-2 complex, which is very specific for lipid A-derived compounds, it remains to be explored whether TLR4 antagonist can be generated that block both classes of TLR4 ligands. Of high interest, B. quintana LPS reveals to inhibit both exogenous and endogenous TLR4 as previously reported 15,48 . Further investigation is warranted to elucidate the structure of B. quintana LPS to obtain insight into the mode of action and the possibility to synthesize this potent TLR4 antagonist.

Materials and Methods
The authors confirm that all experiments were performed in accordance with relevant guidelines and regulations. Written informed consents were obtained from all donors in accordance with the ethical principles set out in the declaration of Helsinki. The ethical review board of the Radboud University Medical Center, Nijmegen, The Netherlands, approved the study in which blood were used for healthy subjects (CMO2299 2010/104). The experimental protocols for murine studies were approved by the ethic committee for animal experiments (DEC) of the Radboud University Medical Center, Nijmegen, The Netherlands.
Reagents and microorganisms. LPS (E. coli serotype O55:B5) was purchased from Sigma Chemical Co and the Bartonella quintana CIP 103739 strain was kindly provided by Dr. Tanja Schulin and grown on sheep blood agar at 37 °C in a 5% CO2 atmosphere. B. quintana LPS was extracted by a two-step extraction method, which eliminates contamination with proteins. B. quintana LPS was extracted by hot phenol-water method as described previously 23,49 . Briefly, Bartonella quintana bacteria were scraped from blood agar plates, resuspended in PBS and heat-inactivated for 60 min in 56 °C. Thereafter, heat killed bacteria were wash twice with PBS and centrifuged for 10 min. at 16,262 × g. 2 grams of bacterial mass was used to isolate the LPS. Warm water (65 °C) was added to the pellet and the solution was vortexed for 10 minutes. Thereafter, the heated phenol (65 °C) was added and the solution was stirred for 2 hours at a temperature between 63-68 °C. Thereafter, solution was centrifuged 4,435 × g for 40 min at 4 °C. The aqueous phase was collected and transferred to a dialysis cassette (3.500 MWCO) and dialyzed against demi water in a 3L glass beaker in the cold room. The distilled water was changed after 30 minutes for the first time and then after 1 hour for 3-4 times. The LPS was dialyzed for two days at 4 °C, changing demi water 3 times a day. The dialyzed LPS was extracted and stored at − 80 °C for until lyophilizing. For re-purification, 5mg Bartonella quintana LPS was added to 1 ml 0.2% TEA (Triethylamine)/0.5% Na-DOC (Natrium deoxycholate). Thereafter, 1 ml warm (60 °C) phenol:water (9:1 V/V) was added and the solution was vortexed for 5 min. After separation of the phases (5 minutes at 4 °C) the solution was centrifuged for 40 min. at 6,652 × g (4 °C). The water phase was collected and transferred to new sterile 15 ml tube. To the first phenol phase again 1 ml 0.2% TEA/0.5% Na-DOC was added and the previous steps were repeated. The second phenol phase was used to repeat the purification steps for the third time. The water phase of last 2 steps were combined with the first step. The LPS was dialyzed as described above, using the 3,500 MWCO cassette. To the dialyzed LPS drop-by-drop 1.5 ml of NaAc/EtOH (0.4 M in 100% EtOH) per each 0.5 ml of LPS was added and the solution was kept for 1 h on ice/water to let the LPS precipitate. Thereafter, the LPS was collected by centrifugation (30 min. at 16,262 × g) and washed twice with 1.5 ml cold EtOH followed by centrifugation (30 min. at 16,262 × g). Thereafter, LPS was dried on air, dissolved in PBS, aliquoted and stored by − 20 °C. E. coli LPS from Sigma was also double purified, as described above.
Isolation of PBMC and stimulation of cytokine production. Peripheral blood mononuclear cells (PBMCs) were isolated healthy individuals (written informed consent was obtained from all subjects), as described earlier 50,51 . Briefly, PBMCs were isolated by density gradient centrifugation using Ficoll-Paque PLUS (GE Healthcare) and collecting the white interphase. Next, PBMCs were washed twice in cold PBS and concentrations were adjusted to 5 × 10 6 cells/ml in RPMI-1640 Dutch Modified culture medium (RPMI supplemented with 2 mM l-glutamine, 1 mM pyruvate; GIBCO Invitrogen, Carlsbad, CA, USA). PBMC (5 × 10 5 ), in a volume of 100 μ l volume in round-bottomed 96-well plates (Greiner, Alphen a/d Rijn, The Netherlands), were incubated with either 100 μ l of culture medium (negative control) or one of the following stimuli: B. quintana LPS and E. coli LPS (10 ng/ml). Animals. C57Bl/6J mice were purchased from Charles River (Sulzfeld, Germany). For the experiments, [8][9][10][11][12] week old mice, weighing 20-25 g, were used. The animals were fed standard laboratory chow (Hope Farms, Woerden, The Netherlands) and housed under specific pathogen-free conditions. The experimental protocols were approved by the ethic committee for animal experiments (DEC) of the Radboud University Medical Center, Nijmegen, The Netherlands.
Experimental endotoxaemia model. The previously reported model of endotoxemia was used 54  Structural analysis of B. quintana LPS. The compositional analysis was done by using combined gas liquid chromatography and mass spectrometry (GLC-MS), as well as electrospray ionization mass spectrometry (ESI-MS). For the GLC-MS the B. quintana LPS was methanolyzed by 2 M HCL/CH3OH for 24 h at 85 °C and for the determination of the hexoses afterwards peracetylated or trimethylsilylated with N,O-bis(trimethylsilyl) trifluoroacetamide for the fatty acids, respectively. The resulting compounds were analyzed in a GLC on a Hewlett-Packard HP 5890 Series II chromatograph, equipped with a 30-m fused silica SPB-5 column (Supelco) using a temperature gradient of 150 °C (3 min) → 320 °C at 5 °C/min, and GLC-MS on a Hewlett-Packard HP 5989A instrument equipped with a 30-m HP-5MS column. Electrospray Ionization Fourier Transform Ion Cyclotron Resonance (ESI FT-ICR) MS was performed in using a hybrid Apex Qe FT-ICR MS instrument (Bruker Daltonics) in the negative ion mode, equipped with a 7 Tesla actively shielded magnet and an Apollo dual ion source 55 . Small-angle X-ray scattering (SAXS) measurements of LPS from Bartonella quintana were performed at the European Molecular Biology Laboratory outstation at the Hamburg synchrotron radiation facility (HASYLAB) using the double-focusing monochromator-mirror camera X33. Scattering patterns in the range of the scattering vector 0.01 < s < 1 nm −1 (s = 2 sinθ /λ , 2θ = scattering angle, λ = wavelength = 0.15 nm) were recorded at 20 and 40 °C with exposure times of 1 min using an image plate detector with online readout (MAR345; MarResearch, Norderstedt/Germany) 56 . The s-axis was calibrated with Ag-Behenate, which has a periodicity of 5.84 nm. We evaluated the diffraction by assigning the spacing ratios of the main scattering maxima to defined three-dimensional structures. For this study, the multi-lamellar structures were the most relevant, for which characteristic spacing's at the periodicity d and further reflections at d/2, d/3 etc. are found.
Statistical analysis. The data are expressed as mean ± SEM. Differences between experimental groups were tested using the two-sided Mann-Whitney U test or one-way ANOVA performed on GraphPad Prism 6.0 software (GraphPad). P values of ≤ 0.05 were considered significant. The IC50 was calculated using non-linear concentration-response curve application within GraphPad Prism 6.0.