This study was carried out in strict accordance with the European Commission legislation for the protection of animals used for scientific purposes (2010/63/EU). The protocol for the treatment of the animals was approved by the Comité de Ética de la Dirección General del Medio Ambiente de la Comunidad de Madrid, Spain (permits PROEX 21/14 and PROEX 148/15). Animals had unlimited access to food and water. They were euthanized in a CO₂ chamber, and all efforts were made to minimize their suffering.

TLR4 ligand LPS from E. coli O111:B4 (Sigma) and TLR2/6 ligand FSL-1 (InvivoGen) were resuspended in sterile 1× PBS. B. abortus and O. anthropi LMG 3331 were obtained from Julian Velasco. O. intermedium LGM 3306 LPS was purified and characterized as previously described (17, 18). The purity of those compounds was assessed by mass spectrometry with a purity level higher than 98%.

The murine macrophage cell line J744 was cultured in RPMI 1640 medium (Gibco) (2 mM L-glutamine, antibiotics), supplemented with 5% FBS (Merck). Cells were cultured in 12-well plates and serum deprived for 16 h prior to ligand stimulation. HEK293T, HEK TLR2, and HEK TLR4/MD-2/CD14 were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) (2 mM glutamine, 2 mM Aminoacids non-essential (AANE) 1% penicillin-streptomycin) with 5% fetal bovine serum (FBS) and incubated overnight before use.

C57BL/6 wild-type (WT) and TLR2 and TLR4 knockout (KO) mice were obtained from S. Akira and maintained in the animal facilities of the Centro de Biologia Molecular Severo Ochoa in Universidad Autonoma de Madrid. Thioglycolate-elicited peritoneal macrophages (PMs) were isolated from 6–8-week-old pathogen-free mice. Cells cultured in RPMI 1640-supplemented 2 mM L-glutamine, antibiotics (Gibco) and with 5% FBS and seeded into 12- or 6-well plates at a density of 1 × 10⁶ cells/well. Cells were allowed to adhere for 2 h, and then the medium was changed to remove non-adherent cells and incubated overnight before use.

Total cellular RNA was isolated using NZyol Reagent (NZYTech). cDNA was prepared by reverse transcription (GoTaq 2-Step RT-qPCR System, Promega) and amplified by PCR using SYBR^® Green PCR Master Mix and ABI Prism 7900HT sequence detection system (Applied Biosystems). Primers used for qPCR analysis of tumor necrosis factor (TNF)-α, interleukin (IL)-6, IL-10, and IL-12 are listed in Supplementary Table S1 . The 2−^ΔΔCt method was applied to analyze the relative changes in expression profiling, and all quantifications were normalized to the housekeeping genes, Glyceraldehyde 3-phosphate dehydrogenase (GADPH) and RPL13A.

Cytokine concentration was determined for IL-10, TNF-α, IL-6, and IL2-p40 using DuoSet ELISA kit from R&D Systems according to the manufacturer’s protocol.

Cells were lysed in ice-cold lysis buffer [50 mM Tris pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM ethylenediaminetetraacetic acid (EDTA), 10% glycerol], phosphatase and protease inhibitors (Roche). Equal protein amount (20 µg) from each cell lysate was separated on sodium dodecyl sulfate (SDS) 10% polyacrylamide gel and transferred to a nitrocellulose membrane (Bio-Rad). Membranes were blocked with 3% bovine serum albumin (BSA) for 1 h and incubated with antibodies against IκBα (9242), p-p38 (Thr¹⁸⁰/Tyr¹⁸²) (9211), total p38 (9212), p-ERK1/2 (Thr²⁰²/Tyr²⁰⁴) (9101), and total ERK 1/2 (9102) from Cell Signaling and β-actin (sc-47778) from Santa Cruz Biotechnology. The membranes were then incubated with the respective horseradish peroxidase (HRP)-conjugated secondary antibody for 1 h and developed using enhanced chemiluminescence (ECL) substrate (BioRad).

Stable immortalized HEK 293T, HEK 293-hTLR2/6, and HEK 293/hTLR4A-MD-2-CD14 cells (InvivoGen, San Diego, CA, USA) were plated at 3 × 10⁶ cells in 6-well plates growing at 37°C in DMEM culture medium supplemented with 5% FBS, 2 mM L-glutamine, 100 U/ml gentamycin, 0.01% pyruvate, and 0.4 mM non-essential amino acids. Then, 24 h later, cells were cotransfected with the pNF3ConA Luc [nuclear factor (NF)-κB] Firefly reporter construct and the thymidine kinase promoter-Renilla reporter plasmid (100:1 ratio) using Metafectene PRO (Biontex, Plannegg, Germany) (19). Transfection medium was changed after 24 h, and cells were seeded at 1.3 × 10⁴ cells per well in 96-well plates. Then, 24 h later, ligands were added. Activities of Firefly and Renilla luciferases were measured 24 h after using TwinLite Firefly and Renilla Luciferase Reporter Gene Assay System (PerkinElmer) in Fluo Star Optima (BMG) plate reader (three replicates per condition). All ratios were compared with the control condition (non-stimulated cells). We used FSL-1 (InvivoGen), TNF-α (InvivoGen), and LPS purified from E. coli (Sigma) and O. intermedium LPS.

The full structure of E. coli LPS was extracted from the (hTLR4/MD-2/E. coli)₂ complex retrieved from the Protein Data Bank (PDB ID 3FXI) (www.rcsb.org; last accessed July 12, 2021) (incluir la REF). The three-dimensional (3D) structure of O. intermedium LPS was constructed using PyMOL (20). O. intermedium LPS structure was then divided into fragments: Ocore (fragment containing the saccharide core of the LPS), O28 (fragment with the two FA Ocore chains containing 12 and 16 carbons and the third 28-carbon acyl chain attached to C16), O19 (fragment containing the two lipid chains with 14 and 18 carbons and the third 19-carbon lipid chain attached to C18), and lipid A. Cuts were introduced as hydrolysis of ether bonds, thus a hydrogen atom was added to the oxygen atom of the fragment. The structures went through a restrained minimization procedure with Maestro using the OPLS3 force field (21). The X-ray crystal structures of hTLR2/6 dimer (PDB ID 2Z7X) and hTLR4 (PDB ID 3FXI) were retrieved from the PDB. PyMOL was used to superimpose TLR2 to one monomer of the TLR4/TLR4 dimer. The final structure comprises TLR4/MD-2 complex dimerized with TLR2, the remaining atoms were deleted. The structures went through a minimization with Amber14 (22) under the ff14SB (23) to minimize the newly constructed TLR4/TLR2 interface.

Gasteiger charges were computed with AutoDockTools 1.5.6 (24) to ligands and receptors, and non-polar hydrogens of the receptors were merged. The structure of the receptors was kept rigid, whereas all ligands were set to be partially flexible by providing rotational freedom to some appropriately selected dihedral angles. The docking calculations were performed with Autodock Vina (25). Each ligand was docked into different regions of hTLR2 monomer and hTLR2/TLR4/MD-2 heterodimer complex by placing different boxes at different regions. For all the docking boxes, the point spacing was set as 1 Å. For hTLR2, a docking was performed with the box covering the TLR2 pocket with a center placed equidistant to the center of mass of residues Phe284, Leu282, and Asn274 and a box size of 40, 36, 30 (X, Y, Z). For hTLR2/TLR4/MD-2 heterodimer, a docking was performed in the region behind hTLR2 containing the N-terminal and central subdomains, and the center of the box was located equidistant to the center of mass of residues Arg321, His318, and Asn290 with a size of 37, 50, 50 (X, Y, Z). Another box was set to cover the interface of TLR2 and TLR4 receptor centered equidistant to the center of mass of Lys324 (TLR4), Tyr376 (TLR2), and Asn379 (TLR2) and of size 37, 50, 50 (X, Y, Z). The docking was also performed inside the MD-2 pocket, setting a box where the center of coordinates was equidistant to the center of mass of Phe119 (TLR4), Ile52 (MD-2), and Ser57 (MD-2) (X, Y, Z) and size 36, 38, 50 (X, Y, Z). A docking covering the MD-2 pocket of the hTLR4/MD-2 homodimer was also performed. A box of 60 Å in size was defined and centered equidistant to the center of mass of residues Arg90 (MD-2), Arg96 (MD-2), and Arg264 (TLR4). The determination of the best result from each docking was based on the predicted binding energy and the mode of interaction of the ligands. Docking poses were analyzed, and structural images were generated in PyMol.

HEK293T cells were seeded in 6-well plates (3 × 10⁵ cells/well) and incubated overnight in DMEM with 5% FBS without antibiotics. On the next day, cells were transiently transfected with 0.5 µg of plasmid mixture of pcDNA3-hTLR4-YFP and pcDNA3-hTLR2-CFP (Addgene) or transfected with each plasmid (molar ratio 1:1) using Metafectene Pro (Biontex), according to the manufacturer’s protocol. Cells were also transfected with the plasmid pCMV-ECFP-EYFP (Addgene) that expresses the tandem CFP : YFP construct, which served as a positive control for FRET. A day after transfection, cells were cultured in 8-well glass bottom chambers and incubated overnight in phenol red-free DMEM with 5% FBS and 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES). The following day, cells were stimulated with O. intermedium LPS for 30 min. FRET between TLR2 and TLR4 proteins was calculated by measuring sensitized emission fluorescence of CFP-YFP pair using NIS Elements 4.40 software on the Nikon Eclipse Ti-E confocal microscope. FRET image acquisition and processing details are provided in the Supplementary Material .

Analysis was performed using GraphPad Prism 5 software. Quantitative results are expressed as means ± SEM or mean ± SD. Statistical analysis between two groups was performed using two-tailed unpaired Student’s t-test. Two or more groups were compared with one-way ANOVA followed by Bonferroni multiple comparisons. A p-value <0.05 was considered statistically significant.

We compared the production of inflammatory cytokines in J774 macrophages treated with E. coli and O. intermedium LPS at different doses. The cytokine induction was dose-dependent for both LPSs, but reaching a plateau for E. coli LPS at 0.1 μg/ml, whereas O. intermedium LPS required much higher doses. However, levels of proinflammatory cytokines TNF-α and IL-6 induced by O. intermedium were much lower than those induced by E. coli LPS, even at 100-fold higher doses ( Figure 1A ) . IL-12 production was also induced at lower levels by O. intermedium. However, in contrast, O. intermedium triggered higher IL-10 expression levels compared to E. coli LPS. In addition, NF-κB and mitogen-activated protein kinase (MAPK) activation induced by O. intermedium LPS increases in a dose-dependent manner; however, the response was always much weaker compared to that of E. coli LPS ( Figure 1B ). Moreover, these results indicate that the atypical LPS from O. intermedium is a weaker proinflammatory inducer compared to enterobacterial LPS.

Atypical LPSs have been proposed to induce inflammatory responses through TLR2 rather than TLR4 (5, 6). Therefore, we investigated the TLR4 and TLR2 dependency for O. intermedium LPS-induced inflammatory response. For this, we first tested their stimulatory effects on WT and TLR2 and TLR4 KO peritoneal macrophages with two different doses of this LPS and compared with two controls: E. coli LPS (TLR4 ligand) and FSL-1 (TLR2 ligand). WT macrophages stimulated with O. intermedium LPS showed a weaker induction of cytokines compared to E. coli LPS. O. intermedium LPS only achieved cytokine levels similar to E. coli LPS at a 100-fold dose of 10 μg/ml ( Figure 2A ). Moreover, O. intermedium LPS showed dependency on TLR4 and TLR2 to induce cytokine production. IL-12, IL-6, and TNF-α were dependent on TLR2 and mostly on TLR4 signaling, whereas IL-10 induction was mostly TLR2-dependent. Next, both LPSs were tested in HEK transfected cells expressing TLR2/TLR6 or TLR4/CD14/MD-2 using NF-κB reporter assays. As expected, E. coli LPS only signals through TLR4, whereas we found that O. intermedium LPS could signal via TLR2 or TLR4 ( Figure 2B ). Nonetheless, as in previous proinflammatory assays, O. intermedium LPS was 100-fold less active than E. coli LPS in TLR4/CD14/MD-2 HEK transfected cells ( Supplementary Figure S1 ).

Moreover, we found that co-activation of macrophages by E. coli LPS, a specificTLR4 ligand, plus specific TLR2 ligands as Pam3CSK4 or FSL-1 induces an additive effect on NF-κB and p38 signaling pathways, observed by an earlier p38 phosphorylation and IκBα degradation ( Supplementary Figure S2 ). The same effect on NF-κB activation was observed when TLR2 ligands were combined with O. intermedium LPS ( Supplementary Figure S3 ). This pattern was not observed with O. intermedium LPS, therefore supporting that this LPS is free of TLR2 contaminants that may be responsible for TLR2 signal. Overall, and more importantly, these results imply that O. intermedium LPS is a TLR4/TLR2-dependent agonist.

E. coli LPS recognition by hTLR4/MD-2 complex was previously described (4) where the inner core composed of KDO I, KDO II, and heptoses established important interactions with MD-2 (e.g., Ser118, Lys122) and TLR4 residues (e.g., Lys341, Tyr296, Asp294). Therefore, O. intermedium LPS was docked in hTLR4/MD-2 homodimer and superimposed with the E. coli LPS to compare possible binding orientations and the interactions of both LPSs with the two TLR4 monomers. The obtained binding poses showed that O. intermedium LPS FA chains were buried in the MD-2 pocket; however, the core saccharide dove toward TLR4*–TLR4 interface ( Figure 3A ). This is more evident when the docked pose of O. intermedium LPS and E. coli LPS are superimposed in the hTLR4/MD-2 homodimer ( Figure 3B ). E. coli LPS core established determinant interactions with TLR4 residues (e.g., Tyr296, Asp294, and Arg264) as well as with TLR4 (Gln436) ( Figure 3C ), whereas these interactions were not detected in the case of O. intermedium LPS ( Figure 3D ). These results indicate that, in contrast to E. coli LPS, O. intermedium LPS does not favor the formation of stable TLR4 homodimers.

Given that O. intermedium LPS is a TLR4 and TLR2-dependent agonist, the ability of this LPS to bind to a hypothetical heterodimer formed by the human TLR2 and TLR4-MD-2 receptors was explored by a molecular docking approach. First, the 3D structure of the hypothetical hTLR2/TLR4-MD-2 heterodimer model was constructed (see Materials and Methods and Supplementary Figure S4 ) and was subsequently used to perform docking calculations with E. coli and O. intermedium LPSs. The O. intermedium LPS was also divided into fragments for specific molecular dockings.

E. coli LPS accommodates five of its six lipid A chains inside the MD-2 pocket, and the remaining chain is at the surface of MD-2, in the TLR4/MD-2 homodimer (4). Therefore, the ability of O. intermedium LPS fragments (O19, OV28, and Ocore) to bind to the MD-2 pocket in the hTLR2/TLR4/MD-2 complex was investigated by docking calculations. Favorable poses for these fragments were found, where the two lipid A fragments showed their three FA chains buried in the MD-2 pocket and the core was near the entrance of MD-2 ( Supplementary Figure S5 ). As for the interactions, the glucosamine phosphate group was interacting with the NH₂ groups of Arg264, which is the same residue that interacts with the glucosamine of the E. coli lipid A in the hTLR4/MD-2 X-ray crystal structure (PDBI ID 3FXI).

In addition, full O. intermedium LPS was docked in the hTLR2/TLR4/MD-2 heterodimer to explore possible theoretical binding modes. The obtained binding poses had five FA chains of the LPS buried in the MD-2 pocket and the remaining sixth C12 chain placed on its surface ( Figure 4A ). In this case, both LPSs had their lipid A part accommodated in the MD-2 pocket. However, E. coli LPS core was diving toward the TLR2–TLR4 interface ( Figure 4B ). The interactions of the E. coli LPS core with the TLR4 residues described in the hTLR4/MD-2 dimer (Tyr296, Lys341, or Asp294) are not observed in the case of the hTLR2/TLR4/MD-2 heterodimer ( Figure 4C ). The TLR4 residues interacting with E. coli LPS core are closer to the TLR4 and TLR2 interface. This suggests that E. coli LPS does not favor the proximity of TLR4 and TLR2 receptors. On the other hand, O. intermedium LPS core maintains interactions with TLR4 residues such as Tyr296 and Arg264, as well as with TLR2 residues far from the interface region ( Figure 4D ). This implies that O. intermedium LPS accommodates in the hTLR2/TLR4/MD-2 complex in such a way that could favor the dimerization of both receptors. The O. intermedium LPS poses obtained from the docking calculations performed on the hTLR2/TLR4/MD-2 heterodimer and on the hTLR4/MD-2 homodimer hint that the core composition and/or structure of this LPS might favor the formation of an hTLR2/TLR4 complex rather than a TLR4 homodimer.

The ability of O. intermedium LPS core to bind in the hTLR2–hTLR4 interface region was also investigated. In this docking, more diversity in the binding poses was observed, where most of the poses were near the entrance of the MD-2 pocket and some poses featured the ligand in the upper region of the TLR2–TLR4 interface and toward the central domain of TLR2 ( Supplementary Figure S6A ).

The position of the O. intermedium LPS core in the hTLR2 interface (region spanning from the central domain, from Leu151 to Arg337, to the C-terminal domain, from Val338 to Iso506) shows interactions of the phosphate and hydroxyl groups with the backbone of TLR2, including residues His398, Ser425, Lys422, Gln396, Lys347, Asn345, Asp286, and Asn257 ( Supplementary Figure S6B ). Moreover, hydrogen bonds established with TLR4 residues Ser415, Gly389, and Lys388 were observed as well. The saccharide core from E. coli LPS was also docked in this region, suggesting that the E. coli core can theoretically interact with the TLR2 C-terminal domain ( Supplementary Figure S5C ) in an energetically favorable manner. The core also established hydrogen bonds with TLR4 Ser415, Lys388 similarly to O. intermedium LPS core; however, it showed fewer hydrogen interactions with TLR2 residues (His398, Gln396, and Asn345). Our calculations predict that the O. intermedium LPS core presents binding poses with more interactions with TLR2 residues located in the central and C-terminal domains. Therefore, O. intermedium LPS core seems better suited to establish interactions at the TLR2 interface compared to E. coli LPS core.

TLR2 is described to associate with TLR1 or TLR6 receptors, and this is required for recognition of the tri- and diacylated lipopeptides, respectively (26–29). TLR2 contains an internal hydrophobic pocket where it accommodates two lipid chains from TLR2-specific lipopeptides (30). Since TLR2 can hypothetically interact with TLR4 by O. intermedium LPS, docking studies were carried out using this LPS fragment in a region that covers the TLR2 pocket. Fragments O19, O28, and Ocore were docked in the TLR2 pocket of the hTLR2 monomer model. The O19 fragment poses showed the two FA chains C19 and C18 buried in the TLR2 pocket and the C14 chain was displayed in the solvent ( Figure 5A ). On the other hand, the O28 fragment that comprises the large 28C chain was fully accommodated inside the pocket, whereas C12 and C16 chains protruded out of the TLR2 pocket ( Figure 5B ). Regarding the Ocore, the obtained docked poses were interacting with TLR2, near its pocket ( Figure 5C ). These results indicate that the lipid A component of O. intermedium LPS is theoretically able to interact with the hydrophobic TLR2 pocket, comprising a maximum of two acyl chains.

As receptor dimerization appears to be required for PAMP recognition and TLR activation, FRET was performed to evaluate the intermolecular distance between TLR4 and TLR2 in the presence of O. intermedium LPS. HEK293T cells were transiently transfected with hTLR4-YFP (acceptor) and hTLR2-CFP (donor) constructs and imaged in vivo. Unstimulated cotransfected cells showed a small FRET signal, with minute punctate-like structures present in the cell membrane and cytoplasm. After stimulation with O. intermedium LPS, a significant FRET signal was observed by the presence of more punctate structures and with higher fluorescence intensity ( Figure 6A ). A tandem vector construct of CFP and YFP was used as a positive control, with a FRET efficiency of around 40% (data not shown). The size of these punctate structures however did not change upon cell stimulation ( Figure 6B ). This effect was mainly seen at high O. intermedium LPS (not shown) in agreement with the biochemical signaling data shown above. These results demonstrate that in resting cells, TLR4 and TLR2 receptors are not close enough. However, O. intermedium LPS stimulation promotes a closer interaction between TLR4 and TLR2 receptors that likely enables the activation of downstream intracellular signaling events. Thus, these data indicate that LPS from O. intermedium can induce the rearrangement of both TLR2 and TLR4 receptors favoring their interaction.