Soluble TLR2 Reduces Inflammation without Compromising Bacterial Clearance by Disrupting TLR2 Triggering

Cells, cell activation, and NF-κB reporter assays

Human embryonic kidney (HEK) 293, Chinese hamster ovary (CHO), mouse RAW264 (American Type Culture Collection), HEK-TLR2, and CHO-CD14 (previously generated in our laboratory (17)) cells were cultured in DMEM (HEK293) or RPMI 1640 medium (Invitrogen) supplemented with 10% FCS (HyClone; <0.06 U/ml endotoxin), 2 mM glutamine, 400 μg/ml hygromycin B (HEK-TLR2), 1 mM pyruvate, 0.5% (v/v) NaHCO₃, and 50 μg/ml l-proline (CHO-CD14). The human monocytic cell line, Mono Mac-6 (provided by H. Ziegler-Heitbrock, Department of Immunology, University of Leicester, Leicester, U.K.), was cultured, as previously described (17). Human monocytes were also prepared, as described (17). Human neutrophil (PMN) preparations were obtained through dextran sedimentation and Ficoll density-gradient centrifugation of blood from healthy donors. Human peritoneal mesothelial cells were prepared and cultured, as described (22). For cell activation experiments (Fig. 2), triplicate cell aliquots (1 × 10⁵ cells/well) were cultured in serum-free medium supplemented or not (Fig. 2, A and B; and monocytes) with 500 ng/ml sCD14 (purified from human milk (11)) and stimulated with ultra-pure LPS (Escherichia coli O111:B4 strain), heat-killed Listeria monocytogenes, peptidoglycan, polyinosinic-polycytidylic acid (poly(I:C)), flagellin (all from InvivoGen), the synthetic bacterial lipopeptide Pam3-Cys-Ser-(Lys)₄ HCl (Pam3Cys; EMC Microcollections), IL-1β, TNF-α (R&D Systems), or PMA plus ionomycin (Sigma-Aldrich), as described in Results. Following a 16-h incubation, the supernatants were tested for IL-8 by ELISA (Duoset; R&D Systems). For NF-κB reporter assays, cells were transiently transfected (Lipofectamine; Invitrogen) with constructs directing expression of the firefly luciferase reporter and the Renilla luciferase. Forty-eight hours posttransfection, the cells were stimulated (16 h, 37°C) with Pam3Cys in the absence or presence of purified sTLR2, and luciferase activity was measured (Promega). sTLR2-depleted (≥90%) AB serum (Fig. 2F) was prepared by sequential immunoprecipitation (see below) with the anti-TLR2 Ab, sc8689, or an irrelevant, anti-plexin-C1 Ab, for mock depletion (both from Santa Cruz Biotechnology), as previously described (17). PBMC (2 × 10⁵ cells) were stimulated overnight with Pam3Cys in the absence or presence of 2% sTLR2- or mock-depleted serum.

Overexpression of sTLR2

A sTLR2 fragment consisting of the putative extracellular domain of human TLR2 (Met¹-Arg⁵⁸⁷) with an N-terminal c-Myc tag was constructed. The plasmid pCRII-TOPO-c-myc-TLR2, previously engineered (17), was used as a template to generate a PCR fragment corresponding to aa 1–587 of the TLR2 molecule. The sTLR2 cDNA was subcloned into the pDR2ΔEF1α expression vector. The recombinant plasmid was transfected into HEK-TLR2 cells. Expression and release of sTLR2-Myc by the HEK-TLR2 plus sTLR2 cells were confirmed by Western blotting of culture supernatants using a rabbit polyclonal anti-TLR2 Ab, TLR2p, generated in our laboratory by immunization with the N terminus 20-mer human TLR2 peptide SKEESSNGASLSGDRNGIGK (17) and anti-c-Myc epitope mAb clone 9E10 (Sigma-Aldrich).

Production of human rsTLR2

A TLR2 construct consisting of the putative human TLR2 extracellular domain (Glu²¹-Arg⁵⁸⁷) with a C-terminal His₆ tag tail was generated. The TLR2 cDNA was obtained by RT-PCR using RNA from Mono Mac-6 monocytes, and cloned into the pCR II-TOPO cloning vector, as previously described (17). The plasmid pCRII-TOPO-TLR2 cDNA was used as a template to generate a PCR fragment corresponding to aa 21–587 with a C-terminal His₆ tag. The resulting sTLR2 cDNA was subcloned into the baculovirus transfer vector pMELBacB (Invitrogen) in frame to the Honeybee Melittin secretion signal. Sf-9 cells were cotransfected with the recombinant pMELBacB-sTLR2 cDNA transfer vector and Bac-N-Blue DNA by using the Bac-N-Blue Transfection and Expression system (Invitrogen). High Five cell cultures (Express Five serum-free medium; Life Technologies) were infected with the recombinant virus. After 72 h postinfection, supernatants were cleared by centrifugation, filtered (0.22-μm filters), and concentrated 25 times (CentriconPlus-70; Millipore) before buffer exchange to 50 mM NaH₂PO₄, 300 mM NaCl, and 10 mM imidazole (pH 8.0; binding buffer). The sTLR2-His protein in the concentrated sample was purified by metal-affinity chromatography using Ni-NTA Superflow resin (Qiagen; 2 h, 4°C, orbital rotation). The protein was eluted by increasing the concentration of imidazole in the binding buffer to 250 mM. sTLR2-containing fractions were pooled and concentrated, and the buffer was exchanged to PBS. The protein concentration in the final sample was determined (Dc Protein Assay; Bio-Rad), and the purity of the sTLR2 preparation was assessed by 10% SDS-PAGE under reducing condition, followed by Coomassie blue G250 staining (Bio-Rad), as described in Results. The sTLR2 preparation was also analyzed by Western blotting, as previously described (17), using either a rabbit polyclonal anti-TLR2 Ab (TLR2p) or an anti-His₅ mAb (Qiagen). Typically, 200 μg of purified sTLR2 was obtained from 200 ml of High Five cell culture supernatant. The sTLR2 preparations were aliquoted and kept at −85°C until use.

Phagocytosis and superoxide production assays

For phagocytosis experiments, RAW264 macrophages (4 × 10⁵) were resuspended in 400 μl of binding buffer (phenol red-free RPMI 1640, 1% sodium azide, 2.5% HEPES) and incubated with FITC-labeled Staphylococcus aureus (Molecular Probes) at a bacteria:cell ratio of 10:1 for 30 min at either 0°C or 37°C. To test the effect of sTLR2, the bacterial suspension was preincubated with 5 μg/ml sTLR2 or BSA for 30 min at 37°C. Following binding, cells were washed with cold washing buffer (PBS/1% sodium azide), resuspended in washing buffer, fixed (2% paraformaldehyde), and analyzed by flow cytometry, as previously described (23). Cell surface fluorescence was quenched with 125 μg/ml trypan blue before flow cytometry.

For superoxide production assays, 150 μl of assay buffer (13 mM Na₂HPO₄, 3 mM NaH₂PO₄, 120 mM NaCl, 4.8 mM KCl, 1.2 mM MgSO₄, 11 mM dextrose, and 0.71 mM CaCl₂ (pH 7.4)) containing 5 μg/ml Pam3Cys or heat-killed Staphylococcus epidermidis strain PCI 1200 (American Type Culture Collection; 5 × 10⁶/well) and 5 μg/ml sTLR2 was added in triplicate to microtiter well plates (microlite 2; Thermo LabSystems) kept at 37°C. To this mixture, 2 μM Luminol and 2 × 10⁵ PMN resuspended in PBS were added, and the chemiluminescence generated was measured at 2-min intervals for 1 h using a fluorometer (Fluorstar Optima plate reader; BMG Labtech).

In vivo models of peritoneal inflammation

Inbred 8- to 12-wk-old C57BL/6 mice (Harlan) were maintained under barrier conditions and pathogen free. All experimental procedures were conducted under a Home Office project license. Lyophilized S. epidermidis cell-free supernatant (SES) was prepared from suspension cultures of S. epidermidis, isolated from an end-stage renal failure patient under continuous ambulatory peritoneal dialysis, as previously described (24). SES preparations were reconstituted in PBS before in vivo use. Peritoneal inflammation was induced in mice by i.p. injection of a defined 500 μl dose of SES, SES plus 100 ng of sTLR2, or control PBS. At the indicated time points, the animals were sacrificed, and their peritoneal cavities were lavaged with 2 ml of ice-cold PBS. Leukocyte numbers in the lavages were assessed by differential cell count on cytospin preparations and Coulter counting (Coulter Z2; Beckman Coulter), or by double staining (anti-F4/80 FITC, Serotec; anti-CD11b allophycocyanin, BD Pharmingen), followed by flow cytometric analysis. Chemokine levels in the cell-free peritoneal lavage samples were quantified by ELISA (R&D Systems). To test for mouse sTLR2 in the peritoneal lavages, equal aliquots were diluted with 3× concentrated Laemmli reducing sample buffer and analyzed by Western blotting using the TLR2-specific polyclonal Ab, TLR2p. Peritoneal inflammation was also induced by i.p. injection (500 μl) of 5 × 10⁷ or 5 × 10⁸ CFU S. epidermidis PCI 1200 strain (American Type Culture Collection) in the absence or presence of sTLR2. At the indicated time points, peritoneal cavity lavages were obtained. Blood was obtained by cardiac puncture. Bacterial CFU were determined by culturing blood and peritoneal lavage samples on Mueller-Hinton agar plates (Oxoid) overnight at 37°C.

Preparation of lipid rafts

Freshly isolated human monocytes (1 × 10⁸ cells) were resuspended in warm phenol red-free RPMI 1640 medium and stimulated (1 h at 37°C) or not with 5 μg/ml Pam3Cys in the absence or presence of 5 μg/ml sTLR2. Subsequently, protein solubilization was conducted (1% (v/v) Triton X-100, 150 mM NaCl, 50 mM Tris-HCl, 1 mM PMSF, 1 μg/ml leupeptin, and pepstatin (pH 7.4)) for 1 h at 0°C. Cell lysates (1.5 ml) were mixed with an equal volume of cold 90% sucrose solution (90% sucrose/50 mM Tris-HCl, 150 mM NaCl (pH 7.4)). Samples were overlaid with 7 ml of 30%, followed by 2 ml of 5% cold sucrose solutions, and centrifuged at 200,000 × g for 16 h at 4°C. One-milliliter fractions were removed from the gradient, and 60 μM n-octylglucoside was added to each fraction. Equal fraction aliquots were analyzed by Western blotting using the anti-CD14 mAb, MY4 (Beckman Coulter), or the anti-TLR2 mAb, IMG319 (Imgenex). To define the lipid raft-containing fractions in the gradient, dot blots of fraction aliquots were tested with HRP-conjugated cholera toxin B (List Biological Laboratories), followed by ECL (Amersham Biosciences) to reveal the presence of the raft-associated ganglioside, GM1.

Binding assays of sTLR2 to Pam3Cys, LPS, and bacteria

Triplicate wells of microtiter well plates (high binding; Costar) were coated (50 μl) with the amounts of Pam3Cys or LPS, indicated in Results, dissolved in ethanol. Following solvent evaporation at room temperature, nonspecific binding was blocked by incubation (2 h, room temperature) with PBS/1% BSA/5% sucrose/0.05% sodium azide. The plates were then washed three times (PBS/0.05% Tween 20) and incubated (2 h, room temperature) with 5 μg/ml sTLR2-His, sCD55-His (donated by C. Harris, Cardiff University, Cardiff, U.K.), or LPS-binding protein (LBP; Alexis Biochemicals) diluted in 0.05% Tween 20, 20 mM Trizma base, and 150 mM NaCl (pH 7.3) buffer (buffer A) supplemented with 0.1% BSA. Subsequently, the wells were washed and incubated (1 h on ice) with an anti-His₅ (5 μg/ml; Qiagen) or anti-LBP (1 μg/ml; Hycult Biotechnology) mAb diluted in buffer A/2% BSA. Following washing, the wells were incubated (1 h on ice) with a biotin-conjugated anti-mouse IgG Ab (DakoCytomation) diluted in buffer A/2% BSA, before washing and incubation (20 min on ice) with streptavidin-HRP (Jackson ImmunoResearch Laboratories) diluted in buffer A/0.5% skim milk. The wells were then washed, and color developed by addition of tetramethylbenzidine substrate (SureBlue; Kirkegaard & Perry Laboratories) was measured at 450 nm.

To test the binding of sTLR2 to bacteria, 5 × 10⁴ heat-killed S. epidermidis PCI 1200 strain was resuspended in 100 μl of PBS/0.05% BSA and incubated (30 min at room temperature) with 1 μg of mouse rsTLR2-human Fc (R&D Systems) or control CD46-Fc fusion protein (provided by C. Harris). Following washing (2× PBS/0.01% sodium azide), samples were incubated (0°C, 30 min) with a biotin-conjugated anti-human IgG Ab (0.5 μg; Southern Biotechnology Associates). After washing and fixing (2% paraformaldehyde), streptavidin-allophycocyanin (0.1 μg; Southern Biotechnology Associates) was added, and the samples were incubated for 20 min on ice before washing and analysis by flow cytometry.

Coimmunoprecipitation and chemical cross-linking experiments

The immunoprecipitation technique was as previously described (17). In this study, for membrane-bound (m)CD14-mTLR2 coimmunoprecipitations, 5 × 10⁶ freshly isolated human monocytes were resuspended in phenol red-free RPMI 1640 medium and incubated (30 min at 37°C) in the presence of 5 μg/ml sTLR2 or 10 μg/ml BSA. After washing and lysis (1% (v/v) Nonidet P-40, 50 mM Tris-HCl, 150 mM NaCl, 1 μg/ml leupeptin and pepstatin, 1 mM PMSF (pH 7.4) buffer), the cell lysate was precleared by successive incubations with the following: 80 μl of protein G-Sepharose (50% suspension; Sigma-Aldrich), 4 μg of the isotype-matched control, mouse IgG2b, and protein G-Sepharose. The precleared samples were incubated (1 h, 4°C) with 5 μg of the anti-CD14 mAb, MY4, and the immunocomplexes were precipitated with 50 μl of protein G-Sepharose. Following washing, samples were analyzed by Western blotting with the anti-TLR2 mAb, IMG319. For chemical cross-linking experiments, 5 × 10⁶ cells were resuspended in 500 μl of cold phenol red-free RPMI 1640 medium and incubated with 5 μg of sTLR2 (sTLR2-His) or the irrelevant His-tagged protein sCD55 for 30 min at room temperature. Following washing (cold RPMI 1640), 3 mg/ml membrane-impermeable and noncleavable cross-linker, bis(sulfosuccinimidyl)suberate (BS³; Pierce), was added to the samples, and the mixture was incubated for an additional 30 min at room temperature. Cross-linking was stopped by the addition of 10 mM Tris-HCl (pH 7.4) buffer and incubation on ice for 15 min. The cells were then lysed, and cell lysates were incubated (2 h, 4°C, orbital rotation) with Ni-NTA beads (10 μl of beads/100 μl of lysate). The beads were washed, and the protein was eluted with Laemmli reducing sample buffer containing 250 mM imidazole. The eluate was analyzed by 7.5% SDS-PAGE and Western blotting using anti-CD14 (69.4, rabbit polyclonal Ab (11)) or anti-TLR2 (sc8689) Abs.

Fluorescence resonance energy transfer (FRET) measurements

For FRET measurements, freshly isolated monocytes were allowed to adhere (1 h; 37°C) to multispot slides (Shandon Multispot; Thermo Electron) in the absence or presence of 5 μg/ml sTLR2 or BSA in phenol red-free RPMI 1640 medium. The slides were then incubated with 20% normal rabbit serum for 15 min at room temperature before washing and labeling (1 h, 0°C) with the anti-CD14 mAb My4-Cy3 or its isotype-matched IgG2b-Cy3 control (acceptor fluorophore; 0.25 μg/spot). Both Abs (Beckman Coulter) were Cy3 conjugated using the FluoroLink mAb Cy3 labeling kit (GE Healthcare). Cell labeling was performed in the absence or presence of 5 μg/ml sTLR2 or BSA. The slides were then washed (2× PBS/0.01% sodium azide), fixed (2% paraformaldehyde), and, following washing, stained with an Alexa 488-conjugated anti-TLR2 (T2.5; eBioscience) or anti-C3aR (hC3aRZ1; Serotec) mAb (donor fluorophore), as described above for MY4-Cy3. After washing and fixing, the slides were mounted (Vectashield; Vector Laboratories). FRET was measured by the release of quenched donor fluorescence after acceptor photobleaching using a previously described technique (25). In this technique, the donor fluorescence intensity before and after acceptor photobleaching in the same cell sample is compared. FRET efficiency was quantified by the following: E (%) = ((I_DA − I_DB)/I_DA) × 100, where E represents percentage of FRET efficiency; I_DA and I_DB, the donor’s intensity after and before photobleaching of the acceptor, respectively. In each cell to be analyzed, FRET efficiency was determined typically in four to five regions of the plasma membrane with different donor intensity. For each experiment, a minimum of 200 cells per condition was analyzed. E values from all regions of interest were averaged. The Cy3 acceptor fluorophore was bleached by repeated excitation (50 times for a total of 2 min), and the bleaching was ≥20% and up to 100% (depending on the experiment and the region of interest). The bleaching conditions were set to avoid bleaching the donor fluorophore. Cells were imaged using the Leica TCS SP2 resonant scanning confocal system (Leica Microsystems). Signal-to-noise ratio was improved by recording images using the frame averaging method (average of 2 frames). The donor fluorophore was excited at 488 nm, and emission was detected between 498 and 540 nm. The acceptor fluorophore was excited at 543 nm and detected between 551 and 669 nm. Under these conditions, negligible fluorescence was observed from an Alexa 488-labeled specimen within the Cy3 emission spectrum, and from a Cy3-labeled specimen within the Alexa 488 emission spectrum. The validity of each FRET dataset was confirmed by the lack of correlation between E% and acceptor or donor fluorescence intensity (data not shown). This suggested that the FRET values observed between mTLR2 and mCD14 were not dependent on acceptor or donor density, and thus resulted from genuine protein-protein interactions and not from randomly associated molecules (26, 27).

Statistics

Statistical analysis of the data was performed by using paired Student’s t test (Minitab 15 statistical software). Values of p of less than 0.05 were considered significant.

sTLR2 renders cells hyposensitive to TLR2-mediated stimulation

To assess the negative regulatory capacity of sTLR2, we engineered a soluble form of human TLR2 consisting of its full extracellular domain, thus resembling the main naturally occurring sTLR2 form found in plasma and milk (17). HEK293 cells stably expressing either the mTLR2 receptor (HEK-TLR2) or both mTLR2 and sTLR2, the latter tagged at the N terminus with a Myc epitope (HEK-TLR2 + sTLR2), were generated. Initial analysis by Western blotting and flow cytometry confirmed that the engineered sTLR2 protein was secreted into the medium by HEK-TLR2 plus sTLR2 cells, and that the HEK-TLR2 plus sTLR2 and HEK-TLR2 cells expressed similar levels of mTLR2 (Fig. 1⇓, A and B). A C terminus His-tagged human sTLR2 protein was also engineered and purified from insect cell culture supernatants. The purity of the sTLR2 preparation was assessed by 10% SDS-PAGE (reducing conditions), followed by Coomassie blue staining. Parallel samples were also analyzed by Western blotting using either anti-TLR2 or anti-His₅ Abs (Fig. 1⇓, C and D). Coomassie staining and Western blots showed a major 72- to 75-kDa sTLR2 band. Minor sTLR2 bands of ∼83 and 90 kDa, whose intensity depended on the preparation, were also detected, mainly with the anti-TLR2 Ab. sTLR2 preparations were estimated to be 85–95% pure (depending on the preparation) and mostly (∼95%) monomeric. Some of the low-intensity <75-kDa bands detected by Coomassie staining may correspond to sTLR2 degradation products, because they can also be detected by either the anti-TLR2 or anti-His Abs.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Expression and purification of human rsTLR2. A, Detection of sTLR2 in HEK-TLR2 plus sTLR2 culture supernatants (2 × 10⁶ cells) by Western blotting with the anti-TLR2 rabbit Ab, TLR2p, or the anti-cMyc epitope mAb, 9E10 (HEK-TLR2 + sTLR2 cells express an N terminus c-Myc-tagged sTLR2 protein). For control experiments, culture supernatants from HEK-TLR2 plus empty expression vector (EV) cell transfectants were tested. B, Fluorescence profiles of mTLR2 expression in HEK-TLR2 and HEK-TLR2 plus sTLR2 cell transfectants stained with the PE-conjugated anti-TLR2 mAb, TL2.1, or the isotype-matched control IgG. C and D, Coomassie blue staining (C) and Western blot (D) pattern of purified His-tagged rsTLR2 following production by insect cells, purification by Ni-NTA chromatography, and 10% SDS-PAGE (reducing conditions). For Western blots, an anti-His₅ mAb and the anti-TLR2 Ab, TLR2p, were used.

The HEK-TLR2 plus sTLR2 cells were found to be markedly insensitive to stimulation with different doses of the TLR2 agonist synthetic bacterial lipopeptide Pam3CysSer(Lys)₄ (Pam3Cys), as judged by the release of the proinflammatory chemokine IL-8 (CXCL8) (Fig. 2⇓A, left). The negative effect was sTLR2 concentration dependent, because HEK-TLR2 cells transiently transfected with increasing amounts of the cDNA encoding sTLR2 showed a concomitant progressive reduction in sensitivity (Fig. 2⇓A, right). TLR2 signaling inhibition by sTLR2 was not limited to Pam3Cys stimulation, because cell activation induced by another TLR2 agonist, peptidoglycan, and by the whole Gram-positive bacterium heat-killed L. monocytogenes, was also affected (Fig. 2⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

sTLR2 renders cells hyposensitive to TLR2-mediated stimulation. A and B, IL-8 levels in culture supernatants of HEK293 cells stably expressing mTLR2 (HEK-TLR2), mTLR2 and sTLR2 (HEK-TLR2 + sTLR2), the empty vector (HEK-EV) (A, left panel), or HEK-TLR2 cells transiently transfected with EV or sTLR2 cDNA (A, right panel) or 250 ng of sTLR2 cDNA (B), and stimulated, as indicated. Results are means ± SD of one experiment representative of four (A) or three (B). The differences in IL-8 release between sTLR2-expressing cells and HEK-TLR2 or HEK-TLR2 plus EV were significant: ∗∗∗, p < 0.0001. C–E, Cells were stimulated with the indicated concentrations of Pam3Cys or 200 ng/ml Pam3Cys (E), dilutions of SES, 80 μg/ml poly(I:C), 10 ng/ml LPS, 5 μg/ml flagellin, 5 ng/ml IL-1β, 10 ng/ml TNF-α, or 50 ng/ml PMA plus 500 ng/ml inonomycin in the absence or presence of 5 μg/ml sTLR2. For NF-κB reporter assays, cells transiently transfected with firefly and Renilla luciferase reporter plasmids were stimulated with Pam3Cys, followed by luciferase activity measurements. Results are of one experiment (±SD) representative of at least three (∗, p < 0.05; ∗∗∗, p < 0.0001 sTLR2 vs control). F, IL-8 levels released by Pam3Cys-stimulated PBMC in the absence or presence of sTLR2- or mocked-depleted 2% AB serum. Results are from one experiment performed in triplicates (±SD) representative of four (∗, p < 0.05; ∗∗, p < 0.01; sTLR2 vs mock-depleted serum).

The purified rsTLR2 protein also showed negative regulatory capacity. The inhibitory effect of rsTLR2 was observed in HEK-TLR2 cell transfectants, human monocytes, PBMC (data not shown), and (mTLR2⁺) peritoneal mesothelial cells (Fig. 2⇑C). The latter cells play a pivotal role during the course of a peritoneal infection, like the one studied in this work (see below), by secreting chemokines that regulate leukocyte infiltration into the peritoneal cavity and by expressing adhesion molecules (22, 28). Fig. 2⇑C shows that release of IL-8, the archetypal human PMN chemoattractant, by mesothelial cells stimulated with Pam3Cys or a cell-free supernatant prepared from the Gram-positive bacterium, S. epidermidis (termed SES), was reduced in the presence of sTLR2, suggesting that during peritoneal infections sTLR2 may also target mesothelial cells for negative regulation. The effect of sTLR2 was not limited to modulation of IL-8 release: the Pam3Cys-driven trans activation of the transcription factor NF-κB was markedly inhibited, indicating that sTLR2 has a wide spectrum of effects (Fig. 2⇑D).

The specificity of the sTLR2 inhibitory effect was evaluated next. We tested whether sTLR2 influences monocyte activation induced by suboptimal doses of the TLR agonists, viral dsRNA mimic poly(I:C) (TLR3), LPS (TLR4), and flagellin (TLR5). In addition, the effect of sTLR2 on signaling via the IL-1R (which shares with TLRs the MyD88-dependent signaling pathway), the TLR-nonrelated receptor TNFR, and nonreceptor-mediated cell stimulation (PMA + ionomycin) was also tested. Fig. 2⇑E shows that only TLR2-mediated monocyte activation was inhibited by sTLR2, indicating that sTLR2 targets monocyte TLR2 signaling specifically.

To evaluate the physiological relevance of the negative regulatory capacity of sTLR2, we compared the sensitivity of PBMC to stimulation via TLR2 in the presence of serum that had been depleted of naturally occurring sTLR2 (≥90%) with that of PBMC stimulated in the presence of mock-depleted serum (Fig. 2⇑F). Reducing the amount of serum sTLR2 resulted in a significant increase in cell sensitivity to TLR2-mediated stimulation. This result confirmed previous findings (17, 20), and suggested that naturally occurring sTLR2 may play an important immunomodulatory role in controlling TLR2-mediated activation in vivo.

Phagocytosis and superoxide production can be affected by sTLR2

To extend the assessment of the negative regulatory potential of sTLR2, we tested the capacity of sTLR2 to affect pathways associated with bacterial killing, namely phagocytosis and superoxide production. RAW264 macrophages were used to test macrophage phagocytic capacity in the absence and presence of sTLR2. The binding and phagocytic uptake of fluorescent bacteria were tested at 0°C and 37°C, respectively, in the presence and absence of trypan blue, to quench cell surface fluorescence. In this way, the amount of bacteria bound (0°C, trypan blue-sensitive fluorescence) and phagocytosed (37°C, trypan blue-resistant fluorescence) by macrophages was evaluated separately. To assess the full potential of sTLR2 as a regulator of the phagocytic process, the experiments were performed in serum-free medium, thereby excluding the contribution of Fc and/or complement receptors. Fig. 3⇓A shows that sTLR2 interfered strongly with the macrophage binding (0°C) of Gram-positive bacteria, S. aureus, while having a comparatively modest effect on phagocytosis (37°C). This effect on phagocytosis was, most likely, a consequence of the marked effect on bacterial binding. At 37°C, the activity of phagocytic receptors (e.g., scavenger receptors, C-type lectins) most likely compensates for the interfering effect of sTLR2.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Phagocytosis and superoxide production can be affected by sTLR2. A, Extent of FITC-labeled bacteria bound (0°C) or phagocytosed (37°C) by RAW264 macrophages preincubated or not with 5 μg/ml sTLR2 or an irrelevant protein (BSA, 2× sTLR2 molarity), as determined by flow cytometry. To distinguish between cell surface-bound and phagocytosed bacteria, the cell surface fluorescence was quenched with trypan blue before flow cytometric analysis. Results are of one representative of three independent experiments. B, Luminol-dependent chemiluminescence generated by superoxide produced over the time by triplicate cultures of human PMN stimulated with 5 μg/ml Pam3Cys or 5 × 10⁶ heat-killed S. epidermidis in the absence or presence of 5 μg/ml sTLR2. Results are from one representative experiment of four.

Freshly isolated human PMN were used to test the effect of sTLR2 on microbial-induced superoxide production. In the presence of sTLR2, the capacity of PMN to generate superoxide over time in response to either the Pam3Cys lipopeptide or whole heat-killed S. epidermidis was substantially reduced (Fig. 3⇑B).

Collectively, the results shown in Figs. 2⇑ and 3⇑ demonstrated the potential of sTLR2 to negatively regulate TLR2-mediated cell signaling and effector functions that are critical during microbial infection.

sTLR2 affects early leukocyte recruitment and endogenous sTLR2 release in a mouse model of peritoneal inflammation

To evaluate the biological activity of sTLR2 and assess its potential as a modulator of inflammation in vivo, we first tested the effect of sTLR2 on a well-established mouse model of acute peritoneal inflammation (24, 29). In this model, the progression of a clinical bacterial peritonitis episode typically seen in end-stage renal failure patients on continuous ambulatory peritoneal dialysis is mimicked by the peritoneal injection of a previously defined dose of the cell-free supernatant SES, derived from cultures of S. epidermidis, the main causative pathogen of this type of peritonitis (30). Intraperitoneal administration of SES to mice resulted in a rapid and transient increase in the peritoneal levels of the PMN chemoattractants, keratinocyte-derived chemokine (KC), and MIP-2, murine functional counterparts of human IL-8 and growth-related oncogene-α (CXCL1), with peak levels occurring at 1 h postinjection (Fig. 4⇓A). Corresponding determinations of PMN numbers recruited to the peritoneal cavity showed peak levels at 2–3 h (depending on the experiment) after SES administration (Fig. 4⇓B). The simultaneous administration of SES and sTLR2 (100 ng) resulted in reduced levels of PMN chemoattractants. These levels were significantly reduced in the case of KC, but not MIP-2 (Fig. 4⇓A). Consistent with the inhibitory effect on PMN chemoattractants, sTLR2 administration resulted in a marked reduction in PMN numbers recruited to the peritoneum either over the whole time course or at the peak of their influx (Fig. 4⇓B). The effect of sTLR2 on the SES-induced peritoneal levels of the mononuclear cell (MNC) chemoattractant, MCP-1 (MCP-1/CCL2), and on the relatively late recruitment of MNC, responsible for the removal of the apoptotic PMN, was also tested (Fig. 4⇓C). Under these conditions, sTLR2 was found to exert a positive and significant effect on MCP-1 levels over the time period post-SES injection. Total MNC recruitment, however, was not found to be affected.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

sTLR2 affects early leukocyte recruitment and endogenous sTLR2 release in a mouse peritoneal inflammation model. A–E, Mice were injected i.p. with a defined dose of SES, SES plus 100 ng of sTLR2, or PBS. At each time point, chemokine expression, PMN, and MNC numbers in the peritoneal lavages were determined (A–C). Cell numbers were determined by differential cell counts on cytospin preparations (B, right; results from four independent experiments) or leukocytes were double stained with anti-F4/80 and anti-CD11b mAbs and analyzed by flow cytometry (B and C, time courses). Values in A–C are expressed as the mean ± SEM (n = 5/condition; ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.0001; SES + sTLR2 vs SES). D, Annexin V/propidium iodide staining of leukocytes present in the lavages at the peak time of PMN influx (shown, 3 h). The representative scatter plots are from analyses of gated PMN. Apoptotic cells were identified according to the annexin V⁺/PI⁻ (lower right quadrant, early apoptosis) and annexin V⁺/PI⁺ (upper right quadrant, late apoptosis/necrosis) staining. Percentage of cells in the apoptotic quadrants is shown (mean ± SEM, n = 5/condition; ∗∗∗, p < 0.0001; significant reduction vs SES). E, Western blot of peritoneal lavages taken at the indicated times and tested for mouse sTLR2 (msTLR2) release. Densitometric scanning of msTLR2 levels at the peak of PMN influx is shown (right; n = 5/condition; ∗, p < 0.05; SES + sTLR2 vs SES).

The suppressive effect of sTLR2 on early (PMN), but not late (MNC), leukocyte recruitment posed the question of whether such a disproportionate leukocyte influx influences PMN survival and thus inflammatory resolution. We compared the apoptotic status of PMN at the peak of their peritoneal influx in SES-challenged mice with that in mice challenged with SES plus sTLR2 (Fig. 4⇑D). Profile comparison of the annexin V/propidium iodide scatter plots showed no difference in the proportion of early apoptotic PMN (lower right quadrant) between SES- and SES plus sTLR2-treated mice. Examination of the proportion of late apoptotic/early necrotic PMN (upper right quadrant), however, showed a marked and significant reduction (∼50%) of their numbers in the SES plus sTLR2-treated mice, suggesting a more efficient clearance of the dying PMN by the MNC in these animals.

The effect of administering sTLR2 together with SES to mice on the levels of endogenous (mouse) sTLR2 in the peritoneal lavage fluid was also tested, because we and others have demonstrated that sTLR2 release is affected by cell activation and infection (17, 18, 21). The detection of endogenous sTLR2 was facilitated by the absence of exogenous sTLR2 (sTLR2-His) in the peritoneal lavages (data not shown). At 1 h postinjection, no differences in the levels of sTLR2 between SES- and SES plus sTLR2-challenged mice were observed (Fig. 4⇑E). At 3 h, i.e., when PMN influx was high, mouse sTLR2 levels in the peritoneal lavages of the sTLR2-treated mice were found increased. By 6 h postinjection, sTLR2 levels between sTLR2-treated and nontreated mice were comparable and similar to those at the 1-h time point. This finding suggested that the administration of sTLR2 together with SES induced a positive feedback for the release of sTLR2, resulting in transiently higher local concentrations of endogenous sTLR2, which may well contribute to maintaining its regulatory effect on inflammation.

sTLR2 reduces peritoneal PMN infiltration without compromising bacterial clearance

The in vitro and in vivo anti-inflammatory effects of sTLR2 described previously raised the question of whether such effects would be detrimental to bacterial clearance during infection. To address this issue, an experimental model of acute peritoneal inflammation consisting of an i.p. challenge with 5 × 10⁷ CFU of S. epidermidis in the absence or presence of sTLR2 (100 ng) was used first. At this bacterial inoculum, the infection almost completely cleared by 12 h. PMN numbers in the peritoneum of mice injected with S. epidermidis showed peak levels at 12 h postinjection (Fig. 5⇓A, inset). In the presence of sTLR2, peritoneal PMN accumulation at the peak time of their influx was significantly reduced (Fig. 5⇓A). Such reduced early PMN influx did not, however, affect the capacity of the mice to clear the infection, because no difference in bacterial load either in the peritoneal cavity (Fig. 5⇓B) or blood (Fig. 5⇓C) between sTLR2-treated and -nontreated mice was observed. To study the inflammation-modulating effect of sTLR2 further, mice were injected with a higher dose of S. epidermidis (5 × 10⁸ CFU), and the effect of increasing doses of sTLR2 (10–1000 ng) was tested (Fig. 5⇓, D–F). All doses of sTLR2 tested showed a similar suppressive effect on the PMN numbers recruited to the peritoneal cavity over the 18-h period postinjection (Fig. 5⇓D). This effect was most significant at the peak time (12 h) of PMN influx. Despite this significant reduction in local PMN numbers, bacterial clearance from the peritoneal cavity was not negatively affected by sTLR2 treatment (Fig. 5⇓E). All mice (sTLR2 treated and nontreated) showed a marked reduction in bacterial load between 3 and 6 h postinjection. In sTLR2-treated mice, there was an apparent increase in bacterial clearance; however, this is most probably not physiologically significant, because it was very modest in magnitude. In the peripheral circulation, although the mice treated with the highest dose of sTLR2 (1 μg) showed increased bacterial load at the earliest time (3 h) postinjection (Fig. 5⇓F), this effect does not appear to be physiologically significant, because none of the animals showed peripheral bacterial abscesses, symptoms of shock, or died from the infection, and all animals had cleared the blood infection almost completely by 6 h postinjection. Moreover, plasma levels of the acute-phase reactant, serum amyloid-A, were similar in control (S. epidermidis only), and sTLR2-treated mice at all time points examined (data not shown), indicating that the relatively small increase in bacterial load in the bloodstream of 1 μg of sTLR2-treated mice was not significant enough to impact on the level of the systemic acute-phase response.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

sTLR2 reduces peritoneal PMN infiltration without compromising bacterial clearance. A–F, Mice (n = 5/condition) were i.p. inoculated with 5 × 10⁷ (A–C) or 5 × 10⁸ (D–F) CFU S. epidermidis alone or together with 100 ng (A–C) or the indicated amounts of sTLR2. At the indicated times, mice were sacrificed, the peritoneal cavity was lavaged, and PMN numbers in the lavages (A and D) were determined by differential cell counts on cytospin preparations. Bacterial titers in the peritoneal fluid and blood (B, C, E, and F) were determined, as described in Materials and Methods. Values in A and D–F are expressed as the mean ± SEM (n = 5/condition; ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.0001; S. epi. + sTLR2 vs S. epi.).

sTLR2 disrupts the interaction of mCD14 with mTLR2, acts as a decoy receptor, and associates with mCD14

We next examined the mechanism underlying the regulatory capacity of sTLR2. We first tested whether sTLR2 affected the ligand-induced clustering of mCD14 and mTLR2 in lipid rafts. The ligand-induced mobilization of TLR2 and TLR4 to lipid rafts and their close proximity to CD14, which resides mainly in the rafts, are believed to be critical to signaling (27, 31, 32, 33). Analysis of lipid raft preparations from freshly isolated nonstimulated (control) monocytes (Fig. 6⇓A) confirmed the preferential association of CD14 with lipid rafts and TLR2 with detergent-soluble (nonraft) fractions (31, 32). Pam3Cys stimulation resulted in an enrichment of TLR2 in lipid rafts and reduced levels of CD14, most likely as a consequence of the activation-induced shedding of soluble CD14 (9). However, when cells were stimulated in the presence of sTLR2, the pattern of CD14 and TLR2 partition into membrane domains resembled that in nonstimulated cells (Fig. 6⇓A), indicating that sTLR2 interferes with the ligand-induced TLR2 mobilization to lipid rafts for signaling, and consequently, with the approximation of TLR2 to CD14 in the rafts.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

sTLR2 disrupts mTLR2 triggering. A, mCD14 and mTLR2 partitioning into lipid raft and nonraft fractions following monocyte stimulation with Pam3Cys ± sTLR2. Dot blots (top) show the position of the raft marker, GM1 ganglioside. Results are from five independent experiments. B, Left, Binding of sTLR2-His to Pam3Cys-, but not to LPS-coated wells. Control assays show no binding of sCD55-His to Pam3Cys, and binding of LBP to LPS. For detection, anti-His₅ or anti-LBP mAbs, anti-mouse IgG biotin, streptavidin-HRP, and substrate were used. Right, Analysis of sTLR2-Fc or CD46-Fc binding to S. epidermidis following detection with anti-IgG biotin and streptavidin-allophycocyanin. Results are from six (Pam3Cys) or five (S. epidermidis) experiments. C, Western blots of CD14 immunoprecipitates following monocyte incubation with sTLR2 or BSA. Monocytes from four donors gave identical results. H, Ig H chain. D, FRET analysis on monocytes labeled with the anti-CD14 mAb, MY4-Cy3 (acceptor), and anti-TLR2 mAb, TL2.5-Alexa 488 (donor). Monocytes were preincubated and labeled in the absence or presence of sTLR2 or BSA. Threshold for significant FRET was determined with an anti-C3aR-Alexa488 mAb used as FRET donor-negative control. Results are from four experiments. E, Chemical cross-linking (BS³) in cells (5 × 10⁶) incubated with 5 μg of sTLR2-His or sCD55-His. Cross-linked His-tagged proteins in cell lysates were Ni-NTA pulled down and analyzed by Western blotting. Head arrows point at Ni-NTA-precipitated, CD14-cross-linked polypeptides. Left panel, Mobility of sTLR2 (5 μg) and mCD14 monomers pulled down (sTLR2) or immunoprecipitated (mCD14) from High Five cell culture supernatants or 5 × 10⁶ CHO-CD14 transfectants, respectively. Right panel, right lane, Control immunoprecipitation of mTLR2. Results are from four experiments.

A decoy receptor activity would explain, at least in part, such interference by sTLR2. We tested this possibility and found that sTLR2 specifically binds Pam3Cys lipopeptide (Fig. 6⇑B, left panel) in a ligand concentration-dependent manner, confirming previous reports (34, 35, 36). We also tested for a possible interaction of sTLR2 with whole bacteria. A sTLR2-Fc fusion protein, but not an irrelevant control, specifically bound heat-killed S. epidermidis (Fig. 6⇑B, right) as well as S. aureus (data not shown). These findings indicated the potential of sTLR2 to act as a decoy receptor; this activity may contribute to sTLR2’s negative regulatory capacity.

We speculated that sTLR2 might also disrupt the close proximity of mCD14 to mTLR2 directly, i.e., in the absence of ligand, by interacting with mTLR2 and/or mCD14. We tested this possibility by first examining the effect of sTLR2 on the ligand-independent natural association of mTLR2 with mCD14 in the detergent-soluble fractions of normal human monocyte cell lysates. The typical ∼110-kDa mTLR2 polypeptide band (17) was consistently detected by Western blotting in mCD14 immunoprecipitates from monocyte cell lysates (Fig. 6⇑C, left track). In addition, TLR2 polypeptide bands most likely corresponding to an intracellular (∼95-kDa) glycoform of the mature protein (37) and to fully glycosylated (∼83-kDa) and intracellularly located sTLR2 (17) were detected. The ∼110-kDa mTLR2 polypeptide band was not detected when the coimmunoprecipitation experiments were performed following preincubation of monocytes with sTLR2 (Fig. 6⇑C, center track), indicating that sTLR2 interfered with the natural interaction of mCD14 with mTLR2, and that such interference takes place at the cell surface. We obtained confirmatory evidence of this interference by performing FRET studies. FRET was used because it allows for the evaluation of interactions between neighboring (colocalized) molecules by determining their proximity within ≤10-nm range (26). In this study, FRET efficiency for the transfer of energy from the anti-TLR2 Alexa488 (donor) mAb to the anti-CD14 Cy3 (acceptor) mAb, used to label mTLR2 and mCD14, was measured. An increase in TLR2 (green, donor) fluorescence after CD14 (red, acceptor) photobleaching was detected in monocytes (Fig. 6⇑D, center), indicating energy transfer and, thus, close proximity between the two molecules. This was in agreement with the results of the coimmunoprecipitation experiments. In the presence of sTLR2, however, no increase in TLR2 fluorescence after CD14 photobleaching was observed, and FRET efficiency between TLR2 and CD14 was reduced to almost background levels (i.e., E% = 2.9 ± 0.5, threshold for significant energy transfer defined with an anti-C3aR Alexa488 mAb used as FRET donor-negative control; Fig. 6⇑D, right). These findings thus confirmed that sTLR2 perturbs the mCD14-mTLR2 interaction.

The interfering effect exerted by sTLR2 in the absence of ligand raised the question of whether this effect results from an interaction of sTLR2 with mCD14 and/or mTLR2. To address this issue, we performed chemical cross-linking experiments by using a noncleavable, membrane-impermeable, cross-linking reagent, BS³ (Fig. 6⇑E). The purified (Ni-NTA pulled-down) rHis-tagged sTLR2 protein (sTLR2-His) and mCD14, immunoprecipitated from CHO-CD14 cell transfectants, showed expected sizes of ∼72–75 kDa and ∼54–56 kDa, respectively, when analyzed by SDS-PAGE, followed by immunoblotting with specific Abs (Fig. 6⇑E, left panel). Incubation of CHO-CD14 transfectants with sTLR2-His, followed by chemical cross-linking, Ni-NTA bead pull-down from the CHO-CD14 cell lysates, and anti-CD14 or anti-TLR2 Western blotting revealed bands of ∼125–130 kDa and ∼250–260 kDa, i.e., of lower mobility than that of mCD14 (Fig. 6⇑E, center panel). The size of these bands was consistent with that estimated for mCD14/sTLR2 heterodimers (∼126–131 kDa) and mCD14/sTLR2 dimer of dimers (∼252–262 kDa). To test for an interaction of sTLR2 with mTLR2, a similar cross-linking strategy was applied to CHO-mTLR2 cell transfectants preincubated with sTLR2-His. In this study, however, Ni-NTA bead pull-down, followed by anti-TLR2 immunoblotting, did not show any cross-linked TLR2 polypeptide band (Fig. 6⇑E, right panel).

Together, these findings indicated that sTLR2 may exert negative regulatory effects by acting as a decoy receptor, and also by disrupting the close proximity between the coreceptor (CD14) and the receptor (TLR2) that is crucial to highly efficient signaling. Such disruption most likely results from the capacity of sTLR2 to interact with the coreceptor.