|Home | About | Journals | Submit | Contact Us | Français|
Toll-like receptor 2 (TLR2) activation induces cellular and organ inflammation, and affects lung function. Since deranged endothelial function and coagulation pathways contribute to sepsis-induced organ failure, we studied the effects of bacterial lipoprotein TLR2 agonists, including peptidoglycan-associated lipoprotein, Pam3Cys, and murein lipoprotein, on endothelial function and coagulation pathways in vitro and in vivo. TLR2 agonist treatment induced diverse human endothelial cells (EC) to produce IL-6 and IL-8, and to express E-selectin on their surface, including human umbilical vein EC (HUVEC), human lung microvascular EC, and human coronary artery EC. Treatment of HUVEC with TLR2 agonists caused increased monolayer permeability and had multiple coagulation effects, including increased production of plasminogen-activator inhibitor 1 (PAI-1) and tissue factor, and decreased production of tissue plasminogen activator (tPA) and tissue factor pathway inhibitor. TLR2 agonist treatment also increased HUVEC expression of TLR2 itself. PAL induced IL-6 production by EC from wild-type, but not from TLR2 knockout mice, indicating TLR2 specificity. Mice were challenged with TLR2 agonists, and lungs and plasmas were assessed for markers of leukocyte trafficking and coagulopathy. Wild-type mice, but not TLR2 mice, that were challenged intravenously with TLR2 agonists had increased lung levels of myeloperoxidase and mRNAs for E-selectin, P-selectin, and MCP-1, and had increased plasma PAI-1 and E-selectin levels. Intratracheally administered TLR2 agonist caused increased lung fibrin levels. These studies show that TLR2 activation by bacterial lipoproteins broadly affects endothelial function and coagulation pathways, suggesting that TLR2 activation contributes in multiple ways to endothelial activation, coagulopathy, and vascular leakage in sepsis.
Although the pathways that are activated during sepsis have been extensively characterized, much remains to be learned about the mechanisms underlying sepsis-induced organ failure. The endothelium, through its effects on inflammation, the coagulation pathways, blood flow, and vascular barrier function, is believed to contribute to the pathogenesis of organ dysfunction in sepsis (1–4). The endothelium produces pro-inflammatory cytokines such as IL-6 and IL8, and participates in leukocyte recruitment to organs (4, 5). It is also critically involved in modulating the coagulation pathways (2). Sepsis causes alterations in multiple limbs of the coagulation system leading to intravascular coagulopathy. Finally, endothelial activation is a critical contributor to vascular leak during sepsis (5, 6). Endothelial activation, coagulopathy, and vascular leak occur at locations that are remote from sites of infection, causing inflammation, edema, and ultimately “bystander organ injury” syndromes.
Toll-like receptors (TLRs) are a family of innate immune receptors that recognize conserved molecular motifs from microorganisms and play a central role in the initiation of inflammatory responses in sepsis (7–10). Although the various TLRs share intracellular pathways, there are differences between the TLRs. For instance, TLR4 recognizes lipopolysaccharide (LPS) (11, 12), whereas TLR2 mediates the effects of bacterial lipoproteins and components of Gram-positive bacteria and fungi (13–16). Also, different TLRs can elicit different inflammatory responses, which is based in part on their differential utilization of the MyD88-dependent versus the TRIF-dependent signaling pathways (8, 17–20). TLRs have been most extensively studied in macrophages, but they are also expressed by many other cell types, including endothelial cells (21–24).
TLR2 has been viewed as being primarily involved in responses to Gram-positive bacterial infections. In addition though, lipoprotein TLR2 agonists like peptidoglycan-associated lipoprotein (PAL), murein lipoprotein (MLP), and outer membrane protein A (OmpA) are ubiquitously expressed by Gram-negative bacteria, and there is increasing evidence that TLR2 is important in Gram-negative bacterial infections (25, 26). We previously observed that PAL, MLP, and OmpA, are shed into the blood of animals with Gram-negative sepsis (27–29). We reported that TLR2 activation modulates inflammatory effects of other TLR agonists, induces systemic, lung and myocardial inflammation in mice, and reduces contractility of cardiac myocytes in vitro (29–32). We have also found that TLR2 activation in vivo causes impaired vasoconstrictive responses to alveolar hypoxia and reduced arterial blood oxygenation (PaO2) in mice (33).
Because systemic endothelia interact continuously with the circulation, components of microorganisms constantly come into contact with endothelial TLRs during sepsis. We therefore performed studies to test the hypothesis that activation of TLR2 contributes to endothelial dysfunction and coagulopathy in sepsis. We found that treatment with TLR2 agonists has broad effects on the endothelium in vitro and in vivo, including upregulation of endothelial inflammatory responses, increased neutrophil trafficking to the endothelium, altered expression of coagulation pathway factors, increased endothelial permeability, and increased lung levels of fibrin. We also found that TLR2 activation caused endothelial cell apoptosis, and potentially upregulates septic responses by increasing expression of TLR2 itself.
PAL and MLP were prepared from E. coli bacteria as described, and were confirmed to contain a single protein band by gold staining (29, 34). Pam3CysSKKK was purchased (EMC Microcollections, Tubingen, Germany). Preparations of PAL and Pam3Cys contained < 5 pg LPS per microgram of protein, and MLP contained < 0.25 ng LPS per microgram of protein based on the Limulus-ameobocyte lysate assay. E. coli O111:B4 LPS was purchased (List Biological Laboratories).
The Institutional Animal Care and Use Committee approved all animal studies. C57BL6/J (wild-type) and TLR2-knockout (TLR2−/−) mice were purchased (Jackson Laboratory).
Human endothelial cells were incubated at 37°C under humidified 5% CO2. Human umbilical vein endothelial cells (HUVEC, passage 2–6), human lung microvascular endothelial cells (HMVEC-L, passage 4–5), and human coronary artery endothelial cells (HCAEC, passage 4–5) were purchased (Cambrex Bio Science). HUVEC were grown in EGM-2, and HMVEC-L and HCAEC were grown in EGM-2MV (Clonetics Corp, Walkersville, MD). Endothelial growth medium was supplemented with 2% FCS.
Cells were seeded into wells (4.5–7.5×104 cells/cm2), grown to 70–80% confluence, and then incubated with TLR2 agonists. ELISAs were used to quantify cytokines (R&D Systems), Tissue Factor Pathway Inhibitor (American Diagnostica), tissue plasminogen antigen activator (tPA, Innovative Research, Novi, MI), and plasminogen activator inhibitor-1 (PAI-1, Innovative Research) in supernatants. A cell-based ELISA assessed surface expression of E-selectin and TLR2. Immunoblots were used to detect TF and TLR2 in cell lysates. Viability and apoptosis were assessed using MTT and TUNEL assays respectively.
Surface expression of E-selectin by HUVEC, HMVEC-L and HCAEC was assessed using a cell-based ELISA at intervals up to 18 hours of stimulation with PAL. After removing the supernatants and gently washing the cells with RPMI/FCS, cells were incubated with mouse anti-human E-selectin (R&D Systems) for 45 minutes and were washed again with RPMI/FCS. They were then incubated with biotinylated anti-mouse IgG (Vectastain, Vector Labs) for 30 minutes. The ELISA was developed using avidin-biotin-peroxidase augmentation, and the absorbance (OD) was read at a wavelength of 405 nm. All steps except for the final step (substrate) were done at 4°C using pre-chilled reagents. ELISA was also used to assess surface expression of TLR2 after 20 hours of incubation with medium or Pam3Cys. HUVEC monolayers were grown to 70–80% confluence in collagen-I -coated 48-well plates, and were then treated for 20 hours with medium or Pam3Cys. The cells were then washed three times with Dulbecco’s PBS with magnesium and calcium and incubated for 1 hr at 37°C with goat anti-human TLR2 (R&D systems), and normal goat IgG control (EMD Gibbstown, NJ). Cells were washed with DPBS, and then were incubated with a 1:2000 dilution of bovine-α-goat-HRP (Jackson ImmunoResearch; West Grove, PA) for 1 hr at 37°C. The HUVEC were then washed and developed by addition of TMB solution (KPL; Gaithersburg, MD). ODs were recorded at a wavelength of 450 nm. Absorbance values were normalized based on the crystal violet staining of adhered cells in each well.
Confluent monolayers of HUVEC were treated for 18 hrs with medium, Pam3Cys or PAL (1 μg/mL and 10 μg/mL) before detaching them at 37°C using Accutase Cell Detachment Solution (Innovative Cell Technolgies; San Diego). HUVEC were then passed through a 40 μM filter, counted, and aliquoted at 1 × 106 cells per sample. The cells were then washed using Flow Cytometry Staining Buffer (FCSB, R&D Systems) and then were incubated with 10 μg of human IgG (R&D Systems) in 0.2 mL of FCSB for 15 minutes at 4°C. After washing twice with FCSB, the cells were incubated for 45 minutes at 4°C with primary antibodies, which included unconjugated goat anti-human TLR2 and normal goat IgG control (CD282, R&D Systems and EMD Gibbstown, NJ respectively; 2 μg/sample) and PE-conjugated mouse anti-human E-selectin and normal mouse IgG1 control (CD62-PE and mouse IgG1–PE, R&D Systems; 1:10 dilution). The TLR2 samples and controls were then washed with FCSB and incubated with FITC-conjugated secondary antibody (donkey-α-goat-FITC, Millipore, Billerica, MA; 2 μg/sample) for 45 minutes at 4°C. All samples were washed two more times with FCSB and then were analysis on a BD LSRII Flow Cytometer (BD Biosciences; San Jose, CA).
After incubation with TLR2 agonists and controls, HUVEC were lysed and protein concentrations of the lysates were estimated using the RCDC Protein Assay kit (Bio-Rad, Hercules, CA). The lysates contained solubilized proteins from all components of the cells, including organelles and membranes, but did not contain nuclear proteins. Total proteins were separated by SDS-PAGE (10% gels for TLR2, 12% gels for TF), and then transferred to PVDF membranes (Bio-Rad). The membranes were blocked in 3% BSA for 45 minutes at room temperature and were then incubated overnight with primary antibodies at 4°C. Primary antibodies used were anti-human Tissue factor mouse monoclonal IgG (1:1000, #05-881, Upstate, Waltham, MA), anti-human TLR2 goat polyclonal IgG (0.5 μg/ml, AF2616, R&D Systems) or anti-actin goat polyclonal antibody (Sigma-Aldrich). After developing the immunoblots, the same membrane was re-probed with anti-actin to confirm equal amounts of protein in the different samples. Binding of the primary antibodies was determined by adding suitable peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch, West Grove, PA). Immunoblots were developed using a chemiluminescent substrate, and then exposing the blots to imaging films.
Studies using human polymorphonuclear neutrophils (neutrophils) were approved by the Institutional Human Studies Committee. Neutrophils were prepared from blood drawn from healthy human volunteers as described (35). The final neutrophil pellet was suspended in HEPES buffer containing Ca2+ (1 mM) and the fluorescent label, calcein (3 μg/ml, neutrophils, 106/ml) (36). Near confluent monolayers of HUVEC were incubated with PAL, Pam3Cys, MLP, or medium for 2 hours, after which the supernatants were removed and the endothelial cells were washed 3 times with medium. The medium contained 10% FCS to block non-specific attachment of neutrophils to the plate. After the final wash, the calcein-labeled neutrophils were added to the wells (105 neutrophils/well), and were allowed to adhere for 60 minutes at 37°C. The plates were then inverted and centrifuged (200 × g, 5 minutes, room temperature) to remove non-adherent neutrophils. Calcein fluorescence of neutrophils was measured using a fluorescent plate reader (Fluostar Bio-Tek Instruments, VT) set at an excitation wavelength of 485 nm and an emission wavelength of 520 nm.
Permeability was determined by measuring fluorescein isothiocyanate (FITC)-albumin (Sigma-Aldrich) flux across the HUVEC monolayer using a transwell system as described (37). HUVEC were seeded into the upper (luminal) chamber of 6.5 mm transwells that contained 0.4 μM pore polyester membrane inserts (Corning Life Sciences, 1 × 105 cells/cm2). Cells were allowed to grow for 2–3 days until the monolayers were uniformly confluent, which was assessed by visualization using inverted microscopy, and confirmed by measuring transendothelial resistance (TER) (38). TER was measured using a tissue resistance measurement chamber, Endohm-6 (World Precision Instruments, Sarasota, FL). Dilutions of Pam3Cys (0 – 15 μg/ml) were added to the luminal chamber in a volume of 100 μl medium. Cytomix, a mixture of equal amounts of recombinant human IL-1β, INF-γ and TNF-α (10 ng/ml each, R&D Systems, Minneapolis, MN), was the positive control. For all conditions, the lower (abluminal) chamber contained 600 μl of EGM-2 medium. After 24 hours, FITC-albumin (200 μg/ml) was added to the luminal chamber in 100 μl medium. 2 hours later samples were removed from both the luminal and abluminal chambers, and their fluorescence was quantified using a FLUOstar OPTIMA plate reader (BMG Labtech, Durham, NC). The excitation and emission wavelengths were 485 nm and 520 nm respectively. Using a standard curve of FITC-albumin, the readings were converted to albumin concentrations that were used in the following equation to determine the permeability coefficient of albumin (Pa):
Where [A] and [L] are the abluminal and luminal FITC-albumin concentrations; t is time (seconds); A is area of membrane (cm2); V is volume of abluminal chamber (mls). Data in the figure are expressed as percents of control (medium alone).
HUVEC viability was measured using a MTT cell viability assay kit as per the manufacturer’s instructions (Biotium, Inc, Hayward, CA). The effects of TLR2 activation on HUVEC apoptosis were assessed using the TUNEL assay as per the manufacturer’s instructions (HT TiterTACS; Trevigen Inc, Gaithersburg, MD). Permeabilized cells treated with endonuclease served as the positive control for the TUNEL assay.
Mouse lung endothelial cells were prepared from lung tissue digests using immunomagnetic selection as described (39). Mice were euthanized by CO2 asphyxiation and blood was removed by cardiac puncture. Lungs were harvested, and then were then rinsed using DMEM/20% FCS, cut into 2mm3 blocks and digested by incubating in type II collagenase (0.2% w/v) for 1 hour at 37°C. The tissue was mechanically disrupted by passing the resultant digest 15–20 times through a blunt-tip needle and 10 ml syringe, and debris were removed by filtering the digested lungs through a 0.7-micron cell strainer. The filtrate was centrifuged (400 × g, 10 minutes, swinging bucket centrifuge, 4°C). The cell pellet was resuspended and then incubated for 2–3 minutes in ice cold RBC Lysis Buffer (Sigma-Aldrich, St. Louis, MO, 3 mls/2 mice), after which 7 mls of Basic Medium (20 mM HEPES, pH 7.4, 20% heat inactivated FCS, 100 U/ml penicillin, 100 μg/ml streptomycin) was added, and the cells were again collected by centrifugation. Cells were then resuspended in Dulbecco’s PBS with Calcium and Magnesium (DPBS, Invitrogen, Carlsbad, CA), counted, and the concentration of cells was adjusted to 3 × 107 cells/ml. Immunoselection of endothelial cells was done by sequentially incubating the cells with rat anti-mouse PECAM-1 (BD/Pharmingen, San Diego, CA) and rat anti-mouse ICAM-2 monoclonal IgGs (BD/Pharmingen) that were covalently attached to magnetic beads (5 μl antibody/50 μl beads) 1–2 days prior to use as described by the manufacturer (Dynabeads, Invivogen, San Diego, CA). The first magnetic immunoselection was done using anti-PECAM IgG. Cells were exposed to magnetic beads conjugated with anti-PECAM-1 IgG (50 μl antibody beads/3×107 cells) for 10 min at room temperature, using end-over-end rotation. Approximately 15 mls of Basic Medium was added to the tube, which was then placed in a magnet, and the liquid was removed from the tube. The beads and the attached cells were then put into gelatin-coated flasks containing Basic Medium and the following additives: 1% Sodium Pyruvate, 1% non-essential amino acids, 1mM L-glutamine, 150 μg/ml Endothelial Cell Growth Supplement (Biomedical Technologies Inc, Stoughton, MA), 12 U/ml Heparin, 50 mM 2-mercaptoethanol (Complete Medium). Medium was changed every 2–3 days. When cells reached 70–80% confluence, which took approximately 10–14 days, they were detached using Trypsin/EDTA. Trypsin was neutralized by addition of Basic Medium (10 ml), and the cells were collected by centrifugation. After this, cells were resuspended in DPBS, and were then incubated with magnetic beads conjugated with anti-ICAM-2 IgG for the second immunoselection, and processed as described above for the first immunoselection. The final cells morphologically resembled endothelial cells and formed a monolayer. To verify that the cells used for these studies expressed endothelial cell markers and to assess purity, we analyzed cells that had been frozen down from these experiments by immunofluorescence staining and flow cytometry essentially as described (39). Staining antibodies included anti-VE-Cadherin (CD144, Cayman Chemical; Ann Arbor, MI), ICAM-2 (CD102, BD Biosciences; San Jose, CA), and control antibodies, including rat and rabbit IgG. Secondary antibodies included FITC labeled goat anti-rabbit IgG and FITC labeled goat anti-rat IgG (Millipore). Both immunofluorescence staining and flow cytometry confirmed that 84–98% of the cells expressed these endothelial cell markers, which is consistent with the purity obtained by other investigators (39).
The mouse lung endothelial cells were placed in wells of 96 well plates at a density of 4.5 × 104 cells/cm2, and were incubated with PAL, LPS (positive control for the TLR2−/− cells), or medium alone. IL-6 levels were quantified in the culture supernatants by ELISA (R&D Systems).
Female mice aged 8–10 weeks were utilized for in vivo studies. For studies measuring plasma levels of PAI-1, and E-selectin, and lung expression of inflammatory mRNA’s, female mice were injected intravenously via the tail vein with carrier (50mM NaPhos, pH 7.4), or TLR2 agonists. Mice were euthanized by CO2 inhalation, and blood and lungs were collected. ELISAs were used to quantify plasma levels of soluble E-selectin (R&D Systems), and PAI-1 (Innovative Research Inc) levels. Lungs were analyzed for E-selectin, P-selectin and MCP-1 mRNAs using quantitative real-time PCR, and for neutrophil activity by measuring myeloperoxidase (MPO) levels (29).
Immunoblotting was used to assess the effects of Pam3Cys challenge on lung fibrin deposition (40). C57BL/6J mice were anesthetized by SQ administration of the combination of ketamine, xylazine, and acepromazine and then were challenged by intratracheal administration of Pam3Cys (200 μg/mouse) or carrier, each in a volume of 40 μl, followed by 1 mL air. After 20 hours, the mice were treated with systemic heparin (20 U IV), and then were euthanized by C02 inhalation. The lungs were perfused with saline, isolated, and homogenized in ice-cold homogenization buffer (20 mM Tris, 150 mM NaCl and 2 U/ml heparin). Lung tissue digests were then incubated with 0.1 U plasmin at 37°C for 4 hours with shaking to release fibrin monomers, and then were centrifuged and the supernatants were collected and stored in −80°C. Fibrin levels were measured using immunoblots. The membranes were stained with a monoclonal antibody against mouse fibrin (American Diagnostics, Stamford, CT). Immunoreactive proteins were developed using SuperSignal West Dura (Thermo Scientific, Rockford, IL) and visualized using FluorChem 5500 Imaging system (Alpha Innotech, San Leandro, CA) and band intensities were quantified via spot densitometry. After developing the immunoblots, the same membrane was stripped and re-probed with anti-actin to confirm equal amounts of protein in the different samples.
Q RT-PCR was performed on lungs of wild-type and TLR2−/− mice after treatment with Pam3Cys, PAL, or saline. Q RT-PCR was done using the ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA) and Taqman pre-made primers for E-selectin, P-selectin, and MCP-1 (Applied Biosystems). Presence of a single amplification product was verified using dissociation curves. Ribosomal RNA (18S) was detected using 18S primers (18S forward: 5′CGGCTACCACATCCAAGGAA, 18S: Reverse: 5′-GCTGGAATTACCGCGGCT) (DNA Oligonucleotide Synthesis Core, Massachusetts General Hospital, Boston, MA). Changes in gene expression were normalized to 18S ribosomal RNA levels using the relative cycle threshold method.
Data are expressed as mean ± SD. The data were analyzed using T-tests when comparing 2 conditions, one-way ANOVA with Bonferroni’s post-hoc analysis when comparing more than 2 conditions, and Mann-Whitney tests for data that was not normally distributed. GraphPad Prism was used for statistical analyses. P values < 0.05 were considered significant. Except where indicated, the following denote p values in the figures: * < 0.05; ** < 0.01; *** < 0.001.
In initial experiments, HUVEC were incubated with culture medium alone, or with serial dilutions of PAL, a naturally occurring TLR2 agonist that is expressed by enteric Gram-negative bacteria. PAL caused a concentration-dependent increase in IL-6 and IL-8 levels in culture supernatants (Figure 1A and 1B). Time course experiments with PAL (1 μg/ml) indicated that both IL-6 and IL-8 levels were similar to baseline at 2 hours, but both were significantly increased by 6 hours, continued to rise through 18 hours and then remained stable through the 24-hour time point (Figure 1C).
Additional experiments were done to assess the effects of 2 other bacterial lipoproteins, MLP a naturally occurring TLR2 agonist in the cell wall of enterobacteriaceae, and Pam3Cys, a synthetic TLR2 agonist. All three TLR2 agonists induced IL-6 production by HUVEC (Figure 1D). Based on the differences in molecular weight of PAL (18–21 kDa), MLP (5–9 kDa), and Pam3Cys (1.5 kDa), on an equimolar basis, PAL was found to be significantly more potent at inducing IL-6 than either MLP or Pam3Cys.
To assess the effects of TLR2 activation on endothelial cells in other vascular beds, we studied effects of PAL on inflammatory responses of human lung microvascular endothelial cells (HMVEC-L) and coronary artery endothelial cells (HCAEC). Treatment with PAL increased production of IL-6 and IL-8 by both HMVEC-L (Figure 1E) and HCAEC (Figure 1F).
To confirm that the effects of PAL on endothelial inflammatory responses are mediated through TLR2, we incubated lung endothelial cells from wild-type and TLR2−/− mice with PAL or with medium alone. There was no detectable production of IL-6 by endothelial cells that were treated with medium alone, suggesting that the process of preparing and growing the mouse lung endothelial cells did not in itself lead to sustained activation of the endothelial cells. Treatment with PAL increased production of IL-6 by lung endothelial cells from wild-type but not from TLR2−/− mice (Figure 1G), indicating that the expression of IL-6 induced by PAL is mediated through TLR2. IL-6 levels were equally increased in the supernatants of endothelial cells from wild-type and TLR2−/− mice that were treated with LPS, indicating intact signaling through intracellular TLR signaling pathways.
HUVEC, HMVEC-L and HCAEC were incubated with PAL or Pam3Cys (1–2 μg/ml), or medium for intervals up to 18 hours, after which the surface expression of E-selectin was assessed using a cell-surface ELISA and flow cytometry. For all three types of endothelial cells, PAL upregulated E-selectin at the cell surface (Figure 2A, shown for 2 hours). In contrast to IL-6 and IL-8 levels, which gradually rose to a maximum at 18 hours, surface E-selectin expression was significantly increased at 2 hours, was maximal at 4 hours, and remained elevated at the 18 hour time point (Figure 2B). Flow cytometry using HUVEC that were treated for 18 hours with PAL (1 μg/ml) or Pam3Cys (1 μg/ml) confirmed that TLR2 agonist treatment upregulates surface expression of E-selectin (Figure 2C).
Neutrophil-endothelial cell interactions are believed to be important in the migration of neutrophils to areas of inflammation. We therefore assessed binding of calcein-labeled human neutrophils to HUVEC that had been preincubated for 2 hours with PAL, MLP and Pam3Cys (1.5–12 μg/ml). The 2 hour time point was chosen based on our data indicating that surface expression of E-selectin on HUVEC was increased by 2 hours of exposure to PAL. All three TLR2 agonists caused a dose-dependent increase in binding of neutrophils to pretreated HUVEC (Figure 2D, shown for TLR2 agonist concentration 1.5 μg/ml). A representative fluorescent micrograph that shows the increased binding of calcein-labeled neutrophils to endothelial cell monolayers is included in Figure 2D.
Wild-type and TLR2−/− mice were injected intravenously with carrier, Pam3Cys (25 μg/mouse) or PAL (25 μg/mouse), and plasmas were collected after 90 minutes. Treatment with PAL or Pam3Cys increased soluble E-selectin levels in the plasma of wild-type mice but not in TLR2−/− mice (Figure 3A, shown for PAL only). Soluble E-selectin levels were increased equally in the plasma of both wild-type and TLR2−/− mice after treatment with LPS, indicating intact signaling through intracellular TLR signaling pathways (data not shown).
Quantitative real-time PCR was used to assess mRNA expression in the lungs of mice. Pam3Cys and PAL both increased expression of mRNAs encoding the adhesion molecules E-selectin and P-selectin, and the chemotactic cytokine, MCP-1 (Figure 3B and 3C). For all three markers, lung mRNA expression peaked 2 hours after challenge with Pam3Cys. However, only MCP-1 mRNA remained elevated at 22 hours (Figure 3B). The time course experiments were performed using high dosages of Pam3Cys (400 μg/mouse). This dose was chosen based on the dosages of LPS that are often utilized to study lung physiology and inflammatory responses in mice. However, since we also found that lower doses induced lung inflammatory mRNAs, we utilized lower dosages (25–50 μg/mouse) for subsequent in vivo experiments that compared responses of wild-type and TLR2−/− mice. Looking at the previously defined 2 hour point of rising E-selectin expression, we found increased expression of E-selectin, P-selectin and MCP-1 mRNAs in the lungs of wild-type but not TLR2 knockout mice (Figure 3C). Again, treatment with LPS equally increased mRNA levels in lungs of wild-type and TLR2−/− mice, indicating that TLR signaling pathways were intact after TLR2 knockout (data not shown).
We previously reported that peripheral blood neutrophil counts are decreased in mice treated with Pam3Cys (33). Because we found that TLR2 activation also increases expression of the adhesion molecules and chemokines, we hypothesized that activation of endothelial TLR2 would stimulate neutrophil trafficking to the lungs, which can be indirectly assessed by measuring MPO levels. Challenge with Pam3Cys (25 μg/mouse) or PAL (25 μg/mouse) increased MPO levels in the lungs of wild-type, but not in the lungs of TLR2−/− mice (Figure 3D, shown for Pam3Cys only). In contrast, lung MPO levels were equally increased in the lungs of LPS-challenged wild-type and TLR2−/− mice, indicating that downstream TLR signaling is intact.
Sepsis causes multiple coagulation abnormalities, including derangements in the fibrinolytic portion of the coagulation system. PAI-1 is an inhibitor of fibrinolysis. We incubated HUVEC monolayers with Pam3Cys and quantified PAI-1 in culture supernatants. Treatment with Pam3Cys increased PAI-1 levels in HUVEC culture supernatants in a dose-dependent manner (Figure 4A).
Tissue plasminogen activator is produced by the endothelium and plays an important role in fibrinolysis. We measured tPA total antigen in HUVEC culture supernatants after 18 hours of treatment with dilutions of Pam3Cys. There was a dose-dependent reduction in tPA in the supernatants of HUVEC treated with Pam3Cys versus medium alone (Figure 4B).
Tissue factor pathway inhibitor (TFPI) regulates the Tissue Factor coagulation pathway. To assess the effects of TLR2 on the modulation of this pathway, TFPI levels were quantified in HUVEC culture supernatants after 24 hours of treatment with PAL (20 μg/ml). TFPI levels were reduced in the supernatants of the PAL-treated versus medium-treated HUVEC (Figure 4C).
Tissue factor (TF) initiates coagulation by activating the extrinsic pathway of plasma coagulation and is the primary initiator of cell-based coagulation. We used immunoblots to assess TF expression in HUVEC lysates following treatment with PAL (20 μg/ml) or medium. A representative immunoblot is shown in Figure 4D. TF expression was increased in the lysates of HUVEC at 1, 6, and 24 hours of treatment with PAL (Figure 4D).
We next tested the expression of PAI-1 in wild-type and TLR2−/− mice after intravenous challenge with TLR2 agonists. We found that PAI-1 levels were increased in the plasma of wild-type mice injected with either PAL (25 μg/mouse) or Pam3Cys (50 μg/mouse) when compared to the plasmas of mice injected with carrier (Figure 5A and 5B). PAI-1 levels were not increased in the plasma of TLR2−/− mice treated with Pam3Cys, indicating that challenge with Pam3Cys induces PAI-1 specifically through TLR2 (Figure 5B).
To further explore the effects of TLR2 on lung coagulation, immunoblots were performed to assess fibrin deposition in the lungs of C57BL/6 mice 20 hours after intratracheal challenge with Pam3Cys (200 μg/mouse). Figure 5C shows a representative immunoblot, and the graph shows the fibrin band intensity relative to the actin band intensity of carrier-treated versus Pam3Cys-treated mice as assessed by spot densitometry (n=4). Fibrin was detected in the lungs of mice that were treated with Pam3Cys but not with carrier (Figure 5C).
Increased vascular permeability to albumin and other macromolecules is a hallmark of septic organ dysfunction. We used a transwell system to assess the effects of Pam3Cys on permeability. HUVEC monolayers were incubated in the upper (luminal) chamber with Pam3Cys (5–15 μg/ml) for 24 hours, after which FITC-labeled albumin was added to the luminal chamber, and flux was quantified based on fluorescent intensity measurements of samples taken from both the luminal and abluminal chamber. Pam3Cys concentrations of ≥ 10 μg/ml increased the permeability of HUVEC monolayers to FITC-Albumin (Figure 6).
To understand mechanisms by which TLR2 agonists change endothelial permeability, we considered the hypothesis that they might induce apoptosis or otherwise cause cell death. The effects of the TLR2 on HUVEC viability were therefore studied using the MTT assay, and the effects on apoptosis were studied using a TUNEL assay (Pam3Cys, 1.67–40 μg/ml). Low concentrations of Pam3Cys minimally affected viability, whereas viability was significantly reduced at Pam3Cys concentrations ≥ 15 μg/ml (Figure 7A). Pam3Cys also induced apoptosis in the TUNEL assay at concentrations ≥ 5 μg/ml (Figure 7B).
To determine whether or not TLR2 agonist treatment regulates TLR2 expression, immunoblots were done to detect TLR2 in lysates of HUVEC that were treated with Pam3Cys at intervals from 1 to 24 hours. TLR2 expression was minimal at baseline, and Pam3Cys upregulated TLR2 expression in a concentration- and time-dependent manner (Figure 8A). While TLR2 was strongly upregulated at the higher concentrations of Pam3Cys, TLR2 expression was visibly increased at the lowest concentration that we tested (0.31 μg/ml). TLR2 expression was strongly increased at 8 hours, and continued to rise until 20 hours, and remained elevated through 24 hours. Cell-based ELISA and flow cytometry studies were done to determine whether or not TLR2 agonist treatment upregulates surface expression of TLR2. Binding of anti-TLR2 IgG, but not isotype control IgG was slightly increased in cells that were treated with Pam3Cys as compared with cells treated with medium alone (Figure 8B). However, much higher concentrations of Pam3Cys were required to upregulate surface expression of TLR2 (10 μg/ml) than were required to upregulate TLR2 in HUVEC lysates. In fact, even the TLR2 was increased in HUVEC lysates at the lowest concentration of Pam3Cys tested (0.31 μg/ml). Flow cytometry studies also showed slight upregulation of TLR2 expression at the HUVEC surface by treatment with the higher concentration (10 μg/ml) of Pam3Cys and PAL (Figure 8C). The differences in concentrations required to induce TLR2 in cell lysates versus at the cell surface suggests that a significant proportion of the TLR2 that we detected in the lysates is intracellular. This finding is consistent with another published study showing predominant expression of TLR2 intracellularly, with slight increased expression at the cell surface after stimulation with inflammatory mediators (41).
During sepsis there is extensive interplay between disparate systems, including those involved in systemic inflammation, coagulation, and vascular barrier function. The endothelium plays a critical role in regulating these systems, and derangements as a result of endothelial activation or dysfunction can contribute to organ failure during sepsis (2–4). Our studies demonstrate that activation of TLR2 has broad effects on endothelial phenotype and function. We found that TLR2 activation increases endothelial cell expression of cytokines and of other inflammatory mediators in vitro and in the lungs of mice. Consistent with the upregulated expression of chemokines and adhesion molecules, more neutrophils adhere to endothelial monolayers after activation of TLR2, and MPO levels are increased in the lungs of mice challenged with TLR2 agonists.
Our studies also indicate that TLR2 activation modulates endothelial cell expression of factors involved in coagulation and in fibrinolysis. A functional effect of TLR2 activation on coagulation is suggested by the increased fibrin levels that we found in the lungs of Pam3Cys-challenged mice. We also found that activation of TLR2 increases endothelial permeability, a novel finding suggesting that TLR2 activation may contribute to the vascular leak seen clinically in sepsis. Finally, on a cellular level, we observed that HUVEC treatment with TLR2 agonists causes upregulation of TLR2 expression, decreased cell viability, and increased apoptosis. Our data show that TLR2 agonists activate endothelial cells from different vascular beds, including from the lung (HMVEC-L) and the heart (HCAEC). The fact that TLR2 agonists induce inflammatory responses in multiple endothelial cell types suggests that TLR2 activation has widespread effects on the endothelium in different tissues during sepsis. We have previously shown that TLR2 agonists are shed into the blood by bacteria in animal models of sepsis (28, 29). Thus TLR2 agonists have access to a vast endothelial cell surface throughout the body during sepsis, and circulating TLR2 agonists could contribute to the induction of diffuse endothelial activation and dysfunction as well as coagulopathy. TLR2 activation is therefore likely to contribute to multiple organ failure during sepsis by these pathways. In addition to playing a role in sepsis, studies have implicated TLR2 in the pathogenesis of atherosclerotic vascular disease (42). Our data support the notion that microbial components acting through endothelial TLR2 might be involved in the progression of coronary artery disease, or even potentially in acute ischemic events.
While some investigators have reported that TLR2 agonists do induce endothelial inflammatory responses (43–47), others have reported that TLR2 agonists do not induce endothelial inflammatory responses (41, 48, 49). There are several plausible explanations for the discrepancies in different studies. The plating density and presence of serum during the period of stimulation both may affect results. Additionally, differences in batches of endothelial cells could potentially contribute to variability in responses, due, for instance, to TLR polymorphisms (50). Our data that baseline TLR2 expression by HUVEC is low corroborates several reports that unstimulated human endothelial cells have only low levels of baseline TLR2 expression (41, 43, 48, 51). Human TLR2 has been shown to be upregulated by cytokines, histamine, and LPS, but has not been heretofore been shown to be upregulated by TLR2 agonists themselves (41, 43, 49, 51). We found that TLR2 expression was upregulated over time by treatment with Pam3Cys. We observed that the degree of upregulation of TLR2 on the cell surface by cell-based ELISA and flow cytometry did not seem to be as marked as that observed in immunoblots of cell lysates. In addition, higher concentrations of Pam3Cys were required to upregulate TLR2 at the cell surface than in the cell lysates, which suggests that much of the increased TLR2 was intracellular. This is in line with another study showing predominantly intracellular localization of TLR2 in HUVEC at baseline, and an increase in TLR2 at the surface of the HUVEC following treatment with IFN-gamma or IL-1beta (41). In contrast to our results, other investigators have found that lipoteichoic acid (LTA), a non-lipoprotein TLR2 agonist, does not upregulate endothelial TLR2 expression (49). Although the cause for the differences in induction of TLR2 between these two TLR2 agonists is not clear, the considerable difference in structure between the two agonists could be responsible. The possibility that different TRL2 agonists have different potencies is supported by the differences that we observed in induction of IL-6 production and neutrophil-endothelial adhesion by PAL, Pam3Cys, and MLP.
Multiple abnormalities of coagulation pathways occur in sepsis, including activation of procoagulant pathways, reduced production of anticoagulants, and decreased fibrinolysis. At the extreme, coagulation disturbance manifest as disseminated intravascular coagulopathy (52). Even in the absence of overt manifestations, disturbances in the balance of coagulation and anticoagulation occur that likely contribute to organ dysfunction during sepsis by impairing microcirculation and affecting the delivery of oxygen and nutrients to organs. PAI-1 inhibits fibrinolysis, thereby favoring perpetuation of already formed clots. Recent studies indicate that higher PAI-1 levels were associated with an increased incidence of organ failure and death in sepsis, pneumonia, and ALI (53–57). Our findings that TLR2 agonists upregulate PAI-1 expression by HUVEC in vitro as well as increase PAI-1 in vivo in the circulation of mice, suggest that TLR2 activation contributes to increased PAI-1 expression during sepsis. We also observed that TLR2 activation increased TF expression and decreased TFPI secretion by EC. We speculate that TLR2 activation may contribute to sepsis-induced coagulopathy through its broad effects on both the coagulation and fibrinolytic sides of coagulation and thrombosis. The global pattern of decreased anticoagulation (TFPI), increased coagulation (TF), decreased production of tPA which facilitates fibrinolysis, and increased inhibition of fibrinolysis (PAI-1), should favor a hypercoagulable state and predispose to microvascular thrombosis.
Respiratory dysfunction is common in patients with septic shock. TLR2 agonists have been shown by us and other investigators to induce systemic and pulmonary inflammation (29, 33, 58, 59). We also have found that activation of TLR2 causes impaired hypoxic pulmonary vasoconstriction (HPV) in mice (33). HPV is a physiological response to alveolar hypoxia that helps to maintain arterial blood oxygenation by facilitating ventilation:perfusion matching. Impairment of HPV, as occurs in sepsis, results in shunting of blood through unventilated areas of the lung, thereby contributing to hypoxemia. In the same study we found that mice treated with Pam3Cys had reduced arterial partial pressure of oxygen (PaO2) (33). Because of these findings, we hypothesize that during sepsis, TLR2 activation may have additional functional effects on the lungs. Both endothelial cell activation and dysfunction, and neutrophil sequestration and activation within the lung contribute to lung injury during sepsis (60). In early sepsis, endothelial cells express chemotactic cytokines and adhesion molecules, which promote neutrophil trafficking and transmigration across the endothelium. Our data indicating that TLR2 activation upregulates multiple processes involved in leukocyte trafficking to the lung and increases lung fibrin deposition support the hypothesis that TLR2 may contribute to the pathogenesis of sepsis-induced respiratory failure.
Increased vascular permeability is also prominent in septic shock. It allows fluid and protein to leak out of the intravascular space into the surrounding tissues, causing tissue edema and intravascular hypovolemia (6). These can contribute to shock and organ hypoperfusion. In the lung, sepsis-induced vascular leak leads to non-cardiogenic pulmonary edema, which contributes to respiratory failure. The mechanisms underlying vascular leak in sepsis have not been fully defined. We observed that TLR2 agonist treatment increases endothelial permeability, which suggests that TLR2 activation might contribute to vascular leakage during sepsis.
The concentrations of TLR2 agonists required to induce responses varied depending on the endpoint. Whereas 40 ng/ml-1 μg/ml concentrations were capable of upregulating the expression of cytokines, adhesion molecules, and some of the coagulation pathway factors, higher concentrations (≥ 10 μg/ml) were required for other effects such as permeability and apoptosis. This suggests that the strength of TLR2 activation may result in variable intracellular signaling pathways being utilized for the different responses observed in endothelial cells. However, further studies will be required to define the precise pathways leading from TLR2 activation to alterations in coagulation, apoptosis, and increased permeability. It seems likely that inflammatory pathways that involve NF-κB will be involved. Apoptosis may have contributed to the increased permeability. We speculate that coagulation pathway intermediaries also may be involved in the increased permeability that we observed. Prior studies support a role for coagulation pathway factors in regulating vascular permeability during inflammation (61–63). For instance, TF activity has been reported to contribute to the increased permeability induced by TNFα and IL-1.
The inflammatory effects of TLR2 activation have been described in a number of studies, but the effects of TLR2 activation on endothelial cell activation and function during sepsis have not been defined. We have demonstrated that bacterial lipoprotein TLR2 agonists broadly affect endothelial functions, including causing increased endothelial permeability, and modulate coagulation pathways in vitro and in vivo. The observed alterations in endothelial function and coagulation pathways closely resemble those that occur in patients with sepsis. Our studies here suggest that TLR2 activation contributes to sepsis-induced endothelial dysfunction, coagulopathy, and organ failure, and raise the possibility that interventions for sepsis should target endothelial TLR2 pathways.
This study was supported by NIH grants K08-AI01722 and R01-AI 058106 and by the Harvard Medical School 50th Anniversary for Scholars in Medicine Award and by Harvard University Milton Awards.
The authors have no disclosures to report.