TLR4 dependent heparan sulphate-induced pancreatic inflammatory response is IRF3-mediated

6-8-week old male C57BL/6J, TLR4^-/- , MyD88^-/- and IRF3^-/- were used in the study. Mice were kept in appropriate facilities at Lund University, under specific pathogen-free conditions. The animals were kept under 12/12 hours light/dark regime in standard mesh cages and handled according to the institute guidelines with approval of the Local Animal Care Ethics Committee.

Heparin by-products from beef lung were treated with alkaline copper sulphate (to remove dermatan sulphate) after papain digestion. The calcium salt was then precipitated with 36% ethanol (to separate from chondroitin sulphate), followed by fractionation of the cetylpyridinium complexes by solubilisation at 0.8 M NaCl to obtain HS3. HS3 is a fraction of low sulphation (HS3, 1.00 sulphate/unit compared to 2.40 of heparin) and has lower anti-coagulant properties than heparin. In order to purify it from danger signals, such as LPS, TGF-β and other contaminating factors, the sulphated glycosaminoglycan was finally subjected to dissociative gel FPLC on Superose 6 (LKB-Pharmacia), which was eluted with 4 M guanidinium chloride 0.05 M sodium acetate, PH 5.8, at a rate of 0.4 ml/min. The effluent was analysed for glycosaminoglycans and pooled material was recovered by ethanol precipitation and converted to sodium salts [9‐11].

Possible LPS contamination, which is an important consideration in the current study, was quantified by using the limulus amoebocyte lysate assay. The analysis showed a concentration of less than 0.5 EU/ml of LPS in the HS preparations. This is a concentration lower than what would be expected to influence the inflammatory response [12].

To determine the appropriate time to sample mice for the second phase of the study, a time-lapse experiment was performed with wild-type mice. The protocol for the in vivo experiments was a modification of a previously described procedure adopted for use in mice [13]. Briefly, the mice were anesthetized using isoflurane (Isoba Vet., Scherling-Plough, Stockholm, Sweden), a midline laparotomy was performed, the proximal end of the biliary duct clamped and the biliary-pancreatic duct was cannulated. HS 100 μl (500 μg/ml) was infused during the course of 2 minutes using an infusion pump. The clamp and the cannulae were then removed and the abdomen was closed in two layers. As a negative control, 100 μl of PBS and as a positive control, 100 μl of LPS (2.5 μg/ml from Escherichia coli 0111:B4, Sigma, St. Louis, MO, USA) were infused. The wild-type mice (10 animals per time point) were sacrificed by exsanguination after 1, 3, 6, 9, and 48 hours, and biopsies of the duodenal lobe were harvested, snap-frozen in liquid nitrogen and stored at -70°C until further processing. For histological and immunostaining studies the samples were fixed in 4% phosphate-buffered formalin.

To investigate possible signalling pathways, experiments were performed on wild-type and 3 different strains of genetically engineered mice, TLR4^-/-, MyD88^-/- and IRF3^-/-. The in vivo model was performed as previously described with the exception that nine hours was used as the time point, having determined this as the peak response time from the first experiment.

The effects of a TLR4 antagonist, eritoran (E5564; Eisai Research Institute, Andover, MA), were studied. Wild-type mice were categorized in three groups (eight per group). In the first two groups, eritoran (5 mg/kg) a synthetic analogue of lipid A and an antagonist of LPS was injected into the tail vein one hour prior to laparotomy. In the last group placebo (lactose monohydrate) was injected in the same way. The first and third (placebo) group were subjected to HS3, 500 μg/ml intraductally, while the second group only underwent laparotomy, leaving the pancreas intact.

Histological and immunohistochemical (IHC) techniques were followed according to standard procedures. Fixed tissue biopsies were dehydrated embedded in paraffin and 5 μm sections were cut. Primary antibodies, directed against either neutrophils (1:300, NIMP-R14, Abcam, Cambridge, UK) or F4/80 (1:500, MCA497GA, AbD Serotec, Oxford, UK) for monocytes, macrophages, and certain subpopulations of dendritic cells were used to visualize inflammatory cells. The slides were then incubated with appropriate secondary antibodies (1:400, ABC Vectastain, Vector Laboratories, Burlingame, CA, USA) and visualized using 3, 3'-diaminobenzidine (DAB, DAB peroxidase substrate Kit, Vectastain; Vector Laboratories). H₂O₂/methanol, levamisol (DakoCytomation) and avidin/biotin blocking (Vector Laboratories) were used to block the endogenous peroxidase, phosphatase and biotin, respectively.

Specificity of the antibodies for neutrophils and monocytes was confirmed by comparisons to morphological characters of the cell types. The slides were photographed using a Nikon Eclipse E800 microscope, Olympus DP70 camera and Nis elements software. The total number of neutrophils and monocytes in the entire sections was calculated and the total area was measured using ImageJ 1.38 (National Institute of Health, USA). Cell counts were expressed as total cells per mm² tissue.

Myeloperoxidase (MPO) activity was measured according to Koike et al [14], with some modifications, as briefly outlined below. Tissue samples were homogenized and washed in gradually increasing concentrations of PBS. The supernatant was reacted with 3,3',5,5'-tetramethylbenzidine in the presence of hydrogen peroxide (H₂O₂) and was subsequently stopped by addition of sulphuric acid (2M, H₂SO₄), after which the colour shift was analysed in a spectrophotometer at 450 nm (and 540 nm as control wavelength). Horseradish peroxidase (HRP) was used as standard and the results were expressed as μU/ml.

Mouse cytokine array kit (R&D Systems, Minneapolis, USA) was used to analyse 40 different murine cytokines. Carefully selected capture antibodies have been spotted in duplicate on nitrocellulose membranes. Samples were mixed with a cocktail of biotinylated detection antibodies. The sample/antibody mixture was then incubated with the array. Chemiluminescent detection reagent was added and the signal was read.

The statistical analysis of the data was performed using the Mann-Whitney U-test. A p-value < 0.05 was considered statistically significant and no corrections for multiple comparisons were made. All statistical analyses were done using SPSS 16.0 (SPSS Inc., Chicago, Ill., USA). Outliers were defined as > 1.5 times the inter-quartile range and excluded from the figures, but included in all calculations.

All comparisons in the treatment groups were made to the PBS control at the corresponding time point.

Immunohistochemical analysis of HS-infused pancreas showed a marked infiltration of neutrophils into the tissue after 9 hours (Figure 1A). LPS induced neutrophil recruitment into the pancreatic tissue to a higher extent (Figure 1B). No significant neutrophil staining was observed in pancreatic tissue from PBS-infused organs (Figure 1C). HS-treated organs contained a 3-fold increase in tissue neutrophil recruitment, measured as number of cells per mm² in pancreatic tissue, only 9 hours following the HS challenge as compared to the PBS group (p < 0.001; Figure 1D). The LPS-treated mice showed an increase of the neutrophil infiltration as early as at 3 hours. A further increase was noted 6 and 9 hours after the treatment (p < 0.001; Figure 1D).

Measurement of MPO activity demonstrated a similar pattern as neutrophil sequestration. MPO activity was increased at 9 hours in the HS group (mean 1.55 μU/ml, SE 0.29) as compared to the PBS group (mean 0.40 μU/ml, SE 0.17; p < 0.05). LPS caused an increase at 3, 6 and 9 hours (mean 8.11 μU/ml, SE 2.32; p < 0.001; Figure 1E).

Macrophage infiltration was also evaluated, but due to the low counts of F4/80 positive cells in each slide, no significant difference was found (data not shown).

The results showed that HS is capable of initiating an inflammatory response in the pancreatic duct, though less pronounced than seen following LPS.

HS as well as LPS induced significant amounts of MPO activity in the pancreatic tissue of wild-type mice (Figure 2; p < 0.05). In order to determine the role of the TLR4 signalling pathways in the HS-induced inflammatory response, TLR4^-/- as well as the adaptor protein, MyD88^-/- and the transcription factor, IRF3^-/- were used.

MPO activity correlated strongly with neutrophil recruitment into the tissue (Figure 1E). MPO activity was abrogated in TLR4^-/- mice in both HS- and LPS-treated animals, as compared to the response in the wild-type mouse. Similarly, LPS and HS did not induce an inflammatory response in MyD88^-/- mice (Figure 2). Interestingly, whereas the HS response in IRF3^-/- mice was completely abrogated, these mice still mounted a strong response to LPS (200-higher than HS-treated IRF3^-/- animals).

The above results show that TLR4-mediated HS inflammation is regulated by IRF3.

To investigate the mechanism by which HS activates TLR4, eritoran, a synthetic Lipid A analogue, which binds to TLR4/MD2, was used. This Lipid IVA has previously been shown to compete with Lipid A in the binding to MD2 and inhibits LPS-mediated inflammatory responses [15].

The MD2 dependence in HS-activated tissues was thus investigated using eritoran. Interestingly, the MPO activity of HS-treated wild-type mouse pancreas was completely abrogated by eritoran pretreatment (mean: 0.04177 ± 0.00405; p < 0.005) as compared to placebo (mean: 7.805 ± 2.674; Figure 3). Eritoran treatment alone did not induce any inflammatory responses.

In search for the involved chemoattractants in the inflammatory process, pancreatic samples of the time-response study were studied for the most well-known chemoattractants of neutrophils and monocytes: the monocyte chemoattractant protein-1 (MCP-1), keratinocyte chemoattractant (KC), and macrophage inflammatory protein-2 (MIP-2).

MCP-1 significantly increased at 3 hours after infusion of HS (p < 0.001; Figure 4A). LPS induced a more pronounced response than HS, but also peaked at 3 hours after stimulation (p < 0.001; Figure 4A). The concentration thereafter decreased and reached background levels 9 hours after infusion.

The concentration of KC in the pancreatic homogenates of the HS-treated group only modestly increased after 1 hour of stimulation, but was significantly different from what was seen in animals administered PBS (p < 0.05; Figure 4B). LPS, on the other hand, induced a pronounced KC production, peaking 3 hours after stimulation (p < 0.001).

Neither HS nor LPS infusion caused any significant change in MIP-2 levels within the pancreas (data not shown).

In order to provide an overview of the cytokine profile of wild-type and IRF3^-/- mice, we performed a cytokine array panel on the pancreas homogenates. From each group, one sample was chosen and the assay was prepared with the same amount of sample protein.

The array showed a difference between HS-treated IRF3 and WT mice regarding two cytokines: RANTES and IP10. Both cytokines were absent in the HS-treated IRF3^-/- mice, while present to a lesser extent in the LPS group.

Time response samples were studied and no demonstrable differences were found in the HS-treated wild-type samples prior to nine hours (Figure 5A).

To confirm the results of the cytokine array, RANTES was measured in all samples of the wild-types, as well as in TLR4^-/- and IRF3^-/- samples. RANTES was significantly increased in the HS-treated wild-type, but not in TLR4^-/- and IRF3^-/- animals, compared to the PBS group (p < 0.001). Moreover, compared to the PBS group, the LPS-treated group had a rise in wild-type and IRF3^-/- mice, but not in TLR4^-/- mice (Figure 5B).