Supported by Crohn's and Colitis Foundation of America Senior Investigator Award and NEI/NIH EY 11654 (to S.C.).
Abnormal innate immune response contributes to inflammatory bowel disease (IBD) and experimental mouse colitis. Colitis studies have focused primarily on key regulators of innate immunity, like pathogen recognition receptors and cytoplasmic mediators. Extracellular matrix (ECM) proteins are emerging as modulators of inflammatory responses by virtue of their interactions with pathogen-associated molecular patterns (PAMPs), cytokines, growth factors, receptors, and ECM fragments that mimic pathogens or cytokines. The ECM proteins have not been investigated in IBD at great depth from this standpoint. We have shown previously that the ECM protein lumican modulates host sensing of bacterial lipopolysaccharides (LPS) by Toll-like receptor (TLR) 4, and neutrophil chemotaxis via integrins.
Here we investigated the role of lumican in the development of colitis mediated by intrarectal administration of the hapten 2-4-5, trinitrobenzene sulfonic acid (TNBS) in Lum+/+ and Lum−/− mice.
The TNBS treated Lum+/+ mouse colons showed marked increases in CXCL1, tumor necrosis factor alpha (TNF-α), and neutrophil infiltration, whereas these responses were significantly dampened in the Lum−/− mice. The nuclear factor kappa B (NF-κB) transcription factor, known to regulate inflammatory genes, showed a robust increase after TNBS treatment in Lum+/+ but not in Lum−/− colons. Also, nuclear translocation of NF-κB was delayed in LPS stimulated Lum−/− primary peritoneal macrophages.
The Lum−/− mice have low innate immune and inflammatory responses, but more severe body weight loss and tissue damage, a phenomenon seen in the innate immune impaired Tlr4−/− and MyD88−/− mice. Therefore, lumican promotes intestinal homeostasis by aiding innate immune and inflammatory responses that are beneficial in the early stages of colitis. (Inflamm Bowel Dis 2011;)
Inflammatory bowel disease (IBD), Crohn's disease (CD) and ulcerative colitis (UC), are chronic inflammatory diseases with certain common inflammatory changes but distinct etiology and pathogenesis.1, 2 In CD, transmural damage can affect any segment of the gastrointestinal tract, while UC manifests mucosal damage of the colon. It is now generally recognized that abnormal mucosal immune response to the commensal flora plays a major role in the etiopathology of IBD.3 In agreement with this view, IBD has been associated with changes in toll-like receptors (TLR) and the nucleotide-binding oligomerization domain containing (Nod) pathogen recognition receptors (PRRs) at the cell surface and cytoplasm, respectively.4, 5 Innate immune response is the first response of host cells to pathogen-associated molecular patterns (PAMPs), through TLRs and the Nod-like receptors.6 Activation of these PRRs ultimately lead to degradation of cytoplasmic inhibitors of nuclear factor kappa B (NF-κB) (IκB), nuclear translocation and activation of NF-κB, and upregulation of proinflammatory cytokine genes.7, 8 Studies investigating innate immune response have focused largely on these cell surface and cytoplasmic regulators. However, the magnitude of innate immune response is modulated by a number of other factors beyond the specific PAMPs and their host receptors. A few studies have identified extracellular matrix (ECM) components and their degradation products as having the potential to modulate innate immune response either by mimicking PAMPs, or by interacting with PAMPs or their receptors. For example, hyaluronan, an ECM glycosaminoglycan,9 and biglycan, an ECM protein, stimulate TLR2- and 4-mediated innate immune response.10 Mindin, an ECM protein of the spondin family, recognizes PAMPs and modulates recruitment of inflammatory cells.11 However, such modulations of innate immune response by non-immune components have not been investigated in IBD.
In a previous study, we developed a mouse model of chronic colitis by administering low doses of 2-4-5, trinitrobenzene sulfonic acid (TNBS) as weekly enemas.12 The TNBS colitis model develops transmural colonic inflammation and tissue damage as seen in Crohn's colitis. We previously investigated gene expression patterns at the early and late stages of this model. As expected, during the early stage (first 6 weeks) of active inflammation, several proinflammatory and innate immune response genes, such as Il1b, Cxcl1, several interferon gamma (IFN-γ) inducible genes, and Cd14 were upregulated.13 Surprisingly, genes for several ECM proteins were also upregulated at this time. Among these, lumican, a leucine-rich repeat ECM protein, was upregulated together with Collagen I and III. Our earlier study shows that lumican binds to bacterial lipopolysaccharides (LPS) and to CD14, the glycerol phosophatidyl inositol-linked cell surface protein that promotes host recognition of LPS through TLR4.14 Gene targeted lumican-null mice (Lum−/−) are hyporesponsive to LPS-mediated septic shock and produce lower levels of proinflammatory cytokines. At the cellular level, Lum−/− macrophages in culture produce lower levels of proinflammatory cytokines in response to LPS. This connection between lumican and innate immune response prompted us in the current study to examine the development of colonic inflammation in the Lum−/− mice.
MATERIALS AND METHODS
Gene targeted lumican-null mice that we developed earlier and maintained in the outbred CD1 background15 were bred into the C57BL/6J background by backcrossing for more than 12 generations. Wildtype (Lum+/+) and Lum−/− C57BL/6J mice 8–10 weeks of age of both genders were used in most experiments. In a few experiments, as indicated in specific figures, the Lum+/+ and Lum−/− mice in the CD1 background were used. The Rag1tm1Mom (Rag1−/−) C57BL/6J and wildtype C57BL/6J mice at 8 weeks of age were purchased from the Jackson Laboratories (Bar Harbor, ME).
Lum+/+ and Lum−/− C57BL/6J mice were randomized into control and TNBS groups. The TNBS mice were sensitized with an epicutaneous application of 1% TNBS (Sigma-Aldrich, St. Louis, MO) in 80% ethanol and the control group was given an epicutaneous treatment of saline (Fig. 1). Seven days after sensitization, the mice were anesthetized with avertin (350 mg/kg body weight) and weighed. The TNBS group received 5 μL/g body weight of 2.5% TNBS in 50% ethanol by intrarectal administration using a syringe fitted with a 20G gavage feeding needle (Fine Science Tools, 18060-20). The control group was treated in a similar manner with 5 μL/g body weight of saline or 50% ethanol. For histology and extraction of proteins, the mice were euthanized 48 hours after intrarectal administration of TNBS, saline or ethanol. The TNBS treatment described above was also used for the wildtype and the Rag1−/− C57BL/6J mice. In a few of our earlier experiments the Lum+/+ and Lum−/− in the CD1 background were not sensitized epicutaneously before intrarectal administration of TNBS as described.12 To determine survival and body weight loss, the mice were weighed daily without anesthesia for 8–10 days after intrarectal administration of TNBS. Animals that died by 24 hours of treatment were not included in determining mortality.
To develop the dextran sulfate sodium (DSS) colitis model, Lum+/+ and Lum−/− in the CD1 outbred strain were given 5% DSS in the drinking water. The mice were weighed daily and harvested on the eighth day.
Colon segments ≈2 mm in length were taken 4 cm away from the anus and fixed in 10% buffered formalin, paraffin-embedded, cut into 7 μm-thick cross-sections, and stained with hematoxylin and eosin (H&E). To examine thickening of the colon associated with colitis, images of H&E-stained colon cross-sections were captured with a Nikon Eclipse E400 microscope fitted with a DXM1200 Nikon Digital camera. The areas of colon cross-sections were measured using the software NIS-Elements. An increase in cross-sectional area was taken as an indication of inflammation and thickening of the colon. The pathology of the H&E-stained tissue sections was scored using a scale of 0–4 as follows: No significant lesions = 0; minimal to mild inflammatory infiltrates in the lamina propria or the submucosa = 1; mild to moderate inflammatory infiltrates, edema, occasional limited focal necrosis/ulceration = 2; moderate to marked transmural inflammation, edema, necrosis affecting up to 70% of the mucosa = 3; marked transmural inflammation, edema, necrosis affecting 70%–100% of the mucosa = 4.
Protein Extraction and Enzyme-linked Immunosorbent Assay (ELISA)
A colon segment of ≈2 mm length was taken from each control and experimental animal. Total protein was extracted by homogenization, followed by a brief sonication (10 seconds at 20% amplitude, 3 times) of the tissue in T-PER containing the Halt proteinase inhibitor cocktail (Thermo Scientific, Pittsburgh, PA) and centrifuged to remove debris. Protein concentrations were estimated using the BCA Protein Assay kit (Thermo Scientific).
The following ELISA kits were used to assay specific proteins and cytokines: mouse MPO kit (HK210) from Hycult Biotechnology (Uden, Netherlands), R&D mouse quantikine TNF-α kit (MTA00), R&D Quantikine mouse IL-4 (M4000B), R&D mouse KC/CXCL1 (MKC00B), total NF-κB p65 Sandwich ELISA kit (#7174) from Cell Signaling (Beverly, MA).
Electrophoretic Mobility Gel Shift Assay (EMSA)
NF-κB binding oligonucleotides (SC2505) and mutant oligonucleotides (SC2511) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and end-labeled with γ32P-ATP and T4 polynucleotide kinase (Fermentas, Burlington, ON, Canada). The labeled oligonucleotides were incubated with macrophage nuclear extracts and resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in a 6% polyacrylamide gel pre-run for 20 minutes. The gels were dried and exposed to film overnight at −80°C.
Measurements taken in control and TNBS-treated animals were compared using the Student's t-test when results were consistent with normally distributed values. Experiments that involved unequal sample size and did not necessarily follow normal distributions were compared using the Mann–Whitney test with P ≤ 0.05 considered significant. GraphPad PRISM software (v. 4.02, San Diego, CA) was used to perform these analyses.
All mouse procedures were evaluated and approved by the Johns Hopkins Animal Care and Use Committee. In case of extreme lethargy and >25% loss in body weight after inducing colitis, the mice were taken out of the study and euthanized.
Macroscopic Disease and the Course of Colitis
TNBS-treated mice of both genotypes showed signs of piloerection, lethargy, and diarrhea which were not seen in the saline-treated mice (data not shown). Ethanol, often used as a vehicle control for the TNBS treatment, leads to colonic tissue damage, and in our experience changes the expression of proinflammatory genes.13 Here we show that intrarectal administration of ethanol leads to similar trends in body weight loss in the Lum+/+ and the Lum−/− mice. The results further show that all ethanol-treated mice and the TNBS-treated Lum+/+ mice recovered their initial body weight by the sixth day after treatment (Fig. 2A). In contrast, the Lum−/− TNBS-treated mice did not recover by day 6, suggesting that the TNBS hapten had additional detrimental effects on lumican deficiency. Intrarectal administration of saline did not lead to loss in body weight (Fig. 2B). In the same experiment, the Lum+/+ and Lum−/− TNBS-treated mice showed up to 18% loss in body weight. However, after 7–8 days the Lum+/+ TNBS group had regained most of the lost weight, while the Lum−/− TNBS group still displayed on average an 11% body weight reduction (Fig. 2B). We also tested the TNBS colitis model in the CD1 outbred strain, and as in the C57BL/6J background, the Lum−/− CD1 mice showed slower recovery from body weight loss (data not shown). As is typical for the TNBS model, we found an overall survival of 30%, with no significant difference between Lum+/+ and Lum−/− mice, while saline-treated controls showed 100% survival (Fig. 2C). Macroscopic appearance of the colon was normal in the saline-treated animals (Fig. 3, top panels), while in the TNBS-treated wildtype and Lum−/− mice it was frequently swollen and bloody, with soft stool (Fig. 3, middle and bottom panels). To gauge edema and inflammation of the colon, we used two parameters: 1) cross-sectional areas of H&E-stained colon sections, and 2) colon to body weight ratios. The cross-sectional areas were higher for the TNBS group over saline-treated controls, with the difference being more pronounced and significant (P = 0.007) in the TNBS Lum−/− group over its respective saline control (Fig. 4A). The colon to body weight ratio was higher in the TNBS over saline-treated mice of both genotypes. However, the increase was more pronounced in the Lum−/− TNBS-treated mice (Fig. 4B). These results suggest that colonic edema and swelling is higher in the inflamed Lum−/− mice.
The TNBS colitis model is considered to involve a T-cell-mediated immune response in the colonic mucosa.16 However, the early acute response to the TNBS treatment may be largely innate immune-driven. To test this we used the TNBS colitis model in the Rag1tm1Mom mice (Rag1−/−). The Rag1−/− mice, deficient in mature B and T cells17, are often used as recipients in T-cell transfer colitis models18 but not directly in TNBS colitis models. In our TNBS model weight loss in the Rag1−/− was similar to that of the wildtype mice (Fig. 5A). The colon, harvested 2 days after the intrarectal TNBS treatment, appeared equally inflamed in both groups (Fig. 5C). Histology also indicated similar degrees of inflammation in the wildtypes and the Rag1−/− mice (Fig. 5B). These results suggest that the early stage of TNBS colitis involves an innate immune inflammatory response.
Given that the TNBS colitis model as used here is primarily an injury and innate immune response, we considered if the increased morbidity in the TNBS-challenged Lum−/− mice was entirely due to epithelial injury and delayed healing. To test this possibility we used the DSS-colitis model where inflammation is initiated by epithelial injury and loss of barrier integrity. Administration of 5% DSS in the drinking water caused diarrhea and equivalent weight loss in the Lum+/+ and the Lum−/− mice (data not shown). This suggests that in the TNBS-colitis model the TNBS hapten itself has an additional detrimental effect in the Lum−/− mice.
Colonic Tissue Damage and Inflammatory Cell Infiltration in the TNBS-treated Lumican-deficient Mice
Histology showed healthy tissue architecture in the saline-treated mice (Fig. 6A,B). The TNBS-wildtype group showed infiltration of inflammatory cells, moderate edema, and ulceration (Fig. 6E). The Lum−/− TNBS group tended to have more edema, ulceration, and frequent necrosis (Fig. 6D,F). However, the histology was quite variable, with some sections showing very little inflammation and tissue damage in the TNBS-treated animals. Therefore, to compare the overall response in the Lum+/+ and the Lum−/− TNBS-challenged mice, the slides were scored using a scale of 0 to 4 (see Materials and Methods). The pathology scores from one set of experiments shows higher inflammation scores in a larger proportion of the TNBS-challenged Lum−/− compared to the Lum+/+ mice (Fig. 7A). Measurements of myeloperoxidase (MPO) released by inflammatory cells, taken as a measure of inflammatory cell infiltrates, showed less than normal increase in colonic extracts of the TNBS-treated Lum−/− mice (Fig. 7B). Taken together, the results suggest increased edema, tissue damage, and necrosis, but overall lower inflammatory cell influx in the Lum−/− TNBS-treated mice.
Induction of Proinflammatory Factors in Inflamed Colonic Tissues
To further examine inflammation at the molecular level, we measured levels of selected cytokines in colonic protein extracts by ELISA. Compared to basal levels in saline-treated mice, the chemokine CXCL1/KC was markedly increased in the TNBS-treated wildtype mice (12-fold over basal level, P = 0.027). By comparison, the increase in Lum−/− TNBS mice was modest (5-fold, P = 0.04) (Fig. 8A). The proinflammatory cytokine TNF-α was increased significantly in the TNBS wildtype mice over saline controls, while the TNBS-treated Lum−/− mice showed very little induction of TNF-α (Fig. 8B). We next investigated IL-4 levels as a Th2 response-related cytokine; no significant change in IL-4 reaffirms a lack of Th2-related immune response in this colitis model (Fig. 8C).
Since NF-κB plays a central role in innate immune response and induction of proinflammatory cytokines and chemokines, we measured total NF-κB p65 levels in colonic protein extracts of each animal using ELISA methods. While total NF-κB p65 increased in both genotypes after TNBS treatment, the increase in TNBS-treated wildtypes was higher (2.5-fold) than that seen in TNBS-treated Lum−/− mice (1.7-fold) (Fig. 9A). To further test if there is a delay in the activation of NF-κB in lumican deficiency, we performed an EMSA test on primary peritoneal macrophages. Lum+/+ and Lum−/− macrophages in culture were exposed to LPS for different times. Nuclear extracts of these macrophages were incubated with radiolabeled NF-κB specific and nonspecific oligonucleotides and analyzed by SDS-PAGE. Maximal specific binding of nuclear proteins with the target oligonucleotide occurred in extracts of Lum+/+ macrophages that were treated with LPS for 10 minutes. In the Lum−/− macrophage extracts, comparable nuclear protein binding occurred after 20 minutes of LPS treatment. Thus nuclear localization of NF-κB in Lum−/− peritoneal macrophages was delayed by 10 minutes compared to wildtypes in response to bacterial LPS treatment (Fig. 9B).
Here we report distinct differences in the response of wildtype and lumican-deficient mice to an intrarectal challenge of TNBS that show weak innate immune and inflammatory responses, and increased morbidity in the absence of lumican. Lumican is an ECM proteoglycan present in large amounts in connective tissues rich in fibrillar collagens such as the intestinal submucosa.19 The lumican core protein has leucine-rich repeat (LRR) motifs, similar to those found in TLR proteins and other members of the LRR superfamily.20 Lumican deficiency disrupts collagen fibril assembly and impairs normal functions of connective tissues.15 In addition, we have shown that lumican promotes innate immune response through its interactions with the CD14-TLR4 pathway and neutrophil migration by interacting with β2 integrin, a subunit of the neutrophil migration receptor MAC1 or CD11b/CD18.14, 21 We also know that lumican expression is induced in epithelial cells after injury,22 and epithelial wound healing in the cornea is delayed in the Lum−/− mice.23, 24 The emerging role of lumican, and possibly some of the other members of this group of ECM proteins, is one of modulating innate immune signals during inflammation and wound repair and maintaining ECM homeostasis.
We used the TNBS colitis model to follow disease manifestation and inflammatory responses during acute and active inflammation of the colon. MPO measurements from colonic protein extracts suggest very little influx of neutrophils in the TNBS-treated Lum−/− mice. CXCL1/KC was induced in all TNBS colons, but the increase in Lum−/− mice was markedly lower than that seen in the Lum+/+ mice. Mouse KC, a functional equivalent of human IL-8, is a strong chemoattractant for neutrophils; its weaker induction in Lum−/− TNBS-treated colons is consistent with low MPO measurements in these animals. The proinflammatory cytokine TNF-α was also not induced to significant levels in the Lum−/− TNBS-treated colons. Lower inductions of KC and TNF-α in the Lum−/− mice suggest weak innate immune response from epithelial cells and macrophages in the colon. An ELISA that detects the p65 subunit of NF-κB showed lower levels of this subunit in the Lum−/− compared to Lum+/+ TNBS-treated colons. By gel shift assays we found that in Lum−/− primary peritoneal macrophage cultures there was a delay in nuclear localization of NF-κB in response to an LPS challenge. The NF-κB transcription factor is a common switch that responds to several cell surface and cytoplasmic pathogen recognition receptors, including TLR4, and mediates the upregulation of many inflammatory genes.25 NF-κB blockade in vivo has been shown to effectively ameliorate inflammation in the TNBS mouse colitis model.26 In a previous study we used this NF-κB blockade treatment in the chronic TNBS colitis model and found effective suppression of Cxcl1.13 Lumican was also inhibited by NF-κB blockade, indicating that its regulation closely mimics that of the inflammatory genes, rather than ECM collagens.13 Our observations in the current study further support the idea that lumican has an inflammation-regulatory role in colitis. The Lum−/− mice show subtle but clearly discernible differences in overall disease response that relate to the differences seen in innate immune signals at the molecular level. While there is no dramatic difference in mortality between wildtype and Lum−/− mice challenged with TNBS, the Lum−/− mice show significantly slower recovery of lost body weight. Histology of the inflamed colons, on average, show more edema and tissue damage in the Lum−/− TNBS colon. Also, there were more instances of necrosis in the Lum−/− TNBS colon that could be due to increased bacterial colonization of the colon.
The TNBS colitis first tested in rats27 and then adapted in the mouse16, 28 is considered to be driven by a T-cell-mediated immune response, as TNBS haptenates host proteins and makes them immunogenic. However, the response to the TNBS-ETOH challenge also involves an innate immune response that in the short term can produce proinflammatory cytokines and an inflammatory response without significant antigen presentation to T cells and the induction of Th1 cells. Thus, in the Rag1−/− mice, deficient in mature T and B lymphocytes, we also detected wildtype-like inflammatory response within 6 days after TNBS treatment. The DSS colitis model, which is T-cell-independent and mediated by epithelial injury and inflammatory cell infiltration, has been shown by others to be equally effective in the Rag1−/− mice.29 We tested the DSS colitis model in the Lum−/− mice, and compared to the wildtype mice, found no difference in disease manifestation. It appears that the DSS-mediated epithelial injury and healing may not be affected by lumican deficiency. In contrast, the TNBS-haptenated host proteins seem to be stimulating an early innate immune response that is affected by lumican deficiency.
In the Lum−/− TNBS-treated mice cytokine induction and inflammatory cell infiltration was reduced, yet disease severity in terms of weight loss and tissue damage was elevated. A more extreme version of reduced inflammatory response but aggravated disease is seen in the Tlr4−/−and MyD88−/− mice. These mutant mice show poor neutrophil infiltration, aggravated commensal Gram-negative bacteria in the lamina propria, and overall increased disease.30 In the same vein, MyD88−/− mice showed increased bacterial burden in a Citrobacter rodentium infection model.31 TLR4 and MYD88 dependent innate immune responses play a central role in epithelial and mucosal health.32–34 The exact protective mechanism is not well understood, but a combination of processes is credited. First, TLR activation by commensal bacteria promotes epithelial integrity and homeostasis. TLR-mediated activation of proinflammatory cytokines help epithelial survival and cell migration for efficacious healing of wounds. Second, infiltrating neutrophils and macrophages are protective by restricting microbial (commensal and pathogenic) overgrowth.
Innumerable studies of IBD point to an etiopathogensis of dysregulated immune response and altered response to environmental and microbial factors, with the concept of altered host innate immune response gaining significant traction.1, 3, 35 NOD2/CARD15, a key regulator of innate immune response, has been established as a CD susceptibility gene; TLR4 and CD14 have also been associated with IBD genetics. Investigations of the major innate immune signaling nodes will no doubt continue to elucidate IBD pathogenesis and significant susceptibility genes. The ECM genes may modify IBD pathogenesis to affect disease severity and possibly susceptibility. ECM proteins bind growth factors, cytokines, and cell surface receptors, and can either enhance or attenuate signals.24, 36, 37 Clearly, the absence of lumican in the mouse affects the outcome of colitis in a small but significant way. Other related ECM proteins that may have similar functions are fibromodulin, decorin, and biglycan. We believe that as more ECM proteins become better characterized, many will be identified as modifiers of key innate immune and inflammatory genes and as having an effect on IBD pathogenesis.