Differential Expression of Inflammatory and Fibrogenic Genes and Their Regulation by NF-κB Inhibition in a Mouse Model of Chronic Colitis¹

Fibrosis is a major complication of chronic inflammation, as seen in Crohn’s disease and ulcerative colitis, two major forms of inflammatory bowel diseases (IBD).³ Crohn’s disease in particular is prone to complications of fibrosis and stenosis (1, 2). The 2,4,6-trinitrobenzene sulfonic acid (TNBS) hapten, given as one or two enemas with ethanol as a carrier to disrupt epithelial integrity, induces acute inflammation and colitis in the mouse with Crohn’s colitis-like transmural tissue damage (3, 4, 5). In an earlier study, we modified the TNBS-induced colitis model by giving multiple low doses of TNBS over a period of 6 wk that resulted in chronic inflammation and fibrotic changes in the colonic tissue that persisted for at least 4 wk without additional TNBS administration (6). Neurath et al. (7) showed that acute inflammation in the TNBS-induced colitis model could be abrogated by blocking NF-κB p65 subunit expression with an antisense oligonucleotide targeted against the translation start site. The NF-κB transcription factor regulates the expression of a variety of genes that encode proinflammatory cytokines and proteins of innate immunity and acute phase response (8, 9). In IBD patients, the lamina propria macrophages have increased NF-κB p65 expression and DNA-binding activity accompanied by an increased production of IL-1, IL-6, and TNF-α (10).

In the chronic TNBS-induce colitis model, we have further shown that NF-κB antisense oligonucleotide enemas, given each time with the TNBS treatment, reduced the severity of disease in terms of body weight loss and colonic inflammation as seen by histology (6). Trichrome-stained histology of colon from TNBS-w8 mice treated with the NF-κB antisense oligonucleotide indicated considerable reduction in collagen deposition, with only 33% of the animals showing residual mild fibrosis. Another study of chronic TNBS-induced colitis achieved increased inhibition of fibrosis using a deoxyoligonucleotide that binds to the general NF-κB-binding consensus sequence, delivered directly into cells by encapsulating the oligonucleotide in a Sendai virus envelope (11).

In the current study, we explored the connection between chronic inflammation and fibrosis in the TNBS-induced colitis model with specific blockade of the p65 NF-κB subunit. Gene expression patterns of the colon were obtained at defined times from chronic inflammatory to late stages of the model. Our results show that high expression of inflammation-related genes at TNBS-w6, a stage with active and chronic inflammation, was followed by their rapid decline. Profibrogenic extracellular matrix genes were also up-regulated at this stage, but their overexpression continued after inflammation had subsided. We further studied the ameliorative effects of NF-κB p65 inhibition on the expression of specific inflammation and fibrosis-related genes.

CD-1 outbred female mice, 10–12 wk old (Charles River Laboratories), were maintained in a conventional animal housing facility at the Johns Hopkins University (Baltimore, MD) following animal protocols approved by the Animal Use and Care Committee at Johns Hopkins University. The chronic colitis model was developed by weekly enemas of TNBS (Sigma-Aldrich) in 0.1 ml of 45% ethanol given once a week for 6 wk (6). For the NF-κB blockade, mice were treated intrarectally with 150 μg of an antisense phosphorothioate oligonucleotide for the NF-κB subunit p65 (5′-GAAACAGATCGTCCATGGT-3′), or control oligonucleotide (5′-GTACTACTCTGAGCAAGGA-3′) in 0.1 ml of dH₂O 1 day before each TNBS enema (7). Mice given saline enemas alone were used as controls. Signs of colitis, diarrhea, rectal bleeding, weight loss, piloerection, lethargy and periorbital exudates were monitored as described previously (5). After the six TNBS enemas, mice were sacrificed 3 days (termed as TNBS-w6), 2 wk (TNBS-w8), 4 wk (TNBS-w10), or 6 wk (TNBS-w12) later (see Fig. 1).The TNBS-w6 stage was considered to reflect active, chronic inflammation and was selected as the first time point of the study. The saline control groups were also sacrificed at week 6 (saline-w6), week 8, and week 10. The last two sets were pooled as saline-w10. Colon tissue samples were collected at 4 ± 0.5 cm from the anus and placed immediately into 1) 4% formalin for histology, 2) TRIzol reagent (Invitrogen Life Technologies) for total RNA isolation, and 3) T-PER Tissue Protein Extraction Reagent (Pierce) for protein isolation.

The Murine Genome_U74Av2 array chip (Affymetrix) containing 12,400 probe sets was used to detect 9,710 unique transcripts (based on LocusLink identification released in September 2004). We established seven microarrays for mouse colons at TNBS-w6, TNBS-w8, TNBS-w10, TNBS-w12, saline-w6 (2 arrays), and saline-w10. The microarray procedures were standard protocols provided by Affymetrix (www.affymetrix.com/ support/technical/manual).

Briefly, equal amounts of total RNA (10 μg) were pooled per experimental group and converted to cDNA followed by synthesizing biotin-labeled cRNA. The fragmented cRNA (10 μg/array) was hybridized to MG-U74Av2 array. The hybridization images were first analyzed by Affymetrix Microarray Suite version 5.0 software to assess hybridization quality and background and to obtain image files of hybridization intensity for each probe set. The raw dataset is available at the National Center for Biotechnology Information Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) (accession no.GSE5864).

The DNA-Chip Analyzer software (dChip) (http://www.dchip.org/) (12) was used to normalize all arrays and generate a model-based expression value (arbitrary unit) for each transcript. A “present” or “absent” call for each transcript was defined according to whether the gene transcript was detected in the sample. The background signal was 76 ± 7 arbitrary units. We compared saline-w6 with TNBS-w6 and saline-w10 with TNBS-w8, TNBS-w10, TNBS-w12, respectively. Compared with the saline control, the expression signal in the TNBS-treated groups was considered significantly different if 1) fold change was >2, 2) a difference in signal intensity of 100 arbitrary units or greater, and 3) received a “present call” in at least one array.

The following web sites were used for functional annotation of novel genes, GeneCards (www.genecards.org/) (13) and National Center for Biotechnology Information (www.ncbi.nlm.nih.gov/entrez).

To confirm the gene expression changes, RT-PCR were performed by using a QuantiTect SYBR Green PCR Kit (Qiagen) with the following specific primers: Gapdh: forward 5′-TTGTCTCCTGCGACTTCA, reverse 5′-CCTGTTGCTGTAGCCGTATT; Il1b: forward 5′-GAATCTATACCTGTCCTGTG, reverse 5′-ACCGCTTTTCCATCTTCT; Cxcl1: forward 5′-ACTGCACCCAAACCGAAGTC, reverse 5′-CAAGGGAGCTTCAGG-GTCAA; Col1a1: forward 5′-GTACTCCTGGTGCTGATG, reverse 5′-GAAGCCTCTTTC-TCCTCTCTGA. Col3a1: forward 5′-CAGGTCCTAGAGGAA-ACAGA, reverse 5′-TCACC-TCCAACTCCAACAATG; Lum: forward 5′-TCGAGCTTGAT-CTCTCCTAT, reverse 5′-TGGTCCCAGGATCTTACAGAA; Pla2g2a: forward 5′-CAAAGG-ATTCCCCCAAGGAT, reverse 5′-CTCCAGGCGCTTGTAGCAA; Saa3: forward 5′-CCTGGGCTGCTAAAGTC-ATCA, reverse 5′-TCTTGAGTCCTCTGCTCCATGTC; Reg2: forward 5′-AGGAGAGTGG-TACTACAGCTTCCAA, reverse 5′-CGACGGTTACTTTTAG-GGTCATG; Reg3g: forward 5′-GGTAACAGTGGCCAATATGTATGG, reverse 5′-ATCCACCTCTGTTGGGTTCATAG; and Ccl2: forward 5′-CCCAATGAGTAGGCTGGAGA, reverse 5′-TCTGGACCCATTCCTT-CTTG. The cycles passing threshold (C_T) were recorded and relative expression level of a target gene = 2^ΔCT, where ΔC_T = C_T of Gapdh − C_T of the target gene.

The ABC kit (Santa Cruz Biotechnology) was used for immunohistochemistry staining on formalin-fixed, paraffin-embedded colonic sections. Primary Abs included: goat polyclonal anti-mouse S100A9 (1/250; Santa Cruz Biotechnology), rabbit polyclonal anti-mouse lumican (10 μg/ml) (14), rabbit anti-human CD3 clone PSI, and mouse anti-human CD4 clone IF6 (Ventana Medical Systems).

Colon tissue homogenates were used to determine IL-1β (Invitrogen Life Technologies) and chemokine CXCL1 (R&D Systems) protein levels by ELISA. The sensitivity of the ELISA kits is <7 pg/ml. The concentration of total protein for each sample was detected using the Bradford assay kit (Bio-Rad) and equivalent amounts were used to measure cytokine level and presented as a ratio of total protein per sample. An NF-κB immunoassay kit (BioSource International) was used to measure total NF-κB in colonic protein extracts.

Chronic colitis was established by six weekly enemas of TNBS as described earlier. The following time points were investigated here: TNBS-w6, 3 days after the sixth TNBS enema and TNBS-w8, TNBS-w10, and TNBS-w12 at weeks 8, 10, and 12 of the model, respectively (Fig. 1). Mice treated with TNBS enemas showed signs of diarrhea, fecal-occult blood, weight loss, piloerection, and lethargy, while 6 of 25 died within the first 6 wk due to severe colitis and weight loss. The signs of severe disease appeared after the third TNBS treatment and usually subsided within 1 wk and manifested briefly again after the next TNBS enema. The colon of the TNBS-w6 group showed hyperemia, edema, and hemorrhage. By TNBS-w8, thickening of the colonic wall was evident. Macroscopic appearance of the colon at TNBS-w12 was similar to the saline group (data not shown). The progression of inflammation and fibrosis was examined over time by histology. Colonic sections stained with H&E showed active inflammation at TNBS-w6. By the next stage, TNBS-w8, inflammatory cell infiltrates were visibly reduced, with the recovery of almost normal tissue architecture by TNBS-w10 and TNBS-w12 (Fig. 2,A). Adjacent colonic sections were stained with Masson’s trichrome to visualize newly synthesized collagen in fibrotic tissues. The patchy trichrome-positive staining of the lamina propria at TNBS-w6 indicated new collagen synthesis. This was replaced with a more regular, stronger staining of the mucosa and submucosa, increased fibroblasts and disorganized tissue architecture, indicative of fibrotic tissues at TNBS-w8 and TNBS-w10 (Fig. 2 B). Trichrome-positive staining of collagen deposits was evident in the submucosa even at TNBS-w12, suggestive of persistent fibrosis after inflammation had subsided. This is consistent with our previous study where collagen staining was evident as late as TNBS-w10, 4 wk after the last TNBS enema (6).

To characterize the inflammatory cells in inflamed colons of TNBS-treated mice, colonic sections were immunostained for S100A9 to visualize polymorphonuclear neutrophils and macrophages (Fig. 3,A). S100A9 (calgranulin B, or MRP14) is a calcium-binding protein expressed in inflamed tissues by macrophages and neutrophils (15). The microarray results (discussed later) also indicated elevated S100a9 transcript at TNBS-w6. In an earlier study, the S100A9 gene was found to be overexpressed in colonic tissue of IBD patients (16, 17). We detected a massive increase in S100A9-positive macrophages and neutrophils in the mucosa and submucosa at TNBS-w6. By TNBS-w8, these inflammatory infiltrates had decreased considerably (Fig. 3,A). A large body of research on IBD and animal models has shown that highly polarized CD4⁺ T cells producing IFN-γ with a Th1-type response are critical to chronic inflammation. Therefore, colonic sections of control and TNBS-induced colitis mice were immunostained with anti-CD3 and anti-CD4 to determine whether T cells are present in the chronic inflammatory stage of this model. CD3⁺ and CD4⁺ T cells were detected at TNBS-w6, as well as at TNBS-w8, in the lamina propria and in the submucosa (Fig. 3, B and C). Thus, macrophages and polymorphonuclear neutrophils that mark active inflammation declined by TNBS-w8, but influx of CD4⁺ T cells, indicative of adaptive immune response, was evident at TNBS-w6 and TNBS-w8.

To investigate gene activities during chronic inflammation and healing of the colon, we obtained gene expression profiles of the mouse colon at TNBS-w6, TNBS-w8, TNBS-w10, and TNBS-w12 and compared these with the saline controls. As compared with the saline control, there were 123, 72, 35, and 29 annotated genes differentially expressed at TNBS-w6, TNBS-w8, TNBS-w10, and TNBS-w12, respectively, comprising a total of 175 nonredundant genes (Table I). The sequential change in gene expression is shown in a heat map of above and below mean expression of these genes (Fig. 4, clusters A–E). A subset of 69 genes up-regulated at TNBS-w6 showed a rapid decline in gene expression within 2 wk, at TNBS-w8. A majority of the genes from this group related to inflammation (Ccl9, Ccr5, Cxcl1, Il1b, Nos2, Pla2g2a, Cd14, S100a8, S100a9), acute-phase immune response/Ag presentation (classical complement activation pathway genes C1qb, C1qg, histocompatibility class II Ag genes H2-Aa), and cell proliferation (Stat1, Anxa). Interestingly, a second group of proinflammatory genes maintained elevated expression beyond TNBS-w6, into TNBS-w8 and TNBS-w10 (25 plus 14 genes from cluster A). These include Ccl2, Ccl7, Cxcl5, Saa3, Pap, and Reg3 and may be important in maintaining chronic inflammation and fibrosis. Others from this late-induction group were related to microbial defense (macrophage-expressed Tgtp encoding T cell-specific GTP-binding protein, mast cell proteases, Mcpt1, Mcpt 2, Cma2), and cell proliferation (Ccnb1, Smc2l1, Mki67, Ptprc).

Genes encoding proteins relating to fibrosis, collagen type I and III (Col1a1, Col1a2, Col3a1), and lysyl oxidase, required for cross-linking newly synthesized collagen chains, were induced at the same time as the proinflammatory cytokine genes at TNBS-w6 (Fig. 4, cluster A). However, unlike the first tier of proinflammatory genes that was down-regulated by TNBS-w8, the ECM-related genes continued above basal expression into TNBS-w8 and later (Table I). Another gene, Lum encoding a proteoglycan, lumican, known to bind collagen and to regulate fibril structure (14, 18), was also expressed at high levels in TNBS-w6 (Fig. 4, cluster A and Table I). The expressions of selected proinflammatory genes and ECM genes Col1a1, Col3a1, and Lum were further investigated by qRT-PCR and immunohistology (presented later).

The heat map of differentially expressed genes revealed additional changes in expression that are relatively novel in the context of colitis. Genes involved in tissue remodeling (Eif4ebp2, Mmp15, Timp1), cell adhesion (Dsc2, L1cam), differentiation (Rdh16), and restoration of normal colonic functions were overexpressed later at TNBS-w10 and TNBS-w12 (Fig. 4, clusters C and D). Finally, a group of genes normally expressed at high levels in the resting healthy colon (Fig. 4, cluster E, saline controls), were down-regulated at TNBS-w6 and TNBS-w8. These include transport related genes, Slc1a1, Slc4a4, Slc20a1, Clca3, genes implicated in regulating cell shape and actin filament formation (Gsn, Scin), and posttranslational protein modification related to epithelial differentiation (Tgm3). Expression of many from this cluster was not restored to baseline, even at the latest TNBS-w12 stage (Table I). This suggests that although known inflammation-related genes resume normal expression late in the model, certain genes required for the restoration of tissue homeostasis do not recover or take much longer to do so.

The microarray results indicated elevated IL-1β transcript at TNBs-w6 and its decline by TNBS-w8. An increased amount of IL-1β mRNA was also detected by qRT-PCR at TNBS-w6 (Fig. 5,A). ELISA indicated statistically significant overexpression of the cytokine at TNBS-w6 and TNBS-w8 (Fig. 5,B). The chemokine ligand CXCL1, also known as GRO-α, is a neutrophil chemoattractant, secreted by activated macrophages, and plays a key role in inflammatory responses. By microarray, Cxcl1 was found to be up-regulated at TNBS-w6. qRT-PCR on total RNA from individual TNBS-treated and control animals indicated statistically significant high levels of the transcript and elevated levels of the chemokine ligand by ELISA at TNBS-w6 (Fig. 5, C and D).

Collagen types I and III are known to be regulated differently in fibrosis (19, 20, 21, 22). Our microarray profiles indicated overexpression of fibrillar collagen genes Col1a1 and Col3a1 during active inflammation at TNBS-w6 and later at TNBS-w8. qRT-PCR on colonic total RNA confirmed elevated transcripts of Col1a1 and Col3a1 at TNBS-w6 and TNBS-w8 and return to basal levels by TNBS-w10 and TNBS-w12 (Fig. 6,A). Lum was also induced at TNBS-w6, and its expression remained elevated above basal level at TNBS-w10 and TNBS-w12 (Fig. 6,A). Immunohistochemistry of colonic sections further showed that there was a marked increase in lumican deposit in the mucosa and submucosa of mice from the TNBS-w6, TNBS-w8, and TNBs-w10 groups (Fig. 6 B).

An antisense oligonucleotide against NF-κB p65, administered as rectal enemas, has been demonstrated to suppress acute inflammation within the first 2 wk of TNBS-induced colitis (7). We further showed that chronic inflammation and fibrosis in week 8 could be inhibited significantly by administration of the antisense NF-κB oligonucleotide during development of colitis (6). Examined at an earlier time point in week 6 in the current study, we found mild inflammation and fibrosis in 60% and moderate inflammation and fibrosis in 40% of the antisense NF-κB-treated colitis group (Fig. 7,A). The control oligonucleotide containing a scrambled sequence of the NF-κB antisense oligonucleotide had no beneficial effects. The influx of CD4⁺ T cells was also partially inhibited by the NF-κB blockade (Fig. 7,B). Taken together, these results indicated that although the NF-κB blockade treatment reduced inflammation and fibrosis, neither was completely abrogated by this method. Residual signs of inflammation and fibrosis in the NF-κB blockade group could be due to incomplete inhibition of NF-κB. Therefore, we determined the extent of NF-κB inhibition after the antisense oligonucleotide treatment by ELISA measurements of total NF-κB from colonic protein extracts. As expected, there was a significant increase in total NF-κB after inducing colitis with TNBS compared with that of saline-treated controls. The NF-κB antisense oligonucleotide treatment reduced total NF-κB by 60% (p < 0.05), while the control oligonucleotide had no such inhibitory effect (Fig. 7 C).

We next investigated how inflammatory and profibrogenic genes responded to the NF-κB inhibition. The microarray gene expression patterns had indicated the presence of several proinflammatory genes, Cxcl1, Il1b, Pla2g2a, Saa3, Reg2, and Reg3g, that are also likely targets of NF-κB regulation (23, 24). Quantitative assessment of these transcripts in TNBS-induced colitis subjected to the NF-κB blockade treatment showed statistically significant suppression of gene expression (Fig. 8). The control oligonucleotide had little to no inhibitory effect. The microarray results had shown Ccl2 to be one of several proinflammatory genes with continued overexpression at later stages of chronic colitis. Ccl2 has recently been suggested to play a significant role in driving fibrosis (25). Therefore, we tested the expression of Ccl2 in mice with TNBS-induced colitis before and after the NF-κB blockade treatment (Fig. 8). Unlike the other proinflammatory genes tested, expression of Ccl2 was not suppressed effectively by partial inhibition of NF-κB.

To investigate the effects of NF-κB antisense treatment on profibrogenic genes, we assayed for the expression of Col1a1, Col3a1, and Lum at TNBS-w6 and TNBS-w8 before and after treatment with NF-κB antisense oligonucleotide (Fig. 9). All three ECM genes had shown a modest increase in expression at TNBS-w6 and TNBS-w8. Partial inhibition of NF-κB-blocked Lum expression effectively, but Col1a1 and Col3a1 expression was not inhibited by the NF-κB blockade treatment.

A major complication of chronic transmural inflammation of the intestine is fibrostenosis as seen in Crohn’s disease (2). We investigated the connection between chronic inflammation and fibrosis using a mouse model of chronic colitis, established by giving six weekly enemas of TNBS as described in our earlier study (6). We examined gene expression patterns in this model from weeks 6–12, spanning a period of late active inflammation to chronic inflammation, fibrosis, resolution of inflammation, and the beginning of normal homeostatic conditions. Another recent study of the chronic colitis model described three phases in the model, acute inflammation from 0 to 14 days, nonprogressive inflammation lasting from days 14 to 35, and the last phase as chronic inflammation from days 35 to 49 or weeks 5 to 7 (26). By comparison, our study focused on chronic inflammation (TNBS-w6) and healing investigated at 2-wk intervals (TNBS-w8, TNBS-w10, and TNBS-w12). Histology and immunohistology indicated a resolution of inflammation by TNBS-w8, while low levels of fibrosis persisted as late as TNBS-w12, 6 wk after the last TNBS treatment. Our temporal gene expression study provided a continuous view of proinflammatory and profibrogenic changes as chronic inflammation shifted to fibrosis. We further investigated the effects of suppressing inflammation on fibrosis by inhibiting NF-κB. An antisense oligonucleotide directed against the NF-κB p65 subunit given intrarectally reduced the total NF-κB pool but did not lower it to basal levels. Several proinflammatory genes were amenable to complete suppression at this reduced NF-κB level, while others were not. Furthermore, the fibrogenic collagen I and III genes were not suppressed adequately at this level of NF-κB inhibition.

At TNBS-w6, the earliest time point of the study, the gene expression profiles indicated up-regulation of cytokines, chemokines, growth factors, receptors, and transcription factors (Il1b, Il1rl1, Ccl2, Ccl7, Ccl9, Ccr2, Ccr5, Cxcl1, Cxcl5, Cxcl9, Reg2, Reg3a, and Stat1) that promoted cell growth and served as inflammatory cell chemoattractants and inducers of stress response. Normal healing is a self-limiting process where proinflammatory signals are balanced by healing and tissue/ECM rebuilding signals. TNF-α is the proinflammatory cytokine prototype (27, 28), while TGF-β is a key immunosuppressive anti-inflammatory cytokine/growth factor that has been shown to induce the profibrogenic myofibroblast phenotype in fibroblasts and fibrosis in experimental colitis models (29, 30, 31, 32, 33, 34, 35, 36, 37, 38). There was no measurable increase in TNF-α at the chronic inflammatory stage in TNBS-treated mice. Because TNF-α is an early cytokine in the inflammation time scale, its expression may be back to normal by TNBS-w6. This may also be the reason for not detecting TNF-α overexpression in earlier microarray studies of long-term IBD patients (16, 39). Fichtner-Feigl (26) reported IL-13 as another cytokine that triggers TGF-β1-dependent fibrosis in the TNBS-induced chronic colitis model examined around the week 6 time point. Our microarray did not detect an increase in IL-13 mRNA and there are several possible explanations for this. Cytokine gene expressions can show subtle, transient changes that may go undetected by microarray profiling. Furthermore, in the Fichtner-Feigl study (26), the cytokines were detected in TNBS-treated mouse colonic lamina propria mononuclear cells simulated in culture with anti-CD3 Ab. The approaches followed by these studies (11, 26, 40) were quite different from our direct study of colonic tissue total RNA and protein. Nevertheless, these different approaches complement each other in their investigations of the initiation, development, and complications of chronic colitis and fibrosis. Emerging concepts highlight IL-23 as an early cytokine that ultimately leads to the induction of IL-13- and IL-4-mediated Th2 responses and TGF-β1-centered profibrogenic events (11, 26). Our differential gene expression profiles are likely to reveal downstream targets regulated by these cytokines.

An imbalance between proinflammatory and anti-inflammatory signals is believed to underlie abnormal ECM buildup, tissue distortion, and fibrosis (2, 41). Thus, we asked whether in our chronic colitis model there was sustained overexpression of genes that might contribute to inflammation and fibrosis. A set of inflammation-related genes continued above mean expression into TNBS-w8 and beyond (Ccl2, Ccl7, Pap, Reg3g, Saa3). The Pap and Reg3g genes have now been consistently associated with IBD and chronic inflammation and may play a role in innate immune functions and bacterial colonization of the gut (42). Pap has been suggested to down-regulate NF-κB and suppress proinflammatory signals (43, 44). Whether these have profibrogenic influence remains to be seen. Saa3 encodes a serum amyloid factor and is recognized as an inflammation marker. Its high expression is maintained for almost 4 wk after the last TNBS treatment. Serum amyloid A factor binds ECM proteins and such ECM-SAA complexes have been suggested to regulate adhesion and activation of CD4⁺ T lymphocytes (45). Continued expression of Saa3 is consistent with the idea that this mouse model of chronic colitis is T cell driven and ECM-bound serum amyloid A could be aiding fibrogenic changes in the tissue. We detected sustained expression of Ccl2 (Mcp1) and Ccl7 (Mcp3) in TNBS-w8 and Ccl2 even 4 wk later at TNBS-w10. These encode chemokine ligands, MCP-1/CCL2 and MCP-3/CCL7, also known as monocyte chemoattractant proteins, which have been linked to recruitment of monocytes and lymphocytes in chronic inflammatory diseases. MCP-1/CCL2 may promote abnormal recruitment of leukocytes and pulmonary fibrosis. Increased MCP-3 has been associated with severity of systemic sclerosis and pulmonary fibrosis (46). A recent study has shown Ccl2 to drive intestinal fibrosis in a mouse model, where intramural transfer of the Ccl2 gene by adenoviral vector injection caused a transient increase in TGF-β1 in the acute phase and increased deposition of collagen in the colon (25). We speculated that residual persistent fibrosis in TNBS-induced colitis that received the NF-κB blockade treatment could be due to poor inhibition of inflammatory genes. Among these, Ccl2, for example, could tip the balance toward fibrosis. Indeed, Ccl2 expression was not inhibited completely by NF-κB blockade unlike many other proinflammatory genes like Cxcl1, Il1b, Pla2g2a, Reg2, Reg3g, and Saa3 that were.

Among fibrogenic ECM genes, Col1a1, Col1a2, and Col3a1 encoding collagen types I and III were overexpressed at the same time as several inflammation mediators. Their expression at the mRNA and protein levels continued after expression of inflammation-related genes had resumed basal levels. Earlier studies have also linked altered amounts of collagen I and III with fibrosis (38, 47, 48, 49). Increase in the ECM protein lumican (Lum) in colitis was a novel finding. Lumican is a proteoglycan that in the stable mature ECM is found in close association with collagen fibrils and ample evidence supports its role in the regulation of the collagen fibril structure (14, 50, 51, 52). The Lum transcript was elevated sharply at TNBS-w6 and remained up-regulated after collagen expression had subsided. Immunohistology indicated increased presence of the lumican protein in the submucosa. Distribution of lumican was similar to the fibrillar collagens. We speculate that altered ratio of lumican:collagen may affect the assembly and architecture of the newly deposited collagen. In addition to regulating collagen structure, recent studies suggest regulation of cellular functions by lumican. Thus, in the injured cornea lumican plays a role in aiding inflammatory cell influx and cellular apoptosis as well (53, 54). It may have a similar role in regulating inflammatory cell influx in chronic colitis.

Increased remodeling of the ECM has been suggested to keep fibrogenic collagen deposition in check. Accordingly, our earlier study has shown that ulcerative colitis, less prone to fibrosis, expressed a wider variety of matrix metalloproteinase (16). Our current temporal gene expression pattern showed a massive overexpression of Timp1, encoding a matrix metalloproteinase inhibitor at the later stages (11 and 118 times the basal level of expression in TNBS-w8 and TNBS-w10, respectively) that may be counterproductive to fibrosis-reducing remodeling of the ECM. Theiss et al. (55) have shown recently that TNF-α, in conjunction with insulin-like growth factor can induce Timp1 in myofibroblasts that deposit fibrotic tissue in the submucosa. Finally, proteolytic changes in the ECM may contribute to altered interactions of the ECM with immune and nonimmune cells that affect fibrosis (56). In agreement with this idea is our observation that many genes encoding proteolytic enzymes, trypsin, lysozyme, mast cell proteases, elastases and serine proteases, were elevated in TNBS-induced colitis. Increased proteolytic activity has also been reported in IBD and considered to be a major cause of tissue damage (57, 58).

In summary, our study identified novel inflammation mediators and ECM proteins in chronic colitis. These may play significant roles in chronic inflammation and establishment of fibrosis, providing potential targets for IBD therapy. Our study further suggests the presence of two groups of proinflammatory genes, one easily suppressed by partial NF-κB inhibition, while a second, as indicated by Ccl2, may maintain high activity at lower levels of NF-κB. This raises the possibility that for effective treatment of inflammation and fibrosis early and complete inhibition of NF-κB may be necessary. However, NF-κB is also required for cell survival and other biologically beneficial events and the long-term effects of complete abrogation of NF-κB may be less than desirable. Therefore, effective therapeutic regimens may be developed by partial inhibition of NF-κB accompanied with additional targeted inhibition of specific proinflammatory signals.

The authors have no financial conflict of interest.