Disease activity index
Dextran sulfate sodium
Inflammatory bowel disease
Mucosal addressin cell adhesion molecule
Terminal deoxynucleotide transferase labeling
Crypt cell renewal is essential for normal intestinal homeostasis as well as mucosal regeneration following injury. However, the factors regulating crypt cell growth in pathological conditionsare not fully understood. We report here that the endogenously produced chemokine CXCL10 regulates crypt cell proliferation. CXCL10 was constitutively expressed by basal crypts in mouse colon, but the expression of CXCL10 as well as CXCR3 was enhanced in the epithelium in the proliferative zone after oral administration of dextran sulfate sodium. Neutralization of CXCL10 protected mice from epithelial ulceration by promoting crypt cell survival without evidence of altered immune cell infiltration. Furthermore, recombinant CXCL10 administration into mice inhibited intestinal epithelial cell proliferation. These findings suggest that CXCL10 regulates crypt cell growth to maintain intestinal homeostasis in an autocrine or paracrine fashion. Thus, CXCL10 can be a new therapeutic target for inflammatory bowel disease by controlling the dynamics of epithelial homeostasis.
Ulcerative colitis (UC) is a chronic, relapsing inflammatory bowel disease (IBD) affecting the colonic mucosa; it is of unknown etiology. UC clinically displays bloody diarrhea, abdominal painand weight loss, and often leads to a severe outcome since the therapeutic approaches, including corticosteroid and 5-aminosalicylic acid, are not always successful in inducing long-term remission 1. Although the incidence of UC varies from 1 to 15 per 100,000 population in different locations, it increases year by year. Moreover, UC generally acts upon young people, and approximately 30% of patients with UC will undergo colectomy in the course of their life-times 2. Thus, UC is a serious disease affecting the quality of life for a long period. It is now considered that IBD is not a simple inflammatory disease but a rather complicated disorder of intestinal components including epithelial cells, immune cells, neural cells, and extracellular matrix 3–6. Particular types of microbes, and genetic factors are also suggested to be involved 3. Global approaches are therefore needed to establishthe mechanisms of IBD.
Recent investigations have highlighted immunological disorders, especially the balance of Th1 and Th2 subsets and their trafficking into the inflamed colon 5–7. Chemokines have been revealed to play a central role in directing immune cell migration 8, 9, and thereby most likely contribute to induction and modulation of IBD 9. Thymus-expressed chemokine (TECK)/CCL25 is selectively expressed by the crypt epithelium in the small intestine, recruits CCR9+ lymphocytes into small bowel (but not colonic mucosa), and is involved in small-bowel Crohn's disease 10, 11. Mucosae-associated epithelial chemokine (MEC)/CCL28 is expressed in the colonic(but not small-bowel) epithelium, and possibly attracts CCR3+ and CCR10+ lymphocytes into the normal colonic mucosa 12. Various chemokines expressed by intestinal epithelium 9–14 are therefore expected to recruit specific types of leukocyte to sites of each intestinal compartment. Among these chemokines, interferon-inducible protein-10 (IP-10)/CXCL10 is reported to be expressed in the colonic epithelium or lamina propria under both physiological and pathological conditions in humans 13, 14. However, the biological roles of chemokines remain largely uncertain in vivo in disease states. To establish the role of CXCL10 in IBD, we investigated the effect of anti-CXCL10 mAb on colonic injury, using a mouse model for UC induced by oral administration with dextran sulfate sodium (DSS) 15.
2.1 Production of CXCL10 by proliferative zone epithelium
Colonic epithelial injury was induced in mice by administration through drinking water containing 5% DSS 15. DSS is reported to initially damage the basal crypt without inflammatory infiltration 15, 16 and to later cause colitis with mucosal inflammatory infiltration predominantly into the distal colon, resembling UC 15, 16. The expression level of CXCL10 mRNA in colon tissues was significantly increased by day 1, reached a peak on day 3, and decreased by day 5 (Fig. 1A). Expression of mRNA for CXCR3, a receptor for CXCL10, was rapidly up-regulated (by day 1) and down-regulated thereafter (Fig. 1A). CXCL10 was selectively expressed at the base of the colonic crypts in normal mice (Fig. 1B), but CXCL10-producing epithelial cells rapidly expanded to the basal one-third/one-half of crypts 1 day after DSS administration (Fig. 1C). Non-epithelial cells producing CXCL10 were scattered in the epithelium but not in the lamina propria (Fig. 1C, data not shown). CXCR3 was constitutively expressed by the luminal surface epithelium and, to a lesser extent, by basal crypts (Fig. 1D), non-epithelial cells in the lamina propria, and cells with dendritic morphology in the colon patch (data not shown). Following DSS administration, however, CXCR3+ cells disappeared from the luminal surface and increased within the basal one-half/two-thirds of crypts (Fig. 1E), similar to the distribution pattern of CXCL10 (Fig. 1C).
2.2 Blockade of CXCL10 protects mice from acute colitis
To reveal the role of CXCL10 in colon injury, we investigated the effect of anti-CXCL10 mAb on DSS-induced colitis. Mice were injected i.v. with anti-CXCL10 mAb or control mAb on days 0, 2, and 4 after DSS administration. Control mAb-treated mice exhibited 10–20% weight loss, together with diarrhea, and gross bleeding; however, neutralization of CXCL10 ameliorated clinical disease severity from day 4 (Fig. 2A). The decrease in colon length on day 5, which reflects the extent of colon damage in this model 15, 16, was minimal in anti-CXCL10 mAb-treated mice compared with the control mAb-treated mice (Fig. 2B). Although broad mucosal ulceration and degradation were observed in control mAb-treated mice at day 5 (Fig. 2C–F), the mucosal architecture was well preserved in anti-CXCL10 mAb-treated mice (Fig. 2G, H).
2.3 Blockade of CXCL10 protects colonic mucosa from massive apoptosis without evidence of controlling immune cell trafficking
Because CXCL10 has been shown to attract activated T, especially Th1, cells 17, 18 and to promote angiostasis 19, 20, two possible effects of anti-CXCL10 mAb were initially considered: inhibition of inflammatory Th1 cell migration or promotion of angiogenesis. However, histological scores revealed that blockade of CXCL10 did not affect the inflammatory infiltrate in the lamina propria (Fig. 2I). We further characterized the inflammatory cell components in situ. Although DSS-induced mucosal inflammation was accompanied with a significant infiltration of F4/80+ macrophages, CD11c+ dendritic cells, and CD4+ T cells (Fig. 3A, C, E), but not of NK cells and CD8+ T cells in the lamina (Fig. 3G), the numbers of these cells were not statistically different between control mice and anti-CXCL10-mAb-treated mice (Fig. 3B, D, F, G). Staining using terminal deoxynucleotide transferase labeling (TUNEL) revealed that most mucosal infiltration was characterized by apoptotic cells (Fig. 4A). Treatment with anti-CXCL10 mAb dramatically reduced the number of apoptotic cells (Fig. 4B). Moreover, we did not detect significant alteration of mucosal addressin cell adhesion molecule (MAdCAM)-1+ endothelial venules in the lamina propria (Fig. 4C, D) throughout the study. These findings necessitated the clarification of other roles of CXCL10 in this model.
2.4 Blockade of CXCL10 enhances the number of surviving crypts
The effect of anti-CXCL10 mAb on colitis appeared to be restricted to the crypt epithelial damage (Fig. 2I) as also evidenced by vicia villosa lectin staining (Fig. 4E, F). Since DSS was also reported to reduce cell proliferation both in vivo21 and in vitro in epithelial cell lines 22, we next examined the epithelial turnover, by in vivo 5-bromo-2′ deoxyuridine (BrdU) labeling 1 h before killing 21, 23, 24. In normal mice, BrdU+ cells were selectively located in the basal one-third of crypts 21, 24. However, mice exposed to DSS for 5 days exhibited a significant reduction in BrdU+ cells (Fig. 5A), suggesting that crypt cell growth stood impaired by this time. In contrast, many BrdU+ cells emerged within the replicative zone of intestinal crypts following blockade of CXCL10 (Fig. 5B). Surviving crypts, as determined by the proportion of BrdU+ crypt cells 25, were markedly increased in anti-CXCL10 mAb-treated mice (Fig. 5C). Interestingly, the numbers of surviving crypts in anti-CXCL10 mAb-treated mice were significantly higher than those in normal mice (Fig. 5C). These findings indicated that neutralization of CXCL10 protected crypt cells from CXCL10-induced growth inhibition or apoptosis, leading to the promotion of re-epithelialization and protection of mice from epithelial injury in this model.
2.5 CXCL10 inhibits crypt cell proliferation and migration in vivo
To clarify whether CXCL10 directly inhibits re-epithelialization or promotes epithelial apoptosis, we investigated the effect of recombinant CXCL10 (rCXCL10) on crypt cell proliferation and death in vivo. Mice were pretreated with vehicle (Fig. 6A) or rCXCL10 (Fig. 6B) for 3 days and received BrdU 2 h before killing. Mice treated with rCXCL10 showed an approximately 50% decrease in number of BrdU+ cells within the proliferative zone (Fig. 6B, C), indicating that crypt cell proliferation was inhibited by rCXCL10. Epithelial cell migration was also evaluated by the position of BrdU+ cells 72 h after BrdU incorporation 21, 24. Most BrdU+ cells disappeared from the epithelium; if any remained, they were located near the luminal surface in control mice (Fig. 6D). In contrast, in rCXCL10-treated mice, BrdU+ cells remained in the proliferative zone from 2–72 h after BrdU administration (Fig. 6E, F). This suggested that rCXCL10 directly inhibited not only crypt cell proliferation but also migration in vivo. On the other hand, treatment with rCXCL10 did not increase the number of apoptotic epithelial cells in TUNEL assays (data not shown).
Intestinal integrity is orchestrated by environmental cellular events and functions. The life-cycle of epithelial cells is of particular interest since mucosal structure is maintained by a constant renewal of the cells within the crypt of Lieberkühn 24. Each intestinal epithelial cell is located strictly along the crypt–villus axis, depending on the maturation stage of the cell. In the colon, pluripotential stem cells at the crypt base have been suggested to reproduce themselves as well as produce differentiated progeny 24. Transit amplifying (TA) cells subsequently migrate upward and locate in the basal one-half of the crypts, a proliferative zone of the colon. During maturation, the postmitotic differentiated cells move to and settle at the villus tip, and are finally shed from the luminal surface 24. A multistep migratory process of epithelial cells is pivotal to steady-state tissue homeostasis and mucosal protection, since impaired migration often leads to intestinal disorders, including IBD and also neoplasm 26, 27. Although several growth factors may contribute to the multistep processes by stimulating proliferation or maturation of the epithelial cells 21, 24, 28, the mechanisms regulating hierarchical positioning and in vivo migration of epithelial cells are not fully understood.
Epithelial repair is immediately accompanied by an inflammatory reaction, and epithelial stem cells are induced to enter DNA synthesis as a result of wounding 29, 30. We have shown here that blockade of CXCL10 enhanced BrdU+ TA cells following acute DSS-induced colon injury (Fig. 5). This might have caused rapid epithelial regeneration, as demonstrated by histological analysis showing typical regenerative changes (Fig. 2G, H) and by higher number of surviving crypts compared with untreated mice (Fig. 5C). Although DSS-induced chronic colitis has been reported to be a Th1-dominant disease 6, 31, we could not confirm significant Th1 cell infiltration in situ32 by day 5. Since acute DSS-induced colitis could be induced even in SCID mice, which lack T cells 22, it is likely that DSS directly injures the epithelium. This suggested that the protective effect of anti-CXCL10 mAb was mostly on the epithelium, but not on the lamina propria, at least during the acute phase of colon injury.
We have also demonstrated that CXCL10 can directly inhibit crypt cell proliferation in vivo (Fig. 6A–C). Since we could not detect increased epithelial cell apoptosis in situ following administration of rCXCL10 (data not shown), the CXCL10-induced reduction in the number of BrdU+ TA cells was virtually due to cytostasis, at least at steady-state, although CXCL10-mediated epithelial apoptosis in inflammatory conditions remains to be elucidated. Interestingly, CXCL10 and its receptor CXCR3 were co-produced by epithelial cells in the proliferative zone (Fig. 1C, E), suggesting a mechanism of autocrine/paracrine regulation. In the normal colon, binding of CXCL10 to CXCR3 may inhibit stem cell proliferation to optimize the number of cycling cells for intestinal homeostasis. Overproduction of CXCL10, possibly induced by proinflammatory cytokines 33, 34, was accompanied with the reduction of CXCR3 expression 3 days after DSS administration (Fig. 1A). In this context, it was reported that CXCR3 is expressed by cycling endothelial cells as well, and thatCXCL10 can inhibit endothelial proliferation after binding to CXCR3 20. A similar mechanism regulating proliferation through CXCL10–CXCR3 interaction in colonic epithelium is suggested.
CXCL10 may also inhibit epithelial cell migration, as evidenced by impaired migration of TA cells (Fig. 6D–F). Although cells are expected to migrate toward chemoattractants even in the epithelium, no direct evidence of chemokines controlling epithelial cellular upward migration is available. In this respect, epidermal growth factor (EGF) is known to promote epithelial cellular migration 35. Since CXCL10 can inhibit EGF-induced fibroblast migration invitro 36, the inhibitory effect of CXCL10 may partially depend on EGF-receptor-mediated activity. On the other hand, chemokines generally have a capacity to activate adhesion molecules on various cell types, leading to cellular arrest or retention 8, 37. CXCL10-mediated control of integrin or cadher in expression may contribute to firm cellular interactions between neighboring cells or with the extracellular matrix immediately beneath the epithelium. Since distribution of epithelial integrins correlates with differentiation of epithelial cell lineages 24, it is also speculated that distinct expressions of chemokines and their receptors may contribute to determine the hierarchical positioning of epithelial cells.
Although several chemokines were reported to be involved in skin wound healing, possibly promoting leukocyte as well as epidermal cell migration 29, our findings suggested the novel role of chemokines in controlling epithelial cell positioning in the gut. Crypt cells are continually renewed to maintain intestinal homeostasis. The appropriate numbers and positions of crypt cell lineages must be strictly regulated, since excessive proliferation may amplify the risk of carcinogenesis. CXCL10–CXCR3 interaction may contribute to maintain the stem cell niche by inhibiting its excessive proliferation through an autocrine/paracrine mechanism. Inhibition of cell migration is a rational operation to leave stem cells where they are. In addition, this study also suggests that UC might result from an epithelial disorder. This may help explain the long-standing questions about why UC shows continuous mucosal lesions and tends to form epithelial carcinoma. Blockade of CXCL10 could become established as a novel and essential therapeutic intervention in the active phase of UC.
4 Materials and methods
4.1 Induction of colitis
Specific pathogen-free female C57BL/6 mice (7–8 weeks old) were obtained from Charles River Japan Inc. (Yokohama, Japan). Colitis was induced in mice by the administration with 5% DSS (Mr 5,000; Wako, Osaka, Japan) in distilled water ad libitum for 5 days. Two mice were housed per cage and DSS/water consumption was monitored daily. Body weight, stool consistency (scores: 0, normal stools; 1, soft stools; 2, liquid stools), hemoccult-positivity and presence of gross blood (scores: 0, negative fecal occult blood; 1, positive fecal occult blood; 2, visible rectal bleeding) were assessed daily. The disease activity index (DAI) was determined as a combination of the above parameters according to the scoring criteria described in 16. For blocking experiments, PBS containing 100 μg/100 μl anti-CXCL10 mAb 34, 38 or anti-human parathyroid-related peptide mAb, which was the IgG1 subclass-matched control mAb, or PBS alone were administered i.v. 0, 2, and 4 days after DSS administration. Results from mice treated with PBS alone were similar to those treated with control mAb, thus they are not described inthe text. All animal experiments complied with the standards set out in the guidelines of Niigata University Graduate School of Medical and Dental Sciences.
4.2 Evaluation of histology
The degree of colonic injury was assessed by colon length and histological score. The entire colon (10 mice per group) was sampled and its length immediately recorded. The entire colon was fixed in 4% formalin, embedded in paraffin, and transverse sections were stained with hematoxylin and eosin 17, 32. According to the preliminary histological study, we presented the distal colon tissue section located approximately 10 mm away from anal verge in all photographs. Sequential high-power fields of the entire colon were evaluated histologically by atleast three external pathologists. Histology was scored as follows 16. Epithelium: 0, normal morphology; 1, loss of goblet cells; 2, loss of goblet cells in large areas; 3, loss of crypts; 4, loss of crypts in large areas. Infiltration: 0, no infiltrate; 1, infiltrate around crypt basis; 2, infiltrate reaching the L. muscularis mucosae; 3, extensive infiltration reaching the L. muscularis mucosae and thickening of the mucosa with abundant edema; 4, infiltration of the L. submucosa.
4.3 Immunohistochemical staining
The following anti-mouse mAb were used: CD4 (clone: RM4–5), CD8α (53–6.7), pan-NK cell (DX5), MAdCAM-1 (MECA-367) (all from BD PharMingen); F4/80 (CI:A3–1; BMA Biomedicals); and CD11c (N418; Serotec). To detect crypt epithelium, anti-cytokeratin polyclonal Ab (pAb) (Biomedical Technologies Inc.) or biotinylated vicia villosa lectin (Vector) were used. As secondary Ab, a horseradish peroxidase (HRP)-labeled anti-rat Ig (Biosource) or anti-Hamster IgG were used 23. For immunostaining of CXCL10 or CXCR3, goat pAb to CXCL10 or CXCR3 (Santa Cruz Biotechnology, Santa Cruz, CA) were used 23, 38–40. The specificity of both anti-CXCL10 mAb and anti-CXCL10 pAb was confirmed immunologically against murine CXCL9,CCL3, CCL21, and CCL22 38. The specificity of the goat pAb was also examined by immunohistochemistry. Anti-CXCL10 reacted with a subset of hepatocytes and lymph node dendritic cells in Propionibacterium acnes-induced Th1 granuloma-laden mice 38. Anti-CXCR3 reacted with P. acnes-treated lymph node cells (Fig. 1F, G); most of these cells were reported to be Th1 cells 38, and CXCR3 was also reported to be expressed in brain-infiltrating CD4+ T cells in mouse hepatitis virus-infected mice 40. Single or triple immunostaining was performed by indirect immunoperoxidase methods 23.
All mice were injected with BrdU (Sigma) (500 μg/100 μl in PBS) 1 h before killing 23. Acetone-fixed 5 μm fresh-frozen colon sections were incubated with primaryAb against CXCL10 or CXCR3 for 30 min at 37°C, and subsequently with FITC-labeled anti-goat IgG 32. They were incubated with anti-cytokeratin pAb followed by Alexa-594 (MolecularProbes Inc.)-conjugated anti-rabbit IgG, and were observed by fluorescence microscopy 32. To detect replicating cells, tissue sections were reacted with the Ab and reagents usinga BrdU staining kit (ZYMED, South San Francisco, CA) according to the manufacturer's instruction 23. Crypts that had five or more BrdU-labeled nuclei were defined as surviving crypts 23. The numbers of surviving crypts were compared with each group.
4.4 Terminal deoxynucleotide transferase labeling
Apoptotic cells were identified using an in situ apoptosis-detection kit (Takara Biomedicals, Japan) according to the manufacturer's instruction. In brief, acetone-fixed 5 μmfresh-frozen colon sections were permeabilized on ice and incubated with the terminal deoxynucleotide transferase mixture for 1 hr at 37°C. FITC-labeled dNTP were treated with anti-FITC HRP for 30 min, visualized with DAB, and counter-stained with hematoxylin.
4.5 Quantitative RT-PCR
Total RNA was extracted from the colon by homogenization using RNAzolB (Biotex Laboratories) and reverse transcribed 17, 23. Thereafter, cDNA was amplified using the ABI 7700 sequence-detector system (Applied Biosystems, Foster City, CA) with a set of primers and probes corresponding to CXCL10, CXCR3, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as previously described 17, 23.
4.6 In vivo epithelial proliferation and migration
Mice were pretreated with rCXCL10 34 (2 mg/kg/day, i.v.) or vehicle for 3 days, and injected with BrdU 24 h after final pretreatment. The mice were killed 2 or 72 h later, and BrdU incorporation in colon tissue was analyzed as described above. Only crypts longitudinally sectioned and visible in their entire length were estimated. At least 15 mm2 areas of stained cryosections were examined for the numbers of BrdU-labeled nuclei per crypt. The distances of the lowest labeled cells from crypts were also measured.
4.7 Statistical analysis
Differences were evaluated using Student's t test. p values <0.05 were considered to be statistically significant.
We thank M. Nomoto for histological analysis, and T. Tsuchida for technical assistance.