Crohn’s disease (CD) is a type of inflammatory bowel disease associated with increased Th1 cytokines and unique pathological features. However, its pathogenesis has not been fully understood. Previous studies showed that homologous to lymphotoxin, exhibits inducible expression, competes with herpesvirus glycoprotein D for HVEM on T cells (LIGHT) transgenic (Tg) mice develop autoimmunity including intestinal inflammation with a variable time course. In this study, we establish an experimental model for CD by adoptive transfer of Tg mesenteric lymph node cells into RAG−/− mice. The recipients of Tg lymphocytes rapidly develop a disease strikingly similar to the key pathologic features and cytokine characterization observed in CD. We demonstrate that, as a costimulatory molecule, LIGHT preferentially drives Th1 responses. LIGHT-mediated intestinal disease is dependent on both of its identified signaling receptors, lymphotoxin β receptor and herpes virus entry mediator, because LIGHT Tg mesenteric lymph node cells do not cause intestinal inflammation when transferred into the lymphotoxin β receptor-deficient mice, and herpes virus entry mediator on donor T cells is required for the full development of disease. Furthermore, we demonstrated that up-regulation of LIGHT is associated with active CD. These data establish a new mouse model resembling CD and suggest that up-regulation of LIGHT may be an important mediator of CD pathogenesis.
Inflammatory bowel disease (IBD)4 is characterized by a chronic relapsing inflammation of the gastrointestinal tract and is divided into two primary forms, Crohn’s disease (CD) and ulcerative colitis (UC) (1, 2). The current working hypothesis suggests that an abnormal mucosal T cell response to normal constituents of the mucosal microflora underlies the etiology of IBD (1). Specifically in CD patients, the T cell response exhibits a dominant Th1 phenotype (2) with production of large amounts of IFN-γ, TNF, and IL-12, which is a Th1-polarizing factor (3). Th1 cytokines have been specifically shown to be elevated in CD patients, whereas the elevation of Th2 cytokines appears to be more related to UC (2, 4, 5). Blockade of TNF with anti-TNF Abs has already proven to be a highly effective treatment in CD, and such treatment achieves positive effects in 60% of patients (6). Previous studies have shown potential mechanisms for anti-TNF action, such as induction of lymphocyte apoptosis (7, 8, 9, 10, 11, 12, 13).
CD is a complex disease with a multifactorial etiology (2, 4). Uncovering the mechanisms underlying CD has been difficult. There are three mouse models that resemble CD in disease location, histological features, and clinical course (i.e., IL-10 deficient, TNF, AU-rich regulatory element (ARE), and SAMP1/Yit mice) (14, 15, 16, 17, 18). Previous studies suggested that a Th1 phenotype is strongly associated with active CD, and the majority of animal models of CD are characterized by overproduction of Th1 cytokines (2). Studies have clearly indicated a convincing role for TNF, IL-12, and IL-18 in the pathogenesis of IBD with results from mice confirmed in human studies (11, 19, 20). However, the cytokines that act on T cells to perpetuate the dominant Th1 response in CD are undefined. Furthermore, the mechanisms by which T cells are activated, expanded, and recruited in the intestine are unclear.
Recent studies in IBD pathogenesis point to an interesting cytokine called homologous to lymphotoxin, exhibits inducible expression, competes with herpesvirus glycoprotein D for HVEM on T cells (LIGHT), a TNF superfamily member (TNF superfamily 14). LIGHT is primarily expressed on activated T cells (21, 22). LIGHT is closely related to TNF and together with lymphotoxin (LT) defines the TNF/LT core family (21, 23). LIGHT is a proinflammatory cytokine and potent T cell costimulatory molecule (24, 25, 26). Previous findings establish a crucial role for LIGHT in T cell activation and expansion by directly stimulating other T cells (26). The role of LIGHT’s involvement in IBD is hinted by several observations. First, LIGHT (27) maps to the region overlapping a susceptibility locus for IBD on human chromosome 19p13.3 (28, 29). Second, transgenic (Tg) mice with enhanced LIGHT expression on T cells develop colitis, suggesting possible contribution in the development of IBD (26, 30). Third, blocking the LT/LIGHT axis by LTβR-Ig in two animal models can ameliorate the severity of colitis (31). Despite the implications of these findings, the unique and direct role of LIGHT in CD pathogenesis has not been directly examined.
LIGHT has two signaling receptors (21); LTβR is expressed on stromal cells and nonlymphoid hemopoietic cells (32, 33, 34), and herpes virus entry mediator (HVEM) is expressed on T and B lymphocytes as well as other hemopoietic cells (25, 35, 36). LIGHT and LTα1β2 cooperate in lymphoid organogenesis and the development of lymphoid structures by signaling through the LTβR (37, 38). It has been shown that signaling LTβR via LIGHT transgene was sufficient to induce up-regulation of chemokines and adhesion molecule expression (38, 39). Presumably, the HVEM receptor is responsible for LIGHT-mediated T cell activation, although that has not been formally proven. Moreover, the distinct role of each receptor, HVEM and LTβR, in LIGHT-mediated colonic pathogenesis is unclear.
We have developed a Tg model overexpressing LIGHT on T cells that develop autoimmunity including intestinal inflammation, but the time course for the development of colitis was variable in LIGHT Tg mice (39). To establish a consistent and reliable animal model in a timely fashion for the understanding of IBD pathogenesis and developing treatment, we transfer Tg mesenteric lymph node (MLN) cells into RAG−/− mice and monitor the intestinal inflammation. The transfer of Tg MLN cells or naive thymic T cells into RAG−/− recipient results in severe intestinal inflammation in 5–6 wk with features reminiscent of CD such as fissuring ulcers and ileitis, which were not present in the original LIGHT Tg mice. Blockade with TNFR-Ig effectively ameliorates the severity of disease. The mechanism by which LIGHT induces CD is to drive a Th1 response and promote activation/expansion of Th1 cells in the intestines via both HVEM and LTβR. Our results support the notion that an initial Th1 response underlies the pathological features of CD. Furthermore, we demonstrate that a dramatic increase of LIGHT expression is associated with active CD in humans. This mouse model not only replicates CD in the intestinal pathology but also helps us to explore the immunologic mechanism of CD, identify downstream target genes, and serve as a platform for therapeutic intervention.
Materials and Methods
LIGHT Tg mice were generated as previously described; proximal lck promoter and CD2 enhancer were used to generate the T cell-specific LIGHT Tg mice (40). C57BL/6 (B6) LIGHT Tg mice were crossed to LTβR−/− mice in B6 background. LIGHT Tg mice were crossed to HVEM−/− mice (crossed to B6 background for six generations) to obtain LIGHT Tg/HVEM−/− mice. RAG-1−/− mice (B6 background) were purchased from The Jackson Laboratory. Animal and human experimentation protocols were consistent with National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee at the University of Chicago.
Clinical samples were obtained from the multi-institutional tissue bank Ardais BIGR Library (Lexington, MA). The samples were selected by searching the Ardais library for cases of UC, CD, and ischemic bowel disease. The diagnosis was determined using all available clinical and pathologic information at the time of collection. Freshly frozen tissue was harvested into optimal cutting temperature embedding solution at the participating institutions according to the Ardais standardized protocol. Each selected sample contained an accompanying series of histological photomicrographs that were secondarily reviewed to confirm the diagnosis. Clinical information beyond the pathologic diagnosis was not available for the samples selected from the Ardais library: active CD (n = 5), inactive CD (n = 5), active UC (n = 5), inactive UC (n = 5), and controls with ischemic colitis (n = 2) and neoplastic disease (n = 5). All active vs inactive samples were paired from the same patient. The demographic characteristics of the patient population and therapy were not available due to the restriction of Institutional Review Board protocol.
The generation of HVEM KO mice
A genomic bacterial artificial chromosome clone containing the HVEM locus was obtained from Genome Systems by screening a bacterial artificial chromosome library with HVEM-specific primers, derived from the HVEM mRNA sequence (muHVEM2: 5′-atg gcc tga gca agt gtc tgc-3′; and Int3Rev: 5′-aca cac cct gag aac ctg ccc ac-3′). A ∼8.8-kb genomic subclone (with HindIII restriction sites) with the coding sequence of HVEM was subcloned and completely sequenced. The targeting vector was constructed using homologous fragments of the HVEM locus. Exons 2, 3, and the 5′ region of exon 4 was replaced by a intron gene trap cassette encompassing an engrailed two-splice acceptor sequence, a β-galactosidase cassette, and a neomycin gene resistance cassette. For negative selection, a viral thymidine kinase cassette was inserted into the targeting vector. Gene targeting in E14.1 ES cells was performed as described previously (41). Screening for homologous recombinant ES cell lines was performed by PCR using the primers HVEM-SP-5′ (5′-tcc ctg agg ctg aga ggt tcc-3′) and pJak2 (5′-ctg aag agg agt tta cgt cca g-3′). Two of 723 G418-resistant ES cell clones gave positive signals for homologous recombination and could be verified by Southern blot analysis (EcoRI digest) using the 5′ flanking probe. Single integration of the targeting vector was verified by Southern blot using a neomycin probe (NheI digest); here only one hybridization band was detectable (data not shown). Both ES cell clones containing the correctly targeted HVEM allele were injected into C57BL/6 blastocysts. The germline transmission of the inactivated HVEM locus of both cell lines was confirmed by PCR and Southern blot analysis. Heterozygous mice were backcrossed five times into the C57BL/6 strain. At the N5 generation, homozygous offspring was obtained by intercrossing HVEM+/− mice. In all experiments, littermates were used as controls. Genotyping for the HVEM alleles was performed by PCR using the primers HVEM-SP-5′/pJak2 for the inactivated allele and the primers HVEM-SP-5′/HVEM-SP-3′-WT (5′-aga ggc agc agg gtc agc tgg-3′) for the wild-type (wt) allele. The mice were housed and crossed to C57BL/6 in a specific pathogen-free animal facility. For verification of the inactivation of the HVEM gene, a Northern blot analysis of spleen RNA was performed using the cDNA encompassing the extracellular domain of the murine HVEM.
Lamina propria lymphocyte (LPL) purification
LPL purification was performed as previously described (42). Briefly, large bowel including cecum and colon was collected from the wt or Tg recipients of the adoptive transfer. Stool was removed, and samples were washed with RPMI 140 (Mediatech). Then intestine was subject to EDTA digestion (1 mM EDTA in 1× PBS) and serial collagenase digestions (1.5 mg/ml collagenase in RPMI 140) (Sigma-Aldrich). The supernatant of collagenase digestion was collected, resuspended in 40% Percoll, and loaded onto 75% Percoll gradient (Sigma-Aldrich). The interface was collected after 20-min centrifugation at room temperature with the brake off. The cells were washed and used for FACS or intracellular cytokine staining. The cell number for LPL is from the entire large intestine including cecum, descending colon, and rectum from either wt or Tg recipients.
Flow cytometry analysis and intracellular cytokine staining
Single-cell suspensions of splenocytes, lymph node (LN) cells, or LPL were collected and stained with the following Abs: PE-anti-CD8, FITC-anti-CD8, FITC-anti-CD4, CyChrome-anti-CD4, FITC-anti-CD69, CyChrome-anti-CD44, rat anti-mouse MAdCAM-1 (MECA 367; a gift from Dr. D. Chaplin, University of Alabama at Birmingham, Birmingham, AL), PE-anti-LPAM-1 (anti-α4β7; clone DATK32), PE-anti-B220, and FITC-anti-CD3 Abs (BD Pharmingen) in PBS plus 0.01% NaN3 for 30 min at 4°C. For biotin-labeled Ab, streptavidin-PE (Immunotech) was used as secondary Ab. For MAdCAM-1 staining, bio-goat-anti-rat IgG (BD Pharmingen) and streptavidin-PE were used. After incubation with Abs, the cells were washed and analyzed on a FACScan (BD Biosciences). Intracellular cytokine staining was performed as previously described (26). Allophycocyanin-anti-IFN-γ and FITC-anti-IL-4 were used for intracellular cytokine staining (BD Pharmingen). LPL purified from wt or Tg recipients was stimulated with 50 ng/ml PMA plus 500 ng/ml ionomycin for 4 h at 37°C in the presence of 10 μg/ml brefeldin A (Sigma-Aldrich).
H&E staining of intestine samples was performed as described previously (26). The mean degree of inflammation in the colon is calculated from observation of 20 different fields of H&E-stained longitudinal sections of colon from each animal. The degree of inflammation was graded as follows (31): 0, normal; 1, minor infiltration in the submucosa, presence of goblet cells, and limited elongation of the crypts; 2, large infiltration of leukocytes in the mucosa and submucosa, rare goblet cells, and elongation of the crypts; and 3, massive infiltration of leukocytes in the muscularis layer, submucosa, and mucosa, and no goblet cells.
Real-time PCR and cytokine beads assay
RNA was purified with TRIzol (Invitrogen Life Technologies) and RNA easy kit (Qiagen) following the instructions provided by the manufacturers. cDNA was synthesized and subject to real-time PCR as described previously (42). Each cDNA sample was amplified for the interested gene and GAPDH with the TaqMan Universal PCR master mixture according to the manufacturer’s instructions (Applied Biosystems). The concentration of the interested gene was determined using the comparative CT (threshold cycle number at a cross point between amplification plot and threshold) method and normalized to the internal GAPDH control. The following primers were used: human LIGHT forward primer, 5′-AGCAGCTGATACAAGAGCGAA-3′; human LIGHT reverse primer, 5′-CCGGTCAAGCTGGAGTTGG-3′; human LIGHT probe, 5′-FAM-AGGTCAACCCAGCAGCGCATCTCACA-TAMRA-3′; human GAPDH forward primer, 5′-GAAGGTGAAGGTCGGAGTCA-3′; human GAPDH reverse primer, 5′-GAAGATGGTGATGGGATTTC-3′; human GAPDH probe, 5′-TET-CAAGCTTCCCGTTCTCAGCC-TAMRA-3′; murine IFN-γ forward primer, 5′-TCAAGTGGCATAGATGTGGAAGAA-3′; murine IFN-γ reverse primer, 5′-TGGCTCTGCAGGATTTTCATG-3′; murine IFN-γ probe, 5′-FAM-TCACCATCCTTTTGCCAGTTCCTCCAG-3′; murine IL-12p40 forward primer, 5′-GGAAGCACGGCAGCAGAATA-3′; murine IL-12p40 reverse primer, 5′-AACTTGAGGGAGAAGTAGGAATGG-3′; murine IL-12p40 probe, 5′-FAM-CATCATCAAACCAGACCCGCCCAA-TAMRA-3′; murine IL-4 forward primer, 5′-ACAGGAGAAGGGACGCCAT-3′; murine IL-4 reverse primer, 5′-GAAGCCCTACAGACGAGCTCA-3′; murine IL-4 probe, 5′-FAM-TCCTCACAGCAACGAAGAACACCACA-TAMRA-3′; murine MAdCAM-1 forward primer, 5′-GACACCAGCTTGGGCAGTGT-3′; murine MAdCAM-1 reverse primer, 5′-CAGCATGCCCCGTACAGAG-3′; murine MAdCAM-1 probe, 5′-FAMCAGACCCTCCCAGGCAGCAGTATCC-TAMRA-3′; murine GAPDH forward primer, 5′- TTCACCACCATGGAGAAGGC-3′; murine GAPDH reverse primer, 5′-GGCATGGACTGTGGTCATGA-3′; murine GAPDH probe, 5′-TET-TGCATCCTGCACCACCAACTGCTTAG-TAMRA-3′. Cytokine beads assay (BD Pharmingen) was used to test the cytokine level of the culture supernatant or intestine homogenate following the instructions by the manufacturers (BD Pharmingen).
Cell culture and proliferation assay
wt (B6) LN cells (2 × 105/well) were stimulated with immobilized anti-CD3 mAb (0.2 μg/ml) in the absence or presence of rLIGHT produced by Escherichia coli (24) (10 μg/ml) for 72 h in 96-well plate. Anti-CD28 mAb (10 μg/ml) was used as positive control (StemCell Technologies). Culture supernatant was collected after 72 h and frozen at −20°C, then subject to cytokine beads assay as described by the instructions of the manufacturer (BD Pharmingen). wt or HVEM−/− splenocytes (2 × 105/well) were stimulated with immobilized anti-CD3 mAb (0.3 μg/ml) in the absence or presence of flag-LIGHT from CHO (12.5 ng/ml) for 72 h in 96-well plate, pulsed with 1 μCi of [3H]thymidine for 16 h, and then harvested for liquid scintillation counting.
Adoptive transfer of LN cells
MLN cells were collected from wt littermates or LIGHT Tg mice (2–3 mo old), and 6 × 106 LN cells were i.v. transferred into RAG-1−/− mice (B6, at the age of 6–7 wk). TNFR-Ig (a generous gift from Dr. J. Browning, Biogen, Boston, MA) was injected i.p. to Tg recipients (150 μg/mouse) every 5 days. LTα−/− or LTβR−/− mice (6–9 wk) were sublethally irradiated (700 rad) and received MLN cells from wt vs Tg mice. Disease progress was monitored including body weight and survival rate. Samples were collected 4 or 5 wk after MLN transfer and subjected to H&E staining.
LIGHT-induced pathology mimics CD
We have generated a Tg line that expresses LIGHT on T cell lineage, and Tg mice spontaneously develop intestinal inflammation several months after birth (39). The clinical presentation of the disease is highly variable, making the comparison between mice and between experiments difficult. To better understand the mechanism of LIGHT-mediated intestinal inflammation and synchronize disease progression, we developed an adoptive transfer model in which wt or Tg MLN-derived lymphocytes (at the age of 2–3 mo) were transferred into RAG-1−/− mice. Transgenic MLN T cells at this age did not increase in number or up-regulate activation markers including CD69 (Fig. 1,A), CD25, and CD44 (data not shown) as compared with wt controls. Adoptive transfer of Tg MLN cells, a major draining LN for both small and large bowel, into RAG-1−/− mice (Tg recipients) led to rapid and synchronized development of severe intestinal inflammation 4–5 wk after the transfer (Fig. 1,B). The inflammation was consistently accompanied by severe diarrhea and sustained weight loss, and often culminated in death (Fig. 1,B). In contrast, adoptive transfer of wt MLN cells into RAG-1−/− mice (wt recipients) did not cause disease (Fig. 1 B). To see whether this effect was primarily dependent upon T cells, naive T cells from wt or Tg thymus (>99% are T cells) were adoptively transferred into RAG-1−/− mice. The mice receiving Tg thymocytes developed diarrhea and wasting disease with similar intensity and time course as that seen in the adoptive transfer of Tg MLN cells (data not shown). These data suggest that up-regulation of LIGHT on T cells can sufficiently cause intestinal inflammation.
Impressively, the disease that develops in this experimental mouse model has several clinical and pathological features identical with CD. In addition to the classic clinical feature of diarrhea and weight loss, these mice demonstrate a number of pathologic features observed in Crohn’s disease: 1) the disease affects the terminal ileum (Fig. 1, M and N), cecum, and colon (Fig. 1, D–F), 2) the ulcerations are focal (Fig. 1,E) and often form deep “knife-like” fissure ulcers (Fig. 1,D, arrow), 3) there is transmural inflammation involving all the layers of the intestine and sometimes resulting in serositis (Fig. 1 E), 4) there are lymphocyte aggregates observed deep within the muscularis propria (Fig. 1 F, arrow), and 5) ulcer abscess was observed in Tg recipients (Fig. 1,O) and granuloma formation could also be identified (data not shown); all of which are typically seen in patients with CD (Fig. 1, H–J). Interestingly, these CD-like features and ileum involvement are not observed in parental Tg mice. Therefore, our data indicated that intestinal inflammation in this novel animal model is transferable through T cells with dysregulated LIGHT expression and closely resembles intestinal pathology observed in CD patients.
LIGHT on T cells promotes Th1 phenotypes in the gut
It has been shown that the responding T cells in CD patients exhibit a Th1 phenotype and produce large amounts of IFN-γ and TNF-α (2). The role of LIGHT in T cell differentiation has not been defined in vivo, although in vitro data suggested that LIGHT costimulation promotes IFN-γ production (24). We first examined the cytokine profiles in the reconstituted RAG-1−/− recipients. Total RNA purified from colon tissues were subjected to real-time RT-PCR for different cytokines. Our data showed that there was a dramatic increase in the amount of IFN-γ and IL-12 produced in Tg recipients. In contrast, the mRNA expression of IL-4 was reduced in the colon of Tg recipients (Fig. 2,A). Cytokine bead analysis confirmed the mRNA analysis with a remarkable amount of TNF-α and IFN-γ produced in the colon of Tg recipients. The production of IL-5 and IL-4 was either reduced or not significantly different between wt and Tg recipients (Fig. 2 B). These data suggest that our mouse model displays a dominant Th1 cytokine production in the colon, consistent with the cytokine profile reported in CD patients.
To further dissect whether T cells in the gut of the Tg recipients are more active and produce higher level of Th1 cytokines, we purified LPL from the large bowel of wt or Tg recipients. There was a dramatic expansion in the percentage of lymphocytes in the large bowel of Tg recipients, in comparison to that of wt recipients (Fig. 2,C, 2.84% in wt vs 21.79% in Tg). The percentage of both CD4+ T cells (Fig. 2,C, 2.04% in wt vs 58.34% in Tg) and CD8+ T cells (Fig. 2,C, 0.75% in wt vs 17.55% in Tg) were markedly increased. The absolute numbers of CD4+ and CD8+ T cells in the LPL were also dramatically elevated in Tg recipients (Fig. 2,C). The expanded Tg T cells were hyperactivated relative to transferred wt T cells as indicated by high levels of CD69 and CD44 (data not shown). To test the IFN-γ production by T cells at the single-cell level, intracellular cytokine staining was performed after stimulating the LPL with PMA and ionomycin. IFN-γ-producing CD4+ T cells were dramatically elevated in Tg recipients, whereas IFN-γ-producing CD4+ T cells were negligible in wt recipients (Fig. 2,D). Moreover, IFN-γ-producing CD8+ T cells were also significantly increased in Tg recipients (Fig. 2,D, CD8 vs IFN-γ), although the increase was not as dramatic as that seen in CD4+ T cells. Thus, our results showed the IFN-γ was produced primarily by CD4+ T cells in Tg recipients. In contrast, IL-4 producing CD4+ or CD8+ T cells were not increased in Tg recipients (Fig. 2,D, CD4 or CD8 vs IL-4). In fact, IL-4 production in wt LPL is much higher than that seen in Tg LPL (Fig. 2 D). This suggests that the LIGHT-mediated environment in the lamina propria might suppress the production of Th2 cytokines. Taken together, dysregulated LIGHT expression on T cells leads to CD-like disease through the promotion of Th1 responses in the gut.
LIGHT directly stimulates Th1 cytokines from T cells
To study whether the cytokine production was mediated by LIGHT costimulation acting directly on T cells, rLIGHT was used to stimulate T cells in combination with anti-CD3 mAb. LIGHT costimulation promoted Th1 cytokine production including IFN-γ and TNF-α but not Th2 cytokines such as IL-4 and IL-5 (Fig. 3,A). Anti-CD28 was used as a positive control, which stimulated both Th1 and Th2 cytokine production (Fig. 3,A). In addition, LIGHT and anti-CD28 can both increase IL-2 production as compared with anti-CD3 stimulation alone (Fig. 3 A). These data show that LIGHT can directly stimulate T cells to produce Th1 cytokines.
TNF-α is a major pathogenic cytokine in CD because blocking TNF is a very effective treatment for CD. Because LIGHT costimulation dramatically increased TNF-α production from T cells in vitro (Fig. 3) and in the colon of Tg recipients in vivo (Fig. 2), we next tested whether LIGHT-induced CD pathology could be attributed to TNF overproduction and attenuated by TNF blockade. Tg MLN cells were transferred into RAG-1−/− mice that were treated with TNFR-Ig. Our results showed that blocking TNF can significantly reduce the severity of disease (Table I, groups 1–3). Impressively, the CD-like pathology in the intestine of Tg recipients was almost entirely reversed by TNFR-Ig treatment (Fig. 1, K and L). These data suggested that, similar to CD, TNF is one of the major downstream effector cytokines in the LIGHT-mediated CD model.
|Group .||Donor .||Recipients .||Survivala .||Disease Score .|
|1||wt||RAG-1−/−||18/18||0.12 ± 0.32|
|2||Tg||RAG-1−/−||11/20||2.93 ± 0.25c|
|3||Tg||RAG-1−/− +TNFR-Ig||8/8||0.15 ± 0.36|
|4||wt||LTα−/−||5/5||0.08 ± 0.28|
|5||Tg||LTα−/−||1/3||1.55 ± 0.69b|
|6||wt||LTβR−/−||4/4||0.03 ± 0.18|
|7||Tg||LTβR−/−||7/7||0.00 ± 0.00|
|8||wt/HVEM−/−||RAG-1−/−||5/5||0.22 ± 0.42|
|9||Tg/HVEM−/−||RAG-1−/−||5/5||1.58 ± 0.73b|
|Group .||Donor .||Recipients .||Survivala .||Disease Score .|
|1||wt||RAG-1−/−||18/18||0.12 ± 0.32|
|2||Tg||RAG-1−/−||11/20||2.93 ± 0.25c|
|3||Tg||RAG-1−/− +TNFR-Ig||8/8||0.15 ± 0.36|
|4||wt||LTα−/−||5/5||0.08 ± 0.28|
|5||Tg||LTα−/−||1/3||1.55 ± 0.69b|
|6||wt||LTβR−/−||4/4||0.03 ± 0.18|
|7||Tg||LTβR−/−||7/7||0.00 ± 0.00|
|8||wt/HVEM−/−||RAG-1−/−||5/5||0.22 ± 0.42|
|9||Tg/HVEM−/−||RAG-1−/−||5/5||1.58 ± 0.73b|
Survival was monitored up to 5 wk after the transfer.
, p < 0.05 group 4 vs group 5, group 8 vs group 9;
, p < 0.01 group 1 vs group 2.
LIGHT-mediated costimulation is dependent on HVEM
Previous studies suggested that LIGHT functions as a costimulatory molecule for T cells (24, 25, 26). Consistent with this notion, rLIGHT can significantly enhance the expression of CD69 on CD4+ and CD8+ T cells upon anti-CD3 mAb stimulation (Fig. 3,B). The receptor mediating LIGHT costimulation has not been experimentally demonstrated, although HVEM on T cells has been considered a strong candidate. HVEM biology has not been well studied due to the lack of reagents and animal models. HVEM-deficient mice were, therefore, generated by standard gene-targeting approach, and there were no significant defects in cellular components of primary and secondary lymphoid tissues (data not shown). We examined whether LIGHT-mediated costimulation is dependent on HVEM signaling. Splenocytes isolated from wt or HVEM−/− mice were stimulated with anti-CD3 and rLIGHT or anti-CD3 mAb in vitro. Our results showed that LIGHT-induced costimulation was absent in HVEM−/− T cells as determined by proliferation assays (Fig. 4,A). In contrast, the proliferation of wt T cells was much higher in the presence of rLIGHT as compared with treatment with only anti-CD3 mAb (Fig. 4 A). Taken together, these results for the first time formally demonstrated that HVEM signaling is essential for LIGHT-mediated costimulation.
We introduced LIGHT transgene into HVEM−/− mice (Tg/HVEM−/−) and observed much reduced intestinal inflammation in Tg/HVEM−/− mice compared with Tg mice (data not shown). In addition, T cells in Tg/HVEM−/− mice did not show significant up-regulation of activation markers as compared with wt controls (data not shown). To examine the requirement for HVEM on T cells for mediating disease in our adoptive transfer model, we transferred the MLN cell from HVEM−/− or Tg/HVEM−/− mice into RAG-1−/− mice. MLN cells from Tg/HVEM−/− mice were not able to cause wasting disease and mortality (Table I, group 9). The RAG-1−/− mice receiving MLN cells from Tg/HVEM−/− mice (Tg/HVEM−/− recipients) developed much milder intestinal inflammation (Fig. 4,B) than that observed in the recipients of Tg MLN cells (Fig. 4,B). Specifically, the inflammation appeared to be confined to the epithelial layer without significant involvement of the muscularis propria (Fig. 4,B). These results suggest that HVEM on donor T cells is required for the full development of disease, and signaling via HVEM on T cells is necessary for the activation and expansion of T cells in our model. In addition, we measured the TNF level by real-time PCR in the intestine of the recipients and found that the recipients of Tg/HVEM−/− MLN cells had much higher level of TNF as compared with that of wt/HVEM−/− recipients (Fig. 4 C), suggesting that this phenotype was likely still TNF-dependent disease.
LTβR is required for LIGHT-mediated intestinal inflammation
LIGHT can signal LTβR expressed primarily on stromal cells (32, 33, 34). We crossed LIGHT Tg mice with LTβR−/− mice (Tg/LTβR−/−). Tg/LTβR−/− mice did not develop intestinal inflammation (Fig. 4,D), but these mice also had severe defects in LN development precluding us from transferring MLN cells into RAG-1−/− mice. Because LTβR is expressed primarily by stromal cells (radioresistant), we used the adoptive transfer approach again by transferring Tg MLN cells into LTβR−/− mice that were sublethally irradiated. LTβR−/− mice receiving Tg MLN cells did not develop wasting disease and lethality (Table I, group 7), and intestinal inflammation was not observed in these recipients (Fig. 4,E). Because LTβR−/− mice lack LN, it is possible the absence of intestinal inflammation was caused by this developmental defect instead of LTβR deficiency itself. To distinguish between these two possibilities, we transferred the Tg MLN cells into LTα−/− mice, which also have defective LN development similar to that seen in LTβR−/− mice (43). The transfer led to severe intestinal inflammation in LTα−/− mice (Table I, group 5). In contrast, the transfer of wt MLN cells into LTα−/− mice did not cause disease (Table I, group 4). Taken together, our results have definitively demonstrated the essential role of LTβR, in LIGHT-mediated intestinal inflammation.
Up-regulation of LIGHT expression in IBD patients
To test whether increased LIGHT expression is associated with CD, we investigated the mRNA expression level of LIGHT in the colonic mucosa of CD patients. We found that LIGHT mRNA was up-regulated significantly in active CD as compared with inactive CD and control patients (Fig. 5), suggesting LIGHT might play a role in the active stage of CD pathogenesis. We show a comparison between samples with active CD (n = 5), inactive CD (n = 5), active UC (n = 5), inactive UC (n = 5), and controls with ischemic colitis (n = 2) and neoplastic disease (n = 5). Interestingly, LIGHT expression was also up-regulated in UC patients as compared with controls. Furthermore, PBMC isolated from CD patients expressed higher levels of LIGHT, which associated with increased T cell proliferation upon anti-CD3 stimulation (data not shown). Taken together, these data implicate that up-regulation of LIGHT on activated T cells might play a role in IBD pathogenesis.
IBD including CD and UC are chronic inflammatory disorders of the gastrointestinal tract. The current working hypothesis is that IBD results from an inappropriate and exaggerated mucosal T cell immune response (1, 2, 4). However, the molecular mechanism of IBD pathogenesis and the exact role of Th cell subsets in the development of IBD are not fully understood. Epidemiological and family studies have provided overwhelming evidence that genetic factors have an important role in determining susceptibility to IBD (29). Since the identification of IBD1 as the first susceptibility locus for CD, several additional chromosomal loci have been identified, indicating the genetic complexity of IBD (29). Interestingly, LIGHT gene (27) is mapped to the region overlapping a susceptibility locus for IBD on human chromosome 19p13.3 (28, 29). In this study, we report that LIGHT expression is significantly increased in the intestine of active CD patients. The colonic pathology in LIGHT-induced experimental model shares several features identical to CD patients (Fig. 1); furthermore, the pathogenic mechanisms identified in our model are highly consistent with the current paradigm of CD pathogenesis. CD has been shown to associate with a Th1 phenotype (2). In our model, there is a significant increase in IFN-γ, IL-12, and TNF production but not IL-4 and IL-5, suggesting a dominant Th1 response (Fig. 2). More interestingly, the expansion and activation of Th1 cells in our model lead to intestinal pathology that resembles what is seen in CD patients, suggesting there is a causal role of Th1 cells in CD pathogenesis. Thus, we established a new experimental model for CD that closely resembles human patients in terms of both intestinal pathology and disease mechanism. More importantly, we have uncovered LIGHT as a potential common mediator of CD pathogenesis. Blocking LIGHT and its downstream signaling pathway may be an effective approach to treat IBD.
Although closely related to TNF and LT, LIGHT is a unique proinflammatory cytokine that acts as a potent costimulatory molecule for T cells. Previous studies showed that LIGHT costimulation promotes the induction of IFN-γ and GM-CSF from purified T cells (22, 24). Dysregulated LIGHT expression leads to abnormal activation and expansion of T cells in vivo (26, 30). The receptor essential for LIGHT-mediated costimulation has not been formally demonstrated, although it has been assumed the HVEM receptor is responsible for this activity. Our data clearly showed that LIGHT costimulation requires HVEM signaling both in vivo and in vitro. The finding that HVEM expression is required for LIGHT’s function suggests an autocrine effect of LIGHT, i.e., both LIGHT and HVEM are expressed by T cells, which bind to each other and mediate the costimulation effect. This hypothesis is also supported by our previous studies (26) although it is unclear whether LIGHT and HVEM need to be expressed on the same T cell or nearby T cells. The LIGHT-HVEM interaction may serve as a molecular mechanism for the expansion and activation of T cells in CD pathogenesis because LIGHT costimulation seems to preferentially promote Th1 cytokine production. A recent study also suggests that LIGHT can synergize with CD40L to stimulate the maturation of dendritic cells and that LIGHT costimulation allows dendritic cells to prime in vitro-enhanced specific CTL responses (44). Thus, LIGHT provides a unique linkage between innate and adaptive immunity.
A number of studies underline the critical importance of both TNF and TNF family members in IBD pathogenesis. The production of TNF is dramatically up-regulated in IBD patients, and TNF blockade achieves 60% effectiveness in CD patients (2). An experimental model of mucosal inflammation has been produced by creating mice that overexpress TNF due to a deletion in the ARE (TNF ARE mice) (16). The intestinal inflammation in this model resembles the pathology observed in CD patients; in particular, granuloma formation is identified. CD40L is another TNF family member implicated in the pathogenesis of IBD. Overexpression of CD40L under the control of the lck promoter leads to severe intestinal inflammation (45). CD40 and CD40L up-regulation was observed in IBD patients (46), and anti-CD40L administration ameliorated the severity of chronic murine colitis (47). Experimental colitis induced by the transfer of CD4+CD45RBhigh cells into SCID mice or reconstitution of Tgε26 mice with wt bone marrow is attenuated by LTβR-Ig treatment (31), and the effect of LTβR-Ig treatment in these two models of colitis was as potent as that of anti-TNF treatment. Although inhibition of TNF/TNFR interactions is effective in ameliorating IBD and has been widely applied to IBD patients, inhibition of interactions between LTαβ/LIGHT with LTβR is currently being investigated as a potential IBD treatment (48).
The adoptive transfer model we used in the current study is performed by transferring MLN cells into RAG-1−/− mice. The Tg MLN cells include pathogenic T cells and other cell types that might have regulatory effects on the disease development. We presented the results using 6 million cells; Tg MLN cells caused severe intestinal inflammation in the recipients, whereas the same number of wt MLN cells did not cause intestinal inflammation. Our FACS data showed that the percentage of T cells is comparable between wt and Tg donor cells, and we used young donor at the age of 2–3 mo that did not develop disease yet and showed no sign of activation marker up-regulation (CD4+CD69+ cells: 3.05% in wt vs 3.47% in Tg). Our preliminary data suggest that as low as 0.5 × 106 purified Tg T cells can also cause lethal IBD in RAG-1−/− mice (data not shown). In the CD45RBhigh transfer model, the activated T cells (i.e., CD45RBhighCD4+, 0.4–0.5 × 106) caused disease in severely immunocompromised mice such as SCID (49, 50, 51) or RAG−/− mice (52). If the regulatory T cells (CD45RBlowCD4+) were present in the transferred T cell population, the recipients would not develop disease. In this transfer model, the recipients develop moderate wasting and extensive inflammation of the large intestine, and in its severest form, these recipients display transmural involvement of the intestine wall. In fact, LTβR-Ig treatment ameliorates the severity of the disease in the CD45RBhighCD4+-reconstituted SCID mice (31). Therefore, endogenous LIGHT might be involved in causing the diseases in this transfer model.
LIGHT Tg mice spontaneously develop intestinal inflammation several months after birth (39). The severity of the inflammation in the parental Tg line is not as dramatic as that seen in the adoptive transfer recipients. Moreover, there was no typical CD-like pathology observed in parental Tg line. The differences observed between the parental Tg line and adoptive transfer model were likely due to the following reasons. First, the time course of the disease pathogenesis is different. The development of intestinal inflammation in the parental Tg line is over several months and is less severe, suggesting multiple compensatory pathways may be activated. The adoptive transfer model is an acute model (4–5 wk) that probably represents the effector phase of the disease with a dominant Th1 response present. Second, the parental Tg line has the transgene turned on in early life (before birth), whereas the adoptive transfer model provides a scenario in which an adult is acutely exposed to the pathogenic cells. Third, the parental Tg line did not show dominant Th1 responses in the gut. Th1 and Th2 cytokines were both increased dramatically (data not shown). We speculate that LIGHT may initiate a Th1 response in the MLN and/or gut of Tg mice; however, the intestinal stromal cells may develop a compensatory mechanism to counteract the dominant Th1 response by up-regulating Th2 cytokines. Therefore, the aged parental Tg line (5–8 mo) did not show a typical CD-like pathology. In addition, the adoptive transfer model offers other advantages: the disease development is more predictable, synchronized, and dramatically shortened. All of these factors further facilitate the investigation of the mechanism underlying CD and aid the development of therapeutic intervention.
Thus, the potency of LIGHT to mediate CD pathogenesis is attributed to its effects on altering both the microenvironment via LTβR and effector T cells via HVEM. Signaling through both receptors by LIGHT eventually results in the transformation of gut mucosa from a normal tertiary immune structure into a pathological lymphoid site. Dysregulated LIGHT expression may well serve as a molecular mechanism to drive the pathogenesis of CD. Therefore, the unique characteristics of LIGHT and its downstream signaling pathways suggest its candidacy as a potential target for CD. In addition, our experimental model of LIGHT- and TNF-α-mediated ileal and colonic inflammation provides a platform for the understanding of the pathogenesis and therapeutic intervention of CD in the future.
We thank Jeff Browning for his advice and TNFR-Ig.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The studies were supported in part by National Institutes of Health (NIH) Grants R01-HD37104, R01-DK58897, and P01-CA09296-01 (to Y.-X.F.) and NIH Grant R01-DK61931 (to J.R.T.). J.W. is a recipient of NIH Training Grant 5T32 HL07237.
Abbreviations used in this paper: IBD, inflammatory bowel disease; CD, Crohn’s disease; UC, ulcerative colitis; LIGHT, homologous to lymphotoxin, exhibits inducible expression, competes with herpesvirus glycoprotein D for HVEM on T cells; LT, lymphotoxin; Tg, transgenic; HVEM, herpes virus entry mediator; wt, wild type; LN, lymph node; MLN, mesenteric LN; LPL, lamina propria lymphocyte; ARE, AU-rich regulatory element.