RANK-RANKL Signaling Pathway Is Critically Involved in the Function of CD4⁺CD25⁺ Regulatory T Cells in Chronic Colitis¹

Intestinal mucosal surfaces are exposed to a large number of dietary and bacterial Ags (1, 2, 3, 4, 5). However, the gut-associated immune system fences off harmful Ags from systemic circulation and induces systemic tolerance against luminal Ags. In contrast, inflammatory bowel diseases (IBD)⁴ and animal models of T cell-mediated chronic colitis are associated with the activation of intestinal and systemic immune responses (2, 3). In this regard, CD4⁺CD25⁺ regulatory T (T_R) cells play a central role in the maintenance of immunological homeostasis (6, 7). CD4⁺CD25⁺ T_R cells have been detected mainly in lymphoid sites, including thymus, lymph nodes, and spleen. Because numerous studies have demonstrated the capacity of CD4⁺CD25⁺ T_R cells to prevent the induction of immune responses and this suppression requires direct cell-cell contact with responder T cells or APCs, it is conceivable that CD4⁺CD25⁺ T_R cells act as a central regulator within lymphoid tissues (6, 7, 8).

The GALT can be divided into effector sites, which consist of lymphocytes scattered throughout the lamina propria (LP) of the intestinal mucosa and organized lymphoid tissues (inductive sites) such as mesenteric lymph nodes (MLNs) and Peyer’s patches, which are responsible for the induction phase of immune responses (2, 9). It is thought that presentation of Ags to naive, effector, and memory T cells is concentrated at these inductive sites of organized mucosal lymphoid follicles, and thus APCs finely tune the balance between intestinal immune tolerance and inflammation.

In addition to the inductive sites, however, it remains unclear where CD4⁺CD25⁺ T_R cells suppress the development of colitis. Although it is reasonable to hypothesize that mechanisms for the induction, maintenance, and suppression of colitis would be centrally controlled in the inductive sites by CD4⁺CD25⁺ T_R cells, two-thirds of which constitutively express the lymph node-homing receptor CD62L (10), we previously demonstrated that human intestinal LP CD4⁺CD25^bright T cells obtained from normal individuals possess T_R activity in vitro and therefore questioned whether these inductive sites alone were involved in the induction and suppression of intestinal inflammation (11). We also reported that peripheral CD4⁺CD25⁺ T_R cells actually migrated to the intestine and suppressed the development of colitis in the CD4⁺CD45RB^high cell transfer model of colitis without the involvement of lymph nodes in lymph node-null LTα^−/− × RAG-2^−/− recipient mice (12). Consistent with our previous reports, it has recently been reported that CD4⁺CD25⁺ T_R cells were detected in peripheral tissues and at sites of ongoing immune responses, such as synovial fluid from rheumatoid arthritis patients (13), tumors (14), transplants (15), skin lesions in mice infected with Leishmania major (16), lungs from mice infected with Pneumocystis carinii (17), and diseased lesions in delayed-type hypersensitivity models (18), as well as in inflamed mucosa of colitic mice (8, 19).

However, it remains largely unknown which molecular mechanisms actually control the function of CD4⁺CD25⁺ T_R cells in the intestine to suppress intestinal inflammation. In the present study, we show that both CD4⁺CD25⁺ T_R cells and colitogenic CD4⁺ T cells preferentially express a TNF family member, receptor activator of NF-κB ligand (RANKL), and that blockade of the signaling pathway via RANKL and its receptor activator of NF-κB (RANK) (20) by administering neutralizing anti-RANKL mAb abrogates the CD4⁺CD25⁺ T_R cell-mediated suppression of colitis induced by adoptive transfer of CD4⁺CD45RB^high T cells into SCID mice, indicating a critical role for the RANKL/RANK signaling pathway in the function of intestinal CD4⁺CD25⁺ T_R cells in attenuating colitis.

Female BALB/c, CB-17 SCID, and C57BL/6-Ly5.2 mice were purchased from Japan Clea. C57BL/6-Ly5.1 mice and C57BL/6-Ly5.2 RAG-2-deficient (RAG-2^−/−) mice were obtained from Taconic Farms and Central Laboratories for Experimental Animals. Mice were maintained under specific pathogen-free conditions in the animal care facility of Tokyo Medical and Dental University. Mice were used at 7–12 wk of age. All experiments were approved by the regional animal study committees.

The following mAbs except anti-CCR9 mAb (R&D Systems) and reagents were purchased from BD Pharmingen: RM4-5-, PE-, or PerCP-conjugated anti-mouse CD4 (rat IgG2a); 7D4, FITC-conjugated anti-mouse CD25 (rat IgM); PC61, PE-conjugated anti-mouse CD25 (rat IgG1); H1.3F3, FITC-conjugated anti-CD69 (Ham IgG1); FJK-16s, allophycocyanin-conjugated anti-mouse Foxp3 (rat IgG2a); DATK32, PE-conjugated anti-integrin α₄β₇ (rat IgG2a); M290, PE-conjugated anti-integrin α_Eβ₇ (rat IgG2a); 242503, PE-conjugated anti-CCR9 (rat IgG2b); isotype control Abs, biotin-conjugated rat IgG2, FITC-conjugated rat IgM, PE-conjugated rat IgG2a, and PE-conjugated mouse IgG2a; PE-conjugated streptavidin; and CyChrome-conjugated streptavidin. The neutralizing anti-mouse RANKL mAb (IK22-5, rat IgG2a), and anti-mouse RANK mAb (12-31, rat IgG2a) were prepared as described previously (21).

CD4⁺ T cells were isolated from spleen cells of BALB/c mice using the anti-CD4 (L3T4) MACS system (Miltenyi Biotec) according to the manufacturer’s instructions. Enriched CD4⁺ T cells (96–97% pure, as estimated by FACSCalibur; BD Biosciences) were then labeled with PE-conjugated anti-mouse CD4 (RM4-5), FITC-conjugated anti-CD45RB (16A), FITC-conjugated anti-CD25 (7D4), and streptavidin-PE. Subpopulations of CD4⁺ cells were isolated by two-color sorting on a FACSVantage (BD Biosciences). All populations were >98.0% pure on reanalysis.

A series of in vivo experiments was conducted to investigate the role of the RANK/RANKL pathway in the expansion and function of CD4⁺CD25⁺ T_R cells in the suppression of murine chronic colitis. In experiment 1, chronic colitis was induced by adoptive transfer of CD4⁺CD45RB^high T cells into SCID mice (22). CB-17 SCID mice were injected i.p. with one or two subpopulations of sorted CD4⁺ T cells in PBS and then administered 250 μg of anti-RANKL mAb or control rat IgG in 250 μl of PBS three times per week for 6 wk as follows: 1) CD4⁺CD45RB^high alone (3 × 10⁵/mouse) plus control IgG (n = 8), 2) CD4⁺CD45RB^high alone (3 × 10⁵) plus anti-RANKL mAb (n = 8), 3) CD4⁺CD45RB^high (3 × 10⁵) plus CD4⁺CD25⁺ (1 × 10⁵) plus control IgG (n = 8), or 4) CD4⁺CD45RB^high (3 × 10⁵) plus CD4⁺CD25⁺ (1 × 10⁵) plus anti-RANKL mAb (n = 8). Mice were sacrificed at 6 wk after T cell transfer. For experiment 2, to further assess the localization of effector T cells and T_R cells in the recipients, we used Ly5.1⁺CD4⁺CD45RB^high cells and Ly5.2⁺CD4⁺CD25⁺ cells as donors and C57BL/6 RAG-2^−/− mice as recipients in the same treatment setting as experiment 1 (12). In experiment 3, to more properly assess the effect of anti-RANKL mAb on the trafficking of T_R cells to inflamed mucosa of colitic mice, we conducted another in vivo setting. First, RAG-2^−/− mice were transferred with CD4⁺CD45RB^high T cells and, 4 wk after transfer, these colitic mice were treated with 250 μg of anti-RANKL mAb or control IgG two times within 1 day. Then they were retransferred with splenic Ly5.1⁺CD4⁺ T cells from normal mice, and we evaluated the cell number of Ly5.1⁺CD4⁺Foxp3⁺ and Foxp3⁻ cells recovered from LP, spleen (SP), and MLNs at 24 h after the retransfer.

The recipient SCID mice after T cell transfer were weighed initially, then three times per week thereafter. They were observed for clinical signs of illness: hunched over appearance, piloerection of the coat, diarrhea, and blood in the stool. Mice were sacrificed and assessed for a clinical score as the sum of four parameters: hunching and wasting, 0 or 1; colon thickening, 0–3 (0, no colon thickening; 1, mild thickening; 2, moderate thickening; 3, extensive thickening); and stool consistency, 0–3 (0, normal beaded stool; 1, soft stool; 2, diarrhea; and an additional point was added if gross blood was noted) (12).

Tissue samples were fixed in PBS containing 6% neutral-buffered formalin. Paraffin-embedded sections (5 μm) were stained with H&E. Three tissue samples from the proximal, middle, and distal parts of the colon were prepared. The sections were analyzed without prior knowledge of the type of T cell reconstitution or treatment. The area most affected was graded by the number and severity of lesions. The mean degree of inflammation in the colon was calculated using a modification of a previously described scoring system (12), as follows: mucosal damage, 0; normal, 1; 3–10 intraepithelial lymphocytes (IEL)/high-power field (HPF) and focal damage, 2; >10 IEL/HPF and rare crypt abscesses, 3; and >10 IEL/HPF, multiple crypt abscesses and erosion/ulceration, submucosal damage, 0; normal or widely scattered leukocytes, 1; focal aggregates of leukocytes, 2; diffuse leukocyte infiltration with expansion of submucosa, 3; diffuse leukocyte infiltration, muscularis damage, 0; normal or widely scattered leukocytes, 1; widely scattered leukocyte aggregates between muscle layers, 2; leukocyte infiltration with focal effacement of the muscularis, and 3; extensive leukocyte infiltration with transmural effacement of the muscularis.

Colonic LP mononuclear cells were isolated using a method described previously (22). In brief, the entire length of intestine was opened longitudinally, washed with PBS, and cut into small (∼5-mm) pieces. To remove epithelium including IEL, the dissected mucosa was incubated two times with Ca²⁺Mg²⁺-free HBSS containing 1 mM DTT (Sigma-Aldrich) for 30 min and then serially incubated two times in medium containing 0.75 mM EDTA (Sigma-Aldrich) for 60 min at 37°C under gentle shaking. The supernatants from these incubations, which included the epithelium and IEL, were deserted, and the residual fragments were pooled and treated with 2 mg/ml collagenase A (Worthington Biomedical) and 0.01% DNase (Worthington Biochemical) in 5% CO₂ humidified air at 37°C for 2 h. The cells were then pelleted two times through a 40% isotonic Percoll solution and further purified by Ficoll-Hypaque (Pharmacia) density gradient centrifugation (40%/75%). Enriched CD4⁺ LP T cells were obtained by positive selection using an anti-CD4 (L3T4) MACS magnetic separation system. The resultant cells when analyzed by FACSCalibur contained >96% CD4⁺ cells.

Total cellular RNA was extracted from 7 × 10⁵ cells using a RNeasy Mini Kit (Qiagen). Five micrograms of total RNA was reverse-transcribed using the Superscript II Reverse Transcriptase (Invitrogen). RANK and RANKL levels were measured with a QuantiTect SYBER green PCR kit using Applied Biosytems I7500 real-time PCR system and 7500 system SDS software with the following primers: RANK: forward (Fwd), 5′-GGT CTG CAG CTC TTC CAT GAC-3′ and reverse (Rev) 5′-TGA GAC TGG GCA GGT AAG CC-3′; RANKL: Fwd, 5′-TTG CAC ACC TCA CCA TCA ATG-3′ and Rev, 5′-TTA GAG ATC TTG GCC CAG CCT-3′; and G3PDH: Fwd, 5′-CTA CTG GCG CTG CCA AGG CAG T-3′ and Rev, 5′-GCC ATG AGG TCC ACC ACC CTG-3′. PCR cycling conditions consisted of 95°C for 15 min, followed by 45 cycles of 95°C for 15 s, 60°C for 30 s, and 72°C for 40 s. Data are expressed as the relative amount of indicated mRNA as normalized against G3PDH.

To detect the surface expression of various molecules, isolated splenocytes or LP mononuclear cells were preincubated with a FcγR-blocking mAb (CD16/32, 2.4G2; BD Pharmingen) for 15 min, then incubated with specific FITC-, PE-, or biotin-labeled Abs for 20 min on ice. Biotinylated Abs were detected with PE- or CyChrome-streptavidin. Intracellular Foxp3 staining was performed with the allophycocyanin-anti-mouse Foxp3 staining set (eBioscience) according to the manufacturer’s instructions. Standard two- or three-color flow cytometric analyses were obtained using the FACSCalibur with CellQuest software. Background fluorescence was assessed by staining with control isotype-matched mAbs.

To measure cytokine production, 1 × 10⁵ LP CD4⁺ T cells were cultured in 200 μl of culture medium at 37°C in a humidified atmosphere containing 5% CO₂ in 96-well plates (Costar) precoated with 5 μg/ml hamster anti-mouse CD3ε mAb (145-2C11; BD Pharmingen) and 2 μg/ml hamster anti-mouse CD28 mAb (37.51; BD Pharmingen) in PBS overnight at 4°C. Culture supernatants were removed after 48 h and assayed for cytokine production. Cytokine concentrations were determined by specific ELISA per the manufacturer’s recommendation (R&D Systems).

Spleen cells from BALB/c mice were separated into unfractionated whole CD4⁺ T cells and CD4⁺CD25⁺ T cells using the anti-CD4 (L3T4) MACS magnetic separation system and/or FACSVantage as described above. Responder CD4⁺ cells (7 × 10⁴) and mitomycin C (MMC)-treated CD4⁻ cells (5 × 10⁴) as APCs, with or without CD4⁺CD25⁺ cells (1 × 10⁴), were cultured in the presence or absence of neutralizing anti-RANKL mAb or anti-RANK mAb (1, 3, or 10 μg/ml) for 72 h in round-bottom 96-well plates in RPMI 1640 medium supplemented with 10% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin, 2 mM glutamine, 1 mM sodium pyruvate, and 50 μM 2-ME. Incorporation of [³H]thymidine (1μCi/well) by proliferating cells was measured during the last 9 h of culture. In some experiments, a CFSE dilution assay was performed. To this end, splenic CD4⁺ T cells were negatively obtained from whole spleen cells of normal C57BL/6 mice using a CD4 T Cell Isolation Kit (Miltenyi Biotec). CD4⁺CD25⁻ responder T and CD4⁺CD25⁺ T_R cells were separately isolated using anti-CD25 MACS beads (Miltenyi Biotec) from the CD4⁺ T cells. CD11c⁺ dendritic cells (DC) were isolated from SP and LP of normal C57BL/6 mice and colitic C57BL/6 background RAG-2^−/− mice previously transferred with CD4⁺CD45RB^high T cells. All populations were >92% pure on reanalysis. Then, CD4⁺CD25⁻ responder T cells were labeled with 1 μM CFSE (Molecular Probes). CFSE-labeled CD4⁺ CD25⁻ responder T cells (1 × 10⁵) were cocultured with unlabeled Ly5.2⁺CD4⁺CD25⁺ T_R cells (0.33 × 10⁵) with CD11c⁺ DC (2 × 10⁴) and anti-CD3 (1 μg/ml) in the presence of anti-RANKL mAb (1 μg/ml) or control IgG (1 μg/ml) in 96-well round-bottom plates for 120 h in triplicates. After incubation, cells were collected and analyzed by FACS. Propidium iodide was added to exclude dead cells. Proliferation analysis was based on division times of responder CFSE⁺CD4⁺ T cells at the condition that can discriminate the unlabeled CD4⁺CD25⁺ T_R cells and CD11c⁺ DC cells.

For the isolation of DC from LP, MLN, and SP of colitic mice, total cells obtained by digestion with collagenase were passed through a 40-μm cell strainer. CD11c⁺ DC were further purified using an anti-CD11c MACS magnetic separation system, resulting in >80% purity. In vitro assay for the induction of gut-homing receptors was performed by the modified protocol established by others (23). In short, 2 × 10⁵ CD4⁺ T cells purified from normal BALB/c mice were cultured with 2 × 10³ purified LP, MLN, or splenic CD11c⁺ DC from colitic mice in addition to a soluble 1 μg/ml anti-CD3 mAb (BD Biosciences), human 10 ng/ml rTGF-β (PeproTech), all-trans retinoic acid (RA), and 5 ng/ml human rIL-2 (Shionogi Pharm) with or without 1 μg/ml anti-RANKL mAb. On day 4, cells were stained with PE-conjugated anti-CD4 mAb and FITC-conjugated anti-integrin α₄β₇, FITC-conjugated anti-integrin α_Eβ₇, or FITC-conjugated anti-CCR9 mAb. In some experiments, 2 × 10⁵ CD4⁺ T cells purified from normal BALB/c mice were cultured with 2 × 10³ purified LP, MLN, or SP CD11c⁺ DC from colitic mice in addition to soluble 1 μg/ml anti-CD3 mAb (BD Biosciences) with or without 1 μg/ml anti-RANKL mAb. On day 4, intracellular Foxp3 staining was performed with the allophycocyanin-anti-mouse Foxp3 staining set (eBioscience) according to the manufacturer’s instructions.

The results are expressed as the mean ± SEM. Groups of data were compared using the Mann-Whitney U test. Differences were considered to be statistically significant when p < 0.05.

To assess the role of the RANK/RANKL axis in the pathogenesis of chronic colitis in terms of stimulatory and inhibitory effects, we first assessed the expression of RANK and RANKL molecules in normal splenic CD4⁺ T cells and CD11c⁺ DC by RT-PCR. As shown in Fig. 1,A, CD4⁺ T cells preferentially expressed RANKL, but not RANK mRNA, while CD11c⁺ cells preferentially expressed RANK, but not RANKL mRNA. We further assessed the expression of RANKL on CD4⁺ T cells obtained from colitic SCID mice induced by adoptive transfer of CD4⁺CD45RB^high T cells and age-matched normal BALB/c mice at a protein level by flow cytometry. As shown in Fig. 1,B, splenic CD4⁺CD25^high and CD4⁺CD25^low T cells in normal mice slightly but substantially expressed RANKL, while CD4⁺CD25⁻ cells did not. We confirmed that almost all CD4⁺CD25^high T cells expressed Foxp3, while CD4⁺CD25^low T cells did not, indicating that CD4⁺CD25^high and CD4⁺CD25^low T cells are T_R and previously activated effector T cells, respectively (Fig. 1,B). Interestingly, although colitic SCID mice were reconstituted with a splenic CD4⁺CD45RB^high T cell population that lacks CD4⁺CD25⁺ T_R cells, inducible CD4⁺CD25^highFoxp3⁺ T cells as well as activated CD4⁺CD25^lowFoxp3⁻ T cells were found in the spleen and these cells expressed the RANKL molecule, in contrast to CD4⁺ CD25⁻ T cells (Fig. 1,B). To exclude the possibility that isolated CD4⁺CD45RB^high T cells were contaminated with CD4⁺Foxp3⁺ T_R cells, we examined the expression of Foxp3 in those cells by flow cytometry. Although almost one-third of the control CD4⁺CD45RB^low T cells expressed Foxp3, we confirmed that CD4⁺CD45RB^high T cells did not (Fig. 1 C), indicating that a small part of CD4⁺CD45RB^highFoxp3⁻ T cells had differentiated into CD4⁺Foxp3⁺ T_R cells in LP or MLNs in the periphery and migrated to the spleen in colitic mice.

Having evidence that activated CD4⁺ T cells or CD4⁺CD25⁺ T_R cells express RANKL, we next assessed whether T_R activity of splenic CD4⁺CD25⁺ T cells is modulated by blocking the RANK/RANKL pathway using neutralizing anti-RANKL and anti-RANK mAbs in in vitro coculture assay. To this end, we used two approaches, [³H]thymidine uptake and CFSE dilution assay, to assess the proliferation of responder CD4⁺ T cells. As shown in Fig. 2,A, splenic CD4⁺CD25⁺ T cells were able to suppress the proliferation of splenic CD4⁺ responder T cells when cocultured at a ratio of 1 T_R to 1 responder in the presence of MMC-treated CD4⁻ APCs, soluble anti-CD3 mAb, and control rat IgG. Similarly, neither anti-RANKL nor anti-RANK mAb at concentrations of 1–10 μg/ml affected the T_R activity of CD4⁺CD25⁺ T_R cells (Fig. 2,A). To further precisely assess the role of the RANK/RANKL pathway in the suppressive activity of CD4⁺CD25⁺ T_R cells to block the proliferation of CD4⁺CD25⁻ responder T cells at the same ratio of the following in vivo adoptive transfer experiment, we next used CD4⁺CD25⁻ T cells as responder T cells and SP or LP CD11c⁺ DC from normal or colitic mice and adopted the CFSE dilution assay to assess the proliferation of responder CD4⁺CD25⁻ T cells without the impact of the proliferation of cocultured CD4⁺CD25⁺ T_R cells. As shown in Fig. 2 B, the proliferation of CD4⁺CD25⁻ responder T cells is seen in the right part of the histogram, and they extensively proliferated in response to anti-CD3 mAb in the absence of CD4⁺CD25⁺ T_R cells with any CD11c⁺ DC regardless of SP or LP and normal or colitic DC. In contrast, the addition of CD4⁺CD25⁺ T_R cells prevented the proliferation of CD4⁺CD25⁻ responder T cells in the presence of anti-RANKL mAb to a similar extent in the presence of control IgG, indicating that the blockade of the RANK/RANKL pathway does not affect the suppressive activity of CD4⁺CD25⁺ T_R cells at least in vitro.

Although blockade of the RANK/RANKL signaling pathway did not affect the T_R function of CD4⁺CD25⁺ cells in the coculture assay (Fig. 2), in vitro assays do not always reflect function in vivo. Thus, we next administered neutralizing anti-RANKL mAb to SCID mice transferred with CD4⁺CD45RB^high cells alone or a mixture of CD4⁺CD45RB^high cells and CD4⁺CD25⁺ T cells (Fig. 3,A). As a positive control, mice transferred with CD4⁺CD45RB^high T cells alone and treated with control rat IgG developed wasting disease, which became evident 3–5 wk after transfer (Fig. 3,B), and clinical (Fig. 3,C) and histological evidence of severe chronic colitis as estimated at 6 wk after transfer (Fig. 3, D and E). As a negative control, mice transferred with both CD4⁺CD45RB^high and CD4⁺CD25⁺ T_R cell populations and administered with the control rat IgG did not develop wasting disease or colitis (Fig. 3, A–E). Although we showed that activated CD4⁺CD25^lowFoxp3⁻ effector T cells express RANKL (Fig. 1), the administration of anti-RANKL mAb to mice transferred with CD4⁺CD45RB^high cells alone did not affect the severity of colitis as compared with the control IgG-treated mice transferred with CD4⁺CD45RB^high cells alone (Fig. 3, A–E), suggesting that the RANK/RANKL pathway between RANKL-expressing activated effector CD4⁺ T cells and RANK-expressing APCs is not critically involved in the development of colitis. Surprisingly, however, the administration of anti-RANKL mAb did induce severe colitis in mice transferred with CD4⁺CD45RB^high cells and CD4⁺CD25⁺ T_R cell populations, indicating that the anti-RANKL mAb treatment abolished the CD4⁺CD25⁺ T_R cell-mediated in vivo suppression of colitis induced by adoptive transfer of CD4⁺CD45RB^high T cells.

A further quantitative evaluation of CD4⁺ T cell infiltration was made by isolating spleen cells and LP cells from the resected bowels. Significantly higher numbers of CD4⁺ T cells were recovered from the spleen (Fig. 4,A) and LP (Fig. 4,B) of the mice transferred with both CD4⁺CD45RB^high and CD4⁺CD25⁺ T_R cell populations and administered with anti-RANKL mAb as compared with control IgG, being comparable to those from mice transferred with CD4⁺CD45RB^high T cells alone and treated with anti-RANKL mAb or the control rat IgG. We also examined the cytokine production by CD4⁺ lamina propria lymphocyte from the four groups of mice. As shown in Fig. 4 C, IFN-γ and IL-17 production by LP CD4⁺ T cells was markedly suppressed by cotransfer of CD4⁺CD25⁺ T_R cells with CD4⁺CD45RB^high T cells, but this suppression was partially but significantly abrogated by the treatment with anti-RANKL mAb to the level of LP CD4⁺ T cells from the mice transferred with CD4⁺CD45RB^high T cells alone and treated with anti-RANKL mAb or control IgG.

To further assess the balance in cell numbers between effector CD4⁺ T cells and CD4⁺CD25⁺ T_R cells in recipient SCID mice, we transferred Ly5.1⁺ CD4⁺CD45RB^high cells and Ly5.2⁺ CD4⁺CD25⁺ T_R cells into C57BL/6 RAG-2^−/− mice to distinguish between effector CD4⁺ T cells and T_R cells (Fig. 5,A). We first confirmed that C57BL/6 RAG-2^−/− mice transferred with CD4⁺CD45RB^high T cells and CD4⁺CD25⁺ T_R cells did develop colitis with the marked expansion of CD4⁺ T cells when treated with anti-RANKL mAb, but not with control IgG (data not shown). As expected, the absolute number of total CD4⁺ T cells (Ly5.1⁺ plus Ly5.2⁺ cells) and Ly5.1⁺ effector CD4⁺ T cells in LP of anti-RANKL mAb-treated mice was significantly increased due to the presence of colitis as compared with that in control IgG-treated mice with no colitis (Fig. 5,B). Also, the absolute number of Ly5.2⁺ T cells and Ly5.2⁺CD4⁺Foxp3⁺ T_R cells in LP of colitic anti-RANKL mAb-treated mice was significantly higher than that in non-colitic control IgG-treated mice (Fig. 5,B), suggesting that both effector CD4⁺ T cells and T_R cells extensively proliferated in LP of anti-RANKL mAb-treated colitic mice. Interestingly, however, the ratio of Ly5.2⁺CD4⁺CD25⁺ T_R cells to total CD4⁺ T cells in the recipient mice treated with anti-RANKL mAb was markedly decreased in the intestine as compared with that in the recipient mice treated with control IgG, although there were no significant differences in the spleen, MLN, or liver between the two groups (Fig. 5 C). These results suggested that the blockade of the RANK/RANKL signaling pathway affected the expansion and/or migration of CD4⁺CD25⁺ T_R cells, resulting in dysregulated cell balance between effectors and T_R cells in the inflamed mucosa.

To assess why blockade of the RANK/RANKL pathway abolished T_R function in mice transferred with CD4⁺CD45RB^high T cells and CD4⁺CD25⁺ T_R cells, we conducted several in vitro and in vivo experiments. First, to assess the possibility that blockade of the RANK/RANKL pathway skews the expression of gut-homing receptors on T_R cells, splenic CD4⁺ T cells were cultured for 72 h in the presence or absence of anti-RANKL mAb with a stimulating mixture including anti-CD3mAb, soluble IL-2, TGF-β, and RA, which is known to induce the gut-homing receptors (integrin α₄β₇, α_Eβ₇, and CCR9) (23). As expected, the mixture strongly induced integrin α₄β₇, α_Eβ₇, and CCR9, but the addition of anti-RANKL mAb did not affect the expression of these molecules (Fig. 6 A).

To further assess the possibility that blockade of the RANK/RANKL pathway affected the T_R cell migration to the inflamed mucosa, we performed an additional in vivo adoptive transfer experiment, since it has been reported that RANK is expressed on endothelial cells (24). To this end, RAG-2^−/− mice were transferred with Ly5.2⁺CD4⁺CD45RB^high T cells and 4 wk after the transfer they were treated with control IgG or anti-RANKL mAb twice in 1 day. On the following day, they were transferred with splenic CD4⁺ T cells obtained from normal C57BL/6-Ly5.1 mice (Fig. 6,B). One day after the second transfer, the number of Ly5.1⁺CD4⁺Foxp3⁺ or Ly5.1⁺CD4⁺Foxp3⁻ T cells in SP, MLN, and LP was found not to be modified by anti-RANKL-treatment at all (Fig. 6 B).

To finally assess the possibility that colitic LP DC cells modulate the expansion of T_R cells in the inflamed mucosa in a RANK/RANKL-dependent manner, we evaluated the expression of RANK on CD11c⁺ DC cells obtained from the spleen, MLN, and LP of colitic and normal mice. As shown in Fig. 7,A, the expression of RANK on colitic LP DC was significantly increased as compared with that on normal LP DC. In contrast, the expression of RANK in SP and MLN was similar in normal and colitic mice. Given the up-regulated expression of RANK on colitic LP DC cells, we next assessed the possibility that the RANK/RANKL pathway is involved in the expansion of T_R cells. To this end, splenic CD4⁺ T cells obtained from normal mice were cultured with colitic LP, MLN, or SP CD11c⁺ DC in the presence of anti-CD3 mAb with or without anti-RANKL mAb for 72 h. The ratio of CD4⁺Foxp3⁺ T_R cells per the total of CD4⁺ T cells (Fig. 7,B) and the number of CD4⁺Foxp3⁺ T_R cells (Fig. 7 C) recovered from culture with colitic LP, but not MLN or SP, CD11c⁺ DC in the presence of anti-RANKL mAb was significantly decreased as compared with that in the presence of control IgG, suggesting that the RANK/RANKL pathway is critically involved in the expansion of LP T_R cells through the direct interaction with colitic RANK-expressing LP DC and T_R cells.

In the present study, we demonstrated that 1) CD4⁺CD25⁺ T cells including CD4⁺CD25^high T_R cells and activated CD4⁺CD25^low effector cells rather than CD4⁺CD25⁻ T cells preferentially express the RANKL molecule and 2) blockade of the RANK/RANKL signaling pathway suppresses the expansion of CD4⁺CD25⁺ T_R cells and subsequently abolishes the T_R cell-mediated suppression of colitis due to dysregulation of the cell balance between effector CD4⁺ T cells and T_R cells in the inflamed intestine. Interestingly, although activated effector CD4⁺ T cells and inducible CD4⁺CD25⁺ T_R cells also express RANKL molecules in SCID mice transferred with CD4⁺CD45RB^high T cells alone, the administration of this mAb did not affect the course of colitis. Collectively, these findings indicate that the RANK-RANKL signaling pathway is critically involved in intestinal mucosal tolerance by controlling the expansion and function of CD4⁺CD25⁺ T_R cells in the inflamed mucosa.

Although many previous reports have established the role of the RANK/RANKL signaling pathway in osteoclastogenesis and bone loss in various chronic T cell-mediated inflammatory diseases including IBDs (20, 25, 26, 27, 28, 29), the role of this pathway in the local inflammation in various models remains unknown. For example, Kong et al. (26) initially reported that the blockade of this pathway by using soluble recombinant osteoprotegerin (OPG) protein at the onset of disease prevented bone and cartilage destruction, but interestingly not inflammation in a T cell-dependent model of rat adjuvant arthritis. In contrast, Ashcroft et al. (30) demonstrated that the administration of RANK-Fc protein not only reverses the bone loss in IL-2^−/− mice, which is a spontaneous model of osteoporosis and colitis, but also reduces the development of colitis by blocking the interaction between RANK-expressing DC and RANKL-expressing activated CD4⁺ T cells in the inflamed mucosa of the colon. In our colitis model induced by adoptive transfer of CD4⁺CD45RB^high T cells into SCID mice, however, administration of neutralizing anti-RANKL mAb did not prevent the development of colitis, although inducible CD4⁺CD25^high T_R cells and previously activated CD4⁺CD25^low T cells expressed RANKL. Consistent with this finding, Byrne et al. (31) previously demonstrated that administration of human osteoprotegerin-Fc increased bone density in this model, but had no effects on the intestinal inflammation). Several explanations have been advanced for the discrepancy, including differences in the species, the type of animal model, the type of blocking agents, and dosing regimens used. Furthermore, it has recently been demonstrated that in vivo administration of neutralizing anti-cytokine mAbs, such as anti-IL-2 mAb, enhances the corresponding cytokine activity due to the formation of cytokine/anti-cytokine mAb complexes, which are more stable and stimulatory (32). Although we previously demonstrated that our anti-RANKL mAb used in vivo successfully worked as a blocking mAb in a model of collagen-induced arthritis (21), further studies will be needed to address this issue.

Although we could not detect a suppressive effect of neutralizing anti-RANKL mAb on the development of colitis in SCID mice transferred with CD4⁺CD45RB^high T cells alone, we found that this treatment induced colitis in mice transferred with CD4⁺CD45RB^high T cells and CD4⁺CD25⁺ T_R cells at a ratio of 3:1, while mice transferred with CD4⁺CD45RB^high T cells and CD4⁺CD25⁺ T_R cells at the same ratio and given control IgG did not develop colitis. This strongly suggested that the target cells for anti-RANKL mAb are CD4⁺CD25⁺ T_R cells rather than CD4⁺CD45RB^high T cells or the differentiated effector CD4⁺ T cells. Consistent with this notion, we found that RANKL was expressed on CD4⁺CD25^high T_R cells, but not on CD4⁺CD25⁻ cells (Fig. 1). In an in vitro coculture assay to further evaluate the role of RANKL on CD4⁺CD25⁺ T_R cells in modulating the T_R activity of CD4⁺CD25^high cells in vitro, however, the addition of anti-RANKL mAb or anti-RANK mAb produced no detectable reduction of T_R activity, suggesting that the direct interaction between RANKL-expressing CD4⁺CD25⁺ T_R cells or activated CD4⁺ T cells and RANK-expressing APCs including DC is not essentially important for abolishing T_R activity at least in vitro. Since in vitro assays do not always reflect the T_R function in vivo, we next performed another adoptive transfer experiment using Ly5.1⁺CD4⁺CD45RB^high T cells and Ly5.2⁺CD4⁺CD25⁺ T_R cells to evaluate the possibility that blockade of the RANK/RANKL signaling pathway affects the recruitment and expansion of specific populations in our model. In this experiment, we found that the ratio of the Ly5.2-derived CD4⁺CD25⁺ T_R cell population per total CD4⁺ T cells in anti-RANKL mAb-treated mice was significantly decreased in the LP, but not in the MLN, spleen, or liver, as compared with the ratio in control-IgG-treated mice. This finding suggests three points. First, it is possible that the migration of CD4⁺CD25⁺ T_R cells to the inflamed mucosa is regulated by the interaction between RANKL-expressing T_R cells and possibly RANK-expressing endothelial cells. Consistent with this hypothesis, it has been reported that RANK is expressed on murine and human endothelial cells (24). However, this is unlikely because our short-term in vivo adoptive transfer experiment (Fig. 7) demonstrated that treatment of anti-RANKL mAb did not affect the recovered cell number of CD4⁺CD25⁺ T_R cells in the inflamed mucosa. Second, it is possible that the in vivo expansion of RANKL-expressing CD4⁺CD25⁺ T_R cells or colitogenic CD4⁺ effector/memory T cells is modulated by RANK-expressing DC in the inflamed mucosa. Consistent with this hypothesis, we found that 1) the expression of RANK on colitic LP DC is significantly increased as compared with that on normal LP DC and 2) the ratio of CD4⁺Foxp3⁺ T_R cells to total CD4⁺ T cells after stimulation with colitic LP, but not MLN or splenic, CD11c⁺ DC was significantly suppressed by the addition of anti-RANKL mAb. Thus, it is possible that the interaction of RANKL-expressing CD4⁺CD25⁺ T_R cells and RANK-expressing activated DC in the inflamed mucosa plays an important role in the maintenance of T_R cells. Third, not only the first suppression of T cell priming in draining lymph nodes by CD4⁺CD25⁺ T_R cells, but also the second line of suppression in the inflamed mucosa by these cells is critically involved in the intestinal homeostasis to suppress the development of colitis. Consistent with our finding, Green et al. (33) previously reported that the blockade of the RANK/RANKL pathway resulted in a decreased frequency of CD4⁺CD25⁺ T_R cells in the draining lymph nodes and pancreas in the NOD mouse (33). However, it remains unknown why the ratio of CD4⁺CD25⁺ T_R cells in MLNs was unchanged in our adoptive transfer model. The exact mechanism of the function of CD4⁺CD25⁺ T_R cells in the inflamed mucosa warrants further investigation.

Finally, the suppressive site of colitis should be discussed. Denning et al. (34) previously demonstrated that integrin β₇-deficient (β7^−/−) CD4⁺CD25⁺ T_R cells that preferentially migrate to MLNs, but are impaired in their ability to migrate to the intestine because of the lack of the gut-homing integrin α₄β₇ and α_Eβ₇ molecules, are capable of preventing intestinal inflammation, suggesting that T_R accumulation in the intestine is dispensable for the protection of this colitis model. In their protection protocol, indeed, it is possible that β₇^+/+CD4⁺CD25⁺ T_R cells are not needed to suppress the development of colitis, because β₇^−/− CD4⁺CD25⁺ T_R cells directly migrate to MLNs and can inhibit naive CD4⁺CD45RB^high T cell activation and proliferation within Ag-draining MLNs, resulting in suppression of the development of the gut-seeking activated effector CD4⁺ T cells instructed to express the gut-homing receptors such as integrins α₄β₇ and α_Eβ₇. However, it remains unknown whether mucosal CD4⁺CD25⁺ T_R cells are necessary for the suppression of mucosal pathogenic effector CD4⁺ T cells ex vivo, especially in an ongoing colitis system in which it can be assessed whether LP CD4⁺CD25⁺ T_R cells as effector T_R cells can suppress the surrounding LP effector CD4⁺ T cells ex vivo.

In this regard, we have previously demonstrated that human CD4⁺CD25^bright and mouse CD4⁺CD25⁺ T cells reside in the intestinal LP, express CTLA-4, GITR, and Foxp3 and possess T_R activity in vitro (11, 12). We also found that the clinical score in SCID mice transferred with CD4⁺CD45RB^high T cells and intestinal LP CD4⁺CD25⁺ T cells at a ratio of 3:1 was significantly decreased as compared with that in SCID mice transferred with CD4⁺CD45RB^high T cells alone (12), indicating that the murine intestinal LP CD4⁺CD25⁺ T cells maintain intestinal homeostasis to suppress the development of colitis. Having evidence that the murine intestinal LP CD4⁺CD25⁺ T cells suppressed the development of colitis induced by the adoptive transfer of CD4⁺CD45RB^high T cells, we further asked whether MLNs are fully essential for the suppression of colitis by splenic CD4⁺CD25⁺ T cells. As a second approach to this issue, we also found that the cotransfer of splenic CD4⁺CD25⁺ T_R cells prevented the development of colitis in the lymph node-null LTα^−/− × RAG-2^−/− mice transferred with CD4⁺CD45RB^high T cells, indicating that splenic CD4⁺CD25⁺ T cells can suppress the development of colitis in the absence of MLNs (12). Moreover, we demonstrated that CD4⁺CD25⁺ T_R cells actually migrated and resided in the colon in LTα^−/− × RAG-2^−/− mice cotransferred with Ly5.2-derived CD4⁺CD45RB^high T cells and Ly5.1-derived splenic CD4⁺CD25⁺ T cells, suggesting that the LP might be a regulatory site between colitogenic effector/memory cells and T_R cells to suppress intestinal inflammation, probably as a second line of suppression (yy). Along with the present findings that the RANK/RANKL interaction is critically involved in the function of CD4⁺CD25⁺ T_R cells in the intestine, our research suggests that therapeutic approaches enhancing the migration of CD4⁺CD25⁺ T_R cells, such as the specific induction of RANKL on CD4⁺CD25⁺ T_R cells, may be feasible in the treatment of IBD.

The authors have no financial conflict of interest.