Antigen-specific suppression of established arthritis in mice by dendritic cells deficient in NF-κB

Authors


Abstract

Objective

NF-κB inhibitors applied to animal models of rheumatoid arthritis (RA) demonstrate the important role of NF-κB in the production of mediators of inflammation in the joint and their antiinflammatory effects. Because NF-κB is involved in the differentiation, activation, and survival of almost all cells, its prolonged inhibition might have unwanted adverse effects. Therefore, we sought to apply NF-κB inhibitors more specifically, targeting dendritic cell (DC) differentiation, in order to influence the outcome of the autoimmune response, rather than to produce a broad antiinflammatory effect. We tested whether DCs treated with the NF-κB inhibitor BAY 11-7082 and exposed to arthritogenic antigen would suppress established arthritis in C57BL/6 mice.

Methods

Antigen-induced arthritis was generated in C57BL/6 mice by injection of methylated bovine serum albumin (mBSA). After mBSA challenge, mouse knee joints were injected with antigen-exposed BAY 11-7082–treated DCs or with soluble tumor necrosis factor receptor (sTNFR). Intraarticular injection of interleukin-1 (IL-1) was used to induce disease flare.

Results

Inflammation and erosion were suppressed in mice that received mBSA-exposed BAY 11-7082–treated DCs, but not in those that received keyhole limpet hemocyanin–exposed BAY 11-7082–treated DCs. Clinical improvement was dependent on IL-10 and was associated with antigen-specific suppression of the delayed-type hypersensitivity (DTH) reaction and switching of anti-mBSA antibody isotype from IgG2b to IgG1 and IgA. Suppression of the DTH reaction or arthritic disease was not impaired by concomitant administration of sTNFR. Suppression could be reversed with intraarticular administration of IL-1β and could be restored by a second injection of mBSA-exposed BAY 11-7082–treated DCs.

Conclusion

BAY 11-7082–treated DCs induce antigen-specific immune suppression in this model of inflammatory arthritis, even after full clinical expression of the disease. Such DCs have potential as antigen-specific therapy for autoimmune inflammatory arthritis, including RA.

Rheumatoid arthritis (RA) is an autoimmune disease characterized by chronic inflammation of synovial tissue and erosion and damage of bone and cartilage (1). As a result of the autoimmune process, the pathology of RA is driven by innate immune inflammatory mediators produced by synoviocytes and macrophages and involving the transcription factor NF-κB. The NF-κB transcription factor pathway plays a key role in the differentiation, activation, and survival of almost all cells in the body. In RA, active NF-κB is localized in the nucleus of hemopoietic and parenchymal cells (2–4). Moreover, previous studies of NF-κB inhibitors in animal models of RA demonstrated the important role played by NF-κB in the production of mediators of inflammation in the joint and the antiinflammatory effect of such inhibitors (5–10).

The clinical manifestations of RA respond well to a variety of disease-modifying drugs. Several of these drugs, such as gold, corticosteroids, sulfasalazine, cyclosporine, and leflunomide, have been shown to inhibit NF-κB, among other effects (11–13). Because of its involvement in the differentiation, activation, and survival of almost all cells, prolonged inhibition of NF-κB might have unwanted adverse effects. Therefore, we sought to apply NF-κB inhibitors more specifically, targeting dendritic cell (DC) differentiation, in order to influence the outcome of the autoimmune response, rather than to produce a broad antiinflammatory effect.

DCs play a central role in initiation of autoimmunity in RA through their capacity to present self antigen to T cells in the thymus and in peripheral sites. For stimulation of an immune response after antigen encounter, DCs differentiate and migrate to draining lymph nodes in response to proinflammatory signals (14). There, DCs normally prime immune responses to foreign antigen but may prime self-antigen–specific responses in autoimmunity (15, 16). DCs differentiated from bone marrow or peripheral blood precursors and exposed to antigen in vitro represent an excellent tool for inducing antigen-specific immunity in recipient animals. Furthermore, trials of ex vivo–differentiated DCs exposed to cancer antigens have shown promise in the stimulation of tumor-specific immunity. Using this tool, we previously showed that application of NF-κB inhibitors to DCs as they differentiate in vitro affects their differentiation. In the presence of a variety of NF-κB inhibitors, DCs express low levels of CD40 and class II major histocompatibility complex molecules relative to the levels expressed by mature DCs. Moreover, when exposed to antigen and injected into mice, these DCs prevent priming of immunity and suppress previously primed immune responses through the induction of antigen-specific CD4+ regulatory T cells (17). These T cells produce interleukin-10 (IL-10), and immune suppression is IL-10 dependent. Inactivation of the NF-κB subunit RelB or lack of expression of cell surface CD40 was shown to be sufficient for DCs to induce tolerance (17).

In the present study, we determined whether antigen-exposed DCs modified with the drug BAY 11-7082 could be used to suppress active inflammatory arthritis in an antigen-specific manner. BAY 11-7082 blocks NF-κB activation through inhibition of IκB phosphorylation (7), and when added to cultures of murine bone marrow or human blood monocyte DC precursors, it modifies their differentiation (17, 18). For these experiments, we used the antigen-induced arthritis (AIA) model in C57BL/6 mice, in which arthritis develops as a result of immunity to the methylated bovine serum albumin (mBSA) antigen. Our data demonstrate that subcutaneous administration of mBSA-exposed BAY 11-7082–treated DCs suppresses the clinical and pathologic features of AIA, even after full clinical expression of the disease has occurred.

MATERIALS AND METHODS

Induction of AIA, DTH reaction, and arthritis flare and clinical scoring.

Monarticular AIA was induced in male C57BL/6 mice obtained from ARC Animal Supplies (Perth, Western Australia, Australia), as previously described (19). Briefly, on day −21 and day −14, mice were primed subcutaneously in each axilla with 100 μg of mBSA (1 mg/ml; Sigma, St Louis, MO) mixed 1:10 with Freund's complete adjuvant (Sigma), and pertussis toxin (400 ng; CSL, Parkville, Victoria, Australia) was injected intraperitoneally. On day 0, with the mouse under anesthesia, 30 μg of mBSA in 10 μl of saline was injected intraarticularly into one knee, and 10 μl of saline was injected into the contralateral knee. The day before the mice were euthanized, 5 μg of mBSA was injected intradermally into the ear to demonstrate a DTH reaction. The following day, DTH responses were read.

Clinical scores were assigned based on caliper measurements taken of lightly anesthetized mice every second day after arthritis induction. Scores were normalized to the initial size of the joint and to the maximum swelling achieved in the experiment and were expressed in quintiles. After the mice were euthanized, the joints were also graded for arthritis severity using a semiquantitative scale of 1–5, where 1 = no difference between the mBSA-injected knee and the saline-injected control knee, 2 = slight discoloration of the mBSA-injected knee, 3 = discoloration of the mBSA-injected knee and mild lateral swelling, 4 = discoloration of the mBSA-injected knee and moderate lateral swelling, and 5 = severe discoloration of the mBSA-injected knee to the point where the ligament is no longer visible and severe lateral swelling. A positive DTH response was recorded in all mice with AIA.

Decalcified joints were fixed in formalin, blocked in paraffin, sectioned, and stained with hematoxylin and eosin or with toluidine blue. In each knee joint, inflammation, erosion, pannus formation, and cartilage integrity were scored on a semiquantitative scale of 0–5, where 0 = normal and 5 = severe, as described in detail elsewhere (20).

To produce an arthritis flare, 100 μg of IL-1β was administered intraarticularly into the mBSA-injected joint on day 12 (21).

For in vivo blocking studies, recipient mice were injected intraperitoneally with 1 mg of anti–IL-10 receptor (anti–IL-10R) monoclonal antibody (mAb) (1B1.3a; BD PharMingen, San Jose, CA) or isotype control mAb 4 hours before transfer of DCs, or injected with 2 μg of soluble tumor necrosis factor receptor (sTNFR) (Etanercept; Wyeth Pharmaceuticals, Baulkham Hills, New South Wales, Australia). All experimental groups contained between 5 and 15 mice.

Preparation and administration of bone marrow–derived DCs.

Bone marrow cells were collected from murine long bones and suspended in RPMI 1640 (Gibco, Mulgrave, Victoria, Australia), passed through nylon mesh, and mononuclear cells were separated by Ficoll gradient centrifugation. Macrophages, class II–positive cells, and lymphocytes were immunodepleted using appropriate mAb followed by magnetic beads (magnetic-activated cell sorting; Miltenyi Biotec, Sunnyvale, CA). Bone marrow cells were incubated for 6–8 days in XCell620 medium (CSL) or RPMI 1640 with 10% fetal calf serum supplemented with 10 ng/ml each granulocyte–macrophage colony-stimulating factor and IL-4 (PeproTech, Rocky Hill, NJ), with fresh medium applied on alternate days. DC preparations routinely contained 80–90% CD11c+ cells.

BAY 11-7082–treated DCs were cultured continuously in the presence of ∼5 μM BAY 11-7082 (Biomol, Plymouth Meeting, PA), exposed to 100 μM mBSA or keyhole limpet hemocyanin (KLH; Sigma) for 12 hours, washed, and then suspended in normal saline. BAY 11-7082–treated DCs (5 × 105 cells) were administered subcutaneously into the base of the tail at various times after the induction of arthritis.

Determination of mBSA-specific antibodies and isotyping.

For mBSA-specific antibody determination by enzyme-linked immunosorbent assay, serum was prepared from blood samples taken from the lateral tail vein. Then, 100 μl of mBSA, at 10 μg/ml in 50 mM carbonate buffer (pH 9.6), was coated onto the wells of 96-well microtiter plates (Greiner, Kremsmünster, Austria). After washing with 0.5% Tween 20–phosphate buffered saline (PBS) and blocking with 200 μl of 3% BSA fraction V, 100 μl of serum in 5-fold dilutions was added to triplicate wells. After washing, each well was incubated with 100 μl of either biotinylated rabbit anti-mouse IgM, IgA, IgG1, IgG2b, or IgG2c secondary antibody (Sigma), followed by washing and incubation with streptavidin–horseradish peroxidase. After incubation with 0.1% ABTS in 0.03% hydrogen peroxide and 150 mM citrate buffer, pH 4.5, the presence of antigen-specific antibodies was determined according to the net absorbance readings at 405 nm and 492 nm. IgE assays were performed by using passive cutaneous anaphylaxis. Doubling dilutions of sera (0.5 ml) were injected intradermally into the dorsal skin of male Sprague-Dawley rats. After 24 hours, the rats were injected intravenously with 2 mg of ovalbumin or papain in 0.5 ml of PBS containing 0.05% Evans blue dye.

Cellular and cytokine responses in draining lymph nodes from mice with AIA and from treated mice.

CD11c+ DCs from the spleens of naive syngeneic mice were purified by positive immunoselection using CD11c microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). T cells derived from draining lymph nodes of each group of mice were purified by passing them through nylon-wool columns. Assessment of mBSA-specific T cell responses in cells from draining lymph nodes obtained from mice with AIA included 2 × 104 CD11c+ DCs from naive mice and 1 × 105 draining lymph node T cells and were incubated in the presence and absence of 1 μg/ml of mBSA for 72 hours in round-bottomed wells. Cells were pulsed for the final 18 hours of culture with 3H-thymidine (ICN, Costa Mesa, CA) and counted using a Packard TopCount NXT instrument (Packard, Meriden, CT).

Levels of interferon-γ (IFNγ), IL-2, IL-10, tumor necrosis factor (TNF), and IL-6 were measured in supernatants using the BD Cytometric Bead Array Mouse Inflammation kit (BD PharMingen) according to the manufacturer's protocol. In some experiments, either anti–IL-10R (BD PharMingen), anti-TGFβ mAb (R&D Systems, Minneapolis, MN), or isotype control mAb (R&D Systems) was added to some wells. Draining lymph nodes were stained with allophycocyanin (APC)–labeled CD4 (BioLegend, San Diego, CA), fluorescein isothiocyanate (FITC)–labeled CD25 (BioLegend), and phycoerythrin-labeled FoxP3 (eBioscience, San Diego, CA), after permeabilization using BD Perm/Wash, or with FITC-labeled CD3 (BioLegend) and APC-labeled NK1.1 (BioLegend). After setting gates based on isotype controls, FoxP3+ and NK1.1+ cell populations were determined.

Statistical analysis.

Differences were analyzed using either Student's unpaired t-test or the Mann-Whitney U test as appropriate.

RESULTS

Antigen-specific suppression of AIA by BAY 11-7082–treated mBSA-exposed DCs.

Antigen-exposed DCs in which the RelB function is inhibited by BAY 11-7082, an NF-κB inhibitor, prevent the priming of immunity and suppress previously primed immune responses (17). To determine whether BAY 11-7082–treated DCs similarly suppress inflammatory arthritis in an antigen-specific manner, monarticular AIA was induced in mBSA-primed C57BL/6 mice by injection of mBSA into 1 knee joint, and DCs were administered 2 days later. Saline was injected into the contralateral control knee joint. We have previously shown that tolerance induced by DCs is dose-dependent and route-independent (17); therefore, BAY 11-7082–treated DCs were administered subcutaneously in a single dose of 5 × 105 cells in all experiments except where indicated otherwise. At 10 days after induction, mBSA-exposed BAY 11-7082–treated DCs suppressed the clinical severity of AIA, whereas neither DCs generated without BAY 11-7082 nor DCs exposed to the irrelevant antigen KLH and treated with BAY 11-7082 had any effect (Figure 1A).

Figure 1.

Antigen-specific suppression of antigen-induced arthritis (AIA) in C57BL/6 mice by BAY 11-7082–treated dendritic cells (DCs). A, Bone marrow–derived DCs were generated in the presence or absence of BAY 11-7082, then exposed to either keyhole limpet hemocyanin (KLH) or methylated bovine serum albumin (mBSA) for 18 hours. Mice were primed with mBSA, Freund's complete adjuvant, and pertussis toxin, and 8 days later, mBSA was injected into one knee joint and saline into the other. Two days later, 5 × 105 bone marrow–derived DCs were injected subcutaneously into the base of the tail. Severity scores were determined 8 days following DC injection, after mice were euthanized (maximum score 5; see Materials and Methods for details). Values are the mean and SEM of 5 mice per group, each tested individually. Results are representative of 2 separate experiments. ∗∗ = P < 0.01 versus each of the other groups. BD, AIA was induced, and 5 × 105 mBSA-exposed BAY 11-7082–treated DCs were injected subcutaneously 2, 4, or 6 days after knee joints were injected. Clinical scores (B) and delayed-type hypersensitivity reactions (D) were monitored in live mice until day 10. Severity scores (C) were assessed on day 10, after mice were euthanized. Values are the mean and SEM of 5–15 mice per group, each tested individually. Results are representative of 3 separate experiments. ∗∗∗ = P < 0.001, comparing treated versus untreated groups, as indicated.

To determine whether the duration and severity of arthritis affected the ability of DCs to suppress inflammation, AIA was induced in groups of mBSA-primed mice by administration of intraarticular mBSA. Mice were left untreated or were injected subcutaneously with 5 × 105 mBSA-exposed BAY 11-7082–treated DCs 2, 4, or 6 days later. Clinical assessments of arthritis were performed until the animals were euthanized (10 days after the intraarticular injection) and were expressed as the clinical score (Figure 1B) or the raw caliper score (data not shown). Arthritis severity was determined on day 6 (after the animals were euthanized), and changes were graded on a scale of 1–5 (Figure 1C). Another measure of antigen-specific immunity, the mBSA-specific DTH response, was tracked simultaneously (Figure 1D).

Regardless of the timing of DC administration, both the clinical score and the DTH response declined over the subsequent 4–6 days, but did not decline in untreated mice with AIA. The raw caliper, clinical, and severity scores were correlated. Suppression of clinical scores was mirrored by a significant reduction in histologic evidence of inflammation, erosion, pannus formation, and cartilage disruption after DC administration (Figure 2). These data indicate that antigen-exposed DCs, in which NF-κB activity is inhibited, suppress clinical and histologic inflammation and destructive changes in an antigen-specific manner, even when delivered after full clinical disease expression has occurred. This antigen-specific suppression of immunity is apparent systemically, as observed by the decrease in the DTH reaction.

Figure 2.

Suppression of the pathologic changes of antigen-induced arthritis (AIA) in C57BL/6 mice by BAY 11-7082–treated dendritic cells (DCs). AIA was induced as described in Figure 1, and 5 × 105 methylated bovine serum albumin (mBSA)–exposed BAY 11-7082–treated DCs were injected subcutaneously on day 6 after knee joint injection. After 10 days, sections from saline-injected and mBSA-injected knee joints were stained with hematoxylin and eosin and with toluidine blue. AC, Representative sections from a saline-injected control joint, an untreated arthritic joint, and a DC-treated arthritic joint (original magnification × 100). D, Histologic scores for synovial inflammation, erosions, pannus formation, and cartilage damage. Values are the mean and SEM of 5 mice per group, each tested individually. Results are representative of 3 separate experiments. ∗∗∗ = P < 0.0005; ∗∗ = P < 0.005; ∗ = P < 0.05 versus treated groups.

Arthritis suppression induced by BAY 11-7082–treated DCs compared with TNF blockade, and effects of IL-1β.

To assess the efficacy of BAY 11-7082–treated DCs relative to the efficacy of TNF blockade, an important biologic disease modifier in RA, mice were treated 6 days after induction of AIA (when disease was fully expressed) with a subcutaneous injection of mBSA-exposed BAY 11-7082–treated DCs or an intravenous infusion of sTNFR (etanercept), or both. Treatment with sTNFR, DCs, or the combination of DCs and sTNFR each suppressed arthritis as compared with controls (P < 0.001) by day 12 (Figure 3). Thus, both BAY 11-7082–treated DCs and TNF blockade are efficacious in the suppression of acute inflammatory AIA, and disease suppression is not impaired by concomitant administration of sTNFR.

Figure 3.

Equivalence of the suppression of antigen-induced arthritis (AIA) in C57BL/6 mice by BAY 11-7082–treated dendritic cells (DCs) and by tumor necrosis factor (TNF) blockade and reversal by intraarticular injection of interleukin-1β (IL-1β). AIA was induced as described in Figure 1, and mice were left untreated or 5 × 105 methylated bovine serum albumin (mBSA)–exposed BAY 11-7082–treated DCs were injected subcutaneously, 2 μg of soluble TNF receptor (sTNFR) was injected intravenously, or both were administered on day 6 after knee joint injection. In each experimental group, 100 μg of IL-1β was injected into the arthritic joint on day 12. Mice were then treated a second time with either a subcutaneous injection of 5 × 105 mBSA-exposed BAY 11-7082–treated DCs or control mBSA-unexposed DC on day 15 or were left untreated. Clinical scores were monitored daily until day 20, when mice were euthanized. Values are the mean ± SEM of 5 mice per group, each tested individually. Results are representative of 3 separate experiments. ∗∗∗ = P < 0.0005; ∗∗ = P < 0.001, comparing treated versus control groups on days 12 and 20, respectively, as indicated. ns = not significant.

Administration of IL-1β to the joints of mice with resolving AIA has been shown to induce disease flare. We therefore tested whether intraarticular administration of IL-1β would overcome the suppression induced by BAY 11-7082–treated DCs. Clinical arthritis flared regardless of whether mice had previously received DCs or sTNFR, within 2 days of IL-1β injection. However, disease could be rapidly resuppressed by a second administration of mBSA-exposed BAY 11-7082–treated DCs (or sTNFR [data not shown]), but not control DCs (Figure 3). After readministration of mBSA DCs, arthritis was suppressed as compared with untreated mice as well as with mice given control DCs (P < 0.01). These data indicate that the antigen-specific suppression of arthritis by BAY 11-7082–treated DCs occurs rapidly and consistently. The disease is still prone to flares in response to IL-1β–induced proinflammatory stress, and the flares require further specific intervention for control.

IL-10 dependence of arthritis suppression by BAY 11-7082–treated DCs and association with immune deviation.

We have previously shown that antigen-exposed BAY 11-7082–treated DCs induce a population of IL-10–producing Treg cells after administration to previously primed or naive mice and that suppression of priming was IL-10–dependent (17). To determine whether a similar mechanism responsible for the suppression of arthritis was also IL-10–dependent, AIA was induced in mice, and after 6 days, groups of mice were treated with either saline, anti–IL-10R alone, or mBSA-exposed BAY 11-7082–treated DCs with and without anti–IL-10R or isotype control mAb delivered in vivo. Severity scores were determined 8 days later. While DCs alone or DCs in the presence of isotype control mAb suppressed arthritis severity, IL-10 blockade reversed the DC-mediated suppression as compared with that in anti–IL-10R–treated mice and in untreated mice (P < 0.05) (Figure 4).

Figure 4.

Interleukin-10 (IL-10) dependence of the suppression of antigen-induced arthritis (AIA) by BAY 11-7082–treated dendritic cells (DCs). AIA was induced as described in Figure 1, and 5 × 105 methylated bovine serum albumin (mBSA)–exposed BAY 11-7082–treated DCs were injected subcutaneously on day 6 after knee joint injection, either alone or 4 hours after intraperitoneal administration of 1 mg of anti–IL-10 receptor (anti–IL-10R) or isotype control monoclonal antibody (mAb). Control groups were either left untreated or were treated with 1 mg of anti–IL-10R mAb. Severity scores were determined on day 10. Values are the mean and SEM of 5 mice per group, each tested individually. Results are representative of 2 separate experiments. ∗ = P < 0.05 versus all other groups.

Sera from treated and untreated mice were analyzed for mBSA-specific antibodies. In untreated mice, antibodies were predominantly Th1-type, with high levels of IgG2b and IgG2c, and with low levels of IgG1. In contrast, in treated mice, IgG2b and 2c were suppressed, and IgG1 and IgA levels were increased (P < 0.0001) (Figure 5A). IgE levels were low in both groups, as was expected (Figure 5B). These data indicate that BAY 11-7082–treated DC suppression of arthritis is IL-10–dependent and is associated with deviation of the mBSA-specific antibody response.

Figure 5.

Association of the suppression of clinical antigen-induced arthritis (AIA) by BAY 11-7082–treated dendritic cells (DCs) with an immunoregulatory response. AIA was induced as described in Figure 1, and 5 × 105 methylated bovine serum albumin (mBSA)–exposed BAY 11-7082–treated DCs were injected subcutaneously on day 6 after knee joint injection. A, Serum anti-mBSA antibody (AB) levels, as measured by enzyme-linked immunosorbent assay using isotype-specific secondary monoclonal antibodies, on day 10. B, Total IgE levels, as measured on day 10. Values are the mean and SEM. ∗∗∗ = P < 0.0001 versus the level with the equivalent antibody in untreated mice with AIA. NS = not significant.

To test the nature of the immune response generated in response to DC treatment, AIA was induced in mice, and after 6 days, groups of mice were treated with saline or with mBSA-exposed BAY 11-7082–treated DCs, then euthanized 8 days later. Inguinal lymph node cells draining the site of DC injection were enriched for T cells with the use of nylon-wool columns and were restimulated with splenic CD11c+ DCs and mBSA. In mice that had previously received DCs, mBSA-specific proliferation and IL-2 production were reduced and the production of IL-10 and IFNγ was increased, as compared with untreated AIA control mice (data not shown). Little IL-4 or IL-5 was produced by cells from either group of mice. These data are consistent with the induction of a regulatory cytokine response and suppression of antigen-specific T cell proliferative responses in draining lymph nodes.

A number of mechanisms of immune suppression have been described in reports of studies of the response to administration or targeting of DCs in vivo with the aim of inducing tolerance, such as the generation of regulatory populations, as well as immunomodulation by natural killer (NK) cells, NK T cells, and myeloid suppressor cells. To determine whether inguinal lymph node T cells from mice treated with DCs could suppress the mBSA response in vitro, inguinal lymph node T cells enriched from mice with untreated AIA were restimulated with splenic DCs and mBSA, and various numbers of T cells enriched from the lymph nodes of DC-treated mice were added to the cultures.

T cells from DC-treated mice failed to suppress the T cell proliferative or cytokine response of untreated mice to mBSA in vitro. When examined ex vivo, the proportion of FoxP3+,CD25+ T cells in inguinal lymph nodes did not differ in these mice with AIA, whether they were untreated (mean ± SEM 4 ± 1.7% of CD4+ T cells) or were treated with DCs (4.1 ± 1.1% of CD4+ T cells). The proportions of NK1.1+,CD3+ and NK1.1+,CD3– T cells were also similar in untreated (0.8 ± 0.2% and 1.4 ± 0.2%, respectively) or treated (0.63 ± 0.1% and 1 ± 0.2%, respectively) mice. However, in lymph node T cells from DC-treated mice restimulated with mBSA, treatment with anti-TGFβ or anti–IL-10 mAb promoted both the proliferative response and the production of IFNγ, TNFα, and IL-6 (Figure 6). These data suggest that following administration of BAY 11-7082–treated DCs, TGFβ and IL-10 produced by regulatory T cells suppress T cell proliferation and proinflammatory cytokines produced in response to antigen.

Figure 6.

Regulation of proliferative and proinflammatory cytokine responses. A, A total of 1 × 105 T cells purified from draining lymph nodes obtained from untreated mice with antigen-induced arthritis (AIA) on day 10 were incubated with 2 × 104 CD11c+ splenic dendritic cells (DCs) from naive mice and with various concentrations of methylated bovine serum albumin (mBSA) for 72 hours, with or without 2.5 × 104 or 5 × 104 T cells purified from draining lymph nodes of DC-treated mice (left). Values are the mean ± SEM 3H-thymidine uptake in 5 mice per group. T cell proliferation and interferon-γ (IFNγ) and interleukin-10 (IL-10) production in pooled supernatants from 5 mice per group are also shown (right). Values are the mean ± SEM cytokine level. B, Lymphocytes were purified from draining lymph nodes (DLNs) obtained on day 10 from DC-treated mice. Cells were incubated for 72 hours with 2 × 104 CD11c+ splenic DCs from naive mice and either isotype control, anti–transforming growth factor β (anti-TGFβ), or anti–IL-10 receptor (anti-IL-10R) monoclonal antibody, with or without 1 μg/ml of mBSA. Shown are the levels of T cell proliferation and production of IFNγ, tumor necrosis factor (TNF), and IL-6 in pooled supernatants from 5 mice per group. Values are the mean and SEM of triplicate wells. Results are representative of 2 separate experiments.

DISCUSSION

In the present study, we examined whether antigen-exposed DCs modified with the drug BAY 11-7082 could be used to suppress active inflammatory arthritis in an antigen-specific manner. For these experiments, we used the AIA model in mice, in which arthritis develops as a result of Th1-type immunity to the mBSA antigen. Our data demonstrate that a single subcutaneous administration of mBSA-exposed BAY 11-7082–treated DCs suppresses the clinical and pathologic features of AIA with an efficacy equivalent to that of TNFα blockade, even when administered after full clinical expression of the disease has occurred. Furthermore, this suppression was antigen-specific and associated with a reduced mBSA-specific immune response in vivo. Disease suppression was IL-10–dependent and was associated with Th2 and regulatory-type immune deviation of the anti-mBSA antibody response and IL-10– and TGFβ-dependent suppression of proinflammatory cytokine production. Of interest, the antigen-specific suppression induced in mice with AIA by BAY 11-7082–treated DCs could be reversed by administration of IL-1β to the site of inflammation in the joint.

The ability of mBSA-exposed BAY 11-7082–treated DCs to suppress AIA through these mechanisms supports previous evidence that antigen-exposed BAY 11-7082–treated or RelB-deficient DCs could suppress a previously primed immune response in vivo (17). Our data demonstrate that DCs that differentiated in the presence of an NF-κB inhibitor and were then exposed to antigen can very efficiently suppress the mBSA-dependent inflammatory arthritis in this model.

Several alternative approaches to the suppression of inflammatory arthritis with DCs have previously been used. After genetic modification using either IL-4 or FasL, DCs injected after type II collagen priming reduced the severity of collagen-induced arthritis (CIA) (22–24). Whereas each of these approaches suppressed Th1-mediated T cell and antibody responses, the immune response typically did not deviate toward a Th2-type or regulatory T cell response. In contrast, DCs generated in the presence of vasoactive intestinal peptide (VIP) were able to suppress several mouse models of autoimmunity, including experimental autoimmune encephalomyelitis (EAE) and CIA (25). Yet another type of regulatory DC exposed to TNFα before injection (referred to as “semimature”) and delivered intravenously in high doses was shown to suppress CIA, EAE, and experimental thyroiditis in a partially IL-10–dependent manner and to be associated with Th2-type immune deviation (26–28).

Although the mechanism by which arthritis is suppressed by IL-4–treated or FasL-treated DCs has not yet been fully elucidated, both TNF-treated DCs and VIP-treated DCs have been shown to induce IL-10–producing T regulatory cell type 1 (Tr1) Treg cells, and in the case of VIP-treated DCs, disease suppression occurs in an IL-10–dependent and TGFβ-dependent manner (25, 27). While systemic administration of VIP expands CD4+,CD25+,FoxP3+ Treg cells in vivo, this has not been shown for VIP-treated DCs. However, VIP has been shown to reduce DC NF-κB activation and CD40 expression, similar to the pattern previously described for BAY 11-7082–treated DCs (17, 25). The current study and previous studies of BAY 11-7082–treated DCs suggest similar mechanisms of suppression are involved, including induction of IL-10–producing Treg cells, without expansion of FoxP3+ Treg cells, which suppress proliferation and IL-2 production (29), IL-10–dependent suppression of pathologic changes, IL-10– and TGFβ-dependent suppression of proliferation and proinflammatory cytokine production, and an antibody profile consistent with immune deviation and regulation. It should be noted that contributions by NK T cells to the observed cytokine production by draining lymph node lymphocytes cannot be excluded, and it will be of interest to determine their role in future experiments.

Depending on the setting of induction, Tr1 cells have been shown to vary in their production of cytokines other than IL-10. For example, IL-15 sustains the production of IFNγ by Tr1 cells (29, 30). Although T cells with a typical Tr1 phenotype from draining lymph nodes of mice given BAY 11-7082–treated DCs were unable to suppress mBSA-specific responses in vitro, it is possible that the induction conditions, migration to sites of pathology, or the presence of proinflammatory cytokines prevented suppression in this assay (31). The speed and antigen-specificity of the clinical response to BAY 11-7082–treated DCs suggest that these DCs rapidly alter the response of existing mBSA-specific immune effector T cells to a Tr1 phenotype, rather than induce Treg cells from naive precursors, very likely as a result of a collaborative interaction with constitutive FoxP3+ Treg cells and innate immune effectors, including NK cells, NK T cells, and macrophages (32–34).

IL-10 and TGFβ are important partners that drive the immunoregulatory response after tolerizing immunotherapy with either mucosal antigen, allergen-specific immunotherapy, or DCs (35–41). After specific immunotherapy, allergen-specific T cell immunity deviates toward the production of IL-10 and TGFβ, and allergen-specific IgA, IgG1, and IgG4 isotypes are induced (35, 36). In humans, IgG4 and IgA represent noninflammatory antibody isotypes (42). We observed a similar induction of mBSA-specific IgG1 and IgA, along with suppression of IgG2b/c Th1-type isotypes, after administration of mBSA-exposed BAY 11-7082–treated DCs. Thus, the induction of antigen-specific immunoregulation after both DC immunotherapy and specific immunotherapy correlates with the concomitant induction of IgA and Th2-type immune deviation of antibody. Data from the current study suggest that immunoregulatory cells stimulated by BAY 11-7082–treated DCs in mice with AIA exert in an IL-10–dependent manner regulatory effects on innate immune pathology (43), concomitant with their effects on B cell isotype switching, through altered proinflammatory cytokine secretion by CD4+ T cells and macrophages (44). The increase in mBSA-specific IgG1 and IgA observed after administration of BAY 11-7082–treated DCs is consistent with the TGFβ- and IL-10–dependent suppressive activity in draining lymph nodes.

Methylation of BSA alters the charge of the BSA molecule from negative to positive, thereby allowing antigen penetration into fibrocartilage and existing within large aggregates in the joint cavity for prolonged periods (45). The AIA model is dependent on CD4+ T cells and Th1-type immunity, as well as the presence of mBSA-specific antibodies and mBSA immune complexes in the serum and joint tissues of animals with established disease (19, 46). A role for presentation of autoantigen has been hypothesized in AIA based on increased T cell responses to type II collagen and proteoglycan in arthritic animals (47). Following administration of BAY 11-7082–treated DCs, pathologic evidence of synovial inflammation, loss of cartilage integrity, pannus formation, and bone erosion were all suppressed by the time the animals were euthanized. Joint histology scores correlated well with clinical arthritis scores. This mirrors a similar lack of radiographic progression in patients with RA correlating with good clinical disease suppression achieved using disease-modifying drugs (48). The current data suggest that such suppression of histologic progression is also achievable in an antigen-specific manner.

Administration of IL-1α or IL-1β to an arthritic joint at a late stage of AIA, when joint inflammation is “smoldering,” has previously been shown to induce a disease flare for ∼24 hours. Delivery of IL-1 and a second dose of antigen produced a more marked and sustained flare (21, 49). While the sustained antigen-induced flare was shown to be T cell–dependent, the transient flare induced by IL-1 still occurred after depletion of T cells and neutrophils with anti–lymphocyte globulin and nitrogen mustard, respectively. In the current studies, local administration of IL-1β induced a marked and sustained flare of AIA that did not spontaneously remit in the absence of DCs or sTNFR. Thus, IL-1 reversed the mBSA-specific tolerance in the joint that had been induced by DCs and Treg cells, and a further DC treatment restored disease suppression.

We have recently shown that IL-1β inhibits the suppressive capacity of CD4+,CD25+ Treg cells in vitro and in vivo. IL-1 stimulated proliferation and IFNγ production by CD4+,CD25+,FoxP3– effector/memory T cells and deactivated Treg cell function (50). This is of clinical relevance, given the role of IL-1 in joint inflammation in RA and juvenile idiopathic arthritis (JIA). For example, IL-1 overexpression signatures have been demonstrated, and recombinant IL-1 receptor antagonist was shown to suppress arthritis in a high proportion of patients with systemic-onset JIA (51). Thus, although IL-1 has important proinflammatory effects on myeloid and joint parenchymal cells and promotes T cell activation and cytokine production, we show here that BAY 11-7082–treated DCs are nevertheless able to restore tolerance even in settings of IL-1 overexpression.

Findings of the present study suggest that antigen-specific immunotherapy, using DCs modified with NF-κB–inhibitory agents, should be applicable to patients with inflammatory arthritides, including RA and JIA, even after inflammatory disease has begun. Recent understanding of the integral role played by citrullinated protein autoantigens, as well as autoantibodies that recognize these proteins, in the pathogenesis of disease, particularly among individuals carrying shared epitope HLA alleles (52), raises the possibility of testing antigen-specific therapeutic strategies in patients with RA and potentially also in at-risk subjects.

AUTHOR CONTRIBUTIONS

Dr. Thomas had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Martin, Capini, O'Sullivan, Thomas.

Acquisition of data. Martin, Capini, Duggan, Lutzky, Stumbles.

Analysis and interpretation of data. Martin, Capini, Pettit, O'Sullivan, Thomas.

Manuscript preparation. Capini, O'Sullivan, Thomas.

Statistical analysis. Martin, Capini, Thomas.

Ancillary