Green Tea Epigallocatechin-3-Gallate Suppresses Autoimmune Arthritis Through Indoleamine-2,3-Dioxygenase Expressing Dendritic Cells and the Nuclear Factor, Erythroid 2-Like 2 Antioxidant Pathway

DBA/1 mice were orally fed with either phosphate buffered saline (PBS) control or EGCG (10 mg/kg, Sigma, in PBS) using an oral gavage Zonde needle (Thermo Fisher Scientific, Vantaa, Finland), nine times over three weeks, beginning a week after booster immunization.

Bovine type II collagen (CII) was dissolved in 0.05 N acetic acid and was emulsified (1:1 ratio) with an equal volume of complete Freund's adjuvant (CFA); both CII and CFA from Chondrex, Redmond, WA [22, 23]. Mice were immunized by tail base injection on day 0 with an emulsion of CII (100 μg) and CFA (1 mg/ml), followed by a booster injection at a separate site at the base of the tail with an emulsion of CII (100 μg) in incomplete freund's adjuvant (1:1) on day 14 [23]. Starting 18 days after the primary immunization, three independent observers examined the severity of arthritis three times a week for up to 6 weeks. The severity of arthritis was recorded as the mean arthritic index on a 0 to 4 scale according to the following criteria: 0 = no edema or swelling; 1 = slight edema and erythema limited to the foot or ankle; 2 = slight edema and erythema from the ankle to the tarsal bone; 3 = moderate edema and erythema from the ankle to the tarsal bone; and 4 = edema and erythema from the ankle to the entire leg [22]. The final score was an average value of three independent joint evaluations.

Blood was collected from the orbital sinus of EGCG-treated and control mice at the peak of clinical disease. Serum specimens were stored at −20 °C until use, and anti-CII IgG1 and anti-CII IgG2a Ab levels were measured by an enzyme-linked immunosorbent assay (ELISA). Microplates were coated with 4 μg/ml of CII overnight and blocked with 1 % bovine serum albumin (BSA) from Sigma-Aldrich, St. Louis, MO and then incubated with sera at a dilution of 1:16,000. Bound total or CII-specific IgG1 or IgG2a were detected by incubation with horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG1 or IgG2a-specific antibodies (cat # A90-205P and A90-207P from Bethyl Laboratories, Inc., Montgomery, TX) for 1 h. Then the plates were washed with phosphate-buffered saline with Tween 20 buffer (PBST) and developed with 3,3',5,5'-tetramethylbenzidine (TMB) substrate according to the manufacturer’s instructions (Sigma-Aldrich). The reaction was terminated with 4.5 N sulfuric acid (H₂SO₄). The optical density (OD) values were measured at 450 nm using an Automatic Microplate Reader (BLx808, BIO-TEK, Winooski, Vermont).

Red blood cells were depleted from splenocytes and lymph node cells using lysis buffer which contained 10 mM potassium bicarbonate (KHCO₃), 0.15 M ammonium chloride (NH₄Cl) and 0.1 M ethylenediaminetetraacetic acid (EDTA), pH 7.2, and single cell suspensions were prepared and flow cytometric analysis was performed using a FACSCalibur (BD Biosciences, San Jose, CA) with BD CellQuest Pro Software (BD Biosciences) and the data was analyzed using FloJo Software (FlowJo, LLC, Ashland, OR). For analysis of lymphocytes the following rat anti-mouse antibodies were used: CD4-PerCP-Cy5.5 (clone RM4-5), CD8-PE (clone 53–6.7), CD21/35-FITC (clone 7G6) and CD23-Biotin (clone B3B4) with Streptavidin- allophycocyanin (APC); all antibodies and second step reagents from BD Biosciences. Tregs were identified using anti-mouse FoxP3-FITC (clone FJK-16a; eBioscience) and CD25-APC (clone 3C7; Biolegend). CD11b⁺ cells were identified using rat anti-mouse CD11b-PerCP (clone M1/70; BD Biosciences) and indoleamine‐2,3‐dioxygenase (IDO)-positive cells were detected with rat anti-mouse IDO (clone mIDO-48; Biolegend) followed by FITC goat anti-Rat Ig secondary antibody (cat# 554016, BD Biosciences).

Single cell suspensions isolated from draining lymph nodes were stimulated with 25 ng/ml PMA (Sigma) and 250 ng/mL ionomycin (Sigma). GolgiStop (BD Biosciences) was added and the cells were harvested after 5 h of culture. Cells were first stained extracellularly with anti-CD4 PerCP-Cy5.5-conjugated (clone RM4-5) antibody, then fixed and permeabilized with Perm/Fix solution, and finally stained intracellularly with anti- IFN-γ FITC-conjugated (clone XMG1.2), and anti- TNF-α PE-conjugated (clone MP6-XT22) antibodies. Directly conjugated isotype-matched rat anti-mouse antibodies were used as controls for nonspecific staining. All reagents for fixation and staining were from BD Biosciences and the protocol was carried out following the manufacturer’s instructions.

Draining inguinal lymph nodes (dLNs) were aseptically excised, minced and single cell suspensions were cultured in triplicate in RPMI 1640 containing fetal calf serum (10 % vol/vol), 2-mercaptoethanol (20 μM), L-glutamine (1 % wt/vol), penicillin (100 U/mL), and streptomycin (100 μg/mL) in the presence or absence of CII (Chondrex) or rat anti-mouse CD3 antibody (clone 17A2; BD Biosciences). During the last 16–18 h of the three-day assay, cells were pulsed with 1 uCi of [³H]-thymidine (Perkin Elmer, Waltham, MA) per well. The incorporation of [³H]-thymidine was determined using a Betaplate scintillation counter (Perkin-Elmer, Waltham, MA).

Hind paws and knees were removed from the sacrificed mice, frozen in liquid nitrogen and lysed using a buffer containing 20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 % Triton X-100, protease inhibitor cocktail (Complete Mini, Roche, Indianapolis, IN) and phosphatase inhibitor cocktail (PhosSTOP, Roche, Indianapolis, IN). The crude extract was then sonicated for 30 s. The homogenate was centrifuged at 20,000 X g for 15 min, and the resulting supernatant was collected. The total protein content of samples was quantified using the Bradford assay (Sigma). 10 μg of each sample was resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) using a 4–15 % Mini-PROTEAN TGX Precast Gel (Biorad, Hercules, CA), transferred to a polyvinylidene difluoride (PVDF) membrane (Biorad, Hercules, CA), and probed with primary antibodies including rabbit antibody to Nuclear Factor, Erythroid 2-Like 2 (anti-Nrf2; catalog no. ab31163), rabbit antibody to phosphorylated Nrf-2, anti-p-Nrf2 (ab76026) and a mouse IgG1 monoclonal antibody that recognizes mouse heme oxygenase-1, anti-HO-1 (ab13248, clone HO-1-1); all from Abcam and all have been previously shown to react with murine proteins. The loading control was an anti-GAPDH (clone 6C5; Advanced ImmunoChemical) which reacts to mouse and other species. Appropriate HRP-conjugated secondary antibodies which included goat anti-rabbit or goat anti-mouse (Jackson ImmunoResearch) were used at 1:5000 and detected with the SuperSignalWest Femto Chemiluminescent Substrate Kit (Thermo Scientific). Protein expression levels were visualized and quantitated using the gel documentation system, G:BOX (Syngene, Frederick, MD).

The indoleamine‐2,3‐dioxygenase (IDO) enzyme activity assay was performed as previously reported [24] with some modifications. In brief, freshly isolated DCs were washed, resuspended in sterile Hank’s Balanced Salt Solution (HBSS; Sigma-Aldrich) containing 500 uM tryptophan (Sigma-Aldrich), and incubated for 4 h. The supernatants were then harvested and assayed for kynurenine. For the assay, 30 ul of 30 % trichloroacetic acid was added to 60 ul of culture supernatant and the mixture was vortexed and centrifuged at 10,000 X g (12,000 rpm) for 5 min. Then, 40 ul of supernatant was added to an equal volume of Ehrlich reagent (5 ml of glacial acetic acid with 100 mg P-dimethylamino-benzaldehyde). The OD was measured at 492 nm using a NanoDrop (LMS). Purified _L-kynurenine (0–500 uM; Sigma-Aldrich) was used as the standard.

Spleens from mice were collected, embedded in Tissue-Tek Optimal Cutting Temperature (O.C.T.) compound and snap-frozen in liquid nitrogen. Cryosections (6 μm thick) were fixed with 4 % paraformaldehyde, blocked with 10 % horse serum for 30 min and stained with various antibodies. Anti-Foxp3-FITC (clone FJK-16a; e-Bioscience) and anti-IDO (clone mIDO-48; Biolegend) or Rat IgG2b isotype control (Biolegend) were detected with goat anti-rat Alexa-555 secondary (catalog # A-21434; Invitrogen) and anti-APC-CD11b (clone M1/70; BD Bioscience). Fluorescence images were acquired using an LSN510 confocal microscope (Carl Zeiss, Oberkochen, Germany). For quantification of immunofluorescence (IF) 10 representative high-powered fields per slide were assessed from at least 4 to 5 samples/group.

Serum specimens were assayed for IL-1β, IL-6, IFN-γ, TNF-α and IL-10 by ELISA using Duoset assay kits from R&D Systems and following the manufacturer’s instructions.

Mice were sacrificed seven weeks after immunization. The spleens obtained from the mice were treated with RPMI 1640 (Invitrogen) containing dithiothreitol (DTT) and EDTA for 90 min at 37 °C to remove the epithelial cells and then washed with HBSS and digested with DNase. CD11b⁺ DCs cells were separated from splenocyte suspensions from EGCG-fed or PBS-fed mice using a mouse CD11b positive selection kit (catalog # 18770; Stem Cell, Canada) and were co-cultured with CD4⁺CD25⁻ T cells isolated from EGCG-fed mice for 3 days in the presence or absence of CII (10 μg/ml, Chondrex Inc, USA). Before the stimulation with CII, the cells were pretreated with the IDO-specific inhibitor 1-methyl dL tryptophan (1-MT), obtained from Sigma-Aldrich, for two hours. To measure the amount of intracellular Foxp3 in CD4⁺CD25⁺ T cells, cells were stained using a Regulatory T Cell Staining Kit (eBioscience, San Diego, CA, USA) as described in flow cytometry above.

Hind paws and knees were obtained from each mouse, the skin was trimmed, and the joints were fixed in 10 % phosphate-buffered formalin for 1 day, decalcified in 15 % EDTA for 3 weeks and embedded in paraffin. Tissue sections (6 μm) were prepared and stained with hematoxylin, eosin and safranin O. Inflammation and joint damage were assessed by scoring five parameters. Disease was scored on a scale of 0 to 3 for inflammation ranging from no inflammation to severe inflammation, loss of proteoglycans ranging from fully stained to destained cartilage, cartilage destruction ranging from appearance of dead chondrocytes to complete loss of the articular cartilage, and was scored on a 0 to 5 scale for loss of bone ranging from no damage to complete loss of bone structure. For histology of both knee and paw, a composite score was calculated by summing the individual parameters.

1-MT was obtained from Sigma-Aldrich and was prepared as a 20-mmol/l stock in 0.1 N NaOH (pH 7.4) and stored at −20 °C in the dark. EGCG fed mice were given 2 mg/ml 1-MT solution, supplemented with aspartame using foil-wrapped, standard autoclaved drinking bottles. Mice drank an average of 5 ml/d, and water was replaced as needed for 3 weeks.

Phenotypes were examined using repeated measures one-way analysis of variance (ANOVA), with treatment and time as fixed factors and mouse number as the random factor. Data from in vitro and ex vivo experiments were analyzed for statistical significance using the Mann-Whitney U test (GraphPad Prism 6, GraphPad Software, Inc., San Diego, CA, USA). A p value < 0.05 was taken as the level of significance. In all experiments, * indicates P < 0.05, ** P < 0.01, and *** P < 0.001.

We determined whether EGCG modulated disease activity in a murine collagen-induced arthritis (CIA) model. DBA/1 mice were immunized on days 0 and 14 with bovine type II collagen (CII), as detailed in the methods. Mice were fed with EGCG nine times over three weeks with 10 mg/kg, starting on day 21 after induction of arthritis throughout the disease course and vehicle-fed or EGCG-fed mice were observed for 49 days for the development of clinical arthritis (Fig. 1). Mice treated with PBS (vehicle control) developed the typical signs of CIA both in terms of clinical score and paw swelling seven weeks after the initial immunization (Fig. 1a, b). In contrast, EGCG-treated mice displayed a significant decrease in severity of arthritis and paw thickness compared to animals treated with vehicle alone (Fig. 1a, b). As expected, vehicle-fed CIA mice developed severe symptoms of arthritis including marked swelling, redness, and erythema of the hind paws and the forepaws. In contrast, EGCG-fed mice exhibited markedly reduced clinical manifestations of fully developed CIA (Fig. 1c).

The observed clinical paw inflammation was also confirmed by assessment of H&E stained histological sections of mouse hind paws. The histological images of the paws shown in Fig. 1d are representative of mice in each group at day 49 when disease was found to be at its maximum. Joint tissue samples from arthritic vehicle-fed mice revealed the expected histopathological changes, including marked synovial hyperplasia, erosion, and loss of articular cartilage and bone. In this group, there was extensive cartilage and bone erosions with massive infiltration of polymorphonuclear and mononuclear leukocytes as indicated by the elevated histopathological score (Fig. 1e). The joint swelling and pannus formation appeared to be dependent on disease severity. A similar analysis of arthritic joints from the EGCG-fed group showed a marked reduction in the number of infiltrating leukocytes, with less visible cartilage or bone erosion when compared with the vehicle-fed CIA mice (Fig. 1d, e). Thus, these results indicate that EGCG reduced infiltration of inflammatory cells and diminished the severity of arthritis as assessed by clinical and histological scores. Consistent with previous reports, our results demonstrate that the oral administration of EGCG successfully ameliorates disease activity in an inflammatory arthritis model.

Previous studies have shown that humoral immunity plays an essential role in the pathogenesis of CIA [25, 26]. Our results indicate that, in EGCG-treated mice CII-specific IgG2a antibodies were dramatically reduced (P < 0.01) when compared to vehicle-fed CIA mice (Fig. 1f). In contrast, the titers of anti-CII specific IgG1 antibodies were increased in EGCG-treated mice compared with vehicle-treated mice (P < 0.5; data not shown). These results suggest an important role for EGCG in modulating B cell responses, as well as Th1/Th2 balance in the ongoing immune response.

Previous reports have shown that arthritis is associated with the presence of Th1 CD4⁺ T effector cells secreting high levels of IFN-γ while there is an absence of Th2 effectors in the arthritic synovium in murine CIA models and RA [20, 25, 27]. IFN-γ induces activation of macrophages that produce proinflammatory cytokines, such as TNF-α and IL-1β which are abundant in arthritic joints in animal models and RA [25, 28, 29]. We therefore measured pro-inflammatory cytokines in the serum and knee homogenates of the experimental mice (Fig. 2). At day 49 after induction of CIA, EGCG significantly reduced serum levels of IL-6, TNF-α, and IFN-γ compared to vehicle-fed control CIA mice. However, there was no statistical difference in the production of IL-1β while IL-10 levels were elevated in the serum of EGCG-fed mice (Fig. 2a). Similar results were obtained when pro-inflammatory cytokines were measured in knee homogenates (Fig. 2b). The levels of IL-1β, IL-6, TNF-α and IFN-γ were significantly lower and IL-10 was higher in EGCG-fed mice when compared with vehicle-fed CIA mice. We also quantitated the frequency of cells producing Th1 cytokines including IFN-γ and TNF-α in the draining lymph nodes of EGCG-fed mice and vehicle-fed control mice (Fig. 2c, d). In concordance with the results for the serum and joint cytokine profiles, Th1 cell frequencies were significantly reduced in the EGCG-fed mice as compared to the controls. Our results suggest that consumption of EGCG limited inflammatory Th1 cell numbers and cytokines, which may have contributed to the altered anti-CII antibody isotype profiles and reduced disease severity.

RA is characterized by chronic inflammation of the synovium which could result in part from the infiltration of activated immune cells including CD4⁺ T cells, B cells, and antigen presenting cells including DC and macrophages [2, 20, 27, 29]. To identify the changes in immune cell populations after EGCG supplementation, we conducted flow cytometric analysis (Fig. 3). EGCG reduced the frequencies of B cells (CD5⁻B220⁺) and the major T-cell subsets (CD4⁺CD8⁻ and CD4⁻CD8⁺) in the dLNs on day 49 in CIA (Fig. 3a, b, d). We next examined the frequencies of different splenic B cell subsets including the follicular (Fo, CD21^intCD23^hi) and marginal zone (MZ, CD21^hiCD23^lo) B cells [30]. EGCG treatment significantly reduced the frequencies of Fo and MZ B cells in the spleen (Fig. 3c, d). Finally, mice treated with EGCG had significantly increased frequencies of CD4⁺CD25⁺Foxp3⁺ T-regulatory (Treg) cells in the dLNs (Fig. 3e, f). There was an average 2-fold increase in Treg frequencies in EGCG-treated mice compared to the vehicle-fed CIA mice.

Activation and cytokine production by CD4⁺ T cells play an important role in the pathogenesis of inflammatory arthritis. Upon antigen encounter, clonal expansion of Ag-specific T cells is a prerequisite for the initiation and development of the T-cell mediated immunopathological features [20, 22, 23]. We next determined whether the EGCG-induced reduction in severity of murine CIA might be mediated by its impact on Ag-specific T-cell responses. Draining lymph nodes (dLN) were collected from collagen-immunized/boosted DBA/1 mice after 49 days following vehicle or EGCG treatment. Cells cultured in the absence of CII, exhibited basal levels of DNA synthesis, as measured by incorporation of [³H]-thymidine, that did not differ between the vehicle-fed and EGCG-fed groups (Fig. 4a, b). In contrast, cells from vehicle-fed mice cultured in the presence of CII exhibited a robust antigen recall response, which was significantly reduced in cells derived from EGCG-fed mice (Fig. 4a). Mice treated with EGCG also demonstrated significantly reduced T cell DNA synthesis when stimulated with anti-CD3 (Fig. 4b). Consistent with the above results, CFSE assays demonstrated that EGCG reduced CD4⁺ T cell proliferation (Fig. 4c, d). These results indicate that administration of EGCG attenuates antigen-induced T cell proliferation in murine CIA.

IDO is an intracellular enzyme that catabolizes the essential amino acid tryptophan, expressed in monocytes, dendritic cells (DC) and macrophages. IDO has been shown to suppress T cell proliferation by mechanisms that remain to be elucidated [31]. Active IDO can be induced by IFN-γ, endotoxin, a CTLA-4 fusion protein via CD80/CD86 ligation and the combination of PGE₂ and TNF-α [24, 32]. A previous study demonstrated that IDO expressing DC in Peyer’s patches play an essential role in the induction of oral tolerance in CIA [31]. To examine whether DC in EGCG-fed mice express IDO as one mechanism for subduing CIA, we performed immunofluorescence staining for IDO and CD11b on tissue sections of dLN obtained on day 49 after disease induction. As shown in Fig. 5a, b, IDO staining (red) was clearly increased in a substantial proportion of the CD11b⁺ DC (blue) from the dLN of EGCG-fed mice as compared to the vehicle controls. Consistent with the above results, flow cytometric analysis of the dLN and spleens revealed that the CD11b⁺IDO⁺ DC were significantly increased after EGCG treatment (Fig. 5c, d). We next examined whether the DC-expressed IDO was enzymatically active. As shown in Fig. 5e, the culture supernatants of CD11b⁺ DC isolated from EGCG-fed mice contained high levels of kynurenine, the first catabolite in the tryptophan metabolic pathway, compared to the controls. These studies demonstrate that the EGCG-induced IDO expressed by CD11b⁺ DCs was functionally active.

We next investigated whether the CD11b⁺ DC induced by EGCG were more potent in generating regulatory T cells (Treg). Highly purified CD4⁺CD25⁻ T cells isolated from the spleen of EGCG-fed mice (purity > 97 %) were cocultured with CD11b⁺ DC from EGCG or vehicle-fed control CIA mice for 3 days in the presence or absence of CII. Figure 6a,b shows that the frequency of CD4⁺CD25⁺Foxp3^bright T cells was increased when CD4⁺CD25⁻ T cells were co-cultured with CD11b⁺ DC from EGCG-fed CIA mice in the presence of CII antigen than when CD4⁺CD25⁻ T cells were co-cultured with CD11b⁺ DC from vehicle-fed CIA mice. To determine whether IDO was mechanistically involved in DC-mediated antigen-specific Treg generation after EGCG treatment, 1-MT, an IDO inhibitor, was added to the CII antigen-stimulated cultures. Indeed, 1-MT abrogated the increase in the proportion of CD4⁺CD25⁺ Foxp3⁺ T cells induced by CD11b⁺ DC from EGCG-fed CIA mice (Fig. 6a, b) which correlated with arthritis scores (Fig. 6d). Thus, our findings demonstrate that highly purified CD4⁺CD25⁻ T cells can be converted into T regulatory cells (Tregs) by splenic DC obtained from EGCG-fed mice by antigen-specific stimulation with CII through an IDO-dependent mechanism. Previous studies have demonstrated the presence of IDO expressing cells frequently juxtaposed to Tregs in other disease models. For example, IDO expressing myeloid DC and macrophages in cutaneous granulomas were adjacent to Foxp3⁺ cells [33]. We next investigated whether EGCG treatment promoted a physical association between IDO expressing CD11b⁺ DC and Tregs in the spleen. Immunofluorescence analysis of EGCG-treated mice revealed that Foxp3 and IDO expressing CD11b⁺ cells were present and adjacent to each other in spleens to a greater extent than in control vehicle-treated mice (Fig. 6c). Thus, these studies correlate with the in vitro observation that IDO expressing CD11b⁺ DC can augment Treg numbers.

Nuclear factor, erythroid 2-like 2 (Nrf-2) is a transcription factor that plays a major role in cellular defense against oxidative stress by inducing proteins, such as Heme oxygenase-1 (HO-1), that inactivate reactive oxygen species. Increased levels of Nrf-2 result in enhanced antioxidant activity, which then suppresses inflammation in several animal models [34‐36]. Although increases in total Nrf2 expression was variable in joint homogenates, there was a consistently dramatic enhancement of phosphorylation of Nrf2 (p-Nrf2), indicating pNrf2 activation, observed in arthritic joint homogenates of EGCG-fed compared with vehicle-fed CIA mice (Fig. 7a). The level of p-Nrf2 was significantly elevated in EGCG-fed CIA mice compared with control vehicle-fed mice as determined by densitometry Fig. 7b. We also examined the expression of HO-1, a downstream target of Nrf2. In EGCG-fed animals, HO-1 protein was significantly increased compared with vehicle-fed CIA mice and by densitometry EGCG treatment increased total HO-1/GAPDH ratios relative to control vehicle-fed mice (Fig. 7a, b).

IDO plays a pivotal role in the inflammatory pathology associated with systemic autoimmunity and disease progression [32]. We next examined whether the function of IDO was in part related to the activation of Nrf2 observed in these mice. Thus, EGCG-fed CIA mice were treated with the IDO inhibitor 1-MT for three weeks and monitored for the development and severity of arthritis. 1-MT treated EGCG-fed mice displayed arthritis that was similar to the vehicle-fed control CIA mice. Immunohistochemical analysis of knee joints demonstrated that p-Nrf2 expression was enhanced in EGCG-fed CIA mice and that this increased expression was blocked by 1-MT (Fig. 7e). Examination of Safranin-O stained paws from CIA mice demonstrated that EGCG-fed mice were protected from cartilage degradation in contrast to vehicle-fed and EGCG-fed mice treated with 1-MT (Fig. 7e). In further support of these results, p-Nrf2, Nrf2 and HO-1 from joint homogenates were examined by western blot (Fig. 7c, d). 1-MT-treated EGCG-fed CIA mice had similar levels of all three molecules as compared to vehicle-fed control CIA mice. EGCG-fed mice expressed increased levels of p-Nrf2 as well as total Nrf2 and HO-1. These results are consistent with an IDO-dependent mechanism of enhanced Nrf2 expression in the inflamed joints of CIA mice.