The anti- or pro-inflammatory effect of an adenosine receptor agonist on the Th17 autoimmune response is inflammatory environmental-dependent

Abstract

Adenosine is a key endogenous signaling molecule that regulates a wide range of physiological functions, including immune system function and inflammation. Studies have shown that adenosine receptor (AR) agonists can be either anti- or pro-inflammatory in immune responses and in inflammation, and the clarification of the mechanisms causing these opposing effects should provide a better guide for therapeutic intervention. While previous studies mostly examined the effects of AR agonists on Th1-type immune responses, in this study, we compared their effect on Th17 and Th1 autoimmune responses in experimental autoimmune uveitis (EAU), a mouse model of human uveitis induced by immunization with the human interphotoreceptor retinoid-binding protein peptide IRBP_1–20. We showed that injection of mice with a non-selective AR agonist, NECA, at an early stage after immunization had an inhibitory effect on both Th1 and Th17 responses, whereas injection of the same amount of NECA at a late stage, inhibited the Th1 response, but had an enhancing effect on the Th17 response. We also showed that the effects of NECA on Th1 and Th17 responses were completely dissociated, that the enhancing effect of NECA on Th17 responses was modulated by γδ T cells, and that the response of γδ T cells to NECA was determined by their activation status. We conclude that the inflammatory environment has a strong impact on converting the effect of AR agonist on the Th17 autoimmune response from anti- to pro-inflammatory. Our observation should help in the designing of better AR-targeted therapies.

Introduction

Previous studies have shown that the appropriate generation and clearance of extracellular accumulated adenosine in inflammation are critical in limiting tissue pathology (1–4). Experimental studies have shown that adenosine receptor (AR) agonists and antagonists are promising pharmacological modulators of disease-associated inflammation and immune responses (5–10), but it has been difficult to achieve reproducible beneficial effects because of a lack of knowledge of how adenosine exerts anti- or pro-inflammatory effects (11–13). Previous studies have proposed that the pro- and anti-inflammatory effects of adenosine are produced by activation of different ARs. For example, the suppressive effects of adenosine are mainly mediated by A2A receptor (A2AR) signaling (14–16), whereas A2BR signaling mostly enhances immune responses (13; 17–20). However, this cannot explain the observations that A2AR^−/− mice show increased susceptibility to autoimmune disease and that administration of an A2AR antagonist to mice with actively induced experimental autoimmune encephalomyelitis (EAE) inhibits, rather that enhances, disease development (21), implying that whether an AR agonist is anti- or pro-inflammatory is sophisticatedly regulated. To identify factors that modulate or convert the enhancing and inhibiting effects of AR agonists, we asked whether T cell subsets at different degrees of activation respond differently to the same AR agonist and whether environmental factors modulate the anti- and pro-inflammatory effects of an AR agonist. Since previous studies examining the effect of AR agonists on immune responses have mainly looked at the IFN-γ-producing (or Th1) cell response and since our ongoing study of the regulation of the Th17 autoimmune response in EAU showed that γδ T cells have a strong regulatory effect on Th17 autoreactive T cells (22–25) and that these cells express high levels of ARs (manuscript submitted), we asked whether AR agonists affect the Th1 and Th17 responses differently and whether the regulation of the Th17 autoimmune response by AR agonists is associated with the regulatory activity of γδ T cells in the Th17 response. Here, using a mouse model of uveitis in which B6 mice are immunized with the human IRBP peptide IRBP_1–20 to induce EAU, thus promoting the in vivo activation of Th1 and Th17 autoreactive T cells (22; 25–27), we showed that, while AR agonists always had an inhibitory effect on the Th1 autoimmune response, their effect on the Th17 autoimmune response could be either inhibitory or enhancing, depending on both environmental conditions and the activation status of the T cells. A single early injection of the immunized mice (day 0–5 after immunization with IRBP_1–20) with an AR agonist had a suppressive effect on the Th17 response, whereas treatment at a later date (day 7–10 post-immunization), when inflammation was already been initiated, had an enhancing effect on the Th17 response. Mechanistic studies showed that the enhancing effect of an AR agonist required the presence of γδ T cells and was greatly diminished when the γδ T cells were functionally deficient. Moreover, an enhancing effect was only seen for Th17 responses. Our observation that the pro- and anti-inflammatory effects of an AR agonist can be converted by environmental factors implies that successful AR-targeting treatments or immunomodulation requires monitoring of existing environmental conditions.

MATERIALS AND METHODS

All animal studies conformed to the Association for Research in Vision and Ophthalmology statement on the use of animals in Ophthalmic and Vision Research. Institutional approval was obtained from the Institutional Animal Care and Use Committee (IACUC) of the Doheny Eye Institute, University of Southern California, and institutional guidelines regarding animal experimentation were followed.

Animals and reagents

Female C57BL/6 (B6) and TCR-δ^−/− mice on the B6 background, purchased from Jackson Laboratory (Bar Harbor, ME), were housed and maintained in the animal facilities of the University of Southern California. Recombinant murine IL-1, IL-7, and IL-23 were purchased from R & D (Minneapolis, MN). Fluorescein isothiocyanate (FITC)-, or phycoerythrin (PE)-conjugated antibodies against the mouse αβ T cell receptor (αβ TCR), γδ TCR, IL-17, IFNγ, or CD44 and isotype control antibodies were purchased from Biolegend (San Diego, CA). The non-selective AR agonist 50-N-ethylcarboxamidoadenosine (NECA), the selective A2AR agonist 2-p-(2-carboxyethyl) phenethylamino-5′-N-ethylcarboxamidoadenosine (CGS21680), and the selective A2BR agonist (BAY 60-6538) were purchased from Sigma-Aldrich (St. Louis, MO, USA).

Immunization and NECA treatment

EAU was induced in B6 mice by subcutaneous injection of 200 μl of emulsion containing 200 μg of human IRBP_1–20 (Sigma-Aldrich, St. Louis, MO) in complete Freund’s adjuvant (CFA; Difco, Detroit) at six spots at the tail base and on the flank and intraperitoneal (i.p.) injection with 300 ng of pertussis toxin, as described previously (23; 25; 27).

For NECA treatment, immunized B6 mice received a single i.p. injection of NECA (5 μg/kg) on different days after immunization; the day of injection was day 0 (early) or day 7 (late) post immunization. Controls were injected with vehicle only.

T cell preparation

αβ and γδ T cells were purified from B6 mice immunized with IRBP_1–20 as described previously (22; 25; 27). Nylon wool-enriched splenic T cells were incubated sequentially for 10 min at 4°C with FITC-conjugated anti-mouse γδ TCR or αβ TCR antibodies and for 15 min at 4°C with anti-FITC Microbeads (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), then the cells were separated into bound and non-bound fractions on an autoMACS™ separator column (Miltenyi Biotec GmbH). The purity of the isolated cells, determined by flow cytometric analysis using PE-conjugated antibodies against αβ or γδ T cells, was >95%.

Assessment of Th1 and Th17 polarized responses

Responder T cells (3 × 10⁶) prepared from IRBP_1–20-immunized B6 or TCR-δ^−/− mice were co-cultured for 48 h with IRBP_1–20 (10 μg/ml) and irradiated spleen cells (2 × 10⁶/well) as antigen-presenting cells (APCs) in a 12-well plate under either Th17 polarized conditions (culture medium supplemented with 10 ng/ml of IL-23) or Th1 polarized conditions (culture medium supplemented with 10 ng/ml of IL-12), then IL-17 and IFN-γ levels in the culture medium were measured using ELISA kits (R & D) and the number of antigen-specific T cells expressing IL-17 or IFN-γ determined by intracellular staining followed by FACS analysis.

Cytoplasmic staining

After 5 days’ of in vitro stimulation of the in vivo primed T cells with the immunizing antigen and APCs, activated T cells were separated using Ficoll gradient centrifugation and stimulated in vitro for 4 h with 50 ng/ml of phorbol myristic acetate (PMA), 1 μg/ml of ionomycin, and 1 μg/ml of brefeldin A (Sigma, St. Louis, MO). The cells were then fixed, permeabilized overnight with Cytofix/Cytoperm buffer (eBioscience, San Diego, CA), and intracellularly stained with antibodies against IFN-γ or IL-17 and analyzed on a FACScalibur.

Immunofluorescence flow cytometry

Aliquots of 2 × 10⁵ cells were double-stained with combinations of FITC- or PE-conjugated monoclonal antibodies. Data collection and analysis were performed on a FACS_calibur flow cytometer using CellQuest software.

Limiting dilution analysis (LDA)

B6 and TCR-δ^−/− mice immunized with IRBP_1–20/CFA were left untreated or were injected with NECA, then the spleen and draining lymph nodes were removed 13 days after immunization and pooled and a single-cell suspension prepared. T cells were enriched by MACS column purification and seeded in 24 replicates in two sets of 96-well flat-bottomed culture plates containing irradiated spleen cells (1 × 10⁵ per well) under Th1 or Th17 polarizing conditions, with one set of plates containing an optimal dose of immunizing peptide (10 μg/ml). Based on preliminary LDA estimates of frequencies, the number of T cells seeded in each well varied from 3 × 10³ to 2 × 10⁵. After 44 h of incubation, the plates were pulsed with 0.5 μCi of [³H]thymidine/well for 6 h, harvested, and assessed for isotope incorporation. Positive microcultures were defined as those in which the level of incorporated thymidine exceeded the mean level in control cultures (no responders) by more than three standard deviations (SDs). The frequency of responder T cells was obtained by estimates of precursor frequency calculated using a program developed to analyze LDA data (28–30) that uses the Poisson distribution to calculate the frequency of responder T cells with 99% confidence limits.

Induction of EAU by cell transfer

For induction of EAU by adoptive transfer, T cells were isolated from lymph node and spleen cells on days 12–14 post-immunization (optimal time for proliferation and cytokine responses) and stimulated for 48 h with 10 μg/ml of IRBP_1–20 in the presence of irradiated syngeneic APCs, then activated T cell blasts were separated by Ficoll gradient centrifugation and transferred into B6 mice (1.5 × 10⁶ activated cells per mouse).

Scoring of EAU

The mice were examined three times a week for clinical signs of EAU by indirect fundoscopy. The pupils were dilated using 0.5% tropicamide and 1.25% phenylephrine hydrochloride ophthalmic solutions and fundoscopic grading of disease performed using the scoring system described previously (31). For histopathological evaluation, whole eyes were collected at the end of the experiment and immersed for 1 h in 4% glutaraldehyde in phosphate buffer, pH 7.4, then transferred to 10% formaldehyde in phosphate buffer until processed. The fixed and dehydrated tissues were embedded in methacrylate, then 5 μm sections were cut through the pupillary-optic nerve plane and stained with hematoxylin and eosin. Presence or absence of disease was evaluated blind by examining six sections cut at different levels for each eye. Disease was graded pathologically based on cellular infiltration and structural changes (32).

Assessment of the γδ T cell response to AR agonists

γδ T cells were isolated from naïve or immunized B6 mice using a MACS column, then 5 × 10⁴/well separated γδ T cells were cultured for 48 h in 96-well plates in medium with or without NECA (100 ng/ml) and the supernatants collected for assessment of IL-17 levels.

Assessment of the enhancing effect of γδTCR⁺ T cells on Th17 autoreactive T cells

αβ T cells (1 × 10⁶) from immunized TCR-δ^−/− mice were stimulated with the immunizing peptide in 24-well plates for 5 days in the presence or absence of γδ T cells (5% of total cell number) isolated from immunized B6 mice that had either been left untreated or had been incubated for 48 h with 100 nM NECA, then the proliferating cells were separated on a Ficoll gradient, intracellularly stained with anti-IL-17 antibodies, and subjected to FACS analysis.

Statistical analysis

Experiments were repeated 4–5 times. Experimental groups were typically composed of four mice and the figures show the data from a representative experiment. The statistical significance of differences between the values for different groups was examined using the two-tailed Student’s t test.

Results

1. Whether an AR agonist has an anti- or proinflammatory effect in mice with actively induced EAU mice depends on whether it is administered before or after onset of inflammation

To examine the effect of an AR agonist on the Th17 autoimmune response and whether this was affected by the time of injection of the agonist, B6 mice were immunized with the uveitogenic peptide IRBP_1–20 in CFA, then randomly divided into three groups (n=4). One group received an i.p. injection of 100 ng/mouse of 5′-N-ethylcarboxamidoadenosine (NECA), a non-selective AR agonist (13) that binds to both A2ARs and A2BRs, during the early stage after IRBP_1–20 immunization (day 0-day 5), another group was treated with NECA 7–10 days after immunization, and the third group was left untreated. A kinetic study of the effects of NECA administration at different time points showed a change in effect from anti- to pro-inflammatory at days 6–7 (not shown), so, in subsequent studies, two representative time points were used, with the agonist being injected on day 0 (during immunization), designated as “early treatment”, or on day 7 after immunization, designated as “late treatment”. At day 13 after immunization (the time point at which the highest T cell response is seen), serum cytokine levels were measured and responder T cells from the spleens and draining lymph nodes were subjected to in vitro stimulation with the immunizing peptide (10 μg/ml) and APCs (irradiated spleen cells) under culture conditions that favor Th17 autoreactive T cell expansion (medium containing 10 ng/ml of IL-23) (25; 28; 33) and the activated T cell blasts separated by Ficoll gradient centrifugation and stained intracellularly with FITC-labeled anti–IL-17 antibodies. Fig. 1 shows that the responses of T cells from mice that received NECA treatment differed greatly from those of T cells from untreated mice, being either inhibited or enhanced, depending on the time point when NECA was injected. Fig. 1A shows that serum IL-17 levels were reduced by early NECA treatment and increased by late treatment. Intracellular staining of in vitro stimulated IRBP-specific T cells showed that late treatment resulted in a significantly higher percentage of IL-17⁺ αβ T cells (Th17 response) than in mice not treated with NECA, whereas early treatment resulted in a significantly depressed Th17 response compared to control mice (Fig. 1B and C). Measurement of secreted IL-17 levels showed that responder T cells from late treatment mice produced much higher levels of IL-17 than T cells from non-NECA-treated mice, whereas cells from early treatment mice produced minimal amounts of IL-17 (Fig. 1D). We also performed limiting dilution assay (LDA), which measures the frequency of in vivo primed antigen-specific Th17 cells before in vitro expansion (25; 28; 33). As shown in Fig. 1E, the frequency of in vivo primed IL-17⁺ IRBP_1–20-specific T cells was significantly increased in late treatment mice (>40 per 100,000 responder T cells compared to 22 in controls) and significantly decreased in early-treated mice (< 5 per 100,000 responder T cells). We also compared the pathogenic activity of the Th17 polarized cells derived from the different groups of mice. Our results showed that the number of IRBP_1–20-specific Th17 cells in late treatment mice was significantly higher (2.5 × 10⁶ per mouse) than that in non-NECA-treated mice (1.0 × 10⁶/mouse) and was reduced in early treatment mice (0.4 × 10⁶/mouse) (Fig. 2A). Moreover, as shown by fundoscopy (Fig. 2B) and pathological examination (Fig. 2C), when 1.5 × 10⁶ activated IRBP-specific T cells were adoptively transferred into naïve mice by i.p. injection, the induced EAU was significantly more severe using cells from late treatment mice than using cells from control mice, while no disease was seen using cells from the early treatment group, demonstrating that the late treatment mice generated more IL-17⁺ pathogenic T cells.

Fig. 1. NECA treatment can either suppress or enhance the Th17 response in mouse EAU.

Groups of B6 mice were immunized with IRBP_1–20/CFA alone or were also injected with NECA (100 ng/mouse) either on the day of immunization (early treatment) or 7 days after immunization (late treatment). All the mice were euthanized 13 days post-immunization and sera were collected and T cells from the draining lymph nodes and spleens separated, counted, and subjected to in vitro stimulation with an optimal dose of immunizing peptide (10 μg/ml) and APCs (irradiated spleen cells) in medium supplemented with 10 ng/ml of IL-23, then the activated T cells were separated on a Ficoll gradient and subjected to various analyses. The results shown (except in 1B) are the mean ± SD for one study using 4 mice and the experiment was repeated 4–5 times with similar results. **, p < 0.01.

A) Serum IL-17 levels measured by ELISA.

B) Cytoplasmic staining assessing the percentage of IL-17⁺ cells among the in vitro antigen-stimulated IRBP-specific T cells. After 5 days’ of in vitro stimulation, the activated T cells were treated with PMA, ionomycin, and brefeldin, then were intracellularly stained with PE-conjugated anti-αβTCR antibodies and FITC-conjugated anti-IL-17 antibodies, followed by FACS analysis.

C) Percentage of IL-17⁺ cells in the αβ T cells after in vitro stimulation of in vivo primed T cells with the immunizing peptide IRBP_1–20.

D) IL-17 levels in the supernatant of in vitro cultured T cells after exposure to immunizing antigen and APCs.

E) LDA results. Responder T cell frequencies were evaluated by LDA as detailed in the Materials and Methods.

Fig. 2. Pathogenic activity of IRBP-specific T cells isolated from immunized mice with or without early or late injection of NECA.

Three groups of donor mice were immunized with a pathogenic dose of IRBP/CFA with or without early or late injection of NECA (100 ng/mouse), then T cells from draining lymph nodes and spleens were separated and subjected to in vitro stimulation with IRBP peptide and syngeneic APCs in medium supplemented with 10 ng/ml of IL-23. Activated T cells (1.5 × 10⁶) were then adoptively transferred by i.p. injection into naïve B6 mice and EAU scored. The results shown are the mean ± SD for one study using 4 mice and the experiment was repeated 3 times with similar results. A) Number of IRBP-specific T cells isolated from spleen and draining lymph nodes per IRBP-immunized mouse in each group. (B) EAU in the recipients of IRBP-specific T cells evaluated by fundoscopy every five days or by pathologic examination of the eye 10 days after cell transfer (C).

2. Comparison of the effect of selective and non-selective AR agonists

Since NECA non-selectively binds to all four types of ARs, we examined whether the effect was mediated via a specific AR. Groups of B6 mice were immunized with IRBP/CFA and either left untreated or injected on day 0 or day 7 with an agonist specific for A2ARs (CGS 21680, 1 mg/kg body weight) or A2BRs (BAY 60-6538) or the non-selective NECA, then, on day 13, serum IL-17 levels (Fig. 3A), the percentage of proliferating IL-17⁺ cells among the in vitro stimulated T cells (Fig. 3B and C), and IL-17 production by the responder T cells after in vitro stimulation (Fig. 3D) were measured. The results showed that the A2AR agonist (CGS21680) and NECA had a similar inhibitory effect on Th17 responses compared to the non-agonist-treated group in the early treatment group, but a stimulatory effect in the late treatment group. In contrast, the A2BR agonist caused a moderate increase in the Th17 responses regardless of the time of treatment, suggesting that the non-selective agonist NECA mainly acts via A2ARs, which can exert either an enhancing or inhibitory effect on the Th17 response.

Fig. 3. Comparison of the effects of a non-selective AR agonist (NECA) and specific A2AR or A2BR agonists.

Groups of B6 mice were immunized with IRBP/CFA, then were either left untreated or were injected on the day of immunization or 7 days later with NECA (non-selective; 5 μg/kg), CGS 21680 (A2AR-specific agonist, 1 mg/kg), or BAY 60-6538 (A2BR agonist, 1 mg/kg) as indicated. All the mice were euthanized on day 13, then serum IL-17 levels were measured (A) and the relative number of in vitro primed, IL-17⁺ IRBP-specific Th17 responders compared to that in the untreated group (control) measured by LDA (B), and in vitro differentiation towards Th17 cells (C) and the production of IL-17 by in vivo primed T cells after in vitro antigenic stimulation (D) assessed as described in Fig. 1. Except in panel 3C, the results are the mean ± SD for one study using 4 mice and the experiment was repeated 4–5 times with similar results. **, p < 0.01.

3. Differential effects of an AR agonist on Th1 and Th17 responses

To determine whether AR agonists have the same or a different effect on different types of responder T cells, we also examined the effect of NECA on Th1 responses. First, we measured serum IFN-γ levels at day 13 in immunized mice with or without early or late NECA treatment. As shown in Fig. 4A, both sets of NECA-treated mice produced significantly less IFN-γ and, as shown in Fig. 4B, the frequency of in vivo primed IFN-γ-producing cells was also significantly decreased in both sets of NECA-treated mice. After stimulation with the immunizing peptide under Th1-polarized conditions in vitro, IFN-γ production (Fig. 4C) and the number of IFN-γ-expressing cells (Fig. 4D) were both significantly decreased in both sets of NECA-treated mice, although late treatment had less of an inhibitory effect. These results contrast markedly with the effect on the Th17 responses shown in Fig. 1.

Fig. 4. The effects of NECA on Th1 responses are different from those on Th17 responses.

The Th1 responses evaluated in the same way as described for Th17 responses in Fig. 1. Except in panel 4D, the results are the mean ± SD for one study using 4 mice and the experiment was repeated 4–5 times with similar results. **, p < 0.01.

A) Serum IFN-γ levels assessed by ELISA.

B) LDA assay of the frequency of IFN-γ-producing cells among responder T cells.

C) IFN-γ levels in the culture supernatants of T cells after exposure to immunizing antigen and APCs.

D) Cytoplasmic staining of IFN-γ⁺ cells among the in vitro antigen-stimulated IRBP-specific T cells.

4. The effect of an AR is modulated by γδ T cells

We recently reported that activation of γδ T cells plays a key role in the activation and expansion of a Th17 autoimmune response (22–25). To determine whether the effect of an AR agonist on Th17 autoreactive T cells involved γδ T cells, groups (n=4) of TCR-δ^−/− mice were immunized with IRBP_1–20/CFA and were left untreated or underwent early or late treatment with NECA. As shown in Fig. 5A, on day 13, both sets of NECA-treated TCR-δ^−/− mice had significantly lower serum levels of IL-17 than the non-NECA-treated group. Likewise, both NECA-treated groups had lower numbers of in vivo primed IRBP_1–20-specific Th17 cells as detected by LDA (Fig. 5B), a lower percentage of IL-17⁺ αβ T cells in the in vitro antigen-stimulated T cells (Fig. 5C), and lower IL-17 production by in vitro antigen-stimulated T cells (Fig. 5D), showing that the enhancing effect of late treatment with an AR agonist on Th17 responder T cells requires a γδ T cell response.

Fig. 5. The enhancing effect of late NECA treatment on the Th17 response requires γδ T cells.

Twelve TCR-δ^−/− mice were immunized with IRBP/CFA then separated into three groups, which were left untreated or received early or late NECA treatment, then were tested at day 13 post immunization. The results shown are the mean ± SD for one study using 4 mice and the experiment was repeated 4–5 times with similar results. **, p < 0.01.

(A) Serum IL-17 levels determined by ELISA, (B) frequency of in vivo primed Th17 cells by LDA, (C) percentage of IL-17⁺ cells among the in vitro antigen-stimulated IRBP-specific T cells assessed by cytoplasmic staining, and (D) in vitro cytokine production by the responder T cells determined by ELISA.

5. γδ T cells from immunized mice respond more vigorously to NECA than γδ T cells from naïve mice

To further examine a possible role of γδ T cells in the immunomodulatory effect of NECA in EAU, we determined the effect of in vivo early or late injection of NECA on γδ T cell activation in IRBP_1–20-immunized B6 mice on day 13 after immunization. As shown in Fig. 6A, only 1.8% of total CD3⁺ splenic T cells in naïve mice were γδ T cells, whereas the corresponding numbers were 7.0% in non-NECA-treated immunized mice, 4.9% in immunized mice with early NECA treatment, and 10% in late NECA treatment mice. In addition, the percentage of γδ T cells expressing the T cell activation molecule CD44 was 5.1% in naïve mice and 32% in immunized non-NECA-treated mice and this percentage fell to 18% with early NECA treatment and increased to 63% with late NECA treatment (Fig. 6A), paralleling the effect on the Th17 response. To determine whether the activation status of γδ T cells had an effect on the enhancing effect of the AR agonist on Th17 response in immunized mice, we compared the effect of in vitro NECA treatment (100 nM) on γδ T cells separated from naïve and immunized mice. As shown in Fig. 6B, γδ T cells isolated from naive mice produced little IL-17 after incubation for 48 h in medium with or without NECA, whereas γδ T cells isolated from immunized mice produced moderate amounts of IL-17 in the absence of NECA and levels were significantly enhanced by culture in medium with NECA. Finally, we performed a study in which the percentage of IL-17⁺ T cells was measured among responder T cells from immunized TCR-δ^−/− mice after 5 days of in vitro stimulation in the absence or presence of γδ T cells (5% of the total cell number) from immunized B6 mice that had been either left untreated or had been stimulated for 48 h in vitro with NECA. The results showed that NECA-treated γδ T cells were much more effective than untreated γδ T cells in increasing the percentage of IL-17⁺ αβ T cells (Fig. 6C).

Fig. 6. In vitro treatment with NECA promotes the enhancing effect of γδTCR+ T cells on Th17 autoreactive T cells.

(A) Groups of B6 mice were left untreated or were immunized with IRBP_1–20/CFA, then were left untreated or received early or late NECA injection and all groups euthanized 13 days post-immunization and CD3⁺ T cells from spleen and draining lymph nodes prepared, then the percentage of αβTCR⁺ and γδTCR⁺ T cells among the separated CD3⁺ cells and the surface CD44 expression of γδTCR cells were determined after staining with anti-γδTCR and anti-αβTCR antibodies (left panels) or anti-CD44 and anti-γδTCR antibodies (right panels), followed by FACs analysis.

B) γδ T cells from immunized mice produce more IL-17 in response to in vitro treatment with NECA than γδ T cells from naïve mice. Pooled γδ T cells isolated from naïve or immunized B6 mice (n=4) were incubated with 100 nM NECA for 48 h, then IL-17 levels in the culture supernatants were measured by ELISA. The results shown are the mean ± SD for one study using 4 mice and the experiment was repeated 4–5 times with similar results. **, p < 0.01.

C) αβ T cells from immunized TCR-δ^−/− mice were stimulated in vitro for 5 days in the absence or presence of γδ T cells (5% of total cell number) isolated from immunized B6 mice that had either been left untreated or had been incubated for 48 h with 100 nM NECA.

Discussion

Previous studies have shown that adenosine can either inhibit or enhance an immune response, depending on which of the four identified ARs (34) is (are) activated (13–20). In general, activation of A2ARs has anti-inflammatory effects (15; 16; 35; 36), whereas activation of A2BRs promoted inflammatory responses (11; 18; 37). However, this does not apply to all immune responses. For example, proinflammatory effects have been seen as a result of A2AR activation (11; 18) and anti-inflammatory effects of A2BR activation have been reported (11; 38–40). A study in A2AR^−/− mice showed that they developed more severe EAE than syngeneic B6 mice, but treatment of wild-type B6 mice with an A2AR antagonist inhibited, rather than enhanced, the development of EAE (21), suggesting that the mechanisms by which adenosine agonists/antagonists affect an immune response is more sophisticated than currently thought. Since AR-based treatments have been promoted for the treatment of autoimmune diseases (7–9) and neurological diseases (10) and in transplantation (41), clarification of the mechanisms involved will help determine how to use such molecules for immune modulation.

Early studies investigating the effect of adenosine on T cell biology mostly examined the response of IFN-γ-producing (Th1-type) T cells, so we were interested in determining whether AR agonists are also able to manipulate Th17 pathogenic autoimmune responses in EAU and whether Th1 and Th17 autoreactive T cells respond similarly to AR agonists. Using our recently established system for studying the role of γδ T cells in Th17 autoimmune responses (22–25), we examined whether AR agonists exert their effect by acting on γδ T cells, leading to altered Th17 responses. We found that injection of mice with actively induced EAU with a non-specific AR agonist during the inducing phase had a suppressive effect, whereas AR agonist injection when autoimmune inflammation had already been initiated (7–10 days post-immunization) enhanced the pathogenic response. A study using agonists specific for A2ARs or A2BRs showed that A2AR activation had the strongest modulating effect on the Th17 autoimmune response. Kinetic studies showed that the inhibitory effect was seen when the agonist was injected during the first week of disease induction, while later administration enhanced the response (data not shown). We also demonstrated that the enhancing effect of late AR agonist treatment on the Th17 response was markedly reduced in TCR-δ^−/− mice lacking γδ T cells, suggesting that the functional conversion requires the involvement of γδ T cells. These observations support our previous reports that γδ T cells play an important role in the Th17 autoimmune response and that activation of γδ T cells converts their effect on Th17 responses from anti-inflammatory to pro-inflammatory (23; 42). Our present finding also suggests that excessive production of adenosine may enhance the in vivo activation of γδ T cells and Th17 autoreactive T cells.

We demonstrated that, regardless of treatment time, an AR agonist consistently suppressed Th1 responses, whereas it either enhanced or suppressed the Th17 response, depending on the status of autoimmune inflammation. This unique effect of an AR agonist on Th17 responses is conceivably related to a synergistic effect on γδ T cell activation of adenosine and proinflammatory molecules, such as cytokines and TLR ligands, which preferentially promotes Th17, but not Th1, responses, as shown in our previous studies (22; 23; 25; 27). In the present study, we also showed that γδ T cells with different activation status varied significantly in their response to an AR agonist. These results show that, in attempts at manipulating inflammation and the immune response using AR agonists or antagonists, we should take into consideration this two-edged sword effect of adenosine on the Th17 response. An AR agonist might exert a strong suppressive effect on the immune response during the quiescent phase of the disease, but this effect can be converted from anti-inflammatory effect to proinflammatory when the agonist is applied during ongoing disease when inflammation has already occurred. The fact that exogenously administered AR agonists have a different effect on Th1 and Th17 immune responses suggests caution in their use in the treatment of diseases involving both Th1 and Th17 responses.

The different effects of AR agonist on Th1 and Th17 responses involve several immune cells, including γδ, α T cells and DCs. In a recent paper, we have demonstrated that γδ T cells from immunized mice expressed the highest levels of A2AR and possess the strongest binding ability to adenosine, as compared to other immune cells examined, including αβ T cells, DCs, and B cells. A very small (3%) number of the activated γδ T cells can effectively compete adenosine binding with a majority (97%) of the co-cultured αβ T cells. Binding of adenosine by γδ T cells not only diminishes the suppressive effect of adenosine on α T cells but also promotes γ T cells activation, rendering these cells more competitive in adenosine binding. (Manuscript in press in PLoS One). To further clarify the mechanisms by which AR agonist regulate autoimmune responses, we have examined the AR agonist effect on DCs and γδ-DC interactions. In another manuscript, in preparation, we demonstrated that adenosine promoted the differentiation of a DC subset co-expressing CD11c and Gr-1, which possess a strong stimulating effect on Th17 autoreactive T cells. We have also determined the effect of NECA treatment on FoxP3⁺ regulatory cells. Our results did not show a significant effect and the change in number of Foxp3⁺ cells.

In vivo, T cells are exposed to a complex of cytokines, Toll-like receptor ligands, and adenosine metabolites, particularly in an inflammatory environment, and display considerable functional plasticity. Recent studies have shown that the activity of regulatory T cells can be reduced in response to environmental triggers, a process designated as “instability of regulatory T cell function” (43–46). Our results provide support for this by showing that, as one of the major regulatory T cells in the Th17 autoimmune response (22; 23; 25; 27), γδ T cell function was also modulated by inflammatory factors. Adenosine is produced under various stress conditions (36; 47–50) and exerts multifaceted effects on various functions, including that of the immune system, and its effect on immune responses deserves further examination.

Abbreviations

AR

Adenosine receptor

A2AR

A2A adenosine receptor

A2BR

A2B adenosine receptor

EAU

experimental autoimmune uveitis

IRBP

interphotoreceptor retinoid-binding protein

LDA

limiting dilution assay

NECA

5′-N-ethylcarboxamidoadenosine