Regulation of Adenosine Deaminase on Induced Mouse Experimental Autoimmune Uveitis

In inflammatory and ischemic conditions, production of endogenous adenosine in the extracellular environment modulates various biological responses, including immune responses (1–4). Newly formed adenosine is rapidly removed from tissues by adenosine-metabolizing enzymes. The discovery of the potent effects of adenosine on inflammation and immune responses led to attempts at treatment of immune dysfunction by targeting adenosine receptor (AR) signaling (2, 3, 5, 6). Unfortunately, the successful application of such treatment has been hindered by our incomplete understanding of the sophisticated purinergic signaling events that occur in various cellular components under different pathophysiological circumstances (3, 7).

Adenosine deaminase (ADA) is an adenosine-degrading enzyme that is expressed in almost all animal tissues (8). Exogenous ADA was initially used to treat immune deficiencies involving ADA dysfunction (4, 9–11), but subsequent studies showed that enhanced ADA function was associated with increased an incidence of autoimmune disease (4, 12) and that suppression of aberrant ADA activity by ADA inhibitors has an anti-inflammatory effect (13, 14). Our interest in the use of ADA to treat autoimmune disease started with our early observation that ligands of adenosine A2A receptors (A2ARs) or adenosine A2B receptors (A2BRs) enhance, rather than suppress, Th17 autoreactive T cell responses (15–17). Because ADA counteracts the effects of adenosine (9, 18, 19), we wished to determine whether it could be used to suppress Th17-type autoimmune responses. In this study, we showed that ADA can inhibit the development of experimental autoimmune uveitis (EAU). We also showed that this effect is dependent on γδ T cell function, is significantly reduced in recipient mice with defective γδ T cell function (TCR-δ^−/− mice), and is restored if the TCR-δ^−/− recipient mice received an injection of γδ T cells before induction of EAU. A kinetic study in which recipient mice were treated during different disease phases confirmed our previous finding (15) that the outcome of AR-targeted treatment is not always consistent and that the effect of treatment can be either pro- or anti-inflammatory, depending upon the immune status of the recipient. The suppressive effect was seen when ADA was administered on days 8–14, shortly before disease expression, whereas disease was exacerbated if ADA was injected before EAU induction or immediately after EAU expression. We conclude that a single injection of ADA can effectively suppress an ongoing autoimmune response; however, to achieve a desirable therapeutic effect and to avoid undesired effects, the immune status of the recipient should first be determined.

Female C57BL/6 (B6) and TCR-δ^−/− mice on the B6 background (The Jackson Laboratory, Bar Harbor, ME) were housed and maintained in the animal facilities of the University of California, Los Angeles and were used at 12–16 wk of age. The experimental protocols were approved by the Institutional Animal Care and Use Committee of the University of California, Los Angeles (Protocol ARC#2014-029-03A). Recombinant murine IL-12 and IL-23 were purchased from R&D Systems (Minneapolis, MN); FITC-, PE-, or allophycocyanin-conjugated mouse mAbs against mouse αβ TCR (clone H57-597), mouse γδ TCR (clone GL3), mouse IL-17, mouse IFN-γ, mouse MHC class II, or mouse CD25, as well as isotype-control Abs, were purchased from BioLegend (San Diego, CA). ADA was a gift from Sigma-Tau Pharmaceuticals (Gaithersburg, MD).

At day 13 postimmunization, CD3⁺ T cells were purified from the spleen or draining lymph nodes of B6 or TCR-δ^−/− mice immunized with peptide IRBP_1–20 (aa 1–20 of human IRBP; Sigma) by positive selection using a combination of FITC-conjugated anti-CD3 Abs and anti-FITC Ab-coated MicroBeads, followed by separation on an autoMACS separator, according to the manufacturer’s suggested protocol (Miltenyi Biotec, Auburn, CA).

αβ T cells, γδ T cells, and dendritic cells (DCs) were isolated from IRBP_1–20-immunized mice at 13 d postimmunization. γδ T cells were separated from the CD3⁺ T cells from IRBP_1–20-immunized B6 mice by positive selection using a combination of FITC-conjugated anti–TCR-δ Abs and anti-FITC Ab-coated MicroBeads, followed by separation using an autoMACS. αβ T cells were prepared from the spleens or draining lymph nodes of IRBP_1–20-immunized B6 or TCR-δ^−/− mice by positive selection, using a combination of FITC-conjugated anti-CD3 Ab and anti-FITC Ab-coated MicroBeads. The purity of the isolated cells was >95%, as determined by flow cytometric analysis using PE-conjugated Abs against αβ or γδ T cells. CD11c⁺ DCs were sorted from spleens of ADA-treated or untreated IRBP_1–20-immunized B6 mice.

To test γδ T responses to DCs (Fig. 6), freshly prepared γδ T cells from immunized TCR-δ^−/− mice were cultured in cytokine-free medium for 5 d (to assure the resting status of the cells, because γδ T cells freshly isolated from immunized mice are activated) and then incubated with DCs for 48 h at a γδ T cell/DC ratio of 10:1.

To induce EAU, B6 mice were injected s.c. at six spots at the tail base and on the flank with a total of 200 μl emulsion consisting of equal volumes of 150 μg peptide IRBP_1–20 in PBS and CFA (Difco) and i.p. with 200 ng pertussis toxin (Sigma). The mice were randomly grouped and injected i.p. with PBS (vehicle), ADA in PBS (5 U/mouse), or erythro-9-(2-hydroxy-3-nonyl) adenine (EHNA) in PBS on day 8 postimmunization. They were examined three times a week until the end of the experiment (day 30 postimmunization). For adoptive transfer, recipient mice were injected i.p. with 2 × 10⁶ IRBP_1–20-specific T cells, prepared as described previously (20, 21), in 0.2 ml PBS.

To examine mice for clinical signs of EAU by indirect fundoscopy, the pupils were dilated using 0.5% tropicamide and 1.25% phenylephrine hydrochloride ophthalmic solutions. Fundoscopic grading of disease was performed using the scoring system described previously (22). For histology, whole eyes were collected at the end of the experiment and prepared for histopathological evaluation. The eyes were immersed for 1 h in 4% phosphate-buffered glutaraldehyde and then transferred to 10% phosphate-buffered formaldehyde until processed. Fixed and dehydrated tissues were embedded in methacrylate, and 5-μm sections were cut through the pupillary-optic nerve plane and stained with H&E.

Responder CD3⁺ T cells (3 × 10⁶), prepared from IRBP_1–20-immunized B6 or TCR-δ^−/− mice, were cocultured for 48 h with IRBP_1–20 (10 μg/ml) and irradiated spleen cells (2 × 10⁶/well) as APCs in a 12-well plate under Th17-polarized conditions (culture medium supplemented with 10 ng/ml IL-23) or Th1-polarized conditions (culture medium supplemented with 10 ng/ml IL-12). IL-17 and IFN-γ levels in the culture medium were measured using ELISA kits (R&D Systems), and the number of Ag-specific T cells expressing IL-17 or IFN-γ ωερε determined by intracellular staining, followed by FACS analysis, as described below (23, 24).

ELISA was used to measure cytokine (IFN-γ and IL-17) levels in the serum on day 13 postimmunization and in the 48-h culture supernatants of responder T cells isolated from immunized B6 or TCR-γ^−/− mice, with or without prior injection of γδ T cells.

In vivo–primed T cells were stimulated with the immunizing Ag and APCs for 5 d, then the T cells were separated using Ficoll gradient centrifugation and stimulated in vitro for 4 h with 50 ng/ml PMA, 1 μg/ml ionomycin, and 1 μg/ml brefeldin A (all from Sigma). Aliquots of cells (2 × 10⁵ cells) were fixed, permeabilized overnight with Cytofix/Cytoperm buffer (eBioscience, San Diego, CA), and intracellularly stained with PE-conjugated anti-mouse IFN-γ Abs or FITC-labeled anti-mouse IL-17 Abs. Data collection and analysis were performed on a FACSCalibur flow cytometer using CellQuest software.

All experiments were repeated four or five times. Experimental groups typically consisted of six 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 Mann–Whitney U test.

We demonstrated previously that Th17 and Th1 autoimmune responses respond differently to treatment with AR agonists, because AR agonists, especially A2AR-specific agonists, suppress Th1 but augment Th17 responses (15, 16). Therefore, we were interested in determining whether removal of adenosine by ADA would suppress the Th17 response and inhibit induction of EAU.

A schematic diagram of disease induction and examination of mice under investigation is shown in Fig. 1A. B6 mice were immunized with the uveitogenic peptide IRBP_1–20 in CFA and randomly divided into two groups (n = 6): one received an i.p. injection of ADA (5 U/mouse) at day 8 postimmunization, and the other received vehicle. At day 13 postimmunization (the time at which the highest T cell response is seen), serum cytokine levels were measured by ELISA; in addition, responder T cells were purified from the spleen and draining lymph nodes and stimulated in vitro with the immunizing peptide and APCs (irradiated spleen cells) under culture conditions that favor Th17 or Th1 autoreactive T cell expansion (medium containing 10 ng/ml of IL-23 or IL-12, respectively) (24, 25), and the T cells were separated by Ficoll gradient centrifugation and stained intracellularly with FITC-labeled anti–IFN-γ or anti–IL-17 Abs. The mice that received ADA had significantly milder disease, as shown by fundoscopy (Fig. 1B) and pathologic examination (Fig. 1C), and recovered significantly earlier than did the untreated mice (Fig. 1B). Measurement of serum cytokines showed that, compared with controls, ADA treatment caused a significant decrease in serum IL-6 and IL-17 levels but a slight increase in serum IFN-γ and IL-10 levels (Fig. 1D). The Th1 and Th17 responses, assessed by intracellular staining of IRBP-specific T cells after 5 d of in vitro stimulation with the immunizing Ag and APCs, revealed that T cells from ADA recipients generated significantly fewer IL-17⁺ αβ T cells than did those from the control mice, whereas the number of IFN-γ⁺ T cells was slightly increased (Fig. 2A). IL-17 and IFN-γ double-positive cells are not abundant in this mouse model. The cytokine-production results measured at 48 h after in vitro stimulation agreed with those obtained by intracellular staining. As shown in Fig. 2B, responder T cells from ADA-treated mice produced significantly less IL-17 than did T cells from nontreated mice when activated under Th17-polarizing conditions (culture medium containing 10 ng/ml IL-23; upper panel), whereas there was little difference in the amount of IFN-γ produced by the two sets of responder T cells under Th1-polarizing conditions (culture medium containing 10 ng/ml IL-12; lower panel). We also compared the pathogenic activity of the IRBP-specific T cells isolated from treated and untreated mice when transferred into naive recipients and found that T cells from ADA-treated mice had decreased EAU-inducing activity (Fig. 2C).

EHNA is a reversible inhibitor of ADA (26, 27). To determine the effect of injecting EHNA into mice after the start of the EAU-induction process, two groups (n = 6) of B6 mice were injected with IRBP_1–20. On day 8 postimmunization, one group received a single i.p. injection of EHNA in PBS (10 mg/kg), and the other received PBS. In contrast to the results with ADA treatment, compared with controls, EHNA-treated mice had a significantly higher EAU clinical score (Fig. 3A), significantly higher serum IL-17 levels (Fig. 3B), and a significantly higher percentage of IL-17⁺ αβ T cells among the in vivo primed responder T cells after 5 d of in vitro stimulation with immunizing peptide and APCs (28% compared with 16% in controls, Fig. 3C), with little difference in the percentage of IFN-γ⁺ T cells (data not shown). Responder T cells produced significantly higher levels of IL-17 than did those from nontreated mice when activated under Th17-polarizing conditions (Fig. 3D).

We reported previously that γδ T cells are important in enhancing Th17 autoimmune responses (15, 24, 28, 29) and that the effect of an AR agonist on EAU is γδ T cell dependent (15, 16). To determine whether the ADA effect was also affected by γδ T cell function, groups (n = 6) of wild-type B6 and TCR-δ^−/− mice, with or without transfer of γδ T cells (2 × 10⁶/recipient) from immunized B6 mice, were immunized with IRBP_1–20/CFA and injected with ADA or PBS on day 8 postimmunization. Samples taken at day 13 postimmunization showed that the immunized TCR-δ^−/− mice had significantly lower serum levels of IL-17 than did the immunized B6 mice, whereas serum IFN-γ levels were the same in both groups (Fig. 4A). Moreover, TCR-δ^−/− mice did not respond to ADA treatment, as assessed by serum IL-17 levels (Fig. 4A), the amount of IL-17 secreted by in vitro–activated Th17 cells (Fig. 4B), or the number of IL-17⁺ T cells generated (Fig. 4C). However, after injection of γδ T cells before immunization, TCR-δ^−/− mice produced increased levels of IL-17 in the serum, and this effect was inhibited by ADA injection (Fig. 4A). The results of intracellular staining for IL-17⁺ αβ T cells among the responder T cells (Fig. 4C) agreed with the serum cytokine study, showing that ADA treatment suppressed the generation of IL-17⁺ T cells in B6 mice, but not in TCR-δ^−/− mice, and that TCR-δ^−/− mice that received an i.p injection of γδ T cells before immunization showed an increased ability to generate IL-17⁺ T cells that was inhibited by ADA.

Mechanistic studies showed that, in ADA-treated immunized B6 mice, γδ T cell activation was inhibited, as demonstrated by the smaller percentage (4.6% versus 7.4%) of γδ TCR–expressing T cells among the CD3⁺ T cells in ADA-treated mice and the smaller percentage of γδ T cells expressing the T cell activation marker CD44 (Fig. 5A). In contrast, mice treated with the ADA inhibitor (EHNA) showed a higher percentage (9.5% versus 6.6%) of γδ T cells among the CD3⁺ T cells, as well as a higher percentage (78% versus 63%) of γδ T cells expressing CD44 (Fig. 5B).

Previous studies showed that ADA promotes T cell activation by affecting DC function (12, 14, 19). To determine whether ADA modulates the Th17 autoimmune response in EAU by its effect on DC functions, we compared the Th1- and Th17-stimulatory activity of DCs isolated from ADA-treated or untreated mice. As shown in Fig. 6A, when splenic DCs isolated from the spleen of ADA-treated and untreated immunized B6 mice were incubated for 48 h with responder CD3⁺ T cells from untreated immunized B6 or TCR-δ^−/− mice, and cytokine levels in the supernatants measured by ELISA, DCs isolated from ADA-treated mice were less able to stimulate IL-17 production (Fig. 6A). We also compared the γδ T cell–stimulating effect of splenic DCs isolated from ADA-treated and untreated immunized B6 mice by incubating them with γδ T cells from immunized B6 mice and measuring IL-17 levels in the 48-h culture supernatants. We found that γδ T cells produced significantly less IL-17 after incubation with DCs from ADA-treated mice than after incubation with DCs from untreated mice (Fig. 6B). Given our previous finding that the CD25⁺ DC subset has a strong stimulatory effect on Th17 autoreactive T cells (28, 30), we examined whether ADA treatment would suppress the differentiation of CD25⁺ DCs, leading to decreased Th17 activation. ADA administration did not change the total numbers of splenic CD11c⁺ cells, but it significantly reduced the absolute and relative numbers of splenic CD25⁺CD11c⁺ cells (data not shown). As shown in Fig. 6C, the percentage of splenic CD11c⁺ cells from ADA-treated immunized B6 mice that expressed CD25 was much smaller (14%) than that for untreated immunized B6 mice (27%).

We reported previously that administration of an AR agonist to mice immunized with IRBP_1–20 can have an enhancing or inhibitory effect on EAU induction, depending on the timing of treatment (15). To determine whether the timing of ADA treatment was important for its effect, we performed a kinetic study in which ADA was administered to mice at different days after immunization with IRBP_1–20. We found that ADA administration on day 8 postimmunization significantly ameliorated EAU, whereas administration on day 1 significantly enhanced EAU (Fig. 7A, 7B). This enhancing effect was seen when ADA was injected between days 1 and 5 postimmunization, whereas an inhibitory effect was seen when ADA was administered between days 8 and 14 postimmunization (data not shown).

To determine how ADA treatment at different time points resulted in opposite effects on EAU, we examined the effect of ADA on γδ T cell activation in vivo, IL-17 levels in the serum, and IL-17 production by autoreactive T cells after in vitro stimulation. Our results showed that the enhancing effect of ADA treatment on day 1 was associated with increased γδ T cell activation in the treated mice, with an increase in the percentage of γδ T cells among the CD3⁺ T cells (10.2% versus 7.4%) and in the percentage of γδ T cells from treated mice that expressed CD44 (80% versus 69%) (Fig. 7C, middle panels). The suppressive effect of ADA treatment on day 8 was associated with decreased γδ T cell activation in the treated mice (Fig. 7C, bottom panels). Measurement of serum IL-17 levels (Fig. 7D) and IL-17 production (Fig. 7E) by in vitro–activated autoreactive T cells supported the conclusion that the Th17 response was significantly enhanced by treatment with ADA on day 1 and significantly inhibited by treatment on day 8.

We reported previously that, in EAU, AR agonists enhance Th17 responses but suppress Th1 autoimmune responses (15–17). To determine how adenosine metabolism regulates Th17 autoimmune responses and whether adenosine-degrading enzymes would reverse the proinflammatory effect of AR agonists, we tested the effect of ADA.

Studies showed that extracellular adenosine damps excessive inflammatory responses (31–34), ADA reduces extracellular adenosine levels (9, 35) and generates proinflammatory effects (12, 19), and treatment with an ADA inhibitor is immunosuppressive (11, 13, 36). Evidence supporting a proinflammatory effect of ADA includes the observations that in vitro treatment promotes human and mouse T cell responses (12, 19) and that an ADA inhibitor reduces tissue injury in a mouse model of enteritis (13). Therapeutic use of exogenous ADA has also been applied in bone marrow transplantation and hematopoietic stem cell gene therapy; it resulted in suppression of the graft-rejection response (9). In the current study, we demonstrated that ADA treatment at days 8–14 postimmunization suppressed development of EAU by inhibiting the Th17 autoimmune response. Our results demonstrate that diseases that preferentially involve Th17 responses are more responsive to ADA treatment.

We reported previously that the CD25⁺CD11c⁺ DC subset has an enhanced stimulating effect on γδ T cells, leading to an augmented Th17 response (28, 30, 37). In a study to determine possible mechanisms leading to increased activation of the CD25⁺ DC subset, we showed that A2BR ligation is an important promoting factor for activation of the CD25⁺CD11c⁺Gr-1⁺ DC subset (17, 37). Therefore, we hypothesize that degradation of adenosine by ADA may restrain differentiation and activation of CD25⁺ DCs and, thereby, inhibit autoreactive Th17 responses. However, further investigations of the mechanism by which ADA differentially affects Th1 and Th17 responses are required.

We reported previously that AR agonists can inhibit or enhance an autoimmune response, depending on when the agonist was administered, with the suppressive effect prevailing if the agonist is administered before the beginning of the inflammatory response and the enhancing effect prevailing if it is administered after the inflammatory response is established (15). To determine whether such a time-dependent effect was also seen with ADA treatment, we injected mice with ADA before the start of inflammation (within a few days after EAU induction; days 0–5) or after the start of the inflammatory response (1 wk after disease induction; days 8–14). We found that the timing was important in determining the effect of ADA treatment, because ADA administration immediately before, or just after, immunization with IRBP_1–20 caused exacerbation of EAU. We conclude that, like the effect of adenosine (15), the effect of ADA is dependent on the immune status of the recipient and environmental factors. The different effects of ADA in different disease models were previously attributed to binding of adenosine to cell type–specific ARs (33, 38–41), as well as to different concentrations of adenosine produced in the local environment (2–4). It was suggested (2–4) that extracellularly accumulated adenosine could be anti- or proinflammatory. Lower levels of extracellular adenosine are preferentially bound by high-affinity A2ARs, leading to an inhibitory outcome; however, as tissue damage develops and local adenosine levels increase, the binding of adenosine by low-affinity A2BRs that are mainly proinflammatory leads to exacerbation of inflammation and damage (2–4). Such a scenario is supported by this study showing that elimination of adenosine by ADA in the induction phases of the disease generates a proinflammatory effect when the existing adenosine is low; however, the effect may become anti-inflammatory when the existing adenosine levels are exceedingly high, such as when the disease approaches its peak.

Our previous studies demonstrated that Th17 responses are compromised under conditions in which γδ T cells are functionally defective (15, 16, 24, 30, 42, 43). In the current study, we showed that suppression of EAU by ADA was closely associated with decreased γδ T cell activation, further suggesting that modulation of γδ T cell function might be an effective means of controlling Th17 autoimmune responses. We attempted to address the question of why both ADA and an ADA inhibitor have different effects on Th1 and Th17 responses. Our unpublished observations (D. Liang, A. Zuo, R. Zhao, H. Shao, H.J. Kaplan, and D. Sun) showed that ADA has a suppressive effect on Foxp3⁺ T cells; conceivably, a low Foxp3⁺ T cell activity offsets the suppressive effect of ADA on Th1 responses. These results agree with previous reports showing that both adenosine and ADA interfere with regulatory T cell function (35, 44, 45). Because of previous concerns that prolonged ADA treatment is more likely to cause immune dysregulation (4), we looked at the therapeutic effect of a single injection of ADA and found that a single, appropriately timed injection could be effective.

A complete understanding of the functional diversity of adenosine requires further intensive investigations. The knowledge acquired will allow us to design improved therapeutics or combined treatments.

The authors have no financial conflicts of interest.