The adenosine metabolite inosine is a functional agonist of the adenosine A_2A receptor with a unique signaling bias

Abstract

Inosine is an endogenous purine nucleoside that is produced by catabolism of adenosine. Adenosine has a short half-life (approximately 10 s) and is rapidly deaminated to inosine, a stable metabolite with a half-life of approximately 15 h. Resembling adenosine, inosine acting through adenosine receptors (ARs) exerts a wide range of anti-inflammatory and immunomodulatory effects in vivo. The immunomodulatory effects of inosine in vivo, at least in part, are mediated via the adenosine A_2A receptor (A_2AR), an observation that cannot be explained fully by in vitro pharmacological characterization of inosine at the A_2AR. It is unclear whether the in vivo effects of inosine are due to inosine or a metabolite of inosine engaging the A_2AR. Here, utilizing a combination of label-free, cell-based, and membrane-based functional assays in conjunction with an equilibrium agonist-binding assay we provide evidence for inosine engagement at the A_2AR and subsequent activation of downstream signaling events. Inosine-mediated A_2AR activation leads to cAMP production with an EC₅₀ of 300.7 μM and to extracellular signal-regulated kinase-1 and -2 (ERK1/2) phosphorylation with an EC₅₀ of 89.38 μM. Our data demonstrate that inosine produces ERKl/2-biased signaling whereas adenosine produces cAMP-biased signaling at the A_2AR, highlighting pharmacological differences between these two agonists. Given the in vivo stability of inosine, our data suggest an additional, previously unrecognized, mechanism that utilizes inosine to functionally amplify and prolong A_2AR activation in vivo.

1. Introduction

Inosine is an endogenous purine nucleoside formed by deamination of adenosine. As with adenosine, it is produced and released into the extracellular space during normal cell metabolism. Adenosine has a short half-life of less than 10 s in vivo [1] and is rapidly deaminated to inosine by adenosine deaminase. Inosine has a much longer in vivo half-life than adenosine of approximately 15 h [2]. In interstitial fluids, the basal level of inosine is in the micromolar range and can be two to seven times that of adenosine [3-6], During ischemia and sepsis interstitial inosine levels rise dramatically and in certain disease states can be greater than 1 mM [7] and remain elevated longer than those of adenosine [8], Consequently, inosine levels become markedly greater than those of adenosine under disease conditions (>10-fold; [3-6]). The biological significance of the high levels of inosine in vivo is poorly understood.

As has been observed for adenosine, inosine exerts a wide variety of anti-inflammatory and immunomodulatory effects in vivo and in vitro. These include inhibition of proinflammatory cytokine and chemokine production [9-11], stimulation of anti-inflammatory cytokine production [9], protection from TNBS-induced colitis [12], LPS-induced endotoxemia [13,14], LPS-induced acute lung injury [10], glycodeoxycholic acid-induced acute pancreatitis [15], streptozotocin-induced and non-obese type 1 diabetes [16] as well as improvement of islet transplant survival [17]. However, the molecular mechanisms underlying the anti-inflammatory and immunomodulatory effects of inosine are incompletely understood. They do not require cellular uptake of inosine, suggesting involvement of cell surface receptors in this process. However, the presence of cell-surface receptors specific for inosine has not been documented. It is widely stated that the biological effects of inosine are mediated through ligation of specific membrane-bound G protein-coupled receptors (GPCRs) termed P₁-purinoceptors, also known as adenosine receptors (AR; [18]).

There are four pharmacologically distinct adenosine receptor subtypes, termed A₁R, A_2AR, A_2BR and A₃R [19], A_2AR and A_2BR are coupled to the stimulatory G protein Gαs while A₁R and A₃R are coupled to inhibitory G protein Gi [19]. Thus, adenosine occupation at the A_2AR and A_2BR leads to an increase in intracellular cAMP levels, whereas adenosine ligation at the A₁R and A₃R results in a reduction in intracellular cAMP levels. Among the four AR subtypes, A_2AR is the most effective in downregulating inflammation through modulation of intracellular cAMP levels. Although inosine is a functional agonist for A₁R and A₃R, with EC₅₀ values of 290 μM and 0.25 μM, respectively, inosine-mediated anti-inflammatory effects resemble those of the activated A_2AR. In vivo studies utilizing receptor knockout mice demonstrate that the anti-inflammatory properties of inosine are mediated in part through the A_2AR [20,21], Additionally, inosine-mediated inhibition of the production of the pro-inflammatory cytokine TNF-α following LPS-stimulation of peritoneal macrophages is partially reversed by the A_2AR-selective antagonist DMPX in vitro [9]. Nevertheless, efforts directed toward a comprehensive demonstration of inosine as a functional agonist at the A_2AR at the molecular and cellular level or inosine’s ability to bind directly to the A_2AR have been unsuccessful. Hence, inosine is largely considered to be inactive at the A_2AR [22] in spite of a large body of in vivo data suggesting otherwise.

To understand the molecular mechanisms underlying the inosine-mediated anti-inflammatory and immunomodulatory effects, we examined inosine agonism at both the human and mouse A_2ARs utilizing a variety of cell-based and membrane-based assays. Here we demonstrate that inosine not only binds to the A_2AR but also activates A_2AR -mediated cAMP production and ERK1/2 phosphorylation. To the best of our knowledge, this is the first comprehensive demonstration of inosine functional agonism at the A_2AR. Our data suggest the existence of a sustained A_2AR receptor signaling mechanism that is driven by inosine long after its more potent predecessor, adenosine, has been degraded, adding a new dimension to the anti-inflammatory and immunomodulatory role of the A_2AR.

2. Materials and Methods

2.1. Materials

NECA, CGS 21680, and ZM241385 were purchased from Tocris Biosciences. [³H]CGS 21680 was purchased from PerkinElmer Corporation. Rolipram, adenosine, inosine, inosine monophosphate, xanthine, hypoxanthine, adenosine 5′ [α,β-methylene] diphosphate and dipyridamole were purchased from Sigma-Aldrich.

2.2. Cell culture

CHO-K1 cells stably expressing human A_2AR (CHO-hA_2AR; [23]) and human A_2BR (CHO-hA_2BR; [23]) were grown in DMEM/F-12 (1:1) supplemented with 10 % FBS, 2 mM glutamine and G418 (0.2 mg/ml). CHO-K1 cells transiently expressing mouse A_2AR (mA_2AR) were grown in the same medium without G418. All cells were maintained at 37 °C in a 5 % CO₂ incubator.

2.3. CHO-K1 transfection

CHO-K1 cells were transiently transfected with either a pCMV6 vector bearing the mA_2AR gene or the pCMV6 vector alone using Lipofectamine 3000 (Life Technologies) according to the manufacturer’s suggested protocol. Transfected cells were cryopreserved in DMEM/F-12 (1:1) containing 5 % DMSO at −195 °C (liquid nitrogen).

2.4. Dynamic mass redistribution (DMR) assay

DMR responses were assessed using an EnSpire multimode plate reader (PerkinElmer) according to a manufacturer-suggested protocol. Briefly, CHO-K1 and CHO-hA_2AR cells were seeded onto EnSpire-LFC, 96-well microplates (PerkinElmer) and cultured for 18 h at 37 °C in a 5% CO₂ incubator. Before the assay, cells were washed with assay buffer (HBSS with 20 mM HEPES containing either 1 % DMSO alone or 100 nM ZM 241385 and 1 % DMSO) and allowed to equilibrate for 2 h at room temperature near the EnSpire multimode plate reader. Initial baseline DMR measurements were collected for 5 min, agonists and inverse agonist were added and inducible DMR measurements were collected for an additional 15 min.

2.5. Cell-based cAMP assay

CHO-hA_2AR [23], CHO-hA_2BR [23] and CHO-K1 cells transiently expressing mA_2AR were seeded in 96-well half-area white plates (Greiner bio-one; 5 × 10³ cells/well for hA_2AR and hA_2BR; 1.5 × 10⁴cells/well for mA_2AR) in the absence of G418 20 h prior to assay. CHO-hA_2AR cells and CHO-K1 cells transiently expressing mA_2AR were washed twice with Hanks’ balanced salt solution (HBSS) and incubated in HBSS containing adenosine deaminase (ADA; 3 U/ml) and adenosine 5′ [α,β-methylene] diphosphate (50 μM) for 15 min at 37 °C. Cells were washed twice with HBSS to remove ADA and incubated with rolipram (50 μM), adenosine 5′ [α,β-methylene] diphosphate (50 μM), adenosine, and inosine in the presence or in the absence of dipyridamole at indicated concentration(s) for 10 min at 37 °C. Assay conditions for CHO-hA_2BR cells were essentially the same as above with the exception of exclusion of ADA and adenosine 5′ [α,β-methylene] diphosphate pretreatment. Pretreatment was not required for cells expressing the A_2BR since no baseline level cAMP signal was detected in the absence of exogenously added adenosine, likely due to the low affinity of the A_2BR towards adenosine and the small amount of adenosine produced by the cells during the assay which was not sufficient to activate the A_2BR. Intracellular cAMP levels were quantified using an HTRF assay kit (Cisbio).

2.6. Cell-free membrane-based cAMP assay

HEK293-hA_2AR cell membranes (PerkinElmer) were incubated in HBSS, containing adenosine 5′ [α,β-methylene] diphosphate (50 μM) and ADA (3 U/ml) at 37 °C for 20 min. Membranes were washed twice with 33mM HEPES containing 0.1 % Tween 20 and stimulated with the same buffer containing 100 μM ATP, 2 μM GTP, 10 μM GDP, 2 μM MgCl₂, 150 mM NaCl, 50 μM adenosine 5′ [α,β-methylene] diphosphate, 50 μM rolipram, and inosine (0–300 μM) or CGS 21680 (100 nM) in half-area white plates (Greiner bio-one; 4.5 μg protein/well) for 30 min at 37 °C. cAMP levels were quantified using an HTRF assay kit (Cisbio).

2.7. ERK1/2 phosphorylation assay

CHO-hA_2AR cells [23] were seeded in 96-well plates (Greiner bio-one; 2.5 × 10⁴ cells/well) in the absence of G418 20 h prior to assay. Medium was replaced with medium without serum and incubated for an additional 3 h. CHO-hA_2AR cells were washed twice with HBSS and incubated in HBSS containing ADA (3 U/ml) and adenosine 5′ [α,β-methylene] diphosphate (50 μM) for 15 min at 37 °C. Cells were washed twice with HBSS to remove ADA and incubated with rolipram (50 μM), adenosine 5′ [α,β-methylene] diphosphate (50 μM), adenosine or inosine at indicated concentration(s) for 20 min at 37 °C. Assay was terminated by aspirating the assay buffer and incubating cells with lysis buffer (50 (μl/well) at room temperature with shaking for 10 min. Phospho ERK1/2 levels were detected using an Alphascreen Surefire kit (PerkinElmer) according the manufacturer suggested protocol. Briefly, 4 μl of the lysate was transferred to a ProxiPlate-384 (PerkinElmer) and incubated with 7 μl of assay detection mixture at room temperature in the dark for 2 h. Fluorescent emissions were quantified using an EnSpire multimode plate reader (PerkinElmer).

2.8. Radioligand displacement assay

The binding of inosine to hA_2AR at functional concentrations (cell- and membrane-based cAMP and ERK1/2 phosphorylation) was evaluated using the [³H]-labeled A_2AR-selective agonist CGS 21680. HEK293-hA_2AR cell membranes (PerkinElmer; 10 μg/well) were pre-incubated in binding buffer (50 mM Tris, 10 mM MgCl₂, 1 mM EDTA, 4 U/ml ADA, pH 7.4) in a 96-well MultiScreen plate (Millipore) at 27 °C for 15 min. After addition of varying concentrations of inosine and [³H]CGS 21680 (100 nM), plates were incubated for an additional 75 min followed by washing with cold PBS. Binding was quantified using a MicroBeta TriLux microplate scintillation counter (PerkinElmer). Specific binding was calculated by subtracting counts obtained in the presence of NECA (50 μM) from total counts.

2.9. Data analysis

Data were analyzed using GraphPad Prism Software. Dose response curves were generated by non-linear regression with a variable slope. Equilibrium binding of [³H]CGS 21680 was analyzed using one-way ANOVA with alpha equal to 0.05. Pooled EC₅₀ values for cAMP production and ERK1/2 activation were compared using a paired t test.

3. Results

3.1. Inosine generates an A_2AR-selective, inverse agonist-sensitive dynamic mass redistribution (DMR) signal in CHO-hA_2AR cells

Signaling through the GPCR family of receptors is complex. Receptor coupling to more than one G protein subunit leading to different signaling outcomes is common. The A_2AR couples to two different classes of Gα subunits. A_2AR coupling to the Gαs and Gαolf subunits that belong to the Gs class stimulates adenylate cyclase to increase intracellular cAMP upon agonist engagement at the receptor [23], Agonist engagement at the A_2AR also leads to ERK1/2 phosphorylation via either cAMP dependent or independent mechanisms [24], However, there are no reports demonstrating that inosine stimulates the A_2AR to produce cAMP or ERK1/2 phosphorylation. This may be suggestive of weak stimulation by inosine, hence small differences in cAMP production and ERK1/2 phosphorylation are not readily detectable by the analyte-specific assay technologies utilized. We reasoned that an inosine-mediated, pathway-independent, global and integrated signal would be more detectable than individual, isolated, pathway-specific signals such as cAMP or phosphorylated ERK1/2. Thus we utilized a label-free, biosensor-enabled DMR assay capable of capturing and translating agonist-stimulated signaling events into real-time, quantitative, whole cell phenotypic responses. Applying DMR technology to characterize GPCR signaling in living cells has been well established [25]. Since the human A_2AR (hA_2AR) is the most pharmacologically characterized A_2AR [26] relative to other species, we utilized the hA_2AR for interrogating whole cell phenotypic responses emanating from inosine agonism at the A_2AR. New knowledge gained here on the hA_2AR will serve as a guide to the A_2ARs of other species as this receptor is evolutionarily conserved in terms of sequence, structure and function. In CHO-K1 cells stably transfected with hA_2AR (CHO-hA_2AR), both adenosine (Fig. 1B) and inosine (Fig. 1C) dose-dependently triggered a whole cell DMR signal that was qualitatively different from that produced by the CHO-K1 cells (Fig. 1A). The A_2AR-selective inverse agonist, ZM 241385, inhibited the inosine-induced DMR signal (Fig. 1D) indicating that inosine is capable of generating whole cell phenotypic responses through the A_2AR.

Fig. 1.

Dynamic mass redistribution analysis of ARs on CHO-K1 and CHO-hA_2AR cells.

DMR response traces of CHO-K1 (A) and CHO-hA_2AR (B, C & D) stimulated with indicated concentrations of adenosine (Ado; A & B), inosine (Ino; C & D) and 100 nM ZM 241385 (ZM; D). All data are mean values ± SEM of a representative independent experiment (n=3). Data with p < 0.001 and p < 0.05 relative to control are indicated in shaded and dotted boxes, respectively.

3.2. Inosine dose-dependently enhances hA_2AR- but not hA_2BR-mediated cAMP production

To examine whether inosine-mediated, A_2AR-dependent whole cell phenotypic responses are manifestations of increased cAMP production, we evaluated the effects of inosine on cAMP production by CHO-hA_2AR. We developed a highly sensitive cell-based cAMP assay by utilizing adenosine deaminase (ADA) to rid the cells of endogenously produced adenosine and adenosine 5′ [α,β-methylene] diphosphate, an inhibitor of ecto-5′-nucleotidase, to halt endogenous/exogenous production of adenosine and inosine. Using this approach we isolate and decrease background signaling and thereby enhance the signal generated by the addition of inosine. In this assay, adenosine increased cAMP production with an EC₅₀ (half maximal effective concentration) of 6 ± 1 nM and an E_max (maximal stimulation) of 48 nM (Fig. 2A; Table 1). Similarly, inosine dose-dependently increased cAMP production with an EC₅₀ of 300.7 ± 48.9 μM (Table 1) and E_max of 80 nM (Fig. 2B; Table 1). However, EC₅₀ values for adenosine and inosine were significantly different (p < 0.0001). To rule out the unlikely possibility that hA_2AR stimulation was due to the presence of adenosine as an impurity, we analyzed an 800 μM inosine stock solution by HPLC and found no trace of adenosine (≤ 0.025%). The inosine-induced increase in cAMP production was completely inhibitable by the A_2AR inverse agonist ZM 241385, demonstrating the specificity of the inosine-hA_2AR interaction (Fig. 2B; Table 1). As expected, ZM 241385 also inhibited adenosine-induced cAMP production however, the inhibition was partial (Fig 2A; Table 1; EC₅₀=2.45 ± 1.29 μM). These results suggest a potency and/or a pharmacological difference between inosine and adenosine at the A_2AR, and rule out the remote possibility of hA_2AR stimulation by undetectable amounts of adenosine that may be present in inosine.

Fig. 2.

Effect of adenosine and inosine on hA_2AR- and hA_2BR-mediated intracellular cAMP production.

Adenosine stimulated hA_2AR-mediated (A) and hA_2BR-mediated (C) cAMP production, while inosine stimulated hA_2AR-mediated (B) but not hA_2BR-mediated (C) cAMP production. CHO-hA_2AR (A and B) and CHO-hA_2BR (C) cells were incubated with adenosine (A and C) and inosine (B and C) in the presence or in the absence of the hA_2AR-selective inverse agonist ZM 241385 (A and B) for 10 min. Mean intracellular cAMP levels ± SEM of a representative experiment are shown (n=3). Shaded boxes indicate data with p < 0.001 relative to control. EC₅₀ values with 95% confidence intervals (in parentheses) were 12(4–33) nM and 259.9(104–651) μM for adenosine and inosine respectively.

Table 1.

Summary of EC₅₀ ± SEM for adenosine- and inosine-stimulated cAMP production and ERK1/2 phosphorylation at the hA_2AR and mA_2AR.

Agonist/ Inverse Agonist	EC50 (μM)
	hA2AR		mA2AR
	cAMP Assay	ERK1/2 Phosphorylation Assay	cAMP Assay
Adenosine	0.006 ± 0.001	0.211 ± 0.036	0.059 ± 0.02
Inosine	300.7 ± 48.9	89.38 ± 15.12	709 ± 135.2
Adenosine + ZM 241385 (100 nM)	2.45 ± 1.29	1.17 ± 0.10	7.34 ± 4.56
Inosine + ZM 241385 (100 nM)	> 300	> 300	>1000

The A_2BR also couples to Gαs, hence it stimulates cAMP production upon activation. Therefore, we addressed whether inosine increased cAMP production by CHO-K1 cells stably expressing the human A_2BR (CHO-hA_2BR). Although adenosine increased cAMP production by CHO-hA_2BR cells, inosine did not, even at concentrations equal to and above those that stimulated cAMP production by CHO-hA_2AR bearing cells (Fig. 2C). These results demonstrate that inosine engagement at the A_2AR, but not at the A_2BR, leads to stimulation of adenylate cyclase and a rise in intracellular cAMP levels.

To examine the notion that inosine actions at the A_2AR are indirect and that inosine influences the extracellular concentration of adenosine through equilibrative nucleoside transporter 1 (ENT1) and ENT2 to activate the A_2AR, we utilized saturating concentrations of an ENT1 and ENT2 inhibitor, dipyridamole. Inhibition of ENT1 and ENT2 with dipyridamole slightly reduced the baseline while slightly increasing the maximal cAMP production in response to adenosine (Fig. 3A) indicating that this cell-based assay a) is minimally influenced by endogenously produced adenosine and b) faithfully reports A_2AR activation by the exogenous agonist. Similarly, dipyridamole marginally reduced baseline cAMP levels while enhancing maximum inosine-mediated cAMP production (Fig. 3B) demonstrating that inosine directly and dose-dependently activates the A_2AR.

Fig. 3.

Effect of ENT1/ENT2 inhibition on hA_2AR-mediated intracellular cAMP production.

The ENT1 and ENT2 inhibitor dipyridamole marginally reduces baseline cAMP production and marginally enhances adenosine- and inosine-mediated cAMP production. CHO-hA_2AR cells were incubated with adenosine (A) and inosine (B) in the presence or in the absence of the indicated concentrations of dipyridamole for 10 min. Mean intracellular cAMP levels ± SEM of a representative experiment are shown (n=4). Shaded boxes indicate data with p < 0.001 relative to control.

We also evaluated the effects of metabolites of inosine on A_2AR function. As shown in Fig. 4, inosine monophosphate is approximately 20-fold less potent than inosine in activating the A_2AR. We also observed that xanthine is more potent than inosine monophosphate whereas hypoxanthine is equipotent to inosine monophosphate at the A_2AR (Fig. 4) indicating that inosine metabolites are poor agonists relative to adenosine (EC₅₀=0.006 ± 0.001 μM) and inosine (EC₅₀= 300.7 ± 48.9 μM) at the receptor.

Fig. 4.

Metabolites of inosine are less potent in inducing hA_2AR-mediated intracellular cAMP production than inosine.

CHO-hA_2AR cells were incubated with the indicated concentrations of inosine (Ino), inosine monophosphate (IMP), xanthine (XAN) and hypoxanthine (HPXAN). Mean intracellular cAMP levels ± SEM of a representative experiment are shown (n=3; **, P < 0.01; *** p < 0.001 vs control).

3.3. Inosine dose-dependently increases mouse A_2AR (mA_2AR)-mediated intracellular cAMP production

In mouse models of human disease, the role of the A_2AR in inosine-mediated inhibition of inflammation [12-15] and pain [27] has been convincingly demonstrated utilizing gene ablation and pharmacological tools. As inosine can be readily converted to adenosine in vivo, it is not clear whether the in vivo therapeutic effects of inosine mediated via the A_2AR are direct or indirect in nature. We therefore investigated the ability of inosine to activate the mA_2AR in CHO-K1 cells transiently expressing the receptor to produce cAMP. CHO-K1 cells were chosen as the host cells as they do not have endogenous expression of any adenosine receptor subtype [23]. Both adenosine and the A_2AR -selective agonist CGS 21680 increased the intracellular production of cAMP by CHO-K1 cells transiently expressing mA_2AR, while the A_2AR-selective inverse agonist ZM 241385 inhibited it (Fig. 5A). In contrast, vector-transfected CHO-K1 cells did not yield measurable cAMP levels under any condition. As seen with adenosine, inosine dose-dependently increased cAMP production (Fig. 5A). Moreover, dose response analyses indicated that the EC₅₀ values of adenosine and inosine were 0.059 ± 0.02 μM and 709 ± 135.2 μM, respectively (Fig. 5B and 5C; Table 1). Relative to adenosine, inosine-induced cAMP production appeared to be more sensitive to the A_2AR inverse agonist ZM 241385. Taken together, these results suggest that inosine, although not as potent as adenosine, is a functional agonist at the mA_2AR.

Fig. 5.

Inosine induces mA_2AR-mediated intracellular cAMP production.

(A) CHO-K1 cells transiently transfected with mA_2AR but not with vector alone exhibit agonist-inducible cAMP production. Transiently transfected CHO-K1 cells were incubated with various agonists and an inverse agonist for 10 min. Mean intracellular cAMP levels ± SEM of representative experiments are shown (n=3; *, p < 0.05; **, p < 0.01 vs DMSO). Adenosine (Ado), inosine (Ino), A_2AR-selective agonist CGS 21680 (CGS) and A_2AR-selective inverse agonist ZM 241385 (ZM). Dose response of adenosine (B) and inosine (C) to mA_2AR. CHO-K1 cells transiently transfected with mA_2AR were incubated with various concentrations of adenosine and inosine. Mean cAMP levels ± SEM of a representative experiment in triplicates are shown. Shaded and dotted boxes indicate data with p < 0.001 and p < 0.05 relative to control, respectively. EC₅₀ values with 95% confidence intervals (in parentheses) were 79(21.5–293) nM and 773(249.5–2396) μM for adenosine and inosine respectively.

3.4. Inosine dose-dependently induces hA_2AR-mediated ERK1/2 activation

Agonist engagement at the A_2AR leads to ERK1/2 phosphorylation and subsequent activation [28,29]. Depending on the cell type, the signaling pathways leading to A_2AR-mediated ERK1/2 activation can vary. Gαs-dependent ERK1/2 activation is reported in CHO-K1 cells, whereas Gs-independent ERK1/2 activation is reported in HEK293 cells [28]. Analysis of receptor truncations indicates that Gαs engagement/cAMP production and ERK1/2 phosphorylation/activation are functionally separable [30]. To investigate whether inosine agonism at the A_2AR results in ERK1/2 activation, we evaluated the effects of inosine on ERK1/2 phosphorylation in CHO-hA_2AR cells. As shown in Fig. 6A and Table 1, adenosine stimulated ERK1/2 phosphorylation with an EC₅₀ of 0.211 ± 0.036 μM. Similarly, inosine dose-dependently enhanced ERK1/2 phosphorylation with an EC₅₀ of 89.38 ± 15.12 μM (Fig. 6B; Table 1). Both adenosine- and inosine-induced ERK1/2 phosphorylation were inhibited by the A_2AR inverse agonist ZM 241385 (Fig. 6A & Fig. 6B; Table 1). Comparison of EC₅₀ values of adenosine indicated a 35-fold preference for cAMP production than for ERK1/2 phosphorylation (p < 0.001). In contrast, inosine exhibited a 3.3-fold preference for ERK1/2 phosphorylation over cAMP production (p < 0.005). These results further demonstrate that inosine is an agonist at the A_2AR and that the inosine-engaged receptor has a different pharmacological profile relative to the adenosine-engaged receptor.

Fig. 6.

Both adenosine and inosine dose-dependently induce hA_2AR-mediated ERK1/2 phosphorylation.

Adenosine- (A) and inosine- (B) stimulated ERK1/2 phosphorylation by CHO-hA_2AR cells. CHO-hA_2AR cells were incubated with various concentrations of adenosine or inosine for 20 min. Phospho ERK1/2 levels were quantified using a FRET-based detection kit. Mean FRET ratios ± SEM of representative experiments are shown (n=3). Shaded boxes indicate data with p < 0.001 relative to control. EC₅₀ values with 95% confidence intervals (in parentheses) were 139 (28–680) nM and 41.9 (5.8–303) μM for adenosine and inosine respectively. The A_2AR-selective inverse agonist ZM 241385 (ZM) blocked adenosine- and inosine-stimulated ERK1/2 phosphorylation.

3.5. Inosine dose-dependently stimulates hA_2AR-mediated cAMP production in a cell-free, membrane-based system

In the cell-based A_2AR functional assay, one can conceptualize a scenario where the exogenously added inosine is taken up, converted to adenosine intracellularly and exported back out to engage and activate the receptor. Such a scenario is highly unlikely given the time scale (10 min) of the cell-based assay and maintenance of a cellular response in the presence of the ENT1 and ENT2 inhibitor dipyridamole in intact cell assays. However, to further rule out the possibility, we evaluated the ability of inosine to activate the A_2AR in HEK293-hA_2AR cytosol-free membrane preparations. To this end, we established a sensitive, cell-free, membrane-based assay for the hA_2AR. The A_2AR-selective agonist CGS 21680 increased cAMP production, while the A_2AR-selective inverse agonist ZM 241385 reduced cAMP production to levels that were not detectable by the assay, demonstrating that this assay reports bona fide A_2AR receptor pharmacology (Fig. 7A). Inosine dose-dependently stimulated cAMP production while ZM 241385 inhibited inosine-induced cAMP production in this cell-free, membrane-based assay (Fig. 7B), indicating that the inosine effects at the A_2AR are direct, and thus providing further evidence for inosine agonism at this receptor.

Fig. 7.

Inosine dose-dependently induces hA_2AR-mediated cAMP production in cell-free membrane preparations.

CHO-hA_2AR cell membranes were incubated with indicated concentrations of CGS 21680 (CGS) and ZM 241385 (ZM) (A) and with inosine (B) in the presence and in the absence of the A_2AR-selective inverse agonist ZM 241385 for 30 min. Mean cAMP production ± SEM of representative experiments are shown (n=6; * and ***, p < 0.05 and p < 0.001 vs DMSO respectively; ## and ###, p < 0.001 and p < 0.001 vs inosine respectively).

3.6. Inosine displaces the hA_2AR-selective agonist, CGS 21680, binding to the receptor

In general, agonist-mediated GPCR stimulation stems from agonist binding at the receptor that is detectable by saturation binding experiments. On the basis of our observations of inosine-mediated hA_2AR activation and subsequent generation of cAMP and ERK1/2 responses in the cell-based assays, we hypothesized that inosine may bind to the hA_2AR with high enough affinity to be detectable in an equilibrium binding assay. As there are no commercially available radiolabeled inosine analogs to evaluate direct binding, we determined whether inosine could displace binding of the high affinity hA_2AR agonist, CGS 21680. In equilibrium binding assays, the specific binding of [³H]CGS 21680 to the A_2AR was reduced by approximately 20% regardless of the inosine concentration (F [4, 73] = 7.287; P<0.0001) in the presence of inosine (Fig. 8) supporting the notion that inosine competes with the high affinity adenosine analog CGS 21680 for the same binding site at the A_2AR. The low-magnitude displacement of the [³H]CGS 21680 is consistent with inosine being a low affinity agonist.

Fig. 8.

Displacement of hA_2AR-bound [³H]CGS 21680 by inosine.

Equilibrium binding of [³H]CGS 21680 to hA_2AR in the presence of indicated concentrations of inosine and vehicle control. CHO-hA_2AR membranes were incubated with ADA and [³H]CGS 21680 in the presence and absence of inosine. Values are mean ± SEM of results of 3 independent experiments performed in triplicate; overall p < 0.0001; *, p < 0.05; ** < 0.01; *** < 0.001.

4. Discussion

Inosine is a ubiquitous purine nucleoside that exerts anti-inflammatory and immunomodulatory effects. Inosine and metabolically stable analogs of inosine [31,32] utilize adenosine receptors to modulate anti-inflammatory effects. There are four adenosine receptor (AR) subtypes termed A₁R, A_2AR, A_2BR and A₃R. Among them, A_2AR is the most effective receptor in downregulating inflammation [33,34]. The role of the A_2AR, however, is less clear in the context of inosine-mediated anti-inflammation and immunomodulation. There is a discrepancy between postulating inosine-mediated in vivo anti-inflammatory effects requiring the A_2AR [20,21] when in vitro studies report no functional activation of the A_2AR [22] or displacement of a high-affinity agonist bound to the A_2AR in the presence of inosine [7], Here we examined the possibility of inosine-mediated A_2AR activation in vitro using a variety of highly sensitive and selective experimental approaches not applied previously to the evaluation of inosine receptor activation.

Our investigations produced several lines of experimental evidence supporting inosine as a specific A_2AR agonist. First, in CHO-K1 cells that do not express any endogenous adenosine receptor subtypes [23], inosine dose-dependently increased cAMP production upon exogenous expression of both mouse (mA_2AR) and human A_2ARs (hA_2AR). This inosine-induced cAMP production was inhibited by the A_2AR-selective inverse agonist ZM 241385. Second, inosine-induced cAMP production by CHO-hA_2AR cells was not altered in the presence of the ENT1 and 2 inhibitor dipyridamole. Third, in CHO-hA_2AR cells, inosine exhibited a dose-dependent enhancement of ERK1/2 phosphorylation that was sensitive to ZM 241385. Fourth, consistent with A_2AR activation in the cell-based assays, inosine exhibited a ZM 241385-sensitive induction of cAMP production in cell-free membrane preparations containing the hA_2AR. This further supports the notion that it is inosine, and not a cellular metabolic byproduct of inosine, that is responsible for the stimulation of the A_2AR. Finally, in equilibrium binding assays, inosine inhibited the binding of the A_2AR-selective agonist CGS 21680 to CHO-hA_2AR membranes.

Our data indicate that inosine is a less potent agonist than adenosine at the A_2AR. In cell-based cAMP assays, inosine concentrations that activate the mA_2AR and the hA_2AR are in the high micromolar range, hence generation of accurate EC₅₀ values are not possible. The approximate EC₅₀ values of inosine for cAMP production for the hA_2AR and mA_2AR s are 300.7 ± 48.9 μM and 709 ± 135.2 μM, respectively, indicating that, relative to adenosine, inosine is approximately four orders of magnitude less potent at the A_2AR. In ERK1/2 phosphorylation assays, EC₅₀ values for adenosine and inosine are 0.211 ± 0.036 μM and 89.38 ± 15.12 μM, respectively, a potency difference of 423-fold. Although the EC₅₀ values of inosine appear relatively high, they are well within the physiological levels achieved in disease settings [7,8].

The low potency of inosine relative to adenosine at the A₂aR may be explained by the nature of its interactions with the critical residues that constitute the agonist-binding pocket of the receptor. Crystal structure predictions in combination with site-directed mutagenesis indicate that Asn²⁵³ in the sixth transmembrane domain of the hA_2AR is the critical residue that defines the hydrophobic pocket in which the adenine ring of the agonist rests [35]. In the case of adenosine, two hydrogen bond interactions of Asn²⁵³ with N⁶ and N⁷ of adenine (Fig. 9; [36]) stabilize adenosine binding at the A_2AR. Inosine, a metabolite of adenosine, differs from adenosine by having only N⁷ but not N⁶ in the adenine ring, and thus is capable of having only one hydrogen bond interaction with Asn²⁵³ in the agonist-binding pocket. This may explain at least in part the observed low affinity interaction between inosine and the receptor relative to that of adenosine as evident from the agonist displacement and functional assays.

Fig. 9.

Schematic representation of hydrogen bonding interaction between adenosine/inosine and Asn²⁵³ in the sixth transmembrane domain of the A_2AR agonist-binding pocket.

We have found that adenosine and inosine possess uniquely different pharmacological profiles. The EC₅₀ values for cAMP production and ERK1/2 activation reveal that both adenosine and inosine exhibit an agonist-specific signaling bias. Adenosine is 35-fold more potent in stimulating hA_2AR-mediated cAMP production relative to hA_2AR-mediated ERK1/2 phosphorylation, thus adenosine is a cAMP-biased agonist. In contrast, inosine is 3.3-fold more potent in stimulating hA_2AR-mediated ERK1/2 phosphorylation than hA_2AR-mediated cAMP production, hence it exhibits an ERK1/2 signaling bias. Additionally, inosine-stimulated ERK1/2 phosphorylation is sensitive to the A_2AR reverse agonist ZM 241385 to a greater extent than adenosine-stimulated ERK1/2 phosphorylation. These findings suggest that adenosine and its primary metabolite, inosine, are uniquely different agonists with signaling biases capable of producing overlapping as well as distinct pharmacological outcomes.

Using an equilibrium competitive binding assay, we demonstrate that inosine displaces the binding of the A_2AR-selective agonist CGS 21689 to the hA_2AR. This observation differs, however, from a previous report [7] wherein inosine failed to displace binding of the A₁R/A₃R agonist N6-(3-iodo-4-aminobenzyl)adenosine [37,38] to the rat A_2AR. Differences in species and in affinity of the agonist utilized in these studies may explain in part the differing outcomes.

Adenosine activates both the A_2AR and A_2BR. The A_2AR is the high affinity receptor that is coupled to Gs and Golf and plays a functionally non-redundant role in down modulating inflammation. The A_2BR is the low affinity receptor that is coupled to Gs and Gq, shows Gs-dependent ERK1/2 activation [24] and produces both pro- and anti-inflammatory effects [39], The anti-inflammatory and immunomodulatory roles of both A_2AR and A_2BRs have been largely attributed to Gs-dependent production of cAMP [33,34], Although the significance of A_2AR-mediated ERK1/2 activation in the context of inflammation and immunomodulation is largely unknown, A_2AR-mediated ERK1/2 activation promotes cell death [40], cell proliferation [40] and inhibits osteoclast formation [41]. Our finding that inosine, the primary metabolite of adenosine, is a signaling-biased agonist at the A_2AR but not at the A_2BR suggests a wide variety of functional outcomes that are dependent upon adenosine and inosine levels and relative receptor abundance in vivo.

The major outcome of our investigation is the identification of inosine, the first metabolite of the breakdown of adenosine, as a functional agonist of the A_2AR. The demonstration of inosine as a functional agonist at the A_2AR increases our understanding of A_2AR signaling with respect to the regulation of inflammation and immunomodulation. The current paradigm of A_2AR is based on the notion that the endogenous immunomodulator, adenosine, is the sole natural agonist of the A_2AR. Adenosine, having a half-life <10 s in vivo [1], is rapidly deaminated to inosine, rendering it inactive at the A_2AR [22]. Hence, adenosine-mediated A_2AR signaling may be short-lived. Identifying inosine as an agonist at the A_2AR expands upon this concept and supports the notion that deamination does not render adenosine completely inactive at the A_2AR, but rather produces an agonist, inosine, with a relatively longer half-life (15 h; [2]) and elevated interstitial levels that may prolong A_2AR signaling to sustain anti-inflammatory and immunomodulatory responses. We envision that these two endogenous modulators acting together produce prolonged and accentuated activation of the A_2AR, an effect that neither one can achieve individually. Adenosine engages the A_2AR to generate initial rapid and robust cAMP-biased responses that are subsequently shifted towards ERK1/2 biased responses by inosine. Such dual agonist-mediated signaling may be crucial for mounting a more robust and effective immunomodulatory response in order to limit the extent of tissue damage in inflammatory diseases. Adenosine-driven immunomodulation may be more focused at the disease site where adenosine is produced, released and degraded. In contrast, inosine, being relatively more stable, accumulates, functions locally and due to its stability may diffuse more broadly. Understanding the pharmacological intricacies of interplay between these two natural anti-inflammatory and immunomodulatory agonists and the A_2AR may provide new insights into potential therapeutic interventions of inflammation and immune-mediated diseases. Additionally, quantification of the relative potency and efficacy of inosine and adenosine at the A_2AR will aid our understanding of agonist-receptor interactions at the binding pocket that in turn may lead to the design of agonists with beneficial pharmacological profiles.

Highlights.

• Inosine directly and physically engages the A_2AR.

• Inosine dose-dependently stimulates cAMP production mediated through the A_2AR.

• Inosine dose-dependently induces A_2AR-mediated ERK1/2 phosphorylation.

• Adenosine and inosine differ in their signaling biases at the A_2AR.

Abbreviations

ADA

adenosine deaminase

Ado

adenosine

AR

adenosine receptor

A₁R

adenosine A₁ receptor

A_2AR

adenosine A_2A receptor

A_2BR

adenosine A_2A receptor

A₃R

adenosine A₃ receptor

cAMP

cyclic adenosine monophosphate

CGS 21680

4-[2-[[6-Amino-9-(N-ethyl-β-D-ribofuranuronamidosyl)-9H-purin-2-yl]amino]ethyl] benzenepropanoic acid hydrochloride

DMPX

3,7-Dimethyl-l-propargylxanthine

DMSO

dimethyl sulfoxide

DMR

dynamic mass redistribution

ENT1

equilibrative nucleoside transporter 1

ENT2

equilibrative nucleoside transporter 2

ERK1/2

extracellular signal-regulated kinase-1 and -2

GPCR

G protein-coupled receptor

HBSS

Hanks’ balanced salt solution

HEPES

4-(2-Hydroxyethyl) piperazine-1-ethanesulfonic acid, N-(2-Hydroxyethyl)piperazine-N’-(2-ethanesulfonic acid

HTRF

homogeneous time resolved fluorescence

Ino

inosine

LPS

lipopolysaccharide

NECA

5′-N-Ethylcarboxamidoadenosine

PBS

phosphate buffered saline

SD

standard deviation

SEM

standard error of the mean

TNBS

2,4,6-trinitrobenzenesulfonic acid

ZM 241385

4-(2-[7-Amino-2-(2-furyl)[l,2,4]triazolo[2,3-a][l,3,5]triazin-5-ylamino]ethyl)phenol