A3 Adenosine Receptors as Modulators of Inflammation: From Medicinal Chemistry to Therapy

Abstract

The A₃ adenosine receptor (A₃AR) subtype is a novel, promising therapeutic target for inflammatory diseases, such as rheumatoid arthritis (RA) and psoriasis, as well as liver cancer. A₃AR is coupled to inhibition of adenylyl cyclase and regulation of mitogen-activated protein kinase (MAPK) pathways, leading to modulation of transcription. Furthermore, A₃AR affects functions of almost all immune cells and the proliferation of cancer cells. Numerous A₃AR agonists, partial agonists, antagonists, and allosteric modulators have been reported, and their structure–activity relationships (SARs) have been studied culminating in the development of potent and selective molecules with drug-like characteristics. The efficacy of nucleoside agonists may be suppressed to produce antagonists, by structural modification of the ribose moiety. Diverse classes of heterocycles have been discovered as selective A₃AR blockers, although with large species differences. Thus, as a result of intense basic research efforts, the outlook for development of A₃AR modulators for human therapeutics is encouraging. Two prototypical selective agonists, N6-(3-Iodobenzyl)adenosine-5′-N-methyluronamide (IB-MECA; CF101) and 2-chloro-N6-(3-iodobenzyl)-adenosine-5′-N-methyluronamide (Cl-IB-MECA; CF102), have progressed to advanced clinical trials. They were found safe and well tolerated in all preclinical and human clinical studies and showed promising results, particularly in psoriasis and RA, where the A₃AR is both a promising therapeutic target and a biologically predictive marker, suggesting a personalized medicine approach. Targeting the A₃AR may pave the way for safe and efficacious treatments for patient populations affected by inflammatory diseases, cancer, and other conditions.

1 INTRODUCTION

The relevance of adenosine in the immune system has been established based on mounting scientific evidence showing that the nucleoside represents a paracrine inhibitor of inflammation, regulating the onset, extension, and termination of the inflammatory process and acting through four G protein coupled receptors (GPCRs), designated as A₁, A_2A, A_2B, and A₃ adenosine receptors (ARs).¹ Following inflammation, metabolic alterations occur leading to an increase of extracellular adenosine that is present in the low nanomolar range under physiological conditions, while in stressful conditions it can rise to micromolar levels.² Adenosine in the extracellular milieu is largely formed by hydrolysis/dephosphorylation of ATP, ADP, and AMP through specific ectonucleotidases termed ectonucleoside triphosphate diphosphohydrolase (CD39) and ecto-5′-nucleotidase (CD73).^3,4 Intracellular levels of adenosine are derived from hydrolysis of AMP and S-adenosylhomocysteine (SAH) through cytosolic 5′-nucleotidase, and SAH hydrolase, respectively. Adenosine activity is extinguished through its phosphorylation to AMP by adenosine kinase (AK) or deamination to inosine by adenosine deaminases (ADA1 and ADA2), with ADA present also extracellularly.² The existence of concentrative nucleoside transporters (CNTs) and equilibrative nucleoside transporters (ENTs) regulates the extra- and intracellular adenosine concentrations.⁵

The A₃AR, the last of the four subtypes to be discovered, was cloned sequentially in rat, sheep, and human,^6–8 but it was not shown to respond as an AR from the outset. One of the first activities of this receptor to be reported was the induction of histamine from rat basophilic cells.⁹ The discovery and initial characterization of the A₃AR, and the exploration of its biological paradoxes, has led to the synthesis and biological characterization of a multitude of receptor probe molecules and clinically relevant candidate molecules, including orthosteric agonists and antagonists as well as allosteric enhancers. This discovery of the A₃AR as a fourth AR has spawned current and projected clinical trials of several A₃AR agonists and potentially of a selective A₃AR allosteric enhancer, as well.¹⁰

Using mechanisms triggered by adenosine to inhibit the immune system is a very exciting area of research, and increasing attention is focused on their elucidation in the context of developing new anti-inflammatory strategies. Thus, today the A₃ARsubtype is considered a novel, very promising therapeutic target and predictive biological marker, given its overexpression in inflammatory and cancer cells, compared to low levels found in healthy cells.¹¹

The aim of this review is to summarize the state and the progress of the field of A₃AR modulators and their clinical development in the context of inflammation and cancer and other conditions, with an emphasis on rheumatoid arthritis (RA), psoriasis, and hepatocellular carcinoma.

2 MOLECULAR BIOLOGY OF A₃AR

The A₃AR, the only AR subtype cloned before its pharmacological identification, was initially isolated from rat testis and then from a variety of species. The A₃AR structure had a sequence homology of only 74% in rat versus sheep and human, versus 85% between sheep and human, suggesting significant interspecies differences in ligand recognition. This is manifested in different pharmacological profiles of the species homologs, especially with respect to antagonist binding, which have made the characterization of this AR subtype difficult.¹² There are also species differences in the biological roles of the A₃AR, for example, as the main mechanism for adenosine–induced release of inflammatory mediators in rat mast cells, but not in those of human.¹³

The A₃ARis located on human chromosome 1p21-p13 and consists of a single chain of 318 amino acids.¹⁴ The A₃AR gene presents two exons separated by a single intron of about 2.2 kb.¹⁵ Its promoter region has putative binding sites for multiple transcription factors: The upstream sequence has a CCAAT sequence, as well as consensus binding sites for SP1, NF-IL6, GATA1, and GATA3 transcription factors, the latter of which is important for the A₃AR-dependent role in immune function. Two species of mRNA code for the hA₃AR (sizes 2 and ~5 kb). Variants of the A₃AR have been shown to be associated with coronary heart disease, autism spectrum disorder, and aspirin-induced urticaria.^16–18 Recently, the A₃AR 3′-UTR (untranslated region) of the mRNA was found to be targeted by the proinflammatory microRNA (miR-206) in ulcerative colitis leading to downregulation of A₃AR mRNA/protein expression in colon cells.¹⁹ The A₃AR is expressed in diverse tissues at relatively low levels, compared to A₁AR and A_2AAR. Genomic analysis of the expression of the A₃AR gene in various human tissues (Table 1A) shows highest levels in testes, the spinal cord, and various brain regions, bladder, lung, adipose tissue, and whole blood. The highest expression reached 12.4 RPKM (reads per kilobase of transcript per million mapped reads), while comparable data for A₁AR and A_2AAR and exceeded 20 RPKM at maximal levels in specific tissues. This suggests that potential use of A₃AR ligands in pain and other nervous system disorders is supported by the presence of the receptor in these tissues, although the cell type is not determined in this RNA sequencing data. Various cancer tumors also show major alteration in A₃AR expression in comparison to normal tissue. As accessed from a public cancer database (Table 1B), in 393 unique genomic analyses of cancerous tumors, 25 showed a significant increase in A₃AR (P < 10⁻⁴) and 28 showed a significant decrease compared to normal tissue of the same type. The most prominent increases were in brain cancer (particularly glioblastoma and astrocytoma) and kidney cancer (particularly renal clear cell carcinoma). Thus, the approach of using of A₃AR ligands in a wide range of cancers coincides with significant changes in the receptor expression level in tumors.

TABLE 1.

Expression of the A₃AR RNA in Normal and Cancerous Tissues from Public Databases

(A) Exon expression for the A3AR gene in various postmortem human tissues, from RNA sequencing dataa,b
Tissue	RPKMb
Testes	12.401
Brain (spinal cord, cervical C-1)	5.612
Brain (substantia nigra)	4.268
Adrenal gland	3.884
Spleen	3.495
Small intestine (terminal ileum)	2.778
Brain (amygdala)	2.405
Brain (hypothalamus)	2.201
Nerve (tibial)	2.102
Brain (hippocampus)	1.99
Bladder	1.764
Lung	1.747
Adipose (subcutaneous)	1.73
Whole blood	1.709
Colon (transverse)	1.604
Artery (coronary)	1.517
(B) Alteration in the level of A3AR in cancerous tumors compared to normal tissuec
Tumor	Percentile (no. of analyses)c
Upregulation
Brain and CNS cancerd	1% (4/29)
Kidney cancer	1% (6/20)
Breast cancer	5% (11/43)
Esophageal cancer	10% (1/9)
Downregulation
Bladder cancer	5% (3/10)
Colorectal cancer	5% (12/33)
Sarcoma	5% (1/31)
Brain and CNS cancer	10% (1/29)
Cervical cancer	10% (1/10)
Myeloma	10% (1/8)

^aData from The Genotype-Tissue Expression (GTEx) Project. The data were accessed from the GTEx portal (http://www.gtexportal.org/home/) on February 9, 2017, GTEx Analysis Release V6p (dbGaP Accession phs000424.v6.p1).

^bRPKM stands for reads per kilobase of transcript per million mapped reads for the A₃AR gene (ADORA3, gencode ID ENSG00000121933.13). The highest 16 values are shown from a total of 53 tissues assayed.

^cPercentile refers to the best gene rank percentile for the analyses within the cell. The data were accessed from http://oncomine.org on February 9, 2017, using gene summary visualization for ADORA3. Ratio refers to the number of analyses out of the total number that met the criterion of p < 10⁻⁴ for the change in expression in cancer versus normal tissue.

^dHighly significant upregulation of ADORA3 noted in numerous analyses of glioblastoma and astrocytoma.

As is common to the GPCR superfamily, the A₃AR is characterized by seven transmembrane (TM) domains and an intracellular C-terminal region, with Ser and Thr residues serving as potential phosphorylation sites relevant for rapid receptor desensitization. Following agonist stimulation, the A₃AR undergoes phosphorylation at the C-terminus by GPCR kinases and subsequent internalization through clathrin-coated pits.^20–24 Interestingly, by mutational studies, it has been reported that the highly conserved Trp (W6.48) in TM6 is essential for the active conformation of A₃AR necessary to trigger a series of intracellular pathways for signal transmission, to interact with β-arrestin2, and to undergo receptor internalization.²⁵ Furthermore, use of a novel fluorescent A₃AR agonist has allowed for the observation of colocalization with internalized receptor–arrestin complexes.²⁶

The energetics of A₃AR ligand interactions has been studied using a thermodynamic approach. The thermodynamic parameters of ligand binding at all ARs are similar within either agonist or antagonist classes, which reflects a common ligand receptor interaction mechanism with other ARs This commonality is proposed to explain the difficulty in designing selective adenosine ligands.^27–29

3 DISTRIBUTION IN IMMUNE AND CANCER CELLS

The A₃AR is highly expressed in several immune cell types, as well as in cancer cells.¹¹ In particular, the native human A₃AR was first revealed in human eosinophils and subsequently in neutrophils, monocytes, macrophages, foam cells, dendritic cells, lymphocytes, splenocytes, bone marrow cells, lymphonodes, synoviocytes, chondrocytes, osteoblasts, and mast cells.^9,13,30–63 Overall, the presence of the A₃ARin almost all the cells involved in inflammatory processes suggests their potential involvement in a number of inflammatory pathologies, spanning from wound healing and remodeling to lung injury, inflammatory bone loss, autoimmune, and eye diseases.² In addition, high A₃AR expression has been observed using biochemical methods in many of types of cancer cells, including astrocytoma, melanoma, lymphoma, sarcoma, glioblastoma, colon, liver, pancreas, prostate, thyroid, lung, breast, and renal carcinomas.^64–89 This expression pattern reflects a demonstrated role for this subtype in tumor biology.

4 MEDICINAL CHEMISTRY OF THE A₃ ADENOSINE RECEPTOR

4.1 Adenosine derivatives as agonists of the A₃ adenosine receptor

The first efforts to develop A₃AR selective agonists (Table 2) were performed at the US National Institutes of Health (NIH),⁹⁰ and culminated in the report of N⁶-(3-iodobenzyl)adenosine-5′-N-methyluronamide (IB-MECA 1, Fig. 1), which is ~50-fold selective for the A₃AR in rat compared to the A₁AR and A_2AAR.^91,92 The first few years of medicinal chemical optimization of the affinity and selectivity of A₃AR agonists relied entirely on comparison of binding affinities at the rat ARs,⁵⁶ because the human homologues were not available initially.⁷ A successful approach was to combine multiple A₃AR enhancing substitutions in adenosine analogues. Thus, IB-MECA contained a substituent that enhanced affinity at the ARs including A₃AR, for example, a 5′-N-alkyluronamide, with an N⁶-benzyl substituent that maintained affinity at this subtype but reduced affinity at the A₁AR and A_2AAR. Initially, an unsubstituted N⁶-benzyl group served this purpose, and later a halo atom at the 3-position of the benzyl ring was shown to increase A₃AR affinity and selectivity.⁹² Optimization of the 5′-N-alkyluronamide demonstrated that methyl was more favorable for A₃AR binding than larger alkyl groups. A combination with 4-amino-3-iodo substitution of the N⁶-benzyl group maintained high affinity, but not high selectivity at the A₃AR; thus, compound 3 became a widely used high-affinity radioligand in cells and membranes highly expressing this receptor.⁵⁶ A 3-isothiocyanatobenzyl group was also tolerated at the N⁶-position, which provided the first selective chemically reactive affinity label of the rat A₃AR, termed ICBM(N6-(3-isothiocyanatobenzyl)-5.-N-methylcarboxamidoadenosine) 4.⁹³ In a subsequent study of the structure–activity relationship (SAR), a third position of derivatization was explored: the C2-position.⁹⁴ It was noted that 2-[p-(2-carboxyethyl)phenyl-ethylamino]-5′-N-ethylcarboxamidoadenosine (CGS21680 16⁹⁵), originally introduced as an A_2AAR-selective agonist, surprisingly displayed affinities (nanomolar, all human homologues) in the order A_2AAR (27) > A₃AR (67) > A₁AR (289).¹⁴⁹ Based on this initial observation, it was apparent that the A₃AR binding site was flexible in the ability to accommodate a variety of C2 substituents, including sterically bulky groups. Thus, the order of potency of CGS21680 was A_2AAR > A₃AR > A₁AR ≫ A_2BAR. Nevertheless, the 2-chloro analogue 2 of IB-MECA was the focus at that time as an A₃AR agonist of increased selectivity, since other C2 modifications were not yet systematically explored. Later, an extended C2-alkynyl group, initially in the form of 6-hexynyl, was shown to be tolerated in adenosine derivatives in binding to the A₃AR.^96,97 This modification was also compatible with a 5′-N-ethyluronamide group, an observation that led to the identification of HE-NECA 7 as a potent, but nonselective A₃AR agonist.^98,99

TABLE 2.

Affinity of Selected Nucleoside Derivatives in Binding at Human ARs

Compound		pKi value
		A1AR	A2AAR	A3AR
Agonists
1120	IB-MECA	7.29	5.50	8.74
2120	Cl-IB-MECA	6.66	5.27	8.85
799,103	HE-NECA	7.22	8.19	8.62
899,103	PE-NECA	6.25	6.21	8.21
999,103	PHP-NECA (R,S)	8.57	8.51	9.38
1099,103	PEMADO	4.48	4.38	8.52
11104		4.27	4.98	8.60
13110	CP-608,039	5.14	<4.3	8.24
14a126		<5	<5	7.81
14b126		6.71	5.36	9.42
15a132	LC257	5.79	<4	8.74
15b132		5.42	<5.30	8.70
1695	CGS21680	6.54	7.57	7.17
20105	MRS3558	6.59	5.64	9.54
21119	MRS5151	4.83	~5	8.62
22117	MRS3630	7.74	5.49	8.43
25120,121	MRS5698	<5	<5	8.52
26121	MRS5679	<5	<5	8.51
27121	MRS5980	<5	<5	9.15
29122	MRS5841	<5	<5	8.72
32134	MRS5919	<5	<5	8.22
Antagonists and partial agonists
33136,142		<4	<4	6.19
35128	MRS1292	ND	ND	7.53a
37109		5.80	5.32	7.91
39139		5.60	6.47	8.38
41143		<4	8.14	7.93
42175	MRS3771	5.23	<5	7.54
45125	MRS5776	<5	<5	7.70
46140		<5	5.13	8.31
47141		9.36	7.11	8.52

^apK_i at rat A₃AR = 7.31.

ND, not determined.

FIGURE 1.

Structures of adenosine or 1,3-dialkylxanthine riboside derivatives that act as agonists of the A₃AR. Compound 16 (CGS21680, not shown) is an A_2AAR agonist

Unlike the C2-position, modification of the 2′ and 3′ hydroxyl groups was highly detrimental to A₃ARbinding affinity of simple adenosine analogues,⁹⁶ but a 3′-deoxy analogue of 2 (structure not shown) was later found to be a selective, full agonist at the rat A₃AR with a binding K_i value of 33 nM.¹⁰⁰ The ribose 5′-position was also amenable to modification beyond 4′-CH₂OH and 5′-N-alkyluronamides. For example, 5′-methyl ether analogue NNC53-0055 6 was an agonist at the A₃AR,¹⁰¹ and 5′-alkylthioethers were tolerated at the A₃AR.¹⁰² Acylation of the N⁶-NH reduced affinity at the A₃AR in comparison to the mono-alkylated analogues.⁹⁷ The flexibility of substitution at the N⁶-position compatible with A₃AR affinity was higher than initially indicated in the report on 1. For example, small alkyl and alkoxy groups, such as N⁶-methyl in 10 and N⁶-methyloxy in 6 and 11, could be appended to the nitrogen.^99,101–104 However, a small alkyl group at the N⁶-position often reduced affinity at the rat A₃AR compared to the human homologue. The human A₃AR tolerated larger N⁶ substituents, such as the preferred 1S,2R stereoisomer of N⁶-cyclopropylphenyl in 17.¹⁰⁵ The adenine moiety could be replaced with other heterocyclic nucleobases, leading to the retention of A₃AR affinity and selectivity, but only in limited cases. For example, xanthine-7-ribosides such as DBXRM 5 were shown to fully activate the A₃AR by virtue of an intact 5′-N-methyluronamide.¹⁰⁶ Recently, in silico screening using an A_2AAR crystal structure identified alternative nucleobases that, when ribosylated, retained receptor affinity and efficacy at the A₃AR and other ARs. ¹⁰⁷ Various pyridine-3,5-dicarbonitrile derivatives also bind to and activate ARs as atypical agonists, but they are not selective for the A₃AR.¹⁰⁸

The prototypical A₃AR selective agonists 1 (CF101, Piclidenoson) and 2-chloro-N6-(3-iodobenzyl)-adenosine-5′-N-methyluronamide (Cl-IB-MECA 2, CF102, Namodenoson) are both in clinical trials for inflammation and cancer, respectively.¹⁰ They have demonstrated safety and clinical efficacy in Phases I and II trials. IB-MECA is now about to enter larger Phase III trials for RA and psoriasis, while Cl-IB-MECA is now about to enter a Phase II trial for primary liver cancer. 1 and 2 have also become widely used pharmacological probes with 2 being more selective for the rat and human A₃AR. However, at the mouse A₃AR, an exceptionally high affinity with K_i value of 87 pM was noted for 1, leading to 69-fold and >10,000-fold selectivity in comparison to mouse A₁AR and A_2AAR, respectively.¹⁰⁹ Two other agonists of the A₃AR, 12 and 13, were considered for clinical application to anti-ischemic cardioprotection.¹¹⁰ Compound 13 was unusually water soluble due to the presence of a 3′-amino group, which is largely protonated in physiological medium. Another highly selective A₃AR agonist 15b containing a C2-pyrazolyl group was reported.¹¹¹

In 2000, Jacobson et al. reported that conformationally constraining a ribose-like moiety in the form of bicyclic ring increased the A₃AR selectivity.¹¹² The methanocarba group, that is, a bicyclo[3.1.0]hexane in place of the tetrahydrofuryl group of native ribose, had been applied earlier to antiviral drugs,¹¹³ and this was the first report of its incorporation in signaling ligands. Two isomeric forms of the methanocarba ring system, depending on the fusion point of the cyclopropane and cyclopentane rings, enforce either a North (N) (Fig. 2) or South (S) envelope conformation of the pseudoribose. A priori, the conformational preference of the ARs in binding ribose was not known, but the (N) isomer of simple adenosine was found to have >100-times the affinity of the corresponding (S) isomer at the A₃AR. Thus, this derivatization approach of using chemically constrained rings can be used to probe the conformational preference of a given receptor. Among the four ARs, the greatest affinity gain with the (N)-methanocarba ring occurred at the A₃AR. Numerous A₃AR selective agonists were subsequently reported that included this major modification that enhanced affinity and selectivity at the A₃AR compared to other ARs. The (N)-methanocarba modification is compatible with many other A₃AR affinity-enhancing groups, such as 5′-N-alkyluronamides and N⁶-benzyl groups.¹¹⁴ Incorporation of the (N)-methanocarba modification in IB-MECA 1 resulted in MRS1898 18, a potent and moderately selective A₃AR agonist that was later radiolabeled to provide a selective radioligand for the rodent A₃AR.¹¹⁵ Other halogens could be placed at the 3-position, such as bromo 19 and chloro 20; the 3-bromo equivalent 19 provided a selective A₃AR agonist as a ⁷⁶Br-labeled positron emitter for receptor imaging studies.¹¹⁶

FIGURE 2.

Structures of (N)-methanocarba-adenosine derivatives that act as agonists of the A₃AR

The A₃AR-favoring (N)-methanocarba-5′-N-methyluronamide scaffold could also be combined with an A₁AR-favoring substituent (N⁶-cyclopentyl) to provide a dual acting A₁/A₃AR agonist 22 that displayed cardioproprotective properties, as both receptors demonstrate anti-ischemic properties in heart.¹¹⁷

The C2-position could also be derivatized with (aryl)alkylthio groups, which resulted in moderate A₃AR affinity but not selectivity.¹¹⁸ The C2-position substitution was expanded to include alkynyl and arylalkynyl groups that were compatible with both the riboside and (N)-methanocarba series.^97,103,104 C2-ethynyl and arylethynyl groups were shown to further increase A₃AR selectivity of adenosine analogues, that is, ribosides 7–11 and (N)-methanocarba derivatives such as 23–29.^119–123 A carboxylic acid congener 21 is an A₃AR agonist that displayed greater water solubility and the option of conjugation without losing receptor affinity. A lower homologue of 21 was used as a carboxy-bearing pharmacophore to condense with an amine-functionalized Cy5 fluorophore in the fluorescent A₃AR-selective agonist MRS5218 23 of high A₃AR affinity. 23 is selective for the A₃AR in human and mouse and is suitable for characterization of the receptor in live cells using flow cytometry.¹²⁴ Compound 24 contains a terminal alkyne group, as well as an alkyne at the C2 fusion position, and is suitable for click coupling to carriers such as gold nanoparticles with the retention of A₃AR affinity and selectivity. A 3,4-difluorophenyl ethynyl group in 25 was particularly conducive to attaining high A₃AR affinity in multiple species. Thus, the K_i value of 25 was approximately 3 nM at both the human and mouse A₃ARs. The tolerance of the receptor for extended C2 substituents was surprising—even a biphenyl substituent in 26 preserved high A₃AR affinity, which was explained based on conformational plasticity of TM2.^120,125 A₃AR-selective agonist 29 contains a sulfonate group that renders it unable to diffuse through biological barriers such as the blood–brain barrier. Thus, it is useful in vivo for distinguishing peripheral and central A₃AR effects.¹²² The preferred placement of a sulfonate group on the scaffold of C2-arylethynyl-methanocarba-adenosine-5′-N-methyluronamides was predicted successfully using computational modeling of the receptor interactions at the human and mouse A₃ARs.

In 2003, the group of Lak Shin Jeong in South Korea synthesized thionucleoside analogues that were shown to be highly potent and selective as A₃AR agonists.¹²⁶ The SAR upon modification of thionucleosides at the C2, N⁶ and 5′-positions was explored in detail. The 4′-thio modification of adenosine analogues was found to be compatible with many other A₃AR affinity-enhancing groups, such as 5′-N-alkyluronamides and a range of N⁶ substitutions. Compounds 14a and 14b are analogues of IB-MECA 1 and Cl-IB-MECA 2, respectively, which were found to be highly potent and selective in A₃AR binding. Compound 14b has been shown to suppress angiogenesis, a property that might be beneficial in treating cancer, diabetic retinopathy, and inflammatory diseases.¹²⁷

To summarize the SAR described, A₃AR affinity and selectivity of agonists are based on substitution at the C2, N⁶, and 5′ positions of adenosine,¹²⁸ and only limited ribose functional group substitution of nucleosides is tolerated at this receptor. Some potent A₃AR agonists such as 13 contain a 3′-amino-3′-deoxy modification of adenosine,¹¹⁰ but this modification does not apply universally. Although highly specific A₃AR agonists were obtained in SAR studies, their binding K_i values were not directly predictive of the magnitude of an in vivo protective response, for example, in reducing chronic neuropathic pain. Thus, it became necessary to measure parameters other than simple binding affinities (such as half-life, duration of response, and maximal efficacy in vivo) to select for molecules with translational potential. An in vivo phenotypic screen in real time of the action of A₃AR agonists to reduce or prevent chronic neuropathic pain was adopted for a comparison of diverse substitutions at these positions on the adenosine scaffold.¹²¹ The data obtained in a mouse model of neuropathic pain, that is, chronic constriction injury (CCI) of Bennett and Xie,¹²⁹ allowed the chemists to steer the SAR in the direction of compounds that displayed high efficacy in reducing hyperalgesia and a long duration of action in vivo upon oral administration. The CCI model was ideally suited for the comparison of antinociceptive activity of A₃AR agonists because of their high potency (greater than the molar potency of other pain medications) and because they lacked activity in tests of acute pain, such as the hot plate test and tail flick assay.¹³⁰ The efficacy and duration of action of novel A₃AR agonists after p.o.per os administration indirectly indicated favorable oral bioavailability and pharmacokinetics, at least with respect to chronic neuropathic pain. Thus, this phenotypic screen proved to be an invaluable guide in the extension of the SAR in this compound series.

This in vivo phenotypic screen confirmed that C2-phenylalkynyl analogues were among the preferred A₃AR agonists. The terminal cyclic group in the C2-alkyne series was then varied to include diverse 6-membered rings, 5-membered heterocyclic rings, and cycloalkyl rings.¹²¹ With respect to A₃AR binding and selectivity, many of these groups, including substituted phenyl rings, maintained high A₃AR affinity and selectivity. However, the in vivo phenotypic screen identified 5-membered heterocyclic rings, such as thienyl derivatives, as being particularly potent and efficacious in vivo in the chronic neuropathic pain model. A 5-chlorothienylethynyl group in MRS5980 27 and MRS7154 28 was found to prolong the protective action of A₃AR agonists in the CCI model. The substitution of the N1 group of the adenine moiety with CH in 31 was well tolerated in A₃AR binding and activation and in the CCI model.

Due to the possibility that the C2-arylethynyl group could serve as a Michael reaction acceptor in nucleophilic attacks, alternative bioisosteric extensions at the C2-position were compared. The aryltriazolyl group in MRS7138 30 was found to mimic the geometry of the corresponding arylethynyl group when the (N)-methanocarba nucleoside was receptor bound,¹³¹ and this substituent would not have liability as a potential Michael acceptor. This conformational relationship was predicted using molecular docking to a homology model of the A₃AR. C2-triazole substitution (two positional isomers) was previously found to be compatible with A₃AR binding in the riboside series, as in 15a and 15b.¹³² As a postscript to that effort to replace the C2-arylethynyl group, this ethynyl group was found to be relatively unreactive toward thiols such as glutathione, and the risk of such compounds depleting liver glutathione was shown to be very small.¹³³

The surprising finding that substitution of the exocylic NH of adenine with H or CH₃ in MRS5919 32 allowed full activation of the A₃AR emphasized that the loss of otherwise important recognition elements in a ligand can be compensated by other affinity enhancing moieties on the nucleoside.¹³⁴ Moreover, the ribose moiety is the main effector of receptor activation, while adenine modifications tend to change the subtype selectivity but usually do not have a major effect on the agonist efficacy. However, there are exceptions to the above generalization, including various N⁶ substituents and C2 substituents that produce partial agonism or antagonism at the A₃AR, as has been summarized,¹³⁵ and nucleoside derivatives with reduced efficacy are discussed below.

4.2 Nucleoside derivatives as partial agonists and antagonists of the A₃ adenosine receptor

Selected nucleoside derivatives that act as antagonists or low efficacy agonists at the A₃AR are shown in Figure 3. The truncation of hydroxyl groups of adenosine nucleosides, that were demonstrated to be A₃AR-selective agonists, was first explored in 1995.^96–99 The goal was to reduce the hydrophilicity of the nucleosides to increase bioavailability without loss of receptor affinity. A secondary goal was to probe the effect on intrinsic efficacy of the truncated nucleosides as A₃AR agonists, although it was not immediately achieved.¹⁰⁰ The conversion of adenosine agonists into antagonists by complete removal of the ribose ring, that is, in adenine derivatives, was previously demonstrated, but the pharmacological characteristics of intermediate structures, that is, those with partially truncated ribose moieties, were unknown. Among the four ARs, the A₃AR appears to be the easiest with respect to conversion of nucleoside agonists into antagonists, and numerous examples have been reported.^{109,128,132,135–141} However, it should be noted that the degree of efficacy can vary, depending on the functional assay and the receptor expression level. Thus, a modified nucleoside that behaves as an A₃AR antagonist in one system, such as binding of a radiolabeled guanine nucleotide, might still activate the receptor under different circumstances, such as measurement of inhibition of adenylate cyclase.¹¹⁵

FIGURE 3.

Structures of (N)-methanocarba and riboside derivatives of adenosine or 1,3-dialkylxanthine that act as partial agonists or antagonists of the A₃AR

Cristalli and co-workers took another approach to achieve A₃AR antagonism. The presence of 8-alkynyl substituents on adenosine (4′-CH₂OH) analogues, such as 33, reduced the ability of the A₃AR-selective nucleoside to activate the receptor, as is consistent with antagonism.^136,142 The theme of reduced efficacy in truncated nucleosides, rigid nucleosides, and otherwise modified ribosides was developed in pharmacological studies of Gao et al.¹³⁷ The requirement of an H-bond donating group at the 5′-position of nucleoside analogues for A₃AR activation was also demonstrated. A spirolactam 35 related structurally to IB-MECA 1 retained selectivity of binding to the rat and human A₃ARs, but completely lacked the ability to activate the receptors and was shown to be a functional antagonist. Thus, a degree of flexibility of the 5′-amide, which is capable of forming multiple H bonds with the receptor, was required for full A₃AR activation. Xanthine-7-riboside 34 was a partial agonist at the rat A₃AR but an antagonist at the A₁AR.¹⁰⁶ An N⁶ substituent that converted A₃AR agonist activity into antagonism was the N⁶-(2,2-diphenyl)ethyl group in 36.¹⁰⁵ Curiously, the corresponding rigidified N⁶-fluorenylmethyl analogue (structure not shown), upon addition to an aryl-aryl bind to 36, became a full agonist. The combination of a N⁶-benzyl-type substituent with 2-chloro in 38 reduced the efficacy at the A₃AR to nearly zero, although its residual efficacy could be expanded to full efficacy in the presence of A₃AR PAM LUF6000 95 (where PAM is positive allosteric modulator).¹⁰⁸ Modification at the 5′-position as an ester 37 produced a low-efficacy partial agonist, while 5′-N,N-dialkyluronamide in 42 resulted in antagonism at the A₃AR.¹⁰⁹ Thus, conformational factors at various regions surrounding the adenosine core and H-bonding around the 5′-position are determinants of A₃AR efficacy.¹²⁸

The 4′-truncation of the A₃AR nucleosides was explored in great detail for the 4′-thionucleosides ^138,139 leading to compounds 39 and 40, which were shown to be antagonists using a functional assay of guanine nucleotide binding. However, the 2-hexynyl group of compound 41 added a second activity to this series of A₃AR ligands, that is, 41 was a combined potent A₃AR antagonist and A_2AAR agonist.¹⁴³ Truncation of the nucleoside ribose-like moiety in the (N)-methanocarba series also led to A₃AR-selective antagonists and partial agonists,¹⁰⁹ including a radiolabeled 3-bromo analogue 44 for positron emission tomography (PET).¹¹⁶ Compound 45 is a selective antagonist of both the human and mouse A₃ARs.^120,125 Compound 46 is a selective A₃AR antagonist with renal protective properties.¹³⁹ Compound 47 is a mixed A₁/A₃AR antagonist that also displays functional agonism at the A_2AAR, which displayed greater potency than predicted from its only moderate affinity in A_2AAR binding assays.¹⁴¹

4.3 Non-nucleoside derivatives as antagonists of the A₃ adenosine receptor

For more than 20 years, the advance of potent and selective A₃AR antagonists as promising therapeutic choices for a range of diseases has been a prime subject of medicinal chemistry research. The pharmaceutical industry and academic communities have focused on the synthesis and screening evaluation of numerous heterocyclic compounds to discover potent and highly selective A₃AR antagonists due to their potential therapeutic applications.¹⁴⁴

A₃AR antagonists belong to different structural groups including monocyclic, bicyclic, and tricyclic aromatic compounds (Table 3). Several xanthine or purine analogues were examined first, but none showed significant affinity or selectivity at rat A₃AR.^{96,100,144,145} Consequently, different classes of compounds, that could be classified as nonxanthine derivatives^{135,144–149} and the lately nucleoside-derived antagonists, have been discovered as highly potent and selective A₃AR antagonists.

TABLE 3.

Affinity of Selected A₃AR Antagonists

	Compound	pKi value or % inhibition at 10 μM
		A1AR	A2AAR	A3AR
Monocyclic systems
48155	MRS1334	5.54 (r)	<10%(r)	5.41 (r)
				8.57 (h)
49158	MRS1505	4.38 (r)	4.62 (r)	6.09 (r)
				8.10 (h)
50162		17% (h)	43%(h)	8.46 (h)
51164	ISVY130	1% (h)	10% (h)	8.44 (h)
52165	SYJA385	7% (h)	10%(h)	6.41 (h)
53167		24% (h)	28% (h)	9.10 (h)
54146		<5 (h)	<5 (h)	9.39 (h)
55146		<6.18 (h)	<6.08 (h)	9.44 (h)
				8.80 (r)
Bicyclic systems
56169	VUF5574	52% (r)	43%(r)	8.39 (h)
57170		4% (h)	1% (h)	7.71 (h)
58171		0% (h)	19%(h)	9.11 (h)
59135		5.10 (h)	6.08 (h)	7.59 (h)
60173		6% (h)	8%(h)	7.60 (h)
61174		6.37 (h)	5.09 (h)	8.22 (h)
62175	MRS3777	26% (h)	16%(h)	7.33 (h)
63176		5.98 (h)	5.50 (h)	9.74 (h)
64176		<5 (h)	<5 (h)	8.54 (h)
65177		5% (h)	1% (h)	8.92 (h)
66179		1% (h)	1%(h)	10.57 (h)
Tricyclic systems
67180	CGS15943	8.46 (h)	9.40 (h)	7.02 (h)
68182	MRS1220	7.28 (r)	8.00 (r)	9.19 (h)
69183		<5 (h)	<5 (h)	8.09 (h)
70184		<6 (h)	6.98 (h)	8.94 (h)
71184		<6 (h)	<6 (h)	8.16 (h)
72185		42% (b)	3% (b)	8.68 (h)
73186		<6 (h)	<6 (h)	8.05 (h)
74188		25% (b)	14% (b)	8.33 (h)
		0% (h)
75189		6.59 (b)	0% (b)	9.10 (h)
		7.96 (h)	2% (h)
76190		0% (b)	8.06 (b)	8.68 (h)
77192		5.57 (h)	<5 (h)	8.80 (h)
78193	MRE3008-F20	<5 (r)	5.70 (r)	9.54 (h)
79203	MRE3005-F20	6.60 (h)	7.22 (h)	10.40 (h)
80149		5.47 (h)	<5.3	8.01 (h)
81204		ND	ND	79% (h)
82204		ND	ND	16% (h)
83196		<6 (h)	<6 (h)	7.74 (h)
84143		32% (h)	49% (h)	9.29 (h)
85143	OT-7999	4% (h)	31%(h)	9.02 (h)
86207	PSB-10	5.77 (h)	5.56 (h)	9.36 (h)
87209	KF-26777	5.74 (h)	6.33 (h)	9.70 (h)
88211		24% (h)	0% (h)	8.66 (h)
89200		<6 (h)	<6 (h)	9.10 (h)
90200		<6 (h)	<6 (h)	8.46 (h)
91195		5.60 (h)	<5.3 (h)	8.84 (h)
92195		5.52 (h)	5.82 (h)	8.71 (h)

h, human; r, rat; and b, bovin.

ND = not determined.

In this section, we summarize the medicinal chemistry of A₃AR antagonists updating our previous reviews on this field.^150–153

4.3.1 Monocycles

1,4-Dihydropyridines and pyridines

After the first evidence that 1,4-dihydropyridines (DHPs) exerted antagonistic activity at the A₃AR in addition to L-type calcium channel inhibition, Jacobson et al. designed a series of substituted DHPs in the attempt to separate the two different activities.¹⁵⁴ In this study, the replacement of the methyl ester at the 5-position of nifedipine with a bulkier 4-nitrobenzyl ester, along with the introduction of phenylethynyl and phenyl moieties at positions 4 and 6, respectively, led to 48 (MRS1334, Fig. 4). This compound showed high affinity and selectivity as an A₃AR antagonist without inhibiting L-type calcium channels.¹⁵⁵ In addition, a 3,5-diacyl-2,4-dialkylpyridine series was delineated by the oxidation of the corresponding DHP derivatives, and the best profile against A₃AR was achieved with 49 (MRS1505, Fig. 4) in which the position 4 of the pyridine ring was substituted with small alkyl groups such as ethyl chain.^156,157 General SAR of pyridine derivatives revealed that structural requirements responsible for enhancement of A₃AR affinity and selectivity did not completely reflect that of the DHP parent compounds.¹⁵⁸ Among this series, were also reported fluorinated and hydroxylated pyridine derivatives¹⁵⁹ and an extension of this study performed by Jacobson and co-workers described a series of N-alkylpyridinium salts as water soluble A₃AR antagonists although with lower potency than the pyridine analogues.¹⁶⁰ A pyridine-based A₃AR antagonist PET ligand [¹⁸F]FE@SUPPY was introduced.¹⁶¹

FIGURE 4.

Monocycle-based A₃AR antagonists

Pyrimidines

Within the classes of bi- and tricyclic ARs antagonists, the pyrimidine nucleus present in the endogenous modulator adenosine, is a frequent substructural scaffold. 4-Amino-6-hydroxy-2 mercaptopyrimidines derived from chain opening of a series of triazolopyrimidinones have been synthesized by Cosimelli et al.¹⁶² Introduction of the propylsulfanyl and p-chlorobenzyloxy moieties at 2 and 6 positions, respectively, combined with an acetamide group at 4-position of the pyrimidine ring led to compound 50 (Fig. 4), a potent and selective human A₃R antagonist (K_i = 3.5 nM). Similar structures characterized by two regioisomeric series of diaryl 2- or 4-amidopyrimidines such as N-[2,6-bis(4-methoxyphenyl)pyrimidin-4-yl]acetamide 51 (ISVY130, Fig. 4) were reported by Sotelo and co-workers. Some of the ligands in this series exhibited good selectivity and affinity with K_i values of < 10 nM at the A₃AR.^163,164

Pyrazin-2(1H)-ones

Very recently, Sotelo and co-workers published a novel series of compounds, including a simplified pyrazin-2(1H)one scaffold, as A₃AR antagonists and with better pharmacokinetic properties.¹⁶⁵ These new derivatives obtained by the Ugi-based multicomponent reaction were less potent than many other A₃AR antagonists reported in the literature. The entire library of compounds, including the most potent compound of this series with a K_i value of 386 nM (52, SYJA385, Fig. 4), was subjected to a computational study to determine a rational hypothesis for their binding model.¹⁶⁵

Thiadiazole and thiazoles

IJzerman co-workers identified thiadiazole and thiazole analogues as A₃AR antagonists by chemical structure simplification of corresponding bicyclic quinazoline and isoquinoline nuclei, respectively.¹⁶⁶ Among them, compound N-[3-(4-methoxyphenyl)-[1,2,4]thiadiazol-5-yl]acetamide 53 (Fig. 4) was claimed as the best compound of the series exhibiting a K_i value of 0.79 nM and acting as antagonist in a cyclic AMP (cAMP) functional assay.¹⁶⁷ SAR optimization by introduction of a 5-(pyridine)-4-yl moiety on the 2-aminothiazole ring revealed a series of potent and selective compounds such as N-[4-(3,4,5-trimethoxyphenyl)-5-(pyridin-4-yl)thiazol-2-yl]-acetamide 54, with subnanomolar affinity for human A₃AR (K_i = 0.4 nM) and 1000-fold selectivity against the other AR subtypes.¹⁶⁸

The QSAR analysis of thiazole and thiadiazole A₃AR antagonists indicated that their binding affinity increased with decreasing lipophilicity and in the presence of small alkyl moieties such as amide functions (acetamide or propionamide).¹⁶⁸ In addition, the introduction of substituents, such as benzoyl, nicotinoyl (e.g., compound 55, Fig. 4) and isonicotinoyl moieties in position 2 of the thiazole ring, led to potent and selective antagonists at both human and rat A₃ARs.

4.3.2 Bicycles

Quinazolines, phthalazines, and quinoxalines

A structure–affinity study reported by IJzerman co-workers indicated that introduction of a phenyl or heteroaryl substituent on the 2-position of the quinazoline scaffold or the equivalent 3-position of the isoquinoline improved the A₃AR affinity in comparison to the unsubstituted derivatives. Combinations of the best substituents in the two series led to the potent and selective human A₃AR antagonist N-(2-methoxyphenyl)-N-(2-(pyridin-3-yl)quinazolin-4-yl)urea 56 (VUF5574; Fig. 5) with a K_i value of 4 nM. ¹⁶⁹

FIGURE 5.

Bicycle-based A₃AR antagonists

Subsequently, the 2-amino/2-oxoquinazoline-4-carboxamide compounds, resulting from an in silico molecular simplification approach of the 2-aryl-1,2,4-triazolo[4,3-a]quinoxalin-1-one skeleton, were published by Morizzo et al. as

A₃AR antagonists. One example of this series is compound 57 (Fig. 5) that showed good affinity and selectivity at the A₃AR.¹⁷⁰ With a similar approach on the triazoloquinazolinone nucleus, a new series of 2-phenylphthalazin-ones was identified as promising A₃AR antagonists. Molecular manipulations by introduction of amide and ureide functional groups at the 4-position of the phthalazinone ring led to compound 58 (Fig. 5) with the best activity profile.¹⁷¹ Additionally, the 2-(4-methyl-1H-benzo[d]imidazol-2-yl)-quinoxaline 59 (Fig. 5) is noteworthy for the novelty of its design strategy utilizing a 3D database searching approach.¹⁷²

Imidazo[1,2-a]pyrazines

Very recently, the imidazo[1,2-a]pyrazine nucleus was reported as a suitable core for the design of new AR antagonists. Within this series of compounds, a N-(2,6-diphenylimidazo[1,2-a]pyrazin-8-yl)-4-methoxybenzamide 60 (Fig. 5) showed good A₃AR affinity with a K_i value of 25 nM. The molecular docking study of these compounds was also carried out to describe the potential binding mode of the new derivatives to their refined target receptor model.¹⁷³

Adenines and adenine-like derivatives

The first class of selective A₃ARantagonists characterized by a bicyclic structure strictly correlated to the adenine core was identified by Biagi et al.¹⁷⁴ The adenine-like structure of these new N⁶-ureidosubstituted-2-phenyl-9-benzyl-8-azaadenines was responsible for their activity as antagonists, while the phenylcarbamoyl group ensured selectivity at the A₃AR. The SAR studies based on the systematic substitutions of 2, 6, and 9 positions of the bicyclic system led to improved A₁/A₃ selectivity with compound 61 (Fig. 5).

Starting from reversine, 2-(4-morpholinoanilino)-N⁶-cyclohexyladenine, with a moderate activity as A₃AR antagonists (K_i = 0.66 μM), Perreira et al. explored the SAR of related derivatives in order to improve A₃AR potency and selectivity.¹⁷⁵ A series of reversine analogues was synthesized by substitution of 2- and N⁶-positions of the adenine core. One of the most remarkable compounds in terms of hA₃AR affinity and selectivity resulted when the N⁶-cyclohexyl moiety of reversine was combined with a 2-phenyloxy group (compound 62, MRS3777, Fig. 5).¹⁷⁵ Some analogues from this study were shown to be inactive at 10 μM in the rat, reflecting the typical species dependence of binding of most known nonnucleoside A₃AR antagonists.

The pyrazolo[3,4-d]pyrimidine scaffold, a bicyclic system structurally connected to the adenine nucleus that resulted from simplification of tricycles pirazolo-triazolo-pyrimidine, was recently described by Taliani et al.¹⁷⁶ The SAR profile of this series indicated that the presence of an amide or ureide functionality at the 4-position (compounds 63 and 64, respectively) along with a phenyl ring at the 6-position was essential for promoting A₃AR affinity and selectivity. The N²-position was characterized by substantial steric tolerance; in fact, both small methyl group (63, Fig. 5) and bulkier benzyl moiety (64, Fig. 5) were well tolerated. Compound 63, which showed a subnanomolar A₃AR affinity and high selectivity versus the other AR subtypes, has been suggested as a promising lead compound for the development of adjuvant agents in glioma chemotherapy. In a related effort, a series of junction isomers of pyrazolo[3,4-d]pyrimidine derivatives were synthesized by Lenzi et al. applying a molecular simplification approach to the tricyclic pyrazolo[3,4-c]quinolin-4-one skeleton.¹⁷⁷ The binding results of the junction isomers were successful and the new derivatives maintained high affinity for the hA₃AR increasing also the selectivity versus the other AR subtypes. Aryl/arylalkyl substitution at the 5-position of such derivatives was poorly tolerated for A₃AR binding affinity, while small groups at the same position were shown to enhance the ligand–receptor interaction. In addition, the substitution of the 2-phenyl ring with a 4-methoxy group led to 2-(4-methoxyphenyl)-5-methyl-2H-pyrazolo[4,3-d]pyrimidin-7(6H)-one 65 (Fig. 5), the most potent compound of this series. Very recently, a large number of 2-arylpyrazolo[4,3-d]pyrimidin-7-amine or 7-acylamine derivatives were synthesized as potent A₃AR antagonists.^178,179 The pyrazolopyrimidines bearing a 4-methoxyphenyl or a 2-thienyl group at the 5-position showed high hA₃AR affinity and selectivity. 4-Methoxy-N-(2-phenyl-5-(thiophen-2-yl)-2H-pyrazolo[4,3-d]pyrimidin-7-yl)benzamide 66 (K_i = 0.027 nM, Fig. 5) is one of the most potent and selective A₃AR antagonists in this structural class.¹⁷⁹

4.3.3 Tricycles

Triazoloquinazoline

Jacobson and co-workers first described the N⁵-acylation of the free amino group of well-known 9-chloro-2-(furan-2-yl)-[1,2,4]triazolo[1,5-c]quinazoline-5-amine 67 (CGS15943, Fig. 6).^180,181 This structural modification yielded 68 (MRS1220, Fig. 6), which was enhanced in both affinity and A₃AR selectivity.^181,182 The removal of the chlorine atom at 9-position of the triazoloquinazoline 67 along with the replacement of the 5-phenylacetamido and the 2-furyl moieties with a linear alkyl chain and a 4-Br-phenyl group, respectively, led to 69 (Fig. 6). This compound was found to be a potent and selective A₃AR antagonist.¹⁸³

FIGURE 6.

Tricycle-based A₃AR antagonists

Very recently, a new series of triazoloquinazolines was reported as A₃AR antagonists. Two examples are the 3,5-diphenyl[1,2,4]triazolo[4,3-c]quinazoline 70 (K_i =1.16 nM, Fig. 6) and the 5′-phenyl-1,2-dihydro-3′H-spiro[indole-3,2′-[1,2,4]triazolo[1,5-c]quinazolin]-2-one 71 (K_i = 6.94 nM, Fig. 6).¹⁸⁴

Pyrazolo[3,4-c]/[4,3-c]quinolines

2-Arylpyrazolo[3,4-c]quinolin-4-ones, 4-amines, and 4-amino-substituted derivatives are reported as potent and selective hA₃AR antagonists.¹⁸⁵ Most of them showed a nanomolar hA₃AR affinity and different degrees of selectivity that were strictly dependent on the presence and nature of the substituent on the 4-amino group. The benzoamide derivative 72 (Fig. 6) was the most potent and selective among the three reported series of compounds.

The pyrazole[4,3-c]quinoline-4-one scaffold was adapted to A₃AR antagonists as structural isomers of the previous nucleus by Baraldi et al. Among the pyrazole[4,3-c]quinoline-4-ones, compounds that contained a 2-p-substituted phenyl (CH₃, OCH₃, and Cl) group showed good hA₃AR affinity and excellent selectivity in comparison to the other AR subtypes (e.g., compound 73, Fig. 6).¹⁸⁶

Triazolo[4,3-a]/[1,5-a]quinoxaline

Colotta et al. identified 1,2,4-triazolo[4,3-a]quinoxalin-1-one derivatives as promising hA₃AR antagonists.¹⁸⁷ The SAR study recognized the appropriate substitutions at 2, 4, and 6 positions of the tricyclic template. In particular, the introduction of the 4-oxo or 4-N-amide functions afforded selective and potent A₃AR antagonists such as compounds 74¹⁸⁸ and 75¹⁸⁹, respectively, confirming the importance of both nuclear or extranuclear carbonyl functionality for A₃AR affinity (Fig. 6). A series of 2-aryl-8-chloro-1,2,4-triazolo[1,5-a]quinoxalines has also been synthesized and evaluated in radioligand binding assay at both bovine and human ARs^190,191 showing a similar SAR profile to that of the 2-arylpyrazolo[3,4/4,3-c]quinolines^{185, 186} and triazolo[4,3-a]quinoxaline.¹⁸⁹ One of the representative derivatives was a 2-(4-methoxyphenyl)-[1,2,4]triazolo[1,5-a]quinoxalin-4(5H)-one 76 (Fig. 6).

1,2,3-Triazolo[1,2-a][1,2,4]benzotriazolones

A₃AR antagonists based on aminophenyl-triazolobenzotriazinone have been reported by Da Settimo et al. A lead optimization strategy focused on the structural modifications provided that the suitable groups were attached to the 5-amino group and in the 4′-and/or 9-positions. The best result was obtained with compound 77 (Fig. 6), which showed a K_i value of 1.6 nM at the A₃AR and no significant affinity at the other ARs.¹⁹²

Pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidines

The pyrazolo-triazolo-pyrimidine (PTP) scaffold, due to its strong structural correlation with the nonselective antagonists CGS15943 (67, Fig. 6) ¹⁸⁰ and to the adenine nucleus present in the endogenous modulator adenosine, has been investigated in depth as a prototypical template for adenosine antagonists.

Rigorous research efforts were made on this scaffold in order to obtain potent A_2A- and A₃AR antagonists.¹⁵³ A series of PTP derivatives (MRE series) reported by Baraldi’s group were obtained by the structure–activity optimization based on the introduction of different substituents at the 5, 7, 8, and 9 positions.^{145,149,193,194,196–200} The N⁷-substituted derivatives proved to be predominantly hA_2AAR antagonists, while the combination of a small alkyl chain at the N⁸-pyrazole position with a (substituted)phenylcarbamoyl chain at the N⁵-position led to potent and selective hA₃AR antagonists. ¹⁹⁸ One of the most active and selective compounds in this series was 78 (MRE3008-F20, Fig. 7) with a K_i value of 0.29 nM.¹⁹³ The corresponding tritium-labeled analogue as the first potent and selective radiolabeled antagonist for the A₃AR was prepared. It bound to the hA₃AR expressed in Chinese hamster ovary (CHO) cells with a K_D value of 0.82 nM (B_max = 297 fmol/mg protein).^201,202 The isosteric replacement of the phenyl ring with a 4-pyridylmoiety yielded 79 (MRE3005-F20, Fig. 7) that maintained the high affinity at the A₃AR with enhanced water solubility.²⁰³ Subsequently, replacement of pyridin-4-ylmoiety of MRE3005-F20 with substituted piperidine rings led to the preparation of the hydrochloride salt of 1-(cyclohexylmethyl)piperidin-4-yl (80, Fig. 7).¹⁴⁹

FIGURE 7.

Tricycle-based A₃AR antagonists

Synthesis of fluorosulfonyl- and bis(β-chloroethyl)amino-phenylamino-pyrazolo[4,3-e]1,2,4-triazolo[1,5-c]pyrimidines as irreversible A₃AR antagonists was performed by Baraldi’s group to provide useful tools for structure–activity studies. Electrophilic groups, specifically sulfonyl fluoride and nitrogen mustard (bis-(β-chloroethyl)amino) moieties, have been incorporated at the 4-position of the aryl urea group (compounds 81 and 82, respectively, Fig. 7).²⁰⁴ While compounds containing a fluorosulfonyl moiety proved to be irreversible antagonists at the hA₃AR (at 100 nM, 79% of inhibition), the corresponding nitrogen mustard derivatives were unable to covalently bind the target receptor subtype. This difference in the receptor interaction between the 81 and 82 series has been explained on the basis of chemical reactivity of the two different groups.

Very recently a new series of N,N′-disubstituted guanidines of the PTP scaffold, prepared in a one-pot reaction, was described as A₃AR antagonists. The best compound of this series was 83 (Fig. 7) bearing a N,N-(4-nitrophenyl)guanidine moiety at 5-position of the tricyclic nuleus.¹⁹⁶

Triazolopurines

[1,2,4]Triazolo[5,1-i]purines structurally related to the triazoloquinazolines have been identified as A₃AR antagonists.^183,205 Within this family, the 5-n-butyl-8-(4-n-propoxyphenyl)-3H-[1,2,4]triazolo[5,1-i]purine 84 (Fig. 7) showed a good potency and selectivity at A₃AR. Among the reported structures, compound 85 (OT-7999, Fig. 7) demonstrated a significant reduction of intraocular pressure in cynomolgus monkeys at 2–4 hr following topical application.²⁰⁵

Tricyclic xanthines

Caffeine and theophylline are the classical nonselective xanthine antagonists of ARs that display micromolar affinity at human AR subtypes. At the rat A₃AR, caffeine and theophylline are weaker in affinity.

Initial SAR studies at the A₃AR were carried out using multiple substituted xanthines, many of which retained selectivity for the A₃AR.⁹⁰ An interesting approach was based on the ring annelation of xanthine derivatives, which permitted several research groups to discover different tricyclic systems that showed dissimilar affinity to AR subtypes.²⁰⁶ A series of imidazo[2,1-i]purinones as tricyclic xanthines (PSB series) was developed by Müller et al. as human A₃AR antagonists with improved water solubility.²⁰⁷ For example, 86 (PSB-10, Fig. 7) showed a subnanomolar A₃AR affinity and a good selectivity compared to the other AR subtypes, and its tritiated form ([³H]PSB-11) exhibited a K_D value of 4.9 nM (B_max = 3500 fmol/mg of protein).²⁰⁸ Another similar compound is 2-(4-bromophenyl)-4-propyl-7,8-dihydro-1H-imidazo[2,1-i]purin-5(4H)-one 87, also designated KF-26777 (Fig. 7), was endowed with high affinity and selectivity.²⁰⁹

The pyrido[2,1-f]purine-2,4-diones, another series of tricyclic xanthines, have been reported to exert affinity at A₃AR in the low nanomolar range. Different substituents at the 1 and 8 positions of the new scaffold were evaluated, and the SAR studies led to 3-(cyclopropylmethyl)-1-(4-methylbenzyl)pyrido[2,1-f]purine-2,4(1H,3H)-dione 88 (Fig. 7) showing the best A₃AR binding profile of the series with total selectivity.^210,211

Replacement of the pyridine nucleus of the pyrido[2,1-f]purine-2,4-dione scaffold with different 5-membered heterocycles has been extensively examined by Baraldi’s group. The SAR studies led to both series of pyrrolo[2,1-f]purine-2,4-dione and imidazo[2,1-f]purine-2,4-diones as A₃AR antagonists.¹⁹⁹ Among the examined molecules, the imidazo[2,1-f]purine-ones were 2- to 10-fold more potent than the corresponding pyrrolo[2,1-f]purine-ones (e.g., 89 vs. 90, respectively, Fig. 7) and the most favorable affinity and selectivity at A₃AR was obtained by introduction of small alkyl chain at the 7-position of the main scaffold (compound 89).²⁰⁰

More recently, replacement of the trichlorophenyl ring at 2-position of PSB-10 86 and congeners with differently substituted five-membered heterocycles, like 1,3- and 1,5-disubstituted pyrazoles or 3-substituted isoxazoles (R-enantiomer 91 and 92, respectively, Fig. 7) was investigated by Baraldi’s group.¹⁹⁵ The 2-heterocyclic substitution induced excellent affinity and selectivity for the hA₃AR subtype. Docking of the most potent compound (91) in complex with a hA₃ARhomology model furnished a general survey of the hypothetical binding mode of the newly described derivatives.¹⁹⁵

4.4 Allosteric modulators of the A₃ adenosine receptor

Allosteric modulators of GPCRs bind at a location that is distinct from the binding site for a native ligand, that is, the orthosteric site, and this phenomenon has been reported for the A₃AR.²¹² In theory, the modulation may be positive, that is, with a PAM enhancing the activity of a directly acting (orthosteric) agonist, or negative, in the case of a NAM. A₃AR PAMs have been shown to enhance the potency and/or efficacy of agonists. When administered in vivo, they would be expected to be silent, with respect to A₃AR activity, unless either an endogenous or exogenous agonist would be present. Therefore, treatment with an A₃AR-selective PAM might display greater event- or site-specific action than an exogenous agonist, because it would magnify the effect of locally released adenosine, that is, in response to stress of an organ or tissue.

The SAR of three classes of A₃AR PAMs have been explored in detail: 1H-imidazo-[4,5-c]quinolin-4-amines (93 – 97),^213–215 2,4-disubstituted quinolines (98), ²¹⁶ and 3-(2-pyridinyl)isoquinolines (99, 100).²¹⁷ Representative key members of these structural classes are shown in Figure 8. In addition to these heterocycles, allosteric modulation of this receptor by compounds and ions that are not specific to the A₃AR, such as amiloride analogues and sodium, have been studied.^212,218

FIGURE 8.

Allosteric modulators of the A₃AR

The principal assay used to screen for allosteric modulators of the A₃AR has been to examine effects on the dissociation rate of an agonist radioligand, [¹²⁵I]I-AB-MECA 3.²¹² Many of the PAMs that have been reported also compete with the radioligand for specific binding. Thus, the objective in early studies was to identify lead molecules that impeded the dissociation rate, with minimal competitive binding potency.^213,217 At a concentration of 10 μM, the imidazoquinolinamines and DU124182 93 and DU124183 94 reduced the dissociation rate of the radioligand from the receptor by roughly half, and the phenylamino derivative 94 was more potent as an allosteric enhancer.²¹³ Dichloro substitution and expanding the cycloalkyl group in LUF6000 95, MRS5049 96, and MRS5190 97 were later shown to produce potent allosteric enhancement with less prominent competitive inhibition.^214,215 An amide LUF6096 98 was particularly selective for allosteric enhancement of the A₃AR compared to 95, but it displayed a short half-life in vivo. The residues of the A₃AR that are associated with the allosteric enhancement compared to orthosteric ligand binding were probed using stite-directed mutagenesis.²¹⁸

When multiple effector mechanisms induced by agonist Cl-IB-MECA were compared, 95 was found to modulate each activity in a different manner, that is, this imidazoquinolinamine appeared to be a functionally biased PAM.^108,219 For example, 95 had no effect on phosphorylation of ERK1/2, a small effect on β-arrestin2 translocation, and intense effects on cAMP inhibition and cell hyperpolarization. The agonist-enhancing effect of 95 was probe-dependent, that is, it had different degrees of modulation of different AR agonists, considered receptor “probes.” Notably, 95 greatly increased the efficacy of the naturally occurring nucleoside inosine, which is normally a weak and only partially efficacious agonist at the A₃AR. Inosine is a metabolite of adenosine and, like adenosine, is elevated in concentration when stress to an organ is present. Thus, 95 could produce therapeutic benefit in vivo partially from its enhancing action on inosine, as well as endogenous adenosine. 95 also enhanced the efficacy of nucleoside antagonists of the A₃AR, such as MRS542 38, to produce full agonism at the A₃AR. This indicates that nucleoside antagonists might behave as antagonists in a given functional model, but there is a latent agonism that can be amplified in the presence of a PAM. In contrast, nonnucleoside A₃AR antagonists did not display any latent agonism in the presence of an A₃AR PAM.

LUF6000 95 has anti-inflammatory effects in rat models of arthritis and a model of liver inflammation in mice,²²⁰ suggesting its potential use in the treatment of autoimmune inflammatory diseases. LUF6096 98 was shown to protect the heart in a model of canine cardiac ischemia,²¹⁸ suggesting the potential use of A₃AR PAMs in the treatment of ischemic conditions.

4.5 Structural characterization of the A₃ adenosine receptor

Although there is currently no X-ray crystallographic structure of the A₃AR available, effective use of modeling techniques has provided a window into its orthosteric binding site (Fig. 9). In particular, docking of agonists in conjunction with data from site-directed mutagenesis has demonstrated the close similarity of the A₃AR to the X-ray crystallographic structure of the hA_2AAR.²²¹ Complexes of the A_2AAR with four different agonists, namely 6-(2,2-diphenylethylamino)-9-((2R,3R,4S,5S)-5-(ethylcarbamoyl)-3,4-dihydroxytetrahydrofuran-2-yl)-N-(2-(3-(1-(pyridin-2-yl)piperidin-4-yl)ureido)ethyl)-9H-purine-2-carboxamide (UK432,097); adenosine; 5′-Nethylcarboxamidoadenosine (NECA); and 2-[p-(2-carboxyethyl)phenylethyl-amino]-5′-N-ethylcarboxamidoadenosine (CGS21680) (PDB IDs: 3QAK, 2YDO, 2YDV and 4UG2, respectively), have been crystallized and their structures determined.^222–224 The complexes revealed a common interaction pattern anchoring the adenosine moiety in the orthosteric binding site of the A_2AAR involving several conserved amino acid residues that are predicted to serve the same function in the A₃AR binding site. For example, the side chain Asn6.55 (using standard notation for numbering of TMs²²⁵) coordinates by H-bonding in a bidentate fashion with the adenine N⁶H and N7 in both ARs. His7.43 is in contact with the 2′-hydroxyl group of ribose. Also, Phe168 (EL2) is predicted to form π–π stacking with the adenine ring, as in the A_2AAR. However, there are some differences in the interaction of nucleoside ligands with A₃AR compared to A_2AAR that account for pharmacological differences of such ligands, with respect to their affinity, selectivity, and efficacy. For example, His6.52, which forms an H-bond to the 5′-carbonyl group of potent A_2AAR agonists occurs as Ser6.52 in the A₃AR and is not predicted to closely associate with typical nucleoside ligands. This might explain why binding of 4′-truncated nucleosides is maintained at the A₃AR relative to other ARs.

FIGURE 9.

(A) Docking pose of 3,4-difluorophenyl agonist analogue MRS5980 (27) at the hA₃AR homology model, in which TM2 is based on its position in the active β² adrenergic receptor. Residues interacting with the ligand (magenta carbon atoms) are labeled. H-bond and π–π interactions are represented as green solid and cyan dashed lines, respectively. Nonconserved ARs residues are in italics. (B) Docking pose of biphenyl agonist analogue MRS5679 (26) at the hA₃AR homology model. Residues interacting with the ligand (violet carbon atoms) are labeled and H-bond interactions are represented as green solid lines. The degree of displacement of TM2 with respect to the TM bundle in hA₃AR homology models based on the hA_2AAR (red ribbon), hybrid hA_2AAR-β₂ adrenergic receptor (purple ribbon), and hybrid hA_2AAR-opsin (violet ribbon) templates is highlighted with an arrow. TM1 is omitted to aid visualization

Conformational plasticity of the A₃AR has been proposed to account for the high-affinity binding of rigid C2-arylalkynyl agonists such as MRS5980 27.¹²⁰ When docking this class of compounds in a homology model of the A₃AR derived exclusively from the agonist-bound “active-like” A_2AAR structure, there was a steric clash with the extracellular tip of TM2. An outward movement of TM2, as observed in several other active GPCR structures such as the β₂-adrenergic receptor and opsin, was therefore hypothesized to occur also for the A₃AR. Thus, a hybrid homology model in which TM2 assumed its highly displaced position in opsin was required to dock biphenyl derivative MRS5679 26, and the other TMs followed their orientation in the agonist-bound A_2AAR.²²⁶ This proposed outward movement of TM2 was also associated with the degree of activational bias of C2-extended A₃AR agonists for cell survival, in a comparison of five different functional readouts.¹³¹

Molecular dynamics (MD) simulation of the A₃AR in complex with agonists have been reported.²²⁷ The analysis of the conformational changes of conserved Trp243 (6.48) as a result of agonist (Cl-IB-MECA) binding suggested that the ligand was able to promote and stabilize an expected conformational switch involved in receptor activation.

A supervised MD (SuMD) study simulated the approach of PAM LUF6000 95 toward a putative allosteric binding site on the A₃AR that contained an adenosine molecule in the orthosteric site.²²⁸ In the modeled ternary complex, 95 occupied the upper region of the orthosteric binding site by directly interacting with the agonist. The study suggested that 95 might exert its PAM activity by acting as “pocket cap.”

GPCRs may form characteristic dimers, either homo- or heterodimers, and the pharmacological implications of such dimerization remain to be fully characterized. Homodimers of the A₃AR have been detected using fluorescent methods.²²⁹ Indications are that in the future, the discovery of A₃AR ligands will rely heavily on computational methods to define the dynamic interaction of the receptor with its ligands and with other proteins, including other GPCR protomers.

5 INTRACELLULAR PATHWAYS IN IMMUNE AND CANCER CELLS

A schematic diagram showing the main signaling pathways triggered by adenosine through A₃AR activation in different cellular types is reported in Figure 10. The A₃AR preferentially couples to Gi protein to inhibit cAMP accumulation, and it may also couple to Gq protein to mediate stimulation of phospholipase C (PLC), which then increases calcium concentrations.²³⁰ The activation of PLC, and even the Ca²⁺ effects observed at high concentrations of A₃AR agonists, could conceivably be triggered by mechanisms other than Gq, such as Gβγ subunits. In addition, a Gq protein kinase C (PKC) dependent mechanism has been related to the apoptosis-inducing factor upregulation mediated by high doses of adenosine and of an A₃AR agonist in a human bladder cancer cell line.⁷⁸ However, PKC was recruited by A₃AR stimulation in order to increase tumor necrosis factor alpha (TNF-α) release in activated macrophages.²³¹

FIGURE 10.

Intracellular pathways of A₃AR in immune and cancer cells. Schematic diagram showing the main signaling pathways triggered by adenosine through A₃AR activation in different cellular types. A₃AR preferentially couples to Gi. PLC activation, and even the Ca²⁺ effects observed at high concentrations of A₃AR agonists, could conceivably be triggered by mechanisms other than Gq, such as Gβγ subunits

A huge amount of data supports a link between A₃AR and MAPKs in several cellular models.²³² A₃AR-mediated activation of extracellular signal-regulated kinases (ERK1/2) and mitogenesis modulation was found in human fetal astrocytes, CHO cells expressing the human A₃AR (CHO-hA₃), microglia, colon carcinoma, glioblastoma, melanoma and in foam cells.^{43,74,233–240} However, an inhibition of ERK activation has been revealed in A375 melanoma, prostate cancer, and glioma cells, leading to a reduction of cell proliferation as well as a decrease of TNF-α release in RAW 264.7 cells.^{22,73,241,242} The A₃AR also activates p38 MAPKs in different cellular models including CHO-hA₃, hypoxic melanoma, glioblastoma, and colon carcinoma cells,^235,237,243 while reducing p38 MAPKs in human synoviocytes.⁵³ Concerning c-Jun N-terminal kinase (JNK), its activation by A₃AR has been retrieved in microglia and glioblastoma cells, leading to cell migration and matrix metalloproteinase 9 (MMP-9) stimulation, respectively.^74,244 MEK/ERK1/2 and PI3K/Akt signaling pathways downstream of the A₃AR control multiple resistance-associated protein 1 (MRP1) transporter substrate in glioblastoma cells, demonstrating a chemosensitizing effect by pharmacological blockade of A₃AR and consequent reduction in tumor size.²⁴⁵

A pathway involving Akt phosphorylation protects RBL-2H3 and glioblastoma cells from apoptosis, while the same pathway produced an antiproliferative effect and increase in MMP-9 in A375 and glioblastoma cells, respectively.^{74,236,246,247} On the other hand, A₃AR-mediated activation of the phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)/Akt signaling pathway enhances pigmentation in B16melanoma cells and in human skin explants.⁸⁵

The PI3K/Akt and nuclear factor (NF) κB signaling pathways are the mediators of the anti-inflammatory A₃AR-mediated effects, as observed in activated BV2 microglial cells, monocytes, adjuvant-induced arthritis, and in mesothelioma.^75,248–253 Akt inhibition is involved in the reduction of A₃AR-mediated of hypoxia inducible factor 1 (HIF-1α) accumulation in murine astrocytes.²⁵⁴ In addition, the block of PI3K/Akt/mammalian target of rapamycin (mTOR) signaling through A₃AR suppresses angiogenesis in endothelial cells.¹²⁷

Furthermore, an A₃AR-mediated decrease in the protein kinase A (PKA) level was responsible for: (i) an increase in glycogen synthase kinase 3β (GSK-3β), leading to a downregulation of beta-catenin and its transcriptional gene products cyclin D1 and c-Myc, and (ii) a decrease in the level of NF-κB DNA-binding capacity. This effect through A₃AR activation has been reported in melanoma, hepatocellular carcinoma, synoviocytes from RA patients, and in adjuvant-induced arthritis rats.^89,255–257 Interestingly, a downregulation of A₃AR mRNA/protein expression in colon cells after ulcerative colitis by miR-206 led to an increase of NF-κB and the downstream cytokine (TNF-α/IL-8/IL-1β) expression in the mouse colon, producing a proinflammatory effect.¹⁹

6. BIOLOGICAL ROLE AND THERAPEUTIC APPLICATIONS IN MODELS OF IMMUNE DISORDERS AND CANCER

6.1 Preclinical studies in immune cells

The A₃AR is expressed in almost all cells of the immune system acting as essential mediators of adenosine’s role in inflammation.^258–260

A growing body of evidence suggests a key role of the A₃AR in neutrophil behavior. In recent years, it has been reported that the A₃AR is distributed in a highly polarized fashion on the neutrophils cell membrane and contribute to their chemotaxis and migration. Specifically, A₃AR is translocated to the neutrophil leading edge, and adenosine generated by ecto-ATPases/nucleotidases results in autocrine stimulation of A₃AR-enhanced polarized migration, following its initiation by ATP activating the P2Y₂ receptor.^34,261–263 This was confirmed also in a sepsis model of A₃ARknockout (KO) mice where a reduction in the recruitment of neutrophils to the lung and peritoneum was observed.²⁶⁴ In contrast, in human breast cancer cells, the A₃AR is expressed in multiple leading edges where it promotes cell migration with numerous directional changes. Indeed, exogenous adenosine may simultaneously stimulate A₃AR on all leading edges of the cell, inducing it to spread out in opposing directions resulting in arrest of cell motility.²⁶³ Furthermore, it has been found that the endogenous A₃AR aggregates into plaque-like microdomains that affect cytoskeletal remodeling. In addition, they promote the formation of membrane protrusions, also named cytonemes, which enable neutrophils to capture pathogens.³⁶ In contrast, data in early literature reported that chemotaxis and, in addition, oxidative burst were inhibited by A₃AR activation with anti-inflammatory effects.^32,33,35

Adenosine modulates monocyte-macrophage functions, being responsible for both inflammatory mediator production and resolution induction. For example, A₃AR stimulation inhibits the respiratory burst, interleukin (IL) 1β, TNF-α, chemokine macrophage inflammatory protein (MIP) 1α, interferon regulatory factor 1, iNOS (inducible nitric oxide synthase), and CD36 gene expression.^{38,40,41,249,265–267} However, adenosine reduced the expression of adhesion molecules on monocytes and decreased cytokine production, effects that were potentiated by an A₃AR antagonist.²⁶⁸ In addition, A₃AR stimulation increased TNF-α production in activated macrophages.²³¹

A functional A₃AR is expressed in dendritic cells, antigen-presenting entities activating naive T lymphocytes and starting primary immune responses.^269,270 In particular, the A₃AR in the human immature elements has been found to induce elevated Ca²⁺ levels, actin polymerization, and chemotaxis, while in mature dendritic cells, the A₃AR is down-regulated and decreases TNF-α release.^44,46

T cells represent the major actors in adaptive immunity and play a crucial role in the battle against infections and tumors. Both cytotoxic (CD8⁺) and helper (CD4⁺) T cells express A₃AR.^48,271 Initial studies attributed to the A₃AR an immunosuppressive role toward T cell mediated immune responses, but the mechanisms involved have not been investigated.^272,273 Interestingly, oral administration of an A₃AR agonist increased the cytotoxic activity of mouse NK cells and serum IL-12, thus reducing in vivo growth of melanoma cells.²⁷⁴ Indeed, ex vivo activation of CD8⁺ T cells with an A₃AR agonist improves adoptive immunotherapy for melanoma.²⁷⁵ A₃AR has been found to be upregulated in peripheral blood mononuclear cells (PBMCs) obtained from patients affected by different autoimmune disorders such as RA, Crohn’s disease, and psoriasis. This effect has been ascribed to the increase of TNF-α resulting in an upregulation of NF-κB and the cAMP response element binding protein (CREB), acting as A₃AR transcription factors.²⁷⁶ Therefore, A₃AR could be indicated as a biological predictive marker in autoimmune inflammatory pathologies. Furthermore, an immunosuppressive effect has been confirmed in lymphocytes derived from patients affected by RA, where the A₃AR, upregulated with respect to healthy subjects, inhibited the NF-κB signaling, inflammatory cytokines, and MMPs production. Their density inversely correlated with indexes used to assess RA disease activity, supporting the importance of A₃AR in the control of RA joint inflammation.²⁷⁷ Furthermore, A₃AR activation in rat models hampered damage of cartilage, formation of osteoclast/osteophyte, bone destruction, and generation of pannus and of lymphocyte formation.^278,279 Similarly, an A₃AR anti-inflammatory effect, that is, NF-κB-TNF-α-dependent has also been found in synoviocytes of RA patients.²⁵⁷ Interestingly, this mechanism has been suggested also in A₃AR-induced hepatoprotection against ischemia/reperfusion (IR) injury.²⁸⁰

The A₃ARhas long been recognized as a major contributor of rodent mast cell activation by increasing their degranulation and, more recently, this factor has been observed also in both primary human and LAD2 mast cells.^{9,13,63,82–84,281} A₃AR activation with an agonist significantly increased IL6, IL8, VEGF, amphiregulin, and osteopontin genes in human mast cells, affecting severe asthma.^285,286 Indeed, prolonged treatment with an agonist produced A₃AR downregulation responsible for the suppression of its basal inhibitory effect on cytokine production. This response was obtained only at a transcriptional level, suggesting that, at variance with rodents, in humans the primary role of the A₃AR is to act as a modulator rather than a stimulator of mast cell responses.²⁸⁷

Activation of the A₃AR induced hypothermia through the induction of mast cell degranulation, consequent histamine release, and activation of central histamine H₁ receptors.²⁸⁸

6.2 Preclinical studies in cancer cells

The role of A₃AR in cancer has been extensively studied using selective agonist and antagonist ligands, often with contrasting results. The efficacy of antagonists as anticancer drugs has been supported starting by the concept that hypoxia, characteristic of solid tumors, increases adenosine levels and stabilizes HIF-1α, the most important factor regulating cellular responses to the lack of oxygen.²⁸⁹ In this context, it has been reported that A₃AR stimulation induced HIF-1α accumulation in different cancer cell lines.^235,243 This has been seen to lead to an increase in angiopoietin-2 and/or VEGF, depending on the cell model investigated.²³⁷ Accordingly, it has been found that A₃AR stimulation increased microvessel density, expression of proangiogenic factors, macrophage tumor infiltration, and cytokine production in in vivo melanoma tumor models.²⁹⁰ Furthermore, a basal stimulation through A₃AR was responsible for an increased MRP1 expression in glioblastoma cells. As a consequence, A₃AR antagonist administration provoked a potentiating effect on the reduction in tumor size induced by the chemotherapic vincristine.²⁴⁵ In addition, an increase of glioblastoma cell invasion, through A₃AR and MMP-9 stimulation, was found, as already shown in macrophages.^74,291 However, a potential therapeutic effect of agonists in cancer has been also reported moving from initial studies on the observation that tumor metastases were infrequent in striated muscles. Interestingly, it has been found that in addition to adenosine, natural agonists of A₃AR were secreted from muscle cells, contributing to the systemic anticancer and chemoprotective activity exerted by muscle-conditioned medium. This evidence explained the rarity of tumor metastases in muscle and represented the rationale for the utility of A₃AR agonists in cancer.^292,293 Further studies reported that A₃AR activation inhibited telomerase activity and exerted cytostatic effects in tumor cells.^{10,255,294,295} Interestingly, cancer tissues and the interstitial fluid of several tumors contained high levels of adenosine able to activate ARs, among which the A₃AR subtype was the most expressed.^230,296,297 For example, an A₃AR upregulation has been found in human colorectal and hepatocellular carcinomas that was reflected in peripheral blood cells, thus making this AR subtype a possible marker for cancer, reflecting receptor status in remote tumor tissue.^87–89 As for the role of A₃AR in tumors, pro- and antiproliferative effects due to their activation have been documented in several cancer cell types.^{69,74,75,298–307} A₃AR activation decreased prostate cancer cell migration in vitro and in vivo and inhibited cell proliferation, inducing G1 cell cycle arrest and apoptosis.^70,73,308

Even though contrasting results about the role of A₃AR agonists and antagonists in cancer still are present in literature data, only the therapeutic utility of A₃AR agonists has been supported by in vivo studies. These studies encompassed syngeneic xenograft, orthotopic and metastatic experimental animal models of colon, prostate, melanoma, and hepatocellular carcinomas, in which IB-MECA and Cl-IB-MECA were administered orally in view of their stability and bioavailability. These agonists inhibited cell growth in syngeneic and lung metastatic models of murine melanoma, and potentiated cyclophosphamide effect.¹⁰ Also in xenograft models, IB-MECA inhibited the development of human colon and prostate tumors in nude mice and increased 5-fluorouracil or taxol antitumor effect. Furthermore, it blocked primary and liver metastases of colon carcinoma cells inoculated in the spleen. Finally, Cl-IB-MECA reduced hepatocellular tumor growth, liver inflammation, and cancer pain in rat bone residing breast cancer.^10,76,86,89

6.3 Clinical trials

The scientific evidence obtained through in vitro and in vivo experiments led to the progression of A₃AR agonists in clinical studies for the therapy of inflammatory and cancer pathologies. Importantly, they were found safe and well tolerated in all preclinical, Phases I and II human clinical studies. According to clinicaltrials.gov, the agonist IB-MECA (Piclidenoson, CF101) entered in the following clinical trials (NCT numbers):

1. RA (Phase II, NCT00280917; Phase II, NCT01034306; Phase II, NCT00556894) in which it showed significant antirheumatic effect as a standalone drug. Interestingly, a direct significant correlation was found between receptor expression at baseline and patients’ response to the drug. CF101 treatment resulted in an ACR20 (American College of Rheumatology Criteria for disease improvement) of 48.6%, statistically significantly higher than that of the placebo group (25%) at week 12. CF101 treatment also showed superiority in ACR50 and ACR70 values versus placebo. These data suggest that the A₃AR is a promising therapeutic target in RA and can be used also as a biologicalmarker to predict patients’ response to CF101. This is a unique type of a personalized medicine approach, which may pave the way for a safe and efficacious treatment for this patient population.

2. Plaque psoriasis (Phase II, NCT00428974; Phase II/III, NCT01265667) in which even though the primary endpoint, which was a 75% reduction in the psoriasis area and severity score (PASI75) at week 12, was not reached, however 24.7% of subjects achieved PASI90 at week 32.^309,310 In addition, CF101 was more efficacious than apremilast (Otezla), a PDE4 inhibitor, on week 32 while having an excellent safety profile, suggesting its promise as a chronic treatment.

3. Ocular hypertension (NCT01033422), keratoconjunctivitis sicca (dry eye disease; Phase II, NCT00349466; Phase III, NCT01235234), RA in cotreatment with methotrexate (NCT00280917) failed to reach their endpoints.

Furthermore, additional trials are expected in the near future for the treatment of: RA (Phase III, NCT02647762); osteoarthritis of the knee (Phase II, NCT00837291).

The agonist Cl-IB-MECA (Namodenoson, CF102) has been in the following clinical trials:

1. Advanced hepatocellular carcinoma (Phase I/II, NCT00790218), finding that it is able to induce a median overall survival (OS) of 7.8 months.³¹¹ In patients with advanced HCC and Child Puh B, CF102 induced an OS of 9.3 months (literature data demonstrate OS of 3.5 months). A global Phase II study in this patient population is currently ongoing.

2. Chronic hepatitis C genotype 1 (Phase I/II, NCT00790673).

Additional trials of CF102 are expected soon for the treatment of: nonalcoholic steatohepatitis (Phase II, NCT02927314) and hepatocellular carcinoma (Phase II, NCT02128958).

An A₃AR allosteric enhancer (CF602, LUF6000 95) demonstrated an anti-inflammatory effect in vivo. It is currently under development for treating sexual dysfunction.

7 CONCLUSIONS

The impact of the A₃AR on the drug discovery process and development is rapidly expanding, and its significance for human health should not be underestimated. Purine scientists are well advised to remain optimistic for the three following reasons. First, A₃AR is overexpressed in cancer and inflammatory cells in comparison to healthy cells, findings that are mirrored in the PBMCs of patients affected by these pathologies. Second, highly selective A₃AR agonists are now available as tool compounds and potential clinical molecules. They induce both anti-inflammatory/anticancer effects in pathological cells, as well as protective functions in normal cells, have been synthesized. Third, clinical data that correlate a high expression of A₃AR at baseline prior to agonist treatment with a beneficial response in patients suggest that modulation of the A₃AR may lead to a personalized medicine approach. Thus, new findings with A₃AR ligands appear to open new opportunities to fight inflammatory diseases, cancer, and other conditions.

Abbreviations

ADA

adenosine deaminase

AK

adenosine kinase

AR

adenosine receptor

cAMP

cyclic AMP

CCI

chronic constriction injury

CD73

ecto-5′-nucleotidase

CHO-hA₃

Chinese hamster ovary cells expressing the human A₃AR

Cl-IB-MECA

2-Chloro-N6-(3-iodobenzyl)-adenosine-5′-Nmethyluronamide

CNT

concentrative nucleoside transporter

CREB

cAMP response elements binding protein

ENT

equilibrative nucleoside transporter

ERK1/2

extracellular signal-regulated kinases

GPCR

G protein coupled receptor

GSK-3β

glycogen synthase kinase-3β

HIF-1α

hypoxia inducible factor 1

IB-MECA

N6-(3-Iodobenzyl)adenosine-5′-N-methyluronamide

IL

interleukin

IR

ischemia/reperfusion

JNK

c-Jun N-terminal kinase

KO

knockout

MIP

macrophage inflammatory protein

MMP-9

matrix metalloproteinase 9

MRP1

multiple resistance-associated protein 1

mTOR

mammalian target of rapamycin

NF-κB

nuclear factor κB

PET

positron emission tomography

PI3K

phosphatidylinositol-4,5-bisphosphate 3-kinase

PKA

protein kinase A

PKC

protein kinase C

PLC

phospholipase C

SAH

S-adenosylhomocysteine

TM

transmembrane domain

TNF-α

tumor necrosis factor α

PBMC

peripheral blood mononuclear cell

RA

rheumatoid arthritis.

Biographies

Kenneth A. Jacobson is a Senior Investigator and Chief of the Laboratory of Bioorganic Chemistry and the Molecular Recognition Section at the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health in Bethesda, Maryland, USA. Dr. Jacobson is a medicinal chemist with interests in the structure and pharmacology of G protein coupled receptors, especially receptors for adenosine and for purine and pyrimidine nucleotides. He was inducted into the American Chemical Soc. Medicinal Chemistry Hall of Fame in 2009.

Stefania Merighi received her doctoral degree in Biology Summa cum Laude in 1997 and her Ph.D. degree in Cellular and Molecular Pharmacology in 2003 from the University of Ferrara. Currently, she is Senior Researcher of Pharmacology at the University of Ferrara. Her research activities primarily focus on the pharmacological, biochemical, and molecular study of GPCRs in oncology and neuroinflammation.

Katia Varani received her doctoral degree in Biology in 1988 and her Ph.D. degree in Cellular and Molecular Pharmacology in 1995 from the University of Ferrara. She is currently Associate Professor of Pharmacology in the Medical Sciences Department of the University of Ferrara. Her research activities primarily focus on the pharmacological, biochemical, and molecular study of adenosine receptors.

Pier Andrea Borea received his degree in Chemistry from the University of Ferrara in 1967. He is currently President of the Evaluation Board of the University of Ferrara. He contributed to some 400 publications in international journals and about 20 chapters in international books. His main field of interest is represented by the study, at the molecular level, of drug–receptor interactions.

Stefania Baraldi received her doctoral degree in Pharmaceutical Chemistry and Technology with 110 cum Laude (University of Ferrara, 2005), her Ph.D. in Pharmaceutical Science/medicinal Chemistry (University of Ferrara, 2009) and her second-level master’s degree in Cosmetic Science and Technology with Top Marks (University of Ferrara, 2012). Her scientific interests have focused on the design and synthesis of ligands in the adenosine field and Endocannabinoid System modulators for the treatment of pain and inflammation. Her research activity has been supported by the company King Pharmaceutical (North Carolina, USA) now part of Pfizer.

Mojgan Aghazadeh Tabrizi received her doctoral degree in Pharmacy (University of Ferrara, 1987), Ph.D. in Medicinal Chemistry (University of Ferrara, 1992). She has worked at the Department of Pharmaceutical Sciences of Ferrara University focusing her interests on the adenosine receptors ligands and antitumor compounds in collaboration with the company King Pharmaceuticals (North Carolina, USA), now part of Pfizer. Since 2008, she has been part of a research project on the treatment of pain and inflammation involving the endocannabinoid system. She is author/coauthor of several scientific publications in the field of medicinal chemistry.

Romeo Romagnoli graduated in Pharmaceutical Chemistry and Technology at the University of Ferrara in 1990, and in 1995 he received his Ph.D. in Organic Chemistry. From 1999 to 2014, he has held a position of Assistant Professor in Pharmaceutical Chemistry at the University of Ferrara. Actually he is Associate Professor in Pharmaceutical Chemistry at the same University. His scientific interests have focused on new synthetic methods for the preparation of natural substances or biologically active analogues, the medicinal chemistry of ligands for adenosine receptors subtypes and antitumor agents. He has published more than 230 research papers and 20 international patents.

Pier Giovanni Baraldi received his doctoral degree in Chemistry in 1974 from the University of Ferrara where he is currently Full Professor of Medicinal Chemistry. He has published more than 400 scientific papers including about 50 patents and he participated in more than 90 Medicinal Chemistry meetings as plenary speaker. His research interests have focused on the design and synthesis of minor groove alkylating agents, combretastatin analogs, ligands for ARs, cannabinoid receptors, and TRP channel modulators. He has been promoter of several scientific collaborations with national and international pharmaceutical companies.

Antonella Ciancetta received hermaster’s degree in Pharmaceutical Chemistry and Technology in 2007 and her Ph.D. degree in Drug Sciences in 2010 from the University of Chieti, Italy. She is currently a Visiting Fellow at the National Institutes of Health in Bethesda, Maryland, USA in Dr. Jacobson’s laboratory. Her research focuses on the application and development of structure-based molecular modeling techniques for the design of GPCR ligands and the elucidation of their binding mechanisms.

Dilip K. Tosh is a Staff Scientist in the Molecular Recognition Section, Laboratory of Bioorganic Chemistry, NIDDK, National Institutes of Health in Bethesda, Maryland, USA. He is a medicinal chemist in the field of G protein coupled receptors and transporters, especially interested in adenosine receptors and other targets of nucleosides and nucleotides. He received his Ph.D. in Synthetic Organic Chemistry from the Indian Institute of Technology, Bombay, India.

Zhan-Guo Gao is a Staff Scientist in the Molecular Recognition Section, Laboratory of Bioorganic Chemistry, NIDDK, National Institutes of Health in Bethesda, Maryland, USA. He is a pharmacologist in the field of G protein coupled receptors, especially interested in adenosine receptors and P2Y receptors.

Stefania Gessi received her Doctor degree in Biology Summa cum Laude from University of Ferrara in 1994 and her PhD in Cellular and Molecular Pharmacology in 2000. She is currently Associate Professor of Pharmacology in the Medical Sciences Department of the University of Ferrara. Her research activity focuses on the pharmacological, biochemical and molecular study of adenosine receptors in health and diseases.