Discovery of New Human A_2A Adenosine Receptor Agonists: Design, Synthesis, and Binding Mode of Truncated 2-Hexynyl-4′-thioadenosine

Abstract

The truncated C2- and C8-substituted-4′-thioadenosine derivatives 4a-d were synthesized from D-mannose, using palladium-catalyzed cross coupling reactions as key steps. In this study, an A₃ adenosine receptor (AR) antagonist, truncated 4′-thioadenosine derivative 3 was successfully converted into a potent A_2AAR agonist 4a (K_i = 7.19 ± 0.6 nM) by appending a 2-hexynyl group at the C2-position of a derivative of 3 that was N⁶-substituted. However, C8-substitution greatly reduced binding affinity at the human A_2AAR. All synthesized compounds 4a-d maintained their affinity at the human A₃AR, but 4a was found to be a competitive A₃AR antagonist/A_2AAR agonist in cyclic AMP assays. This study indicates that the truncated C2-substituted-4′-thioadenosine derivatives 4a and 4b can serve as a novel template for the development of new A_2AAR ligands.

The endogenous cytoprotective modulator adenosine (1) exerts its pharmacological effects against hypoxia, ischemia, and inflammation through binding to four subtypes (A₁, A_2A, A_2B, and A₃) of adenosine receptors (ARs), members of the G protein-coupled receptor (GPCR) family.¹ Among these, activation of A_2AAR has been known to play a role in the suppression of immune and inflammatory responses and in vascular responses to adenosine.² It is also highly localized within the central nervous system (CNS), and selective antagonists have become an attractive target for the treatment of Parkinson’s disease.³ The A₃AR is the most recently identified subtype and is found in the cardiovascular system, CNS, immune cells, lung, and liver.^4,5 The activation of A₃AR is beneficial in models of myocardiac and cerebral ischemia and cancer, while its antagonism is of interest for treating asthma, inflammation, and glaucoma.⁶

Thio-Cl-IB-MECA (2)⁷ which is bioisosteric with 2-chloro-N⁶-(3-iodobenzyl)adenosine-5′-N-methyluronamide (Cl-IB-MECA)⁸ was discovered as a potent and selective A₃AR agonist (K_i = 0.38 nM). This compound showed a potent anticancer activity by inhibiting the Wnt signaling pathway.⁹ On the basis of rational design, truncated 4′-thioadenosine derivative 3 lacking the 5′-uronamide of 2 essential for the receptor activation was discovered as a potent, selective, and species-independent A₃AR antagonist (K_i = 1.66 nM).¹⁰ This compound and other related A₃AR antagonists with nucleoside skeletons are expected to be suitable for evaluation in small animal models and for further development as drugs (Figure 1).

Figure 1.

The rationale for the design of A_2A adenosine receptor agonists

Based on the observation that truncation resulting in the 4′-thioadenosine antagonist derivative 3 preserved AR affininity and selectivity, we designed and synthesized the truncated C2- and C8-substituted 4′-thioadenosine derivatives 4a-d as potential new ligands for the A_2AAR. This expectation was supported by reports that C2- or C8-substitution sometimes leads to substantial enhancement in the binding affinity or selectivity at the A_2AAR or other AR subtype.¹¹ C2- or C8-substitution was readily achieved through Sonogashira¹² and Suzuki¹³ cross coupling reactions. From this study, a C2-alkynyl derivative was found to be potent, mixed A_2AAR agonist and A₃AR antagonist, which is an excellent combination for the anti-asthmatic activity. Herein, we describe the synthesis and pharmacological activity of novel C2- and C8-substituted 4′-thioadenosine derivatives 4a-d from D-mannose.

D-Mannose was converted to the glycosyl donor 5 according to our previously published procedure.¹⁰ The glycosyl donor 5 was condensed with 2-amino-6-chloropurine in the presence of TMSOTf as a Lewis acid to give the β-anomer 6 (30%) as a single stereoisomer (Scheme 1). The anomeric assignment was easily accomplished in a ¹H NMR experiment that showed a NOE between H-8 and 3′-H. Treatment of 2-amino-6-chloro derivative 6 with isoamyl nitrite, iodine, and methylene iodide in the presence of CuI afforded the 2-iodo-6-chloro derivative 7, which was converted to the 2-iodo-6-amino derivative 8 upon treatment with methanolic ammonia. Sonogashira¹² coupling reaction of 8 with 1-hexyne in the presence of bis(triphenylphosphine)palladium dichloride yielded the 2-hexynyl derivative 9. Finally, removal of the isopropylidene of 9 with 1 N HCl produced the final 2-hexynyl-4′-thioadenosine derivative 4a. Suzuki¹³ coupling reaction of the 2-iodo derivative 8 with (E)-1-catecholboranylhexene¹⁴, prepared by treating with 1-hexyne and catecholborane, in the presence of tetrakis(triphenylphosphine)palladium(0) afforded the 2-hexenyl derivative 10. Removal of the acetonide of 10 with 1 N HCl gave the 2-hexenyl-4′-thioadenosine derivative 4b.

Scheme 1. Synthesis of the 2-substituted derivatives 4a and 4b.

Reagents and conditions: a) silylated 2-amino-6-chloropurine, TMSOTf, DCE, rt to 80 °C, 3 h; b) CuI, isoamyl nitrite, I₂, CH₂I₂, THF, 110 °C, 45 min; c) NH₃/MeOH, 80 °C, 2 h; d) 1-hexyne, CuI, TEA, DMF, bis(triphenylphosphine)palladium dichloride, rt, 3 h; e) 1 N HCl, THF, rt, 15 h; f) (E)-1-catecholboranylhexene, then tetrakis(triphenylphosphine)palladium(0), Na₂CO₃, DMF, H₂O, 90 °C, 15 h.

Using a strategy similar to Scheme 1, 8-substituted adenosine derivatives 4c and 4d were synthesized from the glycosyl donor 5 (Scheme 2). Condensation of 5 with 8-bromoadenine¹⁵ under Lewis acid conditions, afforded the 8-bromo derivative 11. Coupling of 11 with 1-hexyne under Sonogashira conditions gave 8-hexynyl derivative 12, which was treated with 1 N HCl to yield the final 8-hexynyl-4′-thioadenosine derivative 4c.

Scheme 2. Synthesis of the 8-substituted derivatives 4c and 4d.

Reagents and conditions: a) silylated 8-bromoadenine, TMSOTf, DCE, rt to 90 °C, 2 h; b) 1-hexyne, CuI, TEA, DMF, bis(triphenylphosphine) palladium dichloride, rt, 3 h; c) 1 N HCl, THF, rt, 15 h; d) (E)-1-catecholboranylhexene, then tetrakis(triphenylphosphine) palladium(0), Na₂CO₃, DMF, H₂O, 90 °C, 15 h.

The 8-bromo derivative 11 was condensed with (E)-1-catecholboranylhexene¹⁴ under Suzuki conditions¹³ to give the 2-hexenyl derivative 13. Removal of the isopropylidene group of 13 under acidic conditions afforded the final 8-hexenyl-4′-thio adenosine derivative 4d.

Binding assays were carried out using standard radioligands and membrane preparations from Chinese hamster ovary (CHO) cells stably expressing the human (h) A₁ or A₃AR or human embryonic kidney cells (HEK-293) expressing the hA_2AAR.¹⁶ Unlike the parent N⁶-substituted compound 3 that only weakly bound to the A_2AAR, C2-substituted variations of compound 3 led to a dramatic increase in the binding affinity (K_i = 7.19 ± 0.6 nM for 4a and 72.0 ± 19.1 nM for 4b) at the hA_2AAR, while maintaining high binding affinity (K_i = 11.8 ± 1.3 nM for 4a and 13.2 ± 0.8 nM for 4b) at the hA₃AR (Table 1).

Table 1.

Binding affinities of known A₃AR antagonist 3 and truncated 2- and 8-substituted-4′-thioadenosine derivatives 4a-d at three subtypes of hARs.

Compounds	Affinitya
	hA1 (% inhibition)	hA2A (% inhibition)	hA3 (Ki, nM)
3	37.9%	17.7%	1.66 ± 0.90
4a	38.9 ± 9.9%	97.2 ± 4.1%	11.8 ± 1.3
4b	16.2 ± 8.4%	95.9 ± 8.7%	13.2 ± 0.8
4c	49.3 ± 4.9%	46.5 ± 4.3%	20.0 ± 4.0
4d	3.7 ± 2.9%	22.8 ± 6.4%	259 ± 10

^aAll binding experiments were performed using adherent mammalian cells stably transfected with cDNA encoding the appropriate hAR (A₁AR and A₃AR in CHO cells and A_2AAR in HEK-293 cells). Binding was carried out using 1 nM [³H]CCPA, 10 nM [³H]CGS21680, or 0.5 nM [¹²⁵I]I-AB-MECA as radioligands for A₁, A_2A, and A₃ARs, respectively. Values are expressed as mean ± sem, n = 3–4 (outliers eliminated) and normalized against a non-specific binder, 5′-N-ethylcarboxamidoadenosine (NECA, 10 μM). Values expressed as a percentage refer to percent inhibition of specific radioligand binding at 10 μM, with nonspecific binding defined using 10 μM NECA.

These results indicate that bulky hydrophobic pockets exist in the binding sites of A_2AAR and A₃AR, allowing the C2-substituent to form favorable hydrophobic interactions. The 2-alkynyl derivative 4a showed a better binding affinity than the 2-alkenyl derivative 4b. Interestingly, C8-substitution on 3 abolished the binding affinity at the hA_2AAR, but the binding affinity at the hA₃AR was maintained although decreased. These findings suggest that a bulky hydrophobic group at position 8 could be tolerated at the binding site of the hA₃, but not hA_2AAR. The ability to enhance affinity at the A_2AAR in the truncated series by extending an unsaturated carbon chain at the 2 position implies a mode of receptor binding in common with the riboside series.¹¹ All compounds showed very weak binding affinity at the hA₁AR.

Compounds 4a and 4b were found to be potent and full antagonists in a cyclic AMP functional assay at the hA₃AR. In this assay, 4a dose-dependently shifted the concentration-response curve for agonist Cl-IB-MECA to the right as an antagonist, corresponding to a K_B value of 1.69 nM calculated by Schild analysis (Figure 2). This is consistent with previous studies in which truncated N⁶-substituted 4′-thioadenosine derivatives have generally displayed A₃ AR antagonist activity.¹⁰ However, in a cyclic AMP functional assay at the hA_2AAR expressed in CHO cells, compound 4a behaved as a full agonist compared to the standard 2-[p-(2-carboxyethyl)phenyl-ethylamino]-5′-N-ethylcarboxamidoadenosine (CGS21680) and displayed an EC₅₀ of 12 nM. At the hA_2BAR expressed in CHO cells, 4a was a weak partial agonist in cyclic AMP accumulation (EC₅₀ ~10 μM). The finding that compound 4a is both a potent and selective agonist at the hA_2AAR and a competitive antagonist at the hA₃AR is similar to the pharmacological profile of a more heavily 2,5′-substituted adenosine derivative¹⁷ which inhibited both formation of ROS and eosinophils degranulation for the anti-asthmatic activity.

Figure 2.

Parallel right shifts induced by compound 4a on the concentration-response curve of a full agonist in the inhibition of cyclic AMP production at the hA₃AR expressed in CHO cells (A), the corresponding Schild plot (B) and the activity of 4a as a full agonist at the h A_2AAR expressed in CHO cells, compared to CGS21680 (C).

To investigate the binding mode, we performed a study of docking the C2- and C8-substituted-4′-thioadenosine derivatives 4a-d in the hA_2AAR X-ray crystallographic structure (PDB code: 3EML)¹⁸ using the GOLD software¹⁹, considering the flexibility of the binding site residues. As shown in Figure 3, the C2-substituted-4′-thioadenosine derivatives 4a and 4b, whose binding affinities are in the nM range, occupied the binding site very well. Their bulky and rigid C2-substituents oriented toward the extracellular region, forming hydrophobic interactions. The adenine moieties appeared to form three H-bonds with Glu169 and Asn253, and the ring systems were in π-π stacking with Phe168. Also, the thio-sugar rings were located deep inside the binding pocket, and the 3′-OH groups were able to donate a H-bond to Ser277. Based on this result, the C8-substituents were expected to be oriented in a hydrophobic pocket deep inside the binding site, which was not occupied by 4a and 4b in Figure 3. However, the C8-substitued derivatives 4c and 4d showed various binding modes (data not shown). In addition to the expected binding mode, the C8-substituents alternately pointed toward the extracellular region through a rotation of the bond between the adenine and thio-sugar rings. It might be due to spatial restriction of the long and rigid C8-substituents in the hydrophobic pocket inside the binding site, which would cause the nucleosides to lose some H-bonding and/or π-π stacking interactions that were shown for the C2-substituted derivatives. These results might explain why the A_2AAR affinities of 4c and 4d were reduced.

Figure 3.

Predicted binding modes of (A) 4a and (B) 4b docked in the hA_2AAR crystal structure. The key interacting residues are marked and displayed in capped-stick, except Phe168 in ball-and-stick, with carbon atoms in white. The ligands are depicted as ball-and-stick with carbon atoms in magenta (4a) and purple (4b). Hydrogen bonds are shown in yellow dashed lines. The Van der Waals surfaces of the ligands were generated by MOLCAD and colored by hydrogen bonding property (red: H-bond donating regions; blue: H-bond accepting regions). The Fast Connolly surface of the protein is Z-clipped and non-polar hydrogens are undisplayed for clarity.

In conclusion, we synthesized the truncated C2-and C8-substituted-4′-thioadenosine derivatives 4a-d, starting from D-mannose, using palladium-catalyzed cross coupling reactions as key steps. Although the antagonist activity of various truncated 4′-thionucleosides at the A₃AR was well explored previously, this is the first characterization of the functional activity of such derivatives at the A_2AAR. From this study, we successfully identified potent and sterically compact A_2AAR agonists, 4a and 4b by placing extended hydrophobic 2-hexynyl or 2-hexenyl groups on truncated and N⁶-unsubstituted 4′-thioadenosine derivatives. This observation was supported by molecular modeling that placed the chain at the 2 position in a hydrophobic region of the A_2AAR. However, C8-substitution greatly reduced binding affinity at the hA_2AAR. Thus, the absence of a 5′-uronamide of typical A_2AAR agonists² or the native -CH₂OH of adenosine did not preclude potent binding and full activation of the hA_2AAR. This suggested a major difference between A_2AAR and the A₃AR in the pathway of receptor activation. All synthesized compounds 4a-d maintained their binding affinity at the human A₃AR, and as for the 2-H or 2-Cl analogues that were N⁶-substituted, competitive A₃AR antagonism was demonstrated. This study establishes that the truncated C2-substituted-4′-thioadenosine derivatives 4a and 4b can serve as a novel template for the development of new A_2AAR ligands, although they still may interact at the A₃AR. This mixed activity as A_2AAR agonist/A₃AR antagonist might also be advantageous in disease models such as asthma.¹⁷ Thorough elucidation of the structure-activity relationship of this series is in progress in our laboratory.