4-Heteroaryl Substituted Amino-3,5-Dicyanopyridines as New Adenosine Receptor Ligands: Novel Insights on Structure-Activity Relationships and Perspectives

A new set of amino-3,5-dicyanopyridines was synthesized and biologically evaluated at the adenosine receptors (ARs). This chemical class is particularly versatile, as small structural modifications can influence not only affinity and selectivity, but also the pharmacological profile. Thus, in order to deepen the structure–activity relationships (SARs) of this series, different substituents were evaluated at the diverse positions on the dicyanopyridine scaffold. In general, the herein reported compounds show nanomolar binding affinity and interact better with both the human (h) A1 and A2A ARs than with the other subtypes. Docking studies at hAR structure were performed to rationalize the observed affinity data. Of interest are compounds 1 and 5, which can be considered as pan ligands as binding all the ARs with comparable nanomolar binding affinity (A1AR: 1, Ki = 9.63 nM; 5, Ki = 2.50 nM; A2AAR: 1, Ki = 21 nM; 5, Ki = 24 nM; A3AR: 1, Ki = 52 nM; 5, Ki = 25 nM; A2BAR: 1, EC50 = 1.4 nM; 5, EC50 = 1.12 nM). Moreover, these compounds showed a partial agonist profile at all the ARs. This combined AR partial agonist activity could lead us to hypothesize a potential effect in the repair process of damaged tissue that would be beneficial in both wound healing and remodeling.


Introduction
The amino-3,5-dicyanopyridines have attracted much attention due to their versatility to behave as AR ligands. In fact, they are endowed with not only a wide range of affinity but also with different degrees of activities, with their profile varying from full to partial agonist or neutral antagonist at the different ARs [1 -7]. As for other G-protein-coupled receptors [8][9][10], some of the antagonists have been proven to be inverse agonists [6]. The interest in modulating the effects of the natural ligand adenosine ensued from the evidence of its involvement in a large variety of physiological functions throughout the body. Interaction of the ubiquitous adenosine with its four G-protein-coupled A 1 , A 2A , A 2B and A 3 ARs produced different responses depending on the type of AR and consequent cellular signaling involved. In fact, the A 1 and A 3 ARs are coupled to G i
The analysis of the 1 H-NMR spectra of the target amino-3,5-dicyanopyridines, bearing a 2-(1H-imidazol-2-yl-methyl)sulphanyl substituent on the 6-side chain (1-10, 12, 18-21), featured a particular behavior. In the 1 H NMR spectra of both the crude and the purified compounds (1-3, 5, 8, 9 and 18-21), the two imidazole protons at the 4 and 5 position appeared as a single signal (singlet) around 7 ppm, which integrated 2 (equivalent protons), while the NH proton was a very broad signal from 12 to 13 ppm. The same applied for compounds 4, 6, 7, 10 and 12 before purification. In this situation, it was possible to hypothesize a fast tautomeric equilibrium, which determines the equivalence of the two protons and influences the relaxation time of the imidazole NH group (broad signal). However, the pattern of the 1 H NMR spectra of compounds 4, 6, 7, 10 and 12 changed after crystallization or purification by silica gel column chromatography. In fact, the two imidazole protons at 4,5 appeared as distinct signals falling around 6.8 and 7.1 ppm, while the signal of the NH proton was a singlet around 12 ppm. This experimental evidence led us to hypothesize that the two distinct signals of the imidazole protons in the purified compounds 4, 6, 7, 10 and 12 indicated the formation of an intramolecular hydrogen bond between the pyridine nitrogen at position 1 and the imidazole NH on the 6-side chain. The two distinct signals could be due to the stiffening of the sulfanylmethyl-imidazole chain as a result of which the neighborhood of the two protons at 4 and 5 becomes different. This hypothesis was supported by a 1 H NMR study of compound 10, taken as reference. A detailed report of the registered spectra is reported in the Supplementary Materi- The synthesis of the target dicyanopyridines 14-17 starting from the common intermediate 32 is reported in Scheme 2. Treatment of the latter with chloroacetic acid or 4-(chloromethyl)-1,3-thiazole-2-carboxylic acid (57) [35] (see Scheme 5 below) in anhydrous DMF, in mild alkaline conditions (NaHCO 3 ), furnished the 2-sulfanyl-acetic acid 42 and the 2-methylsulfanyl-1,3-thiazole-2-carboxylic acid derivative 43, respectively. Compound 42 was isolated from the crude mass reaction by acidification with 6M HCl. Both 42 and 43 were converted respectively into the target derivatives 14-16 and 17, with variable yields, by reacting with (1-(1H-imidazol-2-yl)methanamine hydrochloride (compound 14), 1-(1H-imidazol-4-yl)methanamine (55) (see Scheme 4 below) [36] (compound 15) or 2-(1Himidazol-4-yl)ethanamine (compounds 16 and 17), in anhydrous DMF, in the presence of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride, 1-hydroxybenzotriazole hydrate and Et 3 N.
The analysis of the 1 H-NMR spectra of the target amino-3,5-dicyanopyridines, bearing a 2-(1H-imidazol-2-yl-methyl)sulphanyl substituent on the 6-side chain (1-10, 12, 18-21), featured a particular behavior. In the 1 H NMR spectra of both the crude and the purified compounds (1-3, 5, 8, 9 and 18-21), the two imidazole protons at the 4 and 5 position appeared as a single signal (singlet) around 7 ppm, which integrated 2 (equivalent protons), while the NH proton was a very broad signal from 12 to 13 ppm. The same applied for compounds 4, 6, 7, 10 and 12 before purification. In this situation, it was possible to hypothesize a fast tautomeric equilibrium, which determines the equivalence of the two protons and influences the relaxation time of the imidazole NH group (broad signal). However, the pattern of the 1 H NMR spectra of compounds 4, 6, 7, 10 and 12 changed after crystallization or purification by silica gel column chromatography. In fact, the two imidazole protons at 4,5 appeared as distinct signals falling around 6.8 and 7.1 ppm, while the signal of the NH proton was a singlet around 12 ppm. This experimental evidence led us to hypothesize that the two distinct signals of the imidazole protons in the purified compounds 4, 6, 7, 10 and 12 indicated the formation of an intramolecular hydrogen bond between the pyridine nitrogen at position 1 and the imidazole NH on the 6-side chain. The two distinct signals could be due to the stiffening of the sulfanylmethyl-imidazole chain as a result of which the neighborhood of the two protons at 4 and 5 becomes different. This hypothesis was supported by a 1 H NMR study of compound 10, taken as reference. A detailed report of the registered spectra is reported in the Supplementary Materials. Moreover, an ab initio quantum mechanical calculation on compound 10, and also on the other congeners, namely 4, 6, 7 and 12, was performed. The program used was GAMESS (General Atomic and Molecular Electronic Structure System) [37], and the discussion is reported in the Section 3.2 ("Ab Initio Quantum Mechanical Studies").

Molecular Docking Studies
Molecular docking studies were performed at the hA 1 AR and hA 2A AR 3D structures to analyze the binding data of the synthesized compounds. The cryo-EM structure of the adenosine-bound hA 1 AR (pdb code: 6D9H; 3.6-Å resolution [38]) and the crystal structure of the hA 2A R in complex with the inverse agonist ZM241385 (pdb code: 4EIY; 1.8-Å resolution [39]) were retrieved from the protein data bank (https://www.rcsb.org/, accessed on 6 April 2012) and employed as molecular targets. Docking analyses were performed with the CCDC Gold [40] docking algorithm by using MOE (Molecular Operating Environment, version 2019.0101) suite [41] as the interface.

In Vitro Permeation Studies
Compounds 1 and 5 were evaluated for their capability to penetrate the artificial membrane simulating the epidermal barrier. Permeation flux was assessed by using vertical Franz diffusion cells [42].

Structure-Activity Relationships
The pharmacological results of the newly synthesized amino-3,5-dicyanopyridine derivatives 1-21 and those of the reference compound LUF5833 (2-amino-6-[(1H-imidazol-2-ylmethyl)sulfanyl]-4-phenylpyridine-3,5-dicarbonitrile) [2] are reported in Tables 1 and 2. To better follow the discussion in both this section and in the molecular modeling one, LUF5833 was considered as a reference to define the numbering of the dicyanopyridine core to refer to. The pyridine nitrogen atom represents position 1, while the amino and the sulfanyl function occupy position 2 and 6, respectively.
Most of the reported compounds were devoid of an affinity for the hA 2B AR, with the exception of derivatives 1, 4 and 5-8 showing EC 50 values for this receptor below 63 nM ( Table 1). As observed for other reported set of this series [5][6][7], the A 3 AR affinity was, in general, null or fell in the micromolar range. The only two exceptions were compounds 1 and 5, which displayed a nanomolar K i value for this receptor subtype. In general, the herein reported compounds interacted better with both the hA 1 and A 2A ARs than with the other subtypes. However, compounds 1 and 5, which can be considered as pan ligands binding all the ARs with comparable affinity and in the nanomolar range, were interesting. The analogs 4 and 6 had a similar trend though binding the hA 3 AR with lower affinity with respect to 1 and 5 but similarly to the lead LUF5833. Moreover, compounds 1 and 5, when evaluated in the functional tests, showed a partial agonist profile at all the ARs (Table 2).                                                 In general, keeping constant both the 2-amino function and the 6-(1H-imidazol-2ylmethyl)sulfanyl side chain on the dicyanopyridine core, we can see that the introduction of different heteroaryl groups at the 4 position (compounds 1-10) influenced the binding affinity at the diverse ARs differently. All of these compounds showed an hA 1 AR K i value below 10 nM, with the only two exceptions being compounds 2 and 10. A common feature of these latter two compounds was a hydrophilic hydroxyl group appended on the 4-heteroaryl moiety. Regarding the A 2A receptor, the presence at the 4-position of either furanyl or thienyl rings (compounds 1-6) seemed to better promote the binding interaction with this subtype than a pyridine substituent (derivatives 7-10). Moreover, as observed in previously reported set of compounds of this series [5][6][7], the presence of a substituent on the 4-(hetero)aryl moiety dramatically influenced the hA 2B AR activity (compare compound 1 to 2, 3, and compound 8 to 9, 10, respectively).
As previously reported [6], replacement of the methylsulfanyl linker with a methylamino one led to a high decrease of affinity at all the ARs (compare compound 1 to 12). Moreover, the presence of a longer 6-linker between the 1H-imidazol-2yl group and the dicyanopyridine scaffold in general negatively influenced the affinity at all the ARs, and, to a minor extent, the binding at the A 1 subtype. In fact, compounds 16 and 17 maintained K i values in the low nanomolar range.
The replacement of the methylsulfanyl linker on the 6-side-chain with a methylamino (compound 1 versus 12), or with a longer one (compound 1 versus 16), produced a shift of the pharmacological profile from partial agonist at the A 1 AR to antagonist or inverse agonist, respectively. Similarly, when the 1H-imidazol-2-yl was replaced with the 1Himidazol-4-yl moiety (compound 1 versus 11) a change from partial agonist to antagonist profile was observed ( Table 2). These data confirm that the requirements at the R position are very precise. In support of this, the inverse agonist profile (compound 16) was changed back to partial agonist (compound 17) when the 1,3-thiazol-5-yl moiety was directly attached to the methylsulfanyl linker. This result was in accordance with the literature data [4,[43][44][45]. In fact, many amino-3,5-dicyanopyridines bearing the thiazole feature in this precise position were reported as potent and selective A 1 AR agonists.
The introduction of cycloalkyl substituents on the 2-amino function (compounds 18 and 19) and the inclusion of the latter in a (pyrrolidin-1-yl) moiety (compound 20) or its acetylation (compound 21) differently affected the AR binding affinities with respect to the unsubstituted derivative 1. Interaction with the A 2B receptor was completely lost, while affinity was retained at the other AR subtypes. In particular, while compound 20 maintained A 1 and A 2A AR K i values in the nanomolar range, the 2-acetylamino derivative 21 bound the A 1 subtype with good affinity.

Ab Initio Quantum Mechanical Studies
The hypothesis of the intramolecular H-bond formation between the pyridine nitrogen at position 1 and the imidazole NH hydrogen was supported by ab initio quantum mechanical calculations performed on compounds 4, 6, 7, 10 and 12. The program used was GAMESS [37]. The formation energies (E 1 and E 2 ) of two conformations, 1 and 2, were calculated (Table 3) with Geometry Optimization, i.e., with conformational optimization. Conformation 1 included the formation of an intramolecular H-bond between the imidazole NH hydrogen and the pyridine nitrogen at position 1. As an example, the two minimized conformations, 1 and 2, of the reference compound 10 (10-1 and 10-2) were depicted in Figure 2. Referring to the compounds under study, negative ∆E values (Table 3) indicated that conformations 1 were slightly more stable than the 2 ones, where the H-bond was not supposed. Thus, the observed energy gain was attributable to the intramolecular H-bond formation, but its low value, which was generally observed, indicated that this hypothetical H-bond was very weak. In fact, in the presence of water, a perturbation of the system, was observed, and the intramolecular H-bond was not formed. In this situation, the imidazole NH hydrogen established a stabilizing H-bond with a water molecule, leading to an energy gain of about −11 Kcal/mol.

Ab Initio Quantum Mechanical Studies
The hypothesis of the intramolecular H-bond formation between the pyridine nitrogen at position 1 and the imidazole NH hydrogen was supported by ab initio quantum mechanical calculations performed on compounds 4, 6, 7, 10 and 12. The program used was GAMESS [37]. The formation energies (E1 and E2) of two conformations, 1 and 2, were calculated (Table 3) with Geometry Optimization, i.e., with conformational optimization. Conformation 1 included the formation of an intramolecular H-bond between the imidazole NH hydrogen and the pyridine nitrogen at position 1. As an example, the two minimized conformations, 1 and 2, of the reference compound 10 (10-1 and 10-2) were depicted in Figure 2. Referring to the compounds under study, negative ΔE values (Table 3) indicated that conformations 1 were slightly more stable than the 2 ones, where the Hbond was not supposed. Thus, the observed energy gain was attributable to the intramolecular H-bond formation, but its low value, which was generally observed, indicated that this hypothetical H-bond was very weak. In fact, in the presence of water, a perturbation of the system, was observed, and the intramolecular H-bond was not formed. In this situation, the imidazole NH hydrogen established a stabilizing H-bond with a water molecule, leading to an energy gain of about −11 Kcal/mol. (10-1) (10-2)

Molecular Modeling Studies
Analogously to previously reported dicyanopyridines [5][6][7]46,47], the best-score docking conformations generally observed for the newly synthesized compounds at the hA1AR present the pyridine scaffold being inserted between hydrophobic residues conserved among the human ARs (Phe171, Leu250 6.51 and Ile274 7.39 ) and making non-polar interaction with these amino acids ( Figure 3 reports the binding mode of 9 within the

Molecular Modeling Studies
Analogously to previously reported dicyanopyridines [5][6][7]46,47], the best-score docking conformations generally observed for the newly synthesized compounds at the hA 1 AR present the pyridine scaffold being inserted between hydrophobic residues conserved among the human ARs (Phe171, Leu250 6.51 and Ile274 7.39 ) and making non-polar interaction with these amino acids ( Figure 3 reports the binding mode of 9 within the hA 1 AR cavity). The 2-amino function of 9 makes a double polar interaction with Asn254 6.55 and Glu172. The 3-cyano also binds Asn254 6.55 through a polar interaction, while the 5-cyano moiety points toward the depth of the binding cavity positioning itself close to Ala66 2.61 , Ile69 2.64 , Val87 3.32 and His278 7.43 . The 4-heterocyclic substituent is inserted in the depth of the cavity, close to residues belonging to TM3, EL2, TM5, TM6 and TM7 domains (Val87 3.32 , Leu88 3.33 , Thr91 3.36 , Phe171, Met180 5.38 , Trp247 6.48 , Leu250 6.51 , Ile274 7.39 , Thr277 7.42 and His278 7.43 ). In general, all the AR non-nucleoside agonists reported to date bear an aromatic ring at this position. The affinity data of the compounds synthesized in this work confirm this feature, since the presence of aromatic heterocyclic groups (R 2 ) at the 4-position affects affinity, especially for both the A1 and the A2B subtypes. In particular, as previously observed [2,7], the introduction of a substituent in the para-position of the 4-aromatic ring modulates the binding affinity. In fact, the presence of substituents on the pyridyl moiety (compounds 9 and 10) in a position corresponding to the para one in a phenyl ring, or at the 5-position of a 2furanyl ring (2 and 3), modulates the affinity for the AR subtypes compared to the unsubstituted analogues 1 and 8. Figures 3A and 4A show the interaction of the 2-methoxy-5pyridyl substituent at 4-position of 9 with the receptor residues. The presence of the additional methoxy group allows the compound to completely fill the narrow sub-cavity in the depth of the binding site ( Figure 3A). Since the amino acid residues forming such a sub-cavity are mainly hydrophobic, it is no surprise that the hA1AR affinity data of 9 are comparable to those of 8. When the methoxy group is replaced by a polar hydroxyl function, the affinity gets significantly lower (compound 10). Analogously, the presence of an additional 5-methyl group on a 4-(2-furanyl) substituent maintains high affinity for the hA1AR (derivative 3) compared to the corresponding unsubstituted analogue 1. Replacement of the methyl group with a polar hydroxymethyl function (2) leads to a decrease of the hA1AR affinity.
Similar to what was observed in previously reported docking studies [5][6][7], the 6substituent points toward the entrance of the binding cavity ( Figure 3). As already observed [6], a methylsulfanyl-linker on the 6-side-chain leads to a higher affinity with re- The 4-heterocyclic substituent is inserted in the depth of the cavity, close to residues belonging to TM3, EL2, TM5, TM6 and TM7 domains (Val87 3.32 , Leu88 3.33 , Thr91 3.36 , Phe171, Met180 5.38 , Trp247 6.48 , Leu250 6.51 , Ile274 7.39 , Thr277 7.42 and His278 7.43 ). In general, all the AR non-nucleoside agonists reported to date bear an aromatic ring at this position. The affinity data of the compounds synthesized in this work confirm this feature, since the presence of aromatic heterocyclic groups (R 2 ) at the 4-position affects affinity, especially for both the A 1 and the A 2B subtypes. In particular, as previously observed [2,7], the introduction of a substituent in the para-position of the 4-aromatic ring modulates the binding affinity. In fact, the presence of substituents on the pyridyl moiety (compounds 9 and 10) in a position corresponding to the para one in a phenyl ring, or at the 5-position of a 2-furanyl ring (2 and 3), modulates the affinity for the AR subtypes compared to the unsubstituted analogues 1 and 8. Figures 3A and 4A show the interaction of the 2-methoxy-5-pyridyl substituent at 4-position of 9 with the receptor residues. The presence of the additional methoxy group allows the compound to completely fill the narrow sub-cavity in the depth of the binding site ( Figure 3A). Since the amino acid residues forming such a sub-cavity are mainly hydrophobic, it is no surprise that the hA 1 AR affinity data of 9 are comparable to those of 8. When the methoxy group is replaced by a polar hydroxyl function, the affinity gets significantly lower (compound 10). Analogously, the presence of an additional 5-methyl group on a 4-(2-furanyl) substituent maintains high affinity for the hA 1 AR (derivative 3) compared to the corresponding unsubstituted analogue 1.
Replacement of the methyl group with a polar hydroxymethyl function (2) leads to a decrease of the hA 1 AR affinity.
that, for compounds bearing a longer 4-linker, the imidazolyl moiety gets located in a more external position, leading to a different interaction with the receptor residues. The presence of a 4-methylsulfanyl linker and an imidazole-2yl group (compounds 1 and 5) on the 6-side-chain seems to be correlated to an agonist/partial agonist profile. The replacement of these features respectively with an aminomethyl linker (12) or an imidazole-4-yl moiety (11) causes a shift to antagonist behavior. The docking results do not explain the observed agonist-to-antagonist activity shift as well as the change from an inverse agonist (compound 16) to partial agonist profile when a 1,3-thiazole moiety was included into the 6-side-chain (compound 17). Figure 4C is a top view of the docking pose of 17 within the hA1 receptor cavity. Compounds endowed with hA2AAR affinity showed a trend comparable to that observed for the hA1AR. Docking experiments performed at the hA2AAR crystal structure showed analogous arrangements of the analyzed molecules to those verified at the hA1AR ( Figure 5 reports the binding mode of compound 3 within the hA2AAR cavity). The sets of residues involved in the interaction with the ligand are highly conserved at the two AR subtypes. The dicyanopyridine scaffold is inserted between Phe168, Leu249 6.51 and Ile274 7.39 , with the exocyclic 2-amino function giving polar interaction with Glu169 and Asn253 6.55 . The 4-heterocyclic substituent is positioned close to residues belonging to TM3 (Val84 3.32 , Leu85 3.33 and Thr88 3.36 ), EL2 (Phe168), TM5 (Met177 5.38 ), TM6 (Trp246 6.48 and Leu249 6.51 ) and TM7 (Ile274 7.39 , Ser277 7.42 and His278 7.43 ) domains. These residues are con- Similar to what was observed in previously reported docking studies [5][6][7], the 6substituent points toward the entrance of the binding cavity ( Figure 3). As already observed [6], a methylsulfanyl-linker on the 6-side-chain leads to a higher affinity with respect to the aminomethyl one (compare 1 to 12), probably due to the lack of conjugation effects of the methylsulfanyl-linker (existing instead for the aminomethyl one) that allows this structural feature to better accommodate in the binding site, assuming a non-coplanar position with respect to the dicyanopyridine scaffold. The 1H-imidazol-2yl moiety on the 6-side-chain gets positioned between residues belonging to TM2, EL2 and TM7 domains (Ile69 2.64 , Asn70 2.65 , Glu170, Phe171, Glu172, Tyr271 7.36 and Ile274 7.39 ). The location of this substituent, with respect to the receptor residues, modulates, to some extent, the binding at the A 1 AR. In fact, the presence of a longer 4-linker between this group and the dicyanopyridine scaffold generally leads to a slight decrease of affinity at this subtype with respect to the parent compound 1 (see derivatives 14-17). In this sense, docking results show that, for compounds bearing a longer 4-linker, the imidazolyl moiety gets located in a more external position, leading to a different interaction with the receptor residues. The presence of a 4-methylsulfanyl linker and an imidazole-2yl group (compounds 1 and 5) on the 6-side-chain seems to be correlated to an agonist/partial agonist profile. The replacement of these features respectively with an aminomethyl linker (12) or an imidazole-4-yl moiety (11) causes a shift to antagonist behavior. The docking results do not explain the observed agonist-to-antagonist activity shift as well as the change from an inverse agonist (compound 16) to partial agonist profile when a 1,3-thiazole moiety was included into the 6-side-chain (compound 17). Figure 4C is a top view of the docking pose of 17 within the hA 1 receptor cavity.
Compounds endowed with hA 2A AR affinity showed a trend comparable to that observed for the hA 1 AR. Docking experiments performed at the hA 2A AR crystal structure showed analogous arrangements of the analyzed molecules to those verified at the hA 1 AR ( Figure 5 reports the binding mode of compound 3 within the hA 2A AR cavity). The sets of residues involved in the interaction with the ligand are highly conserved at the two AR subtypes. The dicyanopyridine scaffold is inserted between Phe168, Leu249 6.51 and Ile274 7.39 , with the exocyclic 2-amino function giving polar interaction with Glu169 and Asn253 6.55 . The 4-heterocyclic substituent is positioned close to residues belonging to TM3 (Val84 3.32 , Leu85 3.33 and Thr88 3.36 ), EL2 (Phe168), TM5 (Met177 5.38 ), TM6 (Trp246 6.48 and Leu249 6.51 ) and TM7 (Ile274 7.39 , Ser277 7.42 and His278 7.43 ) domains. These residues are conserved between hA 2A AR and hA 1 AR. Introduction of substituents on the 4-heteroaryl moiety modulates hA 2A AR affinity to a lesser extent than what is observed for the A 1 subtype. However, the hA 2A AR affinity seems generally higher for compounds bearing a 4furan-2yl moiety (compounds 1-3) with respect to those substituted with a 4-pyridyl group. This behavior may be interpreted considering steric factors. A potential slightly smaller cavity at the hA 2A AR, compared to that of the hA 1 AR, could accommodate slightly smaller substituents. Even the results observed for the 6-side-chain are similar to those obtained on the hA 1 AR. The hA 2A AR residues involved in the interaction with the 6-substituent are quite conserved with the hA 1 AR subtype. In detail, the imidazolyl group gets located between residues belonging to the TM2 (Ile66 2.64 and Ser67 2.65 ), EL2 (Leu167, Phe168 and Glu169) and TM7 (Tyr271 7.36 and Ile274 7.39 ) domains. served between hA2AAR and hA1AR. Introduction of substituents on the 4-heteroaryl moiety modulates hA2AAR affinity to a lesser extent than what is observed for the A1 subtype. However, the hA2AAR affinity seems generally higher for compounds bearing a 4-furan-2yl moiety (compounds 1-3) with respect to those substituted with a 4-pyridyl group. This behavior may be interpreted considering steric factors. A potential slightly smaller cavity at the hA2AAR, compared to that of the hA1AR, could accommodate slightly smaller substituents. Even the results observed for the 6-side-chain are similar to those obtained on the hA1AR. The hA2AAR residues involved in the interaction with the 6-substituent are quite conserved with the hA1AR subtype. In detail, the imidazolyl group gets located between residues belonging to the TM2 (Ile66 2.64 and Ser67 2.65 ), EL2 (Leu167, Phe168 and Glu169) and TM7 (Tyr271 7.36 and Ile274 7.39 ) domains. Generally, the introduction of substituents on the exocyclic amine at 2-position of the dicyanopyridine scaffold leads to a decrease of affinity at both AR subtypes with respect to the unsubstituted derivative 1. Only a slight reduction of A1AR affinity was observed for compounds 20 and 21. As previously observed [6], docking results on the herein reported compounds showed that substituents on the 2-amino function lead to a partial disruption of the interaction with the EL2 glutamate and TM6 asparagine residues (see above), with a consequent decrease of affinity. Furthermore, docking results suggested also various binding modes of these compounds at both the AR subtypes, making it difficult to make a clear interpretation of the biological results.

In Vitro Permeation Studies
The stratum corneum is a lipophilic membrane which represents the most important barrier to drug skin diffusion. In vitro permeation studies conducted on compounds 1 and 5 through Franz cells [42] showed an amount of permeated drug after 24 h respectively of 15.0 ± 1.4 µg/mL and 22.7 ± 1.0 µg/mL.
These permeation data were in agreement with the results obtained through solubility studies, carried out on both compounds in testing medium (PBS + Tween 80 (2% w/w)). On the basis of these experiments, compound 5 resulted in being less soluble (27.3 ± 2.52 Generally, the introduction of substituents on the exocyclic amine at 2-position of the dicyanopyridine scaffold leads to a decrease of affinity at both AR subtypes with respect to the unsubstituted derivative 1. Only a slight reduction of A 1 AR affinity was observed for compounds 20 and 21. As previously observed [6], docking results on the herein reported compounds showed that substituents on the 2-amino function lead to a partial disruption of the interaction with the EL2 glutamate and TM6 asparagine residues (see above), with a consequent decrease of affinity. Furthermore, docking results suggested also various binding modes of these compounds at both the AR subtypes, making it difficult to make a clear interpretation of the biological results.

In Vitro Permeation Studies
The stratum corneum is a lipophilic membrane which represents the most important barrier to drug skin diffusion. In vitro permeation studies conducted on compounds 1 and 5 through Franz cells [42] showed an amount of permeated drug after 24 h respectively of 15.0 ± 1.4 µg/mL and 22.7 ± 1.0 µg/mL.
These permeation data were in agreement with the results obtained through solubility studies, carried out on both compounds in testing medium (PBS + Tween 80 (2% w/w)). On the basis of these experiments, compound 5 resulted in being less soluble (27.3 ± 2.52 µg/mL) than the analogous 1 (40.6 ± 0.16 µg/mL).
Therefore, it can be hypothesized that the dicyanopyridine 5, due to its higher lipophilicity with respect to 1 (pkCSM calculated logP, 3.15 versus 2.68, respectively) [48,49] and, consequently, its affinity for stratum corneum, accumulates inside the membrane, generating an into/out concentration gradient, which represents the driving force to its permeation.
From the perspective of a possible application of the two compounds in wound healing [50] as promoters of endothelial cell proliferation and migration [21,22], the lipophilicity of compound 5 could be a desirable property, allowing an accumulation of the compound on the injured skin leading to a limited systemic absorption.

General Methods
The microwave-assisted syntheses were performed by using an Initiator EXP Microwave Biotage instrument (frequency of irradiation: 2.45 GHz). Analytical silica gel plates (Merck F254, Kenilworth, NJ, USA), preparative silica gel plates (Merck F254, 2 mm) and silica gel 60 (Merck, 70-230 mesh) were used for analytical and preparative TLC and for column chromatography, respectively. All melting points were determined on a Gallenkamp melting-point apparatus and are uncorrected. Elemental analyses were performed with a Flash E1112 Thermofinnigan elemental analyzer for C, H and N, and the results were within ±0.4% of the theoretical values. All final compounds revealed purity not less than 95%. The IR spectra were recorded with a Perkin-Elmer Spectrum RX I spectrometer in Nujol mulls and are expressed in cm −1 . NMR spectra were recorded on a Bruker Avance 400 spectrometer (400 MHz for 1 H NMR and 100 MHz for 13 C NMR). The chemical shifts are reported in δ (ppm) and are relative to the central peak of the residual non-deuterated solvent, which was CDCl 3 or DMSOd 6 . The following abbreviations are used: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad, Ar = aromatic protons. Compounds 22 [30], 25, 26 [30], 27 [31], 29 [32], 32 [27], 35, 36 [27], 37 [28], 39 [29], 51, 52 and 56 were synthesized by following the procedures reported by us in Reference [6]. When available, melting point and/or 1 H NMR values were in accordance with the literature data. Equimolar amounts of sodium hydrogen carbonate and commercially available 2-(bromomethyl)-1H-imidazole hydrobromide (4.0 mmol) were consequentially added to a solution of the mercapto-compound (32 [6,27] (2) Sodium hydrogen carbonate (9.3 mmol) and an equimolar amount of the commercially available 2-(bromomethyl)-1H-imidazole hydrobromide were added to a solution of the suitable mercapto-compound (33, 9.3 mmol) in anhydrous DMF (3 mL). The reaction mixture was stirred at RT, in a nitrogen atmosphere, until the disappearance of the starting material (TLC monitoring). Then water was added (20 mL) to precipitate a solid which was collected by filtration and washed with water. For compound 1, the crude product was treated with Et 2 O (2 mL), collected by filtration and recrystallized. For derivative 2, a second crop of product was obtained by extracting the aqueous solution with EtOAc (4 × 10 mL). The collected organic layers were dried (Na 2 SO 4 ), and the solvent was removed under reduced pressure. The two crops of product were triturated with Et 2 O (2 mL) collected by filtration and purified by crystallization. Sodium hydrogen carbonate (2 mmol) and the commercially available 2-(bromomethyl)-1H-imidazole hydrobromide (1 mmol) were added to a solution of the suitable mercaptocompound (35-41 [6,27-29] 1 mmol) in anhydrous DMF (1 mL). The reaction mixture was stirred at RT until disappearance of the starting material (TLC monitoring). Then, water was added (25 mL) to precipitate a solid which was collected by filtration and washed with water. The crude product was triturated with Et 2 O (5 mL), collected by filtration and purified by crystallization (compounds 4-9) or silica gel column chromatography, eluting system cyclohexane/EtOAc/MeOH 2:6:2 (compound 10). in anhydrous DMF (2 mL). The reaction mixture was stirred at RT, under a nitrogen atmosphere, for 3 h. After dilution with cold water (30 mL), a solid precipitated and was collected by filtration and washed with water and Et 2 O. A second crop of product was obtained by extracting the aqueous phase with EtOAc (3 × 30 mL). Then the collected organic layers were anhydrified (Na 2 SO 4 ) and evaporated under reduced pressure to yield a solid which was treated with Et 2 O (5 mL) and then collected by filtration. The crude product was purified by silica gel column chromatography, eluting system CH 2 Cl 2 /MeOH 8:2. Sodium hydrogen carbonate (5.5 mmol) and an equimolar amount of the commercially available 2-(bromomethyl)-1H-imidazole hydrobromide were added to a solution of the suitable mercapto-compound (48-50, 52 [5], 5.0 mmol) in anhydrous DMF (2 mL). The reaction mixture was stirred at RT, a under nitrogen atmosphere, until the disappearance of the starting material (TLC monitoring). Then, water was added (30 mL) to precipitate a solid which was collected by filtration and washed with water. A second crop of product was obtained by extracting the aqueous solution with EtOAc (3 × 10 mL). The collected organic layers were dried (Na 2 SO 4 ) and the solvent removed under reduced pressure. The two crops of product were triturated with Et 2 O (2 mL), collected by filtration and purified by crystallization.  13 13 13 (53) A solution of commercially available 1H-imidazol-4-ylmethanol (8.15 mmol) in a large excess of thionyl chloride (8 mL) was heated at reflux for 3 h. After evaporation under reduced pressure of the excess of the reagent, cyclohexane (10 mL ×2) was added to the residue and then removed under reduced pressure to yield the crude product. The latter was then treated with diethyl ether (5 mL The cAMP levels were then quantified by using the AlphaScreen cAMP Detection Kit (Perkin Elmer, Boston, MA, USA), following the manufacturer's instructions [54]. At the end of the experiments, the plates were read with a Perkin Elmer EnSight Multimode Plate Reader.

Data Analysis
The protein concentration was determined according to a Bio-Rad method, with bovine albumin as a standard reference. Inhibitory binding constant (K i ) values were calculated from those of IC 50 according to the Cheng-Prusoff equation, K i = IC 50 /(1 + [C*]/KD*), where [C*] is the concentration of the radioligand, and KD* its dissociation constant [53]. K i and IC 50 values were calculated by non-linear regression analysis, using the equation for a sigmoid concentration-response curve (Graph-PAD Prism, San Diego, CA, USA).

Ab Initio Quantum Mechanical Studies
The ab initio quantum mechanical calculations were performed by using the program GAMESS (General Atomic and Molecular Electronic Structure System) [37], a general ab initio quantum chemistry package which is maintained by the members of the Gordon research group (Department. of Chemistry) at Iowa State University, USA, which uses an RHF (Restricted Hartree-Fock) calculation and whose basic function for the description of atomic orbitals is average accuracy (basis set 3-21 G) [37, 55,56]

Molecular Modeling
Receptor refinement and energy minimization tasks were carried out by using Molecular Operating Environment (MOE, version 2019.01) suite [41]. Docking experiments were performed with CCDC Gold [40].
A 1 AR and A 2A AR crystal structures refinement: The recently published cryo-EM structure of the h A1 AR and the crystal structure of the human A 2A AR in complex with adenosine and ZM241385, respectively, were downloaded by the Protein Data Bank webpage (http://www.rcsb.org; accessed on 6 April 2012, pdb code, 6D9H, with 3.6-Å resolution [38]; and pdb code, 4EIY, with 1.8-Å resolution [39], respectively). Both structures were checked within MOE and corrected by restoring missing loops and the wild-type receptor sequences and by adding hydrogen atoms. The Homology Modeling tool of MOE was used for these tasks. The protein structures were then energetically minimized with MOE, using the AMBER99 force field, until the RMS gradient of the potential energy was less than 0.05 kJ mol-1 Å-1. The reliability and quality of the models were checked by using the Protein Geometry Monitor application within MOE.
Molecular docking analysis: Docking analyses were performed by using CCDC Gold [40], with default efficiency settings through MOE interface, by selecting Chem-Score as scoring function and 50 poses to be generated for each ligand. Each docking pose was then energetically minimized within the respective receptor target within MOE by keeping fixed the receptor coordinates. The minimized docking poses were then re-scored by using ChemScore as the scoring function.
(2 mL) of compounds solution (PBS pH 7.4 and Tween 80 2% w/w) was placed in the donor compartment. At predetermined time intervals (1, 2, 3, 4, 5, 6 and 24 h), 0.5 mL samples were withdrawn from the receiving chamber, and the drug concentration was assayed by HPLC. A correction for the cumulative dilution, due to the sample replacement with an equal volume of fresh medium, was calculated. All the experiments were performed in triplicate.

Conclusions
This study has produced a new set of amino-3,5-dicyanopyridines as AR ligands which were synthesized to deepen the SARs of this versatile series. With this in mind, molecular modeling studies supported the rationalization of the biological results obtained on this set. In general, the target compounds interacted better with both the hA1 and A2A subtypes than with the other ARs. However, compounds 1 and 5 emerged as pan ligands by binding all the subtypes with similar binding affinity in the nanomolar range. This interesting behavior, together with their partial agonist profile, suggested the need to evaluate them for their potential use in wound healing. Preliminary results on their permeation capability through artificial membrane, simulating the epidermal barrier, indicated that compound 5 could be a candidate for further evaluation as a promoter of skin-wound healing.