Systematic Modification of the Substitution Pattern of the 7-Hydroxy-5-oxopyrazolo[4,3-b]pyridine-6-carboxamide Scaffold Enabled the Discovery of New Ligands with High Affinity and Selectivity for the Cannabinoid Type 2 Receptor

Selective ligands of the CB2 receptor are receiving considerable attention due to their potential as therapeutic agents for a variety of diseases. Recently, 7-hydroxy-5-oxopyrazolo[4,3-b]pyridine-6-carboxamide derivatives were shown to act at the CB2 receptor either as agonists or as inverse agonists/antagonists in vitro and to have anti-osteoarthritic activity in vivo. In this article, we report the synthesis, pharmacological profile, and molecular modeling of a series of twenty-three new 7-hydroxy-5-oxopyrazolo[4,3-b]pyridine-6-carboxamides with the aim of further developing this new class of selective CB2 ligands. In addition to these compounds, seven other analogs that had been previously synthesized were included in this study to better define the structure–activity relationship (SAR). Ten of the new compounds studied were found to be potent and selective ligands of the CB2 receptor, with Ki values ranging from 48.46 to 0.45 nM and CB1/CB2 selectivity indices (SI) ranging from >206 to >4739. In particular, compounds 54 and 55 were found to be high-affinity CB2 inverse agonists that were not active at all at the CB1 receptor, whereas 57 acted as an agonist. The functional activity profile of the compounds within this structural class depends mainly on the substitution pattern of the pyrazole ring.


Introduction
Primarily responsible for keeping body homeostasis, the endocannabinoid system (ECS) is a complex ubiquitous lipid signaling system consisting of two major G-proteincoupled receptors, cannabinoid-1 (CB1) and cannabinoid-2 (CB2) receptors, their endogenous ligands anandamide and 2-arachidonoylglycerol (2-AG), and the anabolic and catabolic enzymes responsible for their synthesis and degradation [1]. Although only CB1R and CB2R are widely recognized as CBRs, many other receptors, from other GPCRs to ion channels and nuclear receptors, have been reported to interact with cannabinoids [2]. This pleiotropic system plays a critical role in regulating a variety of physiological processes and functions, including mood regulation, nociception, cognitive functions, motivation Following these results, a more detailed investigation of pyrazolo [4,3-b]pyridine compounds was carried out by extending the structure-activity relationship (SAR). Starting from structure 3, a small library of 23 compounds was created by changing three parts of the molecule, namely, the amide substituent, the chain at position 4, and the substitution pattern of the pyrazole ring, as well as their combinations. In particular, the design strategy for the new compounds, summarized in Figure 2, was aimed at (i). investigating the effects on activity associated with less bulky aliphatic or planar aromatic amide substituents, chains at N4 with a smaller number of rotatable bonds; (ii). the introduction of substituents, also different from methyl, at other positions on the pyrazole ring; and (iii). the evaluation of their solubility in water after conversion to salts with monovalent cations. The results of these studies are reported herein.

Synthesis
Most of the new compounds were prepared by synthetic methods that we have already described for similar compounds [34].
Briefly, a first set of 15 compounds showing modifications either at the amide moiety or at the N4 chain or at both positions was prepared starting from methyl 4-amino-2methylpyrazole-3-carboxylate (4) [35] (Scheme 1), which was N-alkylated by reaction with appropriate alkyl iodides or by reductive alkylation with aromatic aldehydes to give intermediates 5-8, which were then N-acylated with methyl malonyl chloride to give bicyclic precursors 9-12. These were converted by Dieckmann condensation into methyl 4,5-dihydro-7-hydroxy-5-oxo-2H-pyrazolo [4,3-b]pyridine-6-carboxylates, which were directly subjected to amidation reaction with the corresponding amines without purification and/or characterization to afford the final compounds 13-27.  Following these results, a more detailed investigation of pyrazolo [4,3-b]pyridine compounds was carried out by extending the structure-activity relationship (SAR). Starting from structure 3, a small library of 23 compounds was created by changing three parts of the molecule, namely, the amide substituent, the chain at position 4, and the substitution pattern of the pyrazole ring, as well as their combinations. In particular, the design strategy for the new compounds, summarized in Figure 2, was aimed at (i). investigating the effects on activity associated with less bulky aliphatic or planar aromatic amide substituents, chains at N4 with a smaller number of rotatable bonds; (ii). the introduction of substituents, also different from methyl, at other positions on the pyrazole ring; and (iii). the evaluation of their solubility in water after conversion to salts with monovalent cations. The results of these studies are reported herein.

Synthesis
Most of the new compounds were prepared by synthetic methods that we have already described for similar compounds [34].
Briefly, a first set of 15 compounds showing modifications either at the amide moiety or at the N4 chain or at both positions was prepared starting from methyl 4-amino-2methylpyrazole-3-carboxylate (4) [35] (Scheme 1), which was N-alkylated by reaction with appropriate alkyl iodides or by reductive alkylation with aromatic aldehydes to give intermediates 5-8, which were then N-acylated with methyl malonyl chloride to give bicyclic precursors 9-12. These were converted by Dieckmann condensation into methyl 4,5-dihydro-7-hydroxy-5-oxo-2H-pyrazolo [4,3-b]pyridine-6-carboxylates, which were directly subjected to amidation reaction with the corresponding amines without purification and/or characterization to afford the final compounds 13-27.

Synthesis
Most of the new compounds were prepared by synthetic methods that we have already described for similar compounds [34].
Briefly, a first set of 15 compounds showing modifications either at the amide moiety or at the N4 chain or at both positions was prepared starting from methyl 4-amino-2methylpyrazole-3-carboxylate (4) [35] (Scheme 1), which was N-alkylated by reaction with appropriate alkyl iodides or by reductive alkylation with aromatic aldehydes to give intermediates 5-8, which were then N-acylated with methyl malonyl chloride to give bicyclic precursors 9-12. These were converted by Dieckmann condensation into methyl 4,5-dihydro-7-hydroxy-5-oxo-2H-pyrazolo [4,3-b]pyridine-6-carboxylates, which were directly subjected to amidation reaction with the corresponding amines without purification and/or characterization to afford the final compounds 13-27.
Following the same procedure, the aminopyrazole derivative 28 [36] (Scheme 2) was converted by N-alkylation to intermediates 29, 30. Reaction with methyl malonyl chloride gave the N-acylated products 31, 32, which, in turn, were cyclized to 4,5-dihydro-7-hydroxy-5-oxo-2H-pyrazolo [4,3-b]pyridine-6-carboxylates. Subsequent reaction with the appropriate amines led to the final compounds 33-35, which carried modifications of the amide or chain substituents, in all cases linked to a group other than methyl at the N2 position. Following the same procedure, the aminopyrazole derivative 28 [36] (Scheme 2) was converted by N-alkylation to intermediates 29, 30. Reaction with methyl malonyl chloride gave the N-acylated products 31, 32, which, in turn, were cyclized to 4,5-dihydro-7hydroxy-5-oxo-2H-pyrazolo [4,3-b]pyridine-6-carboxylates. Subsequent reaction with the appropriate amines led to the final compounds 33-35, which carried modifications of the amide or chain substituents, in all cases linked to a group other than methyl at the N2 position. With the aim of deepening the understanding of SAR in the class of 7-hydroxy-3methyl-5-oxopyrazole [4,3-b]pyridine-6-carboxamide compounds (as 3), we set out to investigate the previously unexplored chemical space around position 3 by replacing the hydrogen atom with a methyl group. To this end, compound 36 (Scheme 3), intended as the starting point for the synthesis, was prepared according to a previous literature report [37], but then subjected to N-methylation according to a different and more convenient approach, since we found that treatment at room temperature with dimethyl sulfate in the presence of potassium carbonate in acetone resulted in obtaining the two isomeric products 37 and 38 in the same yield of 49% each after separation of the reaction mixture by flash chromatography. The structure of both regioisomers 37 and 38 was elucidated using 1 H NMR data. These nitro derivatives were reduced with ammonium formate in the Following the same procedure, the aminopyrazole derivative 28 [36] (Scheme 2) was converted by N-alkylation to intermediates 29, 30. Reaction with methyl malonyl chloride gave the N-acylated products 31, 32, which, in turn, were cyclized to 4,5-dihydro-7hydroxy-5-oxo-2H-pyrazolo [4,3-b]pyridine-6-carboxylates. Subsequent reaction with the appropriate amines led to the final compounds 33-35, which carried modifications of the amide or chain substituents, in all cases linked to a group other than methyl at the N2 position. With the aim of deepening the understanding of SAR in the class of 7-hydroxy-3methyl-5-oxopyrazole [4,3-b]pyridine-6-carboxamide compounds (as 3), we set out to investigate the previously unexplored chemical space around position 3 by replacing the hydrogen atom with a methyl group. To this end, compound 36 (Scheme 3), intended as the starting point for the synthesis, was prepared according to a previous literature report [37], but then subjected to N-methylation according to a different and more convenient approach, since we found that treatment at room temperature with dimethyl sulfate in the presence of potassium carbonate in acetone resulted in obtaining the two isomeric products 37 and 38 in the same yield of 49% each after separation of the reaction mixture by flash chromatography. The structure of both regioisomers 37 and 38 was elucidated using 1 H NMR data. These nitro derivatives were reduced with ammonium formate in the Scheme 2. Synthesis of the final compounds 33-35. Reagents and conditions: (a) 1-iodoethane or 1-iodobutane, DMF, 50 • C, 7 h; (b) methyl malonyl chloride, TEA, dry DCM, 0-5 • C, 2-6 h; (c) NaH, THF, MeOH (cat.), 60 • C, 2 h; (d) appropriate amine, toluene, reflux, 1-2 h.
With the aim of deepening the understanding of SAR in the class of 7-hydroxy-3-methyl-5-oxopyrazole [4,3-b]pyridine-6-carboxamide compounds (as 3), we set out to investigate the previously unexplored chemical space around position 3 by replacing the hydrogen atom with a methyl group. To this end, compound 36 (Scheme 3), intended as the starting point for the synthesis, was prepared according to a previous literature report [37], but then subjected to N-methylation according to a different and more convenient approach, since we found that treatment at room temperature with dimethyl sulfate in the presence of potassium carbonate in acetone resulted in obtaining the two isomeric products 37 and 38 in the same yield of 49% each after separation of the reaction mixture by flash chromatography. The structure of both regioisomers 37 and 38 was elucidated using 1 H NMR data. These nitro derivatives were reduced with ammonium formate in the presence of 10% Pd/C to give the amino compounds 39 and 40, which, in turn, could be processed in parallel (Scheme 4) by the same method shown in Schemes 1 and 2. Thus, 39 and 40 were alkylated to give the secondary amine derivatives 41-45, which, after amidation to give 46-50 with methyl malonyl chloride, underwent Dieckmann cyclization to yield the bicyclic ester compounds, whose direct treatment with the appropriate amines led to the final compounds 51-55. However, while compounds 51 and 52 were obtained in two steps with a moderate yield of 45-47%, compounds 53-55 were obtained only with a low yield of 5-20%. This difference in behavior could be justified by the fact that in compounds 53-55, three substituents (i.e., the two methyl groups and the aliphatic chain in position 4) are in close proximity to each other and in a coplanar arrangement, leading to repulsive steric interactions. Therefore, the Dieckmann cyclization reaction is thermodynamically less favored when applied to substrates 48-50 than to substrates 46 and 47. X-ray crystallographic analysis [38,39] of compound 51 (see Supplementary Information, Figure S1: X-ray crystal structure of compound 51) confirmed that the structural assignment performed by 1 H NMR was correct.
Finally, a series of compounds characterized by an adamantylamide and an isopropyl chain at position 4, but having different substituents at positions 1 or 2 of the pyrazole ring, was synthesized by a modified procedure recently described by us [40]. This synthetic approach has allowed the chemical diversity around the pyrazole moiety to increase by introducing substituents at positions 1 or 2 as the final step of the synthesis. The structure of compounds 56-62 is highlighted in Table 1 polysaccharides for hydrogel preparation have advantages such as bioactivity toxicity, biocompatibility, and similarity of structure and function to the glyca extracellular matrix [12,13]. Pectin is a gelling heteropolysaccharide isolated from h plants that is widely used in several scientific fields [14,15]. Pectins include a di group of polyanionic polysaccharides whose backbone is formed by 1,4-linked galacturonic acid (GalA) residues that can be partly methyl esterified. Pectins with degree of methyl esterification form a gel in the presence of multivalent cations in a pH range, whereas high-methyl-esterified pectins form physical gels at pH < 3.5 i presence of co-solutes [16]. Biodegradability, biocompatibility, drug loading/rele capacity, and tunable mechanical properties contribute to the application of pectin as promising materials for biomedicine [17]. Pectin has been found to attach t glycocalyx of visceral mesothelium [18] and was proposed as a serosal bioadhesiv pleural air leaks [19,20,21] and a sealant for anastomotic healing after intestinal su [22]. Recently, our group showed that a hydrogel made of ionically cross-linked pectin adhered to rat intestinal serosa ex vivo with a strength that depended on the of cross-linking cation [23]. However, the serosal adhesion and strength of the ionot pectin gel were found to decrease significantly after incubation in physiological m [23,24,25]. Improvement of adhesivity and strengthening of the pectin gel net appear to be necessary to obtain promising hydrogel adhesives to treat anasto leakage.
Chitosan is a natural polymer derived by the partial deacetylation in an alk solution of chitin, which is the main component of the exoskeleton of crabs, shrimp krill. It is a linear amino-polysaccharide of β (1-4)-linked D-glucosamine and N-acet glucosamine residues and has similarity to glycosaminoglycans of the extrace matrix [26]. Chitosan is used as a hydrogel in many successful applications withi biomedical field, including wound dressing, drug controlled-release systems, engineering, and regenerative medicine [26,27,28]. The use of chitosan as a bioadh substance is one area of considerable attention. The ability to adhere to the skin mucous membranes are the advantages of chitosan for the development of dre materials and targeted drug delivery, respectively [29,30,31,32]. The cationic behav chitosan in acidic solutions and the induction of gelation reactions mediate interac Thus, 39 and 40 were alkylated to give the secondary amine derivatives 41-45, which, after amidation to give 46-50 with methyl malonyl chloride, underwent Dieckmann cyclization to yield the bicyclic ester compounds, whose direct treatment with the appropriate amines led to the final compounds 51-55. However, while compounds 51 and 52 were obtained in two steps with a moderate yield of 45-47%, compounds 53-55 were obtained only with a low yield of 5-20%. This difference in behavior could be justified by the fact that in compounds 53-55, three substituents (i.e., the two methyl groups and the aliphatic chain in position 4) are in close proximity to each other and in a coplanar arrangement, leading to repulsive steric interactions. Therefore, the Dieckmann cyclization reaction is thermodynamically less favored when applied to substrates 48-50 than to substrates 46 and 47. X-ray crystallographic analysis [38,39] of compound 51 (see Supplementary Information, Figure S1: X-ray crystal structure of compound 51) confirmed that the structural assignment performed by 1 H NMR was correct.
Finally, a series of compounds characterized by an adamantylamide and an isopropyl chain at position 4, but having different substituents at positions 1 or 2 of the pyrazole ring, was synthesized by a modified procedure recently described by us [40]. This synthetic approach has allowed the chemical diversity around the pyrazole moiety to increase by introducing substituents at positions 1 or 2 as the final step of the synthesis. The structure of compounds 56-62 is highlighted in Table 1, along with the structure of all the other final pyrazolo [4,3-b]pyridine-6-carboxamide derivatives that have been the subject of subsequent pharmacological investigation. the other final pyrazolo [4,3-b]pyridine-6-carboxamide derivatives that have been the subject of subsequent pharmacological investigation. the other final pyrazolo [4,3-b]pyridine-6-carboxamide derivatives that have been the subject of subsequent pharmacological investigation. the other final pyrazolo [4,3-b]pyridine-6-carboxamide derivatives that have been the subject of subsequent pharmacological investigation. the other final pyrazolo [4,3-b]pyridine-6-carboxamide derivatives that have been the subject of subsequent pharmacological investigation.

Estimated Drug-like Properties
There is a growing interest in making the drug discovery process more effective by incorporating the most favorable physicochemical properties (e.g., molecular weight, solubility, polar surface area, ClogP, and others) early in the design and evaluation of new chemical entities to increase the likelihood of further preclinical/clinical development [41,42]. Among all synthesized compounds, we selected those (15, 18, 26, 35, 54, and 55) with higher receptor affinity, i.e., compounds with K i CB2 < 20 nM (see Table 1). Their physicochemical and pharmacokinetic parameters were evaluated using the free web tool Swis-sADME [43,44]. All compounds showed a very favorable overall profile, as they did not violate Lipinski's "rule of five" in any way and showed no potential as PAINS (see Supplementary Information, Table S1. Calculated physicochemical and drug-like properties of selected compounds). The same compounds were also characterized by a multiparameter score LLE (Liphophilic Ligand Efficiency) within the optimal range for suitable drug candidates (5 < LLE < 7) [45,46], apart from 18 (LLE = 4.90), since the reduction in overall lipophilicity, as predicted by CLogP values, did not affect the CB2R affinity values (pK i CB2).

Water Solubility Evaluation
The results of drug discovery and development can be significantly affected by the poor water solubility of pharmaceutical compounds. The solubility of the drug is mainly influenced by its chemical disposition and the conditions of its solution. The molecular assembly defines the molecular volume, crystal energy, hydrogen bonding, ionizability, and lipophilicity that determine the solubility of the drug [47]. 7-Hydroxy-5-oxopyrazolo[4,3-b]pyridine-6-carboxamide derivatives typically have pKa values in the range of 4.2-5.0 [48], and their salts are expected to be sufficiently soluble in water at physiological pH, while the tautomeric equilibrium in the neutral species (Figure 3a) results in the formation of a strong intramolecular hydrogen bonding network that likely reduces its solubility. and lipophilicity that determine the solubility of the drug [47]. 7-Hydroxy-5-oxopyrazolo[4,3-b]pyridine-6-carboxamide derivatives typically have pKa values in the range of 4.2-5.0 [48], and their salts are expected to be sufficiently soluble in water at physiological pH, while the tautomeric equilibrium in the neutral species (Figure 3a) results in the formation of a strong intramolecular hydrogen bonding network that likely reduces its solubility. Therefore, despite their lower lipophilicity compared to type 1 compounds ( Figure  1), the thermodynamic water solubility of type 3 compounds as well as their dissolution rate is expected to be poor/moderate and a limiting factor for their evaluation in in vitro/in vivo tests. Therefore, we decided to investigate the dissolution of different salts (Na + , K + , Cs + , and Me4N + ) of 63 (Table 1), representative of the whole class of compounds tested, testing their solubility in different EtOH/water mixtures. The 63-cesium salt (85.7 mg in 1 Therefore, despite their lower lipophilicity compared to type 1 compounds (Figure 1), the thermodynamic water solubility of type 3 compounds as well as their dissolution rate is expected to be poor/moderate and a limiting factor for their evaluation in in vitro/in vivo tests. Therefore, we decided to investigate the dissolution of different salts (Na + , K + , Cs + , and Me 4 N + ) of 63 (Table 1), representative of the whole class of compounds tested, testing their solubility in different EtOH/water mixtures. The 63-cesium salt (85.7 mg in 1 mL) proved soluble in a 1:20 EtOH/water mixture, which is approximately equivalent to a 6% v/v solution of EtOH in water. The potassium salt was soluble in a 1:4 EtOH/water mixture, while the sodium salt began to precipitate once the 1:1 ratio of the same mixture was reached. The tetramethylammonium salt was not soluble at all, probably due to its organic nature. These results confirmed the idea that cations with progressively larger ionic radius (sodium < potassium < cesium) can lead to progressively less associated ion pairs with the 63-enolate and, as expected, 63-salts with larger cations showed increased solubility in solvent mixtures with low EtOH content, where salt dissociation predominates and the neutral form is not solubilized by the organic co-solvent.
Moreover, the solubility of the 63-sodium salt in physiological 0.9% NaCl solution, which is the delivery vehicle for in vivo administration, is strongly affected by the action of the common ion (Na + ), resulting in a further reduction in solubility. Plotting the concentration (µM) at different time points of dissolved 63, added to a physiological solution as an excess of the sodium salt (Figure 3b), we observed relatively constant values with an average of 6.37 × 10 −6 mol/L. Considering the concentration of sodium ions (154 mM), the solubility product constant for the sodium salt results: Ks = [Na + ] [63 − ] = 9.81 × 10 −7 mol 2 /L 2 . When added to the same solution, 63-cesium salt initially showed higher concentrations of dissolved 63, which gradually decreased reaching, after 200 min, the value consistent with the Ks of the sodium salt and the excess of sodium ions. However, starting from the cesium salt, the concentration of drug dissolved in the physiological solution maintained high values for at least one hour (around 50 × 10 −6 mol/L, with a peak higher than 100 × 10 −6 mol/L) and, for a further two hours, these values were higher than those obtained with the sodium salt. From this, it can be deduced that the 63-cesium salt would be very suitable for the extemporaneous preparation of injectable solutions in the presence of high concentrations of sodium ions.

In Vitro Pharmacology
The binding affinities (K i values) of the new compounds for human recombinant CB1R and CB2R are reported in Table 1. The compounds were evaluated in parallel with compounds 63 and 3 already reported [34] and SR144528 as reference CB2R ligand, as previously described ( [34] and references cited therein). Among the thirty compounds tested, fourteen of them displayed significant binding affinity for CB2R, with K i values < 200 nM, while twelve of these proved to be potent CB2R ligands with K i values in the range 0.45-61.76 nM. Only compounds 26 and 35 showed also significant affinity for CB1R, with K i values of 115.63 and 102.30 nM, respectively. Nevertheless, all the compounds with affinity for the CB2R had selectivity index values between >70 and >4739.
To investigate the effects of key substitutions at the pyrazolo[4,3-b]pyridine scaffold on functional activity at CB2 receptors, the ability of representative high-affinity and selective compounds (15, 18, 26, 35, 55, and 57) to modulate intracellular cAMP levels in NKH-477-stimulated CHO cells-optimized to overexpress CB2R-was measured using enzyme fragment complementation technology. As shown in Figure 4, only compound 57 activated the receptor with typical agonist behavior (EC 50 = 1.59 µM, Emax = 84.05) by reducing cAMP levels induced by NKH-477 (a water-soluble analog of forskolin) as expected for a Gi protein-coupled receptor agonist. All the other compounds showed a classical antagonist/inverse agonist behavior as suggested by their capability of antagonizing the CB2-ligand challenge (4 µM, JWH-133) and, in the case of 15 and 18, of increasing the level of NKH-477-induced cAMP (Figure 4).

Structure-Activity Relationship (SAR)
A crucial effect on receptor affinity was demonstrated by modifications in the amide function (Table 1). While the rest of the structure remained essentially unchanged, compounds 13, 14, and 17, in which the N-adamantyl group was replaced by aliphatic groups or groups containing an aromatic ring, were without activity up to a concentration of 10,000 nM. However, the introduction of benzhydryl, tetrahydronaphthalene-1-yl, and cyclooctyl groups resulted in compounds 19, 18, and 15, respectively, showing gradually decreasing K i values for the CB2 receptor, with the last two compounds (especially 15) proving to be potent and selective ligands. Apparently, the cyclooctyl system is able to reproduce quite well the properties of an adamantyl group in terms of lipophilicity and steric requirements, while this result cannot be achieved when smaller aliphatic groups (compounds 13 and 14) or an extended aromatic system (e.g., 19) are inserted at the amide nitrogen. However, any modification of the amide moiety appears to be completely ineffective in providing pharmacological activity when combined with the insertion of a short aliphatic chain (namely, isopropyl) at the N4 position, as in the case of compounds 21-25, which are completely inactive even when the adamantyl group is retained (e. g., 21 and 22). Surprisingly, the presence of a benzyl group at N4 makes compound 26 one of the most potent CB2 ligands among the 30 compounds studied, with also nonnegligible selectivity toward the CB1 receptor (SI = 241). Such a striking effect on pharmacological activity, achieved by replacing an n-butyl chain with a benzyl group, might be the consequence of a moderate increase in lipophilicity and a concomitant certain decrease in the conformational freedom of the molecule. On the other hand, caution should be exercised in interpreting this experimental result because the introduction of a 2,4-dimethoxybenzyl group in place of the simple benzyl leads to the completely inactive compound 27.
The application of two simultaneous structural modifications (compounds 33-35) gave positive results. For example, 34 with an ethyl group at N2 and a tert-butylamide showed an affinity for the CB2 receptor increased by two orders of magnitude compared to compound 13, whereas in the case of 33, with an ethyl substituent at both the N2 and N4 positions, the affinity decreased by an order of magnitude but was accompanied by a significant selectivity advantage over 63. The third compound in this subgroup, namely, 35, which differs from compound 63 in having an ethyl instead of a methyl at N2, resulted in even better binding to both CB1 and CB2 receptors. Of all the compounds tested, compound 35 can be considered the best in terms of CB2 affinity. Comparing its in vitro activity with that of 3 and 63, compound 35 actually falls between 3 and 63 in terms of CB1 and CB2 affinity. The lengthening of the N2 substituent in 35 from one to two carbon atoms seems to partially compensate for the shortening of the N4 chain by one carbon atom during the transition from 3 to 63.
Very interesting results were obtained with the introduction of another methyl group in position 3 of the pyrazole ring. Although the length (four or five carbon atoms) of the linear chain in the N4 position seems to be irrelevant, the relative position of the two methyl groups has a very significant effect. While the simultaneous presence of two methyl groups at N1/C3 (compounds 51 and 52) produces ligands that are sufficiently potent but much more selective than the analogous compounds methylated only at the N1 position [34], the combination of two methyl groups at N2/C3 (compounds 54 and 55) maintains affinity for the CB2 receptor (cf. 63), but completely abolishes the ability to interact with the CB1 receptor, with the respective SI values exceeding 4600. The last compound 53, which has two N2/C3 methyl groups and the isopropyl chain in the N4 position at the same time, proved to be inactive, as expected based on the results described above for compounds 21-25. Nevertheless, it should be borne in mind that the overall pharmacological profile of a given molecule is probably due to the combined effects of the various substituents present rather than to the action of any single one of them. This is indeed evident in the case of the subgroup of compounds 56-62: all N4-isopropyl-substituted compounds that do not have substituents in the N1 position of the pyrazole ring (namely, 56, 59, 61, and 62) are inactive, as expected. In contrast, the compounds (57, 58, and 60) that are N1-substituted with a methyl, propargyl, or cyanomethyl group, respectively, all have pharmacological activity that is far from negligible and comparable among themselves.
The results of the functional activity assay are consistent with the presence of a substituent on the pyrazole ring at the N1 position (for agonists) or at the N2 position (for antagonists), independently of its nature (methyl, ethyl, or other chains). The introduction of a second methyl group into the pyrazole ring (compound 55) or even a significant change at N4 of the pyrazolo[4,3-b]pyridine system (compound 26) confirmed or enhanced the antagonistic effect as suggested by the increased levels of NKH-477-induced cAMP upon a challenge with an EC 80 concentration of the CB2 agonist JWH-133. Similarly, we observed that compounds 15 and 18, whose adamantyl group was replaced by different cycles, were still able to antagonize JWH-133-induced inhibition of NKH-477-induced cAMP formation.

Molecular Modeling
Previously, a CB2 3D-QSAR affinity model had been developed [29], which led to the identification of a new CB2 receptor ligand with high affinity and selectivity [30]. In this study, we extended the 3D-QSAR model by creating a structure-based CB2 pharmacophore that comprised the excluded volumes ( Figure 5) and by introducing new compounds to increase the predictability and accuracy of the 3D-QSAR model. The structure of the CB2 receptor co-crystallized with AM-10257 (Protein Data Bank code: 5ZTY, resolution: 2.80 Å) was selected and prepared for calculations using Protein Preparation Wizard to assign bond orders, add missing hydrogen atoms, cap termini, and generate tautomeric states. different cycles, were still able to antagonize JWH-133-induced inhibition of NKH-477induced cAMP formation.
The entire set of ligands, which consisted of 155 molecules, was randomly divided into a training set and a test set, which contained 95 and 60 compounds, respectively. The constructed 3D-QSAR model was tested using two independent test sets of derivatives, and it was decided that it could be used for the development of new high-affinity CB2 compounds. The 3D-QSAR contour plots were analyzed to understand the effects of the spatial arrangement of the structural features. By default, the blue cubes indicate favorable features that contribute to ligand interactions with the CB2 receptor, while the red cubes indicate unfavorable features.
According to our predictions, the alkyl chain at N4 had a major effect on receptor affinity. The presence of blue cubes around the alkyl chain at N4 (Figure 6a) indicates the preference of hydrophobic/nonpolar functional groups at this position. Moreover, the substitution of the alkyl chain at N4 with an isopropyl group leads to a decrease in binding affinity due to the shortness of this group. The presence of red cubes around the alkyl chain suggests that further elongation could decrease the binding affinity at the CB2 receptor.
features that contribute to ligand interactions with the CB2 receptor, while the red cubes indicate unfavorable features.
According to our predictions, the alkyl chain at N4 had a major effect on receptor affinity. The presence of blue cubes around the alkyl chain at N4 (Figure 6a) indicates the preference of hydrophobic/nonpolar functional groups at this position. Moreover, the substitution of the alkyl chain at N4 with an isopropyl group leads to a decrease in binding affinity due to the shortness of this group. The presence of red cubes around the alkyl chain suggests that further elongation could decrease the binding affinity at the CB2 receptor. Indeed, the alkyl chain was not further elongated to avoid an excessive increase in lipophilicity. According to our contour analysis, the substitution of the alkyl chain at N4 with a benzyl group could lead to high-affinity compounds (see compounds 26 and 3), which was confirmed by the experimental results. Surprisingly, this benzyl group also increased the selectivity toward the CB2 receptor. Another result of our 3D-QSAR contour plot analysis was the N1, N2, and N3 methylation of the pyrazolopyridine moiety. Methylation of these positions could lead to high-affinity compounds. The presence of red cubes next to the methyl groups at positions N1, N2, and N3 indicated that more than one methyl group in this region of the molecule could decrease the binding affinity at the CB2 receptor, which was also confirmed by the experimental results (Figure 6a). The blue cubes around the methyl groups indicate that the substituents in these positions may interact as Figure 6. Pictorial representation of the contour plots generated for compound 54 in the context of (a) hydrophobic/non-polar interactions, (b) hydrogen bond donor effects, and (c) electronwithdrawing groups.
Indeed, the alkyl chain was not further elongated to avoid an excessive increase in lipophilicity. According to our contour analysis, the substitution of the alkyl chain at N4 with a benzyl group could lead to high-affinity compounds (see compounds 26 and 3), which was confirmed by the experimental results. Surprisingly, this benzyl group also increased the selectivity toward the CB2 receptor. Another result of our 3D-QSAR contour plot analysis was the N1, N2, and N3 methylation of the pyrazolopyridine moiety. Methylation of these positions could lead to high-affinity compounds. The presence of red cubes next to the methyl groups at positions N1, N2, and N3 indicated that more than one methyl group in this region of the molecule could decrease the binding affinity at the CB2 receptor, which was also confirmed by the experimental results (Figure 6a). The blue cubes around the methyl groups indicate that the substituents in these positions may interact as H-bond donors in the CB2 binding pocket (Figure 6b). The appearance of blue cubes around the hydroxyl group in position 7 and the oxygen of the amide moiety in position 3 (Figure 6c) of the pyrazolopyridine core indicates the preference of electron-withdrawing groups at these positions. On the other hand, the red cubes around the alkyl chain indicate that the electron-withdrawing groups are not preferred in this position.
At this point, we decided to compute the binding affinity of the new CB2 compounds mentioned in this article to obtain some information before planning the synthesis of further series of compounds. The experimental results herein reported were in agreement with our predictions (Figure 7) (further details on the model will be reported elsewhere in due time). These findings on the predictive ability of our newly developed computational approach will allow us to further rationalize the structure-affinity relationship through its microscopic interpretation in view of the design of other CB2 ligands.
with our predictions (Figure 7) (further details on the model will be reported elsewhere in due time). These findings on the predictive ability of our newly developed computational approach will allow us to further rationalize the structure-affinity relationship through its microscopic interpretation in view of the design of other CB2 ligands.  Table 1.

Synthesis
Commercially available reagents were used as received unless otherwise specified. Solvents were treated before use with suitable drying agents and used after distillation in an atmosphere of N2. Reactions requiring anhydrous conditions were performed under N2. Organic solutions were dried over anhydrous Na2SO4. Evaporation was carried out at reduced pressure using a rotating evaporator. Melting points were determined on a Gallenkamp apparatus and are uncorrected. For flash chromatography, Merck Kieselgel 60 (0.040-0.063 mm) was used. Thin layer chromatography (TLC) was performed on silica gel plates (Merck 60 F254) eluting with solvents indicated, visualized by a 254 nm UV lamp (λ = 254 nm), and stained with aqueous KMnO4, Pancaldi, and phosphomolybdic acid solutions. 1 H NMR and 13 C NMR spectra were recorded on a Bruker Avance DPX-400 NMR spectrometer at 400 MHz ( 1 H) or 100 MHz ( 13 C), respectively, or on a Bruker Avance DRX-600 spectrometer operating at 600 MHz ( 1 H) or 150 MHz ( 13 C), respectively. Mass spectra were recoded with an LC-MSD 110 series AGILENT instrument, with electrospray  Table 1.

Synthesis
Commercially available reagents were used as received unless otherwise specified. Solvents were treated before use with suitable drying agents and used after distillation in an atmosphere of N 2 . Reactions requiring anhydrous conditions were performed under N 2 . Organic solutions were dried over anhydrous Na 2 SO 4 . Evaporation was carried out at reduced pressure using a rotating evaporator. Melting points were determined on a Gallenkamp apparatus and are uncorrected. For flash chromatography, Merck Kieselgel 60 (0.040-0.063 mm) was used. Thin layer chromatography (TLC) was performed on silica gel plates (Merck 60 F254) eluting with solvents indicated, visualized by a 254 nm UV lamp (λ = 254 nm), and stained with aqueous KMnO 4 , Pancaldi, and phosphomolybdic acid solutions. 1 H NMR and 13 C NMR spectra were recorded on a Bruker Avance DPX-400 NMR spectrometer at 400 MHz ( 1 H) or 100 MHz ( 13 C), respectively, or on a Bruker Avance DRX-600 spectrometer operating at 600 MHz ( 1 H) or 150 MHz ( 13 C), respectively. Mass spectra were recoded with an LC-MSD 110 series AGILENT instrument, with electrospray interface and 0.4 mL/min flow rate using a binary solvent system of 95:5 methanol/water. The UV detector was set at 254 nm and mass spectra were acquired either in positive or in negative mode scanning over the mass range of 105-1500. Elemental analyses were performed in our analytical laboratory and agreed with theoretical values to within ±0.4%. The appropriate primary alkyl halide (2.26 mmol) was added to a solution of primary amine derivative 4, 28, 39, 40 (2.26 mmol) in dry DMF (3 mL) under a nitrogen atmosphere. The solution was gradually warmed to 50 • C and maintained at this temperature for 7 h. After cooling to rt, the solution was diluted with water (20 mL) and extracted with ethyl acetate. The organic layer was washed with 5% lithium chloride solution (4×), water (2×), then brine, and dried. Removal of solvent gave an oily residue, which was purified by flash column chromatography on silica gel to give the title compounds. 5: prepared from 4 and 1-iodobutane as reported previously [34]. To a solution of 4 (0.1 g, 0.6 mmol) and the appropriate aromatic aldehyde (0.7 mmol) in dry dichloromethane (4 mL) was added anhydrous sodium sulfate (200-300 mg) and the mixture was stirred at room temperature overnight. Filtration of the solid and evaporation of the solution left an oily residue, which was taken up in methanol (4 mL). After adding sodium borohydride (20 mg, 0.6 mmol), the reaction mixture was stirred at rt for 2 h. Removal of methanol gave a residue, which was dissolved in dichloromethane, and this solution was washed with water, dried, and evaporated. The oily product was purified by flash column chromatography on silica gel, eluting with EtOAc/PE (1:1).  (30-40 mL) was added dropwise to an ice-cooled solution of amines 5-8, 29, 30, 41-45 (10 mmol) and triethylamine (20 mL, 15 mmol) in dry dichloromethane (50 mL). After stirring at rt for 2-6 h, the solution was washed with saturated solution of sodium hydrogen carbonate, then 2 N HCl and brine. The organic layer was dried and evaporated to give an oily residue, which was purified by flash column chromatography on silica gel. 9: prepared from 5 as previously described [34]. A solution of the corresponding precursor 9-12, 31, 32, 46-50 (0.96 mmol) in anhydrous THF (35 mL) was added dropwise to a suspension of 60% sodium hydride (77 mg, 1.92 mmol) in the same solvent (20 mL) containing dry methanol (20 µL, 0.48 mmol). The reaction mixture was stirred at 60 • C for 2 h and then evaporated to leave a solid residue, which was taken up in 5% sodium carbonate solution. The aqueous solution was extracted with diethyl ether to remove some starting material, then brought to pH 1 by addition of 6 N HCl and extracted with EtOAc (3 × 20 mL). The organic layer was dried, filtered, and evaporated to give the intermediates of the Dieckmann condensation as a solid residue insoluble in most organic solvents, which was used in the next step without purification. A mixture of the bicyclic ester (0.96 mmol), the corresponding primary amine (1.92 mmol) in THF (3 mL), and toluene (10 mL) was refluxed for 1-2 h, with further toluene added dropwise and the azeotropic mixture methanol/toluene distilled off. After completion of the reaction, the solvent was removed under reduced pressure, and the solid residue was taken up in EtOAc. The organic solution was washed successively with 2 N HCl and brine, then dried, filtered, and concentrated to give a solid residue, which was purified as indicated below to yield the final compounds 13-27, 33-35, 51-55.

Dissolution of 63-Salts in 0.9% NaCl Solution
An excess amount (~4 to 5 mg) of each salt of compound 63 was added to 5 mL of a 0.9% NaCl solution, kept under constant agitation, and thermostated at room temperature (21 ± 2 • C). At fixed time points, an aliquot was withdrawn from the solution, and centrifuged (3000 g, 10 min, RT) to separate undissolved compound and the supernatant, after dilution with MeOH, was analyzed by LC-MS/MS. An Accela UHPLC system coupled with a TSQ Quantum Access Max triple quad mass spectrometer (Thermo Fisher, Waltham, MA, USA), was employed to quantify dissolved 63 over time. UHPLC separation occurred by gradient elution, employing a Synergy Fusion C18 80Å RP-column (2.0 × 100 mm, 4 µm; Phenomenex, Bologna, Italy). Mobile phases A and B were acetonitrile and water, both with 0.1% v/v formic acid added. The following elution conditions were employed: linear gradient from 60 to 95%A in 5 min; 95% A between 5 and 8 min, returning to 60% A in 0.

Competition Binding Assay
Binding at equilibrium was performed as previously reported [49]. Briefly, membranes from HEK-293 cells over-expressing the respective human recombinant CB1 receptor (Bmax = 2.5 pmol/mg protein) and human recombinant CB2 receptor (Bmax = 4.7 pmol/mg protein) were incubated with [ 3 H]-CP-55,940 (0.14 nM/Kd = 0.18 nM and 0.084 nM/Kd = 0.31 nM, respectively, for CB1 and CB2 receptor) as the high-affinity ligand. Competition curves were performed by displacing [ 3 H]-CP-55,940 with increasing concentration of compounds (0.1-10 µM). Nonspecific binding was defined by 10 µM of WIN55,212-2 as the heterologous competitor (K i values 9.2 nM and 2.1 nM, respectively, for CB1 and CB2 receptors). All compounds were tested following the procedure described by the manufacturer (Perkin Elmer, Italy). Displacement curves were generated by incubating compounds with [ 3 H]-CP-55,940 for 90 min at 30 • C. K i values were calculated by applying the Cheng-Prusoff equation to the IC 50 values (obtained by GraphPad) for the displacement of the bound radioligand by increasing concentrations of the test compound. Data represent mean values for at least three separate experiments performed in duplicate and are expressed as K i (nM), average SEM < 5%.

Functional Activity at CB2 Receptor In Vitro
The cAMP Hunter™ assay enzyme fragment complementation chemiluminescent detection kit was used to characterize the functional activity in CB2 receptor-expressing cell lines, as previously reported [67]. Gi-coupled cAMP modulation was measured following the manufacturer's protocol (DiscoveRx, Fremont, CA, USA). Briefly, CHO-K1 cells overexpressing the human CB2 receptor were plated into a 96-well plate (10,000 cells/well), and incubated overnight at 37 • C, 5% CO 2 . Media were aspirated and replaced with assay buffer. In the case of Agonist mode, cells were treated with doseresponse solutions of samples prepared in presence of cell assay buffer containing 25 µM NKH477 solution (a water-soluble analog of Forskolin) to stimulate adenylate cyclase and enhance basal cAMP levels. In the case of Antagonist mode, cells were pre-treated (15 min at 37 • C) with a dose-response solution of samples prepared in presence of cell assay buffer and further treated with a challenge of a known CB2 agonist (JWH-133, 4 µM) in the presence of cell assay buffer containing 25 µM NKH477. Following stimulation, cell lysis and cAMP detection were performed as per the manufacturer's protocol. Luminescence measurements were measured using a GloMax Multi Detection System (Promega, Milan, Italy). Data are reported as mean ± SEM of three independent experiments conducted in triplicate and were normalized considering the NKH477 stimulus alone as 100% of the response. Data were analyzed using PRISM software 9.3.1 (GraphPad Software, Inc., San Diego, CA, USA).

Conclusions
Our efforts to expand the pyrazolo[4,3-b]pyridine-6-carboxamide family have successfully yielded 23 new cannabinoid ligands, 10 of which display remarkable CB2R affinity (K i < 50 nM) and also good selectivity (SI > 200) over CB1R. Most of the compounds in the series also exhibit excellent physicochemical properties, in particular lipophilicity values determining LLE values between 5.15 and 6.21, except for compound 18 (LLE = 4.90). In addition, the conversion of 63, taken as a prototypical exponent of the series, to alkaline salts drastically improved both water solubility and dissolution rate. Both agonists and inverse agonists/antagonists have been identified in this series of compounds, with functional activity determined mainly or even exclusively by the substitution pattern of the pyrazole ring. It is therefore quite clear that in the future it will be possible to obtain both agonists and inverse agonists/antagonists in this class of compounds by N1 and N2 substitution of the pyrazole ring, respectively. The affinity and selectivity of the new compounds for the CB2 receptor can be achieved by wisely choosing the optimal substituents for the other parts of the molecule using the newly developed 3D-QSAR model, which herein was applied to validate its prediction ability on receptor affinity and selectivity of the new compounds: the calculated K i values are in agreement with the experimental K i values.
Taken together, these data suggest that the pyrazolo[4,3-b]pyridine-6-carboxamides described here represent an excellent starting point for the further chemical optimization of CB2R agonists and inverse agonist/antagonists, as well as a remarkable tool for exploring the involvement of CB2R in various physiopathological conditions. Supplementary Materials: The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/molecules28134958/s1, Figure S1: X-ray crystal structure of compound 51; Figure S2: Bioavailability radar plot of selected compounds using the SwissADME software; Figures S3-S48: NMR spectra of compounds 13-27, 33-35, 51-55; Table S1: Calculated physicochemical and drug-like properties of selected compounds.