Discovery of a High Affinity Adenosine A₁/A₃ Receptor Antagonist with a Novel 7-Amino-pyrazolo[3,4-d]pyridazine Scaffold

Abstract

Here we describe the design and synthesis of pyrazolo[3,4-d]pyridazines as adenosine receptor (AR) ligands. We demonstrate that the introduction of a 3-phenyl group, together with a 7-benzylamino and 1-methyl group at the pyrazolopyridazine scaffold, generated the antagonist compound 10b, which displayed 21 nM affinity and a residence time of ∼60 min, for the human A₁R, 55 nM affinity and a residence time of ∼73 min, for the human A₃R and 1.7 μΜ affinity for the human A_2BR while not being toxic. Strikingly, the 2-methyl analog of 10b, 15b, had no significant affinity. Docking calculations and molecular dynamics simulations of the ligands inside the orthosteric binding area suggested that the 2-methyl group in 15b hinders the formation of hydrogen bonding interactions with N^6.55 which are considered critical for the stabilization inside the orthosteric binding cavity. We, therefore, demonstrate that 10a is a novel scaffold for the development of high affinity AR ligands. From the mutagenesis experiments the biggest effect was observed for the Y271^7.46A mutation which caused an ∼10-fold reduction in the binding affinity of 10b.

Adenosine, a naturally occurring purine nucleoside, is the endogenous agonist of adenosine receptors (ARs).¹ ARs are G protein-coupled receptors (GPCRs) comprising four subtypes; A₁, A_2A, A_2B, and A₃. The A_2A and A_2B subtypes act synergistically with Gα_s stimulating adenylyl cyclase and, therefore, increasing 3′,5′-cyclic adenosine monophosphate (cAMP) levels. In contrast, A₁ and A₃ receptor subtypes inhibit adenylyl cyclase and decrease cAMP levels by coupling to the G_i/o family of G proteins.

In the last two decades numerous heterocyclic compounds have been synthesized as AR ligands including xanthines and bi- or tricyclic fused heterocyclic analogues, e.g., purines, deazapurines, pyrazolopyridines, imidazotriazines, thienopyridazines, naphthyridines, pyridopyrimidines, and pyrazoloquinolines.²⁻⁴

Different therapeutic applications have been identified in preclinical and clinical studies for A₁R antagonists as potassium-sparing diuretic agents with kidney-protecting properties,² treatments for chronic lung diseases such as asthma,^5,6 and possible use in Parkinson’s disease.⁷

A₃R has been reported to be overexpressed in several types of cancer cells and is, thus, considered as a biological marker for tumors.⁸ In a recent study, the potent and selective A₃R antagonist LJ-1888 ((2R,3R,4S)-2-[2-chloro-6-(3-iodobenzylamino)-9H-purine-9-yl]tetrahydrothiophene-3,4-diol) blocked the development and attenuated the progression of renal interstitial fibrosis,⁹ while A₃R antagonists have demonstrated efficacy in eye pathologies by lowering intraocular pressure.¹⁰

While the binding mode of several agonists and antagonists at A₁R has been revealed with X-ray crystallography or cryogenic electron microscopy,¹¹⁻¹³ the experimental structures for A₃R and A_2BR have, to date, not been resolved, and only homology models can be used for these AR subtypes.

By the repurposing of antiproliferative aromatic condensed nitrogen heterocycles, we previously identified nanomolar affinity pyrazolo[3,4-c]pyridine A₁R/A₃R antagonists.¹⁴ It has been reported that non-xanthine pyrazolo derivatives that potently bind ARs are pyrazolo[4,3-d]pyrimidines,³ pyrazolo[1,5-c]quinazolines,¹⁵ pyrazolo[3,4-b]pyridines,^16,17 pyrazolo[3,4-b]pyridines, pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidines, pyrazolo[3,4-c]- or -[4,3-c]quinolines, pyrazolo[4,3-d]pyrimidinones, pyrazolo[3,4-d]pyrimidines, and pyrazolo[1,5-a]pyridines.¹⁸ After we previously identified the potent pyrazolo[3,4-c]pyridine A₁R/A₃R antagonists¹⁴ and observed that certain substituted pyrazolo[3,4-b]pyridines had antagonistic potency against A₃R or A₁R,^16,17 we quantified the novel pyrazolo[3,4-d]pyridazine scaffold for activity at ARs. Here, we synthesized a series of new 3-alkyl- or 3-aryl-7-amino-pyrazolo-[3,4-d]pyridazine derivatives and determined their affinities against the different ARs using functional cAMP accumulation assays, fluorescent ligand displacement binding studies, and molecular dynamics (MD) simulations.^19,20 We identified the 21 nM A₁R/55 nM A₃R/<2 μΜ A_2BR antagonist 1-methyl-3-phenyl-7-benzylaminopyrazolo[3,4-d]pyridazine (10b) as a lead compound. Strikingly, compound 15b, the 2-methyl congener of 10b, had lower affinity by >100-fold against 3AR subtypes since, we assumed, it cannot form hydrogen bonding interactions with N^6.55 which are considered critical for stabilization inside the orthosteric binding cavity. Finally, as these new compounds present structural similarity to antiproliferative purine analogues,²¹ we evaluated their cytotoxic potential against the human fibroblasts cell line (WI-38) and prostatic (PC-3) and colonic (HCT116) cancer cell lines.

Similarity Calculations

Searching the CHEMBL²² database to determine if pyrazolo[3,4-d]pyridazine has been used as a scaffold for ligands binding to ARs, using a TanimotoCombo (Tc)²³ coefficient > 0.85, we did not find any pyrazolo[3,4-d]pyridazine derivatives with potency against ARs, suggesting that it is a novel ring system for the development of AR ligands. When we considered the amide 7-benzylamino-3-phenylpyrazolo[3,4-d]pyridazine, we found the 4-(2-phenethyl)amino 1-phenylethylpyrazolo[3,4-b]pyridine (T_c = 0.15) had been reported to bind A₁R.^16,17 Thus, we proceeded with a structural activity relationship study around 7-benzylamino-3-phenyl pyrazolo[3,4-d]pyridazine and synthesized a series of 7-amino-pyrazolo[3,4-d]pyridazines for biological evaluation against ARs.

Chemistry

The synthesis of the target compounds was accomplished through the previously reported pyrazolecarboxylates 4a,b and 5a,b (Scheme 1). Briefly, commercial isopropylmethylketone (1a) or acetophenone (1b), was first converted to the ethyl 2,4-diketocarboxylates 2a and 2b, respectively,^24,25 which upon reaction with hydrazine monohydrate gave the pyrazolecarboxylates 3a,b.²⁶ These were methylated using methyl iodide in the presence of sodium hydride and provided the regioisomers 4a,b(27,28) and 5a,b,²⁸ respectively. Interestingly, when we used tetrahydrofuran as solvent in the place of dimethylformamide (DMF), we exclusively obtained the N¹-methyl-5-carboxylate 4a isomer.

Scheme 1. Synthesis of 4a,b and 5a,b.

Reagents and conditions: (a) diethyl oxalate, NaH 60%, toluene dry, 50°C, 2 h; (b) NH₂NH₂ 80%, EtOH, reflux, 90 min; (c) (i) NaH 60%, DMF dry, 0 °C, 15 min, (ii) CH₃I, rt, 1 h.

Each of the isomeric pyrazoles 4a,b or 5a,b was subsequently treated with paraformaldehyde in the presence of a 33% HBr solution in acetic acid and was converted to the bromides 6a,b (Scheme 2) or 11a,b (Scheme 3), respectively. The bromomethyl group was then oxidized using N-methylmorpholine N-oxide to generate the carbaldehydes 7a,b (Scheme 2) and 12a,b (Scheme 3).

Scheme 2. Synthesis of 10a–c.

Reagents and conditions: (a) paraformaldehyde, 33% HBr in AcOH, 90 °C, 3.5 h; (b) Ν-methylmorpholine-Ν-oxide, MeCN dry, rt, 24 h; (c) NH₂NH₂ (80%), HCl 36%, EtOH, 90°C, 1 h; (d) POCl₃, 110 °C, 2.5–8 h; (e) HNR₁R₂, EtOH, reflux, 2 h.

Scheme 3. Synthesis of 15a–c.

Reagents and conditions: (a) paraformaldehyde, 33% HBr in AcOH, 90 °C, 3.5 h; (b) Ν-methylmorpholine-Ν-oxide, MeCN dry, rt, 24 h; (c) NH₂NH₂ (80%), HCl 36%, EtOH, 90 °C, 1 h; (d) POCl₃, 110 °C, 2.5–8 h; (e) HNR₁R₂, EtOH, reflux, 2 h.

The aldehydes 7a,b and 12a,b were then treated with hydrazine, and upon ring closure the pyrazolopyridazinones 8a,b and 13a,b were obtained. The pyridazinones reacted with phosphorus oxychloride to give the corresponding chloro derivatives 9a,b and 14a,b with suitable purity that they could be introduced to the next reaction. These crude products were then treated with benzylamine or morpholine to result in the target compounds 10a–c and 15a–c (Figures S1–S3).

Assessing Biological Activity of Pyrazolo[3,4-d]pyridazine Derivatives

cAMP Assays Assessing Activity at Adenosine Receptors

Having synthesized compounds 10a–c and 15a–c, we next tested their activity, as antagonists, against the different human AR subtypes using a single high concentration of the compound (1 μM) coadministered with NECA (5′-N-ethylcarboxamidoadenosine) in a cAMP accumulation assay (Figure 1A and B). Note that for A₁R and A₃R 10 μM forskolin was added since these are G_i/o-coupled receptors and reduce cAMP accumulation.^19,29 All compounds lacked efficacy at NECA-stimulated A_2AR (even when tested at 10 μM) (Table S1). Compounds 10c, 15b, and 15c also lacked efficacy at the other 3AR subtypes, with 15a displaying weak efficacy only at A₃R, while compounds 10a and 10b displayed activity at all 3ARs although this was only detectable for A_2BR when a 10 μM concentration of the compound was used (Table S1). Based upon a single concentration of antagonist, we calculated the equilibrium dissociation constant (pK_d) of each compound (Table 1). Of the compounds tested, 10b displayed the highest affinity at the different AR subtypes with greater selectivity toward A₁R and A₃R than A_2BR. We next performed a more extensive Schild analysis using multiple doses of the most potent antagonist, 10b, only at A₁R and A₃R (Figure 1C). In both cases 10b acted as a competitive antagonist, generating a Schild slope that did not significantly differ from unity. Using the Schild plot, we calculated 10b’s affinity (pA₂/pK_b) to be 21 nM at A₁R and 55 nM at A₃R while only 1.7 μΜ at A_2BR (Table S1).

Figure 1.

Characterization of 7-amino-pyrazolo[3,4-d]pyridazines at human A₁R and A₃R. (A and B) Cells expressing either human A₁R (A) or A₃R (B) were exposed to 10 μM forskolin and stimulated with increasing concentrations of NECA for 30 min in the presence of a 1 μM concentration of the test compound, and the cAMP accumulation was quantified. (C) cAMP accumulation was measured as detailed in part A using multiple concentrations of 10b. Using pEC₅₀ values, Schild regression analysis was conducted to calculate pA₂/pK_b values. All values are mean ± SEM expressed as percentage forskolin inhibition, relative to NECA. n ≥ 3 independent experimental repeats were performed in duplicate.

Table 1. Chemical Structures, Antagonistic Potencies (pEC₅₀ in the Presence of NECA^a), and Affinities (pK_i^b) of 7-Amino-pyrazolo[3,4-d]pyridazines 10a–c and 15a–c against A₁R and A₃R.

^aMean ± SEM; functional activities (pEC₅₀ values of NECA in the presence of either 1 μM ligands or vehicle) as mean ± standard error of the mean (SEM) of at least three independent repeats, conducted in duplicate—values obtained from Figure 1.

^bMean ± SEM; equilibrium binding affinities of the ligands measured with NanoBRET against Nluc-A₃R or Nluc-A₁R; NECA was used as positive control.³

^#Due to the high affinity of MRS1220, 10 nM was used to enable measurement of the full dose–response curve of NECA to determine pEC₅₀.

Statistical significance compared to NECA was determined, at p < 0.05, through one-way ANOVA with Dunnett’s post-test (*, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001).

Quantifying Binding Parameters Using a NanoBRET-Based Saturation Binding Assay

We next sought to independently verify the affinities determined using the Shield analysis by directly quantifying the potential antagonists’ binding to A₁R and A₃R using a previously described saturation nano-bioluminescence resonance energy transfer (NanoBRET) binding assay.¹⁹ We determined the ability of all the compounds to displace the specific binding of CA200645,³⁰ a fluorescent antagonist of A₃R and A₁R, using Nluc-A₃R expressing human embryonic kidney 293 (HEK293) and Nluc-A₁R HEK293 cells (Figure 2 and Table 1). A_2BR was not included in this analysis since the pK_d values of 10a and 10b at A_2BR were estimated to be below 1 μM (Figure 1 and Table 1). Consistent with the Schild analysis, compound 10b displayed the highest affinity at A₁R and A₃R (A₁R, pK_i = 7.95 ± 0.09; A₃R, pK_i = 7.89 ± 0.11). Of the remaining compounds, 10a displayed weak affinity at A₃R (pK_i, 6.42 ± 0.28), which agreed with the Schild regression estimate, but failed to fully displace CA200645 at A₁R, making an estimate for its affinity unreliable. All the other compounds failed to displace CA200645 at A₁R or A₃R except for 15a and 15c, which did display some binding at A₃R but, like 10a, also failed to fully displace CA200645 at the concentrations tested. Significantly, 15b, which contains an N-methyl substitution to 1-NH and 2-NMe compared to 1-NMe and 2-NH in 10b, failed to bind either AR subtype.

Figure 2.

Inhibition of BRET between CA200645 at NLuc-A₁R and Nluc-A₃R by 10b and 10a. HEK293 cells expressing Nluc-A₁R (A) or Nluc-A₃R (B) were treated with 5 nM or 20 nM CA200645, respectively, enabling concentration-dependent decreases in the BRET ratio at 10 min to be determined with the response normalized to DMSO. Binding curves were fitted with the Cheng Prusoff equation built into GraphPad Prism 9.3 to enable estimates of the pK_i.¹⁹ Comparison of pK_i values for A₁R (C) and A₃R (D) as determined via BRET binding. Each data point represents the mean ± SEM of at least three experiments performed in duplicate. The statistical significance compared to NECA was determined, at p < 0.05, through one-way ANOVA with Dunnett’s post-test (*, p < 0.05; ***, p < 0.001). ^#Compounds did not fully displace CA200645, so pK_i values are estimates preventing statistical analysis.

Determining Kinetic Parameters of 10b Binding at A₃R and A₁R Using NanoBRET

We next investigated the real-time binding kinetics^19,30 of 10b at A₃R and A₁R using the NanoBRET binding method. Specifically, we quantified 10b’s ability to inhibit specific binding of CA200645 to Nluc-A₃R and Nluc-A₁R expressed in HEK293 cells. The kinetic parameters for CA200645 binding at Nluc-A₃R were previously determined as K_on = 32.5 ± 0.28 × 10⁵ M^–1 min^–1 and K_off = 0.025 ± 0.005 min^–1 with a pK_D of 10.11. Conversely the kinetics of CA200645 binding at Nluc-A₁R were determined as K_on = 14.5 ± 0.4 × 10⁵ M^–1 min^–1, K_off = 0.023 ± 0.001 min^–1, and pK_D = 7.80 ± 0.2 nM.¹⁴ Applying these parameters into the “kinetics of competitive binding” model built into GraphPad Prism9.0, we were able to provide estimates of the kinetics of binding for 10b against A₁R (K_on = 51.4 ± 0.26 × 10⁵ M^–1 min^–1, K_off = 0.019 ± 0.003 min^–1 with a pK_D = 7.46 ± 0.1 and RT = 59.8 ± 12.7 min) and against the A₃R, (K_on = 25.6 ± 0.1 × 10⁵ M^–1min^–1, K_off = 0.0014 ± 0.002 min^–1 with a pK_D = 7.26 ± 0.05 and RT = 72.58 ± 8.8 min). None of the other compounds were analyzed using this method due to their extremely fast K_off rates (>min^–1). For compound 10b there was an excellent agreement between pK_D (K_on/K_off) of the compounds from the kinetics assays and the Schild analysis (pA₂/pK_b) and fair agreement (∼3.16-fold) with the saturation binding assays (pK_i).

Simulations

Investigation of the Binding of the 7-Amino-pyrazolo[3,4-d]pyridazines to A₁R and A₃R

Having pharmacologically evaluated the different compounds, we then used molecular docking to provide insights into how they bind to the ARs. We docked 10a–10c into the orthosteric binding site of A₁R and 10b and 15b into A_2BR and A₃R (the amino acid sequences of A₁R, A₃R, and A_2BR in the orthosteric binding area are shown in Scheme S1) using ChemScore as the scoring function³¹ with the highest score docking pose being inserted into a hydrated phosphatidylethanolamine bilayer. The complexes were subjected to 100 ns MD simulations with amber99sb,³² and then, the MD simulations’ trajectory was analyzed (Table S2). The MD simulations showed that the 7-benzylamino-pyrazolo[3,4-d]pyridazine 10b substituted with N¹Me and a 3-phenyl group formed a stable complex with all 3ARs with RMSD_protein values <2.1 Å. Starting from the same docking pose of 10b in A₁R or A₃R (Figure 3), the mean frame from MD simulations was close to the starting docking pose in A₁R (RMSD_lig = 1.21 Å) while in A₃R (Figure S2) the ligand moved considerably into the cleft between the transmembrane (TM)3, TM5, and TM6 helices (RMSD_lig = 4.88 Å). Thus, starting from the same binding pose for 10b, the MD simulations produced two different binding orientations at A₁R and A₃R. This is due to the fact that A₁R has a broader binding area, expanded toward TM1 and TM2, compared to the other ARs, according to the X-ray structures of A₁R in complex with antagonists.^11,12 A similar AR ligand reported in the literature is 4-(2-phenethyl)amino 1-phenylethyl pyrazolo[3,4-b]pyridine (Tc = 0.15), which binds with a similar docking pose to 10b to A₁R.¹⁶ We also docked a representative adenine derivative (N9-methyl,N6-benzyl adenine) to A₁R and found a similar docking pose (Figure S3).

Figure 3.

(A–C) 100 ns MD simulations of 10a–c inside the orthosteric binding area of A₁R. (D) 100 ns MD simulations of 15b inside the orthosteric binding area of A₁R. Starting structures are shown (docking pose), and representative frames from MD simulations, receptor–ligand interaction frequency histograms, and RMSD plots of proteins (RMSD_protein; blue plots) and ligand heavy atoms (RMSD_ligand; red plots) inside the orthosteric binding area of WT A₁R or A₃R. Bars are plotted only for residues with interaction frequencies ≥0.2. Color scheme: ligand = brown sticks, receptor = white cartoon and sticks, hydrogen bonding interactions = yellow (dashes or bars), π–π interactions = green (dashes or bars), hydrophobic interactions = gray, water bridges = blue. For the protein models of A₁R in complex with 10a–c or 15b, the experimental structure of the inactive form of A₁R in complex with an antagonist (PDB ID 5UEN(4)) was used.

Inside the A₁R orthosteric site, compound 10b formed hydrogen bonds through its pyrazole or pyridazine nitrogen donor groups to the amide side chain of N254^6.5 or the imidazole side chain of H278^7.43. Furthermore, 10b was stabilized in the orthosteric binding site through π–π interactions between its pyrazolo[3,4-d]pyridazine or phenyl rings with F171^5.29, H251^6.2, and W247^6.48, respectively. The benzylamino group of 10b oriented toward the widened TM2 area in A₁R, forming hydrophobic interactions with A66^2.61 and I69^2.64. Furthermore, 10b was found to bind deep in the pocket interacting with V87^3.32 and W247^6.48 while 3-phenyl-pyrazole aligned close to the side chains of M180^5.38 and L250^6.1 (Figure 3A). In A₃R, compound 10b was stabilized through formation of hydrogen bonding interactions with N254^6.5 and H278^7.43 and hydrophobic interactions with L90^3.32, L91^3.33, F168^5.29, M177^5.38, L246^6.1, and I268^7.39 (Figure S2B). Finally, the MD simulations for 10b (Figure S2A) in complex with A_2BR (Figure S2) show weak hydrogen bond interactions with N254^6.5.

Pharmacologically, compounds 10b and 15b differed considerably in their affinity to the ARs (Figures 1 and 2 and Table 1). Comparing MD simulations for 15b with 10b in the orthosteric binding area of A₁R, A₃R (and A_2BR) shows that starting from a similar docking pose, the substitution from N¹ Me and 2-NH (found in 10b) to N¹H and N² Me (in 15b) results in 15b failing to generate hydrogen bonds with N^6.55 because of the steric repulsion between 2-methyl and the amide side chain of N^6.55; for this reason also 15a and 15c were inactive (Figure S2). Although many ligands can have similar docking poses, subtle changes in the ligand substitution pattern can result in significant changes in binding, and this can be followed only with MD simulations. Considering the two active compounds, 10b and 10a, replacement of the 3-phenyl group (found in 10b) with a 3-isopropyl group (generating 10a) results in a remarkable reduction of affinity. This is due to 10a losing significant π–π interactions with H251^6.2 and hydrophobic interactions with residues deeper in the binding site, e.g., W247^6.48, L250^6.1, and V87^3.32 (Figure 3). Finally, substitution of 10b’s 7-benzylamino by the more rigid morpholinyl group (found in 10c) resulted in reduced affinity to the ARs. The more rigid morpholino group in 10c repels F171^5.29, so the ligand rotates and moves to the bottom of the binding area, losing hydrogen bonding interactions with N254^6.5 and weakening its hydrophobic interaction with critical residues, e.g., F171^5.29 and L250^6.1 (Figure 3). With an accuracy of ∼±4 kcal mol^–1, the MM-GBSA method^33,34 (Supporting Information) only provides an approximation when applied to structure–activity relationships for analogs in the same series. Nevertheless, the MM-GBSA binding free energy calculations for ligands 10a–c against A₁R (Table S2), using the OPLS2005 force field^35,36 with a hydrophobic slab as an implicit membrane model and including the waters in the orthosteric binding area, predicted fairly well the stability of 10a–c in complex with A₁R with binding free energy values (after neglecting entropy) ΔG_eff = −94.50, −96.42, and −85.35 kcal mol^–1.

Mutagenesis Experiments to Study 10b Binding to A₁R

We have previously observed that mutation of residues that do not directly interact with the ligands (e.g., V^5.30 for A₃R, which is more than 4 Å apart from the ligand inside the orthosteric binding area) can, through allosteric interactions due to the plasticity of the binding area, significantly affect ligand affinity.^20,21,37 As such it is not always straightforward to determine the effects of a mutation on affinity properties. Despite this caveat, we next used mutational analysis combined with NanoBRET to determine the important residues required for 10b binding to A₁R. The mutation of L250^6.1A resulted in only a slight reduction of binding affinity for 10b (Table 2) despite the MD simulations suggesting that the ligand should be close enough to L250^6.1 to enable hydrophobic interactions. It is possible that residues H251^6.52 and W247^6.48 could contribute to the stabilization of 10b with hydrophobic interactions even if L250^6.1 is mutated to alanine. It is noteworthy that mutation of E172^5.30 (which is also more than 4 Å apart from the ligand inside the orthosteric binding area) to alanine also did not significantly change the binding affinity (Table 2). This contrasts with our studies using 3-phenyl-7-anilinopyrazolo[3,4-c]pyridines which showed a 1.5-fold decrease in affinity due to the E172^5.30A mutation.¹⁴

Table 2. Binding Affinities (pK_i) for 10b Measured Using Saturation NanoBRET Binding with CA200645 as the Fluorescent Tracer against WT A₁R and Mutant A₁Rs.

A1R	pKi	Effect on affinity
WT	7.68 ± 0.11	baseline
T913.36A	7.68 ± 0.07	no change
E1725.30 A	7.34 ± 0.06	no significant change
L2506.51 A	7.57 ± 0.04	no significant change
H2516.52A	7.62 ± 0.06	no significant change
S2677.42A	7.86 ± 0.03	no significant change
Y2717.46A	6.99 ± 0.05	∼10-fold reduction

In addition, mutation of H251^6.2A has been reported to reduce antagonist affinity against A₃R^20,21 although here it did not have any effect on 10b affinity at A₁R. Other residues of interest to mutate were T91^3.36A and S267^7.42A, which are deep in the orthosteric pocket. Interestingly, we found that mutation to alanine of these residues also did not have a significant effect on the binding affinity of 10b (Table 2). This is in contrast to our results for pyrazolo[3,4-c]pyridines which can interact directly with these residues.⁹ The results for 10b suggested that it is positioned above pyrazolo[3,4-c]pyridines,⁹ in the A₁R pocket, and so unaffected by these mutations.

The biggest effect in this study was observed for the Y271^7.46A A₁R mutation, which caused a ∼10-fold reduction in the binding affinity of 10b (Table 2). This effect is in contrast to that observed previously for pyrazolo[3,4-c]pyridines¹⁴ for which we showed that the Y271^7.46A mutation caused a slight increase in binding affinity. Since the MD simulations showed contacts with H278^7.43 and not Y271^7.46, the Y271^7.46A mutation in A₁R might affect the binding of 10b through contact with H278^7.43. We performed the MD simulations of 10b in complex with A₁R-Y271^7.46A and observed that the ligand loses its hydrogen bonding interactions with N254^6.5, which might weaken its binding interactions with the orthosteric binding area (Figure S4).

Preliminary Toxicological Analysis of Pyrazolo[3,4-d]pyridazine Derivatives

Given the high affinity 10b displays for A₁R and A₃R, and thus the potential for it to be a scaffold for future compound development, we wanted to evaluate its antiproliferative nature as an early indicator of its toxicological profile. We therefore evaluated 10b, alongside the other compounds in this study, for cytotoxic activity against human fibroblasts (WI-38) and two cancer cell lines, namely the prostate cancer (PC-3) and colon cancer (HCT116) cell lines. Importantly, 10b alongside all the compounds proved to be not cytotoxic against the cell lines, with IC₅₀ values >10 μM. The only compound that did display any cytotoxicity was 15b, which displayed moderate cytotoxicity against the PC-3 and HCT116 cell lines, showing IC₅₀ values of 5.3 ± 0.1 μM against PC-3 cells and 4.15 ± 0.05 μM against HCT116 cells. As a result of these data, we are confident that 10b is noncytotoxic and can be progressed for further development as a dual A₁R/A₃R antagonist.

Glossary

Abbreviations

ARs

adenosine receptors

BRET

bioluminescence resonance energy transfer

GPCRs

G protein-coupled receptors

HEK

human embryonic kidney

MD

molecular dynamics

NECA

5′-N-ethylcarboxamidoadenosine

PDB

Protein Data Bank

RMSD

root-mean-square deviation

T_c

TanimotoCombo

t_m

mixing time

TM

transmembrane

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00052.

• Scheme S1. Comparison of amino acid residue sequences of the binding area. Table S1. Chemical structures and antagonistic potencies of 7-amino-pyrazolo[3,4-d]pyridazines 10a–c and 15a–c against A2AR and A2BR. Table S2. Mean RMSD values for all compounds against A₁R, A_2AR, and A_2BR and ΔG_eff only for 10–10c against A₁R. Figure S1. ¹H and ¹³C NMR spectra of the target compounds. Figure S2. Results from the MD simulations of 10b and 15b against A₃R and A_2BR. Figure S3. Docking poses of 4-(2-phenethyl)amino-1-phenylethyl pyrazolo[3,4-b]pyridine and N9-methyl,N6-benzyl adenine to A₁R. Figure S4. Representative frames from 100 ns MD simulations of 10b inside the orthosteric binding area of WT A1R and 10b inside mutant Y271A A₁R and the receptor–ligand interaction frequency histogram and RMSD graphs of protein Ca and ligand heavy atoms. Information for the methods and synthetic protocols. (PDF)

Author Present Address

^‡ Cyprus Pharmaceutical Organization Limited, 11 King Paul A’ Street, 1096 Nicosia, Cyprus

Author Contributions

^† A.S., M.S., and K.K. contributed equally. A.K., N.L. and G.L. contributed equally. A.K., N.L., G.L., and M.S. conceived and designed the research; K.K. and A.P. synthesized and characterized the compounds in the P.M., N.P., and N.L. lab; A.S. and L.D. performed the mammalian assays in the G.L. lab; M.S. performed the simulations in the A.K. lab; K.K., A.P., M.S., A.S., N.L., P.M., A.K., and G.L. analyzed the data; A.K., P.M., N.P., N.L., and G.L. wrote the manuscript; A.K., G.L., N.L., and M.S. edited the manuscript.

The authors declare no competing financial interest.