Substituted Purines as High-Affinity Histamine H3 Receptor Ligands

Continuing with our program to obtain new histamine H3 receptor (H3R) ligands, in this work we present the synthesis, H3R affinity and in silico studies of a series of eight new synthetically accessible purine derivatives. These compounds are designed from the isosteric replacement of the scaffold presented in our previous ligand, pyrrolo[2,3-d]pyrimidine ring, by a purine core. This design also considers maintaining the fragment of bipiperidine at C-4 and aromatic rings with electron-withdrawing groups at N-9, as these fragments are part of the proposed pharmacophore. The in vitro screening results show that two purine derivatives, 3d and 3h, elicit high affinities to the H3R (Ki values of 2.91 and 5.51 nM, respectively). Both compounds are more potent than the reference drug pitolisant (Ki 6.09 nM) and show low toxicity with in vitro models (IC50 > 30 µM on HEK-293, SH-SY5Y and HepG2 cell lines). Subsequently, binding modes of these ligands are obtained using a model of H3R by docking and molecular dynamics studies, thus determining the importance of the purine ring in enhancing affinity due to the hydrogen bonding of Tyr374 to the N-7 of this heterocycle. Finally, in silico ADME properties are predicted, which indicate a promising future for these molecules in terms of their physical–chemical properties, absorption, oral bioavailability and penetration in the CNS.


Introduction
The histamine 3 receptor (H 3 R) belongs to a superfamily of G protein-coupled receptors (GPCRs) and has received great interest in recent years due to its distribution, mainly in the central nervous system, acting as an auto-and heteroreceptor and regulating the release of histamine itself and other neurotransmitters, such as acetylcholine and dopamine, among others [1][2][3]. Antagonists or inverse agonists of this receptor prevent the activation of the H 3 R and therefore, due to negative feedback, increase the concentration of histamine and the other neurotransmitters mentioned above. Therefore, these are being investigated as therapeutic alternatives in neurodegenerative diseases such as Alzheimer's and Parkinson's, as well as in disorders such as narcolepsy and attention deficit hyperactivity disorder. In present times, the H 3 R inverse agonist pitolisant (Wakix, Figure 1) is the only drug approved by the EMA for the treatment of narcolepsy with or without cataplexy and for obstructive sleep apnea, which has increased interest in the clinical applications of novel H 3 R inverse agonists or antagonists [4,5]. In this regard, many series of compounds with great structural diversity have been synthesized [6]. However, there is a common pharmacophore consisting of a basic moiety, separated by a linker (often saturated alkyl chains) from an aromatic zone attached to an arbitrary scaffold that is generally hydrophobic [7]. First generation H3R ligands usually contain imidazole-based scaffolds, with regard to histamine structure [8]. Otherwise, the second generation of H3R ligands, the imidazole heterocycle, is replaced by piperidine moieties [9].
In recent years, several H3R ligands that contain the piperidine heterocycle in their structure have been reported. Contilisant (Figure 1) was reported to display H3R antagonism as well as cholinesterase and MAO-B inhibition [10]. Zhang et al. published a series of lactams that contain several basic scaffolds in their structure, in which compound I shows great affinity for the H3R and selectivity against the H1R [11]. On the other hand, our research group has reported several pyrrolo [2,3-d]pyrimidines substituted in position C-4 with the bipiperidine moiety, obtaining affinities for H3R in the nanomolar range and highlighting compounds II and III, which show Ki values of 126 and 6.0 nM, respectively [12,13]. Our previous antecedents suggest that derivatives with a higher affinity for H3Rs are obtained using bipiperidine as a basic moiety, whereas in the arbitrary zone, the best results have used aromatic rings with electron-withdrawing groups (EWG).
In order to improve affinity to the H3R, in this work we propose the use of purine heterocycle as a pyrrolo [2,3-d]pyrimidine bioisostere while maintaining the aforementioned substitution patterns present in III (Figure 1). Purine heterocycle is selected because this scaffold is part of several compounds with diverse biological activities [14][15][16][17]. Moreover, this heterocycle significantly improves pharmacological activities and physicochemical properties [18]. In addition, the effects of EWG on the aromatic ring at N-9 are considered as well as the replacement of a hydrogen atom at the C-2 position by a chlorine atom. Later, the synthesized purine derivatives are evaluated for their affinity to the human H3R and the most promising compounds are evaluated for their cytotoxicity on cell lines. Finally, docking and molecular dynamic studies are carried out to understand the effects of chemical modifications on affinity to the H3R. Considering the great need to find new therapeutic alternatives for the treatment of neurodegenerative diseases, this work contributes to medicinal chemistry through the discovery of easily accessible compounds In this regard, many series of compounds with great structural diversity have been synthesized [6]. However, there is a common pharmacophore consisting of a basic moiety, separated by a linker (often saturated alkyl chains) from an aromatic zone attached to an arbitrary scaffold that is generally hydrophobic [7]. First generation H 3 R ligands usually contain imidazole-based scaffolds, with regard to histamine structure [8]. Otherwise, the second generation of H 3 R ligands, the imidazole heterocycle, is replaced by piperidine moieties [9].
In recent years, several H 3 R ligands that contain the piperidine heterocycle in their structure have been reported. Contilisant (Figure 1) was reported to display H 3 R antagonism as well as cholinesterase and MAO-B inhibition [10]. Zhang et al. published a series of lactams that contain several basic scaffolds in their structure, in which compound I shows great affinity for the H 3 R and selectivity against the H 1 R [11]. On the other hand, our research group has reported several pyrrolo [2,3-d]pyrimidines substituted in position C-4 with the bipiperidine moiety, obtaining affinities for H 3 R in the nanomolar range and highlighting compounds II and III, which show K i values of 126 and 6.0 nM, respectively [12,13]. Our previous antecedents suggest that derivatives with a higher affinity for H 3 Rs are obtained using bipiperidine as a basic moiety, whereas in the arbitrary zone, the best results have used aromatic rings with electron-withdrawing groups (EWG).
In order to improve affinity to the H 3 R, in this work we propose the use of purine heterocycle as a pyrrolo [2,3-d]pyrimidine bioisostere while maintaining the aforementioned substitution patterns present in III (Figure 1). Purine heterocycle is selected because this scaffold is part of several compounds with diverse biological activities [14][15][16][17]. Moreover, this heterocycle significantly improves pharmacological activities and physicochemical properties [18]. In addition, the effects of EWG on the aromatic ring at N-9 are considered as well as the replacement of a hydrogen atom at the C-2 position by a chlorine atom. Later, the synthesized purine derivatives are evaluated for their affinity to the human H 3 R and the most promising compounds are evaluated for their cytotoxicity on cell lines. Finally, docking and molecular dynamic studies are carried out to understand the effects of chemical modifications on affinity to the H 3 R. Considering the great need to find new therapeutic alternatives for the treatment of neurodegenerative diseases, this work contributes to medicinal chemistry through the discovery of easily accessible compounds with high Pharmaceuticals 2022, 15, 573 3 of 14 affinity to the H 3 R, with low toxicity in the cell lines evaluated and with a promising ADME profile.

H3R Affinity
Compounds 3a-h were screened for their affinities to the human H3R by performing competitive radioligand binding assays using [ 3 H]N α -methylhistamine as a radioligand ( Table 1). All compounds presented Ki values in the nanomolar range, confirming that the purine heterocycle provides derivatives with a high affinity to the receptor and that some substitution patterns are important in structure-activity relationships. Firstly, the presence of a chlorine atom at the C-2 position maintains or slightly decreases affinity to H3R, as observed in 3a-d versus 3e-h compounds. In the arbitrary zone, the presence of the pfluorophenyl portion yields the derivatives with the lowest affinity (3a and 3e), and this is an interesting result that correlates with our previous works [13]. Finally, the substitution in N-9 with 2,6-dichlorobenzyl provides the most active derivatives of this series, 3d and 3h, which present Ki values of 2.91 nM and 5.51 nM, respectively, being even better than those presented by pitolisant ( Figure 2). In this case, a purine scaffold appears to

H 3 R Affinity
Compounds 3a-h were screened for their affinities to the human H 3 R by performing competitive radioligand binding assays using [ 3 H]N α -methylhistamine as a radioligand ( Table 1). All compounds presented K i values in the nanomolar range, confirming that the purine heterocycle provides derivatives with a high affinity to the receptor and that some substitution patterns are important in structure-activity relationships. Firstly, the presence of a chlorine atom at the C-2 position maintains or slightly decreases affinity to H 3 R, as observed in 3a-d versus 3e-h compounds. In the arbitrary zone, the presence of the p-fluorophenyl portion yields the derivatives with the lowest affinity (3a and 3e), and this is an interesting result that correlates with our previous works [13]. Finally, the substitution in N-9 with 2,6-dichlorobenzyl provides the most active derivatives of this series, 3d and 3h, which present K i values of 2.91 nM and 5.51 nM, respectively, being even better than those presented by pitolisant ( Figure 2). In this case, a purine scaffold appears to increase affinity to H 3 R compared to derivatives with pyrrolo[2,3-d]pyrimidine [13], but this observation is not decisive, since this tendency is not observed in other derivatives. increase affinity to H3R compared to derivatives with pyrrolo[2,3-d]pyrimidine [13], but this observation is not decisive, since this tendency is not observed in other derivatives.

Cytotoxicity Assays
The potential general nephrotoxic, neurotoxic and hepatotoxic effects of selected H3R ligands 3d and 3h were investigated to determine their influence on the viability of normal

Cytotoxicity Assays
The potential general nephrotoxic, neurotoxic and hepatotoxic effects of selected H 3 R ligands 3d and 3h were investigated to determine their influence on the viability of normal HEK-293, neuroblastoma SH-SY5Y and hepatoma HepG2 cell lines by a colorimetric MTT reduction assay (Table 2). Tested compounds showed similar and weak influences on the viability of all examined cell lines, which may indicate that both compounds would not be potentially cytotoxic in future in vivo studies.

Docking Studies
Although the crystal structure of the H 3 R receptor is not yet available, the inactive H 3 R model was downloaded from the GPCRdb website (https://gpcrdb.org, accessed on 1 November 2021), which comes from homology modelling using the crystal of the histamine H 1 receptor (PDB ID: 3RZE) as a template [19]. Subsequently, an energetic minimization of the residues was performed to adjust the dihedral angles of each amino acid. According to the literature, the critical residues of the orthosteric binding pocket are Asp114 and Glu206. Both residues assist in recognizing the endogenous ligand ( Figure S1), forming salt bridges of the carboxylates of these residues with the cationic amines of the synthesized ligands [20].
Molecular docking studies were carried out to understand the binding mode of the most active compound, 3d, within the H 3 R binding site. A standard protocol was used to perform induced-fit docking, which generated four binding poses for 3d. The four best poses were analyzed based on their docking scores and polar and non-polar interactions. The best poses showed docking scores of −10.23 kcal/mol with interaction with Asp114 and of −9.18 kcal/mol with Glu206 ( Figure S2). Therefore, the lowest energy pose was selected, and the other synthesized compounds were screened. Once the pose was obtained for each ligand, an MM-GBSA calculation was performed, energetically minimizing the residues composing the binding site at about 6 Å from the center of mass of the ligands. Finally, the binding free energy was calculated for each of them (Table 3). With an R 2 of 0.71, the correlation between the estimated affinity energy and the experimental data (-logK i ) was reached, indicating that the resulting posture is the best representative of biological activity and affinity with this receptor (Figure 3).  Ligand-receptor interactions can be divided into three regions constituting the synthesized ligands (Figure 4), where the basic moiety interacts firstly with a hydrophobic pocket (hydrophobic pocket 1, HP1) composed of three aromatic rings of residues Tyr91, Phe398 and Trp402. In addition, there is an electrostatic interaction between the cationic amine of piperidine with Asp114. Next, the aromatic region of purine forms a -stacking interaction with Tyr115 and a hydrogen bond with the N-7 of purine core with the hy- Ligand-receptor interactions can be divided into three regions constituting the synthesized ligands (Figure 4), where the basic moiety interacts firstly with a hydrophobic pocket (hydrophobic pocket 1, HP1) composed of three aromatic rings of residues Tyr91, Phe398 and Trp402. In addition, there is an electrostatic interaction between the cationic amine of piperidine with Asp114. Next, the aromatic region of purine forms a π-stacking interaction with Tyr115 and a hydrogen bond with the N-7 of purine core with the hydroxyl of Tyr374. Finally, the halogenated aromatic region interacts with a second hydrophobic pocket (HP2) consisting of Tyr374, Trp371, Phe207 and Phe367. This region interacts favorably with the electron-deficient ring of the synthesized ligands, and halogen substitution has a direct impact on the favorable hydrophobic interaction, with the para-position being the least favorable due to a steric hindrance it causes with Phe207, which coincides with substitutions in the ortho-position being the most active, as is the case for compounds 3d and 3h. Ligand-receptor interactions can be divided into three regions constituting the synthesized ligands (Figure 4), where the basic moiety interacts firstly with a hydrophobic pocket (hydrophobic pocket 1, HP1) composed of three aromatic rings of residues Tyr91, Phe398 and Trp402. In addition, there is an electrostatic interaction between the cationic amine of piperidine with Asp114. Next, the aromatic region of purine forms a -stacking interaction with Tyr115 and a hydrogen bond with the N-7 of purine core with the hydroxyl of Tyr374. Finally, the halogenated aromatic region interacts with a second hydrophobic pocket (HP2) consisting of Tyr374, Trp371, Phe207 and Phe367. This region interacts favorably with the electron-deficient ring of the synthesized ligands, and halogen substitution has a direct impact on the favorable hydrophobic interaction, with the paraposition being the least favorable due to a steric hindrance it causes with Phe207, which coincides with substitutions in the ortho-position being the most active, as is the case for compounds 3d and 3h. The same protocol was repeated to obtain the best poses with pitolisant, considering the interaction of protonated piperidine with Asp114 and Glu206. However, the difference in affinity energy between both poses was not so vast ( Figure 5). Additionally, apart from an electrostatic interaction of the carboxylate with Asp114 or Glu206 (regardless of what the case may be), the ether oxygen interacts, forming a hydrogen bond with Tyr374, thus highlighting the importance of this residue for the affinity for the binding site. The same protocol was repeated to obtain the best poses with pitolisant, considering the interaction of protonated piperidine with Asp114 and Glu206. However, the difference in affinity energy between both poses was not so vast ( Figure 5). Additionally, apart from an electrostatic interaction of the carboxylate with Asp114 or Glu206 (regardless of what the case may be), the ether oxygen interacts, forming a hydrogen bond with Tyr374, thus highlighting the importance of this residue for the affinity for the binding site.

Molecular Dynamic Studies
To understand the affinity differences between the most active compound and the less active compound, molecular dynamics of 50 ns were performed with the ligand-receptor complexes of compound 3d (most active compound) and 3a (least active compound). We considered the phospholipids that constitute the lipid membrane and a solvated system (water), and we obtained a transmembrane system.
The trajectories were analyzed with the Simulation Interaction Diagram module of

Molecular Dynamic Studies
To understand the affinity differences between the most active compound and the less active compound, molecular dynamics of 50 ns were performed with the ligand-receptor complexes of compound 3d (most active compound) and 3a (least active compound). We considered the phospholipids that constitute the lipid membrane and a solvated system (water), and we obtained a transmembrane system.
The trajectories were analyzed with the Simulation Interaction Diagram module of the Schrodinger Suite. The ligand-protein interactions revealed the residues that persist the most during molecular dynamics. The H 3 R-3d complex interacts with Asp114 with almost 99% during the whole trajectory, with an average distance of 2.06 Å between the amine proton of the piperidine and the carboxylate, whereas 3a has an interaction with 88% of the molecular dynamics with an average distance of 2.22 Å (Figure 6). Another important aspect is the presence of hydrogen bonding of Tyr374 with the N-7 of purine in compound 3d with a presence of 36%, which, in turn, caused an enhanced π-stacking interaction with Trp371 with 45% during molecular dynamics. Considering the molecular dynamics results, both systems, H3R-3d and H3R-3a, reveal positive interactions at the binding site. However, 3d interacts preferentially with TM6 domain residues (Trp371 and Tyr374). In contrast, 3a interacts with Tyr115, a residue belonging to TM3, and with Phe192, an amino acid residue belonging to the extracellular loop ( Figure 6). Therefore, the inverse agonist action of the most active compound (3d) may be explained by the interaction between the TM3 (Asp114) and TM6 (Trp371 and Tyr374) domains, similar to that which occurs with other synthesized ligands that have a similar mechanism of action [21,22].

In Silico ADME and Drug-Likeness Properties
Employing the SwissADME online webserver [23], compounds 3a-h were analyzed to predict their physicochemical properties, in relation to absorption, bioavailability and blood-brain barrier (BBB) permeability. Lipinski's rule establishes that an orally bioavailable drug should not violate the following criteria: ≤5 hydrogen bond donors (HBD); ≤10 hydrogen bond acceptors (HBA); a MW of <500 g/mol; and a log p value of <5 [24]. Moreover, Veber et al. described the magnitude of PSA and the number of rotatable bonds as criteria to rate oral bioavailability. Veber's rule states that, to be orally bioavailable, a molecule should have either a PSA ≤ 140 Å and ≤10 rotatable bonds (NRB) or ≤12 HBD and HBA in total and ≤10 rotatable bonds [25]. As shown in Table 4, compounds 3a-h accomplished the drug-likeness criteria described by Lipinski and Veber; hence, they are expected to have good oral bioavailability. Moreover, according to the brain or intestinal estimated permeation method (BOILED-Egg), all compounds should have good gastrointestinal absorption and should cross the blood-brain barrier [26], and this is an important result considering the CNS distribution of the H3R. Furthermore, the SwissADME server The molecular dynamics of the H 3 R-3a system demonstrates the importance of the halogen atom in the para position of the aromatic ring and its low biological activity, which responds to the steric hindrance generated by the halogen with the Phe207 residue, causing a very poor interaction with Tyr374 and Trp371 but improving hydrophobic interactions with Tyr115 and Phe192.
Considering the molecular dynamics results, both systems, H 3 R-3d and H 3 R-3a, reveal positive interactions at the binding site. However, 3d interacts preferentially with TM6 domain residues (Trp371 and Tyr374). In contrast, 3a interacts with Tyr115, a residue belonging to TM3, and with Phe192, an amino acid residue belonging to the extracellular loop ( Figure 6). Therefore, the inverse agonist action of the most active compound (3d) may be explained by the interaction between the TM3 (Asp114) and TM6 (Trp371 and Tyr374) domains, similar to that which occurs with other synthesized ligands that have a similar mechanism of action [21,22].

In Silico ADME and Drug-Likeness Properties
Employing the SwissADME online webserver [23], compounds 3a-h were analyzed to predict their physicochemical properties, in relation to absorption, bioavailability and bloodbrain barrier (BBB) permeability. Lipinski's rule establishes that an orally bioavailable drug should not violate the following criteria: ≤5 hydrogen bond donors (HBD); ≤10 hydrogen bond acceptors (HBA); a MW of <500 g/mol; and a log p value of <5 [24]. Moreover, Veber et al. described the magnitude of PSA and the number of rotatable bonds as criteria to rate oral bioavailability. Veber's rule states that, to be orally bioavailable, a molecule should have either a PSA ≤ 140 Å and ≤10 rotatable bonds (NRB) or ≤12 HBD and HBA in total and ≤10 rotatable bonds [25]. As shown in Table 4, compounds 3a-h accomplished the drug-likeness criteria described by Lipinski and Veber; hence, they are expected to have good oral bioavailability. Moreover, according to the brain or intestinal estimated permeation method (BOILED-Egg), all compounds should have good gastrointestinal absorption and should cross the blood-brain barrier [26], and this is an important result considering the CNS distribution of the H 3 R. Furthermore, the SwissADME server resolved that compounds 3a-h should have no pan-assay interference liability (PAINS). In addition, the SwissADME server indicated that the bioavailability radar chart for 3d was within the desired range (pink region) for six parameters used for oral absorption prediction: flexibility (FLEX), lipophilicity (LIPO), solubility (INSOLU), size and polarity (POLAR) and saturation (INSATU) (Figure 7), which can foresee their predicted good oral bioavailability.

Chemistry
Melting points were determined on a Thermogeräte Kofler apparatus (Reichert, Werke A.G., Wien, Austria) and were uncorrected. All reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise specified. The NMR spectra were recorded at NMR Bruker AV 400 or 200 MHz. Chemical shifts were given in parts per million relative to TMS [ 1 H and 13 C, δ (SiMe 4 ) = 0]. Most NMR assignments were supported by additional 2D experiments. HRMS-ESI-MS experiments were carried out using a high-resolution mass spectrometer, Exactive™ Plus Orbitrap, ThermoFisher Scientific (Bremen, Germany). Thin layer chromatography (TLC) was performed using Merck GF-254 type 60 silica gel. Column  [27]. 1  The N-benzyl-purines 2a-h (1.0 mmol), 4-piperidinopiperidine (2.0 mmol), triethylamine (3.0 mmol) and ethanol (5 mL) were added to a microwave reaction flask, and the reaction mixture was irradiated for 15 min at 80 • C. Then, the solvent was evaporated under vacuum, and the crude product was purified by column chromatographic on silica gel using a (8:2) chloroform/methanol mixture.

Human Histamine H 3 Radioligand Displacement
Radioligand displacement assays at the human histamine H 3 R were performed, as published previously [28]. In summary, the compounds were incubated with membrane preparations of HEK-293 cells stably expressing the human histamine H 3 R (20 µg/well) and [ 3 H]N α -methylhistamine (2 nM, K D = 3.08 nM) in 200 µL of binding buffer (10 mM MgCl 2 , 100 mM NaCl and 75 mM Tris/HCl, pH 7.4) for 90 min. At least three independent experiments of each compound in duplicate were conducted with 11 concentrations. Obtained data was analyzed with GraphPad Prism 7 (San Diego, CA, USA) using non-linear least squares fit and the equation "one site competition" (representative Figures given in the SI). K i values were calculated according to the Cheng-Prusoff equation [29] and are reported as means with a 95% confidence interval. Statistical analyses were performed on the -logK i values and were converted to mean K i values and a 95% confidence interval. Differences are considered significant if 95% confidence intervals are not overlapping.

Cell Cultures
Cultures of HEPG2 (ATCC-HB-8065 TM), HEK293 (ATCC CRL-1573 TM) and SHSY-5Y (ATCC CRL-2266 TM) cell lines were used as experimental models. Cells were maintained in a DMEM culture medium supplemented with 10% Foetal Bovine Serum (FBS), in an incubator at 37 • C and a humidified atmosphere with 5% CO 2 . Cells were seeded on culture plates, were allowed to grow for 24 h and then were treated with the synthesized compounds for 48 h with 100, 10, 1 and 0.1 µM concentrations. Negative cell death control, corresponding to untreated cells, and positive cell death control, corresponding to cells treated with 10% SDS, were maintained.

Cytotoxicity Assays
In a colorimetric MTT reduction assay [30], sterile 96-well culture plates were used. An amount of 2000-2500 cells per well were seeded. Upon completion of treatment, the culture medium was vacuum aspirated and washed with 1x PBS, and then 100 µL of 0.5 mg/mL MTT solution was added for a period of 4 h at 37 • C for formazan crystal formation. At the end of the 4 h, 100 µL of 10% SDS in HCl was added and kept overnight in an incubator at 37 • C to dissolve the formed crystals. The absorbance of each well was measured on the Synergy HTX microplate reader at 570 nm wavelength.

Docking Studies
The 3D structures of the compounds that were synthesized were built using OECHEM, followed by protonation states, and they were adjusted to a pH of 7.2 using FixpKa from the QUACPAC package (QUACPAC 2.1.3.0: OpenEye Scientific Software, Santa Fe, NM. http://www.eyesopen.com, accessed on 1 November 2021). The conformers were generated using OMEGA software [31]. The H 3 R model was downloaded from the GPCR database (https://gpcrdb.org, accessed on 1 May 2022) [32][33][34]. The preparation of the proteins was performed with Schrodinger Suite 2021-1 [35], and the minimization was performed in the presence of restraints to maintain the protein conformation utilizing force field, OPLS4 [36].
The molecular docking studies were carried out with the induced-fit docking protocol [37,38], considering possible interactions with Asp114 or Glu206. The best poses were refined with the MM-GBSA method [39] in order to obtain the binding free energy by taking into account the solvation energies of the interacting molecules in addition to the molecular mechanics (MM) energies. The contribution of polar solvation energies was computed by the generalized Born (GB) implicit solvent model, whereas the nonpolar contribution of the solvation energy was dependent on solvent-accessible surface area (SA).