Design, synthesis, and anticonvulsant effects evaluation of nonimidazole histamine H3 receptor antagonists/inverse agonists containing triazole moiety

Abstract Histamine H3 receptors (H3R) antagonists/inverse agonists are becoming a promising therapeutic approach for epilepsy. In this article, novel nonimidazole H3R antagonists/inverse agonists have been designed and synthesised via hybriding the H3R pharmacophore (aliphatic amine with propyloxy chain) with the 1,2,4-triazole moiety as anticonvulsant drugs. The majority of antagonists/inverse agonists prepared here exerted moderate to robust activities in cAMP-response element (CRE) luciferase screening assay. 1-(3-(4-(3-Phenyl-4H-1,2,4-triazol-4-yl)phenoxy)propyl)piperidine (3l) and 1-(3-(4-(3-(4-chlorophenyl)-4H-1,2,4-triazol-4-yl)phenoxy)propyl)piperidine (3m) displayed the highest H3R antagonistic activities, with IC50 values of 7.81 and 5.92 nM, respectively. Meanwhile, the compounds with higher H3R antagonistic activities exhibited protection for mice in maximal electroshock seizure (MES)-induced convulsant model. Moreover, the protection of 3m against the MES induced seizures was fully abrogated when mice were co-treated with RAMH, a CNS-penetrant H3R agonist, which suggested that the potential therapeutic effect of 3m was through H3R. These results indicate that the attempt to find new anticonvulsant among H3R antagonists/inverse agonists is practicable.


Introduction
Epilepsy, a very common neurologic disorder, affects about around 1% of world population 1,2 . Presently, antiepileptic drugs (AEDs) are the main strategy of therapy. However, the AEDs available in the clinic such as phenytoin, carbamazepine, sodium valproate, topiramate, and oxcarbamazepine are only effective in approximately 70% of the patients with epilepsy. Moreover, their use is long-term and often accompanied with severely side effects, including naupathia, headache, and ataxia, even threaten the life of patients [3][4][5] . Investigations for more effective and safer AEDs are still a formidable and urgent task for medicinal chemists.
The role of central histaminergic system being concerned in epilepsy have been demonstrated in many experimental and epidemiological studies, in which histamine regulated seizure susceptibility as an anticonvulsant neurotransmitter [6][7][8] . For example, H 1 -antagonists such as pyrilamine, ketotifen that decrease brain histamine levels increased the duration of convulsive phase in electrically-induced convulsions model 9 . Histidine, as the precursor of histamine, showed protection against chemically-induced convulsions in rats, via activating the histamine H 1 receptors 10 .
Histamine H 3 receptors (H 3 R) as a G-protein coupled receptor (GPCR) binding to histamine like other histamine receptors, is expressed mainly in the central nervous system, where it acts as an auto-receptor in histaminergic neurons, and negatively regulates the synthesis and release of histamine 11 . What is more, as a inhibitory heteroreceptor, H 3 R also regulates the release of other neurotransmitters including dopamine, acetylcholine, serotonin, norepinephrine, c-aminobutyric acid, and glutamate. These neurotransmitters, especially c-aminobutyric acid and glutamate, are related to epilepsy inextricably 12,13 . Therefore, more attention has been focussed on H 3 R as an attractive therapeutic target for epilepsy treatment 14 .
A large number of experimental studies involved in acute and chronic models of epilepsy confirmed the anticonvulsive potential of H 3 R antagonists/inverse agonists. They showed the protection against experimental seizures by feedback increase of histamine release and binding with H 1 receptors 15,16 . Besides, other mechanisms might be involved in their anticonvulsive action, such as facilitating of GABA release [17][18][19] , increasing histidine decarboxylase (HDC) activity 20,21 and synergism with AEDs 17,18,22 .
Early, anticonvulsant activity of some imidazole H 3 R antagonists such as thioperamide and clobenpropit was confirmed in models of epilepsy ( Figure 1) 16,19,20,23 . Recently, a large number of nonimidazole H 3 R antagonists such as DL77 (Figure 1) prepared by a group/team of Kiec-Kononowicz exhibited excellent anticonvulsant activity in the electrically-induced seizures model and subcutaneously pentylenetetrazole (PTZ)-induced seizure model at dosedependent, and the therapeutic action was proved through H 3 R 24-28 . Sadek et al. synthesised some histamine H 3 R ligands (Figure 1, I) incorporating different antiepileptic structural motifs to investigate if the H 3 R pharmacophore could be combined to some antiepileptic molecules, and give some new anticonvulsants by the multiple-target approaches. The results were encouraging, which indicated that the H 3 R pharmacophore successfully combined to the antiepileptic molecules, maintaining the H 3 R affinity and anticonvulsant activity, although the anticonvulsant activity decreased compared to the prototypal antiepileptic molecules (Figure 1) 29,30 . Pitolisant (PIT), a H 3 R antagonist/inverse agonist, has been subjected into clinical Phase III for the treatment of epilepsy 31 . When used alone or in combination with other AEDs in the human photosensitivity model at dose ranges of 30-60 mg, PIT showed a favourable EEG profile in a dose-dependent manner 32 .
Supported by the above results, in this work, we designed and synthesised some novel H 3 R antagonists/inverse agonists by hybriding the H 3 R pharmacophore (aliphatic amine with propyloxy chain) with the 1,2,4-triazole, the latter have been identified as an important and effective anticonvulsive fragment in recent years ( Figure 1, II and III) [33][34][35][36][37] . According to Quan's reports, the 1,2,4-triazole derivatives were likely to have several mechanisms of action such as inhibiting voltage-gated sodium ions channel and modulating GABAergic activity [38][39][40] . And a group of Plech illustrated the anticonvulsive effects of 4-alkyl-5-aryl-1,2,4-triazole-3-thione derivatives and suggested that the influence on the voltage-gated Na þ channels was involved in them at least 41,42 . Therefore, in this work, our strategy was to design molecules combining pharmacophores of H 3 R antagonists and another anticonvulsant active pharmacophore (e.g. 1,2,4-triazole moiety) into one skeleton, and then produced a synergism for anticonvulsant active.
Their structures were characterised and confirmed by 1 H-NMR, 13 C-NMR, and HR-MS.

Biological activities
2.2.1. Evaluation of H 3 R antagonistic activity cAMP-response element (CRE) reporter gene assay has been extensively used to evaluate the efficacy of GPCR antagonists or agonists. In this work, the H 3 R antagonistic activities of the prepared 3-(4-(4H-1,2,4-triazol-4-yl)phenoxy)-propylamine derivatives have been screened by CRE-driven luciferase assay, in which the HEK-293 cells expressing the human H 3 R and a reporter gene consisting of the firefly luciferase coding region were used 43,44 . Ciproxifan (CXP) and Pitolisant (PIT) were employed as the positive controls. Initially, compounds and positive controls were tested at two concentrations (100 nM and 1 mM) to obtain the preliminary investigation of their H 3 R antagonistic activities. In the assays, the antagonistic activities was positively correlated with the rise of the fluorescence value and indicated by the % antagonism. For the prominent compounds IC 50 values were determined at additional assays.
As seen in Table 1, majority of the synthesised compounds displayed gentle to robust H 3 R antagonistic activities and eight of them exhibited micromolar inhibitory activity. The antagonistic activities of all compounds depended on the concentration treated. It is worth mentioning that compounds 3l (IC 50 ¼ 7.81 nM) and 3m (IC 50 ¼ 5.92 nM) displayed the most potent H 3 R antagonistic activities, with the much stronger potency than that of CXP (IC 50 ¼ 0.082 mM) and PIT (IC 50 ¼ 0.5 mM) in the CRE reporter gene assay.
Surprisingly, antagonism percent of some compounds as well PIT were above 100%. It is well known that H 3 R is a GPCR coupled with Gai. When the ligand (histamine) binds to H 3 R, the dissociated Gai inhibits the activity of adenylate cyclase (AC) and downregulates the level of intracellular cAMP. In this CREs driven luciferase assay, when cells were pre-treated by H 3 R antagonists, the downregulation of cAMP would be inhibited, and the level of intracellular cAMP would regain to the initial level. However, in some cases, the cAMP levels raised above the initial level, giving the % antagonism greater than 100%. Our first speculation was that these compounds might stimulate the AC directly and upregulate the level of cAMP. However, an additional assay indicated that these compounds didn't have effects on the level of cAMP when pre-treated alone. As shown in Figure 2, forskolin (2 mM) treated group gave more than 200 times rise for the cAMP level when compared to the control group. While Histamine, ciproxifan, and compounds 3a, 3c, 3h, 3k, 3l, and 3m have no significant effects on the level of cAMP when carried out comparisons by ANOVA followed by Dunnett's test.
Another explanation is that these compounds may be inverse agonists when binding to H 3 R, which not only antagonise the function of histamine, but also give the inverse agonistic performance. Actually, PIT is a well-known H 3 R antagonist and reverse agonist. So, the above results make sense. Based on the above, we further assessed the H 3 R inverse agonist activity of compound 3m and PIT by using CRE-luciferase assay. In this experiment, transfected HEK-293 cells were stimulated with 10 mM forskolin or 10 mM forskolin plus different concentrations of compound 3m. The raise of luciferase activity after adding compound represented the inverse agonistic activity. The EC 50 was calculated by seven concentrations. Cytotoxicity appeared at 100 mM. As shown in Figure 3, PIT and compound 3m showed effective H 3 R inverse agonistic activity with an EC 50 value of 403 and 129 nM, respectively.
Simple structure-activity relationships (SARs) could be obtained from Table 1. In the series of 3a-3j, the different tertiary amines significantly influenced the H 3 R antagonistic activities. The N-ethyl derivative 3a showed an IC 50 of 2.9 mM, while the activity declined sharply for the N-propyl derivative 3b. Interestingly, compounds containing piperazine or morpholine (3d-3g) exhibited weaker activities than those with piperidine or pyrrolidine (3c and 3h). This probably attributed to the increase of the molecular polarity. The introduction of phenyl or amide group on the piperidine ring of compound 3h, gave the compounds 3i and 3j, which also decreased the H 3 R antagonistic activities when compared to compound 3h. Based on the facts above, it could be concluded that the N,N-diethyl group, pyrrolidine and piperidine were more of benefit to the H 3 R antagonistic activities of the 3-(4-(4H-1,2,4-triazol-4-yl)phenoxy)-propylamine skeleton, and piperidine derivative (3h) was the best one with the IC 50 of 0.127 mM.
To enrich the structure-activity relationships, we prepared the derivatives of 3h via introducing the substituents at the triazole ring and adjusting the length of the link.
Compounds 3k, 3l, 3m, and 3n were substituted on 3-position of 1,2,4-triazole ring with methyl, phenyl, para-chlorophenyl, and biphenyl, respectively. Encouragingly, the introduction of methyl, phenyl, and para-chlorophenyl groups significantly increased the H 3 R antagonistic activities, giving the two prominent compounds 3l and 3m with nanomolar IC 50 values. While the biphenyl substituted compound 3n showed weaker activity when compared to 3h. Replacing the three-carbon link in the compound 3h with two-carbon, four-carbon and five-carbon links, gave the compounds 3o, 3p, and 3q, respectively. It could be seen that the length of the link had a direct impact on H 3 receptor antagonistic activities of the 3-(4-(4H-1,2,4-triazol-4-yl)phenoxy)-propylamine derivatives. The activity order of the link length of carbon was 3 > 2 > 4 ) 5.
To investigate the molecular determinants that manage the antagonistic activities of the tested compounds, molecular docking studies of PIT, 3h, and 3m with the H 3 R homology model were carried out. The homology model was constructed from the crystal structure of the H 1 receptor (PDB ID: 3RZE) 45 . The docking results are shown in Figure 4.
As shown in Figure 4(A), PIT bound to H 3 R through two critical H-bond interactions with Tyr115 and Glu206, and other interactions with amino acid residues Arg381, Phe193, Met378 and so on. Figure 4(B) revealed that compound 3h had a similar binding pattern to PIT, interacting with the same amino acid residues Glu206, Tyr115, Arg381, and Met378. Surprisingly, the compound  3m with the highest H 3 R antagonistic activity showed a different binding pattern to PIT (as seen in Figure 4(C)). The overlying pattern of PIT, 3h, and 3m was shown in Figure 4(D). The piperidine group of 3m was not involved in the formation of the salt bridge or hydrogen-bond interactions with Glu206, which was generally considered as the critical residue of H 3 R 46,47 . The unexpected binding pattern of 3m might be due to the phenyl group on the triazole ring, which did not fit into the hydrophobic cavity in TMs 3-5-6 region of H 3 R, even though the compound 3m showed a forceful binding with H 3 R via another mode. The triazole nitrogen established an ionic bond with Glu206, and a hydrogen bond was observed between piperidine nitrogen and Tyr115. p-p shaped, and alkyl interactions with Trp371, Tyr343, Arg381, His187, Leu199, and ALA202 were observed to support the forceful binding with H 3 R.

Anticonvulsant activity evaluation
To investigate the anticonvulsive effects, all the target compounds (3a-3q) were screened in the MES-induced and PTZ-induced convulsion models in mice. Compounds were administered intraperitoneally (i.p.) to mice at dosage of 10 mg/kg in the both models. PIT and valproic sodium (VPA) were used as positive controls in the tests.

Protective effects of H 3 R antagonists/inverse agonists 3a-3q on MES-induced convulsions.
Protection for the mice was defined as the reduction or abolition of the tonic hind limb extension (THLE) in the MES model in mice. As seen in Figure 5, compounds 3a, 3 g, 3h, 3l, 3m, and 3o showed moderate protection for the electro-stimulated mice with significant difference from that of the control group (p < 0.05, p < 0.01, or p < 0.001). Mice pre-treated with PIT (10 mg/kg, i.p.) and VPA (300 mg/kg, i.p.) were moderately or potently protected, respectively. Generally, the anticonvulsant activities of these compounds in MES model correlated directly to their H 3 R antagonistic activities. For example, the antiepileptic activity obtained of compound 3m was the highest, and in vitro H 3 R antagonistic activity measured for 3m with IC 50 of 5.92 nM was also the highest. Compounds 3a, 3h, 3l, and 3o, showing anticonvulsant activity in the MES model, also showed good H 3 R antagonistic activities. Compound 3c, 3k, and 3p reduced the average duration of THLE, although they did not achieve a significant difference from the control group.

Protective effects of H 3 R antagonists/inverse agonists 3a-3q on PTZ-induced convulsions. Some experiments indicated that H 3 R antagonists/inverse agonists could protect animals in PTZ-
induced convulsions model 26,29 . So the compounds 3a-3q, PIT, and VPA were also screened in the PTZ model in mice. Unfortunately, all compounds tested at the dose of 10 mg/kg (i.p.) did not show any protection against the seizures induced by PTZ.
PIT also failed to protect the PTZ-treated mice as well at the same conditions. By contrast, anticonvulsant agent VPA showed full protection against the PTZ-induced convulsions ( Figure 6).

Effects of compound 3m on MES-induced convulsions in dose dependent manner.
In a further experiment, compound 3m, as the most active one in the MES-induced seizure model, was chosen to verify its protective effect in different doses. Encouragingly, the 3m-provided protections were observed and were dose dependent. The standard antagonist PIT also displayed anticonvulsive activity dose-dependently at the same condition. Notably, when pre-treated with 20 mg/kg dose, PIT could fully abrogate the tonic hind limb extension induced by electro-stimulation, showing its potential anticonvulsant activity (Figure 7). To exclude the possibility that the anticonvulsant activity of 3m was connected with sedative effect, we carried out a rotarod test for 3m. The result showed that compound 3m had no neurotoxicity at the maximum dose of 10 and 20 mg/kg (the details could be seen in Support Table 1).

2.2.2.4.
Effects of RAMH pre-treatment on the compound 3m-provided protection in MES-induced seizure model. To investigate the correlation between the anticonvulsant activity and H 3 R antagonistic activity of compound 3m, the protection provided by compound 3m against MES-induced seizure was reassessed after the administration of RAMH (10 mg/kg, i.p.), a CNS penetrant histamine H 3 R agonist. The results indicated that when co-administration with RAMH, compound 3m lost its original protective effect ( Figure 8). Administration of RAMH alone also did not affect the duration of THLE of mice with p > 0.05 for saline versus RAMH. The above findings suggested that H 3 R antagonism was the main contributor for the anticonvulsant activity of compound 3m in MES model. When the H 3 R was blocked by H 3 R antagonist 3m, . Protection in the test was defined as the reduction or abolition of the THLE in mice. Results are showed as mean ± SEM with seven animals in each group. Values are considered significant at Ã p < 0.01 as compared to saline-treated group, and # p < 0.01 as compared to 3m þ RAMH treated group. histamine or other neurotransmitter such as GABA in the CNS increased, finally leading to anticonvulsive effects.

Conclusion
To identify novel H 3 R antagonists/inverse agonists with potential anticonvulsant activities, a series of 3-(4-(4H-1,2,4-triazol-4-yl)phenoxy)-propylamine derivatives were designed through combining pharmacophore of H 3 R antagonists and another anticonvulsant active pharmacophore (1,2,4-triazole moiety) into one molecule. The majority of those prepared compounds displayed moderate to robust H 3 R antagonistic activities. The SAR analysis revealed that piperidine and triazolephenol linked by three-carbon chain was benefit for the H 3 R antagonistic activity, and substitution by aromatic nucleus on the 3-position of 1,2,4-triazole further increased the H 3 R antagonistic activities. The most potent H 3 R antagonists/ inverse agonists 3l and 3m exhibited nanomolar H 3 R antagonistic activities with IC 50 of 7.81 nM and 5.92 nM, respectively. Molecular docking analysis demonstrated that 3m strongly bound to H 3 R via interactions with Tyr115, Glu206, Trp371, Tyr343, and so on, although its binding mode was not similar to PIT. The anticonvulsive screens in vivo indicated that compounds with higher H 3 R antagonistic activities showed more protection in the MES-induced convulsant model in mice, while no one was observed protective effect in PTZ-induced convulsant model. Moreover, the protection of 3m in the seizure model was fully abrogated when mice were co-treated with a H 3 R agonist RAMH, which suggested that its potential therapeutic effect was through H 3 R.

Synthesis
All the chemical solvents and reagents were purchased from supplier and used as received. Unless otherwise specified, reactions were monitored by thin-layer chromatography (TLC). All NMR spectrum was carried out on an AV-300 spectrometer with 300 MHz. High resolution mass spectra were measured on an MALDI-TOF/TOF mass spectrometer.

Synthesis of compounds 1a-1e
Taking compound 1a as an example: dimethoxyl-N, N-dimethyl formamide (DMF-DMA, 1.31 g, 11 mmol) and formyl hydrazine (0.65 g, 11 mmol) were added into a flask containing 30 ml of acetonitrile. The mixture was heated up to 60 C for 30 min, then 4-aminophenol (0.60 g, 5.5 mmol) and acetic acid (3 mL) were added and heated up to 120 C for 9 h. The mixture was cooled, filtered and washed by acetonitrile to give the product 1a.

Synthesis of compounds 3a-3q
Taking compound 3a as an example: in a 100 mL round-bottom flask with 15 mL of acetonitrile, compound 2a (0.40 g, 1.68 mmol), diethylamine (0.245 g, 3.36 mmol), K 2 CO 3 (0.46 g, 3.36 mmol) and potassium iodide (0.56 g, 3.36 mmol) were added one by one. The mixture was heated up to 110 C for 12-16 h. After cooing the mixture to 40 C, it was filtered and dried by vacuum to obtain a residue. Purification by column chromatography (silica gel, 0-20% methanol in CH 2 Cl 2 ) gave the compound 3a. The same conditions were used to prepare the compounds 3b-3q. Compounds 3a, 3c, 3d, 3e, 3 g, 3k, 3o, and 3p obtained as oils were transformed into the corresponding hydrochloride by hydrogen chloride in CH 2 Cl 2 .