Histamine is one of the most important chemical transmitters involved in several biological processes. It has been widely studied since its discovery^1-3 due to its role in inflammatory and allergic reactions, but it is also involved in the regulation of gastric acid secretion, sleep, mood, and food intake (Passani and Blandina, 2011). These processes are triggered by the interaction of histamine with its receptors (HRs), named H₁R, H₂R, H₃R, and H₄R. HRs are members of the G protein-coupled receptor (GPCR) class A family, which are also known as 7 transmembrane (7TM) receptors (Seifert et al., 2011; Corrêa and Fernandes, 2015).

The H₃R is mainly found in CNS, and is involved in the inhibition of histamine release, as well as in the inhibition of other neurotransmitters, such as norepinephrine, acetylcholine, and serotonin (Leurs et al., 2005). Therefore, H₃R modulates neurotransmitters release by acting as a presynaptic receptor (auto- and hetero-receptor). It has been previously reported that signal transduction by H₃R occurs mainly by activation of G_i/o proteins, leading to decrease of intracellular cAMP concentration and decrease of Ca²⁺ influx in neurons (Tiligada et al., 2009). As a result of its actions, H₃R could be involved in several neurological disorders, including cognitive, convulsive, and sleep–wake disorders, as well as in obesity (Łażewska et al., 2014). Therefore, H₃R ligands could be promising drugs to treat these conditions. H₃R ligands have been widely explored. Recently, the novel H₃R antagonist pitolisant has been approved for treatment of narcolepsy, highlighting the importance of this receptor as a new target for treatment of certain CNS disorders (Sayed, 2016). Ciproxifan and ABT-239 (Figure 1) are other examples of such well-studied compounds.

In the early 2000s, several research groups described a new isoform of HR, expressed in immune cells (eosinophils, basophils, mast cells, NK cells, DCs, monocytes, and T cells), which is involved in the modulation of chemotaxis as well as other functions (Oda et al., 2000; Liu et al., 2001). This receptor was named H₄R, is also coupled to G_i/o proteins (Corrêa and Fernandes, 2015), and bears a considerable sequence identity to H₃R (∼31% total, 54% in TM domains) (Oda et al., 2000; Liu et al., 2001). Considering the physico-chemical properties of the different amino acids, this similarity in TM domains can reach 68% (de Esch et al., 2005). Since these cells are involved in immune and inflammatory processes, H₄R is considered a promising target to future anti-inflammatory and modulatory immunological drugs. H₄R ligands have been designed and evaluated as drug candidates (Figure 1), with the indolecarboxamides (JNJ-7777120) and 2-aminopyrimidines (A-943931) being the most explored chemotypes in this regard. Some molecules lacking the carbonyl group from the indolecarboxamides were also identified as H₄R ligands in virtual screening experiments (Christopher et al., 2012).

Due to the high sequence identity between H₃R and H₄R, it is likely that several compounds that bind to one of them can present considerable affinity for the other. Hence, some pharmacophore templates to H₃R ligands can also be applied to H₄R ligands (Kottke et al., 2011). In fact, the search for selective ligands that are able to discriminate one of those receptors is frequently reported in the literature, and most medicinal chemists working with HR-ligand research have such selectivity parameter as one of the most important goals. However, dual-acting H₃R/H₄R ligands may also present therapeutic potential in determined pathological conditions, such as neuropathic pain (Smith et al., 2007), cancer (Medina et al., 2008), and have also been reported in Parkinson’s disease (Shan et al., 2015a,b). Such potential applications were already reported for imidazole-containing H₃R/H₄R ligands, such as imetit, immepip, clobenpropit, and thioperamide.

To the best of our knowledge, the functional activity of dihydrobenzofuran-containing molecules have not yet evaluated in the H₃R and H₄R until the present report. The affinities to these receptors were evaluated as well. Considering the chemical similarities observed for several H₃R and H₄R ligands, we have designed a set of 1-[(2,3-dihydro-1-benzofuran-2-yl)methyl]piperazines (LINS01 series – 1a–i), evaluated their affinity and selectivity for those receptors, and characterized their functional activities. Drug-likeness was also assessed to define some structure-activity relationship (SAR) roles in this set of compounds, and an evaluation of the anti-inflammatory potential of a selected molecule was carried out in a mouse asthma model.

Reagents and starting materials were obtained from commercial suppliers (Sigma–Aldrich Co., Saint Louis, MO, United States; LabSynt Co., Diadema, Brazil) and used without further purification. ¹H and ¹³C NMR spectra were recorded in a Bruker Ultrashield 300 spectrometer, operating at 300 and 75 MHz, respectively, using CDCl₃ as solvent with TMS as internal standard. Chemical shifts are reported in parts per million (ppm, δ units). Coupling constants (J) are reported in units of hertz (Hz), if applicable. The high resolution mass spectra (HRMS) were obtained through direct injection after electron-spray ionization in positive mode (ESI+) in a MicroTOF from Bruker Daltonics mass spectrometer. Gas chromatography coupled to mass spectrometer (GC-MS) analysis were done in a Shimadzu GC-2010 coupled to mass spectrometer GCMS-QP2010 plus, using helium as carrier gas in a silica capillary column. The low resolution mass spectra (LRMS) were obtained through electron impact ionization (70 eV). The ion-radical and its fragments are reported in mass/charge ratio (m/z). The purity (>95%) for the final compounds 1a–i were determined and confirmed (H₂O/MeOH 50%) by HPLC in a C-18 column coupled to UV detector (254 nm). Only compounds with >95% purity were considered to the biological assays. The yields for each step are summarized in Table 1. Spectral data and characterization for the intermediates can be found in Supplementary Information.

The design of LINS01 molecules was done considering the presence of an aromatic nucleus (aryl or heteroaryl group) linked to a polar moiety (imidazole, pyrrolidine, or piperazine) observed in H₃R and H₄R ligands. The presence of an extra lipophilic group attached to the polar moiety was also evaluated. In a previous paper from Fernandes et al. (2011), a QSAR study defined the importance of the R2 group for the affinity for H₄R in indolecarboxamides (e.g., chlorine in JNJ-7777120). The role of the R2 group is also important for ligand affinity to H₃R, which are represented by the cyclopropyl ketone and 4-cyanophenyl groups in H₃R ligands such as ciproxifan and ABT-239, respectively (Cowart et al., 2004). Considering this, the present set of molecules includes the substitutions not only in piperazine nitrogen (R1), but also in the dihydrobenzofuran moiety (R2) to determine preliminary SAR data for these compounds. The results show that the R2 substituents can play an important role in the binding affinities, being possibly more important than the substituent R1 to drive the receptor selectivity.

The 2-allylphenol was used as a starting material (Figure 2) for compounds 1a–d, which was converted into the (2-iodomethyl)-2,3-dihydrobenzofuran, as previously described by our group, giving excellent yield (Table 1 and Supplementary Information) (Corrêa et al., 2016, 2017). The iodinated dihydrobenzofuran 4a was then reacted with the corresponding piperazines in considerable excess, giving moderate to good yields. The yields of this step are presented in Table 1. A similar approach was applied to prepare the commercially unavailable 1-allylpiperazine 5 (Supplementary Figure S1) with 40% yield. The 1-methyl and 1-phenyl piperazines are commercially available.

The compounds 1e–i were prepared starting from the corresponding 4-substituted phenol, which was allylated using allyl bromide, and then thermally isomerized through Claisen rearrangement to give the corresponding 2-allylphenols, as reported previously by Corrêa et al. (2017). Both steps resulted in good to excellent yields (Table 1 and Supplementary Information). With the 4-substituted 2-allylphenols, the final compounds were prepared in the same fashion to compounds 1a–d (Corrêa et al., 2016).

The binding analyses were initially performed in the format of a screening, to allow a quick assessment of the compounds presenting relevant affinity for H₃R and/or H₄R (Supplementary Figure S2). The results show that most of the compounds presented higher affinity for H₃R, and that only compounds 1e and 1f displayed considerable binding to H₄R. Accordingly, competition binding assays with full concentration–response curves were performed with the selected compounds in the H₃R and H₄R (Figure 3). The obtained K_i values are in the range of 0.4–10 μM (Table 2).

Drug-likeness is a qualitative evaluation of a certain compound for how drug-like it is with respect to some factors such as bioavailability and potency based on the characteristics of market drugs. The drug-likeness of the compounds was assessed through the Lipinski’s rule-of-five (Lipinski et al., 2001) and Veber’s rules (Veber et al., 2002) parameters as well as water solubility. Lipinski’s rule-of-five proposes that a molecule should have MW < 500, clogP < 5, HBD < 5 and HBA < 10 to have good oral bioavailability. Veber’s rules include in these parameters the TPSA < 140 and RotB < 10. The LINS01 compounds fulfill all these parameters (Table 3), suggesting they should have good bioavailability and possibly adequate pharmacokinetic profile. The water solubility also determines the bioavailability and other ADME processes. LINS01 molecules show adequate water solubility, as indicated by the logS values (acceptable values for > -4). With this regard, it is noteworthy the contribution of methoxy group (1g) to enhanced water solubility. The balanced hydrophilic–lipophilic character of LINS01 molecules (as indicated by logP < 5 and logS > -4 values) suggests they may have good pharmacokinetic behavior in vivo. Drug-likeness scores corroborate to this estimation, since only positive values were obtained (Table 3). Considering that market drugs and non-drugs usually present, respectively, positive and negative score values, LINS01 molecules can be considered promising drug-like compounds to further optimization according to this criteria.

In order to evaluate the potency of these compounds, metric LE, LLE, LELP, and GE values were calculated (Table 4). During drug discovery and development process, the MW and lipophilicity usually increase in consonance with the potency. This can be explained by the contribution of the inserted non-hydrogen atoms to binding energy. LE is considered a weighted value for non-hydrogen atoms contributions to binding energy, which allows to compare affinities of molecules with different sizes (Hopkins et al., 2014). Ligands with LE > 3 are considered promising drug-like compounds. For comparison, the mean value for oral drugs is around 4.5. Considering that lipophilicity also influences on binding affinity, the LLE and LELP values are lipophilicity-weighted LE values. The ideal LLE value for an optimized drug candidate is ∼5–7, while LELP optimal range values are between -10 and 10. GE is a measurement of the binding efficiency for an added functional group in the parent molecule (Verdonk and Rees, 2008). Accordingly, it represents how much the added group contributed to overall affinity of the molecule, considering the number of heavy atoms. For example, a GE = 0.31 should represent at least a gain of 1.7-fold in the potency of a small molecule (MW < 500). The higher the GE value, the higher the contribution of the group to the potency.

Regarding the H₃R, the N-allyl derivative 1c showed the highest affinity in the series, showing a pK_i value of 6.40, followed by the chloro, methoxy, and t-butyl derivatives 1e, 1g, and 1h, with pK_i values around 6.07, and the methyl derivative 1f with pK_i 6.15. The non-alkylated piperazine compound 1a, as well as the N-phenyl derivatives 1d and 1i did not show appreciable affinity for H₃R, as mentioned before. The binding efficiencies of the compounds 1b, 1c, and 1e–h are considered adequate, since they present LE > 3 and with exception of compounds 1g and 1h, ≥4.5, indicating that although their low MW, their efficiencies are comparable to a real drug. LELP and LLE values suggest that lipophilicity is relatively high in these molecules, thus suggesting that polar groups (which can improve binding affinity) can be inserted for improvement in a future set of compounds. We believe that H₃R may have a considerable different binding pocket than H₄R in the R1 and R2 regions, which allow the interaction of bulky hydrophobic groups (such as allyl in R1 and t-butyl in R2). However, some steric hindrance effects might also play a role (especially in the R1 region), limiting the interaction of bulkier groups such as the phenyl group of 1d or 1i, but allows the interaction of the allyl group present in 1c. The results also suggest that the presence of an alkyl group linked to the polar moiety of the molecule increases the affinity for H₃R, as can be seen by the high GE values for methyl (1b) and allyl (1c) groups (Table 4).

Several 2-aminoethylbenzofurans have already been evaluated as H₃R ligands, showing good affinity (Cowart et al., 2004). These compounds can be viewed as rigid analogs of ciproxifan, such as ABT-239. Hence, the compounds here presented can also be considered rigid derivatives of ciproxifan, but lacking aromaticity in the furan ring. As observed in ABT-239 analogs, small alkyl substituents in the polar moiety increase H₃R affinity. This could be the explanation for the poor affinity of 1a and moderate affinities for 1c and the N-methyl derivatives 1e–h (Figure 3 and Table 2).

Since ABT-239 and its analogs possess a 5-aryl substituted benzofuran and better affinity, substituents at the 5-position were explored to evaluate the role of the R2 group in the dihydrobenzofuran derivatives. These data (Table 2) define the importance of the R2 substituent in the binding affinity to H₃R of these compounds. Although there are no significant differences between the affinity of compounds 1e–h to H₃R, the affinity gain caused by the R2 substitution can be noted by the pK_i of the non-substituted molecule 1b (5.57). Although compounds 1e–h present quite similar pK_i, the GE is very different among the substituents (Table 4). Methyl group (1f) has shown the higher contribution to binding efficacy in H₃R, while t-butyl group (1h) gave lower GE value. The presence of a polar group in R2 (such as methoxy in 1g) also suggests that hydrophobic groups are preferred in this moiety, as indicated by GE = 0.34.

In summary, substitutions in R2 seem to play an important role in H₃R binding, since substituting the hydrogen in this position led to improved affinity for the receptor. A rationale SAR for the substitution pattern cannot be defined yet, and further substituents must be explored. However, it is possible to verify that possibly larger volume is tolerated in this region of the molecule, such as the t-butyl group in 1h, which is not tolerated when interacting to the H₄R. Possibly, this group presents the same role than the cyanophenyl group in ABT-239 (Figure 4), and maybe it could generate selectivity toward H₃R. Previous report from Dastmalchi et al. (2008) suggests that bulkier molecules may display increased affinities for H₃R, possibly because there is a hydrophobic pocket nearby the R2 region which can be accessed by these groups, driving the selectivity toward H₃R over H₄R. With exception of compound 1e, the molecules with substituents in R2 presented over 10-fold selectivity for H₃R as compared to H₄R (Table 2). Other molecules presenting other bulky groups in R2 are in progress by our group to verify this possibility.

Regarding the affinity for H₄R, compound 1e was the only one to present affinity in the micromolar range (pK_i 6.06). Compound 1f has a K_i in the range of ∼10,000 nM. The remaining molecules presented negligible binding affinities, as found in our initial screening approach (see Supplementary Information) and aforementioned. Considering the similarity of the presented compounds with the high affinity H₄R antagonist JNJ-7777120, it seems that the absence of the carbonyl group in addition to the change of the indole ring to dihydrobenzofuran is detrimental to the affinity of the ligands for H₄R. In fact, the changes in affinity caused by such changes corroborate previous findings in benzofuran compounds (Engelhardt et al., 2012).

Previous reports (Venable et al., 2005; Engelhardt et al., 2012) suggested that a methyl group is the best substituent in the piperazine nitrogen of indolecarboxamide and benzimidazol carboxamide series. Higher homologs or bulkier groups lead to molecules with lower affinity for H₄R. In the present series, this pattern was also detected, since compounds 1e and 1f are N-methyl derivatives. The LE values suggest these molecules present good efficacy in the H₄R binding affinity. Moreover, LELP and LLE values also suggest that 1e and 1f represent drug-like molecules that can still be optimized regarding the hydrophilic–lipophilic balance. The present results suggest the R2 substituent is more important to the binding affinity than the R1 group, and bulkier groups (such as phenyl in 1d and 1h) lead to low binding affinities (>10,000 nM) when binding H₄R. Venable et al. (2005) evaluated several H₄R ligands with different substituents in R2 position, with chlorine being the group leading to an important increase in the affinity. It is proposed that this atom performs a key interaction with the binding pocket of H₄R. In the present report, this role was also observed, since compound 1e presented the highest affinity in the series for this receptor. The insertion of the chlorine atom resulted in more than 10-fold increase in affinity and a very high GE value (Table 4). This suggests that chlorine atom yielded in improved efficiency for 1e as H₄R ligand. In addition, when methyl is present in R2, the affinity drops significantly, and bulkier groups led to total loss of affinity. Accordingly, the H₄R seems to present steric hindrance in the R2 region, which is not observed to H₃R.

The functional activities of the LINS01 compounds were performed to evaluate their possible actions as agonists of H₃R and/or H₄R. Both H₃R and H₄R are GPCRs known to be G_i-coupled. To determine the possible activity of the compounds as agonists of this pathway, we used a BRET biosensor to assess and quantify G_i1 isoform activation.

Regarding H₃R, Figure 3C shows that all analyzed compounds were unable to trigger G_i activation even in the highest tested concentration (10,000 nM), likewise clobenpropit, a known H₃R antagonist. On the other hand, histamine (used as a positive control) yielded a strong activation signal at 10 and 10,000 nM. The results suggest that the compounds could act as antagonists and block the activation triggered by histamine. This hypothesis was indeed confirmed when cells expressing H₃R were incubated with the LINS01 compounds prior to stimulation with histamine. As can be seen in Figure 3E, preincubation with any of the tested compounds at the concentration of 10,000 nM fully blocked histamine agonistic effect. The compounds were equally efficacious in antagonizing histamine effects when compared to clobenpropit, and therefore can be considered H₃R antagonists. Interestingly, some compounds displayed a profile of reversal of BRET values, as compared to histamine alone, suggesting an inverse agonist activity.

Analyses of functional activity on H₄R showed that similarly to JNJ-777120, a known H₄R antagonist, the chlorinated compound 1e did not trigger G_i activation (Figure 3D). Possibly the specific interaction of chlorine with H₄R does not lead to activation of the receptor, but when methyl group is present an agonist mode interaction may occur and therefore a halogen-bonding interaction is suggested rather than a hydrophobic interaction to reach the antagonist binding mode. The role of chlorine in H₄R ligands was already widely studied, and it has been directly linked to the antagonist activity on G_i as well as β-arrestin activation (Nijmeijer et al., 2013). It is important to stress that histamine itself was a poor stimulant of G_i in H₄R (compare BRET values from Figures 3C,D). When evaluating the potential to block histamine activity, preincubation with the compound 1e at the concentration of 10,000 nM fully blocked histamine agonistic effect, similarly to the effect obtained for the H₄R antagonist JNJ-777120. The compound 1f, as described above, displays a moderate agonistic effect, and therefore only partially reduced histamine effect.

In order to evaluate the anti-inflammatory potential of the best H₄R blocker from the LINS01 set, compound 1e was tested in a mouse asthma model and compared to the previously tested compound 1d in two different doses (3 and 5 mg/kg, Figure 5). Experimental murine asthma was induced by sensitization and provocation with OVA. In a recent report (Corrêa et al., 2017), we verified that 1d showed anti-inflammatory activity only at 5 mg/kg dose.

In the present work, we evaluated in parallel the compounds 1e (LINS01007) and 1d (LINS01005) to rodents. The reduction in cell counts in both groups correlates well to the affinity to H₄R (see Figure 3B), supporting the involvement of this receptor in the mechanism of the observed anti-inflammatory action. The results presented in Figure 5 also suggest a dose-dependent effect for compound 1d on both total cell and eosinophil counts in bronchoalveolar lavage fluids, since a sounder reduction in cellularity was observed in the highest concentration of 5 mg/kg. This reduction can be considered more relevant due to the important reduction in eosinophilia. On the other hand, compound 1e led to low counts in both doses. Possibly even in the lower dose (3 mg/kg) the maximum efficacy was obtained.

In bronchoalveolar lavage fluids with experimental allergic asthma, enhanced cell counts are commonly observed. In particular, there is considerable increase in eosinophils, which are commonly absent or in low density in the non-sensitized animals. The involvement of H₄R in experimental murine asthma model has been demonstrated in several studies, including results with JNJ-7777120, which was capable of reducing the inflammatory infiltrations (especially eosinophils) and other asthma parameters (Neumann et al., 2013). Considering that the pharmacological profile obtained with LINS01 compounds in the total cell and eosinophil counts is very similar to that obtained with the JNJ compound, we believe the infiltration reductions are correlated to H₄R. Moreover, the higher activity observed to 1e also supports this conclusion.

In this study, we present nine LINS01 molecules with quite simple structure, which have been evaluated concerning binding affinities for the H₃R and H₄R, generating corresponding receptor subtype selectivity values. We have also performed functional analyses of G_i activation, which revealed that all tested compounds act as antagonists to different extents on both H₃R and H₄R. This is the first report regarding functional activity of such compounds. SAR information concerning this chemotype was generated and metric analyses suggest that future derivatives with improved affinity can still be obtained. In vivo evaluation in a mouse asthma model was performed and the potential anti-inflammatory activity of such molecules is exemplified by 1e, which showed to be highly effective at low doses.

MC and ÁB performed the synthesis of the compounds, and were supervised by JF at LINS-UNIFESP. LT, DD, SS, and LP performed the binding and functional assays at CC-N’s laboratory, their supervisor. AB and RL did the in vivo assays. MB generated the constructions for BRET assays and discussed obtained data. JF and CC-N wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.