Introduction

Histamine is a biogenic amine and mediates physiological and pathophysiological effects via four histamine receptor subtypes (Foord et al. 2005). The histamine H1 receptor (H1R) is involved in allergic reactions, the H2R is involved in secretion of gastric acid, the H3R is responsible for modulation of neurotransmitter release and the histamine H4 receptor (H4R) acts as an immunomodulator (de Esch et al. 2005; Thurmond et al. 2008). Recent studies support the hypothesis that the H1R and H4R possess a synergistic role in treatment of type-I allergic reactions (Thurmond et al. 2008; Deml et al. 2009). Thus, a detailed study of H1 and H4 receptor ligands at both H1R and H4R is necessary. Furthermore, such studies provide a more detailed insight into the interaction between ligand and receptor on molecular level. Besides, the understanding with regard to receptor subtype selectivity will be increased.

A wide variety of structurally diverse agonists (Hashimoto et al. 2003; Lim et al. 2005; Igel et al. 2009; Smits et al. 2009) and antagonists (Jablonowski et al. 2003; Terzioglu et al. 2004; Thurmond et al. 2004; Venable et al. 2005; Smits et al. 2008, 2009) at the histamine H4 receptor have been identified. A recent study (Deml et al. 2009) has shown that phenylhistamines and suprahistaprodifen, previously classified as “selective H1R agonists” (Leschke et al. 1995; Elz et al. 2000; Menghin et al. 2003; Seifert et al. 2003; Straßer et al. 2008, 2009), exhibit affinity to the human (h)H4R. The aim of this study was to analyse the interactions of phenylhistamines and histaprodifens between hH1R and hH4R in more detail. Therefore, we analysed several new phenylhistamine and histaprodifen derivatives as well as phenoprodifens at the hH4R (Fig. 1).

Fig. 1
figure 1

Structures of histamine, phenylhistamines, histaprodifens and phenoprodifens. Histamine 1, small phenylhistamines 2–9, bulky phenylhistamines with an additional histamine moiety 10–12, dimeric phenylhistamines 13 and 14, small histaprodifens 15–18, bulkier histaprodifens 19, 26–29, suprahistaprodifens 20–21, phenoprodifens 22–25, dimeric histaprodifen 30

Full size image

To study the pharmacology of phenylhistamines and histaprodifens at hH4R, we co-expressed hH4R-RGS19 with Gαi2 and Gβ1γ2 in Sf9 insect cells (Deml et al. 2009; Schneider et al. 2010). For pharmacological characterization, [3H]histamine competition binding assays and steady-state GTPase assays were performed. In order to obtain information about the binding mode of phenylhistamines and histaprodifens at hH4R, phenylhistamines were docked into the active and histaprodifen into the inactive state model of hH4R. Subsequently, molecular dynamic simulations, including the surrounding of the receptor, were performed. Additionally, we performed calculations of ΔΔG°solv(water → binding pocket of hH4R), corresponding to the transfer of a ligand from the aqueous phase into the binding pocket of hH4R, for six selected phenylhistamines.

Materials and methods

Materials

[γ-32P]GTP was synthesized as described (Preuss et al. 2007). [3H]Histamine (14.2 Ci/mmol) was from PerkinElmer Life Sciences (Boston, MA, USA). As liquid scintillation cocktail, Rotiszint ecoplus from Roth (Karlsruhe, Germany) was used. Phenylhistamines, histaprodifens and phenoprodifens were synthesized as described (Straßer et al. 2008, 2009). Sources of all other materials were described earlier (Seifert et al. 2003; Straßer et al. 2008).

Preparation of compound stock solutions

Chemical structures of the analysed compounds are given in Fig. 1. Compounds 1–30 were dissolved as described (Straßer et al. 2008, 2009). The final DMSO concentration in all assays was adjusted to 3% (v/v) or 5% (v/v), as appropriate for the ligands. Control experiments with histamine, dissolved in double-distilled water or dissolved in a solvent containing 50% (v/v) DMSO and 50% (v/v) double-distilled water, showed that a final DMSO concentration of 5% (v/v) did not shift pKi and pEC50 values of histamine.

Pharmacological and biochemical methods

Construction of baculoviruses was described earlier (Kelley et al. 2001; Seifert et al. 2003; Straßer et al. 2008; Schneider et al. 2010). Cell culture, membrane preparation and determination of protein concentration were performed as described previously (Seifert et al. 2003; Straßer et al. 2008). All assays were performed with Sf9 insect cell membranes, coexpressing hH4R-RGS19, Gαi2 and Gβ1γ2. Ligand concentrations were used in the range from 0.1 nM up to 316 μM, where appropriate. Competition binding assay and steady-state GTPase assay were performed as described (Straßer et al. 2008; Schneider et al. 2010). Shortly, competition binding assays were performed in the presence of 10 nM [3H]histamine. Reaction mixtures were incubated for 90 min at room temperature and shaking at 250 rpm. Bound [3H]histamine was determined by filtration through GF/C filters and liquid scintillation counting. The steady-state GTPase assays were performed in the agonist mode as described (Straßer et al. 2008). [35S]GTPγS binding assays were performed, as described (Schneider et al. 2009). Shortly, the studies were conducted in presence of 2 nM [35S]GTPγS, 1 μM GDP and 100 mM NaCl. Incubations were conducted for 120 min at room temperature and at 250 rpm. All samples were filtered through GF/C filters and bound [35S]GTPγS was determined by liquid scintillation counting. For data analysis, the software Prism 4.02 (GraphPad Software Inc., San Diego, CA, USA) was used. pKi values were calculated according to Cheng and Prusoff (1973). All data are the means ± SEM of at least three independent experiments. For comparison of two pairs of data, the significance of the deviation of zero p was calculated using the t test.

Molecular modelling

The inactive state model of hH4R was constructed by homology modelling using SYBYL 7.0 (Tripos Inc.) as described (Deml et al. 2009). For construction of the active state model, the crystal structure of opsin 3DQB (Scheerer et al. 2008) was used as template. For a detailed explanation, see Supplementary material. Phenylhistamines 2, 4, 6–9 were docked manually in the binding pocket of the hH4R active state model, whereas the histaprodifen 15 was docked into the binding pocket of the hH4R inactive state model. Subsequently, molecular dynamic simulations with the software package GROMACS 4.0.2 (Van der Spoel et al. 2005) were carried out. For the receptor, the ffG53A6 force field (Oostenbrink et al. 2004) was used. The GROMACS topologies for the ligands 2, 4, 6–9 were calculated using the PRODRG server (http://davapc1.bioch.dundee.ac.uk/prodrg/) (Schuettelkopf and van Allten 2004). For partial charges of the ligands, Gasteiger–Hückel partial charges were used. The equilibration phase and the productive phase were performed, using the system and simulation parameters, as described (Straßer et al. 2008). For calculation of ΔG°sol(ligand in water) and ΔG°sol(ligand in binding pocket of hH4R), we also used GROMACS 4.0.2 using thermodynamic integration (Straatsma and Mc Cammon 1991; Villa and Mark 2002). We performed calculations, using the following values for the coupling parameter λ, which switches the interaction between ligand and surrounding on or off, respectively: 0.0, 0.05, 0.1, 0.2, 0.3, 0.4, 0.45, 0.5, 0.55, 0.6, 0.7, 0.8, 0.9, 0.95, 0.975, 0.99, 0.995 and 1.0. To avoid abrupt transitions from one λ to the next λ, we used small steps in increasing λ. In order to achieve an acceptable simulation time, simulation times of 100 ps for each λ were used. For calculation of ΔG°sol(L, wat), describing the Gibbs energy of solvation of the ligand in water, and ΔG°sol(L, LR), describing the Gibbs energy of solvation of the ligand in the binding pocket of the ligand–receptor complex, each calculation was performed two times using different starting structures.

Prediction of thermodynamic data

In general, for the binding of a ligand L into the binding pocket of a receptor R, an equilibrium between L and R on the one hand and the ligand–receptor complex LR on the other hand is defined:

$$ {\text{L}} + {\text{R}} \rightleftharpoons {\text{LR}} $$
(1)

The corresponding ΔRG° is given by the equation

$$ {\Delta_{\text{R}}}{\text{G}}^\circ = - R{ }T{ \ln }(K) $$
(2)

wherein K represents the association constant. Additionally, ΔRG° can be described by the following equation:

$$ {\Delta_{\text{R}}}{\text{G}}^\circ = {\text{G}}^\circ \left( {\text{LR}} \right) - {\text{G}}^\circ \left( {{\text{L}},{\text{ wat}}} \right) - {\text{G}}^\circ \left( {\text{R}} \right) $$
(3)

Therein, G°(LR) describes the Gibbs energy of the reference state for the ligand–receptor complex, G°(L, wat) describes the Gibbs energy of the reference state for the ligand solved in water and G°(R) describes the Gibbs energy of the reference state for the ligand-free receptor. These first two terms on the right-hand side of Eq. 3 are given by the following equations:

$$ {\text{G}}^\circ \left( {\text{LR}} \right) = {\text{G}}^\circ \left( {{\text{L}},{\text{ gas}}} \right) + {\text{G}}^\circ \left( {{\text{R}},{\text{ LR}}} \right) + \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{L}},{\text{ LR}}} \right) $$
(4)

and

$$ {\text{G}}^\circ \left( {{\text{L}},\,{\text{wat}}} \right) = {\text{G}}^\circ \left( {{\text{L}},\,{\text{gas}}} \right) + \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{L}},\,{\text{wat}}} \right) $$
(5)

Therein, G°(L, gas) describes the Gibbs energy of reference state of the ligand in gas phase, G°(R, LR) describes the Gibbs energy of reference state of the receptor in the ligand-bound state, ΔG°sol(L, LR) describes the Gibbs energy of solvation of the ligand in the binding pocket of the ligand–receptor complex and ΔG°sol(L, wat) describes the Gibbs energy of solvation of the ligand in water.

Combining Eqs. 3, 4 and 5 results in the following equation:

$$ {\Delta_{\text{R}}}{\text{G}}^\circ = {\text{G}}^\circ \left( {{\text{R}},{\text{ LR}}} \right) + \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{L}},{\text{ LR}}} \right) - \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{L}},{\text{ wat}}} \right) - {\text{G}}^\circ \left( {\text{R}} \right) $$
(6)

Molecular dynamics in combination with thermodynamic integration can be used to predict ΔG°sol(ligand in binding pocket of the receptor) and ΔG°sol(L, wat). Using these values, the change in Gibbs energy for the transfer of the ligand from the aqueous phase into the binding pocket of the receptor (hH4R in this study), ΔΔG°sol(water → hH4R) is given by

$$ \Delta \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{water}} \to {\text{h}}{{\text{H}}_{{4}}}{\text{R}}} \right) = \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{L}},\,{\text{LR}}} \right) - \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{L}},\,{\text{wat}}} \right) $$
(7)

Combining Eqs. 2, 6 and 7 leads to the following equation:

$$ \Delta \Delta {\text{G}}{^\circ_{\text{sol}}}\left( {{\text{water}} \to {\text{h}}{{\text{H}}_{{4}}}{\text{R}}} \right) = - R\;T \times {2}.{3}0{3}\;{\text{p}}{{\text{K}}_i} + {\text{G}}^\circ \left( {\text{R}} \right) - {\text{G}}^\circ \left( {{\text{R}},\;{\text{LR}}} \right) $$
(8)

By fitting ΔΔG°sol(water → hH4R) against the pKi, a linearity, with the slope given by −R T × 2.303 (corresponding to −5.7 kJ/mol for a temperature of 298.15 K) and the y-intercept given by G°(R) − G°(R, LR), should be received. Thus, Eq. 8 is suggested to be used to predict the change in Gibbs energy of the receptor during the process of ligand binding.

Results

Analysis of phenylhistamines, histaprodifens and phenoprodifens at hH4R in the [3H]histamine competition binding assay

Tables 1 and 2 summarize the affinities of 33 studied compounds at hH4R in the [3H]histamine competition binding assay, and Fig. 2 shows representative competition binding isotherms. Phenylhistamine 2 showed low affinity at hH4R. The introduction of an additional methyl group in R- 3R and S-configuration 3S decreased affinity, but there was no difference in affinity between 3R and 3S, as observed at hH1R (Straßer et al. 2009). The introduction of a trifluoromethyl group in meta position 4 increased affinity compared to the unsubstituted phenylhistamine 2 (p = 0.0009). But the introduction of an additional methyl group in R- 5R and S-configuration 5S significantly decreased affinity compared to 4 (p = 0.087). The exchange of the trifluoromethyl group in 4 into a bromine 6 did not affect affinity. The N α-methylated phenylhistamines 7–9 (Fig. 2a) showed a significant (7 compared to 2: p = 0.0004; 8 compared to 4: p = 0.0129; 9 compared to 6: p = 0.0004) increase in affinity compared to the unmethylated compounds 2, 4 and 6. The affinity of phenylhistamine with an additional histamine moiety 10 was increased compared to the small phenylhistamine 2 (p = 0.0007). The affinities of the derivatives with a trifluoromethyl group 11 or a bromine 12 were in the same range as for compound 10. For the dimeric phenylhistamine 13, a pKi value in the range of phenylhistamine 2 was observed. The introduction of one trifluoromethyl moiety into dimeric phenylhistamine 14 slightly increased affinity (p = 0.01).

Table 1 Affinities, potencies and efficacies of histamine, phenylhistamines, histaprodifens and phenoprodifens at hH4R-RGS19 co-expressed with Gαi2 and Gβ1γ2 in Sf9 cell membranes in the competition binding and steady-state GTPase assay
Full size table
Table 2 Affinities of phenylhistamines 7, 8 and 9 at hH1R compared to hH4R
Full size table
Fig. 2
figure 2

Competition binding isotherms and concentration–response curves for selected phenylhistamines and histaprodifens at hH4R. The experiments were performed using Sf9 cell membranes expressing hH4R-RGS19, Gαi2 and Gβ1γ2 as described under “Materials and methods”. Data were analysed by nonlinear regression and were best fit to one-site (monophasic) competition curves. a Competition binding isotherms; the experiments were performed in presence of 10 nM [3H]histamine; b concentration response curves determined in the steady-state GTPase assay; c [35S]GTPγS binding curves for histamine 1 and phenylhistamines 4 and 8

Full size image

The affinity of histaprodifen 15 was in the same range as for phenylhistamine 2. The introduction of a chlorine 16 or fluorine 17 into one phenyl moiety did not lead to differences in affinity. The introduction of a methyl group in N α position of histaprodifen 18 increased affinity (p = 0.0045). Histaprodifen derivative 19 exhibited an affinity in the same range as 18. Thus, the additional imidazolyl moiety had no significant influence onto affinity. The affinity of suprahistaprodifen 20 was in the same range as for phenylhistamine with an additional histamine moiety 10. The introduction of a centre of chirality by a methyl group in compounds 21R and 21S led to a slight decrease in affinity, compared to suprahistaprodifen 20 (p = 0.0027). Compared to suprahistaprodifen 20, the affinity of phenoprodifen 22 to hH4R was decreased (p = 0.0034). The additional methyl moiety in R- 23R and S-configuration 23S did not substantially alter pKi values compared to the unsubstituted compound 22. However, the introduction of a trifluoromethyl group 24 into the phenylhistamine moiety of the unsubstituted phenoprodifen 22 showed a significant (p = 0.017) increase in affinity, whereas the exchange of this trifluoromethyl group to a bromine 25 decreased affinity (p = 0.038), comparable to phenoprodifen 22. An additional ethyl moiety in 26 compared to suprahistaprodifen 20 substantially decreased affinity (p = 0.0001), whereas an additional thienyl moiety in 27 compared to suprahistaprodifen 20 led only to a smaller decrease in affinity. Compounds 28 and 29, possessing a different substitution pattern of the terminal imidazolyl moiety, compared to suprahistaprodifen 20, in combination with a varying length of the CH2-spacer, showed a decreased affinity compared to 20. Dimeric histaprodifen 30 showed an increase in pKi value, compared to histaprodifen 15.

Analysis of phenylhistamines, histaprodifens and phenoprodifens at hH4R in the functional steady-state GTPase and GTPγS binding assay

In Tables 1, 3 and Fig. 2b, the potencies and efficacies of the analysed compounds are given. As well-known (Deml et al. 2009), histamine 1 showed a high potency at hH4R. The potency as well as the efficacy of phenylhistamine 2 was decreased, compared to histamine 1. The chiral phenylhistamines 3R and 3S behaved as neutral antagonists. The trifluoromethyl group in meta position of the phenyl moiety 4 increased potency (p = 0.0015) as well as efficacy (p = 0.01) compared to the unsubstituted phenylhistamine 2. The chiral derivatives 5R and 5S exhibited antagonism. The exchange of the trifluoromethyl 4 into a bromine 6 did not result in differences in efficacy, but potency decreased significantly (p = 0.0006). The more bulky phenylhistamine with an additional histamine moiety 10 showed, compared to phenylhistamine 2, a significantly higher potency (p = 0.0005) and efficacy (p = 0.0072). But in contrast to compounds 2 and 4, the introduction of a trifluoromethyl group into 10, leading to 11, decreased potency and efficacy (p = 0.0045). The exchange of the trifluoromethyl group in 11 into bromine 12 did not affect potency and slightly increased efficacy. Dimeric phenylhistamine 13 showed antagonism, but the introduction of a trifluoromethyl moiety in 14 slightly increased efficacy.

Table 3 Potencies and efficacies of phenylhistamines 2, 4, 6–9 at hH4R
Full size table

The potency of histaprodifen 15 was in the same range as for phenylhistamine 2. However, in contrast to phenylhistamine 2, histaprodifen 15 was an inverse agonist. The introduction of chlorine into one phenyl moiety of the histaprodifen 16 did not affect potency and efficacy, compared to histaprodifen 15 itself. In contrast, fluorine at the corresponding position 17 showed, compared to histaprodifen 15, an increase in potency (p = 0.01). Compared to histaprodifen 15, for the N α-methylated derivative 18, a significant increase in potency (p = 0.0082) and a decrease in efficacy were observed. Histaprodifen derivative 19 with an additional imidazolyl moiety showed antagonistic behaviour. A different substitution pattern of this imidazolyl moiety, as given for suprahistaprodifen 20, significantly increased efficacy (p = 0.005). The potency of suprahistaprodifen 20 was higher than for histaprodifen 15 (p = 0.001). The chiral suprahistaprodifens 21R and 21S with an additional methyl group, in contrast to suprahistaprodifen 20, showed inverse agonism, and the potency decreased (p = 0.037). There were no significant differences in potency and efficacy between 21R and 21S. Phenoprodifen 22, possessing an additional phenyl moiety compared to suprahistaprodifen 20, showed a decreased potency compared to 20 (p = 0.024) and acted, in contrast to 20 as inverse agonist. For the chiral phenoprodifens 23R and 23S, weak inverse agonism was observed. The trifluoromethyl-substituted phenoprodifen 24 revealed, compared to phenoprodifen 22, an increased potency (p = 0.0065) and the inverse agonism of 22 switched to partial agonism of 24. The exchange of the trifluoromethyl group into a bromine 25 led to antagonistic behaviour. The additional ethyl moiety in 26 decreased potency but increased efficacy compared to suprahistaprodifen 20. The additional thienyl moiety of 27 did not affect potency but switched the large partial agonism of 26 into inverse agonism. The histaprodifen derivatives 28 and 29 showed antagonistic or inverse agonistic behaviour, respectively. Dimeric histaprodifen 30, compared to suprahistaprodifen 20, showed a decreased potency and acted, in contrast to 20, as inverse agonist at hH4R.

The pharmacological data of compounds 1, 2, 4, 6, 7–9 obtained with the [35S]GTPγS binding assay are given in Table 3. A comparison of potencies and efficacies of compounds 1, 2, 4 and 6 with the corresponding data, obtained in the steady-state GTPase assay (Table 1), shows a very good accordance. In another context, the good accordance between steady-state GTPase and GTPγS binding assay was shown previously (Seifert et al. 1998; Wenzel-Seifert and Seifert 2000). The functional data revealed a significantly higher potency (p = 0.0052) and efficacy (p = 0.011) for the N α-methylated phenylhistamine 7, compared to phenylhistamine 2 itself. A similar trend was observed for compounds 4 and 8: Again, the N α-methylated phenylhistamine derivative 8 showed a significantly higher potency (p = 0.002) and efficacy (p = 0.049), compared to the unmethylated derivative 4 (Fig. 2c). The bromine-substituted N α-methylated phenylhistamine 9 exhibited a higher potency and efficacy as the bromine-substituted phenylhistamine 6. To summarize the binding and functional data, structure–activity relationships of the most important phenylhistamines, histaprodifens, supra- and phenoprodifens at hH4R is given in Fig. 3.

Fig. 3
figure 3

Summary of the most important structure activity relationships of selected phenylhistamines, histaprodifens, supra- and phenoprodifens at hH4R. The summary is based on the results, presented in Tables 1, 2 and 3

Full size image

Binding mode of phenylhistamines 2, 4, 6–9 and histaprodifen 15

Representative binding modes of 2 and 8 and 15 at hH4R are shown in Fig. 4. Phenylhistamine 2 (Fig. 4a) and 8 (Fig. 4c) fit into the binding pocket. However, only one electrostatic interaction between the positively charged amino moiety of phenylhistamine 2 or 8 and the highly conserved Asp3.32 was detected. The imidazole and phenyl moieties of 2 or 8 are embedded in an aromatic pocket, established by Tyr3.33, Trp6.48 and Tyr6.51. Stable hydrogen bonds between the imidazole moiety of the phenylhistamine and hH4R were not observed during molecular dynamic simulation. However, the molecular modelling revealed two small empty pockets (Fig. 4a, arrows 1 and 2) in case of the smallest phenylhistamine 2. In case of phenylhistamine derivative 8, the additional methyl moiety fits into pocket 1 (Fig. 4a, c, arrow 1), whereas the trifluoromethyl group fits into pocket 2 (Fig. 4a, c, arrow 2). This results in an increase of interaction between hH4R and phenylhistamine derivative 8, compared to phenylhistamine 2. Due to a free small pocket between the phenylhistamine and Trp6.48, an increased flexibility of Trp6.48 was observed during the molecular dynamic simulations, compared to Trp6.48 with histaprodifen 15 bound in the binding pocket. Representative for small phenylhistamines the evolution of the dihedral angles α (CD1 → CG → CB → CA) and β (CG → CB → CA → N) of Trp6.48, with phenylhistamine 2 in the binding pocket, during the productive phase of molecular dynamic simulation is shown. The flip, observed at about 250 ps (Fig. 4b, red box), corresponds to a conformational change of Trp6.48 from a nearly vertical conformation into a more horizontal conformation.

Fig. 4
figure 4

Binding mode of phenylhistamines 2 and 8 and histaprodifen 15 at hH4R. a Phenylhistamine 2 docked into the binding pocket of hH4R; b dihedral angle α (CD1 → CG → CB → CA) and β (CG → CB → CA → C) of Trp6.48 during productive phase of molecular dynamic simulation of phenylhistamine 2 and Trp6.48 in its vertical and horizontal conformation as observed during molecular dynamic simulation; c phenylhistamine 8 docked into the binding pocket of hH4R; d histaprodifen 15 docked into the binding pocket of hH4R and dihedral angle α (CD1 → CG → CB → CA) and β (CG → CB → CA → C) of Trp6.48 during productive phase of molecular dynamic simulation of histaprodifen 15. The amino acids of the transmembrane domains are numbered according to the nomenclature of Ballesteros et al. (2001)

Full size image

The binding mode of histaprodifen 15, based on molecular dynamic simulations, is shown in Fig. 4d. The basic amine moiety interacts electrostatically with the highly conserved Asp3.32. The histaprodifen is embedded in a pocket, established by Trp6.48, Tyr6.51 and Phe168 (second extracellular loop). The neighboured Phe169 is responsible for pharmacological differences between human and mouse H4R (Lim et al. 2008). The highly conserved Trp6.48 is in close contact to one of the phenyl moieties of 15. The phenyl moieties are embedded in the binding pocket in a rigid manner. Consequently, during the molecular dynamic simulation, only a very small flexibility of Trp6.48 was observed.

Prediction of Gibbs free energy of binding for phenylhistamines 2, 4, 6–9

In Table 4, the predicted Gibbs free energy of solvation of a ligand in water (ΔG°sol(water)), Gibbs free energy of solvation of a ligand in the binding pocket of hH4R (ΔG°sol(hH4R)) and the change in Gibbs free energy of solvation for transferring the ligand from the water into the binding pocket of hH4R (ΔG°sol(water → hH4R)) are given. Additionally, in Fig. 5a, the evolution of ΔG°sol(hH4R) for representative ligands as function of the coupling parameter λ is given, whereas in Fig. 5b, the correlation between predicted ΔΔG°sol(water → hH4R) and experimentally determined affinities is shown. The correlation between ΔΔG°sol(water → hH4R) and the pKi values (Fig. 5b) is given by the equation

$$ \Delta \Delta {\text{G}}{^\circ_{\text{solv}}}\left( {{\text{water}} \to {\text{h}}{{\text{H}}_{{4}}}{\text{R}}} \right) = \left( { - {126}.{4}\pm {32}} \right){\text{kJ/mol}} - \left( {{3}0.{3}\pm {5}} \right)\;{\text{kJ/mol}}\;{\text{p}}{{\text{K}}_i} $$

with r 2 = 0.89. Thus, the correlation between predicted ΔΔG°sol(water → hH4R) and experimental affinities is very good. A comparison of the predicted slope of −30.3 kJ/mol in comparison to the theoretical value of −5.7 kJ/mol shows that the prediction is in the same range. Based on Eq. 8, the y-intercept, predicted to be 126.4 kJ/mol, is suggested to correspond to the change in Gibbs energy of the receptor during ligand binding.

Table 4 Calculated Gibbs energy of solvation for phenylhistamines 2, 4, 6–9 in water and in the binding pocket of hH4R
Full size table
Fig. 5
figure 5

Calculation of ΔΔGsol(water → hH4R). a Evolution of Gibbs free energy of solvation ΔG°sol(hH4R) for the ligand in the binding pocket of hH4R as a function of the coupling parameter λ, switching on, respectively off, the interaction between ligand and surrounding. b Predicted ΔΔG°sol(water → hH4R) in correlation to experimentally determined affinities

Full size image

Discussion

Comparison of the pharmacology of phenylhistamines, histaprodifens and phenoprodifens between hH1R and hH4R

Figure 6 shows the correlation of the affinities of the analysed compounds at hH1R and hH4R. This comparison illustrates that the affinity of most of these compounds is equal to the affinities at hH1R or lower than the affinities at hH1R. Only the three N α-methylated phenylhistamines 7–9 exhibit a significantly higher affinity at hH4R than at hH1R. It was shown previously (Schneider et al. 2010) for benzimidazole derivatives at hH4R that the introduction of an additional methyl moiety at the basic amine leads to an increase in affinity. Thus, the phenylhistamine derivatives 7–9 are the first phenylhistamines exhibiting a significantly higher affinity at hH4R, than at hH1R (Table 2). Since the introduction of a bromine or trifluoromethyl in meta position of the phenyl moiety increased affinity for the small phenylhistamines 4 and 6 as well as for the N α-methylated phenylhistamines 8 and 9, we suggest that this additional small moiety fits into a small pocket of the receptor and leads to an increase in affinity due to the better ligand–receptor interaction. The same trend is observed for the same phenylhistamines at hH1R (Straßer et al. 2009; Straßer and Wittmann 2010). The corresponding molecular modelling and molecular dynamic simulation studies revealed that the trifluoromethyl or bromine (Fig. 4a, c, arrow 2) fit into a small pocket. Thus, the pharmacological data can be explained well by molecular modelling studies. The introduction of an additional methyl group in 7–9, compared to 2, 4 and 6, leads to a significant increase in affinity at hH4R. In contrast, at hH1R, this additional methyl group has no influence onto affinity. The binding data suggest that the additional methyl group stabilizes the ligand–receptor complex due to a hydrophobic interaction. The additional methyl group fits into a small pocket of hH4R, leading to an increase in hydrophobic ligand receptor interaction on the one hand. On the other hand, the phenylhistamine is stabilized better in the binding pocket.

Fig. 6
figure 6

Comparison of affinities and efficacies of phenylhistamines and histaprodifens between hH1R and hH4R. a Comparison of the affinities of phenylhistamines and histaprodifens between hH4R and hH1R (Straßer et al. 2008, 2009; Straßer and Wittmann 2010). b Comparison of the efficacies of phenylhistamines and histaprodifens between hH4R and hH1R (Straßer et al. 2008, 2009). The most important compounds are marked in the figure

Full size image

Besides these structural considerations, the predicted ΔΔG°sol(water → hH4R) for six phenylhistamines are in very good correlation to the experimentally determined affinities (Fig. 5b). These predictions include only the change in Gibbs energy for transfer of the ligand from the aqueous phase into the binding pocket of hH4R, but no change in Gibbs energy due to receptor conformation. Thus, based on these results, the analysed phenylhistamines 2, 4, 6–9 may stabilize the hH4R in similar conformations. Additionally, we predicted a change in Gibbs energy of the receptor during binding of phenylhistamines of about 126.4 kJ/mol. To the best of our knowledge, there are neither experimental nor theoretical data available in literature, which can be used for comparison. However, in future studies, further ligands, belonging to highly different ligand classes, should be analysed by a combination of molecular modelling and experimental studies, as performed in this study in order to predict the change in Gibbs energy of the receptor during ligand binding. Different ligand classes, which stabilize the receptor in different conformations, could lead to different y-intercepts. However, in further studies, it has to be analysed if those energetical values can be interpreted absolutely or if they have to be used relative to each other.

Histaprodifens as partial agonism–inverse agonism switch between hH1R and hH4R

The results of the steady-state GTPase assay show that the phenylhistamines act as partial agonists at hH1R (Straßer et al. 2009) as well as at hH4R, except for the chiral phenylhistamines 3R, 3S, 5R, 5S and dimeric phenylhistamine 13, which are neutral antagonists at hH4R. In contrast, the histaprodifens exhibit partial agonism at hH1R but inverse agonism at hH4R. Thus, the exchange of the phenyl moiety into the phenylpropyl moiety induces a partial agonism–inverse agonism switch at hH4R, but not at hH1R. In general, suprahistaprodifens and phenoprodifens show partial agonism at hH1R, but partial agonism or inverse agonism at hH4R. Interestingly, these are the compounds which are able to bind in two different orientations into the binding pocket. Based on the results of phenylhistamines and histaprodifens, we suggest that the inverse or partial agonism of suprahistaprodifens or phenoprodifens is dependent of ligand orientation in the binding pocket. If the histaprodifen partial structure is located in the histaprodifen binding pocket, the compounds act as inverse agonists and as partial agonists otherwise. This result is very important for developing therapeutically relevant hH4R antagonists or inverse agonists. This additional phenyl moiety decreases activity of hH4R on the one hand, but on the other hand stimulates hH1R, resulting in allergic reactions for example. Since histaprodifens bind to hH1R and to hH4R, they are not useful as therapeutics. Nevertheless, our results provide important information about structural properties to be able to construct new H4R antagonists or dual H1R/H4R antagonists.

The molecular modelling studies revealed that Trp6.48 is in close aromatic contact to one of the phenyl moieties of histaprodifen (Fig. 4d). In contrast, for the phenylhistamines, the Trp6.48 is also in proximity to the phenylhistamine (Fig. 4a). But there is space left between the Trp6.48 and the phenylhistamine. Thus, the Trp6.48 can show a higher flexibility and switch between different conformations (Fig. 4b). The highly conserved Trp6.48 is discussed to be involved in an aromatic rotamer toggle switch during receptor activation (Crocker et al. 2006). For phenylhistamines, a switch from a more vertical position, suggested to be characteristic for the inactive conformation, to a more horizontal position, suggested to be characteristic for the active conformation, could be observed in the molecular dynamic simulations (Fig. 4b, red box). However, this switch was only observed for short time scales and seemed not to be stable during the whole simulations. Nonetheless, these simulation data are in good accordance to the experimentally determined small efficacies. Thus, phenylhistamines may stabilize both the inactive and active conformation of hH4R. Due to the close aromatic contact, the flexibility of Trp6.48 is reduced in case of bound histaprodifen, and no conformational change, as for phenylhistamine 2, was observed. This is in very good accordance to the experimental result that histaprodifen 15 acts as inverse agonist.

Conclusions

In this study, we identified the first phenylhistamine derivatives with a higher selectivity with regard to hH4R than to hH1R. This study also revealed the structural basis for the unique agonist–inverse agonist switch at histaprodifens between hH1R and hH4R. Thus, our study, combining experimental and modelling studies, provided important structural information for the future development of dual H1/H4 receptor agonists and insight into functional mechanism of receptor activation. Phenylhistamines can no longer be classified as “selective H1R agonists”.