Correlation of apparent affinity values from H₃-receptor binding assays with apparent affinity (pK_app) and intrinsic activity (α) from functional bioassays

Abstract

Background and purpose:

Agonist apparent affinities (pK_I′) in histamine H₃-receptor binding assays were higher than expected from apparent affinity values (pK_app) estimated in bioassay. Here, we investigate whether the degree of pK_I′ overestimation is related to agonist intrinsic efficacy, by studying the effect of buffer composition on the pK_I′ of ligands with varying intrinsic activity.

Experimental approach:

In the guinea-pig ileum bioassay, intrinsic activity (α) was determined from the maximal inhibition of the contraction produced by increasing agonist concentration. pK_app values were estimated using the method of Furchgott. The pK_L of [³H]clobenpropit in guinea-pig cerebral cortex was estimated by saturation analysis in 20 mM HEPES-NaOH buffer (buffer B_(0,0,0)), or buffer B_(0,0,0) containing 70 mM CaCl₂, 100 mM NaCl and 100 mM KCl (buffer B_{(0.07,0.1,0.1)}). PK_I values were determined in competition studies in both buffers.

Key results:

[³H]clobenpropit saturation isotherms had n _H values of unity in both buffers. In buffer B_{(0.07,0.1,0.1)}, agonist pK_I′ values were closer to pK_app values than in buffer B_(0,0,0) but were associated with n _H values <1. A two-site analysis of agonist data in buffer B_{(0.07, 0.1, 0.1)} provided a better fit than a one-site fit and low affinity values (pK_IL) were comparable to pK_app. Differences between the pK_I′ in buffer B_(0,0,0) and pK_IL values in buffer B_{(0.07,0.1,0.1)} (ΔpK) were correlated with α.

Conclusions and implications:

H₃-receptor binding assays conducted in buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)} can provide a measure of ligand affinity (pK_app) and intrinsic efficacy. The assay predicts that some ligands previously classified as H₃-receptor antagonists may possess residual intrinsic efficacy.

Introduction

Tissue- or assay-dependent expression of intrinsic efficacy is a well-recognized phenomenon and as such, ligands that act as antagonists or partial agonists in one tissue of a species may behave as full agonists in another tissue from the same species. For instance, Black (1988) found that dichloroisoprenaline was an agonist at β-adrenoceptors in the spontaneously beating guinea-pig heart preparation, but an antagonist in the guinea-pig cardiac papillary muscle. Similarly, Kenakin and Beek (1980) showed that prenalterol was almost a full agonist at β₂-adrenoceptors in guinea-pig trachea, a partial agonist in guinea-pig left atria and an antagonist in guinea-pig extensor digitorum longus.

Some years ago, before the histamine H₃-receptor was cloned (Lovenberg et al., 1999) and as part of a programme aimed at the development of high-affinity and selective H₃-receptor antagonists, we developed a radioligand binding assay using the H₃-receptor agonist, [³H]R-α-methylhistamine ([³H]R-α-MH; Harper et al., 1997a, 1997b; 1999a). In the course of evaluating ligands in this assay, we found that the affinity estimates of ligands, previously characterized as competitive antagonists, were on the whole, comparable to those reported in histamine H₃-receptor isolated tissue bioassays. However, we also noticed that there were a number of ligands that expressed a considerably higher affinity in the radioligand binding assays than they expressed in the isolated tissue bioassays. Moreover, for some of the ligands, the degree of affinity overestimation appeared to correlate with the ligand's intrinsic activity (α) measured in the functional assay, as though the overestimation provided a measure of agonist intrinsic efficacy. This observation raised the possibility that the H₃-receptor radioligand binding assay might be a sensitive method of detecting residual intrinsic efficacy of H₃-receptor ligands and, moreover, that it could allow identification of H₃-receptor partial agonism that had remained undetected in the isolated tissue bioassay.

To facilitate estimation of both the apparent affinity (pK_app) and intrinsic efficacy of H₃-receptor ligands, in the radioligand-binding assay, we considered that it would be necessary to manipulate the conditions of the assay to provide ‘high' and ‘low' affinity estimates for each ligand, the low affinity estimate being equivalent to the ligand's pK_app value estimated in isolated tissue bioassays and the difference between the ‘low' and ‘high' affinity estimates being a measure of the ligand's intrinsic efficacy.

We chose to try to manipulate the radioligand-binding assay, to provide the ‘low' affinity (pK_app) estimates for each ligand, by adding salts to the assay buffer rather than by adding guanine nucleotide analogues or using the guinea-pig ileum Krebs–Henseleit (K–H) buffer. This was because it was noted that the affinity of agonists in H₃-receptor radioligand-binding assays, in the presence of high concentrations of guanine nucleotides or in Krebs buffer (e.g. Arrang et al., 1990; [³H]R-α-MH pK_L, 100 μM Gpp(NH)=9.05, Krebs=9.41) were still considerably higher than pEC₅₀ values obtained in functional assays of the same tissue (pEC₅₀=8.40) and therefore could not be equivalent to the agonist pK_app. In addition, in the process of developing the guinea-pig cortex H₃-receptor assay, we had found that increasing the buffer concentrations of a variety of salts (e.g. NaCl, CaCl₂, MgSO₄, NaH₂PO₄ and KCl) produced a concentration–dependent decrease in the specific binding of the agonist radioligand, [³H]R-α-MH (data not shown). Saturation studies performed in 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES–NaOH) buffer (buffer B) containing 70 mM CaCl₂, 100 mM NaCl and 100 mM KCl, alone and in combination (buffer B_{(0.07,0.1,0.1)}), indicated that the salts reduced the specific binding of [³H]R-α-MH by decreasing the pK_L without any change in B_max and the greatest decrease in pK_L was obtained in the presence of 100 mM NaCl, 100 mM KCl and 70 mM CaCl₂ (data not shown). Although competition studies performed in buffer B and in buffer B_{(0.07,0.1,0.1)} indicated that antagonist affinity was unchanged by buffer composition (see Table 1) and that ‘low' affinity estimates could be obtained for H₃-receptor agonists (see Table 1), which were comparable to those obtained in the guinea-pig ileum bioassay, we were concerned that these studies required high concentrations of radioligand (20 nM) and that it was necessary to increase the tissue concentration to obtain a suitable ‘window' of specific binding. Consequently, we chose to try and use the H₃-receptor antagonist, [³H]clobenpropit, to perform the studies on the assumption that an antagonist's affinity would not change appreciably, when the assay conditions were manipulated (see Table 1, Harper et al., 1997c, 1999b).

Table 1.

Parameter estimates for histamine H₃-receptor ligands, obtained in buffer B_(0,0,0) and buffer B _{(0.07,0.1,0.1)}, in guinea-pig cerebral cortex using [³H]R-α-MH as radioligand

	Ligand	Buffer B(0,0,0)		Buffer B(0.07,0.1,0.1)		n
		pKI′	nH	pKI′	nH
Antagonists	thioperamide	8.59±0.27	0.84±0.09	8.76±0.14	0.84±0.06	3
	clobenpropit	10.50±0.08	1.22±0.06	9.82±0.18	1.04±0.16	3
Agonist	iodoproxyfan	9.58±0.05	0.96±0.10	8.47±0.14	1.04±0.09	4
	proxyfan	8.89±0.34	0.95±0.02	7.40±0.05	1.11±0.06	3

Data are the mean±s.e.m. from the number of assays shown (n). Competition experiments were conducted as described by Harper et al. (1999a). A final assay tissue concentration of 6 mg and 0.1 nM concentration of [³H]R-α-MH was used in buffer B_(0,0,0). In buffer B_{(0.07,0.1,0.1),} the concentration of [³H]R-α-MH was 20 nM and a tissue concentration of 12 mg was used to achieve sufficient specific binding.

In this study, we first report the effect that including NaCl, KCl and CaCl₂ (buffer B_{(0.07,0.1,0.1)}) in the assay buffer has on the affinity of [³H]clobenpropit at H₃-receptors in guinea-pig cortex. We subsequently report the effect that this buffer has on the ‘behaviour' of a series of H₃-receptor antagonists and agonists (see Figure 1), which had been described at the time, with varying α as defined by assays performed in the guinea-pig ileum bioassay. Some of these data were previously presented to the British Pharmacological Society (Harper et al, 1997d; Watt et al., 1997).

Figure 1.

Chemical structures of histamine H₃-receptor ligands.

Methods

Preparation of guinea-pig cerebral cortex membranes

Adult male Dunkin–Hartley guinea pigs (200–300 g) were killed by cervical dislocation and the whole brain was removed and immediately placed in ice-cold 20 mM HEPES–NaOH buffer (pH7. 4, 4°C). The cortex was dissected, weighed and homogenized in ice-cold 20 mM HEPES–NaOH buffer (pH7.4, 4°C) (1 g 15 ml⁻¹) using a polytron homogenizer (Kinematica AG, GmbH, Lucerne, Switzerland; PT-DA 3020/2TS; ∼3 s × 3). The homogenate was centrifuged at 100 g for 5 min and the supernatants pooled and stored at 4°C. The pellets were rehomogenized in fresh ice-cold buffer (80 ml) and recentrifuged at 100 g for 5 min at 4°C. The supernatants were centrifuged at 39800 g for 12 min at 4°C and the final pellet was resuspended in 20 mM HEPES–NaOH buffer containing 3 mM metyrapone (21°C) (Harper et al., 1997c), to the required tissue concentration, using a Teflon-in-glass homogenizer (setting 5, 3 ×).

[³H]clobenpropit – saturation studies

Guinea-pig cortical membranes (1.6 mg) were incubated for 165 min at 21±3°C, in a final volume of 0.5 ml with HEPES–NaOH buffer (buffer B_(0,0,0)) and 0.004 to 3 nM [³H]clobenpropit. Total and nonspecific binding of [³H]clobenpropit were defined using HEPES–NaOH buffer and 1μM thioperamide (pK_I at histamine H₃-receptors in guinea-pig cortex ∼9.0, Harper et al., 1999a), respectively. The assay was terminated by rapid filtration through Whatman GF/B filters, presoaked in 0.3% polyethyleneimine, which were washed (3 × 3 ml) with ice-cold 50 mM Tris–HCl (pH7.4, 4°C) using a Brandell Cell Harvester (Brandell, Gaithersburg, MD, USA). Filters were transferred into scintillation vials, 4 ml Meridian Gold–Star liquid scintillation cocktail added and after 3 h the bound radioactivity was determined by counting (3 min) in a Beckman liquid scintillation counter.

To determine the effect of modifying the assay buffer on the binding of [³H]clobenpropit, we performed saturation analysis in buffer B_(0,0,0), and in this buffer containing final assay concentrations of 70 mM CaCl₂, 100 mM NaCl and 100 mM KCl (buffer B_{(0.07,0.1,0.1)}; with final ionic concentrations (M) Ca²⁺, 0.07; Na⁺, 0.1; K⁺, 0.1; Cl⁻, 0.27=buffer B_{(0.07, 0.10, 0.10, 0.27)}). In a further series of experiments, [³H]clobenpropit saturation analysis was performed in the presence of increasing (0, 30, 70, 100, 200, 300 mM) final assay concentrations of CaCl₂ (buffer B_(0,0,0), B_(0.03,0,0), B_(0.07,0,0), B_(0.1,0,0), B_(0.2,0,0) and B_(0.3,0,0), respectively).

[³H]clobenpropit competition studies

Competition studies were conducted using a membrane concentration, which was previously found to result in zone A conditions for [³H]clobenpropit binding at a 0.2 nM concentration (see Harper et al., 1999b). In addition, the competition assay incubation time was previously shown to be sufficient for equilibrium of both radioligand and competitor (see Harper et al., 1999b). Guinea-pig cerebral cortex membranes were resuspended in 20 mM HEPES–NaOH buffer containing 0.3 mM metyrapone. Membranes (1.6 mg) were incubated for 165 min at 21±3°C in a final volume of 0.5 ml with 20 mM HEPES–NaOH buffer containing [³H]clobenpropit (0.2 nM), histamine H₃-receptor ligands and either 20 mM HEPES–NaOH, buffer B_{(0.07,0.1,0.1)}, buffer B_(0.03,0,0), buffer B_(0.07,0,0), buffer B_(0.1,0,0), buffer B_(0.2,0,0) or buffer B_(0.3,0,0). Total and nonspecific binding of [³H]clobenpropit were defined using HEPES–NaOH buffer and 1μM thioperamide, respectively.

Guinea-pig ileum assay

Measurement of intrinsic activity

Adult male Dunkin–Hartley guinea pigs (300–500 g) were killed by cervical dislocation. The ileum was removed at a point 20 cm from the caecum and flushed with and placed in modified K–H buffer of the following mM composition: 118 NaCl, 5.9 KCl, 1.2 CaCl₂, 1.2 MgSO₄, 1, Na₂HPO₄, 25 NaHCO₃ and 10 D-glucose. Ileum segments (2.5–3 cm) were suspended in 20 ml organ baths containing K–H buffer maintained at 37±1°C and gassed with 95% O₂/5% CO₂. The initial resting tension was adjusted to 1 g and the tissues field stimulated (0.1 Hz and 0.5 ms) at supramaximal voltage. A single cumulative concentration–effect curve (E/(A)) curve was obtained in each tissue. Decreases in tissue twitch tension (% inhibition) were recorded using isometric transducers (Grass FTO3). Mepyramine (3 μM) and famotidine (10 μM) were added to the K–H buffer to block postsynaptic H₁ and presynaptic H₂ receptors, respectively.

Measurement of antagonist affinity

Thioperamide, JB16132, JB96134, iodophenpropit, JB97034, GT-2227, JB95130 and GR175737 were preincubated with tissues for 1 h before the change in tension in response to increasing concentrations of the H₃-receptor agonist, R-α-MH, was determined. Clobenpropit was preincubated with tissues for 3 h.

Estimation of apparent agonist affinity (pK_app)

Measurement of the pK_app of agonists at H₃-receptors in guinea-pig ileum was achieved using the method of Furchgott with the irreversible H₃-receptor antagonist, N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline (EEDQ, see Taylor and Kilpatrick, 1992). Agonist E/(A) curves were constructed in untreated tissues and in tissues which had been incubated (15 min) with EEDQ (0.3 μM) and washed (six times at 10 min intervals) before use. In each experiment the effect of EEDQ treatment on agonist E/(A) curves was ascertained for R-α-MH and at least three other agonists.

Data analysis

All data are presented as the mean±s.e.m. unless otherwise stated.

Functional data–agonist concentration effect curves

To obtain estimates of pEC₅₀, and maximal asymptote (α), the Hill equation was fitted to agonist dose–response data, expressed as percentage inhibition of the electrically induced contraction of the ileum. To permit comparison of the agonist α values in different experiments, the α-values for each agonist, in each experiment were expressed as a percentage of the mean α-value obtained for R-α-MH in that experiment. R-α-MH was assigned a α-value of 1.0

Functional data – Schild analysis

When the minimum criteria for competitive antagonism were satisfied, that is, the antagonist produced a parallel, rightward shift in the R-α-MH concentration effect curve with no change in maximum asymptote, data were analysed according to the methods described by Black et al. (1985). pA₂ values were estimated by fitting the individual pEC₅₀ values, obtained in the presence (pEC₅₀′) and absence (pEC₅₀) of antagonist to the following derivative of the Schild equation

If the Schild slope parameter (b) was not significantly different from unity, it was constrained to unity and the data refitted to provide a pK_B estimate.

Functional data – estimation of apparent agonist affinity (pK_app)

Mean data sets obtained for at least three agonists and R-α-MH, in control tissue and in EEDQ-pretreated tissues, were fitted to the operational model of agonism (see Black and Leff, 1983) with shared values of maximal effect (E_max) and transducer slope parameter (n). pK_app values were obtained by fitting all the data to the model using a derivative-free nonlinear, regression programme (BMDP Statistical Software, Module AR: Dixon, 1992).

Radioligand binding – saturation analysis

The Hill equation was fitted to saturation data (Eq. (2)) using Graph-Pad prism where the Hill slope (n_H) was permitted to vary and where this parameter was constrained to unity.

In this equation, L is the radioligand concentration, B_max the receptor density and K_L the equilibrium dissociation constant of the radioligand.

Radioligand binding – competition curve data

To obtain pIC₅₀ and n_H parameter estimates, competition data were fitted to the Hill equation and to the Hill equation with n_H constrained to unity, using Graph-Pad Prism software. When n_H parameter estimates were less than unity and the unconstrained Hill equation provided a significantly better fit of the data than the constrained equation, as determined using an F-test, a two-site model was also fitted to the data.

Notwithstanding the finding of n_H values that were significantly less than unity, dissociation constants were subsequently determined from pIC₅₀ values using the Cheng and Prusoff (1973) to correct for the different receptor occupancy of [³H]clobenpropit in the different buffers. The parameter pK_I′ has been assigned to dissociation constants which were derived from pIC₅₀ values, where competition curve n_H parameter estimates were significantly less than unity. The pK_L values that were used to correct pIC₅₀ values obtained in buffer B_(0,0,0) and Buffer B_{(0.07,0.1,0.1)} were 10.36 and 9.82, respectively. pK_L values used to correct for the occupancy of [³H]clobenpropit in the presence of increasing CaCl₂ concentrations are presented in Table 3.

Table 3.

Effect of increasing CaCl₂ concentration on the affinity (pK_L), n_H and Bmax of [³H]clobenpropit at histamine H₃-receptors in guinea-pig cerebral cortex

Buffer	pKL	Bmax (fmol mg−1)	nH
B(0,0,0)	10.34±0.10	3.19±0.25	1.23±0.11
B(0.03, 0,0)	10.16±0.22	3.54±0.05	1.11±0.05
B(0.07,0,0)	10.05±0.06	4.17±0.38	1.08±0.09
B(0.1,0,0)	9.84±0.15	4.14±0.23	1.04±0.14
B(0.2,0,0)	9.70±0.05	3.53±0.32	1.06±0.12
B(0.3,0,0)	9.60±0.02	4.10±0.23	1.06±0.02

Data are the mean±s.e.m. of three separate experiments. The concentrations of CaCl₂ used in the buffers were (M): 0, 0.03, 0.07, 0.1, 0.2, 0.3. The s.e.m. of the n_H parameter from individual data sets was between 0.36 and 0.09.

Statistical analysis

The effect of antagonist treatment on pEC₅₀ and α-values was assessed by analysis of variance (ANOVA) and the Bonferroni-modified t-test for multiple comparisons. Differences in [³H]clobenpropit pK_L values were determined by ANOVA. P-values of less than 0.05 were considered significant.

Statistical comparison of ligand pK_I′ values obtained in buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)} were performed using Minitab version 13 (Mintab Inc.) by fitting a general linear model with corresponding ANOVA, according to Rosendaal and Stone (2003). This approach was used to avoid the situation that often occurs when dealing with comparisons of large complex data sets by the use of multiple t-tests, that is, by chance significant differences will be found.

Materials

[³H]clobenpropit (VUF9153) was prepared to a specific activity of 1.67TBq mmol^−1. by Amersham International plc. (Little Chalfont, Buckinghamshire, UK).

Iodophenpropit, proxyfan, 4-iodoproxyfan and the chloro and bromo derivatives, JB96132, JB96134, JB97034, GR175737, JB95130 and GT-2227 were synthesized by James Black Foundation chemists. Histamine, 2-methyl-1,2-di-3-pyridyl-1-propanone (metyrapone), HEPES, EEDQ and Trizma base were obtained from Sigma Chemical Co., Poole, Dorset, UK. R-α-MH, S-α-methylhistamine (S-α-MH), thioperamide and imetit were obtained from Research Biochemicals Inc. (Poole, Dorset, UK). N-α-methylhistamine (N-α-MH) was obtained from Tocris Cookson Ltd. (Bristol, UK). All other materials were obtained from Fisher Scientific (Loughborough, Leicestershire, UK).

Results

Guinea-pig ileum bioassay

Histamine, S-α-MH, R-α-MH, N-α-MH, imetit, proxyfan, chloroproxyfan, iodoproxyfan and bromoproxyfan produced dose–dependent inhibition of the electrically induced twitch of the guinea-pig ileum (e.g. Figure 2). Agonist potency values (pEC₅₀) and maximal inhibitory effects (α) are shown in Table 2.

Figure 2.

Effect of histamine H₃-receptor ligands on the electrically induced contraction of the guinea-pig ileum. Each point represents the mean±s.e.m. of determinations in at least three separate preparations (see Table 2).

Table 2.

pEC₅₀, α, pK_app and pK_B values for histamine H₃-receptor ligands in the guinea-pig ileum bioassay

H3-receptor agonist	pEC50 (n)	α (n)	pKapp	H3-receptor antagonist	pA2 (n)
imetit	7.95±0.04 (12)	0.90±0.09 (12)	7.61±0.16	thioperamide	8.53±0.08 (20)
proxyfan	7.29±0.19 (4)	0.35±0.07 (4)	7.66±0.49	iodophenpropit	8.82±0.34 (4)c
			7.34±0.10 (4)a
chloroproyfan	7.85±0.14 (4)	0.45±0.12 (4)	8.04±0.25	JB96132	8.58±0.10 (54)
bromoproxyfan	8.27±0.10 (9)	0.69±0.07 (9)	7.76±0.33	JB96134	6.96±0.10 (6)c
iodoproxyfan	8.33±0.11 (10)	0.90±0.05 (10)	8.11±0.21	JB97034	7.22±0.16 (5)c
R-α-MH	7.64±0.06 (25)	1.00±0.05 (25)	7.19±0.20 (3)b	JB95130	5.44±0.26 (7)c
N-α-MH	7.41±0.07 (5)	0.93±0.10 (5)	7.16e	GR175737	8.29±0.28 (4)c
S-α-MH	6.52±0.03 (10)	0.99±0.08 (10)	5.59±0.14	GT-2227	6.82±0.11 (70)
histamine	7.40±0.04 (6)	1.16±0.11 (6)	6.37±0.16	clobenpropit	9.93±0.12 (3)d

Abbreviations: N-α-MH, N-α-methylhistamine; R-α-MH, R-α-methylhistamine; S-α-MH, S-α-methylhistamine.

Data shown are the mean±s.e.m.; n=number of tissues used unless stated otherwise. The error on the pK_app values is the fitting error.

^aThe pA₂ of proxyfan determined in four separate experiments±s.e.m.

^bMean pK_app from three separate experiments±s.e.m.

^cpA₂ determined from a single antagonist concentration.

^dpA₂ determined in three separate experiments.

^eSee Taylor and Kilpatrick (1992).

Thioperamide, clobenpropit, GR175737, iodophenpropit, JB95130, GT-2227, JB96132, JB96134 and JB97034 had no effect on the electrically induced twitch of the guinea-pig ileum but produced parallel rightward shifts in R-α-MH E/(A) curves without change in maximal response or slope (n_H) (Table 2). Antagonist pK_B or pA₂ values are shown in Table 2.

EEDQ treatment of tissues resulted in an increase in agonist pEC₅₀ and decrease in the maximal inhibitory response to each agonist (Figure 3). The effect of EEDQ on agonist E/(A) curves was prevented following receptor protection with H₃-receptor antagonists (thioperamide 1 μM, pA₂=8.53; clobenpropit 30 nM, pA₂=10.1; data not shown). Agonist pK_app values, shown in Table 2, were obtained by global fitting of the data using the model of agonism (Black and Leff, 1983) and a derivative-free nonlinear, regression programme (BMDP Statistical Software, Module AR: Dixon, 1992).

Figure 3.

Effect of EEDQ (0.3 μM) on (a) R-α-MH, (b) imetit, (c) S-α-MH and (d) histamine-induced inhibition of the electrically induced contraction of the guinea-pig ileum. Each point is the mean±s.e.m. of determinations in at least six separate preparations.

Effect of buffer composition on [³H]clobenpropit saturation analyses

The effect of 70 mM CaCl₂, 100 mM KCl and 100 mM NaCl (buffer B_{(0.07,0.1,0.1)}) or the effect that increasing concentrations of CaCl₂ had on the binding of [³H]clobenpropit to guinea-pig cerebral cortex membranes, was investigated. This was carried out so that it was possible to correct the H₃-receptor ligand pIC₅₀ values, obtained in competition studies, to account for changes in the radioligand receptor occupancy that would result from a change in the radioligand's affinity (pK_L).

The binding of [³H]clobenpropit to guinea-pig cerebral cortex membranes was saturable in buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)} (Figure 4). Saturation isotherms were monophasic and n_H parameter estimates were not significantly different from unity in both buffers (buffer B_(0,0,0) n_H=1.14±0.09, buffer B_{(0.07, 0.1, 0.1)} n_H=0.95±0.06; n=4, t-test P>0.05). In addition, there was no significant difference between the goodness of fit to the Hill equation and that to the Hill equation with n_H constrained to unity (F-test, P>0.05). The affinity (pK_L) of [³H]clobenpropit was significantly lower in buffer B_{(0.07, 0.1, 0.1)} (pK_L=9.82±0.07; n=4) than in buffer B_(0,0,0) (pK_L=10.36±0.07; n=4; ANOVA P<0.002) but there was no significant difference in the H₃-receptor density estimates in buffer B_(0,0,0) (B_max=4.46±0.53 fmol mg⁻¹ original wet weight) and buffer B_{(0.07,0.1,0.1)} (B_max=5.59±0.66 fmol mg⁻¹ original wet weight; ANOVA P>0.05).

Figure 4.

Representative saturation isotherms of [³H]clobenpropit to sites in guinea-pig cerebral cortex membranes in buffer B_(0,0,0) (a and c) and buffer B_{(0.07,0.1,0.1)} (b and d). The lines shown superimposed on the data points are the saturation isotherm obtained by fitting the Hill equation with n_H constrained to unity to the data. Guinea-pig cortical membranes (1.6 mg) were incubated for 165 min at 21±3°C in a final volume of 0.5 ml with HEPES–NaOH buffer and [³H]clobenpropit. Total and nonspecific binding of [³H]clobenpropit were defined using buffer B_(0,0,0) or buffer B_{(0.07,0.1,0.1)} and 1 μM thioperamide, respectively. All determinations were made in triplicate.

Binding of [³H]clobenpropit to guinea-pig cerebral cortex membranes was also saturable in buffer containing concentrations of CaCl₂ up to and including 300 mM (buffers B_(0.03,0,0), B_(0.07,0,0), B_(0.1,0,0), B_(0.2,0,0) and B_(0.3,0,0); Figure 5). Saturation isotherms were monophasic and n_H parameter estimates were not significantly different from unity at all CaCl₂ concentrations (n=3; P>0.05, t-test). There was no significant difference between the goodness of fit to the Hill equation and that to the Hill equation with n_H constrained to unity in all buffers (F-test, P>0.05). CaCl₂ concentration had no significant effect on H₃-receptor density (Table 3, Figures 5 and 6; ANOVA P>0.05). The pK_L of [³H]clobenpropit was significantly decreased in the presence of increasing CaCl₂ concentration (Table 3, Figures 5 and 6; ANOVA P<0.01). The decrease in pK_L appeared to be saturable (Figure 6) such that the greatest change in this parameter was obtained in the presence of 100 mM CaCl₂ (0.51±0.10, n=3) and increasing this concentration by a further 200 mM produced only a further 0.24±0.19 decrease (n=3).

Figure 5.

Representative saturation isotherms of [³H]clobenpropit to sites in guinea-pig cerebral cortex membranes in (a) buffer B_(0,0,0), (b) buffer B_(0.03,0,0), (c) buffer B_(0.07,0,0), (d) buffer B_(0.1,0,0), (e) buffer B_(0.2,0,0) and (f) buffer B_(0.3,0,0). The lines shown superimposed on the data points are the saturation isotherm obtained by fitting the Hill equation with n_H constrained to unity to the data. Guinea-pig cortical membranes (1.6 mg) were incubated for 165 min at 21±3°C in a final volume of 0.5 ml with HEPES–NaOH buffer and [³H]clobenpropit. Total and nonspecific binding of [³H]clobenpropit were defined using appropriate buffer and 1 μM thioperamide, respectively. All determinations were made in triplicate.

Figure 6.

Effect of increasing concentration of CaCl₂ in buffer B on (a) pK_L and (b) B_max of [³H]clobenpropit at sites in guinea-pig cerebral cortex membranes. The line shown superimposed on the data was obtained by fitting a hyperbolic function.

Effect of buffer composition on specific binding of 0.2 nM [³H]clobenpropit

In buffer B_(0,0,0), the percentage specific binding of [³H]clobenpropit (35±1%, n=65) was significantly less than that obtained in the presence of buffer B_{(0.07,0.1,0,1)} (73±1%, n=65; ANOVA P<0.001; see Figure 7). In contrast, the percentage coefficient of variation, calculated from all the triplicate data points forming each competition curve, was significantly greater in buffer B_{(0.07,0.1,0.1)} (6.68±0.24, n=65) than in buffer B_(0,0,0) (4.68±0.14, n=65; t-test, P<0.001).

Figure 7.

Competition curves for H₃-receptor agonists and antagonists at sites labelled with [³H]clobenpropit in guinea-pig cerebral cortex. (a) Effect of increasing concentrations of ligands on [³H]clobenpropit binding (dpm). Data were obtained in a single experiment in buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)} and errors are the mean±s.e.m. of triplicates. (b) Mean competition curve data for ligands, expressed as percentage specific binding, in buffer B_(0,0,0) or (c) buffer B_{(0.07,0.1,0.1)}. Guinea-pig cortical membranes (1.6 mg) were incubated for 165 min at 21±3°C in a final volume of 0.5 ml with HEPES–NaOH buffer, [³H]clobenpropit (0.2 nM) and increasing concentrations of ligands. Total and nonspecific binding of [³H]clobenpropit were defined using appropriate buffer and 1 μM thioperamide, respectively. Data are the mean±s.e.m. of between four and six experiments (see Table 4). The lines shown superimposed on the data for imetit, chloroproxyfan and S-α-methylhistamine were obtained using a two-site fit. The line shown superimposed on the thioperamide data was obtained using a one-site fit.

Changes in ligand pK_app obtained in buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)}

In both buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)} each histamine H₃-receptor ligand produced a concentration–dependent inhibition of the specific binding of [³H]clobenpropit to H₃-receptors in guinea-pig cerebral cortex membranes (e.g. Figure 7). In buffer B_(0,0,0), the estimated mid-point slope parameter estimates (n_H) for the agonist ligands; imetit, N-α-MH, S-α-MH, iodoproxyfan and chloroproxyfan were all significantly less than unity whereas in B_{(0.07,0.1,0.1)} buffer, the n_H values were significantly less than unity for all the agonists (Table 4, t-test P<0.05). The n_H values, estimated from competition curves, for all the antagonist ligands obtained in both buffers were not significantly different from unity (Table 4, t-test, P<0.05).

Table 4.

Parameter estimates for histamine H₃-receptor ligands obtained from analysis of competition experiments performed in buffer containing 3 mM metyrapone (buffer B_(0,0,0)) and in buffer B_{(0.07,0.1,0.1)}

Ligand	n	Buffer B (0,0,0)			Buffer B (0.07,0.1,0.1)			ΔpKI′	ΔpK
		pKI or pKI′	nH	pKIH and pKIL	pKI or pKI′	nH	pKIH and pKIL
Histamine H3-receptor agonists
imetit	4	9.22±0.30	0.63±0.03*	10.20±0.31	8.06±0.07	0.66±0.07*	10.05±0.51	1.16±0.23	1.52±0.29
				8.04±0.34			7.71±0.03
proxyfan	4	8.34±0.10	0.83±0.23		7.68±0.11	0.70±0.09*	9.53±0.11	0.66±0.06	0.98±0.21
							7.36±0.11
chloroproxyfan	4	8.98±0.24	0.77±0.05*	9.47±0.23	8.41±0.02	0.72±0.06*	9.94±0.26	0.58±0.24	0.92±0.20
				8.22±0.36			8.06±0.07
bromoproxyfan	3	8.93±0.25	0.82±0.22		8.37±0.09	0.63±0.03*	10.34±0.19	0.55±0.26	0.99±0.25
							7.94±0.09
iodoproxyfan	5	9.54±0.22	0.71±0.04*	10.04±0.37	8.73±0.12	0.70±0.03*	10.21±0.22	1.21±0.23	1.16±0.13
				8.27±0.36			8.37±0.10
R-α-MH	4	9.74±0.15	1.02±0.31		7.80±0.07	0.61±0.07*	9.07±0.37	1.94±0.11	2.51±0.29
							7.24±0.15
N-α-MH	6	9.58±0.19	0.70±0.04*	10.08±0.10	8.09±0.29	0.51±0.04*	9.82±0.26	1.45±0.23	2.15±0.13
				8.19±0.30			7.43±0.23
S-α-MH	6	7.67±0.14	0.64±0.03*	8.81±0.18	6.29±0.14	0.49±0.02*	8.08±0.22	1.37±0.15	1.98±0.19
				6.90±0.19			5.68±0.12
histamine	4	8.87±0.32	0.83±0.14		7.04±0.15	0.54±0.02*	8.44±0.13	1.50±0.43	2.24±0.32
							6.30±0.09
Histamine H3-receptor antagonists
thioperamide	4	8.41±0.06	0.99±0.08		8.93±0.06	0.92±0.04		−0.52±0.07	−0.52±0.07
iodophenpropit	4	9.72±0.05	1.08±0.09		9.62±0.14	1.03±0.08		0.10±0.14	0.10±0.14
JB96132	4	8.54±0.09	1.23±0.22		8.94±0.07	0.95±0.03		−0.40±0.13	−0.40±0.13
JB96134	4	7.79±0.12	1.15±0.30		7.44±0.03	0.91±0.15		0.35±0.16	0.35±0.16
JB97034	3	7.53±0.24	1.01±0.10		7.58±0.19	0.91±0.10		−0.06±0.05	−0.06±0.05
JB95130	3	6.16±0.31	1.02±0.25		5.97±0.11	1.08±0.03		−0.18±0.20	−0.18±0.20
GR175737	3	8.43±0.03	0.82±0.06		7.95±0.07	0.93±0.07		0.48±0.07	0.48±0.07
GT-2227	3	7.33±0.21	1.26±0.28		6.88±0.21	1.13±0.08		0.45±0.14	0.45±0.14

Abbreviations: N-α-MH, N-α-methylhistamine; R-α-MH, R-α-methylhistamine; S-α-MH, S-α-methylhistamine.

Data are the mean±s.e.m. from the number of assays shown (n). When the mean n_H parameter estimate for a ligand was not different from unity, an affinity (pK_I) value is provided. When n_H is significantly less than unity, pIC₅₀ values were corrected using the Cheng–Prusoff equation but affinity values were assigned the parameter, pK_I′. pK_IH and pK_IL values were obtained by fitting a two-site model to the data. ΔpK_I′ is the difference between pK_I or pK_I′ values in buffer B_(0,0,0) to pK_IL in buffer B_{(0.07,0.1,0.1).} ΔpK is the difference between the pK_I or pK_I′ in buffer B_(0,0,0) to pK_IL in buffer B_{(0.07,0.1,0.1)}.

^*n_H significantly different from unity P<0.05, t-test.

Notwithstanding the non-unit n_H values, obtained with several agonist ligands in both buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)}, in the first instance, we investigated the relationship between the apparent affinity values (pK_I or pK_I′), obtained in the two buffers. The pK_I or pK_I′ values for 14 of the 17 ligands, in buffer B_(0,0,0), were higher than those obtained in buffer B_{(0.07,0.1,0.1)} (Figure 8). All the agonist ligands expressed a lower pK_I or pK_I′ in buffer B_{(0.07,0.1,0.1)} than in buffer B_(0,0,0). In addition, the difference between the estimated pK_I or pK_I′ (ΔpK_I′) of ligands in buffer B_(0,0,0) and in buffer B_{(0.07,0.1,0.1)}, was significantly less (P<0.0001 t-test) for the antagonist ligands (see Figure 8 and Table 4, mean ΔpK_I′=0.03±0.13, n=8; for instance, JB97034 had ΔpK_I′=−0.06±0.05) than for the ligands classified as agonists in the guinea-pig ileum bioassay (mean ΔpK_I′=1.16±0.16, n=9; S-α-MH and R-α-MH, ΔpK_I′=−1.37±0.15 and ΔpK_I′=1.94±0.11, respectively).

Figure 8.

Comparison of the apparent affinities (pK_I or pK_I′) of histamine H₃-receptor agonists and antagonists obtained in buffer B_(0,0,0) and in buffer B containing 70 mM CaCl₂, 100 mM KCl and 100 mM NaCl (buffer B_{(0.07,0.1,0.1)}). The broken line represents the line of identity.

There was a significant effect of tissue preparation on agonist affinity (pK_I or pK_I′) (P<0.001) in buffer B_(0,0,0) but not in buffer B_{(0.7,0.1,0.1)} (P>0.1, ANOVA). ANOVA indicated that there was a significant effect of tissue preparation and agonist on the magnitude of the ΔpK_I′ (tissue preparation P<0.025, agonist P<0.001). There was no significant effect of tissue preparation on antagonist affinity (pK_I) when determined in buffer B_(0,0,0) or buffer B_{(0.07,0.1,0.1)} (ANOVA, P>0.1). There was a significant effect of antagonist but no significant effect of tissue preparation on the magnitude of the ΔpK_I′ (antagonist P<0.001, tissue preparation P>0.1 ANOVA).

Comparison of ligand pK_app values (pK_I′), obtained in buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1),} with pK_app, pA₂ or pK_B values obtained in the guinea-pig ileum bioassay

When the pK_I or pK_I′ value, obtained for each ligand in buffer B_(0,0,0), was compared with the pK_app or pA₂ value estimated in the guinea-pig ileum bioassay, the data points for the ligands appeared randomly scattered and 14 of the 17 data points lay below the expected line of identity (y=x), that is, the ligands expressed an affinity that was higher than expected from the pK_app or pA₂ value obtained in the functional bioassay (Figure 9a). When the same comparison was made using pK_I or pK_I′ values, obtained in buffer B_{(0.07,0.1,0.1)}, the data points were more linearly distributed and closer to the line of identity (Figure 9b). The mean deviation from the line of identity (∑(pK_i – pK_app, pA₂ or pK_B)/n) was greatest when the pK_I or pK_I′ values were estimated in buffer B_(0,0,0) rather than buffer B_{(0.07,0.1,0.1)} (B_(0,0,0), mean deviation from pK_app, pA₂ or pK_B=1.09±0.21; B_{(0.07,0.1,0.1)} mean difference from pK_app, pA₂ or pK_B=0.44±0.08; Figure 9a and b).

Figure 9.

Comparison of the affinities of histamine H₃-receptor agonists (pK_app) and antagonists (pA₂ or pK_B) at H₃-receptors in the guinea-pig ileum bioassay (see Table 2) with those estimated in radioligand binding assays (see Table 4). pK_I or pK_I′ values for (a) agonists and antagonists in standard buffer (B_(0,0,0)), (b) agonists and antagonists in B_{(0.07,0.1,0.1)}, (c) agonists in B_(0,0,0), (d) agonists in B_{(0.07,0.1,0.1)}, (e) antagonists in B_{(0.07,0.1,0.1)} and (f) antagonists in B_{(0.07,0.1,0.1).} The broken line represents the line of identity (y=x).

When the antagonist pK_I values, were compared with pK_B or pA₂ values estimated in the guinea-pig ileum assay, there was no significant change in the deviation from y=x when pK_I values were estimated in buffer B_(0,0,0) or buffer B_{(0.07,0.1,0.1)} (0.41±0.14 and 0.30±0.12, respectively; t-test, P>0.05; Figure 9e and f). In contrast, when only the agonist pK_I or pK_I′ values, obtained in buffer B_(0,0,0) were compared with pK_app values estimated in the guinea-pig ileum assay, the deviation from y=x was 1.69±0.24. When the same comparison was performed using pK_I′ values obtained in buffer B_{(0.07,0.1,0.1)}, the data points were significantly closer to y=x (0.55±0.08; t-test, P<0.0005; Figure 9c and d) but 8 of the 9 data points still lay below the line of identity. In buffer B_(0,0,0), the mean deviation from y=x, was significantly greater for agonists than antagonists (t-test, P<0.005) but there was no significant difference in the degree of deviation from y=x in buffer B_{(0.07,0.1,0.1)} (t-test, P>0.05).

In light of this finding and the observation that all the mean agonist n_H parameter estimates were significantly less than unity in buffer B_{(0.07,0.1,0.1)}, the agonist competition data were analysed further. The goodness of fit of the Hill equation and of the Hill equation with n_H constrained to unity, were compared. The Hill equation with unconstrained slope provided a significantly improved fit to the data for all agonists (F-test, P<0.05) and therefore a two-site model was fitted to the data. The two affinity values (pK_IL and pK_IH) that were obtained are presented in Table 4. When the pK_IL values, obtained from the two-site fit of the agonist competition curves in buffer B_{(0.07,0.1,0.1)}, were compared with pK_app values estimated in the guinea-pig ileum assay, the data points were evenly distributed about the line of identity and the deviation from y=x (0.08±0.06) was significantly less than when pK_I′ values were compared (0.55±0.08, Figure 10).

Figure 10.

Comparison of (a) pK_I′ and (b) pK_IL values for agonists in buffer B_{(0.07,0.1,0.1)} with ileum pK_app estimates.

Comparison of α-values obtained in the guinea-pig ileum with ΔpK values from radioligand binding

There appeared to be a relationship between agonist ΔpK values and α-values estimated in the guinea-pig ileum bioassay such that agonists with higher α expressed higher ΔpK values (Figure 11).

Figure 11.

Comparison of ligand ΔpK values and α measured in the guinea-pig ileum bioassay.

Discussion

In this study, we have used a guinea-pig ileum bioassay to determine the affinity (pA₂) of histamine H₃-receptor antagonists and also the α and pK_app of a series of histamine H₃-receptor agonists at guinea pig H₃-receptors. In addition, we have investigated the effect of altering buffer composition on the apparent affinity (pK_I′) that ligands, with known α, express in competition studies performed in the absence (buffer B_(0,0,0)) and presence of buffer salts (buffer B_{(0.07,0.1,0.1)}); with the aim of establishing whether assays of this type can be used to measure the pK_app and to detect residual intrinsic efficacy of ligands. We refer to ileum H₃-receptor affinity values estimated by the method of Furchgott, as pK_app values, in order to take account of the problem that defining this parameter as pK_A assumes that activation of the receptor has no effect on binding and this is not the case for currently proposed models of agonist action (see Colquhoun, 1998). The term pK_app is used to indicate that the measurement is a macroscopic equilibrium constant describing the overall constant for the binding of agonist to receptor and subsequent isomerization of this receptor to form AR^* (see Neubig et al., 2003).

The pA₂ values of the previously described H₃-receptor antagonist ligands, thioperamide, clobenpropit and GR175737, in the guinea-pig ileum, were comparable to those reported previously (Arrang et al., 1987, 1990; Van der Goot et al., 1992; Barnes et al., 1993; Clitherow et al., 1996; Valentine et al., 1999). The affinity of iodophenpropit and GT-2227 were underestimated compared with those reported previously (iodophenpropit pA₂=9.6, Jansen et al., 1992; GT-2227, pA₂=7.9, Tedford et al., 1998); this underestimation may have been a consequence of the 1 h antagonist preincubation being insufficient for equilibration of these antagonists.

In the radioligand-binding studies, the competition curves for all the agonist ligands, with the exception of R-α-MH and histamine, in buffer B_(0,0,0) and for all the agonists in buffer B_{(0.07,0.1,0.1)}, were associated with n_H parameter estimates that were significantly less than unity. This behaviour has been well described for the binding of agonists at many types of receptor and a number of models have been proposed to explain the phenomenon (e.g. Lefkowitz et al., 1993; Samama et al., 1993; Weiss et al., 1996). In light of n_H values less than unity, it could be argued that only the pIC₅₀ values for R-α-MH and histamine should be corrected using the Cheng–Prusoff equation because the derivation of this correction relies on simple competition between two ligands at a homogenous receptor population and therefore should be applied only when n_H is not different from unity. However, in this study we have corrected all pIC₅₀ values using the Cheng–Prusoff equation to correct for the differential occupancy of ∼0.2 nM [³H]clobenpropit in the two buffers (buffer B_(0,0,0), pK_L=10.36 and buffer B_{(0.07,0.1,0.1)}, pK_L=9.82). Ideally, in order for this not to be a confounding problem, we would have performed competition studies in both buffers at a concentration of [³H]clobenpropit, which was equivalent to its pK_L. However, this was not possible because the low specific activity of the radioligand resulted in too small a specific-binding window in buffer B_(0,0,0) at a ∼0.04 nM concentration. We also corrected all pIC₅₀ values irrespective of n_H parameter because although R-α-MH and histamine had n_H values that were not different from unity, it appeared from functional pEC₅₀ values that their pIC₅₀ values were likely to have been overestimated by as much if not more than other agonists, where the n_H parameter was significantly less than unity (e.g. chloroproxyfan). Thus, the pIC₅₀ value of R-α-MH in buffer B_(0,0,0) was 8.96±0.16 and the pEC₅₀ value in the functional assay was 7.64, whereas the pIC₅₀ value of chloroproxyfan was 8.17±0.24 and the pEC₅₀ value was 7.85. Clearly, if the same mechanism underlies the binding of all agonists it is equally inappropriate to correct an agonist pIC₅₀ value when the n_H value is not different from unity as when the n_H is less than unity. To make a distinction between pIC₅₀ values that have been corrected for [³H]clobenpropit occupancy where n_H was equal to unity and those where n_H was less than unity and where strictly speaking the Cheng–Prusoff equation should not have been applied, we have assigned the latter the parameter pK_I′.

In the radioligand-binding studies, affinity values (pK_I or pK_I′) for the agonist ligands, in the absence of buffer salts (buffer (B_(0,0,0))), were higher than their estimated pK_app or pA₂ values obtained in bioassay studies (Figure 9a), which confirmed our previous observations where we used [³H]R-α-MH as the radioligand (Harper et al., 1999a). This indicated that, despite using an antagonist radioligand, it was still possible to obtain overestimated ‘high' affinity binding values for agonists. The finding that the apparent affinity values (pK_I′) of ligands in buffer B containing salts (buffer B_{(0.07,0.1,0.1)}) were reduced relative to those obtained in buffer B_(0,0,0) was also consistent with our preliminary studies in which [³H]R-α-MH was used as the radioligand and indicated that it was also possible to obtain ‘low' affinity estimates for agonist ligands when using the antagonist radioligand, [³H]clobenpropit.

The observation that the change in pK_I′ between that obtained in buffer B_(0,0,0) and that obtained in buffer B_{(0.07,0.1,0.1)} (ΔpK_I′) was greater for agonists than antagonists (agonists=1.16; antagonists=0.03) and that there was a significant effect of agonist but not antagonist on the ΔpK_I′ (agonist P<0.001, antagonist P>0.1) suggested that the change in pK_I′ value, brought about by modification of the buffer composition, was related to a property of the ligands, which was only expressed by agonists. It is possible to explain the high- and low-affinity binding of agonists in H₃-receptor radioligand binding assays by considering the extended ternary complex model (TCM) developed by Lefkowitz and colleagues (Samama et al., 1993; Lefkowitz et al., 1993; see Figure 12a) and also by the cubic TCM that was described some years later (Weiss et al., 1996; see Figure 12b). In the extended TCM, it is proposed that the receptor can exist in a low-affinity agonist state (R) and a high-affinity state (R^*), which can also interact with a G-protein (R^*G) in the absence of agonist. Thus, the high-affinity H₃-receptor agonist binding may be a consequence of the agonist binding to R^* or R^*G. In the cubic TCM, it is postulated that the receptor can exist in a low-affinity state (R_i equivalent to R in the TCM) and a high-affinity state (R_a, equivalent to R^* in the TCM) and that both R and R^* can exist as high-affinity states as a consequence of interaction with a G-protein (RG and R^*G). Therefore, according to this model, high-affinity H₃-receptor agonist binding could result from either binding to preformed high-affinity receptor states (R^* and R^*G) or from the induction of these high-affinity states through binding to low-affinity receptors (R). Low-affinity agonist binding in buffer B_{(0.07,0.1,0.1)} can be explained by considering that, under these conditions, the agonist binds only to the low-affinity receptor state (R) and cannot bind to or induce the formation of R^* or R^*G. Although, at the time these studies were performed it was not definitely known that the H₃-receptor was G-protein coupled because it had not yet been cloned, the possibility that agonists could induce H₃-receptor ternary complex formation was supported by studies suggesting that these receptors were linked to effector systems through G-proteins (Arrang et al., 1990; West et al., 1990; Zweig et al., 1992; Clark et al., 1993; Litosch et al., 1993; Clark and Hill, 1995; Clark and Hill, 1996).

Figure 12.

(a) The extended TCM as described by Samama et al. (1993). (b) The cubic TCM of Weiss et al. (1996). In both models, H is hormone, G is G protein and R is the inactive receptor state and R^* the ‘active' state. To allow simple comparison of the models, the terms used for ‘inactive', R_i and ‘active' receptor states (R_a) in the model of Weiss et al. (1996) have been modified to R and R^*, respectively.

It seems unlikely that the apparent changes in ligand affinity obtained using buffer B_(0,0,0) and buffer B_{(0.07,0.1,0.1)} could have arisen simply as a consequence of highly variable data because although the percentage specific binding of [³H]clobenpropit was higher in buffer B_{(0.07,0.1,0.1)} (73±1%) compared with that obtained in buffer B_(0,0,0) (35±1%), the % coefficient of variation of the data points in buffer B_(0,0,0) was lower than that in buffer B_{(0.07,0.1,0.1).} Furthermore, we previously found that although the percentage specific binding of [³H]clobenpropit in buffer B_(0,0,0) is low (∼40%), there was a close correlation between histamine H₃-receptor pK_I values estimated using this radioligand and a radioligand exhibiting considerably higher percentage specific binding ([³H]R-α-MH with 89%; Harper et al., 1999a, 1999b).

The pretext for performing competition studies in assay buffer containing salts, that is, that it could allow estimation of agonist affinity (pK_app), was supported by comparisons made between the radioligand binding pK_I′ values and the pK_app or pA₂ values obtained in the guinea-pig ileum bioassay (Figure 9). Thus, the deviation from the expected y=x for all the ligands, was significantly lower in modified buffer (buffer B_(0,0,0)=1.09±0.21; B_{(0.07,0.1,0.1)}=0.44±0.08; Figure 9a and b). In addition, when the same analysis was performed on subsets of ligands, characterized as agonists (Figure 9c and d) or antagonists (Figure 9e and f), in the guinea-pig ileum, the deviation from the expected y=x was increased for the agonists (buffer B_(0,0,0)=1.69±0.24, Figure 9c) and unchanged for the antagonists (buffer B_(0,0,0)=0.41±0.14, Figure 9e) relative to the complete data set. In addition, the deviation from y=x was significantly decreased when agonist but not antagonist pK_I values were determined in the buffer containing salts (buffer B_{(0.07,0.1,0.1)} agonists=0.55±0.08, Figure 9d; antagonists=0.30±0.12, Figure 9e) and, furthermore, the mean deviation from y=x was only significantly greater for agonists than that for antagonists when the pK_I′ or pK_I values were estimated in buffer B_(0,0,0).

Notwithstanding the finding that the deviation from the expected y=x for the antagonists was unchanged by buffer, we noticed that for one antagonist, iodophenpropit, there was a large difference between the pA₂ and pK_I obtained in both buffers (∼1 log unit; pA₂=8.82±0.34; pK_I buffer B_(0,0,0)=9.72±0.05, buffer B_{(0.07,0.1,0.1)}=9.62±0.14). This discrepancy could be explained by considering that the pA₂ value of iodophenpropit was underestimated because it had not been preincubated with tissue for long enough to reach equilibrium. In support of this possibility, Jansen et al. (1992) reported a pA₂ of 9.6 for iodophenpropit and we found that the pA₂ value of another H₃-receptor antagonist with similar structure, clobenpropit, was increased when preincubated with tissue for 3 h.

Despite the similarity between pK_app values in the ileum bioassay and the pK_I or pK_I′ values of ligands in buffer B_{(0.07,0.1,0.1)} (see Figure 9b), it was still apparent that eight of the nine agonist data points still lay below the expected line of identity (y=x, Figure 9d). It seems unlikely that we found that the ileum pK_app values were lower than the pK_I′ values in buffer B_{(0.07,0.1,0.1)} simply because of experimental error and because they were estimated, albeit with the exception of R-α-MH, in a single experiment. This is because, for one agonist, proxyfan, the pK_app (7.66±0.49) from model fitting was comparable to the pA₂ determined by investigating the effect of this ligand on R-α-MH concentration effect curves (7.34±0.10, n=4; Table 4 and Schlicker et al., 1996, pA₂=7.12). In addition, the pK_app values for R-α-MH and histamine were comparable to those previously reported by Taylor and Kilpatrick (1992) (R-α-MH=7.01; histamine=6.13).

The possibility that, even in buffer B_{(0.07,0.1,0.1)}, the agonist pK_I′ values were not equivalent to the pK_app estimated in the ileum bioassay is supported by the observation that all the agonist competition curves in this buffer had n_H parameter estimates, which were significantly less than unity (Table 4). The cubic TCM (Weiss et al., 1996; see Figure 12b) predicts this behaviour if it is considered that the change in composition of the assay buffer (buffer B_{(0.07,0.1,0.1)}) is not sufficient to prevent the agonist inducing some high-affinity receptor states (ARG and/ or AR^*G) or from binding to pre-existing high-affinity receptors (R^* or RG). Therefore, according to this model, the flat competition curves result from competition for labelled R (low-affinity binding) and some induction or binding to R^*G or RG (high-affinity binding component). The data cannot be explained by competition for labelled R and R^* because in this situation competition curves would have unit slope. Interestingly, in support of this explanation, when the buffer B_{(0.07,0.1,0.1)} agonist competition data were analysed using a two-site model, and the low-affinity estimates (pK_IL) were compared with the ileum pK_app values, the data points were closer to y=x (Figure 10).

A possible explanation for the decrease in agonist affinity in buffer B_{(0.07,0.1,0.1)} is that, in this buffer, the metal ions interfere in some way with the ability of agonists to induce high-affinity ternary complex receptor states. Indeed, many studies have demonstrated that agonist, but not neutral antagonist, binding is sensitive to metal ions (Limbird et al., 1982; Puttfarcken et al., 1986; Gowraganahalli et al., 1990) and have suggested that the effect that metal ions have on agonist binding is owing to them, in some way, preventing the formation of ternary complex between agonist, receptor and G-protein (Childers and La Riviere, 1984; Lambert and Childers, 1984; Kim and Neubig, 1985; Lynch et al., 1985; Demoliou-Mason and Barnard, 1986; Puttfarcken et al., 1986; Carraway et al., 1992). However, there are two other possible explanations to explain why the modified buffer reduces the high-affinity agonist binding and why some high-affinity binding remains in the modified buffer (B_{(0.07,0.1,0.1)}). This is because adding NaCl, CaCl₂ and KCl, also alters the osmotic and ionic strength of buffer B_(0,0,0 (buffer B_{(0.07,0.1,0.1)} osmotic strength=290 mM; ionic strength=430 mM; buffer B_(0,0,0) osmotic strength=20 mM; ionic strength=20 mM). In fact, buffer ionic strength has been previously shown to reduce agonist affinity (Arias, 1996). However, our previous observation that 70 mM CaCl₂ produced a greater reduction in [³H]R-α-MH pK_L than 100 mM NaCl (data not shown) suggests that the osmotic strength of the buffer does not result in low-affinity agonist binding. Nonetheless, in retrospect, it would have been interesting to perform further competition studies to establish whether osmotic strength contributed to the reduction in agonist affinity, by adding glucose to buffer B_(0,0,0).

To establish whether agonist competition curves with unit slope could be obtained, which at the same time had affinity estimates equal to ileum pK_app estimates and therefore to test the hypothesis that flat agonist competition curves, obtained in buffer B_{(0.07,0.1,0.1),} resulted from some formation of ARG and/or AR^*G, we performed further competition experiments in which the concentration of just one of the salts, CaCl₂, was increased. The results we obtained were consistent with this hypothesis. The pK_I′ of R-α-MH decreased with increasing CaCl₂ concentration, whereas the pK_I of thioperamide remained unchanged (Table 5), indicating that CaCl₂ alone had the same effect as the combination of NaCl, KCl and CaCl₂ in buffer B_{(0.07,0.1,0.1)}. In addition, the decrease in R-α-MH pK_I′ was associated with an increase in n_H such that at the highest CaCl₂ concentration (300 mM) the n_H parameter estimate was not different from unity and, moreover, at this CaCl₂ concentration, the pK_I of R-α-MH (7.24±0.06, n=3) was not different to the pK_IL estimated from a two-site analysis of data obtained in buffer B_{(0.07,0.1,0.1)} (7.24±0.15, n=4). An alternative approach to elucidate whether the nonunit agonist competition curves in buffer B_{(0.07,0.1,0.1)} had resulted from the agonists binding to, or inducing the formation of ternary complex (ARG), would have been to add guanine nucleotides to the assay buffer. In addition, when these studies were performed, it had been suggested that the H₃-receptor was coupled to either Gi (Clark et al., 1993; Litosch et al., 1993; Clark and Hill, 1996) or Gs-proteins (Cherifi et al., 1992). Therefore, it would also have been interesting to establish if treatment of tissues with pertussis or cholera toxin resulted in unit n_H parameter estimates for agonists in buffer B_{(0.07,0.1,0.1)}.

Table 5.

Apparent affinity (pK_I′) and n_H values expressed by R-α-MH and thioperamide at H₃-receptors in guinea-pig cortex in buffer containing increasing concentrations of CaCl₂

Buffer	R-α-MH		Thioperamide
	pKI′	nH	pKI	nH
B(0,0,0)	9.03±0.08	0.54±0.04*	8.28±0.08	1.04±0.28
B(0.03,0,0)	8.59±0.21	0.52±0.04*	8.55±0.09	0.93±0.14
B(0.07,0,0)	8.45±0.07	0.54±0.02*	8.46±0.08	0.90±0.04
B(0.1,0,0)	8.09±0.13	0.60±0.06*	8.38±0.16	0.87±0.05
B(0.2,0,0)	7.68±0.07	0.68±0.03*	8.44±0.08	1.14±0.13
B(0.3,0,0)	7.24±0.06	0.88±0.08	8.39±0.06	0.95±0.02

Data are the mean±s.e.m. of three experiments. The concentrations of CaCl₂ used in the buffers were the same as those in the experiments described in Table 3.

^*n_H significantly different from unity, P<0.05 t-test.

The observation that CaCl₂ decreased the pK_L of [³H]clobenpropit (Figure 6) was consistent with the effect that buffer containing 70 mM CaCl₂, 100 mM KCl and 100 mM NaCl (buffer B_{(0.07,0.1,0.1)}) had on this parameter. That this effect appeared saturable was consistent with the possibility that the increased buffer CaCl₂ concentration or increased buffer ionic strength prevented clobenpropit from binding to or inducing high-affinity receptor states. These states cannot be equivalent to R^*G or RG because the n_H parameter estimates for clobenpropit at all CaCl₂ concentrations were not significantly different from unity. However, it is possible that the higher affinity of [³H]clobenpropit, in buffer B_(0,0,0), results from it inducing or binding to R^*.

The finding of a relationship between α measured in the ileum bioassay and the ΔpK (Tables 2 and 4 and Figure 11) suggests that H₃-receptor radioligand binding assays can be used to detect residual agonist efficacy. Thus, ligands with ΔpK values of less than 1.00 were partial agonists (α=0.35–0.90) in the guinea-pig ileum assay and those ligands with ΔpK values of >1.1 were full agonists (α∼1.00). The finding that GR175737, an antagonist ligand as defined by the guinea-pig ileum bioassay, had a ΔpK value of 0.48 can be accounted for by considering that the H₃-receptor radioligand binding assay detects intrinsic efficacy that has remained undetected in the bioassay. Thus, if a full agonist (α∼1.0, R-α-MH) has a ΔpK of 2.51 and expresses a pK_I′ value approximately 2.5 log units higher than its pK_app, a partial agonist (α∼0.45, chloroproxyfan) has a ΔpK of 0.92 and expresses a pK_I′ value approximately 1 log unit higher than its pK_app, then it is possible that a weaker partial agonist, which acts as a competitive antagonist in the functional assay, could express a pK_I′ value that is still significantly higher than its pK_B. Interestingly, since these studies were conducted, GR175737 has been shown to be a partial agonist (α∼0.4) (Wulff et al., 2002) in a cAMP assay which also found that proxyfan, a partial agonist in the guinea-pig ileum assay (α=0.35, pEC₅₀=7.29), was a full agonist with over 1 log unit higher potency (pEC₅₀∼8.4).

Conclusion

We have manipulated the conditions of the H₃-receptor radioligand-binding assay to provide a method of obtaining a measure of both the pK_app and intrinsic efficacy of novel H₃-receptor ligands. The assay predicts that some ligands previously classified as H₃-receptor antagonists may possess residual agonist efficacy, so that under certain conditions radioligand-binding assays may be a more sensitive detector of agonist intrinsic efficacy than functional in vitro assays. The prospect that radioligand-binding assays can be used to detect intrinsic efficacy may be useful for the study of human receptors in native tissue, where it may not be possible to develop a functional assay. Studies of this type may be useful for excluding the possibility that, in human tissues, the receptor dimerizes with other receptor types or interacts with tissue-dependent factors (e.g. RAMP and scaffolding proteins) that modify the receptor pharmacology, such that ligands defined in recombinant systems as antagonists are found to express intrinsic efficacy.

Abbreviations

EEDQ

N-ethoxycarbonyl-2-ethoxy-1, 2-dihydroquinoline

N-α-MH

N-α-methylhistamine

PEI

polyethyleneimine

R-α-MH

R-α-methylhistamine

S-α-MH

S-α-methylhistamine

Conflict of interest

The authors state no conflict of interest.