Analysis of the behaviour of selected CCK_B/gastrin receptor antagonists in radioligand binding assays performed in mouse and rat cerebral cortex

Abstract

1. The previously described complex behaviour of the CCK_B/gastrin receptor antagonist, L-365,260, in radioligand binding assays could be explained by a variable population of two binding sites. We have investigated whether other CCK_B/gastrin receptor ligands (PD134,308, PD140,376, YM022 and JB93182) can distinguish between these sites.

2. In the mouse cortex assay, Hill slopes were not different from unity and the ligand pK_I values did not differ when either [¹²⁵I]-BH-CCK-8S or [³H]-PD140,376 was used as label as expected for a single site (G₂).

3. In the rat cortex, where previous analysis of replicate (n=48) L-365,260 data indicated the presence of two CCK_B/gastrin sites (G₁ and G₂), the competition data for PD134,308, PD140,376, YM022 and JB93182 could be explained by a homogeneous population of CCK_B/gastrin sites because the Hill slope estimates were not significantly different from unity. However, the estimated affinity values for JB93182 and YM022 were significantly higher and that for PD134,308 was significantly lower than those obtained in the mouse cortex when the same radioligand was used. In view of our previous data obtained with L-365,260, the rat cortex data were also interpreted using a two-site model. In this analysis, SR27897 expressed ∼9 fold, PD134,308 ∼13 fold and PD140,376 ∼11 fold selectivity for the G₂ site. In contrast, JB93182 expressed ∼23 fold and YM022 ∼4 fold selectivity for the G₁ site. If the two-site interpretation of the data is valid then, because of its reverse selectivity to L-365,260, JB93182 has been identified as a compound which if radiolabelled could provide a test of this receptor subdivision.

Introduction

There is a high degree of sequence homology among the cloned CCK_B/gastrin receptors from several species (Pisegna et al., 1992). However, the CCK_B/gastrin receptor antagonist, L-365,260, expresses species-dependent affinities that were originally attributed to single amino-acid differences within the sixth transmembrane domain of the receptor (Beinborn et al., 1993). More recently, two isoforms of the receptor have been identified which are produced as a consequence of alternate gene splicing (Song et al., 1993). Although agonists are reported not to distinguish between the two receptor isoforms in terms of binding and transduction, L-365,260 has been shown to be 3 fold more potent at competing with ¹²⁵I-Bolton-Hunter-labelled-CCK-8S for the human short receptor isoform than for the long receptor isoform (Ito et al., 1994; Wank et al., 1994).

Until recently, there was no evidence from functional bioassays for CCK_B/gastrin receptor heterogeneity within a species (Presti & Gardener, 1993). However, we recently reported an analysis of the variation in L-356,260 data obtained in replicate experiments performed over a 3 year period in both radioligand binding (Harper et al., 1996a) and in isolated tissue assays (Roberts et al., 1996a). Overall, the data were not consistent with the pharmacological expression of a single CCK_B/gastrin receptor, in both central and peripheral tissues, but they could be accounted for by the variable expression of two CCK_B/gastrin receptor subtypes or states which we referred to as gastrin-G₁ and -G₂. The mouse cortex assay appeared to express a homogeneous population of sites characterized by a pK_I for L-365,260 of 8.41 (G₂) whereas the rat cortex assay expressed a variable proportion of two sites characterized by pK_I values of 8.48 (G₂) and 7.22 (G₁). It was only possible to obtain affinity estimates at the two sites when we simultaneously analysed a larger number of data sets (47/48 individual experiments) because of apparent within-assay variation in the proportion of the two sites and the low subtype selectivity (∼18 fold) of L-365,260.

Although the physiological role of the two proposed CCK_B/gastrin receptors has not been elucidated, the mRNA coding for the two receptor isoforms, which may be identical to these pharmacologically-defined subtypes, has been shown to vary between tissues (Biagini et al., 1995). Therefore, having made the observation that, under our assay conditions, L-365,260 can distinguish between two CCK_B/gastrin receptors we have investigated whether other CCK_B/gastrin receptor antagonists could also distinguish between these sites. Accordingly, we have analysed the behaviour of a selection of previously characterized CCK/gastrin receptor ligands in the mouse and rat cortex assays. As in the previous analysis of the behaviour of L-365,260, the data analysed was obtained over a number of years as part of a drug discovery programme (e.g. Kalindjian et al., 1994). We decided to use the two cortex assays because, if the compounds could distinguish between the two pharmacologically defined subtypes, we expected to obtain simple behaviour in the mouse cortex but variable behaviour in the rat cortex. In addition, we determined the CCK_A receptor-selectivity of the compounds in a guinea-pig pancreas radioligand binding assay. In particular, the CCK_A selective antagonist, SR27897 (Gully et al., 1993), was used to confirm that any complexity in cortex data was not a consequence of the compounds interacting with CCK_A receptors in this tissue.

We selected compounds from the large number of CCK_B/gastrin receptor ligands that have been described including L-365,260 (Lotti & Chang, 1989), PD134,308 (Hughes et al., 1990), L-156,586 (Lam et al., 1990), PD136,450 (Horwell et al., 1991), PD140,376 (Hunter et al., 1993), LY262691 (Howbert et al., 1993), YM022 (Nishida et al., 1994), L-740,093 (Patel et al., 1994), CP-212,454 (Lowe et al., 1994), RP72540 (Bertrand et al., 1994), RP73870 (Bohme et al., 1994; Pendley et al., 1995), RPR101367 (Bertrand et al., 1995), GV150013 (Corsi et al., 1995) and JB93182 (Kalindjian et al., 1996). When these compounds were originally described there was no mention of data complexity and no suggestion of CCK_B/gastrin receptor heterogeneity. Therefore, the compounds used in this study (PD134,308, PD140,376, YM022, JB93182) were selected on the basis of their structural diversity, in the hope that this approach would increase the likelihood of exposing subtype selectivity, and for the compound-specific reasons detailed below.

YM022 was selected because it is one of the highest affinity CCK_B/gastrin receptor antagonists which has been described (Nishida et al., 1994). The two `peptoid' CCK_B/gastrin ligands, PD134,308 and PD140,376, were included in the analysis because they appeared to express lower affinities at CCK_B/gastrin receptors in the isolated lumen-perfused rat stomach assay (Shankley et al., 1997) than at CCK_B/gastrin binding sites in the mouse and guinea-pig cortex (Hughes et al., 1990; Hunter et al., 1993). Moreover, an analysis of the behaviour of L-365,260 in replicate experiments suggested that the rat stomach assay expresses a homogeneous gastrin receptor population with low affinity for L-365,260 (Roberts et al., 1996a). Finally, JB93182, was included because data obtained in the lumen-perfused rat stomach assay suggested that this compound had a higher affinity at receptors present in this tissue (G₁; Roberts et al., 1996b) than for those in the mouse cortex (G₂). Therefore, JB93182 might be expected to exhibit complex behaviour in the rat cortex assay and express opposite selectivity to L-365,260 for the proposed two sites.

A preliminary account of this study was presented to the British Pharmacological Society (Harper et al., 1996b).

Methods

Preparation of mouse and rat cortex cells

The preparation of mouse and rat cortex cells, as judged by light microscopy, was based upon the method of Clarke et al. (1986). Young adult male mice (Charles River 25–30 g) and male rats (Wistar 250–500 g) were killed by cervical dislocation. The cortex was immediately dissected and placed in ice-cold HEPES-NaOH buffer (pH 7.2 at 21±3°C) of the following mM composition: NaCl 130; KCl 4.7; MgCl₂ 5; HEPES 10, EGTA 1 plus bacitracin 0.125 g l⁻¹. The tissue was weighed and homogenized in 40 ml of ice-cold buffer using a Teflon-in-glass homogenizer. The homogenate was centrifuged at 39,800×g for 20 min at 4°C, the supernatant discarded and the pellet resuspended by homogenization in fresh HEPES-NaOH buffer and recentrifuged. For radioligand binding assays using ¹²⁵I-Bolton Hunter labelled cholecystokinin-8S ([¹²⁵I]-BH-CCK-8S), the resulting pellets from rat and mouse cortex were resuspended in HEPES-NaOH buffer to give tissue concentrations of 5 mg ml⁻¹ and 2 mg ml⁻¹ (original wet weight), respectively. At these tissue concentrations the percentage of added ligand bound in the rat and mouse cerebral cortex assay was 9.2±0.7 and 11.8±1.3% of that added, respectively (n=3 to 6; Harper et al., 1996a). The non-specific binding in the mouse was 22.8±1.9 and in the rat 49.0±1.5% of the total bound radioligand. Mouse cortex cells were resuspended to give a tissue concentration of 8 mg ml⁻¹ (original wet weight) for assays using [³H]-PD140,376. At this concentration the percentage of added radioligand bound was 12.3±2.4% and the percentage non-specific binding was 36.0±8.1% (n=5).

Preparation of guinea-pig pancreatic acini

Adult male Dunkin Hartley guinea-pigs (200–300 g) were killed by cervical dislocation. The pancreas were removed, dissected from connective tissue and fat and placed in ice-cold HEPES-NaOH buffer. The tissue was weighed and homogenized in 40 ml of ice-cold buffer using a Polytron PT10 (4×1s; setting 10). The homogenate was centrifuged at 39,800×g for 20 min at 4°C and the supernatant discarded. The pellet was resuspended by homogenization in 80 ml of ice-cold HEPES-NaOH buffer, using a Teflon-in-glass homogenizer, and recentrifuged. The final pellet was resuspended in HEPES-NaOH buffer, containing 0.375 μM PD134,308, to a tissue concentration of 1 mg ml⁻¹ (original wet weight) and filtered through 500 μm pore-size Nytex mesh.

Incubation conditions–CCK_B/gastrin receptor competition studies

All CCK/gastrin receptor antagonists were diluted in HEPES-NaOH buffer. Aliquots (50 μl) of competing ligands at concentrations from 0.10 pM to 100 μM, were incubated in triplicate with mouse or rat cortex cells (400 μl) in a final volume of 500 μl with appropriate buffer containing [¹²⁵I]-BH-CCK-8S (50 μl; 200 pM) or [³H]-PD140,376 (50 μl; 1.5 nM). Non-specific binding was defined with 1 μM L-365,260. [¹²⁵I]-BH-CCK-8S has previously been shown not to label CCK_A-receptors under these assay conditions (Harper et al., 1996a). Assays were incubated for 150 min at 21±3°C and then terminated by rapid filtration through pre-soaked Whatman GF/B filters which were washed (3×3 ml) with ice-cold 50 mM Tris-HCl buffer (pH 6.9 at 21±3°C). Bound [¹²⁵I]-BH-CCK-8S was determined by counting each filter (1 min) in a LKB Clinnigamma counter. Bound [³H]-PD140,376 was determined by liquid scintillation spectroscopy. Filters were transferred into scintillation vials, 5 ml Beckman Ready-Solv HP liquid scintillation cocktail added and after a further 4 h the bound radioactivity determined by counting (5 min) in a Beckman LS6000 liquid scintillation counter.

Incubation conditions CCK_A receptor competition studies

All CCK/gastrin receptor antagonists were diluted in HEPES-NaOH buffer. Aliquots (50 μl) of concentrations from 1 pM to 100 μM, were incubated in triplicate with guinea-pig pancreatic acini (400 μl; containing 0.375 μM PD134,308) in a final volume of 500 μl with appropriate buffer containing [¹²⁵I]-BH-CCK-8S (50 μl; 200 pM). Non-specific binding was defined with 1 μM L-364,718. [¹²⁵I]-BH-CCK-8S does not label CCK_B/gastrin receptors under these assay conditions because a final assay concentration of 0.3 μM PD134,308, which would be expected to produce >95% occupancy of CCK_B/gastrin receptors, was included in the assay buffer. In addition, high concentrations of L-365,260 (>0.1 μM) were required to produce significant inhibition of binding (Table 1). After a 150 min incubation at 21±3°C, assays were terminated by rapid filtration through pre-soaked Whatman GF/B filters which were washed (3×3 ml) with ice-cold 50 mM Tris-HCl buffer (pH 6.9 at 21±3°C). Bound [¹²⁵I]-BH-CCK-8S was determined by counting each filter (1 min) in a LKB Clinnigamma counter.

Table 1.

Analysis of data sets from competition experiments between [¹²⁵I]-BH-CCK-8S and CCK/gastrin receptor ligands for CCK_A binding sites in guinea-pig pancreas

Data analysis

The individual competition curve data were expressed as the percentage of the decrease in specific binding of either [³H]-PD140,376 or [¹²⁵I]-BH-CCK-8S within each experiment. When [³H]-PD140,376 was used as label, dissociation constants (K_I) were determined using the Cheng & Prusoff (1973) equation in order to correct for the receptor occupancy by the radioligand,

In this equation [L] is the radioligand concentration and K_D is the equilibrium dissociation constant of the radioligand. When ∼0.1 nM [³H]-PD140,376 was used as label (pK_D=9.89±0.14, n=4±s.e.mean), the correction factor was ∼2. In practice, it was not necessary to correct for the radioligand occupancy when ∼20 pM [¹²⁵I]-BH-CCK-8S was used as the affinity of CCK-8S at sites in rat and mouse cortex and guinea-pig pancreas was >0.2 nM (i.e. [L]/K_D <0.1).

Competition data were fitted to the following Hill equation, which describes the relationship between the amount of bound ligand (B) and free ligand concentration ([L]), using a derivative-free, non-linear, regression programme (BMDP Statistical Software, Module AR; Dixon, 1992).

In the equation, R, the total number of specific binding sites occupied by the radiolabel in the absence of the competing ligand, was fixed in the fitting procedure at a value of 100% because the data were expressed as the percentage of the decrease in specific binding. n_H is the midpoint slope parameter and IC₅₀ is the midpoint location parameter which was estimated as log₁₀ IC₅₀ on the basis that IC₅₀ values are log-normally distributed (Harper et al., 1996a).

Two-site model

Data sets were also fitted using the BMDP computer programme to the following two-site model,

This equation describes the binding of a ligand to two independent sites (R₁ and R₂), governed by the midpoint location parameters IC₅₀₁ and IC₅₀₂, respectively. In practice, the data were expressed as per cent specific binding (i.e. R₁+R₂=100%) so that R₂ could be substituted by the term (100–R₁). Again, the IC₅₀ values were estimated as base 10 logarithms on the basis that they are expected to be log-normally distributed.

Statistical comparison of model parameter estimates

All data are presented as mean±s.e.mean. Hill equation slope values were tested for differences from unity by Student's t-test as were the differences in individual compound pIC₅₀ values obtained on the mouse and rat cortex assays. P values of <0.05 were considered significant. The goodness-of-fit of the one-site and two-site models were assessed by comparison of the residual variance of the fits to the data using the `extra sum of squares' principle (for details see De Lean et al., 1980).

Materials

¹²⁵I-Bolton Hunter labelled CCK-8S ([¹²⁵I]-BH-CCK-8S) with specific activity of ∼2200 Ci mmol⁻¹ was obtained from NEN™ Life Science Products, Hounslow, U.K. [³H]-PD140,376 ([ [L-3-[(4-aminophenyl)methyl]-N-[α- methyl - N -[(tricyclo[3.3.1.1.^3,7]dec-2-yloxy)carbonyl]-D-tryptophyl]-β-alanine]) with a specific activity of ∼50 Ci mmol⁻¹ was obtained from Amersham International Plc, Little Chalfont, Buckinghamshire, U.K. L-365,260 (3R-(+)-N-(2,3-dihydro-1-methyl - 2 - oxo - 5 - phenyl - 1H - 1,4-benzodiazepin-3-yl)-N′- 3 -methylphenyl urea), PD134,308 (CI988) ([R-(R*,R*)]-4-[[2-[[3-(1H -indol-3-yl)-2- methyl -1-oxo -2- [[(tricyclo[3.3.1.1.^3,7]dec- 2-yloxy)carbonyl] amino] propyl]amino]-1-phenylethyl]amino]-4-oxobutanoic acid), PD140,376 (([[L-3-[(4-aminophenyl)methyl]-N-[α-methyl-N-[(tricyclo[3.3.1.1.^3,7]dec-2-yloxy)carbonyl]-D-tryptophyl]-β-alanine]), SR27897 (1-[[2-(4-(2-chloro-phenyl)thiazol-2-yl)aminocarbonyl]indolyl]acetic acid) and YM022 ((R) -1- [2,3- dihydro- 1-( 2′-methylphenacyl) - 2-oxo -5-phenyl - 1H - 1,4 - benzodiazepin -3-yl]-3-(3-methylphenyl)urea) were synthesized by James Black Foundation chemists.

HEPES (N-[2-hydroxyethyl]piperazine-N′-[2-ethanesulphonic acid), EGTA (ethyleneglycol-bis(β-aminoethylether) N,N,N′,N′-tetraacetic acid), bacitracin and Trizma base® were obtained from the Sigma Chemical Co., Poole, Dorset, U.K. All other materials were obtained from Fisons Scientific Apparatus Loughborough, Leics., U.K.

All compounds were dissolved in DMF to give stock concentrations of 10 mM and further dilutions were made in HEPES-NaOH buffer.

Results

Analysis of competition curves in guinea-pig pancreas

The affinity of compounds at CCK_A binding sites was estimated in the guinea-pig pancreas. This was done to exclude the possibility that any complex data obtained in the cortex CCK_B/gastrin receptor assays was due to [¹²⁵I]-BH-CCK-8S also labelling CCK_A binding sites under our assay conditions.

L-365,260, YM022, SR27897, PD134,308, PD140,376 and JB93182 produced concentration-dependent inhibition of the specific binding of [¹²⁵I]-BH-CCK-8S to CCK_A binding sites in guinea-pig pancreas (Table 1). The mean mid-point slope parameter estimates (n_H) were not significantly different from unity. Of all the compounds, JB93182 had the lowest, sub-micromolar affinity (pK_I=5.29±0.12; n=5) for CCK_A receptors.

Analysis of competition curves in mouse cortex

Our previous analysis of the variation in L-365,260 competition curves indicated that the mouse cortex assay expressed a homogenous population of CCK_B/gastrin receptors. Therefore, our expectation was that the mean mid-point slope parameter estimates of competition curves for all ligands should not be significantly different from unity. In addition, we did not expect significant variation in the location of the competition curves for each ligand between experiments.

The competition curves for JB93182, L-365,260, YM022, PD134,308, PD140,376 and SR27897 all had mean mid-point slope parameters (n_H) which were not significantly different from unity (Table 2). In addition, there was no significant difference between the behaviour of any of the ligands, in terms of mid-point slope parameter and estimated pK_I values (r=0.99, P<0.002), at sites in mouse cortex labelled with either [³H]-PD140,376 or [¹²⁵I]-BH-CCK-8S (Figure 1; Table 2). For instance, the pK_I estimates for JB93182 in the mouse cortex assay, when competing with [¹²⁵I]-BH-CCK-8S and [³H]-PD140,376 for CCK_B/gastrin binding sites, were 8.74±0.15 (n=4) and 8.88±0.10 (n=3), respectively.

Table 2.

Analysis of multiple data sets from competition experiments between [¹²⁵I]-BH-CCK-8S or [³H]-PD140, 376 and CCK/gastrin receptor ligands in mouse cortex (see text for details)

Figure 1.

Competition between [¹²⁵I]-BH-CCK-8S and increasing concentrations of ligands for CCK_B/gastrin binding sites in mouse cortex. Data represent the mean±s.e.mean of 5–47 experiments (see Table 2) where each point was determined in triplicate. The curves shown superimposed on the mean experimental data points were obtained by simulation using equation (2) where the parameters were set at the mean values estimated by fitting each replicate curve to that equation. The parameters used in the simulations are presented in Table 2.

Analysis of competition curves in rat cortex

Our previous analysis of the behaviour of L-365,260 suggested that the rat cortex expressed a variable proportion of two CCK_B/gastrin sites. However, in the first instance, without prejudice to this previous finding, the data were analysed by fitting the Hill equation (2).

Hill equation fitting

YM022, SR27897, PD134,308, PD140,376 and JB93182 produced concentration-dependent inhibition of the specific binding of [¹²⁵I]-BH-CCK-8S to CCK_B/gastrin binding sites in rat cortex (Figure 2). The mean mid-point slope parameter estimates from competition curves for JB93182 (0.82), PD134,308 (0.83) and PD140,376 (0.86) were not significantly different from unity. The mean mid-point slope parameter estimate for L-365,260 (0.85), obtained previously with the exceptionally large data set (n=48), was significantly different from unity (Table 3). The apparent affinity (pIC₅₀) values of PD140,376 and SR27897 for CCK_B/gastrin binding sites in the rat cortex were not significantly different from those estimated in the mouse cortex. In contrast, the mean apparent affinities expressed by L-365,260 and PD134,308 were both significantly lower, while those expressed by JB93182 and YM022 were significantly higher, than those obtained in the mouse cortex.

Figure 2.

Competition between [¹²⁵I]-BH-CCK-8S and increasing concentrations of ligands for CCK_B/gastrin binding sites in rat cortex. Data represent the mean±s.e.mean of 5–48 experiments (see Table 3) where each point was determined in triplicate. The curves shown superimposed on the mean experimental data points were obtained by simulation using equation (2) where the parameters were set at the mean values estimated by fitting each replicate curve to that equation. The parameters used in the simulation are presented in Table 3.

Table 3.

Analysis of multiple data sets for competition experiments between [¹²⁵I]-BH-CCK-8S and CCK/gastrin receptor ligands of CCK_B/gastrin binding sites in rat cortex (see text for details)

Two site model fitting

In view of the previous evidence for the expression of a variable proportion of two CCK_B/gastrin binding sites in the rat cortex, a two-site model (equation 3) was also applied to the data. First, an attempt was made to fit the individual competition curves obtained from replicate experiments for each compound. Given the variation in the proportion of receptors, evident from both the previous analysis of the L-365,260 data and by inspection of the individual curves obtained with the compounds in this study (see Figure 3, for example), the receptor affinity and the receptor ratio was allowed to vary for each replicate curve in the fit.

Figure 3.

Representative PD140,376 competition curves obtained in the rat cortex assay using [¹²⁵I]-BH-CCK-8S as the radioligand. The data shown are examples of individual curves obtained in separate experiments (n=1 in triplicate). The lines shown superimposed on the individual data sets were obtained using the R₁, R₂, pIC₅₀₁ and pIC₅₀₂ parameters obtained from the simultaneous two-site model fitting of the individual curves (equation (3) and see Table 4). In the fit, a global pIC₅₀₁ (8.40±0.16) and pIC₅₀₂ (9.44±0.09) were estimated but the percentage of R₁ and R₂ were allowed to vary for each data set.

A satisfactory fit could not be obtained for all the curves. Thus, as observed previously when the L-365,260 data set was analysed, there was not always sufficient information (i.e. biphasicity) within each curve to allow estimation of each of the parameters. Therefore, a second model fit was performed in which all of the replicate curves for each compound were fitted to the model simultaneously. In the model fit the proportion of R₁ to R₂ was allowed to vary between each replicate curve although single values for IC₅₀₁ and IC₅₀₂ were estimated for each compound. The model fits (equation 3) all converged and estimates of the pIC₅₀ of each compound at the gastrin G₁ (pIC₅₀₁) and gastrin G₂ (pIC₅₀₂) sites were obtained (Table 4). The fit of the data to the two-site model was significantly better than to the Hill equation for each ligand (JB93182, F_(53,46)=10.66; PD134,308, F_(59,52)=4.40; PD140, 376, F_(54,48)=8.18; YM022, F_(86,76)=2.27; SR27897, F_(44,39)=11.46). The compounds all appeared to express some degree of site selectivity (Table 4; JB93182 ∼23 fold, PD134,308 ∼13 fold, PD140,376 ∼11 fold, YM022 ∼4 fold and SR27897 ∼9 fold).

Table 4.

Apparent affinities of CCK/gastrin receptor ligands at gastrin-G₁ and gastrin-G₂ sites in rat cortex (see text for details) using [¹²⁵I]-BH-CCK-8S as radiolabel

The mouse cortex (G₂) pIC₅₀ values were indistinguishable from the corresponding pIC₅₀₂ G₂ values estimated from the two-site model fit of the rat cortex data (F_(6,12)=3.31; Figure 4A) whereas they were significantly different from the rat G₁ values (F_(6,12)=30.91; Figure 4B).

Figure 4.

Comparison of affinity estimates obtained in mouse and rat cortex assays. The line shown superimposed on the data is the line of identity. (A) Comparison of the G₂ pIC₅₀ values obtained from the mouse cortex data ([¹²⁵I]-BH-CCK-8S as radioligand, Table 2) and the G₂ affinity estimate obtained from using the two-site model (equation 3) to analyse the individual replicate curve data obtained in the rat cortex (Table 4). The mouse cortex (G₂) pIC₅₀ values were indistinguishable from the G₂ pIC₅₀ values estimated from the two-site model fit of the rat cortex data (F_(6,12)=3.31). (B) Comparison of the G₂ pIC₅₀ values obtained from the mouse cortex data ([¹²⁵I]-BH-CCK-8S as label, Table 2) and the G₁ estimates obtained using the two-site model (equation 3) to analyse the individual replicate curve data from the rat cortex (Table 4). The mouse G₂ pIC₅₀ values were significantly different from the rat G₁ pIC₅₀ values (F_(6,12)=30.91).

Discussion

In this study, in an attempt to expose CCK_B/gastrin receptor heterogeneity, we investigated the behaviour of a chemically-diverse series of CCK_B/gastrin receptor antagonists in the mouse and rat cortex. The low affinities of PD134,308 (6.08), PD140,376 (6.21), L-365,260 (6.50) and JB93182 (5.29) and the high affinity of SR27897 (9.62) in the guinea-pig pancreas assay suggested that any differences between the rat and mouse cortex assays were not due to [¹²⁵I]-BH-CCK-8S labelling differential proportions of CCK_A binding sites in rat and mouse cortex assays. The high Hill slope value associated with an unusually high variance (1.55±0.21; n=5) obtained with JB93182 in the CCK_A assay could be due to the high concentrations of the compound which were required to define the competition curve inhibiting the non-specific binding of the radioligand.

In the mouse cortex, the L-365-260 and PD134,308 data were consistent with those previously reported in this tissue when either [¹²⁵I]-BH-CCK-8S (Hughes et al., 1990) or [³H]-pBC264 (Durieux et al., 1992; Bertrand et al., 1994) was used as the radioligand. In addition, the data obtained for all the competing compounds were consistent with our previous observation that this tissue appears to contain a homogeneous population of CCK_B/gastrin binding sites. These sites have a high affinity for L-365,260 (pK_I=8.41±0.01, n=47; Harper et al., 1996a) and were provisionally termed gastrin G₂. In this study, the observation that there was a good correlation between the estimated affinities of the compounds regardless of whether an antagonist ([³H]-PD140,376) or agonist ([¹²⁵I]-BH-CCK-8S) was used to label the sites indicated that any complexity in this, and the previous study, was not a consequence of using an agonist as the radioligand.

In the rat cortex assay, PD134,308, YM022 and L-365,260 had estimated affinity values for CCK_B/gastrin binding sites that were similar to those reported previously in rat cortex. For instance, affinity values of 8.20, 10.17 and 7.72 have been reported for PD134,308, YM022 and L-365,260, respectively (Nishida et al., 1994). However, in contrast to our previous study (Harper et al., 1996a), none of these data, at first sight, could be considered to provide compelling evidence for receptor heterogeneity. Thus, the mean Hill slope parameter estimates obtained from the competition curves for PD134,308, PD140,376, YM022, JB93182 and SR27897 were not significantly different from unity (Table 3). However, JB93182 and YM022 expressed significantly higher affinities and PD134,308 expressed significantly lower affinity in the rat cortex assay to that expressed in the mouse cortex assay.

The tissue-dependent affinity of JB93182, YM022 and PD134,308 could be due to these compounds detecting species differences in the amino-acid sequence of the receptor. The species dependence of antagonist affinity has previously been reported for other CCK_B/gastrin receptor ligands such as L-365,260 (Lotti & Chang, 1989; Beinborn et al., 1993) and LY247348 (Hunter et al., 1993). However, in view of data obtained for JB93182 and PD134,308 in rat and mouse isolated, lumen-perfused, stomach assays (Roberts et al., 1996b; Shankley et al., 1997), it seems unlikely that speciation of CCK_B/gastrin receptors could account for the differences in antagonist affinity detected in this study. For instance, the behaviour of JB93182 in the isolated, lumen-perfused rat stomach bioassay (pK_B=9.90±0.18; Shankley et al., 1997), in agreement with our previous conclusion (Roberts et al., 1996a), could be accounted for by the presence of a homogeneous population of CCK_B/gastrin receptors (G₁). By contrast, in the mouse stomach, a tissue which was suggested to contain a variable population of gastrin-G₁ and gastrin-G₂ receptors (Roberts et al., 1996a), the slope of the Schild plot for JB93182 was significantly different from unity (b=0.60±0.09). Moreover, these data could be described by a two-receptor model with pK_B values of ∼9.1 (gastrin-G₂) and ∼9.9 (gastrin-G₁). Therefore, an alternative explanation for the species differences in the apparent affinity of JB93182, PD134,308 and YM022 is that provided by our original analysis of the complexity in L-365,260 data obtained in rat and mouse cortex assays (Harper et al., 1996a) i.e. that the mouse cortex assay expresses one CCK_B/gastrin binding site whereas the rat cortex expresses two sites.

In the current study, the finding of Hill slope estimates in the rat cortex which were not significantly different from unity could simply mean that we had insufficient data to expose the two sites. Hill slope parameter estimates from a single-site model fit of a system expressing two sites are dependent on both the affinity of the compound at the two sites and the numbers of each site labelled. Indeed, in the previous study we were only able to obtain compelling evidence that the mean Hill slope of L-365,260 competition curves was significantly different from unity when we analysed an exceptionally large data set (Harper et al., 1996a). The difficulty in detecting the two sites appeared to result from the relatively low selectivity of L-365,260 (∼18 fold) and the between-experiment variability in the ratio of the two sites was an added complicating factor. The large data set allowed the definition of the relationship between the Hill slope value (n_H) and the location (pIC₅₀) of individual competition curves expected for a system with variable expression of two sites (Harper et al., 1996a). Because of this complexity, we only expected to obtain clear evidence for receptor heterogeneity (i.e. Hill slope parameters significantly different from unity) if either the compounds investigated in this study expressed greater receptor selectivity than L-365,260 or if the selectivity of any of the compounds was equivalent to that of L-365,260 and we performed what might be considered to be an unrealistic number of replicate experiments. Therefore, on the basis of the prior knowledge gained from the studies using L-365,260, we considered it justifiable to analyse the data obtained in this study along lines consistent with our original hypothesis, namely, the presence of two CCK_B/gastrin binding sites (G₁ and G₂) in the rat cortex.

A satisfactory fit was obtained when the proportion of G₁ and G₂ sites was allowed to vary for each replicate competition experiment. This indicated that all of the compounds expressed some degree of selectivity for the two sites and that the proportion of the sites varied between experiments as was found in the original analysis. The finding that the mouse cortex (G₂) pIC₅₀ values were different from the rat cortex G₁ pIC₅₀ values (Figure 4B) but indistinguishable from the rat cortex G₂ values (Figure 4A) was consistent with our original conclusion that one of the sites is common to both the mouse and rat cortex. Moreover, the new data was also consistent with the previous conclusion that the G₁ and G₂ sites characterized in the binding studies were equivalent to the sites characterized in the rat (G₁) and the mouse (G₁/G₂) isolated, lumen-perfused, stomach assays. Thus, the affinity estimates for JB93182 and PD134,308 at the rat cortex G₂ and G₁ sites (Table 4) were similar to those estimated in the functional assays (G₂-receptor: JB93182 ∼9.1, PD134,308 ∼9.2; G₁-receptor: JB93182 ∼9.9; PD134,308 ∼7.6; Roberts et al., 1996b; Shankley et al., 1997).

A reasonable distribution in the variation of the proportion of R₁ and R₂ sites was present in the replicate experiments for each compound (Table 4) as the two-site model fits converged. However, the fit indicated that within the PD134,308 data set the R₁ site was predominant in four of the five replicate experiments. Even in the fifth replicate there were still 32% of the R₁ sites present. If the two-site model interpretation is valid, then reliable estimates for pIC₅₀₁ and pIC₅₀₂ are expected only when the data set for each compound includes individual competition curves that lie at either end of the distribution of the pIC₅₀ values (i.e. under the conditions where ∼100% of either site are present). Thus, for PD134,308 the G₂ site pIC₅₀ estimate is based on relatively less information and, accordingly, the value was the furthest from that predicted by the mouse cortex assay.

The data obtained in this study are consistent with the hypothesis that there are two, pharmacologically-distinguishable, CCK_B/gastrin binding sites in rat cortex (gastrin-G₁ and G₂). To provide further evidence for the presence of two sites, it is clear that more highly selective ligands are required. However, in the absence of such ligands, one suitable approach for subtype discrimination would be to attempt to label and characterize one of the receptor populations with a receptor-selective radioligand. The current analysis suggests that JB93182 expresses an approximate 23 fold selectivity for the gastrin-G₁ site. Therefore, if our interpretation of the rat cortex data is correct it should be possible to label only gastrin-G₁ sites in this tissue with a radiolabelled form of JB93182. Therefore, the affinity values obtained for compounds competing with this radioligand should be similar to the gastrin-G₁ affinity estimates obtained in this study. Characterization of the binding of [³H]-JB93182 forms the subject of the accompanying report (Harper et al., 1999).

Abbreviations

CCK-8S

cholecystokinin sulphated octapeptide

BH-CCK8S

Bolton-Hunter CCK8S

HEPES

(N-[2-hydroxyethyl]piperazine-N′-[2-ethanesulphonic acid])

EGTA

(ethyleneglycol-bis(β-aminoethylether)N,N,N′,N′-tetraacetic acid)

DMF

dimethylformamide