Effects of (2R)-(3-amino-2-fluoropropyl)sulphinic acid (AFPSiA) on transient lower oesophageal sphincter relaxation in dogs and mechanism of hypothermic effects in mice

Abstract

1. The effects of the novel GABA analogue (2R)-(3-amino-2-fluoropropyl)sulphinic acid (AFPSiA) on transient lower oesophageal sphincter relaxations (TLOSRs) were studied in the dog. In addition, the GABA_A/GABA_B selectivity was determined in vitro and in vivo, and the pharmacokinetics and the metabolism of the compound were studied in the dog and rat.

2. TLOSRs were reduced by 55±8% after intragastric administration of AFPSiA at 14 μmol kg⁻¹ and did not decrease further at higher doses. When evaluated 2 and 4 h after administration, the effect declined to 37±6 and 16±9%, respectively. Spontaneous swallowing was only significantly inhibited at 100 μmol kg⁻¹.

3. The oral availability of AFPSiA was 52±17 and 71±4% in the dog and rat, respectively. A fraction of AFPSiA was oxidised to the corresponding sulphonate, (2R)-(3-amino-2-fluoropropyl)sulphonic acid (AFPSoA) after oral administration to the rat and dog.

4. In rat brain membranes, AFPSiA was found to have ten times higher affinity for rat brain GABA_B (K_i=47±4.4 nM) compared to GABA_A (K_i=430±46 nM) binding sites. The compound was a full agonist at human recombinant GABA_B(1a,2) receptors (EC₅₀=130±10 nM). In contrast, the metabolite AFPSoA was considerably more selective for binding to rat brain GABA_A (K_i=37±3.1 nM) vs GABA_B (K_i=6800±280 nM) receptors.

5. In the mouse, high doses (1–8 mmol kg⁻¹) of AFPSiA induced a rapid and mild hypothermia followed by a profound and sustained hypothermia at the higher doses tested (6 and 8 mmol kg⁻¹). This effect was unaffected by the selective GABA_B receptor antagonist CGP62349. AFPSoA (1 and 2 mmol kg⁻¹) produced transient and moderate hypothermia while the hypothermic response was considerably larger at 4 mmol kg⁻¹.

6. It is concluded that AFPSiA inhibits but does not abolish TLOSRs in the dog. High doses of the compound induce hypothermia in the mouse, which probably is attributable to activation of the GABA_A receptor. The latter effect may be caused both by AFPSiA and its oxidised sulphonic acid metabolite AFPSoA.

Introduction

Gastro-oesophageal reflux disease (GORD) is a highly prevalent condition and a major burden to the society as well as the afflicted individual. GORD is nowadays in most cases successfully treated with proton pump inhibitors, but a number of patients do not receive full symptomatic allieviation (Klinkenberg-Knol & Meuwissen 1990; Vakil, 2004). In many of these patients, reflux of bile and other nonacid contents from the stomach is believed to generate symptoms (Koek et al., 2003). Since transient lower oesophageal sphincter relaxation (TLOSR) is the dominant cause of reflux, interventions aimed at inhibiting TLOSRs may have a therapeutic role to play in GORD (Mittal et al., 1995; Hirsch et al., 2002), particularly when acid secretion suppression alone is not sufficiently successful.

The pharmacology of TLOSR has evolved rapidly during the past few years. Among compounds inhibiting TLOSRs, GABA_B receptor agonists are the most promising. The skeletal muscle relaxant baclofen, the only GABA_B agonist approved for clinical use, reduces TLOSRs in dogs (Lehmann et al., 1999), ferrets (Blackshaw et al., 1999), cats (Liu et al., 2002) and humans (Lidums et al., 2000; Zhang et al., 2002). Importantly, data have emerged to show that baclofen also provides reduction in reflux of both acid and nonacid gastric contents and symptomatic relief in GORD patients (Cange et al., 2002; Ciccaglione & Marzio, 2003; Koek et al., 2003; Vela et al., 2003).

While being the preferred pharmacological option in the treatment of spasticity, baclofen may be less useful in GORD due to its central side effects. The need for novel GABA_B agonists with an improved safety and tolerability profile is therefore obvious. GABA analogues in which a sulphinic acid group replaces the carboxylic acid group of GABA have not previously been evaluated to any great extent. The sulphinic acid analogue of baclofen (‘siclofen', Carruthers et al., 1995) and some other 3-aminopropyl sulphinic acids with a large lipophilic group in the 2-position have properties very similar to baclofen with central side effects appearing at doses close to those inhibiting TLOSRs (unpublished own findings). We have synthesised 3-aminopropyl sulphinates with small substituents in the 2-position (for instance hydroxy and keto groups). Based on the potency of the compounds to activate recombinant human GABA_B receptors, we selected the most potent for further in vivo studies. This compound, (2R)-(3-amino-2-fluoropropyl)sulphinic acid (AFPSiA; Figure 1), was evaluated with respect to its effect on TLOSRs in the dog and GABA_A/GABA_B selectivity in vitro. Further, the pharmacokinetics of AFPSiA were determined in dogs and rats. Moreover, the side effect profile of AFPSiA was assessed in vivo in mice, and the relation between side effects and formation of a major metabolite [(2R)-(3-amino-2-fluoropropyl)sulphonic acid, AFPSoA] of the compound was investigated.

Figure 1.

Structure of AFPSiA and AFPSoA.

Methods

Synthesis of AFPSiA and AFPSoA

AFPSiA was synthesised from the tert-butyl [(2R)-2-fluoro-3-hydroxypropyl]carbamate. This hydroxy intermediate was transformed into mesylate by the reaction with methanesulphonyl chloride and this mesylate was then reacted with tert-butyl thiol to form the corresponding sulphide. Oxidation of the sulphide by 3-chloroperoxybenzoic acid gave the sulphone, which was reacted with trifluoromethanesulphonic acid to yield AFPSiA by the simultaneous deprotection of the amino group and the unmasking of the sulphinic acid moiety.

AFPSoA was synthesised from the iodide intermediate, tert-butyl [(2R)-2-fluoro-3-iodopropyl]carbamate. The iodide was transformed into thioacetate through the reaction with potassium thioacetate, and the subsequent reaction with trifluoroacetic acid and hydrogen peroxide then gave AFPSoA.

Determination of binding affinities of AFPSiA and AFPSoA at GABA_A and GABA_B receptor sites in rat brain membranes

Preparation of rat brain synaptic vesicles was performed as described (Jensen et al., 2002). Inhibition of [³H]GABA binding at GABA_B receptor sites in rat brain synaptic membranes by test compounds was measured using a filtration binding assay as described (Jensen et al., 2002). Displacement curves to determine IC₅₀ values were constructed by fitting the 4-parameter logistic equation to the data using XLfit for Microsoft Excel. K_D for GABA was determined on each preparation by homologous competition and used to calculate K_i values from IC₅₀ determinations on that particular preparation using the Cheng–Prusoff equation (Cheng & Prusoff, 1973). The average K_D for GABA was 110±21 nM (mean±s.e.m., n=16 preparations).

The affinity of test compounds at GABA_A receptor sites in rat brain membranes was measured using a [methylamine-³H]muscimol radioligand binding assay (Martini et al., 1983). This assay was performed as described above for the [³H]GABA assay, with the exception that 10 nM [³H]muscimol (274 GBq mmol⁻¹, Amersham Biosciences, Uppsala, Sweden) was used instead of GABA and that isoguvacin was excluded from the incubation buffer. The binding affinity (K_D) for muscimol at GABA_A receptor sites in the rat brain synaptic membrane preparation was determined by saturation binding analysis to 14±1.4 nM (mean±s.e.m., n=3) and used to calculate K_i values from IC₅₀ determinations of test compounds using the Cheng–Prusoff equation.

Determination of effects of GABA_B receptor agonists on GABA_B-receptor-mediated intracellular calcium release

The effect of compounds on GABA_B-receptor-mediated intracellular calcium release was determined in CHO-K1 cells stably coexpressing a GABA_B(1a)–G_αqi5 fusion protein and GABA_B(2) using a fluorescence imaging plate reader (FLIPR, Molecular Devices). GABA_B(1a) and GABA_B(2) were cloned from human brain cDNA and subcloned into pCI-Neo (Promega, Madison, WI, U.S.A.) and pcDNA3.1(−)/Zeo (Invitrogen, Basel, Switzerland), respectively. A GABA_B(1a)-G_αqi5 fusion protein expression vector was constructed using the pCI-Neo-GABA_B(1a) cDNA plasmid and pLEC1-G_αqi5 (Molecular Devices, Menlo Park, CA, U.S.A.). In order to make the G_αqi5 pertussis toxin insensitive, Cys356 was mutated to Gly using standard PCR methodology with primers 5′-GGATCCATGGCATGCTGCCTGAGCGA-3′ (forward) and 5′-GCGGCCGCTCAGAAGAGGCCGCCGTCCTT-3′ (reverse). The G_αqi5mut cDNA was ligated into the BamHI and NotI sites of pcDNA3.0 (Invitrogen). The GABA_B(1a) coding sequence was amplified by PCR from pCI-Neo- GABA_B(1a) using primers, 5′-GGATCCCCGGGGAGCCGGGCCC-3′ (forward) and 5′-GGATCCCTTATAAAGCAAATGCACTCGA-3′ (reverse) and subcloned into the BamHI site of pcDNA3.0-G_αqi5mut.

For generation of a stable cell line expressing the GABA_B receptor heterodimer, GABA_B(1a)−G_αqi5mut plasmid DNA (8 μg), GABA_B(2) plasmid DNA (8 μg) and 24 μl lipofectamine (Life Technologies, Paisley, U.K.) were mixed in 5 ml OptiMEM (Life Technologies), and incubated for 45 min at room temperature. CHO-K1 cells were grown in culture medium containing Nut mix F-12 (Ham) with Glutamax-1 (Life Technologies) supplemented with 10% fetal bovine calf serum (Life Technologies) 100 U ml⁻¹ penicillin (Life Technologies) and 100 μg ml⁻¹ streptomycin (Life Technologies) at 37°C in a humidified CO₂-incubator. The cells were detached with 1 mM ethylenediaminetetraacetic acid (EDTA; Sigma-Aldrich, St Louis, MO, U.S.A.) in phosphate-buffered saline (PBS) without calcium and magnesium (Life Technologies) and 1 million cells were seeded in 100 mm Petri dishes. After 24 h the culture medium was replaced with OptiMEM and incubated for 1 h in a CO₂ incubator. The cells were exposed to the transfection medium for 5 h, which was then replaced with culture medium. After 48 h, the cells were detached and seeded in six-well plates 2000 cells well⁻¹, and grown in culture medium supplemented with geneticin (400 μg ml⁻¹) and zeocin (250 μg ml⁻¹). After 4 days, cells from single colonies were collected and transferred to a 24-well plate. After 10 days, the cell clones were seeded in T-25 flasks and grown for another 16 days before they were tested for GABA_B-receptor-mediated functional response in the FLIPR assay described below. The clones that showed the highest peak response were collected and subcloned by seeding the cells in six-well plates 1000 cells well⁻¹ and repeating the steps described above. The clonal cell line that gave the highest peak response in the FLIPR was used in the present study.

Measurement of GABA_B receptor dependent release of intracellular calcium in the FLIPR was performed as described by Coward et al. (1999), with some modifications. CHO cells transfected with GABA_B(1a)–G_αqi5mut and GABA_B(2) were cultivated in culture medium with the addition of 250 μg ml⁻¹ zeocin and 400 μg ml⁻¹ geneticin. At 24 h prior to the experiment, the cells (35,000 cells well⁻¹) were seeded in black-walled poly-D-lysine-coated 96-well plates (Becton Dickinson, Bedford, U.K.) in culture medium without antibiotics. The cell culture medium was aspirated and 100 μl of Fluo-3 (Molecular Probes, Eugene, OR, U.S.A.) loading solution (4 μM Fluo-3, 2.5 mM probenecid [Sigma-Aldrich] and 20 mM HEPES (Life Technologies) in Nut Mix F-12 Ham) was added. After incubation for 1 h at 37°C in a 5% CO₂ incubator, the dye solution was aspirated and the cells were washed two times with 150 μl of wash solution (2.5 mM probenecid and 20 mM HEPES in Hank's buffered salt solution (HBSS, Life Technologies) followed by addition of 150 μl of wash solution. The cells were then assayed in the FLIPR. The test compounds were diluted in HBSS containing 20 mM HEPES and added to a volume of 50 μl wash solution (final volume in plate well=200 μl). The fluorescence was sampled every second for 60 s (10 s before and 50 s after the addition of test compound) and then every sixth second for 120 s. The peak response values from 2 to 4 determinations at each concentration of test compound and experiment were calculated, after subtraction of basal fluorescence, as the percentage of the maximum response obtained with GABA run in the same experiment and used to construct concentration–response curves for EC₅₀ determinations by fitting the 4-parameter logistic equation to the data using XLfit for Microsoft Excel. Intrinsic activities were calculated by dividing the peak response obtained by test compounds with the peak response obtained by GABA.

Effects on TLOSRs and spontaneous swallowing in the dog

The method used has been described previously (Stakeberg & Lehmann, 1999). In brief, adult Labrador retrievers of both sexes were equipped with an oesophagostomy. After recovery, they were intubated with a water-perfused multilumen Dentsleeve assembly to record pressure of the oesophagus, lower oesophageal sphincter and stomach. An antimony pH catheter was placed alongside the manometric assembly to measure reflux episodes. A thin air-perfused catheter was placed retrogradely in the hypopharynx to measure swallows. Only pharyngeal contractions followed by a peristaltic wave were included in the analysis. TLOSRs were stimulated by infusion of an acidified nutritious soup followed by insufflation of air. All dogs had been used for other studies previously, but there was always a wash-out period of at least 3 days between experiments in which drugs had been administered. The data were related to the average of the five preceding control experiments in each dog, and every fourth experiment was designated as a control. The database generated over a large number of years shows that there is no consistent change with time with regard to TLOSR incidence. AFPSiA (dissolved in 0.9% NaCl) was given intragastrically through the assembly at 2 ml kg⁻¹ 30 min before infusion of soup. In effect duration studies, AFPSiA was administered at 14 μmol kg⁻¹ 2 and 4 h before infusion of soup.

AFPSoA was administered i.v. in two dogs at 13 μmol kg⁻¹. However, the compound proved to be emetogenic and no further experiments were performed.

Effects on behaviour and body temperature in the mouse

Measurement of body temperature was done as described by Quéva et al. (2003). In summary, temperature-sensitive chips were implanted subcutaneously in C57/Bl mice. Drugs were injected subcutaneously and temperature was measured at regular intervals with a transponder so as to produce minimal disturbance to the animals. The centrally active GABA_B receptor antagonist CGP62349 (Froestl et al., 1996) was not a pure diastereomer, but a 50/50 mixture of CGP62349 and its epimer 3-[(1S)-1-[[(2S)-2-hydroxy-3-[hydroxy[(4-methoxyphenyl)methyl]phosphinyl]propyl]amino]ethyl]-benzoic acid, but for the sake of simplicity, this mixture was referred to as CGP62349. This compound was synthesised in-house.

Pharmacokinetics

For analysis of plasma concentrations of AFPSiA in dogs, blood was drawn from a foreleg vein after intravenous administration (7 μmol kg⁻¹) in separate experiments and after intragastric administration (7 and 14 μmol kg⁻¹) in the manometric experiments. Female Sprague–Dawley rats were dosed orally or intravenously at 7 μmol kg⁻¹ and blood samples were drawn from an arterial catheter.

The blood samples for analysis of both AFPSoA and AFPSiA were taken from dogs as described above (dose 56 μmol kg⁻¹, intragastrically) and from the orbital plexus of female Wistar rats after adminstration of AFPSiA (350 μmol kg⁻¹) by gavage.

For collection of urine, female Sprague–Dawley rats (n=2) were placed in metabolism cages immediately after intravenous administration at 14 μmol kg⁻¹ and urine was collected quantitatively for 48 h.

AFPSiA, AFPSoA and an internal standard were isolated from plasma by precipitation of the plasma proteins, separated by liquid chromatography on a BioBasic AX column and measured by tandem mass spectrometry. Determination of AFPSiA in urine was performed after dilution of the urine with the mobile phase. The methods were similar to those reported by Fakt et al. (2003).

The pharmacokinetic parameters of AFPSiA were calculated by non-compartment analysis using WinNonlin Professional (Pharsight Corporation, CA, U.S.A.). The area under the plasma concentration–time curve (AUC_(0−t)) was calculated by a combination of the linear and logarithmic trapezoidal rule from the time of administration to the last sampling time or the sampling time with the last detectable plasma concentration. AUC_(0−t) was extrapolated to infinity (AUC) by adding the residual area Ct/k, where Ct is the predicted plasma concentration at time t and k is the elimination rate constant. Ct and k were obtained by linear least-square regression analysis of the logarithm of the last 3–5 plasma concentrations versus time. The plasma clearance (CL) and volume of distribution at steady state (V_ss) were calculated after intravenous administration. The half-life in plasma was calculated as

After oral administration, the oral bioavailability (F) of AFPSiA was calculated as

For dogs, the indiviual bioavailability was calculated using individual values for AUC_iv, whereas a mean value was used for rats.

All animal experiments were approved by the local Ethical Committee for Animal Experiments.

Statistics

Student's t-test was used to identify statistically significant differences between placebo- and drug-treated groups. In the hypothermia experiments, only nadir temperature was compared between placebo and drug groups.

Results

Binding affinities and agonistic properties of AFPSiA and AFPSoA at native and recombinant GABA_B receptors

The binding affinities (K_i) and agonistic effects (EC₅₀ and intrinsic activity) of the compounds are summarised in Table 1. In summary, AFPSiA was found to have a nine-fold selectivity for GABA_B compared with GABA_A receptors while AFPSoA was almost 200 times more selective for the latter. AFPSiA, but not AFPSoA, was a full agonist at GABA_B receptors.

Table 1.

Binding affinities and agonistic properties of AFPSiA and AFPSoA at GABA_B and GABA_A receptors

Compound	GABAB affinity Ki (nM)	GABAA affinity Ki (nM)	GABAB agonism EC50 (nM)	GABAB intrinsic activity
AFPSiA	47±4.4 (10)	430±46 (3)	130±10 (6)	1.1±0.04 (6)
AFPSoA	6800±280 (7)	37±3.1 (3)	28,000±1300 (7)	0.6±0.08 (3)

Binding affinities (K_i) and agonistic properties (EC₅₀ and intrinsic activity) of AFPSiA and AFPSoA at the GABA_B and GABA_A receptors were measured as detailed in the ‘Methods'. Data are means±s.e.m. of (n) experiments.

TLOSRs and swallows

The number of TLOSRs in placebo experiments was similar to what has been reported previously (Lehmann et al., 1999; Figure 2). AFPSiA reduced the frequency of TLOSRs by 58±22% at 100 μmol kg⁻¹ (NS, Figure 2a). However, the maximal effect seemed to be achieved already at 14 μmol kg⁻¹ which produced 55±8% inhibition. The inhibition was statistically significant even at 7 μmol kg⁻¹. In contrast, spontaneous swallowing was only significantly inhibited at 100 μmol kg⁻¹ of AFPSiA and there was no trend for a dose dependency at lower doses (Figure 2b). Basal lower oesophageal sphincter pressure and duration of TLOSR were unaffected (data not shown), and the effect on acid reflux (defined as an oesophageal pH drop <4) was variable (not shown). The effect of AFPSiA declined with time, and amounted to 37±6 and 15±9%, 2 and 4 h after dosing, respectively (Figure 3). There were no visible side effects in any experiment after administration of AFPSiA to dogs.

Figure 2.

Dose–response curve for the effects of AFPSiA on TLOSRs (a) and spontaneous swallowing (b) in dogs. Note that the maximal inhibition of TLOSRs is about 60% and does not increase in the range of 14–100 μmol kg⁻¹. ^*P<0.05; ^**P<0.01, n (number of dogs) 3 (100 μmol kg⁻¹), 6 (14 μmol kg⁻¹) and 4 (the other groups). The control number (mean±s.e.m.) of TLOSRs/45 min for 7, 14, 28, 56 and 100 μmol kg⁻¹ was 5.9±0.5, 5.9±0.7, 5.5±0.8, 5.3±0.8 and 5.7±0.5, respectively.

Figure 3.

Effect duration of AFPSiA (14 μmol kg⁻¹) on TLOSRs in dogs. The inhibitory effect was abolished 4 h after administration. ^**P<0.01, n (number of dogs)=6. The control number (mean±s.e.m.) of TLOSRs/45 min for 1, 2 and 4 h was 5.9±0.5, 5.9±0.6 and 5.8±0.6, respectively.

Hypothermia in the mouse

Despite the potency of AFPSiA at the GABA_B receptor, high doses of the compound were required to decrease body temperature in the mouse. At 1 and 2 mmol kg⁻¹, there were brief and minor effects, while 4 mmol kg⁻¹ produced a mild but more sustained hypothermia (Figure 4a). In contrast, 6 and 8 mmol kg⁻¹ of AFPSiA induced an initial moderate drop in body temperature which remained at a relatively stable level for about 100 min and then continued to decrease substantially for the remainder of the experiment.

Figure 4.

Dose–response curve for the hypothermic effects of AFPSiA (a) and AFPSoA (b) in the mouse. AFPSiA 1 mmol kg⁻¹ vs vehicle P<0.01; 2 mmol kg⁻¹ vs vehicle P<0.05; 4 mmol kg⁻¹ vs vehicle P<0.05; 6 mmol kg⁻¹ vs vehicle P<0.001; 8 mmol kg⁻¹ vs vehicle P<0.01 (all comparisons at 20 min after dosing); n=4–6. AFPSoA 1 mmol kg⁻¹ vs vehicle P<0.01; AFPSoA 2 and 4 mmol kg⁻¹ vs vehicle P<0.001 (all comparisons at 20 min after dosing); n=5–6. The mice were killed for ethical reasons 60 min after the highest dose.

The oxidised metabolite of AFPSiA, AFPSoA, induced a dose-dependent hypothermia (Figure 4b) and was more potent than the parent compound. At the highest dose tested (4 mmol kg⁻¹) the reduction in body temperature was so prominent that the mice had to be sacrificed 1 h after dosing on ethical grounds.

The GABA_B receptor antagonist CGP62349 (4.9 μmol kg⁻¹) neither affected body temperature in its own right nor the pronounced reduction in body temperature seen after 6 mmol kg⁻¹ of AFPSiA (Figure 5).

Figure 5.

Lack of an effect (P>0.05) of CGP62349 (4.9 μmol kg⁻¹) on AFPSiA (6 mmol kg⁻¹)-induced hypothermia in the mouse; n=6 (vehicle group) and n=5 (all other groups).

Pharmacokinetics

AFPSiA was rapidly absorbed from the gastrointestinal tract and the plasma peak appeared within an hour after administration in dogs and rats and declined with a half-life of 0.5 and 1.5 h, respectively (Table 2). The oral bioavailability was 52% in dogs and 71% in rats (Table 2). In rats, 40% of the administered dose of AFPSiA was recovered in urine after intravenous administration. The plasma concentration time profile for APFSiA and the sulphonate metabolite are presented in Figure 6a (dogs) and Figure 6b (rats administered 350 μmol kg⁻¹). There was a delayed increase in the concentration of AFPSoA, and the levels were below the limit of quantification in some of the early samples (Figure 6). However, the lower limit of quantification was at least ten times higher for AFPSoA than for the parent compound.

Table 2.

Pharmacokinetic constants for AFPSiA

Species	Dose (μmol kg−1)	CL (ml min−1 kg−1)	Vss (L kg−1)	t1/2λz (h)	F (%)	fe
Rat	7 (iv/po)	40±6 (3)	3.0±0.8 (3)	1.5±0.1 (3)	71±4 (4)	0.4a (2)
Dog	7 (iv/po)	19±6 (4)	0.6±0.1 (4)	0.5±0.2 (4)	52±17 (4b)

The data represent arithmetical means±s.d. (n).

^aAfter an intravenous dose of 14 μmol kg⁻¹.

^bF is calculated after an oral dose of 14 μmol kg⁻¹ for two dogs.

Figure 6.

Plasma levels of AFPSiA (solid line) and AFPSoA (interrupted line) after intragastric administration of AFPSiA to dogs (a; 56 μmol kg⁻¹, n=4) and rats (b; 350 μmol kg⁻¹, n=3). In the dog experiments, the levels of AFPSoA were below the limit of quantification until 90 min after dosing, and in the rat study, AFPSoA was below the limit of quantification at 30 and 60 min after dosing.

Discussion

The primary aim of the present work was to assess the effects of the GABA analogue AFPSiA on TLOSRs in dogs. The interest in this effect stems from the well-documented inhibitory actions on TLOSRs that baclofen and other GABA_B agonists have been shown to produce. However, since all such compounds tested previously carry the burden of producing central side effects at low doses, a secondary aim was to study the side-effect profile of AFPSiA. For ethical reasons, this could not be done in dogs. Mice were chosen since hypothermia in our experience is more pronounced in mice than in rats, for instance, and smaller quantities of compound are required. In addition, the suspicion that the sulphinate group may be converted to sulphonate in vivo motivated an investigation into the pharmacokinetics of AFPSiA and possible formation of the corresponding sulphonate, AFPSoA. Since repeated blood sampling is difficult to perform in mice, the pharmacokinetics and metabolite studies were performed in dogs and rats.

In contrast to baclofen and other centrally acting GABA_B agonists that can abolish TLESRs (Blackshaw et al., 1999; Lehmann et al., 1999), AFPSiA reduced them maximally by approximately 60%. However, the compound was found to have a markedly better safety profile in that only extremely high doses in mice could induce CNS effects. We have found in other studies that the threshold dose of baclofen to produce hypothermia in C57/Bl mice is about 20–30 μmol kg⁻¹ s.c., so the threshold dose for AFPSiA was approximately 30–50 times higher. Such a profile is consistent with a peripheral site of action of AFPSiA to inhibit TLOSRs. Presumably, this effect is exerted at the peripheral endings of the gastric vagal mechanosensitive afferents (Partosoedarso et al., 2001; Smid et al., 2001). Racemic baclofen is in our hands approximately five times less potent on the human GABA_B(1a,2) receptor compared with AFPSiA, yet, baclofen is about 10 times more potent on TLOSRs in the dog (Lehmann et al., 1999). Since the potency of GABA_B agonists on vagal afferents (Smid et al., 2001) is considerably lower than that reported for central preparations (Knight & Bowery, 1996), it may be suggested that AFPSiA inhibits TLOSRs by a selective action on the peripheral branches of vagal afferents, while baclofen preferentially acts centrally, possibly on the central terminals of vagal afferents. If this assumption is correct, inhibition of vagal afferents peripherally would then maximally reduce TLOSRs by 50–60%, while centrally mediated inhibition can abolish TLOSRs. Consequently, the advantage of AFPSiA in terms of safety window compared with baclofen is to some extent counterbalanced by the limited maximal effect on TLOSRs. However, for a benign disease such as GORD, safety outweighs efficacy, and inhibition of TLOSRs in humans by some 60% by baclofen (Lidums et al., 2000) has been shown not only to reduce acid (Ciccaglione & Marzio, 2003), nonacid (Vela et al., 2003) and bile reflux (Koek et al., 2003), but also to relieve GORD symptoms (Ciccaglione & Marzio, 2003; Koek et al., 2003; Vela et al., 2003). Another finding arguing for a peripheral site of action of AFPSiA was the lack of an effect on swallowing except for after the highest dose. Previous studies have shown that both baclofen and CGP44532, two centrally acting GABA_B receptor agonists, inhibit swallowing (Lehmann et al., 2002), presumably through an action on the central pattern generator for swallowing. The absence of an effect of AFPSiA on swallowing can be considered advantageous since primary peristalsis is the major mechanism for clearance of gastric refluxate (Allen et al., 1996). Compatible with the relatively short half-life of AFPSiA, the duration of the effect was comparatively brief and essentially absent 4 h after administration of the compound.

Stimulation of the GABA_B receptor on vagal efferents to the lower oesophageal sphincter has been shown to inhibit evoked lower oesophageal sphincter relaxation (Smid & Blackshaw, 2000), and vagal motoneurons innervating the lower oesophageal sphincter express GABA_B(1) receptors abundantly (McDermott et al., 2001). However, it appears unlikely that this effect underlies the inhibitory action of AFPSiA on TLOSRs since TLOSRs are discrete events, and a reduction in transmitter release from vagal efferents would presumably produce a graded inhibition of the depth of TLOSRs and not a diminution in incidence. In the present experiments, there were no indications of hypothetical ‘partial TLOSRs' and, indeed, such phenomena have never been reported after pharmacological interventions.

There was a conversion of AFPSiA to AFPSoA both in dogs and rats and the plasma AUC for the latter was approximately half of that of the former. In the dog, AFPSoA appeared later in blood than AFPSiA and quantifiable levels could only be detected 90 min after administration of the parent compound, that is, after the measurement of TLOSRs. It can thus be concluded that it is unlikely that AFPSoA contributes to the inhibition of TLOSRs after administration of AFPSiA. This notion is also based on the finding that the selective GABA_A receptor agonist muscimol has no effect on TLOSRs in the ferret (Blackshaw et al., 1999). The hypothesis that the effect of AFPSiA on TLOSRs was mediated by GABA_B receptors with no contribution from GABA_A receptors could not be further tested using antagonists since GABA_B antagonists are emetogenic in dogs in the presently used experimental setting and since GABA_A antagonists are epileptogenic. While the possibility that AFPSiA also may act on other targets cannot be excluded, this seems unlikely since other GABA_B agonists based on the GABA structure have never been shown to have any affinity for non-GABA_B(/GABA_A) proteins.

Hypothermia in the mouse accompanied administration of high doses of AFPSiA and AFPSoA. The time course after administration of high doses of AFPSiA was different from that seen for, for example baclofen, γ-hydroxybutyrate and muscimol (Quéva et al., 2003), which produce hypothermia that peaks within an hour and then rapidly returns towards control levels. Also, lower doses of AFPSiA and AFPSoA provoked a rapid and transient hypothermia. When AFPSiA was given at 6 or 8 mmol kg⁻¹, the hypothermic response appeared to consist of two phases. The first phase was characterised by a rapid and moderate hypothermia, and the second by a sustained and profound hypothermia. Notably, the second phase coincided with a prominent increase in plasma levels of AFPSoA noted in the rat experiments. It is suggested that the first phase of hypothermia is related to AFPSiA, and the second both to AFPSiA and AFPSoA. At lower doses, the formation of AFPSoA might not have been large enough to affect body temperature.

Despite the fact that AFPSiA was found to be a potent GABA_B agonist in vitro, the mouse experiments clearly showed that hypothermia was exclusively induced by stimulation of GABA_A receptors or some other non-GABA_B-receptor-mediated mechanism. The latter alternative seems less probable since there is no report on any affinity of GABA_B receptor agonists for other receptors. We have found (unpublished) that CGP62349 at the dose used here completely prevents hypothermia produced by other GABA_B agonists in the mouse. Additionally, CGP62349 inhibits baclofen-induced hypothermia in mice at a four-fold lower dose (Froestl et al., 1996). The dose of CGP62349 used in the present study should therefore have blocked central GABA_B receptors sufficiently to prevent any GABA_B-mediated hypothermia.

In analogy with the low selectivity of GABA for GABA_B over GABA_A receptors, the unsubstituted sulphinate analogue of GABA, 3-aminopropylsulphinic acid or homohypotaurine, has a low selectivity for the GABA_B receptor (Shue et al., 1996). Substitution with fluorine at the 2-position apparently does not alter this profile significantly, in fact, the selectivity profile of AFPSiA is identical to that of its unsubstituted congener. It may also be noted that replacement of the sulphinate group with a sulphonate group markedly directs the selectivity towards the GABA_A receptor. This finding is supported by the fact that 3-aminopropylsulphonic acid or homotaurine is a selective GABA_A receptor agonist (Shue et al., 1996) with a selectivity almost identical to AFPSoA.

While it was suspected that AFPSiA would be oxidised very rapidly and extensively to AFPSoA, AFPSiA was found to have a relatively high oral availability both in rats and dogs. Moreover, 40% of the given dose was recovered in urine in rats. This suggests that substituted 3-aminopropylsulphinic acids represent a group that may have a potential in drug development, particularly since at least AFPSiA may have a peripheral site of action and a low propensity to provoke central side effects. In addition to GORD, other indications of interest include, but are not restricted to, asthma and coughing (Chapman et al., 1993) as well as prevention of metastases (Ortega, 2003).

Abbreviations

AFPSiA

(2R)-(3-amino-2-fluoropropyl)sulphinic acid

AFPSoA

(2R)-(3-amino-2-fluoropropyl)sulphonic acid

EDTA

ethylenediaminetetraacetic acid

FLIPR

fluorescence imaging plate reader

GORD

gastro-oesophageal reflux disease

HBSS

Hank's buffered salt solution

HEPES

N-(2-hydroxyethyl)piperazine-N′-(2-ethanesulphonicacid)

PBS

phosphate-buffered saline

TLOSR

transient lower oesophageal sphincter relaxation