Introduction

Ghrelin is an acylated 28-amino acid peptide that was initially isolated from rat stomach and subsequently identified as the endogenous ligand for the growth hormone secretagogue receptor (GHS-R) (Kojima et al. 1999). Whilst ghrelin is predominately synthesised and secreted into the bloodstream by X/A-like cells found in the gastrointestinal oxynitic mucosa (Dornonville de la Cour et al. 2001), small amounts can also be found in other tissues including the placenta, testis, kidney, pituitary, small intestine, lung and brain (for review, see Gualillo et al. 2003).

In addition to its potent, dose-related growth hormone releasing effects, ghrelin also significantly stimulates prolactin, adrenocorticotropic hormone (ACTH) and cortisol release. Furthermore, ghrelin also has effects on appetite and energy homeostasis, controls stomach acid secretion and gut motility as well as effects on the pancreas, cardiovascular and immune system (for reviews, see Kojima and Kangawa 2005; Hosoda et al. 2006; Kojima et al. 2001a).

The GHS-R is a typical G-protein coupled (GPCR) seven-transmembrane (7-TM) receptor that, upon binding of ghrelin (or exogenous ligand), preferentially couples to Gq, which ultimately leads to increases in intracellular calcium. Two distinct ghrelin receptor cDNAs have been isolated. The first, GHS-R type 1a, encodes a 7-TM GPCR with binding and functional properties consistent with its role as the endogenous receptor for ghrelin. The other GHS-R cDNA, type 1b, is produced by an alternative splicing mechanism (Howard et al. 1996), but unlike the GHS-R type 1a, it does not bind either ghrelin or synthetic ligands, and its function therefore remains unclear (Gnanapavan et al. 2002; Camina 2006).

Peripherally, GHS-R mRNA has been described in numerous organs, including the heart, lung, liver, kidney, pancreas, stomach, small and large intestines, adipose tissue and immune cells (Gnanapavan et al. 2002; Guan et al. 1997; Hattori et al. 2001; Kojima et al. 2001b), highlighting a multifunctional role for ghrelin in these tissues (Broglio et al. 2003). It is worth noting, however, that there may be dissociations between GHS-R mRNA and actual protein expression (Ghelardoni et al. 2006). In the central nervous system (CNS), GHS-R mRNA is predominately expressed in the arcuate and ventromedial nuclei of the hypothalamus and hippocampus. Lower expression has also been noted in multiple hypothalamic nuclei, pituitary, substantia nigra, ventral tegmental area, dorsal and raphe nuclei, cortex as well as the dentate gyrus and CA2/CA3 regions of the hippocampus (Guan et al. 1997; Howard et al. 1996; Nakazato et al. 2001, Hou et al. 2006; Zigman et al. 2006).

Based on the widespread GHS-R mRNA expression in the CNS, it is reasonable to propose that ghrelin may play a role in other aspects of normal brain functioning other than those associated with appetite and energy homeostasis. In support of this, there are a few relatively recent publications demonstrating that ghrelin has a role in cognitive processing. For example, Carlini demonstrated that intracerebroventricular (ICV) administered ghrelin can significantly and dose-dependently improve performance in a rodent step-down avoidance task, and these effects were also associated with increased anxiogenic behaviours (Carlini et al. 2002). Additional studies then characterised the different roles of the hippocampus, amygdala and dorsal raphe nucleus in mediating the observed behavioural responses to ghrelin (Carlini et al. 2004). More recently, peripherally administered ghrelin has been shown to enter the CNS via a passive mechanism and bind to hippocampal neurones, promoting neuronal dendritic spine formation and generation of long-term potentiation (LTP). The electrophysiological and synaptic morphological changes observed in these studies were also associated with enhanced performance in spatial learning and memory tasks (Diano et al. 2006).

Over the last 15 years, there has been considerable effort directed at the discovery and development of selective ghrelin receptor agonists. This has been driven primarily by the hypothesis that agents that mimic the actions of ghrelin may have utility not only in growth hormone replacement therapy but also in disorders requiring increased nutritional intake, such as cancer-induced cachexia, post-operative ileus and potentially in gastrointestinal motility disorders such as neurogenic and diabetic gastroparesis (Binn et al. 2006; Murray et al. 2005). The medicinal chemistry efforts conducted by a number of pharmaceutical companies led to the identification of relatively high molecular weight and complex structured molecules, such as CP-424391-18 (Pan et al. 2001), CP-464709-18 (Carpino et al. 2002) and MK-0677 (Patchett et al. 1995).

Recently, a series of potent, achiral ghrelin receptor agonists exemplified by SB-791016-A were identified via high-throughput screening and data-mining approaches (Heightman et al. 2007). The “drug-like” properties of these molecules were subsequently enhanced in a series of lead optimisation steps, which ultimately led to the identification of GSK894490A, an orally bioavailable and selective non-peptide GHS-R agonist (Witherington et al. 2008). We therefore sought to determine whether two non-peptide ghrelin receptor agonists (CP-464709-18 and GSK894490A) were capable of reproducing the pro-cognitive effects reported with ghrelin peptide itself. We therefore examined their potential efficacy in three well-characterised cognition tests, the novel object recognition test, a modified “Atlantis” water maze test and reversal of a scopolamine-induced deficit in a cued fear conditioning. These tests were chosen as they each probe a different cognitive domain and are mediated by distinct underlying neural substrates, i.e. scopolamine-induced deficits in cued fear conditioning (amygdala dependent), a temporal (24 h) induced deficit in novel object recognition (perirhinal cortex and possibly hippocampal dependent) and the Atlantis water maze (hippocampal dependent).

Materials and methods

Animals

Male Lister hooded rats (Harlan, UK or Biological Resource Centre, Singapore) were housed four per cage in a temperature (18 ± 2°C) and humidity (40 ± 5%) controlled environment on a 12-h light/dark cycle with lights on at 7:30 a.m. Food and water were available ad libitum. All experiments were carried out in accordance with the Singapore National Advisory Committee for Laboratory Animal Research guidelines for the use and care of animals for scientific procedures and also met GlaxoSmithKline animal use ethical standards.

Test compounds

The ghrelin agonists examined in these studies were: GSK894490A: “N-[5-[(3R,5S)-3,5-dimethyl-1-piperazinyl]-2-(methyloxy)phenyl]-4-(5-methyl-2-furanyl) benzenesulfonamide hydrochloride” and CP-464709-18 “(3aR)-5-(2-methylalanyl)-O-[(2,4difluorophenyl)methyl]-d-seryl-3-oxo-3a-(2 pyridinylmethyl)-2-(2,2,2 trifluoroethyl)-3,3a,4,5,6,7-hexahydro-2H pyrazolo[4,3-c]pyridine tartaric acid.”

Both compounds (calculated as free base) were prepared fresh on each experimental day in a 1% methylcellulose solution (Sigma, UK). Compounds or vehicle (1% methylcellulose) was administered using their respective route at a 1 ml/kg dose volume.

Pharmacokinetic studies

Drug administration pre-treatment times were determined prior to all cognition studies to ensure that cognition testing was conducted at or close to the time of maximal CNS drug levels. For these studies, male Lister hooded rats (250–300 g) were administered GSK894490A, 3 mg/kg p.o. or CP-464709-18, 3 mg/kg s.c. and culled at one of five time points (0.5, 1, 2, 3 or 6 h; n = 3).

Post decapitation, trunk blood samples were collected into tubes containing EDTA as an anticoagulant and the contents diluted 1:1 with deionised water. The whole brain was removed and washed free of all blood using sterile saline and homogenised in deionised water [50:50 (v/v)]. Samples were extracted using protein precipitation with an 80% acetonitrile/20% ammonium acetate (10 mM) solution containing an internal standard. The samples were shaken for 20 min and then centrifuged at 4,000 rpm for 15 min. They were analysed for drug concentrations by reverse phase high-performance liquid chromatography-tandem mass spectrometry using a API4000 triple quadrupole mass spectrometer (Applied BioSystems, USA). The lower limit of quantification was 5 ng/ml for blood samples and 10 ng/g for brain tissue.

Novel object recognition 24-h temporal deficit model

Animals were pre-handled and sham dosed before and after a 1-h habituation session to the test caging (Tecniplast, UK) for 2 days before the initial presentation of the objects (T1 trial). The objects used in these studies were custom fabricated black acrylic cubes (5 cm3) and cylinders (diameter and height, 5 cm; Labman Design, Singapore). Magnets were recessed into the base of the objects and the test cage positioned upon a magnetic plate which prevented the animals from moving the objects during the exploration trials. Based on the findings from the pharmacokinetic studies, GSK894490A (0.3 and 3.0 mg/kg p.o.; n = 12) or vehicle (n = 12) was administered 2 h prior to the T1 and T2 (recognition) trials. whilst CP-464709-18 (1 and 3 mg/kg s.c.; n = 12) or vehicle (n = 12) was administered 1 h prior to the T1 and T2 trials.

For the T1 trial, the animals were habituated to the test cage without objects for 3 min. The animals were then briefly moved to an adjacent cage for approximately 10 s, whilst two identical objects were placed into the test cage. The animals were then placed back for a further 3 min and the time spent exploring each object recorded by an experienced observer. For the T2 trial, animals were placed back into the test cage for a further 3 min habituation period 24 h after the T1 trial and then presented with one familiar and one novel object for a total of 3 min and object exploration recorded.

Objects were randomly assigned to ensure that treatment groups were fully balanced for both the novel object and its position within the test cage (either left or right). Object exploration was recorded only when the rat’s nose or mouth was in close contact with the object. Climbing or resting on the objects was not scored as exploration. The discrimination index (d2) was calculated as novel–familiar exploration/total exploration.

Atlantis water maze task

The water maze apparatus comprised a white fibreglass pool (diameter, 1.7 m; height, 0.65 m) housed in a fully partitioned space within the main procedure room. Surrounding the pool were a variety of spatial cues (posters and halogen light sources). The position of these remained constant throughout the entire studies. The water maze was filled with clean water warmed to 26 ± 1°C every morning. The water was made opaque by adding one litre of opacifier (Syntran® 5905, Interpolymer, USA). The pool circumference was arbitrarily marked: north, south, east and west; and using these points, the pool was divided into four imaginary quadrants.

An on-demand platform (diameter, 20 cm) sometimes referred to as an “Atlantis” platform (Buresova et al. 1985; Spooner et al. 1994) was placed in the centre of one of the four quadrants. When the platform was fully raised, it was covered by 2 cm of water and therefore invisible to the rat. A video camera was positioned directly above the tank to record the rat’s swim trajectory, and this was connected to a personal computer in which all of the parameters (latency, path length, swim speed) was acquired using Watermaze™ software (Actimetrics Inc. USA).

During the visual cue (VC) training session run on the first study day, a curtain was completely drawn around the water maze shielding the spatial cues. The Atlantis platform was set in the raised position (2 cm below water level), and a black acrylic cylindrical object was suspended 40 cm directly above the platform. During each of the four VC trials, the animals were trained to locate the platform using the black acrylic object as a visual cue. When the platform was located, the trial was stopped and the rat left on the platform for 30 s. Mean VC latency data was used to evenly distribute good and poor performers throughout each of the treatment groups.

During the spatial cue (SC) training sessions (Tuesday–Friday), the curtain was fully retracted so that the animals could utilise the spatial cues surrounding the pool. Performance in the SC training sessions was determined over a 4-day period (six trials per day). Unlike the fixed platform water maze protocol, the “Atlantis” platform enables the task difficulty to be gradually increased over the four training days, thus avoiding the ceiling effect typically associated with a fixed platform protocol. This is achieved by gradually extending the time that the animal has to dwell within the trigger zone across the 4 days. In these studies, the dwell time was increased from 0.8 s on day 1 to 3 s on day 4 (day 1, 0.8 s; day 2, 1.5 s; day 3, 2.3 s; day 4, 3.0 s).

During the first and second days of training, all animals are able to raise the platform by simply swimming directly over it. However, on the third and fourth days, the animal must circle and/or tread water within the trigger zone to raise the platform. In any particular trial, if the animal fails to locate and/or activate the platform, it is automatically raised after 90 s, and after 2 min, the animal is led to the platform using a pole. Once the animal has located the platform, the trial is stopped and the animal left on the platform for 30 s.

For each study, 20 naïve animals were assigned into two groups (n = 10), based on their VC performance to ensure equal distribution of good and poorer performers across each treatment group. GSK894490A (3 mg/kg p.o.) or vehicle was administered 2 h prior to each of the four spatial cue training sessions, whilst CP-464709-18 (3 mg/kg s.c.) or vehicle was administered 1 h prior. For the second study, CP-464709-18 (1 mg/kg s.c.) was administered twice daily (the first dose, 1 h prior to each spatial cue training session and once again 2 h after the last trial).

Reversal of a scopolamine-induced deficit in cued fear conditioning

Fear conditioning (FC) studies were performed using an automated video acquisition based system (MED Associates, USA). Prior to training, all chambers were individually calibrated so that the conditioning tone sound pressure (dB) and delivered shock were consistent across all of the chambers.

On day 1, animals were habituated to the FC chambers for 5 min to reduce any potential novel environment induced stress. On day 2, animals were trained to associate an auditory tone (CS) with a mild foot shock (US). Prior to training, two internal walls of the conditioning chambers were lightly wiped with a 3% acetic acid solution to provide a novel olfactory context. Animals were then placed into the chambers for a habituation period of 5 min. At the end of this period, the CS (2 kHz tone at 90 dB) was presented for a total of 5 s, and this co-terminated with the US stimulus (0.5 mA foot shock, 1 s duration). After the CS/US pairing, all animals were left in the chambers for 30 s and then returned to their home cages.

GSK894490A (1, 3 or 10 mg/kg p.o.; n = 16) or vehicle (n = 16) was administered 2 h prior to the CS/US training session, whilst CP-464709-18 (1, 3 or 10 mg/kg s.c.; n = 16) or vehicle (n = 16) was administered 1 h prior to CS/US pairing. Scopolamine (0.1 mg/kg s.c.) or vehicle (0.9% saline) was administered 30 min prior to the US/CS pairing sessions in both studies.

On day 3, the contextual environment of the chambers was altered by wiping two of the internal walls with a 3% ammonium hydroxide solution and inserting polka dot panels on the ceiling and rear walls. Animals were then placed back into their respective chambers and habituated for 2 min. The CS tone (2 kHz tone at 90 dB) was then presented for a total of 3 min and total freezing behaviour during this period monitored via video camera and automatically scored when there was an absence of all movement, except those relating to respiration.

Statistical analysis

All graphs were prepared using GraphPad PrismTM (v4), and all data are expressed as means ± SEM. Statistical analysis was performed using StatSoft StatisticaTM (v6), and all data were checked for normality prior to analysis.

For the novel object recognition and fear conditioning studies, a one-way ANOVA followed by planned comparisons was used to compare treatment groups. For the novel object recognition T2 trial data, a repeated-measures ANOVA followed by planned comparisons was used to compare novel vs. familiar object exploration. For the Atlantis water maze studies, the presence of a significant difference between treatment groups was determined by repeated-measures ANOVA followed by planned comparisons.

Results

Pharmacokinetic analysis of GSK894490A and CP-464709-18

Following a single 3 mg/kg oral dose of GSK894490A, maximal blood concentrations of 630 ± 200 nM were achieved 1 h post-administration, and these remained relatively constant for up to 3 h. At the 6 h time point, blood concentrations was reduced to 252 ± 39 nM (Fig. 1a). Maximal brain concentrations of GSK894490A were achieved at 3 h post-dose (240 ± 47 nM), and drug was still present at concentrations of 95 ± 10 nM 6 h post-administration (Fig. 1b).

Fig. 1
figure 1

Concentrations of GSK894490A and CP-464709-18 in a the blood and b the brain of male Lister hooded rats following a single 3-mg/kg dose. Following oral administration, maximal blood concentrations of GSK894490A were achieved after 1 h. Peak blood concentrations of CP-464709-18 which was administered by the subcutaneous route where seen at the 30-min time point. GSK894490A blood concentrations remained relatively constant between the 1- and 3-h time points, whereas CP-464709-18 blood concentrations had fallen by 95% at 3 h. Mirroring the blood data, GSK894490A brain levels were close to maximal concentrations within 1 h and remained relatively constant for up to 3 h. CP-464709-18 brain concentrations had declined by 80% at the 3 h time point. Data expressed as means ± SEM (n = 3)

Full size image

Following a single 3 mg/kg subcutaneous dose of CP-464709-18, maximal blood concentrations of 1,234 ± 229 nM were achieved 30 min post-administration but fell sharply during the three 6 h (Fig. 1a). Maximal brain concentrations of CP-464709-18 were also achieved at 30 min post-dose (70 ± 13 nM) but had fallen to near the lower limit of detection (14 ± 13 nM) at the 3 h time point. No drug was present in the brain at the 6 h time point (Fig. 1b).

Based on the brain exposure data, pre-treatment times of 2 h for GSK894490A and 1 h for CP-464709-18 were employed for all cognition training and testing.

Effect of GSK894490A and CP-464709-18 upon novel object recognition performance 24-h deficit model

There was no effect of GSK894490A treatment on overall object exploration during the T1 (F [2,33] = 0.023, P = 0.98) or T2 trials (F [2,33] = 0.027, P = 0.77; Table 1). However, during the CP-464709-18 study, drug-administered animals showed a highly significant effect of treatment on object exploration duration during T1 trial (F [2,21] = 8.06, P = 0.003; Table 1). Post hoc analysis revealed that both doses of CP-464709-18 significantly (P < 0.01) reduced object exploration by approximately 50% (Table 1). During the T1 trial, it was noted that CP-464709-18-treated animals appeared somewhat anxious immediately after dosing which resulted in the observed reduced exploration of the objects. Interestingly, following their second dose of CP-464709-18 24 h later, all animals appeared normal and were essentially indistinguishable from vehicle-treated animals in both their exploration rates and overall behaviours.

Table 1 Effect of GSK894490A (3 mg/kg p.o.; 2 h ptt) and CP-464709A (1 mg/kg s.c.; 1 h ptt) on novel object recognition performance
Full size table

As to be expected in this 24-h temporal deficit model, vehicle-treated animals explored the novel and familiar objects at similar levels during the T2 trial in both studies (Figs. 2a and 3a). Analysis of data from the GSK894490A study revealed that there was a significant effect of treatment on novel vs. familiar exploration (F [1,33] = 33.02, P < 0.001), and post hoc analysis revealed a significant effect of both doses (0.3 and 3 mg/kg; P < 0.001; Fig. 2a). Furthermore, analysis of the d2 index values also revealed a significant effect of treatment (F [2,25] = 10.07, P < 0.001), and post hoc analysis showed that both dose groups were significantly different from vehicle (0.3 and 3 mg/kg; P < 0.001; Fig. 2b).

Fig. 2
figure 2

Effect of GSK894490A (0.3 and 3 mg/kg p.o.; 2 h ptt) administered prior to the T1 and T2 trials on a novel vs. familiar object exploration and b d2 index (novel–familiar exploration/total exploration). No difference in novel vs. familiar object exploration was observed in the vehicle-treated animals. GSK894490A at both doses significantly enhanced recognition memory as evident by a significant (P < 0.001) increase in novel vs. familiar exploration which also translated to a significant (P < 0.001) increase in the d2 index. Data expressed as means ± SEM (n = 8)

Full size image
Fig. 3
figure 3

Effect of CP-464709-18 (1 and 3 mg/kg s.c.; 1 h ptt) administered prior to the T1 and T2 trials on a novel vs. familiar object exploration and b d2 index (novel–familiar exploration/total exploration). No difference in novel vs. familiar object exploration was observed in the vehicle-treated animals. CP-464709-18 enhanced recognition memory at the highest dose only (3 mg/kg) as evident by a significant (P < 0.001) increase in novel vs. familiar exploration which also translated to a significant (P < 0.001) increase in the d2 index. Data expressed as means ± SEM (n = 8)

Full size image

In the CP-463709-18 study, there was also a significant effect of treatment on novel vs. familiar exploration (F [1,21] = 29.9, P < 0.001), and post hoc analysis revealed a significant effect in one dose group only (1 mg/kg, F [1,21] = 2.83, P = 0.11; 3 mg/kg, F [1,21] = 59.51, P < 0.001; Fig. 3a). Analysis of the d2 index values also revealed a significant effect of treatment (F [2,33] = 12.07, P < 0.001), and post hoc analysis showed that, as with the novel vs. familiar data, only the 3 mg/kg dose group was significantly different from vehicle (1 mg/kg, P = 0.38; 3 mg/kg, P < 0.001; Fig. 3b).

Effect of GSK894490A and CP-464709-18 on the Atlantis water maze task

GSK894490A visual cue data were analysed with a repeated-measures ANOVA, and this revealed no differences in either latency (F [1,12] = 0.15, P = 0.70), path length (F [1,12] = 0.67, P = 0.42) or swim speed (F [1,12] = 0.40, P = 0.54) between the assigned treatment groups prior to dosing (Fig. 4a–c). Analysis of the spatial cue data showed that, whilst there appeared to be a trend for improvement on SC days 2 and 3, this failed to reach significance: SC day 1 (latency, F [1,12] = 2.84, P = 0.12; path length, F [1,12] = 2.48, P = 0.14), SC day 2 (latency, F [1,12] = 0.62, P = 0.45; path length, F [1,12] = 1.85, P = 0.20) and SC day 3 (latency, F [1,12] = 0.60, P = 0.45; path length, F [1,12] = 1.60, P = 0.23). There was, however, a significant reduction in both the latency and path length to locate and activate the submerged platform on SC day 4 (latency, F [1,12] = 16.50, P = 0.002; path length, F [1,12] = 18.36, P = 0.001; Fig. 4a, b). Analysis of the swim speed data showed that there was no effect of treatment on either of the four SC days: day 1, F [1,12] = 2.52, P = 0.14; day 2, F [1,12] = 0.50, P = 0.50; day 3, F [1,12] = 1.57, P = 0.23; or day 4, F [1,12] = 0.001, P = 0.98 (Fig. 4c).

Fig. 4
figure 4

Effect of GSK894490A (3 mg/kg p.o.; 2 h ptt) administered once daily prior to each SC training session on a latency to locate platform, b overall path length and c swim speed in the Atlantis water maze. There was no difference in either latency, path length or swim speed between the assigned vehicle and GSK894490A treatment groups during the visual cue day. GSK894490A-treated animals began to show an improvement in performance as determined by a reduced latency and overall path length to locate and activate the platform from spatial cue day 2 onwards. This improvement reached significance (P < 0.01) on the fourth spatial cue day. There was no treatment effect on swim speed on either of the test days. Data expressed as means ± SEM (n = 10)

Full size image

Analysis of the VC data from the single-administration CP-464709-18 study showed that there were no differences between the assigned treatment groups on either of the three measures: latency F [1,12] = 2.42, P = 0.15, path length F [1,12] = 2.93, P = 0.11 or swim speed F [1,12] = 1.40, P = 0.26 (Fig. 5a–c). Analysis of the SC data showed that there was no significant effect of treatment on either latency or path length to locate the platform on any of the four SC training days: day 1 (latency, F [1,12] = 2.05, P = 0.18; path length, F [1,12] = 3.24, P = 0.10), day 2 (latency, F [1,12] = 0.70, P = 0.42; path length, F [1,12] = 1.94, P = 0.19), day 3 (latency, F [1,12] = 0.007, P = 0.934; path length, F [1,12] = 0.82, P = 0.38) or day 4 (latency, F [1,12] = 0.056, P = 0.47; path length, F [1,12] = 0.054, P = 0.82 (Fig. 5a, b). Similarly, there was no effect of treatment on the swim speed on either of the 4 days: day 1, F [1,12] = 0.11, P = 0.75; day 2, F [1,12] = 0.65, P = 0.44; day 3, F [1,12] = 2.83, P = 0.14; or day 4, F [1,12] = 2.48, P = 0.14 (Fig. 5c).

Fig. 5
figure 5

Effect of CP-464709-18 (3 mg/kg s.c.; 1 h ptt) administered once daily prior to each SC training session on a latency to locate platform, b overall path length and c swim speed in the Atlantis water maze. There was no difference in either latency, path length or swim speed between the assigned vehicle and CP-464709-18 treatment groups during the visual cue day. CP-464709-18 failed to show any improvement in performance as determined by either a reduction in latency or path length to locate and activate the platform. There was no treatment effect on swim speed on either of the test days. Data expressed as means ± SEM (n = 10)

Full size image

As this negative result was unexpected, we hypothesised that the poorer pharmacokinetic properties of CP-464709-18, i.e. its shorter half-life and lower brain concentrations, were responsible for the observed lack of effect in the Atlantis water maze. The second study therefore aimed to determine whether maintaining drug exposure by administering two doses of CP-464709-18 (first dose at 1 h prior to training and the second dose at 2 h after the last trial) was able to improve performance.

Analysis of the VC data from this study showed that there were no differences between the assigned CP-464709-18 treatment groups on either of the three measures: latency F [1,12] = 0.0003, P = 0.99, path length F [1,12] = 0.003, P = 0.96 or swim speed F [1,12] = 0.003, P = 0.96 (Fig. 6a–c). Analysis of the SC data showed that, whilst there was no significant effect of treatment on either latency or path length on SC day 1 (latency, F [1,12] = 3.68, P = 0.08; path length, F [1,12] = 2.90, P = 0.11), day 2 (latency, F [1,12] = 1.30, P = 0.28; path length, F [1,12] = 0.36, P = 0.56) or day 3 (latency, F [1,12] = 0.49, P = 0.50; path length, F [1,12] = 0.42, P = 0.53), there was a significant reduction in both the latency and path length to locate and raise the platform on the fourth SC day 4 (latency, F [1,12] = 9.86, P = 0.009; path length, F [1,12] = 9.66, P = 0.009; Fig. 6a, b). Analysis of the swim speed data showed that there was no effect of treatment on either of the 4 days: day 1, F [1,12] = 0.33, P = 0.58; day 2, F [1,12] = 1.74, P = 0.21; day 3, F [1,12] = 0.33, P = 0.57; or day 4, F [1,12] = 0.64, P = 0.44 (Fig. 6c).

Fig. 6
figure 6

Effect of CP-464709-18 (1 mg/kg s.c.; 1 h ptt) administered once prior to each SC training session and once again 2 h after the last trial on a latency to locate platform, b overall path length and c swim speed in the Atlantis water maze. There was no difference in either latency, path length or swim speed between the assigned vehicle and CP-464709-18 treatment groups during the visual cue day. CP-464709-18-treated animals demonstrated a significant (P < 0.01) improvement in performance as determined by a reduced latency and path length to locate and activate the platform on spatial cue day 4. There was no treatment effect on swim speed on either of the test days. Data expressed as means ± SEM (n = 10)

Full size image

Representative trajectory plots (Fig. 6) illustrate the different swim strategies adopted by the vehicle-, GSK894490A- and CP-464709-18-treated animals during each day of the water maze studies. During their first exposure to the pool, all rats display the characteristic thigmotaxis response (pool wall circling) indicative of novelty-induced anxiety. As exposure to the pool increases, anxiety is attenuated, and the animals begin to explore the centre of the pool, ultimately locating the submerged platform (SC1). During the SC1 and SC2 training days, the dwell time of 0.8 and 1.5 s, respectively, is low enough that given an average swim speed 20 cm/s and a trigger zone diameter of 40 cm, the platform will rise if the animal swims directly across the platform location.

During the SC3 and SC4 training days, the animal is required to dwell within the trigger zone for 2.3 and 3 s, respectively, before the platform will rise. This increasing task difficulty paradigm tests the strength of the memory trace for the platform location. Closer examination of the video and tracking data shows that the GSK894490A and twice daily dosed CP-464709-18 animals begin to slow when approaching the trigger zone and then tread water waiting for the platform to rise, whilst the vehicle- and single-dosed CP-464709-18 animals crossed and circled the platform trigger zone but did not remain in the area for the necessary time to trigger the platform to rise (Fig. 7).

Fig. 7
figure 7

Representative single-animal tracking plots from the Atlantis water maze studies. During their first exposure to the pool, all animals displayed the characteristic thigmotaxis response (pool wall circling), indicative of novelty-induced anxiety. As exposure to the pool increases, anxiety is attenuated, and the animals begin to explore the pool centre, ultimately locating the submerged platform. Visual examination of the video and tracking data from the fourth SC training session shows that the GSK894490A and twice daily dosed CP-464709-18 animals reduce their swim speed when approaching the trigger zone and then tread water waiting for the platform to rise, whilst the vehicle and singly dosed CP-464709-18 animals crossed and circled the trigger zone but failed to remain in the area for the necessary time to raise the platform

Full size image

A comparison of body weights between the first and last days of the water maze studies showed that GSK894490A and CP-464709-18 treatment significantly increased bodyweight after 5 days of dosing when compared to their respective vehicle controls (Fig. 8).

Fig. 8
figure 8

Effect of GSK894490A and CP-464709-18 administration on body weight changes during the water maze study. Both compounds significantly increased bodyweight when compared to their respective vehicle-treated control animals following 4 days of drug treatment. Data expressed as means ± SEM (n = 10)

Full size image

Effect of GSK894490A and CP-464709-18 on a scopolamine-induced deficit in cued fear conditioning

Vehicle-treated animals displayed a clear association of the CS (tone) and US (foot shock) pairing as they spent 72 ± 6% and 82 ± 6% freezing during the 3 min CS tone presentation in the GSK894490A and CP-464709-18 studies, respectively. Scopolamine-treated animals spent a significantly shorter time freezing than their respective vehicle controls (GSK894490A, F [4,75] = 23.42, P < 0.001) and (CP-464709-18, F [4,75] = 18.50, P < 0.001), demonstrating a clear CS/US learning impairment. Neither of the ghrelin agonists attenuated the scopolamine-induced impairment (Table 2).

Table 2 Effect of GSK894490A (1–10 mg/kg p.o.; 2 h ptt) and CP-464709A (1–10 mg/kg s.c.; 1 h ptt) on a scopolamine-induced deficit in cued fear conditioning
Full size table

Discussion

Since its discovery and subsequent identification as the endogenous ligand for the growth hormone secretagogue receptor (Kojima et al. 1999), considerable effort aimed at elucidating the biological role of ghrelin has led to a detailed understanding of its growth hormone releasing properties (for review, see Bowers 1998), its role in energy homeostasis and appetite (Tschop et al. 2000; Wren et al. 2000, Cowley et al. 2003) and its action on the gastrointestinal (Masuda et al. 2000; Peeters 2005), cardiovascular (Cao et al. 2006; Enomoto et al. 2003), pulmonary (Santos et al. 2006; Volante et al. 2002), sleep wake cyles (Weikel et al. 2003; Schussler et al. 2006) and reproductive systems (Gualillo et al. 2001; Tanaka et al. 2003) amongst others (for review, see Lago et al. 2005).

Although it is now well known that GHS-Rs are widely expressed in brain regions other than the hypothalamus (Guan et al. 1997; Howard et al. 1996; Nakazato et al. 2001, Hou et al. 2006; Zigman et al. 2006), studies aimed at elucidating the additional roles that ghrelin may have in the CNS have been hampered by, until quite recently, a lack of non-peptide, centrally penetrant GHS-R ligands. Primarily driven by the recently published observations demonstrating that the peptide ghrelin can improve performance in rodent cognition tests (Diano et al. 2006; Carlini et al. 2002, 2004) and the availability of recently identified selective non-peptide, brain penetrant agonists (Pan et al. 2001; Carpino et al. 2002; Patchett et al. 1995; Heightman et al. 2007; Witherington et al. 2008), the aim of these studies was to determine if structurally diverse small-molecule ghrelin receptor agonists are also pro-cognitive.

GSK894490A and CP-464709-18 both significantly improved rodent object recognition memory as determined by their enhancing effects in the 24-h temporal deficit novel object recognition (NOR) test. Whilst both doses of GSK894490A were significantly active, only the highest dose of CP-46409-18 (3 mg/kg) was efficacious. Subsequent pharmacokinetic analysis of bloods and brains from the test animals confirmed the presence of each compound at levels similar to those observed during the preliminary pharmacokinetic time course studies. For example, brain concentrations of 200 ± 63 nM; n = 3 were achieved in the 3 mg/kg GSK894490A test group, whilst the 3 mg/kg dose of CP-464709-18 resulted in slightly lower brain concentrations of 75 ± 26 nM; n = 3.

Whilst 1 mg/kg of CP-464709-18 failed to significantly improve NOR performance, brain concentrations of 18 ± 4 nM; n = 3 were recorded from the test animals, which is considerably greater than its potency (pEC50 = 0.12 nM) at the rat recombinant ghrelin receptor. It is worth noting, however, that the drug concentrations reported here are “total drug” concentrations and do not take into account either blood or brain protein binding which may significantly reduce free drug levels available to activate the receptor.

An additional interesting difference between GSK894490A and CP-464709-18 effects in the NOR assay was the pronounced anxiogenic-like behaviours seen with CP-464709-18 during the T1 trial. Almost immediately after dosing and throughout the whole T1 trial, CP-464709-18-treated animals appeared anxious and hesitant when approaching the objects. These behaviours ultimately resulted in the significant reduction in object interaction times observed during the T1 trial. Anxiogenic-like behaviours as defined by significant reductions in open-arm entries of an elevated plus maze have previously been reported with both peripheral and ICV administered ghrelin (Asakawa et al. 2001; Carlini et al. 2002, 2008). As the anxiogenic effects were significantly attenuated with a corticotrophin releasing hormone (CRH) antagonist, it is believed that ghrelin may be mediating these anxiety effects by its action on the CRH/ACTH system (Asakawa et al. 2001).

Based on these prior published observations, perhaps the anxiogenic effects seen with CP-464709-18 during the NOR study should not be that surprising. However, why similar effects were not observed with GSK894490A is quite puzzling, especially if one considers that significantly higher blood and brain exposures were achieved with GSK894490A. It is also interesting to note that the CP-464709-18-induced anxiogenic effects were completely absent following the second dose of CP-464709-18 24 h later. Indeed, all of the treated animals appeared quite normal and were indistinguishable from vehicle animals. This attenuation of CP-464709-18-induced anxiogenic effects with repeated dosing is suggestive of tolerance, but in light of recent publication showing that ghrelin can actually elicit anxiolytic and antidepressant-like activity in the elevated plus maze and forced swim test, respectively (Lutter et al. 2008), further work aimed at understanding the link between ghrelin and anxiety is required.

Further support for the cognition-enhancing effects of the ghrelin agonists examined here were the robust effects seen in the Atlantis water maze studies. Whilst a single administration of GSK894490A was sufficient to improve performance, CP-464709-18 required two administrations, presumably because of its poorer CNS exposure and shorter half-life.

Typically, within an Atlantis water maze study, vehicle-treated animals show either a flat or negative learning slope across the four SC training days. Whilst all treated animals are able to locate the correct quadrant, the non-ghrelin-treated animals tended to adopt a more random search strategy within the target quadrant and consequently did not remain within the trigger zone for long enough to raise the platform. The ghrelin-treated animals, however, appear considerably more confident of the platform location and consequently either circled or tread water within the trigger zone.

One possible explanation for this apparent improvement in performance is that ghrelin administration is strengthening encoding of the spatial map during the training phase, but there may also be an effect on rule learning. For example, an animal that demonstrates good performance is not only required to recall the exact spatial location of the submerged platform but also to understand the rule, i.e. stay within this area and wait for the platform to rise. Whatever mechanism, ghrelin-treated animals clearly performed better in both the NOR and Atlantis water maze tasks, and based on the earlier findings by Diano et al. 2006, enhanced LTP and/or increased spine density are likely to be the key responsible neurobiological mediators.

Interestingly, there was no effect of either of the ghrelin agonists in reversing a scopolamine-induced impairment in cued fear conditioning (amygdala dependent). Whilst ghrelin projections have been shown in the amygdala (Cowley et al. 2003), activation of this pathway and other areas associated with a cued fear conditioning response was unable to attenuate the impairment caused by muscarinic receptor antagonism by scopolamine.

In summary, the two molecules evaluated in these studies, GSK894490A and CP-464709-18, are potent (pEC50 of 1.26 and 0.16 nM, respectively) organic ghrelin receptor agonists that show no discernable activity at any other receptors tested. Prior pharmacokinetic studies demonstrated that both compounds were orally bioavailable and readily crossed the blood/brain barrier, albeit with differing properties, and elicit pro-cognitive effects in rodent tests dependent on the hippocampus and rhinal cortices. Based on these observations, the central ghrelin receptor would appear to be a tractable receptor and perhaps should be considered as a drug target for therapeutic approaches to treat diseases affecting cognition such as Alzheimer’s disease and schizophrenia.