Main

Corticotropin-releasing factor (CRF) plays a crucial role in modulation of the release of adrenocorticotrophic hormone from the anterior pituitary (De Souza 1995). In addition, CRF is broadly distributed throughout the mammalian central nervous system (CNS), being concentrated in corticolimbic regions such as amygdala, hippocampus, and periaqueductal gray (PAG) and frontal cortex (FCX), as well as the locus coeruleus (LC) and dorsal raphe nucleus (DRN), the origin of ascending adrenergic and serotonergic projections, respectively (Sawchenko et al. 1993; Van Bockstaele et al. 1996; Price et al. 1998; Steckler and Holsboer 1999). In line with this organization, independently of the hypothalamocorticotrophic axis, cerebral CRF-containing neurones fulfill an important role in the control of emotional behavior (Adamec and McKay 1993; De Souza 1995; Mitchell 1998; Steckler and Holsboer 1999; Smagin and Dunn 2000), actions expressed via two principal subtypes of CRF receptor, both of which positively couple to adenyl cyclase (Grigoriadis et al. 1996). CRF1 receptors bind CRF and a related neuropeptide, urocortin, with high affinity whereas CRF2 receptors display a distinct preference for the latter (Donaldson et al. 1996; Grigoriadis et al. 1996). Of several splice variants of the latter, the CRF2α subtype is found predominantly in the rodent CNS, notably in several corticolimbic regions enriched in CRF itself, such as the hippocampus, medial amygdala, septum, olfactory bulb, and raphe nuclei, although species differences in this regard should be pointed out (Chalmers et al. 1995; Sánchez et al. 1999). Studies of mRNA encoding CRF1 receptors and of the corresponding peptide have established a contrasting pattern of distribution in rats with a predominance of CRF1 over CRF2α sites in anterior pituitary, FCX and basolateral amygdala, as well as high levels in the hippocampus and PAG (Potter et al. 1994; Chalmers et al. 1995; Sánchez et al. 1999; Chen et al. 2000).

These observations provide a neuroanatomical substrate for a potential role of CRF1 and/or CRF2 receptors in the control of mood (Coplan et al. 1996; Steckler and Holsboer 1999), and, in this regard, there is a diversity of evidence implicating CRF1 sites in the modulation of anxious states. First, several studies have reported anxiogenic actions of CRF (and urocortin) upon i.c.v. administration (see Steckler and Holsboer, 1999) and similar effects have been seen upon direct introduction into the dorsal PAG (Martins et al. 1997), basolateral amygdala (Sajdyk et al. 1999), and hippocampus (Radulovic et al. 1999), structures possessing a high density of CRF1 receptors. Second, direct evidence for participation of CRF1 receptors in anxiogenic actions is derived from studies of antisense probes for their neutralization, i.c.v. administration of which attenuates induction of anxiety by CRF (Skutella et al. 1998). Under certain conditions, antisense probes against CRF1, but not CRF2, receptors also display intrinsic anxiolytic activity, although such actions are variable (Skutella et al. 1994; Liebsch et al. 1995; Heinrichs et al. 1997). Third, underpinning these observations, CRF1 receptor-deficient mice display reduced anxiety in a variety of experimental paradigms (Timpl et al. 1998; Contarino et al. 1999; see also Steckler and Holsboer 1999). Fourth, contrariwise, mice over-expressing CRF1 receptors or lacking CRF-binding hormone show an increase in anxious behavior (Skutella et al. 1994; Stenzel-Poore et al. 1994; Karoloyi et al. 1999).

The fifth line of evidence concerns actions of CRF1 receptor antagonists. Peptidergic antagonists, such as 6-hel-CRF9-41 and astressin, attenuate anxiogenic actions of CRF and stress (e.g., Menzaghi et al. 1994; Spina et al. 2000). In analogy, the novel, non-peptidergic CRF1 antagonists, NBI27914, CRA1000, CRF1001 (all anilinopyrimidines), and CP154,526 (a pyrrollopyrimidine), inhibited the anxiogenic actions of CRF (Guanowsky et al. 1997; Smagin et al. 1998; Okuyama et al. 1999). However, like several studies of peptidergic antagonists in rats under non-stressed conditions (Heinrichs et al. 1992; Menzaghi et al. 1994; Spina et al. 2000), with the exception of NBI27914, they all failed to elicit anxiolytic activity alone in a plus-maze procedure (Lundkvist et al. 1996; Griebel et al. 1998; Okuyama et al. 1999). Further, whereas Griebel et al. (1998) reported modest anxiolytic actions of CP154,526 in a light-dark box paradigm in mice, in other studies, CP154,526, CRA1000, and CRA1001 were ineffective in this model unless mice were pre-exposed to stress (Guanowsky et al. 1997; Okuyama et al. 1999). The latter authors also documented their inactivity in a passive avoidance paradigm. As concerns other procedures, CP154,526 was active in “defensive withdrawal” procedures, decreased fear-potentiated startle model in rats, and suppressed separation-induced ultrasonic vocalizations (USV), whereas it was inactive in a conflict procedure in the rat and in a model of conditioned defeat in hamsters (Schulz et al. 1996; Griebel et al. 1998; Jasnow et al. 1999; Arborelius et al. 2000; Kehne et al. 2000). The novel phenylpyrimidine, R121919, is similarly active in a “defensive withdrawal” model (Heinrichs et al. 2000), and anxiolytic actions have also been claimed for a further pyrollopyrimidine derivative, antalarmin (Deak et al. 1999; Fiorino et al. 2000). On the other hand, while the pyrazolopyrimidine, DMP904, and the pyrazolotriazine, DMP696, were active in a rat model of “situational anxiety”, CP154,526 was not effective in this paradigm (Gilligan et al. 1998; 2000; He et al. 2000).

Clearly, the above data are rather disparate, and several reports remain preliminary. To explain contrasting patterns of data regarding the potential anxiolytic actions of non-peptidergic CRF1 antagonists, inter-species (and inter-strain) differences, as well as procedural variables and the level of stress, have been evoked (Griebel et al. 1998; Steckler and Holsboer 1999). Irrespective of underlying factors, the actions of non-peptidergic antagonists in therapeutically-pertinent models of potential anxiolytic activity are of critical importance as concerns the potential utility of CRF1 receptor blockade in the clinical treatment of anxious states (Owens and Nemeroff 1991). To clarify such issues, one instructive approach may be to simultaneously examine the actions of chemically distinct CRF1 antagonists in a number of contrasting procedures. In addition, it would be informative to directly compare their functional profiles to those of other classes of anxiolytic agent.

In light of the above comments, we compared the potential anxiolytic actions of CP154,526 in rats to those of a chemically distinct and novel CRF1 antagonist, the triazolopyridine, DMP695 (Bakthavatchalam et al. 1998). Like CP154,526, DMP695 shows high affinity for both cloned, human (h)CRF1 (Ki, 3.3 nM) and native, rat CRF1 (Ki, 4.6 nM) receptors, at which it potently expresses antagonist properties in suppressing CRF-induced increases in cAMP levels (Bakthavatchalam et al. 1998; Gilligan et al. 2000; Gilligan PJ, personal communication). In contrast, it shows low affinity for CRF2α and diverse other classes of receptor (Bakthavatchalam et al. 1998; Millan MJ et al., unpublished observations). Further, in vivo, DMP695 shows high bioavailability and is active in a model of situational anxiety in the rat—which is, interestingly, irresponsive to CP154,526 (Bakthavatchalam et al. 1998; Gilligan et al. 2000; He et al. 2000). In the present study, the actions of CP154,526 and DMP695 in plus-maze, Vogel conflict, SI, and conditioned USV procedures, were compared to those of the benzodiazepine (BZD), chlordiazepoxide; the serotonin (5-HT)1A agonist, flesinoxan; and the 5-HT2C antagonist, SB242,084 (Barrett and Gleeson 1991; Schreiber and De Vry 1993; Coplan et al. 1995; Griebel 1995; Miczek et al. 1995; Dekeyne et al. 2000; Millan et al. 2000a). Moreover, we determined their potential influence upon extracellular level of 5-HT, norepinephrine (NE) and dopamine (DA) in the FCX of freely moving rats.

METHODS

Animals

Unless otherwise specified, these studies employed male Wistar rats of 200–250 g and NMRI mice of 22–25 g (Iffa Credo, l'Arbresles, France) housed in sawdust-lined cages with unrestricted access to standard chow and water. There was a 12:12 hr light:dark cycle with lights on at 0730. Laboratory temperature and humidity were 21 ± 0.5°C and 60 ± 5%, respectively. Animals were adapted to laboratory conditions for at least a week prior to testing. They were used once only. All animal use procedures conformed to international European ethical standards (86/609-EEC) and the French National Committee (décret 87/848) for the care and use of laboratory animals.

SI Test

As previously described (Dekeyne et al. 2000), male Sprague-Dawley rats of 240–260 g (Charles River, Saint-Aubin-les-Elbeuf, France) were individually housed for 5 days before testing. On the test day, they were placed in weight-matched pairs (±5 g) in opposite corners of a highly illuminated (300 lux), open-topped arena (57 × 36 × 30 cm) for a 10 min session. A camera was mounted 2 m above the arena and was connected to a monitor and a videotape recorder in an adjacent room. The observer recorded from the screen the duration of active social interaction: that is, the time spent in grooming, following, sniffing, biting, jumping, or crawling over or under the other animal. If animals remained adjacent to each other without any movement for more than 10 s, scoring was discontinued until active SI resumed. Animals were administered with drug or vehicle 30 min before testing, with each rat of the same pair receiving the same treatment.

Vogel Test

As previously described (Dekeyne et al. 2000), the test was conducted in polycarbonate cages (32 × 25 × 30 cm) possessing a grid floor with the spout of a water bottle located 6 cm above the floor. Both the grid and the spout were connected to an Anxiometer (Columbus Instruments, Ohio, USA) used to record licks and deliver electrical shocks. During the 3 days preceding testing, rats were housed by four and were restricted to 1 hr-per-day access to tap water (from 0900 to 1000). On day 4, just after water delivery, they were isolated in cages with a grid-floor. Testing took place on day 5. Each rat was placed in the test cage and the session was initiated after the animal had made 20 licks and received a first, mild shock (a single, 0.5 s constant current pulse of 0.3 mA intensity) through the spout. Thereafter, a shock was delivered to the animal every twentieth lick during a period of 3 min. Only animals that initiated the session within 5 min were studied further. Data were the number of licks emitted by the animal during the 3 min session. Certain control (vehicle) animals did not receive shocks during the session and were used to evaluate free drinking behavior. Drugs were given 30 min before testing. The percentage of drug effect was computed as [(drug − vehicle)/(vehicle non-shocked − vehicle)].

Plus-maze Test

As previously described (Millan et al. 1997), the experiments were performed in a white-mat-painted plus-maze constructed of wood and elevated to a height of 50 cm. The apparatus comprised two open arms (50 × 10 cm) and two enclosed arms of the same dimensions, with walls 40 cm high. The two open arms were opposite to each other. On the test day, each rat was administered with drug or vehicle and was placed, 30 min later, in the central square of the maze facing one of the enclosed arms. The number of entries and time spent in open, and enclosed arms were recorded by an observer situated 2 m from the maze. An entry was counted only when the rat had its four limbs in an individual arm. Data were the total number of entries, the percentage entries and the percentage time spent in open arms. Drugs were given 30 min before testing.

USV Test

As previously (Millan et al. 1997), there were 3 different experimental phases performed at intervals of 24 hr. On day 1 (training), rats were placed in a chamber equipped with a grid-floor and were exposed to 6 randomly-distributed electric shocks (800 μA and 8 s) over a 7 min period. On day 2 (selection), they were placed in the chamber for 2 min and received a single shock. They were returned to the chamber 30 min later and ultrasonic vocalizations recorded for 10 min. Only rats emitting ultrasonic vocalizations for a total duration of at least 90 s were examined further. On day 3, the procedure was identical to day 2, but rats were treated with drug or vehicle immediately after the 2 min session. Data were the total duration of ultrasonic vocalizations recorded over the 10 min session.

Rotarod Procedure in Mice

As described previously (Dekeyne et al. 2000), 30 min after drug or vehicle injection, mice were placed on the rotating bar of a Rotarod apparatus (Ugo Basile, Varese, Italy) that gradually accelerated from 4 to 40 rpm over a period of 300 s. The latency of mice to fall was determined with a cut-off of 360 s.

Spontaneous Locomotion in Rats

As previously described (Dekeyne et al. 2000), rats were individually placed for 12 min in transparent polycarbonate cages (45 × 30 × 20 cm) equipped with two rows of photocells 4 cm above the floor and 24 cm apart. A locomotion count corresponded to the consecutive interruption within 2 s of 2 infrared beams. Drugs or vehicle were given 30 min prior to testing.

Determination of Dialysate Levels of Monoamines

The protocol employed is described in detail elsewhere (Gobert et al. 2000). Briefly, the influence of drugs upon levels of DA, NE, and 5-HT in single dialysate samples of the FCX was determined employing HPLC plus coulometric detection in freely moving rats implanted one week before testing with a guide cannula in this region. Samples were taken every 20 min. Following 2 hr “equilibration,” basal monoamine levels were monitored for 1 hr, then drugs were injected, and samples were taken for a further 3 hr. Changes were expressed relative to basal values (defined as 100%).

Drugs

For all drugs in all procedures, extensive dose-response relationships were examined. Incremental doses were tested until (1) statistical significance was attained, (2) the dose-response curve inflected, and/or (3) (for s.c. administration) the solubility limit was reached. All drug doses are in terms of the base. CP154,526 was administered i.p. as a suspension in carboxymethylcellulose (0.1%). Flesinoxan and DMP695 were administered s.c. in solution (sterile water). A few drops of lactic acid were added, and pH adjusted as close to normality (>5.0) as possible. SB242,084, chlordiazepoxide, and, for the Vogel test, DMP695, were administered i.p. as a suspension in water with a few drops of Tween 80. Drugs were injected in a volume of 1 ml/kg (rats) or 10 ml/kg (mice). Drug sources, salts, and structures were as follows: CP154,526 (butyl-ethyl-[2,5-dimethyl-7-(2,4,6-trimethylphenyl)-7H-pyrrolo [2.3-d] pyrimidin-4-yl]amine) HCl, DMP695 (N-(2-chloro-4,6-dimethylphenyl)-1-[1-methoxymethyl-(2-methoxyethyl]-6-methyl-1H-1,2,3,triazolo[4,5-c]pyridin-4-amine) mesylate, SB242,084 (6-chloro-5-methyl-1-[6-(2-methylpyridin-3-yloxy) pyridin-3-yl carbamoyl] indoline) HCl and racemic (±) flesinoxan HCl were synthesized by Servier chemists (P. Casara and G. Lavielle). Chlordiazepoxide HCl was supplied from Produits Roche (Neuilly-sur-Seine, France).

Statistics

In all behavioral studies, dose-effects were analyzed employing one-way analysis of variance (ANOVA) followed by Dunnett's test. Where computable (USV and motor procedures), Inhibitory Dose50s (ID50s) plus 95% confidence limits (CL) were calculated. In the dialysis study, data were analyzed by ANOVA with sampling time as the repeated within-subject factor.

RESULTS

Vogel Conflict Test

Over a dose-range of 5.0–80.0 mg/kg, CP154,526 significantly, dose-dependently, and markedly increased punished responses in the Vogel procedure (Figure 1 and Table 1 ). DMP695 mimicked this effect of CP154,526 in monotonically enhancing punished responses over a dose range of 10.0–40.0 mg/kg, with the latter dose achieving statistical significance. Chlordiazepoxide similarly showed robust activity in the Vogel procedure over a dose-range of 5.0–20.0 mg/kg. Flesinoxan was also active, with statistical significance obtained at doses of 2.5 and 10.0 mg/kg, although the dose-response curve inflected at a dose of 40.0 mg/kg. SB242,084 displayed only modest activity in this procedure, attaining statistical significance at a dose of 15.0 mg/kg.

Figure 1
figure 1

Actions in the Vogel test. VEH = vehicle; NS = non-stressed controls. Data are means ± SEMs. N = 5 per value. ANOVA as follows: CP154,526, F(3,37) = 13.5, p < .01; DMP695, F(3,22) = 3.2, p < .05; chlordiazepoxide, F(4,41) = 5.5, p < .01; flesinoxan, F(6,63) = 3.0, p < .05; and SB242,084, F(3,58) = 2.9, p < .05. Asterisks indicate significance of differences to corresponding vehicle values in Dunnett's test. *p < .05

Full size image
Table 1 Summary of Drug Actions in Models of Potential Anxiolytic versus Motor Activity
Full size table

SI Test

In pairs of unfamiliar rats exposed to an unfamiliar environment, CP154,526 elicited a pronounced, dose-dependent, and significant increase in active SI at doses of 0.16–2.5 mg/kg, with a further increase in the dose to 10.0 achieving no additional effect (Figure 2 and Table 1). DMP695 also elicited a dose-dependent and significant facilitation of active SI. Although chlordiazepoxide evoked a robust increase in SI, its dose-response curve was clearly biphasic, inflecting at the highest dose (10.0 mg/kg). Flesinoxan similarly manifested a biphasic dose-response curve in this paradigm. Finally, SB242,084 was potently active in enhancing SI. Simultaneous monitoring of other behaviours (rearing, locomotion and sleeping) revealed no significant effects of CP154,526 or DMP695 (not shown).

Figure 2
figure 2

Actions in the Social Interaction test. VEH = vehicle. Data are means ± SEMs. N = 5 per value. ANOVA as follows: CP154,526, F(4,31) = 3.2, p < .05; DMP695, F(3,18) = 4.8, p < .05; chlordiazepoxide, F(4,31) = 4.1, p < .01; flesinoxan, F(4,32) = 8.6, p < .01; and SB242,084, F(4,27) = 12.9, p < .01. Asterisks indicate significance of differences to corresponding vehicle values in Dunnett's test. *p < 0.05

Full size image

USV Test

Administered over a dose-range corresponding to doses active in the Vogel and SI models, neither CP154,526 nor DMP695 significantly reduced USV in rats re-exposed to an environment in which they had previously received an aversive stimulus (Figure 3 and Table 1). In distinction, chlordiazepoxide showed dose-dependent activity in this model. Flesinoxan also displayed marked activity, whereas SB242,084 was ineffective.

Figure 3
figure 3

Actions in the USV test. VEH = vehicle. Data are means ± SEMs. N = 5 per value. ANOVA as follows: CP154,526, F(4,32) = 0.6, p > .05; DMP695, F(3,18) = 1.9, p > .05; chlordiazepoxide, F(4,31) = 4.5, p < .01; flesinoxan, F(4,29) = 7.1, p < .01; and SB242,084, F(4,25) = 1.1, p >.05. Asterisks indicate significance of differences to corresponding vehicle values in Dunnett's test. *p < .05

Full size image

Plus-maze Test

Administered over a broad dose-range, CP154,526 did not significantly modify the number or percentage of open arm entries in the plus-maze procedure (Figure 4 and Table 1). It also did not significantly affect the total number of arm entries. DMP695 similarly failed to increase presence in the open arms and, at the highest dose tested, it tended to decrease entries and time in open arms. In distinction, chlordiazepoxide evoked a significant increase in entries and time in open arms, although its dose-response curve was biphasic. Both flesinoxan and SB242,084 were ineffective in increasing presence in open arms. They did not suppress total arm entries at any dose examined.

Figure 4
figure 4

Actions in the plus-maze test. VEH = vehicle. Data are means ± SEMs. N = 6 per value. ANOVA as follows: For % entries in open arms (left panels): CP154,526, F(3,28) = 0.8, p > .05; DMP695, F(4,31) = 1.3, p > .05; chlordiazepoxide, F(6,49) = 4.1, p < .01; flesinoxan, F(5,41) = 1.5, p > .05; and SB242,084, F(4,37) = 1.6, p > .05. For % time in open arms (middle panels): CP154,526, F(3,28) = 0.7, p > .05; DMP695, F(4,31) = 1.2, p > .05; chlordiazepoxide, F(6,49) = 4.0, p < .01; flesinoxan, F(5,41) = 2.1, p > .05; and SB242,084, F(4,37) = 0.8, p > .05. For total entries (right panels): CP154,526, F(3,28) = 0.5, p > .05; DMP695, F(4,31) = 3.1, p < .05; chlordiazepoxide, F(6,49) = 1.6, p > .05; flesinoxan, F(5,41) = 2.0, p > .05; and SB242,084, F(4,37) = 1.1, p > .05. Asterisks indicate significance of differences to corresponding vehicle values in Dunnett's test. *p < .05.

Full size image

Motor Behavior

CP154,526 did not significantly affect behavior in the rotarod test in mice (Table 2 ). It also did not significantly affect spontaneous locomotor activity in rats, although it tended to decrease activity at the highest dose evaluated. DMP695 elicited a dose-dependent reduction in latency to fall in the rotarod test in mice, and a dose-dependent reduction in spontaneous locomotor activity in rats. Chlordiazepoxide elicited a pronounced and dose-dependent ataxia in the rotarod test in mice and also reduced spontaneous locomotor activity in rats. Flesinoxan also showed clear activity in both procedures. Finally, SB242,084 was ineffective in the rotarod procedure and reduced locomotion only at the highest dose tested.

Table 2 Influence of Drugs on Motor Behavior
Full size table

Modulation of Dialysate Levels of Monoamines in FCX

CP154,526 failed to modify extracellular levels of 5-HT, NE or DA in the FCX of freely moving rats (Figure 5 and Table 3 ). In contrast, chlordiazepoxide evoked a marked and sustained diminution of levels of 5-HT, NE and DA. Flesinoxan markedly diminished levels of 5-HT, whereas those of NE and DA were simultaneously elevated. On the other hand, SB242,084 elevated levels of both NE and DA without modifying concentrations of 5-HT. (Because of insufficient quantities available, DMP695 was not examined in this procedure.)

Figure 5
figure 5

Influence upon extracellular levels of 5-HT, NE, and DA in dialysates of frontal cortex. Doses are indicated in mg/kg. Dialysate levels are expressed as a percentage of basal, pre-injection values, which were defined as 100%. These were 0.84 ± 0.09, 0.96 ± 0.06, and 1.79 ± .10 pg/20 μl dialysate for 5-HT, DA and NE, respectively. Data are means ± SEMs. n = 5 per value. ANOVA with dose as between factor and time as within factor was performed over 20–180 min. 5-HT: CP154,526, F(1,10) = 1.0, p > .05; chlordiazepoxide, F(1,9) = 70.9, p < .01; flesinoxan, F(1,12) = 50.6, p < .01; and SB242,084, F(1,10) = 1.1, p > .05. DA: CP154,526, F(1,10) = 0.5, p > .05; chlordiazepoxide, F(1,9) = 14.5, p < .01; flesinoxan, F(1,12) = 24.1, p < .01; and SB242,084, F(1,10) = 9.9, p < .05. NE: CP154,526, F(1,10) = 1.1, p > .05; chlordiazepoxide, F(1,9) = 13.0, p < .01; flesinoxan, F(1,12) = 11.2, p < .01; and SB242,084, F(1,10) = 14.2, p < .05. Asterisks indicate significant differences between the drug-treated group and the vehicle-treated group. *p < .05.

Full size image
Table 3 Area-under-the-Curve (AUC) Analysis of the Influence of Drugs upon Extracellular Levels of 5-HT, NE, and DA in Dialysates of Frontal Cortex
Full size table

DISCUSSION

Vogel Test

In line with a potential role in conflict paradigms, i.c.v. administration of CRF decreased punished responses in pigeons (Zhang and Barrett 1990). Further, employing a Vogel procedure, it was demonstrated that anxiogenic actions of social defeat are abolished in CRF1 knock-out mice (Van Gaalen et al. 1999), although a concurrent reduction of non-punished responses complicated interpretation of these data. In fact, α-helical-CRF9-41 was ineffective in a Geller-Seifter conflict paradigm in rats (Britton et al. 1986) and a similar lack of activity was documented for CP154,526 at doses of 0.6–20.0 mg/kg, i.p. (Griebel et al. 1998). Over this dose-range, CP154,526 similarly did not modify punished responses in the Vogel test, but, at a higher dose (80.0), a robust response was seen herein. Although CRF may enhance nociception (Millan 1999), CP154,526 is inactive in diverse algesiometric models (not shown), and any potential antinociceptive effect of CP154,526 is unlikely to be involved (Barrett and Gleeson 1991). In addition, although inactivation of CRF1 receptors attenuates the suppressive influence of stress upon food intake, non-peptidergic CRF1 antagonists exert little influence upon food and water intake (see Steckler and Holsboer 1999), and, at anxiolytic doses, CP154,526 did not affect food or water consumption (not shown). The observation that DMP695 evoked a comparable increase in punished responses underpins findings with CP154,526 and, collectively, the above-discussed data support a role of CRF1 receptors in modulation of emotionality in conflict procedures.

The activity of chlordiazepoxide in the Vogel test coincides with numerous reports of anxiolytic actions of BZDs in conflict paradigms (Barrett and Gleeson 1991; Griebel et al. 1998; Dekeyne et al. 2000). Further, the enhancement of punished responses by flesinoxan extends observations with various 5-HT1A agonists and with flesinoxan in other conflict models (Coplan et al. 1995; Griebel 1995; Dekeyne et al. 2000). Nevertheless, actions of 5-HT1A agonists are less marked than those of BZDs (Barrett and Gleeson 1991; Sanger 1992; Coplan et al. 1995; King et al. 1997; Millan et al. 1997). 5-HT2C receptor antagonists are also less efficacious than BZDs in conflict procedures (Cervo and Samanin, 1995; Kennett et al. 1996; Griebel et al. 1997). Indeed, SB242,084 displayed only modest activity in the Vogel paradigm herein, although it enhanced punished responses in a Geller-Seifter conflict procedure (Kennett et al. 1997).

Plus-maze Test

Robust anxiogenic actions of CRF, reversible by non-peptidergic and peptidergic CRF1 antagonists as well as antisense probes, have been observed in mice and rats employing the plus-maze procedure (Heinrichs et al. 1992; Menzaghi et al. 1994; Martins et al. 1997; Okuyama et al. 1999; Spina et al. 2000). In addition, a reduction in anxiety was detected in studies of CRF1 knock-out mice (Smith et al. 1998; Contarino et al. 1999), implying that endogenous CRF can mediate anxiety under these conditions. However, in the above-cited studies, intrinsic, anxiolytic actions of CRF1 antagonists were not reported. In fact, only a preliminary report has appeared of anxiolytic actions of the CRF1 antagonists, NBI27914 and NBI30545, in this paradigm (Wilcoxon et al. 1999). These findings contrast with the lack of activity of CRA1000, CRA100 (Okuyama et al. 1999) and, as shown herein, DMP695 and CP154,526. The latter was ineffective over a broad dose-range (0.63-80.0), supporting observations of Griebel et al. (1998). Moreover, although Lundkvist et al. (1996) reported “signs” of anxiolytic activity at 1.0 mg/kg, this action was absent at higher doses and expressed non-specifically inasmuch as closed arm entries also increased. There are several possible explanations for this lack of activity.

First, a preliminary report claimed that CP154,526 has weak partial agonist actions at CRF1 receptors (Grosjean-Piot et al. 1997), but this contention has not been confirmed and does not apply to DMP695, which was also ineffective (Schulz et al. 1996; Steckler and Holsboer 1999; Gilligan et al. 2000). Further, this explanation would not account for a lack of anxiolytic activity of other “pure” antagonists mentioned above. Second, contrasting actions of non-peptidergic antagonists might reflect differences in their mode of interactions at CRF1 receptors, which may possess multiple binding sites and/or various isoforms (Gilligan et al. 2000). Third, there may be species differences between mice and rats inasmuch as anxiolytic effects of CRF1 receptor deletion were demonstrated in the former with the plus-maze procedure, while essentially negative findings with CRF1 antagonists have been described in the latter (Steckler and Holsboer 1999). Fourth, the level of “stress” may be a critical variable, with the effect of CRF1 receptor blockade being proportional to the degree of stress experienced (Griebel et al. 1998; Okuyama et al. 1999; Steckler and Holsboer 1999; Keck et al. 2000).

Irrespective of such considerations, potential anxiolytic actions of CRF1 receptor antagonists in the plus-maze and other procedures reflecting exploratory activity, such as the light-dark box and “defensive withdrawal” models, would be of interest to document further (Griebel et al. 1998; Okuyama et al. 1999; Steckler and Holsboer 1999; Arborelius et al. 2000). These variable data may be contrasted with the reproducible anxiolytic actions of BZDs, such as chlordiazepoxide (Griebel et al. 1997, 1998). On the other hand, data with 5-HT1A agonists have proven highly variable, with both anxiolytic and anxiogenic actions, possibly mediated by pre- and postsynaptic 5-HT1A receptors, respectively (Schreiber and De Vry 1993; Andrews et al. 1994; Coplan et al. 1995; Collinson and Dawson 1997; Millan et al. 1997). Indeed, although flesinoxan enhanced open arm entries in a mouse plus-maze, it concurrently reduced total arm entries (Rodgers et al. 1994) and specific anxiolytic actions of flesinoxan were not found herein. While the 5-HT2B/2C antagonist, SB206,553, showed anxiolytic actions in a plus-maze study (Griebel et al. 1997), these authors underlined its “weaker anxiety-reducing potential” than BZDs, and SB242,084 was inactive in the present model. Nevertheless, for both 5-HT1A and 5-HT2C receptor ligands, quantification of anxiety-related behaviors other than arm entries may enhance test sensitivity (Rodgers and Cole 1994), and the exploitation of such an approach might similarly improve detection of potential actions of CRF1 antagonists.

SI Test

Extensive studies of social stress, such as aggressive encounters leading to “social defeat,” suggest an important role of CRF1 receptors in the modulation of emotionality in interaction with conspecifics (Heinrichs et al. 1992; Menzaghi et al. 1994; Liebsch et al. 1995; 1999; Jasnow et al. 1999; Spina et al. 2000). Such observations are pertinent to the SI model whereby unfamiliar pairs of rats are introduced to an unfamiliar, “stressful” environment. Under these conditions, i.c.v. administration of CRF (or urocortin) acts anxiogenically in suppressing active SI (Dunn and File, 1987; Moreau et al. 1997; Sajdyk et al. 1999). This action involves, at least partially, engagement of CRF1 receptors in the amygdala and is blocked by non-peptidergic CRF1 antagonists (Sajdyk et al. 1999; Steckler and Holsboer 1999). The present studies amplify, thus, such observations in demonstrating dose-dependent increases in SI with both CP154,526 and DMP695.

The enhancement of SI by chlordiazepoxide underpins studies with other BZDs, although dose-response curves inflect at high doses concomitant with onset of motor-suppressive properties (Griebel 1995; Dekeyne et al. 2000). 5-HT1A agonists are likewise effective in the SI model (Schreiber and De Vry 1993; Griebel 1995; Dekeyne et al. 2000), observations supported by the present findings with flesinoxan. Finally, the present data confirm the robust anxiolytic actions of SB242,084, in analogy to other 5-HT2C antagonists in this procedure (Kennett et al. 1996; 1997; Dekeyne et al. 2000).

USV Test

Central administration of CRF potentiates expression of conditioned fear in rats, an action to which CRF1 receptors in the PAG and hippocampus contribute, although a role of other structures should not be excluded (Schulz et al. 1996; Tershner and Helmstetter, 1996; Martins et al. 1997; Deak et al. 1999; Radulovic et al. 1999). Evidence that endogenous pools of CRF facilitate psychological stress was provided by Schulz et al. (1996), who showed that CP154,526 blocks both CRF- and fear-induced potentiation of the acoustic startle response in rats. In contrast, in the present model of conditioned fear, CP154,526 and DMP695 failed to inhibit USV. This lack of activity corresponds to the report by Okuyama et al. (1999) that CRA1000 and CRA1001 do not influence conditioned fear in a passive avoidance paradigm. On the other hand, CP154,526 was reported to reduce USV following separation in young rats (Kehne et al. 2000), while Deak et al. (1999), employing the freezing response, observed significant anxiolytic effects of antalarmin. Thus, the precise measure of anxiety may be a decisive variable determining drug actions, although many other factors, such as procedural details and the level of stress, may also underlie contrasting patterns of data.

This inactivity of CP154,526 and DMP695 contrasts to dose-dependent actions of chlordiazepoxide and other BZDs in this paradigm (Molewijk et al, 1995; Millan et al. 1997). On the other hand, SB242,084 was ineffective, in analogy to other 5-HT2C antagonists (Sànchez and Mørk 1999; Dekeyne A, unpub. obs.). This inactivity may be related to the contention that the USV model mimics a “ panic-like” state (Molewijk et al. 1995; Jenck et al. 1998) since, in contrast to other forms of anxiety, activation of 5-HT2C receptors (probably in the PAG) reduces emotionality under such conditions (Jenck et al. 1998). A relationship to panic states has also been claimed for potent actions of flesinoxan and other 5-HT1A agonists in the USV procedure (Griebel 1995; Millan et al. 1997; Sànchez and Mørk 1999), which were corroborated herein. However, the exacerbation of panic states by flesinoxan in man (Van Vliet et al. 1996; Jenck et al. 1999) questions this interpretation.

Motor Behavior

The potential influence of CRF1 receptors upon motor behavior critically depends upon novelty, suggesting that anxiety itself impacts upon this parameter (Steckler and Holsboer 1999). In fact, no major alteration of motor behavior is apparent in mice lacking the gene encoding CRF1 receptors, their neutralization with antisense, or treatment with selective CRF1 receptors antagonists (Griebel et al. 1998; Smith et al. 1998; Liebsch et al. 1999; Okuyama et al. 1999; Steckler and Holsboer 1999), and CP154,526 modified neither spontaneous locomotor behavior nor rotarod performance in mice. Indeed, anxiolytic actions of CP154,526 were exerted in the absence of marked motor perturbation. On the other hand, DMP695 interfered with motor behavior. The reason for this difference remains to be clarified since receptorial interactions of DMP695 at sites other than CRF1 receptors have not been described (Bakthavatchalam et al. 1998; Newman-Tancredi A and Millan MJ, unpublished observations). In any case, the decrease in motor function elicited by DMP695 clearly cannot underlie an increase in responses in the Vogel procedure and an increase in active SI.

While the influence of CRF1 antagonists upon motor behavior requires further characterization, BZDs, such as chlordiazepoxide, profoundly disrupt motor function. Further, although 5-HT1A receptor agonists, such as flesinoxan, are not sedative in humans; they perturb motor behavior in rodents (Coplan et al. 1995; Millan et al. 1997). Contrariwise, selective 5-HT2C receptor antagonists like SB242,084 compromise motor function only at high doses in rodents (Kennett et al. 1996, 1997; Griebel et al. 1998; Dekeyne et al. 2000).

Monoaminergic Transmission

DRN-serotonergic neurones are susceptible to modulation by CRF, likely acting via CRF1 receptors (Chalmers et al. 1995; Ruggiero et al. 1999). The predominant influence of CRF (at low concentrations) upon serotonergic neurones is inhibitory. This action is attenuated by antalarmin, which itself does not modify electrical activity (Price et al. 1998; Kirby et al. 2000). Correspondingly, CP154,526 did not affect dialysate levels of 5-HT in the FCX. Thus, in distinction to BZDs and 5-HT1A agonists, such as chlordiazepoxide and flesinoxan, respectively, which reduce extracellular levels of 5-HT (Millan et al. 1997), suppression of serotonergic transmission is unlikely to fulfill a major role in the anxiolytic actions of CRF1 antagonists. Nevertheless, very recently, CP154,526 (32 mg/kg, i.p.) was reported to transiently and slightly (15% relative to basal values) suppress extracellular levels of 5-HT in the hippocampus of conscious rats (Isogawa et al. 2000). Thus, the potential influence of various CRF1 antagonists upon 5-HT release in the FCX and other regions under “resting” and “anxious” conditions would benefit from additional study.

In contrast, CRF potently enhances the activity of LC-derived adrenergic pathways (Valentino et al. 1993; Curtis et al. 1997), an action which may, in principal, enhance anxiety (Charney et al. 1995; see Millan et al. 2000b), and which is prevented by intracerebral application of non-peptidergic CRF1 antagonists (Smagin et al. 1995; Curtis et al. 1997; Page and Abercrombie 1999) and CP154,526 (Braselton et al. 1996). These antagonists do not themselves suppress adrenergic activity, and CP154,526 failed to modify FCX levels of NE herein, in line with the lack of influence of local administration of CP154,526 into the LC upon adrenergic transmission (Kawahara et al. 2000). However, a mild (∼15% relative to basal values) diminution in frontocortical release of NE upon i.p. administration of CP154,526 (32 mg/kg) was recently documented by Isogawa et al. (2000). Further, CP154,526 and peptidergic CRF1 antagonists abrogate the induction of corticolimbic NE release provoked by stress (Shimizu et al. 1994; Smagin et al. 1997; Kawahara et al. 2000), an action which may contribute to their anxiolytic properties. Although this provides a parallel with BZDs, which also inhibit LC-adrenergic neurones, the significance of modulation of adrenergic transmission to anxious states remains under discussion (Swiergiel et al. 1992; Valentino et al. 1993; Weiss et al. 1994; Millan et al. 2000b). In any event, a lack of intrinsic influence of CRF1 receptor antagonists upon LC-adrenergic projections contrasts strikingly to the pronounced excitation elicited by 5-HT1A agonists and 5-HT2C antagonists, such as flesinoxan and SB242,084, respectively (Figure 5; Gobert et al. 2000; Millan et al. 2000a). Finally, the lack of influence of CP154,526 upon frontocortical (Figure 5) and hippocampal (Isogawa et al. 2000) levels of DA, which have been implicated in anxious states (Morrow et al. 1999), may be differentiated from their suppression by chlordiazepoxide and other BZDs, and their facilitation by 5-HT1A agonists, such as flesinoxan, and 5-HT2C antagonists, such as SB242,084 (see Millan et al. 2000a).

GENERAL DISCUSSION

As summarized in Table 4 , CP154,526 and DMP695 displayed similar profiles of anxiolytic activity, consistent with a common role of CRF1 receptors in their actions. In line with this contention, in extensive binding studies performed both in our laboratory (Newman-Tancredi A and Millan MJ, unpublished observations) and elsewhere (Schulz et al. 1996; Bakthavatchalam et al. 1998; Gilligan PJ, personal communication), as compared to CRF1 receptors, CP154,526 and DMP695 displayed affinities at least 100-fold lower for multiple serotonergic receptors and diverse other receptors, enzymes and channels. Further, as judged by doses blocking the actions of exogenous CRF, doses active in the present anxiolytic procedures correspond well to those required to occupy CRF1 receptor in vivo (Schulz et al. 1996; see Steckler and Holsboer 1999). Thus, it appears that other factors account for certain differences in the functional profiles of CP154,526 versus DMP695. Notably, DMP695 has a more pronounced influence than CP154,526 upon motor function and a relatively weak activity in the social interaction procedure. Interestingly, while the latter difference between DMP695 and CP154,524 concerns their active dose-ranges, previous studies have found qualitative differences between the anxiolytic profiles of various CRF1 antagonists. For example, as compared to DMP695 and other CRF1 antagonists, CP154,526 is inactive in a rat model of “situational anxiety” (Gilligan et al. 2000; Gilligan PJ, personal communcation). As mentioned above in relation to the plus-maze model, variables underlying such differences may include partial agonist activity, differential involvement of multiple CRF1 receptor isoforms and/or binding sites, and contrasting interactions with “stress”—which is more pronounced for the Vogel test than for the Social Interaction procedure. The resolution of such questions will require further comparative studies of CP154,526, DMP695 and additional CRF1 antagonists in the present and other functional models.

Table 4 Summary of Overall Functional Profiles Characterized in the Present Studies
Full size table

Finally, although chlordiazepoxide attenuates anxiogenic actions of CRF, and similarities between CRF1 antagonists and BZDs have been pointed out, the anxiolytic profiles of CP154,526 and DMP695 contrast with those of chlordiazepoxide and other BZDs. Although the latter may more broadly display anxiolytic properties, certain anxious states might be specifically responsive to CRF1 antagonists (Steckler and Holsboer 1999; Gilligan et al. 2000). The lack of activity of CP154,526 and DMP695 in the USV protocol differentiates them from 5-HT1A agonists. In fact, notwithstanding the lack of intrinsic influence of CRF1 antagonists upon monoaminergic transmission, the anxiolytic profile of CP154,526 and DMP695 most closely resembled that of 5-HT2C antagonists (Table 4). Indeed, in the light of mutual sites of action in the hippocampus and amygdala, studies of a possible interrelationship between CRF1 and 5-HT2C antagonists in the modulation of anxious states would be of interest.

CONCLUSIONS

In conclusion, the selective, non-peptidergic CRF1 antagonists, CP154,526 and DMP695, showed a similar profile of potential anxiolytic activity in rats. Their pattern of action could be distinguished from BZDs and 5-HT1A agonists, and resembled 5-HT2C antagonists. Nevertheless, in contrast to the other drug classes, CP154,526 did not influence extracellular levels of monoamines in FCX, and CRF1 antagonists likely possess a distinctive mechanism of action requiring further elucidation. Indeed, additional work is also necessary to more precisely characterize the potential utility of CRF1 receptor antagonists in the clinical treatment of anxious disorders.