INTRODUCTION

Selective serotonin reuptake inhibitors (SSRIs) are the most frequently prescribed drugs in anxiodepressive disorders. Since SSRIs produce a tonic elevation of serotonin (5-hydroxytryptamine, 5-HT) levels in the extracellular space (Fuller, 1994; Malagié et al, 1995), their therapeutic effect is probably mediated, at least in part, by action at various levels of the serotonergic system, notably at serotonergic receptors (Barnes and Sharp, 1999). In particular, it has been proposed that the antidepressant effects of SSRIs would be related to desensitization of somatodendritic 5-HT1A and terminal 5-HT1B autoreceptors (Blier et al, 1987; Artigas et al, 1994,2001; Pineyro and Blier, 1999), which is induced by and participates indirectly in the increase in 5-HT concentration in the extracellular space (Gartside et al, 1995,1999; Sharp et al, 1997; Trillat et al, 1998; Adell et al, 2001; Malagié et al, 2001).

Another well-known action of SSRIs is their effects on sleep and wakefulness. Indeed, in several species and notably the rat (Pastel and Fernstrom, 1987; Ursin et al, 1989; Lelkes et al, 1994; Maudhuit et al, 1994; Neckelmann et al, 1996b), the hamster (Gao et al, 1992), and also in humans (Van Bemmel et al, 1993; Hendrickse et al, 1994), the most consistent action of SSRIs is a reduction of paradoxical sleep (PS), which is sometimes associated with an enhancement of wakefulness (W) and a secondary increase in slow wave sleep (SWS) (Maudhuit et al, 1994; Ursin, 2002). In the same manner, systemic treatment with selective agonists at 5-HT1A and 5-HT1B receptors induces an inhibition of PS and an enhancement of wakefulness, notably in rodents (Dzoljic et al, 1992; Tissier et al, 1993; Bjorvatn and Ursin, 1994; Monti et al, 1995; Bjorvatn et al, 1997; Boutrel et al, 1999,2002). In contrast, inactivation of 5-HT1A and 5-HT1B receptors (notably in mice with genetic deletions targeted at these receptors) facilitates the expression of PS (Boutrel et al, 1999,2002). Altogether, these data strongly suggest that 5-HT1A and/or 5-HT1B receptors, because they are activated by extracellular 5-HT, might mediate the action of SSRIs on sleep and wakefulness. However, this hypothesis has still to be validated because (i) 5-HT1A receptor antagonists were reported to be unable to prevent the effects of SSRIs on sleep or wakefulness (Bjorvatn et al, 1992; Neckelmann et al, 1996a) and (ii) to date, no study has been published with respect to the possible involvement of 5-HT1B receptors in PS inhibition by SSRIs.

These considerations led us to address this question by using knockout mice that do not express 5-HT1A or 5-HT1B receptors (Saudou et al, 1994; Ramboz et al, 1998). We analyzed the effects of citalopram, the most selective SSRI currently available (Hyttel, 1982; Milne and Goa, 1991), on sleep–wakefulness cycles in 5-HT1A and 5-HT1B knockout mice and their wild-type counterparts. In addition, we investigated whether pharmacological blockade of either 5-HT1A or 5-HT1B receptors with selective antagonists could prevent the effects of citalopram on sleep in the mouse.

MATERIALS AND METHODS

Animals

All the procedures involving animals and their care were conducted in conformity with the institutional guidelines that are in compliance with national and international laws and policies (council directive 87–848, October 19, 1987, Ministère de l'Agriculture et de la Forêt, Service vétérinaire de la santé et de la protection animale, permissions 75–116 to MH and 75–125 to JA). All mice were of the same 129/Sv genetic background. They were produced in the laboratory from homozygous breeding of wild-type and mutant (5-HT1A−/− and 5-HT1B−/−) strains (Saudou et al, 1994; Ramboz et al, 1998; Boutrel et al, 1999,2002), and housed in standard animal care facilities (see below).

Surgery

At 2–3 months of age (body weight: 21–27 g), male wild-type and mutant 5-HT1A−/− and 5-HT1B−/− mice were implanted under sodium pentobarbital anesthesia (70–75 mg/kg i.p.) with the classical set of electrodes for polygraphic sleep monitoring, as previously described (Boutrel et al, 1999). In brief, EEG electrodes were inserted through the skull onto the dura over the right cerebral cortex (2 mm lateral and 4 mm posterior to bregma) and over the cerebellum (at midline, 2 mm posterior to lambda), EOG electrodes were positioned subcutaneously on each side of the orbit, and EMG electrodes were inserted into the neck muscles. All electrodes were anchored to the skull with super-bond cement (Limoge-Lendais et al, 1994), and soldered to a mini-connector secured with acrylic cement. After completion of surgery, animals were housed in individual cages (20 × 20 × 30 cm3) and maintained under standard laboratory conditions: 12–12 h light–dark cycle (light on at 07:00 am), 24±1°C ambient temperature, food and water ad libitum. The animals were allowed at least 10 days to recover and habituate to the recording conditions.

Pharmacological Treatments

Drugs were dissolved in 0.1 ml of saline and injected intraperitoneally (i.p.) between 9:30 and 10:00 am. In case of combined treatments, the antagonist or saline was injected 15 min prior to citalopram. For baseline data, mice were injected with saline only, in single or combined treatment, as appropriate. A washout period of at least 3 days was allowed between two consecutive treatments. One series of animals was used to analyze the effects of citalopram alone, while another one was used in experiments with combined treatments. In both series, each animal received all doses and compounds, in a counter-balanced fashion. However, some recordings were discarded from the results because of insufficient quality. Thus, the number of tests finally contributing to the data was comprised between 4 and 10.

Polygraphic Recording

Polygraphic sleep monitoring was started just after the injection(s) and continued during 8 h thereafter.

Data Analysis and Statistics

Recordings were scored manually every 15 s epoch using the classical criteria for mice (Boutrel et al, 2002). The effects of each treatment on the states of vigilance (W, SWS, PS) were analyzed for every 2 h period after injection, and were expressed (mean±SEM) as minutes and as percentage of baseline (injection of saline). Statistical analyses were performed using ANOVA for the factors treatment and strain of mice. In case of significance, this ANOVA was followed by a post hoc Fisher test or a paired Student's t-test as appropriate, in order to determine statistical significance of the effect of each dose of a given compound.

Chemicals

WAY 100635 (N-[2-[4-(2-methoxyphenyl)-1-piperazinyl]ethyl]-N-(2-pyridinyl) cyclohexane carboxamide) (0.5 mg/kg i.p.) was obtained from Wyeth Research (Princeton, NJ, USA); GR 127935 (2′-methyl-4′-(5-methyl-[1,2,4]oxadiazol-3-yl)-biphenyl-4-carboxylic acid[4-methoxy-3-(4-methyl-piperazine-1-yl)-phenyl]amide) (1 mg/kg i.p.) was from Glaxo-Wellcome (Ware, UK); citalopram (1–10 mg/kg i.p.) was from Lundbeck (Copenhagen, DK).

RESULTS

Pharmacological Blockade of the Serotonin Transporter

In wild-type mice, blockade of the 5-HT transporter by citalopram induced a significant decrease in PS amounts, which lasted for 2–6 h after the injection (Figure 1). This effect was dose-dependent (ANOVA, F3,21=22.9 and 14.7, p<0.0001; and F3,21=2.7, p=0.06 for the three successive 2 h periods after injection, respectively), and significant for the doses of 5 and 10 mg/kg of citalopram until 4 h (p=0.0005, n=7) and 6 h (p=0.01, n=5) after treatment, respectively. This citalopram-induced PS reduction was accounted for by an increase in PS latency (150.3±7.2 and 157.4±13.7 min for 5 and 10 mg/kg vs 68.8±10.8 and 63.8±13.5 min at baseline, respectively; means±SEM for seven and five tests, respectively, p<0.01), and a decrease in the number of episodes with no significant modification of their mean duration (54±4 and 62±7 s during 4 h after 5 and 10 mg/kg of citalopram, compared to 54±6 and 57±6 s after saline, respectively). After this initial decrease, PS amounts returned progressively to baseline values (Figure 1). Citalopram induced no modification of W, but a slight enhancement of SWS was observed between the 4th and the 6th hour after treatment with the dose of 1 mg/kg (Figure 1).

Figure 1
figure 1

Effects of the SSRI citalopram at various doses on wakefulness and sleep in wild-type mice during four successive 2 h periods after injection. Data (mean±SEM of seven animals, five to seven tests for each dose) are expressed as min/2 h. *p<0.05, significantly different from saline (open bars); Post hoc Fisher test.

Full size image

In 5-HT1B knockout mice, citalopram also produced a dose-dependent decrease in PS amounts during the first 4 h after injection (ANOVA, 0–2 h: F3,19=4.1, p=0.02 and 2–4 h: F3,19=6.7, p=0.003) (Figure 2). This effect was not significantly different from that in wild-type mice (ANOVA for factor strain: F1,40=0.90, p=0.35 and F1,40=0.004, p=0.95, for the 0–2 and 2–4 h periods after treatment, respectively) (Figure 2 vs Figure 1). It was because of an increase in PS latency (103.5±12.1 for 10 mg/kg vs 48.7±5.7 min after saline, n=4, p<0.001), and a decrease in the number of PS episodes with no modification of their mean duration (57±6 and 62±7 s during 4 h after 5 and 10 mg/kg of citalopram, compared to 57±6 and 60±4 s after saline, respectively; means±SEM for eight and five tests, respectively).

Figure 2
figure 2

Effects of the SSRI citalopram at various doses on PS in 5-HT1B−/− (left) and 5-HT1A−/− (right) mice during the first and second 2 h period after injection. Data (mean±SEM of, respectively, eight and seven animals, four to eight tests for each dose) are expressed as min/2 h. *p<0.05, significantly different from saline (open bars); Post hoc Fisher test.

Full size image

In contrast, in 5-HT1A knockout mice, citalopram induced a lesser decrease of PS amounts, which was observed only during the first 2 h after the injection (ANOVA: F3,18=6.3, p=0.004), and reached statistical significance only for the highest dose tested, 10 mg/kg (p=0.006, n=4) (Figure 2). Like that observed in wild-type mice, this PS decrease was accounted for by an increase in PS latency (106.9±15.7 vs 51.3±5.4 min after saline, n=4, p<0.001), and by a reduced number of PS episodes with no significant modifications of their mean duration (68±8 s for 10 mg/kg, vs 66±11 s at baseline, n=4). However, even at this highest dose, PS reduction caused by citalopram was significantly less in 5-HT1A−/− mutants than in both wild-type and 5-HT1B−/− mice (ANOVA for PS amounts across strains: F2,58=32.0 and 22.8, p<0.0001 during 0–2 and 2–4 h after injection, respectively) (Figure 2 vs Figure 1).

Finally, no significant modifications of W or SWS were observed after citalopram administration (1, 5, or 10 mg/kg i.p.) in 5-HT1B−/− and 5-HT1A−/− knockout mice (not shown).

Blockade of the Serotonin Transporter After Pharmacological Inactivation of 5-HT1A or 5-HT1B Receptors

Pretreatment with the selective 5-HT1A receptor antagonist, WAY 100635 (0.5 mg/kg i.p.), prevented the PS inhibitory effect of citalopram (5 mg/kg i.p.) in wild-type mice (Table 1 and Figure 3). Similar data were obtained in 5-HT1B−/− mutants. In contrast, in 5-HT1A−/− mice, no significant differences in PS amounts were observed whether or not WAY 100635 was injected before citalopram (p=0.50, n=5) (Table 1 and Figure 3). A significant enhancement of SWS and concomitant decrease in W were also observed in wild-type mice only, during 2 h post-treatment (Table 1).

Table 1 Effects of the SSRI Citalopram (5 mg/kg i.p.), in Association with the 5-HT1A Antagonist WAY 100635 (0.5 mg/kg i.p.) or the 5-HT1B Antagonist GR 127935 (1 mg/kg i.p.), on Wakefulness and Sleep in the Three Strains of Mice
Full size table
Figure 3
figure 3

Effects of inactivation of 5-HT1A and 5-HT1B receptors on the PS inhibition induced by the SSRI citalopram (5 mg/kg i.p.) during 4 h after injection. Data (mean±SEM of 10 mice in the wild-type and the 5-HT1A−/− groups, and six mice in the 5-HT1B−/− group) after saline and citalopram (empty bars), and WAY 100635 (0.5 mg/kg i.p., hatched bars) or GR 127935 (1 mg/kg i.p., black bars) in association with citalopram, are expressed as percentages of PS amounts in saline-treated mice. The number of tests is indicated in each bar. *p<0.05, significantly different from saline treatment (dashed line); Student's t-test. †p<0.05, significantly different from treatment with saline and citalopram (solid bars); Student's t-test.

Full size image

Pretreatment with the 5-HT1B receptor antagonist, GR 127935 (1 mg/kg i.p.), did not induce any modifications of the effects of citalopram on PS amounts in any strain, except for a partial prevention of PS inhibition in wild-type mice for the first 2 h after injection of the SSRI (Table 1). This combined treatment induced during 4 h a decrease in the amounts of PS in both wild-type and 5-HT1B−/− mice, and no modifications in 5-HT1A−/− mutants, these effects being similar to those observed after injection of citalopram alone (Figure 3).

DISCUSSION

The present data indicate that, in the mouse, blockade of the serotonin transporter by citalopram induces an inhibition of PS with no major modifications of the other states of vigilance. This action appears to be mediated mainly through 5-HT1A receptor stimulation resulting from the increase in extracellular 5-HT levels caused by 5-HT reuptake inhibition.

The citalopram-induced decrease in PS that we report herein for the first time in the mouse is similar to that previously described in other rodents with various SSRIs (Pastel and Fernstrom, 1987; Ursin et al, 1989; Gao et al, 1992; Maudhuit et al, 1994; Neckelmann et al, 1996b; Ursin, 2002). In addition, mice exhibited after treatment with moderate doses of citalopram a slight secondary increase in SWS amounts (see Figure 1 and Table 1), which was also found in rats (Ursin et al, 1989; Maudhuit et al, 1994). However, in contrast to the latter studies, no modifications of wakefulness were observed herein.

The PS inhibition induced by the SSRI citalopram is most probably the consequence of the enhancement of extracellular 5-HT levels resulting from blockade of the amine reuptake (Fuller, 1994; Malagié et al, 1995; Fabre et al, 2000). Among the receptors at which 5-HT in excess could act to induce PS inhibition, the 5-HT1A and 5-HT1B types appeared as the most relevant candidates (Dzoljic et al, 1992; Tissier et al, 1993; Bjorvatn and Ursin, 1994; Monti et al, 1995; Bjorvatn et al, 1997; Boutrel et al, 1999,2002). Indeed, we can conclude from the present study that 5-HT1A receptors are primarily involved in such a PS inhibition because the effect of citalopram on sleep was markedly reduced in 5-HT1A−/−, but globally unchanged in 5-HT1B−/−, mutants compared to wild-type mice. The almost complete loss of citalopram action in 5-HT1A−/− mutants cannot be accounted for by a downregulation of the transporter itself because only mild changes of the latter were observed in knockout vs wild-type animals, notably in the brainstem (Ase et al, 2001), a key area for 5-HT-mediated PS regulation (Horner et al, 1997; Bjorvatn and Ursin, 1998). Thus, the presence of functional 5-HT1A receptors seems to be a prerequisite for the effect of citalopram on sleep, at least at moderate doses. Indeed, this effect was abolished not only in 5-HT1A−/− mutants but also in wild-type mice after pharmacological blockade of 5-HT1A, but not 5-HT1B receptors.

The question can be raised of whether the citalopram-induced PS inhibition could be secondary to a decrease in core temperature produced by this drug (Goodrich, 1983; Oerther and Ahlenius, 2001). However, in the rat, citalopram-induced hypothermia can be prevented by pretreatment with either 5-HT1A or 5-HT1B receptor antagonists (Oerther and Ahlenius, 2001), whereas in wild-type mice, citalopram-induced PS inhibition could be antagonized by blockade of 5-HT1A but not 5-HT1B receptors (Figure 3, Table 1). This difference strongly suggests that the effects of citalopram on PS are probably not secondary to its effects on core temperature. However, replication on wild-type and 5-HT1A−/−/5-HT1B−/− mutant mice of the study performed by Oerther and Ahlenius (2001) in rats should provide further answer to this question.

In line with previous studies (Boutrel et al, 1999,2002), the data obtained herein with the selective 5-HT1A receptor antagonist WAY 100635 (Fletcher et al, 1996) associated with citalopram (see Table 1) confirm that this receptor type mediates a tonic inhibition of PS in the mouse (Boutrel et al, 2002). Indeed, as expected from such a 5-HT1A receptor-mediated physiological control of PS, blockade of 5-HT1A receptors by WAY 100635 promoted PS in wild-type mice, when this antagonist was administered alone (Boutrel et al, 2002) or in combination with citalopram (Figure 3, Table 1). Indeed, in the latter case, PS amounts could not be affected by citalopram (because of the blockade of 5-HT1A receptors) and only the enhancing effect of WAY 100635 (see Boutrel et al, 2002) was observed. However, these data are at variance with those obtained in rats where other 5-HT1A antagonists (alprenolol, NAN-190) failed to oppose the action of SSRIs on sleep (Bjorvatn et al, 1992; Neckelmann et al, 1996a). The antagonist properties of the latter ligands at various receptors in addition to 5-HT receptors (Claustre et al, 1991; Hodgkiss et al, 1992; Hamon, 1997) and/or the use of different species, might possibly explain such differences compared to the present data obtained with the highly selective 5-HT1A antagonist WAY 100635.

At the doses used, citalopram after 5-HT1A receptor blockade in wild-type mice induced an increase of PS during 2 h, whereas in the absence of 5-HT1A receptor expression, that is, in 5-HT1A−/− mutants, it induced a slight PS decrease (Table 1). One probable explanation of this difference is that in the first case one deals with a clear-cut pharmacological effect of WAY 100635 (Boutrel et al, 2002), whereas in 5-HT1A−/− mutants, adaptive phenomena might have occurred, notably at the level of 5-HT receptors. Indeed, 5-HT1B receptors were found to be supersensitive in 5-HT1A−/− mutants (Boutrel et al, 2002), and such a functional change might account, at least in part (see below), for the slight PS decrease observed after citalopram treatment (Table 1). In the whole, these data further support the idea that constitutive knock out of a specific gene does not produce exactly the same effects as the pharmacological blockade of the corresponding encoded protein.

In contrast to the effects reported herein on sleep regulations, it has been shown that combined treatment with 5-HT1A antagonists and 5-HT reuptake inhibitors potentiated the effects of the latter drugs, at least with respect to 5-HT release in the hippocampus, food intake, and nociception (Hjorth, 1996; Trillat et al, 1998; Ardid et al, 2001). Such a facilitatory effect of 5-HT1A antagonists can be interpreted as resulting from the blockade of somatodendritic 5-HT1A autoreceptors from which 5-HT can trigger a negative control on serotonergic cell firing, notably after citalopram treatment (Evrard et al, 1999). With regard to sleep, WAY 100635 did not potentiate the action of citalopram, which confirms that the 5-HT1A-mediated action of 5-HT on sleep involves 5-HT1A receptors that differ from somatodendritic 5-HT1A autoreceptors. Indeed, clear-cut evidence of the implication of postsynaptic 5-HT1A receptors in the 5-HT-mediated inhibitory control of PS has already been reported in rats (Tissier et al, 1993). Further studies suggested that these receptors are located in specific brainstem nuclei (Horner et al, 1997; Bjorvatn and Ursin, 1998).

Interestingly, blockade of 5-HT reuptake by citalopram had almost the same effect on PS in wild-type and 5-HT1B−/− mutant mice. This observation is in line with recent data showing that the molecular target of citalopram, the 5-HT transporter, is not modified in the brainstem (Ase et al, 2001) and only slightly enhanced in the nucleus raphe dorsalis (Evrard et al, 1999) of 5-HT1B−/− mutants compared to their wild-type counterparts. We observed that PS amounts were largely reduced after citalopram treatment whether or not 5-HT1B receptors were functional, that is, in 5-HT1B−/− mutants and in wild-type mice pretreated with either saline or the 5-HT1B antagonist GR 127935 (Pauwels, 1997). Accordingly, it can be concluded that, in contrast to 5-HT1A receptors, 5-HT1B receptors do not play a crucial role in the inhibitory effect of citalopram on PS. Nevertheless, some modulation of this effect through the latter receptors cannot be excluded. In particular, it has to be emphasized that PS inhibition induced by treatment with saline+citalopram (5 mg/kg) was significant for only 2 h in 5-HT1B−/− mutants, whereas it lasted for 4 h in wild-type mice (Table 1); in addition, the 5-HT1B antagonist, GR 127935, partly reversed the action of citalopram in the latter animals (Table 1, Figure 3). On the other hand, PS inhibition following treatment with 10 mg/kg of citalopram in 5-HT1A−/− mice (see Figure 2) might result from activation of 5-HT1B receptors, inasmuch as PS in these mutant mice was found to be more sensitive to 5-HT1B receptor agonists (Boutrel et al, 2002). Meanwhile, the fact that 5-HT1B−/− mutants exhibited some increase in PS amounts under baseline conditions compared to wild-type mice (see Table 1) confirms that 5-HT1B receptors are involved in a tonic inhibition of PS in the mouse (Boutrel et al, 1999). However, overall comparison of the respective effects of 5-HT1B vs 5-HT1A receptor inactivation on PS under baseline conditions as well as after citalopram treatment (see Table 1) clearly indicates that 5-HT1A much more than 5-HT1B receptors mediate the 5-HT inhibitory control of this sleep stage in the mouse.

To conclude, given that insomnia, one cardinal symptom of depression, may be exacerbated by SSRIs, notably at initiation of the treatment (Levitan et al, 2000; Londborg et al, 2000), the present data provide further support to the therapeutic strategy consisting of the association of these antidepressants with 5-HT1A antagonists (Artigas et al, 1994). Indeed, such combined treatments would simultaneously accelerate and/or augment the antidepressive action (Artigas et al, 1994,2001; Blier and Bergeron, 1995; Stahl, 1998; Adrien, 2002), and limit the sleep-inhibitory effect of of SSRIs (Thase, 2000).