INTRODUCTION

The clinical treatment of depression is characterized by a high degree of nonresponse. Additionally, the onset of action of antidepressants typically takes 2–4 weeks. In order to enhance efficacy and shorten onset, augmentation strategies have been developed. Most augmentation strategies are derived from the hypothesis that chronic selective serotonin reuptake inhibitor (SSRI) treatment results in a desensitization of presynaptic serotonergic (5-HT) autoreceptors. This blunted autoreceptor control leads to increased effects of SSRIs on central serotonin levels in time, which is postulated to parallel the antidepressant effect (Hjorth et al, 2000; Blier, 2001; Blier et al, 1987).

Augmentation strategies focus on establishing enhanced SSRI responses on central serotonin levels. It should be mentioned, however, that many of these experiments reflect the response of 5-HT to acute SSRI administration, and not chronic administration. Over the last few decades, several such augmentation strategies have been developed. These add-on strategies range from antagonism of serotonergic autoreceptors like 5-HT1A and 5-HT1B (Hjorth et al, 2000; Blier et al, 1998; Artigas et al, 1994, 1996; Bosker et al, 2001; Celada et al, 2001; Cremers et al, 2000a, 2000b), but also comprise heteroceptors such as adrenoceptors (Bengtsson et al, 1998; Szabo and Blier, 2001; Besson et al, 2000; Hopwood and Stamford, 2001; Bortolozzi and Artigas, 2003; Pudovkina et al, 2003; Rouquier et al, 1994; Amargos-Bosch et al, 2003; Weikop et al, 2004; Gobert et al, 1997; Gobert and Millan, 1999; De Boer et al, 1996), 5-HT2A receptors (Szabo and Blier, 2002), GABAB receptors (Abellan et al, 2000; Slattery et al, 2005; Mombereau et al, 2004; Nakagawa et al, 1996), and substance P receptors (Guiard et al, 2004), to mention but a few.

Our group recently reported on the acute augmentation of 5-HT levels by combining SSRIs with 5-HT2C antagonists (Cremers et al, 2004). The 5-HT2C receptor antagonist SB242084 was chosen as it has a 100-fold affinity over other 5-HT receptors (Kennett et al, 1997; Bromidge et al, 1997). As 5-HT2C receptors were not described to be prominently involved in controlling serotonin release, this observation was surprising. Furthermore, as no effects of 5-HT2C antagonists are observed during basal conditions, this receptor is only mediating an effect on serotonin neurotransmission in the presence of an SSRI. It is also of interest that along with the recent finding that 5-HT2C receptors are located on GABA-ergic cells in the raphe (Serrats et al, 2005), a number of investigators had proposed theories involving the manipulation of dopamine (DA) (Alex et al, 2005; Eberle-Wang et al, 1997; Stanford and Lacey, 1996; Giorgetti and Tecott, 2004; Di Giovanni et al, 2001; Pozzi et al, 2002), and noradrenaline (Gobert et al, 2000; Millan et al, 2005) by 5-HT2C receptors localized on GABA-ergic neurons in the VTA, SN, and LC and some terminal areas.

5-HT2C receptors are described to be located postsynaptically to the serotonergic neuron, and as such, the mechanism through which the augmentation of SSRI responses occurs involves alternative transmitter systems. Using microdialysis coupled to HPLC, we evaluated which neurotransmitter systems and receptors are involved in the augmentation strategy.

MATERIALS AND METHODS

Animals

Male albino rats of a Wistar-derived strain (285–320 g; Harlan, Zeist, The Netherlands) were used for the experiments. After surgery, rats were housed individually in plastic cages (35 × 35 × 40 cm3), and had free access to food and water. Animals were kept on a 12 h light schedule (light on 0700 hours). The experiments were concordant with the declarations of Helsinki and were approved by the animal care committee of the faculty of mathematics and natural science of the University of Groningen, The Netherlands.

Drugs

Citalopram hydrobromide, phaclofen, and SB242084 dihydrochloride were obtained from H Lundbeck A/S (Copenhagen, Denmark). Prazosin hydrochloride and DNQX were purchased from Sigma-Aldrich Chemie BV, Zwijndrecht, The Netherlands. Citalopram was dissolve in saline, whereas SB242084 dihydrochloride and prazosin were first prepared in 20% v/v solutol solution and further diluted in saline or Ringer solution depending on whether they were required for systemic injections or local infusions. DNQX and phaclofen were dissolved in Ringer solution for infusions.

Surgery

Microdialysis of extracellular serotonin levels was performed using I-shaped microdialysis probes with a polyacrylonitrile/sodium methyl sulfonate copolymer dialysis fiber (Brainlink, Groningen, The Netherlands). The dialysis probe was stereotactically implanted under the following conditions: isofluorane 2%, N2O 300 ml/min, and O2 300 ml/min. Microdialysis probes were implanted at AP: −0.53, ML: +0.48, VD: −0.80 for hippocampus (4 mm dialysing membrane) and intra aural: +0.12, ML: +0.14, VD: −0.90 (4 mm dialysing membrane) at a 10° angle for raphe nuclei, according to coordinates from bregma (Paxinos and Watson, 1986). The microdialysis probes were permanently fixed to the skull using stainless steel screws and methylacrylic cement. Animals were allowed to recover 18–24 h before microdialysis experiments commenced.

Microdialysis Experiments

Rats were allowed to recover for at least 24 h. Probes were perfused with artificial cerebrospinal fluid containing 147 mM NaCl, 3.0 mM KCl, 1.2 mM CaCl2, and 1.0 mM MgCl2, at a flow rate of 1.5 μl/min by TSE Univentor 802 syringe pump (Technical and Scientific Equipment (TSE), Bad Homburg, Germany). Microdialysis samples were collected every 15 min in HPLC vials containing 32.5 μl 0.02 M acetic acid for monamine analysis, and 7.5 μl for GABA analysis. For drug infusion regimes, the tubing was momentarily disconnected and filled with required drug, and then reattached ensuring no air was present in the tube. The collected samples were stored in a freezer at −80°C.

Monoamine Analysis

Serotonin, norepinephrine (NE), and DA concentrations were determined using HPLC coupled with electrochemical detection. Microdialysis fractions were injected via a Gilson 223 XL autoinjector (Gilson, Villiers Le Bel, France) onto respective columns for 5-HT or NE/DA analysis.

For 5-HT determination, 20 μl of the microdialysis sample was injected onto a 100 × 2.0 mm C18 Hypersil 3 μm column (Bester, Amstelveen, The Netherlands) and separated with a mobile phase consisting of 4.1 g/l sodium acetate, 500 mg/l Na2-EDTA, 50 mg/l heptane sulfonic acid, 4.5% methanol v/v, and 30 μl/l of triethylamine, pH 4.75 at a flow rate of 0.4 ml/min by Shimadzu LC-10 AD pumps (Shimadzu, ‘s Hertogenbosch, The Netherlands). 5-HT was detected amperometrically at a glassy carbon electrode at 500 mV vs Ag/AgCl (Antec Leyden, Leiden, The Netherlands). The detection limit was 0.5 fmol 5-HT per 20 μl sample (signal-to-noise ratio 3).

For the determination of NE and DA concentrations, 20 μl microdialysate fractions were injected onto a 150 × 2.1 mm2 C18 Hypersil Keystone 3 μm BDS column (Bester, Amstelveen, The Netherlands) and separated with a mobile phase consisting of 4.1 g/l sodium acetate, 150 mg/l Na2-EDTA, 150 mg/l octane sulfonic acid, and 2.5% methanol v/v, pH 4.1 at a flow rate of 0.35 ml/min by Shimadzu LC-10 AD pumps (Shimadzu, ‘s Hertogenbosch, The Netherlands). NE and DA were detected amperometrically at a glassy carbon electrode at 500 mV vs Ag/AgCl (Antec Leyden, Leiden, The Netherlands). The detection limit was 0.5 fmol NE per 20 μl sample (signal-to-noise ratio 3).

GABA Analysis

GABA concentrations in the dialysates were determined off-line by precolumn derivatization with o-phtaldialdehyde/mercaptoethanol reagent, and separation by reverse-phase HPLC on a Supercosil LC-18-DB column (Rea et al, 2005). Samples were derivatized as follows, based on the derivitization by Lindroth and Mopper (1979). One hundred milligrams o-phtaldialdehyde was dissolved in 2 ml methanol and added to 200 ml 0.5 mol/l NaHCO3 (pH adjusted to 9.5 with NaOH), containing 20 μl 2-mercaptoethanol. The reagent was freshly prepared daily.

Thirty microliters microdialysate samples were derivatized with 50 μl o-phtaldialdehyde/mercaptoethanol reagent, mixed, and allowed to react for 2 min. Fifty microliters of the reaction mixture was then injected by a Gilson 231 XL sampling injector (Gilson, Villiers Le Bel, France) onto the HPLC apparatus. The mobile phase consisted of 30% methanol, 70 mM di-sodium hydrogen phosphate, 400 μM EDTA, and 0.15% tetrahydrofluoran, and orthophosphoric acid was added dropwise until a pH of 5.25 was obtained. Fluorescent detection was performed off-column using a JASCO FP-1520 detector (excitation λ=350 nm, emission λ=450 nm).

Data Presentation and Statistics

Four consecutive microdialysis samples with less then 20% variation were taken as control and set at 100%. Statistical analysis was performed using Sigmastat for Windows (Jandel Corporation). Treatments were compared using two-way ANOVA for repeated measurements, followed by Student's Newman Keuls post hoc analysis. Effect were compared vs baseline using one-way ANOVA for repeated measurements on ranks, followed by Dunnet's test. Level of significance was set at p<0.05.

RESULTS

Baseline levels in dialysates from hippocampus samples were determined as 5.26±0.49 fmol/sample for 5-HT (n=63), 6.54±0.52 fmol/sample for NE (n=18), and 604.99±41.21 fmol/sample for GABA, respectively (n=17). In raphe samples, baseline levels were determined as 33.93±4.58 fmol/sample for 5-HT, 11.39±1.04 fmol/sample for DA (n=16), and 15.20±1.94 fmol/sample for NE (n=17), respectively. Basal dialysates were not corrected for in vitro recovery.

In order to examine the augmentation of the SSRI-induced 5-HT release by the 5-HT2C antagonist, we performed a number of combination studies at the level of the hippocampus and raphe nuclei. Experiments involving the local infusion of 10 μM citalopram in hippocampus, with concomitant administration of the 5-HT2C antagonist (0.4 mg/kg SB242084 s.c.) caused no change in basal 5-HT levels as compared to citalopram infusion with systemic vehicle administration (F(1, 10)=0.355, p=0.564) (Figure 1). Similarly, no effect was observed with the co-infusion of 100 nM (F(1, 8)=0.977, p=0.352) or 1000 nM (F(1, 8)=0.620, p=0.454) SB242084 with 10 μM citalopram infusion, respectively, as compared to citalopram infusion alone (Figure 2). The absolute value of citalopram infusion in hippocampus was determined as 52.09±8.39 fmol/sample for 5-HT.

Figure 1
figure 1

The effect of systemic co-administration of 0.4 mg/kg SB242084 with 10 μM citalopram infusion on 5-HT levels in ventral hippocampus. The horizontal bar represents the period of citalopram infusion. SB242084 and vehicle were administered systemically at t=0. Data are expressed as percentage of basal levels±SEM (n=6).

Full size image
Figure 2
figure 2

The effect of co-infusion of 100 and 1000 nM SB242084 with 10 μM citalopram infusion on 5-HT levels in ventral hippocampus. The horizontal bars represent the period of infusion of citalopram and SB242084, respectively. Data are expressed as percentage of basal levels±SEM (n=5).

Full size image

The local administration of 1 μM SB242084 in the raphe, and the hippocampus, in combination with systemic citalopram administration produced different responses in hippocampal 5-HT release (Figure 3). It was determined that there was no augmentation of the citalopram-induced response in hippocampus when SB242084 was locally infused in the raphe (F(1, 13)=0.012, p=0.915), but there was a significant increase in 5-HT levels when the SB242084 was locally infused in the hippocampus (F(1, 13)=12.087, p=0.004). Systemic citalopram administration alone significantly increased extracellular 5-HT as compared to pre-drug administration levels (F(1, 13)=11.17, p=0.015).

Figure 3
figure 3

Time course of the effect of co-infusion of 1 μM SB242084 in RN and HC, respectively, with 3.0 mg/kg citalopram s.c. on 5-HT levels in ventral hippocampus. The horizontal bar represents the period of infusion of SB242084. Citalopram was administered systemically at t=90. Data are expressed as percentage of basal levels±SEM (n=5–9). *Represents significance (p<0.05) vs citalopram.

Full size image

The co-administration of systemic SB242084 with systemic citalopram (Figure 4) increased 5-HT levels in hippocampus to almost 800% of basal levels (F(1, 8)=9.19, p=0.013). No effects were observed on extracellular NE levels in hippocampus (F(1, 14)=0.227, p=0.641) or raphe nuclei (F(1, 16)=0.989, p=0.335) with the co-administration of citalopram (3.0 mg/kg s.c.) and SB242084 (0.4 mg/kg s.c.), vs citalopram administration alone (data not shown). However, the systemic administration of the α-1 antagonist, prazosin (0.4 mg/kg), prevented the 5-HT2C-mediated augmentation in 5-HT levels in hippocampus (F(1, 14)=0.227, p=0.641).

Figure 4
figure 4

Reversal of the augmenting effects of the combination of citalopram (3.0 mg/kg s.c.) with SB242084 (0.4 mg/kg s.c.) on 5-HT levels in ventral hippocampus by prazosin (0.4 mg/kg s.c.). Citalopram, prazosin, and SB242084 were administered systemically at t=0. Data are expressed as percentage of basal levels±SEM (n=5–9). *Represents significance (p<0.05) vs citalopram.

Full size image

No effect on extracellular DA release was observed in the raphe nuclei with systemic citalopram administration (3.0 mg/kg) alone (F(1, 12)=0.334, p=0.574) or in combination with systemic SB242084 (0.4 mg/kg) (F(1, 6)=0.641, p=0.339) (data not shown).

To examine the potential involvement of the glutamate system in the 5-HT2C receptor-mediated augmentation, 5-HT levels were monitored while the AMPA/kainate antagonist, DNQX (100 μM), was infused during the administration of SB242084 (0.4 mg/kg s.c.) with citalopram (3.0 mg/kg s.c.) (Figure 5). The infusion of DNQX did not modify the increase in 5-HT observed with the combination of SB242084 and citalopram (F(1, 10)=0.0292, p=0.868).

Figure 5
figure 5

The investigation of the effect of 3.0 mg/kg citalopram s.c. and 0.4 mg/kg SB242084 s.c. combination, with 100 μM DNQX preinfusion on 5-HT levels in ventral hippocampus. Citalopram and SB242084 were administered systemically at t=90. The horizontal bar represents the period of infusion of DNQX, or vehicle. Data are expressed as percentage of basal levels±SEM (n=5–9). *Represents significance (p<0.05) vs citalopram.

Full size image

The local infusion of the GABAB antagonist phaclofen was also shown to augment the effect of the SSRI citalopram on 5-HT in hippocampus (Figure 6). A 50 μM phaclofen infusion augmented the citalopram-induced increase in 5-HT in a manner similar to the augmentation seen with the 5-HT2C antagonist SB242084 (F(1, 12)=12.813, p=0.004). Infusion of the GABAA antagonist bicucculine (50 μM) significantly increased basal 5-HT levels (F(1, 9)=11.97, p=0.007) in hippocampus (Figure 7), although the combination of systemic citalopram (3.0 mg/kg) produced no further effect on 5-HT levels (F(1, 13)=0.158, p=0.698). The combination of SB242084 (0.4 mg/kg s.c.) with citalopram (3.0 mg/kg s.c.) resulted in a slight, yet significant decrease in extracellular GABA in hippocampus (F(1, 18)=5.053, p=0.037), when compared to citalopram alone (Figure 8).

Figure 6
figure 6

Time course of the effect of 3.0 mg/kg citalopram s.c. and 50 μM phaclofen on 5-HT levels in ventral hippocampus. Citalopram was administered systemically at t=90. The horizontal bar represents the period of infusion of phaclofen or vehicle. Data are expressed as percentage of basal levels±SEM (n=5–9). *Represents significance (p<0.05) vs citalopram.

Full size image
Figure 7
figure 7

Time course of the effect of 3.0 mg/kg citalopram s.c. and 50 μM bicucculine on 5-HT levels in ventral hippocampus. Citalopram was administered systemically at t=90. The horizontal bar represents the period of infusion of bicucculine or vehicle. Data are expressed as percentage of basal levels±SEM (n=5–9). *Represents significance (p<0.05) vs citalopram.

Full size image
Figure 8
figure 8

Investigation of the effect of 3.0 mg/kg citalopram s.c. and 0.4 mg/kg SB242084 s.c. co-administration on GABA levels in ventral hippocampus. Citalopram and SB242084 were administered systemically at t=0. Data are expressed as percentage of basal levels±SEM (n=6).

Full size image

DISCUSSION

Classical augmentation of SSRI effects by 5-HT1A and 5-HT1B antagonists is related to serotonin autoreceptor blockade, which, in turn facilitate serotonin release. Interestingly, the augmentation of 5-HT with the combination of citalopram and 5-HT2C receptor antagonists was of similar magnitude as previously reported for compounds that block 5-HT1A and 5-HT1B autoreceptors (Cremers et al, 2000a, 2000b, 2001, 2004). However, the current mechanism of augmentation is more complex, as 5-HT2C receptors are not located on serotonergic neurons (see below). Multiple neurotransmitter systems are likely to be involved in the effects of 5-HT2C antagonists, eventually leading to the augmentation of SSRI effects.

Distribution of 5-HT2C receptors in the brain is abundant. The highest densities are found in the chorioid plexus, cerebral cortex, hippocampus, striatum, and substantia nigra (Barnes and Sharp, 1999; Hoyer et al, 2002). Evidence that 5-HT2C receptors might be involved in the regulation of serotonin release is scarce. In fact, a number of researchers have reported slight attenuations in basal NE and DA levels with no effect on basal 5-HT (Millan et al, 1998, 2005; Gobert et al, 2000).

Similarly, no effect has been found when potassium-evoked serotonin release from brain homogenate was studied in vitro in the presence of 5-HT2C antagonist ketanserin (Bonanno et al, 1986; Maura et al, 1986). These data were confirmed by the absence of effects when both the SSRI and the 5-HT2C antagonist were applied locally in the hippocampus in the present study. Interestingly, these experiments indicate that the augmentation originates from blockade of 5-HT2C receptors in terminal areas, but does not comprise a mere local interaction, as the SSRI has to be administered systemically. Thus, the SSRI activates systems outside the vicinity of the microdialysis probe that are required for 5-HT2C antagonists to become relevant.

The 5-HT2C receptor is located as a heteroceptor on GABA-ergic neurons (Serrats et al, 2005), and may also be located on glutamatergic, noradrenergic, or dopaminergic neurons (see below). Theoretically, all these systems might be involved in the mechanism through which 5-HT2C antagonists augment the effects of SSRIs on serotonin levels in the brain.

Norepinephrine

Several studies have shown that NE release is increased upon systemic administration of 5-HT2C antagonists (Millan et al, 2005, 1998; Gobert et al, 2000). Additional studies show that the manipulation of NE release affects serotonin release through adrenoceptors which are postulated to be present in the dorsal raphe nucleus (Hopwood and Stamford, 2001; Bortolozzi and Artigas, 2003; Pudovkina et al, 2003) and possibly terminal areas (Rouquier et al, 1994; Amargos-Bosch et al, 2003; Linner et al, 2004; Koch et al, 2004; Weikop et al, 2004; Gobert et al, 1997; De Boer et al, 1996). However, if NE would be involved in the present mechanism, it is to be expected that NE levels would be enhanced when the SSRI was administered in conjunction with the 5-HT2C antagonist. As no effects on NE levels were observed in hippocampus or in raphe nuclei, the direct involvement of NE in the mechanism of augmentation is less likely. It is possible that the co-administration of these compounds is affecting the noradrenergic system at regions other than those in this study, which may indirectly be mediating effects on 5-HT release.

Interestingly, after investigation of a series of receptor-specific antagonists, we determined that the α1-adrenoceptor antagonist prazosin completely abolished the augmentation by 5-HT2C receptor antagonists, implying that a norepinephrinergic tone on the serotonergic neurons is necessary but not directly involved in the augmentation at the level of the hippocampus.

Dopamine

Interactions between the serotonergic system and the dopaminergic system have also been described (Martin-Ruiz et al, 2001b; Ferre et al, 1994). Antagonism of 5-HT2C receptors in the prefrontal cortex has been shown to elevate DA and NE (Alex et al, 2005; Giorgetti and Tecott, 2004; Lucas and Spampinato, 2000; Gobert et al, 2000; Millan et al, 1998; De Deurwaerdere et al, 2004; Di Matteo et al, 1999, 2000). As dopaminergic innervation of the hippocampus is low, no augmentation of DA levels were detectable. It has been previously shown that DA D1-specific compounds had no effect on the turnover of 5-HT (Lappalainen et al, 1991), and that the DA D2 receptor agonist increased local 5-HT release in RN (Ferre S and Artigas F, 1993), although this effect may be mediated at D2 receptors localized outside the raphe nuclei (Martin-Ruiz et al, 2001b). However, no effect was observed on extracellular DA levels in the raphe nuclei after a single administration, or the combination of the 5-HT2C antagonist with citalopram, which renders an interaction with DA neurons less likely.

Glutamate

A study by Martin-Ruiz et al (2001a) has shown that modulation of glutamatergic neurotransmission by 5-HT2 receptors might also be involved in control of serotonin release. Evidence suggests that the 5-HT2A/2C receptor agonist, 1-[2,5-dimethoxy-4-iodophenyl]-2-aminopropane, increases glutamate levels in terminal areas (Scruggs et al, 2003), and it has been reported that enhanced glutamatergic release facilitates serotonin release through activation of kainate receptors which are located on the serotonergic neuron (Martin-Ruiz et al, 2001a). The administration of SSRIs would increase extracellular 5-HT, and thus activate both 5-HT2A and 5-HT2C receptors on glutamatergic neurons. The application of SB242084 would block 5-HT2C receptor-mediated regulation of glutamate release, leaving the response to be mediated by 5-HT2A receptors, which have been shown to be located on glutamatergic neurons and increase glutamate release after stimulation (Martin-Ruiz et al, 2001a). However, if this were the underlying mechanism, it should have been possible to prevent the augmentation of serotonin release by glutamate with the local administration of AMPA/kainate antagonist DNQX. As no attenuation of augmentation was observed, glutamatergic involvement is also not likely. In addition, evaluation of the effects of citalopram with and without SB242084 on glutamate levels did not show any effects (data not included). Of course, the limited relevance of glutamate levels as measured with microdialysis should be taken into account. Various authors have expressed serious doubts whether glutamate sampled by microdialysis is of synaptic origin (Timmerman and Westerink, 1997). One key finding to support this is that very few researchers have reported a decrease in glutamate levels with the administration of tetrodotoxin, whereas there is a dramatic decrease in levels of neurotransmitters such as serotonin, noradrenaline, or DA.

GABA

Several studies have shown that 5-HT2C receptors facilitate GABA release (Liu et al, 2000; Hajos et al, 2003; Abi-Saab et al, 1999), and recently 5-HT2C receptors have been shown to be localized on GABA cells in the raphe nucleus (Serrats et al, 2005). Arguably, a diminished GABA-ergic tone owing to 5-HT2C receptor antagonists in this area may contribute to the increased effect of SSRIs. Several studies have shown that inhibition of 5-HT release by GABA is mediated by GABAA and GABAB receptors (Pei et al, 1989; Tao and Auerbach, 2000, 2003). This would indicate that there is a reciprocal relationship between the activity of GABA-ergic and serotonergic neurons. However, to date, there are no studies localizing the 5-HT2C receptors to GABA-ergic neurons in terminal regions.

The present study clearly shows a prominent interaction between the serotonergic and the GABA-ergic system. In the presence of an SSRI, the extent of 5-HT release is governed by a GABAB receptor-mediated feedback control. This effect was not apparent under pre-SSRI administration conditions. In contrast, a GABAA receptor-induced feedback was present under basal conditions, but not in the presence of the SSRI. Although the involvement of the GABAB receptor in the augmentation of the serotonin response by citalopram is apparent, a pronounced increase of GABA levels in hippocampus upon administration of the SSRI was not observed. However, a slight, yet significant decrease in GABA levels was reported with the combination of citalopram and SB242084 as compared to citalopram alone. The mechanism of the 5-HT2C-mediated augmentation of the SSRI response on 5-HT levels may be explained by this decrease in GABA release, which may in turn affect the GABA tone on terminal serotonin neurons. As the GABA receptors have an inhibitory effect on 5-HT release, the co-administration of SB242084 may serve to reverse these effects, thus allowing a further increase in extracellular 5-HT.

Conclusion

In the present study, we have evaluated several mechanisms through which 5-HT2C antagonists might augment the effects of SSRIs on serotonin release. Whereas the involvement of dopaminergic and glutamatergic systems does not seem likely, NE and GABA-ergic systems are crucial for the observed effects. A basal stimulatory input from the NE system is required for the augmentation, although it is not thought to underlie the mechanism under evaluation. Enhanced GABA-ergic feedback through activation of terminal GABAB receptors was found in the presence of systemic SSRI treatment. The absence of pronounced effects of the SSRI on GABA levels in hippocampus, remains to be explained, but might be related to the differential origin of GABA levels as quantified by microdialysis. However, the decrease in GABA levels with the combination of SB242084 and citalopram, as compared to citalopram alone, as well as the fact that phaclofen augmented the citalopram-induced increase in 5-HT, indicates that the GABA system is indeed involved in the mechanism of the augmentation.