Introduction

Difficult-to-treat depressive disorders are extremely debilitating for patients and present a formidable challenge for psychiatrists and researchers alike. The depressive phase of bipolar disorder (bipolar depression) has been especially difficult to treat. Bipolar disorder is characterized by manic, major depressive, and/or mixed episodes. Most controlled clinical research on the treatment of bipolar disorder has focused on the manic phase, with relatively little research on treatment of the depressive phase. Combination and augmentation strategies have emerged as a means to combat bipolar depression (for review, see Keck et al. 2003; Muller-Oerlinghausen et al. 2002) as well as treatment-resistant depression (for review, see Fava 2000; Hirschfeld et al. 2002; Nelson 2003; Shelton 2003) and psychotic depression (for review, see Schatzberg 2003). The combination of an atypical antipsychotic (e.g., olanzapine) and a selective serotonin reuptake inhibitor (SSRI; e.g., fluoxetine) is beneficial for treatment-resistant depression (Corya et al. 2003; Shelton et al. 2001) and psychotic depression (Matthews et al. 2002; Williamson et al. 2001), and the olanzapine/fluoxetine combination (Symbyax) is currently the only approved treatment for the depressive phase of bipolar disorder (Shi et al. 2004; Tohen et al. 2003). While there is clear clinical benefit from this combination, the precise neural mechanisms responsible for its efficacy are not clearly understood. Therefore, it is important to investigate the mechanism of action of this combination in order to not only better understand the etiology of the clinical syndromes, but also to eventually facilitate the development of improved drugs to treat them.

Analysis of potential brain mechanisms underlying the therapeutic response to the olanzapine/fluoxetine combination has focused on extracellular monoamine levels in the prefrontal cortex (PFC) following acute administration of the drugs. Results of these experiments revealed that acute coadministration of olanzapine and fluoxetine leads to significantly higher extracellular levels of dopamine and norepinephrine in rat PFC than after either drug alone (Koch et al. 2004; Zhang et al. 2000). Recently, we reported the impact of both acute (24 h) and chronic (3 weeks) fluoxetine pretreatment on the effect of acute olanzapine administration on the electrophysiological characteristics of locus coeruleus (LC) neurons (Seager et al. 2004). The LC is a nucleus that provides over 90% of the norepinephrine for the neuraxis and innervates the PFC extensively. The data revealed a significant interaction between fluoxetine and olanzapine, with olanzapine having a much greater excitatory effect on LC firing rate when administered in the presence of both acute and chronic fluoxetine. Interestingly, this effect was more pronounced in the chronic fluoxetine condition. The olanzapine/fluoxetine combination not only increased the rate of firing of these neurons, but it also increased the amount of burst firing. This increase in burst firing could lead to a further enhancement of transmitter release (Florin-Lechner et al. 1996; see also Floresco et al. 2003) and possibly the release of peptides colocalized in LC terminals (Grenhoff et al. 1993; Xu et al. 1998).

Since the olanzapine/fluoxetine combination is given to patients using a chronic dosing regimen, and full efficacy is only achieved after several weeks of treatment, it is important to observe the neurobiological effects of chronic coadministration of the two drugs. Therefore, the current experiment sought to model the use of the two drugs clinically by administering both drugs chronically and observing their effects on LC electrophysiology.

Materials and methods

Animals

Male Sprague–Dawley rats were obtained from Harlan Industries (Indianapolis, IN) and weighed 225–310 g at the time of electrophysiological recording. Animals were group housed upon arrival, and food and water were available ad libitum. A 12:12 h light/dark cycle was maintained in the colony room. All procedures were carried out in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals and approved by the Eli Lilly Animal Care and Use Committee.

Drug solutions

Fluoxetine hydrochloride (Eli Lilly and Company, Indianapolis, IN) was dissolved in distilled water. Olanzapine (Eli Lilly and Company) was dissolved in a combination of 5% lactic acid, sodium hydroxide, and distilled water (final pH=5.8).

Pump implantation and injections

Rats were anesthetized with isoflurane (5%), and an osmotic pump (Alzet model 2ML1, Durect Corp., Cupertino, CA) was implanted subcutaneously. Rats received either olanzapine (1 or 10-mg/kg/day) or saline for 3 weeks (delivery rate=10 μl/h). Rats in both the olanzapine and saline conditions were reimplanted with new pumps 1 and 2 weeks after the first implant procedure. The amount of olanzapine in each pump was based on the average weight for that week, ensuring that either 1 or 10 mg/kg/day was administered throughout the 3-week period. Each rat also received daily i.p. injections of either saline or fluoxetine (10-mg/kg) during the 3-week period. Due to the longer half-life of fluoxetine compared to olanzapine, the two drugs were given via different routes of administration: olanzapine—s.c., fluoxetine—i.p. Previous studies in our laboratory have shown that daily i.p. injection of fluoxetine produces similar electrophysiological results as s.c. pump administration of fluoxetine (see Czachura and Rasmussen 2000; Seager et al. 2004).

LC recordings

Rats were anesthetized with chloral hydrate (400-mg/kg, i.p.) and positioned in a stereotaxic apparatus. A 24-gauge cannula was inserted into the lateral tail vein for i.v. delivery of supplemental anesthesia. A homeothermic blanket was used to maintain body temperature above 37°C. An incision was made in the scalp and the periosteum moved aside. A 3-mm diameter hole was drilled into the skull just posterior to lambda and lateral to the midline. Electrodes were fabricated from 2.0-mm OD capillary tubing (Radnoti, Monrovia, CA) on a Narishige (Tokyo, Japan) electrode puller (model PE-2). Electrodes were filled with a 2 M NaCl solution and broken back to impedances of 2.0–3.0 MΩ. Coordinates for the LC (∼1.1 mm lateral to the midline and 0.8 mm posterior to the interaural line) were based upon those of Paxinos and Watson (1998). Electrodes were lowered into the LC with a Burleigh 6000 controller (EXFO/Burleigh, Victor, NY) until a single unit was isolated. Recordings were made unilaterally, and electrode tracks were 0.1–0.2 mm apart. Characteristics of LC neurons have been described previously (Graham and Aghajanian 1971), and the criteria for inclusion in the current study were a regular firing rate with a positive–negative action potential of long duration (∼2 ms) and a clear burst of action potentials followed by a period of quiescence in response to a pinch of the contralateral hind paw. Previous work from our laboratory has histologically confirmed the localization of neurons with these electrophysiological characteristics to the LC (Melia et al. 1992). Sampling of neurons began 2 h after the last i.p. injection of saline or fluoxetine with the osmotic pumps in place. Only data from cells where stable recordings were obtained (>60 s) were used for analysis.

Activity from the electrode was monitored via an oscilloscope and audio monitor, band-pass filtered (300–3,000 Hz; Dagan model 2400, Minneapolis, MN), and passed through a Micro1401 data acquisition unit connected to a PC running Spike2 software (Cambridge Electronic Design, Cambridge, England). A software window discriminator was set to capture only the largest amplitude unit in the recording.

Tissue and plasma preparation and analysis

Separate groups of rats underwent the same dosing regimen as above and were sacrificed via decapitation. Trunk blood (collected on heparin) and a sample of frontal cortex brain tissue were placed in siliconized microcentrifuge tubes on ice. The blood samples were centrifuged, and an aliquot of plasma was transferred to a new set of tubes. Brain and plasma samples were homogenized in four volumes of acetonitrile containing 0.1% formic acid using an ultrasonic probe. Homogenates were centrifuged at 14,000 rpm for 16 min in a benchtop centrifuge (Eppendorf model 5417R, Hamburg, Germany), and the supernatant was assayed for the concentration of administered drugs.

Aliquots of supernatant from both tissue and plasma samples were diluted with water to achieve an acetonitrile concentration less than that used in the mobile phase and injected by an autosampler onto an HPLC. Fluoxetine and olanzapine were assayed using isocratic conditions and a mobile phase containing 40 or 10% acetonitrile in water containing 0.1% formate, respectively. The column employed contained 3.5-μm Zorbax SB-C18 packing material in a 2.1×50-mm format (part # 871700-902, Agilent Technologies, Wilmington, DE). Fluoxetine and olanzapine eluting from the column were measured by mass spectroscopy monitoring a mass-to-charge ratio of 310.1 and 313.1, respectively. An Agilent model 1946 single-quad mass spectrometer (Agilent Technologies) was set to selectively monitor the mass-to-charge ratio of the singly protonated fluoxetine and olanzapine. Plasma olanzapine was measured using a triple-quad mass spectrometer (PE SCIEX API 3000, Foster City, CA) set to selectively monitor the ion transition from parent to daughter ion (311/252.9). Each assay was calibrated by using a standard curve generated by extracting a series of brain tissue and plasma samples from nontreated rats to which known quantities of analyte had been added.

Experimental design and statistical analysis

The final design of the experiment included a total of six groups. Rats received one of two levels of fluoxetine (10-mg/kg fluoxetine or saline) and one of three levels of olanzapine (1-mg/kg olanzapine, 10-mg/kg olanzapine, or saline), making for a 3×2 design. Firing rates were analyzed with a 3×2 between-groups analysis of variance (ANOVA) with fluoxetine and olanzapine as the between-group factors. Post-hoc tests were conducted with the Tukey test. Analysis of burst firing was accomplished with a modified Spike2 script file. The start of a burst was defined as the occurrence of two spikes with an interspike interval of <80 ms, and the end of a burst was defined by the next interval >160 ms (Dawe et al. 2001; Grace and Bunney 1984; Seager et al. 2004). Two parameters of burst firing were calculated for each cell: the percentage of spikes occurring in bursts and bursts per second. Each burst parameter was then analyzed with a 3×2 between-groups ANOVA followed by post-hoc Tukey tests. Brain and plasma levels of olanzapine and fluoxetine were also analyzed with 3×2 between-groups ANOVAs. An alpha level of 0.05 was used for all statistical tests.

Results

A total of 598 LC cells were recorded from 40 rats.

Firing rate

Analysis of firing rates revealed significant main effects of fluoxetine, F(1,592)=10.88, p=0.001, and olanzapine, F(2,592)=26.63, p<0.0001, as well as an interaction of the two factors, F(2,592)=4.95, p=0.007. Chronic administration of fluoxetine produced a significant decrease in baseline firing rate, F(1,165)=18.67, p<0.0001 (see Fig. 1). When fluoxetine was combined with chronic administration of olanzapine, firing rates were still significantly reduced in rats in the 1-mg/kg olanzapine/fluoxetine condition when compared to rats in the 1-mg/kg olanzapine/saline condition, F(1,183)=5.96, p=0.02, but not in rats in the 10-mg/kg olanzapine/fluoxetine condition when compared to rats in the 10-mg/kg olanzapine/saline condition (see Fig. 1). In fact, olanzapine produced significant changes in firing rate in both fluoxetine- and saline-treated animals, F‘s>3.52, p‘s<0.04. Post-hoc Tukey tests revealed that firing rates were significantly higher in the 10-mg/kg olanzapine/saline group than in the saline/saline group and were significantly higher in the 10-mg/kg olanzapine/fluoxetine group than in the saline/fluoxetine and 1-mg/kg olanzapine/fluoxetine groups (see Fig. 1).

Fig. 1
figure 1

Mean (±SEM) firing rate of LC neurons as a function of fluoxetine treatment and olanzapine dose. Chronic administration of fluoxetine produced a significant reduction in firing rate of LC neurons, and chronic olanzapine produced an increase in firing rate in both fluoxetine- and saline-treated groups. *Significantly different from saline/saline group; +significantly different from 1-mg/kg olanzapine/saline group; #significantly different from saline/fluoxetine group; ^significantly different from 1-mg/kg olanzapine/fluoxetine group. The numbers of cells (and rats) for each of the treatment groups (from left to right) are as follows: 79(6), 95(6), 90(6), 88(6), 90(6), 156(10)

Full size image

Burst firing

Although spontaneous burst firing of LC neurons in untreated, anesthetized rats is relatively infrequent (see saline/saline group in Figs. 2, 3), it is measurable and amenable to pharmacological manipulation (see below).

Fig. 2
figure 2

Mean (±SEM) percentage of spikes occurring in bursts of LC neurons as a function of fluoxetine treatment and olanzapine dose. Chronic administration of fluoxetine produced a significant reduction in the percentage of spikes occurring in bursts of LC neurons, and chronic olanzapine produced an increase in the percentage of spikes occurring in bursts in both fluoxetine- and saline-treated groups. *Significantly different from saline/saline group; +significantly different from 1-mg/kg olanzapine/saline group; #significantly different from saline/fluoxetine group; ^significantly different from 1-mg/kg olanzapine/fluoxetine group. The numbers of cells and rats contributing to the figure are the same as in Fig. 1

Full size image
Fig. 3
figure 3

Mean (±SEM) bursts per second of LC neurons as a function of fluoxetine treatment and olanzapine dose. Chronic administration of fluoxetine produced a significant reduction in bursts per second of LC neurons in the 1-mg/kg olanzapine group, and chronic olanzapine produced an increase in bursts per second in both fluoxetine- and saline-treated groups. *Significantly different from saline/saline group; +significantly different from 1-mg/kg olanzapine/saline group; #significantly different from saline/fluoxetine group; ^significantly different from 1-mg/kg olanzapine/fluoxetine group. The numbers of cells and rats contributing to the figure are the same as in Fig. 1

Full size image

Percentage of spikes in bursts

Analysis of the percentage of spikes occurring in bursts revealed significant main effects of fluoxetine, F(1,592)=4.25, p=0.04, and olanzapine, F(2,592)=12.86, p<0.0001. Fluoxetine administration led to a significantly lower percentage of spikes occurring in bursts in the saline/fluoxetine and 1-mg/kg olanzapine/fluoxetine groups, F‘s>4.07, p‘s<0.05, but not in the 10-mg/kg olanzapine/fluoxetine group (see Fig. 2). The percentage of spikes occurring in bursts was significantly greater in the 10-mg/kg olanzapine group than in the saline and 1-mg/kg olanzapine groups for both fluoxetine- and saline-treated animals (see Fig. 2).

Bursts per second

Analysis of bursts per second revealed significant main effects of fluoxetine, F(1,592)=5.34, p=0.02, and olanzapine, F(2,592)=14.15, p<0.0001. Fluoxetine administration led to significantly fewer bursts per second in the 1-mg/kg olanzapine/fluoxetine group, F(1,183)=5.21, p=0.02, but not in the saline/fluoxetine or 10-mg/kg olanzapine/fluoxetine group (see Fig. 3). Bursts per second were significantly greater in the 10-mg/kg olanzapine/saline group than in the saline/saline and 1-mg/kg olanzapine/saline groups, and bursts per second were also significantly greater in the 10-mg/kg olanzapine/fluoxetine group than in the saline/fluoxetine and 1-mg/kg olanzapine/fluoxetine groups (see Fig. 3).

Brain and plasma drug levels

Table 1 contains olanzapine and fluoxetine brain and plasma levels for each treatment condition. Analysis of brain olanzapine levels revealed significant main effects of fluoxetine, F(1,20)=6.10, p=0.02, and olanzapine, F(1,20)=33.83, p<0.0001, as well as an interaction of the two factors, F(1,20)=5.03, p=0.04. Further analysis revealed that olanzapine levels were significantly higher in the 10-mg/kg olanzapine/fluoxetine group than in the 10-mg/kg olanzapine/saline group, F(1,10)=5.57, p=0.04, but differences between the 1-mg/kg olanzapine/fluoxetine group and 1-mg/kg olanzapine/saline group just failed to reach statistical significance, F(1,10)=4.74, p=0.06. Additionally, olanzapine levels were significantly higher in the 10-mg/kg group than in the 1-mg/kg group for both fluoxetine- and saline-treated rats, F‘s>17.50, p‘s<0.002.

Table 1 Mean (±SEM) brain and plasma olanzapine and fluoxetine levels as a function of fluoxetine treatment and olanzapine dose
Full size table

Analysis of plasma olanzapine levels revealed significant main effects of fluoxetine, F(1,14)=8.48, p=0.01, and olanzapine, F(1,14)=42.25, p<0.0001, as well as an interaction of the two factors, F(1,14)=7.83, p=0.01. Further analysis revealed that olanzapine levels were significantly higher in the 10-mg/kg olanzapine/fluoxetine group than in the 10-mg/kg olanzapine/saline group, F(1,10)=26.84, p=0.0004, but were not different between the 1-mg/kg olanzapine/fluoxetine group and 1-mg/kg olanzapine/saline group. Additionally, olanzapine levels were significantly higher in the 10-mg/kg group than in the 1-mg/kg group for both fluoxetine- and saline-treated rats, F‘s>31.95, p‘s<0.003. Finally, there were no differences in brain and plasma fluoxetine levels between olanzapine treatment conditions.

Discussion

As observed previously, chronic administration of fluoxetine produced a significant reduction in baseline and burst firing of LC neurons (Grant and Weiss 2001; Seager et al. 2004), which is consistent with most reports looking at the effects of antidepressants on LC activity (for review, see Grant and Weiss 2001; but see Valentino et al. 1990; Haddjeri et al. 1995). Previous studies have shown that acute administration of olanzapine increases the firing rate and burst activity of LC neurons (Dawe et al. 2001; Seager et al. 2004). In the present study, chronic administration of olanzapine also produced a significant increase in the firing rate and burst firing of LC neurons, indicating that the olanzapine-induced activation of the LC is maintained through several weeks of treatment. Importantly, in the current experiment, chronic olanzapine administration blocked the inhibition of LC activity typically seen after chronic administration of an SSRI alone, and in fact, firing rates were significantly higher in the 10-mg/kg olanzapine/fluoxetine group than in both the saline/fluoxetine and saline/saline control conditions. However, unlike the findings of Seager et al. (2004) with chronic fluoxetine and acute olanzapine administration, firing patterns in the combination condition were not significantly elevated when compared to the 10-mg/kg olanzapine/saline condition. The increased firing rate of LC neurons observed here is consistent with a recent report showing that chronic olanzapine/fluoxetine coadministration can upregulate tyrosine hydroxylase in the LC (Ordway and Szebeni 2004). This enhanced LC activity could lead to greater norepinephrine release in target areas than after fluoxetine alone and the subsequent relief of depressive symptoms. This hypothesis is consistent with reports indicating that norepinephrine is crucial for the antidepressant effect of fluoxetine and other SSRIs (Cryan et al. 2001, 2004; for review, see Lucki and O’Leary 2004). The increase in burst activity observed in the current experiment could further enhance this norepinephrine release (Florin-Lechner et al. 1996; see also Floresco et al. 2003) and possibly lead to the release of peptides that are colocalized in LC terminals (e.g., galanin; Consolo et al. 1994; Grenhoff et al. 1993; Xu et al. 1998). Interestingly, it has recently been suggested that the atypical antipsychotic, clozapine, may induce the co-release of dopamine from norepinephrine-containing neurons (Devoto et al. 2003). Thus, extracellular levels of dopamine may also be elevated in the olanzapine/fluoxetine group. In vivo microdialysis experiments utilizing the current dosing regimen could test this hypothesis. While increases in norepinephrine levels could also occur with olanzapine-alone treatment, it is important to note that increases in norepinephrine levels after treatment with the combination would occur in the context of other changes caused by fluoxetine (e.g., dramatic changes in the 5-HT system; Czachura and Rasmussen, 2000). Thus, the clinical effect with the olanzapine/fluoxetine combination would be different than with olanzapine treatment alone.

Analysis of brain and plasma levels revealed that chronic coadministration with fluoxetine elevated levels of olanzapine in the 10-mg/kg group but not in the 1-mg/kg group. This elevation is not surprising given that fluoxetine is a potent inhibitor of CYP2D6 (Crewe et al. 1992), one of the pathways for olanzapine metabolism (Kassahun et al. 1997; Ring et al. 1996). Coadministration of fluoxetine and olanzapine in humans has been shown to slightly elevate peak plasma concentration (C max) and decrease clearance (CL/F) of olanzapine (Gossen et al. 2002). However, elevated levels of olanzapine were not observed in a previous experiment where rats received acute administration of olanzapine following chronic fluoxetine treatment (Seager et al. 2004) or after acute coadministration of the two drugs in rats (Zhang et al. 2000). Daily dosing in humans revealed that steady-state plasma concentrations of olanzapine are not attained until after ∼1 week of treatment and are twofold higher than after a single dose (Bergstrom et al. 2000). Therefore, the chronic nature of the olanzapine dosing regimen in the current study may have allowed the inhibitory effects of fluoxetine on olanzapine’s metabolism to be revealed. The increased brain and plasma levels of olanzapine observed in animals treated with fluoxetine could have partially contributed to the enhanced LC activity in these co-treated animals. Additional experiments using a variety of doses of olanzapine and fluoxetine would help determine the minimally effective dose of olanzapine necessary to reverse the fluoxetine-induced inhibition of LC activity. Regardless of the exact pharmacokinetic interaction between the two drugs, it appears that in patients, the net effect of a sufficient dose of olanzapine would be an increase in LC activity.

While the precise mechanism responsible for the effects of the olanzapine/fluoxetine combination on LC activity cannot be determined from this experiment, a likely candidate is activity at D2 and 5-HT2A receptors. Both typical and atypical antipsychotics have been shown to activate LC neurons (Dawe et al. 2001; Dinan and Aston-Jones 1984; Nasif et al. 2000; Ramirez and Wang 1986; Seager et al. 2004; Souto et al. 1979), and many of these compounds have D2 or both D2 and 5-HT2A antagonist properties. However, the 5-HT2A antagonist, MDL 100,907, does not affect spontaneous firing of LC neurons but does block the suppressive effect of the 5-HT1A antagonist, WAY 100,635 (Szabo and Blier 2001). Also, unlike olanzapine, when MDL 100,907 is coadministered with fluoxetine, it does not elevate PFC norepinephrine levels more than fluoxetine alone (Zhang et al. 2000). Interestingly, the SSRI/5-HT2A antagonist, YM992, which initially inhibits LC neurons via activation of autoreceptors (Szabo and Blier 2002), significantly elevates PFC extracellular norepinephrine levels (Hatanaka et al. 2000). Further work is needed to address the mechanistic issues involved in the effects of chronic coadministration of olanzapine and fluoxetine. Interestingly, recent findings suggest that the olanzapine/fluoxetine combination can upregulate tyrosine hydroxylase in the LC (Ordway and Szebeni 2004) and can influence the expression of neurotrophic molecules (Maragnoli et al. 2004), neurogenesis (Kodama et al. 2004), and the excitability of PFC neurons (Gronier and Rasmussen 2003).

One problem associated with the use of antidepressants in the treatment of bipolar depression is their potential to induce mania or hypomania or to accelerate cycling (Stoll et al. 1994; for review, see Henry and Demotes-Mainard 2003). During mania, the patient’s mood changes to an overactive state marked by feelings of irritability, elation, expansiveness, or euphoria. Symptoms associated with mania include increased motor and verbal activity, sleep loss, racing thoughts, grandiose ideas, and impulsive acts. One hypothesis is that the activity of the LC may play a role in this antidepressant-induced mania. In addition to SSRIs, other types of antidepressant treatments, including monoamine oxidase inhibitors, tricyclics, and electroconvulsive therapy, decrease firing rates in the LC (Grant and Weiss, 2001), and all have been shown to induce mania in bipolar patients (for review, see Altshuler et al. 1995). Conversely, olanzapine increases LC activity and has demonstrated efficacy in the treatment of acute bipolar mania (Tohen et al. 2000), and the olanzapine/fluoxetine combination does not induce mania in bipolar patients (Tohen et al. 2003). Thus, decreasing LC activity may be permissive to the induction of mania, and the activation of LC firing by olanzapine may both help treat mania and prevent the occurrence of antidepressant-induced mania in bipolar depressed patients.

In summary, chronic coadministration of olanzapine and fluoxetine activated norepinephrine-containing neurons of the LC, which could lead to enhanced neurotransmitter release in target areas and the amelioration of the symptoms of bipolar depression. There are multiple mechanisms potentially responsible for these effects including activity at D2 and 5-HT2A receptors, as well as activation of neurotrophic factors. This activation of the LC, in the context of changes occurring in other systems (e.g., serotonergic), may play an important role in the clinical benefits of the olanzapine/fluoxetine combination in bipolar disorder.