Introduction

Partial response and treatment resistance to antidepressant medications are common (Fava and Davidson 1996). Clinical studies of inter-individual differences in sensitivity to antidepressant drugs have uncovered some mechanisms underlying differences in responsiveness to treatment, especially relating to drug metabolism (Kirchheiner et al. 2003). However, few animal models exist to identify neurobiological differences that lead to differential reactivity to various classes of antidepressants.

Depressive disorders are accompanied by dysregulation of noradrenergic and serotonergic systems, which are thought to mediate many of the symptoms of the illness, and correction or compensation for such abnormalities may be necessary for antidepressant efficacy (Ressler and Nemeroff 2000). Indeed, most of the antidepressants in clinical practice today act on the noradrenergic or serotonergic system or both. In this study, we compared responses to the selective norepinephrine reuptake inhibitor (NRI) desipramine (DMI), or the selective serotonin reuptake inhibitor (SSRI) fluoxetine (FLU) in the forced swim test (FST) in two groups of animals that differ in emotional reactivity.

The animal model we chose assesses individual differences in novelty-seeking behavior, which are associated with significant differences in spontaneous anxiety and depression-like behavior. Measures of locomotor exploration in a novel environment were originally introduced as a tool to investigate individual differences in addiction-related behaviors (Piazza et al. 1989). It has become apparent, however, that high (HR) and low (LR) responders in the novelty test exhibit many other differences in affective behavior. Thus, HR rats show lower levels of anxiety-like behavior in the light-dark box and elevated plus-maze anxiety tests than their LR counterpart (Kabbaj et al. 2000). Interestingly, HR rats explore a novel environment more actively than LR in spite of the fact that this increased activity results in higher levels of corticosterone (CORT; Dellu et al. 1996; Kabbaj et al. 2000; Piazza et al. 1989). Our laboratory has shown that the HR-LR trait is genetically transmitted and that differences in novelty-seeking behavior are closely tied to differences in anxiety-like measures in the selectively bred animals (Stead et al. 2006).

In the present series of studies, we asked whether HR-LR differences in emotional reactivity extend to the FST, a behavioral test of antidepressant efficacy. We assessed whether HR and LR animals showed differences in the FST in their response to two classes of antidepressants, DMI and FLU. As behavioral differences in responsiveness to DMI emerged between groups, we asked whether we could identify neural correlates of these differences. Thus, we studied changes in c-fos messenger RNA (mRNA) levels in noradrenergic brain stem cell groups and their forebrain projection areas after FST in vehicle- and DMI-treated HR and LR rats. Finally, as stress responsiveness in general and CORT secretion in particular are thought to play a role in mediating symptoms of depression (Ressler and Nemeroff 2000), we compared FST-induced CORT secretion in HR vs. LR animals.

Materials and methods

Animals

A total of 152 adult male Sprague–Dawley rats (Charles River Laboratories, Wilmington, MA, USA) weighing approximately 225–250g upon arrival were used across all the experiments. Animals were housed three per cage in a room adjacent to the testing room and maintained on a 12/12-h light/dark cycle (lights on at 0700 hours). Rats were acclimated to the animal quarters for 1week before any experimental procedure. Experiments were conducted between 1300 and 1700 hours during the light portion of the cycle. Food and water were available ad libitum. Animals were treated in accordance with National Institutes of Health guidelines on laboratory animal use and care.

Experiment 1: comparison of the behavioral effects of DMI during FST in HR vs. LR animals

Locomotor activity test

After 7days of habituation to the housing conditions, 45 rats were tested for locomotor activity during a 60-min exposure to the mild stress of a novel environment. Each rat was placed in a 43 L × 21.5 W × 24.5 H (in cm) clear acrylic activity monitor, and locomotor activity was monitored by means of two banks of three photocells each connected to a microprocessor. The locomotion testing rig and motion recording software were created in-house at the University of Michigan. Rats that exhibited locomotor counts in the highest third of the sample were classified as HR (n = 15), whereas rats that exhibited locomotor counts in the lowest third of the sample were classified as LR (n = 15; Cecchi et al. 2002).

Forced swim test

The animals were placed in vertical Plexiglas cylinders 60H × 40D (in cm) containing water (25°C) at a depth of 30cm (Detke et al. 1995). On the first day of the experiment, rats were placed in the water for a 15-min period (pretest phase). After the pretest phase of the FST, HR and LR animals were randomly assigned to two different treatment groups: saline (VEH) or 10mg/kg DMI (DMI) to be administered subcutaneously 5min, 6h, and 23h after the end of the pretest phase (Taghzouti et al. 1999). The experimental design yielded four groups: HR-VEH, LR-VEH, HR-DMI, and LR-DMI (n = 7–8 per group). The selected dose of DMI has been shown to be effective but submaximal in affecting rats’ behavior in the FST (Detke et al. 1995); it has therefore been chosen to highlight possible differences in response between HR and LR animals. Twenty-four hours after pretest, the animals were tested again on the FST for a 5-min period (test phase). Rats’ behavior during pretest and test was recorded to allow later scoring.

Behavior scoring

The time in seconds each rat spent climbing, lying immobile, and swimming was measured for each rat during pretest and test. Climbing was defined as the behavior during which a rat makes vigorous movements with its forepaws in and out of the water, usually in contact with the walls; immobility was defined as the behavior during which a rat is only making those movements required to keep its head above the water; finally, swimming was defined as the behavior during which a rat is actively moving around in the cylinder.

Experiment 2: comparison of the effects of FLU on climbing, immobility, and swimming during FST in HR vs. LR animals

Animals were sorted in HR and LR and tested in the FST as above. Vehicle or fluoxetine (20mg/kg) was administered subcutaneously 5min, 6h, and 23h after the end of the pretest phase (n = 7–8 per group). This dose of fluoxetine was chosen because it had been shown to be differentially effective in HR vs. LR animals (Taghzouti et al. 1999). The time spent by each rat climbing, lying immobile, and swimming during pretest and test was measured as above.

Experiment 3: effects of DMI on c-fos mRNA levels in NE nuclei and projections, and on CORT secretion in HR vs. LR rats

Animals were sorted in HR and LR according to their locomotor activity and tested in the FST. HR and LR rats received a subcutaneous vehicle or DMI injection (10mg/kg) 5min, 6h, and 23h after the end of the pretest phase according to their treatment group as in experiment 2 (n = 6 per group). Animals were euthanized 30min after the test phase of the FST, as it has been shown that this time corresponds to the peak of FST-induced CORT secretion (Connor et al. 1997), and their brain and blood collected. Brains were immediately frozen in isopentane cooled to −30°C and stored at −80°C. Blood samples were separated by centrifugation (3,000rpm for 10min at 4°C), and plasma was removed, frozen and stored at −80°C.

In a separate control experiment, animals were tested to investigate if DMI differentially affects c-fos mRNA levels and CORT secretion in HR and LR animals that are not subjected to the FST. Animals were separated in HR and LR according to their locomotor activity and assigned to one of three treatment groups: no injection, saline, or DMI. The experimental design yielded six groups: HR-no injection (HR-NI), HR-saline (HR-VEH), HR-DMI, LR-no injection (LR-NI), LR-saline (LR-VEH), and LR-DMI (n = 4 per group). According to their treatment group, animals received a subcutaneous injection of saline or DMI, or were simply handled, 24.5h, 18.5h, and 90min before euthanasia. Brains and blood were then processed as above.

In situ hybridization

The in situ hybridization method used in this study is described in detail by Isgor et al. (2003). Briefly, tissue was sectioned at −20°C at a thickness of 12μm, mounted onto poly(l-lysine)-coated slides, and stored at −80°C until use. Before probe hybridization, tissue was fixed in 4% paraformaldehyde at room temperature, rinsed with aqueous buffers, and dehydrated with graded alcohols. A sense and an antisense riboprobe were synthesized with incorporation of 35S-UTP and 35S-CTP from a cDNA fragment for c-fos generously donated by Dr. T. Curran and hybridized to tissue overnight at 55°C. Sections were then washed with increasing stringency, dehydrated with graded alcohols, air-dried, and exposed to film. The sense probe showed no detectable signal in the brain. Exposure time was chosen to maximize signal of the antisense probe. After X-ray film exposure, sections were finally stained with cresyl violet, dehydrated, and coverslipped with a xylene-based mounting medium (Permount).

Image analysis

Brains were processed for in situ hybridization and c-fos mRNA levels compared in HR vs. LR animals in bed nucleus of stria terminalis (BST; bregma −0.26 to −0.80mm), hippocampus (HIPP; bregma −3.14 to −3.80mm), paraventricular nucleus of the hypothalamus (PVN; bregma −1.60 to −2.12mm), medial prefrontal cortex (PFC; bregma +3.20 to +3.70), locus coeruleus (LC; bregma −7.64 to −8.00mm), and nucleus of the solitary tract (NTS; bregma −13.68 to −14.08mm). PFC and PVN have been selected because in a previous study, DMI has been shown to affect FST-induced fos-like immunoreactivity in these nuclei (Duncan et al. 1996). BST receives one of the densest NE fiber inputs in the brain (Moore and Bloom 1979) and has been shown to play a key role in NE modulation of stress and anxiety (Cecchi et al. 2002). Finally, functional and structural abnormalities in the HIPP correlate with the presence and severity of affective disorders (Benedetti et al. 2006).

Six tissue sections were selected for each brain region from each animal. Digital images of the brain sections were captured from X-ray films in the linear range of the gray levels using a CCD camera (TM-745, Pulnix, USA). Optical density for c-fos mRNA was determined for each section using the Micro Computer Imaging Device (Ontario, Canada) image analysis system. For each animal, data from multiple sections were averaged to obtain a representative value (mean optical density).

CORT Assay

For plasma CORT measurement, aliquoted samples (10μl) from each animal were suspended in radioimmunoassay (RIA) buffer and heated for 30min at 70°C to separate CORT from CORT binding globulin. Total CORT was assayed by RIA using a rabbit antiserum raised against B-21-hemisuccinate:BSA. The antiserum cross-reacts 8% with cortisol, 1% with deoxycorticosterone and progesterone, and less that 1% with aldosterone, testosterone, and estradiol. 3H CORT was used as a tracer. The detection limit of the RIA was about 5ng/ml, and the intra- and inter-assay coefficients of variation were less than 5% and 6%, respectively.

Statistical analysis

Statistical comparisons were done by two-way analysis of variance (ANOVA). The factors of variation were animals’ phenotype and drug treatment. Where ANOVA indicated significant main effects or significant interactions, post hoc comparisons were conducted using the Newman–Keuls test.

Results

Experiment 1: DMI significantly decreases immobility in FST in LR but not HR animals

The results of the locomotor activity test followed a unimodal distribution. The average locomotor activity for all rats in experiment 1 (n = 45) was 389 locomotor counts, with locomotor counts of 535 ± 108 and 295 ± 48 for HR and LR rats, respectively.

HR and LR animals behaved similarly during pretest, with no significant differences between groups in climbing, immobility, or swimming time (data not shown).

DMI treatment led to a significant overall increase in climbing time (F 1,25 = 10.65, p < 0.01), with no significant phenotype effect or significant phenotype × treatment interaction (see Fig. 1a). A two-way ANOVA for swimming time revealed a significant treatment effect (F 1,25 = 5.14, p < 0.05) and a significant phenotype × treatment interaction (F 1,25 = 4.36, p < 0.05) with no significant phenotype effect. Subsequent post hoc analyses showed a significant DMI-induced decrease in swimming time in HR but not LR animals (p < 0.05; see Fig. 1b). However, the most profound change was seen in the immobility measure. Statistical analyses for immobility time showed a significant group × treatment interaction (F 1,25 = 4.10, p < 0.05). Subsequent post hoc analyses indicated that DMI significantly decreased immobility time in LR but not HR animals (p < 0.01; see Fig. 1c). Indeed, DMI almost abolished immobility in the LR animals, reducing it to 8.6% of the vehicle score. By contrast, the immobility score of the HR rats was completely unaltered by DMI. Thus, in the case of the HR animals, the drug treatment altered the proportion of their active behaviors—increasing climbing and decreasing swimming. By contrast, in the LR animals, the drug significantly increased the active behavior of climbing and decreased immobility.

Fig. 1
figure 1

Behavioral effects of DMI treatment (10 mg/kg) on the FST. Effects of DMI on climbing (a), swimming (b) and immobility time (c) in HR and LR animals during the test phase of the FST. All values are mean ± SEM (n = 7–8 per group) #p < 0.05, ##p < 0.01 compared to vehicle-treated rats. Two-way ANOVA followed by Newman–Keuls test for post hoc comparisons

Full size image

Experiment 2: FLU affects HR and LR animals’ behavior equally in the FST

Results of the locomotor activity test were similar to the ones described for experiment 1. HR and LR animals behaved similarly during the pretest, also as above (data not shown).

During the test, there were no significant effects of FLU treatment on climbing time (see Fig. 2a). Statistical analysis for swimming time revealed a significant increase in FLU-treated animals (F 1,26 = 5.187, p < 0.05), with no significant phenotype effect or significant phenotype × treatment interaction (see Fig. 2b). Finally, statistical analysis for immobility time showed a significant effect of treatment (F 1,26 = 9.071, p < 0.01), with no significant phenotype effect or significant phenotype × treatment interaction (see Fig. 2c). Thus, fluoxetine was equally effective in both the HR and LR animals, decreasing immobility and increasing swimming time.

Fig. 2
figure 2

Behavioral effects of FLU treatment (20 mg/kg) on the FST. Climbing (a), swimming (b), and immobility time (c) during the test phase of the FST in HR and LR animals in response to saline or FLU treatment. All values are mean ± SEM (n = 7–8 per group). #p <  0.05, ##p < 0.01 compared to vehicle-treated rats. Two-way ANOVA

Full size image

Experiment 3: DMI differentially affects FST-induced c-fos mRNA expression in PFC and PVN in HR vs. LR animals

In animals that were not subjected to the FST, a light but detectable signal for c-fos mRNA was found only in PFC. When data for PFC were quantified and analyzed, two-way ANOVA showed no significant phenotype effect (F 1,18 = 0.676), treatment effect (F 2,18 = 1.409), or phenotype × treatment interaction (F 2,18 = 0.940).

Comparison of c-fos mRNA levels after FST in HR vs. LR animals

Locus coereleus

Two-way ANOVA showed a significant phenotype effect (F 1,16 = 37.950, p < 0.01), a significant treatment effect (F 1,16 = 53.895, p < 0.01), as well as a significant phenotype × treatment interaction (F 1,16 = 10.803, p < 0.01). Subsequent post hoc analyses showed higher FST-induced c-fos mRNA levels in HR than in LR animals. DMI treatment led to a significant decrease in c-fos mRNA expression in the LC in both phenotypes (see Fig. 3a).

Fig. 3
figure 3

Effects of DMI treatment (10 mg/kg) on c-fos mRNA levels after FST. Messenger RNA levels for c-fos after FST in the LC (a), NTS (b), PFC (c), and PVN (d) of HR and LR animals treated with either saline or DMI as measured by in situ hybridization. All values are mean ± SEM (n = 5–6 per group). **p < 0.01 compared to HR animals in the same treatment group. #p < 0.05, ##p < 0.01 compared to vehicle-treated rats. Two-way ANOVA followed by Newman–Keuls test for post hoc comparisons

Full size image

Nucleus tractus solitarius

By contrast, statistical analyses for NTS showed a significant decrease in FST-induced c-fos mRNA expression (F 1,16 = 10.803, p < 0.01), with no significant phenotype effect or phenotype × treatment interaction (see Fig. 3b).

Forebrain

When data for the brain regions in the forebrain were analyzed, DMI showed no phenotype-dependent effects in BST or CA1, CA2 and CA3 subregions of the HIPP (data not shown). By contrast, statistical analysis for PFC revealed a significant treatment effect (F 1,20 = 15.748, p < 0.01) and a significant phenotype × treatment interaction (F 1,20 = 19.200, p < 0.01). Subsequent post hoc analyses showed a significant decrease in c-fos mRNA expression after DMI treatment in LR animals only (p < 0.01; see Fig. 3c). Similarly, statistical analysis for PVN revealed a significant phenotype × treatment interaction (F 1,20 = 7.200, p < 0.05). Once again, post hoc analyses showed a significant decrease in c-fos mRNA expression after DMI treatment in LR animals only (p < 0.01; see Fig. 3d). Figure 4 shows c-fos mRNA expression in representative coronal sections of PFC.

Fig. 4
figure 4

Coronal sections of PFC, hybridized by c-fos mRNA radioactive probe, exposed onto a Kodak XAR film and captured using video-interfaced computer image analysis software. Scale bar = 1 mm. Representative sections from animals in the different experimental groups are shown

Full size image

Corticosterone secretion in HR vs. LR animals

CORT secretion after FST was 181 ± 5.8 SEM ng/ml and 192 ± 13.5 SEM ng/ml for vehicle-injected HR and LR rats, respectively. Statistical analyses indicated a significant decrease in CORT secretion after DMI treatment (F 1,18 = 11.944, p < 0.01). There was a tendency for a bigger DMI effect in LR animals (a 38% decrease in CORT secretion vs. a 12% decrease in HR rats); this phenotype difference, however, did not reach statistical significance.

In animals that were not subjected to FST, CORT levels ranged from 13.6 ± 1.43 SEM ng/ml for the HR-VEH group to 31.8 ± 7.2 SEM ng/ml for the LR-NI group. Two-way ANOVA showed no significant phenotype effect (F 1,18 = 0.371), treatment effect (F 2,18 = 1.711), or phenotype × treatment interaction (F 2,18 = 3.025).

Discussion

The body of work described here suggests that the novelty-seeking trait is useful in predicting responsiveness to certain types of antidepressants, particularly DMI, a prototypical NRI.

Most remarkably, at the dose used, DMI treatment resulted in a profound change in immobility scores, but in LR animals only. It is important to underscore that a decrease in immobility during FST is thought to be the best single index in predicting the antidepressant properties of a given drug (Porsolt et al. 1977). Thus, the findings suggest that LR rats may be sensitive to DMI’s antidepressant effects, while HR rats may be resistant to it.

It should be noted that DMI at the dose used was not inactive in HR animals. Indeed, the drug produced a significant increase in climbing regardless of the animal’s phenotype and a decrease in swimming only in HR rats. Both climbing and swimming behaviors in the FST represent active coping strategies aimed at escape and removal of stress (Thierry et al. 1984), but the significance of each behavior relative to the other remains unclear. Thus, DMI altered the ratio of active escape behaviors in HR animals without reducing immobility, while it primarily led to a decrease in immobility in the LR rats.

At the neural level, the FST test resulted in higher levels of c-fos mRNA expression in the LC in vehicle-injected HR than in vehicle-injected LR animals. These results are consistent with previous reports of higher NE release during restraint stress in the hippocampus of HR versus LR rats (Rosario and Abercrombie 1999). The hippocampus receives NE afferents exclusively from the LC through the dorsal noradrenergic bundle (Loy et al. 1980), and our data add further evidence of higher activation of this NE pathway in HR animals during inescapable stress.

When we looked at how DMI affected c-fos activation after the FST, the decrease in immobility in LR animals was accompanied by significantly lower FST-induced c-fos mRNA levels in PFC and PVN. Our data are consistent with a previous report of a decrease in fos-like immunoreactivity in the PFC and PVN after FST in animals treated with NRIs at an effective dose (Duncan et al. 1996). However, this is the first report to focus on group differences and to demonstrate that the change in c-fos gene expression correlates with a behavioral change in immobility and differs as a function of the animal’s propensity for novelty-seeking and predisposition to spontaneous anxiety.

Beyond these phenotype-specific changes, there were some shared alterations in patterns of neural activation after DMI. Thus, the drug treatment decreased FST-induced c-fos in LC and NTS in both HR and LR rats. These findings are consistent with previous studies suggesting that inhibition of NE reuptake by NRIs causes a stimulation of postsynaptic receptors which, in turn, decreases the activity of the presynaptic neurons by a feedback mechanism (Nyback et al. 1975; Scuvee-Moreau and Dresse 1979). It has been suggested that such inhibition could be critical in mediating the effects of DMI on immobility during forced swim test in rats (Kostowski et al. 1984). However, in our study, FST-induced c-fos expression was similarly reduced in LC and NTS in HR and LR animals regardless of their response to DMI in the FST. Moreover, in LR animals, c-fos mRNA levels were altered by DMI after FST in the PVN, but not the BST, although both nuclei receive their NE input mainly through the ventral bundle (Moore and Bloom 1979). Similarly, FST-induced c-fos mRNA expression was decreased in PFC, but not HIPP, after DMI administration in LR rats, although these brain regions both receive NE innervation through the dorsal bundle (Moore and Bloom 1979). Therefore, our results suggest that the mechanisms through which DMI affects immobility time in the FST are, at least in part, postsynaptic.

In our study, HR and LR animals showed a similar decrease in immobility time and increase in swimming after FLU administration in both groups of animals, suggesting little or no differences in the serotonergic system, at least in response to the dose we used. Consistently, parallel studies in our laboratory show comparable c-fos mRNA levels in the dorsal raphe of HR and LR animals after FST, a further indication of similar regulation of the serotonergic system in these two subpopulations of animals during FST (in preparation). In a previous study, Taghzouti and collaborators have shown a differential response to FLU in HR vs. LR animals in the FST. In their study, FLU caused a significant decrease in immobility time in LR animals only. By contrast, HR animals showed a significant increase in immobility after FLU administration. Similar results were obtained for swimming time. In the overall population of rats, they observed no effect of FLU on immobility time (Taghzouti et al. 1999). Most of the published studies report a decrease in immobility time and increase in swimming behavior in the FST after FLU administration (Cryan et al. 2005). There have been however reports of a lack of effect of SSRIs in the FST in rats (Borsini 1995). It is possible that small differences in experimental protocol and/or scoring of the results might lie behind the heterogeneity of results obtained in the FST after SSRIs administration.

Conclusion

In conclusion, our results show that differences in novelty-seeking behavior can predict inter-individual variability in response to DMI in the FST in rats. In our study, DMI was selectively effective in decreasing immobility only in LR animals. This behavioral effect was accompanied by a decrease in FST-induced c-fos mRNA expression in the PFC and PVN. Our study proposes the HR-LR model as a valuable tool to investigate between-subject variability in responsiveness to DMI in the FST and suggests that further studies based on this model of inter-individual variability in rats could lead to a better understanding of the biological mechanisms underlying individual differences in behavioral and physiological responses to NRIs in humans.