Twelve-week old male C57BL/6 mice (Charles River Laboratories, Maastricht, Netherlands) were maintained under standard lab conditions (12:12 h light/dark cycle, controlled temperature (22 +/− 2°C) and humidity (55+/− 5%), and ad libitum access to food and water). Mice were singly-housed and acclimated to the room for 10 days before the experimental onset. The experiments were carried out in accordance with the European Communities’ Council Directive 2010/63/EU. All efforts were made to minimize animal suffering during the experiments. The protocols were approved by the committee for the Care and Use of Laboratory animals of the Government of Upper Bavaria, Germany.

Mice were randomly assigned to 2 × 2 groups (control vehicle (n = 12), control CMI (n = 12), chronic stress vehicle (n = 12), and chronic stress CMI (n = 11)) counterbalanced by body weight. Clomipramine was administered orally through drinking water (0.12 mg/ml) from the first day of the stress procedure (day 1) until the animals were sacrificed (day 22). An open field (OF) test and forced swim test (FST) were performed on day 16 and day 18, respectively in order to assess CMI efficacy. We were mindful of the behavioral testing order, beginning with the least invasive test before testing more invasive assays (McIlwain et al., 2001; Paylor et al., 2006). Therefore the OF test was carried out before the FST and there was one full day of rest between the two tests. Body weight was measured daily and fluid intake was measured twice weekly throughout the entire experimental time-course. Behavioral testing and animal sacrifice always occurred in the morning at approximately the same time so that levels of metabolites measured at the time of sacrifice are representative of levels at the time of testing.

The CSDS paradigm lasted for 21 days and was conducted as described previously (Wagner et al., 2011, 2012). Briefly, experimental mice were placed in the home cage of a dominant CD1 resident mouse. Interaction between the mice was permitted until the experimental mouse was attacked and defeated by the CD1 aggressor. Mice were subsequently separated by a wire mesh divider that prevented physical contact but maintained sensory contact for 24 h. Each day, for 21 days, the procedure was repeated with a different, unfamiliar CD1 aggressor mouse. Both control and stress animals were handled daily during the course of the stress procedure.

Mice were placed in one corner of a 50 cm × 50 cm × 40 cm arena. Thirty-minute trials were video-recorded by an overhead camera and analyzed using the automated video tracking software ANYmaze 4.9 (Stoelting, Wood Dale, IL, USA). The arena was cleaned with water at the beginning of testing and in between animals. Total distance traveled and time spent in the OF center was measured. For analysis of time spent in the center of the arena, a center zone was virtually defined as a 25 cm × 25 cm central square.

Mice were placed in a 2 L glass beaker filled with room-temperature (22 ± 1°C) water to a height of 15 cm so that the mouse could neither touch the bottom nor escape. The test lasted 6 min and was later analyzed by an experienced experimenter, blind to the experimental group. The first 2 min were designated a habituation period, and therefore time spent immobile and time spent struggling in the final 4 min of the test were scored (Castagne et al., 2009).

The FST additionally served as an acute stressor to examine the stress response. The stress response was performed as previously described (Wagner et al., 2012; Balsevich et al., 2014; Santarelli et al., 2014). Briefly, at the conclusion of the FST, animals were towel-dried and returned to their home cages to recover. At 30-min (stress response) and 90-min (stress recovery) after the onset of the FST, blood samples were taken by tail cut (Fluttert et al., 2000). The 30-min time point was chosen based on previous data indicating a near maximal response of the HPA axis at this time (Droste et al., 2008). Samples were collected in EDTA-coated microcentrifuge tubes (Kabe Labortechnik, Germany) and kept on ice until all samples were centrifuged at 8000 rpm at 4°C for 15 min. Plasma was collected and stored at −20°C. Plasma corticosterone levels were determined by radioimmunoassay using a commercially available kit (MP Biomedicals Inc.; sensitivity 12.5 ng/ml).

On day 22 of the experimental timeline, mice were anesthetized with Isofluorane and then immediately sacrificed by decapitation. Basal trunk blood was collected and subsequently processed (as described above). Brains were removed, snap-frozen, and stored at −80°C until use. Adrenal glands were removed, pruned from fat, and weighed.

Mice brains were weighed and then homogenized in the fivefold volume phosphate buffered saline (PBS), containing “Complete Protease Inhibitor Cocktail Tablets” (Roche, Penzberg, Germany) using a Dispomix Drive (Medic Tools AG, Zug, Switzerland).

The blood plasma and the brain homogenates were analyzed using the combined high-performance liquid chromatography/mass spectrometry (HPLC/MS-MS) technique. Analysis was performed using an Agilent 1100 Series (Agilent, Waldbronn, Germany) liquid chromatograph, which was interfaced to the ESI source of an Applied Biosystems API 4000 (ABSciex, Darmstadt, Germany) triple quadrupole mass spectrometer. All samples were prepared using Ostro protein precipitation and phospholipid removal plates (Waters, Eschborn, Germany).

Deuterated CMI (CMI-D3) was used as internal standard. Chromatography was accomplished using an gradient elution in a Accucore RP-MS 2.6 μm column (2.1 × 50 mm, Thermo Scientific, Dreieich, Germany) at a flow rate of 0.3 ml/min and 30°C.

The composition of eluent A was methanol with 10 mM ammonium formate with 0.1% formic acid and water with 10 mM ammonium formate with 0.1% formic acid as eluent B.

The gradient was 0–0.5 min 20% A, 0.5–2 min 20–90% A, 1 min held at 90% A, 3–3.5 min 90–20% A and 3.5–8 min 20% A. The total run time was 8 min and the injection volume was 5 μl.

The retention time for CMI, DCMI and CMI-D3 were 4.9 min, 4.9 min and 4.9 min, respectively. The ion source was operated in the positive mode at 500°C, and multiple reaction monitoring (MRM) collision-induced dissociation (CID) were performed using nitrogen gas as the collision gas. The collision energy was set to 27 V, 33 V and 27 V for CMI, DCMI and CMI-D3, respectively. The transitions monitored during analysis were m/z 315 → 86 for CMI, m/z 301 → 72 for DCMI and 318 → 89 for CMI-D3. The detection limit for CMI in plasma was 5 ng/ml and 3 ng/g wet weight in brain tissue and 2.5 ng/ml and 3 ng/g wet weight in brain tissue for DCMI.

All variables were evaluated using IBM SPSS Statistics 18 software (IBM SPSS Statistics, IBM, Chicago, IL, USA). Data were analyzed by 2-way ANOVA for significant effects of treatment and stress. Where the initial test yielded a significant interaction, a Bonferroni post hoc test was applied. Where there was only one grouping variable, independent Student’s T-test were used. Finally, correlations were analyzed with the Pearson product-moment test. Statistical significance was set at p < 0.05. Data are presented as mean ± S.E.M.

Body weight gain from the onset of CSDS/treatment exposure until sacrifice was not significantly affected by either stress or CMI treatment. There was a significant effect of stress on fluid consumption (stress: F_(1,44) = 133.195, p < 0.001) in which stress significantly increased daily fluid intake. There was however no effect of CMI treatment on fluid intake. Mice exposed to CSDS therefore received a higher dose of CMI compared to control animals receiving CMI treatment (T_(16,9) = −9.368, p < 0.001).

Stress exposure resulted in increased anxiety-like behavior, as demonstrated by the reduced number of entries into the center (stress: F_(1,44) = 7.563, p = 0.009) and a trend towards reduced center time in the OF test (stress: F_(1,44) = 3.478, p = 0.069). Although there was no significant treatment × stress interaction for either OF center time or center entries, the reduced center time and center entries in the OF test on account of CSDS, appear to be largely mediated by the CMI-treated cohort (Figure 1A). Stress additionally tended to reduce total distance traveled in the OF (stress: F_(1,44) = 3.024, p = 0.089). Likewise heightened depressive-like behaviors resulted from CSDS, shown in the FST as a reduction in time spent struggling (stress: F_(1,40) = 6.054, p = 0.018) and an increase time spent immobile (stress: F_(1,44) = 22.211, p < 0.001; Figure 1B). Stress furthermore enhanced the stress response as shown by elevated basal (F_(1,44) = 7.649, p = 0.008), response (F_(1,44) = 25.136, p < 0.001), and recovery (F_(1,42) = 6.590, p = 0.014) corticosterone levels (Figure 1C). By contrast, CMI had no main effect on any of the parameters (center time, center entries, total distance) measured in the OF test or on corticosterone levels. There was a trend indicating that CMI treatment reduced the time spent immobile (CMI: F_(1,44) = 3.413, p = 0.071) in the FST (Figure 1B). Finally stress significantly increased adrenal size (stress: (F_(1,43) = 100.902, p < 0.001) whereas CMI treatment led to reduced adrenal size (treatment: (F_(1,43) = 4.768, p = 0.034; Figure 1D).

Clomipramine and DCMI levels were measured in the plasma and the brain and the absolute values are presented in Table 1. Briefly, stress exposure significantly lowered absolute levels of CMI in the brain (T_(11,7) = 2.302, p = 0.041), but stress did not significantly affect the levels of CMI in the plasma or DCMI levels in the plasma and brain. Metabolic ratios and partitioning ratios were investigated as an index of CMI pharmacokinetics, thereby also controlling for the observed differences in fluid intake. Assessment of the metabolic ratios (defined as the ratio of CMI to DCMI) revealed an effect of stress. The metabolic ratios were significantly lowered on account of CSDS in the plasma (T₍₂₁₎ = 3.981, p = 0.001) and in the brain (T_(16,41) = 6.600, p < 0.001; Figure 2A). In order to assess whether tissue distribution of CMI/DCMI was affected by stress, we treated tissue and CSDS as independent variables. Significant effects of stress and tissue were detected (stress: F_(1,43) = 57.905, p < 0.001; tissue: F_(1,43) = 4.968, p = 0.031). Importantly, a significant interaction was also detected to reveal that only under control conditions was the CMI/DCMI ratio significantly higher in the brain compared to the plasma (stress × tissue: F_(1,43) = 5.343, p = 0.026). Regardless, control animals presented a higher CMI/DCMI ratio in both plasma and brain compared to stressed-animals.

To measure brain uptake of CMI and DCMI, the partitioning ratio (defined as the brain-to-plasma concentration) was calculated. Chronic social defeat stress significantly decreased the partitioning ratio for CMI (T₍₂₁₎ = 3.663, p = 0.001), whereas significantly increased the partitioning ratio for DCMI (T₍₂₁₎ = −2.460, p = 0.023; Figure 2B).

The Pearson product-moment test was used to describe the relationship between CMI bio-distribution and behavioral responses. The analyses were subdivided by stress condition to account for the main stress effects on both CMI bio-distribution and behavioral readouts, which would bias the relationship. Under basal conditions, the CMI/DCMI metabolic ratio for both the plasma and brain negatively correlated to the time spent immobile (floating) in the FST (plasma: r = −0.693, p = 0.018; brain: r = −0.643, p = 0.024; Figures 3A,B). This correlation was absent under stress conditions. Finally, the DCMI partitioning ratio positively correlated to the time spent immobile (r = 0.610, p = 0.035; Figure 3C) under stress, but not under control conditions.