Bipolar disorder (BD) is a severe, recurrent psychiatric condition characterized by episodes of mania or hypomania and depression. Mania presents with hyperactivity, disinhibited behavior and inflated self-esteem. On the other hand, during depression patients demonstrate anhedonia and loss of motivation and interest in usual activities [1
Pharmacological treatment of BD includes mostly the use of mood stabilizers. The first drug fulfilling criteria for the mood stabilizer such as preventing manic and depressive episodes was lithium. Its antimanic actions in the acute episode were first discovered by Cade et al. in animal model studies [2
]. Its antidepressant potential was reported in the 1970s [3
] and later it was recommended as an augmentation of antidepressants in the treatment-resistant depression and an effective long-term prophylaxis of recurrences in mood disorders (as recently reviewed by Rybakowski [5
Although lithium has been for decades in clinical use, the molecular mechanism of action and normotymic potential of lithium are still not well identified. Moreover, absence of suitable animal model of bipolar disorder impedes studies on lithium action. Previous studies focused on identifying target genes and molecular pathways of lithium indicated its role in reducing inositol signaling [6
], phosphorylation of AKT [7
] and inhibition of GSK-3β (glycogen synthase kinase 3) signaling [6
]. The latter two may lead to the changes in the activation of WNT pathway. These findings were possible to discover using pharmacological and pharmacogenetic studies, which showed involvement of lithium in multiple cellular signaling pathways as reviewed previously [9
Animal studies of the brain transcriptome during lithium administration showed that inconsistent results are mainly due to the different duration of lithium administration (1 week, two weeks and 6 weeks), analysis of the whole brain tissue (whole brain homogenates) and the use of different animals (rat or mouse) [11
]. Moreover, these studies used animals treated with lithium, but without manic-like or depressive-like symptoms, so it is still not known if therapeutic action of lithium in manic and depressive behavior induces changes in the gene expression of similar or different pathways depending on the baseline condition and the brain region.
Therefore, we hypothesized that chronic lithium treatment exerts its therapeutic effect by inducing changes in the brain transcriptome specific for mania and depression. The aim of this study was to investigate if chronic lithium administration in rats presenting manic-like behavior (induced by amphetamine) or depressive like behavior (induced by chronic mild stress) influences specific gene expression profiles in different brain regions (amygdala, frontal cortex and hippocampus).
The main observation of this study shows that the brain transcriptome is influenced by chronic lithium administration. The important finding is that the differentially expressed genes specific for the mania model are involved in olfactory receptor transduction and taste receptor pathways and show altered expression of long non-coding RNA genes.
Comparing depressive-like and manic-like animals, we found that in the depression model, the gene expression profile is more region-specific (different pathways regulated upon lithium in the amygdala, frontal cortex and hippocampus) and time-dependent (2 weeks versus 4 weeks). In the model of mania, on the other hand, lithium significantly affected olfactory receptor and bitter taste receptor pathways, independent of the brain region or time of lithium administration suggesting their role in antimanic action. The genes from these pathways also showed significantly altered expression in “manic” rats before lithium administration. Our observation is consistent with the previous human studies in bipolar disorder patients that presented several olfactory and gustatory dysfunctions [13
] and the sensory enhancement and dysregulation of taste were often reported during an acute manic episode in BD patients [14
Previous animal studies of lithium showed that mice chronically fed with lithium accumulated it mainly in neurogenic brain regions including olfactory bulb [15
]. Similarly, chronic lithium effects in the neurogenic brain region (hippocampus) were also observed in human BD studies showing that lithium influenced hippocampal volume and structural plasticity in BD patients [16
]. These findings suggest that chronic lithium treatment targets olfactory bulb and alters associated pathways (e.g., olfactory transduction).
On the contrary, in “depressive” rats we observed the upregulated expression of genes regulating olfactory receptor activity after 4-week lithium administration in the amygdala, which is consistent with the previous observations that bipolar patients during depressive episode experienced sensory blunting and olfactory deficits [14
]. Thus chronic lithium treatment seems to restore olfactory function in rats presenting depressive-like behavior. Moreover, previous study showed that olfactory bulbectomy induced depressive-like behavior in rodents [19
], in particular when taking into account the close anatomical links between the olfactory system and the brain circuits involved in memory [20
] and emotion [21
]. The recent studies reported the presence of sensory changes during mood swings in BD [22
] and that olfactory assessment may be useful to screen unipolar and bipolar depression [23
]. These genes involved in olfactory receptors activity were also differentially expressed in either manic and depressive rats responding to chronic lithium treatment and showed the opposite effect that also depended on tissue: in the amygdala they were upregulated after chronic lithium in mania, but downregulated in depression; in the hippocampus they showed downregulated expression in mania, but upregulated in depression after chronic lithium. This opposite gene expression profiles between amygdala and hippocampus after chronic lithium treatment may be further supported by results of the recent study that showed the opposite temporal profiles of protein expression between amygdala and hippocampal neurons in long-term response to acute stress [24
Our findings from the animal model indicating the involvement of the taste receptor pathway were further supported by the observations in our clinical sample of bipolar and unipolar patients showing altered expression of taste receptor genes in peripheral blood leukocytes during depressive episode (Dmitrzak-Węglarz et al., under review). However, this pathway was not previously reported in the chronic lithium mechanism of action.
Another pathway significantly influenced by chronic lithium treatment in our animal mania model was bitter taste receptor pathway. Recent transcriptome study by Lee et al. [25
] showed altered taste receptors gene expression during manic episodes. These genes included bitter taste receptors, TAS2R5 and TAS2R3, that were suggested as potential markers specific for the manic state [25
]. These receptors were previously found downregulated (together with olfactory receptors) in the dorsolateral prefrontal cortex of schizophrenia postmortem brain tissues [26
] and this downregulation influenced altered cognition. The taste alteration was also related to psychosocial and cognitive performance in bipolar patients [13
]. However, the previous studies did not investigate the effects of mood stabilizers on taste-related gene expression. Our observation that chronic lithium influences the regulation of the olfactory and taste receptors’ pathway is supported by the previous case studies reporting long lasting impaired taste (dysgeusia) and smell (hyposmia) in lithium users: cluster headache patient [27
] and bipolar patient [28
Our study revealed that a number of differentially expressed genes influenced by chronic lithium treatment in the brain were long non-coding RNAs. The paper by ConLiGen showed significant associations for two long non-coding RNAs (lncRNAs) localized on chromosome 21 with lithium prophylactic efficacy supporting their involvement in modulating a clinical response [29
]. Several studies reported that lncRNAs expression was dysregulated in psychiatric conditions, including BD [30
] whereas Lee et al. found recently that lncRNAs were the predominant category showing upregulation during an acute manic episode [25
]. Here, we report for the first time that lncRNAs expression is altered by chronic lithium treatment and suggest they may be regulators of antimanic lithium action.
Comparing the 2-week and 4-week gene expression profile, we found that the longer the treatment was, the more differentially expressed genes were observed, independent of the brain region. A previous study comparing microarrays from rat brains after 7 and 42 days of lithium treatment showed that, although plasma Li concentration reached therapeutic levels after 2 days of treatment, it required 2 weeks to reach therapeutic levels in the brain [11
]. This finding suggested that the long-term lithium treatment and associated gene expression profile changes better represent the downstream Li effects, which are more relevant to its clinical effect. The other report in the rat fed for 21 days with lithium led to significant changes in the transcriptome in the frontal cortex [31
To investigate the chronic lithium effect in the brain that mimics changes during the episode of depression or mania, we aimed to model mania and depression separately as the animal model of bipolar disorder does not exist [32
]. Animal models are relevant platforms to evaluate the intracellular mechanisms in the brain associated with the pathogenesis of psychiatric disorders and drug studies, including lithium discovery [2
]. In our study, we applied a chronic mild stress (CMS) protocol, the most extensively validated [35
], to induce depressive-like behavior. Moreover, the CMS protocol was also suitable to model recurrent depression (a hallmark of bipolar disorder) [36
]. The choice of amphetamine to mimic mania was supported by the previous preclinical studies that used repeated intraperitoneal injections of psychostimulants, such as amphetamine, to mimic acute manic episodes in rodents, including hyperactivity and risk-taking behavior due to increased sustained dopamine efflux [37
]. Moreover, these behaviors were reversed by the administration of mood stabilizers, including lithium [40
]. This chronic amphetamine-induced mania model showed recently good face, predictive and construct validity in modeling mania in Wistar rats [44
4. Materials and Methods
4.1. Experimental Animals
All experimental procedures were performed in agreement with 3R rule and the study was approved by local ethical committee Poznan University of Life Sciences, Poland (agreement no. 22/2017, 23 June 2017). We used male Wistar rats with the baseline weight of 180 ± 10 g. The rats were housed five animals per cage with food and water available ad libitum and were maintained in a 12 h light/dark cycle (lights on at 7:00 a.m.) at a temperature of 22 ± 1 °C. All experiments were performed at the same time each day to avoid circadian variations. The animals were kept for one week of acclimatization and then were randomly divided into experimental groups with 5 animals in each group (amphetamine-exposed and control group, chronic mild stress-exposed and non-stressed rats; amphetamine-exposed rats receiving lithium or water and stress-exposed rats receiving lithium or water). After the acclimatization period, animals underwent baseline behavioral tests. The experimental design was shown in Figure 7
4.2. Animal Model of Mania
Manic-like behavior was induced by daily intraperitoneal (i.p) injections of dextroamphetamine (d-AMPH) 2 mg/kg for 2 weeks as described previously [37
]. The control group were animals receiving daily intraperitoneal injection with saline (0.9% NaCl, 1 mL/kg) for 14 days. After seven days, the behavior of all animals was assessed (Figure 7
4.3. Animal Model of Depression
Animal model of depression was developed by the chronic mild stress protocol (CMS) as described previously [48
]. To induce depressive behavior the animals were exposed for 4 weeks to different stress stimuli according to the CMS protocol described by Papp [49
]. The control group were rats kept in standard conditions for 4 weeks that were not exposed to CMS. After two weeks of the start of the experiment, behavior of all animals was assessed (Figure 7
4.4. Lithium Administration
The animals for the lithium study, after one week of amphetamine injections or two weeks of the chronic mild stress protocol and behavioral assessment, were randomly divided into lithium treated amphetamine-exposed group (n
= 15) and amphetamine-exposed control group (rats receiving water, n
= 15) and lithium-treated stress-exposed group (n
= 15) and a stress-exposed control group (rats receiving water, n
= 15). The animals were receiving the daily lithium solution in the dose of 1 mg/kg body mass or the same amount of water into the mouth with a syringe [50
]. To minimize the side effects of lithium on kidney function, a saline bottle was provided in all lithium and control cages, as previously reported [51
]. To assess the short term and long term lithium effects on brain transcriptome, the animals (n
= 15 at each time in each group) were sacrificed by decapitation without euthanasia at two time points of chronic lithium administration, 14 and 30 days (long-term effect) after the first dose. The each control group (for stress-exposed and amphetamine-exposed rats) consisted of animals receiving water (Figure 7
4.5. Behaviour Assessment
All the behavioral tests were performed during the light phase (between 08:00 a.m. and 12:00 p.m.), at room temperature in a separate quiet room. A blind evaluator assessed all behavioral parameters. To assess depressive-like or manic-like behavior we used a forced-swim test, open field test and elevated maze test at baseline and after induction of stress-induced depression or amphetamine-induced mania.
A forced swim test was performed to measure despair behavior and the hopelessness in the animal [52
]. Each rat was placed in a glass cylinder 40 cm tall filled with water to a depth of 30 cm (24 ± 1 °C) so the rats could not support themselves by touching the bottom. Two swimming sessions were conducted: a 15-min pretest followed by a 5-min test 24 h later. For each animal, we recorded time spent immobilized (no additional activity other than that required to keep the rat’s head above the water) and active climbing time (upward-directed movements of the forepaws along the side of the swim chamber. The data were analyzed using the video tracking system (Videomot2, AnimaLab, Poznan, Poland). The water in the cylinders was replaced after each trial to remove urine or feces and to avoid confounding results.
Open field test was performed to evaluate the spontaneous locomotory activity in a new environment and exploration [54
]. The rat was placed in the middle of the plastic box measuring 100 cm in diameter with 50-cm walls, divided into 25 squares of 20 × 20-cm size on the bottom (AnimaLab, Poland) and allowed to explore it for 5 min. We recorded the following parameters: time spent immobilized, time spent in center, total distance and exploration time and analyzed the data with the video tracking system (Videomot2, AnimaLab, Poland). After the end of each test, the open field was cleaned to remove any remaining materials (such as feces, urine or smell) that could interfere with the test results.
The elevated maze test was used to test the anxiety level and risk-taking behavior in the rats [42
]. The apparatus consisted of two opposite open arms, two opposite closed arms and a central platform connecting the four arms. The maze was propped up 50 cm away from the ground by the maze feet. The test room was maintained with constant temperature, humidity and illumination. The rats were placed in the central platform facing one of the open arms. The number of entries into the open and closed arms and the total time spent in each arm during a 5-min exploration period were recorded by the video tracking system (Videomot2, AnimaLab, Poznań, Poland). After the end of each test, the open field was cleaned to remove any remaining materials (such as feces, urine or smell) that could interfere with the test results.
4.6. Microarray-Based Gene Expression Analysis
The different regions of brain tissue (amygdala, hippocampus, prefrontal cortex and hypothalamus) were collected immediately after decapitation (without anesthesia). The skull was opened, and the cerebral content was excised and rapidly dissected on a chilled Petri dish. The frontal cortex, hippocampus and amygdala were isolated and cleaned from the subcortical structures and white matter, immediately snap frozen in liquid nitrogen and kept frozen in −80 °C for further processing. Frozen tissues were used for RNA extraction using Nucleospin RNA/Protein kit (Macherey Nagel, Dylan, Germany). RNA integrity (RIN) was assessed using Tape Station 2200 (Agilent) and RNA concentration was measured using fluorimeter (Quantus, Promega). Total RNA from amygdala, hippocampus and frontal cortex in the starting amount of 50 ng was used for microarray experiments. We used SurePrint G3 Rat Gene Expression v2 Microarray Kit in format 8 × 60 k and one-color Low Input Quick Amp Labeling Kit (Agilent Technologies, Cedar Creek, TX, USA) to analyze gene expression profile following the standard protocol provided by the manufacturer. The hybridization signals were detected with SureScan Dx Microarray Scanner (Agilent Technologies, Santa Clara, CA, USA). The images obtained after scanning were analyzed with Agilent Feature extraction software v.18.104.22.168.
4.7. Statistical Analysis
Behavioral data were analyzed using a t
-test for paired samples after checking normality of data distribution using the Shapiro–Wilk test. A cut-off value p
< 0.05 was regarded as significant. Data were shown as mean ± SEM in figures and text if not otherwise stated. For behavioral measurements and gene expression data, 5 animals per experimental group were sufficient to achieve power of 80% (as calculated by the G-power calculator available at https://stats.idre.ucla.edu/other/gpower
). To conform to the 3R’s rule, we included the minimal required number of animals (n
= 5) in the study.
Bioinformatic analysis of gene expression data was analyzed with the use of Gene Spring 14.9 software (Agilent Technologies, Santa Clara, CA, USA). To identify differentially expressed genes, moderated t-statistics from the empirical Bayes method on normalized fluorescence signal was used for the fold of change (FC) calculation. FC was calculated in relation to the adequate control group. The list of significantly differentially expressed transcripts (p < 0.05 and FC > 2.0) was generated using statistical filtering (moderated t-test with multiple test correction FDR). Then, separate lists of up- and downregulated genes were used for functional analysis to identify Gene Ontology (GO) terms and pathways significantly enriched by the set of differentially expressed genes, either upregulated or downregulated (p < 0.05). Gene Ontology and pathway analyses were done in Gene Spring.