Introduction

Stressful life experiences result in biological and behavioral responses that increase the risk to develop stress-related pathologies, including major depression [1].

According to the concept of allostasis [2], acute stress may have positive effects, and in line we have demonstrated that acute stress improves the cognitive performance [3]. Conversely, chronic stress may induce detrimental and protracted changes [4,5,6,7,8] by perturbing the homeostatic network and leading to the so-called “allostatic load” and “allostatic overload”, which drive psychopathologies [9]. However, the consequences of chronic stress are not predictable. Indeed, even under chronic stress, the brain can, or cannot, activate adaptive mechanisms resulting in, respectively, resilience or vulnerability [9]. This is witnessed by the evidence that some subjects exposed to stress experience diseases while others elaborate resilience and maintain normal functions [10].

To date, despite several attempts have been made to unravel the mechanisms responsible for the development of pathological phenotypes when exposed to stress, less effort has been made to fully clarify the determinants that draw the trajectory of stress response toward resilience, i.e. the ability to adapt to adverse context [11], both at central and peripheral level [9, 12,13,14,15].

In the last few years, alterations in brain metabolism have been linked with both the pathogenesis and pathophysiology of psychiatric disorders [16]. This is corroborated by the analysis of blood and urine samples from depressed patients that show an alteration in the levels of metabolites involved in the modulation of energy and neuronal functions [17]. Consistent with this, at preclinical level, as recently reviewed by van Der Kooij [18], accumulating evidence showed metabolic alterations in animal models of psychiatric disorders based on chronic stress exposure. However, the detailed mechanisms underpinning energy metabolism in psychiatric disorders have not yet been exhaustively elucidated.

On these bases and considering the fundamental influence of bioenergetics to stress-related disorders we employed a metabolomic approach to have a more comprehensive understanding of the mechanism that may lead to susceptibility and resilience to chronic stress exposure. For this purpose, we used the chronic mild stress (CMS) paradigm, a well-characterized animal model of MDD [19], which allows stratification of this population into vulnerable and resilient groups by evaluating the hedonic phenotype [4, 5]. Indeed, anhedonia, the inability to derive pleasure from normally rewarding experiences, is one of the core symptoms of depressed patients, as listed in the DSM-5 [20]. The analyses were carried out in the ventral hippocampus (vHip) given its key role not only in the mediation of stress response and in the management of specific pathological phenotypes, including the anhedonic-like behavior [21], but also in energy metabolism [22].

Additionally, to deeper investigate the potential mechanisms underlying the metabolic changes in our experimental setting, we focused on mitochondrial dynamics whose alteration may compromise the well-being of the entire cell, thus causing the pathological conditions connected with mitochondrial homeostasis [23]. Indeed, mitochondria are the primary organelles involved in the regulation of energy production within the cell and sustain stress response system by modulating energy transformation as well as several intracellular signaling pathways [22, 24]. In this context, mitochondrial adenosine triphosphate (ATP) production is fundamental for the support of synaptic transmission and communications, the release of neurotransmitters as well as for the correct maintenance of plasticity [25, 26], all of which are essential for the proper brain functions aimed to cope with stress.

Here, we provide evidence that a peculiar mitochondrial function and energetic metabolite profile contribute to dictate the difference between resilience or vulnerable phenotype in response to stress.

Material and methods

Animals

Adult male Wistar rats (Charles River, Germany) were brought into the laboratory one month before the start of the experiment. Except for the first 10 days after arrival when the animals were housed in groups of 10, they were singly housed in standard laboratory conditions: except for the CMS procedure, food and water was freely available on a 12-h light/dark, constant temperature (22 ± 2 °C) and humidity (50 ± 5%). All procedures used in this study have conformed to the rules and principles of the 86/609/EEC Directive and have been approved by the Local Bioethical Committee at the Maj Institute of Pharmacology, Polish Academy of Sciences, Krakow, Poland. All efforts were made to minimize animal suffering, to reduce the number of animals used and the animal studies comply with the ARRIVE guidelines.

Stress procedure and behavioral test

After 2 weeks of adaptation to the housing conditions, rats were trained to consume 1% sucrose solution as previously described [5] and sucrose consumption was monitored at weekly intervals throughout the duration of the study (Fig. 1A).

Fig. 1: Behavioral characterization of animals exposed to chronic mild stress.
figure 1

A schematic representation of the experimental paradigm; B sucrose intake was measured in the sucrose consumption test (SCT) at weekly intervals in control (no stress) or stressed (CMS-vul/CMS-res) animals. The data are the mean ± SEM: ***P < 0.001 vs no stress (one-way ANOVA with repeated measures, Fisher’s PLSD); C schematic representation of the dissection of the vHip; D analysis performed in vHip.

Full size image

On the basis of their sucrose intakes in the final baseline test, the animals were randomly divided into two matched groups: one group was subjected to the CMS for a period of 2 consecutive weeks (see: Calabrese et al. [4] for details) and the other one was not subjected to the CMS procedure (control group). On the basis of the result of the sucrose consumption test carried out following the first 2 weeks of stress, animals showing the anhedonic phenotype (CMS-vulnerable) were separated by the animals that did not develop anhedonia despite CMS (CMS-resilient) (Fig. 1B).

The animals of each experimental group were decapitated 24 h after the final sucrose test and the ventral hippocampus was dissected from the whole brain according to the plates 34–43 of the atlas of Paxinos and Watson [27] (Fig. 1C) for the subsequent molecular analysis (Fig. 1D).

Behavioral testing was done blindly by an experimenter who was unaware of the experimental group of the animals.

Metabolomic analysis

Metabolomic data were obtained by liquid chromatography coupled to tandem mass spectrometry. We used an API-3500 triple quadrupole mass spectrometer (AB Sciex, Framingham, MA, USA) coupled with an ExionLC™ AC System (AB Sciex, Framingham, MA, USA). 10 mg of ventral hippocampus were used for the analysis. Half tissue was smashed in 250 µl of ice-cold methanol/acetonitrile 50:50, while the second half was lysed in 250 µl of ice-cold water/methanol 20:80, respectively. Both solutions contained [U-13C6]-glucose (Merck Life Science S.r.l, Milano, Italy) 1 ng/µl and [U-13C5]-glutamine (Merck Life Science S.r.l, Milano, Italy) 1 ng/µl as internal standards. Lysates were spun at 20,000×g for 5 min at 4 °C and supernatants were then passed through a regenerated cellulose filter (4 mm Ø). Samples were then dried under N2 flow at 40 °C. Samples were then resuspended in 100 µl of methanol for subsequent analysis.

Quantification of energy metabolites was performed by using a cyano-phase LUNA column (50 mm × 4.6 mm, 5 µm; Phenomenex, Torrance, CA, USA) by a 5 min run in negative ion mode with two separated runs. Protocol A: samples lysed in acetonitrile/methanol were used to analyze lactate, malate, αKetoglutarate, phosphoenolpyruvate (PEP), dihydroxyacetone-P/glyceraldehyde-3P (DHAP/GAP), erytrose-4P (E4P), dTMP, dAMP, dIMP, dCTP, ITP, and GTP. The mobile phase A was: water and phase B was: 5 mM ammonium acetate in MeOH and the gradient was 10% A and 90% B for all the analysis with a flow rate of 500 µl/min. Protocol B: samples lysed in water/methanol solution were used to analyze 3′, 5′-Cyclic GMP, acetyl-CoA, ADP, AMP, ATP, cAMP, Citrate, CMP, CoA, CTP, dADP, dATP, dCDP, dCMP, dGDP, dGMP, dGTP, dITP, dTTP, dUMP, dUTP, FAD, Fructose bis-P, Fumarate, GDP, Glucose, Glucose-6P, GMP, IMP, Iso-citrate, malonyl-CoA, NAD+, NADH, NADP+, NADPH, oxaloacetate, pyruvate, ribose-xylulose-ribulose-5P (R-X-Ru-5P), succinate, succinyl-CoA, UDP, UMP, and UTP. The mobile phase A was: water and phase B was: 5 mM ammonium acetate in MeOH and the gradient was 50% A and 50% B for all the analysis with a flow rate of 500 µl/min.

Carnitine quantification was performed on acetonitrile/methanol extracts by using a Varian Pursuit XRs Ultra 2.8 Diphenyl column. Samples were analysed by a 3 min run in positive ion mode and the mobile phase was 0.1% formic acid in MeOH.

Amino acid and biogenic amine quantification were performed through previous derivatization. Briefly, 20 µl out of 100 µl of acetonitrile/methanol samples were collected and dried under N2 flow at 40 °C. Dried samples were resuspended in 50 µl of phenyl-isothiocyanate (PITC), EtOH, pyridine, and water 5%:31.5%:31.5%:31.5% and then incubated for 20 min at RT, dried under N2 flow at 40 °C for 90 min and finally resuspended in 100 µl of 5 mM ammonium acetate in MeOH/H2O 50:50. Quantification of different amino acids was performed by using a C18 column (Biocrates, Innsbruck, Austria) maintained at 50 °C. The mobile phases for positive ion mode analysis were phase A: 0.2% formic acid in water and phase B: 0.2% formic acid in acetonitrile. The gradient was T0: 100%A, T5.5: 5%A, T7: 100%A with a flow rate of 500 µl/min. All metabolites analyzed in the described protocols were previously validated by pure standards and internal standards were used to check instrument sensitivity.

MultiQuant™ software (version 3.0.3, AB Sciex, Framingham, MA, USA) was used for data analysis and peak review of chromatograms. Raw areas were normalized by the median of all metabolite areas in the same sample. The data were then transformed by generalized log-transformation and Pareto scaled to correct for heteroscedasticity, reduce the skewness of the data, and reduce mask effects [28]. In detail, obtained values were transformed by generalized log (glog) as follows:

$$glog_2\left( x \right) = log_2\frac{{x + \sqrt {x^2 + a^2} }}{2}$$

where a is a constant with a default value of 1 and x is the sample area for a given metabolites [29]. Then, obtained values underwent Pareto scaling as follows:

$${{{\bar{\mathrm x}}}}_{ij} = \frac{{x_{ij} - \bar x_i}}{{\sqrt {s_i} }}$$

where xij is the transformed value in the data matrix (i (metabolites), j (samples)) and si is the standard deviation of transformed metabolite values [30]. Obtained values were considered as relative metabolite levels. Data processing and analysis were performed by MetaboAnalyst 5.0 web tool [22]; however, when we go into details and dissect the energetic status into the major classes of metabolites, i.e. glycolysis and TCA cycle, differences come up that might sustain the response to stress. In fact, among glycolysis and TCA cycle, DHAP/GAP, lactate and acetyl-CoA were all increased in vulnerable but not resilient animals, possibly indicating a different metabolism of glucose and pyruvate between vulnerable and resilient animals. Consistently, NAD+/NADH ratio was increased only in vulnerable rats, indicating that the conversion of pyruvate to lactate might be exploited to regenerate NAD+ levels in the cytoplasm. Noteworthy, altered levels of lactate were previously observed in the hippocampus of depressed rats [36]. Besides ATP production, energy metabolism provides a variety of metabolic intermediates for the generation of nucleotides, e.g. NADPH, rubose-5P, and ATP. Of note, the energetic crisis occurring in vHip of stressed animals was in line with the altered levels of nucleotides observed. In this regard, NTPs/dNTPs ratio was significantly increased in both vulnerable and resilient rats. This alteration was probably due to energetic depletion observed in the vHip of our animals, and might explain, at least in part, the dysregulation of neurogenesis observed in the vHip of several pre-clinical models of depression [12, 37,38,39]. 1C cycle intermediates levels support this hypothesis; indeed, serine levels were higher in vulnerable animals, while glycine was increased only in vulnerable rats.

Fatty acid β-oxidation is a major source of mitochondrial acetylCoA in many tissues. For long, the role of β-oxidation in brain homeostasis has been underestimated. Nevertheless, recent researches highlighted an important role of lipid metabolism, and more specifically of β-oxidation, in several adult brain functions [40,41,42,43]. Strikingly, we observed that many carnitines, indirect indicators of β-oxidation flux and altered in depression, were upregulated by chronic stress, independently from the behavioral phenotype [44, 45]. Specifically, we found increased levels of medium and long chain-acylcarnitines, possibly indicating an incomplete oxidation rather than an efficient fueling of carbons from fatty acids to the TCA cycle [46]. This is in line with the altered energetic profile described above. In addition, among the TCA cycle intermediates citrate and iso-citrate were the lowest abundant under stress, indicating that stressed animals are unable to efficiently convert oxaloacetate and acetylCoA to citrate. Of note, citrate synthase expression levels have been associated to cognitive decline in aged animals [47]. On the contrary, our data showed that α-ketoglutarate and other TCA cycle intermediates were unaffected or even increased, as for succinate, in the vHip of stressed animals. Together with these findings, increased levels of glutamine and glutamate in stressed animals suggest a replenishment of TCA cycle from glutaminolysis rather than from pyruvate and β-oxidation. Altered levels and transport of glutamate were also observed in previous works focused on vHip of stressed animals, suggesting a major role of this amino acid/neurotransmitter in stress adaptation [48, 49]. On the other hand, it is known that glutamine, produced by glutamine synthetase in astrocytes, is involved in the detoxification of brain ammonia [50]. This is due to the presence of incomplete urea cycle in the central nervous system, whose main function seems to be the synthesis of citrulline and arginine rather than ammonia depletion [51, 52]. Strikingly, both citrulline and arginine are involved in nitric oxide metabolism in the hippocampus, with protective effects against stress and cognitive decline [53,54,55]. Consistently, our data suggest a possible role of glutamine and urea cycle metabolites citrulline and arginine in the regulation of ammonia and nitric oxide levels under stress in adult resilient rats.

Accordingly, alterations of metabolites of the purine network, glycolysis, and fatty acid beta oxidation have been found in the vHip of mice exposed to the chronic social defeat animal model of depression [45], as well as changes in several metabolites in the whole hippocampus of rats subjected to the chronic mild unpredictable stress [56]. Furthermore, other authors have also shown lipidomic [56, 57] and proteomic changes [58, 59] in the brain of rodents exposed to chronic stress protocols.

Moreover, metabolic characterization of peripheral blood from MDD patients revealed disturbances of different metabolic pathways [60], including altered plasma neurotransmitter metabolite profile [61], thus increasing the knowledge about the potential molecular pathogenesis of MDD.

Prompted by the metabolic differences between vulnerable and resilient rats, we provided further support to these findings exploring the possibility that mitochondrial oxidative phosphorylation, morphology, and recycling may be part of a strategy to cope with CMS. We observed an upregulation of the protein expression of some subunits of complex I, II, and IV of the electron respiratory chain in resilient animals. In particular, the elevated levels of C-I in CMS-res underlined the unbalance in the NAD+/NADH and lactate levels, suggesting that the mitochondria of resilient animals have a more efficient NADH oxidation. Accordingly, recent findings demonstrated that mice with mutation of the Ndfus4 gene, which encodes for a structural component of C-I, showed increased susceptibility to stress following 3 weeks of chronic unpredictable stress [62]. Moreover, C-II impairment has been related to ROS production and susceptibility to manifest anxiety behavior [63]. Consistently, mitochondria of vulnerable and resilient animals activate different mechanisms of morphology and recycling regulation. Vulnerable rats increase fusion machinery to cope with the excessive production of ROS, as supported by the increased levels of Cat protein and mRNA levels. In addition, vulnerable animals seem to activate the pro-apoptotic BNIPL3/NIX axis that has been found to be highly induced during hypoxia conditions and stressful conditions [64]. On the other hand, resilient animals favor mitochondrial fission likely to guarantee a higher mitochondrial quality and metabolic efficiency. In fact, the increased activation of mitochondrial fission by the pDRP1 Ser616 is associated to the PINK1-mediated mitochondrial quality control in resilient animals, potentially preventing the onset of major depression abnormalities. In line with these findings, DRP1 knockout in mouse embryonic fibroblasts showed suppressed mitophagy mediated by Parkin, while DRP1 and nitric oxide production have been recently linked to corticotrophin-releasing hormone activity in the hippocampal neurons of stressed animals [65, 66].

Taken together, these results suggest that, despite the metabolic and energetic status were profoundly affected by stress exposure independently from the behavioral phenotype, the vulnerability and resilience seem to be linked with the activation of different mitochondrial strategies set in motion to cope with negative challenges in rat vHip.

In conclusion, the modulation of the energetic metabolic profile in vHip under chronic stress exposure may represent a mechanism to explain the difference between vulnerable and resilient rats, unraveling novel and promising targets for effective therapeutic interventions.

These findings could provide novel insights into metabolic changes that could be helpful as diagnostic and predictive markers for the prevention and intervention in MDD as well as for the discovery of candidate drug targets. Indeed, the outcome of the metabolomic profile we determined in vulnerable and resilient animals may be useful in the translation “from bench to bedside” to identify innovative blood metabolite markers associated to MDD.