PPARγ Dysfunction in the Medial Prefrontal Cortex Mediates High-Fat Diet-Induced Depression

Epidemiological studies suggest a bidirectional association between depression and obesity; however, the biological mechanisms that link the development of depression to a metabolic disorder remain unclear. Even though nuclear receptor peroxisome proliferator-activated receptor γ (PPARγ) agonists show anti-depressive effect, and high-fat diet-(HFD)-induced PPARγ dysfunction is involved in the pathogenesis of metabolic disorders, the neuronal PPARγ has never been studied in HFD-induced depression. Thus, we aimed to investigate the effect of neuronal PPARγ on depressive-like behaviors in HFD-induced obese mice. We fed male C57BL/6 J mice with HFD to generate obese mice and conducted a series of behavioral tests to assess the effects of HFD feeding on depression. We generated neuron-specific PPARγ knockout mice (NKO) to determine whether neuronal PPARγ deficiency was correlated with depressive-like behaviors. To further prove whether PPARγ in the medial prefrontal cortex (mPFC) neurons is involved in depressive-like behaviors, we applied AAV- CaMKIIα-Cre approach to specifically knockout PPARγ in the mPFC neurons of LoxP mice and used AAV-syn-PPARγ vectors to overexpress PPARγ in the mPFC neurons of NKO mice. We observed a low mPFC PPARγ level and an increase in depressive-like behaviors in the HFD-fed mice. Moreover, neuronal-specific PPARγ deficiency in mice induced depressive-like behaviors, which could be abolished by imipramine. Furthermore, overexpressing PPARγ in the mPFC reversed the depressive-like behaviors in HFD-fed mice as well as in neuronal-specific PPARγ knockout mice. These results implicate that dysregulation of neuronal PPARγ in the mPFC may contribute to an increased risk for depression in obese populations.


Introduction
Depression is a chronic and debilitating mental illness with a 17% lifetime prevalence and is a major cause of morbidity, disability, and mortality [1]. Clinical studies have demonstrated that depressed patients exhibit multiple symptoms, and a significant percentage of depressed patients are refractory to classical monoamine-based antidepressants, suggesting that the pathophysiology leading to depression may differ among patients [2]. Increasing evidence has suggested that depression may involve different molecular mechanisms [3]. Identifying the pathophysiology and signaling pathways underlying the multiple causes of depression is necessary to effectively control depression.
Previous work has suggested a bidirectional association between depression and obesity. Obese individuals have a 55% increased risk of developing depression over time [4], and diabetes is estimated to double the incidence of depression [5,6]. At the same time, depressed individuals have a 58% increased risk of becoming obese. Depression is considered a systemic disease, and metabolic diseases, such as obesity, insulin, and leptin resistance, may also be responsible for the increased incidence of depression [7,8]. Therefore, elucidating the biological mechanisms that promote depressed moods in individuals with obesity/ diabetes is useful to find targets for new agents.
The nuclear receptor peroxisome proliferator-activated receptor gamma (PPARγ) is a ligand-activated transcription factor that belongs to the nuclear receptor family, and the activity of which is regulated by the direct binding of steroid and thyroid hormones, vitamins, lipid metabolites, and xenobiotics [9]. Impaired PPARγactivity has been observed in the adipocytes of obese mice, and this impairment mediates insulin resistance [10]. Moreover, the Pro12Ala polymorphism in PPARγ is associated with a high risk for developing depression [11]. Furthermore, Ahmed et al. reported that rosiglitazone produces a specific antidepressant-like effect in the forced swim test (FST) and tail suspension test (TST) [12]. Guo et al. demonstrated a novel role for adipose PPARγ in susceptibility and resilience to chronic stress and a functional link between PPARγ and adiponectin in mediating depressivelike behaviors [13]. Neuronal seipin deficiency, which causes reduced PPARγ levels, leads to depressive-like behaviors in male mice that are reversed by PPARγ agonist rosiglitazone [14]. However, whether neuronal PPARγ impairment is sufficient to induce depressive like behaviors or whether neuronal PPARγ deficiency is involved in the etiology of depression has not been proven.
In the current study, we examined whether high-fat diet-(HFD)-induced obesity causes a decrease in brain PPARγ and whether this impairment of PPARγ activity contributes to the onset of depression. We found that HFD-induced obese mice displayed lower PPARγ levels in their medial prefrontal cortex (mPFC) and depressive-like symptoms, which could be blocked by AAV-PPARγ stereotaxically injected into the mPFC. Furthermore, we demonstrated that mPFC neuronal-specific PPARγ-deficiency is sufficient to induce depressive-like behaviors, which could be rescued by overexpressing PPARγ.

Animals
Adult male C57BL/6 J mice (aged 7-8 weeks) were obtained from the Southern Medical University Animal Center (Guangzhou, China). Retired male CD1 (aged 8-9 months) breeders were obtained from Charles River Laboratories (Beijing, China). PPARγ LoxP/LoxP (PPARγ fl/fl ) mice and calcium/calmodulin-dependent protein kinase II alpha (CaMKIIα)-Cre mice [15] were purchased from the Jackson Laboratory (Maine, USA). Neuronal-specific PPARγ knockout mice (NKO) [16] were generated by crossing the PPARγ fl/fl mice with the CaMKIIα-Cre mice. PPARγ fl/ fl (LoxP) mice were used as the control. Mice were housed four per cage in standard laboratory cages except for CD1 breeders, which were individually housed in the same kind of cages. All mice were kept on a standard 12-h light/dark cycle (8:00-20:00 light period) in a temperature-controlled room (21 ± 25 ℃), with free access to food and water. All of the experiments were conducted in accordance with the regulations of the Administration of Affairs Concerning Experimental Animals (China) and were approved by the Southern Medical University Animal Ethics Committee. Male mice were used in all experiments.

Genotype Identification of PPARγ Gene Knockout Mice
The offspring of the transgenic mice were genotyped using PCR with mouse tail DNA, and the DNA was amplified with the following primers: 5′-TGC CCA AGA AGA AGA GGA A-3′ (forward primer for CaMKIIα-Cre) and 5′-TTG CAG GTA CAG GAG GTA GTC-3′ (reverse primer for CaMKIIα-Cre); 5′-TGT AAT GGA AGG GCA AAA GG-3′ (forward primer for LoxP) and 5′-TGG CTT CCA GTG CAT AAG TT-3′ (reverse primer for LoxP). The PCR products were visualized using ethidium bromide staining.

Obesity Model
One week after acclimating to current condition, the male C57BL/6 J mice were randomly divided into 2 groups and fed either with control diet (CD) (10% kcal fat, D12450B, Research Diets) or HFD (60% kcal fat, D12492, Research Diets) for 16 consecutive weeks. After the HFD feeding regimen, all mice were evaluated with glucose tolerance test, behavioral tests, or virus administration.

Glucose Tolerance Test
Mice were fasted overnight (15 h) before undergoing glucose tolerance test (GTT). Glucose (2 g/kg; Sigma-Aldrich) was injected intraperitoneally, and blood samples were obtained before (0) and 30, 60, 90, and 120 min after the injection of glucose. Blood glucose was measured using a glucometer (Bayer HealthCare LLC, 1455).

Primary Cortical Neuron Cultures
The neuronal culture was prepared from the cortices of littermates (within 24 h after birth) of NKO or PPARγ fl/fl mice as described with modifications [17]. Briefly, dissociated cortical neurons were cultured in Neurobasal Medium supplemented with 2% B27 and 25 μM glutamate in poly-llysine-coated plates. After 24 h in culture, the cells were treated for 48 h with 0.1 μM cytosine arabinoside (Ara-C) to inhibit glial growth. PPARγ expression in the neuronal cultures was determined 15 days after culture in vitro.

Tissue Preparation and Immunofluorescence
The brain tissue was prepared as described previously [18]. Adult C57BL/6 J mice injected with AAV were anaesthetized with isoflurane and sacrificed by intracardial perfusion with heparinized 0.9% saline following by ice-cold 4% paraformaldehyde. The brains were removed and post-fixed with 4% paraformaldehyde. The fixed brains were washed with flowing water for 2 h at room temperature and then incubated with a 30% sucrose solution in PBS at 4 °C for at least 2 days. A microtome (Leica CM 1850) was used to coronally Sect. (35 µm) the brain mPFC. The sections were mounted with Vectashield mounting medium containing DAPI (Vector Laboratories Inc.) and visualized using a confocal laser scanning system (Leica DM IRE2).

Open Field Test (OFT)
The OFT was conducted as previously described [19]. The open field apparatus consisted of a rectangular chamber (40 × 40 × 30 cm) which was made of gray polyvinyl chloride. The mice were gently placed at the center and were allowed to explore the area for 30 min. The digitized image of the track taken by each mouse was stored, and the locomotion activity was analyzed post hoc using EthoVision 7.0 software.

Forced Swim Test (FST)
The FST was performed as previously described [20][21][22]. Each mouse was placed in a clear glass cylinder (height 45 cm, diameter 19 cm), which was filled to 23 cm with water (22-25 °C). Mice were forced to swim for 6 min. The immobility time (minimal movements to keep the head above water) during the final 4 min were calculated by an investigator who was blind to the study conditions. This is due to the fact that most mice are very active at the beginning of the FST, and the potential effects of the treatment can be obscured during the first 2 min.

Tail Suspension Test (TST)
The TST is a classical test performed to evaluate depressive-like behaviors [23]. In the present study, it was performed with modified from the version described by Melissa and colleagues [24]. Each mouse was hung 50 cm above the floor for 6 min by adhesive tape at approximately 1 cm from the tip of the tail. Immobility time was evaluated in the 6 min using an automated device (Med Associates, USA). The immobility time can be defined as motionless hanging without any struggling behaviors. Any mice that did climb their tails were excluded from further experimental analysis.

Chronic Social Defeat Stress (CSDS)
CSDS was performed as described previously [25]. Briefly, a singly caged 8-10 weeks old male C57BL/6 J mouse (test mouse) was exposed to a CD1 aggressor mouse (target mouse) for 10 min each day for a total of 10 days. To separate the susceptible and resilient subpopulations, the avoidance behaviors were tested 24 h following the final session, and a social interaction ratio (time spent in the interaction zone in the presence versus the absence of a target mouse) of 1.0 was set as a cutoff. The mice with scores < 1.0 were considered susceptible, and those with scores ≥ 1.0 were considered resilient. The duration of time that the mice spent in the interaction zone was obtained using EthoVision 7.0 software (Noldus).
In the drug treatment experiment, all of the mice were housed individually. Following the CSDS protocol, the defeated mice were randomly divided into three groups and were given daily either saline, rosiglitazone (5 mg/kg, i.p.), or imipramine (15 mg/kg, i.p.) for 10 consecutive days. The rosiglitazone and imipramine were individually dissolved in dimethyl sulfoxide (DMSO) and then diluted in 0.9% saline to a final concentration of 0.5% DMSO. The long-term behavioral consequences of defeat stress were assessed using the social avoidance test 24 h following the final injection.

Sucrose Preference Test (SPT)
The procedure of sucrose preference test was performed as previously described, with modified [26,27]. Mice were singly caged and acclimated to drink from two 50-ml bottles (A and B) for 4 days. One of which was contained with 1% sucrose and the other with water. The sucrose preference was determined for 2 h after a 20 h period of water and food deprivation. The sucrose preference for each mouse was calculated as 100 × (VolA / (VolA + VolB)), and the total fluid intake was calculated as VolA + VolB.

Western Blot Analysis
Mice were decapitated, and the brain tissues were quickly removed and cut into 1-mm-thick slices on ice-cold Rodent Brain Matrix (RBM-2000C, ASI Instruments) as described previously [28]. The slices were arranged on a frozen glass plate, and regions (PFC, mPFC and hippocampus) were dissected using a cold scalpel, with reference to the mouse brain atlas. For a single subject, sections from both hemispheres were pooled together in a tube and immediately frozen on dry ice and stored at − 80 °C. For protein extractions, all tissue samples were homogenized in RIPA-buffer with a protein protease inhibitor cocktail. The membranes were incubated with PPARγ (1:1,000, Santa Cruz, California, USA) or GAPDH (1:4,000, Proteintech, Wuhan, China) antibodies and were visualized using chemiluminescence detection solutions (PerkinElmer). The FluorChemTM SP software was used to measure band densitometry.

Statistical Analysis
Data are shown as the mean ± SEM. Student's t test was applied for comparisons between 2 groups, and one-way and two-way ANOVA were used to compare 3 or more sets of data, followed by Tukey's multiple comparison test or Sadik's multiple comparison test separately, using SPSS 20.0 software. A probability (p) value of < 0.05 was considered to be statistically significant.

A Long-Term HFD Impaired Glucose Tolerance and Increased Depressive-Like Behaviors
To generate obese mice, we subjected 8-week-old male C57BL/6 J mice to the HFD feeding regimen for 16 consecutive weeks (HFD, 60% kcal fat, D12492, Research Diets Fig. 1 A long-term HFD impaired glucose tolerance and increased depressive-like behaviors. A Schematic diagram of the CD/HFD feeding and behavior tests design. B The initial and final body weights of male C57BL/6 J mice after 16 weeks of HFD feeding. n = 12 mice per group, ***p < 0.001. C Blood glucose levels after D-glucose (2 g/kg) injection in mice fed the CD or HFD, n = 5 mice per group. D The percentage of sucrose consumption in the sucrose preference test, n = 14-15 mice per group. E-F The effect of the chronic HFD consumption on depression-like behaviors expressed as the immobility time in the TST and FST (E and F), n = 14-15 mice per group. G The effect of chronic HFD consumption on spontaneous locomotion, as assessed by the OFT, n = 12-13 mice per group. H Western blot images and quantification of PPARγ levels in the PFC, mPFC, and hippocampus of the two groups of mice following 16 weeks with the CD or HFD. n = 6-8 mice per group. I Densitometry values for PPARγ were normalized to those of GAPDH. Data are the means ± SEMs. *p < 0.05, **p < 0.01, and ***p < 0.001.
To assess the effects of a long-term HFD feeding regimen on depressive-like behaviors, we subjected both groups of mice to a series of specific behavioral tests, including the OFT, the SPT, the TST, and the FST (Fig. 1A). After 16 consecutive weeks of feeding, the HFDfed group showed a significant reduction in sucrose consumption ratio compared to the CD-fed group (Fig. 1D, p < 0.05). The total intake of water and sucrose consumption was similar between the two groups, while the intake of sucrose solution was much higher in the HFD-fed mice compared with the CD-fed mice (As shown in Supplementary file 1 and file 2). The results indicated the presence of anhedonic behaviors in the HFD-fed group. The immobility time of the HFD-fed group increased in the TST (Fig. 1E, p < 0.01) and FST (Fig. 1F, p < 0.05) with no significant difference found between two groups in OFT (Fig. 1G, p > 0.05), which was consistent with the results of a previous study [29]. These results indicated that a chronic HFD feeding regimen led to depressive-like behaviors without a change in spontaneous activity.

PPARγ Expression Decreased in Brain of HFD-Fed Mice
To determine whether brain PPARγ might be related to HFD-induced depressive-like behaviors, we first measured PPARγ protein levels in the brain areas correlating with depression, namely, the PFC, mPFC, and hippocampus. Compared to CD-fed mice, HFD-fed mice showed a decreased level of PPARγ protein in the mPFC and PFC but not in the hippocampus (Fig. 1H and I).

Neuronal PPARγ Knockout Mice Displayed Depressive-Like Behaviors
Given that PPARγ expression decreased in brain of HFDfed mice, we generated neuronal-specific PPARγ knockout mice (NKO) using the Cre-LoxP recombination system to determine whether neuronal PPARγ deficiency was correlated with depressive-like behaviors. To achieve neuronal selectivity for PPARγ deletion, we crossed PPARγ fl/fl mice with mice expressing Cre under a well-characterized CaM-KII α promoter to drive Cre expression. The littermate PPARγ fl/fl (LoxP) mice were used as controls. As the PPARγ was observed in both neuronal and glial cells [30], to exclude the impact of glial cells, we checked PPARγ expression in primary cultured cortical neurons from both NKO and LoxP mice. As shown in Fig. 2A and B, on day 15, there was a significant loss of PPARγ protein expression in the cultured neurons from the NKO mice compared with those from the LoxP mice. The NKO mice had a normal appearance, and the neuronal PPARγ deficiency did not affect animal growth or body weight (as shown in Supplementary file 1 and file 3).
To determine whether neuronal-specific knockout of PPARγ affected depressive-like behaviors, TST and FST were performed, in which the duration of immobility time increased in the NKO group comparing with the LoxP group ( Fig. 2 C and D, p < 0.05), with no significant differences observed between groups in terms of the total distance traveled in the OFT (Fig. 2E, p > 0.05). Moreover, the increased immobility time in the FST of the NKO mice were reversed by acute treatment with imipramine (15 mg/kg body weight, by intraperitoneal (i.p.) injection) (F (2, 29) = 4.655, p = 0.0176; one-way ANOVA) (Fig. 2F).
In the social interaction test, two-way repeated-measures ANOVA revealed that neuronal-specific knockout of PPARγ significantly reduced the increased interaction time (interaction, F (1, 27) = 19.89, p = 0.0001). NKO mice showed no significant interest to target (no target 41.5 ± 3.3 s, target 38.7 ± 3.7 s, p > 0.05), while LoxP mice had a longer interaction time when the target was present (no target 34.2 ± 3.15 s, target 55.9 ± 4.1 s, p < 0.001) (Fig. 2G). These results indicated that neuronal PPARγ deficiency induced depressive-like behaviors in adult mice. Fig. 2 Neuronal PPARγ knockout mice displayed depressive-like behaviors. A Western blot showing PPARγ and GAPDH protein levels in cultured neurons at day 15 in culture that were generated from NKO and control (LoxP) mice. B The quantification of PPARγ levels in NKO and LoxP mice. n = 3 mice per group. C The immobility time increased for the NKO group compared with the LoxP group in the FST, n = 12 mice per group. D The immobility time increased for the NKO group compared with the LoxP group in the TST, n = 11 mice per group. E The influence of neuronal PPARγ deficiency on spontaneous locomotion, as assessed in the open field test, n = 8-10 mice per group. F The increased immobility time of the NKO group is reversed by imipramine in the FST, n = 10-11 mice per group. G The avoidance behaviors of the NKO mice in the social interaction test, n = 14-15 mice per group, **p < 0.001, two-way repeatedmeasures ANOVA (interaction, F (1, 27) = 19.89, p = 0.0001) followed by Sidak's multiple comparison test. H Confocal images showed the injection site of the AAV-CaMKIIα-Cre vector into the mPFC to induce PPARγ knockout in pyramidal neurons and produced GFP to mark the infected neurons. I-J PPARγ knockout in the mPFC of LoxP mice with AAV-CaMKIIα-Cre increased depression-like behaviors, as reflected by increased immobility time in the FST and TST, n = 9-11 mice per group. K PPARγ knockout in mPFC of LoxP mice with AAV-CaMKIIα-Cre had no effect on locomotive activity in the open field test, n = 9-11 mice per group. Data are the means ± SEMs. *p < 0.05, **p < 0.01, and ***p < 0.001.

Neuronal PPARγ Deficiency in mPFC Caused Depressive-Like Behaviors in Mice
Given that PPARγ expression was decreased in the mPFC of HFD mice and that the PPARγ-NKO mice showed increased depressive-like behaviors, we used an AAVmediated approach to achieve neuronal knockout of PPARγ in the mPFC to determine the contribution of neuronal PPARγ deficiency in the mPFC to depressive-like behaviors. PAOV-CaMKIIα -EGFP-T2A-Cre (CaMKIIα-Cre) or
CaMKIIα-Cre infection induced PPARγ knockout in pyramidal neurons and produced GFP to mark the infected neurons (Fig. 2H). Three weeks after the vector injections, depressive-like behaviors in the LoxP group were evaluated. We found that the immobility time increased in the CaMKIIα-Cre-LoxP group compared with the CaMKIIα-EGFP-LoxP group in the TST and FST (Fig. 2 I and J, p < 0.05), with no change in the total distance in the OFT (Fig. 2K, p > 0.05), and these behaviors were equivalent to those of the NKO mice. Our results proved that neuronal PPARγ deficiency in mPFC is sufficient to induce depressive-like behaviors in mice.

Neuronal PPARγ Overexpression in mPFC Rescued Depressive-like Behaviors in NKO Mice
Next, we used a viral expression approach to examine whether neuronal-specific overexpression of PPARγ in the mPFC is sufficient to rescue the depressive-like behaviors in NKO mice. To this end, we stereotaxically injected the pAAV-syn-EGFP-T2A-PPARγ-WPR-ES-pAS (AAV-PPARγ) vectors to induce neuronal-specific overexpression of PPARγ. The schematic of the AAV-PPARγ vector is shown in Fig. 3A (top), and the pAAV-syn-EGFP-T2A-WPR-ES-pAS (AAV-EGFP) viruses were used as the control viruses. The injection locations of AAV were illustrated on a brain atlas image as shown in Supplementary file1 and file 4. Western blotting revealed that the expression of PPARγ increased significantly in the mPFC after 2 weeks of injection of the AAV-PPARγ vectors (Fig. 3A, bottom). Next, we microinjected AAV-PPARγ or AAV-EGFP vectors into the mPFC of LoxP and NKO mice. The timeline of the experimental procedure is shown in Fig. 3B (top). Either AAV-PPARγ or AAV-EGFP was injected into the mPFC of NKO mice, and behaviors were examined 3 weeks later. We found that PPARγ overexpression in the mPFC reversed the increased immobility time in the FST (F (2, 33) = 4.305, p = 0.0218) and TST (F (2, 32) = 6.772, p = 0.0035) (Fig. 3C  and D), with no effect on the total distance in the OFT (F (2, 28) = 1.882, p = 0.1710) (Fig. 3E, p > 0.05), indicating that PPARγ deficiency in the mPFC contributed to depressive-like behaviors in NKO mice.

PPARγ Decreased in the mPFC of Susceptible Mice After Chronic Social Defeat Stress
The CSDS model mimics several psychopathological dimensions of depression. In this study, mice were exposed to 10 days of repeated social defeat stress, and social withdrawal behaviors were subsequently analyzed using a social interaction test 24 h after the last defeat episode. Social interaction ratio is the time spent in the interaction zone with the target present divided by the time spent in the interaction zone with the target absent. Usually, social interaction ratio equaling to 1.0 is used as the threshold for dividing defeated mice into susceptible and resilient groups (Fig. 4A). Sidak's multiple comparison test), while no difference was shown among groups when no target existed (no target: control mice, 59.1 ± 3.0 s; susceptible mice, 74.4 ± 3.1 s; resilient mice, 64.8 ± 3.8 s; Sidak's multiple comparison test) (Fig. 4B).
Next, we investigated whether PPARγ expression changed in the mPFC of the CSDS mice. At 48 h after the final defeat episode, susceptible mice showed less PPARγ protein in their mPFC than either the resilient or control mice, and the resilient mice showed increased PPARγ protein in their mPFC compared with the control mice, whereas the PPARγ protein levels in the PFC and hippocampus were comparable among the subgroups (Fig. 4C and D). We found a significant positive correlation between the PPARγ protein and social interaction ratio (Fig. 4E, Pearson correlation, r = 0.646, p < 0.05).
Previous studies showed that socially defeated mice demonstrate long-lasting social avoidance that is reversed by chronic (but not acute) treatment with antidepressants [2]. To determine whether rosiglitazone could reverse social withdrawal behaviors, after 10 days of social defeat stress, the defeated mice were treated with saline, rosiglitazone (5 mg/kg), or imipramine (15 mg/kg) for 10 consecutive days by i.p. injection. Two-way repeated-measures ANOVA (interaction, F (3, 57) = 6.464, p = 0.0008) showed that a 10-day treatment of rosiglitazone was sufficient to stably reverse the avoidance behaviors of defeated mice (target, saline 35.8 ± 4.3 s, rosiglitazone 61.7 ± 7.6 s, p < 0.05, Holm-Sidak's multiple comparison test). Meanwhile, no difference was found among treatments, while no target was introduced (p > 0.05, Holm-Sidak's multiple comparison test) (Fig. 4F). Previous reports have shown that a 4-week treatment with the well-characterized antidepressant imipramine is necessary to increase the interaction time of defeated mice to the levels of the control mice [31], whereas a 10-day treatment with imipramine is ineffective, which is consistent with our results. Taken together, these results indicate that PPARγ decreases in the mPFC of susceptible mice and increased in the mPFC of resilient mice after CSDS and that the agonist of PPARγ alleviates CSDS-induced depressivelike behaviors.

Discussion
In this study, we explored the role of PPARγ in depressivelike behaviors. We demonstrated that HFD-induced PPARγ impairment in the brain leading to depression. First, we found that PPARγ decreased in mPFC of HFD-fed mice (Fig. 1). Because PPARγ is found in astrocytes, neurons, and cerebrovascular endothelial cells, PPARγ of which cell might be involved in depression? We further proved that neuronal-specific PPARγ knockout was sufficient to induce depressive-like behaviors, and mPFC neuronal-specific PPARγ deficiency mice showed similar behavior ( Fig. 2 and Fig. 3), suggesting that PPARγ impairment in mPFC neurons caused depression. More importantly, making up for PPARγ decrease in the mPFC neurons caused by HFD reversed the depressive-like behaviors of HFD-fed mice ( Fig. 1 and Fig. 3). Our data proved that PPARγ dysfunction in the mPFC neurons might be the main causes of depression in HFD mice. Meanwhile, we found that PPARγ expression was decreased in the mPFC of susceptible mice and increased in the mPFC of resistant mice after CSDS (Fig. 4), and a 10-day treatment with PPARγ agonist rosiglitazone blocked these depression-related behaviors. Our data suggested that PPARγ also participated in the pathophysiology of social defeat-induced depression.
Previous studies indicated that dysfunction of mPFC is closely related to depression [32]. In this study, we found HFD-induced PPARγ dysfunction in mPFC mediating depressive-like behaviors. It's interesting to note that PPARγ decreased in mPFC, but not in hippocampus, indicating that PPARγ is regulated differently in neurons of various brain regions of mice under HFD. Moreover, after CSDS, susceptible mice present lower PPARγ protein levels than the control group, while resilient mice present higher PPARγ protein levels than the control group in the mPFC, but not in the hippocampus. The results suggest that PPARγ pathway in the mPFC plays a key role in the depression, and the mechanism of how PPARγ expression is regulated in neurons needs to be further studied.
HFD-induced PPARγ dysfunction is involved in the pathogenesis of metabolic disorders. For example, HFD induces cyclin-dependent kinase 5 (CDK5)-dependent PPARγ phosphorylation, which gives rise to dysregulated expression of a subset of genes including a number of key metabolic regulators, resulting in insulin resistance [10]. In addition, HFD disrupts the normal circadian cycle and causes a reprogramming of the metabolic and transcriptional Fig. 4 PPARγ decreased in the mPFC of susceptible mice after chronic social defeat stress. A Horizontal scatter plot illustrating the distribution of the interaction ratios of the control, susceptible, and resilient mice following the CSDS protocol, n = 10-16 mice per group. B Time spent in the interaction zone after 24 h of chronic defeat stress for the 3 subgroups, n = 10-16 mice per group. C-D Western blot images and quantification of PPARγ levels in the PFC, mPFC, and hippocampus of the three groups of mice following CSDS, n = 6 per group. E Correlation of PPARγ protein level in mPFC with social interaction ratio after CSDS (Pearson correlation, r = 0.646, p < 0.05). F The avoidance behaviors of the mice after 10 days of treatment with saline, imipramine or rosiglitazone (i.p.) following the CSDS. A 10-day treatment with rosiglitazone stably reversed the avoidance behaviors of defeated mice. n = 14-17 mice per group. Two-way ANOVA (interaction, F (3,57) = 6.464, p = 0.0008) followed by Holm-Sidak's multiple comparison test. Data are the means ± SEMs. *p < 0.05, **p < 0.01, and ***p < 0.001. ◂ liver pathways, partly mediated by inducing the de novo oscillation of PPARγ-mediated transcriptional control at otherwise non-cyclic genes [33]. Taken together, HFD not only induces peripheral PPARγ dysfunction mediating somatic manifestations, but also induces central PPARγ dysfunction mediating affective components of depression. Recently, varieties of evidences implicated that depression is a systemic disease with multiple physical symptoms rather than a mental illness. The above findings provide clues to understand why depression frequently occurs comorbidly with diabetes.
As to the high abundance and important role of PPARγ in white fat tissue, studies explored the mechanism how rosiglitazone exerts antidepressant effect in adipose tissue and found that it elicited antidepressant effect through the PPARγ-dependent induction of adiponectin [34,35]. However, the finding does not rule out the possibility that brain PPARγ exerts antidepressant effects. In our study, mPFC neuronal-specific PPARγ knockout is sufficient to induce depressive behaviors, and overexpression of PPARγ in mPFC neurons can rescue the depressive behaviors of HFDfed mice. The literature and our study suggest that PPARγ might play anti-depressant effect by both peripheral and central signaling activation.
In neurons, PPARγ regulates a series of key factors which have been reported to be involved in depression. PPARγ activation in hippocampal neuronal cells increases brain-derived neurotrophic factor (BDNF) expression, a key modulator in synaptic plasticity and depression [36]. PPARγ was also found to bind to the promoter of neurotrophic factor-alpha 1 and upregulate its expression in neurons [37]. Neurotrophic factor-alpha 1 has been shown to mediate its antidepressant effects through fibroblast growth factor 2 (FGF2) and neurogenesis [21]. Furthermore, mass spectrometry and bioinformatics analysis revealed that PPARγ agonists may induce the expression of synaptic vesicle-related genes, such as vesicle-associated membrane protein 2 (VAMP2) and syntaxin-binding protein 5 (STXBP5) [38], and the loss of synapses has been observed in depression. Neuronal PPARγ dysfunction may result in the abnormal expression of diverse neurotrophic factors and synapse-related genes, leading to transmitter release disorders and behavioral abnormalities. In addition, it is reported that PPARγ agonist rosiglitazone reduces corticotroph cell number, proopiomelanocortin (POMC), adrenocorticotrophic hormone (ACTH), and glucocorticoid receptors (GR) content in pituitary of diabetic rats [39]. PPARγ might also work on hypothalamus-pituitary-adrenal (HPA) axis to play antidepressant effect.
PPARγ signaling is well known for its role in inflammation and oxidative stress [40], both implicated in the pathophysiology of depression [41]. Due to the important roles of microglia and astrocytes in brain inflammation and oxidative stress, even our study suggested neuronal PPARγ dysfunction in mPFC leads to depression, it does not exclude the possibility that PPARγ signaling in astrocytes or microglia is involved in antidepressant effects.