Hijacking Dorsal Raphe to Improve Metabolism and Depression-Like Behaviors via BDNF Gene Transfer in Mice

Depression and metabolic disorders, such as diabetes and obesity, are major public health concerns in the modern world. Accumulating evidence shows that a bidirectional relationship exists between depression and diabetes/obesity (1,2). While there is high comorbidity of both conditions, each one increases the risk of developing the other. Long-term use of various antidepressants can cause significant weight gain and increase the relative risk of type 2 diabetes (3). Conversely, neuropsychiatric adverse events, including depression, have emerged as a major issue in developing antiobesity drugs (4). Mechanisms underlying the intertwined links between both conditions remain largely elusive, possibly accounting for the current lack of effective measures to manage comorbid depression and diabetes/obesity. A common feature of the pathophysiology of depression and diabetes/obesity in the brain is synaptic dysfunction (5–7). This suggests that a synaptic repair strategy may be used to combat comorbid depression and diabetes/obesity, which targets pathophysiology rather than pathogenesis.

Brain-derived neurotrophic factor (BDNF) is well known for its role in regulating synaptic plasticity in a variety of ways, shaping the structure and function of neural circuits across the life span (8). A BDNF-based synaptic repair strategy has been proposed for neurodegenerative diseases (9). However, the actions of BDNF in the brain on moods and metabolism are region-specific. BDNF action in the ventral tegmental area-nucleus accumbens pathway is sufficient to develop a depression-like phenotype in rodents, in contrast to its antidepressant role in the hippocampus (10,11). BDNF deletion in the amygdala leads to anxiety but reduces weight gain induced by a high-fat diet (HFD) as a result of increased energy expenditure, in contrast to its antidiabetes/obesity role in the hypothalamus (12–15).

The dorsal raphe nucleus (DRN) is located on the midline of the brainstem and beneath the aqueduct of the midbrain. It contains the largest group of serotonergic neurons in the brain and is a neurochemically heterogeneous nucleus with widespread projections mainly to the forebrain, including the limbic system regulating emotions and the hypothalamus controlling energy balance (16,17). Chemogenetic studies show that activation of DRN serotonergic neurons produces antidepressant-like behavioral responses (18,19). Optogenetic stimulation of the DRN projection to the lateral habenula alleviates depressive symptoms, whereas inhibition induces depression-like behaviors in the chronic unpredictable mild stress (CUMS) rat model of depression (20). In the DRN, interfering with TrkB (tropomyosin-related kinase B, the high-affinity receptor of BDNF) leads to loss of antidepressant efficacy and to abnormal aggression in mice (21). On the other hand, chemogenetically silencing DRN synapses formed with projections from the paraventricular nucleus of the hypothalamus is sufficient to elicit rapid food intake (22). γ-Aminobutyric acid (GABA)ergic and glutamatergic neurons in the DRN exert opposite role in regulating energy balance (23), consistent with previous pharmacological studies (24). Taken together, the role of the DRN in regulating both moods and metabolism makes it a possible candidate site to implement the BDNF-mediated therapy for comorbid depression and diabetes/obesity.

In this study, we hypothesized that BDNF gene transfer in the DRN improves metabolism and depression-like behaviors. We tested this in the CUMS mouse model and models of diabetes and obesity. The CUMS model is one of the most widely used animal models of depression with high validity and translational potential. In this model, animals chronically experience various unpredictable mild stressors, leading to anhedonia and behavioral despair (25). In addition, we used chemogenetic tools to activate neurons via recombinant adeno-associated virus (rAAV)-mediated hM3Dq expression in the DRN and found effects in DIO mice similar to those produced by BDNF.

All animal studies were approved by the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences Institutional Review Board. Adult male C57BL/6J mice (Beijing Vital River Laboratory Animal Technology Co., Ltd., Beijing, China), male leptin receptor mutant db/db mice (Changzhou Cavens Laboratory Animal Co., Ltd., Changzhou, China), and male Tph2^f/f mice (26) were used for this study.

The CUMS model was established as described previously (27). The total experimental design is shown in Fig. 2 A. The daily schedule of the CUMS paradigm is summarized in Supplementary Table 1.

The rAAV-BDNF vector was constructed as described previously (28) and packaged with the capsid of AAV type 9 (Taitool Bioscience). Then, 1 μL of rAAV with a titer of 3–4 × 10⁹ viral genome/µL was injected into the DRN (2.3 mm posterior to bregma with a vertical tilt of 30°, 3.7 mm of injection depth) at a rate of 300 nL ⋅ min⁻¹ using a Micro4 Pump Controller (World Precision Instruments). For the hippocampus, 1 μL of rAAV per side was injected bilaterally (2.2 mm posterior to the bregma, 2.2 mm lateral to the midline, 1.8 mm injection depth).

For immunoblot analysis, brain tissues were lysed by sonication in radioimmunoprecipitation assay lysis buffer with a protease inhibitor cocktail (Sigma-Aldrich), separated by SDS-PAGE gel (10%) and blotted onto nitrocellulose membranes. The following primary and secondary antibodies were used: anti-BDNF antibody (1:200, Alomone Laboratories), anti-GAPDH antibody (1:10000, Sigma-Aldrich), anti-BDNF antibody (1:1000, Abcam), anti-TrkB (phospho Y515) antibody (1:1000, Abcam), anti-TrkB antibody (1:1000, Abcam), anti–β-actin antibody (1:5000, Cell Signaling), and anti-rabbit IgG conjugated with horseradish peroxidase (1: 5000, Promega).

For the sucrose consumption test, mice were preconditioned with a 1% sucrose solution for 48 h. Then each mouse was singly housed and deprived of water for 12 h. In the following 1-h test, the amount of the 1% sucrose solution consumed was recorded. For the water consumption test, there was no preconditioning step and the solution was the regular drinking water.

Mice were dropped individually into a glass cylinder (height: 19 cm; diameter: 14 cm) containing 14 cm-height water maintained at 23–25°C and remained there for 6 min. Immobility was judged as the mouse floated in the water except small movements to keep its head above the water. The duration of immobility was recorded during the last 4 min of the testing period.

Mice were individually placed in a square Plexiglas box (50 cm × 50 cm × 40 cm) in a brightly lit room and remained there for 10 min. The time spent in the center area of the box (12.5 cm × 12.5 cm) and the distance traveled in the box during the testing period were recorded and analyzed by the EthoVision video tracking system (Noldus).

Mice were individually placed in a circular elevated maze with two open and two enclosed arms (height: 43.5 cm; diameter: 45 cm; track width: 6.5 cm) and remained there for 5 min. The time spent in the open arms and the distance traveled in the maze during the testing period were recorded and analyzed by the EthoVision video tracking system (Noldus).

Liver and adipose tissues were fixed with 4% paraformaldehyde and embedded in paraffin. Hematoxylin-eosin (HE) staining was performed on paraffin-embedded sections, and Oil Red O staining was performed on frozen liver sections using an Oil Red O solution (Sigma-Aldrich). Both were performed by the Histology Core at Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences.

For corticosterone, blood was collected from a tail snip before and after the restraint stress exposure. Serum was prepared by allowing the blood to clot for 30 min on ice, followed by centrifugation. For other biomarkers, blood was collected when mice were sacrificed by rapid anesthetization with an intraperitoneal (i.p.) injection of chloral hydrate (10%, wt/vol). According to the manufacturer’s protocol, commercially available kits were used to determine levels of corticosterone (Abcam), leptin (Abcam,), insulin (ALPCO), triglyceride (Wako, Osaka, Japan), free fatty acid (Wako), cholesterol (Wako), glucagon (R&D Systems), and thyroxine (Calbiotech).

Blood glucose levels were measured using an ACCU-CHEK Active Blood Glucose Meter (Roche) on blood from a tail snip. For the glucose tolerance test, mice were fasted overnight (16 h). Glucose (1 g/kg) was administered i.p., and blood glucose levels were measured at 0, 15, 30, 60, and 120 min. For the insulin tolerance test, mice were fasted for 6 h. Insulin (1 unit/kg) was administered i.p., and blood glucose levels were measured at 0, 15, 30, 60, and 120 min.

We measured Vo₂, Vco₂, physical activity, and the respiratory exchange ratio using the Comprehensive Lab Animal Monitoring System (CLAMS-8, Columbus Instruments). Food and water were available ad libitum. Mice were individually housed and allowed to be habituated to the instrument for 12 h. During the following 60 h, the physiological and behavioral parameters were recorded. Data were analyzed using the Oxymax V4.60 software (Columbus Instruments).

This was performed as described previously (28). Primers used in this study are provided in Supplementary Table 2.

This was performed as described previously (28). The following primary and secondary antibodies were used: anti-TPH2 (1:800, Thermo Fisher Scientific), anti–c-Fos (1:2000, Cell Signaling Technology), and anti-rabbit IgG conjugated with Alexa Fluor 594 (1:1000, Thermo Fisher). All images were captured using a confocal laser scanning microscope (Zeiss LSM780).

Two hours after i.p. injection of clozapine N-oxide (CNO) (1 mg/kg) or vehicle solution in mice of different experimental groups, animals were sacrificed by rapid anesthetization with chloral hydrate (10%, wt/vol, i.p.). For chronic activation of DRN neurons, CNO was administered in the drinking water at a concentration of 40 mg/L. The solution was freshly prepared every second day.

All data are represented as mean ± SD. Statistical significance was analyzed with GraphPad Prism 7 software. Unpaired two-tailed Student t tests and two-way ANOVA were used. P < 0.05 was considered statistically significant. All the data were exported into Adobe Illustrator CS5 for preparation of figures.

All data and resources related to the current study are available from the corresponding author upon reasonable request.

We delivered mouse BDNF to the DRN with EGFP (AAV-EGFP) as a control (Fig. 1A–C). Western blot assay further confirmed the higher expression level of BDNF protein in the DRN of BDNF-expressing mice than that in EGFP-expressing mice (Fig. 1D). We detected elevated protein levels of BDNF in the projection areas of DRN axons and also enhanced activation of TrkB in the hypothalamus (Supplementary Fig. 1). One month after virus microinjection, the weight of body and major fat depots was significantly reduced in BDNF-expressing mice compared with control mice (Fig. 1E and F). Food intake by BDNF-expressing mice was significantly less than that by control mice (Fig. 1G). The blood glucose level in unfasted BDNF-expressing mice was significantly lower than that in control mice, whereas there was no significant difference between the two groups after 16 h of food deprivation (Fig. 1H). After 1 h of acute restraint stress exposure, the change in the serum level of corticosterone was higher in EGFP-expressing mice than that in BDNF-expressing mice (Fig. 1I).

We used the CUMS mouse model of depression to assess whether BDNF gene transfer in the DRN improves depression-like behaviors (Fig. 2A). The sucrose solution consumed reflects the anhedonia state of mice after exposure to chronic stress. Before stress, there was no significant difference in the baseline sucrose consumption among the four experimental groups (Supplementary Fig 2A). CUMS resulted in reduced water consumption in both EGFP- and BDNF-expressing mice; however, there was no significant difference between the two groups with or without stress (Fig. 2B), indicating that BDNF gene transfer does not affect normal fluid intake. After CUMS, sucrose consumption in EGFP-expressing mice was significantly decreased compared with that in naïve EGFP-expressing mice (Fig. 2C), indicating an anhedonic phenotype. BDNF gene transfer significantly increased sucrose consumption in CUMS-treated mice compared with that in CUMS-treated EGFP-expressing mice (Fig. 2C), indicating resilience to CUMS. The duration of immobility in the forced swimming test is considered to reflect behavioral despair or the development of a passive coping strategy in response to stress. The forced swimming test also reveals BDNF overexpression in the DRN exerts antidepressant effects (Fig. 2D).

Depression is often associated with anxiety, and one of the main precipitating factors of anxiety is stress. In this study, we further examined anxiety-related behaviors using the open field test (OFT) and elevated zero maze test, both detecting reactivity to environmentally induced conflict situations. In the OFT, there was no significant difference in the total distance among the four experimental groups (Supplementary Fig. 2B). There was no significant difference in the duration spent by animals in the central area in the OFT between naive control-EGFP and control-BDNF groups, whereas CUMS caused marked decrease in the central duration in the CUMS-EGFP group compared with that in the control-EGFP group (Fig. 2E), indicating an anxiety-like phenotype. Interestingly, the central duration was significantly increased in the CUMS-BDNF group compared with that in the CUMS-EGFP group (Fig. 2E), indicating an anxiolytic effect. Similar results were found in the elevated zero maze, confirming that BDNF gene transfer in the DRN has anxiolytic effects after exposure to CUMS (Supplementary Fig. 2C, Fig. 2F).

Next, we tested whether BDNF expression in the DRN plays a role in improving metabolism in the DIO model. Before HFD feeding, the body weight of BDNF-expressing mice remained relatively stable, whereas that of EGFP-expressing mice increased steadily (Fig. 3A). After HFD feeding, the weight gain of control mice accelerated rapidly, whereas BDNF-expressing mice still kept a stable weight (Fig. 3A and B), indicating that BDNF gene transfer induces robust resistance to HFD-induced obesity. Gross anatomy showed markedly reduced accumulation of pericardial and abdominal fat and absence of hepatic steatosis in BDNF-expressing mice (Fig. 3C). The weight of major fat pads in BDNF-expressing mice was also greatly reduced (Fig. 3D). HE staining and Oil red O staining on liver sections showed BDNF gene transfer in the DRN prevented the liver steatosis induced by HFD feeding (Fig. 3E). BDNF expression significantly reduced serum levels of cholesterol, triglyceride, insulin, and leptin (Fig. 3F, G, I, and J), but not free fatty acid (Fig. 3H). BDNF-expressing mice showed better glucose tolerance (Fig. 3K) and insulin sensitivity (Fig. 3L), indicating that BDNF expression restrains hyperglycemia induced by HFD feeding.

These observations revealed that BDNF-expressing mice are resistant to HFD-induced obesity, which may result from reduced energy intake or elevated energy expenditure. BDNF-expressing mice consumed fewer high-fat pellets compared with control mice (Fig. 3M). Vo₂ and Vco₂ in BDNF-expressing mice were markedly increased during both the dark and the light phases (Fig. 3N and O), indicating enhanced energy expenditure. BDNF-expressing mice exhibited more physical activity than control mice (Fig. 3P). Notably, BDNF-expressing mice showed a higher respiratory exchange ratio compared with that in control mice (Fig. 3Q), indicating that BDNF overexpression in the DRN increases carbohydrate utilization as opposed to lipid oxidation.

We further examined the effects of BDNF on metabolism in db/db mice, a genetic model of obesity and phases I to III of diabetes type 2. BDNF gene transfer in the DRN significantly reduced the body weight of db/db mice (Fig. 4A and B). Gross anatomy showed reduced accumulation of abdominal fat and greatly alleviated hepatic steatosis in BDNF-expressing mice (Fig. 4C). BDNF expression also significantly reduced serum levels of cholesterol, free fatty acid, insulin, and leptin (Fig. 4D, F–H), but not triglyceride (Fig. 4E). BDNF-expressing mice showed better glucose tolerance (Fig. 4I) and insulin sensitivity (Fig. 4J). BDNF gene transfer in the DRN prevented the development of hyperphagia in db/db mice (Fig. 4K).

Next, we performed a molecular phenotyping profile of the DRN, hypothalamus, liver, and white and brown adipose tissues by real-time quantitative PCR. In DRN tissues of standard diet (SD)-fed, HFD-fed, and db/db mice, BDNF gene transfer consistently increased expression of Tph2 and Vmat2, the former encoding the rate-limiting enzyme in serotonin synthesis and the latter being responsible for transporting monoamines from cellular cytosol into synaptic vesicles (Fig. 5A–C), implicating enhanced serotonin synthesis and intracellular transport.

In the hypothalamus, expression levels of Pomc and Cartpt, encoding anorexigenic proopiomelanocortin and CART prepropeptide, respectively, were significantly decreased in BDNF-expressing SD and HFD mice, whereas Agrp and Npy, encoding two orexigenic peptide hormones, were upregulated in BDNF-expressing HFD mice (Fig. 5D–F). These changes may reflect a compensatory response to the reduced body weight and fat content, a similar phenotype observed in a previous report (12). Melanocortin 4 receptor (encoded by Mc4r) is a central integrator coordinating energy intake and expenditure, glucose homeostasis, and autonomic outflow (29). Expression levels of Mc4r were significantly upregulated in the hypothalamus of the BDNF-expressing mice (Fig. 5D–F). In addition to participating in stress response, corticotropin-releasing hormone (CRH) also modulates energy balance and feeding behavior (30,31). Expression levels of Crh were also significantly elevated in the hypothalamus of the BDNF-expressing mice compared with control mice (Fig. 5D–F). Expression levels of Trh, encoding another anorexigenic factor, thyrotropin-releasing hormone (TRH) (32), were upregulated as well in the hypothalamus of the BDNF-expressing SD and db/db mice (Fig. 5D–F); however, circulating levels of thyroxine were not significantly different between EGFP- and BDNF-expressing mice (Supplementary Fig. 3), suggesting that TRH may act centrally to modulate metabolism.

Intracerebroventricular infusion of BDNF in rats attenuates diabetic hyperglycemia induced by streptozotocin, which is associated with a decrease in glucagon secretion and hepatic expression of gluconeogenic genes (33). In our study, there was no significant change in expression levels of hepatic G6pc (encoding glucose-6-phosphatase, one of the key gluconeogenic enzymes) between EGFP- and BDNF-expressing mice, whereas Pck1 (encoding phosphoenolpyruvate carboxykinase, a main control enzyme regulating gluconeogenesis), was markedly upregulated in BDNF-expressing SD, HFD, and db/db mice, suggesting a compensatory response to the reduced blood glucose level (Fig. 5G–I). Circulating levels of glucagon were not significantly different between EGFP- and BDNF-expressing mice (Supplementary Fig. 4). These results indicate that other mechanisms underlie the regulation of blood glucose induced by BDNF gene transfer in the DRN.

BDNF gene transfer in the hypothalamus prevents HFD-induced hepatic steatosis by regulating the expression of genes involved in fat metabolism in the liver (12). In our study, we observed expressions of lipogenic genes, including Fasn (encoding fatty acid synthase), Gpam (encoding mitochondrial glycerol-3-phosphate acyltransferase), and Scd1 (encoding stearoyl-CoA desaturase), were downregulated (Fig. 5G–I), whereas expressions of Cpt1a (encoding carnitine palmitoyltransferase 1A) and Acox1 (encoding acyl-CoA oxidase-1, palmitoyl) showed opposite changes in the liver of BDNF-expressing SD and HFD mice (Fig. 5G and H). Expression of Pparg (peroxisome proliferator-activated receptor-γ, a key player involved in regulating fatty acid storage and glucose metabolism) was markedly downregulated in the liver of BDNF-expressing mice (Fig. 5G–I). Expression of hepatic Ucp2 (encoding mitochondrial uncoupling protein 2 involved in lipid oxidation) was significantly upregulated in BDNF-expressing SD and HFD mice (Fig. 5G and H). In BDNF-expressing mice, expression of hepatic Insr (encoding insulin receptor) was uniformly upregulated (Fig. 5G–I). The results may contribute to interpreting the phenotypes observed such as reduced hepatic steatosis in BDNF-expressing mice. A recent report shows that sympathetic innervation in the liver may play a critical role in regulating metabolism (34). Whether the effects of BDNF on the liver are mediated by activation of the sympathetic output remains to be determined.

HE staining showed smaller adipose cells on white adipose tissue (WAT) sections and smaller lipid droplets on brown adipose tissue (BAT) sections in BDNF-expressing HFD mice (Fig. 4J). In WAT, BDNF expression greatly upregulated Adrb3 (encoding β-adrenergic receptor 3) expression, compared with control mice (Fig. 5K). In BAT, BDNF expression markedly upregulated expression of Adrb1 and Adrb2 in HFD mice compared with control mice (Fig. 5L). This suggests that enhanced sympathetic signaling may, at least partly, contribute to the reduced fat mass and increased energy expenditure in BDNF-expressing mice.

Although central serotonergic neurons have been demonstrated to regulate glucose and lipid homeostasis (35), whether DRN serotonin plays a critical role needs to be clarified. Here, we specifically knocked out Thp2 in the DRN by rAAV-mediated Cre expression in Tph2^f/f mice (Fig. 6A). The results showed that DRN-specific knockout of Tph2 cannot block the effects of BDNF on reducing the body weight and improving glucose tolerance and insulin sensitivity in the DIO model (Fig. 6B-D), suggesting that BDNF may act downstream of DRN serotonin signaling in regulating metabolism.

Given that BDNF promotes excitatory neurotransmission (8), we investigated whether activation of DRN neurons improve metabolism as BDNF. We used the chemogenetic tool to activate DRN neurons (Fig. 7A and B). After i.p. injection of the synthetic ligand CNO, DRN neurons were selectively activated in hM3Dq-expressing mice (hM3Dq-CNO) as revealed by c-Fos immunostaining (Fig. 7C). Chronic activation of DRN neurons reduced the body weight and alleviated hepatic steatosis as well as demonstrated better glucose tolerance and insulin sensitivity in the DIO model (Fig. 7D–H).

Here we report that BDNF gene transfer in the DRN exerts antidepressant and anxiolytic effects in the CUMS mouse model, prevents the development of HFD-induced obesity, and improves diabetes/obesity in db/db mice. Our study provides preclinical evidence for BDNF gene transfer in the DRN to improve metabolism and depression-like behaviors in various animal models.

In the adult brain, BDNF can be anterogradely transported from the cell body to the axonal terminals to act on targeted neurons after release (36) as well as can be retrogradely transported to cell bodies of afferent neurons after binding to TrkB localized at axon terminals in the BDNF-releasing site (37). Thus, there are four modes by which BDNF-TrkB signaling pathways can exert their biological effects: locally by acting on the BDNF-expressing neuron (autocrine) or on surrounding neurons (paracrine) and remotely by acting on targeted neurons via anterograde axonal transport or on afferent neurons via retrograde axonal transport. Therefore, in our study, BDNF expression in the DRN may act on DRN neurons and also on their widespread input-output systems, including the hippocampus and the hypothalamus (16,17,38,39).

Although TrkB mRNA is highly expressed in the DRN, no mRNA signal of Bdnf is detected by in situ hybridization (40), implicating that the lightly stained immunoreactive signal of BDNF observed in this area may originate via anterograde/retrograde axonal transport from DRN afferent/efferent systems such as the hippocampus or the hypothalamus. Consistent with the expression patterns, conditional knockout of Bdnf in the DRN displays no behavioral abnormalities, whereas DRN-specific knockout of TrkB results in loss of antidepressant efficacy and increased aggressive-like behavior (21). Thus, endogenous BDNF in the DRN acts in a noncell autonomous way to elicit downstream signaling pathways, which may exert quite different effects from the exogenously administered or overexpressed BDNF as demonstrated in other and our studies (41,42).

Infusion of BDNF protein into the midbrain, near the periaqueductal gray and dorsal/median raphe nuclei, produces an antidepressant effect in the learned helplessness rat model of depression (41), which is accompanied by augmented serotonergic activity in the infusion site (42) and in the forebrain, including the hippocampus and the hypothalamus (43). The increased serotonergic activity is related to BDNF-induced altered firing patterns of DRN neurons (44) and enhanced expression of tryptophan hydroxylase (45). Our study also found that expression levels of Tph2 and Vmat2, responsible for serotonin synthesis and intracellular transport, respectively, were significantly elevated in BDNF-expressing mice, implicating an augmented serotonergic activity. BDNF promotes the survival, growth, and differentiation of serotonergic neurons, and heterozygous Bdnf^+/− mice develop brain serotonergic abnormalities (46). Brain 5-hydroxytryptamine deficiency results in stress vulnerability and blunts responses to fluoxetine in a social defeat stress model of depression (47). Thus, the antidepressant/anxiolytic effects of BDNF we observed in the CUMS model may be related to the enhanced serotonergic activity in the brain.

BDNF is the most widely expressed neurotrophin in the brain, with high abundance in the hypothalamus (40). The hypothalamus and the DRN reciprocally send projections to each other (16,17,38,39), implicating that BDNF gene transfer in the DRN may regulate metabolism by targeting the hypothalamus. By molecular profiling of the hypothalamus, we found increased expression of key genes regulating metabolism such as Crh, Trh, and Mc4r. Antagonizing the CRH signal can counteract the effects of an intracerebroventricular BDNF infusion on reducing food intake and body weight (30). CRH administered via an intracerebroventricular infusion increases plasma levels of epinephrine and norepinephrine, which is associated with an increase in locomotor activity and Vo₂ (31). Apart from its effect on regulating thyroid function, TRH can also act in the brain to regulate feeding behavior, thermogenesis, locomotor activity, and autonomic function (32). The melanocortin 4 receptor–related network plays a critical role in maintaining energy homeostasis (29). Energy expenditure is regulated by the sympathetic nervous system, which heavily innervates both BAT and WAT (48–53). In the brain, CRH, TRH, and melanocortin 4 receptor in common can increase sympathetic activity (29,31,32), which in turn promotes WAT lipolysis and beiging as well as BAT thermogenesis (48–53). Our study shows that BDNF gene transfer in the DRN increases the expression of β-adrenergic receptors in BAT and WAT, implicating enhanced sympathetic activity that leads to increased energy expenditure.

BDNF may also act locally to facilitate excitatory neurotransmission. Previous study reports that microinjection of muscimol, a GABA-A receptor agonist, into the DRN induces intense feeding (24). Consistently, a recent study reports that in the DRN, activation of glutamatergic neurons decreases feeding, whereas activation of GABAergic neurons has opposite effects (23). Global activation of DRN neurons via chemogenetic tools in this study improves obesity and glucose intolerance in DIO mice. These data suggest that there is a balance in the DRN between inhibitory and excitatory neurotransmissions, which controls normal feeding and body weight. In vitro studies have shown that BDNF enhances the efficacy of excitatory synapses but depresses that of GABAergic synapses, which may be mediated by BDNF-induced changes in presynaptic transmitter release, postsynaptic responses, and expression of GABA/glutamate receptors (8). Thus, BDNF gene transfer in the DRN may bias the inhibitory and excitatory balance toward excitatory neurotransmission, reducing food intake and body weight.

In summary, DRN gene transfer of BDNF improves metabolism and depression-like behaviors, which can be exploited to develop new therapeutic interventions for comorbid depression and diabetes/obesity. Specific mechanisms remain to be clarified, especially regarding the intracellular signaling pathways downstream of TrkB activation, as well as changes in synaptic plasticity both in and out of the DRN region.

Funding. This work was supported by the research grants from the National Key Research and Development Program of China (2016YFC1306700), the National Natural Science Foundation of China (81625008, 81930104, and 31970952), Beijing Municipal Science & Technology Commission (Z181100001518001), CAMS Innovation Fund for Medical Sciences (2016-I2M-1-004), Science and Technology Program of Guangdong (2018B030334001), and the Fundamental Research Funds for the Central Universities (3332018001) and by Collaborative Innovation Program of Shanghai Municipal Health Commission (2020CXJQ01), Shanghai Municipal Science and Technology Major Project (No. 2018SHZDZX01), and ZJLab to Y.-Q.D.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.X. and R.H. performed most of the experiments in this study. J.X., Y.-Q.D., and Q.X. designed the experiments and wrote the manuscript. Z.L. contributed to the real-time quantitative PCR analysis of gene expression. J.L and S.L. did the Western blot assay. Y.S. and Q.X. supervised the project. Q.X. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.