STAT3 but Not ERK2 Is a Crucial Mediator Against Diet-Induced Obesity via VMH Neurons

Leptin plays an important role in the control of energy homeostasis. Through its actions in the central nervous system, more specifically in the hypothalamus, leptin promotes a state of negative energy balance, decreasing food intake and increasing energy expenditure (1–3). To exert its actions, leptin binds to its receptor (LepR) and activates different signaling pathways: STAT3, ERK1/2, and PI3K (4). Each one of these signaling pathways has been shown to play an important role in the control of energy balance in several genetically modified mouse models: deletion of STAT3 in LepR-expressing neurons induces extreme obesity (5); deletion of SHP2, a protein that mediates the activation of ERK1/2 via LepR, induces early-onset obesity (6); and mice with deletion of PTEN, a negative regulator of PI3K pathway, in LepR neurons present a leaner phenotype with decreased adiposity (7).

The ventromedial nucleus of the hypothalamus (VMH) mediates some effects of leptin on energy balance, and LepR is highly expressed in VMH neurons (8). Moreover, specific deletion of LepR in SF1 VMH neurons causes a marked increase in body weight of mice subjected to high-fat diet (HFD), due to increased food intake and decreased energy expenditure (9,10). These data suggest that VMH mediates some of leptin’s effects to protect against diet-induced obesity (DIO). However, little is known about the intracellular mechanisms by which VMH neurons mediate these effects. Xu et al. (11) showed that deletion of the P110α, a catalytic subunit of PI3K, in VMH neurons causes an important increase in body weight in animals submitted to HFD, an effect associated exclusively with decreased energy expenditure without any changes in food intake. A similar phenotype was recently observed in animals with deletion of another catalytic subunit of PI3K, P110β, in VMH neurons (12). Nevertheless, the obese phenotype of these animals with reduced PI3K activity in VMH did not recapitulate the higher body weight and increased food intake observed in mice on HFD with the deletion of LepR (9,10), indicating that other signaling pathways might be involved. In fact, the role of the other signaling pathways such as STAT3 and ERK2 to mediate VMH actions on energy balance remains undetermined. To fulfill this gap, we investigated the role of STAT3 and ERK2 in SF1 neurons of the VMH by assessing the effects of conditional deletions of these molecules on energy homeostasis.

Animal care and procedures were approved by the Institutional Animal Care and Use Committee at Ribeirao Preto Medical School. Mice were single housed at 22°C–24°C with a 12-h light/12-h dark cycle. SF1-cre, STAT3^flox, and ERK2^flox mice were generated as previously described (9,13,14). All mice were backcrossed for several generations with C57BL6 mice. Unless otherwise specified, mice were fed a standard chow diet with ad libitum access to food and water. For HFD studies, mice were given a 60% kcal fat diet (5.24 kcal/g, 60% kcal from fat; Research Diets, New Brunswick, NJ). For body weight and food intake recordings, animals were individually caged and fed with standard chow diet or HFD, with body weight and food intake weighed weekly. Littermates were used as controls for all studies.

Mice at 12 weeks of age were anesthetized intraperitoneally with ketamine/xylazine. ALZET Osmotic Pumps, model 2002 (DURECT Corporation, Palo Alto, CA), filled with indicated concentrations of leptin were implanted subcutaneously on day 0. After 6–12 days, mice were euthanized, and blood was collected. During the experiment, body weight and food intake were daily recorded.

Tail vein blood was collected at 10:00 p.m.–12:00 a.m. from ad libitum–fed mice. Blood glucose was assayed with OneTouch Ultra Blood Glucose Monitoring System (Thermo Fisher Scientific, Waltham, MA). ELISA kits were used to measure serum insulin (ALPCO, Salem, NH), leptin (R&D Systems, Minneapolis, MN), and estradiol (Calbiotech, El Cajon, CA) and in-house radioimmunoassay was used for testosterone measurement. Plasma corticosterone was measured as previously described (15).

Mice at 9 weeks were fasted overnight (12 h) and then were injected with glucose (0.75 g/kg body wt i.p.). Blood glucose levels were sampled from the tail nick at 15, 30, 60, and 120 min after injection.

Mice were individually housed at room temperature (22°C–24°C) with an alternating 12-h light/12-h dark cycle. After adaptation of 2 days, VO₂, VCO₂, and locomotor activity (total activity at x-axis) were measured for 24 h with the Comprehensive Lab Animal Monitoring System (CLAMS) (Columbus Instruments, OH). For standard chow diet experiments, we used body weight–matched animals at 12 weeks. For HFD experiments, we subjected body weight–matched animals at 12 weeks to HFD for 1 week and evaluated energy expenditure as described above.

Animals were injected with 5 mg/kg i.p. leptin, and 30 min after they were anesthetized and subjected to cardiac perfusion and the brain was collected. Coronal brain sections (25 μm) were obtained as previously described (16). STAT3 phosphorylation (pSTAT3) analysis was performed as previously described (16). Immunofluorescence for pERK1/2 was performed on free-floating sections. Sections were blocked in 0.01 mol/L PBS containing 10% normal horse serum, 0.1% Triton X-100, and 0.04% NaN₃ for 2 h at room temperature. Briefly, sections were incubated with the primary antibody anti-pERK raised in rabbit (cat. no. 9101, 1:500; Cell Signaling Technology, Danvers, MA) for 48 h at 4°C. After rinsing, the sections were incubated with a biotinylated secondary antibody donkey anti-rabbit (1:200; Vector Laboratories, Burlingame, CA) for 2 h at room temperature. After rinsing, the sections were incubated with Alexa Fluor 488–conjugated streptavidin (1:200; Jackson ImmunoResearch Laboratories, West Grove, PA) for 2 h at room temperature. Images were obtained in a Leica TCS SP5 Confocal microscope system equipped with a 488-nm (argon-krypton) laser line. For each group, all images were detected at identical acquisition settings.

All experimental statistical analyses were performed with GraphPad Prism 8 software (GraphPad, San Diego, CA). For two-group comparisons, the two-tailed Student t test was used. Repeated-measures ANOVA followed by Bonferroni test was used to compare changes over time between two groups. The significance level was set at P < 0.05.

The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

To generate animals with deletion of STAT3 specifically in SF1 neurons, we crossed SF1-cre mice (9) with STAT3^flox animals (13), generating the SF1-cre;STAT3^flox/flox mice. For control animals, we used STAT3^flox/flox littermate mice. We first validated the animals by confirming that STAT3^flox/flox mice exhibited the expected pSTAT3 in the VMH after intraperitoneal injection of leptin, whereas pSTAT3 was absent in the VMH of SF1-cre;STAT3^flox/flox but intact in other hypothalamic nuclei (Fig. 1A). Since SF1 is also expressed in the adrenal cortex (17), we measured plasma corticosterone levels of the SF1-cre;STAT3^flox/flox and STAT3^flox/flox animals to assess glucocorticoid secretion. We observed that STAT3 deletion did not affect the levels or the rhythm of corticosterone between the groups (Fig. 1B). SF1 is also expressed in gonads and gonadotrophs of pituitary gland (17). To assess whether deletion of STAT3 in the SF1-cre;STAT3^flox/flox mice would affect the function of reproductive tissues, we evaluated the weight of testis, seminal vesicle, uterus, and ovaries as well as measured plasma levels of testosterone and estradiol. We found similar values between SF1-cre;STAT3^flox/flox and STAT3^flox/flox mice at 8–12 weeks of age (Fig. 1C–H), suggesting that apparently the deletion did not have adverse effects on gonadal or gonadotroph function.

Disruption of STAT3 in SF1 neurons did not affect body weight, cumulative food intake, VO₂, or leptin levels in male and female SF1-cre;STAT3^flox/flox mice compared with STAT3^flox/flox mice on normal chow diet (Fig. 2A–H). We also showed that high rate of leptin infusion (300 ng/h) induced similar decrease in body weight and food intake in both groups (Fig. 2I–J).

SF1-cre;STAT3^flox/flox male and female mice when submitted to HFD presented a marked increase in body weight, food intake, and energy efficiency compared with STAT3^flox/flox control animals (Fig. 3A–C and E–G). The conditional deletion also resulted in increased serum leptin levels compared with those in control animals (Fig. 3D and H), strongly suggesting that the increase in body weight was secondary to greater body fat mass. Interestingly, in females, compared with male mice, the impact of STAT3 deletion in SF1 was more pronounced in body weight (35% vs. 23% at 15 weeks of age), food intake (21% vs. 14% at 15 weeks of age), and energy efficiency in comparison with their respective female and male controls subjected to HFD.

In addition, after 1 week of HFD, SF1-cre;STAT3^flox/flox female mice showed decreased VO₂ and VCO₂ in the light and dark cycles, but similar locomotor activity and respiratory exchange ratio (RER), compared with STAT3^flox/flox mice with paired body weight (mean body weight ± SD prior to the experiment: 25.4 ± 1.0 g for SF1-cre;STAT3^flox/flox vs. 23.9 ± 1.6 g for STAT3^flox/flox female mice). Deletion did not have an impact on VO₂, VCO₂, RER, or locomotor activity in SF1-cre;STAT3^flox/flox compared with STAT3^flox/flox male mice with paired body weight after 1 week of HFD (mean body weight ± SD prior to the experiment: 29.25 ± 3.6 g for SF1-cre;STAT3^flox/flox vs. 30.0 ± 3.5 for STAT3^flox/flox male mice). These findings help to explain the relatively higher increase in body weight and energy efficiency observed in female SF1-cre;STAT3^flox/flox mice (Fig. 4A–J).

Next, to evaluate the role of the ERK2 signaling pathway in VMH neurons, we generated animals with specific deletion of ERK2 in SF1 neurons. We confirmed that leptin was unable to induce ERK2 phosphorylation in the VMH of SF1-cre;ERK2^flox/flox mice, whereas ERK2^flox/flox mice showed an intact response (Supplementary Fig. 1A). We also showed that ERK2 deletion did not affect circadian corticosterone levels; seminal vesicle, uterus, or gonadal weight; or sexual hormone levels (Supplementary Fig. 1B–H).

SF1-cre;ERK2^flox/flox male and female mice did not show any difference in body weight, food intake, VO₂, or serum leptin levels compared with control animals under regular diet conditions (Fig. 5A–D and Supplementary Fig. 2A–D). Also, no difference was observed in body weight, food intake, VO₂, or leptin levels in male or female mice subjected to HFD (Fig. 5E–H and Supplementary Fig. 2E–H). Finally, SF1-cre;ERK2^flox/flox male mice had a similar response to high doses of leptin infusion on body weight and food intake reduction compared with control ERK2^flox/flox males (Supplementary Fig. 2I–J).

We also evaluated whether the deletion of STAT3 or ERK2 in VMH neurons is able to affect glucose homeostasis under regular diet and HFD conditions. We observed that neither SF1-cre;ERK2^flox/flox nor SF1-cre;STAT3^flox/flox male or female mice had changes in glycemia, glucose tolerance, or insulin levels compared with their respective controls, ERK2^flox/flox or STAT3^flox/flox animals on regular diet (Fig. 6A–C and F–H and Supplementary Fig. 3A–C and F–H). Similar results were also observed for glycemia in animals on HFD (Fig. 6D and I and Supplementary Fig. 3D and I). In animals subjected to HFD, serum insulin levels were elevated in male and female SF1-cre;STAT3^flox/flox mice compared with STAT3^flox/flox mice. In contrast, serum insulin levels were similar between SF1-cre;ERK2^flox/flox and ERK2^flox/flox in both sexes (Fig. 6E and J and Supplementary Fig. 3E and J).

VMH neurons mediate important actions of leptin in the protection against DIO. Deletion of LepR in SF1 neurons of VMH leads to increased body weight in animals submitted to HFD by causing a decrease in energy expenditure and an increase in food intake (9,10). However, the specific intracellular mechanisms involved in these actions still remain partially understood. Here we investigated the role of STAT3 and ERK2 pathways in energy homeostasis via VMH neurons and showed that deletion of STAT3 but not ERK2 in SF1 neurons causes impairment in energy homeostasis in animals challenged with HFD in a sex-specific pattern.

Deletion of STAT3 in LepR-expressing neurons results in an obese phenotype that is similar to that observed in mice lacking functional LepR (db/db mice) or leptin-deficient mice (ob/ob) (5). This study indicates that STAT3 is the main signaling pathway responsible for mediating leptin’s metabolic actions. So, our findings indicating that deletion of STAT3 did not have any impact on body weight in SF1-cre;STAT3^flox/flox mice on regular diet were unexpected. However, there are data suggesting that VMH neurons in fact do not exert important effects under normocaloric conditions. As an example, deletion of SF1, an essential factor for the development and function of VMH, does not affect body weight of mice fed with chow diet (18). In addition, Bingham et al. (10) showed that animals with deletion of LepR in SF1 neurons did not present any changes in body weight on regular diet. In this direction, we also showed intact leptin responsiveness to infusion of high levels of leptin in SF1-cre;STAT3^flox/flox mice. Based on these studies from the literature and our data, we hypothesize that leptin and its main signaling pathway STAT3 do not control energy balance on regular chow diet via VMH neurons.

Remarkably, deletion of STAT3 in SF1 neurons caused an important increase in body weight and food intake in animals submitted to HFD, a phenotype that was more pronounced in females that also presented decreased energy expenditure. These data support the relevant role of STAT3 in SF1 neurons to mediate leptin’s effects against DIO mainly by the control of food intake. Some studies have shown that SF1 neurons have a role in the control of food intake. As an example, deletion of LepR in SF1 neurons or dysfunction of glutamate release from these neurons causes increased food intake (9,10,19). Considering these findings from the literature, we can hypothesize that STAT3 signaling might be recruited in the LepR-mediated glutamate activation of high-order hypothalamic circuitry involved in the control of food intake in obesogenic conditions.

Previous studies have shown that animals with reduced PI3K activity in SF1 neurons have an increase in body weight when challenged with HFD; however, this impact on body weight was secondary only to reduction in energy expenditure, without changing food intake (11,12). Taking together the data from the literature and the current study, we can propose that, at least in males, STAT3 and PI3K mediate protection against DIO via VMH neurons by different mechanisms: the STAT3 signaling pathway is necessary to reduce food intake and the PI3K is required to increase energy expenditure. In fact, mice with LepR deletion in SF1 neurons show a marked increase in body weight on HFD secondary to both increased food intake and decreased energy expenditure, probably reflecting the absence of the activation of PI3K and STAT3 simultaneously by LepR in VMH neurons (9,10).

However, it seems that the body weight and energy efficiency start to diverge before food intake changes and remains elevated, suggesting that additional mechanisms are impaired in the SF1-cre;STAT3^flox animals besides hyperphagia. Our hypothesis is that this finding involves lipid metabolism and storage. VMH electrical stimulation leads to lipolysis of white adipose tissue (WAT), and the neurons of this nucleus can alter the response of WAT lipid metabolism to different diets (20–22). Also, leptin infusion can reduce body weight and lipogenesis (23). Based in these previous studies and our findings, we can hypothesize that a possible leptin-VMH-WAT axis could require the STAT3 signaling for the decreased lipolysis or increased lipogenesis in the SF1-cre;STAT3^flox animals.

Concerning the sex-specific phenotype, we suggest that the more prominent impairment in body weight, food intake, and VO₂ observed in SF1-cre;STAT3^flox/flox female mice under HFD could be explained by estradiol (E2) effects on energy homeostasis. E2 regulates body weight in females mainly by its actions on VMH neurons (24). In addition, it was previously shown that estradiol requires STAT3 to exert its effects on energy balance (25). Based on these data, we suggest that the deletion of STAT3 in females in our study is reflecting not only the lack of leptin’s effects but also the lack of estradiol effects in SF1 neurons on energy homeostasis, justifying the more pronounced impact on energy balance, including suppression of energy expenditure of female mice. In support of this hypothesis, Saito et al. (26) have recently shown that deletion of PI3K in SF1 neurons caused increased body weight in females even when subjected to regular diet but not in males (11). These authors postulated that this effect was possibly secondary to estradiol actions, indicating that in females, estradiol might be a confounding factor in our mouse models, since it has similar effects and activates the same signaling pathways of leptin in the control of energy balance.

LepR activation is negatively regulated by several molecules. One of these molecules is SOCS3, expression of which occurs in response to STAT3 signaling (27). In a negative feedback mechanism, SOCS3 binds to the LepR-JAK2 complex and decreases leptin signaling (27). Deletion of SOCS3 in SF1 neurons was shown to induce a reduction of food intake, but unexpectedly it also reduced energy expenditure, leading to unchanged body weight in male mice treated with chow and HFD (28,29). Lack of SOCS3 in SF1 neurons increases leptin-induced STAT3 phosphorylation (28), which might contribute to the decrease of food intake in these animals. These results from Zhang et al. (28) show that SOCS3 effects on energy expenditure are independent of STAT3 signaling in SF1 neurons of males. The results shown here reinforce that STAT3 in these neurons is not required for the regulation of energy expenditure in male mice. Also, considering the relevant effects of deletion of STAT3 in energy homeostasis observed in our study, we would expect that, in an opposite way, the animals with deletion of SOCS3 in SF1 neurons would have a more expressive protection against DIO. In our view, this minor effect on body weight found in those studies could be explained by the fact that there are other negative regulators of LepR signaling, such as PTP1B and TCPTP (30–32), that might be compensating the lack of SOCS3 action in those animals.

An important consideration is that since our study assessed the deletion of STAT3 in SF1 neurons, we cannot exclude that the phenotype found in the animals could also be explained by other mediators besides leptin and estradiol that could also activate STAT3. Several cytokines, like IL-6, are able to activate STAT3 and affect energy homeostasis (33,34); therefore, they could also be involved in the findings of the current study. Moreover, it is important to consider that although we showed that deletion of STAT3 in extrahypothalamic tissues that express SF1, like adrenal and gonads, did not affect hormone levels or reproductive organ sizes, we cannot completely rule out that the phenotype could be at least partially explained by changes in other features of those tissues (such as development and microarchitecture) not assessed in this study.

We also showed that deletion of ERK2 in SF1 neurons does not affect energy homeostasis in both normocaloric and high fat conditions. These data indicate that the ERK2 signaling pathway does not have an important role in the control of energy homeostasis via VMH neurons. Rahmouni et al. (35) showed that the preadministration of ERK pathway inhibitor blocked the hypophagic effects of leptin injected in the third ventricle of rats. Also, deletion of SHP2, a protein that mediates the activation of ERK1/2, causes an early obesity phenotype in mice (6). Hence, we suggest that these effects of ERK2 on energy homeostasis previously described possibly involve other hypothalamic nuclei. A potential candidate is the arcuate nucleus of the hypothalamus, since it is known that the ERK activation in this nucleus is influenced by different nutritional triggers (36,37). However, it is important to consider that we have investigated in this study only the effects of deletion of ERK2, and therefore we cannot rule out a compensatory mechanism through the ERK1 pathway due to the deletion of ERK2 in SF1 neurons.

Our study showed that lack of STAT3 and ERK2 in SF1 neurons did not have an impact on basal glycemia, glucose tolerance, or insulin levels in male or female mice on chow. Minokoshi, Haque, and Shimazu (38) have shown that VMH neurons regulate glucose uptake in peripheral tissues, and this effect could involve STAT3 and ERK2 (39). However, other studies have demonstrated that deletion of LepR in SF1 neurons (9) or reexpression of LepR in the otherwise LepR-deficient mice (40) does not affect glucose homeostasis, supporting our data that leptin or its signaling pathways activated by LepR are not important in the control of glucose homeostasis via VMH neurons in normocaloric conditions. Concerning the animals submitted to HFD, we found that deletion of STAT3, but not ERK2, causes an increase in insulin levels of SF1-cre;STAT3^flox animals. However, because of the confounding factor of the early difference in body weight between control and study animals, we were not able to perform more sensitive tests, like glucose tolerance tests and insulin tolerance tests, to differentiate whether the animals developed more severe insulin resistance or the increased levels of insulin were secondary to the higher body mass.

In summary, our study shows that STAT3 but not ERK2 plays a crucial role in the protection against DIO via VMH neurons. These results add further understanding of the dual role of VMH to protect against DIO through at least two pathways: PI3K in the control of energy expenditure (11) and STAT3 in the control of food intake. A mouse model with the double deletion of STAT3 and PI3K in SF1 neurons would be helpful to prove the hypothesis that these two pathways are not redundant and in fact control energy balance by different mechanisms and also to evaluate whether the double knockout mice would recapitulate the phenotype of the LepR-deficient mice in the VMH (9,10).

Acknowledgments. The authors are grateful to Milene Mata, Maria Valci dos Santos, Vanessa da Cunha and Carla de Melo for their help with experimental protocols and procedures.

Funding. This work was supported by grants from the Sao Paulo Research Foundation (Fundação de Amparo à Pesquisa do Estado de São Paulo [FAPESP], Brazil: 2018/10090-0 scholarship to G.H.M.G. and grants 2014/17248-8 and 2013/09799-1), the National Council for Scientific and Technological Development (CNPq), Coordination for Enhancement of Higher Education Personnel (CAPES), and Fundação de Apoio ao Ensino, Pesquisa e Assistência (FAEPA) (Brazil).

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

Author Contributions. G.H.M.G. and L.L.K.E. designed the study. G.H.M.G. performed the experiments and wrote the manuscript. G.H.M.G., M.d.C., J.A.-R., and L.L.K.E. interpreted the data. R.E.V. and S.M.T. helped with the experiments. G.A.-P. and B.d.C.B. performed the immunofluorescence studies. J.D. provided some of the breeder animals. All coauthors revised the manuscript. G.H.M.G. and L.L.K.E. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.