Estradiol Regulates Energy Balance by Ameliorating Hypothalamic Ceramide-Induced ER Stress

Article Estradiol Regulates Energy Balance by Ameliorating Hypothalamic Ceramide-Induced ER Stress Graphical Abstract Highlights dCentral E2 regulates BAT thermogenesis via the sympathetic nervous system dCentral E2 reduces hypothalamic ceramides and ER stress dHypothalamic ceramides/ER stress mediate E2 actions on BAT Authors Ismael Gonza ´lez-Garcı ´a, Cristina Contreras, A ´nxela Este ´vez-Salguero, ..., Christophe Magnan, Manuel Tena-Sempere, Miguel Lo ´pez Correspondence [email protected] In Brief Gonza ´lez-Garcı ´a et al. demonstrate that estradiol (E2) acts in a precise area of the hypothalamus, named the ventromedial nucleus (VMH), to regulate brown fat thermogenesis. The actions of E2 are mediated by modulation of hypothalamic ceramides and ER stress. Gonza ´lez-Garcı ´a et al., 2018, Cell Reports 25, 413–423 October 9, 2018 ª2018 The Author(s). https://doi.org/10.1016/j.celrep.2018.09.038 Cell Reports Article Estradiol Regulates Energy Balance by Ameliorating Hypothalamic Ceramide-Induced ER Stress Ismael Gonza ´lez-Garcı ´a, 1,2 Cristina Contreras, 1,2 A ´nxela Este ´vez-Salguero, 1,2 Francisco Ruı ´z-Pino, 2,3,4 Benoit Colsh, 5 Iva ´n Pensado, 1,2 Laura Lin ˜ares-Pose, 1,2 Eva Rial-Pensado, 1,2 Pablo B. Martı ´nez de Morentin, 1,2 Johan Fernø, 6 Carlos Die ´guez, 1,2 Rube ´n Nogueiras, 1,2 Herve ´Le Stunff, 7,8 Christophe Magnan, 8 Manuel Tena-Sempere, 2,3,4,9 and Miguel Lo ´pez 1,2,10, * 1 Department of Physiology, CiMUS, University of Santiago de Compostela-Instituto de Investigacio ´n Sanitaria, Santiago de Compostela, 15782, Spain 2 CIBER Fisiopatologı ´a de la Obesidad y Nutricio ´n (CIBERobn), Santiago de Compostela, 15706, Spain 3 Department of Cell Biology, Physiology, and Immunology, University of Co ´rdoba, Co ´rdoba, 14004, Spain 4 Instituto Maimo ´nides de Investigacio ´n Biome ´dica (IMIBIC)/Hospital Reina Sofı ´a, Co ´rdoba, 14004, Spain 5 CEA-Centre d’Etude de Saclay, Laboratoire d’e ´tude du Me ´tabolisme des Me ´dicaments, Gif-sur-Yvette, France 6 Hormone Laboratory, Haukeland University Hospital, Bergen, 5021, Norway 7 Paris-Saclay Institute of Neuroscience, CNRS UMR 9197, Universite ´Paris-Sud, University Paris Saclay, Orsay 91405 Cedex, France 8 Unite ´de Biologie Fonctionnelle et Adaptative, CNRS UMR 8251, Universite ´Paris Diderot, Sorbonne Paris Cite ´, Paris, 75205, France 9 FiDiPro Program, Research Centre for Integrative Physiology and Pharmacology, University of Turku, Kiinamyllynkatu 10, 20520 Turku, Finland 10 Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.celrep.2018.09.038 SUMMARY Compelling evidence has shown that, besides its putative effect on the regulation of the gonadal axis, estradiol (E2) exerts a dichotomic effect on the hypothalamus to regulate food intake and energy expenditure. The anorectic effect of E2 is mainly mediated by its action on the arcuate nucleus (ARC), whereas its effects on brown adipose tissue (BAT) thermogenesis occur in the ventromedial nucleus (VMH). Here, we demonstrate that central E2 decreases hypothalamic ceramide levels and endoplasmic reticulum (ER) stress. Pharmacological or genetic blockade of ceramide synthesis and amelioration of ER stress selectively occurring in the VMH recapitulate the effect of E2, leading to increased BAT thermogenesis, weight loss, and metabolic improvement. These findings demonstrate that E2 regulation of ceramideinduced hypothalamic lipotoxicity and ER stress is an important determinant of energy balance, suggesting that dysregulation of this mechanism may underlie some changes in energy homeostasis seen in females. INTRODUCTION One of the most interesting and least understood aspects of energy balance modulation is gender dimorphism. Although some common mechanisms control energy balance in both males and females, elevated levels of ovarian steroids deeply affect metabolic networks in females (Mauvais-Jarvis et al., 2013, 2017; Palmer and Clegg, 2015; Mauvais-Jarvis, 2015; Morselli et al., 2016; Lo ´pez and Tena-Sempere, 2017). For example, diminished levels of estradiol (E2) after ovarian insufficiency of any etiology, including physiological (menopause) or surgical (ovariectomy [OVX]), are associated with hyperphagia, reduced energy expenditure, and weight gain (Mauvais-Jarvis et al., 2013, 2017; Palmer and Clegg, 2015; Mauvais-Jarvis, 2015; Morselli et al., 2016; Lo ´pez and Tena-Sempere, 2017). E2 replacement in these conditions precludes or reverts OVX-induced obesity by reducing energy intake and elevating energy expenditure; in keeping with this, E2 replacement also prevents metabolic complications, such as glucose intolerance and diabetes. (Mauvais- Jarvis et al., 2013, 2017; Palmer and Clegg, 2015; Mauvais-Jar- vis, 2015; Morselli et al., 2016; Lo ´pez and Tena-Sempere, 2017). Although these beneficial effects of E2 were assumed to be mostly exerted at the peripheral level, recently gleaned data have shown that, to a large extent, they are exerted at the hypothalamus. E2 directly acts on the CNS to modulate energy balance (Lo ´- pez and Tena-Sempere, 2015, 2017). Estrogen receptors (ERs) are widely expressed throughout the brain, particularly in hypothalamic sites, such as the arcuate (ARC), ventromedial (VMH), and paraventricular (PVH) nuclei, as well as the preoptic (POA) and lateral (LHA) hypothalamic areas (Simerly et al., 1990; Simonian and Herbison, 1997; Voisin et al., 1997; Osterlund et al., 1998; Merchenthaler et al., 2004), all having critical roles in the regulation of energy metabolism (Schneeberger et al., 2014; Scott et al., 2014; Magnan et al., 2015; Lo ´pez et al., 2016; Cui et al., 2017). Several lines of evidence have shown that E2 exerts a nucleus-specific action in the hypothalamus to modulate energy homeostasis. Thus, while most of the effects of E2 on feeding occur in proopiomelanocortin (POMC) neurons in the ARC (Xu et al., 2011), E2 within the VMH modulates brown adipose tissue (BAT) thermogenesis (Musatov et al., 2007; Xu et al., 2011; Martı ´nez de Morentin et al., 2014, 2015). However, Cell Reports 25, 413–423, October 9, 2018 ª2018 The Author(s). 413 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). the exact molecular mechanism by which E2 exerts its actions in this nucleus remains unknown. Current evidence has pointed to hypothalamic endoplasmic reticulum (ER) stress as a central pathophysiological mechanism leading to insulin and leptin resistance and subsequently to obesity (Zhang et al., 2008; Martı ´nez de Morentin and Lo ´pez, 2010; Ozcan et al., 2009; Schneeberger et al., 2013). It also has been shown that central ceramide-induced lipotoxicity affects energy balance (Ramı ´rez et al., 2013; Picard et al., 2013; Turpin et al., 2014; Contreras et al., 2014; Magnan et al., 2015). In the hypothalamus, elevated ceramide concentration elicits ER stress, leading to weight gain, insulin resistance, hepatic steatosis, decreased sympathetic tone, and BAT thermogenesis (Contreras et al., 2014, 2017). Notably, the central action of ceramides can be reversed by decreasing hypothalamic ER stress, resulting in increased BAT thermogenesis and browning of white adipose tissue (WAT), which ultimately ameliorates obesity (Contreras et al., 2014, 2017). It also has been reported that peripheral E2 treatment reduces serum ceramide concentration in OVX rats (Vinayavekhin et al., 2016). However, despite this evidence, it is currently unknown (1) whether ovarian estrogens modulate hypothalamic ceramide-induced lipotoxicity, and, more importantly, (2) whether impairment of this mechanism may be the cause of obesity in conditions of estrogen deficiency. Therefore, in this study we aim to determine the importance of hypothalamic ceramides and ER stress on the central effects of E2 on energy homeostasis. RESULTS Central E2 Inhibits Hypothalamic Ceramide-Induced Lipotoxicity and ER Stress OVX rats gained significantly more weight 15 days after the procedure (sham: 12.7 ±2.54 g; OVX: 48.63 ±2.38 g; p < 0.001) and developed a marked hyperphagia (sham: 16.1 ±0.42 g; OVX: 18.25 ±0.40 g after 15 days of OVX; p < 0.001). OVX rats showed the expected decrease in uterus weight and an increase in serum luteinizing hormone (LH) (data not shown), confirming the efficiency of the OVX procedure (Martı ´nez de Morentin et al., 2014, 2015). As expected, central (intracerebroventricular, ICV) E2 treatment elicited a marked decrease in body weight and feeding (Figures 1A and 1B), as well as a reduction in adiposity and lean mass (Figure 1C). As reported before (Cavalcanti-de-Albuquer- que et al., 2014), central E2 administration did not affect oxygen consumption (VO 2 72-hr sham vehicle: 121.16 ±2.76 mL/g lean mass; VO 2 72-hr OVX vehicle: 114.66 ±3.13 mL/g lean mass; VO 2 72-hr OVX E2: 118.17 ±3.1 mL/g lean mass). However, central E2 significantly augmented the energy expenditure (EE) of OVX rats (Figure 1D). In keeping with this evidence, OVX rats receiving ICV E2 showed a reduced respiratory quotient (RQ; Figure 1E), which is indicative of higher lipid oxidation, as well as an increase in body temperature and in BAT temperature (Figures S1A and S1B), uncoupling protein 1 (UCP1) BAT expression (Figure S1C) and a reduction in the hepatic lipid content (Figure S1D). The selected dose of E2 was formerly demonstrated to be in the physiological range and not to leak to periphery from the cerebrospinal fluid (Martı ´nez de Morentin et al., 2014, 2015). Next, we investigated the effect of OVX and central E2 replacement on ceramide content in the mediobasal hypothalamus (MBH). We found that OVX induced a marked elevation of hypothalamic ceramide levels and that E2 ICV restored them to the levels observed in sham-operated rats (Figure 1F). Of note, in association to the rise in ceramide concentration, OVX rats showed increased hypothalamic ER stress in the MBH, as demonstrated by the augmented levels of unfolded protein response (UPR) markers, such as phosphorylated inositol-requiring enzyme (pIRE), phosphorylated PKR-like ER kinase (pPERK), phosphorylated eukaryotic initiation factor 2 alpha (peIF2a), and C/EBP homologous protein (CHOP) (Figure 1G). Again, central E2 administration reversed the increased ER stress by blunting the elevated levels of the UPR markers to the levels of those detected in the sham controls. Particularly, the hypothalamic expression of the chaperone glucose regulated protein 78 kDa (GRP78; also called binding immunoglobulin protein [BiP]; see below) was increased after central treatment with E2, as a cellular mechanism to reduce ER stress (Figure 1G). Pharmacological Inhibition of Hypothalamic Ceramide Synthesis Recapitulates the Effect of Central E2 on Energy Balance through the SNS Ceramide-induced lipotoxicity has been reported to promote positive energy balance due to reduced thermogenic capacity (Contreras et al., 2014, 2017). Therefore, we aimed to investigate whether inhibition of ceramide synthesis in the hypothalamus of OVX rats might improve their metabolic phenotype. Chronic ICV administration of the ceramide inhibitor myriocin (Ramı ´rez et al., 2013) to OVX rats decreased body weight and adiposity independent of feeding (Figures 1H–1J). This effect was associated with normalization of ceramide and ER stress markers levels in the MBH (Figures 1K and 1L), increased body temperature (Figure 1M), elevated BAT temperature (Figure 1N) and raised UCP1 protein levels in the BAT (Figure 1O), as well as reduced hepatic lipid levels (Figure S1E) of OVX rats. Given that central ceramides have been reported to induce a marked decrease in the activity of the sympathetic nervous system (SNS) innervating BAT (Contreras et al., 2014), we aimed to investigate whether adrenergic receptor blockade affected the central effect of ICV myriocin on BAT thermogenesis of OVX rats. The effect of central administration of myriocin was reversed (in a feeding-independent manner) by pharmacological blockade of beta 3 adrenergic receptor (b3-AR) with the specific antagonist, SR59230A (Lo ´pez et al., 2010; Martı ´nez de Morentin et al., 2014; Contreras et al., 2014, 2017; Martı ´nez-Sa ´nchez et al., 2017a; Seoane-Collazo et al., 2018)(Figures 2A and 2B). The increase in body weight induced by SR59230A was associated with equivalent reversal of ICV myriocin-induced activation of body temperature (Figure 2C), BAT temperature (Figure 2D), and BAT UCP1 expression (Figure 2E). Silencing of SPTLC Specifically in the VMH Ameliorates ER Stress and the Metabolic Phenotype of OVX Rats To further investigate the role of de novo ceramide synthesis pathway and to dissect in which hypothalamic nucleus those actions take place, we targeted serine palmitoyltransferase long chain base subunit 1 (SPTLC1). This enzyme catalyzes the 414 Cell Reports 25, 413–423, October 9, 2018 limiting first step in the de novo synthesis of ceramides: the condensation of palmitoyl-CoA and serine, producing 3-ketos- phinganine. The SPTLC is composed of two subunits (SPTLC1 and 2); both are essential for enzyme function because they constitute its catalytic core (Hanada, 2003; Yard et al., 2007; Watson et al., 2009). Therefore, we silenced SPTLC1 expression by using adenoviruses harboring a small hairpin RNA (shRNA) or control adenoviruses expressing green fluorescence protein (GFP) alone (Watson et al., 2009), specifically in the VMH, a key hypothalamic nucleus modulating thermogenesis (Morrison et al., 2014; Contreras et al., 2015). Infection in the VMH was assessed by visualization of GFP expression (Figure 3A) and by decreased protein levels of SPTLC1 (Figure 3B). Given that the stability of the SPTLC2 subunit is inherently dependent on the expression of SPTLC1 and both subunits associate with a 1:1 molar stoichiometry (Hanada, 2003; Watson et al., 2009), silencing SPTLC1 also led to an attendant reduction in the expression of SPTLC2 (Figure 3B), as expected and as previously shown (Watson et al., 2009). Administration of adenoviruses encoding shSPTLC1 in the VMH induced a feeding-independent weight loss in OVX rats, but not in sham rats (Figures 3C–3F). This effect was associated Figure 1. Effect of Central E2 and Myriocin on Energy Balance in OVX Rats (A–G) Body weight change (A), daily food intake (B), fat mass change (left panel) and lean mass change (right panel) (C), time course energy expenditure (EE) (left panel) and total energy expenditure change (right panel) (D), respiratory quotient (E), ceramide levels in the MBH (F), and representative western blot autoradiographic images (left panel) and MBH protein levels of UPR (right panel) (G) of sham rats or OVX rats ICV treated with vehicle or E2 (n = 7–9 animals per group). (H–O) Body weight change (H), daily food intake (I), fat mass change (left panel) and lean mass change (right panel) (J), ceramide levels in the MBH (K), representative western blot autoradiographic images (left panel) and MBH protein levels of UPR (right panel) (L), body temperature (M), representative infrared thermal images (left panel) and temperature of BAT area (right panel) (N), and representative western blot autoradiographic images (left panel) and protein levels of UCP1 in the BAT (right panel) (O) of sham rats or OVX rats ICV treated with vehicle or myriocin (n = 7–16 animals per group for all the analyses, but the body weight and food intake measurements were n = 28–38). All data are expressed as mean ±SEM. *, **, and ***p < 0.05, 0.01, and 0.001 versus sham vehicle; #, ##, and ### p < 0.05, 0.01, and 0.001 versus OVX vehicle. For the western blot analyses, representative images for all proteins are shown; in the case of the loading controls a representative gel is displayed for clarity, although each band of each protein was always corrected by its own internal control band (b-actin or a-tubulin). The bands for each picture always come from the same gel, although they have been spliced for clarity. See also Figure S1. Cell Reports 25, 413–423, October 9, 2018 415 with a marked relief of ER stress in the VMH of OVX rats, as demonstrated by the reduced protein levels of pIRE, pPERK, peIF2a, ATF6a, and CHOP, which were also observed in sham rats, but to a lesser extent (Figures 3G–3H). In keeping with these data, injection of shSPTLC1 adenoviruses led to increased body temperature (Figures 4A and 4B), BAT temperature (Figures 4C and 4D) and UCP1 protein levels in the BAT (Figures 4E and 4F) of OVX, but not of sham, rats. These effects were associated with an improvement in the metabolic phenotype of OVX rats, as demonstrated by decreased hepatic steatosis (Figure S2A). Overall, these results, alongside the myriocin data, indicate that the hypothalamic and, more precisely, VMH ceramide levels mediated the central actions of E2 on BAT thermogenesis and energy balance. Pharmacological Inhibition of Hypothalamic ER Stress Recapitulates the Effect of Central E2 on Energy Balance Increased hypothalamic ER stress has been associated with the development of insulin and leptin resistance, leading to obesity (Zhang et al., 2008; Ozcan et al., 2009; van Dam et al., 2015; Schneeberger et al., 2013; Contreras et al., 2014, 2017). Bearing in mind that hypothalamic ceramides elicit ER stress (Contreras et al., 2014, 2017), we aimed to investigate whether ceramideinduced ER stress may be a mechanism mediating the central actions of E2 deficiency on BAT thermogenesis and energy homeostasis. Thus, OVX rats were ICV treated with the chemical chaperone tauroursodeoxycholic acid (TUDCA) (Zhang et al., 2008; Ozcan et al., 2009; Schneeberger et al., 2013; Imbernon et al., 2016; Contreras et al., 2017; Porteiro et al., 2017). Central administration of this drug to OVX, but not sham, rats induced feeding-independent weight loss (Figures 5A–5D), decreased hypothalamic ER stress (Figures 5E and 5F), a trend to increase body temperature (Figure 5G), elevated BAT temperature (Figure 5H) and UCP1 protein levels in BAT (Figure 5I). Of note, none of the metabolic changes were found in sham rats (Figures 5G–5I), which is in line with the fact that they did not show elevated ER stress (Figures 1G and 1L). Furthermore, the central injection of TUDCA reduced the hepatic lipid content of OVX rats but had no effects in sham rats (Figure S3A). Figure 2. Effect of Myriocin and b3-AR Antagonism on BAT Thermogenesis in OVX Rats (A–E) Body weight change (A), daily food intake (B), body temperature (C), representative infrared thermal images (left panel) and temperature of BAT area (right panel) (D), and representative western blot autoradiographic images (left panel) and protein levels of UCP1 in the BAT (right panel) (E) of OVX rats ICV treated with myriocin and subcutaneously treated with vehicle or the b3-AR antagonist SR59230A (n = 7–18 animals per group). All data are expressed as mean ±SEM. ** and ***p < 0.01 and 0.001 versus OVX vehicle vehicle; #, ##, and ### p < 0.05, 0.01, and 0.001 versus OVX myriocin vehicle. For the western blot analyses, representative images for all proteins are shown; in the case of the loading controls a representative gel is displayed for clarity, although each band of each protein was always corrected by its own internal control band (a-tubulin). The bands for each picture always come from the same gel, although they have been spliced for clarity. 416 Cell Reports 25, 413–423, October 9, 2018 GRP78 in the VMH Decreases Body Weight and Improves the Metabolic Phenotype of OVX Rats through the SNS Our first set of data showed that central treatment with E2 decreased ER stress and augmented the hypothalamic protein levels of GRP78 (Figure 1G). This was of importance because this protein is a chaperone, located in the ER, that facilitates the protein folding upstream of the UPR and therefore decreases ER stress levels (Gregor and Hotamisligil, 2011; Fu et al., 2012). Thus, a gain of function experiment was undertaken using an adenovirus encoding GRP78 or control adenoviruses expressing GFP alone injected into the VMH of OVX rats and their sham controls. Infection efficiency in the VMH was assessed by expression of GFP (similarly to Figure 3A; data not shown), and by increased concentration of GRP78 in the VMH (Figures 6E and 6F). GRP78 adenoviruses into the VMH induced feeding-inde- pendent weight loss in OVX rats, but not in sham rats (Figures 6A–6D). That effect was linked to decreased ER stress in the VMH of OVX, but not sham, rats (Figures 6E and 6F), increased body temperature (Figure 6G), BAT temperature (Figure 6H), and augmented UCP1 protein levels in BAT (Figure 6I). Notably, central administration of GRP78 in the VMH reduced the hepatic lipid content of OVX rats (Figure S3B). Finally, we investigated whether regulation of BAT following administration of GRP78 adenoviral particles in the VMH of OVX rats was mediated by the SNS. Pharmacological inactivation of b3-AR by SC administration of the specific antagonist, SR59230A (Lo ´pez et al., 2010; Martı ´nez de Morentin et al., 2014; Contreras et al., 2014, 2017; Martı ´nez-Sa ´nchez et al., 2017a; Seoane-Collazo et al., 2018), prevented the effect on body weight associated with central administration of GRP78 viruses (Figure 7A) without affecting feeding (Figure 7B). Consistently, the treatment with SR59230A blunted the GRP78-induced Figure 3. Effect of SPTLC1 Downregulation in the VMH of OVX Rats on Energy Balance (A–H) Direct fluorescence of GFP (A), representative western blot autoradiographic images (left panel) and VMH protein levels of serine palmitoyltransferase, long chain subunit 1-2 (right panel) (B), body weight change (C and E), daily food intake (D and F), and representative western blot autoradiographic images (left panel) and VMH protein levels of UPR (right panel) (G and H) of sham or OVX rats stereotaxically treated with adenoviruses encoding GFP or shSPTLC1 into the VMH (n = 7–20 animals per group). All data are expressed as mean ±SEM. *, ** and ***p < 0.05, 0.01, and 0.001 versus sham GFP or OVX GFP. For the western blot analyses, representative images for all proteins are shown; in the case of the loading controls a representative gel is displayed for clarity, although each band of each protein was always corrected by its own internal control band (b-actin). The bands for each picture always come from the same gel, although they have been spliced for clarity. See also Figure S2. Cell Reports 25, 413–423, October 9, 2018 417 increase in body temperature (Figure 7C), BAT temperature (Figure 7D), and UCP1 protein levels in the BAT (Figure 7E). Overall, this evidence demonstrates that E2-induced decrease of ER stress in the VMH is an important modulator of energy balance by controlling BAT thermogenesis. DISCUSSION This study identifies a link between the effects of E2 on hypothalamic ceramide-induced lipotoxicity and ER stress with BAT thermogenesis. We demonstrate that central E2 replacement restored hypothalamic ceramide levels and ER stress in OVX female rats. Of note, pharmacological or genetic blockade of ceramide synthesis and relief of ER stress recapitulate the effects of E2, namely increased BAT thermogenesis, weight loss, and reduced hepatic lipids. Conditions of estrogen deficiency, such as ovariectomy or menopause, are associated with a positive energy balance, as result of hyperphagia and decreased energy expenditure, leading to enhanced adiposity (Mauvais-Jarvis et al., 2013, 2017; Palmer and Clegg, 2015; Mauvais-Jarvis, 2015; Morselli et al., 2016; Lo ´pez and Tena-Sempere, 2017), hepatic lipid accumulation, and steatosis (Paquette et al., 2007; Vo ¨lzke et al., 2007; Paquette et al., 2008) in animal models and humans. Peripheral and central estrogen replacement therapy reverts this phenotype both in women and in female rodents (Mauvais-Jarvis et al., 2013, 2017; Palmer and Clegg, 2015; Mauvais-Jarvis, 2015; Morselli et al., 2016; Lo ´pez and Tena-Sempere, 2017). Recent evidence indicates that AMP-activated protein kinase (AMPK) in the VMH mediates the central effect of E2 on BAT thermogenesis through the SNS (Martı ´nez de Morentin et al., 2014, 2015). Considering the key role of AMPK on fatty acid metabolism (Kahn et al., 2005; Lage et al., 2008; Lo ´pez et al., 2016), this evidence directly links the central effects of estrogens with hypothalamic lipids. However, despite recent data showing sexual dimorphism in brain fatty acid content (Rodriguez-Navas et al., 2016) and even though peripheral E2 treatment reduces serum ceramide levels in OVX rats (Vinayavekhin et al., 2016), whether complex lipids metabolism is involved in the central actions of E2 remains unclear. This is relevant, because the hormonal regulation of lipid metabolism usually shows tissue-specific effects (Lo ´pez et al., 2010; Martı ´nez-Sa ´nchez et al., 2017a). In this sense, we have recently reported that thyroid hormones (THs) modulate ceramide metabolism in an opposite fashion within different brain areas, such as the hypothalamus and the cerebral cortex (Martı ´nez-Sa ´nchez et al., 2017a). Therefore, the fact that systemic E2 decreased serum ceramide levels (Vinayavekhin et al., 2016) did not imply a similar hypothalamic effect. Compelling evidence indicates that hypothalamic ceramideinduced lipotoxicity is a pathological mechanism leading to obesity by suppressing BAT thermogenesis (Contreras et al., 2014, 2017). Importantly, pharmacological or genetic targeting Figure 4. Effect of SPTLC1 Downregulation in the VMH of OVX Rats on BAT Thermogenesis (A–F) Body temperature (A and B), representative infrared thermal images (left panel) and temperature of BAT area (right panel) (C and D), and representative western blot autoradiographic images (left panel) and protein levels of UCP1 in the BAT (right panel) (E and F) of sham or OVX rats stereotaxically treated in the VMH with adenoviruses encoding GFP or shSPTLC1 (n = 7–20 animals per group). All data are expressed as mean ±SEM. * and ***p < 0.05 and 0.001 versus sham GFP or OVX GFP. For the western blot analyses, representative images for all proteins are shown; in the case of the loading controls a representative gel is displayed for clarity, although each band of each protein was always corrected by its own internal control band (a-tubulin). The bands for each picture always come from the same gel, although they have been spliced for clarity. See also Figure S2. 418 Cell Reports 25, 413–423, October 9, 2018 of this pathway, causing reduced ceramide concentration and/or decreased ER stress ameliorates obesity (Contreras et al., 2014, 2017). However, it remains totally unknown whether fluctuations in ceramide levels and/or ER stress reflect brain changes associated with, for example, neuronal and/or glial metabolic activity (Picard et al., 2013, 2014; Magnan et al., 2015), or whether they might respond to or integrate nutritional and hormonal cues from the periphery (Lo ´pez et al., 2016). Supporting this latter idea, several endocrine signals, such as THs (Lo ´pez et al., 2010; Alvarez-Crespo et al., 2016; Martı ´nez-Sa ´nchez et al., 2017a, b), leptin (Wolfgang et al., 2007; Tanida et al., 2013), bone-morpho- genetic protein 8B (BMP8B) (Whittle et al., 2012; Martins et al., 2016), glucagon-like peptide 1 (GLP-1) (Beiroa et al., 2014), and, importantly, E2 (Martı ´nez de Morentin et al., 2014, 2015), have been shown to regulate hypothalamic fatty acid metabolism to modulate energy balance. Thus, we hypothesized that E2-induced actions on energy balance could be mediated by modulation of hypothalamic ceramides and ER stress. Our data showed that central treatment with E2 induced a negative energy balance in OVX rats, associated with a reduction of hypothalamic ceramide-induced lipotoxicity and ER stress. Of note, central pharmacological inhibition of ceramide synthesis by using the SPTLC inhibitor myriocin suppressed hypothalamic ER stress and fully recapitulated the effects of E2 on energy balance. Next, we aimed to further investigate the specific hypothalamic nuclei where those actions take place. It is known that the VMH plays a major role in the modulation of BAT thermogenesis (Morrison et al., 2014; Figure 5. Effect of Central TUDCA on Energy Balance in OVX Rats (A–I) Body weight change (A and C), daily food intake (B and D), representative western blot autoradiographic images (left panel) and MBH protein levels of UPR (right panel) (E and F), body temperature (G), representative infrared thermal images (left panel) and temperature of BAT area (right panel) (H), and representative western blot autoradiographic images (left panel) and protein levels of UCP1 in the BAT (right panel) (I) of sham or OVX rats ICV treated with vehicle or TUDCA (n = 7–8 animals per group). All data are expressed as mean ±SEM.* and **p < 0.05 and 0.01 versus sham vehicle or OVX vehicle. For the western blot analyses, representative images for all proteins are shown; in the case of the loading controls a representative gel is displayed for clarity, although each band of each protein was always corrected by its own internal control band (b-actin or a-tubulin). The bands for each picture always from come from the same gel, although they have been spliced for clarity. See also Figure S3. Cell Reports 25, 413–423, October 9, 2018 419 Contreras et al., 2015) and that ERais the site of action for E2 in the VMH to drive the modulation of BAT thermogenic function (Musatov et al., 2007; Xu et al., 2011; Martı ´nez de Morentin et al., 2014, 2015). Therefore, following a genetic strategy we silenced SPTLC1 expression in the VMH of OVX obese rats. Specific downregulation of SPTLC exclusively in the VMH ameliorated ER stress and obesity and improved the metabolic health of OVX rats, including a reduction in hepatic lipid content. Overall, these data suggest that central E2 induces a negative energy balance by diminishing ceramide-induced lipotoxicity and ER stress, a hypothesis that was proven by treating OVX rats centrally with the chemical chaperone TUDCA or by genetically overexpressing the chaperone GRP78 within the VMH. Importantly, all the above effects were reversed by pharmacological blockage of b3-AR, therefore suggesting that VMH changes in ceramide signaling and hypothalamic ER stress induced by E2 to increase thermogenesis are conveyed via the sympathetic outflow to BAT. In summary, in this study we demonstrate that E2 promotes an amelioration of ceramide-induced lipotoxicity and ER stress, specifically in the VMH. This action leads to increased BAT thermogenesis through the activation of sympathetic b3-AR signaling, an action that is associated with feeding-inde- pendent weight loss and reduced hepatic steatosis, as well as with a substantial attenuation (or reversal) of some of the metabolic hallmarks of OVX and menopause, namely, body weight gain, and hepatic steatosis. This evidence is of relevance because proper understanding of gender differences in the Figure 6. Effect of GRP78 Overexpression in the VMH of OVX Rats on Energy Balance and BAT Thermogenesis (A–I) Body weight change (A and C), daily food intake (B and D), representative western blot autoradiographic images (left panel) and VMH protein levels of UPR (right panel) (E and F), body temperature (G), representative infrared thermal images (left panel) and temperature of BAT area (right panel) (H), and representative western blot autoradiographic images (left panel) and protein levels of UCP1 in the BAT (right panel) (I) of sham or OVX rats stereotaxically treated in the VMH with adenoviruses encoding GFP or GRP78 (n = 7–18 animals per group). All data are expressed as mean ±SEM. *, **, and ***p < 0.05, 0.01, and 0.001 versus sham GFP or OVX GFP. For the western blot analyses, representative images for all proteins are shown; in the case of the loading controls a representative gel is displayed for clarity, although each band of each protein was always corrected by its own internal control band (b-actin or a-tubulin). The bands for each picture always come from the same gel, although they have been spliced for clarity. See also Figure S3. 420 Cell Reports 25, 413–423, October 9, 2018 every 30 min. Animals were placed for adaptation for 1 week before starting the measurements. For the measurement of body composition, we used nuclear magnetic resonance (NMR) (Whole Body Composition Analyzer; EchoMRI; Houston, TX) (Imbernon et al., 2013; Martı ´nez de Morentin et al., 2012, 2014; Martins et al., 2016; Martı ´nez-Sa ´nchez et al., 2017a; Seoane-Collazo et al., 2018). Animals do not need to be anesthetized neither other special preparation before measurement. They are placed in a holder of custom-defined size during the measurement (measuring time: 0.5-3.2 min). Two measurements were done twice for animal 72 h apart. Sample processing Rats were killed by cervical dislocation and decapitation. From each animal, the MBH, VMH, liver and the interscapular BAT were collected for western blotting and Oil Red O staining and immediately homogenized on ice to preserve phosphorylated protein levels. Those samples and the serum were stored at 80C until further processing. Dissection of the VMH was performed by micro-punch procedure under the microscope, as previously described (Lo ´pez et al., 2010; Whittle et al., 2012; Contreras et al., 2014, 2017; Martı ´nez de Morentin et al., 2014; Martins et al., 2016; Martı ´nez-Sa ´nchez et al., 2017a; Seoane-Collazo et al., 2018). Ceramide quantification Ceramide analysis reported in this manuscript were performed in two different laboratories and according to the next methods. In the first one, ceramides were extracted and analyzed using liquid chromatography-mass spectrometer consisted of a Waters Acquity UPLC System connected to a Waters LCT Premier Orthogonal Accelerated Time of Flight Mass Spectrometer (Waters; Millford, MA, USA), operated in positive electrospray ionization mode. Full scan spectra from 50 to 1500 Da were acquired and individual spectra were summed to produce data points each 0.2 s. Mass accuracy and reproducibility were maintained by using an independent reference spray by the LockSpray interference. The analytical column was a 100 mmL 2.1 mm i.d., 1.7 mm C8 Acquity UPLC BEH (Waters; Millford, MA, USA). The two mobile phases were phase A: methanol; phase B: water, both contained 0.2% formic acid (v/v) and 2 mM ammonium formate. A linear gradient was programmed (0.0 min: 20% B; 3 min: 10% B; 6 min: 10% B; 15min: 1% B; 18 min: 1% B; 20 min: 20% B; 22 min: 20% B). The flow rate was 0.3 mL/min. The column was held at 30C. Quantification was carried out using the extracted ion chromatogram of each compound, using 50 mDa windows. The linear dynamic range was determined by injecting standard mixtures. Positive identification of compounds was based on the accurate mass measurement with an error < 5 ppm and its LC retention time, compared to that of a standard (±2%). Concentrations were measured by multiple reaction monitoring experiments using N-lauroyl-D-erythro-sphingosine (C12-ceramide) and N-Heptadecanoyl-D-erytro-Sphingosine (C17-ceramide) as internal standards and using as patrons N-Palmitoyl-D-erytro-sphingosine (C16-ceramide), N-Stearoyl-D-erytro-sphingosine (C18-ceramide), N-lignoceroyl-D-erythro-sphingosine (C24-ceramide) and N-nervonoyl-D-erythro-sphingosine (C24:1-ceramide) (Avanti Polar Lipids; Alabaster, AL, USA). The ceramide analyses from Figure 1K were performed in the Research Unit on Bioactive Molecules (RUBAM) of the Institut de Quı ´mica Avanc¸ ada de Catalunya (IQAC-CSIC) according to this method. In the second one, total lipid species were extracted according to the method of Folch et al. (1957). Ceramide lipid species were enriched in the lower phase (organic phase) and analyzed using an untargeted lipidomic approach by liquid chromatography coupled with high-resolution mass spectrometry (LC–HRMS) as described by Seyer et al. (2016). The ceramide analyses from Figure 1F were performed in the Unite ´de Biologie Fonctionnelle et Adaptative (Univertsite ´Paris Diderot) according to this method. Histological analyses Hepatic lipid content was analyzed by Oil Red O staining, as previously shown (Seoane-Collazo et al., 2014; Contreras et al., 2014, 2017; Imbernon et al., 2016; Porteiro et al., 2017; Martı ´nez-Sa ´nchez et al., 2017a). Hepatic frozen sections were cut (8 mm) and fixed in 10% buffered formaldehyde. Sections were stained in filtered Oil Red O (Sigma; St. Louis, MO, USA), washed in distilled water, counterstained with Harris hematoxylin (Bio-Optica; Milan, Italy) and washed in distilled water again. Sections were mounted in aqueous mounting medium (Bio-Optica; Milan, Italy). Images were taken with a digital camera Olympus XC50 (Olympus Corporation; Tokyo, Japan) at 20X. Digital images of liver were quantified with ImageJ Software (National Institutes of Health; USA). Direct detection of GFP fluorescence was performed after perfusion of the animals; on 40 mm brain sections. Images were taken with a fluorescence microscope Olympus IX51, as previously shown (Contreras et al., 2017; Seoane-Collazo et al., 2018). Western blotting Protein lysates from hypothalamus (MBH or VMH) and BAT were homogenized in lysis buffer (consisting of a mix of 0.05 M Tris-HCl, 0.01 M EGTA, 0.001 M EDTA, 0.016 M Triton X-100, 0.001 M sodium orthovanadate, 0.05 M sodium fluoride, 0.01 M sodium pyrophosphate and 0.25 M sucrose, made up with distilled water and adjusted to 7.5 pH; all of them from Sigma; St. Louis, MO, USA) and freshly added protease inhibitor cocktail tablets (Roche Diagnostics; Indianapolis, IN, USA). The protein concentration was determined by the Bradford Method (Protein assay dye concentrate, Bio-Rad Laboratories; Hercules, CA, USA), and the total protein content of the tissues was calculated. The protein lysates were subjected to SDS-PAGE, electrotransferred to polyvinylidene difluoride membranes (PVDF; Millipore; Billerica, MA, USA) with a semidry blotter and probed with antibodies against GRP78 (Cell Signaling; Danvers; MA, USA); ATF6a, CHOP, peIF2a(Ser52), pPERK (Thr981) (Santa Cruz; Santa Cruz, CA,USC); pIREa(Ser724), SPTLC1, SPTLC2, UCP1 (Abcam; Cambridge, UK); a-tubulin or b-actin (Sigma; St. Louis, MO, USA) as described (Lo ´pez et al., 2010; Varela et al., 2012; Martı ´nez de Morentin et al., 2014; Contreras et al., 2014; Martı ´nez-Sa ´nchez et al., 2017a; Seoane-Collazo et al., 2018). Cell Reports 25, 413–423.e1–e5, October 9, 2018 e4 Each membrane was then incubated with the corresponding secondary antibody: anti-mouse or anti-rabbit (all of them from DAKO; Glostrup, Denmark). The membranes were exposed to an X-ray film (Fujifilm; Tokyo, Japan) and developed using developer (Developer G150; AGFA HealthCare: Mortsel, Belgium) and Fixator (Manual Fixing G354; AGFA HealthCare: Mortsel, Belgium). Autoradiographic films were scanned and the bands signal was quantified by densitometry using ImageJ-1.33 software (NIH; Bethesda, MD, USA). Values were expressed in relation to b-actin (hypothalamus) or a-tubulin (BAT). Representative images for all proteins are shown; in the case of the loading controls a representative gel is displayed for clarity, although each band of each protein was always corrected by its own internal control band (b-actin or a-tubulin). In all the Figures showing images of gels, all the bands for each picture come always from the same gel, although they may be spliced for clarity. QUANTIFICATION AND STATISTICAL ANALYSIS ImageJ-1.33 software (NIH; Bethesda, MD, USA) was used to process western blot analyses and liver histologic images (Lo ´pez et al., 2010; Varela et al., 2012; Martı ´nez de Morentin et al., 2014; Contreras et al., 2014; Martı ´nez-Sa ´nchez et al., 2017a, b; Seoane-Collazo et al., 2018). FLIR-Tools-Software; FLIR; West Malling, Kent, UK) was used to process the thermal images (Whittle et al., 2012; Martı ´nez de Morentin et al., 2012; Contreras et al., 2014, 2017; Martı ´nez de Morentin et al., 2014, 2015; Martins et al., 2016; Martı ´nez- Sa ´nchez et al., 2017a; Seoane-Collazo et al., 2018). Statistical analysis was conducted using GraphPad Prism 6 Software (GraphPad Software, Inc.; La Jolla, CA, USA). Data are expressed as mean ±SEM. Protein data were expressed in relation (%) to control (Sham or OVX, vehicle or GFP treated) rats. Error bars represent SEM. Statistical significance was determined by Student’s t test (when two groups were compared) or ANOVA (when more than two groups were compared) followed by post hoc Bonferroni test. p < 0.05 was considered significant. The number of animals used in each experimental setting and analysis are specified in each figure legend. e5 Cell Reports 25, 413–423.e1–e5, October 9, 2018