Mitochondrial Dynamics Mediated by Mitofusin 1 Is Required for ...

12 downloads 0 Views 3MB Size Report
Jun 6, 2017 - Loss of MFN1 in POMC neurons alters central glucose sensing d. Mfn1 deletion in POMC neurons causes defective pancreatic insulin release.
Short Article

Mitochondrial Dynamics Mediated by Mitofusin 1 Is Required for POMC Neuron Glucose-Sensing and Insulin Release Control Graphical Abstract

Authors Sara Ramı´rez, Alicia G. Go´mez-Valade´s, Marc Schneeberger, ..., Ramon Gomis, Antonio Zorzano, Marc Claret

Correspondence [email protected] (A.Z.), [email protected] (M.C.)

In Brief Ramı´rez et al. report that mitochondrial fusion protein MFN1 in POMC neurons is necessary for adequate mitochondrial dynamism in response to metabolic challenges, central glucose-sensing and pancreatic insulin release. The latter is mechanisticallyassociated with enhanced hypothalamic ROS production. These alterations confer susceptibility to diabetes in a pathophysiological context.

Highlights d

POMC neuron MFN1-dependent mitochondrial dynamics is required for metabolic shifts

d

Loss of MFN1 in POMC neurons alters central glucose sensing

d

d

Mfn1 deletion in POMC neurons causes defective pancreatic insulin release Impaired insulin secretion is caused by enhanced central ROS production

Ramı´rez et al., 2017, Cell Metabolism 25, 1390–1399 June 6, 2017 ª 2017 Elsevier Inc. http://dx.doi.org/10.1016/j.cmet.2017.05.010

Cell Metabolism

Short Article Mitochondrial Dynamics Mediated by Mitofusin 1 Is Required for POMC Neuron Glucose-Sensing and Insulin Release Control Sara Ramı´rez,1,16 Alicia G. Go´mez-Valade´s,1,16 Marc Schneeberger,1 Luis Varela,2 Roberta Haddad-To´volli,1 Jordi Altirriba,3 Eduard Noguera,4,5,6 Anne Drougard,7 A´lvaro Flores-Martı´nez,8 Mo´nica Imberno´n,9,10 In˜igo Chivite,1 Macarena Pozo,1 Andre´s Vidal-Itriago,1 Ainhoa Garcia,6,11 Sara Cervantes,11 Rosa Gasa,6,11 Ruben Nogueiras,9,10 Pau Gama-Pe´rez,12 Pablo M. Garcia-Roves,12,13 David A. Cano,8 Claude Knauf,7 Joan-Marc Servitja,6,11 Tamas L. Horvath,2,14 Ramon Gomis,6,11,15 Antonio Zorzano,4,5,6,17,* and Marc Claret1,6,17,18,* 1Neuronal

Control of Metabolism (NeuCoMe) Laboratory, Institut d’Investigacions Biome`diques August Pi i Sunyer (IDIBAPS), 08036 Barcelona, Spain 2Program in Integrative Cell Signaling and Neurobiology of Metabolism, Department of Comparative Medicine, Yale University School of Medicine, New Haven, CT 06520, USA 3Laboratory of Metabolism, Department of Internal Medicine Specialties, Faculty of Medicine, University of Geneva, 1211 Geneva, Switzerland 4Institute for Research in Biomedicine (IRB Barcelona), 08028 Barcelona, Spain 5Departament de Bioquı´mica i Biologia Molecular, Facultat de Biologia, Universitat de Barcelona, 08028 Barcelona, Spain 6CIBER de Diabetes y Enfermedades Metabo ´ licas Asociadas (CIBERDEM), 08036 Barcelona, Spain 7Institute of Research in Digestive Health (IRSD) – INSERM U1220, European Associated Laboratory ‘‘NeuroMicrobiota’’, University Paul Sabatier, 31024 Toulouse, France 8Unidad de Gestio ´ n Clı´nica de Endocrinologı´a y Nutricio´n, Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocı´o/Consejo Superior de Investigaciones Cientı´ficas/Universidad de Sevilla, 41013 Sevilla, Spain 9Instituto de Investigaciones Sanitarias (IDIS), CIMUS, University of Santiago de Compostela, Santiago de Compostela 15782, Spain 10CIBER Fisiopatologı´a de la Obesidad y Nutricio ´ n (CIBERobn), 15706 Santiago de Compostela, Spain 11Diabetes and Obesity Research Laboratory, Institut d’Investigacions Biome ` diques August Pi i Sunyer (IDIBAPS), 08036 Barcelona, Spain 12Department of Physiological Sciences, University of Barcelona, 08907 Barcelona, Spain 13Institut d’Investigacio ´ Biome`dica de Bellvitge (IDIBELL), L’Hospitalet de Llobregat, 08908 Barcelona, Spain 14Department of Anatomy and Hystology, University of Veterinary Medicine, Budapest 1078, Hungary 15Department of Endocrinology and Nutrition, Hospital Clı´nic. School of Medicine, University of Barcelona, 08036 Barcelona, Spain 16These authors contributed equally 17Senior author 18Lead Contact *Correspondence: [email protected] (A.Z.), [email protected] (M.C.) http://dx.doi.org/10.1016/j.cmet.2017.05.010

SUMMARY

Proopiomelanocortin (POMC) neurons are critical sensors of nutrient availability implicated in energy balance and glucose metabolism control. However, the precise mechanisms underlying nutrient sensing in POMC neurons remain incompletely understood. We show that mitochondrial dynamics mediated by Mitofusin 1 (MFN1) in POMC neurons couple nutrient sensing with systemic glucose metabolism. Mice lacking MFN1 in POMC neurons exhibited defective mitochondrial architecture remodeling and attenuated hypothalamic gene expression programs during the fast-to-fed transition. This loss of mitochondrial flexibility in POMC neurons bidirectionally altered glucose sensing, causing abnormal glucose homeostasis due to defective insulin secretion by pancreatic b cells. Fed mice lacking MFN1 in POMC neurons displayed enhanced hypothalamic mitochondrial oxygen flux and reactive oxygen species generation. Central delivery of antioxidants was able to

normalize the phenotype. Collectively, our data posit MFN1-mediated mitochondrial dynamics in POMC neurons as an intrinsic nutrient-sensing mechanism and unveil an unrecognized link between this subset of neurons and insulin release.

INTRODUCTION Proopiomelanocortin (POMC) neurons in the arcuate nucleus (ARC) of the hypothalamus are critical regulators of food intake, energy expenditure, and glucose metabolism (Schneeberger et al., 2014). POMC neurons sense signals informing about the energy status of the organism. However, the molecular underpinnings underlying nutrient sensing in POMC neurons remain incompletely understood. A major organelle implicated in cellular nutrient and energy management is the mitochondria, which bioenergetically adjusts to different metabolic situations (Gao et al., 2014). This flexibility entails changes in mitochondrial respiratory status, causing energy and redox alterations, thereby integrating sensing and signaling processes. Mitochondria-derived signals are critical

1390 Cell Metabolism 25, 1390–1399, June 6, 2017 ª 2017 Elsevier Inc.

for POMC neuron function. For example, reactive oxygen species (ROS) activate POMC neurons, promoting satiety and energy expenditure (Diano et al., 2011). Furthermore, variations in AMP:ATP or NAD:NADH in POMC neurons regulate metabolism through classical energy sensors such as KATP channels (Ibrahim et al., 2003; Parton et al., 2007), AMPK (Claret et al., 2007), mTOR (Mori et al., 2009; Smith et al., 2015), or SIRT1 (Ramadori et al., 2010). Therefore, mitochondria can be considered energysensing organelles as they are primary sources of metabolites communicating cellular energy status. A remarkable feature of mitochondria is their ability to modify their morphology through fusion and fission events. Mitofusin (MFN) proteins are key regulators of mitochondrial fusion (Schrepfer and Scorrano, 2016). MFN1 mediates this process more efficiently than its homologous MFN2 (Ishihara et al., 2004), which exhibits additional functions (de Brito and Scorrano, 2008; Mun˜oz et al., 2013). Evidence suggests that mitochondrial dynamics is a mechanism connecting nutrient availability and demand with bioenergetic adaptations (Gao et al., 2014; Liesa and Shirihai, 2013; Schrepfer and Scorrano, 2016). Thus, we reasoned that such a highly flexible process could represent a core mechanism to sense nutrient fluctuations and to metabolically adjust to energy demands. Here, we show that loss of MFN1 in POMC neurons hinders adequate adaptations to food availability, prevents fed-associated mitochondrial fusion, and affects glucose sensing. Our data also unveil MFN1-mediated mitochondrial dynamics in POMC neurons as a regulatory mechanism of systemic glucose homeostasis via modulation of insulin release. RESULTS Mitochondrial Dynamics in POMC Neurons Is Required for Glucose-Sensing and Metabolic Adaptations to the Fast-to-Fed Transition To examine the role of MFN1 in POMC neurons, we generated conditional knockout mice (hereafter POMCMfn1KO) as previously described (Schneeberger et al., 2013). Tissue-specific recombination of floxed allele was confirmed by PCR (Figure S1A). No perturbations in POMC neuron architecture, area, or number were observed (Figures S1B–S1D). Corticosterone values were equivalent after antecedent hypoglycemia (Figure S1E). Electron microscopy did not reveal mitochondrial ultrastructural alterations (Figure 1A) or degeneration (Figure S1F) in POMC neurons from mutant mice. Although mitochondrial density was unchanged (Figure 1B), mitochondrial cytosol coverage (Figure 1C) and area (Figure 1D) were reduced. To assess whether mitochondrial dynamics in POMC neurons was involved in sensing metabolic changes, we measured mitochondrial aspect ratio (AR; a measure of elongation) in control and POMCMfn1KO mice under fed and fasting conditions. Positive energy balance was associated with increased AR in POMC neurons from control animals, but not from mutant mice (Figure 1E). To ascertain the consequences of impaired mitochondrial dynamics on energy status monitoring, we analyzed the global transcriptomic response in the hypothalamus from control and mutant mice under both fed and fasting conditions. We observed a marked attenuation in the differential expression of genes

during the transition from fast to fed in POMCMfn1KO mice. In control hypothalamus, 200 genes were negatively regulated, while 98 were increased. However, only 40 and 9 of those transcripts were significantly changed in mutant mice (Figure 1F). Representative genes were validated (Figure S1G). The transcriptomic signatures and genes dependent on MFN1 were identified by examining enriched functional categories. The pleiotropic effect of Mfn1 deletion on gene expression was illustrated by the fact that only one out of 19 significantly enriched categories was independent of MFN1 (Figure 1G; Tables S1 and S2). Notably, upregulated pathways were most affected by MFN1 loss. These results indicate that POMCMfn1KO mice are unable to adequately respond to the shift to fed state. The transition from fasted to post-prandrial state causes remarkable changes in circulating glucose levels. As POMC neurons are key glucose sensors (Claret et al., 2007; Ibrahim et al., 2003; Parton et al., 2007), we assessed the effects of intracerebroventricular (i.c.v.) glucose on appetite. POMCMfn1KO mice showed enhanced satiety in response to a low glucose concentration that was ineffective in control counterparts (Figure 1H). Next, we assessed the effect of glucopenia by i.c.v. delivery of 2-deoxyglucose (2-DG). While control mice significantly increased food intake, this effect was blunted in POMCMfn1KO mice (Figure 1I). To specifically evaluate glucose sensing in POMC neurons, we measured C-FOS induction. Glucose increased C-FOS localization in POMC neurons from control mice (Figure 1J). Co-expression in POMCMfn1KO mice was basally upregulated and no further increase was observed upon stimulation (Figure 1J). These results demonstrate that appropriate responsiveness of POMC neurons to bidirectional variations in glucose concentration and subsequent functional and behavioral responses rely on MFN1-mediated mitochondrial dynamics. Altered Glucose Metabolism in POMCMfn1KO Mice POMCMfn1KO mice exhibited normal energy homeostasis (Schneeberger et al., 2013). However, these mice showed glucose intolerance (Figure 2A) but normal insulin sensitivity (Figure 2B). Loss of the homologous MFN2 in POMC neurons enhanced hepatic glucose production (Schneeberger et al., 2015). Nevertheless, liver glucose output and expression of gluconeogenic genes were normal in POMCMfn1KO mice (Figures S2A and S2B). Circulating and hepatic triglycerides (Figures S2C and S2D), very low-density lipoprotein secretion (Figure S2E), and expression of lipid metabolism genes (Figure S2F) were unchanged in POMCMfn1KO mice. These results indicate that MFN1 in POMC neurons is required for systemic glucose homeostasis by liver-independent mechanisms. Lack of MFN1 in POMC Neurons Reveals Abnormal Pancreatic Insulin Release Glucose intolerance in the face of normal insulin sensitivity in POMCMfn1KO mice was suggestive of defective insulin release. Indeed, glucose-stimulated insulin secretion (GSIS) was impaired (Figure 2C), albeit mutant mice responded normally to L-arginine (Figure 2D), indicating secretagogue specificity and functionality of the secretory machinery. POMCMfn1KO mice exhibited equivalent pancreas weight (Figure S3A), b cell mass (Figure 2E), islet density (Figure S3B), and size distribution (Figure 2F). Cell Metabolism 25, 1390–1399, June 6, 2017 1391

Figure 1. Intact Mitochondrial Dynamics in POMC Neurons Is Required for Glucose Sensing and Fast to Fed Transition (A–D) POMC neuron electron microscopy images (A) (bar, 500 nm), mitochondrial density (B), coverage (C) (n = 29–31 neurons/5 mice/genotype), and area (D) (>1,335 mitochondria/5 mice/genotype were analyzed). (E) POMC neuron mitochondrial aspect ratio (AR) under fed and fasting conditions (n = 60–107 neurons/3–4 mice/genotype/condition). Representative skeletonized images of POMC neuron mitochondrial network are shown. (F) Volcano plot of hypothalamic gene expression in control and POMCMfn1KO mice. Dashed lines represent the threshold for fold-change (±1.4) and false discovery rate (FDR) (%0.05). Differentially expressed genes during the fast to fed transition are depicted in red. The number of down and upregulated genes are stated in the upper left and right side, respectively. Unchanged genes are represented in gray. n = 3–4 mice/genotype/condition. (G) Top enriched functional categories and number of genes differentially expressed. (H and I) Food intake after i.c.v. vehicle (V) (H and I), glucose (G) (H), or 2-DG (I) administration (n = 5–7/genotype/treatment). (J) Quantification of C-FOS localization in POMC neurons after vehicle (V) or glucose (G) (n = 2–3/genotype). All studies have been conducted in control and POMCMfn1KO mice between 10 and 14 weeks of age. Data are expressed as mean ± SEM. *p < 0.05; ***p < 0.001. See also Figure S1 and Tables S1 and S2.

1392 Cell Metabolism 25, 1390–1399, June 6, 2017

(legend on next page)

Cell Metabolism 25, 1390–1399, June 6, 2017 1393

Pancreatic islets from POMCMfn1KO mice displayed normal insulin release (Figure 2G). We next investigated glucagon contribution to POMCMfn1KO mice phenotype. Basal and insulin-induced glucagon levels were unaltered in mutant mice (Figure 2H). Likewise, POMCMfn1KO mice responded normally to a glucagon challenge (Figure 2I), and a cell mass was unchanged (Figure 2E). These results are supportive of acute neuronal regulation of GSIS irrespective of glucagon action. Next, we examined the impact of MFN1 loss in POMC neurons upon glucose metabolism in a pathophysiological setting. Fasting glycemia and HOMA-IR values were identical in control and POMCMfn1KO mice fed with standard chow (Figure 2J). However, high-fat diet feeding exacerbated both parameters in mutant mice (Figure 2K) without affecting body weight (Figure S3C). Enhanced Sympathetic Function Blunts Insulin Release in POMCMfn1KO Mice A key modulator of pancreatic islet function is the autonomic nervous system (Thorens, 2014). We therefore measured endocrine parasympathetic and sympathetic innervation using vesicular acetylcholine transporter (VAChT) and tyrosine hydroxylase (TH) as markers, respectively. Nevertheless, no changes were observed in nerve fiber density in islets from mutant mice (Figure 2L). POMC neurons modulate the sympathetic outflow to different tissues (Chhabra et al., 2016; do Carmo et al., 2013; Ramadori et al., 2010). Enhanced sympathetic activity inhibits insulin secretion through activation of a2-adrenergic receptors in b cells (Thorens, 2014). Thus, we investigated the involvement of the sympathetic nervous system (SNS) by assessing the effect of the a2-adrenergic antagonist Idazoxan. This compound was able to rescue defective GSIS in POMCMfn1KO mice (Figure 2M). Consistently, epinephrine content was specifically increased in pancreas (Figures S3D– S3H). In contrast, the cholinergic agonist Carbachol did not improve GSIS (Figure 2N), further suggesting that excessive sympathetic stimulation caused reduced insulin release. No recombination and equivalent Mfn1 transcript levels were observed in pancreatic islets from mutant mice (Figures S3I and S3J). These results rule out unspecific deletion of Mfn1

in b cells as the cause of the POMCMfn1KO mice pancreatic phenotype. Excessive ROS Production Underlies Defective GSIS in POMCMfn1KO Mice To establish the primary defect driving abnormal GSIS, we conducted high-resolution respirometry in ARC-enriched microdissections (Schneeberger et al., 2013). The respiratory capacity through mitochondrial complex I and II was enhanced in POMCMfn1KO mice under fed, but not fasting, conditions (Figure 3A). Consistently, both fluorometric and amperometric analysis showed increased ROS production from POMCMfn1KO hypothalamic explants in response to high glucose (Figures 3B and 3C). We were unable to detect hypothalamic protein damage (Figure 3D), suggesting low-grade and transient ROS production. Increased ROS enhances POMC neuron activity (Diano et al., 2011), which may influence the synthesis of POMC processing products a-melanocyte-stimulating hormone (a-MSH) and b-endorphin. Nevertheless, hypothalamic content of these neuropeptides (Figure 3E) and a-MSH fiber immunoreactivity in the paraventricular nucleus (PVN) were unaltered (Figure 3F). To test whether defective GSIS in POMCMfn1KO mice resulted from excessive ROS production in the hypothalamus, we administered i.c.v. two structurally different ROS scavengers. Remarkably, acute central honokiol (Diano et al., 2011; Schneeberger et al., 2013) and N-acetylcysteine (NaC) administration normalized GSIS in POMCMfn1KO mice (Figures 3G and 3H). Hypothalamic markers of endoplasmic reticulum (ER) stress were unchanged (Figure 3I), and central treatment with the chemical chaperone Tauroursodeoxycholic acid (TUDCA) did not rescue GSIS in mutant mice (Figure 3J). These results demonstrate that transient excessive ROS production in the hypothalamus, and not enhanced ER stress, is the underlying cause of impaired GSIS in POMCMfn1KO mice. Deletion of Opa1 in POMC Neurons Does Not Recapitulate POMCMfn1KO Phenotype Next, we generated mice lacking optic atrophy 1 (OPA1), another key mitochondrial fusion protein (Schrepfer and Scorrano, 2016), in POMC neurons (POMCOpa1KO). Mice with Opa1 exon 5 (ENSMUSE00001245216) flanked by loxP

Figure 2. Defective Glucose Metabolism in POMCMfn1KO Mice Is Caused by Abnormal Insulin Release (A) Glucose tolerance test (n = 6–11/genotype). (B) Insulin sensitivity test (n = 7/genotype). (C) GSIS (n = 6/genotype). (D) Arginine-stimulated insulin release (n = 5/genotype). (E) Representative immunofluorescence images displaying glucagon (red) and insulin (green) staining in pancreatic sections. a and b cell mass are shown (n = 3/ genotype). (F) b cell size distribution frequency (n = 3/genotype). (G) Ex vivo islet GSIS (n = 5/genotype). The average of two independent experiments is shown. (H) Basal and insulin-induced plasma glucagon levels (n = 7–8/genotype). (I) Glucagon challenge test (n = 4–5/genotype). (J and K) Fasting glycemia and HOMA-IR values in mice fed with standard (J) or high-fat diet (K) (n = 5–7/genotype/diet). (L) Pancreatic islet sympathetic (TH) and parasympathetic (VAChT) nerve density (n = 4–5/genotype). (M) GSIS after vehicle (V) or Idazoxan (IDA) administration (n = 6/genotype/treatment). (N) GSIS after vehicle (V) or Carbachol (CAR) administration (n = 6/genotype/treatment). All studies have been conducted in control and POMCMfn1KO mice between 10 and 14 weeks of age or otherwise stated. Data are expressed as mean ± SEM. *p < 0.05; **p < 0.01; ****p < 0.0001. NS: non-significant. See also Figures S2 and S3.

1394 Cell Metabolism 25, 1390–1399, June 6, 2017

(legend on next page)

Cell Metabolism 25, 1390–1399, June 6, 2017 1395

Figure 4. Deletion of Opa1 in POMC Neurons Does Not Recapitulate POMCMfn1KO Phenotype (A) Body weight between 4 and 6 weeks of age (n = 13–20/genotype/age). (B) Aspect ratio of POMC neuron mitochondria (n = 67–108 neurons/3–4 mice/genotype). Representative skeletonized images of POMC neuron mitochondrial network are shown. (C and D) Food intake after vehicle (V), glucose (G), or 2-DG administration (n = 8–13/genotype/treatment). (E) Quantification of C-FOS localization in POMC neurons after vehicle (V) or glucose (G) (n = 4/genotype). (F) GSIS (n = 6–10/genotype). (G) Fluorometric ROS production in hypothalamic explants (n = 7–10/genotype/condition). All studies have been conducted in control and POMCOpa1KO mice between 4 and 6 weeks of age. Data are expressed as mean ± SEM. NS: non-significant. *p < 0.05; **p < 0.01; ***p < 0.001. See also Figure S4.

sites were crossed with POMC-cre mice (Figures S4A–S4C). Tissue-specific recombination of floxed allele was confirmed (Figure S4D), and POMC neuroanatomy was normal (Figures S4E–S4G). POMCOpa1KO mice exhibited obesity from 7 weeks of age (data not shown). To avoid obesity as confounder, we used mice aged 4–6 weeks for the studies herein included (Fig-

ure 4A). Despite reduced mitochondrial fusion (Figure 4B), mutant mice exhibited unaffected glucose sensing, neuronal activation, GSIS, and hypothalamic ROS production (Figures 4C–4G). Thus, the main phenotypic features of POMCMfn1KO mice are not caused by altered mitochondrial fusion itself but to MFN1-specific effect.

Figure 3. Defective GSIS in POMCMfn1KO Mice Is Caused by Enhanced Hypothalamic ROS Production (A) High-resolution mitochondrial respirometry in ARC-enriched samples from fed (n = 9/genotype) or fasted (n = 4–5/genotype) mice. (B) Fluorometric ROS production in hypothalamic explants (n = 7–8/genotype/condition). (C) Amperometric ROS production in hypothalamic explants (n = 4/genotype/condition). (D) Hypothalamic carbonyl protein content (n = 6–8/genotype). (E) Neuropeptide hypothalamic content (n = 5–6/genotype). (F) a-MSH immunoreactivity in the PVN and integrated density (n = 2–3/genotype). (G) Effects of acute i.c.v. honokiol on GSIS (n = 5–7/genotype/treatment). V, vehicle. (H) Effects of acute i.c.v. N-acetylcisteine (NaC) on GSIS (n = 5–7/genotype/treatment). V, vehicle. (I) Hypothalamic expression of ER stress markers measured by qPCR (n = 6–7/genotype). (J) Effects of acute i.c.v. TUDCA on GSIS (n = 5–6/genotype/treatment). V, vehicle. All studies have been conducted in control and POMCMfn1KO mice between 10 and 16 weeks of age. Data are expressed as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

1396 Cell Metabolism 25, 1390–1399, June 6, 2017

Assessment of Heterogeneity of Fusogenic Proteins in POMC Neurons The striking phenotypic divergences among POMCMfn1KO, POMCOpa1KO, and POMCMfn2KO mice (Schneeberger et al., 2013, 2015) could be due to the heterogeneous expression of fusogenic proteins in POMC neuron subsets. To address this, we used publically available single-cell RNA sequencing data for POMC neurons (Lam et al., 2017). Analysis of 163 individual POMC neurons showed that Mfn1, Mfn2, and Opa1 were expressed in 98% of POMC neurons, albeit at variable levels (Figure S4H). No correlation was observed among these genes or Pomc expression (Figure S4I). Immunofluorescence studies confirmed that MFN1 and MFN2 colocalized in all POMC neurons examined (Figure S4J); hence, the divergent phenotypes are not due to POMC neuron heterogeneity. DISCUSSION Our studies provide evidence that (1) the inability to properly maintain POMC neuron mitochondrial dynamism in response to metabolic challenges alters hypothalamic gene expression programs and glucose sensing; (2) loss of MFN1 in POMC neurons increases mitochondrial respiration, ROS production, and neuronal activity; (3) defective MFN1-mediated mitochondrial fusion in POMC neurons impairs glucose metabolism and unmasks a link between these neurons and insulin release via the SNS; and (4) elevated ROS plays a causal role. Mitochondrial dynamics is responsive to metabolic challenges and constitutes a central mechanism for bioenergetic adaptations to cell requirements (Gao et al., 2014; Liesa and Shirihai, 2013; Schrepfer and Scorrano, 2016). However, the potential nutrient-sensing functions of mitochondrial architecture remodeling upon whole-body physiology remain largely incomplete. Our data show that fed state is associated with mitochondrial elongation in ARC POMC neurons and that MFN1 is required for appropriate mitochondrial shape adjustments and global hypothalamic transcriptional response during metabolic challenges. We also demonstrate that loss of MFN1 in POMC neurons impairs central glucose sensing and insulin release, conferring higher susceptibility to develop diabetes in an obesogenic environment. In contrast, perturbed mitochondrial fission through deletion of dynamin-related peptide 1 (DRP1) in POMC neurons improves glucose metabolism via counterregulatory mechanisms and hepatic glucose output (Santoro et al., 2017). These results suggest that mitochondrial dynamics in POMC neurons couples metabolic status with appropriate glucose homeostasis by different effects to peripheral tissues. The involvement of the hypothalamus in insulin release control has been known for decades, albeit the identity of the neurons mediating such effects has remained elusive. Previous studies had hinted that POMC neurons could be implicated as central delivery of a-MSH or a melanocortin receptor agonist lowered serum insulin (Fan et al., 2000; Mansour et al., 2010). However, more direct evidence was lacking. Here, we unveiled a functional link between POMC neurons and the endocrine

pancreas. Indeed, insulin secretion after a glucose challenge was blunted in POMCMfn1KO mice. This effect appeared to have a neural basis, as sympathetic antagonism normalized insulin release, and ex vivo GSIS in mutant mice islets was normal. In contrast to studies linking autonomic defects with alterations in islet architecture and maturation (Borden et al., 2013; Tarussio et al., 2014), POMCMfn1KO mice did not exhibit abnormal islet anatomy, suggesting a mild and transient sympathetic deregulation. Altered GSIS in POMCMfn1KO mice was associated with increased mitochondrial respiration and ROS production in response to glucose. Consistently, deletion of Mfn1 is sufficient to enhance respiratory capacity in different cell types in a cell-autonomous manner (Kulkarni et al., 2016). ROS have emerged as key signals in the hypothalamus regulating energy metabolism (Shadel and Horvath, 2015; Drougard et al., 2015). Central ROS mediates GSIS in a model of intracarotid glucose delivery, a process that relies on DRP1 in the ventromedial hypothalamus (VMH) (Leloup et al., 2006; Carneiro et al., 2012). This is in contrast with our data showing that enhanced hypothalamic ROS impairs insulin release, which could be attributed to the lack of anatomical specificity of the intracarotid glucose load or the different biological properties of VMH and POMC neurons. Of note, enhanced ROS in POMCMfn1KO mice was not associated with appetite or body weight changes. The moderate and transient ROS levels of our mutant mice, in comparison with pharmacological doses (Diano et al., 2011), may not be sufficient to engage a feeding response. Divergent ROS sources could also explain this discrepancy, as it has been reported different physiological ROS effects depending on the site of production (Scialo` et al., 2016). Strikingly, loss of each of the three main fusion proteins in POMC neurons leads to divergent outcomes (Schneeberger et al., 2013, 2015), suggesting that these disparate phenotypes are not caused by impaired mitochondrial fusion per se. This disparity is not the consequence of heterogeneous expression of fusion proteins in subsets of POMC neurons, which is in line with their ubiquitous and critical role (Schrepfer and Scorrano, 2016). These contrasting phenotypes denote qualitatively distinct fusogenic properties and/or additional functions of these proteins. An alternative explanation could be related with different susceptibility to genetic mutations of distinct subpopulations of POMC neurons, which could be ultimately coupled to specific pathways that regulate diverse biological functions. We suggest a model by which loss of MFN1 in POMC neurons impacts on mitochondrial dynamics flexibility upon metabolic challenges, thereby altering local bioenergetic and redox homeostasis in a transient and moderate manner. This may interfere with adequate POMC neuron activational responses, affecting glucose sensing and subsequent physiological adaptations to insulin release. Our data support the notion that neuronal sensors play a key role in this biological process (Osundiji and Evans, 2013). Although the contribution of POMC neurons to the regulation of insulin secretion is likely modest, especially when compared with pancreatic mechanisms, our results suggest that POMC neurons may be critical in promoting and finetuning GSIS. Thus, defective POMC neuronal function may be Cell Metabolism 25, 1390–1399, June 6, 2017 1397

involved in the pathophysiology of altered insulin release in type 2 diabetes.

Received: May 31, 2016 Revised: April 3, 2017 Accepted: May 23, 2017 Published: June 6, 2017

STAR+METHODS REFERENCES

Detailed methods are provided in the online version of this paper and include the following: d d d d

d

KEY RESOURCES TABLE CONTACT FOR REAGENT AND RESOURCE SHARING EXPERIMENTAL MODEL AND SUBJECT DETAILS B Animals METHOD DETAILS B Physiological Measurements B Physiological Tests B High-Fat Diet Studies B I.c.v. Cannulation and Treatments B Sympathetic and Parasympathetic Function B Double POMC and C-FOS Immunofluorescence B POMC Neuron Count and Area B Double MFN1 and MFN2 Immunofluorescence B Mitochondrial Aspect Ratio Analysis in POMC Neurons B Electron Microscopy and Mitochondrial Analysis B Hypothalamic a-MSH, b-endorphin, and Protein Carbonyl Analysis B Global Transcriptomic Analysis B qPCR B RNA-Seq Data Analysis B Pancreas Morphometry B Islet Isolation and Insulin Secretion B Sympathetic and Parasympathetic Innervation B Mitochondrial Respirometry B ROS Measurements STATISTICAL ANALYSIS

Borden, P., Houtz, J., Leach, S.D., and Kuruvilla, R. (2013). Sympathetic innervation during development is necessary for pancreatic islet architecture and functional maturation. Cell Rep. 4, 287–301. Carneiro, L., Allard, C., Guissard, C., Fioramonti, X., Tourrel-Cuzin, C., Bailbe´, D., Barreau, C., Offer, G., Ne´delec, E., Salin, B., et al. (2012). Importance of mitochondrial dynamin-related protein 1 in hypothalamic glucose sensitivity in rats. Antioxid. Redox Signal. 17, 433–444. Chen, H., McCaffery, J.M., and Chan, D.C. (2007). Mitochondrial fusion protects against neurodegeneration in the cerebellum. Cell 130, 548–562. Chhabra, K.H., Adams, J.M., Fagel, B., Lam, D.D., Qi, N., Rubinstein, M., and Low, M.J. (2016). Hypothalamic POMC deficiency improves glucose tolerance despite insulin resistance by increasing glycosuria. Diabetes 65, 660–672. Claret, M., Smith, M.A., Batterham, R.L., Selman, C., Choudhury, A.I., Fryer, L.G., Clements, M., Al-Qassab, H., Heffron, H., Xu, A.W., et al. (2007). AMPK is essential for energy homeostasis regulation and glucose sensing by POMC and AgRP neurons. J. Clin. Invest. 117, 2325–2336. de Brito, O.M., and Scorrano, L. (2008). Mitofusin 2 tethers endoplasmic reticulum to mitochondria. Nature 456, 605–610. Diano, S., Liu, Z.W., Jeong, J.K., Dietrich, M.O., Ruan, H.B., Kim, E., Suyama, S., Kelly, K., Gyengesi, E., Arbiser, J.L., et al. (2011). Peroxisome proliferationassociated control of reactive oxygen species sets melanocortin tone and feeding in diet-induced obesity. Nat. Med. 17, 1121–1127. do Carmo, J.M., da Silva, A.A., Rushing, J.S., Pace, B., and Hall, J.E. (2013). Differential control of metabolic and cardiovascular functions by melanocortin-4 receptors in proopiomelanocortin neurons. Am. J. Physiol. Regul. Integr. Comp. Physiol. 305, R359–R368. Drougard, A., Fournel, A., Valet, P., and Knauf, C. (2015). Impact of hypothalamic reactive oxygen species in the regulation of energy metabolism and food intake. Front. Neurosci. 9, 56.

SUPPLEMENTAL INFORMATION

Fan, W., Dinulescu, D.M., Butler, A.A., Zhou, J., Marks, D.L., and Cone, R.D. (2000). The central melanocortin system can directly regulate serum insulin levels. Endocrinology 141, 3072–3079.

Supplemental Information includes four figures and three tables and can be found with this article online at http://dx.doi.org/10.1016/j.cmet.2017.05.010.

Gao, A.W., Canto´, C., and Houtkooper, R.H. (2014). Mitochondrial response to nutrient availability and its role in metabolic disease. EMBO Mol. Med. 6, 580–589.

AUTHOR CONTRIBUTIONS S.R., A.G.G.-V., M.S., L.V., R.H.-T., J.A., A.D., A.F.-M., M.I., P.M.G.-R., S.C., R.G., R.N., D.A.C., C.K., J.-M.S., and M.C. designed/performed experiments and analyzed data. E.N., I.C., M.P., A.G., A.V.-I., and P.G.-P. performed experiments. A.Z., T.L.H., and R.G. provided reagents and intellectual input. M.C. conceived the study, supervised research, secured funding, and wrote the manuscript with input from all authors. ACKNOWLEDGMENTS We thank Gregory S. Barsh (Stanford University, USA) for providing Pomc-Cre mice and Oscar Yanes (Rovira i Virgili University, Spain) for technical support. This work has been funded by: PI13/01604, integrated in the Plan Nacional I+D+I and co-funded by ISCIII-Subdireccio´n General de Evaluacio´n and European Regional Development Fund (ERDF): a way to build Europe (M.C.); EFSD/ Lilly Fellowship Award (EFSD/LILLY_12_1_002) (M.C.); 2014SGR659 (R.G.) and 2014SGR48 (A.Z.) Generalitat de Catalunya; SAF2013-40987R from MINECO (A.Z.); NIH Grants DK111178, AG051459, and AG052986 (T.L.H.). M.C. is a recipient of a Miguel Servet 2 contract (MSII15/00025) and R.H.-T. of a FAPESP fellowship (2016/01868-2). This work was carried out in part at the Esther Koplowitz Centre.

1398 Cell Metabolism 25, 1390–1399, June 6, 2017

Ibrahim, N., Bosch, M.A., Smart, J.L., Qiu, J., Rubinstein, M., Rønnekleiv, O.K., Low, M.J., and Kelly, M.J. (2003). Hypothalamic proopiomelanocortin neurons are glucose responsive and express K(ATP) channels. Endocrinology 144, 1331–1340. Ishihara, N., Eura, Y., and Mihara, K. (2004). Mitofusin 1 and 2 play distinct roles in mitochondrial fusion reactions via GTPase activity. J. Cell Sci. 117, 6535–6546. Kulkarni, S.S., Joffraud, M., Boutant, M., Ratajczak, J., Gao, A.W., Maclachlan, C., Hernandez-Alvarez, M.I., Raymond, F., Metairon, S., Descombes, P., et al. (2016). Mfn1 deficiency in the liver protects against diet-induced insulin resistance and enhances the hypoglycemic effect of metformin. Diabetes 65, 3552–3560. Lam, B.Y.H., Cimino, I., Polex-Wolf, J., Nicole Kohnke, S., Rimmington, D., Iyemere, V., Heeley, N., Cossetti, C., Schulte, R., Saraiva, L.R., et al. (2017). Heterogeneity of hypothalamic pro-opiomelanocortin-expressing neurons revealed by single-cell RNA sequencing. Mol. Metab. 6, 383–392. Leloup, C., Magnan, C., Benani, A., Bonnet, E., Alquier, T., Offer, G., Carriere, A., Pe´riquet, A., Fernandez, Y., Ktorza, A., et al. (2006). Mitochondrial reactive oxygen species are required for hypothalamic glucose sensing. Diabetes 55, 2084–2090. Liesa, M., and Shirihai, O.S. (2013). Mitochondrial dynamics in the regulation of nutrient utilization and energy expenditure. Cell Metab. 17, 491–506.

Mansour, M., White, D., Wernette, C., Dennis, J., Tao, Y.X., Collins, R., Parker, L., and Morrison, E. (2010). Pancreatic neuronal melanocortin-4 receptor modulates serum insulin levels independent of leptin receptor. Endocrine 37, 220–230. €nzberg, H., Opland, D., Faouzi, M., Villanueva, E.C., Mori, H., Inoki, K., Mu Ikenoue, T., Kwiatkowski, D., MacDougald, O.A., Myers, M.G., Jr., and Guan, K.L. (2009). Critical role for hypothalamic mTOR activity in energy balance. Cell Metab. 9, 362–374. Mun˜oz, J.P., Ivanova, S., Sa´nchez-Wandelmer, J., Martı´nez-Cristo´bal, P., Noguera, E., Sancho, A., Dı´az-Ramos, A., Herna´ndez-Alvarez, M.I., Sebastia´n, D., Mauvezin, C., et al. (2013). Mfn2 modulates the UPR and mitochondrial function via repression of PERK. EMBO J. 32, 2348–2361. Novak, A., Guo, C., Yang, W., Nagy, A., and Lobe, C.G. (2000). Z/EG, a double reporter mouse line that expresses enhanced green fluorescent protein upon Cre-mediated excision. Genesis 28, 147–155. Osundiji, M.A., and Evans, M.L. (2013). Brain control of insulin and glucagon secretion. Endocrinol. Metab. Clin. North Am. 42, 1–14. Parton, L.E., Ye, C.P., Coppari, R., Enriori, P.J., Choi, B., Zhang, C.Y., Xu, C., Vianna, C.R., Balthasar, N., Lee, C.E., et al. (2007). Glucose sensing by POMC neurons regulates glucose homeostasis and is impaired in obesity. Nature 449, 228–232. Ramadori, G., Fujikawa, T., Fukuda, M., Anderson, J., Morgan, D.A., Mostoslavsky, R., Stuart, R.C., Perello, M., Vianna, C.R., Nillni, E.A., et al. (2010). SIRT1 deacetylase in POMC neurons is required for homeostatic defenses against diet-induced obesity. Cell Metab. 12, 78–87. Santoro, A., Campolo, M., Liu, C., Sesaki, H., Meli, R., Liu, Z.W., Kim, J.D., and Diano, S. (2017). DRP1 suppresses leptin and glucose sensing of POMC neurons. Cell Metab. 25, 647–660. Schneeberger, M., Dietrich, M.O., Sebastia´n, D., Imberno´n, M., Castan˜o, C., Garcia, A., Esteban, Y., Gonzalez-Franquesa, A., Rodrı´guez, I.C., Bortolozzi,

A., et al. (2013). Mitofusin 2 in POMC neurons connects ER stress with leptin resistance and energy imbalance. Cell 155, 172–187. Schneeberger, M., Gomis, R., and Claret, M. (2014). Hypothalamic and brainstem neuronal circuits controlling homeostatic energy balance. J. Endocrinol. 220, T25–T46. Schneeberger, M., Go´mez-Valade´s, A.G., Altirriba, J., Sebastia´n, D., Ramı´rez, S., Garcia, A., Esteban, Y., Drougard, A., Ferre´s-Coy, A., Bortolozzi, A., et al. (2015). Reduced a-MSH underlies hypothalamic ER-stress-induced hepatic gluconeogenesis. Cell Rep. 12, 361–370. Schrepfer, E., and Scorrano, L. (2016). Mitofusins, from mitochondria to metabolism. Mol. Cell 61, 683–694. Scialo`, F., Sriram, A., Ferna´ndez-Ayala, D., Gubina, N., Lo˜hmus, M., Nelson, G., Logan, A., Cooper, H.M., Navas, P., Enrı´quez, J.A., et al. (2016). Mitochondrial ROS produced via reverse electron transport extend animal lifespan. Cell Metab. 23, 725–734. Shadel, G.S., and Horvath, T.L. (2015). Mitochondrial ROS signaling in organismal homeostasis. Cell 163, 560–569. Smith, M.A., Katsouri, L., Irvine, E.E., Hankir, M.K., Pedroni, S.M., Voshol, P.J., Gordon, M.W., Choudhury, A.I., Woods, A., Vidal-Puig, A., et al. (2015). Ribosomal S6K1 in POMC and AgRP neurons regulates glucose homeostasis but not feeding behavior in mice. Cell Rep. 11, 335–343. Tarussio, D., Metref, S., Seyer, P., Mounien, L., Vallois, D., Magnan, C., Foretz, M., and Thorens, B. (2014). Nervous glucose sensing regulates postnatal b cell proliferation and glucose homeostasis. J. Clin. Invest. 124, 413–424. Thorens, B. (2014). Neural regulation of pancreatic islet cell mass and function. Diabetes Obes. Metab. 16 (Suppl 1 ), 87–95. Xu, A.W., Kaelin, C.B., Takeda, K., Akira, S., Schwartz, M.W., and Barsh, G.S. (2005). PI3K integrates the action of insulin and leptin on hypothalamic neurons. J. Clin. Invest. 115, 951–958.

Cell Metabolism 25, 1390–1399, June 6, 2017 1399

STAR+METHODS KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Antibodies Alexa 488 chicken anti-rabbit

Thermo Fisher Scientific

Cat# A21441; RRID: AB_2535859

Alexa 488, Donkey Anti-Rabbit

Jackson ImmunoResearch

Cat# 711-546-152; RRID: AB_2340619

Alexa 555, Goat anti-Guinea Pig

Thermo Fisher Scientific

Cat# A21435; RRID: AB_2535856

Alexa 594 goat anti-rabbit

Thermo Fisher Scientific

Cat# A11012; RRID: AB_10562717

Alexa 647 streptavidin-conjugated

Thermo Fisher Scientific

Cat# S21374; RRID: AB_2336066

Guinea pig polyclonal anti-VACht

Millipore

Cat# AB1588; RRID: AB_2187981

Guinea Pig polyclonal anti-insulin

Dako

Cat# 0564; RRID: AB_10013624

Mouse monoclonal anti-Mfn2

Abcam

Cat# AB56889; RRID: AB_2142629

Rabbit polyclonal anti-c-FOS

Santa Cruz

Cat# Sc-52; RRID: AB_2106783

Rabbit polyclonal anti-glucagon

Dako

Cat# A0565; RRID: AB_10013726

Rabbit polyclonal anti-Mfn1

Santa Cruz

Cat# Sc-50330; RRID: AB_2250540

Rabbit polyclonal anti-POMC

Phoenix Pharmaceuticals

Cat# H-029-30; RRID: AB_2307442

Rabbit polyclonal anti-TH

Novus Biologicals

Cat# NB300-109; RRID: AB_350437

Rabbit polyclonal anti-Tom-20

Santa Cruz

Cat# Sc-11415; RRID: AB_2207533

Sheep polyclonal anti-a-MSH

Millipore

Cat# AB5087; RRID: AB_91683

TSA Biotin tyramide reagent

PerkinElmer

Cat# SAT700001EA

2-deoxy-D-glucose

Sigma-Aldrich

Cat# D6134

3-30 -Diaminobenzidine

Sigma-Aldrich

Cat# D5637

Adenosine diphosphate (ADP)

Calbiochem

Cat# 117105-1GM

Amplex Red Reagent

Thermo Fisher Scientific

Cat# A12222

Antimycin A

Sigma-Aldrich

Cat# A8674

Aprotinin

Sigma-Aldrich

Cat# A6279

Artificial cerebrospinal fluid (aCSF)

Tocris Bioscience

Cat# 3525

Carbachol

Sigma-Aldrich

Cat# PHR1511

Carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCF)

Sigma-Aldrich

Cat# C2920

Chemicals, Peptides, and Recombinant Proteins

Collagenase

Sigma-Aldrich

Cat# C5138

Digitonin

Sigma-Aldrich

Cat# D5628

Durcupan

Electron Microscopy Sciences

Cat# 14040

Glucagon

Novo Nordisk

Cat# 965616.4

Glucose 40%

Fresenius Kabi

Cat# 620724

Glutamate

Sigma-Aldrich

Cat# G1626

Glutaraldehyde solution 50%

Panreac AppliChem

Cat# A3166,0100

Histopaque-1119

Sigma-Aldrich

Cat# 11191

Honokiol

Sigma-Aldrich

Cat# H4914

Human insulin

Lilly

C.N 710008.9

Hydrogen peroxide solution

Sigma-Aldrich

Cat# H1009

Idazoxan hydrochloride

Sigma-Aldrich

Cat# I6138

L-Arginine

Sigma-Aldrich

Cat# W381918

Lipofundin

Fresenius Kabi

C.N 600120 0 H

Malate

Sigma-Aldrich

Cat# M1000

N-acetylcisteine

Sigma-Aldrich

Cat# A9165

Osmium tetroxide

Electron Microscopy Sciences

Cat# 19190 (Continued on next page)

e1 Cell Metabolism 25, 1390–1399.e1–e6, June 6, 2017

Continued REAGENT or RESOURCE

SOURCE

IDENTIFIER

Peroxidase from horseradish

Sigma-Aldrich

Cat# P8375

Picric acid solution

Sigma-Aldrich

Cat# P6744

Rotenone

Sigma-Aldrich

Cat# R8875

Sodium pyruvate

Sigma-Aldrich

Cat# P2256

Succinate

Sigma-Aldrich

Cat# S2378

Tauroursodeoxycholic acid (TUDCA)

Calbiochem

Cat# 580549

Trizol

Thermo Fisher Scientific

Cat# 15596026

Tyloxapol

Sigma-Aldrich

Cat# T0307

Uranyl acetate

Electron Microscopy Sciences

Cat# 22400

Vectastain Elite ABC HRP Kit

Vector Laboratories

Cat# PK-6100

Critical Commercial Assays 2-CAT Plasma ELISA High Sensitive

Labor Diagnostika Nord

Cat# BA E-4500

Corticosterone EIA

Immunodiagnostic Systems

Cat# AC-14F1

Endorphin, beta - EIA

Phoenix Pharmaceuticals

Cat# EK-022-06

GeneChip Mouse HT_MG-430_PM Array Plate

Affymetrix

Cat# 901257

Glucagon ELISA

Mercodia

Cat# 10-1281-01

MSH, alpha - EIA

Phoenix Pharmaceuticals

Cat# EK-043-01

Protein Carbonyl ELISA

Enzo Life Sciences

Cat# ALX-850-312-KI01

RNeasy MinElute Cleanup kit

QIAGEN

Cat# 74204

Serum Triglyceride Determination kit

Sigma-Aldrich

Cat# TR0100-1KT

Ultra Sensitive Mouse Insulin ELISA

Crystal Chem

Cat# 90080

This paper; Gene Expression Omnibus

GEO: GSE81993

Deposited Data Transcriptomic analysis Experimental Models: Organisms/Strains LacZ/EGFP mouse

Novak et al., 2000

N/A

Mfn1loxP mouse

Chen et al., 2007

N/A

Opa1loxP mouse

Dr. Antonio Zorzano laboratory

N/A

POMC-Cre mouse

Xu et al., 2005

N/A

N/A

N/A

DAVID Functional Annotation Tool

NIH, Open source

https://david.ncifcrf.gov/

ImageJ software

NIH, Open source

https://imagej-nih-gov.sire.ub.edu/ij/

LIMMA software package

Bioconductor, Open source

https://www.bioconductor.org/

Prism

GraphPad Software

https://www.graphpad.com/scientificsoftware/prism/

Research Diets

Cat# D12451

Oligonucleotides See Table S3 Software and Algorithms

Other High-fat diet 45% Kcal from fat

CONTACT FOR REAGENT AND RESOURCE SHARING Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Marc Claret ([email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS Animals The generation of POMCMfn1KO and POMCZ/EG mice has been previously reported (Schneeberger et al., 2013). Mice with floxed alleles for Opa1 were generated by the Mouse Mutant Core Facility of the IRB Barcelona and crossed with POMC-Cre mice to

Cell Metabolism 25, 1390–1399.e1–e6, June 6, 2017 e2

generate POMCOpa1KO animals. Colonies were maintained by breeding POMC-Cre; loxp/loxp mice with loxp/loxp mice. Mice were maintained on a light and temperature controlled conditions with free access to water and standard chow (Harlan). In vivo studies were performed with approval of the University of Barcelona Ethics Committee, complying with current Spanish and European legislation. METHOD DETAILS Physiological Measurements Commercially available ELISA kits were used to measure plasma levels of insulin (Crystal Chem) and glucagon (Mercodia). For glucagon, 200 ml of blood was collected in EDTA coated tubes with 500 KIU/ml of aprotinin (Sigma-Aldrich) and centrifuged at 3600 rpm for 20 min at 4 C. Total epinephrine and norepinephrine was measured in tissue extracts and subsequent ELISA (Labor Diagnostika Nord). For the extraction of catecholamines, tissues were digested in 800 ml of 0.01N HCl with 0.3 mg/ml of ascorbic acid using a homogenizer. Samples were centrifuged at 6000 rpm for 20 min at 4 C. Blood triglycerides (TAG) were measured using a hand-monitor device. For VLDL measurement, food deprived (6h) mice were i.p. injected with 500 mg/Kg of Tyloxapol (Sigma-Aldrich) and blood samples taken at the indicated time points. Plasma and liver TAG’s were measured using quantitative enzymatic determination TAG kit (Sigma-Aldrich). Plasma corticosterone levels after insulin (i.p., 0.4 IU/Kg) or 2-DG (i.p., 200 mg/Kg) were measured using ELISA (ImmunoDiagnostic Systems). Physiological Tests Glucose tolerance (2 g/kg), GSIS (3 g/kg), pyruvate (sodium pyruvate, 1 g/Kg) and arginine-stimulated insulin release (1 g/Kg) tests were performed in 16h fasted mice. Insulin (0.4 IU/kg), glucagon sensitivity (1 mg/Kg) and glucagon secretion by insulin-induced hypoglycemia (0.7 IU/Kg) tests were performed on 6h food-deprived mice. All compounds were intraperitoneally (i.p.) injected. High-Fat Diet Studies Control and mutant mice were fed with high-fat diet (45% Kcal derived from fat; Research Diets), starting at 6 weeks of age for 12 consecutive weeks. Mice were fasted overnight (16h), basal glycemia was measured using a glucometer and plasma obtained for insulin analysis. The homeostatic model assessment of insulin resistance (HOMA-IR) was calculated using the equation [(G0 3 I0)/405)], where G0 and I0 are fasting glucose and insulin, respectively. Measurements were taken twice (after 10 and 12 weeks of diet) and average results plotted. I.c.v. Cannulation and Treatments I.c.v. surgery was performed as described (Schneeberger et al., 2013). On experimental days, mice were overnight (16h) fasted and infused with 2 ml of vehicle (aCSF or Lipofundin), Honokiol (37.6 mM; Sigma-Aldrich), N-acetylcysteine (2.5 mg/l; Sigma-Aldrich) or TUDCA (2.5 g/l; Calbiochem) just after lights on. Two hours later, a GSIS was performed. For glucose-sensing tests, 16h fasted mice were injected with 2 ml of either saline or D-glucose (25 g/l) and food intake measured. Ad libitum fed mice were administered with 2-DG (50 g/l). Sympathetic and Parasympathetic Function Mice were overnight (16h) fasted. Saline, Idazoxan (0.4 mg/Kg; Sigma-Aldrich) or Carbachol (0.1 mg/Kg; Sigma-Aldrich) were i.p. injected just after lights on. Thirty minutes later a GSIS was performed. Double POMC and C-FOS Immunofluorescence Mice were i.p. injected with either vehicle or D-glucose (2 g/Kg) and transcardially perfused (4% paraformaldehyde) 60 min later. Brains were cryoprotected (30% sucrose), frozen at 80 C and subsequently cut into 25 mm-thick slices. Brain sections were blocked with 2% goat serum in KPBS + 0.4% Triton X-100 and incubated with rabbit anti C-FOS antibody (1:5000; Santa Cruz) in blocking solution for 16h at 4 C. As secondary antibody, a goat anti-rabbit Alexa Fluor 594 (1:400; Life Technologies) in KPBS + 0.4% Triton X-100 was used (90 min at room temperature (RT)). After extensive washing, slices were incubated with rabbit antiPOMC precursor antibody (1:1000; Phoenix Pharmaceuticals) 16h at 4 C. After washing, slices were incubated with chicken antirabbit Alexa Fluor 488 (1:400; Life Technologies) for 90 min. Imaging was performed using a Leica DMI 6000B microscope with a 20x objective. A minimum of 650 POMC neurons from 3 mice/genotype were included in the analysis. POMC Neuron Count and Area The number of POMC neurons were determined as described (Claret et al., 2007). Briefly, brain slices from control and mutant mice were stained for POMC as described above. Average somatic area was analyzed in > 500 POMC neurons (n = 3 mice/genotype). The area occupied by POMC neurons was manually scored using ImageJ software. Double MFN1 and MFN2 Immunofluorescence To co-localize mitofusin proteins with POMC neurons, we used brain sections from adult POMCZ/EG reporter mice that express the green fluorescent protein in this subset of neurons (Schneeberger et al., 2013). Mice were transcardially perfused (4% e3 Cell Metabolism 25, 1390–1399.e1–e6, June 6, 2017

paraformaldehyde) and brain sections were obtained as described above. To ensure reliable protein staining, we conducted primary antibody controls to show its specificity in binding to the antigen (using tissue slices from MFN1 and MFN2 knock-out mice) and secondary antibody controls to show specific labeling to the primary antibody. Twenty micrometer brain sections were blocked with 2% goat serum in KPBS + 1% Triton X-100 and incubated with rabbit anti MFN1 antibody (1:100; Santa Cruz) in blocking solution overnight at 4 C. As secondary antibody, a goat anti-rabbit Alexa Fluor 594 (1:300; Life Technologies) in KPBS + 1% Triton X-100 was used (2h at RT). After extensive washing, slices were blocked in 2% chicken serum in KPBS + 1% Triton X-100 and then incubated (overnight at 4 C) with mouse anti-MFN2 (1:200; Abcam) in KBPS + 1% Triton X-100. Afterward, biotinylated tyramide amplification and fluorescent detection with streptavidin-conjugated Alexa Fluor 647 (1:700; Life Technologies) was conducted. Images were taken using a Leica DMI 4000B confocal microscope. Twenty micrometer Z stack images were generated. For analysis and data presentation, the entire arcuate nucleus of the hypothalamus was divided in 3 regions identified as rostral (1.2 - 1.5 mm from bregma), mid (1.6 - 2.2 mm from bregma) and caudal (2.3 - 2.7 mm from bregma). The analysis was done in 508 (rostral), 1229 (mid) and 119 (caudal) POMC neurons from 4 animals. Mitochondrial Aspect Ratio Analysis in POMC Neurons Analysis of cellular mitochondrial aspect ratio by confocal microscopy is a difficult process, and precautions are needed for reliable image acquisition and analysis. Brain sections (30 mm thick) were blocked with 2% chicken serum in KPBS + 0.4% Triton X-100 and incubated with rabbit anti-POMC precursor (1:1000; Phoenix Pharmaceuticals) in blocking solution overnight at 4 C. As secondary antibody, a chicken anti-rabbit Alexa Fluor 488 (1:300; Life Technologies) in KPBS + 0.4% Triton X-100 was used (120 min at RT). After extensive washing, slices were blocked with 2% goat serum in KPBS + 0.4% Triton X-100 and incubated with rabbit antiTOM20 (1:200; Santa Cruz) for 48h at 4 C. After washing, slices were incubated with goat anti-rabbit Alexa Fluor 594 (1:300; Life Technologies) for 120 min at RT. Images were acquired sequentially, using 405,488, 532 nm laser lines, with a Leica TCS SPE laser scanning confocal system with a 40x oil immersion objective. The confocal pinhole was set at 1 Airy unit. The format was 1024x1024. Specific settings for frame average (6 frames per image), laser gain (1000 for 488 nm laser and 830 for 532 nm laser) and smart offset (1.2% offset) were selected for all pictures to improve image quality. Images were segmented by threshold to select the cellular area of study. TOM20 staining was submitted to background subtraction and filtering processes with Gaussian Blur filter set at 0.8. Segmentation of POMC neuron mitochondria was performed by local standard threshold using a radius of 7. Mitochondria of each POMC neuron were then subjected to particle analysis for acquiring aspect ratio (AR = major axis/minor axis). An AR value of 1 indicates a perfect circle, and as mitochondria elongates and become more elliptical AR increases. Particles smaller than 1.2 mm2 were excluded. For analysis, 2 mm z stacks over the diameter of each POMC neuron were performed with a zoom of 4x. Examination of R 50 neurons/mouse from at least 3 mice per group is advisable. The analysis was performed using the ImageJ Launcher software. Electron Microscopy and Mitochondrial Analysis Mice were transcardiacally perfused with 0.9% saline with heparine followed by fixative solution (paraformaldehyde 4%, gluteraldehyde 0.1%, picric acid 15% in phosphate buffer (PB) 0.1M, pH = 7.4). Brains were removed and fixed overnight at 4 C with the same fixative without gluteraldehyde. Brains were washed vigorously with ice-cold PB 0.1 M, and sliced at 50 mm in a vibratome. Sections containing the ARC were stained for POMC (1:4000; 48h at 4 C with gentle shaking; Phoenix Pharmaceuticals). After extensive washes, slices were incubated with secondary antibody, then with ABC and finally developed using 3,30 -Diaminobenzidine. After developing, slices were put in osmium tetroxide (1%, 15 min) and then dehydrated in an ethanol gradient. Uranyl acetate (1%) was added to 70% ethanol to enhance ultrastructural contrast. Slices were then embedded in Durcupan, cut in an ultra microtome and collected in grids for posterior analyzes. A Tecnai 12 Biotwin electron microscope was used to visualize the ultrastructure of the samples, and POMC neurons were imaged at 13000x magnification for posterior offline analyzes. For mitochondria analyzes, random sections of POMC neurons cut throughout the middle of the cell body were analyzed. Most of these sections contained the nucleus. ImageJ software was used to manually outline each individual mitochondrion in the digital images. All samples were checked twice for consistency of mitochondria labeling. We used mitochondria cross-sectional area as a measurement of mitochondria size. Mitochondria density was estimated by dividing the number of mitochondria profiles by the cell area. Mitochondria coverage was estimated by dividing the total area of mitochondria (sum of all mitochondria profiles in a given cell) by the cytosol area. Hypothalamic a-MSH, b-endorphin, and Protein Carbonyl Analysis Hypothalami from fasted (16h) mice were harvested and immediately frozen in liquid nitrogen. Hypothalami were sonicated in 500 ml of 0.1N HCl solution. Lysates were centrifuged and supernatants used for neuropeptide (Phoenix Pharmaceuticals) or protein carbonyl content (Enzo Life Sciences) determination by ELISA. Total protein concentration was determined by Bradford. For immunofluorescence analysis, mice were transcardially perfused and brain processed as above. Sections containing the PVN were stained for a-MSH (1:20000; 48h at 4 C; Chemicon) and fluorometric density measured using ImageJ software. Global Transcriptomic Analysis Hypothalamus mRNA from control and POMCMfn1KO mice under ad libitum fed or fasting (16h) conditions (n = 3-4/genotype/condition) was isolated using Trizol (Ambion) and RNeasy MinElute Cleanup kit (QIAGEN). RNA was hybridized onto Affymetrix GeneChip Mouse HT_MG-430_PM Array Plate at the Functional Genomics Core Facility (IDIBAPS). Expression data were normalized with a Cell Metabolism 25, 1390–1399.e1–e6, June 6, 2017 e4

robust multi-array average (RMA). The LIMMA software package from Bioconductor (http://www.bioconductor.org) was used for statistical analysis to identify differences in gene expression using a multiple test-adjusted p value. A fold-change of ± 1.4 and a FDR % 0.05 was considered significant. The DAVID Functional Annotation Tool (https://david.ncifcrf.gov/) was used to identify enriched functional categories among differentially expressed genes. Non-redundant biological process categories and KEGG pathways significantly (p < 0.05) enriched (up- and downregulated) in control animals during the fast to fed transition were ranked. The accession number for the transcriptomic data reported in this paper is GEO: GSE81993. qPCR Tissues were harvested and immediately frozen in liquid nitrogen. mRNA was isolated using Trizol and retrotranscription performed using reagents from Applied Biosystems. qPCR was conducted using Premix Ex Taq (Takara) in ABI Prism 7900 HT system (Applied Biosystems). Proprietary Taqman Gene Expression assay FAM/TAMRA probes (Applied Biosystems) used for qPCR analysis are listed in Table S3. RNA-Seq Data Analysis To address whether transcripts encoding for mitochondrial fusion proteins (Mfn1, Mfn2, Opa1) are expressed in POMC neurons we used publically available single-cell RNA sequencing raw data for POMC neurons (Lam et al., 2017; GEO: GSE92707). According to the recommendations of the software (edgeR) used in the original analysis, we considered a gene expressed when the raw count was > 10. Data is presented as log2 (CPM+1). 3D graph was performed under R language with the rgl package. R squared and P values for correlations were obtained from Pearson analysis. Pancreas Morphometry Pancreases were dissected, fixed in 10% formalin neutral buffered solution (16h) and embedded in paraffin. Sections (4 mm) from three different levels (> 150 mm apart) for each pancreas were deparaffinised, rehydrated, microwaved in citrate buffer (10 mM; pH = 6.0) and permeabilized with 1% Triton X-100 in PBS. Sections were incubated with guinea pig anti-insulin (1:2500; Dako) and rabbit-anti glucagon (1:1000; Dako). As secondary antibodies, Alexa 555 goat anti-guinea pig (1:200; ThermoFisher) and Alexa 488 donkey anti-rabbit (1:200; Jackson Immunoresearch) were used. Imaging was performed using a Leica DMI 6000B microscope and the analysis performed using ImageJ software. Total alpha and beta-cell mass was calculated by multiplying their fractional area per pancreas weight. Islet Isolation and Insulin Secretion Islets were isolated by gentle collagenase digestion (1 mg/ml) followed by Histopaque gradient (Sigma-Aldrich) and manually handpicked under stereomicroscope. Insulin secretion was evaluated using freshly isolated islets in static incubation assays and normalized by content. Insulin was measured by ELISA (Crystal Chem). Sympathetic and Parasympathetic Innervation For measurement of sympathetic and parasympathetic innervation, sections (30 mm apart) from adult pancreata were immunostained for tyrosine hydroxylase (rabbit anti-TH antibody, 1:1000; Novus Biologicals) or VAChT (guinea pig anti-VAChT, 1:100; Chemicon), respectively. Randomly chosen 45-50 islets were analyzed. Sympathetic innervation was measured as the density of the TH+ area per endocrine area, whereas parasympathetic innervation was measured as the density of VAChT+ puncta per endocrine area. Mitochondrial Respirometry Mitochondrial function was measured by high-resolution respirometry (Oxygraph-2k, Oroboros Instruments), as described previously (Schneeberger et al., 2013). Fresh hypothalamic arcuate-enriched microdissections were obtained and mechanically homogenized and permeabilized (with digitonin) in cold respiration media. Leak respiration was measured by adding malate (2 mM) and pyruvate (10 mM), in the absence of ADP. Complex I oxidative phosphorylation was measured by the addition of ADP (5 mM, Oxphos I). Subsequently, glutamate (20 mM) and succinate (10 mM) were added to assess electron transfer in the NADH and succinated-linked pathway (Oxphos I+II). Next, carbonylcyanide-4-(trifluoromethoxy)-phenyl-hydrazone (FCCP) was titrated to achieve maximum flux through the electron transfer system (ETS I+II) (1 mM). Finally, respiration was inhibited by the sequential addition of rotenone (0.1 mM) and antimycin A (2.5 mM). The remaining O2 flux after inhibition with antimycin A (O2 flux independent of the electron transfer system) was subtracted to calculate the different respiratory states. Oxygen flux values are expressed relative to protein content determined by Bradford method. ROS Measurements Fluorometric Methods Amplex Red (10-acetyl-3,7-dihydrophenoxazine) was used as a fluorescent probe for the detection of hydrogen peroxide (H2O2). Hypothalamic explants from food deprived (6h) mice were washed in Krebs–Ringer bicarbonate/glucose (2.8 mM) buffer (pH = 7.4) in an atmosphere of 95% O2–5% CO2 and then incubated in Eppendorf tubes containing 400 ml of the same medium plus Amplex Red and horseradish peroxidase (4 U/ml; Sigma) at 37 C. Basal fluorescence (excitation: 544 nm, emission: 590 nm) was measured in a Biotek Synergy HT plate reader. Subsequently, the media were replaced with the same chromogenic buffer with high (16.7 mM) glucose. e5 Cell Metabolism 25, 1390–1399.e1–e6, June 6, 2017

After 30 min of incubation at 37 C, a sample was collected and measured. A standard H2O2 curve was generated to quantify H2O2 release by the hypothalamus in response to high glucose. Real-Time Amperometric Measurements For real-time amperometric H2O2 analysis, hypothalami were processed as above. After a 10 min recovery period, the spontaneous H2O2 release was measured (at 37 C for 20 min) using an H2O2-specific amperometric probe (ISO-HPO, 100 mm diameter, and 5 mm length; World Precision Instruments) implanted directly in the hypothalamus. The calibration of the probe was done as detailed in the manufacturer’s instructions (ISO-HPO-100; World Precision Instruments [WPI]; Aston). The probe was left in 0.1 M phosphate-buffered saline (PBS) buffer until stabilization. Afterward, a H2O2 solution (ranging from 100 to 800 nM) was added in the PBS solution. The current recorded was directly proportional to the H2O2 concentration. Probes were tested before each experiment to validate their sensitivity, which should be at least 1 pA/nM. Real-time concentration of H2O2 gas in the tissue was measured (Apollo1000; World Precision Instruments) at a sampling rate of 10 values/s and data acquired by Labscribe2 software (World Precision Instruments). Data are expressed as a delta variation of H2O2 release from basal. STATISTICAL ANALYSIS Data are expressed as mean ± SEM. P values were calculated using unpaired Student’s t test, two-way ANOVA with Tukey’s multiple comparisons test or one-way ANOVA with Sidak multiple comparisons test as appropriate. p < 0.05 was considered significant.

Cell Metabolism 25, 1390–1399.e1–e6, June 6, 2017 e6