Injury to hypothalamic Sim1 neurons is a common feature of obesity by exposure to high fat diet in male and female mice

Abstract

The hypothalamus is essential for regulation of energy homeostasis and metabolism. Feeding hypercaloric, high-fat (HF) diet induces hypothalamic arcuate nucleus injury and alters metabolism more severely in male than in female mice. The site(s) and extent of hypothalamic injury in male and female mice are not completely understood. In the paraventricular nucleus (PVN) of the hypothalamus, Single-Minded Family BHLH Transcription Factor 1 (Sim1) neurons are essential to control energy homeostasis. We tested the hypothesis that exposure to HF diet induces injury to Sim1 neurons in the PVN of male and female mice. Mice expressing membrane-bound enhanced green fluorescent protein (mEGFP) in Sim1 neurons (Sim1-Cre:Rosa-mEGFP mice) were generated to visualize the effects of exposure to HF diet on these neurons. Male and female Sim1-Cre:Rosa-mEGFP mice exposed to HF diet had increased weight, hyperleptinemia, and developed hepatosteatosis. In male and female mice exposed to HF diet, expression of mEGFP was reduced by >40% in Sim1 neurons of the PVN, an effect paralleled by cell apoptosis and neuronal loss, but not by microgliosis. In the arcuate nucleus of the Sim1-Cre:Rosa-mEGFP male mice, there was decreased alpha-melanocyte-stimulating hormone in pro-opiomelanocortin neurons projecting to the PVN, with increased cell apoptosis, neuronal loss, and microgliosis. These defects were undetectable in the arcuate nucleus of female mice exposed to the HF diet. Thus, injury to Sim1 neurons of the PVN is a shared feature of exposure to HF diet in mice of both sexes, while injury to pro-opiomelanocortin neurons in arcuate nucleus is specific to male mice.

GRAPHICAL ABSTRACT

Hypothalamic Sim1 neurons are essential to control energy homeostasis, but it is not known whether exposure to high fat diet induces injury to these neurons. We propose that injury to Sim1 neurons in the paraventricular nucleus (PVN) of hypothalamus is a common feature of obesity by exposure to high fat diet in male and female mice. Injury to Sim1 neurons is not accompanied by microgliosis in the PVN. High fat diet induces injury to Pro-opiomelanocortin neurons in the arcuate nucleus (ARC) with decreased α-MSH and microgliosis in male, but not in female mice. Thus, this study establishes Sim1 neurons in the PVN as a target of HF diet-induced neuronal injury in mice of both sexes

INTRODUCTION

Exposure to high fat (HF) diets causes injury to peripheral tissues and to the brain in humans and rodents (Flier 2004, De Souza et al. 2005, Velloso et al. 2009, Moraes et al. 2009, Thaler et al. 2012, Dorfman & Thaler 2015). In the melanocortin pathway, proopiomelanocortin (POMC) neurons localized in the arcuate nucleus of the hypothalamus are involved in the regulation of energy balance by responding to the peripheral hormones leptin and insulin. These hormones induce increased release of the anorexigenic hormone α-Melanocyte-stimulating hormone (α-MSH) by the POMC neurons. On the other hand, leptin and insulin inhibit secretion of the orexigenic hormones Agouti-related peptide (AgRP) and Neuropeptide Y (NPY) by the AgRP/NPY neurons, which are also localized to the arcuate nucleus (Cone 2005, Krashes et al. 2016). In the melanocortin system, POMC neurons projecting to the paraventricular nuclei (PVN) of the hypothalamus release α-MSH, which binds to Melanocortin-4 receptor (MC4R), a G-protein-coupled receptor expressed by MC4R neurons localized in the PVN, to reduce food intake (Chen et al. 2012, Li et al. 2016). Conversely, the MC4R antagonist/inverse agonist AgRP binds to MC4R to promote feeding (Cone 2005, Krashes et al. 2016). Male rodents treated with HF diet have hypothalamic inflammation, gliosis, and neuronal loss. These changes take place in the arcuate nucleus, mediobasal hypothalamus, and lateral hypothalamus, thus indicating that these hypothalamic areas are targets of brain injury (Flier 2004, De Souza et al. 2005, Velloso et al. 2009, Moraes et al. 2009, Thaler et al. 2012, Dorfman & Thaler 2015). It has also been reported that, in male rodents, exposure to HF diet decreases the number of POMC neurons in the arcuate nucleus, impairs post-translational processing of POMC to generate α-MSH, and reduces POMC neuron synapses as well as secretion of α-MSH (Levin 1999, Enriori et al. 2007, Horvath et al. 2010, Thaler et al. 2012, Schneeberger et al. 2013, Cakir et al. 2013, Yi et al. 2017). Other reports have instead found that, in male mice, HF diet exposure alters the volume, rather than the total number of arcuate nucleus neurons, including the POMC neurons (Lemus et al. 2015). Female rodents exposed to HF diet, while having less severe adverse metabolic consequences and less prominent microgliosis in the arcuate nucleus and mediobasal hypothalamus than male mice, nevertheless develop obesity (Clegg et al. 2011, Hong et al. 2009, Atamni et al. 2016, Dorfman et al. 2017, Qiu et al. 2018). These observations pose the question of whether, at least in female mice, exposure to HF diet disrupts energy homeostasis by inducing injury to neurons in hypothalamic areas other than the arcuate nucleus, mediobasal hypothalamus, and lateral hypothalamus that are injured in male mice (Shah et al. 2014, Balthasar et al. 2005, Xu et al. 2013). MC4R neurons in the PVN co-express Single-Minded Family BHLH (Basic Helix-Loop Helix) Transcription Factor 1 (Sim1). Importantly, in rodents, loss of Sim1 neurons causes obesity with hyperphagia and decreased energy expenditure (Xi et al. 2012, Tolson et al. 2014). Moreover, mutation of Sim1 also causes obesity and hyperphagia, without decreased energy expenditure, both in rodents and humans (Holder et al. 2000, Michaud et al. 2001, Kublaoui et al. 2006, Traurig et al. 2009, Ghoussaini et al. 2010, Ramachandrappa et al. 2013, Tolson et al. 2010, El Khattabi et al. 2015). These observations indicate that Sim1 neurons have a central role in energy homeostasis.

In respect to diets with elevated fat, the predominant source of saturated fat in the human diet is palmitic acid, which is abundant in palm oil. Palm oil has been increasingly used to replace dietary trans fatty acids, and is the major food oil in world markets (Hayes & Pronczuk 2010). Nevertheless, palm oil consumption has adverse metabolic consequences and promotes tissue injury (Vega-Lopez et al. 2006, Sun et al. 2015, Perfilyev et al. 2017, Fattore et al. 2014, Rosqvist et al. 2014). Here, we asked whether exposure to a hypercaloric HF diet, with excess fat derived from palm oil, induces injury to Sim1 neurons in the PVN of male and female mice. To this end, we monitored injury to Sim1 neurons in mice of both sexes exposed to HF diet by measuring abundance of membrane-bound enhanced green fluorescent protein (mEGFP) expressed specifically in the cell body and processes of Sim1 neurons in the PVN. We also measured cell apoptosis, neuronal loss and microgliosis in the same population of neurons.

MATERIALS AND METHODS

Pre-registration statement: the study was not pre-registered

Materials

Pierce Bicinchoninic Acid (BCA) Protein Assay Kit (Cat #: 23225) was purchased from Thermo Scientific (Rockford, IL, USA). Ultra-Sensitive Mouse Insulin ELISA (Enzyme-Linked Immunosorbent Assay) Kit (Cat #: 90080) and Leptin ELISA Kit (Cat #: 90030) were obtained from Crystal Chem (Elk Grove Village, IL, USA). Free Fatty Acid (FFA) Quantification Colorimetric/Fluorometric Kit (Cat #: K612) was obtained from BioVision (Milpitas, CA, USA, RRID: SCR_005057). Nile Red (Cat# N3013) and bovine serum albumin (BSA) (Cat # A7511–10G) were from Sigma-Aldrich (St. Louis, MO, USA, RRID: SCR_008988). Heparin (Cat # P87721) was purchased from Braun Medical. (Bethlehem, PA, USA). Click-iT TUNEL (Terminal deoxynucleotidyl transferase dUTP Nick End Labeling) Assay kit (Cat # C10619) and ProLong Gold antifade reagent with 4,6-diamidino-2-phenylindole (DAPI) (Cat # P36935) were purchased from Molecular Probes (Eugene, OR, USA, RRID: SCR_013318). Capillary tubes for retro-orbital bleeding (Cat #: 22–260943) and formaldehyde (Cat # BP531–500) were purchased from Fisher Scientific (Pittsburgh, PA, USA, RRID: SCR_008452). Donkey Cy5 conjugated anti-rabbit antibodies (dilution: 1/500, Cat # 711–175-152, RRID: AB_2340607) were from Jackson Immunoresearch (West Grove, PA, USA, RRID: SCR_010488). Rabbit polyclonal anti-Ionized calcium binding adaptor molecule 1 (Iba1) antibodies (dilution: 1/1000, Cat# 019–19741, RRID: AB_839504) were from Wako Pure Chemical Industries, Ltd (Richmond, VA, USA, RRID: SCR_013651). Normal donkey serum (Cat # Ab7475), rabbit polyclonal anti-α-MSH antibodies (dilution: 1/1000, Cat# ab123811, RRID: AB_10976325) and rabbit polyclonal anti-NeuN antibodies (dilution: 1/1000, Cat# Ab128886, RRID: AB_2744676) were purchased from Abcam (Cambridge, MA, USA, RRID: SCR_012931). The lipid standards 1, 2-diheptadecanoyl-sn-glycero-3-phosphoethanolamine (PE C17:0/C17:0, Cat #: 830756, CAS #: 140219–78-9) and 1, 2-diheptadecanoyl-sn-glycero-3-phosphocholine (PC C17:0/C17:0, Cat #: 850360, CAS #: 70897–27-7) were from Avanti Polar Lipids, Inc. (Alabaster, AL, USA, RRID: SCR_016391). The 2-propanol (i-PrOH) Optima LC/MS grade (Cat # A461–500, CAS #: 67–63-0), methyl tert-butyl ether (MTBE) HPLC grade (Cat #: AA41839AK, CAS #: 1634–04-4), and sodium chloride (Cat #: S271–500, CAS #: 7647–14-5) were from Fisher Scientific (Fair Lawn, NJ, USA, RRID: SCR_008452). Formic acid (Cat #: 33015, CAS #: 64–18-6), water (Cat # 34877, CAS #: 7732–18-5) and methanol of HPLC grade (Cat #: 34860, CAS #: 67–56-1), ammonium acetate (Cat #: A1542, CAS# :631–61-8), sodium fluoride of BioXtra grade (Cat #: S7920, CAS #: 7681–49-4) were purchased from Sigma-Aldrich, Co. (St. Louis, MO, USA, RRID: SCR_008988).

Generation of Sim1-Cre: Rosa-mEGFP mice

All studies used male and female C57BL/6 background mice. To generate Sim1-Cre:Rosa-mEGFP mice with expression of membrane bound EGFP (mEGFP) in Sim1 neurons, Tg(Sim1-Cre) 1 Low L/J mice (RRID:IMSR_JAX:006451) expressing Cre recombinase under the Sim1 promoter (Balthasar et al. 2005) obtained from the Jackson Laboratory (Bar Harbor, ME, USA, RRID:SCR_004633) were crossed with Gt(ROSA-MEGFP)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J ( RRID:IMSR_JAX:007676) (Rosa-mEGFP ) mice (Muzumdar et al. 2007) also obtained from the Jackson Laboratory.

Diet and animal care

All Sim1-Cre:Rosa-mEGFP mice used in the study were bred and conditioned with low fat (LF) and HF diets in the University of Arkansas for Medical Sciences (UAMS) vivarium. Mice were housed in a temperature-controlled environment with 12h: 12h light-dark cycle, “lights on” at 0600 h and “lights off” at 1800 h and given ad libitum access to food and water. Each shoebox cage hosted 2–4 mice and had the addition of one “mouse hut” per cage for enrichment (from Bioserv, Cat # K3102). UAMS veterinarians checked on the animals daily to monitor their health. Pups were genotyped at day 19–21 of age by following the Jackson Laboratories recommended protocols and primers. The DNA used for genotyping was derived from a small piece of tail (<2 mm). All mice with the correct genotype were used for the study. Animals were weaned at 21 days of age. No randomization methods were used. No blinding was performed. Eight week-old mice were fed in parallel either with D12450H LF diet (3.82 kcal/g, fat = 10 kcal %) from Research Diets, Inc., or with the custom HF diet D15012001 (4.70 kcal/g, fat = 45 kcal %) from Research Diets, Inc. for 15 weeks (Table I and Table II). To make the D15012001 HF diet, palm oil was included in the ingredients and cornstarch was decreased as compared to that of the D12450H LF diet, while sucrose content is approximately the same. Food intake and body weight were recorded every week.

Table I.

Mouse diet	D12450H	D15012001
	10 % LF diet	45% HF diet
Caloric Information
Protein:	20% kcal	20% kcal
Carbohydrate:	70% kcal	35% kcal
Fat:	10% kcal	45% kcal
Energy Density:	3.82 kcal/g	4.7 kcal/g
Diet ingredients
Casein	200.00 g	200.00 g
L-Cystine,	3.00 g	3.00 g
Corn Starch	452.20 g	72.8 g
Sucrose	176.80 g	172.80 g
Maltodextrin	75.00 g	100 g
Cellulose	50.00 g	50.00 g
Soybean Oil, USP	25.00 g	25.00 g
Lard	20.00 g	20.00 g
Palm Oil	0.00 g	157.5 g
Fatty acid profile
Saturated (g)	10.1 g	80.2 g
Monounsaturated (g)	12.8 g	79.1 g
Polyunsaturated (g)	20.2 g	41 g
Total fat	43.1 g	200.3 g

Table II.

Mouse diet	D12450H	D15012001
	10 % LF diet	45% HF diet
Fatty acid
composition (g)
C2, Acetic	0	0
C4, Butyric	0	0
C6, Caproic	0	0
C8, Caprylic	0	0
C10, Capric	0	0
C12, Lauric	0	0.5
C14, Myristic	0.3	1.5
C14:1, Myristoleic	0	0
C15	0	0
C16, Palmitic	6.4	65.8
C16:1, Palmitoleic	0.3	0.5
C16:2	0	0
C16:3	0	0
C16:4	0	0
C17	0.1	0.1
C17:1	0	0
C18, Stearic	3.1	11.3
C18:1, Oleic	12.3	78.5
C18:2, Linoleic	17.8	37.8
C18:3, Linolenic	2.1	2.9
C18:4, Stearidonic	0	0
C20, Arachidic	0.1	0.9
C20:1	0.2	0.2
C20:2	0.2	0.2
C20:3	0	0
C20:4, Arachidonic	0.1	0.1
C20:5, Eicosapentaenoic	0	0
C21:5	0	0
C22, Behenic	0.1	0.1
C22:1, Erucic	0	0
C22:4, Clupanodonic	0	0
C22:5, Docosapentaenoic	0	0
C22:6, Docosahexaenoic	0	0
C24, Lignoceric	0	0

For liver and brain mass spectrometry analysis, male C57BL/6J DIO mice (Stock #: 380050, RRID:IMSR_JAX:000664) were received from Jackson Laboratory (Bar Harbor, ME, USA, RRID:SCR_004633), and housed in polycarbonate cages on a 12-hour light-dark cycle at University of Arkansas for Medical Sciences. Mice were administered control diet D12450B (3.85 kcal/g, 10 kcal% fat) and a HF diet D12492 (5.24 kcal/g, 60 kcal% fat) starting from week 6 of age and continued for a total of 9 weeks (one group per cage, n = 4 mice). Mice (15-week old) were fasted for 4h, retro-orbital blood was collected and mice were euthanized by CO₂ asphyxiation and decapitated by guillotine. The brain was harvested, with the hypothalamus dissected as a ~ 2 mm wide block. The liver was harvested from the same mice and dissected as a ~ 2 mm wide block. Tissues were snap frozen in liquid nitrogen and stored at −80°C. The UAMS Institutional Animal Care and Use Committee (IACUC) approved the protocol used for the animal studies (AUP FILE approval #: 3788). Insulin and leptin in serum collected at sacrifice were measured by using Ultra-Sensitive Mouse Insulin and leptin ELISA kits according to the manufacturer’s instructions.

Measuring food intake

Male and female mice on LF and HF diets were housed 2–4 mice per cage. Three cages were used for each group. Food intake per cage was measured every week. To calculate the average weekly food intake per mouse we divided the amount of food eaten (in grams) by the number of mice in the cage. To calculate caloric intake per mouse, the amount of food in grams was multiplied by the amount of kcal/g of food. Collection of blood from the retro-orbital sinus is done at sacrifice.

Measuring blood glucose levels

Three drops of blood were individually collected on the Bayer Contour Glucose Test Strip (Save Rite Medical, Cat #: 56–7080), which was previously inserted in the blood glucose meter (Save Rite Medical, Cat #: 567189). Three glucose meter readings per mouse were recorded and data are reported as the average of the three readings.

Measures to minimize animal suffering

Blood collection:

Mice were brought to the laboratory from the UAMS vivarium for fasting and euthanasia. The animal was first deeply anesthetized in the induction chamber connected to the isofluorane vaporizer (VetEquip, Inc. Livermore, CA, and USA). The oxygen flow meter was set to 1.5 liters/min and the isofluorane vaporizer dial was set at the maximal level of 5%. The animal, when lying on its side and breathing rhythmically, was removed from the chamber and deep anesthesia was monitored by paw pinch, tail pinch, and eye blink tests. Approximately 0.2 ml of blood was collected from the retro-orbital sinus using a capillary inserted at the medial canthus of the eye under the nictitating membrane and the animal was immediately returned to the induction chamber connected to the isofluorane vaporizer. Heart exposure: The animal was removed from the induction chamber and placed on a tray with its face in the nose cone of a circuit to maintain the isoflurane/oxygen administration. Deep anesthesia was again monitored by paw pinch, tail pinch, and eye blink tests. The heart was then exposed and the animal was perfused through the left ventricle with heparinized saline for 30 min, and then with fixative using a Phosphate Buffered Saline (PBS) solution with 4% formaldehyde for another 30 min.

Detection of mEGFP fluorescence, immunofluorescence, apoptosis in brain tissues, and lipid staining in liver

Detection of mEGFP fluorescence: Brains were harvested from male Sim1-Cre:Rosa-mEGFP mice kept in parallel on LF diet and HF diet that were initiated at 8-weeks of age. Male mice on LF and HF diets were sacrificed alternately after 4h starvation between 1100h and 1500h on the same day. Brains from male and female mice were not collected on the same day. Mice deeply anesthetized with isofluorane were perfused though the left ventricle with heparinized saline (0.9% NaCl containing 2 units of heparin/ ml at a rate of 3–4 ml/min for 30 minutes and then with 4% formaldehyde in PBS, pH 7.4 for another 30 minutes. The brain was then harvested, post-fixed in PBS containing 4% formaldehyde for 48 hours at room temperature, washed with PBS and stored in PBS containing 0.01% sodium azide until being cut by vibratome to obtain 50 μm serial sections. Sections containing: a) paraventricular nucleus (PVN) and nucleus of lateral olfactory tract (LOT, Bregma −0.70 to −1.06 mm) and b) medial amygdala (MeA) (Bregma −1.34 to −2.06) (Paxinos & Franklin) were transferred to microscope slides to directly visualize mEGFP fluorescence.

Immunofluorescence:

Serial adjacent sections of the PVN and of the arcuate nucleus (Bregma −1.22 to −1.94 mm) of Sim1-Cre:Rosa-mEGFP mice were immunostained with antibodies against α-MSH, NeuN, and Iba1, respectively. Sections of male mice treated with LF and HF diets were immunostained at the same time. The same protocol of staining was used for female mice. Sections were incubated for 1h at room temperature on a plate shaker set at 200 rpm with 0.5% Triton x-100 in PBS (permeabilization step). Sections were then incubated for 1h at room temperature on a plate shaker set at 200 rpm with PBS containing 0.1% Triton x-100 (PBST) and normal 10% donkey serum (blocking step). Then samples were incubated with PBST containing 1% BSA and primary antibody on a plate shaker set at 200 rpm at 4°C for 48 hours, washed four times for 10 minutes with PBS at room temperature, and incubated with PBST 0.1% containing 1% BSA and secondary antibody on a plate shaker set at 200 rpm at 4°C overnight. Sections were then washed four times with PBS for 10 minutes at room temperature and transferred to gelatin-coated microscope slides. Tissues on microscope slides were dried in the dark for 15 minutes before adding 1,4-diazabicyclo[2.2.2]octane (DABCO) mounting medium (40 ml containing 100 mg DABCO dissolved in 10 ml PBS with the addition of 30 ml glycerol) and mounting the coverslip and mounting the coverslip.

Detection of apoptotic cells

Apoptotic cells were visualized in alternate sections of the PVN and of the arcuate nucleus of Sim1-Cre:Rosa-mEGFP mice used for the immunofluorescence studies. Apoptotic cells were visualized by using Click-iT TUNEL Assay kit according to the manufacturer’s instructions with few modifications. Free-floating sections were permeabilized using PBST 0.5% for an hour at room temperature. Sections were then transferred to chrome gelatin-coated microscope slides, placed in humid chambers at room temperature and incubated with 100 µl of the Terminal Deoxynucleotidyl Transferase (TdT) reaction buffer (component A of the kit) for 10 minutes at 37°C. A cocktail containing TdT enzyme, TdT reaction buffer, and EdUTP (a dUTP modified with a small bio-orthogonal alkyne moiety) was prepared according to the manufacturer’s instructions. Terminal Deoxynucleotidyl Transferase (TdT) reaction buffer was aspirated from the tissues. Each section was incubated with 100 µl TdT cocktail spread evenly on the tissue for 60 minutes at 37°C in a humidified chamber. Tissues were rinsed with deionized water and washed with 3% BSA, and 0.1% Triton X-100 in PBS for 5 minutes, followed by one rinse with 1X PBS. Click-iT Plus TUNEL reaction cocktail containing Alexa Fluor 647 picolyl azide, copper protectant, Click-iT Plus TUNEL Reaction Buffer, and Buffer Additive was prepared according to the manufacturer’s instructions. Each brain section was incubated with 100 µl of the Click-iT Plus TUNEL reaction cocktail for 30 minutes at 37°C in dark. The cocktail was aspirated from the tissues and the tissues were washed with 3% BSA in PBS for 5 minutes at room temperature. The tissues were washed with 1X PBS for 5 minutes at room temperature. Tissues were incubated with 300ng/µl DAPI for 15 minutes at room temperature in dark. The DAPI solution was aspirated and the tissues were washed twice with PBS for 5 minutes each. The tissues were dried in the dark. One drop of ProLong Gold Antifade Mountant was added to each brain section and the coverslip was mounted.

Lipid staining in liver by Nile Red

Livers were harvested from the same animals immediately after the brain. After 48 hours post-fixation in PBS containing 4% formaldehyde, the liver was washed with PBS and stored in PBS containing 30% sucrose and 0.01% sodium azide in a 50 ml conical tube and kept at 4° C for the following 48 hours, when the tissue sank to the bottom of the tube. The liver was embedded in Optimum Cutting Temperature (OCT) medium at −20 degrees and cut into 30 µm sections using a cryostat. Sections were stored in PBS containing 0.01% sodium azide in a 24 well plate. Free floating liver sections were permeabilized by using 0.5% TritonX-100 in PBS for an hour at room temperature. The liver sections were then incubated with 0.5 µM Nile Red overnight at 4° C on a plate shaker set at 200 rpm. Liver sections were washed 4 times with PBS for 10 minutes each and incubated with 14.3 µm DAPI in PBS. Sections were washed 3 times with PBS and transferred gelatin-coated microscope slide. Sections were dried in dark for 10 minutes before adding DABCO mounting medium.

Microscopy

Images were taken using a confocal microscope (Olympus Fluoview FV1000) equipped with 20 X/0.85 N.A Plan Apochromatic oil objective. Images of tissues were collected as z-stacks of 15 optical slices of 4.0 µm thickness. The Z-stacks were converted into a two-dimensional maximum intensity projection image (MIP) by using the Olympus software. Higher magnification images were taken with the 60X/1.42 N.A Plan Apochromatic oil objective. All images of the same experiment were obtained with identical acquisition parameters. “Regions of Interest” (ROIs) were drawn to exclude edge artifacts. When indicated, brain images were taken using the 4 x objective of the EVOS® FL Auto Imaging System. Each image was quantified using FIJI (ImageJ) software.

Data analysis.

Matching brain coronal sections from mice on LF diet and HF diet were used for the analysis. MIP images were exported from the Olympus Fluoview FV10-ASW software (RRID: SCR_014215) as multi-tiff images. The multi-tiff images were opened with the ImageJ software (NIH, RRID: SCR_003070) and ROIs with same areas were drawn in matching sections of the PVN or arcuate nucleus of mice treated with LF or HF diet. The integrated fluorescence intensity of the ROIs was measured. To count the number of neurons and apoptotic cells within the ROIs, multi-tiff images were opened with ImageJ software. Images were adjusted for analysis by setting identical thresholds for male and female brains, respectively (for NeuN staining upper threshold = 4095 pixels for males and females and lower threshold = 600 pixels for males and 400 pixels for females; for Tunel staining, upper threshold = 3895 pixels, lower threshold = 800 pixels for males and females). Using the “Analyze Particles” application within the Analyze menu of ImageJ, the computer generated a table containing the number of neurons or of TUNEL positive cells (objects counts) in the ROI. To measure microglia area coverage within the ROIs, multi-tiff MIP images were opened in ImageJ, and ROIs were drawn in the ARC and PVN. Images were adjusted for analysis (upper threshold = 3898 pixels and the lower threshold = 1000 pixels). Within the “Analyze” menu, and “Analyze Particle” application, tables were generated containing the number of microglia “objects” counted, and the percent area within the ROI covered by microglia. Thickness of microglia processes was measured using ImageJ software by drawing, at a distance of 2 μm from the cell body, a segment across the process. By using Plot Profile application within the “Analyze” menu of ImageJ software, the computer generated a plot of pixel intensity (y-axis) along the length of the line (x-axis). The thickness of the processes was obtained by measuring the length of the X-axis under the peak. To analyze lipid droplet in livers, MIP images were exported from Olympus Fluoview software as indicated above. Multi-tiff images were opened in ImageJ and adjusted by setting the lower threshold to 1000 pixels and upper threshold to 3109 pixels. ROI was selected to exclude holes at the central vein and destroyed are of the tissues from the analysis. Using the Analyze particle Application within the analyze menu of ImageJ, the computer generated a summary table containing the number of lipid droplets, total area, average size of lipid droplets, and % area coverage. The table was exported as Excel sheet to compare the average size and the % area coverage.

Extraction of PC and PE from hypothalamus and liver

Hypothalamus.

Homogenization of the hypothalamus was carried out according to the method of Borg et al. (Borg et al. 2012) with some modifications. The hypothalamus (10–20 mg) was defrosted on ice, homogenized in 100 µl RIPA buffer (Tris-HCl 50 mM, NaCl 150 mM, EDTA 1 mM, NaF 1 mM, containing protease inhibitors, pH 7.4) in a 2 ml teflon pestle tissue homogenizer. The homogenizer was rinsed once with additional 150 µl RIPA buffer to obtain a total homogenate volume of 250 µl. Homogenates were frozen in 20 µl aliquots and kept at −80°C until their use. PC and PE extraction from the hypothalamus was carried out according to the method of Matyash et al. (Matyash et al. 2008) with some modifications. A solution for PC and PE extraction, methyl-tert-butyl (MTBE)/methanol (MeOH)/H₂O) (10:3:2.5, v/v/v) (washing solution), was mixed by shaking and the top phase was used as described below. Twenty µl of defrosted hypothalamus homogenate plus 5 µl of RIPA buffer were added to a mixture of 150 µl MeOH and 500 µl MTBE. After addition of internal standards (PC 17:0/17:0, PE 17:0/17:0, 10 μl volume in chloroform (CHCl₃)/MeOH/H₂O (60:30:4.5, v/v/v) samples were vortexed and mixed on rotary shaker continuously for 1 h at room temperature. To induce phase separation 120 µl HPLC grade water was added, samples were incubated for 10 min at room temperature and then centrifuged for 10 min at 1,000 g at 4°C. The upper organic phase was collected and PC and PE extraction was repeated once more by adding 200 µl of the washing solution to the lower water phase. The organic phases from these two extractions were then combined and dried in a vacuum centrifuge. The lipid film was reconstituted in 200 µl CHCl₃/MeOH/H₂O (60:30:4.5, v/v/v), sonicated for 15 minutes in an ice cold water bath and then stored at −20°C in glass auto sampler vials until analysis.

Liver.

Liver pieces (10–20 mg) were defrosted on ice, homogenized as described by others (Matyash et al. 2008) in 100 µl volume of a solution containing 0.1% (w/v) ammonium acetate using a Teflon pestle tissue homogenizer, which was then rinsed twice with 100 µl of the ammonium acetate solution. The 300 µl sample was frozen immediately at −80°C as 20 µl aliquots until use. In brief, 50 µl of the liver homogenate was added to a solution containing 225 µl MeOH and 750 µl MTBE. After addition of internal standards, brief vortexing, and mixing on rotary shaker as described above, phase separation was induced by adding 188 µl HPLC grade water to the samples, which were then incubated for 10 minutes at room temperature and centrifuged for 10 min at 1,000 g at 4°C. The upper organic phase was collected and PC and PE extraction was repeated once by adding 300 µl of the washing solution to the lower water phase. The organic phases were combined, dried and the lipid film was dissolved in 200 µl CHCl₃/MeOH/H₂O. (60:30:4.5, v/v/v), sonicated and then stored as described above.

Protein quantification

To determine the amount of protein in the sample, 10 µl aliquots from the tissue homogenate (see above) were analyzed by using the BCA protein assay kit according to the manufacture instructions.

Liquid chromatography tandem mass spectrometer quantification

PC and PE extracted from tissues were quantitated by liquid chromatography-tandem mass spectrometry using the Agilent 1290 Infinity LC system equipped with the Agilent Zorbax Eclipse Plus C18 2.1 × 100 mm 1.8-micron column and connected to the Agilent 6490 triple quadrupole mass analyzer. The column was kept at 50°C and the flow rate was maintained at 0.5 ml/min. Lipids extracts (5 µl) were directly injected onto the column operated with solvent A, consisting of 0.1% acetic acid and 1 mM ammonium acetate in 2-propanol (i-PrOH)/MeOH/H₂O (5:1:4, v/v/v) and solvent B consisting of HPLC grade MeOH containing 0.1% acetic acid and 1 mM ammonium acetate. Separation was performed by using the linear gradient as follows: from 0 to 2 min hold 0% solvent B; from 2 to 10 min a linear gradient (0 – 100% solvent B); from 10 to 17 min hold at 100% solvent B; and then return to 0% solvent B over 3 min. The electrospray ion source was set in positive ion mode. Agilent 6490 triple quadrupole tandem mass spectrometer parameters: MS1 heater: 100°C; MS2 heater: 100°C; Rough Vac.: 2.86E+0; High Vac: 4.08E-5; Source parameters (electrospray ionization): Gas Temp.: 200°C; Gas Flow: 14 l/min; Nebulizer: 20 psi; Sheath Gas Temp.: 250°C; Sheath Gas Flow: 11 l/min; Capillary: positive 3000V, negative 3000V; Nozzle Voltage: positive 1500V, negative 1500V.

Lipid species present in extracts of mouse liver and hypothalamus were measured using ion transitions corresponding to the neutral loss of the ethanolamine head group (M-141 Da, PE species) and the PC-specific fragment (M = 184 Da, PC species). A list of corresponding ion transition was assembled as described by others (Tsugawa et al. 2014, Sud et al. 2007). Quantitative data was then acquired in multiple reaction monitoring (MRM) mode monitoring 66 ion transitions specific for PC and PE species (Table III). Lipid concentrations were calculated based on a calibration made of PC and PE standards (PC C17:0/C17:0 and PE C17:0/C17:0). All phospholipid concentrations were normalized to total protein in their corresponding homogenates.

Table III.

MRM transitions of the phosphatidylcholine (PC) and phosphatidylethanolamine (PE) species for measurements in positive ion mode. The ion transition list was assembled as described by others (Tsugawa et al. 2014, Sud et al. 2007).

PC species	Precursor ion	Product ion		PE species	Precursor ion	Product ion
PC (28:0)	678.5	184.1		PE (28:0)	636.5	495.5
PC (30:0)	706.5	184.1		PE (30:0)	664.5	523.5
PC (30:1)	704.5	184.1		PE (30:1)	662.5	521.5
PC (32:0)	734.6	184.1		PE (32:0)	692.5	551.5
PC (32:1)	732.5	184.1		PE (32:1)	690.5	549.5
PC (32:2)	730.5	184.1		PE (32:2)	688.5	547.5
PC (34:0)	762.6	184.1		PE (34:0)	720.6	579.6
PC (34:1)	760.6	184.1		PE (34:1)	718.5	577.5
PC (34:2)	758.6	184.1		PE (34:2)	716.5	575.5
PC (34:3)	756.6	184.1		PE (34:3)	714.5	573.5
PC (34:4)	754.6	184.1		PE (34:4)	712.5	571.5
PC (36:0)	790.6	184.1		PE (36:0)	748.6	607.6
PC (36:1)	788.6	184.1		PE (36:1)	746.6	605.6
PC (36:2)	786.6	184.1		PE (36:2)	744.6	603.6
PC (36:3)	784.6	184.1		PE (36:3)	742.5	601.5
PC (36:4)	782.6	184.1		PE (36:4)	740.5	599.5
PC (36:5)	780.6	184.1		PE (36:5)	738.5	597.5
PC (36:6)	778.5	184.1		PE (36:6)	736.5	595.5
PC (38:1)	816.6	184.1		PE (38:1)	774.6	633.6
PC (38:2)	814.6	184.1		PE (38:2)	772.6	631.6
PC (38:3)	812.6	184.1		PE (38:3)	770.6	629.6
PC (38:4)	810.6	184.1		PE (38:4)	768.6	627.6
PC (38:5)	808.6	184.1		PE (38:5)	766.5	625.5
PC (38:6)	806.6	184.1		PE (38:6)	764.5	623.5
PC (38:7)	804.6	184.1		PE (38:7)	762.5	621.5
PC (40:1)	844.7	184.1		PE (40:1)	802.6	661.6
PC (40:2)	842.7	184.1		PE (40:2)	800.6	659.6
PC (40:3)	840.6	184.1		PE (40:3)	798.6	657.6
PC (40:4)	838.6	184.1		PE (40:4)	796.6	655.6
PC (40:5)	836.6	184.1		PE (40:5)	794.6	653.6
PC (40:6)	834.6	184.1		PE (40:6)	792.5	651.5
PC (40:7)	832.6	184.1		PE (40:7)	790.5	649.5
PC (40:8)	830.6	184.1		PE (40:8)	788.5	647.5

Statistical Analysis.

Our study was based on the assumption that the extent by which mEGFP abundance would decrease by exposure to HF diet would be similar to that of α-MSH in POMC neurons in the arcuate hypothalamus of male mice treated with HF diet (Schneeberger et al. 2013). We estimated that we could detect a change in mEGFP abundance at the p ≤ 0.05 level with a statistical power of ≥ 80% with n ≥ 5 mice/group. The data met the assumption of the statistical approach. The predefined criterion of the analysis has been to present all data including outliers. All statistical analyses were performed using GraphPad Prism 6 software (GraphPad Prism, RRID: SCR_002798). Statistical significance was calculated by two-tailed unpaired t-test with Welch correction on two groups or, where indicated, by two-way ANOVA with Holm-Sidak multiple comparison test on multiple groups. A value of P < 0.05 was considered statistically significant. Data were expressed as mean ± S.D, unless noted otherwise. No test for normality was performed.

Sample size differences.

Six male mice were treated with LF diet and eight male mice were treated with HF diet. Seven female mice were treated with LF diet and eight female mice were treated with HF diet. In the following experiments the sample size is different. Insulin in male and female mice (Fig. 2E and M, respectively) and glucose in male and female mice (Fig. 2F and N respectively), are not measured for all mice because we did not have enough samples for the analysis; the sample size for analysis of mEGFP in LOT of male and female mice (Fig. 3B and E, respectively) and MeA of male and female mice (Fig. 3C and F, respectively) is reduced because we did not have enough samples for the analysis; the sample size for TUNEL assay in male and female mice (Fig. 4A and E, respectively) and NeuN in female mice (Fig. 4F) is reduced because we did not have enough samples for the analysis; sample size for α-MSH, NeuN and Iba1 in female mice (Fig. 5F, G, H, I) is reduced because we did not have enough samples for the analysis.

Figure 2. Exposure to HF diet induces increased weight gain, hyperleptinemia and hepatosteatosis in male and female Sim1-Cre: Rosa-mEGFP mice.

A-H, metabolic parameters of male Sim1-Cre: Rosa mice treated with LF diet (LFD) and HF diet (HFD) for 15 weeks. A-B) Body weight (g) and weight gain (weight gain (g)/ weight (g) at the beginning of diet treatment x 100) of male mice exposed to LF diet (n = 6 mice) and HF diet (n = 8 mice) are expressed as mean ± SEM. Statistical analysis is carried out by using two-way ANOVA with Holm-Sidak multiple comparison test. C) Average weekly caloric intake per mouse is measured as described in “Material and Methods”. Data are derived from male mice exposed to LF diet (n = 6 mice) and HF diet (n = 8 mice) and are expressed as mean ± SEM. D) Leptin is measured in the serum of mice treated with LF diet (n = 6 mice) and HF diet (n = 8 mice). E) Insulin is measured in the serum of mice treated with LF diet (n = 5 mice) and HF diet (n = 8 mice). F) Blood glucose is measured in the serum of mice treated with LF diet (n = 6 mice) and HF diet (n = 7 mice). G) FFA level is measured in the serum of mice treated with LF diet (n = 6 mice) and HF diet (n = 8 mice). H) Lipid droplet size and area of liver occupied by lipid droplets (expressed as -lipid droplet area/liver area x 100-) of male mice treated with LF diet (n = 6 mice) and HF diet (n = 8 mice) were calculated from 3 MIP images per mouse, scale bar = 80 µm. I-P, metabolic parameters of female Sim1-Cre: Rosa mice. I-J) Body weight and weight gain of female mice treated with LF diet (n = 7 mice) and HF diet (n = 8 mice) are measured as in A and B respectively. K) Average weekly caloric intake is expressed as in C. L) Leptin in the serum of female mice treated with LF diet (n = 7 mice) and HF diet (n = 8 mice) is measured as in D. M) Insulinemia is measured in the serum of female mice treated with LF diet (n = 7 mice) and HF diet (n = 7 mice) as in E. N) Fasting blood glucose is measured in the serum of female mice treated with LF diet (n = 6 mice) and HF diet (n = 7 mice) (N). O) FFA in serum of female mice treated with LF diet (n = 7 mice) and HF diet (n = 8 mice) is measured as in G. P) Lipid droplet size and lipid percentage area coverage in livers of female mice treated with LF diet (n = 7 mice) and HF diet (n = 8 mice) were calculated as in H, scale bar = 80 µm. CV indicates the central vein. * Indicates P < 0.05, ** indicates P < 0.01, *** indicates P < 0.001; ****, P < 0.0001.

Figure 3. Expression of mEGFP in Sim1 neurons of the PVN is reduced in male and female mice exposed to HF diet, indicating neuronal injury.

A, Triangular ROIs with the same areas (white dotted lines) were drawn in MIP images to outline the PVN in matching sections of male mice treated with LF diet (LFD) and with HF diet (HFD). In the same sections, rectangular ROIs with the same areas (white dotted lines) were drawn to outline the AHC (upper panel, scale bar = 60 µm; middle and lower panels scale bar = 10 µm). The integrated pixel intensity of mEGFP fluorescence in PVN (within the triangular ROIs) and within the AHC (rectangular ROIs) is measured in mice treated with LF diet and HF diet (LF diet, n = 6 mice; HF diet, n = 8 mice). Data were derived from images taken at the right and left side of the 3^rd ventricle (3V) of three consecutive hypothalamic sections per mouse (6 MIP images per mouse, coefficient of variation per mouse ≤ 0.20). Data are normalized to control. B, Integrated pixel intensity of mEGFP fluorescence in the lateral olfactory tract is measured in ROI (LOT, LF diet, n = 3 mice; HF diet, n = 3 mice). Data are expressed as in A. C, Integrated pixel intensity of mEGFP fluorescence in the medial amygdala is measured in ROI (MeA, LF diet, n mice = 3; HF diet, n mice = 3). Data are expressed as in A, scale bar = 100 μm. D-F, female Sim1-Cre: Rosa mice. D, Integrated pixel intensities of mEGFP fluorescence in PVN and AHC are measured in ROIs as in A (LF diet, n = 7 mice; HF diet, n = 8 mice, coefficient of variation per mouse = ≤ 0.31). E, Integrated pixel intensity of mEGFP fluorescence in LOT is measured in ROI (LF diet, n = 4 mice; HF diet, n = 4 mice). Integrated pixel intensity of mEGFP fluorescence in MeA is measured in ROI (LF diet, n = 4 mice; HF diet, n = 5 mice).

Figure 4. HF diet induces apoptosis and neuronal loss in the PVN of male and female Sim1-Cre: Rosa mice without concomitant microglia activation.

A-D, Sim1-Cre: Rosa male mice; for all graphs data are normalized to control. A, TUNEL positive cells, within the PVN (ROI enclosed by white dotted line), appear with red fluorescence in the nuclei (upper panels, low magnification, scale bar = 60 µm and lowest panel, high magnification, scale bar = 16 µm). TUNEL positive cells in PVN ROIs were counted for each mouse using two MIPs, taken at the right and left side of the 3rd ventricle of male Sim1-Cre: Rosa mice treated with LF diet (LF diet, n = 6 mice) and with HF diet (HF diet, n = 7 mice). B, Cells positive for NeuN immunofluorescence within the PVN ROI (red fluorescence) were counted (LF diet, n = 6 mice; HF diet, n = 8 mice), scale bar = 60 µm. C, The quantification of Iba1 raw pixel density within the ROIs was carried out using two MIPs (LF diet, n = 6 mice; HF diet, n = 8 mice), scale bar = 60 µm. D, High-resolution image of individual microglia cells. Red bars indicate the region of the process where thickness of microglia processes was measured. The average microglia process thickness per mouse (LF diet, n = 6 mice; HF diet, n = 8 mice, 30 microglia cells per mouse); scale bar = 10 µm. E-H, Sim1-Cre: Rosa female mice; for all graphs data are normalized to control. E) TUNEL positive cells in the ROIs (LF diet, n = 4 mice; HF diet, n = 5 mice) were counted as in A. F, NeuN immunofluorescence (red fluorescence), within the PVN was analyzed as in B (LF diet, n mice = 4; HF diet, n mice = 5). G, The quantification of Iba1 raw pixel density within the ROIs was carried out as in C (LF diet, n mice = 7; HF diet, n mice = 8). H, Average microglia process thickness is measured as in D (LF diet, n = 7 mice; HF diet, n = 8 mice, 30 cells per mouse).

Figure 5. HF diet induces loss of α-MSH abundance, neuronal loss, and microglial activation in arcuate nucleus of male but not of female Sim1-Cre: Rosa mice.

A-E, Sim1-Cre: Rosa male mice, for all graphs data are normalized to the LF diet control. A, upper panel, dtTomato fluorescence in a hypothalamic section from HF diet-treated male mouse (Bregma – 1.70 mm, scale bar = 1 mm; Arc, arcuate nucleus; MeA, medial amygdala; mt, mammillothalamic tract; OPT, optic tract; VMH, ventromedial hypothalamus). Lower panels, the α-MSH immunostaining is pseudo-colored in red. The HF diet section used for α-MSH immunostaining is the same section as that visualized in the mtTomato fluorescence channel (upper panel). The quantification of α-MSH raw pixel density in ROIs (white dotted lines) was carried out using two MIPs, taken at the right and left side of the 3rd ventricle of male mice treated with LF diet and HF diet (LF diet, n = 6 mice; HF diet, n = 8 mice), scale bar = 60 µm. B, Immunostaining of α-MSH in the PVN of male Sim1-Cre: Rosa mice. The quantification of α-MSH raw pixel density was carried out using two MIPs, as described in A (LF diet, n = 6 mice; HF diet, n = 8 mice), scale bar = 60 µm. C, TUNEL positive cells, within the arcuate nucleus, visualized as in Fig. 3A, scale bar = 60 µm; an image at higher magnification (8X, scale bar = 16 µm). In the graph, the number of TUNEL positive cells were counted as in Fig. 3A (LF diet, n = 6 mice; HF diet, n = 8 mice). D, The number NeuN positive cells (red fluorescence) within the arcuate nucleus (ROI enclosed by white dotted line) were counted as in Fig. 3B (LF diet, n = 6 mice; HF diet, n = 8 mice), scale bar = 60 µm. E, Upper panels, the quantification of Iba1 raw pixel density within the ROIs and microglia coverage in ROI was carried out as in Fig. 3C (LF diet, n = 6 mice; HF diet, n = 8 mice), scale bar, 60 µm. Lower panels, high-resolution image of individual microglia cells. The average microglia process thickness per mouse (LF diet, n = 6 mice; HF diet, n = 8 mice, 30 microglia cells per mouse); scale bar = 10 µm. F-I, Sim1-Cre: Rosa female mice, for all graphs data are normalized to the control. F, quantification of α-MSH raw pixel density in ROIs is as in A (LF diet, n = 6 mice; HF diet, n = 7 mice). G, quantification of α-MSH in the PVN of female Sim1-Cre: Rosa mice is as in B (LF diet, n = 7 mice; HF diet, n = 7 mice). H, The number NeuN positive cells (red fluorescence) within the arcuate nucleus (ROI enclosed by white dotted line) were counted as in in D (LF diet, n = 4 mice; HF diet, n = 5 mice). I, quantification of Iba1 raw pixel density within the ROIs, microglia coverage, and in ROI was carried out as in E upper panels (LF diet, n = 6 mice; HF diet, n = 8 mice). Lower panels, high-resolution image of individual microglia cells. The average microglia process thickness per mouse (LF diet, n = 6 mice; HF diet, n = 8 mice, 30 microglia cells per mouse) was carried out as in E; scale bar = 10 µm.

RESULTS

In Sim1-Cre:Rosa-mEGFP mice, Sim1 neurons of the PVN express brightly fluorescent mEGFP targeted to neuronal membranes

The PVN of the hypothalamus includes Sim1 neurons, which are essential for regulation of food intake and weight (Xi et al. 2012, Tolson et al. 2014, Tolson et al. 2010). Sim1 neurons have been visualized by crossing mice expressing Cre-recombinase under the Sim1 promoter (Sim1-Cre mice) with Rosa-EGFP reporter mice, which express cytosolic EGFP, which is detectable by immunohistochemistry after Cre-mediated deletion of a loxP-flanked transcriptional blocker (Balthasar et al. 2005). MARCKS is a cytosolic protein kinase C substrate that associates to the cell membrane by acyl groups (Wiederkehr et al. 1997). Adding the plasma membrane targeting sequence of MARCKS to EGFP (mEGFP) results in expression of a brightly fluorescent protein, which can be detected at the cell plasma membrane (De Paola et al. 2003, Muzumdar et al. 2007). To generate Sim1-Cre:Rosa-mEGFP mice expressing mEGFP in Sim1 neurons, the Tg(Sim1-Cre) 1 Low L/J) mice expressing Cre recombinase under the Sim1 promoter were crossed with Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J mice, which express Cre-dependent cell membrane localized red and green fluorescence (Fig. 1A). In Sim1-Cre: Rosa-mEGFP mice, the mEGFP was evident by its intrinsic fluorescence in the PVN (Fig. 1B). In the PVN, mEGFP was localized to cell membranes of neuronal cell bodies (Fig. 1C arrows), and to neuronal processes detectable as individual fibers in the Anterior Hypothalamic Central Area (AHC) (Fig. 1C arrowheads). mEGFP was also detected in other brain regions, including the MeA and LOT (Fig. 1D), consistent with Sim1 neurons being localized to these areas (Balthasar et al. 2005). Male and female 8-week-old Sim1-Cre: Rosa-mEGFP mice were treated with HF and LF diet for 15 weeks (Fig. 1E).

Figure 1: Sim1-Cre: Rosa mice express mEGFP in Sim1 neurons of the PVN.

A, Scheme to describe the generation of Sim1-Cre: Rosa mice. B, Fluorescence microscopy of the hypothalamus of Rosa and Sim1-Cre: Rosa mice. Brain sections of the PVN from 8 week old male mice were imaged with the 20 X Olympus objective to visualize mEGFP in Sim1-Cre: Rosa mice expressing Cre recombinase under the Sim1 promoter (green fluorescence) and widespread dtTomato fuorescence (red fluorescence), but not mEGFP, in Rosa mice. Scale bars = 60 µm. C, mEGFP at the plasma membrane of Sim1 neuronal cell bodies (arrows, bar = 10 µm) and at Sim1 neuronal processes (arrowheads, bar = 5 µm) visualized using the 60 X objective of the Olympus microscope. D, mEGFP is also expressed medial amygdala (MeA) and nucleus of lateral olfactory tract (LOT); bar = 60 µm. E, Timeline of animal treatment. Mice were genotyped when they were 3 weeks old and started LF diet or HF diet when they were 8 weeks old (n = 6 male mice on LF diet, n= 8 male mice on HF diet, n= 7 female mice on LF diet, and n= 8 female mice on HF diet).

Exposure to HF diet induces increased weight and induces hyperleptinemia and hepatosteatosis in male and female Sim1-Cre: Rosa-mEGFP mice

A group of 8-week old male Sim1-Cre:Rosa-mEGFP mice (starting weight = 24.0 ± 1.3 grams), were treated for 15 weeks with control LF diet (LF diet, Table I). Another group of 8-week old male Sim1-Cre:Rosa-mEGFP mice (starting weight = 22.1 ± 1.5 grams), were treated in parallel with the HF diet. The body weight of male mice treated with HF diet was increased by 40.2 ± 16.4 % (p = 0.0001) as compared to that of male mice exposed to LF diet. The male mice exposed to HF diet gained 93.3 ±19.0 % (p < 0.0001) of their initial weight over the 15 weeks of treatment (Fig. 2A-B). In male mice exposed to HF diet, as compared to mice treated with LF diet, food intake was significantly increased at almost every time point of the treatment, starting from week 3 (Fig. 2 C). At sacrifice, male mice exposed to HF diet had increased serum leptin by 3.0 ± 0.5 fold (p = 0.0002) and insulin levels by 2.5 ± 0.8 fold (p = 0.0090) (Fig. 2D-E), while blood glucose and free fatty acids (FFA) levels were similar to those of mice treated with the control LF diet (Fig. 2F-G). Exposure to HF diet induces hepatosteatosis in mice and humans (Nam et al. 2015, Rosqvist et al. 2014). The lipophilic dye Nile Red, which becomes fluorescent in hydrophobic environments, has been used to monitor lipid content in liver and primary liver cells (Li et al. 2009, Martius et al. 2014). When livers of male mice exposed to HF diet were stained with Nile Red, the size of lipid droplets was increased by 14.2 ± 1.0 fold (p < 0.0001) and so was the fraction of liver tissue occupied by fat (by 4.5 ± 0.2 fold p < 0.0001) as compared to that of mice treated with LF diet (Fig. 2H). Thus, exposure to HF diet induces increased weight, hyperleptinemia, hyperinsulinemia, and hepatosteatosis in male mice. Female rodents exposed to HF diet have been reported to have less severe adverse metabolic consequences and hypothalamic injury than male mice (Hong et al. 2009, Clegg et al. 2011, Atamni et al. 2016, Dorfman et al. 2017, Qiu et al. 2018). A group of 8-week old female Sim1-Cre:Rosa-mEGFP mice were treated for 15 weeks with LF diet (starting weight = 17.8 ± 0.58 grams) and HF diet (starting weight =18.6 ± 0.90 grams). The body weight of female mice treated with HF diet was increased by 64.5 ± 28.0 % (p = < 0.0001) as compared to that of female mice exposed to LF diet. The female mice exposed to HF diet gained 86.5 ± 25.1% (p < 0.0001) of their initial weight over the 15 weeks of treatment (Fig. 2I-J). After 15 weeks of treatment with HF diet, male and female mice gained weight to a similar extent (p =0.52). While female mice treated with HF diet had increased caloric intake as compared to that of female mice treated with LF diet, differences reached statistical significance only at some time points (Fig. 2K). In female mice treated with HF diet, serum leptin was elevated as compared to that of female mice treated with the control LF diet by 5.9 ± 1.2 fold (p = 0.0003) while serum insulin level was unchanged (Fig. 2L-M). Female mice treated with HF diet had elevated levels of blood glucose as compared to that of female mice treated with LF diet (by 24.0 ± 6.0 %, p = 0.0023) and unchanged levels of FFA (Fig. 2N-O). In livers of female mice on HF diet, the size of lipid droplets was increased by 9.1 ±1.3 fold (p < 0.0001) as compared to that of mice on LF diet and so was the fraction of liver tissue occupied by fat (by 2.3 ± 0.2 fold, p = 0.0002) (Fig. 2P). In conclusion, it appears that exposure to HF diet induces increased caloric uptake, weight gain, hyperleptinemia, and hepatosteatosis both in male and female mice. On the HF diet, increased insulinemia appears in male mice and blood glucose is significantly increased in female mice.

Exposure to HF diet with elevated palm oil content induces loss of mEGFP abundance in the PVN of male and female Sim1-Cre:Rosa-mEGFP mice

To monitor whether exposure to HF diet affects Sim1 neurons, mEGFP fluorescence pixel density was measured within the PVN and AHC of male Sim1-Cre: Rosa-mEGFP mice. ROIs of the same areas were drawn to outline PVN and AHC in images derived from matching sections of Sim1-Cre: Rosa-mEGFP mice treated with the LF diet and HF diet (Fig. 3A). In the PVN and AHC of the male mice treated with HF diet, the mEGFP fluorescence intensity was decreased by 47.0 ± 10.3 % (p = 0.0006) and by 47.7 ± 11.7 %, (p = 0.0015) respectively, compared to that of mice treated with LF diet (Fig. 3A). Thus, in Sim1 neurons of the PVN, the HF diet induces loss of mEGFP expression at the cell body and processes. Conversely, in the same brain sections that include the PVN, mEGFP fluorescence intensity in Sim1 neurons of the LOT was unchanged (Fig. 3B). This was also the case in the MeA, where Sim1 neurons are also localized (Balthasar et al. 2005) (Fig. 3C). Thus, in male mice, exposure to the HF diet induces damage to Sim1 neurons localized specifically to the PVN.

In the PVN and AHC of the same female Sim1-Cre: Rosa-mEGFP mice treated with HF diet, mEGFP fluorescence intensity was reduced by 42.2 ± 11.3 % (p = 0.0025) and by 56.8 ± 12.0 % (p = 0.0010), respectively, compared to that of female mice treated with LF diet (Fig. 3D). Conversely, mEGFP fluorescence intensity in Sim1 neurons of LOT and MeA was not changed (Fig. 3E-F). Thus, in female mice, like in male mice, increased weight gain by exposure to HF diet selectively induces loss of mEGFP fluorescence intensity in Sim1 neurons localized to the PVN.

Exposure to HF diet induces apoptosis and neuronal loss in the PVN of male and female Sim1-Cre:Rosa-mEGFP mice without concomitant microgliosis

In male and female Sim1-Cre:Rosa-mEGFP mice treated with HF diet, loss of mEGFP expression in the Sim1 neurons of the PVN (Fig. 3) may result from neuronal injury. To test this possibility, matching sections from the male Sim1-Cre:Rosa-mEGFP mice treated with LF diet and HF diet were co-stained using the TUNEL assay, which detects apoptotic DNA fragmentation. In the PVN of male mice treated with HF diet, Tunel-positive cells were significantly increased by 8.0 ± 0.9 fold (p< 0.0001) as compared to male mice treated with LF diet (Fig. 4A). To determine whether increased DNA fragmentation also leads to neuronal loss in the PVN, sections from male Sim1-Cre:Rosa-mEGFP mice treated with LF and HF diets were stained with the neuron-specific nuclear protein NeuN (Mullen et al. 1992). Cells detected as positive for NeuN, in a ROI that included PVN visualized by the mEGFP, were significantly decreased by 54.7 ± 10.3 % (p = 0.0002) in male Sim1-Cre:Rosa-mEGFP mice treated with HF diet compared to that of mice treated with the LF diet (Fig. 4B). Brain injury induces changes in microglia cell morphology including retraction and thickening of cell processes (Davalos et al. 2005, Nimmerjahn et al. 2005, Svahn et al. 2013). Exposure to HF diet also induces neuronal injury and reactive microgliosis in the arcuate nucleus, and this process is thought to be initially protective, but over prolonged time may contribute to hypothalamic injury (Dorfman & Thaler 2015). Abundance and distribution of the microglia marker Iba1 is used to monitor hypothalamic microgliosis (Ito et al. 1998, De Souza et al. 2005, Moraes et al. 2009, Horvath et al. 2010, Thaler et al. 2012, Valdearcos et al. 2014, Valdearcos et al. 2015, Baufeld et al. 2016, Yi et al. 2017). However, in a ROI including the PVN visualized by the mEGFP in Sim1 neurons, the abundance of Iba1 did not increase in the PVN of male Sim1-Cre:Rosa-mEGFP mice exposed to HF diet (Fig. 4C) and individual microglia cells appeared to have similar morphology and process thickness compared to control (Fig. 4D). These experiments indicate that loss of mEGFP in Sim1 neurons in the PVN of male mice is paralleled by apoptosis and neuronal injury, but not by microgliosis.

In female Sim1-Cre:Rosa-mEGFP mice treated with HF diet, TUNEL-positive cells in the PVN were significantly increased by 23.4 ± 6.0 fold ( p = 0.0118) compared to female mice treated with LF diet (Fig. 4E). Cells positive for NeuN staining in a ROI including the PVN visualized by the mEGFP were decreased by 43.2 ± 0.1 % (p = 0.0034) (Fig. 4F). In the PVN of female Sim1-Cre:Rosa-mEGFP mice treated with HF diet, abundance of Iba1, microglia morphology and thickness of microglia processes did not change (Fig. 4G-H). Together, these experiments indicate that HF diet induces apoptosis and neuronal injury in the PVN of male and female mice without concomitant microgliosis.

Male, but not female, Sim1-Cre:Rosa-mEGFP mice exposed to HF diet have neuronal injury and microgliosis in the arcuate nucleus

In male mice, exposure to HF diet induces neuronal loss and reactive microgliosis in the arcuate nucleus and mediobasal hypothalamus (De Souza et al. 2005, Moraes et al. 2009, Thaler et al. 2012, Valdearcos et al. 2014, Dorfman & Thaler 2015, Valdearcos et al. 2017). On the other hand, here it is found that in the male Sim1-Cre:Rosa-mEGFP mice exposed to the palm oil-based HF diet, neuronal damage in the PVN is not paralleled by reactive microgliosis (Fig. 4). We therefore tested directly whether neuronal injury and concomitant microgliosis instead occurred in the arcuate nucleus of male Sim1-Cre:Rosa-mEGFP mice exposed to the diet with elevated fat content. In respect to neuronal injury, it has been found that in male rodents exposed to HF diet, POMC neurons in the arcuate nucleus are decreased and so are their synapses. Moreover, in male mice treated with HF diet, the expression of α-MSH in POMC neurons is also decreased because ability to process the POMC precursor to α-MSH is impaired (Levin 1999, Enriori et al. 2007, Horvath et al. 2010, Thaler et al. 2012, Schneeberger et al. 2013, Cakir et al. 2013, Mercer et al. 2014). To measure α-MSH abundance in male Sim1-Cre:Rosa-mEGFP mice, matching sections from a region of the brain that includes the arcuate nucleus and the ventromedial hypothalamus (Fig. 5A, upper panel) were immunostained with antibodies against the hormone (Fig. 5A, lower panels). In male mice treated with the HF diet, abundance of α-MSH was decreased by 60.0 ± 12.1 % (p = 0.0004) compared to mice treated with LF diet (Fig. 5A lower panels). POMC neurons have their cell bodies residing in the arcuate nucleus and their axons project to the PVN where α-MSH is released in the proximity of Sim1 neurons that co-express MC4R (Garfield et al. 2015, Krashes et al. 2016). Exposure to HF diet decreased, (by 56.3 ± 13.0 %, p = 0.0010), abundance of α-MSH in POMC fibers projecting to the PVN, which can be visualized by the Sim1 neurons expressing mEGFP (Fig. 5B). Moreover, in the arcuate nucleus of male Sim1-Cre:Rosa-mEGFP mice exposed to HF diet, the number of Tunel-positive cells was significantly increased by 3.5 ± 0.52 fold (p = 0.0028) (Fig. 5C), and the number of cells stained with the neuronal marker NeuN was instead decreased by 54.7 ± 16.0 % (p = 0.0048) (Fig. 5D). Thus, in the male Sim1-Cre:Rosa-mEGFP mice exposure to HF diet indeed induces cell apoptosis and neuronal loss in the arcuate nucleus. At this location, injury to POMC neurons was paralleled by increased Iba1 immunoreactivity by 2.3 ± 0.1 fold (p< 0.0001), with a concomitant increase of the fraction of arcuate nucleus area covered by microglia by 2.4 ± 0.2 fold (p< 0.0001) (Fig. 5E, upper panels). At high magnification, microglia cells had a different morphology than those of mice treated with LF diet, by having thicker processes (by 1.9 ± 0.1 fold, p < 0.0001) (Fig. 5E, lower panels). Thus, in the arcuate nucleus of male Sim1-Cre:Rosa-mEGFP mice, differently than in the PVN of the same mice, neuronal injury is paralleled by microgliosis. Together, the data indicate that, in male Sim1-Cre:Rosa-mEGFP mice, exposure to HF diet induces injury to two sets of neurons that regulate appetite, namely the POMC neurons in the arcuate nucleus and the Sim1 neurons in the PVN.

To determine whether exposure to HF diet also affects POMC neurons in the arcuate nucleus of Sim1-Cre:Rosa-mEGFP female mice, expression of α-MSH was measured by immunostaining matching sections derived from the region of the brain that includes arcuate nucleus and ventromedial hypothalamus. Exposure of female Sim1-Cre:Rosa-mEGFP mice to HF diet did not change abundance of α-MSH either in the arcuate nucleus or in the PVN, where POMC neuron fibers project and where mEGFP in Sim1 neurons is instead profoundly decreased (Fig. 5F and G). Consistent with this finding, exposure to HF diet also did not decrease NeuN positive cells in the arcuate nucleus of female Sim1-Cre:Rosa-mEGFP mice (Fig. 5H). Moreover, in the arcuate nucleus of female Sim1-Cre:Rosa-mEGFP mice exposed to HF diet, Iba1 pixel density, as well as the fraction of the nucleus area covered by microglia, and the thickness of microglia cell processes were not significantly increased (Fig. 5I). Thus, while in female mice exposed to HF diet there is injury to PVN neurons (Fig. 3 and 4), damage to POMC neurons of the arcuate nucleus is undetectable (Fig. 5).

Male mice treated with a high-palmitate diet have changed phosholipid fatty acid composition in liver, but not hypothalamus.

It has been shown by Thaler et. al. that neuronal loss in the arcuate nucleus of the hypothalamus of male mice takes place also in response to exposure to another palmitate-rich HF diet based on lard (D12492) (Thaler et al. 2012). Exposure to HF diet and obesity change the phospholipid composition of the liver in rodents and induce a rise of PC/PE ratio resulting in endoplasmic reticulum stress and hepatocyte injury (Nam et al. 2015, Sanyal & Pacana 2015) (Fu et al. 2011). Here we asked whether exposure to the D12492 HF diet has parallel effects to alter fatty acid composition of abundant phospholipids in murine liver and hypothalamus. To this end, male mice were treated for 9 weeks with the HF diet and LF diet. The mice treated with the HF diet did not have significant increase in body weight, but already increased levels of circulating leptin (by 80.44 ± 13.5-fold, p<0.0001) and insulin (by 2.09 ± 0.3-fold, p<0.05), as compared to mice treated with the LF diet. In the liver of mice treated with HF diet there was a minimal decrease of total PC abundance (by 7.8% ± 2.4, p<0.05) as compared to that of mice treated with LF diet (Fig. 6A). In the liver of mice treated with the LF diet the most abundant PC species were those containing a total of 34-and 36-carbon atoms in the acyl chains (33.9 % ± 1.1 and 32.9% ± 0.5 of total PC, respectively). When mice were treated with HF diet, abundance of PC species with 34-carbon atom acyl chains was significantly decreased (by 42.7% ± 2.7 p<0.0001) and abundance of longer PC species with 38-carbon atom acyl chains was significantly increased (by 43.2% ± 2.4, p<0.0001). These data indicate, that treatment with HF diet promotes, in the liver, formation of PC species bearing longer fatty acids. In the liver of mice fed LF diet, PC species with one, two and three double bonds were the most abundant (62.7% ± 0.67 of total PC species), and those PC species were decreased when mice were fed with the HF diet (by 23.8 % ± 1.6, p< 0.0001), with a concomitant increase of PC species carrying four or more double bonds (by 44.0% ± 2.9, p< 0.0001). These data indicate that administration of HF diet promotes in the liver formation of PC species with greater degree of unsaturation and that such effect appears to involve most abundant PC components. In the hypothalami of the same cohort of mice treated with the LF diet and HF diet abundance of total PC was the same (Fig. 6B). In the hypothalami of the mice treated with LF diet, the most abundant PC species were those containing acyl chains with a total length of 34- and 36-carbon atoms (26.1% ± 0.3 and 30.2% ± 0.4 of total PC, respectively) and containing a single double bond per molecule (44.0% ± 0.2 of total PC). Exposure to the HF diet did not change relative abundance of any PC species detected in the hypothalamus.

Figure 6. Male mice treated with a high-palmitate diet have changed phosholipid fatty acid composition in liver, but not hypothalamus.

A) Abundance of individual PC species in the liver of mice treated with LF diet (LFD) and palmitate-rich HF diet (HFD) is expressed as nmoles/mg protein. Abundance of total PC species in the liver of mice treated with HFD, calculated as the sum of abundance of each PC species (nmoles/mg protein), is expressed as percent of that of mice treated LFD. Liver PC species of mice treated with HFD and LFD, respectively, are grouped by the number of carbon atoms and their relative abundance (nmoles/mg protein) is expressed as percent of total liver PC (nmoles/mg protein) for each condition. Liver PC species of mice treated with HFD and LFD, respectively, are grouped by the number of double bonds of the two fatty acyl groups combined and their relative abundance (nmoles/mg protein) is expressed as percent of total liver PC (nmoles/mg protein) for each condition. B) Abundance of PC species in the hypothalamus of mice treated with LFD and HFD are expressed as in A. C) Abundance of PE species in the liver of mice treated with LFD diet and HFD are expressed as for PC in A. D) Abundance of PE species in the hypothalamus of mice treated with LFD diet and HFD are expressed as for PC in A. The ratio -total PC species / total PE species- in liver and hypothalamus of mice treated with HFD is derived from A-C and expressed as percent of that of mice treated with LFD. Values are mean ± SD. Data are derived from n = 4 mice for each group. *p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, Student’s t-test. F, Timeline of animal treatment for the mass spectrometry analysis shown in panels A-E. Mice were weaned when they were 3 weeks old and fed the chow diet for 3 weeks before starting LFD and HFD when they were 6 weeks old (n = 4 male mice on LF diet and n = 4 male mice on HF diet).

In the liver of mice treated with HF diet, abundance of total PE was profoundly decreased as compared to that of mice treated with LF diet (by 34.8% ± 2.4) (Fig. 6C), leading to an increased PC/PE ratio (by 41.2 ± 5.4, p<0.001) (Fig 6E). When mice were treated with HF diet, the relative abundance of PE species with 34- and 36-carbon atom acyl chains in liver were significantly decreased (by 32.7% ± 1.0, p<0.0001) and longer PE species with 38-carbon atom acyl chains were instead significantly increased (by ~21.1% ± 1.8, p<0.0001) (Fig. 6C). In the liver of mice fed with the HF diet, there was also decreased relative abundance of PE species with one (by 38.4% ± 1.8, p< 0.0001, two (by 46.7% ± 1.8, p<0.0001), and three double bonds (by 24.9% ± 3.1, p<0.0001), while PE with four double bonds were instead increased (by 24.2% ± 1.8, p<0.0001). These data indicate that HF diet induces in the liver of male mice the formation of PE species bearing longer fatty acids and a greater degree of unsaturation, similar to what we observed in the case of the PC species. In the hypothalamus of male mice treated with HF diet, abundance of all PE species as well as acyl chain length, degree of unsaturation and the PC/PE ratio remained unchanged (Fig. 6D). These data indicate that exposure to palmitate-rich HF diet induces major changes in PE abundance and PC and PE fatty acid composition in liver, but not brain.

DISCUSSION

Exposure for 15 weeks to a hypercaloric palm oil-based diet with elevated fat content induces increased weight gain, hyperleptinemia, and hepatosteatosis in male and female mice. In addition, male mice exposed to HF diet are hyperinsulinemic, and female mice exposed to HF diet have increased blood glucose. Thus, exposure to HF diet induces adverse metabolic consequences in mice of both sexes. In male and female mice exposed to HF diet, there is injury to Sim1 neurons of the PVN, without local microgliosis. Conversely, in the male, but not female mice, exposure to HF diet induces injury to POMC neurons in the arcuate nucleus of the hypothalamus and microgliosis. Thus, injury to Sim1 neurons in the PVN, rather than to POMC neurons in the arcuate nucleus, is a shared consequence of diet-induced hypothalamic damage in mice of both sexes.

In male mice, exposure to HF diet induces inflammation, gliosis, and neuronal loss in the arcuate nucleus, mediobasal hypothalamus and lateral hypothalamus, indicating these hypothalamic areas as main targets of brain injury (Flier 2004, De Souza et al. 2005, Velloso et al. 2009, Moraes et al. 2009, Thaler et al. 2012, Dorfman & Thaler 2015, Yi et al. 2017). Female rodents exposed to HF diet, while having less severe adverse metabolic consequences than male mice, nevertheless develop obesity. Sites of hypothalamic damage in female mice are currently under investigation (Hong et al. 2009, Clegg et al. 2011, Atamni et al. 2016, Dorfman et al. 2017, Qiu et al. 2018). In our study, we used a genetically encoded reporter, mEGFP, to monitor whether exposure to a hypercaloric HF diet with elevated palm oil content induces injury to Sim1 neurons of the hypothalamus. In this respect, it has been reported that pre and postnatal loss of Sim1, of Sim1 neurons, or of essential components of Sim1 neurons, causes obesity (Michaud et al. 2001, Holder et al. 2004, Xi et al. 2012, Tolson et al. 2014, Tolson et al. 2010, Kohno et al. 2014). On the other hand, it has also been reported that mice with postnatal ablation of Sim1 neurons by injection of diphtheria toxin in the PVN are resistant to HF diet (Xi et al. 2013). We find that, when exposed to an hypercaloric HF diet for 15 weeks, both adult male and female Sim1-Cre:Rosa-mEGFP mice have increased caloric intake and body weight. In the PVN of the male and female mice, abundance of a genetically encoded reporter, mEGFP, which is specifically expressed in Sim1 neurons, is decreased. This effect is paralleled by apoptosis, detected by the Tunel assay, and decreased population of mature PVN neurons, detected by the NeuN staining (Mullen et al. 1992, Chareyron et al. 2016). These results suggest that exposure to HF diet induces injury, rather than complete loss, of Sim1 neurons, which may in turn disrupt control of energy homeostasis in mice of both sexes. Sim1 neurons localized to the PVN and amygdala are necessary and sufficient to regulate food intake (Balthasar et al. 2005, Shah et al. 2014). Sim1 neurons, at multiple brain locations including the PVN, also regulate energy expenditure (Balthasar et al. 2005, Xi et al. 2012, Kohno et al. 2014, Cardinal et al. 2015, Sutton et al. 2014, Sutton et al. 2016, Timper & Bruning 2017). It is likely, but not directly tested here, that diet-dependent injury to Sim1 neurons of the PVN underlies obesity by altering, in addition to food intake, also energy expenditure. Exposure to HF diet appears to induce injury to multiple neuronal populations of the melanocortin pathway at different hypothalamic sites. These neuronal populations include POMC neurons (Huang et al. 2003, Cakir et al. 2013, Schneeberger et al. 2013, Williams et al. 2014) and steroidogenic factor 1-expressing neurons of the ventromedial hypothalamus (Klockener et al. 2011), as well the Sim1 neurons of the PVN, as reported here. A complete map of the type of neurons affected by exposure to HF diet in male and female mice has yet to be completed. It is possible that in mice exposed to HF diet, injury to other neuronal populations besides Sim1 neurons contribute to disrupt energy balance in mice of both sexes. Sim1 neurons include the MC4R neurons, and Sim1 deficiency appears to regulate MC4R expression (Balthasar et al. 2005, Kublaoui et al. 2006, Tolson et al. 2010, Shah et al. 2014). It has been reported that exposure to HF diet increases MC4R mRNA in the rodent hypothalamus (Enriori et al. 2007). However, in neuronal and hypothalamic cells exposed to elevated concentration of palmitate, MC4R protein abundance and function is reduced (Cragle & Baldini 2014). A similar analysis to visualize possible diet-dependent changes in MC4R protein abundance in the rodent brain is difficult because of a lack of antibodies that can detect hypothalamic MC4R (Krashes et al. 2016). Whether HF feeding induces injury to the population of Sim1 neurons that express MC4R remains yet to be resolved.

We find here that male and female mice treated with and without exposure to HF diet had similar levels of FFA in serum. The mice used in this study were fasted for 4 h prior to sacrifice and tissue harvesting, to avoid possible confounding effects by prolonged starvation and related stress to the hypothalamus. However, because in obesity increased fatty acids are released by adipocytes under starvation (Koo 2013), the 4h time interval without food may be too short to detect possible differences in serum FFA levels by HF diet exposure. On the other hand, exposure to HF diet induces hepatosteatosis in mice and humans (Nam et al. 2015, Rosqvist et al. 2014). In humans, most of the triacylglycerol that accumulate in the liver of obese individuals is contributed by circulating FFA (Donnelly et al. 2005). Thus, fat accumulation in liver, taking place both in male and female mice, is likely to be a consequence of altered whole body lipid homeostasis and dyslipidemia by exposure to HF diet. We find here that in male and female mice exposed to HF diet, accumulation of excess fat in liver parallels the increase in body weight. In a recent report, mass spectrometry analysis of male mice fasted for 4h finds that levels of triglycerides are increased in the hypothalamus of rodents treated with HF diet, without concomitant increase of circulating FFA (Borg et al. 2012). However, we could not detect increased lipid in the hypothalamus by using the same lipid stain approach as that used for the liver. Together, these observations indicate that exposure to HF diet disrupts whole body lipid homeostasis in male and female mice, which may then induce injury to both Sim1 and POMC neurons in male mice, and to Sim1 neurons in female mice.

It has been reported that consumption of a diet with elevated fat content for 8 weeks induces, in male rats, an approximately 7 fold increase in apoptosis (Moraes et al. 2009). Similar to this report, we find here that in the arcuate nucleus of male Sim1-Cre:Rosa-mEGFP mice exposed for 15 weeks to a palm-oil based HF diet, apoptosis in the arcuate nucleus is increased by 3.5 fold. Others have found that in male rodents exposure to HF diet decreases the number of arcuate nucleus POMC neurons by 25%−50% (Thaler et al. 2012, Yi et al. 2017). Similarly, we find here that in the arcuate nucleus of male mice treated with HF diet, neurons are decreased by 54.7%. In male rodents, exposure to HF diet also induces increased abundance of inflammatory cytokines and reactive microgliosis in the arcuate nucleus (Moraes et al. 2009, De Souza et al. 2005, Thaler et al. 2012, Dorfman & Thaler 2015, Valdearcos et al. 2014, Valdearcos et al. 2017). Moreover, in male rodents, exposure to HF diet induces impaired post-translational processing of POMC to generate α-MSH, loss of POMC neuron synapses, and impaired secretion of α-MSH (Levin 1999, Enriori et al. 2007, Horvath et al. 2010, Schneeberger et al. 2013, Cakir et al. 2013). On the other hand, another report using stereological analysis finds that treating male mice with HF diet does not affect total number of hypothalamic neurons identified by cresyl violet staining (Namavar et al. 2012). It has also been reported by using stereological analysis that exposure of male mice to HF diet does not change the total number of POMC neurons or microglia cells in the arcuate nucleus identified by immunostaining, but changes neuronal cell volume distribution at this location (Lemus et al. 2015). These variable outcomes indicate that exposure to HF diets may induce injury to the hypothalamus and to POMC neurons to a different extent, depending on the type and duration of the diet. To monitor HF diet-induced injury in the arcuate nucleus, we have measured, on consecutive sections of the arcuate nucleus of male Sim1-Cre:Rosa-mEGFP mice, cell apoptosis by Tunel staining; loss of mature neurons by NeuN immunostaining; and loss of hormone in POMC neurons by α-MSH immunostaining. We have extended this analysis to measure loss of α-MSH abundance in the PVN, where POMC neurons project their axons in the proximity of Sim1 neurons identified here by the intrinsic fluorescence of the mEGFP reporter. Data converge to indicate that in the arcuate nucleus of male mice exposed to HF diet profound loss of α-MSH and of mature neurons is paralleled by microgliosis. The extent of neuronal loss in the arcuate nucleus of male mice detected in this study (54.7%) is similar, but more profound than that of POMC neurons described by others (Thaler et al. 2012). It is possible that, in the arcuate nucleus, POMC neurons are less affected by exposure to HF diet than other neuronal populations detected here by the NeuN staining, or that the palm oil-based HF diet used here is most effective to cause hypothalamic injury. In this respect, the fatty acid composition of obesogenic diets is considered a determinant of obesity and metabolic alteration in rodents (Buettner et al. 2007, Buettner et al. 2006, Catta-Preta et al. 2012, Picklo & Murphy 2016, Sun et al. 2015, Fattore et al. 2014). Palm oil is the major food oil in world markets (Hayes & Pronczuk 2010). Therefore, for the HF diet employed here to monitor neuronal damage in arcuate nucleus and PVN, we used palm oil, rather than lard, as major fat ingredient. The palm oil based HF diet and another HF diet (D12492 from Research Diets, Inc.) where instead the excess fat is derived from lard (Cintra et al. 2012, Qiu et al. 2014, Klockener et al. 2011, Thaler et al. 2012), have, for the same amount of saturated fat (80.2g and 81.5g, respectively), similar amount of monounsaturated fat (79.1 and 91.5g, respectively). However the palm oil based HF diet used here has reduced levels of polyunsaturated fatty acids as compared to the lard based D12492 HF diet (41 g and 81.5 g, respectively). Partial substitution of the saturated fatty acid component of D12492 diet by flax seed oil with increased polyunsaturated fatty acids appears to revert hypothalamic inflammation (Cintra et al. 2012). Because the palm oil diet used here (Table I) has an increased ratio of saturated to polyunsaturated fatty acids than the D12492 lard based diet used by Thaler et al. (Thaler et al. 2012), it is possible that adverse effects on the arcuate nucleus by the palm oil HF diet are also more severe. However, in this work, we did not directly compare effects of different HF diets on hypothalamic injury.

It has been found that female rodents exposed to HF diet, while having decreased hypothalamic microglial activation and less severe adverse metabolic consequences than male mice, nevertheless eventually develop obesity with increased caloric intake (Clegg et al. 2011, Hong et al. 2009, Atamni et al. 2016, Dorfman et al. 2017, Qiu et al. 2018). This work presents the novel finding that, in female mice exposed to HF diet, profound injury to Sim1 neurons of the PVN is not paralleled, in the arcuate nucleus, by neuronal loss or by microgliosis. There is also no loss of α-MSH abundance in POMC neurons. Together, these observations identify injury to Sim1 neurons in the PVN, rather than to POMC neurons in the arcuate nucleus, as a common feature underlying defects in energy homeostasis induced by exposure to a hypercaloric HF diet in male and female mice. Exposure to HF diet induces whole body insulin resistance and, in the hypothalamus, impaired neuronal function, synaptogenesis, and damaged mitochondria (Kleinridders et al. 2014, Timper & Bruning 2017, Pimentel et al. 2014, Schneeberger et al. 2013, Jin & Diano 2018). In male mice, exposure to HF diet impairs insulin-dependent depolarization of POMC neurons (Qiu et al. 2018). Conversely, POMC neurons of female mice exposed to HF diet remain instead insulin responsive by a mechanism dependent on estrogen (Qiu et al. 2018). Here, we find that exposure to HF diet induces both hyperinsulinemia and hypothalamic injury to POMC neurons in male, but not female mice. Together, these data suggest that a more severe whole body insulin resistance induced by HF feeding in the male mice may contribute, in the arcuate nucleus, to injury to POMC neurons. Here we also find sex-dependent response of the microglia in male and female mice, with clearly detectable microgliosis in the arcuate nucleus of male, but not of female mice. Interestingly, it has been recently reported that there are sex-specific differences in hypothalamic microglial activation via the CX3CL1-CX3CR1 pathway and that such differences are not mediated by estrogen (Dorfman et al. 2017). Here we find concomitant injury to POMC neurons and microglia activation in the arcuate nucleus of male mice, but not female mice, exposed to HF diet. These observations open up the possibility, to be tested in the future, that insulin sensitivity of POMC neurons, regulated by estrogen levels, is a contributing factor to activation of microglia. Our work indicating that exposure to the HF diet induces microgliosis in a regional and sex specific manner is consistent with other reports (Milanski et al. 2009, Thaler et al. 2012, Chowen et al. 2013, Baufeld et al. 2016, Valdearcos et al. 2017). Together, these findings indicate the possibility that microglia populations with specific transcriptional identities exist in different areas of the brain and have different reactivity to injury (Grabert et al. 2016).

In the PVN, damage to Sim1 neurons may take place by direct and indirect effects induced by disrupted whole body lipid metabolism (Brookheart et al. 2009, Cnop et al. 2012). With respect to the effects by exposure to diets with elevated palmitate, free fatty acids and lipoproteins in the circulation can cross the blood/brain barrier to enter the brain (Robinson et al. 1992, Murphy 2017, Bazinet & Laye 2014, Schonfeld & Reiser 2013). It has been reported that, when exposed to HF diet with excess saturated fat, mice have increased levels of palmitate in the hypothalamus, an effect thought to induce inflammatory signaling by the microglia and to alter neuronal function (Valdearcos et al. 2014, Timper & Bruning 2017). Exposing cultured neurons and immortalized hypothalamic neurons to increased palmitate can also induce direct cell injury and impairs MC4R signaling (Cragle & Baldini 2014, Cooney et al. 2017, McFadden et al. 2014, Campana et al. 2018). In mice, exposure to HF diet with elevated palmitate has also been shown to increase abundance of diacylglycerol and ceramide in the hypothalamus (Borg et al. 2012, Lee et al. 2018, Gonzalez-Garcia et al. 2017). Further, increased diacylglycerol and downstream activation of protein kinase C-theta promotes hypothalamic insulin and leptin resistance (Benoit et al. 2009, Timper & Bruning 2017). Increased ceramide induces neuronal ER stress by a pathway suppressed by central estradiol (Contreras et al. 2014, Gonzalez-Garcia et al. 2017). On the other hand, our data indicate that, while exposure to HF diet induces major changes in the fatty acid composition of PC and PE and a profound loss of PE species in liver, thereby leading to increased PC/PE ratio, as observed by others (Nam et al. 2015, Sanyal & Pacana 2015, Fu et al. 2011, Engin 2017), virtually none of these changes are detectable in the hypothalamus. In the liver of obese mice, increased PC/PE ratio induces endoplasmic stress and injury (Fu et al. 2011). Conversely, the absence of changes in hypothalamic phospholipid fatty acid composition observed here in mice exposed to HF diet suggests that injury to Sim1 neuron may instead originate from aberrant signaling by systemic lipid dysmetabolism and inflammation (Fu et al. 2012, Timper & Bruning 2017). With this respect, exposure to HF diet triggers, in hypothalamic neurons, increased signaling of the IkB kinase-β/ nuclear factor kappa-light-chain-enhancer of activated B cells pathway and endoplasmic reticulum stress (Zhang et al. 2008, Yang & Hotamisligil 2008, Timper & Bruning 2017). Another possibility is that, under HF feeding, increased lipids from the circulation are catabolized, rather than incorporated into phospholipids, and contribute to Sim1 neuron injury. With this respect, it has been reported that β-oxidation of fatty acids takes place in neurons, limits fatty acid incorporation into brain lipids, and is a main source of superoxide and oxidative stress (Ebert et al. 2003, Freed et al. 1994, Schonfeld et al. 2011, Schonfeld & Reiser 2013). In summary, our data suggest that, in mice exposed to HF diet, injury to Sim1 neurons in the PVN occurs by indirect mechanisms originated by lipid dysmetabolism and inflammation taking place in the liver and other peripheral tissues, rather than by direct lipid deposition or local microglia activation.

In conclusion, in male and female mice exposed to HF diet, increased weight gain, hepatosteatosis and hyperleptinemia are paralleled by PVN injury without concomitant microgliosis. Conversely, injury to pro-opiomelanocortin neurons in arcuate nucleus is specific to male mice. The data indicate that Sim1 neuronal damage in the PVN is a major common feature of disrupted energy balance in mice of both sexes.

Abbreviations used:

AgRP

Agouti-Related Peptide

AHC

Anterior Hypothalamic Central Area

α-MSH

α-Melanocyte-Stimulating Hormone

ARC

Arcuate nucleus

BCA

Bicinchoninic Acid

BSA

Bovine Serum Albumin

BHLH

Basic Helix-Loop Helix

CAS

Chemical Abstracts Service

DABCO 1

4-diazabicyclo[2.2.2]octane

DAPI 4

6-diamidino-2-phenylindole

ELISA

Enzyme-Linked Immunosorbent Assay

FFA

Free Fatty Acids

HF

high fat

HFD

HF diet

IACUC

Institutional Animal Care and Use Committee

Iba1

Ionized calcium binding adaptor molecule 1

i-PrOH

2-propanol

LF

low fat

LFD

LF diet

LOT

Lateral Olfactory Tract

MC4R

Melanocortin-4 receptor

MeA

Medial Amygdala

mEGFP

membrane-bound Enhanced Green Fluorescent Protein

MIP

Maximum Intensity Projection

MTBE

methyl tert-butyl ether

MRM

multiple reaction monitoring

NPY

Neuropeptide Y

OCT

Optimal Cutting Temperature

PBS

Phosphate Buffered Saline

PC

phosphocholine

PE

phosphoethanolamine

POMC

Proopiomelanocortin

PVN

paraventricular nucleus

ROI

Region Of Interest

RRID

Research Resource Identifiers

Sim1

Single Minded 1

TdT

Terminal Deoxynucleotidyl Transferase

TUNEL

Terminal deoxynucleotidyl transferase dUTP Nick End Labeling