1 INDUCED BROWN ADIPOSE TISSUE DYSFUNCTION AND ...

2 downloads 0 Views 2MB Size Report
Nov 16, 2015 - 9Center for Regenerative Therapies Dresden, Dresden Germany. 17 ... adipocyte HIF2α is protective against maladaptation to obesity and ...
MCB Accepted Manuscript Posted Online 16 November 2015 Mol. Cell. Biol. doi:10.1128/MCB.00430-15 Copyright © 2015, American Society for Microbiology. All Rights Reserved.

1

ADIPOCYTE-SPECIFIC

HIF2α

DEFICIENCY

EXACERBATES

OBESITY-

2

INDUCED BROWN ADIPOSE TISSUE DYSFUNCTION AND METABOLIC

3

DYSREGULATION

4

Running title: Garcia-Martin, Adipocyte HIF2α regulates WAT and BAT angiogenesis.

5 6

Rubén García-Martín1,#, Vasileia I. Alexaki1, Nan Qin2, María F. Rubín de Celis3, Matina

7

Economopoulou4, Athanasios Ziogas1, Bettina Gercken1, Klara Kotlabova1, Julia Phieler1,

8

Monika Ehrhart-Bornstein3, Stefan R. Bornstein3, Graeme Eisenhofer2, Georg Breier5,

9

Matthias Blüher8, Jochen Hampe6, Ali El-Armouche7, Antonios Chatzigeorgiou1,2,10,

10

Kyoung-Jin Chung1 and Triantafyllos Chavakis1,2,9,10,#

11 12

1

Department of Clinical Pathobiochemistry; 2Institute of Clinical Chemistry and Laboratory

13

Medicine; 3Department of Medicine III; 4Department of Ophthalmology; 5Department of

14

Psychiatry; 6Department of Medicine I; 7Department of Phamacology and Toxicology;

15

1-7

16

8

Department of Endocrinology and Nephrology, University of Leipzig, Germany

17

9

Center for Regenerative Therapies Dresden, Dresden Germany

18

10

Medical Faculty, Technische Universität Dresden, Dresden, Germany

Paul-Langerhans Institute Dresden, Dresden Germany

19 20

#Correspondence:

21

[email protected]; [email protected]

22

Word count for Material and Methods section: 1756

23

Word count for Introduction, Results and Discussion: 4616

24 1

25

ABSTRACT

26

Angiogenesis is a central regulator for white (WAT) and brown adipose tissue (BAT)

27

adaptation in the course of obesity. Here we show that deletion of hypoxia-inducible factor

28

2α (HIF2α) in adipocytes (by using Fabp4-Cre transgenic mice), but not in myeloid or

29

endothelial cells, negatively impacted on WAT angiogenesis and promoted WAT

30

inflammation, WAT dysfunction, hepatosteatosis and systemic insulin resistance in obesity.

31

Importantly, adipocyte HIF2α regulated vascular endothelial growth factor (VEGF)

32

expression and angiogenesis of the obese BAT as well as its thermogenic function.

33

Consistently, obese adipocyte-specific HIF2α-deficient mice displayed BAT dysregulation,

34

associated with reduced levels of uncoupling protein-1 (UCP1) and a dysfunctional

35

thermogenic response to cold exposure. VEGF administration reversed WAT and BAT

36

inflammation and BAT dysfunction in adipocyte HIF2α-deficient mice. Together,

37

adipocyte HIF2α is protective against maladaptation to obesity and metabolic dysregulation

38

by promoting angiogenesis in both WAT and BAT and by counteracting obesity-mediated

39

BAT dysfunction.

40 41 42

2

43

INTRODUCTION

44

The hypoxia response is mediated by the heterodimeric hypoxia-inducible factors (HIFs)

45

comprising an α-subunit (HIF1α or HIF2α) that is regulated by oxygen and an oxygen-

46

insensitive β-subunit. Hypoxia prevents hydroxylation of HIFα subunits by prolyl

47

hydroxylases (PHDs), thereby leading to inhibition of degradation of the HIFα subunits,

48

their stabilization and the subsequent HIF-dependent upregulation of genes essential for

49

cellular adaptation to and cell survival under hypoxic conditions (reviewed in (1)).

50 51

In obesity, excessive lipid storage and adipocyte hypertrophy are thought to result in

52

hypoxia in white adipose tissue (WAT) and brown adipose tissue (BAT) (2-6), although

53

there is some controversy concerning oxygen levels in obese AT (7). The presence of

54

hypoxia in WAT may be associated with increased inflammation (2, 4, 5); hypoxia also

55

triggers an angiogenic response, including upregulation of the major angiogenic factor,

56

vascular endothelial growth factor (VEGF) in adipocytes (4). The angiogenic response is

57

crucial for the adaptation of the WAT in obesity. Previous studies have demonstrated that

58

defective angiogenesis in obese WAT may promote insulin resistance, inflammation and

59

adipocyte apoptosis (3, 8, 9). On the other hand, mice with adipocyte overexpression of

60

VEGF family member A (VEGF-A) are protected against the adverse effects of high-fat

61

diet (HFD) (8, 10, 11). In obese WAT, enhanced expression of both HIF1α and HIF2α is

62

observed (2, 6, 12). Previous studies have functionally implicated HIF1α (6, 12-14) and

63

different PHDs (15, 16) in the process of obesity. In contrast, less is known about the role

64

of HIF2α. HIF2α heterozygous-null mice were recently shown to have reduced insulin

65

sensitivity and enhanced AT inflammation upon HFD (17) and HIF2α was found to

3

66

regulate lipid metabolism in hepatocytes (18-20), whereas less data exist on the specific

67

role of HIF2α in adipocytes (6). This is particularly important given several findings

68

demonstrating that regulation of VEGF-A and angiogenesis in WAT is not dependent on

69

HIF1α. Whereas adipocytes lacking HIF1β (the common and obligate partner for HIF1α

70

and HIF2α) showed reduced VEGF expression (21), transgenic mice overexpressing

71

HIF1α in adipocytes did not show any upregulation in VEGF-A expression or other

72

proangiogenic factors (12). In addition, mice lacking adipocyte HIF1α showed no vascular

73

alterations, as compared to HIF1α-proficient mice (6). These observations point to a

74

potential role of HIF2α in the WAT for the adaptive response to obesity that remains to be

75

established.

76 77

BAT is also a highly vascularized tissue; the grade of vascularization determines its ability

78

for lipid consumption and its thermogenic function (3, 22). Increased BAT activity results

79

in improved insulin sensitivity and glucose homeostasis (reviewed in (23)). Norepinephrine

80

derived from sympathetic nerves is a central player in inducing expression of both VEGF

81

and the major thermogenic factor, uncoupling protein-1 (UCP1) (24). Catecholamine-

82

mediated induction of UCP1 requires a signaling cascade involving cAMP, protein kinase

83

A and PGC1α (reviewed in (23)). VEGF resulting from beta adrenergic receptor

84

stimulation promotes BAT angiogenesis and functionality (3, 8, 10, 11, 25-27).

85

Interestingly, UCP1-deficient mice display angiogenesis despite the absence of hypoxia in

86

their BAT (25, 28, 29); thus, hypoxia in BAT is not an absolute prerequisite for stimulation

87

of angiogenesis. Hypoxia may collaborate with norepinephrine in upregulating VEGF

88

expression in brown adipocytes (24), whereas activation of HIF1α in these cells may occur

4

89

even without hypoxia (28). However, HIF1α does not regulate expression of VEGF (3) or

90

UCP1 (6) in the BAT. On the other hand, although HIF2α expression is induced by cold

91

exposure (25), its potential role in the adaptive response of the BAT to obesity and cold

92

exposure, has not been addressed thus far.

93 94

To address the aforementioned questions pertinent to the role of HIF2α in both WAT and

95

BAT, we generated mice with adipocyte-specific HIF2α deletion. We found that lack of

96

HIF2α in adipocytes resulted in systemic insulin resistance associated with reduced

97

vascularization and a pro-inflammatory phenotype in both WAT and BAT in the course of

98

obesity. In contrast, myeloid or endothelial HIF2α did not affect obesity-related metabolic

99

dysregulation. In addition to reduced angiogenesis in the BAT, adipocyte HIF2α deficiency

100

was associated with reduced expression of the major thermogenic factor UCP1 in the obese

101

BAT. Treatment with VEGF reversed WAT and BAT inflammation and BAT dysfunction

102

in obese mice lacking adipocyte HIF2α, suggesting that the metabolic dysregulation

103

observed in adipocyte HIF2α deficiency, was, at least in part, mediated by diminished

104

VEGF production. Thus, adipocyte HIF2α was identified as a factor contributing to the

105

metabolic adaptation to diet-induced obesity in both WAT and BAT.

106 107 108

MATERIALS AND METHODS

109

Mice

110

Mice carrying a floxed HIF2α (Epas1) allele (Jackson Laboratories, Bar Harbor, ME)

111

(HIF2αfl/fl) were bred with mice carrying Cre recombinase under control of Fabp4 promoter

5

112

(Jackson Laboratories) to generate adipocyte-specific HIF2α KO mice (AdHIF2KO)

113

(Fabp4-Cre+ Epas1fl/fl). Similarly, Epas1fl/fl mice were crossed with LysM-Cre mice (30)

114

(Jackson Laboratories) to generate myeloid-specific HIF2α KO mice (MyeHIF2KO)

115

(LysM-Cre+ Epas1fl/fl). Fabp4-Cre- Epas1fl/fl and LysM-Cre- Epas1fl/fl littermates were used

116

as controls. For the study of endothelial HIF2α, we engaged a tamoxifen-mediated

117

inducible deletion by using CreERT, whose expression is driven by stem-cell leukemia

118

promoter (Scl)-5’, which is specifically expressed in endothelial cells (31, 32). Scl-

119

CreERT+ Epas1fl/fl (EndHIF2KO) and Scl-CreERT- Epas1fl/fl littermate control mice

120

received i.p. tamoxifen (Sigma-Aldrich, Munich, Germany) (2 mg/mouse/day) at the age of

121

8 weeks. Eight-ten week-old male mice were fed a normal diet (ND) or high-fat diet (HFD)

122

with 10% Kcal from fat or 60% Kcal from fat, respectively (Research Diets, New

123

Brunswick, NJ) and feedings were conducted for up to 24 weeks. Animal experiments were

124

approved by the Landesdirektion Sachsen, Germany.

125 126

In vivo metabolic analyses

127

For glucose tolerance test, mice were fasted overnight before intraperitoneal injection of D-

128

(+)-glucose (Sigma-Aldrich, Munich, Germany) (1g/Kg body weight). At the indicated

129

times, blood was collected via tail vein for measuring glucose levels with an Accu-Chek

130

glucose meter (Roche, Manheim, Germany). For insulin tolerance test, mice were fasted 6

131

hours before intraperitoneal injection of insulin (Lilly, Bad Homburg, Germany) (1 U/Kg

132

body weight) and blood glucose levels were measured at the indicated times. For glucose-

133

stimulated insulin secretion, mice were fasted overnight before intraperitoneal injection of

134

D-(+)-glucose (Sigma-Aldrich, Munich, Germany) (1g/Kg body weight). At the indicated

6

135

times, blood was collected via tail vein for measuring plasma insulin with an ELISA kit

136

(Chrystal Chem, Cologne, Germany). Blood triglycerides and cholesterol were determined

137

using an Accutrend Plus system (Roche). For analysis of fasted plasma samples, mice were

138

fasted overnight (16-18 hours), blood was collected via tail vein and plasma leptin,

139

adiponectin, insulin and FGF21 were determined using ELISA kits (R&D system,

140

Wiesbaden-Nordenstadt, Germany; Chrystal Chem) following manufacturer’s instructions.

141

For the lipid tolerance test, mice were fasted overnight before oral gavage of olive oil

142

(Sigma-Aldrich) (5 μL/g of body weight) and blood triglycerides were determined as

143

described above. For the determination of free fatty acids, serum from 16-18 h fasted mice

144

was collected and a free fatty acid fluorometric assay kit was used (Cayman Chemical, Ann

145

Arbor, MI). For in vivo insulin signalling pathway analysis, mice were fasted for 6 hours

146

before intraperitoneal injection of insulin (Lilly) (5 U/Kg); 8 min thereafter mice were

147

euthanized and tissues were harvested and snap frozen for further analysis. Lean and fat

148

mass were measured in mice fed a HFD for 22 weeks by using computed tomography (CT)

149

(33) (Skyscan 1178, Bruker, Rheinstetten, Germany). For cold exposure experiments, obese

150

mice fed for at least 19 weeks with HFD were placed at 4ºC overnight. Body temperature

151

was measured at the indicated times with a thermometer (Bioseb, Vitrolles, France).

152

Thereafter, mice were sacrificed and tissues isolated and processed at 4ºC.

153 154

VEGF administration

155

Controlled administration of VEGF was achieved via subcutaneous implantation of mini-

156

osmotic pumps (Alzet, CA, USA). Pumps were filled with either recombinant murine

157

VEGF (Peprotech, Hamburg, Germany) diluted in PBS with 0.1% BSA or with PBS with

158

0.1% BSA as control. Delivery rate was set at 75 ng/hour. 5-week old mice were fed a HFD 7

159

for 5 weeks before pump implantation; after pump implantation, mice were fed for

160

additional 3 weeks prior to euthanasia and further analysis.

161 162

Protein detection

163

Adipose tissues (AT) were excised after euthanasia and proteins isolated as described

164

elsewhere (34). Briefly, AT were homogenized and digested in RIPA lysis buffer [1%

165

TritonX-100; 0.5% sodium deoxycholate; 0.1% SDS; 50 mM Tris-HCL, pH 7.5; 150 mM

166

NaCL; Mini Protease Inhibitor and Phosphatase Inhibitor Cocktail Tablet (Roche)],

167

incubated on ice for 20 minutes and centrifuged to remove cellular debris and fat. Protein

168

concentration was determined using BCA Protein Assay Kit (Thermo Scientific, Schwerte,

169

Germany). Antibodies against UCP1 (Abcam, Cambridge, U.K.), phospho-Ser473 Akt,

170

total Akt (Cell Signaling/New England Biolabs, Frankfurt am Main, Germany) and tubulin

171

(Sigma-Aldrich) were used for immunoblotting. For blot quantification, densitometry was

172

performed with ImageJ software (National Institute of Health, Bethesda, MD, USA);

173

tubulin was used for UCP1 normalization, whereas total Akt was used for phospho-Ser473

174

Akt normalization.

175 176

VEGF-A was determined in WAT and BAT lysates using an ELISA kit (R&D system).

177 178

Gene expression

179

Total RNA from tissues or cells was isolated using TRIzol (Invitrogen, Darmstadt,

180

Germany) following manufacturer’s instructions. After purification with DNase I treatment

181

(Thermo Scientific), 1 µg of RNA was reverse-transcribed using an iScript cDNA

182

Synthesis kit (Bio-Rad, Munich, Germany), and real time PCR was performed with SsoFast 8

183

EvaGreen Supermix (Bio-Rad) using a Bio-Rad CFX384 Touch Real Time PCR Detection

184

System (Bio-Rad). Calculation was based on ΔΔCt method (35) and normalization was

185

performed to 18s RNA.

186 187

Ex vivo lipolysis assay in adipose tissue explants

188

Subcutaneous and gonadal fat depots (scWAT and gonWAT, respectively) were surgically

189

removed from 24-week HFD-fed control and AdHIF2KO mice and washed with ice-cold

190

PBS. A piece of 100 mg was incubated for 2 hours at 37ºC in 250 μL DMEM containing

191

2% fatty acid-free BSA (Sigma-Aldrich) and in the presence or absence of 10 μM

192

isoprenaline (Sigma-Aldrich). Fatty acids released to the medium were quantified using a

193

free fatty acid fluorometric assay kit (Cayman Chemical).

194 195

Cell culture

196

For the isolation of bone-marrow derived macrophages (BMDM), we followed our

197

previously published protocol (36).

198 199

Histology

200

Fresh tissues were excised, fixed in 4% paraformaldehyde, and paraffin embedded.

201

Sections were stained with H&E. Histological scoring for liver NAFLD/NASH was read

202

blinded to the experimental design using H&E staining. The degree of steatosis, ballooning

203

and lobular inflammation was evaluated according to previously published criteria

204

following the NASH-CRN Committee scoring system (37, 38). The NAFLD Activity Score

205

(NAS) consists of the sum of steatosis, ballooning and lobular inflammation. A value of

9

206

NAS>5 correlates with the presence of NASH (38). For immunohistochemistry, sections

207

were de-paraffinized and antigen retrieval was done by incubation with hot citrate buffer.

208

Inhibition of intrinsic peroxidase activity was performed with 0.5 % H2O2 treatment. In

209

addition, F4/80 staining required proteinase K (Sigma-Aldrich) treatment at 37°C, whereas

210

for UCP1 staining, sections were also incubated with pronase (Sigma-Aldrich) at 37°C.

211

Sections were permeabilized with 0.4% TritonX100 and blocked with serum from ABC kit

212

(Vector Laboratories, Peterborough, U.K.). Sections were then incubated overnight with

213

antibodies against UCP1 (Abcam), or against F4/80 (Novus Biologicals, Herford,

214

Germany). The avidin–biotin complex was detected with an AEC Peroxidase Substrate Kit

215

(Vector Laboratories). F4/80 quantification was performed by counting the number of

216

F4/80 positive cells per adipocyte per field in at least 6 random low-magnification fields

217

per sample. TUNEL staining was performed using a Dead End Colorimetric TUNEL kit

218

following manufacturer’s instructions (Promega, Mannheim, Germany). TUNEL

219

quantification was performed by counting the number of TUNEL positive cells per

220

adipocyte per field in at least 6 random low-magnification fields per sample. Pictures were

221

visualized by a computerized microscope (Zeiss, Oberkochen, Germany) and analysed with

222

AxioVision Rel 4.8 software (Zeiss). For Oil Red O staining, 10 μm cryosections were

223

prepared. Slides were fixed in 4% paraformaldehyde for 1 hour, rinsed in dH2O, and stained

224

for 15 min in Oil Red O in 60% isopropanol. They were rinsed with 60% isopropanol and

225

counterstained with Mayer’s hematoxylin. Quantification was performed as described (39).

226

Fibrosis staining was performed by using a Masson trichrome staining kit (Sigma-Aldrich)

227

following manufacturer's instructions. For hypoxia staining, mice were intraperitoneally

228

injected with 60 mg/kg of CCl-103F (Hypoxyprobe F6, Hypoxyprobe, Burlington, MA)

229

and sacrificed 90 min after the injection. Tissues were fixed and processed as described 10

230

above using Hypoxyprobe F6 kit (Hypoxyprobe) following manufacturer´s intructions;

231

quantification was performed by determining the percentage of the area staining positive

232

for hypoxia in 5 random low-magnification fields per sample. AT whole mount staining

233

was performed following a previously described method (40) with some modifications.

234

Briefly, WAT and BAT were fixed with paraformaldehyde and cut in small pieces of 2 mm

235

x 2 mm. After blocking and permeabilization with BSA (1%) and Triton-X100 (0.5%),

236

mounts were stained with FITC-conjugated isolectin B4 (Sigma-Aldrich) at 4ºC. After

237

several washings with PBST, samples were visualized using a confocal microscope (Leica

238

TCS SP5, Leica, Wetzlar, Germany). Z-stacks of 5 µm depth were analysed and

239

quantification of fluorescence intensity was performed using LAS software (Leica).

240 241

Measurement of liver triglyceride content

242

Triglyceride content quantification in the liver was done using a commercially available kit

243

(Abcam). Briefly, 100 mg of liver tissue was homogenized in 1 ml of 5% Triton-X100. The

244

samples were then heated to 95ºC and cooled to room temperature twice. Thereafter,

245

samples were centrifuged, and triglyceride content in the supernatant was quantified using

246

enzymatic determination.

247 248

Flow cytometry

249

After sacrificing mice, scWAT and gonWAT were excised and the lymph nodes, immersed

250

in the fat depot, were removed. AT was then digested using collagenase type I (2 mg/ml per

251

mg of tissue; Life Technologies) for 60 min at 37ºC. The suspension was resuspended in

252

DMEM containing 0.5% fatty acid–poor BSA (Sigma-Aldrich) and centrifuged to separate

253

the floating adipocyte fraction from the pelleted stromal vascular fraction (SVF). For FACS 11

254

analysis, the following antibodies were used: Fc receptor–blocking Ab 2.4G2, CD45–Alexa

255

Fluor 488 (Biolegend, Fell, Germany); CD31-APC (eBioscience, Frankfurt, Germany).

256

FACS was carried out on a FACSCanto II (BD, Heidelberg, Germany) and analyzed with

257

FACSDiva Version 6.1.3 software.

258

Sorting of lung endothelial cells from endothelial-specific HIF2α KO mice was

259

performed as previously described (32). The solution was filtered and stained for CD31

260

(eBioscience) and sorting for CD31-positive and CD31-negative cells was performed using

261

a sorter FACS Aria II (BD).

262 263

Metabolic cage analysis

264

Metabolic cage analysis was done with mice that were fed with HFD for 6 weeks. Mice

265

were individually housed in metabolic cages (PhenoMaster, TSE Systems, Bad Homburg,

266

Germany) with free access to water and food, maintaining a 12h:12h light-dark cycle. A

267

period of at least 16 hours acclimatization in the metabolic cages preceded initiation of the

268

experiment and data collection. Volume of oxygen consumption (VO2) and carbon dioxide

269

production (VCO2) were determined every 20 minutes. Respiratory exchange ratio (RER)

270

was calculated as VCO2/VO2. Energy expenditure (EE) was calculated as 3.941xVO2 +

271

1.106xVCO2 (41). Food intake was also monitored. Data were normalized with respect to

272

body weight using ANCOVA analysis.

273 274

Statistical analyses

275

Data are expressed as mean ± s.e.m. and statistically analyzed by Student’s t-test, or Mann-

276

Whitney U as appropriate. Body temperature during cold exposure experiment was

277

analyzed by ANOVA. Significance was set at P < 0.05. 12

278

RESULTS

279

Adipocyte HIF2α deficiency promotes HFD-induced metabolic dysregulation.

280

Mice with adipocyte-specific deletion of HIF2α were generated by crossing mice with a

281

floxed Epas1 (HIF2α) allele with mice expressing Cre recombinase under the control of the

282

Fabp4 promoter. This resulted in Fabp4Cre+ Epas1fl/fl mice, in which HIF2α was deleted in

283

white and brown adipocytes (AdHIF2KO) (Fig. 1A) (42), while Fabp4Cre- Epas1fl/fl mice

284

were used as littermate HIF2α-proficient mice (referred to as control mice). Efficient

285

deletion of HIF2α was observed in subcutaneous (sc) and visceral gonadal (gon) WAT, and

286

in BAT, whereas HIF2α expression was unaffected in non-adipose tissues/cells such as the

287

liver, skeletal muscle, hypothalamus and bone marrow-derived macrophages (BMDM) (Fig.

288

1A-1C). HIF1α expression was not affected in any of these tissues by adipocyte-specific

289

HIF2α deletion (Fig. 1A-1C).

290 291

AdHIF2KO and control mice were fed a HFD or ND. Mice lacking HIF2α in adipocytes

292

showed a significant increase in HFD-induced body weight gain (Fig. 2A), accompanied by

293

increased weight of the scWAT, liver and BAT upon 16 and 24 weeks of HFD (Fig. 2B,

294

data from mice fed a HFD for 24 weeks; data from mice fed a HFD for 16 weeks not

295

shown). In contrast, the weight of gonWAT did not differ between HFD-fed AdHIF2KO

296

mice and HFD-fed control mice. The increase in the mass of scWAT, liver and BAT was

297

not simply due to enhanced body weight, since a significant increase in the weight of these

298

tissues was also found when weights were plotted as percentage of total body weight (Fig.

299

2B, data from mice fed a HFD for 24 weeks; data from mice fed a HFD for 16 weeks not

300

shown). In keeping with these data, fat body mass determined by computed tomography

13

301

(CT) was elevated in AdHIF2KO mice, as compared to control mice, whereas no difference

302

was found in lean body mass between both genotypes (Fig. 2C).

303 304

AdHIF2KO mice on a HFD showed increased glucose intolerance and insulin resistance

305

compared to control mice (Fig. 2D-2E). Analysis of glucose-stimulated insulin secretion

306

(GSIS) demonstrated elevated insulin secretion upon glucose administration in AdHIF2KO

307

mice (Fig. 2F), reflecting the impaired insulin sensitivity. Furthermore, insulin signaling

308

was assessed, and BAT of AdHIF2KO mice exhibited reduced Akt phosphorylation in

309

response to insulin administration, as compared to control mice (Fig. 2G). In contrast,

310

insulin-induced Akt phosphorylation was not different in sc and gon WAT, liver and

311

muscle due to adipocyte HIF2α deficiency (data not shown). Fasting plasma insulin,

312

glucose, leptin and cholesterol levels were enhanced in obese AdHIF2KO mice, while

313

adiponectin levels were significantly lower in obese AdHIF2KO mice compared to control

314

mice (Fig. 2H-2L). In contrast, there was no significant difference in blood triglycerides or

315

FGF-21 (43) in AdHIF2KO mice (Fig. 2M and 2N). When fed a ND, AdHIF2KO mice

316

displayed no significant changes in body weight, fat mass and insulin sensitivity, as

317

compared to control mice (Fig. 3A-3C). These findings indicate that adipocyte HIF2α

318

contributes to the adaptive response of WAT to obesity and protects against HFD-induced

319

metabolic dysregulation.

320 321

We then analyzed the alterations in WAT induced by adipocyte HIF2α-deficiency. We

322

found that the mRNA expression of the adipocyte lipase Atgl was significantly reduced in

323

the gonWAT of obese AdHIF2KO mice; moreover, the expression of the lipase MgII was

14

324

lower (although not significantly), as compared to obese control mice (Fig. 4A). In the

325

gonWAT of AdHIF2KO mice, we also found decreased expression of genes involved in

326

peroxisomal (Acox, Crot) and mitochondrial lipid oxidation (Cpt1, Acsl1), of genes

327

involved in β-oxidation of long fatty acids (Vlcad, Lcad) and reduced expression of the

328

lipolysis regulator (Ppar-α) owing to adipocyte HIF2α-deficiency (Fig. 4A). In addition,

329

scWAT showed reduced expression of genes involved in lipolysis (MgII) and peroxisomal

330

lipid oxidation (Acox) (data not shown). Explants of scWAT from obese AdHIF2KO mice

331

showed reduced basal lipolysis, whereas gonWAT explants from obese AdHIF2KO mice

332

displayed diminished isoprenaline-stimulated lipolysis (Figure 4B), as compared to

333

respective explants from obese control mice. Furthermore, we found elevated RER in obese

334

AdHIF2KO mice indicating reduced lipid consumption, as compared to obese control mice,

335

while the food intake, oxygen consumption and energy expenditure (EE) were not affected

336

by adipocyte HIF2α−deficiency (Fig. 4C-4F). In addition, AdHIF2KO mice showed

337

elevated triglyceride levels upon an oral lipid tolerance test (Fig. 4G), suggesting defective

338

lipid metabolism due to adipocyte HIF2α deficiency.

339 340

Adipocyte HIF2α deficiency decreases WAT angiogenesis and enhances WAT

341

inflammation in obesity.

342

Given the strong link between insulin resistance and AT inflammation (44-46), we sought

343

to determine the inflammatory status of the obese WAT in HIF2α-deficiency.

344

Immunohistochemical staining for F4/80, a marker for macrophages, showed increased

345

abundance of macrophages in the gonWAT and scWAT of obese AdHIF2KO mice, as

346

compared to control mice (Fig. 5A). AT inflammation is associated with formation of

15

347

crown-like structures, found mostly in gonWAT, consisting of apoptotic adipocytes

348

surrounded by macrophages (47). Besides higher macrophage accumulation, we found

349

more apoptotic cells, as measured by TUNEL immunohistochemistry in the gonAT of

350

obese AdHIF2KO mice (Fig. 5B). Furthermore, WAT dysfunction is linked to WAT

351

fibrosis, which occurs by macrophage-mediated remodeling of the extracellular matrix (34,

352

48). By Masson´s trichrome staining, which is specific for collagen deposition, we found

353

increased fibrosis of the gonWAT in obese AdHIF2KO mice (Fig. 5C). Together, adipocyte

354

HIF2α protects against obesity-mediated WAT inflammation and fibrosis and thus, WAT

355

dysfunction.

356 357

Adipocyte hypertrophy in obesity leads to hypoxic areas in the WAT of men and mice (4,

358

5). AT hypoxia stimulates vascular growth in the AT (8). Interestingly, WAT of obese

359

AdHIF2KO mice displayed marked hypoxia compared to the WAT of obese control mice

360

(Fig. 5D and 5E). Limited vascularization of WAT is linked to elevated presence of

361

hypoxia, as well as insulin resistance, enhanced AT inflammation and adipocyte death (8).

362

On the other hand, enhanced vascularization of AT prevents obesity-related metabolic

363

dysregulation of AT (8, 11). HIF2α is a major regulator of proangiogenic responses in

364

several physiological and pathological situations (18, 49, 50). We found reduced WAT

365

vascularization in obese AdHIF2KO mice, as compared to control mice, which was

366

reflected by reduced numbers of endothelial cells in both WAT depots, as assessed by flow

367

cytometry (Fig. 5F). Moreover, vascularization was studied by staining with isolectin B4

368

and was found to be decreased in the scWAT and gonWAT of obese AdHIF2KO mice, as

369

compared to controls upon 16 and 24 weeks of HFD (Fig. 5G and 5H, data from mice fed a

16

370

HFD for 24 weeks; data from mice fed a HFD for 16 weeks not shown). A major HIF2α

371

target regulating angiogenesis is VEGF-A (18, 49, 50). Consistently, its expression was

372

reduced in the gonWAT of obese AdHIF2KO mice (Fig. 5I and 5J). Thus, the absence of

373

HIF2α resulted in insufficient angiogenic response in the obese WAT, thereby contributing

374

to enhanced WAT hypoxia, inflammation and cell death.

375 376

No role of HIF2α in myeloid or endothelial cells for WAT adaptation to obesity

377

Our findings so far indicate that adipocyte HIF2α deletion contributes to obese WAT

378

dysfunction, including reduced vascularization and enhanced inflammation in the obese

379

WAT, thereby resulting in insulin resistance. A major expression of HIF2α is found in

380

myeloid and endothelial cells, and HIF2α in these cells is capable of regulating angiogenic

381

responses (50, 51). As Fabp4-Cre transgenic mice have been previously reported to

382

potentially delete target genes in non-adipocytes, e.g. in macrophages (13, 52, 53), we

383

sought to clarify if myeloid HIF2α played any role for the WAT phenotype observed in

384

obese AdHIF2KO mice. To this end, we generated myeloid-specific HIF2α-deficient mice

385

(MyeHIF2KO) by engaging LysM-Cre transgenic mice. Compared to their littermate

386

controls, mice lacking HIF2α in myeloid cells did not show any metabolic changes in diet-

387

induced obesity, as assessed by analysis of body weight, fat mass or insulin tolerance test

388

(Fig. 6A-6D). These data suggest that myeloid HIF2α does not substantially contribute to

389

obesity-related WAT dysfunction.

390 391

Moreover, HIF2α in endothelial cells regulates angiogenesis-related functions (51). To

392

assess the role of endothelial HIF2α, we generated endothelial cell-specific HIF2KO mice

17

393

(EndHIF2KO) and subjected them to diet-induced obesity. In contrast to AdHIF2KO mice,

394

mice lacking HIF2α in endothelial cells did not develop any metabolic alterations, as

395

compared to littermate HIF2α-proficient mice (Fig. 6F-6I). Importantly, the number of

396

endothelial cells in the obese WAT did not change in both MyeHIF2KO and EndHIF2KO

397

mice (Fig. 6E and 6J), as compared to control littermate mice, thus, indicating that obese

398

WAT angiogenesis is regulated predominantly by adipocyte HIF2α.

399 400

Ectopic accumulation of fat in adipocyte-HIF2α deficient mice

401

Deficient lipolytic activity of WAT has been linked to ectopic accumulation of fat (e.g. in

402

the liver) along with reduced insulin sensitivity (10, 54). Livers of obese AdHIF2KO mice

403

showed not only increased size (Fig. 2B) but also increased lipid accumulation. As assessed

404

by histology analysis, we found enhanced abundance of lipid droplets in livers due to

405

adipocyte HIF2α deficiency in mice fed a HFD for 16 or 24 weeks (Fig. 7A-7B, data from

406

mice fed a HFD for 24 weeks; data from mice fed a HFD for 16 weeks not shown). This

407

finding was confirmed by measuring liver triglyceride content, which was higher in

408

AdHIF2KO mice (Fig. 7C). By histological assessment, AdHIF2KO mice displayed

409

enhanced steatosis and ballooning and an enhanced NAFLD Activity Score (NAS) (Fig.

410

7D); a NAS >5, as observed in AdHIF2KO mice, correlates with the presence of NASH

411

(38). Gene expression analysis showed unaltered expression of lipolytic markers, such as

412

Ppar-α, Mcad or Cpt1 (Fig. 7E), but significantly increased expression of lipogenic

413

markers, such as Ppar-γ and Scd1, as well as a tendency towards increased expression of

414

Srebp1c and Fas in livers from obese AdHIF2KO mice, as compared to littermate control

415

mice (Fig. 7F). These findings could explain the enhanced steatosis observed in adipocyte

18

416

HIF2α−deficiency under obese conditions. While the expression of the main hepatic

417

glucose transporter, Glut2, was unchanged, fatty acid transporter (Cd36) expression was

418

significantly increased in livers of obese AdHIF2KO mice, thereby suggesting enhanced

419

uptake of lipids by the liver (Fig. 7G). The enhanced hepatic lipid uptake was also in

420

keeping with reduced serum free fatty acid (FFA) levels in obese AdHIF2KO mice (Fig.

421

7H).

422 423

Deficiency of adipocyte HIF2α leads to BAT dysfunction in obesity.

424

We next analyzed the BAT of AdHIF2KO, as BAT is one of the main targets of the activity

425

of Cre recombinase in Fabp4-Cre-transgenic mice (Fig. 1A). The main function of BAT is

426

heat generation via lipid consumption in brown adipocytes. Recently, BAT has gained

427

attention in the context of obesity (reviewed in (23)); with the progression of obesity, BAT

428

gains weight, accumulates lipids and brown adipocytes become bigger, switching from the

429

typical brown appearance (several lipid droplets per cell) to a “white fat-like” shape

430

comprising a huge fat drop.

431 432

As described above (Fig. 2B), the BAT of mice lacking HIF2α in adipocytes displayed

433

higher mass. In addition to its enhanced mass, the BAT of obese AdHIF2KO displayed a

434

marked pale color, suggesting lipid accumulation (Fig. 8A). Microscopic analysis of the

435

BAT revealed that brown adipocytes from HFD-fed AdHIF2KO were larger than those in

436

littermate control mice (Fig. 8B and 8C, data from mice fed a HFD for 24 weeks; data from

437

mice fed a HFD for 16 weeks not shown). Interestingly, this was not observed under ND

438

conditions (data not shown). Accordingly, BAT from HFD-fed AdHIF2KO mice showed

19

439

enhanced expression of lipogenic markers (Fig. 8D), suggesting that excessive lipogenesis

440

could contribute to the elevated cell size of brown adipocytes in HIF2α−deficiency.

441 442

Consistent with recent reports with regards to induction of inflammation in the BAT during

443

obesity (55, 56), we found that dysfunction of the BAT in obese AdHIF2KO mice was

444

linked to higher BAT inflammation, as assessed by increased accumulation of F4/80-

445

positive macrophages (Fig. 8E, data from mice fed a HFD for 24 weeks; data from mice fed

446

a HFD for 16 weeks not shown). Moreover, gene expression of F4/80 and other

447

inflammatory factors, such as Tnf or Mcp-1 was upregulated in the BAT of obese

448

AdHI2KO mice, as compared to littermate control mice (Fig. 8F).

449 450

BAT function largely depends on angiogenesis (3, 57); by staining with fluorochrome-

451

labeled isolectin-B4, we found reduced vascularity in the BAT of obese AdHIF2KO mice

452

(Fig. 8G), as compared to littermate control mice, suggesting that HIF2α supports

453

angiogenic processes in the BAT in obesity. Consistently, VEGF-A levels were

454

downregulated in the BAT of obese AdHIF2KO (Fig. 8H). Together, these data

455

demonstrate that HIF2α is necessary for promoting angiogenesis not only in WAT, but also

456

in BAT in obesity, by regulating BAT VEGF-A levels.

457 458

The main function of BAT is thermogenesis, which requires the action of UCP1. UCP1

459

increases the permeability of the inner mitochondrial membrane to reduce mitochondrial

460

membrane potential and thereby uncouple ATP generation from the respiratory chain

461

leading to heat generation (reviewed in (58)). Interestingly, we found reduced UCP1

20

462

protein and mRNA in the BAT of AdHIF2KO mice, as compared to littermate HIF2α-

463

proficient mice, under obese (Fig. 9A, 9B, 9F, data from mice fed a HFD for 24 weeks;

464

data from mice fed a HFD for 16 weeks not shown) but not lean conditions (data not

465

shown). These data suggested that HIF2α-mediated regulation of UCP1 is operative only

466

during HFD-induced metabolic deterioration of the BAT. In contrast, other BAT markers

467

were not changed between obese control and AdHIF2KO mice (Fig. 9C).

468 469

Because of the essential role of UCP1 in heat generation, we performed a cold resistance

470

test in HFD-fed AdHIF2KO and control littermate mice. Obese AdHIF2KO mice were

471

unable to maintain adequate body temperature, as compared to HFD-fed control mice (Fig.

472

9D and 9E). Thus, not only HIF2α expression is induced in mice subjected to cold

473

exposure (25), but adipocyte HIF2α contributes to cold adaptation. Furthermore, gene

474

expression analysis revealed a significant reduction of UCP1 expression in the BAT of

475

obese AdHIF2KO mice, as compared to littermate control mice under both cold and room

476

temperature conditions (Fig. 9F). Additionally, reduced expression of Vegf-a was observed

477

in the BAT of obese AdHIF2KO mice under both cold and room temperature conditions

478

(Fig. 9F), as compared to obese HIF2α-proficient mice. Together, these findings

479

demonstrate that adipocyte HIF2α deficiency promotes obesity-related BAT dysfunction.

480 481

BAT dysfunction and enhanced AT inflammation at early stages of obesity in

482

AdHIF2KO mice.

483

In order to understand the mechanisms underlying the metabolic alterations seen in

484

AdHIF2KO mice, we analyzed mice fed with a HFD for only 4 weeks, hence, before body

21

485

mass differences are established between AdHIF2KO and control mice. In contrast to

486

longer feedings (16-24 weeks; Fig. 2B and data not shown), AdHIF2KO fed for only 4

487

weeks did not show any difference in WAT weights while BAT weight was already

488

significantly elevated (Fig. 10A). Moreover, brown adipocytes were enlarged in

489

AdHIF2KO mice already at early stages of obesity (Fig. 10B). In addition, UCP1

490

expression was reduced and macrophages were more abundant in the BAT from

491

AdHIF2KO mice after only 4 weeks of diet (Fig. 10B-10C). Despite unchanged tissue mass,

492

sc and gon WAT from AdHIF2KO mice displayed enhanced accumulation of macrophages,

493

associated with reduced vascularization, indicating WAT dysfunction due to HIF2α

494

deficiency in adipocytes in early stages of obesity (Fig. 10D-10E). In contrast, lipid

495

accumulation in the liver, as well as markers for lipolysis, lipogenesis and FA uptake were

496

not altered in AdHIF2KO, as compared to control mice, upon 4 week HFD feeding (Fig.

497

10F-10I). Together, these data indicate that enhanced inflammation and dysfunction

498

especially of the BAT is already present at early stages of diet-induced obesity due to

499

adipocyte HIF2α deficiency and is likely the primary event leading to metabolic

500

dysregulation in these mice.

501 502

VEGF rescues BAT dysfunction in obese AdHIF2KO mice.

503

Mice lacking HIF2α in adipocytes showed BAT dysfunction already at early stages of

504

obesity (Fig. 10A-10C) accompanied by reduced VEGF expression in the BAT (Fig. 8H).

505

We therefore addressed next, whether the impaired levels of this major proangiogenic

506

factor could contribute to elevated BAT dysfunction and inflammation. To this end, obese

507

control and AdHIF2KO were subjected to VEGF administration via mini-osmotic pumps or

22

508

PBS as control treatment. VEGF treatment reversed the enhanced weight of obese BAT due

509

to adipocyte HIF2α deficiency (Fig. 11A). Consistently, VEGF treatment reversed the

510

HIF2α deficiency-associated reduced UCP1 expression of the BAT and elevated

511

macrophage accumulation in the BAT (Fig. 11B-11C). In other words the BAT dysfunction

512

in adipocyte HIF2α deficiency was reversed, at least partially, by VEGF administration.

513

Additionally, VEGF administration efficiently reversed the exacerbated macrophage

514

accumulation and the reduced vascularization in the WAT of obese AdHIF2KO mice (Fig.

515

11D-11E). Together, these data demonstrate that the reduced levels of VEGF resulting from

516

adipocyte HIF2α-deficiency contribute to BAT inflammation and dysfunction (reduced

517

UCP1 expression) in obese AdHIF2KO mice.

518 519

DISCUSSION

520

We identified here HIF2α in white and brown adipocytes as an important factor

521

counteracting the maladaptation of WAT and BAT to obesity. Especially, our present work

522

describes for the first time that adipocyte HIF2α contributes to the regulation of the

523

adaptation of BAT to obesity. Accordingly, obese AdHIF2KO mice displayed BAT

524

dysfunction, such as increased BAT mass, enlarged brown adipocytes and decreased UCP1

525

gene and protein expression.

526 527

Metabolic dysregulation of BAT in obese AdHIF2KO mice was linked to a dysfunctional

528

thermogenic response of these mice to cold exposure. In particular, obese AdHIF2KO,

529

exposed to cold, failed to sustain their body temperature. These data are in keeping with the

530

previous observation that cold exposure induced HIF2α expression (25). In fact, BAT

23

531

hypoxia is not only present in the course of obesity (3), but is also one of the earliest events

532

during cold challenge, likely resulting from the excessive demand of oxygen that is

533

required for UCP1 activity and heat generation (25).

534 535

Noteworthy, we found that the expression of the major thermogenic factor UCP1 was

536

altered in adipocyte HIF2α deficiency. In contrast, a recent report did not show alterations

537

in UCP1 expression in the BAT of adipocyte-specific HIF1α-deficient mice compared to

538

adipocyte HIF1α-proficient mice in obesity (6). Although the underlying mechanisms of

539

UCP1 regulation by adipocyte HIF2α are not known and merit future investigations, our

540

findings are consistent with previous work showing that mice with adipocyte-specific

541

deficiency of PHD2, which is a negative regulator of HIF2α, showed enhanced UCP1

542

expression in obese BAT (15). VEGF has been previously shown to promote UCP1

543

expression (3, 11, 57). The fact that VEGF treatment rescued the reduced UCP1 expression

544

in HIF2α-deficient brown adipocytes suggests that the regulation of UCP1 by HIF2α may

545

involve VEGF. Moreover, UCP1 expression was not affected in lean AdHIF2KO mice, as

546

compared to control mice, indicating that the HIF2α-mediated regulation on UCP1 may be

547

operative in adaptive rather than constitutive BAT responses, such as under chronic

548

excessive lipid consumption characteristic of obesity or under acute cold exposure. Taken

549

together, our data demonstrate that adipocyte HIF2α is one of the factors that contribute to

550

BAT adaptation to obesity.

551 552

Furthermore, deletion of adipocyte HIF2α resulted in enhanced WAT dysfunction only in

553

the obese but not in the lean state, as indicated by reduced WAT angiogenesis, enhanced

24

554

WAT inflammation, increased WAT fibrosis and adipocyte death. Consistent with the

555

reduced expression of lipases and of genes involved in lipid oxidation, with the reduced ex

556

vivo lipolysis and the reduced lipid tolerance in adipocyte HIF2α-deficiency, we found

557

enhanced blood cholesterol levels, reduced serum free fatty acids and ectopic fat

558

accumulation in the liver (54). AdHIF2KO mice subjected to a short HFD feeding (4 weeks)

559

displayed BAT dysfunction and enhanced AT inflammation, as compared to control mice,

560

whereas liver lipid accumulation was not affected at the early stage of obesity. A recent

561

study briefly described that lack of HIF2α in adipocytes enhanced body weight and

562

gonWAT inflammation and promoted insulin resistance (6) without addressing WAT

563

angiogenesis or assessing further metabolic organs apart from the gonWAT, such as the

564

scWAT and the liver, and more importantly without reporting the role of adipocyte

565

HIF2α in BAT, as we have thoroughly done here. We provide here the crucial detailed

566

insights with regards to WAT and BAT dysfunction in obese AdHIF2KO mice.

567 568

A recent report has shown that heterozygous HIF2α+/- mice are prone to insulin resistance

569

and AT inflammation in obesity (17); these authors attributed their findings to the absence

570

of HIF2α from macrophages because clodronate-mediated macrophage depletion improved

571

glucose intolerance in HIF2α+/- mice. However, the more specific strategy employed here

572

(myeloid-specific HIF2α-deficiency) did not reveal a role of macrophage HIF2α in obesity-

573

related WAT inflammation and metabolic dysregulation. It is likely that the phenotype of

574

heterozygous HIF2α+/- mice in obesity resulted from the partial deficiency of adipocyte

575

HIF2α. Macrophage-mediated WAT inflammation is a common downstream event of

576

several pathways leading to WAT dysfunction and several examples exist that amelioration

25

577

of WAT inflammation improves obesity-related insulin resistance (59). Thus, it is quite

578

possible that obesity-related metabolic dysregulation initiated by WAT dysfunction due to

579

adipocyte-specific HIF2α-deficiency could be improved by manipulating a downstream

580

effector, e.g. by macrophage ablation. Together, our data clearly demonstrate that adipocyte

581

but not endothelial or myeloid HIF2α orchestrates the intimate crosstalk between

582

adipocytes, macrophages and the endothelium within the WAT.

583 584

Although Fabp4-Cre transgenic mice have been extensively used to achieve adipocyte -

585

specific deletion, some degree of recombination has been observed in other cells/tissues,

586

such as macrophages or endothelial cells, has been reported (53). In the present work we

587

did not find significant HIF2α deletion in the liver, skeletal muscle, heart, hypothalamus or

588

bone marrow macrophages of AdHIF2KO mice. This finding together with the absence of

589

any phenotype in the myeloid- and endothelium-specific HIF2α deficient mice in diet-

590

induced obesity, allow us to conclude that the observed phenotypes in AdHIF2KO mice

591

essentially derive from the deletion of HIF2α in adipocytes. Together, adipocyte but not

592

myeloid or endothelial HIF2α is responsible for regulating the angiogenic response within

593

the WAT in obesity and for reducing obesity-related WAT inflammation and metabolic

594

dysregulation.

595 596

Our findings support and extend a previous report demonstrating that vascular rarefaction

597

by adipocyte-specific deletion of VEGF promotes BAT dysfunction and whitening (3).

598

Although previous reports have shown that hypoxia in obesity induces HIFs (2, 6, 12) and

599

have illustrated the importance of proangiogenic responses in WAT and BAT adaptation to

26

600

obesity (3, 8, 9, 11, 57), the involvement of HIF1α or HIF2α in this process was not

601

clarified so far. In fact, several lines of evidence have suggested that HIF1α does not

602

participate in regulating Vegf-a expression and in inducing angiogenesis in both WAT and

603

BAT. On the other hand, adipocytes lacking HIF1β, and thereby HIF1α and HIF2α

604

signaling,

605

overexpression of HIF1α in adipocytes did not enhance VEGF-A expression (12) and

606

adipocyte-specific HIF1α-deleted mice had no alterations in endothelial cell numbers, as

607

compared to HIF1α-proficient mice (6). Similarly, HIF1α overexpression in brown

608

adipocytes did not enhance Vegf-a expression (3). These findings on the role of HIF1β and

609

HIF1α, together with our present findings of reduced VEGF-A levels, endothelial cell

610

numbers and vascularity of WAT and BAT in obese adipocyte-specific HIF2α-deficient

611

mice unequivocally underline the primacy of adipocyte HIF2α as the major HIF isoform

612

orchestrating the angiogenic response in the WAT and BAT in obesity. The clarification of

613

the distinct actions of HIF1α and HIF2α in adipocytes adds yet another example to the

614

variable, non-redundant and often opposite functions these two transcription factors have in

615

several biological processes (1).

showed

reduced

VEGF-A

expression

(21).

Additionally,

transgenic

616 617

In conclusion, we demonstrate that adipocyte HIF2α regulates angiogenesis in the obese

618

WAT and BAT. Furthermore, adipocyte HIF2α is integral to the thermogenic response of

619

the BAT in obesity by regulating UCP1 expression. Through these complementary

620

mechanisms adipocyte HIF2α counteracts BAT dysfunction, AT inflammation and

621

metabolic dysregulation and insulin resistance in obesity.

622 27

623 624

AUTHORS´ CONTRIBUTIONS

625

RGM designed and performed experiments, analysed data and wrote manuscript. KJC and

626

VIA designed experiments and analysed data. NQ, ME, MFRC, BG, KK, AZ, AC and JP

627

performed experiments. MB provided critical reagents. GE, GB, MEB, SRB, AEA and JH

628

participated in data interpretation and editing of the manuscript. TC conceived the project,

629

designed experiments and wrote the manuscript.

630 631

FUNDING SOURCES

632

This work was supported by the Deutsche Forschungsgemeinschaft (CH279/5-1 and

633

CH279/6-2 to TC), by the ERC (ENDHOMRET to TC), by the Else Kröner Fresenius-

634

Stiftung (to TC) and by a seed grant from the Center for Regenerative Therapies Dresden

635

(to TC, GB and MEB).

636 637

ACKNOWLEDGEMENTS

638

We would like to acknowledge S. Korten, S. Grossklaus, J. Gebler, M. Prucnal and C.

639

Mund for technical support.

640 641 642

DISCLOSURES

643

Authors declare no conflict of interest.

644 645

REFERENCES

28

646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691

1. 2. 3. 4. 5.

6.

7.

8.

9.

10. 11.

12.

13.

Majmundar, A. J., W. J. Wong, and M. C. Simon. 2010. Hypoxia-inducible factors and the response to hypoxic stress. Mol Cell 40:294-309. Rausch, M. E., S. Weisberg, P. Vardhana, and D. V. Tortoriello. 2008. Obesity in C57BL/6J mice is characterized by adipose tissue hypoxia and cytotoxic T-cell infiltration. Int J Obes (Lond) 32:451-463. Shimizu, I., T. Aprahamian, R. Kikuchi, A. Shimizu, K. N. Papanicolaou, S. MacLauchlan, S. Maruyama, and K. Walsh. 2014. Vascular rarefaction mediates whitening of brown fat in obesity. J Clin Invest 124:2099-2112. Ye, J., Z. Gao, J. Yin, and Q. He. 2007. Hypoxia is a potential risk factor for chronic inflammation and adiponectin reduction in adipose tissue of ob/ob and dietary obese mice. Am J Physiol Endocrinol Metab 293:E1118-1128. Pasarica, M., O. R. Sereda, L. M. Redman, D. C. Albarado, D. T. Hymel, L. E. Roan, J. C. Rood, D. H. Burk, and S. R. Smith. 2009. Reduced adipose tissue oxygenation in human obesity: evidence for rarefaction, macrophage chemotaxis, and inflammation without an angiogenic response. Diabetes 58:718-725. Lee, Y. S., J. W. Kim, O. Osborne, Y. Oh da, R. Sasik, S. Schenk, A. Chen, H. Chung, A. Murphy, S. M. Watkins, O. Quehenberger, R. S. Johnson, and J. M. Olefsky. 2014. Increased Adipocyte O2 Consumption Triggers HIF-1alpha, Causing Inflammation and Insulin Resistance in Obesity. Cell 157:1339-1352. Goossens, G. H., A. Bizzarri, N. Venteclef, Y. Essers, J. P. Cleutjens, E. Konings, J. W. Jocken, M. Cajlakovic, V. Ribitsch, K. Clement, and E. E. Blaak. 2014. Increased adipose tissue oxygen tension in obese compared with lean men is accompanied by insulin resistance, impaired adipose tissue capillarization, and inflammation. Circulation 124:67-76. Sung, H. K., K. O. Doh, J. E. Son, J. G. Park, Y. Bae, S. Choi, S. M. Nelson, R. Cowling, K. Nagy, I. P. Michael, G. Y. Koh, S. L. Adamson, T. Pawson, and A. Nagy. 2013. Adipose vascular endothelial growth factor regulates metabolic homeostasis through angiogenesis. Cell Metab 17:61-72. Gealekman, O., N. Guseva, C. Hartigan, S. Apotheker, M. Gorgoglione, K. Gurav, K. V. Tran, J. Straubhaar, S. Nicoloro, M. P. Czech, M. Thompson, R. A. Perugini, and S. Corvera. 2011. Depot-specific differences and insufficient subcutaneous adipose tissue angiogenesis in human obesity. Circulation 123:186194. Sun, K., I. Wernstedt Asterholm, C. M. Kusminski, A. C. Bueno, Z. V. Wang, J. W. Pollard, R. A. Brekken, and P. E. Scherer. 2012. Dichotomous effects of VEGF-A on adipose tissue dysfunction. Proc Natl Acad Sci U S A 109:5874-5879. Elias, I., S. Franckhauser, T. Ferre, L. Vila, S. Tafuro, S. Munoz, C. Roca, D. Ramos, A. Pujol, E. Riu, J. Ruberte, and F. Bosch. 2012. Adipose tissue overexpression of vascular endothelial growth factor protects against diet-induced obesity and insulin resistance. Diabetes 61:1801-1813. Halberg, N., T. Khan, M. E. Trujillo, I. Wernstedt-Asterholm, A. D. Attie, S. Sherwani, Z. V. Wang, S. Landskroner-Eiger, S. Dineen, U. J. Magalang, R. A. Brekken, and P. E. Scherer. 2009. Hypoxia-inducible factor 1alpha induces fibrosis and insulin resistance in white adipose tissue. Mol Cell Biol 29:4467-4483. Jiang, C., A. Qu, T. Matsubara, T. Chanturiya, W. Jou, O. Gavrilova, Y. M. Shah, and F. J. Gonzalez. 2011. Disruption of hypoxia-inducible factor 1 in

29

692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737

14.

15.

16.

17. 18.

19.

20. 21. 22. 23. 24.

25.

adipocytes improves insulin sensitivity and decreases adiposity in high-fat diet-fed mice. Diabetes 60:2484-2495. Zhang, X., K. S. Lam, H. Ye, S. K. Chung, M. Zhou, Y. Wang, and A. Xu. 2010. Adipose tissue-specific inhibition of hypoxia-inducible factor 1{alpha} induces obesity and glucose intolerance by impeding energy expenditure in mice. J Biol Chem 285:32869-2877. Matsuura, H., T. Ichiki, E. Inoue, M. Nomura, R. Miyazaki, T. Hashimoto, J. Ikeda, R. Takayanagi, G. H. Fong, and K. Sunagawa. 2013. Prolyl hydroxylase domain protein 2 plays a critical role in diet-induced obesity and glucose intolerance. Circulation 127:2078-2087. Rahtu-Korpela, L., S. Karsikas, S. Horkko, R. Blanco Sequeiros, E. Lammentausta, K. A. Makela, K. H. Herzig, G. Walkinshaw, K. I. Kivirikko, J. Myllyharju, R. Serpi, and P. Koivunen. 2014. HIF prolyl 4-hydroxylase-2 inhibition improves glucose and lipid metabolism and protects against obesity and metabolic dysfunction. Diabetes 63:3324-3333. Choe, S. S., K. C. Shin, S. Ka, Y. K. Lee, J. S. Chun, and J. B. Kim. 2014. Macrophage HIF-2alpha Ameliorates Adipose Tissue Inflammation and Insulin Resistance in Obesity. Diabetes 63:3359-3371. Taniguchi, C. M., E. C. Finger, A. J. Krieg, C. Wu, A. N. Diep, E. L. LaGory, K. Wei, L. M. McGinnis, J. Yuan, C. J. Kuo, and A. J. Giaccia. 2013. Cross-talk between hypoxia and insulin signaling through Phd3 regulates hepatic glucose and lipid metabolism and ameliorates diabetes. Nat Med 19:1325-1330. Wei, K., S. M. Piecewicz, L. M. McGinnis, C. M. Taniguchi, S. J. Wiegand, K. Anderson, C. W. Chan, K. X. Mulligan, D. Kuo, J. Yuan, M. Vallon, L. C. Morton, E. Lefai, M. C. Simon, J. J. Maher, G. Mithieux, F. Rajas, J. P. Annes, O. P. McGuinness, G. Thurston, A. J. Giaccia, and C. J. Kuo. 2013. A liver Hif2alpha-Irs2 pathway sensitizes hepatic insulin signaling and is modulated by Vegf inhibition. Nat Med 19:1331-1337. Rankin, E. B., J. Rha, M. A. Selak, T. L. Unger, B. Keith, Q. Liu, and V. H. Haase. 2009. Hypoxia-inducible factor 2 regulates hepatic lipid metabolism. Mol Cell Biol 29:4527-4538. Lee, K. Y., S. Gesta, J. Boucher, X. L. Wang, and C. R. Kahn. 2011. The differential role of Hif1beta/Arnt and the hypoxic response in adipose function, fibrosis, and inflammation. Cell Metab 14:491-503. Cao, Y. 2010. Adipose tissue angiogenesis as a therapeutic target for obesity and metabolic diseases. Nat Rev Drug Discov 9:107-115. Harms, M., and P. Seale. 2013. Brown and beige fat: development, function and therapeutic potential. Nat Med 19:1252-1263. Fredriksson, J. M., J. M. Lindquist, G. E. Bronnikov, and J. Nedergaard. 2000. Norepinephrine induces vascular endothelial growth factor gene expression in brown adipocytes through a beta -adrenoreceptor/cAMP/protein kinase A pathway involving Src but independently of Erk1/2. J Biol Chem 275:13802-13811. Xue, Y., N. Petrovic, R. Cao, O. Larsson, S. Lim, S. Chen, H. M. Feldmann, Z. Liang, Z. Zhu, J. Nedergaard, B. Cannon, and Y. Cao. 2009. Hypoxiaindependent angiogenesis in adipose tissues during cold acclimation. Cell Metab 9:99-109.

30

738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783

26.

27. 28. 29.

30. 31.

32.

33. 34.

35. 36.

37.

Asano, A., M. Morimatsu, H. Nikami, T. Yoshida, and M. Saito. 1997. Adrenergic activation of vascular endothelial growth factor mRNA expression in rat brown adipose tissue: implication in cold-induced angiogenesis. Biochem J 328 ( Pt 1):179-183. Bagchi, M., L. A. Kim, J. Boucher, T. E. Walshe, C. R. Kahn, and P. A. D'Amore. 2013. Vascular endothelial growth factor is important for brown adipose tissue development and maintenance. Faseb J 27:3257-3271. Nikami, H., J. Nedergaard, and J. M. Fredriksson. 2005. Norepinephrine but not hypoxia stimulates HIF-1alpha gene expression in brown adipocytes. Biochem Biophys Res Commun 337:121-126. Matthias, A., K. B. Ohlson, J. M. Fredriksson, A. Jacobsson, J. Nedergaard, and B. Cannon. 2000. Thermogenic responses in brown fat cells are fully UCP1dependent. UCP2 or UCP3 do not substitute for UCP1 in adrenergically or fatty scid-induced thermogenesis. J Biol Chem 275:25073-25081. Cross, M., I. Mangelsdorf, A. Wedel, and R. Renkawitz. 1988. Mouse lysozyme M gene: isolation, characterization, and expression studies. Proc Natl Acad Sci U S A 85:6232-6236. Gothert, J. R., S. E. Gustin, J. A. van Eekelen, U. Schmidt, M. A. Hall, S. M. Jane, A. R. Green, B. Gottgens, D. J. Izon, and C. G. Begley. 2004. Genetically tagging endothelial cells in vivo: bone marrow-derived cells do not contribute to tumor endothelium. Blood 104:1769-1777. Langer, H. F., V. V. Orlova, C. Xie, S. Kaul, D. Schneider, A. S. Lonsdorf, M. Fahrleitner, E. Y. Choi, V. Dutoit, M. Pellegrini, S. Grossklaus, P. P. Nawroth, G. Baretton, S. Santoso, S. T. Hwang, B. Arnold, and T. Chavakis. 2011. A novel function of junctional adhesion molecule-C in mediating melanoma cell metastasis. Cancer Res 71:4096-4105. Judex, S., Y. K. Luu, E. Ozcivici, B. Adler, S. Lublinsky, and C. T. Rubin. 2010. Quantification of adiposity in small rodents using micro-CT. Methods 50:14-19. Phieler, J., K. J. Chung, A. Chatzigeorgiou, A. Klotzsche-von Ameln, R. Garcia-Martin, D. Sprott, M. Moisidou, T. Tzanavari, B. Ludwig, E. Baraban, M. Ehrhart-Bornstein, S. R. Bornstein, H. Mziaut, M. Solimena, K. P. Karalis, M. Economopoulou, J. D. Lambris, and T. Chavakis. 2013. The complement anaphylatoxin C5a receptor contributes to obese adipose tissue inflammation and insulin resistance. J Immunol 191:4367-4374. Livak, K. J., and T. D. Schmittgen. 2001. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25:402-408. Choi, E. Y., V. V. Orlova, S. C. Fagerholm, S. M. Nurmi, L. Zhang, C. M. Ballantyne, C. G. Gahmberg, and T. Chavakis. 2008. Regulation of LFA-1dependent inflammatory cell recruitment by Cbl-b and 14-3-3 proteins. Blood 111:3607-3614. Chatzigeorgiou, A., K. J. Chung, R. Garcia-Martin, V. I. Alexaki, A. Klotzsche-von Ameln, J. Phieler, D. Sprott, W. Kanczkowski, T. Tzanavari, M. Bdeir, S. Bergmann, M. Cartellieri, M. Bachmann, P. Nikolakopoulou, A. Androutsellis-Theotokis, G. Siegert, S. R. Bornstein, M. H. Muders, L. Boon, K. P. Karalis, E. Lutgens, and T. Chavakis. 2014. Dual role of B7 costimulation in

31

784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830

38.

39. 40. 41. 42.

43.

44.

45. 46. 47. 48. 49. 50.

obesity-related non-alcoholic steatohepatitis (NASH) and metabolic dysregulation. Hepatology 60:1196-1210. Kleiner, D. E., E. M. Brunt, M. Van Natta, C. Behling, M. J. Contos, O. W. Cummings, L. D. Ferrell, Y. C. Liu, M. S. Torbenson, A. Unalp-Arida, M. Yeh, A. J. McCullough, and A. J. Sanyal. 2005. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 41:1313-1321. Mehlem, A., C. E. Hagberg, L. Muhl, U. Eriksson, and A. Falkevall. 2013. Imaging of neutral lipids by oil red O for analyzing the metabolic status in health and disease. Nat Protoc 8:1149-1154. Xue, Y., S. Lim, E. Brakenhielm, and Y. Cao. 2010. Adipose angiogenesis: quantitative methods to study microvessel growth, regression and remodeling in vivo. Nat Protoc 5:912-920. Weir, J. B. 1949. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol 109:1-9. He, W., Y. Barak, A. Hevener, P. Olson, D. Liao, J. Le, M. Nelson, E. Ong, J. M. Olefsky, and R. M. Evans. 2003. Adipose-specific peroxisome proliferatoractivated receptor gamma knockout causes insulin resistance in fat and liver but not in muscle. Proc Natl Acad Sci U S A 100:15712-15717. Xu, J., D. J. Lloyd, C. Hale, S. Stanislaus, M. Chen, G. Sivits, S. Vonderfecht, R. Hecht, Y. S. Li, R. A. Lindberg, J. L. Chen, D. Y. Jung, Z. Zhang, H. J. Ko, J. K. Kim, and M. M. Veniant. 2009. Fibroblast growth factor 21 reverses hepatic steatosis, increases energy expenditure, and improves insulin sensitivity in dietinduced obese mice. Diabetes 58:250-259. Chatzigeorgiou, A., T. Seijkens, B. Zarzycka, D. Engel, M. Poggi, S. van den Berg, O. Soehnlein, H. Winkels, L. Beckers, D. Lievens, A. Driessen, P. Kusters, E. Biessen, R. Garcia-Martin, A. Klotzsche-von Ameln, M. Gijbels, R. Noelle, L. Boon, T. Hackeng, K. M. Schulte, A. Xu, G. Vriend, S. Nabuurs, K. J. Chung, K. Willems van Dijk, P. C. Rensen, N. Gerdes, M. de Winther, N. L. Block, A. V. Schally, C. Weber, S. R. Bornstein, G. Nicolaes, T. Chavakis, and E. Lutgens. 2014. Blocking CD40-TRAF6 signaling is a therapeutic target in obesityassociated insulin resistance. Proc Natl Acad Sci U S A 111:2686-2691. Chatzigeorgiou, A., K. P. Karalis, S. R. Bornstein, and T. Chavakis. 2012. Lymphocytes in obesity-related adipose tissue inflammation. Diabetologia 55:25832592. Osborn, O., and J. M. Olefsky. 2012. The cellular and signaling networks linking the immune system and metabolism in disease. Nat Med 18:363-374. Altintas, M. M., A. Azad, B. Nayer, G. Contreras, J. Zaias, C. Faul, J. Reiser, and A. Nayer. 2011. Mast cells, macrophages, and crown-like structures distinguish subcutaneous from visceral fat in mice. J Lipid Res 52:480-488. Khan, T., E. S. Muise, P. Iyengar, Z. V. Wang, M. Chandalia, N. Abate, B. B. Zhang, P. Bonaldo, S. Chua, and P. E. Scherer. 2009. Metabolic dysregulation and adipose tissue fibrosis: role of collagen VI. Mol Cell Biol 29:1575-1591. Rankin, E. B., J. Rha, T. L. Unger, C. H. Wu, H. P. Shutt, R. S. Johnson, M. C. Simon, B. Keith, and V. H. Haase. 2008. Hypoxia-inducible factor-2 regulates vascular tumorigenesis in mice. Oncogene 27:5354-5358. Imtiyaz, H. Z., E. P. Williams, M. M. Hickey, S. A. Patel, A. C. Durham, L. J. Yuan, R. Hammond, P. A. Gimotty, B. Keith, and M. C. Simon. 2010. Hypoxia32

831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867

51.

52.

53.

54.

55.

56. 57. 58. 59.

inducible factor 2alpha regulates macrophage function in mouse models of acute and tumor inflammation. J Clin Invest 120:2699-2714. Skuli, N., A. J. Majmundar, B. L. Krock, R. C. Mesquita, L. K. Mathew, Z. L. Quinn, A. Runge, L. Liu, M. N. Kim, J. Liang, S. Schenkel, A. G. Yodh, B. Keith, and M. C. Simon. 2012. Endothelial HIF-2alpha regulates murine pathological angiogenesis and revascularization processes. J Clin Invest 122:14271443. Tang, T., J. Zhang, J. Yin, J. Staszkiewicz, B. Gawronska-Kozak, D. Y. Jung, H. J. Ko, H. Ong, J. K. Kim, R. Mynatt, R. J. Martin, M. Keenan, Z. Gao, and J. Ye. 2010. Uncoupling of inflammation and insulin resistance by NF-kappaB in transgenic mice through elevated energy expenditure. J Biol Chem 285:4637-4644. Lee, K. Y., S. J. Russell, S. Ussar, J. Boucher, C. Vernochet, M. A. Mori, G. Smyth, M. Rourk, C. Cederquist, E. D. Rosen, B. B. Kahn, and C. R. Kahn. 2013. Lessons on conditional gene targeting in mouse adipose tissue. Diabetes 62:864-874. Haemmerle, G., A. Lass, R. Zimmermann, G. Gorkiewicz, C. Meyer, J. Rozman, G. Heldmaier, R. Maier, C. Theussl, S. Eder, D. Kratky, E. F. Wagner, M. Klingenspor, G. Hoefler, and R. Zechner. 2006. Defective lipolysis and altered energy metabolism in mice lacking adipose triglyceride lipase. Science 312:734-737. Nguyen, K. D., Y. Qiu, X. Cui, Y. P. Goh, J. Mwangi, T. David, L. Mukundan, F. Brombacher, R. M. Locksley, and A. Chawla. 2011. Alternatively activated macrophages produce catecholamines to sustain adaptive thermogenesis. Nature 480:104-108. Herrero, L., H. Shapiro, A. Nayer, J. Lee, and S. E. Shoelson. 2010. Inflammation and adipose tissue macrophages in lipodystrophic mice. Proc Natl Acad Sci U S A 107:240-245. Sun, K., C. M. Kusminski, K. Luby-Phelps, S. B. Spurgin, Y. A. An, Q. A. Wang, W. L. Holland, and P. E. Scherer. 2014. Brown adipose tissue derived VEGF-A modulates cold tolerance and energy expenditure. Mol Metab 3:474-483. Cannon, B., and J. Nedergaard. 2004. Brown adipose tissue: function and physiological significance. Physiol Rev 84:277-359. Feng, B., P. Jiao, Y. Nie, T. Kim, D. Jun, N. van Rooijen, Z. Yang, and H. Xu. 2011. Clodronate liposomes improve metabolic profile and reduce visceral adipose macrophage content in diet-induced obese mice. PLoS One 6:e24358.

33

868

FIGURE LEGENDS

869

Figure 1. HIF2α deletion in AdHIF2KO mice.

870

A, Epas1 (Hif2α) (left) and Hif1α (right) gene expression in subcutaneous (sc) and gonadal

871

(gon) white AT (WAT), brown AT (BAT), liver, quadriceps skeletal muscle and heart

872

(n=5-8/group). Gene expression of control mice was set as 1. B-C, Epas1 and Hif1α gene

873

expression in hypothalamus (B) and bone marrow-derived macrophages (BMDM) (C).

874

Gene expression of control mice was set as 1 (n=3-4/group).

875

Data are expressed as mean ± s.e.m. * P < 0.05.

876 877

Figure 2. Mice lacking HIF2α in adipocytes display worsened obesity-related

878

metabolic dysregulation.

879

A, Body weight of control and AdHIF2KO mice subjected to HFD. B, Mice were sacrificed

880

after 24 weeks of HFD, and subcutaneous (sc) and gonadal (gon) WAT, liver, spleen and

881

BAT were weighed. The tissue weights are presented as percentage of total body weight. C,

882

Lean and fat body mass was determined in obese (24 weeks on HFD) control and

883

AdHIF2KO mice by computed tomography (CT); data were normalized to control mice. D,

884

Glucose tolerance test of control and AdHIF2KO mice fed with HFD for 12 weeks. E,

885

Insulin tolerance test of control and AdHIF2KO mice fed with HFD for 14 weeks. F,

886

Glucose-stimulated insulin secretion of control and AdHIF2KO mice fed with HFD. G,

887

Control and AdHIF2KO mice were fed with a HFD for 15 weeks. Mice were fasted for 6

888

hours before an intraperitoneal injection of insulin (5 U/Kg of body weight), BAT was

889

harvested and analysed for phosphorylation of Akt at Ser473 and total Akt. Representative

890

blots (left) and densitometric quantification of the blots (right panel) are shown. Data

34

891

(densitometric analysis) are expressed as mean ± s.e.m. (n=6 mice per group). H-N, Plasma

892

insulin (H), blood glucose (I), plasma leptin (J), blood cholesterol (K), plasma adiponectin

893

(L), blood triglycerides (M) and plasma FGF-21 (N) of mice of the indicated genotypes fed

894

for 15 weeks with HFD and fasted overnight.

895

In A-F and H-N, data are expressed as mean ± s.e.m. (n=6-15 mice/group). * P < 0.05.

896 897

Figure 3. Lean AdHIF2KO do not show metabolic dysregulation.

898

A, Body weight of control and AdHIF2KO mice fed with normal diet (ND) (10% Kcal

899

from fat). B, Weights of subcutaneous (sc) and gonadal (gon) WAT, liver, spleen and BAT

900

from control and AdHIF2KO fed with ND for 24 weeks. C, Insulin tolerance test of control

901

and AdHIF2KO fed with ND for 13 weeks.

902

Data are expressed as mean ± s.e.m. (n=6-9 mice/group).

903 904

Figure 4. Obese AdHIF2KO mice show altered lipid metabolism.

905

A, Expression of metabolic genes in gonadal adipocyte fraction from control and

906

AdHIF2KO mice fed a HFD for 24 weeks is shown. Gene expression of control mice was

907

set as 1 (n=6-7/group). B, Subcutaneous (sc) WAT and gonadal (gon) WAT explants from

908

obese control and AdHIF2KO mice were cultured for 2h at 37ºC in the presence or absence

909

of 10 μM isoprenaline and release of FFA to the medium was quantified (n=4-6/group). C-

910

F, Control and AdHIF2KO were fed with a HFD for 6 weeks and placed in metabolic cages.

911

After 16 hours of adaptation, respiratory exchange ratio (RER) (C), food intake (D), oxygen

912

consumption (E) and energy expenditure (EE) (F) were monitored (n=6/group). G, Oral

913

lipid tolerance test of control and AdHIF2KO mice fed with a HFD for 15 weeks (n=12-

914

13/group). 35

915

Data are expressed as mean ± s.e.m. * P < 0.05.

916 917

Figure 5. Obese AdHIF2KO mice show enhanced WAT inflammation and reduced

918

WAT vascularity.

919

A-C, Mice were sacrificed after 24 weeks on a HFD. A, Representative pictures in gonadal

920

(gon) WAT (left) and quantification from subcutaneous (sc) and gon WAT (right panel) of

921

immunohistochemistry for F4/80 (n=5/group). Macrophage accumulation of control mice

922

was set as 1. Scale bar is 200 μm. B, Representative pictures of TUNEL

923

immunohistochemistry from the gonWAT of obese control and AdHIF2KO mice (24 weeks

924

on HFD), and the respective quantification (right panel) are shown (n=5/group). WAT

925

apoptosis of control mice was set as 1. Scale bar is 200 µm. C, Representative pictures of

926

Masson Trichrome staining for fibrosis detection in gonWAT from control and AdHIF2KO

927

mice. Scale bar is 200 μm. D and E, Representative images of staining for CCl-103F

928

(Hypoxiprobe-F6) in gonWAT (D) as well as quantification of hypoxia staining (E) from sc

929

and gonWAT from obese control and AdHIF2KO mice (16 weeks on HFD) are shown

930

(n=5-6/group). Black arrows indicate hypoxic areas. Scale bar is 200 µm. F-J, Mice were

931

sacrificed after 24 weeks on a HFD. F, ScWAT and gonWAT were digested with

932

collagenase, and flow cytometry analysis for CD31+CD45- cells was performed to analyse

933

endothelial cell numbers (n=10-17). The absolute endothelial cell number per g of tissue

934

was quantified. Data are shown relative to control; data of control mice were set as 1. G and

935

H, Representative images of scWAT (G) of isolectin B4 staining in whole mounts and

936

quantification (H) of isolectin B4 staining in whole mounts from scWAT and gonWAT

937

(n=5-7). Vascularization of control mice was set as 1. Scale bar is 200 μm. I, Vegf-a gene

36

938

expression in subcutaneous and gonadal adipocyte fractions from control and AdHIF2KO

939

mice (n=6-7). Gene expression of control mice was set as 1. J, VEGF-A protein levels

940

measured in scWAT and gonWAT lysates from control and AdHIF2KO mice, normalized

941

over total protein content (n=4-7).

942

Data are expressed as mean ± s.e.m. * P < 0.05.

943 944

Figure 6. No role of myeloid or endothelial HIF2α in obesity-related metabolic

945

dysregulation.

946

A, Effective deletion of Epas1 (Hif2α) but not of Hif1α in bone marrow-derived

947

macrophages from control and MyeHIF2KO mice. Gene expression of control mice was set

948

as 1. B, Body weight of HFD-fed control and MyeHIF2KO mice. C, Tissue weights of

949

obese control and MyeHIF2KO mice. D, Insulin tolerance test of obese MyeHIF2KO and

950

control mice. E, FACS staining for endothelial cells (CD31+CD45-) in the SVF of

951

subcutaneous (sc) and gonadal (gon) WAT of obese MyeHIF2KO and littermate control

952

mice. The absolute endothelial cell number per g of tissue was quantified. Data are shown

953

relative to control; data of control mice were set as 1. A-E: Data are expressed as mean ±

954

s.e.m. (n=4-5 per group). * P < 0.05.

955

F, Effective deletion of Epas1 (Hif2α) but not of Hif1α was assessed in sorted lung

956

endothelial cells from control and EndHIF2KO mice. Gene expression of control mice was

957

set as 1. G, Body weight of HFD-fed control and EndHIF2KO mice. H, Tissue weights of

958

obese EndHIF2KO and littermate control mice. I, Insulin tolerance test of obese

959

EndHIF2KO and littermate control mice. J, FACS staining for endothelial cells

960

(CD31+CD45-) from the SVF of scWAT and gonWAT of obese EndHIF2KO and

37

961

littermate control mice. The absolute endothelial cell number per g of tissue was quantified.

962

Data are shown relative to control; data of control mice were set as 1. F-J: Data are

963

expressed as mean ± s.e.m. (n=5 per group). * P < 0.05.

964 965

Figure 7. Obese AdHIF2KO mice develop enhanced hepatosteatosis.

966

Mice were sacrificed after 24 weeks on a HFD. A, Representative pictures of Oil-Red-O

967

(ORO) staining of livers from control and AdHIF2KO mice and quantification (n=5/group).

968

Steatosis (ORO staining) of control mice was set as 1. Scale bar is 200 μm. B,

969

Representative images (H&E staining) of obese control and AdHIF2KO livers. Scale bar is

970

200 µm. C, Triglyceride content was determined in livers from control and AdHIF2KO

971

mice fed for 24 weeks with HFD (n=6-7). D, Histological scoring for steatosis,

972

hepatocellular ballooning, lobular inflammation and NAFLD Activity Score (NAS) of

973

livers from obese control and AdHIF2KO mice (n=5). E-G, Gene expression analysis for

974

lipolytic (E) and lipogenic markers (F), as well as transporters for FFA (Cd36) and glucose

975

(Glut2) (G) in livers from obese control and AdHIF2KO mice (n=6-7). Gene expression of

976

control mice was set as 1. H, Serum free fatty acids (FFA) from control and AdHIF2KO

977

mice fed a HFD (n=8-12).

978

Data are expressed as mean ± s.e.m. * P < 0.05.

979 980

Figure 8. Dysfunction of BAT in obese AdHIF2KO mice.

981

A, Representative images of BAT of obese control and AdHIF2KO mice. B, Representative

982

H&E staining of BAT from control and AdHIF2KO mice upon HFD for 24 weeks. Scale

983

bar is 100 μm. C, Quantification of brown adipocyte cell diameter and fitting curve of

38

984

control (grey line) and AdHIF2KO mice (black line). D, Gene expression analysis for

985

lipogenic markers in BAT from obese control and AdHIF2KO mice fed a HFD for 24

986

weeks. Gene expression of control mice was set as 1. E, Quantification of F4/80

987

immunohistochemistry in BAT from control and AdHIF2KO mice upon HFD for 24 weeks.

988

Macrophage accumulation of control mice was set as 1. F, Gene expression of pro-

989

inflammatory markers in BAT from control and AdHIF2KO mice on HFD for 24 weeks.

990

Gene expression of control mice was set as 1. G, Representative images of isolectin B4

991

staining in whole mounts from BAT of control and AdHIF2KO mice on HFD for 24 weeks.

992

Scale bar is 100 μm. H, VEGF-A protein levels, normalized over total protein content, were

993

measured in BAT lysates from control and AdHIF2KO mice on HFD for 24 weeks.

994

Data in C-F and H are expressed as mean ± s.e.m. (n=5-7 mice per group). * P < 0.05.

995 996

Figure 9. Reduced UCP1 expression in the BAT of obese adipocyte-specific

997

HIF2α deficient mice.

998

A, Representative images of UCP1 immunohistochemistry in BAT from 24 week-HFD-fed

999

control and AdHIF2KO mice. Scale bar is 100 μm. B, Representative western blot for

1000

UCP1 and tubulin (left) and quantification (right) from BAT protein lysates from control

1001

and AdHIF2KO mice upon HFD for 24 weeks. The UCP1/tubulin ratio of control mice was

1002

set as 1 (n=4-5/group). C, Gene expression of Pgc1α, Prdm16, Dio2, Cidea and Adrb3 in

1003

BAT from control and AdHIF2KO mice fed a HFD for 24 weeks is shown. Gene

1004

expression of control mice was set as 1 (n=7-14/group). D and E, Obese control and

1005

AdHIF2KO mice were exposed to either room temperature (RT) (D) or 4ºC (E) and

1006

temperature was measured (n=3-4 for RT, n=9-12 for 4ºC). F, Gene expression analysis of

39

1007

the BAT from the mice displayed in D and E. Gene expression of control RT mice was set

1008

as 1.

1009

Data in B-F are expressed as mean ± s.e.m. * P < 0.05.

1010 1011

Figure 10. Metabolic dysregulation of WAT and BAT in AdHIF2KO mice in early

1012

stages of diet-induced obesity.

1013

Control and AdHIF2KO mice were fed with HFD for 4 weeks. A, Subcutaneous (sc) and

1014

gonadal (gon) WAT and BAT were weighed. The tissue weights are presented as

1015

percentage of total body weight. B, Representative images from H&E staining (upper

1016

panels) and UCP1 immunohistochemistry (lower panels) in BAT from control and

1017

AdHIF2KO mice. Scale bar is 100 μm. C, Quantification of immunohistochemistry for

1018

F4/80 in BAT from both genotypes. Macrophage accumulation of control mice was set as 1.

1019

D, Quantification of immunohistochemistry for F4/80 in sc and gonWAT from control and

1020

AdHIF2KO mice. Macrophage accumulation of control mice was set as 1. E, Vessels were

1021

stained for isolectin B4 in whole mounts of scWAT and gonWAT from control and

1022

AdHIF2KO mice and quantified. Vascularization of control mice was set as 1. F,

1023

Representative images from H&E staining in livers from control and AdHIF2KO. Scale bar

1024

is 200 μm. G-H, Gene expression analysis for lipolytic (G) and lipogenic markers (H), as

1025

well as transporters for FFA (Cd36) and glucose (Glut2) (I) in livers from obese control and

1026

AdHIF2KO mice. Gene expression of control mice was set as 1.

1027

Data in A, C-E and G-I are expressed as mean ± s.e.m. (n=6 mice per group). * P < 0.05.

1028

40

1029

Figure 11. VEGF administration reverses metabolic dysregulation of the BAT

1030

associated with adipocyte HIF2α deficiency.

1031

Control and AdHIF2KO were fed with HFD for a total of 8 weeks; for the last 3 weeks of

1032

the experiment, mini-osmotic pumps were implanted subcutaneously for administration of

1033

murine VEGF or PBS as a control. A, BAT weight after PBS or VEGF treatment is shown.

1034

The tissue weight is presented as percentage of total body weight. B, Representative images

1035

for UCP1 staining in BAT from control and AdHIF2KO mice. Scale bar is 100 μm. C,

1036

Quantification of immunohistochemistry for F4/80 in BAT from control and AdHIF2KO

1037

mice treated with either PBS or VEGF. Macrophage accumulation of control mice was set

1038

as 1. D, Quantification of immunohistochemistry for F4/80 in subcutaneous (sc) and

1039

gonadal (gon) WAT from control and AdHIF2KO mice treated with either PBS or VEGF.

1040

Macrophage accumulation of control mice was set as 1 in each case. E, Flow cytometry

1041

analysis for CD31+CD45- cells was performed to analyse absolute endothelial cell numbers.

1042

The absolute endothelial cell number per g of tissue was quantified. Data are shown relative

1043

to control mice; data of control mice were set as 1 in each case.

1044

Data in A, and C-E are expressed as mean ± s.e.m. (n=3-6 mice per group). * P < 0.05.

41