The necroptosis-inducing kinase RIPK3 dampens adipose tissue inflammation and glucose intolerance

Abstract

Receptor-interacting protein kinase 3 (RIPK3) mediates necroptosis, a form of programmed cell death that promotes inflammation in various pathological conditions, suggesting that it might be a privileged pharmacological target. However, its function in glucose homeostasis and obesity has been unknown. Here we show that RIPK3 is over expressed in the white adipose tissue (WAT) of obese mice fed with a choline-deficient high-fat diet. Genetic inactivation of Ripk3 promotes increased Caspase-8-dependent adipocyte apoptosis and WAT inflammation, associated with impaired insulin signalling in WAT as the basis for glucose intolerance. Similarly to mice, in visceral WAT of obese humans, RIPK3 is overexpressed and correlates with the body mass index and metabolic serum markers. Together, these findings provide evidence that RIPK3 in WAT maintains tissue homeostasis and suppresses inflammation and adipocyte apoptosis, suggesting that systemic targeting of necroptosis might be associated with the risk of promoting insulin resistance in obese patients.

ARTICLE

Received 21 Oct 2015 | Accepted 9 May 2016 | Published 21 Jun 2016

Jrmie Gautheron1,2,*, Mihael Vucur1,2,*, Anne T. Schneider1,2, Ilenia Severi3, Christoph Roderburg1,Sanchari Roy1,2, Matthias Bartneck1, Peter Schrammen1, Mauricio Berriel Diaz4, Josef Ehling5, Felix Gremse5, Felix Heymann1, Christiane Koppe1,2, Twan Lammers5, Fabian Kiessling5, Niels Van Best6, Oliver Pabst6,Gilles Courtois7, Andreas Linkermann8, Stefan Krautwald8, Ulf P. Neumann9, Frank Tacke1, Christian Trautwein1, Douglas R. Green10, Thomas Longerich11, Norbert Frey12, Mark Luedde12, Matthias Bluher13,Stephan Herzig4,14, Mathias Heikenwalder15 & Tom Luedde1,2

Receptor-interacting protein kinase 3 (RIPK3) mediates necroptosis, a form of programmed cell death that promotes inammation in various pathological conditions, suggesting that it might be a privileged pharmacological target. However, its function in glucose homeostasis and obesity has been unknown. Here we show that RIPK3 is over expressed in the white adipose tissue (WAT) of obese mice fed with a choline-decient high-fat diet. Genetic inactivation of Ripk3 promotes increased Caspase-8-dependent adipocyte apoptosis and WAT inammation, associated with impaired insulin signalling in WAT as the basis for glucose intolerance. Similarly to mice, in visceral WAT of obese humans, RIPK3 is overexpressed and correlates with the body mass index and metabolic serum markers. Together, these ndings provide evidence that RIPK3 in WAT maintains tissue homeostasis and suppresses inammation and adipocyte apoptosis, suggesting that systemic targeting of necroptosis might be associated with the risk of promoting insulin resistance in obese patients.

1 Department of Medicine III, University Hospital RWTH Aachen, Aachen 52074, Germany. 2 Division of GI and Hepatobiliary Oncology, University Hospital RWTH Aachen, Aachen 52074, Germany. 3 Department of Experimental and Clinical Medicine, University of Ancona, Ancona 60020, Italy. 4 Institute for Diabetes and Cancer IDC Helmholtz Center Munich, Neuherberg 85764 and Joint Heidelberg-IDC Translational Diabetes Program, Inner Medicine I, Heidelberg

University, Heidelberg 69120, Germany. 5 Department for Experimental Molecular Imaging, University Clinic and Helmholtz Institute for Biomedical Engineering RWTH Aachen, Aachen 52074, Germany. 6 Institut of Medical Microbiology, University Hospital RWTH Aachen, Aachen 52074, Germany. 7 Inserm U1038, BIG, CEA, Grenoble 38054, France. 8 Division of Nephrology and Hypertension, Christian-Albrechts-University, Kiel 24105, Germany. 9 Department of Visceral and Transplantation Surgery, University Hospital RWTH Aachen, Aachen 52074, Germany. 10 Department of Immunology, St Jude Childrens Research Hospital, Memphis, Tennessee 38105, USA. 11 Institute of Pathology, University Hospital RWTH Aachen, Aachen 52074, Germany. 12Department of Cardiology and Angiology, University Hospital Schleswig-Holstein, Campus Kiel, Kiel 24105, Germany. 13 Department of Medicine, University of Leipzig, Leipzig 04103, Germany. 14 German Center for Diabetes Research (DZD), Neuherberg 85764, Germany. 15 Division of Chronic Inammation and Cancer, German Cancer Research Center (DKFZ), Heidelberg 69120, Germany. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to T.L. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

DOI: 10.1038/ncomms11869 OPEN

The necroptosis-inducing kinase RIPK3 dampens adipose tissue inammation and glucose intolerance

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

Diabetes mellitus is increasing at an alarming rate and represents a major global health burden affecting 8.3% of the adult population worldwide1. Type 2 diabetes

mellitus has been associated with the pandemic spreading of overweight and obesity owing to the transition in lifestyle and dietary habits as well as ageing of the population in the setting of a genetical predisposition2. Excess adipose tissue predisposes towards the development of insulin resistance by an increased secretion of various adipocyte-derived proteins (referred as adipokines)3,4. Thereby, adipose tissue acts on many systemic processes including energy metabolism, inammation and the complications of the metabolic syndrome. In mouse models of obesity, high-fat diet (HFD) feeding results in an adipose tissue inammatory response characterized by the invasion of proinammatory macrophages (also referred to as M1), accompanied by the secretion of a variety of inammatory cytokines into the circulation5. This inammatory cascade is believed to impair insulin signal transduction, thereby causing glucose intolerance6. In parallel, obesity triggers the organization of macrophages around dead and/or dying adipocytes in so-called adipocyte crown-like structures (CLSs)7. On the basis of morphologic criteria, it was shown that adipocytes within the CLS structures display both characteristics of necrosis8 and apoptosis9. However, up to now the exact contribution of distinct forms of regulated cell death within white adipose tissue (WAT) to glucose intolerance is not well understood.

Historically, cell death was divided into two forms: the programmed cell death apoptosis, widely considered to prevent inammation, and the unregulated and accidental cell death necrosiswhich was considered to induce inammation10. Apoptosis represents a highly synchronized intracellular signalling pathway depending on activation of aspartate-specic proteases known as caspases11. Of these, Caspase-8 represents a key upstream Caspase that engages to the death-inducing signalling complex via the adaptor molecule Fas-Associated protein with Death Domain11. The receptor-interacting protein kinase 3 (RIPK3) and its substrate mixed lineage kinase-like (MLKL)12 mediate necroptosis, a newly discovered form of programmed cell death, which is activated upon tumour necrosis factor (TNF)-, antigen- or Toll-like-receptor stimulation13. Necroptosis contributes to various pathological conditions such as alcoholic- and methionine choline-decient-diet-induced non-alcoholic steatohepatitis14,15, atherosclerosis16, Gauchers disease17 and ischaemic heart and kidney injuries18,19, suggesting that this pathway might be a privileged pharmacological target in various diseases. However, the role of this pathway in obesity, metabolic syndrome and type 2 diabetes (T2D) has remained elusive. Our results show that obesity triggers RIPK3 overexpression in the WAT of mice and humans, where it dampens adipocyte apoptosis and inammation, thereby preventing impaired insulin signalling as the basis of glucose intolerance. These data provide evidence thatin contrast to its pro-inammatory functions in other organs and diseasesRIPK3 maintains WAT homeostasis.

ResultsRIPK3 dampens glucose intolerance in obese mice. To explore the functional role of RIPK3 in obesity and associated metabolic disorders, 6-week-old male C57BL/6 mice genetically ablated for Ripk3 (knockout, KO)20, and age- and sex-matched wild-type (WT) control mice were fed for 16 weeks with either normal chow diet (NCD) or a choline-decient HFD (CD-HFD), which is known to efciently recapitulate the key features of human metabolic syndrome and non-alcoholic steatohepatitis (NASH)21,22. As shown previously23, KO mice

on NCD displayed a slightly but signicantly reduced body weight gain compared with WT controls, while upon CD-HFD feeding, WT and KO mice showed similar body weight gain over time (Supplementary Fig. 1a,b). In line, X-ray computed tomography revealed a similar and signicant gain in whole-body fat mass, subcutaneous and visceral fat, upon CD-HFD feeding in WT and KO mice (Fig. 1a).

We next examined the effect of Ripk3 deciency on glucose homeostasis in CD-HFD-fed mice. As shown previously21,22, WT mice fed with CD-HFD displayed an impaired glucose tolerance in a standard glucose tolerance test compared with mice on NCD (Fig. 1b,c). Strikingly, CD-HFD-fed KO mice developed more pronounced glucose intolerance than WT animals (Fig. 1b,c). To evaluate the time point of impaired glucose tolerance development, we examined new independent groups of mice at 1, 4 and 7 months of CD-HFD feeding and could conrm impaired glucose tolerance in KO mice compared with WT mice at all examined time points (values in KO mice often exceeded the indicated threshold of 600 mg dl 1 of blood glucose (Fig. 1d,e)).

Thus, genetic ablation of Ripk3 promotes the development of glucose intolerance in obese mice. Of note, KO mice did not show signicant changes in ghrelin levels and food intake compared with WT mice on CD-HFD (Supplementary Fig. 1c,d), arguing against an inuence of Ripk3 deletion on hypothalamic appetite regulation or calorie consumption as a basis for the observed phenotype. Moreover, gut microbiota composition in WT versus KO mice kept under CD-HFD was similar but differed signicantly from an independent cohort of WT mice kept under NCD (Supplementary Fig. 2), arguing against a major inuence of the microbiota on the phenotypic differences between CD-HFD-fed WT and KO mice.

Impaired insulin signalling in Ripk3-decient obese mice. Current theories of the pathogenesis of glucose intolerance and T2D include a defect in glucose uptake and insulin action in the skeletal muscle and liver as well as disturbed adipocyte functions24. Therefore, to evaluate whether one of these tissues contributes to the impaired glucose tolerance in Ripk3-decient mice, we measured serum insulin levels and performed insulin tolerance tests (ITTs) in WT and KO mice fed with NCD or CD-HFD. In NCD-fed mice, we detected no signicant differences in insulin levels (Fig. 2a) and ITT (Fig. 2b) between WT and KO mice. In contrast, basic insulin levels were signicantly higher in obese KO compared with WT mice (Fig. 2c). Moreover, obese KO mice showed a signicantly more impaired ITT than WT mice (Fig. 2d), suggesting a higher degree of insulin resistance upon Ripk3 ablation after CD-HFD feeding.

To further characterize insulin-dependent signalling, mice were injected intraperitoneally with 1 U kg 1 insulin and euthanized 10 min later. Liver, skeletal muscle and epididymal WAT (epiWAT) protein samples were analysed by western blot analysis for phosphorylation of AKT, GSK3 and ERKknown intracellular mediators of insulin signalling25 (Fig. 2e). As expected from human T2D (ref. 26), CD-HFD feeding resulted in impaired phosphorylation of all examined mediators as a surrogate for insulin resistance in the skeletal muscle. In contrast, no clear difference was detected between WT and KO mice (Fig. 2e). In contrast, AKT and GSK3 phosphorylations in the liver tissue were not signicantly inuenced by CD-HFD feeding in both WT and KO mice, whereas ERK phosphorylation was impaired in KO mice (Fig. 2e). Interestingly, in epiWAT, KO mice showed a decrease in insulin-induced AKT and GSK phosphorylation on CD-HFD feeding compared with WT mice, while effects on ERK phosphorylation were not different (Fig. 2e). These ndings indicated a broad spectrum of aberrations in the phosphorylation

2 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869 ARTICLE

a

WT RIPK3/ WT RIPK3/

NCD

***

***

Fat mass (%)

60

40

20

0

WT

RIPK3/

RIPK3/

WT

NCD CD-HFD

Subcutaneous fat

Visceral fat

Lung Adipose tissue

CD-HFD

b c

800

400

200

GTT

Blood glucose (mg dl1 )

Blood glucose (mg dl1 )

*** ***

*** **

NS NS

NS

150

100

50

0

600

** *

WT

CD-HFD

Glycaemic response

(AUC, x103 )

RIPK3/

WT RIPK3/

NCD

WT

RIPK3/

RIPK3/

WT

0 0 15 30 60 90 120 150 180

Time (min)

Threshold

NCD

CD-HFD

d

WT RIPK3/

800

600

400

200

0 0 15 30 60 90 120 150 180 0 15 30 60 90 120 150 180 0 15 30 60 90 120 150 180

CD-HFD

**

* ***

**

800

600

400

200

* ** *** ***

***

** *** ***

**

1 Month 4 Months 7 Months Time (min)

Figure 1 | Ripk3 deciency induces glucose intolerance in obese mice. Data were obtained from Ripk3 constitutive knockout mice (referred to as KO in the main text and RIPK3 / in the gures) and WT control mice fed with a NCD or a CD-HFD for 16 weeks. Differences between WT and KO mice were determined by analysis of variance (ANOVA) with Bonferronis post hoc test. All data are expressed as means.e.m. (a) Left panel: three-dimensional volume renderings of segmented bones (white), lungs (pink) and fat (blue) upon in vivo mCT imaging (upper panel) as well as 2D cross-sectional mCT images in transversal planes of the abdomen of the mice (lower panel). Subcutaneous and visceral fat tissue is indicated with blue arrows. Scale bar, 1 cm.

Right panel: fat mass quantication (n 6 in each group) of WT and RIPK3 / mice after 16 weeks of NCD or CD-HFD. ***Po0.001. (b) Mice were

examined by glucose tolerance test (GTT). Results are expressed as mean with s.e.m., **Po0.01, ***Po0.001, n.s.: not signicant. (c) Area under the curve (AUC) for the glycaemic response was calculated using the trapezoidal rule (n 5 in each group). *Po0.05, **Po0.01, ***Po0.001, n.s., not signicant.

(d) WT and RIPK3 / mice fed with CD-HFD were examined at different time points 1 (n 4), 4 (n 6) and 7 (n 6) months with GTT. *Po0.05,

**Po0.01, ***Po0.001.

of mediators of insulin signalling in different tissues of KO mice, but also showed that Ripk3 deletion inuenced insulin sensitivity mainly in the adipose tissue.

RIPK3 in WAT of obese mice prevents inammation. To provide additional evidence for a specic function of RIPK3 in WAT, we examined expression levels of RIPK3 in epiWAT, liver and muscle tissue of NCD- and CD-HFD-fed mice. In line with

the lack of effects of Ripk3 deletion in the liver and muscle insulin signalling, 4 months of CD-HFD feeding did not alter RIPK3 expression in the liver and skeletal muscle of WT mice. In contrast, 4 months of CD-HFD feeding led to a strong induction of RIPK3 expression in epiWAT of WT animals (Fig. 3a), suggesting a specic regulatory function in adipose tissue. Next, we tested RIPK3 expression in epiWAT at different time points of CD-HFD feeding by western blot analysis. This analysis revealed that maximum expression occurred at 4 months of age (Fig. 3b).

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

a b

ITT

WT

RIPK3/

NCD

NS

Insulin (pg ml1 )

Insulin (pg ml1 )

NCD

4,000

3,000

2,000

1,000

0

Blood glucose (mg dl1 )

Blood glucose (mg dl1 )

180

160

140

120

80

100 Insulin

WT

0 15 30 60 90 120

RIPK3/

Time (min)

Time (min)

c d

ITT

220

200

180

160

120

WT

RIPK3/

CD-HFD

*

CD-HFD

***

***

***

*

*

15,000

10,000

5,000

0

140 Insulin

WT

0 15 30 60 90 120

RIPK3/

e

Liver Skeletal muscle epiWAT

(kDa)

WT RIPK3/ WT RIPK3/ WT RIPK3/

1.68 1.48 2.15 2.16 6.96 5.83 9.25 3.97 5.33 2.94 3.31 1.17

1.47 1.51 1.51 2.59 0.92 0.97 1.39 0.88

0.83 0.20 0.41 0.39

pS473

pY204

AKT

GSK-3

ERK

GAPDH

3.81 4.39 4.47 1.52

2.79 1.65 2.43 1.46

pS9/21

0.50 0.45 0.19 0.23

pT202

Insulin

+ + + + + + + + + + + +

60

60

50

50

50

50

40

NCD CD-HFD NCD CD-HFD NCD CD-HFD NCD CD-HFD NCD CD-HFD NCD CD-HFD

Figure 2 | Ripk3 deciency promotes insulin resistance in obese mice. Data were obtained from RIPK3 / and WT mice fed with NCD or CD-HFD for different time points. Differences between WT and RIPK3 / mice were determined by ANOVA with Bonferronis post hoc test. All data are shown as means.e.m. (a) Fasting serum concentrations of insulin in NCD-fed WT (n 18) and RIPK3 / (n 12) mice. n.s., not signicant. The feeding period

was 16 weeks. (b) WT and KO mice fed with NCD diet were examined by insulin tolerance test (ITT). (n 3) n.s., not signicant. (c) Fasting serum

concentrations of insulin in CD-HFD-fed WT (n 11) and RIPK3 / (n 12) mice. *Po0.05. The feeding period was 16 weeks. (d) Obese mice fed

for 24 weeks with CD-HFD were examined by ITT. n 6 for WT and n 7 for KO. *Po0.05, ***Po0.001. (e) WT and RIPK3 / mice fed with NCD and

CD-HFD for 16 weeks were injected intraperitoneally with 1 U kg 1 insulin or NaCl. Ten minutes later mice were euthanized and tissue samples harvested. Immunoblot analyses of skeletal muscle (left), liver (middle) and epiWAT (right) tissue samples using antibodies specic for AKT, phospho-AKT (pS473),

GSK-3, phospho-GSK-3 (pS9/21), ERK, phospho-ERK (pT202, pY204) and GAPDH as loading control. Numbers in the blot indicate relative insulin induction of AKT, GSK-3 and ERK phosphorylation.

4 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869 ARTICLE

a b c

WT RIPK3/

WT RIPK3/ WT RIPK3/

epiWAT WT

WT

epiWAT iWAT

Liver

(kDa)

(kDa)

(kDa)

NS

RIPK3

NS

RIPK3

GAPDH

50 40

RIPK3

GAPDH

GAPDH

Skeletal muscle

50 50

30

50

40

40

CD-HFD

1 4 12 Months

NCD

HFD NCD

HFD

RIPK3

RIPK3

* *

RIPK3 protein

(relative to GAPDH)

4

3

2

1

0

2.0

1.5

1.0

0.5

GAPDH

GAPDH

CD-HFD

(relative to GAPDH)

epiWAT

50

30

epiWAT iWAT

RIPK3 protein

+ +

d

H&E

H&E

H&E

Relative mRNA expression

2.5

2.0

1.5

1.0

0.5

0.0

Liver

Skeletal muscle

epiWAT

NCD

CD-HFD

e

NCD

CD-HFD

* *

WT RIPK3/

col11

Figure 3 | Obesity induces RIPK3 overexpression. Data were obtained from RIPK3 / and WT mice fed with NCD, CD-HFD or HFD for different time points. Differences between WT and RIPK3 / mice were determined by Students t-test. All data are expressed as means.e.m. (a) Protein levels of

RIPK3 in the liver, skeletal muscle and epiWAT from lean and diet-induced obese mice. GAPDH was used as loading control. n.s., not specic. Western blots are representative of three independent experiments. (b) Protein levels of RIPK3 in epiWAT of mice fed for 1, 4 and 12 months with CD-HFD. (c) Upper panel: protein levels of RIPK3 in epiWAT and inguinal WAT (iWAT) from WT mice fed for 6 months with NCD or HFD. Lower panel: bar graphs for quantication of relative expression of RIPK3 in these respective tissues (n 3). *Po0.05. (d) Representative haematoxylin and eosin (H&E) images of

liver (upper panel), skeletal muscle (middle panel) and epiWAT (lower panel) from WT and KO mice. Scale bars, 200 mm. (e) col1a1 mRNA levels were assessed by RTPCR in WT and KO fed for 16 weeks after 8 weeks of methionine choline-decient (MCD) diet feeding. Values were calculated relative to

WT mice fed with NCD, and b-actin was used as an internal standard, n 6 per group. Error bars represent s.e.m.

We also wanted to examine the compartment-specic regulation of RIPK3 expression in mice fed for 6 months with a conventional HFD. As shown in Fig. 3c, RIPK3 showed a similar upregulation of RIPK3 in epiWAT on conventional HFD feeding, as seen in CD-HFD-fed mice, suggesting that, independent of choline deciency, obesity represents a general stimulator for RIPK3 overexpression in epiWAT. Moreover, RIPK3 on 6 months of HFD was upregulated in inguinal WAT (iWAT), but less pronounced than in epiWAT, suggesting that RIPK3 upregulation in obesity occurs in a compartment-specic manner.

In addition, we performed haematoxylin and eosin stainings on the liver, skeletal muscle and epiWAT to examine whether differences in RIPK3 expression went along with clear histological alterations in these respective tissues. In liver tissue, we observed a trend towards higher steatosis in KO mice compared with WT animals (Fig. 3d), while no clear differences in muscle tissue were observed between both groups (Fig. 3d). Finally, absence of Ripk3 surprisingly was associated with a strong increase in focal areas of increased cellularity and density in epiWAT upon CD-HFD feeding compared with WT mice (Fig. 3d). This nding was not

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

associated with increased collagen col1a1 expression (Fig. 3e), which would indicate brosis, suggesting that the histological changes in the epiWAT of KO mice reected increased inammation.

Next, we performed immunohistochemical (IHC) stainings for immune cell subsets to characterize local inammatory responses in epiWAT of WT and KO mice. As shown in Fig. 4a, obese KO mice displayed multiple focal areas comprising numerous adipocytes surrounded by macrophages and B and T lymphocytes. Inammatory inltrates in the WAT of these mice comprised activated M1 and M2 macrophages as demonstrated by IHC stainings for F4/80 and CD206 (Fig. 4a). In line, blinded pathological examination revealed a signicantly increased inammatory score only in the WAT of obese KO mice compared with obese WT mice and lean WT and KO mice (Fig. 4b). Moreover, uorescence-activated cell sorting (FACS) analysis conrmed the increase in neutrophils, macrophages and natural killer (NK) cells in the WAT of obese KO mice compared with WT mice (Fig. 4c). Interestingly, it was recently shown that NK cells link obesity-induced adipose stress to inammation and insulin resistance27, suggesting that NK cells might contribute to glucose intolerance in KO mice. Correlating with increased inammation, we detected increased levels of TNF and monocyte chemoattractant protein-1 (MCP-1)a fundamental chemokine-regulating macrophage inltration into WAT28on qRTPCR in epiWAT of KO mice compared with WT mice (Fig. 4d). Interestingly, other cytokines such as interleukin (IL)-6, IL-1a, IL-1b and granulocyte

macrophage colony-stimulating factor (GM-CSF) were moderately upregulated in NCD-fed KO mice, but not in CDHFD-fed KO and WT mice (Fig. 4d). While these factors obviously are not involved in the mediation of the phenotype of obese RIPK3-KO mice, they indicate that lean KO mice have an aberrant cytokine pattern in their WAT, which does not induce a signicant phenotype. Of note, this difference in WAT inammation was not accompanied by signicant changes in systemic inammation; levels of circulating cytokines, such as TNF or IL-1b, were not altered between WT and KO mice after CD-HFD feeding (Supplementary Fig. 3). Together, these data suggest that Ripk3 deletion induced inammation in the adipose tissue of obese mice.

Increased adipocyte apoptosis in obese Ripk3-KO mice. It was suggested that adipocyte apoptosis is an important contributor to inammation and insulin resistance9,29. As such, genetic inactivation of the apoptosis mediator Bid prevented adipose tissue macrophage inltration and systemic insulin resistance in obese mice9. Thus, to analyse whether ablation of Ripk3 might be associated with increased adipocyte apoptosis in WAT upon CD-HFD feeding, we performed IHC stainings for the cleaved form of Caspase-3 (Cl-Casp-3), showing an increased presence of Cl-Casp-3 cells in CD-HFD-fed KO mice compared with WT mice (Fig. 5a). Apoptosis of adipocytes is associated with loss of the adipocyte-specic lipid droplet protein perilipin9,30. Therefore, to specically identify apoptotic adipocytes, we performed a double staining for cleaved Cl-Casp-3 and perilipin, revealing numerous regions of adipocytes that were devoid of perilipin labelling and had Cl-Casp-3 nuclei in KO mice when compared with WT mice (Fig. 5a). In addition, we performed a transmission electron microscopy analysis on the WAT of obese WT mice, which revealed that adipocytes showed ultrastructural signs of degeneration and cell death in CLSs (Supplementary Fig. 4). Together, these data suggest that, upon CD-HFD feeding, deletion of Ripk3 sensitizes WAT adipocytes to apoptosis in vivo.

To evaluate whether glucose intolerance and inammation in adipose tissue of KO mice are associated with the increased

presence of apoptosis in the WAT of KO mice, we next used mice with combined constitutive ablation of both Ripk3 and Caspase-8 (double-knockout (DKO))31 (Fig. 5b). DKO mice are viable but develop lymphadenopathy and splenomegaly because of impaired apoptosis signalling through the death receptor Fas3133. DKO mice showed similar body weight gain and cumulative food intake as age-, sex- and background-matched WT controls (Fig. 5c) and similar systemic levels of inammatory cytokines (Supplementary Fig. 5). Additional genetic inactivation of Caspase-8 protected KO mice from glucose intolerance and insulin resistance (Fig. 5d,e). In line, DKO displayed neither signicant inammation nor cleaved-Casp-3 cells in their WAT (Fig. 5f,g).

We aimed at providing further evidence that Ripk3 ablation might sensitize adipocytes towards apoptosis activation. For this, we used 3T3-L1 cells generated from murine broblasts that can be trans-activated into an adipocyte-like phenotype and are a well-established model system to study fat cells in culture34. Of note, stimulation of trans-differentiated 3T3-L1 cells with TNF alone or in combination with inhibitors of apoptosis (zVAD), RIPK1-dependent necroptosis (Nec-1s) and RIPK3 (GSK-872) did not induce any cell death (Supplementary Fig. 6a,b), suggesting that additional factors other than TNF putatively present in the peri-adipocyte milieu in obese WAT might be necessary for induction of adipocyte cell death. We therefore used another murine broblast cell line, L929, which is known to be sensitive to necroptosis upon TNF stimulation35. On TNF stimulation or in combination with zVAD, L929 cell death showed morphological signs of necroptosis but not apoptosis (Supplementary Movies 1 and 2). As shown in Supplementary Fig. 7, treatment of L929 cells with GSK-872 at a low dose (without spontaneous toxicity at this dose) induced apoptosis upon TNF treatment, which was rescued by additional zVAD treatment (Supplementary Fig. 7). Moreover, western blot analysis of epiWAT at 4 months of CD-HFD feeding showed that, similarly to RIPK3, Caspase-8 expression was strongly induced (Supplementary Fig. 6c), suggesting that absence of RIPK3 might tip the balance of cell death towards apoptosis. Finally, based on our in vivo data on RIPK3 upregulation in WT WAT, we engineered an adenoviral vector for overexpression of murine RIPK3 before and after trans-differentiation of 3T3-L1 cells and measured necroptosis using an antibody detecting the phosphorylated version of the executer MLKL (Supplementary Fig. 6d,e). Interestingly, phosphorylation of MLKL was only detected before trans-differentiation, arguing that RIPK3 can be overexpressed in adipocytes without cell toxicity (Supplementary Fig. 6d,e). Collectively, these ndings provide evidence that RIPK3 prevents glucose intolerance in obese mice by inhibiting Caspase-8-dependent adipocyte apoptosis in WAT.

The Ripk3-KO phenotype is not mediated by immune cells. Despite the results in DKO mice, it is possible that RIPK3 signalling in immune cells contributes to the metabolic phenotype of KO mice on CD-HFD feeding. As such, in CD-HFD-fed WT mice, macrophages in WAT could undergo necroptosis, which may limit inammation. To functionally test this hypothesis, we transferred WT bone marrow (BM) into irradiated KO mice (WT-RIPK3 / ) as well as the BM from Ripk3-KO mice into

WT mice (RIPK3 / -WT; Fig. 6a). As expected, isolated macrophages from the BM of RIPK3 / -WT mice at the end of CD-HFD feeding (18 weeks after bone marrow transfer (BMT)) showed no RIPK3 expression, while macrophages from WT-RIPK3 / mice showed RIPK3 (Fig. 6b). More importantly, all WT-RIPK3 / mice showed RIPK3 expression in their epiWAT on western blot analysis (Fig. 6b), arguing that WT

6 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869 ARTICLE

a

WT RIPK3/

WT RIPK3/

B220 CD4 F4/80 CD206

b c

NCD CD-HFD

Inflammatory score

Neutrophils

iMacrophages

10

10

10 KO WT KO

10

10

10

**

10

5

4

3

2

1

10

10

Ly6G

10

10

CD11b

10

10

10

10

10

10

10

10

10

10

10

10

10

10 10

10

10

10

10

10

10

10

10

10

10

10

CD11b

F4/80

WT RIPK3

40 35 30 25 15 10

5 0

CD-HFD

0 WT

RIPK3/

WT

RIPK3/

% of cells in epiWAT

related to CD45+ cells

10

10 WT KO

10

10

NCD CD-HFD

10 WT

10

10

*

10

10

CD4

10

10

10

10

*

10

10

10

10

10

10

10

10

NK1.1

Natural killer cells

Neutrophils

iM

NK cells CD4 CD8

d

#

Relative mRNA expression

30 25 20 15 10

5 4 3 2 1 0

WT KO WT KO WT KO WT KO WT KO WT KO WT KO

NCD

CD-HFD

**

* * * * *

#

*

TNF MCP-1 IL-6 IL-1 IL-1 IFN- GM-CSF

Figure 4 | RIPK3 overexpression in obese mice prevents epiWAT inammation. Data were obtained from RIPK3 / and WT mice fed with NCD and CD-HFD for different time points. All data are shown as means.e.m. (a) Representative images of immunohistochemical stainings for B220 , CD4,

F4/80 and CD206 cells in epiWAT from WT and KO mice fed for 16 weeks with NCD and CD-HFD. Scale bars, 100 mm. (b) Inammatory score of epiWAT of WT and KO mice fed for 16 weeks with NCD and CD-HFD. **Po0.01. H&E and F4/80 stains were evaluated blinded by an experienced pathologist as stated in the Methods section. Score 4, high inammation; score 1, no inammation. Differences between WTand KO mice were determined by ANOVA with Bonferronis post hoc test. (c) FACS data for intra-epiWAT levels of neutrophils, inammatory macrophages, natural killer cells, CD4-, CD8- and B-cells in WTand KO mice fed for 24 weeks with CD-HFD. n 6 for WTand n 8 for KO. *Po0.05. Differences between WTand KO mice were

determined by Students t-test. (d) TNF, MCP-1, IL-6, IL-1a, IL-1b, IFN-g and GM-CSF mRNA levels were assessed by RTPCR in WT and KO fed for 16 weeks with NCD or CD-HFD. Values were calculated relative to WTmice fed with NCD, and b-actin was used as an internal standard, n 6 per group. Differences

between groups were determined by ANOVA with Bonferronis post hoc test. y indicates that mRNA levels of TNF and MCP-1 are signicantly increased in KO mice compared with WT. # indicates that mRNA levels of MCP-1 in KO mice fed with CD-HFD were signicantly increased compared with all other groups. *Po0.05, **Po0.01.

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

a

WT RIPK3/

WT RIPK3/

Cl-Casp-3

*

* *

Cl-Casp-3

perilipin

* *

NCD

CD-HFD

b c

WT

WT

NCD

NCD

CD-HFD

epiWAT

Casp-8/ RIPK3/

120

Caspase-8/RIPK3

Caspase-8/RIPK3

WT

WT

Caspase-8/RIPK3

WT

RIPK3

Casp-8

GAPDH

CD-HFD + +

50

60

30

Body weight gain (%)

40

0

Cumulative food intake (g)

500

400

300

200

100

0

CD-HFD

(KDa)

Caspase-8/RIPK3

80

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16Time (week)

Time (week)

d e

Insulin (pg ml1)

C8/R3/

Blood glucose (mg dl1 )

800

600

400

200

0 0 15 30 60 90 120 150 180 Time

(min)

WT

*

***

NCD

CD-HFD

***

***

NS

NS NS

WT

Casp-8/RIPK3

Casp-8/RIPK3

***

8,000

6,000

4,000

2,000

***

0 WT C8/R3/

WTNCD CD-HFD

CD-HFD

f g

Casp-8/RIPK3/ Casp-8/RIPK3/

WT

**

Inflammatory score

4

3

2

1

Cl-Casp-3F4/80H&E

0 C8/R3/ C8/R3/

WT NCD

NCD CD-HFD

Figure 5 | Additional Caspase-8 deletion rescues the phenotype of Ripk3-KO mice. Data were obtained from Caspase-8/RIPK3 constitutive DKO (Caspase-8/RIPK3 / ) mice, from WTcontrol mice and from RIPK3 / /Caspase-8LPC-KO fed with NCD or CD-HFD for 16 weeks. Differences between

WT and DKO mice were determined by ANOVA with Bonferronis post hoc test. All data are shown as means.e.m. (a) Immunohistochemical analyses of cleaved-Caspase-3 and double stainings of cleaved Caspase-3 (brown stained) with perilipin (pink stained) in epiWAT. The asterisks designate dying adipocytes stained negatively for perilipin and positively for cleaved-Caspase-3. Scale bars, 100 mm. (b) Western blot analyses of RIPK3 and Caspase-8 in

WT and Caspase-8/RIPK3 / mice fed with NCD or CD-HFD for 16 weeks. GAPDH is used as loading control. (c) Relative body weight gain (%) of WT (n 6) and Caspase-8/RIPK3 / (n 7) mice (left panel) and analysis of cumulative food intake (g) in WT (n 6) and DKO (n 7) mice after 16 weeks

of feeding (right panel). (d) WT and Caspase-8/RIPK3 / mice were examined by GTT (n 5). ***Po0.001, n.s., not signicant. (e) Fasting blood

concentrations of insulin in NCD-fed WT (n 8) and Caspase-8/RIPK3 / (n 6) mice and in CD-HFD-fed WT (n 6) and Caspase-8/RIPK3 /

(n 7) mice. *Po0.05. (f) Representative images of immunohistochemical stainings for H&E, F4/80 and cleaved-Caspase-3 in epiWAT from WT

and Caspase-8/RIPK3 / mice. Scale bars, 100 mm. (g) Histological quantication of inammation from IHC stains of EpiWAT of WT and Caspase-8/RIPK3 / mice. H&E and F4/80 stains were evaluated blinded by an experienced pathologist as stated in the Methods section. Score 4: high inammation; score 1, no inammation. **Po0.01.

8 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869 ARTICLE

a b c

RIPK3/ WT

WT RIPK3/

RIPK3/ WT

WT RIPK3/

800

600

400

200

RIPK3/ WT

WT RIPK3/

CD-HFD

(kDa)

Threshold

**

WT

epiWATMacrophages

RIPK3

GAPDH

RIPK3

GAPDH

50

30

50 40

Blood glucose (mg dl1 )

+

***

bp 850 650 500 400 300 200 100

**

KO

WT

DNA from blood

Glycaemic response (AUC, 103)

120

100

80

60

0 0 15 30 60 90 120 150 180Time (Min)

d e f

RIPK3/ WT WT RIPK3/

**

2,500

2,000

1,500

500

0

Insulin (pg ml1)

H&E

F4/80

CI-Casp-3

**

1,000

RIPK3/ WT

WT RIPK3/

WT RIPK3/

RIPK3/ WT

WT RIPK3/

CD-HFD

g h

RIPK3/ WT WT RIPK3/

* *

CD-HFD

Inflammatory score

5 4

3 2

1 0

Circulating cytokine levels (pg ml1 )

200

150

100

50 30

20

10

0

TNF GMCSF

*

**

RIPK3/ WT

IL-1

IL-10 IL-2 IL-5 IFN

Figure 6 | The function of immune cells in the metabolic Ripk3-KO phenotype. Data were obtained from RIPK3 / mice reconstituted with WT bone marrow upon irradiation at the age of 6 weeks (WT-RIPK3 / ) and WT mice reconstituted with Ripk3-decient bone marrow (RIPK3 / -WT). Two weeks after BM transfer, mice were put on CD-HFD for 16 weeks. Differences between groups were determined by Students t-test. All data are expressed as means.e.m. (a) PCR analysis from whole-blood DNA using primers indicating the presence or absence (KO) of WT Ripk3. (b) Western blot analysis of RIPK3 from bone marrow-derived macrophages and adipose tissue of WT-RIPK3 / and RIPK3 / -WT animals. GAPDH is used as a control loading. (c) Mice were examined by GTT. **Po0.01, ***Po0.001. (d) AUC for the glycaemic response was calculated using the trapezoidal rule (n 4

in each group). **Po0.01. (e) Fasting blood concentrations of insulin in RIPK3 / -WT (n 6) and WT-RIPK3 / (n 7). **Po0.01. (f)

Representative images of histological and immunohistochemical stainings for H&E, F4/80 and Cl-Casp-3 cells in epiWAT from RIPK3 / -WT and WT-RIPK3 / animals. Arrowheads indicate cl-Casp-3 nuclei of adipocytes. Scale bars, 100 mm. (g) Histological quantication of inammation from the IHC stains H&E and F4/80 stains were evaluated blinded by an experienced pathologist as stated in the Methods section. Score 4: high inammation;

score 1, no inammation. *Po0.05. (h) Fasting serum concentrations of TNF, IL-1b, GM-CSF, IL-10, IL-2, IL-5 and IFN-g in CD-HFD-fed RIPK3 / -WT (n 6) and WT-RIPK3 / (n 7) mice were assessed using Bio-plex Assay. *Po0.05, **Po0.01.

bone-marrow-derived immune cells inltrated into the WAT of obese WT-RIPK3 / mice. Interestingly, WT-RIPK3 /

mice after 16 weeks of CD-HFD presented with a signicantly stronger glucose intolerance and higher insulin levels than RIPK3 / -WT animals (Fig. 6ce). In line with this, they

displayed more macrophage inltration and apoptosis in their WAT (Fig. 6f,g). On a systemic level, both BMT lines showed similar levels of inammatory cytokines (Fig. 6h), and TNF levels in RIPK3 / -WT were even slightly increased compared with

WT-RIPK3 / mice, while IL-10 showed a reciprocal

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

downregulation. Importantly, primary macrophages isolated from WT and KO mice did not show any difference in the expression of inammatory cytokines (TNF, IL-1a and IL-6) or Cyclin D1 when stimulated with TNF (Supplementary Fig. 8a,b), arguing against altered cytokine expression of Ripk3-decient macrophages. Moreover, supernatants from scratched cultured 3T3-L1 cells only without clearance of cell debris resulted in a similar transcriptional activation of IL-6 and Cyclin D1 (Supplementary Fig. 8a,b), suggesting that debris from dead adipocytes might be involved in the inammatory activation in the WAT of KO mice (Supplementary Fig. 8c). Together, these ndings provided evidence that adipocyte apoptosis is an important driver of glucose intolerance in CD-HFD-fed KO mice.

Ripk3-decient hepatocyte apoptosis in glucose intolerance. On the basis of these previous ndings, we next examined whether systemic glucose intolerance in KO mice was associated with increased signs of NAFLD and NASH. Blinded histological analysis (NAS score) revealed a nonsignicant trend to higher NAS score in KO mice compared with WT mice after 4 months of CD-HFD feeding (Supplementary Table 1). While, in marked difference to WAT, inammation was not increased in KO livers, they showed a higher lipid content than WT livers (Supplementary Table 1). Further analysis of serum markers of liver injury (aspartate aminotransferase (AST), alanine amino-transferase (ALT) and glutamate dehydrogenase (GLDH)) revealed a moderate increase in KO mice (Fig. 7a) that went along with increased numbers of hepatocytes that stained positive for the apoptosis marker cytokeratin-18-cleavage product M30 (Fig.7b). This nding indicated thatsimilarly to WATablation of Ripk3 led to a switch towards apoptosis in liver cells. On the basis of these ndings, we next tested whether additional deletion of Caspase-8 in hepatocytes (Caspase-8LPC-KO) could also rescue KO mice from glucose intolerance (RIPK3 / /Caspase-8LPC

KO). Importantly, this approach did not rescue glucose intolerance and insulin resistance of KO mice (Fig. 7c,d). Moreover, RIPK3 / /Caspase-8LPC-KO mice still showed the same level of

WAT inammation as single KO mice (Fig. 7e). These ndings suggested that Ripk3 ablation has a similar effect in adipocytes and hepatocytes upon CD-HFD feeding. However, our genetic and functional data indicated that the effects of Ripk3 deletion in WAT are the primary determinants of the metabolic phenotype on CD-HFD feeding.

RIPK3 is overexpressed in WAT of obese humans and mice. We nally assessed whether RIPK3 expression is also altered in the visceral fat tissue (visWAT) of human patients with obesity and T2D (Supplementary Table 2). Similarly to obese mice, we found an upregulation of RIPK3 in the visWAT of obese patients with and without T2D compared with non-obese controls (Fig. 8a,b). We also evaluated whether RIPK3 upregulation in the visWAT of obese and diabetic patients was associated with activation of necroptosis. To test this, we performed western blot analyses with an antibody against phosphorylated MLKL, which is considered to be the best available biomarker for activation of necroptosis36. Interestingly, MLKL phosphorylation closely followed RIPK3 overexpression in the visWAT of these patients (Fig. 8c,d and Supplementary Fig. 9a) and it also showed a correlation with metabolic serum markers (Supplementary Fig. 9b), suggesting that necroptosis is activated in the WAT of obese humans. Of note, RIPK3 expression correlated with metabolic serum markers, such as HbA1c and levels of circulating insulin, in these patients (Supplementary Fig. 9c); however, it is presently unclear whether the downstream effects of RIPK3 inhibition in humans would be similar to our ndings in

mice, requiring further investigations. Given the upregulation that we had detected in western blot analyses on WAT of obese humans and mice, we nally wanted to conrm histologically in murine WAT that adipocytes can upregulate RIPK3. As hypothesized, this analysis revealed high expression of RIPK3 (Fig. 8e) in WT mice, suggesting that, next to apoptosis, obesity triggers activation of RIPK3-dependent signalling in adipocytes.

DiscussionOur data in obese mice suggest that RIPK3 in adipose tissue maintains homeostasis and dampens inammation by inhibiting Caspase-8-dependent apoptosis of adipocytes, thereby preventing glucose intolerance (Fig. 9). In many tissues, RIPK3 upregulation is known to sensitize cells to necroptosis, which triggers inammation37,38. However, it has been suggested that in certain tissues and disease models necroptosis might not be associated with induction of inammation3941 or might even inhibit the pro-inammatory response to apoptosis42. Thus, apoptosis and necroptosis might either trigger or block inammation, depending on the cell type and cell death stimulus, and specic forms of programmed cell death in adipose tissue have opposing inuences on systemic glucose homeostasis.

Importantly, our ndings together with previous studies suggest that different forms of regulated cell death can be detected simultaneously in the WAT of obese mice and humans8,9. The nding that Ripk3 deletion triggers increased apoptosis in mice is unexpected and might be related to the specic trigger of cell death (triglyceride overload) or the specic cell type (adipocytes). Of note, it was shown previously that mice with constitutive ablation of Ripk1 display spontaneous apoptosis in the lymphoid and adipose tissue43, suggesting that adipocytes might be very sensitive to a switch towards apoptosis on genetic or pharmacological modulations of RIPK3 (ref. 44). Together, these data support the hypothesis that RIPK3 can have an anti-apoptotic function in certain circumstances, which might be of relevance when considering RIPK3 inhibition as a pharmacological strategy for treatment of necroptosis-related diseases. Interestingly, our data pointed towards an age-related expression pattern of RIPK3, with highest expression at 4 months of CD-HFD feeding. Age-specic differences in WAT expression have been shown for numerous genes45. Of note, expression of Caspase-8 remained high even at older age (Supplementary Fig. 10), suggesting that loss of the protective effect of even lower RIPK3 levels might be sufcient to trigger apoptosis also at that age.

In addition to adipocyte apoptosis, we detected at a lower extent the apoptosis of innate immune cells in the inamed WAT of obese mice as demonstrated by IHC (nuclei of immune cells stained for Cl-Casp-3 (Fig. 5)) and transmission electron microscopy (macrophage with cytoplasmic signs of degeneration (Supplementary Fig. 4)). While our BMT experiments argued against a primary role of immune cells in the phenotype, it was previously demonstrated in models of atherosclerosis that macrophages can form foam cells containing lipid droplets that can undergo necroptosis as well as apoptosis16,46,47, thereby aggravating and perpetuating the inammatory process. In line, our RIPK3-western blots in the WAT of mice that had undergone BMT (Fig. 6b) conrmed that RIPK3 was expressed in both adipocytes and immune cells. A similar process might therefore occur in the WAT of KO mice. Importantly, it was previously shown that adipocyte death is sufcient to initiate macrophage inltration48, supporting our data that adipocyte apoptosis in KO mice is the main trigger for inammation, which can be blunted by additional deletion of Caspase-8 (Fig. 5). In the same line, it was previously suggested that induction of fat cell apoptosisfor example, by applying Leptin, TNF or natural compoundsmight

10 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869 ARTICLE

*** *** ***

*

*

a

150

100

50

0

**

AST (IU l1 )

400

300

200

100

0

AST (IU l1 )

GLDH (IU l1 )

WT

RIPK3/

WT

RIPK3/

NCD CD-HFD

400

300

200

100

0

WT

RIPK3/

WT

RIPK3/

NCD CD-HFD

WT

RIPK3/

WT

RIPK3/

NCD CD-HFD

b

(per 20x magnification)

WT RIPK3/ WT RIPK3/

M30 positive cells

10

8 6

4 2

M30

ND ND

*

NCD CD-HFD

0 WT

RIPK3/

WT

RIPK3/

NCD CD-HFD

c d

WT NCD WT

800

600

400

200

GTT

CD-HFD

Threshold

RIPK3//Caspase-8LPC-KO

Insulin (pg ml1)

*

Blood glucose (mg dl1 )

20,000

15,000

10,000

5,000

0

NCD

***

***

**

***

***

***

***

***

WT WT

CD-HFD

0 0 15 30 60 90 120 150 180

Time (min)

e

WT WT

RIPK3//Casp-8LPC-KO

R3//C8LPC-KO

H&E

NCD CD-HFD

Figure 7 | Increased liver injury and apoptosis in Ripk3-decient livers. Data were obtained from WT, KO and RIPK3 / /Caspase-8LPC-KO mice fed with NCD or CD-HFD for 16 weeks. Differences between the mice were determined by ANOVA with Bonferronis post hoc test. All data are shown as means.e.m. (a) Analysis of serum levels of AST, ALT and GLDH after 16 weeks of NCD or CD-HFD for WT and KO mice (n 11). *Po0.05,

**Po0.01, ***Po0.001. (b) Left panel: representative images of immunohistochemical staining for M30 cells in the liver from WT and KO mice fed for 16 weeks with NCD and CD-HFD. Scale bars, 100 mm. Right panel: statistical analysis of M30-positive hepatocytes (n 6). *Po0.05, ND, non detectable.

(c) RIP3 / /Caspase-8LPC-KO mice fed with CD-HFD were examined by GTT and compared with WT animals fed either with NCD or CD-HFD (n 5).

**Po0.01, ***Po0.001. (d) Fasting serum concentrations of insulin in NCD-fed WTmice (n 9) and in CD-HFD-fed WTand RIPK3 / /Caspase-8LPC-KO

mice (n 5) mice. *Po0.05. (e) Representative H&E images of epiWAT from WT animals fed with NCD and WT and RIPK3 / /Caspase-8LPC-KO mice

fed with CD-HFD for 16 weeks. Scale bars, 100 mm.

represent a novel therapeutic strategy against obesity and a realistic alternative to caloric restriction in order to achieve fat loss49. However, our present ndings in Ripk3-decient mice underline that activation of apoptosis in WAT adipocytes is linked with detrimental unwanted metabolic side effects including inammation and glucose intolerance. It is presently unclear and might be interesting to determine whether activation of necroptosis might be an alternative in this respective setting. It is

unexpected that the strong local differences in WAT inammation were not associated with a marked difference in the systemic inammatory markers between WT and KO mice (Supplementary Fig. 3). However, it is possible that these differences might get clearer at later time points of CD-HFD or that alternative factors and adipokines other than the ones we measured connected local WAT inammation with systemic glucose intolerance.

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

a b

** r = 0.4679

P < 0.01

r = 0.5489 P < 0.01

6

4

2

6

4

2

0 0 20 40 60 80 BMI

(kDa) Lean Obese T2D

50

30

RIPK3 expression

p-MLKL expression

0 Lean Obese T2D

0 Lean Obese T2D

c d

(kDa) Lean Obese T2D

50

30

RIPK3

GAPDH

GAPDH

p-MLKL protein/GAPDH

(relative to lean)

8

6

4

2

* 8

6

4

2

0 0 20 40 60 80 BMI

p-MLKL

RIPK3 protein/GAPDH

(relative to lean)

e

WT

RIPK3

NCD CD-HFD

Figure 8 | Activation of RIPK3-dependent signalling pathways in obesity. Data were obtained from lean controls, obese non-diabetic patients and obese patients with T2D. WT mice were fed with CD-HFD for 4 months. Differences between groups were determined by ANOVA with Bonferronis post hoc test and correlations were assessed by non-parametric Spearmans test. All data are shown as means.e.m. (a) Representative western blot analysis of RIPK3 on protein lysates of the visWAT from lean controls and obese and T2D patients (n 5 for each group). GAPDH is used as a loading control. (b) Left panel:

relative protein quantication of RIPK3 levels in the different groups of human samples (n 10 for each group in total), standardized to the loading control

GAPDH. **Po0.01. Right panel: correlation analysis between RIPK3 protein levels and body mass index (BMI). **Po0.01. (c) Representative western blot analysis of p-MLKL in the visWAT of lean controls and obese and T2D patients (n 5 for each group). GAPDH is used as a loading control. (d) Left panel:

relative protein quantication of p-MLKL levels in the different groups of human samples (n 10 for each group), standardized to the loading control

GAPDH. *Po0.05. Right panel: correlation analysis between p-MLKL protein levels and the BMI. **Po0.01. (e) Representative images of immunohistochemical stainings for RIPK3 in epiWAT from WT mice fed 16 weeks with NCD and CD-HFD. Scale bars, 100 mm.

Our ndings on the association between RIPK3 overexpression and obesity and metabolic serum parameters in humans clearly indicated that RIPK3 has a similar function in the WAT of obese humans and mice. Of note, we showed that RIPK3 overexpression in human patients strictly correlated with phosphorylation of MLKL, the direct target of RIPK3 mediating necroptosis. It was previously shown that adipocyte apoptosis is markedly increased in omental fat from obese subjects, correlating with the magnitude of adipose tissue inltration by macrophages and systemic markers of insulin resistance9. Moreover, it is known that TNF is strongly expressed in adipocytes of obese subjects and correlates with body mass index (BMI)50, providing further evidence for the hypothesis that necroptosis in WAT might primarily act to limit immune responses triggered by inammatory cytokines such as TNF or initiated and aggravated by adipocyte apoptosis9.

We and others recently showed that in patients with NASH an important complication of obesity and major risk factor for the development of liver cirrhosis and hepatocellular carcinoma in western industrialized countries51,52hepatocytes with the phosphorylated form of MLKL can be detected53,54. Given that

pharmacological inhibition of apoptosis was even tested in clinical trials with NASH patients55, targeting of necroptosis appeared a promising new strategy in this respective clinical context. However, our present ndings on the unexpected role of RIPK3 in WAT suggest that systemic pharmacological targeting of RIPK3 in NASH patientsas well as in other necroptosis-related diseases10might cause or aggravate glucose intolerance and insulin resistance in obese individuals.

Methods

Human material. Samples of visceral (omental) adipose tissue were obtained from 30 individuals (19 women and 11 men; Supplementary Table 2). Their age ranged from 22 to 78 years and the BMI from 16 to 70.3 kg m 2. The entire cohort was subdivided into subgroups of healthy lean (BMIo25 kg m 2; n 10) individuals

and patients with obesity (BMI430 kg m 2; n 20) either with (n 10) or

without (n 10) T2D. The latter two subgroups were matched for age and BMI. All

adipose tissue samples were collected during open or laparoscopic abdominal surgery as described previously56. Adipose tissue was immediately frozen in liquid nitrogen and stored at 80 C. The study was approved by the Ethics Committee

of the University of Leipzig (approval no: 159-12-21052012), and was performed in accordance to the declaration of Helsinki. All subjects gave written informed consent before taking part in this study. Measurement of body composition,

12 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869 ARTICLE

Weight gain

Lean adipose tissue Obese adipose tissue

WT adipocyte

Blood vessel

Blood vessel

Glucose tolerance

0

Glucose tolerance

0

RIPK3

RIPK3

Casp8

Casp8

RIPK3

RIPK3-KO adipocyte

RIPK3

Casp8

Death

RIPK3

Casp8

M1-macrophages

Foam/dying macrophages

Figure 9 | RIPK3 dampens WATapoptosis and systemic glucose intolerance. In WTmice, weight gain induced overexpression of RIPK3 in adipocytes that counterbalanced the activation of Caspase-8-dependent apoptosis, thereby dampening WAT inammation and glucose intolerance.

WT

Adipocytes

Apoptotic adipocytes

RIPK3-KO

Adipocytes

Apoptotic adipocytes

metabolic parameters and adipose tissue-related parameters was performed as described previously56.

Animals. Ripk3-decient mice were obtained from Genentech20, and Caspase-8/Ripk3-decient mice were described previously31. For the breeding of Ripk3-KO- and WT mice, heterozygous co-founders (C57BL/6) were interbred.

/ and / mice from the F1 generation were separated and interbred in

parallel. The phenotype was analysed from the F4 generation on. Mice were housed under exactly the same conditions (cage inlays were exchanged between cages once a week during the experiments). Mice with constitutive deletion of Ripk3 and conditional deletion of Caspase-8 in the liver (liver parenchymal cell knockout (LPC-KO)) were described previously14. For these, age- and sex-matched littermates were used as controls. Animals were allocated to the experimental groups based on their genotypes. All animal experiments were approved by the Federal Ministry for Nature, Environment and Consumers Protection of the state of North Rhine-Westphalia and were performed in accordance to the respective national, federal and institutional regulations.

Diet. In all experiments, 6-week-old male mice were fed with a CD-HFD (Research Diets; D05010402) or with a NCD for varying time intervals as indicated. Mice were single housed, and food intake and weight body gain were measured weekly by weighing food hoppers and animals, respectively. At the end of the feeding protocol, mice were fasted at least for 6 h before being euthanized. Similarly, in Fig. 3c, fat samples were gained from mice that had been fed for 6 months with NCD or conventional HFD (Research Diets; D12492).

Faecal Microbiota Analysis. Faecal samples were collected from mice with different genotype, diet and time points. DNA from B20 mg faecal pellets was isolated with the QIAamp DNA Stool Mini Kit (QIAGEN) as instructed by the manufacturer with additional bead beating steps. For in-depth microbiota proling, 16S rRNA gene V4 region amplicons from faecal DNA isolates were amplied by the 515F/806R primers and sequenced on an Illumina MiSeq platform using2 300 bp according to the established protocols57. Data were analysed using the

QIIME software pipeline (v1.9.1)58. The UCLUST algorithm was used with default variables to cluster the sequences into operational taxonomic units based on 97%

identity against the Greengenes reference database (version 13_8)59. The diversity between samples was measured by the weighted UniFrac distance, and a principal coordinate analysis of the weighted UniFracs was calculated and constructed with QIIME.

Imaging. In vivo mCT imaging of WT and RIPK3 / mice fed with NCD or CD-HFD (n 68) for 16 weeks was performed using a dual-energy gantry-based

at-panel microcomputed tomography scanner (TomoScope 30s Duo, CT Imaging, Erlangen, Germany)60. The dual-energy X-ray tubes of the mCT were operated at voltages of 40 and 65 kV with currents of 1.0 and 0.5 mA, respectively. To cover the entire mouse, three sub-scans were performed, each of which acquired 720 projections with 1,032 1,012 pixels during one full rotation with durations of

90 s. Animals were anaesthetized using 2% isourane in air for the entire imaging protocol (ow rate 1 l min 1). After acquisition, volumetric data sets were reconstructed using a modied Feldkamp algorithm with a smooth kernel at an isotropic voxel size of 35 mm. The fat-containing tissue regions, which appear hypo-intense in the mCT data, were segmented using an automated segmentation method with interactive correction of segmentation errors61,62. The volumetric fat percentage was computed as the ratio of (subcutaneous and visceral) fat volume to the entire body volume.

L929 cells were seeded in standard 12-well cell culture plates with 1.5 ml of DMEM and were subjected to in vitro time-lapse microscopy, using an AxioObserver Z1 (Zeiss) modied with large chamber cell culture incubation unit (Pecon). After 1 h of live imaging, cells were stimulated by adding zVAD (20 mM)

or dimethylsulphoxide for 1 h and then TNF (20 ng ml ). Cells were followed over a time period of up to 78 h, taking images from eight different target positions within the well every 3 min. Cells were kept at 37 C and 5% CO2 atmosphere over the time course of the experiment. Time-lapse movies were generated using ZEN Blue 2.0 (Zeiss).

Metabolic studies. Mice were fasted overnight for 16 h. Then, 2 mg of glucose per g body weight was intraperitoneally injected and tail blood was taken at the indicated time points. Glucose levels were determined using a hand-held glucose analyser (Contour XT Bayer). Insulin (1 U kg 1) was intraperitoneally injected for 10 min. Mice were then euthanized and tissues were collected for western blot

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

analysis. Insulin tolerance tests were performed on tail blood taken at the indicated time points.

Serum analysis. Insulin and ghrelin levels were measured using suspension bead array immunoassay kits following the manufacturers specications (Bio-Plex pro Mouse Diabetes Assay 8-plex, Bio-Rad) on a Luminex 100/200 analyser. Bio-Plex Pro Mouse Cytokine GrpI panel 8-plex kit was used to measure TNF, IL-1b, IL-2,

IL-5, IL-10, GM-CSF and interferon (IFN)-g in the serum of mice. The data were analysed with Bio-Plex Manager Instrument Control Version 6 (Bio-Rad). Measurement of body composition, metabolic parameters and adipose tissue-related parameters from human patients was performed as described previously56.

BM transplantation. BM from WT and RIPK3 / donors (n 4 mice per

group) was transplanted into 6-week-old RIPK3 / and WT recipient mice, respectively. After ablative g-irradiation of the recipient mice, 2.5 106 of the BM

cells from donors were injected via the tail vein. After BM transfer, mice were treated with antibiotics (0.02% Borgal) for 2 weeks before the feeding experiments were started. At the end of the feeding experiment, DNA from the blood of recipient mice was extracted using the QIAamp kit (Qiagen) and analysed for conrming the success of the BM transfer.

Western blot analysis. Tissue samples were homogenized in NP-40 lysis buffer using a tissue grind pestle (Kimble/Chase) or with a bead ruptor 12 (Omni International) to obtain protein lysates. These were separated by SDSPAGE, transferred to polyvinylidene diuoride membrane and analysed by immunoblotting. Membranes were probed with the following antibodies: anti-p-AKT (#4060), anti-AKT (#4685), anti-p-GSK-3 (#8566), anti-GSK-3 (#5676), anti-p-ERK (#4377) and anti-ERK (#4695; Cell Signaling), anti-Caspase-8 (Enzo #ALX-804-447), anti-RIPK3 mouse (IMGENEX #IMG-5523), anti-RIPK3 human (#ab56164), anti-phospho-MLKL human (#ab187091) and mouse (#ab196436; Abcam) and anti-GAPDH (ABD Serotec #MCA4739). All primary antibodies were used at the dilution 1:2,000. As secondary antibodies, anti-rabbit-horseradish peroxidase (HRP; #NA934V) and anti-mouse-HRP (#NA931V; Amersham) and anti-rat-HRP (Santa Cruz #sa2956) were used. All secondary antibodies were used at the dilution 1:5,000. Unedited scans of all western blot images are shown in Supplementary Fig. 11.

Quantitative real-time PCR. Total RNA was puried from fat tissue using Trizol reagent (Invitrogen) and an RNeasy Mini Kit (Qiagen). The quantity and quality of the RNA was determined spectroscopically using a nanodrop (Thermo Scientic). Total RNA (0.5 mg) was used to synthesize cDNA using the Transcriptor cDNA

First-Strand Synthesis Kit (Roche) according to the manufacturers protocol. cDNA samples (2 ml) were used for real-time PCR in a total volume of 25 ml using SYBR

Green Reagent (Invitrogen) and specic primers on a qPCR machine (Applied Biosystems 7,300 Sequence Detection System). All real-time PCR reactions were performed in duplicates. Data were generated and analysed using the SDS 2.3 and RQ manager 1.2 software. Primer sequences are available on request. All values were normalized to the level of beta-actin mRNA. The expression of TNF, MCP-1, IL-6, IL-1a, IL-1b, CycD1, IFN-g, GM-CSF and col1a1 was tested using the primers as follows: TNF: 50-ACCACGCTCTTCTGTCTACTGA-30 (for), 50-TCC ACTTGGTGGTTTGCTACG-30 (rev); MCP-1: 50-GTGTTGGCTCAGCCAGAT GC-30 (for), 50-GACACCTGCTGCTGGTGATCC-30 (rev); IL-6: 50-GCTACC AAACTGGATATAATCAGGA-30 (for), 50-CCAGGTAGCTATGGTACTCCAG AA-30 (rev); IL-1a: 50-GCACCTTACACCTACCAGAGT-30 (for), 50-AAACTT CTGCCTGACGAGCTT-30 (rev); IL-1b: 50-GCAACTGTTCCTGAACTCAACT-30 (for), 50-ATCTTTTGGGGTCCGTCAACT-30 (rev); CycD1: 50-GCGTACCCTG ACACCAATCTC-30 (for), 50-CTCCTCTTCGCACTTCTGCTC-30 (rev); IFN-g: 50-ATGAACGCTACACACTGCATC-30 (for), 50-CCATCCTTTTGCCAGT TCCTC-30 (rev); GM-CSF: 50-GGCCTTGGAAGCATGTAGAGG-30 (for), 50-G GAGAACTCGTTAGAGACGACTT-30 (rev); col1a1: 50-GCTCCTCTTAGG GGCCACT-30 (for), 50-CCACGTCTCACCATTGGGG-30 (rev).

Histological examination and evaluation. Parafn sections (2 mm) were stained with haematoxylin and eosin or various primary and secondary antibodies. Paraformaldehyde (4%) xed and parafn embedded liver, skeletal muscle and epiWAT were incubated in Bond Primary antibody diluent (Leica) and stainings were performed on a BOND-MAX immunohistochemistry robot (Leica Biosystems) using BOND polymer rene detection solution for DAB. The following antibodies were used: anti-F4/80 (BMA Biomedicals AG, 1:120), anti-CD206 (AbD Serotec 1:200), anti-B220 (BD Biosciences; 1:3,000), anti-CD4 (eBioscience; 1:1,000), anti-cl-Casp-3 (Cell Signaling; 1:300), anti-perilipin(RDI Division of Fitzgerald, 1:1,000), anti-RIPK3 (Abcam; 1:500) and anti-p-MLKL (Abcam; 1:500). Image acquisition was performed on an Olympus BX53 microscope with a Leica SCN400 slide scanner. Stains were evaluated blinded by an experienced pathologist and inammatory scores were determined using the following system: 0 basically no inammation, 1r400-fold eld of view,

2 400200-fold eld of view, 3Z200100-fold eld of view, 4 up to 40-fold

eld of view. The histological scoring system for non-alcoholic fatty liver disease (NAFLD) was performed according to the NAS score system63.

Transmission electron microscopy. Small adipose tissue fragments from lean and obese mice were xed in 2% gluteraldheyde2% paraformaldehyde in phosphate buffer for 4 h at room temperature, and then post-xed in 1% osmium tetroxide and embedded in an Epon-Araldite mixture. Semithin sections (2 mm) were stained with toluidine blue. Thin sections were obtained with an MT-X ultratome (RMC; Tucson, AZ), stained with lead citrate and examined with a CM10 transmission electron microscope (Philips; Eindhoven, the Netherlands).

Cell isolation and ow cytometry. To isolate cells from WAT, the tissue was minced into small pieces sizing below 1 mm. Lymph nodes were excised before mincing. Next, the tissue was digested for 30 min with 2% collagenase IV (Worthington) in Hanks buffered salt solution (HBSS). HBSS was supplemented with 5 mM ethylene-diamine-tetra-acetic acid and 0.5% bovine serum albumin to stop the collagenase activity. To prepare for ow cytometry, cells were ltered using a 100-mm mesh. Staining was performed using the following antibodies: NK1.1 (BioLegend), CD8a, CD4, F4/80 and CD11b (eBioscience), CD45, Gr1/Ly6C and Ly6G (BD Biosciences). Before analysis, count beads (Allophycocyanine calibrite beads, BD Biosciences) were added to calculate cell numbers. Flow cytometry was carried out using a FACS Canto II (BD Biosciences). Flow cytometric data are given as percentage of cells related to the number of CD45 cells. Data were analysed with the FlowJo software (TreeStar Inc.).

Cell culture. L929 (ATCC CCL-1) cells were cultured in DMEM supplemented with 10% fetal calf serum (FCS), penicillin (100 IU ml 1), streptomycin(0.1 mg ml 1) and L-glutamine (0.03%). 3T3-L1 CARD1 cells64 were maintained at a non-differentiated state in DMEM supplemented with 10% bovine calf serum, and the medium was changed every 2 days without ever reaching conuence. Adipocyte conversion was induced by treating 2-day post-conuent cultures with DMEM supplemented with 10% FCS and dexamethasone, isobutylmethylxanthine and insulin according to the manufacturers specications (Differentiation Kit DIF001 Sigma). Cells were treated by zVAD (Merck Millipore20 mM), Nec-1s (Santa Cruz10 mM) and GSK-872 (Merck Millipore3 mM).

For isolation of macrophages, BM cells were isolated from the femur and tibia of 16-week-old C57BL6 and RIPK3 / mice. To obtain broblast-conditioned medium, which is known to contain the macrophage colony-stimulating factor (CSF1), L929 broblasts were cultured in RPMI medium containing 10% FCS for 3 days, and the supernatant was collected, ltered and stored until usage at 80 C.

BM cells were cultured in RPMI medium containing 10% FCS and 20% broblast-conditioned medium for 1 week on bacterial grade plastic plates (Greiner). On day 7, cells were left untreated, stimulated for additional 16 h with TNF (20 ng ml 1)

or placed with medium containing adipocyte debris or cleared from them. Allin vitro experiments are representative of three independent experiments. All cell lines were tested and were free of mycoplasma infection.

Generation of a recombinant adenovirus encoding RIPK3. Cloning of mouse full-length RIP3 (NCBI Reference Sequence: NM_019955.2) into a pDonR201 gateway vector (Invitrogen) was carried out by PCR. Generation of an adenovirus encoding full-length RIP3 cDNA was carried out using the ViraPower Adenoviral Expression System (Invitrogen) according to the manufacturers protocol. A lacZ encoding adenovirus served as control.

Statistical analysis and general experimental design. We calculated sample size using size power analysis methods (GraphPad StatMate) for a priori determination based on the s.d. of previous experiments in WT mice. We calculated the minimal sample size for each group as seven animals. For some experiments, fewer animals were found sufcient to obtain statistical differences. In the serum analyses, at some time points all results from different longitudinally followed groups of mice were combined. Animals with same sex and same age were employed to minimize physiological variability and to reduce s.d. from the mean. The exclusion criteria for animals were established in consultation with a veterinarian and on the basis of experimental outcomes. In case of death or sickness, the animal was excluded from analysis. Tissue samples were excluded in cases of failure in extraction of RNA or protein of suitable quality and quantity. Animal experiments were blinded in terms of genetic background of mice by using ear number codes during the CD-HFD-feeding periods. Statistical tests were used as described in the Figure legends. Statistical analyses were performed using the GraphPad Prism software(version 5.0). All data are presented as means.e.m. and were analysed by analysis of variance (ANOVA) with Bonferronis post hoc multiple comparison test. Analysis of two groups of samples was performed using Students t-test. Correlations were assessed by non-parametric Spearmans test. Statistical signicance was indicated as follows: ***Po0.001; **Po0.01; *Po0.05; n.s., not signicant.

Data availability. The authors declare that the data supporting the ndings of this study are available within the article and its Supplementary Information les.

14 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869 ARTICLE

References

1. Hameed, I. et al. Type 2 diabetes mellitus: from a metabolic disorder to an inammatory condition. World J. Diabetes 6, 598612 (2015).

2. Kahn, S. E., Cooper, M. E. & Del Prato, S. Pathophysiology and treatment of type 2 diabetes: perspectives on the past, present, and future. Lancet 383, 10681083 (2014).

3. Scherer, P. E. Adipose tissue: from lipid storage compartment to endocrine organ. Diabetes 55, 15371545 (2006).

4. Capurso, C. & Capurso, A. From excess adiposity to insulin resistance: the role of free fatty acids. Vasc. Pharmacol. 57, 9197 (2012).

5. Feng, D. et al. High-fat diet-induced adipocyte cell death occurs through a cyclophilin D intrinsic signaling pathway independent of adipose tissue inammation. Diabetes 60, 21342143 (2011).

6. Musi, N. & Goodyear, L. J. Insulin resistance and improvements in signal transduction. Endocrine 29, 7380 (2006).

7. Murano, I. et al. Dead adipocytes, detected as crown-like structures, are prevalent in visceral fat depots of genetically obese mice. J. Lipid Res. 49, 15621568 (2008).

8. Cinti, S. et al. Adipocyte death denes macrophage localization and function in adipose tissue of obese mice and humans. J. Lipid Res. 46, 23472355 (2005).

9. Alkhouri, N. et al. Adipocyte apoptosis, a link between obesity, insulin resistance, and hepatic steatosis. J. Biol. Chem. 285, 34283438 (2010).

10. Silke, J., Rickard, J. A. & Gerlic, M. The diverse role of RIP kinases in necroptosis and inammation. Nat. Immunol. 16, 689697 (2015).

11. Ashkenazi, A. & Salvesen, G. Regulated cell death: signaling and mechanisms. Annu. Rev. Cell Dev. Biol. 30, 337356 (2014).

12. Sun, L. et al. Mixed lineage kinase domain-like protein mediates necrosis signaling downstream of RIP3 kinase. Cell 148, 213227 (2012).

13. Newton, K. RIPK1 and RIPK3: critical regulators of inammation and cell death. Trends Cell Biol. 25, 347353 (2015).

14. Gautheron, J. et al. A positive feedback loop between RIP3 and JNK controls non-alcoholic steatohepatitis. EMBO Mol. Med. 6, 10621074 (2014).15. Roychowdhury, S., McMullen, M. R., Pisano, S. G., Liu, X. & Nagy, L. E. Absence of receptor interacting protein kinase 3 prevents ethanol-induced liver injury. Hepatology 57, 17731783 (2013).

16. Lin, J. et al. A role of RIP3-mediated macrophage necrosis in atherosclerosis development. Cell Rep. 3, 200210 (2013).

17. Vitner, E. B. et al. RIPK3 as a potential therapeutic target for Gauchers disease. Nat. Med. 20, 204208 (2014).

18. Linkermann, A. et al. Two independent pathways of regulated necrosis mediate ischemia-reperfusion injury. Proc. Natl Acad. Sci. USA 110, 1202412029 (2013).

19. Luedde, M. et al. RIP3, a kinase promoting necroptotic cell death, mediates adverse remodelling after myocardial infarction. Cardiovasc. Res. 103, 206216 (2014).

20. Newton, K., Sun, X. & Dixit, V. M. Kinase RIP3 is dispensable for normal NF-kappa Bs, signaling by the B-cell and T-cell receptors, tumor necrosis factor receptor 1, and Toll-like receptors 2 and 4. Mol. Cell Biol. 24, 14641469 (2004).

21. Wolf, M. J. et al. Metabolic activation of intrahepatic CD8 T cells and NKT

cells causes nonalcoholic steatohepatitis and liver cancer via cross-talk with hepatocytes. Cancer Cell 26, 549564 (2014).22. Raubenheimer, P. J., Nyirenda, M. J. & Walker, B. R. A choline-decient diet exacerbates fatty liver but attenuates insulin resistance and glucose intolerance in mice fed a high-fat diet. Diabetes 55, 20152020 (2006).

23. Linkermann, A. et al. Synchronized renal tubular cell death involves ferroptosis. Proc. Natl Acad. Sci. USA 111, 1683616841 (2014).

24. Lin, Y. & Sun, Z. Current views on type 2 diabetes. J. Endocrinol. 204, 111 (2010).

25. Shao, J., Yamashita, H., Qiao, L. & Friedman, J. E. Decreased Akt kinase activity and insulin resistance in C57BL/KsJ-Leprdb/db mice. J. Endocrinol. 167, 107115 (2000).

26. Nathan, D. M. et al. Impaired fasting glucose and impaired glucose tolerance: implications for care. Diabetes Care 30, 753759 (2007).

27. Wensveen, F. M. et al. NK cells link obesity-induced adipose stress to inammation and insulin resistance. Nat. Immunol. 16, 376385 (2015).

28. Suganami, T. & Ogawa, Y. Adipose tissue macrophages: their role in adipose tissue remodeling. J. Leukoc. Biol. 88, 3339 (2010).

29. Grant, R. W. & Stephens, J. M. Fat in ames: inuence of cytokines and pattern recognition receptors on adipocyte lipolysis. Am. J. Physiol. Endocrinol. Metab. 309, E205E213 (2015).

30. Herrero, L., Shapiro, H., Nayer, A., Lee, J. & Shoelson, S. E. Inammation and adipose tissue macrophages in lipodystrophic mice. Proc. Natl Acad. Sci. USA 107, 240245 (2010).

31. Oberst, A. et al. Catalytic activity of the caspase-8-FLIP(L) complex inhibits RIPK3-dependent necrosis. Nature 471, 363367 (2011).

32. Kaiser, W. J. et al. RIP3 mediates the embryonic lethality of caspase-8-decient mice. Nature 471, 368372 (2011).

33. Newton, K. et al. Activity of protein kinase RIPK3 determines whether cells die by necroptosis or apoptosis. Science 343, 13571360 (2014).

34. Green, H. & Meuth, M. An established pre-adipose cell line and its differentiation in culture. Cell 3, 127133 (1974).

35. Vercammen, D. et al. Inhibition of caspases increases the sensitivity of L929 cells to necrosis mediated by tumor necrosis factor. J. Exp. Med. 187, 14771485 (1998).

36. Wang, H. et al. Mixed lineage kinase domain-like protein MLKL causes necrotic membrane disruption upon phosphorylation by RIP3. Mol. Cell 54, 133146 (2014).

37. Pasparakis, M. & Vandenabeele, P. Necroptosis and its role in inammation. Nature 517, 311320 (2015).

38. Moriwaki, K. & Chan, F. K. Necrosis-dependent and independent signaling of the RIP kinases in inammation. Cytokine Growth Factor Rev. 25, 167174 (2014).

39. Kearney, C. J. et al. Necroptosis suppresses inammation via termination of TNF- or LPS-induced cytokine and chemokine production. Cell Death Differ. 22, 13131327 (2015).

40. Wang, Q., Ju, X., Zhou, Y. & Chen, K. Necroptotic cells release nd-me signal and are engulfed without proinammatory cytokine production. In vitro cellular & developmental biology Animal 51, 10331039 (2015).

41. Davidovich, P., Kearney, C. J. & Martin, S. J. Inammatory outcomes of apoptosis, necrosis and necroptosis. Biological chemistry 395, 11631171 (2014).

42. Vucur, M. et al. RIP3 inhibits inammatory hepatocarcinogenesis but promotes cholestasis by controlling caspase-8- and JNK-dependent compensatory cell proliferation. Cell reports 4, 776790 (2013).

43. Kelliher, M. A. et al. The death domain kinase RIP mediates the TNF-induced NF-kappaB signal. Immunity 8, 297303 (1998).

44. Festjens, N., Vanden Berghe, T., Cornelis, S. & P., Vandenabeele RIP1 a kinase on the crossroads of a cells decision to live or die. Cell Death Differ. 14, 400410 (2007).

45. Wronska, A. et al. White adipose tissue depot-specic activity of lipogenic enzymes in response to fasting and refeeding in young and old rats. Gerontology 61, 448455 (2015).

46. Seimon, T. & Tabas, I. Mechanisms and consequences of macrophage apoptosis in atherosclerosis. J. Lipid Res. 50(Suppl): S382S387 (2009).

47. Angelovich, T. A., Hearps, A. C. & Jaworowski, A. Inammation-induced foam cell formation in chronic inammatory disease. Immunol. Cell Biol. 93, 683693 (2015).

48. Fischer-Posovszky, P., Wang, Q. A., Asterholm, I. W., Rutkowski, J. M. & Scherer, P. E. Targeted deletion of adipocytes by apoptosis leads to adipose tissue recruitment of alternatively activated M2 macrophages. Endocrinology 152, 30743081 (2011).

49. Zhang, Y. & Huang, C. Targeting adipocyte apoptosis: a novel strategy for obesity therapy. Biochem. Biophys. Res. Commun. 417, 14 (2012).

50. Kern, P. A. et al. The expression of tumor necrosis factor in human adipose tissue. Regulation by obesity, weight loss, and relationship to lipoprotein lipase.J. Clin. Invest. 95, 21112119 (1995).51. Yki-Jarvinen, H. Non-alcoholic fatty liver disease as a cause and a consequence of metabolic syndrome. Lancet Diabetes Endocrinol. 2, 901910 (2014).

52. Dyson, J. et al. Hepatocellular cancer: the impact of obesity, type 2 diabetes and a multidisciplinary team. J. Hepatol. 60, 110117 (2014).

53. Afonso, M. B. et al. Necroptosis is a key pathogenic event in human and experimental murine models of non-alcoholic steatohepatitis. Clin. Sci. 129, 721739 (2015).

54. Gautheron, J., Vucur, M. & Luedde, T. Necroptosis in nonalcoholic steatohepatitis. Cell. Mol. Gastroenterol. Hepatol. 1, 264265 (2015).

55. Ratziu, V. et al. A phase 2, randomized, double-blind, placebo-controlled study of GS-9450 in subjects with nonalcoholic steatohepatitis. Hepatology 55, 419428 (2012).

56. Kloting, N. et al. Insulin-sensitive obesity. Am. J. Physiol. Endocrinol. Metab. 299, E506E515 (2010).

57. Caporaso, J. G. et al. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 6, 16211624 (2012).58. Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335336 (2010).

59. McDonald, D. et al. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J. 6, 610618 (2012).

60. Ehling, J. et al. Micro-CT imaging of tumor angiogenesis: quantitative measures describing micromorphology and vascularization. Am. J. Pathol. 184, 431441 (2014).

61. Gremse, F. et al. Absorption reconstruction improves biodistribution assessment of uorescent nanoprobes using hybrid uorescence-mediated tomography. Theranostics 4, 960971 (2014).

62. Gremse, F. et al. Hybrid microCT-FMT imaging and image analysis. J. Vis. Exp. 100, e52770 (2015).

NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 15

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11869

63. Kleiner, D. E. et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 41, 13131321 (2005).

64. Orlicky, D. J., DeGregori, J. & Schaack, J. Construction of stable coxsackievirus and adenovirus receptor-expressing 3T3-L1 cells. J. Lipid Res. 42, 910915 (2001).

Acknowledgements

We thank R. Hillermann, D. Kull, V. Mangels, K. Kreggenwinkel, J. Ruske, N. Koch, C. Glantschnig and M. Scheideler for their excellent technical and experimental support and Professor T. Pufe (Institute of Anatomy and Cell BiologyRWTH Aachen) for technical assistance. We thank C. Gautheron for expert artwork. Research in the laboratory of T.L. was supported by a Mildred-Scheel Endowed Professorship and a single project grant (110043) from the German Cancer Aid (Deutsche Krebshilfe), the German-Research-Foundation (SFB-TRR57/P06 and LU 1360/3-1), an ERC Starting Grant (ERC-2007-Stg/ 208237-Luedde-Med3-Aachen), the EMBO Young Investigator Program, the Interdisciplinary-Centre-for-Clinical-Research (IZKF) Aachen-Germany and the Ernst-Jung-Foundation Hamburg. M.H. was supported by an ERC starting grant, The Stiftung fr Experimentelle Biomedizin, the Helmholtz foundation and the German-Research-Foundation (SFB-TR36). S.H. was supported by the Helmholtz Cross-Program Topic Metabolic Dysfunction and the Helmholtz Alliance ICEMED. M.V. was supported by a START-Grant from the RWTH Aachen. J.G. was supported by a grant from the dean of the medical faculty and a START-Grant from the RWTH Aachen.

Author contributions

J.G. and T.L. conceived and designed the experiments; J.G. and M.V. performed and analysed most of the experiments; A.T.S., C.R., S.R., M.B., P.S., M.B.D., C.K. contributed to the performance and analyses of experiments; I.S. performed the electron microscopy

experiment; J.E., F.G., T.La. and F.K. designed and performed the imaging experiments; N.V.B. and O.P. performed the microbiota experiments; F.H. helped to set up the live imaging movies; M.Bl. provided human fat samples and clinical data; A.L., S.K. and D.R.G. provided mouse lines and reagents; G.C., U.P.N., A.L., F.T., C.T., D.R.G., N.F., M.L., S.H. and M.H. provided intellectual input; T.Lo. performed histological analyses and provided intellectual input, M.L. and M.H. provided important technical support; J.G. and T.L. wrote the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Gautheron, J. et al. The necroptosis-inducing kinase RIPK3 dampens adipose tissue inammation and glucose intolerance. Nat. Commun. 7:11869 doi: 10.1038/ncomms11869 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

16 NATURE COMMUNICATIONS | 7:11869 | DOI: 10.1038/ncomms11869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications