a-ketoglutarate orchestrates macrophage activation through...

12
NATURE IMMUNOLOGY VOLUME 18 NUMBER 9 SEPTEMBER 2017 985 M1 macrophages elicit rapid proinflammatory responses to infec- tion and tissue damage by sensing microbial components such as lipopolysaccharide (LPS) and damage-associated molecular patterns released from injured tissues. In contrast, M2 macrophages induced by interleukin 4 (IL-4) and IL-13 show anti-inflammatory and repara- tive activities to resolve inflammation and promote tissue repair 1,2 . Macrophages are thought to engage the M1 phenotype initially in infection, to orchestrate host immunity, and to then obtain the M2 phenotype to restrain proinflammatory responses and repair dam- aged tissues. However, deregulated activation and a phenotypic switch in macrophages can lead to ineffective immune responses against infections and severe tissue damage 3 . For example, failure to restrain excessive production of proinflammatory cytokines in macrophages in response to systemic bacterial infection results in septic shock accompanied by tissue and organ damage. In contrast, after initial infection, the prolonged restriction of M1 activation can paralyze the immune system against subsequent microbial infections 3–5 . Therefore, a sophisticated regulatory network in macrophages is critical to guid- ing activation in the context of infection, cytokine milieu and other microenvironment cues. The different regulatory mechanisms that orchestrate macrophage activation, including signaling cascades and epigenetic programming, are increasingly understood. Among these regulatory mechanisms, differences in the bioenergetic demands of M1 and M2 macrophages are emerging as regulatory circuits that adjust macrophage behavior in response to nutrient states in its habitation and the infected tissues. M1 macrophages rely on aerobic glycolysis for ATP generation and increased glucose and glutamine consumption, but they suppress oxi- dative metabolism 6–9 . M1 macrophages have a broken tricarboxylic acid (TCA) cycle, allowing accumulation of citrate and succinate. Accumulated citrate can be used for production of fatty acids and itaconate, an antimicrobial molecule 6–8 . Succinate acts as a proinflam- matory metabolite that stabilizes transcription factor HIF-1α through inhibition of prolyl hydroxylase (PHD) activity and promotion of reactive oxygen species (ROS) production 8,10 . In contrast, M2 macro- phages maintain an intact TCA cycle and favor oxidative metabolism, especially FAO, as a mode of ATP production 11,12 . Similarly to block- ing electron transport chain (ETC) activity, inhibition of lysosomal lipolysis or fatty acid oxidation impairs IL-4-induced anti-inflamma- tory responses and the reparative ability of M2 macrophages, which suggests that oxidative phosphorylation (OXPHOS) and FAO con- trol the regulatory circuits of M2 activation 11,12 . M2 macrophages consume more glucose and glutamine than naive macrophages. This glycolytic metabolism strengthens M2 activation by generating fatty acids to fuel FAO and serving as an acetyl-CoA donor for epigenetic reprogramming 13,14 . However, it remains unclear how glutamine metabolism supports M2 activation 15 . These findings reveal the importance of metabolic remodeling during macrophage activation, but whether or not these metabolic activities intervene in canonical signaling cascades and epigenetic reprogramming for macrophage 1 Department of Fundamental Oncology, Faculty of Biology and Medicine, University of Lausanne, Epalinges, Switzerland. 2 Ludwig Lausanne Branch, Epalinges, Switzerland. 3 Metabolomics Unit, Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland. 4 Laboratory of Cellular Metabolism and Metabolic Regulation, Center for Cancer Biology, VIB, Leuven, Belgium. 5 Laboratory of Cellular Metabolism and Metabolic Regulation, Department of Oncology, KU Leuven and Leuven Cancer Institute (LKI), Leuven, Belgium. 6 Institute of Bioinformatics and Systems Biology, National Chiao Tung University, Hsin-Chu, Taiwan. 7 Department of Biological Science and Technology, National Chiao Tung University, Hsin-Chu, Taiwan. 8 Laboratory of Tumor Inflammation and Angiogenesis, Vesalius Research Center, VIB, Leuven, Belgium. 9 Laboratory of Tumor Inflammation and Angiogenesis, Department of Oncology, KU Leuven, Leuven, Belgium. Correspondence should be addressed to P.-C.H. ([email protected]). Received 2 May; accepted 21 June; published online 17 July 2017; doi:10.1038/ni.3796 a-ketoglutarate orchestrates macrophage activation through metabolic and epigenetic reprogramming Pu-Ste Liu 1,2 , Haiping Wang 1,2 , Xiaoyun Li 1,2 , Tung Chao 1,2 , Tony Teav 3 , Stefan Christen 4,5 , Giusy Di Conza 1,2 , Wan-Chen Cheng 1,2 , Chih-Hung Chou 6,7 , Magdalena Vavakova 1,2 , Charlotte Muret 1,2 , Koen Debackere 8,9 , Massimiliano Mazzone 8,9 , Hsien-Da Huang 6,7 , Sarah-Maria Fendt 4,5 , Julijana Ivanisevic 3 & Ping-Chih Ho 1,2 Glutamine metabolism provides synergistic support for macrophage activation and elicitation of desirable immune responses; however, the underlying mechanisms regulated by glutamine metabolism to orchestrate macrophage activation remain unclear. Here we show that the production of a-ketoglutarate (aKG) via glutaminolysis is important for alternative (M2) activation of macrophages, including engagement of fatty acid oxidation (FAO) and Jmjd3-dependent epigenetic reprogramming of M2 genes. This M2-promoting mechanism is further modulated by a high aKG/succinate ratio, whereas a low ratio strengthens the proinflammatory phenotype in classically activated (M1) macrophages. As such, aKG contributes to endotoxin tolerance after M1 activation. This study reveals new mechanistic regulations by which glutamine metabolism tailors the immune responses of macrophages through metabolic and epigenetic reprogramming. ARTICLES © 2017 Nature America, Inc., part of Springer Nature. All rights reserved.

Upload: others

Post on 05-Feb-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

nature immunology  VOLUME 18 NUMBER 9 SEPTEMBER 2017 985

M1 macrophages elicit rapid proinflammatory responses to infec-tion and tissue damage by sensing microbial components such as lipopolysaccharide (LPS) and damage-associated molecular patterns released from injured tissues. In contrast, M2 macrophages induced by interleukin 4 (IL-4) and IL-13 show anti-inflammatory and repara-tive activities to resolve inflammation and promote tissue repair1,2. Macrophages are thought to engage the M1 phenotype initially in infection, to orchestrate host immunity, and to then obtain the M2 phenotype to restrain proinflammatory responses and repair dam-aged tissues. However, deregulated activation and a phenotypic switch in macrophages can lead to ineffective immune responses against infections and severe tissue damage3. For example, failure to restrain excessive production of proinflammatory cytokines in macrophages in response to systemic bacterial infection results in septic shock accompanied by tissue and organ damage. In contrast, after initial infection, the prolonged restriction of M1 activation can paralyze the immune system against subsequent microbial infections3–5. Therefore, a sophisticated regulatory network in macrophages is critical to guid-ing activation in the context of infection, cytokine milieu and other microenvironment cues.

The different regulatory mechanisms that orchestrate macrophage activation, including signaling cascades and epigenetic programming, are increasingly understood. Among these regulatory mechanisms, differences in the bioenergetic demands of M1 and M2 macrophages are emerging as regulatory circuits that adjust macrophage behavior

in response to nutrient states in its habitation and the infected tissues. M1 macrophages rely on aerobic glycolysis for ATP generation and increased glucose and glutamine consumption, but they suppress oxi-dative metabolism6–9. M1 macrophages have a broken tricarboxylic acid (TCA) cycle, allowing accumulation of citrate and succinate. Accumulated citrate can be used for production of fatty acids and itaconate, an antimicrobial molecule6–8. Succinate acts as a proinflam-matory metabolite that stabilizes transcription factor HIF-1α through inhibition of prolyl hydroxylase (PHD) activity and promotion of reactive oxygen species (ROS) production8,10. In contrast, M2 macro-phages maintain an intact TCA cycle and favor oxidative metabolism, especially FAO, as a mode of ATP production11,12. Similarly to block-ing electron transport chain (ETC) activity, inhibition of lysosomal lipolysis or fatty acid oxidation impairs IL-4-induced anti-inflamma-tory responses and the reparative ability of M2 macrophages, which suggests that oxidative phosphorylation (OXPHOS) and FAO con-trol the regulatory circuits of M2 activation11,12. M2 macrophages consume more glucose and glutamine than naive macrophages. This glycolytic metabolism strengthens M2 activation by generating fatty acids to fuel FAO and serving as an acetyl-CoA donor for epigenetic reprogramming13,14. However, it remains unclear how glutamine metabolism supports M2 activation15. These findings reveal the importance of metabolic remodeling during macrophage activation, but whether or not these metabolic activities intervene in canonical signaling cascades and epigenetic reprogramming for macrophage

1Department of Fundamental Oncology, Faculty of Biology and Medicine, University of Lausanne, Epalinges, Switzerland. 2Ludwig Lausanne Branch, Epalinges, Switzerland. 3Metabolomics Unit, Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland. 4Laboratory of Cellular Metabolism and Metabolic Regulation, Center for Cancer Biology, VIB, Leuven, Belgium. 5Laboratory of Cellular Metabolism and Metabolic Regulation, Department of Oncology, KU Leuven and Leuven Cancer Institute (LKI), Leuven, Belgium. 6Institute of Bioinformatics and Systems Biology, National Chiao Tung University, Hsin-Chu, Taiwan. 7Department of Biological Science and Technology, National Chiao Tung University, Hsin-Chu, Taiwan. 8Laboratory of Tumor Inflammation and Angiogenesis, Vesalius Research Center, VIB, Leuven, Belgium. 9Laboratory of Tumor Inflammation and Angiogenesis, Department of Oncology, KU Leuven, Leuven, Belgium. Correspondence should be addressed to P.-C.H. ([email protected]).

Received 2 May; accepted 21 June; published online 17 July 2017; doi:10.1038/ni.3796

a-ketoglutarate orchestrates macrophage activation through metabolic and epigenetic reprogrammingPu-Ste Liu1,2, Haiping Wang1,2, Xiaoyun Li1,2, Tung Chao1,2, Tony Teav3, Stefan Christen4,5, Giusy Di Conza1,2, Wan-Chen Cheng1,2, Chih-Hung Chou6,7, Magdalena Vavakova1,2, Charlotte Muret1,2, Koen Debackere8,9, Massimiliano Mazzone8,9, Hsien-Da Huang6,7, Sarah-Maria Fendt4,5, Julijana Ivanisevic3 & Ping-Chih Ho1,2

Glutamine metabolism provides synergistic support for macrophage activation and elicitation of desirable immune responses; however, the underlying mechanisms regulated by glutamine metabolism to orchestrate macrophage activation remain unclear. Here we show that the production of a-ketoglutarate (aKG) via glutaminolysis is important for alternative (M2) activation of macrophages, including engagement of fatty acid oxidation (FAO) and Jmjd3-dependent epigenetic reprogramming of M2 genes. This M2-promoting mechanism is further modulated by a high aKG/succinate ratio, whereas a low ratio strengthens the proinflammatory phenotype in classically activated (M1) macrophages. As such, aKG contributes to endotoxin tolerance after M1 activation. This study reveals new mechanistic regulations by which glutamine metabolism tailors the immune responses of macrophages through metabolic and epigenetic reprogramming.

A rt i c l e s©

201

7 N

atu

re A

mer

ica,

Inc.

, par

t o

f S

pri

ng

er N

atu

re. A

ll ri

gh

ts r

eser

ved

.

Page 2: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

986  VOLUME 18 NUMBER 9 SEPTEMBER 2017 nature immunology

A rt i c l e s

activation remains unclear. Most importantly, the therapeutic poten-tial of fine-tuning deregulated macrophage activation via metabolic targeting is still largely unexplored.

We report here that αKG generated from glutaminolysis promotes M2 activation via Jmjd3-dependent metabolic and epigenetic repro-gramming. In contrast, αKG impairs proinflammatory responses in M1 macrophages by suppressing the nuclear factor-κB (NF-κB) pathway through PHD-dependent proline hydroxylation of protein kinase IKKβ. Although glutaminolysis restricts M1 polarization, we found that αKG produced from glutaminolysis during LPS stimula-tion has a crucial role in promoting LPS-induced endotoxin tolerance. These results highlight the mechanisms controlled by glutaminolysis to fine-tune macrophage activities and suggest that modulation of glutamine metabolism would be an attractive strategy for harnessing macrophage-mediated immune responses.

RESULTSaKG modulates macrophage polarizationTo investigate how glutamine metabolism regulates macrophage acti-vation, we first determined the effect of glutamine deprivation in mouse bone-marrow-derived macrophages (BMDMs) treated with IL-4 or LPS. As reported15, glutamine deprivation impaired expres-sion of M2-specific marker genes, including Arg1, Ym1 (Chil3), Retnla and Mrc1, and suppressed arginase 1 activity (Fig. 1a,b). In contrast, BMDMs deprived of glutamine showed higher expression of M1-spe-cific marker genes, including Il1b, Tnf, Il6 and Il12 (Il12b), upon LPS

stimulation than BMDMs activated in glutamine-replete conditions (Fig. 1c). Because glutamine is catabolized mainly into αKG, which enters the TCA cycle to replenish TCA cycle intermediates via glutami-nolysis16, we assessed the transcriptomes of naive and LPS- and IL-4-stimulated BMDMs to determine the expression of genes involved in glutaminolysis and αKG metabolism. Treatment with LPS, but not IL-4, inhibited the expression of genes promoting αKG production (Fig. 1d and Supplementary Fig. 1a). To test whether glutamine promotes M2 but suppresses M1 activation through production of αKG, we treated BMDMs with BPTES, an inhibitor of glutaminase 1 (GLS1), to suppress glutaminolysis in the absence or presence of dimethyl-αKG (DM-αKG), a cell-permeable analog of αKG, during IL-4 or LPS stimulation17. Treatment with BPTES suppressed expression of M2-specific genes and arginase 1 activity in IL-4-treated BMDMs, but the addition of DM-αKG restored the M2 phenotype (Fig. 1e,f). In con-trast, BPTES treatment boosted proinflammatory gene expression and cytokine production in LPS-stimulated BMDMs, and the presence of DM-αKG impaired this induction (Fig. 1g,h). The effects of targeting glutaminolysis in both M1 and M2 macrophages were confirmed with compound 968, another GLS1 inhibitor (Supplementary Fig. 1b,c). Furthermore, DM-αKG augmented M2 marker gene expression and arginase 1 activity (Supplementary Fig. 1d,e) but impaired the M1 proinflammatory phenotype (Supplementary Fig. 1f–h) in a dose-dependent manner. Together, these results indicate that αKG gener-ated by glutaminolysis is crucial for supporting an anti-inflammatory phenotype in macrophages.

2.0

1.0

0

3.0

2.0

1.0

0

1.2

0.8

0.4

0

1.5

1.0

0.5

0

Il12Il6TnfIl1b

1.5

1.0

0.5

0

1.2

0.8

0.4

0

5

4

3

2

1

0

2.0

1.5

1.0

0.5

0

Ctrl LPS IL-4

Ccbl1

Got1

Gpt2

Gpt1

Idh3g

Idh3b

Bcat1

Idh3a

Bcat2

Psat1

Ccbl2

Glud1

Gls1

Got2

Idh1

Idh2

Rel

ativ

e fo

ld c

hang

e in

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

e in

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

e in

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

e in

argi

nase

act

ivity

Rel

ativ

e fo

ld c

hang

e in

argi

nase

act

ivity

Rel

ativ

e fo

ld c

hang

e in

mR

NA

exp

ress

ion

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

1.5

1

0.5

0BPTES – + +

– – +DM-αKGBPTES – + +

– – +DM-αKGBPTES – + +

– – +DM-αKGBPTES – + +

– – +DM-αKGBPTES – + +

– – +DM-αKG

BPTES

Z-score+1.75–1.5

e – + +– – +DM-αKG

BPTES – + +– – +DM-αKG

BPTES – + +– – +DM-αKG

BPTES – + +– – +DM-αKG

BPTES – + +– – +DM-αKG

BPTES

3

2

1

0

60

40

20

0

900

600

300

0– + +– – +DM-αKG

3.0

2.0

1.0

0

40

30

20

10

0

12

8

4

0

2.0

2.5

1.5

1.0

0.5

0

TN

F (

pg/m

l)

IL-1

β (p

g/m

l)

********

* * * *

**

**

**

**

**

****

**

*

**

Argl Yml Retnla Mrcl

Argl Yml Retnla Mrcl

w/o Gln

w/ Gln

Il12Il6TnfIl1�

w/o Gln

w/ Gln

w/o Gln

w/ Gln

a

d

b c

e

g

f

h

Figure 1 Glutamine metabolism modulates macrophage activation via αKG production. (a–c) qPCR analysis of mRNA expression of M2 marker genes (a), arginase activity (b) or M1 marker genes (c) in BMDMs stimulated with IL-4 (a,b) or LPS (c) under various culture conditions for 6 h. (d) Expression of genes encoding metabolic enzymes contributing to αKG metabolism in untreated (ctrl) or LPS- or IL-4-treated BMDMs, assessed by next-generation RNA-seq. (e–h) qPCR analysis of relative mRNA M2 marker gene expression (e), arginase activity (f), M1 marker gene expression (g) or production of proinflammatory cytokines (h) in BMDMs treated with IL-4 (e,f) or LPS (g,h) under various culture conditions for 6 h. *P < 0.05, unpaired, two-tailed Student’s t-test. Data are representative of 3 independent experiments with 3 samples per group (a–c, e–h; mean ± s.d.) or cumulative results from 2 independent experiments (d).

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 3: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

nature immunology  VOLUME 18 NUMBER 9 SEPTEMBER 2017 987

A rt i c l e s

Glutaminolysis supports metabolic reprogramming in M2 macrophagesM2 activation induces metabolic remodeling to promote oxygen consumption and drive OXPHOS as the major mode of ATP pro-duction11,12. However, the metabolic checkpoints that determine the engagement of OXPHOS in M2 macrophages are unknown. To investigate whether glutamine metabolism provides such metabolic checkpoints, we determined the oxygen-consumption rates (OCR) of IL-4-stimulated BMDMs using a Seahorse extracellular flux analyzer. As expected, IL-4-treated BMDMs showed higher spare respiratory capacity (SRC) than naive BMDMs11. Whereas IL-4 treatment did not increase SRC in glutamine-deprived BMDMs, the addition of DM-αKG allowed glutamine-deprived BMDMs to acquire this metabolic change (Fig. 2a,b), suggesting that αKG is a metabolic checkpoint produced by glutaminolysis to support metabolic reprogramming of M2 mac-rophages. We next investigated whether αKG selectively boosts spe-cific nutrient-mediated OXPHOS by treating BMDMs with UK5099, which blocks mitochondrial import of pyruvate18, or etomoxir, which suppresses FAO7,11. IL-4 stimulation and glutamine deprivation did not affect pyruvate-dependent OCR in macrophages (Fig. 2c,d). In contrast, IL-4 treatment markedly augmented FAO-depend-ent OCR in glutamine-replete conditions. In glutamine-deprived BMDMs, FAO-dependent OCR was not induced by IL-4 stimulation,

but it was restored by DM-αKG supplementation (Fig. 2e,f), sug-gesting that αKG serves as a regulator governing the engagement of FAO in M2 macrophages. Further, we found that M2 activation promotes the expression of carnitine palmitoyltransferase 1a (Cpt1a), a rate-limiting enzyme in FAO, and increases the rate of fatty acid uptake in an αKG-dependent manner (Fig. 2g,h). Collectively, these results indicate that glutamine metabolism is crucial for supporting metabolic reprogramming of M2 macrophages through αKG-medi-ated regulations.

M2 polarization relies on the aKG–Jmjd3 pathwayWe next investigated whether glutaminolysis promotes M2 activa-tion by influencing activation of transcription factor STAT6. Neither glutamine deprivation nor BPTES treatment suppressed IL-4-induced STAT6 phosphorylation in BMDMs; DM-αKG supplementation in also failed to increase STAT6 phosphorylation in glutamine-deprived BMDMs (Supplementary Fig. 2a–c), ruling out the possibility that glutaminolysis promotes M2 activation by boosting STAT6 activa-tion. In addition to STAT6 signaling, epigenetic reprogramming orchestrates M2 activation and sustains macrophage polarization flexibility2,19. We next examined trimethylation of histone H3 K27 (H3K27me3), a repressive epigenetic mark that prevents expression of M2 marker genes19, on the promoters of M2 marker genes Arg1,

80

60

40

20

0

160

140

100

120

80

60

40

20

0

60

40

20

0

14

10

12

8

6

4

2

0

10

8

6

4

2

0

UK

-sen

sitiv

e O

CR

(pm

ol/m

in/m

g pr

otei

n)

Eto

mox

ir-se

nsiti

ve O

CR

(pm

ol/m

in/m

g pr

otei

n)

OC

R (

pmol

/min

/mg

prot

ein)

Bod

ipy

C12

upt

ake

(MF

I × 1

03 )

Rel

ativ

e fo

ld c

hang

e in

Cpt1a

mR

NA

exp

ress

ion

60

40

20

0– – – ++ + – –– + + +

DM-αKGGlnIL-4

Time (min)

18 30 42 54 66 78 90

– – – ++ + – –– + + +

DM-αKGGlnIL-4

SR

C (

pmol

/min

/mg

prot

ein)

OC

R (

pmol

/min

/mg

prot

ein)

– – – ++ + – –– + + +

DM-αKGGlnIL-4

– – – ++ + – –– + + +

DM-αKGGln

IL-4

– – – ++ + – –– + + +

DM-αKGGlnIL-4

Time (min)

6 18 30 42 54 66 78

Time (min)6 18 30 42 54 66 78 90

Eto

Rot/AAOligo

FCCP

**

* *

**

*

**

n.s.

n.s.n.s.

*

**

180

160

140

100

120

80

60

40

20

090

Oligo

SRC

Rot/AA

FCCP

Rot/AA

UK

Oligo

FCCP

180

160

140

100

120

80

60

40

20

0

OC

R (

pmol

/min

/mg

prot

ein)

w/ Glnw/ Gln+IL-4w/o Gln+IL-4w/o Gln+DM-αKG+IL-4

6

w/ Glnw/ Gln+IL-4w/o Gln+IL-4w/o Gln+DM-αKG+IL-4

w/ Glnw/ Gln+IL-4w/o Gln+IL-4w/o Gln+DM-αKG+IL-4

a b c d

e f g h

Figure 2 αKG promotes metabolic changes induced by M2 activation. (a,b) OCR (a) and SRC (b) of BMDMs under various culture conditions, with or without 6 h IL-4 stimulation before treatment with oligomycin (oligo), FCCP, and rotenone plus antimycin A (Rot/AA). (c) OCR of BMDMs treated as in a followed by addition of FCCP, UK5099 (UK) and Rot/AA. (d) UK-sensitive OCR of BMDMs treated as in a, calculated as the reduction of OCR after UK5099 treatment. (e) OCR of BMDMs treated as in a then treated with oligomycin, FCCP, etomoxir (eto) and Rot/AA. (f) Etomoxir-sensitive OCR of BMDMs treated as in a, calculated as the reduction of OCR after etomoxir treatment. (g,h) Cpt1a mRNA expression (g) and uptake of BODIPY-labeled C12-fatty acids (h) in BMDMs treated as indicated for 6 h. *P < 0.05, unpaired, two-tailed Student’s t-test. Data are representative of 3 independent experiments with 3 samples per group (mean ± s.d.).

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 4: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

988  VOLUME 18 NUMBER 9 SEPTEMBER 2017 nature immunology

A rt i c l e s

Ym1, Retnla and Mrc1. IL-4 treatment decreased H3K27me3 on the promoters of these genes in a glutamine-dependent fashion (Fig. 3a), which indicates that glutamine metabolism may support M2 activa-tion by facilitating demethylation of H3K27. Given that Jmjd3 is a key enzyme for demethylation of H3K27 and controls M2 activa-tion through epigenetic regulations19,20, we asked whether glutamine metabolism supports M2 activation by promoting Jmjd3-dependent demethylation of H3K27 on the promoters of M2-specific marker genes. DM-αKG promoted loss of H3K27me3 in glutamine-deprived BMDMs, and treatment with GSK-J4, a Jmjd3 selective inhibitor21, suppressed demethylation of H3K27 in the presence of glutamine or DM-αKG (Fig. 3b). Moreover, GSK-J4 suppressed expression of M2-specific marker genes and arginase 1 activity (Supplementary Fig. 2d,e) in BMDMs stimulated with IL-4 in glutamine-replete conditions or with DM-αKG supplementation, but not in BMDMs cultured in glutamine-depleted conditions. This suggests that αKG produced from glutamine metabolism promotes M2 activation via Jmjd3-dependent epigenetic reprogramming. To further test whether DM-αKG augments M2 activation in a Jmjd3-dependent manner, we transduced BMDMs from mice in which Cas9 is selectively expressed in macrophages (LysM-Cre Cas9 mice)22 with a lentivirus harboring singe-guide RNAs (sgRNAs) targeting Jmjd3 or scramble sgRNAs to generate Jmjd3-deficient BMDMs or control BMDMs, respectively. Jmjd3 expression was lower in Jmjd3-deficient BMDMs than in con-trol BMDMs (Supplementary Fig. 2f), and DM-αKG treatment did not promote M2-specific gene expression or arginase 1 activity in

Jmjd3-deficient BMDMs (Fig. 3c,d). Furthermore, supplementation of glutamine or DM-αKG did not augment demethylation of H3K27me3 in Jmjd3-deficient BMDMs (Fig. 3e).

Next, we compared the expression of IL-4-induced genes, including Ccl22, Hr, Il4i1 and Irf4, which have been reported to be suppressed by glutamine deprivation15. Deprivation of glutamine suppressed the expression of these genes, and supplementation with DM-αKG restored it in BMDMs. However, supplementation of glutamine or DM-αKG did not promote expression of these genes in Jmjd3-defi-cient BMDMs (Fig. 3f). Next, we found that Jmjd3-deficient BMDMs did not increase oxygen consumption in response to IL-4 in the pres-ence of glutamine or DM-αKG (Fig. 3g,h), which suggests that the αKG–Jmjd3 signaling pathway is responsible for metabolic repro-gramming in M2 macrophages. Together, these results indicate that the αKG–Jmjd3 pathway functions as a metabolic checkpoint by which glutamine metabolism supports macrophage M2 activation.

aKG/succinate ratio modulates macrophage activationαKG is a co-stimulator factor for Jmjd3, and succinate is an inhibi-tor; therefore, a high αKG/succinate ratio induces Jmjd3-dependent demethylation of H3K27 to support pluripotency of embryonic stem cells17,23. To test whether αKG/succinate ratios modulate macrophage activation, we measured the αKG/succinate ratios in naive, M1 and M2 BMDMs and found that IL-4 stimulation resulted in a higher αKG/succinate ratio than LPS treatment, (Fig. 4a and Supplementary Fig. 3a), suggesting that regulation of αKG dehydrogenase activity

0.3

0.2

0.1

0 H3K

27 m

e3 e

nric

hmen

t(%

of i

nput

)

H3K

27 m

e3 e

nric

hmen

t(%

of i

nput

)

Rel

ativ

e fo

ld c

hang

e in

mR

NA

exp

ress

ion

H3K

27 m

e3en

richm

ent

(% o

f inp

ut)

SR

C (

pmol

/min

/m

g pr

otei

n)

OC

R (

pmol

/min

/m

g pr

otei

n)

1.5

1.2

0.9

0.6

0.3

0

0.8

0.6

0.4

0.2

0

4

3

2

1

0

0.3

0.4

0.2

0.1

0

0.3

0.5

0.4

0.2

0.1

0

0.8

0.6

0.4

0.2

0

0.3

0.5

0.4

0.2

0.1

0

DM-αKGGlnIL4

– – – ++ + – –– + + +

DM-αKG

DM-αKG

Gln

Gln

IL4

– – –

+ – – + – –– – + – – +DM-αKG

Gln + – – + – –– – + – – +DM-αKG

Gln + – – + – –– – + – – +DM-αKG

Gln + – – + – –– – + – – +

DM-αKGGln + – – + – –

– – + – – +DM-αKGGln + – – + – –

– – + – – +DM-αKGGln + – – + – –

– – + – – +DM-αKGGln + – – + – –

– – + – – +

++ + – –– + + +

50

40

30

20

10

0

50

40

30

20

10

0

20

15

10

5

0

3

2

1

0

100

80

60

40

20

0

100

80

60

40

20

0

1.2

0.8

0.4

0.0

2.0

1.5

1.0

0.5

0.0

1.5

1.0

0.5

0.0

1.5

1.0

0.5

0.0

Arg1

Arg1

2.5

2

1.5

0.5

1

0

1.5

0.5

1

0

Ym1

Ym1

Retnla

Retnla

Mrc1

Mrc1

Arg1

Ccl22 Hr Il4il Irf4

Ym1 Retnla Mrc1

Jmjd3-sgRNA Jmjd3-sgRNACtrl CtrlRot/AA

Rot/AA

OligoOligo

FCCPFCCP

n.s.n.s.n.s.n.s.n.s. n.s. n.s.

* ***

***

*

n.s.

w/ Gln

w/ Gln+IL-4

w/o Gln+IL-4

* * * *

w/o Gln+DM-αKG+IL-4

* ***

*

*

*

**

**

*

** *

*

* *

**

**

* *

**n.s.n.s.n.s.

n.s.*

*

***

n.s.n.s.n.s.

DM-αKGGln

GSK-J4

– – + + – –– – – – + +– + – + – +

DM-αKGGln

GSK-J4

– – + + – –– – – – + +– + – + – +

DM-αKGGln

GSK-J4

– – + + – –– – – – + +– + – + – +

DM-αKGGln

GSK-J4

– – + + – –– – – – + +– + – + – +

GlnIL-4

+ + –– + +

GlnIL-4

+ + –– + +

GlnIL-4

+ + –– + +

GlnIL-4

+ + –– + +

Rel

ativ

e fo

ld c

hang

e in

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

e in

arg

inas

e ac

tivity

4

3

2

1

0DM-αKG – + – +DM-αKG – + – +DM-αKG – + – +DM-αKG – + – + DM-αKG – + – +

9

6

3

0

20

15

10

5

0

16

12

8

4

0

10

8

6

4

2

0

Argl Yml Rentla Mrcl

Jmjd3-sgRNACtrl

n.s. n.s.

n.s.n.s.

*

n.s.

****

Jmjd3-sgRNACtrl

Jmjd3-sgRNACtrl

Jmjd3-sgRNACtrl

18 36

Time (min)

54 72 18 36

Time (min)

54 72

a b

c d e

f g h

Figure 3 αKG promotes IL-4-induced epigenetic changes in a Jmjd3-dependent manner. (a,b) Chromatin-immunoprecipitation (ChIP) analysis of H3K27me3 in BMDMs with or without IL-4 stimulation (a) or stimulated with IL-4 with or without DM-αKG or GSK-J4 (b) after real-time qPCR of promoter regions on M2 marker genes. Data are normalized to input. (c–e) Relative mRNA expression (c) or H3K27me3 (e) of promoter regions on M2 marker genes and arginase activity (d) in Jmjd3-deficient (Jmjd3-sgRNA) or control (ctrl) IL-4-stimulated BMDMs with or without DM-αKG (1 µM). (f) Relative mRNA expression in IL-4 stimulated BMDMs under various culture conditions. (g) OCR (g) or SRC (h) of BMDMs stimulated with IL-4 under various conditions for 6 h then treated with oligomycin (oligo), FCCP and rotenone plus antimycin A (Rot/AA). *P < 0.05, unpaired, two-tailed Student’s t-test. Data are representative of 3 independent experiments (a,b,e–h; mean ± s.d.) or from 2 independent experiments with 3 samples per group (c,d; mean ± s.d.).

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 5: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

nature immunology  VOLUME 18 NUMBER 9 SEPTEMBER 2017 989

A rt i c l e s

may differ between M1 and M2 macrophages to affect αKG-to- succinate conversion and establish different αKG/succinate ratios. Indeed, LPS-stimulated BMDMs showed higher αKG dehydrogenase activity than IL-4-treated BMDMs (Fig. 4b), which indicates that M2 macrophages may favor αKG accumulation by suppressing αKG dehy-drogenase activity. To further examine whether αKG/succinate ratios affect M2 activation, we treated glutamine-deprived BMDMs with a fixed concentration of diethyl-succinate (DE-Suc), a cell-permeable succinate, in the presence of increasing doses of DM-αKG. Increasing αKG/succinate ratios in this manner promoted M2-specific marker gene expression (Fig. 4c) and arginase 1 activity (Supplementary Fig. 3b). In contrast, although DE-Suc acts as a proinflammatory metabolite to stimulate Il1b gene expression and cytokine produc-tion in LPS-treated macrophages8, treatment with escalating doses of DM-αKG gradually abolished IL-1β expression in BMDMs (Fig. 4d and Supplementary Fig. 3c). In addition, increasing doses of DE-Suc impaired DM-αKG-induced M2 marker gene expression (Fig. 4e) and arginase 1 activity (Supplementary Fig. 3d). However, DE-Suc did not restore IL-1β expression in DM-αKG-treated macrophages (Fig. 4f and Supplementary Fig. 3e). Together, these data indicate that αKG/succinate ratio affects macrophage polarization programs, and manipulation of this ratio may allow tailoring of macrophage immune responses.

aKG suppresses IKKb activationWe tested whether αKG inhibits production of proinflammatory cytokines by suppressing a conserved upstream signal responsible for M1 activation. Both glutamine deprivation and BPTES treatment in BMDMs promoted nuclear translocation of transcription factor NF-κB in response to LPS treatment, an essential event to drive broad proinflammatory functions (Fig. 5a and Supplementary Fig. 4a). Further, DM-αKG suppressed nuclear translocation of NF-κB in glutamine-deprived BMDMs (Fig. 5b). As assessed by a PCR array, supplementation with DM-αKG suppressed expression of many

NF-κB signaling targets, such as Cd80 and Sele, in LPS-stimulated BMDMs but not in BMDMs treated with control vehicle (Fig. 5c), which suggests that αKG derived from glutaminolysis may impair proinflammatory responses by restraining the NF-κB signaling path-way. Because αKG can interfere with cellular function by activating PHD enzymes24, we examined whether αKG suppresses M1-specific marker gene expression in a PHD-dependent manner. Treatment with dimethyloxalyglycine (DMOG), an inhibitor of PHD activity, restored M1 marker gene expression in DM-αKG-treated BMDMs (Fig. 5d). BPTES treatment in LPS-stimulated BMDMs increased the amount of phosphorylated IKK, whereas DM-αKG reduced it (Fig. 5e). Co-treatment with DMOG restored IKK phosphorylation in DM-αKG-treated BMDMs, suggesting that αKG impedes IKK activation by augmenting PHD activity. PHD has been reported to inhibit IKKβ activation through hydroxylation of IKKβ on P191 (refs. 25,26). We next overexpressed wild-type IKKβ or P191A-IKKβ, a proline-to-alanine substitution mutant that cannot be hydroxylated by PHD, in BMDMs (Supplementary Fig. 4b) and treated them with LPS. DM-αKG did not suppress LPS-induced M1 marker gene expression in BMDMs transduced with P191A-IKKβ, in contrast to non-transduced BMDMs and BMDMs overexpressing wild-type IKKβ (Fig. 5f). Similarly, DM-αKG did not suppress LPS-induced nuclear translo-cation of NF-κB in P191A-IKKβ−transduced BMDMs, as assessed by imaging flow cytometry (Fig. 5g,h). These results suggest a role for αKG in restricting M1 activation by intervening with the NF-κB path-way via PHD-mediated post-translational modification of IKKβ.

Priming-phase glutaminolysis modulates endotoxin toleranceGlutamine uptake in macrophages is greatly enhanced by LPS treat-ment27, while exposure to LPS induces endotoxin tolerance, in which macrophages become resistant to subsequent exposure after reso-lution of the acute response, to prevent excessive proinflammatory responses3,28,29. To test whether the engagement of glutamine metabo-lism during initial LPS activation contributes to the establishment

2.5 * * *

2.0

1.5

αKG

/suc

cina

tear

bitr

ary

ratio

Rel

ativ

eαK

G d

ehyd

roge

nase

activ

ity (

fold

)

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

1.0

0.5

0.0

2.5

2.0

1.5

1.0

0.5

0.0

2.05

Argl Yml Retnla Mrcl

4

3

2

1

0

4

3

2

1

0

3

2

1

0–– –

+ + +0.1 1

–– –

+ + +0.1 1

DE-Suc. DE-Suc.DM-αKG (mM) DM-αKG (mM)

–– –

+ + +0.1 1

DE-Suc.DM-αKG (mM)

–– –

+ + +0.1 1

DE-Suc.DM-αKG (mM)

n.s.n.s. n.s.

n.s.** *

**

* **

1.5

1.0

0.5

0.0Ctrl IL-4 IL-4LPS LPS

*** * **

Yml Retnla Mrcl

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion 4

II1b II1bArgl

3

2

1

0 Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion 3

2

1

0–– –

+ + +0.1 1

DE-Suc.DM-αKG (mM)

–– –

+ + +0.1 1DE-Suc. (mM)

DM-αKG

* **

* **

* ** 1.2

0.8

0.4

0.0

n.s.

n.s.

*

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

3

4

2

1

0 Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

3

4

2

1

0

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion 3

2

1

0–– –

+ + +0.1 1DE-Suc. (mM)

DM-αKG –– –

+ + +0.1 1DE-Suc. (mM)

DM-αKG –– –

+ + +0.1 1DE-Suc. (mM)

DM-αKG –– –

+ + +0.1 1DE-Suc. (mM)

DM-αKG

a b c

d e f

Figure 4 Integration of αKG/succinate ratio in macrophage determines macrophage immune responses. (a) Intracellular αKG/succinate ratio in BMDMs stimulated with vehicle (ctrl), IL-4 or LPS for 18 h, measured by MS after metabolite extraction (Online Methods). (b) Activity of mitochondrial αKG dehydrogenase in BMDMs stimulated with IL-4 or LPS for 18 h. (c,d) qPCR analysis of relative mRNA expression of M2 marker genes in glutamine-deprived BMDMs stimulated with IL-4 (c) and Il1b mRNA in BMDMs stimulated with LPS (d) under various conditions for 6 h. (e,f) qPCR analysis of relative mRNA expression of M2 marker genes in glutamine-deprived BMDMs stimulated with IL-4 (e) and Il1b mRNA in BMDMs stimulated with LPS (f) for 6 h under various conditions. *P < 0.05, unpaired, two-tailed Student’s t-test. Data are representative of 3 independent experiments with 3 samples per group (a–f; mean ± s.d.).

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 6: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

990  VOLUME 18 NUMBER 9 SEPTEMBER 2017 nature immunology

A rt i c l e s

of endotoxin tolerance, we deprived BMDMs of glutamine during first LPS priming and then re-challenged them with LPS to examine production of proinflammatory genes, including Il1b, Tnf, Il6 and Il12 (Supplementary Fig. 5a). Macrophages primed with LPS in glutamine-deprived conditions show high expression of proinflam-matory genes compared to BMDMs primed in glutamine-replete conditions (Fig. 6a), which indicates that LPS-induced endotoxin tolerance requires glutamine. Furthermore, inhibition of glutami-nolysis by BPTES during the LPS priming phase sustained BMDM’s ability to produce proinflammatory genes upon re-challenge with LPS (Supplementary Fig. 5b), which suggests that glutaminolysis dur-ing initial LPS treatment is essential to induce endotoxin tolerance. Glutamine-deprived BMDMs treated with DM-αKG during LPS prim-ing (Supplementary Fig. 5c) showed suppression of proinflammatory gene expression (Fig. 6b) and cytokine production (Supplementary Fig. 5d,e) upon re-exposure to LPS. However, DE-Suc treatment did not restore endotoxin tolerance in glutamine-deprived macrophages (Fig. 6c), implying that αKG promotes endotoxin tolerance independ-ently of its conversion into succinate.

In addition to resolving proinflammatory responses, endotoxin-tolerant macrophages express tissue-repairing and antimicrobial

genes2,3,28. RNA sequencing of BMDMs primed with LPS in glutamine-replete medium, or glutamine-depleted medium with or without DM-αKG, showed that glutamine deprivation during LPS priming suppressed expression of reparative and antimicrobial gene expression, while DM-αKG treatment sustained the expression of most of this gene set (Supplementary Fig. 5f). Further comparison of global gene expression patterns showed that glutamine depriva-tion altered the expression patterns of more than 1,200 genes in LPS-primed BMDMs, and supplementation with DM-αKG restored the expression patterns of 506 genes in this set (Fig. 6d). Furthermore, gene cluster analysis revealed that those genes regulated by αKG and glutamine metabolism controlled biological processes involved in inflammatory responses, exosome release and mitochondrial activ-ity (Fig. 6e). Because mitochondrial activity controls energy pro-duction and metabolic profiles, and immunometabolic paralysis—a global impairment of major metabolic pathways—has been reported to be a metabolic phenotype during endotoxin tolerance in human monocytes30, we next used metabolomics analysis to determine whether glutamine metabolism during LPS priming supports immu-nometabolic paralysis. As expected, tolerant macrophages showed a systemic reduction of metabolic activities compared to nontolerant

LPS (min)

LPS (min)

+Gln

Ctrl DM-αKG0 15 30 3060 6015

–Gln210

–1–2–3–4–5–6

–20–24

Ifng

Egr2

Cd80 Tnf

II6Sele

Lta

Csf2

Birc3

Birc2

Plau

II12b

Cxcl3

Csf1

Irf1

II1a

Ifnb1

Myd88

Nfkbia

II1r2

Cd83

Nr4a2 Ltb

F8

C3

Cfb

Rel

CxcI1

CcI12

Gad

d45b

Icam

1Fas

Egfr

Bcl2a1a

Nfkb2

Ccr5

CcI22

Pdgfb

Vcam1

Ncoa3

Rela

Nfkb1

Selp

Relb

Mitf

Tnfrsf1b

Tnfsf10

Cxcl9

Myc

Sod

2Ptgs2

Aldh3a2

Xiap

Traf2

Cd40

II1m

Stat3

Cd74

II1b

BcI2I1

Trp53

Csf2rp

Stat1

Stat5b

II15

Csf3

CcI5

Akt1

Cxcl10

Cdkn1a

Nqo1

Map

2k6

Mmp9

II2ra

Ccnd1

Adm F3

Fol

d re

gula

tion

0 15 30 60 15 30 60NF-κB

NF-κB

NF-κB

NF-κB

Lamin-A/C

Lamin-A/C

β-actin

β-actin

β-actin

IKKα/β

p-IKKα/β

II1b II1bTnf Tnf

WCL

Nuclearfraction

WCL

3 3.0

3.0

2.52.0

2.0

1.51.0

1.0

0.50.0

2.5

1.5

2.0

1.5

1.2 1.2

1.6

0.8

0.4

0.0

0.9

0.6

0.3

0.0

1.5

1.2

0.9

0.6

0.3

0.0

* **

**

**

*n.s.

n.s.

n.s.

n.s.

CtrlWT IKKβP191A-IKKβ

1.0

0.5

0.0

1.2*

n.s.

1.0

Rel

ativ

e si

mila

rity

inde

x

0.8

0.60.3

0.0

WT IK

P191A

-IKKβ

WT IKKβ

P191A-IKKβ0.0

* * *

*

*

*n.s.*

** * *

***

n.s.

II6

II6

II12

II12

2

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

1

0

4

3

2

1

0

DMOG(mM)DM-αKG

BPTES

–––

––

–+++ +

+ ++

0.5 1

DMOG(mM)DM-αKG

BPTES

–––

––

–+++ +

+ ++

0.5 1 DMOG(mM)DM-αKG

BPTES

–––

––

–+++ +

+ ++

0.5 1

DMOG(mM)DM-αKG

DM-αKG

Ctrl DM-αKGCtrl Ctrl

DAPI MergeNF-κBWT IKKβ

36.5% 46.4%

48.1%26.3%

P191A-IKKβ

DM-αKG

DM-αKG

DMOGLPS

––– – –

––

–– –

––

+ + +++

++

+ – + – + DM-αKG

DM-αKG

Ctrl

DM-αKG

– + – + – + DM-αKG – + – + – + DM-αKG – + – + – +

+ +++

+

30 min 60 min

BPTES

BPTES

–––

––

–+++ +

+ ++

0.5 1

Nuclearfraction

a

b

c

d e f

g h

Figure 5 αKG suppresses M1 activation through a PHD-dependent post-translational regulation of IKKβ. (a,b) Immunoblot analysis of NF-κB p65 in BMDMs stimulated with LPS in medium with or without glutamine (a) or in glutamine-depleted medium supplemented with vehicle (ctrl) or DM-αKG (1 mM) (b). WCL, whole-cell lysate. (c) qPCR array analysis of NF-κB target gene expression in BMDMs stimulated with LPS in glutamine-depleted medium with or without DM-αKG (1 mM) for 6 h. Genes showing significant differences (P < 0.05) after DM-αKG treatment are shown in red. (d) qPCR analysis of M1 marker gene expression in BMDMs stimulated with LPS in glutamine-replete medium with or without DM-αKG and/or BPTES. DMOG, dimethyloxalyglycine. (e) Immunoblot analysis of phosphorylated (p-) IKKα/β and IKKα/β in BMDMs stimulated with LPS under various culture conditions. (f,g) qPCR analysis of M1 marker gene expression (f) and imaging flow analysis of NF-κB (g) in untransduced (ctrl) BMDMs or BMDMs overexpressing wild-type (WT) IKKβ or P191A-IKKβ, stimulated with LPS in glutamine-depleted medium. (g) Left, representative histograms of similarity profiles of NF-κB nuclear translocation. Right, combined quantitative results of similarity index. (h) Representative images of cells from g. *P < 0.05, unpaired, two-tailed Student’s t-test. Data are representative of 3 (a,b,d,e; mean ± s.d.)), 4 (g,h; mean ± s.d.) or 2 (f) independent experiments or are cumulative results from 4 independent experiments (c) with 3 (a,b,d–f) or 4 (c) samples per group.

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 7: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

nature immunology  VOLUME 18 NUMBER 9 SEPTEMBER 2017 991

A rt i c l e s

macrophages (Supplementary Fig. 5g), while glutamine deprivation during initial LPS treatment prevented induction of endotoxin tol-erance but did not affect reduction of global metabolic activities in BMDMs. In agreement with published findings30, LPS-primed BMDMs produced lower amounts of lactate and pyruvate upon re-exposure to LPS than did naive BMDMs exposed to LPS (Supplementary Fig. 5h); however, nontolerant, glutamine-deprived BMDMs also generated lower amounts of lactate and pyruvate during initial LPS treatment, which suggests that glutamine metabolism may not impede the induc-tion of endotoxin tolerance by preventing immunometabolic paralysis. Together, these data demonstrate that glutamine metabolism during M1 activation acts as a metabolic checkpoint to induce endotoxin tolerance and regulate the transcriptomic landscape associated with endotoxin tolerance in macrophages.

Glutaminolysis impairs the NF-kB pathway in tolerant macrophagesImpairment of LPS-activated proximal signaling pathways is a hall-mark event in the induction of endotoxin tolerance3,31. Glutamine deprivation and BPTES treatment during LPS priming did not prevent the impairment of proximal signaling pathways upon re-exposure to

LPS (Supplementary Fig. 6a,b). Moreover, the addition of DM-αKG during LPS priming did not affect activation of proximal signaling pathways in glutamine-deprived BMDMs (Supplementary Fig. 6c), which indicates that glutamine metabolism affects establishment of endotoxin tolerance by mechanisms independent of LPS proximal signaling pathways. Next, we examined expression of negative regu-lators of the NF-κB signaling pathway, which are known to induce endotoxin tolerance3. Glutamine-deprived BMDMs expressed lower amounts of these regulators during LPS priming than did BMDMs primed with LPS in the glutamine-replete cultures. Notably, this phe-nomenon was prevented by DM-αKG treatment (Fig. 7a). Moreover, glutamine deprivation and BPTES treatment during LPS priming sustained NF-κB nuclear translocation upon re-exposure to LPS in BMDMs (Fig. 7b and Supplementary Fig. 6d). In contrast, DM-αKG treatment diminished the capacity of glutamine-deprived BMDMs to enhance NF-κB nuclear translocation upon re-challenge with LPS (Fig. 7c). Of note, supplementation of DM-αKG altered nuclear trans-location of NF-κB only in BMDMs primed with LPS in glutamine-deprived cultures (Supplementary Fig. 6e). These results suggest that αKG modulates endotoxin tolerance by affecting NF-κB-mediated

3.0

2.5

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion

2.0

1.5

1.5

4.5

3.0

**

* *

II1b Tnf II6 II12II1b Tnf II6 II12

1.0

0.5

2

Fol

d ch

ange

s

Fol

d ch

ange

sF

old

chan

ges

Cell adhesion

Upregulated genes

184 202

Gln supplementation Gln supplementation

DM-αKG supplementation DM-αKG supplementation

Overlap Overlap

55 589 304 124

Downregulated genes

Chemotaxis Wound healing

Cell surfaceProtein binding

1

0

Clu

ster

1

Clu

ster

2

2

1

0

2

1

0

3

2

1

0

ET w/ G

ln

ET w/o

Gln

ET w/o

Gln

+ DM

-αKG

ET w/ G

ln

ET w/o

Gln

ET w/o

Gln

+ DM

-αKG

ET w/ G

ln

ET w/o

Gln

ET w/o

Gln

+ DM

-αKG

ET w/ G

ln

ET w/o

Gln

ET w/o

Gln

+ DM

-αKG

ET w/ G

ln

ET w/o

Gln

ET w/o

Gln

+ DM

-αKG

ET w/ G

ln

ET w/o

Gln

ET w/o

Gln

+ DM

-αKG

3

2

1

0

2

3

1

0

0Response to LPS Mitochondrion

Integral component ofplasma membrane

–0.5

–1–1

–1

–1.5

–0.5

–2

0 0

–1.5

Extracellular exosome

0

3.0 w/ Glnw/o Gln2.5

1.6n.s.

n.s.

n.s.

n.s. n.s.

n.s.

n.s.

n.s.

1.2

0.8

0.4

0

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

2.0

2.5 DE-Suc.Ctrl

1.5

1.0

0.5

0

2.5

2.0

1.5

1.0

0.5

00+ + + +

++– –

2nd LPS

1st LPS

+ + + +

++– –

2nd LPS

1st LPS

+ + + +

++– –

2nd LPS

1st LPS

+ + + +

++– –

2nd LPS

1st LPS

+ + + +

++– –

2nd LPS

1st LPS

+ + + +

++– –

2nd LPS

1st LPS

+ + + +

++– –

2nd LPS

1st LPS

+ + + +

++– –

2nd LPS

1st LPS

Rel

ativ

e fo

ld c

hang

ein

mR

NA

exp

ress

ion 2.5

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

2.0

1.5

1.0

0.5

0

2.0

2.5

1.5

1.0

0.5

0

* ** *

** *

*

**

**n.s.II1b Tnf II6 II12

w/ Gln + Ctrl

w/o Gln + Ctrlw/o Gln + DM-αKG

w/ Gln + DM-αKG

n.s.

n.s.

n.s.

2nd LPS + + +

+

+ + + + +

+++– – – –1st LPS

2nd LPS + + +

+

+ + + + +

+++– – – –1st LPS

2nd LPS + + +

+

+ + + + +

+++– – – –1st LPS

2nd LPS + + +

+

+ + + + +

+++– – – –1st LPS

a c

b

d e

Figure 6 Glutamine metabolism supports induction of endotoxin tolerance in an αKG-dependent manner. (a) qPCR analysis of relative mRNA expression in LPS-stimulated BMDMs with or without LPS priming in glutamine-replete or glutamine-depleted medium. (b,c) qPCR analysis of relative mRNA expression of M1 marker genes in BMDMs stimulated with LPS after priming in glutamine-depleted medium with or without DM-αKG (1 mM) (b) or DE-Suc (5 mM) (c). (d) Numbers of unique and overlapping genes significantly (P < 0.05) upregulated (left) or downregulated (right) in BMDMs stimulated with LPS for 18 h in glutamine-depleted medium then supplemented with glutamine or DM-αKG. (e) Clustering of differentially expressed genes in BMDMs primed with LPS in glutamine-replete (ET w/ Gln) glutamine-depleted (ET w/o Gln) medium or in ET w/o Gln medium + DM-αKG (centered to the mean expression of ET w/o Gln group). Black lines represent expression patterns of individual genes. Red lines indicate mean changes of indicated gene clusters. Data are representative of 3 independent experiments (a–c; mean ± s.d.; *P < 0.05, unpaired, two-tailed Student’s t-test) or are cumulative from 2 independent experiments with 3 samples per group (d,e).

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 8: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

992  VOLUME 18 NUMBER 9 SEPTEMBER 2017 nature immunology

A rt i c l e s

transcriptional events. Because LPS priming can stimulate Jmjd3 expression32, and Jmjd3 promotes expression of IRF4, which is an important regulator for establishment of endotoxin, we tested whether the αKG–Jmjd3 pathway regulates the establishment of endotoxin tolerance. Supplementation with glutamine and DM-αKG treatment suppressed induction of proinflammatory genes in Jmjd3-deficient BMDMs upon re-exposure to LPS (Supplementary Fig. 6f), which suggests that αKG promotes tolerance on LPS-induced proinflam-matory gene expression in a Jmjd3-independent manner. In contrast, supplementation with glutamine and DM-αKG treatment increased the expression of some genes, including Arg1, Mmp9 and Mrc1, which regulate reparative activity, in endotoxin-tolerant control BMDMs but not in Jmjd3-deficient BMDMs (Fig. 7d), which suggests that their induction by DM-αKG was Jmjd3-dependent. Together, these data suggest that αKG produced by glutamine metabolism in macrophages controls induction of endotoxin tolerance via different mechanisms for proinflammatory genes and reparative genes.

To further assess the physiological relevance of endotoxin tol-erance regulated by glutamine metabolism, we pre-injected mice with PBS, a low dose of LPS or a low dose of LPS plus BPTES to induce endotoxin tolerance, then we re-challenged the mice with a lethal dose of LPS 18 h later. Pre-exposure to LPS resulted in pro-longed survival due to the establishment of endotoxin tolerance, as expected, whereas treatment with BPTES abrogated endotoxin tolerance-induced protection (Fig. 7e). However, the addition of DM-αKG to the primary injection of LPS with BPTES markedly improved survival (Fig. 7f). Because BPTES and DM-αKG treat-ments might affect multiple cell types in these assays, we examined induction of endotoxin tolerance in mice injected with clodronate liposome to deplete macrophages and monocytes before first LPS challenge5 (Online Methods). BPTES treatment with or without DM-αKG during primary injection of LPS resulted in similar sur-vival rates as those seen with control vehicle treatment (Fig. 7g), which suggests that targeting glutamine metabolism in macrophages

Trim30a

+Gln

Gln

DM-αKG

Arg1 Mmp9 Mrc1

Rel

ativ

e fo

ld c

hang

e in

mR

NA

exp

ress

ion

Sur

viva

l (%

)– – – – + + + + + +

6060 3030 151560

Z-score–1.8

100

80

60

40

20

0

Sur

viva

l (%

)

100

80

60

40

20

0S

urvi

val (

%)

100

80

60

40

20

0

1.5

1.0

0.5

0

ETET-BPTES

ET-BPTES

ET-BPTES + DM-αKG ET-BPTES

ET-BPTES + DM-αKG

ET

Septic shock

***

1.5

1.0

0.5

0

1.5

2.0

1.0

0.5

0+

+

+

+

––

––

––

*

* * *

**

––

Gln

DM-αKG

+

+

+

+

––

––

––

––

Gln

Ctrl

Jmjd3-sgRNA

DM-αKG

+

+

+

+

––

––

––

––

+1.8

0 3015

1st LPS

WCL

Lamin-A/C

NF-κB

NF-κB

β-actin

Nuclearfraction

2nd LPS (min)

–Gln

Time (h)

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 10 11 0 1 2 3 4 5 6 7 8Time (h) Time (h)

ET w/ G

ln

ET w/o

Gln

ET w/o

Gln

+

DM-α

KG

Irf7Irf4

Trafd1Zfp36Socs3Inpp5d

Irf3Maf1

Homer1Ltbp1II1rl1Bcl3

Dusp1Socs1Nlrc5

Dusp4Mt2

Nlrx1Nr4a2Tnaip3

Kremen1Tfnaip8l2

+++++ + +

603015600 3015

Ctrl DM-αKG

1st LPS

WCL

Lamin-A/C

NF-κB

NF-κB

β-actin

Nuclearfraction

2nd LPS (min)

P < 0.001

a b c

d

e f g

Figure 7 αKG supports endotoxin tolerance in macrophages by modulating NF-κB signal and Jmjd3-dependent regulation. (a) Expression of genes encoding regulators contributing to induction of endotoxin tolerance in BMDMs stimulated with LPS for 18 h in glutamine-replete (ET w/ Gln) or glutamine-depleted (ET w/o Gln) medium or in ET w/o Gln medium plus DM-αKG, assessed by RNA-seq. (b,c) Immunoblot analysis of LPS-primed (1st LPS) or unprimed BMDMs cultured with or without glutamine (b) or DM-αKG (c) after re-stimulation with LPS in glutamine-replete medium. (d) qPCR analysis of relative mRNA expression in control (ctrl) or Jmjd3-deficient (Jmjd3-sgRNA) BMDMs after LPS stimulation under various culture conditions for 18 h. (e) Kaplan–Meier survival curves of mice after injection with vehicle (septic shock), LPS (ET) or LPS plus BPTES (ET-BPTES) followed by injection of LPS 18 h later. *P < 0.02; **P < 0.005, Mantel–Cox test (n = 9–10 mice per group). (f) Kaplan–Meier survival curves of mice after injection with ET-BPTES or ET-BPTES + DM-αKG (ET-BPTES-DM-αKG) followed by injection with LPS 18 h later (n = 10 mice per group). (g) Kaplan–Meier survival curves of macrophage-depleted mice after injection with LPS (ET), ET-BPTES or ET-BPTES-DM-αKG followed by injection with LPS 18 h later (n = 8–11 mice per group). Data are representative of 3 independent experiments (b–d; mean ± s.d.; *P < 0.05, unpaired, two-tailed Student’s t-test) or are cumulative from 2 (a,e,f) or 3 (g) independent experiments.

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 9: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

nature immunology  VOLUME 18 NUMBER 9 SEPTEMBER 2017 993

A rt i c l e s

may modulate induction of endotoxin tolerance in vivo, as we observed here. Together, these results demonstrate that glutamine metabolism during M1 activation acts as a metabolic checkpoint to induce endotoxin tolerance by affecting NF-κB-mediated transcriptional regulation.

DISCUSSIONOur study shows that αKG produced by glutaminolysis is an anti-inflammatory metabolite that augments M2 activation and controls metabolic reprogramming of M2 macrophages through Jmjd3-dependent regulations. Similarly to a metabolic checkpoint used in embryonic stem cells to determine differentiation and pluripo-tency17,23, glutamine metabolism supported either proinflammatory or anti-inflammatory activation in response to the αKG/succinate ratio. Furthermore, glutamine metabolism restricted M1 activation by hampering the NF-κB pathway via an αKG–PHD-dependent mecha-nism to modulate IKKβ activity. Last, targeting glutamine metabolism could prevent induction of endotoxin tolerance and sustain a macro-phage’s proinflammatory potential.

Epigenetic and metabolic reprogramming orchestrate macro-phage polarization and contribute to macrophage functional plas-ticity2,6. However, the mechanisms by which macrophages integrate these complicated cellular activities have not been established. Our study reveals that αKG prevents M1 activation by suppressing IKKβ activation, and this mechanism is controlled via PHD-dependent proline hydroxylation on IKKβ. PHDs belong to the dioxygenase proteins, in which enzyme activity is positively regulated by αKG but antagonized by succinate24. We found that succinate did not antagonize αKG-induced impairment of the IKKβ–NF-κB pathway. Although the molecular details remain unclear, this may result from the different affinity of PHD isoforms for succinate binding and association with IKKβ25. In addition, we showed that αKG acts as a metabolic regulator that instructs M2 macrophages to augment FAO. Jmjd3 has been reported to promote formation of brown adipocytes through epigenetic regulation of genes involved in mitochondrial biogenesis and FAO33. In support of this result, our data indicate that the αKG–Jmjd3 pathway modulates M2 macrophage meta-bolic reprogramming. It is also interesting to note that succinate can stabilize HIF-1α, but αKG destabilizes it8. Therefore, in M1 macro-phages, the reduced αKG/succinate ratio might further strengthen the metabolic program of M1 macrophages by boosting HIF-1α- supported aerobic glycolysis.

We showed that αKG production during the LPS priming phase determines the induction of endotoxin tolerance on proinflam-matory genes; however, this was independent of Jmjd3. It is likely that other epigenetic regulators in the dioxygenase protein family are responsible for αKG-induced endotoxin tolerance. For example, DNA demethylase Tet2, a target of αKG, controls proinflammatory responses in macrophages34. Future studies are needed to examine whether Tet2 or other dioxygenase proteins regulate αKG-induced endotoxin tolerance. The results of such studies could warrant the development of novel approaches to fine-tune macrophage immune responses during sepsis and immune paralysis induced by endotoxin tolerance. Notably, glutaminolysis supports β-glucan-induced trained immunity, a nonspecific immune response that protects against re-infection35,36. Because LPS cannot induce trained immunity but promotes endotoxin tolerance35, it will be of interest to compare the metabolic demands and metabolic fate of glutamine in β-glucan- and LPS-treated macrophages. The results of these comparisons may uncover how these macrophages elicit distinct consequences on immune responses.

Tumor-associated macrophages have been suggested to behave sim-ilarly to endotoxin tolerant macrophages because of their elevated tis-sue-repairing activity and defective proinflammatory activation37–39. Because glutaminolysis supports proliferation and the integrity of TCA cycle in tumor cells40, the antitumor effects of glutaminase inhibitors are under intensive investigation in animal models and clinical trials41. It will be important to explore whether glutaminase inhibitor treat-ment can affect the phenotypes of tumor-associated macrophages and stimulate antitumor immune responses in the tumor microenviron-ment. Furthermore, 2-hydroxyglutarate (2-HG), a chemical analog of αKG, is an oncometabolite produced by mutated isocitrate dehy-drogenase to promote tumor progression42. Given that 2-HG could affect Jmjd3 (ref. 43), it will be interesting to investigate whether production of 2-HG by tumor cells can drive M2-like phenotypes of tumor-associated macrophages. Notably, PHD activity impairs the tumoricidal functions of tumor-specific T cells44. Given that αKG is the activator of PHD enzymes and inhibiting glutaminolysis can sup-press production of αKG, it will be worthwhile to further investigate how glutaminase inhibition affects antitumor responses by T cells in the tumor microenvironment. The results of such studies would pro-vide important information for development of combination therapies using glutaminase inhibitors with cancer immunotherapy.

METHODSMethods, including statements of data availability and any associated accession codes and references, are available in the online version of the paper.

Note: Any Supplementary Information and Source Data files are available in the online version of the paper.

ACKnoWLeDGMenTSWe thank F. Cottard and C.-P. Lin for technical help and P. Romero and C. Hess for discussion. Supported by Swiss National Science Foundation project grant (31003A_163204), the Swiss Institute for Experimental Cancer Research (26075483), the Harry J. Lloyd Charitable Foundation, the Swiss Cancer Foundation (KFS-3949-08-2016) and a Melanoma Research Alliance Young Investigator Award to P.-C.H. S.-M.F. is supported by a Flanders Research Foundation (FWO) research grant and by Eugène Yourassowsky Schenking. J.I. is supported by the University of Lausanne. H.-D.H. is supported by the Ministry of Science and Technology of Taiwan (MOST105-2627-M-009-007 and MOST103-2628-B-009-001-MY3). T.C. is supported by a University of Lausanne FBM PhD fellowship. M.V. is supported by the National scholarship program of the Slovak Republic.

AUTHoR ConTRIBUTIonSP.-S.L., H.W., X.L., T.C., T.T., S.C., G.D.C.,W.-C.C., M.V., C.M., K.D. and J.I. performed experiments. P.-S.L., H.W., S.C., C.-H.C., M.M., H.-D.H., S.-M.F., J.I. and P.-C.H. analyzed results. P.-S.L., H.W. and P.-C.H. designed the studies. P.-S.L. and P.-C.H. wrote the manuscript.

CoMPeTInG FInAnCIAL InTeReSTSThe authors declare no competing financial interests.

reprints and permissions information is available online at http://www.nature.com/reprints/index.html. Publisher’s note: springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

1. Murray, P.J. et al. Macrophage activation and polarization: nomenclature and experimental guidelines. Immunity 41, 14–20 (2014).

2. Ivashkiv, L.B. Epigenetic regulation of macrophage polarization and function. Trends Immunol. 34, 216–223 (2013).

3. Biswas, S.K. & Lopez-Collazo, E. Endotoxin tolerance: new mechanisms, molecules and clinical significance. Trends Immunol. 30, 475–487 (2009).

4. Sawhney, S., Woo, P. & Murray, K.J. Macrophage activation syndrome: a potentially fatal complication of rheumatic disorders. Arch. Dis. Child. 85, 421–426 (2001).

5. Ho, P.C., Tsui, Y.C., Feng, X., Greaves, D.R. & Wei, L.N. NF-κB-mediated degradation of the coactivator RIP140 regulates inflammatory responses and contributes to endotoxin tolerance. Nat. Immunol. 13, 379–386 (2012).

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 10: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

994  VOLUME 18 NUMBER 9 SEPTEMBER 2017 nature immunology

6. O’Neill, L.A. & Pearce, E.J. Immunometabolism governs dendritic cell and macrophage function. J. Exp. Med. 213, 15–23 (2016).

7. O’Neill, L.A., Kishton, R.J. & Rathmell, J. A guide to immunometabolism for immunologists. Nat. Rev. Immunol. 16, 553–565 (2016).

8. Tannahill, G.M. et al. Succinate is an inflammatory signal that induces IL-1β through HIF-1α. Nature 496, 238–242 (2013).

9. Liu, L. et al. Proinflammatory signal suppresses proliferation and shifts macrophage metabolism from Myc-dependent to HIF1α-dependent. Proc. Natl. Acad. Sci. USA 113, 1564–1569 (2016).

10. Mills, E.L. et al. Succinate Dehydrogenase Supports Metabolic Repurposing of Mitochondria to Drive Inflammatory Macrophages. Cell 167, 457–470.e413 (2016).

11. Huang, S.C. et al. Cell-intrinsic lysosomal lipolysis is essential for alternative activation of macrophages. Nat. Immunol. 15, 846–855 (2014).

12. Vats, D. et al. Oxidative metabolism and PGC-1beta attenuate macrophage-mediated inflammation. Cell Metab. 4, 13–24 (2006).

13. Covarrubias, A.J. et al. Akt-mTORC1 signaling regulates Acly to integrate metabolic input to control of macrophage activation. eLife 5, e11612 (2016).

14. Huang, S.C. et al. Metabolic Reprogramming Mediated by the mTORC2-IRF4 Signaling Axis Is Essential for Macrophage Alternative Activation. Immunity 45, 817–830 (2016).

15. Jha, A.K. et al. Network integration of parallel metabolic and transcriptional data reveals metabolic modules that regulate macrophage polarization. Immunity 42, 419–430 (2015).

16. Fendt, S.M. et al. Reductive glutamine metabolism is a function of the α-ketoglutarate to citrate ratio in cells. Nat. Commun. 4, 2236 (2013).

17. Carey, B.W., Finley, L.W., Cross, J.R., Allis, C.D. & Thompson, C.B. Intracellular α-ketoglutarate maintains the pluripotency of embryonic stem cells. Nature 518, 413–416 (2015).

18. Bricker, D.K. et al. A mitochondrial pyruvate carrier required for pyruvate uptake in yeast, Drosophila, and humans. Science 337, 96–100 (2012).

19. Ishii, M. et al. Epigenetic regulation of the alternatively activated macrophage phenotype. Blood 114, 3244–3254 (2009).

20. Satoh, T. et al. The Jmjd3-Irf4 axis regulates M2 macrophage polarization and host responses against helminth infection. Nat. Immunol. 11, 936–944 (2010).

21. Kruidenier, L. et al. A selective jumonji H3K27 demethylase inhibitor modulates the proinflammatory macrophage response. Nature 488, 404–408 (2012).

22. Platt, R.J. et al. CRISPR-Cas9 knockin mice for genome editing and cancer modeling. Cell 159, 440–455 (2014).

23. TeSlaa, T. α-Ketoglutarate Accelerates the Initial Differentiation of Primed Human Pluripotent Stem Cells. Cell Metab. 24, 485–493 (2016).

24. Fraisl, P., Aragonés, J. & Carmeliet, P. Inhibition of oxygen sensors as a therapeutic strategy for ischaemic and inflammatory disease. Nat. Rev. Drug Discov. 8, 139–152 (2009).

25. Cummins, E.P. et al. Prolyl hydroxylase-1 negatively regulates IkappaB kinase-beta, giving insight into hypoxia-induced NFkappaB activity. Proc. Natl. Acad. Sci. USA 103, 18154–18159 (2006).

26. Takeda, Y. et al. Macrophage skewing by Phd2 haplodeficiency prevents ischaemia by inducing arteriogenesis. Nature 479, 122–126 (2011).

27. Rodríguez-Prados, J.C. et al. Substrate fate in activated macrophages: a comparison between innate, classic, and alternative activation. J. Immunol. 185, 605–614 (2010).

28. Shalova, I.N. et al. Human monocytes undergo functional re-programming during sepsis mediated by hypoxia-inducible factor-1α. Immunity 42, 484–498 (2015).

29. Su, X. et al. Interferon-γ regulates cellular metabolism and mRNA translation to potentiate macrophage activation. Nat. Immunol. 16, 838–849 (2015).

30. Cheng, S.C. et al. Broad defects in the energy metabolism of leukocytes underlie immunoparalysis in sepsis. Nat. Immunol. 17, 406–413 (2016).

31. Chen, J. & Ivashkiv, L.B. IFN-γ abrogates endotoxin tolerance by facilitating Toll-like receptor-induced chromatin remodeling. Proc. Natl. Acad. Sci. USA 107, 19438–19443 (2010).

32. De Santa, F. et al. Jmjd3 contributes to the control of gene expression in LPS-activated macrophages. EMBO J. 28, 3341–3352 (2009).

33. Pan, D. et al. Jmjd3-Mediated H3K27me3 Dynamics Orchestrate Brown Fat Development and Regulate White Fat Plasticity. Dev. Cell 35, 568–583 (2015).

34. Zhang, Q. et al. Tet2 is required to resolve inflammation by recruiting Hdac2 to specifically repress IL-6. Nature 525, 389–393 (2015).

35. Saeed, S. et al. Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity. Science 345, 1251086 (2014).

36. Arts, R.J. et al. Glutaminolysis and Fumarate Accumulation Integrate Immunometabolic and Epigenetic Programs in Trained Immunity. Cell Metab. 24, 807–819 (2016).

37. del Fresno, C. et al. Tumor cells deactivate human monocytes by up-regulating IL-1 receptor associated kinase-M expression via CD44 and TLR4. J. Immunol. 174, 3032–3040 (2005).

38. Hagemann, T. et al. “Re-educating” tumor-associated macrophages by targeting NF-kB. J. Exp. Med. 205, 1261–1268 (2008).

39. Mantovani, A. & Sica, A. Macrophages, innate immunity and cancer: balance, tolerance, and diversity. Curr. Opin. Immunol. 22, 231–237 (2010).

40. Wise, D.R. & Thompson, C.B. Glutamine addiction: a new therapeutic target in cancer. Trends Biochem. Sci. 35, 427–433 (2010).

41. Altman, B.J., Stine, Z.E. & Dang, C.V. From Krebs to clinic: glutamine metabolism to cancer therapy. Nat. Rev. Cancer 16, 619–634 (2016).

42. Ye, D., Ma, S., Xiong, Y. & Guan, K.L. R-2-hydroxyglutarate as the key effector of IDH mutations promoting oncogenesis. Cancer Cell 23, 274–276 (2013).

43. Black, J.C., Van Rechem, C. & Whetstine, J.R. Histone lysine methylation dynamics: establishment, regulation, and biological impact. Mol. Cell 48, 491–507 (2012).

44. Clever, D. et al. Oxygen sensing by T cells establishes an immunologically tolerant metastatic niche. Cell 166, 1117–1131.e14 (2016).

A rt i c l e s

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 11: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

nature immunologydoi:10.1038/ni.3796

ONLINE METHODSCell culture and differentiation of BMDMs. OT-1 CD8+ T cells were isolated from splenocytes of OT-1 mice and cultured in RPMI medium with 10% FBS and β-mercaptoethanol. To activate OT-1 cells, OT-1 splenocytes were treated with OVA-peptide and IL-2 for 3 d, then cultured in the presence of IL-2 for another 3 d before adoptive transfer into tumor-bearing mice. Bone marrow cells were collected and cultured in DMEM supplemented with 10% FBS and 20% L929 cell culture supernatant for macrophage differentiation for 7 d. On day 7, differentiated BMDMs were re-plated with DMEM (without L929 cell culture supernatant) overnight and then stimulated with 10 ng/ml LPS for M1 activation and 20 ng/ml IL-4 (Bioconcept; 214-14) for M2 activation with indicated duration.

RNA purification, RT-PCR, qPCR, NF-kB signal target gene quantitative PCR array, RNA-seq and bioinformatic analysis. Total RNAs from BMDMs were isolated with TRIzol reagent (Life Technologies). 1 µg total RNA was con-verted into cDNA using First Strand cDNA synthesis kit (Life Technologies). qPCR was performed in triplicate on a LightCycler 480 Instrument II machine (Roche Life Science) using SYBR Green PCR mixture (KAPA Biosystems) for quantification of the target gene expression. Relative expression was nor-malized to β-actin for each sample. The primers for qRT-PCR amplification are summarized in Supplementary Table 1. For NF-κB signal target gene PCR array, RNA was extracted by using RNA Miniprep Plus Kit (Direct-zol, ZymoResearch) and then converted into cDNA with RT2 First Strand Kit (Qiagen). RT2 Profiler qPCR array for NF-κB signaling targets was pur-chased from Qiagen (PAMM-225Z). Real-time PCR was performed according to the manufacturer’s instructions with a Roche Light Cycler 480 detector. Data were analyzed by ∆∆Ct method to normalize with GAPDH expression. Fold regulation in gene expression was calculated using the analysis web software from Qiagen.

For next-generation RNA-seq, total RNA was extracted with TRIzol and RNeasy Mini Kit (Qiagen). mRNAs were then isolated from purified DNA-free RNA for library construction. Libraries were next sequenced on an Illumina HISEQ 2500 (Illumina). Mappable results were analyzed using the DESeq2 package. A minimum expression of each gene (mean of counts > 20) was applied as a cutoff before analysis. Significantly differentially expressed genes was defined as a 1.5-fold change with a false discovery ratio (FDR) ≤ 0.05.

Chromatin immunoprecipitation (ChIP). ChIP was performed using the ChIP Assay Kit (Millipore) according to the manufacturer’s instructions. Briefly, 1 × 107 macrophages were fixed with 10% formaldehyde to cross-link histones to DNA. Cells were lysed with hypotonic buffer (0.3% NP40, 0.1 mM EDTA, 10 mM HEPES (pH 7.9), 10 mM KCl) to enrich nuclei. Chromatin was sheared by micrococcal nuclease (Mnase; NEB (Bioconcept); M0247S) for 10 min at 37 °C, and the reactions were stopped by addition of sonication buffer (1% Triton X-100, 1 mM EDTA, 50 mM HEPES (pH 7.9), 0.4 M NaCl and proteinase inhibitor cocktail). The soluble chromatin supernatant was immu-noprecipitated with anti-H3K27me3 (Millipore 07-449). Immunoprecipitated DNA and input DNA were analyzed by q-RT-PCR, and results are presented as percentage of input. The primers were used for amplification of promoters of M2 gene are summarized in Supplementary Table 2.

Plasmids, reagents and antibodies. The retro-gfpIkkb-puro vector was purchased from Addgene (#58251). The P191A-IKKβ mutant was generated using Agilent QuikChange II site-directed mutagenesis kit according to the manufacturer’s instructions. The lentiviral JMJD3 sgRNA expression vector was constructed using a BbsI site on pKLV-U6gRNA(BbsI)-PGK-puro2ABFP (Addgene #50946). The primers used for cloning the IKKβ P191A mutation and Jmjd3 sgRNA vectors are listed in Supplementary Table 3. The follow-ing chemical reagents were used in this study: LPS (Sigma; L4391), mouse IL-4 (Peprotech; 214-14), BPTES (Sigma; SML0601), glutaminase inhibitor 968 (Sigma; SML1327), DMOG (Sigma; D3695), GSK4-J4 (Sigma; SML0701), diethyl-succinate (Sigma; 112402), dimethyl αKG (Sigma; 349631), etomoxir (Sigma; E1905), oligomycin A (Sigma; 75351), FCCP (Sigma; C2920), roten-one (Sigma; R8875), and antimycin A (Sigma; A8674). Antibodies used in this study were purchased from the following sources: NF-κB p65 (Santa Cruz; sc-8008, 1:1,000), lamin A/C (Santa Cruz; sc-6215, 1:5,000), β-actin

(Sigma; A5316, 1:5,000), IKKβ (Cell Signaling; 2678P, 1:1,000), phospho-IKKα/β (Cell Signaling; 2694P, 1:1,000), JNK (Cell Signaling ; 9252, 1:2,000), phospho-JNK (Cell Signaling; 9251, 1:1,000); p38α (Cell Signaling; 9218, 1:2,000), phospho-p38 (Cell Signaling; 9211, 1:1,000), phospho-p44/42 ERK (Cell Signaling; 4370S, 1:5,000), ERK (Cell Signaling; 4696, 1:5,000), and Jmjd3 (Cell Signaling; 3457, 1:1,000).

Imaging flow analysis of NF-κB translocation. After stimulation, macro-phages were fixed with 4% paraformaldehyde at room temperature for 10 min and then incubated with anti-NF-κB p65 (sc-8008, Santa Cruz) in permea-bilization buffer for 30 min. After washing, cells were then incubated with anti-mouse antibody conjugated with Alexa Fluor 647 (A-21235, Thermo Fisher Scientific) and anti-GFP conjugated with Alexa Fluor 488 (338008, BioLegend) with DAPI. These cells then were resuspended in 1% paraformal-dehyde, and NF-κB nuclear translocation was analyzed by an ImageStreamX (Amnis Corporation) flow cytometer. The quantification of nuclear transloca-tion was done with IDEAS Software.

Animals, in vitro and in vivo endotoxin tolerance assays. C57BL/6, LysM-Cre (B6.129p2-Lyz2tm1(cre)lfo/J) and Cas9 knock-in (B6J.129(B6N)-Gt(ROSA)26Sortm1(CAG-cas9-EGFP)Fezh/J) mice were purchased from Jackson Laboratory and maintained at the animal facility of University of Lausanne. We generated macrophage-specific Cas9 knock-in mice by crossing LysM-Cre mice (myeloid-specific overexpression of Cre recombinase) with Cas9 knock-in mice. All experiments were performed in accordance with Swiss federal regulations and procedures approved by veterinary authority of Canton Vaud.

For the in vitro endotoxin tolerance assay, BMDMs were pretreated with 10 ng/ml LPS under various culture conditions. After 18 h, cells were washed with PBS once then incubated in glutamine-replete medium with 10% FBS. After 1 h recovery, cells were restimulated with 10 ng/ml LPS to examine their proinflammatory responses. For in vivo endotoxin tolerance, C57BL/6J mice were pre-injected with PBS or BPTES (12.5 mg per kg body weight). In the DM-αKG treatment, mice also received control vehicle or DM-αKG (0.6 g per kg body weight). 1 h later, mice were injected intraperitoneally with LPS (0.1 µg per 25 g body weight), to induce endotoxin tolerance, or PBS, as control treatment. After 18 h, mice were injected intraperitoneally with LPS (100 µg per 25 g of body weight) plus d-galactosamine (0.5 mg/g of body weight) to induce acute septic shock. In the experiment with macrophage depletion, mice were injected intravenously with 200 µl clodronate-containing liposomes (ClodronateLiposomes). 24 h later, mice were injected intraperitoneally with 100 µl clodronate-containing liposomes. 6 h later, mice were treated with control vehicle, BPTES or BPTES + DM-αKG. After 1 h, mice were treated to examine in vivo endotoxin tolerance as described above.

Fatty acid uptake assay and seahorse extracellular flux analysis. BMDMs were plated in 24-well plates and allowed to attach overnight. After stimulation with IL-4 for 24 h, cell were washed with DMEM and then incubated with or without 0.5 µM BODIPY 500/510 C1, C12 (Thermo Fisher; D3823) in DMEM with 10% dialyzed FBS for 30 min at 37 °C incubator. Fatty acid uptake was determined by measuring fluorescence intensity within cells with flow cytom-etry and analyzed by FlowJo. For extracellular flux assay, 1 × 105 BMDMs were plated in a Seahorse Bioscience culture plate for overnight. Cells were then activated for 6 h. OCR was measured by an XF96 Seahorse Extracellular Flux Analyzer following the manufacturer’s instruction. In seahorse assay, cells were treated with oligomycin (4 µM), FCCP (1.6 µM), rotenone (0.5 µM), antimycin A (0.5 µM), UK5099 (3.6 µM) or etomoxir (18 µM). Each condition was performed with 4–6 replicates, and the readings of OCR of each well were normalized to protein amount.

Untargeted and targeted metabolomic analyses. To assess targeted metabo-lite levels, treated BMDMs were washed once with 0.9% saline and quenched in liquid nitrogen. The cells were then extracted with water–methanol and chlo-roform as described45. The dried samples with the polar metabolite fraction were derivatized for 90 min at 37 °C with 20 mg/ml methoxyamine in pyridine. Subsequently, 7.5 µl was transferred into glass vials and derivatized for 60 min at 60 °C with 15 µl N-(tert-butyldimethylsilyl)-N-methyltrifluoroacetamide,

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.

Page 12: a-ketoglutarate orchestrates macrophage activation through ...crid.fmrp.usp.br/wp-content/...orchestrates-macrophage-activation_2017.pdf · 8Laboratory of Tumor Inflammation and Angiogenesis,

nature immunology doi:10.1038/ni.3796

with 1% tert-butyldimethylchlorosilane. 1 µl of sample was injected into a 7890A GC system (Agilent Technology) combined with a 5975C Inert MS system (Agilent Technology) to determine expression levels of metabolites.

For untargeted metabolic profiling, cells were extracted using a pre-cooled methanol and H2O (4:1, v/v) solvent mixture to extract a broad range of polar and nonpolar metabolites. After lysing cells and spinning down protein pellets, the resulting supernatant was collected and dehydrated in a vacuum con-centrator (LabConco). The dry metabolome extracts were reconstituted in acetonitrile and H2O (1:1, v/v), sonicated for 30 s, and centrifuged 15 min at 13,000 r.p.m. and 4 °C to remove insoluble debris. The supernatants were ana-lyzed by HILIC ESI-Q-TOF-MS in MS only and auto MS/MS (data-depend-ent analysis (DDA)) acquisition mode on LC-MS 6550 iFunnel Q-TOF mass spectrometer interfaced with 1290 UPLC system (Agilent Technologies). Raw LC/MS data were converted to mzXML files using ProteoWizard and then uploaded to the XCMS Plus server platform for data processing including peak detection, retention time correction, profile alignment and isotope annotation. Data were processed as a two-group and multi-group experiments and the parameter settings were as follows: centWave for feature detection (∆m/z = 15 p.p.m., minimum peak width = 5 s and maximum peak width = 30 s); obi-warp settings for retention time correction (prof Step = 1); and parameters for chromatogram alignment, including mzwid = 0.015, minfrac = 0.5 and bw = 5. The relative quantification of metabolite features was based on EIC (Extracted Ion Chromatogram) areas. An orthogonal partial least-squares discriminant

analysis (OPLS-DA) was employed for modeling the discrimination among groups and to determine the metabolite features (VIP variables) that drove the separation. Multivariate analysis was performed using both SIMCA-P+ version 12 (Umetrics AB) and MetaboAnalyst 3.0. In addition, one-way analysis of variance (ANOVA) was used to filter out the significantly altered metabo-lite features among groups. Accurate masses were searched against HMDB and METLIN databases. Targeted MS/MS analyses were further performed to obtain the high-quality MS/MS data for significantly altered metabolite features (or ions) of interest. Finally, metabolite identification was carried out by matching the acquired MS/MS data from macrophage cell extracts against recorded MS/MS data for standards, assembled in METLIN database (https://metlin.scripps.edu/index.php).

Statistical analysis. All results are presented as mean ± s.d. and analyzed for statistical significance by an unpaired Student’s t-test. P < 0.05 was considered statistically significant.

Data availability. Data have been deposited in the Gene Expression Omnibus under accession code GSE99296. Other data that support the findings of this study are available from the corresponding author upon request.

45. Christen, S. et al. Breast cancer-derived lung metastases show increased pyruvate carboxylase-dependent anaplerosis. Cell Rep. 17, 837–848 (2016).

© 2

017

Nat

ure

Am

eric

a, In

c., p

art

of

Sp

rin

ger

Nat

ure

. All

rig

hts

res

erve

d.