LETTER

doi:10.1038/nature13322

Negative regulation of the NLRP3 inflammasome by A20 protects against arthritis Lieselotte Vande Walle1,2, Nina Van Opdenbosch1,2, Peggy Jacques3, Amelie Fossoul1,2, Eveline Verheugen3, Peter Vogel4, Rudi Beyaert5,6, Dirk Elewaut3, Thirumala-Devi Kanneganti4, Geert van Loo5,6* & Mohamed Lamkanfi1,2*

Rheumatoid arthritis is a chronic autoinflammatory disease that affects 1–2% of the world’s population and is characterized by widespread joint inflammation. Interleukin-1 is an important mediator of cartilage destruction in rheumatic diseases1, but our understanding of the upstream mechanisms leading to production of interleukin-1b in rheumatoid arthritis is limited by the absence of suitable mouse models of the disease in which inflammasomes contribute to pathology. Myeloid-cell-specific deletion of the rheumatoid arthritis susceptibility gene A20/Tnfaip3 in mice (A20myel-KO mice) triggers a spontaneous erosive polyarthritis that resembles rheumatoid arthritis in patients2. Rheumatoid arthritis in A20myel-KO mice is not rescued by deletion of tumour necrosis factor receptor 1 (ref. 2). Here we show, however, that it crucially relies on the Nlrp3 inflammasome and interleukin-1 receptor signalling. Macrophages lacking A20 have increased basal and lipopolysaccharide-induced expression levels of the inflammasome adaptor Nlrp3 and proIL-1b. As a result, A20deficiency in macrophages significantly enhances Nlrp3 inflammasome-mediated caspase-1 activation, pyroptosis and interleukin-1b secretion by soluble and crystalline Nlrp3 stimuli. In contrast, activation of the Nlrc4 and AIM2 inflammasomes is not altered. Importantly, increased Nlrp3 inflammasome activation contributes to the pathology of rheumatoid arthritis in vivo, because deletion of Nlrp3, caspase-1 and the interleukin-1 receptor markedly protects against rheumatoid-arthritis-associated inflammation and cartilage destruction in A20myel-KO mice. These results reveal A20 as a novel negative regulator of Nlrp3 inflammasome activation, and describe A20myel-KO mice as the first experimental model to study the role of inflammasomes in the pathology of rheumatoid arthritis. A20 was deleted in myeloid cells by crossing A20 flox/flox mice into lysozyme M (LysM)-Cre-recombinase-expressing mice. Unlike wild-type macrophages, A20myel-KO macrophages failed to induce A20 messenger RNA (mRNA) and protein expression in response to the Toll-like receptor-4 (TLR4) ligand lipopolysaccharide (LPS) (Fig. 1a, b), demonstrating the effectiveness of LysM-driven deletion of A20 in myeloid cells. Arthritis development in A20myel-KO mice was shown to be independent of tumour necrosis factor receptor 1 (TNF-R1), whereas deletion of MyD88 markedly protected against pathology of rheumatoid arthritis2. As this signalling adaptor operates downstream of both TLRs and interleukin-1 receptor (IL-1R), we crossed Il1r12/2 mice into A20myel-KO mice to assess the contribution of IL-1 signalling to arthritis pathogenesis. As expected, wild-type mice (A20 flox/floxIl1r11/1) did not develop arthritis, whereas A20myel-KO mice spontaneously developed an arthritic phenotype (Fig. 1c). The incidence of A20myel-KO mice developing arthritis was 100% (Fig. 1d). In sharp contrast, A20myel-KOIl1r12/2 mice were virtually devoid of clinical signs of arthritis (Fig. 1c, d). In agreement, clinical scoring of disease severity confirmed A20myel-KO Il1r11/1 mice as developing severe arthritic disease (high clinical score) whereas A20myel-KOIl1r12/2 mice were markedly protected (clinical

score 0). A20myel-KO littermates that were heterozygous for IL-1R1 expression showed an intermediate arthritic phenotype between those of A20myel-KOIl1r11/1 and A20myel-KOIl1r12/2 mice (Fig. 1d, e). These clinical assessments were supported by a histological examination of representative ankle joints. Tissue sections of diseased A20myel-KOIl1r11/2 mice stained by haematoxylin and eosin showed significant synovial and periarticular inflammation and high levels of infiltrated mononuclear cells, which was associated with extensive cartilage and bone destruction (Fig. 1f). In marked contrast, ankle joints of A20myel-KOIl1r12/2 littermates were strongly protected from arthritic histopathology and contained significantly reduced numbers of infiltrating inflammatory cells (Fig. 1f). Semi-quantitative scoring of these histological parameters confirmed that the severity of arthritis was substantially lower in A20myel-KOIl1r12/2 mice relative to A20myel-KOIl1r11/2 littermates (Fig. 1g and Extended Data Table 1). These results demonstrate that IL-1 production is detrimental for arthritis pathogenesis in mice with a myeloid cell-restricted deletion in A20. Macrophages are a prime source of proIL-1b, and generally depend on caspase-1 for maturation and secretion of the biologically active cytokine. Caspase-1 is produced as a cytosolic zymogen, the activation of which is controlled by different inflammasomes3. To study the role of A20 in inflammasome signalling, we assessed caspase-1 processing in bone-marrow-derived macrophages (BMDMs) of wild-type and A20myel-KO mice. Notably, caspase-1 activation was substantially increased in LPS-primed A20myel-KO macrophages that were treated with soluble (ATP and nigericin) or crystalline (silica) stimuli of the Nlrp3 inflammasome compared with wild-type BMDMs (Fig. 2a). Concurrently, the levels of secreted IL-1b in the culture medium of ATP- and nigericintreated A20myel-KO macrophages were about twice those of wild-type cells, and silica triggered nearly four times higher levels of secreted IL1b (Fig. 2b). Enhanced caspase-1 autoprocessing (Fig. 2c, d) and IL-1b secretion (Fig. 2e, f) by the Nlrp3 inflammasome was evident within 10 min after ATP or nigericin addition, and continued to increase in a time-dependent fashion. Similarly, the induction of caspase-1-dependent pyroptosis was enhanced in A20myel-KO macrophages (Fig. 2g, h). Despite their hypersensitivity for Nlrp3 inflammasome activation, A20myel-KO macrophages failed to process caspase-1 and secrete IL-1b and IL-18 when treated with LPS, ATP or nigericin alone (Extended Data Fig. 1). The increased responsiveness of A20myel-KO macrophages towards inflammasome activation was restricted to the Nlrp3 inflammasome because caspase-1 processing and pyroptotic cell death by the Nlrc4 inflammasome were similarly induced in wild-type and A20myel-KO macrophages that had been infected with Salmonella enterica serovar Typhimurium (Fig. 2i, j). Similarly, stimulation of the AIM2 inflammasome with cytosolic double-stranded DNA (dsDNA) did not result in differential levels of caspase-1 processing and pyroptosis induction in wild-type and A20myel-KO macrophages (Fig. 2k, l). It is worth noting, however, that despite normal caspase-1 activation and pyroptosis levels in response

1

Department of Medical Protein Research, VIB, Ghent B-9000, Belgium. 2Department of Biochemistry, Ghent University, Ghent B-9000, Belgium. 3Department of Rheumatology, Ghent University, Ghent B-9000, Belgium. 4Department of Immunology, St. Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA. 5Inflammation Research Center, VIB, Ghent B-9052, Belgium. 6Department of Biomedical Molecular Biology, Ghent University, Ghent B-9052, Belgium. *These authors contributed equally to this work. 0 0 M O N T H 2 0 1 4 | VO L 0 0 0 | N AT U R E | 1

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER c

a

d

e 2.5

100

50

0.0

β-actin

+/

f

– +/

–/–

+ +/

A20myel-KOIl1r1+/–

A20myel-KOIl1r1–/–

200 μm

200 μm

r1

r1

–/–

r1

0 0 0 A2 A2 A2

g 8 P = 0.046

Histological score

87 kDa 5 μm

– +/

+ +/

r1

l1 l1 l1 KO Il1 fl/fl I KO I KO I 0 yel- myel- myelA2 m

A20myel-KO

6h A20fl/fl

A20myel-KO

A20fl/fl A20

100 75

+

+/

1 1 1 r1 l1r l1r l1r KO Il1 fl/fl I KO I KO I 0 myel- myel- myel2 A 0 0 0 A2 A2 A2

b 0h

1.0 0.5

+

LPS

P = 0.034 1.5

0

240 bp

P < 0.0001

2.0

Clinical score

Clinical incidence (%)

240 bp

A20myel-KOIl1r1–/–

A20fl/flIl1r1+/+

A20

605 bp

A20myel-KOIl1r1+/+

150

A20myel-KO

A20fl/fl

6h

A20myel-KO

0h A20fl/fl

LPS

5 μm

6

4

2

β-actin

0 50

– +/

r1

42 kDa

el-

my

0

A2

l1 KO I

my

0

A2

–/–

r1

l1 KO I el-

Figure 1 | Il1r1 deficiency rescues the arthritis phenotype of A20myel-KO mice. a, b, A20 and b-actin mRNA (a) and protein (b) levels of LPS-stimulated BMDMs. c, Hind paws of 20-week-old mice. d, e, A20fl/flIl1r11/1 (n 5 19), A20myel-KOIl1r11/1 (n 5 8), A20myel-KOIl1r11/2 (n 5 15) and A20myel-KO Il1r12/2 (n 5 9) mice aged 21–30 weeks were clinically scored for arthritis

incidence (d) and severity (e). f, Ankle joint sections stained with haematoxylin and eosin; magnification 340 (top) and 3100 (bottom). g, Histological scores of ankle sections of A20myel-KOIl1r11/2 (n 5 10) and A20myel-KOIl1r112/2 (n 5 8). P values in e and g were determined by Student’s t-test.

to S. enterica serovar Typhimurium infection and dsDNA transfection, secretion of IL-1b in response to these treatments was consistently higher in A20myel-KO macrophages compared with wild-type controls (Fig. 2m, n), which could be explained by increased induction of proIL-1b mRNA in A20myel-KO macrophages (data not shown). Together, these results demonstrate that A20 negatively regulates activation of caspase-1 by the Nlrp3 inflammasome, but not by the Nlrc4 and AIM2 inflammasomes. Activation of the Nlrp3 inflammasome in wild-type macrophages is tightly regulated at different levels. A priming signal (referred to as step 1 and usually provided by TLRs) upregulates Nlrp3 expression levels along with the inflammasome substrate proIL-1b via the pro-inflammatory transcription factor NF-kB4. A20 negatively regulates LPS-induced NF-kB activation (refs 5–8 and Extended Data Fig. 2a), which was also reflected in increased secretion of the NF-kB-dependent cytokines IL-6 and TNF in A20myel-KO macrophages (Extended Data Fig. 2b, c). We further showed A20-deficiency to markedly enhance LPS-induced mRNA expression levels of Nlrp3 (Fig. 3a) and proIL-1b (Fig. 3b). In contrast, LPS-induced transcript levels of caspase-1 and the inflammasome adaptor ASC were respectively mildly upregulated and normal in A20myel-KO macrophages (Extended Data Fig. 3a, b). Analysis of protein expression levels confirmed Nlrp3 and proIL-1b to be significantly higher in LPSprimed A20myel-KO macrophages than wild-type cells, whereas caspase-1 and ASC were not differentially modulated in the two genotypes (Fig. 3c). TLR stimulation during brief time intervals (10 min or less) was recently shown to license activation of the Nlrp3 inflammasome independently of new protein synthesis9–12. Rapid Nlrp3 inflammasome activation resulted in procaspase-1 processing and secretion of pre-synthesized proIL-18 in the absence of the NF-kB-dependent cytokines IL-1b, TNF and IL-6 (refs 9–11). To address whether A20 modulated rapid Nlrp3 inflammasome activation, cells were exposed to ATP or nigericin after being primed with LPS for 10 min. As reported9–12, wild-type BMDMs

activated caspase-1 (Fig. 3d) and secreted significant amounts of IL-18, but not IL-1b, TNF or IL-6 (Fig. 3e). Moreover, we noted Nlrp3 protein levels were lowered after stimulation both in wild-type and A20myel-KO macrophages (Fig. 3d), supporting the notion that acute Nlrp3 inflammasome activation occurred independently of LPS-induced new protein synthesis. Both caspase-1 processing and IL-18 secretion were markedly increased in A20myel-KO macrophages in the absence of substantial IL1b, TNF and IL-6 secretion (Fig. 3d, e). This was probably due to increased basal expression of Nlrp3 and proIL-18 in these cells (Fig. 3c, d). In agreement, basal mRNA levels of Nlrp3, proIL-1b and proIL-18 were increased in untreated A20myel-KO macrophages (Fig. 3a, b and Extended Data Fig. 3c). Together, these results suggest that A20 negatively regulates Nlrp3 inflammasome signalling by suppressing NF-kB-dependent production of Nlrp3 and the inflammasome substrates proIL-1b and proIL-18. In agreement, the pharmacological inhibitor of kappa B kinase (IKK) inhibitor BMS-345541 significantly reduced Nlrp3 levels in LPSprimed A20myel-KO macrophages (Fig. 3f). Moreover, both BMS-345541 and the selective IKK2 inhibitor TCPA-1 significantly reduced ATPand nigericin-induced caspase-1 autoprocessing, IL-1b secretion and pyroptosis induction in LPS-primed A20myel-KO macrophages (Fig. 3g–i). Having established A20 as a negative regulator of Nlrp3 inflammasome activation, we hypothesized that excessive Nlrp3 activation might drive pathology of rheumatoid arthritis in A20-deficient mice upstream of IL-1R1. To test this hypothesis, Nlrp32/2 mice were crossed into A20myel-KO mice and the levels of four cytokines (IL-1a, IL-1b, IL-6 and TNF) relevant to rheumatoid arthritis were monitored. Although IL1a levels were not significantly different in A20-sufficient and A20myel-KO mice (Extended Data Fig. 4a), the latter group of mice had significantly higher levels of IL-1b in circulation (Fig. 4a). In addition, the levels of IL-6 and TNF were also significantly higher in A20myel-KO mice compared with A20 flox/flox littermates (Extended Data Fig. 4b, c). Notably,

2 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 4

©2014 Macmillan Publishers Limited. All rights reserved

20 0

on C

Cytokine (pg ml–1)

WT A20myel-KO

***

Relative proIL-1β expression 400 300 200

*

100

Figure 2 | Hyperactivation of the Nlrp3 but not the Nlrc4 and AIM2 inflammasomes, in A20-deficient macrophages. a–h, BMDMs were stimulated as described in Methods. Lysates were immunoblotted for caspase-1 (a, c, d) and supernatants analysed for IL-1b (b, e, f) and LDH (g, h). Nig, nigericin. i–n, BMDMs were treated as described in Methods. Lysates were immunoblotted for caspase-1 (i, k) and supernatants analysed for LDH (j, l) and IL-1b (m, n). Black arrow, procaspase-1; white arrow, p20. MOI, multiplicity of infection. Data represent mean 6 s.d. of one out of three biological replicates, with three technical replicates each (*P , 0.05; ***P , 0.001; Student’s t-test).

deletion of Nlrp3 in A20myel-KO mice markedly reduced IL-1b secretion to baseline levels of A20 flox/flox mice, thereby demonstrating that the Nlrp3 inflammasome contributes critically to excessive IL-1b production in A20myel-KO mice in vivo (Fig. 4a). Intriguingly, A20myel-KONlrp32/2 mice also were protected from excessive IL-6 production, suggesting that high IL-6 levels are consequent to excessive inflammasome-mediated IL-1b production (Extended Data Fig. 4b). In contrast, TNF production was not significantly affected in A20myel-KONlrp32/2 mice (Extended Data Fig. 4c). In agreement, TNF-R1 signalling was previously shown to be dispensable for pathology of rheumatoid arthritis in A20myel-KO mice, whereas IL-6 neutralization provided protection2. Based on these findings, we assessed the contribution of Nlrp3 to pathogenesis of rheumatoid arthritis in A20myel-KO mice. Swelling and redness of the hind paws of A20myel-KONlrp31/1 mice became evident around 11 weeks of age (Fig. 4b), and had afflicted all animals of this genotype when they were 20 weeks old (Fig. 4c). Disease severity continued to progress, and became increasingly pronounced in A20myel-KO Nlrp31/1 mice aged 21–40 weeks (Fig. 4d). In contrast, A20myel-KONlrp32/2 mice of similar age were markedly protected from rheumatoid arthritis,

Relative Nlrp3 expression

A20myel-KO

A20myel-KO

A20myel-KO

LPS

100

IL-18

TNF

IL-1β

*

80

***

60 40 20 0 1.25 2.5

C LP on t LP S + rol S N + ig A C LP on TP t LP S + rol S N + ig C AT LP on P t LP S + rol S N + ig C AT LP on P tr LP S + ol S N + ig AT P

p24 p18

DMSO BMS-345541 Control

LPS

IL-6

37

*** ***

15,000 10,000 5,000 0

DMSO BMS-345541 TPCA-1 *** *** *** ***

40 30 20 10

ig

l

P LP

S

+

N

AT

tro

LP

S

+

on

P LP

S

+

AT + S

LP

N ig

0 l

25

i

C

TPCA-1

20,000 IL-1β (pg ml–1)

37

BMS-345541

DMSO

TPCA-1

DMSO

DMSO BMS-345541 TPCA-1 *** ***

LPS + ATP LPS + Nig

LDH release (%)

h Control

on

20,000

25

ASC

C 40,000

tro l M oc ds k DN A

ds k DN A

M

***

50 0

tro

l tro on

WT A20myel-KO ***

60,000

0

on C

Casp1

2,000

500

50

S. Typhimurium MOI 10

100 37 25 50 37

proIL-1β

f

g

5,000

Nlrp3

10,000

0

25 20

WT A20myel-KO – + – +

LPS

e

100

WT A20myel-KO ***

10,000

C 40

n IL-1β (pg ml–1)

WT A20myel-KO

tro l

dsDNA

Mock

Control

dsDNA

60

LPS + ATP

c

WT A20myel-KO ***

Control

50 37

DMSO

15,000

oc

25

l LDH release (%)

37

Mock

Control 50

Casp1

A20myel-KO

0 10 20 30 60 Time after nigericin (min)

18,000

LPS

BMS-345541

*

0

0 5 10 S. Typhimurium (MOI) WT

0

WT

50 0

k

***

+

Casp1

m

***

50

0 10 20 30 60 Time after ATP (min)

WT A20myel-KO

WT A20myel-KO

100

0

100

25

si lic a

N ig +

*** 0

150

LDH release (%)

10

j

LP S

LP S

50

0 10 20 30 60 Time after nigericin (min)

5

0

10

5

0

100

0

A20myel-KO

WT

***

150

β-actin IL-18 Nlrp3

2,000

0 10 20 30 60 Time after ATP (min)

i

***

WT A20myel-KO

Casp1

IL-1β (pg

2,000

4,000

150

LDH release (%)

ml–1)

***

6,000

WT A20myel-KO *** ***

20

25 20

h LDH release (%)

g 8,000

6,000

25

IL-1β (pg ml–1)

f

WT A20myel-KO *** *** ***

10,000

IL-1β (pg ml–1)

AT P

l 25

+

on tro

Casp1

Casp1

37

e

LP S

C LPS + Nig (min) 50 37

40

0 10 20 30 60 0 10 20 30 60

0 10 20 30 60 0 10 20 30 60

LPS + ATP (min) 50

60

LPS Control + Nig

d

b

WT A20myel-KO ***

Control

A20myel-KO

WT

80

WT

5,000

d

A20myel-KO

WT

Casp1

***

0

c

MOI 50 37

***

WT

10,000

25

0

a

WT A20myel-KO ***

TPCA-1

IL-1β (pg ml–1)

50 37

Casp1

15,000

BMS-345541

b

LPS + silica

LPS + Nig

Control

LPS + silica

LPS + Nig

LPS + ATP

Control

LPS + ATP

A20myel-KO

WT

a

Relative Nlrp3 expression

LETTER RESEARCH

Figure 3 | A20 inhibits Nlrp3 inflammasome priming. a, b, Nlrp3 (a) and proIL-1b (b) mRNA levels of LPS-treated BMDMs. c, Expression of the indicated proteins in BMDMs 6 h after LPS treatment. d, e, Rapid Nlrp3 inflammasome activation as described in Methods. Expression of the indicated proteins (d) and secreted cytokines (e) were determined. f–i, A20myel-KO BMDMs were treated as indicated. Nlrp3 mRNA levels (f), caspase-1 expression (g), secreted IL-1b (h) and LDH activity (i) were determined. Black arrow, procaspase-1; white arrow, p20. Data represent mean 6 s.d. of one out of three biological replicates, with three technical replicates each (*P , 0.05; ***P , 0.001; Student’s t-test).

and their hind paws had a normal appearance and lacked clinical signs of pathology of rheumatoid arthritis (Fig. 4b–d). Histological analysis of the ankle joints of these mice showed significantly reduced synovial and periarticular inflammation, and substantially less infiltrated mononuclear cells compared with tissue sections of A20myel-KONlrp31/1 mice of comparable age (Fig. 4e, f and Extended Data Table 2). In agreement, three-dimensional micro-computed tomography imaging showed that the extent of bone erosion in hind paws of representative A20myel-KO Nlrp31/1 and A20myel-KONlrp32/2 mice was markedly different. Unlike in Nlrp3-deficient A20myel-KO mice, hind paws of Nlrp3-sufficient mice exhibited severe loss of bone density in the metatarsal region (Fig. 4g), demonstrating a key role for Nlrp3 in the pathology of rheumatoid arthritis. We also analysed the impact of caspase-1/11 deficiency on IL-1b secretion and pathology of rheumatoid arthritis in A20myel-KO mice. IL1b levels in circulation were significantly reduced in A20myel-KOCasp1/ 112/2 mice compared with A20myel-KOCasp1/111/1 mice (Fig. 4h). As in A20myel-KONlrp32/2 mice, serum levels of IL-6 were markedly reduced in A20myel-KOCasp1/112/2 mice, whereas TNF production was not significantly different compared with A20myel-KOCasp1/111/1 mice (Extended Data Fig. 4). Moreover, hind paws of all analysed A20myel-KO Casp1/111/1 mice were clearly inflamed and swollen (Fig. 4i–k). In contrast, 50% of A20myel-KOCasp1/112/2 mice were devoid of clinical signs of arthritis, and disease symptoms in the remaining A20myel-KO Casp1/112/2 mice were very mild (Fig. 4i–k). In agreement, analysis of 0 0 M O N T H 2 0 1 4 | VO L 0 0 0 | N AT U R E | 3

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER b P = 0.025 P = 0.035

100

A20myel-KONlrp3–/–

0 0 A2

+

+/

3

rp

m

0

A2

lye

+

+/

3

lrp

KO N

0

–/–

11–20

21–30 31–40 Age (weeks)

0

A2

d

A20myel-KONlrp3+/+

e

A20myel-KONlrp3–/–

200 μm

A20myel-KONlrp3+/+

f

200 μm

&'

A20myel-KONlrp3–/–

3

A20myel-KONlrp3+/+ A20myel-KONlrp3–/–

50

3

lrp

KO N

l-

e my

150

100

50

l fl/fl N

Clinical score

c

A20myel-KONlrp3+/+

P < 0.0001 P < 0.0001 2 5 μm

5 μm

1

P = 0.042

6 Histological score

IL-1β (pg ml–1)

150

Clinical incidence (%)

a

4

METHODS SUMMARY

2

0 11–20

21–30 31–40 Age (weeks)

my

0

h IL-1β (pg ml–1)

A20myel-KONlrp3+/+ A20myel-KONlrp3–/–

150

P = 0.0002 P = 0.0002

100 50 0

fl

0 A2

j

k Clinical score

Clinical incidence (%)

150

100

50

0

e

my

0

A2

+/+

e my

0 A2

e

my

0

A2

K

elmy

l

A20myel-KOCasp1/11+/– A20myel-KOCasp1/11–/–

1.0 0.5 ,' +/+

+/–

–/–

1 1 1 1/1 1/1 1/1 sp asp asp OC O Ca KO C l-K ll-K

e

my

0

A2

m

0

A2

1.5

0.0

–/–

+/–

1 1 1 1/1 1/1 1/1 sp asp asp OC O Ca KO C l-K ll-K

A2

–/–

3

lrp

KO N

lye

0

A2

P = 0.002

2.0

1 1/1

sp O Ca

P < 0.0001

2.5

–/–

+/+

1

1/1

sp O Ca

K elmy

0

A2

+/+

1

1/1

asp /fl C

i A20myel-KOCasp1/11+/+

g

+

+/

3

lrp

ON

K el-

A20myel-KOCasp1/11–/–

0

e my

0 A2

e my

0

A2

be caused by different combinations of genetic and environmental triggers. As such, available mouse models of rheumatoid arthritis may each be relevant to distinct subsets of patients with rheumatoid arthritis, or suitable for studying certain aspects of the pathogenesis of rheumatoid arthritis. Pathology in the widely used collagen- and antigen-induced mouse arthritis models occurs independently of inflammasome signalling13,14. In contrast, we demonstrated here that arthritis pathology in A20myel-KO mice critically relies on the Nlrp3 inflammasome/IL-1 signalling axis. Because both A20/Tnfaip315–18 and Nlrp319–21 are rheumatoid arthritis susceptibility genes, this suggests that A20myel-KO mice might be a suitable pre-clinical model for validating the effectiveness of experimental therapies for rheumatoid arthritis targeting inflammasomes and/or IL-1 signalling.

Figure 4 | Nlrp3 and caspase-1 deletion rescues arthritis in A20myel-KO mice. a, IL-1b serum levels of A20fl/flNlrp31/1 (n 5 10), A20myel-KONlrp31/1 (n 5 10) and A20myel-KONlrp32/2 (n 5 10) mice aged 20–35 weeks. b, Hind paws of 30 week old mice. c, d, A20myel-KONlrp31/1 (n 5 20) and A20myel-KONlrp32/2 (n 5 17) mice were scored for arthritis incidence (d) and severity (e). e, Representative ankle joint sections; magnification 340 (top) and 3100 (bottom). f, Histological scores of ankle sections of A20myel-KONlrp31/1 (n 5 5) and A20myel-KONlrp32/2 (n 5 5) mice. g, microcomputed tomography images of hind paws of A20myel-KONlrp31/1 and A20myel-KONlrp32/2 mice. Solid arrow, bone erosion; empty arrow, intact cartilage and bone. h, IL-1b serum levels of A20fl/flCasp1/111/1 (n 5 11), A20myel-KOCasp1/111/1 (n 5 26) and A20myel-KOCasp1/112/2 (n 5 16) mice aged 20-35 weeks. i, Hind paws of 25 weeks old mice. j, k, 15–30 weeks old A20myel-KOCasp1/111/1 (n 5 12), A20myel-KOCasp1/111/2 (n 5 15) and A20myel-KOCasp1/112/2 (n 5 24) mice were clinically scored for arthritis incidence (j) and severity (k). l, Representative ankle joint sections; magnification 310 (top) and 320 (bottom). P values were determined by Student’s t-test (a, d, f, k) and Mann–Whitney U test (h).

Mice. Nlrp32/2 (ref. 22), Casp1/112/2 (ref. 23), Il1r12/2 (ref. 24) and A20myel-KO mice2 with a lysozyme M-Cre-targeted deletion of A20 in myeloid cells were as described. BMDM studies. BMDMs were isolated and the Nlrp3 inflammasome was activated by LPS in combination with ATP (Roche), nigericin (Sigma-Aldrich) or Silica (US Silica). The AIM2 or Nlrc4 inflammasomes were activated by infection with Salmonella enterica serovar Typhimurium or transfection with plasmid DNA, respectively. Antibodies. The following antibodies were used: A20 (Santa Cruz Biotechnology), caspase-1, Nlrp3, ASC (Adipogen), IL-1b (Genetex), IL-18 (Biovision), IkBa, Phospho-IkBa (Ser32) (Cell Signaling) and b-actin (Novus Biologicals). Cytokine analysis and lactate dehydrogenase measurement. Cytokine levels in culture medium of BMDMs and in serum were determined by Luminex assays (Bio-Rad) and IL-1b enzyme-linked immunosorbent assay (ELISA) (R&D Systems). Lactate dehydrogenase (LDH) activity (Promega) was used to quantify pyroptosis. Reverse transcription PCR. RNA was isolated using RNeasy kit (Qiagen) and mRNA levels were determined by quantitative reverse transcription PCR (qRT– PCR). Clinical scoring. Mice were randomly scored in a blinded fashion for development of peripheral arthritis. The severity of arthritis was assessed using a visual scoring system. Histology and histological scoring. Paraffin sections of murine paws were stained with haematoxylin and eosin for evaluation of inflammation and bone erosion. Histological scores were based on evaluation of two parameters, calcaneal erosion and inflammation at the synovio-entheseal complex, each ranging from 0 (normal) to 3. Online Content Methods, along with any additional Extended Data display items and Source Data, are available in the online version of the paper; references unique to these sections appear only in the online paper. Received 20 October 2013; accepted 10 April 2014. Published online 29 June 2014. 1. 2. 3. 4. 5. 6. 7.

joint sections of arthritic A20myel-KOCasp1/111/2 mice stained with haematoxylin and eosin showed massive mononuclear cell infiltration associated with marked articular, periarticular and synovial inflammation and reactive fibrosis. In contrast, joints of A20myel-KOCasp1/112/2 mice were markedly protected from the histopathology of rheumatoid arthritis (Fig. 4l). Collectively, these results demonstrate that A20 puts a brake on Nlrp3 inflammasome activation by reducing basal and LPS-induced Nlrp3 expression levels (Extended Data Fig. 5). Moreover, we showed that excessive Nlrp3 inflammasome activation drives arthritis pathogenesis in A20myel-KO mice. Arthritis in humans is a complex disease that may

8. 9. 10. 11. 12. 13.

Burger, D., Dayer, J. M., Palmer, G. & Gabay, C. Is IL-1 a good therapeutic target in the treatment of arthritis? Best Pract. Res. Clin. Rheumatol. 20, 879–896 (2006). Matmati, M. et al. A20 (TNFAIP3) deficiency in myeloid cells triggers erosive polyarthritis resembling rheumatoid arthritis. Nature Genet. 43, 908–912 (2011). Lamkanfi, M. & Dixit, V. M. Inflammasomes and their roles in health and disease. Annu. Rev. Cell Dev. Biol. 28, 137–161 (2012). Bauernfeind, F. G. et al. Cutting edge: NF-kB activating pattern recognition and cytokine receptors license NLRP3 inflammasome activation by regulating NLRP3 expression. J. Immunol. 183, 787–791 (2009). Wertz, I. E. et al. De-ubiquitination and ubiquitin ligase domains of A20 downregulate NF-kB signalling. Nature 430, 694–699 (2004). Vereecke, L., Beyaert, R. & van Loo, G. The ubiquitin-editing enzyme A20 (TNFAIP3) is a central regulator of immunopathology. Trends Immunol. 30, 383–391 (2009). Lee, E. G. et al. Failure to regulate TNF-induced NF-kB and cell death responses in A20-deficient mice. Science 289, 2350–2354 (2000). Boone, D. L. et al. The ubiquitin-modifying enzyme A20 is required for termination of Toll-like receptor responses. Nature Immunol. 5, 1052–1060 (2004). Fernandes-Alnemri, T. et al. Cutting edge: TLR signaling licenses IRAK1 for rapid activation of the NLRP3 inflammasome. J. Immunol. 191, 3995–3999 (2013). Juliana, C. et al. Non-transcriptional priming and deubiquitination regulate NLRP3 inflammasome activation. J. Biol. Chem. 287, 36617–36622 (2012). Lin, K. M. et al. IRAK-1 bypasses priming and directly links TLRs to rapid NLRP3 inflammasome activation. Proc. Natl Acad. Sci. USA 111, 775–780 (2014). Schroder, K. et al. Acute lipopolysaccharide priming boosts inflammasome activation independently of inflammasome sensor induction. Immunobiology 217, 1325–1329 (2012). Ippagunta, S. K. et al. Inflammasome-independent role of apoptosis-associated speck-like protein containing a CARD (ASC) in T cell priming is critical for collageninduced arthritis. J. Biol. Chem. 285, 12454–12462 (2010).

4 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 4

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH 14. Kolly, L. et al. Inflammatory role of ASC in antigen-induced arthritis is independent of caspase-1, NALP-3, and IPAF. J. Immunol. 183, 4003–4012 (2009). 15. Consortium, W. T. C. C. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 447, 661–678 (2007). 16. Dieguez-Gonzalez, R. et al. Analysis of TNFAIP3, a feedback inhibitor of nuclear factor-kB and the neighbor intergenic 6q23 region in rheumatoid arthritis susceptibility. Arthritis Res. Ther. 11, R42 (2009). 17. Plenge, R. M. et al. Two independent alleles at 6q23 associated with risk of rheumatoid arthritis. Nature Genet. 39, 1477–1482 (2007). 18. Thomson, W. et al. Rheumatoid arthritis association at 6q23. Nature Genet. 39, 1431–1433 (2007). 19. Ben Hamad, M. et al. Association study of CARD8 (p.C10X) and NLRP3 (p.Q705K) variants with rheumatoid arthritis in French and Tunisian populations. Int. J. Immunogenet. 39, 131–136 (2012). 20. Kastbom, A. et al. Genetic variation in proteins of the cryopyrin inflammasome influences susceptibility and severity of rheumatoid arthritis (the Swedish TIRA project). Rheumatology (Oxford) 47, 415–417 (2008). 21. Mathews, R. J. et al. Evidence of NLRP3-inflammasome activation in rheumatoid arthritis (RA); genetic variants within the NLRP3-inflammasome complex in relation to susceptibility to RA and response to anti-TNF treatment. Ann. Rheum. Dis. 73, 1202–1210 (2014). 22. Mariathasan, S. et al. Cryopyrin activates the inflammasome in response to toxins and ATP. Nature 440, 228–232 (2006).

23. Schott, W. H. et al. Caspase-1 is not required for type 1 diabetes in the NOD mouse. Diabetes 53, 99–104 (2004). 24. Glaccum, M. B. et al. Phenotypic and functional characterization of mice that lack the type I receptor for IL-1. J. Immunol. 159, 3364–3371 (1997). Acknowledgements We thank R. Flavell and V. Dixit for supplying mutant mice. L.V.W. is a postdoctoral fellow with the Fund for Scientific Research-Flanders (FWO). This work was supported by the Ghent University Concerted Research Actions (grant BOF14/ GOA/013) and grants from the European Research Council (Grant 281600) and the FWO (grants G030212N, 1.2.201.10.N.00 and 1.5.122.11.N.00) to M.L., and by FWO research grants (Odysseus-G091908, G061910N and G016812N) and a Queen Elisabeth Medical Foundation grant to G.V.L., and by the Ghent University Group-ID MRP to G.V.L., R.B. and D.E. T.-D.K. is supported by grants from the National Institutes of Health (AR056296, CA163507 and AI101935) and the American Lebanese Syrian Associated Charities. Author Contributions L.V.W., G.V.L. and M.L. designed the study; L.V.W., N.V., A.F., P.J., E.V. and P.V. performed experiments; L.V.W., N.V., P.J., P.V., G.V.L., R.B., D.E., T.-D.K. and M.L. analysed data and wrote the manuscript; T.-D.K. provided essential reagents and scientific insight; M.L. oversaw the project. Author Information Reprints and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests. Readers are welcome to comment on the online version of the paper. Correspondence and requests for materials should be addressed to M.L. ([email protected]).

0 0 M O N T H 2 0 1 4 | VO L 0 0 0 | N AT U R E | 5

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER METHODS Mice. Nlrp32/2 (ref. 22) and Casp1/112/2 (ref. 23) mice have been described, and were provided by V. Dixit and R. Flavell. Il1r12/2 (ref. 24) mice were purchased from Jackson Laboratories and bred in house. A20myel-KO mice with a lysozyme M-Cretargeted deletion of A20 in myeloid cells were as described2. In vivo experiments were controlled with age- and sex-matched littermates. The sample size was chosen to validate statistical analyses. Mice were housed in individually ventilated cages and kept under pathogen-free conditions at the animal facilities of Ghent University. All animal experiments were conducted with permission of the ethical committees on laboratory animal welfare of Ghent University. Macrophage differentiation and stimulation. BMDMs were generated by culturing bone marrow cells in IMDM medium containing 10% heat-inactivated FBS, 30% L cell-conditioned medium, 100 U ml–1 penicillin, and 100 mg ml–1 streptomycin at 37 uC in a humidified atmosphere containing 5% CO2. Six days later, cells were collected and seeded in 12-well plates in IMDM containing 10% heat-inactivated FBS and 1% non-essential amino acids in the presence of antibiotics. The next day BMDMs were either left untreated or treated with 5 mg ml–1 ultrapure LPS from Salmonella minnesota (InvivoGen) for 3 h followed by 5 mM ATP (Roche) or 20 mM nigericin (Sigma-Aldrich) for 30 min or 0.5 mg ml–1 silica (Min-U-Sil 5, US Silica) for 6 h. BMDMs were also separately infected with Salmonella enterica serovar Typhimurium at the indicated multiplicity of infection for 3 h, the last hour of which was in the presence of gentamycin (50 mg ml–1). To activate the AIM2 inflammasome, BMDMs were left untreated or treated with 5 mg ml–1 LPS for 3 h. After removal of LPS, BMDMs were exposed to Lipofectamine 2000 (mock, Life Technologies) or Lipofectamine with 1.25 mg ml–1 pCDNA3 plasmid DNA for 19 h. To induce rapid Nlrp3 inflammasome activation, BMDMs were stimulated with LPS for 10 min, and then treated with ATP or nigericin for 45 min. For inhibition of NF-kB, A20myel-KO BMDMs were pre-incubated for 30 min with 1.25 or 2.5 mM of the selective IKK inhibitors BMS-345541 (ref. 25) or TPCA-1 (ref. 26) (SigmaAldrich) before stimulation with LPS. Western blotting. Cell lysates and culture supernatants were incubated with cell lysis buffer (20 mM Tris HCl pH 7.4, 200 mM NaCl, 1% NP-40) and denatured in Laemmli buffer. Subsequently the protein samples were boiled at 95 uC for 10 min and separated by SDS–polyacrylamide gel electrophoresis. Separated proteins were transferred to PVDF membranes. Blocking, incubation with antibody and washing of the membrane were done in PBS supplemented with 0.05% Tween-20 (v/v) and 3% (w/v) non-fat dry milk. Immunoblots were incubated overnight with primary antibodies against caspase-1 (AG-20B-0042-C100, Adipogen), Nlrp3 (AG-20B-0014C100, Adipogen), ASC (AG-25B-0006, Adipogen), IL-1b (GTX74034, Genetex), IL-18 (5180R-100, Biovision), IkBa (9242S, Cell Signaling), Phospho-IkBa (2859S, Ser32) (Cell Signaling), b-actin (NB600-501H, Novus Biologicals) and A20 (sc166692, Santa Cruz Biotechnology). Horseradish-peroxidase-conjugated goat antimouse (115-035-146, Jackson Immunoresearch Laboratories) or anti-rabbit secondary antibody (111-035-144, Jackson Immunoresearch Laboratories) was used to detect proteins by enhanced chemiluminescence (Thermo Scientific). Cytokine analysis and LDH measurement. Cytokine levels in culture medium and serum were determined by magnetic bead-based multiplex assay using Luminex technology (Bio-Rad) and IL-1b ELISA (R&D Systems), according to the manufacturers’ instructions. Cell death was determined by measuring LDH activity in culture medium according to the manufacturer’s instructions (Promega). Reverse transcription PCR. Total RNA was isolated from BMDMs using RNeasy kit (Qiagen). Complementary DNA was synthesized using the QuantiTect Reverse

Transcription Kit (Qiagen). A20 mRNA expression in wild-type and A20myel-KO BMDMs, treated with 5 mg ml–1 LPS for 6 h, was determined by RT–PCR using the following primers: 59-CTCGGAACTTTAAATTCCGC-39 and 59-GGGTAAGTT AGCTTCATCC-39. To analyse mRNA levels of Nlrp3, proIL-1b, ASC and caspase-1, wild-type and A20myel-KO BMDMs were left untreated or treated with 5 mg ml–1 LPS for 3 h and qRT–PCR was performed using a LightCycler 480 SYBR Green I Master Mix Kit (Roche Applied Science) in a LightCycler 480 real-time PCR machine (Roche Applied Science). The cycling conditions were 95 uC for 5 min followed by 45 cycles of 95 uC for 10 s, 60 uC for 15 s and 72 uC for 20 s. Gene expression levels were normalized to GAPDH. qRT–PCR sense/antisense primers used were as follows: mouse GAPDH, 59-GGTGAAGGTCGGTGTGAACG-39 and 59-CTCGCTCCT GGAAGATGGTG-39; mouse Nlrp3, 59-ATTACCCGCCCGAGAAAGG-39 and 59-TCGCAGCAAAGATCCACACAG-39; mouse proIL-1b, 59-TGGGCCTCAA AGGAAAGA-39 and 59-GGTGCTGATGTACCAGTT-39; mouse ASC, 59-CTT GTCAGGGGATGAACTCAAAA-39 and 59-GCCATACGACTCCAGATAGTA GC-39; mouse Casp1, 59-ACAAGGCACGGGACCTATG-39 and 59-TCCCAGTC AGTCCTGGAAATG-39. Quantitative analysis of mouse proIL-18 mRNA was performed by Taqman gene expression assay (Mm00434225_m1, Applied Biosystems). Clinical scoring. Mice were randomly scored in a blinded fashion for development of peripheral arthritis once a week and scores were subsequently linked to genotypes. A score ranging from 0 to 3 was assigned to each paw, with 0 being normal, 0.5 being swelling of one or more toes, 1 being mild swelling of the wrist and/or ankle or carpus and/or tarsus, 2 being moderate swelling of the wrist and/or ankle or carpus and/or tarsus or mild swelling of both, and 3 being severe swelling of the entire paw. Micro-computed tomography imaging. Micro-computed tomography micrographs of paws fixed in formalin were made using an ex vivo micro-computed tomography scanner (LocusSP Specimen CT, GE Healthcare) at 28-mm isotropic voxel size, with 720 projections, an integration time of 1,700 ms, photon energy of 80 keV and a current of 70 mA. Histology and histological scoring. Murine paws were dissected, fixed in 4% formaldehyde and de-calcified in 5% formic acid until bones were pliable. Paraffin sections were stained with haematoxylin and eosin for evaluation of inflammation and bone erosions. Histological sections were evaluated by two blinded assessors, and scores were determined by combining assessment of two parameters, calcaneal erosion and inflammation, at the synovio-entheseal complex, each ranging from 0 (normal) to 3 (erosion: 0, none; 1, minimal; 2, intermediate; 3, into bone marrow; inflammation: 0, none; 1, minimal one or two cell layers; 2, two to five cell layers; 3, more than five cell layers). Statistics. GraphPad Prism 5.0 software was used for data analysis. Data are shown as mean 6 s.d. Data were compared by an unpaired two-tailed Student’s t-test when values followed a Gaussian distribution, or with the non-parametric Mann–Whitney U test. P , 0.05 was considered to indicate statistical significance. 25.

26.

Burke, J. R. et al. BMS-345541 is a highly selective inhibitor of I kappa B kinase that binds at an allosteric site of the enzyme and blocks NF-kappa B-dependent transcription in mice. J. Biol. Chem. 278, 1450–1456 (2003). Podolin, P. L. et al. Attenuation of murine collagen-induced arthritis by a novel, potent, selective small molecule inhibitor of IkB kinase 2, TPCA-1 (2[(aminocarbonyl)amino]-5-(4-fluorophenyl)-3-thiophenecarboxamide), occurs via reduction of proinflammatory cytokines and antigen-induced T cell proliferation. J. Pharmacol. Exp. Ther. 312, 373–381 (2005).

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 1 | A20 deficiency does not cause spontaneous Nlrp3 inflammasome activation. a–c, Wild-type and A20myel-KO BMDMs were stimulated with 5 mg ml–1 LPS for 3 h, treated with 5 mM ATP or 20 mM nigericin for 60 min, or stimulated with 5 mg ml–1 LPS for 3 h and then treated with 5 mM ATP or 20 mM nigericin for 60 min. Cell extracts were immunoblotted for caspase-1 (a) and culture supernatants were analysed for secretion of IL-1b (b) and IL-18 (c). Black arrows on western blots denote procaspase-1 (p45); white arrows denote the processed p20 subunit (p20). ELISA data are shown as mean 6 s.d. of one out of three biological replicates, with three technical replicates each (***P , 0.001; Student’s t-test).

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

Extended Data Figure 2 | A20 negatively regulates NF-kB activation. a, Wild-type and A20myel-KO BMDMs were incubated with 5 mg ml–1 LPS for the indicated durations before cell extracts were prepared and immunoblotted with the indicated antibodies. b, c, Wild-type and A20myel-KO BMDMs were stimulated with 5 mg ml–1 LPS for 3 h, treated with 5 mM ATP or 20 mM nigericin for 60 min, or stimulated with 5 mg ml–1 LPS for 3 h and then treated with 5 mM ATP or 20 mM nigericin for 60 min. Culture supernatants were analysed for IL-6 secretion (b) and TNF secretion (c). ELISA data are shown as mean 6 s.d. of one out of three biological replicates, with three technical replicates each (***P , 0.001; Student’s t-test).

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 3 | Quantification of ASC, caspase-1 and proIL-18 mRNA expression in wild-type and A20 deficient macrophages. a–c, Wildtype and A20myel-KO BMDMs were stimulated with 5 mg ml–1 LPS for 3 h before mRNA levels of caspase-1 (a), ASC (b) and proIL-18 (c) were analysed by qRT–PCR. Data are shown as mean 6 s.d. of one out of three biological replicates, with three technical replicates each (*P , 0.05; ***P , 0.001; Student’s t-test).

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

Extended Data Figure 4 | Comparison of serum titres of inflammatory cytokines between A20myel-KO mice and A20myel-KONlrp32/2 or A20myel-KOCasp1/112/2 mice. a–c, Levels of IL-1a (a), IL-6 (b) and TNF (c) in serum of A20fl/fl (n 5 10), A20myel-KO (n 5 10), A20myel-KOCasp1/112/2 (n 5 12) and A20myel-KONlrp32/2 (n 5 9) mice between 20 and 35 weeks of age. P values were determined by Mann–Whitney U test (b) and Student’s t-test (c).

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 5 | The ubiquitin-editing enzyme A20 negatively regulates Nlrp3 inflammasome activation. The NLR member Nlrp3, the inflammatory cytokine proIL-1b and the ubiquitin-editing enzyme A20 are expressed at low levels in resting macrophages. Binding of the TLR4 ligand LPS to its receptor triggers phosphorylation and rapid degradation of IkBa, allowing translocation of NF-kB to the nucleus, and NF-kB-mediated

upregulation of proIL-1b, proIL-18, Nlrp3 and A20. A20 prevents excessive Nlrp3 inflammasome activation by dampening basal and LPS-induced NF-kBmediated upregulation of Nlrp3. As such, A20 reduces the pool of Nlrp3 that is available for inflammasome assembly. In addition, it limits the levels of the inflammasome substrates proIL-1b and proIL-18.

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER Extended Data Table 1 | Arthritic histopathology of A20myel-KOIl1r11/2 and A20myel-KOIl1r12/2 mice

Raw data of histological scores of ankle sections of A20myel-KOIl1r11/2 (n 5 10) and A20myel-KO Il1r12/2 mice (n 5 8). Arthritic histopathology of ankles was scored on two parameters, calcaneal erosion and inflammation at the synovio-entheseal complex, as described in Methods.

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH Extended Data Table 2 | Arthritic histopathology of A20myel-KO Nlrp31/1 and A20myel-KONlrp32/2 mice

Raw data of histological scores of ankle sections of A20myel-KONlrp31/1 (n 5 5) and A20myel-KONlrp32/2 mice (n 5 5). Arthritic histopathology of ankles was scored on two parameters, calcaneal erosion and inflammation at the synovio-entheseal complex, as described in Methods.

©2014 Macmillan Publishers Limited. All rights reserved