Articles

The transcription factor IRF1 and guanylate-binding proteins target activation of the AIM2 inflammasome by Francisella infection © 2015 Nature America, Inc. All rights reserved.

Si Ming Man1,2, Rajendra Karki1, R K Subbarao Malireddi1, Geoffrey Neale3, Peter Vogel4, Masahiro Yamamoto5, Mohamed Lamkanfi6,7 & Thirumala-Devi Kanneganti1 Inflammasomes are critical for mounting host defense against pathogens. The molecular mechanisms that control activation of the AIM2 inflammasome in response to different cytosolic pathogens remain unclear. Here we found that the transcription factor IRF1 was required for activation of the AIM2 inflammasome during infection with the Francisella tularensis subspecies novicida (F. novicida), whereas engagement of the AIM2 inflammasome by mouse cytomegalovirus (MCMV) or transfected double-stranded DNA did not require IRF1. Infection of F. novicida detected by the DNA sensor cGAS and its adaptor STING induced type I interferon–dependent expression of IRF1, which drove the expression of guanylate-binding proteins (GBPs); this led to intracellular killing of bacteria and DNA release. Our results reveal a specific requirement for IRF1 and GBPs in the liberation of DNA for sensing by AIM2 depending on the pathogen encountered by the cell. Inflammasomes are cytosolic multimeric protein complexes that provide host defense against microbial pathogens and have a role in the development of autoinflammatory and metabolic diseases1. Formation of the inflammasome is initiated by sensors of the innate immune system, including nucleotide-binding oligomerization domain (Nod)–like receptors (NLRs) or AIM2-like receptors (ALRs), following detection of microbe-associated or damage-associated molecular patterns in the cytoplasm of a cell2. In most cases, NLRs and ALRs engage the adaptor ASC, whose caspase-recruitment domain is critical for subsequent interaction with the caspase-recruitment domain of procaspase-1 (ref. 3). Activation of caspase-1 induces proteolytic cleavage and release of mature interleukin 1β (IL-1β) and IL-18, as well as the induction of a type of pro-inflammatory cell death called ‘pyroptosis’2. Therefore, inflammasomes are able to mount innate immunological effector functions in response to a range of stimuli encountered by the cell. AIM2 is a member of the ALR family that contributes to the host defense against bacterial pathogens and viral pathogens, such as Francisella tularensis and cytomegalovirus, respectively4–9. AIM2 contains a HIN-200 domain that directly binds double-stranded DNA (dsDNA) released into the cytosol of a cell during a microbial infection4–7. Similarly, self DNA that is released into and accumulates in keratinocytes also activates the AIM2 inflammasome to drive the release of IL-1β in lesions of patients with psoriasis10, which indicates that AIM2 also responds to endogenous dsDNA released ­

during ­ cellular damage. Structural analysis has demonstrated that the positively charged HIN-200 domain of AIM2 embraces the dsDNA, whereas the pyrin domain of AIM2 facilitates the recruitment of ASC11. The molecular mechanisms that govern assembly of the AIM2 inflammasome are less clear than those for the NLRP3 inflammasome. NLRP3 is activated by a plethora of microbe-associated and damage-associated molecular patterns. Activation of the NLRP3 inflammasome requires two signals3. The first signal, or ‘priming’, is provided by the engagement of Toll-like receptors (TLRs), the receptor Nod2 or the receptor for tumor-necrosis factor to activate the transcription factor NF-κB and induce NLRP3 expression. The second signal is provided by an NLRP3 activator; these include ATP, uric acid crystals, silica, asbestos, bacterial messenger RNA, bacterial DNA-RNA hybrids and muramyldipeptide3. In addition, signaling via type I interferons is critical for activation of the so-called ‘noncanonical’ NLRP3 inflammasome12,13. In this context, extracellular lipopolysaccharide (LPS) from Gram-negative bacteria activates TLR4 and its adaptor TRIF to induce the expression of type I interferons and pro-caspase-11 (refs. 12–15). LPS released into the cytosol by means of vacuolar lysis binds caspase-11, which leads to caspase-11dependent cell death and activation of the non-canonical NLRP3 inflammasome16–20. In contrast, AIM2 is constitutively expressed and does not require NF-κB-mediated priming for its activation6. Instead, signaling via type I interferons contributes to activation of

1Department

of Immunology, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA. 2School of Biotechnology and Biomolecular Sciences, The University of New South Wales, Sydney, Australia. 3Hartwell Center for Bioinformatics & Biotechnology, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA. 4Animal Resources Center and the Veterinary Pathology Core, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA. 5Department of Microbiology and Immunology, Osaka University, Osaka, Japan. 6Department of Medical Protein Research, Vlaams Instituut voor Biotechnologie, Ghent, Belgium. 7Department of Biochemistry, Ghent University, Ghent, Belgium. Correspondence should be addressed to T.-D.K. ([email protected]). Received 11 November 2014; accepted 5 February 2015; published online 16 March 2015; doi:10.1038/ni.3118

nature immunology  aDVANCE ONLINE PUBLICATION



Articles by the TLR adaptor TRIF is required for robust activation of the caspase-11–NLRP3 inflammasome in macrophages infected with Citrobacter rodentium, Escherichia coli or Vibrio cholerae12,13,15; however, we found that TRIF and caspase-11 were dispensable for activation of the AIM2 inflammasome by F. novicida (Fig. 1b). Indeed, F. novicida synthesizes a tetra-acylated lipid A that differs to the hexa-acylated species of many enteric bacteria that normally activates caspase-11, which explains the ability of F. novicida to evade caspase-11-mediated detection18. These findings suggested that an as-yet-undefined type I interferon signature independent of TRIF and caspase-11 was needed to engage the AIM2 inflammasome by infection with F. novicida but not in response to transfected dsDNA. We performed microarray analysis to identify genes regulated differentially in wild-type and IFNAR1-deficient (Ifnar1−/−) BMDMs that had been infected with F. novicida. The expression of many members of the IRF family of interferon-regulatory transcription factors, including IRF1, IRF7 and IRF9, was significantly reduced in the absence of IFNAR1 (Fig. 1c). IRF9, a subunit of the ISGF3 complex, contributed to activation of AIM2 inflammasome by infection with F. novicida (Fig. 1a and Supplementary Fig. 1a), and we excluded the possibility of a role for IRF7 in activation of the AIM2 inflammasome upon infection with F. novicida (Supplementary Fig. 1b). Confirmation of IRF1 protein expression indicated that it was robustly upregulated by infection with F. novicida through a mechanism that required IFNAR1, IFNAR2, STAT1 and IRF9 (Fig. 1d). Consistent with that, BMDMs stimulated with recombinant mouse interferon-β (IFN-β) also robustly upregulated IRF1 expression, which was critically dependent on type I interferon receptors (Supplementary Fig. 1c). To investigate whether IRF1 is an upstream molecule that leads to activation of the AIM2 inflammasome, we infected unprimed wildtype and Irf1−/− BMDMs with F. novicida and analyzed inflammasome responses 20 h later. Notably, unprimed Irf1−/− BMDMs failed to respond to F. novicida–induced, AIM2-dependent activation of caspase-1, release of IL-1β and IL-18 and cell death, whereas their

0 60

* ** ** ** *

40 20

** ***

*** *** *** /–

W Ai T m Ifn 2 –/– a Ifn r1 –/– ar St 2 –/– at 1 –/ Irf – 9–

0

0.8

40

2

8 16 24

40

Irf9

dT

)

a

0

2

8 16 24 0

A: ly (d

ov WT 0

Irf1 Irf4

2

8 16 24

Irf6 –/–

WT F. novicida (h)

0

Irf8 Irf5

GAPDH

20

Irf2

Ifnar2–/–

IRF1

**

c Irf7

20

F. n

Po

Ifnar1–/– 2

60

NS

ic

:d dA

ic ov 8 16 24 0

NS

NS

id

T)

a id

T)

IL-18 (ng/ml)

0

:d ly ( 0

NS

NS

1

WT F. novicida (h)

0.4 0 60

Po

d

2

dA

ci vi no

****

NS

NS

Po

1.2

0

2

8 16 24 0

Stat1 2

8 16 24

–/–

WT 0

2

8 16 24 0

Irf3

Irf9 2

8 16 24

Expression (log2)

IRF1

Ai WT m Ifn 2 –/– a Ifn r1 –/– ar St 2 –/– at 1 –/ Irf – 9–

0.4

IL-18 (ng/ml)

0.8

Cell death (% LDH release)

IL-18 (ng/ml)

1.2

Poly(dA:dT)

F.

Med 1.6

–2

/–

F. novicida

4 3 2 1 0

da

Casp-1 p20

NS WT Trif –/– Casp11–/–

F. n

Pro-Casp-1

Med

NS NS

ly (

b IL-1β (ng/ml)

W T W T Ai m Ifn 2 –/– a Ifn r1 –/– a St r2 –/– at Irf 1 –/– 9 –/– W T W T Ai m Ifn 2 –/– a Ifn r1 –/– a St r2 –/– a Irf t1 –/– 9 –/–

a

Cell death (% LDH release)

RESULTS IRF1 mediates AIM2 activation by F. novicida F. tularensis, a pathogen that causes a rapid and lethal infections in humans and mice, is recognized by the AIM2 inflammasome 8,9,22. In unprimed primary mouse bone marrow–derived macrophages (BMDMs), F. novicida induced AIM2-dependent activation of caspase-1, release of IL-1β and IL-18 and cell death in a manner that required type I interferon receptor 1 (IFNAR1) and IFNAR2 and components of the transcription factors ISGF3, STAT1 and IRF9 (refs. 8,9,21,22) (Fig. 1a and Supplementary Fig. 1a). In contrast, activation of the AIM2 inflammasome by transfection of the dsDNA ligand poly(dA:dT) occurred independently of signaling via type I interferons (Fig. 1a). The type I interferon signature that leads to activation of the AIM2 inflammasome in response to infection with F. novicida is unknown. The production of type I interferons induced

Cell death (% LDH release)

© 2015 Nature America, Inc. All rights reserved.

the AIM2 inflammasome in response to certain pathogens, such as the F. tularensis subspecies novicida (F. novicida), but not in response to the DNA virus mouse cytomegalovirus (MCMV)8,21. Therefore, the downstream mechanism that leads to activation of the AIM2 inflammasome in response to different pathogens is probably different and remains to be resolved3. Here we identified two distinct pathways that led to activation of the AIM2 inflammasome. We found that engagement of the AIM2 inflammasome by infection with F. novicida required the transcription factor IRF1 but not TRIF or caspase-11. IRF1 was required for robust expression of small interferon-inducible GTPases known as ­‘guanylatebinding proteins’ (GBPs), which killed cytosolic F. ­ novicida and mediated the release of DNA for access by AIM2. In addition, IRF1deficient (Irf1−/−) mice failed to control infection with F. novicida. In contrast, engagement of AIM2 via MCMV or transfected dsDNA did not require IRF1. Therefore, we have identified a specific requirement for IRF1 and GBPs in F. novicida–induced activation of the AIM2 inflammasome.

0

2

GAPDH

Figure 1  Infection with F. novicida induces IRF1 expression in a manner that requires signaling via type I interferons. (a) Immunoblot analysis of pro-caspase-1 (Pro-Casp-1) and the caspase-1 subunit p20 (Casp-1 p20) (top), as well as analysis of the release of IL-18 (middle) and cell death (assessed as release of lactate dehydrogenase (LDH)) (bottom), in unprimed wild-type (WT) or mutant BMDMs (above lanes and below plots) left uninfected or untreated (medium alone (Med)) or assessed 20 h after infection with F. novicida (multiplicity of infection (MOI), 100) (left) or 5 h after transfection of poly(dA:dT) (right). (b) Release of IL-1β and IL-18 and death of unprimed BMDMs after treatment as in a. (c) Microarray analysis of the expression of genes encoding interferon-regulatory factors (left margin) in Ifnar1−/− BMDMs 8 h after infection with F. novicida, relative to that of their wild-type counterparts. (d) Immunoblot analysis of IRF1 in unprimed BMDMs 0–24 h (above lanes) after infection with F. novicida (MOI, 50). NS, not significant; *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 (one-way analysis of variance (ANOVA) with Dunnett’s multiple-comparisons test). Data are representative of two (b) or three (a,d) independent experiments (mean and s.e.m.) or one experiment with two biological replicates per genotype (c).



aDVANCE ONLINE PUBLICATION  nature immunology

–/ –

Ai m 2

/–

–/ –

–/ –

30 20 10

D N A

0

pc

–/ –

g

/–

–/–

W

Ai m 2

Aim2

T

Irf1

Irf 1 –/– Ai m 2 –/– W T Irf 1–

Po

no F.

ND

0

)

vi ci da

:d T ly (d A Po

–/–

WT

0.4

WT –/– Irf1 –/– Aim2

40

pc

ND

0

)

vi ci da

0

0.5

0.8

NS

50

Cell death (% LDH release)

ND

1.2

pcDNA

D N A

4

1.0

NS

1.6

:d T

8

2.0

ly (d A

12

f

NS

**

1.5

WT –/– Irf1 –/– Aim2

IL-18 (ng/ml)

IL-1β (ng/ml)

***

Med

F. novicida Poly(dA:dT)

IL-18 (ng/ml)

Med

no

Ai m 2 W T Irf 1–

T Casp-1 p20

Casp-1 p20

c

1

W Pro-Casp-1

Pro-Casp-1

b 16

Irf

T

Irf

W

e

1 –/– Ai m 2 –/– W T Irf 1– Ai /– m 2 –/– W T Irf 1 –/ Ai – m 2 –/–

a

F.

Pro-Casp-1 F. novicida

Casp-1 p20 Med

MCMV

NS

4 2

nature immunology  aDVANCE ONLINE PUBLICATION

A:

0.5 MCMV –

0

WT Irf1–/– F. novicida – – + + + + – – + + + + + + IFN-β – + – – + – + + Pro-Casp-1

Casp-1 p20

ly

(d

i

MCMV

WT Irf1–/– Aim2 –/–

1.0

Po

F.

no

vi

ci

dT

)

da

40 wild-type counterparts exhibited robust AIM2 inflammasome responses (Fig. 2a–d). 20 In contrast, Irf1−/− BMDMs responded 0 normally to the transfected dsDNA ligands poly(dA:dT) and plasmid pcDNA3.1 DNA and induced AIM2-dependent maturation of caspase-1, release of IL-18 and cell death (Fig. 2a–f). Consistent with that, Irf1−/− BMDMs had an impaired ability to generate an inflammasome speck in response to infection with F. novicida but not in response to transfected poly(dA:dT) (Supplementary Fig. 1d,e). Analysis of mRNA transcripts of Aim2 and Il1b by quantitative RT-PCR revealed similar expression of these genes in wild-type and Irf1−/− BMDMs infected with F. novicida (Supplementary Fig. 2a). The levels of the pro-inflammatory cytokines tumor-necrosis factor and IL-6 were slightly lower in Irf1−/− BMDMs than in wild-type BMDMs (P > 0.05), whereas the level of the chemokine CXCL1 (KC) was similar in Irf1−/− and wild-type BMDMs (Supplementary Fig. 2b). Accordingly, the level of IL-12 was lower in F. novicida–infected Irf1−/− BMDMs than in F. novicida–infected wild-type BMDMs (Supplementary Fig. 2c). We next investigated whether IRF1 was required for activation of the AIM2 inflammasome in response to MCMV9. Notably, engagement of the AIM2 inflammasome with MCMV did not require IRF1 (Fig. 2c,d,g,h). This indicated that detection of bacteria and viruses by AIM2 occurred via distinct pathways governed by the requirement for IRF1. IRF1 has been identified as a transactivator of IFN-β23, which suggests that defective production of IFN-β might lead to impaired activation of the AIM2 inflammasome. Although we detected reduced expression of the gene encoding IFN-β in Irf1−/− BMDMs

NS

V

NS

*

60

M

80

C

d

WT Irf1–/– Aim2 –/–

NS 1.5

6

0

MCMV

Ifn a Ai r1 –/– m 2 –/

IL-1β (ng/ml)

h

IL-18 (ng/ml)

Poly(dA:dT)

M

Figure 2  IRF1 is essential for activation of the AIM2 inflammasome by infection with F. novicida. (a,b) Immunoblot analysis of caspase-1 (as in Fig. 1a) (a) and the release of IL-1β and IL-18 (b) of unprimed BMDMs left untreated or assessed 20 h after infection with F. novicida (MOI, 100) or 5 h after transfection of poly(dA:dT). (c,d) Microscopic analysis (c) and quantification (b) of the death of unprimed BMDMs 20 h after infection with F. novicida, 5 h after transfection of poly(dA:dT) or 10 h after infection with MCMV (MOI, 10). Arrowheads (c) indicate dead cells. Original magnification (c), ×40. (e,f) Immunoblot analysis of caspase-1 (as in Fig. 1a) (e) and release of IL-18 and death (f) of unprimed BMDMs left untransfected (med) or assessed 5 h after transfection of plasmid DNA (pcDNA). (g,h) Immunoblot analysis of caspase-1 (as in Fig. 1a) (g) and release of IL-1β and IL-18 (h) of unprimed BMDMs left uninfected (Med) or assessed 10 h after infection with MCMV (MOI, 10). (i) Immunoblot analysis of caspase-1 (as in Fig. 1a) in BMDMs left uninfected (–) or infected (+) with F. novicida with (+ or wedge) or without (–) co-stimulation with recombinant mouse IFN-β (at a concentration of 500 U/ml or at increasing concentrations (wedges) of 25, 250 and 500 U/ml). *P < 0.01, **P < 0.001 and ***P < 0.0001 (two-tailed t-test). Data are representative of three (a–f,i) or two (g,h) independent experiments (mean and s.e.m. in b,d,f,h).

Cell death (% LDH release)

© 2015 Nature America, Inc. All rights reserved.

Articles

infected with F. novicida (Supplementary Fig. 2d), priming of Irf1−/− BMDMs with recombinant IFN-β failed to ‘rescue’ activation of the AIM2 inflammasome by F. novicida, whereas stimulation with IFN-β increased AIM2-dependent maturation of caspase-1 in wild-type cells in a dose-dependent way (Fig. 2i). These findings suggested that the defective activation of the AIM2 inflammasome in Irf1−/− BMDMs was not due solely to reduced production of IFN-β but largely reflected defects in additional IRF1-mediated processes. IRF1 is dispensable for the NLRP3 and NLRC4 inflammasomes To investigate whether IRF1 is required for activation of additional inflammasomes, we stimulated unprimed wild-type and Irf1−/− BMDMs with activators of the canonical NLRP3 inflammasome (ATP and nigericin) and activators of the non-canonical NLRP3 inflammasome (C. rodentium and Salmonella enterica serovar Typhimurium (S. Typhimurium) lacking the flagellin subunits fliC and fljB)17. We found that Irf1−/− BMDMs underwent normal activation of caspase-1, release of IL-1β and IL-18 and cell death in response to activators of the canonical or non-canonical NLRP3 inflammasome (Fig. 3a,b). In addition, wild-type and Irf1−/− BMDMs infected with S. Typhimurium grown to log phase, which activates the NLRC4 inflammasome via an SPI-1 type III secretion system23, showed similar levels of activation of caspase-1, release of IL-1β and IL-18 and cell death (Fig. 3c,d), which indicated that IRF1 did not interfere with activation of the NLRC4 inflammasome. 

–/ – –/ –

1

N Irp 3

Irf

1 Irf

N Irp 3 W T

–/ –

–/ –

–/ – –/ –

T

1 Irf

W

N Irp 3 W T

–/ – –/ –

1

N Irp 3

Irf

–/ –

1

N Irp 3 W T

Pro-Casp-1

Casp-1 p20

NS

NS

1.0

0.2

1

Irc

4

–/

12

20

4 0

0.4 0

40 30

m

m tiu

∆f

liC

∆f lj

B

en od .r

C

1.2 0.8

NS

ST

ig N

P

∆f li

1.6 IL-18 (ng/ml)

IL-1β (ng/ml)

8

AT +

C

LP

∆f lj

S

B

en

NS

N

T W

Irf

4 Irc N

NS WT Irf1–/– –/– NIrp3

40

ST m

m tiu

N +

od

S

C –/ –

d

–

–/ –

–

–/

1

.r

LP

ig

P AT + S

B

∆f lj C

∆f li

Irf

T

LP

tiu en od

C .r

ST m

m

ig N

P

+ S

AT

LP

+

Activation of the AIM2 inflammasome requires GBPs Genes encoding GBPs are located on mouse chromosomes 3 and 5 (ref. 24). Those located on chromosome 3 (Gbp1, Gbp2, Gbp3, Gbp5 and Gbp7) encode GBPs linked to the control of bacterial and parasitic replication19,24. To investigate the role of these GBPs in activation of the AIM2 inflammasome, we obtained BMDMs from mice lacking the GBP-encoding locus on chromosome 3 (called ‘Gbp chr3’ here)19,24 and infected the cells with F. novicida. We observed significantly decreased levels of activation of caspase-1, release of IL-1β and IL-18 and induction of cell death in F. novicida– infected BMDMs in which Gbp chr3 was deleted, relative to their levels in F. novicida–infected wild-type BMDMs (Fig. 4c). Gbpchr3deficient BMDMs, however, displayed normal levels of activation of caspase-1, release of IL-18 and induction of cell death in response to transfected poly(dA:dT)19 (Fig. 4c). Of the GBP-encoding genes located on chromosome 3, IRF1 controlled upregulation of the expression of Gbp2, Gbp3 and Gbp5 (Fig. 4a,b). Gbp1 was not identified in our microarray data set (Fig. 4a), consistent with published studies showing that Gbp1 is not expressed in mice on the C57BL/6 background25. Given the importance of IRF1 for activation of the AIM2 inflammasome, we set out to identify which of the remaining GBPs was responsible for this activity.

NS

0

0

STm

NS

60

0.4

0 S

NS

+

0.6

NS

∆fliC∆fljB STm

C. rodentium

S

2.0 IL-18 (ng/ml)

20 3

NS

NS 3.0

Med

LP

NS

NS

LPS + Nig

Cell death (% LDH release)

NS 40

LP

IL-1β (ng/ml)

b

LPS + ATP

Cell death (% LDH release)

Med

GBP-encoding genes require IRF1 c We further investigated the IRF1 signature that led to activation of the AIM2 inflamPro-Casp-1 masome by F. novicida. For this we performed microarray analysis to identify Casp-1 p20 genes expressed differentially in F. novic−/− −/− ida–infected wild-type, Irf1 , Ifnar1 and Med AIM2-deficient (Aim2−/−) BMDMs. Analysis of the expression of genes encoding molecules involved in macrophage-mediated immunity revealed that the gene set with reduced expression in both Irf1−/− BMDMs and Ifnar1−/− BMDMs relative to their expression in wild-type BMDMs was dominated by those encoding GBPs, including Gbp2, Gbp3, Gbp4, Gbp5, Gbp6, Gbp8, Gbp9 and Gbp10 (Fig. 4a). Aim2−/− BMDMs, in contrast, showed elevated expression of genes encoding GBPs and other interferon-inducible genes, relative to their expression in wild-type BMDMs (Fig. 4a,b), an observation supported by the negative regulatory effect of AIM2 on interferon responses8,9. We further confirmed by real-time quantitative RT-PCR analysis the downregulation of GBP-encoding genes in the absence of IRF1 or IFNAR1 (Fig. 4b).



Irf

–/ –

1

T

Irf

W

N Irp 3 W T

–/ –

a

W

© 2015 Nature America, Inc. All rights reserved.

Figure 3  IRF1 is not required for activation of the canonical or non-canonical NLRP3 or NLRC4 inflammasome. (a,b) Immunoblot analysis of caspase-1 (as in Fig. 1a) (a) and release of IL-1β and IL-18 and death (b) of unprimed BMDMs (Med) or LPS-primed BMDMs stimulated with ATP (LPS + ATP) or nigericin (LPS + Nig) or BMDMs left uninfected (Med) or assessed 20 h after infection with C. rodentium (MOI, 20) or ∆fliC∆fljB mutant S. Typhimurium (MOI, 20) (∆fliC∆fljB STm). (c,d) Immunoblot analysis of caspase-1 (as in Fig. 1a) (c) and release of IL-1β and IL-18 and death (d) of unprimed BMDMs left uninfected or assessed 4 h after infection with a wild-type strain of S. Typhimurium grown to log phase (MOI, 1). NS, not significant (two-tailed t-test). Data are from three independent experiments (mean and s.e.m. in b,d).

–/ –

Articles

NS WT –/– Irf1 –/– NIrc4

20 10 0

For this, we individually knocked down each of the remaining GBPencoding genes in the chromosome 3 locus (Gbp2, Gbp3, Gbp5 and Gbp7) in primary wild-type BMDMs by an approach involving small interfering RNA and found that silencing of Gbp2 or Gbp5 led to a significant reduction in the activation of caspase-1 and secretion of IL-1β in response to infection with F. novicida (Supplementary Fig. 3a–c). To confirm the role of GBP2 and GBP5 in activation of the AIM2 inflammasome by infection with F. novicida, we stimulated GBP2deficient (Gbp2−/−) and GBP5-deficient (Gbp5−/−) BMDMs with F. novicida, poly(dA:dT), LPS plus ATP (to activate the NLRP3 inflammasome) or S. Typhimurium grown to log phase (to activate the NLRC4 inflammasome). Both Gbp2−/− and Gbp5−/− BMDMs showed significantly reduced activation of caspase-1, secretion of IL-1β and IL-18 and cell death following infection with F. novicida relative to that in their wild-type counterparts (Fig. 4d–f), which confirmed a role for both GBP2 and GBP5 in activation of the AIM2 inflammasome. In contrast, stimulation of wild-type, Gbp2−/− and Gbp5−/− BMDMs with poly(dA:dT), LPS plus ATP or S. Typhimurium grown to log phase resulted in levels of activation of caspase-1, secretion of IL-1β and IL-18 and cell death similar to those in their wild-type counterparts19 (Fig. 4d–f). These results indicated a non-redundant role for GBP2 and GBP5 in activation of the AIM2 inflammasome in response to infection with F. novicida. GBPs mediate killing of F. novicida to induce AIM2 activation To investigate whether the expression of GBP2 and GBP5 is entirely under the control of IRF1, we infected wild-type and Irf1−/− BMDMs with F. novicida and monitored the dynamics of the expression of GBP2 and GBP5 over the course the infection. We found that the expression of GBP2 and GBP5 was delayed compared with the expression of IRF1 in wild-type BMDMs and was reduced in the absence of IRF1 (Fig. 5a). However, the residual expression of GBP2 and GBP5 in Irf1−/− BMDMs indicated that an alternative IRF1-independent pathway might exist to induce the expression of these GBPs. It is possible that signaling via type I interferons might trigger an ­alternative aDVANCE ONLINE PUBLICATION  nature immunology

T) dA :d y( Po l

Po

**

NS

NS

A: d

ic ov F. n

S.

f

T)

id a

T)

id a ic

ic

dA :d

y(

ov

Po l

F. n Ty + A T p (lo him P g u ph riu as m e)

dT A:

S

WT Gbp2–/– Gbp5–/–

LP

ly

(d

vi

ci

M

no F.

30 20 10 0

)

da

ed

p P (lo him g u ph riu as m e)

)

Ty

Ai m

0

0

AT

dT

+ S.

LP

S

A: (d

Po

ly

vi

ci

da

ed M

no F.

****

NS

NS

NS

****

NS

NS

NS

NS

nature immunology  aDVANCE ONLINE PUBLICATION

AT

+

S

Ty

S.

LP

A:

(d

ly

Po

p (lo him g u ph riu as m e)

P

)

dT

da

ci

vi

no

F.

AT P ph (lo im g u ph riu as m e) S.

Ty

S

LP

A:

(d

+

dT

da

ci

ly

vi

no

F.

Po

p (lo him g u ph riu as m e)

S.

Ty

A:

+

S

LP

(d

ly

AT

dT

P

)

da

vi

ci

Po

no

F.

pathway that leads to GBP expression independently of IRF1. To investigate this, we infected wild-type and Ifnar1−/− BMDMs with F. novicida and found that a lack of signaling via type I interferons completely abolished the expression of GBP2 and GBP5 (Fig. 5a). Together these findings provided evidence demonstrating that two different members of the GBP family, GBP2 and GBP5, were under the control of IRF1 and signaling via type I interferons and were expressed to specifically engage the AIM2 inflammasome. GBPs can target vacuolar bacteria, such as Salmonella, and induce the recruitment of antimicrobial peptides to kill the bacteria19,26. Whether GBPs mediate killing of cytosolic bacteria such as F. novicida has not been investigated. We therefore used confocal microscopy to observe the spatial distribution of GBP5 relative to that of F. novicida in infected BMDMs. GBP5 was recruited to and engulfed F. novicida bacteria in wild-type BMDMs (Fig. 5b and Supplementary Fig. 4). The absence of IRF1 led to a significantly reduced prevalence of GBP5-associated bacteria (Fig. 5c). BMDMs lacking IRF1 or Gbpchr3 also failed to control bacterial replication over time (Fig. 5d). To investigate whether GBP5 directly affected the viability of bacteria in macrophages, we infected wild-type and Gbp5−/− BMDMs with F. novicida expressing green fluorescent protein (GFP) and quantified bacteria in these cells over time. Single-cell analysis revealed

)

IL-18 (ng/ml)

Cell death (% LDH release)

Figure 4  IRF1 controls the 60 40.0 2.5 **** *** ** expression of GBPs for activation 20.0 2.0 40 of the AIM2 inflammasome. 5.0 1.5 (a) Microarray analysis of genes 1.0 20 2.5 0.5 encoding molecules involved in 0 0 0 macrophage-mediated immunity with higher or lower expression in unprimed Irf1−/−, Ifnar1−/− or Aim2−/− BMDMs 8 h after infection with F. novicida than in their unprimed wild-type counterparts. (b) Real-time quantitative RT-PCR analysis of GBP-encoding genes in BMDMs 8 h after infection with F. novicida, presented relative to that of the gene encoding β-actin. (c–f) Immunoblot analysis of caspase-1 (as in Fig. 1a) (c, far left, and d,e), release of IL-1β and IL-18 (c, middle, and f, left and middle) and death (c and f, right) of unprimed wild-type BMDMs, Gbpchr3-deficient BMDMs (Gbpchr3-KO) or Aim2−/−, Gbp2−/− or Gbp5−/− BMDMs left untreated (Med) or assessed 20 h after infection with F. novicida (MOI, 100), after transfection of poly(dA:dT) or stimulation with LPS plus ATP, or 4 h after infection with S. Typhimurium grown to log phase (MOI, 1). *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 (two-tailed t-test (c) or one-way ANOVA with Dunnett’s multiple comparisons test (f)). Data are representative of one experiment with two biological replicates per genotype (a) or are from three independent experiments (b–f; mean and s.e.m. in b,c,f). IL-1β (ng/ml)

© 2015 Nature America, Inc. All rights reserved.

0.2

(d

0.5

0.4

y

1.0

1.0

Po l

1.5

NS

*

100 75 50

Cell death (% LDH release)

2.0

NS

*

2.0

WT Gbpchr3-KO

dA :d

IL-1β (ng/ml)

–/ –

–/ –

NS

*

2.5

ov

Casp-1 p20

–

ch r3

0

F. n

Casp-1 p20

id a

5

T)

Pro-Casp-1

ic ov F. n

10

id a

Pro-Casp-1

W T G bp ch r3 Ai m -KO 2 –/– W T G bp ed

15

–/ –

–/ –

e

Casp-1 p20

W T G bp W 2 T G bp W 2 T G b W p2 T G bp W 2 T G bp 2

3

ND

F. Me no d vi ci da

0

0

y(

5

10

Po l

10

Pro-Casp-1

20

M

15

20

30

c

WT lrf1–/– lfnar1–/– Aim2–/–

IL-18 (ng/ml)

0

40

F. Me no d vi ci da

ND

–

d

0

W T G b W p5 –/ – T G bp 5 W – T /– G bp W 5 –/– T G bp W 5 –/– T G bp 5 –/

0

5

5

F. Me no d vi ci da

Expression (log2) –3

10

20

10

Gbp4 expression (fold)

15

ND

15

–/ –

10 8 6 4 2 0

Gbp9 expression (fold)

b

–/–

Aim2

Gbp3 expression (fold)

–/–

Gbp8 expression (fold)

lfnar1

Gbp2 expression (fold)

–/–

Irf1

Adrb3 Cxcl10 Cxcl11 Cxcl9 Fcgr1 Fcgr4 Gbp2 Gbp3 Gbp4 Gbp5 Gbp6 Gbp8 Gbp9 Gbp10 lfna1 Ifna5 Ifna6 ll10 ll12 Stat2

Gbp5 expression (fold)

a

-K O 2 –/– W T G bp ch r3 Ai m -KO 2 –/

Articles

that wild-type BMDMs restricted bacterial replication from 4 h to 16 h after infection, whereas Gbp5−/− BMDMs failed to control bacterial replication at 16 h and harbored a significantly larger number of bacteria than wild-type BMDMs had (Fig. 5e and Supplementary Fig. 4). In addition, GBP5-associated bacteria tended to lose their GFP expression, in contrast to non–GBP5-associated bacteria (Supplementary Fig. 4), which might suggest a loss of bacterial viability following recruitment of GBP5 (ref. 19). To investigate the hypothesis that IRF1 mediates the AIM2-dependent killing of intracellular bacteria, we infected unprimed wild-type, Irf1−/− and Aim2−/− BMDMs with GFPexpressing F. novicida and quantified bacteria in these cells over time. In agreement with that hypothesis, Irf1−/− and Aim2−/− BMDMs failed to suppress bacterial replication over 24 h of infection (Fig. 5f). For AIM2 to detect dsDNA, F. novicida or its DNA must be able to escape the vacuole and enter the cytoplasm. Indeed, co-staining of the inflammasome and DNA in F. novicida-infected BMDMs revealed colocalization of DNA with the inflammasome speck (Supplementary Fig. 5a). To investigate whether F. novicida that has escaped from the vacuole engages IRF1- and GBP-dependent activation of the AIM2 inflammasome, we infected wild-type BMDMs with wild-type F. novicida or an isogenic mutant form of F. novicida that fails to disrupt the vacuole as required for escape into the cytoplasm (lacking 

Articles Irf1–/–

Ifnar1–/– 24

WT

0 2 8 16 24

F. novicida (h)

0 2 8 16 24

WT

0 2 8 16 24 0 2 8 16

a

b

WT Irf1–/–

WT

GBP5

DNA

Merge

Inset

IRF1 GBP2 GBP5 GAPDH

© 2015 Nature America, Inc. All rights reserved.

0

4h 8h

NS

4h

**

8h

**** *

24 h

WT Irf1–/– Gbpchr3-KO

e

NS

****

30

WT Gbp5 –/–

20 10 0

4h

16 h

g

f

NS 50 40 30 20 10 0

NS

4h

**** ****

**** ****

8h

24 h

WT Irf1–/– –/– Aim2

–

5

108 7 10 106 1054 10 3 10 2 10

F. novicida per BMDM

10

WT Irf1–/–

NS

W cGT A Ai S –/– m W 2 –/– T cG Ai A S – m /– 2 –/

**** *

NS

F. novicida per BMDM

15

GBP5 + DNA

d

F. novicida (CFU)

GBP5+ bacteria (%)

c

the gene encoding MglA, which regulates various pathogenicity island proteins, including IglC (F. novicida ∆mglA))27. F. novicida ∆mglA retained the ability to induce IRF1 and the expression of GBP2 and GBP5, although at a lower capacity relative to that of wild-type F. novicida (Supplementary Fig. 5b). However, F. novicida ∆mglA failed to activate the AIM2 inflammasome to trigger the release of IL-1β and IL-18 and cell death (Supplementary Fig. 5c,d). The production of other proinflammatory cytokines, tumor-necrosis factor, IL-6 and KC by wild-type BMDMs infected with F. novicida ∆mglA was similar to that of wild-type BMDMs infected with wild-type F. novicida (Supplementary Fig. 5e), which demonstrated that the inability of F. novicida ∆mglA to activate the AIM2 inflammasome was not due to a general defect in its ability to stimulate cytokines in macrophages. These results suggested that escape from the vacuole into the cytosol was essential for activation of the AIM2 inflammasome and that such escape enhanced but was not indispensable for the induction of IRF1-mediated expression of GBPs. It is possible that additional processes that complement GBP expression are required for full engagement of the AIM2 inflammasome. Indeed, deficiency in IRF1 did not completely abolish the expression of GBP2 and GBP5 (Fig. 5a), which suggested that IRF1 might direct additional processes other than driving GBP expression to activate the AIM2 inflammasome. We next investigated whether the absence of IRF1 prevented escape of F. novicida into the cytosol, which might be an additional mechanism that governs the IRF1-dependent activation of the AIM2 inflammasome. Cytosolic pathogens such as F. novicida escape the vacuole and can be used as a mechanism for delivering ligands into the cytosol of a cell18. We took advantage of this cytosolic delivery method and infected wild-type, Irf1−/− and Aim2−/− BMDMs with F. novicida in the presence of ultrapure LPS from Salmonella to investigate whether Salmonella LPS can be efficiently introduced into the cytosol to ­activate the inflammasome in the absence of IRF1. We found that wild-type, Irf1−/− and Aim2−/− BMDMs all underwent induction of 

IL-1β (ng/ml)

*** 1.5 Figure 5  GBPs target bacteria to mediate killing during infection with F. novicida. WT (a) Immunoblot analysis of IRF1, GBP2, GBP5 and GAPDH (loading control) in unprimed 1.0 Pro-Casp-1 cGAS –/– BMDMs at various times (above lanes) after infection with F. novicida (MOI, 50). 0.5 Casp-1 p20 (b) Immunofluorescence staining of GBP5 in unprimed BMDMs 16 h after infection 0 with F. novicida; outlined area (right) indicates region enlarged at far right. Scale bars, F. novicida Med F. novicida 10 µm (main images) or 1 µm (inset, far right). (c) Frequency of bacteria co-localized with GBP5 in wild-type BMDMs (n = 3,575 bacteria) and Irf1−/− BMDMs (n = 3,620 bacteria). (d) Recovery of F. novicida (in colony-forming units (CFU)) from wild-type, Irf1−/− and Gbpchr3-deficient BMDMs at 4, 8 or 24 h after infection with F. novicida (MOI, 100). (e) Quantification of F. novicida in wild-type BMDMs (n = 758) and Gbp5−/− BMDMs (n = 679) (f) or wild-type BMDMs (n = 715), Irf1−/− BMDMs (n = 710) and Aim2−/− BMDMs (n = 706) at various times (horizontal axes) after infection, assessed by confocal microscopy. Each symbol represents an individual cell. (g) Immunoblot analysis of caspase-1 (as in Fig. 1a) (left) and release of IL-1β (right) of unprimed wild-type, cGAS-deficient (Mb21d1−/−; called ‘cGAS−/−’ here) and Aim2−/− BMDMs 20 h after infection with F. novicida (MOI, 100). *P < 0.05, **P < 0.005, ***P < 0.001 and ****P < 0.0001 (two-tailed t-test (c,g) or one-way ANOVA with Dunnett’s multiple-comparisons test (d) or Tukey’s multiple comparisons test (e,f)). Data are from three independent experiments (a–c,g; mean and s.e.m. in c,d,g) or are from one experiment representative of two independent experiments (d–f).

the activation of caspase-1 in response to infection with F. novicida in the presence of Salmonella LPS (Supplementary Fig. 6a); this indicated that the LPS was delivered into the cytosol, which probably activated the non-canonical NLRP3 inflammasome. These findings suggested that in the absence of IRF1, escape into the cytosol was not fully compromised at 20 h after infection. DNA sensor cGAS mediates activation of the AIM2 inflammasome Although signaling via type I interferons that engages the AIM2 inflammasome requires IRF1, the pattern-recogntion receptors that induce the production of type I interferons in response to infection with F. novicida have remained largely unexplored. A major pathway that mediates the recognition of cytosolic bacteria for the generation of IFN-β production is the signaling axis of the DNA sensor cyclic GMP-AMP synthase (cGAS) and the adaptor STING22,28. In agreement with that, both cGASdeficient BMDMs and STING-deficient BMDMs (with nonfunctional (‘Gt’) alleles encoding STING that result in a lack of detectable protein)28 had impaired ability to robustly activate caspase-1 or induce the release of IL-1β and IL-18 or cell death in response to infection with F. novicida (Fig. 5g and Supplementary Fig. 6b,c). In addition, cGAS-deficient BMDMs failed to robustly induce IRF1 expression (Supplementary Fig. 6d). The residual inflammasome responses in cGAS-deficient and STING-deficient BMDMs infected with F. novicida could have been due to a minor source of type I interferons induced by TLRs, RLRs or other PRRs. Collectively, these findings indicated that the cGAS signaling axis induced type I interferon–dependent expression of IRF1 necessary for the engagement GBP-mediated killing of F. novicida, which ultimately resulted in the release of bacterial DNA for activation of the AIM2 inflammasome (Supplementary Fig. 6e). IRF1 provides protection against infection with F. novicida We extended our findings to an in vivo setting and infected wild-type mice, Irf1−/− mice, Aim2−/− mice and mice deficient in both ­caspase-1 aDVANCE ONLINE PUBLICATION  nature immunology

Articles

90 80 70

IL-18 (pg/ml)

© 2015 Nature America, Inc. All rights reserved.

1,000 800 600 400 200 0

*** *** ***

012345678 Time (d)

f

WT

lrf1

–/–

g

c

WT

108

Liver

Spleen

**

**

107 10

6

105 10

4

–/–

lrf1

10109 108 107 106 105 104 103 10

h

d

WT lrf1–/–

WT

Spleen

Liver 107

* F. novicida (CFU/g tissue)

100

0 1 2 3 4 5 6 7 8 Time (d)

e

WT lrf1–/– Aim2–/– Casp1–/– Casp11–/–

110

F. novicida (CFU/g tissue)

b

F. novicida (CFU/g tissue)

****

WT lrf1–/– Aim2–/– Casp1–/– Casp11–/–

F. novicida (CFU/g tissue)

100 80 60 40 20 0

Body weight (%)

Survival (%)

a

106 105 104 3

10 –/–

lrf1

i

WT

109 108 107 106 105 104

*

WT Aim2–/–

–/–

lrf1

WT –/– lrf1 –/– Aim2 Casp1–/– Casp11–/–

Figure 6  IRF1 provides host protection against infection with F. novicida in vivo. (a) Survival of 8-week-old wild-type mice (n = 28), Irf1−/− mice (n = 17), Aim2−/− mice (n = 13) and Casp1−/−Casp11−/− mice (n = 5) infected subcutaneously with 7.5 × 104 CFU of F. novicida. (b) Body weight of 8-week-old wild-type mice (n = 16), Irf1−/− mice (n = 9), Aim2−/− mice (n = 9) and Casp1−/−Casp11−/− mice (n = 5) 0–8 d (horizontal axis) after infection as in a, presented relative to initial body weight at day 0, set as 100%. (c,d) Bacterial burden in the liver and spleen of 8-week-old male wild-type mice (n = 8) and Irf1−/− mice (n = 6) (c) or 8-week-old female wild-type mice (n = 12) and Aim2−/− mice (n = 9) (d) on day 2 after infection with 1 × 105 CFU of F. novicida. (e) Concentration of IL-18 in the serum of wild-type mice (n = 15), Irf1−/− mice (n = 11), Aim2−/− mice (n = 11), and Casp1−/−Casp11−/− mice (n = 5) 24 h after infection as in c,d. (f) Hematoxylin-and-eosin staining (f), Gram staining (g), myeloperoxidase staining (h) and TUNEL staining (i) of livers from wild-type and Irf1−/− mice on day 3 after infection with F. novicida: dashed circled (f) indicate granulomas; arrowheads (g) indicate bacterial colonies. Original magnification, 10× (top row) or 40× (bottom row). Each symbol (c–e) represents an individual mouse; small horizontal lines indicate the mean (±s.e.m.). *P < 0.05, **P < 0.001, ***P < 0.0001 and ****P < 0.0001 (log-rank test (a), two-tailed t-test (c,d) or one-way ANOVA with Dunnett’s multiple-comparisons test (e)). Data are pooled from two independent experiments (a,c–e; mean and s.e.m. in b) or are from one experiment representative of two independent experiments (b).

and caspase-11 (Casp1−/−Casp4−/−; called ‘Casp1−/−Casp11−/−’ here) with F. novicida and monitored their susceptibility to infection. Irf1−/−, Aim2−/− and Casp1−/−Casp11−/− mice lost more body weight than wild-type mice did, and all succumbed to infection within 6 d, whereas 75% of the wild-type mice survived beyond day 6 (Fig. 6a,b). Analysis of bacterial burdens showed that Irf1−/− mice harbored a significantly larger burden of F. novicida in the liver and spleen than did wild-type mice (Fig. 6c). Similarly, Aim2−/− mice were significantly more susceptible to infection with F. novicida than were wildtype mice8,22 (Fig. 6d). The greater susceptibility of Irf1−/− mice than Aim2−/− mice to infection with F. novicida might suggest that IRF1 is important in multiple antimicrobial defense mechanisms. Analysis of serum IL-18 showed that Irf1−/− mice had an impaired ability to produce this cytokine following infection with F. novicida; Aim2−/− and Casp1−/−Casp11−/− mice had a similar phenotype for this (Fig. 6e). Histopathology analysis of liver tissues showed that Irf1−/− mice failed to control the dissemination of F. novicida (Fig. 6f,g). In contrast, granulomas that surround infectious particles were present in the liver of wild-type mice (Fig. 6g). In addition, Irf1−/− mice had more dead cells than wild-type mice had, but their granulocyte recruitment in the liver was similar to that of wild-type mice, as revealed by TUNEL staining and staining of myeloperoxidase, respectively (Fig. 6h,i). The greater prevalence of TUNEL staining in Irf1−/− mice was probably the result of more damage from their greater bacterial burden, which leads to the death of many cell types in the liver. Together these results highlighted a role for IRF1 in the host defense against F. novicida infection via engagement of the AIM2 inflammasome. DISCUSSION Various pathways exist for the recognition of DNA for triggering innate immune responses in the cell. TLR9 recognizes CpG dinucleo­ tide DNA in the endosomal compartment to activate the transcription of genes encoding pro-inflammatory cytokines29, and STING responds directly to cyclic dinucleotides or serves as an adaptor for a range of DNA sensors to induce type I interferon responses30. nature immunology  aDVANCE ONLINE PUBLICATION

AIM2 mediates the sensing of cytosolic dsDNA to initiate assembly of the inflammasome. However, the precise mechanisms by which different pathogens trigger AIM2 have remained unclear3. In addition, the function of signaling via type I interferons in activation of the AIM2 inflammasome has remained unclear, given that such signaling is important for mediating activation of the AIM2 inflammasome in response to infection with F. novicida but not in response to infection with MCMV3. Here we have identified an IRF1-dependent pathway downstream of signaling via type I interferons that was responsible for activation of the AIM2 inflammasome by F. novicida. The existence of this ‘noncanonical’ AIM2 inflammasome pathway defined by its requirement for IRF1, GBPs and signaling via type I interferons is analogous to the non-canonical NLRP3 inflammasome and its requirement for signaling via type I interferons and caspase-11 (refs. 12,13,15,17). IRF1 was first identified as a virus-induced transcription factor that interacts with regulatory DNA elements of the gene encoding human IFN-β and that contributes to the transcription of the gene encoding IFN-β31. Subsequent studies have revealed that both human and mouse IRF1 are involved in the expression of IFN-α, major histocompatibility class I, the GTPase MX (in mice), 2′-5′-oligo A synthetase (in mice) and the regulator ISG54 (in humans)32. We have shown here that IRF1 was not absolutely required for the production of IFN-β and certain subtypes of IFN-α in F. novicida–infected macrophages. Instead, we found that cGAS and STING were probably the upstream sensors that initiated the production of type I interferons. Notably, IRF1 and type I interferons induce GBPs to kill and lyse cytosolic bacteria to drive the release of bacterial DNA for access by AIM2. Although our microarray analysis suggested that IRF1 and IFNAR induced a similar expression profile of interferon-stimulated genes, protein-expression analysis revealed that the expression of GBP2 and GBP5 was completely abolished in the absence of IFNAR1, whereas we observed some residual expression of these proteins in the absence of IRF1. This suggested that IRF1 and IFNAR1 probably each regulated their own set of genes in 

© 2015 Nature America, Inc. All rights reserved.

Articles F. novicida–infected ­macrophages, some of which might be commonly regulated by both IRF1 and IFNAR1. Herpesviruses, including MCMV, enter the host cell via a mechanism that requires binding and fusion of the viral envelope with the host cell membrane, which results in release of the viral capsid into the host cell. Under these circumstances, viral DNA may have access to the cytoplasm directly for detection by AIM2. As shown before33, we found that mutant F. novicida that fails to escape the vacuole did not activate the AIM2 inflammasome. In addition, we showed that escape of F. novicida from the vacuole into the cytoplasm enhanced the expression of IRF1 and GBP, which suggested that this process might provide an additional signal for enhancing GBP-mediated killing of the bacteria in the cytosol. F. tularensis live vaccine strain escapes the vacuole as early as 1 h after infection34; this suggests that escape from vacuoles precedes GBP expression, which occurs mostly beyond 16–24 h after infection, following signaling via type I interferons. However, the precise method by which GBPs kill F. novicida is unknown. Published studies have found that GBP1 or GBP7 can recruit additional effector proteins to mediate the killing of pathogens. GBP7 interacts with and recruits p22 phox and other components of the NADPH oxidase to the mycobacteria-containing vacuole to mediate the killing of bacteria, whereas GBP1 facilitates delivery of ubiquitinylated cargo (for example, bacteria decorated with ubiquitin) to autolysosomes to mediate killing26. GBP1, GBP2 and GBP5 share a common feature in that they harbor a carboxyterminal ‘CaaX’ motif (a cysteine residue followed by two aliphatic residues and then any amino acid residue) that is targeted by prenylation, a post-translational modification process that results in the attachment of a lipid hydrophobic moiety to mediate docking of the protein to cellular membranes35. It is possible that GBP2 and GBP5 might directly target cytosolic F. novicida to induce recruitment of an antimicrobial process to mediate the killing of this bacteria. Indeed, IRF1 controls the expression of inducible nitric oxide synthase36. The contribution of inducible nitric oxide synthase and its connection with GBPs in the activation of the AIM2 inflammasome warrants further investigation. We observed that macrophages lacking either IRF1 or AIM2 failed to restrict F. novicida replication, whereas mice lacking IRF1 were more susceptible to infection with F. novicida than were mice lacking AIM2. These results suggested that IRF1 controls multiple antimicrobial pathways to suppress the bacterial burden in vivo, which ultimately leads to lysis of the bacterial pathogen for activation of the AIM2 inflammasome. IRF1 has been linked to the production of IL-12. However, published studies have shown that at 3 d after infection with F. tularensis live vaccine strain, mice lacking the IL-12 subunit p40 or p30 or mice treated with antibodies to IL-12 have a bacterial burden similar that of their wild-type or PBS-treated counterparts37. In our study, mice lacking IRF1 had a significantly higher bacterial burden than did wild-type mice on day 2, and by day 3, more than 80% of the mice lacking IRF1 succumbed to infection. Thus, we speculate that IL-12 probably does not account for the substantial difference between wild-type mice and Irf1−/− mice in their bacterial burden. Additional mechanisms have been identified that control assembly and degradation of the AIM2 inflammasome. Polyubiquitination of ASC recruits the autophage marker p62 to target the AIM2 inflammasome for degradation by autophagy38. The human pyrin domain– ­containing protein POP3 interacts with the pyrin domain of AIM2 and competes with ASC to inhibit activation of the AIM2 inflammasome in response to poly(dA:dT) and MCMV39. The existence of these pathways indicates that controlling activation of the AIM2 inflammasome is important for maintaining homeostasis of the cell. 

Indeed, aberrant activation of the AIM2 inflammasome in response to dsDNA could lead to considerable physiological consequences. Increased expression of AIM2 is associated with the development of psoriasis, an abdominal aortic aneurysm and systemic lupus ­erythematosus, whereas reduced expression of AIM2 is linked to colorectal and prostate cancer40. Our findings have identified a previously unknown layer of regulation that governs cytoplasmic sensing of DNA after infection with intracellular pathogens. Therapies that modulate IRF1 activities might lead to enhanced protection against rapid and lethal infection with F. tularensis. Methods Methods and any associated references are available in the online version of the paper. Accession codes. GEO: microarray data, GSE66461. Note: Any Supplementary Information and Source Data files are available in the online version of the paper. Acknowledgments We thank V.M. Dixit (Genentech) and N. Kayagaki (Genentech) for Aim2-deficient mice and Gbp5-deficient mice; K.A. Fitzgerald (University of Massachusetts Medical School) for femurs from cGAS-deficient mice; P. Broz (University of Basel) for femurs from Gbpchr3-deficient mice; K. Pfeffer (Heinrich-HeineUniversity Duesseldorf) for femurs from GBP2-deficient mice; P.G. Thomas (St. Jude Children’s Research Hospital) for the MCMV Smith MSGV strain; and X. Qi and M. Barr for technical assistance. Supported by the US National Institutes of Health (AR056296, CA163507 and AI101935), the American Lebanese Syrian Associated Charities (T.-D.K.), European Research Council (281600), the Fund for Scientific Research-Flanders (G030212N to M.L.), St. Jude Children’s Research Hospital and the National Health and Medical Research Council (Neoma Boadway Endowed Fellowship and R.G. Menzies Early Career Fellowship to S.M.M.). AUTHOR CONTRIBUTIONS S.M.M., R.K. and T.-D.K. designed the study; S.M.M., R.K., R.K.S.M., G.N. and P.V. performed experiments and analyzed the data; M.Y. contributed reagents; and S.M.M., M.L. and T.-D.K. wrote the manuscript. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html. 1. Strowig, T., Henao-Mejia, J., Elinav, E. & Flavell, R. Inflammasomes in health and disease. Nature 481, 278–286 (2012). 2. Lamkanfi, M. & Dixit, V.M. Mechanisms and functions of inflammasomes. Cell 157, 1013–1022 (2014). 3. Rathinam, V.A.K., Vanaja, S.K. & Fitzgerald, K.A. Regulation of inflammasome signaling. Nat. Immunol. 13, 333–342 (2012). 4. Bürckstümmer, T. et al. An orthogonal proteomic-genomic screen identifies AIM2 as a cytoplasmic DNA sensor for the inflammasome. Nat. Immunol. 10, 266–272 (2009). 5. Fernandes-Alnemri, T., Yu, J.W., Datta, P., Wu, J. & Alnemri, E.S. AIM2 activates the inflammasome and cell death in response to cytoplasmic DNA. Nature 458, 509–513 (2009). 6. Hornung, V. et al. AIM2 recognizes cytosolic dsDNA and forms a caspase-1activating inflammasome with ASC. Nature 458, 514–518 (2009). 7. Roberts, T.L. et al. HIN-200 proteins regulate caspase activation in response to foreign cytoplasmic DNA. Science 323, 1057–1060 (2009). 8. Fernandes-Alnemri, T. et al. The AIM2 inflammasome is critical for innate immunity to Francisella tularensis. Nat. Immunol. 11, 385–393 (2010). 9. Rathinam, V.A. et al. The AIM2 inflammasome is essential for host defense against cytosolic bacteria and DNA viruses. Nat. Immunol. 11, 395–402 (2010). 10. Dombrowski, Y. et al. Cytosolic DNA triggers inflammasome activation in keratinocytes in psoriatic lesions. Sci. Transl. Med. 3, 82ra38 (2011). 11. Jin, T. et al. Structures of the HIN domain:DNA complexes reveal ligand binding and activation mechanisms of the AIM2 inflammasome and IFI16 receptor. Immunity 36, 561–571 (2012). 12. Rathinam, V.A. et al. TRIF licenses caspase-11-dependent NLRP3 inflammasome activation by Gram-negative bacteria. Cell 150, 606–619 (2012). 13. Broz, P. et al. Caspase-11 increases susceptibility to Salmonella infection in the absence of caspase-1. Nature 490, 288–291 (2012).

aDVANCE ONLINE PUBLICATION  nature immunology

© 2015 Nature America, Inc. All rights reserved.

Articles 14. Sander, L.E. et al. Detection of prokaryotic mRNA signifies microbial viability and promotes immunity. Nature 474, 385–389 (2011). 15. Gurung, P. et al. Toll or interleukin-1 receptor (TIR) domain-containing adaptor inducing interferon-β (TRIF)-mediated caspase-11 protease production integrates Toll-like receptor 4 (TLR4) protein- and Nlrp3 inflammasome-mediated host defense against enteropathogens. J. Biol. Chem. 287, 34474–34483 (2012). 16. Kayagaki, N. et al. Noncanonical inflammasome activation by intracellular LPS independent of TLR4. Science 341, 1246–1249 (2013). 17. Kayagaki, N. et al. Non-canonical inflammasome activation targets caspase-11. Nature 479, 117–121 (2011). 18. Hagar, J.A., Powell, D.A., Aachoui, Y., Ernst, R.K. & Miao, E.A. Cytoplasmic LPS activates caspase-11: implications in TLR4-independent endotoxic shock. Science 341, 1250–1253 (2013). 19. Meunier, E. et al. Caspase-11 activation requires lysis of pathogen-containing vacuoles by IFN-induced GTPases. Nature 509, 366–370 (2014). 20. Shi, J. et al. Inflammatory caspases are innate immune receptors for intracellular LPS. Nature 514, 187–192 (2014). 21. Henry, T., Brotcke, A., Weiss, D.S., Thompson, L.J. & Monack, D.M. Type I interferon signaling is required for activation of the inflammasome during Francisella infection. J. Exp. Med. 204, 987–994 (2007). 22. Jones, J.W. et al. Absent in melanoma 2 is required for innate immune recognition of Francisella tularensis. Proc. Natl. Acad. Sci. USA 107, 9771–9776 (2010). 23. Man, S.M. et al. Inflammasome activation causes dual recruitment of NLRC4 and NLRP3 to the same macromolecular complex. Proc. Natl. Acad. Sci. USA 111, 7403–7408 (2014). 24. Yamamoto, M. et al. A cluster of interferon-γ-inducible p65 GTPases plays a critical role in host defense against Toxoplasma gondii. Immunity 37, 302–313 (2012). 25. Staeheli, P., Prochazka, M., Steigmeier, P.A. & Haller, O. Genetic control of interferon action: mouse strain distribution and inheritance of an induced protein with guanylate-binding property. Virology 137, 135–142 (1984). 26. Kim, B.H. et al. A family of IFN-γ inducible 65-kD GTPases protects against bacterial infection. Science 332, 717–721 (2011). 27. Santic, M., Molmeret, M., Klose, K.E., Jones, S. & Kwaik, Y.A. The Francisella tularensis pathogenicity island protein IglC and its regulator MglA are essential for modulating phagosome biogenesis and subsequent bacterial escape into the cytoplasm. Cell. Microbiol. 7, 969–979 (2005).

nature immunology  aDVANCE ONLINE PUBLICATION

28. Sauer, J.D. et al. The N-ethyl-N-nitrosourea-induced Goldenticket mouse mutant reveals an essential function of Sting in the in vivo interferon response to Listeria monocytogenes and cyclic dinucleotides. Infect. Immun. 79, 688–694 (2011). 29. Hemmi, H. et al. A Toll-like receptor recognizes bacterial DNA. Nature 408, 740–745 (2000). 30. Ishikawa, H., Ma, Z. & Barber, G.N. STING regulates intracellular DNA-mediated, type I interferon-dependent innate immunity. Nature 461, 788–792 (2009). 31. Fujita, T. et al. Evidence for a nuclear factor(s), IRF-1, mediating induction and silencing properties to human IFN-β gene regulatory elements. EMBO J. 7, 3397–3405 (1988). 32. Miyamoto, M. et al. Regulated expression of a gene encoding a nuclear factor, IRF-1, that specifically binds to IFN-beta gene regulatory elements. Cell 54, 903–913 (1988). 33. Mariathasan, S., Weiss, D.S., Dixit, V.M. & Monack, D.M. Innate immunity against Francisella tularensis is dependent on the ASC/caspase-1 axis. J. Exp. Med. 202, 1043–1049 (2005). 34. Checroun, C., Wehrly, T.D., Fischer, E.R., Hayes, S.F. & Celli, J. Autophagy-mediated reentry of Francisella tularensis into the endocytic compartment after cytoplasmic replication. Proc. Natl. Acad. Sci. USA 103, 14578–14583 (2006). 35. Britzen-Laurent, N. et al. Intracellular trafficking of guanylate-binding proteins is regulated by heterodimerization in a hierarchical manner. PLoS ONE 5, e14246 (2010). 36. Kamijo, R. et al. Requirement for transcription factor IRF-1 in NO synthase induction in macrophages. Science 263, 1612–1615 (1994). 37. Elkins, K.L., Cooper, A., Colombini, S.M., Cowley, S.C. & Kieffer, T.L. In vivo clearance of an intracellular bacterium, Francisella tularensis LVS, is dependent on the p40 subunit of interleukin-12 (IL-12) but not on IL-12 p70. Infect. Immun. 70, 1936–1948 (2002). 38. Shi, C.S. et al. Activation of autophagy by inflammatory signals limits IL-1β production by targeting ubiquitinated inflammasomes for destruction. Nat. Immunol. 13, 255–263 (2012). 39. Khare, S. et al. The PYRIN domain-only protein POP3 inhibits ALR inflammasomes and regulates responses to infection with DNA viruses. Nat. Immunol. 15, 343–353 (2014). 40. Connolly, D.J. & Bowie, A.G. The emerging role of human PYHIN proteins in innate immunity: implications for health and disease. Biochem. Pharmacol. 92, 405–414 (2014).



ONLINE METHODS

© 2015 Nature America, Inc. All rights reserved.

Mice. Irf1−/− mice41, Irf7−/− mice42, Irf9−/− (p48) mice43, Ifnar1−/− mice44, Ifnar2−/− mice44, Stat1−/− mice45, Aim2−/− mice22, Nlrp3−/− mice46, Nlrc4−/− mice47, Casp1−/−Casp11−/− mice17, Casp11−/− mice17, mice in which Gbpchr3 was deleted24, Gbp5−/− mice19, Trif−/− mice48 and STING-deficient mice28 have been described. Mice were bred at the St. Jude Children’s Research Hospital. Animal studies were conducted under protocols approved by the St. Jude Children’s Research Hospital on the Use and Care of Animals. Cell culture and stimulation. Primary BMDMs were grown for 5–6 d in IMDM (Gibco) supplemented with 1% non-essential amino acids (Gibco), 10% FBS (Atlanta Biologicals), 30% medium conditioned by L929 mouse fibroblasts, and 1% penicillin and streptomycin (Sigma). BMDMs in antibiotic–free medium were seeded onto 12-well plates at a density of 1 × 106 cells per well, followed by incubation overnight. F. novicida strain U112 or its isogenic mutant F. novicida ∆mglA were grown overnight under aerobic conditions at 37 °C in BBL Trypticase Soy Broth (211768; BD) supplemented with 0.2% l-cysteine (Fisher). Bacteria were subcultured (1:10) for 4 h in fresh Trypticase Soy Broth supplemented with 0.2% l-cysteine and resuspended in PBS. S. Typhimurium strain SL1344, an isogenic mutant lacking fliC and fljB (∆fliC∆fljB STm) and C. rodentium (51459; American Type Culture Collection) were inoculated into LB broth and were incubated overnight under aerobic conditions at 37 °C. S. Typhimurium SL1344 was subcultured (1:10) for 3 h at 37 °C in fresh LB broth to generate bacteria grown to log phase. The following conditions were used: F. novicida or F. novicida ∆mglA, an MOI of 100 and 20 h of incubation (for activation of caspase-1) and an MOI of 50 and 2, 8, 16 and 24 h of incubation (for expression of IRF1 or GBP); S. Typhimurium, an MOI 1 and 4 h of incubation; ∆fliC∆fljB STm, an MOI of 20 and 20 h of incubation; and C. rodentium, an MOI 20 and 20 h of incubation. Gentamicin (50 µg/ml, Gibco) was added at 2 h (S. Typhimurium), 4 h (C. rodentium and ∆fliC∆fljB STm) or 8 h (F. novicida) after infection to kill extracellular bacteria. The MCMV Smith MSGV strain (VR-1399; American Type Culture Collection) was obtained from P.G. Thomas. Virus was added to unprimed BMDMs at an MOI of 10, followed by 10 h of incubation, for activation of the AIM2 inflammasome. For activation of the canonical NLRP3 inflammasome, BMDMs were primed for 4 h with 500 ng/ml ultrapure LPS from Salmonella minnesota R595 (InvivoGen) and were stimulated for 45 min with 5 mM ATP or 10 µM nigericin (Sigma). For transfection of DNA, each reaction consisted of 2.5 µg of poly(dA:dT) (InvivoGen) resuspended in PBS and mixed with 0.6 µl of Xfect polymer in Xfect reaction buffer (Clontech Laboratories). After 10 min, DNA complexes were added to BMDMs in Opti-MEM (Gibco), followed by incubation for 5 h. For CFU analysis, supernatants from BMDMs infected for 3 or 7 h with F. novicida were replaced with medium containing 50 µg/ml gentamicin (Gibco). Cells were incubated for an additional 1 h, then were washed twice with PBS and scraped before being plated onto Trypticase Soy Broth agar. For 24 h of infection, medium containing 50 µg/ml gentamicin (1 h incubation) was replaced with medium containing 10 µg/ml gentamicin and cells were further incubated for 16 h. Levels of lactate dehydrogenase released by cells were measured with a CytoTox 96 Non-Radioactive Cytotoxicity Assay according to the manufacturer’s instructions (Promega). Cell culture supernatants were collected for enzyme-linked immunosorbent assay. Knockdown via small interfering RNA. BMDMs were transfected for 48 h with small interfering RNA (siRNA) from siGENOME smart pools (Dharmacon) with the GenMute siRNA Transfection Reagent according to the manufacturer’s instructions (Fisher Scientific). siGENOME SMARTpool siRNA specific for mouse Gbp2 (M-040199-00-0005), Gbp3 (M-063076-010005), Gbp5 (M-054703-01-0005), Gbp7 (M-061204-01-0005) were used in the study (sequences, Supplementary Table 1). A control siRNA pool was used. Transfected cells were infected with F. novicida as described above. Immunoblot analysis. For immunoblot analysis of caspase-1, BMDMs and supernatants were lysed in RIPA lysis buffer (1% vol/vol NP-40, 12.7 mM sodium deoxycholate, 0.1% wt/vol SDS, 150 mM sodium chloride, 3.3% vol/vol 1.5 M Tris (pH8.8)) and sample loading buffer containing SDS and 100 mM DTT.

nature immunology

For immunoblot analysis of signaling via IRF1 or GBP, supernatants were removed and BMDMs were washed once with PBS, followed by lysis in RIPA buffer and sample loading buffer containing SDS and 100 mM DTT. Proteins were separated by electrophoresis through 8–12% polyacrylamide gels. Following electrophoretic transfer of proteins onto PVDF membranes, nonspecific binding was blocked by incubation with 5% skim milk, then membranes were incubated with primary antibodies (antibody to caspase-1 (anticaspase-1) (1:3,000 dilution; Casper-1; AG-20B-0042; Adipogen), anti-IRF1 (1:1,000 dilution; D5E4; 8478; Cell Signaling Technologies), anti-GBP2 (1:1,000 dilution; 11854-1-AP; Proteintech), anti-GBP5 (1:1,000 dilution; 13220-1-AP; Proteintech) or anti-GAPDH (1:10,000 dilution; D16H11; 5174, Cell Signaling Technologies), followed by incubation with secondary antibodies (horseradish peroxidase–conjugated anti-rabbit (111-035-047) or anti-mouse (315-035-047) (1:5,000 dilution; Jackson Immuno Research Laboratories). Immunofluorescence staining. F. novicida strain U112 or a strain expressing GFP was used for infection. For visualization of inflammasomes, BMDMs were infected for 20 h and were further incubated for 1 h at 37 °C in fresh medium containing 1× FAM (5-carboxyfluorescein) FLICA (fluorochrome inhibitor of caspase) active caspase-1 (ImmunoChemistry Technologies). Cells were washed three times with PBS and were fixed for 15 min at room temperature in 4% paraformaldehyde, followed by blockade of nonspecific binding by incubation for 1 h in 10% normal goat serum (Dako) supplemented with 0.1% saponin (Sigma). Cells were incubated for 50 min at 37 °C with a rabbit anti-ASC (1:500 dilution; AL177; AG-25B-0006-C100; AdipoGen). For staining of GBP5, BMDMs were infected for the appropriate time and were washed three times with PBS. Cells were fixed and blocked as described above and were stained overnight at 4 °C with rabbit anti-GBP5 (1:500 dilution; 13220-1-AP; Proteintech). The secondary antibody used was Alexa Fluor 568–conjugated antibody to rabbit immunoglobulin G (1:250 dilution; A11036; Life Technologies). Cells were counterstained in DAPI mounting medium (Vecta Labs). Bacteria, inflammasomes and BMDMs were visualized, counted and imaged with a Nikon C2 confocal microscope. Real time quantitative RT-PCR analysis. RNA was extracted with TRIzol according to the manufacturer’s instructions (Life Technologies). The RNA isolated was reverse-transcribed into cDNA with a First-Strand cDNA Synthesis Kit (Life Technologies). Real-time quantitative PCR was performed on an ABI 7500 real-time PCR instrument with 2× SYBR Green (Applied Biosystems) and the appropriate primers (sequences, Supplementary Table 2). Cytokine analysis. Cytokines were measured by multiplex enzyme-linked immunosorbent assay (Millipore) or enzyme-linked immunosorbent assay for IL-18 (MBL international) according to the manufacturer’s instructions. Microarray. Transcripts were profiled for F. novicida–infected unprimed BMDMs obtained from wild-type and mutant mice. Total RNA (100 ng) was converted into biotin-labeled cRNA with an Ambion Wild-Type Expression kit (Life Technologies) and was hybridized to a Mouse Gene 2.0 ST GeneChip (Affymetrix). After staining and washing, array signals were normalized and transformed into log2 transcript expression values with the robust multi-array average algorithm (Partek Genomics Suite version 6.6)49. Differential expression was defined by application of a difference in expression of 0.5-fold (log2 signal) between conditions. Lists of differentially expressed transcripts were analyzed for ‘functional enrichment’ with the DAVID bioinformatics database50 and Ingenuity Pathways Analysis software. Animal infection. F. novicida strain U112 was grown as described above. For analyses of survival and weight change, mice were given subcutaneous injection of 7.5 × 104 CFU of F. novicida in 200 µl PBS. For CFU analysis, mice were given subcutaneous injection of 1 × 105 CFU of F. novicida in 200 µl PBS. After 2 d, liver and spleen were collected and then were homogenized for 2 min in PBS with metal beads with a TissueLyser II apparatus (Qiagen). CFU were quantified by plating of lysates onto Trypticase Soy Broth agar supplemented with 0.2% l-cysteine, followed by incubation overnight. No randomization or blinding was used. Histopathology. Formalin-preserved livers were processed and embedded in paraffin according to standard procedures. Sections 5 µm in thickness

doi:10.1038/ni.3118

were stained with hematoxylin and eosin and were examined by a pathologist ‘blinded’ to experimental sample identity. For immunohistochemistry, formalin-fixed, paraffin-embedded livers were cut into sections 4 µm in thickness. For staining of granulocytes, samples were incubated for 30 min with anti-MPO (1:500 dilution; A0398; Dako) followed by incubated for 30 min with Rabbit-on-Rodent HRP-Polymer (RMR622L; BioCare Medical). TUNEL (terminal deoxynucleotidyl transferase deoxyuridine triphosphate nick end labeling) staining was performed with the Dead End kit (PRG7130) according to the manufacturer’s instructions (Promega). Statistical analysis. GraphPad Prism 6.0 software was used for data analysis. Statistical significance was determined by t-tests (two-tailed) for two groups or one-way ANOVA (with Dunnett’s or Tukey’s multiple comparisons tests) for three or more groups. Survival curves were compared with the log-rank test. P values of