NLRC4 inflammasomes in dendritic cells regulate ... - Nature

6 downloads 0 Views 1MB Size Report
Jan 8, 2012 - form multimeric complexes known as inflammasomes, which ... an in vivo link between flagellin-induced NLRC4 inflammasome activation in ...
ARTICLES

NLRC4 inflammasomes in dendritic cells regulate noncognate effector function by memory CD8+ T cells

npg

© 2012 Nature America, Inc. All rights reserved.

Andreas Kupz1, Greta Guarda2, Thomas Gebhardt1, Leif E Sander3, Kirsty R Short1, Dimitri A Diavatopoulos1,6, Odilia L C Wijburg1, Hanwei Cao1, Jason C Waithman4, Weisan Chen4, Daniel Fernandez-Ruiz1, Paul G Whitney1, William R Heath1, Roy Curtiss III5, Jürg Tschopp2,7, Richard A Strugnell1,8 & Sammy Bedoui1,8 Memory T cells exert antigen-independent effector functions, but how these responses are regulated is unclear. We discovered an in vivo link between flagellin-induced NLRC4 inflammasome activation in splenic dendritic cells (DCs) and host protective interferon-g (IFN-g) secretion by noncognate memory CD8+ T cells, which could be activated by Salmonella enterica serovar Typhimurium, Yersinia pseudotuberculosis and Pseudomonas aeruginosa. We show that CD8a+ DCs were particularly efficient at sensing bacterial flagellin through NLRC4 inflammasomes. Although this activation released interleukin 18 (IL-18) and IL-1b, only IL-18 was required for IFN-g production by memory CD8+ T cells. Conversely, only the release of IL-1b, but not IL-18, depended on priming signals mediated by Toll-like receptors. These findings provide a comprehensive mechanistic framework for the regulation of noncognate memory T cell responses during bacterial immunity. CD8+ T cells typically gain effector function in response to T cell antigen receptor (TCR)-mediated recognition of cognate antigen. After elimination of antigen, effector T cells contract to a small but stable population of long-lived memory cells. Memory CD8+ T cells potently and rapidly respond to rechallenges with cognate antigen 1. However, some lines of evidence suggest that memory CD8+ T cells also exert functions that operate independently of TCR engagement. Proliferation and IFN-γ secretion by memory CD8+ T cells2 have been reported in response to the synthetic RNA duplex poly(I:C) or lipopolysaccharide (LPS)3 and noncognate aspects of memory T cell responses contribute to immunity against Listeria monocytogenes2 and influenza virus infections4. However, little is known about the mechanisms that regulate these responses. For example, it is unclear which microbial components the cells respond to, which host signals drive their effector function and whether these responses are regulated at a T cell–intrinsic level or rather involve interactions with additional cell types. The observation that antigen-independent responses by memory CD8 + T cells can be elicited with LPS suggests a role for pattern recognition receptors such as Toll-like receptors (TLRs) 5. Other more recently discovered pattern recognition receptors include the group of NOD-like receptors (NLRs)6. These cytosolic receptors can associate with proteins and nonprotein ligands to form multimeric complexes known as inflammasomes, which through caspase-1 activation catalyze the processing of members of the IL-1 family into their active forms 6,7. TLR stimulation 8 and

NLR-mediated signals affect the magnitude and duration of tumor-specific CD8 + T cell responses 9. However, it is unknown whether inflammasomes are also involved in regulating antigenindependent T cell functions. Inflammasome activation has been largely studied in vitro with bone marrow–derived DCs, and little is known about whether DCs activate inflammasomes in vivo. The spleen contains CD8α+ DCs, CD4+ DCs and a heterogeneous group of DCs that express neither CD8α nor CD4 (double-negative DCs; DN DCs). Whereas CD8α+ DCs efficiently cross-present exogenous antigen on major histocompatibility (MHC) class I molecules10, the other subsets excel at presenting exogenous antigen on MHC class II molecules. Considering that bone marrow–derived DCs are representative of inflammatory DCs, which are absent from lymphoid organs at steady state11, it is unclear whether the various types of DCs found at steady state12 can also activate inflammasomes. Similarly, our understanding of how inflammasomes are regulated is mostly derived from in vitro experiments. Inflammasome activation is regulated by at least two distinct signals. A first ‘priming signal,’ which is typically mediated by TLR stimulation, is thought to induce transcription of the inactive cytokine precursor. In the case of the NLRP3 inflammasome, this signal also induces expression of NLRP3 itself. NLRC4 is constitutively expressed13 and therefore does not require earlier induction through priming signals. A ‘second signal’ then activates the inflammasome in a ligand-like way. Although exceptions to this model have been noted in vitro in human blood monocytes, in which single agonists could

1Department of Microbiology and Immunology, The University of Melbourne, Parkville, Victoria, Australia. 2Department of Biochemistry, University of Lausanne, Epalinges, Switzerland. 3Department of Infectious Diseases and Pulmonary Medicine, Charité University Hospital, Berlin, Germany. 4Ludwig Institute of Cancer Research, Heidelberg, Victoria, Australia. 5Center of Infectious Diseases, Arizona State University, Tempe, Arizona, USA. 6Present address: Department of Pediatrics, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands. 7Deceased. 8These authors contributed equally to this work. Correspondence should be addressed to S.B. ([email protected]) or R.A.S. ([email protected]).

Received 21 October 2011; accepted 28 November 2011; published online 8 January 2012; doi:10.1038/ni.2195

162

VOLUME 13

NUMBER 2

FEBRUARY 2012

NATURE IMMUNOLOGY

ARTICLES S. Typhimurium (Fig. 1e). Thus, noncognate memory CD8+ T cells secreted IFN-γ in response to S. Typhimurium. To assess whether these responses had a role in the control of S. Typhimurium infections, we transferred in vitro–activated effector gBT-I cells into Ifng−/− mice that are highly susceptible to S. Typhimurium infection17. As these effector gBT-I cells convert into CD62L hi memory T cells within weeks after transfer18, IFN-γ could be provided only by noncognate memory CD8+ T cells under these conditions. Notably, Ifng−/− mice that received memory gBT-I cells survived intravenous infection with S. Typhimurium BRD509 significantly longer than control Ifng−/− mice did (Fig. 1f), with median survival times of 31 and 24 d, respectively (P < 0.0001). The improved survival was accompanied by lower bacterial burden in Ifng−/− mice bearing memory gBT-I cells (Fig. 1g). Memory gBT-I cells also conferred a significant survival advantage to Ifng−/− mice infected orally with S. Typhimurium BRD509 (P = 0.0038; Supplementary Fig. 1a), and we observed an even more pronounced protective effect when lymphocyte-deficient mice (Rag2−/− Il2rg−/− mice) bearing IFN-γ-competent memory gBT-I cells were intravenously infected with this strain of S. Typhimurium (Supplementary Fig. 1b; P < 0.0001). We used the attenuated aroA- and aroD-deficient S. Typhimurium strain BRD509 for the experiments described above, as infections with this slower growing S. Typhimurium strain mimic more closely the natural course of typhoid fever in humans. We also assessed the role of noncognate memory CD8+ T cells during infections with the wild-type strain of S. Typhimurium SL1344, which, in contrast to the prolonged, less severe course of BRD509 infection, leads to a rapidly lethal disease. Despite the fulminant nature of wild-type infections (control Ifng−/− mice succumbed to the infection within 4 d), the presence of memory gBT-I cells provided a median survival advantage of 1 d (Supplementary Fig. 1c; P = 0.0028) to

RESULTS Noncognate memory CD8+ T cells respond to S. Typhimurium IFN-γ is critical for the control of Salmonella species infections in mice and humans16,17. Noncognate memory CD8+ T cells act as potent sources of IFN-γ during L. monocytogenes infections2, therefore we questioned whether memory CD8+ T cells also secrete IFN-γ in response to S. Typhimurium and whether such responses have a role in host defense against S. Typhimurium infection. We initially analyzed splenic CD8+ T cells for their capacity to secrete IFN-γ in response to S. Typhimurium with an antibody that detects IFN-γ as it is being released from the cell. Although mice treated with vehicle did not secrete IFN-γ, ~10% of all CD8+ T cells secreted IFN-γ 2 h after intravenous injection of S. Typhimurium (Fig. 1a,b). These responses were restricted to CD8+ T cells with a memory phenotype (positive for the activation and memory marker CD44, the chemokine receptor CXCR3 and lymph node–homing receptor CD62L; Fig. 1a,c,d) and did not require cognate TCR signals, as transgenic MHC class I–restricted memory CD8+ T cells specific for the immunodominant peptide of herpes simplex virus type 1 glycoprotein B (gBT-I) also secreted IFN-γ in response to

b

20

c

*

0.2

10

5

3

IFN-γ

10

10

80

0 102 103 104 105

40

10

20

10

2

3

0 10 10 CXCR3

0 PBS

CD44

105

3

0

2

10 0

28.3

4

10

60

IFN-γ

10

4

STM

0.3 PBS

5

Events (% max)

14.4

IFN-γ+ (%)

5

Gated on CD8+ CD44hi

100

15

Gated on + CD45.1 gBT-I

e

CD44hi IFN-γ+

PBS

10

d

lo – CD44 IFN-γ

4

10

5

10

2

STM

0 0

2

10

3

10

4

10

IFN-γ

Gated on CD3+CD8+

a

npg

5

10

CD62L

10

4

10

3

98.1

2

10 0

2

0 10

STM

3

4

5

10 10 10

CD44

f 100

g

9

Survival (%)

Figure 1 Noncognate memory CD8+ T cells secrete 80 IFN-γ in response to S. Typhimurium and contribute to the * * control of S. Typhimurium infections. (a,b) Flow cytometry 8 60 and frequency of IFN-γ+ cells among total CD3+CD8+ T cells in spleens of naive B6 mice intravenously injected 40 with PBS or 1 × 108 colony-forming units (CFU) of 7 S. Typhimurium (STM); IFN-γ secretion by CD8+ T cells –/– 20 Ifng + memory gBT-I was assessed 2 h later. Numbers adjacent to outlined –/– Ifng + PBS areas indicate percent cells in each throughout. 6 0 (c,d) Expression of CXCR3 by CD44hiIFN-γ+ CD8+ –/– 5 10 15 20 25 30 35 Ifng–/– Ifng T cells (black line) and CD44loIFN-γ− CD8+ T cells Time after infection (d) + PBS + memory gBT-I + + (grey line) (c) and frequency of CD62L and IFN-γ cells among CD3+CD8+CD44hi T cells (d) 2 h after injection of S. Typhimurium. (e) Flow cytometry and frequency of CD45.1+CD3+CD8+ gBT-I cells in spleens of B6 mice given transfer of activated gBT-I cells and, 6 weeks after transfer (when gBT-I cells converted into CD62Lhi memory CD8+ T cells), given intravenous injection of PBS or S. Typhimurium; IFN-γ secretion by CD45.1+CD3+CD8+ gBT-I was assessed 2 h later. (f,g) Survival (f) and bacterial counts on day 23 after infection in the spleen (g) of Ifng−/− mice given transfer of activated gBT-I cells (n = 15) or treated with PBS (n = 15), 6 weeks after transfer (when gBT-I cells converted into CD62Lhi memory CD8+ T cells), given intravenous injection of 2 × 102 CFU S. Typhimurium BRD509. *P < 0.0001 (paired Student’s t-test (b) or log-rank (Mantel-Cox) test (g)). Data are representative of three (a,b,e), two (c,d) or at least three (f,g) experiments.

NATURE IMMUNOLOGY

VOLUME 13

NUMBER 2

FEBRUARY 2012

Bacteria (CFU log10)

© 2012 Nature America, Inc. All rights reserved.

also elicit IL-1β release14,15, it remains unclear how these different signals contribute to inflammasome activation in vivo. Building on our observation that noncognate memory T cell responses have an important protective role during infections with the intracellular pathogen Salmonella enterica serovar Typhimurium (S. Typhimurium), we sought to delineate the regulation of noncognate memory CD8+ T cell responses in vivo. We show that a complex mechanism involving bacterial flagellin, inflammasome activation and cytokine production by DCs is responsible for the regulation of protective noncognate memory CD8+ T cell responses in vivo .

163

ARTICLES

IFN-g secretion by memory CD8+ T cells requires flagella Having established a role for noncognate memory CD8+ T cell responses during S. Typhimurium infection, we aimed to delineate the underlying molecular and cellular mechanisms that control these responses in vivo. We first focused on the microbial stimuli that drive these responses in various S. Typhimurium mutants. As LPS has been suggested to elicit IFN-γ secretion by effector CD8+ T cells in vitro3,19, we injected ΔmsbB S. Typhimurium , which produces a modified form of LPS with impaired stimulatory activity at TLR4 owing to a deletion-insertion mutation in the gene encoding the enzyme that catalyzes complete lipid A biosynthesis20. Notably, these mutant S. Typhimurium induced IFN-γ secretion in memory CD8+ T cells as potently as wild-type S. Typhimurium did (Fig. 2a). In a complementary approach, we tested whether wild-type S. Typhimurium could induce IFN-γ secretion by memory CD8+ T cells without TLR4. Consistent with the findings obtained above with ΔmsbB S. Typhimurium, we observed that memory CD8+ T cells still secreted IFN-γ in Tlr40/0 mice (a congenic strain with a naturally occurring mutation in Tlr4) in response to wild-type S. Typhimurium (Supplementary Fig. 2), albeit in slightly smaller amounts than those of control mice. These findings indicate that LPS-TLR4 interactions were not the dominant factor required for IFN-γ secretion in response to S. Typhimurium. Flagellin, the major structural subunit of bacterial flagella, not only has high expression in S. Typhimurium but also acts as a potent pathogenassociated molecular pattern21,22. To assess whether flagella contributed to IFN-γ secretion by memory CD8+ T cells, we exposed mice to S. Typhimurium mutants (ΔfliC ΔfljB) lacking flagella. ΔfliC ΔfljB S. Typhimurium did not elicit IFN-γ secretion by memory CD8+ T cells (Fig. 2a) but did revert to do so when mutants were complemented with a plasmid encoding flagellin (ΔfliC ΔfljB + pLS408). These findings indicate that flagellin was the dominant factor in S. Typhimurium responsible for eliciting IFN-γ secretion. S. Typhimurium–induced activation of bone marrow–derived DCs requires bacteria to inject flagellin into the cytosol through their type III secretion system21.

10 5 0

c

20

25 20

15

15

+

15

b

IFN-γ (%)

*

+

20

IFN-γ (%)

+

IFN-γ (%)

a

Δ Δ ms Δf fliC bB liC Δ f pL Δflj ljB S4 B + ss 08 aR ::T Δin n vA ph oA ST M

npg

© 2012 Nature America, Inc. All rights reserved.

Ifng−/− mice infected intravenously with wild-type S. Typhimurium. These findings establish that noncognate memory CD8+ T cells carried out protective roles during infections with both growthattenuated and wild-type S. Typhimurium strains. Therefore, we identified a protective role for noncognate CD62Lhi memory CD8+ T cells during S. Typhimurium infections.

10 5

10 5

0

0 0

0.25 2.5

25 250

Flagellin (mg/mouse)

Ultrapure flagellin

Figure 2 Bacterial flagellin is required for S. Typhimurium–induced IFN-γ secretion by memory T cells. (a) Frequency of IFN-γ+ cells among total splenic CD3+CD8+ T cells 2 h after injection of B6 mice with S. Typhimurium mutants or wild-type S. Typhimurium. (b,c) IFN-γ secretion by CD3+CD8+ T cells in the spleens of naive B6 mice at 2 h after intravenous injection of various doses of flagellin purified from S. Typhimurium (b) or commercially available ultrapure flagellin (10 μg per mouse; c). *P < 0.0001 (One-way analysis of variance (ANOVA) followed by Bonferroni multiple comparison test). Data are representative of three (a) or at least three (b,c) experiments.

164

To test whether a similar mechanism operated in our system, we injected ΔinvA and ssaR::Tn phoA S. Typhimurium mutants, which have mutations in genes encoding the two type III secretion systems (SPI-1 and SPI-2, respectively) and cannot inject effector molecules into host cells23. Notably, at this early time point, these mutants induced IFN-γ secretion by memory CD8+ T cells as potently as their wildtype counterparts did (Fig. 2a). IFN-γ secretion was induced with similar efficiency when we injected heat-killed instead of live S. Typhimurium (data not shown), which indicated that bacterial viability was not essential for the induction of IFN-γ secretion by memory CD8+ T cells in vivo. In-house purified (Fig. 2b) or commercially available ultrapure flagellin (Fig. 2c) also potently elicited IFN-γ secretion by memory CD8+ T cells 2 h after intravenous injection. Therefore, these results suggest that attached, polymeric flagella (present in live bacteria) and monomeric flagellin (liberated through heat inactivation and present in purified preparations) were equally potent in inducing IFN-γ secretion by memory CD8+ T cells. Taken together, these findings show that flagellin was the dominant factor in S. Typhimurium driving IFN-γ secretion by memory CD8+ T cells in vivo. IFN-g secretion requires NLRC4 inflammasomes and IL-18 Antigen-independent IFN-γ secretion by effector CD8+ T cells is thought to involve IL-12 and IL-18 in vitro24,25. To determine the involvement of these cytokines in noncognate IFN-γ responses by memory CD8+ T cells in vivo, we re-evaluated this effect in IL-12p40deficient (Il12b−/−) and Il18−/− mice. We continued to use heat-killed S. Typhimurium) (HKST), as it elicited IFN-γ secretion as efficiently as viable bacteria did and allowed us to exclude the possibility of dose variations owing to further bacterial replication after injection. In contrast to the IFN-γ secretion in C57BL/6 (B6) mice challenged with HKST, Il18−/− mice did not secrete IFN-γ, whereas Il12−/− mice had IFN-γ secretion similar to that of challenged B6 mice (Fig. 3a). Conversely, injection of recombinant IL-18 induced the dose-dependent release of IFN-γ from memory CD8+ T cells in wild-type mice, indicating that IL-18 not only was required for S. Typhimurium– induced IFN-γ secretion, but also was sufficient to elicit IFN-γ release by itself (Fig. 3b). Although IL-18 shares many functional features with IL-1β (ref. 26), we found that IL-1β was not required for innate IFN-γ secretion (Fig. 3a). Thus, IL-18, but not IL-1β or IL-12, was a critical host element in the induction of IFN-γ secretion by memory CD8+ T cells in vivo. The findings reported above suggested that memory CD8+ T cells responded directly to IL-18. We thus compared IL-18 receptor (IL-18R) expression on CD44hiIFN-γ+ and CD44loIFN-γ− CD8+ T cells and found higher expression of IL-18R on CD44hiIFN-γ+ CD8+ T cells than on CD44loIFN-γ− CD8+ T cells (Fig. 3c). Consistent with the finding that innate IFN-γ secretion was a specific function of CD62Lhi memory CD8+ T cells, IL-18R was coexpressed with CD62L and CXCR3 on CD44hi memory CD8+ T cells in naive B6 mice (Supplementary Fig. 3). To assess whether memory CD8+ T cells responded directly to IL-18, we used the fact that IL-18R signaling requires the adaptor MyD88 (ref. 26) and found that IFN-γ secretion was completely inhibited in Myd88−/− mice (Fig. 3d). However, because MyD88 is also involved in the signaling of many TLRs5, it was necessary to delineate whether MyD88 was required in T cells or other host cells involved in the process. As above (Fig. 1e,f), we transferred wild-type gBT-I cells into Myd88−/− mice. Hence, only the donor memory T cells (gBT-I cells) were responsive to IL-18 and IL-18R signals (MyD88 competent). Both memory gBT-I cells (CD45.1+) and endogenous memory T cells (CD45.1−) responded to HKST in

VOLUME 13

NUMBER 2

FEBRUARY 2012

NATURE IMMUNOLOGY

ARTICLES

3

4

5

mAb αIL-18

0

4

0.7

2.2

0.2

10

3

10 2 10 0

2.9

* * *

15 10 5 0

2

3

4

5

0 10 10 10 10

CD45.1

8 IL-18 (ng/ml)

10

IFN-γ+ (%)

5

0 10 10 10 10

IL-18R

20

10 5

g

f

11.7

* * *

6 4 2 0

C as p N 1 −/− lrp N 3 −/− Py lrc4 − ca /− rd − /− B6

2

0.6

C as p N 1 −/− lrp N 3 −/− l Py rc4 − ca /− rd − /− B6

20 0

11.2

PBS

IFN-γ



B6



−/



−/

−/

Il1 2 Il1 b Il1 8

0.1 1 5 10 IL-18 (mg/mouse)

40

Myd88−/−

B6

15

B6

0

0

60



5

5

100 80

−/

10

10

*

20

88

15

e 8.4

CD44lo IFN-γ− CD44hi IFN-γ+

15 IFN-γ+ (%)

IFN-γ+ (%)

20

d

+ IFN-γ (%)

*

25

c

M yd

b

Events (% max)

a

npg

© 2012 Nature America, Inc. All rights reserved.

Figure 3 T cell–intrinsic IL-18R-MyD88 signaling and NLRC4 inflammasomes are required for IFN-γ production by memory CD8+ T cells. (a) Frequency of IFN-γ+ cells among total CD3+CD8+ T cells of mice lacking cytokines and control B6 mice given injection 2 h earlier with HKST. (b) Frequency of IFN-γ+ cells among total CD3+CD8+ T cells of B6 mice given injection of recombinant IL-18 2 h earlier. (c) IL-18R expression in splenic CD44hi IFN-γ+ CD8+ T cells (black line) and CD44lo IFN-γ− CD8+ T cells (grey line) 2 h after injection of HKST into B6 mice. (d) Frequency of IFN-γ+ cells among total CD3+CD8+ T cells 2 h after injection of Myd88−/− mice or B6 mice with HKST. (e) Flow cytometry and frequency of IFN-γ+CD45.2+ endogenous memory T cells and CD45.1+ transgenic memory gBT-I cells among total CD3+CD8+ T cells 2 h after injection of HKST into mice pretreated with PBS or monoclonal antibody to IL-18. (f,g) Frequency of IFN-γ+ cells among total CD3+ CD8+ T cells (f) and serum IL-18 concentrations (g) of mice lacking inflammasome components and B6 mice injected 2 h earlier with HKST. *P < 0.0001 (one-way ANOVA followed by Bonferroni multiple-comparison test (a,f,g) and paired Student’s t-test (d)). Data are representative of three (a,d,f,g) or two (b,c,e) experiments.

wild-type mice (Fig. 3e and Supplementary Fig. 4a,b). Notably, although the IFN-γ response by endogenous memory CD8+ T cells in Myd88−/− hosts was again abrogated, MyD88-competent memory gBT-I cells responded to HKST when MyD88 was absent from all other cells (Fig. 3e and Supplementary Fig. 4a,b). Those findings were further substantiated by the observations that IFN-γ responses by memory gBT-I cells in Myd88−/− mice and B6 were inhibited by treatment with antibody to IL-18 (Fig. 3e and Supplementary Fig. 4a,b) and that memory CD8+ T cells did not secrete IFN-γ in response to HKST in Il18r1−/− mice (Supplementary Fig. 4c). Collectively, these data show that noncognate IFN-γ secretion by memory CD8+ T cells involved IL-18R and MyD88 signaling, but did not require T cell-extrinsic, MyD88-dependent (that is, TLR-mediated) signals. Having identified flagellin and IL-18 as critical mediators, we next investigated the molecular mechanisms that controlled IL-18 production in response to S. Typhimurium. IL-18 requires enzymatic activation by specific enzymes, such as caspase-1 (ref. 26). To test whether caspase-1 was involved in IFN-γ secretion, we injected Casp1−/− mice with HKST and assessed IFN-γ secretion by memory CD8+ T cells 2 h later. Casp1−/− mice showed no IFN-γ secretion (Fig. 3f), which replicated the results obtained with Il18−/− mice. Activation of caspase-1 requires autocatalytic cleavage, which typically occurs in multimeric protein complexes called inflammasomes. Among the different inflammasomes described6, we were interested in the NLRC4 inflammasome, comprising NLRC4, the adaptor PYCARD and caspase-1, as it is critical for sensing flagellin, along with neuronal apoptosis inhibitor protein 5 (NAIP5; refs. 21,27,28). When we injected mice deficient for NLRC4, NLRP3 or PYCARD (also known as ASC) with HKST and assayed IFN-γ release by memory CD8+ T cells 2 h later, we found that IFN-γ secretion was completely inhibited without NLRC4, significantly inhibited in mice lacking the adapter Pycard (Pycard−/−), but not different from controls in Nlrp3−/− mice (Fig. 3f). Further highlighting the critical role of IL-18 in these responses (Fig. 3a,b), higher IL-18 serum concentrations induced by HKST similarly required caspase-1, NLRC4 and PYCARD (Fig. 3g). Given that Casp1−/−, Nlrc4−/− and Pycard−/− mice contained memory CD8+ T cells (Supplementary Fig. 5), these findings show that S. Typhimurium–induced, IL-18-mediated IFN-γ secretion by memory CD8+ T cells required NLRC4 and caspase-1 in vivo, whereas NLRP3, IL-1β and IL-12 were dispensable. These observations establish a direct functional link between NLRC4 inflammasomes and IFN-γ secretion by T cells.

NATURE IMMUNOLOGY

VOLUME 13

NUMBER 2

FEBRUARY 2012

In addition to activating NLRC4, flagellin can also be sensed through TLR5. To address potential involvement of TLR5, we injected Tlr5−/− mice with HKST and assayed IFN-γ secretion 2 h later. Although IFN-γ secretion by memory CD8+ T cells was somewhat lower in Tlr5−/− mice than in wild type mice (Supplementary Fig. 6a), we found it difficult to discern whether this somewhat blunted response was due to a functional contribution of TLR5 or was the result of the severe constitutive proinflammatory conditions prevalent in mice lacking TLR5 (ref. 29). To address the question of involvement of TLR5 from a different angle, we used the observation30 that alanine substitutions at amino acid positions 415 or 89 of flagellin abrogated the stimulatory activity of flagellin on TLR5. Reconstitution of ΔfliC ΔfljB S. Typhimurium (as used in Fig. 2) with either TLR5-incompetent flagellin (ΔfliC ΔfljB + pLS408-L415A; ΔfliC ΔfljB + pLS408-Q89A) or wild-type flagellin (ΔfliC ΔfljB + pLS408-WT) elicited IFN-γ secretion by memory CD8+ T cells with similar efficiency (Supplementary Fig. 6b). Together with our observation that T cell–extrinsic MyD88 was not required for IFN-γ secretion in response to S. Typhimurium (Fig. 3d,e), we conclude that the TLR5-MyD88 pathway did not have a critical role in this type of innate IFN-γ production by memory CD8+ T cells. Yersinia and Pseudomonas also elicit innate IFN-g secretion Our observations so far had shown that S. Typhimurium–derived flagellin elicited secretion of IFN-γ by memory CD8+ T cells through NLRC4 inflammasome–mediated release of IL-18. To assess whether this mechanism is unique to S. Typhimurium or is a more widely used host defense strategy, we examined whether other flagellated bacteria also elicited IFN-γ secretion by memory CD8+ T cells. Heat-killed flagellated Escherichia coli and Helicobacter pylori did not elicit IFN-γ secretion, however heat-killed Y. pseudotuberculosis and P. aeruginosa induced the release of IFN-γ by memory CD8+ T cells 2 h after intravenous injection (Fig. 4a). Analogous to our observations with S. Typhimurium, IFN-γ secretion by memory CD8+ T cells in response to Y. pseudotuberculosis or P. aeruginosa depended on the expression of flagella by the bacteria (Fig. 4a), involved increases in serum IL-18 concentrations (Fig. 4b) and required caspase-1 in the host (Fig. 4c). Thus, the inflammasomemediated induction of IFN-γ secretion by memory CD8+ T cells not only occurred in response to S. Typhimurium, but seemed to be a more broadly used host defense strategy against some types of flagellated bacteria.

165

ARTICLES Figure 4 NLRC4 inflammasomes are HKST HKYP HKPA 15 4 ** required for IFN-γ secretion and IL-18 ** 5.3 6.7 3.1 release in response to S. Typhimurium, 3 B6 10 Y. pseudotuberculosis and P. aeruginosa. * 2 (a) Frequency of IFN-γ+ cells among total * 5 + + CD3 CD8 T cells of B6 mice given injection 1 105 8 0.4 0.3 0.3 2 h earlier of 1 × 10 CFU HKST, heat-killed 104 0 0 −/− Casp1 103 E. coli, H. pylori or P. aeruginosa (HKEC, 102 HKHP and HKPA, respectively), flagellin0 deficient P. aeruginosa (fliC− HKPA), heat-killed 0 102 103 104 105 CD44 Y. pseudotuberculosis (HKYP) or heat-killed Y. pseudotuberculosis not expressing flagellin (flagellin− HKYP; details in Online Methods). (b) Serum IL-18 concentrations in B6 mice given injection 2 h earlier of PBS, HKPA or HKYP. (c) Flow cytometry and frequency of IFN-γ+ cells among CD3+CD8+ T cells in B6 and Casp1−/− mice 2 h after injection of heat-killed bacteria. **P < 0.0001 and *P < 0.05 (one-way ANOVA followed by Bonferroni multiple-comparison test). Data are representative of three experiments.

b

IFN-γ

P

A

H KY

S

H KP

PB

IFN-γ+ (%)

H KS H T KE H C KH P Fl HK iC − P A H Fl KP ag el HK A lin − Y P H KY P

© 2012 Nature America, Inc. All rights reserved.

npg

DC provide IL-18 after in vivo inflammasome activation Given the findings presented above, we investigated which cell type activated the NLRC4 inflammasome and provided IL-18 in response to flagellin. As IL-18 is secreted by DCs, and in vitro studies show NLRC4 inflammasome activation in bone marrow–derived DCs21, we speculated that DCs drove the observed T cell response. To test this hypothesis in vivo, we reconstituted lethally irradiated mice with a 1:1 mix of Il18−/− bone marrow and bone marrow from transgenic mice expressing diphtheria toxin receptor (DTR) under the control of the promoter of the gene encoding CD11c, which can be monitored by expression of green fluorescent protein (GFP)31. These mixed chimeras now contained two types of DCs, IL-18-competent DCs (GFP+) that were susceptible to diphtheria toxin (DTX) depletion owing to their transgenic expression of DTR; and IL-18-deficient DCs (GFP−) derived from the Il18−/− bone marrow, which did not express DTR. PBS-treated mixed chimeras contained both GFP+ (IL-18-competent) DCs and GFP− (IL-18-deficient) DCs, whereas DTX-treated chimeras had only GFP− (IL-18-deficient) DCs (Fig. 5a). When we challenged DTX-treated and control mice with HKST, we observed IFN-γ secretion (Fig. 5b) and increases in serum IL-18 concentrations (Fig. 5c) only in test chimeras treated with PBS (Il18−/− + CD11c-DTR > B6 + PBS) or control chimeras treated with DTX (CD45.1 + CD11c-DTR > B6 + DTX), and not in test chimeras treated with DTX (Il18−/− + CD11c-DTR > B6 + DTX). These observations are consistent with the view that DC-derived IL-18 is required for IFN-γ secretion by memory T cells in vivo, although we could not exclude a contribution by other CD11cexpressing cells, such as macrophages. In support of the possibility that DCs were involved, DCs isolated from HKST-exposed B6 mice contained activated caspase-1, as indicated by the presence of its cleavage fragment p10 (Fig. 5d). In further support of the proposed importance of NLRC4 in S. Typhimurium–induced inflammasome activation, caspase-1 activation was much lower in Nlrc4−/− mice than in control mice and did not require NLRP3 (Fig. 5d). Together these findings demonstrated in vivo NLRC4 inflammasome activation in splenic DC. Splenic DCs comprise several functionally distinct subsets12; therefore, we wanted to determine whether all DC subsets activated NLRC4 inflammasomes in response to flagellin or whether this was specific to a certain subset. We addressed this question first through uptake studies, in which we injected B6 mice with heat-killed GFP-expressing S. Typhimurium. Significantly more CD8α+ DCs than CD4+ DCs or DN DCs contained GFP+ bacteria 2 h after injection (Fig. 5e). To understand whether this ‘preferential’ uptake affected in vivo inflammasome activation, we injected mice that had received flagellin with a fluorescence-labeled caspase-1 inhibitor (YVAD) that allows the

166

c

IL-18 (ng/ml)

a

identification of activate caspase-1 by flow cytometry. Although labeling with this reagent was weak32, we consistently found that CD8α+ DCs contained significantly more active caspase-1 than did CD4+ DCs or DN DCs (Fig. 5f and Supplementary Fig. 7a). Thus, our results showed that ‘preferential’ access of CD8α+ DCs to S. Typhimurium was correlated with in vivo inflammasome activation, which suggested that CD8α+ DCs were the dominant splenic DC subset that sensed intravenously administered flagellin. Although the findings presented above showed that CD8α+ DCs were particularly efficient at sensing flagellin through NLRC4 inflammasomes, we also wanted to assess their contribution to the regulation of innate IFN-γ secretion by memory CD8+ T cells. CD8α+ DCs are very efficient at cross-presenting exogenous antigen on MHC class I molecules33. Because this biochemical pathway requires the translocation of extracellular material into the cytosol, we speculate that this feature enables CD8α+ DCs to sense extracellular flagellin through cytosolic receptors without requiring flagellin to be delivered into the cytosol by the bacteria (Fig. 2a,b). We could test experimentally whether cross-presenting cells were important for the responses we noted, as uptake and subsequent cytosolic transfer of exogenously administered cytochrome c in cross-presenting cells leads to their selective Apaf-1-dependent apoptosis33. Treatment of mice with cytochrome c not only led to selectively fewer splenic CD8α+ DCs (Supplementary Fig. 7b), but also significantly impaired the ability of memory CD8 + T cells to secrete IFN-γ in response to HKST or flagellin (Fig. 5g), which supported the view that the efficiency with which CD8α+ DCs sensed flagellin through NLRC4 inflammasomes was linked to their crosspresentation abilities. IL-18 secretion does not require a TLR-mediated priming signal Notably, the NLRC4 inflammasome-dependent IFN- γ secretion described here was very rapid. We therefore questioned whether such kinetics were compatible with the two-step concept of inflammasome activation34. Our finding that IFN-γ secretion did not require T cell extrinsic MyD88 (Fig. 3d,e) suggested that NLRC4-dependent IL-18 secretion might not require a TLR-mediated priming signal. To test this notion, we assessed serum IL-18 concentrations in Myd88−/− mice and found that these concentrations increased in response to S. Typhimurium with efficiency similar to that in wildtype mice (Fig. 6a). To ensure that a potential TLR-mediated priming signal was not signaling through pathways associated with the alternative TLR adapter TRIF5, we also determined IL-18 serum concentrations in mice lacking TRIF or the TLRs able to signal through this adapter, that is, TLR3 or TLR4. Similar to the observations obtained with Myd88−/− mice, IL-18 serum concentrations in these mice were

VOLUME 13

NUMBER 2

FEBRUARY 2012

NATURE IMMUNOLOGY

ARTICLES d

+DTX

5

B6

4

10

7 0.6

5

5

CD86

HKST-GFP

3 Il18

–/–

+ CD11c-DTR > B6 + DTX

10

4

10

3

5.6

Il18

1

2

2

3

4

5 0 10 10 10 10 GFP

+ CD11c-DTR > B6 + PBS

f

0

0

3

1

2.2

10 0

CD45.1 + CD11c-DTR > B6 + DTX –/–

4

0

2

2

2.3

160

Active caspase-1 (% MFI)

+

10

*

4

IL-18 (ng/ml)

15

p10

5

5

C D 8α + C D D C 4+ s D DC N s D C s

c

10

**

g

*

20

**

*

15

140 IFN-γ+ (%)

4

**

20

IFN-γ (%)

6

0.3

p45 3

H KS T fla ge PB llin S + fla ge llin +

C

yt

C

+

S

Induction (fold)

h5

Pro-IL-1β Pro-IL-18

4 3 2 1 CD DC 8α + s DC s

DN

s

0 DC

– –/

88 yd M

– –/

B6

B6

0

0

*

+

50

100 80 60 40 20 0

CD 4

100

IL-1β (pg/ml)

**

M + yd fla 8 ge 8 –/– llin + fla ge B6 llin

150

IL-1β (pg/ml)

**

200 –

0

g

f

400



B6



0 0/

–/

r3

r4

Tl

Tl



ca

m

1

–/

B6

Ti

yd

88

–/



0

600

–/

0

50

** **

800

–/

2

100

e

sp Nl 1 r Py c4 ca rd

4

**

IL-1β (pg/ml)

6

IL-1β (pg/ml)

1

IL-18 (ng/ml)

2

150

M + yd fla 8 ge 8 –/– llin + fla ge B6 llin

IL-18 (ng/ml)

IL-18 (ng/ml)

3

d 10 8 6 4 2 0

Ca

c 8

Δf li Δf C Δ liC flj B pL Δflj S4 B + 08 HK ST

b 4

PB

C

yt

C

directly compared. Release of IL-1β involved bacterial expression of flagellin (Fig. 6d), required caspase-1, PYCARD and NLRC4 (Fig. 6e) and was elicited by injection of ultrapure flagellin (Fig. 6f), which confirmed that flagellin recognition through the NLRC4 inflammasome not only elicited the release of IL-18 (Fig. 3g), but also led to IL-1β release. In line with the two-signal model, but in contrast to the regulation of IL-18 release, the higher IL-1β serum concentrations in response to HKST (Fig. 6f) or ultrapure flagellin (Fig. 6g) strictly depended on MyD88. This disparate regulation of IL-18 and IL-1β secretion in response to S. Typhimurium could be explained molecularly by the fact that pro-IL-18 was preformed, whereas IL-1β requires synthesis before inflammasome-dependent activation, as has been implied26,35. To account for the possibility that different subtypes of DC could express cytokine precursors differently, we chose to assess

not different from those in wild-type mice (Fig. 6b). Similar increases in serum IL-18 were even elicited by injection of ultrapure flagellin into Myd88−/− mice and wild-type control mice (Fig. 6c). Thus, these findings demonstrate that in contrast to the predominant ‘two signal’ model proposed for NLRP3 inflammasome activation, NLRC4 inflammasome–dependent in vivo release of IL-18 in response to S. Typhimurium could occur without a TLR-mediated priming signal. The finding noted above prompted us to evaluate whether TLR-mediated priming signals were required for IL-1β release, as studies reporting the necessity of a priming signal for inflammasomemediated cytokine release have largely examined IL-1β34. It was first necessary to establish whether the release of IL-1β in response to S. Typhimurium was also mediated through NLRC4 inflammasomes recognizing flagellin, so that release of IL-18 and IL-1β could be

a

+

H KS T

+

4

D

C

C

D



+

D C s D D Cs N D C s

10 Figure 5 DCs provide IL-18 in response to inflammasome activation. (a) Splenic DCs from chimeras 120 treated with PBS or DTX. Dot plots of CD11chi MHC-IIhi events and frequency of CD11c+ GFP+ cells. 5 (b,c) Frequency of IFN-γ+ cells among total CD3+CD8+ T cells (b) and serum IL-18 concentrations 100 (c) in DTX-treated and PBS-treated chimeras 2 h after injection of HKST. (d) Pro-caspase-1 (p45) 80 0 0 and active caspase-1 (p10) in extracts from splenic DCs 2 h after injection of PBS or HKST into B6, − / − − / − hi hi + Nlrc4 or Nlrp3 mice. (e) CD11c MHC-II events and frequency of GFP DCs (n = 7) in B6 mice given injection of heat-killed GFP-expressing S. Typhimurium (HKST-GFP) 2 h earlier. (f) Active caspase-1 in DCs of B6 mice given injection of flagellin 2 h earlier, presented relative to that in control mice given injection of PBS. (g) Frequency of IFN-γ+ cells among total CD3+CD8+ T cells 2 h after injection of HKST into B6 mice treated with cytochrome c (Cyt C) or PBS. *P < 0.001 and **P < 0.0001 (one-way ANOVA followed by Bonferroni multiple comparison test (b,e–g) or paired Student’s t-test (c)). Data are representative of two (a–e,g; mean and s.e.m. in e) or four (f; mean and s.e.m.) experiments.

M

© 2012 Nature America, Inc. All rights reserved.

0.5

GFP+ DC (%)

2

2 0 10 10 10 10 GFP

npg

DN DCs

0.3

23.7

10 0

b

+

CD4 DCs

PBS

3

10 CD11c

CD8α+ DCs

B6

+

10

e

PB S H KS N lrc T 4 –/– N + lrp 3 –/– HK ST + H KS T

+PBS

+

a

Figure 6 Release of IL-1β, but not of IL-18, depends on TLR stimulation. (a,b) Serum IL-18 concentrations 2 h after injection of HKST into Myd88−/− mice (a), TRIF-deficient (Ticam1−/−) mice, Tlr3−/− mice, Tlr40/0 mice and wild-type B6 mice (b). (c) Serum IL-18 concentrations 2 h after intravenous injection of 10 μg/mouse commercially available ultrapure flagellin. (d–f) Serum IL-1β concentrations 2 h after injection of heat-killed S. enterica mutants (d) or HKST (e,f) into wild-type mice, mice lacking inflammasome components (e) or MyD88 (f). (g) Serum IL-1β concentrations 2 h after intravenous injection of 10 μg/mouse commercially available ultrapure flagellin into Myd88−/− or wild-type mice. (h) Cytokine gene expression in DC subsets isolated from B6 mice treated with 1 × 108 CFU HKST, presented relative to that in control mice treated with PBS. *P < 0.001 and **P < 0.0001 (one-way ANOVA followed by Bonferroni multiple comparison test (b,d,e) and paired Student’s t-test (a,c,f,g)). Data are representative of three (a,d–f), at least two (b), two to three (c,g) or five (h; mean and s.e.m.) experiments.

NATURE IMMUNOLOGY

VOLUME 13

NUMBER 2

FEBRUARY 2012

167

ARTICLES

npg

© 2012 Nature America, Inc. All rights reserved.

pro-IL-1β and pro-IL-18 transcription by real-time PCR in splenic DC subsets (Fig. 6h). CD4+ DCs, DN DCs and CD8α+ DCs were sorted from spleens of mice that were injected with HKST 2 h earlier. HKST induced pro-IL-1β transcription in all DC subsets, whereas there was negligible induction of pro-IL-18 (Fig. 6h). Together these findings showed that the requirement for TLR-mediated priming signals did not simply depend on the type of inflammasome activated34,36,37 but also depended on the type of cytokine processed downstream of the inflammasome. DISCUSSION In this study, we have identified a mechanism that established a direct link between flagellin-dependent NLRC4 inflammasome activation and protective IFN-γ production by noncognate memory CD8+ T cells. We found that S. Typhimurium–derived flagellin activated NLRC4 inflammasomes in CD8α+ DCs in vivo. This recognition elicited the release of IL-18, which directly signaled to memory CD8+ T cells to induce their secretion of IFN-γ, underscoring the idea that direct DC-T cell interactions are not limited to antigen-specific T cell responses but also orchestrate innate immune responses. These findings not only unify inflammasome-mediated bacterial recognition by DCs with protective IFN-γ responses by memory T cells but also provide a comprehensive mechanistic framework for the regulation of noncognate memory T cell responses. The critical role of NLRC4 in the induction of protective IFN-γ by noncognate memory T cells is in agreement with reports of greater S. Typhimurium susceptibility without both NLRP3 and NLRC4 (ref. 38). In studies of S. Typhimurium mutants that permanently express flagella, a mechanism has been proposed by which NLRC4 inflammasomes contribute to S. Typhimurium clearance39. Induction of NLRC4-dependent pyroptosis in infected cells exposes bacteria to the extracellular space, where they are subsequently eliminated by neutrophils39. Although this study has uncovered a notable pathway, this mechanism not only relies on fixed expression of flagella39, but also is limited to cells that express NLRC4 and can undergo pyroptosis. We think our observation that NLRC4-dependent flagellin recognition is rapidly translated into IFN-γ secretion reflects a host defense strategy that acts synergistically with pyroptosis-induced elimination of S. Typhimurium. IFN-γ helps infected cells eliminate intracellular pathogens40. Thus, by providing IFN-γ to infected cells that do not express NLRC4, noncognate memory CD8+ T cells enable these cells to purge S. enterica although they may not sense the presence of the bacteria through inflammasomes themselves. Despite these beneficial antibacterial effects, stimulating memory CD8+ T cells in an antigen-independent fashion also carries the risk of activating memory CD8+ T cells with unwanted TCR specificities, such as those specific for self-antigens. Given the well-established association between several bacterial infections and subsequent autoimmune episodes, the mechanisms idenitfied in this study could contribute to such autoimmune reactions. Notably, the IFN-γ responses identified here occur rapidly. Our observation that inflammasome-dependent IL-18 signals do not require time-consuming TLR-mediated priming signals explains how such rapid responses to flagellin can occur. Conversely, the much lower concentrations of IL-1β observed early after HKST, which may increase later in the response as priming signals are received, might explain why IL-1β did not have an important role early in the response. Other bacteria, such as Y. pseudotuberculosis or P. aeruginosa, could also initiate similar responses, suggesting that this mechanism represents a more widely applied host defense strategy. Therefore, the rapid downregulation of flagella by P. aeruginosa and S. Typhimurium after successful invasion of cells37,38, or

168

the cessation of production of flagella in Y. pseudotuberculosis on exposure to mammalian host temperature41, probably represents bacterial strategies aimed at evading this rapid host defense mechanism. The observation that only certain flagellated bacteria elicited IFN-γ secretion by memory CD8+ T cells is also notable in the context of the observation that NAIP molecules are the key factors that confer specificity of the NLRC4 inflammasome for certain bacterial ligands27,28. Although NAIP5 (also known as Birc1e) recognized flagellin from S. enterica, Y. pseudotuberculosis and P. aeruginosa, it did not bind flagellin from E. coli. This binding pattern is markedly similar to the panel of bacteria that we have found to elicit innate IFN-γ secretion by memory CD8+ T cells, suggesting that the responses described here might also involve NAIP5. Conversely, given our finding that the type III secretion apparatus was not required for early IFN-γ secretion, we speculate that NAIP2, which seems to be an important link in the activation of NLRC4 by rod structures from type III secretion systems27,28, is not required for the early regulation of innate IFN-γ secretion by memory CD8+ T cells. This study also yields insights about the function of inflammasomes. Our in vivo study confirmed key mechanisms that have been described before only in vitro, such as the critical role of NLRC4 for flagellin-mediated inflammasome activation and a requirement for TLR-mediated priming signals for flagellin-induced secretion of IL-1β, and also uncovered some important differences. For example, in bone marrow–derived DCs, which are considered in vitro equivalents of inflammatory DCs11, NLRC4 inflammasome activation has required the delivery of flagellin through bacterial type III secretion systems21,42. Our study showed that inflammasome activation by CD8α+ DCs in vivo did not have such requirements and was elicited by purified flagellin. Thus, although both DC subtypes recognized flagellin through NLRC4 inflammasomes, one subset responded to flagella only in the context of viable bacteria, whereas the other could sense soluble flagellin irrespective of the presence of bacteria. We speculate that these distinct DC subset-specific requirements represent functional adaptations of the DCs to their respective microenvironments, with inflammatory DCs differentiating at sites of bacterial replication and CD8α+ DCs directly exposed to the blood in the marginal zone of the spleen43. The selective ability of CD8α+ DCs to sense soluble flagellin in vivo also raises the question of how flagellin gains access to the cytosolic NLRC4 inflammasomes. Our cytochrome c depletion studies suggest that this may be linked to the specialization of CD8α+ DCs for crosspresentation. As this process requires the translocation of exogenous antigen into the cytosol for subsequent MHC class I presentation33, flagellin could access the cytosol through this shunt. Further work is required to delineate the biochemical events that enable CD8α+ DCs to sense extracellular material through cytosolic receptors. Our observations suggest that the dependence of inflammasome activation on a priming signal not only is determined by the type of stimulus6 or the type of responding inflammasome34, but also depends on which cytokine is being activated. Such disparate priming signal requirements after stimulation of the same inflammasome by the same stimulus further highlight the complexity of inflammasome activation and regulation. In comparing our insights with the regulation of NLRP3 inflammasomes, we find that priming signals seem to affect distinct inflammasomes in different ways. For example, priming signals required for NLRP3 inflammasome activation not only are important for the production of cytokine precursors, but also regulate the expression of NLRP3 itself 44. Consistent with a report of constitutive expression of NLRC4 in mouse bone marrow–derived DCs13, our finding that NLRC4-dependent IL-18 release occurred without TLR-mediated priming signals suggests that NLRC4 does not

VOLUME 13

NUMBER 2

FEBRUARY 2012

NATURE IMMUNOLOGY

ARTICLES require induction. This indicates that priming of NLRC4-dependent responses occurs through cytokine precursor expression rather than through the induction of NLRC4 itself. In summary, our results show a mechanism by which bystander memory CD8+ T cells contribute to antibacterial immunity. These findings provide a detailed description of the mechanistic framework for noncognate memory CD8+ T cell responses in vivo. They also identify a role for splenic CD8α+ DCs as important sensors of flagellin through NLRC4 inflammasomes. METHODS Methods and any associated references are available in the online version of the paper at http://www.nature.com/natureimmunology/.

npg

© 2012 Nature America, Inc. All rights reserved.

Note: Supplementary information is available on the Nature Immunology website. ACKNOWLEDGMENTS We thank D. Maskell (University of Cambridge) for mutant S. Typhimurium strain SL1344 ΔmsbB; R. Robins-Browne (University of Melbourne) for motile E. coli MG1655 and Y. pseudotuberculosis B14; C. Whitchurch (University of Technology, Sydney) for Pseudomonas aeruginosa and P. aeruginosa PAKMS591 (fliC− P. aeruginosa); A. Walduck (University of Melbourne) for Helicobacter pylori SS1; F.R. Carbone, E.L. Hartland, A.M. Lew, L. Alexopoulou, S. Akira, R.A. Flavell, D.I. Godfrey and A.G. Brooks for reagents, bacteria, mice and discussions; and A. Turner for assistance. Supported by the National Health and Medical Research Council of Australia (Career Development Awards to T.G., O.L.C.W. and S.B.), the Louis-Jeantet Foundation (G.G. and J.T.) and the Gates Foundation, through the Grand Challenges in Health program (R.C., O.L.C.W. and R.A.S.). AUTHOR CONTRIBUTIONS A.K., R.A.S. and S.B. conceived of the study, designed and undertook experiments and wrote the manuscript; G.G. and J.T. contributed reagents and intellectual input; L.E.S. did immunoblots on splenic DCs; K.R.S., H.C., P.G.W. and R.C. did PCR and generated mutants; J.C.W. and W.C. did experiments and provided reagents; D.A.D. and O.L.C.W. were involved in the initial finding; D.F.-R. generated flt3L DCs and T.G. and W.R.H. provided intellectual input and critically reviewed the manuscript. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at http://www.nature.com/natureimmunology/. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html. 1. Williams, M.A. & Bevan, M.J. Effector and memory CTL differentiation. Annu. Rev. Immunol. 25, 171–192 (2007). 2. Berg, R.E., Crossley, E., Murray, S. & Forman, J. Memory CD8+ T cells provide innate immune protection against Listeria monocytogenes in the absence of cognate antigen. J. Exp. Med. 198, 1583–1593 (2003). 3. Tough, D.F., Sun, S. & Sprent, J. T cell stimulation in vivo by lipopolysaccharide (LPS). J. Exp. Med. 185, 2089–2094 (1997). 4. Kohlmeier, J.E., Cookenham, T., Roberts, A.D., Miller, S.C. & Woodland, D.L. Type I interferons regulate cytolytic activity of memory CD8(+) T cells in the lung airways during respiratory virus challenge. Immunity 33, 96–105 (2010). 5. Akira, S., Uematsu, S. & Takeuchi, O. Pathogen recognition and innate immunity. Cell 124, 783–801 (2006). 6. Schroder, K. & Tschopp, J. The inflammasomes. Cell 140, 821–832 (2010). 7. Franchi, L., Eigenbrod, T., Munoz-Planillo, R. & Nunez, G. The inflammasome: a caspase-1-activation platform that regulates immune responses and disease pathogenesis. Nat. Immunol. 10, 241–247 (2009). 8. Manicassamy, S. & Pulendran, B. Modulation of adaptive immunity with Toll-like receptors. Semin. Immunol. 21, 185–193 (2009). 9. Ghiringhelli, F. et al. Activation of the NLRP3 inflammasome in dendritic cells induces IL-1β-dependent adaptive immunity against tumors. Nat. Med. 15, 1170–1178 (2009). 10. Shortman, K. & Heath, W.R. The CD8+ dendritic cell subset. Immunol. Rev. 234, 18–31 (2010). 11. Shortman, K. & Naik, S.H. Steady-state and inflammatory dendritic-cell development. Nat. Rev. Immunol. 7, 19–30 (2007). 12. Heath, W.R. & Carbone, F.R. Dendritic cell subsets in primary and secondary T cell responses at body surfaces. Nat. Immunol. 10, 1237–1244 (2009).

NATURE IMMUNOLOGY

VOLUME 13

NUMBER 2

FEBRUARY 2012

13. Sander, L.E. et al. Detection of prokaryotic mRNA signifies microbial viability and promotes immunity. Nature 474, 385–389 (2011). 14. Netea, M.G. et al. Differential requirement for the activation of the inflammasome for processing and release of IL-1β in monocytes and macrophages. Blood 113, 2324–2335 (2009). 15. Piccini, A. et al. ATP is released by monocytes stimulated with pathogen-sensing receptor ligands and induces IL-1β and IL-18 secretion in an autocrine way. Proc. Natl. Acad. Sci. USA 105, 8067–8072 (2008). 16. Jouanguy, E. et al. IL-12 and IFN-γ in host defense against mycobacteria and Salmonella in mice and men. Curr. Opin. Immunol. 11, 346–351 (1999). 17. VanCott, J.L. et al. Regulation of host immune responses by modification of Salmonella virulence genes. Nat. Med. 4, 1247–1252 (1998). 18. Waithman, J., Gebhardt, T., Davey, G.M., Heath, W.R. & Carbone, F.R. Cutting edge: enhanced IL-2 signaling can convert self-specific T cell response from tolerance to autoimmunity. J. Immunol. 180, 5789–5793 (2008). 19. Tough, D.F., Borrow, P. & Sprent, J. Induction of bystander T cell proliferation by viruses and type I interferon in vivo. Science 272, 1947–1950 (1996). 20. Khan, S.A. et al. A lethal role for lipid A in Salmonella infections. Mol. Microbiol. 29, 571–579 (1998). 21. Miao, E.A. et al. Cytoplasmic flagellin activates caspase-1 and secretion of interleukin 1β via Ipaf. Nat. Immunol. 7, 569–575 (2006). 22. Salazar-Gonzalez, R.M. & McSorley, S.J. Salmonella flagellin, a microbial target of the innate and adaptive immune system. Immunol. Lett. 101, 117–122 (2005). 23. Haraga, A., Ohlson, M.B. & Miller, S.I. Salmonellae interplay with host cells. Nat. Rev. Microbiol. 6, 53–66 (2008). 24. Kambayashi, T., Assarsson, E., Lukacher, A.E., Ljunggren, H.G. & Jensen, P.E. Memory CD8+ T cells provide an early source of IFN-γ. J. Immunol. 170, 2399–2408 (2003). 25. Raué, H.P., Brien, J.D., Hammarlund, E. & Slifka, M.K. Activation of virus-specific CD8+ T cells by lipopolysaccharide-induced IL-12 and IL-18. J. Immunol. 173, 6873–6881 (2004). 26. Arend, W.P., Palmer, G. & Gabay, C. IL-1, IL-18, and IL-33 families of cytokines. Immunol. Rev. 223, 20–38 (2008). 27. Kofoed, E.M. & Vance, R.E. Innate immune recognition of bacterial ligands by NAIPs determines inflammasome specificity. Nature 477, 592–595 (2011). 28. Zhao, Y. et al. The NLRC4 inflammasome receptors for bacterial flagellin and type III secretion apparatus. Nature 477, 596–600 (2011). 29. Vijay-Kumar, M. et al. Metabolic syndrome and altered gut microbiota in mice lacking Toll-like receptor 5. Science 328, 228–231 (2010). 30. Smith, K.D. et al. Toll-like receptor 5 recognizes a conserved site on flagellin required for protofilament formation and bacterial motility. Nat. Immunol. 4, 1247–1253 (2003). 31. Jung, S. et al. In vivo depletion of CD11c+ dendritic cells abrogates priming of CD8+ T cells by exogenous cell-associated antigens. Immunity 17, 211–220 (2002). 32. Osuka, A., Hanschen, M., Stoecklein, V. & Lederer, J.A. A protective role for inflammasome activation following injury. Shock 37, 47–55 (2011). 33. Lin, M.L. et al. Selective suicide of cross-presenting CD8+ dendritic cells by cytochrome c injection shows functional heterogeneity within this subset. Proc. Natl. Acad. Sci. USA 105, 3029–3034 (2008). 34. Latz, E. The inflammasomes: mechanisms of activation and function. Curr. Opin. Immunol. 22, 28–33 (2010). 35. Puren, A.J., Fantuzzi, G. & Dinarello, C.A. Gene expression, synthesis, and secretion of interleukin 18 and interleukin 1β are differentially regulated in human blood mononuclear cells and mouse spleen cells. Proc. Natl. Acad. Sci. USA 96, 2256–2261 (1999). 36. Mariathasan, S. & Monack, D.M. Inflammasome adaptors and sensors: intracellular regulators of infection and inflammation. Nat. Rev. Immunol. 7, 31–40 (2007). 37. Elinav, E., Strowig, T., Henao-Mejia, J. & Flavell, R.A. Regulation of the antimicrobial response by NLR proteins. Immunity 34, 665–679 (2011). 38. Broz, P. et al. Redundant roles for inflammasome receptors NLRP3 and NLRC4 in host defense against Salmonella. J. Exp. Med. 207, 1745–1755 (2010). 39. Miao, E.A. et al. Caspase-1-induced pyroptosis is an innate immune effector mechanism against intracellular bacteria. Nat. Immunol. 11, 1136–1142 (2010). 40. Vazquez-Torres, A., Jones-Carson, J., Mastroeni, P., Ischiropoulos, H. & Fang, F.C. Antimicrobial actions of the NADPH phagocyte oxidase and inducible nitric oxide synthase in experimental salmonellosis. I. Effects on microbial killing by activated peritoneal macrophages in vitro. J. Exp. Med. 192, 227–236 (2000). 41. Minnich, S.A. & Rohde, H.N. A rationale for repression and/or loss of motility by pathogenic Yersinia in the mammalian host. Adv. Exp. Med. Biol. 603, 298–310 (2007). 42. Franchi, L. et al. Cytosolic flagellin requires Ipaf for activation of caspase-1 and interleukin 1β in salmonella-infected macrophages. Nat. Immunol. 7, 576–582 (2006). 43. Neuenhahn, M. et al. CD8α+ dendritic cells are required for efficient entry of Listeria monocytogenes into the spleen. Immunity 25, 619–630 (2006). 44. Bauernfeind, F.G. et al. Cutting edge: NF-κB activating pattern recognition and cytokine receptors license NLRP3 inflammasome activation by regulating NLRP3 expression. J. Immunol. 183, 787–791 (2009).

169