NOD2 stimulation induces autophagy in dendritic cells ... - Nature

Articles

NOD2 stimulation induces autophagy in dendritic cells influencing bacterial handling and antigen presentation

© 2010 Nature America, Inc. All rights reserved.

Rachel Cooney1,2,5, John Baker1,5, Oliver Brain1,2, Benedicte Danis1, Tica Pichulik1, Philip Allan1,2, David J P Ferguson3, Barry J Campbell4, Derek Jewell2 & Alison Simmons1,2 Nucleotide-binding oligomerization domain–containing-2 (NOD2) acts as a bacterial sensor in dendritic cells (DCs), but it is not clear how bacterial recognition links with antigen presentation after NOD2 stimulation. NOD2 variants are associated with Crohn’s disease, where breakdown in self-recognition of commensal bacteria leads to gastrointestinal inflammation. Here we show NOD2 triggering by muramyldipeptide induces autophagy in DCs. This effect requires receptor-interacting serine-threonine kinase-2 (RIPK-2), autophagy-related protein-5 (ATG5), ATG7 and ATG16L1 but not NLR family, pyrin domain containing-3 (NALP3).We show that NOD2-mediated autophagy is required for both bacterial handling and generation of major histocompatibility complex (MHC) class II antigen-specific CD4+ T cell responses in DCs. DCs from individuals with Crohn’s disease expressing Crohn’s disease—associated NOD2 or ATG16L1 risk variants are defective in autophagy induction, bacterial trafficking and antigen presentation. Our findings link two Crohn’s disease–associated susceptibility genes in a single functional pathway and reveal defects in this pathway in Crohn’s disease DCs that could lead to bacterial persistence via impaired lysosomal destruction and immune mediated clearance.

NOD2 is an intracellular pattern recognition receptor (PRR) expressed in a limited number of tissues that includes intestinal epithelial cells or Paneth cells and monocyte-derived cells of the immune system. NOD2 is distinguishable in that polymorphisms in the ligand recognition domain are associated with Crohn’s disease 1–3, an inflammatory disease of the human intestine thought to result from a breakdown in self-recognition of commensal gut flora together with defects in mucosal barrier function. PRRs such as NOD2 represent one of the first gateways of immune processing in the body, and signals deriving collectively from the sequential ligation of these after pathogen recognition dictate the maturation profile of the cells in which they are expressed. In the case of antigen-presenting cells, the correct processing of these signals is crucial to the nature of priming of the subsequent adaptive immune response. It is not known exactly how PRRs that recognize the same structural building blocks on pathogenic and commensal bacteria can effectively distinguish between harmless and harmful bacteria, nor is it known how this recognition breaks down in diseases such as Crohn’s. There are known links between Toll-like receptors (TLRs) and the antigen presentation pathway in that TLRs are often localized to the phagosome on internalization of pathogen, which eventually fuses with lysosomal-associated membrane protein-1 (LAMP-1)-positive and MHC class II–positive compartments. In addition, continued TLR signaling within phagosomes facilitates increased class II– associated invariant chain peptide exchange in LAMP-1+ MHC class II compartments leading to enhanced MHC class II surface expression

concomitant with DC maturation4. NOD2 presents a particular conundrum with respect to antigen presentation as it is generally localized to the cytoplasm, and it is so far not clear whether NOD2 is capable of direct intersection with the antigen presentation machinery. Recently, it has been shown that signaling through TLRs can lead to autophagosome formation 5,6. During macroautophagy (hereafter referred to as autophagy), cytoplasmic material, including organelles, protein aggregates or bacteria, is sequestered into double membrane–coated autophagosomes that subsequently fuse with endosomes and lysosomes to form autolysosomes where lysosomal degradation can occur7. Autophagy has been shown to have a major role in antigen presentation, with constitutive fusion of autophagosomes with multivesicular MHC class II–loading compartments in antigen-presenting cells8. Recent large-scale population studies have revealed further Crohn’s disease susceptibility variants in addition to NOD2, two of which function in autophagosome formation normally, namely ATG16L1 and immunity-related GTPase family M3,9–13. Here we examined whether NOD2 activation by its bacterial ligand, muramyldipeptide, is capable of inducing autophagy in primary human antigen-presenting cells, monocyte-derived DCs, and show that muramyldipeptide treatment of DCs leads to induction of autophagy. This phenomenon requires NOD2 and the NOD2 signaling mediator RIPK-2, but not NALP3, a PRR that also recognizes muramyldipeptide. In addition, we found that NOD2 autophagy induction requires known autophagy proteins including phosphatidylinositol 3-kinase

1Medical

Research Council Human Immunology Unit, Weatherall Institute of Molecular Medicine, 2Department of Gastroenterology and 3Nuffield Department of Clinical Laboratory Sciences, John Radcliffe Hospital, Headington, Oxford, UK. 4School of Clinical Sciences, University of Liverpool, Duncan Building, Liverpool, UK. 5These authors contributed equally to this work. Correspondence should be addressed to A.S. ([email protected]). Received 8 June; accepted 13 November; published online 6 December 2009; doi:10.1038/nm.2069

90

volume 16 | number 1 | january 2010 nature medicine

P D

EB SS

C

80

*

*

60

*

40 20

C

on tro l PA MD P M 3C YS 4 LP S

0

f LC3-I LC3-II

LPS

on tro l

80 60 40 20 0

M

EBSS

Viability (%)

d 100

PAM3CYS4

C

Merge

on tro l M D P PA M 3C YS LP S 4

Actin

Actin

TO-PRO-3

C

D P EB SS

LC3-I LC3-II

LC3-I LC3-II

LC3

Cells with autophagosomes (%)

e

M

D P M E6 DP 4D + EB SS E E6 BS 4D S +

c

MDP

MDP

b

**

**

Control

Control

100 80 60 40 20 0

M

Merge

on tro l

TO-PRO-3

C

LC3

on tro l M D P EB SS

a

Cells with autophagosomes (%)

Articles

Actin Figure 1 Muramyldipeptide (MDP) induces autophagy in DCs. DCs were incubated in the presence or absence of MDP (1 µg ml−1) for GFP-LC3 Lysotracker Merge 8 h or incubated in EBSS for 2 h. (a) Left, cells stained with antibody to LC3 (green) and TO-PRO-3 nuclear stain (blue). Scale bar, 10 µm. Right, percentage of LC3-positive autophagosome forming cells. (b) Immunoblot of LC3 isoforms with antibody to LC3 or antibody to β-actin. (c) Immunoblot of LC3 isoforms with antibody to LC3 or antibody to β-actin using DCs left untreated or treated N with E64D. (d) Percentage of viable cells after * Trypan blue staining. (e) Comparison of MDP60 * mediated autophagy with that of PAM3CYS4 * 40 or LPS. DCs were left unstimulated (control) 20 or treated with MDP, PAM3CYS4 or LPS at 0 1 µg ml−1 for 8 h. Right, quantification of AV cells containing LC3-positive autophagosomes. Left, representative confocal images. Scale bars, 10 µm. (f) Immunoblot of LC3 after MDP, PAM3CYS4 or LPS treatment. LC3-I, unmodified LC3. (g) DCs transfected with GFP-LC3 for 24 h and stained with Lysotracker left unstimulated (control) or stimulated with MDP, PAM3CYS4 or LPS. Scale bars, 10 µm. (h) Transmission electron micrograph showing autophagy in a MDP treated DC with enlargement of the enclosed area illustrating autophagic vacuoles (AV). N, nucleus. Scale bars represent 1 µm (for the large image) and 100 nm (for the enclosed area). Data are means of three replicates of random counting of 100 DCs and identifying those with one or more autophagic vacuoles. *P < 0.05, **P < 0.01. For quantification of autophagosome-forming cells, data are mean ± s.d. from eight independent counts. Data are representative of more than three independent experiments.

g

h

D P YS 4 LP S

M

PA

M

3C

l

tro

on

PAM3CYS4

C

LPS

DCs with autophagosomes (%)

MDP


Control

(PI3K), ATG5, ATG7 and ATG16L. We investigated whether NOD2mediated autophagy provides a means by which NOD2 can link with the MHC class II antigen presentation machinery in DCs. NOD2 stimulation led to MHC class II DM (HLA-DM) localization with microtubule-associated protein 1A/1B/light chain-3 (LC3) autophagy dependent upregulation of surface MHC class II and generation of antigen-specific CD4+ T cell responses. In contrast, we found that DCs isolated from individuals with Crohn’s disease expressing associated susceptibility variants in NOD2 (1007fsinsC, R702W or G908R) or ATG16L1 T300A were deficient in these effects. Examining the effect of Crohn’s disease–variant NOD2 on bacterial handling revealed reduced localization of both Salmonella enterica serovar typhimurium (S. enterica) and Crohn’s-associated, adherent-invasive Escherichia coli with autophagolysosomes, an effect that was reversible after treatment with the autophagy activator rapamycin. These findings provide a mechanism in which NOD2 influences bacterial degradation and interacts with the MHC class II antigen presentation machinery within DCs and link two genes implicated in Crohn’s susceptibility within one functional pathway. A combination of defective bacterial lysosomal degradation and induction of antigen-specific CD4+ T cell responses in the presence of variant NOD2 would allow the persistence of bacterial components that could trigger inflammatory responses in Crohn’s disease.

nature medicine volume 16 | number 1 | january 2010

RESULTS Muramyldipeptide induces autophagy in DCs To investigate a link between NOD2 and autophagy induction, we used primary immature human DCs that expressed wild-type (WT) NOD2 and treated the cells with muramyldipeptide, measuring autophagosome induction by assessing the degree of LC3 relocalization from diffuse to punctate staining. LC3 is a modifier protein conjugated to phosphatidylethanolamine, analogous to ATG8 in yeast14. Phosphatidylethanolamine-conjugated LC3 (LC3-II) is localized in the inner and outer membranes of autophagosomes, and the population associated with the inner membrane is degraded after fusion of autophagosomes with lysosomes15. After muramyldipeptide treatment of DCs, we observed efficient induction of autophagy in comparison with nonstimulated cells. The degree of LC3 relocalization was comparable to that induced after growth of DCs in the starvation medium Earle’s balanced salt solution (EBSS) (Fig. 1a). Phosphatidylethanolamine-conjugated forms of LC3 can be used to detect the degree of autophagy in stimulated cells by western blotting. We detected increased LC3-II in DCs after muramyldipeptide stimulation to an equivalent level to that of EBSS in comparison with nonstimulated cells (Fig. 1b). This accumulation was exaggerated after treatment of DCs with the lysosomal protease inhibitor E64D, suggesting that the observed increase was not due

91

Articles a

b

Figure 2 NOD2 signaling is required for NS NS NALP3 NALP3 NOD2 NOD2 MDP-Induced autophagy. (a) DCs were siRNA siRNA siRNA siRNA siRNA siRNA NS NOD2 NS NALP3 + control + MDP + control + MDP + control + MDP siRNA siRNA siRNA siRNA transfected with NS siRNA or NOD2-specific siRNA for 24 h and then left unstimulated WB WB WB NALP3 or stimulated with MDP for 8 h. Top left, NOD2 LC3 western blot (WB) analysis using antibody WB WB WB to NOD2; bottom left, western blot analysis actin actin actin using antibody to β-actin. DCs were transfected with NS siRNA or NALP3-specific siRNA for TO-PRO-3 Merge LC3 24 h and then left unstimulated or stimulated with MDP for 8 h. Top right, *** 80 *** western blot analysis with antibody to *** NS siRNA 60 NALP3; bottom right, western blot analysis with antibody to β-actin. (b) Immunoblot 40 of LC3 isoforms after MDP stimulation of DCs transfected with NS, NALP3–specific 20 NOD2 or NOD2-specific siRNA for 24 h, using siRNA MDP antibody to LC3 (top) or antibody to 0 β-actin (bottom). (c) Representative confocal images of LC3 staining in DCs transfected with NS, NOD2-specific or NALP3-specific siRNA for 24 h and treated with MDP for NALP3 8 h. Cells were stained with antibody to siRNA LC3 (green) and TO-PRO-3 (blue). Scale bars, 10 µm. Quantification of the percentage of LC3-positive autophagosome-forming DCs transfected with NS, NOD2-specific or NALP3-specific siRNA for 24 h and either left unstimulated or treated with MDP for 8 h. ***P < 0.001. Results are means ± s.d. of three independent experiments. Autophagosome-forming si R DCs (%) N A N + S c N on si O R tro D N 2 l A si + R M N N D O A P D + 2 co N si AL nt R ro N P3 A l si + R M N N D AL A P P3 + co si nt R ro N A l + M D P

N S

to decreased degradation of lipidated LC3 (Fig. 1c). The autophagy induction mediated by muramyldipeptide was not associated with any increase in cell death above background levels in comparison with nonstimulated cells (Fig. 1d) and was as robust as that induced by stimulation of TLR4 with LPS or TLR1 or TLR2 with PAM3CYS4 (Fig. 1e,f).

b

MDP

a

LC3

TO-PRO-3

Merge

NS ATG5 siRNA siRNA WB ATG5

0 mM 3-MA

WB actin NS ATG7 siRNA siRNA

2 mM 3-MA

5 mM 3-MA

NS ATG16L1 siRNA siRNA WB ATG16L1

10 mM 3-MA

80 60 40 20

60

***

***

20

N

A

+

7

G

AT

G

5

si

si

R

A N

R si

S

5 G

AT

co P nt ro l R N +M AT AT G G7 A + DP 16 si c o R N ntr AT L1 G siR A + ol 16 N M L1 A + D P si R con N A tro l + M D P

ro l

D

nt

M

co

+

N si

R

A N si R

N

S

A

+

m 10

N

AT

A M

A M

3M

3M m

5

+ P D

M

P D M

P D M

+

+

0

2

m

m

M

M

3-

M

3M

A

A

0

+ P D

***

40

0

92

WB actin

NOD2 is required for muramyldipeptide-mediated autophagy NOD2 is not the sole ligand for muramyldipeptide in monocytederived cells, as muramyldipeptide is also recognized by the NALP inflammasome16. To determine to what extent NOD2 contributes to muramyldipeptide-mediated autophagy induction in DCs, we used siRNAs to knock down expression of NOD2 or NALP3 in these cells. We transfected WT NOD2–expressing immature DCs with nonspecific nonsilencing (NS, control), NOD2-specific or NALP3-specific siRNAs and stimulated them with muramyldipeptide. We confirmed efficient

***

80



c

** **

100

We used a GFP-LC3 fusion protein to assess muramyldipeptidemediated autophagy induction further. We transfected immature DCs with either GFP-LC3 and left them unstimulated or stimulated with muramyldipeptide, LPS or PAM3CYS4. After stimulation, we observed marked redistribution of GFP-LC3 from diffuse to punctate staining (Fig. 1g). We also quantified the degree of autophagosome formation induced by muramyldipeptide treatment of DCs by electron microscopy, which confirmed increased autophagosome formation after stimulation of DCs with muramyldipeptide (Fig. 1h). The quantity of autophagosome-forming cells within the population of cells was equivalent to that found by confocal analysis (Fig. 1h). Thus, muramyldipeptide is capable of directly inducing autophagosome formation in primary human DCs to levels equivalent to those seen after stimulation of other PRRs and of starvation stimuli.

WB ATG7 WB actin

M


c

Figure 3 MDP-induced autophagy requires PI3K, ATG5, ATG7 and ATG16L1. (a) DCs were incubated for 8 h in varying concentrations of 3-methyladenine (3-MA) and stimulated with MDP. Top, representative confocal images of autophagosome-forming cells. Scale bars, 10 µm. Bottom, quantification of the percentage of LC3-positive autophagosomeforming cells. (b) DCs were transfected with NS siRNA, ATG5-specific, ATG7-specific or ATG16L1-specific siRNAs. Western blot performed with antibodies to ATG5, ATG7 or ATG16L1 and β-actin. (c) Quantification of the percentage of LC3-positive autophagosome-forming cells after transfection of NS siRNA, ATG5-specific, ATG7-specific or ATG16L1-specific siRNAs for 24 h and MDP stimulation for 8 h. **P < 0.01, ***P < 0.001. Results are means ± s.d. of three independent experiments.


Figure 4 NOD2-mediated autophagosome formation is required for MDP-mediated MHC class II upregulation and antigen presentation. (a) FACS analysis of HLA-DR on DCs left unstimulated or stimulated with MDP or PAM3CYS4 for 24 h. (b) Confocal analysis of DCs left unstimulated or stimulated with MDP for 8 h. Cells were stained with antibody to LC3 and antibody to HLA-DM. Scale bars, 10 µm. (c) Confocal analysis of DCs transfected with NS, ATG5-specific, ATG7-specific or ATG16L1specific siRNAs, left unstimulated or stimulated with MDP and stained with antibody to LC3 and antibody to HLA-DM. Scale bars, 10 µm. (d) Top, FACS analysis of HLA-DR surface expression after stimulation of DCs transfected with NS, ATG5–specific, ATG7–specific or ATG16L1– specific siRNA for 24 h and subsequently left unstimulated or stimulated with MDP for 24 h. Bottom, FACS analysis of MHC I surface expression on DCs left unstimulated or stimulated with MDP or PAM3CYS4 for 24 h. (e) Proliferation of autologous CD4+ T cells incubated with DCs transfected with NS siRNA, NOD2-specific siRNA or ATG16L1–specific siRNA for 24 h and exposed to live attenuated S. enterica or live attenuated S. enterica expressing the C-fragment (Cfr) of tetanus toxin. Data are shown as means ± s.d. of triplicate cultures. **P < 0.01. Data are representative of three independent experiments.

b

a

LC3

HLA-DM

TO-PRO-3

Merge

Control

HLA-DR MDP

PAM3CYS4 MDP Control Isotype

c

d

MDP LC3

HLA-DM

TO-PRO-3

Merge

Isotype NS siRNA control

NS siRNA

NS siRNA + MDP

Isotype ATG5

ATG5 siRNA control

siRNA

ATG5 siRNA + MDP

ATG7 siRNA

Isotype ATG7 siRNA control ATG7 siRNA + MDP

ATG16L1 siRNA

Isotype ATG16L1 siRNA control

e

ATG16L1 siRNA + MDP

**

** 500 HLA-DR knockdown of either NOD2 or NALP3 by *** Salmonella 400 immunoblotting (Fig. 2a). We then examined Salmonella Cfr 300 Isotype the degree of autophagosome formation after NS siRNA + MDP 200 muramyldipeptide treatment in these cells. ATG5 siRNA + MDP 100 NOD2 but not NALP3 knockdown resulted ATG7 siRNA + MDP 0 in a decrease in the accumulation of LC3-II as ATG16L1 siRNA + MDP NS siRNA NOD2 ATG16L1 MHC I siRNA siRNA observed by western blotting after muramyl dipeptide treatment (Fig. 2b) and in LC3positive autophagosome–forming DCs (Fig. 2c). As autophagy has been shown to be responsible for constitutive We then investigated to what extent RIPK-2, a known mediator of and efficient delivery of cytosolic proteins for MHC class II presNOD2 signaling17,18, is involved in this process by siRNA-mediated entation in DCs8,19,20 and NOD2 has been shown to induce antigen knockdown and with a drug inhibitor of RIPK-2. Our analysis specific immunity in mice21, we investigated whether NOD2-induced confirmed a role for RIPK-2 as a downstream mediator in NOD2- autophagy influences antigen presentation. We compared the extent mediated autophagy (Supplementary Fig. 1). In contrast, we found no to which muramyldipeptide influences MHC class II upregulation, a requirement for the key TLR signal transducer myeloid differentiation function of DC maturation, in immature DCs expressing WT NOD2 factor-88 in muramyldipeptide-mediated autophagosome formation to that induced by PAM3CYS4. The level of MHC class II surface (Supplementary Fig. 1). These findings confirm that muramyldipep- expression after stimulation with muramyldipeptide was increased tide induced autophagy is progressing to a large extent through NOD2 above the levels observed on immature DCs, consistent with DC in muramyldipeptide treated DCs. maturation stimuli (Fig. 4a). To determine whether this increase in surface MHC class II represented an increase in protein expression Autophagy is required for NOD2 antigen presentation or redistribution, we biotinylated the cell surface of stimulated DCs We investigated to what extent conventional autophagy mediators are before immunoprecipitating biotin-conjugated surface proteins and required for NOD2-mediated autophagy. Class III PI3K is involved at western blotting for HLA-DR. A comparison of surface expression the vesicle elongation step of autophagy and can be inhibited by the PI3K of HLA-DR with the total HLA-DR detected in whole-cell lysates inhibitor 3-methyladenine. We tested the effect of 3-methyladenine from the same cells indicated that the increase in HLA-DR surface treatment on NOD2-mediated autophagy induction and found expression after DC stimulation was a result of redistribution from it to inhibit autophagosome formation (Fig. 3a). We then used within the cell (Supplementary Fig. 2). MHC class II surface egress is siRNAs to knock down ATG5, ATG7 or ATG16L1 expression within DCs facilitated by HLA-DM–mediated class II–associated invariant chain (Fig. 3b) and observed a marked reduction in autophagosome forma- peptide exchange for antigenic peptide. We examined DCs for subceltion after muramyldipeptide stimulation (Fig. 3c). Therefore, several lular redistribution of LC3 with HLA-DM after muramyldipeptide known components of autophagosome formation, PI3K, ATG5, ATG7 stimulation and observed relocalization of LC3 to HLA-DM– and ATG16L1, are essential for NOD2-mediated autophagy. containing compartments (Fig. 4b). [H3] incorporation (c.p.m. × 103)


Articles


93

Articles LC3

TO-PRO-3

TO-PRO-3

1007fsinsC NOD2 MDP

1007fsinsC PAM3CYS4

R702W or G908R NOD2 PAM3CYS4

***

d

3

500 400

***

***

***

R702W or G908R NOD2 MDP

200 100

sC

R

fs 07 10 TO-PRO-3

Merge ATG16L1 control

f

LC3

HLA-DM

TO-PRO-3

ATG16L1 T300A control

***

40

ATG16L1 PAM3CYS4

ATG16L1 T300A PAM3CYS4

20 ATG16L1 T300A control ATG16L1 T300A MDP

h

3

400

***

PAM3CYS4 MDP Control Isotype HLA-DR

g ATG16L1 T300A Isotype Control MDP PAM3CYS4 HLA-DR ATG16L1

***

Isotype Control MDP PAM3CYS4

Salmonella Salmonella Cfr

300 200 100

HLA-DR

16 L1 G T3 16L 00 1 AT A G T3 16L 00 1 A AT

L1 16

G AT

ATG16L1 T300A PAM3CYS4

G

G

0

AT

0

[H ] incorporation 3 (c.p.m. × 10 )

Autophagosomeforming DCs (%)

60

ATG16L1 T300A MDP

*** ***

AT T 16 G1 300 L1 6L A 1 T3 T + c 00 30 on A 0A tro AT + P + M l G AM D 16 3 P AT A L1 CY G TG + c S4 16 1 o L1 6L ntr + 1 + ol PA M M DP 3C YS 4

L1 16 G AT

80

HLA-DR R702W/G908R NOD2

Merge

ATG16L1 MDP 100

PAM3CYS4 MDP Control Isotype

R702W or G908R NOD2 PAM3CYS4

70 G 2W 90 o 8R r

T W

nt ro l PA MD M fs P in 3C 10 10 sC Y + S4 07 07 c fs fs in on in sC sC tro l + + PA MD P M 3C YS 4

co

+

T

W

+

T

10

W

LC3

HLA-DR 1007fsinsC NOD2

R702W or G908R NOD2 control

Salmonella Salmonella Cfr

300

WT NOD2 PAM3CYS4 MDP Control Isotype

1007fsinsC NOD2 PAM3CYS4

0

e

c

Merge

R702W or G908R NOD2 MDP

in


HLA-DM

1007fsinsC MDP

0

+

LC3

1007fsinsC control

20

T

b

Merge

1007fsinsC NOD2 control

40

W

TO-PRO-3

R702W or G908R NOD2 control

60

AT


LC3

[H ] incorporation 3 (c.p.m. × 10 )

*** ***

80

Merge

07

a

Figure 5 Failure in NOD2-mediated autophagy induction and antigen presentation in DCs expressing Crohn’s disease–associated NOD2 and ATG16L1 variants. Crohn’s disease DCs homozygous for 1007fsinsC and homozygous or compound heterozygous for R702W or G908R NOD2 were obtained. (a) Top, representative confocal images after stimulation with MDP or PAM3CYS4, stained with antibody to LC3 and TO-PRO-3. Scale bars, 10 µm. Bottom, percentage of LC3-positive autophagosome–forming DCs. (b) Localization of LC3 with HLA-DM with cells treated as in a. Scale bars, 10 µm. (c) FACS analysis of HLA-DR surface expression on DCs after activation with MDP for 24 h. (d) DCs were exposed to live attenuated S. enterica or S. enterica expressing the C-fragment (Cfr) of tetanus toxin for 2 h before being cultured with autologous CD4 + T cells and T cell proliferation determined by [H3] incorporation. (e) Representative confocal images and quantification of LC3-positive autophagosomes in Crohn’s disease DCs expressing ATG16L1 T300A and WT NOD2 left unstimulated or stimulated with MDP or PAM3CYS4. Scale bars, 10 µm. (f) DCs treated as in e, stained with antibody to LC3, antibody to HLA-DM and TO-PRO-3. Scale bars, 10 µm. (g) FACS analysis of HLA-DR surface expression on DCs after activation with MDP or PAM3CYS4 for 24 h. (h) T cell proliferation of autologous CD4+ T cells after incubation with DCs exposed to live attenuated S. enterica or live attenuated S. enterica expressing the C-fragment (Cfr) of tetanus. Results are representative of three independent experiments from different donors. **P < 0.01, ***P < 0.001.

To assess whether muramyldipeptide-induced autophagosome formation contributes to muramyldipeptide-mediated HLA-DR surface expression, we transfected DCs were with NS, ATG5-specific, ATG7-specific or ATG16L1-specific siRNAs and left them unstimulated or stimulated with muramyldipeptide. There was robust HLADM colocalization with LC3 after muramyldipeptide treatment of DCs expressing NS siRNA but not in DCs transfected with ATG5-specific, ATG7-specific or ATG16L1-specific siRNAs (Fig. 4c). Furthermore, FACS analysis of cell surface HLA-DR revealed a reduction in HLA-DR surface expression in cells where autophagy protein expression was diminished (Fig. 4d). In contrast, we did not observe a change in surface expression of MHC I in these conditions (Fig. 4d). We then investigated the effect of NOD2-mediated autophagy on antigen presentation. We used a live attenuated recombinant S. enterica serovar

94

typhimurium aroA aroD mutant expressing the carboxy-terminal fragment of tetanus toxin (Cfr). We infected DCs transfected with NS or NOD2-specific siRNAs with recombinant S. enterica Cfr before exposure to autologous CD4+ T cells of tetanus-immune individuals. After knockdown of either NOD2 or ATG16L1, there was a significant reduction in antigen-specific proliferative responses (Fig. 4e). These results indicate that NOD2 autophagy induction influences NOD2mediated MHC class II trafficking, surface expression and MHC class II antigen presentation in DCs. Defective autophagy in Crohn’s disease DCs We tested whether DCs obtained from individuals with Crohn’s disease expressing variant NOD2 were capable of inducing autophagy to the same degree as WT NOD2–expressing DCs. We selected


Articles

Merge

20 0

R702W or G908R NOD2

d E. coli

Lysotracker

TO-PRO-3

N

R702W or G908R NOD2 + rapamycin

**

**

E. coli localizing with lysosomes (%)

100 80 60 40 20 0

*** ***

Merge 1007fsinsC NOD2 + vehicle

O

R702W or G908R NOD2 + vehicle

E. coli localizing with lysosomes (%)

1007fsinsC NOD2

40

1007fsinsC NOD2 + vehicle 1007fsinsC NOD2 + rapamycin

1007fsinsC NOD2 + rapamycin R702W or G908R NOD2 + vehicle R702W or G908R NOD2 + rapamycin

e

80

***

***

60 40 20 0

200

**

**

*

T

10 W mi n fs WT 1 R 10 ins T h 70 C 2 2W 1 07f 10 h 0 s R or 07 ins mi 70 G fs C n R 2W 90 ins 1 h 70 8 C 2W or R 1 2 G h or 90 0 m G 8R in 90 1 8R h 2 h

150 Figure 6 Crohn’s disease–variant NOD2 is associated with aberrant bacterial handling in DCs. (a) WT or 1007fsinsC NOD2 expressing donor DCs were incubated with GFP–S. enterica and Lysotracker for 2 h. Left, representative confocal images. 100 Right, percentage of bacteria localizing with lysosomes. Scale bars, 10 µm. (b) Crohn’s disease–variant NOD2 donor DCs 50 were incubated with vehicle (DMSO) or 10 ng ml−1 rapamycin before culture with GFP–S. enterica and Lysotracker for 2 h. 0 Left, representative confocal images. Right, percentage of bacteria localizing with lysosomes. Scale bars, 10 µm. (c) WT or Crohn’s disease–variant NOD2 expressing donor DCs were incubated with GFP–E. coli and Lysotracker for 2 h. Left, representative confocal images. Right, percentage of bacteria localizing with lysosomes. Scale bars, 10 µm. (d) Crohn’s disease–variant NOD2 donor DCs were incubated with vehicle (DMSO) or 10 ng ml−1 rapamycin before culture with GFP– E. coli and Lysotracker for 2 h. Left, representative confocal images. Right, percentage of bacteria localizing with lysosomes. One hundred cells were counted per condition within randomly chosen fields. Scale bars, 10 µm. (e) WT or Crohn’s disease– variant NOD2 DCs were exposed to E. coli at an MOI of 25 for 1 h and then in gentamicin (50 µg ml−1)-containing medium for the remaining time periods before lysis and quantification of intracellular CFU. Results are means ± s.d. of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

10

07

W


60

100 80 60 40 20 0

fs in 10 veh sC 07 ic + R 70 ra fs le 2W pa ins m C or yc + R G in 70 2W v 908 or ehic R + ra G9 le pa 08 m R yc + in

TO-PRO-3

WT NOD2

**

07

Lysotracker

Merge

10

Salmonella

b

**

TO-PRO-3

Adherent-invasive E. coli 2 bacteria counts (CFU × 10 )

R702W or G908R NOD2

Bacteria localizing with lysosomes (%)

1007fsinsC NOD2

80

W 10 07 T N fs O i D n R 2 70 sC 2W NO D or 2 G 90 8R

WT NOD2

Lysotracker

W T N O D 10 2 07 fs R 70 N ins 2W OD C or 2 G N 908 O R D 2

c E. coli

Merge

Bacteria localizing with D 10 lysosomes (%) 2 07 + fs v D eh ins 2 R + 100 icle C 70 ra 7 2W pa fsi m n or yc sC R G in 70 2W v 90 8 or ehi R + c ra G9 le pa 08 m R yc + in

TO-PRO-3

O

Lysotracker

N

a

Salmonella

s ubjects on the basis of whether their Crohn’s disease was quiescent and whether they had been free of immunotherapy for more than 12 months. After stimulation of WT- or variant NOD2–expressing DCs, the degree of LC3 relocalization from diffuse to punctate staining was equivalent after activation with PAM3CYS4, but there was no relocalization of LC3 after treatment of variant NOD2–expressing DCs with muramyldipeptide (Fig. 5a). We examined whether there was also a deficiency in MHC class II processing in these cells. Treatment of variant NOD2–expressing DCs with muramyldipeptide resulted in failure of HLA-DM to colocalize with LC3 and did not lead to the upregulation in MHC class II surface expression that would normally be expected after NOD2 stimulation (Fig. 5b,c). We then assessed the ability of variant NOD2 to influence antigen-specific CD4+ T cell responses with S. enterica Cfr. There was a reduction in tetanus toxin–specific CD4+ T cell proliferative responses in NOD2 variant–expressing donors in comparison with WT NOD2–expressing controls (Fig. 5d). We then investigated these effects in Crohn’s disease DCs expressing the Crohn’s disease–associated susceptibility variant ATG16L1 T300A. As before, we selected donors with the requirement that their Crohn’s disease was quiescent and that they had been off immunomodulator treatment for 12 months. In addition, these donors expressed WT NOD2. ATG16L1 T300A–expressing DCs also showed a defect in autophagy induction and HLA-DM localization with LC3 after


muramyldipeptide treatment but not after exposure to PAM3CYS4 (Fig. 5e,f). Furthermore, these cells also failed to induce MHC class II (Fig. 5g) or generate antigen-specific CD4+ T cell responses after exposure to recombinant S. enterica Cfr (Fig. 5h). These results show that Crohn’s disease DCs expressing variant NOD2 and ATG16L1 are both defective in autophagy induction after muramyldipeptide stimulation and show defects in MHC class II processing and in CD4+ T cell antigen specific responses. Defective bacterial handling in Crohn’s disease DCs As combinatorial PRR triggering could not compensate for Crohn’s disease–variant NOD2 defects in autophagy in the context of antigen presentation, we tested whether NOD2-mediated autophagy could affect bacterial handling in DCs. We assessed this with S. enterica serovar typhimurium expressing GFP (GFP–S. enterica) by comparing the amount of GFP–S. enterica localizing with lysosomes in WT- or variant NOD2–expressing DCs. We found substantially more bacteria localizing to lysosomes in WT NOD2–expressing DCs compared to those expressing Crohn’s disease variants (Fig. 6a). We tested whether the autophagy activator rapamycin could overcome this defect and found that treatment with rapamycin resulted in a substantial increase in bacteria localizing to lysosomal compartments (Fig. 6b). Autophagy-mediated engulfment of bacteria (xenophagy) provides a major mechanism of pathogen control in the case of the

95


Articles bacteria S. enterica serovar typhimurium22. To explore whether autophagy is required for lysosomal trafficking of bacteria present in the gut of individuals with Crohn’s disease, we examined the effect of Crohn’s disease–variant NOD2 on bacterial handling of E. coli. There is increasing evidence for the presence of a particular form of E. coli, adherent-invasive E. coli, in mucosa biopsies of individuals with Crohn’s disease23,24. We used GFP-labeled adherent-invasive E. coli derived from the mucosa of individuals with Crohn’s disease and exposed DCs from WT or variant NOD2–expressing donors to GFP–E. coli before visualizing the extent of bacterial localization within lysosomes. Similar to the results with GFP–S. enterica, we found a significant reduction in lysosomal localization in the presence of Crohn’s disease–variant NOD2 that was reversible with rapamycin (Fig. 6c,d). Finally, to assess whether this defect in localization with lysosomes resulted in reduced bacterial cell death, we quantified viable intracellular bacteria. We exposed DCs expressing WT or Crohn’s disease–variant NOD2 to E. coli at a multiplicity of infection (MOI) of 25 for 1 h and then in gentamicin-containing medium to kill extracellular bacteria for the remaining time periods before lysis and quantification of intracellular bacteria by colony-forming unit (CFU) counting. Crohn’s disease– variant NOD2–expressing DCs showed an increase in bacterial colony formation in comparison with WT NOD2–expressing DCs, indicating diminished bacterial killing within DCs at early time points after infection in the presence of Crohn’s disease–variant NOD2 (Fig. 6e). The inability of Crohn’s disease–variant NOD2–expressing DCs to traffic bacteria correctly via defects in muramyldipeptide-mediated autophagy induction provides a mechanism for bacterial persistence in these cells that could lead to abnormal priming of antigen-specific immune responses and tissue inflammation in Crohn’s disease. DISCUSSION Here we demonstrate that NOD2 induces autophagy in DCs that is required for NOD2-mediated antigen presentation and bacterial handling and is defective in the presence of Crohn’s disease–associated NOD2 or ATG16L1 variants. NOD2 is predominantly localized to the cytosol, and our findings provide a mechanism whereby it interacts with the antigen presentation pathway in DCs. In addition, to our knowledge, these findings are the first to link two of the strongest genetic risk factors in Crohn’s disease to a single signaling pathway. Aberrant bacterial destruction and inadequate generation of CD4+ T cell immune responses in the context of variant NOD2 expression could facilitate bacterial persistence and a mechanism for generation of secondary inflammatory changes characteristic of Crohn’s disease. It will be crucial to define the exact molecular complex by which NOD2 acts to induce autophagy in human DCs. So far, the mole cular basis by which any of the PRRs trigger autophagy and exactly what molecular interaction releases inhibition of mammalian target of rapamycin to initiate autophagosome formation in such circumstances is not known. Here we have shown that a known mediator of NOD2 signaling, RIPK-2, is required for NOD2-mediated autophagy, and it is possible that either NOD2 or RIPK-2 may link directly with proteins that govern autophagosome formation, such as mammalian target of rapamycin or other downstream autophagy pathway components. In this study, we have concentrated on defining the effect of NOD2 autophagy induction in DCs, but a key area of further investigation will be to establish whether a similar effect occurs in other cells in which NOD2 is expressed, such as macrophages and Paneth cells.

96

Macrophages, specialized in bacterial handling and destruction, use pyroptosis and caspase-1–mediated cell death as well as autophagy to handle Shigella flexneri. It has been shown induction of autophagy protects macrophages from pyroptosis25, so it is possible that defective autophagy induction by Crohn’s disease–variant NOD2 or ATG16L1 T300A increases the level of pyroptosis of bacterially infected cells, which might tip the balance toward a proinflammatory state. A series of Crohn’s disease susceptibility genes have now been mapped, and it remains to be established whether particular combinations of these, in addition to NOD2 and ATG16L1, contribute to specific cellular pathways such as autophagy and antigen presentation. Little information exists at present about the degree to which combinations of Crohn’s disease susceptibility variants contribute to either particular molecular pathways or clinical phenotypes in Crohn’s disease. Large-scale multicenter studies will be required to generate this information in the future, which will help in identifying molecular pathways that could act as therapeutic targets in Crohn’s disease. Methods Methods and any associated references are available in the online version of the paper at http://www.nature.com/naturemedicine/. Note: Supplementary information is available on the Nature Medicine website. Acknowledgments We would like to acknowledge the previous researchers in our department involved in genotyping of the Oxford Crohn’s disease cohort and thank A.-K. Simon and A. McMichael for helpful discussions. Most particularly, we would like to thank all of our donors who generously gave blood samples for this study. We are supported by grants from the UK Medical Research Council (J.B., T.P. and A.S.), the UK Higher Education Funding Council (A.S.), Oxford National Institute for Health Research Biomedical Research Centre (A.S.) and the National Institute of Health Research Specialist Biomedical Research Centre in Microbial Disease (01CD1) (B.J.C.), Action Medical Research (R.C.), Digestive Disorders Foundation and National Association for Crohn’s and Colitis and The Barbour Trust in Memory of Simon Ash (O.B.), Fondation Philippe Weiner Maurice Anspach (B.D.) and the Wellcome Trust (P.A.). The GFP-LC3 construct was a kind gift from H. Mellor (University of Oxford) GFP–S. enterica typhimurium was a kind gift from D. Holden (Imperial College, University of London) and SL5338 pTECH1 was a kind gift from K. Turner (Sanger Centre). AUTHOR CONTRIBUTIONS A.S. conceived the idea; A.S., J.B. and R.C. designed the experiments and prepared the manuscript; R.C., J.B., O.B., P.A. and A.S. did RNAi, flow cytometry, immunoblot and confocal experiments. J.B., R.C., O.B. and P.A. were involved with gene sequencing of NOD2 and ATG16L1. T.P. and B.D. provided technical assistance and participated in flow cytometry experiments; D.J.P.F. did the electron microscopy analysis; B.J.C. provided GFP–E. coli, and B.J.C. and D.J. provided additional intellectual input. Published online at http://www.nature.com/naturemedicine/. Reprints and permissions information is available online at http://npg.nature.com/ reprintsandpermissions/. 1. Ogura, Y. et al. A frameshift mutation in NOD2 associated with susceptibility to Crohn’s disease. Nature 411, 603–606 (2001). 2. Hugot, J.P. et al. Association of NOD2 leucine-rich repeat variants with susceptibility to Crohn’s disease. Nature 411, 599–603 (2001). 3. Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 447, 661–678 (2007). 4. Blander, J.M. & Medzhitov, R. Toll-dependent selection of microbial antigens for presentation by dendritic cells. Nature 440, 808–812 (2006). 5. Xu, Y. et al. Toll-like receptor 4 is a sensor for autophagy associated with innate immunity. Immunity 27, 135–144 (2007). 6. Delgado, M.A., Elmaoued, R.A., Davis, A.S., Kyei, G. & Deretic, V. Toll-like receptors control autophagy. EMBO J. 27, 1110–1121 (2008). 7. Levine, B. & Kroemer, G. Autophagy in the pathogenesis of disease. Cell 132, 27–42 (2008).


Articles 17. Park, J.H. et al. RICK/RIP2 mediates innate immune responses induced through Nod1 and Nod2 but not TLRs. J. Immunol. 178, 2380–2386 (2007). 18. Yang, Y. et al. NOD2 pathway activation by MDP or Mycobacterium tuberculosis infection involves the stable polyubiquitination of Rip2. J. Biol. Chem. 282, 36223–36229 (2007). 19. Dengjel, J. et al. Autophagy promotes MHC class II presentation of peptides from intracellular source proteins. Proc. Natl. Acad. Sci. USA 102, 7922–7927 (2005). 20. Nimmerjahn, F. et al. Major histocompatibility complex class II–restricted presentation of a cytosolic antigen by autophagy. Eur. J. Immunol. 33, 1250–1259 (2003). 21. Magalhaes, J.G. et al. Nod2-dependent TH2 polarization of antigen-specific immunity. J. Immunol. 181, 7925–7935 (2008). 22. Birmingham, C.L., Smith, A.C., Bakowski, M.A., Yoshimori, T. & Brumell, J.H. Autophagy controls Salmonella infection in response to damage to the Salmonellacontaining vacuole. J. Biol. Chem. 281, 11374–11383 (2006). 23. Darfeuille-Michaud, A. Adherent-invasive Escherichia coli: a putative new E. coli pathotype associated with Crohn’s disease. Int. J. Med. Microbiol. 292, 185–193 (2002). 24. Darfeuille-Michaud, A. et al. High prevalence of adherent-invasive Escherichia coli associated with ileal mucosa in Crohn’s disease. Gastroenterology 127, 412–421 (2004). 25. Suzuki, T. et al. Differential regulation of caspase-1 activation, pyroptosis and autophagy via Ipaf and ASC in Shigella-infected macrophages. PLoS Pathog. 3, e111 (2007).


8. Schmid, D., Pypaert, M. & Münz, C. Antigen-loading compartments for major histocompatibility complex class II molecules continuously receive input from autophagosomes. Immunity 26, 79–92 (2007). 9. Hampe, J. et al. A genome-wide association scan of nonsynonymous SNPs identifies a susceptibility variant for Crohn disease in ATG16L1. Nat. Genet. 39, 207–211 (2007). 10. Rioux, J.D. et al. Genome-wide association study identifies new susceptibility loci for Crohn disease and implicates autophagy in disease pathogenesis. Nat. Genet. 39, 596–604 (2007). 11. Prescott, N.J. et al. A nonsynonymous SNP in ATG16L1 predisposes to ileal Crohn’s disease and is independent of CARD15 and IBD5. Gastroenterology 132, 1665–1671 (2007). 12. Cummings, J.R. et al. Confirmation of the role of ATG16L1 as a Crohn’s disease susceptibility gene. Inflamm. Bowel Dis. 13, 941–946 (2007). 13. Parkes, M. et al. Sequence variants in the autophagy gene IRGM and multiple other replicating loci contribute to Crohn’s disease susceptibility. Nat. Genet. 39, 830–832 (2007). 14. Ichimura, Y. et al. A ubiquitin-like system mediates protein lipidation. Nature 408, 488–492 (2000). 15. Kabeya, Y. et al. LC3, a mammalian homologue of yeast Apg8p, is localized in autophagosome membranes after processing. EMBO J. 19, 5720–5728 (2000). 16. Pétrilli, V., Dostert, C., Muruve, D.A. & Tschopp, J. The inflammasome: a danger sensing complex triggering innate immunity. Curr. Opin. Immunol. 19, 615–622 (2007).


97

ONLINE METHODS


Dendritic cell isolation and transfection. We obtained buffy coats from the UK National Blood Centre, and we obtained donor samples with informed consent following local ethical guidelines granted by Milton Keynes Research Ethics Committee Ref. 07/H0603/43. We isolated monocytes from Lymphoprep (Axis-Shield) gradient-enriched mononuclear cells by MACS CD14-positive selection (Miltenyi Biotech), and we cultured them together with IL-4 and granulocyte-macrophage colony–stimulating factor (Peprotech) for 4 d before use. We performed transfections with the Amaxa primary cell DC transfection reagent (Amaxa) using an Amaxa nucleofector. Genotyping. We genotyped NOD2 polymorphisms R702W, G908R and 1007fsinsC as follows. We extracted genomic DNA from blood with a Flexigene DNA Extraction Kit (Qiagen) and performed PCR with the following primers: R702W forward 5′-GAATTCCTTCACATCACTTTCCAGT -3′ and reverse 5′-GTCAACTTGAGGTGCCCAACATT-3′; G908R forward 5′CCCAGCTCCTCCCTCTTC-3′ and reverse 5′-AAGTCTGTAATGTAAAGCCAC3′; 1007fsinsC forward 5′-CTGAGCCTTTGTTGATGAGC–3′ and reverse 5′-TCTTCAACCACATCCCCATT-3′. We performed ATG16L1 rs2241880 polymorphism PCR with the primers forward 5′-GGCAGTAGCTGGTACCCTCA-3′ and reverse 5′-CCACAGGTTAGTGTGCAGGA-3′. We purified PCR products with a PCR purification kit (Qiagen) before sequencing. Antibodies, constructs and siRNAs. The GFP-LC3 construct was a kind gift from H. Mellor. TO-PRO-3 nuclear stain was from Invitrogen. We obtained LC3-specific antibodies from Nanotools (clone 5F10), Santa Cruz Biotechnology (MAP LC3 (H-50) sc-28266) and Abgent (clone RB7481); antibody to NOD2 (clone 2D9) from Santa Cruz or Novus Biologicals; antibody to ATG16L1 (R-158-100) from Biosensis; antibodies to HLA-DR and HLA-DM from BD Biosciences; antibody to ATG7 (EP17597) from Abcam and rabbit polyclonal antibody to ATG5 (2630) from Cell Signaling. Secondary conjugates were from Invitrogen. We obtained the following siRNAs from Qiagen: NS control target sequence 5′-AATTCTCCGAACGTGTCACGT-3′ NOD2 target sequence 5′-CAACATGGCCGTGAACTTTAT-3′; NALP3 target sequence 5′-CACGCTAATGATCGACTTCAA-3′; ATG16L1 target sequence 5′CAGGACAATGTGGATACTCAT-3′; ATG5 target sequence 5′-AACCTTTGG CCTAAGAAGAAA-3′; ATG7 target sequence 5′-ATCAGTGGATCTAA ATCTCAA-3′. 3-methyladenine and rapamycin were from Sigma. For immuno blot analyses, we resolved cell lysates by SDS-PAGE and transferred them to polyvinylidene fluoride membranes (Amersham Biosciences), followed by detection with enhanced chemiluminescent reagent (Amersham Biosciences). For LC3 immunoblotting, some cell lysates contained the additional protease inhibitor E64D (Sigma). For both LC3 and NOD2 immunoblotting, lysis buffer contained 2% Triton X-114. T cell functional assays and bacteria. We isolated T cell populations with microbeads coated with antibodies to CD4 (Miltenyi Biotec). For tetanus

nature medicine

toxin–specific T cell proliferation assays, we cultured 1 × 104 autologous CD4+ T cells with 2.5 × 104 DCs after addition of live attenuated S. enterica or live attenuated S. enterica expressing Cfr of tetanus toxin for 2 h. We pulsed cocultures with [3H] after 72 h for the last 16 h of the assay. We cultured GFP–S. enterica typhimurium (kind gift from D. Holden), S. enterica Cfr (SL5338 pTECH1) (kind gift from K. Turner) and adherent-invasive GFP–E. coli to logarithmic phase in LB and ampicillin where appropriate at 37 °C. For bacterial viability assays, we exposed 1 × 105 DCs to E. coli at an MOI of 25 for 1 h and then in gentamicin (50 µg ml−1)-containing medium for the remaining time periods before lysis with 0.5% sodium deoxycholate and quantification of intracellular CFU. We quantified CFU by plating quadruplicates of serial dilutions of lysates on LB agar to quantify the number of live intracellular bacteria. Confocal microscopy. We seeded DCs at 2 × 105 cells per well onto eight-well poly-D-lysine–coated Tissue Culture Slides (BD Biosciences). We allowed the cells to adhere for 2 h before stimulation or starvation in EBSS. Where used, we added Lysotracker (Invitrogen) to the wells for the last 2 h of stimulation at a final concentration of 0.05% (vol/vol). After stimulation, we washed the cells with PBS, fixed them by incubation for 10 min at 37 °C in 4% paraformaldehyde, permeabilized them with 0.1% (vol/vol) Triton X-100 and washed them with PBS containing 2% BSA. We added TO-PRO-3 for 10 min at a 1 in 2,000 concentration. After staining, we mounted cover slips using Vectashield (Vector Laboratories). We used a Radiance 2000 laserscanning confocal microscope for confocal microscopy, followed by analysis with LaserSharp 2000 software (Bio-Rad). We acquired images in sequential scanning mode. Electron microscopy. We washed DCs in ice-cold PBS, pelleted them and fixed them in 4% glutaraldehyde in 0.1 M phosphate buffer (consisting of seven parts disodium hydrogen orthophosphate (14.2 g in 1 l distilled water) and three parts potassium dihydrogen orthophosphate (6.8 g in 500 ml distilled water) generating 0.1 M buffer at pH 7.2). We then post-fixed them in 2% osmium tetroxide, dehydrated them and treated them with propylene oxide before embedding them in Spurr’s epoxy resin (TAAB Laboratories Equipment Ltd.). We cut thin sections and stained them with uranyl acetate and lead citrate before examination in a Jeol 1200EX electron microscope. We quantified autophagy on the basis of three replicates of random counting of 100 DCs and identifying those with one or more autophagic vacuoles. Statistical analyses. We used Prism software (GraphPad) to determine the statistical significance of differences in the means of experimental groups with unpaired, two-tailed Student’s t tests. P values of less than 0.05 were considered significant. Additional methods. Detailed methodology is described in the Supplementary Methods.

doi:10.1038/nm.2069