nature immunology  VOLUME 11 NUMBER 8 AUGUST 2010 701

A rt i c l e s

Innate-like lymphocytes have features of both the innate and adaptive immune systems. In particular, B-1 cells1,2, certain types of γδ T cells3 and natural killer T cells4 express antigen receptors with limited diver-sity. These invariant lymphocyte subsets follow specific ontogenic pathways, home to particular tissues and have larger clonal sizes than do conventional lymphocytes5. These innate cells show immediate effector functions after stimulation, such as antibody production or cytokine secretion2,5.

Mucosal-associated invariant T cells (MAIT cells) are another lymphocyte subset that expresses an evolutionarily conserved invariant T cell antigen receptor (TCR) α-chain, composed of the invariant α-chain variable region 7.2 (iVα7.2) segment in humans and iVα19 in mice associated with the α-chain joining region 33 (Jα33) segment. They are selected on the highly phylogenetically conserved major histocompatibility complex (MHC) class I–related molecule MR1 (α1 and α2 domains 90% homologous between mouse and human, compared with 60–70% for other MHC class I molecules)6,7. MAIT cells are abundant in human blood (1–8% of T cells versus 0.01–1% for natural killer T cells), the intestinal mucosa and mesenteric lymph nodes8. Human MAIT cells have a memory phenotype early in life but are few in number and naive in cord blood, which suggests that MAIT cell populations expand after birth and acquire their memory phenotype in the presence of commensal flora8. In support of that hypothesis, MAIT cells are

not detectable in germ-free mice7. MAIT cells are selected in the thymus independently of the classical MHC molecule processing pathway. In fact, their numbers are greater in mice doubly defi-cient in the transporter associated with antigen processing (TAP; Tap1−/−) and the invariant chain (Ii; Cd74−/−; doubly deficient mice called ‘Tap−/−Ii−/−’ here), in which conventional T cell selection is impaired7,9. Despite the availability of transgenic mice expressing MAIT cell–specific Vα19 and Vβ6 chains8,10,11, little is known about the function of these cells. However, their location in the gut and their considerably conserved features in mice and humans suggest an important role in the response to microbes.

Here we show that in both humans and mice, MAIT cells responded in an MR1-dependent manner to antigen-presenting cells (APCs) cultured with bacteria. This activation was induced by a wide variety of bacteria and yeasts but not by viruses. APCs acquired their stimu-latory ability very quickly after infection, and cognate interaction between the invariant TCR and MR1 was required. Moreover, this interaction had antibacterial potential in vivo, as MR1-deficient mice were more susceptible to infection with Mycobacterium abscessus or Escherichia coli. Finally, in humans, the number of MAIT cells was lower in peripheral blood from patients with infectious diseases such as tuberculosis, in whom we detected them at the site of infection. We conclude that MAIT cells are evolutionarily conserved innate-like T cells with antimicrobial properties.

1Département de Biologie des Tumeurs, Institut Curie, Paris, France. 2Institut National de la Santé et de la Recherche Médicale U932, Paris, France. 3Centre d′Investigation Clinique en Biothérapie 507, Institut Gustave Roussy and Institut Curie, Villejuif and Paris, France. 4Université Pierre et Marie Curie, University of Paris, Institut National de la Santé et de la Recherche Médicale, Unité Mixte de Recherche 945, Hôpital Pitié-Salpêtrière, Laboratoire d’Immunologie Cellulaire et Tissulaire, Service des Maladies Infectieuses et Tropicales, Paris, France. 5Centre National de la Recherche Scientifique, Unité Propre de Service 44, Transgenesis, Archiving and Animal Models, Institut de Transgenèse, Orléans, France. 6Département de Chirurgie, Institut Curie, Paris, France. 7Unité de Virologie et Immunologie Moléculaires, Unité de Rercherche 892, Institut Scientifique de Recherche Agronomique, Domaine de Vilvert, France. 8Department of Pathology and Immunology, Washington University, St. Louis, Missouri, USA. 9Laboratoire de Microbiologie, Hôpital R. Poincaré and Equipe Associé 3647, University Versailles Saint-Quentin, Garches, France. 10These authors contributed equally to this work. Correspondence should be addressed to O.L. (olivier.lantz@curie.net).

Received 25 March; accepted 18 May; published online 27 June 2010; corrected after print 13 August 2010; doi:10.1038/ni.1890

Antimicrobial activity of mucosal-associated invariant T cellsLionel Le Bourhis1,2, Emmanuel Martin1,2, Isabelle Péguillet1,3,10, Amélie Guihot4,10, Nathalie Froux5, Maxime Coré1,2, Eva Lévy1,2, Mathilde Dusseaux1,2, Vanina Meyssonnier4, Virginie Premel1,2, Charlotte Ngo6, Béatrice Riteau7, Livine Duban1,2, Delphine Robert1,3, Shouxiong Huang8, Martin Rottman9, Claire Soudais1,2 & Olivier Lantz1–3

Mucosal-associated invariant T lymphocytes (MAIT lymphocytes) are characterized by two evolutionarily conserved features: an invariant T cell antigen receptor (TCR) -chain and restriction by the major histocompatibility complex (MHC)-related protein MR1. Here we show that MAIT cells were activated by cells infected with various strains of bacteria and yeast, but not cells infected with virus, in both humans and mice. This activation required cognate interaction between the invariant TCR and MR1, which can present a bacteria-derived ligand. In humans, we observed considerably fewer MAIT cells in blood from patients with bacterial infections such as tuberculosis. In the mouse, MAIT cells protected against infection by Mycobacterium abscessus or Escherichia coli. Thus, MAIT cells are evolutionarily conserved innate-like lymphocytes that sense and help fight off microbial infection.

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

702  VOLUME 11 NUMBER 8 AUGUST 2010 nature immunology

A rt i c l e s

RESULTSMR1- and bacteria-dependent activation of human MAIT cellsThe population expansion and acquisition of a memory phenotype by human MAIT cells shortly after birth8 suggested that these cells could respond to bacteria. We isolated monocytes from peripheral blood mononuclear cells and cultured them together with E. coli. We seeded autologous sorted Vα7.2 cells on these APCs at a ratio of 1:1 and cultured them together overnight. Most MAIT cells, which are Vα7.2+CD161+ and either CD8βint cells or CD4−CD8β− double-negative (DN) cells8, were activated, as indicated by CD69 upregula-tion (Fig. 1a). This activation was specific, as neither Vα7.2+CD161− conventional T cells from the same culture (Supplementary Fig. 1a) nor Vα7.2−CD4+ and Vα7.2−CD8+ T cells cultured in parallel (Supplementary Fig. 1b) were activated. MAIT cell activation was modulated by the ratio of bacteria to monocyte and the multiplicity of infection (MOI; Fig. 1b) and was completely abrogated by antibody to MR1 (anti-MR1; Fig. 1c). In contrast to mainstream memory T cells, MAIT cells secreted interferon-γ (IFN-γ; Fig. 1d) but not large amounts of interleukin 2 (IL-2), IL-4, IL-5, IL-13 or IL-17 (data not shown) after being stimulated with E. coli–infected monocytes. Blockade of both CD69 upregulation (Fig. 1c) and IFN-γ secretion (Fig. 1e) by anti-MR1

indicated that cognate interaction between the TCR and the MHC Ib molecules was required for MAIT cell activation.

Fewer MAIT cells in bloods from bacteria-infected patientsThe bacterial reactivity of MAIT cells suggested their involvement in antibacterial defense. We observed significantly smaller proportions and lower absolute numbers (P < 0.001 ) of MAIT cells in the blood of patients with pulmonary bacterial pathologies, including tuberculosis, than in the blood of healthy donors or patients with cancer (Fig. 2a and Supplementary Fig. 2a). The frequency of γδ T cells was unchanged, whereas their absolute numbers were slightly lower than those of healthy controls because of the lymphopenia in some of these patients (Fig. 2b and Supplementary Fig. 2b). Together these results indicate that the lower MAIT cell frequency represented a true decrease in numbers and not dilution by infection-reactive mainstream T cells. One possibility was that after infection, MAIT cells infiltrate tissues, which would decrease their number in blood. To test that hypothesis, we visualized MAIT cells by tissue immunofluorescence with anti-CD3, anti-Vα7.2 and anti-IL-18Rα, the last of which correlates with CD161 expression (Fig. 2c). We observed MAIT cells in lung lesions from two patients infected with Mycobacterium tuberculosis (Fig. 2d).

Vα7.2

CD

161

a

CD69CD8

CD

4

EC

Ctrl

CD8+ + DN

c

CD69

IsoCtrl

α-MR1

CD8+ + DN

b0

0.1

1.0

10

CD69

CD8+ + DN

EC (MOI) d

–

–

110100

103

102

101

CD4+ CD8+ MAIT100

α-CD3α-CD28

IFN

-γ (

pg/m

l)

>103 >103

Memory

EC (MOI)

– +

e

α-MR1

IFN

-γ (

ng/m

l)

5

010

0

MAIT

10

*

EC (MOI)

Figure 1 Human MAIT cells respond to bacteria-infected monocytes in an MR1-dependent manner. (a) Activation of Vα7.2+ sorted cells by E. coli (EC)-infected APCs (far right). Ctrl, control (uninfected). Data are representative of more than five experiments. (b) Activation of human MAIT cells by monocytes cultured together with E. coli (MOI, key). Data are representative of three experiments. (c) Blockade of the activation of sorted Vα7.2+ cells by anti-MR1 (α-MR1; 26.5) or isotype-matched control antibody (Iso). Ctrl, uninfected. Data are representative of three experiments. (d) Production of IFN-γ by flow cytometry–sorted memory (CD45RO+) CD4+ and CD8+ conventional T cells and MAIT (Vα7.2+CD161+) cells (CD3+) after 48 h of culture with autologous monocytes infected with E. coli (MOI, key) or uninfected monocytes (–); T cells stimulated with anti-CD3 and anti-CD28 serve as controls. Data are representative of three experiments (mean and s.e.m. of two different donors). (e) Blockade of IFN-γ production by flow cytometry–sorted MAIT cells as in d with or without blocking anti-MR1 in control conditions (MOI, 0) and after E. coli infection (MOI, 10). *P < 0.01 (paired t-test). Data are representative of three independent experiments (mean and s.e.m.).

MA

IT (

% a

mon

gC

D3+

γδ–

)γδ

+ (%

am

ong

CD

3+)

a100

10

1

0.01

0.1

**

NS

100

10

Health

y

Tuber

culos

is

Pneum

opat

hies

Cance

r

1

0.1

NSNS

NS

Gated onCD3+Vα7.2+ CD8+ + DN

IL-18Rα

c CD3

IL-18Rα

d

Peripheral blood (n = 10)

Ascites: cancer(n = 9)

Ascites: MT(n = 1)

32

e7.8 1.7

b

CD

161

Vα7.2

Vα7.2

CD

161

4.6–11.6 0.1–20.5

Figure 2 Human MAIT cells are less abundant in the blood of bacteria-infected patients. (a,b) Pro-portion of MAIT (Vα7.2+CD161+CD3+TCRγδ−) cells (a) and γδ T cells (b) in the blood of healthy donors or patients (pathology, horizontal axes). Each symbol represents an individual patient. NS, not significant; *P < 0.001 (Mann-Whitney test). (c) Expression of CD161 and IL-18Rα by human T cells gated on CD3+Vα7.2+CD4− cells. (d) Hematoxylin and eosin staining (left) and immunostaining (right) of M. tuberculosis–infected lung tissue. Arrows indicate triple-stained MAIT cells. Original magnification, ×400. (e) Expression of CD161 and Vα7.2 by cells from peripheral blood or ascites fluid (n (above plots), number of patients). Outlined areas indicate CD161+Vα7.2+ gate; numbers adjacent indicate percent cells in gate (above, geometric median; below, range). MT, M. tuberculosis. Data are representative of one experiment per patient (a,b,e) or two experiments (c,d).

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

nature immunology  VOLUME 11 NUMBER 8 AUGUST 2010 703

A rt i c l e s

Similarly, we obtained ascites fluid from patients with suspected ovarian malignancy. The ascites fluid of one patient had a higher MAIT cell frequency than that of blood and other ascites fluid (32% versus 0.1–18.6%; Fig. 2e). Notably, this patient was subsequently diagnosed as having disseminated tuberculosis. These results indicate that human MAIT cells seem to migrate into infected tissues, which suggests that MAIT cells might have antimicrobial functions.

Infected APCs rapidly stimulate mouse MAIT cellsTo investigate the activation of mouse MAIT cells by infected APCs, we cultured bone marrow–derived dendritic cells (BMDCs) from MR1-sufficient mice (called ‘Mr1+ mice’ here) or MR1-deficient (Mr1−/−) mice and cultured them with E. coli. As a source of MAIT cells, we used iVα19-transgenic and iVα19-Vβ6-transgenic TCRα-chain–deficient mice, which have greater frequencies of these cells8. T cells from iVα19-Vβ6-transgenic mice (Fig. 3a) and iVα19-transgenic mice (Fig. 3b) strongly upregulated CD69 expression when cultured with infected Mr1+ BMDCs but not when cultured with Mr1−/− BMDCs. As with human MAIT cells, only CD8+ and DN T cells were activated (Fig. 3a). We also observed MR1-dependent upregulation of CD25 (Fig. 3c) and low secretion of IL-2 by both iVα19 single-transgenic and iVα19-Vβ6 double-transgenic MAIT cells (Fig. 3d), consistent with their low proliferation capacities (Fig. 3e). Peritoneal macrophages cultured together with E. coli were also able to induce MAIT cell activation (data not shown), which indicated that mouse MAIT cells responded to various types of infected APCs in an MR1-dependent manner.

To examine the role of soluble factors in MAIT cell activation, we fixed Mr1+ or Mr1−/− BMDCs with glutaraldehyde at various times after infection, then added iVα19-Vβ6-transgenic T cells. At 3 h after infec-tion, Mr1+ BMDCs were already able to stimulate MAIT cells (Fig. 4a), which indicated that BMDCs acquired the MAIT cell–stimulatory ability very rapidly. This result also demonstrated that soluble factors such as cytokines are not necessary for MAIT cell activation.

To further examine the cellular events required for MAIT cell acti-vation, we treated BMDCs with several drugs that modify antigen

processing, then infected the BMDCs. Treatment of the BMDCs with dynasore or cytochalasin D (inhibitors of endocytosis and/or phagocytosis) or chloroquine (inhibitor endosomal acidification) strongly inhibited MAIT cell activation (Fig. 4b), which indicated that bacteria or bacteria-derived compounds need to be internal-ized and processed. The proteasome inhibitor lactacystin did not show any effect (Fig. 4b), although it prevented the activation of ovalbumin-reactive OT-I TCR–transgenic cells (data not shown). Additionally, infected BMDCs derived from Tap−/−Ii−/− mice acti-vated MAIT cells as well as wild-type Tap+/+Ii+/+ cells did (Fig. 4c). Together these results indicate that activation of MAIT cells by BMDCs is independent of the conventional MHC class I and class II antigen-presentation pathways.

Conserved interactions between invariant TCR and MR1To study the evolutionary conservation of the TCR-MR1 interactions, we assessed whether human MAIT cells were activated by infected mouse APCs and vice versa. Sorted human MAIT cells were acti-vated by E. coli–infected APCs from Mr1+ mice but not by those from Mr1−/− mice (Fig. 4d). Similarly, mouse hybridomas (6C2 and 8D12) expressing the iVα19-Vβ8 chains were stimulated by infected human monocytes, as indicated by IL-2 secretion, and this was completely abrogated by the addition of anti-MR1 (Fig. 4e). These results indicate that mouse and human MAIT cell TCRs interact similarly with MR1 on infected APCs.

Looking at the TCR repertoire involved in MR1 reactivity, we observed that in iVα19 single-transgenic mice, most CD8+ T cells and DN T cells expressing the Vβ6 or Vβ8 segments (known to be ‘preferentially’ found in MAIT cells8) responded to infected Mr1+ BMDCs but not to infected Mr1−/− BMDCs (Fig. 3b). In contrast, 80% of the CD8+ T cells and DN T cells expressing neither of these Vβ segments did not respond (Fig. 3b), which excluded the possibility of an as-yet-unknown iVα19-binding superantigen. Additionally, MAIT cells did not require selection by MR1 but required only expression of permissive Vβ segments in asso-ciation with the iVα19 TCRα chain. Indeed, T cells from iVα19-Vβ6

a c

CD69

CD

25

Mr1+

Mr1–/–

d

681

9

CD8

CD

4

TCRβ

Vβ6

CD69

0

0.2

0.1

iVα19-Vβ6 TG

Mr1+ Mr1–/–

*

0

0.2

0.1

iVα19 TG

Vβ8V

β6CD69

Mr1+ Mr1–/–

CD8

CD

4

Vβ6+

EC (MOI)

EC (MOI)

b

Mr1+ Mr1–/–

Day 2

Day 5

EC (MOI) :

CFSE

CD

25

e

iVα19-Vβ6 TG

CD8+

CD4+

DN

Mr1+ Mr1–/–

iVα 19 TG

0 10 0 10

Vβ8+

Vβ6–

Vβ8–

100

IL-2

(ng

/ml)

EC (MOI) :

0

*

100 1

100 1

Figure 3 T cells from mice transgenic for the specific MAIT TCR Vα and/or Vβ chains respond to E. coli–infected APCs through MR1. (a) Activation of mouse iVα19-Vβ6-transgenic (TG) MAIT cells by uninfected (grey histograms) or E. coli–infected (black lines) Mr1+ or Mr1−/− BMDCs. Data are representative of more than five experiments. (b) Activation of iVα19-transgenic T cells by uninfected BMDCs (gray histograms) or E. coli–infected BMDCs (black lines), gated on Vβ6+, Vβ8+ or Vβ6−Vβ8− T cells (left margin). Data are representative of more than five experiments. (c) T cells (as in a) costained for CD25 and CD69 after overnight coculture with uninfected (0) or E. coli–infected (10) Mr1+ or Mr1−/− BMDCs. Numbers in plots indicate percent CD25+CD69+ cells. Data are representative of more than five experiments. (d) IL-2 production by MAIT cells after stimulation by E. coli–infected Mr1+ or Mr1−/− BMDCs. *P < 0.05 (analysis of variance). Data are representative of two experiments (mean ± s.e.m.). (e) Activation of iVα19-Vβ6 bead-sorted MAIT cells stained with the cytosolic dye CFSE and stimulated as in a, assayed as CD25 upregulation and CFSE dilution (proliferation) at days 2 and 5. Data are representative of two experiments.

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

704  VOLUME 11 NUMBER 8 AUGUST 2010 nature immunology

A rt i c l e s

double-transgenic or iVα19 single-transgenic MR1-deficient mice were activated by E. coli–infected BMDCs (Supplementary Fig. 3a,b). Finally, 10% of DN T cells from Tap−/−Ii−/− Vβ6 single-transgenic mice were activated by infected BMDCs in an MR1-dependent man-ner (Supplementary Fig. 3c), in line with published quantification of iVα19 by quantitative RT-PCR8. Thus, bacteria-infected APCs activate all mouse MAIT cells through MR1, regardless of the fine specificity imparted by the TCRβ chain, which indicates that expression of the iVα19 TCRα chain together with a permissive Vβ chain is sufficient for recognition of MR1 on infected APCs.

MAIT cells sense a wide variety of microbesThe results presented above suggested extreme conservation of the bacterial component(s) involved in MAIT cell activation. To begin this characterization, we used heat-killed or paraformaldehyde (PFA)-fixed bacteria. Live bacteria were more efficient than dead microbes in inducing a MAIT cell response (Fig. 5a), but PFA-treated bacteria

were able to trigger better activation than were heat-killed organisms (Fig. 5a), which indicated that the compound responsible for MAIT activation was partially heat labile but was preserved by fixation.

We then assessed whether infecting BMDCs with one specific class of bacteria induced MAIT cell activation. The Gram-negative bacilli Pseudomonas aeroginosa and Klebsiella pneumoniae, Gram-positive Lactobacillus acidophilus and Gram-positive cocci Staphylococcus aureus and Staphylococcus epidermidis all induced a strong response in both iVα19-Vβ6-transgenic and iVα19-transgenic MAIT cells (Fig. 5b and data not shown, respectively). Unexpectedly, Enterococcus faecalis and Streptococcus group A, two bacteria closely related to staphylococci, did not induce MAIT cell activation, even though all bacteria induced effec-tive BMDC activation (data not shown). We obtained similar patterns for the activation of human MAIT cells by monocytes for the same panel of bacteria (data not shown). As streptococci and staphylococci share many microbe-associated molecular patterns12,13, these results suggest that some bacteria provide the necessary trigger for MAIT cell response,

WT

CD69

ca

CD69

Mr1+

Mr1–/– Tap–/–Ii–/–

10

d

0.6

0

IL-2

(ng

/ml)

0.4

0.2

6C2BW

α-MR1 – + – + – +

100 1EC (MOI)

50

25

0

mMr1–/–mMr1+hMono

10

e

b50

0

25

Ctrl

Chloro

DMSO

Dyn

CytoD

Lacta

–EC30

0

15*

**

NS

Time (min) : 30 120 180

CD

69+ c

ells

(%

)

CD

69+ c

ells

(%

)

EC (MOl)

Mr1+ Mr1–/–

CD

69+ c

ells

(%

)

100

8D12

Figure 4 MAIT cell activation requires a highly conserved cognate interaction between MR1 and the TCR, as well as internalization of exogenous compounds, but is independent of the classical MHC class I and class II presentation pathways. (a) Activation of iVα19-Vβ6-transgenic MAIT cells by uninfected (grey histograms) or E. coli–infected (black lines) Mr1+ or Mr1−/− BMDCs fixed at various time points after infection (above plots) before overnight coculture. (b) Activation of MAIT cells by BMDCs left untreated (dimethylsulfoxide (DMSO)) or treated with cloroquine buffer alone (control (Ctrl)), chloroquine (Chloro), dynasore (Dyna), cytochalasin D (CytoD) or lactacystin (Lacta) and then left uninfected (–) or infected with E. coli. *P < 0.05 (Mann-Whitney test). (c) Activation of T cells by BMDCs from wild-type (WT) or Tap−/−Ii−/− mice on an Mr1+

a 4

3

2

1

0

0 1 10 102 103 0 1 10 102 103 0 1 10 102 103

CD

69 (

MF

I × 1

03 )

CD

69 (

MF

I × 1

03 ) Live

PFA

HK

MOI

Uninf bEC

PA

KP

LA

Uninf

SASE

Uninf2

1

0

2

1

0

MOI

c

MOI

CA

CG

Uninf

2

1

0

EFSGA

SC

3

EC

101

dMOI

2

5

1

4

100

50

0

CD

69+C

D25

+ (

%)

3

+ ++ ++ +++

ECEF

MAIT activation –

51 2 3 4

EF

0

0

CD

69 (

MF

I × 1

03 )

0 1 10 102 103

Figure 5 MAIT cells respond to phylogenetically diverse microbes. (a) Activation of iVα19-Vβ6-transgenic T cells by uninfected control BMDCs (Uninf) or by BMDCs cultured with live (Live), PFA-treated (PFA) or heat-killed (HK) E. coli. MFI, mean fluorescent intensity. (b) Activation of iVα19-Vβ6-transgenic T cells by BMDCs infected with E. coli (EC), P. aeroginosa (PA), K. pneumoniae (KP), L. acidophilus (LA), Streptococcus group A (SGA), E. faecalis (EF), S. aureus (SA), S. epidermidis (SE) or by uninfected control BMDCs (Uninf). (c) Activation of iVα19-Vβ6-transgenic T cells by wild-type BMDCs infected with mixed suspensions of PFA-fixed bacteria at various ratios of E. coli to E. faecalis (wedges; total MOI constant at 1 or 10), assessed as upregulation of CD69 and CD25. Right, summary of data at left (1–5 above bars corresponds to 1–5 above data points at left); minus and plus symbols below indicate degree of MAIT cell activation. (d) Activation of iVα19-Vβ6-transgenic T cells by BMDCs left uninfected (Uninf; control) or infected with C. albicans (CA), C. glabrata (CG) or S. cerevisiae (SC) and fixed 4 h after infection, assessed after overnight stimulation. Data are representative of two (a) or three (b–d) experiments.

or Mr1−/−background, infected with E. coli (black lines) or left uninfected (gray histograms) before overnight coculture. (d) Activation of sorted human MAIT cells seeded on human monocytes (hMono) or mouse Mr1+ (mMr1+) or Mr1−/− (mMr1−/−) macrophages infected with E. coli (MOI, horizontal axis), followed by overnight coculture, assessed by CD69 staining. (e) Activation of hybridomas expressing mouse MAIT–specific TCRs (responding cells) by human monocytes infected as in a, assessed by IL-2 secretion. Data are representative of two (a,c–e) or three (b) experiments (error bars (b), s.e.m.).

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

nature immunology  VOLUME 11 NUMBER 8 AUGUST 2010 705

A rt i c l e s

whereas others do not. Alternatively, E. faecalis and Streptococcus could actively inhibit the presentation of MR1 and its ligand. However, this hypothesis was refuted by the findings that PFA-fixed E. faecalis still did not induce MAIT cell activation (Fig. 5c) and that infection with a mixed suspension of PFA-killed E. coli and E. faecalis at a ratio of 1:10 still did (Fig. 5c). In these experiments, we fixed bacteria with PFA to avoid differences in growth rates. These data suggest that E. faecalis lacks the necessary compound for MR1-dependent MAIT activation.

To address the specificity of MAIT cell responses to bacteria rela-tive to that of other microbes, we tested viruses and yeasts in our in vitro model. BMDCs infected with Saccharomyces cerevisiae, Candida glabrata or Candida albicans were able to induce a strong MAIT response in an MR1-dependent manner (Fig. 5d). However, several unrelated viruses (including encephalomyocarditis virus, Sendai virus, Newcastle disease virus, herpes simplex virus and parain-fluenza 3 virus), despite their ability to activate BMDCs, failed to acti-vate MAIT cells (Supplementary Fig. 4). These results suggest that MAIT cells respond ‘preferentially’ to bacteria and yeasts.

MAIT cells respond to a conserved exogenous antigenAll the microbes assessed above are able to stimulate APCs by innate receptors such as Toll-like receptors (TLRs) and Nod-like receptors through the recognition of common microbe-associated molecular patterns. We tested whether TLR ligands could recapitulate the bacteria-driven activation of MAIT cells. Despite efficient activation of the BMDCs by Pam3CSK4, poly(I:C), lipolysaccharide (LPS) or CpG (ligands of TLR2, TLR3, TLR4 or TLR9, respectively14; Supplementary Fig. 5a), we detected no MR1-dependent activation on iVα19-Vβ6-transgenic MAIT cells (Supplementary Fig. 5b). We obtained similar results with human monocytes and autologous MAIT cells (data not shown). We then used BMDCs singly or doubly deficient in the adap-tors MyD88 and TRIF, which are impaired in their response to many TLR ligands and IL-1β15–17, to study the response to whole bacteria. We found that iVα19-Vβ6-transgenic MAIT cells were still receptive to stimulation by infected BMDCs of all three genotypes, although MyD88-deficient BMDCs were less efficient than the wild-type cells (Supplementary Fig. 6a). This result could be explained by impaired cellular activation of the BMDCs, as indicated by lack of upregulation of the maturation marker CD40 (Supplementary Fig. 6b). Several other innate pathways were also not involved, as MAIT cell activation was unaffected when we used BMDCs from mice deficient in intra-cellular bacteria sensors Nod1 and Nod2 (refs. 18,19), the cytoplasmic receptor NLRP3 and adaptor ASC20–22 or the signaling adaptor IPS-1 (MAVS)23–25 (Supplementary Fig. 6a,b). These results indicate that DC activation through the main microbe-recognition pathways is not sufficient for MAIT cell activation.

Both human MAIT cells (Fig. 1a) and mouse iVα19-transgenic MAIT cells (Fig. 2b) were activated by autologous E. coli–infected APCs. However, published reports have shown that these cells do not react to transduced cell lines with high surface expression of human or mouse MR1 (refs. 26,27). This discrepancy suggests that a ligand presented by MR1 is provided during the APC-bacteria coculture and is necessary for MAIT activation. Therefore, we assessed the capacity of cell lines to activate MAIT cells after bac-terial coculture. WT3 mouse fibroblasts cultured together with E. coli, which does not invade this cell type, activated purified iVα19-transgenic T cells and iVα19-Vβ6-transgenic T cells (Fig. 6a and Supplementary Fig. 7a, respectively). This activation was

– +

WT3mWT3m

WT3WT3

0102

103

–

20

40

0

CD

69+

CD

25+ (

%)

α-MR1

a

*

*

+

10

5

0

** *

b EC (MOI)Live cells Fixed cells

Figure 6 Fibroblastic cells can induce MAIT cell activation through MR1, which can present an exogenous bacteria–derived ligand. (a) Activation of iVα19-transgenic MAIT cells by WT3 cells or WT3-mMR1 cells (WT3m) cultured with E. coli (MOI, key), which serve as APCs, assessed as upregulation of CD69 and CD25 in the presence or absence of blocking antibody to MR1. (b) Activation of iVα19-transgenic T cells seeded on WT3 and WT3-mMR1 cells fixed by glutharaldehyde before coculture for 3 h with E. coli (MOI, key), assessed as upregulation of CD69 and CD25 after overnight culture. *P < 0.05 (paired t-test). Data are pooled from four independent experiments (mean and s.e.m.).

Figure 7 Bacteria-dependent population expansion of MAIT cells, activation at the site of infection and antibacterial function. (a) Analysis of iVα19 mRNA in the mesenteric lymph nodes of Mr1+ or Mr1−/− mice on a Tap−/−Ii−/− background; germ-free Mr1+Tap−/−Ii−/− mice were reconstituted with E. cloacae (EClo), L. acidophilus or E. faecalis. Results are presented in arbitrary units (AU) relative to to the expression of TCRα constant region mRNA. Symbols represent individual mice. *P < 0.05, and **P < 0.01 (Student′s t-test). Data are pooled from eight experiments with individual mice analyzed in different batches (n = 4–6 per group). (b) Bacterial counts (colony-forming units (CFU)) in the spleens of iVα19-transgenic or Vβ6-transgenic mice on an Mr1+ or Mr1−/− background after intraperitoneal injection of 1 × 107 colony-forming units of E. coli. Symbols represent individual mice; small horizontal lines indicate the median. *P < 0.05, and **P < 0.01 (Mann-Whitney test). Data are pooled from two independent experiments (n = 8–16 mice per group). (c) Total T cells, Vβ6+ or Vβ8+ cells and Vβ6−Vβ8− cells in the peritoneal lavage fluid of E. coli–infected or uninfected Mr1+ and Mr1−/− mice. Symbols represent individual mice; small horizontal lines indicate the mean. *P < 0.01 (Mann-Whitney test). Data are representative of three independent experiments (n = 3–4 mice per group). (d) Proportion of activated (CD25+) cells among Vβ6+ or Vβ8+ cells or Vβ6−Vβ8− cells in the peritoneal lavage fluid of uninfected or infected Mr1+ or Mr1−/− mice. Symbols represent individual mice; small horizontal lines indicate the mean. *P < 0.01 (Mann-Whitney test). Data are representative of two independent experiments (n = 2–12 mice per group).

Germ-free

– ECloMr1+ Mr1–/– Mr1+

Mr1+ (EC)

Mr1–/–

Mr1–/– (EC)

Mr1+ (EC)

Mr1–/– (EC)

Mr1+ Mr1–/–

3

2

1

4

iVα1

9 m

RN

A (

log

AU

)

***

**

a

4

3

2

5

b

iVα19 TG Vα6 TG

Vβ6+ or Vβ8+Vβ6+ orVβ8+

Vβ6–Vβ8–Vβ6–Vβ8–

0E. c

oli (

CF

U (

log 10

))

1

***

6

c30

20

10

0

CD

25+ (

%)

**

Uninf

2

1

0

T c

ells

(×1

05 )

*

Total

*

* **

d

EF

3 Uninf

LA

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

706  VOLUME 11 NUMBER 8 AUGUST 2010 nature immunology

A rt i c l e s

substantially greater with WT3 cells overexpressing mouse MR1 (WT3-mMR1 cells) and was completely blocked by anti-MR1, which indicated that this response depended on MR1 expression by the WT3 cells. We obtained similar results with human MAIT cells stimulated by HeLa human cervical cancer cells expressing human MR1 (Supplementary Fig. 7c). To test for the presence of an exogenous ligand, we fixed WT3 and WT3-mMR1 cells with glutaraldehyde before bacterial coculture so that no processing and/or upregulation of an endogenous ligand could occur. We observed that a substantial proportion of iVα19-transgenic and iVα19-Vβ6-transgenic MAIT cells were activated by WT3-mMR1 cells but not by WT3 cells (Fig. 6b and Supplementary Fig. 7b). The activation of iVα19-transgenic cells by WT3-mMR1 cells, even when fixed, in the presence of bacteria, indicated that MAIT cells recognize an exogenous ligand presented by MR1.

Antibacterial function of MAIT cells in vivoThe reactivity of MAIT cells to cells cultured together with bacteria, the acquisition of a memory phenotype after birth in humans and their absence in germ-free mice7 suggest that colonization by the commen-sal flora induces population expansion of these T cells in vivo. To test that hypothesis, we reconstituted germ-free mice with monomicro-bial flora. Consistent with published findings7, the mesenteric lymph nodes of Tap−/−Ii−/−Mr1+ germ-free mice and Tap−/−Ii−/−Mr1−/− mice had similar amounts of iVα19-Jα33 mRNA (Fig. 7a). Feeding germ-free Tap−/−Ii−/− mice a single strain of bacteria (Enterobacter cloacae, Lactobacillus casei, Bacteroides thetaiotaomicron or Bifidobacterium animalis) increased iVα19-Jα33 mRNA in the mesenteric lymph nodes to the amount found in conventional Tap−/−Ii−/−Mr1+ mice (Fig. 7a and data not shown), which demonstrated that single microbial flora are sufficient to induce MAIT cell population expansion. In contrast, feeding mice E. faecalis failed to generate a detectable MAIT popula-tion (Fig. 7a), which further suggested that this bacterium does not provide the ligand necessary for MAIT cell activation.

We then determined whether MAIT cells have antibacterial prop-erties in vivo. We monitored MAIT cell activation and E. coli clear-ance after intraperitoneal injection of 1 × 107 colony-forming units of E. coli. Both iVα19 and Vβ6-transgenic Mr1−/− mice had more colony-forming units in the spleen 3 d after infection than did their Mr1+ counterparts (Fig. 7b). The iVα19-transgenic mice showed better pro-tection than did Vβ6-transgenic mice, which correlated with more MAIT cells8. T cells in iVα19-transgenic Mr1+ mice accumulated in the peritoneal cavity (Fig. 7c) and upregulated their surface expression of CD25 (Fig. 7d). Moreover, on an Mr1+ background, T cells expressing the Vβ6 segment or the Vβ8 segment had higher CD25 expression than did cells that expressed neither the Vβ6 segment nor the Vβ8 segment (Fig. 7d), which further suggested activation is MAIT cell specific. Furthermore, consistent with our in vitro data (Supplementary Fig. 5), MR1-deficient mice were not more sensitive to influenza virus infec-tion than were wild-type mice (data not shown). Together these results indicate that MAIT cells are recruited and activated at the site of infec-tion and have antibacterial properties.

We next examined the role of MAIT cells in a mycobacterium infection model using M. abscessus28. M. abscessus–infected BMDCs strongly stimulated mouse MAIT cells in a dose- and MR1-dependent manner (Fig. 8a), whereas Mr1+ and Mr1−/− BMDCs were activated similarly by the bacteria (data not shown). We injected 1 × 106 bacteria intravenously into nontransgenic, iVα19-transgenic or Vβ6-transgenic mice, as well as into Tap−/−Ii−/− and Cd3−/− mice on an MR1-sufficient or MR1-deficient background. We found that iVα19-transgenic and Vβ6-transgenic Mr1−/− mice had a higher bacterial load in the spleen 15 d after infection than did their Mr1+ littermates (Fig. 8b). We found no significant difference between nontransgenic Mr1+ and Mr1−/− mice. Tap−/−Ii−/− mice (which have impaired conventional T cells and more MAIT cells) had higher bacterial burdens than did wild-type mice, but even more so in the absence of MR1, which suggested that the greater proportion of MAIT cells found in these mice led to some protection. Similarly, Cd3−/− mice had an impaired response to M. abscessus relative to that of wild-type mice; however, we found no difference between Mr1+ and Mr1−/− mice of this genotype, which confirmed the idea that the protection observed was mediated by T cells. Together these data indicate an important role for MAIT cell–mediated immune responses to mycobacterium infection.

DISCUSSIONMAIT cells are highly conserved between species and are very abun-dant in humans, but no function has yet been described for these cells. We have now shown that human and mouse MAIT cells responded to bacteria- or yeast-infected APCs in an MR1-dependent manner. MAIT cells reacted to APCs cultured together with various bacteria, including enterobacteriacae, staphylococci and Mycobacterium, but not to those cultured with streptococci or E. faecalis. Activation was independent of the classical MHC class I and II pathways and of the main innate immune receptors. In mice, MAIT cell activation medi-ated antibacterial functions in two different infection models. Finally, MAIT cell frequencies were much lower in the blood of bacteria-infected patients. This observation can be explained by the migration of these cells into infected tissues, which suggests that as in mice, human MAIT cells have a role in the antibacterial response.

The acquisition of a memory phenotype by human MAIT cells after birth suggested that the commensal flora induces MAIT cell maturation and population expansion. In mice, MAIT cells are naive and are present in small numbers8, which could result from a missing genetic element arising from the ‘genetic bottleneck’ of laboratory

CF

U (

log1

0)

5

3

4

iVα19 TG Vβ6 TG WTB6

Mr1+B6

Mr1–/–

Mr1+

Mr1–/–

b

Tap–/–li–/– Cd3–/–

*

*

*

CD

69+ (

%)

00 1 10 100

40

80a

CD69M. abscessus (MOI)

Mr1+ Mr1–/–

Figure 8 Antimycobacterial activity of MAIT cells. (a) Activation of MAIT cells by Mr1+ or Mr1−/− BMDCs infected with M. abscessus (MOI, horizontal axis, right), assessed as CD69 upregulation (left) and frequency of CD69+ MAIT cells among all MAIT cells (right) after overnight culture. Data are representative of five experiments. (b) M. abscessus in the spleen of iVα19-transgenic, Vβ6-transgenic, wild-type, Tap−/−Ii−/− or Cd3−/− mice on an Mr1+ or Mr1−/− background, as well as in C57BL/6 controls (B6), 15 d after intravenous injection of 1 × 106 M. abscessus. Symbols represent individual mice; small horizontal lines indicate the median. *P < 0.05 (Mann-Whitney test). Data are pooled from two independent experiments.

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

nature immunology  VOLUME 11 NUMBER 8 AUGUST 2010 707

A rt i c l e s

mice or from the cleanliness of mouse facilities. We have shown that MAIT cells were absent from germ-free mice but were recovered after colonization with bacteria, which indicated an interaction between MAIT cells and the commensal flora. Hence, the gut flora could drive the survival or population expansion of MAIT cells directly by MR1-dependent interactions or indirectly through the constitution of trophic niches. The finding that human and mouse MAIT cells were activated by ubiquitous bacteria does not exclude the possibil-ity that in vivo, the population expansion of MAIT cells could be driven by one particular undefined microbe with special properties, as observed in the case of Bacteroides fragilis29 or segmented filamen-tous bacteria30,31.

The wide specificity of MAIT cells for phylogenetically distant microbial entities and the observed cross-reactivity between humans and mice suggest that MAIT cells recognize a conserved antigen. Human MAIT cells did not react to cells overexpressing human MR1 (ref. 26), which suggested that a ligand might have been missing from the MR1 groove. We have now shown that human MAIT cells recognized MR1 expressed on APCs or fibroblasts overexpressing human MR1, cultured together with bacteria. Similarly, Vβ6+ and Vβ8+ T cells from iVα19 single-transgenic mouse recognized mMR1 on infected mouse APCs but not on uninfected MR1-transduced cell lines. Together these results suggest that human MR1 and mouse MR1 present a ligand that is either provided or induced by the bacteria.

Characterization of the ligand presented by MR1 is the next main challenge in MAIT cell research. Published reports have proposed that MAIT cells are activated by the lipidic compound α-mannosyl- ceramide presented by MR1 (ref. 32). We were unable to reproduce the data obtained with that compound in our experimental settings (data not shown). We also failed to find any effect of N-butyldeoxynojirimy-cin, a drug that inhibits glycolipid biosynthesis33 (data not shown), which challenged the possibility of a lipidic ligand loaded onto MR1. Even though the human MAIT cell population is oligoclonal8, most of these cells are activated by various types of bacteria and yeasts, which suggests the absence of fine TCR specificity toward various unrelated microbes. One possible explanation would be that all these microbes induce upregulation of an endogenous ligand, which might also be necessary for thymic selection. However, our observation that fixed cells expressing MR1 activated MAIT cells in the presence of bacteria indicates the existence of an exogenous antigen. This ligand could be of multiple compositions, but the MR1–invariant TCR interaction would be nondiscriminative, or it could be an extremely conserved compound among microbes.

The induction of immunological responses by a highly conserved microbe-derived product is reminiscent of the ligands of the innate immune system34,35. It was therefore unexpected that none of the tested cells deficient in the main innate immune pathways (MyD88, TRIF, Nod1 and/or Nod2, NLRP3 or ASC, or IPS-1) had a substan-tial effect on MAIT cell activation. Also, pure TLR ligands did not recapitulate the activation triggered by whole bacteria. Thus, cellular activation alone cannot trigger the APC to become competent to activate MAIT cells. These results contrast with the activation of natural killer T cells, in which ligands for TLR9 and TLR4 induce the upregulation of an endogenous ligand loaded on the antigen- presenting molecule CD1d36,37.

B cells have been linked to MAIT cell population expansion in the periphery7,8. However, so far there are no satisfactory results showing direct activation of MAIT cells by B cells. Further experi-ments are needed to formally address this question. Nonetheless, we have shown that MAIT cells can be activated by various types of cells (dendritic cells, macrophages, epithelial and fibroblastic cells), which

suggests that the pathways necessary for MR1 processing and loading are broadly expressed. The transporter TAP and invariant chain Ii, essential participants in the MHC class I and class II pathways, are not required for MAIT cell ontogeny7,9. In this study, we have estab-lished that they are also not necessary for MR1-dependent activa-tion. Studies have reported that some exogenous antigen processing could occur in the absence of TAP, a process called ‘vesicular process-ing’38,39. MR1 could use such a pathway to stimulate MAIT cells after bacterial triggering. Our results showing that fixed cells overexpress-ing MR1 activated MAIT cells suggest that a bacterial product can be directly loaded onto the MR1 molecule without processing. However, such circumstances are nonphysiological because MR1 expression at the cell surface has not been detected in nontransduced settings26. Accordingly, loading of MR1 with the putative ligand could occur in a specific compartment of the endosomal pathway. The microbial trigger could induce the production and trafficking of MR1 through this compartment en route to the cell membrane, similar to what has been described for the MHC class Ib molecule H-2M3 (refs. 40,41).

MAIT cells have an antibacterial function that could be attributed to the production of IFN-γ42. Their IFN-γ production is moderate compared with that of conventional memory T cells stimulated with anti-CD3 and anti-CD28. However, the absence of activation of con-ventional CD4+ and CD8+ T cells with infected APCs is striking. In mice, larger numbers of MAIT cells correlate with greater protection against infection of E. coli. In the M. abscessus infection model, we observed no substantial difference between nontransgenic Mr1+ and Mr1−/−mice with low MAIT cell frequency, whereas we did find a dif-ference between iVα19-transgenic and Vβ6-transgenic mice, which have more MAIT cells, closer to what is observed in humans. We speculate that MAIT cell function resides in the prompt response of these cells at the site of infection. MAIT cell ontogeny and the ability of these cells to respond to a wide variety of bacteria, includ-ing commensals in the gut, might suggest a dual role: in the defense against infection, as shown in our study here, and possibly in mucosal homeostatic mechanisms.

The results presented here have shown that MAIT cells respond to microbes and participate in antibacterial immune responses in both humans and mice. MAIT cells represent an evolutionarily conserved innate-like lymphocyte population that senses and participates in immune responses to microbes. MAIT cells represent the first T cell population identified, to our knowledge, that distinguishes between microbial and viral challenges. Given the abundance of this cell type in humans, their wide microbial specificity and their protective capacity, the manipulation of MAIT cells could have a considerable effect on the development of vaccines and therapeutic drugs for infectious diseases.

METHODSMethods and any associated references are available in the online version of the paper at http://www.nature.com/natureimmunology/.

Note: Supplementary information is available on the Nature Immunology website.

ACkNOwLEDGMENtSWe thank T. Hansen (Washington University in St. Louis) for anti-MR1 and cell lines overexpressing MR1; J. Cadranel (Tenon Hospital) for lung biopsies; M. Chignard (Institut Pasteur) for MyD88-deficient bone marrow; L. Alexopoulou (Centre d’Immunologie de Marseille-Luminy) for TRIF-deficient bone marrow; D. Philpott (University of Toronto) for Nod1-Nod2–deficient bone marrow; J. Tschopp (University of Lausanne) for NLRP3-ASC–deficient bone marrow; M. Albert (Institut Pasteur) for IPS-1-deficient bone marrow; the microbiology department of the Curie Institute for bacterial strains; V. Sancho-Shimizu (Faculté Necker–Institut National de la Santé et de la Recherche Médicale U550) for viruses; M. Garcia, C. Billerit, S. Boissel and I. Grandjean for managing the

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

708  VOLUME 11 NUMBER 8 AUGUST 2010 nature immunology

mouse colonies; Z. Maciorowski and A. Viguier for cell sorting; and L. Guerri, V. Soumelis and S. Amigorena for critical reading of the manuscript. Supported by Institut National de la Santé et de la Recherche Médicale, Centre d′Investigation Clinique en Biothérapie 507 Institut National de la Santé et de la Recherche Médicale and Agence National de Recherches.

AUtHOR CONtRIBUtIONSL.L.B., E.M., M.C., M.D., E.L. and C.S. did most experiments; M.R. provided M. abscessus and did M. abscessus experiments with C.S.; A.G., V.M. and C.N. provided patient samples; I.P., D.R. and V.P. analyzed patient samples; N.F. managed germ-free mice and did the reconstitution experiments with L.D. and E.M.; B.R. did in vivo influenza experiments; L.L.B., E.M., C.S. and O.L. designed experiments; and L.L.B., C.S. and O.L. wrote the manuscript.

COMPEtING FINANCIAL INtEREStSThe authors declare competing financial interests: details accompany the full-text HTML version of the paper at http://www.nature.com/natureimmunology/.

Published online at http://www.nature.com/natureimmunology/. reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions/.

1. Kearney, J.F. Innate-like B cells. Springer Semin. Immunopathol. 26, 377–383 (2005).

2. Montecino-Rodriguez, E. & Dorshkind, K. New perspectives in B-1 B cell development and function. Trends Immunol. 27, 428–433 (2006).

3. Hayday, A. & Tigelaar, R. Immunoregulation in the tissues by γδ T cells. Nat. Rev. Immunol. 3, 233–242 (2003).

4. Bendelac, A., Savage, P.B. & Teyton, L. The biology of NKT cells. Annu. Rev. Immunol. 25, 297–336 (2007).

5. Bendelac, A., Bonneville, M. & Kearney, J.F. Autoreactivity by design: innate B and T lymphocytes. Nat. Rev. Immunol. 1, 177–186 (2001).

6. Riegert, P., Wanner, V. & Bahram, S. Genomics, isoforms, expression, and phylogeny of the MHC class I-related MR1 gene. J. Immunol. 161, 4066–4077 (1998).

7. Treiner, E. et al. Selection of evolutionarily conserved mucosal-associated invariant T cells by MR1. Nature 422, 164–169 (2003).

8. Martin, E. et al. Stepwise development of MAIT cells in mouse and human. PLoS Biol. 7, e54 (2009).

9. Tilloy, F. et al. An invariant T cell receptor alpha chain defines a novel TAP-independent major histocompatibility complex class Ib-restricted α/β T cell subpopulation in mammals. J. Exp. Med. 189, 1907–1921 (1999).

10. Kawachi, I., Maldonado, J., Strader, C. & Gilfillan, S. MR1-restricted V alpha 19i mucosal-associated invariant T cells are innate T cells in the gut lamina propria that provide a rapid and diverse cytokine response. J. Immunol. 176, 1618–1627 (2006).

11. Croxford, J.L., Miyake, S., Huang, Y.Y., Shimamura, M. & Yamamura, T. Invariant V(alpha)19i T cells regulate autoimmune inflammation. Nat. Immunol. 7, 987–994 (2006).

12. Santos-Sierra, S., Golenbock, D.T. & Henneke, P. Toll-like receptor-dependent discrimination of streptococci. J. Endotoxin Res. 12, 307–312 (2006).

13. Texereau, J. et al. The importance of Toll-like receptor 2 polymorphisms in severe infections. Clin. Infect. Dis. 41 Suppl 7, S408–S415 (2005).

14. Kawai, T. & Akira, S. TLR signaling. Semin. Immunol. 19, 24–32 (2007).15. Adachi, O. et al. Targeted disruption of the MyD88 gene results in loss of IL-1- and

IL-18-mediated function. Immunity 9, 143–150 (1998).16. Zhang, F.X. et al. Bacterial lipopolysaccharide activates nuclear factor-κB through

interleukin-1 signaling mediators in cultured human dermal endothelial cells and mononuclear phagocytes. J. Biol. Chem. 274, 7611–7614 (1999).

17. Hoebe, K. et al. Identification of Lps2 as a key transducer of MyD88-independent TIR signalling. Nature 424, 743–748 (2003).

18. Girardin, S.E. et al. Nod1 detects a unique muropeptide from gram-negative bacterial peptidoglycan. Science 300, 1584–1587 (2003).

19. Girardin, S.E. et al. Nod2 is a general sensor of peptidoglycan through muramyl dipeptide (MDP) detection. J. Biol. Chem. 278, 8869–8872 (2003).

20. Martinon, F., Petrilli, V., Mayor, A., Tardivel, A. & Tschopp, J. Gout-associated uric acid crystals activate the NALP3 inflammasome. Nature 440, 237–241 (2006).

21. Petrilli, V. et al. Activation of the NALP3 inflammasome is triggered by low intracellular potassium concentration. Cell Death Differ. 14, 1583–1589 (2007).

22. Eisenbarth, S.C., Colegio, O.R., O’Connor, W., Sutterwala, F.S. & Flavell, R.A. Crucial role for the Nalp3 inflammasome in the immunostimulatory properties of aluminium adjuvants. Nature 453, 1122–1126 (2008).

23. Meylan, E. et al. Cardif is an adaptor protein in the RIG-I antiviral pathway and is targeted by hepatitis C virus. Nature 437, 1167–1172 (2005).

24. Seth, R.B., Sun, L., Ea, C.K. & Chen, Z.J. Identification and characterization of MAVS, a mitochondrial antiviral signaling protein that activates NF-κB and IRF 3. Cell 122, 669–682 (2005).

25. Xu, L.G. et al. VISA is an adapter protein required for virus-triggered IFN-β signaling. Mol. Cell 19, 727–740 (2005).

26. Huang, S. et al. MR1 antigen presentation to mucosal-associated invariant T cells was highly conserved in evolution. Proc. Natl. Acad. Sci. USA 106, 8290–8295 (2009).

27. Huang, S. et al. MR1 uses an endocytic pathway to activate mucosal-associated invariant T cells. J. Exp. Med. 205, 1201–1211 (2008).

28. Rottman, M. et al. Importance of T cells, gamma interferon, and tumor necrosis factor in immune control of the rapid grower Mycobacterium abscessus in C57BL/6 mice. Infect. Immun. 75, 5898–5907 (2007).

29. Mazmanian, S.K., Liu, C.H., Tzianabos, A.O. & Kasper, D.L. An immunomodulatory molecule of symbiotic bacteria directs maturation of the host immune system. Cell 122, 107–118 (2005).

30. Ivanov, I.I. et al. Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell 139, 485–498 (2009).

31. Gaboriau-Routhiau, V. et al. The key role of segmented filamentous bacteria in the coordinated maturation of gut helper T cell responses. Immunity 31, 677–689 (2009).

32. Shimamura, M. et al. Modulation of Vα19 NKT cell immune responses by α-mannosyl ceramide derivatives consisting of a series of modified sphingosines. Eur. J. Immunol. 37, 1836–1844 (2007).

33. Schumann, J. et al. Differential alteration of lipid antigen presentation to NKT cells due to imbalances in lipid metabolism. Eur. J. Immunol. 37, 1431–1441 (2007).

34. Fritz, J.H., Ferrero, R.L., Philpott, D.J. & Girardin, S.E. Nod-like proteins in immunity, inflammation and disease. Nat. Immunol. 7, 1250–1257 (2006).

35. Kawai, T. & Akira, S. TLR signaling. Cell Death Differ. 13, 816–825 (2006).36. Mattner, J. et al. Exogenous and endogenous glycolipid antigens activate NKT cells

during microbial infections. Nature 434, 525–529 (2005).37. Paget, C. et al. Activation of invariant NKT cells by Toll-like receptor 9-stimulated

dendritic cells requires type I interferon and charged glycosphingolipids. Immunity 27, 597–609 (2007).

38. Chen, L. & Jondal, M. TLR9 activation increases TAP-independent vesicular MHC class I processing in vivo. Scand. J. Immunol. 70, 431–438 (2009).

39. Stober, D., Trobonjaca, Z., Reimann, J. & Schirmbeck, R. Dendritic cells pulsed with exogenous hepatitis B surface antigen particles efficiently present epitopes to MHC class I-restricted cytotoxic T cells. Eur. J. Immunol. 32, 1099–1108 (2002).

40. Gulden, P.H. et al. A Listeria monocytogenes pentapeptide is presented to cytolytic T lymphocytes by the H2–M3 MHC class Ib molecule. Immunity 5, 73–79 (1996).

41. Lenz, L.L., Dere, B. & Bevan, M.J. Identification of an H2–M3-restricted Listeria epitope: implications for antigen presentation by M3. Immunity 5, 63–72 (1996).

42. 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).

A rt i c l e s©

201

0 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunologydoi:10.1038/ni.1890

ONLINE METHODSMice. The mice used here have been described8. Vβ6-transgenic mice were generated on the B6 background, and iVα19-transgenic mice and MR1-deficient mice were backcrossed onto the B6 background for more than ten generations. Mice doubly deficient in TAP and Ii were on a mixed B6-129 background. All iVα19-transgenic mice were on a TCRα-chain–deficient B6 background to avoid endogenous Vα expression. All experiments included littermates. MyD88-deficient, TRIF-deficient, Nod1-Nod2-deficient, NLRP3-ASC-deficient and IPS-1-deficient bone marrow were provided by M. Chignard, L. Alexopoulou, D. Philpott, J. Tschopp and M. Albert, respectively, and has been described15,20,43–45. All mice were housed at the Institut Curie in an accredited specific pathogen–free colony and were genotyped by PCR or flow cytometry staining. Live animal experiments were done in accordance with the guidelines of the French Veterinary Department. Germ-free mice were housed at the Transgenesis, Archiving and Animal Models, Centre National de la Recherche Scientifique, in Orléans, France, and were fed single bacteria species. Colonization was verified several times during the experiments. At 2–3 months after reconstitution, mesenteric lymph nodes were collected and MAIT cells were quantified by real-time quantitative RT-PCR as described7.

Cell preparation. Cell suspensions were prepared from the spleen or the peripheral or mesenteric lymph nodes by mechanical disruption on cell strain-ers. Cells were cultured in DMEM and Glutamax supplemented with 10% (vol/vol) FCS, penicillin and streptomycin, nonessential amino acids, HEPES, pH 7.4 and sodium pyruvate (all from Gibco).

BMDCs were prepared as described46. Bone marrow was obtained from legs of mice 6–12 weeks of age and was cultured for 7 d in the presence of 20% (vol/vol) supernatants of J558 mouse myeloma cell cultures or recombinant granulocyte-macrophage colony-stimulating factor (20 ng/ml; Peprotech). Cells were given fresh medium on days 3 and 5.

For in vitro T cell activation, splenocytes were incubated with magnetic beads (anti-CD11c (N418) and anti-CD19 for activation experiments, with anti-CD4 (L3T4) for cytokine-secretion experiments; all from Miltenyi) before magnetic separation with the MACS Pro system according to the manufac-turer’s recommendation (Miltenyi). The purity and yield of the depletion or enrichment were checked by flow cytometry.

The cell lines WT3 and WT3-mMR1 have been described27.

Human samples. Blood samples were obtained from healthy donors from the blood bank in accordance with institutional regulations. Peripheral blood mononuclear cells were obtained with a standard Ficoll gradient according to the manufacturer’s protocol (GE Healthcare). Monocytes were isolated by adherence on plastic culture plates. MAIT cells were isolated by magnetic-activated cell sorting with biotinylated anti-Vα7.2 and anti-biotin magnetic beads according to the manufacturer’s specifications (Miltenyi).

Patient blood samples were obtained as leftover materials after hemato-logical analysis of patients treated for infectious disease in the Department of Immunology of La Pitié-Salpétrière Hospital. All patients with known immuno-deficiency or low CD4+ cell counts were excluded from analysis. Similarly, blood was obtained from patients monitored for various solid tumors at the Institut Curie who had not received previous treatment (no chemotherapy or radiotherapy). In our institution, all patients were informed that pathological specimens may be used for research purposes.

Microbes, infection and activation. All bacteria were provided by the clini-cal microbiology laboratories of the Institut Curie and were American Type Culture Collection reference strains. Bacteria were cultured overnight at 37 °C in Luria broth, were washed in PBS and were diluted as needed. M. abscessus strain CIP 104536T-R was grown and used as described28. Where needed,

bacteria were fixed for 5 min in 1% (wt/vol) paraformaldehyde and then were washed extensively before use.

For in vitro infection, cells were washed and put in DMEM without supple-mentation. Dilutions of bacteria in DMEM were added for 3 h or other periods of time. Then, cells were washed two times in DMEM with 10% (vol/vol) FCS supplemented with penicillin and streptomycin (complete medium). Purified T cells were added for overnight coculture or longer where needed, then cells were collected and stained for flow cytometry.

For drug treatment, cells were cultured for 1 h in medium supplemented with cytochalasin D (5 μg/ml), dynasore (80 μM), lactacystin (5 μM) or dime-thyl sulfoxide or chloroquine (50 μM) or were left untreated before infection. After 3 h of infection, cells were washed. In some experiments, APCs were fixed for 2 min with 0.05% (wt/vol) glutaraldehyde and quenched with glycine and were washed thoroughly with complete medium before the addition of responder cells or bacteria.

For in vivo infection, mice were injected intraperitoneally or intravenously with 1 × 107 E. coli or 1 × 106 M. abscessus, respectively, in PBS.

Flow cytometry. Directly conjugated antibodies (BD Pharmingen) were used for flow cytometry according to standard techniques on a FACSAria and LSR II (Becton Dickinson). DAPI (4,6-diamidino-2-phenylindole) and excitation at 405 nm were used for the exclusion of dead cells. The following antibodies were used in mice: phycoerythrin- or fluorescein isothiocyanate–conjugated anti-Vβ6 (RR4-7) or anti-Vβ8 (F23.1; both from BD Pharmingen); phycoerythrin- or allophycocyanin-conjugated anti-CD44 (IM-7; ebiosciences); phycoerythrin-indodicarbocyanine-conjugated anti-TCRβ (H57-597; Biolegend); allophycocyanin-indotricarbocyanine–conjugated anti-CD8α (Ly-2; eBiosciences); Alexa Fluor 70–conjugated anti-CD4 (L3T4; eBiosciences); phycoerythrin–Texas red– or allophycocyanin-conjugated anti-CD69 (H1.2F3; Invitrogen), phycoerythrin–Texas red–conjugated anti-CD25 (PC61; Invitrogen); allophycocyanin-conjugated anti-CD11b (MI/70; eBio-sciences); phycoerythrin-conjugated anti- (3/23; BD Pharmingen); allophyco-cyanin-indotricarbocyanine–conjugated anti-CD11c (HL3; eBiosciences); and phycoerythrin-conjugated anti-CD86 (6L1; BD Pharmingen).

For staining of human cells, the following antibodies were used: allophyco-cyanin-indotricarbocyanin–conjugated anti-CD4 (RP4-T4; Biolegend), Alexa Fluor 700–conjugated anti-CD3ε (UCHT1; BD Pharmingen), phycoerythrin-indodicarbocyanin–conjugated anti-TCRγδ (IMMU510; Immunotech), phycoerythrin–Texas red–conjugated anti-CD8β (2ST8.5H7; Immunotech), fluorescein isothiocyanate–conjugated anti-CD45RO (UCHL1; Beckman Coulter), allophycocyanin-conjugated anti-CD161 (DX12; Miltenyi Biotec), allophycocyanin-conjugated anti-CD69 (CH/4; BD Pharmingen) and fluores-cein isothiocyanate–conjugated anti-CD11b (B-Ly6; BD Pharmingen). Anti-Vα7.2 (3C10) has been described8.

For quantification of cytokines, cytometric bead array technology was used according to the manufacturer’s specifications (BD Biosciences).

Statistical analysis. All quantitative data were analyzed by Prism software with paired or unpaired t-tests, nonparametric tests (U or Mann-Whitney) or analysis of variance.

43. Fritz, J.H. et al. Nod1-mediated innate immune recognition of peptidoglycan contributes to the onset of adaptive immunity. Immunity 26, 445–459 (2007).

44. Kanneganti, T.D. et al. Critical role for Cryopyrin/Nalp3 in activation of caspase-1 in response to viral infection and double-stranded RNA. J. Biol. Chem. 281, 36560–36568 (2006).

45. Paz, S. et al. Induction of IRF-3 and IRF-7 phosphorylation following activation of the RIG-I pathway. Cell. Mol. Biol. 52, 17–28 (2006).

46. Yao, Y., Li, W., Kaplan, M.H. & Chang, C.H. Interleukin (IL)-4 inhibits IL-10 to promote IL-12 production by dendritic cells. J. Exp. Med. 201, 1899–1903 (2005).

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

Corrigendum: Antimicrobial activity of mucosal-associated invariant T cellsLionel Le Bourhis, Emmanuel Martin, Isabelle Péguillet, Amélie Guihot, Nathalie Froux, Maxime Coré, Eva Lévy, Mathilde Dusseaux, Vanina Meyssonnier, Virginie Premel, Charlotte Ngo, Béatrice Riteau, Livine Duban, Delphine Robert, Martin Rottman, Claire Soudais & Olivier LantzNat. Immunol. 11, 701–708 (2010); published online 27 June 2010; corrected after print 13 August 2010

In the version of this article initially published, the author Shouxiong Huang (Department of Pathology and Immunology, Washington University, St. Louis, Missouri, USA) was not included. This author should be listed as author 15 (and affiliation 8). The error has been corrected in the HTML and PDF versions of the article.

CORR IGENDA©

201

0 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.