Natural killer T cells recognize diacylglycerol antigens from ... - Nature

© 2006 Nature Publishing Group http://www.nature.com/natureimmunology

ARTICLES

Natural killer T cells recognize diacylglycerol antigens from pathogenic bacteria Yuki Kinjo1,11, Emmanuel Tupin1,11, Douglass Wu2, Masakazu Fujio2, Raquel Garcia-Navarro3, Mohammed Rafii-El-Idrissi Benhnia4,10, Dirk M Zajonc5,10, Gil Ben-Menachem6,10, Gary D Ainge7, Gavin F Painter7, Archana Khurana1, Kasper Hoebe8, Samuel M Behar9, Bruce Beutler8, Ian A Wilson5, Moriya Tsuji3, Timothy J Sellati4, Chi-Huey Wong2 & Mitchell Kronenberg1 Natural killer T (NKT) cells recognize glycosphingolipids presented by CD1d molecules and have been linked to defense against microbial infections. Previously defined foreign glycosphingolipids recognized by NKT cells are uniquely found in nonpathogenic sphingomonas bacteria. Here we show that mouse and human NKT cells also recognized glycolipids, specifically a diacylglycerol, from Borrelia burgdorferi, which causes Lyme disease. The B. burgdorferi–derived, glycolipid-induced NKT cell proliferation and cytokine production and the antigenic potency of this glycolipid was dependent on acyl chain length and saturation. These data indicate that NKT cells recognize categories of glycolipids beyond those in sphingomonas and suggest that NKT cell responses driven by T cell receptor–mediated glycolipid recognition may provide protection against diverse pathogens.

Natural killer T (NKT) cells are T lymphocytes that express both ab T cell receptors (TCRs) and NK cell receptors1,2. Unlike conventional ab T cells, which are activated by antigenic peptides, NKT cells recognize glycolipids presented by CD1d, a nonclassical antigenpresenting molecule. Also unlike conventional ab T cells, NKT cells express an invariant TCRa chain. In mice, most NKT cells express an invariant Va14-Ja18 TCR rearrangement, whereas the homologous human population expresses an invariant Va24-Ja18 TCR rearrangement1,2. These two populations are therefore sometimes referred to as ‘Va14 invariant NKT cells’ (or ‘Va14i NKT cells’) and ‘Va24i NKT cells’, respectively, whereas the term ‘iNKT cells’ is used to refer to both populations2. By several means, including use of the synthetic glycolipid a-galactosylceramide (a-GalCer), many studies have shown that iNKT cells are pivotal during a variety of immune responses in mice, including the development of autoimmunity, maintenance of self-tolerance, and responses to tumors and infectious agents1,2. The glycolipid a-GalCer, originally isolated from a marine sponge during a screen for compounds that could prevent tumor metastasis3, is a potent iNKT cell agonist. Unlike most glycosphingolipids in nature, however, for a-GalCer, an a-anomeric linkage connects the sugar to

the ceramide lipid4. Therefore, it is thought that a-GalCer is not likely to be a natural antigen for iNKT cells. Because Va14i NKT cells have been shown to participate in the protection of mice from bacterial infections1,2,5–7, many groups have searched for microbial antigens for invariant iNKT cell TCRs. Some microbes such as salmonella stimulate iNKT cells in a way not involving direct recognition of foreign glycolipids by iNKT cell TCRs8. Instead, during salmonella infection, the combination of interleukin 12 (IL-12) from Toll-like receptor (TLR)–stimulated antigen-presenting cells and a weak TCR signal sent by CD1d-bound self glycolipids activates Va14i NKT cells8,9. However, it has been reported that most iNKT cells directly recognize natural a-linked glycosphingolipids from sphingomonas bacteria9–11, which are Gram-negative a4 proteobacteria that lack lipopolysaccharide. Their a-linked glycosphingolipids are similar in structure to a-GalCer12,13. Mouse and human iNKT cell TCRs directly recognize CD1d-bound glycosphingolipids from sphingomonas, and such recognition stimulates cytokine release from mouse and human iNKT cells9,10. Moreover, mice lacking Va14i NKT cells show delayed clearance of sphingomonas bacteria9,10. Those data support the hypothesis that the iNKT cell TCR can directly recognize microbial antigens and that recognition of glycosphingolipids can rapidly contribute to

1Division of Developmental Immunology, La Jolla Institute for Allergy & Immunology, La Jolla, California 92037, USA. 2Department of Chemistry and Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, California 92037, USA. 3HIV and Malaria Vaccine Program, Aaron Diamond AIDS Research Center, Rockefeller University, New York, New York 10016, USA. 4Center for Immunology and Microbial Disease, Albany Medical College, Albany, New York 12208, USA. 5Department of Molecular Biology and Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, California 92037, USA. 6Laboratory of Developmental and Molecular Immunity, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland 20892, USA. 7Carbohydrate Chemistry Team, Industrial Research, Gracefield Research Center, Lower Hutt 6008, New Zealand. 8Department of Immunology, The Scripps Research Institute, La Jolla, California 92037, USA. 9Division of Rheumatology, Immunology, and Allergy, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts 02115, USA. 10Present addresses: Division of Vaccine Discovery (M.R.-E.-I.B.) and Division of Cellular Biology (D.M.Z.), La Jolla Institute for Allergy & Immunology, La Jolla, California 92037, USA, and Paramount Biosciences, Cambridge, Massachusetts 02138, USA (G.B.-M.). 11These authors contributed equally to this work. Correspondence should be addressed to M.K. ([email protected]).

Received 14 June; accepted 25 July; published online 20 August 2006; doi:10.1038/ni1380

978

VOLUME 7

NUMBER 9

SEPTEMBER 2006

NATURE IMMUNOLOGY

ARTICLES

Relative number of cells


B. burgdorferi, expression of both CD25 and CD69 was increased on Va14i NKT cells in mice injected with dendritic cells (DCs) ‘pulsed’ with B. burgdorferi lysate (Fig. 1b). These data demonstrated that B. burgdorferi activates Va14i NKT cells in vivo, either indirectly through antigen-presenting cell activation or by direct presentation of a glycolipid antigen to Va14i NKT cell TCRs.

b

a

CD25

CD25

CD69

CD69

Figure 1 In vivo activation of Va14i NKT cells by B. burgdorferi. (a) Expression of activation markers CD25 and CD69 on gated TCRb+ liver mononuclear cells positive for the a-GalCer–CD1d tetramer at 1 week after infection with B. burgdorferi via tick bites. Data are representative of two uninfected, PBS-treated mice (gray histograms) and four infected mice (solid lines); similar data were obtained in two experiments. (b) Expression of activation markers CD25 and CD69 on Va14i NKT cells gated as described in a, at 14 h after intravenous injection of DCs pulsed with B. burgdorferi lysate (solid lines), a-GalCer (dashed lines) or PBS (gray histograms). Data are representative of three mice in each group; similar data were obtained in three experiments.

host defense against lipopolysaccharide-negative microorganisms. Sphingomonas are not highly pathogenic, however, and a-linked glycosphingolipids are believed to be unique to sphingomonas and related bacteria. It remains a mystery whether iNKT cells can recognize other classes of glycolipids that might be derived from pathogenic microbes. Borrelia burgdorferi spirochetes, Gram-negative bacteria that lack lipopolysaccharide, are the causative agent of Lyme disease. With 15,000 cases each year, Lyme disease is the most common vector-borne disease in the US14. After infection with B. burgdorferi, mice with homozygous deficiency in CD1d, which lack Va14i NKT cells, have an higher bacterial burden than that of wild-type mice and develop increased thickening of the tibiotarsal joint indicative of arthritis15. Here we show that Va14i NKT cells were activated in vivo during B. burgdorferi infection and iNKT cells recognized a B. burgdorferi– derived diacylglycerol. These data demonstrate that iNKT cells recognize antigens other than glycosphingolipids and from pathogenic microbes. In addition, these results raise the possibility that the conserved TCR of iNKT cells may be involved in host defense against many pathogens. RESULTS NKT cell activation during B. burgdorferi infection To determine whether Va14i NKT cells are activated during B. burgdorferi infection, we infected C57BL/6 mice with B. burgdorferi by means of Ixodes scapularis tick bites and isolated lymphocytes from the liver at various times after infection. We evaluated Va14i NKT cell activation by measuring CD25 and CD69 on the surface of Va14i NKT cells, which bind to CD1d tetramers loaded with a-GalCer16,17. At 1 week after infection, expression of both CD25 and CD69 was increased on liver Va14i NKT cells (Fig. 1a), although we did not detect this increase at 3 d or 2 weeks after infection. We obtained similar results after syringe inoculation of B. burgdorferi (data not shown). As with the results obtained with mice infected with live

NATURE IMMUNOLOGY

VOLUME 7

NUMBER 9

SEPTEMBER 2006

Galactosyl diacylglycerols activate Va14i NKT cells B. burgdorferi expresses two main glycolipids: B. burgdorferi glycolipid 1 (BbGL-I) and BbGL-II. BbGL-I, which represents 23% of the total lipid, is a cholesteryl 6-O-acyl-b-galactoside (data not shown) and BbGL-II, which represents approximately 12% of the total lipid, is a 1,2-diacyl-3-O-a-galactosyl-sn-glycerol18,19 (Fig. 2). Notably, sera from patients with Lyme disease react to these compounds20–22. To determine if these glycolipids could be responsible for the Va14i NKT cell activation noted during B. burgdorferi infection, we tested their ability to stimulate iNKT cells in a cell-free antigen-presentation assay. As expected, galacturonic acid–containing sphingomonas glycosphingolipid (GalA-GSL)23, stimulated the release of IL-2 from the Va14i NKT cell hybridoma 1.2 when added to a plate coated with soluble CD1d molecules (Fig. 3a). The same hybridoma also released IL-2 in response to purified BbGL-II (Fig. 3a), although BbGL-II seemed to be a less potent antigen than GalA-GSL. In contrast, the cholesterolcontaining BbGL-I did not stimulate IL-2 release. Purified BbGL-II contains a mixture of C14:0, C16:0, C18:0, C18:1 and C18:2 fatty acids, with C16:0 and C18:1 being the most common18. Because it was not possible to specify which fatty acids were in the sn-1 and sn-2 positions of the glycerol, we chemically synthesized several variants of BbGL-II (Fig. 2 and Table 1) and tested these compounds in the cell-free antigen-presentation assay. All of the synthetic BbGL-II compounds were able to induce some IL-2 production from the 1.2 hybridoma when high doses were used. However, BbGL-IIc, which had an oleic acid in the sn-1 position and a palmitic acid in the sn-2

α-GalCer

BbGL-II HO

HO

OH

OH

O C25H51

O HO

O HO HO

HN HO

OR2 O

OH

O

C14H29

OR1 OH

Figure 2 Structure of BbGL-II. BbGL-II is a monogalactosyl diacylglycerol containing a mixture of fatty acids. R1, sn-1 position of glycerol; R2, sn-2 position of glycerol (Table 1).

Table 1 Fatty acids of synthetic BbGL-II compounds Compound

R1

R2

BbGL-IIa BbGL-IIb

C16:0 C16:0

C18:1 C18:2

BbGL-IIc BbGL-IId

C18:1 C18:1

C16:0 C18:2

BbGL-IIe BbGL-IIf

C18:2 C18:2

C16:0 C18:1

BbGL-IIg BbGL-IIh

C16:0 C16:0

C18:0 C16:0

Data represent the acyl chains of the synthetic compounds BbGL-IIa–BbGL-IIh. R1, sn-1 position of glycerol; R2, sn-2 position of glycerol (Fig. 2); these indicate acyl chain length and the number of unsaturated bonds (C18:2 indicates 18 carbon fatty acids containing 2 unsaturated bonds).

979

ARTICLES

a 3

IL-2 (10 U/ml)

3.0

b

0 125 500 2,000

2.5 2.0 1.5 0.5

BbGL-II

BbGL-I

GSL 4

40 20

ND

0

II PIM4GSL α- CD1GC A20

Veh IIa IIb IIc BbGL

2

1

0 IIa

IIb

IIc

IId

IIe

IIf

IIg

IIh

α-GC

12

50

2,

0 00 0 5 G SL

ND

0

50

0

ND

2,

Ve h

1,200 1,000 800 600 400 200 0

00

d

3

3

60

IL-2 (10 U/ml)

IL-2 (U/ml)

80

IL-2 (U/ml)


5

NT

0

c

0 125 500 2,000

1.0

PIM4 BbGL-IIc

Figure 3 Va14i NKT cell hybridomas respond to B. burgdorferi glycolipids. (a) ELISA of IL-2 in supernatants after 16 h of culture of Va14i NKT cell hybridoma 1.2 on CD1d-coated plates with purified B. burgdorferi glycolipids BbGL-I or BbGL-II or with GalA-GSL (GSL; amounts (in ng) in key). NT, not tested. (b) Response of Va14i NKT cell hybridoma 1.2 to CD1d-coated plates containing synthetic B. burgdorferi glycolipids BbGL-IIa–BbGL-IIh (IIa–IIh) or a-GalCer (a-GC; amounts (in ng) in key), assessed by ELISA of IL-2 in culture supernatants. (c) Response of CD1d-reactive Va14i – hybridoma 19 to CD1d-transfected A20 cells alone (CD1-A20) or to CD1d-coated plates containing vehicle alone (Veh), BbGL-IIa, BbGL-IIb, BbGL-IIc or BbGL-II (IIa–IIc and II; 2 mg), PIM4 (2 mg), GalA-GSL (0.5 mg), assessed by ELISA for IL-2 in culture supernatants. (d) Response of Va14i NKT cell hybridoma 1.2 to CD1d-coated plates containing vehicle alone, PIM4, BbGL-IIc or GalA-GSL (amounts (in ng), horizontal axis), assessed by ELISA for IL-2 in culture supernatants. ND, not detected. All data are means ± s.d. for triplicate wells and represent three (a,b) or two (c,d) independent experiments.

position, was the most potent antigen (Fig. 3b). BbGL-II compounds also stimulated the Va14i hybridomas 24.9E and 1.4 (Supplementary Fig. 1 online) and 2C12 (data not shown). However, BbGL-II compounds failed to stimulate three CD1d-autoreactive hybridomas without expression of the Va14i TCR chain (Fig. 3c and Supplementary Fig. 2 online), although these cells produced IL-2 when cultured either with CD1d transfectants or with plates coated with CD1d loaded with cognate antigen. These data suggested that an a-galactosyl diacylglycerol derived from B. burgdorferi could specifically stimulate Va14i NKT cells. It has been reported that another non-glycosphingolipid antigen, the glycolipid PIM4 purified from Mycobacterium bovis Bacillus Calmette-Guerin, stimulates a subset of iNKT cells24. To compare BbGL-II activity to PIM4 activity, we chemically synthesized PIM4 and tested its antigenic potency in the cell-free antigen-presentation assay. PIM4 did not elicit IL-2 production from any of three Va14i NKT cell hybridomas tested (Fig. 3d and Supplementary Fig. 1). To determine if primary Va14i NKT cells can also directly recognize BbGL-II, we stained liver mononuclear cells with CD1d tetramers Figure 4 Va14i NKT cells bind CD1d tetramers loaded with BbGL-II. (a) Staining of liver mononuclear cells from wild-type C57BL/6 mice (WT) or Ja18-deficient mice (Ja18-KO) with CD1d tetramers loaded with vehicle, purified BbGL-I, purified BbGL-II or a-GalCer. (b) Staining of liver mononuclear cells from wild-type mice with CD1d tetramers loaded with vehicle, synthetic BbGL-IIc or a-GalCer. Numbers beside boxed areas indicate percent tetramer-positive, TCRb+ cells. Data represent two independent experiments in which two individual mice were analyzed.

980

loaded with purified BbGL-II, BbGL-II or a-GalCer or with vehicle containing 0.5% polysorbate-20 in 0.9% NaCl. CD1d tetramers loaded with purified BbGL-II bound to approximately 12% the number of cells that bound to a-GalCer-loaded tetramers (Fig. 4a).

a

WT

Vehicle

Jα18-KO

0.11

0.13

0.22

0.22

4.07

0.36

BbGL-I

BbGL-II

b 0.1 Vehicle

8.7 BbGL-IIc

α-GalCer

37.8

TCRβ α-GalCer

32.80

0.23

TCRβ

VOLUME 7

NUMBER 9

SEPTEMBER 2006

NATURE IMMUNOLOGY

ARTICLES

BbGL-IIe 0.9

BbGL-IIc 3.4

BbGL-IIg

BbGL-IIh

0.9

1.1

α-GalCer

Vehicle

5.2

TCRβ

b

Vehicle

3.2

BbGL-IId

2.1

BbGL-IIc

α-GalCer

60.3

42.2

BrdU

BbGL-II-reactive cells were absent from the livers of Ja18-deficient mice, which lack Va14i NKT cells because of the absence of the Ja gene segment required for the formation of the Va14i TCR. This suggested that BbGL-II-reactive cells and a-GalCer-reactive cells are overlapping populations. Furthermore, CD1d tetramers loaded with the most potent synthetic BbGL-II analog, BbGL-IIc, stained approximately 23% as many liver Va14i NKT cells as did a-GalCer-loaded tetramers (Fig. 4b). These results showed that a substantial fraction of the Va14i NKT cells can directly recognize a-galactosyl diacylglycerols derived from B. burgdorferi. As a functional assay of the ability of BbGL-II to stimulate primary Va14i NKT cells, we assessed in vitro proliferation of splenic Va14i NKT cells. We cultured whole spleen cells in the presence of BbGL-II variants, a-GalCer or the vehicle dimethyl sulfoxide and determined the proportion of Va14i NKT cells by using a-GalCer-loaded CD1d tetramers. After 5 d of culture with BbGL-IIc, Va14i NKT cells represented 3.4% of total lymphocytes (Fig. 5a). In contrast, Va14i NKT cells constituted only 0.9% of total lymphocytes after 5 d of culture with the vehicle dimethyl sulfoxide or other variants of BbGL-II. Incorporation of 5-bromodeoxyuridine (BrdU) demonstrated that cells binding a-GalCer-loaded tetramers proliferated in the presence of a-GalCer and BbGL-IIc but not in the presence of BbGL-IId (Fig. 5b). To determine whether B. burgdorferi–derived glycolipids also stimulate Va14i NKT cells in vivo, we immunized mice with bone

Figure 6 In vivo response of Va14i NKT cells to B. burgdorferi antigens. (a) Expression activation markers CD25 and CD69 on gated TCRb+ liver mononuclear cells positive for the a-GalCer–CD1d tetramer at 14 h after intravenous injection of wild-type DCs pulsed with BbGL-IIc (thick solid lines), a-GalCer (broken lines), BbGL-IIa (thin solid lines) or vehicle (gray histograms). (b) Percentage of a-GalCer–CD1d tetramer–positive, B220– liver mononuclear cells expressing intracellular IFN-g (numbers in top right quadrants) 14 h after injection of DCs pulsed with various compounds (above plots). Data represent three independent experiments with two or three mice in each group.

NATURE IMMUNOLOGY

VOLUME 7

NUMBER 9

SEPTEMBER 2006

Figure 5 In vitro proliferation of Va14i NKT cells by a B. burgdorferi antigen. (a) Flow cytometry of splenocytes from wild-type mice stimulated for 5 d with various compounds (above plots). Numbers beside outlined areas indicate the percentage of cells positive for the a-GalCer–CD1d tetramer. (b) Flow cytometry of splenocytes from wild-type mice stimulated as described in a; BrdU was added to cultures 18 h before analysis. Cells positive for a-GalCer–CD1d tetramer were gated; numbers above bracketed lines indicate the percentage of BrdU+ cells. Data represent three independent experiments.

marrow–derived DCs pulsed with BbGL-II glycolipids or a-GalCer and 14 h later analyzed the expression of activation markers on cells binding a-GalCer-loaded tetramers. Injection of BbGL-IIc-pulsed DCs resulted in increased expression of CD25 and CD69 on most liver mononuclear cells positive for a-GalCer–CD1d tetramer (Fig. 6a). In contrast, BbGL-IIa-pulsed DCs did not induce the expression of activation markers (Fig. 6a). We obtained similar results with splenocytes (data not shown). DCs pulsed with a-GalCer or BbGL-IIc, but not those pulsed with BbGL-IIa, triggered production of intracellular interferon-g (IFN-g), which could be detected directly ex vivo without the use of brefeldin A or similar reagents (Fig. 6b). It has been shown that DCs pulsed with a-GalCer or GalA-GSL ‘preferentially’ stimulate the production of IFN-g rather than IL-4 (refs. 10,25). Consistent with those results, DCs pulsed with BbGL-IIc also induced more cells to produce IFN-g than IL-4 (Supplementary Fig. 3 online). However, liver Va14i NKT cells did not upregulate activation markers or produce IFN-g after immunization with PIM4-pulsed DCs (Supplementary Fig. 3) and we obtained similar results in the spleen (data not shown). These data showed that the synthetic B. burgdorferi glycolipid, but not synthetic PIM4, stimulates a substantial fraction of Va14i NKT cells in vivo. TLR-independent BbGL-II-induced NKT cell activation In addition to glycolipid antigen recognition by the invariant TCR, called ‘direct recognition’7, iNKT cells can be activated indirectly during microbial infection by a combination of recognition of self antigen and production of IL-12 by activated antigenpresenting cells8,9. We therefore investigated whether B. burgdorferi glycolipids triggered TLR signaling and activation of antigenpresenting cells. We cultured bone marrow–derived DCs in the

a

CD25

CD69 BbGL-IIa

Vehicle

b

1.6

IFN-γ

BbGL-IIb 0.9

Relative number of cells

α-GalCer tetramer

BbGL-IIa 0.9

0.9

Cells


a

1.2

α-GalCer

BbGL-IIc 17.8

73.7

TCRβ

981

ARTICLES

a

Vehicle

WT

b

BbGL-IId

Vehicle

BbGL-IId

BbGL-IIc

α-GalCer

0.9 Cells

1.1

1.7

1.2

40.1

31.2

BrdU

c

BbGL-IIc

Vehicle

1.6

2.7

2.1

α-GalCer

BbGL-2c

11.3

62.7

WT

Myd88

IFN-γ

α-GalCer tetramer

–/–

Vehicle

BbGL-IIc

1.2

1.5

12.9

68.7

3.0 Myd88 –/–

TCRβ

TCRβ

Figure 7 MyD88-independent activation of Va14i NKT cells by B. burgdorferi antigens. (a) Percentage of cells positive for the a-GalCer–CD1d tetramer (numbers beside outlined areas) among splenocytes from wild-type or Myd88–/– mice stimulated for 5 d with various compounds (above plots). (b) Flow cytometry of splenocytes from wild-type or Myd88 –/– mice stimulated as described in a; BrdU was added to cultures 18 h before analysis. Numbers above bracketed lines indicate percentage of BrdU+ cells in the gated a-GalCer–CD1d tetramer–positive population. Data (a,b) represent three independent experiments. (c) Flow cytometry of intracellular IFN-g in B220– liver mononuclear cells positive for the a-GalCer–CD1d tetramer (percent IFN-g+, top right quadrants), assessed 14 h after wild-type or Myd88 –/– DCs were pulsed with various compounds (above plots) and injected intravenously into wild-type or Myd88 –/– mice, respectively. Cells binding isotype control antibody were less 2% (vehicle or BbGL-IIc) and less than 7% (a-GalCer). Data represent four independent experiments with two mice in each group.

presence of BbGL compounds, a-GalCer, lipopolysaccharide or CpG and assessed the expression of maturation markers. Culture with lipopolysaccharide or CpG but not culture with BbGL-IIc induced an increase in the expression of CD40, CD80, CD86 and major histocompatibility complex class II molecules on DCs (data not shown). Furthermore, BbGL-IIc did not stimulate the release of tumor necrosis factor or IL-12p70 from DCs (Supplementary Fig. 4 online). These data suggested that BbGL-IIc induced Va14i NKT cell activation only through direct TCR recognition, in a way similar to sphingomonas GalA-GSL or to a-GalCer. To confirm those findings, we obtained spleen cells from mice lacking the adaptors MyD88 or TRIF and cultured the cells in the presence of BbGL-IIc. All known TLR signaling proceeds through MyD88 and/or through TRIF26. BbGL-IIc induced the proliferation of Va14i NKT cells independently of the presence of MyD88 (Fig. 7a,b). Furthermore, Myd88–/– DCs pulsed with BbGL-IIc also stimulated the release of IFN-g from Myd88–/– liver Va14i NKT cells (Fig. 7c). Similarly, BbGL-IIc induced in vitro proliferation of Va14i NKT cells in spleen cell cultures from TRIF-deficient ‘Trif Lps2/Lps2’ mice, which have the ‘Lps2’ mutation resulting in a distal frameshift error in TRIF26

Va24i NKT cells respond to B. burgdorferi glycolipids To determine if the iNKT cell response to B. burgdorferi glycolipids is conserved in humans, we tested Va24i human NKT cell lines grown in vitro with a-GalCer. These lines produced IFN-g and IL-4 after culture with cells transfected with CD1d and pulsed with the synthetic galactosyl diacylglycerol compounds (Fig. 8), but we noted no response after culture with untransfected control cells (data not shown). The response pattern of Va24i NKT cells differed from that of Va14i NKT cells. Minimal cytokine release was induced by culture with BbGL-IIc, and we obtained maximum responses after culture with compounds having a higher degree of unsaturation, particularly with BbGL-IIf (Fig. 8).

a 50

0 1 10

40 30 20

b 120

0 1 10

100

10 0

982

(data not shown). DCs from Trif Lps2/Lps2 mice also stimulated the release of IFN-g from Trif Lps2/Lps2 liver mononuclear cells positive for the a-GalCer–CD1d tetramer (Supplementary Fig. 5 online). These data confirmed that Va14i NKT cell activation induced by BbGL-IIc is not dependent on TLR signals. We conclude that a-galactosyl diacylglycerol compounds from B. burgdorferi stimulate Va14i NKT cells by direct engagement of their invariant TCR.

IL-4 (pg/ml)

Figure 8 Activation of human Va24i NKT cells by diacylglycerol glycolipids. Cytokines IFN-g and IL-4 in supernatants of cells from a human Va24i + cell line expanded in a-GalCer and cultured in the presence of HeLa cells transfected with human CD1d and ‘pulsed’ with BbGL-IIa–BbGL-IIh (IIa–IIh; concentration (in mg/ ml) in keys), measured after 24 h. Data are the mean ± s.d. of duplicate wells and represent three independent experiments.

IFN-γ (ng/ml)


α-GalCer

80 60 40 20

IIa

IIb

IIc

IId

IIe

IIf

VOLUME 7

IIg

IIh

NUMBER 9

0

IIa

IIb

IIc

SEPTEMBER 2006

IId

IIe

IIf

IIg

IIh

NATURE IMMUNOLOGY

ARTICLES α2

α1 Arg79

Asp153

Thr156


Asp80

F′

Figure 9 Binding of B. burgdorferi glycolipid to mouse CD1d. The modeled BbGL-IIc compound (yellow) is presented binding to mouse CD1d; blue dashed lines indicate potential hydrogen bond interactions to CD1d a2 helix residues. For comparison, binding of CD1d to the short-chain a-GalCer analog PBS-25 and to sphingomonas GalA-GSL (Protein Data Bank accession number, 2FIK) is presented in blue and green, respectively.

Modeling of the binding of BbGL-II to CD1d The functional data reported above indicated that many iNKT cells respond to BbGL-II compounds and that NKT cell responsiveness depends to a great degree on the acyl chains of each compound. To understand the structural basis of those observations, we produced a model for the binding of BbGL-II to CD1d based on the crystal structure of the mycobacterial glycolipid PIM2 (phosphatidylinositol dimannoside) bound to mouse CD1d27. From the binding of the PIM2 glycerol moiety in the crystal structure with CD1d, we oriented the BbGL-II glycolipids with the sn-1 fatty acid in the A¢ pocket and the sn-2 fatty in the F¢ pocket. BbGL-IIc could be modeled easily in the mouse CD1d binding groove (Fig. 9) with the C16 alkyl chain in the F¢ pocket. The cis double bond on the nine position of the sn-1 oleoyl group, which results in a ‘kink’ in the sn-1 acyl chain, is located in a region of the A¢ pocket, where it facilitates encircling of the A¢ pocket pole at Cys12. Notably, the resulting model is reminiscent of the binding of a-GalCer to CD1d in that similar polar interactions could be formed between the galactose in the head group of BbGL-IIc and CD1d a2-helix residues Asp153 and Thr156 (Fig. 9). Unlike a-GalCer, however, BbGL-IIc does not form hydrogen bonds with Arg79 or Asp80 of the a1 helix. Although this model cannot explain the differences noted in antigenic potency for the entire series of BbGL-II compounds, the other two lipids with a C16 sn-2 alkyl chain do not make such an optimal fit. For example, the structurally related BbGL-IIe ligand, which has an additional cis unsaturation in the sn-1 alkyl chain between C12 and C13, could not be easily modeled without the introduction of steric ‘clashes’ with mouse CD1d (data not shown). DISCUSSION The invariant TCR expressed by iNKT cells is highly conserved28, but the selection pressure for the maintenance of this specificity has not been fully elucidated. The iNKT cells recognize a-GalCer, which was derived from a marine sponge, but it seemed implausible that a conserved T cell population was selected for the recognition of a rather atypical antigen from a marine organism. Three studies have demonstrated that most iNKT cells recognize glycosphingolipids from sphingomonas bacteria9–11. Moreover, those glycosphingolipids have a-linked sugars similar to a-GalCer12,13. Sphingomonas are very abundant in the environment29, but they are not pathogenic, and it

NATURE IMMUNOLOGY

VOLUME 7

NUMBER 9

SEPTEMBER 2006

remained a key issue whether iNKT cells could recognize other classes of glycolipids that might be more widely distributed in pathogenic microbes. Indirect evidence has suggested that pathogenic ehrlichia contain an antigen that stimulates iNKT cell TCRs9, but the structure of that antigen is unknown. Here we have shown that microbial iNKT cell antigens are not limited either to atypical a-branched glycosphingolipids or to sphingomonas and related bacteria. Indeed, we have provided evidence that a-galactosyl diacylglycerols from B. burgdorferi, the causative agent of Lyme disease, directly stimulated iNKT cells through TCR engagement rather than iNKT cell activation through antigen-presenting cell stimulation. We focused on the spirochete B. burgdorferi for two reasons. First, CD1d-deficient mice have an increased burden of B. burgdorferi bacteria15. The more chronic infection in these mice is probably related to an enhanced inflammatory response and increased joint swelling15, which is considered to be the hallmark of human Lyme disease. Here we have demonstrated that Va14i NKT cells were activated in vivo during infection with live B. burgdorferi or after injection of DCs pulsed with B. burgdorferi lysate. In addition to augmenting the antimicrobial response, iNKT cells can regulate the inflammation resulting from microbial infection1,2,5–7. For example, in a mouse model of Chagas disease, Va14i NKT cell–deficient Ja18deficient mice die from severe inflammation after Trypanosoma cruzi infection, whereas control mice survive30. Therefore, Va14i NKT cell activation after B. burgdorferi infection may be involved in both bacterial clearance and the prevention of reactive arthritis by altering the balance of cytokines, which is thought to be a key factor in the regulation of Lyme arthritis31. A second reason we focused on B. burgdorferi was that the structures of abundant B. burgdorferi glycolipids that are the target of the serological response in human patients with Lyme disease have been determined18,19. We found that one of those glycolipids, BbGL-II, activated a substantial proportion of iNKT cells in mice and humans, including each of four Va14i NKT cell hybridomas and four human Va24i NKT cell lines. Tetramers of CD1d loaded with BbGL-IIc detected nearly 25% of liver Va14i NKT cells. It is not possible to estimate the exact number of BbGL-IIc-reactive Va14i NKT cells in this way, because of the overall low amount of BbGL-IIc tetramer staining. BbGL-II induced the upregulation of activation markers on most Va14i NKT cells in vivo, even in MyD88- or TRIFdeficient mice, which are likely to have impaired antigen-presenting cell activation. Those data suggested that most iNKT cells may be capable of responding to BbGL-II. Notably, BbGL-II is not a glycosphingolipid but is instead a monogalactosyl diacylglycerol. Although glycerol-based phospholipids can be presented by CD1b to T cells expressing diverse TCRs6, all of the previously defined antigens that activate most iNKT cells are glycosphingolipids9–11,32. Another antigen that is not a glycosphingolipid has been detected in a cell wall fraction of M. bovis Bacillus Calmette-Guerin enriched for PIM4 (ref. 24). Tetramers loaded with this PIM4-enriched fraction stain only 1% of the number of liver T cells positive for the a-GalCer–CD1d tetramer, although a much higher percentage is detected in the spleen24. Those data suggest that only a subset of the Va14i NKT cells recognize PIM4 and therefore that PIM4 reactivity is unlikely to account for the conservation of the invariant TCR. Relatively minor subsets of Va14i NKT cells have been reported to react to other antigens, such as phosphatidylinositol, phosphatidylethanolamine33–35 or ganglioside GD3 (ref. 36), probably reflecting TCR CDR3b sequence heterogeneity. In our study here, we were unable to demonstrate bioactivity of synthetic PIM4 in stimulating Va14i NKT cells either in vitro or in vivo. We cannot rule out

983


ARTICLES the possibility that synthetic PIM4 has properties different from those of the purified material used in the previous study24. Based on the PIM2-CD1d crystal structure, synthetic PIM4 is likely to bind CD1d, as it has the same lipid component as the PIM2 that was crystallized together with CD1d. Moreover, the synthesized material, which contained two palmitic acid acyl chains, was designed to correspond to the main component of the purified PIM4, as detected by mass spectrometry24. It is possible, however, that a minor portion of the mixture, inevitably present in a purified glycolipid fraction, contained the bioactivity. Regardless of which of those possibilities is true, consistent with the hypothesis that BbGL-II acts as a broad antigen for iNKT cells, a model of a BbGL-IIc–CD1d complex localized the exposed galactose in a position similar to that in the previously solved glycosphingolipid-CD1d structures. It is difficult to conceive how the more complex, phosphate-containing carbohydrate of PIM4 could form a similar epitope, although it is likewise difficult to understand how the trisaccharide isoglobotrihexosylceramide activates so many iNKT cells. The glycerol moiety of purified BbGL-II contains a mixture of fatty acid chains and, although it has antigenic activity, it was less potent than the optimal synthetic compound. We found that small changes in length and saturation of BbGL-II alkyl chains had a large effect on antigenic potency. As the acyl chains are not solvent exposed, the differences in antigenic potency must be related to the efficiency or stability of CD1d loading or to lipid-dependent differences in the position of the galactosyl head group or, as a-GalCer causes an induced fit of CD1d37,38, it is possible that lipid antigen influences the conformation of CD1d. Although it remains to be determined how the fatty acid composition might correlate with those various factors, modeling of BbGL-II lipids in the mouse CD1d binding groove suggested a size restriction of C16 for the alkyl chain bound in the F¢ pocket; longer bound alkyl chains would protrude and perhaps alter the galactose moiety. That prediction is not unexpected, as all of the ligands present mouse CD1d crystal structures have a maximum alkyl chain length of C16 that can be accommodated in the F¢ pocket23,37–40. Otherwise, it is difficult to model a particular ‘cis’ unsaturation in a flexible alkyl chain, and some of the differences in ligand potencies could result from a variety of factors, including more efficient loading of those compounds onto CD1d. Although we consider the substantial changes in antigenic potency due to subtle alkyl chain alterations unexpected, it is known that truncation or unsaturation of the lipid chains of a-GalCer can influence antigenic potency as well as the pattern of cytokine production41–43. Some of those a-GalCer alterations, however, truncate a relatively large number of carbons, such as the nine residues deleted from the C18 sphingosine of a-GalCer in an OCH ((2S,3S,4R)-1-O-(a-D-galactopyranosyl)-Ntetracosanoyl-2-amino-1,3,4-nonanetriol) analog41. Furthermore, a Va14i NKT cell hybridoma and a Va24i NKT cell clone recognize phosphatidylethanolamine in addition to a-GalCer, although most iNKT cells do not have this specificity. The phosphatidylethanolamine-reactive cells require unsaturated bonds in the acyl chains of the phosphatidylethanolamine antigen, but the reactivity increases as the degree of unsaturation in both chains increases34,35. We found no evidence of that type of specificity in our study. Instead, a particular combination of saturated and unsaturated chains of precise length was required. In summary, we have shown that the invariant TCRs expressed by iNKT cells can recognize glycoglycerol lipids from B. burgdorferi. That finding has several notable implications for understanding the biology of these unconventional T lymphocytes. Because bacteria other than B. burgdorferi have glycoglycerol lipids, the invariant TCRs could then

984

have a broad reactivity to a variety of microbes, which might explain in part the evolutionary selection for this TCR specificity. Moreover, bacteria probably produce a mixture of antigenic and nonantigenic glycoglycerol lipids, and we speculate that some of those could even serve as TCR antagonists. Therefore, alterations in alkyl chain composition possibly could serve as a mechanism of immune evasion. Finally, glycoglycerol lipids have been found in mammalian cells44 and thus compounds of this type could constitute a previously unknown category of self antigens for iNKT cells. METHODS Reagents. The a-GalCer was a gift from the Kirin pharmaceutical research corporation (Gunma, Japan). B. burgdorferi glycolipids were purified as described18. The synthesis of BbGL-II compounds is described in the Supplementary Methods online. PIM4 was synthesized as described45. Mice. C57BL/6 mice were obtained from the Jackson Laboratory or Taconic. Ja18-deficient mice and Myd88–/– mice on the C57BL/6 background were gifts of M. Taniguchi (Riken Research Center for Allergy & Immunology, Yokohama, Japan) and S. Akira (Osaka University, Osaka, Japan) respectively. TrifLps2/Lps2 mice have been reported26. All mice were housed in specific pathogen–free conditions, and experiments were approved by the Institutional Animal Care and Use Committee of the La Jolla Institute of Allergy & Immunology (La Jolla, California). B. burgdorferi infection. Virulent B. burgdorferi strain 297, originally isolated from the cerebrospinal fluid of a patient with Lyme disease, was cultured as described46,47. Spirochetes were grown in Barbour-Stoenner-Kelly medium to mid- to late-log phase at 34 1C before being grown at 23 1C. Organisms grown at 23 1C were then shifted to a temperature of 37 1C and were grown to mid- to late-log phase before use in experiments. Spirochetes were counted by darkfield microscopy. Rearing, infection and infestation of I. scapularis ticks were done as described48. Nymphal ticks derived only from feedings that resulted in essentially 100% infection were used in this study. For B. burgdorferi infection, mice were infested with four nymphs confined in a capsule placed on the back. B. burgdorferi lysate. For the generation of B. burgdorferi lysates, spirochetes were washed three times with PBS, followed by resuspension in PBS, sonication and storage at –20 1C until use. Mouse DCs were prepared by culture of bone marrow progenitor cells with mouse GM-CSF (PeproTech) for 6 d, after which DCs were incubated with B. burgdorferi lysate at a concentration equivalent to 1 107 spirochetes/ml. After being washed with PBS, 5 105 lysate-pulsed DCs were injected intravenously into mice. Liver mononuclear cells positive for the a-GalCer–CD1d tetramer were analyzed 14 h later for activation marker expression and intracellular cytokine production. Cell-free antigen presentation assay. CD1d-reactive hybridomas have been described10,49. Published protocols10,49 with slight modifications were used for stimulation of T cell hybridomas on CD1d-coated plates. Various amounts of compounds were incubated for 24 h in microwells coated with 0.5–1.0 mg of mouse CD1d. After being washed, 5 104 to 1 105 Va14i NKT cell hybridomas or control hybridomas were cultured in plates for 16 h, and IL-2 in supernatants was measured by enzyme-linked immunosorbent assay (ELISA; BD PharMingen). CD1d tetramers. Tetramers of mouse CD1d loaded with a-GalCer and BbGL compounds were produced as described16,23, with slight modifications. Purified BbGL-I and BbGL-II, synthetic BbGL-IIc and a-GalCer were dissolved in vehicle (0.5% polysorbate-20 in 0.9% NaCl). CD1d molecules were incubated at a molar ratio of 1:3 (CD1d/glycolipid) with a-GalCer and at a molar ratio of 1:15 with BbGL-I, BbGL-II and BbGL-IIc. As a negative control, CD1d molecules were incubated with an equivalent volume of vehicle. After 24 h, CD1d-glycolipid complexes were formed into tetramers by the addition of phycoerythrin-conjugated streptavidin (BD PharMingen or Molecular Probes). In vitro NKT cell activation. Splenocytes (4 105) from wild-type C57BL/6, Myd88–/– or Trif Lps2/Lps2 mice were cultured for 5 d with 0.1 mg/ml of a-GalCer,

VOLUME 7

NUMBER 9

SEPTEMBER 2006

NATURE IMMUNOLOGY

ARTICLES


10 mg/ml of BbGL compounds or dimethyl sulfoxide vehicle in 96-well roundbottomed plates containing RPMI 1640 medium supplemented with glutamine, penicillin and streptomycin, b-mercaptoethanol and 10% FBS. For BrdU labeling, BrdU (Sigma) was added to the culture at a final concentration of 10 mM at 18 h before analysis. Cells were collected and were stained with a BrdU flow cytometry kit (BD PharMingen). Autofluorescent cells were gated out before measurement of BrdU incorporation by cells positive for the a-GalCer–CD1d tetramer. In vivo NKT cell response and flow cytometry. Published protocols10 were used for staining of a-GalCer–CD1d tetramer–positive cells for activation markers and intracellular cytokines. Mouse DCs were incubated for 24 h with 0.1 mg/ml of a-GalCer or 10 mg/ml of BbGL-IIc. After being washed with PBS, glycolipid-pulsed DCs (5 105) were injected intravenously into mice. Liver mononuclear cells positive for the a-GalCer–CD1d tetramer were analyzed directly ex vivo 14 h later for activation markers and for intracellular cytokines without pretreatment with brefeldin. Cells were analyzed with a FACSCalibur (BD Bioscience) and FlowJo software. Human iNKT cell response. Human Va24i NKT cell lines were generated with modifications to a published protocol10. Va24i+ T cells were isolated from Leukopaks (New York Blood Center) using magnetic beads (Miltenyi Biotec) coupled to a monoclonal antibody to Va24i and were cultured for 10 d with immature DCs in the presence of 100 ng/ml of a-GalCer and 10 IU/ml of IL-2. After a second stimulation with a-GalCer-pulsed irradiated immature DCs, cell lines were 99% Va24i+. Va24i iNKT cells (3 104) were cultured with irradiated (10,000 rads) human CD1d–transfected HeLa cells (3 104) in the presence of glycolipids. The concentration of IFN-g or IL-4 in the supernatants was determined by ELISA after 24 h (BD PharMingen). Molecular modeling. Crystal structures of mouse CD1d-lipid complexes, including CD1d-bound to the short-chain a-GalCer analog PBS-25 (Protein Data Bank accession number, 1Z5L37) and to PIM2 (Protein Data Bank accession number, 2GAZ27) were superimposed. All B. burgdorferi glycolipids were modeled with their alkyl chains mimicking the PIM2 lipid structure, in but with the galactose headgroup adopting an orientation similar to that of a-GalCer. Individual structures were then subjected to the structure idealization procedure as implemented in the program Refmac 5 (ref. 50). Note: Supplementary information is available on the Nature Immunology website.

ACKNOWLEDGMENTS We thank M.J. Caimano (University of Connecticut Health Science Center, Farmington, Connecticut) for B. burgdorferi–infected ticks; E. Janssen and S. McBride (La Jolla Institute for Allergy & Immunology, La Jolla, California) for TrifLps2/Lps2 mice; V. Kumar (Torrey Pines Institute for Molecular Studies, San Diego, California) for the sulfatide-reactive T cell hybridoma; N. Nagarajan, R. Severins, Y.W. Zhu (La Jolla Institute for Allergy & Immunology, La Jolla, California) and P. Rogers (Gemini Science, San Diego, California) for technical assistance; and G. Kim (La Jolla Institute for Allergy & Immunology, La Jolla, California) and M.A. Poles (New York University School of Medicine, New York, New York) for suggestions. Supported by the National Institutes of Health (AI45053 and AI71922 to M.K.; GM62116 to M.K. and I.A.W.; GM44154 to C.-H.W.; AI054546 to T.J.S.; CA58896 to I.A.W.; and AI062842 to M.T.), the Arthritis Foundation (T.J.S.) and the Cancer Research Institute (Y.K.). AUTHOR CONTRIBUTIONS Y.K. and E.T. designed and did most of the experiments and prepared the manuscript; D.W., M.F. and C.-H.W. synthesized the BbGL-II compounds and helped with the manuscript; R.G.-N. and M.T. did the human NKT cell experiments; M.R.B., T.J.S. and E.T. did the experiments with live bacteria; D.M.Z. and I.A.W. did the molecular modeling of the binding of BbGL-IIc in the CD1d structure and helped with the manuscript; G.B.-M. provided purified BbGL-I and BbGL-II; G.D.A. and G.F.P. provided the PIM4 compounds; A.K. prepared CD1d proteins; K.H. and B.B. generated the TrifLps2/Lps2 mice; S.M.B. generated the hybridoma 24.9E; and M.K. provided overall supervision, helped design all the experiments and prepared the manuscript. COMPETING INTERESTS STATEMENT The authors declare that they have no competing financial interests.

NATURE IMMUNOLOGY

VOLUME 7

NUMBER 9

SEPTEMBER 2006

Published online at http://www.nature.com/natureimmunology/ Reprints and permissions information is available online at http://npg.nature.com/ reprintsandpermissions/ 1. Godfrey, D.I. & Kronenberg, M. Going both ways: immune regulation via CD1ddependent NKT cells. J. Clin. Invest. 114, 1379–1388 (2004). 2. Kronenberg, M. Toward an understanding of NKT cell biology: progress and paradoxes. Annu. Rev. Immunol. 23, 877–900 (2005). 3. Morita, M. et al. Structure-activity relationship of a-galactosylceramides against B16bearing mice. J. Med. Chem. 38, 2176–2187 (1995). 4. Kawano, T. et al. CD1d-restricted and TCR-mediated activation of Va14 NKT cells by glycosylceramides. Science 278, 1626–1629 (1997). 5. Skold, M. & Behar, S.M. Role of CD1d-restricted NKT cells in microbial immunity. Infect. Immun. 71, 5447–5455 (2003). 6. Brigl, M. & Brenner, M.B. CD1: antigen presentation and T cell function. Annu. Rev. Immunol. 22, 817–890 (2004). 7. Kinjo, Y. & Kronenberg, M. Va14i NKT cells are innate lymphocytes that participate in the immune response to diverse microbes. J. Clin. Immunol. 25, 522–533 (2005). 8. Brigl, M., Bry, L., Kent, S.C., Gumperz, J.E. & Brenner, M.B. Mechanism of CD1drestricted natural killer T cell activation during microbial infection. Nat. Immunol. 4, 1230–1237 (2003). 9. Mattner, J. et al. Exogenous and endogenous glycolipid antigens activate NKT cells during microbial infections. Nature 434, 525–529 (2005). 10. Kinjo, Y. et al. Recognition of bacterial glycosphingolipids by natural killer T cells. Nature 434, 520–525 (2005). 11. Sriram, V., Du, W., Gervay-Hague, J. & Brutkiewicz, R.R. Cell wall glycosphingolipids of Sphingomonas paucimobilis are CD1d-specific ligands for NKT cells. Eur. J. Immunol. 35, 1692–1701 (2005). 12. Kawahara, K., Moll, H., Knirel, Y.A., Seydel, U. & Zahringer, U. Structural analysis of two glycosphingolipids from the lipopolysaccharide-lacking bacterium Sphingomonas capsulata. Eur. J. Biochem. 267, 1837–1846 (2000). 13. Kawahara, K., Kubota, M., Sato, N., Tsuge, K. & Seto, Y. Occurrence of an agalacturonosyl-ceramide in the dioxin-degrading bacterium Sphingomonas wittichii. FEMS Microbiol. Lett. 214, 289–294 (2002). 14. Orloski, K.A., Hayes, E.B., Campbell, G.L. & Dennis, D.T. Surveillance for Lyme disease–United States, 1992–1998. MMWR CDC Surveill. Summ. 49, 1–11 (2000). 15. Kumar, H., Belperron, A., Barthold, S.W. & Bockenstedt, L.K. Cutting edge: CD1d deficiency impairs murine host defense against the spirochete. Borrelia burgdorferi. J. Immunol. 165, 4797–4801 (2000). 16. Matsuda, J.L. et al. Tracking the response of natural killer T cells to a glycolipid antigen using CD1d tetramers. J. Exp. Med. 192, 741–754 (2000). 17. Benlagha, K., Weiss, A., Beavis, A., Teyton, L. & Bendelac, A. In vivo identification of glycolipid antigen-specific T cells using fluorescent CD1d tetramers. J. Exp. Med. 191, 1895–1903 (2000). 18. Ben-Menachem, G., Kubler-Kielb, J., Coxon, B., Yergey, A. & Schneerson, R. A newly discovered cholesteryl galactoside from Borrelia burgdorferi. Proc. Natl. Acad. Sci. USA 100, 7913–7918 (2003). 19. Schroder, N.W. et al. Acylated cholesteryl galactoside as a novel immunogenic motif in Borrelia burgdorferi sensu stricto. J. Biol. Chem. 278, 33645–33653 (2003). 20. Cinco, M., Banfi, E., Balanzin, D., Godeas, C. & Panfili, E. Evidence for (lipo) oligosaccharides in Borrelia burgdorferi and their serological specificity. FEMS Microbiol. Immunol. 3, 33–38 (1991). 21. Eiffert, H., Lotter, H., Jarecki-Khan, K. & Thomssen, R. Identification of an immunoreactive non-proteinaleous component in Borrelia burgdorferi. Med. Microbiol. Immunol. (Berl.) 180, 229–237 (1991). 22. Wheeler, C.M. et al. Nonprotein antigens of Borrelia burgdorferi. J. Infect. Dis. 167, 665–674 (1993). 23. Wu, D. et al. Design of natural killer T cell activators: structure and function of a microbial glycosphingolipid bound to mouse CD1d. Proc. Natl. Acad. Sci. USA 103, 3972–3977 (2006). 24. Fischer, K. et al. Mycobacterial phosphatidylinositol mannoside is a natural antigen for CD1d-restricted T cells. Proc. Natl. Acad. Sci. USA 101, 10685–10690 (2004). 25. Fujii, S., Shimizu, K., Kronenberg, M. & Steinman, R.M. Prolonged IFN-g-producing NKT response induced with a-galactosylceramide-loaded DCs. Nat. Immunol. 3, 867–874 (2002). 26. Hoebe, K. et al. Identification of Lps2 as a key transducer of MyD88-independent TIR signalling. Nature 424, 743–748 (2003). 27. Zajonc, D.M., Ainge, G.D., Painter, G., Severn, W.B. & Wilson, I.A. Structural characterization of mycobacterial phosphatidylinositol mannoside binding to mouse CD1d. J. Immunol. (in the press). 28. Brossay, L. et al. CD1d-mediated recognition of an a-galactosylceramide by natural killer T cells is highly conserved through mammalian evolution. J. Exp. Med. 188, 1521–1528 (1998). 29. Neef, A., Witzenberger, R. & Kampfer, P. Detection of sphingomonads and in situ identification in activated sludge using 16S rRNA-targeted oligonucleotide probes. J. Ind. Microbiol. Biotechnol. 23, 261–267 (1999). 30. Duthie, M.S., Kahn, M., White, M., Kapur, R.P. & Kahn, S.J. Critical proinflammatory and anti-inflammatory functions of different subsets of CD1d-restricted natural killer T cells during Trypanosoma cruzi infection. Infect. Immun. 73, 181–192 (2005). 31. Wooten, R.M. & Weis, J.J. Host-pathogen interactions promoting inflammatory Lyme arthritis: use of mouse models for dissection of disease processes. Curr. Opin. Microbiol. 4, 274–279 (2001).

985


ARTICLES 32. Zhou, D. et al. Lysosomal glycosphingolipid recognition by NKT cells. Science 306, 1786–1789 (2004). 33. Gumperz, J.E. et al. Murine CD1d-restricted T cell recognition of cellular lipids. Immunity 12, 211–221 (2000). 34. Rauch, J. et al. Structural features of the acyl chain determine self-phospholipid antigen recognition by a CD1d-restricted invariant NKT (iNKT) cell. J. Biol. Chem. 278, 47508–47515 (2003). 35. Brigl, M. et al. Conserved and heterogeneous lipid antigen specificities of CD1drestricted NKT cell receptors. J. Immunol. 176, 3625–3634 (2006). 36. Wu, D.Y., Segal, N.H., Sidobre, S., Kronenberg, M. & Chapman, P.B. Cross-presentation of disialoganglioside GD3 to natural killer T cells. J. Exp. Med. 198, 173–181 (2003). 37. Zajonc, D.M. et al. Structure and function of a potent agonist for the semi-invariant natural killer T cell receptor. Nat. Immunol. 6, 810–818 (2005). 38. Koch, M. et al. The crystal structure of human CD1d with and without a-galactosylceramide. Nat. Immunol. 6, 819–826 (2005). 39. Zeng, Z. et al. Crystal structure of mouse CD1: An MHC-like fold with a large hydrophobic binding groove. Science 277, 339–345 (1997). 40. Zajonc, D.M. et al. Structural basis for CD1d presentation of a sulfatide derived from myelin and its implications for autoimmunity. J. Exp. Med. 202, 1517–1526 (2005). 41. Miyamoto, K., Miyake, S. & Yamamura, T. A synthetic glycolipid prevents autoimmune encephalomyelitis by inducing TH2 bias of natural killer T cells. Nature 413, 531–534 (2001).

986

42. Goff, R.D. et al. Effects of lipid chain lengths in a-galactosylceramides on cytokine release by natural killer T cells. J. Am. Chem. Soc. 126, 13602–13603 (2004). 43. Yu, K.O. et al. Modulation of CD1d-restricted NKT cell responses by using N-acyl variants of a-galactosylceramides. Proc. Natl. Acad. Sci. USA 102, 3383–3388 (2005). 44. Slomiany, B.L., Murty, V.L., Liau, Y.H. & Slomiany, A. Animal glycoglycerolipids. Prog. Lipid Res. 26, 29–51 (1987). 45. Ainge, G.D. et al. Phosphatidylinositol mannosides: synthesis and adjuvant properties of phosphatidylinositol di- and tetramannosides. Bioorg. Med. Chem. (in the press). 46. Sellati, T.J. et al. Activation of human monocytic cells by Borrelia burgdorferi and Treponema pallidum is facilitated by CD14 and correlates with surface exposure of spirochetal lipoproteins. J. Immunol. 163, 2049–2056 (1999). 47. Benhnia, M.R. et al. Signaling through CD14 attenuates the inflammatory response to Borrelia burgdorferi, the agent of Lyme disease. J. Immunol. 174, 1539–1548 (2005). 48. Yang, X. et al. Identification, characterization, and expression of three new members of the Borrelia burgdorferi Mlp (2.9) lipoprotein gene family. Infect. Immun. 67, 6008–6018 (1999). 49. Sidobre, S. et al. The Va14 NKT cell TCR exhibits high-affinity binding to a glycolipid/ CD1d complex. J. Immunol. 169, 1340–1348 (2002). 50. Murshudov, G.N., Vagin, A.A. & Dodson, E.J. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240–255 (1997).

VOLUME 7

NUMBER 9

SEPTEMBER 2006

NATURE IMMUNOLOGY