Interplay of and Microbial Signals in Regulation of CD1d Expression and NKT Cell Activation

This information is current as Markus Sköld, Xiaowei Xiong, Petr A. Illarionov, Gurdyal S. of September 27, 2021. Besra and Samuel M. Behar J Immunol 2005; 175:3584-3593; ; doi: 10.4049/jimmunol.175.6.3584 http://www.jimmunol.org/content/175/6/3584 Downloaded from

References This article cites 46 articles, 28 of which you can access for free at: http://www.jimmunol.org/content/175/6/3584.full#ref-list-1 http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 27, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2005 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Interplay of Cytokines and Microbial Signals in Regulation of CD1d Expression and NKT Cell Activation1

Markus Sko¨ld,* Xiaowei Xiong,* Petr A. Illarionov,† Gurdyal S. Besra,† and Samuel M. Behar2*

In this study we show that like MHC class I and class II molecules, cell surface CD1d expression on APC is regulated and affects activation under physiological conditions. Although IFN-␥ alone is sufficient for optimum expression of MHC, CD1d requires two signals, one provided by IFN-␥ and a second mediated by microbial products or by the proinflammatory TNF. IFN-␥-dependent CD1d up-regulation occurs on macrophages following infection with live or exposure to microbial products in vitro and in vivo. APC expressing higher CD1d levels more efficiently activate NKT cell hybridomas and primary NKT cells independently of whether the CD1d-restricted TCR recognizes foreign or self-lipid Ags. Our findings support a model in which CD1d induction regulates NKT cell activation. The Journal of Immunology, 2005, 175: 3584–3593. Downloaded from

ntigen-presenting molecules play a fundamental role in itself a suitable target of CD8ϩ T cells by increasing the efficiency T cell development, T cell priming, and regulation of of Ag processing and the number of Ag-MHC complexes ex- A immunity. The cell surface expression of Ag-presenting pressed at the cell surface. The importance of class I MHC-anti- molecules is particularly critical for host defense against intracel- genic peptide complexes for the host is best appreciated by the

lular pathogens as the presentation of microbial Ags by class I or number of viruses for which their success as pathogens is http://www.jimmunol.org/ class II MHC molecules is the principle way that the immune related to their interference with class I MHC expression and Ag system identifies infected cells. Although Ag-presenting molecules processing (7). are constitutively expressed by many cells, they are also regulated Like class I and class II MHC, the Ag-presenting molecule by various stimuli such as cytokines and microbial products. CD1d is constitutively expressed by many cell types. It is prom- MHC class II is constitutively expressed by B cells, dendritic inently expressed by splenic B cells, and CD1d is found on M␾, cells (DC),3 and macrophages (M␾); it is also induced by cyto- DC, and even T cells (8). Whether cell surface expression of CD1d kines, principally IFN-␥, on a variety of cells including M␾, en- is subject to additional regulation by cytokines or microbial prod- dothelial, and epithelial cells (1). MHC class II is also modulated ucts, as observed for MHC, is an important question. Although during cell differentiation as exemplified by its redistribution from class I and class II MHC induction increases the efficiency of T cell by guest on September 27, 2021 the intracellular MHC class II compartment to the cell surface activation, it is unknown whether increased surface expression of during DC maturation (2). The class II MHC redistribution that CD1d will do the same. The observation that mice lacking CD1d- accompanies DC maturation can be induced by TNF or by micro- restricted NKT cells have diminished host resistance to certain bial products such as the TLR4 ligand LPS (3–5). pathogens, impaired tumor immunity, and alterations in their pre- Although MHC class I is constitutively expressed by nearly ev- disposition to autoimmune disease indicates that NKT cells par- ery nucleated cell, it too is regulated by cytokines. IFN-␥ increases ticipate in these immunological responses (9, 10). There are two cell surface levels of MHC class I on a variety of cell types and emerging models of how lipid Ag presentation by CD1d activates regulates transcription of many involved in the MHC class NKT cells. I Ag-processing pathway (1, 6). These effects of IFN-␥ are critical NKT cell recognition of foreign lipids has been modeled using in T cell responses to viral pathogens as the infected cell makes the synthetic Ag ␣-galactosylceramide (␣-GalCer), which binds to CD1d and specifically activates invariant NKT (iNKT) cells that are defined by their canonical V␣14-J␣18 TCR-␣ chain (murine) *Division of Rheumatology, Immunology, and Allergy, Brigham and Women’s Hos- or V␣24-J␣18 TCR-␣ chain (human) (11, 12). In theory, microbial pital and Harvard Medical School, Boston, MA 02115; and †School of Biosciences, glycolipid Ags are so sufficiently different from host lipids that University of Birmingham, Edgbaston, United Kingdom CD1d-restricted NKT cells can recognize them as foreign. Re- Received for publication May 17, 2005. Accepted for publication June 28, 2005. cently, several microbial lipid Ags have been identified that are The costs of publication of this article were defrayed in part by the payment of page presented by CD1d and activate NKT cells including monoglyco- charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sylceramides from species, phosphatidylinositol 1 This work was supported by National Institutes of Health Grants R03 AI49093 and mannosides from Mycobacterium tuberculosis, and lipophospho- R01 HL80312 and by an Arthritis Foundation Investigator Award (to S.M.B.). G.S.B. glycan from Leishmania donovani (13–16). is a Lister-Jenner Research Fellow and acknowledges support from The Medical In addition to recognizing microbial lipids, both iNKT and NKT Research Council Grants G9901077 and G0400421 and The Wellcome Trust Grant 072021/Z/03/Z. cells expressing diverse TCR recognize CD1d in the absence of 2 Address correspondence and reprint requests to Dr. Samuel M. Behar, Division of exogenously added Ags (17). CD1d autoreactivity is Ag-depen- Rheumatology, Immunology, and Allergy, Brigham and Women’s Hospital, Smith dent and the lysosomal glycosphingolipid isoglobotrihexosylcer- Building Room 516B, One Jimmy Fund Way, Boston, MA 02115. E-mail address: amide (iGb3) has recently been suggested to be the primary [email protected] self-Ag for CD1d-autoreactive human and murine iNKT cells (18, 3 Abbreviations used in this paper: DC, dendritic cell; M␾, macrophage; ␣-GalCer, ␣-galactosylceramide; iNKT, invariant NKT; MFI, mean fluorescence intensity; MOI, 19). Cellular phospholipids also bind to CD1d, and reactivity to multiplicity of infection. phosphatidylethanolamine, phosphatidylinositol, and related Ags

Copyright © 2005 by The American Association of Immunologists, Inc. 0022-1767/05/$02.00 The Journal of Immunology 3585

does occur for some iNKT and NKT cells (18, 20). NKT cell Middlebrook 7H9 medium containing 10% albumin/dextrose/catalase en- recognition of self-lipids may be particularly important during richment (both from BD Biosciences). Bacteria were opsonized using noninfectious immunological responses including tumor immunity RPMI 1640 medium with 2% human serum (Gemini Bio-Products), 10% FBS, and 0.05% Tween 80 and then washed twice with complete medium and autoimmunity. Recognition of self-lipid Ags may even be im- without antibiotics. Bacteria were passed through a 5-␮m syringe filter portant for NKT cell activation during infection because IL-12 (Millipore), counted in a Petroff-Hausser chamber and added to purified induced by microbial pathogens costimulates activation of CD1d M␾ at different multiplicity of infection (MOI) as indicated. M␾ and bac- autoreactive human iNKT cells in vitro (21). Heat killed Salmo- teria were cultured in complete medium without antibiotics in the presence or absence of rIFN-␥ (2.5 U/ml). In some experiments, bacteria were heat nella presented by DC lacking the iGb3 endogenous Ag or an killed at 80°C for 20 min. The M␾ phenotype was analyzed by flow cy- intact TLR signaling pathway does not activate murine iNKT cells tometry, and cytokine production was analyzed by ELISA after 24 or 72 h (16). These data support the paradigm that one pathway for iNKT incubation. Mice were infected with virulent Mtb (Erdman strain) via the cell activation is the costimulation of weak self-Ag recognition by aerosol route as previously described (22). pathogen-induced IL-12 production. Importantly, recognition of T cell hybridoma and primary NKT cell assays self-lipid Ags also requires that NKT cell activation must be reg- All CD1d-restricted T cell hybridomas used have been previously de- ulated to avoid inappropriate responses. scribed (18, 23, 24). Hybridomas KT/7, KT/12, KT/22, and KT/23 were In studies designed to better understand how NKT cells are ac- kindly provided by Dr. S. Cardell (Lund University, Lund, Sweden) and tivated by infection, we discovered an IFN-␥-dependent mecha- hybridoma DN32.D3 by Dr. A. Beandelac (University of Chicago, Chi- nism by which bacterial infection or exposure to microbial prod- cago, IL). M␾ were plated in flat-bottom 96-well plates and kept in com- plete medium alone, or with rIFN-␥ (2.5 U/ml), Pam Cys (1 ng/ml), or ucts, including TLR ligands, induces CD1d expression on M␾ in 3 both for 3 days before the T cell hybridoma cells were added. CD1d vitro and in vivo. The effect of TLR ligands and live bacteria on autoreactivity was tested by seeding M␾ at different cell densities. A con- Downloaded from CD1d induction is mediated by the proinflammatory cytokine stant number of M␾ were used to test for ␣-GalCer reactivity. TNF. This observation suggested that modulation of cell surface Gal(␣132)GalCer (␣-GalGalCer) was added to M␾ thelast6hofthe CD1d levels on APC could be a mechanism that regulates NKT 3-day culture period. The complete synthesis of ␣-GalGalCer will be re- ported elsewhere (P. A. Illarionov and G. S. Besra, unpublished observa- cell activation. We found that both NKT cell hybridomas and pri- tions). A complete structural analysis of ␣-GalGalCer was performed by mary NKT cells are activated more efficiently by APC expressing nuclear magnetic resonance and electrospray mass spectrometry (data not 4 higher cell surface CD1d levels. Both iNKT cell and diverse NKT shown). Cells were washed once before the T cell hybridomas (5 ϫ 10 / http://www.jimmunol.org/ cell populations share this mechanism and it is independent of well) were added. RMA-S.CD1d cells were sorted into different sublines whether the CD1d-restricted TCR recognizes foreign or self-lipid based on their CD1d surface expression. These RMA-S.CD1d sublines or RMA-S nontransfected cells were seeded into 96-well plates (105/well). In Ags. These data provides a new basis for understanding how some assays, RMA-S cells were pulsed with 10 ng/ml ␣-GalCer for6hat CD1d-restricted NKT cells are activated during inflammation. 37°C before adding ␣-GalCer-reactive T cell hybridomas (105/well). T cell

hybridoma assays were incubated for 24 h at 37°C, 5% CO2, before IL-2 Materials and Methods was measured in the culture supernatants by ELISA (reagents from BD ␾ Mice Pharmingen). CD1d cell surface expression by M and RMA-S cells was monitored in each experiment using flow cytometry. B6 and B6.129S7-Ifngrtm1Agt/J (IFN-␥RϪ/Ϫ) mice were obtained from The Primary NKT cells were enriched by depleting thymocytes of CD8ϩ Jackson Laboratory. Mice were housed under specific pathogen-free con- cells using magnetic cell sorting with CD8␣-microbeads (Miltenyi Biotec). by guest on September 27, 2021 ditions and were used in a protocol approved by the institution. Six- to Enriched thymic NKT cells (0.2–1 ϫ 106) were cultured with ␣-GalCer/ 8-wk-old B6 mice were used as a source of primary NKT cells and as vehicle-pulsed APC (105/well), with or without rIL-2 (Chiron) at 100 U/ml recipient mice in the M␾ adoptive transfer experiments described below. as indicated. RMA-S cells were irradiated at 2000 rad before they were Mycobacterium tuberculosis (Mtb)-infected mice were housed in a bio- used as APC. IL-4 and IFN-␥ was measured in the culture supernatants by safety level 3 facility at the Animal Biohazard Containment Suite (Dana- ELISA (reagents from BD Pharmingen) after incubation for 48 h at 37°C, Farber Cancer Institute) and were used in an approved protocol. B6 and 5% CO2. IFN-␥RϪ/Ϫ mice were used as a source of inflammatory M␾. Detection of CD1d regulation in vivo M␾ and in vitro cultures An adoptive transfer system was used to examine CD1d and class II MHC Inflammatory M␾ were elicited by i.p. injection of 1-ml sterile 3% thio- regulation at the site of infection in vivo. Purified thioglycolate elicited glycolate medium (REMEL). Peritoneal exudate cells were harvested by peritoneal M␾ were obtained from B6 or IFN-␥RϪ/Ϫ mice as described lavage after 4 days. M␾ were purified from peritoneal exudate cells by earlier and labeled with 1 ␮M CFSE (Molecular Probes). A total of 1 ϫ 107 magnetic cell sorting with CD11b microbeads (Miltenyi Biotec). Positively CFSE-labeled M␾ were injected i.v. via the tail vein into Mtb-infected B6 selected cells routinely contained ϳ95% F4/80ϩCD11bϩ M␾ as deter- recipients or age- and sex-matched uninfected B6 mice. Transfers were mined by flow cytometry. Enriched M␾ (1 ϫ 106) were seeded into 24- performed 4–5 wk after aerosol infection during the peak of the immune well plates in complete culture medium (RPMI 1640 (Invitrogen Life response. Three or five recipient mice were used per group in five separate Technologies) supplemented with 10% FCS (HyClone), penicillin/strepto- experiments. At 48 or 72 h after cell transfer, single cell suspensions were mycin, L-glutamine, sodium-pyruvate, 2-ME, nonessential amino acids, es- prepared from lung tissue as described with the modification that collage- sential amino acids, and HEPES buffer, all from Invitrogen Life Technol- nase digested lung tissue was treated with 200 U/ml DNase I (Sigma- ϫ ogies). Synthetic TLR2 ligand Pam3Cys-Ser-(Lys)3 3 HCl was obtained Aldrich) for 10 min at 37°C (22). from EMC and ultra pure TLR4 ligand Escherichia coli LPS (O111:B4 strain) from InvivoGen. Recombinant mouse IFN-␥, IL-12p70, and TNF Detection of iNKT cell activation in vivo (US Biological) were used at the concentrations indicated. Supernatants An adoptive transfer system using ␣-GalCer-pulsed M␾ and an IFN-␥ ␾ and M were harvested at the time points indicated after cell culture at ELISPOT assay was adapted from Fujii et al. (25) and used to detect ␾ 37°C, 5% CO2.M were incubated with 2 mM EDTA in PBS for 10 min activation of iNKT cells in vivo. Purified thioglycolate-elicited peritoneal at 37°C to detach adherent cells for flow cytometry. NaN3 (2 mM) was M␾ were obtained from B6 mice as described above and cultured in me- added to the supernatant before their cytokine content was measured using ␥ dium alone or with rIFN- (2.5 U/ml) and Pam3Cys (1 ng/ml) for 3 days. ELISA (reagents from eBioscience). To examine the contribution of TNF M␾ were pulsed with ␣-GalCer (100 ng/ml) for 6 h, washed with PBS and in CD1d regulation by TLR agonists and live Mtb, a blocking anti-TNF injected i.v. via the tail vein into naive B6 recipients (106 M␾/mouse). At mAb (MP6-XT3) and an isotype control Ab (no azide/low endotoxin; BD 48 h after M␾ adoptive transfer, single cell suspensions were prepared ␮ Pharmingen) were used (40 g/ml). from the spleens of the recipient mice or from naive B6 control mice. Total Bacteria and infections splenocytes were cultured with ␣-GalCer (100 ng/ml) or with vehicle (DMSO) in ELISPOT plates precoated with anti-IFN-␥ capture mAb (all An in vitro infection model was used to analyze the effect of live Mtb on ELISPOT reagents from BD Pharmingen). The cells were incubated for ␾ CD1d and class II MHC expression by inflammatory M . Virulent Mtb 16 h at 37°C, 5% CO2, before they were discarded and the plates washed (H37Rv) or avirulent Mtb (H37Ra strain) was grown to mid-log phase in with deionized water and PBS/Tween 20. The plates were incubated with 3586 REGULATION OF CD1d EXPRESSION MODULATES NKT CELL ACTIVATION a biotinylated secondary mAb, washed, and incubated with streptavidin- HRP. After several washes the plates were developed using 3-amino-9- ethylcarbazole substrate. The spots were enumerated using an ImmunoSpot plate reader and ImmunoSpot software version 3 (Cellular Technology). Flow cytometry

PBS with 1% w/v BSA and 2 mM NaN3 were used to wash the single-cell suspensions and to dilute Abs and second step reagents. Cells were incu- bated with purified anti-CD16/CD32 (2.4G2; ATCC HB-197) at 25 ␮g/ml to inhibit nonspecific staining. The following PE-, FITC-, or PE-Cy5-con- jugated, or biotinylated mAbs and second step reagents were obtained from BD Pharmingen: anti-CD1d (1B1), anti-I-A/I-E (M5/114.15.2), anti- CD11b (M1/70), anti-CD11c (HL3), anti-CD22.2 (Cy34.1), anti-CD40 (3/ 23), anti-CD80 (16-10A1), anti-CD86 (GL1), anti-Ly6C/G (RB6-8C5), and appropriate isotype control Abs and streptavidin. PE-conjugated anti- CD115 (AFS98) was purchased from eBioscience and FITC-conjugated anti-CD205 (MCA949F) was purchased from Serotec. The F4/80 mAb (ATCC HB-198) and the M1/42 mAb (ATCC TIB-126) were purified and biotinylated using standard protocols or were conjugated to Alexa Fluor 488 using a labeling from Molecular Probes. Stained cells were washed and analyzed directly or fixed overnight at 4°C in 1% paraformaldehyde in PBS. Cells were collected using a FACSort or a FACSCanto flow cyto- meter (BD Biosciences) and analyzed using CellQuest (BD Biosciences) or Downloaded from FlowJo software (Tree Star). Real-time PCR M␾ were purified and cultured as described above. After 24 h of culture, total RNA was prepared using TRIzol Reagent (Invitrogen Life Technol- ogies) followed by DNase I (amplification grade) treatment according to the manufacturer’s instructions (Invitrogen Life Technologies). RNA was http://www.jimmunol.org/ reverse transcribed to single-stranded cDNA using SuperScript II reverse transcriptase and oligo(dT) to prime first-strand synthesis (Invitrogen Life Technologies). The following primers were used for real-time PCR am- plification using an iTaq SYBR Green Supermix kit from Bio-Rad and an iCycler real-time PCR detection system (Bio-Rad): CD1d forward 5Ј-TGT CACCTAAAGAAGACTATCCCATTG-3Ј, reverse 5Ј-CCGAAGCATTC CCAGGGTA-3Ј; and ␤-actin forward 5Ј-CATCTTGGTTAGACTTGC CCAT-3Ј, reverse 5Ј-GGAGACCACGGACAAATAGGG-3Ј. Samples were incubated at 50°C for 2 min followed by 10 min incubation at 95°C. FIGURE 1. CD1d up-regulation on Mtb-infected M␾ is IFN-␥-depen-

Each amplification cycle consisted of 15 s at 95°C and 1 min at 60°C and dent. A, Thioglycolate-elicited purified peritoneal M␾ were infected in by guest on September 27, 2021 was repeated 40 cycles. The relative amount of CD1d mRNA was deter- vitro with live Mtb (H37Ra) at the indicated MOI in the presence (circles) ⌬⌬ mined by the threshold cycle Ct method relative to the expression of the or absence (squares) of rIFN-␥ (2.5 U/ml). CD1d (closed symbols, left) and housekeeping ␤-actin. ϩ class II MHC (closed symbols, right) expression on F4/80 M␾ was an- Results alyzed on day 3. Values are the MFI. B, Representative dot plots show the expression of CD1d (top panels) and class II MHC (bottom panels)by Cytokine mediated up-regulation of CD1d on elicited peritoneal ϩ ␾ ␾ F4/80 purified M cultured for 3 days in vitro alone, or in the presence M of rIFN-␥, live Mtb (MOI 5:1), or both rIFN-␥ and bacteria. C, CD1d Several pathogens cause worse disease in mice that lack CD1d or (closed symbols, left) or class II MHC (closed symbols, right) expression ϩ iNKT cells, suggesting that CD1d-restricted NKT cells play a by purified F4/80 inflammatory M␾ was analyzed 4 days after culture ␥ physiologic role in host defense (9). Implicitly, these studies indi- with rIFN- (circles), rIL-12-p70 (triangles), or rTNF (squares). Open cate that following infection CD1d-restricted NKT cells must be- symbols show Ab isotype control staining. come activated. Modulation of CD1d cell surface expression on APC may be one mechanism that regulates NKT cell activity. induce cell surface expression of CD1d and class II MHC. When Therefore, we developed an in vitro model using elicited peritoneal M␾ were infected with Mtb in the presence of low amounts of M␾ to test whether bacterial infection and inflammatory mediators rIFN-␥, CD1d was induced in a dose-dependent manner (Fig. 1A). modulate cell surface expression of CD1d. Uninfected M␾ cultured in the absence of rIFN-␥ expressed low Highly purified F4/80ϩCD11bϩ M␾ expressed low levels of CD1d levels as mean fluorescence intensity (MFI ϭ 22), whereas CD1d and were heterogeneous for class II MHC (data not shown). heavily infected M␾ cultured in the presence of rIFN-␥ expressed These M␾ expressed CD115 and low levels of CD40, CD80, and high CD1d levels (MFI ϭ 238), representing more than a 10-fold CD86, but not Ly6C/G, CD11c, or CD205 (data not shown). M␾ increase in CD1d cell surface expression. Under these conditions, were infected with Mtb, and CD1d and class II MHC expression the induction of class II MHC by rIFN-␥ was unaffected by the was analyzed 72 h later. Infection of highly purified M␾ did not bacteria, except at the highest MOI where Mtb had a slight inhib- affect cell surface expression of CD1d or MHC class II (Fig. 1A). itory effect on IFN-␥-induced class II MHC surface expression IFN-␥ is the principal M␾ activating factor and affects multiple (Fig. 1, A and B). Still, the inhibition of class II MHC was modest aspects of M␾ function including induction of bactericidal activity compared with induction of CD1d under the same culture condi- and cell surface expression of class II MHC, which facilitates rec- tions. Similar results were obtained using the virulent H37Rv ognition by CD4ϩ T cells. Microarray analysis of bone marrow- strain of Mtb (data not shown). These data demonstrate that CD1d derived M␾ has previously shown that transcription of CD1d and expression by M␾ is not static and that microbial pathogens and MHC class II are induced following stimulation with IFN-␥ IFN-␥ act synergistically to induce cell surface CD1d. and Mtb (see supplemental data in Ref. 26). Therefore, we tested Even in the absence of bacteria, rIFN-␥ induced CD1d and class the possibility that IFN-␥ treatment of Mtb-infected M␾ would II MHC on uninfected M␾ (Fig. 1, A and B). To determine the The Journal of Immunology 3587

optimal conditions for the maximal induction of CD1d and class II viable bacteria, CD1d was not significantly induced in the absence MHC by rIFN-␥ on cultured inflammatory M␾, rIFN-␥ was ti- of rIFN-␥, even at a ratio of 50:1 (heat killed bacteria to M␾) (data trated over a large concentration range. rTNF and rIL-12p70 were not shown). A reverse correlation between CD1d and class II MHC also individually tested as these cytokines can be produced by expression was observed following treatment of M␾ with heat infected M␾ and might also modulate CD1d and class II MHC killed bacteria and rIFN-␥. Although this combination dramati- expression. Even at low concentrations, rIFN-␥ was sufficient to cally induced CD1d expression, class II MHC was inhibited in a up-regulate CD1d on M␾ and did so in a dose-dependent manner dose-dependent manner (Fig. 2A). Microbial pathogens including (Fig. 1C). In contrast, rTNF and rIL-12p70 had no effect on CD1d Mtb are detected by the via TLR signaling. Al- expression. Treatment with rIFN-␥ led to a 4-fold induction of though TLR agonists alone did not affect CD1d expression, low ␾ CD1d compared with medium treated M . MHC class II was also concentrations of Pam3Cys (a TLR2 agonist) or LPS (a TLR4 up-regulated by rIFN-␥ as expected, and to some extent by rIL- agonist) both dramatically induced CD1d in the presence of 12p70 at the highest concentration (Fig. 1C). rIFN-␥ (Fig. 2B). Similar to heat killed bacteria, TLR agonists The comparison between CD1d and class II MHC expression inhibited IFN-␥-induced class II MHC up-regulation as has been identifies a difference in the regulation of these two Ag-presenting previously reported (27). In the absence of IFN-␥, LPS but not

molecules. Maximum cell surface expression of class II MHC was Pam3Cys, induced class I MHC expression in a dose-dependent elicited by rIFN-␥ alone, whereas optimal induction of CD1d re- manner (Fig. 2B). IFN-␥ alone also induced class I MHC, and quires two signals mediated by rIFN-␥ and a microbial signal. addition of either TLR agonist had a minor effect on class I MHC cell surface expression. Under these conditions, CD1d induction ␥ Synergistic effect by IFN- and TLR agonists on CD1d cell could be detected within 24 h and was maximal within 2–3 days Downloaded from surface expression (Fig. 2C). Regulation of CD1d expression appears to be at the

What bacterial factor(s) regulate CD1d? The observed induction of RNA level because Pam3Cys alone had little effect on expression. CD1d did not require actively metabolizing bacteria and the in- In contrast, treatment with IFN-␥ led to an increase in CD1d RNA, ␥ ducing factor was heat stable because heat killed Mtb also induced and IFN- plus Pam3Cys synergistically increased the amount of CD1d in an IFN-␥-dependent manner (Fig. 2A). As observed with CD1d RNA relative to ␤-actin (Fig. 2D). http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 2. IFN-␥ and bacterial products have a synergistic effect on CD1d induction. CD11bϩ inflammatory M␾ were cultured alone or with heat killed (HK) Mtb (H37Ra strain) at the indicated MOI (A); or with the

TLR2 agonist Pam3Cys or the TLR4 agonist LPS at the indicated concentrations (B)inthe presence (circles) or absence (squares) of rIFN-␥ (2.5 U/ml). CD1d (closed symbols, left), class II MHC (closed symbols, center), and class I MHC (closed symbols, right) ex- pression by F4/80ϩCD11bϩ M␾ was exam- ined on day 3. The mean MFI Ϯ SEM of three to four experiments is shown in B. C, Purified M␾ were cultured alone (squares) or

stimulated with Pam3Cys (1 ng/ml) and rIFN-␥ (2.5 U/ml) (circles) for 24, 48, or 72 h. CD1d expression (filled symbols) by F4/80ϩ M␾ was determined by flow cytom- etry. Open symbols show Ab isotype control staining. D, Alteration of CD1d RNA levels relative to ␤-actin were determined by real- time RT-PCR. Fold increase was calculated by the ⌬⌬Ct method. 3588 REGULATION OF CD1d EXPRESSION MODULATES NKT CELL ACTIVATION

Synergistic effect of TNF and IFN-␥ in CD1d regulation TNF modulates CD1d expression by M␾ because of the consistent In addition to its synergistic effect on CD1d and class II MHC correlation between TNF production and CD1d induction and the ␥ expression, the combination of rIFN-␥ and heat killed bacteria or IFN- -dependence of TNF production (Fig. 3A). Although rTNF TLR agonists also modulates cytokine production by M␾ (Fig. 3A was not sufficient to up-regulate CD1d or MHC class II cell surface ␥ and data not shown). Highly purified M␾ cultured with TLR ago- expression (Fig. 1C), we hypothesized that TNF and IFN- may nists did not secrete detectable amounts of IL-12p40 and TNF, and act synergistically to regulate CD1d expression. In the presence of ␥ only small amounts of IL-6. Interestingly, the induction of all three rIFN- , rTNF induced CD1d in a dose-dependent manner and at cytokines by TLR agonists was IFN-␥-dependent. 10 ng/ml rTNF there was a 6.4-fold increase in surface levels (Fig. When M␾ infected at a low MOI were treated with rIFN-␥, 3B). Although at the highest concentration of rTNF there was a ␥ ␾ CD1d was induced on all cells, even though not all cells were 40% reduction in IFN- -dependent class II MHC expression, M infected (Fig. 1A). This result suggested that optimum CD1d in- treated in this way still expressed high class II MHC levels com- ␥ duction requires a second soluble factor. We considered whether pared with cells cultured with IFN- and TLR agonists (compare Figs. 2B and 3B). Thus, in the presence of rIFN-␥, neither rTNF nor viable Mtb dramatically alter class II MHC expression com- pared with uninfected M␾ cultured with rIFN-␥ alone (Figs. 1A and 3B). Purified M␾ infected with Mtb, treated with heat killed bacteria or cultured with TLR ligands, up-regulates CD1d in the presence

of rIFN-␥. Under the same conditions, these various stimuli also Downloaded from induce TNF production. Although neither molecule induced CD1d alone, both TLR agonists and rTNF could synergize with IFN-␥ to induce CD1d. To directly test the role of TNF in CD1d regulation by TLR agonists or Mtb in combination with rIFN-␥, the effect of TNF blockade on CD1d and class II MHC expression was exam-

ined. Blocking of TNF inhibited CD1d induction by Pam3Cys and http://www.jimmunol.org/ rIFN-␥ (Fig. 3C) and by Mtb infection and rIFN-␥ (Fig. 3D)by ϳ65%, but had only a minor influence on MHC class II expression by M␾ cultured in the absence of blocking anti-TNF Abs or iso- type control Abs. We speculate that the residual induction of CD1d was mediated by rIFN-␥ alone and was resistant to inhibition by anti-TNF mAb, or that the blocking conditions were not 100% efficient. These experiments demonstrate that the physiological amounts of TNF produced by M␾ following intracellular infection or after encounter with TLR ligands is sufficient to induce CD1d in by guest on September 27, 2021 the presence of rIFN-␥. We conclude that signaling via IFN-␥R and TLR is not sufficient to induce high CD1d levels on M␾, but also requires TNF production to mediate the effect. Taken together, these findings identify a mechanism that modulates CD1d expres- sion during inflammation caused by microbial pathogens or other etiologies. Finding that the proinflammatory cytokine TNF can replace the requirement for microbial products in the induction of high CD1d levels on M␾ may explain how noninfectious inflam- matory conditions can regulate CD1d surface expression and pro- mote the activation of self-reactive CD1d-restricted NKT cells.

FIGURE 3. TNF and IFN-␥ synergize to up-regulate CD1d on inflam- matory M␾. A, rIFN-␥ modulates cytokine production by inflammatory CD1d is induced on M␾ recruited to sites of inflammation in M␾ in response to TLR ligands. M␾ were stimulated with the TLR2 ag- vivo onist Pam Cys in the absence (E) or presence (F) of rIFN-␥ (2.5 U/ml). 3 Our data show that in the presence of proinflammatory cytokines IL-6, TNF, and IL-12p40 were measured in the culture supernatants on day ␾ 3. B, Purified M␾ were cultured with rTNF in the presence (circles) or or microbial ligands, CD1d is up-regulated on inflammatory M in absence (squares) of rIFN-␥ (2.5 U/ml). CD1d (F, left) and class II MHC vitro. Given the important role of NKT cells in pulmonary immu- (F, right) expression were determined on day 3. Open symbols denote nity (28–31), we designed an adoptive transfer experiment to mea- isotype control Ab staining. C, CD1d induction by rIFN-␥ and the sure CD1d induction on M␾ recruited to sites of pulmonary in- ␾ ␥ ␾ Pam3Cys is TNF dependent. M were cultured for 24 h with rIFN- (2.5 flammation. Purified M labeled with CFSE were adoptively U/ml) and the TLR2 agonist Pam3Cys (1 ng/ml) in the presence of a block- transferred into either uninfected or Mtb-infected B6 recipients. ing anti-TNF mAb (F) or an isotype-matched control mAb (E). After 24 h, The lungs were removed 48 h after cell transfer and donor CD1d (left, p Ͻ 0.0001) and class II MHC (right, p ϭ 0.0463) expression CFSEϩF4/80ϩ cells were detected by flow cytometry (Fig. 4A, left ϩ ␾ by F4/80 M was analyzed by flow cytometry. Each symbol represents column). Higher levels of CD1d and class II MHC were induced a separate experiment (line represents mean percentage of inhibition). D, ϩ on wild-type M␾ transferred into infected recipients compared CD1d up-regulation by rIFN-␥ and Mtb is TNF dependent. CD11b M␾ were infected with Mtb (MOI 5:1) in the presence of rIFN-␥ (2.5 U/ml) and with the same cells transferred into uninfected recipient mice (Fig. blocking anti-TNF mAb (F) or an isotype-matched control mAb (E). After 4A, top and middle row, and B). In parallel, to determine whether 24 h, M␾ expression of CD1d (left, p ϭ 0.0003) and class II MHC (right, CD1d induction was dependent on IFN-␥ in vivo, highly purified ϩ ϩ Ϫ/Ϫ p ϭ 0.4427) was analyzed. Each symbol represents a separate experiment F4/80 CD11b M␾ obtained from IFN-␥R mice were trans- (line represents the mean percentage of inhibition). ferred into infected B6 mice (Fig. 4, A, bottom row, and B). The Journal of Immunology 3589

23, which are not known to be autoreactive but show strong reac- tivity to ␣-GalCer presented by CD1d (18). M␾ were cultured in medium alone, or treated with rIFN-␥,

Pam3Cys, or both before the hybridomas were added. Because ␥ the combination of rIFN- and Pam3Cys is the most potent inducer of CD1d we expected these M␾ to present CD1- restricted Ags more efficiently to the NKT cell hybridomas (Fig. 5A). For the ␣-GalCer-specific CD1d-restricted NKT cell hy- bridomas, Gal(␣132)GalCer (␣-GalGalCer) was used as a model exogenous Ag (Fig. 5B and data not shown) (12). This requires uptake and lysosomal processing by ␣-galactosidase A to convert it into ␣-GalCer (32). For the CD1d-autoreactive NKT cell hybridomas, the M␾ were plated at different cell densities in the absence of Ag (Fig. 5C and data not shown). Interestingly, all three groups of NKT cell hybridomas significantly recognized only the ␾ ␥ M treated with a combination of rIFN- and Pam3Cys. The results obtained using M␾ as APC support our model that CD1d cell surface expression can regulate NKT cell activity. Al-

though the activation of T cell hybridomas is not dependent upon Downloaded from ␾ ␥ accessory signals, treatment of M with IFN- and Pam3Cys is likely to induce changes that may affect their function in ways that are independent of CD1d surface levels. To conclusively show that CD1d cell surface expression affects NKT cell activation, we sorted CD1d-transfected RMA-S cells (RMA-S.CD1d) to obtain

sublines that stably express different CD1d levels and used these http://www.jimmunol.org/ sublines as APC for the NKT cell hybridoma assays. A direct correlation was observed between the CD1d level expressed by the FIGURE 4. CD1d is up-regulated on M␾ at the site of inflammation in ␥ Ϫ/Ϫ ␾ RMA-S cells and the amount of IL-2 produced by three CD1d- vivo. A, Purified CFSE-labeled wild-type (WT) or IFN- R M were ␣ adoptively transferred into uninfected (top row) or Mtb-infected (middle autoreactive and three -GalCer-reactive NKT cell hybridomas and bottom row) mice. Transferred M␾ identified in recipient lung tissue (Fig. 5, D–F). Importantly, a clear dose response was observed for 48 h after cell transfer (CFSEϩF4/80ϩ) were analyzed for CD1d and class the activation of NKT cell hybridomas by RMA-S.CD1d sublines II MHC expression (thick lines). The thin lines represent control Ab stain- expressing only moderate CD1d levels (MFI ϭ 25–100) (Fig. 5F). ing. B, CD1d (left) and class II MHC (right) expression on transferred These lower levels of CD1d were comparable to the physiological wild-type (WT) M␾ from infected or uninfected recipients, or transferred levels of CD1d expressed by the M␾ used in our experiments. by guest on September 27, 2021 Ϫ Ϫ IFN-␥R / M␾ from infected recipients using three or five mice per group. Thus, in agreement with our model, APC expressing higher levels Ͻ ءءء Ͻ ءء Ϯ Bars show mean values SD. , p 0.001; , p 0.001 by one-way of cell surface CD1d activate NKT cell hybridomas more ANOVA with Bonferroni post test. The difference in CD1d and class II efficiently. MHC expression by wild-type M␾ from uninfected recipients and IFN- Ϫ Ϫ In conclusion, it appears that CD1d levels contribute to the ac- ␥R / M␾ from infected recipients was not statistically significant. tivation of both invariant and TCR diverse NKT cells. Further- more, this was true not only for autoreactive CD1d-restricted NKT cell hybridomas, but also for ␣-GalCer-specific NKT cell hybrid- Whereas both CD1d and class II MHC were up-regulated on wild- omas. These data suggest that even in the presence of a potent Ag, type M␾ recruited to infected lung tissue, minimal CD1d and class the CD1d-TCR avidity must surpass a minimum threshold to ac- Ϫ Ϫ II MHC expression was observed on donor IFN-␥R / M␾ found tivate the NKT cell. Such a threshold makes possible the regulation in the lungs of infected recipient mice. These results confirm our of NKT cell activation by modulating the expression of CD1d in vitro data and show that CD1d is induced in an IFN-␥-depen- on APC. dent manner on M␾ at sites of inflammation in vivo. CD1d levels affect activation of primary NKT cells in vitro and Level of CD1d cell surface expression on APC regulates NKT in vivo cell activity To determine whether CD1d levels affect the activation of primary Our results show that inflammatory signals regulate CD1d expres- NKT cells, enriched thymic NKT cells were tested for their ability sion by M␾ in vitro and in vivo. To determine whether M␾ ex- to recognize ␣-GalCer-pulsed target cells. Thymic NKT cells rec- ␥ ␾ pressing higher levels of CD1d activate NKT cells more efficiently ognized rIFN- plus Pam3Cys-treated M expressing high CD1d we tested a panel of CD1d-restricted NKT cell hybridomas. One levels, better than M␾ cultured in medium alone (Fig. 6A). Thus, group of iNKT cell hybridomas includes 24.7, 24.8, 24.9, and a higher CD1d level expressed by M␾ enhances Ag recognition DN32.D3, which are CD1d-autoreactive (23, 24). All recognize and activation of primary NKT cells. The capacity of thymic NKT ␣-GalCer presented by CD1d except 24.8, which instead recog- to recognize RMA-S cells expressing different CD1d levels was nizes cellular phospholipids (18, 20). A second group consists of also tested. Similar to our findings using NKT cells hybridomas, CD1d-autoreactive hybridomas expressing a diverse TCR reper- thymic NKT cell activation, as measured by their production of toire. Neither 14S.6 nor 14S.15 express an invariant TCR or rec- IL-4 and IFN-␥, correlated with the CD1d level expressed by the ognize ␣-GalCer; instead both hybridomas recognize CD1d-trans- RMA-S cells (Fig. 6B and data not shown). It is noteworthy that fected tumor cell lines including RMA-S, but the CD1d-restricted IL-4 and IFN-␥ production by primary iNKT cells in vitro required Ags remain unidentified (18, 23). The third group includes ␣-Gal- exogenous IL-2. IL-2 has also been shown to affect the generation Cer-reactive iNKT cell hybridomas KT/7, KT/12, KT/22, and KT/ of cytokine producing iNKT cells in vivo (25). The requirement 3590 REGULATION OF CD1d EXPRESSION MODULATES NKT CELL ACTIVATION

FIGURE 5. CD1d cell surface expression regulates NKT cell activation. A, Four groups of M␾ were com- pared for their ability to stimulate CD1d-restricted NKT cell hybridomas. Purified M␾ were cultured in medium ␥ alone, with rIFN- (2.5 U/ml), Pam3Cys (1 ng/ml), or both for 3 days. The dot plots show representative F4/80 and CD1d expression by M␾ used as APC in the T cell hybridoma assays. B, CD1d presentation of an exoge- nous Ag was tested using ␣-GalGalCer. M␾ were pulsed with ␣-GalGalCer at the indicated concentra- tions before culture with ␣-GalCer-reactive NKT hy- bridomas. Treatment conditions for the M␾ medium E ␥ f alone ( ), Pam3Cys ( ), rIFN- ( ), and rIFN- ␥ϩ F ␾ Pam3Cys ( ). C,M treated as in A and B were Downloaded from tested for their recognition by CD1d-autoreactive hy- bridomas. D–F, RMA-S sublines that differed in their cell surface CD1d levels were used as APC for ␣-Gal- Cer-reactive (D and F) or autoreactive (E) CD1d-re- stricted NKT cell hybridomas. IL-2 production (pico- gram per milliliter) by the NKT cell hybridomas was

measured by ELISA after 24 h. One representative ex- http://www.jimmunol.org/ periment of two to six is shown. by guest on September 27, 2021 for exogenous IL-2 in vitro may reflect that the APC used in the observe that CD1d is up-regulated on M␾ recruited to the lungs of culture system fail to produce IL-2 or fail to provide a costimula- Mtb-infected mice. On average, the CD1d induction we observed tory signal that leads to IL-2 production by the iNKT cells. Finally, on M␾ transferred into Mtb-infected recipient mice was lower than we asked whether NKT cell activation in vivo was affected by the what we observed on M␾ in vitro after treatment with rIFN-␥ and CD1d level expressed by APC. To answer this question, splenic bacterial products or TNF. Transferred M␾ may not encounter iNKT cells specific for ␣-GalCer were enumerated by an IFN-␥ high enough concentrations of bacteria or TNF in the lung tissue to ELISPOT assay following i.v. administration of ␣-GalCer-pulsed induce high levels of cell surface CD1d. Although we observed M␾ that had been cultured in medium alone or in the presence of that live bacteria had little effect on IFN-␥-induced class II MHC ␥ ␾ rIFN- and Pam3Cys (25). Injection of M expressing higher expression, our finding that dead bacteria or TLR agonists inhibit CD1d levels elicited a greater expansion of iNKT cells compared IFN-␥-induced class II MHC up-regulation are in agreement with with M␾ expressing lower CD1d levels (Fig. 6C). For example, the literature (27, 33). These results emphasize how differently whereas 43 spot forming cells/105 cells were detected after injec- CD1d is regulated because it is induced by the same conditions tion of ␣-GalCer-pulsed CD1dlow M␾, this number increased 2.5- that inhibit class II MHC expression. Down-regulation of IFN-␥- fold to 109 spot forming cells/105 cells after injection of ␣-GalCer- induced class II MHC by TLR ligands may be one mechanism by pulsed CD1dhigh M␾. Taken together, these data show that the cell which microbes evade class II MHC-restricted immune responses. surface level of CD1d expressed by M␾ influences the activation The inhibition observed using pure TLR ligands may be due to of primary iNKT cells in vitro and in vivo. high concentrations in the in vitro cultures that do not accurately reflect the amount of TLR ligands exposed to the M␾ when in- Discussion fected with live bacteria. Alternatively, other microbial products Both CD1d and class II MHC traffic through endosomal pathways expressed by viable bacteria may counter the inhibitory effects of and present exogenous Ags to T cells. However, our comparison of TLR agonists. In contrast, TLR agonists are potent inducers of CD1d and class II MHC expression by inflammatory M␾ reveals CD1d in combination with IFN-␥ at physiological concentrations. a fundamental difference in their regulation. Maximal cell surface Maximum CD1d induction was observed using 1 ng/ml LPS or ␾ ␥ expression of class II MHC on M is achieved using IFN- alone; Pam3Cys, the lowest concentration tested. Induction of CD1d by addition of TLR agonists inhibits maximal class II MHC induction TLR agonists may therefore be achieved in the picogram per mil- by IFN-␥. In contrast, maximal CD1d expression requires two sig- liliter range where the effect on class II MHC expression is less nals: one provided by IFN-␥ and the second provided by either pronounced. Interestingly, we found the effect of TLR ligands on TNF or TLR agonists. In vitro infection of M␾ with live bacteria CD1d induction is mediated by TNF. The ability of IFN-␥ and in the presence of exogenously added IFN-␥ led to dramatic up- TNF to act synergistically in CD1d up-regulation is important as it regulation of CD1d. CD1d levels are also modulated in vivo as we suggests that even in the absence of infection, inflammation can The Journal of Immunology 3591

FIGURE 6. CD1d levels affect activation of primary NKT cells in vitro and in vivo. A, Two groups of M␾ were compared for their ability to stimulate ␾ ␥ enriched thymic NKT cells. Purified M were cultured in medium alone (untreated) or with rIFN- (2.5 U/ml) and Pam3Cys (P3C) (1 ng/ml) for 3 days. M␾ were pulsed with ␣-GalCer (100 ng/ml) or vehicle before culture with enriched thymic NKT cells in the presence of rIL-2. IL-4 (mean Ϯ SD) was p Ͻ 0.001. B, RMA-S sublines that differed in their cell surface CD1d levels ,ءءء ;p Ͻ 0.01 ,ءء .measured in the culture supernatants after 48 h by ELISA were pulsed with ␣-GalCer (10 ng/ml, closed symbols) or vehicle (open symbols) and used as APC with enriched thymic NKT cells in the presence (squares) or absence (circles) of rIL-2. IL-4 (mean Ϯ SD) was measured in the culture supernatants after 48 h by ELISA. C,M␾ treated as in A were tested for their capacity to stimulate expansions of activated NKT cells in vivo. ␣-GalCer-pulsed M␾ were injected i.v. into mice and after 48 h the spleen was retrieved. Splenocytes were treated with ␣-GalCer (ϩ) or vehicle (Ϫ) and the number of IFN-␥-secreting cells (spot forming cells (SFC)/105) was Downloaded from .p Ͻ 0.001. ns, Not significant. These results are representative of two to three experiments ,ءءء .enumerated by an ELISPOT assay induce CD1d, which in turn can promote activation of CD1d-au- pothesis that the cell surface CD1d level modulates NKT cell ac- toreactive NKT cells. tivation in vivo. CD1d-restricted NKT cells are believed to be an early source of The advent of mouse models that allow the transient ablation of

cytokines such as IL-4 and IFN-␥ in various disease models. DC in adult animals has allowed investigators to determine the in http://www.jimmunol.org/ Whether IFN-␥ is required for optimum CD1d induction and sub- vivo requirement for DC during immune responses (42–44). These sequent NKT cell activation, this paradigm suggests that there models have been used to show that splenic NKT cell activation is must be another cellular source of early IFN-␥ production. Neu- dependent upon DC following systemic administration of ␣-Gal- trophils and M␾ appear to be the primary source of IFN-␥ follow- Cer in vivo (45, 46). Interestingly, in the liver, Kuppfer cells, and ing Salmonella infection, and they outnumber IFN-␥-producing not DC, are the critical APC for iNKT cell activation (45). Thus, NK, NKT, and T cells (34). Furthermore, Brigl et al. (21) show whether DC or M␾ are required for NKT cell activation may de- that in addition to iNKT cells, the main IFN-␥-producing lympho- pend on the tissue, the resident cell types, and the degree of cel- cyte population during the first 3 days of Salmonella infection is lular activation. For example, although diphtheria toxin-mediated CD3Ϫ and potentially NK cells. DC deletion leaves the splenic M␾ population untouched, splenic by guest on September 27, 2021 Determining the cellular requirements that lead to NKT cell ac- M␾ do not support rapid iNKT cell activation, and even ex vivo, tivation is essential for understanding how NKT cells affect im- splenic M␾ were not efficient APC for iNKT cells (46). Although munity to tumors, infection, and to self. We find that CD1d up- these results may at first appear to contradict our data, they are in regulation promotes activation of NKT cells both in vitro and in fact consistent because we find that highly purified unstimulated vivo. We find this to be true for both invariant and diverse CD1d- M␾ poorly stimulate NKT cell hybridomas and primary NKT restricted NKT cell hybridomas, including ones that are autoreac- cells. However, we clearly show that the inflammatory cytokines tive and ones that recognize exogenous Ag, indicating that this IFN-␥ and TNF lead to CD1d up-regulation on M␾, which makes mechanism is independent of the Ag presented by CD1d. We en- them effective at activating NKT cells both in vitro and in vivo. vision that increasing CD1d levels favor NKT cell activation by In addition to inducing CD1d, treatment of M␾ with rIFN-␥ and increasing the avidity of the CD1d-NKT cell interaction. The avid- bacterial products or TNF is likely to up-regulate the expression of ity of the TCR-CD1d interaction is known to be important for costimulatory molecules such as CD40, CD80, and CD86, and NKT cell activation. The reduced potency of ␣-GalCer analogues soluble mediators such as IL-12, which may promote NKT cell to activate iNKT cells correlates with reduced avidity of the in- activation. Our data supports the conclusion that the CD1d surface variant TCR for the Ag-CD1d complex (35). In addition to the Ag, level modulates NKT cell activation. First, the CD1d levels cor- the TCR repertoire also influences the avidity of the CD1d-TCR related with activation of CD1d-restricted T cell hybridomas irre- interaction (36, 37). An increase in the CD1d-TCR avidity may spective of their Ag specificity or TCR type. Importantly, the ac- explain why iNKT cells expressing V␤8 are disproportionately tivation of these CD1d-restricted T cell hybridomas is independent deleted during T cell ontogeny in mice overexpressing CD1d (38). of soluble factors and cell-cell contact dependent accessory signals CD1d levels also affect tissue inflammation in vivo. CD1d-re- (18). Furthermore, we derived sublines of CD1d-transfected stricted NKT cells have the capacity to prevent or ameliorate some RMA-S tumor cells that express different cell surface levels of forms of experimental autoimmunity. In the NOD mouse model of CD1d. This allowed us to directly correlate the cell surface CD1d human type I diabetes, NKT cells are dysfunctional and correction level with NKT cell activation, independently of any potential dif- of this defect by adoptive transfer of NKT cells, ␣-GalCer admin- ferences in expression of costimulatory signals that may affect istration, or overexpression of a CD1d-restricted TCR protects NKT cell activation. This allowed us to circumvent the need to use NOD mice from diabetes (39, 40). Recently, overexpression of rIFN-␥ and bacterial products to induce CD1d, and enabled us to CD1d specifically within the pancreatic islets of NOD mice was isolate the contribution of CD1d levels to NKT cell activation. found to rescue them from diabetes (41). By increasing the number The capacity of self-lipid Ags to activate CD1d-restricted NKT of CD1d-self-Ag complexes on the cell surface, the avidity for the cells may be dependent on certain accessory signals. The activa- TCR may be sufficiently raised to trigger NKT cell activation un- tion of iNKT cells by Salmonella is an instructive example. Our der physiological conditions. These experiments support our hy- current understanding is that Salmonella LPS activates DC by 3592 REGULATION OF CD1d EXPRESSION MODULATES NKT CELL ACTIVATION

MyD88-dependent TLR signaling resulting in IL-12 secretion (16, 6. Kloetzel, P. M. 2004. Generation of major histocompatibility complex class I 21). IL-12 costimulates iNKT cell recognition of the CD1d-re- : functional interplay between proteasomes and TPPII. Nat. Immunol. 5: 661–669. stricted endogenous self-lipid Ag iGb3 (16). CD1d induction could 7. Petersen, J. L., C. R. Morris, and J. C. Solheim. 2003. Virus evasion of MHC play an important role in augmenting NKT cell activation by weak class I molecule presentation. J. Immunol. 171: 4473–4478. endogenous Ags. This result is true for the 24.8 NKT cell hybrid- 8. Brigl, M., and M. B. Brenner. 2004. CD1: presentation and T cell func- tion. Annu. Rev. Immunol. 22: 817–890. oma that recognizes phospholipids, as well as for other autoreac- 9. Sko¨ld, M., and S. M. Behar. 2003. Role of CD1d-restricted NKT cells in micro- tive NKT cell hybridomas that possibly recognize iGb3. Impor- bial immunity. Infect. Immun. 71: 5447–5455. tantly, iNKT cell activation and IFN-␥ secretion stimulated by 10. Godfrey, D. I., and M. Kronenberg. 2004. Going both ways: immune regulation via CD1d-dependent NKT cells. J. Clin. Invest. 114: 1379–1388. Salmonella LPS is predicted to increase cell surface CD1d expres- 11. Bendelac, A., M. N. Rivera, S. H. Park, and J. H. Roark. 1997. Mouse CD1- sion and lead to a positive feedback loop that would further stim- specific NK1 T cells: development, specificity, and function. Annu. Rev. Immu- ulate iNKT cell activation. nol. 15: 535–562. 12. Kawano, T., J. Cui, Y. Koezuka, I. Toura, Y. Kaneko, K. Motoki, H. Ueno, For microbial Ags that are presented by CD1d, the affinity of the R. Nakagawa, H. Sato, E. Kondo, et al. 1997. CD1d-restricted and TCR-mediated iNKT TCR for the CD1d-Ag complex may be sufficiently high that activation of V␣14 NKT cells by glycosylceramides. Science 278: 1626–1629. 13. Fischer, K., E. Scotet, M. Niemeyer, H. Koebernick, J. Zerrahn, S. Maillet, additional accessory signals are unnecessary. NKT cells recogni- R. Hurwitz, M. Kursar, M. Bonneville, S. H. Kaufmann, and U. E. Schaible. tion of DC presenting CD1d-restricted Sphingomonas Ags is in- 2004. Mycobacterial phosphatidylinositol mannoside is a natural antigen for dependent of MyD88, and presentation of Leishmania Ags is in- CD1d-restricted T cells. Proc. Natl. Acad. Sci. USA 101: 10685–10690. 14. Amprey, J. L., J. S. Im, S. J. Turco, H. W. Murray, P. A. Illarionov, G. S. Besra, dependent of IL-12 (14, 16). Thus, additional costimulation may S. A. Porcelli, and G. F. Spath. 2004. A subset of liver NK T cells is activated not be required for the activation of iNKT cells by bona fide during Leishmania donovani infection by CD1d-bound lipophosphoglycan.

CD1d-restricted foreign lipid Ags. Nevertheless, CD1d induction J. Exp. Med. 200: 895–904. Downloaded from 15. Kinjo, Y., D. Wu, G. Kim, G. W. Xing, M. A. Poles, D. D. Ho, M. Tsuji, may enhance Ag recognition because our experiments show that K. Kawahara, C. H. Wong, and M. Kronenberg. 2005. Recognition of bacterial greater iNKT cell activation can be achieved even for very potent glycosphingolipids by natural killer T cells. Nature 434: 520–525. Ags such as ␣-GalCer. 16. Mattner, J., K. L. Debord, N. Ismail, R. D. Goff, C. Cantu, III, D. Zhou, P. Saint-Mezard, V. Wang, Y. Gao, N. Yin, et al. 2005. Exogenous and endog- We favor a model in which lipid Ags promote formation of enous glycolipid antigens activate NKT cells during microbial infections. Nature stable CD1d-Ag-TCR ternary complexes. Up-regulation of CD1d 434: 525–529. 17. Bendelac, A., O. Lantz, M. E. Quimby, J. W. Yewdell, J. R. Bennink, and cell surface levels would enhance the initial APC-NKT cell inter- http://www.jimmunol.org/ R. R. Brutkiewicz. 1995. CD1 recognition by mouse NK1ϩ T lymphocytes. Sci- action and promote full NKT cell activation. The Ags may be of ence 268: 863–865. microbial origin, or self-lipids that are normally sequestered or 18. Gumperz, J. E., C. Roy, A. Makowska, D. Lum, M. Sugita, T. Podrebarac, become altered during cellular stress. In the present study exam- Y. Koezuka, S. A. Porcelli, S. Cardell, M. B. Brenner, and S. M. Behar. 2000. Murine CD1d-restricted T cell recognition of cellular lipids. Immunity 12: ining NKT cell activation following infection, we found that proin- 211–221. flammatory mediators regulate cell surface CD1d expression on 19. Zhou, D., J. Mattner, C. Cantu, III, N. Schrantz, N. Yin, Y. Gao, Y. Sagiv, M␾. Because these mediators could be of either microbial or cel- K. Hudspeth, Y. P. Wu, T. Yamashita, et al. 2004. Lysosomal glycosphingolipid recognition by NKT cells. Science 306: 1786–1789. lular origin, we hypothesize that regulation of CD1d cell surface 20. Rauch, J., J. Gumperz, C. Robinson, M. Sko¨ld, C. Roy, D. C. Young, M. Lafleur, expression is a general mechanism by which the immune system D. B. Moody, M. B. Brenner, C. E. Costello, and S. M. Behar. 2003. Structural

features of the acyl chain determine self-phospholipid antigen recognition by a by guest on September 27, 2021 controls NKT cell activation and function. APC expressing higher CD1d-restricted invariant NKT (iNKT) cell. J. Biol. Chem. 278: 47508–47515. CD1d levels activate CD1d-restricted NKT cells more efficiently. 21. Brigl, M., L. Bry, S. C. Kent, J. E. Gumperz, and M. B. Brenner. 2003. Mech- Once a stable interaction between APC and NKT cell is estab- anism of CD1d-restricted activation during microbial infec- tion. Nat. Immunol. 4: 1230–1237. lished, we envision that contact dependent costimulatory signals 22. Chackerian, A. A., J. M. Alt, T. V. Perera, C. C. Dascher, and S. M. Behar. 2002. and soluble factors such as CD40/CD40L and IL-12 may be re- Dissemination of Mycobacterium tuberculosis is influenced by host factors and quired to fully activate CD1d-restricted NKT cells. Our model precedes the initiation of T-cell immunity. Infect. Immun. 70: 4501–4509. 23. Behar, S. M., T. A. Podrebarac, C. J. Roy, C. R. Wang, and M. B. Brenner. 1999. provides important insight into how CD1d is regulated and how Diverse TCRs recognize murine CD1. J. Immunol. 162: 161–167. CD1d levels expressed by APC influence NKT cell activation fol- 24. Lantz, O., and A. Bendelac. 1994. An invariant T cell receptor ␣ chain is used by lowing infection and possibly during other inflammatory a unique subset of major histocompatibility complex class I-specific CD4ϩ and CD4Ϫ8Ϫ T cells in mice and . J. Exp. Med. 180: 1097–1106. conditions. 25. Fujii, S., K. Shimizu, M. Kronenberg, and R. M. Steinman. 2002. Prolonged IFN-␥-producing NKT response induced with ␣-galactosylceramide-loaded DCs. Nat. Immunol. 3: 867–874. Acknowledgments 26. Ehrt, S., D. Schnappinger, S. Bekiranov, J. Drenkow, S. Shi, T. R. Gingeras, We acknowledge helpful scientific discussions with Michael Brenner and T. Gaasterland, G. Schoolnik, and C. Nathan. 2001. Reprogramming of the mac- thank Mi Xiao Donovan for technical assistance. rophage transcriptome in response to interferon-␥ and Mycobacterium tubercu- losis: signaling roles of nitric oxide synthase-2 and phagocyte oxidase. J. Exp. Med. 194: 1123–1140. Disclosures 27. Noss, E. H., R. K. Pai, T. J. Sellati, J. D. Radolf, J. Belisle, D. T. Golenbock, W. H. Boom, and C. V. Harding. 2001. Toll-like receptor 2-dependent inhibition The authors have no financial conflict of interest. of macrophage class II MHC expression and antigen processing by 19-kDa li- poprotein of Mycobacterium tuberculosis. J. Immunol. 167: 910–918. 28. Nieuwenhuis, E. E., T. Matsumoto, M. Exley, R. A. Schleipman, J. Glickman, References D. T. Bailey, N. Corazza, S. P. Colgan, A. B. Onderdonk, and R. S. Blumberg. 1. Boehm, U., T. Klamp, M. Groot, and J. C. Howard. 1997. Cellular responses to 2002. CD1d-dependent macrophage-mediated clearance of Pseudomonas aerugi- interferon-␥. Annu. Rev. Immunol. 15: 749–795. nosa from lung. Nat. Med. 8: 588–593. 2. Pierre, P., S. J. Turley, E. Gatti, M. Hull, J. Meltzer, A. Mirza, K. Inaba, 29. Kawakami, K., N. Yamamoto, Y. Kinjo, K. Miyagi, C. Nakasone, K. Uezu, R. M. Steinman, and I. Mellman. 1997. Developmental regulation of MHC class T. Kinjo, T. Nakayama, M. Taniguchi, and A. Saito. 2003. Critical role of V␣14ϩ II transport in mouse dendritic cells. Nature 388: 787–792. natural killer T cells in the innate phase of host protection against Streptococcus 3. Cella, M., A. Engering, V. Pinet, J. Pieters, and A. Lanzavecchia. 1997. Inflam- pneumoniae infection. Eur. J. Immunol. 33: 3322–3330. matory stimuli induce accumulation of MHC class II complexes on dendritic 30. Kawakami, K., Y. Kinjo, K. Uezu, S. Yara, K. Miyagi, Y. Koguchi, cells. Nature 388: 782–787. T. Nakayama, M. Taniguchi, and A. Saito. 2001. Monocyte chemoattractant pro- 4. Delamarre, L., H. Holcombe, and I. Mellman. 2003. Presentation of exogenous tein-1-dependent increase of V␣14 NKT cells in lungs and their roles in Th1 antigens on major histocompatibility complex (MHC) class I and MHC class II response and host defense in cryptococcal infection. J. Immunol. 167: molecules is differentially regulated during dendritic cell maturation. J. Exp. Med. 6525–6532. 198: 111–122. 31. Akbari, O., P. Stock, E. Meyer, M. Kronenberg, S. Sidobre, T. Nakayama, 5. Turley, S. J., K. Inaba, W. S. Garrett, M. Ebersold, J. Unternaehrer, M. Taniguchi, M. J. Grusby, R. H. DeKruyff, and D. T. Umetsu. 2003. Essential R. M. Steinman, and I. Mellman. 2000. Transport of peptide-MHC class II com- role of NKT cells producing IL-4 and IL-13 in the development of allergen- plexes in developing dendritic cells. Science 288: 522–527. induced airway hyperreactivity. Nat. Med. 9: 582–588. The Journal of Immunology 3593

32. Prigozy, T. I., O. Naidenko, P. Qasba, D. Elewaut, L. Brossay, A. Khurana, 40. Duarte, N., M. Stenstro¨m, S. Campino, M.-L. Bergman, M. Lundholm, T. Natori, Y. Koezuka, A. Kulkarni, and M. Kronenberg. 2001. Glycolipid an- D. Holmberg, and S. L. Cardell. 2004. Prevention of diabetes in nonobese dia- tigen processing for presentation by CD1d molecules. Science 291: 664–667. betic mice mediated by CD1d-restricted nonclassical NKT cells. J. Immunol. 173: 33. Noss, E. H., C. V. Harding, and W. H. Boom. 2000. Mycobacterium tuberculosis 3112–3118. inhibits MHC class II antigen processing in murine bone marrow macrophages. 41. Falcone, M., F. Facciotti, N. Ghidoli, P. Monti, S. Olivieri, L. Zaccagnino, Cell. Immunol. 201: 63–74. E. Bonifacio, G. Casorati, F. Sanvito, and N. Sarvetnick. 2004. Up-regulation of 34. Kirby, A. C., U. Yrlid, and M. J. Wick. 2002. The innate immune response differs CD1d expression restores the immunoregulatory function of NKT cells and pre- in primary and secondary Salmonella infection. J. Immunol. 169: 4450–4459. vents autoimmune diabetes in nonobese diabetic mice. J. Immunol. 172: 35. Sidobre, S., K. J. Hammond, L. Be´nazet-Sidobre, S. D. Maltsev, 5908–5916. S. K. Richardson, R. M. Ndonye, A. R. Howell, T. Sakai, G. S. Besra, 42. Jung, S., D. Unutmaz, P. Wong, G. Sano, K. De los Santos, T. Sparwasser, S. Wu, S. A. Porcelli, and M. Kronenberg. 2004. The T cell antigen receptor expressed S. Vuthoori, K. Ko, F. Zavala, et al. 2002. In vivo depletion of CD11cϩ dendritic ␣ ϩ by V 14i NKT cells has a unique mode of glycosphingolipid antigen recognition. cells abrogates priming of CD8 T cells by exogenous cell-associated antigens. Proc. Natl. Acad. Sci. USA 101: 12254–12259. Immunity 17: 211–220. 36. Schu¨mann, J., R. B. Voyle, B.-Y. Wei, and H. R. MacDonald. 2003. Cutting 43. Probst, H. C., and M. van den Broek. 2005. Priming of CTLs by lymphocytic edge: influence of the TCR V␤ domain on the avidity of CD1d:␣-galactosyl- choriomeningitis virus depends on dendritic cells. J. Immunol. 174: 3920–3924. ceramide binding by invariant V␣14 NKT cells. J. Immunol. 170: 5815–5819. 44. van Rijt, L. S., S. Jung, A. Kleinjan, N. Vos, M. Willart, C. Duez, 37. Stanic, A. K., R. Shashidharamurthy, J. S. Bezbradica, N. Matsuki, ϩ Y. Yoshimura, S. Miyake, E. Y. Choi, T. D. Schell, L. Van Kaer, S. S. Tevethia, H. C. Hoogsteden, and B. N. Lambrecht. 2005. In vivo depletion of lung CD11c et al. 2003. Another view of T cell antigen recognition: cooperative engagement dendritic cells during allergen challenge abrogates the characteristic features of of glycolipid antigens by V␣14J␣18 natural T(iNKT) cell receptor. J. Immunol. asthma. J. Exp. Med. 201: 981–991. 171: 4539–4551. 45. Schmieg, J., G. Yang, R. W. Franck, N. Van Rooijen, and M. Tsuji. 2005. Gly- 38. Chun, T., M. J. Page, L. Gapin, J. L. Matsuda, H. Xu, H. Nguyen, H. S. Kang, colipid presentation to natural killer T cells differs in an organ-dependent fashion. A. K. Stanic, S. Joyce, W. A. Koltun, et al. 2003. CD1d-expressing dendritic cells Proc. Natl. Acad. Sci. USA 102: 1127–1132. but not thymic epithelial cells can mediate negative selection of NKT cells. 46. Bezbradica, J. S., A. K. Stanic, N. Matsuki, H. Bour-Jordan, J. A. Bluestone, J. Exp. Med. 197: 907–918. J. W. Thomas, D. Unutmaz, L. Van Kaer, and S. Joyce. 2005. Distinct roles of 39. Van Kaer, L. 2004. Natural killer T cells as targets for immunotherapy of auto- dendritic cells and B cells in Va14Ja18 natural T cell activation in vivo. J. Im- Downloaded from immune diseases. Immunol. Cell Biol. 82: 315–322. munol. 174: 4696–4705. http://www.jimmunol.org/ by guest on September 27, 2021