of July 7, 2015.This information is current as

NephritisHydrocarbon Oil-Induced Model of Lupus Immunoregulatory Role of CD1d in the

Sebastian Joyce, Luc Van Kaer and Ram Raj SinghKakumanu, Westley H. Reeves, Vincenzo Cerundolo, Hong, Stephan D. Gadola, Akiei Mizutani, Srinivasa R.Satoh, Aleksandar K. Stanic, Jang-June Park, Seokmann Jun-Qi Yang, Avneesh K. Singh, Michael T. Wilson, Minoru

http://www.jimmunol.org/content/171/4/2142doi: 10.4049/jimmunol.171.4.2142

2003; 171:2142-2153; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/171/4/2142.full#ref-list-1

, 30 of which you can access for free at: cites 50 articlesThis article

Subscriptionshttp://jimmunol.org/subscriptions

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/ji/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/cgi/alerts/etocReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2003 by The American Association of9650 Rockville Pike, Bethesda, MD 20814-3994.The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on July 7, 2015

http://ww

w.jim

munol.org/

Dow

nloaded from

Immunoregulatory Role of CD1d in the HydrocarbonOil-Induced Model of Lupus Nephritis1

Jun-Qi Yang,2* Avneesh K. Singh,2† Michael T. Wilson,† Minoru Satoh,‡ Aleksandar K. Stanic,†

Jang-June Park,† Seokmann Hong,3† Stephan D. Gadola,§ Akiei Mizutani, ‡

Srinivasa R. Kakumanu,* Westley H. Reeves,‡ Vincenzo Cerundolo,§ Sebastian Joyce,†

Luc Van Kaer,† and Ram Raj Singh4*

Systemic lupus erythematosus (SLE) is a systemic autoimmune disease that is accompanied by the emergence of autoreactive Tcells and a reduction in regulatory T cells. Humans and mice with SLE have reduced numbers of CD1d-restricted NK T cells,suggesting a role for these cells in the regulation of SLE. In this study, we show that CD1d deficiency exacerbates lupus nephritisinduced by the hydrocarbon oil pristane. This exacerbation in disease is associated with: 1) reduced TNF-� and IL-4 productionby T cells, especially during the disease induction phase; and 2) expansion of marginal zone B cells. Strikingly, inoculation ofpristane in wild-type mice resulted in reduced numbers and/or functions of NK T cells and CD1d-expressing dendritic cells. Thesefindings suggest that CD1d may play an immunoregulatory role in the development of lupus in the pristane-induced model.TheJournal of Immunology, 2003, 171: 2142–2153.

Systemic lupus erythematosus (SLE,5 lupus) is a systemicautoimmune disease characterized by the loss of toleranceto a variety of self-Ags. Development of SLE is associated

with the emergence of autoreactive Th cells (1–4) accompaniedwith a reduction in regulatory T cells (5–7). The nature and spec-ificity of regulatory T cells that inhibit autoantibody productionand development of lupus remain largely undefined. Some regu-latory T cells are�� T cells or VH peptide-reactive CD8� T cellsthat inhibit autoantibody production by ablating autoreactive Bcells (3, 7). Recent reports show that humans and mice with lupusand related autoimmune diseases have reduced numbers of NKTcells (8–11), which recover with improvement in disease activityin patients with SLE (10). These observations suggest that NKTcells may be part of a regulatory T cell network that inhibits theinduction of SLE.

Murine NKT cells coexpress NK cell markers (e.g., CD161) andT cell markers (i.e., invariant TCR V�14J�18 chains paired pre-dominantly with V�8 chains), are mostly CD4� or double nega-tive, and are specific for the MHC class I-like molecule CD1d (12).CD1d molecules present a yet unknown ligand, mimicked by theglycolipid �-galactosylceramide (�-GalCer), to NKT cells (12–15). CD1d-reactive NKT cells, which can be tracked using CD1d/�-GalCer tetramers (16–18), appear to play protective rolesagainst a variety of immune-mediated conditions including auto-immune diabetes (19–21).

NZB/NZW F1 and MRL-lpr/lpr mice that spontaneously de-velop autoantibodies and nephritis have served as useful models tostudy the pathogenesis of SLE (2–4, 7, 22). Recent introduction ofinduced models of lupus, generated in otherwise normal mousestrains by exposure to hydrocarbon oils such as pristane, has fur-ther facilitated investigations into SLE (23, 24). Mechanisms bywhich pristane induces lupus-like autoimmunity are poorlyunderstood.

To address the role of CD1d in the pathogenesis of lupus, wehave crossed aCD1d1-null genotype (CD1do) (25) onto theBALB/c background and have investigated the effect of CD1d de-letion on the development of nephritis, autoantibody production,and cytokine responses in the pristane-induced model. We havealso examined the effects of pristane inoculation in wild-typeBALB/c mice on the numbers and/or functions of NKT cells,CD1d-expressing dendritic cells (DCs), and marginal zone B cells.Our results indicate that CD1d deficiency exacerbates pristane-induced lupus and that pristane inoculation in CD1d-sufficientmice suppresses the numbers and/or functions of CD1d-expressingDCs and NKT cells and enhances the numbers of marginal zone Bcells. These findings suggest an immunoregulatory role of CD1d inlupus.

Materials and MethodsMice

CD1do 129 � C57BL/6 mice (25) were crossed onto the BALB/cJ back-ground (The Jackson Laboratory, Bar Harbor, ME) for nine generations. Ateach backcross, heterozygous (CD1d�/�) mice were identified by Southern

*Department of Internal Medicine, Autoimmunity and Tolerance Laboratory, Uni-versity of Cincinnati College of Medicine, Cincinnati, OH 45267;†Department ofMicrobiology and Immunology, Vanderbilt University School of Medicine, Nash-ville, TN 37232; ‡Department of Medicine, University of Florida, Gainesville, FL32610;§Tumor Immunology Unit, Weatherall Institute of Molecular Medicine, Uni-versity of Oxford, John Radcliffe Hospital, Oxford, United Kingdom

Received for publication November 15, 2002. Accepted for publication June17, 2003.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby markedadvertisementin accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported in part by grants from the National Institute of Health(AR47322 to R.R.S., AI44074 to W.H.R., HL68744 and AI50953 to L.V.K., andAI42284 to S.J.) and by Cancer Research U.K. Program Grant C3999/A2291 (to V.C.).2 J.-Q.Y. and A.K.S. contributed equally to this work.3 Current address: Department of Bioscience and Biotechnology, Sejong University,Seoul 143-747, Korea.4 Address correspondence and reprint requests to Dr. Ram Raj Singh, Department ofInternal Medicine, Division of Immunology, University of Cincinnati College ofMedicine, MSB Room 7464, 231 Albert Sabin Way, Cincinnati, OH 45267-0563.E-mail address: singhrm@email.uc.edu5 Abbreviations used in this paper: SLE, systemic lupus erythematosus;�-GalCer,�-galactosylceramide; DC, dendritic cell; KBS, kidney biopsy score; GAS, glomer-ular activity score; TIAS, tubulointerstitial activity score; BUN, blood urea nitrogen;CLS, chronic lesions score.

The Journal of Immunology

Copyright © 2003 by The American Association of Immunologists, Inc. 0022-1767/03/$02.00

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

blotting (25) or by PCR using the neo (sense, 5�-CTTGGGTGGAGAGGCTATTC-3�; antisense, 5�-AGGTGAGATGACAGGAGATC-3�) andCD1d primers (sense, 5�-AATTACACCTTCCGCTGCC-3�; antisense, 5�-CTTCGTGAAGCTGATGGTGG-3�) under the following conditions:94°C for 30 s, 56°C for 1 min, and 72°C for 1 min for 35 cycles. The N9CD1d�/� BALB/c mice were intercrossed to establish CD1do BALB/cmice. The CD1do phenotype was further confirmed by flow cytometry ofPBLs stained with a conjugated anti-CD1 mAb, 1B1 (BD PharMingen, SanDiego, CA).

Establishment of pristane-induced lupus

BALB/cJ mice were inoculated once with 0.5 ml of pristane (2,6,10,14-tetramethyl-pentadecane; Sigma-Aldrich, St. Louis, MO) (23) Mice werebled before and at 3 and 6 mo after pristane inoculation, and sera werefrozen for analysis of autoantibodies. All mice were monitored for pro-teinuria once a month and were sacrificed at 10 mo of age to harvestkidneys.

Assessment of lupus disease

Kidney disease was assessed in CD1do and wild-type littermates, as de-scribed previously (2, 7). Proteinuria was measured on a 0–4� scale usinga colorimetric assay strip for albumin (Albustix; Bayer, Elkhart, IN), where0 � absent, 1� � 30 (mild), 2� � 100 (moderate), 3� � 300, and 4�� 2000 mg/dl (severe). Blood urea nitrogen (BUN) levels were measuredby impregnating Azostix (Bayer) with a drop of fresh blood and using thefollowing scale: 1� (normal) � 5–15, 2� (mild) � 15–26, and 3� (se-vere) elevation � 30 mg/dl.

Renal histology

Paraffin sections of kidneys fixed in 4% paraformaldehyde were stainedwith H&E, periodic acid-Schiff, and Masson’s trichrome. Stained sectionswere scored for the following features on a 0–3 scale by three of us(R.R.S., J.-Q.Y., S.R.K.) in a blind fashion, as described previously (22):1) Glomerular activity score (GAS) that included glomerular proliferation,karyorrhexis, fibrinoid necrosis, cellular crescents, inflammatory cells, andhyaline deposits; 2) tubulointerstitial activity score (TIAS) that includedinterstitial inflammation, tubular cell pyknosis, nuclear activation, cell ne-crosis and cell flattening, and epithelial cells or macrophages in tubularlumens; 3) chronic lesions score (CLS) that included glomerular scars,glomerulosclerosis, fibrous crescents, tubular atrophy, and interstitial fi-brosis; and 4) vascular lesion score that included arterial/arteriolar lesions.The raw scores assigned by various readers were averaged to obtain a meanscore for each of the individual features. The mean scores for individualfeatures were summed to obtain the four main scores (GAS, TIAS, CLS,and vascular lesion score) and then all four scores were summed to deter-mine a composite kidney biopsy score (KBS).

Renal immunostaining

Frozen kidney sections were fixed with methanol and acetone (1:1) for 5min. Slides were washed and incubated with biotinylated rat anti-mouse-Thy1.2, -CD4, -CD8, -CD11b and -B220 mAbs or control rat IgG (BDPharMingen). Sections were then stained using Vectastain ABC-AP kit andVector red alkaline phosphatase substrate kit I (Vector Laboratories, Bur-lingame, CA) following the manufacturer’s instructions. For immunoflu-orescence, sections were stained with FITC-conjugated goat anti-mouseIgG (Sigma-Aldrich). Slides were read by three of us in a blind fashion.

Autoantibodies against nuclear and cytoplasmic Ags

Abs against cellular proteins were analyzed by immunoprecipitation of35S-radiolabeled K562 cell extract using 4 �l of serum/sample (23). Spec-ificity was confirmed using reference sera containing anti-nRNP, Sm, Su,or ribosomal P Abs. For anti-ribosomal P peptide ELISA, a carboxyl-ter-minal 22-aa peptide carrying the major epitope of ribosomal PO recognizedby human and murine autoimmune sera was synthesized by F-moc chem-istry using a Rainin Symphony/Multiplex peptide synthesizer and purifiedby reversed-phase HPLC. Microtiter plate wells (Maxisorp Immunoplate;Nunc, Naperville, IL) were coated with 50 �l of peptide (2 �g/ml) in 20mM Tris-HCl (pH 8) at 4°C for 16 h. The wells were washed once withNET/Nonidet P-40 (150 mM NaCl, 2 mM EDTA, 20 mM Tris (pH 7.5),and 0.3% Nonidet P-40) and blocked with 0.5% BSA in NET/Nonidet P-40for 1 h at 22°C. They were then incubated with 100 �l of 1/500 mouseserum in blocking buffer for 2 h. Wells were washed, incubated with 100�l of 1/1000 alkaline phosphatase-conjugated goat anti-mouse IgG (South-ern Biotechnology Associates, Birmingham, AL) for 2 h. Plates were de-veloped with p-nitrophenyl phosphate substrate and OD was determined at405 nm using an ELISA plate reader (Molecular Devices, Menlo Park,

CA). Anti-DNA Ab were measured by ELISA, as previously described (6,7). Anti-DNA Ab titers are expressed as units per milliliter using a refer-ence-positive standard of pooled serum from MRL-lpr/lpr mice.

Flow cytometry

For liver NKT cells, liver was perfused with PBS via the portal vein untilopaque and pressed through a 70-�m cell strainer (BD Biosciences, Moun-tain View, CA). Hepatocytes were pelleted by centrifugation at 30 � g for3 min. The remaining liver cells in the supernatant were pelleted at 300 �g for 5 min and then resuspended in a 40% isotonic Percoll solution (Am-ersham Pharmacia Biotech, Piscataway, NJ). This suspension was under-laid with a 60% isotonic Percoll solution. After centrifugation for 20 minat 1500 � g, mononuclear cells were isolated at the 40/60% interface, andthen washed once with RPMI 1640 medium (Life Technologies, GrandIsland, NY) with 5% FCS (HyClone Laboratories, Logan, UT). The cellswere stained with murine CD1d/�-GalCer tetramers that were generated asdescribed elsewhere (18) or with human CD1d/�-GalCer tetramers thatalso recognize murine NKT cells (17). Stained cells were analyzed by flowcytometry. Spleen or thymus cells were incubated with anti-CD16/32(2.4G2; BD PharMingen) to block FcR�II/III, followed by staining withvarious conjugated mAbs (all BD PharMingen), as indicated in the figurelegends. Stained cells were analyzed using a BD Biosciences FACSCaliburflow cytometer and CellQuest software.

Activation of NKT and T cells

For in vitro NKT cell activation, spleen cells (1–2 � 106/ml) were incu-bated with titrated doses of synthetic �-GalCer (KRN7000; Kirin Brewery,Gunma, Japan) (15). For T cell activation, splenocytes (2 � 105/well) werestimulated with plate-bound anti-CD3 mAb (1–10 �g/ml) and supernatantswere collected after 48 h of culture for the measurement of cytokines.

Detection of cytokines

A standard sandwich ELISA was used to measure cytokines (Figs. 3 and5a), as previously described (4). TNF-� levels were measured using theBD PharMingen mouse cytokine cytometric bead array kit (Fig. 4) accord-ing to the manufacturer’s instructions. To examine the cellular sources ofcytokines in �-GalCer-stimulated cultures (Fig. 5, b–e), a cytokine secre-tion assay was performed using the MACS cytokine secretion assay kit(Miltenyi Biotec, Auburn, CA) according to the manufacturer’s protocolwith some modifications. Briefly, stimulated or control spleen cells (1 �107) were incubated at 37°C for 45 min with the cytokine Catch Reagent,which attaches to all leukocytes via CD45 Ag and binds to the specificcytokine. After washing, cells were stained with PE-conjugated cytokinedetection Ab, followed by incubation with anti-PE microbeads. Cytokine-secreting cells were then positively selected using AutoMACS (MiltenyiBiotec). Cytokine-enriched cells were counterstained with DX5, CD1d/�-GalCer tetramer, and TCR� and analyzed by flow cytometry. Dead cellsand B cells, which can nonspecifically bind to cytokine detection Ab viaPE, were excluded by staining with propidium iodide and PerCP-conju-gated B220 (BD PharMingen), respectively.

Statistical analysis

Levels of Abs and cytokines, lymphocyte percent and numbers, and renalscores were compared using Student’s t or Mann-Whitney U test. Frequen-cies of Abs, proteinuria, and BUN were compared using the two-sidedFisher’s exact test.

ResultsCD1d deficiency accelerates pristane-induced lupus nephritis

Inoculation of hydrocarbon oils such as pristane induces lupus-likeautoantibody production and mild glomerulonephritis in otherwisenormal mouse strains such as BALB/c (Refs. 23 and 24, and seedata in CD1d� mice in Figs. 1 and 2). To determine whether CD1dis involved in the development of pristane-induced lupus, we back-crossed CD1d1o 129/B6 mice onto the BALB/c background fornine generations and inoculated the final CD1do and controlBALB/c mice with pristane or PBS. All mice were bled before andat 3 and 6 mo postinoculation and monitored for proteinuria.

Proteinuria developed early and was more severe in the pristane-inoculated CD1do mice than in the CD1d� littermates (Fig. 1a). At6 mo postinoculation, 54% of the mice in the CD1do, but none in

2143The Journal of Immunology

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

the CD1d� group, had �100 mg/dl (moderate to severe) protein-uria ( p � 0.0001). The frequency of moderate to severe protein-uria was still high in CD1do mice at 10 mo (75% in CD1do vs 16%in CD1d�; p � 0.005). None of CD1d�, but 63% of CD1do, micedeveloped severe (�300 mg/dl) proteinuria at 10 mo postinocula-tion ( p � 0.0001).

BUN was also elevated in pristane-inoculated CD1do mice (Fig.1b), suggesting an advanced renal disease in these mice. Nine(82%) of 11 CD1do, but 0 of 7 CD1d�, mice had elevated (�15mg/dl) BUN ( p � 0.002). Two (18%) of 11 CD1do mice had aseverely elevated BUN (�30 mg/dl).

Mice were sacrificed at 10 mo of age to harvest kidneys and theirrenal histology was analyzed (Fig. 1, c and d). Mild and focal me-sangioproliferative glomerulonephritis was found in 60% of pristane-inoculated CD1d� mice (Fig. 1d, middle panels), the remaining 40%of mice had no evidence of nephritis by light microscopy, and none ofthe CD1d� mice had diffuse proliferative or chronic lesions. CD1do

mice, however, developed diffuse proliferative glomerulonephritiswith fibrous crescents, glomerulosclerosis, tubular atrophy, and inter-stitial fibrosis in 50% of mice (Fig. 1d, right panels); another 25% ofpristane-inoculated CD1do mice had mild to moderate mesan-gioproliferative lesions; and the remaining 25% of mice had

FIGURE 1. CD1d deficiency accelerates ne-phritis in pristane-inoculated BALB/c mice. a,CD1do (n � 24) and CD1d� (n � 19) mice wereinoculated with pristane and monitored for pro-teinuria. Results are shown as the percentage ofmice with 0� to 4� proteinuria at 6 and 10 mopostinoculation. (0�, 0 mg/dl; 1�, 30 mg/dl;2�, 100 mg/dl; 3�, 300 mg/dl; and 4�, �2000mg/dl). b, BUN levels in 10-mo-old CD1do (n �11) and CD1d� (n � 7) mice. Results are shownas the percentage of mice with normal (1�, 5–15mg/dl) or elevated (2�, 15–26 and 3�, 30–40mg/dl) BUN. c, Kidney biopsy scores: Individualcomponents (GAS, TIAS, and CLS) and a com-posite KBS (see Materials and Methods) in 10-mo-old animals are shown as the mean � SEscores (�, p � 0.05, Mann-Whitney U test). d,Representative periodic acid-Schiff (upper pan-els) and H&E-stained (lower panels) kidney sec-tions from 10-mo-old mice are shown. Left pan-els, PBS-injected CD1do mice show normal-appearing glomeruli (G) and tubules (T).Although pristane-inoculated CD1d� mice havemild focal and mesangial proliferative glomeru-lonephritis (FP, MP, middle panels), pristane-in-oculated CD1do mice (right panels) develop amore advanced glomerular and tubulointerstitialdisease with diffuse glomerular proliferation andglomerulosclerosis (GS) and almost complete oc-clusion of capillary lumens, severe interstitial in-flammation (INF), dilated atrophic tubules (AT)with flattened tubular nuclei, and severe fibrouscrescents (CR). Magnification, �400. Resultsrepresent two independent experiments. e, Kid-ney sections from pristane-inoculated CD1do andCD1d� mice stained for total T cells (Thy1.2), itsCD4� and CD8� subsets, and macrophages(CD11b). Note increased staining in CD1do ani-mals. Few cells stained for B cells in both groups(data not shown).

2144 CD1d/NKT CELLS IN LUPUS NEPHRITIS

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

mild focal or no lesions. A composite KBS (see Materials andMethods) was increased in pristane-inoculated CD1do mice( p � 0.05). Further analysis of the KBS revealed an increase inactive (GAS) as well as chronic kidney lesions (CLS) in CD1do

mice (Fig. 1c). None of the PBS-injected mice (six CD1d�, fiveCD1do) had proteinuria or renal histological changes (Fig. 1d,left panels).

As shown in Fig. 1d, inflammatory cell infiltration was in-creased in CD1do mice. To determine the phenotype of kidney-infiltrating cells, kidney sections were stained with conjugated anti-Thy1.2, anti-CD4, anti-CD8, anti-CD11b, and anti-B220 Abs.Infiltrating cells that were mostly T cells and macrophages wereincreased in CD1do mice as compared with CD1d� mice (Fig. 1e).B220� cells were rarely detected in kidney sections from bothgroups of animals (data not shown).

CD1d deficiency enhances pristane-induced autoantibodyproduction

Pristane-inoculated BALB/c mice develop autoantibodies to sev-eral cellular Ags (23, 24). We detected these Abs with an immu-noprecipitation assay using cell extract from an erythroleukemiacell line, K562, as a source of autoantigens. Overall, reactivity to

cellular Ags was higher in sera from pristane-inoculated CD1do

mice than in sera from pristane-inoculated CD1d� or PBS-injectedCD1do or CD1d� mice (Fig. 2a). Anti-ribosomal P (Fig. 2a) andanti-OJ (isoleucyl tRNA synthetase complex) Abs (Fig. 2b), whichare generally not induced in pristane-inoculated wild-type BALB/cmice (23, 24, 26), were detected in 3 (13%) of 23 CD1do, but innone of 16 CD1d�, mice. Levels of some of these autoantibodieswere quantitated by ELISA. As shown in Fig. 2c (left panel), anti-ribosomal P peptide Abs were significantly increased in the CD1do

mice as compared with wild-type littermates ( p � 0.01). SerumIgG anti-dsDNA Ab levels, as measured by ELISA, were alsohigher in CD1do than in CD1d� mice at 6 mo postinoculation( p � 0.05; Fig. 2c).

T cell cytokine responses in pristane-injected CD1d-deficientBALB/c mice

Abnormalities in cytokine production contribute to the develop-ment of lupus (24). To determine whether exacerbation of lupus inCD1do animals is related to abnormalities in cytokine production,we measured cytokine responses in the spleens of pristane-inocu-lated CD1do and CD1d� BALB/c mice. After 12–24 h, 10–12days, 6 wk, and 6 mo of pristane inoculation, spleen cells were

FIGURE 2. CD1d deficiency increases serum autoantibody production in the pristane-induced lupus model. Sera collected from BALB/c mice inoculatedwith pristane or PBS were tested for Abs against certain cellular Ags by immunoprecipitation (a and b) or ELISA (c). a, 35S-Labeled K562 cell extractwas immunoprecipitated and analyzed in 12.5% gel with sera from CD1do (left panel) or CD1d� (right panel) BALB/c mice 6 mo after injection with PBS(lanes 1–3 in both panels) or pristane (lanes 4–11 in the left and lanes 4–10 in the right panel). Left panel, Sera in lanes 4–9 and 11 are anti-nRNP Abpositive (indicated by appearance of A, B�/B, C, D, E/F, and G bands); sera in lanes 5 and 6 show three (P0, P1, and P2) bands of anti-ribosomal P Ab.Right panel, In CD1d� mice, sera in lanes 4–7 are anti-nRNP Ab positive. None of the CD1d� sera show anti-ribosomal P Ab bands. None of thePBS-injected control mice are positive for any of these autoantibodies (lanes 1–3 in both panels). On the left are shown molecular mass and the bands areindicated on the right. b, Immunoprecipitation using 8% gel with sera from CD1do mice shown in lanes 5–8 of the left panel of a. Sera in lanes 5–7 areanti-Su Ab positive (100-kDa band; long arrow); sera in lanes 5 and 6 are anti-OJ Ab positive (short arrows), and sera in lanes 5–7 are anti-Sm Ab positive(U5–200 kDa doublet; bold arrow). c, Results of ELISA for IgG anti-ribosomal P (�, p � 0.01) and anti-dsDNA (�, p � 0.05) Abs are expressed as themean � SE. Pristane-inoculated, n � 23 CD1do, 16 CD1d�; PBS-injected, n � 5 CD1do, 6 CD1d� mice. Similar results were obtained in two independentexperiments.

2145The Journal of Immunology

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

stimulated in vitro with �-GalCer, anti-CD3 Ab, or Con A (Figs.3 and 4 and data not shown). As expected, CD1do mice exhibitedno cytokine response to �-GalCer stimulation. Upon stimulationwith anti-CD3 (Fig. 3) or Con A (data not shown), IL-4 levels weresignificantly decreased, whereas IFN-� and IL-2 levels were un-changed in CD1do mice as compared with CD1d� mice; there wasno difference in the levels of these cytokines between PBS vspristane-inoculated mice at 12 h and 11 days after pristane inoc-ulation. This change in cytokine profile in CD1do mice was notassociated with significant alterations in various spleen cell pop-ulations, including CD4, CD8, DC, B220, macrophages, and neu-trophils (Ref. 25 and data not shown). At 6 mo, IL-4 and IL-13production was lower in pristane-injected CD1do and CD1d� micethan in PBS-injected CD1d� mice, while there was no effect ofCD1d deficiency or pristane injection on T cell production ofIFN-� or IL-2. Such a cytokine profile, i.e., decreased type 2 withunchanged type 1 cytokines, may contribute to the developmentand exacerbation of autoimmune disease in wild-type and CD1do

mice, respectively.Results of TNF-� production by T cells revealed a striking pat-

tern (Fig. 4). TNF-� levels, upon Con A or anti-CD3 stimulation,were similar between PBS-injected CD1do and CD1d� mice. In

pristane-inoculated mice, however, TNF-� levels in Con A or anti-CD3-stimulated cultures significantly decreased on days 10 –12postinoculation ( p � 0.05), regardless of the CD1d status of themice. At 6 mo postinoculation, TNF-� levels were slightly in-creased in pristane-inoculated mice ( p � NS). Such selective de-ficiency in T cell production of TNF-� during the initial phases ofdisease development may contribute to the development and ex-acerbation of disease in wild-type and CD1do mice, respectively.

NKT cell functions and numbers are reduced in pristane-injected animals

Although the lack of cytokine production upon �-GalCer stimu-lation was expected in CD1do mice, it was surprising that in wild-type mice �-GalCer-induced TNF-� production was significantlylower in pristane-inoculated than in PBS-injected animals (Fig. 4).This suggested that pristane inoculation itself alters NKT cell func-tions. To further evaluate the effect of pristane on NKT cell func-tions, spleen cells were harvested from BALB/c mice at varioustime points (12–24 h, 10–12 days, 6 wk, and 6 mo) after pristaneor PBS injection and stimulated with �-GalCer for 40–48 h. Cul-ture supernatants were assayed for various cytokines, which weresignificantly decreased in the pristane group at 10 –12 days (p �

FIGURE 3. Effect of CD1d deficiency on T cell cytokine production in the pristane-induced lupus model. PBS- or pristane-injected CD1d� or CD1do

mice were sacrificed at 12–24 h, 11 days, or 6 mo after pristane inoculation. Their spleen cells were stimulated with anti-CD3 for 48 h and culturesupernatants were tested for cytokines. �, p � 0.05 to � 0.0001, n � 3–7 mice/group.

FIGURE 4. T cell production of TNF-� is selectively decreased during the initial phases of development of pristane-induced lupus. Ten- to 14-wk-oldCD1do and CD1d� BALB/c mice were injected with PBS or pristane (n � 3–5 mice/group). Their spleen cells were cultured with medium alone, �-GalCer,Con A, or anti-CD3 Ab. TNF-� levels were measured in culture supernatants. Note that PBS-injected CD1do mice had no TNF-� response to �-GalCer,but had normal TNF-� response when their spleen cells were stimulated with anti-CD3 or Con A. Pristane-injected CD1d� mice, however, had decreasedTNF-� response to �-GalCer stimulation at all times tested (��, p � 0.01, pristane- vs PBS-injected CD1d� mice). TNF-� response to Con A or anti-CD3stimulation was decreased on days 10 and 11 (�, p � 0.05, pristane- vs PBS-injected CD1d� or CD1do mice), but not at 12–20 h or 6 mo time points. Resultsshown are from one of two similar experiments.

2146 CD1d/NKT CELLS IN LUPUS NEPHRITIS

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

0.05–� 0.01; Fig. 5a, left panel), 6 wk, and 6 mo postinoculation(data not shown); TNF-� (Fig. 4) and IL-2 levels (data not shown)decreased as early as 12–24 h postinoculation in the pristane groupas compared with the PBS group.

We then assessed the effect of pristane on in vivo NKT cellfunctions. Mice were injected with pristane or PBS and 12–24 h,10–11 days, or 6 mo later injected i.v. with 4 �g �-GalCer andbled 2 h later for detection of cytokines in serum samples (Fig. 5a,

FIGURE 5. �-GalCer-induced cytokine responses are markedly reduced in pristane-inoculated mice. a, Effect of �-GalCer stimulation on cytokineproduction: 10- to 14-wk-old BALB/c mice were inoculated with pristane or PBS (n � 24/group). Left panels, Ten- to 11-day after inoculation, spleenswere harvested and single spleen cells were cultured with �-GalCer for 40–48 h. Culture supernatants were assayed for cytokines (IFN-�, IL-2, IL-4, IL-5,and IL-10). Values of p � 0.05–0.01 at all �-GalCer concentrations, n � 4–5 mice/group. Cytokine levels in control cultures containing synthetic �-GalCeror vehicle in which galactosylceramides were dissolved were similar to or �2-fold of the baseline value (cells with medium alone). Right panels, Twelvehours, 10 days, or 6 mo after PBS or pristane inoculation, wild-type and CD1do mice were injected i.v. with 4 �g of �-GalCer, bled 2 h later, and serumsamples were assayed for IFN-�, IL-2, IL-4, IL-10, and IL-13. �, p � 0.05; ��, p � 0.001, n � 3–5 mice/group, �SE. In pristane-inoculated CD1do miceat 6 mo postinoculation, the mean IFN-�, IL-2, IL-4, IL-10, and IL-13 levels were 0.3, 0.1, 0.03, 0.3, and 0.02 ng/ml, respectively (data not shown). Thebackground serum cytokine levels in wild-type mice injected with vehicle alone ranged from undetectable (�0.02) to 0.15 ng/ml. Data are from one of threesimilar experiments. b–e, Cytokine-secreting cells that are mostly NKT cells in �-GalCer-exposed animals are reduced after pristane injection. PBS- orpristane-inoculated mice were injected i.v. with �-GalCer (b–d). Two hours after injection, mice were sacrificed; their spleen cells were stained for IFN-�(upper two rows), IL-2 (3rd and 4th rows), and IL-4 (5th and 6th rows) using a cytokine secretion assay (see Materials and Methods). b, CD1d/�-GalCertetramer and cytokine staining cells are shown as the percentage of gated live B220� lymphocytes. Tetramer/cytokine double-positive cells are decreasedin pristane-injected mice (highlighted in bold). c, Spleen cells (1 � 107) were enriched for cytokine-secreting cells and stained for the tetramer. Numbersof cells (�102) in each quadrant are shown. Most enriched cytokine� cells are tetramer�. The cytokine� cells (right upper and lower quadrants) were thengated to determine the percentage of TCR��tetramer� cells (d, left panels) or of TCR��DX5� cells (d, right panels). e, Unstimulated spleen cells fromcontrol PBS-injected BALB/c mice were stained with tetramer and cytokines. Stained cells are indicated as the percentage of gated live B220� lymphocytes.f, Cytokine expression on CD1d/�-GalCer tetramer� cells. �-GalCer-stimulated spleen cells were enriched for cytokine� cells as shown in c. The enrichedcells were then analyzed for cytokine expression on gated CD1d/�-GalCer tetramer� cells. Note decreased IL-4 and increased IFN-� and IL-2 expressionon tetramer� cells.

2147The Journal of Immunology

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

right panel). All cytokines tested were markedly reduced at 6 mopostinoculation ( p � 0.002–0.00003, pristane vs PBS-injectedmice). Serum IL-2 and IL-13 levels decreased as early as 12–24 hafter pristane inoculation, while serum levels of other cytokineswere variable at 12 h and 10 days postinoculation. Thus, NKT cellcytokine responses, as assessed by brief in vivo or in vitro expo-sure to �-GalCer, markedly decline after pristane inoculation.

�-GalCer-induced responses shown in Fig. 5a may reflect itsdirect effect on NKT cells as well as the secondary effects of NKTcell activation on other immune cells (27). To investigate this fur-ther, PBS- or pristane-inoculated mice were injected i.v. with 4 �g�-GalCer. Two hours later, mice were sacrificed and their spleencells were enumerated for cytokine-secreting CD1/�-GalCer tet-ramer� cells using a cytokine secretion assay. Less than 2% of liveB220� lymphocytes in �-GalCer-primed (Fig. 5b) and few cells inunprimed (Fig. 5e) animals were positive for IFN-�, IL-2, or IL-4.Most cytokine� cells were tetramer positive (Fig. 5b). To furtherconfirm this, spleen cells (1 � 107) were enriched for cytokine�

cells (Fig. 5c), which was highly efficient (�98.5%) for all threecytokines tested. Among all IFN-�� cells in PBS-injected mice,88% were NKT cells (TCR��tetramer�) and 6% each were con-ventional T cells (TCR��tetramer�) or NK cells (TCR��DX5�)(Fig. 5d). Only a few cytokine-secreting tetramer� cells expressedDX5 (Fig. 5d). Intriguingly, IL-4-secreting cells expressed more

DX-5 than IL-2 or IFN-�-secreting NKT cells (Fig. 5d). Gadueand Stein (28) made a similar observation that more DX5� thymicNKT (tetramer�) cells secrete IL-4 than DX5� thymic NKT cells.Thus, most IFN-�-, IL-2-, or IL-4 secreting cells after brief in vivo�-GalCer exposure are NKT cells. The percent (Fig. 5b) and num-bers (Fig. 5c) of these cells were lower in pristane-inoculated micethan in PBS-injected mice. Interestingly, the remaining NKT cellsin pristane-injected mice had a Th1 phenotype, i.e., decreased IL-4and increased IFN-� expression as compared with NKT cells fromPBS-injected mice (Fig. 5f): The mean fluorescence intensities ontetramer� cells were 157 vs 202 for IL-4 and 220 vs 194 for IFN-�in pristane- vs PBS-injected mice, respectively.

�-GalCer stimulation induces the expression of activation mark-ers on spleen cells (27). To assess whether this function of NKTcells is also compromised in pristane-induced lupus, �-GalCer-stimulated spleen cells from PBS- or pristane-inoculated wild-typeand CD1do mice were analyzed for activation and memory mark-ers, CD25, CD44, CD62 ligand, CD69, CD80, and CD86, by flowcytometry (Fig. 6 and data not shown). Significant decreases in theinduction of CD25 and CD69 on B and T cells were observed at 6mo postinoculation ( p � 0.03–0.0005; Fig. 6), but not signifi-cantly at earlier time points, i.e., 12 h and 10–12 days postinocu-lation (data not shown). On CD1d/�-GalCer tetramer� cells, how-ever, decreases in CD25 and CD69 expression were observed as

FIGURE 6. �-GalCer-induced expression of acti-vation markers on spleen cells is reduced in pristane-inoculated mice. Six months after pristane inoculation,mice were injected with 4 �g of �-GalCer i.v., theirspleens were harvested 2 h later, and the spleen cellswere cultured with �-GalCer for 40 h. Expression ofactivation markers, CD25, CD44, CD69, CD86, andCD62 ligand, on B and/or T cells is shown. The(mean � SE) percentage of cells expressing thesemarkers is shown in the table. �, p � 0.05 and ��, p �0.01 compared with PBS-injected wild-type mice(WT/PBS/�-GC group); n � 3–5 mice/group. Similarlevels of the expression of activation markers wereobserved in CD1do mice injected with �-GalCer andwild-type mice injected and cultured with vehiclealone (WT/PBS/Vehicle group). Results are from onerepresentative of three independent experiments.

2148 CD1d/NKT CELLS IN LUPUS NEPHRITIS

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

early as 10 days postinoculation. For example, the percentCD69�tetramer� cells (6.6 � 0.6 vs 12 � 1% of gated TCR��

cells, p � 0.05) as well as the mean fluorescent intensity of CD69on tetramer� cells (51 � 2.3 vs 63 � 5.7, p � 0.05) were signif-icantly reduced in pristane-inoculated as compared with PBS-in-jected mice. Pristane inoculation also affected the induction ofother activation and memory markers. For example, the inductionof CD44 and CD86 expression was significantly decreased onTCR� and B220� cells, respectively ( p � 0.01–0.001; Fig. 6). Asexpected, induction of all activation markers was markedly lowerin CD1do mice (Fig. 6).

Results shown in Figs. 4–6 describe the effect of pristane on�-GalCer-induced NKT cell responses. To examine whether NKTcell numbers decline spontaneously (i.e., without any �-GalCerpriming) in pristane-injected mice, we enumerated CD1d/�-Gal-Cer tetramer�TCR�� cells in the thymus, spleen, and liver ofBALB/c mice at 12–24 h, 10–12 days, 6–8 wk, or 12 mo afterpristane injection. We found that the most severe deficiency inNKT cells found was in thymus 6–8 wk after (Fig. 7a), but not atearlier time points (12–24 h and 10–12 days) after pristane injec-tion (data not shown). The percent and total thymic tetramer� cellswere significantly reduced at 6 –8 wk in pristane-inoculated miceas compared with PBS-injected control animals ( p �0.05-�0.01);the percent and total splenic tetramer� cells, however, were notsignificantly different between the two groups ( p � 0.06–0.08;Fig. 7a). In the liver, results on tetramer� cells at 12 mo postin-oculation were difficult to interpret (data not shown) because ofpristane-induced changes in the liver architecture and cellularity.

NKT cells comprise diverse phenotypes of T cells (19); somesubsets that may contribute to immune regulation and protectionagainst autoimmunity might be particularly deficient in lupus mice.To examine such a possibility, we analyzed CD4 and CD8 expres-sion on tetramer� cells at 6–8 wk after pristane injection (Fig. 7b).In the thymus, the reduction in percent and total tetramer� cells inpristane-injected animals was seen in all three subsets, CD4�,CD4�CD8�, and CD4�CD8� double-positive cells. In the spleen,the percent, but not total, CD4� and CD4�CD8�tetramer� cellswere reduced in pristane-inoculated mice ( p � 0.05). Thus, allsubsets of invariant NKT cells decline in pristane-injected animals.

Thus, primary NKT cell functions as well as secondary effectsof NKT cells on other immune cells are markedly compromised inpristane-inoculated BALB/c mice. Although a significant decreasein NKT cell numbers and most NKT cell functions are detectable6 wk after pristane inoculation, some impairment of NKT cellfunctions, such as TNF-� and IL-2 production upon �-GalCerstimulation, begins as early as 12–24 h after pristane exposure.

To explore the mechanisms of decline in NKT cell functions andnumbers after pristane inoculation, we examined the effect ofpristane inoculation on CD1d-expressing APC (Fig. 8). Overall,the expression of CD1d was decreased in spleen cells of pristane-inoculated mice as compared with untreated control animals (meanfluorescence intensity, 79 � 0.8 vs 98 � 3.6, p � 0.05). Specifically,the number of CD1d-expressing CD11c� DCs was significantly de-creased in pristane-injected mice as compared with control animals(6.7 � 0.2 vs 9.4 � 0.3, p � 0.01). The CD1d�CD11c� cells couldbe subdivided into three subsets, namely, CD1d�CD11chigh,CD1dhighCD11clow, and CD1dlowCD11clow cells. Significant differ-ences between the pristane-inoculated and control mice were seen inthe numbers of CD1d�CD11chigh (1.7 � 0.2 vs 2.5 � 0.3, p � 0.05)and CD1dhighCD11clow subsets (3.3 � 0.3 vs 5.2 � 0.1, p � 0.02) ofCD11c� cells. The CD1d-expressing B220� cells were also slightlydecreased in pristane-inoculated mice ( p � 0.04–0.06). There was nosignificant effect of pristane injection on the numbers of CD1d-ex-pressing macrophages (CD11b�) and T cells. Thus, pristane inocu-lation results in decreased numbers of CD1d-expressing DCs and B

FIGURE 7. Decreased NKT cell numbers in pristane-inoculated mice.Eight weeks after inoculation with pristane or PBS, 16-wk-old BALB/cmice were sacrificed and their thymus and spleen cells were analyzed forCD1d/�-GalCer tetramer� cells. a, TCR��tetramer� cells are expressedas the mean � SE (percent) of gated lymphocytes or as the absolute num-bers �104. b, NKT cell subsets (CD4�tetramer�, CD4�CD8�tetramer�,and CD4�CD8�tetramer� cells) are expressed as the mean percentage ofgated lymphocytes. Note that TCR��tetramer� thymocytes (a) and bothCD4� and CD4� subsets of tetramer� thymocytes or splenocytes (b) weresignificantly lower in pristane-injected than in control animals. A minorpopulation of double-positive (CD4�CD8�) tetramer� thymocytes wasalso decreased in pristane-injected mice as compared with control mice (b).�, p �0.05; ��, p �0.01, n � 3–6 mice/group, �SE. Results are repre-sentative of three independent experiments.

FIGURE 8. Reduced numbers of CD1d-expressing DCs in pristane-inoc-ulated mice. Twelve days after inoculating with pristane or PBS, BALB/c micewere injected i.v. with 4 �g of �-GalCer and sacrificed 2 h later. Total spleencell numbers were similar between the two groups of animals. CD1d�

CD11c�, CD1d�B220�, CD1d�CD11b�, and CD1d�TCR�� cells are ex-pressed as the mean percentage of spleen cells. �, p � 0.01, n � 5 mice/group.Results are representative of three independent experiments.

2149The Journal of Immunology

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

cells, which, in turn, may be responsible for the decreased NKT cellfunctions in pristane-induced lupus.

Effect of CD1d deficiency on marginal zone B cells inpristane-inoculated BALB/c mice

Results in Fig. 8 show that, while overall CD1d expression wasdecreased on B cells, the CD1dhighB220� cells that correspond tomarginal zone B cells (29) were increased in pristane-injectedmice in two separate experiments (total number of cells (�106)were 3.7 � 0.7 and 8.1 � 0.9 in PBS- vs pristane-injected mice,p � 0.05). To further examine the effect of pristane inoculation andCD1d deficiency on marginal zone B cells, we enumeratedCD21highCD23low B cells in the spleens of CD1do and CD1d�

mice on day 11 or 30 (Fig. 9, a and b) or at 6 or 12 mo (data notshown) after pristane or PBS injection. We found that marginalzone B cells were increased by 1.5- to 2-fold in the pristanegroup at 1, 6, and 12 mo after injection, regardless of the CD1dstatus of mice. Interestingly, this expansion of marginal zone Bcells was restored to normal in pristane-injected BALB/c micetreated with a CD1d ligand, �-GalCer (A. K. Stanic, J.-Q. Yang,L. V. Kaer, and R. R. Singh, manuscript in preparation). Addi-tionally, although serum IgG1 and IgG2a levels were similar be-tween the two groups of animals (data not shown), serum levels of

IgG3 isotype that is known to be preferentially secreted by mar-ginal zone B cells (29) were significantly higher in pristane-in-jected CD1do mice than in pristane-injected CD1d� mice (Fig. 9c).Previous studies have reported reduced CD23 expression on Bcells in lupus-prone NZB mice (30). We found that the mean flu-orescent intensity of CD23 expression on B cells was also reducedin pristane-injected mice as compared with PBS-injected mice, re-gardless of the CD1d status of mice, although the differences werenot statistically significant (data not shown).

Effect of �-GalCer-activated NKT cells on autoantibodyproduction in vitro

Our findings suggest that the development of lupus-like disease inpristane-inoculated wild-type mice and its exacerbation in CD1do

animals is related, at least in part, to the induced or genetic defi-ciency of NKT cells, respectively. To provide further support forthis possibility, we tested whether NKT cell activation can sup-press autoantibody production by B cells from pristane-injectedmice. For this purpose, B cells isolated from the spleens ofpristane-inoculated BALB/c mice were cocultured with T cells iso-lated from the spleens of PBS-injected control animals, with orwithout �-GalCer. Supernatants collected on day 6 were tested forIgG anti-DNA Ab and rheumatoid factor. IgG anti-DNA Ab, andrheumatoid factor levels in these cultures were generally low andwere not affected by the addition of �-GalCer (data not shown).However, in LPS-stimulated cultures, which had high levels ofrheumatoid factor, addition of �-GalCer significantly decreasedrheumatoid factor levels (Fig. 10).

DiscussionIn this article, we report that lupus nephritis, which is generallymild in pristane-inoculated BALB/cJ mice, is quite severe and ad-vanced in CD1do mice. Anti-ribosomal P and anti-OJ Abs that arenot normally induced in pristane-inoculated BALB/cJ mice (26)are detected in CD1do mice. This exacerbation of disease activityis associated with: 1) reduced TNF-� and IL-4 production by Tcells, especially during the disease induction phase; and 2) expan-sion of marginal zone B cells. Strikingly, pristane inoculation byitself results in reduced NKT cell numbers and functions and de-creased numbers of CD1d-expressing DCs and B cells.

Our observations suggest a regulatory role of CD1d-restrictedevents and that depletion of NKT cells or reduction in their func-tions may participate in the development of lupus. Indeed, patientswith lupus and other systemic autoimmune diseases have reducednumbers of NKT cells (9–11). The CD161, a marker of NK/NKTcells, is the most significantly decreased lineage marker in the

FIGURE 9. Effect of pristane inoculation and CD1d deficiency on mar-ginal zone B cells. CD1do or CD1d� mice were injected with PBS orpristane and sacrificed on day 11 or 30. Their spleen cells were stained withCD21, CD23, and B220 and analyzed by flow cytometry. Marginal zone Bcells (CD21highCD23low) are shown as the mean � SE (percent) of B220�

cells on day 30 (a) or on days 11 and 30 (b) after injection with PBS orpristane. Note 1.5- to 2-fold expansion of marginal zone B cells on day 30,but not on day 11, in pristane-injected mice. Results represent three inde-pendent experiments (n � 2–4 mice/group). c, Serum total IgG3 levels(mean � SE, �g/ml) in CD1do and CD1d� BALB/c mice 9–12 mo afterpristane injection (n � 13–16 mice/group, p � 0.02).

FIGURE 10. �-GalCer-stimulated T cells can decrease autoantibodyproduction by LPS-activated B cells from pristane-inoculated mice in vitro.Splenic B cells from BALB/c mice 6 mo after pristane inoculation werecocultured with splenic T cells from PBS-injected control animals, with orwithout �-GalCer. Supernatants collected on day 6 were tested for rheu-matoid factor, which are expressed as the mean � SD triplicate OD.

2150 CD1d/NKT CELLS IN LUPUS NEPHRITIS

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

peripheral blood cells of patients with SLE, as revealed in a recentgene expression study (31). Importantly, lupus disease activity ap-pears to inversely correlate with numbers of circulating NKT cellsin patients with SLE (10). Lupus-prone NZB/NZW F1 mice, whenrendered deficient in CD1d, also experience an exacerbation oflupus nephritis (J.-Q. Yang, L. V. Kaer, and R. R. Singh, manu-script in preparation), albeit less profound than in the pristanemodel. Old (�1.5 years) CD1d0 BALB/c mice also have increasedserum anti-DNA Ab levels as compared with age-matched controls(S. Porcelli, personal communication). A more direct evidence ofNKT cell involvement in regulation of lupus-like autoimmunitycomes from aging B6 J�18 knockout mice which have elevatedanti-DNA and anti-cardiolipin Abs and renal IgG and complementdeposition as compared with age-matched B6 controls (S. Porcelli,F. Dieli, and G. Sireci, personal communication). In MRL-lpr/lprmice, however, CD1d deficiency does not worsen kidney diseaseor anti-DNA Ab production (32).6 The regulatory effect of CD1d-restricted events on nephritis may require intact Fas signaling thatis absent in MRL-lpr/lpr mice or the antiapoptotic effects of mu-tant Fas ligand are able to bypass the role of CD1d-reactive T cells.Furthermore, mechanisms of tissue damage appear to differ be-tween different manifestations of lupus in the same or in differentanimal models of lupus (22, 32), which probably represent differ-ent subsets of this heterogeneous disease in humans. For example,mouse strains that develop anti-DNA Abs and kidney disease rep-resent 50% patients with SLE who develop these manifestations(reviewed in Ref. 33). Pristane-injected BALB/c mice, which de-velop mesangial and focal kidney lesions (Ref. 24 and Fig. 1), mayrepresent 50% of patients with lupus nephritis, whereas NZB/NZW F1 and MRL-lpr/lpr mice, which develop diffuse prolifera-tive nephritis (7, 22), may represent 30% patients with lupusnephritis (reviewed in Ref. 33). It is, therefore, not surprising thatdifferent mechanisms may operate in the development of lupus in

various animal models. It is also possible that many functionallydistinct subsets of CD1d-reactive T cells may exist; while somemay promote autoimmune responses (34), others inhibit autoim-mune disease (34, 35). For example, implantation of transgenic Tcells that express the TCR-� and -� chain genes from a T cellclone, which is CD1d specific but does not express the invariantV�14 NKT TCR, induces a lupus-like disease in irradiatedBALB/c nude recipients, whereas another subset of CD1d-reactiveT cells prevents the development of lupus in the same model (34).

There are multiple mechanisms that may explain why impair-ment or absence of CD1d-regulated events in pristane-injectedwild-type and CD1do mice, respectively, would render them moresusceptible to lupus (Fig. 11). First, abnormalities in cytokine pro-duction may contribute to the induction of autoimmunity inpristane-inoculated mice and its exacerbation in CD1do animals.We found that invariant NKT cells in pristane-inoculated miceexpress lesser amounts of IL-4 and higher amounts of IFN-� (Fig.5f). In patients with another autoimmune disease, multiple sclero-sis, remission from disease is associated with a Th2 bias of CD4�

NKT cells (35), supporting our interpretation that lack of IL-4-secreting NKT cells may contribute to the exacerbation of auto-immune disease. Since activation of NKT cells by CD1d and�-GalCer may direct conventional T cells to the acquisition of aTh2 phenotype (27), decrease or absence of NKT cells may resultin a Th1 bias of conventional T cells. Indeed, anti-CD3 or ConA-stimulated T cells from pristane-injected CD1d-deficient miceexhibit decreased IL-4 production along with stable IFN-� pro-duction (Fig. 3). Such a cytokine milieu may be detrimental inpristane-inoculated mice, since IL-4 deficiency is known to accel-erate autoantibody production in pristane-induced lupus (24). Fur-thermore, T cells from pristane-inoculated wild-type and CD1do

mice produce smaller amounts of TNF-� during the early phase ofdisease development (Fig. 4, middle panel). This finding is impor-tant since low levels of TNF-� are believed to trigger an initial steptoward the development of renal disease in lupus-prone mice (36,

6 J.-Q. Yang, V. Saxena, H. Xu, L.Van Kaer, C.-R. Wang, and R. R. Singh. Repeated�-galactosylceramide administration results in expansion of NKT cells and alleviatesinflammatory dermatitis in MRL-lpr/lpr mice. Submitted for publication.

FIGURE 11. Proposed model for the role of CD1d and NKT cell deficiency in the development of pristane-induced lupus-like autoimmunity. Pristaneinoculation itself induces a state of CD1d deficiency in BALB/c mice (I). Decreased numbers of CD1d-expressing DC subsets that participate in theestablishment of immune tolerance and regulation may contribute to the deficiency of regulatory invariant NKT cells (II). The remaining NKT cells inwild-type mice exhibit a Th1 bias (III), which may direct the conventional T cells toward a Th1 cytokine milieu, i.e., decreased IL-4 with stable IFN-�production (IV). Decreased T cell production of TNF-� may further perpetuate the disease induction phase of lupus in these animals (V). Finally, adiminution or lack of NKT cell regulation of marginal zone B cells via CD1d may cause expansion and activation of this subset of B cells (VI), which areenriched in autoreactive cells. Thus, CD1d and NKT cell deficiency may result in the activation of autoreactive T and B cell subsets, and a Th1-biasedcytokine milieu may further contribute to the activation of these cells and development of inflammation.

2151The Journal of Immunology

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

37). When rendered deficient in TNF-�, NZB mice, which gener-ally develop only a very mild autoimmune disease, exhibit a moresevere form of nephritis (38). In our experiments, TNF-� levelsincrease at 6 mo after pristane injection, i.e., the time when theseanimals begin to develop kidney disease (Fig. 4, right panel).These findings are consistent with previous reports showing thattreatment with TNF-� initiated at a young age improves lupusnephritis (36), whereas administration of TNF-� at a later age ac-celerates nephritis (39). Thus, reduced TNF-� and IL-4 productionby T cells during the early phase of disease development maycontribute to the induction of lupus in wild-type mice and to ex-acerbation of disease in CD1do mice.

Second, NKT cells appear to play a critical role in the devel-opment of immune tolerance (40), which is promoted by the ex-pression of CD1d on APC (41). Thus, a decrease in CD1d-ex-pressing DCs in pristane-injected wild-type mice (Fig. 8) and alack of these cells in CD1do mice might make these mice moresusceptible to the loss of immune tolerance and induction of patho-logic autoimmunity. In fact, our preliminary results suggest thatCD8��CD11b�CD11c� cells that correspond to CD1dhighCD11c�

cells (42) are reduced in the bone marrow of pristane-injected CD1do

mice (data not shown).Finally, CD1dhigh B cells that correspond to marginal zone

(CD21�CD23low) B cells are increased in pristane-injected mice,and IgG3 isotype, which is generally produced by marginal zone Bcells (29), is significantly increased in CD1do mice compared withwild-type littermates (Fig. 9). These observations lead us to spec-ulate that NKT cells may regulate the expansion and activation ofmarginal zone B cells via CD1d, which is highly expressed onthese cells (Fig. 11). In conditions of NKT cell deficiency such asin CD1do and pristane-injected wild-type mice, the lack of NKTcell-mediated regulation of marginal zone B cells may result inexpansion and activation of this subset of B cells, which is en-riched in autoreactive cells (29, 43, 44). Recent studies have sug-gested a role of these B cells in the development of lupus (45, 46).

The mechanisms by which NKT cell numbers and functionsdecline in the pristane-induced lupus model remain unclear. First,decreases in NKT cell functions such as �-GalCer-induced TNF-�responses begin as early as 12–24 h after pristane inoculation,excluding the possibility that changes in lymphocyte functions aredue to kidney dysfunction. Even at 6 mo when pristane-injectedmice have marked decreases in NKT cell responses, their kidneyfunctions are relatively intact. Second, the direct recognition ofpristane by NKT cells, which may lead to the activation and sub-sequently activation-induced cell death of NKT cells, is a possi-bility. However, pristane incubated with soluble plate-boundCD1d or pristane-pulsed DCs neither activated transgenic NKTcells or NKT cell hybridomas nor did it inhibit the stimulatoryactivity of �-GalCer in three separate experiments (data notshown). These experiments, however, do not absolutely excludethe possibility of pristane binding to CD1d, as pristane may not befully soluble in aqueous or detergent buffers. Third, another pos-sibility is that pristane induces expression of endogenous lipidAgs, which in turn activate CD1d-restricted T cells. Fourth, themost severe deficiency of NKT cells found was in thymus (Fig. 7),indicating that defects in thymic production was probably respon-sible for decreased NKT cell numbers in pristane-injected animals.The double-positive (CD4�CD8�) NKT cells that comprised6% of all tetramer staining cells in the BALB/c thymus weredecreased in pristane-inoculated animals. This very minor subsetof NKT cells (16, 47) has been suggested to represent an earlyprecursor stage in the NKT cell lineage (48). Impaired generationof such NKT cells in the thymus may be responsible for the declinein NKT cell numbers in pristane-induced lupus. Fifth, decreased

CD1d expression on DCs (Fig. 8) may also contribute to the de-creased NKT cell functions in pristane-inoculated mice, sinceCD1d-expressing DCs are known to play a role in the activationand maintenance of NKT cells (49). Finally, it is also possible thatthe effects of CD1d deficiency on the lupus disease process areunrelated to any direct effects of pristane on the responses ofCD1d-restricted T cells.

In summary, pristane inoculation that causes lupus-like diseaseitself induces a state of invariant NKT cell deficiency in BALB/cmice (Fig. 11). Some invariant NKT cell functions begin to declinewithin hours of pristane injection. We then begin to detect a de-crease in numbers of invariant NKT cells in the thymus and offunctional (cytokine-secreting) NKT cells in the spleen along witha bias in the remaining NKT cells toward a Th1 phenotype. Fur-thermore, induction of NKT cell deficiency by germline deletion ofthe CD1d gene in BALB/c mice exacerbates lupus-like disease,presumably through one or more of the following potential mech-anisms: 1) decreased T cell production of IL-4 along with stableIFN-� production; 2) decreased TNF-� production by T cells inthe disease induction phase; 3) a decrease in DC subsets that par-ticipate in the establishment of immune tolerance and regulation;and 4) an expansion and activation of marginal zone B cells. Ourfindings in the hydrocarbon oil-induced model may be relevant tospontaneous lupus, since pristane is known to exacerbate nephritisin genetically lupus-susceptible NZB/NZW F1 mice (50) that alsoexperience exacerbation of spontaneous lupus nephritis in the ab-sence of the CD1d1 gene (J.-Q. Yang, L. V. Kaer, and R. R. Singh,manuscript in preparation). Patients with SLE also have a reduc-tion in invariant NKT cell numbers, which appears to correlatewith lupus disease activity (10). Interestingly, we have recentlyfound that the NKT cell ligand �-GalCer protects BALB/c andSJL/J mice against pristane-induced lupus nephritis and protectsMRL-lpr/lpr mice against spontaneous inflammatory dermatitis(5). Taken together, while the precise mechanism for these find-ings remains to be confirmed, they corroborate the overall conclu-sion of our article that NKT cells play a suppressive role in thepristane-induced model of lupus nephritis. Further studies into therole of CD1d in human lupus and mechanisms by which it mayregulate the development of lupus nephritis should proveinvaluable.

AcknowledgmentsWe thank M. Hirakata for help with anti-OJ Ab confirmation; K. Behney,V. Saxena, and H. Liu for technical help; D. Adams for advice; M. Shlom-chik for reagents for rheumatoid factor assay; and Kirin Brewery Co. Ltd.for providing synthetic �-GalCer.

References1. Shivakumar, S., G. C. Tsokos, and S. K. Datta. 1989. T cell receptor �� ex-

pressing double-negative (CD4�/CD8�) and CD4� T helper cells in humansaugment the production of pathogenic anti-DNA autoantibodies associated withlupus nephritis. J. Immunol. 143:103.

2. Singh, R. R., V. Kumar, F. M. Ebling, S. Southwood, A. Sette, E.E. Sercarz, andB. H. Hahn. 1995. T cell determinants from autoantibodies to DNA can upregu-late autoimmunity in murine systemic lupus erythematosus. J. Exp. Med. 181:2017.

3. Peng, S. L., M. P. Madaio, A. C. Hayday, and J. Craft. 1996. Propagation andregulation of systemic autoimmunity by �� T cells. J. Immunol. 157:5689.

4. Singh, R. R., B. H. Hahn, and E. E. Sercarz. 1996. Neonatal peptide exposure canprime T cells, and upon subsequent immunization induce their immune deviation:implications for antibody- vs. T cell-mediated autoimmunity. J. Exp. Med. 183:1613.

5. Filaci, G., S. Bacilieri, M. Fravega, M. Monetti, P. Contini, M. Ghio, M. Setti,F. Puppo, and F. Indiveri. 2001. Impairment of CD8� T suppressor cell functionin patients with active systemic lupus erythematosus. J. Immunol. 166:6452.

6. Singh, R. R., F. M. Ebling, D. A. Albuquerque, V. Saxena, V. Kumar,E. H. Giannini, T. N. Marion, F. D. Finkelman, and B. H. Hahn. 2002. Inductionof autoantibody production is limited in nonautoimmune mice. J. Immunol. 169:587.

2152 CD1d/NKT CELLS IN LUPUS NEPHRITIS

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from

7. Fan, G., and R. R. Singh. 2002. Vaccination with minigenes encoding VH-derivedMHC class I-binding epitopes activates cytotoxic T cells that ablate autoanti-body-producing B cells and inhibit lupus. J. Exp. Med. 196:731.

8. Mieza, M. A., T. Itoh, J. Q. Cui, Y. Makino, T. Kawano, K. Tsuchida, T. Koike,T. Shirai, H. Yagita, A. Matsuzawa, H. Koseki, and M. Taniguchi. 1996. Selec-tive reduction of V�14� NK T cells associated with disease development inautoimmune-prone mice. J. Immunol. 156:4035.

9. Kojo, S., Y. Adachi, H. Keino, M. Taniguchi, and T. Sumida. 2001. Dysfunctionof T cell receptor AV24AJ18�, BV11� double-negative regulatory natural killerT cells in autoimmune diseases. Arthritis Rheum. 44:1127.

10. Oishi, Y., T. Sumida, A. Sakamoto, Y. Kita, K. Kurasawa, Y. Nawata,K. Takabayashi, H. Takahashi, S. Yoshida, M. Taniguchi, Y. Saito, andI. Iwamoto. 2001. Selective reduction and recovery of invariant V�24J�Q T cellreceptor T cells in correlation with disease activity in patients with systemic lupuserythematosus. J. Rheumatol. 28:275.

11. van der Vliet, H. J., M. E. von Blomberg, N. Nishi, M. Reijm, A. E. Voskuyl,A. A. van Bodegraven, C. H. Polman, T. Rustemeyer, P. Lips,A. J. van den Eertwegh, et al. 2001. Circulating V�24�V�11� NKT cell num-bers are decreased in a wide variety of diseases that are characterized by auto-reactive tissue damage. Clin. Immunol. 100:144.

12. Bendelac, A., M. N. Rivera, S. H. Park, and J. H. Roark. 1997. Mouse CD1-specific NK1 T cells: development, specificity, and function. Annu. Rev. Immu-nol. 15:535.

13. Porcelli, S. A., and R. L. Modlin. 1999. The CD1 system: antigen-presentingmolecules for T cell recognition of lipids and glycolipids. Annu. Rev. Immunol.17:297.

14. Gumperz, J. E., C. Roy, A. Makowska, D. Lum, M. Sugita, T. Podrebarac,Y. Koezuka, S. A. Porcelli, S. Cardell, M. B. Brenner, and S. M. Behar. 2000.Murine CD1d-restricted T cell recognition of cellular lipid. Immunity 12:211.

15. Morita, M., K. Motoki, K. Akimoto, T. Natori, T. Sakai, E. Sawa, K. Yamaji,Y. Koezuka, E. Kobayashi, and H. Fukushima. 1995. Structure-activity relation-ship of �-galactosylceramides against B16-bearing mice. J. Med. Chem. 38:2176.

16. Matsuda, J. L., O. V. Naidenko, L. Gapin, T. Nakayama, M. Taniguchi,C. R. Wang, Y. Koezuka, and M. Kronenberg. 2000. Tracking the response ofnatural killer T cells to a glycolipid antigen using CD1d tetramers. J. Exp. Med.192:741.

17. Karadimitris, A., S. Gadola, M. Altamirano, D. Brown, A. Woolfson,P. Klenerman, J. L. Chen, Y. Koezuka, I. A. Roberts, D. A. Price, et al. 2001.Human CD1d-glycolipid tetramers generated by in vitro oxidative refolding chro-matography. Proc. Natl. Acad. Sci. USA 98:3294.

18. Stanic, A. K., A. D. De Silva, J. J. Park, V. Sriram, S. Ichikawa, Y. Hirabyashi,K. Hayakawa, L. Van Kaer, R. R. Brutkiewicz, and S. Joyce. 2003. Defectivepresentation of the CD1d1-restricted natural V�14J�18 NKT lymphocyte antigencaused by �-D-glucosylceramide synthase deficiency. Proc. Natl. Acad. Sci. USA100:1849.

19. Hammond, K. J., and D. I. Godfrey. 2002. NKT cells: potential targets for au-toimmune disease therapy? Tissue Antigens 59:353.

20. Hong, S., M. T. Wilson, I. Serizawa, L. Wu, N. Singh, O. Naidenko, T. Miura,T. Haba, D. C. Scherer, J. Wei, et al.. 2001. The natural killer T-cell ligand�-galactosylceramide prevents autoimmune diabetes in non-obese diabetic mice.Nat. Med. 7:1052.

21. Wang, B., Y. B. Geng, and C. R. Wang. 2001. CD1-restricted NK T cells protectnonobese diabetic mice from developing diabetes J. Exp. Med. 194:313.

22. Singh, R. R., V. Saxena, S. Zang, L. Li, F. D. Finkelman, D. P. Witte, andC. O. Jacob. 2003. Differential contribution of IL-4 and STAT6 vs STAT4 to thedevelopment of lupus nephritis. J. Immunol. 170:4818.

23. Satoh, M., and W. H. Reeves. 1994. Induction of lupus-associated autoantibodiesin BALB/c mice by intraperitoneal injection of pristane. J. Exp. Med. 180:2341.

24. Richard, H. B., M. Satoh, J. C. Jennette, B. P. Croker, H. Yoshida, andW. H. Reeves. 2001. Interferon-� is required for lupus nephritis in mice treatedwith the hydrocarbon oil pristane. Kidney Int. 60:2173.

25. Mendiratta, S. K., W. D. Martin, S. Hong, A. Boesteanu, S. Joyce, andL. Van Kaer. 1997. CD1d1 mutant mice are deficient in natural T cells thatpromptly produce IL-4. Immunity 6:469.

26. Satoh, M., H. B. Richards, V. M. Shaheen, H. Yoshida, M. Shaw, J. O. Naim,P. H. Wooley, and W. H. Reeves. 2000. Widespread susceptibility among inbredmouse strains to the induction of lupus autoantibodies by pristane. Clin. Exp.Immunol. 121:399.

27. Singh, N., S. Hong, D. C. Scherer, I. Serizawa, N. Burdin, M. Kronenberg,Y. Koezuka, and L. Van Kaer. 1999. Cutting edge: activation of NK T cells byCD1d and �-galactosylceramide directs conventional T cells to the acquisition ofa Th2 phenotype. J. Immunol. 163:2373.

28. Gadue, P., and P. L. Stein. 2002. NK T cell precursors exhibit differential cyto-kine regulation and require Itk for efficient maturation. J. Immunol. 169:2397.

29. Martin, F., and J. F. Kearney. 2002. Marginal-zone B cells. Nat. Rev. Immunol.2:323.

30. Wither J. E., A. D. Paterson, and B. Vukusic. 2000. Genetic dissection of B celltraits in New Zealand black mice: the expanded population of B cells expressingup-regulated costimulatory molecules shows linkage to Nba2. Eur. J. Immunol.30:356.

31. Bennett, L., A. K. Palucka, E. Arce, V. Cantrell, J. Borvak, J. Banchereau, andV. Pascual. 2003. Interferon and granulopoiesis signatures in systemic lupus er-ythematosus blood. J. Exp. Med. 197:711.

32. Chan, O. T., V. Paliwal, J. M. McNiff, S. H. Park, A. Bendelac, andM. J. Shlomchik. 2001. Deficiency in �2-microglobulin, but not CD1, acceleratesspontaneous lupus skin disease while inhibiting nephritis in MRL-Faslpr mice: anexample of disease regulation at the organ level. J. Immunol. 167:2985.

33. Wallace, D. J. 2002. The clinical presentation of systemic lupus erythematosus.In Dubois’ Lupus Erythematosus, Vol. 6. D. J. Wallace and B. H. Hahn, eds.Lippincott Williams & Wilkins, Philadelphia, PA, p. 621.

34. Zeng, D., M. Dick, L. Cheng, M. Amano, S. Dejbakhsh-Jones, P. Huie, R. Sibley,and S. Strober. 1998. Subsets of transgenic T cells that recognize CD1 induce orprevent murine lupus: role of cytokines. J. Exp. Med. 187:525.

35. Araki, M., T. Kondo, J. E. Gumperz, M. B. Brenner, S. Miyake, andT. Yamamura. 2003. Th2 bias of CD4� NKT cells derived from multiple scle-rosis in remission. Int. Immunol. 15:279.

36. Jacob, C. O., and H. O. McDevitt. 1988. Tumor necrosis factor-� in murineautoimmune ’lupus’ nephritis. Nature 331:356.

37. Fujimura, T., S. Hirose, Y. Jiang, S. Kodera, H. Ohmuro, D. Zhang, Y. Hamano,H. Ishida, S. Furukawa, and T. Shirai. 1998. Dissection of the effects of tumornecrosis factor-� and class II gene polymorphisms within the MHC on murinesystemic lupus erythematosus (SLE). Int. Immunol. 10:1467.

38. Kontoyiannis, D., and G. Kollias. 2000. Accelerated autoimmunity and lupusnephritis in NZB mice with an engineered heterozygous deficiency in tumornecrosis factor. Eur. J. Immunol. 30:2038.

39. Brennan, D. C., M. A. Yui, R. P. Wuthrich, and V. E. Kelley. 1989. Tumornecrosis factor and IL-1 in New Zealand Black/White mice: enhanced gene ex-pression and acceleration of renal injury. J. Immunol. 143:3470.

40. Sonoda, K. H., M. Exley, S. Snapper, S. P. Balk, and J. Stein-Streilein. 1999.CD1-reactive natural killer T cells are required for development of systemictolerance through an immune-privileged site. J. Exp. Med. 190:1215.

41. Sonoda, K. H., and J. Stein-Streilein. 2002. CD1d on antigen-transporting APCand splenic marginal zone B cells promotes NKT cell-dependent tolerance. Eur.J. Immunol. 32:848.

42. Naumov, Y. N., K. S. Bahjat, R. Gausling, R. Abraham, M. A. Exley,Y. Koezuka, S. B. Balk, J. L. Strominger, M. Clare-Salzer, and S. B. Wilson.2001. Activation of CD1d-restricted T cells protects NOD mice from developingdiabetes by regulating dendritic cell subsets. Proc. Natl. Acad. Sci. USA 98:13838.

43. Grimaldi, C. M., D. J. Michael, and B. Diamond. 2001. Expansion and activationof a population of autoreactive marginal zone B cells in a model of estrogen-induced lupus. J. Immunol. 167:1886.

44. Li, Y., H. Li, and M. Weigert. 2002. Autoreactive B cells in the marginal zonethat express dual receptors. J. Exp. Med. 195:181.

45. Zeng, D., M. K. Lee, J. Tung, A. Bredolan, and S. Strober. 2000. A role of CD1in the pathogenesis of lupus in NZB/NZW mice. J. Immunol. 164:5000.

46. Wither, J. E., V. Roy, and L. A. Brennan. 2000. Activated B cells express in-creased levels of costimulatory molecules in young autoimmune NZB and(NZB � NZW)F1 mice. Clin. Immunol. 94:51.

47. Pellicci, D. G., K. J. Hammond, A. P. Uldrich, A. G. Baxter, M. J. Smyth, andD. I. Godfrey. 2002. A natural killer T (NKT) cell developmental pathway in-volving a thymus-dependent NK1.1�CD4� CD1d-dependent precursor stage.J. Exp. Med. 195:835.

48. Gapin, L., J. L. Matsuda, C. D. Surh, and M. Kronenberg. 2001. NKT cells derivefrom double-positive thymocytes that are positively selected by CD1d. Nat. Im-munol. 2:971.

49. Ikarashi, Y, R. Mikami, A. Bendelac, M. Terme, N. Chaput, M. Terada, T. Tursz,E. Angevin, F. A. Lemonnier, H. Wakasugi, and L. Zitvogel. 2001. Dendritic cellmaturation overrules H-2D-mediated natural killer T (NKT) cell inhibition: crit-ical role for B7 in CD1d-dependent NKT cell interferon � production. J. Exp.Med. 194:1179.

50. Yoshida, H, M. Satoh, K. M. Behney, C. G. Lee, H. B. Richards, V. M. Shaheen,J. Q. Yang, R. R. Singh, and W. H. Reeves. 2002. Effect of an exogenous triggeron the pathogenesis of lupus in (NZB � NZW)F1 mice. Arthritis Rheum.46:2235.

2153The Journal of Immunology

by guest on July 7, 2015http://w

ww

.jimm

unol.org/D

ownloaded from