Molecular studies have shown that CD1 proteins present self and foreign lipid Ags to T cells, but the possible roles of CD1 in human autoimmune diseases in vivo are not known, especially for the group 1 CD1 isoforms (CD1a, CD1b, and CD1c). To investigate the hypothesis that CD1-restricted T cells might be activated and home to target tissues involved in Hashimoto’s thyroiditis and Graves’ disease, we performed ex vivo analysis of lymphocytes from peripheral blood and autoinflammatory lesions of thyroid tissue. Immunofluorescence analysis identified two types of CD1-expressing APCs in inflamed thyroid tissues. CD1a, CD1b, and CD1c were expressed on CD83+ dendritic cells, and CD1c was expressed on an abundant population of CD20+IgD+CD23CD38 B cells that selectively localized to the mantle zone of lymphoid follicles within the thyroid gland. CD1c-restricted, glycolipid-specific T cells could not be detected in the peripheral blood, but were present in polyclonal lymphocyte populations isolated from affected thyroid glands. In addition, polyclonal thyroid-derived lymphocytes and short-term T cell lines were found to recognize and lyse targets in a CD1a- or CD1c-dependent manner. The targeting of CD1-restricted T cells and large numbers of CD1-expressing APCs to the thyroid gland during the early stages of autoimmune thyroiditis suggests a possible effector function of CD1-restricted T cells in tissue destruction and point to a new model of organ-specific autoimmune disease involving lipid Ag presentation.

There has been rapid progress in identifying the molecular structures of self and foreign lipid Ags and their mode of presentation by CD1 to T cells. At least four human CD1 proteins, CD1a, CD1b, CD1c, and CD1d, bind lipid Ags within a hydrophobic groove to form antigenic CD1-lipid complexes. To date, diacylgycerols, sphingolipids, polyisoprenoid lipids, polyketides, lipopeptides, and mycolyl glycolipids found in bacterial pathogens are known to be presented to human T cells (1, 2, 3, 4, 5). Other Ags, such as gangliosides, sulfatides, and phosphatidylinositols, are present in uninfected human cells (6, 7, 8). Increasing appreciation of the diversity of the CD1-restricted T cell repertoire has raised basic questions about how the immune system regulates responses to different types of self and foreign Ags.

After translation in the endoplasmic reticulum, newly synthesized CD1 proteins traverse the secretory pathway and bind phosphatidylinositol-containing and other endogenous lipids to form CD1-lipid complexes, raising the possibility that the presentation of self Ags could lead to T cell activation (9). Self-lipid Ag recognition was first suggested by the isolation of T cells described as CD1-autoreactive because they were activated in a CD1-dependent manner in the absence of any added exogenous Ag (10). Although such CD1 autoreactivity might in some cases represent ligand-independent recognition of CD1, increasing evidence shows that CD1 proteins normally present self-lipids. For example, several classes of mammalian lipids, including GM1 and GD3 gangliosides, sulfatide, phosphatidylinositol, phosphatidylethanolamine, glycosyl phosphatidylinositols, and dolichyl phospholipids, can activate CD1-restricted T cells when they are added to cultures at high concentrations in vitro (3, 6, 7, 11, 12, 13, 14). These studies provide proof of concept that self-lipid reactive T cells exist in the periphery of mice and humans, prompting in vivo studies on the role of CD1-restricted T cells in the generation of autoimmune diseases.

In general, CD1 proteins have been proposed to influence autoimmune diseases by activation of either regulatory or effector T cells. The most extensively studied model involves the ability of CD1d-restricted NK T cells with invariant Vα14 (mice) or Vα24 (human) TCRs to prevent autoimmunity. Studies of mice lacking CD1d proteins or Vα14Jα281 TCRs have generally found accelerated progression of autoimmune diabetes, experimental autoimmune encephalomyelitis, collagen-induced inflammatory arthritis, and anterior chamber-associated immune deviation (as reviewed in Ref. 15). Conversely, activation of NK T cells with α-galactosyl ceramide has been found to ameliorate autoimmune responses in these same models and a mouse model of Graves’ disease (16). Also, a protective role of NK T cells in human autoimmunity has been supported by correlation of reduced frequency of peripheral blood NKT cells with more rapid progression of autoimmune diabetes (17, 18, 19), although this is controversial (20, 21). These studies support the view that CD1d-restricted NK T cells are generally immunoprotective and carry out their effects indirectly through cytokine-mediated regulation of other effector cells, such as MHC-restricted T cells and dendritic cells (DC)3 (17, 21, 22, 23).

Quite separate from any role of immunoregulatory NK T cells, autoreactive effector T cells with diverse TCRs might recognize gangliosides, ceramides, dolichols, or other self-Ags and target tissues that express structurally altered or high levels of these lipids. This new hypothesis is akin to well-studied mechanisms by which self peptide-reactive T cells attack pancreatic islets, myelin, thyroid, or other tissues that express tissue-specific protein Ags. There is evidence that T cell clones from human patients with multiple sclerosis can specifically recognize ganglioside and sulfatide (6, 7). These and other candidate self-lipid Ags are expressed at the sites of organ-specific autoimmune responses. For example, GM1 ganglioside, GD3 ganglioside, and sulfatide are found in central or peripheral nervous tissues, and dolichyl glycolipids, which are related in structure to known CD1c-presented mannosyl phosphodolichols, are expressed at 10-fold higher levels in thyroid gland compared with other tissues (24, 25). The possibility that self-lipids might serve as a target of organ-specific autoimmune disease in vivo has not been extensively studied, although a recent study has shown that sulfatide treatment can improve outcomes in experimental autoimmune encephalomyelitis in mice (26). In contrast to the proposed immunoprotective roles of CD1d and NK T cells, this view argues that diverse lipid-reactive T cells would be expected to express varied TCRs, recognize multiple CD1 isoforms, and carry out their effects locally at the sites of tissue destruction.

Such diverse lipid Ag-specific T cells are difficult to analyze in vivo. Whereas NK T cells can be measured and manipulated based on their conserved TCR, there are no known cell surface markers for diverse CD1-restricted T cells, so they must be studied as polyclonal populations ex vivo. Likewise, mice lack orthologs of CD1a, CD1b, and CD1c, which present candidate lipopeptide, ganglioside, and dolichyl autoantigens in humans, necessitating the development of human model systems. To address these issues, we initiated studies of human CD1 proteins in two related forms of autoimmune thyroiditis, Graves’ disease (GD) and Hashimoto’s thyroiditis (HT). Both conditions commonly affect women in the third to fifth decade of life and involve a predominantly lymphocytic, tissue-specific invasion of the thyroid gland, which leads to T cell-dependent B cell production of autoantibodies against thyroid Ags. These related diseases differ in that HT typically destroys thyroid hormone-producing epithelia, resulting in hypothyroidism, whereas GD involves the production of autoantibodies that agonize the thyroid-stimulating hormone receptor, typically resulting in hyperthyroidism. These common diseases are well suited for basic studies of human lymphocyte responses, because thyroiditis is treated with hormone replacement rather than immunosuppressive drugs that may alter the function of lymphocytes in situ or ex vivo. Also, serum thyroid stimulating hormone measurements and autoantibodies against thyroid peroxidase (TPO) and other thyroid Ags allow early and definitive diagnosis, so that the evolution of the disease can be monitored.

To determine whether CD1 plays a role in the generation of thyroid autoimmunity, we identified CD1+ APCs and CD1-restricted T cells within the thyroid glands of affected patients. CD1-expressing APCs infiltrated the thyroid gland in large numbers during the acute and chronic stages of both GD and HT. Two types of APCs were identified in part by their distinct anatomic locations within the gland. CD1a+CD1b+CD1c+CD83+ DCs were scattered throughout the gland, and CD1c+ B cells were selectively localized to the marginal zone of secondary lymphoid follicles in the thyroid tissue. Fresh thyroid-derived lymphocytes and short-term T cell lines isolated from the thyroid gland were able to lyse target cells that expressed CD1a and CD1c proteins, and we found evidence of CD1c-mediated recognition of a synthetic dolichyl glycolipid. The infiltration of CD1-expressing APCs and CD1-restricted T cells provide evidence for a role of diverse CD1-restricted T cells in the intrathyroidal T cell response during human autoimmune thyroiditis, providing a new perspective for studying how CD1 and lipids play a role in this common human autoimmune disease.

Thyroid tissue was obtained from surgical biopsy specimens of patients with HT and GD and was compared with thyroid glands from patients with a noninflammatory thyroid disease, multinodular goiter. Clinical diagnosis of HT was based on palpable thyroid enlargement by an experienced endocrinologist and elevated TPO (>100 IU/ml) or thyroglobulin (>115 IU/ml) autoantibodies measured by ELISA (Immunowell). Clinical diagnosis of GD was made based on thyroid enlargement, serum free thyroxine measurements (>1.5 IU/L), and thyroid-stimulating hormone receptor Abs measured by RIA (Brahms Diagnostica) (Table I). In all cases in which a thyroid biopsy was obtained, diagnoses were confirmed by histopathological examination of the tissue.

Table I.

Summary of patients’ dataa

SampleAnti-Tg (IU/ml)Anti-TPO (IU/ml)Anti-TSHR (IU/L)
HT-1 774.0 105,000 − 
HT-2 584.0 766 − 
GD-1 598.0 1,155 67 
GD-2 − − 118.0 
GD-3 − − 4.1 
GD-4 − − 68 
GD-5 − 686 79 
GD-6 − 100 ND 
MNG-1 − − ND 
SampleAnti-Tg (IU/ml)Anti-TPO (IU/ml)Anti-TSHR (IU/L)
HT-1 774.0 105,000 − 
HT-2 584.0 766 − 
GD-1 598.0 1,155 67 
GD-2 − − 118.0 
GD-3 − − 4.1 
GD-4 − − 68 
GD-5 − 686 79 
GD-6 − 100 ND 
MNG-1 − − ND 
a

Tg, Thyroglobulin; TSHR, thyroid-stimulating hormone receptor; MNG, multinodular goitre; −, negative.

Thyroid tissue from surgical biopsy specimens was enzymatically digested with 2.5 mg/ml trypsin (Sigma-Aldrich), 0.2 mg/ml collagenase (type P; Roche), and 0.05 mg/ml DNase (Sigma-Aldrich) at 37°C and passed through a 500-μm pore size mesh to yield a single cell suspension as previously described (27). Cell viability and number were assessed under a UV microscope by ethidium bromide/acridine orange staining. Preparations containing >85% viable cells were cryopreserved in heat-inactivated FCS with 10% DMSO and stored in liquid nitrogen for further analysis. Total thyroid-derived cells were subjected to plastic adherence by overnight culture in RPMI 1640 medium supplemented with 2 mM l-glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 10% heat-inactivated FCS at 37°C in 5% CO2 in tissue culture flasks. Thyroid-derived lymphocytes were recovered in the nonadherent cell populations and were additionally purified by centrifugation over Ficoll-Paque Plus (Amersham Biosciences). The population of adherent cells, which contained mostly thyroid epithelial cells and DCs, was collected for culture or cryopreservation.

Thyroid-derived T cells were cultured at 106 cells/well in 24-well plates at a 2:1 ratio with autologous adherent cells. One week later, 10 IU/ml human rIL-2 (Hoffmann-La Roche) was added, and cells were additionally cultured for 1 wk, yielding cell populations that were 70% CD3+ (28). In some cultures, T cells were partially enriched for Vα24+ T cells using goat anti-mouse IgG-coated magnetic beads (Dynabeads) and Vα24-specific mAb (C15), although resulting cultures contained a significant percentage of Vα24 cells. The resulting T cells were cloned by limiting dilution in U-bottom, 96-well plates (Costar) and were maintained by biweekly stimulation with feeder cells consisting of a 10:1 mixture of irradiated, allogeneic human PBMC (50,000 cells/well) and autologous EBV-transformed B cells together with PHA (1/2000 dilution; Difco). IL-2 (10 IU/ml) was added 24 h after each stimulation. None of the T cell lines obtained had Vα24 TCR, as determined by flow cytometry using the C15 mAb.

PBMCs were isolated from eight thyroiditis patients (GD, n = 5; HT, n = 3) and two healthy donors by Ficoll-Paque gradient density. After adherence of total PBMC to plastic for 30–40 min, nonadherent lymphocytes were harvested as a source of T cells. Adherent cells were additionally cultured for 3 days in the presence of GM-CSF (300 IU/ml) and IL-4 (250 IU/ml) to obtain autologous. monocyte-derived DCs.

Lipid Ag suspensions were prepared freshly for each assay after removal from −20°C storage by transferring them to glass tubes, drying under nitrogen to remove organic solvents, and resuspending in T cell medium by sonication in a water bath for 2 min. Mannosyl-phosphodolichols (MPD C35 and MPD C95) were prepared by coupling hexose dinucleotides to pure human (C95) or synthetic (C35) phosphodolichol of the indicated chain length as previously described (3).

Cytotoxicity was measured by lysis of 51Cr-labeled C1R lymphoblastoid cell lines mock-transfected with vector alone or stably transfected with CD1a, CD1b, or CD1c genes or thyroid follicular cells (29). Chromium labeling was conducted for 2–16 h, and target cells were additionally incubated for 2 h in the presence or the absence of the glycolipid Ag. Target cells were seeded in duplicate into 96-well microtiter plates (Costar) prefilled with effector cells according to desired E:T cell ratios. Supernatants were harvested after 4 h, and 51Cr release was assessed by counting 50 μl of the supernatant in a liquid scintillation counter.

T cell proliferation was measured by culturing T cells in U-bottom wells at a 1:1 ratio with irradiated autologous monocyte-derived DCs (100,000 cells/well) in the presence or the absence of the lipid Ags for 72 h. This was followed by the addition of 1 μCi of [3H]thymidine (PerkinElmer) and an additional 6 h of culture before cells were collected and β-emissions were counted. IL-2 release was measured using a similar method, except that 50 μl of supernatant was removed after 24 h of culture. Supernatants were transferred to wells containing 10,000 HT-2 cells, which are dependent on IL-2 for their growth. HT-2 cells were cultured for 24 h before 1 μCi of [3H]thymidine was added for an additional 6 h of culture before counting β emissions.

For immunohistochemistry, frozen thyroid tissue was cut to 4-μm cryosections and fixed in cold acetone. Tissue sections were stained with Abs specific for CD1c (F10/21A3.1) (30), CD3 (OKT3; American Type Culture Collection), CD20 (L26; DakoCytomation), or isotype control (P3) at 10 μg/ml. Stained sections were visualized with an anti-mouse avidin-biotin-peroxidase complex (Vector Laboratories) and counterstained with hematoxylin. For immunofluorescence, cryostat sections were stained by indirect fluorescence with Abs to CD1a (OKT6; American Type Culture Collection), CD1b (BCD1b3.1) (31), CD1c (F10/21A3.1) (31), and CD1d (CD1d 42) (32) or isotype control (P3) at 10 μg/ml, followed by a PE-labeled goat anti-mouse IgG (1/100 dilution; BioSource International) and FITC-labeled mAb for CD20 (L27; BD Biosciences), CD3 (UCHT1; BD Pharmingen), CD83 (HbB15e; BD Pharmingen), CD23 (MHM6; DakoCytomation), CD38 (HIT2; BD Pharmingen), and IgD (F(ab′)2 rabbit anti-human IgD, DakoCytomation). Images were acquired with a Nikon Eclipse 800 confocal microscope, digitally captured using a SPOT RT digital camera, and compiled using Adobe Photoshop software.

For flow cytometry, single cell suspensions were incubated for 15 min with human serum and additionally incubated with anti-CD1 mAb or isotype-matched control (P3) for 30 min on ice. Cells were then washed, incubated with PE-labeled goat anti-mouse IgG serum (1/100 dilution; BioSource International), washed again, and incubated with FITC-labeled mAb for CD20 (L27; BD Biosciences), CD83 (HbB15e; BD Pharmingen), or isotype control. Samples were analyzed on a FACScan using the CellQuest software (BD Biosciences). Dead cells were excluded from the analysis by propidium iodine staining.

To investigate the hypothesis that CD1-restricted T cells might be involved in autoimmune thyroiditis, we first identified CD1-expressing cells in the thyroid glands of patients with GD and HT. Surgical biopsy specimens of normal thyroid tissue and multinodular goiters showed little or no infiltrate of mononuclear cells. The predominantly lymphocytic infiltrate present in the HT samples resulted in severe disruption of the thyroid architecture, such that only small remnants of thyroid epithelia persisted, which is typical of three patients analyzed (Fig. 1,A, lower panel). The immune infiltrates were organized into lymphoid follicles and contained CD20+ cells in the germinal center and mantle zone (Fig. 1,B). GD samples showed hypertrophy of follicular epithelial cells and prominent lymphocytic infiltration, although lymphocytes were less numerous and lymphoid follicles were less abundant than those in HT (Fig. 1,A, middle panel). Immunohistochemical staining of thyroid glands with autoimmune thyroiditis detected the presence of cells staining with mAb for CD1a, CD1b, and CD1d and a particularly large population of cells staining with anti-CD1c (Fig. 1 B and data not shown).

FIGURE 1.

CD1c+ cells infiltrate autoimmune thyroid glands. A, Representative frozen thyroid tissue sections (4–5 μm) stained with H&E from a patient with multinodular goiter (MNG) shows intact thyroid follicles, whereas samples from GD and HT show infiltrates of small round cells and destruction of thyroid follicles. B, Immunohistochemical staining of thyroid tissue sections with anti-CD1c (F10/21A3.1), anti-CD20 (L27), and anti-CD3 (OKT3) shows CD1c+ cells that specifically localize to the mantle zone (MZ) located between T cell and B cell areas of lymphoid follicles as well as less abundant CD1c+ cells outside the follicle in both GD and HT. A higher magnification of CD1c+ cells shows a small, round cell from the marginal zone (upper inset) and a cell with more abundant cytoplasm and branched morphology typical of those located outside the lymphoid follicle (lower inset).

FIGURE 1.

CD1c+ cells infiltrate autoimmune thyroid glands. A, Representative frozen thyroid tissue sections (4–5 μm) stained with H&E from a patient with multinodular goiter (MNG) shows intact thyroid follicles, whereas samples from GD and HT show infiltrates of small round cells and destruction of thyroid follicles. B, Immunohistochemical staining of thyroid tissue sections with anti-CD1c (F10/21A3.1), anti-CD20 (L27), and anti-CD3 (OKT3) shows CD1c+ cells that specifically localize to the mantle zone (MZ) located between T cell and B cell areas of lymphoid follicles as well as less abundant CD1c+ cells outside the follicle in both GD and HT. A higher magnification of CD1c+ cells shows a small, round cell from the marginal zone (upper inset) and a cell with more abundant cytoplasm and branched morphology typical of those located outside the lymphoid follicle (lower inset).

Close modal

To determine a more precise phenotype and possible functions of CD1-expressing cells in the thyroid gland, we conducted immunohistochemical analysis of serial sections, immunofluorescence microscopy, and flow cytometric analysis with markers of B cells (CD20, IgD, CD23, CD38, and CD5), T cells (CD3), DCs (CD11c, CD83, and DC-SIGN), macrophages (CD68), and thyroid epithelial cells (TPO). Most CD1c+ cells were small, round cells that selectively localized to the T cell-B cell interface of the lymphoid follicles (Fig. 1,B). The scant cytoplasm suggested that these were lymphocytes, and localization to the outer rims of these lymphoid follicles suggested that they might be mantle zone B cells. Consistent with this, two-color flow cytometric analysis of thyroid-derived lymphocytes confirmed the coexpression of CD1c and CD20 and provided additional evidence that these CD1c-expressing B cells were abundant, such that they comprised more than one-third of all lymphocytes in a representative lesion (Fig. 2,A). Additional analysis by two-color immunofluorescence microscopy showed that CD1c is coexpressed with CD20 and IgD on cells located at the marginal zone of the lymphoid follicles. CD1c was not expressed on centroblasts (CD20+CD38+CD23) and centrocytes (CD20+CD38+CD23+) on CD5+ B-1 B cells (Fig. 2 B and data not shown). Taken together, these studies provided evidence that the immune infiltrate in HT and GD included an abundant population of mature CD1c-expressing B cells that selectively localize to the mantle zone of lymphoid follicles.

FIGURE 2.

CD1c+ cells are mantle zone B cells and DCs. A, Two-color flow cytometry on single cell suspensions of infiltrating cells using anti-CD1c, anti-CD20, and anti-CD83 Abs reveal CD1c+CD20+ B cells and CD1c+CD83+ cells in the infiltrates, which are representative of experiments on four thyroid glands. B, Double-immunofluorescence staining of CD1c+ cells (red) and anti-CD20, anti-CD23, rabbit anti-human IgD, or anti-CD38 (green) in a lymphoid follicle shows colocalization of CD1c+ cells with CD20 and IgD staining, but not with markers for germinal center B cells (CD23 and CD38). C, Double-immunofluorescence staining of CD1a, CD1b, and CD1c cells (red) with anti-CD11c (green) shows CD1 on cells located adjacent to the thyroid follicle (∗) or scattered on the infiltrates in a representative thyroid gland from a HT patient.

FIGURE 2.

CD1c+ cells are mantle zone B cells and DCs. A, Two-color flow cytometry on single cell suspensions of infiltrating cells using anti-CD1c, anti-CD20, and anti-CD83 Abs reveal CD1c+CD20+ B cells and CD1c+CD83+ cells in the infiltrates, which are representative of experiments on four thyroid glands. B, Double-immunofluorescence staining of CD1c+ cells (red) and anti-CD20, anti-CD23, rabbit anti-human IgD, or anti-CD38 (green) in a lymphoid follicle shows colocalization of CD1c+ cells with CD20 and IgD staining, but not with markers for germinal center B cells (CD23 and CD38). C, Double-immunofluorescence staining of CD1a, CD1b, and CD1c cells (red) with anti-CD11c (green) shows CD1 on cells located adjacent to the thyroid follicle (∗) or scattered on the infiltrates in a representative thyroid gland from a HT patient.

Close modal

Low power micrographs of autoimmune thyroid tissue identified a second population of CD1c-staining cells that were scattered outside the lymphoid follicles (Fig. 1,B). These CD1c+ cells were initially discriminated from B cells based on their more abundant cytoplasm and branched morphology (Fig. 1,B, inset). Two-color flow cytometry demonstrated the presence of CD1c- and CD83-coexpressing cells, which were much less abundant than the CD1c-expressing B cells, as they comprised only 3% of the nonadherent cell fraction (Fig. 2,A, lower panel). Additional analysis by fluorescence microscopy showed evidence of the presence of CD1a, CD1b, and CD1c on cells that coexpressed CD11c (Fig. 2,C) and in some cases had a clearly dendritic morphology. Based on these criteria, these cells were identified as tissue DCs that localized to two main locations, interstitial DCs surrounding the thyroid follicles or scattered in the lymphocytic infiltrates (Fig. 2 C, upper panels).

To determine the specificity of T cells that invade the thyroid gland in autoimmune thyroiditis, single cell suspensions of infiltrating mononuclear cells were prepared from thyroid biopsy specimens. Fresh thyroid cells were separated by plastic adherence to yield responder cells that were composed of polyclonal T cells and B cells, as well as adherent cells composed mainly of TPO-expressing epithelial cells and myeloid cells. In initial experiments, polyclonal thyroid-derived lymphocytes of known MHC haplotype (patient HT-1: A2,−; B15,44) were activated and expanded with autologous, irradiated thyroid cells and examined for their ability to recognize target thyroid-derived cells from patients that were either matched (patient GD-4: HLA-A2,−; B44,−) or unmatched (patient GD-5: HLA-A1, A3, B5, B62) at MHC class I, A and B loci. Substantial cell lysis was observed for both matched and unmatched target cells, but not for the human fibroblast cell line M1 (HLA-A1,3; B8,−) that does not express any CD1 molecule (33). This suggested that activated lymphocytes infiltrating the thyroid gland might lyse target cells using a mechanism that is independent of MHC class I (Fig. 3 A).

FIGURE 3.

Infiltrating T cells are restricted by CD1, but not by MHC class I, molecules. A, T cells from patient HT-1 (HLA-A2,-; B15,44) were tested with a panel of 51Cr-labeled MHC class I-matched (HLA-A2,-; B44,-) and MHC class I unmatched (HLA-A1, A3, B5, B62) thyroid-derived APCs and the CD1 human fibroblast cell line M1 (HLA-A1,3; B8,-). b, Infiltrating T cells from patient HT-1 cultured for 2 wk in autologous thyroid-derived APCs lysed transformed B cell lines (C1R) transfected with CD1a, CD1c, or CD1d, but not mock-transfected, cells. Results are expressed as the percent lysis for the E:T cell ratios shown.

FIGURE 3.

Infiltrating T cells are restricted by CD1, but not by MHC class I, molecules. A, T cells from patient HT-1 (HLA-A2,-; B15,44) were tested with a panel of 51Cr-labeled MHC class I-matched (HLA-A2,-; B44,-) and MHC class I unmatched (HLA-A1, A3, B5, B62) thyroid-derived APCs and the CD1 human fibroblast cell line M1 (HLA-A1,3; B8,-). b, Infiltrating T cells from patient HT-1 cultured for 2 wk in autologous thyroid-derived APCs lysed transformed B cell lines (C1R) transfected with CD1a, CD1c, or CD1d, but not mock-transfected, cells. Results are expressed as the percent lysis for the E:T cell ratios shown.

Close modal

To directly determine whether human CD1 proteins could function as the restriction element for the recognition by thyroid-derived lymphocytes, lymphocytes were tested for their ability to lyse C1R lymphoblastoid B cells transfected with human CD1 proteins after a single in vitro stimulation by autologous thyroid-derived APCs. Although mock-transfected C1R cells were not lysed, there was moderate lysis of targets expressing CD1a, CD1c, and CD1d (Fig. 3 B). These studies provided evidence that lymphocytes infiltrating thyroid glands from patients with autoimmune thyroiditis have reactivity for CD1 proteins.

To investigate this further, bulk lymphocyte cultures from patients with HT (HT-1) and GD (GD-6) were stimulated in vitro with thyroid-derived APCs to generate short-term T cell lines. The resulting oligoclonal T cell lines were analyzed for their ability to lyse CD1-transfected C1R cells. We derived many lines that did not lyse C1R cells expressing CD1 proteins, and HT-1.4 is one such example (Fig. 4,A). Two independently derived lines from an HT gland (HT-1.1 and HT-1.2) lysed CD1a-transfected cells, but not mock-transfected C1R cells, demonstrating autoreactivity against CD1a (Fig. 4,B). Likewise, three lines derived from a gland with GD pathology (GD-6.1, GD-6.3, and GD6.5) selectively lysed target cells expressing CD1a (Fig. 4 C). Also supporting the CD1 dependence of the response, an anti-CD1a Ab resulted in 28% inhibition of the cytotoxicity of the T cell line HT-1.2 compared with a control Ab (data not shown). Taken together, these data provided evidence that CD1a-reactive T cells infiltrate the glands of patients with both forms of autoimmune thyroiditis.

FIGURE 4.

T cell lines derived from autoimmune thyroids recognize CD1a+ APCs. Oligoclonal T cell lines were isolated from thyroid infiltrates, and reactivity was tested using 51Cr-labeled C1R B cells expressing the indicated CD1 protein.

FIGURE 4.

T cell lines derived from autoimmune thyroids recognize CD1a+ APCs. Oligoclonal T cell lines were isolated from thyroid infiltrates, and reactivity was tested using 51Cr-labeled C1R B cells expressing the indicated CD1 protein.

Close modal

Two T cell lines, one from a thyroid gland affected with HT (HT-1.3) and one from a GD sample (GD-6.6), were found to weakly, but selectively, lyse transfectants expressing CD1c (Fig. 5,A). These T cells were tested for their ability to recognize previously described CD1c-presented lipid Ags. Previous studies had shown that CD1c presents mycobacterial mannosyl-β-1-phosphomycoketide lipids and that CD1c-restricted T cells are also reactive with structurally related lipids in which the unusual mycobacterial lipid tail is replaced with a synthetic lipid that resembles the polyunsaturated, polyisoprenyl tail to yield MPD (3). This synthetic MPD resembles the structure of common eukaryotic MPDs, except that it has a shorter lipid tail. Treatment of CD1 transfectants with semisynthetic MPD greatly increased the lysis of CD1c-transfected targets, but not of CD1b, CD1d, or mock-transfected C1R cells, providing evidence that lipid-mediated T cell activation requires CD1c. Using this method, we found that these thyroid-derived T cells did not recognize the fully saturated mannosyl-β-1-phosphomycoketide lipid from mycobacteria (data not shown). In addition, we found that full-length human C95 mannosyl phosphodolichols or phosphodolichols failed to activate either blood- or thyroid-derived lymphocytes from thyroiditis patients (Fig. 5 C and data not shown). Also, proliferation assays that had been previously shown to detect responses to C35 MPD in a majority of tuberculosis patients failed to elicit detectable peripheral blood T lymphocyte responses from patients with GD and HT, even though MPD-reactive T cell lines could be derived from the thyroid gland. This may reflect the homing of such T cells to the thyroid gland in this organ-specific disease, and future studies using single cell detection techniques may resolve this issue. In summary, these data provide evidence that both CD1a and CD1c are recognized by thyroid-infiltrating T cells and that CD1c restricts the recognition of a dolichyl phospholipid.

FIGURE 5.

Thyroid-infiltrating T cells recognize a dolichyl glycolipid presented by CD1c. A, The T cell line GD-6.6 was derived from a GD patient and tested for lysis with 51Cr-labeled C1R lines pulsed for 2 h with semisynthetic mannosyl-β-1-dolichol (10 μM). B, T cell line HT-1.3 derived from an HT patient was tested as described in A. C, Proliferation of fresh, polyclonal peripheral blood T cells from thyroid autoimmune patients or healthy donors to autologous, CD1-expressing monocyte-derived DCs in the absence or the presence of synthetic phosphodolichol (PD) and MPD with C35 or C95 lipid components at 10 μM.

FIGURE 5.

Thyroid-infiltrating T cells recognize a dolichyl glycolipid presented by CD1c. A, The T cell line GD-6.6 was derived from a GD patient and tested for lysis with 51Cr-labeled C1R lines pulsed for 2 h with semisynthetic mannosyl-β-1-dolichol (10 μM). B, T cell line HT-1.3 derived from an HT patient was tested as described in A. C, Proliferation of fresh, polyclonal peripheral blood T cells from thyroid autoimmune patients or healthy donors to autologous, CD1-expressing monocyte-derived DCs in the absence or the presence of synthetic phosphodolichol (PD) and MPD with C35 or C95 lipid components at 10 μM.

Close modal

Increasing evidence for TCR and Ag diversity within the CD1-restricted T cell repertoire and the discovery of self-lipid Ags have given rise to the hypothesis that encountering self- or altered self-lipids might lead to autoimmune tissue destruction (15). There are now many examples of long-term T cell lines that show autoreactivity to CD1 or recognize self-lipids, but there is little information about whether CD1 or lipid-reactive T cells and CD1-expressing APCs migrate into the sites of autoimmune tissue destruction in vivo. Autoimmune thyroiditis represents an attractive model for characterizing lesional T cells and APCs, because these diseases are known to be mediated by lymphocytes that home specifically to this organ while sparing all nearby tissues. The results reported in this study provide evidence that all the elements required for the CD1-mediated autoimmune response are present in thyroid glands during the acute and chronic stages of the autoimmune disease, including CD1-restricted T cells, CD1c+ mantle zone B cells, and CD1+ DCs.

We found that fresh thyroid-derived T lymphocytes can recognize and lyse autologous thyrocytes in a MHC class I-independent manner and lyse CD1-transfected cells in a CD1a-, CD1c-, or CD1d-dependent manner. Thus, the natural history of HT and GD involves the homing of CD1-restricted T cells to the site of tissue destruction. As with all known CD1 autoreactive T cells, this probably represents recognition of ligands by CD1, but the molecular identity of any natural self-lipid Ags has not yet been solved. The ability of a synthetic C35 mannosyl phosphodolichol (C35 MPD) to augment the CD1c autoreactive response provides some insight into the types of endogenous ligands that may be involved (Fig. 5). Synthetic C35 MPD cannot be considered a lipid autoantigen, because its lipid tail is much shorter than the C95 dolichols present in the thyroid gland (34). However, dolichyl lipids are present at very high levels in thyroid epithelial cells, where dolichol phosphates function as carbohydrate donors and acceptors during N-linked glycosylation of thyroglobulin (25, 35). Although CD1c-autoreactive T cells did not recognize any C95 form of dolichyl lipids, the augmentation of the response by C35 MPD provides support for the hypothesis that the response may involve an as yet unknown natural polyisoprenoid lipid autoantigen.

To date, the molecular analysis of the activation requirements of CD1-restricted T cells has relied extensively on CD1-transfected tumor cells or monocytes treated with GM-CSF and IL-4 to yield monocyte-derived DCs, which mimic many features of true tissue DCs (36). In an attempt to better understand the natural, tissue-based APCs, we isolated lesional cells and used them to generate short-term T cell lines and determine their phenotypes. The prominent formation of lymphoid follicles within the thyroid gland gives the DCs access to self-Ags and an organized secondary lymphoid structure at the site of tissue destruction, in contrast to other models in which tissue-specific Ags are transported to lymph nodes. Also, prior studies have shown that thyroidectomy leads to the loss of autoantibodies and other peripheral signs of the immune response, demonstrating the importance of local intrathyroid immune responses in driving the disease (37, 38). These considerations suggested that the thyroid gland itself would be a good source for identifying physiologically relevant, CD1-expressing APCs that are involved in activating CD1-restricted T cells.

Lesional APCs were found to be sufficient to promote the outgrowth of short-term, CD1-restricted T cell lines, and a detailed immunophenotype identified at least two subpopulations of APCs expressing group 1 CD1 proteins. The most abundant population is CD1c+ B cells that localize to the B cell-T cell interface of lymphoid follicles. This location along with their coexpression of IgD and lack of CD23 and CD5 indicate that these cells are mantle zone B cells. Although the precise functions of mantle zone B cells are not yet well understood, previously published and current observations of high CD1 expression in these cells point toward a possible role in glycolipid presentation to CD1c-restricted T cells. B cells from intrathyroidal lymphoid follicles are in a privileged location where they could capture tissue Ags and present them to T lymphocytes. A recent study has shown that CD1c-restricted T cells from patients with systemic lupus erythematosus can promote B cell maturation and Ig class switching (39). Recently, B cells with a predominantly mantle zone phenotype that infiltrate the thyroid in human GD have been shown to produce thyroid autoantibodies. Although these data do not prove that mantle zone B cells or their products are pathogenic, they suggest that they are committed to the response to thyroid self-Ags (40, 41). Furthermore, because they are outside the limits of lymphoid organs, thyroidal B cells may bypass normal peripheral tolerance mechanisms more easily, as has been proposed to explain the presence of rheumatoid factor autoantibodies in rheumatoid arthritis patients (42). In addition, we consistently detected CD11c+ or CD83+ cells that coexpressed group 1 CD1 proteins, CD1a, CD1b, or CD1c. These cell surface markers, along with the prominent dendritic morphology of CD1-expressing cells, identify these as tissue DCs.

In contrast with the proposed systemic immunoregulatory functions of invariant NK T cells, this study provides evidence that noninvariant NK T cells recognizing group 1 CD1 isoforms infiltrate the lesions of autoimmune thyroiditis. These results implicate group 1 CD1 proteins in modulating these common autoimmune diseases and provide a new model for the study of group 1 CD1 proteins ex vivo in humans. The precise mechanisms by which lipids and CD1 could influence the final steps in immunopathogenesis, which clearly involve the production of autoantibodies against thyroid-specific proteins, have not been determined. However, CD1-restricted T cells have been recently shown to influence DC and B cell maturation, providing possible insights into pathogenic mechanisms. In particular, CD1a, CD1b, and CD1c are expressed on myeloid DCs at early stages of maturation, and CD1-restricted T cells have been shown to promote this maturation process, so that DCs become fully able to prime MHC-restricted T cells (43). In this regard, it is of interest that CD1a and CD1c autoreactive T cells were isolated from lesions with large numbers of CD1+CD83 + mature DCs. These observations form the basis for new ideas about the pathogenesis of human autoimmune disease in which tissues provide both lipid and protein Ags that are simultaneously processed and presented by CD1 and MHC proteins, leading to the activation of lipid- and peptide-reactive T cells that influence one another.

The authors have no financial conflict of interest.

We thank Manuela Cernadas, Nancy Kedersha, and Marisa Nucci for their advice, support, and kindly providing reagents; M. Gately for providing human rIL-2, and A. Lanzavecchia for providing mAb C15. We acknowledge the help of Drs. A. Lucas and A. Alastrue and their teams of the Endocrinology Service of the Hospital Germans Trias i Pujol (Badalona, Spain), for providing the surgical thyroid samples.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by grants from the Cancer Research Institute; the American College of Rheumatology Research, and Education Foundation; the Pew Foundation Scholars in the Biomedical Sciences Program; National Institutes of Health Grants AI50216 and AR48632 (to D.B.M.); and Grant FISS99-1231 from the Fondo de Investigaciones Sanitarias of the Spanish Ministry of Health (to D.J.). G.S.B., a Lister Jenner Research Fellow, acknowledges support from the Medical Research Council and the Wellcome Trust.

3

Abbreviations used in this paper: DC, dendritic cell; GD, Graves’ disease; HT, Hashimoto’s thyroiditis; MPD, mannosyl phosphodolichol; TPO, thyroid peroxidase.

1
Sieling, P. A., D. Chatterjee, S. A. Porcelli, T. I. Prigozy, R. J. Mazzaccaro, T. Soriano, B. R. Bloom, M. B. Brenner, M. Kronenberg, P. J. Brennan, et al
1995
. CD1-restricted T cell recognition of microbial lipoglycan antigens.
Science
269
:
227
.
2
Naidenko, O. V., J. K. Maher, W. A. Ernst, T. Sakai, R. L. Modlin, M. Kronenberg.
1999
. Binding and antigen presentation of ceramide-containing glycolipids by soluble mouse and human CD1d molecules.
J. Exp. Med.
190
:
1069
.
3
Moody, D. B., T. Ulrichs, W. Muhlecker, D. C. Young, S. S. Gurcha, E. Grant, J. P. Rosat, M. B. Brenner, C. E. Costello, G. S. Besra, et al
2000
. CD1c-mediated T-cell recognition of isoprenoid glycolipids in Mycobacterium tuberculosis infection.
Nature
404
:
884
.
4
Moody, D. B., D. C. Young, T. Y. Cheng, J. P. Rosat, C. Roura-Mir, P. B. O’Connor, D. M. Zajonc, A. Walz, M. J. Miller, S. B. Levery, et al
2004
. T cell activation by lipopeptide antigens.
Science
303
:
527
.
5
Moody, D. B., B. B. Reinhold, V. N. Reinhold, G. S. Besra, S. A. Porcelli.
1999
. Uptake and processing of glycosylated mycolates for presentation to CD1b-restricted T cells.
Immunol. Lett.
65
:
85
.
6
Shamshiev, A., A. Donda, I. Carena, L. Mori, L. Kappos, G. De Libero.
1999
. Self glycolipids as T-cell autoantigens.
Eur. J. Immunol.
29
:
1667
.
7
Shamshiev, A., H. J. Gober, A. Donda, Z. Mazorra, L. Mori, G. De Libero.
2002
. Presentation of the same glycolipid by different CD1 molecules.
J. Exp. Med.
195
:
1013
.
8
Joyce, S., A. S. Woods, J. W. Yewdell, J. R. Bennink, A. D. De Silva, A. Boesteanu, S. P. Balk, R. J. Cotter, R. R. Brutkiewicz.
1998
. Natural ligand of mouse CD1d1: cellular glycosylphosphatidylinositol.
Science
279
:
1541
.
9
De Silva, A. D., J. J. Park, N. Matsuki, A. K. Stanic, R. R. Brutkiewicz, M. E. Medof, S. Joyce.
2002
. Lipid protein interactions: the assembly of CD1d1 with cellular phospholipids occurs in the endoplasmic reticulum.
J. Immunol.
168
:
723
.
10
Porcelli, S., M. B. Brenner, J. L. Greenstein, S. P. Balk, C. Terhorst, P. A. Bleicher.
1989
. Recognition of cluster of differentiation 1 antigens by human CD4CD8 cytolytic T lymphocytes.
Nature
341
:
447
.
11
Wu, D. Y., N. H. Segal, S. Sidobre, M. Kronenberg, P. B. Chapman.
2003
. Cross-presentation of disialoganglioside GD3 to natural killer T cells.
J. Exp. Med.
198
:
173
.
12
Gumperz, J. E., C. Roy, A. Makowska, D. Lum, M. Sugita, T. Podrebarac, Y. Koezuka, S. A. Porcelli, S. Cardell, M. B. Brenner, et al
2000
. Murine CD1d-restricted T cell recognition of cellular lipids.
Immunity
12
:
211
.
13
Schofield, L., M. J. McConville, D. Hansen, A. S. Campbell, B. Fraser-Reid, M. J. Grusby, S. D. Tachado.
1999
. CD1d-restricted immunoglobulin G formation to GPI-anchored antigens mediated by NKT cells.
Science
283
:
225
.
14
Rauch, J., J. Gumperz, C. Robinson, M. Skold, C. Roy, D. C. Young, M. Lafleur, D. B. Moody, M. B. Brenner, C. E. Costello, et al
2003
. Structural features of the acyl chain determine self-phospholipid antigen recognition by a CD1d-restricted invariant NKT (iNKT) cell.
J. Biol. Chem.
278
:
47508
.
15
Godfrey, D. I., M. Kronenberg.
2004
. Going both ways: immune regulation via CD1d-dependent NKT cells.
J. Clin. Invest.
114
:
1379
.
16
Nagayama, Y., K. Watanabe, M. Niwa, S. M. McLachlan, B. Rapoport.
2004
. Schistosoma mansoni and α-galactosylceramide: prophylactic effect of Th1 Immune suppression in a mouse model of Graves’ hyperthyroidism.
J. Immunol.
173
:
2167
.
17
Wilson, S. B., S. C. Kent, K. T. Patton, T. Orban, R. A. Jackson, M. Exley, S. Porcelli, D. A. Schatz, M. A. Atkinson, S. P. Balk, et al
1998
. Extreme Th1 bias of invariant Vα24JαQ T cells in type 1 diabetes.
Nature
391
:
177
.
18
Sumida, T., A. Sakamoto, H. Murata, Y. Makino, H. Takahashi, S. Yoshida, K. Nishioka, I. Iwamoto, M. Taniguchi.
1995
. Selective reduction of T cells bearing invariant Vα24JαQ antigen receptor in patients with systemic sclerosis.
J. Exp. Med.
182
:
1163
.
19
Van der Vliet, H. J., B. M. 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(α24+) Vβ11+ NKT cell numbers are decreased in a wide variety of diseases that are characterized by autoreactive tissue damage.
Clin. Immunol.
100
:
144
.
20
Lee, P. T., A. Putnam, K. Benlagha, L. Teyton, P. A. Gottlieb, A. Bendelac.
2002
. Testing the NKT cell hypothesis of human IDDM pathogenesis.
J. Clin. Invest.
110
:
793
.
21
Hammond, K. J., M. Kronenberg.
2003
. Natural killer T cells: natural or unnatural regulators of autoimmunity?.
Curr. Opin. Immunol.
15
:
683
.
22
Naumov, Y. N., K. S. Bahjat, R. Gausling, R. Abraham, M. A. Exley, Y. Koezuka, S. B. Balk, J. L. Strominger, M. Clare-Salzer, S. B. Wilson.
2001
. Activation of CD1d-restricted T cells protects NOD mice from developing diabetes by regulating dendritic cell subsets.
Proc. Natl. Acad. Sci. USA
98
:
13838
.
23
Beaudoin, L., V. Laloux, J. Novak, B. Lucas, A. Lehuen.
2002
. NKT cells inhibit the onset of diabetes by impairing the development of pathogenic T cells specific for pancreatic β cells.
Immunity
17
:
725
.
24
Ishizuka, I..
1997
. Chemistry and functional distribution of sulfoglycolipids.
Prog. Lipid Res.
36
:
245
.
25
Carroll, K. K., N. Guthrie, K. Ravi.
1992
. Dolichol: function, metabolism, and accumulation in human tissues.
Biochem. Cell. Biol.
70
:
382
.
26
Jahng, A., I. Maricic, C. Aguilera, S. Cardell, R. C. Halder, V. Kumar.
2004
. Prevention of autoimmunity by targeting a distinct, noninvariant CD1d-reactive T cell population reactive to sulfatide.
J. Exp. Med.
199
:
947
.
27
Tolosa, E., C. Roura, M. Catalfamo, M. Marti, A. Lucas-Martin, A. Sanmarti, I. Salinas, G. Obiols, M. Foz-Sala, R. Pujol-Borrell.
1992
. Expression of intercellular adhesion molecule-1 in thyroid follicular cells in autoimmune, non-autoimmune and neoplastic diseases of the thyroid gland: discordance with HLA.
J. Autoimmun.
5
:
107
.
28
Roura-Mir, C., M. Catalfamo, M. Sospedra, L. Alcalde, R. Pujol-Borrell, D. Jaraquemada.
1997
. Single-cell analysis of intrathyroidal lymphocytes shows differential cytokine expression in Hashimoto’s and Graves’ disease.
Eur. J. Immunol.
27
:
3290
.
29
Martin, R., D. Jaraquemada, M. Flerlage, J. Richert, J. Whitaker, E. O. Long, D. E. McFarlin, H. F. McFarland.
1990
. Fine specificity and HLA restriction of myelin basic protein-specific cytotoxic T cell lines from multiple sclerosis patients and healthy individuals.
J. Immunol.
145
:
540
.
30
Grant, E. P., M. Degano, J. P. Rosat, S. Stenger, R. L. Modlin, I. A. Wilson, S. A. Porcelli, M. B. Brenner.
1999
. Molecular recognition of lipid antigens by T cell receptors.
J. Exp. Med.
189
:
195
.
31
Behar, S. M., S. A. Porcelli, E. M. Beckman, M. B. Brenner.
1995
. A pathway of costimulation that prevents anergy in CD28 T cells: B7-independent costimulation of CD1-restricted T cells.
J. Exp. Med.
182
:
2007
.
32
Exley, M., J. Garcia, S. P. Balk, S. Porcelli.
1997
. Requirements for CD1d recognition by human invariant Vα24+ CD4CD8 T cells.
J. Exp. Med.
186
:
109
.
33
van Seventer, G. A., H. Spits, H. Yssel, C. J. Melief, P. Ivanyi.
1988
. Differential recognition by human cytotoxic T cell clones of human M1 fibroblasts transfected with an HLA-B7 gene (JY150) suggests the existence of two different HLA-B7 alleles in the cell line JY (HLA-A2,2;B7,7;Cw-,-;DR4, w6).
J. Immunol.
141
:
417
.
34
Burgos, J., F. W. Hemming, J. F. Pennock, R. A. Morton.
1963
. Dolichol: a naturally-occurring C100 isoprenoid alcohol.
Biochem J.
88
:
470
.
35
Spiro, M. J., R. G. Spiro.
1986
. Control of N-linked carbohydrate unit synthesis in thyroid endoplasmic reticulum by membrane organization and dolichyl phosphate availability.
J. Biol. Chem.
261
:
14725
.
36
Sallusto, F., A. Lanzavecchia.
1994
. Efficient presentation of soluble antigen by cultured human dendritic cells is maintained by granulocyte/macrophage colony-stimulating factor plus interleukin 4 and downregulated by tumor necrosis factor α.
J. Exp. Med.
179
:
1109
.
37
Cho, B. Y., S. K. Oh, Y. K. Shong, S. E. Kim, H. Y. Yoo, H. K. Lee, C. S. Koh, H. K. Min.
1990
. Changes in thyrotropin receptor antibody after subtotal thyroidectomy in Graves’ disease: comparison with the degree of lymphocytic infiltration in the thyroid.
Autoimmunity
8
:
143
.
38
Tew, J. G., R. P. Phipps, T. E. Mandel.
1980
. The maintenance and regulation of the humoral immune response: persisting antigen and the role of follicular antigen-binding dendritic cells as accessory cells.
Immunol. Rev.
53
:
175
.
39
Sieling, P. A., S. A. Porcelli, B. T. Duong, F. Spada, B. R. Bloom, B. Diamond, B. H. Hahn.
2000
. Human double-negative T cells in systemic lupus erythematosus provide help for IgG and are restricted by CD1c.
J. Immunol.
165
:
5338
.
40
Armengol, M. P., M. Juan, A. Lucas-Martin, M. T. Fernandez-Figueras, D. Jaraquemada, T. Gallart, R. Pujol-Borrell.
2001
. Thyroid autoimmune disease: demonstration of thyroid antigen-specific B cells and recombination-activating gene expression in chemokine-containing active intrathyroidal germinal centers.
Am. J Pathol.
159
:
861
.
41
Segundo, C., C. Rodriguez, M. Aguilar, A. Garcia-Poley, I. Gavilan, C. Bellas, J. A. Brieva.
2004
. Differences in thyroid-infiltrating B lymphocytes in patients with Graves’ disease: relationship to autoantibody detection.
Thyroid
14
:
337
.
42
Pulendran, B., R. van Driel, G. J. Nossal.
1997
. Immunological tolerance in germinal centres.
Immunol. Today
18
:
27
.
43
Vincent, M. S., D. S. Leslie, J. E. Gumperz, X. Xiong, E. P. Grant, M. B. Brenner.
2002
. CD1-dependent dendritic cell instruction.
Nat. Immunol.
3
:
1163
.