THYROIDVolume 11, Number 3, 2001Mary Ann Liebert, Inc.

Differential Regulation of Fas-Mediated Apoptosis in BothThyrocyte and Lymphocyte Cellular Compartments

Correlates with Opposite Phenotypic Manifestations ofAutoimmune Thyroid Disease

Carla Giordano,1 Pierina Richiusa,1 Marcello Bagnasco,2 Giuseppe Pizzolanti,1 Francesco Di Blasi,3

Maria S. Sbriglia,1 Antonina Mattina,1 Gianpaola Pesce,2 Paola Montagna,2 Francesca Capone,1

Gabriella Misiano,1 Alessandro Scorsone,1 Alberto Pugliese,4 and Aldo Galluzzo1

Several mechanisms are probably involved in determining the evolution of autoimmune thyroid disease (AITD)towards either hypothyroidism and the clinical syndrome known as Hashimoto’s thyroiditis (HT) or towardhyperthyroidism and the symptoms of Graves’ disease (GD). To gain further insight into such mechanisms weperformed an exhaustive comparative analysis of the expression of key molecules regulating cell death (Fas,Fas ligand [FasL], Bcl-2) and apoptosis in both thyrocytes and thyroid infiltrating lymphocytes (TILs) from pa-tients with either GD or HT. GD thyrocytes expressed less Fas/FasL than HT thyrocytes, whereas GD TILs hadhigher levels of Fas/FasL than HT TILs. GD thyrocytes expressed increased levels of the antiapoptotic mole-cule Bcl-2 compared to the low levels detected in HT thyrocytes. The opposite pattern was observed in GD(low Bcl-2) and HT (high Bcl-2) TILs. The patterns of apoptosis observed were consistent with the regulationof Fas, FasL, and Bcl-2 described above. Our findings suggest that in GD thyroid the regulation of Fas/FasL/Bcl2favors apoptosis of infiltrating lymphocytes, possibly limiting their autoreactive potential and impairing theirability to mediate tissue damage. Moreover, the reduced levels of Fas/FasL and increased levels of Bcl-2 shouldfavor thyrocyte survival and favor the thyrocyte hypertrophy associated with immunoglobulins stimulatingthe thyrotropin (TSH) receptor. In contrast, the regulation of Fas/FasL/Bcl2 expression in HT promotes thy-rocyte apoptosis, tissue damage, and a gradual reduction in thyrocyte numbers leading to hypothyroidism.These findings help define key molecular mechanisms contributing to the clinical outcome of thyroid autoim-munity.

233

Introduction

LOSS OF IMMUNOLOGICAL TOLERANCE to thyroid autoantigens(thyroid peroxidase, thyroglobulin) is associated with

two forms of autoimmune thyroid disease (AITD):Hashimoto’s thyroiditis (HT) and Graves’ disease (GD) (1).HT is characterized by lymphocytic infiltration of the thy-roid, severe tissue damage, and hypothyroidism. In contrast,the main features of GD are thyroid hypertrophy and hy-perthyroidism resulting from the effects of stimulating immunoglobulins (TSIs) against the thyrotropin (TSH) re-ceptor and autoimmune tissue damage (2). Although au-toimmune responses against specific antigens appear as ma-

jor determinants in AITD pathogenesis, other molecularmechanisms may play a role in determining the oppositephenotypic outcomes of AITD. Great attention has been fo-cused on the expression of Fas, Fas ligand (FasL), and Bcl-2,molecules that are involved in the regulation of apoptoticcell death (3–5). Growing evidence suggests that the abnor-mal expression of Fas/FasL and impaired regulation of Fas-mediated apoptosis may be critical in a number of autoim-mune diseases, including systemic lupus erithematosus(SLE), multiple sclerosis, type 1 diabetes, and AITD (3,6–8).We and others have recently characterized Fas/FasL ex-pression in thyrocytes from normal thyroids and nontoxicgoiters (NTG). Both normal and NTG thyrocytes have been

1Laboratory of Immunology, Endocrinology, Institute of Clinica Medica, University of Palermo, Italy.2Allergology and Clinical Immunology, D.I.M.I., University of Genoa, Italy.3Institute of Biology, CNR, Palermo, Italy.4Immunogenetics, Diabetes Research Institute, University of Miami School of Medicine, Miami, Florida.

shown to express FasL, an observation we originally madeusing an antibody that was associated with cross-reactivityartifacts but that was later confirmed by different groups us-ing a variety of well-characterized monoclonal antibodies(9–15). Moreover, we reported that thyrocytes from HTglands, in contrast to NTG and normal thyroids, hyperex-press Fas and FasL on their surface (9). Interleukin-1b (IL-1b), a cytokine abundantly produced in HT glands, inducesFas expression in normal thyrocytes, and cross-linking of Fasresults in massive thyrocyte apoptosis that can be preventedby antibodies that block Fas. Thus, the interaction ofFas/FasL on the surface of neighboring thyrocytes may trig-ger their “fratricidal” apoptosis and represent an importantmechanism of thyrocyte destruction in HT.

Similarly, autoimmune responses against several thyroid-specific antigens and lymphocytic infiltration of the thyroid,although to a lesser extent than in HT, are also observed inGD. Although TSIs may explain thyroid hypertrophy andthe dramatic symptoms of hyperthyroidism, the presence ofautoreactive lymphocytes in GD thyroid leads to less signif-icant tissue damage (2). The enlargement of the thyroid glandtypical of active GD can be related to the increased rates ofcell proliferation and decreased rates of cell death, processesthat are regulated by the expression of different moleculesincluding Fas and FasL. Thus, a different pattern of Fas/FasLexpression may characterize GD thyroids compared to HTglands, and such differences may be responsible for the re-duced thyroid cell apoptosis observed in GD (16). To test thishypothesis, we purified and independently examined thetwo main cellular components of the disease, thyroid infil-trating lymphocytes (TILs) and follicular thyrocytes, fromsurgical thyroid specimens obtained from patients with GD,HT, or NTG. We evaluated the expression of Fas, FasL, Bcl-2, and apoptosis in both thyrocytes and lymphocytes andfound evidence that differential regulation of apoptosis inthese two cellular compartments correlates with the oppo-site phenotypic outcomes (hypothyroidism and hyperthy-roidism) of thyroid autoimmunity.

Materials and Methods

Patients and tissue specimens

Thyroid specimens were obtained from 25 patients withAITD (16 with GD, 9 with HT) who underwent total or par-tial thyroidectomy at our clinic. We also obtained specimensfrom 25 patients with NTG. Most patients underwentsurgery because of the presence of large goiters with trachealdislocation (Table 1). Three patients with HT had their thy-roids removed because cytology examination of fine-needleaspiration biopsy specimens suggested the presence of a thy-roid lymphoma, but this possibility was excluded in all casesby subsequent histological examination. All patients wereCaucasian, 7 were males, 43 were females, the age range was21–73 years. All GD patients (n 5 16) were treated with anantithyroid agent (methimazole) and b-adrenergic antago-nists. Methimazole has been associated with upregulation ofFasL expression in thyrocytes both in vitro and in vivo (17,18).The diagnosis was established on the basis of commonly ac-cepted clinical and laboratory parameters and the typical his-tological features associated with each condition (1). All pa-tients were euthyroid at the time of subtotal or totalthyroidectomy. None of the specimens were obtained from

patients treated with radioactive iodine (131I), Lugol, or SSKIbecause these agents may affect apoptosis in thyrocytes(Giordano C, personal observation).

Preparation and flow-cytometry analysis of thyrocytes andintrathyroidal lymphocytes

Thyrocyte isolation was performed using the collagenasemethod as previously described (9). The digest was filteredthrough a 200-mm mesh, and cells were plated in 75-cm2

flasks in complete medium (RPMI 1640 supplemented with10% fetal calf serum, antibiotics, and glutamine). Afterovernight adhesion, unattached cells (intrathyroidal lym-phocytes) were collected, filtered (Falcon strainer: 100 mm,70 mm, 40 mm) and purified by Ficoll Hypaque density gra-dient separation to eliminate debris, red blood cells, anddead cells. Cell viability was tested by orange acridine ex-clusion. For lymphocyte staining, cells were incubated for 30minutes at 4°C with anti-Fas (UB2, immunoglobulin G1(IgG1); Kamiya Biomedical Company, Thousand Oaks, CA)or anti-FasL (NOK-2, IgG2a; PharMingen, San Diego, CA;H11, IgG2a, Alexis, San Diego, CA) monoclonal antibodies(mAbs), or with isotype control IgG1 and IgG2a mAbs. Cellswere washed and treated with fluorescein isothiocyanateconjugated (FITC) goat anti-mouse (Dako, Copenhagen, Den-mark) or rabbit anti-rat IgG (Southern Biotechnology Asso-ciates, Inc., Birmingham, AL). After another washing, cellswere incubated for 10 minutes with 6% species-specific nor-mal serum and treated with saturating concentrations ofphycoerythrin (PE)-conjugated anti-CD3 (IgG1) or controlPE-IgG1 (Becton Dickinson, San Jose, CA). Alternatively,cells were incubated for 30 minutes at 4°C with a combina-tion of FITC anti-CD3, and the following PE mAbs: anti-CD4(IgG1), anti-CD8 (IgG1), anti-CD19 (IgG1), anti-CD25 (IgG1),anti-CD69 (IgG1), anti-CD14 (IgG2b), anti-CD45 (IgG1) (allfrom Becton Dickinson). Intracellular Bcl-2 was detected inpermeabilized cells, as previously described (19). The anti-Bcl-2 mAb (C2, IgG1; Santa Cruz Biotechnology, CA) wasused at the concentration of 1 mg per 106 cells. Adherent cells(thyrocytes) were removed from the flasks with trypsin eth-ylenediaminetetraacetic acid (EDTA) (0.05% and 0.02%, re-spectively). For immunodepletion of contaminating hema-topoietic cells, single-cell thyrocyte suspensions were treatedwith anti-CD45 (anti-Leukocyte; Becton Dickinson) for 30minutes at 4°C. CD451 cells were then removed after 30 min-utes binding to sheep anti-mouse IgG-coupled beads (Dynal,Wirral Merseyside, UK) and magnetic depletion (9). For thy-rocyte staining, cells were incubated for 30 minutes at 4°Cwith human serum containing anti-TPO antibodies or anti-TPO mAb (IgG1; Biocytex, Marseille, France), or isotype-matched control Ig. Cells were washed and labeled with PE-conjugated goat anti-human Ig (Dako), washed again,saturated, and labeled as already described with anti-Fas oranti-FasL mAbs. All samples were analyzed with a FACScanflow cytometer (Becton Dickinson).

Immunohistochemistry and double staining

Immediately after surgical removal thyroid specimenswere reduced to fragments, quick-frozen in 2-methylbutaneand stored at 280°C. Frozen sections (4 mm) were cut in aReichert-Jung cryotome and kept at 270°C until use. Im-munohistochemical staining was performed according to the

GIORDANO ET AL.234

streptavidin alkaline phosphatase (AP) method, as previ-ously described, with minor modifications (20). Briefly, sec-tions were pretreated with the avidin/biotin blocking kit(BioGenex, San Ramon, CA) and then incubated with the ap-propriate dilution of each mAb (30 minutes). Tissue sectionswere washed in Tris-buffered saline (TBS, pH 7.6) and incu-bated for 20 minutes at room temperature with biotinylatedanti-mouse immunoglobulins, followed by streptavidin-AP

complex (all from BioGenex). The substrate solution con-tained basic new fucsin, naphthol As-Bi phosphate, and lev-amisole (all from Sigma Chemical Co., St Louis, MO). Con-trol sections were incubated with irrelevant isotype-matchedmAbs. The preparations were counterstained with Carazzi’shematoxylin. For double staining, streptavidin-peroxidaseand streptavidin-AP were used in combination, as previ-ously described (20). The preparations were mounted in

APOPTOSIS PATTERN IN AITD 235

TABLE 1. CLINICAL CHARACTERISTICS OF THE PATIENTS STUDIED

Case # Gender Age Disease TPO-Ab (U/mL) TRAb (U/L) TSH (mU/L)

1 F 48 HT 247.17 Absent ,0.012 F 58 HT 680 0.763 F 59 HT .400 “ 0.234 M 42 HT 460.18 “ 0.915 F 60 HT 083.99 “ 1.866 F 30 HT 0480 1.247 F 52 HT .400 “ 0.48 F 69 HT 071.78 “ 0.349 F 69 HT 305.3 “ 1.6110 F 27 GD 045 14 2.211 F 31 GD 058.84 43 0.0512 F 43 GD 156.02 52 0.913 F 63 GD 247.13 71 0.614 F 21 GD .400 12 0.3415 F 44 GD 091.88 54 0.0716 M 49 GD 033.92 16 0.917 F 64 GD 391.8 30 0.9318 F 24 GD .400 105 ,0.0119 F 66 GD 199.6 45 ,0.0120 F 38 GD 395 13 0.3121 F 57 GD 166.7 110 0.0222 F 32 GD 072.02 45 0.3323 F 44 GD 103.27 50 ,0.0124 M 31 GD 101.6 95 0.7225 F 27 GD .400 160 ,0.0126 M 31 NTG Absent Absent 1.5627 F 69 NTG “ “ 0.4228 F 46 NTG “ “ 1.429 F 52 NTG “ “ 0.130 F 40 NTG “ “ 1.0831 F 44 NTG “ “ 0.0532 F 49 NTG “ “ 4.133 F 54 NTG “ “ 0.5434 F 63 NTG “ “ 0.7635 F 29 NTG “ “ 2.236 F 48 NTG “ “ 0.0437 F 35 NTG “ “ 0.1238 F 57 NTG “ “ 0.0539 F 59 NTG “ “ 1.3740 M 47 NTG “ “ 0.441 F 73 NTG “ “ 0.242 M 54 NTG “ “ 0.343 F 39 NTG “ “ 2.7444 M 69 NTG “ “ 0.9845 F 31 NTG “ “ 0.9646 F 33 NTG “ “ 1.6447 F 60 NTG “ “ 1.348 F 51 NTG “ “ 0.549 F 40 NTG “ “ 0.9550 F 35 NTG “ “ 2.2

Normal values: TPO-Ab , 18 U/mL; TRAb , 9 U/L; TSH 0.23–4 mU/L.TPO-Ab, thyroid peroxidase antibodies; TRAb, Thyrotropin receptor antibodies; TSH, thyrotropin; HT, Hashimoto’s thy-

roiditis; GD, Graves’ disease; NTG, nontoxic goiter.

buffered glycerol and examined by means of a Leitz Labor-lux microscope (Wild Microscopes, Rockleigh, NJ) equippedfor microphotography.

Apoptosis detection

Cell death or apoptosis was assessed with several meth-ods. The TUNEL method was used to detect apoptotic cellson frozen tissue sections (Boehringer Mannheim). Briefly,sections were fixed in 4% paraformaldehyde for 10 minutesand, after washing, blocking of endogenous peroxidase(0.3% H2O2 in methanol) and cell permeabilization (0.1% Tri-ton-X100 in 0.1% sodium citrate) were performed. Sectionswere covered with coverslips and incubated in a humidifiedchamber for 60 minutes at 37°C with 50 mL TUNEL reactionmixture (terminal deoxynucleotidyl-transferase from calfthymus and nucleotide mixture buffer). After three wash-ings, sections were incubated in a humidified chamber for30 minutes at 37°C with 50 mL convert-POD (antifluoresceinantibody, Fab fragment from sheep, conjugated with horse-radish peroxidase). DAB was used as substrate. Sectionswere counterstained with Carazzi’s hematoxylin, mountedin buffered glycerol, and examined with a Leitz Laborlux mi-croscope (Wild Microscopes, Rockleigh, NJ). For culturedcells, the percentage of hypodiploid nuclei was determinedby hypotonic fluorochrome staining by flow-cytometry ac-cording to Nicoletti’s method (21).

RNA extraction and cDNA synthesis

Total RNA was extracted from purified thyrocytes usingTRIzol reagent (Life Technologies, Paisley, UK) according tothe manufacturer’s instructions. RNA was treated withRNAse-free DNAse I and further purified using the S.N.A.P.kit (Invitrogen, Carlsbad, CA). Two micrograms of total RNAand 1 mL of oligo(dT)15 (final concentration 1 mM; PerkinElmer, Foster City, CA) were heated to 65°C for 5 minutes fol-lowed by cooling on ice. cDNA synthesis was performed byadding 100 units of Moloney murine leukemia virus (MMLV)reverse transcriptase (Amersham, Buckinghamshire, UK), 40URNAsin (Amersham), 1 mM dNTPs (Amersham) in a finalvolume of 20 mL, and the reaction was allowed to proceed at25°C for 10 minutes and then at 37°C for 60 minutes. Heatingat 95°C for 5 minutes terminated the reaction.

Semiquantitative multiplex RT-PCR for Fas, FasL and b-actin genes

We chose a multiplex approach to analyze the relative ex-pression levels of Fas gene compared to b-actin, as alreadydescribed by Otsuki et al. (22). Primer sequences were pre-viously described (22). Each reverse transcription-poly-merase chain reaction (RT-PCR) contained 1 mL of cDNA(equivalent to a cDNA amount from 100 ng of initial totalRNA), 200 mM of each deoxynucleotide triphosphates, 1.5mM of MgCl2, 13 TaqGold buffer (Perkin-Elmer), 200 nM ofFas primers and 20 nM of actin primers, 1.25 units of Taq-Gold in a final volume of 50 mL. After an initial period of 10 minutes at 95°C, 29 cycles of denaturation (95°C for 1minute), annealing (57°C for 1 minute) and extension (72°Cfor 1 minute) were performed on a MasterCycler (Eppen-dorf, Milan, Italy) followed by a final extension of 7 minutesat 72°C. Our choice of 29 PCR cycles allowed analyzing re-sults in the exponential phase of the amplification curve.

Products were electrophoresed on 2% agarose gel and thebands corresponding to Fas (366 bp), FasL (294 bp), and b-actin (661 bp) visualized by ethidium bromide staining. Gelswere photographed using 667 Polaroid film, and picturesscanned using a Epson GT5000 scanner.

Western blot

Purified thyrocytes or lymphocytes were washed twicewith Ca21- and Mg21-free PBS buffer. Cell extracts were pre-pared by lysing 107 cells in 100 mL of modified RIPA buffercontaining 50 mM Tris-HCl pH 7.4, 1% NP-40, 0.25% sodiumdeoxycholate, 150 mM sodium chloride, 1 mM EDTA, 1 mMsodium ortovanadate, 1 mM sodium fluoride, 1 mM phenyl-methylsulfonyl fluoride and aprotinin, leupeptin, pepstatinA, E-64 1 mg/mL of each (all from Sigma), at 4°C. Insolublematerials were removed by centrifugation for 20 minutes at14,000 rpm at 4°C. Protein content was determined using theBradford method (Bio-Rad protein assay kit, Hercules, CA).Protein samples (10 or 20 mg of protein) were boiled for 5minutes in a sodium dodecyl sulfate (SDS) buffer and sepa-rated in 12% SDS-polyacrylamide denaturing gels (SDS-PAGE) according to Laemmli (23). Proteins were elec-troblotted to nitrocellulose membranes (Bio-Rad). Theloading and transfer of equal amounts of protein was con-firmed by staining the membrane with Ponceau S. The blotswere blocked by incubation in PBS with 0.05% Tween 20(PBST) and 5% nonfat dry milk at 4°C overnight. After a briefrinse, blots were incubated for 1 hour at room temperatureor overnight at 4°C in PBST-5% milk with one of the fol-lowing primary antibodies: rabbit anti-Fas polyclonal anti-body (Fas-Ab1, Calbiochem, La Jolla, CA) at 0.5 mg/mL; rab-bit anti-FasL polyclonal antibody (FasL C-178, Santa Cruz)at 1 mg/mL; mAb for Bcl-2 (Bcl2 C-2, Santa Cruz) at 1 mg/mL.Blots were washed four times with PBST, incubated with an-tirabbit or antimouse IgG-peroxidase conjugate (Santa Cruz)(1:4000) in PBST-5% milk for 1 hour at room temperature,washed again, and developed with a enhanced chemilumi-nescence reagent (Pierce, Rockford, IL).

Quantitative analysis of mRNA

Fas and FasL mRNA was measured by S1 quantitativeanalysis as described (24). In brief, two oligonucleotides of 77bases with the first 72 bases complementary to the Fas anti-gen mRNA and 65 bases with the first 60 bases complemen-tary to FasL mRNA (Fas oligo 59-CCC CAA GTT AGA TCTGGA TCC TTC CTC TTT GCA CTT GGT GTT GCT GGT GAGTGT GCA TTC CTT GAT GAT TCC GCA CC-39; Fas ligandoligo 59-ACC CCG GAA GTA TAC TTT GGA ATA TAC AAAGTA CAG CCC AGT TTC ATT GAT CAC AAG GCC CAAAG-39) were radiolabeled with T4 polynucleotide kinase(Amersham). The Fas probe was hybridized with 15 mg of to-tal RNA extracted from thyrocytes, intrathyroidal lympho-cytes, or 5 mg from HeLa (positive control). The FasL probewas hybridized with 40 mg of total RNA extracted from thy-rocytes or 10 mg from phorbol hyristate acetate (PMA)-acti-vated Jurkat cells (positive control). Both Fas and FasL probeswere then digested with 100 units of S1 nuclease (Roche Mo-lecular System, Mannheim, Germany). Samples were resolvedon 6% denaturing polyacrylamide gels and subject to autora-diography. The loading of equal amounts of RNA was con-firmed by Northern Blot analysis using a radiolabeled probespecific for human b-actin (data not shown).

GIORDANO ET AL.236

Densitometry

Autoradiograms from S1 RNA analysis and Western blotswere scanned and the intensity of the bands quantified indensitometric units (DU) using a Image Quant Software(Molecular Dynamics). Results are expressed as ratio of theintensity of the band of interest over the intensity of the ap-propriate controls. Data are expressed as mean values 6 SD.

Statistical analysis

Statistical significance was evaluated using the Student’st test for independent samples. Multiple comparisons be-tween groups were done by analysis of variance (ANOVA).Differences between groups with p , 0.05 were consideredstatistically significant.

Results

Fas expression in thyrocytes and TILs

Fas mRNA expression was evaluated by quantitative S1analysis in both thyrocytes and TILs isolated from GD andHT thyroids (Fig. 1A). The levels of Fas transcript in GD thy-rocytes were approximately half than in HT thyrocytes(Table 2) (p , 0.001). Similar findings were obtained by semi-

quantitative RT-PCR (Fig. 2A), with Fas mRNA being clearlymore abundantly expressed in HT (mean Fas/actin ratio0.99 6 0.2) compared to GD (mean Fas/actin ratio 0.55 6 0.3,p , 0.001) and NTG thyrocytes (mean Fas/actin ratio 0.42 60.22). Western blot analysis (Fig. 2B) confirmed that the dif-ferences in Fas mRNA expression in purified thyrocytestranslated into similar differences at the protein level. In-deed, Fas protein was highly expressed in all HT thyrocytesamples (p , 0.001 vs. GD and NTG) while considerably re-duced Fas protein levels characterize GD thyrocytes. Only 6of 18 NTG thyroids exhibited trace amounts of Fas protein(Table 2).

We also analyzed and compared Fas expression in TILsfrom HT and GD glands (Table 2). As shown in Figure 2B,GD TILs expressed Fas protein at much higher level than HTlymphocytes (p , 0.001). Similar findings were obtained atthe RNA level (Fig. 1A). By immunohistochemistry, Fas ex-pression was mostly visible in the areas of disrupted folli-cles in HT glands (Fig. 3B). Fas protein was not detected incytokeratin-positive follicular cells in GD thyroids (Fig. 3D)but it was seen near largely infiltrated areas (Fig. 7A). Allspecimens were also analyzed by flow cytometry as shownin Fig. 4A. Thyrocytes purified from NTG thyroids showednegligible Fas surface expression, similar to thyrocytes puri-

APOPTOSIS PATTERN IN AITD 237

FIG. 1. A: Fas mRNA expression in thyrocytes (THY) and infiltrating lymphocytes (TILs) by quantitative S1 analysis. Thy-rocytes were immunodepleted from CD451 cells as described in Methods. TILs were isolated and purified through deple-tion of adherent cells and examined by flow cytometry (. 99% as CD31 cells). RNA (15 mg) was analyzed from Graves’disease (GD) thyrocytes (n 5 5, lanes 2–4 and 6–7; from #14, #18, #19, #23, #25 human glands; see Table 1), GD TILs (n 51, lane 5, from #23 human gland), Hashimoto’s thyroiditis (HT) thyrocytes (n 5 4, lanes 8–11, from #2, #4, #6, #7 humanglands) and TILs (n 5 1, lane 12, from #2 human gland). No S1 nuclease (undigested probe) was added to the radioactiveprobe and a 50-fold lower amount of probe was loaded (lane 1). Positive (5 mg HeLa RNA) and negative (100 mg of Es-cherichia coli tRNA) controls are shown in lanes 13 and 14, respectively. B: FasL mRNA expression in thyrocytes (THY). Aquantitative S1 analysis of FasL mRNA expression in hematopoietic cell-depleted thyrocytes is shown. Data from GD thy-rocytes (n 5 5, lanes 2–6, from #12, #14, #20, #23, and #24 human glands; see Table 1), HT thyrocytes (n 5 1, lane 7, from#2 human gland) and nontoxic goiter (NTG) thyrocytes (n 5 1, lane 8, from #46 human gland) are shown as representativeof three independent experiments. RNA extracted from PMA-activated Jurkat cells was included as positive control (lane9), while a negative control is shown in lane 10. No nuclease S1 (undigested probe) was added to the radioactive probe anda 50-fold lower amount of probe was loaded (lane 1).

A

B

GIORDANO ET AL.238

TABLE 2. MEAN VALUES 6 SD OF FAS AND FASL EXPRESSION IN THE DISEASE SPECIMENS

AVAILABLE FOR EACH ANALYTICAL METHOD

Flow cytometry S1 analysis Western blot RT-PCR(%) (Fas/actin ratio) (D.U.) (D.U.) (Fas/actin ratio) D.U.)

Cases Fas1 THY Fas1 TIL Fas1 THY Fas1 TIL Fas1 THY Fas1 TIL Fas1 THY Fas1 TIL

HT 60 6 8.4* 95 6 12* 0.66 6 0.05* 0.60 6 0.02* 8550 6 176* 3630 6 612* 0.99 6 0.2* NA(n 5 9) (n 5 6) (n 5 5) (n 5 2) (n 5 8) (n 5 8) (n 5 8)GD 8.5 6 2.1* 47 6 7.6* 0.29 6 0.03* 0.80 6 0.02* 1220 6 714* 9350 6 412* 0.55 6 0.3* NA(n 5 12) (n 5 8) (n 5 7) (n 5 3) (n 5 13) (n 5 8) (n 5 12)NTG 7.0 6 2.8* NA 0.14 6 0.02* NA 800 6 523* NA 0.42 6 0.2* NA(n 5 22) (n 5 10) (n 5 18) (n 5 16)

FasL1 THY FasL1 TIL FasL1 THY FasL1 TIL FasL1 THY FasL1 TIL FasL1 THY FasL1 TIL

HT 78 6 18* 20.8 6 8**** 0.60 6 0.05* NA 7759 6 235* 3132 6 445* 0.89 6 7.2* NA(n 5 8) (n 5 5) (n 5 7) (n 5 8) (n 5 7) (n 5 8)GD 16 6 5.7* 65 6 15* 0.41 6 0.06* NA 3053 6 465* 5777 6 428* 0.47 6 6.8* NA(n 5 14) (n 5 8) (n 5 8) (n 5 12) (n 5 8) (n 5 12)NTG 14 6 4.8* NA 0.45 6 0.02* NA 2875 6 343* NA 0.32 6 9.2* NA(n 5 20) (n 5 18) (n 5 18) (n 5 16)

*p , 0.001 vs. HT.Numbers in parentheses show the number of AITD glands examined.TIL, thyroid infiltrating lymphocytes; HT, Hashimoto’s thyroiditis; GD, Graves’ disease; NTG, nontoxic goiter.For the data expression, see Methods Section.

FIG. 2. A: Fas mRNA expression in thyrocytes. Reverse transcription-polymerase chain reaction (RT-PCR) analysis of FasmRNA expression in hematopoietic cell-depleted thyrocytes from nontoxic goiter (NTG) (n 5 2, lanes 1–2 from #28 and #42;see Table 1), Graves’ disease (GD) thyroids (n 5 4, lanes 3–6, from #15, #17, #20, #24 human glands) and Hashimoto’s thy-roiditis (HT) glands (n 5 4, lanes 7–10, from #1, #4, #8, #9, human glands). Lanes 11 and 12 show positive controls (HeLaand spleen). B: Fas and FasL protein expression by Western blot. Fas and FasL proteins were compared in hematopoieticcell-depleted thyrocytes from NTG (lane 1, from #31 human gland; see Table 1), GD (lane 3, from #18 human gland) andHT (lane 5, from #5 human gland) specimens, as well as in infiltrating lymphocytes from GD (lane 2, from #18 humangland) and HT (lane 4, from #5 human gland) glands.

A

B

fied from GD glands (Table 2, p 5 NS). In contrast, increasedproportions of thyrocytes purified from HT glands expressedFas (p , 0.001). Flow cytometry analysis of TILs (Fig. 4B)showed that most CD31 lymphocytes from GD glands(90%–99%) expressed Fas compared to approximately 50%of the lymphocytes from HT thyroids (40%–53.5%). Of note,TILs (CD31) were equally represented in both GD and HTsamples when they were examined by gating in flow cy-tometry (.90%), confirming that with our method of purifi-cation cell populations were similarly enriched. Fas was sim-ilarly expressed in CD41 and CD81 lymphocytes purifiedfrom both GD and HT thyroids (data not shown). Only mod-est proportions of B cells (CD191) expressed Fas (mean 515% 6 7%) in both GD and HT glands (data not shown). Col-lectively, the data presented here indicate that Fas expres-sion is reduced in purified thyrocytes from GD glands com-pared to HT glands. In contrast, Fas expression is increasedin TILs isolated from GD glands compared to those isolatedfrom HT thyroids.

FasL expression in thyrocytes and infiltrating lymphocytes

As shown in Figure 1B, FasL mRNA expression was higherin HT compared to NTG (p , 0.001 vs. HT) and GD thyro-cytes (p , 0.001 vs. HT; p 5 NS vs. NTG). Similar resultswere obtained by Western blot (Fig. 2B). FasL protein washighly expressed in all HT thyrocyte samples (p , 0.001 vs.GD and NTG) (Table 2). Although to a lesser extent, all GDand NTG thyroids examined exhibited FasL protein andmRNA (Table 2). As shown in Figure 4A, flow cytometryanalysis of TPO-positive GD thyrocytes demonstrated that10%–20% expressed FasL, similarly to NTG thyrocytes

(range, 13%–25%). A higher proportion of thyrocytes ex-pressed FasL in HT glands (63%–92%), as previously re-ported (9). The highest expression of FasL was found in TILs(CD31) from GD glands (p , 0.001) whereas FasL expressionwas much lower in HT lymphocytes (Fig. 4B). The highestFasL protein in GD TILs was confirmed by Western blot (Fig.2B, p , 0.001).

Bcl-2 expression in thyrocytes and lymphocytes

We also evaluated the expression of Bcl-2, a known anti-apoptotic molecule. As shown in Figure 5A, Western blotanalysis revealed that Bcl-2 was highly expressed in GD thy-rocytes (mean densitometric units (DU) 6 SD 5 9356 6 725)at levels that were much higher than in NTG specimens(6225 6 482). In contrast, Bcl-2 was almost absent in HT thy-rocytes (16296 212, p, 0.001 vs. GD and NTG). Remarkabledifferences in Bcl-2 expression were also seen in TILs (Fig. 5A).The Bcl-2 levels found in HT lymphocytes (7220 6 678) werehigher than the reduced levels observed in GD lymphocytes(3276 175, p, 0.001 vs. HT). By flow cytometry (Fig. 5B), HTlymphocytes expressed high Bcl-2 levels compared to GD lym-phocytes (mean values 70% 6 16% vs. 6.2% 6 7.8%, respec-tively, p, 0.001). Similar findings were obtained for Bcl-2mRNA levels by semiquantitative RT-PCR (data not shown).

Apoptosis in thyrocytes and infiltrating lymphocytes

Low levels of apoptosis were detected in PI-stained GDthyrocytes by flow cytometry (mean 8% 6 2.6%) (Fig. 6A).In contrast, high levels of thyrocyte apoptosis were found inHT glands (mean 56% 6 10%), as previously reported (Fig.

APOPTOSIS PATTERN IN AITD 239

FIG. 3. Fas expression in tissue sections. Immunostaining of thyroid cryostat sections from Hashimoto’s thyroiditis (HT)(A and B) and Graves’ disease (GD) (C and D) with negative control (A and C) and anti-Fas and anticytokeratin mono-clonal antibodies (B and D). B: Fas staining (arrow) in HT tissue [(from #1 thyroid gland: positivity (red)] appears to be ho-mogeneous in follicular structures; D: Fas staining (arrow) in GD tissue (from #24 human gland): follicular structures arepositive only for cytokeratin (brown). Magnification: 503.

6B) (9). However, the opposite was observed for TILs, withhigher proportions of lymphocytes being apoptotic in GD(mean 60.3% 6 18.4%) compared to HT glands (mean15.6%6 6.2%, p , 0.001) by flow cytometry analysis per-formed immediately after purification (Fig. 6C and 6D).TUNEL staining of thyroid sections confirmed that TILs butnot thyrocytes were undergoing apoptosis in GD glands (Fig.7). CD31 lymphocytes expressed Fas in markedly infiltratedareas (Fig. 7A) and were intensely apoptotic (Fig. 7B). Dou-ble-staining for apoptosis (TUNEL) and cytokeratin showedthat apoptosis was predominantly localized in cytokeratin-

negative cells and in nonfollicular areas (Fig. 7C and 7D),confirming that apoptosis in GD mainly affects TILs. Thus,apoptosis in GD is markedly different than in HT, where itdramatically affects thyrocytes and to some degree also in-volves TILs (9).

Discussion

Abundant literature shows that variation in Fas/FasL ex-pression levels may critically alter apoptosis, a mechanismof cell death that appears to play a critical role in several au-

GIORDANO ET AL.240

FIG. 4. A: Flow cytometry analysis of Fas and FasL expression in purified thyrocytes. Thyroid follicular cells from non-toxic goiter (NTG), Graves’ disease (GD), and Hashimoto’s thyroiditis (HT) glands were purified and depleted of contam-inating hematopoietic cells as described in Methods. Cells were labeled with control serum or control isotype-matched mon-oclonal antibody (mAb) (immunoglobulin G1 [IgG1]) (dotted lines), anti-thyroid peroxidase (TPO) mAb (left, solid lines),anti-Fas mAb (middle, solid lines), and FasL mAb (right, solid lines). Cells were gated on thyrocyte physical parametersand TPO positivity (upper panels). Numbers shown indicate the percentage of positive cells. One representative experi-ment is shown (from #33, #12, and #1 human glands, see Table 1). B: Flow cytometry analysis of Fas and FasL expressionin thyroid infiltrating lymphocytes (TILs). TILs were isolated from GD and HT specimens. Fas and FasL expression in elec-tronically gated CD31 cells is shown (left). Cells were labeled with isotype-matched antibodies (dotted lines) or anti-FasmAb and anti-FasL mAb (solid lines). Numbers shown indicate the percentage of positive cells. One representative exper-iment is shown (from #7 and #18 human glands; see Table 1). Numbers shown indicate the percentage of positive cells.

A

B

APOPTOSIS PATTERN IN AITD 241

A

B

FIG. 5. Bcl-2 expression. A: Western blot in hematopoietic cell-depleted thyrocytes from nontoxic goiter (NTG) (lane 1,from #48 human gland; see Table 1), Hashimoto’s thyroiditis (HT) (lane 3, from #4 human gland), and Graves’ disease (lane5, from #23 human gland) specimens. Lanes 2 and 4 show Bcl-2 protein expression in thyroid infiltrating lymphocytes (TILs)isolated from HT (from #4) and GD (from #23) glands, respectively. Data from one of five experiments are shown. B: Flowcytometry analysis of infiltrating lymphocytes from GD and HT glands. Bcl-2 expression was measured in permeabilizedand electronically gated CD31 cells. Cells were labelled with isotype-matched antibody (dotted lines), or with anti-Bcl-2monoclonal antibody (mAb) (solid lines). Numbers shown indicate the percentage of positive cells. One representative ex-periment is shown from #23 and #4 human glands.

FIG. 6. Detection of apoptosis in thyrocytes: Purified thyrocytes from Graves’ disease (GD) (A, from #15 human gland;see Table 1) and Hashimoto’s thyroiditis (HT) glands (B, from #8 human gland) were PI-stained and analyzed by flow cy-tometry for their DNA content in nuclei from purified thyrocytes cultured for 48 hours. Detection of apoptosis in infiltrat-ing lymphocytes: PI-stained, purified lymphocytes from GD (C, from #15 human gland) and HT glands (D, from #8 hu-man gland) were analyzed by flow cytometry for their DNA content of nuclei. Numbers shown indicate the proportion ofapoptotic cells. One representative experiment is shown.

toimmune diseases, including AITD (3,16). Thus, it is con-ceivable that differences in the expression of Fas and FasLmay contribute to the heterogeneous phenotypes observedin AITD and may play a role in determining the oppositeclinical syndromes (hypothyroidism and hyperthyroidism)observed in AITD. Although several studies have evaluatedFas/FasL expression and apoptosis in AITD (9,11,25–28), thepattern of expression in the two relevant cell populations,thyrocytes and TILs, has not been fully characterized. In thisstudy we report an extensive evaluation of Fas/FasL/Bcl-2expression and apoptosis in GD and HT thyroid specimens.The purification techniques utilized in this study allowed usto separately examine both thyrocytes and TILs. To ourknowledge, there has not yet been such an analysis of bothcellular compartments from both HT and GD glands in thesame study, and this is the first time that Fas/FasL/Bcl-2 ex-pression and apoptosis are simultaneously evaluated in TILsand thyrocytes purified from GD glands.

The comparison between GD and HT specimens revealedsignificant differences in the expression of Fas, FasL, Bcl-2,and in the occurrence of apoptosis. Thyrocytes from a rela-tively large series of GD patients were found to express sig-nificantly less Fas than HT thyrocytes, suggesting that an an-tiapoptotic milieu predominates in GD thyroids. Thisinterpretation is also supported by the higher levels of theantiapoptotic molecule Bcl-2 that we found expressed by GDthyrocytes compared to HT thyrocytes. Consistent with ourfindings, increased Bcl-2 expression has also been reportedin GD thyrocytes by Hiromatsu et al. (15). The patterns ofapoptosis observed in GD and HT thyrocytes also fit wellwith this interpretation, because significant levels of thyro-

cyte apoptosis were detected in HT but not GD glands (Figs.6 and 7). Thus, the constitutive expression of Bcl-2 and thelow expression of Fas observed in both GD and NTG thy-roids may help explain why GD thyrocytes are resistant toapoptosis and is consistent with a report suggesting that Fas-mediated apoptosis is normally blocked in thyrocytes (15).

In all the cases examined, GD thyrocytes expressed FasLat levels that were similar to those detected in normal thy-roids and NTG specimens. Similarly, comparable levels ofFasL were recently reported in GD thyrocytes by Hiromatsuet al. (15) and a recent report by Mitsiades et al. (17) suggeststhat FasL expression is upregulated in methimazole-treatedGD thyrocytes. The expression of FasL by thyrocytes mayhelp explaining the lack of tissue damage observed in GDthyroids, since FasL expression could enable thyrocytes toprotect themselves from autoreactive lymphocytes. This hy-pothesis is consistent with recent in vitro studies demon-strating the ability of FasL-positive GD thyrocytes to induceFas-mediated apoptosis of lymphocytes (17). Unlike HT lym-phocytes, GD lymphocytes expressed high levels of Fas andlow levels of Bcl-2, and predictably should be more prone toapoptosis. Indeed, significant proportions of GD lympho-cytes were found to be apoptotic in this study (Figs. 6 and7). Moreover, we have previously shown that FasL expressedon human thyrocytes is functional and capable of mediatingapoptosis of Fas-positive lymphoid cells in vitro (9), and thy-rocytes expressing FasL can reportedly induce apoptosis ofinfiltrating lymphocytes in experimental autoimmune thy-roiditis (29,30). Thus, our findings are consistent with the hy-pothesis that GD thyrocytes expressing FasL may induceapoptosis of Fas-positive TILs and lend support to the con-

GIORDANO ET AL.242

FIG. 7. Fas expression and Apoptosis in Graves’ disease (GD) thyroids. Tissue sections from GD specimens (from #18 hu-man gland; see Table 1) were stained for Fas and CD3 (A) or CD3 and apoptosis (TUNEL staining, B). A: CD3 and Fasstaining shows numerous double-stained lymphocytes (red and brown, and fusion amaranth color, see arrow). B: Almostall CD3-positive cells (see arrow) are also TUNEL-positive, i.e., apoptotic. C: Lack of colocalization of apoptosis (TUNELstaining, brown) with cytokeratin (red) in GD sections: cytokeratin-positive follicular cells are TUNEL-negative (see blackarrow) while cytokeratin-negative infiltrating cells are apoptotic (TUNEL-positive, see white arrow). D: Magnification 503.Arrow shows the TUNEL negativity of thyrocytes.

cept that FasL expression may confer immune privilege tothyrocytes, similar to the eye and testis (9,31–33). Moreover,this mechanism may help explain the less destructive formof autoimmunity observed in GD thyroids.

However, an imbalance in the expression of Fas and FasLcan turn a protective mechanism into a potentially damag-ing one, and we have recently produced evidence that sucha mechanism may play a key role in the development of HT(9). The simultaneous expression of both Fas and FasL onthe surface of neighboring thyrocytes appears to trigger frat-ricidal apoptosis of thyrocytes, and this may represent animportant mechanism of thyrocyte destruction in HT (33).Thus, the severe tissue damage associated with HT can beseen as the consequence of abnormal Fas/FasL expressionand reduced Bcl-2 levels in thyrocytes, both of which maybe induced by proapoptotic cytokines secreted by infiltrat-ing mononuclear cells. The observation that lymphocytes ex-press low FasL levels suggests that they may be unable tomediate significant tissue damage and that further supportsthe concept that thyrocytes may be the main mediators oftheir own demise.

An important implication of this and other related stud-ies is that we may have to modify our perception of the tar-get organ in organ-specific autoimmunity. The data dis-cussed here suggest that the phenotypic outcomes of AITDmay be significantly affected by the thyrocyte response tothe stimuli provided by the infiltrating lymphocytes (34).Similar observations have recently been made in type 1 di-abetes, another organ-specific autoimmune disorder inwhich pancreatic b-cells are destroyed by autoreactive lym-phocytes infiltrating the endocrine pancreas (6). Both a be-nign, nondestructive, and a malignant form of pancreatic au-toimmunity have been described (35), and a recent study hassuggested that the switch of the disease process from the be-nign to the malignant state does indeed require a responsefrom the target cells. In animal models of diabetes, such aresponse is mediated through the stimulation of the tumornecrosis factor receptor and, similar to what we describe herefor the thyroid, appears to involve changes in apoptosis ofpancreatic b-cells (6,36). In conclusion, this study illustratescritical differences in Fas/FasL expression and apoptosis thatmay influence key molecular mechanisms modulating thephenotypic outcome of AITD in man and contribute to thepathogenesis of thyroid autoimmunity.

Acknowledgments

We thank Drs. G. Modica, A. Lo Monte, and M. Romano,Department of Surgery, University of Palermo, Italy, for pro-viding the surgical specimens used in this study. This workwas supported by grants from MURST 98 (A.G.) and JDFI#197022 (C.G.).

References

1. De Groot LJ, Quintans J 1989 The causes of autoimmune thy-roid disease. Endocr Rev 10:537–562.

2. Weetman AP, McGregor AM 1994 Autoimmune thyroid dis-ease: Further developments in our understanding. EndocrRev 15:788–830.

3. Thompson CB 1995 Apoptosis in the pathogenesis and treat-ment of disease. Science 267:1456–1462.

4. Nagata S, Golstein P 1995 The Fas death factor. Science267:1449–1456.

5. Adams JM, Cory S 1998 The Bcl-2 protein family: Arbitersof cell survival. Science 281:1322–1325.

6. Stassi G, De Maria R, Trucco G, Rudert W, Testi R, GalluzzoA, Giordano C, Trucco M 1997 Nitric oxide primes pancre-atic b-cells for Fas-mediated destruction in insulin-depen-dent diabetes mellitus. J Exp Med 186:1193–1200.

7. French LE, Tschopp J 1997 Thyroiditis and hepatitis: Fas onthe road to disease. Nat Med 3:387–388.

8. Ohsako S, Elkon KB 1999 Apoptosis in the effector phase ofautoimmune diabetes, multiple sclerosis and thyroiditis.Cell Death Differ 6:13–21.

9. Giordano C, Stassi G, De Maria R, Todaro M, Richiusa P,Papoff G, Ruberti G, Bagnasco M, Testi R, Galluzzo A 1997Potential involvement of Fas and its ligand in the patho-genesis of Hashimoto’s thyroiditis. Science 275:960–963.

10. Papoff G, Stassi G, De Maria R, Giordano C, Galluzzo A,Bagnasco M, Ruberti G, Testi R 1998 Constitutive expressionof FasL in thyrocytes (technical comments). Science279:2015a, http://www.Science.com.

11. Smith D, Sieg S, Kaplan D 1998 Technical note: Aberrant de-tection of cell surface Fas ligand with anti-peptide antibod-ies. J Immunol 160:4159–4160.

12. Herr I, Posovsky C, Böhler T, Debatin KM 2000 mAb33 from transduction laboratories specifically binds to human CD95-L. Cell Death Differ 7:129–130.

13. Baker JR, Bretz JD 2000 Specificity questions concerning theclone 33 anti-Fas ligand antibody. Cell Death Differ 7:8–9.

14. Fiedler P, Eibel H 2000 Antibody mAb33 from TransductionLaboratories detects human CD95L in ELISA but not in im-munoblots. Cell Death Differ 7:126–128.

15. Hiromatsu Y, Hoshino T, Yagita H, Koga M, Sakisaka S,Honda J, Yang D, Kayagaki N, Okumura K, Nonaka K 1999Functional Fas ligand expression in thyrocytes from patientswith Graves’ disease. J Clin Endocrinol Metab 84:2896–2902.

16. Baker JR 1999 Dying (apoptosing?) for a consensus on theFas death pathway in the thyroid [editorial]. J Clin En-docrinol Metab 84:2593–2595.

17. Mitsiades N, Poulaki V, Tseleni-Balafouta S, Chrousos GP,Koutras DA 2000 Fas ligand expression in thyroid follicularcells from patients with thionamide-treated Graves’ disease.Thyroid 10:527–532.

18. Davies TF 2000 A new role for methimazole in autoimmunethyroid disease: Inducing T cell apoptosis. Thyroid10:525–526.

19. Giordano C, Stassi G, Todaro M, De Maria R, Richiusa P,Scorsone A, Giordano M, Galluzzo A 1995 Low Bcl-2 ex-pression and increased spontaneous apoptosis in T-lym-phocytes from newly diagnosed patients with insulin-de-pendent diabetes. Diabetologia 38:953–958.

20. Bagnasco M, Pesce GP, Caretto A, Paolieri F, Pronzato C, Villaggio BS, Giordano C, Betterle C, Canonica GW 1995 Follicular thyroid cells of autoimmune thyroiditismay coexpress ICAM-1 (CD54) and its natural ligand LFA-1 (CD11a/CD18). J Allergy Clin Immunol 95:1036–1043.

21. Nicoletti I, Migliorati G, Pagliacci MC, Grignani F, RiccardiC 1991 A rapid and simple method for measuring thymo-cyte apoptosis by propidium iodide staining and flow cy-tometry. J Immunol Methods 139:271–279.

22. Otsuki T, Sakaguki H, Tomokuni A, Aikoh T, Matsuki T,Kawakami Y, Kusaka M, Ueki H, Kita S, Ueki A 1998 Solu-ble Fas mRNA is dominantly expressed in cases with sili-cosis. Immunology 94:258–262.

APOPTOSIS PATTERN IN AITD 243

23. Laemmli UK 1970 Cleavage of structural protein during theassembly of the head of bacteriophage T4. Nature227:680–685.

24. Ausubel FM, Kingston RE, Moore DD, Seidman JG, SmithJA, Struhl KC 1987 In: Current Protocols in Molecular Biol-ogy. John Wiley & Sons, New York, pp. 16–17.

25. Arscott PL, Knapp J, Rymaszewski M, Bartron JL, Bretz JD,Thompson NW, Baker JR 1997 Fas (APO-1, CD95)-mediatedapoptosis in thyroid cells is regulated by a labile protein in-hibitor. Endocrinology 138:5019–5027.

26. Hammond LJ, Lowdell MW, Cerrano PG, Goode AW, Bot-tazzo GF, Mirakian R 1997 Analysis of apoptosis in relationto tissue destruction associated with Hashimoto’s autoim-mune thyroiditis. J Pathol 182:138–144.

27. Kawakami A, Eguchi K, Matsuoka N, Tsuboi M, Kawabe Y,Ishikawa N, Ito K, Nagataki S 1996 Thyroid-stimulating hor-mone inhibits Fas-antigen-mediated apoptosis of humanthyrocytes in vitro. Endocrinology 137:3163–3169.

28. Mitsiades N, Poulaki V, Kotoula V, Mastorakos G, Tseleni-Balafouta S, Koutras DA, Tsokos M 1998 Fas/Fas ligand up-regulation and BCL-2 down-regulation may be significantin the pathogenesis of Hashimoto’s thyroiditis. J Clin En-docrinol Metab 83:2199–2203.

29. Batteux F, Tourneur L, Trebeden H, Charreire J, Chioccia G.1999 Gene therapy of experimental autoimmune thyroiditisby in vivo administration of plasmid DNA coding for Fasligand. J Immunol 162:603–608.

30. Batteux F, Lores P, Bucchini D, Chioccia G 2000 Transgenicexpression of Fas ligand on thyroid follicular cells preventsautoimmune thyroiditis. J Immunol 164:1681–1688.

31. Griffith TS, Brunner T, Fleitcher SM, Green DR, FergusonTA 1995 Fas ligand-induced apoptosis as a mechanism ofimmune privilege. Science 270:1189–1192.

32. Bellgrau D, Gold D, Selawry H, Moore J, Franzusoff A, DukeRC 1995 A role for CD95 ligand in preventing graft rejec-tion. Nature 377:630–632.

33. Williams N 1997 Thyroid disease: A case of cell suicide? [edito-rial] Science 275:926.

34. Prabhakar BS, Fan JL, Seethramaiah GS 1997 Thyrotropin-receptor-mediated disease: A paradigm for receptor au-toimmunity. Immunol Today 18:437–442.

35. Dlits SM, Lafferty KJ 1999 Autoimmune diabetes: The in-volvement of benign and malignant autoimmunity. J Au-toimmun 12:229–232.

36. Pakala SV, Chivetta M, Kelly CB, Katz JD 1999 In autoim-mune diabetes the transition from benign to pernicious in-sulitis requires an islet cell response to tumor necrosis fac-tor alpha. J Exp Med 189:1053–1062.

Address reprint requests to:Carla Giordano, M.D.

Laboratory of ImmunologyEndocrinology Section

Institute of Clinica MedicaUniversity of PalermoPiazza delle Cliniche 2

90127 PalermoItaly

E-mail: cgiordan@unipa.it

GIORDANO ET AL.244