Overexpression of Metallothionein I/II: A New Featureof Thyroid Follicular Cells in Graves’ Disease

M. Ruiz-Riol,* M. J. Martínez-Arconada,* N. Alonso, B. Soldevila, D. Marchena,M. P. Armengol, A. Sanmartí, R. Pujol-Borrell, and E. M. Martínez-Caceres

Laboratory of Immunobiology for Research and Applications to Diagnosis (M.R.-R., M.J.M.-A., D.M.,M.P.A., R.P.-B., E.M.M.-C.), Banc de Sang i Teixits, and Division of Endocrinology and Nutrition (N.A.,B.S., A.S.), Hospital “Germans Trias i Pujol”, Institut d’Investigacio Germans Trias i Pujol; andDepartment of Cell Biology (M.P.A., E.M.M.-C.), Vall d’Hebron Institut de Recerca, Barcelona (R.P.-B.),Physiology and Immunology, Medical Faculty, Universitat Autonoma de Barcelona, 08916 Badalona,Spain

Context: One salient feature of autoimmune thyroid disease is the inappropriate expression of humanleukocyteantigen(HLA)classIImoleculesbythyroidfollicularcells.Metallothioneins(MT)aresmallproteinsinducedbytissue stress that cancontribute to restoringhomeostasisof tissue inflammationandhavebeenfound to be increased in a transcriptomic analysis of Graves’ disease (GD) glands.

Methodology: To assess the role of MT in the pathogenesis of GD, we analyzed MT-I and -IIexpression and distribution in GD-affected thyroid glands (n � 14) compared with other thyroiddiseases (n � 20) and normal thyroid glands (n � 5). Two-color indirect immunofluorescence andsemiquantitative morphometry were applied. The relationship between MT and HLA class II ex-pression was analyzed by their degree of colocalization in GD sections, and in vitro inductionkinetics and expression of these molecules on the HT93 thyroid cell line were compared by quan-titative RT-PCR and flow cytometry using interferon-� and zinc as stimuli.

Results: MT were clearly overexpressed in nine of 14 GD glands. MT expression distribution in GDwas almost reciprocal to that of HLA class II. In vitro analysis of MT and HLA class II demonstratedthat MT is induced more slowly and at a lower level than HLA. Moreover, the main MT inducer, zinc,reduces interferon-�-induced class II expression.

Conclusions: These findings show that MT and HLA class II play very different roles in the autoimmuneprocess by affecting the thyroid gland, thereby pointing to the possible role of MT as a marker of cellstress and homeostasis restoration in GD. (J Clin Endocrinol Metab 97: 446–454, 2012)

Autoimmune thyroid diseases (AITD) are chronic dis-orders stemming from a sustained immune response

to thyroid-specific antigens and are paradigms of organ-specific autoimmune diseases (1). Despite extensive effortsto understand the pathogenesis of AITD, a number of ma-jor questions, e.g. the factors that propagate the autoim-mune response over long periods of time, remain un-known. This point is crucial, because chronicity is a

distinctive feature of autoimmunity in humans that dif-fers from animal autoimmunity models. In fact, exper-imental autoimmune thyroiditis is not triggered by im-munization with the self-antigen [thyroglobulin (Tg)] initself, but instead, the administration as a mixture withadjuvant is required. Once the effect of the adjuvantceases, thyroiditis subsides (2). This finding demon-strates that breaking tolerance to a self-antigen is a nec-

ISSN Print 0021-972X ISSN Online 1945-7197Printed in U.S.A.Copyright © 2012 by The Endocrine Societydoi: 10.1210/jc.2011-1429 Received May 5, 2011. Accepted October 17, 2011.First Published Online November 16, 2011

* M.R.-R. and M.J.M.-A. contributed equally to this work.Abbreviations: AITD, Autoimmune thyroid diseases; CD40, cluster of differentiation-40;EAE, experimental autoimmune encephalomyelitis; GD, Graves’ disease; HLA, human leu-kocyte antigen; IFL, indirect immunofluorescence; MNG, multinodular goiter; moAb,monoclonal antibodies; MS, multiple sclerosis; MT, metallothioneins; TFC, thyroid follicularcells; Tg, thyroglobulin; TPO, thyroid peroxidase.

O R I G I N A L A R T I C L E

E n d o c r i n e R e s e a r c h

446 jcem.endojournals.org J Clin Endocrinol Metab, February 2012, 97(2):446–454

essary, but not sufficient, condition for eliciting an au-toimmune disorder (3).

It is plausible that, in clinical AITD, stimuli generatedin the thyroid gland itself maintain the autoimmune re-sponse. Thyroid follicular cells (TFC) within thyroidglands affected by autoimmunity have been shown to ex-press a variety of molecules not detected in healthy glands:human leukocyte antigen (HLA) class II, intercellular ad-hesion molecule-1, transporter in antigen processing, lowmolecular mass polypeptides (in the HLA region), clusterof differentiation-40 (CD40), and others (4–11). Thesefindings led to propose 25 yr ago that inappropriate oraberrant expression of major immunological molecules,such as HLA-DR, could be responsible for triggering andperpetuating the autoimmune response that leads to clin-ical AITD (5, 12).

In the last decade, it has become clear that innate im-munity circuits play a key role in preceding, inducing, anddirecting the adaptive immune response. Associated stim-uli might not always need to be authentic pathogen-associated molecular pattern-bearing molecules (PAMP).In fact, a number of endogenous molecules can exertthe same effect. These endogenous stimuli, collectivelydubbed DAMP (danger-associated molecular patterns),include small molecules such as uric acid and ADP (13).Based on this observation, our group has been seekingmolecules that may serve as markers for innate immuneactivation in Graves’ disease (GD) glands (14). Amongthe molecules detected in this transcriptomic analysis,we observed increased expression of metallothioneins(MT) in GD compared with normal thyroid glands.

MT are a family of low-molecular-mass (6 kDa) pro-teins with high cysteine content. Their thiol groups areable to bind heavy metallic ions such as Zn2�, Cu�, Cd2�,or Hg� and transfer Zn2� and Cu� to the catalytic sites ofnumerous enzymes. MT most likely protect against theeffects of toxic metals such as Cd2� or Hg� and have beenpostulated as Zn2� chaperones (15). Their high cysteinecontent confers antioxidant properties on MT that may alsobe important in regulating inflammation;however, their rolein the physiology of inflammation is not yet fully understood(16). MT have also been implicated in carcinogenesis andhave been studied in detail in that context (17, 18).

The MT family comprises four members (reviewed inRef. 16). Cytokines, free oxygen radicals, hormones,and various intracellular heavy metal ions (Zn2�, Cu�,and Cd2�) coordinately induce MT-I and -II by pro-cesses currently under study. In contrast, MT-III and -IVare constitutively expressed. This report on the role ofMT in AITD is limited to the inducible MT, MT-I, andMT-II. The MT-I and MT-II genes, located in chromo-some 16q13, include several closely related genes that

encode 11 MT-I isoforms and a single gene that encodesMT-II.

Previous studies reported elevated MT levels in acute andchronic multiple sclerosis (MS) lesions and in the central ner-vous system of mice with experimental autoimmune enceph-alomyelitis (EAE). In these contexts, MT appear to be in-duced by inflammatory cytokines (IL-1, IL-6, and TNF-�),but not necessarily by IFN-� (19–21). In the central ner-vous system, MT act as antioxidants, reduce inflamma-tion, and provide a general protective effect (22).

The thyroid is a complex endocrine organ rich in met-alloproteins suchas thyroidperoxidase (TPO), the enzymethat catalyzes the iodination of Tg in its tyrosine residues.Sudden changes in iodine intake can chemically stress thethyroid gland and trigger an autoimmune response (re-viewed in Ref. 23). We reasoned that MT may be usefulmarkers for tissue stress in AITD, because they are forother chronic inflammatory conditions, and that elevatedMT levels might indicate induction of danger-associatedmolecular patterns in general.

In an attempt to gain insight into the possible role ofMT in GD, we compared expression levels between TFCfrom thyroid glands affected by GD and those from glandsaffected by other thyroid diseases and normal glands; inparallel, we also evaluated HLA class II levels and lym-phocyte infiltration to determine whether MT levels cor-related with either. Furthermore, thyroid HT93 cells werecultured with different stimuli to compare MT and HLAclass II expression regulation, and the changes were as-sessed by indirect immunofluorescence (IFL) and quanti-tative RT-PCR (qRT-PCR).

Patients and Methods

PatientsThyroid patients were enrolled from the Endocrinology

Clinic and Nutrition Department of the Hospital UniversitariGermans Trias i Pujol and Hospital Universitari Vall d’Hebron,both affiliated with the Faculty of Medicine of the UniversitatAutonoma de Barcelona. All patients in this study fulfilled theusual clinical and laboratory criteria for GD (n � 14), Hashimo-to’s disease (HT, n � 5), benign thyroid nodules [multinodulargoiter (MNG), n � 12], and thyroid carcinoma (n � 3). Thyroidcarcinoma patients were characterized histopathologically hav-ing follicular carcinoma (n � 1) or papillary carcinoma (n � 2).Normal thyroid tissue was harvested from patients undergoingsurgery for laryngeal carcinomas (n � 5).

Total T4, free T4, T3, TSH, and thyroid antibody measure-ments supported the diagnosis. Thyroid ultrasound and/or scin-tiscans were available. TPO and Tg autoantibodies were mea-sured by a commercially available ELISA kit (Orgentec, Mainz,Germany), assuming no higher than 50 and 100 IU/ml as thenormal cutoff values, respectively. TSH receptor autoantibodies

J Clin Endocrinol Metab, February 2012, 97(2):446–454 jcem.endojournals.org 447

were measured with a competitive ELISA kit (DRG, Marburg,Germany) (normal values �1 IU/liter).

Patients with GD had been treated with carbimazole and pro-pranolol and exhibited normal thyroid function at the time ofsurgery. Patient selection was based on the following criteria:definitive diagnosis, availability of clinical and laboratory data,and existence of tissue and peripheral blood samples. Patientsgave their written informed consent to participate in the study,and the Ethics Committees of the two hospitals approved theproject. A summary of the relevant clinical and laboratory dataare provided in Table 1. Tissues were processed as previouslydescribed (24). Briefly, blocks (approximately 0.5 cm) weresnap-frozen in isopentane cooled in liquid nitrogen immediatelyafter collection. Tissues were stored at �70 C until processed.

HT93 cell culturesThe HT93 thyroid cell line had been generated by transfection

with the plasmid pX-8 containing the SV40 early region as pub-lished (25); for these experiments, HT93 cells were cultured inRPMI 10% fetal calf serum plus antibiotics. Cells were grown inslide chambers with different concentrations of IFN-� (0–1000U/ml; Prospec, Rehovot, Israel) and/or ZnCl2 (20–200 �M;Sigma-Aldrich, St. Louis, MO). Cultures with RPMI 10% fetalcalf serum served as controls. Fluorescence intensity of MT andHLA class II staining on monolayers was examined by UV mi-croscopy or flow cytometry as indicated below.

Immunofluorescence stainingIFL was used to stain 4-�m thyroid cryosections. Monoclonal

antibodies (moAb) against MT-I/MT-II, phenotypic markers forB and T lymphocytes (CD20 and CD3, respectively), and thepan-leukocyte marker CD45 were used as primary antibodies.Expression of HLA class II was detected by staining with a moAbagainst the HLA class II nonpolymorphic determinant. All re-

agent details are given in Table 2. Primary moAb were detectedwith Alexa-488 or tetraethylrhodamine isothiocyanate-conju-gated antimouse IgG subclass-specific antiserum from Molecu-lar Probes Invitrogen (Carlsbad, CA) or Dako (Carpinteria, CA),respectively, following standard protocols. Slides were mountedwith Fluoprep (Biomerieux SA, Marcy l’Etoile, France), and im-ages were acquired and analyzed using Openlab software (Im-provision Ltd., Coventry, UK).

Two independent observers (M.J.M.-A. and E.M.M.-C.) ex-amined sections of at least 5 � 5 mm in area using �20 and �40objectives on a Zeiss Axioplan microscope equipped with a UVlamp and connected to a Leica high-sensitivity digital photo-graphic camera. To assess the intensity of MT and HLA class IIstaining, a semiquantitative scale was applied, as described (24):�, undetectable staining; �, limited positive staining [only a fewTFC per follicle in less than 10% of the follicles (focal)], ��,10–50% positive follicles (multifocal); and ���, more than50% of all the follicles were positive (generalized). The followingsemiquantitative scale was applied to assess the intensity of thelymphocytic infiltration: �, no lymphocytes detected; �, scat-tered lymphocytes; ��, small but well-defined infiltration foci;���, medium to large infiltrates; and ����, loss of normaltissue architecture due to lymphocytic infiltration.

For staining of the HT93 cells, d-3 to -5 monolayer cultureswere fixed for 2 min in chilled acetone/methanol and stained for30 min by incubation with moAb to HLA class II followed by thecorresponding conjugated secondary antibody. Slides weremounted with Fluoprep and examined under the UV microscopeas above using an �63 immersion oil objective for detailed celldistribution observation.

Dilution-adjusted normal mouse serum was used for all con-trols samples instead of the primary moAb, and/or each layer wasomitted in turn in the double immunofluorescence stainingprotocols.

Flow cytometryHLA-DR and MT expression in the HT93 cell line was de-

termined using flow cytometer (FACSCanto, FACSDiva soft-ware; Becton Dickinson, San Jose, CA) and analyzed withFlowJo version 7 software (TreeStar, Inc., Ashland, OR). HLA-DR-fluorescein isothiocyanate (Becton Dickinson) moAb wasused. Indirect staining of MT was made with unconjugated MT(Dako), and the secondary antibody R-phycoerythrin-conju-gated antimouse IgG1 from Southern Biotechnology (Bir-mingham, AL). For the intracellular MT staining, a fixation-permeabilization kit (Dako) was included in the protocol. Cell

TABLE 1. Summary of patients’ clinical and demographic data

DiagnosisMean (range)

age (yr)Sex

(M/F)No. of

subjects

No. of serumsamplestested

TPO Ab(%)

Tg Ab(%)

TSH-R Ab(%)

GD 35 (16–60) 0/14 14 9 6/9 (66%) 5/9 (55%) 9/9 (100%)Multinodular goiter 45 (19–57) 2/10 12 6 0 0 0Hashimoto’s thyroiditis 39 (26–67) 0/5 5 3 2/3 (66%) 2/3 (66%) 0Thyroid carcinoma 56 (48–72) 0/3 3 1 1/3 (33%) 1/3 (33%) 0Normal thyroid tissue

(laryngealcarcinoma)

48 (17–65) 4/1 5 0 0 0 0

Ab, Antibody; F, female; M, male.

TABLE 2. Monoclonal antibodies used in the study

Specificity Clone Isotype SourceHLA class II EDU I IgG2b Dr. R. Vilellaa

MT I and II E9 IgG1K DakoCD20 L26 IgG2a DakoCD3 33-2A3 IgG2a Dr. R. VilellaCD45 72-503 IgG2a Dr. R. VilellaCytokeratin 8–18 5D3 IgG1 Novocastrab

a Immunology Department, Hospital Clínic, Barcelona, Spain.b Novocastra Labs, New Castle, UK.

448 Ruiz-Riol et al. Metallothionein Overexpression in Graves’ Disease J Clin Endocrinol Metab, February 2012, 97(2):446–454

viability was also assessed using AnexinneV-7AAD kit (BectonDickinson) in all culture conditions.

RNA extractionRNA was extracted from HT93 cell line cultures using Chom-

czynski’s technique (26). Briefly, cultured cells were homoge-

nized in lysis buffer consisting of guani-dinium thiocyanate (4 M; Sigma),N-lauroyl-sarcosine (0.5% wt/vol; Serva,Heidelberg, Germany), sodium citrate (pH4, 50 mM; Sigma), and 2-mercapto-ethanol(5%; Sigma). All samples were treated withAmbion DNA-free (Applied Biosystems, Fos-terCity,CA) toavoid interferencebycontam-inating genomic DNA. Each RNA samplewas quantified in a spectrophotometer(NanoDrop; Thermo Scientific, Wilmington,DE) and its integrity tested by electrophoresisin denaturalizing 1% agarose gel withethidium bromide staining.

Reverse transcription andquantitative real-time PCR

RNA (2 �g) was retro-transcribed witholigo-(dT15–18mer) (Pharmacia-Biotech,Uppsala, Sweden) and SuperScript-II (Phar-macia-Biotech). Real-time PCR to quantifythe expression of GAPDH (Hs99999905_m1), HLA-DRA (Hs00219578_m1), andMT1G (Hs01584215_g1) were performedin triplicate in a 96-well plate (LightCycler-480; Roche, Mannheim, Germany) using aTaqMan Universal PCR Master Mix andthe Assay-on-Demand (TaqMan MGBprobe; Applied Biosystems) protocol fol-lowing recommended procedures. For rela-tive quantification, the expression for eachgene was normalized to the housekeepinggene GAPDH following the 2���Ctmethod and expressed as arbitrary units.

Results

TFC in GD glands overexpress MTIFL staining revealed strong MT ex-

pression in the cytoplasm of the TFC innine of 14 GD thyroid glands. TFCwere easily identified by their shape andlocation around the colloid space of thethyroid follicles; however, to furtherconfirm the identity of the MT-positivecells, sections were double stained withantibodies to TPO, a highly specificmarker for TFC. TPO and MT clearlycolocalized within the cells, but theirintracellular distributions differed mark-edly. TPO accumulated in the colloid/apical pole, whereas MT staining ex-

hibited no such polarity (Fig. 1A, see detail). The MTstaining pattern was cytoplasmic and irregular in distri-bution. MT overexpression in the TFC did not affect theentire section; positive cells were confined to some areas(Fig. 1A). We examined and scored series of sections: two

FIG. 1. MT expression in the thyroid glands of GD patients. A, Double-stained, overlappedimages for MT (red) and TPO (green) showing TPO-positive cells that also overexpress MT.Note that in the upper part of the field, the TFC in two large thyroid follicles are completelynegative, but in the lower part, the TFC in the small follicles are clearly positive (�200). B,Double-stained, overlapped images for MT (red) and for HLA class II (green). In this area on aGD thyroid gland, there is clear overexpression of MT in the TFC in the absence ofinappropriate HLA class II expression or lymphocytic infiltration (�200). C, Representativeexamples of areas exhibiting strong inappropriate expression of HLA class II in the TFC andintense lymphocyte infiltration. In these areas, very few cells with features of TFC express MT(�200). D, Cytoplasmic staining of MT in TFC of GD (�400). E, MT and HLA class II stainingdo not coincide in the same cells (�400). F, Relative expression of MT, HLA class II, andlymphocyte infiltrates in 14 glands from Graves-Basedow patients. The heat map diagramfollows 14 different GD glands sorted by MT overexpression (scored from 0 to ���),lymphocyte infiltration (scored 0 to ����), and HLA class II expression (scored 0 to ���).This representation makes it easier to appreciate the reciprocal relationship between MToverexpression and HLA class II inappropriate expression. TB, Thyroid Barcelona collection.

J Clin Endocrinol Metab, February 2012, 97(2):446–454 jcem.endojournals.org 449

glands scored ���, four glands ��, and three �. Incontrast, MT was not detected in any of the other glands:HT (zero of five), MNG (zero of 12), or normal thyroidtissue (zero of five). Interestingly, in two of three thyroidcarcinoma glands, MT expression was detected in the neo-plastic TFC [one follicular carcinoma (scored �) and onepapillary carcinoma (scored ��)], both with cytoplasmicand perinuclear localization. These results are summa-rized in Fig. 1F (GD) and Table 3.

MT and HLA class II expression patterns do notcorrelate or bear a strict relationship withlymphocytic infiltration

In TFC, inappropriate expression of HLA class II orintercellular adhesion molecule-1 has previously been as-cribed to the local cytokine production by infiltrating lym-phocytes (6, 9, 24). Thus, we sought a correlation betweenMT overexpression and the proximity and intensity oflymphocytic infiltrates, with or without activated lym-phoid follicles or inappropriate HLA class II expression inthe TFC. Individual glands exhibited different distribu-tions for MT and HLA class II expression in the TFC,which did not bear an obvious interrelationship. More-over, MT distribution was not related to the lymphocyticinfiltration of the gland. On the one hand, and as shown inFig. 1, B and 1D, areas of MT overexpression were observedthatdidnotcontain infiltrating lymphocytes,whereas,ontheother hand, as shown in Fig. 1C, there were areas of inap-propriate HLA class II expression with varying degrees oflymphocytic infiltration that contained few MT-positiveTFC.Thepresenceof lymphocytic infiltrationcoexistedwithMT overexpression in some TFC (not shown).

When the intensities of MT overexpression, inappro-priate HLA class II expression, and lymphocytic infiltra-tion were compared among the different glands, it wasagain difficult to conclude whether or not a direct rela-tionship existed among them. The graphs in Fig. 1F rep-resent the scores for each GD gland; among the six glandswith an MT score of 2 or higher, none had an HLA classII score higher than 1; however, only three of the six glandswith an HLA class II score of 2 or higher overexpressedMT (all scored 1). Overall, glands whose TFC overex-

pressed MT were those in which inappropriate HLAclass II expression was less prominent, even when in-tense lymphocytic infiltration was present. Conversely,glands with TFC that prominently expressed aberrant HLAclass II exhibited little MT expression and, on average, lessinfiltration. MT-negative glands exhibited greater inappro-priate HLA class II expression that appeared to be indepen-dent of the degree of lymphocytic infiltration. In addition, norelationship was observed between MT overexpression andthe results of thyroid autoantibody titers or the clinical fea-tures of the GD patients (i.e. goiter size or presence of ex-ophthalmia; data not shown). In six of 12 MNG and one offive normal thyroid tissues, low levels of HLA class II ex-pression (score �) associated with a discrete inflammatoryinfiltration (score�)werealsoobserved.Conversely, theHTglands analyzed showed important lymphocytic infiltrates(score � ��) with four of five inappropriate HLA class IIexpression (score ��). In the two thyroid carcinomas withMToverexpression, lowHLAclass II expression (score�) inabsenceof infiltrateswasfound(Table3).ThedistributionofMT and HLA class II expression in the thyroid carcinomasfollowed the same pattern as in GD patients (not shown).

In vitro expression of MT and HLA class II in thethyroid cell line HT93

To gain more insight into the relationship between MTand HLA class II expression, cells of the thyroid cell lineHT93 were cultured for 0, 24, 48, or 72 h with IFN-� (0–1000 U/ml) in the presence/absence of ZnCl2 (160 �M) asinductor of MT expression. Analysis by qRT-PCR showed adose-dependent increase in transcription of HLA class II al-ready at 24 h (not shown), reaching its plateau at 48 h (Fig.2B). The addition of ZnCl2 to the culture medium did notalter the chronology of HLA class II induction. MT1G tran-scriptionshowedaslower induction in thepresenceofZnCl2athighdoses in combinationwith IFN-� (�500U/ml).Moreimportantly, HLA class II had already reached its maximalexpression at 100 U IFN-�, whereas MT required 1000 Uplus ZnCl2 (Fig. 2B). Therefore, HLA class II and MTshowed clear differences in both the time course and doseresponse of induction by Zn2� and IFN-�.

TABLE 3. Frequency of MT, inappropriate HLA class II expression, and lymphocytic infiltration in the glands fromdifferent groups of patients as assessed by IFL

GD (%) HT (%) MNG (%) Carcinoma (%) Normal thyroid tissue (%)Overexpression of MT in

TFC9/14 (64) 0/5 (0) 0/12 (0) 2/3 (66) 0/5 (0)

Inappropriate HLA class IIexpression in TFC

10/14 (71) 4/5 (80) 6/12 (50) 2/3 (66) 1/5 (20)

Lymphocytic infiltration 13/14 (93) 5/5 (100) 6/12 (50) 1/3 (33) 1/5 (20)

Samples of normal thyroid tissue came from patients undergoing surgery for laryngeal carcinoma.

450 Ruiz-Riol et al. Metallothionein Overexpression in Graves’ Disease J Clin Endocrinol Metab, February 2012, 97(2):446–454

These results were complemented by immunofluores-cence staining of HT93 cells, which showed that the max-imal induction of MT protein expression at 48 h coincidedwith a small decline in the level of the induced HLA classII expression (Fig. 2A). At 72 h, a decline was observed inthe expression of both molecules, being more marked forHLA class II.

The expression of HLA class II and MT in HT93 cellswas also assessed by flow cytometry after 48 h stimu-lation with different concentrations of ZnCl2 (20 –200�M) in the presence or absence of IFN-� (1000 U/ml). Asshown in Fig. 3, a moderate increase was found in MTexpression in a Zn2� dose-dependent manner. By con-trast, Zn2� clearly reduced the level of HLA class IIexpression induced by 1000 U IFN-�. In concentrationsof ZnCl2 higher than 160 �M, the viability of cultured

cells was less than 50%, and the re-sults were not considered (data notshown).

Discussion

MT I/II are cysteine-rich, heavy met-al-binding proteins that presumablyfunction in metal ion regulation anddetoxification of peripheral tissues.Accumulating evidence suggests thatMT may protect against the demyeli-nation and neuronal damage associ-ated with MS, a chronic autoimmunedisease of the central nervous system(22, 27). In contrast to the largeamount of data implicating MT inEAE/MS, little is known on their pos-sible roles in other autoimmune dis-eases. Herein, we report for the firsttime that MT are overexpressed in theparenchyma of thyroid glands fromGD patients. These results concurwith our microarray gene expressionanalyses, in which we used an Af-fymetrix U199 chip to analyze geneexpression in thyroid glands from GDpatients (14).

MT overexpression was clearly ob-served in the TFC of most GD glands.When MT levels, HLA class II expres-sion, and lymphocytic infiltration wereanalyzed in parallel, we were unable toestablish a correlation among the threeparameters; overexpression of MT andaberrant expression of HLA class II

demonstrated an almost reciprocal intensity distribution,either within the same glands or among different glands.The reciprocal expression of MT and HLA class II mole-cules could be explained in part by the differences in ki-netics detected by in vitro stimulation of the HT93 cell lineand analyzed by qRT-PCR, IFL, and flow cytometry.

MT overexpression was not detected in any of the fiveglands from HT patients; MT expression could be assessedeven in those sections in which the structure of the TFCwas preserved and lymphocyte infiltrates less prominent(not shown), suggesting that different pathogenic mech-anisms are involved in GD and HT that may account fordifferences in MT induction.

Reportedly, the infiltrating cells secrete proinflamma-tory cytokines such as IL-6 or TNF-� that induce MT

FIG. 2. HLA class II and MT expression in HT93 cells cultured under different stimuli ofIFN-� in presence/absence of ZnCl2. A, HT93 cells were cultured for 48 h with nostimulus, IFN-� (1000 U/ml), or IFN-� (1000 U/ml) plus ZnCl2 (800 nM). Note that HLAclass II expression (green) is decreased under ZnCl2 stimulus, the same one that inducesMT (red). Original magnification, �630. B, Relative expression level by qRT-PCR of HLA-DR and MT1G in HT93 cell line under stimulation with IFN-� and ZnCl2. Cells werecultured for 48 h with different doses of IFN-� (indicated in the x-axis) or ZnCl2 (160 �M)plus IFN-�. Results show the relative expression level of HLA-DR and MT from threeindependent experiments.

J Clin Endocrinol Metab, February 2012, 97(2):446–454 jcem.endojournals.org 451

expression (28). In contrast, IFN-� is not required for MTinduction (19), even though this cytokine is arguably themost potent inducer of inappropriate HLA class II expres-sion in TFC afflicted with AITD (29). During EAE, MTexpression is induced by proinflammatory cytokines, buttheir expression remains high during the remission phaseof the disease in relation to the secretion of immunoregu-

latory or antiinflammatory cytokines (21). These resultssuggest that in areas of the GD glands where MT expres-sion is high, there are immunoregulatory mechanisms at-tempting to resolve the inflammatory process.

In six MNG and one normal thyroid tissue, low levelsof HLA class II expression associated with a discrete in-flammatory infiltration was also observed. It is not un-

FIG. 3. HLA class II and MT expression in the thyroid cell line HT93 cultured with different concentrations of ZnCl2 with or without IFN-�. Theexpression of MTI/II and HLA-DR was assessed by flow cytometry at 48 h. Cultured cells had viabilities ranging from 60–90%. Resultscorresponding to one representative experiment of four are shown. FITC, Fluorescein isothiocyanate; SSC, side scatter; FSC, forward side scatter;PE, phycoerythrin.

452 Ruiz-Riol et al. Metallothionein Overexpression in Graves’ Disease J Clin Endocrinol Metab, February 2012, 97(2):446–454

common for necrotic foci to be present in local nodularthyroid lesions (24). Necrosis may activate resident mac-rophages to produce IL-1, which would induce an increasein CXCL12 expression by TFC (30) and in turn attractIFN-producing lymphoid cells to the foci. Such a mecha-nism would explain the existence of inflammatory infil-trates and aberrant HLA class II expression in thyroiddiseases (31). In contrast, MT expression was negative inall MNG cases analyzed (n � 12), probably due to theabsence of the inducing stimuli. It could be argued that thedifferences in MT expression in MNG and GD could bedue to thyrotoxicosis, antithyroid medications, or both.However, all GD patients had been treated with carbima-zole and propranolol and had thyrotoxicosis, and the factthat only nine of 14 overexpressed MT in their thyroidglands renders this hypothesis less plausible.

Interestingly, in two of the three primary carcinomacases included as additional controls, MT overexpressionwas detected in the TFC with cytoplasmic and perinuclearlocalization. Other authors have reported this finding inthyroid carcinomas (32, 33), as well as in other carcino-mas, where it was deemed a sign of malignancy. The neo-plastic factors involved in MT induction are not well de-fined (34, 35).

The reciprocal expression of MT and HLA class II in theglands of GD patients suggests that MT may in fact neg-atively regulate HLA class II. Because HLA class II ex-pression by TFC is involved in (auto)antigen presentationand probably contributes to perpetuation of the autoim-mune response (12, 36), it could be postulated that MTinduction in GD might actually be involved in resolvingthe inflammatory process, as has been proposed for otherautoimmune conditions (22).

In summary, herein we report the new observation thatMT are overexpressed in TFC of thyroid glands from pa-tients with GD. MT overexpression had a different inter-and intragland distribution from that of inappropriateHLA class II expression or lymphocytic infiltration, thusindicating different mechanisms of induction. We proposethat MT overexpression could reflect activation of a repairmechanism and should undergo further analysis alongwith other mechanisms of tissue repair underlying the au-toimmune process in GD glands.

Acknowledgments

We express our thanks to the Surgical and Pathology Depart-ments of Germans Trias i Pujol and Vall d’Hebron UniversityHospitals who provided us with the thyroid tissue samples. WearealsoverygratefultoDrs.JorgeCarrilloandMarcoA.Fernandez-Sanmartín, head of the flow cytometry facility, for their continuoushelp and suggestions.

Address all correspondence and requests for reprints to:Ricardo Pujol Borrell, Immunocat Program, Serveis Clinics Cen-trals, Hospital Universitari Vall d’Hebron, Passeig de la Valld’Hebron, 119-129, 08035 Barcelona, Barcelona, Spain. E-mail:ricardo.pujol@uab.es.

This work was supported in part by Grants PI031601 andPI051510 from the Fondo de Investigacion Sanitaria, Instituto deSalud Carlos III of the Spanish Ministry of Health. Research inthe Laboratory of Immunobiology for Research and Applica-tions to Diagnosis is supported in part by the Grups Consolidatsde Recerca de la Generalitat de Catalunya “Grup de Qualitat”2005SGR00139, AGAUR.

Disclosure Summary: The authors have nothing to disclose.

References

1. Weetmann AP 2007 The genetics of autoimmune thyroid diseases In:Wiersinga WM, Drexhage HA, Weetman AP, Butz S, eds. The thy-roid and autoimmunity. Stuttgart, Germany: Georg Thieme VerlagKG; 54–59

2. Roitt IM, Doniach D 1960 Thyroid auto-immunity. Br Med Bull16:152–158

3. Matzinger P 2002 The danger model: a renewed sense of self. Science296:301–305

4. Hanafusa T, Pujol-Borrell R, Chiovato L, Russell RC, Doniach D,Bottazzo GF 1983 Aberrant expression of HLA-DR antigen on thy-rocytes in Graves’ disease: relevance for autoimmunity. Lancet2:1111–1115

5. Bottazzo GF, Pujol-Borrell R, Hanafusa T, Feldmann M 1983 Roleof aberrant HLA-DR expression and antigen presentation in induc-tion of endocrine autoimmunity. Lancet 2:1115–1119

6. Tolosa E, Roura C, Catalfamo M, Martí M, Lucas-Martín A,Sanmartí A, Salinas I, Obiols G, Foz-Sala M, Pujol-Borrell R 1992Expression of intercellular adhesion molecule-1 in thyroid follicularcells in autoimmune, non-autoimmune and neoplastic diseases of thethyroid gland: discordance with HLA. J Autoimmun 5:107–118

7. Bagnasco M, Pesce GP, Caretto A, Paolieri F, Pronzato C, VillaggioB, Giordano C, Betterle C, Canonica GW 1995 Follicular thyroidcells of autoimmune thyroiditis may coexpress ICAM-1 (CD54) andits natural ligand LFA-1 (CD11a/CD18). J Allergy Clin Immunol95:1036–1043

8. Watson PF, Pickerill AP, Davies R, Weetman AP 1995 Semi-quan-titative analysis of interleukin-1�, interleukin-6 and interleukin-8mRNA expression by human thyrocytes. J Mol Endocrinol 15:11–21

9. Sospedra M, Obiols G, Babi LF, Tolosa E, Vargas F, Roura-Mir C,Lucas-Martin A, Ercilla G, Pujol-Borrell R 1995 Hyperinducibilityof HLA class-II expression of thyroid follicular cells from Graves’disease. A primary defect? J Immunol 154:4213–4222

10. Catalfamo M, Serradell L, Roura-Mir C, Kolkowski E, Sospedra M,Vives-Pi M, Vargas-Nieto F, Pujol-Borrell R, Jaraquemada D 1999HLA DM and invariant chain are expressed by thyroid follicularcells, enabling the expression of compact DR molecules. Int Immu-nol 11:269–277

11. Sospedra M, Tolosa E, Armengol P, Ashhab Y, Urlinger S, Lucas-Martin A, Foz-Sala M, Jaraquemada D, Pujol-Borrell R 1997 Hy-perexpression of transporter in antigen processing-1 (TAP-1) in thy-roid glands affected by autoimmunity: a contributing factor to thebreach of tolerance to thyroid antigens? Clin Exp Immunol 109:98–106

12. Pujo-Borrell R, Hanafusa T, Chiovato L, Bottazzo GF 1983 Lectin-induced expression of DR antigen on human cultured follicular thy-roid cells. Nature 304:71–73

13. Bianchi ME 2007 DAMPs, PAMPs and alarmins: All we need toknow about danger. J Leukoc Biol 81:1–5

J Clin Endocrinol Metab, February 2012, 97(2):446–454 jcem.endojournals.org 453

14. Ruiz-Riol M, Barnils Mdel P, Colobran Oriol R, Pla AS, BorrasSerres FE, Lucas-Martin A, Martínez Caceres EM, Pujol-Borrell R2011 Analysis of the cumulative changes in Graves’ disease thyroidglands points to IFN signature, plasmacytoid DCs and alternativelyactivated macrophages as chronicity determining factors. J Auto-immun 36:189–200

15. Vasak M 2005 Advances in metallothionein structure and functions.J Trace Elem Med Biol 19:13–17

16. Coyle P, Philcox JC, Carey LC, Rofe AM 2002 Metallothionein: themultipurpose protein. Cell Mol Life Sci 59:627–647

17. Theocharis SE, Margeli AP, Klijanienko JT, Kouraklis GP 2004Metallothionein expression in human neoplasia. Histopathology45:103–118

18. Dziegiel P 2004 Expression of metallothioneins in tumor cells. PolJ Pathol 55:3–12

19. Espejo C, Carrasco J, Hidalgo J, Penkowa M, Garcia A, Saez-TorresI, Martínez-Caceres EM 2001 Differential expression of metallo-thioneins in the CNS of mice with experimental autoimmune en-cephalomyelitis. Neuroscience 105:1055–1065

20. Lock C, Hermans G, Pedotti R, Brendolan A, Schadt E, Garren H,Langer-Gould A, Strober S, Cannella B, Allard J, Klonowski P, Aus-tin A, Lad N, Kaminski N, Galli SJ, Oksenberg JR, Raine CS, HellerR, Steinman L 2002 Gene microarray analysis of multiple sclerosislesions yields new targets validated in autoimmune encephalomy-elitis. Nat Med 8:500–508

21. Penkowa M, Espejo C, Ortega-Aznar A, Hidalgo J, Montalban X,Martínez Caceres EM 2003 Metallothionein expression in the cen-tral nervous system of multiple sclerosis patients. Cell Mol Life Sci60:1258–1266

22. Chung RS, Hidalgo J, West AK 2008 New insight into the molecularpathways of metallothionein-mediated neuroprotection and regen-eration. J Neurochem 104:14–20

23. Bartalena L Tanda ML, Piantanida E, Lai A, Compri E, LombardiV 2007 Environment and thyroid autoimmunity In: Wiersinga WM,Drexhage HA, Weetman AP, Butz S, eds. The thyroid and autoim-munity. Stuttgart, Germany: Georg Thieme Verlag KG; 60–73

24. Lucas-Martin A, Foz-Sala M, Todd I, Bottazzo GF, Pujol-Borrell R1988 Occurrence of thyrocyte HLA class-II expression in a widevariety of thyroid diseases: Relationship with lymphocytic infiltra-tion and thyroid autoantibodies. J Clin Endocrinol Metab 66:367–375

25. Belfiore A, Mauerhoff T, Pujol-Borrell R, Badenhoop K, BuscemaM, Mirakian R, Bottazzo GF 1991 De novo HLA class II and en-hanced HLA class I molecule expression in SV40 transfected humanthyroid epithelial cells. J Autoimmun 4:397–414

26. Chomczynski P, Sacchi N 1987 Single-step method of RNA isolationby acid guanidinium thiocyanate-phenol-chloroform extraction.Anal Biochem 162:156–159

27. Espejo C, Martínez-Caceres EM 2005 The role of methallothioneinsin experimental autoimmune encephalomyelitis and multiple scle-rosis. Ann NY Acad Sci 1051:88–96

28. Zheng RQ, Abney E, Chu CQ, Field M, Grubeck-Loebenstein B,Maini RN, Feldmann M 1991 Detection of interleukin-6 and inter-leukin-1 production in human thyroid epithelial cells by non-radio-active in situ hybridization and immunohistochemical methods.Clin Exp Immunol 83:314–319

29. Todd I, Pujol-Borrell R, Hammond LJ, Bottazzo GF, Feldmann M1985 Interferon-� induces HLA-DR expression by thyroid epithe-lium. Clin Exp Immunol 61:265–273

30. Armengol MP, Cardoso-Schmidt CB, Fernandez M, Ferrer X, Pujol-Borrell R, Juan M 2003 Chemokines determine local lymphoneo-genesis and a reduction of circulating CXCR4� T and CCR7 B andT lymphocytes in thyroid autoimmune diseases. J Immunol 170:6320–6328

31. Rasooly L, Burek CL, Rose NR 1996 Iodine-induced autoimmunethyroiditis in NOD.H-2h4 mice. Clin Immunol Immunopathol 81:287–292

32. Nartey N, Cherian MG, Banerjee D 1987 Immunohistochemicallocalization of metallothionein in human thyroid tumors. Am JPathol 129:177–182

33. Liu ZM, Chen GG, Shum CK, Vlantis AC, Cherian MG, Koropat-nick J, van Hasselt CA 2007 Induction of functional MT1 and MT2isoforms by calcium in anaplastic thyroid carcinoma cells. FEBS Lett581:2465–2472

34. Ostrakhovitch EA, Olsson PE, Jiang S, Cherian MG 2006 Interac-tion of metallothionein with tumor suppressor p53 protein. FEBSLett 580:1235–1238

35. Pastuszewski W, Dziegiel P, Krecicki T, Podhorska-Okolow M, Cie-sielska U, Gorzynska E, Zabel M 2007 Prognostic significance ofmetallothionein, p53 protein and ki-67 antigen expression in laryn-geal cancer. Anticancer Res 27:335–342

36. McLachlan SM, Rapoport B 2007 Thyroid peroxidase as an au-toantigen. Thyroid 17:939–948

454 Ruiz-Riol et al. Metallothionein Overexpression in Graves’ Disease J Clin Endocrinol Metab, February 2012, 97(2):446–454