changes in integrin and e-cadherin expression in neoplastic

Changes in Integrin andE-Cadherin Expression inNeoplastic Versus NormalThyroid Tissue

Guido Serini, Livio Trusolino,Enrico Saggiorato, OttavioCremona, Marco De Rossi,Alberto Angeli, Fabio Orlandi,Pier Carlo Marchisio*

Background: The functional organiza-tion of polarized epithelia dependsmostly on adhesion molecules belong-ing to the integrin and cadherinfamilies. These molecules either recog-nize basement membrane components,such as laminins, or form intercellularjunctions via homotypic interactions.Such tissue organization is often dis-rupted upon neoplastic transformation,and the resulting loss of functionalpolarization and cell cohesion might bea prerequisite for the invasive andmetastatic behavior of carcinomas.Purpose: We studied modifications ofthyroid adhesive mechanisms atvarious stages of neoplastic progressionin terms of adhesion molecule expres-sion, topography, and functionalregulation by tyrosine kinases. Startingfrom this working hypothesis, wesought to identify one or more biologi-cal markers that would be suggestive ofmalignant transformation and poorerprognosis and that could be developedas a reliable indicator(s) in early diag-nostic steps. Methods: The study wascarried out on both surgical samplesand the corresponding fine-needleaspiration biopsy smears (numbers ofspecimens collected: 19 adenomas,seven follicular carcinomas, 13 papil-lary carcinomas, and 39 normal tis-sues). Immunohistochemistry of tissuesections and smears and immuno-precipitation and western blot analysisof protein extracts were done with abattery of monoclonal and polyclonalantibodies. Northern blotting was per-formed on RNA extracts from frozentissue samples and use of an integrinsubunit P4 complementary DNA probe.

Results: Our findings can be sum-marized as follows: 1) In normalthyroid cells, the cooperative role of in-tegrin a ^ and laminin 5/kalinin inhemidesmosome-mediated adhesion ismissing, and recognition of the basallamina occurs via integrin a3pi andlaminin 1 and/or 2 (this pattern beingmaintained in adenomas but altered incarcinomas regardless of their his-totype or differentiation grade); 2) onlyin carcinomas with clinical and/or his-tologic aggressiveness do neoexpressionof integrin subunit (34 and loss oflaminin 2/merosin occur, indicating denovo assembly of integrin a ^ ; 3)pericellular redistribution and cyto-skeletal disconnection of the E-cad-herin-catenin complex occur; and 4)basal E-cadherin tyrosine phosphoryla-tion decreases in carcinomas ascompared with that in normal andadenomatous tissues. Conclusions: Themalignant progression of thyroidtumors involves marked rearrange-ments of cell-basement membrane andcell-cell adhesion molecules and chan-ges in their cytoskeleton linkage. Theserearrangements are also easily andreproducibly detected on fine-needleaspiration biopsy smears. Implications:Immunodetection of adhesion mole-cules in sections and/or fine-needlesmears may complement the toolbox ofthyroid surgical pathologists; it mayexpand the possibilities of achieving acorrect early diagnosis of thyroidtumors and of gaining some prognosticinformation on thyroid tumors. [J NatlCancer Inst 1996; 88:442-9]

The induction and maintenance of apolarized and differentiated epithelialphenotype depend on expression, mem-brane sorting, and function of surface ad-hesion molecules (/) that provide do-main-dependent mechanical and chemi-cal information and are responsible fordefining and maintaining cell topologyand the functional polarity typical ofepithelia. Such tissue organization isoften disrupted upon neoplastic transfor-mation, and the resulting loss of thepolarized distribution and/or functionalproperties of adhesion molecules mightbe a prerequisite for the invasive andmetastatic behavior of carcinomas (2,J).

Epithelial cells adhere to the basementmembrane zone mostly via integrins (4-6); these integrins are transmembrane ot/pheterodimeric proteins that mediaterecognition of the basal lamina and at-tachment to laminins (represented inepithelial basal lamina by laminin 1/EHS1

[aipiyl] , laminin 2/merosin [a2piYl],and laminin 5/kalinin [a3p3y2]) (7). Themajor integrin restricted to the basaldomain of the basal layer in squamousepithelia (8-11) and in most columnarepithelia (J2J3) is 0^4, which usuallycodistributes with hemidesmosomes (14-16) and laminin 5/kalinin.

Epithelial cell-cell association is pri-marily mediated by E-cadherin (17-19), astructural component of zonula adherensinvolved in forming homotypic bonds andin linking the microfilament network(20,21) via a chain of cytoskeletal pro-teins termed "catenins" (22-24).

Differentiated thyroid carcinomas orig-inating from follicular cells are classifiedas a papillary type or a follicular type.Their poorly differentiated variants en-compass the tall-cell carcinoma for papil-lary types (25) and insular (26) andHiirthle cell (27) carcinomas for folliculartypes.

So far, no conclusive data have beenprovided about a simple and reliablemethod for correct, early diagnosis ofthyroid tumors. Ideally, the preoperativedifferential diagnosis between benign andmalignant thyroid neoplasms, performedon fine-needle aspiration biopsy smears,could be improved by two potential find-ings: 1) changes in integrin expressionand topography and 2) aberrant synthesis,assembly, and/or adhesive status of the E-

* Affiliations of authors: G. Serini, L. Trusolino,P. C. Marchisio, Department of Biomedical Scien-ces and Human Oncology, University of Torino,Italy, and Department of Biological and Technologi-cal Research, San Raffaele Scientific Institute,Milano. Italy: E. Saggiorato, A. Angeli, F. Orlandi,Department of Clinical and Biological Sciences,University of Torino; O. Cremona, Department ofBiomedical Sciences and Human Oncology, Univer-sity of Torino, and Department of Medical Sciences,University of Torino, Novara Branch; M. De Rossi,Department of Biological and Technological Re-search, San Raffaele Scientific Institute.

Correspondence to: Pier Carlo Marchisio, M.D.,Ph.D., Department of Biological and TechnologicalResearch, San Raffaele Scientific Institute, Via Ol-gettina 58, 20132 Milan, Italy.

See "Notes" section following "References."

442 REPORTS Journal of the National Cancer Institute, Vol. 88, No. 7, April 3, 1996

Downloaded from https://academic.oup.com/jnci/article-abstract/88/7/442/1011862by gueston 17 February 2018

cadherin-catenin complex. Moreover,these modifications might provide a bet-ter prognostic definition of thyroidtumors. Accordingly, the aim of thisstudy was to understand thyroid adhesivemechanisms during neoplastic progres-sion in terms of adhesion molecule ex-pression and functional regulation bytyrosine kinases. Starting from this work-ing hypothesis, we sought to identify oneor more biological markers suggestive ofmalignant transformation and poorerprognosis that would be suitable fordevelopment as reliable indicators inearly diagnostic steps. The main mor-phologic-functional changes observed inthyroid cancers are pericellular redistri-bution of the integrin a3f3i complex,integrin a$4 neoexpression, and E-cadherin disorientation as well as a reduc-tion in its tyrosine phoshorylation level.

Materials and Methods

Surgical Specimens

After clinical staging and fine-needle aspirationbiopsies, fresh neoplastic thyroid tissue was surgi-cally removed from patients previously treated withLugol's iodine solution. All patients underwentthyroidectomy at San Luigi Gonzaga Hospital, Or-bassano (Torino, Italy). The tumor tissue was eitherdirectly embedded in OCT 4583 (Miles Scientific,

Naperville, IL) and snap-frozen for further im-munohistochemical studies or directly frozen in liq-uid nitrogen for protein and RNA extraction. Aportion of this tissue was used for routine his-topathology. Tumor-free tissue was excised fromeach patient and processed in parallel. Frozen sec-tions (5 UJTI) were serially cut in a Reichert-Jungcryomicrotome, transferred onto poly-L-lysine-coated slides, air-dried, and stored overnight at roomtemperature.

Histopathologic Classification

In parallel with immunohistochemistry, serialsections of each specimen were stained withhematoxylin-eosin to determine the histotype andsubhistotype. We examined 19 surgical samples offollicular adenomas, seven of follicular carcinomas,and 13 of papillary carcinomas, hi addition, we ex-amined a total of 39 normal tissue specimens. Of theseven follicular carcinomas examined, four wereclassified as poorly differentiated malignant neo-plasms (HUrthle cell subtypes) and three as well-dif-ferentiated carcinomas; among the 13 papillarycarcinomas, we observed five poorly differentiatedvariants (four tall-cell carcinomas and one sclerotictumor), two follicular variants, five well-differen-tiated cancers, and one dedifferentiated carcinomawith wide anaplastic fields. The histologic featuresof each case, together with sex and age information,TNM staging (28), follow-up, and additional dataare summarized in Table 1.

Fine-Needle Aspiration Biopsies

The fine-needle aspiration biopsies were per-formed with a 22-gauge x 1.5-inch needle attachedto a 30-mL plastic syringe (29). After aspiration, the

small fluid specimen was expelled from the needleand smeared onto a polylysine-coated slide. Thesmears were air-dried for 2 hours, pre-fixed in ab-solute methanol at -20 "C, and stored at -80 'C. Weexamined aspiration biopsy smears from 10 ade-nomas, five follicular carcinomas, and five papillarycarcinomas. Written informed consent was obtainedfrom each subject with the approval of the San LuigiGonzaga Hospital review board.

Antibodies

The primary monoclonal antibodies (MAbs) usedin this study (with the investigators who providedthem) were as follows: MAR4 to integrin subunit P,and MAR6 (30) to integrin subunit a^ (from S.M6nard, Istituto Nazionale Tumori, Milano, Italy),GOH3 to integrin subunit a$ (from A. Sonnenberg,The Netherlands Cancer Institute, Amsterdam), F2to integrin subunit ct3 (from L. Zardi, Istituto Scien-tifico per lo Studio e la Cura dei Tumori, Genova,Italy), AA3 and S3-41 to integrin subunit P4 (31)(from V. Quaranta, Scripps Research Institute, LaJolla, CA), IA9 to integrin subunit P5 (from M.Hemler and R. Pasquahni, Dana-Farber Cancer In-stitute, Boston, MA), GB3 to laminin BM600/nicein(32) (from P. Verrando, Laboratoire.de RecherchesDermatologiques, Facultd de Meclecine, Nice,France), and 5H2 to laminin 2/merosin (33) (from E.Engvall, Wenrter Gren Institute, Stockholm, Swe-den). Other MAbs were commercially obtained:SAM-1 to integrin subunit Oj and Gi9 to integrinsubunit <x2 (Immunotech, Marseille, France), LAM-89 to laminin 1 and PT-66 to phosphotyrosine-con-taining proteins (Sigma Chemical Co., St. Louis,MO), HECD-1 to human E-cadherin (Takara ShuzoCo., Kyoto, Japan), and MAbs to ot-catenin and P-

Table 1. Clinicopathologic features of patients with thyroid carcinomas

TNMt

Case*

FC1FC2FC3FC4FC5FC6PC7PC8PC9PC10PCUPC12PC13PC14PC15PCI6PC17PC18FC19PC2O

Age, y/sex

39/female75/female27/female51/female31/male33/female46/female30/female45/female41/female70/male27/female31/female16/female61/female60/male37/female73/female34/female65/female

Histotypet

DFCPDCDFCDFCPDCPDCDPCPDCPDCDPCDPCPDCPDCDPCPDCDPCDPCDDCPDCDPC

Subhistotype

Well-differentiated carcinomaHUrthle cell carcinomaWell-differentiated carcinomaWell-differentiated carcinomaHUrthle cell carcinomaHUrthle cell carcinomaFollicular variantTall-cell carcinomaTall-cell carcinomaFollicular variantWell-differentiated carcinomaSclerotic variantTall-cell carcinomaWell-differentiated carcinomaTall-cell carcinomaWell-differentiated carcinomaWell-differentiated carcinomaWide anaplastic fieldsHUrthle cell carcinomaWell-differentiated carcinoma

At presentation

T2aN0M0T4aNlbM0T3aN0M0T2aN0M0T2aN0M0TlaNOMOT4bN0M0T2aNlaM0T4bNlbM0T4aNlbM0T2bN0M0

T4aNlbM0T4bN0M0T2aN0M0T2aN0M0T1N0M0T2aN0M0

T4aN0M+ (OTHflT2N0M0T4N1M0

At recurrence

M+ (PUL)

M+(OTH)II

Follow-up!

NED (32)AWD (32)NED (31)NED (30)NED (29)NED (15)NED (27)NED (26)AWD (26)NED (25)NED (24)NED (24)NED (23)NED (40)NED (21)NED (13)NED (5)DOD(5)NED (2)NED (3)

p4 expression

*FC = follicular carcinoma; PC = papillary carcinoma.tDFC = differentiated follicular carcinoma; PDC = poorly differentiated carcinoma; DPC = differentiated papillary carcinoma; DDC = dedifferentiated carcinoma.t According to Hermanek and Sobin (28). M+ = presence of metastases; PUL = pulmonary metastases; OTH = metastases in other organs.§NED = no evidence of disease; AWD = alive with disease; DOD = dead of disease. Months after surgical resection are indicated in parentheses.IIMediastinum.^Trachea.

Journal of the National Cancer Institute, Vol. 88, No. 7, April 3, 1996 REPORTS 443


catenin (Affiniti, Nottingham, U.K.). Rabbit poly-clonal antiserum R571O to integrin subunit f54 wasprovided by V. Quaranta; rabbit antiserum R1542 tointegrin subunit (3, was a gift of L. Languino (LaJolla Cancer Research Foundation, CA). Forimmunoperoxidase staining and immunofluores-cence, MAbs were used at a final immunoglobulinconcentration of 10-40 u.g/mL. Immunoprecipita-tions were performed with 4 u.g MAbs per sample.For immunoblotting, 2 Ug/mL MAbs or 10 Ug/mLpolyclonal antibodies were used. For control pur-poses, irrelevant antibodies were routinely used.

Indirect Immunoperoxidase Technique

Experiments were performed as previouslydescribed (34). Briefly, cryostat sections were fixedfor 10 minutes in a chloroform-acetone mixture(1:1) at 4 'C, air-dried, and incubated for 10 minutesin phosphate-buffered saline (PBS) supplementedwith 1 % serum from the same species as that for thesecondary antibody. Serial sections were overlaidwith 50 u.L of different antibodies in Tris-bufferedsaline (TBS) in 0.4% bovine serum albumin (BSA)and incubated at room temperature for 30 minutes ina moist chamber. After a thorough wash in PBS, thesections were incubated with the appropriatebiotinylated secondary antibody and processed forthe ABC method (avidin—biotin—peroxidase com-plex) using the Vectastain ABC Kit (VectorLaboratories, Inc., Burlingame, CA). After the sec-tions were washed three more times, 100 uL of sub-strate was added for 5-10 minutes and was preparedas follows: 5 mg of 3-amino-9-ethylcarbazole(Sigma Chemical Co.) was dissolved in I mL ofAVV-dimethylformamide (Merck, Darmstad, FederalRepublic of Germany) supplemented with 9 mL of100 nW sodium acetate (pH 5.2) and 100 uL of12% H2O2. All samples were counterstained withMayer's hemalum solution, mounted in Kaiser'sglycerol gelatin (Merck), and examined with a ZeissAxiophot photomicroscope (Zeiss, Jena, FederalRepublic of Germany) equipped with 16x and 63xplanapochromatic lenses.

Indirect Immunofluorescence Microscopy

Smears were fixed in a chloroform-acetone mix-ture (1:1) for 10 minutes at 4 "C and air-dried. Aftera 15-minute saturation with TBS-BSA (0.4% at37 *C), the primary antibodies (MAbs AA3 and S3-41 to P4 and MAR6 to o^) were layered onto slidesand incubated in a moist chamber for 30 minutes.After rinsing in TBS containing 0.4% BSA, theslides were incubated with the appropriate rhoda-mine-tagged secondary antibody (Dakopatts, Copen-hagen, Denmark) for 30 minutes at 37 "C.Coverslips were mounted in Mowiol 4-88 (HoechstAG, Frankfurt/Main, Federal Republic of Germany).

Routine observations were carried out in a ZeissAxiophot photomicroscope equipped for epifluores-cence. Fluorescence images were recorded onKodak T-Max 400 films exposed at 1000 ISO anddeveloped in T-Max Developer for 10 minutes at20 -C.

Detergent Solubilization

Surgical samples were directly snap-frozen inliquid nitrogen, pulverized in a B-Braun Mikro-Dis-membrator (B-Bran. Melsungen. Federal Republic

of Germany), and lysed in extraction buffer (TBS[pH 8], 1% Triton X-100, 0.5% Nonidet P-40, and 2nW CaCl2) containing a mixture of phosphatase andprotease inhibitors (2 mM sodium orthovanadate, 50mM sodium fluoride, 1 mM phenylmethyl sulfonylfluoride, 2 Hg/mL leupeptin, and 2 ug/mL pepstatinA) by several passages in a Dounce homogenizer(PBI International, Milan, Italy) on ice. Tissuelysates and insoluble material were collected andcentrifuged for 10 minutes at 20 000 rpm at 4 'C.Supematants were saved; in some experiments,detergent-insoluble pellets were resuspended in abuffer containing 50 mM Tris-HCl (pH 8.8) and 1%sodium dodecyl sulfate (SDS), sonicated, boiled for5 minutes, and recentrifuged. SDS extracts werediluted 10-fold in extraction buffer, and supematantswere adjusted to 0.1% SDS before immunopre-cipitation.

Immunoprecipitation

Experiments were earned out as previouslydescribed (35). Tissue extracts were adsorbed ontoprotein A-Sepharose CL-4B (Pharmacia LKBBiotechnology AB, Uppsala, Sweden) previouslyincubated with normal mouse serum (Sigma Chemi-cal Co.). Precleared lysates were incubated withMAb HECD-1 to E-cadherin, and immuno-complexes were collected by protein A-SepharoseCL-4B coupled with rabbit anti-mouse immuno-globulin G (Pierce Chemical Co., Rockford, IL).After seven washes with extraction buffer, the finalpellet was boiled in Laemmli buffer (36) in thepresence of 4% (3-mercaptoethanol, and proteinswere processed for SDS-polyacrylamide gel elec-trophoresis (PAGE) (8% polyacrylamide gels). Forquantitative recovery of E-cadherin-catenin com-plexes, protein concentrations were normalized(BCA Protein Assay Reagent Kit; Pierce ChemicalCo.), MAb HECD-1 was titrated in normal thyroidlysates, and saturating amounts of MAb were usedin each immunoprecipitation experiment.

Western Blot Analysis

Frozen surgical samples were pulverized using aB-Braun Mikro-Dismembrator (37). For semiquantita-tive recovery of epithelium-specific components fromthe crude extract, each sample was divided into 200-mg pieces and processed in parallel for western blotanalysis and for routine histology to evaluate the ratioof stromal contamination according to histotogic ex-amination. The pulverized tissues were solubilized inboiling Laemmli buffer and sonicated; 300 |ig ofpresumptive epithelial proteins was loaded onto eachlane. Alternatively, immunocomplexes, recoveredfrom immunoprecipitation experiments after elution inboiling Laemmli buffer, were directly loaded ontogel lanes. Materials were fractionated by SDS—PAGE (8% polyacrylamide gels) under nonreducingconditions, and proteins were electrophoreticallytransferred to nitrocellulose filters (Hybond; Amer-sham Life Science. Inc.. Arlington Heights, IL) andanalyzed as described previously (38). Filters wereprobed with rabbit antiserum R1542 to integrin sub-unit (3|, R5710 to integrin subunit f54, and MAbs toE-cadherin, a-catenin, pVcatenin, and phosphotyro-sine-containing proteins. Specific binding wasdetected by the Enhanced Chemiluminescence Sys-tem (Amersham Life Science, Inc.).

Northern Blot Analysis

Total RNA was isolated from pulverized, snap-frozen tissues by the acid guanidium method (39)using a Dounce homogenizer on ice, and northernblots were performed with 10 ug total RNA perlane. Ethidium bromide at a concentration of 0.2Ug/mL was added before electrophoresis to 1%agarose gels containing formaldehyde to verify theintegrity of the RNA by short-wavelength UVdetection and to monitor the equivalence of loadingbefore and after transfer to GeneScreen Plus filters(Du Pont NEN, Boston, MA) (i.e., the integrity andrelative amounts of ribosomal RNAs were assessed).For additional RNA quantitation, filters werestained with 0.04% methylene blue in 0.5 M sodiumacetate. The integnn subunit P4 complementaryDNA (cDNA) clone (1.5-kilobase [kb] £coRI frag-ment) was labeled with random priming (Mega-prime DNA labeling system; Amersham LifeScience, Inc.) and [3 P]deoxycytidine triphosphate(3000 Ci/mmol; Amersham Life Science, Inc.).Membranes were pretreated and hybridized in 50%formamide (Merck) and 10% dextran sulfate (SigmaChemical Co.) at 42 "C. Blots were washed twicewith 2x sodium chloride-sodium citrate (SSC) atroom temperature for 10 minutes, then twice with2x SSC-1 % SDS at 60 #C for 30 minutes, and final-ly twice with O.lx SSC at room temperature for 30minutes followed by exposure to autoradiographyfor 24 hours at -80 'C with intensifying screens.

Results

Expression and Regulation ofIntegrins and Laminins in Normal andNeoplastic Thyroid Tissues

In normal thyroid tissue, integrinchains a3 and p, were exclusively ex-posed at the basal domain of follicularthyroid cells (Fig. la, panel A), whereas<x2 and Oj could not be detected (notshown). The a$4 complex was found tobe absent in thyroid cells (Fig. le, panelA).

The basement membrane of normalthyroid tissue was composed of laminins1 and 2 but not laminin 5 (not shown).The parallel lack of expression of both in-tegrin Oep^ and laminin 5 at the basalaspect of thyroid cells strongly argues forthe absence of hemidesmosomes inthyroid follicular cells. In fact, transmis-sion electron microscopy as well as im-munostaining with human sera reactingwith hemidesmosome-specific bullouspemphigoid antigens revealed that thy-roid cells do not assemble hemidesmo-somes (not shown). The integrin subunitsp3, P5, and (X, did not show any obviousimmunoreactivity.

The above distribution pattern was to-tally retained in follicular adenomas (Fig.



Fig. 1. Adhesion molecule expression and topography in normal and neoplastic thyroid tissues. Panel A: hi normal (a) and adenomatous (b) thyroid tissues, the in-tegrin subunit 03 is exposed at the basal domain of thyroid cells, whereas it is aspecifically redistributed along the cell margins in follicular (c) and papillary (d) car-cinomas. The p4 integrin chain is not present in normal (e) and adenomatous (f) follicular cells, but it is strongly expressed in poorly differentiated and/or clinicallyaggressive follicular (g, case FCl) and papillary (h, case PC 13) carcinomas. The integrui subunit a* expression entirely matches that of p4 (not shown). Bar denotes16 um. Panel B: hi normal thyroid tissue (a) and in adenoma tissue (not shown), E-cadherin is selectively enriched at cell-cell contacts, hi malignant tumors (b), themolecule is conserved but pericellularly redistributed. The expression of P-catenin entirely matches that of E-cadherin (not shown). Bar denotes 16 |im and, for bothinsets, 4 |im.

lb and f, panel A) but was subverted inboth papillary and follicular carcinomas(Fig. lc, d, g, and h, panel A).

To support immunohistochemical dataand to check the molecular mass of theintegrin subunit Pi, we performed im-munoblot analysis on some normal tissueand tumor samples. A Pi polyclonal an-tiserum identified a 110-kd band presentat similar intensity levels (Fig. 2, panelA).

Surprisingly, we observed neoexpres-sion of the integrin 0^4 heterodimer in11 of 20 malignant tumors (Fig. lg and h,panel A; Table 1). More precisely, two ofthree well-differentiated follicular car-

cinomas and nine of 13 poorly differen-tiated and/or clinically aggressive cancerswere found to be p4 and 0$ positive. Ex-pression of the P4 chain in these specifictumors was confirmed both by im-munofluorescence in fine-needle aspira-tion biopsy smears (Fig. 2, panel B) andby western blot analysis (Fig. 2, panel C).Northern blot analysis revealed a specific5.5-kb P4 transcript with variable band in-tensity in those carcinomas that werefound to express p4 protein in immuno-histochemistry and western blot experi-ments (Fig. 2, panel D). Thus, the dataindicate that, in some thyroid cancers, atranscriptionally regulated expression of

P4 occurs, which leads to a ^ membraneexposure.

As with normal thyroid tissue, we didnot observe any P3 or p5 immunoreac-tivity in thyroid cancers (not shown).

Among basal lamina components,laminin 1 retained the normal patternwithin rumors, whereas laminin 5 re-mained undetectable in carcinomas (notshown). Interestingly, laminin 2 wasdetected at high levels in integrin subunitP4-negative tumors, but it was totally ab-sent in integrin subunit p4-positive can-cers (not shown). The inverse relationshipbetween integrin 0^4 and laminin 2 im-munoreactivity was strikingly replicable:



CO ,_ N

t .< o oz u. u. a.

—110 kDa

WB: beta 1C CM

CO I*- T -co

t - H H < < < < O O O O O O O O O Oz z z u . u . L u . u . u . u . u . u . Q . a a . Q . a .

— 190 kDa

WB: beta 4o r» co

co in CM r- i- i- a> co v-< O O O O O O O O OU . U . U . U . Q . a . Q . a a . 0 .

:*•

•5.5kb

28S

18S

NB: beta 4

Fig. 2. Panel A: western blot (WB) analysis of pi integrin chain expression in representative cases from nor-mal (NT) and neoplastic thyroid (FA, follicular adenoma; FC, follicular carcinoma; PC, papillary car-cinoma). Panel B: indirect immunofluorescence observation of the integrin subunit p4 in fine-needleaspiration biopsy smears. p4 is undetectable in smears from nodular goiters (a) and adenomas (b), but it ishighly immunoreactive in some papillary (c, case PC 13) and follicular (d. case FC1) carcinomas. Bardenotes 2 urn. Panel C: western blot (WB) analysis of pV» integrin chain expression in nonnal and neoplasticthyroid tissues. Panel D: northern blot (NB) analysis of (It transcripts in neoplastic thyroid samples. A 5.5-kilobase (kb) transcript corresponding to p4 messenger RNA is detected in the same samples expressing p4protein but not in p4 protein-negative extracts. kDa = kilodaltons.

All carcinomas showing integrin subunitP4 neosynthesis did not express laminin 2.

Expression and Functional Statusof the E-cadherin-Catenin Complexin Normal and Neoplastic ThyroidTissues

E-cadherin was detected in thyroid fol-licular epithelium at the boundaries be-tween adjacent cells (Fig. la, panel B).This strictly lateral topography was main-tained in adenomas; however, it was criti-cally subverted in malignant tumors,where E-cadherin was diffusely dis-tributed over the entire membrane in allneoplastic cells (Fig. lb, panel B). Thedistribution of (J-catenin was identical tothat of E-cadherin in normal, adenoma-tous, and cancerous thyroid tissues (notshown).

E-cadherin protein levels followingSDS extraction were identical in normal,adenomatous, and cancerous thyroid tis-sues. Quantitative recovery of a- andp-catenins after non-ionic detergent ex-traction was also identical in all samples(Fig. 3, panel A).

The loss of lateral exposure and thepericellular distribution of E-cadherin in-dicate that the association of E-cadherinwith the cytoskeleton could be impairedin malignant thyroid rumors. To verifythat this is actually the case, we measuredthe relative fractions of total E-cadherinthat either were or were not resistant toextraction by non-ionic detergents (Fig. 3,panel B). In fact, in normal thyroid tis-sues and in thyroid adenomas, E-cadherinwas only partially extractable in non-ionic detergents, with approximately 50%of the total pool being recovered from thecytoskeleton-associated fraction; conver-sely, in malignant tumors, most E-cad-herin could be extracted by non-ionicdetergents, and the insoluble fraction rep-resented only a minor portion. Theseresults support morphologic data and in-dicate that the complex is disconnectedfrom the microfilament network and un-dergoes pericellular relocalization only incarcinomas.

Finally, we examined the tyrosine-phosphorylation status of the E-cadherin—catenin complex in detergent extractsfrom normal and neoplastic thyroid tis-sues (Fig. 3, panel Q . In E-cadherin im-munoprecipitates, three bands of 130,100, and 88 kd, respectively, comigrating



A CM CM i -

K < O OZ u. U. 0.

200^

92 .5 • - •

46*

IP: E-cadherin

WB: E-cadherin

B NT7 FASol. Ins. Sol.

92.5 • « *

IP:WB

toCM CM 1 -H < O OZ u. u. 0.

mmMmPwww

E-cadherin

alpha-catenin

FC3Ins. Sol. Ins

•*-

E-cadherin: E-cadherin

CM CM » -t- < o oz u. u. Q-

^*

E-cadherin

beta-catenin

PC7. Sol. Ins.

-

C NT2

92.5 •

69 • ^

46»-H•IP

WB:

FA FC2PC16

: E-cadherin

Phosphotyrosine

E-cadherin

alpha-cateninbeta-catenin

Fig. 3. Panel A: western blot analysis of the E-cadherin-catenin complex in nor-mal and neoplastic thyroid tissues. The complex is expressed at comparablelevels, with a conserved stoichiometry, in extracts from both benign and malig-nant specimens. Panel B: detergent resistance of E-cadherin in normal andneoplastic thyroid tissues. In normal thyroid tissues and adenomas, E-cadherin islargely resistant to Triton X-100 extraction; in contrast, in carcinomas, it isreadily extracted by non-ionic detergents. Panel C: tyrosine-phosphorylationstatus of the E-cadherin-catenin complex in normal and neoplastic thyroid tis-

sues. In normal thyroid tissues and adenomas, three bands, comigrating with E-cadherin, a-catenin, and fJ-catenin, can be observed. In contrast, no tyrosinephosphorylation of E-cadherin is detectable in follicular and papillary car-cinomas. IP = immunoprecipitation; WB = western blot; NT = normal thyroidtissue; FA = follicular adenoma; FC = follicular carcinoma; PC = papillary car-cinoma; Sol. = Triton X-100 soluble; Ins. = Triton X-100 insoluble. Molecularmass markers are indicated on the left in kilodaltons.

with E-cadherin, a-catenin, and P-catenin,were identified by western blot analysisusing a phosphotyrosine-specific antibody.No tyrosine phosphorylation of y-cate-nin-plakoglobin was observed. One strik-ing observation was that the level oftyrosine phosphorylation of E-cadherinwas markedly decreased and almostabolished in thyroid carcinomas; thisfinding was consistently observed in allmalignant neoplasms, and it was alwaysassociated with unmodified E-cadherinprotein expression.

Discussion

The epithelial phenotype is charac-terized by the following two discretemembrane domains involved in tissuepolarization and cell cohesion: the basaldomain and the lateral domain. The basaldomain is responsible for recognition ofmatrix ligands and attachment to thebasement membrane zone by means ofspecific receptors. One major basal recep-tor is the integrin 0^4 heterodimer that isalso responsible for hemidesmosome as-

sembly (14-16). The lateral domainmediates epithelial cell alignment bymeans of junction-associated adhesionmolecules, e.g., cadherins (40,41) and in-tegrins of the p^ subfamily (4-6).

Spatial disorganization of, functionalimpairment of, or quantitative alterationsin the levels of several adhesionmolecules have been widely reported inepithelial tumors [for references, seeQ J,42)1

The aim of our study was to investigatethe adhesive mechanisms involved in es-

Joumal of the National Cancer Institute, Vol. 88, No. 7, April 3, 1996 REPORTS 447


tablishing the polarized phenotype ofthyroid cells and to observe the topo-graphic and functional modifications oftwo classes of relevant adhesionmolecules at various stages of neoplasticprogression. In our working hypothesis,alterations of cellular adhesiveness occur-ring upon malignant transformationwould result in a rearrangement of ad-hesion molecule expression and functionthat could be readily detected inpreoperative fine-needle aspiration biopsysmears as well as in surgical samples.

The fact that the thyroid carcinomasthat switch on P4 expression and switchoff laminin 2 synthesis are those that areassociated with the most aggressive be-havior could be of major importance bothfrom a biological viewpoint and from aclinical viewpoint. The role of integrinsubunit p4 is not yet fully understood.The 0^4 heterodimer has been shown tomediate stable adhesion to laminin 1,laminin 2, and laminin 5 (43-45). In con-trast, integrin subunit o^-transfected cellscan adhere to laminin 2, but they fail toattach to and spread on laminin 1 (46).Therefore, the fact that normal thyroidtissue expresses one integrin receptor((X3P1) and two laminin isoforms(laminins 1 and 2) and yet only oneligand is recognized by the integrinheterodimer seems paradoxical. We sug-gest that, under normal conditions invivo, recognition between follicular cellsand the basal lamina is mediated by theinteraction of integrin (X3P1 and laminin 2(and possibly by unknown integrins ornonintegrin receptors). When synthesis oflaminin 2, which has been shown to berelated to a highly differentiated pheno-type (33,47), is switched off in a subset ofaggressive thyroid tumors, the de novoexpression of 0^4 might confer a selec-tive advantage for invasion by providinga complementary and more versatilelaminin receptor capable of triggeringrecognition events and attachment tolaminin 1 that precede infiltration.

Detection of 0^4 exposure can beachieved easily in smears from fine-needle aspiration biopsies. If validated bya large-scale clinical investigation that isnow in progress (Orlandi F, Saggiorato E,Serini G, Trusolino L, Marchisio PC, An-geli A: manuscript in preparation), thisfinding might be relevant for the presur-gical detection of aggressive carcinomas

and for the differential diagnosis betweenfollicular adenomas and carcinomas.

E-cadherin-p-catenin localization innormal and adenomatous thyroid tissuesis restricted to the lateral domain of juxta-posed thyroid cells. In contrast, in carci-nomas, the complex undergoes peri-cellular diffusion. This subverted topog-raphy is accompanied by a different pat-tern of detergent extractability betweennoncancerous and cancerous lesions thatis consistent with disconnection of the E-cadherin-catenin complex from the actinmicrofilaments. This surface rearrange-ment is not associated with any changedsynthesis of E-cadherin and a- and P-catenins. To our knowledge, this is thefirst report in which loss of E-cadherin-catenin lateral polarization and removalof the complex from cytoskeleton-asso-ciated adhesion sites, without reduced ex-pression of the molecules involved, havebeen described in the progression frombenign to malignant lesions.

Tyrosine phosphorylation of the E-cad-herin-catenin complex plays a key role inperturbation of the cadherin cell adhesionsystem (3,48,49). In normal adult tissues,specific proto-oncogenic tyrosine kinasesof the src family are enriched at zonulaadherens, where the level of tyrosinephosphorylation is high (50). Since apositive association exists in our data be-tween the three parameters of decreasedtyrosine phosphorylation, loss of deter-gent insolubility, and pericellular re-distribution of E-cadherin, we suggestthat tyrosine phosphorylation of the E-cadherin cytodomain may play a role incontrolling not only the structure of thewhole molecule and, hence, intercellularadhesion, but also its stability within themembrane plane. In addition, dissociationof the E-cadherin-catenin complex fromjunctional regions could drive the com-plex away from kinases (57) and phos-phatases (52) specifically acting onE-cadherin and responsible for the phos-phorylation status of the molecule undernormal conditions.

In summary, the addition of a panel ofMAbs to adhesion molecules in the tool-box of thyroid surgical pathologists, ifvalidated in larger studies, may expandthe options to reach a reasonably correctearly diagnosis and even gain some prog-nostic information from routine needlebiopsies.

References

(/) Wollner DA, Nelson WJ. Establishing andmaintaining epithelial cell polarity. Roles ofprotein sorting, delivery and retention. J CellScil992;102(Pt2):185-90.

(2) Juliano RL, Vamer JA. Adhesion molecules incancer the role of integrins. Curr Opin CellBioll993;5:812-8.

(3) Takeichi M. Cadherins in cancer: implicationsfor invasion and metastasis. Curr Opin CellBioll993;5:806-ll.

(4) Ruoslahti E. Integrins. J Clin Invest 1991 ;87:1-5.

(5) Hynes RO. Integrins: versatility, modulation,and signaling in cell adhesion. Cell 1992;69:11-25.

(6) Marchisio PC. Integrins and tissue organiza-tion. Adv Neuroimmunol 1991; 1:213-30.

(7) Burgeson RE, Chiquet M, Deutzmann R,Ekblom P, Engel J, Kleinman H, et al. A newnomenclature for laminins. Matrix Biol 1994;14:209-11.

(8) De Luca M, Tamura RN, Kajiji S, BondanzaS, Rossino P, Cancedda R, et al. Polarized in-tegrin mediates keratinocyte adhesion to basallamina. Proc Natl Acad Sci U S A 1990;87:6888-92.

(9) Marchisio PC, Bondanza S, Cremona O, Can-cedda R, De Luca M. Polarized expression ofintegnn receptors (ot6{J4, a2pi, a3pi andocv|35) and their relationship with the cyto-skeleton and basement membrane matrix incultured human keratinocytes. J Cell Biol1991; 112:761-73.

(10) Larjava H, Peltonen J, Akiyama SK, YamadaSS, Gralnick HR, Umo J, et al. Novel functionfor p"l integrins in keratinocyte cell-cell inter-actions. J Cell Biol 1990; 110:803-15.

(//) Carter WG, Wayner EA, Bouchard TS, KaurP. The role of integrins a2f)l and rx3p"l incell-cell and cell-substrate adhesion of humanepidermal cells. J Cell Biol 1990; 110:1378404.

(12) Sonnenberg A, Linders CJ, Daams JH, KennelSJ. The a6pl (VLA-6) and a6P4 protein com-plexes: tissue distribution and biochemicalproperties. J Cell Sci 1990;96(Pt 2):207-17.

(13) Natali PG, Nicotra MR, Bigotti A, De MartinoC. Localization of the a6 and (}4 integrin sub-units in normal human non-lymphoid tissues. JCell Sci l992:103(Pt4):1243-7.

(14) Stepp MA, Spurr-Michaud S, Tisdale A, El-well J, Gipson IK. oc6[}4 integrin heterodimeris a component of hemidesmosomes. Proc NatlAcad Sci U S A 199O;87:897O-4.

(15) Sonnenberg A, Calafat J, Janssen H, Daams H,van der Raaij-Helmer LM, Falcioni R, et al.Integrin ot6/P4 complex is located in hemides-mosomes, suggesting a major role in epider-mal cell-basement membrane adhesion. J CellBiol 1991:113:907-17.

(16) Jones JC, Kurpakus MA, Cooper HM, QuarantaV. A function for the integrin alpha 6 beta 4 inthe hemidesmosome. Cell Regul 1991:2:427-38.

(17) Mansouri A, SpurrN, Goodfellow PN, KemlerR. Characterization and chromosomal local-ization of the gene encoding the human celladhesion molecule uvomorulin. Differentiation1988;38:67-71.

(18) Ringwald M. Schuh R, Vestweber D, EistetterH, Lottspeich F, Engel J, et al. The structure ofcell adhesion molecule uvomorulin. Insightsinto the molecular mechanisms of Ca + de-pendent cell adhesion. EMBO J 1987:6:3647-53.

(19) GallinWJ, SorkmBGEdebnanGM, CunninghamBA. Sequence analysis of a cDNA done encoding



the liver cell adhesion molecule, L-CAM. ProcNatl Acad Sci U S A 1987;84: 2808-12.

(20) Gumbiner B. Simons K. A functional assay forproteins involved in establishing an epithelialoccluding barrier identification of a uvo-morulin-like polypeptide. J Cell Biol 1986;102:457-68.

(27) Hirano S, Nose A, Hatta K, Kawarkami A,Takeichi M. Calcium-dependent cell-cell ad-hesion molecules (cadherins): subclass speci-ficities and possible involvement of actinbundles. J Cell Biol 1987;105(6Pt l):2501-10.

(22) Ozawa M, Baribault H, Kemler R. Thecytoplasmic domain of the cell adhesion mole-cule uvomorulin associates with three inde-pendent proteins structurally related indifferent species. EMBO J 1989;8:1711-7.

(23) Ozawa M, Ringwald M, Kemler R.Uvomorulin-catenin complex formation isregulated by a specific domain in the cytoplas-mic region of the cell adhesion molecule. ProcNatl Acad Sci U S A 1990,87:4246-50.

(24) Nagafuchi A, Takeichi M. Transmembranecontrol of cadherin-mediated cell adhesion: a94 kDa protein functionally associated with aspecific region of the cytoplasmic domain ofE-cadherin. Cell Regul 1989; 1:37-44.

(25) Johnson TL, Lloyd RV, Thompson NW,Beierwaltes WH, Sisson JC. Prognostic im-plications of the tall cell variant of papillarythyroid carcinoma. Am J Surg Pathol 1988;12:22-7.

(26) Carcangiu ML, Zampi G, Rosai J. Poorly dif-ferentiated (insular) thyroid carcinoma. A rein-terpretation of Langhans' wuchernde Struma.Am J Surg Pathol 1984;8:655-68.

(27) Carcangiu ML, Bianchi S, Savino D, VoynickIM, Rosai J. Follicular Hiirthle cell tumors ofthe thyroid gland. Cancer 1991;68:1944-53.

(28) Hermanek P, Sobin LH, editors. TNM clas-sification of malignant tumors. 4th ed. NewYork: International Union Against Cancer andSpringer-Verlag, 1987.

(29) Lowhagen T, Granberg PO, Lundell G, Skin-nari P, Sundblad R, Willems JS. Aspirationbiopsy cytology (ABC) in nodules of thethyroid gland suspected to be malignant. SurgClin North Am 1979;59:3-18.

(30) Bottini C, Miotti S, Fiorucci S, Facheris P,Menard S, Colnaghi MI. Polarization of thecc6f34 integrin in ovarian carcinomas. Int JCancer 1993:54:261-7.

(31) Kajiji S, Tamura RN, Quaranta V. A novel in-tegrin (ae-p^) from human epithelial cells sug-gests a fourth family of integrin adhesionreceptors. EMBO J 1989;8:673-8O.

(32) Verrando P, Hsi BL, Yeh Gl, Pisani A, SerieysN, Ortonne JP. Monoclonal antibody GB3, anew probe for the study of human basement

membrane and hemidesmosomes. Exp CellRes 1987; 170:116-8.

(33) Leivo I, Engvall E. Merosin, a protein specificfor basement membranes of Schwann cells,striated muscle, and trophoblast, is expressedlate in nerve and muscle development. ProcNatl Acad Sci U S A 1988;85:1544-8.

(34) Savoia P, Trusolino L, Pepino E, Cremona O,Marchisio PC. Expression and topography ofintegrins and basement membrane proteins inepidermal carcinomas: basal but not squamouscell carcinomas display loss of a6()4 and BM-600/nicein. J Invest Dermatol 1993; 101:352-8.

(35) Rabino M, Trusolino L, Prat M, Cremona O,Savoia P, Marchisio PC. A monoclonal an-tibody identifies a novel GPI-anchored glyco-protein involved in epithelial intercellularadhesion. J Cell Sci 1994;107(Pt 6): 1413-28.

(36) Laemmli UK. Cleavage of structural proteinsduring the assembly of the head of bac-teriophage T4. Nature 1970;277:680-5.

(37) Di Renzo MF, Olivero M, Ferro S, Prat M,Bongarzone I, Pilotti S, et al. Overexpressionof the c-MET/HGF receptor gene in humanthyroid carcinomas. Oncogene 1992;7:2549-53.

(38) Towbin H, Staehelin T, Gordon J. Electro-phoretic transfer of proteins from poly-acrylamide gels to nitrocellulose sheets:procedure and some applications. Proc NatlAcad Sci U S A 1979;76:435O-3.

(59) Chomczynski P, Sacchi N. Single-step methodof RNA isolation by acid guanidium thio-cyanate-phenol-chloroform extraction. AnalBiochem 1987; 162:156-9.

(40) Takeichi M. Cadherins: a molecular familyimportant in selective cell-cell adhesion. AnnuRev Biochem 1990;59:237-52.

(41) Takeichi M. Cadherin cell adhesion receptorsas a morphogenetic regulator. Science 1991;251:1451-5.

(42) Brabant G, Hoang-Vu C, Cetin Y, Dralle H,Scheumann G, Molne J, et al. E-cadherin: adifferentiation marker in thyroid malignancies.Cancer Res 1993;53:4987-93.

(43) Lee EC, Lotz MM, Steele GD Jr, MercurioAM. The integrin a6f}4 is a laminin receptor. JCell Biol 1992;117:671-8.

(44) Niessen CM, Hogervorst F, Jaspars LH, deMelker AA, Delwel GO, Hulsman EH, et al.The 0t6f}4 integrin is a receptor for bothlaminin and kalinin. Exp Cell Res 1994;211:360-7.

(45) Spinardi L, Einheber S, Cullen T, Milner TA,Giancotti FG. A recombinant tail-less integrin(J4 subunit disrupts hemidesmosomes, but doesnot suppress a6P4-mediated cell adhesion tolaminins. J Cell Biol 1995; 129:473-87.

(46) Delwel GO, de Melker AA, Hogervorst F,Jaspars LH, Fles DL, Kuikman I, et al. Dis-tinct and overlapping ligand specificities of thea3Api and a6Af}l integrins: recognition oflaminin isoforms. Mol Biol Cell 1994;5:203-15.

(47) Simon-Assmann P, Duclos B, Orian-RousseauV, Arnold C, Mathehn C, Engvall E, et al. Dif-ferential expression of laminin isoforms anda6-(}4 integrin subunits in the developinghuman and mouse intestine. Dev Dyn 1994;201:71-85.

(48) Matsuyoshi N, Hamaguchi M, Taniguchi S,Nagafuchi A, Tsukita S, Takeichi M. Cad-herin-mediated cell-cell adhesion is perturbedby v-src tyrosine phosphorylation in metastaticfibroblasts. J Cell Biol 1992; 118:703-14.

(49) Behrcns J, Vakaet L, Friis R, Winterhager E,Van Roy F, Mareel MM, et al. Loss ofepithelial differentiation and gain of invasive-ness correlates with tyrosine phosphorylationof the E-cadherin/p^-catenin complex in cellstransformed with a temperature-sensitive v-SRC gene. J Cell Biol 1993; 120:757-66.

(50) Tsukita S, Oishi K, Akiyama T, Yamanashi Y,Yamoto T, Tsukita S. Specific proto-onco-genic tyrosine kinases of src family are en-riched in cell-to-cell adherens junctions wherethe level of tyrosine phosphorylation iselevated. J Cell Biol 1991;113:867-79.

(5/) Hoschuetzky H, Aberle H, Kemler R. p-Catenin mediates the interaction of the cad-herin-catenin complex with epidermal growthfactor receptor. J Cell Biol 1994;127:1375-80.

(52) Brady-Kalnay SM, Rimm DL, Tonks NK.Receptor protein tyrosine phosphatase PTP|lassociates with cadherins and catenins in vivo.J Cell Biol 1995;130:977-86.

Notes

EHS—laminin is derived from Engelbreth-Holm-Swarm sarcoma.

G. Serini and L. Trusolino contributed equally tothe experimental work described in this report.

Supported by Target Project "ApplicazioniCliniche della Ricerca Oncologica" ConsiglioNazionale delle Ricerche (Rome), by AssociazioneItaliana per la Ricerca sul Cancro (Milano), and bythe Ministero per l'Universita e la Ricerca Scien-tifica e Tecnologica (Rome).

We gratefully acknowledge the skillful technicalassistance of Germana Cecchini. We are indebted toAndrea Graziani and Daniela Gramaglia for provid-ing the monoclonal antibody to phosphotyrosine.

Manuscript received August 8, 1995; revisedNovember 3, 1995; accepted January 25, 1996.

A message brought to you by the National Cancer Institute



changes in integrin and e-cadherin expression in neoplastic

Documents