[CANCER RESEARCH 54, 4539-4546, August 15, 1994]

Normal Human Tissues, in Addition to Some Tumors, Express Multiple

Different CD44 Isoforms

Stephen B. Fox, Jonathan Fawcett, David G. Jackson, Ian Collins, Kevin C. Gatter, Adrian L. Harris,Andrew Gearing, and David L. Simmons1

Nuffield Department of Pathology; University of Oxford ¡S.B. F., K. C. G.], Cell Adhesion Laboratory /J. F., D. L. S.¡,and Molecular Oncology Unit ¡J.F., A. L. //./, ImperialCancer Research Fund, Laboratories, and Molecular Immunology Group [D. G. J.], Institute of Molecular Medicine, John Radcliffe Hospital, Oxford OX3 9DU, UnitedKingdom: and R&D Systems Europe [I. C./ and British Biotechnology Ltd, ¡A.G./, Abingdon, Oxon, United Kingdom

ABSTRACT

At least 20 different isoforms of the human CD44 lymphocyte-homing

receptor/hyaluronan receptor have been described to date that arise fromthe differential splicing of up to 10 alternative exons (termed vl-vlO)encoding the membrane-proximal extracellular domain. Although numerous analyses at the in UNA level have indicated tissue-specific expression of

CD44 variants, few analyses have been performed at the protein levelbecause of limited availability of suitable monoclonal antibodies. Recently,however, exon-specific monoclonal antibodies have been generated using

bacterial fusion proteins, and these have been reported to detect highlevels of vCD44 containing the v6 exon on human tumors. Together withearlier evidence linking this particular exon with tumor metastasis in therat, these latter experiments have led to the interpretation that v6 splicevariants play a causative role in tumor dissemination.

In this paper we describe the use of a new and comprehensive panel ofCD44 exon-specific monoclonal antibodies generated against a recombi-nant CD44(v3-10)-immunoglobulin chimera to study vCD44 expression in

a large number of normal and neoplastic tissues. We show that theexpression of vCD44 varies greatly among different human tumors andthat some express either very low levels of vCD44 or no CD44 at all.Furthermore, we demonstrate that expression is not limited to isoformscontaining the v6 exon but includes variants carrying v3, v4/5, and v8/9.Additionally, normal epithelial tissues are shown to express considerable levels of these same vCD44 isoforms. Such results argue against aubiquitous role for vCD44 isoforms in promoting tumor growth andmetastasis.

INTRODUCTION

CD44 is the major cell surface receptor for hyaluronan (1-4) and

exists as multiple isoforms generated by the alternative splicing of upto 10 exons (vl-vlO) encoding parts of the extracellular domain(5-7). An abundant M, 90,000 isoform termed the hemopoietic variant

(CD44H) lacks all 10 variable exons and is ubiquitously expressed oncells of mesodermal and hemopoietic origin. In contrast, the CD44isoforms, which contain multiple alternatively spliced exons, range insize from M, 120,000 (CD44 v8-10) to 250,000 (CD44 v3-10) and

are predominantly expressed on cells and tumors of epithelial origin(2, 6, 7).

While CD44H appears to function in such diverse processes aslymphocyte homing, lymphocyte activation, and extracellular matrixadhesion (8, 9), the precise functions of each of the alternativelyspliced CD44 isoforms are less clear. Interestingly, a rat isoformcontaining exons v4-v7 has been shown to promote metastasis aftertransfection of the cDNA2 into nonmetastatic rat carcinoma and

sarcoma cell lines (10-13). In a different set of experiments, human

Received 12/28/93; accepted 6/8/94.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance withIS U.S.C. Section 1734 solely to indicate this fact.

1To whom requests for reprints should be addressed, at Cell Adhesion Laboratory,

Imperial Cancer Research Fund, Institute of Molecular Medicine. John Radcliffe Hospital,Headington, Oxford, OX3 9DU, United Kingdom.

2 The abbreviations used are: cDNA. complementary DNA; PCR, polymerase chain

reaction; mAb, monoclonal antibody; SDS-PAGE, sodium dodecyl sulfate-polyacrylam-ide gel electrophoresis; ELISA, enzyme-linked immunosorbent assay.

CD44H has also been shown to enhance metastasis when transfectedinto a human B-cell lymphoma (14). Clearly, these results indicate an

important role for CD44 in regulating tumor progression. However,the relative contribution of each individual CD44 variant in differenttypes of tumor is not yet clear.

In a recent series of studies, a number of polyclonal and monoclonalantisera directed to human CD44 alternatively spliced exons havebeen used to detect vCD44 expression in different human tumors byimmunohistochemical staining. In one group of studies, high levelexpression of CD44 isoforms carrying the v6 exon was reported incolorectal carcinomas, non-Hodgkins lymphomas, and certain typesof gastric adenocarcinomas (15-18). Indeed, the level of CD44v6

expression appeared to indicate the stage of tumor progression incases of colorectal carcinoma. However, other workers have shown adown-regulation of CD44v6 expression in tumors of squamocellularorigin and an up-regulation of CD44v9 expression in primary gastric

tumors (17, 18).In order to gain a clearer picture of the role and expression of CD44

variants in different tumors, we have generated a panel of mousemonoclonal antibodies to the human CD44 exons v3, v4/5, v6, andv8/9 using soluble CD44-immunoglobulin Fc fusion proteins as the

immunogen. We have used these reagents in a comprehensive analysisof human vCD44 expression which has revealed that expression ofCD44 isoforms in tumors is not restricted to those carrying the v6exon but also includes those carrying v3, v4, and v8/9. Furthermore,we have shown that vCD44 expression by human tumors is extremelyheterogeneous and that some malignant tumors appear to express noCD44 at all. In agreement with a recent study in which v4, v6, and v9expression profiles in normal tissues were analyzed (19), we havefound that the expression of vCD44 by normal human tissues islargely restricted to the epithelia, indicating that in most normal cellsthe process of CD44 alternative splicing is tightly regulated.

MATERIALS AND METHODS

Construction of pCD44H-Fc, pCD44(v8-10)-Fc, and pCD44(v3-10)-Fc.

Two strategies were used to clone vCD44 cDNAs. In the first strategy, theCD44 mAb F. 10.44.2 (20) was used to isolate full-length CD44 clones from amixture of four cDNA libraries constructed from phorbol 12-myristate 13-acetate-activated U937 cells, cytokine-activated HUVEC human umbilicalvein endothelial cells, full-term human placenta cells, and HT29 colon carci

noma cells after transient transfection and expression in COS 1 cells (21). Theonly CD44 isoforms isolated by this method were CD44H and CD44(v8-10)

(CD44E) described previously by Stamenkovic et al. (l, 2). In a secondstrategy, cDNAs representing the extracellular domains of vCD44 cDNAswere generated by reverse transcriptase-PCR from the breast carcinoma lineZR75 and cloned into the CD33-immunoglobulin Fc fusion vector (see below).The forward amplification primer 5'-TGTAAGATCTCGCG CAGATC-GATTTGAATATAACC-3' was located at position 175 of the published

CD44H sequence and incorporated a Bglll site. The reverse amplificationprimer 5'-TGTAAGATCTACTTACCTG TCCATTCTGGAATTTGGGGT-GTCCTTAT-3' was located immediately 5' of the transmembrane domain at

position 920 and incorporated a Bg/II site and a splice donor consensus site.The PCR conditions used were: 95°C,30 s; 55°C,30 s; 72°C,30 s, cycled 30

4539

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

CD44 VARIANT-SPECIFIC MONOCLONAL ANIIIiODIIS

times. PCR products were end filled with Escherichia coli DNA polymerase(Klenow fragment), blunted with T4 polynucleotide kinase. and cloned into£roRV-cut, phosphatase-treated pBluescript. Individual transformants containing CD44H, CD44(v8-10), and CD44(v3-10) inserts were identified by

double-stranded sequencing and cloned into the the HinMl/BglU site of

pCD33 signal immunoglobulin. a derivative of the original plgGI expression

vector (21, 22).The pCD33 signal immunoglobulin vector was constructed by triple ligation

of a 72-nucleotide Hind\\\IBam\\\ fragment encoding the signal peptide ofhuman CD33 (23) together with ßamHI-digested PCR products for the extracellular domains of either CD44H, CD44(v8-10), or CD44(v3-10) and

BawHI-cut plgGI vector (21. 22).Expression of Chimeric Fusion Proteins. pCD44H-Fc, pCD44(v8-10)-

Fc, and pCD44(v3-l())-Fc constructs were transfected into COS cells (10fig/107 COS cells) using DEAE-dextran as a facilitator. The medium was

changed at 24 h to 1% fetal calf serum, and supernatants were harvested at 7days. Fusion proteins were affinity isolated on protein A-Sepharose, eluted at

pH 3.0. neutralized immediately in 10% (v/v) 1 M Tris base, buffer exchangedinto 100 mM Tris (pH 7.5). and concentrated by centrifugation dialysis (Cen-

tricon 10 filtration units. Amicon). For production of labeled proteins. COStransfectants were grown in methionine-free medium containing 5% dialyzedfetal calf serum and 50 /iCi/ml [15S]methionine/cysteine (Translabel. ICN), for

18 h. Supernatants were" harvested, and labeled secreted proteins were isolated

by affinity purification on columns of protein A-Sepharose. Bound proteins

were eluted by boiling in Laemmli sample buffer under reducing conditionsand resolved on 10% SDS-PAGE. The gel was fixed, impregnated withAmplify (Amersham), dried, and exposed to X-ray film for 12 h at —80°C.

Generation of CD44 mAbs. Mice were immunized with a course of threeinjections of 10 fig CD44(v8-10)-Fc or CD44(v3-10)-Fc proteins, the first incomplete Freund's adjuvant and the subsequent two injections in Freund's

incomplete adjuvant. Spleens were isolated and hybridomas produced bystandard methods (24). Hybridoma supernatants were screened by solid phaseELISA using the fusion protein immunogens, pCD44H-Fc, pCD44v8—10-Fc,and pCD44v3-10-Fc, and pCD14-Fc as a control for Fc reactivity. Reactivity

profiles were established for all positive wells against these 4 proteins. Allhybridomas were cloned 3 times by limiting dilution and then isotyped.

Determination of Antibody Specificity. The exon specificity of thevCD44 monoclonal antibodies was determined by indirect immunofluorescentantibody staining of a panel of transfected COS 1 cells each expressingdifferent full-length human vCD44 cDNAs including CD44(vlO), CD44(v8-10), CD44(v7-10), CD44(v6-10), CD44(v3, 8-10), and CD44(v3-10). Fulldetails concerning the construction of the full-length vCD44 cDNAs and thederivation of the CD44*pRcCMV "cassette cloning vector" used for expres

sion will be published elsewhere. Briefly, PCR products containing the appropriate alternative exon combinations amplified from human leukocyte or tumorcell cDNAs were ligated into a unique ß#/II,WarI-cloning site within a full-

length CD44(vlO) cDNA in the expression vector pRcCMV. Individual constructs ( 1 mg DNA) were introduced into recipient COS cells plated in 3.5-cmtissue culture dishes by incubation (2 h, 37°C) with DEAE-dextran and

chloroquinc in serum-free RPMI, and 10% Nu-Serum (Collaborative Re

search), followed by a 10% dimethyl sulfoxide shock (21). After 48 h inculture the transfected cells were incubated with samples of undiluted hybri-

doma culture supernatants or purified immunoglobulin (10 fig/ml) for 20 minat 5°C,followed by staining with fluorescein isothiocyanate-conjugated goat

anti-mouse IgG or IgM antibodies (Sigma Chemical Co., Poole, United King

dom), as appropriate.Cell Surface Protein Labeling and Immunoprecipitations. Samples (10s

cells) of the ZR75 breast carcinoma were surface labeled with Na-'25I (1 mCi;

Amersham, United Kingdom), using immobilized glucose oxidase/lactoperox-idase (Enzymobeads, Bio-Rad) according to the protocols of the manufacturers, and extracted in 100 mM Tris (pH 7.4)-2 mM EDTA-2 mM phenylmeth-ylsulfonyl fluoride-1% Nonidet P-40, and immunoprecipitations were carried

out using 10 jig purified antibodies or. in the case of mAb 1E8 (IgM), as a10-fold concentrate of tissue culture supernatant. Antigen-mAb complexeswere recovered with affinity-purified goat anti-mouse IgG, IgA, and IgM

Antibody, coupled to agarose (Sigma), and following elution, proteins wereresolved on 7.5% SDS-PAGE gels. Control mAbs included F. 10.44.2 (aframework CD44 antibody: 20), 9G11 (anti-CD31, R&D Systems Europe),and mAb 38 (anti-CD 11a; Nancy Hogg, Imperial Cancer Research Fund).

Preparation of Normal and Neoplastic Tissues and Immunohistochem-

ical Staining. Fresh tissue samples were obtained from the histopathologydepartment at the John Radcliffe Hospital and immediately snap frozen inliquid nitrogen prior to storage at -70°C. Cryostat sections (8 firn) were cut

onto multiwell slides, dried overnight at room temperature, fixed in acetone for10 min, and then air dried. Slides were then stained or wrapped in aluminumfoil and stored at —20°Cuntil stained. Tissue samples were also available as

routinely fixed specimens in paraffin blocks. Tissues studied included skin,

bladder, thymus, spleen, tonsil, lymph node, thyroid, adrenal, pancreas, sali

vary gland, liver, kidney, heart, esophagus, stomach, duodenum, colon, lung,

ovary, testes, uterus, cervix, placenta, brain, spinal cord, prostate, breast, and

muscle. Sections from a series of breast, colorectal, and bronchogenic carcinomas either snap frozen or formalin-fixed and paraffin-embedded were also

examined.For immunohistochemical staining each of the anti-CD44 monoclonal an

tibodies listed in Table 1 was applied to tissue sections for detection by thealkaline phosphatase/anti-alkalin e phosphatase or streptavidin-biotin-peroxi-

dase (Dako Duet Kit; Dako, United Kingdom) techniques. For tissues withendogenous biotin a two-stage immunoperoxidase procedure was used (25).

Omission of the primary antibody was used as a negative control. A microwaveretrieval technique was performed with all antibodies on a selection of routinely processed normal and neoplastic tissues (26, 27).

Table 1 ELISA profilas of CD44\- mAbs on CD44Fc fusion proteins

2C53G53D22F101E8CD44

H-Fc1.9360.0380.0320.031(1.020CD44(v8-10)-Fc1.7310.0380.0380.0351.035CD44(v3-10)-Fc0.9811.4201.0121.290.031CD14-FC0.0240.0320.0390.0380.000CDSO-Fcn.d."n.d.n.d.n.d.0.040

' n.d., not determined.

£I

rí

I

Size(bp)

1900-

1350-

700-

Fig. 1. CD44 isoform PCR products generated from first-strand cDNA made from

RNA extracted from cell lines: A. BstEll size markers: lane l, ZR75. breast carcinoma;lane 2, MCF7 breast carcinoma; lane 3, U937. promonocytic leukemia. Products wereresolved by electrophoresis on a 1% agarose gel.

4540

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

CD44 VARIANT-SPECIFIC MONOCLONAL ANTIBODIES

RESULTS

Generation of Soluble CD44-Fc Fusion Proteins. Two different

approaches were used to generate CD44 isoforms for the productionof CD44 exon-specific mAbs. The first approach, which involved

antibody panning of COS cells transfected with a pool of cDNAlibraries in the pCDMS expression vector, yielded CD44H andCD44(v8-10). However, despite repeated attempts, this method failed

to yield cDNAs encoding other larger CD44 isoforms because of thelow levels of expression found in most cell types. Consequently, twobreast tumor cell lines, MCF7 and ZR75, that were known to expresshigh molecular weight CD44 isoforms (data not shown) werescreened for vCD44 mRNA levels by reverse transcriptase-PCR using

primers that encompass the entire coding region for the extracellular

domain. Analysis of the products (Fig. 1) confirmed the presence ofmultiple CD44 splice variant RNAs in the ZR75 and MCF7 cell linesand their absence from the hemopoietic cell line U937. The largestPCR product generated from ZR75 cells was subsequently cloned,sequenced, and identified as CD44(v3-10).

Next, all three CD44 isoforms were expressed as soluble immuno-

globulin Fc chimeras by subcloning the region encoding the extracellular domain in each case into the expression vector pCD33 signalimmunoglobulin (Fig. 2) for expression and secretion from COS 1cells. High levels of soluble CD44 secretion (approximately 10 /j.g/ml) were confirmed for each construct by SDS-PAGE analysis oftissue culture supernatants after biosynthetic labeling with [35S]me-

thionine/cysteine (Fig. 3).

CD44 gene structure showing exon numbering (Screaton et al, 1992)

Signal Peptide VariablysplicedexonsT-M Tallregion region

I A6 7 8 9 10 11 12 13 14 15 16 17 18 19

I II I 1 I V2 I V3 |V4 | V5 | V6 | V7 | V8 | V9 | VIO | |\~\ =universally present exon u= cytoplasmlcdomainexons

PCR generation of extracellular regions of CD44 variants

••--t11 i iHemopoietic

- Splice _ Bel IIDonor ^

Epithelial

I I I I I" Splice Bgl IIDonor

Bgl II

"Metastatic"

-I I I I !X>1 I I I I I I I I » t" Splice Bgl IIDonor

CD33 signal peptidsequence

Cut PCR products with Bgl IIand subdone into Bam H Irestricted pCD33slgIg

Fig. 2. Strategy for construction of CD44 variants as soluble chimeric IgG 1-Fc fusion proteins.

/fuñan IgOl Fc

pCD33signal-Igl (5.7kb)

coIElrcflico

SV40 splice undpulyidenylatum

SVIOorigin

origin•The.CD33 lignai peptìdela »coded »yIm«nt l - 72 uà «.i »kcloned latopl£l u a HinD in - BamH I fragment

4541

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

CD44 VARIANT-SPECIFIC MONOCLONAL ANTIBODIES

Generation of a New Panel of CD44 Exon-specific mAbs. Purified CD44(v8-10)-Fc and CD44(v3-10)-Fc fusion proteins wereused as immunogens to generate CD44 alternative exon-specific

monoclonal antibodies. As an initial screen, hybridoma supernatantswere tested by solid phase ELISA with CD44H-Fc, CD44(v8-10)-Fc,and CD44(v3-10)-Fc and a control CD14-Fc, to define reactivity witheither the v8-10 or v3-7 regions of the CD44 molecule (Table 1).

Each mAb was then screened for reactivity against a panel of COScell transfectants expressing cloned CD44 cDNAs carrying differentcombinations of the v3-vl() alternative exons. As a result of thisanalysis (Table 2) five different CD44 exon-specific mAbs were

identified that recognized specifically v3, v4/5, v6, and v8/9. Inaddition, all five mAbs reacted with unfixed ZR75 breast carcinomacells and revealed strong expression of v3, v4/5, and v6 but smalleramounts of v8/9 (data not shown). Finally, all five antibodies, inaddition to the framework mAbs F. 10.44.2 and 2C9, immunoprecipi-

tated the same wide spectrum of protein bands ranging in size from Mr100,000-200,000 from ZR75 cells, consistent with the presence of

high levels of CD44 isoforms carrying multiple alternative exoncombinations (Fig. 4).

Normal Tissues Express a Diverse Repertoire of CD44 Iso-

Mr

200-

92-

12345678

Fig. 4. Immunoprecipilalion of '^[-surface labeled CD44 isoforms from ZR75 breastcarcinoma cells. Lane 1, mAb F44H-10; lane 2, mAb 2C5; lane 3, mAb 3G5; lane 4, mAb

3D2; lane 5, mAb 4B3; lane 6, mAb 1E8; lane 7, control mAb 9G11 ; lane 8, control mAb38. Cells were surface labeled with 125I,lysed. and immunoprecipitatcd with mAbs; either

10 ng of pure mAb or 10-fold concentrated tissue culture supernatant from the IgM mAb1E8. mAbs were isolated with goat anti-mouse IgGAM agarose, and proteins wereresolved on 7.5% SDS-PAGE.

Table 3 Expression of CD44H and CD44 variants in normal human tissues

See text for cellular distribution.çyconsiderableexpression

of vCD44 isoforms that was largely confinedto the epithelia (Table 3). In particular, respiratory epithelium,tran-M

HEtafllBK_•

^B200^BBP^'B^B^B^B^BÌ92^B^^b.

^^^H^^^••l^k67^•^^^44***!Fig.

3. SDS-PAGE analysis of variant CD44-Fc fusion proteins. Lane M,CD44meta-Fc;lane H. CD44H-Fc; lane E. CD44E-Fc. Proteins were metabolically labeledbyaddition

of |"S]methionine/cysteine to COS cells transiently expressing plasmid con

structs. Proteins were resolved on 10%SDS-PAGE.Table

2 Assignment of mAbs to CD44 variantexonsmAb

name IsotypeVariant2C5

G2aHemopoietic/backbone3G5G2bv3•ar\">,\lA/^jU¿\JlV't/j4B3

Mvf)2F10Gìv61E8M v8/92C5

3G5 3D2 2F10/4B3 1E8Tissue and type (total) (v3) (v4/5) (v6)(v8/9)HemopoieticTonsilLymphocytes

+ + +" +* ±6 +b+bCrypts

+ + + + + + + + + + + ++Lymphnode + + + ++b ib ++b++bSpleen

+ ++Thymus+ + + ±b±bEndocrineSalivary

gland + + + + + ++ ++Adrenal++Pancreas

+ + + ++Thyroid+ + + + + + + + + + ++Breast+ + + + + + ++ + ++Prostate+ + + + + + ++ + + ++ReproductiveEndometrium

+ + + + + ++Decidua+ + + ± ± ++Cervix+ + + + + + + + + + + + ++OvaryPlacenta

+ + + + ++Testis ++GastrointestinalTongue

+ + + + + + + + + + + + ++Esophagus+ + + + + + + + + + + + ++Stomach

++ + ±+Duodenum++ + ± ++Colorectal++ + ±+Central

nervoussystemCerebralcortex ++Spinal

cord ++CardiorespiratoryPneumocytes

++Epithelium+ + + + + + + + + + + + ++MyocytesOthersLiverKidney

+MuscleSmooth

+ ++Striated++Skin++ + + + + + + + + + ++Bladder

+ + + ++ ++ ++ ++Endothelium+-t- ± ± ±±°

±.equivocal; +, weak; + + , moderate; + + +, strong staining; -, no staining.b Scattered cellsonly.sitional

epithelium, and keratinizing and nonkeratinizingsquamousepitheliumdisplayed immunoreactivity with each of the mAbstoexons

v3, v4/5, v6, and v8/9 (Figs. 5 and 6). In addition,placentalcvtotroohoblast.thvroid follicular eoithelium. skin adnexae.mvoeni-4542

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

B

•V-- WK

Fig. 5. CD44H (2C5) expression in tonsil demonstrating intense immunoreactivity of lymphoid tissue (A, magnification X 40) and overlying crypt epithelium (B. magnification X 400). C, CD44H in placenta localized to cytotrophoblast and endothelium with weaker matrix staining. D, CD44H in thymus strongly expressed by most medullary lymphocytesand occasional cortical thymocytes (magnification X 40). £.CD44H (2C5) positively in cerebral cortex (magnification X 40) and endothelium. F, Nuclear stippling with 2C5 in testes(magnification X 400). G, Intense CD44E (1E8) membrane staining of cells in salivary gland (magnification X 400).

Fig. 6. A, CD44E (1E8) staining of macrophages and lymphocytes in a reactive lymph node (magnification X 40). B, v3 (3G5) immunoreactivity in stratified squamous epitheliumof skin (magnification X 40). C, v3 (3G5) immunoreactivity in transitional epithelium of bladder (magnification X 400). D, v4/5 (2G9) expression in glands of secretory phaseendometrium (magnification X 400). E, intense v6 (2F10) immunoreactivity of cervical squamous epithelium (magnification, X 400). F, strong v6 (2F10) staining of respiratoryepithelium (magnification X 400). G, strong v6 (2F10) staining of prostatic epithelium (magnification X 400) with basal layer accentuation (arrows).

4543

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

CD44 VARIANT-SPECIFIC MONOCLONAL ANTIBODIES

C * >!

-'Fig. 7. A, intense v6 (2F10) positivity of esophageal squamous epithelium (magnification X 40). B, strong bu! focal v6 (2F10) expression in duodenal epithelium (magnifica

tion X 40, arrows). C, normal breasl stained for v6 (2F10) demonstrating strong staining of myoepithelial cells of acini and ducts (magnification X 40). D, v6 (4B3) of reactive lymphnode showing moderate staining of the germinal centers (magnification X 40). £,pan v8/9 (1E8) tumor staining of lung carcinoma (magnification X 40).

Fig. 8. A, coloréela)carcinoma stained for v3 (3G5; magnification X 40). B, coloréela)carcinoma stained for v4/5 (2G9; magnification X40) showing intense cell membraneimmunoreactivity with accentuation of the infiltrating margin (arrow). C and D, breast carcinoma negative for v6 (2F10) in both the intraduct (X) and invasive element (*)

(magnification x 40) but with positive staining of the residual myoepithelial layer of the duct (arrow).

4544

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

CD44 VARIANT-SPECIFIC MONOCLONAL ANTIBODIES

Table 4 Expression of CD44 and CD44 variants in lung, colorectal, and breast carcinomas

TumorLungColorectalBreastTotalNo.12510272C5(total)12510273G5(v3)1229233D2(v4/5)1219222F10/4B3(v6)1239241E8(v8/9)6°0S'

il" Usually focal and weak ¡mmunoreactivity.

thelial layer of breast and prostate, and Hassals corpuscles of thymuswere also strongly positive with antibodies to v3, v4/5, and v6(Fig. 5). In contrast, only weak focal reactivity was observed for thesevCD44 isoforms in the crypt epithelium of the gastrointestinal tractand pancreatic ducts (Fig. 6). Similarly, CD44 variants carrying thev8/9 exons were weakly expressed in the myoepithelial layer ofprostate glands and Hassals corpuscles. Weak staining for vCD44 wasalso observed in endocervical epithelium, secretory and proliferativephase basal endometrial glands, and glandular elements in decidualtissue. Stroma (cellular and matrix elements), liver, heart, kidney,ovary, testes, adrenal gland, and smooth and striated muscle werenegative.

Normal leukocytes were found to express a much more restrictedrepertoire of variant exons than were found in epithelial cells. OnlyCD44 v4/5 and v6 expression was consistently seen in hematopoietictissues. For example, medullary thymocytes (and less frequently cortical thymocytes) were positive for v4/5 and v6 (Fig. 6 and 7). Inaddition, only occasional tissue macrophages were positive for v4/5and v8/9 (Fig. 6). Interestingly, CD44 v3 expression was seen inlymphocytes within reactive lymph nodes (Fig. 6), indicating thatexpression of this exon may be up-regulated during in vivo lympho

cyte activation.These results provide a clearer and more comprehensive definition

of CD44 alternative exon usage in normal tissues than those of earlierstudies which relied on polyclonal sera. In addition to expressing theso-called metastatic exon v6, normal epithelial cells clearly express

CD44 isoforms carrying v3, v4/5, and v8/9.CD44 Variant Expression in Tumors. The expression of vCD44

isoforms detected in a comprehensive panel of tumors includingbreast, lung, and colorectal carcinomas is summarized in Table 4. Themajority of tumors expressed CD44H together with vCD44 isoformscarrying the v3 and v4/5 exons and the so-called metastatic v6 exon.

Although many tumor cells were strongly immunoreactive with different exon-specific antibodies, there was considerable heterogeneity

in staining intensity among individual tumors (Fig. 8). For example,staining for CD44 v8/9 was weak and focal in many tumors, and yetsome tumors were intensely positive (Fig. 8). Interestingly, strongexpression of CD44(v8/9) was often seen at the invading edges ofcarcinomas and around the periphery of intraductal elements of breastcarcinomas (Fig. 8), suggesting a possible role for CD44 molecules inthe regulation of tumor cell adhesion.

A surprisingly complex pattern of CD44v expression in differenttumors, compared to the CD44 expression of the normal cell of origin,was seen. A striking result was the high levels of expression of manyof the variant exons by breast carcinomas that arise from breast ductalepithelium, which do not normally express CD44. Conversely, normalgastrointestinal epithelium expressed low levels of many variants, butthe cognate colon tumors expressed low and variable levels of thevariants. A third pattern was demonstrated by respiratory epithelium,in which variants were expressed at high levels in normal cells andcontinued to be expressed at those levels in lung carcinomas. Significantly, wherever variants were expressed at high levels, the wholerange of isoforms was usually found (v3, v4/5, v6, and v8/9), and notjust v6.

These results indicate that differential RNA splicing of CD44transcripts may be deregulated in tumor cells, leading to the inclusionof alternative exons in some tumor types and their exclusion in others.Such complex patterns underline the need for caution in using vCD44expression as a diagnostic indicator for malignancy.

DISCUSSION

In this paper we have described the production of a panel of CD44alternative exon-specific mAbs using recombinant CD44-immuno-

globulin Fc fusion proteins as immunogens. A total of five antibodiesspecific for the alternative exons v3, v4/v5, v6, and v8/v9 have beencharacterized, all of which recognize both native and formalin-fixed

cells. These have allowed us to characterize the expression patterns ofCD44 variants, and they have revealed a surprisingly complex patternof vCD44 expression in both normal and neoplastic tissues.

Epithelial cells express a wide range of variants at high levels.Many types of epithelium in many different organs express v3, v4/5,v6, and v8/9. In contrast, activated leukocytes express a more restricted repertoire of variants. The predominant isoforms contain onlyv4/5 and v6. Only occasionally was v3 seen in normal activatedlymphocytes.

In contrast to numerous reports describing CD44H expression,there have been only three immunohistochemical studies of CD44vexpression in normal and neoplastic human tissues (15-19). We have

generated monoclonal antibodies specific for v3, v4/5, v6, and v8/9and have described their distribution in an extensive survey of humantissues. The findings among these studies are generally concordant,where it is possible to directly compare. All detected CD44H inhemopoietic cells and v6 in epithelia. A minor disagreement is thepattern of v6 staining in squamous epithelium; we observed pan-

epithelial expression rather than preferential staining of the superficialportion of squamous epithelium noted by Heider et al. (18). Thesediscrepancies are probably due to minor epitope variations as recognized by the different antibodies used. Although gastrointestinal glandular epithelium was generally vCD44 negative, endometrial glandswere positive during all phases of the menstrual cycle and duringdecidualization of the stroma.

The tissue distribution of CD44(v8-10) is not known. This partic

ular CD44 isoform was originally designated as the epithelial variant(CD44E) since it was initially identified in epithelium and carcinomas. We did observe strong expression of CD44E(v8/9) in manyepithelia, but it was not present in all types examined (Table 3). Arecent study using PCR (28) detected CD44(v8-10) in 15.4% (2 of 13

specimens) and 33% (2 of 6 specimens) of normal colon and liver,respectively, whereas we could demonstrate no CD44(v8/9) expression in these tissues. This is unlikely to be due to the binding affinityof mAb 1E8 since a range of staining intensities was observed withthis antibody in both normal and neoplastic tissues. A more likelyexplanation is the rare expression of this variant together with thesensitivity of PCR.

Recently, there have been many reports concerning the role ofCD44 variants in primary tumors and their métastases(10-14). In

model systems, expression of isoforms carrying the v6 exon confers a

4545

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

CD44 VARIANT-SPECIFIC MONOCLONAL ANTIBOD1I-S

metastatic phenotype, and antibodies directed against this particularexon prevent tumor spread. Overexpression of vCD44 is also reportedto be a feature of human neoplastic cell lines and tumors. Indeed,recent studies using PCR have suggested that the expression ofparticular CD44 splice variants is restricted to tumors (29).

However, in contrast to these reports and in accordance with otherimmunohistochemical and PCR-based techniques, we identified a

number of different CD44 variants in the three tumor types examinedbut could not demonstrate expression of vCD44 in all tumors. Thefinding of vCD44 in tumors might be due either to up-regulation or to

aberrant expression by neoplastic cells. Thus, bronchogenic carcinomas, which arise from respiratory epithelium which normally expresses CD44H, CD44(v8-10), and all vCD44s examined, also ex

press these CD44 isoforms. In contrast, breast carcinomas which arisefrom ductal epithelium, which does not normally express any CD44isoform, express vCD44 (Table 3). Furthermore, whereas expressionof vCD44 in normal tissues was predominantly membranous, inneoplastic cells, expression was both membranous and cytoplasmicwith accentuation of immunoreactivity at the infiltrating tumor margin. The biological significance of this altered cellular distribution andits preferential location at the invading tumor edge conforms to therole of CD44 as a candidate adhesion molecule involved in tumorinvasion and metastasis.

In conclusion, we have generated a panel of monoclonal antibodieswhich specifically recognize variants encoded by the alternativelyspliced exons of CD44 by using recombinant CD44(v3-10)-Fc as an

immunogen. mAbs have been produced which recognize v3, v4/5, v6,and v8/9. These have allowed us to characterize the expression patterns of CD44 variants in normal and neoplastic tissues and haverevealed a surprisingly complex pattern of vCD44 expression. We arecurrently testing whether any of the variant-specific mAbs inhibit CD44v

isoform function in a range of adhesion and cell proliferation assays.

REFERENCES

1. Stamenkovic. 1., Amiol. M., Pesando, J. M., and Seed. B. A lymphocyte moleculeimplicated in lymph node homing is a member of the cartilage link protein family.Cell, 56: 1057-1062, 1989.

2. Stamenkovic. I.. Aruffo. A., Amiot. M.. and Seed. B. The hematopoietic and epithelial forms of CD44 are distinct polypeptides with different adhesion potentials forhyaluronate-bearing cells. EMBO J., 10: 343-348. 1991.

3. Aruffo, A., Stamenkovic, I.. Melnick. M.. Underhill, C. B.. and Seed. B. CD44 is theprincipal cell surface receptor for hyaluronale. Cell. 61: 1303-1313, 1990.

4. Peach, R. J., Hollenbaugh. D., Stamenkovic, I. and Aruffo. A. Identification ofhyaluronic acid binding sites in the extracellular domain of CD44. J. Coll Biol., 122:257-264, 1993.

5. Jackson, D. G.. Buckley, J.. and Bell, J. 1. Multiple variants of the human lymphocytehoming receptor CD44 generated by insertions at a single site in Ihe extracellulardomain. J. Biol. Chem.. 267: 4732-4739. 1992.

6. Screaton, G. R., Bell, M. V., Bell, J. I., and Jackson. D. G. The identification of a newalternative exon with highly restricted tissue expression in transcripts encoding themouse Pgp-1 (CD44) homing receptor. J. Biol. Chem.. 17: 12235-12238, 1993.

7. Screaton. G. R., Bell, M. V.. Jackson, D. G., Cornells, F.. Gerth, U., and Bell J. I.Genomic structure of DNA encoding the lymphocyte homing receptor CD44 revealsat least 12 alternatively spliced cxons. Proc. Nati. Acad. Sci. USA, 89: 12160-12164,1992.

8. Naujokas. M. F., Morin, M., Anderson M. S.. Peterson, M.. and Miller. J. Thechondroitin sulfate form of invariant chain can enhance stimulation of T cell responses through interaction with CD44. Cell, 74: 257-268, 1993.

9. Seth, A., Goye, L., Nagarkatti, M.. and Nagarkatti, S. T cell receptor independentactivation of cytolytic activity of cytotoxic T lymphocytes mediated through CD44and gp9(lM" -|4 Proc. Nati. Acad. Sci. USA, 88: 7877-7881. 1991.

10. Gunthcrt, U., Hofman. M.. Rudy, W.. Reber, S., Zoller, M.. Haubmann. !.. Matzku,S.. Wenzel. A.. Ponía,H., and Herrlich, P. A new variant of glycoprotein CD44confers metastatic potential to rat carcinoma cells. Cell. 65: 13-24. 1991.

11. Hofman, M.. Rudy. W., Zoller. M.. Tolg, C.. Poma, H.. Herrlich, H., and Gunthcrt.U. CD44 splice variants confer metastatic behavior in rats: homologous sequences areexpressed in human tumor cell lines. Cancer Res.. 5/: 5292-5297. 1991.

12. Seiter, S.. Arch, R.. Reber, S., Komitowski, D., Hofman, M.. Ponta, H.. Herrlich. P.,Mutzku. S., and Zoller, M. Prevention of tumour metastasis formation by anti-variantCD44. J. Exp. Med., 177: 443-455, 1993.

13. Rudy. W.. Hofman, M., Schwartz-Albiez, R., Zollcr, M., Heider. K. H.. Ponta, H. and

Herrlich. P. The two major CD44 proteins expressed on a metastatic rat tumor cellline arc derived from different splice variants: each one individually suffices to confermetastatic behavior. Cancer Res., 53: 1262-1268, 1993.

14. Sy. M. S., Guo, Y. J., and Stamenkovic. I. Distinct effects of two CD44 isoforms ontumour growth in vivo. 3. Exp. Med., 174: 859-866, 1991.

15. Heider. K. H., Hofman, M., Hors, E., van den Berg, F., Ponta. H.. Herrlich, P., andPals, S. T. A human homologue of the rat metastasis-associated variant of CD44 is

expressed in colorectal carcinomas and adenomatous polyps. J. Cell Biol.. 120:623-627, 1992.

16. Koopman, G., Heider. K. H., Horst, E., Adolf, G. R., van den Berg. F., Ponta, H.,Herrlich. P.. and Pals. S. T. Activated human lymphocytes and aggressive non-Hodgkin's lymphomas express a homologue of the rat metastasis-associated variant

of CD44. J. Exp. Med., 177: 897-904, 1993.17. Salmi. M.. Gron-Virta. P.. Sointu. P., Grenman, R., Kalimo. H., and Jalkanen S.

Regulated expression of exon v6 containing isoforms of CD44 in man: downrcgula-

tion during malignant transformation of tumours of squamocellular origin. J. CellBiol., 122: 431-442, 1993.

18. Heider. K. H.. Dammrich. J., Skroch-Angel, P., Muller-Hermelink, H. K.. Vollmers.

H. P., Herrlich, P., and Ponta, H. Differential expression of CD44 splice variants inintestinal- and diffuse-type human gastric carcinomas and normal gastric mucosa.Cancer Res., 53: 4197-4203, 1993.

19. Mackay. C. R., Terpe, H. J., Stauder, R.. Marston. W., Stark, H.. and Gunthert, U.Expression and modulation of CD44 variant isoforms in humans. J. Cell Biol.. 124:71-82, 1994.

20. Dalchau, R., Flanagan, B. F., Allen, A. K., Daar. A. S., and Fabre. J. W. Chemicalcomposition and tissue distribution of the human CD44 glycoprotein. In: W. Knapp,B. Dorken, W. R., Gilks, E. P., Rieber, R. E., Schmidt, H., Stein, and A. E. G. K. vondem Borne (eds.). Leucocyte Typing IV, pp 622-625. Oxford, United Kingdom:

Oxford University Press, 1989.21. Simmons. D. L. Cloning cell surface molecules by transient expression in mammalian

cells. In: D. Hartley (ed). Cellular Interactions in Development, pp 93-127. Oxford.United Kingdom. Oxford University Press. 1993.

22. Fawcett. J.. Holness, C. L. L.. Needham. L. A., Turley, H., Gatter, K. C., Mason.D. Y., and Simmons, D. L. Molecular cloning of 1CAM-3. a third ligand for LFA-1,constitutively expressed on resting leukocytes. Nature (Lond.). 360: 481-484, 1992.

23. Simmons, D. L., and Seed. B. Isolation of a cDNA encoding CD33. a differentiationantigen of myeloid progenitor cells. J. Immunol., 141: 2797-2801), 1988.

24. Galfre, G., and Milstein, C. Preparation of monoclonal antibodies: strategics andprocedures. Methods Enzymol., 73: 3-46, 1981.

25. Harlow, E., and Lane, D. Antibodies: A Laboratory Manual, pp 363-419. ColdSpring Harbor. NY: Cold Spring Harbor Laboratory Press, 1988.

26. Kok. L. P.. and Boon, M. E. Microwave Cookbook for Microscopists: Art andScience of Visualisation. Leiden, the Netherlands: Coulomb Press, 1992.

27. Cattoretti. G., Becker, M. H. G., Key, G.. Duchrow, M.. Schlater, C., Galle, J., andGerdes, J. Monoclonal antibodies against recombinant parts of the KÌ67antigen(MiBI and MÌB3)detect proliferating cells in microwave processed formalin fixedsections. J. Pathol.. 168: 357-364, 1993.

28. Tanabe. K. K.. Ellis, L. M.. and Saya. H. Expression of CD44R1 adhesion moleculein colon carcinomas and metastasis. Lancet (Lond.). 341: 725-726. 1993.

29. Matsumura. Y., and Tarin, D. Significance of CD44 gene products for cancerdiagnosis and disease evaluation. Lancet (Lond.), 340: 1053-1058, 1992.

4546

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1994;54:4539-4546. Cancer Res   Stephen B. Fox, Jonathan Fawcett, David G. Jackson, et al.   Multiple Different CD44 IsoformsNormal Human Tissues, in Addition to Some Tumors, Express

  Updated version

  http://cancerres.aacrjournals.org/content/54/16/4539

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/54/16/4539To request permission to re-use all or part of this article, use this link

on May 15, 2018. © 1994 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from