Normal Keratinocytes Suppress Early Stages of Neoplastic … · characterize small numbers of cells with malignant potential in an environment of more normal cells. We have developed

(CANCER RESEARCH 58. 2200-2208. May 15. 1998|

Normal Keratinocytes Suppress Early Stages of Neoplastic Progression inStratified Epithelium1

Ashkan Javaherian, Michael Vaccariello, Norbert E. Fusenig, and Jonathan A. Garlick2

Department of Oral Biology and Pathology. School of Dental Medicine. Westchester Hall, SUNY at Stany Brook. Stony Brook, New York 11794-8702 Â¡A.J., M. V.. 3. A. G.Â¡.andGerman Cancer Research Center. Division of Differentiation and Carcinogenesis, D-69120 Heidelberg, Germany Â¡N.E. F.]

ABSTRACT

The importance of interactions between potentially neoplastic cells andtheir normal neighbors on malignant progression of precancerous lesionsis not well understood. In this study, we have established novel humantissue models that simulate intraepithelial neoplasia in stratified epitheliato investigate the fate and phenotype of neoplastic keratinocyte clones innormal cell context during clonal expansion and early malignant progression. This was accomplished by mixing genetically marked keratinocyteswith malignant potential (II-4) with normal keratinocytes at ratios of 1:1,4:1, 12:1, and 64:1 (normal:II-4) to visualize nests of marked, dysplastic

cells in organotypic cultures and in cultures transplanted to nude mice.Four weeks after transplantation of 4:1 mixtures, grafts were normal anddemonstrated no ÃŸ-galactosidase (ÃŸ-gal)-positive cells, suggesting that

cells with malignant potential were eliminated from the tissue at thismixing ratio. However, grafted 1:1 mixtures demonstrated persistence ofexpanded foci of dysplastic cells (4 weeks) and invasion (8 weeks). Thisdemonstrated that the capacity of a keratinocyte clone with neoplasticpotential to persist and invade is directly related to the threshold numberof such keratinocytes present in the tissue. To explain the failure of 11-4 to

persist in vivo, the intraepithelial dynamics between the two populationswere studied before grafting. Double-stain immunofluorescence for bro-modeoxyuridine/ÃŸ-gal and filaggrin/0-gal of mixtures grown in organo

typic cultures for 7 days demonstrated that when increasing numbers ofnormal cells were added (12:1), II-4 ceased to proliferate and expressed

filaggrin. This suggests a novel mechanism of tumor suppression whereincontact with normal cells induces cell cycle withdrawal and terminaldifferentiation of potentially malignant cells. These findings support theview that normal tissue architecture acts as a dominant suppressor ofearly neoplastic progression in stratified epithelium.

INTRODUCTION

Squamous cell cancer is initiated as a small nest of aberrant cellsexpands to dominate a tissue and form a macroscopic tumor. Duringearly neoplastic progression, preinvasive lesions demonstrate dysplastic foci that are initially surrounded by normal, undisturbed tissue (1,2). There is in vivo (3-5) and in vitro (6-10) evidence that normal

cells in the microenvironment can alter the neoplastic phenotype ofcarcinogen-initiated epithelium. These studies have proposed that

normal cells limit the progression of initiated cells so that only a smallfraction of potentially neoplastic cells express the neoplastic phenotype. However, mechanisms through which interactions with normalepithelial cells regulate the malignant potential, fate, and phenotype oftransformed epithelial cells have not been elucidated.

The role of normal cell context in controlling the growth of cellswith malignant potential has been difficult to study in vivo. Moststudies of in vivo carcinogenesis, including transgenic models, followthe progression of initiated cells that are surrounded by cells express-

Received 11/21/97; accepted 3/18/98.The costs of publication of this article were defrayed in pan by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1This work was supported by a grant from the National Institutes of Dental Research

(DE-11250-02).2 To whom requests for reprints should be addressed, at Department of Oral Biology

and Pathology. School of Dental Medicine, Westchester Hall, SUNY at Stony Brook.Stony Brook. NY 11794-8702. Fax: (516)632-9707; E-mail: [email protected].

ing the same phenotype (11). This does not accurately reflect the earlyprogression of spontaneous tumors in stratified epithelial tissues because cells with neoplastic potential may not be contacting normalcellular neighbors. Investigation of the role of cell interactions in earlyneoplastic progression therefore requires the capacity to detect andcharacterize small numbers of cells with malignant potential in anenvironment of more normal cells.

We have developed novel tissue models to study early neoplasticprogression in stratified squamous epithelium in which normal cellcontext is respected and cells with malignant potential are geneticallymarked to study their fate and phenotype. We have used organotypiccultures in which normal human keratinocytes are grown on a collagen matrix containing dermal fibroblasts at an air-liquid interface to

generate fully stratified squamous epithelia with a basal rate of proliferation ( 12). This culture system is optimal for studying cell interactions and is advantageous over monolayer cultures, where keratinocytes undergo limited differentiation and are hyperproliferative.The malignant cell line used in our studies (II-4) was derived by

transfection of the spontaneously immortalized human keratinocyteline HaCaT (13) with an activated c-Harvey-ra.v oncogene (14). These

cells have been shown previously to display severe dysplasia inorganotypic culture and low-grade malignant behavior after in vivo

transplantation (15). Upon grafting organotypic cultures to the dorsalsurface of nude mice, we have generated normal and malignant tissuesthat manifest their characteristic behavior in vivo. By geneticallymarking these potentially malignant keratinocytes and mixing themwith normal keratinocytes, we generated three dimensional tissuesthat simulate intrapeithelial neoplasia to study interactions betweenthese cell types during early neoplastic progression.

In this report, we study the role of epigenetic factors in the determination of the fate of cells with neoplastic potential and propose thatnormal tissue architecture acts as a dominant suppressor of earlyneoplastic progression in stratified epithelium. We demonstrate amechanism wherein potentially invasive keratinocyte clones are normalized by contact with surrounding normal keratinocytes and areeliminated from the tissue. The size of such clones is crucial indetermining their ability to expand and invade, supporting a requirement for a high critical number of neoplastic cells to overcome growthsuppression. This novel mechanism of tumor suppression was theinduction of cell cycle withdrawal and terminal differentiation ofpotentially malignant cells by contact with normal cells and maintenance of normal tissue organization. This suggests that the signalingnetwork inherent in cell interactions in stratified epithelia may beeffective in tumor control. Malignant behavior is, therefore, controlledby tissue architecture, and tissue phenotype can predominate overcellular genotype in early neoplastic progression.

MATERIALS AND METHODS

Cell Culture. Human epidermal keratinocytes were cultured from newbornforeskin by the method of Rheinwald and Green ( 16) in keratinocyte mediumdescribed by Wu et al. (17). Cultures were established through trypsinizationof foreskin fragments and grown on irradiated 3T3 fibroblasts. 3T3 cells weremaintained in DMEM containing 10% bovine calf serum. H-4 keratinocyteswere grown in DMEM containing 5% FCS. Organotypic cultures were pre-

2200

Research. on October 31, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

NORMAL KERATINOCYTES SUPPRESS Nl.ol'l ASTIC PRÃ•XÃ•RESSION

pared by adapting the protocol developed by OrganogÃ©nesis. Inc. ( 18). Briefly,early-passage, human dermal fibroblasts were added to neutralized type Icollagen (OrganogÃ©nesis, Canton. MA) to a final concentration of 2.5 X IO4

cells/ml. Three ml of this mixture were added to each 35-mm well of a 6-well

plate and incubated for 4 to 6 days, until the collagen matrix showed no furthershrinkage. At this time, a total of 5 x 10* keratinocytes, II-4-gal3 cells, or

mixtures of these two cell types were plated on the contracted collagen gel.Cultures were maintained submerged in low calcium EGM (OrganogÃ©nesis)for 2 days, submerged for 2 days in normal calcium EGM. and raised to theair-liquid interface by feeding from below with normal calcium EGM for 3

days. For proliferation assays. BrdUrd (Sigma Chemical Co., St. Louis. MO)was added to organotypic cultures at a final concentration of 10 ^M andincubated for 8 h prior to harvesting.

Retroviral Vectors and Transduction of Keratinocytes. The MFG-galvector is a Moloney murine leukemia virus-based vector that contains the genefor bacterial ÃŸ-galactosidase (19). Transduction of II-4 cells with this vector

was performed as described previously (20). Briefly. 11-4cells were transduced24 h after plating I X 10" cells in a 10-cm dish using fresh, filtered (0.45 urn:

Gelman, Ann Arbor. Michigan) supernatant from confluent amphotropic producer cells making the MFG-gal vector. Transduced keratinocytes were pas

saged at clonal density, and clones were screened for persistence of transgeneexpression after three passages. Only clones maintaining transgene expressionin 100% of cells were expanded and used for organotypic culture and graftingexperiments. No deleterious effect on cell growth or phenotype were seen aftertransduction. Producer lines were maintained in 10% bovine calf serum.

Transplantation of Organotypic Cultures to Nude Mice. Organotypiccultures were trimmed using a surgical punch 1.4 cm in diameter. Six-week-old male Swiss nude mice (N:NIHS-nuf DF; Taconic Farms. Germantown,

NY) were anesthetized using Xylazine:Ketamine (1:1). and 1.3 cm of dorsalskin was removed. Organotypic cultures were placed onto this area and werecovered with petrolatum gauze (Sherwood Pharmaceuticals. St. Louis. MO)and secured with bandages (Baxter Scientific). These dressings were changedafter 7 days and removed completely after 14 days. Animals were sacrificed at

4 and 8 weeks after transplantion.Immunofluorescence. Specimens were frozen in embedding media (Tri

angle BiomÃ©dical.Durham, NC) in liquid nitrogen vapors after being placed in2 M sucrose for 2 h at 4Â°C.Tissues were serial sectioned at 6 /xm and mounted

onto gelatin-chrome, alum-coated slides. Tissue sections were washed with

PBS and blocked with 10 Â¿ig/mlgoat IgG. 0.05% goat serum, and 0.2% BSA(v/v) in PBS without fixation. Sections were incubated with rabbit polyclonalantiserum to bacterial ÃŸ-galactosidase (Cortex Pharmaceuticals. San Leandro.CA) and detected with FITC-conjugated goat anti-rabbit IgG (Vector Labora

tories, Burlingame. CA). Double stain immunofluorcscence was performed bydouble-staining for /3-gal detection and by using either a monoclonal antibody

to BrdUrd (Boehringer Mannheim. Indianapolis. IN) or to human filaggrin(Biosearch Technologies, San Rafael, CA). which were detected with TexasRed-conjugated horse anti-mouse IgG (Vector Laboratories). Slides were

coverslipped with Vectashield containing 1 (xg/ml DAPI (Vector Laboratories). Fluorescence was visualized using a Nikon OptiPhot microscope, anddouble exposure photomicroscopy was performed using FITC and Texas Red.For routine light microscopy, tissues were fixed in 10% neutral bufferedformalin and embedded in paraffin, and 4-^irn sections were stained with H&E.

RESULTS

Characterization of Normal and Neoplastic KeratinocyteTransplants. Figs. 1 and 2 describe experimentsdesigned to determine the behavior and fate of cells with malignant potential (II-4-gal)after mixing with normal keratinocytes and grafting to nude mice.Grafts of II-4-gal cells generated a lesion with an uneven, verrucoussurface 8 weeks after grafting (Fig. 1A).H&E-stained sections of thislesion revealed a dysplastic epithelium with focal invasion into theunderlying connective tissue 4 weeks after grafting (Fig. 2/4). Thisepithelium demonstrated basal cell hyperplasia, cells with an alterednuclear to cytoplasmic ratio, suprabasal mitoses, altered tissue orga-

1The abbreviations used are: gal, galactosidase; EGM. epidermal growth medium;

BrdUrd. bromodeoxyuridinc.

nization, hyperparakeratosis. and elongated rete pegs. Immuno-staining for ÃŸ-galexpression showed invasion of both single cells andcell clusters (Fig. 3/1,arrows). By 8 weeks after grafting, II-4-gal cellsdemonstrated islands of malignant epithelium in the connective tissueand severe dysplasia of the surface epithelium (Figs. 2B and 3fl,arrows). These cells therefore generated invasive squamous cell carcinoma with a low-grade biological behavior after transplantation.Grafts showed persistent transgene expression in all II-4 cells, demonstrating that inactivation of the transgene did not occur in vivo.

Grafts of normal foreskin keratinocytes were clinically normal (Fig.IÃŸ)and showed a hyperkeratotic epithelium 8 weeks after grafting(Fig. 2D). No /3-gal expression was seen in normal human keratino-cyte grafts (Fig. 3D). In all grafts, adjacent mouse tissue could bedistinguished from grafted human cells on a morphological basis andwas confirmed by immunostaining with a species-specific antibodyfor human involucrin (data not shown).

Expansion and Invasion Require a Critical Number of Neoplastic Keratinocytes. Whennormalcells outnumberedII-4-galcells incellmixtures (4:1), the graft was clinically (Fig. 1C) and morphologicallynormal (Fig. 2C) and demonstrated no ÃŸ-gal-positivecells (Fig. 3C) asearly as four weeks after grafting. This suggested that cells with malignant potential were eliminated from the tissue at this 4:1 mixing ratio.II-4-gal cells were only detected in 1:1 mixtures after grafting. After 4weeks, these transplantsgenerated a tissue with morphologicallydistinctfoci of dysplastic cells, which were surrounded by normal appearingepithelium (Fig. 2Â£,arrows). Immunohistochemistrydemonstrated thatthese foci were II-4-gal cells that persisted and extended from the basalto upper layers of the epithelium (Fig. 3Â£).

Eight weeks after transplantation, these grafts showed an irregularsurface covered with a thin scale in the center and more exophyticregions adjacent to mouse tissue (Fig. ID). At this time, 1:1 mixturesdemonstrated larger foci of dysplastic cells and cells invading into theunderlying connective tissue (Fig. IF, arrow). Staining for /3-galshowed continued expansion and invasion of II-4-gal cells (Fig. 3F).These results suggested that a relatively high, critical number of cellswith malignant potential needed to be present for these cells to persist,undergo clonal expansion, and form a focally dysplastic and earlyinvasive tumor in vivo.

Dominant Suppression of Neoplastic Keratinocytes by Interactions with Normal Cells. Basedon the observationthatII-4-galcellsfailed to persist when a greater number of normal cells were presentin the mixture, the fate and distribution of cells with malignantpotential was studied after growth in organotypic culture. If cells werebeing lost, perhaps some of these changes were resulting from theintraepithelial dynamics before grafting. One-week cultures of II-4-gal cells demonstrated a poorly differentiated, dysplastic epithelium(Fig. 4Â«),whereas cultures of normal keratinocytes showed a well-differentiated, stratified epithelium (Fig. 4c). Mixtures demonstrateddistinct foci of altered cells in the context of normal cells, therebysimulating intraepithelial neoplasia (Fig. 4). As increasing numbers ofnormal cells were added to the mixtures, tissues showed fewer foci ofdysplasia, more normalized tissue architecture, the presence of astratum granulosum, and thickened stratum corneum (Fig. 4, e, g, i,and k). These cultures demonstrated small clusters and individualdysplastic cells, which were more commonly seen when more II-4-galwas present in the tissue constructs.

Immunostaining for ÃŸ-galexpression verified that these dysplasticfoci were II-4-gal, which were sorted to a position above the basallayer. With increasing numbers of II-4-gal cells (1:1), larger supra-basal clusters were seen, some of which extended to the basal layer(Fig. 4/). ÃŸ-galpositivity in 64:1 mixtures was seen only as singlecells surrounded by normal keratinocytes (Fig. 41), whereas 12:1 and4:1 mixtures demonstrated only suprabasal clusters (Fig. 4, h and j).

2201



NORMAL KERATINOCYTES SUPPRESS NEOPLASTIC PROGRESSION

Fig. 1. Clinical appearance Ã¶ltransplants 8 weeks alter graltmg to nude mice. Seven-day organolypic cultures were transplanted to mice. A, 11-4-gal gralt demonstrating verrucous,exophytic surface; B. normal human keratinocyte graft showing distinct borders with adjacent mouse skin: C. normal keratinocytes and II-4-gal mixed in a ratio of 4:1 (normal:II-4-gal)demonstrating a smooth surface: D, normal keratinocytes and II-4-gal mixed in a ratio of 1:1 (normal:II-4-gall showing a scale in center and pebbly surface near margin with mouseskin. Grafts were repeated three limes, and representative grafts are shown.

All cells were ÃŸ-galpositive in pure cultures of II-4-gal (Fig. 4b), andno ÃŸ-galexpression was detected in control cultures of normal kera

tinocytes (Fig. 4d).It was then studied whether cells with malignant potential were

undergoing changes in their biological behavior after mixing thatresulted in their subsequent loss from the tissue after grafting. Tissuedynamics were studied after growth for 1 week in organotypic cultureby assessing the proliferation and differentiation of II-4-gal aftermixing. Double immunofluorescent staining for ÃŸ-galand BrdUrd

demonstrated many proliferating basal and suprabasal cells in pureII-4-gal cultures (Fig. 5A), whereas normal keratinocytes showed

proliferation limited to the basal layer (Fig. 5ÃŸ).Double immunofluorescent staining for ÃŸ-galand BrdUrd of 12:1 cultures demonstrated

that proliferation was limited to normal basal keratinocytes in apattern similar to that seen in control cultures of normal keratinocytes(Fig. 5C). In addition, individual ÃŸ-gal-positive cells were seen su-prabasally without BrdUrd label, suggesting that II-4-gal proliferation

was limited when surrounded by normal cells. This showed that whenrelatively low numbers of II-4-gal were added to the tissue, they

ceased to proliferate. In contrast, 1:1 mixtures demonstrated numerous, suprabasal ÃŸ-gal-positive clusters that were BrdUrd positive and

continued to proliferate and expand (Fig. 5D). This demonstrated thatclusters containing larger numbers of II-4-gal cells continued toproliferate. It appeared that when the number of contiguous II-4-gal

cells was sufficiently high, cell proliferation was not affected, becausemost larger clusters were contacting other II-4-gal cells.

It was next determined whether II-4-gal cells were undergoing

terminal differentiation, as determined by the presence of filaggrinwhen surrounded by normal keratinocytes. Double immunofluores-cence of pure cultures of II-4-gal did not demonstrate filaggrin ex

pression (Fig. 5Â£),whereas normal keratinocyte cultures expressedthis marker of terminal differentiation in the upper third of theepithelium (Fig. 5F). However, when mixed with normal keratinocytes in a 12:1 ratio, II-4-gal cells underwent terminal differentiationand expressed filaggrin, as seen by Â«Â»localizationof ÃŸ-galand filaggrin (Fig. 5G). Such expression was only seen when II-4-gal cells

were surrounded by normal keratinocytes that were expressing filaggrin and not in individual II-4-gal cells in contact with normal kera

tinocytes that did not express filaggrin (Fig. 5C, long arrow). Thissuggested that II-4-gal cells were being normalized by keratinocytesthat were undergoing terminal differentiation. In distinction, II-4-gal

in cultures grown with a 1:1 ratio were not induced to expressfilaggrin by normal cells in their microenvironment (Fig. 5//).

DISCUSSION

A Novel Model for Early Neoplastic Progression. Althoughmuch is known about the effect of genetic changes in individual cellson the early development of skin cancer, the importance of interactions of such cells with normal neighbors on the regulation of incipient neoplasia is less clear. Because early neoplasia is expressed at thetissue level in the context of normal cells, study of the role of cellinteractions in early tumor progression requires models that respectcell context and tissue organization. We established organotypic cul-

2202



NORMAL KF.RATINOCYTKS SUPPRESS NI-OP1.ASTK1 I'R<KM SSII IN

Fig. 2. Light microscopy of transplants 4 (A. C. and Â£")and 8 weeks (B. /). and F\ after grafting to nude mice. A, grati oÃII-4-g.il demonstrating moderate to se\ere dysplasia. epithelial

thickening. luperorthokeratosis. and tongue-like extension ol epithelium into the underlung connective tissue itirrou }.H. graft of ll-4-gal showing islands of malignant epithelium (wiiit-arrows'! and severe dysplasia of the surface epithelium (thin arrows). C, graft of 4:1 mixture of normal and II-4-gal 4 weeks after grafting; arrow marks horder with mouse tissue. I).

8-week graft of normal cells showing normal epithelium with hyperorthokeratosis. Arrow marks horder with mouse tissue, and asterisk marks mouse skin appendage. K, graft of 1:1mixture of normal and II-4-gal. demonstrating a central cluster of dysplastic cells demarcated by arrows. F. graft of I: I mixture of normal and ll-4-gal demonstrating islands of invasiveepithelium (Â«mm-)beneath a dysplastic surface epithelium.

ture and in vivo transplantation systems where both normal andtransformed keratinocytes manifest their characteristic morphologyand behavior. In monolayer culture conditions, keratinocytes arehyperproliferative and undergo limited differentiation, whereas orga-

notypic culture closely mimics the in vivo tissue (21), thereby facilitating study of interactions between clones of potentially malignantcells and surrounding normal cells. The malignant cell line used in ourstudies (HaCaT-ra.Ã§clone II-4) was derived by transfection of the

immortalized human keratinocyte line HaCaT (13) with an activatedc-Harvey-ra.c oncogene ( 14). This tumorigenic cell line has been well

characterized previously and displays severe dysplasia in vitro andlow-grade malignant behavior after in vivo transplantation (15). By

genetically marking these clones, we were able to follow the fate andphenotype of dysplastic foci during clonal expansion and early invasion. Grafting these cells with restoration of an invasive phenotype

allowed study of the earliest events in neoplastic progression bygenerating a stratified epithelium that evolved from a noninvasive,focally dysplastic tissue to one demonstrating invasion in vivo. Thesefeatures provided an advantage over previous culture and graftingmodels used to study interactions between normal and potentiallymalignant epithelia because interactions between normal and neoplastic cells are studied in a regular, preformed tissue architecture.

Normal Keratinocytes Are Dominant Suppressors of NeoplasticPotential. Our results demonstrate that normal cell context and tissuearchitecture exert strong growth-inhibitory and normalizing effects onra.v-transformed keratinocytes with malignant potential. This suggeststhat the signaling network inherent in cell-cell interactions plays an

important role in tumor control. Our studies extend previous findingson the regulatory role of normal cells on neoplastic progression. It wasshown previously that normal mouse keratinocytes suppressed the

2203



NORMAL KERATINOCYTES SUPPRESS NROPLASTIC PROGRESSION

Fig. 3. Immunofluorcsccm staining of transplants 4 (A. C. and E} and 8 (B. D. and F] weeks after grafting. A. graft of Rll-gal showing expression of transgene in all cells and earlystromal invasion (ÃÃmm'.vl.H. graft of Rll-gal. demonstrating islands of invasive epithelium (Â«mnt'.v).D, graft of normal keratinocytes. C. graft of 4:1 mixture shows no II-4 cellsremaining. Â£.graft of 1:1 mixture showing persistent foci of II-4 cells (arrows}, all of which extend to the basal layer. F, graft of 1:1 mixture showing intracpithelial clusters (longttrriiH'i and invasive clusters in the slroma (*ht)rt amwi. Blue nuclear fluorescence due to counterstain with 3.3'-diaminoben/.idine (Vector Laboratories).

growth ot"papillomas in athymic mice after being mixed with initiated

keratinocytes (3) or mouse papilloma cells (22). In addition, normaltrachÃ©alepithelium demonstrated a selective survival advantage whenmixed with initiated trachÃ©alcells and transplanted to syngenic rats(5). For neoplastic progression to occur, potentially malignant cellsmay first need to overcome a restrictive environment of normal cellcontext that limits their growth and prevents clonal persistence andexpansion.

We report for the first time a mechanism for such suppression, i.e.,that transformed clones with malignant potential cease to proliferate,undergo terminal differentiation, and are eliminated from the tissuedue to interactions with their normal cellular neighbors. The suppres-

sive effect of the microenvironment may be due to several types ofintercellular communication, including the presence of diffusible,paracrine-acting mitotic inhibitors or to direct cell-cell contact (23-

26). For example, the clonal growth of a small number of initiated

mouse keratinocytes was suppressed by coculture with differentiated,normal keratinocytes (8), and the growth of carcinogen-altered rat

trachÃ©alepithelium was limited by conditioned media from normalcells (9). In addition, transforming growth factor ÃŸfrom adjacentfibroblasts can suppress the tumorigenic potential of ra.v-transformedmouse keratinocytes (26). whereas the normal keratinocyte-inducedsuppression of papilloma formation can be overcome by overexpres-

sion of transforming growth factor a (22).In the above studies, the population of cells with malignant poten

tial was not directly identified, as has been shown in our report.Through colocalization of ÃŸ-galexpression with markers of prolifer

ation and differentiation, we have characterized the phenotype ofneoplastic cells undergoing suppression. By following the fate ofthese cells, we found that proliferation of tumor cells is abrogatedwhen isolated, malignant cells are outnumbered and surrounded bynormal keratinocytes. We propose this as a mechanism of tissue-based

2204



.z@-

@ @IIIQ

- -@

@.* -

u@I 0

@flI

c'ii 0

@-@

NORMAL KERATINOCYTESSUPPRESS NEOPLASTIC PROGRESSION

.,.@ .

- .@ .@

.@ @â€˜:@

.@@

.. ______

V..@@@@ .uA. 0

,â€” .,, .@ .- ,â€” . â€”.-@ @-. -@

- . â€˜ . . @..

.@ , ..-.

%.$t -

a

C

ll-4-( .\l

N I IK

@ @,@

@ ,.@ .Aw@@ â€˜a

@@ .â€”@--@

:@@@ @;i:, @.â€˜, is.@ â€¢ :

â€”. S. @) â€¢,,@ -,

% .. .@ . .â€˜@

. . . , (U'

, .@@@ e. @. @.

. -

4@I â€” â€”

.-

@ @?!

. 0III

Fig. 4. Organotypic cultures of mixed 11-4and normal keratinocytes demonstrate foci of genetically marked dysplastic cells innormal cell context. Cultures were grown for 1 week in organotypicculture either as pure cultures or as mixtures of varying ratios[(normal:ll-4-Gal) H&E, a, c, e, g. i, and k, and immunofluorescentstain for a-gal, b, d, f h, j, and 1]. 11-4-GAl.,pure culture ofHaCaT.ras-ll GAL cells; NHK, normal keratinocytes. a and b,control cultures of neoplastic cells (11-4-GAL); c and d, controlculturesof normalforeskinkeratinocytes(NHK);e and f@1:1mixture; g and h, 4:1 mixture; i and j, 12:1 mixture; k and I, 64:1mixture. Arrows, foci of dysplastic cells.

I

@ , â€˜@@

I 2.1

() I : Ik

growth control during early neoplastic progression because local cellinteractions lead to normalization and elimination of sufficiently smallnumbers of potentially malignant cells. However, caution should beexercised in extrapolating the finding of suppression of a geneticallyaltered, immortalized keratinocyte line such as 11-4 to the broader

study of tumor development in vivo, because it has been shownpreviously that small clones of mutated cells can persist in carcinogen-altered skin (27). It appears that the capacity of a potentiallyneoplastic cell to progress is associated with a more advanced stage ofphenotypic and genotypic alteration. Gillett et a!. (4) reported that a

2205

- -@@

@-

.-:â€˜.,@@



A

.E .. ..@-@BFC

.

[email protected].@0

., @-. CH

*


Fig. 5. Double-immunofluorescence stain of organotypic cultures for markers of proliferation and differentiation. A, B, C, and D, superimposed immunofluorescence signalsfor @3-galactosidase(FITC channel, green) and BrdUrd (Texas Red channel, red). E, F, G, and H, superimposed immunofluorescence signals for 13-gal(FITC channel, green)and filaggrin (Texas Red channel). A. pure 11-4 cultures demonstrating BrdUrd-positive nuclei in basal and suprabasal layers that extend to mid-spinous layer (arrow). B, normalkeratinocyte culture demonstrating BrdUrd-positive nuclei limited to the basal layer and no f3-gal expression. C, mixture of normal keratinocytes:II-4 (12: 1) demonstratingindividual @3-galcells in a suprabasal position and BrdUrd-positive nuclei limited to the normal basal keratinocytes. Individual a-gal cells lack colocalization of a-gal, andBrdUrd demonstrates withdrawal of 11-4cells from cell cycle. D, mixture of normal keratinocytes:lI-4 (1:1) showing @-ga1-positive11-4cells that are also BrdUrd positive(arrow), as seen by red nuclei and green cytoplasm. This suggests continued 11-4 proliferation in these larger clusters. E, pure 11-4 cultures demonstrating that all 11-4 cellsexpress /3-gal but not filaggrin, because no yellow is seen. F. normal keratinocyte culture demonstrating the presence of filaggrin in the superficial layers of the epitheliumand no f3-gal expression. The dermal-epithelial interface is marked with the white dotted line. G. mixture of normal keratinocytes:Il-4 (12:1) demonstrating yellow-orangesuprabasal cells (short arrow) resulting from superposition of signals for /3-gal and filaggrin. Long arrows, individual suprabasal j3-gal cells that have yet to migrate tosuperficial layers and are not expressing filaggrin. H. mixture of normal keratinocytes:II-4 (1 : 1) showing larger clusters of a-gal cells that are not expressing filaggrin. Redstaining below the epithelium represents proliferating fibroblasts.

2206




tumorigenic trachÃ©alcell line persisted in the tissue when mixed withnormal cells, whereas a nontumorigenic, preneoplastic cell line didnot. Furthermore, tumorigenicity of ras-transformed mouse keratino-

cytes has been shown to be suppressed by diffusible factors producedby dermal fibroblasts, but additional genetic events allow keratino-

cytes to escape this inhibition, allowing skin tumors to develop (26).Because II-4 keratinocytes harbor an activated ras oncogene, which is

known to be involved in the very early stages of neoplastic transformation in stratified epithelium (28), it is possible that only keratinocytes in the initial stages of carcinogenesis can still be suppressed bythe microenvironment. To explore whether keratinocytes with additional genetic alterations are more likely to overcome the inhibitoryeffects of the microenvironment, keratinocyte lines at various stagesof neoplastic progression will need to be tested in the tissue modelsdescribed in this report.

Requirement for Critical Number of Neoplastic Cells for Expansion and Invasion. Neoplastic progression may occur only if therestrictive microenvironment is modified to allow proliferation andexpansion of initiated cells. When groups of tumor cells are present at

numbers allowing the formation of foci of sufficient size, they maintain their ability to proliferate, even in suprabasal position, and areresistant to induction of differentiation. We determined that equalnumbers of premalignant and normal cells were required for earlyprogression and lesion development, suggesting that the spatial dis-

peared to compete for attachment to extracellular matrix more effectively than II-4 cells in the 64:1, 12:1, and 4:1 ratios, resulting in thefailure of II-4 cells to persist in a basal position. In previous in vivo

studies, when trachea! grafts were seeded with cell mixtures so thatlarge, contiguous areas of trachea were populated with carcinogen-

initiated cells, no growth inhibition was seen, and tumors developed(5). Our findings suggest that attachment to basement membrane maybe crucial to preserving early tumor cell phenotype.

In summary, invasive neoplasia in stratified squamous epitheliummay develop only when changes in tissue architecture are sufficient toovercome the dominant, tumor suppressor role of normal cell context.Maintenance of normal tissue architecture is sufficient to abrogateearly neoplastic progression by repressing the malignant phenotype.These findings have significant biological implications for understanding how precancer is controlled within a stratified epithelium.Further insights into the inhibitory signaling pathways inherent in cellinteractions may lead to the development of cancer prevention strategies designed to redirect potentially malignant cells toward normaldevelopment.

ACKNOWLEDGMENTS

We thank N. Parenteau, J. Teumer. and S. Proske of OrganogÃ©nesis, Inc. fortechnical assistance with organotypic cultures and grafting; L. Taichman,

tribution of II-4 cells and the degree to which normal tissue architec- R. A. F. Clark. I. Spector. and N. Markova for critical discussions; K. Galron

ture is disturbed determines the ability to evade suppression. Inmixtures, most II-4 cells present in large clusters and in contact withother II-4 cells were not inhibited by normal cells because the numberof contiguous II-4 cells was sufficiently large. As a result, cell cyclewithdrawal and differentiation did not occur, and II-4 cells persisted

and invaded when grafted. Our inability to detect genetically markedII-4 cells in both short- and long-term grafts in 4:1 mixtures and the

normal clinical appearance of these grafts suggested elimination ofras-transformed keratinocytes rather than persistence of latent or"dormant" tumor cells. This also suggested that direct contact of the

whole or most of the tumor cell surface with normal cells is needed forsuppression because growth inhibition and induction of differentiationwas not observed in large II-4 clusters in organotypic culture, in spite

of the presence of normal cells in the periphery.The size of an initiated clone is apparently crucial in determining its

survival potential during development of early neoplasia. Progressionto malignancy may therefore require the presence of a number ofpotentially malignant cells that is above a critical threshold. Thisresults in a critical clone size in which larger numbers of aberrant cellsinteract, resulting in escape from growth suppression. Contact between tumor cells has been shown to result in suppression of a deathsignal (29), whereas inhibition of contact between carcinoma cellsresults in elimination of these cells through induction of apoptosis(30). Such intercellular growth control may be due to the presence ofmembrane channels in gap junctions that transmit growth-regulatory

signals between cells, because increased gap junction communicationcan normalize the growth of transformed cells (31-33). TumorigenicHaCaT-ras clones have been shown to exhibit significantly reduced

gap junction intercellular communication (34), and this may partiallyexplain why large clusters of these cells have a reduced sensitivity togrowth-controlling influences of normal cells. Additional studies areneeded to determine the importance of cell-cell communication on

differentiation and proliferation of cells with malignant potential.In addition, it appears that the capacity of the 11-4cells to attach to

basement membrane and persist in a basal position may preserve theinvasive potential of these cells after grafting. As seen in Figs. 3, 4,and 5, cells grown in a 1:1 ratio persisted in a basal position andinvaded after transplantation. On the other hand, normal cells ap-

and R. Harrington for technical assistance; M. Simon for providing antiserumagainst human involucrin; R. Mulligan for the MFC retroviral vector: and J.Neville and C. Dowling for artwork.

REFERENCES

1. Farber. E. The step-by-step development of epithelial cancer: from phenotype togenotype. In: Advances in Cancer Research, pp. 21-48. London: Academic Press,

Inc.. 1996.2. Weinberg, R. A. Oncogenes. tumor suppressor genes, and cell transformation: trying

to put it all together. In: J. Brugge. T. Curran. E. Harlow. and F. McCormick (eds.).Origins of Human Cancer: A Comprehensive Review, pp. 1-16. Cold Spring Harbor.NY: Cold Spring Harbor Laboratory. 1991.

3. Strickland. J. E.. Ueda, M.. Hennings. H., and Yuspa, S. H. A model for initiatedmouse skin: suppression of papilloma but carcinoma formation by normal epidermalcells in grafts on alhymic nude mice. Cancer Res.. 52: 1439-1444. 1992.

4. Gillett. N. A.. Thomassen, D. G.. Buchholz. A., and Kelly. G. Cell-line-dependenlinhibition of phenotypic expression of preneoplastic and neoplastic trachÃ©alepithelialcells by normal trachÃ©alcells in denuded trachÃ©altransplants. Carcinogenesis (Lond.l,10: 1147-1154. 1989.

5. Terzaghi-Howe. M. Inhibition of carcinogen-altered rat trachÃ©alepithelial cell proliferation by normal epithelial cells in vivo. Carcinogenesis (Lond.). 8: 145-150,

1987.6. Hennings, H.. Lowry, D. T.. Robinson. V. A.. Morgan. D. L.. Fujiki. H., and Yuspa.

S. H. Activity of diverse tumor promoters in a keratinocyte co-culture model ofinitiated epidermis. Mol. Carcinog., 13: 2145-2151, 1992.

7. Hennings. H.. Lowry. D. T.. and Robinson, V. A. Coculture of neoplastic and normalkeratinocytes as a model to study tumor promotion. Skin Pharmacol.. 4: 79- 84. 1991.

8. Hennings, H.. Robinson. V. A.. Michael. D. M.. Petti!. G. R.. Jung. R., and Yuspa.S. H. Development of an in vitro analogue of initiated mouse epidermis to studypromoters and antipromoters. Cancer Res., 50: 4794-4800. 1990.

9. Terzaghi-Howe. M., and McKeown. C. Inhibition of carcinogen-altered rat trachÃ©alepithelial cells by normal cell-conditioned medium. Cancer Res., 4f>:917-921, 1986.

10. Terzaghi-Howe. M. Inhibitor production by normal rat trachÃ©alepithelial cells influences the frequency of spontaneous and x-ray-induced enhanced growth variants.Carcinogenesis (Lond.), 10: 967-971, 1989.

11. Adams, J. M.. and Corey, S. Transgenic models of tumor development. Science(Washington DC). 254: 1161-1166. 1991.

12. Garlick. J. A., and Taichman. L. B. Fate of human keratinocytes during reepitheli-alization in an organotypic culture model. Lab. Invest., 70: 916-924, 1994.

13. Boukamp. P.. Petrussevka. R. T., Breitkreutz, D.. Hornung. J.. Markham. A., andFusenig, N. E. Normal keratinization in a spontaneously immortalized aneuploidhuman keratinocyte cell line. J. Cell Biol.. 106: 761-771. 1988.

14. Boukamp. P.. Stanbridge, E. J., Yin-Foo. D., Cerutti. P. A., and Fusenig. N. E.c-Ha-rn.5 oncogene expression in immortalized human keratinocytes (HaCaTl altersgrowth potential in vivo but lacks correlation with malignancy. Cancer Res., 50:2840-2847, 1990.

15. Fusenig, N. E., Boukamp, P., Breitkreutz. D.. HÃ¼lsen.A.. Petrusevska. S., Cerutti, P..and Stanbridge. E. In \-ilrt> transformation of the human skin epithelial cells: role ofras oncogene in malignant progression. Toxic. In Vitro. 4: 627-634. 1990.

2207



NORMAL KERATINOCYTES SUPPRESS NEOPLASTIC PKCH.KI SS1ON

In. Khcinwald. J. G., and Green, H. Serial cultivation of strains of human epidermalkeratinocytes: the formation of keratinocyte colonies from single cells. Cell. 6:331-344, 1975.

17. Wu, Y-J., Parker. L. M., Binder, N. E.. Beckett. M. A.. Sinard. J. H., Griffiths, C. T..and Rheinwald. J. G. The mesothelial keratins: a new family of cyloskeletal proteinsidentified in cultured mesolhelial cells nonkcratiniiing epithelia. Cell, 31: 693-703.

1982.18. Parenleau. N. L. Skin equivalents. In: I. M. Leigh and F. M Watt (eds.). The

Keratinocyte Handbook, pp. 45-56. Cambridge: Cambridge University Press, 1994.

19. Dranoff, G., Jaffee. E.. Lazenby. A.. Golumbek. P.. Levitsky. H.. Brose. K., Jackson.V.. llamada H.. Pardoll. D.. and Mulligan, R. C. Vaccination with irradiated tumorcells engineered to secrete murine granulocyte-macrophage colony-stimulating factorstimulates potent, specific, and long-lasting anti-tumor immunity. Proc. Nail. Acad.Sci. USA. 90: 3539-3543. 1993.

20. Garlick, J. A. Retroviral vectors. In: I. M. Leigh and F. M. Watt (eds.). TheKeratinocyte Handbook, pp. 181-183. Cambridge: Cambridge University Press.

1994.21. Parcmeau. N. L.. Bilbo. P.. Nolle. C. J.. Mason. V. S.. and Rosenberg. M. The

organotypic culture of human skin keralinocvtes and fibroblasts to achieve form andfunction. Cytolechnology, 9: 163-171. 1992.

22. Pinzi, E., Kilkenny, A.. Strickland. J. E.. Balaschak. M.. Bringman. T.. Derynck. R..Aaronson. S.. and Yuspa. S. H. TGFa stimulates growth of skin papillomas byautocrine and paracrine mechanisms but does not cause neoplastic progression. Mol.Carcinog.. /: 7-12. 1988.

23. Potter, V. Initiation and promotion in cancer formation: Ihe importance of studies onintercellular communication. Yale J. Biol. Med., 53: 367-384. 1980.

24. Osgood, E. E. A unifying concept of the etiology of the leukemias. lymphomas. andcancers. J. Nail. Cancer Inst.. Ili: 155-166. 1957.

25. Bell. G. I. Models of carcinogenesis as an escape from milotic inhibitors. Science(Washington DC). 192: 569-572. 1976.

26. Missero, C.. Ramon. Y., Cajal. S., and Dotto, G. P. Escape from transforming growthfactor j3 control and oncogene cooperation in skin tumor development. Proc. Nati.Acad. Sci. USA. KX:9613-9617. 1991.

27. Jonason, A. S., Kunala, S., Price. G. J., Restifo. R. J.. Spinelli. H. M.. Persing, J. A.,Leffell. D. J., Tarone. R. E.. and Brash. D. E. Frequent clones of p53-mutatedkeratinocytes in normal human skin. Proc. Nati. Acad. Sci. USA. 93: 14205-14209.

1996.28. Balmain. A.. Ramsden M.. Bowden. G. T.. and Smith. J. Activation of the mouse

cellular llarvey-ru.v gene in chemically induced benign skin papillomas. Nature(Lond.). 307: 658-660. 1984.

29. Kataoka. S., Naito. M.. Fujita. N.. Ishii. H.. Ishii. S., Yamori. T., Nakajima, M., andTsuruo. T. Control of apoptosis and growth of malignant T lymphoma cells by lymphnode stromal cells. Exp. Cell Res.. 207. 271-276. 1993.

30. Bates. R. C.. Buret. A., van Helden. D. F.. Horion. M. A., and Bums. G. F. Apoptosisinduced by inhibition of intercellular contact. J. Cell Biol., 125: 403-415. 1994.

31. Kumar. N. M., and Gilula. N. B. The gap junction communication channel. Cell. <V-/.-381-388. 1996.

32. Mehta. P. P.. Bertram. J. S.. and Loewenstein. W. R. Growth inhibition of transformed cells correlates with their junclional communication wilh normal cells. Cell,44: 187-196, 1986.

33. Musil. L. S.. and Goodenough. D. A. Gap junctional intercellular communication andthe regulation of connexin expression and function. Curr. Opin. Cell Biol.. 2:875-880, 1990.

34. Fitzgerald, D. J., Fusenig. N. E., Boukamp. P.. Piccoli. C.. Mesnil. M., and Yamasaki,H. Expression and function of connexin in normal and transformed human keratinocytes in culture. Carcinogenesis (Lond.), 15: 1859-1865. 1994.

2208



1998;58:2200-2208. Cancer Res Ashkan Javaherian, Michael Vaccariello, Norbert E. Fusenig, et al. Progression in Stratified EpitheliumNormal Keratinocytes Suppress Early Stages of Neoplastic

Updated version

http://cancerres.aacrjournals.org/content/58/10/2200

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

[email protected] 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/58/10/2200To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Normal Keratinocytes Suppress Early Stages of Neoplastic … · characterize small numbers of cells with malignant potential in an environment of more normal cells. We have developed