UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Human ovarian surface epithelial cells in culture : characterization and matrix interrelationships Kruk, Patricia Ann 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1992_spring_kruk_patricia_ann.pdf [ 4.62MB ]
Metadata
JSON: 831-1.0098894.json
JSON-LD: 831-1.0098894-ld.json
RDF/XML (Pretty): 831-1.0098894-rdf.xml
RDF/JSON: 831-1.0098894-rdf.json
Turtle: 831-1.0098894-turtle.txt
N-Triples: 831-1.0098894-rdf-ntriples.txt
Original Record: 831-1.0098894-source.json
Full Text
831-1.0098894-fulltext.txt
Citation
831-1.0098894.ris

Full Text

HUMAN OVARIAN SURFACE EPITHELIAL CELLS IN CULTURE:CHARACTERIZATION AND MATRIX INTERRELATIONSHIPSbyPATRICIA ANN KRUKB.Sc., Concordia University, 1 980M.Sc., University of Alberta, 1 983A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSPHYinTHE FACULTY OF GRADUATE STUDIESDepartment of AnatomyWe accept this thesis as conformingto th required standardTHE UNIVERSITY OF BRITISH COLUMBIAMarch 1992c. Patricia Ann Kruk, 1 992Signature(s) removed to protect privacyIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of________________The University of British ColumbiaVancouver, CanadaDate / i’’DE-6 (2/88)Signature(s) removed to protect privacyIIABSTRACTThe human ovarian surface epithelium (HOSE) is thought to give rise to over 85%of human ovarian carcinomas. In spite of its clinical importance, experimentalsystems for investigations of HOSE are lacking and available culture methods haveyielded limited success. In this study, HOSE from normal ovarian biopsy specimenswas used: 1) to improve and simplify the methodology for HOSE culture; 2) tofurther characterize HOSE cells; 3) to characterize and isolate an ovarian-derivedextracellular matrix (ECM); and 4) to study the interactions between cultured HOSEcells and ECM as a means to examine the dynamic, pleomorphic, and morphogeneticnature of HOSE.Improved tissue culture techniques, developed during the course of this study,allowed for a better method for the culture of HOSE cells and the decontamination ofmold-infected cultures. An improved explantation method was developed which takesadvantage of the tenuous attachment of HOSE to underlying tissues: the surfaceepithelium is scraped off the ovarian surface, generating epithelial fragments whichproduce monolayers in culture, with little contamination by other cell types. Thescrape method is superior to the explant method previously described in terms ofspeed, simplicity, higher purity of cultures, and increased cell yield. This improvedculture system provided conditions for the in depth studies of HOSE described herein.A technique for the elimination of mold contamination by a simple one-step Percollgradient centrifugation was developed, using the fungus Ciadosporium as a prototypeand intentionally infected cell lines. Centrifugation in Percoll did not affect cellmorphology, growth, or the cells’ ability to produce ECM. Percoll decontaminationmay be a generally applicable technique for decontamination of mold infectedcultures.Characterization of HOSE cells in culture showed that they are variably positivefor mucin and contain lipid, vimentin, and keratin subtypes #7, 8, 1 8 and 19 liketheir in vivo counterparts. Additionally, HOSE cells produce basement membraneand stromal ECM components as demonstrated by the presence of laminin and collagentypes I, Ill, and IV in HOSE cultures.ECM derived from rat ovarian surface epithelial cell cultures (ROSE 1 99-ECM)also consisted of basement membrane and stromal matrix components as shown bythe presence of laminin, fibronectin, and collagen types I and Ill. This predominantlyfibrillar ECM supports the attachment, spreading, and even growth of several celltypes. Although HOSE cells remain epithelial, but dispersed throughout the ROSE1111 99-ECM, HOSE cells plated on a combination of ROSE 1 99-ECM and collagen gelcontract the matrices into organoids providing evidence that HOSE cells are capableof physically remodelling ECM.Examination of HOSE-ECM interactions revealed that substrata influence HOSEmorphology, growth, and proteolytic activities. On plastic and fibrin clots, HOSEcells formed epithelial monolayers while they are spindle-shaped on rat tail tendon-derived collagen gels. On Matrigel, the cells form aggregates that penetrate and lysethe gel. HOSE cells grow rapidly on plastic, remain stationary on collagen gels andfibrin clots and eventually decrease in number when maintained on Matrigel. Thedifferent morphologic phenotypes did not translate into the expression of differentintegrins at the cell surfaces except for collagen gels. With the exception of cells oncollagen gels, HOSE surveyed for various integrins express receptors for vitronectin(VNR), collagen, laminin, and fibronectin (VLA-2,-3,-5, and 131), but lack 134 andVLA-6, a laminin receptor. There appears to be a downregulation of integrins whencells are maintained on collagen gels. Conditioned medium of HOSE cultures on thesesubstrata contains neutral proteases: 1.) chymotrypsin-like and elastase-likeactivities that are inversely related to growth; and 2.) a 30 KD and a 42 KDgelatinase.These results suggest that normal HOSE cells play an active role in ECMremodelling, that is, its deposition, degradation, and physical reorganization. Thesecapabilities may be important for normal ovarian development and normal adultfunctioning such as ovulation. Furthermore, while proteolytic activities andinvasiveness are characteristics of the malignant phenotype, they also appear to bepart of the normal HOSE phenotype. With regards to why HOSE is a preferred site ofcancer development, it is perhaps aberrations of these functions that contribute topathological states.ivTABLE OF CONTENTSABSTRACT iiTABLE OF CONTENTS ivLIST OF FIGURES viLIST OF TABLES I xLIST OF ABBREVIATIONS xACKNOWLEDGEMENTS x iiINTRODUCTION 1A. Overview 1B. HOSE In Vivo 2(1.) Embryology 2(2.) Adult Morphology and Function 2C. The Role of HOSE in Ovarian Cancer 80. Development of Culture Conditions for HOSE 11E. Rationale/Objective 1 3MATERL’LS AND METHODS 14A. Tissue Culture 14(1.) Tissues 14(2.) HOSE Cultures Established by Explantation from Biopsy Material 1 4(3.) HOSE Culture Purification by Differential Adhesion to Collagen Gel 1 6(4.) Culture Purification by Percoll Density Gradient Centrifugation 1 6(a.) Purification of Mixed HOSE Cultures 1 6(b.) Continuous Percoll Gradients 1 7(c.) Percoll Decontamination of Mold-Infected Cultures 1 8(5.) HOSE Cultures Established by Scraping Biopsy Material 1 9(6.) Media and Culture Conditions 1 9B. Culture Characterization 20(1.) Histochemical Staining 2 1(a.) Oil Red 0 for Neutral Fats 2 1(b.) Periodic Acid-Schiff (PAS) and Amylase + PAS for Neutral Sugars andGlycogen 21(2.) Immunofluorescence Staining 2 1V(a.) Collagen Types I, Ill, andI .21(b.) Keratin 22(c.) Laminin 23(d.) Plasminogen Activator Inhibitor (PAl-i) 23(3.) Staining for Mucin and Keratin 23(4.) Scanning Electron Microscopy and Transmission Electron Microscopy 24(5.) SDS-PAGE and Western Immunoblots for Keratin and Vimentin 24C. Ovarian-Derived Extracellular Matrix and HOSE Organoids 25(1.) Preparation of ROSE 199-ECM 25(2.) Characterization of ROSE 1 99-ECM 26(a.) Alcian Blue Staining for Suiphated and Non-Suiphated Sugars 27(b.) Masson Stain for Collagen 27(c.) Silver Staining for Reticulin 27(d.) Double Staining for Actin and Laminin 28(e.) Hoechst Staining for DNA 28(f.) Western Immunoblots for Laminin, Fibronectin, and Collagen 28(3.) Adhesion Assays 29(4.) HOSE Organoids 30D. Maintenance of HOSE on Various Biological Substrata 3 1(1.) Preparation of Placental-Derived, Collagen, Fibrin Clot, and MatrigelSubstrata 31(2.) Morphological Examination 32(3.) Immunoprecipitations for Integrins 32(4.) Growth Curves 34(5.) Protease Assays 34(a.) Chymotrypsin, Elastase, and Trypsin Chromogenic Assays 34(b.) Plasminogen Activator Assay 35(c.) Collagenase Type I:3H-Collagen Type I Degradation 35(d.) Gelatin Zymography 36RESULTS 37A. Improved HOSE Yield and Tissue Culture 37(1.) HOSE Cultures Established by Explantation from Biopsy Material 37(2.) HOSE Culture Purification by Differential Adhesion to Collagen Gel 37(3.) Culture Purification by Percoll Density Gradient Centrifugation 37(a.) Purification of Mixed HOSE Cultures 37vi(b.) Percoll Decontamination of Mold Infected Cultures.41(4.) HOSE Cultures Established by Scraping Biopsy Material 42B. Characterization of HOSE Cells in Culture. 48(1.) To What Extent do Cultured HOSE Cells Reflect Their In Vivo Counterparts?48(2.) Do HOSE Cells Produce Extracellular Matrix? 56C. Characteristics of an Ovarian-Derived ECM 60(1.) What are the Characteristics of ROSE 1 99-ECM? 60(2.) Does ROSE 1 99-ECM Have Biological Activity? 70(3.) What is the Response of HOSE Cells to ROSE 1 99-ECM? 70(4.) HOSE Organoids: An Ovarian Tissue Culture Model 70D. HOSE-ECM Interactions 75(1.) What is the Morphologic Response of HOSE Cells to Substrata2 75(2.) Do Substrata Influence HOSE Cell Integrin Expression? 79(3.) Do Substrata Influence HOSE Cell Growth? 83(4.) Do HOSE Cells Produce Proteases? 87DISCUSSION 100A. Tissue Culture Techniques 100(1.) Scrape-Derived HOSE Cultures 100(2.) Percoll Decontamination 1 02B. HOSE Cells Cultured from Scraped Biopsies are a RelevantModel System 103C. HOSE:Matrix Interrelationships 1 05(1.) Ovarian Surface Epithelial-Derived Extracellar Matrix 1 05(2.) Physical ECM Remodelling 1 08(3.) Effect of Substrata on HOSE Morphology, Growth, Integrin Expression,and Protease Production 111SUMMARY 117REFERENCES 119vIILIST OF FIGURESFigure 1. Diagramatic Culture of Human Ovarian Surface Epitheliurn 1 5Figure 2. Keratin Expression in HOSE Cells After Differential Adhesion 39Figure 3. Cell Lines Before and After Percoll Decontamination 43Figure 4. Effect of Centrifugation in Percoll on Cell Growth 44Figure 5. Matrix Production Following Percoll Decontamination 45Figure 6. Establishment of HOSE Cultures by the Scrape Method 46Figure 7. Section of a Scraped Ovarian Biopsy 49Figure 8. HOSE Cultures Stained for Oil Red 0 50Figure 9. HOSE Cultures Stained with PAS and Amylase + PAS 52Figure 10. Keratin Expression in Early and Late Passage HOSE Cultures 53Figure 11. Keratin and Vimentin Western Immunoblots 55Figure 1 2. HOSE Cultures Stained for Laminin 57Figure 1 3. HOSE Cultures Stained for Collagen Type IV 58Figure 14. HOSE Cultures Stained for Collagen Type I and Type Ill 59Figure 1 5. ROSE 1 99 Cultures 61Figure 1 6. Light Microscopic Appearance of ROSE 1 99-ECM 62Figure 1 7. Ultrastructural Appearance of ROSE 1 99-ECM 63Figure 18. Histochemical Comparison of ROSE 199 Cultures and ECM 65Figure 1 9. ROSE 1 99-ECM Stained for Laminin 66Figure 20. ROSE 1 99-ECM Doubly Stained for Laminin and Actin 67Figure 21. ROSE 1 99-ECM Stained for Collagen Type I 68Figure 22. Western Immunoblots for Laminin, Fibronectin, and Collagen 69Figure 23. Cellular Adhesion to ROSE 199-ECM 71Figure 24. Cellular Spreading and Growth on ROSE 1 99-ECM 72Figure 25. HOSE Cells on ROSE 1 99-ECM 73Figure 26. Organoid - Schematic Set up 74Figure 27. Contraction into Organoids 76Figure 28. Factors Affecting Organoid Contraction 77Figure 29. Morphological Response of HOSE Cells to Substrata 80Figure 30. Invasion into Matrigel by HOSE Cells 81Figure 31. Cross-sections of HOSE on Collagen and Matrigel 82Figure 32. Survey of HOSE Integrins 84Figure 33. Integrin Time Course 85Figure 34. HOSE - Growth Curves and Chymotrypsin-like Activity 86vifiFigure 35. HOSE - Gelatin Zymography 1 93Figure 36. HOSE - Gelatin Zymography 2 94Figure 37. HOSE -3H-Collagen Type I Digestion 96Figure 38. Expression of Plasminogen Activator Inhibitor by HOSE Cells 97ixLIST OF TABLESTable 1. HOSE Cultures Established by Explantation from Biopsy Material 38Table 2. Percoll Centrifugation to Separate Mixed HOSE Cultures 40Table 3. HOSE Cultures Established by Scraping Biopsy Material 47Table 4. Co-expression of Mucin and Keratin in HOSE Cultures 54Table 5. Summary of HOSE Organoids 78Table 6. Chymotrypsin-like Activity in HOSE Cultures 88Table 7. Elastase-like Activity in HOSE Cultures 89Table 8. Chymotrypsin-like Activity in Conditioned Medium from Day 9HOSE Cultures 90Table 9. Elastase-like Activity in Conditioned Medium from Day 9 HOSE Cultures91Table 1 0. Summary of lmmunofluorescence Studies 98xLIST OF ABBREVIATIONSAB.Alcian blueAH Ammonium hydroxideAPMA p-Aminophenylmercuric acetateBAPNA N-Benzoyl-DL-arginine p-nitroanilideBCIP 5-Bromo-4-chloro-3-indolyl phosphate p-toluidine saltBSA Bovine serum albuminCE Coelomic epitheliumCEA Carcinoembryonic antigenCOLL Rat tail tendon-derived collagen gelDAB DiaminobenzidineDC DeoxycholateDM Defined mediumDMF N,N-DimethylformamideDMSO Dimethyl sulfoxideECM Extracellular matrixEDTA Ethylene-diaminetetra-acetic acidEGF Epidermal growth factorEHS MatrigelFB Fibrin clotFBS Fetal bovine serumFITC Fluorescein isocyante conjugatedEN FibronectinFT Freeze thawHBSS Hank’s balanced salt solutionHC HydrocortisoneH&E Hemotoxylin and eosinH202 Hydrogen peroxideHOSE Human ovarian surface epitheliumHRP Horseradish peroxidaseHUDF Human dermal fibroblastsLN LamininMDPF 2-Methoxy-2,4-diphenyl-3(2H) furanoneMeOH MethanolNBD PhallacidinxiNBT.Nitro blue tetrazoliumOSE Ovarian surface epitheliump Passage numberPA Plasminogen activatorPAl-i Plasminogen activator inhibitor type 1PAS Periodic acid SchiffPBS Phosphate buffered salinePC-i Pederson’s fetuinPD Placental-derived matrixPH PhalloidinPL PlasticPMSF Phenylmethyl-sulfonyl fluorideRLF Rat lung fibroblastsROSE Rat ovarian surface epitheliumROTC Rhodamine isocyante conjugatedRT Room temperatureS1ANA N-Succinyl-ala-ala-ala p-nitroanilideSAPNA N-Succinyl-ala-ala-pro-phe p-nitroanilideSDS-PAGE Sodium dodedecyl sulfate polyacrylamide gel electrophoresisSE Surface epitheliumSEM Scanning electron microscopySIP Stock iso-osmotic PercollTBS Tris-HCITEM Transmission electron microscopyTTBS TBS containing 0.05% Tween-20VNR Vitronectin receptorWM Waymouth’s medium10 Primary20 SecondaryxiiACKNOWLEDGEMENTSI would like to thank Dr. Nelly Auersperg for her advice, guidance, and supportthroughout this research. To the other members of my research committee, Drs. B.Crawford, S. Dedhar, J.T. Emerman, and P.B. Clement, I wish to express myappreciation for expert advice on this research project and for critical examinationof the thesis manuscript. I would like to thank Drs. Booth, Deane, Dunnett, Gomell,Kalyapur, Korn, Lee, Louves, Mahlab, Mitchell, Nickerson, Pendleton, Rowe, andYoshida for their cooperation in supplying ovarian biopsy specimens. To Drs. W.M.Elliott, P.E. Reid, C.H. Siemens, E. Turley, V.-J. Uitto, Mrs. C. Burr, Mrs. L.Mornin, Mrs. B. Muelchen, Mrs. B. Thomas, Mrs. L. Trueman, Mr. M. lagallo, Mr. I.MacLaren, Mr. M. Weiss, Ms. V. Gray, and Ms. S. Maines-Bandiera I extend thanksfor expert technical help. I am grateful to Mrs. L. Stein (Provincial Laboratory, B.C.Centre for Disease Control) for identifying the culture contaminants and Dr. J.T.Emerman for kindly providing rat tail tendon-derived collagen gel. A special thanksgoes to Stephen L. Katzwhöse encouragement and support were essential for thecompletion of this thesis.Portions of this thesis have been reproduced with written permission from theTissue Culture Association and the American and Canadian Society of Pathology.This research was supported by grants from the Medical Research Council andNational Cancer Institute of Canada to Dr. N. Auersperg and fellowships from theBritish Columbia Foundation for Non-Animal Research, MacMillan FamilyFellowship Foundation, and University of British Columbia Graduate FellowshipFoundation.1INTRODUCTIONA. OverviewThe human ovary is invested with a simple, yet dynamic, epithelium which isintimately involved with ovarian development, function, and pathology. Duringdevelopment, the human ovarian surface epithelium (HOSE) undergoes periods ofintense proliferation and invades the ovarian stroma as the sex cords to becomeassociated with oogonia and contribute to the formation of the primordial follicles. Inthe adult, HOSE undergoes cyclic changes in morphology related to ovulation. As thepreovulatory follicle matures, HOSE is flattened due to mechanical stresses and atovulation, the continuity of HOSE is disrupted. Although there is little evidence tosuggest that HOSE plays a role in the initiation of ovulation, following ovulation,HOSE cells proliferate at the edges of the ovulatory wound, migrate over the woundsite, and contribute to the repair of the ovulatory defect. When repair is completed,the HOSE returns to its original, resting morphology of a simple cuboidal epithelium.The attrition of folilcles during menopause results in ovarian remodell[ng andshrinkage so that the senile ovary appears wrinkled and the ovarian surface is foldedinto a gryus-like appearance. Further, with increasing age the number of HOSE-lined crypts and cysts increases.Yet, the most notable change in HOSE is its transformation to malignancy which isresponsible for over 85% of ovarian cancers, despite the fact that HOSE comprisesonly a minute fraction of the total ovarian mass. HOSE-derived carcinomas arecommon, and in general, have a poor prognosis. In spite of its clinical importance,studies on the biology of normal HOSE and the role of HOSE in ovarian carcinogensishave been severely limited by the lack of experimental systems. Animals havecontributed little to the understanding of HOSE-derived malignancies because, withfew exceptions, ovarian cancers in animals do not arise in the ovarian surfaceepithelium. Most of the information about ovarian carcinogenesis is based uponhistopathological examination. These studies suggest that early malignant changesoccur in HOSE-lined cysts and invaginations rather than on the ovarian surface.Undoubtedly many factors contribute to the dynamic nature of HOSE such that it iscapable of proliferation, migration, invasion, cyst formation, and wound repair, butone common factor appears to be an interaction between HOSE and extracellularmatrix. This report examines the dynamic, pleomorphic, and morphogenetic natureof HOSE by: 1) improving HOSE culture methodology; 2) further characterizing2HOSE cells; 3) isolating and characterizing an ovarian-derived matrix; and 4)studying the interactions between cultured HOSE cells and various extracellularmatrices.B. HOSE In Wvo(1.) EmbryologyThe dynamic nature of HOSE is evidenced throughout life. Even early indevelopment HOSE cells demonstrate pleomorphism, a proliferative capacity, andinvasive behaviour. Gonadal development is first seen around the 5th week ofdevelopment with the formation of the indifferent gonad that is covered with coelomicepithelium (CE) and which develops on the medial side of the mesonephros. The CEproliferates, possibly in response to induction by mesonephric cells (Wartenberg‘82) to form a bulging structure referred to as the gonadal ridge. Cords of surfaceepithelial cells, the primary sex cords, penetrate the mesenchyme of the indifferentgonad and form a rudimentary rete ovarii in female embryos. While the primary sexcords in female embryos normally regress and degenerate, vestiges of the rete ovariioccur as epithelial cords in the region of the hilum of the ovary while the cortex ofthe indifferent gonad will differentiate into the cortex of the adult ovary (Byskov‘86, Moore ‘82).At 4-5 months of development the epithelium covering the fetal ovary (i.e. HOSE)contributes to the formation of the somatic components of the ovarian cortex.Following intense mitotic activity HOSE cells become stratified (Gondos ‘75, Motta &Makabe ‘82, Nicosia ‘83, Pinkerton et al. ‘61). It is suggested that the stimulus forsuch proliferation in the fetal ovary may be the onset of steroid production andsecretion by the interstitial cells (Gondos ‘75). At the light microscopic level thethickening of the ovarian surface epithelium into a multilayered structure is usuallyuniform in nature, but occasionally irregular papillary projections are present. Inthis situation, HOSE cells can be oriented in many different directions and exhibitpleomorphism. Further, groups of HOSE cells, referred to as the secondary sexcords, invade and penetrate the fetal ovary, associate with germ cells which havemigrated from the yolk sac, and contribute to the formation of primordial follicles.Whether follicular cells are derived solely from HOSE or mesonephric cells or fromboth (Byskov ‘86, Gondos ‘75, Hirshfield ‘91, Nicosia ‘83, Wartenberg ‘82) isunclear so that the contribution of HOSE to the various components of the adult ovaryis still controversial. With the formation of the primodial follicles completed, HOSE3is reduced to a simple epithelium and remains relatively quiescent throughout theremainder of development and childhood.While the mesonephric kidneys are the fetal functional kidneys in both sexes andthe mesonephric ducts give rise to the male genital ducts and their associatedderivatives, it is the paramesoneophric or Mullerian ducts which give rise to thegenital ducts and their associated derivatives in female embryos (Moore ‘82). Theparamesonephric ducts arise as invaginations of the same CE that gives rise to theovarian surface epithelium just laterally to the mesonephric ducts and in closeproximity to the site of development of the ovary. The edges of the invaginationsapproach each other and fuse to form the paramesonephric ducts proper. Cranially,the ducts open into the coelomic cavity which will be the future peritoneal cavity.The cranial unfused portions of these ducts will develop into the oviducts. Theparamesonephric ducts run caudally parallel to the mesonephric ducts. In the caudalregion they come together in the midline and fuse forming the uterovaginalprimordium, i.e. the primordium for the epithelium of the uterus and part of thevagina.Hence arises an important relationship. The CE gives rise to both the HOSE andMullerian duct, so that the lining of the Mullerian duct derivatives and HOSE areembryologically closely related. This proximity is reflected in the various directionsof Mullerian differentiation pursued by the HOSE when it undergoes neoplasia(Blaustein ‘77a). The most frequent forms of metaplasia, in order of decreasingfrequency, are to cells resembling oviduct, endometrium, and endocervix (Blaustein‘77a,b, Scully ‘70,’77). This is of clinical importance in the interpretation of thephenotypes of HOSE-derived ovarian carcinomas.(2.) Adult Morphology and FunctionHOSE remains relatively unchanged throughout childhood. At the light microscopiclevel, the normal resting HOSE (such as prior to ovulation) investing the adult ovaryis a single layer of squamous to columnar epithelium depending upon its positionrelative to follicles, corpora lutea, and crypts (Anderson et al. ‘76, Balboni ‘80,Centola ‘80,’83, Motta et al. ‘80, Papadaki & Beilby ‘71, Van Blerkom & Motta ‘79).The ovarian surface epithelium is also separated from underlying ovarian structuresby a basement membrane below which is a layer of dense collagenous materialreferred to as the tunica albuginea and in humans the ovarian surface epithelium istenuously attached to its underlying tunica albuginea (Anderson et al. ‘76, Clement‘87, Van Blerkom & Motta ‘79). Although rare in human ovaries, surface papillae4interrupt the otherwise smooth ovarian surface in other animals such as rabbits(Motta et al. ‘80, Nicosia et al. ‘89, Nicosia & Johnson ‘84b, Van Blerkom & Motta‘79). Areas of the surface of the human ovary are occasionally invaginated into thesubjacent cortical layers in the form of crypts, cords, and inclusion cysts lined byHOSE cells (Motta et al. ‘80, Nicosia & Nicosia ‘88, Van Blerkom & Motta ‘79, Younget al. ‘89). Crypts are hollow, tubular invaginations in which the lumens opendirectly onto the surface of the ovary. The cells lining the crypts are continuous withHOSE on the ovarian surface and are generally cuboidal or columnar in shape(Papadaki & Beilby ‘71). Some invaginations of HOSE terminate as small, irregularsolid masses or cords of cells that lack a lumen and are directly connected to theovarian surface. In addition to crypts and cords, HOSE-lined inclusion cysts occur inthe cortical stroma which are not in contact with the surface HOSE (Blaustein ‘81b,Van Blerkom & Motta ‘79). Characteristics of HOSE in vivo include the presence of:glycogen; glycosaminoglycans; epithelial mucin; lipid (Blaustein ‘84, Blaustein &Lee ‘79); glycoproteins (Guraya ‘80); epidermal growth factor receptors (Berchuket al. ‘91); 17-3-hydroxysteroid dehydrogenase activity (Blaustein & Lee ‘79);keratin (Czernobilsky ‘85, Czernobilsky et al. ‘84), and vimentin filaments(Czernobilsky ‘85).While appearing uniform at the light microscopic level, Gillet et al. (‘91)demonstrated the pleomorphic nature of HOSE in vivo . SEM examination of ovariansamples revealed two morphologically distinct types of HOSE cells. Both cell typesare common to all surfaces of the ovary including the distending preovulatoryfollicle. Type A cells are cuboidal and covered with numerous microvilli while type Bcells are flat squamous cells with fewer microvilli. It is suggested that type B cellsare the result of squamous metaplasia in response to surface injury associated withovulation. Once repair is completed and the ovulatory defect is re-epithelialized,zones of type B cells become permanantly established. Other ultrastructural studiesreveal that the apical surface of HOSE is generally covered by numerous microvilliof uniform size and shape (Anderson et al. ‘76, Blaustein & Lee ‘79, Gillet et al. ‘91,Motta & Van Blerkom ‘80, Van Blerkom & Mota ‘79). The microvilli may bebranched and are coated with glycocalyx (Motta & Van Blerkom ‘80). Although HOSEis continuous with the peritoneal mesothelium at the mesovarium, it is functionallyand morphologically somewhat different from mesothelial linings (Andrews & Porter‘73, Bebbehani et al. ‘82, Blaustein ‘84, Whitaker et al. ‘80a, ‘80b, ‘82). Forexample, a single cilium is seen on some HOSE cells (Anderson et al. ‘76, Nicosia &Nicosia ‘88, Papadaki & Beilby ‘71, Van Blerkom & Motta ‘79), a feature lacking on5mesothelial cell surfaces. Mesothelial cells also lack 17-f3-hydroxysteroiddehydrogenase activity. Adjoining lateral cell borders of HOSE cells are connected bydesmosomes along the length of the cell and by incomplete tight junctions near theapical free surface of the cell (Anderson et al. ‘72,76, Blaustein & Lee ‘79, Motta etal. ‘80, Papadaki & Bielby ‘71). Anderson et al. (72) demonstrated the occurence ofincomplete tight junctions in the mammalian ovarian surface epithelium whenhorseradish peroxidase injected intraperitoneally made its way into the ovary. Gapjunctions have been observed in the ovarian surface epithelium of the rabbit ovary(Anderson et al. ‘76), but no specific reference has been made to their presence inHOSEFunctionally, HOSE appears to provide a protective and lubricated covering over theovary as denuding the ovary of its HOSE has been implicated in the pathogenesis ofadhesion formation (Gillet ‘91). Histochemical and ultrastructural studies indicatethat HOSE cells are involved in both secretory (presence of RER, Golgi complexes,secretory vesicles) and endocytotic (presence of endocytotic pits, lysosomes,multivesicular bodies) activities so that HOSE may be involved in the transport ofmaterial to and from the peritoneal cavity (Papadaki & Beilby ‘71). Further, anincreased number of lysosomes has been noted in HOSE just prior to ovulation withapparent basal release of the lysosomal contents at ovulation suggesting that HOSEmay produce proteolytic enzymes which play a role in follicular rupture by thedissolution of the follicular apex at ovulation (Bjersing & Cajander ‘75, Cajander &Bjersing ‘77). However, direct evidence for the production of proteases by HOSEcells has not yet been reported. The presence of 17 3-hydroxysteroid dehydrogenaseactivity has been reported in HOSE (Auersperg et al. ‘84, Blaustein & Lee ‘79) and3-f3-hydroxysteroid dehydrogenase and 1 7-3-hydroxysteroid dehydrogenaseactivities have been reported in the ovarian surface epithelium of mice, rats, andrabbits (Rembiszewska & Brynczak ‘85) indicating that the ovarian surfaceepithelium is capable of progestin and androgen metabolism. However, HOSE does notappear to be capable of steroid synthesis. HOSE cells show an absence of cholesterolor its esters (Guraya ‘80) as well as steroidogenic enzymes, such as P450 side chaincleavage activity, necessay for de novo steroid synthesis (Blaustein & Lee ‘79,Hoyer ‘80, Rembiszewska & Brynczak ‘85).At puberty the onset of menses marks the beginning of the reproductive phase oflife. From then on the adult ovary undergoes repetitive periods of folliculogenesisand steroidogenesis for the development of a mature follicle and oocyte which isreleased at ovulation. For ovulation to occur the follicle must penetrate formidable6barriers. These include, from inward to outward, granulosa cells, the basementmembrane underlying the granulosa cells, layers of theca interna and externa, theovarian stroma, the tunica albuginea, the basement membrane underlying HOSE, andfinally HOSE itself. Structural alterations at ovulation restricted to the stigma areain the follicular wall have been reported in mammals and avians. There is a looseningand decomposition of the collagen fibers and glycosaminoglycans in the tunicaalbuginea and thecal layers (Jackson et al. ‘91, Yoshimura et al. ‘87) of thepreovulatory follicle. The degradation of glycosaminoglycans and collagen fibers inthe stigma region of the preovulatory follicle prior to ovulation is thought to reducethe tensile strength of the ovulatory barrier, and hence, loosen the barrier and easeovulation.In ovulation the oocyte transverses the surrounding connective tissue and basementmembrane barriers which have been disrupted by the proteolytic activities of thefollicular cells. This process could be considered similar to tumour invasion, inwhich malignant cells disrupt basement membranes. Therefore, it is not surprisingthat collagenolytic (including type IV) activities are found in the follicular fluid(Puistola et al. ‘86). In many species including humans, rats, rabbits, and chickens,there is an increase of many proteases including stromal collagenases, type IVcollagenases, and plasminogen activator as ovulation approaches, followed by adecrease in protease production after ovulation (Cajander ‘89, Cajander et al. ‘89,Curry et al. ‘86, Politis et al. ‘90a,b, Puistola et al. ‘86, Tilly & Johnson ‘87).While ovulation requires protein synthesis (Brannstrom et al. ‘89), proteaseproduction in the ovary appears to be hormonally regulated and can be induced inhuman and rat follicular fluid following the administration of LH or hCG (Curry etal. 88, ‘89, Dhanasekaran & Moudgal ‘88, Reich et al. ‘85b, Thomas & Sernia ‘90,Yoshimura et al. ‘87). The importance of proteases such as collagenase in ovulationis evidenced where the application of collagenase inhibitors completely inhibitsovulation (Brannstrom et al. ‘88, Reich et al. ‘85a, Yoshimura et al. ‘87).Interestingly, a preovulatory increase in both serum-derived and tissue-derivedprotease inhibitor activities parallels the proteolytic and collagenolytic changes(Zhu & Woessner ‘91). Metalloproteinase inhibitors acting on collagenase,gelatinase, and proteoglycanase have also been found in the follicular fluid and theirlevels change with follicular development (Curry et al. ‘88, ‘89, ‘90). Likewise,plasminogen activator inhibitor activity has been found in increasing amountsassociated with follicular maturation (Politis et al. ‘90a,b). Like protease activities,inhibitor activities in the ovary correlate with the hormonal environment. That is,7concentrations of inhibitors increase with increasing estradiol and progesteroneconcentrations and thus, with the stage of folliclular development (Curry et al. ‘88,‘89, Ny et al. ‘85). The major source of protease and inhibitor production appears tobe the granulosa cells (Cajander et al. ‘89, Fukumoto et al. ‘81, Ny et al. ‘85).Concomitant increases in both proteases and their inhibitors have been proposed tomaintain proteolytic homeostasis and provide localized control of extracellularmatrix degradation (Unemori et al. ‘90). Follicular degradation may occur due to anexcess of enzyme in relation to its inhibitor at the apex of the follicle. For example,Fukumoto et al. (‘81) reported that collagenase activities in the apex of the folliclewere higher than at the base of the follicle throughout the ovarian cycle.In addition to changes occuring in the maturing follicle related to ovulation, HOSEundergoes dramatic morphological and functional changes, such as proliferation andmigration, related to the processes occuring in the ovarian cortex (Papadaki &Beilby ‘71). These changes may occur in response to hormonal changes (VanBlerkom & Motta ‘79). For example, the rat ovarian surface epithelium has beenshown to be susceptibile to hormones (Hamilton ‘80) and to exhibit estrogenreceptors (Adams & Auersperg ‘83). As the preovulatory follicle matures, it causesdistension of the overlying tunica albuginea and HOSE. HOSE is composed of cuboidalcells with numerous microvilli at the base of the preovulatory apex, but is flattenedwith fewer microvilli on the lateral and apical surfaces of the protruding follicle dueto mechanical stresses (Dietl et al. ‘87, Motta & Van Blerkom ‘80). Intercellularaccumulations of fluid appear to be correlated to disruption of the apical wall of thepreovulatory follicle (Dietl et al. ‘87). Cajander and Bjersing (‘77) and Rawson andEspey (‘77) noted the accumulation of numerous lysosomes in the cytoplasm of theHOSE cells just prior to ovulation and the subsequent disappearance of lysosomesshortly after ovulation. In addition, irregular infoldings of the basal plasmamembrane indicated that the active substance, presumably proteolytic enzymes, hadbeen released by exocytosis to contribute to the dissolution of the follicular apexbefore rupture. Others have also implicated HOSE in the production of proteolyticenzymes responsible for the degradation of the surrounding stroma (Guraya ‘80,Okamaura et al. ‘80). While the contributions of HOSE to the initiation of ovulationremain unknown, the net result is that the continuity of HOSE is disrupted atovulation.Following ovulation, repair is initiated to the ovulatory wound composed of plasmaclot, follicular fluid, and cell debris (Nicosia ‘83, Van Blerkom & Motta ‘79). Theovulatory defect is invaded by stromal cells which produce ECM components that act8as a scaffold for the adjacent ovarian HOSE cells (Nicosia ‘83, Papadaki & Beilby71, Harrison & Weir 77). HOSE proliferates at the wound edges and migrates overthe wound site as very flattened cells essentially devoid of microvilli. When repair iscompleted and the ovulatory defect is re-epithelialized, HOSE returns to its restingmorphology (Gillet et al. ‘91, Motta & Van Blerkom ‘80). During the reproductiveperiod the ovarian surface becomes increasingly scarred with each subsequentovulation.If fertilization and implantation of the ovum occurs, the corpus luteum of thatovulatory cycle does not regress, but is transformed into a temporary endocrinegland, the corpus luteum of pregnancy, which produces progesterone for six weeksand then degenerates (Balboni ‘83, Nicosia ‘83). Ultrastructurally, in the pregnantcondition, HOSE differs from HOSE in the non-pregnant condition in that its basalsurface is deeply infolded and penetrates well into the underlying tunica albuginea(Papadaki & Bielby ‘71). There appear to be few if any cytoplasmic differencesexcept for the presence of more lipid droplets (Papadaki & Bielby ‘71).Attrition of follicles during menopause results in ovarian remodelling andshrinkage so that, grossly, the senile ovary appears wrink[ed and cerebriform [nappearance (Blaustein ‘77a, Nicosia ‘83, ‘87, Papadaki & Beilby ‘71, Sauramo ‘52).HOSE follows the contours of the gryus-like senile ovary and often lines surfaceinvaginations or crypts that can penetrate deep into the ovarian cortex. Further, anincreased number of HOSE-lined inclusion cysts and other anomalies such aspapillary or cord-like projections are common in the post-menopausal ovary(Clement ‘87, Nicosia ‘83, ‘87). Histopathological studies suggest that HOSE liningsuch crypts and cysts is frequently atypical and appears to be the site of earlymalignant changes in HOSE (Scully ‘70,’77). Lastly, many reports suggesting thatHOSE is lost from the ovarian surface in the post-menoapusal ovary and limited tolining crypts and cysts (McKay et al. ‘61, Sauramo ‘52) appear to be erroneous.Rather, the absence of HOSE from the ovarian surface can be attributed to ovarianhandling with the resultant removal of HOSE from the ovarian surface (Clement ‘87,Gillet ‘91).C. The Role of HOSE in Ovarian CancerOver 80% of ovarian cancers are thought to arise from HOSE (Czernobilisky ‘85,Fox ‘80, Nicosia & Nicosia ‘88, Parmley & Woodruff ‘74, Scully ‘70, ‘77, Young etat. ‘89) despite the fact that HOSE comprises only a small fraction of the total9ovarian mass. HOSE-derived carcinomas are commom, and in general, have a poorprognosis. For example, HOSE-derived carcinomas are the 5th most frequentlyoccuring malignancy among American women (Richardson et al. ‘85). Although lessfrequent than cervical and endometrial cancers, the mortality attributed to ovariancancers far exceeds that of both the others combined. The 5-year survival of patientswith ovarian cancer is no better than 37% (Nicosia & Nicosia ‘88). The poorprognosis is largely due to inadequate means to detect ovarian cancers in earlystages, so that generally by the time the initial diagnosis is made, the cancer isinoperable. Ovarian carcinoma cells express epidermal growth factor receptors atthe cell surfaces and receptor expression is associated with poor prognosis (Berchuket al. ‘91, Rodriguez et al. ‘91). Additionally, HOSE-derived tumours respond poorlyto the chemotherapy methods currently available. Epidemiologic studies indicate thatracial, geographic, and genetic (Christian ‘71, Dazois et al. ‘71, Kinbrough ‘29,Liber ‘50, Lingeman ‘74, Weiss ‘80) factors may be involved. Also, it appears thatovarian cancers increase with ovulatory age (Casagrande et al. ‘79, Joly et al. ‘74,Papadaki & Beilby ‘71, Weiss ‘80), that is, the total time in a woman’s reproductivelife during which her ovarian cycle is not surpressed by pregnancy, lactation, use oforal contraceptives, and certain pathologies. Environmental agents includingpolycyclic aromatic hydrocarbons, industrial pollutants, smoking, asbestos, talc,and infectious agents may be involved in HOSE carcinogenesis (Cramer et al. ‘82,‘83, Gerard et al. ‘78, Henderson et al. ‘71, Longo et al. ‘79, Mattison &Thorgiersson ‘78, Newhouse et al. ‘72, Weiss ‘80).In spite of its clinical importance, studies of the biology of normal HOSE and therole of HOSE in ovarian carcinogenesis have been severely limited by the lack ofexperimental systems. Most information about ovarian carcinogenesis is based uponhistopathological examinations of ovaries that usually come from women withdisease. These histopathological studies suggest that early malignant changes occur inHOSE-lined crypts or inclusion cysts in the ovarian stroma rather than on theovarian surface (Scully ‘70, ‘77). The close embryonic relationship between HOSEand Mullerian duct derivatives is exemplified in the epithelium lining inclusioncysts. These common inclusion cysts histologically exhibit a variety of Mulleriantype epithelia including cuboidal, columnar, secretory, ciliated, and squamous cells.However, their common characteristic is that they form more or less rounded,dilated structures consisting of a single layer of epithelium surrounding a centrallumen. Little has been reported about the contents of inclusion cysts. Further,inclusion cysts are normally well separated from each other by abundant stroma. An10example of such a common ovarian inclusion cyst is illustrated in figure 7. Theliterature on the formation and pathogenesis of inclusion cysts is also sparsealthough following examination of sections of 1000 normal ovaries Radisavljevic(‘77) suggested that inclusion cysts are almost always initiated by ovulation. Theprevailing theory is that as a woman approaches menopause, and sometimes earlierin her reproductive life, but rarely before puberty, the HOSE extends downward intothe ovarian stroma to form surface epithelial inclusion cysts. So, basically, with agethere is an invagination or indentation of the HOSE which may be subsequently nippedoff by surrounding stroma. Cancers, then, are thought to arise from HOSEinvaginations which may be continuous with the outer surface or which may beisolated and located within the stroma. Some researchers consider this process ofinclusion cyst formation as a return of the surface epithelium to a state whichrecapitulates Mullerian duct development (Blaustein ‘77a, Scully ‘77).While the predominant theory is that the development of common ovarian inclusioncysts is post-pubertal and associated with normal and abnormal ovulations or withrearrangements in the postmenopausal ovary, inclusion cyst formation is not only alate occuring event. The existence of ovarian inclusion cysts in fetal ovaries has beenreported (Blaustein ‘81b, Meizner et al. ‘91). This suggests, then, that there may betwo groups of ovarian inclusion cysts in the adult human ovary. One develops withadvancing age as a result of normal and abnormal ovulations and the other occursduring fetal development.Fetal inclusion cysts as one of a possible dual origin of ovarian inclusion cysts is anattractive idea and it may relate to the re-expression or increased expression ofonco-developmental genes during carcinogenesis. For example, increasedcarcinoembryonic antigen (CEA) serum levels are used diagnostically as tumourmarkers. The veritability of this marker is seen in mucinous cystadenocarcinomaswhere 77% of the patients with this carcinoma show elevated CEA serum levels(Blaustein et al. ‘82). Moreover, Blaustein et al. (‘82) found that CEA was presentin fetal inclusion cysts, papillary cystadenomas, and papillary serouscystadenocarcinomas. While this does not prove the origin of tumours from thesecysts, it demonstrates a shared characteristic, CEA production, among all three.The exact mechanism by which the HOSE extends downward into the ovarian stromato form inclusion cysts is not known, It remains spectulative as to whether there isan active ingrowth and lumen-formation by the normal HOSE or whether inclusionglands form passively through rearrangement of the ovarian surface throughovulation, scarring, or shrinkage of the ovary with age. It is also not known what11conditions must be present for inclusion cyst formation and further progression toneoplasms.Interestingly, a variety of HOSE-derived inclusion cysts have been reported. Forexample, Kerner et al. (‘81) distinguished between common ovarian epithelialinclusion cysts described above and a particular type of HOSE-derived inclusioncysts found in patients with endometriosis. Although of the same origin as thecommon ovarian epithelial inclusion cysts (i.e. from HOSE), these ovarianmesothelial inclusions were smaller that common ovarian inclusion cysts andconsisted of closely packed cell nests lacking a central lumen found in common HOSE-lined cysts. These ovarian mesothelial inclusions were found to occur deeper in thecortex than common inclusion cysts. The reason for the appearance of thesemesothelial inclusion cysts only with endometriosis is not known, but is notfortuitous as such inclusions were not found in any of the ovaries examined withoutendometriosis. The occurence of these mesothelial inclusions and endometriosissuggests that a common stimulus could be responsible for the development of bothconditions.D. Development of Culture Conditions for HOSETo date there is no direct evidence that human ovarian carcinomas arise in theHOSE. The inference of HOSE as the source of ovarian epithelial cancers is basedprimarily on the histopathological finding of premalignant and malignant changes inthe HOSE of ovaries which may sometimes be seen to be in direct continuation withfully malignant tumours.Animal models have contributed little to the understanding of HOSE cancers becauseovarian tumours in species other than man most typically arise in granulosa, theca,stromal, or germ cells (Cotchin ‘77, Fox ‘80, Marchant ‘80, Murphy ‘80, Stevens‘80). Ovarian surface epithelial-derived tubular adenomas in mice with a geneticdeletion of oocytes (Murphy ‘80), adenocarcinomas in hens and adenomas andcarcinomas in dogs (Marchant ‘80) are the exceptions. Ovarian epithelial tumourscan be induced in rodents by irradiation (Miller ‘72) and carcinogenic chemicals(Marchant ‘80, Miller ‘72, Murphy ‘80, Normura ‘77). However, because of thesmall amount of ovarian surface epithelium present and the complex cellularcomposition of the ovary in these models, it is difficult to be certain of the preciseorigin of the malignant cells in these few studies. Thus, there is no satisfactory12animal model for the study of the role of surface epithelium in human ovariancarcinogenesis.Rat ovarian surface epithelial cells were the first culture model for ovariansurface epithelium (Adams & Auersperg ‘81, Hamilton et al. ‘80). These cells havesince been shown to have estrogen receptors (Adams & Auersperg ‘83, Hamilton etal. ‘80) and hydroxysteroid dehydrogenase activity (Adams & Auersperg ‘81).Androgens (testosterone and 5-dihydrotestosterone) stimulate their growth whileantiestrogens inhibit their growth (Hamilton et al. ‘80, ‘83). Further, rat ovariansurface epithelial cells are susceptible to transformation by oncogenic retrovirus(Adams & Auersperg ‘81). More recently, methods for the isolation and culture ofrabbit ovarian surface epithlium have been reported (Nicosia et al. ‘84, ‘85,Piquette & Timms ‘90, Setrakian et al. ‘90). In this species, several proteinhormones (hCG, FSH, LH, and prolactin) stimulate surface epithelial cell growth(Olsterholzer et al. ‘85a,b). These cells are also susceptible to mineral fiberirritation (Nicosia & Johnson ‘84a) and demonstrate morphological plasticity(Nicosia et al. ‘85, ‘89).Until recently, attempts to culture normal HOSE have been unsuccessful, althoughcells from benign HOSE-derived tumours have been grown in culture and establishedtumour-derived cell lines have been easily maintained in vitro (Courtenay ‘80). In1984, Auersperg et al. first reported the growth of HOSE in primary culture adaptedfrom the culture methods originally developed for rat cells. HOSE cells were grownin plastic dishes using Waymouth’s medium 752/1 + 25% fetal bovine serum (FBS)from explants of ovarian surface from biopsy material obtained from premenopausalwomen undergoing surgery for non-malignant disorders. HOSE cells weredistinguished from other morphologically similar cells (endothelial and mesothelial)and from other ovarian cells by the presence of keratin and 17-f3-hydroxysteroid-dehydrogenase.While the initial growth of HOSE under the conditions described above was limitedand unpredictable, further work improved these methods (Siemens & Auersperg‘88). Of several culture media examined, the medium best suited for the serialcultivation and clonal growth of HOSE contained medium 199:MCDB 202 mixed 1:1supplemented with 15% FBS, 20 ng/ml epidermal growth factor (EGF), and 0.40ug/mI hydrocortisone. In this medium, HOSE; 1) underwent 20-25 populationdoublings before senscence; 2) had a population doubling time of approximately 48hours in the log phase of growth; 3) could be subcultured 8 to 10 times; 4) had aseeding efficiency of up to 53%; and 5) had a cloning efficiency of up to 13%.13Rodriguez et al. (‘91) also demonstrated the mitogenic effect of EGF on cultured HOSEcells as well as the presence of EGF receptors on the cell surfaces. Yet, there is norelationship between the number of EGF receptors and responsiveness to the mitogen.More recently, rete ovarian epithelial cells from primary cultures of ovaries fromtwenty-week old fetuses have been grown in chemically-defined, serum-freeconditions (Dubeau et al. ‘90) and HOSE cells from adult women have beenmaintained in serum-free medium (Elliott & Auersperg ‘90).E. Rationale/ObjectiveRecent work developed the culture conditions to sustain growth of HOSE inmonolayer and allowed propagation of sufficient cell numbers for subsequent studieson normal HOSE growth (Siemens & Auersperg ‘88). However, a persistant problemwas contamination by ovarian cells other than HOSE cells which was difficult toeliminate in explant-derived cultures. The objectives of this project were: 1.) toimprove and simplify the methodology for HOSE culture; 2.) to further characterizeHOSE cells; 3.) to characterize and isolate an ovarian-derived matrix; and 4.) tostudy the interactions between cultured HOSE cells and extracellular matrix as ameans to examine the dynamic, pleomorphic, and morphogenetic nature of HOSE thatmight account for normal physiological changes and for cyst formation. The responseof HOSE to various three-dimensional substrata was examined in the hope that thismight mimic HOSE-ECM interactions naturally occuring in the ovary and provideinsight as to why HOSE is a preferred site of cancer development. The work reportedhere, then, may provide a better understanding of HOSE physiology and of the etiologyof ovarian epithelial cancers and be of clinical relevance as it may provide usefulinformation for the diagnosis of ovarian carcinomas.14MATERIALS AND METHODSA. Tissue Culture(1.) TissuesThirty five biopsy specimens of normal ovary were obtained at surgery fromwomen undergoing surgery for nonmalignant gynecological disorders (ages 21 to 66years). The specimens were collected under aseptic conditions and transported to thelaboratory in Waymouth’s 752/1 medium (WM; Sigma, St. Louis, MO) with 10% or15% fetal bovine serum (FBS; Hyclone, UT) at ambient temperature. Each specimenwas assigned a code name to maintain the patient confidentiality.(2.) HOSE Cultures Established by Explantation from Biopsy MaterialThe first 9 ovarian specimens were processed for culture by the explantationmethod of Auersperg et al. (‘84). Briefly, the tissue was transferred to a glassdissecting dish containing routine culture medium, Medium 199 (Sigma): MCDB202 (Irvine Scientific, Irvine, CA) mixed 1:1 and supplemented with 15% FBS and25 ug gentamicin/mi (GIBCO, Grand Island, NY) or 100 IU penicillin G (GIBCO) and100 ug streptomycin (GIBCO). Under a dissecting microscope, extraneous stromawas trimmed from beneath the ovarian surface (figure 1). The tissue was then cutinto small pieces or explants 1-2 mm3. These explants were placed, ovarian surfacefacing up, into 35 mm culture dishes (Corning, Corning NY), weighed down withcoverslips and glass cloning cylinders, and two mis medium was gently added. Theexplants were arranged near the periphery of the coverslips to allow adequateexchange of nutrients and gases. In addition, the culture medium used to collect andprepare the tissues was spun at 1000 RPM (180g) in a clinical centrifuge and thecells within these washings were plated into a separate culture dish.In total 469 explants were cultured in this manner. Cell outgrowths extended outonto the culture dishes and coverslips. The outgrowths were identified as epithelial(either typical cobblestone epithelial or atypical epithelial) or fibroblastic(fusiform cells arranged in parallel arrays) by the morphological criteria ofSiemens & Auersperg (‘88). When HOSE outgrowths began to approach each other,the coverslips were removed and the explants transferred to new dishes. Theexplants were again covered with coverslips and cloning cylinders and immersed intwo ml medium to obtain additional HOSE growth. In most cases, however, outgrowths15Figure 1. Diagramatic Culture of Human Ovarian Surface Epithelium(A), Ovarian biopsy specimens are obtained at surgery, transported to thelaboratory in sterile culture medium, and examined under a disecting microscopeso that only areas devoid of anomalies are chosen for subsequent culture. In theexplantation method (B,C), biopsies are trimmed of excess stroma and cut with ascapel into 1-2 mm3 explants (B) which are placed epithelial side up in aculture dish and weighed down (C). HOSE outgrowths extend out onto the dish oroverlying coverslip. In the scraping method (D,E), the ovarian biopsy is held,surface side down, over a culture dish containing medium. The surface epitheliumis scraped firmly 2 to 3 times with a rubber policeman (D). Scraping generatessheets of HOSE and these are rinsed from the scraper and biopsy surface into oneor two culture dishes (E). The epithelial sheets attach to the culture dish andHOSE cells grow out from the attached sheets.CI1-r? JDiSAA.*LJE.16contained contaminating fibroblasts. Contaminating fibroblasts were removed fromcultures by either scraping the fibroblasts off with a rubber policeman or cuttingthe fibroblastic outgrowths from the coverslips with sterile scissors. Both of thesemethods of decontamination were only temporarily successful. Eventuallyfibroblastic contamination returned. Often, fibroblasts overgrew HOSE cultures.Dishes badly contaminated with fibroblasts were discarded. Because of the problemsassociated with HOSE cultures derived from explants: 1) fibroblastic overgrowth; 2)low HOSE cell yield; and 3) laborious technique, attempts were made to improve theculture of surgical specimens.(3.) HOSE Culture Purification by Differential Adhesion to Collagen GelTo increase the purity of cultured HOSE cells from fibroblastic contaminants, thepossibility that HOSE cells might demonstrate differential adhesion to rat tailcollagen gel as compared to that of stromal fibroblasts was examined. Rat tail tendon-derived collagen gels were prepared according to the method of Emerman and Pitelka(‘77) and described in Materials and Methods section D. (1.). HOSE from 2 flasks ofWelch HOSE passage 4 (p4) and one flask of Solo HOSE p1 containing mixtures ofepithelial and fibroblastic cells, were plated onto the unrimmed collagen gels. AtT=0, 10, 15, 20, 25, 30, 40, 45, and 60 minutes following plating, the cultureswere observed, photographed, and the non-adherant (floating) cells were collectedand replated onto coverslips. At the end of the experiment, the cells adherent to thecollagen gels were harvested as described in Materials and Methods section D. (1.)and plated onto coverslips. The following day all coverslips were fixed and stored in-20°C methanol until they were stained for keratin by immunofluorescence withanti-keratin antibodies AE1 and AE3 as described in Materials and Methods B.(2.)(b.).(4.) Culture Purification by Percoll Density Gradient Centrifugation(a.) Purification of Mixed HOSE CulturesA second approach to purifying mixed epithelial and fibroblastic HOSE cultures wasto separate epithelial and fibroblastic cell populations by Percoll density gradientcentrifugation. Mixed cultures were obtained from 6 different cases (Anon, Solo,Sub, Mac, Bent, Welch) ranging from p1 to p4. As controls, a line of mesothelialcells (LP-9) (Connell & Rheinwald ‘83) was used to represent an equivalent of theepithelial component of HOSE cultures while human dermal fibroblasts (HUDF) or17rat lung fibroblasts (RLF) were used to represent equivalents of the fibroblasticcomponent of the HOSE cultures.In addition to contaminating stromal fibroblasts, the addition of epidermal growthfactor (EGF) to HOSE cultures causes a modulation of epithelial HOSE cells to afibroblast-like morphology (Siemens & Auersperg ‘88). This phenotypic modulationmakes it difficult to identify EGFIHC treated fibroblast-like HOSE cells fromcontaminating stromal fibroblasts. Experiments to test if the EGF/HC-treatedfibroblast-like HOSE cells have different cell densities from epithelial HOSE cells orfrom stromal fibroblasts were performed in the following manner. LP-9 cells werecentrifuged in Percoll under the following conditions: 1.) LP-9 cells maintainedwith 20 ng/ml EGF + 0.04 ug/mI HG for 5 days prior to Percoll centrifugation toestablish fibroblastic LP-9 populations (Connell & Rheinwald ‘83) reminiscent ofEGF/HC modulated HOSE cells; 2.) LP-9 cells were maintained without EGF/HG torepresent epithelial HOSE cells; and 3.) LP-9 revertants, that is cells modulated to afibroblastic-like morphology when grown with EGF/HC and then allowed to revert toan epithelial morphology following 3 days in culture without EGF/HC. HUDF werealso treated with and without EGFIHC to see if EGF/HC altered HUDF cell densities.(b.) Continuous Percoll GradientsStock iso-osmotic Percoll (SIP) was made by mixing 9 parts Percoll (Pharmacia,Uppsala, Sweden) with 1 part of 1 Ox complete Hanks’ Balanced Salt Solution (HBSS,GIBCO). Twenty five percent and 40% SIP were prepared by further dilution withlx HBSS. Thirty six ml of 25% or 40% SIP was placed in a 40 ml round bottomedpolycarbonate test tube (Nalgene, Richester, NY). RLF, mixed HOSE, HUDF or LP-9cells, each in 100 ul HBSS supplemented with l%FBS, were gently layered on top ofthe Percoll solution and centrifuged at 10,000 g for 30 minutes at RT in a SorvallSuperspeed RC2-B centrifuge equipped with an SS34 rotor. The density of the cellbands was determined with density marker beads (Pharmacia) in parallel gradients.Following centrifugation, each cell band was collected with a Pasteur pipet andplated directly, without washing, into tissue culture dishes lined with glasscoverslips. The following day the cultures were examined morphologically and thecoverslips fixed in -20°C methanol and stained by immunofluorescence for keratinas described in Materials and Methods B.(2.) (b.).18(C.) Percoll Decontamination of Mold Infected CulturesIn the course of one of the Percoll experiments described above, it was noted thatPercoll treatment appeared to separate cells from mold infections. To examine thepossible use of Percoll centrifugation as a means to decontaminate cultures of moldand yeast infections more closely, decontamination experiments were carried out ontwo cell types: an immortalized, non-tumorigenic rat ovarian surface epithelial cellline ROSE 239 (Adams & Auersperg ‘85) and HUDF. ROSE 239 grow as epithelialmonolayers and produce basement membrane and interstitial extracellular matrixcomponents (Auersperg et al. ‘91b). All cultures were maintained in the absence ofany fungicides.An environmental mold and an environmental yeast were obtained by exposingopened petri dishes of culture medium to the circulating air within a tissue cultureincubator. The infecting mold was identified as belonging to the fungal Cladosporiumspecies and the yeast was identified as Torulopsis candida. Both occur commonly inthe environment. On two separate occasions, confluent ROSE 239 and confluent HUDFcultures were intentionally infected with mold and yeast for 24 hours. Contaminated,as well as uncontaminated cultures, were prepared for Percoll treatment by rinsingthe cultures three times with complete HBSS and resuspending 3-9 x 106 singlecells in 100 ul of complete HBSS supplemented with 1% FBS. These suspensions ofROSE 239 and HUDF were gently layered on 40% SIP and centrifuged as described inMaterials and Methods A.(4.)(b.). Following centrifugation the cells were counted,tested for viability by trypan blue exclusion, and plated directly, without washing,into tissue cuture dishes. To compare the effects of high speed Percoll centrifugationon cell morphology, growth, and extracellular matrix production with the effects ofcentrifugation as used in routine subculturing, uninfected ROSE 239 and HUDF cellswere spun for 5 minutes at bOg in a counter-top clinical centrifuge. These cells,designated LoC, were compared with uninfected, mold-infected, and yeast-infectedcultures centrifuged at 10,000g in 40% SIP, designated HiC, HiM, and HIYrespectively. To ensure that decontamination had taken place following Percolltreatment, HiM cultures of ROSE 239 and HUDF were maintained for up to one monthwithout antifungal agents.To assess the effects of the various centrifugation treatments on cell growth, ROSE239 and HUDF cells collected following centrifugation (H1C, LoC, and HiM) wereplated at 5000 cells/12 mm well. Cultures were maintained for 12 days and themedium was changed as required. Cells from triplicate wells were harvested atintervals and counted.19ROSE 239 and HUDF, collected from all centrifugation treatments, were grown onglass coverslips and stained by immunofluorescence for extracellular matrixproduction to examine the effect of centrifugation treatments on the differentiation ofROSE 239 cells and HUDF cells. HUDF cultures were supplemented daily for one weekwith 50 ug/mI ascorbic acid (GIBCO) and HUDF cultures and ROSE 239 cultureswere stained for collagen type I as described in Materials and Methods B. (2.) (a.).ROSE 239 cultures were also stained for laminin as described in Materials andMethods B. (2.) (c.).(5.) HOSE Cultures Established by Scraping Biopsy MaterialAn improved method to culture HOSE cells was established by taking advantage ofthe tenuous attachment of HOSE to its underlying tissues (Kruk et al. ‘90). By thismethod, the biopsy specimen was examined under a dissecting microscope and areasdevoid of anomalies, blood vessels, and papillae were selected for culture. The entirespecimen was rinsed with medium, held surface down over a 35 mm culture dishcontaining two ml of culture medium and the ovarian surface was scraped firmly twoto three times with a white rubber scraper (Canlab, Mississauga, ONT) attached to aglass rod. Scraping generated sheets of HOSE cells and these were rinsed from thescraper and the biopsy surface into one or two culture dishes (figure 1). Like theexplantation method, the medium used for the transport and the rinsing of thespecimen was collected, centrifuged, and cells plated into a separate 35 mm dish.Cultures were left undisturbed for 48 hr and incubated as described in Material andMethods A. (6.).(6.) Media and Culture ConditionsAll cultures were incubated at 37°C in a humidified incubator with 5% C02:95%air. Initially, HOSE cells were routinely grown and maintained in Medium 199:MCDB202 (Irvine Scientific, Irvine, CA) (1:1) with 15 % fetal bovine serum (FBS,Hyclone, UT) (199:202/15%FBS) supplemented with either 25 ug gentamicin(GIBCO)/ml or 100 IU penicillin (GIBCO) and 100 ug streptomycin (GIBCO). In thecourse of this study, Medium 202 became unavailable commercially, so Medium 202was replaced with Medium 105 (Sigma) (personal communication, Dr. J.Rheinwald) (199:105/15%FBS). In some cases, 20 ng/ml epidermal growth factor(EGF, Daymar Laboratories, Toronto, ONT) and 0.04 ng/mI hydrocortisone (HC,Sigma) were added to cultures to increase the growth rate and growth potential(Siemens & Auersperg ‘88).20For culture in low-serum containing medium HOSE cells were maintained inMedium 199:105 (1:1) supplemented with 1%FBS (199:105/1%FBS) or 0.5%FBS (199:105/0.5%FBS). For culture in medium containing 0.05% serumproteins (199:105/PC-i) Medium 199:105 (1:1) was supplemented with 2%Pederson’s fetuin (PC-i, Ventrex, 500ug/ml), 15 ug/mI insulin (Sigma), 200ug/ml transferrin (Sigma), 0.33 ug/mI ethanolamine (Sigma), 20 ng/mlphosphatidylcholine (Sigma), and 0.1 ug/mI lipoic acid (Sigma) (Elliott &Auersperg ‘90). For culture in defined medium (199:105/DM) (Elliott &Auersperg ‘90), HOSE cells were maintained in Medium 199:105 (1:1)supplemented with insulin (1 5ug/ml), transferrin (200ug/ml), ethanolamine(0.33ug/ml), phosphatidylcholine (200 ng/ml), lipoic acid (0.lug/ml), HC (400ng/ml), and purified fetuin (Hyclone, 500 ug/ml).The following cell lines were also used in this study: 1.) the mesothelial cell line,LP-9 (from Dr. J. Rheinwald) and human dermal fibroblasts (HUDF) maintained inmedium 199:105/15% FBS; 2.) immortalized HOSE cells (IOSEVan) maintained inmedium 199:105 supplemented with 5% FBS; 3) rat ovarian surface epithelial celllines 199 (ROSE 199) and ROSE 239 (Adams & Auersperg ‘85), ROSE 199subclones Eii/A4 and C8/D10, cervical carcinoma cells C4-l and C4-ll (Auersperget al. ‘89) and rat lung fibroblasts (RLF) were maintained in Waymouth’s medium752/1 (WM) (Flow laboratories, VA) supplemented with 10% FBS.To harvest or subculture cells, cultures were dissociated in 0.06% trypsin(250:1, GIBCO) and 0.01% ethylene-diaminetetra-acetic acid (EDTA, BiologicalResearch Laboratories, Burlington, ONT) in calcium/magnesium-free Hanks’balanced salt solution (HBSS) and spun at bOg in a clinical centrifuge for 5minutes. HOSE cells, LP-9 cells, lOSEVan cells, and HUDF cells were frozen underliquid nitrogen in WM + 25% FBS + 10% dimethylsulfoxide (DMSO, BDH) while allother cell types were frozen under liquid nitrogen in WM + 10% FBS + 10% DMSO.All cell counts were made using a hemocytometer and cell viability was determinedby trypan blue dye exclusion. Cultures were routinely examined using a Wild M40inverted photomicroscope or a Leitz Laborlux K inverted photomicroscope andphotographed on Kodak 2415 Technical Pan black and white film.B. Culture CharacterizationFor histological, immunocytochemical, and immunofluorescent studies cells weregrown to the desired densities on glass coverslips. HOSE cultures between passages210-4 were considered early passage cultures, while those between passages 6-10were considered late passage HOSE cultures. Stained specimens were examined eitheron a Zeiss Photomicroscope II or on a Zeiss Axiophot photomicroscope andphotographed using Kodak T-Max black and white film or Kodak Ektachrome colourslide film.(1.) HistologyThe histochemical procedures used were those described by Culling (‘74).(a.) Oil Red 0 for Neutral FatsCultured HOSE cells, from 5 separate cases, 2 early passage and 3 late passagecultures, were stained with Lillie and Ashburn’s Isopropanol Oil Red 0 method asdescribed in Culling (‘74). Immediately prior to use a saturated stock solution of0.5% Oil Red 0 in isopropyl alcohol was diluted 3:2 with distilled water and filteredwith Whatman #1 filter paper. Unfixed cultures were rinsed briefly with HBSS,stained for 20 minutes with the diluted solution of Oil Red 0, rinsed quickly with70% alcohol, washed for 3 minutes in running tap water, counterstained lightlywith hematoxylin for 2 minutes, rinsed in running tap water, and mounted inglycerine jelly or Gelvatol pH 7.2 (Monsanto) (O’Guin et al. ‘85).(b.) Periodic Acid-Schiff (PAS) and Amylase + PAS for Neutral Sugars and GlycogenCoverslips from 2 early passage and 3 late passsage HOSE cultures were fixed informalin, brought to water, and oxidized with freshly made 1% aqueous periodic acidfor 10 minutes. The specimens were washed in running tap water for 10 minutes,treated with pararosaniline Schiff reagent for 30 minutes, washed again in runningtap water for 10 minutes. The coverslips were counterstained briefly withhematoxylin, differentiated in base, rinsed in water, dehydrated in alcohol, clearedin xylol, and mounted.Treatment with human saliva (amylase) for 60 minutes prior to periodic acidoxidation constituted glycogen digestion so that staining attributable to glycogen wasselectively removed. Control slides consisted of sections of human ovarian tissueobtained from the Dept. of Pathology, UBC.(2.) Immunofluorescence Staining(a.) Collagen Types I, Ill, IVTo promote collagen secretion, HOSE cells of both early passage (4 cases) and latepassage (3 cases) as well as HUDF were grown to confluence and maintained for one22week with daily additions of 50 ngfml ascorbate (Gibco). The cultures were fixed for15 minutes in 3.7% paraformaldehyde-PBS. After a 30 minute PBS wash, thecultures were treated with 0.1% Triton X-100 (Sigma) and then rinsed. Thecultures were then incubated appropriately for 15 minutes at room temperature in5% normal rabbit serum or 5% normal goat serum (both from JacksonImmunochemicals, PA) followed by 60 minutes at 37°C with appropriate primaryantiserum. Primary antisera included sheep anti-type I collagen (1:200), rabbitanti-type IV collagen (1:50) (previously absorbed with 0.1 mg laminin(Sigma)/mI antiserum and 0.1 mg gelatin (Sigma)/ml antiserum) (both anti-type Iand anti-type IV antiserum kindly provided by Dr. H.K. Kleinman), or goat anti-typeIII collagen (1:100) (Southern Biotechnologies, Burmingham, AL). After a 30minute wash in PBS the cultures were stained with fluorescein isothiocyanate(FITC)-conjugated rabbit anti-sheep lgG (1:400) (Miles Scientific, Naperville,IL), rhodamine isothiocyanate (ROTC)-conjugated goat anti-rabbit lgG (1:75)(Cooper Biomedical, Mississauga, ON), and FITC-conjugated rabbit anti-goat IgG(1:100) (Miles-Yeda Ltd., Elkhart, md.) respectively for 60 minutes. Following a30 minute wash in PBS the specimens were mounted in Gelvatol pH 7.2 (O’Guin et al.‘85). Antibody controls consisted of specimens stained with normal serum instead ofthe primary antiserum. ROSE 239 and HUDF served as control cells.(b.) KeratinAs described previously (Auersperg et al. ‘84), the cultures were rinsed in PBS,fixed and stored in -20°C methanol (MeOH). The cultures were permeabilized with-20°C methanol:acetone (1:1) for 5 minutes and air dried. They were rehydrated inPBS for 30 minutes at room temperature and incubated at 37°C for 1 hour with themouse monoclonal anti-keratin antibodies (hybridoma-conditioned medium) AE1(1:5) and AE3 (1:2), generously provided by Dr. T.-T. Sun, or AE1 + AE3 (1:100)(Hybritech, SanDiego, CA). Regardless of the source, AE1 and AE3 antibodies werediluted in PBS containing 1% BSA (Sigma). Following incubation with the primaryantibodies, the cells were washed for 30 minutes with PBS at room temperature,incubated with ROTC-conjugated goat anti-mouse lgG (1:20) (Hyclone) for 1 hour at37°C, washed again for 30 minutes and mounted in Gelvatol pH 6.5. Controlsconsisted of cells stained with PBS instead of primary antibody. C-4l or C4-II cellsserved as positive controls and HUDF served as negative controls. Occasionally,cultures of human amniotic cells, which contain both keratin-positive epithelialcells and keratin-negative cells, were used as controls.23(c.) LamininCells from 6 early passage HOSE and 5 late passage HOSE cases were stained forlaminin (LN) using a modification of Auersperg et al. (‘89). Coverslips were fixedin 3.7% paraformaldehyde/PBS, rinsed in PBS, permeabilized by treatment with0.1% Triton X-100 for 3 minutes at room temperature, rinsed again in PBS, andthen incubated with 5% normal goat serum for 15 minutes at room temperature. Thecoverslips were drained, incubated with rabbit anti-mouse laminin antiserum(1:150) (Biological Research Laboratories) for 1 hour at 37°C, washed in PBS for30 minutes. The cells were then incubated with ROTC-conjugated goat anti-rabbitlgG (1:200) for 1 hour at 370C, rinsed for 30 minutes with PBS, and mounted inGelvatol pH 7.2. Controls consisted of cells stained with normal rabbit serum insteadof primary antiserum and positive controls consisted of the rat ovarian surfaceepithelial cell line, ROSE 239, and HUDF were negative controls.(d.) Plasminogen Activator Inhibitor (PAl-i)Early (from 2 cases) and late (from 3 cases) passage HOSE cultures were stainedfor PAM according to the method of Rheinwald et al. (‘87). Cultures were fixed in-20°C methanol, air dried, and rehydrated in PBS as described for keratinimmunofluorescence. The coverslips were incubated with rabbit anti-PAl-iantiserum (1:25) (provided by Dr. Rheinwald) for 30 minutes at roomtemperature, washed in PBS for 15 minutes, incubated with ROTC-conjugated goatanti-rabbit lgG (1:100) for 30 minutes at room temperature, washed for 30minutes, and mounted in Gelvatol pH 7.4. Controls consisted of cells stained withnormal rabbit serum instead of primary antibody and cultures of human mesothelialcells LP-9 served as positive controls.(3.Staining for Mucin and KeratinHOSE cultures were doubly stained for the epithelial markers, mucin and keratinaccording to the method of Wilson et al. (‘83). HOSE cultures from Iwo p3 and threep10 cases were rinsed with HBSS and fixed in either 4°C 70% alcohol or -20°CMeOH. The coverslips were brought to 95% alcohol and blocked for endogenousperoxidase activity with 0.6% hydrogen peroxide (H2O2, BDH) in MeOH for 30minutes at room temperature and then washed in running water for 10 minutes andwere subsequently incubated with human saliva (amylase) for 1 hour at 37°C andthen washed again for 10 minutes in running water. The coverslips were thenoxidized in 1% periodic acid for 10 minutes, washed in running water for 1024minutes, stained in Schiff reagent for 30 minutes and then washed again for 10minutes in running water. The cultures were then rinsed in 0.05M Tris pH 7.2 andincubated with 5% normal goat serum in 0.05M Tris. The coverslips were drainedand incubated at 37°C for 1 hour with mouse anti-human keratin antibodies AE1(1:5) and AE3 (1:5). This was followed by a brief rinse with Tris, an incubation for1 hour at 37°C with horse radish peroxidiase (HRP) conjugated goat anti-mouse IgG(1:100 or 1:200) (Biorad, Richmond, CA), a rinse with Tris, and the reactiondeveloped with 3,3’diaminobenzidine (DAB, Sigma). The coverslips were develpedfor 1-20 minutes at room temperature in 0.24 g DAB dissolved in 400 mlTris:0.24 ml of 50% H202. The reaction was stopped with water and the coverslipswere counterstained in haematoxylin, rinsed in water, blued in 1.5% sodiumbicarbonate (Sigma), washed in water, dehydrated, cleared, and mounted inPermount. Controls consisted of cells stained with PBS instead of primary antibodyas well as cultures of C4-l cells and frozen sections of human cervix and oviductwhich served as controls for keratin and mucin.(4.) ScanninQ Electron Microscopy and Transmission Electron MicroscopyFor transmission electron microscopy (TEM), specimens were fixed in 2.5%glutaraldehyde/0.1 M P04 buffer, post-fixed in 1% 0s04/0.2M P04 buffer,dehydraded in an ethanol and propylene oxide series, embedded in Epon 812,sectioned, stained 15-20 mm in uranyl acetate (2% in water) and lead citrate, andexamined on a Zeiss EM 30 electron microscope. For SEM, specimens were fixed inthe same solutions used for TEM. They were dehydrated in a graded series of aqueousethanol solutions to 100% ethanol. The specimens were then critical point dried in aCPD 020 critical-point drying system using liquid carbon dioxide. Samples weremounted on stubs with silver paint, coated with gold using a SEMPREP 2 sputtercoater, and examined in a Cambridge Steroscan 250-T electron microscope.(5.) SDS-PAGE Western Immunoblots for Keratin and VimentinPreparations of enriched intermediate filaments (Achtstaetter et al. ‘86) wereobtained from two cases of confluent HOSE cultures at early passage (p3) and fromtwo cases of confluent HOSE cultures at late passages (p10). They were subjected tosodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) in 8.5% gels(Laemmli ‘70) using the high ionic strength buffer system of Thomas and Kornberg(‘75). After electrophoresis, the gels were washed for 30 minutes at RT in transferbuffer and then, electrophorectically transferred to nitrocellulose (Towbin et al.25‘79). The nitrocellulose was dried for 30 minutes, blocked for 1 hour at RT with 20mM Tris-HCI buffer (TBS, pH 7.5) plus 3% BSA and 1% normal goat serum, andthen incubated with either mouse monoclonal antibodies to all human acidic (AE1)keratins except #18 (1:10), mouse monoclonal antibodies to all human basic (AE3)keratins (1:10), or with mouse monoclonal anti-cytokeratin No. 18 (1:100)(Boehringer Mannheim Biochemica, Laval, QUE) in TBS containing 0.05% Tween20 (TTBS) plus 1% BSA overnight at 4°C. After washing for 10 minutes in TTBS,the nitrocellulose was incubated for two hours at RT in goat anti-mouse horse radishperoxidase- (HRP) conjugated lgG (heavy and light chains, Biorad) diluted at1:2000. After 2x10 minute TTBS and lxlO minute TBS washes, the keratin bandswere visualized with 4-chloro-1-naphthol (Sigma) prepared fresh by dissolving60 mg in 20 ml cold methanol and mixing with 100 ml of TBS containing 60 ul of a30% stock solution of hydrogen peroxide. The reaction was stopped with water.For vimentin detection, the nitrocellulose sheets were blocked with 1% normalrabbit serum + 3% BSA in TTBS, incubated with goat anti-vimentin (1:10)(Polysciences Inc, Warrington, PA), followed by incubation with rabbit anti-goatalkaline phosphatase-conjugated lgG (1:2000) (Southern Biotechnologies), and thereaction developed in 100 ml of 0.1M NaHCO3, 1 mM MgCI.6HO carbonate bufferpH 9.5 containing 15 mg 5-bromo-4-chloro-3-indolyl phosphate p-toluidine salt(BCIP, Sigma) in 1.0 ml N,N-dimethylformamide (DMF) plus 30 mg nitro bluetetrazolium (NBT, Sigma) in 1.0 ml 70% DMF. The reaction was stopped withEDTA/Tris-HCI pH 7.5.The nitrocellulose sheets were later stained with amido black (0.1% inmethanol:acetone:water, 45:10:45, vol%) to locate molecular weight markers. Inaddition to the molecular weight markers, the human cervical carcinoma cell line(C4-l) was used as a positive control for keratin and a negative control forvimentin. C4-l cells express keratins #5,6,8,16,18,19 (Auersperg et al. ‘89).HUDF were used as negative controls for keratin and positive controls for vimentin.C. Ovarian-Derived Extracellular Matrix and HOSE Organoids(1.) Preparation of ROSE 199-ECMTo pursue HOSE-ECM interactions, it was decided to characterize an ovarianderived matrix, isolate such a matrix, and examine the response of HOSE cells tosuch a matrix, It was thought that HOSE might mimic the normal resting state of theovarian surface if maintained on such an ovarian-derived matrix. To this end, ECM26was derived from the immortalized, non-tumorigenic rat ovarian surface epithelialcell line (ROSE 199). ROSE 199 secrete an extracellular matrix which, at theultrastruclural level, contains as one of its major components banded fibrilscharacteristic of interstitial collagen. ROSE 199 cells grow as a characteristiccobblestone monolayer, but, when crowded, these cells form ridges and multiple celllayers by virture of their ability to deposit ECM between cell layers. ROSE 199subclone E11/A4 cells form ridges when crowded, but the ROSE 199 subcloneC8/D1 0 forms cobblestone monolayers and remains monolayered when crowded.For ECM production, cells were seeded sparsely in 35 mm petri dishes (Costar)or 12 well plates (NUNC) in WM + 10%FBS + 25 ug of gentamicin/mi (Gibco). Itwas found that ROSE 199 and ROSE 199 subclones E11/A4 and C8/D10 grew equallywell in medium supplemented with either 10% FBS or 10% OMNI serum (AdvancedBiotechnologies Inc, Columbia, MD), so when cultures became confluent, they wereswitched from WM supplemented with 10% FBS and gentamicin to WM supplementedwith 10% OMNI and gentamicin. The cultures were maintained for 6 weekspostconfluence, and the medium changed as required.Cell-free ECMs were prepared from ROSE 199 and ROSE 199 subclones E1i/A4and C8/D10 by modifications of one of three treatments, two of which were chemicaland one which was a non-chemical procedure. Cell-free ECMs have been previouslydescribed by treatment with ammonium hydroxide (AH) (Gospodarowicz ‘84),sodium deoxycholate (DC) (Liotta et al. ‘80), and repeated freeze thaws (FT)(Carley et al. ‘88). ROSE cultures were rinsed briefly with sterile ddH2O and thentreated with either 20 mM AH for 5 minutes, 1% DC for 5 minutes, or three freezethaws. The ECMs were thoroughly washed with ddH2O and then with complete HBSSand finally, stored at 4°C for up to one week.To estimate the amount of ECM material produced by six week old post-confluent35 mm cultures, triplicate ECMs prepared by all three treatments (AH, DC, FT)were dried in a 2000C oven and weighed.(2.) Characterization of ROSE 199-ECMROSE 199-ECMs were stained in situ or as frozen sections of detached matrix.ROSE 199-ECMs were examined by both SEM and TEM and ECMs were stainedhistologically according to the procedures of Culling (‘74) or Kiernan (‘81) foracidic and suiphated sugars, collagen, and reticulin. By immunofluorescencemicroscopy ECMs were stained for laminin and collagen type I to see if ROSE 199-derived ECM, like the ECM produced by HOSE cells, contained both basement27membrane and stromal matrix components. ROSE 1 99-ECMs were also either stainedfor DNA or doubly stained for actin and laminin. The presence of actin or DNA wouldbe an indication of remaining cellular debris in the ROSE 199-ECM preparations.Additionally, ROSE 199-ECMs derived from all three preparative treatments wereprobed for laminin, fibronectin, and collagen types I, Ill, and IV by westernimmunoblotting.(a.) Alcian Blue for Sulphated and Non-Sulphated Acidic SugarsFormalin fixed sections of ROSE 199-ECM were rinsed in water, stained witheither freshly filtered 1% Alcian Blue 8 GX in 3% acetic acid (pH 2.5 ) or 1%Alcian Blue 8 GX in 0.1 N hydrochloric acid (pH 1 .0) for 30 minutes. The specimenswere then dehydrated, cleared, and mounted.(b.) Masson for CollagenFrozen ROSE 199-ECMs were sectioned, fixed in formalin, rinsed in water, stainedfor 20 minutes in Weigert’s iron haematoxylin, washed in water and differentiatedin 1% acid alcohol. The sections were washed again in water, treated with 1%phosphomolybdic acid for 5 minutes and then stained for 5 minutes with 2.5%aniline blue in 2.5% acetic acid. The sections were differentiated in 1% acetic acidfor 1-2 minutes, then dehydrated, cleared, and mounted.(c.) Silver Staining for ReticulinFrozen sections of ROSE 199-ECM were fixed in Bouin’s fixative and stained forreticulin according to the method of Gordon & Sweets ‘36 (described in Kiernan ‘81).Sections were brought to water, oxidized for 1 minute in acid permanganate, washedin water, dipped in 1% oxalic acid, washed again, and treated for 10 minutes with aniron alum solution. Following further washing, the specimens were immersedbriefly in an ammonical silver solution, rinsed in water, placed in a formaldehydereducer solution for 30 seconds, and washed once again in water. The sections werethen toned in 0.2% yellow gold chloride solution, washed in water, immersed in asodium thiosulphate solution for 3 minutes, washed in water, and, finally,dehydrated and mounted.(d.) Double Staining for Actin and LamininROSE 199-ECMs derived by treatment with ammonium hydroxide were doublystained for actin and laminin. Laminin staining would identify ROSE 1 99-ECM while28actin staining would serve as an indication of residual cellular debris present in ECMpreparations. Frozen sections were fixed in paraformaldehyde (3.7%) for 15minutes at RT and then plunged into 20o C acetone for 5 minutes. The slides werethen air dried and rehydrated for 10 minutes with PBS and incubated with 5%normal goat serum at RT for 15 minutes. Sections were then doubly stained withanti-laminin (as in Material and Methods 2. (c.)) and with 1.luM NBD-phallacidin(NDB, Sigma) for 20 minutes at RT. After washing, the sections were incubated withsecondary antibody, goat anti-rabbit ROTC-conjugated lgG (1:200), washed, andmounted. Control sections of ROSE 199-ECM and rat intestine, which served as acontrol, had the anti-laminin antibody replaced with normal rabbit serum and NBDphallacidin neutralized with 82.5 ug/mI phalloidin (PH, Sigma).(e.) Hoechst Staining for DNAFrozen sections of AH-derived ROSE 1 99-ECMs, half of which were treated with25 ug DNAse (Sigma)/ml WM at 37°C for one hour and washed overnight in PBS,were frozen, sectioned, and fixed in formalin. The sections were washed in PBS for30 minutes, stained for 3 minutes at room temperature with 0.02 ug/ml Hoechst33258 bisbenzimide chromosomal dye (Sigma), washed in PBS for 15 minutes, andthen mounted in Gelvatol pH 7.2.(f.) Western lmmunoblots for Laminin, Fibronectin, and CollagenPreparations of ROSE 199 ECM from all treatments (AH, DC, FT) were subjected toseparation by continuous SDS-PAGE in 12% acrylamide gels for fibronectin (EN)and laminin (LN). After electrophoretic transfer to nitrocellulose, the nitrocellulosewas blocked with TBS + 0.5% milk powder (TBS milk) for 1 hour at RT and thenincubated with anti-laminin (1:50) or anti-fibronectin (1:50) (GIBCO) antiserumovernight at 4°C. After washing, the nitrocellulose was incubated with goat anti-rabbit HRP (1:500) (Biorad) in TBS milk. After further washing, LN and EN werevisualized with 4-chloro-1-naphthol prepared as described for keratin westerns(Materials and Methods B. (5.)).To detect collagen subtypes, ROSE 199 ECMs from all treatment types weresubjected to separation by continuous SDS-PAGE in 8% acrylamide gels according tothe method of Grinnell et al. (‘89). After electrophoretic transfer to nitrocellulose,the nitrocellulose was blocked with TTBS + 3% BSA + 1% normal rabbit serum for1 hour at RT and then incubated with anti-collagen antiserum against collagen types I(1:100), III (1:500), IV (1:500) in TTBS + 1% BSA overnight at 4°C. After29washing, the nitrocellulose was incubated at room temperature in rabbit anti-goat(1:2000) (for collagen types Ill, IV) or rabbit anti-sheep (1:2000) (for collagentype I) alkaline phosphatase conjugated lgG (Southern Biotechnologies) diluted inTTBS + 1%BSA for 2 hr. After further washing, the collagens were visualized withBCIP/NBT as described for vimentin westerns (Materials and Methods B. (5.)).(3.) Adhesion AssaysFor a quantative determination of cellular adhesion to ROSE-ECMs, ROSE 199,HUDF, human HT-29 colon carcinoma, human CAOV3 ovarian carcinoma, and humanmelanoma 6278 cells were grown to confluence (i.e. for up to one week) with 100uCi methyl-3H-thymidine (NEN, Boston, MA, 74 GBq/mmol 2.0 Ci/mmol)/ T-25flask of cells. Five thousand to 10,000 radioactively labelled cells were seeded in 12mm wells onto: (1) ECMs derived from six week old postconfluent ROSE 199; (2)ECMs derived from six week old postconfluent ROSE subclone cultures; and (3)uncoated plastic wells. The proportion of cells attached to ECM was determined byliquid scintillography counting of the radioactivity associated with the supernatant,the ECM, and the scraped underlying plastic fractions, collected at different timepoints. Early adhesion was at 10 minutes, intermediate adhesion between 20-30minutes, and late adhesion, when most cells had adhered to plastic, was 45 minutes to8 hours depending on the cell line. Non-adherent cells in the supernatant fraction aswell as those scraped off the plastic underlying the ECM were collected as a pelletresulting from spinning at 10,000 rpm for 5 seconds in an Eppendorf centrifuge.Each fraction, along with the ECM with adherent cells, was removed from triplicatewells and placed into glass scintillation vials. The cells and ECMs were solubilizedwith Protosol (NEN) according to manufacturers instructions. Briefly, 0.5 ml ofProtosol was added to each vial, sealed tightly, incubated at 55°C for six hours withoccasional shaking, followed by the addition of 100 ul of 30% hydrogen peroxide todecolourize the sample, and heated again for 30 minutes at 55°C. Once the sampleshad cooled, 50 UI of glacial acetic acid (BDH) was added to neutralize the Protosol,followed by the addition of 14 ml Econofluor (NEN). The vials were shakenvigorously and allowed to equilibrate for one hour in a Philips PW 4700scintillation counter before counting on a tritium channel. The percentage of cellsattached to ECM at each time point was expressed as the percentage of radioactivityassociated with the ECM fraction over the total radioactivity associated with theentire well and the results were expressed as the mean of triplicate wells ±. thestandard error of the mean.30For verification, concurrent with the liquid scintillography adhesion assays,either the cells remaining in the supernatant were counted with a hemocytometer orthe wells were examined morphologically to estimate the proportion of nonadherantcells.(4.) HOSE OrganoidsThe ROSE 199-derived ECMs were used in conjunction with collagen gels to make acomplex three-dimensional tissue culture model that will be referred to as anorganoid. The organoid set up was designed to mimic conditions that might crudelyreflect a miniature ovary in culture. The components of this model are outlined infigure 26. To ensure that HOSE cells interacted with the components of the organoidmodel and not with the underlying plastic, wells were layered with 1% agarose(GIBCO), allowed to gel at RT, 0.4 ml collagen gel (prepared as described inMaterials and Methods D. (1.)) was laid on top of the agarose layer and also allowedto gel. Finally, an AH-ROSE 1 99-ECM was deposited on top of the collagen gel. SingleHOSE cell suspensions, from ii different cases, varying from 50,000 to 300,000cells were placed on top of the organoid set up. Organoids were incubated understandard conditions and medium changed as required. Organoids were periodicallyexamined morphologically and were maintained until there was no further change inthe amount of contraction. Organoid contraction was quantified by measuring themaximum and minimum axes of the organoids. The average diameter of each organoidwas calculated and expressed as a percentage of the diameter of the culture well. Theorganoids were then fixed in alcohol and examined following either histologicalstaining with haematoxylin and eosin or immunocytochemical staining for keratin.Various culture conditions have been shown to affect collagen gel contraction inother systems (Anderson et al. ‘90, Bell et al. ‘79, Montesano & Orci ‘88, Schafer etal. ‘89). To test similar effects: 1.) organoids were seeded with varying numbers ofHOSE cells; 2.) 24 hours following organoid set up the medium was replaced withmedium containing 20 ng/ml EGF and 0.4 ug/mI hydrocortisone to stimulate HOSEproliferation; 3.) 24 hours after organoid set up, the medium was replaced with thelow serum containing medium 199:105/PC-i (see Methods A. (6.)); and 4.) someorganoids were set up with NIH 3T3 fibroblasts in the collagen gel layer to mimicovarian stroma.31D. Maintainance of HOSE on Various Biological Substrata(1.) Preparation of Placental-Derived. Collagen. Fibrin Clot, and MatrigelSubstrata.The morphology and growth patterns of HOSE cells were compared on plastic (PL),placental-derived ECM (PD, Collaborative Research, Bedford, MA), rat tail tendon-derived collagen gel (COLL) (Emerman & Pitelka ‘77), PD-coated collagen gels(PD+Coll), fibrin clots (FB), and Matrigel (EHS, Collaborative Research)(Kleinman et al. ‘82). These substrata were chosen to represent some of thedifferent types of ECM that HOSE may interact with in vivo. For example, placental-derived matrix (PD), a commercially available laminin rich ECM derived fromplacenta was employed to mimic basemement membranes. HOSE on rat tail tendon-derived collagen gels, which consist mostly of denatured collagen type I, mightrepresent a situation where HOSE are in contact with the tunica albuginea or stromalECM during invasion into the stroma. Fibrin clots were used to mimic the post-ovulatory defect within the follicle and Matrigel, which is made up of basementmembrane components, was employed to mimic the basement membrane that HOSEnormally rests upon. Also, by virtue of its jelly-like nature, Matrigel may mimicloose mesenchymal connective tissue comparable to that found in the developingovary.Collagen gels and Matrigel gels were prepared on ice using pro-chilled pipets andplates to prevent premature gelation. To prepare collagen gels, I part 0.34N NaOHwas mixed with 2 parts lOx medium 199 and then added to 12 parts rat tail tendon-derived collagen gel solution (provided by Dr. J.T. Emerman). This solution wasadded to 24 well plates (NUNC) at 0.4 mI/well and allowed to gel overnight at 37°C.One hour prior to use, the collagen gels were primed with complete medium, whichwas removed prior to the addition of cells. PD-coated dishes and PD-coated collagengels were prepared by thawing stock solutions of PD-derived ECM at 4°C and dilutingthe stock solution to 20 ug/mI with medium 199. On ice, 5 ug/cm2 PD solution wasdispensed per plastic well and collagen gels, allowed to incubate at RT for 2 hoursafter which the excess PD solution was removed. Frozen Matrigel was thawedovernight at 4°C, added undiluted at 0.4 ml well, allowed to gel at 37°C for 30minutes, and primed for one hour with complete medium before the addition of HOSEcells. Fibrin clots were prepared according to a modification of Kaibara et al. (‘81).Stock solutions of fibrinogen (Sigma #F4883) at 20 mg/mI in ddH2O were dilutedin HBSS without calcium and without magnesium to 10 mg/mI and stored at -70°C32until needed. To prepare fibrin clots, stock solutions of 10 mg/mI fibrinogen werethawed and added at 0.4 mI/well using 1 mg/mI poly-D-lysine (Sigma) coated pipetsto prevent the fibrinogen from sticking to the pipets. Immediately afterwards, 0.25ml of thrombin (Sigma #T6759) at 1.2 U/mI was quickly added to the fibrinogen ineach well. The wells were swirled to mix the contents and immediately incubated at37°C for 30 minutes for clot formation. As above, the fibrin clots were primed withcomplete medium for one hour prior to the addition of cells.Single cell suspensions of HOSE from 12 different cases were plated onto thesesubstrata and maintained for up to 12 days. The medium was changed as required. Toharvest cells from the substrata, cells on PL & PD-coated plastic were collected byroutine trypsinization in 0.06% trypsin (1:250)10.01% EDTA. To harvest cellsfrom Matrigel and fibrin clots, the culture medium was removed and the wellsincubated at 37°C for 30 minutes with 0.5 mI/well of undiluted Dispase(Collaborative Research). This treatment dissolved both Matrigel and fibrin clotswithout damaging cells. After the dissolution of these substrata, the cells, often inclumps, were disociated into single cells suspensions by briefly resuspending thecell clumps in trypsin/EDTA. To recover cells from collagen gels, the gels wereincubated at 37°C for 30 minutes in 4 mg/mI collagenase type II (Sigma) inDME:F12 (Sigma) supplemented with 1% BSA (Sigma). The dissolved collagen gelscontained clumps of cells which were resuspended in trypsin/EDTA as describedabove to obtain single cell suspensions.(2.) Morphological ExaminationHOSE cells plated onto substrata were examined by phase microscopy and onparaffin sections stained with either haemotoxylin and eosin orimmunocytochemically for keratin. Additionally, some HOSE cultures maintained onMatrigel were examined with scanning electron microscopy.(3.) Immunoprecipitations for IntegrinsThe integrins expressed by HOSE cells on plastic, collagen gels, fibrin clots, andMatrigel, were examined from two low passage (p3) cobblestone epithelial cases(Van, Boo) and one late passage (p9) atypical epithelial case (Hed). Further, one ofthe low passage cases (Van) was tested again in passage 4. Due to differences in thenumbers of cells obtained from each cases, depending upon the case, 0.5-2.6 x 1 g6cells were seeded per T-25 flask (Corning, NY) previously coated with 4 mIs ofcollagen gel, fibrin clot, or Matrigel. However, each HOSE case was seeded at the33same density on all four substrata. The cultures were maintained for 48 hours. Then,the cells were photographed and surveyed for integrins as previously described(Dedhar et al. ‘87, ‘89, Dedhar & Saulnier ‘90). The cells were harvested, washedthree times in PBS containing 1mM CaCI2 and 1mM MgCl and surface labelled with5-7 ul (0.5mCi-0.7mCi) 1125 (Amersham, Oakville, ONT) for 30 minutes at roomtemperature. The cells were washed three times with PBS and the cell pellet lysed inRIPA containing 1mM phenylmethyl-sulfonyl fluoride (PMSF, Sigma) for 30minutes at 4°C, then spun at 12,000 rpm for 15 minutes at 4°C in an Eppendorfmicrofuge. The lysate was normalized to that volume that would give 6 x 106 cpmper immunoprecipitate reaction. An appropriate volume of lysate was incubated withanti-integrin antiserum for one hour at 40C. The primary antibodies used were:rabbit anti-vitronectin receptor (VNR), mouse anti-VLA2, anti-VLA3, anti-VLA5,anti-34 (all from Telios, San Diego, CA), and rat anti-VLA6 (provided by Dr. C.Damsky, UCSF, CA). Following this incubation, 50 ul of Protein A Sepharose(Sigma) (0.1 g pre-swelled in 15 ml PBS, then pelleted and resuspended to 1 ml)were added to each immunoprecipitate and left to mix overnight at 40C. If the antiintegrin antiserum used had not been raised in rabbits, then 4 ul of rabbit anti-mouse antiserum (BIOCAN) or rabbit anti-rat antiserum (BIOCAN) were added toeach immunoprecipitate at the same time as the addition of Protein A sepharose. Theimmunoprecipitates were each washed twice with 1 mI/wash of RIPA + 0.5M NaCI +PMSF (10 ul of 100mM PMSF per 1 ml RIPA) and followed with two washes oflml/wash of RIPA + PMSF. The antigen-antibody complexes in the final pellet weredissociated by boiling in 45 ul of sample buffer (200mM Tris-HCI pH 6.8containing 3% SDS, 10% glycerol, and 0.001% bromophenol blue). Samples wereanalyzed by electrophoresis in 7.5% SDS polyacrylamide gels followed byautoradiography.For time course studies, a large number of HOSE cells was required so a culture ofVan HOSE (p2) was transfected by CaCl2 precipitation using the SV4O large Tantigen-containing plasmid PX-8 (Fromm & Berg ‘82) at 5 ug DNA per 35 mmculture dish. The transfected cells, IOSEVan, expressed large T-antigen, werecobblestone in morphology, keratin positive, and showed enhanced growth potentialcompared to untransfected cells. IOSEVan were cultured for 48 hours on plastic andcollagen gel coated flasks. Cells were harvested, surface labelled, lysed, andimmunoprecipated with rabbit anti-vitronectin receptor antiserum.34(4i Growth CurvesTo determine growth potential on the various substrata in short term culture,15,000-20,000 cells from 3 HOSE cases (Any, Natty, Mull) were plated per well.Cells were harvested from triplicate wells at various intervals, counted, and countsexpressed as the mean ± SEM.(5i Protease ProductionFor most protease assays, HOSE cells were plated onto PL, PD, FB, PD + COLL,Coil, EHS in complete culture medium 199:105/15%FBS. After 24 hours theculture medium was removed, the cultures washed with HBSS, and the mediumreplaced with: 1.) medium 199:105/1%FBS (3 cases-Any,Natty,Mull); 2.)medium 199:105/0.5%FBS (2 cases-Dose,Pere); or 3.) medium 199:105/PC-i(2 cases-Dose,Pere). Conditioned media were collected following a further 48 hourincubation and tested for chymotrypsin, elastase, trypsin, plasminogen activatorinhibitor, and collagenase type I activities. Further, chymotrypsin and elastaseactivities were compared in 2 cases (Dose,Pere) following 3 (D3) and 9 (D9) daysof culture to see if proteases detected during the first few days of culture could alsobe detected later. Controls consisted of conditioned medium collected from substrata-coated wells alone, but treated identically to coated wells containing HOSE cells.For initial gelatinase assays, cells from 2 cases (Casa & Gar) were plated for 24hr on PL, FB, COLL, and EHS in medium 199:105/15% FBS. The cultures werewashed with HBSS and the medium replaced with 199:105/DM. This conditionedmedium was collected following a further 24 hr incubation, concentrated usingCentricon 30 microconcentrators with a 30,000 MW cutoff (Amicon, Danvers, MA)and subjected to gelatin zymography. For subsequent gelatinase assays, cells fromone case (Cello) were plated for 24 hr on PL, FB, COLL, and EHS in 199:1 05/DM. Asabove, the cultures were then washed with HBSS, incubated for a further 24 hr in199:105/DM which was then collected, concentrated, and subjected to gelatinzymography.(a.) Chymotrypsin, Elastase, and Trypsin Chromogenic AssaysChymotrypsin, elastase, and trypsin-like activities in conditioned media weredetermined by spectrophotometric assay (Uitto et al. ‘89). Peptidase activities wereassayed in 96 well flat bottomed plates (Linbro, Flow Lab) by incubating 50 ulconditioned media with 1 mM chromogenic p-nitoanilide amino acid substrates (madeas 40-50 mM stock solutions in DMSO) in a final volume of lOOul with 0.05M Tris35HCI/0.2M NaCI buffer (pH 7.8) for up to 24 hours at 37°C with continuous shaking.The change in absorbance at 405 nm was recorded spectrophotometrically using aBIO-TEK microplate autoreader EL311 (Mandel Scientific Co. Ltd), was correctedfor cell number and for the level of background activity associated with the controls,and recorded as the relative absorbance at 405 nm. The synthetic peptides Nsuccinyl-ala-ala-pro-phe p-nitroanilide (SAAPNA, Sigma), N-succinyl-ala-alaala p-nitroanilide (SAANA, Sigma), and N-benzoyl-DL-arginine p-nitroanilide(BAPNA, Sigma) were used to demonstrate chymotrypsin-like, elastase-like, andtrypsin-like activities respectively. The results of triplicate wells are reported asthe mean relative change in absorbance per hour per io cells.(b.) Plasminogen Activator AssayConditioned media were also assayed for plasminogen activator (PA) activity usinga casein-agarose diffusion plate assay (Uitto et al. ‘89). Agarose substrate tabletscontaining casein (Biorad) were soaked overnight in dH2O to give a 1% agar gelcontaining casein. The agarose solution was boiled for 3 minutes until the solutionbecame opaque and then allowed to cool to 50°C. Twenty micrograms of plasminogen(Sigma) was quickly added per ml agarose solution. Fifteen ml ofagarose/plaminogen solution was poured per 100 cm plastic petri dish (Corning),the solution swirled to cover the dish, and allowed to gel at room temperature. Oncegelled, wells were bored into the agarose. Thirty ul of conditioned medium were addedper well and incubated for up to 72 hours in a humidified chamber. Whenplasminogen is converted to plasmin by plasminogen activator, it degrades casein andforms a clear area of lysis in the agarose gel. Human urokinase (Sigma) at 0.2 U/mIand 0.02 U/mI was used as standard plasminogen activator activity. The reaction wasstopped by fixing the agarose plates in 4% acetic acid for 1 hour at RT.(c.) Collagenase Type I:3H-Collagen Type I DegradationFifty ul of conditioned medium was incubated with soluble3H-proline-labelledchicken type I collagen (provided by Dr. V.-J. Uitto, UBC, BC) in 0.05M Tris HCIbuffer, pH 7.8, containing 0.2M NaCI and 5 mM CaCl2 for 18 hours at 25°C (Uitto etal. ‘89). The specific acitivity of the labelled collagen was 2 x i04 dpm/mg and toactivate latent collagenase in that might be in the conditioned media, 1 mM paminophenylmercuric acetate (APMA, provided by Dr. V.-J. Uitto) was added to eachassay. The degradation products of collagen were analyzed by SDS-PAGE using 10%acrylimide gels, followed by autoradiography.36(d.) Gelatin ZymographyGelatinolytic activity was assayed by a modification of Heussen and Dowdle (‘80).The conditioned media were mixed with sample buffer and applied directly, withoutprior heating or reduction, to 10% acrylamide gels containing 1 mg/mI 2-methoxy-2,4-diphenyl-3(2H)furanone (MDPF, Calbiochem #444899)-conjugated gelatin(Sigma, #G-6269) (O’Grady et al. ‘84 and provided by Dr. Uitto). Followingelectrophoresis, the SDS was removed from the gels by incubation in 50 mM TrisHCI pH 7.5 containing 2.5% Triton X-100 and 0.02% NaN3 for 30 minutes at RTfollowed by a second incubation at RT for 30 minutes in 50 mM Tris-HCI pH 7.5containing 5 mM CaCI2, 1 uM ZnCl2, 2.5% Triton X-100 and 0.02% NaN3. The gelswere then incubated overnight at 37°C in 50 mM Tris-HCI pH 7.5 containing 0.2MNaCI, 5 mM CaCI2, luM ZnCI2, 0.02% NaN3. The gels were fixed and stained in 50%MeOH/7% acetic acid containing 0.2% Coomassie Brillant Blue G250 (Sigma) andbriefly destained in 10% acetic acid. The gelatinolytic activity was evident as clearbands against the blue background. MDPF conjugated to gelatin provided a means tomonitor gel lysis under ultraviolet illlumination during incubation of the gels.37RESULTSA. Improved HOSE Yield and Tissue Culture(1.) HOSE Cultures Established by Explantation from Biopsy MaterialHOSE from 9 ovarian biopsy specimens was prepared by the explantation method(Table 1). By three weeks of culture, 422 out of 469 explants had cellularoutgrowths (90%). Only 150 (34%) of all the explants produced epithelialoutgrowths. Outgrowths from 113 explants (24%) contained mixtures of epithelialand stromal fibroblasts and 151 out of the total explants (32%) producedcompletely fibroblastic outgrowths. Only 47 (10%) explants failed to grow HOSEcells.Although cells grew out from explants, few cultures contained pure populations ofHOSE cells. Described below are the attempts to overcome the major problemsassociated with this technique which are: 1.) labourious technique; 2.) low HOSEcells yield; and 3.) fibroblastic overgrowth.(2.) HOSE Culture Purification by Differential Adhesion to Collagen GelHOSE cultures (from 2 cases, Solo p1 and Welch p4) that contained mixtures ofepithelial and fibroblastic cells were plated onto unrimmed collagen gels. By onehour post plating most cells had attached to the collagen gels. Keratinimmunofluorescence of the cells recovered from the gels as well as the non-adherentcells remaining in the supernatant revealed mixtures of keratin positive and keratinnegative cells, often regardless of morphology (figure 2 and summarized in table10). Therefore, the epithelial component of mixed HOSE cultures could not beseparated from the fibroblastic component by differential adhesion to collagen gels.Controls of amniotic cells, which contain populations of keratin positive and negativecells, stained as mixtures of keratin positive and negative cells. All coverslipstreated with PBS instead of primary antibody failed to stain.(3.) Culture Purification by Percoll Density Gradient Centrifugation(a.) Purification of Mixed HOSE CulturesFollowing centrifugation, all cell types tested were visible as one or two bands inthe range of 1.05-1.08g/ml (Table 2). HOSE cultures from 6 cases banded variablyranging from a broad band at 1 .05-1 .08g/ml to two bands at 1 .05 & 1 .06-Table1.HOSECulturesEstablishedbyExplantationf,t>mBiopsyMate,ialLpIafi#Outgowths#Epithelial#Mixed#Fibroblasts#NoGrowth(A)onAnon35140210Bent1155310457Dew82139402Dose4439005Felice46530110Mac3011973Mal40917104Myc200785Sub5726192139Figure 2. Keratin Expression in HOSE Cells After Differential AdhesionPhotomicrographs representative of mixed HOSE cell populations derived fromdifferential adhesion to collagen gels and stained by immunofluorescence forkeratin. (A), Immunofluorescence of cells recovered from collagen gelsdemonstrate mixtures of keratin-positive and keratin-negative cells. (B), Phasemicroscopy of the same field as in A. Note that the presence of keratin isindependent of either an epithelial or fibroblast-like cellular morphology. X475.1-Table2.PerollCentrifugationtoSeparateMixedHOSECulturesMixedHOSECultures%SIPCentfugationCellBand(q/mI)Anon2510,000g100’1.05Solo2510,000g100’1.06Solo4010,000g30’1.05Mac4010,000g30’1.05Bent4010,000g30’1.07-1.08Welch4010,000g30’1.05&1.06-1.07Sub4010,000g30’1.05&1.06-1.07Other CellTypesRLF2510,0009100’1.05-1.08HUDF+EGF/HC2510,000930’1.06HUDF-EGFIHC2510,000g30’1.06HUDFrevertants2510,000g30’1.06LP-9+EGFIHC2510,000g30’1.06LP-9-EGF/HC2510,000930’1.06LP-9revertants2510,000930’1.06minutes411.07g/ml. RLF banded in a broad band at 1.05-1.08g/ml and HUDF and LP-9 cellsbanded as a discrete band at 1 .06g/ml. The various EGF/HC treatments of LP-9 cellsand HUDF did not alter the cells’ density in gradient centrifugations.The morphology of each HOSE cell band recovered following Percoll centrifugationresembled its parent culture. That is, bands recovered from mixed HOSE culturesdemonstrated mixed epithielial and fibroblastic phenotypes similar to cellsrecovered following differential adhesion to collagen gels. HOSE cultures stained asmixtures of keratin-positive and keratin-negative cells following centrifugationsimilar to the tests for differential adhesion on collagen gels where keratin stainingwas not restricted to any particular morphological phenotype. That is, somefibroblastic cells stained positively for keratin while others did not. Similarly, someepithelial cells stained positively for keratin and some did not. RLF and HUDFremained fibroblastic after centrifugation and LP-9 cells were fibroblastic orepithelial following centrifugation depending on whether they were maintained withor without EGFIHC respectively. RLF and HUDF failed to stain for keratin. LP-9 cellswithout EGF/HC or revertants uniformly stained positive for keratin followingPercoll centrifugation, while those treated with EGF/HC did not stain. The resultsindicate that centrifugation in Percoll gradients did not separate the stromalfibroblasts in mixed HOSE cultures from the epithelial or fibroblastic HOSEcomponent and that the addition of EGF/HC did not alter cellular density in Percollgradient centrifugations. Coverslips stained with PBS instead of primary antibodyfailed to stain for keratin.(b.) Percoll Decontamination of Mold Infected CulturesWhile centrifugation in Percoll failed to separate stromal fibroblasts from HOSEcells in mixed cultures, it appeared to provide a method to remove mold infectionsfrom cultures. To develop this method, ROSE 239 and HUDF cells were intentionallyinfected with mold and yeast contaminants in an attempt to decontaminate thecultures by Percoll centrifugation. Following centrifugation, all cell types testedwere visible as single bands with the lower limit of the heaviest band approximately1-2 cm above the bottom of the centrifuge tubes. ROSE 239 banded at a density of1.056-1.074 g/ml and HUDF banded at 1.048-1.056 g/ml, while the mold pelletedat the bottom of the test tubes. The percentage of ROSE 239 and HUDF cells recoveredfrom Percoll centrifugation varied between 40-50%. However, the viability was94% or better regardless of the centrifugation treatment employed.42Yeast infections, using Torulopsis candida as a prototype, could not be eliminatedby Percoll gradient centrifugation. Also, Percoll centrifugation did not enhancedecontamination of yeasts when used in conjunction with Fungizone (data not shown).This is similar to the results reported by Behrens and Paronetto (‘84).Figure 3 shows the presence of mold in ROSE 239 and HUDF cultures prior toPercoll centrifugation. Microscopic examination of cells recovered following Percollcentrifugation revealed no contaminants for up to one month in cultures maintainedwithout antifungal agents (figure 3). The morphology of ROSE 239 and HUDF cellsremained epithelial and spindle-shaped, respectively. There was no consistentdifference in growth between the high speed-centrifuged Percoll-treated culturesand low speed-centrifuged controls (figure 4). After either treatment, ROSE 239cells grew to confluence quickly and were subcultured weekly at a split ratio of 1:5while HUDF, which grew more slowly, were subcultured every second week at a splitratio of 1:2.Percoll treatment of mold-infected ROSE 239 and HUDF also did not alter theability of these cells to produce and secrete extracellular matrix. ROSE 239 andHUDF from all treatments, HiC, LoC, and HIM, stained positively for collagen type I(figure 5). Similarly, ROSE 239 retained their capacity to secrete laminin and thepattern of extracellular matrix components was indistinguihable amongcentrifugation treatments (figure 5). All cultures failed to stain when incubatedwith normal serum and secondary antibody and HUDF failed to stain for laminin.(4.) HOSE Cultures Established by Scraping Biopsy MaterialWhile neither differential adhesion to collagen gel nor centrifugation in Percollgradients yielded pure HOSE cultures, scraped-derived cultures were highlysuccessful. In contrast to the explantation method, scraping generates epithelialfragments that settle and attach to plastic culture dishes and produce epithelialoutgrowths (figures 1 and 6). Of the HOSE cultures established from 24 scrapedbiopsies (table 3), 3 (12.5%) were discarded due to contamination and one (4%)was discarded because it failed to provide any HOSE cells. HOSE cultures fromscraped biopsies were scored for growth as described previously (Kruk et al. ‘90). 0denotes no growth; +, some HOSE cells in primary culture, but little if any cellproliferation; ++, proliferation to cells, adequate for short-termexperiments and characterization, but the cells did not reach confluence, +++, cellsreached confluence (i.e. approximately cells/35 mm dish) in 2 to 4 weeks andunderwent up to 12 population doublings in medium 199/MCDB1O5/15%FBS.43Figure 3. Cell Lines Before and After Percoll DecontaminationPhase micrographs of confluent ROSE 239 (A,B) and HUDF (C,D) before (A,C)and after (B,D) centrifugation in Percoll. Xl 50.44Figure 4. Effect of Centrifugation in Percoll on Cell GrowthGrowth curves of ROSE 239 (A) and HUDF (B) following Percoll centrifugationof uninfected cultures (H1C), mold-infected cultures (HiM) and low speedcentrifugation of uninfected cultures (LoG). ROSE 239 and HUDF, collectedfollowing centrifugation (HiC, LoG, and HiM) were plated at 5000 cells/i 2 mmwell. Cultures were maintained for 1 2 days and the medium was changed asrequired. Cells from triplicate wells were harvested at intervals and counted.Triplicate counts are expressed as the mean ± SEM.12010040200205044 A006O0z-.3-JuJC.)0 2 4 6 8 10 12 14015000:b00 2 4 6 8 10 12 14DAYS45Figure 5. Matrix Production Following Percoll DecontaminationLaminin (A) and collagen type I (B,C) in the extracellular matrix of ROSE 239(A,B) and HUDF (C). lmmunofluorescence microscopy. X450.46Figure 6. Establishment of HOSE Cultures by the Scrape Method(A) An ovarian surface epithelial fragment newly detached from the ovary byscraping. Note red blood cells in the background; (B) 5-day-old outgrowth froman epithelial fragment; (C) 10-day-old primary HOSE culture, epithelialfragment in the center. Phase microscopy. Xl 20.4L‘kp•..I47Table 3. HOSE Cultures Established by Scraping Biopsy MaterialCase Morpholoqy in Primary Culture1 HOSE Cell Growth2Ape Flat, epithelial, cobblestoneAny Compact, epithelial, cobblestoneCal Discarded due to contaminaon-Cap Flat, epithelial, cobblestone ++Casa Compact, epitheal, cobblestoneCor Flat, epithelial, cobblestone ++Dose Compact, epithelial, cobblestoneEng Atypical epithelial +FellaHa Flat, epithelial, cobblestone +Hig Flat, epithelial, cobblestone +McHOSE Discarded due to contaminaon -Mull flat, epithelial, cobblestoneNag Flat, epithelial, cobblestone ++Natty Flat, epithelial, cobblestonePere Atypical epithelialRic Flat, epithelial, cobblestone +Ste Atypical epithelial ++Tea FibroblasbcTuc Flat, epithelial, cobblestoneVan Compact, epithelial, cobblestoneVery Flat, epithelial, cobblestoneWee Atypical epithelial ++Yar Discarded due to contaminaon -1= HOSE cultures were scored according to the morphological criteria of Siemens & Auersperg(‘88).2= 0, No HOSE growth; +, some HOSE cells in primary culture, but little if any cell proliferation;++, proliferation to I -i O cells, adequate for short-term experiments and characterizaon, butthe cells did not reach confluence; +++, HOSE cells reached confluence (i.e. approximatelycells/35mm dish in 2 to 4 weeks and underwent up to 12 populaon doublings in Medium199:MCDB 105/15% FBS.4816/24 cases or 67% showed good (++) to excellent (+++) growth with 1 1 cases or46% showing excellent growth, and only 4 cases or 16.5% yielded few HOSE cells(+). Also, HOSE cultures from scraped biopsies were classified according to theirmorphology in primary culture (Auersperg et al. ‘84, Siemens & Auersperg ‘88).The vast majority of scrape-derived HOSE cultures, 19/20 or 95%, of culturesgrew as epithelial cells. Of these epithelial cultures, 20% or 4/20 were compactcobblestone, 55% or 11/20 grew as flat epithlial cells, and 20% or 4/20 grew asatypical epithelial cells. Cells from only a single case were spindle-shaped andfibroblast-like. Therefore, in addition to the improved yield, the scrape method alsogreatly increased the purity of the cultures.In contrast to procedures that use explants (Auersperg et al. ‘84) or enzymaticdissociation (Hamilton et al. ‘80) of the original tissue, the ovarian surface belowthe HOSE remains undisturbed by the scrape method. As a result, contamination bystromal and follicular cells was rare. Occasional groups of contaminating fibroblastscould be removed with a rubber scraper. Histologic examination of scrapedspecimens confirmed that scraping denuded 90 to 100% of the ovarian surface ofHOSE with little disruption of underlying structures (figure 7). Thus, normalsurface-lined HOSE is selected over cyst and invagination-derived HOSE which isfrequently atypical. In addition to the improved yield and greater uniformity of thecultures, the scrape method has the obvious advantages of being considerably fasterand simpler than either explant culture or culture of enzymatically dissociatedtissues and HOSE cultures are now routinely cultured by the scrape method.B. Characterization of HOSE Cells in Culture(1 i To What Extent do Cultured HOSE Cells Reflect Their In Vivo Counterparts?To characterize and compare HOSE cells in culture with what has been reported oftheir in vivo counterparts (McKay et al. ‘61, Blaustein ‘81a, Blaustein & Lee ‘79),HOSE cultures were stained for lipid, mucin, and keratin. Additionally, preparationsof intermediate filaments from HOSE cultures were examined by westernimmunoblotting for keratin subtypes and vimentin.Like their in vivo counterparts (Blaustein ‘81a, Blaustein & Lee ‘79), culturedHOSE cells from both early and late passage stained positively for lipid with Oil Red 0indicating the presence of neutral fats as small red droplets at varying depths in thecytoplasm (figure 8). The amount of positive material varied among cells within49Figure 7. Section of a Scraped Ovarian BiopsyAll HOSE, except for a small area (arrowhead) has been removed. C, inclusioncyst lined with epithelium. Hematoxylin and eosin. X60.:‘:•:;—.(V.L—-—.—50Figure 8. HOSE Cultures Stained with Oil Red 0HOSE cultures stain positively for Oil Red 0 as small droplets that appear singlyor in clusters. X300.vos51cultures from a few droplets scattered in the cytoplasm to small clusters of lipiddroplets to large accumulations found in groups of piled up cells.Cultured HOSE, of both early and late passage, also stained intensely with routinePAS indicating the presence of neutral sugars, like their in vivo counterparts(McKay et al. ‘61). There are three distinct collections of PAS positive material: 1)fine, uniform, pink granular staining; 2) clusters of coarse, magenta granules atvarious locations in cells; and 3) large, dark pink vacuole-like areas adjacent to thenucleus (figure 9A). Following digestion with amylase (figure 9B), the clusters ofcoarse granules failed to stain demonstrating that they were glycogen. Occasionallarge vacuoles also failed to stain, but some mucin, defined as amylase-resistent PASpositive material, was detected in the large vacuoles. The cytoplasms of HOSE cellsin ovarian sections also contained PAS positive amylase insensitive materialindicating that both cultured HOSE cells and HOSE in vivo are variably positive formucin.The epithelial nature of both early and late passage HOSE cells was confirmed bythe presence of keratin similar to previous reports (Auersperg et al. ‘84) and liketheir in vivo counterparts (Czernobilisky ‘85). However, the proportion ofkeratin-positive cells generally decreased with increasing passage (figure 1 0).Also, with increasing passage, HOSE cells modulated from an epithelial morphology toa more fibroblast-like phenotype (figure 10). HOSE cells incubated with PBSinstead of primary antibodies failed to stain for keratin. As controls, C4-l cellsstained positively for keratin while HUDF failed to stain for keratin.Table 4 illustrates the degree of co-expression of mucin and keratin in HOSEcultures. HOSE cells, of both early and late passages, stained variably for mucin.HOSE cultures also contain variable amounts of keratin positive cells (11.25-66%). Mucin was present in less than 10% of HOSE cells, regardless of passage. Thepercentage of mucin vacuoles that were contained within keratin positive cellsranged from 43-100% and was not passage dependent. Furthermore, mucin andkeratin expression were independent of morphology, as these markers of epithelialdifferentiation were found in epithelial and fibroblast-like HOSE cells. For controls,C4-l cells, oviductal epithelium, and cervical epithelium were keratin positive andthe cervical epithelium was also positive for mucin.To determine if the keratin subtypes expressed by HOSE cultures are the same astheir in vivo counterparts as well as to determine if cultured HOSE cells co-expressvimentin and keratin like their in vivo counterparts, preparations of intermediatefilaments were probed by western immunoblotting. As shown in figure 11, low52Figure 9. HOSE Cultures Stained with PAS and Amylase + PASHOSE cultures stained positively with PAS as fine uniform pink granularmaterial, clusters of coarse magenta granules, and large pink vacuole-like areasadjacent to the nucleus (A). Following digestion with amylase, PAS positivematerial (mucin) is restricted to large pink vacuole-like material in the pennuclear region (B). X300.52A.1153Figure 1 0. Keratin Expression in Early and Late Passage HOSE CulturesEarly passage HOSE cells are keratin positive (A) and epithelial (B) while latepassage HOSE cells are largely keratin negative (C) and often fibroblast-like inmorphology (D). A, C, Immunofluorescence microscopy. C, D, Phase microscopy.X300.‘3/,rw.Table4.Co-ExpressionofKeratinandMucininHOSECulturesCase#KeratinPositiveCells(%)#MucinPositiveCells(%)#MucinPositiveCellsthatExpressKeratinU,Anyp3124/243(51)16/243(7)1 16(63)Perep3157/237(66)11/237(5)9/11(81)Gar p1012142(29)2/42(5)2/2(100)Hedplo2W85(24)7/85(8)3/7(43)Tyncp1018/160(11)16/160(8)5/16(38)55Figure 11. Keratin and Vimentin Western ImmunoblotsPreparations of intermediate filaments were subjected to SDS-PAGE followed bytransfer to nitrocellulose. (A-C), lane 1, C4-l; lanes 2&3, early passage HOSE;lanes 4&5, late passage HOSE; lane 6, HUDF. (D), lane 1, HUDF; lanes 2&3HOSE; lane 4, C4-1. Nitrocellulose sheets were immunoblotted for acidickeratins with AE-1 (A), keratin #1 8 (B), basic keratins with AE-3 (C), andvimentin (D). All HOSE cells express keratins of 40, 44, 52, and 54 KD, exceptfor some late passage HOSE cells that lack the 44 KD keratin. HUDF serve asnegative controls for keratin while C4-l cells serve as positive controls forkeratin. All HOSE cells and HUDF express vimentin while C4-l cells do not.cc4AKD 1 2 3 4 5 6B 1 2 3 4 5 644JC 1 2 3 4 5 640—D 1 2 3 457- -56passage HOSE cells expressed keratins of 40, 44, 52, and 54 KDa corresponding tokeratins #7, 8, 18, and 19 characteristic of simple epithelia (Moll et al. ‘82) andidentical to the keratins expressed by the human peritoneal cells, LP-9 (Connell &Rheinwald ‘83) and HOSE in vivo (Czernobilisky ‘85). While keratins derived fromone case of late passage HOSE cells expressed keratins #7, 8, 18, and 19 like lowpassage HOSE cells, the keratins derived from a second late passage HOSE case clearlyexpressed keratins #7 and #18 while keratin # 19 was barely detecteable andkeratin # 8 was absent. This suggests that with increasing passage specific keratinsare gradually lost. These results support the immunofluorescence studies whichdemonstrated decreased staining for keratin with increasing passage (figure 10).The cultured HOSE cells also co-express vimentin along with keratin (figurel 1).Vimentin was demonstrated as a single band at 57 KD. Human dermal fibroblastswere devoid of keratin, but demonstrated the presence of vimentin (figure 11). C4-Icells expressed keratins #5, 6, 8, 16, 18, 19 (figure 11), as demonstratedpreviously (Auersperg et al. ‘89), but C4-l cells did not express vimentin (figure11).(2.) Do HOSE Cells Produce Extracellular Matrix?Further characterization of HOSE cells in culture involved staining byimmunofluorescence for laminin and various collagens to determine if HOSE cellsproduce ECM.Both early and late passage HOSE cells stained positively for laminin (figure 12)as compared to serum controls. Laminin was only present in the cytoplasm of HOSEcultures even where cultures were crowded. As positive controls, ROSE 239 cellsstained intensely for laminin, in both the cytoplasm and in the extracellular matrixas has already been reported (Auersperg et al. ‘91b) while HUDF failed to stain forlaminin.Both early and late passage HOSE cells in culture stained more intensely with anti-collagen type IV antibodies than normal serum controls. Collagen type IV stainingappeared as some staining over the cell bodies, but also as punctate material at theedges of cells, often forming an outline of the cells (figure 13). HOSE cells of bothearly and late passage keratin-positve cultures also stained more intensely for thestromal collagens types I and Ill, than their respective serum controls. Collagen typeIll was present as fine fibrillar material at the cell edges and as punctate andglobular staining over the cell bodies (figure 14). Collagen type I stained as punctatematerial seen within the cell boundary and also occuring beyond the cell boundary57Figure 1 2. HOSE Cells Stained for LamininHOSE cells stain positively for laminin (A,B) compared to non-immune serumcontrols (C,D). A,C Immunofluorescence microscopy. B,D Phase microscopy offields shown in A,C respectively. X400.0IIii.,.‘1if,:’’VtI4I-a,IIV%*I58Figure 1 3. HOSE Cells Stained for Collagen Type IVHOSE cells stain positively for collagen type IV as punctate material outlining thecells (A,B) compared to serum controls (C,D). A,C Immunofluorescencemicroscopy. B,D Phase microscopy of fields shown in A,C respectively. X575.SMPtSaI.59Figure 14. HOSE Cells Stained for Collagen Types I and IllHOSE cells stain more strongly for collagen types I (A,B) and Ill (C,D) thantheir respective serum controls (E,F). The epithelial nature of these HOSEcultures producing stromal collagen types I and Ill was confirmed by thepresence of keratin in parallel cultures (G,H). A,C,E-G, lmmunofluorescencemicroscopy. B,D,H, Phase microscopy. A,8,E, X575. C,D,F, X300. G,H, Xl 00.60suggesting extracellular deposition. These results were consistent not only amongHOSE cases, but also within HOSE cases. Gar HOSE in both p4 and p9 stainedpositively with anti-collagen antibodies against collagens types I and Ill. ROSE 239and HUDF controls also stained more strongly for collagen types I and Ill than theirserum controls. These results indicate that HOSE cells in culture clearly demonstratethe ability to deposit extracellular matrix components and that these components areof both basement membrane matrix and stromal matrix.C. Ovarian-Derived ECM(1.) What Are the Characteristics of the Ovarian-Derived ROSE 1 99-ECM?A spontaneously-derived immortalized, but non-tumorigenic rat ovarian surfaceepithelial cell line, ROSE 199, was used as the material from which to isolate andcharacterize an ovarian-derived ECM. Throughout this study, the growth patterns ofthe cell lines ROSE 199 and their subclones, E11/A4 & C8/D1O remained constantand resembled those reported from the time of inital explantation (Adams &Auersperg 85 and unpublished results). When ROSE 199 cultures became crowded,they multilayered and formed ridged structures (figure 15A). SEM and TEM clearlydemonstrated the multilayered nature of ROSE 199 cultures and the ECMinterspersed between the cell layers (figure 15B, C). Due to the abundant ECMproduced by ROSE 199 cultures, ROSE 199 cells were used to prepare andcharacterize an ovarian-derived ECM.Treatment of ROSE 199 cultures with ammonium hydroxide (AH), deoxycholate(DC), or freeze thaws (FT) resulted in the gradual and progressive rounding andlysis of cells. As the cells lysed, the extracellular material became evident. ROSE1 99-ECM was detected as fine wispy material one week post-confluence and becameabundant with prolonged culture (figure 16). All ROSE 199-ECMs appeared similarwhen prepared by treatment with either AH or DC or FT, except for the presence ofresidual nuclei in FT preparations (figure 16). SEM depicted the ROSE 199-ECM asfibrous material interspersed with globular material (figure 17A). ECMs preparedby all the of three treatments examined at the ultrastructural level demonstratedstriated fibrillar collagenous material of comparable size and shape, averaging 30nm in diameter (figure 17B,C).Sections of ROSE 199 cultures and ECMs of all types of preparations were stainedhistochemically for hematoxylin and eosin, Masson for collagen, and silver stain forreticulin. ROSE 199-ECMs were also stained for non-suiphated and suiphated sugars61Figure 1 5. ROSE 1 99 CulturesCrowded ROSE 199 cultures become multilayered and form ridges (A). SEMfurther illustrates the multilayered nature of ROSE 1 99 cultures and thatfibrillar ECM is deposited between the cell layers (B). TEM also illustrates thefibrillar matrix deposited in ROSE 1 99 cultures (C, From Adams & Auersperg‘84). A, Phase microscopy, X40. B, SEM, X4000. C, TEM, X22,000.62Figure 1 6. Light Microscopic Appearance of ROSE 1 99-ECMIn situ appearance of ROSE 1 99-ECMs prepared by treatment of ROSE 1 99cultures with AH (A), DC (B), and FT (C). Toluidine blue. X250.b’1963Figure 1 7. Ultrastructural Appearance of ROSE 1 99-ECMSEM depicts ROSE 1 99-ECM as predominantly fibrillar with some globularmaterial (A). TEM illustrates the striated fibers, characteristic of interstitialcollagen, in ROSE 199-ECM (B,C). A, X5000. B, Xl 2,000. C, X64,000.64with Alcian Blue at pH 2.5 and 1.0. The abundance of collagen in both ROSE 199cultures and ECMs is seen as broad, uniform bands of staining with eosin and Massonstain (figure 18A-D). ROSE 199-ECMs also stain intensely for reticulin. Thereticulin is evident as fine dark layers between cell layers in intact cultures (figure18E, arrows), but somewhat more dispersed in ECM preparations (figure 18F).ROSE 199-ECMs were found to contain acidic sugars by positive staining with AlcianBlue at pH 2.5 (figure 18G) and that, at least, some of the acidic sugars aresulphated sugars as illustrated by an increased intensity of staining with Alcian Blueat pH 1.0 (figure 18H). ROSE 199-ECMs stained in situ for laminin and collagentype I by immunofluorescence stained more intensely than their respective serumcontrols (figures 19 & 21) demonstrating that ROSE 199-ECM contains basementmembrane as well as stromal matrix components. As controls, ROSE 239 cellsstained positively for laminin and failed to stain when the primary antibody wasreplaced with normal serum. Similarly, HUDF stained positively for collagen type I,but failed to stain when the primary antibody was replaced with normal sheepserum. HUDF also failed to stain for laminin. ROSE 199-ECMs doubly stained forlaminin and actin demonstrated the presence of laminin, but not actin suggesting theabsence of cellular debris in the ECM preparation (figure 20). However, ECMsstained histologically with hematoxylin demonstrated the presence of nuclearmaterial contaminating ECM preparations (figure 18B&D, arrowheads). Thepresence of DNA in ECM preparations was confirmed following staining with Hoechstdye and there was a noticable decrease in DNA following treatment of ECMs withDNAse suggesting that at least some of the residual cellular debris can be removedfrom ECM preparations (data not shown).The presence of laminin, fibronectin and collagen types I and Ill in all three typesof ROSE 199-ECMs was confirmed by western immunoblots. Laminin and fibronectinwere present as single bands on nitrocellulose with molecular weights 400 KD and250 KD respectively (figure 22A, B). Collagen type Ill was visible in western blotsas a single band at 95KD in all ROSE 199 ECM preparations, while the major bandsassociated with collagen type I occured as a doublet at 95 and 92KD compared to the97 and 95 KD of the o1 and a2 chains of human collagen type I standard (figure22C). Collagen type IV was not detected by western immunoblotting in any of theROSE 199-ECMs (figure 22D).ECMs derived from six-week post-confluent 35 mm cultures produced roughly 0.9mg of material from AH and FT preparations, but only 0.5 mg of ECM material fromDC preparations as determined by dry weights.65Figure 1 8. Histochemical Comparison of ROSE 1 99 Cultures and ECMSections of ROSE 1 99 cultures (A,C,E) and ECM (B,D,F,G,H) were stainedhistochemically with hematoxylin and eosin (A,B), with Masson stain forcollagen (C,D), and with silver stain for reticulin (E,F). ROSE 1 99 culturesstained for reticulin were prepared from plastic sections as this preparationclearly demonstrated the fine layers of reticulin in ROSE 1 99 cultures (arrowsin E). ECMs stained uniformly for acidic sugars (G) and as distinct linearmaterial for sulphated sugars (H) with Alcian Blue pH 2.5 and 1 .0 respectively.Note the hematoxylin stained material in ECMs indicative of nuclear remnants(arrowheads in B,D). X250.65A211—,-—166Figure 19. ROSE 199-ECM Stained for LamininROSE 1 99-ECMs stained more intensely for laminin (A,B) than serum controls(C). A, Phase microscopy. B,C, Immunofluorescence microscopy. X300.67Figure 20. ROSE 1 99-ECM Doubly Stained for Laminin and ActinSection of ROSE 1 99-ECM (A-C) stained for actin with NBD-phalloidin (B) andlaminin (C). Controls sections of rat intestine were stained for actin with NBDphalloidin (D), NBD-phalloidin neutralized with phallocidin (E), or treated withphallocidin (F). Control sections of rat intestine stained more intensely forlaminin (G) than serum controls (H). A, Phase microscopy. B-H,Immunofluorescence microscopy. X200.— —68Figure 21. ROSE 1 99-ECM Stained for Collagen type IROSE 199-ECM stained more intensely for collagen type I (B) than serumcontrols (C). A, Same field as B, phase microscopy. B,C, Immunofluorescencemicroscopy. X325.6P169Figure 22. Western Immunoblots for Laminin, Fibronectin, and CollagenROSE 1 99-ECMs from all three preparative treatments were subjected to SDSPAGE followed by transfer to nitrocellulose. The nitrocellulose was blotted forlaminin, fibronectin, and collagen types I, Ill, and IV. (A), lane 1, lamininstandard; lanes 2&6, AH-ROSE 1 99-ECM; lanes 3&7, DC-ROSE 1 99-ECM; lanes4&8, FT-ROSE 199-ECM; lane 5, fibronectin standard. Lanes 1-4 wereimmunoblotted for laminin and lanes 5-8 were immunoblotted for fibronectin.(B-D), lane 1, AH-ROSE 1 99-ECM; lane 2, DC-ROSE 1 99-ECM; lane 3, FTROSE 1 99-ECM; lane 4, collagen type I standard; lane 5, collagen type Illstandard; lane 6, collagen type IV standard. Nitrocellulose sheets wereimmunoblotted for collagen types I (B), Ill (C), and IV (D). All ROSE 1 99-ECMscontain laminin, fibronectin, and collagen types I and Ill, but they lack collagentype IV.00(0Co(0p.p.‘t!TirCoC’4jC)(0c--ILOCo‘IIIC.JCOc’JC-)70(2.) Does ROSE 1 99-ECM Have Biological Activity?Biological activity of ROSE 199-ECM was demonstrated by the adhesion, spreading,and replication of several cell types on ROSE 199-ECM as compared to plastic. Asshown in figure 23, three catagories of activity were found, as determined byenhanced cellular adhesion to ROSE 199-ECM, by liquid scintillography assay: 1) allECM preparations enhanced adhesion (ROSE 199, HT-29); 2) AH-ECM and FT-ECMwere superior to DC-ECM and plastic (CAOV3, melanoma 6278); 3) all ECMpreparations resembled plastic (HUDF). Cell counts and morphological examinationof cellular adhesion to ROSE 199-ECM were in agreement with these results.However, there were no differences in morphological appearance between plastic andECM among the cell lines tested. Further, ROSE 199-ECM supported the growth ofsome cell lines as indicated by the presence of mitotic figures in cultures maintainedon ROSE 199-ECM (figure 24). ROSE 199 cells plated onto ROSE 199-ECM adhered,spread, and grew over the entire ECM so that these cultures soon resembled normalROSE 199 cultures which had never been depleted of cells.To determine if the co-production of basement membrane and stromal matrixcomponents is a feature of ROSE 199 cells and not due do the presence of a mixed,heteogeneous population of stromal and epithelial cells, two ROSE 199 subclones,E11/A4 and C8/D1O were examined. Both clones produce ECM that contains bothbasement membrane and stromal components, like their parental line. The ECMderived from the clones is also biologically active as it also supports the adhesion,spreading and growth of HUDF and ROSE 199 cells (data not shown).(3.) What is the Response of HOSE Cells to ROSE 199-ECM?HOSE atttach and spread on the ROSE 199-ECM, but they do not form a confluent,intact cuboidal epithelium like HOSE in vivo. They remain as single cells dispersedthroughout the ROSE 199 matrix (figure 25).(4). HOSE Organoids: An Ovarian Tissue Culture ModelTo prepare organoids (figure 26), plastic dishes were coated with a nonadhesivesubstance, 1% agarose, to encourage cells to interact with the overlying substratumrather than underlying plastic. HOSE cells and ROSE 199 cells plated onto agarose didnot adhere to the surface and remained as floating aggregates. When transferred toplastic after 24 hr, cells remained viable as indicated by their ability to adhere toand spread onto the plastic (data not shown).71Figure 23. Cellular Adhesion to ROSE 1 99-ECMBiological activity of ROSE 1 99-ECMs was shown by determining the percentageof HUDF, ROSE 1 99, human HT-29 colon carcimona, human CAOV3 ovariancarcimona, and human melanoma 6278 cells which attached to ROSE 1 99-ECMscompared to plastic. Early adhesion was measured 1 0 minutes, intermediateadhesion was measured between 20-30 minutes, and late adhesion was measured,depending upon the cell line, between 45 minutes to 8 hours after plating cellsonto ROSE 1 99-ECMs and plastic. Cellular adhesion on plastic (striped bars) wascompared with cells plated on AH-derived (closed bars), DC-derived (dottedbars), and FT-derived (open bars) ROSE 199-ECMs. All ECM preparationsenhanced adhesion of ROSE 1 99 and HT-29 cells compared to plastic, AH-ECMand FT-ECM enhanced adhesion of CAOV3 and melanoma cells compared to DC-ECMand plastic, and HUDF adhesion to all ECMs was not different from adhesion toplastic.7]AIzLUC)I—IIzLUC)IIIzLii=C)EARLY ADHESION (10 Mm)100806040200HUDF ROSE 199 HT-29 CAOV3 MEL 6278INTERMEDIATE ADHESION (20-30 Mm)100806040200HUDF ROSE 199 HT-29 CAOV3 MEL 6278LATE ADHESION (45 Mm - 8 Hr)100806040200HUDF ROSE 199 HT-29 CAOV3 MEL 627872Figure 24. Cellular Spreading and Growth on ROSE 1 99-ECMThe attachment, spreading, and growth of several cell lines was examined onROSE 1 99-ECM. (A), HUDF attached to ROSE 1 99-ECM and became dispersedthroughout the matrix. (B), HT-29 cells attached and formed epithelialmonolayers on ROSE 1 99-ECM. (C), CAOV3 cells also attached, spread, and grewas indicated by the presence of mitotic figures in the cultures (arrows). A&C,Cresyl violet, X300. B, phase microscopy, X200.rA..14r4t[;:73Figure 25. HOSE Cells on ROSE 1 99-ECMWhile HOSE cells formed epithelial monolayers on plastic (A) they dispersed assingle epithelial cells when plated on ROSE 199-ECM (B). Cresyl violet. X185.It74Figure 26. Organoid - Schematic Set upA diagramatic cross-section through a culture dish illustrates the components ofan organoid. Cultures dishes are first layered with agarose to prevent HOSE cellsfrom attaching to the underlying plastic. A layer of rat tail tendon-derivedcollagen gel is then layered over the agarose and allowed to dry overnight. Thecollagen gel is primed for one hour with medium, then a ROSE 1 99-ECM is placedon top of the collagen gel, and finally, a suspension of HOSE cells is placed on topof the ROSE 1 99-ECM.L)wU)>a)L)wQa)G)a)U)(0—r-.75When HOSE cells were placed on ROSE 1 99-ECM coated collagen gels, the HOSE cellsspread over the entire mixture of ROSE 199-ECM and unrimmed collagen gel andcontracted the mixture into small organoids (figure 27), demonstrating their abilItyto physically remodel ECM. Figure 27A illustrates three organoids in varyingdegrees of contraction and a control well without HOSE cells. Cross-sections oforganoids revealed that HOSE cells envelop the organoid, that the majority of HOSEcells remain on the surface of the organoids with only a few scattered cells foundwithin the stroma, and that HOSE cells in organoids remain epithelial as they retainkeratin (figure 27B). Table 5 summarizes the number of organoids prepared. Cellsfrom three cases, Syk p2, Hack p1, Any p6, failed to contract organoids. Contractionappeared to occur only when ROSE 199-ECM5 were in direct contact with andanchored to the collagen gels. In a few instances (Any p6, Pere p3, Sam p3, Solo p5)ROSE 199 ECMs remained floating when the organoids were prepared. However, theygenerally contracted if these floating ROSE 199-ECMs were anchored into thecollagen gels, usually by placing a cloning cylinder on top of the ROSE 199-ECM for24 hr. The ROSE 199-ECMs used in the organoids set up with Any HOSE cells p6failed to attach to the collagen gels even following attempts to remedy this situation.The degree to which HOSE cells contracted organoids varied among cases. However,the degree of contraction was directly related to the number of HOSE cells seeded perorganoid (figure 28A & table 5). For example, HOSE cells seeded at densities lessthan 50,000 cell per organoid generally did not contract organoids maximally. Theaddition of EGF/HC or reduced serum-containing medium did not affect the degree towhich HOSE cells contracted organoids (figure 28B). However, the inclusion of 3T3fibroblasts within the collagen gel portion of the organoid markedly increased theoverall degree of organoid contraction (figure 28B).D.) HOSE-ECM Interactions(1.) What is the Morphologic Response of HOSE Cells to Substrata?Having demonstrated that HOSE cells can remodel ECM by production of ECMcomponents and by physically modifying (contracting) ECM it was decided to expandupon HOSE-matrix interactions as a means to study HOSE pleomorphism andmorphogenesis. The interactions between HOSE cells and simple substrata wereexamined because the complexity of the organoid model made such analyses difficult.To this end, HOSE cells were cultured on various substrata commonly used in tissue76Figure 27. Contraction into OrganoidsWhen HOSE cells are plated onto ROSE 1 99-ECM coated collagen gels, the cellsextend over both matrices and contract them into organoids. A, photograph ofthree HOSE organoids demonstrating different degrees of organoid contractionalong with one control well devoid of cells (lower left well). The upper right wellillustrates a maximally contracted organoid. Note that in the lower right well thecells are clearly contracting the outer, uniform, opaque collagen gel (arrowhead)about the centrally located dense ROSE 1 99-ECM (arrow). X2.5. B, a crosssection of an organoid stained immunocytochemically for keratin illustrates thatthe many of HOSE cells are on the surface of the organoid and retain keratin.X210.biwSp-ta’ t’-:I77Figure 28. Factors Affecting Organoid ContractionThe effect of cell number on the degree of organoid contraction was comparedamong organoids seeded with 5.7 x i0 cells (open squares), 1.4 x i0 cells(closed diamonds), and 3.0 x 1 0 cells (closed squares) (A). The degree oforganoid contraction was compared among organoids maintained: in199:105:15%FBS (open squares); in 199:105/15%FBS supplemented with 20ng/ml EGF and 0.4 ng/ml hydrocortisone (closed diamonds); in 199:105/PC-i(closed squares); with 3T3 fibroblasts (open diamonds) (B). The degree oforganoid contraction was related to cell number (A) and, of the culture variablesexamined, only the addition of fibroblasts enhanced organoid contraction (B).77A120CCEC80C60402000 2 4 6 8DaysB4-a)Ea)a)L.0120’10080604020 -0-0I..I..2 4 6Days878Table 5. Contraction of HOSE OrganoidsCase # HOSE Cells Seeded per Oraoid # Contracted Orqaoids Degree of Contracon:# Organoids Set up % Oginal DiameterDose p4 200,000 414 10Hack p1 100,000 CV3 100Sub p3 50,000 3/3 75Sykp2 50,000 CV2 100Wdchp4 80,000 212 13Any p6 20,000 CV4’ 100Perep3 100,000 2/32 22Sam p3 100,000 2/32 15Solo p5 100,000 5/73 60Natty p3 50,000 5/5 60Natty p1 150,000 5/5 15Garp3 80,000 4/4 15Gar p4 50,000 4/44 30-42Mehal p5 50,000 4/44 23-35Solo p5 170,000 4/44 12-20Van p5 250,000 4/44 3-11Pere p5 57,000 8/8 25-37Pere p8 140,000 8/8 14-32Pere p6 300,000 8/8 3.5-51 = All ROSE 1 99-ECMs failed to attach to the collagen gel, even following treatment with cloning cylinders.2= One out of three ECMs was odginally attached to the collagen gel, but following treatment with cloningcylinders, two out of three ROSE 199-ECMs were attached to the collagen gel.3= Two out of seven ECMs were odginally attached to the collagen gel, but following treatment with cloningcylinders, five out of seven ROSE 199-ECMs were attached to the collagen gel.4= Each organoid was maintained under different conditions: 1) one in 199:105/15% FBS; 2) one in199:105/15% FBS supplemented with EGF/HC; 3) one in 199:105/PC-i; and 4) one in 199:105/155 FBSwith 3T3 fibroblasts added to the collagen gel component of the organoid. See also figure 28.5= Same as in 4, except that each condition was set up in duplicate organoids.79culture which have been used elsewhere to stimulate and simulate cellulardifferentiation or morphogenesis (Emerman et al. 77, Hadley et al. ‘85, Taub et al.‘90, Vigier et al. ‘89).Within 2 days of culture on the various substrata HOSE cells from 12 differentcases showed consistent changes in morphology (figure 29). On PL, PD, and FB,HOSE attached, spread, and formed flat cobblestone, epithelial monolayers. Theresponse of HOSE to PD-ECM was no different from HOSE on plastic. While cells onfibrin were epithelial, they were more dispersed than on plastic. HOSE attached,spread, but assumed a spindle-shaped morphology on unrimmed collagen gels. HOSEplated upon PD-coated collagen gel demonstrated a phenotype identical to that onunrimmed collagen gel alone.On Matrigel, HOSE cells formed aggregates that were joined to each other viabranching structures (figure 29). Granular areas of Matrigel near HOSE cellssuggested degradation of the Matrigel matrix by HOSE cells. This impression wassupported by SEM and paraffin sections. SEM showed that HOSE aggregates becameembedded and completely covered by Matrigel (figure 30A). HOSE cells cultured forone week or more on Matrigel invaded the Matrigel, and finally attached and spreadon the underlying plastic (figure 30B). Cross-sections of HOSE on all substratarevealed that HOSE remained on the apical surface of all matrices (figure 31A)except Matrigel where nests of cells were found within and completely enveloped bythe Matrigel (figure 31B,C). HOSE aggregates on Matrigel may represent adifferentiated phenotype or an attempt at morphogenesis. The epithelial nature ofHOSE was confirmed by the retention of keratin (figure 31 C).These results demonstrate specific morphological responses by HOSE cells todifferent culture conditions and may crudely mimic morphological changes in HOSEas they occur in vivo. The PD-ECM did not provide any advantage over plastic and so,after preliminary experiments, PD-ECM was eliminated from further experiments.Further, the results showed that HOSE cells in culture have the ability to remodelECM not only by the contraction of organoids, but also by invasion of Matrigel.(2.) Do Substrata Influence HOSE Cell Integrin Expression?To determine if the different HOSE morphologies observed on the various substratawere related to the expression of different integrins, the pattern of specificintegrins expressed at cell surfaces was examined.l25-labeled HOSE cells fromthree separate cases (two low passage cobblestone cultures and one late passageatypical epithelial culture) were immunoprecipitated with several anti-integrin80Figure 29. Morphological Response of HOSE Cells to SubstrataTwo days after plating, HOSE cells are epithelial on plastic (A) and fibrin clots(B) while they are spindle-shaped on collagen gels (C). On Matrigel, the cellsform aggregates that are joined to each other via branching structures (D).Phase microscopy. Xl 50.t81Figure 30. Invasion into Matrigel by HOSE CellsA scanning electron micrograph demonstrates that HOSE cells maintained onMatrigel penetrate the matrix and become covered with it (A), X300. HOSE cellslyse and invade Matrigel and eventually adhere and spread onto the underlyingplastic (area enclosed by three arrowheads in B). B, Phase microscopy, Xl 00.——:82Figure 31. Cross-sections of HOSE on Collagen and MatrigelCross-sections of HOSE cells maintained on collagen gels indicate that the HOSEcells are flattened and remain on the apical surface of the gel (arrows in A) whileHOSE cells invade Matrigel (B,C). In cross-section, HOSE cells in Matrigelappear as cell nests (B) or as cords of cells (C). The epithelial nature of HOSE ismaintained as indicated by the retention of keratin (A & B). A & B,Immunocytochemically stained for keratin, X375. C, Hematoxylin and eosin,X250.‘9-‘L-‘IVC)83antibodies. Figure 32 illustrates a typical profile of integrins expressed by HOSEfollowing 48 hours on plastic, collagen gels, fibrin clots, and Matrigel. The patternof integrins was the same: 1.) among early and late passage HOSE cells; 2.) betweencobblestone and atypical epithelial HOSE cells; 3.) within the same case, Van HOSE,tested at passage 2 and passage 4; and 4.) between HOSE cells and IOSEVan cells (datanot shown). VLA-6 and 34, comprising a laminin receptor, were absent in all cases.With the exception of collagen gels, receptors for vitronectin (VNR), collagen,fibronectin, and laminin (VLA-2,-3, -5) were present in all cases, although VLA-2was present in low amounts. Integrin expression was almost always greatest onplastic and least or absent on collagen gels.To see if the diminished amount of integrins expressed on collagen at 48 hours wasan indication of integrin receptor downregulation or merely artifactual, IOSEVanwere plated on PL and COLL gels for 1-48 hours and immunoprecipated for VNR onthree separate occasions. This integrin was chosen for time course study as it is oneof the integrins that is consistently expressed at high levels on PL and low levels onCOLL at 48 hours. If the downregulation was artefactual (i.e. due to the method of cellharvesting via collagenase treatment), then diminished levels of integrins would bepresent throughout 48 hours of culture on COLL. However, if the downregulation wasreal, then the cells, coming from parental stocks maintained on plastic andexpressing high levels of integrins on their cell surfaces, should express high levelsof integrins during the early intervals on COLL and then show a downregulation ofreceptors. Figure 33 shows that the downregulation of integrin expression on COLLdoes not appear to be an artefact. Levels of VNR remained relatively high on IOSEVancells throughout 48 hours of culture on PL. On COLL, though, the level of VNR washigh during the first 24 hours of culture a diminished expression clearly evident by48 hours. While VNR expression appeared to increase during the first 24 hr oncollagen gels, this upregulation of integrin expression was not found consistently.(3i Do Substrata Influence HOSE Cell Growth?Growth curves indicated that the substrata could be divided into three groups: 1)those that supported intense HOSE cell growth (PL, PD); 2) those which supportedlittle or no growth (COLL, PD+COLL, FB); 3) those where the cell numbereventually decreased (EHS) (figure 34A).84Figure 32. Survey of HOSE IntegrinsImmunoprecipitations of specific integrins from 1251-labelled HOSE cells. Cellswere labelled, extracted, and immunoprecipitated with the appropriateantibodies. lmmunoprecipitates were analyzed by SDS-PAGE under non-reducingconditions and the bands were visualized by autoradiography. Lanes 1 ,5,9, 1 3 areHOSE cells maintained on plastic; lanes 2,6,10,14 are HOSE cells maintained oncollagen gels; lanes 3,7,11, 1 5 are HOSE cells maintained on fibrin clots; lanes4,8,12,16 are HOSE cells maintained on Matrigel. (A), lanes 1-4, HOSE cellsimmunoprecipitated with anti-human monoclonal antibody; lanes 5-8, HOSEcells immunoprecipitated with anti-human VLA-2 monoclonal antibody; lanes 9-1 2, HOSE cells immunoprecipitated with anti-human VLA-3 monoclonalantibody; lanes 1 3-1 6, HOSE cells immunoprecipitated with anti-human VLA-5monoclonal antibody. (B), lanes 1-4, HOSE cells immunoprecipitated with antihuman vitronectin receptor monoclonal antibody; lanes 5-8, HOSE cellsimmunoprecipitated with anti-human l4 antibody; lanes 9-1 2, HOSE cellsimmunoprecipitated with anti-human VLA-6 monoclonal antibody. Molecularweight standards are myosin (200KD), B-galactosidase (11 6KD),phosphorylase B (97KD), and albumin (67KD).8i’AAKD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16200-116-‘97- — aB1 2 3 4 5 6 7 8 9 10 11 12200-• ay a6116— -‘4(3497- d •i85Figure 33. Integrin Time CourseImmunoprecipitations of vitronectin receptor from l1 25 labelled IOSEVancells. Cells were labelled, extracted, and immunprecipitated with rabbit antivitronectin receptor antibody. lmmunoprecipitates were analyzed by SDS-PAGEunder non-reducing conditions and the bands were visualized by autoradiography.Lanes 1 -4 are lOSEVan cells maintained on plastic and lanes 5-8 are lOSEVancells maintained on collagen gels. Cells were immunoprecipitated with antivitronectin receptor antiserum following 1 hour (lanes 1 & 5), 4 hours (lanes2 & 6), 1 2 hours (lanes 3 & 7), and 48 hours incubation on plastic and collagengels. The molecular weight standards are the same as those described in figure32.g;qKD200-116-12345678.ei97- 13386Figure 34. HOSE - Growth Curves and Chymotrypsin-like ActivityGrowth curves (A) and chymotrypsin-like activity detectedspectrophotometrically from conditioned media (B). HOSE cells were plated onplastic (open squares); placental matrix (closed squares); collagen gel (closeddiamonds); collagen gel coated with placental matrix (open diamonds); fibrin clot(open circles); and Matrigel (closed circles). Cell growth is inversely related toprotease production.RelAbs405nm/1O,000CellsCELLNO.X10,0000°..._cx0_I\ \0\C\(4)o.“3a0087(4.) Do HOSE Cells Produce Proteases?Since the morphological data suggested that HOSE cells invaded Matrigel, thecultures were examined to determine if these cells produced proteolytic enzymeswhich may degrade ECM and allow invasion. For most assays, HOSE cells were platedonto the various substrata in medium 199:105/15% FBS overnight and thenreplaced with reduced serum-containing medium (199:105/1% FBS,199:105/0.5% FBS, 199:105/PC-i) for the remainder of the experiment. Theconditioned medium was collected following two to three days incubation on thevarious substrata because by this time the different morphologies were clearlyestablished while the cell numbers between the various matrices were stillcomparable. Analyses of conditioned culture medium from all substrata demonstratedthat normal HOSE secrete proteolytic enzymes.A chymotrypsin-like peptidase with phenylalanine specificity (figure 34B andtable 6) and an elastase-like peptidase with alanine specificity (table 7) weredetected spectrophotometrically. Chymotrypsin-like activity and elastase-likeactivities were inversely related to cell growth (figure 34 and tables 6 & 7). Thatis, while the greatest amount of HOSE cell growth was seen on plastic and placentalECM, the least amount of chymotrypsin-like and elastase-like activities was seen onthese substrata. Conversely, while Matrigel, fibrin clots, and collagen gels supportedlittle or no cell growth, chymotrypsin-like and elastase-like activities weregreatest on these substrata. Trypsin-like activity was not detected in any of theconditioned media. Chymotrypsin-like and elastase-like activities were stilldemonstrable 9 days after subculture and while the amounts were slightly less thanprotease activities detected at D3, the relationship of protease activities to differentsubstrata remained the same (tables 8 & 9). Further, there was no difference in thepattern of chymotrypsin-like and elastase-like activities regardless of serumsupplementation used (tables 6-9). These results suggest that HOSE cellsconstitutively produce chymotrypsin-like and elastase-like proteases although theamount of protease production is influenced by substratum. Only background amountsof protease activities were detected from conditioned media collected from substratacontrols.Major bands of gelatinolytic activity were consistently found in the conditionedmedium of HOSE maintained on the various substrata. For the first 2 cases tested, thecells were plated in 199:105/15%FBS for 24 hrs which was then replaced with199:105/DM and collected after an additional 24 hr. Substrata controls consisted ofsubstrata-coated wells alone incubated for 24 hours in 199:105/DM having beenTable 6.Chymotrypsin-Uke ActivityinHOSECulturesMatgChangeinAbsorbance@405nm/hr periOCells(X104co*=valuescorrectedfor backgroundinTables6-9Cases:#1#2#3#4#4#5#5MediumSupplement:1%FBS1%F8Sl%FBS0.5%FBSPC-I0.5%FBSPC-iPlastic3.01.74.30.10.60.40.0Collagen8.08.76.7Fibrin9.65.616.830.460.47.023.713.017.425.312.614.549.577.571.4126.6933.953.0Table7.Elastase-LikeActivityinHOSECulturesChangeinAbsorbance@405nrnlhrperi04Cells(x10)Cases:#1#2#3#4#5MediumSupplement:1%FBS1%FBS1%FBS0.5%FBS0.5%FBSPlastic0.852.00.00.40.6Collagen3.128.072.04.01.7Fibrin22.437.4167.44.0N.D.Matngel36.6167.3172.1146.869.5coN.D.=not doneTable8.Chymotrypsin-LikeActivityinCondionedMediumFromDay9HOSECulturesChangeinAbsorbance@4O5nmfhrper i04Cells(xlOjCases:#4#4-#5#5MediumSupp’ement:0.5%FBSPC-i0.5%FBSPC-iPlastic1.11.20.01.1Collagen4.05.21.82.2Fibrin4.99.31.51.9Matngel10.524.24.76.5Table9.Elastase-LikeActivityinConditionedMediumFromDay9HOSECulturesChangeinAbsorbance@405nrrVbrper10’Cells(x104)Cases:#4#4#5#5MediumSupplement:0.5%FBSPC-i0.5%FBSPC-iPlastic7.812.52.619.0Collagen15.212.52.920.1Fibrin13.925.53.425.8Matrigel31.3155.756.049.1I-.92previously incubated for 24 hours in 199:105/15% FBS for 24 hours. Agelatinolytic band at 3OKD was expressed exclusively by HOSE cells and was absentin substrata controls (figure 35A). A broad band of lysis was also found at roughly42KD in both conditioned media from cells and conditioned media from substratacontrols, although there appeared to be a greater amount of gelatinase associated withthe cell-derived conditioned medium (figure 35A). A clear doublet of lysis was alsopresent at approximately 97KD, but was similar in conditioned medium from HOSEcells as well as from substrata controls (figure 35A) suggesting the carryover ofserum proteases. Essentially all gelatinolytic activity was abolished followingincubation of gels in 10 mM EDTA (figure 35B), indicating that these gelatinases aremetal loproteases.To distinguish further between gelatinases secreted by HOSE cells and those presentin and carried over from serum, HOSE cells from one case, Cello, were washedextensively in 199:105/DM and plated on plastic, fibrin clots, collagen gels,Matrigel in 199:105/DM overnight. The following day the medium was replaced withfresh 199:105/DM, and after a further 24 hour of culture the conditioned mediumwas collected, concentrated, and subjected to gelatin zymography. HOSE cells platedand maintained in 199:105/DM attached and spread on plastic, collagen gels, andfibrin clots and were epithelial, spindle-shaped, and epithelial respectively inresponse to these substrata. However, cells plated in 199:105/DM failed to adhere toMatrigel and remained as floating rounded cells, so that the subsequent mediumchange at 24 hours resulted in removal of HOSE cells from Matrigel wells. It appearsthat some factor(s) present in serum is necessary for HOSE cell adhesion toMatrigel. Again, controls consisted of substrata-coated wells plated in DM and treatedas those wells that had received cells. Similar to the results described above, gelatinzymography revealed that HOSE cells secrete a 3OKD gelatinase as well as a 42 KDgelatinase that may be the same as that found in serum (figure 36A). However, ascells plated in 199:105/DM did not adhere to Matrigel, no gelatinase activity couldbe detected in Matrigel-coated wells. Substrata controls did not show anygelatinolytic activity (figure 36A), abrogating any endogenous gelatinase activitiesfrom within any substrata that might have leeched out into the conditioned medium.As with the other gelatin zymograms, the gelatinases produced by HOSE cells platedin 199:105/DM are metalloproteases as activity was abolished following incubationin 10 mM EDTA (figure 36 B).Neither collagenase nor plasminogen activator activities were detected. During 2days of culture HOSE cells secreted no detectable collagenase or proteases capable of93Figure 35. HOSE - Gelatin Zymography 1Conditioned media collected from HOSE cells plated in 199:105/15%FBS for 24hours and then maintained for 24 hours in 199:105/DM were analyzed bygelatin zymography. (A), conditioned media collected from HOSE cells maintainedon: plastic (lane 1); collagen gel (lane 2); fibrin clots (lane 3); and Matrigel(lane 4). Conditioned media collected from substrata controls maintained withoutHOSE cells: plastic (lane 5); collagen gel (lane 6); fibrin clots (lane 7); andMatrigel (lane 8). (B), same as in A except that the gel was incubated in 1 0mMEDTA. HOSE cells maintained on the various substrata express a 97, 42 and 30KD gelatinase while substrata controls express a 97 and 42 KD gelatinase. Allgelatinolytic activity was abolished following incubation with EDTA, a divalentcation chelator, indicating that the gelatinases are metalloproteases.CCoococoQ0O)(QQ-coc0coNCoLC)coCOLOCoCOC4J94Figure 36. HOSE - Gelatin Zymography 2Conditioned media collected from HOSE cells plated in 1 99:1 05/DM for 24 hoursand then maintained for an additional 24 hours in fresh 199:105/DM wereanalyzed by gelatin zymography. (A), conditioned media collected from HOSE cellsmaintained on: plastic (lane 1); collagen gel (lane 2); fibrin clots (lane 3); andMatrigel (lane 4). Conditioned media collected from substrata controlsmaintained without HOSE cells: plastic (lane 5); fibrin clots (lane 6); collagengel (lane 7); and Matrigel (lane 8). (B), lanes 1-4 same as in A, except that thegel was incubated in 10mM EDTA. When maintained in defined medium, HOSEcells, except those on Matrigel, express a 42 KD gelatinase not present insubstrata controls. As in figure 35, gelatinase activity was abolished followingtreatment with EDTA.V99L695degrading collagen type I regardless of the substratum on which the cells weremaintained. Figure 37 shows that conditioned medium from HOSE cells maintained onplastic, placental-ECM, fibrin clots, collagen gels, collagen gels coated withplacental-ECM, and Matrigel failed to degrade the al and cz2 chains of native collagentype I.HOSE cells secreted no detectable levels of plasminogen activator (PA) into theirconditioned medium during 2 days of incubation on the various substrata whilestandards of human urokinase produced lysis in the agarose. The lack of PA detectionmay be due to the presence of plasminogen activator inhibitor (PAl-i) detected inboth early and late passage HOSE cultures (figure 38). HOSE cells stained intenselyfor PAl-i, while serum controls did not stain. Further, LP-9 cells served aspositive controls and also stained strongly for PAl-i compared to their respectiveserum controls. As PAl-i binds PA, it possibly made PA undetectble in the assaysused in this study.96Figure 37. HOSE -3H-Collagen Type I DigestionConditioned media from HOSE cells maintained on various substrata wereanalyzed by SDS-PAGE and the bands visualized by autoradiography. Conditionedmedia collected from HOSE cells maintained on: plastic (lane 1); placental matrix(lane 3); collagen gel (lane 5); collagen gel coated with placental matrix (lane7); fibrin clots (9); and Matrigel (lane 11). Conditioned media collected fromsubstrata controls maintained without HOSE cells: plastic (lane 2); placentalmatrix (lane 4); collagen gel (lane 6); collagen gel coated with placental matrix(lane 8); fibrin clots (lane 1 0); Matrigel (lane 1 2). Lane 1 3, collagen type Istandard. All HOSE cells failed to degrade collagen type I.KD1 23 4 5 6 7 8 9101112 13200—.a1(I)t.,i.i... 2(I)116—97Figure 38. Expression of Plasminogen Activator Inhibitor by HOSE CellsHOSE cells (A,B), like LP-9 cells (C,D), stain uniformly for plasminogenactivator inhibitor (PAl-i). HOSE cells failed to stain when the primaryantibody was replaced with normal serum (E,F). A,C,E, Immunofluorescencemicroscopy. B,D,F, Phase microscopy of same fields shown in A,C,E respectively.X500.974TABLE10. Summa,yofImmunofluorescenceStudiesFqureAncienTestMatenalControlfor1°AbControlMateal2KeitinHOSEcells:1°+2°Ab,+vePBS+2°Ab,-yeAmniocCells:10+2Ab,+1-yeAmniocCells: PBS+2°Ab,-yeHUDF:1°+2°Ab-yeHUDF:PBS+2°Ab,-ye5LamininROSE239Cells: 1°+2°Ab,+veNRbS+2°Ab,-ye5CollagenTypeIROSE239Cells:1°+2°Ab,+veNShS+2°Ab, -yeHUDF:1°+2°Ab,+veNShS+2°Ab,-ye10KeratinHOSECells:1°+2°Ab,+1-yePBS+2°Ab,-yeC4-lCells: 1°+2°Ab,+veC4-lCells: PBS+2°Ab,-yeHUDF:1°+2°Ab,-yeHUDF:PBS+2°Ab,-ye12LamininHOSECells:1°+2°Ab,+veNRbS+2°Ab,-yeROSE239Cells:1°+2°Ab,+veROSE239Cells: NRbS+2°Ab,-yeHUDF:1°+20Ab-yeHUDF:NRbS+2°Ab,-ye13CollagenTypeIVHOSECells:1°+2°Ab,+veNRbS+2°Ab,-ye14CollagenTypeIHOSECells:1°+2°Ab,+veNShS+2°Ab,-yeROSE239Cells:1°+2°Ab,+veROSE239Cells: NShS+2°Ab,-yeCollagenTypeIllHOSECels:1°+2°Ab,+1-NGS+2°Ab,-yeROSE239Cells:1°+2°Ab, +veROSE239Cells:NGS+2°Ab,-yeKeratinHOSECells:1°+2°Ab,+vePBS+2°AbC4-lCells:1°+2°Ab,+veC4-lCells: PBS+2°Ab,-yeHUDF:1°+2°Ab-yeHUDF:PBS+2°Ab,-yeTable10.Continued19LaaininROSE199-ECM:1°+2°Ab, +veNRbS+2°Ab,-yeROSE239Cells:t°+2°Ab,+veROSE239Cells:NRbS+2°Ab,-yeHUDF:1O+2O-yeHUDF:NRbS+2°Ab,-ye20ActinROSE199-ECM:NBD,-yePH, -yeRat Intestine:NBD,+veNBD+PH, -yeRat Intestine:PH,-yeRatIntestine:NBD+PH,-yeLamininROSE199-ECM:1°+2°Ab, +veNRbS+2°Ab,-yeRat Intestine:1°+2°Ab,+veRatIntestine:NRbS+2°Ab,-ye21CollagenTypeIROSE199-ECM:1°+2°Ab,i-yeNShS+2°Ab,-yeHUDF:1°+2°Ab,i-yeHUDF:NShS+2°Ab,-ye38PAl-iHOSECells:1°+2°Ab, +veNRbS+2°Ab,-yeLP-9Cells:1°+2°Ab,+veLP-9Cells:NRbS+2°Ab,-ye+ve=positive-ye=negative+1-ye=postiveandnegative10=piin’.y2°=secondary100DISCUSSIONA. Tissue Culture Techniques(1.) Scrape-Derived HOSE CulturesIn spite of its clinical importance, studies on the role of HOSE in ovariancarcinogensis have been severely limited by the lack of experimental systems. Toadapt culture methods for HOSE from methods used to culture ovarian surfaceepithelium from rat and rabbit, two problems had to be overcome. First, the meansto isolate the epithelium from the rest of the ovary had to be modified because in thehuman biopsy specimens it is present in very small amounts compared with, forexample, the papillary ovarian surface epithelium of the rabbit ovary. Secondly, theculture medium used to propagate animal surface epithelium proved inadequate forthe culture of its human counterpart (Auersperg et al. ‘84). Recent methods allowedfor the isolation, propagation, and characterization of HOSE cells in culture, but thetechniques used were laborious and the cultures were frequently contaminated withstromal and follicular cells. Furthermore, the potential of the cells to proliferatewhile retaining their normal differentiated form was limited (Auersperg et al. ‘84,Siemens & Auersperg ‘88). One of the tissue culture techniques developed in thecourse of this study was an improved method for the preparation and culture of HOSEcells from normal human ovaries. Instead of preparing explants, advantage was takenof the tenuous attachment of HOSE to its underlying connective tissue and sheets ofHOSE cells were scraped off the biopsies. In addition to the improved yield, thescrape method also greatly increased the purity of the cultures. In contrast toprocedures that use explants (Auersperg et al. ‘84) or enzymatic dissociation(Hamilton et al. ‘80) of the original tissue, the ovarian tissue below the surfaceepithelium remained undisturbed by the scrape method. As a result, contaminationby stromal and follicular cells was rare. Occasional groups of contaminatingfibroblasts could be removed with a rubber scraper. Histologic examination ofbiopsy specimens confirmed that scraping denuded 90 to 100% of the ovariansurface of HOSE with little disruption to underlying structures. Thus, normalsurface-derived HOSE was selected over cyst and invagination-lined HOSE which isfrequently atypical. In additon to the improved yield and greater purity of thecultures, the scrape method has the obvious advantages of being considerably fasterand simpler than either explant culture or culture of enzymatically dissociated cells.101The ease with which HOSE is scraped off biopsy specimens suggests that previousreports on the desquamation of HOSE with increasing age (Mckay et al. ‘61, Sauramo‘52) are likely an artefact. Also, the tenuous attachment of HOSE to its underlyingstroma has clinical and physiological implications as HOSE may be denuded duringany gynecological surgery. Since adhesion formation is common following ovariansurgery, standard ovarian surgical techniques may promote adhesion formation byremoving HOSE which is important for covering, protecting, and re-epithelializingthe ovary. Lastly, there are contradictory reports on the effects of denuding theovarian surface and subsequent ovulations. For example, Gillet (‘91) showed thatremoval of rabbit ovarian surface epithelial cells reduced the number of conceptionscompared to the undenuded contralateral ovary while Rawson & Espey (‘77)demonstrated that follicles proceeded to ovulate even after the rabbit ovarian surfaceepithelium had been denuded. Therefore, although there is no direct evidence thatHOSE contributes to ovulation, the possiblity that HOSE contributes to normalovarian functions such as ovulation should not be discounted prematurely.By previous methods, the best culture medium to maintain HOSE in its tissue-specific epithelial form was medium 199:202/15% FBS, but the cells grew slowlyand could only be subcultured once or twice, approximately 6 population doublings.Addition of EGF/HC increased the growth rate and growth potential considerably, butcaused modulation of the HOSE cells to an atypical, fibroblastic phenotype (Siemens& Auersperg ‘88). This modulation was initially reversible, but eventually becamepermanent. It interfered with cell identification and with the interpretation ofexperimental results. In the culture medium used at present, the substitution of asynthetic component, MCDB 105 (McKeehan et a!. ‘78) for MCDB 202 haseliminated the need for EGF/HC to support rapid growth for up to 12 populationdoublings, but results in the modulation to a fibroblast-like phenotype withincreasing passage in culture (Kruk et al. ‘90). MCDB 202 and MCDB 105 are bothmodifications of medium F12. Their ingredients are the same, but a number ofcomponents differ in relative concentrations. Several of these differences may havecontributed to the improved growth of HOSE in MCDB 105-supplemented medium.Compared with MCDB 202, MCDB 105 has cysteine and calcium concentrations thatare closer to the optimum for the growth of other epithelial cells (McKeehan et al.‘84). Furthermore, the higher levels of glutamine and proline in MCDB 105 mayenhance extracellular matrix synthesis and, consequently, contribute to theepithelial morphology of HOSE cells. Finally, the buffering capacity of MCDB 105/ 102minimizes pH fluctuations in response to shifts between air and the incubationatmosphere and, thus, provides an improved microenvironment.This simplified and more successful methodology for the culture of HOSE cellsopened the way for the in depth investigations of HOSE described herein.(2). Percoll DecontaminationThe second tissue culture technique developed during the course of this study,stemmed from Percoll centrifugation experiments performed to separate HOSE cellsfrom stromal fibroblasts and resulted in a method to remove mold infections fromcultures. Microbial contaminations in tissue culture are usually of environmentalorigin and include bacteria, yeast, and molds. Most commonly, contaminated culturesare discarded, though they can, at least temporarily, be controlled with the use ofantibiotics and fungicides. However, the potential exists for the development of drugresistant infections and cytotoxic effects on the cultured cells (Adams et al. ‘86,Dickman et al. ‘90, Freshney ‘83, Hanas & Simpson ‘86, Laska et al. ‘90, Ramsammyet al. ‘88, Van der Auwera & Meunier ‘89). Further, antibiotics such asstreptomycin and gentamicin and fungicidal agents such as fungizone, can adverselyaffect cell metabolism, growth, morphology, differentiation, and protein synthesis(Adams et al. ‘86, Dickman et al. ‘90, Freshney ‘83, Laska et al. ‘90, Ramsammy etal. ‘88, Van der Auwera & Meunier ‘89). Percoll centrifugation provides analternate means of removing fungal contaminants from cultures. Similar to thereport by Behrens and Paronetto (‘84) yeast infections could not be eliminated byPercoll gradient centrifugation in this study and Percoll centrifugation did notenhance decontamination of yeast-infected cultures when used in conjunction withfungizone. The difficulty in removing yeast infections may, in part, result fromcomplex yeast-host cell interactions such as the internalization of yeast by hostcells, which renders yeast inaccessible to decontamination (Merkel et al. ‘88).In contrast, the Percoll method proved highly successful for the elimination ofmolds. Percoll is a gradient material composed of non-toxic polyvinylpyrrolidonecoated colloidial silica particles (Pertoft et al. ‘77, Pertoft & Laurent ‘77) and iswidely used for the separation of different types of cells, subcellular particles,viruses, and bacteria (Kreamer et al. ‘86, Pertoft & Laurent ‘77, Raber &D’Ambrosio ‘86, Roskelley & Auersperg ‘90). The non-toxic nature of Percollallowed cell bands, devoid of mold, to be recovered and plated directly, withoutwashing, into dishes containing tissue culture medium. There did not appear to be anyacute toxic effects on cells following Percoll treatment since viability was always103greater than 94%. By the present method, approximately 50% of the centrifugedcells were retrieved, but it is possible that recovery could be increased withimproved cell band retrieval. Following centrifugation in Percoll, there were nodifferences in rates of growth, morphology, and capacity for differentiation, asshown by the production of extracellular matrix material. The maintenance of thesecharacteristics among cultures is not unexpected as it has been shown elsewhere thatcentrifugation in Percoll does not affect such cellular parameters (Knecht et al. ‘89,Kreamer et al. ‘86. Pertoft et al. ‘77, Roskelley & Auersperg ‘90).This method is now used routinely in our laboratory and, to date, several primarycultures from different sources have been successfully decontaminated of mold. Thesimilar banding characteristics of a variety of cells suggest that Percolldecontamination can accommodate different cell types and may, therefore, begenerally applicable.B. HOSE Cells Cultured from Scraped Biopsies are a Relevant ModelSystemHaving developed a simpler and more successful method to culture HOSE cells, itwas important to establish the relevance of this model system as a means to study thebiology of normal HOSE and its role in ovarian carcinogenesis. The greater thenumber of characteristics retained by HOSE cells in culture compared to their invivo counterparts the greater would be the relevance. Initial characteristics of HOSEcells in culture included the presence of keratin, numerous microvilli, 17-f3-hydroxysteroid dehydrogenase activity, and pleomorphism (Auersperg et al. ‘84)similar to HOSE in vivo (Blaustein & Lee ‘79, Czernobilsky ‘85, Czernobilsky et al.‘84, Gillet et al. ‘91, Motta et al. ‘80, Van Blerkom & Motta ‘79). In this study thelist of characteristics of cultured HOSE cells was expanded. Cultured HOSE cells wereshown to express vimentin, keratin subtypes #7, 8, 18, 19, glycogen, lipid,variable amounts of mucin like their in vivo counterparts (Blaustein & Lee ‘79,Czernobilsky ‘85, Czernobilsky et al. ‘84, Viale et al. ‘88). The pleomorphism wasanalyzed and specific cell shape could be related to the presence of matrix. CulturedHOSE cells express the four keratin subtypes characteristic of simple epithelia(MoN et al. ‘82, Steinert & Parry ‘85, Steinert et al. ‘85) like HOSE in vivo(Czernobilsky ‘85, Czernobilsky et al. ‘84), but there was a gradual loss of keratinexpression with increasing passage which is in agreement with other reports(Auersperg et al. ‘84) while the remaining markers were unaffected with104subsequent cell passaging. The keratins #8 and #19 appear to be prefentially lostwith increasing passage in HOSE cultures. Many epithelial cell types co-expressvimentin and keratin in culture (Ben-Ze’ev ‘84a,b,’86, Cells et at. ‘85, Rheinwald& O’Connell ‘85), however, co-expression of vimentin and keratin in normalepithelium in vivo has so far been reported only in mesothelium, HOSE, oviductalepithelium, and parietal endoderm (Czernobilsky ‘85, Lane et at. ‘83, Rheinwald &O’Connell ‘85, Viale et al. ‘88). Czernobilsky (‘85) has suggested that the coexpression of vimentin and keratin in HOSE may reflect the mesodermal origin ofthis epithelium. The mesothelium, which is embryologically related to HOSE in thatthey both arise from the coelomic epithelium, is another mesodermally-derivedepithelium which co-expresses vimentin and keratin in culture (Connell &Rheinwald ‘83). Further, the retention of lipid and glycogen by HOSE cells in culturemimics their in vivo counterparts (Blaustein ‘84, Blaustein & Lee ‘79). Lipiddroplets were found distributed in varying amounts among cultured HOSE cells. Thepresence of lipid in cultured HOSE cells may indicate favourable nutritional cultureconditions such that the cells store excess glucose and fatty acids in the insoluble,non-toxic forms of glycogen and triglycerides. Glycogen was clearly localized asabundant, amylase-sensitive, cytoplasmic granules. This is not surprising sincecultured cells are exposed to an excess of glucose, a nutrient source in the culturemedium, which is often stored by cells as glycogen (Freshney ‘83). The healthystatus of these cells is further reflected by the presence of mitotic figures and largenuclei with prominent nucleoli indicative of ribosomal protein synthesis.HOSE cells in culture demonstrate tremendous pleomorphism under differentcircumstances. HOSE outgrowths have been catagorized into one of three forms: (1.)compact, monolayered epithelial cells; (2.) flat epithelial cells; and (3.) atypicalepithelial cells (Auersperg et al. ‘84). The compact and flat epithelial HOSE cellsdescribed in culture appear to correspond to the type A and type B cells respectivelyshown by Gillet et al. (‘91) in vivo so that it would appear that the heterogeneity ofHOSE cells in vivo (Gillet et at. ‘91) is maintained in culture. Interestingly, ratovarian surface epithelial cells (Adams & Auersperg ‘81, Hamilton et al. ‘80) retaina homogeneous pavementlike appearance for long periods in culture and then eitherbecome stationary or progress to continuous lines (Adams & Auersperg ‘85) whileHOSE cells acquire a heterogeneous appearance in culture and have, to date, neverprogressed to continuous cells lines. The present studies showed that theheterogeneous morphology was independent of the expression of two markers ofepithelial differentiation, mucin and keratin, as these markers could be found in105HOSE cells expressing phenotypes ranging from epithelial to fibroblast-like. HOSEassumed various phenotypes ranging from flat epithelial to fibroblast-like to groupsof rounded cells depending upon the kind of substratum they were plated upon. Therelationship of these changes to events that occur in vivo remains to be investigated.Except for the apparent loss of tight junctions the methods presented here permitthe culture of normal HOSE cells which exhibit most of the characteristics of their invivo counterparts, suggesting that cultured HOSE cells derived from scraped biopsymaterial are relevant for the study of normal HOSE biology and the role of HOSE inovarian carcinogenesis.Yet, it is important to note that even though the list of HOSE cell characteristicshas expanded, there still does not exist a single, specific, unique marker for HOSEcells. For example, mammary epithelial cells in culture can be identified by theirability to produce milk components (Emerman et at. ‘77, Li et al. ‘87, Seely &Aggler, ‘91, Wicha et at. ‘82). Cultured HOSE cells share some features with othercell types, and not surprisingly most closely mimic mesothelial cells as HOSE itselfis a modified mesothelium. Presently, there exists a battery of tests that candistinguish HOSE cells from other cell types which might arise from the ovarianbiopsy and which might contaminate HOSE cultures. For example, if the cells arekeratin positive, they are not fibroblasts or endothelial cells. Similarly, thepresence of 1 7-3-hydroxysteroid dehydrogenase activity excludes the possibilty thatthe cells are mesothelial. Lastly, HOSE cells can be distinguished from granulosacells because human granulosa cells contain small amounts of keratin subtype #7(Czernobilsky ‘85, Czernobilsky et al. ‘84) in contrast to HOSE cells and granulosacells contain a number of steroidogenic enzymes (Goldring et at. ‘86) that are absentin HOSE cells (Blaustein & Lee ‘79, Hoyer ‘80, Rembiszewska & Brynczak ‘85).C. HOSE:Matrix Interrelationships(1 i Ovarian Surface Epithelial-Derived Extracellular MatrixFurther characterization of cultured HOSE cells in this study showed that HOSEcells secrete extracellular matrix made of both basement membrane and stromalmatrix components. Although this study only examined matrix synthesis by HOSE inculture, there is no reason to believe that the cells might not have the ability toexpress a similar phenotype in vivo. Cultured rat ovarian surface epithelial cellsalso secrete basement membrane and stromal matrix components (Auersperg et at.‘91b). ECM isolated from ROSE 199 cultures in the present study also contains both106basement mambrane and stromal components. It is particularly interesting thatROSE 199 cells in crowded cultures form ridges and papillae, as well as multiplecell monolayers separated by layers of collagenous matrix (Adams & Auersperg ‘85).Thus, increasing numbers of proliferating, monolayered rat ovarian surfaceepithelial cells per unit area of plastic substratum are accommodated through acontinuous increase in available ECM so that ECM derived from these cells mayprovide a means to circumvent the restraints of density-dependent growth inculture. In contrast to some other species (Nicosia & Johnson ‘84), papillaryoutgrowths of normal ovarian surface epithelium are scarce in the rat and human(Hamilton et al. ‘80, Siemens & Auersperg ‘88). Only a modest increase in thesestructures has been reported to occur at the site of follicular rupture and inpostmenopausal women (Makabe et al. ‘80, Motta & Van Blerkom ‘80, Nicosia ‘83,‘87). However, in a high proportion of serous ovarian tumours that are benign or ofborderline malignancy, the HOSE is greatly amplified and forms papillae withstromal cores (Young et al. ‘89). Thus, in vivo, increased numbers of HOSE cellsmay be accommodated on the limiting ovarian surface through the formation ofpapillae. It is tempting to speculate that the autonomous deposition of an ECM by ratand human ovarian surface epithelial cells might, at least in part, provide thestromal cores for papillae that occur physiologically and pathologically. The capacityof ovarian cells to form collagenous matrix autonomously would render the cells lessdependent on stroma contributed by normal connective tissue cells and might,therefore, be an important factor for ovarian neoplastic progression. These resultssuggest that the deposition of basement membrane and stromal matrix componentsmay be a general feature of all mammalian ovarian surface epithelial cells.The capability of ovarian surface epithelial cells to secrete stromal matrix is anunusal feature among epithelia and may be related to the location of this epithelium.Perhaps the ovarian surface epithelium, like all other epithelia, contributes to itsunderlying basement membrane, but may also contribute directly to the formation ofeither the specialized layer of dense collagenous material commonly referred to asthe tunica albuginea or to elements of the ovarian stroma. Alternatively, the capacityof the ovarian surface epithelium to secrete stromal matrix may be related to itsorigin. The expression of fibroblastic characteristics by HOSE (morphologicalmodulation between epithelial and fibroblastic phenotypes, depositon of stromalmatrix) may be related to the close developmental relationship of ovarian stromaand surface epithelium, both of which are derived from the gonadal ridge and may bea new example of epithelial-mesenchymal transformation. The retention of a degree107of plasticity that permits cells to express stromal as well as epithelialcharacteristics would be in keeping with other properties of HOSE cells, such astheir lability of keratin expression and their propensity to assume fibroblast-likeshapes and growth patterns in culture (Siemens & Auersperg ‘88). Very similarphenotypic modulations occur in cultured mesothelial cells (Connell & Rheinwald‘83). Various degrees of epithelial-mesenchymal interconversions have also beenreported in a number of other epithelia of mesenchymal origin, including thedeveloping kidney (Ekblom ‘84), and interestingly, regressing Mullerian duct(Trelstad et al. ‘82). The developing cornea and Iongterm cultures of rat intestinalepithelium represent other examples of epithelial cells that produce a stroma rich inconnective-tissue forms of collagen (Hay ‘82, Sambuy & De Angelis ‘86).A highly reproducible and abundant ECM was derived from an immortalized, nontumorigenic rat ovarian surface epithelial cell line. The ECM was prepared andcompared following three different preparative techniques, two chemical treatmentsand one physical treatment of ROSE 199 cultures. The ECM was complex in terms ofboth its component elements and morphology. The three forms of ECM appear similarwith the exception that FT-derived ECMs showed the presence of residual nuclei.While none of the preparations demonstrated cytoplasmic debris following stainingfor actin, which is a standard method to ascertain cellular debris in ECMpreparations (Aggler ‘88), residual nuclear material was present in allpreparations as determined by DNA staining. Preliminary experiments suggest thatimproved washing of ROSE 199-ECMs and treatment with DNAse may eliminate thesecontaminants. ROSE 199-ECM consists of complex bundles of fibers, amorphousmaterial, and striated collagen fibers. It is a unique ECM in that it contains bothbasment membrane and stromal elements. Unusual ECMs have been reportedelsewhere. For example, Brauer & Keller (‘89) reported on the isolation of basmentmembrane material from a murine teratocarcinoma cell line that lacked collagentype IV. Interestingly, collagen type IV is also lacking in ROSE 199-ECM. ROSE 199-ECM demonstrated biological activity by supporting cellular adhesion, spreading, andgrowth. ROSE 199-ECM, then, is a novel ovarian-derived biological substratum thatmay prove useful for the maintenance of cell growth or differentiation similar tosuch commonly used substrata as fibrin clots (Kadish et al. ‘79), collagen gels(Emerman et al. ‘77, Emerman & Pitelka ‘77, Lee et al. ‘84, Li et al. ‘87, Montesano‘86, Wicha et al. ‘79), and Matrigel (Davis et al. ‘90, Hadley et al. ‘85, Kleinman etal. ‘86, Madison et al. ‘85). Other potential uses of ROSE 199-ECM include a reliable108source of matrix components and the abundance of collagen in this matrix which mayprove useful for studies in collagen fibrillogenesis and matrix assembly.(2.) Physical ECM RemodellingMany cell types contract floating collagen gels (Emerman et al. ‘77, Emerman &Pitelka ‘77), but only a few cell types have been shown to remodel unrimmedcollagen gels (Bell et al. ‘79, Schafer et al. ‘89). It is well established thatfibroblasts contract unrimmed collagen gels (Bell et al. ‘79) and this process hasserved as a model for wound repair in culture (Buttle & Ehrlich ‘83). Recently,Schafer et al. (‘89) reported on the ability of keratinocytes to contract unrimmedcollagen gels and suggested that keratinocytes also play a role in wound repair. On thebasis of the results of this study, HOSE cells can now be added to the list of nonfibroblastic cells capable of physically remodelling matrix as evidenced by thecontraction of organoids. The physical ability of HOSE cells to contract ECM parallelsthe fibroblast model of wound repair and may be important for repair of theovulatory defect. Interestingly, Schafer et al. (‘89) noted that the degree of gelcontraction increased with inceasing age of donor keratinocytes and the authorssuggested that the changes in dermal collagen and the wrinkling of skin with age maybe related to keratinocyte function. It is tempting to speculate something similarwith HOSE cells. Perhaps the ovarian shrinkage observed with age is facilitated, inpart, by the matrix remodelling capabilities of HOSE cells.The morphological events associated with organoid contraction by HOSE cellsappear to be the same to those described for collagen gel contraction by fibroblastsand suggests, perhaps, that contraction occurs by the same process in both systems.In the fibroblast model, gel contraction is analogous to squeezing water out of asponge (Bell et al. ‘79). Gel contraction does not involve collagen degradation, butrather consolidation of collagen fibers (Bell et al. ‘79, Guidry & Grinnell ‘85).Morphologically, during contraction, collagen fibrils are bound individually and inclusters at the cell surface and are surrounded by cellular microvilli and blebs(Grinnell & Lamke ‘84). Subsequently, contraction of the collagen gels occurs in twosteps. There is movement and re-arrangement of the collagen fibers by an actindependent mechanical process (Buttle & Ehrlich ‘83, Guidry & Grinnell ‘85, ‘86).This is followed by changes in intermolecular bonds between collagen fibrils whichlead to fibril stabilization by non-covalent collagen-collagen interactions (Buttle &Ehrlich ‘83, Guidry & Grinnell ‘85, ‘86). Grossly, contraction is seen as aconsolidation of collagen fibers along the periphery of the contracting gel. A dense109cellular ring appears along the edge of the contracting gel and the cells along the edgealign themselves in a circumferential pattern on the perimeter of the gel (Buttle &Ehrilch ‘83).Contraction of collagen gels is thought to result from the tractional forces exertedby cells on collagen fibers (Nogawa & Nakanishi ‘87). The cells do not simplyshorten in length like contracting muscle cells. Instead of pulling on the substratumby shortening themselves, the cells exert a shearing force tangential to theirsurface. This force is traction and should be distinguished from contraction whichwould involve shortening of the cell itself even though cellular traction is caused bycytoplasmic contractions. Because it is traction rather than simple contractionwhich distorts rubber membranes, collagen gels, and other flexible substrata, thecells actually elongate in the direction of the contractile force they exert rather thanshorten (Harris et al. ‘80, Stopak & Harris ‘82).All organoids in which the ROSE 1 99-ECMs were not attached to the collagen gelfailed to contract. Similarly, HOSE cells plated onto unrimmed collagen gels failed tocontract suggesting that HOSE cells alone on collagen gel may not be able to generatesufficient tractional force to contract matrix. If non-attached ROSE 199-ECMs werere-attached to the collagen gels, HOSE cells often then contracted the organoids. Thisindicates that one definite requirement for organoid contraction is that the ROSE199-ECM had to be in direct contact with the collagen gel component of the organoid.The role of ROSE 199-ECM in organoid contraction is unknown. It is possible thatinteraction between the collagen gel and components of the ROSE 199-ECM (i.e.fibronectin) may result in matrix-driven translocation. Matrix-driventranslocation is a biophysical process in which there is rapid movement of cells orinert particles from one region of a non-uniform collagen/fibronectin matrix toanother (Newman et al. ‘85, ‘87). Matrix-driven translocation involves theinteraction and movement of matrix components among each other so that particlesare carried along with this matrix movement. In organoids a matrix-driventranslocation-like process, generated by the interaction of collagen gel and ROSE199-derived fibronectin or laminin may serve to distribute HOSE cells quicklyaround the organoid so that HOSE cells could then contract the organoid.Alternatively, matrix-driven translocation-like processes may add to the existingtractional forces exerted by HOSE cells to generate sufficient total force to contractthe organoid. If matrix-driven translocation-like forces are present they could notbe solely responsible for organoid contraciton as controls organoids (i.e. withoutcells) failed to contract. Further, interaction between collagen gel and ROSE 199-110ECM components may re-align or cross-link the existing matrices such that all thecomponents are interlocked. Such matrix interactions might serve to enhanceorganoid contraction as the tractional forces exerted by HOSE cells could betransmitted throughout the re-organized interlocking matrix network.For cellular processes such as migration, wound healing, and morphogenesis tooccur, signals from the ECM to the cell, often via integrins, must be translated intoevents which re-organize the existing ECM. The induction of collagenase productionby fibroblasts following binding to fibronectin via the fibronectin receptor (Werb etal. ‘89) illustrates an example by which integrins indirectly mediate matrixremodelling. Recently, reports that c2/f31, a collagen receptor, directly mediatescollagen gel contraction (Gullberg et al. ‘90, Schiro et al. ‘91) provide evidence thatintegrins can mediate matrix remodelling by cueing physical events. Cells lacking a2or expressing mutant forms of this integrin failed to contract collagen gels. Yet,when a2 cDNA was transfected into cells lacking VLA-2, the cells expressed VLA-2at their surfaces and acquired the ability to contract collagen gels. Interestingly,HOSE cells express VLA-2, so this integrin may mediate organoid contraction bymediating interactions between HOSE cells and collagen. It would be interesting to seeif the HOSE cells which failed to contract organoids have insufficient or altered VLA2 expression. Others suggest that collagen gel contraction requires the co-operativeinteraction between VLA-2 and a second integrin, presumably VLA-5, a fibronectinreceptor (Clark ‘90). This is interesting as HOSE cells express both VLA-2 and VLA5 and they, or other integrins, may function co-operatively to mediate organoidcontraction.While the exact mechanism of gel contraction is unknown, a number of factors havebeen reported to influence gel contraction. For example, cell number andconcentration of collagen gel relate directly to the degree of gel contraction (Bell etal. ‘79, Ura et al. ‘91). Similarly, serum (Anderson et al. ‘90, Buttle & Ehrlich ‘83,Gullberg et al. ‘90), transforming growth factor-n (Montesano & Orci ‘88, Ura et al.‘91), and platelet-derived growth factor (Anderson et al. ‘90, Gullberg et al. ‘90)increase the degree of gel contraction while factors such as heparin (Graham et al.‘87, Guidry & Grinnell ‘87) decrease the degree of gel contraction. EGF andtransforming growth factor-a appear to have no effect on gel contraction (Montesano& Orci ‘88, Ura et al. ‘91) which is in aggreement with the results in the presentstudy. The addition of EGF/HC, which are mitogenic for HOSE cells, and reducedserum levels had no effect on the degree of organoid contraction. However, HOSE cellsmay be independent of the effects of some exogenous growth factors which influence111gel contraction because HOSE cells are capable of autonomous production of growthfactors (Auersperg et al. ‘91 a). Potential mechanisms by which growth factors suchas platelet-derived growth factor or transforming growth factor-f3 influence gelcontraction include increased synthesis or reorganization of actin or matrixreceptors, and stimulation of cell migration (Heino et al. ‘89, Ignotz & Massague‘86, Leof et al. ‘86). In the present study, the addition of 3T3 fibroblasts to HOSEorganoids increased the degree of contraction. This is in aggreement with otherreports which employed the co-culture of stromal and epithelial components to moreaccurately represent a reconstruction of an in vivo situation (Schafer et al. ‘89, Uraet al. ‘91).The use of fibroblast mediated collagen gel contraction and HOSE mediated organoidcontraction appear to be relevant models systems for wound healing (Bell et al. ‘79).However, other potential uses of collagen gel contraction as a model for many normaland abnormal cellular activities exist. These include: morphogenesis as exemplifiedby gland formation (Nogawa & Nakanishi ‘87); disease states such as scirrhouscarcinoma of the stomach (Ura et al. ‘91); normal skin wrinkling associated withaging (Schafer et al. ‘89); and abnormal skin wrinkling associated withepidermolysis bullosa (Ehrlich et al. ‘83). Hopefully, HOSE organoids may provide amodel system for the study of normal as well as abnormal ovarian conditions, suchovarian shrinkage with age and cyst formation.(3.) Effect of Substrata on HOSE Morphology. Growth. Integrin Expression andProtease ProductionThe complexitity of the organoid system made it difficult to examine therelationships between HOSE cells and specific matrix components. To overcome thisproblem HOSE cells were maintained on simpler substrata. These substrata (plastic,collagen gels, fibrin clots, and Matrigel) had profound effects on HOSE cellmorphology, growth, integrin expression, and protease production. These substratainduced morphological changes in HOSE cells that resemble their in vivocounterparts during various morphogenetic processes and reflect the dynamic natureof HOSE as well as its plasticity and heterogeneity both in vivo and in vitro. HOSEcells remained epithelial on both plastic and fibrin clots, however, while they were acohesive monolayer varying from flat epithelial cells to compact epithelial cells onplastic, they assumed a more dispersed epithelial phenotype on fibrin clots. Oncollagen gels, HOSE cells assumed a spindle-shaped morphology, and they formedaggregates of rounded cells that joined other aggregates via branching structures on112Matrigel. HOSE cells proliferated rapidly on plastic only. Interestingly, in spite ofthe dramatic morphological changes in response to substrata, the cells remainedkeratin positive indicating that they were still epithelial cells.In its normal resting state (non-proliferating and non-migratory), HOSE consistsof a cohesive simple monolayer that varies from squamous to columnar (Blaustein‘81a,’84, Blaustein & Lee ‘79, Clement ‘87, Nicosia ‘83, Papadaki & Beilby ‘71).Following ovulation, the continuity of HOSE is disrupted and it has been suggestedthat ovarian surface epithelial cells surrounding the ovulatory site proliferate andmigrate over and into the ovulatory defect to contribute to wound repair (Bullough‘42, Osterholzer et al. ‘85, Van Blerkom & Motta ‘79). Those cells which proliferatefrom the wound edge and migrate into the ovulatory defect are the source of type Bcells according to Gillet et al. (‘91). Once repair is completed, the continuity of HOSEis restored and HOSE returns to its resting state. There has been no detailed reportthat HOSE cells proliferate only at the edges of the ovulatory wound site and that theymigrate over and into the ovulatory defect as dispersed epithelial cells. However,endothelial cells and other epithelial cells which grow as cohesive monolayers inculture, proliferate only at the wound edges when the continuity of the monolayer isdisrupted. Following wounding, cells migrate from the wound edges to cover thedefect (Kadish et al. ‘79, Wong & Gotlieb ‘88). Endothelial cells, mesodermallyderived like HOSE, respond to fibrin clots in the same manner as HOSE cells. Fibrinclots stimulate a morphological change in endothelial cell monolayers so thatcohesive epithelial monolayers become more dispersed (Kadish et al. ‘79). Fibrinclots also stimulate endothelial cell migration (Kadish et al. ‘79, Wong & Gotlieb‘88). Therefore, fibrin clots do not stimulate endothelial cell growth, but provide amigratory cue and cause an alteration in cellular morphology. It is tempting tospeculate that, in vivo, HOSE cells also proliferate only at the wound edges and thenmigrate onto the fibrin clot as dispersed epithelial cells, and that this behaviour isrepresented in vitro in the response of HOSE cells to fibrin clots. Fibrin clots, then,may provide similar migratory cues to HOSE cells as to endothelial cells.Collagen gels induce a phenotypic modulation of HOSE cells to a spindle-shapedmorphology, a loss of integrin expression, and an inhibition of proliferation whilekeratin expression is retained. These collagen gel-induced responses may representan epithelial-mesenchymal transformation of HOSE cells similar to the epithelialmesenchymal transformations that can be induced by suspending embryonic anteriorlens epithelial cells (Greenburg & Hay ‘82,’86), adult thyroid follicular cells(Greenburg & Hay ‘88), or Madin-Darby canine kidney cells (Zuk & Hay ‘90, Zuk et113al. ‘89) within collagen gels. In particular, during the collagen-induced epithelialmesenchymal transformation of thyroid follicular cells, the epithelial cells assumeda spindle-shaped morphology, lost the capacity to produce thyroglobulin and lost theexpression of x3f1 and a3i integrins while keratin expression was retained(Greenburg & Hay ‘88). Further, it was shown that transformation of thyroidepithelium to mesenchyme-like cells did not require cellular proliferation(Greenburg & Hay ‘88). There may be other explanations for the lack of HOSE cellproliferation on collagen gels. For example, many epithelial cells acquire adifferentiated phenotype on rimmed collagen gels and do not proliferate on suchcollagen gels (Emerman et al. ‘77, Emerman & Pitelka ‘77). Alternatively, HOSEexposed to stromal matrix in vivo might be stimulated to proliferate. Disruption ofmatrix in vivo is often associated with the release of growth factors sequestered inthe matrix (Gospodarowicz et al. ‘83, Roberts et al. ‘88, Vlodavsky et al. ‘87).Perhaps growth factors are sequestered from the culture medium by the collagen gel,thereby depriving HOSE cells of the necessary growth stimulatory signals.Like HOSE cells on fibrin clots and collagen gels, HOSE cells on Matrigel did notgrow. A lack of proliferation may reflect inadequate culture conditions or, morelikely, deficient cell anchorage. However, if the cell aggregates formed on Matrigelare a reflection of in vivo morphogenesis as seen in sex cord formation (Gondos ‘75),or Mullerian duct formation (Josse & Picard ‘86, Vigier et al. ‘89), or, evenpossibly, invasiness as demonstrated by ovarian carcinoma cells (Blaustein ‘81a,Scully ‘77) these morphogenetic events may not normally be associated withproliferation. Indeed, at midgestation HOSE undergoes periods of intense mitoticactivity, however, this appears to be limited to HOSE at the ovarian surface and notto HOSE cells within sex cords (Gondos ‘75). An important point demonstrated by thecapacity of HOSE cells to penetrate Matrigel is that invasion appears to be part of thenormal HOSE phenotype. Therefore, HOSE cells can be added to the increasing list ofnormal cells capable of invading Matrigel (Allen et al. ‘90, Fisher et al. ‘89, Noel etal. ‘91, Sanders & Prasad ‘89). This also emphasizes that invasion into Matrigel isnot a characteristic restricted to malignant cells as is so often reported (Erkell &Schirrmacher ‘88, Starkey et al. ‘87). Care must be employed when using thiscriterion as a marker for malignancy. It also emphasizes the necessity for normalcellular counterparts when studying pathological and malignant disease states. Theculture of normal HOSE, then, should not be excluded from studies utilizing ovarianepithelial cancer cells.114The different HOSE morphologies on the various substrata did not translate into theexpression of different integrins at the cell surfaces. Instead, HOSE cells expressedthe integrins VLA-2,-3,-5, and VNR, receptors for collagen, laminin, fibronectin,and vitronectin when maintained on plastic, fibrin clots, and Matrigel. However, allHOSE cells lacked the VLA-6 (a6/4) integrin, the so-called epithelial integrin(Kajiji et al. ‘89, Tamura et a!. ‘90). The lack of expresion of the epithelial integrinby HOSE cells may be attribulable to the mesodermal origin of this epithelium. Morelikely, the term epithelial integrin is a misnomer as non-epithelial cells, such asosteosarcoma cells (Dedhar & Saulnier ‘90), express this integrin. Interestingly,time course studies suggested that there is a downregulation of integrin receptorswhen HOSE cells are maintained on collagen gels. The mechanism of thedownregulation remains unclear, but may be analogous to the EGF:receptor model(Cohen ‘87, Gill et al. ‘87). EGF-responsive cells often upregulate their receptorsin the absence of ligand. When the ligand, EGF, is provided, the ligand-receptorcomplex is internalized. The net effect is to downregulate the number of receptors atthe cell surface. Maybe, in the case of HOSE cells and integrins, the cells maintain ahigh level of integrins at their cell surface on plastic and fibrin because ligands fortheir receptors are unavailable. Although Matrigel contains ECM components, byvirtue of its non-adherent effect on HOSE the cells on Matrigel are hardly in contactwith the matrix and might again maintain a high level of integrins at their cellsurfaces. Even when HOSE cells penetrate Matrigel, there always appears to be azone, probably lytic in nature, separating HOSE cells from Matrigel (see figure 31).Therefore, like plastic and fibrin clots, HOSE cells in Matrigel may not be in contactwith laminin or collagen, but perhaps it is only on collagen gels that HOSE cells arein contact with ECM. There may be other mechanisms involved as suggested by theliterature (Dedhar & Saulnier ‘90, Guo et a!. ‘91). It would be interesting to see ifsimilar changes occur on fibrin clots probed for fibrin receptors (Du et a!. ‘91,Philips et al. ‘91, Rouslahti ‘91). Alternatively, since integrin downregulation isseen only in spindle-shaped HOSE cells on collagen gel, perhaps integrindownregulation is associated with this specific morphology. Alterations in integrinexpression, then, may be related to cell shape similarly to the effect of cell shape onprotease production by epithelial cells, where the more rounded the epithelialperiodontal ligament cells the greater the amount of proteases produced (Hong &Brunette ‘87). It would be interesting to compare, then, the integrin profilesbetween epithelial HOSE (i.e. routinely maintained on plastic) and spindle-shapedHOSE cells induced by other means than collagen, such as following treatment with115EGF (Siemens & Auersperg ‘88). While the expression of HOSE integrins in vivo isunkown, it is tempting to speculate that, at least in culture, VLA-2, -3, and -5,mediate HOSE cell-cell interactions as has been demonstrated between normalkeratinocytes (Carter et al. ‘90, Larjava et al. ‘90) and several human tumour celllines (Kaufmann et al. ‘89). Such integrin-mediated cell-cell interactions mayprovide cohesiveness between HOSE cells for the maintainence of monolayers.This is the first demonstration of protease production by HOSE cells. HOSE cellsproduce a chymotrypsin-like peptidase, an elastase-like peptidase, and 30 KD and42KD gelatinases which are metalloproteases. This has implications for normalovarian functioning such as the initiation and repair of ovulation, and for normaldevelopment such as sex cord formation. Like invasion, protease production appearsto be part of the normal HOSE phenotype and not limited to the malignant phenotype.However, the 72KD and/or the 94KD type IV collagenase, recognized to be importantfor normal events such as wound healing (Granda et al. ‘90) and tumour invasion(Liotta et at. ‘83) and most often associated with normal connective tissue cells(Emonard & Grimaud ‘90, Murphy et at. ‘89, Overall et al. ‘89) and malignant cells(Brown et at. ‘90, Davis & Martin ‘90, Emonard & Grimaud ‘90, Liotta et al. ‘79,Murphy et al. ‘89, Okada et al. ‘90, Zucker et al. ‘89) were not found associated withHOSE cells. The exact mechanism of regulation of protease production by HOSE cellsremains unknown. However, it would be interesting to see if protease production byHOSE cells is under the same hormonal influences as proteases produced by otherovarian cell types such as the granulosa and thecal cells (Curry et at. ‘89,Dhanasekaran & Moudgal ‘88, Ny et al. ‘85, Reich et al. ‘85a,b, Tilly & Johnson‘87).PA-I is common among many cell types (Cicilia et al. ‘89, Ginsburg et at. •86,Levin ‘83, Nielsen et aI. ‘82, Ny et al. ‘86, Pannekoek et al. ‘86, Takahashi et at.‘91) and, so, it is not surprising that it is present in HOSE cells. Recently,Takahashi et al. (‘91) showed that human omental mesothelial cells in culturerelease urokinase plasminogen activator (uPA) and plasminogen activator inhibitor(PAl-i), although they could not detect tissue type plasminogen activator (tPA). Theinability to detect tPA, like the lack of plasminogen activator (PA) detection fromHOSE cells in this study, may be due to the co-expression of large amounts of PAl-iwhich is known to regulate PA activity by binding to PA (Blasi et al. ‘87, ‘90,Krukithof ‘88, Monge et at. ‘89, Seitfert et al. ‘90). Endothelial cells also producePA and PAl and it is suggested that tightly regulated PA and PAl secretion serve tomaintain the luminal surface of endothelial cells free of adherent material so that116blood flow is unimpeded (Collen ‘80, Van Hinsbergh et al. ‘87). Further, endothelia,mesothelia, and HOSE have nonadhesive surfaces and it has been suggested that PA andPAl secretion serves to prevent adhesion among organs coated with epithelia bypreventing or reducing the adhesion of material to these epithelial surfaces(Buckman et al. ‘76). Possibly, then, these same functions of PA and PAl could beattributed to lubricative functions of HOSE cells. Recently, an interestingrelationship was described between PAl-i and vitronectin in that PAl-i binds tovitronectin (Salonen et al. ‘89) in a uniform distribution on the substratum (Jensenet al. ‘90, Pollanen et al. ‘87). The components of the vitronectin:PAI-i complexretain cell attachment-promoting activity and the capacity to inhibit plasminogenactivator. This suggests that the complex regulates plasminogen activator-inducedlysis by cells at the sites, such as focal contacts, where vitronectin promotescellular adhesion and confines plasminogen activator activity to discrete sites ofpericellular proteolysis (Jensen et al. ‘90, Pollanen et al. ‘87, Seiffert et al. ‘90).Perhaps the vitronectin receptor on HOSE cells co-distributes with PAl-iexpression to similarly limit plasminogen activator-mediated cellular proteolysisand maintain cellular adhesion.In summary, the role of HOSE in the remodelling of the ovarian cortex appears tobe broader than is generally assumed, involving synthetic, physical, and proteolyticfunctions. During the postovulatory phase, in particular, these cells may not onlyproliferate, migrate, and deposit a basement membrane, but may also play an activerole in the rebuilding of the underlying connective tissues.117SUMMARYThe human ovarian ovarian surface epithelium undergoes dynamic morphologicalchanges associated with development, ovulation, aging, and malignant transformation.Interactions between HOSE cells and ECM may influence these changes. In this study:1) improved methodology to culture HOSE cells from biopsy material was developed;2) cultured HOSE cells were further characterized; 3) an ovarian-derived ECM wasisolated and characterized; and 4) the interactions between HOSE cells and varioussubstrata were examined in an attempt to study the dynamic, pleomorphic, andmorphogenetic nature of HOSE.An improved methodology to culture HOSE cells from biopsy material was developedby changing the method of biopsy handling. By taking advantage of the tenuousattachment of HOSE to its underlying stroma, HOSE was scraped off the ovariansurface with a rubber policeman. Scraping generated epithelial fragments thatsettled in culture dishes and from which HOSE cells grew out as epithelialmonolayers. Scraping is superior to the explantation method previously described interms of speed, simplicity, and greater purity and yield of HOSE cells. An additionalculture technique was developed to eliminate mold from mold-infected cultures. Asimple one-step Percoll gradient centrifugation was developed, using the fungusCladosporium as a prototype. Centrifugation in Percoll removed mold contaminationand did not affect cell morphology, growth, or the ability of several cell types toproduce ECM suggesting that Percoll decontamination may be a generally applicabletechnique for decontamination of mold infected cultures.The list of HOSE characteristics in culture was expanded. HOSE cells maintained inculture continued to express a number of characteristics seen in their in vivocounterparts such as the co-expression of vimentin and keratin subtypes #7, 8, 18,and 19 and the expression of variable amounts of mucin and lipid. Unfortunately, todate, there is no unique marker for HOSE cells although it is possible to discriminateHOSE from contaminating cell types that might originate from biopsy material.A predominantly fibrous ECM was derived from an immortalized, but nontumorigenic rat ovarian surface epithelial cell line. This ROSE 199-ECM containedboth basement membrane and stromal ECM components and supported the attachment,spreading, and growth of several cell lines. It was hoped that HOSE cells maintainedon ROSE 199-ECM would assume a morphology similar to their in vivocounterparts, that is, a cuboidal monolayer, however, HOSE cells plated onto HOSE199-ECM dispersed as single, cells throughout the ROSE 199-ECM.118In terms of HOSE-ECM interactions, the results from this study also suggest thatHOSE plays an active role in ECM remodelling, that is, both the deposition anddegradation of ECM. HOSE cells produce both basement membrane and stromal matrixcomponents and this may be a common feature of ovarian surface epithelia. Thisstudy is the first report that HOSE cells secrete proteases. HOSE cells produce achymotrypsin-like protease, an elastase-like protease, and 3OKD and 42KDgelatinases which are metalloproteases. In addition to ECM remodelling via depositionof ECM material or enzymatic degradation of ECM, HOSE cells are also capable ofphysically remodelling ECM as evidenced by the ability of HOSE cells to contractspecific mixtures of ECM into organoids. HOSE-ECM interactions are notunidirectional as substrata also influence HOSE morphology, growth, and proteaseproduction. These capabilities may be important for normal ovarian development andnormal adult functions. While such proteolytic activities and invasiveness arecharacteristics of the malignant phenotype, they appear to be part of the normalHOSE phenotype. With regards to why HOSE is a preferred site of ovarian cancerdevelopment there may be many factors, but perhaps aberrations of these normalfunctions contribute to pathological states. HOSE cells in culture appear to be arelevant and a potentially valuable model system for the study of ovarian biology andcarcinogenesis.119REFERENCESAdams, A.M., J.V. Soames, and R.F. Searle. Effect of Amphotericin B on HumanPeriodontal Ligament Cells Cultured In Vitro. Archives of Oral Biology, 31:529-533. 1986.Adams, A.T. and N. Auersperg. Transformation of Cultured Rat Ovarian SurfaceEpithelial Cells by Kirsten Murine Sarcoma Virus. Cancer Research, 41:2063-2072. 1981.Adams, A. T. and N. Auersperg. Autoradiographic Investigation of Estrogen Binding inCultured Rat Ovarian Surface Epithelial Cells. Journal of Histochemistry andCytochemistry, 31:1321-132. 1983.Adams, A.T. and N. Auersperg. A Cell Line, ROSE 199, Derived from Normal RatOvarian Surface Epithelium. Experimental Cell Biology, 53:181-188. 1985.Achtstaetter, T., M. Hatzfeld, R.A. Quinlan, D.C. Parmelee, and W.W. Franke.Separation of Cytokeratin Polypeptides by Gel Electrophoretic and ChromatographicTechniques and Their Identification by Immunoblotting. Methods in Enzymology,134:355-371. 1986.Aggler, J. Three-Dimensional Organization of the Extracellular Matrix Secreted byCultured Rat Smooth Muscle Cells. In Vitro Cellular and Developmental Biology,24:633-638. 1988.Allen, L.E., L. Dubeau, 0. Alvarez, and P.A. Jones. Rapid Degradation of ExtracellularMatrix Proteins by Normal Human Uroepithelial Cells. Cancer Research, 50:1897-1904. 1990.Anderson, E. The Localization of Acid Phosphatase and the Uptake of HorseradishPeroxidase in the Oocyte and Follicle Cells of Mammals. In: Oogenesis. J.D. Biggersand A.W. Schuetz, eds. University Park Press, Baltimore, pp.87-118. 1972.Anderson, E., G. Lee, R. Letournea, D.F. Albertini, and S.M. Meller. CytologicalObservation of the Ovarian Epithelium in Mammals During the Reproductive Cycle.Journal of Morphology, 150:135-164. 1976.Anderson, S.N., Z. Ruben, and G.C. Fuller. Cell-Mediated Contraction of CollagenLattices in Serum-Free Medium: Effect of Serum and Nonserum Factors. In VitroCellular and Development Biology, 26:61-66. 1990.Andrews, P.M. and K.P. Porter. The Ultrastructureal Morphology and PossibleFunctional Significance of Mesothelial Microvilli. Anatomical Record, 177:409-426. 1973.Auersperg, N., E. Hajen, S.L. Maines-Bandiera, and J.W. Schrader. Production ofCytokines and Colony Stimulating Factors by Cultured Human Ovarian SurfaceEpithelium (OSE). Biology of Reproduction, 44:134. 1991a.Auersperg, N., P.A. Kruk, l.A. MacLaren, F.M. Watt, and S.E Myrdal. HeterogeneousExpression of Keratin, Involucrin, and Extracellular Matrix among Subpopulations120of a Poorly Differentiated Human Cervical Carcimona: Possible Relationships toPatterns of Invasion. Cancer Research, 49:3007-3014. 1989.Auersperg, N., MacLaren, l.A., and P.A. Kruk. Ovarian Surface Epithelium:Automonous Production of Connective Tissue-Type Extracellular Matrix. Biology ofReproduction, 44:717-724. 1991b.Auersperg, N., C.H. Siemens, and S.E. Myrdal. Human Ovarian Surface Epithelium inPrimary Culture. In Vitro Cellular and Developmental Biology, 20:743-755. 1984.Balboni, G.M. Structural Changes: Ovulation and Luteal Phase. In: The Ovary. G.B.Serra, ed. Raven Press, New York, pp.123-142. 1980.Balboni, G.M. Structural Changes: Ovulation and Luteal Phase. In: The Ovary. G.B.Serra, ed. Raven Press, New York, pp.177-190. 1980.Behbehani, A.M., W.J. Hunter, A.L. Chapman, and F. Lin. Studies of a HumanMesothelioma. Human Pathology, 9:862-866. 1982.Behrens, U.J. and F. Paronetto. Elimination of Yeast Canadida parapsilosis fromLymphoid Cells and Monolayer Cells in Culture. In Vitro Cellular and DevelopmentalBiology, 20:391-395. 1984.Bell, E., B. Ivarsson, and C. Merrill. Production of a Tissue-Like Structure byContraction of Collagen Lattices by Human Fibroblasts of Different ProliferativePotential in Vitro. Proceedings of the National Academy of Sciences of the UnitedStates of America, 76:1274-1278. 1979.Ben-Ze’ev, A. Control of Intermediate Filament Protein Synthesis by Cell-CellInteraction and Cell Configuration. FEBS Letters, 171:107-110. 1 984a.Ben-Ze’ev, A. Differential Control of Cytokeratins and Vimentin Synthesis by Cell-Cell Contact and Cell Spreading in Cultured Epithelial Cells. Journal of Cell Biology,99:1424-1433. 1984b.Ben-Ze’ev, A. The Relationship Between Cytoplasmic Organization, Gene Expression,and Morphogenesis. Trends in Biochemical Sciences, 11:478-481. 1986.Berchuk, A., G.C. Rodriguez, A. Kamel, R.K. Dodge, J.T. Soper, D.L. Clarke-Pearson,and R.C. Bast. Epidermal Growth Factor Receptor Expression in Normal OvarianEpithelium and Ovarian Cancer. I. Correlation of Receptor Expression withPrognostic Factors in Patients with Ovarian Cancer. American Journal of Obstetricsand Gynecology, 164:669-674. 1991.Bjersing, L. and S. Cajander. Ovulation and the Role of the Ovarian SurfaceEpithelium. Experientia, 31:605-608. 1975.Blasi, F., N. Behrendt, N.V. Cubellis, V. Ellis, L.R. Lund, M.T. Masucci, L.B. Moller,D.P. Plson, N. Pedersen, M. Ploug, E. Ronne, and K. Dano. The Urokinase Receptor andRegulation of Cell Surface Plasminogen Activation. Cell Differentiation andDevelopment. 32:247-254. 1990.Blasi, F., J. Vassalli, and K. Dano. Urokinase-Type Plasminogen Activator:Proenzyme, Receptor, and Inhibitors. Journal of Cell Biology, 104:801-804. 1987.121Blaustein, A. Anatomy and Histology of the Human Ovary. In: Pathology of the FemaleGenital Tract. A. Blaustein, ed. Springer-Verlag, New York, pp. 365-386. 1977a.Blaustein, A. Inflammatory Diseases of the Ovary. In: Pathology of the Female GenitalTract. A. Blaustein, ed. Springer-Verlag, New York, pp. 393-403. 1977b.Blaustein, A. Surface (Germinal) Epithelium and Related Ovarian Neoplasms.Pathology Annual, 16:247-294. 1981a.Blaustein, A. Surface Cells and Inclusion Cysts in Fetal Ovaries. GynecologicalOncology, 12:222-233. 1981b.Blaustein, A. Peritoneal Mesothelium and Ovarian Surface Cells - SharedCharacteristics. International Journal of Gynecological Pathology, 3:361 -375.1984.Blaustein, A., A. Kaganowicz, and J. Wells. Tumor Markers in Inclusion Cysts of theOvary. Cancer, 49:722-726. 1982.Blaustein, A. and H. Lee. Surface Cells of the Ovary and Pelvic Peritoneum: AHistochemical and Ultrastructure Comparison. Gynecologic Oncology, 8:34-43.1979.Brannstrom, M., B.M. Boberg, J. Tornell, P.O. Janson, and K. Ahren. Effects ofInhibitors of Protein Synthesis on the Ovulatory Process of the Perfused Rat Ovary.Journal of Reproduction and Fertility, 85:451-459. 1989.Brannstrom, M., J.F. Woessner, R.D. Koos, C.H.J. Sear, and W.J. LeMaire. Inhibitorsof Mammalian Tissue Collagenase and Metalloproteinases Suppress Ovulation in thePerfused Rat Ovary. Endocrinology, 122:1715-1721. 1988.Brauer, P.R. and J.M. Keller. Ultrastructure of a Model Basement Membrane LackingType IV Collagen. Anatomical Record, 223:376-383. 1989.Brown, P.D., A.T. Levy, I.M.K. Margulies, L.A. Liotta, and W.G. Stetler-Stevenson.Independent Expression and Cellular Processing of Mr 72,000 Type IV Collagenaseand Interstitial Collagenase in Human Tumorigenic Cell Lines. Cancer Research,50:6184-6191. 1990.Buckman, R.F., M. Woods, L. Sargent, and S. Gervin. A Unifying PathogeneticMechanism in the Etiology of lntraperitoneal Adhesions. Journal of SurgicalResearch, 20:1-5. 1976.Bullough, W.S. Oogenesis and Its Relation to the Oestrous Cycle in the Adult Mouse.Journal of Endocrinology, 3:141-149. 1942.Buttle, D.J. and H.P. Ehrlich. Comparative Studies of Collagen Lattice ContractionUtilizing a Normal and a Transformed Cell Line. Journal of Cellular Physiology,116:159-166. 1983.Byskov, A.G. Differentiation of Mammalian Embryonic Gonad. Physiological Reviews,66:71-117. 1986.122Cajander, S.B. Periovulatory Changes in the Ovary. Morphology and Expression ofTissue-Type Plasminogen Activator. In: Developments in Ultrastructure ofReproduction. P.M. Motta, ed. Alan R. Liss, Inc., New York, pp. 91-101. 1989.Cajander, S. and L. Bjersing. Ovarian Histochemsitry. In: The Ovary 1: GeneralAspects. P.L. Zuckerman and B.J. Weir, eds. Academic Press, New York, pp:303-391. 1977.Cajander, S.B., M.P. Hugin, P. Kristensen, and A.J.W. Hsueh. ImmunohistochemicalLocalization of Tissue-Type Plasminogen Activator in Ovaries Before and AfterInduced and Spontaneous Ovulation in the Rat. Cell and Tissue Research, 257:1 -8.1989.Carley, W.W., A.J. Milici, and J.A. Madri. Extracellular Matrix Specificity for theDifferentiation of Capillary Endothelial Cells. Experimental Cell Research,178:426-434. 1988.Carter, W.G., E.A. Wayner, T.S. Bouchard, and P. Kaur. The Role of Integrins afand a3J31 in Cell-Cell and Cell-Substrate Adhesion of Human Epidermal Cells.Journal of Cell Biology, 110:1387-1404. 1990.Casagrande, J.T., W.E. Louie, M.C. Pike, S. Roy, R.K. Ross, and B.E. Henderson.“Incessant Ovulation” and Ovarian Cancer. Lancet, 2:170-173. 1979.Cells, J.E., S.J. Fey, P.M. Larsen, and A. Cells. Preferential Phosphorylation ofKeratins and Vimentin During Mitosis in Normal and Transformed Human AmnionCells. Annals of the New York Academy of Sciences, 455:268-281. 1985.Centola, G.M. Structural Changes: Atresia. In: The Ovary. G.B. Serra, ed. Raven Press,New York, pp:113-122. 1980.Centola, G.M. Structural Changes: Follicular Development and HormonalRequirements. In: The Ovary. G.B. Serra, ed. Raven Press, New York, pp. 95-112.1983.Christian, C.D. Ovarian Tumors- An Extension of the Peutz-Jeghers Syndrome.American Journal of Obstetrics and Gynecology, 111:529-534. 1971.Cicilia, G.T., T.M. O’Connell, W.C. Hahn, and J.G. Rheinwald. Cloned cDNA Sequencefor the Human Mesothelial Protein “Mesosecrin” Discloses Its Identity as aPlasminogen Activator Inhibitor (PAl-i) and a Recent Evolutionary Change inTranscript Processing. Journal of Cell Science, 94:1-10. 1989.Clark, R.A.F. Fibronectin Matrix Deposition and Fibronectin Receptor Expression inHealing and Normal Skin. Journal of Investigative Dermatology, 94:128S-134S.1990.Clement, P.B. Histology of the Ovary. American Journal of Surgical Pathology,11:277-303. 1987.Cohen, S. Epidermal Growth Factor. In Vitro Cellular and Developmental Biology,23:239-246. 1987.123CoHen, D. On the Regulation and Control of Fibrinolysis. Thrombosis andHaemostasis, 43:77-89. 1980.Connell, N.D. and Rheinwald, J.G. Regulation of the Cytoskeleton in Mesothelial Cells:Reversible Loss of Keratin and Increase in Vimentin During Rapid Growth in Culture.Cell, 34:245-253. 1983.Cotchin, E. Spontaneous Tumors of the Uterus and Ovaries in Animals. In: Pathologyof the Female Genital Tract. A. Blaustein, ed. Springer Verlag, New York, pp. 822-861. 1977.Courtenay, V.D. Studies of Ovarian Tumours in Vitro. UICC Technical Report Series,50:121-130. 1980.Cramer, D.W., W.R. Welch, S. Cassells, and R.E. Scully. Mumps, Menarche,Menopause, and Ovarian Cancer. American Journal of Obstetrics and Gynecology,147:1-6. 1983.Cramer, D.W., W.R. Welch, R.E. Scully, and C.A. Wojciechowcki. Ovarian Cancer andTalc. Cancer, 50:372-376. 1982.Culling, C.F.A. Handbook of Histopathological and Histochemical Techniques. ThirdEdition. Butterworth & Co. Ltd., Toronto, p. 361. 1974.Curry, T.E., M. R. Clark, D.D. Dean, J.F. Woessner, and W.J. LeMaire. ThePreovulatory Increase in Ovarian Collagenase Activity in the Rat is Independent ofProstaglandin Production. Endocrinology, 118:1823-1828. 1986.Curry, T.E., D.D. Dean, S.L. Sanders, N.G. Pedigo, and P.B.C. Jones. The Role ofOvarian Proteases and Their Inhibitors in Ovulation. Steroids, 54:501-521. 1989.Curry, T.E., J.S. Mann, R.S. Estes, and P.B.C. Jones. c-2-Macroglobulin and TissueInhibitor of Metalloproteinases: Collagenase Inhibitors in Human PreovulatoryOvaries. Endocrinology, 127:63-68. 1990.Curry, T.E., S.L. Sanders, N.G. Pedigo, R.S. Estes, E.A. Wilson, and M.W. Vernon.Identification and Characterization of Metalloproteinase Inhibitor Activity in HumanOvarian Follicular Fluid. Endocrinology, 123:1611-1618. 1988.Czernobilsky, B. Co-expression of Cytokeratin and Vimentin Filaments inMesothelial, Granulosa, and Rete Ovaril Cells of the Human Ovary. European Journalof Cell Biology, 37:175-190. 1985.Czernobilsky, B., R, Moll, W.W. Franke, G. Dallenbach-Hellweg, and P. HohlwegMajert. Intermediate Filaments of Normal and Neoplastic Tissues of the FemaleGenital Tract with Emphasis on Problems of Differential Tumor Diagnosis. Pathology:Research and Practice, 179:31-37. 1984.Davis, C.M., V. Papadopoulos, C.L. Sommers, H.K. Kleinman, and M. Dym. DifferentialExpression of Extracellular Matrix Components in Rat Sertoli Cells. Biology ofReproduction, 43:869-869. 1990.1 24Davis, G.E. and B.M. Martin. A Latent Mr 94,000 Gelatin-Degrading MetalloproteaseInduced During Differentiaiton of HL-60 Promyelocytic Leukemia Cells: A Memberof the Collagenase Family of Enzymes. Cancer Research, 50:1113-1120. 1990.Dazois, R.R., R.D. Kempers, D.C. Dahlin, and L.G. Bartholomew. Ovarian TumorsAssociated with Peutz-Jaaghers Syndrome. Annals of Surgery, 172:233-238.1971.Dedhar, S., W.S. Argraves, S. Suzuki, E. Ruoslahti, and M.D. Pierschbacher. HumanOsteosarcoma Cells Resistant to Detachment by an Arg-Gly-Asp-Containing SyntheticPeptide Overproduce the Fibronectin Receptor. Journal of Cell Biology, 105:1175-1182. 1987.Dedhar, S., C. Haqq, and V. Gray. Specific Overproduction of Very Late Antigen 1Integrin in Two Human Neuroblastoma Cell Lines Selected for Resistance toDetachment by an Arg-Gly-Asp-Containing Synthetic Peptide. Journal of BiologicalChemistry, 264:4832-4836. 1989.Dedhar, S. and R. Saulnier. Alterations in Integrin Receptor Expression onChemically Transformed Human Cells: Specific Enhancement of Laminin and CollagenReceptor Complexes. Journal of Cell Biology, 110:481-489. 1990.Dhanasekaran, N. and N.R. Moudgal. Gonadotropin Regulation of Rat OvarianLysosomes: Existence of a Hormone Specific Dual Control Mechanism. BioscienceReports, 8:279-285. 1988.Dickman, K.G. and J.L. Renfro. Streptomycin Toxicity in Primary Cultures ofFlounder Renal Proximal Tubule Cells. In Vitro Cellular and Developmental Biology,26:571-578. 1990.Dietl, J., J. Henrich, and F. Buchholz. Surface Morphology of the Perfused RabbitOvary. Archives of Gynecology, 240:33-43. 1987.Dii, X., E.F. Plow, A.L. Frelinger, T.E. O’Toole, J.C. Loftus, and M.H. Ginsberg. Ligands“Activate” Integrin cxllbJ33 (Platelet GPllb-IIla). Cell, 65:409-416. 1991.Dubeau, L., M. Velicescu, A.E. Sherrod, G. Schreiber, and G. Holt. Culture of HumanFetal Ovarian Surface Epithelium in a Chemically-Defined, Serum-Free Medium: AModel for Ovarian Carcinogenesis. Anticancer Research, 10:1233-1240. 1990.Ehrlich, H.P., D.J. Buttle, R.L. Trelstad, and K. Hayashi. Epidermolysis BullosaDystrophica Recessive Fibroblasts Altered Behaviour Within a Collagen Matrix.Journal of Investigative Dermatology, 80:56-60. 1983.Ekblom, P. Basement Membrane Proteins and Growth Factors in KidneyDifferentiation. In: The Role of Extracellular Matrix in Development. Trelstad, R.L,ed. Alan R. Liss, New York, pp. 173-206. 1984.Elliott, W.M. and N. Auersperg. Growth of Normal Ovarian Surface Epithelial Cells(HOSE) in Reduced Serum and Defined Medium. Journal of Cell Biology, 111 :58A.1 990.125Emerman, J.T., J. Enami, D.R. Pitelka, and S. Nandi. Hormonal Effects onIntracellular and Secreted Casein in Cultures of Mouse Mammary Epithelial Cells onFloating Collagen Membranes. Proceedings of the National Academy of Sciences of theUnited States of America, 74:4466-4470. 1977.Emerman, J.T. and D.R. Pitelka. Maintenance and Induction of MorphologicalDifferentiation in Dissociated Mammary Epithelium on Floating Collagen Membranes.In Vitro Cellular and Developmental Biology, 13:316-328. 1977.Emonard, H. and J. Grimaud. Matrix Metalloproteinases. A Review. Cellular andMolecular Biology, 36:131-153. 1990.Erkell, L.J. and V. Schirrmacher. Quantitative in Vitro Assay for Tumor CellInvasion Through Extracellular Matrix or into Protein Gels. Cancer Research,48:6933-6937. 1988.Fisher, S.J., T. Cui, L. Zhang, L. Hartman, K. Grahl, Z. Guo-Yang, J. Tarpey, and C.H.Damsky. Adhesive and Degradative Properties of Human Placental CytotrophoblastCells in Vitro. Journal of Cell Biology, 109:891-902. 1989.Fox, H. Human Ovarian Tumours: Classification, Histogenesis, and Criteria forExperimental Models. UICC Technical Report Series, 50:22-33. 1980.Freshney, R.l. Contamination. In: Culture of Animal Cells: A Manual of BasicTechnique. Alan R. Liss, Inc., New York, pp. 181-188. 1983.Fromm, M. and P. Berg. Deletion Mapping of DNA Regions Required for SV4O EarlyRegion Promoter Function in Vivo. Journal Molecular and Applied Genetics, 1:457-481. 1982.Fukumoto, M., Y. Yajima, H. Okamura, and 0. Midorikawa. Collagenolytic EnzymeActivity in the Human Ovary: An Ovulatory Enzyme System. Fertility and Sterility,36:746-750. 1981.Gerard, G.E., P.M. Loenstein, and M.Green. Detection in Human Ovary and ProstateTumors of DNA Polymerase Activity That Copies Poly (2’-O-Methycytidylate)Oligodeoxyguanylate. Cancer Research, 38:1008-1011. 1978.Gill, G.N., P.J. Bertics, and J.B. Santon. Epidermal Growth Factor and Its Receptor.Molecular and Cellular Endocrinology, 51:169-186. 1987.Gillett, W.R. Artefactual Loss of Human Ovarian Surface Epithelium: PotentialClinical Significance. Reproduction Fertility and Development, 3:93-98. 1991.Gillet, W.R., A. Mitchell, and P.R. Hurst. A Scanning Electron Microscopic Study ofthe Human Ovarian Surface Epithelium: Characterization of Two Cell Types. HumanReproduction, 6:645-650. 1991.Ginsburg, D., R. Zeheb, A.Y. Yang, U.M. Rafferty, P.A. Andreasen, L. Nielsen, K. Dano,R.V. Lebo, and T.D. Gelehrter. cDNA Cloning of Human Plasminogen ActivatorInhibitor from Endothelial Cells. Journal of Clinical Investigation, 78:1673-1680.1986.126Goldring, N.B., Y. Farkash, D. Goldschmit, and J. Orly. Immunofluorescent Probing ofthe Mitochondrial Cholesterol Side-Chain Cleavage Cytochrome P-450 Expressed inDifferentiating Granulosa Cells in Culture. Endocrinology, 119:2821-2832. 1986.Gondos, B. Surface Epithelium of the Developing Ovary. American Journal ofPathology, 81:303-320. 1975.Gospodarowicz, D. Preparation of Extracellular Matrices Produced by CulturedBovine Corneal Endothelial Cells and PF-HR-9 Endodermal Cells: Their Use in CellCulture. In: Methods for Preparation of Media, Supplements, and Substrata forSerum-Free Animal Cell Culture. D.W. Barnes, D.A.Sirbasku, and G.H, Sato, eds. AlanR. Liss, Inc., New York, pp. 275-293. 1984.Gospodarowicz, D., R. Gonzalez, and D.K. Fujii. Are Factors Originating From Serum,Plasma, or Cultured Cells Involved in the Growth-Promoting Effect of theExtracellular Matrix Produced by Cultured Bovine Corneal Endothelial Cells?Journal of Cellular Physiology, 114:191-202. 1983.Graham, M.F., D.E.M. Drucker, H.A. Perr, R.F. Diegelmann, and H.P. Ehrlich.Heparin Modulates Human Intestinal Smooth Muscle Cells Proliferation, ProteinSynthesis, and Lattice Contraction. Gastroenterology, 93:801 -809. 1987.Granda, J.L., M.A. Lande, and R.L. Karvonen. A Human Cartilage Metalloproteinasewith Elastolytic Activity. Connective Tissue Research, 24:249-263. 1990.Greenburg, G. and E.D. Hay. Epithelia Suspended in Collagen Gels Can Lose Polarityand Express Characteristics of Migrating Mesenchymal Cells. Journal of CellBiology, 95:333-339. 1982.Greenburg, G. and E.D. Hay. Cytodifferentiation and Tissue Phenotype Change DuringTransformation of Embryonic Lens Epithelium to Mesenchyme-like Cells in Vitro.Developmental Biology, 115:363-379. 1986.Greenburg, G. and E.D. Hay. Cytoskeleton and Thyroglobulin Expression ChangeDuring Transformation of Thyroid Epithelium to Mesenchyme-Like Cells.Development, 102:605-622. 1988.Grinnell, F., H. Fukamizu, P. Pawelek, and S. Nakagawa. Collagen Processing,Crosslinking, and Fibril Bundle Assembly in Matrix Produced by Fibroblasts inLongterm Cultures Supplemented with Ascorbic Acid. Experimental Cell Research,181:483-491. 1989Grinnell, F. and C.R. Lamke. Reorganization of Hydrated Collagen Lattices by HumanSkin Fibroblasts. Journal of Cell Science, 66:51-63. 1984.Guidry, C. and F. Grinnell. Studies on the Mechanism of Hydrated Collagen GelReorganization by Human Skin Fibroblasts. Journal of Cell Science, 79:67-81.1985.Guidry, C. and F. Grinnell. Contraction of Hydrated Collagen Gels by Fibroblasts:Evidence for Two Mechanisms by Which Collagen Fibrils are Stabilized. Collagen andRelated Research, 6:515-529.1986.127Guidry, C. and F. Grinnell. Heparin Modulates the Organization of Hydrated CollagenGels and Inhibits Gel Contraction by Fibroblasts. Journal of Cell Biology,104:1097-1103. 1987.Gullberg, D., A. Tingstrom, A.C. Thuresson, L. Olsson, L. Terracia, T.K. Borg, and K.Rubin. f Integrin-Mediated Collagen Gel Contraction is Stimulated by PDGF.Experimental Cell Research, 186:264-272. 1990.Guraya, S.S. Histochemistry of the Ovary. In: The Biology of the Ovary. P.M. Mottaand E.S.E. Hafez, eds. Martinus Nijoff Publishers, Boston, pp. 33-51. 1980.Gou, M., L.T. Kim, S.T. Akiyama, H.R. Gralnick, K.M. Yamada, and F. Grinnell. AlteredProcessing of Integrin Receptors during Keratinocyte Activation. Experimental CellResearch, 195:315-322. 1991.Hadley, M.A., S.W. Byers, C.A. Suarez-Quian, H.K. Kleinman, and M. Dym.Extracellular Matrix Regulates Sertoli Cell Differentiaiton, Testicular CordFormation, and Germ Cell Development In Vitro. Journal of Cell Biology, 101 :1511-1522. 1985.Hamilton, T.C., P. Davies, and K. Griffiths. Steroid-Hormone Receptor Status of theNormal and Neoplastic Ovarian Surface Germinal Epithelium. In: Factors RegulationOvarian Function. G.S. Green and P.F. Terranova, eds. Raven Press, New York, pp.81-85. 1983.Hamilton, T.C., W.J. Henderson, and C. Eaton. Isolation and Growth of the Rat OvarianGerminal Epithleium. In: Tissue Culture in Medical Research (II). R.J. Richards andK.T. Rajan, eds. New York, Pergamon Press, New York, pp. 237-244. 1980.Hanas, J.S. and M.V. Simpson. Fluorescence Studies on a Streptomycin-InducedConformational Change in Ribosomes Which Correlates with Misreading. Journal ofBiological Chemistry, 261:6670-6676. 1986.Harris, A.K., P. Wild, and D. Stopak. Silicone Rubber Substrata: A New Wrinkle inthe Study of Cell Locomotion. Science, 208:177-179. 1980.Harrison, R.J. and B.J. Weir. Structure of the Mammalian Ovary. In: The Ovary 1:General Aspects. P.L. Zuckerman and B.J. Weir, eds. Academic Press, New York. pp.113-217. 1977.Hay, E.D. Matrix-Cytoskeletal Interactions in the Developing Eye. Journal ofCellular Biochemistry, 27:143-156. 1982.Heino, J., R.A. Ignotz, M.E. Hemler, C. Crouse, and J. Massague. Regulation of CellAdhesion Receptors by Transforming Growth Factor-f’. Journal of BiologicalChemistry, 264:380-388. 1989.Henderson, W.J., C.A.F. Joslin, and A.C. Turnbull. Talc and Carcinoma of the Ovaryand Cervix. Journal of Obstetrics and Gynecology, 78:266-272. 1971.Heussen, C. and E. B. Dowdle. Electrophoretic Analysis of Plasminogen Activators inPolyacrylamide Gels Containing Sodium Dodecyl Sulfate and CopolymerizedSubstrates. Analytical Biochemistry, 102:196-202. 1980.128Hirshfield, A.N. Development of Follicles in the Mammalian Ovary. InternationalReview of Cytology, 124:43-101. 1991.Hong, H.L. and D.M. Brunette. Effect of Cell Shape on Proteinase Secretion byEpithelial Cells. Journal of Cell Science, 87:259-267. 1987.Hoyer, P.E. Histoenzymology of the Human Ovary: Dehydrogenases Directly Involvedin Steroidogenesis. In: The Biology of the Ovary. P.M. Mofta and E.S.E. Hafez, eds.Martinus Nijoff Publishers, Boston, pp. 52-67. 1980.lgnotz, R.A. and J. Massague. Transforming Growth Factor-f Stimulates theExpression of Fibronectin and Collagen and Their Incorporation into theExtracellular Matrix. Journal of Biological Chemistry, 261 :4337-4345. 1986.Jackson, J.A., A.C. Friberg, and J.M. Bahr. Preovulatory Changes inGlycosaminoglycans and Collagen Content in the Stigma Region of the Follicle of theDomestic Hen. Biology of Reproduction, 45:301-307. 1991.Jensen, P.J., M. John, and J. Baird. Urokinase and Tissue Type PlasminogenActivators in Human Keratinocyte Culture. Experimental Cell Research, 187:162-169. 1990.Joly, D.J., A.M. Lilienfeld, E.L. Diamond, and I.D.J. Bross. An Epidemiologic Study ofthe Relationship of Reproductive Experience to Cancer of the Ovary. AmericanJournal of Epidemiology, 99:190-209. 1974.Josse, N. and J.Y. Picard. Anti-Mullerian Hormone. Physiological Reviews,66:1038-1090. 1986.Kadish, J.L., C.E. Butterfield, and J. Folkman. The Effect of Fibrin on CulturedVascular Endothelial Cells. Tissue and Cell, 11:99-108. 1979.Kaibara, M., E. Fukada, and K. Sakaoku. Rheological Study on Network Structure ofFibrin Clots Under Various Conditions. Bioreheology, 18:23-35. 1981.Kajiji, S., R. Tamura, and V. Quaranta. A Novel Integrin (aE4) from HumanEpithelial Cells Suggests a Fourth Family of Integrin Adhesion Receptors. EMBOJournal, 8:673-680. 1989.Kaufmann, R., D. Frosch, C. Westphal, L. Weber, and C.E. Klein. Integrin VLA-3:Ultrastructural Localization at Cell-Cell Contact Sites of Human Cell Cultures.Journal of Cell Biology, 109:1807-1815. 1989.Kerner, H., E. Gaton, and B. Czernobilsky. Unusual Ovarian, Tubal, and PelvicMesothelial Inclusions in Patients with Endometriosis. Histopathology, 5:277-283.1981.Kiernan, J.A. Histological and Histochemical Methods: Theory and Practice. 2ndEdition. Pergamon Press, Toronto, pp. 92-102. 1981.Kinbrough, R.A. Coincident Carcinoma of Ovary in Twins. American Journal ofObstetrics and Gynecology, 18:148-149. 1929.129Kleinman, H.K., M.L. McGarvey, J.R. Hassell, V.L. Star, F.B. Cannon, G.W. Laurie,and G.R. Martin. Basement Membrane Complexes with Biological Activity.Biochemistry, 25:312-318. 1986.Kleinman, H.K., M.L. McGarvey, L.A. Liotta, P.G. Robey, and K. Tryggvason. Isolationand Characterization of Type IV Procollagen, Laminin, and Heparan SulfateProteoglycan from the EHS Sarcoma. Biochemistry 21:6188-6193. 1982.Knecht, E., J.L. Vargas, F. Aniento, J. Cervera, and S. Grisolia. Effects ofCentrifugation on the Degradation of Short-Lived Proteins in Exponentially GrowingCultured Cells. Experimental Cell Research, 182:307-320. 1989.Kreamer, B.L., J.L. Staecker, N. Sawada, G.L. Sattler, M.T.S. Hsia, and H.C. Pitot. Useof Low-Speed, Iso-Density Percoll Centirfugation Method to Increase the Viability ofIsolated Rat Hepatocyte Preparation. In Vitro Cellular and Developmental Biology,22:201-211. 1986.Kruithof, E.K.O. Plasmingen Activator Inhibitors - A Review. Enzyme, 40:113-121.1 988.Kruk, P.A., S.L. Maines-Bandiera, and N. Auersperg. A Simplified Method to CultureHuman Ovarian Surface Epithelium. Laboratory Investigation, 63:132-136. 1990.Laemmli, U.K. Cleavage of Structural Proteins During the Assembly of the Head ofBacteriophage T4. Nature, 227:680-685. 1970.Lane, E.B., B.L.M. Hogan, M. Kurkinen, and J.l. Garrels. Co-Expression of Vimentinand Cytokeratins in Parietal Endoderm Cells of Early Mouse Embryo. Nature,303:701-704. 1983.Laska, D.A., P.D. Williams, and S.L. White. In Vitro Correlation of UltrastructuralMorphology and Creatine Phosphokinase Release in L6 Skeletal Muscle Cells AfterExposure to Parenteral Antibiotics. In Vitro Cellular and Developmental Biology,26:393-398. 1990.Larjava, H., J. Peltonen, S.K. Akiyama, S.S. Yamada, H.R. Gralnick, J. Uitto, and K.M.Yamada. Novel Function for Bi Integrins in Keratinocyte Cell-Cell Interactions.Journal of Cell Biology, 110:803-815. 1990.Lee, E.Y.H., G. Parry, and M.J. Bissell. Modulation of Secreted Proteins of MouseMammary Epithelial Cells by the Collagenous Substrata. Journal of Cell Biology,98:146-155. 1984.Leof, E.B., J.A. Proper, M.J. Getz, and H.L. Moses. Transforming Growth Factor Type3 Regulation of Actin mRNA. Journal of Cellular Physiology, 127:83-88. 1986.Levin, E.G. Latent Tissue Plasminogen Activator Produced by Human Endothelial Cellsin Culture: Evidence for an Enzyme-Inhibitor Complex. Proceedings of the NationalAcademy of Sciences of the United States of America, 80:6804-6808. 1983.Li, M.L., J. Aggeler, D.A. Farson, C. Hatier, J. Hassell, and M.J. Bissell. Influence of aReconstituted Basement Membrane and Its Components on Casein Gene Expression and130Secretion in Mouse Mammary Epithelial Cells. Proceedings of the National Academyof Sciences of the United States of America, 84:136-140. 1987.Liber, A.F. Ovarian Cancer in Mother and Five Daughters. Archives of Pathology,49:280-290. 1950.Lingeman, C.H. Etiology of Cancer of the Human Ovary: A Review. Journal of theNational Cancer Institue, 53:1673-1681. 1974.Liotta, L.A., S. Abe, R.P. Gehron, and G.R. Marlin. Preferential Digestion of BasementMembrane Collagen by an Enzyme Derived from a Metastatic Murine Tumor.Proceedings of the National Academy of Sciences of the United States of America,76:2268-2272. 1979.Liotta, L.A., C.W. Lee, and D.J. Morkakis. New Method for Preparing Large Surfacesof Intact Human Basement Membrane for Tumor Invasion Studies. Cancer Letters,11:141-152. 1980.Liotta, L.A., C.N. Rao, and S.H. Barsky. Tumor Invasion and the Extracellular Matrix.Laboratory Investigation, 49:636-649. 1983.Longo, D.L. and R.C. Young. Cosmetic Talc and Ovarian Cancer. Lancet, 2:349-351.1979.Madison, R., C.F. DaSilva, P. Dikkes, T.H. Chiu, and R.L. Sidman. Increased Rate ofPeripheral Nerve Regeneration Using Bioresorbable Nerve Guides and a LamininContaining Gel. Experimental Neurology, 88:767-772. 1985.Makabe, S., A. Iwaki, E.S.E. Hafez, and P.M. Motta. Physiomorphology of Fertile andInfertile Human Ovaries. Develoments in Obstetrics and Gynecology, 23:279-290.1 980.Marchant, J. Animal Models for Tumors of the Ovary. UICC Technical Report Series,50:50-65. 1980.Mattison, D.R. and S.S. Thorgeirsson. Smoking and Industrial Pollution, and TheirEffect on Menopause and Ovarian Cancer. Lancet, 1:187-188. 1978.Mckay, D.G., J.M.H. Pinkerton, A.T. Hertig, and S. Danziger. The Adult Human Ovary:A Histochemical Study. Obstetrics and Gynecology, 18:13-39. 1961.Mckeehan, W.L., D.P. Genereux, and R.G. Ham. Assay and Purification of FactorsFrom Serum that Control Multiplication of Human Diploid Fibroblasts. Biochemicaland Biophysical Research Communications, 80:1013-1021. 1978.Meizner, I., A. Levy, M. Katz, A.J. Maresh, and M. Glezerman. Fetal Ovarian Cysts:Prenatal Ultrasonographic Detection and Postnatal Evaluation and Treatment.American Journal of Obstetrics and Gynecology, 164:874-878. 1991.Merkel, G.J. and C.L. Phelps. Factors Influencing the Interaction of Canadida albicanswith Fibroblast Cell Cultures. Infection and Immunity, 56:792-801. 1988.Miller, R.W. Radiation Induced Cancer. Journal of the National Cancer Institute,49:1221-1227. 1972.131Moll, R., Franke, W.W., Schiller, D.L., Geiger, B., and Krepler, R. The Catalog ofHuman Cytokeratins: Review Patterns of Expression in Normal Epithelia, Tumorsand Cultured Cells. Cell, 31:11-24. 1982.Monge, J.C., C.L. Lucore, E.T.A. Fry, B.E. Sobel, and J.J. Billadello. Characterizationof Interaction of Active-Site Serine Mutants of Tissue-Type Plasminogen Activatorwith Plasminogen Activator-i. Journal of Biological Chemistry, 264:10922-10925. 1989.Montesano, R. Cell-Extracellular Matrix Interactions in Morphogenesis: An In VitroApproach. Experientia, 42:977-985. 1986.Montesano, R. and L. Orci. Transforming Growth Factor J3 Stimulates Collagen-MatrixContraction by Fibroblasts: Implications for Wound Healing. Proceedings of theNational Academy of Sciences of the United States of America, 85:4894-4897.1988.Moore, K. The Urogenital System. In: The Developing Human. 3rd ed. W.B. SaundersCo., Philidelphia, pp. 255-297. 1982.Motta, P.M., J. Van Blerkom, and S. Makabe. Changes in the Surface Morphology ofOvarian ‘Germinal’ Epithelium During the Reproductive Cycle and in SomePathological Conditions. Journal of Submicroscopic Cytology, 12:407-425. 1980.Motta, P.M. and J. Van Blerkom. Scanning Electron Miscroscopy of the MammalianOvary. In: Biology of the Ovary. P.M. Motta and E.S.E. Hafez, eds. MartinusPublishers, Boston. pp. 162-175. 1980.Motta, P.M. and S. Makabe. Development of the Ovarian Surface and Associated GermCells in the Human Fetus. Cell and Tissue Research, 226:493-510. 1982.Murphy, E.D. Major Experimental Models: Histogenesis and Evaluation. UICCTechnical Report Series, 50:66-73. 1980.Murphy, G., J.J. Reynold, and R.M. Hembry. Metalloproteinases and Cancer Invasionand Metastasis. International Journal of Cancer, 44:757-760. 1989.Newhouse, M.L., G. Berry, J.C. Wagner, and M.E. Turok. A Study of the Mortality ofFemale Asbestos Workers. British Journal of Industrial Medicine, 29:134-141.1972.Newman, S.A., D.A. Frenz, E. Hasegawa, and S.K. Akiyama. Matrix-DrivenTranslocation: Dependence on Interaction of Amino-Terminal Domain of Fibronectinwith Heparin-Like Surface Components of Cells or Particles. Proceedings of theNational Academy of Sciences of the United States of America, 84:4791-4795.1987.Newman, S.A., D.A. Frenz, and J.J. Tomasek. Matrix-Driven Translocation of Cellsand Nonliving Particles. Science, 228:885-889. 1985.Nicosia, S.V. Morpholgoical Changes in the Human Ovary Throughout Life. In: TheOvary. G.B. Serra, ed. Raven Press, New York, pp. 57-81. 1983.132Nicosia, S.V. The Aging Ovary. Medical Clinics of North America, 71:1-9. 1987.Nicosia, S.V. and J.H. Johnson. Effects of Two International Union Against CancerStandard Reference Samples of Asbestos on Cultured Rabbit Ovarian SurfaceEpithelial Cells. Scanning Electron Microscopy, 3:1329-1336. 1984a.Nicosia, S.V. and J.H. Johnson. Surface Morphology of Ovarian Mesothelium (SurfaceEpithelium) and of Other Pelvic and Extrapelvic Mesothelial Sites in the Rabbit.International Society Gynecological Pathology, 3:249-260. 1 984b.Nicosia, S.V., J.H. Johnson, and E.J. Streibel. Isolation and Ultrastructure of RabbitOvarian Mesothelium (Surface Epithelium). International Journal of GynecologicalPathology, 3:348-360. 1984.Nicosia, S.V., J.H. Johnson, and E.J. Streibel. Growth Characteristics of RabbitOvarian Mesothelial (Surface Epithelial) Cells. International Journal ofGynecological Pathology, 4:58-74. 1985.Nicosia, S.V., R.J. Narconis, and B.O. Saunders. Regulation of Temporal Sequence ofSurface Epithelium Morphogenesis in the Postovulatory Rabbit Ovary. Progress inClinical and Biological Research, 296:111-119. 1989.Nicosia, S.V. and R.F. Nicosia. Neoplasms of the Ovarian Mesothelium. In: Pathology ofHuman Neoplasms. H.A. Azat, ed. Raven Press, New York, pp. 435-486. 1988.Nielsen, L.S., J.G. Hansen, L. Shriver, E.L. Wilson, K. Kaltoft, J. Zeuthen, and K.Dano. Purification of Zymogen to Plasminogen Activator from Human GlioblastomaCells by Affinity Chromatography with Monoclonal Antibody. Biochemistry,21:6410-6415. 1982.Noel, A.C., A. Calle, H.P. Emonard, B.V. Nusgens, L. Simar, J. Foidart, C.M. Lapiere,and J.M. Foidart. Invasion of Reconstituted Bssement Membrane Matrix Is NotCorrelated to the Malignant Metastatic Cell Phenotype. Cancer Research, 51:405-414. 1991.Nogawa, H. and Y. Nakanishi. Mechanical Aspects of the Mesenchymal Influence onEpithelial Branching Morphogenesis of Mouse Salivary Gland. Development,101:491-500. 1987.Normura, T. Carcinogenesis by Urethane Via Mother’s Milk and Its Enhancement ofTranspiacental Carcinogenesis in Mice. Cancer Research, 33:1677-1683. 1977.Ny, T., L. Bjersing, A.J.W. Hsueh, and D.J. Loskutoff. Cultured Granulosa CellsProduce Two Plasminogen Activators and An Antiactivator, Each RegulatedDifferently by Gonadotropins. Endocrinology, 116:1666-1668. 1985.Ny, T., M. Sawdey, D. Lawrence, J.L. Millan, and D.J. Loskutoff. Cloning and Sequenceof a cDNA Coding for the Human B-Migrating Endothelial-Cell-Type PlasminogenActivator Inhibitor. Proceedings of the National Academy of Sciences of the UnitedStates of America, 83:6776-6780. 1986.O’Grady, R.L., A. Nethery, and N. Hunter. A Fluorecent Screening Assay forCollagenase Using Collagen Labeled with 2-Methoxy-2,4-Diphenyl-3(2H)-Furanone. Analytical Biochemistry, 140:490-494. 1984.133O’Guin, W.M., Schemer, A., and T.-T. Sun. Immunofluorescence staining of KeratinFilaments in Cultured Epithelial Cells. Journal of Tissue Culture Methods, 9:123-128, 1985.Okada, Y., H. Tsuchiya, H. Shimizu, K. Tomita, I. Nakanishi, H. Sato, M. Seiki, K.Yamashita, and T. Hayakawa. Induction and Stimulation of 92-KDA Gelatinase/Type IVCollagenase Production in Osteosarcoma and Fibrosarcoma Cell Lines by TumorNecrosis Factor-cL. Biochemical and Biophysical Research Communications,171:610-617. 1990.Okamaura, H., A. Takenaka, Y. Yajima, and T. Nishimura. Ovulatory Changes in theWall at the Apex of the Human Graafian Follicle. Journal of Reproduction andFertility, 58:153-155. 1980.Olsterholzer, H.O., J.H. Johnson, and S.V. Nicosia. An Autoradiographic Study ofRabbit Ovarian Surface Epithelium Before and After Ovulation. Biology ofReproduction, 33:729-738. 1 985a.Olsterholzer, H.O., E.J. Streibel, and S.V. Nicosia. Growth Effects of ProteinHormones on Cultured Rabbit Ovarian Surface Epithelial Cells. Biology ofReproduction, 33:247-258. 1985b.Overall, C.M., J.L. Wrana, and J. Sodek. Independent Regulation of Collagenase, 72-kDa Progelatinase, and Metalloendoproteinase Inhibitor Expression in HumanFibroblasts by Transforming Growth Factor-13. Journal of Biological Chemistry,264:1860-1869. 1989.Pannekoek, H., H.V. Veerman, H. Lambers, P. Diergaarde, C.L. Verweij, A.J.Zonneveld, and J.A. van Mourik. Endothelial Plasminogen Activator Inhibitor (PAl):A New Member of the Serpin Gene Family. EMBO Journal, 5:2539-2544. 1986.Papadaki, L. and J.O.W. Beilby. The Fine Structure of the Surface Epithelium of theHuman Ovary. Journal of Cell Science, 8:445-465. 1971.Parmley, T.H. and J.D. Woodruff. The Ovarian Mesothelioma. American Journal ofObstetrics and Gynecology, 120:234-241. 1974.Pertoft, H, K. Rubin, L. Kjellen, T.C. Larent, and B. Klingeborn. The Viability of CellsGrown or Centrifuged in a New Centrifugation Medium, Percoll (TM). ExperimentalCell Research, 110:449-457. 1 977.Pertoft, H. and T.C. Laurent. lsopycnic Separation of Cells and Cell Organelles byCentrifugation in Modified Colloidal Silica Gradients. In: Methods of Cell Separation.Voll. N. Catsimpoolas, ed. Plenum Press, New York, pp. 25-65. 1977.Philips, D.R., I.E. Charo, and R.M. Scarborough. GPlIb-Illa: The Responsive Integrin.Cell, 65:359-362. 1991.Pinkerton, J.H.M., D.G. McKay, E.C. Adams, and A.T. Hertig. Development of theHuman Ovary- A Study Using Histochemical Technics. Obstetrics andGynecology,18:152-181. 1961.134Piquette, G.N. and B.G. Timms. Isolation and Characterization of Rabbit OvarianSurface Epithelium, Granulosa Cells, and Peritoneal Mesothelium in PrimaryCulture. In Vitro Cellular and Developmental Biology, 26:471-481. 1990.Politis, I., A. Srikandakumar, J.D. Turner, B.K. Tsang, L. Ainsworth, and B.R.Downey. Changes in and Partial Identification of the Plasminogen Activator andPlasminogen Activator Inhibitor Systems During Ovarian Follicular Maturation inthe Pig. Biology of Reproduction, 43:636-642. 1990a.Politis, I., L. Wang, J.D. Turner, and B.K. Tsang. Changes in Tissue-TypePlasminogen Activator-Like and Plasminogen Activator Inhibitor Activities inGranulosa and Theca Layers During Ovarian Follicle Development in the DomesticHen. Biology of Reproduction, 42:747-754. 1990b.Pollanen, J., 0. Saksela, E. Salonen, P. Andreasen, L. Nielsen, K. Dano, and A. Vaheri.Distinct Localizations of Urokinase-Type Plasminogen Activator and Its Type 1Inhibitor Under Cultured Human Fibroblasts and Sarcoma Cells. Journal of CellBiology, 104:1085-1096. 1987.Puistola, U., T. Salo, H. Martikainen, and L. Ronnberg. Type IV CollagenolyticActivity in Human Preovulatory Follicular Fluid. Fertility and Sterility, 45:578-580. 1986.Raber, J.M. and S.M. D’Ambrosio. Isolation of Single Cell Suspensions from the RatMammary Gland: Separation, Characterization, and Primary Culture of Various CellPopulations. In Vitro Cellular and Developmental Biology, 22:429-439. 1986.Radisavljevic, S.V. The Pathogenesis of Ovarian Inclusion Cysts and Cystomas.Obstetrics and Gynecology, 49:424-429. 1977.Ramsammy, L.S., C. Josepovitz, and G.J. Kaloyznides. Gentamicin Inhibits AgonistStimulation of the Phosphatidylinositol Cascade in Primary Cultures of RabbitProximal Tubular Cells and in Rat Renal Cortex. Journal of Pharmacology andExperimental Therapeutics, 247:989-996. 1988.Rawson J.M.R. and L.L. Espey. Concentration of Electron Dense Granules in the RabbitOvarian Suface Epithelium During Ovulation. Biology of Reproduction, 17:561-566.1977.Reich, R., R. Muskin, and A. Tsafriri. Follicular Plasminogen Activator: Involvementin Ovulation. Endocrinology, 116:516-521. 1985a.Reich, R., A. Tsafriri, and G.L. Mechanic. The Involvement of Collagenolysis inOvulation in the Rat. Endocrinology, 116:522-527. 1985b.Rembiszewska, A. and A. Brynczak. Steroidogenic Activity in Surface Epithelium ofthe Human Ovary. Foliz Histochemica et Cytobiologica, 23:159-164. 1985.Rheinwald, J.G., J.L. Jorgensen, W.C. Hahn, A.J. Terpstra, T.M. O’Connell, and K.K.Plummer. Mesosecrin: a Secreted Glycoprotein Produced in Abundance by HumanMesothelial, Endothelial, and Kidney Cells in Culture. Journal of Cell Biology,104:263-275. 1987.135Rheinwald, J.G. and T.M. O’Connell. Intermediate Filament Proteins as DistinguishingMarkers of Cell Type and Differentiated State in Cultured Human Urinary TractEpithelia. Annals of the New York Academy of Sciences, 455:259-267. 1985.Richardson, G.S., R.E. Scully, N. Nikrui, and J.H. Nelson. Common Epithelial Cancerof the Ovary. New England Journal of Medicine, 312:415-424. 1985.Roberts, R., J. Gallagher, E. Spooncer, T.D. Allen, F. Bloomfield, and T.M. Dexter.Heparan Sulphate Bound Growth Factors: A Mechanism for Stromal Cell MediatedHaemopoiesis. Nature, 332:376-378. 1988.Rodriguez, G.C., A. Berchuk, R.S. Whitaker, D. Schlossman, D.L. Clarke-Pearson, andR.C. Bast. Epidermal Growth Factor Recptor Expression in Normal OvarianEpithelium and Ovarian Cancer. II. Relationship Between Receptor Expression andResponse to Epidermal Growth Factor. American Journal of Obstetrics andGynecology, 164:745-750. 1991.Roskelley, C.D. and N. Auersperg. Density Separation of Rat Adrenocortical Cells:Morphology, Steroidogenesis, and P-45Oscc Expression in Primary Culture. InVitro Cellular and Developmental Biology, 26:493-501. 1990.Ruoslahti, E. Integrins. Journal of Clinical Investigation, 87:1-5. 1991.Salonen, E.M., A. Vaheri, J. Pollanen, R. Stephens, P. Andreasen, M. Mayer, K. Dano,J. Gailit, and E. Ruoslahti. Interaction of Plasminogen Activator Inhibitor (PAl-i)with Vitronectin. Journal of Biological Chenistry, 264:6339-6343. 1989.Sambuy, Y. and I. DeAngelis. Formation of Organoid Structures and ExtracellularMatrix Production in an Intestinal Epithelial Cell Line During Long-term in VitroCulture. Cell Differentiation, 19:139-147. 1986.Sanders, E.J. and S. Prasad. Invasion of a Basement Membrane Matrix by ChickEmbryo Primitive Streak Cells in Vitro. Journal of Cell Science, 92:497-504.1989.Sauramo, H. Histology, Histopathology, and Function of the Senile Ovary. AnnalesChirurgiae et Gynaecologiae Fenniae, 41:5-65. 1952.Schafer, l.A., A. Shapiro, M. Kovach, C. Lang, and R.B. Frantianne. The Interaction ofHuman Papillary and Reticular Fibroblasts and Human Keratinocytes in theContraction of Three-Dimensional Floating Collagen Lattices. Experimental CellResearch, 183:112-125. 1989.Schiro, J.A., B.M.C. Chan, W.T. Roswit, P.D. Kassner, A.P. Pentland, M.E. Hemler,A.Z. Eisen, and T.S. Kupper. Integrin VLA-2 Mediates Reorganization and Contractionof Collagen Matrices by Human Cells. Cell, 67:403-410. 1991.Scully, R.E. Recent Progress in Ovarian Cancer. Human Pathology, 1:73-98. 1970.Scully, R.E. Ovarian Tumors. American Journal of Pathology, 87:686-720. 1977.Seely, K.A. and J. Aggeler. Modulation of Milk Protein Synthesis Through Alterationof the Cytoskeleton in Mouse Mammary Epithelial Cells Cultured on a ReconstitutedBasement Membrane. Journal of Cellular Physiology, 146:117-130. 1991.136Seiffert, D., J. Mimuro, R.R. Schleef, and D.J. Loskutoff. Interactions Between Type1 Plasminogen Activitor Inhibitor, Extracellular Matrix, and Vitronectin. CellDifferentiation and Development, 32:287-292. 1990.Setratian, S., B.C. Saunders, and S.V. Nicosia. Isolation and Characterization of RabbitPeritoneal Mesothelial Cells. Acta Ctyologica, 34:92-102. 1990.Siemens, C.H. and N. Auersperg. Serial Propagation of Human Ovarian SurfaceEpithelium in Tissue Culture. Journal of Cellular Physiology, 134:347-356. 1988.Starkey, J.R., D.R. Stanford, J.A. Magnuson, S. Hamner, N.P. Roberston, and G.J.Gasic. Comparison of Basement Membrane Matrix Degradation by Purified Proteasesand by Metastatic Tumor Cells. Journal of Cellular Biochemistry, 35:31-49. 1987.Steinert, P.M. and D.A. Parry. Intermediate Filaments: Conformity and Diversity ofExpression and Structure. Annual Reviews in Cell Biology, 1:41-65. 1985.Steinert, P.M., A.C. Steven, and D.R. Roop. The Molecular Biology of IntermediateFilaments. Cell, 42:411-419. 1985.Stevens, L.C. The Origin and Development of Ovarian Teratomas in Mice. UICCTechnical Report Series, 50:74-81. 1980.Stopak, D. and A.K. Harris. Connective Tissue Morphogenesis by Fibroblast Traction:1. Tissue Culture Observations. Developmental Biology, 90:383-398. 1982.Takahashi, K., J. Hata, K. Mukai, and Y. Sawaski. Close Similarity Between CulturedHuman Omental Mesothelial Cells and Endothelial Cells in Cytochemical Markers andPlasminogen Activator Production. In Vitro Cellular and Developmental Biology,27A:542-548. 1991.Tamura, R.N., C. Rozzo, L. Starr, J. Chamgers, L.F. Reichardt, H.M. Cooper, and V.Quaranta. Epithelial Integrin cx6f34: Complete Primary Structure of a and VariantForms of 34. Journal of Cell Biology, 111:1593-1604. 1990.Taub, M., V. Wang, T.M. Szczesny, and H.K. Kleinman. Epidermal Growth Factor orTransforming Growth Factor-a is Required for Kidney Tubulogenesis in MatrigelCultures in Serum-Free Medium. Proceedings of the National Academy of Sciences ofthe United States of America, 87:4002-4006. 1990.Thomas, J.O. and Kornberg, R.D. An Octamer of Histones in Chromatin and FreeSolution. Proceedings of the National Academy of Sciences of the United States ofAmerica, 72:2626-2630. 1 975.Thomas, W.G. and C. Sernia. The Immunocytochemical Localization of Angiotensinogenin the Rat Ovary. Cell Tissue Research, 261 :367-373. 1990.Tilly, J.L. and A.L. Johnson. Presence and Hormonal Control of Plasminogen Activatorin Granulosa Cells of the Domestic Hen. Biology of Reproduction, 37:1156-1164.1987.137Towbin, H., T. Staehelin, and J. Gordon. Electrophoretic Transfer of Proteins fromPolyacrylamide Gels to Nitrocellulose Sheets: Procedure and Some Applications.Proceedings of the National Academy of Sciences of the United States of America,76:4350-4354. 1979.Trelstad, R.L., A. Hayashi, K. Hayashi, and P.K. Donahoe. The EpithelialMesenchymal Interface of the Male Rat Mullerian Duct: Loss of Basement MembraneIntegrity and Ductal Regression. Developmental Biology, 92:27-40. 1982.Uitto, V.-J., H. Larjava, H. Jyrki, and T. Sorsa. A Protease of Bacteroides gingivalisDegrades Cell Surface and Matrix Glycoproteins of Cultured Gingival Fibroblasts andInduces Secretion of Collagenase and Plasminogen Activator. Infection and Immunity,57:213-218. 1989.Unemori, E.N., K.S. Bouhana, and Z. Werb. Vectorial Secretion of ExtracellularMatrix Proteins, Matrix-Degrading Proteinases, and Tissue Inhibitor ofMetalloproteinases by Endothelial Cells. Journal of Biological Chemistry, 265:445-451. 1990.Ura, H., T. Obara, K. Yokota, Y. Shibata, K. Okamura, and M. Namiki. Effects ofTransforming Growth Factor-n Released from Gastric Carcinoma Cells on theContraction of Collagen-Matrix Gels Containing Fibroblasts. Cancer Research,51:3550-3554. 1991.Van Blerkom, J and P.M. Motta. The Ovary Prior to Ovulation. In: The Cellular Basisof Mammalian Reproduction. Urban & Schwarzenberg, Baltimore, pp. 5-51. 1979.Van der Auwera, P. and F. Meunier. In-Vitro Effects of Cilofungin (Ly121019),Amphotericin B and Amphotericin B-Deoxycholate on Human PolymorphonuclearLeucocytes. Journal of Antimicrobial Chemotherapy, 24:747-763. 1989.Van Hinsbergh, V.W.M., E.D. Sprengers, and T. Kooistra. Effect of Thrombin on theProduction of Plasminogen Activators and PA Inhibitor-i by Human ForeskinMicrovascular Endothelial Cells. Thrombosis and Haemostasis, 57:148-153. 1987.Viale, G., M. Gambacorta, P. Dell’Orta, and G. Coggi. Coexpression of Cytokeratins andVimentin in Common Epithelial Tumours of the Ovary: An Immunocytochemical Studyof Eighty-Three Cases. Virchows Archiv A Pathological Anatomy and Histopathology,413:91-101. 1988.Vigier, B., M.G. Forest, B. Eychenne, J. Bezard, 0. Garrigou, P. Robel, and N. Josso.Anti-Mullerian Hormone Produces Endocrine Sex Reversal of Fetal Ovaries.Proceedings of the National Academy of Sciences of the United States of America,86:3684-3688. 1989.Vlodavsky, I., J. Folkman, R. Sullivan, R. Fridman, R. lshai-Michaeli, J. Sasse, andM. Klagsbrun. Endothelial Cell-Derived Basic Fibroblast Growth Factor: Synthesisand Deposition into Subendothelial Extracellular Matrix. Proceedings of the NationalAcademy of Sciences of the United States of America, 84:2292-2296. 1987.Wartenberg, H. Development of the Early Human Ovary and the Role of theMesonephros in the Differentiation of the Cortex. Anatomy and Embryology,165:253-280. 1982.138Weiss, N.S. Epidemiology of Ovarian Cancer. UICC Technical Report Series, 50:34-49. 1980.Werb, Z., P.M. Tremble, 0. Behrendtsen, E. Crowley, and C.H. Damsky. SignalTransduction Through the Fibronectin Receptor Induces Collagenase and StromelysinGene Expression. Journal of Cell Biology, 109:877-889. 1989.Whitaker, D., J.M. Papadimitriou, and M.N.I. Walters. The Mesothelium: Techniquesfor Investigating the Origin, Nature, and Behaviour of Mesothelial Cells. Journal ofPathology, 132:263-271. 1980a.Whitaker, D., J.M. Papadimitriou, and M.N.I. Walters. The Mesothelium: AHistochemical Study of Resting Mesothelial Cells. Journal of Pathology, 132:272-284. 1980b.Whitaker, D., J.M. Papadimitriou, and M.N.I. Walters. The Mesothelium: ACytochemical Study of “Activated” Mesothelial Cells. Journal of Pathology, 136:169-179. 1982.Wicha, M.S., L.A. Liotta, S. Garbisa, and W.R. Kidwell. Basement Membrane CollagenRequirements for Attachment and Growth of Mammary Epithelium. Experimental CellResearch, 124:181-190. 1979.Wicha, M.S., G. Lowrie, E. Kohn, P. Bagavandoss, and T. Mahn. Extracellular MatrixPromotes Mammary Epithelial Growth and Differentiation in Vitro. Proceedings ofthe National Academy of Sciences of the United States of America, 79:3213-3217.1982.Wilson, T.S., E.M. McDowell, and B.F. Trump. An Alcian Blue (pH 2.5)-PAS-KeratinImmunoperoxidase Method for the Simultaneous Demonstration of Keratin andNeutral and Acidic Mucosubstances. Stain Technology, 58:225-229. 1983.Wong, K.K. and A.l. Gotlieb. The Reorganization of Microfilaments, Centrosomes, andMicrotubles During In Vitro Small Wound Reendothelialization. Journal of CellBiology, 107:1777-1783. 1988.Yoshimura, Y., R. Santulli, S.J. Atlas, S. Fijii, and E.E. Wallach. The Effects ofProteolytic Enzymes on In Vitro Ovulation in the Rabbit. American Journal ofObstetrics and Gynecology, 157:468-475. 1987.Young, R.H., P.B. Clement, and R.E. Scully. The Ovary. In: Diagnostic SurgicalPathology. S.S. Sternberg, ed. Raven Press, New York, 49:1655-1734. 1989.Zhu, C. and J.F. Woessner. A Tissue Inhibitor of Metalloproteinases and aMacroglobulins in the Ovulating Rat Ovary: Possible Regulators of Collagen MatrixBreakdown. Biology of Reproduction, 45:334-342. 1991.Zucker, S., J. Wieman, R.M. Lysik, B. Imhof, H. Nagase, N. Ramamurthy, L.A. Liotta,and L.M. Golub. Gelatin-Degrading Type IV Collagenase Isolated from Human SmallCell Lung Cancer. Invasion Metastasis, 9:167-181. 1989.Zuk, A. and E.D. Hay. Changes in Expression of the 131 Integrin Family Occur DuringTransformation of Lens Epithelium to Mesenchyme in 3D Collagen Gels. Journal ofCell Biology, 111:2651. 1990139Zuk, A., K. Matlin, and E.D. Hay. Type I Collagen Gel Induces Madin-Darby CanineKidney Cells to Become Fusiform in Shape and Lose Apical-Basal Polarity. Journal ofCell Biology, 1 08:903-920. 1989.

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0098894/manifest

Comment

Related Items