The Human Myoepithelial Cell Is a Natural Tumor Suppressor1

Myoepithelial cells, which surround ducts and acini of glandular organs, form a natural border separating probiferating epitheliab cells from basement membrane and underlying stroma. Myoepithelial cells in situ and in vitro constitutively express high amounts of proteinase inhibitors that include tissue inhibitor of metalboproteinase 1, protease nexin-Il, ct-i antitrypsin, and maspin. Human myoepithelial xenografts (HMS-X, HMS-3X, and HMS-4X), which our laboratory has established, accumulate an abundant extracellular matrix containing sequestered proteinase inhibitors. Humatrix, a gel that we have derived from HMS-X, inhibits tumor cell invasion (down to 25% ± 10% of Matrigel control; P < 0.01), and our recently established human myoepithelial cell lines, HMS-1, HMS-3, and HMS-4, inhibit tumor cell invasion in cellular invasion (down to 42% ± 7% of control; P < 0.05) and in conditioned media assays (down to 30% ± 8% of control; P < 0.01). The anti-invasive effects of HMS-1, HMS-3, and HMS-4 can be enhanced by phorbol 12-myristate 13-acetate (down to 2% ± 1% of control) by a maspin-dependent mechanism and abolished by dexamethasone (up to 95% ± 5% of control) by a maspin-independent mechanism (P < 0.01). HMS-X, HMS-3X, HMS-4X, and Humatrix inhibit tumor invasion and metastasis in severe combined immunodeficient mice (P < 0.001). The cumulative data suggest that myoepithelial cells are natural para. crine suppressors of invasion and metastasis and may specifically inhibit the progression of precancerous disease states to invasive cancer in vivo. Introduction It has become clear that cancer cells come under the influence of important paracrine regulation from the host microenvironment ( 1 ). Such host regulation may be as great a Received 6/24/97; accepted 7/14/97. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by USPHS Grants CA7I 195, CA40225, and CAO1351, and by the Carolan Foundation. M. D. S. was supported by USPHS National Institutional Research Service Award CA-09056. S. H. B. is the recipient of a National Cancer Institute Research Career Development Award. 2 To whom requests for reprints should be addressed. Phone: (3 10) 8255588: Fax: (310)441-1248. determinant of tumor cell behavior in vito as the specific oncogenic or tumor suppressor alterations occurring within the malignant cells themselves and may be mediated by specific extracellular matrix molecules, matrix-associated growth factors, or host cells themselves (2, 3). Both positive (fibroblast, myofibroblast, and endotheliab cell) and negative (tumor-infibtrating lymphocyte and cytotoxic macrophage) cellular regulators exist that profoundly affect tumor cell behavior in viva (4, 5). One host cell, however, the myoepithelial cell, has escaped the paracrine onlooker’s attention. Ihe myoepithelial cell, which lies on the epithelial side of the basement membrane, is thought to contribute largely to both the synthesis and remodcling of this structure. This cell lies in juxtaposition to normally proliferating and differentiating epitheliab cells in health and to abnormally proliferating and differentiating epithelial cells in precancerous disease states. such as dysplasia and carcinoma in situ. This anatomical relationship suggests that myoepitheliab cells may exert important paracrine effects on normal glandular epithelium and may regulate the progression of carcinoma in situ to invasive carcinoma. Circumstantial evidence suggests that the myoepithelial cell naturally exhibits a tumor suppressor phenotype. Myoepithelial cells rarely transform, and when they do, they generally give rise to benign neoplasms that accumulate rather than degrade extracellular matrix (6). Myoepitheliab cells directly or indirectly through their production of extracebbular matrix are thought to regulate branching morphogenesis that occurs in the developing breast and salivary gland during embryobogicab development (7). The cell biology of the human myoepithelial cell has not been well investigated, and only a few isolated studies exist in the literature despite the near ubiquitous presence of this cell surrounding epithebiab ducts and acini in numerous glandular organs. Because potentially important paracrine relationships might exist between myoepitheliab cells and epithelial cells in both developmental biology and cancer, our laboratory has been very much interested in this cell and has established immortalized myoepithebial cell lines and transplantable xenografts from benign human myoepitheliomas of the salivary gland gland (HMS3-l, HMS-X; HMS-3, HMS-3X) and breast (HMS-4, HMS-4X; Refs. 8 and 9). Ihese cell lines and xenografts cxpressed identical myoepithebial markers as their in situ counterparts and displayed an essentially normal diploid karyotype. Unlike the vast majority of human tumor cell lines and xcnografts that exhibited matrix-degrading properties, these myoepithelial lines/xenografts, bike their myoepithebiab counterparts in situ, retained the ability to secrete and accumulate an abundant extracellular matrix composed of both basement membrane 3 The abbreviations used are: HMS. human matrix secreting: CM, conditioned medium; liMP, tissue inhibitor of metalloproteina.se; PAl, plasminogen activator inhibitor; a 1-AT, a I -antitrypsin; PN-IIIAPP, protease nexin 11./13 amyboid precursor protein; MMP, matrix metalboproteina.se: PMA. phorbol 12-myristate 13-acetate: CHX, cyclohexamide; DCIS, ductal carcinoma in situ: HMEC. human mammary epithelial cell. Research. on November 11, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 1950 Tumor Suppression by Myoepithelial Cells and non-basement membrane components. When grown as a monolayer, the myoepitheliab cell lines exerted profound and specific effects on normal epithelial and primary carcinoma morphogenesis (8). These studies support our position that our established myoepithelial lines/xenografts recapitulate a normal differentiated myoepithebiab phenotype and can, therefore, be used experimentally as a primary myoepithelial surrogate. Prompted by these studies and by the conspicuous absence of studies examining the role of the myoepithelial cell in tumor progression, we decided to examine the myoepithelial cell from this perspective. Experiments with these cell lines/xenografts, together with relevant in situ observations, form the cornerstone of the present study, which observes that the human myoepitheliab cell is a natural tumor suppressor. Materials and Methods Cell Lines and Xenografts. Human myoepithelial cell lines and xenografts including HMS-l, HMS-X; HMS-3, HMS-3X (salivary gland); and HMS-4, HMS-4X (breast) were established recently in our laboratory (8, 9). The murine EHS tumor was grown as an albograft in severe combined immunodeficient mice. Other cell lines used in this study included human melanoma lines C816l (Dr. Mary J. C. Hendrix, University of Arizona Cancer Center, Tucson, AZ), which was transfected by us with pSV2neo in a previous study (3), MlS (Dr. Don Morton, John Wayne Cancer Center, Santa Monica, CA), and A375 (American Type Culture Collection, Rockville, MD); normal human mammary epithelial cells (HMEC; Clonetics, San Diego, CA); HsS78Bst human breast myofibrobbasts ( l0) ; HIG-82 rabbit synoviocytes; human breast carcinoma lines MCF-7, I47D, BI-549, MDA-MB1 57, MDA-MB-23 1, and HsS78I; human squamous cell carcinoma lines of salivary gland and vulva, A253 and A43 I (American Type Culture Collection, Rockville, MD); and a human mucoepidermoid carcinoma cell line (M/E; Dr. Adi Gazdar, University of Texas Southwestern Medical School, Dallas, IX). Serum-free CM was collected from many of these lines over 24 h and concentrated 10-1000-fold using Centriprep-iO concentrators (Amicon, Beverby, MA). Antibodies and Probes. I!MP-l and maspin cDNA probes were kindly provided (Dr. Judith C. Gasson, UCLA Jonsson Comprehensive Cancer Center, Los Angeles, CA; Dr. Ruth Sager, Dana-Farber Cancer Institute, Boston, MA). Five p g of oligo(dI)-selected mRNA were separated on denaturing agarose gels, transferred to nylon membranes, and probed with fa-32PJdCIP-labeled probes using standard Northern blot protocols (8). Primary antibodies used to known proteinase inhibitors included rabbit anti-human maspin (PharMingen, San Diego, CA), rabbit anti-human al-Al (DAKO, Carpinteria, CA); mouse monoclonal anti-human PA! types 1 , 2. and 3 (American Diagnostica, Greenwich, CT); mouse monoclonal anti-human PN-II/APP (Boehringer Mannheim, Indianapolis, !N); rabbit anti-human a2-antipbasmin (Alexis, San Diego. CA); rabbit 4 Although Hs578Bst was represented as a myoepithelial cell line, we found that it possessed a myofibroblast rather than myoepithelial phenotype (unpublished data). anti-rIIMP-l (Dr. Judith C. Gasson, UCLA Jonsson Comprehensive Cancer Center, Los Angeles, CA); and mouse monoclonal anti-human protease nexin I (Dr. Dennis Cunningham, University of California, Irvine, CA). Concentrated CM was resolved on nonreducing 10% SDS-PAGE mini-gels and transferred to nitrocellulose membranes for Western blotting using standard protocols (9); in some instances, cellular bysates instead of CM were analyzed. Standardized immunoprecipitation protocols using Sepharose-protein A ( I 1) were used in the evaluation of maspin activities. Immunocytochemicab Studies. Formalin-fixed paraffinembedded human tissues of normal breast, DC!S, and the human myoepithelial xenografts, HMS-X, HMS-3X and HMS-4X, were incubated with rabbit anti-cow S-100 (1 : 1000; DAKO), smooth muscle actin (1 :2000: DAKO), and the previously mentioned antibodies to maspin (1:500), al-Al (1:2000), PN-!IJ APP ( 1 :50), lIMP1 (1 : 100), and PA!1 ( 1 :20). Peroxidaseconjugated sheep anti-mouse !gG and goat anti-rabbit IgG were used as secondary antibodies at 1 :200 and 1 :25 dilutions, respectively. Coborimetric detection of peroxidase-conjugated secondary antibody was with diaminobenzidine. Zymography and Related Studies. Proteinases and their inhibitors were profiled by standard zymographic and reverse zymographic methods (12-15). To visualize MMPs, renatured gelatinand casein-containing gels were equilibrated for 30 mm in enzyme buffer [SO msi Tris-HC1 (pH 7.6), 0.2 M NaCl, S mM CaCl2, and 0.02% Brij-35] and incubated overnight in fresh buffer at 37#{176}C. To detect MMP inhibitors, serum-free medium containing 25 ng/ml PMA was conditioned over 24 h by HIG-82 cells and activated with 1 msi para-aminophenylmercuric acetate. To visualize inhibitors of trypsin-like 5crinc proteinases, renatured gelatin gels were equilibrated in 10 mM Iris-HCI (pH 7.5) for 30 mm at room temperature and incubated for 4 h at 37#{176}C in fresh buffer containing 0.2-1.0 pg/ml trypsin. Plasminogen activators were detected by adding io sg/mb plasminogen to casein-containing gels. Plasminogen activators and PAI-l were also analyzed by fibrin direct and reverse zymography (16). PA!-2 could not be detected by this method, because unlike PAI-l, it could not be reactivated foblowing SDS denaturation (17). Net antiplasmin activity in the various CM was determined using chromogenic substrate S2251 (H-D-Val-Leu-Lys-p-nitroanaline 2HCI) according to the acid-stopped method described by the supplier (Pharmacia Hepar, Franklin, OH). in Vitro Invasion Assays of Invasion. Invasion experiments were conducted using an invasion chamber containing either Matrigel control (18) (Collaborative Biomedical Products, Bedford, MA) or Humatrix, a gel extracted from HMS-X according to previous methods (8). !n a subsequent set of Matrigel invasion experiments, the invasion set up was slightly modified by us to test the effects of HMS-l, HMS-3, and HMS-4 CM (concentrated up to 100-fold) and HMS cells on invasion. Each well insert was layered with 60 p.1 of a 1 :3 mixture Matrigel: basal keratinocyte serum-free medium (700 ig Matrigellcm2) or Matrigeb:CM and allowed to gel. On top of this cell-free Matrigel layer, a second 60 sb of the 1 :3 Matrigel mixture (700 sg Matrigel/cm2) was added with or without suspended 10#{176} HMS-l or control cells. Highly invasive C8161 or MDA-MB-23l cells (l0 ) were added on top of this second layer. Six hundred p.1 of Research. on November 11, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Clinical Cancer Research 1951 Table 1 Myoepithelial-re lated immunoreactivity in situ HMS-X. H MS-3X, HMS-4X Normal breast DCIS ME Epi Cells Matrix ME” Epi” 5-100 ++++( ++++ ++++ Maspin ++++ ++++ + ++++ -