Int. J . Cancer: 45, 1105-1 112 (1990) 0 1990 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I‘Union lnternationale Contre le Cancer

SELECTIVE IMMORTALIZATION OF A PHENOTYPICALLY DISTINCT EPITHELIAL CELL TYPE BY MICROINJECTION OF SV40 DNA INTO CULTURED HUMAN MILK CELLS Jiri BARTEK’.~, Jirina BARTKOVA’S~, El-Nasir LALANI’,Vitezslav BREZINA~ and Joyce TAYLOR-PAPADIMITRIOU~~~ ‘Imperial Cancer Research Fund, P.O. Box 123, Lincoln’s Inn Fields London WC2A 3PX, UK; 21nstitute of Medical Research, Research Institute of Clinical and Experimental Oncology, Zluty kopec 7 , 656 01 Brno, Czechoslovakia;and 31nstitute of Medical Research, Paediatric Research Institute, Cernopolni 9, 662 62 Brno, Czechoslovakia. An immortal cell line, MMSV- I, has been developed which exhibits many features of the common luminal epithelial cell of the human mammary gland. The cell line was developed by microinjection of SV40 DNA into individual cells in selected colonies in cultures of human milk epithelial cells. Immunohistochemical staining shows that the MMSV-I cells express keratins 7,8,18 and 19 homogeneously in organized filaments which lead into well-developed desmosomes. They do not express vimentin or keratins found in stratified epithelia or keratin 14 found in basal cells in the mammary gland. The PEM mucin, recognized by the antibody HMFG- I, is also expressed and appears to be processed normally. Fibronectinis detected but shows the punctate pattern typical of cultured normal milk epithelial cells. MMSV-I cells show a reduced requirement for added growth factors, including cyclic AMPelevating agents, but do not grow in agar or form tumours in nude mice. Since the transfected cells could be selected on the basis of an extended in vitro life span, antibiotic resistance markers were not introducedand the cells remain sensitive to hygromycin and neomycin.

A prerequisite for studying the process of carcinogenesis in human mammary epithelial cells in vitro is the availability of a system for culturing the appropriate normal cells. Available evidence suggests that the cells which are transformed in vivo, and which eventually become the invasive cells in breast cancer, reside in the terminal ductal lobular units (Wellings et al., 1975; Jenson, 1981; Russo et al., 1987) and belong to the luminal epithelial cell lineage (Taylor-Papadimitriouand Lane, 1987). The phenotype of the major population of luminal cells found in the terminal ductal lobular unit (TDLU) has been characterized, using immunohistochemical markers, by the profile of keratins expressed (7+, 8 + , 18+, 19+, 14-) (Debus et al., 1982; Lane and Klymkovsky, 1982; Ramaekers et al., 1983a; Bartek et al., 1985a; Taylor-Papadimitriou et al., 1989), and by the expression of a polymorphic epithelial mucin (PEM) (Taylor-Papadimitriou et al., 1983; Gendler et al., 1988). This cell phenotype is readily distinguished from the basal cells in the TDLU which express keratins 5 and 14 (and do not express keratins 8, 18 and 19) (Dairkee et al., 1985; Taylor-Papadimitriou et al., 1989) and from a subclass of lumind epithelial cells which do not express keratin 19 (Bartek et al., 1985~).The dominant phenotype of the cells in invasive breast cancer corresponds to that of the common luminal phenotype in the profile of keratins expressed and in the expression of high levels of PEM (Bartek et al., 1985b, 1987; TaylorPapadimitriou and Lane, 1987; Taylor-Papadimitriou et al., 1989). This being so, the normal cell we have attempted to culture is the luminal epithelial cell which expresses keratins 7, 8, 18 and 19 and PEM. Luminal epithelial cells are shed into milk and can be cultured from early lactation or post-weaning milks (Buehring, 1972; Taylor-Papadimitriou et al., 1980) but they have a short in vitro life span. On the other hand, the epithelial cells from reduction mammoplasty tissue which show long-term growth in medium MCDB 170 (Hammond et al., 1984; Stampfer, 1985) derive originally from the basal layer and most of them

do not have the phenotype of the luminal epithelial cells (Taylor-Papadimitriou et al., 1989). We have therefore attempted to immortalize cells of the common luminal phenotype by introducing SV40 DNA into cultured milk cells. In the milk cell cultures all the epithelial colonies express keratins 7, 8 and 18, and 85% also express keratin 19 homogeneously (Bartek et al., 1985~).In spite of this, the cell lines which were developed previously by infection with whole virus (Chang et al., 1982) or by calcium phosphate transfection of SV40 DNA did not express keratin 19 at all or expressed it only in a minor subpopulation of cells at early passages and not at all at later passages (data not shown). This presumably reflects the high proliferative potential of the keratin 19- cells which form large colonies (Bartek et al., 1985~).Since we have previously related the morphology and growth rate of the milk epithelial colonies to their immunohistochemically defined phenotype, we attempted to selectively immortalize the luminal phenotype expressing keratin 19, by micro-injecting SV40 DNA into microscopically identified colonies. Although there is, to our knowledge, no report of immortalizing or transforming human epithelial cell by this method, its feasibility was suggested by the successful immortalization of rabbit mammary epithelial cells with SV40 DNA (Garcia et al., 1986). Here we report the selective establishment and characterization of a human mammary epithelial cell line with phenotypic features similar to those of the major luminal epithelial cell population. Although the cell line appears to be immortal, it does not grow in soft agar or in the nude mouse, and can therefore serve as a recipient for other oncogenes, and for in vitro studies of human mammary carcinogenesis. MATERZAL AND METHODS

Culture of epithelial cells from humn milk and microinjectionprocedure Cells from pooled samples of early lactation milks were washed 3 times in serum-free RPMI 1640 medium and plated on 60-mm tissue culture dishes (Nunc, Roskilde, Denmark) in RPMI 1640 supplemented with 10% foetal calf serum (FCS) (Gibco, Paisley, Scotland), 10% human serum (Flow, Irvine, Scotland), 10 pg/ml insulin (Sigma, St Louis, MO), 5 p,g/ml hydrocortisone (Sigma), 50 ng/ml cholera toxin (Schwarz/ Mann, Spring Valley, NY) and antibiotics (Taylor-Papadimitriou et al., 1980). This medium is referred to as milk mix or MX. The cultures were grown at 37°C in a humidified atmosphere of 10% COz in air and the medium was changed every 3-4 days. After 5-7 days, small colonies of epithelial cells were observed surrounded by spread macrophages.

4T0whom reprint requests should be addressed. Received: November 10, 1989 and in revised form January 31, 1990.

1106

BARTEK ET AL.

Phenotypic analysis of various colony types by immunofluorescence staining of primary milk cultures was performed on cells fixed after 12-25 days of growth in vitro. For the microinjection experiment, 4 dishes from a 12-day culture with epithelial colonies of about 102-103 cells were used. SV40 DNA purchased from BRL (Gaithersburg, MD) was diluted to a concentration of 0.5 mg/ml in sterile tissue-culture-grade water. Microinjection into nuclei of cuboidal epithelial cells in several colonies showing a slower growth rate (colonies of about 1-3 X 10' cells) was performed under a Zeiss IM 35 inverted phase-contrast microscope equipped with a Zeiss micromanipulator and the automatic injector (Inject Matic, Gabay Electronique, Geneva). SV40 DNA solution or sterile water was injected into about 200 cell nuclei per dish in 2 experimental dishes and 2 control dishes, respectively. Monoclonal antibodies A wide range of mouse monoclonal antibodies (MAbs) against various cytokeratins, vimentin, several mucin epitopes, desmoplakin, fibronectin, HLA antigens, SV40 large T antigen and p53 oncoprotein was used to characterize phenotypes of primary milk cultures and the MMSV-1 cell line and its subclones. All antibodies employed, their corresponding target molecules and appropriate references are listed in Table I. Immunojluorescence Primary cultures of milk cells on plastic dishes or cultures of MMSV-1 cells on either plastic or glass coverslips were rinsed twice with phosphate-buffered saline (PBS) at room temperature, fixed in a pre-cooled (-20°C) mixture of methanol and TABLE I - MONOCLONAL ANTIBODIES USED

Target antigen

Antibody

HMFG-1 HMFG-2 SM-3 BA16 BA17 C04 DA7 LLOO 1 IC7 K,13 RKSE 60 M20 C-15 RCK 105 C-18 6B10 DP 15 v9 FN-3 W6l32

PEM Keratin 19 Keratin 18 Keratin 14 Keratin 13 Keratin 10 Keratin 8 Keratin 7 Keratin 4 Desmoplakin Vimentin Fibronectin HLA class I

Reference

Burchell et al., 1983 Burchell et al., 1987 Bartek et al., 1985a Bartek et al., 1985a Bartek et al., 1989 Lauerova et al., 1988 Taylor-Papadimitriou etal., 1989 Van Muijen et al., 1986 Moll et al., 1987 Ramaekers et al., 1983 Van Muijen et al., 1987 Bartek et al., 1987 Ramaekers et al., 1987 Bartek et al., 1987 Van Muijen et al., 1986 Cowin et al., 1985 Osborn et al., 1984 Keen et al., 1984 Brodsky and Parham, 1982 Adams et al., 1983 Harlow et al., 1981 Harlow et al., 1981 Harlow et al., 1981

acetone (1:l by volume) and air-dried. The indirect immunofluorescence method consisted of a primary incubation with MAb for 1 hr at room temperature, followed by washing with PBS and incubation for 40 min with either fluorescein isothiocyanate (F1TC)-conjugated (Dako, Copenhagen, Denmark) or Texas red-conjugated (Dianova, Hamburg, FRG) antisera to mouse immunoglobulins. After a final washing in PBS, the stained cells were mounted in Gelvatol and examined with a Zeiss photomicroscope III. Gel electrophoresis and immunoblots The total cellular protein lysates were prepared by directly harvesting and resuspending confluent cells in hot Laemmli electrophoresis sample buffer. Solubilized proteins were then separated by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulphate (SDS-PAGE) on a 10% gel. Prestained molecular weight markers (BRL) were run in parallel. The transfer of separated proteins onto nitrocellulose membrane and visualization of blotted proteins by an indirect immunoperoxidase procedure were performed essentially as described by Bartek et al. (1985a). Doubling time and drug sensitivity estimation For construction of growth curves, freshly trypsinized and washed cells were plated in 24-well plates at 1 X 104 per well in 1 ml of DMEM with 10% FCS, 10 pg/ml insulin and 5 pg/ml hydrocortisone and re-fed every 2 days. Cells from 3 wells were individually harvested by trypsinization and counted in a haemocytometer every second day until day 12 of the culture period. Doubling time was estimated from the logarithmic phase of the growth curve. Similar 24-well-plate cultures were used to test the sensitivity of the MMSV-1 cells to hygromycin (Sigma) and G418 (Geneticin, Gibco BRL) at 100,200,400 and 800 Fg per ml in the above growth medium. Viability and cell numbers in triplicate cultures were established every 2 days by phase-contrast examination, Trypan blue exclusion and counting of live cells in a haemocytometer. Soft-agar growth assay Anchorage dependence of growth was tested by seeding a single-cell suspension of 5 X 104 cells in a 0.3% agar layer on top of a 0.5% agar base in 60-mm Nunc Petri dishes. Both agar layers contained DMEM medium with 10%FCS and antibiotics. After 4-5 days of culture at 37°C in a humidified atmosphere of 10% CO, in air, the agar cultures were overlaid with fresh DMEM with 10%FCS, the overlays being changed every 3 4 days until termination of the experiment (usually after 4 weeks in culture). Colonies larger than 20 cells were scored in duplicate dishes and compared with parallel cultures of the breast cancer cell line T47D used as a positive control. For control of viability and growth potential in this simple medium, 2 dishes were also seeded with 5 X lo4 cells to be analyzed and were cultured on plastic in parallel with the agar cultures. Tumorigenicity in athymic nude mice Tumorigenicity of the MMSV-1 cell line was tested by in. athymic nude mice that were less jecting 1 X lo7 cells s . ~into than 60 days old. Five mice injected at 2 sites were used in each experiment and the animals were checked at weekly intervals for a period of at least 5 months.

Karyology Metaphase chromosome spreads were prepared from subconfluent cultures treated with 0.1 pg/ml colcemid for 4 hr at 37"C, harvested by trypsinization and pelleted by low-speed P53 centrifugation. The cell pellet was resuspended in 0.075 M KC1 Antibodies were kindly donated by Drs. W. Bodmer (W6/32, TALIBS), D.Lane for 10 min and fixed in methano1:acetic acid (3:l). Slides with (PAb419,PAb421,PAb423), W.Franke@P 15,K, 13),B.Lane(LL001,LL002), F. Ramaekers (RCK 105, RKSE 60). G.N.P. Van Muijen (MZO, 6B10, 1C7). The metaphase spreads were then stained with conventional Giemsa solution and chromosome numbers were counted in 200 cells. anti-vimentin antibody V9 was purchased from Boehringer, Mannheim, FRG. TALlB5 PAb 419 PAb 423 PAb 421

HLA-DR SV40 T antigen

HUMAN MAMMARY EPITHELIAL CELL LINE

RESULTS

Development of the cell line MMSV-I Epithelial cells from human milk were cultured in a medium (Milk Mix, MX) containing human serum, a cyclic AMPelevating agent, insulin and hydrocortisone (Taylor-Papadimitriou et al., 1980; see “Methods”). In this culture system the milk macrophages act as feeders, and epithelial colonies appear after 5-7 days in culture, when they displace the macrophages. Although large colonies of epithelial cells develop, their growth is limited to only a few divisions after passage so that a direct selection for cells with an extended life span can be made after introduction of an oncogene. We have previously shown that the keratin 19- cells form larger colonies than keratin 19+ cells (Bartek et al., 1985a) which form colonies of about 100-300 cuboidal cells after 10-14 days in culture. We therefore injected SV40 DNA into 400 cells showing the cuboidal morphology in colonies containing 100-300 cells 12 days after seeding. A typical colony chosen for microinjection is illustrated in Figure la. Cells from plates containing colonies injected with SV40 DNA, and from control plates, were passaged. After 2 passages, cells from the control plates senesced, whereas growth was still apparent in one of the 2 plates containing the cells injected with SV40 DNA. The phenotype was assessed by immunohistochemical staining at passage 3 (Table 11) and the cells were cloned by limiting dilution at passage 4. At around passage 9, the cells appeared to stop growing but after 3 weeks of “crisis”, focal growth resumed. The cells have now been in culture for 20 months (over 200 population doublings) and appear to be immortal. The appearance of the cells before and after the crisis event is illustrated in Figure l b and lc. It can be seen that the cells show a more homogeneous morphology in the later passages, suggesting that there may have been selection for a particular phenotype during the crisis period. Phenotype of MMSV-1 cells Using immunohistochemistry and Western blotting, the MMSV-1 cells were examined for their expression of epithelium-specific antigens. The results of this staining are summarized in Table I1 which also describes the profile of antigen expression found in the luminal epithelial cells in vivo and in cultured milk cells. In the profile of keratins expressed, the cells in the MMSV-1 line resemble the dominant luminal phenotype seen in the terminal ductal lobular units; they express keratins 7, 8, 18 and 19 homogeneously and do not express keratin 14 which is found in basal cells and also in the luminal cells in the large ducts (Taylor-Papadimitriou et al., 1989). Moreover, vimentin which can be expressed by some epithelial cells in culture is completely absent from MMSV-1 cells as are keratins 4, 13 and keratin 10 (markers of non-keratinizing and keratinizing squamous epithelia, respectively). Figure 2 shows the detection of keratin 19 in milk cells and in MMSV-1 cells both by Western blot analysis (Fig. 2a) and by indirect immunofluorescence staining (Fig. 2b). Figure 3a illustrates the uniformly positive staining of well-defined intermediate filaments with an antibody to keratin 18, and similar staining patterns were seen with antibodies to keratins 7 and 8 (not shown). The intermediate filaments lead into well-developed desmosomes which show positive staining with an antibody to desmoplakin (Fig. 3b). Another point of interest is that the pre-crisis cultures expressed keratin 14 heterogeneously, as did the primary cultures of milk epithelial cells. The cells emerging after crisis, however, show no expression of keratin 14. The expression of an epithelial mucin was detected using 3 antibodies (HMFG-1, HMFG-2 and SM-3) which react with defined epitopes on the polymorphic epithelial mucin (PEM)

1107

(Gendler et al., 1988; Taylor-Papadimitriou and Gendler, 1989). This mucin is expressed in high amounts by the luminal cells in the lactating gland and by breast cancers, and in smaller quantities by the normal resting gland. The mucin is aberrantly glycosylated in carcinomas and one of the antibodies, SM-3, reacts with a core protein epitope which is selectively exposed in breast cancers and masked in the mucin produced by the normal gland (Burchell et al., 1987). The MMSV-1 cells showed no reaction with this antibody. The epitopes recognized by the other 2 antibodies, HMFG-1 and HMFG-2 (Taylor-Papadimitriou et al., 1981b; Burchell et al., 1983) are expressed by the normally processed mucin (Burchell et al., 1987), and 80% of cells stained positively and somewhat variably with these antibodies, the stronger reaction being seen with HMFG-1 (see Fig. 3c). Although fibronectin is expressed by cultured milk epithelial cells, it shows a punctate surface staining pattern (TaylorPapadimitriou et al., 1 9 8 1 ~ )which was maintained in the MMSV-1 cells. While HLA class-I antigens were expressed, class-I1 antigens were not detectable. To test the stability of the phenotype defined by immunohistochemical staining, 7 clones were isolated at passage 30 and examined individually with a panel of MAbs. A similar profile of reactivity was observed in all the clones. The morphological appearance of one of these clones is shown in Figure Id. From the results of immunohistochemical staining, we conclude that in developing the cell line MMSV- 1, we have immortalized a cell with a phenotype corresponding to the dominant luminal epithelial cell and that many features of this phenotype have been retained in the immortalized line. Expression of SV40 T antigen Although the complete SV40 DNA was injected into the milk epithelial cells, their immortalization is almost certainly due to the action of expressed SV40 T antigen. The expression of T antigen was monitored using antibodies specifically reactive with the N terminal and the C terminal end of the molecule (Mole et al., 1987) both in immunofluorescence and in Western blots. Figure 4a illustrates the nuclear staining seen in immunofluorescence and Figure 2 shows the detection of T antigen in a Western blot. From the fact that the T antigen expressed in MMSV-1 showed the same electrophoretic mobility as the purified wild-type SV40 T antigen (Fig, 2a) and from positive staining by antibodies to either N-terminus or C-terminus of the molecule, it seems likely that the complete SV40 T antigen protein is synthesized by the MMSV-1 cells. All cells showed positive nuclear staining with the antibodies at all stages, including the crisis period, and this continues to be true in cells passaged for over a year. Thus, it seems likely that T antigen expression is necessary for the extended proliferation of the MMSV-1 cells. Since p53 is elevated in SV40transformed cells, MMSV-1 cells were stained using indirect immunofluorescence with an antibody to this antigen. Figure 4b shows the strong nuclear fluorescence seen after staining. Growth properties of MMSV-I cells Unlike the primary milk epithelial cells, the MMSV-1 cell line grows well in the absence of human serum and cholera toxin. The cells have been grown routinely in DMEM containing 10% FCS, insulin and hydrocortisone with a doubling time of 50 hr. The hormones are not essential for growth but a higher proliferation rate is observed in their presence. Although apparently immortal, the MMSV-1 cells did not form colonies in agar, and did not induce tumours in the nude mouse when tested either at passage 7 (pre-crisis) or at later passages (20 passages post-crisis).

1108

BARTEK ET AL.

FIGURE1 - Phase-contrast micrographs illustrating morphology of a cuboidal type colony in primary milk cell culture on day 20 (a), heterogeneous population of pre-crisis MMSV-1 cells at passage 3 (b), post-crisis MMSV-1 cell line at passage 30 ( c ) and a colony of one of the MMSV-1 subclones (d). Scale bars: 40 Fm.

1109

HUMAN MAMMARY EPITHELIAL CELL LINE

TABLE n - PHENOTYPE OF MMSV-1 CELLS Staining patterns2 with antibodies to: Cells examined

Keratin

7

Lumind epithelial' cells in vivo

Cultured milk epithelial cells MMSV-1 cells P. 3 (before crisis) MMSV-1 cells p. 29 (after crisis)

desmo- vimenplakin tin

18

19

+ + +

+/-4

-

+/-4

+

-

+ + +

+/-4

+/-

+/-

+

-

+ + + +

+/-

+/-4

+

-

+ + + +

-

+

-

8

14

+

/

k

4

ncr

IJ,

1

2

3 4 5

6 7 8 9

'Refers to luminal cells in terminal ductal lobular units. Lurninal cells in large ducts show some expression of keratin 14.-2Forantibodies used see "Material and Methods"; (+) homogeneous positivity, ( + / - ) heterogeneous staining, ( - ) negati~e.-~Thepolymorphic epithelial mu& (PEM) was detected with antibodies HMFG-1 and HMFG-2.-4More than 80% of the cells showed a positive reaction.

Since the line was selected only on the basis of extended proliferation in vitro, it was not necessary to introduce antibiotic resistance markers. MMSV-1 cells were sensitive to both hygromycin (killed by 100 @ml after 7 days) and G418 (killed by 400 pg/ml after 10-12 days) which made them potentially useful recipients for other genes, since transfectants expressing the genes can be selected by using one of the above antibiotics. Chromosome number in MMSV-1 cells Metaphase spreads of colchicine-treated MMSV- 1 cells at passage 32 were examined and the number of chromosomes counted in a total of 200 cells. Figure 5 shows that, although a diploid peak was observed, there was considerable variation in the number of chromosomes per cell and an aneuploid peak was also observed along with some hypodiploid cells. It may be possible to selectively clone cells expressing a diploid karyotype and such experiments are underway. DISCUSSION

The mammary gland is a complex tissue, and even within the epithelial compartment there is a considerable degree of complexity. Although it is not clear whether the basal and luminal cells represent 2 distinct lineages, or derive in the adult from a common stem cell, the mature phenotypes are very different. Moreover, when analyzed by immunohistochemical procedures using a panel of MAbs, subcompartments can be identified within the 2 major phenotypes, and these relate to the position in the mammary tree (Taylor-Papadimitriou and Lane, 1987). Perhaps surprisingly, the phenotype of the invasive breast cancer cells, when defined with the same immunohistochemical reagents, shows a remarkable consistency and resembles that of the common luminal epithelial cell found in the TDLU. This is true for primary breast cancers, metastatic lesions and most breast cancer cell lines which have been in culture for many years. Therefore, the malignant features of the cell are expressed on the background of a specific cell type at a definite stage of differentiation. For this reason, we have attempted to culture and immortalize normal cells with the phenotype of the common luminal epithelial cell in order to

FIGURE 2 - Expression of keratin 19 in MMSV-1 cells. (a) Immunoblot analysis of SV40 large T antigen and keratin 19 polypeptide in the whole-cell lysates of primary milk cells (lanes 3,4,5) and MMSV1 cells at passage 26 (lanes 6 , 7 , 8 , 9 ) . Purified SV40 large-T antigen (lanes 1, 2) was run in parallel and nitrocellulose strips with blotted proteins separated on a 10% SDS PAGE were stained with PAb 423 MAb against the C terminus of the large T antigen (lanes 1,3,6), PAb 419 recognizing the N terminus of the large T antigen (lanes 4, 7), BA17 against the human 40-kDa keratin 19 (lanes 2, 5, 8) and a negative control antibody to pig transferrin (lane 9). (b) Indirect immunofluorescence staining of post-crisis MMSV- 1 cells (p.22) with antibody BA17. Note one cell in mitosis with fragmented filaments. Scale bar: 20 pm.

study the effect of carcinogens and oncogenes on this specific phenotype. The MMSV-1 cell line described here was obtained by microinjection of SV40 DNA into selected colonies in cultures of human milk. To our knowledge, this is the first human epithelial cell line developed by microinjection, and the first non-tumorigenic human mammary cell line with the specific phenotype of the common luminal epithelial cell found in the TDLU. Our results demonstrate the advantage of using the technique of microinjection to selectively introduce a target gene into a subpopulation of cells in cultures of mixed phenotype. Since milk epithelial cells go through a limited number of cell divisions in culture, it was possible in this case to select transfectants merely by passaging. Where this is not possible, a plasmid carrying a gene confemng resistance to neomycin or hygromycin can be co-injected. This technique could be readily applied to introducing DNA into epithelial cell cultures contaminated with fibroblasts which can be distinguished morphologically. When a subpopulation cannot be distinguished on morphological grounds, immunohistochemical staining of a surface antigen can be used to indicate which cells should be micro-injected.

1110

BARTEK ET AL.

FIGURE4 - Detection of the SV40 large T antigen (a) and the p53 oncoprotein (b) in nuclei of the post-crisis MIviSV-1 cell colonies by the antibodies PAb 423 and PAb 421, respectively. Scale bars: 20 pm.

FIGURE3 - Indirect immunofluorescence staining of post-crisis MMSV-I cells shows a typical keratin intermediate filament network decorated by the antibody C04 specific for keratin 18 (a); desmosomes as visualized by the anti-desmoplakinantibody DP15 (b); and heterogeneous expression of a polymorphic epithelial much recognized by the antibody HMFG-1 (c). Scale bars: 20 pm.

In defining the phenotype of the MMSV-1 cells, we have relied on the use of immunohistochemical markers to relate it to that of the luminal cell in vivo. The expression of the simple epithelial keratins is maintained in culture and therefore monospecific antibodies to these proteins are extremely useful reagents in distinguishing cell types in culture (Taylor-Papadimitriou et al., 1989). We were particularly interested in obtaining a cell line expressing keratin 19, since this appears to be

I Ill 20

30

40

60 70 60 CHROMOSOME NUMBER

50

90

I

I 100

110

FIGURE5 - Histogram illustrating the distribution of chromosome number of MMSV-1 cells at passage 32 (200 metaphases were counted).

a distinctive and consistent marker for invasive breast cancer cells which therefore correspond to the 19+ luminal cell (Bartek et d . , 198%). Earlier attempts to obtain cell lines from milk cultures by SV40 virus infection (Chang et al., 1982) or transfection of SV40 DNA into mass cultures (data not shown) have resulted in the development of keratin 19- cell lines, suggesting that the keratin 19- cells, which proliferate well, were preferentially immortalized. The selection of suitably

1111

HUMAN MAMMARY EPITHELIAL CELL LINE

small colonies in the early milk cultures for microinjection, however, proved to be an effective method for immortalizing a cell expressing keratin 19. In addition to expressing keratins 7, 8, 18 and 19 in well-defined filaments, the cell line MMSV-1 also resembles the luminal epithelial cells in the TDLU in its lack of expression of keratin 14 and of the other keratins expressed by stratifying epithelia. The polymorphic epithelial mucin expressed by the luminal epithelial cells is extremely immunogenic and many antibodies are available which react with this component, which has been called PAS-0 (Shimizu and Yamauchi, 1982), NPGP (Ceriani et al., 1983), Mam-6 (Hilkens et al., 1984) and EMA (Ormerod et al., 1983). High levels of the mucin are expressed by breast cancers, but the carbohydrate side-chains are shorter (Bardales et al., 1989) than those found in the normal mucin produced in abundance at lactation (Hanisch etal., 1989). This means that some epitopes (such as that seen by the HMFG-1 antibody) are preferentially expressed on the normal mucin and others (such as that seen by antibody SM-3) on the cancerassociated mucin. The mucin produced by the cell line MMSV1 shows a strong reaction with antibody HMFG-1 and does not react with antibody SM-3. The cells therefore exhibit a normal function of the luminal cell. With the immortalization event, the cell giving rise to the MMSV-1 cell line appears to have lost its dependence on several factors which were found to be important for the growth of primary milk cells. A cyclic AMP-elevating agent and human serum are not necessary for growth and, although the cell cycle time is longer in the absence of hydrocortisone and insulin, the cells can grow in the absence of these hormones. However, although the cells show a lower requirement for added growth factors, they do not show anchorage-independent growth, neither do they form tumours in the nude mouse. It will be of interest to attempt to analyze the loss of dependence on an exogenous inducer of cyclic AMP, since this pathway to mitogenesis appears to be important in a variety of normal epithelial cells. The ability of SV40 virus to immortalize or transform cells has long been recognized and appears to be mediated through the SV40 T antigen. The MMSV-1 cells uniformly express T antigen and it is likely that they are dependent on expression of

this protein for continued growth. In this feature, the MMSV-1 cells differ from the rabbit mammary epithelial cell line derived by microinjection of SV40 DNA since these cells lost expression of T antigen with passage (Garcia et al., 1986). That T antigen is sufficient to immortalize milk epithelial cells is evident from some recent experiments in which we have obtained immortal cell lines by infection with a retrovirus vector containing as insert only the sequences coding for T antigen (data not shown). It has been suggested that T antigen may act on cell proliferation by binding to the p53 protein (Lane and Crawford, 1979), or to the retinoblastoma gene product (De Capri0 et al., 1988) which are thought to act as negative regulators of cell growth. As has been noted in other transformed cells (Crawford et al., 1981) the level of p53 was increased in MMSV-1, presumably due to stabilization of the protein by binding to T antigen. One disadvantage of using SV40 to immortalize cells is that the chromosomes can be modified, resulting in aneuploidy, possibly due to the helicase activity of the T antigen. Analysis of Southern blots of genomic DNA from MMSV-1 cells cut with restriction enzymes indicates that the SV40 DNA is present at a single integration site (data not shown). Moreover, the cell line shows a major diploid peak, even though cells are present with lower and higher numbers of chromosomes. We are at present attempting to clone out the cells showing the diploid karyotype, and also to use the technique described here to develop other cell lines using a truncated SV40 T antigen without helicase activity. In spite of the mixed karyotype, the MMSV-1 cell line exhibits many features of the luminal epithelial cell from the human mammary gland and as such should be appropriate for use in studies investigating the effect of carcinogens and oncogenes on this cell type. ACKNOWLEDGEMENTS

The authors thank Dr. P. Riddle for his instruction in the technique of microinjection. They are also grateful to Dr. D. Lane for supplying purified SV40 T antigen protein, to Dr. J. Kovarik and Dr. Z. Staskova (RICEO, Bmo) for their help in karyotype analysis, to Dr. S . Chang for valuable discussions, to Dr. S . Watt and Dr. F. Watt for reading the manuscript, and to Ms. L. Eaton for preparing it.

REFERENCES ADAMS,T.E., BODMER, J.G. and BODMER, W.F., Production and char- clonal antibodies detecting molecular subunits and combinatorial determiacterization of monoclonal antibodies recognizing the a-chain subunits of nants. J. Immunol., 128, 129-135 (1982). human Ia alloantigens. Immunology, 50, 613-620 (1983). BUEHNNG,G.C., Culture of human mammary epithelial cells: keeping BARDALES, R., BHAVANANDAN, V.P., WISEMAN,G. and BRAMWELL, abreast of a new method. J. nut. Cancer Insf., 49, 1433-1434 (1972). M.E., Purification of the epitectin from human laryngeal carcinoma cells. BURCHELL, J., Complexity J., DURBIN,H. and TAYLOR-PAPADIMITNOW, J. biol. Chem., 264, 1980-1987 (1989). of expression of antigenic determinants recognised by monoclonal antiR.C. and TAYLOR-PAPA-bodies HMFG-1 and HMFG-2, in normal and malignant human mammary BARTEK,J., DURBAN,E.M., HALLOWES, epithelial cells. J. Immunol., 131, 508-513 (1983). DIMITNOU, J., A subclass of luminal epithelial cells in the human mamBURCHELL, J., GENDLER, S., TAYLOR-PAPADIMITRIOU, J., GIRLING,A,, mary gland, defined by antibodies to cytokeratins. J . Cell Sci., 75, 17-33 LEWIS,A., MILLIS,R. and LAMPORT, D., Development and characteriza(1985~). tion of breast cancer reactive monoclonal antibodies directed to the core BARTEK,J., KOVARIK, J., BURCHELL, J., TAYLOR-PAPADIMITNOU, J., protein of the human milk mucin. Cancer Res., 47, 5476-5482 (1987). BARTKOVA, J., VOJTESEK, B., REJTHAR, A., SCHNEIDER, J., PETREK, M., J.A., LEE, J.Y.,MONCADA, R. and BLANK, STASKOVA, Z. and MILLIS,R., Monoclonal antibodies to breast epithelial CENANI,R.L., PETERSON, antigens in the study of differentiation and malignancy. In: K. Lapis and E.W., Characterisation of cell surface antigens of human mammary epiS. Eckhardt (eds.), Molecular biology and diSferentiafion of cancer cells, thelial cells with monoclonal antibodies prepared against human milk fat Vol. 2, pp. 123-130, Karger, Basel, and Akademiai Kiado, Budapest globule. Somat. Cell Genet., 9, 415427 (1983). CHANG,S.E., KEEN,J., LANE,E.B. and TAYLOR-PAPADIMITNOU, (1987). J., and characterizationof SV40-transformedhuman breast epBARTEK, J., KOVARIK, J., VOJTESEK, B., BARTKOVA, J., STASKOVA, Z., Establishment ithelial cell lines. Cancer Res., 42, 2040-2053 (1982). REJTHAR, A. and LAUEROVA, L., Subclassificationof human tumous by W. W., The complement of monoclonal antibodies to keratins. In: G.I.Abelev (ed.), Monoclonal an- COWIN, P., KAPPRELL,H.-P. and FRANKE, tibodies fo tumour associated anfigens and their clinical application, Aka- desmosomal plaque proteins in different cell types. J . Cell Biol., 101, 1442-1454 (1985). demiai Kiado, Budapest (1989). (In press). CRAWFORD, L.V., PIM, D.C., GURNEY, E.G., GOODFELLOW, P. and TAYJ., MILLER,N. and MILLIS,R., BARTEK,J., TAYLOR-PAPADIMITNOW, LOR-PAPADIMITRIOW, J., Detection of a common feature in several human Patterns of expression of keratin 19 as detected with monoclonal antibodies tumor cell lines-a 53,000 dalton protein. Proc. naf.Acad. Sci. (Wash.), in human breast tissues and tumours. Int. J . Cancer, 36,299-306 (1985b). 78, 41-45 (1981). C., SMITH,H.S. and HACKETT, A.J., MonoBRODSKY, F.M. and PARHAM, P., Monomorphic anti-HLA-A,B,C, mono- DAIRKEE,S.H., BLAYNEY,

1112

BARTEK ET AL.

clonal antibody that defines human myoepithelium. Proc. nat. Acad. Sci. Epithelial membrane antigen: partial purification assay and properties. (Wash.), 82, 7409-7413 (1985). Brit. J . Cancer, 48, 533-541 (1983). DEBUS,E., WEBER,K. and OSBORN, M., Monoclonal cytokeratin anti- OSBORN, M., DEBUS,E. and WEBER,K., Monoclonal antibodies specific bodies that distinguish simple from stratified squamous epithelia: charac- for vimentin. Europ. J . Cell Biot.,34, 137-143 (1984). terization of human tissues. EMBO J., 1, 1641-1647 (1982). G., MOESKER, 0. and RAMAEKERS, F.C.S., HUYSMANS, A., SCHAART, DECAPRIO,J.A., LUDLOW,J.W., FIGGE,J., SHAW,J.-Y., HUANG, VOOIJS,P., Tissue distribution of keratin 7 as monitored by a monoclonal C.-M., LEE,W.-H., MARSILIO, E., PAUCHA, E. and LIVINGSTON, D.M., antibody. Exp. Cell Res., 170, 235-249 (1987). SV40 large tumor antigen forms a specific complex with the product of the RAMAEKERS, O., KANT,A,, HUYSMANS, F.C.S., PUTS,J. J.G., MOESKER, retinoblastoma susceptibility gene. Cell, 54, 275-283 (1988). A . , HAAG,D., JAP,P.H.K., HERMAN, C.J. and VOOIJS,G.P., Antibodies GARCIA, I., SORDAT,B., RAUCCIO-FARINON, E., DUNAND, M., KRAE- to intermediate filament proteins in the immunohistochemical identificaHENBWHL, J.-P. and DIGGELMANN, H., Establishment of two rabbit mam- tion of human tumours: an overview. Hisrochem. J . , 15,691-713 (1983). mary epithelial cell lines with distinct oncogenic potential and differenti- Russo, J., CALAF,G.,ROI, L. and Russo, I.H., Influence of age and ated phenotype after microinjection of transforming genes. Mol. cell. gland topography on cell kinetics of normal human breast tissue. J. nat. Biol., 6, 1974-1982 (1986). Cancer Inst., 78, 413-418 (1987). S., TAYLOR-PAPADIMITMOW, J., DUHIG,T., ROTHBARD, J. and SHIMIZW, GENDLER, M. and YAMAUCHI, K., Isolation and characterizationof mucinBWRCHELL, J., A highly immunogenic region of a human polymorphic like glycoproteinsin human milk fat globule membranes. J . Biochem., 91, epithelial mucin expressed by carcinomas is made up of tandem repeats. J. 515-519 (1982). biol. Chem., 263, 1282G12823 (1988). STAMPFER, M.R., Isolation and growth of human mammary epithelial HAMMOND, S.L., HAM,R.G. and STAMPFER, M.R., Serum-free growth of cells. J. Tiss. Culr. Meth., 9, 107-116 (1985). human mammary epithelial cells: rapid clonal growth in defined medium J., BURCHELL, J. and HURST,J., Production of and extended serial passage with pituitary extract. Proc. nat. Acad. Sci. TAYLOR-PAPADIMITMOU, fibronectin by normal and malignant human mammary epithelial cells. (Wash.), 81, 5435-5439 (1984). Cancer Res., 41, 2491-2500 (1981a). J., EGGE,H., HANISCH,F.-G., UHLENBRUCK, G., PETER-KATALINIC, J. and GENDLER, S.J., Molecular aspects of muDABROWSKI, J. and DABROWSKI, U., Structures of neutral 0-linked poly- TAYLOR-PAPADIMITRIOU, lactosaminoglycans on human skim milk mucins. J. biol. Chem., 264, cins. Cancer Rev., 11-12, 11-24 (1989). 872-883 (1989). J. and LANE,E.B., Keratin expression in the TAYLOR-PAPADIMITRIOU, HARLOW,E., CRAWFORD,L.V., RM, D.C. and WILLIAMSON, N.M., mammary gland. In: M.C. Neville and C.W. Daniel (eds.), The mammary Monoclonal antibodies specific for simian virus 40 tumour antigens. J. gland, pp. 181-215, Plenum, New York (1987). Virol., 39, 861-870 (1981). TAYLOR-PAPADIMITMOU, J., LANE,E.B. and CHANG,S.E., Cell lineages HILKENS,J., BUUS,F., HILGERS, J . , HAGEMAN, mi., CALAFAT, J., SON- and interactions in neoplastic expression in the human breast. Zn: M.A. Rich, J.C. Hager and P. Furmanski (eds.), Understanding breast cancer, NENBERG, A. and VAN DER VALK,M., Monoclonal antibodies against human milk-fat globule membranes detecting differentiation antigens of M. Dekker, New York (1983). the mammary gland and its tumors. In?. J. Cancer, 34, 197-206 (1984). TAYLOR-PAPADIMITMOU, J., PETERSON, J.A., ARKLIE,J., BURCHELL, I., W.F., Monoclonal antibodies to epitheliumJENSEN,H.M., Breast pathology emphasizing precancerous and cancer- CERIANI,R.L. and BODMER, associated lesions. In: R.D. Bulbrook and D.J. Taylor (eds.), Commen- specific components of the human milk fat globule membrane: production and reaction with cells in culture. Znt. J . Cancer, 28, 17-21 (1981b). taries on breast disease, pp. 41-86, A.R. Liss, New York (1981). J., PURKJS,P. and FENTIMAN, I S . , Cholera S.E. and TAYLOR-PAPADIMITMOU, J., Monoclonal an- TAYLOR-PAPADIMITMOU, KEEN,J., CHANG, tibodies that distinguish between human cellular and plasma fibronectin. toxin and analogues of cyclic AMP stimulate the growth of cultured mammary epithelial cells. J. cell. Physiol., 102, 317-321 (1980). Mol. biol. Med., 2, 15-27 (1984). J., STAMPFER, M., BARTEK,J., LEWIS, A., LANE,D.P. and CRAWFORD, L.V., T antigen is bound to a host protein in TAYLOR-PAPADIMITRIOU, BOSHELL, M., LANE,E.B. and LEIGH,I.M., Keratin expression in human SV40-transformed cells. Nature (Lond.),278, 261-263 (1979). mammary epithelial cells cultured from normal and malignant tissue: reLANE,E.B. and KLYMKOWSKY, M.W., Epithelial tonofilaments: investi- lation to in vivo phenotypes and influence of medium. J. Cell Sci., 94, gating their form and function using monoclonal antibodies. Cold Spr. 403-413 (1989). Harb. Symp. quant. Biol.,46,387402 (1982). W.W., ACHTSTATTER, T., VANMUIJEN,G.N.P., RWITER, D.J., FRANKE, LAUEROVA, L., KOVARIK, J., BARTEK, J., REJTHAR, A. and VOJTESEK, HAASNOOT, W.H.B., PONEC,M. and WARNAAR, S.O., Cell type heteroB., Novel monoclonal antibodies defining epitopes of human cytokeratin geneity of cytokeratin expression in complex epithelia and carcinomas as 18 molecule. Hybridoma, 7, 495-504 (1988). demonstrated by monoclonal antibodies specific for cytokeratin numbers 4 MOLE,S.E., GANNON, J.V., FORD,M.J. and LANE,D.P., Structure and and 13. Exp. Cell Res., 162, 97-113 (1986). function of SV40 large T antigen. Phil. Trans. roy. SOC.Lond. B , 317, VANMUIJEN,G.N.P., WARNAAR, S.O. and PONEC,M., Differentiation455469 (1987). related changes of cytokeratin expression in cultured keratinocytes and in A. and FRANKE, W.W., fetal, newborn, and adult epidermis. Exp. Cell Res., 171,331-345 (1987). MOLL,R., LEE, I., GOWLD,V.E., ROESSNER, Immunocytochemical analysis of Ewing’s tumors. Amer. J . Path., 127, WELLINGS, S.R., JENSEN,H.M. and MARCUM, R.G., An atlas of subgross 288-304 (1987). pathology of the human breast with special reference to possible precanORMEROD, M.G., STEELE,K., WESTWOOD, J.H. and HAZZINI,M.N., cerous lesions. J . nat. Cancer Inst., 55, 231-273 (1975).