Expression of epithelial antigens in primary cultures of normal human breast analysed with monoclonal antibodies

Differentiation a Springer-Vcrlag 1984

Differentiation (1984) 2 5 : 247-258

Expression of epithelial antigens in primary cultures of normal human breast analysed with monoclonal antibodies Paul A.W. Edwards, Isobel M. Brooks, and Paul Monaghan Ludwig Institute for Cancer Research (London Branch), Royal Marsden Hospital, Sutton, Surrey, SM2 SPX, U.K.

Abstract. Primary cultures of normal human breast were stained with monoclonal antibodies to see if antigens characteristic of luminal epithelial cells are retained in culture. Three monoclonal antibodies were used, LICR-LON-M8, LICR-LON-MI 8, and LICR-LON-M24, all specific for the cell surface of luminal epithelial as opposed to myoepithelial or stromal cells in the breast, and each staining a different subset of the epithelial cells in the intact tissue. Cultures were prepared from reduction mammoplasty samples by digestion with collagenase. The surface layer of cells was stained by immunofluorescence without fixation. (Cells underneath the surface layer were not accessible to this mode of staining). The antibodies stained patches of cells resembling flattened epithelium. These patches of cells cannot be distinguished by phase contrast microscopy without reference to the staining, in fact the boundaries of the cells are not usually resolved by phase contrast microscopy. Electron microscopy of sections through these cells show they are very flattened. They lie on top of the polygonal and elongated cells that dominate the phase contrast image. Two of the antibodies, M8 and M24, stain subsets of these epithelial-like cells at all stages of culture. The third antibody, M18, stains such cells initially, but after the first few days staining is predominantly found on the polygonal and elongated cells, then this also gradually disappears. It is possible that the cells stained by antibody M18 are converting from the epithelial-like morphology to the cuboidal and elongated morphology. Many cells are not stained by any of the antibodies, so appear either to by myoepithelial in origin or to have lost their luminal epithelial surface antigens at an early stage. This analysis draws attention to the variety of cell types in these cultures and the limitations of phase contrast microscopy as a means of analysing them.

Introduction

A major aim of cell culture is to maintain and manipulate cell differentiation in vitro. Two obstacles to this are the difficulty of identifying cells in culture and the tendency of at least some cells to lose or alter their differentiated state in culture [l, 10, 181. Monoclonal antibodies to cell surface antigens [33] offer a new approach to these problems, as cells in different states of differentiation can often be recognised by characteristic differences in their surface antigens 171. The aim of this investigation was to see wheth-

er monoclonal antibodies that mark characteristic antigens of the luminal epithelial cells in intact tissue could be used to analyse the fate of epithelial cells in primary cultures of normal human breast. There are two main histological types of cell in breast ducts and alveoli: luminal epithelial and myoepithelial cells. A series of monoclonal antibodies have been raised in this laboratory that bind specifically to the surface of luminal epithelial cells as opposed to myoepithelial cells, or stromal or vascular cells, in the normal ‘resting’ breast [14] (Fig. 1 and Table 1). They are therefore a powerful set of markers for identifying luminal epithelial cells. Not only do these antibodies distinguish luminal epithelial cells but, quite unexpectedly, they each stain one of three different, partly overlapping, subpopulations of luminal epithelial cells in vivo : this has been established using two-colour immunofluorescence staining of unfixed and viable duct epithelium obtained by dissection [8,9a]. The three antibodies between them stain most but perhaps not all luminal epithelial cells. When primary cultures of breast are prepared by digestion with collagenase, the stroma is digested away leaving fragments of duct and alveoli which are usually plated as explants without further dissociation (Figs. 2 and 3). Although such cultures have been extensively studied both in human and rodent [reviewed in 23, 34, see also 6, 10, 20, 29, 351, it is not clear which cells in the culture come from where in the tissue, nor how much of their in vivo differentiation is retained. The only method widely available for identifying cells has been electron microscopy [6, 10, 19, 291 but many of the cells lose their characteristic ultrastructure in culture - particularly on plastic surfaces - and few cells can be sampled. Recently, conventional antisera to human milk fat globule membrane have been used to show that a proportion of cells in such a primary culture express antigens characteristic of epithelial cells [3, 61 and also retain epithelial ultrastructure (61, but their topographic distribution was not determined. We therefore stained primary breast cultures with the monoclonal antibodies, simply staining the surface layer of cells in the intact cultures without prior fixation, to see if the antigens that they recognise were still expressed by cells on the surface of the cultures.

Methods Monoclonal antibodies. The production of the monoclonal antibodies to human breast epithelial cells (Table 1) has

248

been described previously [14]. Briefly, mice were immunized with human milk fat globule membrane. Spleen cells were fused with mouse myeloma NS1 cells by essentially conventional methods [7, 161 and clones were selected for production of antibody to milk fat globule membrane. Unless otherwise stated, ascitic fluid was the source of antibody, and was used at a dilution of 1 : 100. The nonspecific monoclonal antibody LICR-LONFIB75 was raised similarly but using mice immunised with breast fibroblasts [9]. It binds to most adult human cells except haemopoietic cells (P.A.W. Edwards and R.A.J. McIlhinney, unpublished work). Culrures. Tissue from over 30 reduction mammoplasty (cosmetic) operations was prepared for culture by conventional methods as previously described [6]. Briefly, diced tissue was incubated with stirring overnight at 37” C in collagenase (Type IA, Sigma Chemical Co., St. Louis, Mo.) at 1 mg/ ml in complete culture medium: Dulbecco’s modified Eagle’s medium (DMEM) with 10% (v/v) foetal bovine serum, 100 pg/ml kanamycin, and 1 pg/ml Fungizone (all from Gibco Europe, Paisely, Scotland). Tissue fragments were recovered by gentle centrifugation, washed in medium, and redigested for 2 h with collagenase at 2.5 mg/ml in complete medium. “Organoids” (Fig. 2) [29], i.e. fragments of duct and lobular alveolar units, largely free from fibroblasts and vascular cells, were then recovered by repeated sedimentation for $ to 1 h at 1 g , leaving most of the stromal cells and vessel fragments in suspension. At this stage organoids were sometimes, where indicated, cryopreserved in 10% (v/ v) dimethyl sulphoxide, 10% (v/v) foetal bovine serum in DMEM [6]. Cryopreservation made no obvious difference to the results except that the organoids took one or two days longer to attach and initiate spreading. Organoids were plated in complete medium at a low density on “Petriperm” dishes (Herdeus, Stockport, UK) that had been coated with dried collagen by wetting with a 3% (w/v) solution of rat tail collagen [lo] (largely Type I collagen [20]) in 0.1% (v/v) acetic acid, draining, and allowing them to dry thoroughly before use. This collagen coating made no obvious difference to the patterns of antibody staining seen but it accelerated and facilitated attachment and spreading of the organoids [20]. Cultures rich in vascular fragments used for comparative purposes were obtained by plating the supernatant after sedimentation at 1 g . Cultures were stained between 3 and 14 days after plating. Immunofluorescence. Cultures were washed with “assay medium” (Dulbecco’s modified Eagle’s medium with 10 mM Hepes replacing bicarbonate, pH 7.4, with 1% (v/v) foetal bovine serum, 1% (w/v) bovine serum albumin, plus 3 mM NaN,), cut up into l-cm squares and transferred to 24-well tissue culture dishes (Costar, obtained via L.H. Engineering, Stoke Poges, Bucks., UK). They were incubated unfixed on ice for $-I h with monoclonal antibody in assay medium, washed, incubated with fluorescent-conjugated second antibody, washed, and mounted in 9% (w/w) polyvinyl alcohol (Goshenol, Polaron Ltd., Watford, UK), 23% (w/w) glycerol in 50 mM Tris-HC1 (pH 8.5 at 20” C) [21]. Fluorescence was photographed with exposures generally of 1 min on XP1 film (Ilford, Mobberley, Cheshire, UK), using a Laborlux microscope (Leitz, Wetzlar, Germany), with Ploem illumination through filter blocks L2 or M2 respectively, for fluorescein (FITC) or rhodamine

(tetramethylrhodamine, TRITC, and substituted rhodamine, XTRITC) fluorescence.At low power (XI0 objective) an additional K610 suppression filter and increased exposure were used with the M2 filter to improve contrast. Where indicated, photographs were taken on a Polyvar microscope (Reichert, Vienna). Two-colour immunofluorescence. IgG,- and IgM-specific fluorochrome-conjugated second antibodies were used for double immunofluorescence where appropriate (Table I). To compare the IgM monoclonal antibodies with each other, biotinyl or dinitrophenyl (DNP) derivatives were made and the second antibody replaced respectively by rhodamine-conjugated avidin or fluorescein-conjugated rabbit anti-DNP antibody. (Avidin is a protein with a very high affinity for biotin). The fluorescence obtained with the antiDNP method was further amplified by adding a third layer of fluorescein-conjugated goat anti-rabbit antibody. To prepare derivatives the monoclonal antibodies were partially purified by precipitation with an equal volume of saturated ammonium sulphate. Biotinyl derivatives were prepared by incubating antibody at 1 mg/ml in phosphate buffered saline (150 mM NaCl, 3 mM NaN,, 10 mM Na phosphate buffer, pH 7.4) with 200 bg/ml biotinyl-N-hydroxysuccinimide ester (kindly provided by Dr. J. Westwood, Institute of Cancer Research, Sutton) for 4 h at 20” C, then dialysing. DNP derivatives were prepared by incubating antibody at 1 mg/ml in phosphate buffered saline rebuffered to about pH 8.0 with 5% w/v NaHCO,, with 200 pg/ml fluoro-2’,4‘dinitrobenzene (Sigma Chemical Co., St. Louis, Mo. added in ethanol at 10 mg/ml) for 4 h at 20” C and dialysed. Fluorescent-conjugated second antibodies were obtained from Nordic Immunological Laboratories (Maidenhead, Berks., UK) and Meloy (via Gibco Europe, Paisley, Scotland), and substituted tetramethylrhodamine isothiocyanate (XTR1TC)-conjugated avidin from Vector Laboratories (via Sera-Lab Ltd., Crawley Down, West Sussex, UK). Rabbit anti-DNP antibody was generously provided by Dr. B. Ponder, Institute of Cancer Research, Sutton, and was conjugated with fluorescein isothiocyanate (FITC) (Nordic Immunological Laboratories, Maidenhead, Berks., UK) as described in [22]. Fluorescent conjugates were selected for specificity, and generally residual cross-reactions and background staining were removed by absorbing on either mouse IgG (Miles Labs., Stoke Poges, Slough, UK) or IgM (TEPC 183 myeloma protein from Uniscience, Cambridge, UK) followed by human serum proteins, coupled to CNBr-activated Sepharose (Pharmacia Ltd., Uxbridge, UK), and centrifuging for 1 h at 150,000 gav. Sectioning and electron microscopy. Organoids were plated on collagen-coated ‘Thermanox ’ coverslips (Flow Labs., Irvine, Scotland). They were lightly stained by the indirect immunoperoxidase method: after incubation with a mixture of antibodies M8 and M24 as for immunofluorescence, staining was completed at 20” C in 150 mM NaC1/20 mM Na phosphate buffer (pH 7.4). Coverslips were washed, incubated with horseradish peroxidase-conjugated rabbit anti-(all mouse immunoglobulin) second antibody for 1 h, washed, and incubated in 0.05% (w/v) diaminobenzidine 0.05% (v/v) hydrogen peroxide for 5 min, and fixed in 2% (w/v) glutaraldehyde for 1 h and 1% osmium tetroxide for 2 h. Both fixatives were phosphate buffered and the osmotic pressure was adjusted to 350 mOsM by addition of sucrose.

249

The coverslips were dehydrated in ethanol and embedded in Epon/Araldite [25] via propylene oxide. After polymerisation the coverslips were removed from the resin blocks and 1.5 pm sections were cut vertical to the plane of the coverslip. These were stained with toluidine blue and photographed. Thin sections were cut with a diamond knife on a Reichert OMU-4 microtome and examined without further staining at 60 KV in a Phillips EM400 electron microscope. Immunocytochemical staining of tissue sections was on conventional formalin-fixed paraffn-embedded sections using alkaline-phosphatase-conjugatedgoat anti-(mouse immunoglobulin) as second antibody and visualising the enzyme with fast red [14]. Nuclei were counterstained with haemalum.

Results Culture conditions

Digesting human breast with collagenase produces intact fragments of epithelium, ‘organoids’ [29], corresponding to that part of the tissue inside the basement membrane (Fig. 2). When cultured on plastic dishes, these organoids attach and cells migrate out to form characteristic islands (Fig. 3) [6, 15, 23, 24, 291. Cultures were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% (v/v) foetal bovine serum and antibiotics. Growthpromoting additives [29, 30, 34, 351 such as insulin, hydrocortisone, and epidermal growth factor (EGF) were omitted, to avoid cell proliferation which might lead to loss of differentiation characteristics or overgrowth of unresponsive cells. Under these conditions few cells divide and the islands of cells develop by spreading from the organoids [6]. They are a complex mixture of morphological types of cells, multilayered near the remnants of the organoids, thinning to monolayer at periphery [6] (Fig. 3). Significant spreading has occurred by three or four days; after about 14 days in this medium the bulk of the cells appear elongated by phase-contrast microscopy, and eventually the monolayered areas tend to detach and roll up. Immunojluorescence staining shows epithelium-like patches of cells

In order to understand the staining given by the antibodies specific for epithelial cells, we first show the result of staining virtually all the cells on the surface of the cultures, i.e. cells accessible to immunofluorescence staining of the intact cultures. For this we used (Fig. 4) an antibody cocktail that stains virtually all the cells including myoepithelial and stromal cells. This cocktail was a mixture of the antiepithelial antibodies (Table 1) together with monoclonal antibody LICR-LON-FIB75, which binds to all or virtually all cells in the breast. All antibodies are visualised simultaneously with an FITC-conjugated anti-(all mouse immunoglobulin) second antibody. The phase contrast image (Fig. 4b) suggests that most of the island of cells consists of highly aligned elongated cells. Comparison with the fluorescence image (Fig. 4a) shows that some of the surface cells correspond to the phase contrast image, but in places the elongated cells are covered by patches of quite different cells. These ‘patch’ cells are arranged in a pavement, like sheets of epithelium but very flattened compared to epitheli-

Table 1. Properties of monoclonal antibodies specific for luminal epithelial cells’

Designation

Ig subclass

Apparent mol. wt. of antigen in kilodaltonsb

Epitope

LICR-LON-M8 LICR-LON-M18

IgG,

2200,000

Uncertain

IgM

~200,000

Galg (1 ’4) GlcN, A@( 1+6)

TgM

slightly smaller than M8 antigen 39 to 59,000 predominant broad band@)

LICR-LON-M24

[13, 171

Carbohydrate as also present on glycolipid

* Within the breast the antibodies bind only to the luminal epithelial cells, and to only some of these. They bind to the apical membranes and in some cases also the lateral membranes. They do not obviously distinguish alveolar from ductal cells. They are not breast-specific: cells in some other epithelia are also stained. Cell specificity data is from [14] Identification of antigens (principal bands) by immunoblotting and lipid thin layer chromatography overlay of milk fat globule membrane (from R.A.J. McIlhinney, S. Patel, and M. Gore submitted for publication). In addition M8 but not MI8 binds to a purified preparation of the high molecular weight glycoprotein bearing the epithelial membrane antigen described by Ormerod and coworkers [28, M.G. Ormerod, K. Steele, and J. Westwood, manuscript in preparation, personal communication] which is probably the molecule described in Ref. [27l

al cells in vivo. It is not possible to distinguish these cells by phase contrast microscopy, whether mounted after immunofluorescence staining or not, without reference to the staining, except in rare cases where their outlines can just be discerned as characteristic black lines (Figs. 4c and d). Their nuclei are probably seen but because the cultures are multilayered they can only be tentatively identified by reference to the fluorescence image. The boundaries of the sheets and of some of the more rounded-up cells can often be found in phase contrast but only by reference to the staining: we can find no distinctive features of the edges. These patch cells are present in all preparations at all stages of culture. The proportion of the surface they occupy is very variable between islands and ranges from a confluent sheet covering a whole island to isolated groups of cells, but few islands have none. Their appearance in section is shown below (Fig. 9). Staining by antibodies LICR-LON-M8 and LICR-LON-M24

Antibodies LICR-LON-M8 and LICR-LON-M24, which stain subsets of luminal epithelial cells in vivo, stain subpopulations of the epithelium-like ‘patch’ cells identified in Fig. 4. Their staining is compared by two-colour immunofluorescence in Fig. 5. In some areas the cells stained by antibody M24 are a subset of those stained by M8 (this is generally true in Fig. 5a-c); sometimes the reverse is seen, and sometimes quite separate areas of culture may be stained (not illustrated). Most dramatically, the patterns of staining can be complementary to each other (Fig. 5d-f). The cells stained by M8 and M24 appear similar in distribution and morphology, except that in some areas the cells

2 50

Fig. 1. Staining of a section of normal breast by the monoclonal antibody LICR-LON-MI as detailed previously [14]. Staining is %en on the apical (luminal) membrane of the luminal epithelial cells and some cells are positive (arrowed) while others are unstained [14]. x 600 Bar 10 pm Fig. 2. Fragmcnt of epithelium, an ‘organoid’ obtained by collagenasc digestion of normal breast, as put into culture. This is a large example consisting of duct with branches and some alveoli. Phase contrast x 80. Bur 100 pm Fig. 3. Phase contrast appearance, before immunofluorescence staining and mounting, of an island of cells spreading out eight days after plating a typical fragment of breast epithelium or ‘organoid’ on the simple medium used in this study. The body top centre is the remnant of the organoid. x 120. Bar 100 pm

stained by M8 have less exposed surface area than M24stained cells (Fig. 5 b and c). Staining by untibody LICR-LON-M18

phologically different types of cell at different times after plating. In the first few days of culture it stains patch cells like M8 and M24, indeed its staining often overlaps with staining by M8 or M24. However, &is patch cell staining rapidly - - disappears (by four to seven days) and instead staining is predominandy; and later exclusively, found on a second type of cell. Figure 6 contrasts the two types of cell _

Antibody LICR-LON-M18, which stains the third subpopulation of luminal epithelial cells in vivo, stains two mor-

,

251

Fig.4. a Singlecolour immunofluorescence staining of a l0-day primary culture of human breast with an antibody ‘cocktail’ that stains virtually all cells that are accessible to staining. Both ‘basal’ cells and patch cells are clearly visible. x 130. b Phase contrast of the same field. Note the patch cells are not visible. x 130. c Immunofluorescence staining of a 5 d a y primary culture of human breast with the antibody cocktail. This is an unusual example where the boundaries of the patch cells can be Seen in the phase contrast image. Arrows mark a patch cell. x360. d Phase contrast of the same field as (c). The outline of the patch cell marked in (c) is indicated by the arrows. x 360. Bur 100 pm

stained by M18 at a stage where both are present and shows staining of patch cells by M18 overlapping with the staining by M8. Figure 7 contrasts the second type of cell stained by MI8 at about one week with patch cell staining by M8 and M24 at this stage. Referring back to Fig. 4, there is a clear distinction between patch cells and the remaining cells in the culture, which dominate the phase contrast image and which we will refer to here for convenience as

basal cells. The second type of M18-stained cells are a subpopulation of these basal cells, morphologically indistinguishable from the unstained basal cells. They are clearly seen in the phase-contrast image and vary in shape from polygonal to highly elongated (Figs. 7b and e). They are often in apparently monlayered regions of the islands of cells, frequently right at the edge (Fig. 7). Their fluorescence staining is characteristically more coarsely stippled than the

252

Fig. 5a-f. Two examples, (a-c) and (d-f), of cultures stained by two-colour immunofluorescence using antibodies M8 and M24. a Phase contrast. b M8 staining (fluorescein fluorescence). c M24 staining of the same field (rhodamine fluorescence). The population of M24-staining cells is almost entirely contained within the MI-staining cells. d Phase contrast. e M8 staining. f M24 staining. In this second example the cell populations stained by the two antibodies are almost completely complementary. a-c Nine-day culture photographed on Reichert Polyvar microscope. x 50. d-f Thirteen-day culture. x 225. Bur 100 pm

staining of patch cells and the cell's outline is prominent and ragged. Staining by antibody M18 of basal cells is observed increasingly up to about seven days in culture under the conditions used here; thereafter, it either becomes weaker or fewer cells are stained, or both, and it has completely disappeared by 14 days. Combined staining by the three anti-epithelial antibodies together

In the intact tissue all or almost all luminal epithelial cells are stained by one or more of the antibodies M8, M18,

and M24. Therefore, staining a culture with a mixture of M8, M18, and M24 should show almost all the cells on the surface of a culture that have retained antigen expression characteristic of epithelial cells (Fig. 8). Many surface cells are unstained, showing that a large proportion of them are either not of luminal epithelial origin or else have lost their differentiated antigenic phenotype. This is even more striking after the antigen recognised by antibody M18 has disappeared, at 10 to 14 days, when the only cells accessible to staining that retain epithelial surface phenotypes are the patch cells stained by M8 and M24. In confirmation of the in vivo specificity of the three

253

with a mixture of antibodies M8 and M24, using the immunoperoxidase method instead of immunofluorescence, to mark groups of patch cells. The cultures were embedded in resin and chosen areas sectioned in a plane perpendicular to the culture surface. Figure 9a, b shows an area of culture with typical patch cells revealed by the immunoperoxidase staining but whose presence and location could not be predicted from the phase contrast image without the slight darkening due to the peroxidase staining. (The outlines of the patch cells are, in fact, made more prominent by this technique which causes some rounding up and retraction of the cells). Figure 9c is a resin section showing the immunoperoxidase-stained patch cells and negative cells. Figures 9d and e show a section subsequent to that in Fig. 9c examined in the electron microscope. The antibody-stained patch cells are seen to be generally more spread and flattened than the underlying layer of cells. Most of them have many surface microvilli and appear more electrondense than other cells in the culture. The staining is not simply related to the presence of microvilli as not all cells with microvilli on the surface of the culture were stained.

Fig. 6a-c. Staining by antibody MI8 at an early stage in culture contrasted by twocolour immunofluorescence with M8 staining. a Phase contrast. b Mi8 staining. c M8 staining. Whilst M18 stains some of the M8-positive patch cells (arrows), it also stains cells of a more ragged outline that are never stained with M8 or M24. Six-day culture from cryopreserved orgdnoids (comparable to 4 5 day cultures from fresh organoids). x 225. Bar 100 pm

anti-epithelial cell antibodies (Table I), when a culture rich in vessel and stromal cells was prepared, endothelial, pericyte, and fibroblast cells were not stained. Electron microscopy of the epithelium-like patch cells

Typical patch cells were examined in section by light and electron microscopy (Fig. 9). Cultures were lightly stained

Discussion Immunofluorescence staining of the surface layer of cells in these cultures has revealed patches of cells resembling flattened epithelium lying on top of the remaining cells in the culture. Not only do these patches of cells look like epithelium but they also express the antigens recognised by our monoclonal antibodies that are characteristic of the apical membrane of luminal epithelial cells in vivo. The different antibodies stain subsets of these ‘patch’ cells, just as they stain subsets of the epithelial cells in vivo. Surprisingly, the boundaries of the cells within the sheets are rarely resolved by phase contrast microscopy, and their presence and location cannot be predicted without reference to the staining so they may have been neglected in previous studies. Sections through the cells suggest that their boundaries are often very thin compared to the layer of cells underneath, and this perhaps accounts for their minimal contribution to the phase contrast image. Two of the antibodies M8 and M24 continue to stain cells in the epithelium-like sheets throughout the period of observation. However, staining by the third antibody, M18, declines in the patches and is increasingly found on cells in the ‘basal’ layer of cells (all cells in the culture that are not patch cells), though this staining also declines after about a week in culture. The loss of antigen expression might reflect a progressive loss of epithelial differentiation by these cells. A speculative interpretation of the staining by M18, consistent with the observations but impossible at present to substantiate, is that the M18-stained patch cells convert to the M18-stained basal cell type and then gradually lose the antigen recognised by M18, simply because they no longer continue to synthesise it. Such a conversion might seem unlikely were it not for the well-documented conversion of cuboidal to elongated basal cell types in cultured rat mammary gland [I, 5, 261 which could be subsequent to the conversion postulated here. However, an alternative interpretation would be simply that MI 8 stains two distinct morphological types of cell and that the patch cell type loses M18-antigen expression faster than the basal cell type. Our staining suggests that phase contrast microscopy

254

Fig.7. Staining of 6-day cultures by antibody M18 contrasted by two-colour immunofluorescence with staining by M8 and M24. a Phase contrast. b M18 staining. c M8 staining of the same field. The round body bottom left of the photographs is the organoid. Most of its fluorescence in (b) is autofluorescence. Asreri.sks mark a corresponding point in the three photographs. d Phase contrast. e M18 staining. f M24 staining. M24-stained cells appear to lie on top of the cells stained by M18. Note the similarity between M8 and M24-stained cells in (c) and (f) in contrast to the appearance and more peripheral location of M18-stained cells in (b) and (e). a-c x 225, d-f x 90. Bar 100 pm

gives incomplete information about these cultures. It usually fails to show patch cells, so that previous studies have concentrated on the remaining cells that dominate the phase contrast image. It has been suggested [I, 5, 26, 291 that these cuboidal and elongated cells correspond respectively to epithelial and myoepithelial cells. This interpretation does not take account of patch cells and among our antiepithelial cell antibodies only M18 stains any of these cells, and it does not appear to distinguish cuboidal from elongated cells (Figs. 7 b and e). Chang and Taylor-Papadimitriou [4] have stained epithelial cells cultured from human milk with an anti-(epithelial cell) monoclonal antibody and

have also found that staining did not correlate with morphology of the cells. At present we are unable to characterise the surface cells that do not express any of the epithelial antigens. There is no apparent morphological difference between cells on the surface of the basal cell sheet that are or are not stained by antibody M18, so it may be that both M18-negative and M18-positive basal cells are luminal epithelial cells that have lost their characteristic antigens. Some may instead be myoepithelial in origin: if so their similarity to M18stained cells could suggest that epithelial and myoepithelial cells can end up with apparently similar phase contrast mor-

255

Fig. 8. a Single.colour immunofluorescence staining by a mixture of antibodies M8, M18,and M24, all visualised simultaneously with an anti-(all mouse immunoglobulin) fluorescent second antibody, of an 8-day culture of cryopreserved organoids, to show the distribution of all the cells that can be stained by these antibodies. Sheets of stained patch cells are concentrated around organoid remnants while other cells, probably stained by M18, are found right out to the edges of the spreading sheets (arrowed). b Phase contrast. x 110. Insef c shows boxed area at higher power. x 400. Bar 100 pm

256

Fig. 9a-e. Sectioning of typical ‘patch’ cells. a Phase contrast after immunoperoxidase staining (which is, however, scarcely visible in phase contrast) before fixing. The image is different from Figs. 4-8 because the culture has not been mounted and some retraction and rounding-up of the cells occurs during peroxidase staining. b Bright field with cyan filter to show immunoperoxidase staining of the same field. x 190 c Resin section (1.5 pm) cut perpendicular to the plane of the culture surface, approximately along the line shown in (b), at about twice the magnification and countcrstained with toluidine blue. Irnrnunoperoxidase stain is visible as a dark line. The correspondence between (a) and (c) is confirmed by thc organoid remnant and the rounded, staincd cell that is arrowed. x440. d and e Electron micrograph of a section subsequent to that in (c) of the area approximately between the arrows in (c). e This shows a continuation of the right-hand end of (d). x 3,400

257

phologies in these cultures. This would be consistent with observations on rat mammary cell lines [l, 5, 261. Recently, considerable progress has been made towards recovering physiological responses of mammary gland to hormones in vitro by plating cells on or in collagen gel [1&12, 31, 341, or on extracellular matrix [32]. It has been suggested [2, 101 that on a released collagen gel, epithelial cells are able to remain columnar rather than flattened and so retain functions such as secretory responses to hormones. It is possible that the patch cells described here correspond to the cells that respond to hormones when they recover columnar shape on a collagen gel. The analysis presented here has two limitations. The technique does not stain cells that are underneath the surface layer, although it does give clear topographical information, and staining cultures unfixed gives low background staining. Second, a simple culture medium was used to avoid cell proliferation : we are currently staining cultures on medium supplemented with various additives including insulin, hydrocortisone, epidermal growth factor, and cholera toxin to stimulate proliferation [29, 30, 34, 351. The staining patterns are qualitatively as reported here [P.A.W. Edwards and M.J. O’Hare, unpublished observations). In a wider context, these observations show that monoclonal antibodies are useful for identifying differentiated cells in cultures of solid tissues, and that the information they provide is quite different from that provided by morphology or ultrastructure. Acknowledgements. We thank Dr. C.S. Foster, Miss A. McDonald, and Dr. G.C. Easty for help in the early stages of this work; Dr. M.G. Ormerod and Dr. R.A.J. Mcllhinney for permission to quote unpublished work on antigen structures; the surgeons and their staff who kindly provided tissue samples; Professor A.M. Neville for advice and support; and most particularly, Dr. M.J. OHare for advice and with Mrs. Hilary Bunnage for the supply of cultures.

References 1. Bennett DC, Peachey LA, Durbin H, Rudland PS (1978) A possible mammary stem cell line. Cell 15:283-298 2. Burwen SJ, Pitelka DR (1980) Secretory function of lactating

mouse mammary epithelial cells cultured on collagen gels. Exp Cell Res 126: 249-262 3. Ceriani RL, Thompson K, Peterson JA, Abraham S (1977) Surface differentiation antigens of human mammary epithelial cclls carried on the human milk fat globule. Proc Natl Acad Sci USA 74: 582-586 4. Chang SE, Taylor-Papadimitriou J (1983) Modulation of phenotype in cultures of human milk epithelial cells and its relation to the expression of a membrane antigen. Cell Differ 12: 143-1 54 5 . Dulbecco R, Bologna M, Unger M (1979) Differentiation of a rat mammary cell line in vitro. Proc Natl Acad Sci USA 76:12561260 6. Easty GC, Easty DM. Monaghan P, Ormerod MG, Neville AM (1980) Preparation and identification of human breast epithelial cells in culture. Int J Cancer 26: 577-584 7. Edwards PAW (1981) Some properties and applications of monoclonal antibodies. Biochem J 200: 1-10 8. Edwards PAW (1983) Differentiation antigens in an epitheli-

um? Subsets of human breast epithelial cells distinguished by monoclonal antibodies. Biochem SOCTrans 11 :171-172 9. Edwards PAW, Foster CS, McIlhinney RAJ (1980) Monoclonal antibodies to teratomas and breast. Transplant Proc 12:398402

9a. Edwards PAW, Brooks IM (to be published) Antigenic sub-

sets of human breast epithelial cells distinguished by monoclonal antibodies. J Histochem Cytochem 10. Emerman JT, Pitelka DR (1977) Maintenance and induction of morphological differentiation in dissociated mammary epithelium on floating collagen membranes. In Vitro 13 :316328 11. Emerman JT, Enami J, Pitelka DR, Nandi S (1977) Hormonal effects on intracellular and secreted casein in cultures of mouse mammary epithelial cells on floating collagen membranes. Proc Natl Acad Sci USA 74:44664l70 12. Flynn D, Yang J, Nandi S (1982) Growth and differentiation of primary cultures of mouse mammary epithelium embedded in collagen gel. Differentiation 22: 191-194 13. Foster CS (1983) Excess sialylation of breast cancer surface determinant identified by monoclonal antibody LICR-LONM18. Biochem SOCTrans 11 :299-300 14. Foster CS, Edwards PAW, Dinsdale EA, Neville AM (1982) Monoclonal antibodies to the human mammary gland: I. Distribution of determinants in non-neoplastic mammary and extra mammary tissues. Virchows Arch A 394: 279-293 15. Gaffney EV, Polanowski FP, Blackburn SE, Lambidse JT, Burke RE (1976) Cultures of normal human mammary cells. Cell Differ 5:69-81 16. Galfre G, Milstein C (1981) Preparation of monoclonal antibodies : strategies and procedures. Methods Enzymol73 :3 4 6 17. Gooi HC, Uemura K-I, Edwards PAW, Foster CS, Pickering N, Feizi T (1983) Two mouse hybridoma antibodies against milk fat globules recognise the I(Ma) antigenic determinant: GalB1-4GlcNacB1-6.Carbohydr Res 120: 293-302 18. Greenburg G, Vlodavsky I, Foidart JM, Gospodarowicz D (1980) Conditioned medium from endothelial cell cultures can restore the normal phenotypic expression of vascular endothelium maintained in vitro in the absence of fibroblast growth factor. J Cell Physiol 103 :333-347 19. Hallowes RC, Peachey LA (1980) The mammary gland and human breast. In: Hodges G, Hallowes RC (eds) Biomedical application of SEM vol 11. Academic Press, London, pp 167195 20. Hallowes RC, Bone EJ, Jones W (1980) A new dimension in the culture of human breast. In: Richards RJ, Rajan KT (eds) Tissue culture in medical research vol 11. Pergamon Press, Oxford New York, pp 213-220 21. Heimer GV, Taylor CED (1974) Improved mountant for immunofluorescence preparations. J Clin Path0127 :254-256 22. Hudson L, Hay FC (1980) Practical immunology. Blackwell Scientific Publications Ltd, London 23. Janss DH, Hillman EA, Malan-Shibley LB, Ben TL (1980) Methods for the isolation and culture of normal human breast epithelial cells. Methods Cell Biol21: 107-1 34 24. Lasfargues EY (1975) New approaches to the cultivation of human breast carcinomas. In: Fogh J (ed) Human tumor cells in vitro. Plenum Press, New York London, pp 51-77 25. Mollenhauer HH (1964) Plastic embedding mixtures for use in electron microscopy. Stain Techno1 39 : 111-1 14 26. Rudland PS, Bennett DC, Warburton MJ (1979) Hormonal control of growth and differentiation of cultured rat mammary gland epithelial cells. In: Hormones and cell culture, Cold Spring Harbor Conferences on Cell Proliferation vol6, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, pp 677699 27. Shimizu M, Yamauchi K (1982) Isolation and characterization of mucin-like glycoprotein in human milk fat globule membrane. J Biochem 91 :515-524 28. Sloane JP, Ormerod MG (1981) Distribution of epithelial membrane antigen in normal and neoplastic tissues and its value in diagnostic tumor pathology. Cancer 47: 1 7 8 6 1795 29. Stampfer M, Hallowes RC, Hackett AJ (1980) Growth of normal human mammary cells in culture. In Vitro 16:415425 30. Taylor-Papadimitriou J, Purkis P, Fentiman LS (1980) Cholera toxin and analogues of cyclic AMP stimulates the growth of

258

S (1980)

cultured human mammary epithelial cclls. J Cell Physiol

34. Yang J, Richards J, Guzman R, Imagawa W, N a n d

102: 317-321

Sustained growth in primary culture of normal mammary epithelial cells embedded in collagen gels. Proc Nall Acad Sci USA 77 :2088-2092 35. Yang NS, Kube D, Park C, Furmanski P (1981) Growth of human mammary epithelial cells on collagen gel surfaces. Cancer Res 41 :40934100

31. Tonelli QJ, Sorof S (1982) Induction of biochemical differentia-

tion in three-dimensional collagen cultures of mammary epithelial cells from virgin mice. Differentiation 22: 195-200 32. Wicha MS, Lowrie G, Kohn E, Bagavandoss P, Mahn T (1982) Extracellular matrix promotes mammary epithelial growth and differentiation in vitro. Proc Natl Acad Sci USA 79:3213-3217 33. Williams AF, Galfre G, Milstein C (1977) Analysis of cell surfaces by xenogeneic myeloma-hybrid antibodies : Differentiation antigens of rat lymphocytes. Cell 12:663-673

Received April 1983 / Accepted in revised form August 1983