Journal of Dairy Research (1992) 59 491-498 Printed in Great Britain
491
Characteristics of ruminant mammary epithelial cells grown in primary culture in serum-free medium BY STEVEN J. WINDER*, ALAN TURVEYf AND ISABEL A. FORSYTHft Endocrinology and Animal Physiology Department, AFRC Institute for Grassland and Animal Production, Hurley, Maidenhead, Berks SL6 SLR, UK (Received J2 December 1991 and accepted for publication 17 March 1992) Cells were obtained from the mammary glands of sheep and cows by collagenase-hyaluronidase digestion. Characterization of cells as epithelial was by reaction with a monoclonal antibody to cytokeratin. A subpopulation of spindleshaped or stellate cells reacted with a monoclonal antibody to desmin and may be related to myoepithelial cells. The development is described of a simple serum-free culture system for these cells on gels of rat tail (type 1) collagen. A commercial medium (M199) was used, buffered with Hepes and with bovine serum albumin as the sole protein supplement, plus fibronectin for the first 18 h only as an attachment factor. The cell cultures showed stimulated DNA synthesis in response to mitogens on attached gels and also responded as floating cultures to lactogenic hormones with production of a-lactalbumin. SUMMARY.
Growth of isolated mammary epithelial cells in vitro has required the presence of serum or of complex additions to defined media as serum substitutes. Serum in maintenance medium introduces uncontrolled and varying factors which may lead to synergic or antagonistic effects, preventing clear recognition of the action of added hormone(s) or growth factors. The replacement of plastic by type 1 collagen as a substrate has been an important development and mimics better the in vivo environment of the mammary epithelial cell. This has helped to achieve the growth of mammary cells for longer periods while still retaining some original functions (Emerman & Pitelka, 1977). In many cases, nevertheless, where serum-free media have been developed, unphysiologically high concentrations of insulin, epidermal growth factor (EGF), prolactin, cortisol, progesterone, oestradiol and other supplements have been used (Imagawa et al. 1982; Hammond et al. 1984; Ethier et al. 1987; Hansen & Knudsen, 1991). Attempts to evaluate the role of various growth factors and hormones may be frustrated by the presence of factors needed to maintain viable cells. In our efforts to examine fundamental aspects of the regulation by hormones and growth factors of growth and differentiation of ruminant mammary cells, we have developed a simple culture system for the maintenance of ovine and bovine mammary epithelial cells at low rates of DNA synthesis, in which serum is not used at any stage. Under appropriate conditions, these mammary cells can be stimulated to synthesize DNA or differentiate in the presence of physiological concentrations of * Present, address: Department of Medical Biochemistry, University of Calgary, Alberta, Canada T2N 4N1. t Present address: Department of Cell Biology, AFRC Institute of Animal Physiology and Genetics Research, Cambridge Research Station, Babraham CB2 4AT, UK. | For reprints.
492
S. J . W I N D E R AND OTHERS
various growth regulatory factors or hormones and retain features characteristic of mammary epithelial cells as determined by several criteria. MATERIALS AND METHODS
Culture viedia The standard growth medium was Medium 199 (M199, Gibco, Paisley, Renfrewshire) containing 1-1 g NaHCO3/l and 100 mg L-glutamine/1 supplemented with penicillin (120 mg/1), streptomycin sulphate (100 mg/1), 25 mM-Hepes (Sigma, Poole BH17 7TG), 5 mM-sodium acetate and 5 mg/ml bovine serum albumin (BSA; fraction V, 98-99% albumin, Sigma). Additional acetate was provided because of its importance as a major metabolic substrate for ruminant mammary tissue. The BSA preparations used were prepared by heat shock, charcoal treatment and dialysis to remove low molecular mass substances, and contained < 1 ng/mg contaminating insulin or insulin-like growth factor-I (IGF-I) as determined by radioimmunoassay. Digestion medium was prepared from growth medium but contained 400 U/ml hyaluronidase (EC 3.2.1.35, type I-S from bovine testes, 290 U/mg, Sigma) and 0-3 U/ml collagenase (collagenase A, EC 3.4.24.3, ~ 0-3 Wunsch units/mg, Boehringer Corporation, London). Attachment medium was growth medium containing 10 /tg/ml ovine fibronectin (purified from sheep plasma as previously described (Winder et al. 1989)). Other medium supplements tested were cholera toxin (Sigma), crystalline bovine insulin (25-6 i.u/mg, Sigma) and mouse EGF (supplied by Dr K. D. Brown, AFRC Institute of Animal Physiologjr and Genetics Research). Cortisol was from Glaxo (Greenford, Middlesex) and sheep prolactin (NIH-P-S-12) from NIADDK (Bethesda, MD, USA). Substrates All cell cultures were performed in 16 mm diam. 24-well culture dishes (Sterilin, Feltham, Middlesex). For cells grown directly on plastic, mammary acini (0-2 ml) were added directly to the dishes in attachment medium. Mammary gland extracellular matrix was prepared from sheep mammary gland essentially as described by Wicha et al. (1982) and used to coat culture dishes before addition of acini in attachment medium (0-2 ml). Rat tail collagen gels were prepared as previously described (Mackenzie et al. 1982). For culture on collagen gels, 04 ml of neutralized rat tail collagen was added to each well, allowed to polymerize and then acini in attachment medium (0-2 ml) were added. For cultures within collagen gels, acini were resuspeiided into 0-5% (w/v) rat tail collagen before addition on ice of 10 x concentrated M199 (9:1, v/v) and 0-34 M-NaOH to neutralize (Emerman & Pitelka, 1977); 0-4 ml of this suspension was added to each well of the culture dish. Following polymerization, 0-2 ml attachment medium was added to each well. After 18 h in attachment medium, all cultures were changed to growth medium with the addition of the factors under test. Tissue culture Mammary tissue was obtained from Poll Dorset ewes before or during their first pregnancy and from non-pregnant, non-lactating Friesian cows. After stunning by captive bolt and exsanguination, tissue was transferred to the laboratory and immediately (usually within 5 min of slaughter), cut into 1 cm cubes and injected, via a 20 gauge needle, with digestion medium until fully distended (Kraehenbuhl, 1977). Tissue (20 g) was transferred to a Bellco (A R Horwell, London) spinner flask with a further 30 ml digestion medium and incubated at 37 °0 with gentle stirring
Ruminant mammary epithelial cell culture
493
(~ 1 rev./s) for 6-18 h. After digestion, mammary acini were recovered by filtration through a graded series of nylon mesh filters (Henry Simon, Stockport, Cheshire) from 1000 /on, 400 /im, 210/tm, 90 /HII, through to 20 /tm. The retentate on the 20 fim filter was recovered and resuspended in growth medium. The suspension of acini was centrifuged briefly (1000 £, 10 s) to pellet acini and washed to remove digestion medium and remaining single cells as determined by microscopic examination. The acini were then resuspended in either attachment medium for growth on collagen gels, plastic or mammary gland extracellular matrix, or 0 - 5% (w/v) rat tail collagen for growth within collagen gels as required. Plating density was ~ 1 fig DNA per well. After attachment (18 h) all media were changed to growth medium (0-4 ml) containing hormones and growth factors (see Results). Cells were grown at 37 °C in a humidified air environment. Synthesis of DNA was quantified by the incorporation of [3H]thymidine (Winder et al. 1989) and expressed as d.p.m.//ig DNA. a-Lactalbumin in medium was measured by specific radioimmunoassay (Forsyth et al. 1985). Immunocytochemistry Cytokeratin and desmin were demonstrated on tissue sections and in cells after culture using monoclonal antibodies and a biotin-streptavidin detection system. The reagents used were supplied by Amersham International pic (Amersham, Bucks). Samples of mammary tissue from pregnant and lactating sheep were frozen in isopentane cooled in liquid N2. Cryostat sections (5 /jm) were cut and stored unfixed at — 20 °C. Sections were brought to room temperature, air-dried for 2 h, treated with non-immune rabbit serum (5% v/v) in phosphate-buffered saline (0"145 MNaCl-0-01 M-phosphate, PBS, pH 7-1) for 30 min and the excess poured off without rinsing. Mouse monoclonal anticytokeratin (RPN. 1100) or antidesmin (RPN. 1101) was applied at 1:40 dilution in PBS and left overnight at 4 °C. Washing in PBS was followed by incubation in biotinylated sheep anti-mouse Ig (RPN. 1001, 1:40 dilution). The anti-mouse Ig was demonstrated using streptavidin-biotinylated horseradish peroxidase complex (RPN. 1051, 1:50) and diaminobenzidine (0-5 mg/ml) as substrate in PBS, pH 7-4, containing 0-02% (v/v) H2O2 and 0-03% (w/v) NiCl2- Collagen gels with mammary cells attached were treated briefly with acetone at —20 °C, then brought rapidly to room temperature and washed well in PBS. Other procedures were as for sections, the gels being slide mounted after the final step. RESULTS
Cell attachment Mackenzie et al. (1982) allowed bovine mammary epithelial cell clumps to attach in the presence of 5% (v/v) fetal calf serum. In growth medium (see Methods) without serum, attachment and subsequent growth of bovine and ovine mammary epithelial cells on collagen was found to be poor. In order to facilitate cell attachment and outgrowth, sheep plasma fibronectin (5-10 /^g/ml) was added to growth medium. The presence of fibronectin was no longer necessary after the first medium change at 18 h. Synthesis of DNA and colony growth then continued in the absence of fibronectin in serum-free growth medium. A medium change every 24 h was necessary for the rate of DNA synthesis to increase progressively over time until confluence was approached. At this time, spontaneous detachment and contraction of the gels occurred through change in shape of the attached cells (Emerman & Pitelka, 1977) and DNA synthesis dramatically declined.
494
S. J. W I N D E R AND OTHERS
Time, d
Fig. 1. Effect of substrate on DNA synthesis by epithelial cells, prepared from a non-pregnant sheep, in response to insulin (500 ng/ml). Cells were grown on plastic (•), on extracellular matrix (O), in collagen gels (A) or on collagen gels (•)• See text for details. Results are mean + SEM for triplicate observations.
Effect of substrate on mammary epithelial cell growth Various substrates have been successful for the cultivation of mammary epithelia from several species. In order to assess which substrate would be most suitable for the study of ruminant mammary epithelial cells, the synthesis of DNA on four different substrates in response to insulin (500 ng/ml) was compared. Fig. 1 shows [3H]thymidine incorporation into DNA by mammary epithelial cells from a nonpregnant ewe grown on one of four substrates: tissue culture plastic, mammary gland extracellular matrix and on or within rat tail collagen gels. On all substrates it can be seen that there was a time-dependent increase of thymidine incorporation into DNA. However, the response of the cells to the dose of insulin used was greatest in the cells grown on 0-5% rat tail collagen gels. This substrate was chosen for future studies. Optimization of serum-free medium Bovine mammary epithelial cells had previously been successfully cultured on collagen gels in our laboratory (Mackenzie et al. 1982) in the presence of 5 % fetal calf serum. Imagawa et al. (1982) had developed a serum-free culture system for mouse mammary cells. The effects of constituents of the mouse mammary epithelial cell culture medium were examined on the synthesis of DNA by ruminant mammary epithelial cells in the absence of serum. For each constituent, the dose—response relationships were examined in the presence of constant concentrations of the others (BSA 5 rag/ml; EGF 10 ng/ml; cholera toxin 10 ng/ml; insulin 5/tg/ml). By contrast with mouse mammary cells, neither EGF, nor cholera toxin which increases intracellular cyclic AMP concentrations, produced any clear dose-dependent effect on DNA synthesis, while BSA and insulin both stimulated [3H]thymidine incorporation into DNA in a dose-dependent manner. Results for bovine mammary epithelial cells are shown in Fig. 2 and similar results were obtained in sheep (results not shown). For studies of the role of insulin and IGF as mammary mitogens
Ruminant mammary epithelial cell culture
495
18
Q
2
14
.11
•2-
2
0 1 10 BSA, mg/ml
0 1 10100 00-11 10 EGF, CT, ng/ml ng/ml
00-11 10 Insulin, pg/ml
Fig. 2. DNA synthesis by mammary epithelial cells prepared from a non-pregnant, non-lactating cow and cultured on collagen in various concentrations of bovine serum albumin (BSA, 9 ) , epidermal growth factor (EGF, O)> cholera toxin (CT, • ) and insulin ( • ) . The concentration of each medium supplement was varied in the presence of a constant concentration of each of the other three factors (BSA, 5 mg/ml; EGF, 10 ng/ml; CT, 10 ng/ml; insulin 5/tg/ml). Growth in unsupplemented medium is also shown (A)- D&y 4 of culture, mean + SEM for triplicate observations.
(Winder el al. 1989), we therefore optimized the concentration of BSA in the absence of other components. The incorporation of [3H]th3'midine into DNA was maximum at 5 mg BSA/ml and this concentration was used in subsequent studies. As a percentage of maximum the responses at 0, 0-5 and 50 mg BSA/ml were 75-1 ±4-5%, 72-4 + 3-2% and 91-9±1-4% respectively (mean + SEM results for three sheep). Immunocytochemical characterization of cultured mammary cells In sections of mammary tissue from pregnant and lactating sheep, the anticytokeratin antibody localized to secretory and basally located epithelial cells within the basement membrane. There was no staining in cells of the stroma. The antidesmin antibody demonstrated basally located cells at the periphery of alveoli, identified from their shape and position as myoepithelial cells, and also cells in the walls of blood vessels. When applied to cells from collagen cultures, the anticytokeratin antibody recognized almost all cells, although the intensity of reaction varied (Fig. 3a). The majority of cells were polygonal in shape, but spindleshaped or stellate cells were also seen especially at the periphery of colonies. These cells were detected with the monoclonal antibody to desmin (Fig. 36). Accumulation of a-ladalbumin by mammary cells in serum-free medium The identity of cells as of epithelial origin was also shown by their ability to produce a-lactalbumin under hormonal stimulation. Mammary cells were isolated from late pregnant sheep (110 d, n = 5) and grown on attached collagen gels in the presence of IGF-I (10 ng/ml) until confluent (4-5 d). Gels were then rimmed and allowed to float (Emerman & Pitelka, 1977). After 24 h, during which time the gels contracted, medium containing combinations of cortisol, sheep prolactin and insulin was added and culture continued for 3 d. Cells cultured with cortisol (500 ng/ml) and
496
S. J. WINDER AND OTHERS
Fig. 3. Reactivity of cultured cells from sheep mammary gland with monoclonal antibody to (a) cytokeratin (b) desmin. Bar indicates 50/(in.
prolactin (200 ng/ml) released into the medium 5-5 + 2-5 ng a-lactalbumin//tg DNA. This was increased 8-fold to 48-2 +16-8 ng/fig DNA (P < 0-02) by the further addition of insulin (10 ng/ml). The amount of a-lactalbumin released per fig DNA was comparable to that produced by mammary explants from late-pregnant sheep cultured similarly (range 4-50 ng//ig DNA depending on hormones added, results not shown). DISCUSSION
The introduction of collagen (Emerman & Pitelka, 1977) and other extracellular matrix components as a substrate has permitted growth and function of mammary epithelial cells in primary culture from such species as mouse and rat (see Imagawa
Ruminant mammary epithelial cell culture
497
et al. 1990) and ruminants (Mackenzie et al. 1982, 1985; McGrath, 1987; Shamay etal. 1988; Winder et al. 1989;TalhoukeZaZ. 1990;Hansen&Knudsen, 1991; McGrath et al. 1991). Several different commercially available media have been used, but in the earlier studies serum was required for cell attachment, survival and sustained growth. A number of more or less complex supplements to replace serum have been described. In order to study the effects of growth factors, our aim was to use a medium with the simplest formulation possible, consistent with retaining viability and response to mitogens. This has been achieved by using purified fibronectin as an attachment factor and BSA as a sole supplement. What BSA contributes is at present unknown, but the preparation used was low in fatty acids, low molecular mass components removed by charcoal treatment and dialysis, and was measured to contain < 1 ng/mg IGF-I and insulin. A possible role of BSA is to prevent adsorption to plastic of autocrine factors (Lippman & Dickson, 1989; Wheatley et al. 1989) produced by the cells themselves. Even in the absence of any added hormones or growth factors, ruminant mammary epithelial cells show an exponential increase in DNA synthesis with time (Winder et al. 1989). The cytokeratins are a complex group of polypeptides forming intermediate filaments in epithelial cells. The cytokeratins present in bovine udder and a cultured cell line from bovine mammary epithelium have been characterized by Schmid et al. (1983). Using a monoclonal antibody, we have demonstrated cytokeratin in primary cultures of ovine mammary cells grown on collagen, demonstrating their epithelial nature. A subset of epithelial cells of different morphology (spindle-shaped or stellate, not cuboidal) reacted also with an antibody to desmin, suggesting their relationship to myoepithelial cells. The prolactin-stimulated production of a-lactalbumin by cells on floating collagen gels confirm their mammary epithelial origin and functional activity in serum-free medium. Cells prepared and cultured by this method have been shown to respond to mitogens (IGF-I and insulin, Winder et al. 1989; IGF-II, Wheatley & Forsyth, 1990). The response is retained until confluence is reached, usually at ~ 5-6 d of culture. Most importantly, the in vitro response reflects the physiological state of the animals from which the tissue is taken, with a shorter lag phase for tissue from late-pregnant sheep compared with nonpregnant or early-pregnant sheep (Winder et al. 1989) or lactating sheep (S. D. Wheatley & I. A. Forsyth, unpublished results). Ruminant mammary cells can be grown embedded in collagen gels (McGrath, 1987; Fig. 1) as well as on attached gels. In gels the cells become organized as ductlike colonies. McGrath et al. (1991) have suggested that the method of culture may affect the relative potency of the variants of IGF, but direct comparisons within the same species would be needed to establish this. It is not known at present which culture method best reflects the in vivo situation. In the present stud}', only a limited response to EGF was observed. Preliminary studies (C. M. Moorby & I. A. Forsyth, unpublished results) indicate that transforming growth factor-a (TGF-a), which also acts through the EGF receptor, is more active than EGF on sheep mammary cells cultured on collagen, synergizing with IGF-I. Similarly, Zurfluh et al. (1990) found a greater maximum response of DNA to TGF-a than to EGF in the presence of IGF-I for bovine mammary cells cultured in collagen. In summary, this simple culture system allows mammary cells which are predominantly presecretory epithelial cells to be cultured under conditions in which the effects of added growth factors on DNA synthesis can be studied and secretion of a milk component can be induced. The system therefore provides a method for the
498
S. J. W I N D E R AND OTHERS
elucidation of the basic requirements of mammary epithelial cells for growth and differentiation. We are grateful to NIADDK for supptying sheep prolactin and to Dr K. D. Brown for the gift of mouse epidermal growth factor. REFERENCES EMERMAN, J. T. & PITELKA, D. R. 1977 Maintenance and induction of morphological differentiation in dissociated mammary epithelium on floating collagen membranes. In Vitro 13 316-328 ETHIER, S. P., KUDLA, A. & CUNDIFF, K. C. 1987 Influence of hormone and growth factor interactions on the proliferative potential of normal rat mammary epithelial cells in vitro. Journal of Cellular Physiology 132 161-167 FORSYTH, I. A., BYATT, J. C. & ILEY, S. 1985 Hormone concentrations, mammary development and milk yield in goats given long-term bromocriptine treatment in pregnancy. Journal of Endocrinology 104 77 85 HAMMOND, S. L., HAM, R. 0 . & STAMPFER, M. R. 1984 Serum-free growth of human mammary epithelial tells: rapid clonal growth in defined medium and extended serial passage with pituitary extract. Proceedings of the National Academy of Sciences of the USA 81 5435-5439 HANSEN, H. O. & KNUDSEK, J. 1991 Lactating goat mammary gland cells in culture. Comparative Biochemistry & Physiology 99A 129-135 IMAGAWA, W., BANDYOPADHYAY, G. K. & NANDI, S. 1990 Regulation of mammary epithelial cell growth in mice and rats. Endocrine Reviews 11 494-523 IMAGAWA, W., TOMOOKA, Y. & NANDI, S. 1982 Serum-free growth of normal and tumor mouse mammary epithelial cells in primary culture. Proceedings of the National Academy of Sciences of the USA 79 4074 4077 KRAEHENBUHL, J. P. 1977 Dispersed mammary gland epithelial cells. I. Isolation and separation procedures. Journal of Cell Biology 72 390-405 LIPPMAN, M. E. & DICKSON, R. B. 1989 Mechanisms of growth control in normal and malignant breast epithelium. Recent Progress in Hormone Research 45 383-435 MCGRATH, M. F. 1987 A novel system for mammary epithelial cell culture. Journal of Dairy Science. 70 1907-1980 MCGRATH, M. F., COLLIER, R. J., CLEMMONS, D. R., BUSBY, W. H., SWEENY, C. A. & K R I V I , G. G. 1991 The
direct in vitro effect of insulin-like growth factors (IGFs) on normal bovine mammary cell proliferation and production of IGF binding proteins. Endocrinology 129 671-678 MACKENZIE, D. D. S., BROOKER, B. E. & FORSYTH, J. A. 1985 Ultrastructural features of bovine mammary epithelial cells grown on collagen gels. Tissue & Cell 17 39-51 MACKENZIE, D. D. S., FORSYTH, I. A., BROOKER, B. E. & TURVEY, A. 1982 Culture of bovine mammary
epithelial cells on collagen gels. Tissue & Cell 14 231-241 SCHMID, E., SCHILLER, D. L., GRUND, C , STADLER, J . & FRANKE, W. W. 1983 Tissue type-specific expression
of intermediate filament proteins in a cultured epithelial cell line from bovine mammary gland. Journal of Cell Biology 96 37-50 SHAMAY, A., COHEN, N., NIWA, M. & GERTLER, A.
1988 Effect of insulin-like growth factor
I on
deoxyribonucleic acid synthesis and galactopoiesis in bovine undifferentiated and lactating mammary tissue in vitro. Endocrinology 123 804-809 TALHOUK, R. S., NEISWANDER, R. L. & SCHANBACHER, F. L. 1990 In vitro culture of cryopreserved bovine mammary cells on collagen gels: synthesis and secretion of casein and lactoferrin. Tissue k Cell 22 583-599 WHEATLEY, S. D., COLES, M. L., TURVEY, A., MORRELL, D. J. & FORSYTH, L A .
1989 Insulin-like growth
factor-1 release by ovine mammary epithelial cells. Journal of Endocrinology 123, Suppl. Abstr. no. 118 WHEATLEY, S. D. & FORSYTH, I. A. 1990 Both 1GF-II and the truncated form of 1GF-I stimulate DNA synthesis in ovine mammary epithelial cells in vitro. Journal of Reproduction and Fertility, Abstract Series 5 114 WICHA, M. S., LOWRIE, G., KOHN, E., BAGA VAN DOSS, P. & MAHN, T.
1982 Extracellular matrix promotes
mammary epithelial growth and differentiation in vitro. Proceedings of the National Academy of Sciences of the USA 79 3213-3217 WINDER, S. J., TURVEY, A. & FORSYTH, I. A. 1989 Stimulation of DNA synthesis in cultures of ovine mammary epithelial cells by insulin and insulin-like growth factors. Journal of Endocrinology 123 319 326 ZURFLUH, L. L., BOLTEN, S. L., BYATT, J. C , MCGRATH, M. F., TOU, J. S., ZUPEC, M. E. & KRIVI, G. G. 1990
Isolation of genomic sequence encoding a biologically active bovine TGF-a protein. Growth Factors 3 257-266
491
Characteristics of ruminant mammary epithelial cells grown in primary culture in serum-free medium BY STEVEN J. WINDER*, ALAN TURVEYf AND ISABEL A. FORSYTHft Endocrinology and Animal Physiology Department, AFRC Institute for Grassland and Animal Production, Hurley, Maidenhead, Berks SL6 SLR, UK (Received J2 December 1991 and accepted for publication 17 March 1992) Cells were obtained from the mammary glands of sheep and cows by collagenase-hyaluronidase digestion. Characterization of cells as epithelial was by reaction with a monoclonal antibody to cytokeratin. A subpopulation of spindleshaped or stellate cells reacted with a monoclonal antibody to desmin and may be related to myoepithelial cells. The development is described of a simple serum-free culture system for these cells on gels of rat tail (type 1) collagen. A commercial medium (M199) was used, buffered with Hepes and with bovine serum albumin as the sole protein supplement, plus fibronectin for the first 18 h only as an attachment factor. The cell cultures showed stimulated DNA synthesis in response to mitogens on attached gels and also responded as floating cultures to lactogenic hormones with production of a-lactalbumin. SUMMARY.
Growth of isolated mammary epithelial cells in vitro has required the presence of serum or of complex additions to defined media as serum substitutes. Serum in maintenance medium introduces uncontrolled and varying factors which may lead to synergic or antagonistic effects, preventing clear recognition of the action of added hormone(s) or growth factors. The replacement of plastic by type 1 collagen as a substrate has been an important development and mimics better the in vivo environment of the mammary epithelial cell. This has helped to achieve the growth of mammary cells for longer periods while still retaining some original functions (Emerman & Pitelka, 1977). In many cases, nevertheless, where serum-free media have been developed, unphysiologically high concentrations of insulin, epidermal growth factor (EGF), prolactin, cortisol, progesterone, oestradiol and other supplements have been used (Imagawa et al. 1982; Hammond et al. 1984; Ethier et al. 1987; Hansen & Knudsen, 1991). Attempts to evaluate the role of various growth factors and hormones may be frustrated by the presence of factors needed to maintain viable cells. In our efforts to examine fundamental aspects of the regulation by hormones and growth factors of growth and differentiation of ruminant mammary cells, we have developed a simple culture system for the maintenance of ovine and bovine mammary epithelial cells at low rates of DNA synthesis, in which serum is not used at any stage. Under appropriate conditions, these mammary cells can be stimulated to synthesize DNA or differentiate in the presence of physiological concentrations of * Present, address: Department of Medical Biochemistry, University of Calgary, Alberta, Canada T2N 4N1. t Present address: Department of Cell Biology, AFRC Institute of Animal Physiology and Genetics Research, Cambridge Research Station, Babraham CB2 4AT, UK. | For reprints.
492
S. J . W I N D E R AND OTHERS
various growth regulatory factors or hormones and retain features characteristic of mammary epithelial cells as determined by several criteria. MATERIALS AND METHODS
Culture viedia The standard growth medium was Medium 199 (M199, Gibco, Paisley, Renfrewshire) containing 1-1 g NaHCO3/l and 100 mg L-glutamine/1 supplemented with penicillin (120 mg/1), streptomycin sulphate (100 mg/1), 25 mM-Hepes (Sigma, Poole BH17 7TG), 5 mM-sodium acetate and 5 mg/ml bovine serum albumin (BSA; fraction V, 98-99% albumin, Sigma). Additional acetate was provided because of its importance as a major metabolic substrate for ruminant mammary tissue. The BSA preparations used were prepared by heat shock, charcoal treatment and dialysis to remove low molecular mass substances, and contained < 1 ng/mg contaminating insulin or insulin-like growth factor-I (IGF-I) as determined by radioimmunoassay. Digestion medium was prepared from growth medium but contained 400 U/ml hyaluronidase (EC 3.2.1.35, type I-S from bovine testes, 290 U/mg, Sigma) and 0-3 U/ml collagenase (collagenase A, EC 3.4.24.3, ~ 0-3 Wunsch units/mg, Boehringer Corporation, London). Attachment medium was growth medium containing 10 /tg/ml ovine fibronectin (purified from sheep plasma as previously described (Winder et al. 1989)). Other medium supplements tested were cholera toxin (Sigma), crystalline bovine insulin (25-6 i.u/mg, Sigma) and mouse EGF (supplied by Dr K. D. Brown, AFRC Institute of Animal Physiologjr and Genetics Research). Cortisol was from Glaxo (Greenford, Middlesex) and sheep prolactin (NIH-P-S-12) from NIADDK (Bethesda, MD, USA). Substrates All cell cultures were performed in 16 mm diam. 24-well culture dishes (Sterilin, Feltham, Middlesex). For cells grown directly on plastic, mammary acini (0-2 ml) were added directly to the dishes in attachment medium. Mammary gland extracellular matrix was prepared from sheep mammary gland essentially as described by Wicha et al. (1982) and used to coat culture dishes before addition of acini in attachment medium (0-2 ml). Rat tail collagen gels were prepared as previously described (Mackenzie et al. 1982). For culture on collagen gels, 04 ml of neutralized rat tail collagen was added to each well, allowed to polymerize and then acini in attachment medium (0-2 ml) were added. For cultures within collagen gels, acini were resuspeiided into 0-5% (w/v) rat tail collagen before addition on ice of 10 x concentrated M199 (9:1, v/v) and 0-34 M-NaOH to neutralize (Emerman & Pitelka, 1977); 0-4 ml of this suspension was added to each well of the culture dish. Following polymerization, 0-2 ml attachment medium was added to each well. After 18 h in attachment medium, all cultures were changed to growth medium with the addition of the factors under test. Tissue culture Mammary tissue was obtained from Poll Dorset ewes before or during their first pregnancy and from non-pregnant, non-lactating Friesian cows. After stunning by captive bolt and exsanguination, tissue was transferred to the laboratory and immediately (usually within 5 min of slaughter), cut into 1 cm cubes and injected, via a 20 gauge needle, with digestion medium until fully distended (Kraehenbuhl, 1977). Tissue (20 g) was transferred to a Bellco (A R Horwell, London) spinner flask with a further 30 ml digestion medium and incubated at 37 °0 with gentle stirring
Ruminant mammary epithelial cell culture
493
(~ 1 rev./s) for 6-18 h. After digestion, mammary acini were recovered by filtration through a graded series of nylon mesh filters (Henry Simon, Stockport, Cheshire) from 1000 /on, 400 /im, 210/tm, 90 /HII, through to 20 /tm. The retentate on the 20 fim filter was recovered and resuspended in growth medium. The suspension of acini was centrifuged briefly (1000 £, 10 s) to pellet acini and washed to remove digestion medium and remaining single cells as determined by microscopic examination. The acini were then resuspended in either attachment medium for growth on collagen gels, plastic or mammary gland extracellular matrix, or 0 - 5% (w/v) rat tail collagen for growth within collagen gels as required. Plating density was ~ 1 fig DNA per well. After attachment (18 h) all media were changed to growth medium (0-4 ml) containing hormones and growth factors (see Results). Cells were grown at 37 °C in a humidified air environment. Synthesis of DNA was quantified by the incorporation of [3H]thymidine (Winder et al. 1989) and expressed as d.p.m.//ig DNA. a-Lactalbumin in medium was measured by specific radioimmunoassay (Forsyth et al. 1985). Immunocytochemistry Cytokeratin and desmin were demonstrated on tissue sections and in cells after culture using monoclonal antibodies and a biotin-streptavidin detection system. The reagents used were supplied by Amersham International pic (Amersham, Bucks). Samples of mammary tissue from pregnant and lactating sheep were frozen in isopentane cooled in liquid N2. Cryostat sections (5 /jm) were cut and stored unfixed at — 20 °C. Sections were brought to room temperature, air-dried for 2 h, treated with non-immune rabbit serum (5% v/v) in phosphate-buffered saline (0"145 MNaCl-0-01 M-phosphate, PBS, pH 7-1) for 30 min and the excess poured off without rinsing. Mouse monoclonal anticytokeratin (RPN. 1100) or antidesmin (RPN. 1101) was applied at 1:40 dilution in PBS and left overnight at 4 °C. Washing in PBS was followed by incubation in biotinylated sheep anti-mouse Ig (RPN. 1001, 1:40 dilution). The anti-mouse Ig was demonstrated using streptavidin-biotinylated horseradish peroxidase complex (RPN. 1051, 1:50) and diaminobenzidine (0-5 mg/ml) as substrate in PBS, pH 7-4, containing 0-02% (v/v) H2O2 and 0-03% (w/v) NiCl2- Collagen gels with mammary cells attached were treated briefly with acetone at —20 °C, then brought rapidly to room temperature and washed well in PBS. Other procedures were as for sections, the gels being slide mounted after the final step. RESULTS
Cell attachment Mackenzie et al. (1982) allowed bovine mammary epithelial cell clumps to attach in the presence of 5% (v/v) fetal calf serum. In growth medium (see Methods) without serum, attachment and subsequent growth of bovine and ovine mammary epithelial cells on collagen was found to be poor. In order to facilitate cell attachment and outgrowth, sheep plasma fibronectin (5-10 /^g/ml) was added to growth medium. The presence of fibronectin was no longer necessary after the first medium change at 18 h. Synthesis of DNA and colony growth then continued in the absence of fibronectin in serum-free growth medium. A medium change every 24 h was necessary for the rate of DNA synthesis to increase progressively over time until confluence was approached. At this time, spontaneous detachment and contraction of the gels occurred through change in shape of the attached cells (Emerman & Pitelka, 1977) and DNA synthesis dramatically declined.
494
S. J. W I N D E R AND OTHERS
Time, d
Fig. 1. Effect of substrate on DNA synthesis by epithelial cells, prepared from a non-pregnant sheep, in response to insulin (500 ng/ml). Cells were grown on plastic (•), on extracellular matrix (O), in collagen gels (A) or on collagen gels (•)• See text for details. Results are mean + SEM for triplicate observations.
Effect of substrate on mammary epithelial cell growth Various substrates have been successful for the cultivation of mammary epithelia from several species. In order to assess which substrate would be most suitable for the study of ruminant mammary epithelial cells, the synthesis of DNA on four different substrates in response to insulin (500 ng/ml) was compared. Fig. 1 shows [3H]thymidine incorporation into DNA by mammary epithelial cells from a nonpregnant ewe grown on one of four substrates: tissue culture plastic, mammary gland extracellular matrix and on or within rat tail collagen gels. On all substrates it can be seen that there was a time-dependent increase of thymidine incorporation into DNA. However, the response of the cells to the dose of insulin used was greatest in the cells grown on 0-5% rat tail collagen gels. This substrate was chosen for future studies. Optimization of serum-free medium Bovine mammary epithelial cells had previously been successfully cultured on collagen gels in our laboratory (Mackenzie et al. 1982) in the presence of 5 % fetal calf serum. Imagawa et al. (1982) had developed a serum-free culture system for mouse mammary cells. The effects of constituents of the mouse mammary epithelial cell culture medium were examined on the synthesis of DNA by ruminant mammary epithelial cells in the absence of serum. For each constituent, the dose—response relationships were examined in the presence of constant concentrations of the others (BSA 5 rag/ml; EGF 10 ng/ml; cholera toxin 10 ng/ml; insulin 5/tg/ml). By contrast with mouse mammary cells, neither EGF, nor cholera toxin which increases intracellular cyclic AMP concentrations, produced any clear dose-dependent effect on DNA synthesis, while BSA and insulin both stimulated [3H]thymidine incorporation into DNA in a dose-dependent manner. Results for bovine mammary epithelial cells are shown in Fig. 2 and similar results were obtained in sheep (results not shown). For studies of the role of insulin and IGF as mammary mitogens
Ruminant mammary epithelial cell culture
495
18
Q
2
14
.11
•2-
2
0 1 10 BSA, mg/ml
0 1 10100 00-11 10 EGF, CT, ng/ml ng/ml
00-11 10 Insulin, pg/ml
Fig. 2. DNA synthesis by mammary epithelial cells prepared from a non-pregnant, non-lactating cow and cultured on collagen in various concentrations of bovine serum albumin (BSA, 9 ) , epidermal growth factor (EGF, O)> cholera toxin (CT, • ) and insulin ( • ) . The concentration of each medium supplement was varied in the presence of a constant concentration of each of the other three factors (BSA, 5 mg/ml; EGF, 10 ng/ml; CT, 10 ng/ml; insulin 5/tg/ml). Growth in unsupplemented medium is also shown (A)- D&y 4 of culture, mean + SEM for triplicate observations.
(Winder el al. 1989), we therefore optimized the concentration of BSA in the absence of other components. The incorporation of [3H]th3'midine into DNA was maximum at 5 mg BSA/ml and this concentration was used in subsequent studies. As a percentage of maximum the responses at 0, 0-5 and 50 mg BSA/ml were 75-1 ±4-5%, 72-4 + 3-2% and 91-9±1-4% respectively (mean + SEM results for three sheep). Immunocytochemical characterization of cultured mammary cells In sections of mammary tissue from pregnant and lactating sheep, the anticytokeratin antibody localized to secretory and basally located epithelial cells within the basement membrane. There was no staining in cells of the stroma. The antidesmin antibody demonstrated basally located cells at the periphery of alveoli, identified from their shape and position as myoepithelial cells, and also cells in the walls of blood vessels. When applied to cells from collagen cultures, the anticytokeratin antibody recognized almost all cells, although the intensity of reaction varied (Fig. 3a). The majority of cells were polygonal in shape, but spindleshaped or stellate cells were also seen especially at the periphery of colonies. These cells were detected with the monoclonal antibody to desmin (Fig. 36). Accumulation of a-ladalbumin by mammary cells in serum-free medium The identity of cells as of epithelial origin was also shown by their ability to produce a-lactalbumin under hormonal stimulation. Mammary cells were isolated from late pregnant sheep (110 d, n = 5) and grown on attached collagen gels in the presence of IGF-I (10 ng/ml) until confluent (4-5 d). Gels were then rimmed and allowed to float (Emerman & Pitelka, 1977). After 24 h, during which time the gels contracted, medium containing combinations of cortisol, sheep prolactin and insulin was added and culture continued for 3 d. Cells cultured with cortisol (500 ng/ml) and
496
S. J. WINDER AND OTHERS
Fig. 3. Reactivity of cultured cells from sheep mammary gland with monoclonal antibody to (a) cytokeratin (b) desmin. Bar indicates 50/(in.
prolactin (200 ng/ml) released into the medium 5-5 + 2-5 ng a-lactalbumin//tg DNA. This was increased 8-fold to 48-2 +16-8 ng/fig DNA (P < 0-02) by the further addition of insulin (10 ng/ml). The amount of a-lactalbumin released per fig DNA was comparable to that produced by mammary explants from late-pregnant sheep cultured similarly (range 4-50 ng//ig DNA depending on hormones added, results not shown). DISCUSSION
The introduction of collagen (Emerman & Pitelka, 1977) and other extracellular matrix components as a substrate has permitted growth and function of mammary epithelial cells in primary culture from such species as mouse and rat (see Imagawa
Ruminant mammary epithelial cell culture
497
et al. 1990) and ruminants (Mackenzie et al. 1982, 1985; McGrath, 1987; Shamay etal. 1988; Winder et al. 1989;TalhoukeZaZ. 1990;Hansen&Knudsen, 1991; McGrath et al. 1991). Several different commercially available media have been used, but in the earlier studies serum was required for cell attachment, survival and sustained growth. A number of more or less complex supplements to replace serum have been described. In order to study the effects of growth factors, our aim was to use a medium with the simplest formulation possible, consistent with retaining viability and response to mitogens. This has been achieved by using purified fibronectin as an attachment factor and BSA as a sole supplement. What BSA contributes is at present unknown, but the preparation used was low in fatty acids, low molecular mass components removed by charcoal treatment and dialysis, and was measured to contain < 1 ng/mg IGF-I and insulin. A possible role of BSA is to prevent adsorption to plastic of autocrine factors (Lippman & Dickson, 1989; Wheatley et al. 1989) produced by the cells themselves. Even in the absence of any added hormones or growth factors, ruminant mammary epithelial cells show an exponential increase in DNA synthesis with time (Winder et al. 1989). The cytokeratins are a complex group of polypeptides forming intermediate filaments in epithelial cells. The cytokeratins present in bovine udder and a cultured cell line from bovine mammary epithelium have been characterized by Schmid et al. (1983). Using a monoclonal antibody, we have demonstrated cytokeratin in primary cultures of ovine mammary cells grown on collagen, demonstrating their epithelial nature. A subset of epithelial cells of different morphology (spindle-shaped or stellate, not cuboidal) reacted also with an antibody to desmin, suggesting their relationship to myoepithelial cells. The prolactin-stimulated production of a-lactalbumin by cells on floating collagen gels confirm their mammary epithelial origin and functional activity in serum-free medium. Cells prepared and cultured by this method have been shown to respond to mitogens (IGF-I and insulin, Winder et al. 1989; IGF-II, Wheatley & Forsyth, 1990). The response is retained until confluence is reached, usually at ~ 5-6 d of culture. Most importantly, the in vitro response reflects the physiological state of the animals from which the tissue is taken, with a shorter lag phase for tissue from late-pregnant sheep compared with nonpregnant or early-pregnant sheep (Winder et al. 1989) or lactating sheep (S. D. Wheatley & I. A. Forsyth, unpublished results). Ruminant mammary cells can be grown embedded in collagen gels (McGrath, 1987; Fig. 1) as well as on attached gels. In gels the cells become organized as ductlike colonies. McGrath et al. (1991) have suggested that the method of culture may affect the relative potency of the variants of IGF, but direct comparisons within the same species would be needed to establish this. It is not known at present which culture method best reflects the in vivo situation. In the present stud}', only a limited response to EGF was observed. Preliminary studies (C. M. Moorby & I. A. Forsyth, unpublished results) indicate that transforming growth factor-a (TGF-a), which also acts through the EGF receptor, is more active than EGF on sheep mammary cells cultured on collagen, synergizing with IGF-I. Similarly, Zurfluh et al. (1990) found a greater maximum response of DNA to TGF-a than to EGF in the presence of IGF-I for bovine mammary cells cultured in collagen. In summary, this simple culture system allows mammary cells which are predominantly presecretory epithelial cells to be cultured under conditions in which the effects of added growth factors on DNA synthesis can be studied and secretion of a milk component can be induced. The system therefore provides a method for the
498
S. J. W I N D E R AND OTHERS
elucidation of the basic requirements of mammary epithelial cells for growth and differentiation. We are grateful to NIADDK for supptying sheep prolactin and to Dr K. D. Brown for the gift of mouse epidermal growth factor. REFERENCES EMERMAN, J. T. & PITELKA, D. R. 1977 Maintenance and induction of morphological differentiation in dissociated mammary epithelium on floating collagen membranes. In Vitro 13 316-328 ETHIER, S. P., KUDLA, A. & CUNDIFF, K. C. 1987 Influence of hormone and growth factor interactions on the proliferative potential of normal rat mammary epithelial cells in vitro. Journal of Cellular Physiology 132 161-167 FORSYTH, I. A., BYATT, J. C. & ILEY, S. 1985 Hormone concentrations, mammary development and milk yield in goats given long-term bromocriptine treatment in pregnancy. Journal of Endocrinology 104 77 85 HAMMOND, S. L., HAM, R. 0 . & STAMPFER, M. R. 1984 Serum-free growth of human mammary epithelial tells: rapid clonal growth in defined medium and extended serial passage with pituitary extract. Proceedings of the National Academy of Sciences of the USA 81 5435-5439 HANSEN, H. O. & KNUDSEK, J. 1991 Lactating goat mammary gland cells in culture. Comparative Biochemistry & Physiology 99A 129-135 IMAGAWA, W., BANDYOPADHYAY, G. K. & NANDI, S. 1990 Regulation of mammary epithelial cell growth in mice and rats. Endocrine Reviews 11 494-523 IMAGAWA, W., TOMOOKA, Y. & NANDI, S. 1982 Serum-free growth of normal and tumor mouse mammary epithelial cells in primary culture. Proceedings of the National Academy of Sciences of the USA 79 4074 4077 KRAEHENBUHL, J. P. 1977 Dispersed mammary gland epithelial cells. I. Isolation and separation procedures. Journal of Cell Biology 72 390-405 LIPPMAN, M. E. & DICKSON, R. B. 1989 Mechanisms of growth control in normal and malignant breast epithelium. Recent Progress in Hormone Research 45 383-435 MCGRATH, M. F. 1987 A novel system for mammary epithelial cell culture. Journal of Dairy Science. 70 1907-1980 MCGRATH, M. F., COLLIER, R. J., CLEMMONS, D. R., BUSBY, W. H., SWEENY, C. A. & K R I V I , G. G. 1991 The
direct in vitro effect of insulin-like growth factors (IGFs) on normal bovine mammary cell proliferation and production of IGF binding proteins. Endocrinology 129 671-678 MACKENZIE, D. D. S., BROOKER, B. E. & FORSYTH, J. A. 1985 Ultrastructural features of bovine mammary epithelial cells grown on collagen gels. Tissue & Cell 17 39-51 MACKENZIE, D. D. S., FORSYTH, I. A., BROOKER, B. E. & TURVEY, A. 1982 Culture of bovine mammary
epithelial cells on collagen gels. Tissue & Cell 14 231-241 SCHMID, E., SCHILLER, D. L., GRUND, C , STADLER, J . & FRANKE, W. W. 1983 Tissue type-specific expression
of intermediate filament proteins in a cultured epithelial cell line from bovine mammary gland. Journal of Cell Biology 96 37-50 SHAMAY, A., COHEN, N., NIWA, M. & GERTLER, A.
1988 Effect of insulin-like growth factor
I on
deoxyribonucleic acid synthesis and galactopoiesis in bovine undifferentiated and lactating mammary tissue in vitro. Endocrinology 123 804-809 TALHOUK, R. S., NEISWANDER, R. L. & SCHANBACHER, F. L. 1990 In vitro culture of cryopreserved bovine mammary cells on collagen gels: synthesis and secretion of casein and lactoferrin. Tissue k Cell 22 583-599 WHEATLEY, S. D., COLES, M. L., TURVEY, A., MORRELL, D. J. & FORSYTH, L A .
1989 Insulin-like growth
factor-1 release by ovine mammary epithelial cells. Journal of Endocrinology 123, Suppl. Abstr. no. 118 WHEATLEY, S. D. & FORSYTH, I. A. 1990 Both 1GF-II and the truncated form of 1GF-I stimulate DNA synthesis in ovine mammary epithelial cells in vitro. Journal of Reproduction and Fertility, Abstract Series 5 114 WICHA, M. S., LOWRIE, G., KOHN, E., BAGA VAN DOSS, P. & MAHN, T.
1982 Extracellular matrix promotes
mammary epithelial growth and differentiation in vitro. Proceedings of the National Academy of Sciences of the USA 79 3213-3217 WINDER, S. J., TURVEY, A. & FORSYTH, I. A. 1989 Stimulation of DNA synthesis in cultures of ovine mammary epithelial cells by insulin and insulin-like growth factors. Journal of Endocrinology 123 319 326 ZURFLUH, L. L., BOLTEN, S. L., BYATT, J. C , MCGRATH, M. F., TOU, J. S., ZUPEC, M. E. & KRIVI, G. G. 1990
Isolation of genomic sequence encoding a biologically active bovine TGF-a protein. Growth Factors 3 257-266
