Similar growth pattern of mouse mammary epithelium cultivated in collagen matrix in vivo and in vitro

DEVELOPMENTAL

BIOLOGY

104,57-64 (1984)

Similar Growth Pattern of Mouse Mammary Epithelium Cultivated in Collagen Matrix in Viva and in Vitro CHARLES W. DANIEL, JOSEPHJ. BERGER,’ PHYLLIS STRICKLAND, AND RICHARD GARCIA Lkpartmmt

of Biology, Thimann

LaboraknW,

University of Califarnia, Santa Cruz, California

95064

Received July 26, 1985; accepted in revised form February 2, 1984 Mouse mammary ductal cells cultured in type I collagen gels give rise to three-dimensional multicellular outgrowths consisting of thin spikes which are often branched, and which may have pointed or blunt ends. The significance of these spikes to normal ductal morphogenesis has been unclear, since identical structures are not known to occur in viwconversely, it has not been possible to maintain in gel culture the highly structured end buds which are characteristic of ductal elongation in the animal. In order to evaluate whether the pattern of radiating spikes observed in collagen gel cultures results from chemical or physical peculiarities of the culture environment, a small volume of unpolymerized type I collagen solution was injected into mammary gland-free fat pads of young adult mice. After the bubble of collagen had polymerized, an implant of mammary ductal epithelium was introduced into the center of the gel. Histological examination of the implants after 3 to 6 days of growth revealed numerous small epithelial spikes, similar to those observed in gel culture, extending into the fibrous matrix. The early stages of regeneration of mammary implants placed in gland-free fat pads were then examined without the addition of exogenous collagen. In cases where the epithelium happened to contact a fibrous region of the fatty stroma, spikes were also seen to form in these natural collagenous substrates. Whether or not exogenous collagen was used, normal end buds were formed only when epithelial spikes contacted adipocytee. It was concluded that the three-dimensional pattern of radiating tubules in collagen gels in vitro is not merely an artifact of culture, but has a counterpart in vivo whereever regenerating mammary epithelium is surrounded by fibrous stroma. A model is presented in which the pattern of epithelial outgrowth is determined by the physical characteristics of the surrounding stroma; in collagen matrix a comparatively primitive and unspecialized type of morphogenesis occurs which may not require the participation of stromal cells. In contrast, epithelial-adipocyte interactions appear to be necessary for the formation of end buds and subsequent morphogenesis of fully structured mammary ducts. INTRODUCTION

regress (unpublished observations) and significant ductal elongation has not been reported. More recently, an alternative method of culture has become available in which epithelia are attached to, or embedded in, a matrix of reconstituted type I collagen. First applied to the differentiation of mammary alveolar cells by Emerman and Pitelka (1977), collagen gels have now been used as a three-dimensional matrix for the growth of a variety of mammary tissues and cell types. The general finding has been that mammary epithelial cells, whether in suspension or clumped as “organoids,” give rise to three-dimensional outgrowths consisting of thin spikes, or in some cases tube-like structures with lumens. Such epithelial outgrowths have been reported for mouse (Yang et aL, 198Oc), rat (Bennett, 1980), and human (Foster et CCL,1983; Yang et aL, 1980d) mammary gland, and the various patterns observed have been conveniently classified (Lawler et d, 1983). Mammary spikes, which may exhibit a degree of biochemical specialization (Tonelli and Sorof, 1982), and whose structure is suggestive of tubules formed during the early stages of mammary development, differ substantially from mammary ducts in situ, Specifically,

The growth of mammary ducts is the direct result of activities of mammary end buds. These bulbous enlargements of the ductal tips are of considerable developmental significance, since in addition to supplying cells for differentiation into mature ducts, they provide a focus for the action of regulatory influences which modify both rate and direction of growth, thereby determining the pattern of mammary ductal arborization and setting a limit upon the extent of glandular expansion. Attempts to maintain end buds in culture have been unsuccessful, and as a consequence organotypic growth of mammary ducts in vitro has not been achieved. Isolated buds grown on solid substrates quickly lose their structure and grow as featureless cell sheets, resembling monolayers initiated from mammary alveolar tissue (Daniel, 1965). In whole gland organ cultures, consisting of both epithelia and surrounding fatty stroma, end buds ’ Current address: Department of Physiology and Biophysics, Harvard Medical School, Boston, Mass. 02115. 57

0012-1606E-4 $3.00 Copyright 63 1994 by Academic Press. Inc. All rights of reproduction in any form reserved.

58

DEVELOPMENTAL

BIOLOGY

spikes terminate in points composed of one or a few undifferentiated epithelial cells (Foster et aL, 1983), or in blunt-tipped structures (Ormerod and Rudland, 1982; Pasco et al, 1982) which appear to lack the morphological complexity of end buds. Also, mammogenic hormones added to gel cultures, while stimulating cellular proliferation (Yang et aL, 1980d) or casein production (Tonelli and Sorof, 1982), do not induce the differentiation of lobuloalveolar structures. If the potential of the collagen gel system of culture as a technique for exploring ductal morphogenesis is to be fully developed, it is important to learn why end buds are not maintained in culture, and why the observed pattern of radiating spikes is often similar regardless of tissue of origin. Are differences between in v&o and in vitro growth habits due to nutritional/hormonal deficiencies in the medium, or are they a consequence of the physical properties of the collagen gel culture system? To help answer this question we attempted to simulate in viva the physical conditions used in gel culture by injecting a bubble of type I collagen into the mammary fat pad. Our results indicate that end bud cells incubated in collagenous matrix while exposed to the chemical environment of the host animal grow as epithelial spikes which are similar to those seen in culture; this suggests that this pattern of growth is primarily a response to physical characteristics of the collagen gel environement, which may include changes in tissue interactions, and that structures previously seen only in culture may be generated in the animal under appropriate conditions. MATERIALS

AND

METHODS

Animals and surgm. Female mice of the Balb/c or C57Bl strains were maintained on a 12-hr/12-hr light/ dark cycle. Donor mice were between the ages of 4 and 6 weeks. Female mice to be used as hosts underwent surgery at 3 weeks of age to clear the No. 4 mammary fat pads of host mammary tissue (DeOme et al, 1959) and received mammary transplants at approximately 3 months of age. Histology. At necropsy the mammary glands were removed and fixed for several hours in Tellyesniczky’s fixative and whole mounts were prepared by defatting the glands with acetone and staining with hematoxylin (Williams and Daniel, 1983). Whole mounts were photographed and areas of interest were cut out, embedded in paraffin, and sectioned at 5 pm. Alcian blue staining was carried out under conditions which demonstrated both hyaluronate and sulfated glycosaminoglycans. The dye solution contained 0.1% alcian blue (MCB Manufacturing Chemists, Inc., Cincinnati, Ohio) in 0.025 M sodium acetate buffer (pH 5.8) with 0.3 MMgClz (Bern-

VOLUME

104, 1984

field and Banerjee, 1972). After examination and photography the slides were immersed overnight in xylene to remove the coverslips and restained with hematoxylin-eosin. Adjacent sections were mounted on separate slides and stained with Masson’s trichrome to aid in identification of collagen. Collagen gel preparation Collagen solution was prepared as described by Michalopoulos and Pitot (1975). Rat tails were soaked in 95% ethanol for 15 min and cut into l-in. pieces. The skin was removed and the pieces were placed in sterile distilled water. The silverywhite collagen fibers were removed, picked apart into finer filaments with forceps, and placed in distilled water under uv light for 24 hr to sterilize. Approximately 1g (wet weight) of fibers were then suspended in 200 ml of a sterile solution of 1:lOOO acetic acid and stirred at 4°C for ‘72 hr. The solution was then centrifuged gently to remove undissolved material and the supernatant, containing approximately 2 mg/ml collagen, was stored in the cold. Collagen gels were made by combining complete 10X medium F-12 (K.C. Biologicals) with collagen solution at 0°C and adjusting the pH to 7.2-7.4. For cultures the gel mixture was poured into multiwell plates and placed at 37°C to gel. For animal experiments the cold collagen solution was drawn into a l-ml syringe and 0.01 to 0.02 ml was injected into previously cleared No. 4 fat pads of host mice. In both cases time to gel was approximately 5 min. Tissue. End buds were removed from the fat pads surgically or by mild collagenase digestion. Donor mice were anesthetized with Nembutal and the ventral skin reflected to expose the mammary glands. Using forceps and fine needles, end buds and a short length of subtending duct could be dissected out. As reported (Williams and Daniel, 1983) the end bud tip always separated cleanly from the surrounding stroma while the end bud flank and duct could be removed only by cutting or tearing away the adhering fibrous tissues. For collagenase digestion, a portion of gland containing several end buds was removed, rinsed in sterile saline, and placed in 10 ml of collagenase solution (Worthington type II, 1 mg/ml in serum-free F-12 medium) at 37°C for 15 min. The flask was then swirled vigorously, incubated for another 10 min, swirled again, and emptied into a petri dish. End buds completely free of adherent stroma were removed with a micropipet and placed in sterile saline prior to use. Cultures. Fine jewelers forceps were used to pierce the surface of the collagen gel and an end bud was transferred with micropipet into the pocket created. The gel was covered with medium consisting of complete F-12 (K.C. Biological) with penicillin (100 U/ml), streptomycin (100 pg/ml), 5% porcine serum, epidermal growth

DANIEL

ET AL.

Mouse

Mammary

Epithdium

59

FIG. 1. End bud and duct segments from B-week-old female mouse. End buds similar to these were isolated by either microsurgery or collagenase digestion. X55. FIG. 2. End bud isolated by digestion in collagenase and cultured for 5 days in collagen gel matrix. Branching epithelial outgrowths cover the edge of the end bud (eb) and extend into the collagen gel (cg). The thinner epithelial spikes (arrows) correspond in size and general morphology to those formed in collagen gel implants in viva (Fig. 5). X87.

factor (0.1 pg/ml, Collaborative Research), insulin (5 pg/ml), and transferrin (5 pg/ml), and incubated at 37°C in an atmosphere of 5% CO2 in air. Gels remained attached to the culture chamber throughout incubation. Transplants. Ice-cold collagen solution was drawn into a syringe and 0.01-0.02 ml was injected into the center of each previously cleared No. 4 mammary fat pad. The needle was inserted into the dorsal end of the fat pad, creating a long channel through the tissue which discouraged leakage during the period of polymerization. After the collagen was firmly gelled, end buds were implanted with forceps or micropipet into the approximately center of the collagen mass.

to the surface of a layer of prepolymerized collagen, and covered by a second layer of unpolymerized collagen which was then allowed to gel. Both surgically and enzymatically derived end buds maintained their characteristic morphology for approximately 24 hr in gel culture, after which the tissue masses developed irregular contours and small spike-

.

RESULTS

Grmth

in Vitro in Collagen Gels

Cultures were initiated from two types of end bud preparations. First, isolated end buds with short segments of subtending ducts were obtained by collagenase digestion of mammary glands from 5-week-old female mice (Fig. 1). The second type of preparation consisted of end buds that were surgically removed from the fat pad by fine needles and forceps. These were characterized by a naked epithelial tip which separated easily from the surrounding stroma, and a layer of adherent fibrous tissue which could not be separated from the posterior region of the end buds or their subtending ducts (Williams

and Daniel,

1983). In both

cages, isolated

were briefly washed in BSS, transferred

end buds

by micropipette

FIG. 3. Normal mammary ductal outgrowth arising from end bud cultured in collagen gel matrix. A gland-free fat pad (De Ome et al, 1965) was implanted 15 weeks previously with an end bud cultured for 7 days. X12.

60

DEVELOPMENTAL BIOLOGY

like projections began to form. By 5 days the projections were well developed and displayed considerable variation in diameter and degree of branching (Fig. 2). The epithelial spikes displayed both blunt and pointed ends and were observed to branch through both bifurcation and budding. In some cases small lumens were observed (not pictured). In general these observations were comparable to those obtained by others (Richards et CCL,1982). After 7 days in culture, samples of collagen gel cultures were excised with scalpel and forceps and implanted into the mammary gland-free fat pads of young adult female mice. Normal glandular outgrowths were obtained in one out of four transplants (Fig. 3).

FIG. matrix (at) of (cc), a

VOLUME

104, 1984

Gmwth in Viva in Collagen gels Ice-cold unpolymerized collagen solution was drawn into a syringe and a 0.01-0.02 ml bubble was injected into the center of gland-free mammary fat pads of 3-month mice. Within 5 min at body temperature the collagen had polymerized, forming a gel that was firm enough to withstand gentle manipulation. And end bud, derived from either collagenase-dissociated gland (12 end buds transplanted) or by microdissection (14 end buds transplanted), was implanted into the center of each mass of polymerized collagen by means of jeweler’s forceps. Implants appeared to have little tendency to

4. (a) Whole mount of regenerated end bud appearing 6 days after implantation of a collagenase-isolated end bud into collagen gel in tiwo. The newly formed end bud (solid arrow) has appeared at the border (open arrows and dashed lines) between adipose tissue the host fat pad and the collagen gel implant (cg). X87. (b) Longitudinal section through the end bud pictured in Fig. 4a. Cap cells unique structural feature of end buds, are present at the leading edge (solid arrow). X525.

DANIEL ET AL.

Mouse

escape through the channel created by the forceps, and most transplants could be located at necropsy. When implanted fat pads were removed at 4 or 6 days the volume of the injected collagen was much reduced and the collagen no longer formed a conspicuous swelling, probably due to removal of water. In whole mount preparations the injected collagen was visible as an amorphous region in which the transplanted end bud could be identified as a darkly stained mass. No details of cellular outgrowths could be seen in these whole mounts, but in several cases new end buds had regenerated (Fig. 4a). In each instance where new end buds had formed, they appeared in adipose tissue surrounding the injected collagen (Fig. 4b); in no case did an end bud form within collagen. In sectioned material the injected collagen was found to be heavily infiltrated with cells from the surrounding fat pad (Figs 4b, 5a, b). These cells appeared to be of fibroblastic or mesenchymal type, and unusual numbers of neutrophils or other reactive cell types were not ob-

Mammary

Ep’thelium

61

served. The implanted end buds were easily identified, but had lost much of their characteristic structure, becoming irregular in shape, often with scalloped edges. Epithelial spikes frequently radiated out of the epithelial cell mass into the surrounding collagen (Figs. 5a, b). Typically, these spikes were smaller in diameter than normal mammary ducts (Table 1) and were simpler in construction. Lumens were sometimes observed close to the implant (not shown), but usually the projections consisted of solid cords of epithelial cells one to three cells in thickness (Figs. 5a, b). Alcian blue staining was irregular, and in many cases spikes lacked a discernible basement membrane. Epithelial spikes always grew parallel to the direction of alignment of collagen fibers. Because of the cellularity of the collagen matrix it was difficult to follow the very fine epithelial spikes in serial section, but where this could be done their length was found to be from 0.25 to 1.2 mm. No differences were observed between end buds obtained by microdissection and those isolated using collagenase.

FIG. 5. Epithelial spikes arising in tivo from a collagenase-isolated end bud grown for 6 days in implanted collagen gel in a mammary gland-free No. 4 mammary fat pad. The spikes are slender and contain no apparent lumens (open arrows). Study of serial sections revealed that isolated spike segments (Fig. 5b, open arrow) were connected to the main body of the end bud forming a three-dimensional array. Spikes always appeared to align themselves in parallel with the general pattern formed by implanted collagen fibrils. The collagen implants were extensively invaded by host cells. The end buds (solid arrows) usually lost their characteristic morphology and the lumens were often obliterated. X330.

62

DEVELOPMENTAL BIOLOGY TABLE

RELATIVE

DIAMETERS

1

OF DUCTS AND SPIKES IN Vrvo No.

Tissue

observations

Ducts in fat pad Spikes in collagen gel implants in viva Spikes in host collagen in vivo

Spikes in in vitro Spikes in in vitro

collagen gel 6 days collagen gel 15 days

Regeneration

of Transplants

AND IN VITRO

Diameters (mm X 10e3) + SD

15

30.0 z!z10.8

28

12.8 f

20

11.0 rt 3.1

18

18.0 f

23

24.8 + 8.7

in Untreated

3.5

5.6

Fat Pads

It was of interest to determine if spike-like outgrowths might also occur in the gland-free mammary fat pad without the addition of exogenous collagen. End buds and fragments of ducts were transplanted into the center

VOLUME

104, 1984

of No. 4 fat pads which had been previously cleared of the mammary rudiment. By examining sections of 4and 6-day transplants it was occasionally possible to find examples where the cut end of the duct happened to lie against one of the regions of fibrous collagen which are interspersed through the fat pad. In these instances the epithelium at the cut ends sent out epithelial spikes which penetrated the collagen in much the same manner as observed with injected collagen (Figs. 6, 7). Where the cut ends bordered on adipose tissue, end buds formed directly. DISCUSSION

We have investigated the question of whether spikelike outgrowths arising from mammary tissue cultured in collagen gel matrix are entirely a result of conditions of culture, or if there might exist a comparable growth pattern in viva which could aid in their analysis and interpretation. The results of these experiments are summarized in Fig. 8. Mammary tissue transplanted into gland-free mammary fat pads regenerated thin ep-

FIG. 6. Epithelial spikes arising from cut end of duct segment 72 hr after implantation into a cleared mammary fat pad. The implanted cells are surrounded by a fibrous matrix which stains for collagen with Massons trichrome. Spikes are one to two cells in thickness, have no observable lumen, and become narrow toward the tip (arrows). X658. FIG. 7. Epithelial spike arising 48 hr after implantation into a cleared mammary fat pad (arrows). The dense collagenous sheath which surrounds the mature duct is intact (upper left). The tip of the spike is beginning to broaden as it makes contact with surrounding adipocytes. X658.

DANIEL

ET AL.

Mouse

Mammary

EpithAum

63

FIG. 8. Summary figure depicting regeneration of mammary end bud placed in the center of a mass of collagen injected into a gland-free No. 4 mammary fat pad. (Left) Day 0: the end bud epithelium is in direct contact with collagen fibers formed by polymerization of injected collagen. The short duct segment remains firmly attached to its tunic of fibrous connective tissue. (Center) Day 4: epithelial projections arise at the end bud tip and at the cut end of the duct, where the basement membrane is interrupted. A branching three-dimensional outgrowth is formed which aligns with the general orientation of the collagen fibrils of the matrix. Projections do not occur along the sides of the duct segment, where the basement membrane and surrounding extra cellular matrix remain intact. (Right) Day 6: end buds begin to appear where epithelial outgrowths contact adipose tissue of the host fat pad.

ithelial spikes, similar in form and size to those seen in culture, when implanted epithelium was embedded in a collagen matrix. This collagen could be exogenous, obtained by injection of unpolymerized type I collagen, or could be a normal component of the host fat pad. Where mammary cells were in contact with fibrous collagen, spike formation was observed, and when mammary epithelium encountered regions of adipose tissue, end buds were formed. The appearance of radiating spikes in collagen gels in vivo indicates that this growth habit is a direct response to contact with collagen and is only secondarily influenced by the chemical environment. This is consistent with the results obtained by Ormerod and Rudland (1982), who found that the maximum length and diameter of structures formed in culture by Rama 25 cells was related to the concentration of collagen used in forming the gel, and by Foster et al. (1983), who found that tubules were formed only if the collagen gel was released from contact with the culture dish and allowed to float. If the growth pattern of mammary epithelium cells is to be attributed to the presence of a collagenous matrix, an explanation is needed for the observation that in our present experiments cellular outgrowths arose only from end bud tips or from the cut ends of ducts, and never from the lateral surfaces of ducts, even though

they are encased in previously synthesized collagen. A plausible interpretation is that spikes arise only where the basal lamina is thin or disrupted. We observed a reduction in end bud alcianophilia following implantation into collagen (Fig. 5), and at the cut end of ducts the lamina was mechanically damaged; the lateral sides of ducts, in contrast, displayed intense alcian blue staining. This interpretation is consistent with the observation that the basal lamina of mammary ducts is rich in sulfated glycosaminoglycans (Silberstein and Daniel, 1982), which in other systems have been associated with cell differentiation and tissue stabilization (Flint, 1972; Trelstad et aL, 1974). Once formed, mammary spikes in collagen gel implants showed little alcianophilia (Fig. 5), and thus may resemble spikes in culture which at the ultrastructural level displayed no basal lamina (Foster et aL, 1983) or, at best, a patchy and incomplete lamina (Richards et ak, 1982). Hoshino (1976) has reported that when mammary ductal fragments are implanted in gland-free mammary fat pads the basal lamina was damaged and the ductal epithelial cells migrated into the surrounding adipose tissue, forming isolated clumps which then regenerated ductal outgrowths. We have found that the basement membrane and the fibrous stroma surrounding ductal implants remains apparently intact except, as indicated above, at the end bud tip and at the severed ends of

64

DEVELOPMENTAL BIOLOGY

VOLUME 104, 1984

lagen in connective tissue remodelling. J. Em&& Exp. Morphol. ducts. We are in agreement, however, with other reports 27,481-495. indicating the importance of epithelium-adipocyte inC. S., SMITH, C. A., DINSDALE, E. A., MONAGHAN, P., and teractions in ductal morphogenesis (Hoshino, 1967; Sil- FOSTER, NEVILLE, A. M. (1983). Humman mammary gland morphogenesis berstein and Daniel, 1982; Sakakura et aL, 1979). in vitro: The growth and differentiation of normal breast epithelium Our results indicate that spike or tubule formation in collagen gel cultures defined by electron microscopy, monoclonal antibodies, and autoradiography. Dev. Biol 96,19’7-216. occurs when mammary cells are free to interact with a GROBSTEIN,C. (1967). Mechanisms of organogenetic tissue interaction. collagen matrix either in viva or in vitro, and suggest Natl. Cancer Inst. Mmogr. 26,279+X99. that the formation of radiating spikes and tubules rep- HOSHINO, M. D. (1962). Morphogenesis and growth potentiality of resents a comparatively simple and perhaps primitive mammary glands in mice. I. Transplantability and growth potentype of morphogenesis. The observation that spikes are tiality of mammary tissue of virgin mice. J. Nat1 Cancer Inst. 29, 835-849. seen in gel culture regardless of whether the tissue of LAWLER, E. M., MILLER, F. R., and HEPPNER, G. H. (1983). Significance origin is mammary end bud (Richards et GA, 1982), alveoli of three-dimensional growth patterns of mammary tissues in col(Yang et aZ., 1980b), or even tumor (Yang et al, 1980a), lagen gels. In Vitro 19, 600-610. suggests that spikes or simple tubules provide a therMICHALOPOULOS, G., and PITOT, H. C. (1975). Primary culture of pamodynamically stable configuration for a variety of eprenchymal liver cells on collagen membranes. Exp. Cell. Res. 94,7078. ithelial cell types, perhaps along the lines suggested by ORMEROD, J. E., and RUDLAND, P. S. (1892). Mammary gland morBennett (1980) for tube formation. phogenesis in vitro: formation of branched tubules in collagen gels The comparative simplicity of this interaction between by a cloned rat mammary cell line. Da. Bid 91,369-375. epithelium and matrix is apparent from the observations PASCO, D., QUAN, A., and NANDI, S. (1982). Effects of hormones and that epithelial spikes are obtained from cloned lines of EGF on proliferation of rat mammary epithelium enriched for alveoli. Exp. Cell Res. 141.313-324. rat mammary cells, indicating that the participation of RICHARDS,J., GUZMAN, R., KONRAD, M., YANG, J., and NANDI, S. (1982). mesenchyme is not required (Bennett, 1980; Ormerod Growth of mouse mammary gland end buds cultured in a collagen and Rudland, 1982). In contrast, the formation of more gel matrix. Exp. Cell Res. 141, 433-443. complex, organ-specific structures is generally associated SAKAKURA, T., NISHIZUKA, Y., and DAWE, C. J. (1979). Capacity of with tissue interactions (Grobstein, 1967); in the case mammary fat pads of adult C3H/HeMs mice to interact morphogenetically with fetal mammary epithelium. J. Notl. Cancer Inst. of mammary ductal morphogenesis in the young mouse, 63,733-736. this appears to involve epithelium-adipocyte contact SILBERSTEIN, G. B., and DANIEL, C. W. (1982). Glycosaminoglycans in with perhaps the additional participation of mesenchythe basal lamina and extracellular matrix of the developing mouse ma1 or fibroblastic cell types, leading to the formation mammary duct. Dev. Bid SO, 215-222. of end buds. TONELLI, Q. J., and SOROF,S. (1982). Induction of biochemical differWe thank Linda Wilson for technical assistance and for preparing illustrations. This research was supported by PHS Grant AG 01050 from the National Institute on Aging.

REFERENCES BENNETT, D. C. (1980). Morphogenesis of branching tubules in cultures of cloned mammary epitbelial cells. Nature (London) 285,657-659. BERNFIELD,M. R., and BANERJEE,S. D. (1972). Acid mucopolysaccharide (glycosaminoglycan) at the epithelial-mesenchyme interface of mouse embryo salivary glands. J. Cell Bid 52,664-673. DANIEL, C. W., and DEOME, K. B. (1965). Growth of mouse mammary gland in wivo after monolayer culture. Science 149, 634-636. DEOME, K. B., FAULKIN, L. J., BERN, H. A., and BLAIR, P. B. (1959). Development of mammary tumors from hyperplastic alveolar nodules transplanted into gland-free mammary fat pads of female C3H mice. Cancer Res. 19, 515-520. EMERMAN, J. T., and PITELKA, D. R. (1977). Maintenance and induction of morphological differentiation in dissociated mammary epithelium on floating collagen membranes. In Vitro 13,316-328. FLINT, M. (1972). Interrelationships of mucopolysaccharides and col-

entiation in three dimensional collagen cultures of mammary epithelial cells from virgin mice. DQ%wntiation 22,195-200. TRELSTAD, R. L., HAYASHI, K., and TOOLE, B. P. (1974). Epithelial collagen and glycosaminoglycans in the embryonic cornea: Molecular order and morphogenesis in the basement membrane. J. Cell Bid 62,815~830. WILLIAMS, J. M., and DANIEL, C. W. (1983). Mammary ductal elongation: Differentiation of myoepithelium and baaal lamina during branching morphogenesis. Dev. Bid 97,274~290. YANG, J., GUZMAN, R., RICHARDS, J., IMAGAWA, W., MCCORMICK, K., and NANDI, S. (1980a). Growth factor- and cyclic nucleotide-induced proliferation of normal and malignant mammary epithelial cells in primary culture. Eru&wrimlogy 10’7,35-41. YANG, J., GUZMAN, R., RICHARDS, J., and NANDI, S. (1980b). Primary culture of mouse mammary tumor epithelial cells embedded in collagen gels. In Vitro 16, 502-506. YANG, J., RICHARDS, J., GUZMAN, R., IMAGAWA, W., and NANDI, S. (198Oc). Sustained growth in primary culture of normal mammary epitbelial cells embedded in collagen gels. Proc Nat1 Acd Sci USA 77,2088-2092.

YANG, J., GUZMAN, R., RICHARDS, J., JENTOFT, V., DEVAULT, M. R., WELLS, S. R., and NANDI, S. (1980d). Primary culture of human mammary epithelial cells embedded in collagen gels. J. NatL Cancer. Inst.

65,337-343.