Differentiation of mammary epithelial stem cells to alveolar-like cells in culture: Cellular pathways and kinetics of the conversion process

Differentiation of mammary epithelial stem cells to alveolar-like cells in culture: Cellular pathways and kinetics of the conversion process

DEVELOPMENTAL BIOLOGY 107, 301-313 (19%) Differentiation of Mammary Epithelial Stem Cells to Alveolar-like Cells in Culture: Cellular Pathways and...

4MB Sizes 0 Downloads 13 Views



107, 301-313


Differentiation of Mammary Epithelial Stem Cells to Alveolar-like Cells in Culture: Cellular Pathways and Kinetics of the Conversion Process FRIEDA C. PATERSON, MICHAEL J. WARBURTON, Ludwig





(London Branch), Royal







in revised form September

29, 1984; accepted







10, 1984

The cuboidal epithelial stem cell line Rat Mammary (Rama) 25 can differentiate in culture to droplet, alveolar-like cells that form domes, secrete small amounts of casein, and bind peanut lectin after treatment with neuraminidase. Differentiation to droplet cells is accelerated by dimethyl sulfoxide (DMSO). Morphologically intermediate states (gray and dark) which occur in the order: cuboidal - gray - dark - dark droplet - doming cells have been identified along this pathway by time-lapse cinematography. The dark and dark droplet states are associated with increased peanut lectin binding capacity whereas casein is secreted mainly by cells in domes. Cells in cultures containing low concentrations of DMSO (~56 m&f) acquire droplets predominantly in the dark state, whereas with higher concentrations of DMSO droplet formation is seen mainly in the gray state. Kinetic analysis both from time-lapse films and conventional microscopy, shows that increasing the concentration of DMSO prolongs the time spent in the gray state, decreases the time of initial appearance of droplet cells, and increases their subsequent rate of formation, without detectable effects on the rates of the remaining morphological transitions. DMSO also reduces the average rate of DNA synthesis and increases the average cell cycle time, particularly in the second (and subsequent) cell cycles after its addition. However, neither droplet nor doming cells are terminally differentiated. Thus a linear sequence of morphological states exists between the Rama 25 stem cells and the alveolar-like or more probably alveolar bud cells in vitro, and DMSO accelerates the overall conversion predominantly by truncating one of the steps in this pathway. 0 1985 Academic Press. Inc. INTRODUCTION

The neonatal mammary gland is an epithelial tissue which is initially composed of hollow tubes or ducts of relatively uniform ductal epithelial cells. The neonatal tissue then undergoes differentiation processes, initially leading to the formation of myoepithelial cells and then to alveolar cells under the correct hormonal stimulation, both processes occurring after the development of the major organs of the animal. Based on histological and ultrastructural evidence, it has been suggested (Rudland et al, 1980a) that stem cells exist (Radnor, 1972) at ductal termini (Williams and Daniel, 1983; Ormerod and Rudland, 1984) of young virgin rodents that are capable of giving rise to both ductal epithelial and myoepithelial cells, and eventually to alveolar cells in the mature gland of pregnant and lactating animals (Russo et al, 1982; Chepko, 1984). This suggestion has been based on the isolation of a duct-like (Ormerod and Rudland, 1982) epithelial cell line from a carcinogen-induced rat mammary tumor that can differentiate in culture to myoepithelial-like or alveolar-like cells (Bennett et al, 1978). Much work has been done in the past to elucidate the effects of various mammotrophic hormones on the production of alveolar cells and milk products in organ and in tissue cultures (Elias, 1957; Lasfargues and 301

Ozzello, 1958; Raynaud, 1961; Turkington et al, 1967). However, one problem with the use of both these systems for studying processes of differentiation is that they are composed of a mixture of cell types, both stromal and epithelial, and the epithelial cells tend to die out after more than one subculture (Elias, 1957). A more detailed analysis of the cellular and biochemical events involved in differentiation of the mammary gland requires a system in which a homogeneous population of cells can be triggered to differentiate in a synchronous manner. With this end in view, a triply single-cell-cloned epitheiial cell line Rama 25l has been isolated which, under the appropriate conditions is initially homogeneous and can be triggered to produce alveolar-like cells in a relatively synchronous way. This cell line has been obtained from a DMBA-induced mammary epithelial tumor which contains both myo’ Abbreviations used: AB, alveolar buds; CCT, cell cycle time; DEM, Dulbecco’s modified Eagle’s medium; DMBA, dimethylbengalanthracene; DMSO, dimethyl sulfoxide; E, estradiol; FCS, fetal calf serum; HC, hydrocortisone; 4-hr-LI, percentage of cell nuclei labeled with [‘Hlthymidine in 4 hr; I, insulin; PBS, Dulbecco’s phosphate-buffered saline, Ca’+- and Me-free; Prl, prolactin; Rama 25, Rat mammary cell line clone 25; RM, routine medium; RM(D), routine medium for doming containing the standard concentrations of DMSO; TD, terminal duct; TEB, terminal end bud; WB, washing buffer. 0012-1606/85 $3.66 Copyright All rights

0 1986 by Academic Press. Inc. of reproduction in any form reserved.




epithelial-like cells and, after mating the animals, alveolar-like cells (Young and Hallowes, 1973; Supowit and Rosen, 1982). Morphologically similar cell lines have also been obtained from rudimentary rat mammary glands (Rudland et a& 1980b; Ormerod, 1983) showing that cells similar to Rama 25 occur in normal mammary tissues. Rama 25 cells grown on plastic resemble a low cuboidal epithelium, henceforward termed cuboidal cells. Although these lines were single-cell cloned and when plated initially consisted of only cuboidal cells, they can still give rise to elongated, myoepithelial-like cells at very low frequencies under the appropriate conditions (Bennett et ab, 1978; Rudland et al, 1980a; Ormerod and Rudland, 1982). Under different conditions dense cultures of cuboidal cells form groups of small, dark, polygonal cells with small vacuoles or “droplets” at their peripheries. These have been termed droplet cells. Conversion of an initially homogeneous culture of cuboidal cells to droplet cells can be accelerated in a relatively synchronous manner with agents which stimulate differentiation of Friend erythroleukemic cells, notably DMSO (Friend et al, 19’71), or retinoic acid in the presence of the mammotrophic hormones Prl, E, HC, and I (Bennett et al, 1978; Rudland et al, 1983). These droplet cell cultures also contain hemispherical blisters or domes within the cell monolayer (McGrath, 1975), synthezise increased amounts (20-40 fold) of immunoreactive casein which has been authenticated as the 42 kDa component present in rat milk by peptide mapping techniques (Warburton et aL, 1983), and demonstrate increased staining with fluorescently labeled peanut lectin (Newman et al, 1979). Based on the above criteria these cultures have been classified as alveolar-like cells (Rudland et aL, 1980a). Since conversion of the stem cells to droplet cells/doming cultures is fully reversible by removing the DMSO, the droplet cell population cannot be isolated by simple cell-cloning techniques. Cellular conversions which may occur along the alveolar-like pathway in culture can, therefore, be followed only by using time-lapse cinematography. In this paper we describe the identification of different morphological forms of epithelial cell and their interconnecting pathways, the kinetics of these interconversions, and the corresponding changes in cellular proliferation rates using time-lapse and conventional microscopic analysis for different concentrations of DMSO. MATERIALS



Tissue Culture Rama 25 cells were isolated, grown, and passaged in DEM, 10% FCS, 50 rig/ml HC, and 50 rig/ml I (RM) at


107, 1985

37°C in an atmosphere of 10% COz, 90% air as previously described (Bennett et aL, 1978). For obtaining droplet cells/domes in confluent Rama 25 cultures the cells were plated at 105/5-cm petri dish in 5 ml of RM and grown until confluent. After a further 2 days fresh RM, 500 rig/ml Prl, 2.5 rig/ml E were added and the cells incubated for a further 2 days. Medium was then replaced with RM, 500 rig/ml Prl, 2.5 rig/ml E, and different, henceforward termed standard concentrations of DMSO (0, 56, 112, 167, 223 mM) were added as indicated (RM(D)) (Rudland et al, 1982). The number of domes per 14 mm2 microscopic field were counted from the time of addition of DMSO, using an Olympus CK inverted microscope at a magnification of x40. Results were the average of four fields per dish from a minimum of two dishes. For the fixed and random microscopic field experiments cultures were set up in triplicate 5-cm dishes, grown to confluency, and the hormones and the standard concentrations of DMSO added as above. For fixed fields, four small areas were marked on the bottom of the petri dishes and constantly examined. For random fields four fields per dish were selected each time at random. Cultures were examined with a Reichert Biostar phase-contrast inverted microscope (Xl00 magnification), and cells were classified according to their morphological form, a minimum of 100 cells being observed. Similarly the number of domes in the four fixed or four random fields were counted as above using X40 magnification. Triplicate dishes for each concentration of DMSO were scored and the average results were recorded. The zero time corresponded to the time of addition of DMSO. Photo- and ImmunoJluorescence


Phase-contrast pictures of living cells were taken on a Zeiss ICM 405 automatic inverted microscope fitted with a green filter on Ilford Pan F film. Immunofluorescence microscopy was performed by growing cells on plastic coverslips (Lux, purchased from Flow Laboratories, UK) and after the appropriate culture manipulations the medium was removed, and coverslips were washed twice with PBS. For binding to peanut lectin (Newman et ab, 1979) coverslips were incubated first with 50 ~1 of 0.1 units/ml neuraminidase (Behring, Marburg, West Germany) in 0.2 M sodium acetate, pH 5.5, for 30 min at 37°C before washing 5X with PBS. For binding to anticasein serum (Bennett et aZ., 1978) cells were fixed in 3.7% (w/v) formaldehyde in PBS followed by methanol at -2O”C, and then airdried. Coverslips were then incubated with 50 ~1 of either 20 pg/ml FITC-conjugated peanut lectin (Sigma, Poole, UK) or l/10 diluted rabbit anticasein serum





(Warburton et al, 1983) for 30 min at room temperature or 37”C, respectively. Coverslips were then washed 5X with PBS, the peanut-treated ones mounted in Univert mountant (Gurr, Poole, UK), and the anticasein-treated ones incubated with 50 ~1 of l/5 diluted FITC-conjugated goat anti-rabbit IgG (Nordic Immunological Laboratories) for 30 min at 37°C. The latter coverslips were washed 5X before mounting, and viewed in a Reichert Polyvar microscope with epifluorescence optics and an FITC B4 filter block. Photographs were recorded on Ilford XPl 400 film, those in phase contrast were recorded in green light. Inclusion of 2% (w/v) galactose with FITC-peanut lectin or prior absorption of casein antiserum with 1 mg/ml purified casein for 3 hr at 37°C (Bennett et a& 1978) abolished completely any fluorescence observed. Time-Lapse


Stem Cell LXffwentiaticm


grown and filmed as for Set 2 above, except that fresh RM, Prl, E, and the standard concentrations of DMSO were added to the washed cells. Filming was again commenced immediately, one film per DMSO concentration being recorded and each film lasted 50-60 hr. In Set 1 cell genealogies or “family-trees” were constructed by following 20 cells, their daughters, and granddaughters through successive cell divisions. The intermitotic times and times to change from one morphological form to another were recorded; the total number of cells observed per film varied from 100 to 185. In Sets 2 and 3 of confluent cultures, 200 cells per film were followed, their time to divide and the times when changes occurred in their morphological forms were recorded. In all experiments time zero corresponded to the time of addition of DMSO and the start of the film.

Cinematography Biochemical

Films were recorded with a Bolex movie camera on 16-mm black and white Recordak (Kodak) film at 200 ASA, using an Olympus inverted IMT microscope fitted with phase-contrast optics and a green filter, at a magnification of ~12.5 (field width: 800 pm). The Bolex camera was controlled by an Olympus Intervalometer taking 1-set exposures every 2 min, except for the 6-min intervals of films with 0,112, and 167 mM DMSO in Set 2 (see below). Cells were filmed in a 3-cm petri dish in an enclosed plastic chamber which was supplied with water-saturated gases of 15% C02/85% air at 37°C. Negatives and positives of the 16-mm film were viewed with a LW Photo-optic Analyzer (l-in. lens), Model 224A Mark V (LW International, Calif.), and a Gordon A-V frame counter. Film-to-screen distance was 165 cm, giving a film magnification of X115, and the cells’ positions were recorded on tracing paper (Riddle, 1979). Three different sets of films of Rama 25 cells were made: (Set 1) Rama 25 cells were plated in RM for 24 hr and the media changed to RM(D) containing the standard concentrations of DMSO (0, 56, 112, 167, 223 mM). Filming then commenced immediately after addition of DMSO, and continued for the next 5.5 days, one film per concentration of DMSO. (Set 2) Rama 25 cells were grown to confluency to obtain droplet cells/ doming cultures as described in Tissue Culture. After 2 days in RM, Prl, and E, the medium was removed, filtered, and a standard concentration of DMSO was added as above. At the same time the cell monolayer was washed twice with PBS. Filtered medium with the standard concentration of DMSO was then added to the cells, and filming commenced. Each film lasted a minimum of 50-60 hr, one film for each concentration of DMSO being recorded. (Set 3) Rama 25 cells were


For binding of ‘?-labeled peanut lectin or concanavalin A to Rama 25 cells, triplicate cultures were seeded and grown in 1 ml of media in l-cm-diameter multiwell plates (see Tissue Culture). On termination of the experiment the medium was removed, and the cell monolayers were first incubated with neuraminidase prior to peanut binding (see Photo- and Immunofluorescence microscopy). All monolayers were then washed twice with WB: DEM, 0.01 MHepes, pH 7.4,0.1% (w/v) bovine serum albumin. Cells were then incubated for 30 min at room temperature with 200 ~1 ‘%I-peanut lectin or I’25-concanavalin A (100,000 cpm approx, sp act 1.1 &i or 0.6 &i/pg, respectively) which had been iodinated by the chloramine-T method (Hunter and Greenwood, 1962) in WB. The cell monolayers were then washed 5X with WB, detached in 1 ml of 0.25% (w/v) trypsin in PBS for 18 hr at 37”C, and 0.9 ml counted in a Packard Multi Prias 4 gamma counter (65% efficiency). Protein content of parallel wells was estimated by the method of Lowry et al. (1951). Inclusion of 2% (w/v) galactose or a-methyl mannoside with the 1251-peanut lectin or concanavalin A, respectively, depressed the binding to background levels (~5% of specific binding, P. S. Rudland, A. Twiston Davies, and S. Jamieson, unpublished results). Casein was determined as described previously (Warburton et aL, 1983). The average amount of casein secreted from three cultures in 5-cm diameter dishes, was recorded +SEM. The protein content of the confluent cultures was roughly constant within 30% under the conditions of the experiment (Table 5). For radioautography, parallel cultures to those described in Tissue Culture were set up and radioactively labeled with 3 &i/ml of [3H]thymidine at 1 PM for




4-hr pulses at the appropriate times. Separate cultures were also continuously labeled with rH]thymidine as above. At the termination of each labeling time, the medium was removed and the dish was processed for radioautography as described previously (Rudland et aL, 1977). r3H]thymidine-labeled cell nuclei were scored using a Dynascope at X40 magnification. The average percentage of labeled cells from four randomly chosen fields *SEM was recorded. The average cellular DNA content per confluent culture was obtained as described previously (Jimenez de Asua et al, 1977). Measurements of cell number with a Coulter counter and protein content per culture were carried out in parallel dishes using the methods of Rudland et al (1982) and Lowry et al. (1951), respectively. RESULTS

Identification of &&rent Morphological and Their Interconnecting Pathways

Cell Types

From phase microscopy studies, intermediate cells along the cuboidal cell (Fig. 1A) to droplet cell/doming (alveolar-like cell) (Fig. 1B) pathway had already been provisionally identified in Rama 25 cultures, and in primary cultures of normal and tumorous rat mammary glands (Bennett et ab, 1978; Rudland et aC, 1980a). Two intermediate classes were identified by their apparent contrast as viewed under the phase-contrast microscope-gray and dark cells (Fig. 1C). Whereas in cuboidal cells and dark cells the nuclei were the darkest part of the cell, gray cells had nuclei of the same contrast or slightly paler than the cytoplasm and possessed very indistinct cell borders. The dark cells were polygonal in shape and like the cuboidal cells had distinct cellular peripheries (Fig. 1C). Both gray and dark cells appeared often in circular patches in locally confluent cultures. In more dense cultures roughly circular patches of gray and dark cells with pale beads or droplets around their margins could be identified (Figs. lB, C), and these patches frequently contained domes or hemispherical blisters formed by local detachment of the cell monolayer (Fig. 1B) (McGrath, 1975). At the ultrastructural level the increased density of the cytoplasm in gray and dark cells observed in phase contrast correlated with increased numbers of cell organelles, particularly rough endoplasmic reticulum and Golgi, and the droplet cells contained numerous large intra- and (usually) intercellular vacuoles which generally lacked osmiophilic or electron-dense contents (Bennett et a& 1978). Cells with intermediate appearances between cuboidal cells, gray cells, dark cells, and droplet-containing cells were also visible. However, each of the different morphological forms viz. cuboidal,


107, 1985

gray, gray plus droplets, dark, dark plus droplets, and domes were readily identifiable in all the time-lapse films of growing cultures of Rama 25, as they approached confluency and prior to doming (Set 1). Large droplets were also seen in droplet cells (Fig. 1C) at a low frequency, but they did not always precede doming. FITC-conjugated peanut lectin bound to the surface of dark cells whether or not they contained droplets, but only after removal of the terminal sialic acid by treating the cells with neuraminidase (P. S. Rudland, A. Twiston Davies, and S. Jamieson, unpublished results; Newman et al., 1979). No detectable fluorescence was observed on the cuboidal or gray cells (Figs. lE, F). Specific membrane fluorescence due to FITC-conjugated concanavalin A was not detected in any of the different cellular morphologies of Rama 25 (P. S. Rudland, A. Twiston Davies, and S. Jamieson, unpublished results). In droplet cell cultures, antisera to casein stained only the cells in domes (Fig. lD), no specific staining was observed in cultures containing only cuboidal, gray, or dark cells with or without droplets (P. S. Rudland, A. Twiston Davies and S. Jamieson, unpublished results; Bennett et al., 1978). From a complete analysis of all the cellular pathways observed by time-lapse cinematography in Sets 1, 2, and 3 (see Materials and Methods, Time-Lapse Cinematography section) for the standard concentrations of DMSO, two major pathways, or rather attenuated sequences of the same pathway were adopted (Table 1). The major difference was the morphological state (gray or dark) in which the differentiating cells first acquired their droplets. Thus in the zero and the lowdose (56 m&f) DMSO-containing cultures, droplet acquisition occurred predominantly in the dark state, (cuboidal - gray - dark - dark + droplets), whereas with the higher DMSO concentrations (112, 167, 223 mM) the majority of the cells acquired their droplets in the gray state (cuboidal - gray - gray + droplets - dark + droplets) (Table 1). Some rare, minor pathways appeared to be unique for a particular concentration of DMSO, although their statistical significance is open to question (Paterson, 1983). Thus DMSO may have accelerated the conversion process by forwarding the cells’ morphological state in which the droplets were first acquired, as shown in Fig. 2. Kinetics of Cellular Interconversion of Sparse& Plated Cells Using Time-Lapse Cinematography Sparsely plated cuboidal cells were followed by timelapse cinematography (Set 1) for the standard concentrations of DMSO and analyzed for their different morphological forms: cuboidal, gray, dark, droplet. Unless otherwise stated the zero time referred to the

FIG. 1. Light and fluorescent micrographs of different cell types. In (A-C) living cells were photographed using phase-contrast optics. (A) Cuboidal Rama 25 cells. (B) Confluent Rama 25 cultures after treatment with 250 mM DMSO and all the hormones. DD, dark droplet cell; GD, gray droplet cell; DO, dome. (C) Rama 25 cells after several days at confluency without DMSO, Prl, and E. C, cuboidal cell; G, gray cell; D, dark cell; DC, droplet cell; VC, dark cell with large vacuole or droplet; I, more elongated cell. Bars for (A-C) = 100 pm, X28. In (D-F) confluent cultures of Rama 25 were treated with 250 mM DMSO and all the hormones for ‘72 hr. (D) Fixed cultures fluorescently stained with anticasein serum, the micrograph is of a collapsed dome. Only cells associated with the dome stain. Bar = 20 pm, ~144. (E) Droplet cell-containing area desialylated and fluorescently stained with peanut lectin. (F) Phase-contrast micrograph of the same area showing D, dark cell; G, gray cell; and C, cuboidal cell. Only those cells corresponding to the dark cells are stained in (E). Bars for (E, F) = 50 pm, X81. 305





TABLE 1 ACQUISITION INTHE INTERMEDIATE~TATESWITH DIFFERENT CONCENTRATIONS OF DMSO Percentage of cells which acquire droplets along the pathways depicted in Fig. 2”

Concentration of DMSO NM) 0 56 112 167 223



I+* 632 5 62 f 11 27f 3 13+ 3 4+ 2



l-+1* 29 13 66 79 74

+ f f + k

4 3 4 5 8

‘Results were the average + SD of the percentage from three independent sets of films for each standard concentration of DMSO. Set 1 were films of sparse Rama 25 cells filmed for g days. Set 2 were films of confluent Rama 25 cells with no fresh medium change and one of the standard concentrations of DMSO; and Set 3 were films as in Set 2 but with the addition of DMSO accompanied by a change of fresh medium (see Materials and Methods, Time-Lapse cinematography section). The collection of pathway data was achieved by following single, uninterrupted branches of the cell family trees from the start of the film to the end. The percentage of all possible pathways connecting the intermediate states were computed, but only the results for the two major pathways are shown. The remaining pathways were traversed by less than 5% of the cells (not shown), and hence the line totals do not add up to 100% (Paterson, 1983). * Symbols refer to Fig. 2.

time of addition of DMSO and the start of filming (see Materials and Methods, Time-Lapse Cinematography section). The cultures reached confluency (per field) in almost the same time in medium containing zero and low concentrations of DMSO (0 to 112 mM DMSO inclusive), but this time was increased by about 10 hr for cultures in 167 and 223 mM DMSO (Figs. 3A-E). The rate of loss of cuboidal cells as they converted to gray cells, and the rate of loss of gray cells as they converted to dark cells were virtually constant, independent of the concentration of DMSO (Table 2B, C). The maximum number of gray cells occurred before the complete disappearance of cuboidal cells (Figs. 3AE). The duration of the gray state in the presence of DMSO up to 167 mM was significantly greater than that without DMSO (Table 2A). The reduction at 223 mM to a time approaching that without DMSO may be due to a toxic effect at that concentration. The time of first appearance of droplet cells when measured from that of confluency occurred earlier, while times for 100% of the cells in the cultures to acquire droplets were reduced with increasing concentrations of DMSO (Figs. 3A-E). The maximum rate of formation of droplet cells was increased with increasing concentrations of DMSO, rising to fourfold with 223 mM DMSO (Table 2D). Thus, although addition of DMSO did not


107, 1985

drastically affect the rates at which the cells changed from cuboidal to dark cells via the gray cell state, it increased dramatically the time spent in the gray state and the rate of droplet cell formation. We next investigated whether droplets appeared preferentially in either of the intermediate states. Both gray and dark cells disappeared at the same time and were replaced by droplet-containing cells. The time of disappearance was dependent on the concentration of DMSO (Figs. 3A-E). The rate of formation of droplets in dark and in gray cells was the same for a given concentration of DMSO, when expressed as a fraction of dark cells (Table 2E) or of gray cells (Table 2F), respectively. However, the rate of loss of dark cells when calculated as a percentage of the total number of cells, was a minimum of 3~ the rate of loss of the equivalent gray cells in every case. The sum of both these rates-the loss of dark cells and the loss of gray cells-for a particular concentration of DMSO, was equal to the overall rate of formation of droplet cells for that DMSO concentration (not shown), since no other cell types were present (Figs. 3A, E). Thus, the droplets showed no preference for the dark or gray cell intermediate state. In parallel cultures the increase in peanut lectinbinding capacity of desialylated Rama 25 cells at the highest concentration of DMSO closely followed the appearance of the dark cells, whether or not they contained droplets (Figs. 3E, F). This increase preceded the increase in casein secretion by about 40 hr, and no similar increase was observed in the binding capacity for concanavalin A. However, a good proportion of droplet cells were still capable of cell division since 20-40s of them divided at every standard concentration of DMSO, with or without a medium change (not shown). Kinetics of Cellular Interumversim of Cmfluent Cultures Using Ccmventional Microscopic Analysis

The relationships among the increase rate of formation of dark droplet and gray droplet cells, doming DARK CELL










FIG. 2. Summary of the major pathways. The are described in Fig. 1. One route (-) represents cells in the control and 56 mM DMSO-containing other (-) the majority of cells in 112-223 mM cultures as described in Table 1.

various cell types the majority of cultures and the DMSO-containing







160 I


0 0LIe.4


- 40


Stem Cell merentiation







Time (hours)

Time (hours)



FIG. 3. Changes in morphological cell types analyzed by time-lapse cinematography. Nonconfluent Rama 25 cells were filmed 24 hr after plating for $ days, and the DMSO was added just prior to filming. (A) control 0 mM; (B) 56 mM; (C) 112 mM; (D) 167 rnM, (E) 223 mM DMSO (see Materials and Methods, Time-Lapse Cinematography Section, Set 1). The cells were subdivided into their morphological states (Fig. 1) and 1 film per DMSO concentration was analyzed. Cells became confluent in films A, B, C, D, and E, after 33, 34, 28, 43, and 42 hr, respectively, and droplet cells appeared after 80, 74, 54, 68, and 55 hr, respectively. Key: (- - - A - - -) cuboidal cells; (- A -) gray cells without droplets; (- l -) dark cells without droplets; (- - - 0 - - -) gray cells with and without droplets; (- - - @ - - -) dark cells with and without droplets; and (- *- n - * -) dark and gray droplet cells. (F) Binding capacity of Rama 25 cultures for iZ51-labeled peanut lectin, for concanavalin A, and casein secretion were measured in parallel l-cm wells for the lectin bindings or 5-cm dishes for casein determinations (see Materials and Methods, Biochemical Techniques section). Key: (- + -) percentage increase in binding of peanut lectin; (- 0 -) percentage increase in binding of concanavalin A, both in cultures containing 250 mlM DMSO; (- - - A - - -) ng casein secreted per dish without DMSO; and (-A -) ng casein secreted per dish with 250 mM DMSO. All results are the average f SEM from triplicate cultures, the binding at the start for ‘%I-peanut lectin and concanavalin A were 160 f 30 and 56 + 11 cpm/pg protein, and this remained constant in cultures with no DMSO.

frequencies, and the standard concentrations of DMSO were investigated further in the following two sorts of experiment, using data obtained completely indepenTABLE RATES



dent of that from time-lapse films. In contrast to the previous section DMSO was added to confluent rather than to sparse cultures of Rama 25, and the appearance 2





Rate of cell formation





Concentration of DMSO (mM)

Time in gray state’ (hr)



56 112 167 223

46 42 61 38

B Cuboidal cells*

C Grey cells*

D Droplet cells*

E Dark droplet cells”

F Gray droplet cells’

-6.5* -6.8* -6.9* -6.O* -6.4:

-2.4” -2.4* -2.P -2.5* -2.P

0.81* 1.6* 1.9s 3.1* 3.2*

0.15* 0.20* 0.20 0.45 0.63*

0.10* 0.19 0.18 0.50 0.62*

a Time in gray state was measured by the time between cross-over points of cuboidal cells/gray cells and gray cells/dark cells of Figs. 3A-E. The times are accurate to within 10%. *Rates were calculated using the linear regression equation from the plots of the percentage of different morphological forms of cell against time, and correspond to the maximum rate of change (slope) of each line drawn in Figs. 3A-E. Data were taken from the films of sparsely plated cells of Set 1 (see Materials and Methods, Time-Lapse Cinematography section). Data were also plotted as log of percentage of cells against time, but the fit to a straight line was no better than that shown above. ‘Rates in columns E, F are calculated from the number of dark droplet cells divided by the number of dark cells (E), or the number of gray droplet cells divided by the number of gray cells (F), plotted against time. * Correlation coefficient r 2 0.90. for the remainder 0.80 d T < 0.90.






Tin-Z bCUrs)

VOLUME 107, 1985


Time (hours)

Time (hours)





FIG. 4. Transitions of dark and grey cells to droplet cells and domes analyzed in a fixed microscopic field. Five different concentrations of DMSO: (A) 0 mM; (B) 56 mM; (C) 112 mM; (D) 167 mM; and (E) 223 mM were added to I-day confluent Rama 25 cells. One-hundred cells were scored for their different morphological forms in each of four fixed microscopic fields per dish in each of three dishes for each concentration of DMSO. Key: (- A - -) gray cells; (- - - A - - -) dark cells; (- *- 0 - * -) droplet cells; and (- - - n - - -) No. of domes per field. (F) Binding capacity of [‘SI]peanut lectin and casein secretion of parallel l-cm wells and 5-cm dishes, respectively, were recorded as in Fig. 3. Key: (- - - 0 - - -) percentage increase in binding of peanut lectin in cultures with 250 mM DMSO; and ng of casein secreted per dish with 0 mM (-V -), ‘75 mM (- A -), 150 mM (- n -), and 250 mM DMSO (- + -). All results are the average + SEM from triplicate cultures.

of different morphological forms of cell was investigated in random or fixed fields of a conventional microscope; time zero corresponded again to the time of addition of DMSO. Results from the fixed (Figs. 4A-E) and random (not shown) fields were similar to each other and to those obtained from the time-lapse studies of sparsely plated cells (Set 1). Thus the time to first appearance of droplet cells after the addition of DMSO decreased (Figs. 4A-E) and the rate of formation of droplet cells increased with increasing concentrations of DMSO (Table 3A). As in the time-lapse studies the rate of appearance of dark droplet cells and gray droplet cells was the same, for a given concentration of DMSO, when measured as a fraction of the total number of dark or gray cells, respectively (Table 3B, C). When measured from a fixed or random field of cells the rates of formation of droplet cells were virtually identical (Table 3A, D). However, for cultures without DMSO the rate of droplet cell formation obtained from microscopic analysis (0.2%/hr) was fourfold lower than that calculated from the timelapse films of the previous section (Set 1) (0.8%/hr). This discrepancy was probably due to the different experimental protocols employed (see Materials and

Methods). The top rates achieved with cultures containing 223 mM DMSO were, however, comparable. As in the time-lapse studies, the time from the first appearance of droplet cells to that for cultures containing 100% droplet cells decreased with increasing concentrations of DMSO, and both the gray cell and dark cell populations fell to zero at the same time (Figs. 4A-E). As reported in the previous section, increases in the binding capacity of the confluent cultures for peanut lectin at the high concentration of DMSO closely followed the total number of dark cells with or without droplets (Fig. 4F). At that concentration of DMSO, domes and casein secretion occurred after the increased binding capacity of the cultures for peanut lectin was observed. At every concentration of DMSO the droplet cell population had reached a minimum of 50% before doming and casein secretion increased significantly. Both doming and casein secretion increased with increasing concentrations of DMSO after a lag period of 8-10 hr, as previously reported (Warburton et aL, 1983). Both maxima were achieved after loo-120 hr in the DMSO-containing cultures (Figs. 4B-F). However, some cells in domes were still capable of cell division, since









Rate of droplet cell formation (%/hr) Random fields

Fixed fields

Concentration of DMSO MM) 0 56 112 167 223

A Total droplet cells”

B Dark droplet cells”

C Gray droplet cells’

D Total droplet cells*

0.19 0.93* 1.60* 3.9** 5.8*

0.033 0.11** 0.16** 0.24 0.60*

0.029 0.10** 0.15** 0.21 0.58**

0.18 NDd 1.68* 2.61 4.89*

a Rates were calculated using the linear regression equation from the plots of the percentage of droplet-containing cells against time and correspond to the maximum rate of change in Figs. 4A-E for fixed fields. Data were obtained from 100 cells per field from four fixed or four random microscopic fields from three separate dishes per DMSO concentration (see Materials and Methods, Tissue Culture section). The fit to a straight line for semi-log plots was no improvement on the linear plots as in Table 2. b Rates in columns A, D are calculated from the total droplet cells (gray droplet + dark droplet) divided by the total cells plotted against time. ‘Rates in columns B, C are calculated from the number of dark droplet cells divided by the number of dark cells (B) or the number of gray droplet cells divided by the number of gray cells (C) plotted against time as in Table 2. Random field results for the rate of dark droplet cell and gray droplet cell formation were virtually identical to the rates shown in columns B and C, respectively. dND = not determined. * Correlation coefficient T 2 90. ** 0.80 d r < 0.90; for the remainder 0.69 < r < 0.80.

up to 12 out of 40 cells in the time-lapse films of Sets 2, 3 were seen to divide while uplifted in domes (not shown). Thus addition of DMSO to confluent cultures caused an earlier first appearance and a faster rate of formation of droplet cells, as observed earlier with time-lapse studies of sparsely plated cells (Set 1). Moreover the droplet cells were always formed before domes and casein secretion became apparent. Effect of DMSO on the Rates of DNA Synthesis and Cell Division After addition of fresh medium to confluent cultures of Rama 25 there was a constant lag period of 8 f 2 hr before any rise in the 4 hr-LI was observed. The maximum 4 hr-LI occurred at 12-16 hr after additions, independent of the concentration of any added DMSO, as reported previously (Rudland et al, 1982; Warburton et al, 1983). Cultures containing DMSO had a slightly altered 4 hr-LI at that time. With fresh medium alone



Stem Cell Differentiation

a second maximum was seen at 62-66 hr, and this was reduced roughly in proportion to the concentration of DMSO present. There was a corresponding increase of about 35% in cell numbers, DNA and protein content, and the percentage of labeled nuclei from a continuous labeling experiment after cultures were exposed to fresh medium for 48 hr. This increase was only about 22% for cultures containing 223 mM DMSO (Table 4). To identify any change in the division time of confluent cultures with DMSO, time-lapse films of confluent Rama 25 cells were analyzed (Sets 2,3, under Materials and Methods). All cultures showed a steady decline with time in the number of cells dividing (Fig. 5). The culture with fresh medium alone showed a 15-hr lag before an increased number of cells started to divide. For cultures which contained DMSO this lag was not so sharply defined, although there was a tendency to shorter lag times with higher concentrations of DMSO (Fig. 5). The culture with fresh medium alone had the highest number of cells dividing within the subsequent lo- to 30-hr period (Table 5). All cultures containing DMSO had 2-3X fewer cells dividing in the lo- to 30hr period and 5-15X fewer divisions between 30 and 50 hr. The rate of cell division during the 30- to 50-hr period decreased roughly in proportion to the concentration of DMSO (Table 5). To obtain CCTs of growing cultures, the films of Set 1 of sparsely plated Rama 25 cells were utilized (see Materials and Methods). In the absence of DMSO the CCTs of Rama 25 cells remained relatively constant



Cell No./ dish” (x10-6)

DNA/ dish” bd

Protein/ dishb ha

Labeled nuclei’ (W)

Time zero

5.2 + 0.3

91 * 10



48 hr With fresh medium

7.2 f

Conditions of culture

Percentage change 48 hr With fresh medium and 223 mM DMSO Percentage change



+ 9

6.0 2 1.0 24


122 Z!I16

225 + 3

35 Ik 7

34 k 14

33 r 2

35 + 7

111 *


200 rt 6

20 f 4



19 + 1

20 f 4

‘Cell numbers and DNA content of confluent cultures were the average f SD of three, g-cm petri dishes. The protocol used is described under Materials and Methods (Tissue Culture). bTotal protein was the average + SD of three, 5-cm dishes. ‘Percentage of [‘Hlthymidine-labeled nuclei was the average f SD of four fields per dish from two dishes for cultures radioactively labeled for 48 hr.






107, 1985




Time (hours) FIG. 5. Effect of different concentrations of DMSO on division times of confluent cultures. Various additions were made to 4-day confluent cultures of Rama 25 (see Materials and Methods, Time-Lapse Cinematography section, Set 3), and then filming commenced immediately after changing the medium and addition of DMSO where appropriate. In the unchanged culture the old medium was removed, filtered, and then replaced (see Time-Lapse Cinematography section, Set 2). Two-hundred cells per film were followed and their times of division were recorded, one film per DMSO concentration. Key: (-) control, no medium change; (- - -) control with fresh medium; (- - -) fresh medium plus 56 mM DMSO; (-*-) fresh medium plus 112 mM DMSO; (-* *-) fresh medium plus 16’7 mM DMSO; and (-. -) fresh medium plus 223 mM DMSO.

between 5 and 15 hr, up to 40 hr of culture. After this time Rama 25 cells tended to have longer CCTs, due possibly to the effects of density-dependent inhibition of growth (Holley, 1975). In the presence of 223 mM DMSO cells that were “born” within the first 5 hr had CCTs between 12.5 and 20 hr. During the subsequent g hr no mitoses occurred, then three cells divided, and a further gap of 4; hr was seen, until the division rates of cells approached those observed in cultures without

DMSO. Addition of increasing concentrations of DMSO led to a progressive increase in the minimum CCTs (not shown). Similarly, increasing DMSO concentrations led to a progressively lower percentage of cells with CCTs between 0 and 10 hr (Table 6). DMSO treatment did not dramatically affect the percentage of cells with CCTs between 10 and 20 hr, although the percentage of cells with CCTs between 20 and 30 hr and greater than 30 hr were both increased with DMSO








Rate of division” (cell divisions/hr)


No medium change Fresh medium 0 56 112 167 223


2.6* 6.4* 2.9* 2.2* 3.3* 2.7*


Cells dividing between 30 and 50 hr

Cells dividing between 10 and 30 hr Concentration DMSO (m&f)


No. cells involved 49 87 43 36 48 30

Rate of division” (cell divisions/hr) 1.20* 1.8 0.34* 0.26* 0.21* 0.11*

No. cells involved 25 35 6 5 5 3

’ Rates of division were calculated using the linear regression equation from the plots of the number of undivided cells against time and correspond to the weighted average rate of change of each line in Fig. 5 for the appropriate time period (lo-30 hr or 30-50 hr). Films were of confluent cultures with no medium change or with fresh medium and DMSO as indicated (see Materials and Methods, Time-Lapse Cinematography section, Sets 2,3), 200 cells/film were followed. The fit to a straight line for semi-log plots gave no improvement on the above. * Correlation coefficient T > 0.94, for the remaining rate r = 0.85.





upto 167 mM (Table 6). At higher concentrations this trend was reversed, possibly because of the toxic effects of these high concentrations of DMSO. DISCUSSION

The use of time-lapse cinematography is one method of investigating cellular interconversions and population growth. A possible limitation of this technique is the small number of cells available for individual study which may not be representative of the entire cell population. However, the study of cellular conversions requires just this type of approach, where a single cell can be continually monitored. In the present investigation three different sets of films have been analyzed and the qualitative and quantitative results for the identification of the preferential cellular routes adopted by differentiating Rama 25 cells are in agreement. The qualitative results show that two major pathways, or rather a truncated version of one pathway, are adopted by these cells. The ratio of the two pathways, or the degree of truncation, depends upon the concentration of DMSO. These results have been confirmed for a more physiological agent than DMSO, the vitamin A metabolite retinoic acid in the concentration range 0.1 to 5 PM (Paterson, 1983). Since similar kinetic results are obtained from conventional microscopic analysis of cells in fixed fields and in random fields, both with DMSO and retinoic acid (Paterson, 1983), results from cells within a fixed field, such as those of the timelapse films, can be taken as a representative sample of a particular culture. Rates obtained with the higher concentrations of DMSO are nearly the same when


Concentration of DMSO WW 0

56 112 167 223




of cells

O-10 hr


28 20 5.4 0.5 1.6

51 45 51 44 63




20-30 8.7 11 12 21 18


ceil cycle


times” 30-132 hr 13 24 31 35 18

“Number of cells for a given concentration of DMSO with cell cycle times of the described length were taken from plots of the intermitotic times against time of birth for each concentration of DMSO. Films were of sparsely plated Rama 25 cells as described under Materials and Methods (Time-Lapse Cinematography section, Set 1). The intermitotic or cell cycle time is defined as the time between two subsequent divisions, and the time of birth as the time at which the daughter cells become completely separated from each other.






obtained from microscopic or time-lapse analysis. However, a difference in the rate of formation of droplet cells in cultures without DMSO is observed between these two methods, but this discrepancy is probably due to the different experimental protocols employed. Similarly, qualitative and quantitative agreement is found between the LI, the DNA content, and the cell number during the decline in cellular proliferation rate with DMSO or with retinoic acid (Rudland et al, 1983). Qualitative agreement also exists between the 48 hr-LI and the average CCTs. Quantitatively however, the 48 hr-LI decreased by threefold while the average CCT only doubled in the presence of 223 mM DMSO or 3.3 PLM retinoic acid. Confirmation and biological relevance of the morphological results obtained with differentiating Rama 25 cells in vitro comes, in part, from the use of peanut lectin. This lectin, after desialylation of appropriate histological sections, stains specifically a subpopulation of epithelial cells which increases in number during development of the rat mammary gland. This increase is observed at all stages of development from neonates up to pregnant animals, where its maximum occurs prior to the maximum of casein secretion (Newman et aZ., 1979). In desialylated cultures of Rama 25, peanut lectin stains the dark cells, whether or not they contain droplets; the cuboidal cells and gray cells are unstained. Anticasein serum, however, stains the droplet cells in domes only weakly. This is consistent with the observations in vitro that peanut lectin-binding capacity of the desialylated Rama 25 cells increases with the increase in dark and dark droplet cells, and that these events occur prior to the almost simultaneous increase in doming and casein secretion in DMSO-treated cultures. The sequential order of the increase in lectinbinding capacity and casein secretion in vitro is the same as that observed in vivo, which suggests that peanut lectin binds preferentially to desialylatable epithelial cells that are the precursors of the casein secretory alveolar cells. Confluent Rama 25 cells are predominantly in the G1 phase of the cell cycle, since the maximum 4 hr-LI occurs mainly before the increase in cell numbers is observed. Addition of fresh medium to confluent Rama 25 cells produces two maxima for the 4 hr-LI, and these probably represent two rounds of DNA synthesis. The substantial reduction in the LI with DMSO or retinoic acid (Rudland et al, 1983) occurs mainly during the second maximum. This effect may be due to a lengthening of the second (and subsequent) cell cycles, as shown by the increase in the average CCT after the first division. By comparison, the CCTs of the first divisions in growing cultures, and the time to first division in confluent cultures are relatively unaffected




by increasing concentrations of DMSO, within the first 10 hr. Thus the increase in the average CCTs may be mainly due to the extension of the G1 period of the second (and subsequent) cell cycles. Previous studies have shown that inhibitors of DNA synthesis largely prevent doming and casein synthesis in Rama 25 cells, and that DMSO has to be present throughout the 8-hr lag period prior to doming, and probably beyond, for the production of the maximum number of domes (Warburton et al, 1983). These results together suggest that DMSO must be present at least throughout one G1 and S phase before differentiation to alveolar-like cells can occur. Similarly in the Friend erythroleukemic cell system, DMSO is thought to exert its differentiating effect near the G1/S phase boundary (Levy et al, 1975; Conkie et aL, 1981), and to lengthen the subsequent G1 phase(s) of the cell cycle (Gusella and Housman, 1976; Terada et CAL,1977). However, unlike the Friend system (Fibach et aL, 1977) Rama 25 differentiated cells (both dark droplet cells and cells forming domes) can divide, and the whole process can be reversed by removal of DMSO, showing that differentiation and DNA synthesis or cell division need not be mutually exclusive. Similar results have been obtained with DMSO replaced by retinoic acid (Paterson, 1983). When the rat mammary gland differentiates to alveoli, a progressive lengthening of the average CCT is seen, mainly due to a lengthening of G1 (Russo and Russo, 1980). The magnitude of the change in CCT in viva is similar to that for differentiating Rama 25 cells in vitro. Thus in young virgin rats the CCT in termal end buds (TEBs) is 9.9 hr, in terminal ducts (TDs) 17.3 hr, and in alveolar buds (ABs) 28.2 hr. In old virgin rats, the CCT is 20.6 hr in TDs, and 30.8 hr in ABs. In parous rats the CCT is 23.9 hr in TDs and 49.6 hr in ABs. From time-lapse data of Rama 25 cells growing with 223 mM DMSO the CCT for cuboidal cells is 17 + 4 hr, and that for droplet cells is 30 + 10 hr. In the absence of DMSO, the CCT for cuboidal cells is 13 f 2 hr and for droplet cells is 31 -t 5 hr. Moreover, there is also a two- to threefold reduction in the LI in the differentiated rat mammary gland accompanying the disappearance of TEBs and the decrease in TDs, in both the young and old virgin rats (Russo and Russo, 1980). This is consistent with our threefold difference in LI between undifferentiated Rama 25 cells and differentiated droplet cell and doming cultures containing 223 mM DMSO. In the mammary gland the ABs regress, but do not disappear, after pregnancy and lactation (Russo et aL, 1982). This finding implies a certain degree of cellular reversibility. The mature alveolar cells are thought to be terminally differentiated (Russo et al, 1982). On four accounts (a) peanut lectin-binding ability, (b) cell


107, 1985

cycle times, (c) reduction in LIs, and (d) cell reversibility, growing Rama 25 cuboidal cells seem to be similar to cells in TEBs and TDs, while the dark + droplet cells seem to be more like the AB cells. This similarity between the dark droplet cells and some of the epithelial cells within ABs, may account for the dark droplet cells’ low level (1%) of casein synthesis compared with that found in lactating mammary gland explants (Warburton et ah, 1983). In conclusion, a linear pathway of morphological intermediates can be identified between the cuboidal epithelial stem cells and alveolar-like or more probably AB-like cells in vitro, and DMSO accelerates this process predominantly by truncating one of the steps in the pathway. We thank




for setting

up and maintaining the Davies for expert biochemvaluable discussion on the data. During the course of this work Miss receipt of a Ludwig Institute Research

time-lapse apparatus, Mrs. Anna Twiston ical assistance, and Dr. John Smith for analysis of the time-lapse Frieda Paterson was in Studentship.

REFERENCES BENNE’IT, D. C., PEACHEY, L. A., DURBIN, H., and RUDLAND, P. S., (1978). A possible mammary stem cell line. CeU 15,296-308. CHEPKO, G. (1984). Fine structural evidence for a stem population in rat mammary gland. Submitted for publication. CONKIE, D., HARRISON, P. R., and PAUL, J. (1981). Cell-cycle dependence of induced hemoglobin synthesis in Friend erythroleukemia cells temperature-sensitive for growth. Proc. Natl. Acad Sci USA 78, 3644-3648. ELIAS, J. J. (1957). Cultivation of adult mouse mammary gland in hormone-enriched synthetic medium. Science (Washington, D. C.) 126, 842-844. FIBACH, E., REUBEN, R. C., RIFKIND, R. A., and MARKS, P. A. (1977). Effect of hexametbylene bisacetamide on the commitment to differentiation of murine erythroleukemia cells. Cancer Res 37, 440-444. FRIEND, C., SCHER, W., HOLLAND, J. G., and SATO, T. (1971). Hemoglobin synthesis in murine virus-induced leukemic cells in vitro: Stimulation of erythroid differentiation by dimethyl sulfoxide. Proc Natl. Acad. Sci. USA 68, 378-382. GUSELLA, J., and HOUSMAN, D. (1976). Induction of erythroid differentiation in vitro by purines and purine analogues. Cell 8, 263-269. HOLLEY, R. W. (1975). Control of growth of mammalian cells in culture. Nature (limdm) 258,487-490. HUNTER, W. M., and GREENWOOD, R. C. (1962). Preparation of iodine131 labelled human growth hormone of high specific activity. Nature (Lundam) 194,495-496. JIMENEZ DE ASUA, L., CARR, B., CLINGAN, D., and RUDLAND, P. S. (1977). Specific glucocorticoid inhibition of growth promoting effects of prostaglandin Fti on 3T3 cells. Nature (Lmdm) 265,450-452. LASFARGUES, E. Y., and OZZELLO, L. (1958). Cultivation in human breast carcinomas. J. Natl. Cancer Inst. 21,1131-1147. LEVY, J., TERADA, M., RIFKIND, R. A., and MARKS, P. A. (1975). Induction of erythroid differentiation by dimethyl sulfoxide in cells infected with Friend virus: Relationship to the cell cycle. Proc. Natl. Acad Sci USA 72, 28-32. LOWRY, 0. H., ROSENBROUGH, N. J., FARR, A. L., and RANDALL, R. J.





(1951). Protein measurements with Folin-phenol reagent. J. Biol. Chem+ 193, 265-275. MCGRATH, C. M. (1975). Cell organisation and responsiveness to hormones in vitro. Genesis of domes on mammary cultures. Amer. J. 2001 15,231-236. NEWMAN, R. A., KLEIN, P. J., and RUDLAND, P. S. (1979). Binding of peanut lectin to breast epithelium, human carcinomas and a cultured rat mammary stem cell: Use of the lectin as a marker of mammary differentiation. J. NatL Cancer Inst. 63,1339-1346. ORMEROD, E. J. (1983). “The Study of Rat Mammary Gland Morphogenesis Using Clonal Cell Lines.” pp. 56-97. Ph.D. Thesis, University of London. ORMEROD, E. J., and RUDLAND, P. S. (1982). Mammary gland morphogenesis in vitro: Formation of branched tubules in collagen gels by a cloned rat mammary cell line. Dev. Biol 91, 360-375. ORMEROD, E. J., and RUDLAND, P. S. (1984). Cellular composition and organisation of ductal buds in developing rat mammary glands: Evidence for morphological intermediates between epithelial and myoepithelial cells. Amer. J. Anat. 170, 631-652. PATERSON, F. C. (1983). “Cellular and Polypeptide Changes Associated with the Differentiation of a Mammary Stem Cell Line.” Ph.D. Thesis, University of London. RADNOR, C. J. P. (1972). Myoepithelial cell differentiation in rat mammary glands. J. Anut. 111,381-398. RAYNAUD, A. (1961). Morphogenesis of the mammary gland. In “Milk, the Mammary Gland and its Secretions” (S. K. Kon and A. T. Cowie, eds.), Vol. 1, pp. 3-46. Academic Press, New York. RIDDLE, P. N. (1979). “Time-Lapse Cinematography.” Academic Press, New York/London. RUDLAND, P. S., BENNETT, D. C., RIT~ER, M. A., NEWMAN, R. A., and WARBURTON, M. J. (1980a). Differentiation of a rat mammary stem cell line in culture. In “Control Mechanisms in Animal Cells” (L. Jimenez de Asua et d, eds.), pp. 341-365. Raven Press, New York. RUDLAND, P. S., BENNETT, D. C., and WARBURTON, M. J. (1980b). Growth and differentiation of cultured rat mammary epithelial cells. In “Hormones and Cancer” (S. Iacobelli et al, eds.), pp. 255269. Raven Press, New York. RUDLAND, P. S., HALLOWES, R. C., DURBIN, H., and LEWIS, D. (1977). Mitogenic activity of pituitary hormones on cell cultures of normal






and carcinogen-induced epithelium from rat mammary glands. J. Cell Bid 73,561-577. RUDLAND, P. S., PATERSON, F. C., TWISTON DAVIES, A., and WARBURTON, M. J. (1983). Retinoid-specific induction of differentiation and reduction of the DNA synthesis rate and tumor-forming ability of a stem cell line from a rat mammary tumor. .Z. Nat1 Cancer Inst. 70,949-958. RUDLAND, P. S., TWISTON DAVIES, A., and WARBURTON, M. J. (1982). Prostaglandin-induced differentiation or dimethyl sulfoxide-induced differentiation: Reduction of the neoplastic potential of a rat mammary tumor stem cell line. .Z. Natl. Cancer Inst. 69.1083-1093. RUSSO, J., and RUSSO, I. H. (1980). Influence of differentiation and cell kinetics on the susceptibility of the rat mammary gland to carcinogenesis. Cancer Res. 40,2677-2687. Russo, J., TAY, I. K., and Russo, I. H. (1982). Differentiation of the mammary gland and susceptibility to carcinogenesis. Breast Cancer Res. Treat. 2, 5-73. SUPOWIT, S. C., and ROSEN, J. M. (1982). Hormonal induction of casein gene expression limited to a small subpopulation of 7,12dimethylbenz[a]anthracene-induced mammary tumor cells. Cancer Res. 42, 1355-1360. TERADA, M., FRIED, J., NUDEL, U., RIFKIND, R. A., and MARKS, P. A. (1977). Transient inhibition of initiation of S-phase associated with dimethyl sulfoxide induction of murine erythroleukemia cells to erythroid differentiation. Proc Natl. Acad Sci USA 74, 248252. TURKINGTON, R. W., LOCKWOOD, D. H., and TOPPER, Y. J. (1967). The induction of milk protein synthesis in post-mitotic epithelial cells exposed to prolactin. B&him. Biophys. Actu 148,475-480. WARBURTON, M. J., HEAD, L. P., FERNS, S. A., and RUDLAND, P. S. (1983). Induction of differentiation in a rat mammary epithelial stem cell line by dimethyl sulphoxide and mammotrophic hormones. Eur. J. B&hem. 133,707-715. WILLIAMS, J. M., and DANIEL, C. W. (1983). Mammary ductal elongation: Differentiation of myoepithelium and basal lamina during branching morphogenesis. Dev. Biol. 97,274-290. YOUNG, S., and HALLOWES, R. C. (1973). Tumours of the mammary gland. In “Pathology of Tumours in Laboratory Animals” (V. S. Turusov, ed.), Vol. 1, pp. 31-74. IARC Scientific Publications, Lyon.