A morphologically distinct candidate for an epithelial stem cell in mouse

mammary gland

GILBERT H. SMITH

Laboratory of Tumor Immunology and Biology, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA

and DANIEL MEDINADepartment of Cell Biology, Baylor College of Medicine, Houston, TX 77030, USA

Summary

Transplantation studies demonstrate that an epi-thelial stem cell component must exist in themouse mammary gland throughout life. Samplestaken from any portion of the mammary gland atany age and at any developmental stage, includ-ing full functional differentiation, give rise tomammary epithelial outgrowths with completedevelopmental capacity. Cytological examinationof mouse mammary gland explants revealed thepresence of morphologically distinct cells dis-tributed sporadically among the mammary epi-thelium, whose behaviour in vivo and in vitrosuggested that they might represent a latent epi-thelial stem cell population. These pale-stainingcells possessed large spherical nuclei, a clearcytoplasm and a round smooth-contoured shape.Electron microscopy confirmed their pale-stain-ing characteristics and revealed a cytoplasmsparsely populated with organellar structures,such as mitochondria and endoplasmic reticu-lum. Their epithelial genealogy was demon-strated by the presence of terminal bars and tightjunctions formed with their epithelial cell neigh-bours. In vivo, these cells were found amongmammary epithelial cell populations in 16-day-old embryos onward in both ductal or lobular

structures during all stages of pregnancy, lac-tation and involution. In explant cultures, thesecells did not undertake a secretory morphology inthe presence of lactogenic hormones, althoughoccasionally they became immunologically posi-tive for casein. They did not incorporate [3H]-thymidine into their nuclei under any of theexperimental conditions used; however, they ap-peared to undergo mitosis within 4 h regardless ofthe presence or absence of hormone(s). At 24 hincreased numbers of pale cells were found inpairs or in groups. At 72 h in the presence of IFPrl(medium containing insulin, hydrocortisone andprolactin), the pairs and groups of pale cellsobserved at 24 h were not found. Instead, indi-vidual pale cells were seen among groups ofcytologically and functionally differentiated se-cretory epithelial cells. When lactogenic hor-mones were not present, groups of pale cells werestill present in the explants at 72 h. These findingssuggest that the pale cells are arrested at G2 phaseof the cell cycle and that they give rise by mitosisto daughter cells capable of differentiating in thepresence of lactogenic stimuli. Inhibition of DNAsynthesis in the explants did not alter these cellu-lar events.

Key words: stem cell, mammary gland, epithelium.

Introduction

The mouse mammary gland originates from a thicken-ing of the embryonic ectoderm, which subsequentlyinvades the underlying mesenchyme. During theearliest stages of development, the ectoderm forms aclosely packed knob of cells that projects into theunderlying mesenchyme. This knob lengthens in thefemale mouse embryo, beginning at 14 days to form an

Journal of Cell Science 89, 173-183 (1988)Printed in Great Britain © The Company of Biologists Limited 1988

elongated solid cord of epithelial cells that laterbranches in a manner characteristic of the adult mam-mary gland. At birth, the female mouse possessesmammary glands composed of branched epithelialcords that contain lumina, i.e. a system of branchinghollow epithelial tubes that open at the nipple (Hogg etal. 1983). Little subsequent mammary growth anddevelopment occur until the onset of puberty, when the

173

immature gland grows rapidly to produce the tree-likepattern of ducts, upon which secretory alveolar lobuleswill develop during pregnancy. During this period ofgrowth and invasion of the mammary fat pad by themammary ductal system, beginning at approximately 3weeks of age and continuing until approximately 12weeks of age in virgin females, intense mitotic activityoccurs within the mammary end buds (Bresciani,1965). These bulb-like structures at the ends of theexpanding ducts are specialized at their advancing edgeto permit rapid penetration of the surrounding fattystroma (Silberstein & Daniel, 1982). The posteriorregions of the end buds provide differentiating paren-chymal and myoepithelial cells that contribute to theformation and elongation of the subtending ducts(Williams & Daniel, 1983). The growing end budsbranch at regular intervals and generate the character-istic pattern of the mature gland. Local factors appearto regulate the spacing between the growing branchesso that competition of extending ducts is avoided.When insufficient gland-free fat pad is available forcontinued growth, the end buds disappear and thesubtending ducts become mitotically quiescent(Faulkin & DeOme, 1960).

Despite the disappearance of the end buds and theirspecialized structures, including the undifferentiated'cap cells' (Williams & Daniel, 1983), which appear togive rise to both myoepithelial and parenchymal epi-thelial cells, portions of mammary gland from virginfemale mice of any age are capable of regenerating acomplete functional epithelial tree upon transplan-tation into an epithelium-free mammary fat pad(DeOme et al. 1959, 1960; Hoshino, 1962, 1978;Daniel et al. 1968; Medina, 1973). These observationsargue that mammary epithelial-specific stem cells per-sist within the mammary gland population throughouttheir lifetime. Serial transplantation of mammary epi-thelium demonstrates that normal mouse tissue has afinite proliferative capacity (Daniel et al. 1968) and thatthe rate of senescence is governed by the length of timebetween transplant generations, suggesting that mitoticevents determine the rate of loss of proliferative ca-pacity (Daniel, 1972).

In the present study, we have examined the 'repopu-lating capacity' of various portions of the mousemammary gland during various stages of its develop-ment and differentiation. In addition, we haveattempted to identify the potential mammary epithelialstem cells among these mammary populations in situand within explant cultures by employing auto-radiographic, immunocytochemical and morphologicalcriteria. In our concept, the term mammary epithelialstem cell describes an uncommitted mammary cell (i.e.not functioning as myoepithelium or glandular epi-thelium) that can give rise through mitosis to commit-ted epithelial cell progeny without losing its own

uncommitted status. It is implicit in this definition thatthe mitotic event has been unequal in the sense that theprogeny are sensitive to certain mammogenic stimuliwhereas the stem cell is not.

Materials and methods

MiceBalb/c/CRGL, C3H/He and C3H/Sm mice were main-tained at 23°C on a 12 h / l2h , light/dark cycle and given foodand water ad libitum. Females from each strain were evalu-ated in explant culture and/or by transplantation and mor-phological techniques, at all stages of mammary glandfunction and development, for the presence and fate ofpotential mammary epithelial stem cells.

Explant cultureThe abdominal mammary glands were aseptically removedand cultured as described (Smith & Vonderhaar, 1981;Vonderhaar & Smith, 1982; Smith, 1987). The pooledexplants from 6-13 animals were cultured in medium 199containing insulin (I; 5jUgml~'), hydrocortisone (F;l/Ugml"1) and prolactin (Prl; ljUgml"1). Where indicated,cytosine arabinoside (araC) was added at 15/igml~' to virginmouse explants and at 30jUgml~' to explants of pregnantgland. These levels of araC inhibit DNA synthesis by 95%without toxic side effects (Owens et al. 1973; Smith &Vonderhaar, 1981; Vonderhaar & Smith, 1982). The culturemedium was changed routinely after 24 and 72 h in eachexperiment; part of the pooled explants was used for bio-chemical analyses and the remainder was prepared formorphological studies.

Preparation of epithelial cell pelletA fraction enriched in mammary epithelial cells was obtainedfrom freshly excised tissue or from explants by modificationof methods described previously (Topper et al. 1975). Fatcells were removed by incubating up to 600 mg of tissue in5 ml of medium 199 (pH7-4) containing l-5mg of crudecollagenase per ml and 4% bovine serum albumin (BSA).Incubation was in a shaking 37°C water bath for 30—45 min.At 5-min intervals, the contents of the incubation vesselswere aspirated into Pasteur pipettes; the final two aspirationswere done using fine-bore pipettes. The dispersed cells werecollected by centrifugation for 5 min at 800 g at roomtemperature and washed three times with phosphate-bufferedsaline (PBS). This gentle disruption procedure allows manyductal and alveolar structures to remain intact (Wicha et al.1979).

Preparation of mammary cells for light and electronmicroscopyEpithelial cell pellets or intact mammary explants were fixedovernight at 4 °C in 3 % formaldehyde (freshly prepared fromparaformaldehyde) in 0-1 M-sodium cacodylate buffer(pH7-4). The explants or cell pellets were trimmed, thenrinsed thoroughly in 0-1 M-sodium cacodylate prior to post-fixation for 1 h in Dalton's (1955) chrome-osmium fixative.The specimens were then stained en bloc with 0-05 % uranyl

174 G. II. Smith and D. Medina

acetate at p H 4 9 , dehydrated through a graded series ofethanol and propylene oxide, infiltrated and subsequentlyembedded in Epon-Araldite (Mollenhauer, 1964). For eachexperimental or control group, at least eight blocks weresectioned and studied by light and electron microscopy.Sections (l-2jum) thick, were cut from the block with glassknives, picked up on glass slides, dried, stained with Tol-uidine Blue-Azure II and examined under a light micro-scope. Comparable areas in all groups were selected forelectron-microscopic examination. Profiles of at least 300epithelial cells were evaluated ultrastructurally in each exper-imental group. Sections with a silver-silvergold interferencecolour were cut for electron microscopy. The sections werepicked up on Formvar-carbon-coated grids and stained withuranyl acetate and lead citrate. The stained sections wereviewed in a Philips 400 electron microscope at 60 kV acceler-ating voltage. Electron image photographs were taken atinitial magnification of 1500-20000 diameters.

Autoradiographic analysts of thymidine-labellingpatterns

Explain cultures were labelled with 0 1 ^Cirnl"' of [3H]thy-midine for 72 h and washed free of label as described above.The tissue was fixed and embedded in the same manner as forelectron microscopy except that OsO4 fixation was omitted.Sections ( l -2^m thick) were cut and dried on gelatin-coatedglass slides, and the plastic was removed as mentioned above.The slides were dipped in NTB-2 Kodak liquid emulsion,stored in the dark for 10-14 days and then developed inKodak D-19. Distribution of the autoradiographic grains wasdetermined on haematoxylin-stained sections with a Zeissphotomicroscope.

Vaginal smears from mature virgin female mice werechecked daily at 8.00 a.m. to ascertain their oestrous cycle.Mice at various stages in the oestrous cycle were inoculatedwith 0 1 mCi of [3H]thymidine at 8.30 a.m. and were killed1 h later. The no. 2 and no. 3 mammary fat pads were excisedand fixed as whole mounts on Millipore filters in 4-0%paraformaldehyde. Subsequently, the glands were dehy-drated, embedded, sectioned and stained. Stained sectionswere dipped in emulsion and autoradiography was performedfor 12-14 days. The sections were examined for radionucleo-tide-labelled nuclei under phase-contrast optics.

Tissue localization of casein

Explants of mammary tissue that had been incubated withvarious hormone combinations were fixed and embedded inAraldite for immunocytochemical studies as described byHogan & Smith (1982). Sections (1-2/^m thick) were cut andplaced on glass slides. Embedding plastic was removed with asolution of benzene :methanolic KOH: acetone (1:1:1, byvol.). Casein in the sections was localized by an avidin-biotinindirect immunoperoxidase staining technique as described(Smith & Vonderhaar, 1981; Hogan & Smith, 1982). Thespecificity of the anti-casein antibody was confirmed byabsorption with purified mouse caseins and by immuno-precipitation (Smith & Vonderhaar, 1981). Non-specificabsorption and background peroxidase activity were assessedby examining the staining of sections after stepwise omissionof each individual reagent used in the technique. Sectionswere cut consecutively, exposed to rabbit preimmune serum

and to anti-MMTV and served as controls. The slides werecoded and evaluated by two separate viewers.

Transplantation studies

Female Balb/c mice of different ages whose mammary glandswere in various stages of development and differentiationwere injected with 0-5 % Trypan Blue 16 h before being usedas donors so that the gland was more easily visualized in theliving fat pad. The donors were anaesthetized and themammary glands were exposed and examined under adissecting microscope. In all experiments, the smallest dis-sectable piece of duct, alveolus or end bud was collected andtransplanted into virgin Balb/c mice (3 weeks old) whoseinguinal mammary fat pads were surgically exposed andcleared of all epithelial elements as described by DeOmee/ al.(1959) and Medina (1973). In the majority of experiments,the transplant recipients were maintained as virgins for 6weeks after transplantation. In some experiments, the recipi-ents were subsequently mated and nursed their young for 3-5days before the glands were collected. Whole mounts weremade of all fat pads containing mammary epithelial growthand stained with haematoxylin. The extent of the cleared fatpad filled by growth of the implanted epithelium wasestimated according to the method described by Daniel el al.(1968). The estimate of the number of epithelial cells pertransplant piece was performed as described by Medina et al.(1978). Briefly, 40-60 transplant-sized pieces were dissectedfrom the gland, pooled and then digested with collagenase toprovide a single cell suspension. For each estimation, three tofour pooled samples were evaluated. The collagenase-digested samples were washed in serum-free Dulbecco'sminimal essential medium by centrifugation, resuspended,and the number of cells was determined in a Coulter counter.The total number of cells in each pooled sample was dividedby the number of pieces placed in the original pool to give theaverage number of cells per transplant-sized piece.

Results

Transplantation studies

Different segments of the mammary parenchyma weretransplanted into the cleared fat pads of young mice toassess their morphogenic potential. The results areshown in Table 1. The portions of the mammary glandtaken from primary ducts of mature female virgin mice,tertiary ducts of immature and mature female virginmice, tertiary ducts from uniparous non-pregnant non-lactating mice, 15-day pregnant mice and 10-day lactat-ing mice, and end buds from 6-week-old mice all gaverise to well-developed ductal outgrowths in the mam-mary fat pad at 6 weeks after transplantation. In themajority of experiments, the values for percentage ofsuccessful transplants and percentage of fat pad filledwere greater than 80%. These two parameters wereused to assess growth potential of the mammaryparenchyma. The outgrowths from lactating mammarygland provided the major exception to the generalpattern of growth and morphogenesis. In experiment 1

A latent stem cell in mammary epithelium 175

Table 1. Morphogenic potential of different segments of mammary parenchyma of Balb/c female mice

Type ofmammary parenchyma

3° duct3° duct3° duct3° duct3° duct1 ° ductAlveoliAlveoliEnd bud

Donor group

6-week virgin6-week virgin16-week virginUniparous regressedMultiparous regressed16-week virgin15-day pregnant10-day lactating6-week virgin

Successful/totaltransplants (%)

13/14 (93)19/20| (95)11/12 (92)16/20 (80)19/20 (95)19/20 (95)14/16 (88)19/20J (95)15/201! (75)

X(% Fat pad filled)*

100999985"80"949560b

86"

Range

(100)(90-100)(90-100)(50-100)(10-100)(40-100)(50-100)(10-100)§(25-100)

•The groups with letter superscripts (a or b) are significantly different from 3° duct of 6-week virgin. Groups with superscript b arcsignificantly different from groups with superscript a. The data were analysed by one-way analysis variance.

f Approximately 7200 cells per transplant.| Approximately 33 000 cells per transplant.§ Few branch points were observed in the epithelial ducts compared with the outgrowths derived from mammary implants in adult virgin

mice.II Approximately 2200 cells per transplant.

(Table 1), transplants of lactating alveoli were success-ful in 95% (19 of 20) of the cases, but the meanpercentage of fat pad filled, the successful transplants,was only 60%. In addition to the slower growth rate,the outgrowth exhibited fewer branch points comparedwith outgrowths from either virgin ductal or pregnantalveolar tissue. The slower growth could not be attribu-table to fewer cells in the original transplant, sincethere were significantly greater numbers of cells pertransplant in the lactating group (33 000/transplant)compared with ducts from virgins (7200/transplant) orend buds (2200/transplant). The slow growth rate anddecreased number of branches in outgrowths fromlactating gland were observed in three separate exper-iments (Tables 2 and 3), where the percentage ofsuccessful transplants was high (49/60, 82%) but themean percentage of fat pad filled was low (53 %).

In the second set of experiments, mammary tissuefragments containing tertiary ducts from mature vir-gins or alveoli from lactating mice were transplantedinto recipients who subsequently became pregnant andlactating, in order to examine the differentiation poten-tial of the two groups of transplants (Table 2). Theextent of growth of mammary epithelial cells derivedfrom lactating mice once again showed a reducedcapacity for repopulation of the fat pad in virgin mice.However, mammary cells from either source com-pletely filled the fat pad in recipients who subsequentlybecame pregnant and lactated. Full lactogenic re-sponses were observed in all the outgrowths at 10 daysof lactation regardless of the source of the implantedtissue.

To determine whether a further loss in the morpho-genic capacity of lactating epithelium was experienced

Table 2. Differentiation potential of cells derived from virgin and lactating mammary gland

Donor

3° duct, virgin3° duct, virgin10-day lactating10-day lactating

Recipient

Virgin10-day lactatingVirgin10-day lactating

Time(weeks)

612-14

612-14

Successful/totatransplants (%)

20/20 (100)20/20 (100)17/20 (85)20/20 (100)

Table 3. Morphogenic potential of different segments

Donor

10-day lactating10-day lactating20-day lactating20-day lactating

•Zone 1, most peripheral

Group*

Zone 1Zone 2Zone 1Zone 2

region of gland; zone 2,

Successful/totaltransplants (%)

19/20 (95)14/20 (70)13/20 (65)13/20 (65)

(% Fat

intermediate region of gland.

11 (%

A'Fat pad

9810057

100

filled)

of lactating glands

Xpad filled)

67733468

Range

25-10010-10010-7520-95

Range

(60-100)(100)

(10-100)(100)

176 G. H. Smith and D. Medina

Fig. 1. Large pale-staining epithelial cells (arrows) are readily identified in Toluidine Blue-stained 1 jum thick sections ofmouse mammary glands at all stages of development and differentiation: A, in the virgin mouse; B, 12 days pregnant;C, 6-day lactating; and D, in male mammary anlage that have been induced to grow by chronic stimulation withdiethylstilboestrol. Bars, lOjUm.

IT5 AFig. 3. Mitotic figures are found in the pale-staining mammary epithelial cells, just 4h after explain culture in the presenceor absence of araC. Bar, lOjum.Fig. 4. Electron microscopy of the dividing cells in the 4-h cultures confirms that they possess an ultrastructuralmorphology identical with the large pale-staining epithelial cells shown in Fig. 2A. Bar, 1-Ojum.Fig. 5. In A, after 24 h in culture, pale cells appear in groups of two or more among the darker staining epithelium,supporting the conclusion that they have divided at least once during this culture period. In B, electron-microscopicanalysis of these groups of pale cells suggests that, in the groups of three, a second mitotic event has occurred in one of thepale cell daughters leading to the production of two smaller clear-staining cells. Close examination shows that cellularmembranes have not yet been re-established between these newly formed cells. Bars: A, lOftm; B, 1-0 jum.

at later stages of lactation, fragments from differentportions of 10-day and 20-day lactating mammaryglands were transplanted (Table 3). A significant varia-bility in repopulation potential was observed when allthe lactating transplants were evaluated. However, thepoorest repopulation potential was observed in trans-plants taken from the most peripheral portion (zone 1)of the lactating gland after 20 days of lactation.Histological examination showed that no differences incellularity were present with respect to zone 1 and zone2 at either 10 or 20 days of lactation.

Light and electron microscopy

Microscopic examination of the functional and cyto-logical differentiation of hormone-stimulated mousemammary epithelium in explants cultured in serum-free chemically defined medium revealed the presenceof a morphologically distinct mammary epithelial cell.These cells were sporadically distributed among mam-mary epithelia in both ductal and lobuloalveolar struc-tures (Fig. 1A). They possessed spherical nuclei and apale-staining cytoplasm in Toluidine Blue-stained sec-tions from plastic-embedded mammary explants. Asurvey of mouse mammary epithelium at various stagesin the development and functional differentiation of themammary gland indicated that these cells were presentin the 16-day foetal gland (Hogg et al. 1983) throughpuberty, pregnancy, lactation (Fig. 1B,C) and invol-ution and in glands of aged non-breeding parous ornon-parous females. When the rudimentary mammaryanlagen of mature male mice are stimulated to growand develop under the influence of the syntheticoestrogen, diethylstilboestrol, pale cells identical tothose found in the female glands are found distributedprominently within the newly formed mammary epi-thelium (Fig. ID).

Ultrastructural examination of these morphologic-ally distinct cells confirms their pale-staining natureand demonstrates that their cytoplasm is sparselypopulated with mitochondria, ribosomes and vacuoles(Fig. 2). The mitochondria are in a condensed intra-cristal matrix configuration, suggesting low levels ofintracellular ATP and an inactive oxidative phos-phorylation and metabolism (Munn, 1974). The nucleiare large, mainly spherical and have relatively smallamounts of condensed heterochromatic regions. Thenucleolar region is not defined and is often not appar-ent. Examination of these cells in functionally differen-tiated mammary epithelia, such as pregnant or fullylactating glands, reveals little alteration in their ultra-structural appearance under these conditions, althoughtheir relative number when compared with the totalepithelial composition of the gland is much reduced.Once again, the pale cells are found within functionallobular structures as well as in ducts and ductules.Their epithelial genealogy is demonstrated in the

formation of terminal bars, tight junctional complexesand desmosomes with their functionally differentiatedepithelial cell neighbours (Fig. 2B, inset). They alsoformed hemidesmosomes with adjacent myoepithelialcells (Fig. 2A, inset). In addition to the possession ofthese ultrastructural features characteristic of epithelialcells, pale cells stained positive for epithelial keratinswhen tested by immunocytochemical technique (notshown). These immunological observations were partof a larger study, which demonstrates that differentialexpression of epithelial keratins occurs during mam-mary gland growth and within preneoplastic and malig-nant mammary epithelial populations (unpublisheddata).

Explant culture

Explants of mammary tissue from mature virgin femalemice were incubated in medium 199 in the presence ofvarious hormone combinations. The explants werecollected after culture periods of 4, 24 and 72 h andwere examined for cell division, [3H]thymidine incor-poration, cytological differentiation and milk proteinsynthesis. Mitosis was determined by direct obser-vation of semithin sections of plastic embeddedexplants, stained with Toluidine Blue-Azure II, afterincubation in different hormone combinations for vari-ous time periods. Cellular DNA synthesis was evalu-ated by autoradiography of the same sections. Induplicate cultures, araC was added (15jUgml~') toblock DNA synthesis and [3H]thymidine incorpor-ation.

Prominent mitotic figures were observed among theepithelium of virgin, parous or pregnant mouse mam-mary gland explants cultured for 4h in serum-freemedium (Figs 3, 4). The numbers of mitotic figureswere not influenced to a significant degree by thepresence or absence of hormones or the hormonecombinations used. However, the duration of mitoticactivity or the mitotic event seemed to be prolonged inthe presence of IFPrl-containing medium, as judgedfrom the continued observation of mitotic cells in the24-h cultures containing these hormones comparedwith those with insulin alone or no hormones. Themitotic cells observed in the 4-h cultures appeared to bederived from the pale cells described in the previoussection. This conclusion is based on the distribution ofthe mitotic cells within the ducts and acini of theexplants, the manifestation of pairs and groups of palecells (presumptive evidence for pale cell division) induplicate cultures maintained for 24 h under identicalconditions (Fig. 5) and electron microscopic analysis ofthe very same dividing cells observed in semithinsections of 4h explants (see Fig. 4). This examinationwas accomplished using ultrathin sections taken fromthe same areas contained in these 1 [im sections. All ofthese mitotic events and the subsequent appearance of

A latent stem cell in mammary epithelium 177

Fig. 2. The ultrastructural morphology of the pale-staining epithelial cells seen in Fig. 1A,C shows that these cells containvery few cytoplasmic organelles, such as mitochondria and ribosomes, compared with their dark-staining neighbours. Ascan be seen in B, very little change in the pale cell ultrastructure is observed even in the 6-day lactating gland. Theepithelial pedigree of these cells is demonstrated by the formation of hemidesmosomes (arrows) with adjacent myoepithelialcells (inset in A) and terminal bars at their luniinal surface when in contact with their dark-staining, fully differentiatedneighbours (inset in B). Bars, l,um.

178 G. /7. Smith and D. Medina

pairs and small groups of pale cells at 24 h within theepithelial population of the mammary explants oc-curred to the same extent in cultures where DNAsynthesis was inhibited by araC. At 72 h, the pairs andgroups of pale cells, conspicuous at 24 h, were notobserved any longer in explants incubated in IFPrl-containing medium. Instead, the number of identifi-able pale cells was much decreased. This observationimplies that at least some of the cells produced from themitotic events observed at early culture times hadbecome differentiated (Fig. 6). Electron-microscopicexamination of the epithelium in IFPrl-treatedexplants confirmed that nearly all the parenchymalepithelium had taken on a secretory ultrastructuralappearance as described (Vonderhaar & Smith, 1982).In those explant cultures incubated with insulin aloneor with hydrocortisone, groups of pale-staining epi-thelial cells remained evident at 72 h (Fig. 6B), nocytological differentiation was apparent.

DNA and IZNA synthesis

Autoradiographic study of explants incubated in thepresence of [3H]thymidine for the entire 72 h cultureperiod demonstrated that 49-2-56-4% (Table 4) of themammary epithelial cells incorporated thymidine intotheir nuclei in the presence of IFPrl-containing me-dium (Smith et al. 1981). Somewhat smaller numbersof epithelial cells were labelled when explants incu-bated with either insulin alone or no hormone for 72 hwere evaluated. In no case was label discerned in thenuclei of the pale cells, although the epithelium im-mediately adjacent to these cells was at times heavilylabelled (Fig. 7A). However, no matter what hormoneswere present, labelled nuclei were only rarely observedwhen cultures were incubated for 72 h with araC

Table 4. Labelling index of epithelial cellpopulations organ culture

Mammarytissue source Hormones present

Cellscounted

Cellslabelled

Virgin

Primiparous

Virgin

IFPrlIFPrl + araCIFPrl + 0-01 fig m

colchine

IFPrlIFPrl+ araC

I*IFPrl*IFPrl+ araC

790500600

802620

701835420

422

328

400

53-40

54-1

49-90

157 22-3422 50-5

0

Explants were cultured as described in Materials and methods.[3H]Thymidinc was present throughout the 72h culture periodand was chased for 12h. Tissue was processed for autoradiographyas described in Materials and methods.

* Labelled non-epithelial cells in cultures with insulin aloneaccounted for an additional 456/1651 (27-2%) positive cells. InIFPrl, only 77/1300 (5-9%) non-epithelial cells were labelled.

(15 jug ml ' ) . Similar results, i.e. absence of labellingin the pale cell nuclei, were obtained when [3H]thymi-dine was administered in vivo at various stages in theoestrous cycle.

Exposure of similar cultures to a 5-h pulse of[3H]uridine at 48-53 h showed that pale cells incorpor-ated and subsequently retained this radionucleotideduring the 72h culture period, suggesting that RNA-synthetic activity was present. This observation sup-ports the conclusion that these cells are not completelydormant metabolically (Fig. 7B). No difference in thelevels of incorporation of uridine into the mammaryepithelium could be detected under different hormonalconditions by autoradiography with reference to thecomparison of pale- or dark-staining epithelial cells.

Casein synthesis and cytological differentiationFunctional differentiation in these explant cultures wasdetermined by indirect immunoperoxidase detection ofcasein. Casein synthesis was detectable by immunoper-oxidase as early as 24 h after incubation of virgin mousemammary explants in IFPrl; additional amounts ofcasein and an increase in positive cells were found at72h. At 72h, an occasional pale cell was stainedpositive for intracellular casein (not shown); however,most pale cells remained negative and unaffected by thepresence of lactogenic hormone stimulation. Caseinwas not detected in virgin mouse mammary glandexplants incubated in insulin alone, in hydrocortisoneor without hormone stimulation. Coincidentally, theepithelial cells in these explants did not take on asecretory cytological and ultrastructural morphology asdid those in explants incubated in IFPrl. The conver-sion to a secretory morphology occurred in the pres-ence of IFPrl even when DNA synthesis was inhibitedthroughout the duration of the experiment by araC.Thus, most of the cells produced by the mitotic eventsobserved at 4 h of culture and seen as pairs and groupsof pale-staining epithelial cells at 24 h were converted tosecretory cells in the presence of IFPrl. In the presenceof insulin or hydrocortisone, when no casein synthesisor secretory activity was initiated, these pale cellsremained non-secretory and were easily distinguishedfrom their dark-staining neighbours (Fig. 6A).

Discussion

Normal mouse mammary epithelium has a finite ca-pacity to regenerate functional mammary parenchymaupon serial transplantation into epithelia-free mam-mary fat pads (Daniel et al. 1968). Further, the rate ofsenescence (as measured by regenerative capacity) wasproportional to the replication frequency imposedupon the regenerating mammary tissue as opposed tothe passage of time; e.g. senescence could be reachedmore quickly by shortening the period between the

A latent stem cell in mammary epithelium 179

Fig. 6. Electron microscopy of explants cultured for 72h in IF (A) or in IFPrl (B). In the presence of I and F, the pale-staining cells are still recognizable in groups after 72h. They do not show any extensive differentiation. In contrast, in thepresence of IFPrl, only occasional pale cells are found among neighbouring epithelium that has uniformly undertaken asecretory ultrastructural morphology, which includes a highly developed rough endoplasmic reticulum, apically locatedintracytoplasmic lipid droplets and casein micelles. Bar,

180 G. //. Smith and D. Medina

Fig. 7. Explants were cultured in IFPrl and [3H]thymidine for 72h and then with cold thymidine for an additional 16 h.Autoradiographic grains were found over the nucleus in a high percentage of the dark-staining secretory epithelial cells butnever over the pale-staining cells (A). Conversely, when IFPrl-treated cultures were pulsed with [3H]uridine for 5h(48-53 h) during explant culture and subsequently in cold medium for 20 h, all the epithelial cells, including the pale cells(arrows), contained autoradiographic grains indicative of an active RNA synthetic metabolism. Bars, lOjiim.

A latent stem cell in mammary epithelium 181

transplantation generations (Daniel, 1972, 1975). Thisfact suggests that the indigenous mammary epithelialstem cells undergo only a finite number of cell divisionsbefore they themselves become replicatively defunct.Very similar results are obtained when normal mam-mary epithelium is cultured /';/ vitro, i.e. the culturedepithelial cells lose their ability to repopulate thecleared mammary fat pads of syngeneic or nude mice asa consequence of the number of subculture passagesrather than the duration of their maintenance in vitro(Ehmann

mammary fat pads of female C3H mice. Cancer Res. 19,515-520.

EHMANN, U. K., GUZMAN, R. C , OSBORN, R. C ,

YOUNG, L. J. T., CARDIFF, R. D. & NANDI, S. (1987).

Cultured mouse mammary epithelial cells: normalphcnotype after implantation. J. natn. Cancer hist. 78,751-757.

FAULKIN, L. J. & DEOME, K. B. (1960). Regulation ofgrowth and spacing of gland elements in the mammaryfat pad of the C3H mouse. J. natn. Cancer Inst. 24,953-963.

HELMINEN, I-I. J. & ERICSSON, J. L. E. (1968). Studies on

mammary gland involution. 1. On the ultrastructure ofthe lactating mammary gland. jf. Ultrastntct. Res. 25,193-213.

HOGAN, D. L. & SMITH, G. H. (1982). Unconventionalapplication of standard light and electronimmunocytochemical analysis to aldehyde-fixed Araldite-embedded tissues. J. Hislochem. Cvtochem. 30,1301-1306.

HOGG, N. A. S., HARRISON, C. J. & TICKLE, C. (1983).

Lumen formation in the developing mouse mammarygland. J . Embryol. exp. Morph. 73, 39-57.

HOSHINO, K. (1962). Morphogenesis and growthpotentiality of mammary glands in mice. I.Transplantability and growth potentiality of mammarytissue of virgin mice. J. natn. Cancer Inst. 29, 835-851.

HOSHINO, K. (1978). Mammary transplantation and itshistogenesis in mice. In Physiology of Mammary Glands(ed. A. Yokoyama, H. Mizuno & H. Nagasawa),pp. 163-228. Baltimore: University Park Press.

MEDINA, D. (1973). Preneoplastic lesions in mousemammary tumorigenesis. Meth. Cancer Res. 7, 3-53.

MEDINA, D., SHEPHERD, F. & GROPP, T. (1978).

Enhancement of the tumorigenicity of preneoplasticmammary nodule lines by enzymatic dissociation.jf. natn. Cancer Inst. 60, 1121-1126.

MOLLENHAUER, H. H. (1964). Plastic embedding mixturesfor use in electron microscopy. Stain Technol. 39,111-114.

MUNN, E. A. (1974). Relationships between mitochondrialstructure, metabolic control and ion transport. In TheStmcture of Mitochondria (ed. E. A. Munn),pp. 320-350. London, New York: Academic Press.

OWENS, I., VONDERHAAR, B. K. & TOPPER, Y. J. (1973).

Concerning the necessary coupling of development toproliferation of mouse mammary epithelial cells. J. biol.Chew. 248, 472-477.

SILBERSTEIN, G. B. & DANIEL, C. W. (1982).

Glycosaminoglycans in the basal lamina and extracellularmatrix of the developing mouse mammary duct. DeviBiol. 90, 215-222.

SMITH, G. H. (1987). Functional differentiation of virginmouse mammary epithelium in explant culture isdependent upon extracellular proline. J. cell. Physiol.131, 190-199.

SMITH, G. H. & VONDERHAAR, B. K. (1981). Functionaldifferentiation in mouse mammary gland epithelium isattained through DNA synthesis inconsequent of mitosis.Devi Biol. 88, 167-179.'

ToKER, C. (1967). Observations on the ultrastructure of amammary ductule. J. Ultraslmct. Res. 21, 9-25.

TOPPER, Y. J., OKA, T. & VONDERHAAR, B. K. (1975).

Techniques for studying development of normalepithelial cells in organ culture. Meth. linsvw. 39,443-454.

VONDERHAAR, B. K. & SMITH, G. H. (1982). Dissociationof cytological and functional differentiation in virginmouse mammary gland during inhibition of DNAsynthesis. J . Cell Sci. 53, 97-114.

WICHA, M. S., LIOTTA, L. A. & KIDWELL, W. R. (1979).

Effects of free fatty acids on the growth of normal andneoplastic rat mammary epithelial cells. Cancer Res. 39,426-435.

WILLIAMS, J. M. & DANIEL, C. W. (1983). Mammaryductal elongation: differentiation of myoepithelium andbasal lamina during branching morphogenesis. Devi Biol.97, 274-290.

{Received 24 November 1987 -Accepted 15 January 198S)

A latent stem cell in mammary epithelium 183