Mammary epithelial cells undergo secretory differentiation in cycling virgins but require pregnancy for the establishment of terminal differentiation

The mammary gland provides an excellent system to study questions pertaining to organogenesis, cell differentiation and oncogenesis. The development of the gland is to a great extent governed by hormonal stimuli (Forsyth, 1986; Topper and Freeman, 1980). Development of the gland progresses gradually and is completed only in the mature organism. In newborn mice, the mammary anlagen consist of only a few ducts which under the influence of gonadal hormones during puberty and adolescence grow into the mammary fat pad. In virgin mice, the mammary parenchyma is composed of a highly organized system of ducts with terminal end buds which are the major sites of growth and lateral buds spaced regularly along the ducts. With each estrous cycle, the lateral buds differentiate. They subdivide progressively thus giving rise to small alveolar buds (Daniel and Silberstein, 1987; Russo et al., 1989). Further development and differentiation of the gland takes place during pregnancy and lactation. Between day 8 and 10 of gestation, the mammary alveoli start to form larger alveolar lobules. A rapid increase in the number and size of alveoli occurs during the second half of pregnancy resulting in the development of fully differentiated secretory lobules (Pitelka et al., 1973).

These morphogenetic events are accompanied by cellular differentiation processes leading to the development of secretory epithelial cells which are capable of synthesizing and secreting milk proteins. Low levels of milk protein genes are expressed already in virgin mice, but their synthesis increases dramatically during pregnancy following a characteristic time course: β-casein is seen earlier than the whey acidic protein (WAP) and α-lactalbumin is found only toward the end of pregnancy. Recently, WDNM1 has been identified as a milk protein gene whose expression also increases strongly during pregnancy (Morrison and Leder, 1994). Maximal expression levels of the milk protein genes are reached only during lactation.

Although lactogenic hormones are responsible for the overall induction of milk protein gene expression, they do not seem sufficient to explain the complex temporal regulation observed in the developing gland during pregnancy. Studies employing cell and organ culture as well as transgenic animals have provided information about some of the regulatory elements involved in mammary-specific expression (Burdon et al., 1991a; Pittius et al., 1988; Shamay et al., 1992). Response elements for steroid hormones are present in the promoters of β-casein and WAP (Li and Rosen, 1994, 1995). Stat 5, a mammary gland factor involved in mediating the prolactin response has been identified and shown to bind to a specific region of the β-casein (Wakao et al., 1994) and WAP gene promoter (Li and Rosen, 1994, 1995). Interestingly, differences in the regulation between milk protein genes have been noted. Whereas the ‘early’ casein genes rely for their expression to a large degree on prolactin and to a lesser extent on glucocorticoids, the role of the two hormones is reversed for the ‘late’ WAP gene (Burdon et al., 1991a; Pittius et al., 1988; Shamay et al., 1992). Furthermore, the requirements for expression of β-casein and WAP in mammary cell culture have been found to be different. While β-casein is expressed by dissociated mammary epithelial cells embedded into extracellular matrix (Streuli et al., 1991), cell-to-cell contact and formation of a closed lumen are required for expression of WAP (Chen and Bissell, 1989). This strongly suggests that in addition to the lactogenic hormones, local growth factors, extracellular matrix components and a proper three-dimensional structure of the epithelium are involved in their transcriptional control (Lin and Bissell, 1993). Mammary tissue from embryos and immature virgins can be induced to express milk protein genes and to synthesize the corresponding proteins, such as β-casein and WAP, when cultured in the presence of lactogenic hormones (Pittius et al., 1988; Shamay et al., 1992). This demonstrates that alveolar cells in non-primed virgins have the capability to undergo functional differentiation in response to lactogenic hormones.

Based on the ultrastructural appearance in mid-pregnancy, it has been suggested that alveolar development is not synchronized (Mills and Topper, 1970). By morphological and immunological criteria, such as the binding of antibodies and lectins, mammary alveolar cells within individual alveoli appeared homogeneous (Ormerod and Rudland, 1984; Rudland and Hughes, 1989). However, differential keratin expression was seen among lobular alveolar cells and was attributed to the presence of dispersed latent mammogenic stem cells (Smith et al., 1990). In situ hybridization studies with ³⁵S-labeled probes for α-lactalbumin, α-S1-casein and lactoferrin have revealed large-scale regional differences throughout the mammary glands of sheep and cattle (Molenaar et al., 1992) but, due to the lower resolution of this technique, it was not possible to identify intra-alveolar differences. In another study, fluorescently labeled oligonucleotides were used to colocalize transcripts of different milk protein genes in the mammary glands of the tammar wallaby at late lactation (Joseph and Collet, 1994). At this stage, all epithelial cells in a given alveolus synthesize all milk proteins simultaneously.

In an attempt to understand the complex regulation of mammary cell differentiation, we studied alveolar cell differentiation in normal mice and in transgenic mice, which cannot develop a functional mammary gland because of the aberrant expression of a developmental protein (Burdon et al., 1991b). We performed a detailed study of the expression pattern of a mammary-gland-specific reporter gene and four major milk protein genes during pregnancy. We employed in situ hybridization utilizing non-radioactive detection methods to analyze gene transcription on a cellular level. Our results showed activation of milk protein genes in virgin animals during the estrous cycle in a small number of mammary cells. The development of fully differentiated secretory epithelial cells required additional stimuli arising during pregnancy. Perturbation of the temporal expression of a milk protein gene in transgenic mice inhibited terminal differentiation.

C57BL/6 females were used for these studies. To obtain dated pregnancies 4-to 6-week-old virgins were superovulated and mated. The day after copulation was counted as day 0 of pregnancy. The transgenic mice carrying a mouse WAP transgene have been described earlier (Burdon et al., 1991b). The WAP-lacZ transgene was generated by ligating a 4.2 kb fragment spanning the mouse WAP gene promoter from −4200 to +24 to the bacterial β-galactosidase gene (Pharmacia). This promoter has been shown to target transcription exclusively to mammary tissue (Paleyanda et al., 1994) Transgenic mice were generated that carried approximately 2 to 10 copies of the transgene. Two lines were obtained that expressed the transgene and one was analyzed in detail.

Tissues were fixed for 2 hours in 2% paraformaldehyde, 0.25% glutaraldehyde, 0.01% NP40 in PBS, rinsed in PBS and washed in X-gal staining buffer (30 mM K₃Fe(CN)₆, 30 mM K₄Fe(CN)₆, 2 mM MgCl₂, 0.01% sodium deoxycholate, 0.02% NP40, in PBS) for 2 hours. The wash buffer was replaced with staining buffer containing 1 mg/ml X-gal. After sufficient color development, the tissues were transferred to 70% ethanol and processed for paraffin sectioning according to standard procedures. 5 μm sections were prepared and stained with nuclear fast red.

Mammary gland tissue was fixed in 4% paraformaldehyde in PBS and embedded in paraffin by standard methods. 5 μm sections were mounted on silan-coated slides and used for in situ hybridization. Sections were deparaffinized, treated with 2 μg/ml proteinase K for 10 minutes at room temperature, postfixed in 4% paraformaldehyde and acetylated. cRNA probes were hybridized overnight at 50°C in 50% formamide, 5× SSC, 0.1 M sodium phosphate buffer pH 7, 1× Denhardt’s solution, 100 μg/ml sonicated salmon sperm DNA and 100 μg/ml yeast tRNA. Washes were performed at 65°C in 50% formamide/2× SSC (2 times 30 minutes) and 25% formamide/1× SSC/1× PBS (2 times 30 minutes). After treating with 5 μg/ml RNAse A (Boehringer) for 10 minutes at 37°C the sections were subject to color development using the Boehringer Genius 3 kit according to the manufacturer’s recommendations.

Digoxigenin-labeled probes were prepared from appropriately linearized plasmids using the Boehringer Genius 4 kit. A 250 bp cDNA sequence was used for β-casein, a 450 bp cDNA sequence for WAP, a 450 bp cDNA sequence for α-lactalbumin, and a 450 bp cDNA for WDNM1. The cDNA sequences spanned protein coding regions, and they were inserted into pBS (Stratagene, La Jolla).

Paraffin sections were prepared as described above. Anti-WAP antibodies (Shamay et al., 1991) were diluted 1:500 and applied overnight at 4°C. Biotinylated secondary antibodies were detected with a Vectastain ABC kit.

RNA was prepared according to Chomczynski and Sacchi (1987). Total RNA was separated on 1.2% formaldehyde-agarose gels, transferred to Nylon membranes (GeneScreen Plus) and hybridized. ³²P-labeled probes (see above) were endlabeled (oligonucleotides) or labeled by random priming (cDNAs) and used in the following order: transgene-specific oligonucleotide (Burdon et al., 1991a), α-lactalbumin, WDNM1, WAP, β-casein, 18S rRNA oligonucleotide. Hybridized probes were removed by dipping the filter in boiling water for 3 to 5 minutes. Removal of probes was checked by exposing filters overnight between hybridizations.

To understand the differentiation of mammary alveolar cells during puberty, pregnancy and lactation, we have analyzed the expression patterns of a mammary-specific transgene and of endogenous milk protein genes in normal mice, and in transgenic mice carrying a milk protein transgene whose expression results in aberrant mammary development.

We have generated transgenic mice that carry a hybrid gene consisting of 4.2 kb of the WAP gene promoter and the bacterial lacZ gene. This permitted gene expression studies with high sensitivity and single cell resolution. Expression of the transgene was detected in mammary alveolar and ductal cells of virgin mice during estrus (Fig. 1A). The activity per expressing cell was less than 10% of that found during lactation (see legend of Fig. 1). Transgene expression was transient, as we did not find β-galactosidase activity in mammary glands of transgenic mice in diestrus (Fig. 1C). No β-galactosidase activity was detected in ductal and alveolar mammary cells of non-transgenic mice in estrus (Fig. 1B).

Both the number of β-galactosidase-positive cells and the level of transgene expression increased during pregnancy, and on day 15 of pregnancy substantial expression was seen in a large number of alveolar cells (Fig. 1D). The β-galactosidase staining was preferentially observed in those cells that had a more differentiated appearance based on the presence of vacuoles. On day 18 of pregnancy, the majority of alveolar cells expressed the transgene (Fig. 1E), and on day 1 of lactation bacterial β-galactosidase was found in all alveolar cells (Fig. 1F). However, even at this stage expression was not uniform, and quantitative differences were observed between individual alveolar cells.

To corroborate the data obtained with the WAP-lacZ reporter gene, we next analyzed the activation of endogenous milk protein genes (WDNM1, β-casein, WAP and α-lactalbumin) using in situ hybridization of digoxigenin-labeled riboprobes to tissue sections. Similar to the lacZ reporter gene, this method permits the precise cellular localization of the detection signal. A distinct hybridization signal was obtained in mammary tissue at day 9 of pregnancy with probes for WDNM1 (Fig. 2A) and β-casein (Fig. 2B). It is evident that the expression pattern for these genes is not homogeneous. WDNM1 expression was seen in about half of the alveoli. The signal was strictly localized to alveolar cells and was not found in ductal or myoepithelial cells or in the mammary stroma. β-casein was expressed only in a few sporadic alveolar cells. On day 11 of pregnancy, alveolar development had progressed, and WDNM1 expression was seen in most of the alveoli and in the interlobular terminal ducts (Fig. 2C). The number of alveoli that expressed β-casein had increased to approximately 30% (Fig. 2D). The same irregular distribution of β-casein transcripts was seen as before, and frequently only a few cells of one alveolus showed a strong hybridization signal. No WAP-or α-lactalbumin-specific signals were observed at these stages (data not shown).

In 14 day pregnant mice, the number of cells and the size of the alveoli had increased, and a lumen was clearly visible (Fig. 2E-G). In addition to an increase of the WDNM1 (Fig. 2E) and β-casein (Fig. 2F) signals, expression of WAP was now detected (Fig. 2G). Expression of WDNM1 was essentially uniform within alveoli, but differences in intensity occurred between alveoli. In general, a stronger hybridization signal was seen in more distal alveoli at the periphery of the gland. This regionalization of the gland has also been observed for the formation of alveolar buds (Russo et al., 1989). It appears that the most recently formed alveoli, (i.e. distal to the nipple) are more active than the more proximal older structures. Heterogeneous hybridization signals were obtained with a WAP-specific probe (Fig. 2G). WAP RNA was found sporadically and in only a few cells within any given alveolus. This pattern resembled that observed for β-casein expression at earlier stages of pregnancy (Fig. 2B,D).

Little change in the expression pattern was observed for WDNM1 between day 14 and 16 of pregnancy (Fig. 2H). The expression of β-casein RNA was slightly increased (Fig. 2I). A hybridization signal was seen already after 2-4 hours of color reaction while it took between 16 and 20 hours for a signal to develop in the earlier stages (see legend of Fig. 2). Occasional low expression of β-casein was seen also in small ducts (Fig. 2I). The signal for WAP was still uneven at this stage, but the number of WAP-expressing alveoli had increased (Fig. 2J). Analysis of closely adjacent sections showed that WAP expression was mainly found in alveoli that showed strong expression of β-casein (data not shown). The heterogeneity of milk protein expression and cellular differentiation at this stage was clearly visible at a higher magnification of sections after a short color reaction (Fig. 3A-C). Mosaic expression is particularly dramatic for the WAP gene where, in some cases, RNA was found in only 1 or 2 cells within an alveolus (Fig. 3C). The cells within one alveolus displayed great variability in the intensity of hybridization as well as in the size and number of intracellular vacuoles. It appears that the highest expression levels of any milk protein gene were found in cells with large vacuoles. This indicates a coordinated morphological and functional differentiation of the alveolar cells.

On day 18 of gestation the mammary alveoli are widely dilated, suggesting extensive secretory activity. A strong and more uniform signal for WDNM1 (Fig. 2K), β-casein (Fig. 2L) and WAP (Fig. 2M) was obtained. A much weaker signal was now also seen for α-lactalbumin (Fig. 2N).

In the lactating gland, all the alveolar secretory cells expressed the four milk proteins at a high, but still not at a uniform level (Fig. 2O-R). Lower levels of expression were also seen in ductal cells.

The heterogeneous expression of WAP transcripts seen in midpregnancy was compared to the distribution of WAP detectable by immunohistochemical methods (Fig. 4). On day 16 (Fig. 4A,B) and 18 (Fig. 4C,D) of pregnancy, antibodies directed against WAP showed an irregular distribution of expressing cells within individual alveoli. However, most of the immunoreactive material was concentrated in the alveolar lumen. Homogeneous staining of the alveolar cells for WAP was seen during lactation (Fig. 4E,F). The absence of staining in the lumen suggests that the gland had been emptied by continuous suckling.

These results demonstrate that the mosaic expression pattern seen by in situ hybridization reflects heterogeneities in the synthetic activities of the alveolar cells and is not an artifact created by fixation and tissue processing. In addition, we saw the same distribution with frozen sections (data not shown). Lastly, we obtained differential staining patterns characteristic for a particular gene in closely adjacent sections; i.e., while WDNM1 expression in 16-and 18-day glands is uniform in many alveoli, probes for WAP and α-lactalbumin give very spotty patterns (see Fig. 3).

Experiments employing transgenic animals have shown that the deregulated expression of putative developmental genes under the control of the mouse WAP gene promoter can alter and interfere with mammary gland development (Burdon et al., 1991b; Jhappan et al., 1993; Sympson et al., 1994). To evaluate how such changes affect cell differentiation during pregnancy, we took advantage of a line of transgenic mice that carries a mouse WAP transgene (Burdon et al., 1991b). Female mice from this line cannot lactate (the ‘milchlos’ phenotype) and exhibit incomplete mammary gland development during pregnancy. This defect appears to be correlated with a precocious expression of the transgene (Burdon et al., 1991b). In the original description of the ‘milchlos’ mouse, an incomplete penetration of the agalactic phenotype was observed. A wide range of histological variability was also seen during all stages of mammary development (Fig. 6D-I). It is possible that the histological variability results from mosaic expression of the transgene comparable to the activity of endogenous genes. We, therefore, analyzed the expression patterns of the WAP transgene and of endogenous milk protein genes by RNA blotting (Fig. 5) and by in situ hybridization (Fig. 6).

In Fig. 5, we compared the expression of milk protein genes in non-transgenic and WAP transgenic animals. In non-transgenic mice, only WDNM1 was expressed at significant levels on day 9 of gestation. The steady state level of this RNA peaked around day 14 of pregnancy, and no further increase was observed at later stages. β-casein RNA was detected on day 11 of pregnancy and its amount increased throughout pregnancy. A WAP signal was seen on day 16 of pregnancy, followed by at least a 30-fold induction on day 1 of lactation. The induction pattern of α-lactalbumin followed that of WAP, but at a lower level (note that the exposure time for the α-lactalbumin RNA in Fig. 5 was approximately 10-fold longer than that for the other milk protein genes). After a long exposure, low signals for all four genes were seen in virgin animals (not shown).

The expression of milk proteins in the WAP/WAP-transgenic mice was much more variable than in non-transgenic littermates (Fig. 5). This was particularly emphasized with the WDNM1 and α-lactalbumin genes. Precocious expression of the α-lactalbumin gene was see on day 14 of pregnancy. However, the level of RNA at late pregnancy and at parturition was only 5% of that found in non-transgenic mice. Furthermore, expression of the WAP, β-casein and WDNM1 genes remained steady throughout the second part of pregnancy and declined at parturition. This decline was particularly steep in the case of WDNM1. A possible explanation for the sharp decrease of WDNM1 would be the inability of alveolar cells to maintain a differentiation stage compatible with WDNM1 expression. The fact that α-lactalbumin RNA levels never increase during pregnancy further suggests that the cells never progress into the status of terminal differentiation.

By in situ hybridization, WAP RNA was seen on day 9 of pregnancy (Fig. 6C). As in the non-transgenic mice, when the milk protein genes are first expressed, transcripts are seen in only a few cells. This signal most likely comes from the transgenic RNA (see legend to Fig. 5). At day 9 of pregnancy, the expression level and pattern of WDNM1 (Fig. 6A) and β-casein (Fig. 6B) were similar to those in non-transgenic animals. On day 11 to 14, WAP was expressed in more alveoli but in the same irregular pattern as before. The hybridization signal for WDNM1 and β-casein was reduced compared to the non-transgenic animals (data not shown). In addition, from day 16 of pregnancy onwards, we observed distinct morphological differences between non-transgenic and transgenic animals. The alveoli of the transgenic mice appeared much more compact. The epithelial cells retained a cuboidal shape and contained more vacuoles (Fig. 6D-I; compared to non-transgenic animals in Fig. 2H-N). At these stages, the expression of WDNM1 (Fig. 6D,G) and β-casein was reduced (Fig. 6E,H), while the expression of WAP had increased (Fig. 6F,I). A weak signal was also obtained for α-lactalbumin (data not shown). It is important to emphasize that the expression pattern for all the milk protein genes remained much more heterogeneous. The alveoli never enlarged and the alveolar cells maintained their cuboidal shape. This clearly reflects the inability of these mammary glands to reach the fully differentiated state, which becomes evident in the inability of the WAP transgenic mice to nurse their pups. Toward the end of pregnancy and immediately after parturition, we saw an increase in apoptotic alveolar cells compared to non-transgenic animals (data not shown). It is possible that the cells exhibiting a deregulated expression of WAP and α-lactalbumin are eliminated and a loss of cells may contribute to the lack of alveolar development.

We have demonstrated that mammary epithelial cells in virgin mice go through a transient differentiation process: expression of milk protein genes occurs during estrus and then abates. Full differentiation is established and maintained only during pregnancy, and finally results in a nearly homogenous population of secretory cells. During pregnancy, the expression of individual genes varies greatly between neighboring cells within alveoli. This suggests that the onset and progression of differentiation occurs asynchronously among individual cells. Based on these results, we propose two distinct phases in mammary alveolar epithelial development (Fig. 7). First, a transient differentiation phase which is induced by ovarian hormones in the cycling virgin, and secondly, the establishment of terminal differentiation during pregnancy. The asynchronous development of alveolar cells during pregnancy indicates that the differentiation status of individual alveolar cells, as determined by the presence of systemic hormones, is modulated by local factors. Further asynchrony may result from temporal differences in cell proliferation events which give rise to terminally differentiated cells in the G₀ phase of the cell cycle.

The WAP gene promoter is active in approximately 30% of ductal and alveolar cells of virgin mice in estrus but not in diestrus or proestrus. The fate of these differentiated cells is not clear. They may cease expressing milk protein genes because of a change in hormonal signaling, but may persist and subsequently progress through pregnancy where they would resume a permanently differentiated state. Alternatively, these cells may die after having undergone differentiation. Evidence that differentiated cells persist in the virgin gland comes from transgenic mice that express growth modulators, such as the SV40T-antigen (Tzeng et al., 1993), TGFα (Eric Sandgren et al., unpublished data), TGFβ1 (Jhappan et al., 1993; Kordon et al., 1995;), a WAP transgene (Burdon et al., 1991b) or stromelysin (Sympson et al., 1994), under control of the WAP gene promoter. In these mice, phenotypic changes can be seen already in mature virgins.

The differentiation of mammary epithelial cells, and the concomitant expression of milk protein genes, is controlled by a network of regulatory signals. Two processes are required to produce a differentiated cell from its predetermined precursor. A program to cause the cell to be determined and a subsequent environment that allows the cell to differentiate along its predetermined course. During organ development, a predetermined stem cell produces two daughters. One remains a stem cell, the other differentiates. However, differentiating cells are capable of additional rounds of proliferation before eventually giving rise to terminally differentiated cells. The sequential expression of milk proteins in the proliferating and differentiating mammary epithelium is indicative of this process. The delay in onset of expression of α-lactalbumin and WAP in the majority of the differentiating secretory cell population suggests that these genes are expressed late in the differentiative pathway. In fact, their expression may be linked to terminal differentiation in the sense of the loss of proliferative capacity, i.e. entering a state of G₀. We propose that precocious expression of WAP results in terminal differentiation of alveolar cells already at mid-pregnancy, which in turn prevents the alveolar structures from proliferation.

The primary stimuli of mammary gland development during pregnancy are provided by progesterone, estrogen and lactogenic hormones. In addition, these hormones participate in the control of milk protein gene expression. The regulatory features imposed by these hormones vary between different milk protein genes. For example, the induction of WAP gene expression requires a synergistic activity of prolactin and glucocorticoids, but the β-casein gene is transcribed in the presence of prolactin alone (Hennighausen et al., 1988; Burdon et al., 1991a; McKnight et al., 1991; Shamay et al., 1992). Although response elements for the two signalling pathways have been identified in the promoter regions of both genes (Li and Rosen, 1994, 1995), their mode of action in the differential gene induction is not understood. Additional signals, possibly of autocrine or paracrine nature, are required to achieve the temporal coordination of mammary secretory development, differentiation and gene expression. This is supported by experiments with transgenic mice in which the expression of growth modulators is deregulated. TGFβ1 (Jhappan et al., 1993; Pierce et al., 1993; Kordon et al., 1995), stromelysin (Sympson et al., 1994), and WAP (Burdon et al., 1991b; Hennighausen et al., 1994) have been shown to influence mammary gland development and differentiation in vivo. Depending on the growth factor used and the temporal onset of expression, different phenotypes were observed. This suggests that these molecules intersect at different stages in the differentiation process (Robinson et al., unpublished data). TGFβ1 and WAP elicit their effect only over a very short range, and probably act through an intracrine or autocrine mechanism (Kordon et al., 1995; Hennighausen et al., 1994).

The phenotypic changes in ductal and alveolar development, induced by the deregulated expression of a particular growth regulator, can vary dramatically between individual animals. The penetrance of a phenotype is not only dependent on the site of integration, but likely also on the heterogenous and asynchronous expression imposed by the promoter in the transgene. As shown here in mice transgenic for the mouse WAP gene, the expression pattern of differentiation markers can vary between individual mice. Although the overall differentiation and development of mammary tissue in these mice is inhibited and results in the same milchlos phenotype, variations in the state of alveolar differentiation are observed between animals. Common to all transgenic mice, however, is the precocious expression of the WAP transgene and the inability of alveolar cells to enter a state of differentiation which permits full activation of the α-lactalbumin gene. This deregulated activation of growth regulators has been used successfully to study mechanisms underlying mammary development and differentiation. However, a full utilization of this approach requires a temporal control of transgene activation. Further experiments employing precise temporal regulation of gene expression using an inducible system such as the tetracycline responsive system (Furth et al., 1994; Hennighausen et al., 1995) will define more clearly the role of the extracellular matrix, growth factors and their receptors in the regulation of mammary gland development and lactation.