Laboratory of Mammary Gland Development


Why do we study mammary gland development?
Our past research activities in mammary gland development
Our current research activities in prenatal mammary gland development
Our current research activities in postnatal mammary gland development
Wider implications of our research activities

Why do we study mammary gland development?

figurineFigure 1. Statue of ‘La Nature’ at Chateau de Fontainebleau, France. Her polymastia indicates she is a Goddess of Fertility. Photo: J.M. Veltmaat.

The mammary glands, or breasts, are essential for the survival of mammalian species, as they provide the only - or for humans the healthiest- source of nutrients for the offspring. But did you ever wonder why goddesses of fertility are often depicted with polymastia, i.e. supernumerary breasts (Fig. 1), instead of with bigger breasts or many children? Could the choice for many breasts be based on a natural phenomenon of many pairs of mammary glands in other mammals, e.g. up to ten pairs in pigs?

sites of polymastia

Figure 2. Sites of polymastia in humans. The dots, representing the normal nipples, and sites of polymastia can be connected by imaginary lines called milk lines or mammary lines. Modified from: ABC of breast diseases. JM Dixon (Ed.).


In fact, the choice for a polymastic goddess of fertility could rest on the observation of polymastia occurring in two to ten percent of the human population. The position of these supernumerary mammary glands (either unilateral or bilateral) can range from the axilla (armpit) to the vulva.

Imaginary lines through all these possible positions on either side of the adult body (Fig. 2) are called mammary lines or milk lines. During embryonic life, these lines are actually anatomically or histologically discernable structures, extending from the axilla to the inguen (groin). They are local thickenings of the surface ectoderm (i.e. the precursor of the epidermal component of the skin), and can in most studied mammalian species be discriminated as an elevated ridge in the ectodermal landscape of the flank.

Depending on the species, one or several positions on this so called mammary line proceed to form mammary glands. So how does the surface ectoderm make a mammary line? What determines the position of the mammary line, and the variation and number of mammary glands that develop on this line in the various mammalian species? Which developmental events are responsible for anomalies such as polymastia and amastia, i.e. the absence of breasts, or other growth and morphogenetic defects?

These questions drive the research in the Veltmaat laboratory. We use the mouse with its five pairs of mammary glands as a research model, and combine forward and reverse genetics with molecular and developmental biology techniques for phenotypic analyses.




















Our past research activities in mammary gland development

Identification of the mammary line
We started with the question how the mammary line is formed and how mammary glands are formed on this line. Zoologists in the late 19th and throughout the 20th century had not been able to discriminate an elevated ridge in the mouse embryo. Instead, some proposed that a band of enlarged ectodermal cells on the flank in embryonic day 11 (E11) mouse embryos represented the mammary line, which was debated by others (reviewed in Veltmaat et al., 2003).

By performing a whole mount in situ hybridization screen on mouse embryos with a large panel of candidate molecular markers for the mammary line, we demonstrated the co-localization of Wnt10b expression with this band of enlarged ectodermal cells (Fig. 3; Veltmaat et al., 2004). Along and within this line of Wnt10b expression subsequently appear five disc-shaped placodes that undergo a series of morphogenetic changes to become branched mammary trees before birth (Fig. 4; Veltmaat et al., 2003), confirming that this band of enlarging and multilayering surface ectodermal cells is indeed the mammary line in mouse embryos.

Interestingly, Wnt10b is first expressed as a series of fragments, overlying the tips of the somites - i.e. the subdermal precursors for vertebrae, ribs, pectoral muscles, and dermal mesenchyme -, suggesting the involvement of somitic signals in mammary line and placode formation.

figure3   Figure 3. Identification of the mammary line in E11.5 mouse embryos by local multilayering and Wnt10b expression. (A) Cartoon of an E10.5 mouse embryo showing the position of the prospective five mammary rudiments as dots by number, with rudiments 1 and 5 hidden behind forelimb and hindlimb respectively. Rudiments 2, 3 and 4 will form in close apposition to the tip of the somites (grey outlined repetitive structures with numbers in the back of the embryos and green outlined in C)). Red boxed area in (A) is expanded in (B) and the solid red line indicates the plane of section in (C), (D) and (E). (B) Whole mount in situ hybridization for Wnt10b showing the mammary line with emerging 3rd and 4th mammary rudiment. Note that the mammary line is fragmented. (C) Whole transversal section at the level of the red line in (A) showing the position of these local Wnt10b-positive sites of ectoderm on the flank. (D) High magnification of the red-boxed area in (C) visualizing Wnt10b expression (purple) in the multilayered ectodermal area, while the adjacent single-layered ectoderm and the supporting dermal mesenchyme are negative. (E) High magnification of the same area as in (D) but subjected to a histological stain to illustrate that Wnt10b expression co-localizes with local multilayering of the surface ectoderm at the boundary between the squamous ventral surface ectoderm (left) and cuboidal lateral surface ectoderm (right). Modified from, and for more details see: Veltmaat et al., 2004.
Figure 4. Cartoon depicting cross-sections through mammary rudiments at sequential morphogenetic stages of their development in the female mouse embryo between E11 and birth. The five mammary placodes appear asynchronously and not in rostro-caudal or dorsoventral sequence. They go through the sequential morphogenetic stages at their own speed and with individual minor deviations from the depicted shapes, see also Fig. 9. Modified from, and for more details see: Veltmaat et al., 2003.
back to top

Unique identity of each of the mammary gland pairs
Two to ten percent of the human population exhibit polymastia. The positions of these supernumerary mammary glands vary (Fig. 2), and some positions predominate over others depending on ethnicity. This dependency strongly suggests that different signalling thresholds or pathways regulate mammary gland development at different levels along the rostro-caudal (head to tail) body axis. Indeed, our studies on a variety of genetic mouse mutants confirm this position specific requirement for certain genes, as indicated by the induction defect of some but not all mammary placodes in the absence of the genes encoding Fibroblast Growth Factor 10 (FGF10) or its main receptor FGFR2b, the transcription factors Pax3 or Gli3, or in FGF10 hypomorphs expressing reduced levels of Fgf10. The subset of affected mammary placodes varies dependent on the gene that is mutated (Fig. 5; Veltmaat et al., 2006), confirming the hypothesis that at different positions along the body axis, different molecular mechanisms regulate the induction of mammary rudiments. This is a particularly interesting developmental phenomenon, because other repetitive structures such as limbs and somites share their molecular mechanisms of induction.

Figure 5. Lack of placode induction in the context of gene ablation. (Two upper panel rows): Transversal sections through the mammary line at the level of placode 3 in 43 to 45 somite (s) stage (i.e. E11.25-E11.5) embryos show that these mammary induction defects are due to a failure of the surface ectoderm to enlarge and multilayer in Fgf10-/-and Fgfr2b-/- embryos; or failure to multilayer in Gli3Xt-J/Xt-J and Fgf10 hypomorphs; or to multilayer late in Pax ILZ/ILZembryos. Abbreviations: Sg = stratum generativum; si = stratum intermedium; p = periderm. Asterisks demarcate the width of the mammary line, as determined by the formation of the stratum intermedium. Note the differences in size and orientation of nuclei in the Sg, and numbers of cell layers in the surface ectoderm. (Two lower panel rows): Whole mount in situ hybridization of wild type control (ct) and various mutant mouse embryos with a Wnt10b or Lef1 probe visualizes the mammary rudiments (indicated with numbers). In the absence of Fgf10 (and similar for the absence of Fgfr2b), only the 4th mammary rudiment forms, while Fgf10 hypomorphs induce mammary rudiments 1, 2, 4 and 5. Gli3Xt-J/Xt-J (null) mutants fail to induce mammary placodes 3 and 5, while Pax ILZ/ILZ (null) embryos form a late and hypoplastic rudiment 3. Modified from, and for more details see: Veltmaat et al., 2006.


Somitic signals are required for induction of mammary rudiment 3 and to determine its position on the dorso-ventral body axis
The phenotypes of null mutants for Fgf10, Fgfr2b and Gli3 partially overlap in the absence of mammary placodes 3 and 5. Our further studies to understand the relationship between these genes in induction of these placodes revealed an epistatic interaction between Gli3 and Fgf10 in the somites, required for the induction of mammary rudiment 3. On one hand, this confirmed the implication of somitic signals in the mammary line and rudiment formation, albeit it restricted to mammary rudiment 3 as far as somitic Gli3 and Fgf10 are concerned. Moreover, the extent of somitic elongation towards the ventral body part determines the position of mammary placode 3 on the dorso-ventral body axis. On the other hand, these data showed that the ectoderm is dosage-sensitive to somitic Fgf10: While low levels of somitic Fgf10 can still induce low levels of Wnt10b and evoke ectodermal cell enlargement as an indication of mammary line formation, Wnt10b expression is highest above the somite (approximately somite 16) with the highest Fgf10 expression; and only there the surface ectoderm will multilayer as the initial step in placode formation (Fig. 6; Veltmaat et al., 2006). The mechanism of action of Gli3 and Fgf10 in placode 5 remains to be uncovered.

Figure 6

Figure 6. Somitic signals induce mammary placode 3 in the mouse embryo.
Somitic Gli3 expression is indirectly required, together with an unknown factor or factors X, for sufficient somitic Fgf10 expression. This leads to sufficient FGFR2b-activation and downstream Wnt10b expression and Wnt signalling in the surface ectoderm, concomitant with enlargement and multilayering of the surface ectoderm as the initiation of mammary gland formation. This model is only true for the induction of mammary rudiment 3. Modified from, and for more details see: Veltmaat et al., 2006.



Cellular mechanisms of mammary rudiment induction and growth
Given the ‘molecular individuality’ of each of the five mammary placodes in the mouse embryo, we investigated whether the differences in molecular and tissue interactions required for each mammary placode result in differences in cellular mechanisms of mammary placode formation. We identified that although mammary rudiments vary in their growth speed (Fig. 7 and 9; Lee et al., 2011), the cellular mechanisms of growth are very similar: Cell proliferation within the mammary rudiment hardly contributes to rudiment growth (Fig. 8; Lee et al., 2011). Instead, all mammary rudiments grow initially by recruitment of ectodermal cells (Fig. 9; Lee et al., 2011), soon complemented by hypertrophy of the basal cells within the mammary rudiment, and growth is hardly counteracted by apoptotic cell death. The molecular regulation of these growth mechanisms vary per mammary rudiment, as indicated by a differential contribution of Gli3 to each of these mechanisms per mammary rudiment (Fig. 10; Lee et al., 2011).


Figure 7. Growth rates vary among the five mammary rudiments (MR) in wild type mouse embryos. Mammary rudiment 3 is the first to emerge, at around E11.25 (not shown); but is soon exceeded in size by rudiment 1. Mammary rudiment 4 emerges later, but within two days it is the largest mammary rudiment. Mammary rudiment 5 appears next, but because it remains flat for a long time, its size can only be measured from E12.5 onwards, and then grows in pace with mammary rudiment 1. Mammary rudiment 2 emerges at around E11.75 as the last one to form, but grows so rapidly that it almost catches up with rudiments 1 and 3 by E13.5. Modified from, and for more details see: Lee et al., 2011.

figure8 Figure 8. Cell proliferation is not a major mechanism of mammary rudiment growth. Between E11.25 and E13.5, the percentage of BrdU-labelled cells (black) in mammary rudiments, as outlined by the yellow dashed lines, is very low compared to that of the adjacent surface ectoderm and surrounding mesenchyme. In collaboration with imaging analyst Dr. Victor Racine and software engineer Peter Jagadpramana, we developed software that segments histological images in the following compartments: a neck (n), periphery (p) and core (c), as well as the surrounding mammary mesenchyme (mm) and dermal mesenchyme (dm) and adjacent ventral and dorsal ectoderm (v ect and d ect, respectively), as shown in the lower panel. The software calculates the volume of the mammary rudiment (as shown in Fig. 7) by cumulative segmentation data of all consecutive sections through a rudiment, and quantifies BrdU incorporation in each compartment. Such analysis indicates that the mammary rudiments do not primarily grow by cell proliferation. Tracing BrdU-labelled cells 24 hours after labelling showed a significant increase in labelling within the mammary rudiments (not shown), indicative of ectodermal influx during that period as a primary contributor to mammary rudiment growth. Modified from, and for more details see: Lee et al., 2011.
Figure 9. 3D-reconstructions of five mammary rudiments reveal differences in growth and morphogenesis. Mammary rudiments were exposed in vivo to BrdU at E12.5 and harvested 2 or 24 hours later, sectioned, stained for BrdU-immunodectection, and 3D-reconstructed. The ectoderm is depicted in green, mammary rudiments in red, and BrdU positivity in blue within the mammary bud and in black within the ectoderm. The differences in shape and growth rate of the individual rudiments become readily visible by comparison of the red shapes. The low BrdU-incorporation rate of the mammary bud compared to the ectoderm at E12.5 (and similar at E11.5 and E13.5; see Fig. 8) indicate that cell proliferation within the mammary rudiment hardly contributes to rudiment growth. By contrast, the high labelling density of the rudiments at 24 hours post labelling indicates cells have been recruited from the surface ectoderm. Data generated in collaboration with imaging analysts Dr. Weimiao Yu and Dr. Tiehua Du. Modified from, and for more details see: Lee et al., 2011.
figure10 Figure 10. Mammary rudiments grow predominantly by ectodermal influx and cell hypertrophy, which are differentially regulated by Gli3 among the mammary rudiments. Mammary rudiments grow primarily by ectodermal recruitment/migration, soon followed by cell hypertrophy of the basal cells of the mammary rudiment, as depicted by cumulative estimated values along the y-axis in the graph above. While the relative contributions of these cellular mechanisms of growth are similar for each of the five mammary rudiments, the regulatory function of Gli3 in these mechanisms varies among the five rudiments, as indicated with the activating or blocking arrows flanked by the identity (number) of mammary rudiments to which the level of action applies. Modified from, and for more details see: Lee et al., 2011.
back to top
Our current research activities in prenatal mammary gland development



Figure 11. Each mammary rudiment displays a distinct phenotype in the absence of Gli3. Haematoxylin/Eosin counterstained transversal sections through the five mammary rudiments in E12.5 wild type (wt) and Gli3Xt-J/Xt-J (null) embryos. The mammary epithelium is stained blue, which reveals the distinct morphological and growth aberrancies for each mammary rudiment. For more details, see Lee et al., 2011.


Besides the failure to induce mammary rudiments 3 and 5, Gli3Xt-J/Xt-J (null) mutants also form hypoplastic mammary rudiments 2 and 4, and their rudiment 2 is moreover protruding (Fig. 11; Lee et al., 2011). We are currently investigating the molecular mechanisms downstream of Gli3 in the formation of each of the five mammary rudiments, and how these downstream mechanisms contribute to the cellular mechanisms of growth and morphogenesis.

In order to identify more markers for and molecular regulators of the early phases of mammary gland development, we have developed a technique that facilitates the isolation of these three tissues with minimal cross-contamination (Fig. 12) and high integrity of the transcription profile and RNA quality (Sun et al. 2011).

figure12Figure 12. Microsurgically dissected mammary rudiments and their adjacent ectoderm and surrounding mesenchyme of E12.5 mouse embryos. Modified from, and for more details see: Sun et al., 2011.

Our RNA microarrays on the five mammary rudiments versus each other, the adjacent ectoderm and surrounding mesenchyme have revealed that each mammary rudiment has a unique expression profile (Fig. 13). We are currently validating these microarray results by whole mount in situ expression studies, and performing functional studies on validated genes to determine whether and which role they play in the formation and further development of mammary placodes.
figure13 Figure 13. The five mammary rudiments each have a unique expression pattern. The five mammary rudiments at bud stage (B1-B5), the adjacent surface ectoderm (E) and surrounding mesenchyme (M) were isolated as shown in Fig. 12 and subjected to RNA microarray. All genes that are differentially expressed among the 8 tissue samples are arranged from left to right. Circled red areas indicate gene clusters that are highly expressed in mammary epithelium compared to ectoderm and mesenchyme, and moreover differentially expressed among the five mammary rudiments. For more details, see Sun et al., 2011.
back to top
Our current research activities in postnatal mammary gland development
Most of mammary gland development occurs during postnatal life, most notably the elongation and further branching of the mammary tree during puberty, as well as side branching and alveologenesis during pregnancy (Fig. 14). Currently, we are extending our studies on our genes of interest for prenatal mammary development to their roles in postnatal mammary development.
figure 14
Figure 14. Normal postnatal development of the murine mammary gland. Defatted and carmine-red stained inguinal sub-dermal fat pads of wild type C57BL/6J female mice, visualizing the epithelial tree of the mammary gland. Bold arrows indicate the position of the nipple (N), thin arrow indicates the lymph node (LN). Small black arrowheads indicate ductal ends. Note that the mammary tree occupies less than 1/10 of the fat pad in a 3-week old virgin. During puberty, the ductal ends form club shaped terminal end buds (TEBs) from which new branches form and rapidly elongate between the ages of 4 and 7 weeks. The ductal ends reach the distal edge of the fat pad before 9 weeks of age. From then, the epithelial tree grows congruently with the fat pad and the mouse itself. Under the influence of hormonal surges during repeated oestrous cycles, the epithelial tree starts to show signs of alveologenesis (formation of small structures that branch off the ducts) between 26 and 52 weeks of age.
back to top
One major difference between prenatal and postnatal mammary gland morphogenesis is that the postnatal morphogenetic events such as branching and alveologenesis are so numerous, that it is too tedious to quantify them manually. Different mammary biology research groups have developed different shortcut methods, which all still rely on manual, and moreover subjective, input of the biologist. To facilitate the analysis of postnatal mammary gland development and to provide an opportunity for standardization of such analyses among the various mammary gland biology laboratories, we have developed algorithms for automated analysis of images of whole mounted, carmine-red stained mammary glands (Fig. 15), in collaboration with Dr. Yan Nei Law and Dr. Hwee Kuan Lee at A*STAR BioInformatics Institute, Singapore. We are currently implementing these algorithms in a user-friendly on-line software named MammoQuant.
Figure 15. MammoQuant detects and quantifies a wide range of morphogenetic parameters of the mammary gland. (A) Image of a mouse mammary gland prior to analysis.
MammoQuant-generated Quality Control image of image A, visualizing the segmentation of the fat pad (red contour), epithelially invaded area (green contour) and mammary epithelium proper (blue contour). (C and D) MammoQuant-generated Quality Control images visualizing the fat pad and epithelial contour (both in red) as well as the longitudinal and lateral axes of growth for the fat pad (yellow and green) and epithelium (light and dark blue). (E) Higher magnification image showing part of a carmine-red stained mammary gland prior to analysis. (F) MammoQuant-generated Quality Control image of image E with high magnification inset visualizing segmentation of the main mammary branches (yellow), bifurcation points (red), side branches/alveoli (blue) and ductal end points (green). Areas within the contours in B-D, length of the growth axes in C and D, length and average width of individual ductal segments, number of branching and endpoints are calculated and exported to an Excel file for further analysis by the biologist.
Wider implications of our research activities
The mammary gland is a fascinating study object for developmental biologists: It is a repetitive structure, but very different from other repetitive structures in the sense that its multiple copies arise in a non-linear fashion along the main body axes, and that the molecular mechanisms of induction of these multiple copies vary considerably and not in a gradient. The initiation of mammary gland formation poses interesting questions with regards to pattern formation in the ectoderm. The anatomical similarity between mammary glands and other ectodermal appendages such as teeth, feathers and hair follicles in their earliest morphogenetic stages raises the question what the distinctive signals are that give these organs their own identity. Furthermore, mammary glands undergo extensive morphogenetic events, the most obvious of all branching morphogenesis. However, the branching patterns of the five mammary gland pairs in the mouse differ from each other, there is also no mirror symmetry between the two glands of one pair, and the pattern of each individual gland differs notably between individual mice. This lack of stereotypy is in stark contrast to for example the stereotypic branching pattern of the lung. And the list of reasons doesn’t end here.

Understanding the mechanisms of mammary gland induction and development will help us to understand breast anomalies in the human population. For example, patients with Poland Syndrome exhibit breast hypoplasia with an absent or underdeveloped underlying pectoral muscle (mostly unilateral), sometimes combined with ipsilateral syndactyly. While the cause for this birth defect is unknown, many surgeons hypothesize it is due to a lack of arterial blood supply during embryonic life. Our data showing the requirement of somitic signals for induction of the third mammary gland pair, which resides at a similar location as the human breasts, may indicate that loss of heterozygosity in one body half could explain this defect as well. In other syndromes, breast anomalies occur with defective internal organs which may create pathological conditions later in life. The breast anomaly can serve as an early diagnostic marker for those internal defects, while understanding the molecular cause of the breast anomaly may pave ways for the development of interventive or preventive therapeutics for those internal defects.

Besides being an appealing model for developmental biologists, the mammary gland is also a highly regenerative organ, containing many stem cells that are reactivated with every sequential pregnancy. It is also an organ that is highly susceptible to tumour formation. The fields of developmental biology, stem cell biology and tumour biology are closely related, as they revolve about fate decisions that cells make. In other words, progress in one of these research fields will benefit the other. For example, while the molecular mechanisms of most of the non-familial breast tumours still remain elusive, there is a general consensus that the signalling pathways governing organogenesis are often, with deregulated activity, associated with tumourigenesis. Indeed, most of the genes that we and others have shown to be indispensable for the induction of mammary gland formation are associated with breast cancer. While the complexity of the adult mammary gland and heterogeneity within and among breast tumours impedes the identification of new breast tumour genes, the embryonic mammary rudiment is simple in its tissue composition and well experimentally accessible. Therefore, the study of the embryonic phase of mammary gland formation can also provide us important insights in molecular mechanisms of cell fate decisions relevant to breast cancer.

back to top