J. Embryol. exp. Morph. 73, 39-57, 1983
Printed in Great Britain © The Company of Biologists Limited 1983
39
Lumen formation in the developing mouse
mammary gland
By N. A. S. HOGG 1 , C. J. HARRISON 2 AND C. TICKLE 1 *
From the Department of Biology as applied to Medicine, The Middlesex
Hospital Medical School
SUMMARY
The mammary gland is a system of hollow interconnecting tubes which develops from an
invasive branching cord of epithelial cells. This ultrastructural study of the developing mammary gland focuses on how the lumen forms and establishes the polarized epithelial lining of
the gland. The earliest signs of lumen formation are many small cavities and crevices lined with
microvilli which appear at scattered sites throughout the branching cords and neck of the
gland. It is suggested that these initial small lumina form quite simply by separation of cells
whose opposing faces are non-adhesive. The continuous central lumen of the gland develops
by fusion and enlargement of the many small lumina. The cells adjacent to the developing
lumen will form the polarized epithelial lining of the gland. Excess, more basal, epithelial cells
degenerate.
The lumen begins to appear when the branching pattern is almost complete. Thus, during
morphogenesis, invasion by the mammary gland epithelium involves penetration of the
mesenchyme by a solid cord of cells. We suggest that this cellular organization may be a
fundamental characteristic of invasive epithelia and that a crucial step in the development of
malignant epithelial tumours is a change in cell organization from a polarized cell sheet to a
solid cord of cells which can invade.
INTRODUCTION
The mouse mammary gland originates from a thickening of the ectoderm
which invades the underlying mesenchyme. During the earliest stages of
development, the ectoderm forms a solid knob of epithelial cells protruding into
the mesenchyme. This knob then elongates to form a solid cord of epithelial cells
which penetrates the mesenchyme more deeply and later branches in a characteristic fashion (Kratochwil, 1969). By the time the female mouse is born, the
solid epithelial cords have developed lumina and the mammary gland is a system
of branching hollow tubes opening at the nipple (see Ceriani, Pitelka, Bern &
Colley, 1970, for a description of the morphology of the mammary gland of the
newborn rat).
1
Authors' address: Department of Biology as applied to Medicine, The Middlesex Hospital
Medical School, Cleveland Street, London W1P 6DB, U.K.
2
Author's present address: Department of Medical Genetics, St. Mary's Hospital, Manchester, U.K.
* Author to whom requests for reprints should be sent.
40
N . A. S. HOGG, C. J. HARRISON AND C. TICKLE
One of the most striking features of the developing mammary gland, and of
the development of many other glands and organs, is that the early branching
epithelium is a solid cord of cells - tubes are not formed until quite late in
development. The generation of the system of hollow interconnecting tubes,
lined with epithelium, involves the formation of a lumen. It is surprising that this
aspect of gland development seems to have been largely ignored since the formation of a sheet of polarised cells seems to be a fundamental property of epithelial
cells. For example, mammary epithelial cells grown in two-dimensional culture
form a polarised cell sheet (Pickett, Pitelka, Hamamoto & Misfeldt, 1975).
Our long-term aim in studying mammary gland development is to gain insights
into the mechanisms involved in tumour invasiveness. In this respect, the study
of lumen formation during development is relevant to the suggestion that there
is a relationship between epithelial cell organization and the malignancy of
various tumours including those of the breast. For example, tubular breast carcinoma - this is a type of tumour in which the cells typically form organized
epithelial sheets surrounding a lumen (Erlandson & Carstens, 1972) — is often
considered to be, clinically, less malignant than other types of breast tumours in
which the tumour cells form solid cords (Scarff & Torloni, 1968). For these
reasons, our ultrastructural study of the development of the mouse mammary
gland pays close attention to the morphology of lumen formation.
One way in which a solid cord of cells could be transformed into a hollow tube
lined by epithelium would be by the death of the cells in the core of the cord (see
Gliicksmann, 1951). Such localized cell death would present the surviving rim of
epithelial cells with a free surface, which could lead to cell polarization and generation of an organized epithelial sheet. Van Scott & Flaxman (1968) suggested that
cell death was involved in the generation of tubes by epithelial skin cells in culture.
The epithelial cells piled up and tube formation was accompanied by signs of cell
degeneration in the interior of the cell aggregates. However, from observations on
developing salivary glands, Borghese (1950) concluded that, although some
epithelial cells degenerated during morphogenesis, cell death was not involved in
the formation of the lumen of the gland. More recent studies of tubule formation
by cultured capillary endothelial cells (Folkman & Haudenschild, 1980) also show
that cell death is not involved in the formation of the lumen. The first stages of
lumen formation of a capillary appear to involve the segregation of a part of the cell
surface which will form the luminal surface of the cell. Similarly, our studies on the
developing mouse mammary gland suggest that it is local changes in cell surface
properties which lead to lumen formation. If localized regions of the surfaces of
opposing cells in the core of an epithelial cord become nonadhesive, then the
layers of cells will separate (see discussion, Elsdale & Bard, 1974).
MATERIALS AND METHODS
E09 and C57 Black and Tan mice were used in this study. Following mating,
Lumen formation in mammary gland
41
the appearance of a vaginal plug was taken as day 0 of pregnancy. On days 12 and
13, foetuses were sacrificed and their mammary glands with surrounding
ectoderm and mesenchyme were dissected out in a mixture of 50 % horse serum
and 50 % Tyrodes balanced salt solution. By 14 days of gestation, male and
female foetuses can be distinguished by their gonadal development and from this
day onwards, at successive day intervals until birth, mammary glands were
dissected from the female foetuses only. The mammary gland tissue was transferred to half-strength Karnovsky's fixative (Karnovsky, 1965), fixed for 2h at
room temperature, washed in 0-lM-cacodylate buffer pH7-3, and then
osmicated in 1 % osmium tetroxide in 0-lM-cacodylate buffer for l h . After
dehydration in graded ethanols and clearing in propylene oxide, the tissue was
embedded in Araldite CY212. Semithin and ultrathin sections were cut on a
Mark 2 Cambridge Huxley microtome. The semithin sections were stained with
1 % toluidine blue in 1 % borax and the ultrathin sections with uranyl acetate and
lead citrate (Reynolds, 1963).
Glands for freeze fracture were fixed in half-strength Karnovsky's and
cryoprotected by immersion in 30 % glycerol in 0-1 M-cacodylate buffer pH7-3,
for a minimum of 45min. The glands were pushed into a simple tube snap
fracture device and rapid freezing achieved by immersion in nitrogen slush at
—190 °C. The specimens were stored in a Dewar flask under liquid nitrogen.
Freeze fracture was achieved using a modified version of an Edwards freeze
fracturing and etching accessory (Day & Ubee, in preparation) with an Edwards
306 coating apparatus. The specimens were coated with carbon and platinum and
cleaned in a series of bleach solutions prior to washing in distilled water.
Both freeze fractures and ultrathin sections were examined in a Philips
E.M.300.
RESULTS
Phases in mammary gland development of female mice
1. 12-14 days, formation of epithelial knob and resting phase
At 12 days the mammary gland is a small, nearly spherical, knob of cells which
has the appearance of being bodily inserted into the ectoderm (Figs la and 2).
The orientation and columnar morphology of the cells of the knob allow the
periphery of the developing gland to be traced. The cells in this peripheral region
of the mammary knob are tightly packed with the membranes of adjacent cells
having numerous interdigitating folds. Specialised junctions are, however, rare:
there being only a few scattered gap junctions and desmosomes. In contrast to
the peripheral cells, those in the centre of the knob do not adopt any particular
orientation. At the interface of the gland and the mesenchyme, there is a welldefined basal lamina continuous with that of the ectoderm.
The morphology of the mammary gland at 13 days is little changed, except that
42
N. A. S. HOGG, C. J. HARRISON AND C. TICKLE
12 days
.'ctoderm
mammary epithelial cell
knob
mammary mesenchyme
*14 *>V>T'iV
elongated mammary knob
ectoderm developing into
mature epidermis
epithelial collar of nipple
sheath
lumen just beginning
to develop
branching cords of
mammary epithelium
55. epidermis
branching mammary
epithelium has developed
a continuous lumen and is
now a system of
interconnected tubes
Lumen formation in mammary gland
43
a short neck of cells has developed between the body of the gland and the
ectoderm giving the gland a stalked appearance. The cells of this neck are indistinguishable at this stage from those of the gland proper. The body of the gland
itself has increased in size in comparison to that at 12 days, but retains its essentially spherical morphology and intact basal lamina.
By 14 days, the neck of the gland has lengthened further, and the mammary
knob now lies deeper in the mesenchyme. In all other aspects the gland remains
similar to the previous stages and these stages together represent the so-called
'resting' phase of the gland (Kratochwil, 1969).
2. 15-18 days, invasive phase and start of lumen formation
By day 15, changes in the body of the gland signal the end of the 'resting' phase
of development. The gland is now more pear shaped, and has developed a cleft
at the base, which is the first sign of a branch in the invading epithelium (Figs lb,
2 and 3).
The cells in the neck of the gland are now distinct from the cells in the body
of the gland. The neck cells have followed the same pathway of differentiation
as the cells in the lower layers of the ectoderm. Thus, the cells in the neck have
begun to accumulate large concretions of glycogen (Fig. 3a, b) as have the cells
in the lower layers of the ectoderm. In contrast, very low levels of glycogen are
seen in the gland body. In parallel with the accumulation of glycogen in both
the neck of the gland and the ectoderm, keratin production is now apparent at
the ultrastructural level (Fig. 4). Keratin is not seen in the mammary knob at
this stage. Hemidesmosomes are numerous on the surfaces of cells abutting the
basal lamina in both the neck of the gland (Fig. 5) and the ectoderm, whereas
these are found rarely on basal cells in the gland body. These observations
suggest to us that the neck of the gland arises by an invagination of ectoderm
which pushes the mammary knob deep into the mesenchyme, where it then goes
on to branch.
By 16 days, the initial small cleft has developed further, resulting in branching
Fig. 1(A). 12-day mammary gland: an almost spherical knob of epithelial cells
protruding from the ectoderm (e) into the mesenchyme (m).
Fig. 1 (B). 15-day mammary gland. The gland has increased in size and a clear neck
(n) is present. At the base of the knob is a small cleft (arrowed): the first sign of
branching.
Fig. 1 (C). 17-day mammary gland: branching is established. Note intercellular
spaces within the gland body (gb). Localised proliferation of epithelial cells in the
skin has led to the formation of the nipple sheath (ns).
Fig. 1 (D). 20-day mammary gland with a continuous lumen (1) lined by a polarized
epithelium one to three cells thick.
Scale bars (A) = 10 fxm (B) = 20/im (C) & (D) = 40|Um.
Fig. 2. Sequence of diagrams showing the development of the mouse mammary
gland.
44
N . A. S. HOGG, C. J. HARRISON AND C. TICKLE
. " >,
ectoderm
neck
body
Fig. 3 (A). A low-power electron micrograph of a 15-day mammary gland showing
the transition between the cells of the neck (n) which contain large amounts of
glycogen, and those of the body of the gland (b) which have little glycogen. The first
sign of a cleft at the base of the gland is indicated by an arrow. Scale bar = 20/im.
Fig. 3(B). Drawing of mammary gland illustrated in Fig. 3A to show outline of
gland and its component parts.
Fig. 4. Part of a cell typical of the neck of a 15-day gland showing bundles of keratin
in the cytoplasm. Scale bar = 0-5 jum.
Fig. 5. The interface between the epithelial cells of the neck (e) and the mesenchyme
(m) of a 15-day mammary gland. Hemidesmosomes and a well-defined basal lamina
are present. Scale bar = 0-25 jum.
Lumen formation in mammary gland
45
of the knob to give rise to solid cords of epithelial cells which invade the mesenchyme. Two types of cells, distributed apparently at random within the cords,
can be distinguished on the basis of density of staining (Fig. 6). Intercellular
spaces now begin to develop within the cords. Into some of these spaces the cells
extend long cellular projections (Fig. 7). At this stage, the epithelial/
mesenchyme interface is ruffled both on the gland and the neck, and the basal
lamina is less distinct.
'
At 17 days, formation of the nipple sheath begins: a circular zone of thickened
epidermis projects into the mesenchyme with the mammary neck at its centre
(Fig. lc and Fig. 2). The intercellular spaces (Fig. 8) in the invading epithelial
cords have increased in size and number over those at 16 days and, for the first
time in development, small intercellular spaces can be seen in the neck of the
mammary gland heralding the onset of lumen formation in this region too.
By day 18 the nipple sheath is well developed and growth of the mammary
cords into the mesenchyme has continued. The branching structure of the gland
is now quite apparent, with the original mammary body remaining distinct as a
larger group of cells from which the cords extend. The loose nature of the cells
within the centre of the cords persists while those cells on the periphery remain
Fig. 6. The two cell types, dark and pale, which can be distinguished in the mammary gland from 16 days onwards. Scale bar = 1 /im.
Fig. 7. Small intercellular spaces whichfirstappear at 16 days within the gland body.
Numerous cellular extensions can be seen traversing these spaces. Scale bar = 1 jum.
EMB73
46
N . A. S. HOGG, C. J. HARRISON AND C. TICKLE
more cohesive. Cell death is now also evident and is confined to the body of the
gland. The tips of the developing branches are irregular in outline and the basal
lamina is poorly defined (Fig. 9). Within the mesenchyme the cells of the fat pad
containing large lipid droplets are now visible in close association with the
epithelial cords.
3. 19?-20 days, invasion and lumen formation completed
The mammary gland is now a modestly ramifying system of ducts in an extensively developed fat pad. The discontinuous lumina of the earlier stages have
become unified to form a single continuous ductal system (Fig. Id and Fig. 2) and
the loose aggregation of cells within the cords, seen during development is
replaced by an ordered epithelium from one- to three-cells thick. Many of the
lining epithelial cells have active Golgi apparati producing large numbers of
small vesicles and the lumen is often filled with extracellular material. Occasionally, free cells are found within the lumen. No specialised myoepithelial cells can
be recognized at this stage, although the dark- and light-staining cell types persist. The branched outline of the ductal system remains crenelated, but the basal
lamina is well defined (Fig. 10), and numerous hemidesmosomes are present
along its length.
Lumen formation
Up until 16 days, the invasive epithelium is solid. Then with the onset of rapid
growth of the gland into the surrounding mesenchyme, intercellular spaces begin
to appear within the mammary knob. EM sections revealed considerable variation in the size and organization of these spaces. Desmosomes are sometimes
present between adjacent cells at this stage. Towards the periphery of the invading cords the cells appear more cohesive than those in the centre. In contrast,
the cells in the neck of the gland remain tightly packed together.
By 17 days, in the cords of the gland, which may now be 14-20 cells wide, the
intercellular spaces have increased both in size and number. For the first time,
'true' microlumina appear at scattered sites within the branching cords. These
microlumina, which will go on to form the cavity of the ducts of the gland, are
edged by epithelial cells bearing microvilli on their free surfaces and linked by
developing apical junctional complexes (see Fig. 11 and Fig. 12 for possible
stages in the development of the initial small lumina). However, large numbers
of intercellular spaces remain bordered by cells, with neither microvilli nor
junctional complexes. Within the neck of the gland, which remains four to six
cells wide, small intercellular crevices now appear similar to those spaces seen
in the gland body on day 16. These never dilate, as they do in the body but the
cells bordering them bear microvilli and form junctional complexes to give rise
to short discrete lumina.
At day 18, the crevices have developed into narrow discontinuous slits along
the centre of the neck. At these sites, the neck is transformed into a thin tube with
Lumen formation in mammary gland
9
Fig. 8. Tip of an epithelial cord at 17 days. Note the range and size in organization
of the intercellular spaces. Some 'true' lumina with microvilli on apical cell surfaces
can be identified. Note a pale cell (arrowed). Scale bar = 5/im.
Fig. 9. Developing branch of the 18-day mammary gland. The basal lamina is very
poorly defined. A process of a mesenchymal cell (m) is very close to the epithelium.
Scale bar = 1 ^m.
Fig. 10. Epithelial mesenchymal interface of the body of the gland at 20 days. In
contrast to 18 days (Fig. 9), the basal lamina is well defined and collagen is present
in large amounts. Scale bar =
47
48
N . A. S. HOGG, C. J. HARRISON AND C. TICKLE
ri
Fig. 11. High power of Fig. 8 (the tip of an epithelial cord at 17 days) showing the
first stages of lumen formation. Early junctions can be seen between adjacent cells
and the free surfaces of some cells bear microvilli. Scale bar =
Fig. 12. High power of a different region of Fig. 8 showing a more advanced stage
of lumen formation with well-developed junctional complexes linking neighbouring
epithelial cells whose free surfaces now bear numerous microvilli. Scale bar = 1 pim.
Lumen formation in mammary gland
49
polarized epithelial walls two to three cells thick. Radical changes now begin to
take place within the gland body. The small discrete lumina become progressively larger to link up to form the continuous lumen of the gland. The course that
is adopted by the lumen as it becomes continuous is not always a central one but
instead meandering. The epithelial layer of one to three cells deep surrounding
the developing lumen will form the epithelial lining of the gland. These cells
remain healthy (Fig. 13) while the excess cells at greater distances from the
lumen begin to degenerate: they stain less heavily with both toluidine blue and
lead citrate than their healthy counterparts.
By day 19 a lumen is almost continuous along the length and well developed
throughout most of the branching cords and neck of the gland. There is now
extensive cell death in the basal layers of the epithelial cords. Many nuclei display
greatly swollen nuclear membranes and large numbers of intracellular spaces
develop into which cytoplasmic extensions are often elaborated. In many cases
adjoining cell membranes have broken down and extensive syncitia (Fig. 14) are
Fig. 13. Oblique section through an epithelial cord just below the base of the neck of
a 19-day mammary gland. A well-developed lumen (1) can be seen surrounded by
healthy epithelial cells (dark staining). Epithelial cells further removed from the
lumen are degenerating (pale staining). Scale bar = 10jum.
Fig. 14. High power of degenerating mammary epithelial cells from a 19-day mammary gland. Note the breakdown of plasma membranes and swollen nuclear envelopes. Scale bar =
50
N. A. S. HOGG, C. J. HARRISON AND C. TICKLE
formed. The basal lamina beneath the necrosing cells that abuts the mesenchyme
remains patent however and seemingly without breaks.
On day 20 no evidence of necrotic cells remain, and the mammary gland is transformed from a network of solid cords with discontinuous lumina to a system of
interconnecting ducts with a common opening to the skin via the neck of the gland.
As a consequence of the cell death in earlier stages the walls of the gland now
consist of a polarized sheet of epithelial cells one to three cells thick, and are indistinguishable from those of the neck. The cells are closely apposed in the sheet with
only a few small intercellular spaces remaining, and those with luminal surfaces
possess microvilli and well-developed junctional complexes (Fig. 15 and 16).
DISCUSSION
The first signs of lumen formation in the mammary gland can be detected after
the initial branching pattern of the epithelial sprout has been established. Small
crevices lined with microvilli appear between the apposing surfaces of cells at
Fig. 15. Lumen of a 20-day mammary gland showing microvilli on the free surfaces
of the lining epithelial cells which are linked by sub-apical junctional complexes. An
active Golgi apparatus (g) can be seen towards the apex of one of the cells. Scale
bar = 1/im.
Fig. 16. Freeze-fracture replica of the border of the lumen of a 20-day mammary
gland. The free surface of the epithelium bears microvilli and well-developed tight
junctions can be seen. Scale bar =
Lumen formation in mammary gland
51
scattered sites in the central region of the neck and in the cores of the branching
sprouts. Initially, these crevices and spaces are bounded by only a few epithelial
cells, but these soon become larger, fusing with neighbouring crevices to lead to
the formation of the continuous central lumen of the gland.
Development of non-adhesive cell surfaces leads to the formation of scattered
microlumina
The initial step in lumen development appears to be the development of
localized regions of non-adhesive surface membrane on cells in central regions
of the gland. Once the inner faces of opposing cells in the core of the gland
become non-adhesive, a cell layer surrounding a cavity will form. The evidence
for the development of localized regions of non-adhesive surface membrane is
the appearance of microvilli on the cell surfaces lining the initial crevices and
cavities between the epithelial cells. As we have previously argued (Tickle,
Crawley & Goodman, 1978) there is a good correlation between the presence of
a dense array of microvilli on a surface and the non-adhesiveness of that surface.
The development of a localised non-adhesive surface gives that cell a polarity.
One surface is no longer in contact with other cells and becomes the apical
surface of the cell. The development of a region of non-adhesive surface membrane bearing microvilli appears to be accompanied by the development of tight
junctions around its perimeter. A possible scheme for the development of a nonadhesive membrane domain would be as follows. The adhesive membrane
proteins diffuse away from a local region of the cell surface. Tight junctions then
develop at the perimeter of this non-adhesive membrane domain which forms
the apical surface of the cell and becomes elaborated into microvilli.
This scheme suggests that apical surface specialization will be rapidly followed
by lateral cell surface specialization. Indeed, the development of lateral tight
junctions may be essential to maintain the separation of the non-adhesive apical
surface and the adhesive lateral and basal cell surfaces. Evidence for tight junctions having such a role comes from experiments in which the distribution of
membrane proteins has been followed in individual epithelial cells isolated from
their neighbours. For example, in isolated gut cells, with disrupted tight junctional seals, cell surface proteins normally confined to just the apical surfaces of
the cells soon become distributed evenly all around the cell surface (Pisam &
Ripoche, 1976). Again, in kidney cells (cell line MCDK) the normal differential
distribution of intramembranous particles between the apical and basolateral
surface membranes is only maintained when tight junction integrity is preserved
(Sang, Saier & Ellisman, 1980). Thus, the simultaneous appearance in the
developing mammary gland of non-adhesive microvilli on the apical surfaces of
cells lining the crevices and tight junctions on their lateral surfaces may not be
coincidental. Surprisingly, however, tight junctions may not be required to maintain the form of the highly folded surface covering the microvilli: isolated gut
cells retain a localised brush border (Ziomek, Schulman & Edidin, 1980) and
52
N . A. S. HOGG, C. J. HARRISON AND C. TICKLE
isolated amphibian blastomeres retain distinctive differences in cell surface morphology (Roberson, Armstrong & Armstrong, 1980).
Establishment of the epithelial lining of the gland and the role of cell death
The initial small lumina must join up to form the continuous central lumen of
the gland. Each small lumen appears to enlarge until it fuses with a neighbouring
lumen. Enlargement of the lumen must require recruitment of cells into its
surrounding polarized epithelium. Whether such recruitment would require
tight junction disassembly as well as assembly is not clear.
But the formation of the central lumen of the gland is not the only consequence
of the joining up of the many small lumina. By the time the lumen is continuous
throughout the gland it is lined by a polarized epithelium one to three cells thick.
In the neck of the gland, the epithelial cord is only four to six cells wide before
the lumen forms. Thus, the epithelial lining is generated quite simply as the cord
of epithelial cells separates down the middle. However, in other regions of the
gland, extensive cell rearrangement must be occurring since the epithelial cords
prior to lumen completion can be up to 20 cells across. We propose that the cell
death that we observed eliminates excess cells, so that only those required to
make up a thin epithelial lining (one to three cells thick) are left. The two to three
layers of epithelial cells immediately adjacent to the developing lumen survive:
the more basal layers of cells down to the basal lamina appear to be eliminated.
A similar modelling of the epithelial sprout may occur during morphogenesis of
other glands, such as the salivary gland (Borghese, 1950), and cell death has been
found to be involved in modelling the shape of other structures such as developing digits (Ballard & Holt, 1968).
Initiation of cell polarization during lumen formation
This account of how the lumen forms raises many questions. For example, how
are the local cell surface changes in adhesiveness, which lead to lumen formation
and appear to effect the first stage in polarization of the epithelial cells, initiated?
In principle, epithelial cells could be polarized by denning either their apical
surfaces or their basal surfaces. For example, if one considers the formation of
polarized cell sheets by epithelial cells plated in two-dimensional culture (Pickett
et ai, 1975), it may be that the provision of free space above the cells serves to
define their apical surfaces. Alternatively, attachment to the substratum could
define the basal surfaces of the cells. It is clear that the apical surfaces of the
epithelial cells of the developing mammary gland are not defined by the
provision of free space. Thus, there is no central region of cell death in the
sprout, that would create a free space above the rim of surviving cells. This
focuses attention on the basal surfaces of the epithelial cells.
The mammary epithelium is provided with a basal lamina from the earliest
stages of gland development. Nevertheless, the basal lamina at these early stages
does not effect polarization of the enclosed mass of epithelial cells into a sheet
Lumen formation in mammary gland
53
although a local effect on cell orientation in the immediately adjacent layer of
epithelial cells is apparent. One possibility is that a change in the composition of
the basal lamina triggers the polarization of the adjacent layer of mammary
epithelial cells. This polarization could be propagated into deeper epithelial
layers by cell interactions until, in the central region, unlike surfaces confront
and become non-adhesive. In support of the idea that the composition of the
basal lamina could effect cell polarization, recent studies on the developing
kidney show that the onset of synthesis of laminin, a glycoprotein characteristic
of the basal lamina in adult tissues, is correlated with the initiation of tubule
formation in condensed nephrogenic mesenchyme (Ekblom et al., 1980).
Alternatively, in the developing mammary gland, the concentration of
metabolites diffusing into the epithelial cord from the mesenchyme may determine where non-adhesive surfaces should develop.
Lumen formation in other glands and in tubular organs follows the same scheme
The embryonic and post natal development of the lumen of salivary glands is
strikingly similar (Borghese, 1950; Redman & Sreebny, 1971). There is also a
detailed parallel between formation of the lumen of the thyroid follicle (Hilfer,
1964) and the lumen of the mammary gland, further supporting the idea that the
cellular mechanisms of lumen formation may indeed be universal. Without drawing up a comprehensive list, we can also mention that the development of
metanephric tubules also involves the development of tight junctions and nonadhesive microvilli (Saxen & Wartiovaara, 1966).
Invasive morphogenesis always involves solid cords of cells
A survey of the development of tubular glands and organs shows that, invasion
of mesenchyme by epithelia commonly involves penetration by solid cords of
cells (see Table 1). This is dramatically illustrated by the development of the
thyroid (Shain, Hilfer & Fonte, 1972) in which the epithelial cells which will form
the lining of the follicles of the mature gland are first arranged in a polarized cell
sheet, then form a solid cord of cells which penetrates the mesenchyme, and
finally become polarized again as the lumina of the gland develop. As far as we
know, embryonic epithelial invasion of mesenchyme always involves a solid cord
of unpolarized cells. Furthermore, the sprouts of mammary end buds which
develop during puberty and pregnancy may also be made up of solid cords of
epithelial cells (Sekhri, Pitelka & De Ome, 1967).
Embryonic mammary gland morphogenesis and the development of epithelial
tumours
A striking feature of the cellular organization of breast tumours is that invasion
here too involves solid cords of epithelial cells (Willis, 1967). In vitro, the invasion
of collagen gels by mammary epithelial cells from normal tissue or from tumours
54
N. A. S. HOGG, C. J. HARRISON AND C. TICKLE
Table 1. The organization of invasive epithelia
Invasive epithelium
Organization of invading epithelium
Embryonic mammary gland
Solid primary sprout. Invasive epithelial branches
are solid too.
Embryonic hair1
Embryonic salivary glands2'3
Solid peg of epidermal cells penetrates dermis
Solid primary sprouts. Invasive epithelial branches
are solid
Embryonic thyroid4
Initial invagination involves buckling of an
organized cell sheet. Invasive epithelial sprout
penetrates mesenchyme as a solid cord of cells.
Embryonic liver5
Initial diverticula form solid and 'luminated' buds
which anastomose. Solid masses of hepatic cells
rearrange to form tubules
Embryonic pancreas6
Primary islet masses which protrude from initial
hollow diverticulum appear solid
Embryonic endothelium7
New blood vessels develop from solid cords of
endothelial cells
Mammary gland during puberty
and pregnancy
In mice, solid end buds develop during puberty
and solid alveoli are present during gland growth
in pregnancy8. In rats, the lumen of the alveoli
develop after growth of the gland which occurs
in early pregnancy9.
Mouse hyperplastic mammary
lobuloalveolar tissue10
Solid cords of cells invaded virgin fat pads. A short
distance behind invasive sprout, cells were
organized into epithelial lining of tubules
Human breast carcinomas11
(typically, tumours show varying proportions of the following types of cell arrangement)**:
i) Adenocarcinoma
Simple acinar structure with polarized epithelial
lining
ii) Spheroidal-cell carcinoma
simplex
Solid clumps or cords of cells
iii) Diffuse anaplastic carcinoma
Solid cords of highly disorganized cells, strands of
cells.
** Tumours are rated according to their morphology: those with highest proportion of
arrangement (i) are least malignant, those with highest proportion of arrangement (iii) are
most malignant1213.
^engel, 1976
Borghese, 1950
3
Redman and Sreebny, 1970
2
4
5
Kingsbury Alexanderson & Kornstein
1956
6
Wessells and Cohen, 1967
7
Cliff, 1963
Sekhriefa/.,1967
9
Murad, 1970
10
Ashley etai, 1980
11
Willis, 1967
12
Scarff and Torloni, 1967
13
Wellings Jensen & Marcum, 1975
8
Lumen formation in mammary gland
55
are also initiated by solid cords or string of cells (Yang et al., 1979; Bennett, 1980:
Hallowes, Bone & Jones, 1980). Indeed, one gets the overwhelming impression
that this is a general principle of epithelial invasion (see Table 1) suggesting that
solid structures may be able to penetrate into surrounding tissues more easily
than hollow structures can.
A comparison of the morphology of the morphogenesis of the mammary gland
and the development of breast tumours shows that in morphogenesis, solid cords
of epithelial cells develop into hollow tubes lined by polarized epithelium while,
in tumours, solid cords of cells are generated from polarized epithelial cells.
Thus, it is interesting that in some forms of intraductal breast carcinoma - usually
considered to be an early stage in the development of frankly invasive tumours
- the cells are arranged in what is described as a cribriform pattern (Haagensen,
1971): many small cavities are apparent in the solid cord of tumour cells. Indeed,
electron microscopic observations of the cribriform pattern of a specimen of
breast tissue with atypical ductal hyperplasia (possibly an early form of intraductal breast carcinoma) show that the several slit-like lumina are lined by
microvillous cell surface (Goldenberg, Goldenberg & Sommers, 1969). This
pattern is reminiscent of the early stages of lumen formation in the developing
gland, in which several small cavities form. Development of invasive carcinoma
may also occur without passing through a cribriform phase. In lobular neoplasia,
another possibly precancerous breast lesion, regions of the epithelial lining of the
mammary ducts form solid buds of cells that protrude into the surrounding tissue
(Haagenson, 1971) and are reminiscent of the earliest stages in mammary gland
development. Nevertheless, whether invasive carcinoma develops indirectly by
occluding the lumen of the gland or directly by budding from the epithelial lining,
the cells of the invasive tumour are arranged in solid cords.
Changes in the basal lamina may be crucial in allowing the growth of solid
cords of unpolarized cells. According to this idea, one might expect that, if
carcinoma cells are supplied with a normal basal lamina, this could effect their
polarization. Indeed such polarization of carcinoma cells and their participation
in the formation of an organized sheet of epithelial cells has been noted when
tumour cells have metastasized to normal epithelia (Brooks, 1970; Pitelka,
Hamomoto & Taggart, 1980). Such an effect of the basal lamina may also explain
the positioning of carcinoma cells in embryonic ectoderm which we observed
when we tested the invasiveness of epithelial tumours in the developing chick
wing (Tickle etal., 1978).
We thank Professor L. Wolpert and Mr R. P. Gould for their encouragement and comments
on the manuscript, Mr A. Day for help with freeze fracture and Ms A. Crawley for help with
sectioning. This work was supported by the Cancer Research Campaign.
REFERENCES
R. L., CARDIFF, R. D., MITCHELL, D. J., FAULKIN, L. J. & LUND, J. K. (1980).
Development and characterization of mouse hyperplastic mammary outgrowth lines from
ASHLEY,
56
N . A. S. HOGG, C. J. HARRISON AND C. TICKLE
BALB/cfC3H hyperplastic alveolar nodules. Cancer Res. 40, 4232-4243.
K. J. & HOLT, S. J. (1968). Cytological and cytochemical studies on cell death and
digestion in foetal rat foot: the role of macrophages and hydrolytic enzymes. /. Cell Sci.
3, 245-262.
BENNETT, D. C. (1980). Morphogenesis of branching tubules in cultures of cloned mammary
epithelial cells. Nature 285, 657-659.
BORGHESE, E. (1950). The development in vitro of the submandibular and sublingual glands
of Mus musculus. J. Anat. 84, 287-303.
BROOKS, R. E. (1970). Mouse mammary tumour metastases in lung: an electron microscopic
study. Cancer Res. 30, 2156-2165.
CERIANI, R., PITELKA, D. R., BERN, H. A. & COLLEY, V. B. (1970). Ultrastructure of rat
mammary-gland anlagen in vivo and after culture with hormones. /. exp. Zool. 174,79-100.
CLIFF, W. J. (1963). Observations on healing tissue: a combined light and electron microscopic
investigation. Phil. Trans. Roy. Soc. (London) Series B. 246, 305-325.
EKBLOM, P., ALITALO, K., VAHERI, A., TIMPL, R. & SAX£N, L. (1980). Induction of a basement
membrane glycoprotein in embryonic kidney: possible role of laminin in morphogenesis.
Proc. natn. Acad. Sci., U.S.A. 77, 485-489.
ELSDALE, T. & BARD, J. (1974). Cellular intractions in morphogenesis of epithelial mesenchymal systems. J. Cell Biol. 63, 343-349.
ERLANDSON, R. A. & CARSTENS, P. H. B. (1972). Ultrastructure of tubular carcinoma of the
breast. Cancer 29, 987-995.
FOLKMAN, J. & HAUDENSCHILD, C. (1980). Angiogenesis in vitro. Nature 288, 551-556.
GLUCKSMANN, A. (1951). Cell deaths in normal vertebrate ontogeny. Biol. Rev. 26, 59-87.
GOLDENBERG, V. E., GOLDENBERG, N. S. & SOMMERS, S. C. (1969). Comparative ultrastructure of atypical ductal hyperplasia, intraductal carcinoma and infiltrating ductal carcinoma
of the breast. Cancer 24, 1152-1169.
HAAGENSEN, C. D. (1971). Diseases of the Breast. 2nd edn. W. B. Saunders Company,
Philadelphia, London and Ontario.
HALLOWES, R. C., BONE, E. J. & JONES, W. (1980). A new dimension in the culture of human
breast. In Tissue Culture in Medical Research 2 (R. J. Richards & K. T. Rajan, eds.), pp.
213-220. Pergamon Press, Oxford and New York.
HILFER, R. S. (1964). Follicle formation in the embryonic chick thyroid. I. Early morphogenesis. /. Morph. 115, 135-153.
KARNOVSKY, M. J. (1965). A formaldehyde glutaraldehyde fixative of high osmolarity for use
in electron microscopy. /. Cell Biol. 27, 137a (Abstr.).
KINGSBURY, J. W., ALEXANDERSON, M. & KORNSTEIN, E. S. (1956). The development of the
liver of the chick. Anat. Rec. 124, 165-188.
KRATOCHWIL, K. (1969). Organ specificity in mesenchymal induction demonstrated in the
embryonic development of the mammary gland of the mouse. Devi Biol. 20, 46-71.
MURAD, T. M. (1970). Ultrastructural study of rat mammary gland during pregnancy Anat.
Rec. 167, 17-36.
PICKETT, P. B., PITELKA, D. R., HAMAMOTO, S. T. & MISFELDT, D. S. (1975). Occluding
junctions and cell behaviour in primary cultures of normal and neoplastic mammary gland
cells. /. Cell Biol. 66, 316-332.
PITELKA, D. R., HAMOMOTO, S. T. & TAGGART, B. N. (1980). Basal lamina and tissue recognition in malignant mammary tumours. Cancer Res. 40, 1600-1611.
PISAM, M. & RIPOCHE, P. (1976). Redistribution of surface macromolecules in dissociated
epithelial cells. J. Cell Biol. 71, 907-920.
REDMAN, R. S. & SREEBNY, L. M. (1970). The prenatal phase of the morphosis of the rat
parotid gland. Anat. Rec. 168, 127-138.
REDMAN, R. S. & SREEBNY, L. M. (1971). Morphologic and biochemical observations on the
development of the rat parotid gland. Devi Biol. 25, 248-279.
REYNOLDS, E. S. (1963). The use of lead citrate at high pH as an electron opaque stain in
electron microscopy. /. Cell Biol. 17, 208-212.
ROBERSON, M., ARMSTRONG, J. & ARMSTRONG, P. (1980). Adhesive and non-adhesive membrane domains of amphibian embryo cells. /. Cell Sci. 44, 19-31.
BALLARD,
Lumen formation in mammary gland
57
H. U., SAIER, M. H. & ELLISMAN, M. H. (1980). Tight junction formation in the
establishment of intramembranous particle polarity in aggregating MDCK cells. Effect of
drug treatment. Expl Cell Res. 128, 223-237.
SAX£N, L. & WARTIOVAARA, J. (1966). Cell contact and cell adhesion during tissue organization. Int. J. Cancer 1, 271-290.
SCARFF, R. W. & TORLONI, H. (1968). Histological typing of breast tumours. International
Hbtological Classification of Tumours No. 2. Geneva: World Health Organization.
SEKHRI, K. K., PITELKA, D. R., & D E O M E , K. B. (1967). Studies of mouse mammary glands.
I. Cytomorphology of the normal mammary gland. J. natn. Cancer Inst. 39, 459-490.
SENGEL, P. (1976). Morphogenesis of the skin. Cambridge University Press, Cambridge.
SHAIN, W. G., HILFER, S. R. & FONTE, V. G. (1972). Early organogenesis of the embryonic
chick thyroid. I. Morphology and biochemistry. Devi Biol. 28, 202-218.
TICKLE, C , CRAWLEY, A. & GOODMAN, M. (1978). Mechanisms of invasiveness of epithelial
tumours: ultrastructure of the interactions of carcinoma cells with embryonic mesenchyme
and epithelium. /. Cell Sci. 33, 133-155.
VAN SCOTT, E. J. & FLAXMAN, B. A. (1968). Environmental control of epithelial cells in vivo
and in vitro. In Epithelial-MesenchymalInteractions (R. Fleishmajer and R. E. Billingham,
eds.) pp. 280-295. Baltimore: Williams and Wilkins.
WELLINGS, S. R., JENSEN, H. M. & MARCUM, R. G. (1975). An atlas of subgross pathology
of the human breast with special reference to a possible precancerous lesion. J. natn. Cancer
Inst. 55, 231-245.
WESSELLS, N. K. & COHEN, J. H. (1967). Early pancreas organogenesis: morphogenesis,
tissue interactions and mass effects. Devi Biol. 15, 237-270.
WILLIS, R. A. (1967). Pathology of Tumours, 4th Edn. London: Butterworths.
SANG,
YANG, J., RICHARDS, J., BOWMAN, P., GUZMAN, R., ENAMI, J., MCCORMICK, K., HAMAMOTO,
S., PITELKA, D. & NANDI, S. (1979). Sustained growth and 3-dimensional organization
primary mammary tumour epithelial cells embedded in collagen gels. Proc. natn. Acad.
Sci., U.S.A. 76, 3401-3405.
ZIOMEK, C. A., SCHULMAN, S. & EDIDIN, M. (1980). Redistribution of membrane proteins in
isolated mouse intestinal epithelial cells. /. Cell Biol. 86, 849-858.
(Accepted 25th July 1982)