/. Embryol. exp. Morph. 96, 65-77 (1986)
65
Printed in Great Britain © The Company of Biologists Limited 1986
Scanning electron microscopic observation of mouse
embryonic submandibular glands during initial
branching: preferential localization of fibrillar
structures at the mesenchymal ridges participating in
cleft formation
YASUO NAKANISHI, FUKASHI SUGIURA
Department of Chemistry, Faculty of Science, Nagoya University, Chikusa,
Nagoya 464, Japan
JUN-ICHI KISHI AND TARO HAYAKAWA
Department of Biochemistry, School of Dentistry, Aichi-Gakuin University,
Chikusa, Nagoya 464, Japan
SUMMARY
Branching submandibular glands of 12-day mouse embryos and those cultured in the presence
and absence of a collagenase inhibitor from the culture medium of bovine dental pulp or a
Clostridial collagenase were examined with the scanning electron microscope. Fracturing of
fixed and dried glands with the tip of a fine needle succeeded in exposing the surfaces of the
lobules and of their mesenchymal replicas at different stages of branching. At the beginning of
branching, corresponding parts of the mesenchyme formed ridges on or in which the fibrillar
structures were often found. At the stage forming deeper clefts thicker fibres, 0-5-2-5 jum in
diameter, were observed between two adjacent lobules. On the contrary, no apparent differences in thefibrillarstructures on the epithelial surfaces were detected between the shallow
cleft and noncleft regions at the initial phase of branching. These fibrillar structures were very
abundant in glands cultured with collagenase inhibitor and were completely lost in glands
cultured with bacterial collagenase, strongly indicating that these materials consisted of
collagen. The possible involvement of mesenchyme in epithelial branching is discussed with
special reference to mesenchymal traction forces that would be elicited by fibrillar collagens.
INTRODUCTION
Submandibular gland branching at early morphogenetic stages of the embryo is
initiated by the formation of sharp furrows in the epithelium followed by their
progressive broadening. This type of epithelial branching is characteristic of
submandibular organogenesis and is different from other organs such as lung.
Rodent embryonic submandibular glands provide an excellent system to study
Key words: mouse submandibular gland, epithelial branching, collagen fibril, scanning electron
microscopy, collagenase inhibitor, bacterial collagenase, mesenchymal ridges, cleft formation,
fibrillar structures.
66
Y. NAKANISHI, F. SUGIURA, J. KISHI & T. HAYAKAWA
branching morphogenesis in vitro since cleft formation in the epithelium can be
easily observed (Grobstein, 1954).
Collagen, one of the major constituents of extracellular matrix, is thought to be
important in cleft formation due to the following facts: (1) the presence of
commercial bacterial collagenase preparations in the culture medium inhibited
epithelial branching (Grobstein & Cohen, 1965; Wessells & Cohen, 1968), and (2)
a proline analogue, L-azetidine 2-carboxylic acid, in the culture medium suppressed
epithelial morphogenesis (Spooner & Faubion, 1980). The former experiments
were, however, hampered by a possible contamination of glycosaminoglycan
degrading enzymes and proteases (Bernfield, Banerjee & Cohn, 1972). Although
the latter authors demonstrated that L-azetidine 2-carboxylic acid inhibited total
protein synthesis to a much lesser degree than the collagen synthesis of 13-day
salivary glands, it was possible that the impaired noncollagenous proteins synthesized in the presence of the drug might affect the proper morphogenesis.
To overcome the above possibilities, we have recently used a highly purified
Clostridial collagenase and an interstitial collagenase-specific inhibitor (CI) obtained from the culture medium of bovine dental pulp, the use of which resulted
in modulation of epithelial branching in mouse submandibular glands. The
collagenase in the medium completely inhibited the branching, whereas the
collagenase inhibitor dramatically stimulated it within 14 h culture of 12-day
glands without affecting rates of cell proliferation (Nakanishi, Sugiura, Kishi &
Hayakawa, 1986a). This result confirmed earlier findings that collagen plays a
crucial role in the branching morphogenesis of mouse embryonic submandibular
gland.
There have been a few reports documenting the presence and distribution of
collagen in salivary gland by using electron microscopy and immunostaining
(Bernfield & Wessells, 1970; Spooner & Wessells, 1972; Spooner & Faubion, 1980;
Bernfield, Banerjee, Koda & Rapraeger, 1984). These studies were, however, not
sufficient in describing the localization and the three-dimensional orientation of
collagen fibrils during early morphogenetic stages.
How does collagen act as a morphogenetic component? Recently, embryonic
chick heart mesenchyme and several other fibroblastic cells have been shown to
contract silicone rubber as well as collagen gel (Bell, Ivarsson & Merrill, 1979;
Steinberg, Smith, Colozzo & Pollack, 1980; Harris, Stopak & Wild, 1981; Stopak
& Harris, 1982). It is, therefore, reasonable to assume that some insoluble or
fibrillar components of the substratum could be prerequisites for such cells to exert
traction. To examine a possible involvement of such traction force in branching
process, it is of importance to know the three-dimensional alignment of fibrillar
architectures at the epithelial-mesenchymal interfaces of the salivary gland. In the
present study, we examined control specimens of 12-day glands and those cultured
in the presence and absence of either CI or a Clostridial collagenase with the
scanning electron microscope. We present evidence showing the preferential
localization of fibrillar architectures at the mesenchymal ridges participating in
cleft formation.
SEM studies of submandibular gland branching
67
MATERIALS AND METHODS
Materials
A collagenase inhibitor (CI), which is known to be specific for the interstitial collagenases,
was purified from the culture medium of bovine dental pulp by the method of Kishi & Hayakawa
(1984).
A highly purified Clostridial collagenase devoid of caseinolytic activity was kindly donated
by Drs T. Ohya and N. Yokoi of Amano Pharmaceutical Co., Japan. No hyaluronidase or
chondroitinase activities were detected.
Embryos
12-day embryos were obtained from DDY mice. The day of discovery of the vaginal plug was
designated as day 0.
Organ culture
Embryonic salivary glands were dissected from 12-day embryos in Tyrode's solution. Organ
culture was routinely carried out on ultrathin Millipore filter (THWP, 0-45 ptm pore size)
assemblies (Banerjee, Cohn & Bernfield, 1977) in BME medium supplemented with ascorbic
acid (SOjUgmP1) and 10% foetal calf serum. When CI- and collagenase-treated glands were
prepared, the final concentration of either CI or the bacterial collagenase in the medium was
S^gml" 1 (Nakanishi etal. 1986a). Living cultures of salivary glands were monitored by taking
photographs at appropriate intervals on an Olympus light microscope BH-2 equipped with an
automated exposure unit.
Processing for scanning electron microscopy
Intact 12-day glands or those cultured under various conditions were fixed in 1-6% glutaraldehyde and 1-6 % paraformaldehyde in 0-1 M-phosphate buffer, pH 7-2, and postfixed in 2 %
OsO4 in 0-1 M-phosphate buffer, pH 7-2. The specimens were then dehydrated with ethanol,
transferred to isoamyl acetate and dried in a critical-point drier using CO2 as transitional fluid.
The dried specimens thus prepared were mounted onto aluminium stubs, cracked with the tip of
a fine needle under the microscope, coated with gold and examined with an Akashi ALPHA-30
scanning electron microscope at 30 kV.
RESULTS
Scanning electron microscopy was employed to determine if regional differences in fibrillar structures exist on the surfaces of the lobular epithelia and of their
complementary mesenchyme during the first cleft formation. We therefore examined 12-day glands, which were at the stage just before forming clefts, and those
cultured under normal conditions for 17h, reaching the mid 13-day stage. In
addition, glands cultured in the presence of either CI or Clostridial collagenase for
11 h were also examined.
Intact 12-day gland
An intact 12-day gland was cracked and the exposed surfaces of the epithelium
including lobular and stalk regions and its complementary mesenchyme were
examined with the scanning electron microscope (Figs 1-6). An overall view of
the specimen is shown in Fig. 1, in which epithelia of lobular and stalk regions and
their mesenchyme are clearly seen. The lobule had several indentations (Figs 2, 3)
68
Y. NAKANISHI, F. SUGIURA, J. KISHI & T. HAYAKAWA
SEM studies of submandibular gland branching
69
as reported by Nogawa (1983). Some of these preclefts grew to definitive clefts
while others disappeared during the subsequent cultivation (Nakanishi etal.
19866). Fine fibrils, 80-150nm in diameter, were scattered over the distal end of
the lobule and similar fibrils were seen within the preclefts (Figs 3, 4).
Much denser distribution of fibrillar materials was found on the stalk region
adjacent to the lobule (Figs 2, 5). Their diameter, 100-200nm, was greater than
that of the lobular surface (Fig. 4). This observation is entirely consistent with the
statement by Grobstein & Cohen (1965) that the morphogenetically inactive stalk
region is covered with dense collagen fibrils.
Likewise, on the surface of stalk mesenchyme we saw random arrays of fibrils
(Fig. 1). The arising mesenchymal ridges complementary to the preclefts in the
epithelium had fibrillar material attached to the mesenchyme (Fig. 6), suggesting
that these structures were synthesized by the mesenchyme. However, mesenchyme that was not confronted with epithelium did not contain so many fibrillar
structures (Figs 1,2), implying that there would be some specific interactions
between epithelium and mesenchyme leading to the deposition of fibrillar material
at the interface (Kallman & Grobstein, 1965; Grobstein & Cohen, 1965; Bernfield,
1970; von der Mark, 1981).
Cultured 12-day glands
12-day glands were cultured in the normal medium for 17 h, during which period
two clefts were formed on average, and the glands were processed as described
in Materials and Methods. Scanning electron micrographs of the two specimens
are presented (Figs 7-12). An overall view of the lobule and its mesenchymal
counterpart is shown in Fig. 7. Judging from the top view of the specimen (Fig. 8),
together with the picture in Fig. 7, it is reasonable to conclude that the mesenchyme at the top had covered the lobule seen at the bottom in Fig. 7. It is of
particular interest that a narrow but long indentation like a tract runs over the
lobule (Fig. 8), which would probably be growing to a new cleft. There were no
obvious differences in the fibrillar structures between the lobular surfaces of the
shallow cleft and noncleft regions (Fig. 9). The indentation in the epithelium in
Figs 1-6. Scanning electron micrographs of intact 12-day gland, e, epithelium; m,
mesenchyme. *, defects made by the needle tip during fracturing of the fixed gland.
Fig. 1. Overall view of the specimen. Lobular and stalk epithelia and their mesenchymal replica are clearly seen, in addition to a part of the broken stalk at the
bottom. x250.
Fig. 2. Exposed surfaces of the epithelium and mesenchyme of the 12-day gland
seen at the right-hand side in Fig. 1. Several indentations in the lobule and dense fibrils
on the stalk can be seen. x340.
Figs 3, 4. Top view of the lobule in Fig. 2. There are several indentations and fibrils
with different diameter in the higher magnification view (Fig. 4) of the area indicated
by an arrow in Fig. 3. Fig. 3, x660; Fig. 4, X3370.
Fig. 5. High magnification view of the area indicated by an arrow in Fig. 2, in which
a dense fibrillar network is visible. X4680.
Fig. 6. High magnification view of the rectangular area in Fig. 1. Numerous fibrils
are attached to the ridges corresponding to the indentations in Figs 2, 3. X1240.
70
Y. NAKANISHI, F. SUGIURA, J. KISHI & T. HAYAKAWA
SEM studies of submandibular gland branching
71
Fig. 9 corresponds to the ridge in the mesenchyme in Fig. 10. Collagen-like
fibrillar structures are seen preferentially very close to or on the mesenchymal
ridge (Figs 10,11). Essentially the same results were obtained when intact mid 13day glands were examined with the scanning electron microscope.
In a micrograph of another specimen of the 12-day gland cultured in normal
medium (Fig. 12), bundles of fibrils, indicated by arrows, were detected at the
neck of the lobules and between lobules. The maximal diameter of such bundles
was approximately 2-0 /mi. These results suggest that the presence of large bundles
of fibrils is related to epithelial clefting induced by mesenchyme.
12-day glands cultured in the presence of CI
As reported previously, CI from the culture medium of bovine dental pulp when
included into the culture medium strongly enhanced cleft formation in submandibular glands, i.e. five clefts were formed on the average compared with two
in control cultures (Nakanishi etal. 1986a). Therefore, the relationship between
the localization of the fibrillar structures and the cleft formation was examined.
As is evident in Fig. 13, two shallow clefts are visible on one of the four lobular
surfaces of the gland cultured for 11 h with CI (5/igmP 1 ) and the corresponding
ridges are observed in the mesenchymal replica (Fig. 14). Micrographs of these
areas taken at higher magnifications (Figs 15, 16) indicate that the mesenchymal
ridges contain thick fibrils along them and are consistent with Figs 6 and 11.
Figs 17-21 show a gland cultured for 11 h in the presence of CI (5/igmr 1 ),
which was fractured at the base of the clefts. These figures provide relevant
information about the three-dimensional alignment of the fibrillar structures. In
Fig. 17, lateral surfaces of three lobular epithelia, consisting of columnar and
rather round cells, are seen and many fibrils are associated with the basal surfaces
of the epithelia facing the mesenchyme (Fig. 18). The most important finding is
that upper and lower parts of the mesenchyme are connected by large bundles of
Figs 7-12. Scanning electron micrographs of specimens of 12-day glands cultured in
normal medium for 17h. e, epithelium; m, mesenchyme.
Fig. 7. Overall view of the cracked specimen. A lobule and the corresponding
mesenchyme can be seen. x410.
Fig. 8. Top view of the same specimen in Fig. 7. Both micrographs indicate that a
fragment of mesenchyme at the top had covered the lobule at the bottom in Fig. 7. A
shallow cleft indicated by small arrows, which would be becoming a new cleft, traverses
the lobular surface. x570.
Fig. 9. High magnification view of the area indicated by a large arrow in Fig. 8.
There are no defined arrays of thefibrillarstructures on the lobular surface. A shallow
cleft is indicated by an arrow. X4060.
Fig. 10. Side view of the mesenchymal replica of the lobule in Fig. 8. A vertically
running ridge is shown to rise from the mesenchymal surface. X1990.
Fig. 11. High magnification of the front view of the mesenchymal ridge indicated by
an arrow in Fig. 10. An arrow in Fig. 7 also shows the same area. Bundles offibrilsare
associated with the ridge. X5790.
Fig. 12. Overall view of the lobular surface of another specimen. Arrows indicate
the thick bundles offibrilsaround the base of each lobule. x410.
72
Y. N A K A N I S H I , F . S U G I U R A , J. K I S H I & T.
HAYAKAWA
Figs 13-16. Scanning electron micrographs of a specimen of 12-day glands cultured in
the presence of CI (5 jugim""1) for 11 h.
Fig. 13. Surface view of the lobule having two shallow clefts indicated by arrows a
and b. X1050.
Fig. 14. Surface view of the mesenchymal replica of the lobule in Fig. 13. Two ridges
(b and a), corresponding to the shallow clefts (a and b), respectively, in Fig. 13, are
arising from the mesenchymal surface. X1000.
Figs 15, 16. Fibrillar structures at the mesenchymal ridges observed in Fig. 14.
Figs 15 and 16 are higher magnifications of the ridges b and a, respectively. Fig. 15,
X4950; Fig. 16, X5020.
fibrils (0-5-2-5/im in diameter) running vertically between two adjacent lobules
(Figs 17, 20, and see also Fig. 12). Evidence against the possibility that such large
fibres might be formed by artefactual collapsing of the fibre meshwork is provided
by the finding of similar fibres between two lobules (Fig. 19, and see also Fig. 12).
The figure clearly shows that the fibre consists of much thinner fibrils gathering
together from the lower part of the mesenchyme (Figs 17, 19-21). These observations indicate that there must be some unknown mechanism for the formation
of such bundle structures working between two adjacent lobules.
Some glands that were cultured for 17 h in the presence of CI were found to
accumulate heavy fibrillar structures at the epithelial-mesenchymal interfaces
Figs 17-21. Scanning electron micrographs of a specimen of 12-day glands cultured in
the presence of CI (5 ^gml"1) for 11 h. e, epithelium; ra, mesenchyme.
Fig. 17. Lateral surfaces of the mesenchyme and epithelial exposed after vertical
cracking of the specimen at the base of the clefts. X570.
Fig. 18. High magnification view of the epithelial-mesenchymal interface and the
lateral surface of the lobule of the right rectangular area in Fig. 17. It is clear that
lobular epithelia consist of both columnar cells outside and rather round cells inside.
X1710.
Fig. 19. High magnification view of the middle rectangular area in Fig. 17, showing
the presence of a bundle of fibrils between two adjacent lobules. X2430.
Fig. 20. High magnification view of the left rectangular area in Fig. 17, showing a
long fibre connecting the mesenchyme at the upper and lower parts of the figure.
X1610.
Fig. 21. High magnification view of the rectangular area in Fig. 20, indicating that
the fibre consists of many thinner fibrils. X4030.
74
Y. N A K A N I S H I , F . S U G I U R A , J. K I S H I & T.
HAYAKAWA
Figs 22-25. Scanning electron micrographs of specimens of 12-day glands cultured in
the presence of Clostridial collagenase (Sjugml"1) for 11 h. e, epithelium; m, mesenchyme.
Fig. 22. Surface view of the lobule with the smooth contour. x750.
Fig. 23. High magnification view of the rectangular area in Fig. 22. No fibrillar
structures are found on the lobular epithelium. X5340.
Fig. 24. Surface view of the mesenchyme originally apposed to the lobule in Fig. 22.
X2120.
Fig. 25. Surface view of the mesenchyme obtained from another specimen. No
fibrillar structures are observed as in Fig. 24. X3760.
(Nakanishi etal. 1986«), consistent with the presence of collagenase activity in
12-day glands and also with its inhibition by CI in vitro (Nakanishi et al. 1986a).
12-day glands cultured in the presence of bacterial collagenase
We then analysed the Clostridial collagenase-treated glands (5/zgml" 1 , 11 h),
which have no clefts or smooth contours (Nakanishi etal 1986a,fr). The cracking
method used in this study was only occasionally successful in obtaining specimens
that were fractured at the epithelial-mesenchymal interface. In contrast to those
of normal and Cl-treated glands, fibrillar structures were not found on the
epithelial surfaces and short, micro villi-like cell processes were seen (Figs 22-25).
The mesenchymal replica of the lobule (Fig. 24) and that from another specimen
SEM studies of submandibular gland branching
75
(Fig. 25) did not have a significant number of fibrillar structures either. The
fibrillar material in the collagenase-treated glands was extremely scarce, indicating
that such specific structures found at the epithelial and mesenchymal surfaces
of control and Cl-treated glands consisted of collagenase-susceptible materials.
Since the collagenase-treated glands did not as frequently expose epithelial and
mesenchymal surfaces originally apposed, it is likely that the presence of these
fibrillar materials at the epithelial-mesenchymal interfaces (Figs 17,18) facilitated
the separation of the two different tissues by a needle tip.
DISCUSSION
The process of the cleft formation in epithelium consists of two major parts:
initiation and stabilization. Two findings in the present study seem to be important
Collagen fibre
J
Mesenchymal
cells
Epithelial
cells
Collagen
fibrils
Fig. 26. A possible model for the initial cleft formation in the epithelium of mouse
embryonic submandibular gland. The formation of one cleft in a lobule is hypothetically depicted. The lobule of 12- and 13-day glands is packed with columnar and rather
round cells. In this model the narrow cleft, like a furrow, is made by contracting a
bundle of fibrils (schematically drawn as a rope-like structure) at the epithelialmesenchymal interface by mesenchymal cells. Another terminal of the mesenchymal
cell mass, which takes part in the traction, may be anchored to the lateral aspect of the
lobule or to the stalk where heavy deposition of collagen takes place (see Figs 1, 2, 5).
Although mesenchymal cells are depicted to be widely distributed, the epithelium is
sheathed in a capsule of mesenchymal cells in situ.
76
Y. NAKANISHI, F. SUGIURA, J. KISHI & T. HAYAKAWA
in the initiation process. First, at the beginning of branching no defined arrays of
fibrillar structures were observed on the surfaces of the lobules (Figs 9,13, and see
also Spooner & Faubion, 1980). Rather, the fibrillar structures located either on or
in the mesenchyme ran mostly along the preclefts or clefts (Figs 6, 11, 15, 16).
Secondly, we found fibres, 0-5-2-5 /mi in diameter, consisting of many thinner
fibrils between two adjacent lobules with relatively deep clefts (Figs 12, 17,
19-21), strongly supporting an earlier result obtained by transmission electron
microscopy in which a large bundle of collagen fibrils (diameter 0-8 jum) was seen
at the very base of the deep cleft (Bernfield & Wessells, 1970).
These observations are reminiscent of the recent findings that mesenchyme and
several fibroblastic cells are able to contract silicone rubber as well as collagen gel
(Bell etal. 1979; Steinberg etal. 1980; Harris etal. 1981; Stopak & Harris, 1982).
Judging from the localized and oriented distribution of fibrillar collagens described
so far, it is highly probable that the mesenchyme initiates clefting by establishing
ridges containing fibrillar architectures on the surfaces of lobular epithelium. A
possible model for cleft formation in the submandibular epithelium is presented in
Fig. 26. At the beginning of the cleft formation mesenchymal cells align collagen
fibrils by their traction to form ridges which cut into the lobular epithelium in the
region of the forming clefts. This process may accelerate the bundle formation
which facilitates the production of a narrow, deep cleft (Fig. 26). In this case, the
collagen fibres or fibrils may provide the substratum for the mesenchymal cells,
consequently generating the deforming force on the epithelial surface. The alignment of fibrils by fibroblastic cells is supported by the finding of Harris et al. (1981)
that collagen fibrils in the gel, which show random arrays at the onset of culture,
become axially oriented between two cultured cell masses, possibly resulting in
the formation of fibres. However, important questions as to whether the submandibular gland mesenchyme can contract collagen gel and as to how the lobular
epithelium containing basal lamina behaves in the initiation process remain to be
resolved.
The authors wish to thank Dr K. Takata of Radioisotope Center of Nagoya University for the
helpful suggestions and the use of scanning electron microscope, Dr K. Yamazaki-Yamamoto
of Nagoya University for help with electron microscopic observation and Mr H. Yamauchi for
drawing the model. Y. N. and F. S. wish to express their thanks to Professor S. Suzuki for the
continuous support. This work was partly supported by grants from the Ministry of Education,
Science and Culture of Japan and the Ishida Foundation.
REFERENCES
S. D., COHN, R. H. & BERNFIELD, M. R. (1977). The basal lamina of embryonic
salivary epithelia: Production by the epithelium and role in maintaining lobular morphology.
/. Cell Biol. 73, 445-463.
BELL, E., IVARSSON, B. & MERRILL, C. (1979). Production of a tissue-like structure by contraction
of collagen lattices by humanfibroblastsof different proliferative potential in vitro. Proc. natn.
Acad. Sci. U.S.A. 76, 1274-1278.
BERNFIELD, M. R. (1970). Collagen synthesis during epithelio-mesenchymal interactions. Devi
Biol. 22, 213-231.
BANERJEE,
SEM studies of submandibular gland branching
11
M. R., BANERJEE, S. D. & COHN, R. H. (1972). Dependence of salivary epithelial
morphology and branching morphogenesis upon acid mucopolysaccharide-protein (proteoglycan) at the epithelial surface. /. Cell Biol. 52, 674-689.
BERNFIELD, M., BANERJEE, S. D., KODA, J. E. & RAPRAEGER, A. C. (1984). Remodeling of the
basement membrane as a mechanism of morphogenetic interaction. In The Role of Extracellular Matrix in Development (ed. R. L. Trelstad), pp. 545-572. New York: Alan R. Liss.
BERNFIELD, M. R. & WESSELLS, N. K. (1970). Intra- and extracellular control of epithelial
morphogenesis. Devi Biol. (Suppl.) 4, 195-249.
GROBSTEIN, C. (1954). Tissue interaction in the morphogenesis of mouse embryonic rudiments
in vitro. In Aspects of Synthesis and Order in Growth (ed. G. Rudnick), pp. 233-256. Princeton,
NJ: Princeton University Press.
GROBSTEIN, C. & COHEN, J. (1965). Collagenase: Effect on the morphogenesis of embryonic
salivary epithelium in vitro. Science 150, 626-628.
HARRIS, A. K., STOPAK, D. & WILD, P. (1981). Fibroblast traction as a mechanism for collagen
morphogenesis. Nature, Lond. 290, 249-252.
KALLMAN, F. & GROBSTEIN, C. (1965). Source of collagen at epitheliomesenchymal interfaces
during inductive interaction. Devi Biol. 11,169-183.
KISHI, J. & HAYAKAWA, T. (1984). Purification and characterization of bovine dental pulp
collagenase inhibitor. J. Biochem. 96, 395-404.
NAKANISHI, Y., SUGIURA, F., KISHI, J. & HAYAKAWA, T. (1986a). Collagenase inhibitor stimulates
cleft formation during early morphogenesis of mouse salivary gland. Devi Biol. 113, 201-206.
NAKANISHI, Y., SUGIURA, F., KISHI, J. & HAYAKAWA, T. (19866). Local effects of implanted
Elvax chips containing collagenase inhibitor and bacterial collagenase on branching
morphogenesis of mouse embryonic submandibular glands in vitro. Zool Sci. (in press).
NOGAWA, H. (1983). Determination of the curvature of epithelial cell mass by mesenchyme in
branching morphogenesis of mouse salivary gland. /. Embryol. exp. Morph. 73, 221-232.
SPOONER, B. S. & FAUBION, J. M. (1980). Collagen involvement in branching morphogenesis of
embryonic lung and salivary gland. Devi Biol. 11, 84-102.
SPOONER, B. S. & WESSELLS, N. K. (1972). An analysis of salivary gland morphogenesis: Role of
cytoplasmic microfilaments and microtubules. Devi Biol. 27, 38-54.
STEINBERG, B. M., SMITH, K., COLOZZO, M. & POLLACK, R. (1980). Establishment and transformation diminish the ability of fibroblasts to contract a native collagen gel. /. Cell Biol. 87,
304-308.
STOPAK, D. & HARRIS, A. K. (1982). Connective tissue morphogenesis byfibroblasttraction. 1.
Tissue culture observations. Devi Biol. 90, 383-398.
VON DER MARK, K. (1981). Localization of collagen types in tissues. Int. Rev. Connective Tissue
Res. 9, 265-316.
WESSELLS, N. K. & COHEN, J. H. (1968). Effects of collagenase on developing epithelia in vitro:
Lung, ureteric bud, and pancreas. Devi Biol. 18, 294-309.
BERNFIELD,
{Accepted 25 March 1986)