FINE STRUCTURE AND MORPHOGENIC

MOVEMENTS IN THE GASTRULA

OF THE TREEFROG, HYLA REGILLA

PATRICIA C. BAKER, Ph.D.

From the Department of Zoology, University of California, Berkeley, California. Dr. Baker's present
address is Department of Pathology, University of Oregon Medical School, Portland, Oregon

ABSTRACT

The blastoporal groove of the early gastrula of the treefrog, Hyla regilla, was examined with
the electron microscope. The innermost extension of the groove is lined with invaginating
flask- and wedge-shaped cells of entoderm and mesoderm. The distal surfaces of these cells
bear microvilli which are underlain with an electron-opaque layer composed of fine granular
material and fibrils. The dense layer and masses of vesicles proximal to it fill the necks of the
cells. In flask cells bordering the forming archenteron the vesicles are replaced by large
vacuoles surrounded by layers of membranes. The cells lining the groove are tightly joined
at their distal ends in the region of the dense layer. Proximally, the cell bodies are separated
by wide intercellular spaces. The cell body, which is migrating toward the interior of the
gastrula, contains the nucleus plus other organelles and inclusions common to amphibian
gastrular cells. A dense layer of granular material, vesicles, and membranes lies beneath the
surface of the cell body and extends into pseudopodium-like processes and surface undulations
which cross the intercellular spaces. A special mesodermal cell observed in the dorsal lining
of the groove is smaller and denser than the surrounding presumptive chordamesodermal
cells. A long finger of cytoplasm, filled with a dense layer, vesicles and membranes. extends
from its distal surface along the edge of the groove, ending in a tight interlocking with
another mesodermal cell. Some correlations between fine structure and the mechanics of
gastrulation are discussed, and a theory of invagination is proposed, based on contraction
and expansion of the dense layer and the tight junctions at distal cell surfaces.

INTRODUCTION

The ordered movement of embryonic cells in been thought to exert a pulling force on the in-
gastrulation has long been a subject for speculation voluting mesodermal cells by means of a surface
and investigation. In the last sixty years various coat covering all exterior cells of the gastrula. It
experimental approaches have been taken, some was hoped that by observing the ultrastructure of
reviewed below, in an attempt to understand the the invaginating cells some new insight might be
correlative movements of gastrulation as well as gained into the old problem of the mechanics of
the role of individual cells. These studies show that morphogenic movements.
one group of cells initiate gastrulation, namely
the flask or bottle cells of the entoderm, which by MATERIALS AND METHODS
directional movement cause an initial insinking, Pregastrular stages of the Pacific Treefrog, Hyla regilla
the blastoporal groove, that is deepened by further (Baird and Girard), were collected from ponds in the
invagination of these cells. The flask cells have also Berkeley hills and stored at {Ce. The embryos were

95
removed from the refrigerator and allowed to develop
to early gastrular stages (Figs. 1 A and 8 A) at room Early Invagination-Stage lOA
temperature. The dorsal half of the gastrula, includ- At Stage lOA (Fig. I A) the blastoporal groove
ing the blastopore and part of the future yolk plug, in sagittal section has the appearance of a U with
was isolated in Niu-Twitty solution (30). The opera-
long arms (Figs. I B and I C), the base of the U
tion was performed with a hair loop and fine glass
corresponding to the deepest extension of the
needle on the smooth surface of a wax-lined petri
dish. Great care was taken to cut the embryo cleanly blastoporal groove. The cells lining the groove in
and to avoid stretching or damaging the cells. this early period may assume one of several shapes
The specimens were then fixed for 1 to 2 hours at (Fig. I C): (I) flask shape with long, narrow neck
4°C in Dalton's solution (7) adjusted to pH 7.4 with distally and roughly spherical body proximally, (2)
potassium hydroxide. The samples were quickly de- wedge shape with convoluted exterior surface, and
hydrated in acetone using Parson's method (31) and (3) cuboidal shape with a surface which usually is
flat-embedded in Epon (28). The block of plastic was smooth, but which may bear microvilli in those
trimmed so that only the blastoporal area was cells near the deepest part of the groove.
sectioned. Excising the dorsal half of the embryo, The flask-shaped blastoporal cells are found at
rather than the blastoporal region alone, not only
the base of the U with their necks oriented perpen-
reduced possible operative damage to the blastoporal
dicular to the blastoporal cavity (fe, Figs. I C and
cells, but facilitated orientation of the specimen.
Sections less than 100 m~ in thickness were cut 2). They are mostly presumptive entoderm, al-
with a glass knife and stained with lead citrate (33) though some mesodermal cells may become flask-
for 20 to 30 minutes. The electron micrographs were shaped as they reach the deepest extension of the
made with an RCA EMU-3G. groove (16). The wedge-shaped cells of the en-
toderm and mesoderm are found to either side of
OBSERVATIONS
the flask cells (we, Fig. I C). The cube-shaped
The blastoporal cells have a striking appearance in cells are characteristic of the involuting mes-
electron micrographs. Their exterior surfaces and oderm on the dorsal arm of the U and of the
necks are elaborately structured-densely packed entoderm on the ventral arm.
with cytoplasmic particles or filled with mem-
1. FLASK CELLS
branes, vesicles and vacuoles. This feature con-
trasts markedly with that of other cells of the gas- Parasagittal sections of the blastoporal crescent
trula which are largely undifferentiated and in reveal flask cells with very long stalk-like necks at
which cytoplasmic organelles are just beginning to the base of the U (fe, Fig. 2). A given neck may
increase in number and complexity and in which reach a length exceeding 40 J.I. while being only
the most prominent inclusions are yolk, lipid drop- 0.3 to 2.0 J.I. in width. As one continues parasagittal
lets, and cytoplasmic particles (1,13,22). Observa- sectioning toward the median plane of the embryo,
tions were made on the early gastrula, Stage 10 of flask cells with shorter, broader necks (fe, Fig. 5)
Shumway (37). For convenient reference, early replace the long-necked cells. The number of flask
Stage 10 is divided into Stages lOA and lOB; in cells increases until, in a saggital section of the
Stage lOA the blastoporal groove is shorter than blastoporal groove, the necks of as many as ten
that in Stage lOB (compare Figs. I A and 8 A). flask cells border on the base of the U (Fig. 6).

FIGUREI A Early gastrula at Stage lOA with blastoporal groove just beginning.

FIGUREI B Sagittal section of Stage lOA. Flask and wedge cells may be seen bordering
the blastoporal groove; the cavity is the blastocoel.

FIGURE1 C Drawing from large montage of electron micrographs of a parasagittal sec-

tion of blastoporal groove at Stage lOA. The groove (bg) has the appearance of a U, with
the base of the U being the innermost extension of the groove. Long-neck flask cells (fe),
wedge cells (u'C), and cuboidal cells (cc) line the groove. The distal surface of these in-
vaginating cells contains a dense layer (dl) and a vesicular layer (vl). Note the polymor-
phous lipid droplets (pi) peculiar to invaginating cells at Stage lOA. er, endoplasmic reticu-
lum; m, mitochondria; mv, microvilli; pg, pigment granules; y, yolk platelets.

96 THE JOURNAL OF CELL BIOLOGY' VOLUME 24, 1965

 Fig.IA

.'we.' '.' ..... :

. .

bg

Fig.le

PATRICIA C. BAKER Fine Structure in Amphibian Gastrula 97

 The flask cells have long microvilli on their
In addition to the lip junctions mentioned ear-
exterior surfaces (Figs. 4 to 7). These projections
lier, the distal ends of the flask cells adhere to
contain fine cytoplasmic particles and a few ves-
adjacent cells in a variety of ways. In Fig. 4 the
icles, but no pigment, yolk, lipid, or other inclu-
end of the neck is complexly interdigitated on both
sions. In flask cells with extremely long necks (Figs.
sides with wedge cells. In Fig. 6 the edges of the
2 and 4), both microvilli and necks are filled with
cells are so interlocked as to make cell boundaries
the same dense granular substance. The microvilli
hard to distinguish. Cell membranes may be quite
adhere, along part or all of their length, to the
close together with the intercellular space filled
surface of adjacent cells, forming an overlapping
with a dense substance (see arrows, Fig. 7). More-
lip (fl, Figs. 2 and 4). over, interpretation is sometimes difficult because
In the shorter, broader flask cells the microvilli of folds in the cell membrane or because of tan-
are filled with particles 30 to 40 mu in diameter gential sections.
(g, Fig. 7), which are presumably glycogen gran- Beneath the dense layer, the necks of flask cells
ules (12). Occasionally, a vesicle is seen among the are packed with masses of vesicles (v, Figs. 5 to 7),
particles (Fig. 7). These shorter cells may adhere which may be round or crescent-shaped with an
to adjacent ones at their distal surfaces by a trun- interior matrix denser than the surrounding cyto-
cated lip (comparefl, Figs. 4 and 5) which may fit plasm. The vesicles are interpsersed with particles
snugly into a depression in the surface of a neigh- 10 to 40 mu in diameter, in the size range of ribo-
boring cell (fl, Fig. 5). nucleoprotein and glycogen particles (g, Figs. 5
An electron-opaque layer, which hereafter will and 6). In the distal part of the vesicular layer are
be referred to as the dense layer, varying in thick- vacuoles, some surrounded by double membranes
ness from 0.08 to 1.5 ~ extends in a band beneath (vc, Figs. 5 and 6), a few pigment granules, and
the cell surface of all the cells of the blastopore at some granular bodies similar to the ones seen in
the deepest extension of the groove (dt, Figs. I C wedge cells in Figs. 2 and 5. At a distance of some
and 2). This layer is thinner (50 to 80 mu) and 5 to 20 ~ into the neck of the flask cells, the usual
discontinuous in cells near the exterior of the blas- organelles and inclusions appear: yolk platelets,
toporal groove. It is prominent, however, beneath mitochondria, lipid droplets, and Golgi apparatus
the exterior surface of involuting presumptive (Fig. 3). The lipid droplets in the distal parts of the
chordamesodermal cells just as they turn under the flask and wedge cells at this stage are unusual in
dorsal lip (not shown in these figures). In the long being polymorphous (pl, Figs. I C and 2). The
flask cells (fe, Fig. 2) the dense layer extends into lipid droplets in more proximal areas of the cells
the microvilli and up the neck of the cell as far as (Fig. 3). and in other cells of the gastrula at this
40 ~ (dt, Fig. 3). At higher magnification, the sub- stage are roughly spherical in shape (I).
stance filling the neck appears uniformly granular
(Fig. 4). In the shorter, broader flask cells the 2. '''EDGE CELLS
dense layer may widen into a round or triangular
The dense layer in wedge cells is not so thick as
mass (dt, Fig. 5). In these cells the dense layer also that in flask cells (Figs. 2 and 5), and it contains
appears to be granular, but traces of fibrils may faint fibrils oriented parallel to the outer cell sur-
be seen oriented parallel to the long axis of the cell face (j, Figs. 4 and 5). The microvilli are more
(f, Fig. 7). lobular in shape and less densely packed with par-

98 THE JOURNAL OF CEI,L BIOLOGY' VOLUME 24, 1965

 FIGURE 2 Parasagittal section of inner extension of blastoporal groove at Stage lOA,
showing arrangement of long-neck flask cell (fc), wedge cells (we), and cuboidal cell (cc).
The ventral side of the groove is at the top of the figure, the dorsal side at the bottom.
bg, blastoporal groove; dl, dense layer; er, endoplasmic reticulum; fl, flask cell lip; gb,
granular body; m, mitochondria; mo, microvilli; pg, pigment granule; pi, polymorphous
lipid droplets; vi, vesicular layer; y, yolk platelets. X 17,000.

FIGliRE 3 Part of neck of flask cell seen in Fig. 2 (fc), 30 ~ from distal cell surface. The
flask cell is bordered by proximal parts of wedge cells (we). Notice the dense layer (dl) still
present beneath the lateral cell membrane. " lipid droplets; m, mitochondria; v, vesicles;
y, yolk platelet. X 24,000.

98 THE JOURNAL OF CEI,L BIOLOGY' VOLUME 24, 1965

 ticles than those of the flask cells (Fig. 2). Whereas sions presumably correspond to the pseudopodia
the vesicular layer in wedge cells (Fig. 4) has formed by isolated blastoporal cells (18, 19). Al-
fewer vesicles and is narrower than that in flask though widely separated at their proximal edges,
cells (Fig. 5), the vesicles are similar in size and most of the flask cells are closely joined at their
appearance. distal ends (Fig. 9). Some cells at the very tip of
The distal ends of the wedge cells are tightly the blastopore, however, appear to have loosened
joined with the same type of junctions as in flask their intercellular connections (see arrows, Fig.
cells: interdigitation of lateral cell membranes 10). As invagina ting cells are known to release their
(Fig. 4), overlapping lips (wi, Fig. 5), and dense connections and migrate into the blastocoel during
intercellular "cement" (Fig. 5). Proximally ad- archenteron formation (16), intercellular con-
jacent cells may be separated by intercellular nections of the flask cell to the right in Fig. 10 may
spaces 20 to 80 tiu» wide (Fig. 4). be loosening prior to such a migration.
The microvilli on the surface of the cells now
Formation of the Archenteron-Stage lOB
appear thinner and longer than in Stage lOA, and
At a slightly later period in gastrulation, the they are filled with the substance of the dense
blastoporal groove assumes a longer crescent layer. Again, within the granular matrix of the
shape (compare Figs. I A and 8 A); sagittal and layer faint fibrils may be detected running parallel
parasagittal sections of the embryo reveal that to the surface (f, Fig. II). Whorls of membranes
morphogenic movements of the cells have deep- and stacks of vesicles are occasionally found em-
ened and narrowed the groove (Figs. 8 B and C). bedded in the dense layer (Fig. 12).
The archenteron is beginning to from as a roughly Beneath the dense layer are large vacuoles en-
spherical cavity at the anterior-most extension of closed by a single membrane (vc, Figs. 9 and 10).
the groove, i.e. at the base of the U seen in sec- Some vacuoles are embedded in a network of
tional view (Fig. 8 C). This cavity is bounded an- membranes and tubules and appear to be filled
teriorly by entodermal and mesodermal flask cells, with a homogeneous electron-transparent material.
dorsally by cells of the involuting mesoderm, and The masses of membranes which surround the
ventrally by flask and wedge cells of the entoderm vacuoles and continue into the cell neck are not
(Figs. 8 C and 9). Although the gastrulating cells arranged in any discernible pattern. In some areas,
appear to have moved a relatively short distance the membranes appear to be aggregations of
in the period of time between Stages lOA and lOB, vesicles (v, Fig. II); other areas contain highly
marked changes have occurred in their structure folded membranes (ml, Fig. 10) interspersed with
and arrangement. small oblong vesicles or spaces. Farther along the
neck the membranous layer gives way to lipid
1. FLASK CELLS
droplets, yolk platelets, mitochondria, pigment
The flask cells seen at Stage lOB are presumably granules, cytoplasmic particles, and endoplasmic
the same ones which initiated invagination, now reticulum (Figs. 9 and 10).
augmented by a few involuting mesodermal cells The body of an invaginating cell contains the
converted to flask cells (16, 35, 36). By Stage lOB,
nucleus as well as the above-mentioned organelles
the necks of the cells have shortened and become
and inclusions. Its dense cortex contains mem-
more complexly structured (compare Figs. 9 and
branes and small vesicles (mv, u, Fig. 13). This
10 with Figs. 2 and 5). Proximally, the flask cells
are now separated by wide intercellular spaces layer seems a thinner version of the dense layer
(Figs. 8 C, 10, and 13), across which finger-like in the cell neck. Finger-like psuedopodia which
extensions of the cytoplasm may reach from one bridge the intercellular gaps are also filled with the
cell to another (p, Figs. 10 and 13). These exten- dense substance (p, Fig. 13).

FIGURE 4 Higher magnification of distal end of neck of flask cell in Fig. 2, showing lip
(ft) of cell, interlocking of flask (fc) and wedge cells (wc) in region of dense layer (d!), and
faint fibrils (f) in dense layer. bg, blastoporal groove; v, vesicles in vesicular layer. X 65,000.

100 THE JOURNAL OF CELL BIOLOGY' VOLUME 24, 1965

 ~. WEDGE CELLS
15 to 40 mu in diameter, presumably ribonucleo-
At Stage lOB, wedge-shaped cells are confined to protein and glycogen, may be seen in abundance
the ventral entodermal lining of the archenteron in all parts of the cell (g, Figs. 14 and 15). Pigment
and are situated slightly distal to the deepest ex- granules are more concentrated than in other
tension of the groove (we, Figs. 8 C and 9). The mesodermal cells (compare adjacent cells in Fig.
surface of the cells bordering the groove is skewed 14).
toward the base of the groove, and the microvilli
DISCUSSION
are longer and more numerous than in the wedge
cells of Stage lOA (compare Figs. 2 and 9). The History
dense layer, not so dark as in the adjacent flask
cells, extends into the microvilli (Fig. 9). The The flask cells of the blastopore have been
cytoplasm proximal to the dense layer is filled studied with the light microscope by many inves-
with tubules and vesicles. tigators (14--16, 34, 35, 38), and numerous theories
have been advanced to explain their movement
3. INVOLUTING MESODERMAL CELLS (see reviews in 2, 8, 16, 17,36,38). Vogt as early as
1929 (39) suggested that invagination was due in
The mesodermal celis, easily distinguished because
part to an active contraction of the distal ends of
of their pigment granules, line the dorsal side of the
the blastoporal cells. Waddington (40) developed
forming archenteron and extend along the dorsal
this idea further by speculating that the causative
wall of the blastopore to the exterior (Figs. 8 C and
agent for invagination of the entoderm might be
9). Whereas most involuting cells appear cuboidal
fibers of protein at the outer edge of the blastoporal
in shape, mesodermal cells lining the forming
cells which actively contract and reduce the distal
archenteron are roughly rectangular with their
parts of the blastoporal cells, thereby initiating
long axis parallel to the arms of the U. Like the
invagination. In an attempt to obtain direct evi-
flask and wedge cells in this area, the mesodermal
dence for this theory, Picken and Waddington
cells are separated by intercellular spaces 0.5 to
(40) showed that the yolk-free necks of the flask
3 J.I. in width (Fig. 8 C); involuting mesodermal
cells exhibit a weak birefringence indicating the
cells distal to the point bg on Fig. 8 C, however,
presence of fibers running along the length of the
are closely packed.
processes and perpendicular to the outer surface of
In the area of the forming archenteron, a type of
the egg. Lewis (25, 26) developed a similar theory
cell appears among the involuting mesodermal working largely with models. He attributed invagi-
cells which bears some resemblance to the invagi- nation in Amblystoma punetatum to an active contrac-
nating flask cells. These special cells are smaller and tion of a "gel layer" at the outer ends of the ento-
denser than the surrounding mesoderm, and long dermal cells.
fingers of cytoplasm extend from their distal sur-
Holtfreter (15-21) has contributed the most
faces along the edge of the blastoporal groove (Fig. detailed experimental analysis of structure and
14). The tip of an extended finger is interlocked
movement of the blastoporal cells. He discounts the
with cytoplasmic extensions of other elongate idea that contraction of cell surfaces plays a part
mesodermal cells (cj, Figs. 14 and 15). Proximally, in invagination; rather, he thinks that invagination
the cell body adheres to adjacent mesodermal cells depends on the inward migration of the blastoporal
by pseudopodia (p, Fig. 14) which bridge the cells and on the properties of a "surface coat" cov-
intercellular spaces (i, Fig. 14). Microvilli on the ering the exterior of the embryo. The results of
surface are filled with extensions of the dense layer; his work pertinent to this discussion may be sum-
beneath the dense layer may be seen masses of marized as follows.
membranes and vesicles. Cytoplasmic particles The driving force initiating gastrulation is 10-

FIGURE 5 Another type of flask cell (fc) bordered by wedge cells at the innermost exten-
sion of the groove (Stage lOA, parasagittal section). by, blastoporal groove; dl, dense layer;
er, endoplasmic reticulum; f, fibrils in the dense layer; fl, flask cell lip; g, cytoplasmic
(glycogen ?) granules; gb, granular body; mv, microvilli; v, vesicles in vesicular layer; vc,
vacuole; wl, wedge cell lip; y, yolk platelet. X 33,000.

102 THE JOURNAL OF CELL BIOLOGY' VOLUME 24, 1965

cated within the flask cells of the entoderm. At the gastrulae and neurulae by means of focused ul-
time of gastrulation, the inner parts of these cells trasound. Electron micrographs have not sup-
begin to migrate inward, owing in part to some ported Holtfreter's hypothetical picture of a con-
difference in milieu between the proximal and tinuous cell covering congruent with the cytoplasm
distal surfaces of the cells (e.g. alkalinity of the or closely amalgamated to the cell membrane.
blastocoelic fluid) and partly to an inherent tend- Karasaki (22) found no evidence for a surface coat
ency of these cells to elongate and migrate along in electron micrographs of ectoderm of Triturus
their proximo-distal axis. As the blastoporal cell pyrrhogaster embryos ranging in age from early
migrates inward, by a mechanism of contraction blastula to late tail-bud, nor did Matsumoto (29)
and expansion of the cell surface, the distal part who examined neural plate cells and cells of the
remains firmly attached to a surface coat and is blastoporal groove in the same species. The elec-
drawn out into a flask shape. "When cellular de- tron micrographs of the blastoporal groove in the
formation has reached a certain point, the tensile frog Phrynobatraehus natalensis (3) reveal no extra-
strength of the elastic neck portion forces the cellular covering or coat.
coated surface to yield and to recede into the
interior in the form of the archenteron" (17). The
Structure of the Blastoporal Cells
tangential pull exerted by the invaginating blasto- The present study, in agreement with the studies
poral cells throughout the coat then pulls in the on Triturus and Phrynobatraehus, shows no evidence
marginal cells. of a surface coat. It may be possible that an extra-
Holtfreter (15) stated that the surface coat is a cellular mucopolysaccharide coat, such as that
layer, under elastic tension, which surrounds the seen in the light microscope surrounding fertilized
exterior of the embryo and binds the peripheral eggs (9) or removed by ultrasound (5), is present
cells together into a supercellular unit, integrating but not discernible by fixation methods used in
their movements. In an early paper (15) he said electron microscopy. It is doubtful, however, that
the coat was contractile, but in a later paper (38) such a covering plays a direct part in morphogenic
he stated the opposite. Holtfreter speculated that movements. It is hard to imagine a covering,
the coat either closely adhered to the cell mem- closely adhered to the cell surface and under elastic
brane or was an actual continuation of the cyto- tension, which could be penetrated by the long,
plasm. wavy microvilli seen in Figs. 2, 5, and 6 or which
As a surface coat is an essential feature in Holt- would allow the irregularity of cell surfaces seen
freter's theory of gastrulation, various investigators in Fig. 9. Presumably, the exterior cell surface is
have attempted to verify its existence. Dollander thrown into microvilli as the cuboidal cells change
(9, II) observed an extracellular coat surrounding to wedge and then flask cells. The microvilli are
the fertilized egg of Triton which stained with Nile longest in the cells whose distal surface is most
blue sulfate, but he was unable to demonstrate such compressed. Compare the microvilli (mv) in flask
a coat in later stages (10). Levtrup (27) stated that (ie) and adjacent wedge cell (we) with the surface
an extracellular coat exists in Rana temporaria em- of a cuboidal cell (ee) in Fig. 2.
bryos which takes part in osmoregulation by My findings support those of Holtfreter (16) that
mechanically opposing osmotic swelling. Bell (5,6) a long stalk -like neck is characteristic of flask cells
reported that a surface coat composed of muco- in earliest invagination; these cells are found in
polysaccharides can be removed from Rana pipiens Stage lOA, but not in Stage lOB. The necks of the

FIGURE 6 Flask cells at base of U in sagittal section of blastoporal groove (Stage lOA).
Sections are from the same specimen as that shown in Fig. 2. Notice the interlocking of
cell membranes. bg, blastoporal groove; dl, dense layer; g, cytoplasmic (glycogen?) granules
mv, microvilli; v, vesicle of vesicular layer; vc, vacuole. X 26,000.

FIGURE 7 Higher magnification of flask cells at base of U in sagittal section. Arrows point
out tight cell adhesions with a dense substance or "cement" between the cell membranes.
Note fibrils (f) in the dense layer (dl). bg, blastoporal groove; g, cytoplasmic (glycogen ?)
granules; mv, microvilli; v, vesicles in vesicular layer and one in a microvillus. X 67,000.

104 THE JOUR~AL OF CELL BroWGY. VOLUME 24, 1965

flask cells appear quite dark in light micrographs. Triturus pyrrhogaster show a dense band, 25 to 60
This was thought to be due to heavy pigmentation mu wide, composed of microparticles beneath the
of these regions (16, 26, 36), but the electron microvilli of involuting mesodermal cells inside the
microscope reveals that the opacity is due to blastoporal groove (the exact loca tion of the cells
densely packed particles, membranes, and vesicles is not made clear). A similar layer 150 to 200 mu
in the necks and distal ends of the flask cells (Figs. in thickness is seen beneath the cell membrane in
2, 5, 9, and 10), and not to pigment granules. his micrographs of the middle part of the neural
The general structure of the central part of the plate in Stage 19 embryos. Karasaki (22) reported
blastoporal pit in Phrynobatrachus natalensis (3) re- a dense layer at the base of the microvilli in elec-
sembles the same area in Hyla reg ilia (Fig. 6); the tron micrographs of neural ectoderm in Triturus
necks of flask cells, bearing long microvilli, crowd pyrrhogaster at Stage 14. Neither Matsumoto nor
the base of the groove. Balinksy did not observe the Karasaki speculates on the function of this layer.
long-necked flask cells seen in Figs. 2 and 5, pre- The dense bands described in the literature re-
sumably because his sections were from the sagittal semble the dense layer seen in my micrographs. In
plane of the blastoporal crescent. He observed no all cases, the band is composed of closely packed
dense layer or vesicular layer comparable to that particles and lies beneath the microvilli at the
seen in Figs. 5 and 6. The blastoporal cells in distal cell surface. The layer seems to present a
Phrynobatrachus appear to be tightly joined at their barrier to cellular inclusions, and has been found
outermost edges; often the cell membranes are only in flask and wedge cells of the embryo under-
more electron-opaque in these regions. Balinsky going morphogenic movement. Only in Hyla were
thinks that the electron-opaque substance may be fibrils seen in the dense layer (Figs. 5, 7, and II).
a cement. I agree with his interpretation that the This may be because the earlier micrographs of
adhesion of cells at their distal surfaces, whether Balinsky, Karasaki, and Matsumoto were not of
by "cement" (arrows, Fig. 7) or by interlocking sufficient resolution or magnification to reveal
(Figs. 4, 6, and 15), could account in part for the fibrils if present.
integrated mass movements of involuting cells. Holtfreter (16) observed that, as invagination
Although Balinsky (3) did not observe a dense proceeds and the blastopore becomes deeper, more
layer beneath the blastoporal cell surface in the and more wedge cells are transformed into flask
very early gastrula of Phrynobatrachus, his micro- cells, taking the place of flask cells which have mi-
graphs show a band of particles 0.08 to 0.2 p. wide grated into the interior cell mass or into the blas-
beneath the microvilli of cells in the middle of the tocoel. Some of the wedge cells seen in Fig. 2 prob-
neural plate. He postulates that this band is a ably represent intermediate forms in the conversion
contractile layer which is responsible for the wedge of an involuting mesodermal or entodermal cell
shape of the neural plate cells. Matsumoto's (29) into a flask cell. As the neck narrows and length-
electron micrographs of early Stage 12 gastrulae of ens, the dense layer widens and the vesicles become

FIGURE 8 A Stage lOB with blastoporal crescent slightly more pronounced than in Stage
lOA (Fig. 1 A).

FIGUHE 8 B Sagittal section of Stage lOB. The blastoporal groove has lengthened, and
the archenteron (arrow) is beginning to form at the inner end of the groove.

FIGURE 8 C Drawing from large montage of electron micrographs of inner extension of

blastoporal groove in sagittal section of Stage lOB. The archenteric cavity (ac) is beginning
to enlarge. The entodermal flask (fe) and wedge cells (we) forming the ventral lining of the
archenteron are seen in longitudinal section. The anterior end of the archenteron is lined
by flask cells of the entoderm and mesoderm, some cut in cross-section. Roughly rectangu-
lar and cuboidal mesodermal cells form the dorsal roof of the archenteron. Note the large
intercellular spaces (i) bridged by pseudopodium-like process (p) between cells, and the
specialized mesoderm cell (sm) with a dense cytoplasmic finger extending along the surface
of the involuting mesoderm. bg, blastoporal groove; I, lipid droplets; m, mitochondria;
ml, membranous layer; ve, large vacuoles in flask cells; y, yolk platelets.

PATRICIA C. BAKER Fine Structure in Amphibian Gastrula 107

Fig. SA

Fig.8e
tightly packed in the cell neck. The other inclusions trulation. The dense layer also offers a structure for
are displaced proximally. It is possible that the Waddington's (40) predicted mechanism of gas-
vesicles at Stage lOB coalese as the neck retracts to trulation (i.e., fibers or a protein matrix which
form the membranous layer seen at Stage lOB.The could undergo "fibrisation" and by contraction
vesicular layer, like the dense layer, is found only and expansion cause morphogenic movements), as
in those gastrular cells involved in morphogenic well as a mechanism for contraction and expansion
movements. It is likely that the vesicles of Stage of the proximal cell surface (19-21). If we assume
lOA and the membranes of Stage lOB are elements that the dense layer undergoes contraction and
of the endoplasmic reticulum involved in synthesis expansion, then invagination (i.e., insinking, blas-
of the dense layer. This will be discussed in more toporal groove formation, and subsequent archen-
detail in a later paper. teron formation) could be accomplished by the
The large vacuoles in the necks of the flask cells flask and wedge cells in a twofold manner: alter-
lining the archenteron may have an excretory or nate shortening and elongation of the dense layer
secretory function. The shortening and retracting in the neck and distal ends of the cells, and migra-
of the cell neck, which is presumably occurring in tion of the inner ends of the cells by periodic con-
those flask cells retreating into the interior (16), traction and expansion of the dense layer beneath
could be accomplished, in part, by an excretion of the outer cell surface. In the following discussion,
fluids from the cytoplasm of the cell neck. Or the reference will be made to the line diagram in Fig.
vacuoles may be pinocytotic. Balinsky and Walther 16 which offers a hypothetical scheme for the
(4) have observed large vesicles in flask cells of the action of flask cells in gastrulation.
primitive streak in the chick; they presume the Whether initial inpocketing of the entoderm is
vesicles to be pinocytotic, but they do not comment caused by a contraction of the dense band or by a
on the significance of such a process. pull exerted by the migra ling proximal ends, or
Holtfreter (17) was unable to explain what hap- both, is unknown. I have not observed the first
pened to the blastocoelic fluid when the archen- stages of invagination. Balinsky's micrographs (3)
teric cavity replaced the blastocoel, as the coat of what is probably an earlier stage than my Stage
covering the cells of the archenteron was relatively lOA do not reveal a dense layer in the blastoporal
impermeable to water. In the light of the present cells, which suggests that the initial insinking may
study it would seem that, as some of the flask cells be due to migration alone. By the time the blasto-
release their connection distally and retract into pore appears as a distinct crescent (Fig. I A),
the interior, a temporary canal is opened into the however, the dense layer is present beneath the
archenteron (see arrows, Fig. 10) which, in the surface of the cells lining the anterior-most exten-
absence of a coat, would allow blastocoelic fluid sion of the blastopore and in the long necks of the
under pressure from the invaginating cell mass to flask cells (Fig. 2). A contraction of the dense
pass into the archenteric cavity. layer in a plane parallel to the cell surface (Fig. 2)
would narrow the ends of the cells and deepen the
Mechanics of Gastrulation blastoporal groove (b and c, Fig. 16). A contraction
Conclusions concerning dynamic processes of the long neck in a plane perpendicular to the
drawn from studies of cell structure must be tenta- distal cell surface would further deepen the groove
tive, but the preceding micrographs suggest sup- (e andf, Fig. 16).
port for some of the theories which have been Although action in the neck may be partially
advanced to explain morphogenic movements. The responsible for invagination, the main inward pull
contractile band mechanism proposed by Balinsky is exerted by the migrating proximal ends of the
(3) for neurulation can now be postulated for gas- cells (16-19, 38). Such a pull presupposes an

FIGURE9 Archenteric cavity (ac) forming at inner end of blastoporal groove in Stage lOB,
lined dorsally by mesoderm, ventrally by entoderm, and anteriorly by flask cells (fc) of
mesoderm and entoderm. (See Fig. 8 C). dl, dense layer; gb, granular body; i, intercellular
space; I, lipid droplets; mv. microvilli; pg, pigment granule; vc, vacuoles in flask cells;
Ule, wedge cell; y, yolk platelets. X 5,000.

108 THE JOURNAL OF CELL BIOLOGY' VOLUME 24, 1965

anchoring of the cell surface in some way. The substance denser than the cell cytoplasm (24).
arrows in Fig. 13 may point out cellular adhesions One may speculate, then, that migration of the
at the tips of microvilli and at surface undulations. proximal ends of cells occurs in the following way.
Recently Lesseps (24) has shown that reaggregat- The dense layer expands to form projections of
ing heart and pigmented retinal cells in the chick the cell surface which temporarily anchor to other
embryo make initial contact at the crests of surface cells (p and arrows, Fig. 13). The dense layer then
undulations or at the tips of pseudopodium-like contracts. If the net movement forward owing to
processes. He suggests that these protrusions func- expansion exceeds the net movement backward
tion to reduce the repulsive force between ap- owing to contraction, the result will be the "three-
proaching cells, allowing the formation of adhesive steps-forward, two-steps-backward" movement
bonds which can then expand over a broader area. described by Holtfreter (19, 20, 38) in isolated
Such bonding may occur temporarily in the blasto- blastoporal cells. New pseudopodia expand and
poral cells, giving the cell traction as it moves adhere, and the flask cell continues to move in-
inward. ward, deepening the blastoporal groove. The net
After extensive experimental work on isolated inward movement of the proximal end is rep-
embryonic cells, Holtfreter (19, 20) concluded that resented by the longer dashed arrows in Fig. 16.
they do not move by sol-gel transformations sup- Because of tight junctions between adjacent
posedly responsible for amoeboid locomotion, but cells, the movement of the flask cells would ac-
by alternating contractions and expansions of the count, in part, for the coordinated movements of
cell surface. In the blastoporal cells this "three- involuting cells. As the invaginating cells contract
steps-forward, two-steps-backward" succession of distally and migrate inward they drag their neigh-
movements is confined to the inner end, resulting bors behind them, just as if they were attached to a
in a net movement toward the interior. The ques- supracellular covering. This passive movement is
tion immediately arises as to how the structure of not the whole story of involution, however, as the
the cell surface of a migrating blastoporal cell chordamesodermal cells have been shown to have
compares to pseudopodia of an ameba. In con- self-stretching tendencies (16, 25, 36), and the
trast to the highly structured migrating end of special mesoderm cell seen in Fig. 14 may play an
the blastoporal cell, electron microscopy has re- active role in involution. It is possible that contrac-
vealed little structure in the moving areas of cyto- tion and expansion of the dense layer in the long
plasm in amoebae (32). Electron micrographs of finger of the cell pulls adjacent mesodermal cells
Wohlfarth-Botterman (41, 42) show a lightly toward the interior of the embryo.
granular cytoplasm in pseudopodia which pre- The question remains as to the mechanism by
sumably were moving at the time of fixation, and which the dense layer contracts. Presumably, the
Lehmann (23) has shown the presence of a fibrous layer is made up of a proteinaceous matrix inter-
material beneath the plasmalemma of Amoeba spersed with protein fibers; whether the matrix or
proteus+ The pseudopodia of reaggregating heart the fibers undergo contraction is a matter for spec-
and retinal cells in the chick are also filled with a
moving portion of the ground plasm of the amoeba
Since acceptance of this manu-
) Note added in proof: Hyalodiscus simplex, Amoeba proteus, and the slime mold
script for publication, electron micrographs have Physarum (43). This morphological evidence adds
been published demonstrating the presence of a dense support to the contractility theory of amoeboid
matrix and thread-like or fibrillar elements in the movement.

FrGURE 10 Higher magnification of neck of flask cells at anterior surface of archenteron

(ae). Presumably these cells, the same ones which initiated insinking of the blastopore,
have reached the limits of invagination and are retracting into the interior. Arrows mark
the loosening of distal connections. Pseudopodium-like processes (p), which may anchor
the migrating cell, cross intercellular spaces (i). The vesicular layer seen in Figs. 5 and 6 is
replaced by a membranous layer (ml) which surrounds large vacuoles (ve). The charac-
teristic dense layer (dl) extends into the microvilli (mv). Proximal to the membranous layer
the cell neck contains granular bodies (gb), lipid (I), mitochondria(m), and yolk platelets
(y). X 17,000.

110 THE JOURNAL OF CELL BIOLOGY· VOLUME 2.4, 1965

ulation. It is hoped that better methods of fixation This paper is a section of a thesis submitted in partial
combined with experimental work on isolated fulfillment of the requirements for the Ph.D. degree.
flask cells will assist in the solution of these prob- Grateful acknowledgement is made to Prof.
lems. The theory of a contracting and expanding Richard M. Eakin for his generous and invaluable
assistance in preparing this paper for publication, to
dense layer offers, of course, only a partial expla-
Prof. William E. Berg for a critical reading of the
nation of the mechanics of gastrulation. Nothing
manuscript, to Mrs. Emily E. Reid for the drawings,
is known about why the flask cells begin invagina- and to the United States Public Health Service for
tion, what establishes their axial polarity, or what fellowship support.
determines which cells will elongate. Received Jor publication, Feb. 10, 1961.

REFERENCES

I. BAKER, P. C., Changes in lipid bodies during electronmicroscopy, Anat. Rec., 1955, 121,
gastrulation in the treefrog, Hyla regilla, Exp. 281.
Cell Research, 1963, 31, 451. 8. DEHAAN, R. L., Cell migration and morpho-
2. BAKER, P. C., Ultrastructure of the gastrula of genetic movements, in The Chemical Basis of
Hyla regilla, Doctoral thesis, University of Development, (W. D. McElroy and B. Glass,
California, Berkeley, 1964. editors), Baltimore, The Johns Hopkins Press,
3. BALINSKY,B. 1., Ultrastructural mechanisms of 1958.
gastrulation and neurulation, in Symposium 9. DOLLANDER, A., Observations concernant la
on Germ Cells and Development, Institut structure du cortex de l'oeuf de certains
International d'Embryologie and Fondazione amphibiens urodeles mise en evidence du
A. Baselli (1960), Pallanza, 1961, 550. "coat" de Holtfreter, Compt. rend. Assoc. Anat.,
4. BALINSKYB, . 1., and WALTHER, H., The immigra- 1951, 38,430.
tion of presumptive mesoblast from the prim- 10. DOLLANDER, A., Observations relatives a cer-
itive streak in the chick as studied with the taines proprietes du cortex de l'oeuf d'amphi-
electron microscope, Acta Embryol. et Morphol. bien, Arch. Anat. Micr. et Morphol, Exp., 1953,
Exp., 1961, 4, 261. 42, 185.
5. BELL, E., Dissociation of ectodermal cells and of II. DOLLANDER,A., La structure du cortex de l'oeuf
the "surface coat" from amphibian embryos de Triton observee sur coupes fines et ultra-
by means of focused ultrasound: the character
fines au microscope ordinaire, et au microscope
of the "surface coat," Anat. Rec., 1958, 131,
electronique, Compt. rend. Soc. biol., 1954, 148,
532.
152.
6. BELL, E., Some observations on the surface coat
and intercellular matrix material of the 12. EAKIN. R. M., Actinomycin D inhibition of cell
amphibian ectoderm, Exp. Cell Research, 1960, differentiation in the amphibian sucker, Z.
20, 378. ZellJorsch., 1964, in press.
7. DALTON, A. J., A chrome-osmium fixative for 13. EAKIN, R. M., and LEHMANN,F. E., An electro-

FIGURE 11 Distal surface of flask cell of Stage lOB at higher magnification to show part
of membranous layer (ml), vesicles (v), vacuoles (ve) surrounded by a single membrane,
and faint fibrils (f) in granular matrix of dense layer (al). X 48,000.

FIGURE 12 Part of distal surface of flask cell at Stage lOB showing whorls of membranes
(m) and vesicles (v) or cross-sections of tubules embedded in the dense layer (al). vc, vacuole.
X 46,000.

FIGURE 13 Parts of migrating inner ends of several adjacent flask cells at Stage lOB.
The pseudopodium-like processes (p) which cross the intercellular spaces (i) may adhere
temporarily to adjacent cells (arrows) serving as an anchor for the invaginating flask cell.
The mechanism for movement may reside in the structure of the granular matrix, vesicles
(v), and membranes (mv) which make up the dense layer (dl); gb, granular body; I, lipid
droplet; m, mitochondria; y, yolk platelets. X 21,000.

112 THE JOURNAL OF CELL BIOLOGY' VOLUME24, 1965

inInsiitnia
kliOng
0
mIingwraa
triodn;
neck
elongation
miFgurartthioenrQ;
neck
contraction
-- - - - -:;..elongation
- contraction
~'.
\
,

- --
-
Q b c d e f g

FIGURE 16 Theoretical scheme for the transformations of a blastoporal cell during gastrulation, based
on the hypothesis of contraction and expansion of the dense layer. The end of the cell neck remains securely
adhered to adjacent cells. (a) The cell as it appears on the surface prior to gastrulation. (b) The distal
surface contracts causing initial insinking of the blastoporal pit. (c), (d) The proximal cell surface begins
to migrate inward, deepening the groove; the distal surface continues to contract and the neck elougates.
(e), (f), (g) The neck contracts after it reaches maximum elongation. This shortening, together with the
pull exerted by the migrating inner ends, furthers invagination and pulls adjacent cells into the groove.
Some flask cells bypass stages d and c.

FIGURE 14 Special type of cell seen among the involuting mesodermal cells at Stage lOB.
(See sm, Fig. 8 C.) Part of the cell extends in a finger-like process along the surface of the
blastoporal groove (bg); the tip of the process forms an interlocking junction (cj) with an
extension of a similar cell. The body of the cell contains many pigment granules (pg) and
cytoplasmic particles (g) in the size range of glycogen granules; gb, granular body; i, inter-
cellular space; l, lipid droplets; p, pseudopodium-like processes; y, yolk platelets. X 8,000.

FIGURE 15 Higher magnification of area outlined in Fig. 14. The complex cell junction (cj)
between the tips of two extended mesodermal cells is indicated by arrows. Microvilli (mv)
filled with the granular matrix of the dense layer (dl) extend from the surface of the process
facing the blastoporal groove (bg). Beneath the dense layer are vesicles (e), membranes (m)
and cytoplasmic (glycogen ?) granules (g); i, intercellular spaces; pg, pigment granule.
X 33,000.

PATRICIA C. BAKER Fine Structure in Am/phibian Gastrula 1I5

 microscopic study of developing amphibian embryo in early developmental stages, J.
ectoderm, Roux' Arch., 1957, 150, 177. Electronmicroscopy, 1961, 10, 39.
14. HIS, W., Unsere Korperform und das physiolo- 30. NIU, M. C., and TWITTY, V. C., The differentia-
gische Problem ihrer Entstehung, Leipzig, tion of gastrula ectoderm in medium condi-
1874. tioned by axial mesoderm, Proc. Nat. Acad. Sc.,
15. HOLTFRETER,]., Properties and functions of the 1953, 39,985.
surface coat in amphibian embryos, J. Exp. 31. PARSONS,D. F., A simple method for obtaining
Zool., 1943, 93, 251. increased contrast in Araldite sections by using
l6. HOLTFRETER, ]., A study of the mechanics of postfixation staining of tissues with potassium
gastrulation. Part I, J. Exp, Zool., 1943, 94, permangenate, J. Biophysic. and Biochem,
261. Cytol., 1961, 11, 492.
17. HOLTFRETER, ]., A study of the mechanics of 32. PITELKA, D., Electron Microscope Study of
gastrulation. Part II, J. Exp. Zool., 1944, 95, Protozoa, New York, MacMillan Co., 19G3.
171. 33 REYNOLDS,E. S., The use of lead citrate at high
18. HOLTFRETER,]., Structure, motility and locomo- pH as an electron-dense stain in electron
tion in isolated embryonic amphibian cells, J. microscopy, J. Cell BioI., 1963, 17,208.
Morphol., 1946, 79,27. 34. RHUMBLER,L., Zur Mechanik des Gastrulations-
19. HOLTFRETER,]., Observations on the migration, vorganges, insbesondere der Invagination,
aggregation and phagocytosis of embryonic Arch. Enuocklngsmechn., 1902, 14,401.
cells, J. Morphol., 1947, 80, 25. 35. RUFFINI, A., Fisiogenia, Milano, F. Vallardi,
20. HOLTFRETER, ]., Changes of structure and the 1925.
kinetics of differentiating embryonic cells, J. 36. SCHECHTMANA, . M., The mechanism of
Morphol., 1947,80,57. amphib- ian gastrulation, Univ. California
21. HOLTFRETER, ]., Significance of the cell mem- Pub. Zool.
brane in embryonic processes, Ann. New York 1942, 51, I.
37. SHUMWAYW, ., Stages in the normal
Acad. Sc., 1948, 49, 709.
development of Rana pipiens. 1. External form,
22. KARASAKl, S., Electron microscope studies on
Anat. Rec.,
cytoplasmic structures of ectoderm cells of the
1940,78, 139.
Triturus embryo during the early phase of
38. TOWNES, P. L., and HOLTFRETER, ]., Directed
differentiation, Embryologia, 1959, 4,247.
movements and selective adhesion of em-
23. LEHMANN,F. E., Der Feinbau der Organoide bon
bryonic amphibian cells, J. Exp. Zool., 1955,
Amoeba proteus und seine Beeinflussung durch 128,53.
verschiedene Fixierstoffe, Ergebn. BioI., 1958, 39. VOGT, W., Gestaltungsanalyse am Amphibien-
21,88. keim mit ortlicher Vitafarbung. II. Teil,
24. LESSEPS,R.]., Cell surface projections: their role Gastrulation und Mesodermbildung bei
in the aggregation of embryonic chick cells as Urodelen und Anuren, Arch. Entwcklngsmechn.,
revealed by electron microscopy, J. Exp. Zool., 1929, 120, 384.
1963, 153, 171. 40. WADDINGTON, C. H., Organizers and Genes,
25. LEWIS, W. H., Mechanics of invagination, Anat. Cambridge University Press, 1940.
Rec., 1947, 97, 139. 41. WOHLFARTH-BoTTERMAN,K. E., Protistenstu-
26. LEWIS, W. H., Gel layers of cells and eggs and dien. X. Licht- und electronen-mikroskopische
their role in early development, Lect. Ser. Untersuchungen an der Amobe Hyalodiscus
Roscoe B. Jackson Mem. Lab., 1949, 59. simplex n. sp., Protoplasma, 1960, 52, 58.
42. WOHLFARTH-BoTTERMANNK, . E., Cytologische
27. L¢vTRuP, S., On the surface coat in the amphib-
Studien. VII. Strukturaspekte der Grund-
ian embryo, J. Exp. Zool., 1962, 150, 197.
substanz des Cytoplasmas nach Einwirkung
28. LUFT, ]. H., Improvements in epoxy resin em-
verschiedener Fixierungsmittel, Protoplasma,
bedding methods, J. Biophysic. and Biochem.
1961, 53,259.
Cytol., 1961, 9, 409.
43. WOHLFARTH-BOl'TERMANK, . E., Cell structures
29. MATSUMOTO,A., Electron microscopic observa-
and their significance for ameboid movement,
tions on the cytoplasmic processes in Triturus in International Review of Cytology, (G. H.
Bourne and]. F. Danielli, editors), New York,
The Academic Press, Inc., 1964.

116 THE JOURNALOF CELL BIOLOGY' VOI,UME24, 1965