/. Embryol. exp. Morph. Vol. 67, pp. 113-125, 1982
Printed in Great Britain © Company of Biologists Limited 1982
Cell interactions in the
developing somite: in vitro comparisons between
amputated (am/am) and normal
mouse embryos
By O. P. FLINT 1 AND D. A. EDE 2
From the Department of Zoology, University of Glasgow
SUMMARY
Facial, axial and limb development are all abnormal in the homozygous mutant mouse
embryo (amputated). An interpretation of cell behaviour in vivo based on sectioned material
which may explain these abnormalities has been previously suggested. In this study, somite
cells cultured in vitro were found to behave exactly as predicted in this interpretation: they
clump together, forming extensive areas of cell contact, and this has a profound effect on
their mobility as measured by time-lapse cinemicrography. The similarity of cell behaviour
in vitro and in vivo under two distinct sets of environmental conditions suggests that the
abnormal cell behaviour is intrinsic to the cell, and directly linked to the mutation. The
more extensive areas of cell contact formed between mutant cells suggests that the mutation
changes the adhesive properties of the cell surface, but it cannot be excluded that the cells'
motile apparatus is also affected.
INTRODUCTION
In a number of previous studies we have examined the role of abnormal cell
behaviour in the development of the mutant mouse amputated (Flint, 1977a, b;
Flint & Ede, 1978 a, b and Flint, Ede, Wilby & Proctor, 1978), concentrating
on in vivo aspects of morphogenesis. In these, interpretation of embryonic cell
behaviour was based upon a series of micrographs taken from TEM sections or
from SEM preparations. In this paper we report observations on living mutant
and normal embryonic cells in in vitro culture, using, as the main tool, time
lapse cinemicrography. The cells chosen for study were those leaving explants
prepared from somites of 9-5-day mice, and our observations confirm and
amplify our interpretations of the mutant cell behaviour in vivo. The basis of
this altered cell behaviour is further investigated in an SEM analysis of cell-cell
contacts in the cultured cells.
1
Author's address: ICI Pharmaceuticals Division, Mereside, Alderley Park, Macclesfield,
Cheshire SK10 4TG, U.K.
2
Author's address: Developmental Biology Building, 124 Observatory Road, Glasgow
G12 9LU, U.K.
113
114
O. P. FLINT AND D. A. EDE
METHODS
The mice
The amputated gene is kept on a background of a C3H/101 hybrid intercross.
Details of matings for the production of embryos of known age can be found in
Flint & Ede (1978 a). All pregnant females were opened under sterile conditions,
9-5 days after noting a copulation plug. Each female was killed by cervical dislocation and dipped, for external sterilisation, in a solution of cetrimide (2 %)
and sodium hypochlorite (2 %) in water (Bleby, 1972). Unless otherwise
stated, all the solutions used in the preparation of the tissue cultures described
below were warmed to 37-5 °C.
Somite culture
The technique for preparation of mouse somites for culture is a modification
of that described by Cooper (1965). Each conceptus (decidua plus embryo) was
dissected out into fresh Tyrode's solution, and transferred to a solution of horse
serum plus Tyrode's (1:1, v/v) where the embryos were removed. Embryos with
18 somites were chosen and taken through the explant preparation as pairs of
normal and amputated littermates. All the results were recorded as littermate
pairs and the statistical analysis takes this into account. Any littermate pairs
not prepared for explants were fixed for histological examination.
The embryos were treated in the following way. Any ventral trunk tissue up
to the level of the somites was dissected away with fine forceps and pipetted off
with a Pasteur pipette. The caudal half of the body was then cut off and placed
in the well of a cavity slide in a solution of 3 % trypsin (DIFCO, 1:250) in
calcium- and magnesium-free Tyrode's solution (CMF) for one minute at room
temperature. After gently aspirating the tissue two or three times through the
mouth of a Pasteur pipette to loosen the epidermis the tiunk was transferred to
a drop of horse serum and Tyrode's. It was then possible to remove the remaining epidermis and cleanly dissect the somites from the neural tube and notochord
with finely sharpened tungsten needles. In every case the last four somites alone
were dissected free from both sides of the trunk. These somites were then transferred quickly to a drop of 0-25 % trypsin in CMF for one minute at room
temperature. It was found that if this second exposure to trypsin was omitted
the explants would not later adhere to the culture dish (in a trial run, 1 out of 20
adhered if this step was omitted, 20 out of 20 if included). The somites were then
transferred to a fresh drop of horse serum and Tyrode's and chopped up into
six to eight pieces, which were then dropped into culture medium to form the
explants and cultured at 37-5 °C in 5 % CO 2 : 95 % air. The medium was Eagle's
Minimal Essential Medium (MEM) plus medium supplements of non-essential
amino acids, vitamins, sodium bicarbonate buffer and 10 % foetal calf serum
(FCS). Antibiotics were also added to the medium to a final concentration of
0-02 % streptomycin and 0-012 % benzyl penicillin.
Mouse somite cell behaviour in culture
115
Filming
The cultures were incubated for 12 h while the explants adhered to the substrate and cells migrated centrifugally from the explant edge. Filming of each
culture lasted a further 11 h and 7 min. Cultures were filmed as littermate
derived pairs of normal and amputated somite explants on two Wild inverted
microscopes connected by a comparison tube. If the cultures had been filmed
sequentially the second culture, at the start of filming, would have been 11 h
older than the first and it is possible that ageing may cause changes of cell
motility. Abercrombie & Heaysman (1953) could detect no effect of culture age
on heart fibroblast movement after 5 h of culture but Martz (1973) could not
exclude an age-dependent effect reducing 3T3 motility over much longer
periods in culture. It was therefore decided that filming of pairs would remove
a possible and unnecessary bias. Only one pair of littermate-derived cultures
was chosen per experiment for filming. Other culture pairs were left in the
incubator and later fixed (see below).
Filmed cultures were maintained on the micioscope stage in a hot room at
37-5 °C. Each petri dish was held in a small perspex and glass chamber which
was gassed with 5 % CO2. Illumination was either by phase or optical shadow
casting (Hlinka & Sanders, 1970). Fields for filming were chosen at random
within the margin of the moving cell front. A lapse rate of one frame in 10 sec
was chosen and controlled by a Payard-Wild Variometer. Twenty culture pairs
were filmed in this way.
Fixation of cultures
After filming the cultures were returned to the incubator and left to allow
further spreading of the explant. After a total of 44 h of culture all the explants,
including those not filmed, were fixed following the technique of Revel &
Wolken(1973).
Preparation for light microscopy
Petri dish bases were coated with carbon under vacuum at an angle of 60° in
a Speedivac Coating Unit (Edwards High Vacuum Ltd).
Preparation for scanning electron microscopy
It was not possible to dry the cultures by the critical-point technique because
the solvents used attacked the plastic culture dishes. Glass coverslips were not
used for the sake of uniformity in the culture of filmed cells and the other
cultures. The results of air drying were critically compared with living cells
observed under optical shadow casting (see below). The drying technique
employed did not appear to cause any significant distortion or cracking. Revel
& Wolken (1973) compared the scanning EM appearance of rapidly dried tissue
cultured cells with critical-point preparations of similar cultures. They found
O. P. FLINT AND D. A. EDE
116
Normal
Amputated
50|im
0
\
0
Mouse somite cell behaviour in culture
117
that both techniques gave very good results, though slightly more shrinkage
occurred in cells not dried by the critical-point method. Each piece of plastic
with an explant was cut out from the petri dish base and mounted on a 12 mm
aluminium stub with colloidal silver (Polaron) and uniformly coated with
500 nm of gold in a Polaron sputter coater. The explants were examined with
a Cambridge S600 scanning electron microscope in the Anatomy Department of
Glasgow University at accelerating voltages of 15 kV.
Film analysis
The films were analysed with a Specto Mark III motion analysis projector.
Film was projected onto tracing paper on a glass-topped table via a mirror
placed under the table at an angle of 45°. Cells were chosen at random at the
beginning of each film and the position of the nucleolus marked. The new
position of the nucleolus was marked every fifty frames (8-33 min) unless the
cell either left the field, entered mitosis or lost its adhesion with the substrate
(usually prior to mitosis). If any of these events occurred a new cell was chosen.
If the nucleolus did not move after 50 frames, the length of time it remained
stationary was recorded in units of 50 frames (rest length). If a cell moved, the
length of each step was measured for each 50-frame unit (step-length). The cell
density in each measured frame was also recorded.
Of the 20 pairs of cultures filmed it was possible to analyse 18 normal cultures
and 19 amputated cultures. 158 normal cells were followed (an average of
8-78 ± 2-24 (s.D.) cells per film) and 151 amputated cells (an average of 7-95 ± 2-27
cells per film). Normal cells were followed on average for paths of 57-76 ±
23-99 jam (2-12 ± 0-88 h), and amputated cells for 42-14 ± 28-35 pm (2-19 ±
1-47 h). Normal cells moved an average number of 12-16 ±5-05 steps, and
amputated cells 8-28 ± 5-57 steps in this time.
RESULTS
Living cells in vitro
The explant consists of a central piece of somite which flattens slowly as cells
migrate away from its edge onto the substrate. Cells were bipolar or multipolar,
with a higher proportion of bipolar cells in normal cultures, and movement was
normally in the direction of the most active lamellipodium in each cell (Fig. 1).
Movement also occurred when cells in contact by extended filopodiawere hauled
together by contraction of these filopodia. Within explant borders in normal
Fig. 1. A sequence of stills from time-lapsefilmsof cells moving away from explants
and amputated somites. Illumination is by optical shadow casting (see Methods).
Taking the top frame as time-zero, the interval between each subsequent frame is one
hour, up to a total of 4 h. This is indicated by the figure in the bottom right hand
corner of each frame.
* - <£*
Fig. 2. Scanning electron micrographs of (a, c) Normal and (b, d) amputated cells at the edge of somites
explanted from 9-5 day embryos. / = filopodium.
I
!i
1
II
JP
a
r
00
Mouse somite cell behaviour in culture
119
cultures cells tended to move away from one another, leaving large spaces
between cells. The filopodia connecting them became stretched over quite long
distances but they did not form permanent contacts, and soon broke away. By
comparison, in the amputated cultures, cell contacts were maintained over much
longer periods. Cells leaving the edge of the central solid explant remained close
together, with extensive marginal cell-cell contact, and near the margin of the
explant border, cells were far less dispersed than in normal cultures. Gaps
eventually formed between cells so that the explant border came to consist of
a series of interconnected clumps rather than a continuous cell sheet. This was
in marked contrast to the loosely dispersed pattern of cells within normal
explant borders (Fig. 1).
Scanning electron microscopy of cells
Optical shadow casting gives a true three-dimensional relief image of the cell
surface (Hlinka & Sanders, 1970) so that the living cells in Fig. 1. may legitimately be compared with scanning electron micrographs of cultured cells. The
chief difference is that these cells in SEM appear to be flatter than the living
cells, indicating that the external structure of the cell has collapsed to some
extent during drying. But details of cell contact and morphology are much more
highly resolved in scanning'electron micrographs than by optical shadow-casting.
The much longer filopodial connections between normal cells are clearly
shown in Fig. 2. Normal cells have a much more elongated, often bipolar,
appearance and are separated more widely than amputated cells, which form
characteristic clumps. The extent of cell margin in contact between adjacent
cells is much less in normal than in amputated cultures (Fig. 2 a, b). Three types
of cell contact were observed: (1) the apposition of one cell surface against
another without much deformation of the cell. Only very small areas of this
type of contact can be seen in normal cultures, for example in the lower right
corner of Fig. 2a, but very large areas of this kind of contact can be seen anywhere in the amputated cultures (Fig. 2b). (2) Fairly wide and tapering filopodia.
These are found in both normal and amputated cultures, but they are usually
much longer in normal cultures (compare Fig. 2 a, b). (3) Thin filopodia of
uniform diameter (microspikes). These are much longer in normal cultures but
short microspikes are much more numerous in amputated cultures. In general
there is a greater area of cell contact in amputated cultures.
There are also differences between mutant and normal at the points of cell
contact. When one normal cell makes contact with another, the actively moving
lamellipodium or filopod can move a considerable way over or under the
surface of the other cell. In amputated cultures filopodia hardly overlap or
underlap the other cell at all (Fig. 2 c, d), though the number of amputated filopodial contacts can be so great as to form a mesh work between neighbouring
cells (Fig. 2 d).
120
O. P. FLINT AND D. A. EDE
y
•
•
„•
O
o
o
u
I
100 90
Edge
80 70 60
50 40
30 20
I
I
10
0
Centre
Distance from explant centre
(% of total distance)
Fig. 3. Change of cell density with distance from the centre of somite explants, in
the case of the Normal (O) and amputated (•) mouse. These results were obtained
from carbon-coated cultures, observed on a glass stage on the Wild M20 microscope.
The explant border is roughly radially symmetrical about the explant. Using
a camera-lucida attachment the position of each cell within the border was recorded
by drawing a point. Because of the radial symmetry of the explant border it was
possible to estimate the centre of the explant on the drawing by fitting a circle
round the border margin. The average radius of normal explants was 488 ± 152 /tm
and of amputated explants was 470± 152 fim. A series of rectangles, 300 x 100/tm
were drawn along a randomly chosen radius, until no more cells were included in
a rectangle. The cell density in each rectangle was estimated. Over all the rectangles
the average cell density in normal cultures was 19-60± 11-52 cells per 105 /*m2
(477 cells, 83 rectangles counted) and in amputated cultures was 21 10± 11-73 cells
per 105 /im2 (500 cells, 94 rectangles counted). There is no significant difference
between linear regressions for normal and amputated (slope: P > 0-25, elevation:
P > 010). The regression line calculated from both sets of data is y = 39-993203 x (y = cell density, x = distance from explant centre). No results are given
for the most central 15% of the radius. This represents the flattened and very
dense remains of the original explant, where cell density could not be accurately
measured.
Cell density in the explant border
Measurements on fixed and carbon-coated cultures indicated no significant
difference in the average cell density of normal and amputated explant borders.
A clear linear regression emerged when cell density was plotted against distance
from the explant centre in both normal and amputated explants (Fig. 3), and no
difference was found between amputated and normal regressions (see legend to
Fig. 3). This indicates that cell density within the explant border is inversely related to distance from the explant centre and that there is no difference of cell density between normal and amputated explants at any point in the explant border.
Mouse somite cell behaviour in culture
121
Cell movement in cultured somite cells
Three parameters were assessed in measuring the cell paths from time-lapse
films:
(1) Cell speed.
(2) Length of time the cell was at rest.
(3) Length of each step taken by the cell when moving in the chosen interval
of 8-33 min.
This follows the method of analysis of Ede & Flint (19756). Additional
measurements were taken of the cell density in the film frame over the period of
path measurement. Cell density varied only slightly over this period of path
measurement in each film, as cells entered or left the field.
Cell speed
Cell speed (/on/hour) was calculated both for distance taken as the sum of the
lengths of all the steps making up each cell's path and for distance 'as the crow
flies', i.e. the linear distance between the cell's first and last positions. In the
first case there was no change of cell speed with cell density in either amputated
or normal cultures by Bartlett's three group method* (Sokal & Rohlf, 1969,
p. 481). Normal cells moved significantly faster than amputated cells (Table 1).
In the second case, of cell speed 'as the crow flies', there was again no change of
cell speed with cell density by Bartlett's three group method in amputated or
normal cultures, and no difference between normal or amputated average
speeds (Table 1).
Time at rest
Time at rest was measured in two ways: as a percentage of the whole time
a cell was observed and as the average length of time in minutes a cell stayed at
rest. There was no change in amputated or normal cultures in the average
length of rest with cell density in either case (analysed by Bartlett's three group
method). Normal cells spend on average significantly less time at rest when this
is calculated as a percentage of the whole time the cell was observed (Table 1),
or as average length of each rest taken in minutes (Table 1). Clearly mutant
cells pause more often and for longer periods than do normal cells.
Step length
There is no change of average step length with cell density in all cultures by
Bartlett's three group method nor is there any significant difference between
amputated and normal (Table 1).
* It was not possible to sample cultures on a blind trial basis, since it was always obvious
from cell morphology which culture was beingfilmed.The method of statistical analysis
chosen takes into account bias that might have been introduced as a result of this knowledge,
by entailing stricter criteria for assigning significance.
5375-76
46-29
12-90± 8-73 17-34± 14-18
509± 1-21
3498-30
6234-51
10-46± 7-81 ll-34± 9-80
33-51 ±15-37 55-32± 18-25
4-75± 118
6811-30
27-54± 11-37 19-24± 10-10
No
No
No
No
12-44
4547-52
No
(3)
S
2158-73
1129-80
2965-86
N(2)
normal
(4)
5216-51
39-73
2990-22
2751-36
253216
Amputated
No
No
No
No
No
(5)
S
(6)
1932-82
77-91
61-95
36723-86
12630-99
N/am
No
Yes
No
Yes
Yes
(7)
S
so?T»ites. A statistical p.vsiliiatinn
p
p mted norrrial and aimnutated
p
of this data by the sum of squares simultaneous test procedure follows under columns 1-7. In column 1 is the critical sum of squares for
comparison with the computed sum of squares between groups to be compared (columns 2, 4 and 6). If the critical sum of squares is larger
than the comparison value then there is no significant difference between groups. The groups compared are: all normal groups (column 2),
and all amputated groups (column 4); i.e. a comparison of the change of the measured values with cell density. A comparison is then made
between the sets of normal and amputated results in column 6. Whether or not the comparisons reveal significant differences is shown in
columns 3, 5 and 7.
Speed (/tm/hr) over whole
path
Speed (/tm/hr) in a straight
line between start and finish
% of time spent at rest
Average time of each rest
(minutes)
Length of each step taken by
cell in chosen interval of
8-33 minutes
Normal
Critical sum Amputated of squares
(1)
Table 1
J
z
0
O
H
1
to
to
A. ED
Mouse somite cell behaviour in culture
123
CONCLUSIONS
Cell morphology
In both normal and amputated cultures, cells migrating from the explant are
at first confluent but tend to disperse the further they are from the explant
centre. The pattern of cells within the explant border suggests that whereas
normal cells disperse individually, amputated cells disperse as small clumps,
i.e. as groups of cells which remain in contact over long periods of observation
in time-lapse films. In amputated the cells are strongly bound together by large
numbers of adhesive contacts, including short filopodia in large numbers,
sometimes amounting to a meshwork, preventing their moving apart. In
normal cultures the few and relatively extensible filopodia allow the cells they
connect to move apart. The smaller degree of overlapping and underlapping by
filopodia and lamellipodia in amputated cultures suggests that the stronger
adhesions tend to immobilise moving cell projections more rapidly than the
weaker adhesions formed by normal cells.
There is a striking similarity between these cultured somite cells and amputated
mesenchyme cells observed in the embryo in the sclerotome (Flint & Ede,
1978 a), the facial mesenchyme (Flint, 1977 a; Flint & Ede, 19786) and in the
palate (Flint, 1980). In all these places amputated cells clump together, far
greater areas of cell contact are observed between amputated cells than between
normal cells, and filopodia tend to mesh together into knots as they do in the
cultures of amputated cells (Fig. 3d).
Cell movement
Taking the age of each culture when fixed as 44 h, and the average width of
the explant border in normal and amputated cultures as 408 jum (480 ju,m
average explant radius, less 15 % for the explant itself) the rate of advance of
the cell border margin is 9-3 /im per hour. This is in very close agreement with
the measurement of cell speed for amputated and normal cells 'as the crow
flies' (11-0 /*m per hour). Cell speed as measured in this way is therefore approximately equivalent to the rate of radial migration of cells within the explant
border. Normal and amputated explant cultures expand at the same rate, but it
is the details of individual cell movement within the border that differentiate
between the two.
Normal cells move significantly further, over the whole path, in an hour than
amputated cells (Table 1), because they spend less time at rest. This follows from
the looser, more dispersed arrangement of cells in the normal explant border
and from the normal cells forming weaker contacts when they meet other cells,
so that the resumption of cell movement after contact inhibition is not delayed.
Rather similar observations on cell movement have been made in other
mutants in which morphogenesis is disturbed by abnormalities of cell-cell
contact behaviour. In the chick mutant talpid* Ede & Flint (19756) found that
124
O. P. FLINT AND D. A. EDE
cells migrating from primary explants of talpid3 chick wing mesenchyme, moved
more slowly than normal chick cells because they spent more time at rest. This
was correlated with other measurements showing that talpid3 cell-cell adhesions
in rotation reaggregation experiments were stronger than normal cell adhesions
(Ede & Flint, 1975 a). The talpid3 cell, like the amputated cell, appears to form
more extensive and stronger adhesions than normal cells, inhibiting cell movement in culture, but not to such a marked degree, so that clumping is not
visible in vivo (Ede & Flint, 19756). Another example is the t9 mutant of the
mouse, in which Spiegelman & Bennett (1974) found that the movement of
cells through the primitive streak of t9/t9 embryos was retarded because the
ingressed cells formed extensive contacts and because of this, tended not to
disperse as a primary mesenchyme. Yanigasawa & Fujimoto (1977) in a study
of t9 cell aggregation in vitro have been able to confirm that the more extensive
cell contacts observed indicate stronger adhesions between mutant cells.
Relation to morphogenesis
In each of these examples a good case has been made for the effects upon cell
contact relations, and consequently on cell behaviour, being the basic cause of
morphogenetic abnormalities produced by the mutant genes - in amputated,
abnormalities of somitogenesis (Flint et al. 1978), facial development (Flint &
Ede, 19786) and cleft palate (Flint, 1980); in t9, abnormalities of early development and axis formation; in talpid3, abnormalities of the developing limbs
(Ede & Flint, 1975a, b) and the somites (Ms in preparation). In all of them
abnormalities of cell movement and cell adhesion have been demonstrated in
in vitro culture, but it would be too simple to say merely that in all of them the
cells are abnormally sticky and relatively immobile. If this implied that the
abnormalities at the cellular level were identical it would be difficult to understand why their effects on morphogenesis are different. But this is not the case;
e.g. in their behaviour and morphology amputated and talpid3 cells have quite
distinct characteristics, and this would explain why their effects upon morphogenesis are not the same. Bellairs, Sanders & Portch (1980) have shown how
subtly the behavioural properties of chick embryo cells alter with time in the
course of mesodermal differentiation, with each type of mesoderm exhibiting
characteristic patterns of cellular behaviour in vitro. Alterations in these
behavioural properties produced by mutant genes will be equally subtle and
diverse, and through them we may ultimately expect to discover a complex
dynamic programme of gene-controlled cell behaviour patterns underlying all
of the morphogenic aspects of embryogenesis.
The authors wish to thank the Science Research Council for financial support.
Mouse somite cell behaviour in culture
125
REFERENCES
M. & HEAYSMAN, J. E. M. (1953). Observations on the social behaviour of
cells in tissue culture. I. Speed of movement of chick heart iibroblasts in relation to their
mutual contacts. Expl Cell Research 5, 111-131.
BELLAIRS, R., SANDERS, E. J. & PORTCH, P. A. (1980). Behavioural properties of chick
somitic mesoderm and lateral plate when explanted in vitro. J. Embryol. exp. Morph. 56,
41-58.
BLEBY, J. (1972). Disease-free (S.P.F.) animals. In The UFAW Handbook on the Care and
Management of Laboratory Animals. Churchill Livingstone, Edinburgh and London.
COOPER, G. W. (1965). Induction of somite chondrogenesis by cartilage and notochord:
A correlation between inductive activity and specific stages of cytodifferentiation. Devi
Biol. 12, 185-212.
EDE, D. A. & FLINT, O. P. (1975 a). Intercellular adhesion and the formation of aggregates
in normal and talpid3 mutant chick limb mesenchyme. /. Cell Science 18, 97-111.
EDE, D. A. & FLINT, O. P. (19756). Cell movement and adhesion in the developing chick
wing bud, studies on cultured mesenchyme cells from normal and talpid3 mutant embryos.
/. Cell Science 18, 301-313.
FLINT, O. P. (1977a). Cell interactions in facial development in the mouse. /. Anat. V2A,
225-226.
FLINT, O. P. (19776). Cell interactions in the developing axial skeleton in normal and mutant
mouse embryos. In Vertebrate Limb and Somite Morphogenesis (ed. D. A. Ede, J. R.
Hinchliffe & M. Balls), pp. 464-484. Cambridge University Press.
FLINT, O. P. (1980). Cell behaviour and cleft palate in the mutant mouse, amputated. J.
Embryol. exp. Morph. 58, 131-142.
FLINT, O. P. & EDE, D. A. (1978a). Cell interactions in the developing somite: in vivo comparisons between amputated {am/am) and normal mouse embryos. J. Cell Science 31,
275 -292.
FLINT, O. P. & EDE, D. A. (19786). Facial development in the mouse: A comparison between
normal and mutant {amputated) mouse embryos. /. Embryol. exp. Morph. 48, 249-267.
FLINT, O. P., EDE, D. A., WILBY, O. K. & PROCTOR, J. (1978). Control of somite number in
normal and mutant mouse embryos. /. Embryol. exp. Morph. 45, 189-202.
HLINKA, J. & SANDERS, F. K. (1970). Optical shadowcasting of living cells. Trans. N.Y.
Acad. Sci. 32, (6) 675-682.
MARTZ, E. (1973). Contact inhibition of speed in 3T3 and its independence from postconfluence inhibition of cell division. /. Cell Physiol. 81, 39-48.
REVEL, J. P. & WOLKEN, K. (1973). Electron microscope investigations of the underside of
cells in culture. Expl Cell Research 78, 1-14.
SOKAL, R. R. & ROHLF, F. J. (1969). Biometry. The Principles and Practice of Statistics in
Biological Research. W. H. Freeman and Co., San Francisco.
SPIEGELMAN, M. & BENNETT, D. (1974). Fine structural study of cell migration in the early
mesoderm of normal and mutant mouse embryos (r-locus: f9//9). J. Embryol. exp. Morph.
32, 723-738.
YANIGASAWA, K. O. & FUJIMOTO, H. (1977). Differences in rotation-mediated aggregation
between wild-type and homozygous Brachyury (T) cells. /. Embryol. exp. Morph. 40,
277-283.
ABERCROMBIE,
(Received 19 May 1981, revised 28 July 1981)
EMB
6;