/. Embryo/, exp. Morpli. Vol. 45, pp. 189-202, 1978
Printed in Great Britain © Company of Biologists Limited 1978
\ gO,
Control of somite number in normal and
amputated mutant mouse embryos: an experimental
and a theoretical analysis
By O. P. FLINT, 1 D. A. EDE, 1 O. K. WILBY 1 AND J. PROCTOR 2
From the Departments of Zoology and Mathematics,
University of Glasgow
SUMMARY
A regulation is shown for size and number of serially repeated axial structures, the somites,
in a mammalian embryo. The mammalian embryo is normally inaccessible to operation at
post-implantation stages. This problem is resolved by the quantitative analysis of somite
size, number and development in a recessive mutant of the mouse, amputated, whose axial
length is greatly reduced. The effect of the gene simulates an experiment ablating part of the
embryonic tissue available for somitic segmentation. Regulation occurs at the time when the
somite is first formed, by control of the quantity of cells included in each new somite. A
model is devised for the control of somitic segmentation which explains most of the features
observed and which can be simulated on a computer.
INTRODUCTION
Among the problems of morphogenesis, the generation of serially repeated
structures is an especially challenging one and the most striking examples of
such structures in vertebrates are the somites - the blocks of mesodermal cells
which arise in sequence along each side of the embryonic axis from the posterior
boundary of the head (otic capsule) to the tail. Hamilton (1969) studied their
development in the toad Xenopus and showed, using haploid embryos in which
cell size is reduced, that the quantity of material included in each somite is a
measured length of presomitic mesoderm rather than a specified number of
cells. Further studies by Cooke (1975) showed that Xenopus which have been
reduced in size by removal of some vegetal material at the blastula stage show
a capacity for regulation of somite number; that is they produce the normal
number of somites from a greatly reduced number of cells.
This raises the question of whether the capacity for such regulation exists
also in the embryogenesis of higher vertebrates or whether, like the capacity for
1
Authors' address: Department of Zoology, University of Glasgow, Developmental
Biology Building, 124 Observatory Road, Glasgow G12 9LU, U.K.
2
Author's address: Department of Mathematics, University of Glasgow, Glasgow Gl 2 9QQ,
U.K.
13
12 MB 45
190
O. P. FLINT AND OTHERS
limb regeneration in post-embryonic stages, it is limited to certain amphibians.
Studies we are making on a new mutant gene in the mouse have given us the
opportunity of investigating this aspect of somitogenesis in the mouse and also
analysing the relation of somitogenesis to some other embryonic processes.
The mutant, amputated, is a single recessive gene (Meredith, 1965; Flint, 1976),
one of whose effects is to reduce the embryonic length thus simulating the
experimental lesions of Cooke (1975). In interpreting our results we have had
to take into account one basic difference between the amphibians and higher
vertebrates; in amphibians somitogenesis does not begin until all the presumptive
mesoderm has been invaginated at the blastopore lip, but in higher vertebrates it begins anteriorly while presumptive mesoderm is still being invaginated
at the primitive streak, and more somites are initiated in sequence as the node
at the anterior end of the streak moves backwards. In amphibians, therefore,
the embryonic axis is present as a measurable length of mesoderm before the
beginning of somitogenesis and regulation of somite size may occur by direct
relation to it as Cooke (1975) suggests, but if regulation does occur in higher
vertebrates it must be governed by some different relationship since in them
the length of axial mesoderm is being continuously extended as somite initiation
takes place.
METHODS
The mice used were from a CBA/101 hybrid control stock and a mutant
stock derived from this by introducing into it the recessive lethal gene amputated (Flint, 1976). To obtain mutant embryos, known heterozygote males were
mated to possible female heterozygotes. Embryos were dissected out into
Tyrode solution. Measurements were made while they were still living, following the scheme shown in Fig. 1. Length of each embryo was measured along
its axis (i.e. neural tube) from the most anterior surface of the mesencephalon
(arrowed) to the posterior neuropore. The antero-posterior length of each
somite was also measured midway between the dorsal and the ventral surface.
Length of presomitic paraxial mesoderm was also measured. The number of
embryos used to measure each variable can be seen in Table 1.
Statistical analyses of the results consisted of regression analysis, analysis
of variance and covariance. These are routine techniques and descriptions can
be found in Snedecor & Cochran (1967) and Sokal & Rohlf (1969). We have
accepted probabilities (P) of 0-05 and less as an indication of significance.
RESULTS
A description of the embryos. Mutant homozygotes are shorter than their
wild-type littermates (Fig. 1, 4, Table 3). This includes all stages we have
measured up to 26 somites when, in the normal embryo the tail begins visible
outgrowth (Griineberg, 1956). The tail fails to grow in the mutant but at later
Analysis of somitogenesis in the mouse
191
Table 1. The numbers of embryos used for the measurements described in the
text. Except where indicated these are separate groups of embryos
Embryos (age in days, approximately)
8-5 days
Measurements
(+/+)
Axis length
Same
embryos
Most recently
formed somite
Ant/post lengths
of somites 5, 7, 9, 11
Length of each
somite in 18-somite
embryos
Total cell number
in most recently
formed somite (from
serial sections)
Somite number
Total
9-5 days
(+ / +)
9-5 days
(+/ +,
+/am)
9-5 days
(am/
am)
Total
16
—
32
36
88
16
—
32
36
88
—
—
36
36
72
—
—
8
6
14
—
—
7
7
14
—
80
79
30
189
377
stages than 26 somites its trunk is obviously shorter than in the normal embryo
(Flint, 1976). Accompanying a reduced axis there are anomalies of vertebral
cartilage condensation in the mutant from their earliest appearance, presaging
later gross vertebral fusions. These anomalies are not paralleled by any fusions
among the somites themselves before sclerotome differentiation begins (Fig. 1;
Flint, 1977).
Comparisons of somite size in mutant and normal embryos. At all levels of the
body axis mutant somites are smaller than those in normal embryos (Fig. 2).
This is confirmed by analysis of variance (P < 0-001). At some levels of the
body axis mutant somites are closer in size to normal somites than elsewhere.
But even where mutant and normal somites are closest in size, at the level
of somites 3, 4 and 5, the difference is statistically significant (P < 0-001).
There is a significant overall change in somite size with level in the body axis
(normal and amputated, P < 0-001). Anterior somites are larger than posterior
somites.
There is no difference between the sizes of somites in normal or in mutant
embryos at the time of segmentation, whether they are destined to be anterior
or posterior. Direct measurement of living embryos shows that over all the
stages measured there is no change in the size of the most recently formed somite
(Table 3, Fig. 4). But at all times this somite is 50 % smaller in amputated
than in normal embryos. The cell number included in this most recent somite is
13-2
192
O. P. FLINT AND OTHERS
correspondingly reduced in the mutant {amputated, 134 + 22-96; normal,
277±6-81 cells; P < 0-001).
Since anterior somites are larger than posterior somites it follows that somites
grow. This can be seen in Fig. 3 where the change in size of somites 5, 7, 9 and
11 is plotted against developmental age (somite number) in mutant and normal.
Each particular mutant somite remains consistently smaller than its normal
counterpart through all the stages we have measured. This is confirmed by
statistical analysis (Table 2).
How therefore do somites grow? Is it by growth of the cells themselves as
appears to be the case in Xenopus (Hamilton, 1969) or is it by an increase in
cell number? An analysis of cell density in the somite dermatome of every
somite in the axis of a pair of amputated and normal littermates shows that
somites grow mainly by increase in cell number. Cell density does not change from
somite to somite (17-somite normal embryo, average dermatome cell density:
11-32 ±2-99 cells per 105 /tm2 and no significant difference between somites,
P > 0-75; 13-somite amputated littermate, c.d.: 12-66 + 3-19 cells per 105 /im2
and P > 0-75; 7-8 sections counted per somite). But, as has already been
shown, anterior somites are much larger than posterior somites (Fig. 2).
Clearly, only if the cell density were to decrease with increasing somite size
could we say that cell growth played a major role in the growth of the
dermatome.
Regulation of somite size and number. We have counted somite number in
189 embryos at 9-5 days of development. There is no difference between amputated and normal littermates (amputated = 19-04±4-26; normal = 18-90±
3-51, P > 0-50). Nor is there any difference between these normal litter mates
( + / + , +/am) and wild-type embryos ( + / + ) from the control stock ( + / +
= 18-31 +4-68, P > 0-05) thus excluding any heterozygous effect. In spite of
a reduction in axial length in the mutant, it produces the same number of
somites as in the normal embryo. A regulation for somite size and number has
taken place. We can deduce from the evidence already presented how and at
what stage this regulation is effected.
From the earliest stage of their formation there is a difference of size between
mutant and normal somites. This difference is maintained, without deviation,
through all the stages we have observed (Fig. 3, Table 2). Regulation for
size must therefore have occurred at the time of somite formation. It is clear that
the regulation is effected by controlling the number of cells included in each
somite because normal somites at the time of their formation are about twice
the size of amputated somites (Fig. 4) and (see above) contain a proportionately
greater number of cells.
Possible mechanisms for the regulation of somite size. The amputated embryo
reaches the same developmental stage (i.e. somite number) as the normal
embryo at approximately the same time after fertilization. Development in the
amputated embryo therefore keeps pace with the normal embryo; organs emerge
Analysis of somitogenesis in the mouse
193
(b)
Fig. 1. Normal (a) and amputated (b) embryo littermates, unfixed at 9-5 days of
gestation. PM and S indicate, respectively, the lengths we measured when estimating
the size of the presomitic paraxial mesoderm and the somites.
194
O. P. FLINT AND OTHERS
220-
200 -
180-
I
160-
140-
E
JZ
"So
ju
mit
u
_
120-
I
i
10080-
6040-
20-
1
2
3
4
5
6
7
8 9 10 11 12 13 14 15 16 17 18
Somite
Fig. 2. Change of somite size with position in the embryonic axis in 18-somite
embryos (see Table 1). The white bars give the average value for the group of normal
embryos, the black bars for amputated. The vertical line at the head of each bar
shows one standard deviation about the mean.
at the same rate. But the amputated embryo is smaller than the normal embryo
and in order that the same proportion of the shorter embryonic axis should
emerge at the same time the rate of axis formation in the mutant must be
retarded. Streak regression must therefore be retarded. We have both in vivo
and in vitro evidence to show that amputated cells are less motile than normal
cells (Flint, 1976; Flint & Ede, 1978). We believe that it is because of this
reduced motility that streak regression, a morphogenetic movement involving
individual cells (Nicolet, 1965; Spratt, 1957), is retarded in the mutant.
Somite size may regulate for embryo size because the genetic programme in
each cell may precisely define every future parameter of the embryo: axis
length, somite number, somite size, available presomitic paraxial mesoderm
length, etc. The grounds we have for rejecting such an hypothesis of completely
<
<
<
<
<
<
<
<
0001
0001
0005
0010
0001
0050
0025
0050
No, P
No, P
No, P
No, P
No, P
No, P
No, P
No, P
>
>
>
>
>
>
>
>
0-25
0-75
0-50
0-75
0-75
0-75
0-75
0-75
No
P > 0-75
No
P > 0-75
No
P > 005
No
P > 0-25
Of slope?
A.
Yes
P < 0001
Yes
P < 0001
Yes
P < 0001
Yes
P < 0001
Of elevation?
Significant changes between
regressions
Last somite
length
Presomitic
mesoderm
length
Axis length
Mouse, normal
Mouse, amp
Mouse, normal
Mouse, amp.
Mouse, normal
Mouse, amp.
Y=
y =
y=
y=
Y=
Y=
1-70 + 0-22*
0-26 + 0-20*
113-69 + 0-99*
49-86+1-26*
1110-99 + C-10*
545-82 + 0-25*
Yes, P
Yes, P
No, P
No, P
No, P
No, P
<
<
>
>
>
>
0010
0025
010
010
010
0-25
No, P >
No, P >
No, P >
No, P >
No, P >
No, P >
0-25
0-75
0-25
0-25
010
0-25
No
P > 0-50
No
P > 0-75
No
P > 0-25
Yes
P < 0001
Yes
P < 0001
Yes
P < 0001
A similar statistical analysis to that described under Table 2 is carried out. In this case, Y is the variable parameter described in the column
headed Y, and * is always somite number (developmental stage).
Significant differences between
Significant
regressions
Does y change with deviations from
somite number ?
linear regression?
Regression equation
Of elevation?
y
Of slope?
Embryo
11
9
Yes, P
Yes, P
Yes, P
Yes, P
Yes, P
Yes, P
Yes, P
Yes, P
Table 3. Regression equations calculated for the graphs shown in Fig. 4
y = 37-99 + 6-5*
Y= 0-72 +6-88*
Y = 49-41 + 5-27*
y = 4-81 +5-71*
y = 32-61+5-95*
Y= 24-14 + 4-52*
y = 36-81 + 5-23*
Y = 13-72 + 4-51*
Normal
Amputated
Normal
Amputated
Normal
Amputated
Normal
Amputated
5
7
Regression equation
Embryo
Somite
Significant
deviations from
Does y change with * ? linear regression?
The statistical analysis asks whether there is a significant increase in somite length (Y) with somite number (*) and whether there are significant deviations of the points from a linear regression (would a curvilinear regression make a better fit?) In addition we have by analysis of
covariance compared normal and amputated regressions for each pair of somites to see whether there is any difference in somite growth rate (slope)
or somite size at any stage (elevation). There is nowhere any difference in growth rate but amputated somites are consistently smaller than normal
somites.
Table 2. Equations describing the growth of somites 5, 7, 9 and 11 in normal one? amputated
embryos. The graphs of these equations are drawn in Fig. 3
©
196
O. P. FLINT AND OTHERS
240 r
Somite 5
240 —
i
200
200
160
160
120
120
80
80
40
I
40
.£
ill
l
I
~r12
CJ)
c
o
u
Somite 7
i
i
14
i
I
16
i
I
18
i
i
20
i
i
22
i
i i
24
26
I
I I
Somite 9
"p 200
5
160
120
80
40
111
n /
I
1 2
I
I
14
I
I
16
I
I
18
I
I
20
I
I
22
I
24
III I I I I I I I I I I I I I I I
~f12
14 16 18 20 22 24 26
Somite number
26
Fig. 3. Change in size of somites 5, 7, 9 and 11 with somite number. O, Normal;
• , amputated. Table 2 gives the formulae of the regression lines and a statistical
analysis of the results.
mosaic development are those of the complexity of genetic programme required
and that such mosaic development is not characteristic of amniote development.
Given that this is the case, somite size is most probably regulated at the time
of somite formation by control mechanisms acting on the presomitic paraxial
mesenchyme, and these mechanisms are either directly or indirectly related to
the size of the embryo.
While axis length increases throughout the somite stages we have examined,
the size of the most recently formed somite remains constant (Fig. 4, Table 3).
Therefore there can be no direct relationship between axis length and somite
size. The quantity of available paraxial mesenchyme on the other hand exactly
parallels most recently formed somite size in that it too remains constant throughout the somite stages we have observed (Fig. 4, Table 3), and this suggests a more
direct relationship. There is also a comparable difference in both the size of
mutant and normal most recently formed somites, and available paraxial mesoderm lengths (somite length: normal = 133-82 ± 32-57 fixn, amputated = 74-32 ±
15-55 /.tm; parax. mesoderm length: normal = 1112-93 ± 122-33 //m, amputated
= 550-69 ± 86-56 jum). The ratio of normal to amputated in both cases is
approximately 2:1.
If these correlations are not entirely coincidental it is also interesting that the
length of presomitic paraxial mesenchyme in mutant and normal is of the same
Analysis of somitogenesis in the mouse
197
10
_ (a) Axis length
E
£ 6
¥
5
g
4
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
200
180 _ (b)
Last somite
160
^
140
|
120
S 80
J
60
40
20
J
I
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
_ (c) Presomite mesoderm length
2 4 6
8 10 12 14 16 18 20 22 24 26 28 30
Somite number
Fig. 4. Changes of axis length, last (most recently formed) somite length and
presomitic paraxial mesoderm length with somite number. O, Normal; • , amputated. The formulae for the regression lines and a statistical analysis can be found in
Table 3.
order as that proposed by Crick (1970) for a morphogenetic field whose properties might be determined by a simple diffusion gradient. The average length
of the whole embryo at 9-5 days, however, would not permit such a mechanism
(normal = 6-14 ±0-78 mm, amputated = 4-15 ±0-92 mm).
A model for the control of somite size through node regression. The argument:
It is possible to explain the generation of many types of developmental pattern,
198
O. P. FLINT AND OTHERS
100 r
(«)
100 |- (b)
g> 6-0
60
20
a b c 20
Cells
j
m
20
m m H'l
SA B
DOL 1
Cells
i
I
40
c
20
60
i
in
D
\
PM
\
)
(
i
40
M
60
80
H M M M I-ITT
CDooaoooocri PM
Fig. 5. Results of a computer simulation based on the model discussed in the text,
indicating the effect of reduced node regression rate on somite size. Graph (a) is the
simulation for the normal embryo. Three stages (a, b and c) in the development of
the standing morphogen wave are shown. Directly underneath P is a representation
of the graph as a linear strip of cells. The black dotted cells are those engaged in the
continuous destruction of morphogen (the troughs) while in between cells are
actively synthesizing (stable peaks). Two clear bands are indicated where the next
trough will appear. Beneath P, I shows how we think this pattern of morphogen
synthesis affects the appearance of somites. There is a delay between the formation
of the trough and the full formation of thefissure.Therefore though there is a set of
zones within the paraxial mesoderm corresponding to the troughs (B, C and D),
there will be a fairly long strip of unsegmented paraxial mesoderm (PM) between
the node and the first fully-formed fissure (A). We have observed in time-lapse films
zones of increased cellular activity within the PM corresponding to the troughs
and to the later formedfissures.S is the most recently formed somite. The effect of reducing node regression rate in the computer simulation on the form of the standing
wave and consequently somite size is shown in (b).
at least in principle, by appeal to the concept of 'positional information'
(Wolpert, 1969), but it is much less plausible to do so in the case of any sort
of periodic pattern and Wilby & Ede (1975, 1976) have proposed a wave-form
gradient model to account for the development of such patterns in the cartilage
skeleton of the limb-bud. Here there is no regulation, but we show that this
model may be extended to situations where it does occur, and in particular,
that it is capable of generating the patterns of somitogenesis we have described
above, including the modifications which occur in the mutant mouse.
Our work suggests the involvement of the unsegmented paraxial presomitic
mesenchyme, specifically its length, in the control of somite size at segmentation.
The model we propose and which we have simulated successfully on the computer (see Fig. 5) assumes firstly that two bands of paraxial mesoderm are
created on either side of the notochord by the passage of the regressing node
through the primary mesenchyme (Bellairs, 1963; Nicolet, 1970; Lipton &
Jacobson, 1974tf, b), and secondly that as each new group of cells is recruited
Analysis of somitogenesis in the mouse
199
into the presomitic mesenchyme band the cells start synthesis of a morphogen.
The morphogen is created faster than it can diffuse away. The first recruited
anterior cells will therefore initially accumulate most morphogen. At a certain
threshold level irreversible destruction of the morphogen is initiated and the
resultant local instability creates a trough of morphogen with apeak immediately
behind it just below the threshold level. The peak is stable because diffusion
into the morphogen destruction region never allows the adjacent peak cells
to reach threshold level. However, further away from the first trough of
morphogen destruction morphogen levels can rise in the paraxial mesoderm to
threshold level and accordingly a series of troughs and peaks are created in the
unsegmented mesenchyme in front of the node. Each trough of morphogen at
a later stage initiates fissure formation, defining the somites. The distance
between peaks depends upon the rate of node regression so that the faster the
node regresses the greater the distance between peaks, and the larger the
somites. We have already argued that node regression in amputated must be
retarded and our model therefore can explain the smaller somites.
The troughs of morphogen defining where fissures are about to appear in the
mesenchyme give a certain mosaic character to the unsegmented paraxial
mesenchyme which is confirmed by experiment. For example, when Lanot
(1971) reversed strips of chick paraxial mesenchyme in situ he found that
segmentation did not begin at the new anterior end but at the old anterior,
i.e. the new posterior end. We have also seen in time-lapse films (unpublished
results) of cultured chick material bands of increased cellular activity in the
paraxial mesenchyme which correspond to the fissures which will appear later
at the site of each band.
The computer simulation: a formal description. The basic model has four
stages of cellular activity: (1) initiation of synthesis of a morphogen, producing a (2) rise in morphogen concentration to a critical threshold level, at
which there is (3) an irreversible switch from synthesis to destruction of the
morphogen, linked to cell determination and differentiation and leading to (4)
a fall in morphogen concentration and the establishment of a sink. Since there
is free diffusion between cells at all stages the sink developed in Stage 4 maintains the morphogen concentration in neighbouring cells below the threshold
level and so allows a periodic pattern to develop. In the original limb cartilage
model (Wilby & Ede, 1975) the pattern develops across a static tissue from an
initiation region located at one edge and there is no capacity for regulation
since all the relevant interactions are either within cells or between neighbouring
cells so that there is no reasonable way in which these can be modified at long
range. But with a simple change in the pattern initiation process, making it
subject to an external control, a sufficient capacity for regulation is introduced.
This external control is, as we have proposed, the passage of the node.
A computer simulation programme based upon this model, with the appropriate modifications for somitogenesis, has been run on the IBM 375-series
200
O. P. FLINT AND OTHERS
computer at the Edinburgh Computer Centre. Because of restrictions imposed by
the complexity of the programme we have been forced to limit the simulation to
a single string of cells. This considerably reduces the flexibility of the system,
which nevertheless still illustrates the way in which somites are generated in this
model. The string of cells, representing the presomitic mesoderm, is continuously lengthened by addition of cells at one end as the node regresses, and
shortened by subtraction of cells at the other as they become incorporated in
each new somite.
Changes in morphogen concentration with time (A[M]/) and the resultant
patterns of cell differentiation are found by the stepwise application of the
following simple algorithmic equations:
(i) For \ < i < n
AM=
S2 ([M]ix(\-0-2d) + dx([M]i_1)+fi(S)xfi(D)xDKxt).
i = n—l
(ii) For the terminal cells i = 1 and i — n the diffusion terms are [M]t x
(hO-d) + dx[M]2 and [M]nx(h0-d)
+ dx [M]n_x. t = Time step, {M\ =
morphogen concentration in the ith cell, d = rate of cell to cell diffusion, S rate of morphogen synthesis, (/)) = rate of morphogen destruction, DK =
rate of spontaneous decay, ft = state function.
For a cell in the 'synthetic state' where (M)i has never exceeded the threshold
T, the value of/* (S) = S and/* (D) = 1-0. Where Thus been reached, the cell
is in the destructive state and/^ (S) = 0-0 and/ £ (D) = D. To achieve gradient
stability and minimize error accumulation a small time step is used - t = 0-1.
Cells are added stepwise to the model system, one at a time; the node regression rate R is therefore the reciprocal of the elapsed time between additions.
To date, 112 simulations have been run in which the following ranges of
variables have been used: S = 0-01 to 0-05; D = 0-7 to 0-999; d = 0-1 to 0-33;
DK = 0-996 to 10; R = 0-25 to 10.
Not all, but 36 % of the simulations showed the direct dependence of somite
length on node regression rate shown in Fig. 5 a, b, indicating that the selection
of appropriate parameters cannot be arbitrary. In 28 % the relation between
node regression rate and somite length was inverted and in the remaining 36 %
an unstable or irregular pattern was produced, ranging from the inclusion of
simple errors, e.g. a single 1/8/2 in a 2/10/2 string, to complex forms such as
2/11/1/7/1/7/2 and 2/5/1/2/1/3/2/2, which represent imbalances between
various components of the simulations. The imbalances may be paralleled in
reality in various mouse mutants such as pudgy (Griineberg, 1961) and Crookedtail (Matter, 1957) in which somites are of variable size and intersomitic
fissures are irregular and ill-defined.
A serious restriction of the simulation, though not of the model, is that its
one-dimensional quantized character limits fissure width to two alternatives
(one cell and two cells) despite a range of somite lengths from 2 to 17 cells, so
Analysis of somitogenesis in the mouse
201
that the regulative ability of the model is limited to two basic states- 1-cellwide fissure/short somite and 2-cell-wide fissure/larger somite. Consequently, if
all variables except node regression rate are kept constant a reduction of somite
size is not produced until node regression rate is reduced by half. Though this
is, in fact, the magnitude of change in our mutant and operated embryos, the
biological situation will obviously be more complex and if our computer
simulation had been 3-dimensional, with the greatly increased cell interactions
which that would give, regulation would be continuous as it appears to be in
the real embryo.
Control of somite number. As to a mechanism for achieving the normal
species-specific number of somites, for which some sort of global control must
be postulated, we have presented no suggestions. The number does follow,
given appropriate modification of the rate of node regression with varying
axis length. A control of node regression rate is required; we believe this may
involve a modification of cell behaviour, such as would certainly account for
the reduced node regression rate observed in amputated. A possible global
control does emerge from the clock and wave front model proposed by Cooke
& Zeeman (1976) for somite formation in Xenopus. This model requires that
axis length be laid down prior to somite formation, but in the mouse, as in
the chick, the primitive streak is still regressing and the embryonic axis still
forming while the first somites are being produced. A modification of the CookeZeeman model may be conceived which would fit the experimental results
reported here, but more general and experimentally unsupported assumptions
are required of the castrophe theory type of model than for the one which we
present.
The wave-form gradient can therefore successfully model somitogenesis in
its essential aspects, notably the definition of somite boundaries and the development of intersomitic fissures, including the capacity for regulation in adjusting
somite number to overall axial length. More data is required to put real values
to the model variables, and it is, of course, one of the chief functions of such
a model to stimulate the necessary investigations. Our observations on mouse
embryos have shown that node regression plays a key role in controlling somite
size and number, and our simulations have suggested how this may come about,
but of course the problem of how node regression rate itself is first specified
and then regulated remains; in this connexion a detailed analysis of node
regression and factors affecting it is clearly called for.
We wish to thank the SRC and ARC forfinancialsupport.
REFERENCES
R. (1963). The development of somites in the chick embryo. /. Embryol. exp.
Morph. 11, 697-714.
COOKE, J. (1975). Control of somite number during morphogenesis of a vertebrate, Xenopus
laevis. Nature, Lond. 254, 196-199.
BELLAIRS,
202
O. P. FLINT AND OTHERS
COOKE, J.
& ZEEMAN, E. C. (1976). A clock and wavefront model for control of the number of
repeated structures during animal morphogenesis. /. theor. Biol. 58, 455-476.
CRICK, F. (1970). Diffusion in embryogenesis. Nature, Lond. 225, 420-422.
FLINT, O. P. (1976). Morphogenesis and cell behaviour in the development of the axial skeleton
of normal and mutant (amputated) mouse embryos. Ph.D. thesis, Glasgow University.
FLINT, O. P. (1977). Cell interactions in the developing axial skeleton in normal and mutant
mouse embryos. Third Symposium of the British Society for Developmental Biology.
Cambridge University Press. (In the Press.)
FLINT, O. P. & EDE, D. A. (1978). Cell interactions in the developing somite: In vivo
comparisons between amputated {am/am) and normal mouse embryos. /. CellSci. 31, 275291.
GRUNEBERG, H. (1956). A ventral ectodermal ridge of the tail in mouse embryos. Nature,
Lond. Ill, 787-788.
GRUNEBERG, H. (1961). Genetical studies on the skeleton of the mouse Pudgy. Genet. Res. 2,
384-393.
HAMILTON, L. (1969). The formation of somites in Xenopus. J. Embryol. exp. Morph. 22,
253-264.
LANOT, R. (1971). La formation des somites chez l'embryon d'oiseau: etude experimentale.
J. Embryol. exp. Morph. 26, 1-20.
LIPTON, B. H. & JACOBSON, A. G. (1974a). Analysis of normal somite development. Devi
Biol. 38, 73-90.
LIPTON, B. H. & JACOBSON, A. G. (19746). Experimental analysis of the mechanisms of
somite morphogenesis. Devi Biol. 38, 91-103.
MATTER, H. (1957). Die formale Genese einer vererbten wirbelsaulen Missbildung am
Beispiel der mutante crooked-tail der Maus. Rev. Suisse Zool. 64, 1-38.
MEREDITH, R. (1965). Genetic and embryological studies of some congenital malformations in
the mouse. A thesis submitted to the Institute of Biology for non-graduate admission to
Full Membership.
NICOLET, G. (1965). Etude autoradiographique de la destination des cellules invaginees
au niveau du nceud de Hensen de la ligne primitive achevee de l'embryon de Poulet.
Acta Embryologiae et Morphologiae Experimentalis 8, 213-220.
NICOLET, G. (1970). Is the presumptive notochord responsible for somite genesis in the chick.
/. Embryol. exp. Morph. 24, 467^78.
SNEDECOR, G. W. & COCHRAN, W. G. (1967). Statistical Methods, 6th ed. Ames, Iowa:
Iowa State University Press.
SOKAL, R. R. & ROHLF, F. J. (1969). Biometry. The Principles and Practice of Statistics in
Biological Research. San Francisco: W. H. Freeman.
SPRATT, N. T. (1957). Analysis of the organiser centre in the early chick embryo. II. Studies
of the mechanics of notochord elongation and somite formation. /. exp. Zool. 134, 577-612.
WILBY, O. K. & EDE, D. A. (1975). A model generating the pattern of cartilage skeletal
elements in the embryonic chick limb. /. theor. Biol. 52, 199-217.
WILBY, O. K. & EDE, D. A. (1976). Computer simulation of vertebrate limb development,
the effect of cell division control patterns. In Automata, Languages, Development (ed.
A. Lindenmeyer & G. Rozenburg).
WOLPERT, L. (1969). Positional information and the spatial pattern of cellular differentiation. /. theor. Biol. 25, 1-47.
(Received 31 October 1977, revised 19 December 1977)