J. Embryol. exp. Morph. 76, 283-296 (1983)
Printed in Great Britain © The Company of Biologists Limited 1983
283
Development of the lateral line system in Xenopus
laevis.
II. Cell multiplication and organ formation in the supraorbital
system
By RUDOLF WINKLBAUER AND PETER HAUSEN 1
From the Max-Planck-Institut fur Virusforschung, Tubingen
SUMMARY
Cell multiplication was studied during development of the supraorbital lateral line system
in Xenopus laevis. The increase in cell number is biphasic. Thefirstphase extends from the
beginning of primordial elongation to the end of primary organ formation. Cell number
increases linearly during this interval. Throughout this phase, a constant number of cells is in
S phase of the cell cycle at a given time, despite a more than 10-fold increase in total cell
number. After their formation, the number of the primary organs remains essentially constant. The individual primary organs are not clones of cells. Different organs grow at different
rates, and become more and more heterogeneous in size. The second phase which is correlated
with accessory organ formation is characterized by an elevated growth rate. This phase was
not studied in detail. If developing larvae are starved, growth is normal up to completion of
the first growth phase but is arrested at this point. The frequency distribution of the sizes of
such growth-arrested organs approximates a binominal distribution. From its characteristics,
a detailed model of cell proliferation and organ formation can be deduced: cell multiplication
occurs through asymmetrically dividing stem cells, which become allocated to the forming
organs at random and go through a fixed number of cell divisions.
INTRODUCTION
In the previous communication, the gross morphological changes during
supraorbital (SO) lateral line development were described. Evidence obtained
from these observations allowed the discussion of questions concerning cell
movements during this morphogenetic process. In the present article we extend
these studies to the kinetics of cell multiplication and its relation to the different
morphogenetic events during the development of the SO lateral line system.
Changes in cell and organ number during SO lateral line development are quantified; it is asked whether primary organs are cell clones, and how the obvious
heterogeneity in organ size within a SO line (Stone, 1933; accompanying communication) changes with time. The frequency distribution of the sizes of primary organs is analysed in larvae, whose growth was arrested at a defined stage
in development by starvation. The peculiar organ size frequency distribution of
1
Author's address: Max-Planck-Institut fur Virusforschung, Abt. fur Zellbiologie (V),
Spemannstrasse 35, D-7400 Tubingen, Federal Republic of Germany.
284
R. WINKLBAUER AND P. HAUSEN
such animals allows us to draw detailed conclusions on the mode of cell proliferation and its interplay with organ formation.
MATERIALS AND METHODS
X. laevis and X. borealis embryos and larvae were obtained, kept and staged,
skin preparations were made and transplantation experiments and autoradiography were carried out as described (Winklbauer & Hausen, 1983).
Haploid-triploid chimaerae were produced according to the method of Jacobson
& Hirose (1978). X. laevis females were injected with 500 units of human
chorionic gonadotropin. About 12 h later, eggs were stripped from such females
and immediately fertilized with fresh macerated testes in 1/10 MBS-H. 20min
after fertilization, eggs were heated to a temperature of 35-5 °C for 6min and
then rapidly cooled to 20 °C. Shortly before closure of the neural tube, haploidtriploid embryos were identified by the left-right asymmetry of the neural folds
and by the difference in cell size between left and right lateral ectoderm.
RESULTS
Clonal vs. multicellular origin of primary organs
When X. laevis eggs are exposed to elevated temperature 20 min after fertilization, some of the eggs develop as haploid-triploid chimaerae (Jacobson &
Hirose, 1978). The haploid and triploid cells can be distinguished by their different nuclear volume (Fankhauser, 1945). In such chimaerae the epidermis
often contains regions where haploid and triploid cells are intermingled
(Fischberg, 1949). If a lateral line primordium arises in such a region and
develops into primary organs, it is frequently found that haploid and triploid cells
occur together within a single organ. From this observation, a clonal origin of the
primordium as well as of the individual primary organs can be excluded. Fig. 1
shows examples of such mixed organs from the SO lateral line system, as compared to a pure haploid and a pure triploid organ.
In another set of experiments, X. borealis epidermis was inserted homotypically into X. laevis embryos at stage 22, as described (Winklbauer & Hausen,
1983), in such a way that the border of the transplant crossed the prospective ear
region and a mixed placode could develop (the exact positioning of the implant
became obvious only in the later analysis). After development to stage 47 the
primary organs often consisted of a mixture of X. laevis and X. borealis cells
(Fig. 2), as recognized by the X. borealis nuclear marker (see accompanying
communication). Thus both types of experiment indicate the polyclonal origin
of the whole supraorbital lateral line primordium and of each individual primary
organ.
Cell multiplication in lateral line system of Xenopus
1A
285
B
•v^'.
*D
Fig. 1. SO lateral line organs of haploid-triploid chimaerae. Haploid (A) and
triploid (B) epidermal and lateral line organ cells can be distinguished in DAPIstained skin preparations (stage 47£) due to the difference in nuclear size. In
chimaerae where the ear region of the epidermis (origin of the lateral line placodes)
consists of a mixture of haploid and triploid cells (a condition which is found very
frequently), lateral line organs can be found which contain both small haploid and
large triploid cells (C, D). Bar = 50 fxm.
Fig. 2. SO lateral line organs of X. laevis-X. borealis chimaerae. If the border of a
X. borealis epidermal transplant crosses the ear region at the time of lateral line
placode formation, organs surrounded by X. laevis epidermal cells may be found at
stage 47 which contain both X. borealis (hatched) and X. laevis cells. X. borealis cells
were identified in DAPI stained skin preparations through the nuclear marker
described in the accompanying communication. Four experimental animals contained a total of 16 mixed SO organs.
The kinetics of cell multiplication in the SO lateral line system
Skin preparations of carefully staged embryos and larvae were used to determine changes in cell number within the SO lateral line system between stages
33/34 and 49. Fig. 3A (closed circles) shows a continuous but biphasic increase
in cell number throughout the period examined.
From stage 33/34 to stage 47|, cell number increases from about 50 cells to
about 580. This growth is not exponential, but linear with a rate of 4-6 cells/h.
After stage 47| the rate of cell multiplication increases significantly to 7-3
EMB76
286
R. WINKLBAUER AND P. HAUSEN
1800
T
1600 ••
1400 - -
1200 - -
1000 •"
5
6
7
11
12
Days after fertilization
Fig. 3. Kinetics of cell multiplication in the SO lateral line system. (A) Skin preparations of carefully staged (arrows mark the stages) embryos and larvae were used to
determine the cell number of SO primordia and lateral lines. Feeding larvae exhibit
a biphasic growth curve (closed circles). Regression analysis yields a slope of 4-6
cells/h between stages 33/34 and 47 (correlation coefficient r = 0-99) and of 7-3
cells/h between stages 48 and 49 (correlation coefficient r = 0-90). The difference
between the slopes is statistically significant (significance level a = 0-01). Starving
animals reach a plateau after stage 47 (open circles). Regression analysis gives a
slightly negative slope of -0-27 cells/h (correlation coefficient r = -0-13) between
days 7 and 12 (dashed line), which is, however, not significantly different (significance level a = 0-05) from zero (solid line). A total of 128 skin preparations was
evaluated. Each point represents the mean value of five to six (closed circles) or
about eight (open circles) measurements, respectively. Standard deviations are indicated by bars. (B) Number of DNA synthesizing cells in the SO system at different
stages of development. After a 30min [3H]thymidine pulse and subsequent autoradiography, the number of labelled cells per SO system was determined. Each point
represents a single measurement. The slope of the straight line is —0-004 cells/h
(correlation coefficient r = -0-04).
Cell multiplication in lateral line system of Xenopus
287
cells/h. Due to cell crowding in the budding organs an accurate determination
of cell number becomes increasingly difficult at stages later than 49. Therefore
the exact linearity during this second phase cannot be safely concluded.
Larvae of X. laevis usually begin to feed at about 4 days of development
(Nieuwkoop & Faber, 1967), but are able to survive without food uptake for at
least 20 days. Larvae kept under starvation conditions develop normally at first,
but between stage 47 and 48 development ceases, as judged from the external
criteria given in the Normal Table of Nieuwkoop & Faber (1967).
This arrest is also obvious in the cell multiplication curve of the SO lateral line
system of starved animals (open circles in Fig. 3A). At stage 47, no difference
in cell number between feeding and starving larvae is seen, but later, between
days 1\ and 12, cell multiplication is arrested in starved animals. It seems as if in
these animals the SO system cannot enter into the second growth phase normally
characterized by a raised growth rate, but instead reaches a plateau at the level
of the transition point between the two phases of normally growing tadpoles. The
cell number remains constant at 579 ± 77 cells.
DNA synthesizing cells in the SO system
To label DNA synthesizing cell nuclei in the SO system, [3H]thymidine was administered to larvae of different ages for 30 min before fixation. Skin preparations
of the animals were processed for autoradiography as described (Winklbauer &
Hausen, 1983). In Fig. 3B the number of labelled cells found in individual SO
systems is plotted against developmental time. Obviously, the number of cells in
the SO system that are in S-phase at a given time does not increase between
stages 35/36 and 48, although the total cell number increases more than tenfold.
On the average, 29-6 ± 5-4 labelled cells are found within a single SO lateral line
system irrespective of its developmental state.
The number of SO organs
The total number of organs produced by one primordium remains constant at
an average of 16-4 ± 1-5 between stage 43, when individual organs first become
reliably discernible, and stage 47. It increases to 17-1 ± 1-5 between stages 47 and
48 due to the occasional addition of an organ to the anterior end of the SO line
(Winklbauer & Hausen, 1983). The number of parietal organs remains constant
at 3-0 ± 0-8 (data from 77 preparations). It becomes difficult to count accurately
the number of primary organs in larvae older than 10 days, since now the formation of accessory organs obscures the situation, especially in the parietal line and
the anterior part of the SO line.
Divergence of organ and plaque size
Even superficial inspection of the lateral line system at more advanced stages
reveals a striking heterogeneity in the size of the primary organs, or of the
plaques derived from them (see accompanying paper, Fig. 3 and Fig. 5). The
288
R. WINKLBAUER AND P. HAUSEN
extent of this heterogeneity in the SO system of feeding larvae, measured as the
difference in cell number between the smallest and the largest organ or plaque
ever found in skin preparations of a given stage, changes with time (Fig. 4).
From stage 43, when individual organs are first reliably discernible, up to stage
47J, the size of the smallest organs remains constant at 6-7 cells/organ. The size
of the largest organs, on the other hand, increases considerably from 26 to 65
cells/organ. After stage 47|, the smallest organs increase in size, first slowly,
then more pronouncedly, reaching a size of 76 cells/organ at stage 54. The
largest organs grow still faster, however, at least up to stage 52, forming plaques
by the process of accessory organ formation and thus extending further the range
of plaque sizes. The largest plaque ever found contained nearly 300 cells. Obviously, organs within a single lateral line grow at strikingly different rates,
leading to an increasing heterogeneity of organ sizes.
Frequency distribution of organ sizes after growth arrest
As mentioned above, growth of the SO system is arrested in starved larvae
after stage 47J. Fig. 5A shows the frequency distribution of the sizes of different
organs from such animals.
The organ size ranges from 6 to 65 cells per organ. The frequency of organ sizes
300 T
1
50
4
250-•
/
r
49
I
2 0 0 ••
0>
150"
•3
48
I
47i
4
80"
60-
43
40-
I1J.U
20-
4
a
6
54
r
52
H
f-
10 12 14 16 18
Days after fertilization
20
H
h
22
24
26
Fig. 4. Divergence of organ size. The range of organ sizes at a given stage is
represented by a bar. The end points of each bar are determined by the largest and
the smallest SO organ ever found in a skin preparation of the respective stage. The
organ size range at stage \1\ comprises the values for starved animalsfixedafter day
7 of development, since the cell number of such larvae reaches a plateau in the SO
system midway between stages 47 and 48.
Cell multiplication in lateral line system of Xenopus
289
forms a series of distinct peaks rather than a smooth distribution. The cell
numbers of adjacent peaks always differ by about 7 cells. If we assume that the
peaks do in fact occur at fixed cell numbers separated by a fixed interval, then
a best fit to the experimental data is obtained when that interval between two
neighbouring peaks is 7 cells and the first of a total of nine peaks is at an organ
size of 8 cells. If the nine peaks are superimposed according to this frame to
equalize random irregularities, a narrow, symmetrical peak is in fact obtained
(Fig. 5B). The empirical distribution approximates reasonably closely to a
slightly asymmetrical binominal distribution (n = 8 and P = 0-43) (Fig. 5C).
These unexpected numerical results turned out to be the key to the understanding of the mode of growth and its relation to organ segregation in the SO
system.
30--
15 22293643 50 57 64
Organ size class
20--
10--
29
36
43
Organ size (cells/organ)
50
57
64
Fig. 5. Frequency distribution of organ sizes. (A) The cell number per organ from
477 SO organs was determined from DAPI-stained skin preparations. Growtharrested larvae (starvation; plateau in Fig. 3) from five different matings were used.
(B) Superposition of peaks. It is assumed that the first peak should be at 8 cells per
organ and that the difference between neighbouring peaks is 7 cells per organ. Organ
numbers at 8, 15, 22 etc. cells per organ of (A) are summed and give the organ
number at zero in (B). Organ numbers at 9,16,23 etc. cells per organ are added and
give the organ number at +1 in (B), etc. This superposition yields a symmetrical
peak. (C) Frequency of organ size classes. According to the frame indicated in (B),
organ size classes can be defined, each ranging over an organ size difference of 7 cells
per organ, with 8,15,22 etc. cells per organ as the medians of the respective classes.
The frequency of the organ size classes is indicated by bars. This empirical
distribution approximates a binominal distribution with number of trials n = 8 and
probability of success P = 0-43 (solid line).
290
R. WINKLBAUER AND P. HAUSEN
DISCUSSION
A model of cell multiplication and organ segregation in the SO lateral line system
Any attempt to integrate the data provided by our observations into a unifying
hypothesis has to take into account that the peak cell numbers in the frequency
distribution of organ sizes are not multiples of a unit; that the binominal
distribution indicates a stochastic process during development of the system; and
that the total cell number of the SO system increases at a constant rate. The
model described below integrates these preconditions and enables us to explain
numerous independent observations which in turn lend support to the model. We
have been unable to devise any other model which describes our observations
with sufficient accuracy and simplicity, but we acknowledge that we can on
principle not exclude all other possibilities.
One may consider a mixed population composed of 'stem cells', able to divide,
and 'terminal cells', unable to divide. A stem cell shall produce 7 terminal cells
by successive unequal cell divisions and then become terminal itself. Such a
population is divided into groups of 8 cells, and stem cells and terminal cells are
randomly allocated to these groups. After completion of growth a frequency
distribution of the final sizes of these groups will be obtained which is well
approximated by a binominal distribution with n = 8 and with a size difference
of 7 cells between two neighbouring size classes (if the original cell population
is large). If the cell cycle time remains constant, the increase in cell number will
be linear.
If applied to the SO system, the above assumptions can explain the frequency
distribution of organ sizes of Fig. 5 and the linear cell kinetics of Fig. 3. Based
on these considerations, calculations are possible concerning some important
parameters of the system.
The number of cells at the time of organ segregation can be calculated as the
number of organs formed in the SO system, 16-4, multiplied by 8, the number
of cells originally contained in a nascent organ. From this value of 131 cells, one
can determine from the kinetics of Fig. 3 A the time of organ segregation, which
is 61 h of development (stage 394).
Between organ segregation and growth arrest, 579 minus 131 cells must be
produced. Since each stem cell supplies 7 cells, the number of stem cells must be
64 to account for the increase in cell number.
Since growth is linear with a rate of 4-6 cells/h, the duration of the cell cycle,
13-9 h, can be calculated by dividing the number of stem cells by the rate of
growth. Together with the fact that on the average 29-6 cells in the SO system
are in S-phase of the cycle, the duration of the S-phase can be calculated to be
6-4h (provided that the terminal cells are arrested in Gl).
On the basis of the above assumptions and calculations which use only data
from the frequency distribution of organ sizes, from the cell kinetics, and the
Cell multiplication in lateral line system o/Xenopus
291
Hours after
fertilization
40-•
60 •
37/38
80--
394
100- •
43
120-•
140- •
474
Tl T2 T3 T4 T5 T6 T7 T8 S
160-
Fig. 6. Model of growth and cell multiplication in the SO system. Stem cells (S,
closed circles) divide to produce in succession 8 terminal cells (Tl, T2 etc., open
circles). Random allocation of stem cells and terminal cells to forming organs at stage
394 leads to differences in organ size.
number of organs formed, a detailed model of growth and organ segregation in
the SO lateral line system up to stage 47| can be designed (Fig. 6).
In the considerations made above, the origin of the stem cells and terminal
292
R. WINKLBAUER AND P. HAUSEN
cells present at stage 39| is irrelevant. As recruitment of cells from the epidermis
does not occur, new cells must be derived from preexisting primordial cells
(Winklbauer & Hausen, 1983). The simplest assumption is that at the beginning
of primordial elongation between stages 33/34 and 35/36 (Winklbauer &
Hausen, 1983), only stem cells are present which produce, during elongation of
the primordium, terminal cells by the same mechanism of asymmetrical cell
division, which is thought to operate after stage 39£ (Fig. 6). About 64 stem cells
are required according to our calculations. In fact this cell number is present in
the primordium shortly after stage 33/34 (Fig. 3). The calculated cell cycle
parameters of these stem cells seem reasonable: 13-9 h cell cycle duration, 6-4 h
duration of S-phase.
After one round of cell division, when stem cells and terminal cells are about
equal in number, organ segregation commences. The exact probability for a cell
to be one of the 64 stem cells in the population of 131 cells at stage 39i is
p = 0-488. This means that the frequency distribution of organ sizes should be
fairly symmetrical with only a slight skewness favouring smaller organs. A
binominal distribution which fits best the experimental data has a P = 0-433.
Organ segregation is calculated to occur at stage 39£. This fits very well to the
description of morphological changes in the SO lateral line system given in the
accompanying communication. First signs of organ formation were noticed at
stage 39£, and at stage 40 overt fragmentation of the primordium was seen to be
well under way. Most of the increase in cell number in the SO system occurs at
a time when the primary organs are already segregated (Fig. 3 and Fig. 6). Since
stem cells are apportioned at random to the segregating organs, which are initially of equal size (about 8 cells), these organs grow at rather different rates, due
to their differences in stem cell content, and become more and more
heterogeneous in size. The divergence in organ size shown in Fig. 4 is thus
explained. The smallest organs, those which contain no stem cells at all, remain
constant in size during the whole time period covered by the model, i.e. up to
stage 47i Organs in which all the 8 initial cells are stem cells grow fastest.
Extrapolation of the straight lines of Fig. 4, which mark the size of the smallest
organs and the size of the largest organs, respectively, at different stages, yields
that at stage 39|, all organs must be of a size between 6 and 10 cells.
The primary organs become separated from each other only gradually. Three
days elapse between stage 39! and stage 47, until virtually all organs are
separated (previous communication). On the other hand, to give the distinct
frequency distribution of organ sizes shown in Fig. 5, complete segregation of all
organs has to occur at stage 39i. Otherwise the discontinuity of the distribution
would be obscured. This means that organ anlagen do not exchange cells, even
if they are still contiguous and no borderline between the forming organs is
visible.
An obvious implication of the model, as depicted in Fig. 6, is that an organ is
not a cell clone. As stated repeatedly, cells shall be allocated to organs in a
Cell multiplication in lateral line system of Xenopus
293
random fashion, irrespective of their descent. Direct evidence for the non-clonal
status of the primary SO organs was obtained (Figs 1 and 2). The occasional
occurrence of only a single X. borealis cell in an organ where all other cells are
of X. laevis origin (Fig. 2) supports the notion, that there may be cells in an organ
which do not divide after stage 39£.
When growth is arrested at stage 47£, each stem cell has given off by
asymmetrical cell divisions 8 terminal, non-dividing daughter cells, with 7 of
them produced after the allocation of the respective stem cell to a primary organ.
With an initial organ size of 8 cells, only nine organ size classes should be possible
after completion of growth, namely 8 cells (no stem cell), 8 + 1 x 7 cells (one
stem cell), 8 + 2 x 7 cells (two stem cells) and so on until 8 + 8 x 7 cells. The
empirical organ size distribution shows indeed these nine peaks at the expected
positions, but the peaks are of a certain broadness. Three possible sources of
inaccuracy may contribute to the broadening of the peaks.
From independent recounting of the same specimens we know that counting
errors are small, but not negligible. The peak with the smallest organs ranges
from 6 to 11 cells. Since no cell proliferation takes place in these organs (Fig. 4),
and since the observed variance exceeds the one expected from counting errors
alone, inaccuracies in grouping exactly eight cells for the formation of an organ
must occur. Inaccuracies in the proliferation schedule can also contribute to the
broadening of the peaks. Two kinds of errors must be considered here. First, a
stem cell may erroneously divide seven or nine times instead of eight times.
Second, preliminary results suggest that organ segregation is somewhat delayed
in the more anterior part of the primordium. A few stem cells may become
included into anterior organs which have already given off their second terminal
cell, thus being capable of producing only six further daughter cells in the organ.
Since the smallest, non-proliferating organs already exhibit a considerable
variance, inaccuracies in the proliferation schedule may be small.
The latter notion leads to another consideration. If the timing of organ
segregation along the primordium deviates too much from simultaneousness, the
frequency distribution pattern of organ sizes would become too much obscured
and extra peaks could appear. Since this is not found, one must conclude that a
delay in organ segregation between different parts of the primordium must be
short relative to the cell cycle duration of 13-9 h.
Growth in relation to lateral line development
In the present study of growth in the SO lateral line system of X. laevis, two
growth phases could be separated. The first extends from the beginning of
primordial elongation at about stage 33/34 to stage 47i Primary organ formation which occurs during this interval is not marked by any detectable changes
in the growth pattern. This pattern changes, however, at stage 47i The rate of
growth increases, the smallest organs, arrested in size up to this point, resume
cell multiplication. Growth becomes arrested at this stage if larvae are starved.
294
R. WINKLBAUER AND P. HAUSEN
An essential feature of our model (Fig. 6) is that stem cells produce a defined
and fixed number of terminal cells until stage 47?. Whether they are programmed
for a certain number of cell divisions, or whether they cease dividing at this time
due to the effect of starvation cannot be decided safely. The stem cells do not
divide synchronously. With a cell cycle length of 13-9 h, and if cell division starts
at the same stage of development in all embryos, such a synchrony would have
been detected in the kinetics of Fig. 3 between stages 33/34 and 43, since samples
were taken close enough in that time interval. Furthermore, the pulse-labelling
experiments provided no evidence for synchronous divisions.. Finally, a
frequency distribution of organ sizes from an earlier stage, e.g. stage 45, does not
show the regular peaks to be expected when synchronous division is assumed
(not shown). Since such regular peaks are apparent after growth arrest, and since
growth is arrested just at the transition between two growth phases, this arrest
is probably not an arbitrary halt, but stem cells are allowed to finish their
program, and only entrance into the second phase of growth is prevented.
This second phase of accelerated growth is possibly related to the process of
accessory organ formation, which becomes visible after stage 48. Which mode
of cell proliferation occurs in this phase remains to be elucidated. Reactivation
of 'terminal' cells may be involved. In favour of this supposition is the observation that the smallest organs, which do not grow before stage 47£ and hence are
thought to contain no stem cells, increase gradually in size during the second
phase of growth.
Functional organs are made up of receptor cells and supporting cells. According to Stone (1933), these two cell types are already present in primary organs
in Ambystoma. No data are available on their first differentiation in Xenopus.
From the present results, no conclusions can be drawn as to the origin of these
two cell types, since receptor cells could not be reliably identified with our
method.
Stem cells and the asymmetry of cell division
The concept of stem cells was originally developed for adult self-renewing
tissues, e.g. germ cells in the testis, blood cells, intestinal epithelial cells. Stem
cells are, by definition, cells which have the ability to divide throughout the life
of an organism so as to produce more stem cells and to produce differentiated
cells to counter the loss of differentiated cells from these adult tissues (e.g.
Leblond, Clermont & Nadler, 1967; Papaioannou, Rossant & Gardner, 1978;
Nothiger, Schiipbach, Szabad & Wieschaus, 1978; Lajtha, 1979).
In development, expanding cell populations, which account for the growth of
tissues and organs, are of fundamental importance. Some of these expanding
populations may become built up by cells which divide so as to produce more
cells of their own kind and to produce differentiating cells. Proliferation of such
cells may stop when the respective tissue has reached its final size and becomes
static (e.g. nervous system). It is convenient and justifiable to call these cells
Cell multiplication in lateral line system of Xenopus
295
'stem cells', too, even if they do not persist throughout the lifetime of their host
organism. Since their pattern of proliferation and differentiation is identical to
adult stem cells, they behave according to a 'stem cell logic' (Papaioannou et al.
1978; Prothero, 1980; Sulston & Horvitz, 1977). It is in this extended sense that
we use the expression 'stem cell' in the present communication.
The concept of a stem cell does not imply necessarily a fixed mode of cell
division. One can imagine two possibilities how stem cell behaviour of a cell
population and the mode of division of individual cells are connected.
Asymmetric cell division - one daughter cell becoming differentiated and the
other one being able to divide again asymmetrically like the mother cell - results
in a stem cell population in which the relative amount of cells destined to differentiate and that destined to remain in the stem cell compartment is controlled
intrinsically. Examples for such a mechanism can be found in nematode development (Sulston & Horvitz, 1977), Drosophila oocyte (Nothiger et al. 1978) and
ganglion cell production (Seecoff, Donady & Teplitz, 1973) and in other invertebrate systems.
Alternatively a stem cell population may be constituted of symmetrically dividing cells, both daughter cells inheriting the stem cell potential. Some of the cells
become determined to leave the stem cell compartment, either by an essentially
probabilistic process (e.g. Till, McCulloch & Siminovitch, 1969) or, perhaps
more important, by external control signals. The mammalian haemopoietic stem
cell system may turn out to be an example for the latter mechanism (Schofield,
1978).
The SO lateral line system of X. laevis represents a case of strong evidence for
the existence of an asymmetrically dividing stem cell in a vertebrate. Detection
was facilitated here by the combination of stem cell growth with a process, organ
formation, which involves only small numbers of cells. Likewise, other known
examples of asymmetric stem cell division, from invertebrates, usually show this
feature of limited cell numbers. The question is whether asymmetric stem cell
division is simply more readily detected when the cell number of the system is
small, as it often is in invertebrates, or whether accidental depletion of the stem
cell compartment in such small populations is best prevented by applying the
strict intrinsic control mechanism of unequal cell division.
We thank Bernd Strebel for his participation in the experimental work, Gunilla Weinraub,
Roswitha Gromke and Susanne Haase for preparing the figures and Aiko Tanaka-Ingold for
the careful typing of the manuscript. We are indebted to Donald Newgreen for many suggestions to improve the manuscript.
REFERENCES
G. (1945). The effects of changes in chromosome number on amphibian
development. Q. Rev. Biol. 20, 20-78.
FISCHBERG, M. (1949). Experimentelle Auslosung von Heteroploidie durch Kaltebehandlung
der Eier von Triton alpestris aus verschiedenen Populationen. Genetica 24, 213-329.
FANKHAUSER,
296
R. WINKLBAUER AND P. HAUSEN
R. G. (1904). Experimented Untersuchung iiber die Entwicklung der Sinnesorgane der Seitenlinie bei den Amphibien. Arch mikrosk. Anat. EntwMech 63, 35-149.
JACOBSON, M. & HIROSE, G. (1978). Origin of the retina from both sides of the embryonic
brain: A contribution to the problem of crossing at the optic chiasma. Science 202,637-639.
LAJTHA, L. G. (1979). Stem cell concepts. Differentiation 14, 23-34.
LEBLOND, C. P., CLERMONT, Y. & NADLER, N. J. (1967). The pattern of stem cell renewal in
three epithelia (esophagus, intestine and testis). P. Can. Cancer Res. C. 7, 3-30.
NIEUWKOOP, P. D. & FABER, J. (1967). Normal Table o/Xenopus laevis (Daudin). Amsterdam: North Holland Publishing Company.
NOTHIGER, R., SCHUPBACH, T., SZABAD, J. & WIESCHAUS, E. (1978). Stem cells and tissue
homeostasis in insect development. In Stem Cells and Tissue Homeostasis (ed. B.I. Lord,
C. S. Potten & R. J. Cole), pp. 71-85. Cambridge University Press.
PAPAIOANNOU, V. E., ROSSANT, J. & GARDNER, R. L. (1978). Stem cells in early mammalian
development. In Stem Cells and Tissue Homeostasis (ed. B. I. Lord, C. S. Potten & R. J.
Cole), pp. 49-69. Cambridge University Press.
PROTHERO, J. (1980). Control of stem cell proliferation: A density-dependent commitment
model. /. theor. Biol. 84, 725-736.
SCHOFIELD, R. (1978). The relationship between the spleen colony-forming cell and the
haemopoietic stem cell. A hypothesis. Blood Cells 4, 7-25.
SEECOFF, R. L., DONADY, J. J. & TEPLITZ, R. L. (1973). Differentiation of Drosophila
neuroblasts to form ganglion-like clusters of neurons in vitro. Cell Differ. 2, 143-149.
STONE, L. S. (1933). The development of lateral-line sense organs in amphibians observed in
living and vital-stained preparations. /. comp. Neurol. 57, 507-540.
SULSTON, J. E. & HORVTTZ, H. R. (1977). Post-embryonic cell lineages of the nematode,
Caenorhabditis elegans. Devi Biol. 56, 110-156.
TILL, J. E., MCCULLOCH, E. A. & SIMINOVITCH, L. (1964). A stochastic model of stem cell
proliferation, based on the growth of spleen colony-forming cells. Proc. natn. Acad. Sci.,
U.S.A. 51, 29-36.
WINKLBAUER, R. & HAUSEN, P. (1983). Development of the lateral line system in Xenopus
laevis. I. Normal development and cell movement in the supraorbital system. /. Embryol.
exp. Morph. (accompanying communication).
HARRISON,
{Accepted 22 March 1983)