5309
Development 126, 5309-5317 (1999)
Printed in Great Britain © The Company of Biologists Limited 1999
DEV1436
Single cells can sense their position in a morphogen gradient
J. B. Gurdon*, H. Standley, S. Dyson, K. Butler, T. Langon, K. Ryan, F. Stennard, K. Shimizu and A. Zorn
Wellcome CRC Institute, Tennis Court Road, Cambridge CB2 1QR and Department of Zoology, University of Cambridge, UK
*Author for correspondence (e-mail: jbg1000@hermes.cam.ac.uk)
Accepted 9 September; published on WWW 9 November 1999
SUMMARY
Xenopus blastula cells show a morphogen-like response to
activin by expressing different genes according to the
concentration of activin to which they are exposed. To
understand how cells recognize their position in a
concentration gradient, it is essential to know whether each
cell responds individually to activin concentration. An
alternative idea, proposed by previous work, is that cells
need to interact with their neighbours to generate a
concentration-related response. To distinguish between
these ideas, we have cultured blastula cells under
conditions which provide different degrees of contact with
other cells, allowing nil to maximum communication with
their neighbours. The cultures include cells attached to
fibronectin and cells resting unattached on an agarose base.
The cultures also include cells that have no contact with any
cell except their clonal progeny, cells that have lateral
contact to neighbouring cells, and cells that are completely
enveloped by other cells in a reaggregate. We have used
RNase protection and in situ hybridization to assay the
expression of the activin-responsive Xenopus genes Xbra,
Xgsc, Xeomes, Xapod, Xchordin, Mix1, Xlim1 and Cerberus.
We find no difference in gene expression between cells
attached to fibronectin and those unattached on agarose.
Most importantly, we find that cells respond to activin in a
concentration-related way irrespective of their degree of
contact with other cells. Therefore interaction among cells
is not required for the interpretation of morphogen
concentration, at least in the case of the early genes studied
here. We conclude that isolated blastula cells can sense and
respond individually to activin by expressing genes in a
concentration-dependent way.
INTRODUCTION
understanding of morphogen gradients is achieved. This asks
whether a single cell can determine its position in a
concentration gradient. Does it know its ‘positional value’
(Wolpert, 1969) in the absence of other cells? A much more
complicated situation would exist if a cell can establish its
position and determine gene expression only by reference to
other cells in the same concentration gradient. Indeed exactly
this last conclusion has been reached as a result of experiments
on dissociated Xenopus blastula cells. Green et al. (1994) and
Wilson and Melton (1994) have both argued that the major part
of the concentration-dependent response does indeed require
interactions among cells of the responding population. They
confirmed the previous results of Green et al. (1992), that the
choice of genes expressed at 17 hours after activin addition is
related to the initial activin concentration, but found that this
was not true of the response seen at 3 hours, when most genes
showed a similar progressive dose response.
We describe here a more detailed investigation of the shortterm response of cells to different activin concentrations. This
is for two reasons. First, in the course of our work, we reached
conclusions that differ significantly from those published.
Second, there has been no analysis so far of gene response to
morphogen concentration in single cells independently of their
neighbours. We have cultured activin-treated cells in different
configurations so as to provide increasing degrees of cell
contact and potential communication after activin treatment,
Activin, a member of the TGFβ family of signalling factors, is
present as mRNA and protein in the early embryos of Xenopus
(Asashima et al., 1991; Thomsen et al., 1990), and is a
candidate for one of the molecules involved in the mesodermforming induction that takes place a few hours after
fertilization (Smith, 1993). Activin has attracted much interest
in recent years, because it can activate genes in a concentrationdependent manner, thereby behaving like a morphogen. The
identification of mature activin, its cloned receptors, at least
some of its transduction molecules and several of its immediate
response genes has enabled its concentration-related behaviour
to be analyzed in more detail than that of any other vertebrate
morphogen.
The concentration-dependent effects of activin were first
established by Green and Smith (1990) using dissociated
blastula cells of Xenopus; these were exposed to different
concentrations of activin, reaggregated and found to express a
range of different genes in a dose-dependent way (Green et al.,
1992). It was later shown that an activin-loaded bead can cause
waves of gene expression to spread out through a static
population of undissociated blastula cells, establishing the
point that each cell can make a nil, low or high gene response
to activin concentration (Gurdon et al., 1994). There is,
however, a key question that needs to be answered before a full
Key words: Xenopus, Activin, Morphogen gradient, Eomesodermin,
Xbrachyury
5310 J. B. Gurdon and others
and have assayed gene responses at the single cell level by in
situ hybridization, as well as quantitatively by RNase
protection. We conclude that each cell can individually assess
different activin concentrations and can activate early genes
independently of any cell communication. This clarifies and
simplifies our understanding of the mode of action of a
morphogen.
MATERIALS AND METHODS
Cell dissociation and activin treatment
Animal caps were removed from blastulae at about stage 8.5, and their
cells dissociated by incubation in Modified Barth Saline (MBS;
Gurdon, 1977) lacking Ca2+ and Mg2+, but supplemented with EDTA
(0.5 mM) and with 0.1% bovine serum albumin (BSA, fraction V;
Sigma), for 10-15 minutes. Cells were dispersed by gentle pipetting,
incubated for 10 minutes in the desired concentration of activin
protein, washed twice in the above dissociation medium (10 caps per
1 ml), and then washed once in normal (Ca2+- and Mg2+-containing)
MBS with 0.5% BSA, before final resuspension in the desired volume
of MBS containing 0.1% BSA and 3× the normal MBS concentration
of Ca2+ (that is at 6.5 instead of the normal 2.2 mM Ca2+).
Activin was prepared as in Dyson and Gurdon (1998), and was
unlabelled. The concentrations of activin used here are given as µl
activin per 100 µl medium. This refers to a percentage dilution of our
stock preparation, prepared from the medium of mRNA-injected
oocytes. We have assayed our activin preparations for their ability to
induce mesoderm in animal caps. As in our earlier work (McDowell
et al., 1997), the concentration of activin in our oocyte medium is
about 20 nM. After Sephadex purification it is about 7 nM. A 0.45%
concentration (that induces Xbra) therefore corresponds to 30 pM.
Green et al. (1992) used Xbra activation by activin at 3 units/ml,
corresponding to 15 pM. Since we incubate our cells in activin
solution for only 10 minutes, compared to the continuous incubation
of Green et al. (1992), the efficacy of our activin treatment is
comparable to theirs.
Plating of cells on substrates and cell culture
An agarose layer was prepared on Nunclon 35 mm plastic dishes,
using LMP, ultrapure, Gibco agarose at 1% in MBS. Fibronectin was
adsorbed onto BDH superfrost plus microscope slides by layering 2
ml of a 20 µg/ml solution of fibronectin (from bovine plasma, Sigma)
in water onto each 25×75 mm slide. After incubation for 2-4 hours at
23°C in a humidified atmosphere, the fibronectin solution was washed
off and replaced by 1 ml of culture medium. Cells were pipetted at
the desired density onto the fibronectin-coated slides within 10
minutes of removal of the excess fibronectin solution.
The culture medium for all substrates and configurations of cells
was MBS with 3× its normal Ca2+ concentration, 0.5% BSA and 1
µg/ml Gentamycin (Sigma). All cells were cultured in a humidified
atmosphere at 19-23°C until control embryos had reached the desired
stage (usually stage 10.5). For freezing, cells were removed from
fibronectin substrates by incubation for about 5 minutes in MBS
lacking Ca2+ and Mg2+, but containing 0.5 mM EDTA. Preparations
were fixed for 1 hour in Memfa (Harland, 1991), transferred to 100%
methanol, and kept at −20°C in methanol until ready for further
processing.
RNase protection
This was carried out according to Ryan et al. (1996). Quantitation was
achieved by use of a Fujifilm Phosphorimager and MacBAS 2.5
software.
In situ hybridization and single cell density quantitation
Cells were hybridized on fibronectin slides according to Lemaire and
Gurdon (1994) with slight modifications as described by Butler et al.
(1999). The most important of these is a reduction in the duration of
proteinase K treatment from 30 to 12 minutes. We used Boehringer
BM purple to visualize hybridized probes. The quantitation of spread
cell in situ hybridization results was carried out on low density
preparations (see Fig. 1C). A small circular window in the centre of
a photomicroscope eyepiece was placed over a cell (at ×20
magnification), and the light meter reading recorded. A low light
intensity and appropriate film speed setting on the photographic
monitor provided readings of 5-6 for an intensely stained cell, with
background values of about 1. The background values were subtracted
in preparing the results shown in Fig. 6. Colour may have reached
saturation in the most strongly stained cells, but if so this would only
have reduced the magnitude of the effects seen in Fig. 6. It would not
change the conclusions.
Cell staining
The cells shown in Fig. 1, C-E were stained with rhodaminephalloidin (Molecular Probes R-415) to mark cytoskeletal actin, and
TOTO-3 (Molecular Probes T-3604) to stain nuclear DNA.
RESULTS
Configuration and lineage of cell cultures
To test the importance of cell contact and cell communication
we have grown activin-treated animal cap cells on agarose or
fibronectin substrates at different densities. We have cultured
cells either singly or as aggregates on agarose, to which they
do not attach (Fig. 1A,B). We have also cultured cells on
fibronectin, to which they attach, either at low density (most
cells not in contact with another cell), at middle density (most
cells in lateral contact with other cells), or at high density
(most cells surrounded by other cells; see Fig. 1C-E). When
first deposited on a substrate at low density, the great majority
of cells are singles or sister pairs (Fig. 1A, inset). During the
course of the culture period, that is from stage 8.5 to stage
10.5 or 11 (about 4 hours), many cells divide once or twice,
as is true of cells in a normal blastula. Thus the lowest density
cell preparations will contain cells in groups of 2, 4 or 8 from
sister pairs, in addition to a majority of undivided singles, as
seen in Fig. 1A,C. Cells plated at middle or high densities
tend to form clusters during the culture period, such that most
cells are in contact with, or wholly surrounded by, their
neighbours (Fig. 1D,E). We find, like previous authors, that
animal cap cells treated with activin spread out on a
fibronectin subtrate, simulating the migratory behaviour of
mesoderm cells (Howard and Smith, 1993; Ramos et al.,
1996; Smith et al., 1990; Wacker et al., 1998). The extent to
which cells spread out after attaching to a fibronectin
substrate varies somewhat from one batch of embryos to
another, but this does not affect expression of the genes
studied here (see below).
We need to ensure that, when adjacent cells are seen in low
density cultures, these are the clonal progeny of a single cell
or sister-pair, rather than being a mixed colony resulting from
two or more original cells that happened to be plated next to
each other and could therefore interact. To test this we mixed
unlabelled and rhodamine-labelled (RlDx) blastula cells in a
ratio of 3:1, respectively, and made low density cell cultures.
Each small colony was seen to consist entirely of RlDx or
unlabelled cells (not shown), and was therefore formed by
division of one original cell or cell-pair (Fig. 1A, inset).
Single cells sense morphogen concentration 5311
Quantitative analysis of gene expression by RNase
protection
We first made a quantitative analysis of gene expression in low
density populations of cells attached to fibronectin after
treatment with different concentrations of activin. Cells were
cultured for 3-4 hours to the equivalent of stage 10.5, and were
analyzed for gene expression by RNase protection. Fig. 2 shows
that Xbra (Smith et al., 1991) is activated at a low concentration
of activin, and reaches maximum expression at 0.45% activin.
Xeomes (Ryan et al., 1996) and Xgsc (Cho et al., 1991), which
has a low level of maternal mRNA, are both activated at a
threefold higher concentration of activin, and continue to show
progressively stronger expression up to 4% activin. Fig. 3A,B
compares graphically the activin dose-response in low density
(fibronectin) and reaggregated (agarose) cultures. It can be seen
that the activin concentration response of cells cultured under
these two conditions is indistinguishable. A similar result (not
shown) was obtained with cells dispersed at low density on an
agarose substrate, to which the cells did not attach. Clearly
RNase protection analysis shows no effect of cell density or the
degree of cell contact on the concentrations of activin at which
Xbra and Xgsc are activated. We conclude that cells can sense
small differences in activin concentration, and can activate
different genes in a concentration-related way, in the absence of
cell communication.
These results have the benefit of accurate quantitation by
RNase protection, but they apply to populations of cells and do
not tell us about the behaviour of individual cells.
In situ analysis of single cells
The RNase protection results just described are equally
consistent with two quite different interpretations. One is that
each cell senses and responds to a particular activin
concentration in the same way as every other cell in the same
preparation. Thus at a low activin concentration each cell
will express Xbra but not Xgsc or Xeomes; at a higher
concentration, each cell will express Xeomes strongly, and
Xbra more weakly than at the low concentration. This would
show that each cell can assess its position in a concentration
gradient independently of its neighbours. The other possibility
is that, while the cell population as a whole shows the doseresponse described, individual cells make very different
responses to the same activin concentration. It is very
important to distinguish these possibilities, if we are to
understand the mechanism of cell response to a morphogen.
Dissociated cells were treated with different concentrations
of activin as usual, and plated onto fibronectin slides at low
density and therefore in the absence of cell communication.
Fig. 4 shows representative cells in situ-hybridized to probes
for Xbra, Apod, Eomes or Chordin. It can be seen that the
intensity of the colour reaction corresponds well with the
concentration response observed in the RNase protection
analyses above (Fig. 2); Xbra expression is strongest at a
0.45% dose of activin, decreasing at higher concentrations,
whereas Xeomes expression increases progressively from being
very weak at 0.45%, to strong expression at 4% and above.
Xgsc (not shown) follows broadly the same pattern as Xeomes,
but its maternal content of mRNA reduces the clarity of the
effect at low activin concentrations. We have carried out a
similar series of in situ hybridizations with Apod, a gene
expressed at stage 10.5 (Stennard et al., 1996) as a zygotic
isoform of VegT (Stennard et al., 1999). VegT (Zhang and King,
1996) is a gene also known as Xombi (Lustig et al., 1996), and
Brat (Horb and Thomsen, 1997; see Stennard et al., 1997, for
a review). Apod behaves like Xbra in dispersed cell
preparations, in that it responds strongly to low concentrations
of activin; however it does not decrease to the same extent as
Xbra at higher activin concentrations (Fig. 4). We have also
carried out in situ hybridization on dispersed cells using
Xchordin (Sasai et al., 1994) as a probe. Xchordin shows a
response pattern very similar to Xeomes (Fig. 4). The results
shown in Fig. 4 were all obtained in the same experimental
series. Altogether seven series with embryos from different
batches of eggs were hybridized with two or more of the same
probes, and gave comparable results.
We now ask how uniformly cells respond to activin
concentration; are the single cells shown in Fig. 4 representative
of other cells in the same samples? Low power views of the in
situ hybridization preparations show apparent uniformity of
expression among cells that received the same activin
concentration (Fig. 5). However, the low power views seen in
this figure do not reveal the detailed differences in staining
intensity clearly visible in Fig. 4. Therefore to quantitate more
precisely the variation within a cell population, we measured
the intensity of stain in single cells in each low density culture.
Using the procedure described in Materials and Methods, we
determined the density of in situ hybridization staining for 40
cells in each sample. The results seen in Fig. 6 show substantial
uniformity of gene expression among cells that received a
particular concentration of activin. The validity of our method
of quantitating the staining intensity in each cell is indicated by
the agreement between assessment of gene expression by
RNase protection (Fig. 2) and in situ hybridization (Fig. 6). As
a further check on this procedure, we have used a probe for the
ubiquitously expressed gene product EF1α, commonly used as
a quantitative control in RNase protection experiments. The
slight decline in EF1α expression at increasing concentrations
of activin, as seen in Fig. 6, is expected in view of the increased
cell spreading that results from activin treatment.
In conclusion, assays of gene expression at the single cell
level show the same concentration-related responses as
reaggregated cells. This agrees with our RNase protection
results, and, most importantly, it establishes that each cell can
determine its position in a concentration gradient without
reference to its neighbours.
Other early genes undergo cell contact-independent
activation in response to activin
In addition to the genes tested above, a number of other
mesodermal and endodermal genes are expressed after activin
treatment of animal cap cells. To assess the generality of our
conclusions, we have tested activin responsiveness in a
number of other early Xenopus genes under isolated, as
compared to reaggregated, cell culture conditions. First, the
magnitude of activin response, and the ability of cells to
express these genes after cell dissociation, was tested using
RNase protection (1) in animal caps taken from embryos
which had been injected with activin mRNA (10 or 35 pg/2
cell), (2) in animal caps which had been implanted with
activin protein-coated beads, or (3) in animal caps which had
been dispersed, treated with activin protein (4.0% or 90 pM),
and then reaggregated for culture. The transcriptional
5312 J. B. Gurdon and others
Fig. 1. Configurations of cell culture. Cells from stage 8.5-9 animal
caps are dissociated, incubated in activin at the required
concentration and washed, as described by Dyson and Gurdon
(1998). The two substrates used are 1% agarose or fibronectin (see
Materials and Methods). In all cases the MBS culture medium
contains Ca2+, Mg2+ and 0.1% bovine serum albumin. (A) Single
cells on agarose are distributed sparsely so that, initially, nearly all
cells are present as single cells or as sister pairs (inset). After culture,
each cell or sister-pair has formed well separated clonal groups of 2,
4 or 8 cells or, more often, has remained as a single cell. (B) To form
reaggregates, several thousand cells are centrifuged into a pellet,
which is then transferred to an agarose substrate. (C-E) Cells are
distributed at low (C), medium (D) or high density (E) on to a
fibronectin substrate to which they attach. The right hand column for
Figs C-E show cells stained with rhodamine-phalloidin and TOTO-3
(see Methods). At the equivalent of stage 10.5, cells are frozen or
fixed.
activation of Chordin, Cerberus (Bouwmeester et al., 1996),
Xlim1 (Taira et al., 1992) and Mix1 (Rosa, 1989) was detected
in these animal caps, while none of these genes was expressed
in caps that had no exposure to activin. All these genes are
therefore activin-responsive and, for most genes, their levels
of expression are similar in all of the above treatments (not
shown). The response to activin by Hex (Newman et al.,
1997) and to Sox17β (Hudson et al., 1997) was too low in all
three samples to test the importance of cell interaction for
response to activin.
For those genes that were strongly induced by activin
treatment when reaggregated after dissociation (namely
Chordin, Cerberus, Xlim1 and Mix1), the magnitude of
response was then compared in low density and reaggregated
animal cap cells, to determine whether cell communication is
required for gene activation following exposure to activin.
Stage 8.5 animal caps were disaggregated, treated with 4%
activin protein, and then cultured to stage 11 either as isolated
low density cells or as reaggregates. They were then frozen
for analysis by RNase protection. Whole embryos were used
as positive controls, and animal caps which had no activin
treatment as negative controls. The results were quantitated
and a ratio calculated for each gene to show the level of
expression in reaggregated cells compared to low density
Fig. 2. RNase protection analysis of gene expression under low
density culture conditions. Xbra is activated maximally at 0.45%
activin. Xgsc and Xeomes reach maximum activation at 1.35%
activin. WE, whole embryo.
Single cells sense morphogen concentration 5313
cells. As shown in Table 1, Chordin, Cerberus,
Xlim1 and Mix1 show a difference of less than
twofold in response to activin in reaggregated
versus low density cells. The same is true for
Xgsc, included in the analysis as its activation in
mesoderm cells had already been demonstrated
to be largely independent of cell contact
(Lemaire and Gurdon, 1994).
These results were confirmed at the single cell
level using in situ hybridization to test whether
isolated cells respond uniformly to activin.
Dissociated animal cap cells were treated with 4%
activin protein and then cultured in low density
configuration on fibronectin-coated slides. We find
that Xgsc, Xcerberus, Xlim1 and Xmix1 show
uniform expression at the single cell in situ level
(not shown); results with Xbra, Xapod, Xeomes
and Xchordin have already been shown in Fig. 5.
Fig. 3. Graphic representation of gene activation response to activin concentration,
for (A) low density and (B) reaggregated cell cultures. The values shown were
obtained by phosphorimager assessment of gel analyses by RNase protection. B is
taken from Dyson and Gurdon (1998, with permission), since the results shown in A
and B are from the same experimental series. Comparable results were obtained in
two other experimental series.
Fig. 4. Single cell analysis of gene activation in low density cell cultures. Low density preparations of cells attached to a fibronectin substrate
(as in Fig. 1C) were in situ-hybridized to the probes stated (Xbra, Xapod, Xeomes and Xchordin). The colour reaction is BM purple. Each figure
shows a representative clonal colony of 1-4 cells, fixed at stage 10.5. The + or – values under each figure are based on densitometry readings
(see Materials and Methods). −, background value; ±, density value 2; +, ++, +++, density values 3, 4 and 5 or more respectively (see Fig. 6).
Altogether seven independent experimental series gave results similar to these (see text).
5314 J. B. Gurdon and others
Fig. 5. Low-power views of in situ hybridization to low density cultures on fibronectin slides. There is a uniformity of gene expression among
cells at each activin concentration, although these low magnification views do not reveal the more precise differences in staining intensity seen
in Fig. 4, and quantitated in Fig. 6.
We conclude that these early genes, whether normally expressed
in the mesoderm (Xbra, Xgsc, Xeomes, Xapod, Xcerberus,
Xlim1, Xchordin) or endoderm (Xmix1) show a strong
transcriptional response to activin, without the need for cell
interactions. We cannot exclude a small effect of cell
communication on the magnitude of the response, but we find
that this affects gene expression by less than a factor of two
(Table 1). Therefore our results with these other genes are in
accord with those for Xeomes, Xbra, Xapod and Xgsc, namely
that cells can respond to morphogen in the absence of cell
communication.
DISCUSSION
Comparison with previous work on Xenopus
The conclusions we reach in this work clearly differ from those
of the only comparable previous work (Green et al., 1994;
Wilson and Melton, 1994; Symes et al., 1994). It should first
be said that the main theme of the papers cited was to compare
gene expression at stage 10.5 with that at stage 17. In contrast,
the aim of our experiments reported here is to determine
whether or not there are real differences between genes in their
early (stage 10.5) dose-response to activin, and whether their
response depends on cell contact and communication. The
experiments described here cannot therefore be compared
directly with those of the previous authors. For this reason none
of our RNase protection results are inconsistent with those
described by others. We should point out that Symes et al.
(1994) interpret their RNase protection assays to show a broad
overlapping dose response for Xbra and Xgsc, a conclusion not
in accord with our results. However, their relevant figure (their
Fig. 4, upper gel) shows the strongest Xbra expression at an
activin concentration 4-8 times lower than their highest level
Single cells sense morphogen concentration 5315
Table 1. Gene expression in reaggregated and low density
cell cultures
Xgsc
Chordin
Cerberus
Xlim1
Mix1
Ratio
Reaggregated
Low density
Reaggregated : Low density
125
297
76
85
226
145
279
120
108
244
0.90±0.22
1.03±0.21
0.52±0.19
0.83±0.13
1.02±0.22
Gene induction by activin in both reaggregated and low density cultures of
animal cap cells was analyzed by RNase protection. The values for
normalized expression are percentages of the FGF receptor loading control in
each gel track. The results for each gene are mean values from nine analyses
in three independent experimental series. The ratios shown are ± s.d.
For all these genes, there is a difference of less than twofold when
comparing expression in reaggregated and low density cultures.
of Xgsc expression. We do not therefore consider their results
to be inconsistent with ours.
The most important contribution of our work has been to
assay gene expression at the single cell level (by in situ
hybridization) on isolated (low density) cultured embryonic
cells. We are not aware of work in which this has been done
before, possibly because it is necessary to ensure that the
procedures used to permeabilize whole cells sufficiently for
probe penetration do not cause detachment of cells from their
substrate. We cannot therefore compare our results of this kind
to previous work since that has not assessed gene expression
in isolated cells. We believe that it is only by inclusion of this
kind of analysis that a true understanding of cell response to
morphogen concentration can be obtained.
We should comment on an apparent technical difference
between our experimental procedures and those of the authors
cited above. We believe that Xenopus embryo cells are
particularly sensitive to the lack of calcium and magnesium in
the medium when cultured beyond stage 10, and that this is
particularly important, not surprisingly, when cells are cultured
in a low-density configuration. In all our experiments, we
provide calcium at or slightly above the 2 mM level
characteristic of most amphibian culture media. It appears that,
in the work of Green et al. (1994) and Wilson and Melton
(1994), dissociated cells were cultured continuously in calcium
and magnesium-free medium up till the stage of analysis. This
factor may contribute to some of the differences between our
results and theirs. In other work involving single cell culture
but addressing different questions, calcium and magnesium
were present in the culture medium (e.g. Godsave and Slack,
1991).
The results obtained here for the response of animal cap cells
to activin may be compared with those described previously
using equatorial cells not treated with activin (Lemaire and
Gurdon, 1994). Dorsal equatorial cells, whether dissociated or
in whole embryos, express the mesodermal gene Xgsc (Cho et
al., 1991), as do activin-treated animal cap cells. Xgsc has a
promoter with two elements that respond to two different
signalling pathways (Watabe et al., 1995). The proximal
element responds to the Wnt/β catenin pathway. The distal
element is required for Xgsc to respond to activin-class signals,
which are likely to originate from a distant source such as the
Nieuwkoop Centre, and therefore to involve communication
3
2
Intensity of in situ stain over background
Normalized expression
5 Xbra
4
1
5 Apod
4
3
2
1
5 Eomes
4
2
3
1
2
1
5 Chordin
4
3
2
1
5 EF1α
4
3
2
1
0.15 0.45 1.35 4.0 12.0
Activin concentration (%)
Fig. 6. Summary of in situ hybridization results with low density cell
preparations on a fibronectin substrate. For each probe and for each
concentration, 40 single cells were scored for the intensity of stain
(see Materials and Methods). The cells scored were present as single
cells or in clonal colonies of 2-4 cells, but the density of stain was
recorded over single cells. Standard deviations are shown. These
results derive from three independent experimental series.
between cells for establishing a concentration gradient. In the
experiments described here, we provide dissociated cells with
different concentrations of activin, thereby bypassing a normal
cell communication stage. Once cells have received an activin
signal, we now find that they do not need to communicate with
each other to make an appropriate concentration-related
response such as Xgsc. Our results are therefore consistent with
previous results using dissociated Xenopus cells.
Finally it will be helpful to relate our results obtained here
to the community effect described previously. The community
effect applies to Xenopus embryonic mesoderm cells, but is
operative only from a later stage, the mid-gastrula, onwards
(Gurdon et al., 1993). In support of this point, Symes et al.
(1998) observed out that cells cultured at low density in Ca2+deficient medium, and therefore unable to communicate with
each other, will not activate muscle genes. Likewise, Green et
al. (1994) and Wilson and Melton (1994) suggested that a
community effect may help sharpen response to activin
concentrations. Thus a complete response to activin
concentration may have two phases. First, cells respond
5316 J. B. Gurdon and others
individually to activin concentrations during blastula stages.
Then there is a secondary response during gastrulation, when
the community effect may sharpen borders between cells
expressing different genes.
Comparison with non-amphibian work
In Drosophila wing discs it has often been useful to generate
clones of cells that are mutant in respect of signalling
molecules or their receptors. In some cases, the mutant clones
are very small, consisting of less than 10 cells. For example,
the Drosophila mutation Saxophone encodes a defective type
I receptor for DPP. In the wing disc small clones of cells
lacking sax show the phenotype expected of their own genetic
deficiency, and not that appropriate to their position in the DPP
gradient as indicated by their surrounding wild-type cells
(Singer et al., 1997). Therefore these cells determine their
position in a concentration gradient according to their own
genetic contribution, and do not depend on communication
with their wild-type neighbours (Gurdon et al., 1998).
Future directions
Leading on from the results reported, future work will ask what
later genes are expressed in isolated cells in the absence of cell
communication. Our previous experiments (e.g. Gurdon et al.,
1993; Carnac and Gurdon, 1997) have argued that muscle
genes are not activated in the absence of cell contact and that
their activation depends on a community effect. Our results
presented here show that the earliest gene responses to activin
do not depend on a community effect. In future work, we will
want to know which other later (cycloheximide-sensitive)
genes require cell communication for a dose-dependent
response to activin. Most significantly, we have now
established conditions that lead to a uniform gene response to
different activin concentrations, and we can ask whether all
cells that make an early concentration-related response (e.g.
Xbra or Xeomes) to activin behave uniformly in respect of later
gene activations, or whether more complex cell interactions are
involved.
We thank the Cancer Research Campaign for a programme grant
to J. B. Gurdon. This work was also supported by funds from the
Medical Research Council (S. D.), the European Community (T. L.),
a Royal Society Howard Florey Fellowship (F. S.) and the Japanese
Society for the Promotion of Science (K. S.). We thank the following
for probes: J. C. Smith (Xbra); E. De Robertis (Xgsc, Chordin,
Cerberus); I. B. Dawid (Xlim); P. Krieg (Hex); F. Rosa (Mix1).
REFERENCES
Asashima, M., Nakano, H., Uchiyama, H., Sugino, H., Nakamura, T.,
Eto, Y., Ejima, D., Nishimatsu, S., Ueno, N. and Kinoshita, K. (1991).
Presence of activin (erythroid differentiation factor) in unfertilized eggs
and blastulae of Xenopus laevis. Proc. Natl. Acad. Sci. USA 88, 65116514.
Bouwmeester, T., Kim, S.-Y., Sasai, Y., Lu, B. and De Robertis, E. M.
(1996). Cerberus is a head-inducing secreted factor expressed in the anterior
endoderm of Spemann’s organizer. Nature 382, 595-601.
Butler, K., Zorn, A. and Gurdon, J. B. (1999). Non-radioactive on situ
hybridization to Xenopus tissue sections. Methods: in situ hybridization in
developmental biology. Academic Press (in press).
Carnac, G. and Gurdon, J. B. (1997). The community effect in Xenopus
myogenesis is promoted by dorsalizing factors. Int. J. Dev. Biol. 41, 521524.
Cho, K. H. Y., Blumberg, B., Steinbeisser, H. and De Robertis, E. M.
(1991). Molecular nature of Spemann’s organizer: the role of the Xenopus
homeobox gene goosecoid. Cell 67, 1111-1120.
Dyson, S. and Gurdon, J. B. (1998). The interpretation of position in a
morphogen gradient as revealed by occupancy of activin receptors. Cell 93,
557-568.
Godsave, S. F. and Slack, J. M. W. (1991). Single cell analysis of mesoderm
formation in the Xenopus embryo. Development 111, 523-530.
Green, J. B. and Smith, J. C. (1990). Graded changes in dose of a Xenopus
activin A homologue elicit stepwise transitions in embryonic cell fate.
Nature 347, 391-394.
Green, J. B., New, H. V. and Smith, J. C. (1992). Responses of embryonic
Xenopus cells to activin and FGF are separated by multiple dose thresholds
and correspond to distinct axes of the mesoderm. Cell 71, 731-739.
Green, J. B., Smith, J. C. and Gerhart, J. C. (1994). Slow emergence of a
multithreshold response to activin requires cell- contact-dependent
sharpening but not prepattern. Development 120, 2271-2278.
Gurdon, J. B. (1977). Methods for nuclear transplantation in Amphibia. In
Methods in Cell Biology 16 (ed. G. Stein, J. Stein and L. J. Kleinsmith), pp.
125-139. New York: Academic Press.
Gurdon, J. B., Tiller, E., Roberts, J. and Kato, K. (1993). A community
effect in muscle development. Curr. Biol. 3, 1-11.
Gurdon, J. B., Harger, P., Mitchell, A. and Lemaire, P. (1994). Activin
signalling and response to a morphogen gradient. Nature 371, 487492.
Gurdon, J. B., Dyson, S. and St Johnson, D. (1998). Cells’ perception of
position in a concentration gradient. Cell 95, 159-162.
Harland, R. M. (1991). In situ hybridisation: an improved whole-mount
method for Xenopus embryos. Meth. Cell Biol. 36, 685-695.
Horb, M. E. and Thomsen, G. H. (1997). A vegetally localized T-box
transcription factor in Xenopus eggs specifies mesoderm and endoderm and
is essential for embryonic mesoderm formation. Development 124, 16891698.
Howard, J. E. and Smith, J. C. (1993). Analysis of gastrulation: different
types of gastrulation movement are induced by different mesoderm-inducing
factors in Xenopus laevis. Mech. Dev. 43, 37-48.
Hudson, C., Clements, D., Friday, R. V., Stott, D. and Woodland, H. R.
(1997). Xsox17alpha and -beta mediate endoderm formation in Xenopus.
Cell 91, 397-405.
Lemaire, P. and Gurdon, J. B. (1994). A role for cytoplasmic determinants
in mesoderm patterning: cell- autonomous activation of the goosecoid and
Xwnt-8 genes along the dorsoventral axis of early Xenopus embryos.
Development 120, 1191-1199.
Lustig, K. D., Kroll, K. L., Sun, E. E. and Kirschner, M. W. (1996).
Expression cloning of a Xenopus T-related gene (Xombi) involved in
mesodermal patterning and blastopore lip formation. Development 122,
4001-4012.
McDowell, N., Zorn, A. M., Crease, D. J. and Gurdon, J. B. (1997). Activin
has a direct longe range signalling activity and can form a concentration
gradient by diffusion. Curr. Biol. 7, 671-681.
Newman, C. S., Chia, F. and Krieg, P. A. (1997). The XHex homeobox gene
is expressed during development of the vascular endothelium:
overexpression leads to an increase in vascular endothelial cell number.
Mech. Dev. 66, 83-93.
Ramos, J. W., Whittaker, C. A. and DeSimone, D. W. (1996). Integrindependent adhesive activity is spatially controlled by inductive signals at
gastrulation. Development 122, 2873-2883.
Rosa, F. M. (1989). Mix.1, a homeobox mRNA inducible by mesoderm
inducers, is expressed mostly in the presumptive endoderm cells of Xenopus
embryos. Cell 57, 965-974.
Ryan, K., Garrett, N., Mitchell, A. and Gurdon, J. B. (1996).
Eomesodermin, a key early gene in Xenopus mesoderm differentiation. Cell
87, 989-1000.
Sasai, Y., Lu, B., Steinbeisser, H., Geissert, D., Gont, L. K. and De
Robertis, E. M. (1994). Xenopus chordin: a novel dorsalizing factor
activated by organizer-specific homeobox genes. Cell 79, 779-790.
Singer, M. A., Penton, A., Twombly, V., Hoffmann, F. M. and Gelbart, W.
M. (1997). Signaling through both type I DPP receptors is required for
anterior-posterior patterning of the entire Drosophila wing. Development
124, 79-89.
Smith, J. C., Symes, K., Hynes, R. O. and DeSimone, D. (1990a). Mesoderm
induction and control of gastrulation in Xenopus laevis: the roles of
fibronectin and integrins. Development 108, 229-238.
Smith, J. C., Price, B. M. J., Green, J. B. A., Weigel, D. and Herrmann,
Single cells sense morphogen concentration 5317
B. G. (1991). Expression of a Xenopus homolog of Brachyury (T) is an
immediate-early response to mesoderm induction. Cell 67, 79-87.
Smith, J. C. (1993). Mesoderm-inducing factors in early vertebrate
development. EMBO J. 12, 4463-4470.
Stennard, F., Carnac, G. and Gurdon, J. B. (1996). The Xenopus T-box
gene, Antipodean, encodes a vegetally localised maternal mRNA and can
trigger mesoderm formation. Development 122, 4179-4188.
Stennard, F., Ryan, K. and Gurdon, J. B. (1997). Markers of vertebrate
mesoderm induction. Curr. Opin. Genet. Dev. 7, 620-627.
Stennard, F., Zorn, A. M., Ryan, K., Garrett, N. and Gurdon, J. B. (1999).
Differential expression of VegT and Antipodean protein isoforms in
Xenopus. Mech. Dev. (in press).
Symes, K., Yaqoob, M. and Smith, J. C. (1988). Mesoderm induction in
Xenopus laevis: responding cells must be in contact for mesoderm formation
but suppression of epidermal differentiation can occur in single cells.
Development 104, 609-618.
Symes, K., Yordan, C. and Mercola, M. (1994). Morphological differences
in Xenopus embryonic mesodermal cells are specified as an early response
to distinct threshold concentrations of activin. Development 120, 2339-2346.
Taira, M., Jamrich, M., Good, P. J. and Dawid, I. B. (1992). The LIM
domain-containing homeobox gene Xlim-1 is expressed specifically in
the organizer region of Xenopus gastrula embryos. Genes Dev. 6, 356366.
Thomsen, G., Woolf, T., Whitman, M., Sokol, S., Vaughan, J., Vale, W. and
Melton, D. A. (1990). Activins are expressed early in Xenopus
embryogenesis and can induce axial mesoderm and anterior structures. Cell
63, 485-493.
Wacker, S., Brodbeck, A., Lemaire, P., Niehrs, C. and Winklbauer, R.
(1998). Patterns and control of cell motility in the Xenopus gastrula.
Development 125, 1931-1942.
Watabe, T., Kim, S., Candia, A., Rothbacher, U., Hashimoto, C., Inoue, K.
and Cho, K.W.Y. (1995). Molecular mechanisms of Spemann’s organizer
formation: conserved growth factor synergy between Xenopus and mouse.
Genes Dev. 9, 3038-3050.
Wilson, P. A. and Melton, D. A. (1994). Mesodermal patterning by an inducer
gradient depends on secondary cell-cell communication. Curr. Biol. 4, 676686.
Wolpert, L. (1969). Positional information and the spatial pattern of cellular
differentiation. J. Theor. Biol. 25, 1-47.
Zhang, J. and King, M. L. (1996). Xenopus VegT RNA is localized to the
vegetal cortex during oogenesis and encodes a novel T-box transcription
factor involved in mesodermal patterning. Development 122, 4119-4129.