- Email: [email protected]
Activin as a morphogen in Xenopus mesoderm induction
seminars in C E L L & D E V E L OP M E N T A L B I OL OG Y , Vol 10, 1999: pp. 311]317 Article No. scdb.1999.0307, available online at http:rrwww.idealibrary.com on
Activin as a morphogen in Xenopus mesoderm induction Natasha McDowell and J. B. Gurdon
Activin, a member of the Transforming Growth Factor beta (TGF-b ) superfamily, can behave as a morphogen in cells of the early Xenopus embryo by inducing a range of mesodermal genes in a concentration-dependent manner. This review examines the behaviour of activin as it forms a morphogen gradient. It also discusses how a cell can perceive its position in a concentration gradient in order to activate appropriate mesodermal gene responses.
However, often in development, the long-range signal appears to act as a morphogen. Classically defined, this is a molecule which, when secreted by cells at one location in an embryo, diffuses away from its source to create a concentration gradient across the surrounding tissue. Thus cells near the source receive a high concentration of signal and follow one pathway of development, whilst cells further away receive a lower concentration of signal and adopt a different cell fate. Hence, an individual cell must not only assess whether signalling molecules are present in its local environment, but it must also quantify the amount of signal it perceives in order to elicit a concentration-dependent response. Furthermore, although only one type of molecule is involved, such a mechanism can generate a diversity of cell-types in a defined spatial relationship to each other. In contrast to this classical definition, a concentration-dependent response might also be established by an antagonising molecule modulating the activity of the inducing molecule in a graded fashion.1] 3 In such a case, the inducing molecule could be locally produced over the entire responding field, but possess a gradient of activity due to the long-range diffusion of inhibitor from a localised source 2,3 ŽDale and Wardle, this issue..
Key words: activin r morphogen r mesoderm induction r Xenopus Q1999 Academic Press
AS THE EMBRYO DEVELOPS, it divides from a single cell, the fertilised egg, into a complex organism with an array of tissue types and an organised body plan. Hence, not only does cellular diversity increase, but the cells must also become spatially arranged in relation to one another, a process termed pattern formation. How is this diversity and pattern formation achieved? In very early development, coarse spatial cues are often established by the use of cytoplasmic maternal determinants, proteins or mRNAs which are asymmetrically localised within the fertilised egg and early embryo. The other major mechanism is that of cell-to-cell signalling. While short-range interactions require direct contact between inducing and responding cells, long-range signalling molecules include proteins capable of passing across many cell diameters. In both cases, the response can ‘theoretically’ be a simple binary decision; provided the responding cells perceive sufficient signal, their development will be diverted to a new cell fate.
What makes a molecule a morphogen? Most importantly, a morphogen must induce differ ent cell fates at different concentrations. However, it must be further shown that the morphogen can elicit at least two qualitatively different responses, apart from the nil response, from each responding cell. Otherwise the concentration-dependent patterning could simply be due to the responding cells being a heterogeneous population, with some cells capable of responding to only a high signal concentration and others to only a low concentration.4 The response to the putative morphogen must also be demonstrated to be direct; if indirect, the concentration-dependent effect would apply to a different
From the Wellcomer CRC Institute of Cancer and Developmental Biology, Tennis Court Road, Cambridge CB2 1QR, UK and Department of Zoology, University of Cambridge, Cambridge, England, UK Q1999 Academic Press 1084-9521r 99 r 030311q 07 $30.00r 0
311
N. McDowell and J. B. Gurdon
molecule.5 For example, it must be shown that the signal does not simply act at short-range to activate a second long-range signal, or alternatively that it does not initiate a relay of local short-range interactions, in which each subsequent cell in the relay releases a different inducing molecule in addition to adopting a particular cell fate.
internal organs. The traditional model of mesoderm induction is that maternal morphogen-like signals emanate from the vegetal pole Žprospective endoderm. during the early cleavage stages to induce the mesoderm from the overlying animal cap cells Žprospective ectoderm..10 We have used the induction of mesoderm by activin as a model system to analyse the behaviour of a morphogen during vertebrate development and how a cell perceives its position in a gradient.
Putative morphogens Since 1897, when Thomas Hunt Morgan6 first used the concept of a gradient of ‘stuff’ to explain how polarity and pattern formation during regeneration may be controlled, there has been an extensive history of theoretical discussion and experimental contribution concerning the concept of a morphogen gradient. Many of the ideas have been summarised in a number of reviews.5,7 ] 12 In Drosophila the most convincing morphogens are Wingless ŽWg. and Decapentaplegic ŽDpp. in patterning the wing ŽStrigini and Cohen, this issue., and Hedgehog ŽHh. in patterning the adult abdominal segments. All three have been shown to directly activate downstream genes in a concentration-dependent manner.13 ] 19 In vertebrates, a lot of work has focused on activin, a member of the Transforming Growth Factor Beta ŽTGF-b . family, and on its role as a morphogen in the induction of the mesoderm in the frog, Xenopus laevis. The mesoderm is the middle germ layer of the body and later gives rise to such diverse tissue types as notochord, muscle, blood and parts of nearly all
Is activin a morphogen? Does it have a concentration-dependent response? By dissociating and then incubating blastula animal cap cells in various activin concentrations, Green and Smith20 and Green et al.21 demonstrated that the higher the concentration of activin used, the more dorsal the induced mesodermal cell fate. However, the use of dissociated cells cannot establish whether each individual cell has the potential to make a range of responses. We therefore sandwiched intact blastula tissue around activin loaded beads to create a directional source of putative morphogen and assayed gene response by in situ hybridisation. While low activin concentrations induced low response genes such as Xbrachury (Xbra) and Xpo adjacent to the source, high concentration beads induced Xbra and Xpo in more distant cells, the intervening cells expressing the high response genes Goosecoid (Xgsc),
Figure 1. A concentration-dependent wave of gene-expression. Animal cap tissue is sandwiched around either low or high concentration activin beads and gene response assayed by in situ hybridisation. Continuous regions of gene expression are observed under these conditions in which there is no cell movement. Based on ref 17.
312
Activin as a morphogen in Xenopus mesoderm induction
Eomesodermin, (Eomes), Antipodean and Mix122,4 ŽFigure 1.. The existence of such a concentration-dependent wave of gene expression, as well as the observation that labelled and unlabelled cells were never interspersed with each other in the same region, suggests that each individual cell can elicit either a high, low or nil response.4 We have also used oocytesynthesised 35 S-labelled activin to provide visual evidence Žto a distance of approx. seven cell diameters. that activin can form a concentration gradient.23
a non-activin second signal would presumably have induced a response in these distant cells unable to respond to activin. In a related study, Reilly and Melton24 conclude that secondary signalling does take place. The discrepancy between the two conclusions is presumably because we observe immediate responses ŽStage 10.5. to the putative morphogen activin, whereas they do not analyse response genes until Stage 35. During this time period, it is likely that many secondary signalling events will have occurred.
Can activin act directly and at a distance?
Diffusion or relay?
By using lineage-labelled layers of responding cells and cytochalasin, we showed that neither cell movement or cell division can account for the long-range effects of activin.22 However, this does not tell us whether activin acts directly on distant cells, or through the intermediary of another signalling molecule. We have therefore used a dominant-negative receptor specific to activin to show that cells distant from the activin source require functional activin receptors to elicit a response.23 This indicates that the long-range effects of activin are direct, since
As an extension of this work, we were interested in whether signal passage across tissue occurs by diffusion or relay ŽFigure 2a,b.. The fact that activin itself is required by all the responsive cells does not preclude the possibility that a relay of activin signalling is involved, activin being released, at an attenuated level, by each step in the relay series. Initially we showed that activin can pass through reaggregated tadpole endoderm which had been cycloheximide treated so as to prevent protein synthesis and thus the production of new activin.22 However, this does not
Figure 2. Possible mechanisms of signal passage. Ža. Activin molecules diffuse in the extracellular-matrix surrounding cells to create a concentration gradient across the tissue. Žb. A relay of activin signalling from cell to cell; each subsequent cell transmits less activin than the previous one. Žc. Transfer-mediated movement; activin might bind with low affinity to the protein core of betaglycan to aid its passage across tissue.
313
N. McDowell and J. B. Gurdon
examine the situation in intact Žnon-dissociated. blastula tissue, the stage at which mesoderm induction occurs in vivo. Three lines of evidence argue against a relay mechanism. Jones et al.25 used a constitutively active activin receptor to show that, in the absence of ligand, response genes are autonomously induced; if a relay was operative, non-autonomous induction would be expected. Our observation of exogenous 35 Slabelled activin molecules in cells distant from the activin bead source also demonstrates that a relay of activin need not occur.23 In addition, we have shown that the closely-related, but non-endogenous signal, TGF-b2, can be transmitted across intact blastula stage tissue that cannot respond to or synthesise it and can then induce mesodermal genes in distant responsive celIs.23 TFG-b 2 was used since blastula cells have to be injected with the Type II TGF-b receptor in order to respond to it and so normal wild-type tissue can be used as the non-responsive intervening cell layer.23,24 In contrast, Reilly and Melton24 advocate a process of relay for TGF-b molecules in Xenopus since they did not observe such mesodermal gene induction in a clone of distant responsive cells. We believe that the discrepancy can be explained by their use of TGF-b 1 mRNA rather than protein as the source of ligand. In our hands and unlike the case for activin-mRNA, cells containing TGF-b 1 mRNA cannot signal over a longrange, perhaps due to inefficient processingrsecretion of a protein that is a non-endogenous molecule.23 Although it appears that a relay mechanism is not operative, it is still an open question as to whether activin diffuses through or between cells. If the latter occurs, then it must also be asked whether the morphogen diffuses passively through the extracellular matrix or whether diffusion is aided. For example, a transfer-mediated mechanism could be envisaged whereby the signal is propelled along cell membranes, as well as passed from cell to cell, by binding to low affinity proteoglycan receptors on the cell membrane ŽFigure 2c; see also ref 26. ŽKersberg et al; Pfeiffler and Vincent, this issue..
cell to activate different signal transduction pathways in response to different concentrations of the morphogen. For example, a cell might possess receptors of differing affinities: Only high affinity receptors would be bound by low ligand concentrations, thereby activating a signal transduction pathway leading to the expression of low response genes. At high ligand concentrations low affinity receptors would also be bound, thus activating the high response pathway as well. Alternatively, the various concentrations of the morphogen might activate the same signal transduction pathway but to differing degrees. We have used 35 S-activin to determine how cells use their receptors to perceive a change in morphogen concentration.27 We initially investigated the binding of 35 S-activin to cell surface receptors in cells overexpressing the activin Type II B receptor Ž ActRIIB .. This was then compared to ligand binding in normal cells. As might be expected, in both cases an increase in activin concentration leads to an increase in receptor occupancy and causes a corresponding switch from Xbra to Xgsc expression. Interestingly though, Scatchard analysis showed that in both the injected cells and in the wild-type cells a single type of receptor is bound; in wild-type cells, this receptor species had a similar affinity to the introduced ActRIIB in the overexpressing cells. Furthermore, by determining the amount of activin required to saturate all a cell’s available receptors, we have shown that only 2% receptor occupancy of this receptor species is required to switch cells from a nil to a low Ž Xbra. gene response, while 6% receptor occupancy is necessary to induce a high response Ž Xgsc .. Hence, it appears that the activin morphogen response is elicited by a differing activation of the same signal transduction pathway. This leads to the question of what is the absolute number of activin molecules required to induce Xbra and Xgsc? From the values for saturation binding, as well as from the specific activity of the activin preparation, we calculated that wild-type cells have approximately 5000 activin binding receptors per cell. Hence, Xbra is induced when 100 activin molecules are bound, while 300 molecules are needed to induce Xgsc. We then asked whether cells detect an absolute rather than relative occupancy of receptors. To do this we compared the receptor occupancy at which genes are activated in injected versus uninjected cells ŽFigure 3b.. The injection of 1 ng ActRIIB-mRNA, which corresponds to approximately 38,000 receptors per cell, showed Xbra and Xgsc to be induced at 0.3% and 0.8% receptor occupancy,
How does a cell perceive its position in a morphogen gradient? For a cell to respond to a morphogen gradient, it must convert a quantitative extracellular signal concentration into a qualitative pattern of gene expression in the nucleus. One mechanism would be for the 314
Activin as a morphogen in Xenopus mesoderm induction
imaginal disc, where it appears that although signal transduction involves a quantitative difference in the same transduction pathway, two Type 1 receptors participate.15 To transduce the activin signal, Smad2 and Smad4 are thought to be phosphorylated by the Type 1 receptor and move from cytoplasm to nucleus. Graff et al 30 have shown that overexpression of Smad2 causes the induction of Xgsc at a higher concentration than that required to induce Xbra. However, unlike the effect of activin induction, Xbra is not down-regulated by high Smad concentrations. Hence Smad2 appears insufficient to transduce the normal concentration response, a question at present under investigation. Future work must also investigate the variety and binding affinities of transcription factors for the immediate response genes, to see how a quantitative difference at the receptor level is converted into a qualitative difference in gene activation.
Figure 3. Ratio versus absolute model of receptor occupancy. If cells sense a ratio of bound to unbound receptors, receptor injection will require an increase in the number of activin molecules bound to elicit the same gene response. If cells sense absolute numbers of bound ligand, there would be no increase in ligand binding upon receptor over-expression. Based on ref 21.
A mechanism for morphogen gradient perception respectively. These values again correspond to 100 and 300 bound ligands per cell, even though the ratio of bound to unbound receptors has been greatly reduced. We therefore conclude that cells perceive the activin concentration by counting the absolute number of occupied receptors, and not by the ratio of occupied to unoccupied receptors. Consistent with published results for other signalling molecules, the ligand receptor interaction for activin was found to be of high affinity. This would explain our previously noted ratchet effect. By the use of bead replacement, we showed that cells can alter their response from a low to a high level of gene expression, but not vice versa.28 The fact that activin does not vacate its receptors quickly, but remains largely bound for a few hours, can explain why a cell responds to the highest activin concentration it is exposed to.
Traditionally, gradients have been discussed in terms of a source and a sink, and it is assumed that the cells only respond once the gradient has reached a steady state. We rather suggest that a dynamic gradient may be used in early development.23,27 According to this model, the responding cells continuously monitor the concentration gradient as it is formed and respond, using only a fraction of their receptors, to the highest concentration they perceive during their period of competence to the signal. Such a model has a number of advantages.23,27,31 It would enable the embryo to respond very rapidly to concentration changes, which correlates well with the rapid rate of early development. There need not be a mechanism to ensure that the cells do not respond prematurely to the gradient before it reaches equilibrium. Furthermore, titration of Type I receptors will not occur and so increased occupancy can cause a proportional increase in signal transduction. A pre-requisite for a concentration-dependent response is that ligand is limiting. However, in order for it to reach distant cells and not be sequestered by receptors, the ligand must be in excess. The knowledge that a cell needs only a handful of receptors to be occupied to generate a response might explain this apparent paradox. For example, most of the activin might be envisaged as passing through the intercellular space and, because only a small percent-
Downstream of ligand perception Our results show that a cell can use a single species of Type II receptor to perceive different concentrations of activin. Armes and Smith,29 by comparing two constitutively active Type I receptors, similarly suggest that only one, ALK-4, elicits the activin-induced concentration-dependent response. This contrasts with Dpp signal transduction in the Drosophila wing 315
N. McDowell and J. B. Gurdon
References
age is bound by receptors, cells can generate the concentration-dependent response without significantly reducing the surrounding ligand concentration and hence without disturbing the gradient.27,31
1. Dosch R, Gawantka V, Delius H, Blumenstock C, Niehrs C Ž1997. Bmp-4 acts as a morphogen in dorsoventral mesoderm patteming in Xenopus. Development 124:2325]2334 2. Wilson PA, Lagna G, Suzuki A, Hematti-Brivanlou A Ž1997. Concentration-dependent patterning of the Xenopus ectoderm by BMP-4 and its signal transducer Smad1. Development 124:3177]3184 3. Jones CM, Smith JC Ž1998. Establishment of a BMP-4 morphogen gradient by long-range inhibition. Dev Biol 194:12]17 4. Gurdon JB, Mitchell A, Ryan K Ž1996. An experimental system for analysing response to a morphogen gradient. Proc Natl Acad Sci USA 93:9334]9338 5. Neumann C, Cohen S Ž1997. Morphogens and pattern formation. Bioessays 19:721]729 6. Morgan TH Ž1897. Regeneration in Allolobophora foetida. Wilhelm Roux’s Arch 5:570]586 7. Wolpert L Ž1969. Positional information and the spatial pattern of cellular differentiation. J Theor Biol 25:1]47 8. Wolpert L Ž1989. Positional information revisited. Development ŽSuppl..:3]12 9. Slack JMW Ž1987. Morphogenetic gradients-past and present. Trends Biochem Sci 12:200]204 10. Slack JMW Ž1991. From Egg to Embryo, 2nd ed. Cambridge University Press. 11. Lawrence PA, Struhl G Ž1996. Morphogens, compartments, and pattern: lessons from Drosophila. Cell 85:951]61 12. Dale L Ž1997. Morphogen gradients and mesodermal patterning. Curr Biol 7:R698]R670 13. Lecuit T, Brook WJ, Calleja M, Sun H, Cohen SM Ž1996. Two distinct mechanisms for long-range patterning by Decapentaplegic in the Drosophila wing. Nature 381:387]39 3 14. Nellen D, Burke R, Struhl G, Basler K Ž1996. Direct and long-range action of a Dpp morphogen gradient. Cell 85:357]368 15. Singer MA, Penton A, Twombly V, Hoffmann FM, Gelbert WM Ž1997. Signalling through both Type 1 Dpp receptors is required for anterior]posterior patterning of the entire Drosophila wing. Development 124:79]89 16. Zecca M, Basler K, Struhl G Ž1996. Direct and long-range action of a wingless morphogen gradient. Cell 87:833]844 17. Neumann CJ, Cohen SM Ž1997. Long-range action of Wingless organises the dorsal]ventral axis of the Drosophila wing. Development 124:871]880 18. Struhl G, Barbash DA, Lawrence PA Ž1997. Hedgehog acts by distinct gradient and signal relay mechanisms to organise cell type and cell polarity in the Drosophila abdomen. Development 124:2155]2165 19. Struhl G, Barbash DA, Lawrence PA Ž1997. Hedgehog organises the pattern and polarity of epidennal cells in the Drosophila endoderm. Development 124:2143]2154 20. Green JBA, Smith JC Ž1990. Graded changes in dose of a Xenopus activin A homologue elicit stepwise transitions in embryonic cell fate. Nature 347:391]394 21. Green JB, New HV, Smith JC Ž1992. 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 22. Gurdon JB, Harger P, Mitchell A, Lemaire P Ž1994. Activin signalling and response to a morphogen gradient. Nature 371:487]492 23. McDowell N, Zorn AM, Crease DJ, Gurdon JB Ž1997. Activin has a direct long range signalling activity and can form a concentration gradient by diffusion. Curr Biol 7:671]681
An in vivo role for activin as a morphogen in mesoderm induction Activin protein has been shown to be present in Xenopus eggs 32 and the early embryo.33 Furthermore, the use of a dominant negative receptor specific for activin has suggested that activin has an essential role in early Xenopus development and in the induction of the mesoderm.34,35 Zhang et al 36 recently depleted the embryo of maternal VegT, a vegetally localised mRNA and this not only abolished endoderm formation, but also redirected the prospective endoderm cells towards a mesodermal fate, while cells normally fated to form mesoderm became ectoderm. A similar conversion of the fate map occurs when TGF-b-type signalling is disrupted by means of a dominant negative receptor.37 The latest model of mesoderm induction therefore suggests that it is biphasic, consisting of a weak maternal signal and a later strong VegT activated TGF-b-type signal.38 Activin can induce both mesoderm and endoderm and could thus play a role as either the VegT activated TGF-b-type signal, the maternal signal, or both. However, it is likely that in such a capacity activin would only specify broad domains of gene expression, creating a coarse regional pre-pattern of the mesoderm, with secondary processes, such as short-range signals or cell autonomous effects subsequently refining this pattern and creating more precise boundaries, e.g. Xgsc can bind the Xbra promoter to inhibit Xbra transcription.39 In addition, a subsequent BMP-4 activity gradient appears to further pattern the mesoderm along a ventral]dorsal axis.1 ] 3
Acknowledgements This work has been supported by the Cancer Research Campaign. NMcD is supported by a Magdalene College Manifold Research Fellowship.
316
Activin as a morphogen in Xenopus mesoderm induction
24. Reilly KM, Melton DA Ž1996. Short-range signaling by candidate morphogens of the TGF-b family and evidence for a relay mechanism of induction. Cell 86:743]754 25. Jones CM, Armes N, Smith JC Ž1996. Signalling by TGF-b family members: short-range effects of Xnr-2 and BMP-4 contrast with the long-range effects of activin. Curr Biol 6:1468]1675 26. Kerszberg M, Wolpert L Ž1998. Mechanisms for positional signalling by morphogen transport: A theoretical study. J Theor Biol 191:103]114 27. Dyson S, Gurdon JB Ž1998. The interpretation of position in a morphogen gradient as revealed by occupancy of activin receptors. Cell 93:557]568 28. Gurdon JB, Mitchell A, Mahony D Ž1995. Direct and continuous assessment by cells of their position in a morphogen gradient. Nature 376:520]521 29. Armes NA, Smith JC Ž1997. The ALK-2 and ALK-4 activin receptors transduce distinct mesoderm-inducing signals during early development but do not cooperate to establish thresholds. Development 124:3797]3804 30. Graff JM, Bansal A, Melton DA Ž1996. Xenopus Mad proteins transduce distinct subsets of signals for the TGF-b superfamily. Cell 85:479]487 31. Gurdon JB, Dyson S, St Johnson D Ž1998. Cells’ perception of position in a concentration gradient. Cell 95:159]162 32. Asashima M, Nakano H, Uchiyama H, Sugino H, Nakamura
33.
34.
35. 36.
37.
38. 39.
317
T, Eto Y, Ejima D, Nishimatsu S-I, Ueno N, Kinoshita K Ž1991. Presence of activin Žerythroid differentiation factor. in unfertilized eggs and blastulae of Xenopus laevis. Proc Natl Acad Sci USA 88:6511]6514 Fukui A, Nakamura T, Uchiyama H, Sugino K, Sugino H, Asashima M Ž1994. Identification of activins A, AB, and B and follistatin proteins in Xenopus embryos. Develop Biol 163:279]281 Schulte-Merker S, Smith JC, Dale L Ž1994. Effects of truncated activin and FGF receptors and of follistatin on the inducing activities of BVg1 and activin: does activin play a role in mesoderm induction? EMBO J 13:3533]3541 Dyson S, Gurdon JB Ž1997. Activin signalling has a necessary function in Xenopus early development. Curr Biol 7:81]84 Zhang J, Houton DW, King ML, Payne C, Wylie C, Heasman J Ž1998. The role of maternal VegT in establishing the primary germ layers in Xenopus embryos. Cell 94:515]524 Henry GL, Brivanlou IH, Kesler DS, Hematti-Brivanlou A, Melton DA Ž1996. TGF-b signals and a prepattern in Xenopus laevis endodermal development. Development 122:1007]1015 Kimelman D, Griffin KJP Ž1998. Mesoderm induction: a postmodern view. Cell 94:419]421 Artinger M, Blitz I, Inoue K, Tran U, Cho KW Ž1997. Interaction of goosecoid and brachyury in Xenopus mesoderm patterning. Mech Dev 65:187]196
Activin as a morphogen in Xenopus mesoderm induction Natasha McDowell and J. B. Gurdon
Activin, a member of the Transforming Growth Factor beta (TGF-b ) superfamily, can behave as a morphogen in cells of the early Xenopus embryo by inducing a range of mesodermal genes in a concentration-dependent manner. This review examines the behaviour of activin as it forms a morphogen gradient. It also discusses how a cell can perceive its position in a concentration gradient in order to activate appropriate mesodermal gene responses.
However, often in development, the long-range signal appears to act as a morphogen. Classically defined, this is a molecule which, when secreted by cells at one location in an embryo, diffuses away from its source to create a concentration gradient across the surrounding tissue. Thus cells near the source receive a high concentration of signal and follow one pathway of development, whilst cells further away receive a lower concentration of signal and adopt a different cell fate. Hence, an individual cell must not only assess whether signalling molecules are present in its local environment, but it must also quantify the amount of signal it perceives in order to elicit a concentration-dependent response. Furthermore, although only one type of molecule is involved, such a mechanism can generate a diversity of cell-types in a defined spatial relationship to each other. In contrast to this classical definition, a concentration-dependent response might also be established by an antagonising molecule modulating the activity of the inducing molecule in a graded fashion.1] 3 In such a case, the inducing molecule could be locally produced over the entire responding field, but possess a gradient of activity due to the long-range diffusion of inhibitor from a localised source 2,3 ŽDale and Wardle, this issue..
Key words: activin r morphogen r mesoderm induction r Xenopus Q1999 Academic Press
AS THE EMBRYO DEVELOPS, it divides from a single cell, the fertilised egg, into a complex organism with an array of tissue types and an organised body plan. Hence, not only does cellular diversity increase, but the cells must also become spatially arranged in relation to one another, a process termed pattern formation. How is this diversity and pattern formation achieved? In very early development, coarse spatial cues are often established by the use of cytoplasmic maternal determinants, proteins or mRNAs which are asymmetrically localised within the fertilised egg and early embryo. The other major mechanism is that of cell-to-cell signalling. While short-range interactions require direct contact between inducing and responding cells, long-range signalling molecules include proteins capable of passing across many cell diameters. In both cases, the response can ‘theoretically’ be a simple binary decision; provided the responding cells perceive sufficient signal, their development will be diverted to a new cell fate.
What makes a molecule a morphogen? Most importantly, a morphogen must induce differ ent cell fates at different concentrations. However, it must be further shown that the morphogen can elicit at least two qualitatively different responses, apart from the nil response, from each responding cell. Otherwise the concentration-dependent patterning could simply be due to the responding cells being a heterogeneous population, with some cells capable of responding to only a high signal concentration and others to only a low concentration.4 The response to the putative morphogen must also be demonstrated to be direct; if indirect, the concentration-dependent effect would apply to a different
From the Wellcomer CRC Institute of Cancer and Developmental Biology, Tennis Court Road, Cambridge CB2 1QR, UK and Department of Zoology, University of Cambridge, Cambridge, England, UK Q1999 Academic Press 1084-9521r 99 r 030311q 07 $30.00r 0
311
N. McDowell and J. B. Gurdon
molecule.5 For example, it must be shown that the signal does not simply act at short-range to activate a second long-range signal, or alternatively that it does not initiate a relay of local short-range interactions, in which each subsequent cell in the relay releases a different inducing molecule in addition to adopting a particular cell fate.
internal organs. The traditional model of mesoderm induction is that maternal morphogen-like signals emanate from the vegetal pole Žprospective endoderm. during the early cleavage stages to induce the mesoderm from the overlying animal cap cells Žprospective ectoderm..10 We have used the induction of mesoderm by activin as a model system to analyse the behaviour of a morphogen during vertebrate development and how a cell perceives its position in a gradient.
Putative morphogens Since 1897, when Thomas Hunt Morgan6 first used the concept of a gradient of ‘stuff’ to explain how polarity and pattern formation during regeneration may be controlled, there has been an extensive history of theoretical discussion and experimental contribution concerning the concept of a morphogen gradient. Many of the ideas have been summarised in a number of reviews.5,7 ] 12 In Drosophila the most convincing morphogens are Wingless ŽWg. and Decapentaplegic ŽDpp. in patterning the wing ŽStrigini and Cohen, this issue., and Hedgehog ŽHh. in patterning the adult abdominal segments. All three have been shown to directly activate downstream genes in a concentration-dependent manner.13 ] 19 In vertebrates, a lot of work has focused on activin, a member of the Transforming Growth Factor Beta ŽTGF-b . family, and on its role as a morphogen in the induction of the mesoderm in the frog, Xenopus laevis. The mesoderm is the middle germ layer of the body and later gives rise to such diverse tissue types as notochord, muscle, blood and parts of nearly all
Is activin a morphogen? Does it have a concentration-dependent response? By dissociating and then incubating blastula animal cap cells in various activin concentrations, Green and Smith20 and Green et al.21 demonstrated that the higher the concentration of activin used, the more dorsal the induced mesodermal cell fate. However, the use of dissociated cells cannot establish whether each individual cell has the potential to make a range of responses. We therefore sandwiched intact blastula tissue around activin loaded beads to create a directional source of putative morphogen and assayed gene response by in situ hybridisation. While low activin concentrations induced low response genes such as Xbrachury (Xbra) and Xpo adjacent to the source, high concentration beads induced Xbra and Xpo in more distant cells, the intervening cells expressing the high response genes Goosecoid (Xgsc),
Figure 1. A concentration-dependent wave of gene-expression. Animal cap tissue is sandwiched around either low or high concentration activin beads and gene response assayed by in situ hybridisation. Continuous regions of gene expression are observed under these conditions in which there is no cell movement. Based on ref 17.
312
Activin as a morphogen in Xenopus mesoderm induction
Eomesodermin, (Eomes), Antipodean and Mix122,4 ŽFigure 1.. The existence of such a concentration-dependent wave of gene expression, as well as the observation that labelled and unlabelled cells were never interspersed with each other in the same region, suggests that each individual cell can elicit either a high, low or nil response.4 We have also used oocytesynthesised 35 S-labelled activin to provide visual evidence Žto a distance of approx. seven cell diameters. that activin can form a concentration gradient.23
a non-activin second signal would presumably have induced a response in these distant cells unable to respond to activin. In a related study, Reilly and Melton24 conclude that secondary signalling does take place. The discrepancy between the two conclusions is presumably because we observe immediate responses ŽStage 10.5. to the putative morphogen activin, whereas they do not analyse response genes until Stage 35. During this time period, it is likely that many secondary signalling events will have occurred.
Can activin act directly and at a distance?
Diffusion or relay?
By using lineage-labelled layers of responding cells and cytochalasin, we showed that neither cell movement or cell division can account for the long-range effects of activin.22 However, this does not tell us whether activin acts directly on distant cells, or through the intermediary of another signalling molecule. We have therefore used a dominant-negative receptor specific to activin to show that cells distant from the activin source require functional activin receptors to elicit a response.23 This indicates that the long-range effects of activin are direct, since
As an extension of this work, we were interested in whether signal passage across tissue occurs by diffusion or relay ŽFigure 2a,b.. The fact that activin itself is required by all the responsive cells does not preclude the possibility that a relay of activin signalling is involved, activin being released, at an attenuated level, by each step in the relay series. Initially we showed that activin can pass through reaggregated tadpole endoderm which had been cycloheximide treated so as to prevent protein synthesis and thus the production of new activin.22 However, this does not
Figure 2. Possible mechanisms of signal passage. Ža. Activin molecules diffuse in the extracellular-matrix surrounding cells to create a concentration gradient across the tissue. Žb. A relay of activin signalling from cell to cell; each subsequent cell transmits less activin than the previous one. Žc. Transfer-mediated movement; activin might bind with low affinity to the protein core of betaglycan to aid its passage across tissue.
313
N. McDowell and J. B. Gurdon
examine the situation in intact Žnon-dissociated. blastula tissue, the stage at which mesoderm induction occurs in vivo. Three lines of evidence argue against a relay mechanism. Jones et al.25 used a constitutively active activin receptor to show that, in the absence of ligand, response genes are autonomously induced; if a relay was operative, non-autonomous induction would be expected. Our observation of exogenous 35 Slabelled activin molecules in cells distant from the activin bead source also demonstrates that a relay of activin need not occur.23 In addition, we have shown that the closely-related, but non-endogenous signal, TGF-b2, can be transmitted across intact blastula stage tissue that cannot respond to or synthesise it and can then induce mesodermal genes in distant responsive celIs.23 TFG-b 2 was used since blastula cells have to be injected with the Type II TGF-b receptor in order to respond to it and so normal wild-type tissue can be used as the non-responsive intervening cell layer.23,24 In contrast, Reilly and Melton24 advocate a process of relay for TGF-b molecules in Xenopus since they did not observe such mesodermal gene induction in a clone of distant responsive cells. We believe that the discrepancy can be explained by their use of TGF-b 1 mRNA rather than protein as the source of ligand. In our hands and unlike the case for activin-mRNA, cells containing TGF-b 1 mRNA cannot signal over a longrange, perhaps due to inefficient processingrsecretion of a protein that is a non-endogenous molecule.23 Although it appears that a relay mechanism is not operative, it is still an open question as to whether activin diffuses through or between cells. If the latter occurs, then it must also be asked whether the morphogen diffuses passively through the extracellular matrix or whether diffusion is aided. For example, a transfer-mediated mechanism could be envisaged whereby the signal is propelled along cell membranes, as well as passed from cell to cell, by binding to low affinity proteoglycan receptors on the cell membrane ŽFigure 2c; see also ref 26. ŽKersberg et al; Pfeiffler and Vincent, this issue..
cell to activate different signal transduction pathways in response to different concentrations of the morphogen. For example, a cell might possess receptors of differing affinities: Only high affinity receptors would be bound by low ligand concentrations, thereby activating a signal transduction pathway leading to the expression of low response genes. At high ligand concentrations low affinity receptors would also be bound, thus activating the high response pathway as well. Alternatively, the various concentrations of the morphogen might activate the same signal transduction pathway but to differing degrees. We have used 35 S-activin to determine how cells use their receptors to perceive a change in morphogen concentration.27 We initially investigated the binding of 35 S-activin to cell surface receptors in cells overexpressing the activin Type II B receptor Ž ActRIIB .. This was then compared to ligand binding in normal cells. As might be expected, in both cases an increase in activin concentration leads to an increase in receptor occupancy and causes a corresponding switch from Xbra to Xgsc expression. Interestingly though, Scatchard analysis showed that in both the injected cells and in the wild-type cells a single type of receptor is bound; in wild-type cells, this receptor species had a similar affinity to the introduced ActRIIB in the overexpressing cells. Furthermore, by determining the amount of activin required to saturate all a cell’s available receptors, we have shown that only 2% receptor occupancy of this receptor species is required to switch cells from a nil to a low Ž Xbra. gene response, while 6% receptor occupancy is necessary to induce a high response Ž Xgsc .. Hence, it appears that the activin morphogen response is elicited by a differing activation of the same signal transduction pathway. This leads to the question of what is the absolute number of activin molecules required to induce Xbra and Xgsc? From the values for saturation binding, as well as from the specific activity of the activin preparation, we calculated that wild-type cells have approximately 5000 activin binding receptors per cell. Hence, Xbra is induced when 100 activin molecules are bound, while 300 molecules are needed to induce Xgsc. We then asked whether cells detect an absolute rather than relative occupancy of receptors. To do this we compared the receptor occupancy at which genes are activated in injected versus uninjected cells ŽFigure 3b.. The injection of 1 ng ActRIIB-mRNA, which corresponds to approximately 38,000 receptors per cell, showed Xbra and Xgsc to be induced at 0.3% and 0.8% receptor occupancy,
How does a cell perceive its position in a morphogen gradient? For a cell to respond to a morphogen gradient, it must convert a quantitative extracellular signal concentration into a qualitative pattern of gene expression in the nucleus. One mechanism would be for the 314
Activin as a morphogen in Xenopus mesoderm induction
imaginal disc, where it appears that although signal transduction involves a quantitative difference in the same transduction pathway, two Type 1 receptors participate.15 To transduce the activin signal, Smad2 and Smad4 are thought to be phosphorylated by the Type 1 receptor and move from cytoplasm to nucleus. Graff et al 30 have shown that overexpression of Smad2 causes the induction of Xgsc at a higher concentration than that required to induce Xbra. However, unlike the effect of activin induction, Xbra is not down-regulated by high Smad concentrations. Hence Smad2 appears insufficient to transduce the normal concentration response, a question at present under investigation. Future work must also investigate the variety and binding affinities of transcription factors for the immediate response genes, to see how a quantitative difference at the receptor level is converted into a qualitative difference in gene activation.
Figure 3. Ratio versus absolute model of receptor occupancy. If cells sense a ratio of bound to unbound receptors, receptor injection will require an increase in the number of activin molecules bound to elicit the same gene response. If cells sense absolute numbers of bound ligand, there would be no increase in ligand binding upon receptor over-expression. Based on ref 21.
A mechanism for morphogen gradient perception respectively. These values again correspond to 100 and 300 bound ligands per cell, even though the ratio of bound to unbound receptors has been greatly reduced. We therefore conclude that cells perceive the activin concentration by counting the absolute number of occupied receptors, and not by the ratio of occupied to unoccupied receptors. Consistent with published results for other signalling molecules, the ligand receptor interaction for activin was found to be of high affinity. This would explain our previously noted ratchet effect. By the use of bead replacement, we showed that cells can alter their response from a low to a high level of gene expression, but not vice versa.28 The fact that activin does not vacate its receptors quickly, but remains largely bound for a few hours, can explain why a cell responds to the highest activin concentration it is exposed to.
Traditionally, gradients have been discussed in terms of a source and a sink, and it is assumed that the cells only respond once the gradient has reached a steady state. We rather suggest that a dynamic gradient may be used in early development.23,27 According to this model, the responding cells continuously monitor the concentration gradient as it is formed and respond, using only a fraction of their receptors, to the highest concentration they perceive during their period of competence to the signal. Such a model has a number of advantages.23,27,31 It would enable the embryo to respond very rapidly to concentration changes, which correlates well with the rapid rate of early development. There need not be a mechanism to ensure that the cells do not respond prematurely to the gradient before it reaches equilibrium. Furthermore, titration of Type I receptors will not occur and so increased occupancy can cause a proportional increase in signal transduction. A pre-requisite for a concentration-dependent response is that ligand is limiting. However, in order for it to reach distant cells and not be sequestered by receptors, the ligand must be in excess. The knowledge that a cell needs only a handful of receptors to be occupied to generate a response might explain this apparent paradox. For example, most of the activin might be envisaged as passing through the intercellular space and, because only a small percent-
Downstream of ligand perception Our results show that a cell can use a single species of Type II receptor to perceive different concentrations of activin. Armes and Smith,29 by comparing two constitutively active Type I receptors, similarly suggest that only one, ALK-4, elicits the activin-induced concentration-dependent response. This contrasts with Dpp signal transduction in the Drosophila wing 315
N. McDowell and J. B. Gurdon
References
age is bound by receptors, cells can generate the concentration-dependent response without significantly reducing the surrounding ligand concentration and hence without disturbing the gradient.27,31
1. Dosch R, Gawantka V, Delius H, Blumenstock C, Niehrs C Ž1997. Bmp-4 acts as a morphogen in dorsoventral mesoderm patteming in Xenopus. Development 124:2325]2334 2. Wilson PA, Lagna G, Suzuki A, Hematti-Brivanlou A Ž1997. Concentration-dependent patterning of the Xenopus ectoderm by BMP-4 and its signal transducer Smad1. Development 124:3177]3184 3. Jones CM, Smith JC Ž1998. Establishment of a BMP-4 morphogen gradient by long-range inhibition. Dev Biol 194:12]17 4. Gurdon JB, Mitchell A, Ryan K Ž1996. An experimental system for analysing response to a morphogen gradient. Proc Natl Acad Sci USA 93:9334]9338 5. Neumann C, Cohen S Ž1997. Morphogens and pattern formation. Bioessays 19:721]729 6. Morgan TH Ž1897. Regeneration in Allolobophora foetida. Wilhelm Roux’s Arch 5:570]586 7. Wolpert L Ž1969. Positional information and the spatial pattern of cellular differentiation. J Theor Biol 25:1]47 8. Wolpert L Ž1989. Positional information revisited. Development ŽSuppl..:3]12 9. Slack JMW Ž1987. Morphogenetic gradients-past and present. Trends Biochem Sci 12:200]204 10. Slack JMW Ž1991. From Egg to Embryo, 2nd ed. Cambridge University Press. 11. Lawrence PA, Struhl G Ž1996. Morphogens, compartments, and pattern: lessons from Drosophila. Cell 85:951]61 12. Dale L Ž1997. Morphogen gradients and mesodermal patterning. Curr Biol 7:R698]R670 13. Lecuit T, Brook WJ, Calleja M, Sun H, Cohen SM Ž1996. Two distinct mechanisms for long-range patterning by Decapentaplegic in the Drosophila wing. Nature 381:387]39 3 14. Nellen D, Burke R, Struhl G, Basler K Ž1996. Direct and long-range action of a Dpp morphogen gradient. Cell 85:357]368 15. Singer MA, Penton A, Twombly V, Hoffmann FM, Gelbert WM Ž1997. Signalling through both Type 1 Dpp receptors is required for anterior]posterior patterning of the entire Drosophila wing. Development 124:79]89 16. Zecca M, Basler K, Struhl G Ž1996. Direct and long-range action of a wingless morphogen gradient. Cell 87:833]844 17. Neumann CJ, Cohen SM Ž1997. Long-range action of Wingless organises the dorsal]ventral axis of the Drosophila wing. Development 124:871]880 18. Struhl G, Barbash DA, Lawrence PA Ž1997. Hedgehog acts by distinct gradient and signal relay mechanisms to organise cell type and cell polarity in the Drosophila abdomen. Development 124:2155]2165 19. Struhl G, Barbash DA, Lawrence PA Ž1997. Hedgehog organises the pattern and polarity of epidennal cells in the Drosophila endoderm. Development 124:2143]2154 20. Green JBA, Smith JC Ž1990. Graded changes in dose of a Xenopus activin A homologue elicit stepwise transitions in embryonic cell fate. Nature 347:391]394 21. Green JB, New HV, Smith JC Ž1992. 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 22. Gurdon JB, Harger P, Mitchell A, Lemaire P Ž1994. Activin signalling and response to a morphogen gradient. Nature 371:487]492 23. McDowell N, Zorn AM, Crease DJ, Gurdon JB Ž1997. Activin has a direct long range signalling activity and can form a concentration gradient by diffusion. Curr Biol 7:671]681
An in vivo role for activin as a morphogen in mesoderm induction Activin protein has been shown to be present in Xenopus eggs 32 and the early embryo.33 Furthermore, the use of a dominant negative receptor specific for activin has suggested that activin has an essential role in early Xenopus development and in the induction of the mesoderm.34,35 Zhang et al 36 recently depleted the embryo of maternal VegT, a vegetally localised mRNA and this not only abolished endoderm formation, but also redirected the prospective endoderm cells towards a mesodermal fate, while cells normally fated to form mesoderm became ectoderm. A similar conversion of the fate map occurs when TGF-b-type signalling is disrupted by means of a dominant negative receptor.37 The latest model of mesoderm induction therefore suggests that it is biphasic, consisting of a weak maternal signal and a later strong VegT activated TGF-b-type signal.38 Activin can induce both mesoderm and endoderm and could thus play a role as either the VegT activated TGF-b-type signal, the maternal signal, or both. However, it is likely that in such a capacity activin would only specify broad domains of gene expression, creating a coarse regional pre-pattern of the mesoderm, with secondary processes, such as short-range signals or cell autonomous effects subsequently refining this pattern and creating more precise boundaries, e.g. Xgsc can bind the Xbra promoter to inhibit Xbra transcription.39 In addition, a subsequent BMP-4 activity gradient appears to further pattern the mesoderm along a ventral]dorsal axis.1 ] 3
Acknowledgements This work has been supported by the Cancer Research Campaign. NMcD is supported by a Magdalene College Manifold Research Fellowship.
316
Activin as a morphogen in Xenopus mesoderm induction
24. Reilly KM, Melton DA Ž1996. Short-range signaling by candidate morphogens of the TGF-b family and evidence for a relay mechanism of induction. Cell 86:743]754 25. Jones CM, Armes N, Smith JC Ž1996. Signalling by TGF-b family members: short-range effects of Xnr-2 and BMP-4 contrast with the long-range effects of activin. Curr Biol 6:1468]1675 26. Kerszberg M, Wolpert L Ž1998. Mechanisms for positional signalling by morphogen transport: A theoretical study. J Theor Biol 191:103]114 27. Dyson S, Gurdon JB Ž1998. The interpretation of position in a morphogen gradient as revealed by occupancy of activin receptors. Cell 93:557]568 28. Gurdon JB, Mitchell A, Mahony D Ž1995. Direct and continuous assessment by cells of their position in a morphogen gradient. Nature 376:520]521 29. Armes NA, Smith JC Ž1997. The ALK-2 and ALK-4 activin receptors transduce distinct mesoderm-inducing signals during early development but do not cooperate to establish thresholds. Development 124:3797]3804 30. Graff JM, Bansal A, Melton DA Ž1996. Xenopus Mad proteins transduce distinct subsets of signals for the TGF-b superfamily. Cell 85:479]487 31. Gurdon JB, Dyson S, St Johnson D Ž1998. Cells’ perception of position in a concentration gradient. Cell 95:159]162 32. Asashima M, Nakano H, Uchiyama H, Sugino H, Nakamura
33.
34.
35. 36.
37.
38. 39.
317
T, Eto Y, Ejima D, Nishimatsu S-I, Ueno N, Kinoshita K Ž1991. Presence of activin Žerythroid differentiation factor. in unfertilized eggs and blastulae of Xenopus laevis. Proc Natl Acad Sci USA 88:6511]6514 Fukui A, Nakamura T, Uchiyama H, Sugino K, Sugino H, Asashima M Ž1994. Identification of activins A, AB, and B and follistatin proteins in Xenopus embryos. Develop Biol 163:279]281 Schulte-Merker S, Smith JC, Dale L Ž1994. Effects of truncated activin and FGF receptors and of follistatin on the inducing activities of BVg1 and activin: does activin play a role in mesoderm induction? EMBO J 13:3533]3541 Dyson S, Gurdon JB Ž1997. Activin signalling has a necessary function in Xenopus early development. Curr Biol 7:81]84 Zhang J, Houton DW, King ML, Payne C, Wylie C, Heasman J Ž1998. The role of maternal VegT in establishing the primary germ layers in Xenopus embryos. Cell 94:515]524 Henry GL, Brivanlou IH, Kesler DS, Hematti-Brivanlou A, Melton DA Ž1996. TGF-b signals and a prepattern in Xenopus laevis endodermal development. Development 122:1007]1015 Kimelman D, Griffin KJP Ž1998. Mesoderm induction: a postmodern view. Cell 94:419]421 Artinger M, Blitz I, Inoue K, Tran U, Cho KW Ž1997. Interaction of goosecoid and brachyury in Xenopus mesoderm patterning. Mech Dev 65:187]196