A strong case can be made for arguing that the hardest but most important task for the developmental biologist is explaining the mcchanisrns of gastrulation, the series of events that turns a ball of cells into a primitive embryo with an overt body plan. No matter which organism one considers, the process involves the setting up of a prepattern in the blastula, differential gene cxprcssion, t h e coordinated activity of spccific groups of cells and much of the morphogenetic repertoire of the cpithclium, in addition to which, there is often coordinated mesenchymal movement. Gastrulation thus has a dual importaiice: not only is it the key event in development, but explanations for the phenomena underpinning it would help to answer almost every other developmental question. The problems of working with gastrulating embryos arc, however, as hard as they are important: embryos are small when they gastrulate, they are usually opaque and there tends to be too much going on in too short a timc for thc development biologist to dissect out particular aspects for study. Even when we think we know what is going on, gastrulation can hold surprises. Consider progress in the best understood case, the sea urchin, where a transparent ball of cells forms an intcrnal tubc, the gut. In the sixties, Gustavsoii and Wolpert(’) showed that this gut formed when the basal cells of the blastula invaginated and the most inward of these cells then put out long processes which extended to reach and adherc to the inner, opposite sui-face of the blastula where they then contracted. At the same time, the invagination extended t o form a tube which met the anterior surface where the processcs adhered. It was clearly sensible to hypothesise that the initial curvature was due to a change in cell-surfacc adhesivity and that the contractions of the active processes pulled the basal epithelium out into a tube. Even though this picture of gastrulation is plausible, much of it turns out to be wrong as Hardin demonstrated by showing that the tube would form even if the filopodia were removed by laser ablation and that the gut would extend outwards rather than inwards if the embryo was cultured in litliium(2,3). Other mechanisms were clearly involved in elongating t h c tube and we now know that both aspects of sea urchin gastrulation are examples of cell ueurumgcment, a general phenomenon deriving from actin-myosin inter-
actions(4) and which causes epithelia to display autonomous extensive bchaviour (SCC 4 and 5 for reviews). The inechanism was first identified by the Fristroms in their study of the everting Drosophilcr leg disc(‘,’) where the constituent cells seem to rearrange their nearest neighbours to effect a process that seems analogous to the opening of a telescope. More recently, they have shown that morphogenesis is preceded by a flattening and elongation of the participating cells(’) and is thus a two-stage process. Cell rearrangement has now been documented in vertebrate as well as invertebrate development. It accounts for FunduZus epibolyC9)where the bcginnings of a theoretical analysis of the mechanism have been made(’”) and Xeizopus blastula morphogenesis(”) and may also Flap a role in cell thinning in heart formation(’-) hydra regeneration(’’) and various other phenomena[l4). It is also one of the driving forces of amphibian gastrulation where the phenomenon is demonstrated by the extensive buckling shown b the ectoderm when mesoderm involution is inhibited(”). It is now clear that autonomous cell rearrangement is the key mechanism in controlling many examples of epithelial morphogenesis. We are however, short of accessible embryos in which to study thc process and it is therefore a pleasure to welcome to the canon a set of papers on Drosophila gastrulation, particularly because the pictures are so good. Two of these papers are from Leptin’s laboratory and investigate the phenomenon in sectioned(“) and in living(”) material and the other is from Wieschaus’ group; this paper not oiily details SEM observations on Lormal and mutant embryos(”), but also includes information on the posterior midgut invagination. Although they usc differcnt techniqucs. the two laboratories provide very similar views of the phenomenon. As Leptin’s group can claim temporal precedence, I shall initially concentrate on its work and refer later to the observations on the mutants and the foregut invagination of the Wieschaus laboratory as they help to clarify a problem highlighted by thc morphology. The study of both sectioned and SEM material show that, over a 20 min period, the ventral epithelium of this relatively transparent embryo develops a furrow whose internalised cells become mesenchymal in nature simple: by the standards of gastrulation. It might be thought surprising that no one had previously looked at this behaviour in the living embryo, but the sophisticated time-lapse techniques rcquired mercly to describe the dynamics at thc cellular level(17)demonstrate why it would have been hard to undertake the work even five years ago. The classic way to study cell behaviour in vivo involves the biologist immobilising an embryo or organ under a microscope and then, with the use of appropriate optics and frequent changes of focus, taking regular pictures which then slowly have to be analysed. l i f c is easier now! Kam el uL.(~’) injected ~
fluorescein- and rhodamine-labelled markers into Ihe embryos, the one for nuclei and the othcr for highlighting cell borders. Pairs of photographs for the two fluorochromes were then takcn every minute at 10 planes of focus 3pm apart until gastrulation was complete. the whole shooting match being driven by a computer work station. The authors then used the computer to reconstruct the distinct dynamic behaviours of the participating cells and their nuclei. It turns out that furrow formation involves two distinct processes which take place in the central region of the ventral epithelium. The first is the migration of about half of these nuclei from their apical to their basal surface with the apical end of the ccll then narrowing. The second process involves the rapid inward movement of the remainder of these nuclei and, as they move, the epithelium seems to collapse inwards and peripheral cells fold over to form the furrow. Four obvious mechanisms could. in principlc at least, generate this behaviour. Local changes of intercellular adhesivity in the potential folding domain might cause the epithelium to bend and then invaginate('); a propagating wave of local contraction could lcad to foldingc1'); the phenomenon might have a simple mechanical explanation with the cells whose nuclei had moved basally collapsing inwards so causing the peripheral cells to fold around them; or, fourth, we might have another example of cell rearrangement. As the evidence shows neither changes in curvature nor a wave of contractions, both studies exclude the first two possibilities, while the speed of the event and direct observation rule out growth or cell division as a participating mechaniqm. In the original paper, the morphological evidence led Leptin and Grunewald to conclude that furrow formation was caused by an internal reorganisation of the epithelium analogous to cell rearrangement. The evidence from the living organism seems to make Leptin and her co-workers draw back from thi5 conclusion as the data are compatible with folding deriving from a mechanical instability caused by the movement of nuclei. Were this so, the authors suggest that the global behaviour of the epithclium is no morc than the sum of the behaviours of its individual cells and that there is no nced to postulate coordinated behaviour beyond, of course, the specification of a domain where nuclear movement takes place. So the reader is left hanging, wondering which of these alternatives is correct or indeed whether they are quite as mutually exclusive as they seem at first sight. For the recent work €rom the Fristrom laboratory@) now makes it likely that cell rearrangement is a twostage process where overt epithelial shape change i s preceded by intracellular events that will then drive that rearrangement. Furrow formation in Drosophilu gastrulation is also a two-stage proccss with the first event being an intracellular change in the position of the nucleus and the second overt morphogenesis. It thus fits
with our expectation of cell rearrangement, but the actual mechanism of furrow formation remains unclear. One possibility is certainly mechanical collapse while another is organised epithelial movement. The former would derive pa.isively from the intracellular reorganisation while the latter would be the active result. Although Kam el a/.(I7) do not commit themselves, their evidence inclines me to favour the latter alternative as the form of the developing furrow when it invaginates into the fluid-filled blastocoel seeins too well-organised and to require too much energy to derive from mechanical instability (the folds that form by buckling in the anterior retinal epithelium of ciliary body of the eye are much less well-organised(2"). Moreover. the morphology is very similar to the early stages of sea-urchin gastrulation, albeit that the fold there is two-dimensional rather than one. Nevertheless, the evidence is still inadequate to discriminate betwcen the options. It is here that the observations of the Wieschaus group come into their own because they provide several new lines of evidence suggesting that thc second stage of furrow formation is a programmed event rather than the mechanical consequences of the initial nuclear movement. They first show that the second stage is accompanied by the shortening of the elongated cells and argue that this is an active event that helps form the furrow rather than a passive consequence of it. The second line comes from the behaviour of the mutants con( ertina (now cloncd)(21) and folded gastrulation: both fail to complete gastrulation, even though the ventral cells in both mutants undergo the preliminary elongation and apical constriction that is the first step in the process. This evidence argues that furrow formation is an active event rather than a passive response. a view supported by the fact that the concertina gene encodes part of a G protein and hence is likely to play a ccllcommunication('') rather than a mechanical role. The analysis of posterior midgut invagination adds weight to this vie": in the mutants. the cells similarly undergo their initial constriction and again the invagination fails to form. Here, however. the complexity of the normal invagination makes it very hard to argue for a morphogenetic mechanism based on mechanical collapse. It thus seems that furrow formation is another example where gastrulation is underpinned by cell rearrangement. If so, then the phenomenon may turn out to be unexpectedly uscful: epithelial rearrangement is probably the key rnorphogenetic movement in early embryogenesis and we need a model system in which to analyse it. As the Drowphilu blastula is accessible, transparent and undergoes a gastrulation in which nothing seems to happen except epithelial folding, these two laboratories may have the organism, the phenomenon and the technology which will allow the process of cell rearrangement to yield some of its secrets.
References 1 GCSTAVSUN,T . .AKU WOLPIRT,L. (1962). Cellular mechanisms in the morphogenesis of the sea urchin larva. Changes in the 3hape of cell sheets. Exp
Cell Res. 21. 260-279. 2 HARDIN, J. D. (1988). The role of secondary mesenchyme cells during sea
urchin gavtrulatioii studicd by lascr ablation. Developnienr 103, 317-324. 3 HARDIN. J. D. NO CHFNG.L. Y . (1986). The nicchanisni and niechanicc of archenteron elongalion in the sea urchin embryo. I ) w R i d . 115, 490-501. 4 FRISTR~M. D. (1988). The cell~clarbasis of epithelial rnnrphogeneris. Ti537re R Cc4 20. 645 690. 5 UARD,J . H . L. (19Y@). Morphogenois. I'iw Cellular and Molecuinr Processes of Drrchpmenml A n a t o m y . Cambridge Linivrrsity Press, Cambridge. 6 PLKL'IL, b.. FRISTROX, U . , KISS, 1. A N D FRlSTROhl, J . W . (1975). The mechaniam of evagination of studiea on trypsin-accelerated ebaginatwn. Wdh. Roux' Arch. 178, 123-138. 7 FKIS~ROM. D. (1976). The mechanism of evagination of imaginal discs of Drosopidu ntelunogusrer. 111. Evidence for cell rearrangement. Dev. Biol. 54, 163-171. 8 COXDIC, M. L ., FKISIKO.M, D. AND FRISTROM. J. W. (1991). Apical cell 4iape changes during Drosopkilu imaginal lcg disc elongation: a iiovcl morphogenetic mechanism. Development 111. 23-31. 9 KELLER.R. E. AXD TRIKKAUS. .I. P. (1987). Rcarrangcrncnt of enveloping cclls without disruption of the epithelial permeability harrier as a factor in Fiindulw spiboly. Dev. Bid 120. 12-24. 10 WELIKY.6'.AND OSTBK,G . (1990). 'The niechanical basis of cell rearrangement. I. Epithelial morphogenesis during Fundulus epiboly. Developinent 109, 373-386. 11 KEILFR, R. E. (19S6). The cellular hacis nf amphibian gastrulation. In De~elopmentnl hiology, u rvmzpr.chmsivr ~ynthrxis.II. Tkc r.e/lirlnr basis of morpkog.rxesi.r.cd. L. W. Rrowder, pp. 241-327. New York: Plenum Press. 12 XfANASEK; F. J., BL-RNSIUb, M. B. AND WAILK~UN, R. E. (1972). hlyocardial cell shape change as a mechanism of embryonic heart looping. Dev. Uiol. 29, 349-371.
h
13 Born, P. M. AND BUD^^, H . R. (1984). Formation of pattern in regenerating tissue pieces of HyiIrn utlenuutu. 111. 'The shaping of the body column. Dev. Bid. 106. 315-325. 14 KELLER.R. AND IiAKDm. J. (1987). Cell behaviour during active cell rearrangement: evidence and speculation. J . Cell Sci., Suppl. 8 . 369-393. 15 BOlTrAIJT. J. c..DARRIRERE. T.. PoorF, T. J. A O Y A MH., ~ , YAMAIM,K . LI. &ND THIERY, J. P. (19%). Biologically active b nthetic peplidrs as probes 01 embryonic development: a competitive peptide inhibitor of fibronectin function inhibits gastrulation in amphibian embryos and neural crest migration in avian embryos. J . Cell Biol. 99, 1812-1830. 16 LEPTIN.A . A K ~GRTTP;F.WAI.D, B. (1991). Cell shape changes during gastrulation in I),o.wphilu. Devc4opmml 110, 73-M. 17 KAM, Z., MINDEN.3 . S., AGAKD, D. A , , SEDAT,J. W. AND LEITIN.A. (1991). Drosophilu gastrulation: analysis of cell shape changes in living embryos by three-dimensional fluorescence microscopy. Development 112, 365-370. 18 SWF.FTOP;, D., PARKS, S . . COSTA. M. AND W I ~ S C H A U SE. . (1991). Gastrulation in Drosophilu: the formation of the ventral furrow and posterior midgrit invaginationb. L)evelopmmf 112. 365-370. 19 ODELL,G. M . , OSTFR,(3..ALBERCH. P. AND RIJRUSIDF,R. (1981). The nicchaiiical hasic of morphogenesi?. I. Epithelial Ioliling and invaginalion. Dev. B i d . 85. 446-462. 20 UARU. J . B. L. A X D Ross, A. S. A. (19826). Tlic morphogenesis of thc ciliary body ot the avian eye. 11: Differential enlargcmcnt causes an cpithclium to bucklc. Dev. B i d . 92, 87-96. 21 PARKS. S . A N D WIISCHAUS. E. (1991). The D.rosophi!a gastrulation genc copmi-tinu encodes a G alpha-like protciii. Celi 64. 447-458.
1
Jonathan Bard i s at the MRC Human Genetics Unit, Wcstcrn General Hospital, Edinburgh EH4 2XU,
UK.
actions(4) and which causes epithelia to display autonomous extensive bchaviour (SCC 4 and 5 for reviews). The inechanism was first identified by the Fristroms in their study of the everting Drosophilcr leg disc(‘,’) where the constituent cells seem to rearrange their nearest neighbours to effect a process that seems analogous to the opening of a telescope. More recently, they have shown that morphogenesis is preceded by a flattening and elongation of the participating cells(’) and is thus a two-stage process. Cell rearrangement has now been documented in vertebrate as well as invertebrate development. It accounts for FunduZus epibolyC9)where the bcginnings of a theoretical analysis of the mechanism have been made(’”) and Xeizopus blastula morphogenesis(”) and may also Flap a role in cell thinning in heart formation(’-) hydra regeneration(’’) and various other phenomena[l4). It is also one of the driving forces of amphibian gastrulation where the phenomenon is demonstrated by the extensive buckling shown b the ectoderm when mesoderm involution is inhibited(”). It is now clear that autonomous cell rearrangement is the key mechanism in controlling many examples of epithelial morphogenesis. We are however, short of accessible embryos in which to study thc process and it is therefore a pleasure to welcome to the canon a set of papers on Drosophila gastrulation, particularly because the pictures are so good. Two of these papers are from Leptin’s laboratory and investigate the phenomenon in sectioned(“) and in living(”) material and the other is from Wieschaus’ group; this paper not oiily details SEM observations on Lormal and mutant embryos(”), but also includes information on the posterior midgut invagination. Although they usc differcnt techniqucs. the two laboratories provide very similar views of the phenomenon. As Leptin’s group can claim temporal precedence, I shall initially concentrate on its work and refer later to the observations on the mutants and the foregut invagination of the Wieschaus laboratory as they help to clarify a problem highlighted by thc morphology. The study of both sectioned and SEM material show that, over a 20 min period, the ventral epithelium of this relatively transparent embryo develops a furrow whose internalised cells become mesenchymal in nature simple: by the standards of gastrulation. It might be thought surprising that no one had previously looked at this behaviour in the living embryo, but the sophisticated time-lapse techniques rcquired mercly to describe the dynamics at thc cellular level(17)demonstrate why it would have been hard to undertake the work even five years ago. The classic way to study cell behaviour in vivo involves the biologist immobilising an embryo or organ under a microscope and then, with the use of appropriate optics and frequent changes of focus, taking regular pictures which then slowly have to be analysed. l i f c is easier now! Kam el uL.(~’) injected ~
fluorescein- and rhodamine-labelled markers into Ihe embryos, the one for nuclei and the othcr for highlighting cell borders. Pairs of photographs for the two fluorochromes were then takcn every minute at 10 planes of focus 3pm apart until gastrulation was complete. the whole shooting match being driven by a computer work station. The authors then used the computer to reconstruct the distinct dynamic behaviours of the participating cells and their nuclei. It turns out that furrow formation involves two distinct processes which take place in the central region of the ventral epithelium. The first is the migration of about half of these nuclei from their apical to their basal surface with the apical end of the ccll then narrowing. The second process involves the rapid inward movement of the remainder of these nuclei and, as they move, the epithelium seems to collapse inwards and peripheral cells fold over to form the furrow. Four obvious mechanisms could. in principlc at least, generate this behaviour. Local changes of intercellular adhesivity in the potential folding domain might cause the epithelium to bend and then invaginate('); a propagating wave of local contraction could lcad to foldingc1'); the phenomenon might have a simple mechanical explanation with the cells whose nuclei had moved basally collapsing inwards so causing the peripheral cells to fold around them; or, fourth, we might have another example of cell rearrangement. As the evidence shows neither changes in curvature nor a wave of contractions, both studies exclude the first two possibilities, while the speed of the event and direct observation rule out growth or cell division as a participating mechaniqm. In the original paper, the morphological evidence led Leptin and Grunewald to conclude that furrow formation was caused by an internal reorganisation of the epithelium analogous to cell rearrangement. The evidence from the living organism seems to make Leptin and her co-workers draw back from thi5 conclusion as the data are compatible with folding deriving from a mechanical instability caused by the movement of nuclei. Were this so, the authors suggest that the global behaviour of the epithclium is no morc than the sum of the behaviours of its individual cells and that there is no nced to postulate coordinated behaviour beyond, of course, the specification of a domain where nuclear movement takes place. So the reader is left hanging, wondering which of these alternatives is correct or indeed whether they are quite as mutually exclusive as they seem at first sight. For the recent work €rom the Fristrom laboratory@) now makes it likely that cell rearrangement is a twostage process where overt epithelial shape change i s preceded by intracellular events that will then drive that rearrangement. Furrow formation in Drosophilu gastrulation is also a two-stage proccss with the first event being an intracellular change in the position of the nucleus and the second overt morphogenesis. It thus fits
with our expectation of cell rearrangement, but the actual mechanism of furrow formation remains unclear. One possibility is certainly mechanical collapse while another is organised epithelial movement. The former would derive pa.isively from the intracellular reorganisation while the latter would be the active result. Although Kam el a/.(I7) do not commit themselves, their evidence inclines me to favour the latter alternative as the form of the developing furrow when it invaginates into the fluid-filled blastocoel seeins too well-organised and to require too much energy to derive from mechanical instability (the folds that form by buckling in the anterior retinal epithelium of ciliary body of the eye are much less well-organised(2"). Moreover. the morphology is very similar to the early stages of sea-urchin gastrulation, albeit that the fold there is two-dimensional rather than one. Nevertheless, the evidence is still inadequate to discriminate betwcen the options. It is here that the observations of the Wieschaus group come into their own because they provide several new lines of evidence suggesting that thc second stage of furrow formation is a programmed event rather than the mechanical consequences of the initial nuclear movement. They first show that the second stage is accompanied by the shortening of the elongated cells and argue that this is an active event that helps form the furrow rather than a passive consequence of it. The second line comes from the behaviour of the mutants con( ertina (now cloncd)(21) and folded gastrulation: both fail to complete gastrulation, even though the ventral cells in both mutants undergo the preliminary elongation and apical constriction that is the first step in the process. This evidence argues that furrow formation is an active event rather than a passive response. a view supported by the fact that the concertina gene encodes part of a G protein and hence is likely to play a ccllcommunication('') rather than a mechanical role. The analysis of posterior midgut invagination adds weight to this vie": in the mutants. the cells similarly undergo their initial constriction and again the invagination fails to form. Here, however. the complexity of the normal invagination makes it very hard to argue for a morphogenetic mechanism based on mechanical collapse. It thus seems that furrow formation is another example where gastrulation is underpinned by cell rearrangement. If so, then the phenomenon may turn out to be unexpectedly uscful: epithelial rearrangement is probably the key rnorphogenetic movement in early embryogenesis and we need a model system in which to analyse it. As the Drowphilu blastula is accessible, transparent and undergoes a gastrulation in which nothing seems to happen except epithelial folding, these two laboratories may have the organism, the phenomenon and the technology which will allow the process of cell rearrangement to yield some of its secrets.
References 1 GCSTAVSUN,T . .AKU WOLPIRT,L. (1962). Cellular mechanisms in the morphogenesis of the sea urchin larva. Changes in the 3hape of cell sheets. Exp
Cell Res. 21. 260-279. 2 HARDIN, J. D. (1988). The role of secondary mesenchyme cells during sea
urchin gavtrulatioii studicd by lascr ablation. Developnienr 103, 317-324. 3 HARDIN. J. D. NO CHFNG.L. Y . (1986). The nicchanisni and niechanicc of archenteron elongalion in the sea urchin embryo. I ) w R i d . 115, 490-501. 4 FRISTR~M. D. (1988). The cell~clarbasis of epithelial rnnrphogeneris. Ti537re R Cc4 20. 645 690. 5 UARD,J . H . L. (19Y@). Morphogenois. I'iw Cellular and Molecuinr Processes of Drrchpmenml A n a t o m y . Cambridge Linivrrsity Press, Cambridge. 6 PLKL'IL, b.. FRISTROX, U . , KISS, 1. A N D FRlSTROhl, J . W . (1975). The mechaniam of evagination of studiea on trypsin-accelerated ebaginatwn. Wdh. Roux' Arch. 178, 123-138. 7 FKIS~ROM. D. (1976). The mechanism of evagination of imaginal discs of Drosopidu ntelunogusrer. 111. Evidence for cell rearrangement. Dev. Biol. 54, 163-171. 8 COXDIC, M. L ., FKISIKO.M, D. AND FRISTROM. J. W. (1991). Apical cell 4iape changes during Drosopkilu imaginal lcg disc elongation: a iiovcl morphogenetic mechanism. Development 111. 23-31. 9 KELLER.R. E. AXD TRIKKAUS. .I. P. (1987). Rcarrangcrncnt of enveloping cclls without disruption of the epithelial permeability harrier as a factor in Fiindulw spiboly. Dev. Bid 120. 12-24. 10 WELIKY.6'.AND OSTBK,G . (1990). 'The niechanical basis of cell rearrangement. I. Epithelial morphogenesis during Fundulus epiboly. Developinent 109, 373-386. 11 KEILFR, R. E. (19S6). The cellular hacis nf amphibian gastrulation. In De~elopmentnl hiology, u rvmzpr.chmsivr ~ynthrxis.II. Tkc r.e/lirlnr basis of morpkog.rxesi.r.cd. L. W. Rrowder, pp. 241-327. New York: Plenum Press. 12 XfANASEK; F. J., BL-RNSIUb, M. B. AND WAILK~UN, R. E. (1972). hlyocardial cell shape change as a mechanism of embryonic heart looping. Dev. Uiol. 29, 349-371.
h
13 Born, P. M. AND BUD^^, H . R. (1984). Formation of pattern in regenerating tissue pieces of HyiIrn utlenuutu. 111. 'The shaping of the body column. Dev. Bid. 106. 315-325. 14 KELLER.R. AND IiAKDm. J. (1987). Cell behaviour during active cell rearrangement: evidence and speculation. J . Cell Sci., Suppl. 8 . 369-393. 15 BOlTrAIJT. J. c..DARRIRERE. T.. PoorF, T. J. A O Y A MH., ~ , YAMAIM,K . LI. &ND THIERY, J. P. (19%). Biologically active b nthetic peplidrs as probes 01 embryonic development: a competitive peptide inhibitor of fibronectin function inhibits gastrulation in amphibian embryos and neural crest migration in avian embryos. J . Cell Biol. 99, 1812-1830. 16 LEPTIN.A . A K ~GRTTP;F.WAI.D, B. (1991). Cell shape changes during gastrulation in I),o.wphilu. Devc4opmml 110, 73-M. 17 KAM, Z., MINDEN.3 . S., AGAKD, D. A , , SEDAT,J. W. AND LEITIN.A. (1991). Drosophilu gastrulation: analysis of cell shape changes in living embryos by three-dimensional fluorescence microscopy. Development 112, 365-370. 18 SWF.FTOP;, D., PARKS, S . . COSTA. M. AND W I ~ S C H A U SE. . (1991). Gastrulation in Drosophilu: the formation of the ventral furrow and posterior midgrit invaginationb. L)evelopmmf 112. 365-370. 19 ODELL,G. M . , OSTFR,(3..ALBERCH. P. AND RIJRUSIDF,R. (1981). The nicchaiiical hasic of morphogenesi?. I. Epithelial Ioliling and invaginalion. Dev. B i d . 85. 446-462. 20 UARU. J . B. L. A X D Ross, A. S. A. (19826). Tlic morphogenesis of thc ciliary body ot the avian eye. 11: Differential enlargcmcnt causes an cpithclium to bucklc. Dev. B i d . 92, 87-96. 21 PARKS. S . A N D WIISCHAUS. E. (1991). The D.rosophi!a gastrulation genc copmi-tinu encodes a G alpha-like protciii. Celi 64. 447-458.
1
Jonathan Bard i s at the MRC Human Genetics Unit, Wcstcrn General Hospital, Edinburgh EH4 2XU,
UK.
