A role for cytoplasmic determinants in the early development of several organisms has been suggested by results from a variety of studies. Ideally, a demonstration of the existence of determinants requires their transfer between regions of an embryo, but this has been achieved only rarely (see chapters by Nusslein-Volhard and Jackie, this volume). In many cases, the existence of cytoplasmic determinants is inferred indirectly, for example where procedures that disturb the asymmetric arrangement of the cytoplasm also disrupt the normal allocation of cell types, or where a correspondence exists between the prospective fate of a cell and the cytoplasmic endowment it has received (see chapters by Gurdon, Schierenberg, Strome and Kalthoff, this volume). In these latter examples, a heterogeneity within the cell may indeed exist and influence further development, but its nature may preclude direct testing by cytoplasmic transfer, or a regulatory capacity may exist whereby the localization and/or expression of putative determinants can be modified (Jeffery & Raff, 1983; Johnson & Pratt, 1983). In this paper, we define three minimal operational criteria that must be satisfied for the demonstration of a role for cytoplasmic inheritance in early development. First, a consistent asymmetry in the organization of individual cells of the early embryo should be identifiable. Second, at least some of the properties characterizing the distinct regions of each asymmetric cell should be distributed differentially at cell division. Third, the properties that characterize the asymmetric organization of the parent cell and its daughters should lead causally and predictably to differences in the fates of these cells and their descendants. We examine the evidence as to whether these criteria can be met for the earliest events of mouse development. We conclude that by these criteria cytoplasmic determinants are indeed present, but we question the value of the notion of cytoplasmic determinants for an understanding of developmental mechanisms.
AXES OF DEVELOPMENT
The acquisition of developmental axes occurs relatively late in mammalian development when compared with most, if not all, other types of embryo. The earliest identified axis is radial and develops at the 8-cell stage about 48 h into development, the internal or core region of the embryo becoming structurally and developmentally distinguishable from the outer peripheral region (Fig. 1; Tarkowski & Wroblewska, 1967; Wilson, Bolton & Cuttier, 1972; Hillman, Sherman & Graham, 1972; Ducibella & Anderson, 1975). This radial axis marks the positions of future potential allocations to the embryonic and extraembryonic lineages. A secondary axis, also relating to embryonic: extraembryonic divergence, develops around 36 h (or three cell cycles) later when the embryonic: abembryonic axis forms (Fig. 1; Gardner, Papaioannou & Barton, 1973; Gardner & Papaioannou, 1975; Smith & McLaren, 1977; Wiley & Eglitis, 1981). It is not until around 3 days further into development that definitive anteroposterior and dorso ventral embryonic axes become evident. We know little of the mechanism by which the embryonic: abembryonic axis is laid down (Surani & Barton, 1984) and less about how the definitive embryonic axes are established (Smith, 1985). This chapter therefore concentrates on the establishment of the earliest, radial axis.
DOES THE RADIAL AXIS OF DEVELOPMENT INVOLVE THE ASYMMETRIC ORGANIZATION OF INDIVIDUAL CELLS?
It is possible to identify many differences in organization between the central and peripheral regions of the embryo in advance of the earliest establishment of internal and external cell populations at the 16-cell stage. These differences include structural and functional regionalization of individual cells (Fig. 2) with regard to their cytoplasmic organelles (Reeve, 1981a; Reeve & Kelly, 1983; Lois & Izquierdo, 1984; Fleming & Pickering, 1985; Maro, Johnson, Pickering & Louvard, 1985), the major cytoskeletal elements of the cytoplasmic matrix (Ducibella, Ukena, Karnovsky & Anderson, 1977; Sobel, 1983, 1984; Johnson & Maro, 1984; Sobel & Alliegro, 1985; Houliston, Pickering & Maro, in preparation; Fig. 3) and the cytocortex defined here as embracing the cell membrane plus its associated material including membrane-associated cytoskeletal proteins (Calarco & Brown, 1969; Ducibella & Anderson, 1975; Ducibella, Albertini, Anderson & Biggers, 1975; Izquierdo & Marticorena, 1975; Magnuson, Demsey & Stackpole, 1977; Handyside, 1980; Izquierdo, Lopez & Marticorena, 1980; Reeve & Ziomek, 1981; Izquierdo & Ebensperger, 1982; Kimber & Surani, 1982; Kimber, Surani & Barton, 1982; McLachlin, Caveney & Kidder, 1983; Goodall & Johnson, 1984; Fleming & Pickering, 1985; Nuccitelli & Wiley, 1985; Pratt, 1985; Johnson, Maro & Takeichi, 1986). Some of these differences appear first at the 4-cell stage, but most develop and become stabilized at the 8-cell stage in the mouse. The result is an embryo in which each cell is polarized with apical (outward facing) and basolateral (inward facing) domains. Moreover, when the late 8-cell embryo is disaggregated into its constituent cells, each retains most elements of its polar phenotype (Figs 2,3). Thus, the first criterion for a role for cytoplasmic inheritance appears to be met.
CAN DIVISION OF A POLARIZED 8-CELL BLASTOMERE GENERATE CELL DIVERSITY?
Most of the features identified as being polarized in individual 8-cell blastomeres are dispersed at entry into mitosis, congregate in and around the mitotic spindle or its poles, and become distributed to both daughter cells. Such features include cytoplasmic organelles (Reeve, 1981b; Fleming & Pickering, 1985; Maro et al. 1985), cytoplasmic actin (Johnson & Maro, 1984) and both cytoplasmic tubulin and microtubule-organizing centres (MTOCs; Houliston et al., in preparation). It is not possible to say that the distribution to each daughter cell is exactly equivalent, indeed this is unlikely to be the case especially where daughter cells differ in size (Ziomek & Johnson, 1981). However, it is clear that the asymmetric distribution of the main cytoskeletal elements of the cytoplasmic matrix, as well as the organelles associated with them, depends upon the continuing interphase state for its stability. Such a finding is not unexpected for components essential to a cell’s survival.
In contrast, at least some features of cytocortical polarity are relatively stable. Throughout mitosis, which lasts some 45–60 min from prophase to cytokinesis, it is possible to discern an inhomogeneity within the cytocortex (Fig. 4; Johnson & Ziomek, 1981a). As a result, if the cleavage plane is oriented approximately perpendicular to this axis of polarity, one of the two cells that result inherits an apical domain and remains polar, the other lacks such a domain and is apolar. Alternatively, if the cleavage plane runs approximately parallel to the axis of polarity, two polar cells result, and the poles on each cell are contiguous across the residue of the cleavage furrow connecting the two daughter cells (Johnson & Ziomek, 1981a; Reeve, 1981b). Immediately after the completion of cytokinesis, the cytoskeletal matrix and cytoplasmic organelles in those 16-cell blastomeres that have inherited an apical cytocortical domain also resume a polarized distribution about an axis of polarity coincident with that of the cytocortex (Fleming & Pickering, 1985; Maro et al. 1985). The relative stability during mitosis of the cytocortical organization compared to that of the cytoplasmic matrix has also been inferred from observations on the behaviour of exogenously introduced markers (Wilson et al. 1972; Graham & Deussen, 1978; Karasiewicz & Modlinski, 1985). These results suggest that the cytocortex is characterized by two properties. First, it manifests a polar organization with a long half life in relation to the period of mitosis. Second, some aspect of this cytocortical polarity appears to remain spatially discrete in that it is susceptible to division by the cleavage furrow and serves as a defined organizing focus for the internal axis of cell polarity.
We must ask whether this relatively stable polar cytocortical domain coincides exactly and exclusively with the most prominent observable polarized feature of the cytocortex, namely the pole of microvilli, or whether there is some underlying basis for the various visible manifestations of polarity both cytocortical and cytoplasmic. Three lines of evidence make it unlikely that the observable pole of microvilli perse constitutes an exclusive basis for stable polarity. First, during mitosis the clearly demarcated apical pole of microvilli that characterizes the interphase state becomes less-sharply defined. The region of the surface occupied by microvilli expands, occupying progressively more of the cell surface as division proceeds (Fig. 4). During a normal mitosis of 45–60min, microvillous pole expansion is relatively limited. However, if mitosis is arrested at metaphase for up to 6 h by placing prophase cells into medium containing the microtubule-depolymerizing agent nocodazole, the dispersal of microvilli extends over the whole surface in many cells, although a gradation remains evident in most cells giving them a continuing inhomogeneity (Fig. 5). If the drug is removed, mitosis resumes immediately and couplets of cells form with identical phenotypes to those derived from nonarrested controls i.e. polar-apolar pairs or polar-polar pairs with polar continuity at the remnant of the cleavage furrow, and in each case the pole that forms is clearly demarcated from the nonmicro villous areas of the cell (Fig. 5, Table 1; Dhiman, Pickering, Maro & Johnson, in preparation). Thus, the observed ‘diffusion’ of surface polarity did not prevent the restoration of a sharply defined and appropriately oriented polar phenotype subsequently.
A second line of evidence suggesting that the axis of cell polarity is separable from overt microvillous polarity derives from the analysis of interphase polar cells in which the overt polar region of microvilli has been disturbed experimentally in various ways but the underlying polar axis to the cell nonetheless persists (Nuccitelli & Wiley, 1985).
A third line of evidence leading to the same conclusion comes from a study of the generation of polarity de novo at the 8-cell stage. When 8-cell blastomeres are cultured together, polarity in each always develops on an axis that is normal to the geometric centre of the various asymmetric points of contact that each cell makes (Ziomek & Johnson, 1980; Johnson & Ziomek, 1981b). It has been shown that these asymmetric cell contact patterns do indeed determine the orientation of the axis of polarity that develops. By use of pairs of 8-cell blastomeres of known age after their derivation from a dividing 4-cell blastomere, it has been shown that at any time over the first few hours of the 8-cell stage, changing the contact pattern between the cells changes the axis of polarity that develops, such that it is normal to the new point of contact. However, at later time points, such a regulative response does not occur, and changing the contact patterns results in axes of polarity that are related to the old rather than the new contact points (Fig. 6; Johnson & Ziomek, 1981b). The axis of polarity induced within each cell appears to become stabilized as progress through the fourth cell cycle occurs. A similar increase in axial stability with development can be inferred from the results of comparable studies on blastomeres in situ (Sepulveda, Doggenweiler & Izquierdo, 1985). The establishment of this stable axis of polarity occurs at about the time that the earliest evidence of polarity in the cytoplasmic matrix can be detected, but develops in advance of the earliest detection of microvillous polarity (Johnson & Maro, 1984; Maro étal. 1985; Fleming & Pickering, 1985; Fleming, Pickering, Qasim & Maro, 1986b). However, the temporal sequence in which the overt features of polarity develop naturally does not seem to reflect a causal sequence, since the disruption of microtubules and/or microfilaments, and thereby of the organized cytoplasmic matrix, does not prevent the generation of microvillous polarity (Ducibella, 1982; Maro & Pickering, 1984; Johnson & Maro, 1985; Fleming, Cannon & Pickering, 1986a).
Taken together with the evidence from mitotic cells, these results suggest that an early event in the process of polarization is the induction and stabilization of an axis of polarity that provides the framework about which the various visible elements of cellular asymmetry are organized (Fig. 7). As the development and persistence of this polar axis appears not to require an organized cytoplasmic matrix, it may reside in whole or in part within the cytocortex. Such a stable polar cytocortical domain would satisfy the second criterion for the existence of a role for cytoplasmic inheritance during early development in the mouse. Our current understanding of the nature of this domain, and how it might be achieved, is discussed in the last section.
DOES THE GENERATION OF POLAR AND APOLAR 16-CELL BLASTOMERES REPRESENT THE FOUNDATION OF TWO CELL LINEAGES?
The newly formed 16-cell embryo is composed of two cell types : a population of outer, polar cells each derived from the apical region of a polarized 8-cell blastomere and a population of inner, apolar cells each derived from the basal region of a polarized 8-cell blastomere. The ratio of polar to apolar cells will depend on the relative frequency of cleavage plane orientations (see later). Amongst the properties that distinguish the apolar and polar cells, perhaps the most important in the context of lineage establishment is their differing patterns of adhesivity. Apolar cells appear uniformly adhesive, whereas polar cells are preferentially adhesive over their basolateral surfaces (Burgoyne & Ducibella, 1977; Ziomek & Johnson, 1981; Kimber et al. 1982; Johnson & Ziomek, 1983). This distinction ensures the maintenance of the relative internal and external positions within the embryo at which each cell type arrives following cleavage (Ziomek & Johnson, 1982; Randle, 1982; Surani & Handyside, 1983). It also ensures that the polar zone of microvilli is directed outwards at the surface of the embryo, thus orienting each cell radially.
In the definitive blastocyst, which forms at the late 32-cell stage and achieves its mature phenotype by the late 64-cell stage, two cell subpopulations exist that resemble in many ways the subpopulations at the earlier 16-cell stage (Handyside, 1981; Johnson & Ziomek, 1982; Randle, 1982; Edirisinghe, Wales & Pike, 1984; Fleming & Pickering, 1985; Fleming & Goodall, 1986). Thus, the outer troph-ectodermal cells form a polar epithelium whilst the inner cell mass (ICM) cells are apolar and are totally enveloped within trophectodermal processes (Fleming, Warren, Chisholm & Johnson, 1984). However, comparability of phenotype does not prove continuity of lineage. For this to be established, the fate of the progeny of the apolar and polar cells at the 16-cell stage must be followed. This can be achieved by marking cells in each subpopulation at the 16-cell stage and examining the distribution of the marked cells at later stages. Of direct relevance to the subject of this paper is the role of the stable cytocortical domain. Do all cells possessing this domain contribute all their progeny to the trophectoderm? Conversely, do those cells lacking it ever contribute cells to the trophectoderm?
A number of lineage studies has been undertaken in which a single 16-cell polar blastomere has been marked, the cell incorporated into an unmarked intact embryo or aggregate of unmarked cells, and the distribution of the marked cells examined one to two cell cycles later. These studies have revealed that a polar 16-cell blastomere almost always contributes progeny to the trophectoderm, can under some circumstances contribute cells to the ICM as well, but contributes cells exclusively to the ICM so rarely as to be explicable in terms of misidentification of, or damage to, the marked cell or its progeny (Table 2 - which also includes references). This result indicates that the stable cytocortical domain may place a restriction on developmental fate, in that it evidently debars exclusive contribution to the ICM, but the result does not tell us whether the domain is developmentally determinative in the strictest sense of the word, namely that possession of the zone necessarily results in trophectoderm formation. This question has been addressed by examining the lineage behaviour of a single marked polar 16-cell blastomere cultured either alone or in combination with a second polar or apolar cell. It is clear that when the polar cell is associated with an apolar cell it divides conservatively and generates two polar trophectodermal daughter cells, whereas a polar cell alone or aggregated with a second polar cell can divide differentiatively to yield one polar and one apolar cell (Fig. 8; Johnson & Ziomek, 1983). These results indicate that (i) cell interaction influences cleavage planes and thus lineage, and (ii) transmission of all or part of the apical cytocortical domain, or some closely associated property, does appear to be determinative for a trophectodermal fate.
An equivalent series of studies has been undertaken for apolar 16-cell blastomeres. As a result, two conclusions may be drawn. First, apolar cells by virtue of their adhesive properties tend to locate internally and generate only ICM progeny (Randle, 1982; Ziomek & Johnson, 1982). Second, if an apolar cell is forced to occupy an external position, by, for example, its aggregation with a population of other apolar cells, then it responds to the asymmetric array of contacts in just the same way as early 8-cell blastomeres do, and polarizes. Once polarized, it may then contribute progeny to the trophectoderm (Fig. 9; Ziomek & Johnson, 1981; Ziomek, Johnson & Handyside, 1982; Johnson & Ziomek, 1983; Johnson, 1986). This series of experiments leads to the conclusion that the possession of an apical cytocortical domain guarantees a contribution to the trophectoderm through its effect on cell interaction, whilst in its continuing absence an ICM fate is assumed.
THE RELATIONSHIP BETWEEN DETERMINACY AND PLASTICITY
The lineage analyses described in the previous section indicate that the possession of an apical cytocortical domain is indeed associated with the determination of cell fate, and is so because of the polar phenotype imposed on the cells that possess it. The analyses also show that the existence of this determinant of trophectodermal development is compatible with the developmental plasticity observed in early mouse development. Indeed, it is now clear that the plasticity within the system revealed by experiment is used regulatively in the intact embryo. Thus, when the number of differentiative divisions amongst the 8-cell blastomeres within a single embryo is low, so that the allocation of apolar cells to the inside of the 16-cell embryo is correspondingly low, then at the next division the frequency of differentiative divisions by polar 16-cell blastomeres is relatively high, yielding a compensatory second allocation of cells to the ICM. The converse also applies (Fig. 10; Fleming, 1986). Moreover, in the intact 16-cell morula, the differential adhesive properties of polar and apolar cells are sufficient to maintain their relative outer and inner positions so that apolar cells do not in fact express their potential to polarize and therefore they contribute their progeny exclusively to the ICM (Fleming, 1986).
Taken together, the studies summarized here have (i) established the lineage relationships of the ICM and trophectoderm, (ii) quantified the extent to which, and defined the conditions under which, plasticity is expressed within this pattern of lineage relationships, and (iii) defined the cell dynamics underlying both the determinative and plastic aspects of development. We now ask whether the presence or absence of the apical cytocortical domain leads not just to cells with different properties but also to cells with different and stabilized patterns of genetic activity.
DO THE CELLS IN THE TWO CELL LINEAGES FORMED AT THE 16-CELL STAGE EXPRESS DIFFERENT GENES?
It has been known for a number of years that the cells of the ICM and the trophectoderm synthesize different proteins (Van Blerkom, Barton & Johnson, 1976; Dewey, Filler & Mintz, 1978; Howe & Solter, 1979). These differences in biosynthetic pattern appear to be associated with differences in the prevalence of coding transcripts. Thus, when ICM cells from newly expanding blastocysts are removed from within their enclosed trophectodermal environment and are exposed to in vitro culture either in isolation or after aggregation in groups, they undergo a transformation to yield miniature blastocysts that include trophectoderm cells (Johnson, Handyside & Braude, 1977; Handyside, 1978; Hogan & Tilley, 1978; Spindle, 1978; Surani, Torchiana & Barton, 1978). The earliest events in this transformation involve flattening and polarization of the most superficial cells of the exposed ICM (Fleming et al. 1984). Over the ensuing 18 to 24 h period in culture, the polarized cells yield a definitive trophectodermal vesicle that may contain an ICM-like cluster. At the time of their isolation, the ICMs do not synthesize the marker proteins characteristic of trophectoderm. However, the synthesis of these proteins increases with time in culture, corresponding to the phenotypic transformation (Johnson, 1979). If transcription in the newly isolated ICM cells is blocked by addition of α-amanitin, both the morphological transformation to trophectoderm and the synthesis of the mature pattern of trophectodermal proteins are blocked. Some early trophectodermal proteins are, however, synthesized (Johnson, 1979). These results suggest that (i) at least some of the proteins characteristic of trophectoderm are synthesized on templates that are unavailable in ICM cells, and (ii) blocking the synthesis of mRNA transcripts is associated with a failure of the ICM cells to transform to a trophectodermal phenotype. Thus, the segregation of the two cell lineages may indeed have become associated with a difference in prevalence of transcripts. Such a difference in transcript availability could arise via differential inheritance, differential synthesis or a combination of both.
Among the various trophectodermal marker proteins, the most prominent of those that are sensitive to the action of a-amanitin have been identified as cytokeratins (Brulet, Babinet, Kemler & Jacob, 1980). Correspondingly, assembled cytokeratin intermediate filaments have been described in trophectodermal cells but are reportedly absent in ICM cells of the early blastocyst (Brulet et al. 1980; Jackson et al. 1980; Lehtonen et al. 1983). Both synthesis and presence of the cytokeratins can be detected at earlier morula stages, although the distribution of cell types responsible for their synthesis has not been ascertained (Handyside & Johnson, 1978; Lehtonen et al. 1983; Oshima et al. 1983). The cytokeratins therefore seem to offer a potentially useful molecular marker for studying differences in genetic expression.
Interestingly, careful analysis of the distribution of assembled cytokeratin filaments during cleavage has revealed their presence in a few blastomeres as early as the 8-cell stage (Fig. 11; Chisholm, in preparation). Their presence at this stage is preceded by the detection of the 55 × 103Mr cytokeratin protein in immunoblots at the 4-cell stage (Houliston & Maro, in preparation). In the early 16-cell embryo almost as many inside apolar cells have assembled filaments as have outside polar cells, although in general the filaments are less extensive in the apolar cells. When individual pairs of 16 cells derived by division of a single 8-cell blastomere are analysed, most are found to have either two filament-positive or two filamentnegative cells, rather than one negative and one positive cell (Fig. 11). As the embryos progress through the 16- and 32-cell stages, the proportion of outer polar cells with a filament network rises towards 100 % whilst that of inner apolar cells declines.
These results indicate that the capacity to assemble filaments and to elaborate filament networks is not acquired in all cells of the embryo at the same time. Moreover, different cell phenotypes and/or positions appear not to be correlated with the initiation of differences in filament assembly, but rather with a subsequent modulation up or down of filament protein synthesis and/or assembly as cells diverge further. Differences in genetic expression may derive from, rather than generate, the differences in phenotype and position of cells in the two subpopulations. A somewhat analogous ‘product driven’ regulation of cytoskeletal protein synthesis and assembly has been reported for other types of cell (BenZe’ev, 1984). Although these analyses on the early mouse embryo are restricted to one identified marker protein, and are far from being complete, they are consistent with the possession of an apical cytocortical domain leading ultimately to the setting up of genetic differences between cell lineages via its effect on cell surface properties.
WHAT IS KNOWN ABOUT THE STABLE APICAL CYTOCORTICAL DOMAIN?
The evidence relating to the induction and organization of polarity has been reviewed in detail recently (Johnson & Maro, 1986). Here we will present a summary of the conclusions together with a model based upon them. We need to consider (i) the temporal component of polarization - why does it occur at the 8-cell stage? and (ii) the spatial component of polarization - how is the axis of polarity in each blastomere oriented radially?
Polarization, and the associated events of intercellular flattening and gap and tight junction formation, occur characteristically at the 8-cell stage and together constitute the process of compaction. The proteins required for this major transition are present in the embryo considerably in advance of compaction itself (McLachlin et al. 1983; Kidder & McLachlin, 1985; L. Johnson, 1986; Levy, Johnson, Goodall & Maro, 1986).For example, it is possible to elicit cytocortical polarization and intercellular flattening prematurely at the early 4-cell stage by inhibiting protein synthesis (Levy et al. 1986). The results of this type of experiment tell us two things about the process of polarization. First, it must presumably be organized at a post-translational level as a result of a change in the stability, conformation and/or state of chemical modification of pre-existing proteins. Such a level of response may be important in facilitating the efficient integration of spatial signals in the organization of polarization. Second, since the proteins involved are present in adequate quantities some 12 or more hours in advance of their normal activation, the process of polarization (as well as the other events of compaction) must be prevented from occurring prematurely, and activation will therefore involve release from such a restraint. The temporal component to polarization appears to be permissive.
This permissive temporal component may involve a change in state intrinsic to the cells, since the capacity to polarize seems to develop in the absence of other cells (see below). Moreover, this intrinsic change of state does not appear to depend upon any time-measuring process involving recognition of the number of rounds of cytokinesis or karyokinesis, or to changing nuclear:cytoplasmic ratio, since it is insensitive to proximate manipulations of these cellular functions (Surani, Barton & Burling, 1980; Pratt, Chakraborty & Surani, 1981; Smith & Johnson, 1985; contrast with early Xenopus or Drosophila development, Newport & Kirschner, 1982; Edgar, Kiehle & Schubiger, 1986). How intrinsic time is measured is not clear.
A basis for the spatial component to polarization is more clearly established. The axis of polarity in a blastomere is evidently determined by the pattern of cell contacts perceived by the cell. As was reported earlier, changing the contact pattern of a cell during the early labile induction period leads to a change in the axis of polarity that develops. Moreover, in the absence of effective cell contact, blastomeres nonetheless polarize but do so more protractedly and with a random axis (Ziomek & Johnson, 1980; Johnson et al. 1986). In the intact embryo, the intrinsic and permissive temporal signal must coincide with an extrinsic spatial signal which is instructive and which determines the axis of polarity. In the absence of such a positional signal, the cell appears to generate polarity at random.
What is the nature of the spatial signal? Contact between cells in the early mouse embryo is mediated largely by the Ca2+ -sensitive cell adhesion system involving the glycoprotein uvomorulin (also called cadherin, L-CAM, gpl20 and cell-CAM 80/120; Hyafil, Morello, Babinet & Jacob, 1980; Damsky, Richa, Solter, Knudsen & Buck, 1983; Gallin, Edelman & Cunningham, 1983; Shir ay oshi et al. 1984; Vestweber & Kemler, 1984). The neutralization of this system leads to the loss of positional recognition and to delayed and randomly oriented polarization (Shirayoshi, Okada & Takeichi, 1984; Johnson et al. 1986). Uvomorulin would therefore appear to be implicated in the induction of polarity, whether or not other cell surface macromolecules are also required. Such a central role for cell adhesion in the earliest events of positional recognition is appropriate, since, as was seen earlier, one important consequence of polarization is the establishment of a mosaic pattern of surface adhesivity that can be distributed differentially to the daughter cells at cytokinesis. The difference in adhesivity of the two cell subpopulations so generated then reinforces and stabilizes the lineage differences. Cell adhesion is therefore involved in the establishment, confirmation and regulation of the two cell lineages (Kimber & Surani, 1982).
How is the receipt of an uvomorulin-mediated signal at one end of the cell converted to a stable cytocortical domain at the opposite end of the cell? We have suggested that since this process can be achieved in the absence of any obvious organized cytoplasmic matrix (Johnson & Maro, 1985; Fleming et al. 1986a), the transduction of information across the cell must rely either on the passage of a signal circumferentially around the cytocortex by, for example, a spreading enzymic activity, or on the setting up within the cell of axial gradients of small molecules such as could be provided by ion currents (Fig. 12; Johnson & Maro, 1985, 1986; Nuccitelli & Wiley, 1985). These mechanisms are not of course mutually exclusive and could be mutually reinforcing. We can gain some clues as to the mechanisms involved by a close examination of the local consequences of intercellular adhesion during earlier cleavage stages.
Uvomorulin is present and active in mediating adhesion on 1-, 2- and 4-cell blastomeres as well as on 8-cell blastomeres (Hyafil et al. 1980; Shirayoshi et al. 1984; Johnson et al. 1986). At these earlier developmental stages, sites of intercellular contact are characterized by a number of features including the reorganization of intramembranous lipid (Pratt, 1985), presence of certain enzymic activities (Lois & Izquierdo, 1984), reduction or loss of microvilli (Calarco & Brown, 1969; Dvorak et al. 1985), subcortical depolymerization of myosin, actin and tubulin (Sobel, 1983; Johnson & Maro, 1984; Maro & Pickering, 1984; Houliston etal., in preparation), and absence subcortically of several types of cell organelle (Maro et al. 1985). All of these features could arise as a result of localized subcortical changes in the concentration of relatively nonspecific molecules such as ions or cAMP. These various responses are also observed in the compact 8-cell embryo. However, whereas at earlier stages all the responses are focal and limited to the region of contact, at the 8-cell stage many extend beyond the immediate area of contact which itself becomes more extensive anyway. We have proposed that the remarkable global change of cell organization that occurs at compaction could be achieved quite simply by allowing a hitherto local response to cell contact to become propagated beyond the immediate contact point (Johnson & Maro, 1986). For example, tubulin polymerization is evidently not stabilized in regions of contact compared with elsewhere in the cell. If the failure to stabilize microtubules was extended progressively beyond the immediate point of intercellular contact, perhaps by a propagated change in the organization of the cytocortex, then progressive depolymerization would lead to a rise in the concentration of free monomer. Eventually, the contact-driven depolymerization basally would become balanced by a concentration-driven polymerization apically. With this local apical stabilization of microtubules would come recruitment of MTOCs which could further promote apical polymerization (Maro, Johnson, Webb & Flach, 1986; Houliston et al., in preparation). Cell asymmetry would thereby be achieved. A similar contact-driven propagation of actin depolymerization basally could lead to its polymerization apically, associated with apical cytocortical localization of actin-associated stabilizing and organizing proteins (Fleming et al. 1986a). In consequence of these cytoskeletal reorganizations, cell organelle and microvillous distribution would also change, the latter promoting intercellular flattening. This model has the advantage that it invokes a relatively small change in the global state of the cell as a trigger for the magnification and propagation of a hitherto local contact-induced response. A major transition in cell organization could be achieved independently of any requirement for protein synthesis or gene expression. Which features of the local response are critical for the successful achievement of the transition?
It is useful to divide the local responses to contact into those that remain restricted to the region of contact at the 8-cell stage and those that extend beyond it. Amongst the former are the activities of surface enzymes and the modified lipid organization, whilst the latter include the redistribution of microvilli and of subcortical organelles and cytoskeletal elements. It is tempting to speculate that a local contact stimulation of, for example, membrane-linked kinase, phosphatase or calcium channel activities constitutes part of a primary response, and that it is the extent to which the secondary consequences of such an activity are allowed to spread beyond the immediate contact zone that determines whether the cell flips to a polarized state. The chemical basis of these earliest contact-dependent responses will hopefully lead us to an understanding of the chemical basis of the stabilized apical polarizing domain that results. Moreover, since organisms as diverse as ascidians, seaweeds and amphibia also seem to generate, stabilize and use defined cytocortical domains to regulate early development (Gerhart et al. 1984; Brawley & Robinson, 1985; Jeffery, 1985), the underlying principles may be universal. It seems to us that we will achieve a more coherent understanding of development by identifying cellular mechanisms, rather than by searching for elusive and possibly illusory determinants. The question posed by this chapter may therefore be false.
We wish to acknowledge the advice of Dr B. Maro and the technical assistance of Raith Overhill, Tim Crane, Gin Flach, Martin George and Brendan Doe. We thank Dr Ph.Brulet for the gift of TROMA-1, and Dr M. Kirschner for the gift of the serum reactive with pericentriolar material. The work reported in this paper was supported by an MRC grant to M. H. Johnson and P. R. Braude, and by CRC grants to M. H. Johnson. J.C.C. and E.H. hold MRC Research Training Awards.