ABSTRACT
The subject of this review is the nature of regulatory processes underlying the spatial subdivision of morphogenetic regions in later embryogenesis. I have applied a non-classical definition of morphogenetic field, the progenitor field, which is a region of an embryo composed of cells whose progeny will constitute a given morphological structure. An important feature of such fields is that they have sharp spatial boundaries, across which lie cells whose progeny will express different fates. Two examples of the embryonic specification and development of such fields are considered. These are the formation of the archenteron in the sea urchin embryo and the formation of dorsal axial mesoderm in the Xenopus embryo. From these and a number of additional examples, from vertebrate, Drosophila, Caenorhabditis elegans and sea urchin embryos, it is concluded that the initial formation of the boundaries of morphogenetic progenitor fields depends on both positive and negative transcription control functions. Specification of morphogenetic progenitor fields, organization of the boundaries and their subsequent regionalization or subdivision are mediated by intercellular signaling. Genes encoding regionally expressed transcription factors that are activated in response to intercell signaling, and that in turn mediate signaling changes downstream, appear as fundamental regulatory circuit elements. Such [signal →transcription factor gene → signal] circuit elements appear to be utilized, often repetitively, in many different morphogenetic processes.
Brief historical note: embryonic field concepts before 1940:Spemann (1938) reviewed what was even then a series of quite diverse concepts and definitions of “embryonic field.” To a modern eye most of these seem unacceptably metaphysical. Spemann’s preferred meaning for what he called “the embryonic field concept” was essentially that it is an inductive system, that will give rise to some structure. The term and some of the meaning was borrowed originally from “inductive field” as used in physics (p. 297, 303 op cit). In Spemann’s usage, for example, the archenteron roof of the urodele embryo is “at least one of the sources of the field” (p. 304 op cit) that produces the neural plate; the mesenchymal cells that Ross Harrison (1918) had shown to be capable of inducing forelimb buds produce a limb field; in the eye the optic cup produces a lens field, as shown by transplanting it to other regions (Spemann, 1938, p. 325). An essential, original property of Spemann’s embryonic field concept was that the field is “an equipotential system” (a concept derived from Driesch, 1905), in that if divided the field might recreate a complete new structure (Spemann, 1938, pp. 34, 302). This of course is true only transiently, but the phenomenon prominently affected the metaphysical flavor of argument surrounding classical field ideology even as late as 1938. Some of this flavor can be gleaned from Spemann’s remark that (in respect to) “the structure of the field,.. by its remaining whole after experimental partitions or augmentation of the material substrates, our intellect is puzzled by the same difficulties by which Driesch was led to establish the notion of entelechy” (his emphasis; entelechy was Driesch’s early vitalistic notion that parts of embryos possess an innate capacity to determine their own perfect and complete form).
The classical embryonic field has no boundaries, but rather displays a graded potential for developmental morphogenesis. Huxley and de Beer (1934, pp. 221-238) gave many examples of embryonic and postembryonic fields in which, as they defined it, the extent of the field for different organs always exceeds the actual anlage or rudiment. Thus the Huxley-de Beer fields of a given embryo at any one stage often overlap. In their treatment the key feature of the field is the presence of morphogen gradients within it. This feature is used to explain the ability of the classically conceived embryonic field to recreate complete structures when the field is sectioned, as well as its apparent properties of orientation and self-organization. To quote Huxley and de Beer, “The original control of differentiation in all cases appears to be exerted in relation to what may be called a ‘morphogenetic field.’ Within these fields various processes concerned with morphogenesis appear to be quantitatively graded …” (Huxley and de Beer, 1934, p. 274). Their ‘gradient fields’ could be of two forms i.e., either dynamic gradients, or passive gradients of morphogens (1934, op cit, pp. 310-320). Dynamic gradient theory was later developed in elegant ways to include parameters that would shape the morphogen concentration functions and might produce spatial discontinuities and shaped patterns (e.g., Turing, 1952; Meinhardt, 1982). To relate such morphogen concentration gradients to the behavior of cells, the cells are required to know what to do by their position in the shaped morphogen gradient, the intensity and composition of which they read and respond to (see e.g., Wolpert, 1971). These concepts of embryonic or morphogenetic fields are considerably different from those of Harrison and Spemann, and in fact Spemann himself expressed great reservations about gradient-based interpretations of embryonic fields (Spemann, Chapter 16, 1938). Furthermore, in modern usage the meaning of the term embryonic field is often again permuted. For example, in a recent review De Robertis et al. (1991) discuss limb buds, feather buds, fin buds, and wing buds. Apparent gradients of homeobox gene expression are observed within the buds, and these are interpreted by the authors as molecular manifestations of Huxleyde Beer embryonic gradient fields. However, in all of these fields sharp boundaries are manifest, and except for cells destined to become necrotic the areas giving rise to these buds, and the buds themselves, consist of the cellular progenitors of the respective structures. Thus this usage represents another redefinition of the classical concept, since homeobox regulators are nuclear proteins, and do not diffuse across the space occupied by the field, and since Huxley-de Beer fields do not correspond to anlagen. The classical terms ‘embryonic field’, ‘morphogenetic field’, and their relatives, now even more than in Spemann’s day, have accumulated too many different designata, so that they fail to communicate any unique scientific concept. The classical field might (depending upon the author) include not only all the cells that give rise to a given structure, but any other cells that are responsible for inducing it, plus all cells in that area that are competent to respond to this induction in ectopic experimental recombinations. Furthermore, classical field concepts do not easily lend themselves to mechanistic molecular interpretations based on cell function. Ingham and Martinez-Arias (1992) concluded a discussion of morphogenetic boundaries in the insect cuticle with the observation that while we are now achieving a molecular level understanding of cellular function at morphogenetic boundaries, “Fields are still a classical embryological concept. Perhaps that is what they will remain …” The reason that this prediction is likely to be accurate lies in the essential nature of classical embryonic field concepts.
Other coherent models for the process by which the dorsal axis arises in amphibian embryos have previously been put forward, in which the spatial aspects of mesodermal specification depend essentially on concentration clines of a few factors (e.g., Smith and Slack, 1983; Smith et al., 1985; Green et al., 1992). This section of the present essay concludes with discussion of a form of signaling regulatory circuitry that may underlie the various steps of the process described; however, that conclusion is not dependent on the details of these different models of the inductive process. The current molecular and cellular evidence for sequential inductive interactions would, I think, indicate the same form of underlying regulatory circuitry at the intracellular level, had this evidence instead been organized from the starting point of the theories of Smith and Slack (1983) or Green et al. (1992).
The animal cap of stage 7-9 (mid to late blastula) embryos is not an entirely naive, unspecified test system (for an excellent discussion of the problems of interpreting animal cap explant experiments see the recent review of Jessell and Melton, 1992). The animal cap already displays a dorsoventral polarization (presumably the result of early signals from the Nieuwkoop Center and/or from the prior rotational ‘activation’ of the cytoplasm on the dorsal side at first cleavage). Thus the response to various growth factors in dorsal animal cap fragments is different from that in ventral animal cap fragments (Sokol and Melton, 1991; Ruiz i Altaba and Jessell, 1991). Kimelman and Maas (1992) found that injection of bFGF mRNA into eggs causes the dorsal region of animal caps, isolated at gastrula stage, to express dorsal mesoderm markers, while the ventral regions express ventral mesodermal markers. In other words, in these experiments the presumptive ectodermal cells on the dorsal side of the animal cap behave like the initially pluripotential dorsal blastomeres overlying the Nieuwkoop Center. These blastomeres receive both the dorsal signal from the Nieuwkoop Center and the signals producing general marginal zone mesoderm specification, i.e., including bFGF. In contrast, the cells on the ventral side of the animal cap resemble ventral marginal zone cells, after bFGF mRNA injection.
