The process of cellular differentiation in developing multicellular organisms culminates in the formation of specialized somatic cell types. Once cell types acquire a specific cellular phenotype for performing a specialized function, no further changes usually occur. We call this the stability of the differentiated state. It is now widely accepted that the process and stability of cell specialization are under genetic control. But the frequency and extent to which this genetic control involves irreversible genetic changes among eukaryotic animals remain unanswered.

Cell differentiation in the development of multicellular organisms occurs as a consequence of the generation of chronologically and spatially distinct patterns of protein synthesis. These unique constellations of proteins confer on cells the functional and structural characteristics that enable them to perform their specialized roles in the organism. However, in addition to the overt protein synthetic profile, cells may retain the potential to produce proteins that they would not normally produce in significant amounts. During periods of stress, this potential is realized and is evidenced by the synthesis of a set of stress proteins. As discussed in this chapter, the pattern of stress protein synthesis is also subject to developmental regulation. Thus, cells are simultaneously engaged in two parallel developmental processes: regulation of the patterns of overt and stress-inducible protein synthesis. Stress-inducible protein synthesis can be evoked by regulation at either the transcriptional or translational levels, or both. Thus, regulation at both levels must be subject to developmental modulation.

In amphibian oogenesis, essential preconditions are established for a strategy of early development, in which growth and cell division are uncoupled. The fertilized egg cleaves rapidly to form a blastula without a significant expression of the embryonic genes during the first 12 cell cycles and without net intake of food during the first several days. The resources required for fast autonomous embryogenesis are accumulated during the several months of oocyte growth. The axial organization of the embryo is first established in the oocyte, since the direction of the animal-vegetal axis of the egg can be traced back to the nuclear-centriolar axis of the oogonium (Witschi, 1956; Coggins, 1973; Tourte et al., 1981). Also the dorsoventral axis is predetermined in the oocyte of some amphibian species (Wittek, 1952; for a review, see Gerhart, 1980).

The genome of most eukaryotic organisms is constant and perpetuates itself through numerous cell generations by replication. Yet, during embryogenesis, there is increasing cellular diversity, yielding a multitude of cell-specific phe- notypes. Cells become different from one another, and their developmental potential is progressively restricted. The determined state and the overtly differentiated state are exceedingly stable and heritable during normal development. However, various perturbations, both natural and artificially induced, can result in a modulation of cellular phenotypes. It is the purpose of this volume to analyze the nature of the stability of the differentiated state and the ability to reverse the differentiated phenotype. It is not intended to be all- inclusive and, in fact, we have chosen certain biological systems that require further exploration but that have the potential for helping us to understand the cellular and molecular mechanisms underlying these processes.

Transdifferentiation may be generally defined as the change of one recognizable cell type to another different cell type. The term was first used by Selman and Kafatos (1974) to denote the change of the cuticular cells of the moth larval silk gland to those producing HCO3 during metamorphosis/development and has since been used in many different contexts. So as not to produce a welter of semantics to replace the term already in use, I shall instead categorize the phenomenon of transdifferentiation by levels: primary, secondary, and tertiary transdifferentiation. Primary (or true) transdifferentiation would include the cell-type conversion or cell metaplasia that is so well documented to occur in some amphibian eye tissues in vitro and in amphibian (newt) eye tissues in situ (Fig. 1). This level is characterized by verifiably postmitotic cells, terminally differentiated and producing a specific cell product, transforming into a completely different cell type with differing cell product(s). Secondary transdifferentiation is marked by the conversion of those cells or tissues not definitely demonstrable as terminally differentiated, i.e., from an embryonic or possible stem-cell source. Also included is the concept of transdetermination (Hadorn, 1965), in which certain groups of cells in Drosophila occasionally become determined or committed to a developmental fate different from that expected. Tertiary transdifferentiation would encompass other purported/ reported changes of tissue types, e.g., that of muscle to cartilage (Namenwirth, 1974), and of striated to smooth muscle in Anthomedusa as reported by Schmid and Alder (1984) and Weber et al (1987). The well-known plasticity of plant tissues, especially in vitro, is the rule, rather than the exception, and as a topic of transdifferentiation is beyond the scope of this chapter.

For at least a century and a half, biologists have been aware that, relative to normal cells, many cancer cells appear morphologically undifferentiated. While some have concluded that cancer cells must arise by the dedifferentiation of normal cells, others consider it more likely that many cancer cells develop from normal cells that never fully differentiated. Since these undifferentiated cancer cells can be shown to share many fundamental characteristics with normal cells, it is not altogether surprising that some neoplasms can be induced to differentiate and become postmitotic benign cells. If only one or a few malignant tumors had the potentiality for differentiation, the phenomenon would be of interest primarily because of its rarity. However, many types of malignant neoplasms, such as plant teratomas, fish tumors, amphibian cancers, and malignant neoplasms of all three embryonic germ layers of mammals, can be induced to differentiate. This indicates that the phenomenon is not rare, and it suggests the possibility that the control of differentiation could be exploited in the treatment of cancer.

In retrospect, perhaps it is not surprising that in mammals the two parental genomes have different influences on the developing embryo. However, it is only recently that the functional nonequivalence of the maternal and paternal genomes has been established conclusively and so has commanded a considerable amount of interest. This chapter reviews experiments that have demonstrated the complementary roles of the two parental genomes in the mouse, thus establishing the existence of genomic imprinting in murine development. Also, we indicate some of the approaches being employed to elucidate the molecular basis of this phenomenon.

The free-living hermaphroditic soil nematode Caenorhabditis elegans has become a widely used organism for study of invertebrate development. Its advantages include ease of cultivation in the laboratory and a short life cycle of about 72 hr at 22°C. In addition, it is transparent at all stages, so that development can be followed microscopically in living specimens; it is also cellularly simple, having only 959 somatic cells in the adult hermaphrodite. These latter two features have permitted determination of the complete cell lineage, from the zygote to the adult animal (Sulston and Horvitz, 1977; Kimble and Hirsh, 1979; Sulston et al., 1983). The pattern and timing of cell divisions is almost completely invariant from animal to animal. Perhaps most important, C. eJegans is well suited to genetic analysis of development; mutational identification and mapping of about 700 loci on its six chromosomal linkage groups has already been accomplished (Wood et al., 1988).

Plant development differs from animal development in several fundamental respects. Since plant cells are immobilized in a rigid cell wall, morphogenesis is dependent upon control of cell growth and plane of cell division, rather than cell migration as occurs in animal development. The meristematic cells of plants display a degree of plasticity not found in animal cells; they remain embryonic throughout the life of the plant and produce both adult organs and germ cells. Furthermore, individual differentiated cells from vegetative plant parts are able to dedifferentiate and regenerate into new, fertile plants.

Seeds are complex structures, made up of genetically and physiologically distinct components whose functions change during development (Bewley and Black, 1978, 1982; Rubenstein et al, 1978; Johri, 1984; Murray, 1984). The regulation of gene expression can thus be expected to be equally complex, as each part of the seed differentiates according to its own program. Most of the research effort to date has been directed toward describing the kinds of genes expressed in seeds and their temporal and spatial patterns of expression. Our knowledge of factors controlling the observed patterns is limited and comes from genetic analyses, studies of the effects of environmental and hormonal perturbations, and in vitro culture of isolated seed parts. In no case has the product of a regulatory gene been isolated and characterized, and the mechanisms by which hormones and environmental stimuli exert their effects on specific genes remain obscure. This chapter describes the systems and approaches being used to study regulation of expression of genes in each part of the seed, with special emphasis on evidence for interactions between tissues that are likely to involve trans-acting regulatory molecules, even though the identities of such molecules are unknown. My hope is that by pointing to our areas of ignorance, more systematic research will be undertaken, resulting in some clear answers about gene regulation in seeds.