Molecular Biology of the Cell - Volume 3 - Alberts B., Bray D., Lewis J., Raff M., Roberts K., Watson J. 1994

From Cells to Multicellular Organisms
Cellular Mechanisms of Development
Developmental Programs of Individual Cells: Cell Lineage Analysis Using Nematodes as an Example

Possessing memory much like computers, Cells are capable of executing complex developmental programs. As a result of these programs, a complex adult animal body emerges from a multitude of cells, each operating according to its own program controlled during development. Some types of cellular behavior during development are quite autonomous, whereas others are determined by signals originating from surrounding cells. Thus, embryonic cells can be likened to a network of computers acting in parallel and exchanging information. Each Cell possesses an identical genome—that is, it contains the same program—yet this program can exist in a multitude of variants that direct development along different paths depending on the signals The Cell perceives from its environment.

Cellular memory plays an extraordinarily crucial role, which is precisely why cell behavior at a given time depends on the choices made by its ancestors in preceding division cycles. Thus, to fully understand the entire developmental program, it is necessary to know how cell divisions were carried out—in other words, the genealogy of the individual embryonic Cells must be known. This is The Essence of Lineage analysis, the execution of which, particularly in large and complexly organized animals (such as vertebrates), is a rather formidable task. If individual cells are specifically labeled at early Selection/3.html">Stages of development, the cells themselves or their progeny can be identified at later stages. One labeling method involves the microinjection of easily traceable specific molecules (such as fluorescent Dyes or the enzyme horseradish peroxidase; see Section 19.1.5) into the cells of early embryos (Fig. 16-29). Alternatively, individual embryonic cells can be genetically labeled, for example, by inducing their infection with a specially chosen retrovirus (see Section 17.5.4) or by exposing embryos to ionizing radiation that causes random somatic Mutations (see Section 16.5.13). These lineage analysis Methods are highly labor-intensive, and each experiment yields only limited information about the descendants of embryonic cells. The situation is further complicated by the fact that in vertebrates and many other organisms, heredity is subject to random Variability even in genetically identical animals.

Class="center">

Fig. 16-29. Tracing the cell lineage in the ascidian Halocynthia roretzi. A. An embryo at the 64-cell stage. At this stage, horseradish peroxidase is injected as a label into one of the blastomeres. This makes it possible to identify the descendants of this blastomere, much like what is shown in the diagram below the photograph. B. Larvae developing from 6 such embryos were subjected to a special Procedure to reveal the enzyme (shown in black) in the descendant Cells of the injected blastomeres. In the two larvae shown on the left (top and bottom), the ancestral blastomere that received the label was the upper one shown in red in Fig. A; in the two larvae shown in the middle, it was the lower of these blastomeres; for the two larvae shown on the right, the ancestral blastomere in Fig. A is not visible. In this species, specific PARTS OF THE embryo always develop from the same blastomeres (see photographs). (N. Nishida, Dev. Biol., 121, 526–541, 1987.)

However, in representatives of certain lower phyla, including Mollusks, Annelids, and nematodes, Cell Division and movement are highly ordered and occur identically in all individuals. Such complete reproducibility has been exploited in studies of the tiny, transparent nematode Caenorhabditis elegans. This animal is characterized by a simple and virtually invariant Structure, and its development can be traced cell by cell all the way from the egg to the adult Organism. Here, the complete lineage of every cell can be mapped out. Against this Background, the effects of mutations and other experimental interventions can be noted with great precision. Such an approach makes it possible to link specific genes to particular steps in the execution of the programs controlling cell development. Nevertheless, as we shall see, unraveling the internal logic of the program is by no means a simple task, even under such favorable conditions; its complexity is largely due to the presence of intercellular interactions. We will conclude this section with an example illustrating the application of culture experiments to directly analyze small fragments of the program governing The Development of individual mammalian cells.

16.3.1. Anatomically and genetically, the nematode Caenorhabditis elegans is remarkably simple [27]

The length of an adult C. elegans is about 1 mm. The worm's body consists of approximately 1,000 somatic cells and 1,000–2,000 Germ Cells (Fig. 16-30). Using serial-section Electron Microscopy, the animal's anatomy has been completely reconstructed, cell by cell. The overall body plan of this primitive worm is fundamentally similar to that of most higher animals. The elongated body exhibits bilateral Symmetry and is composed of conventional Tissues (nerves, Muscles, intestine, hypodermis) organized in a conventional manner (Mouth and Brain at the anterior end, anus at the posterior). The outer body wall consists of two layers: a protective hypodermis ("Skin") and an underlying Muscle layer. A simple tube formed by endodermal cells constitutes the intestine. Located between the intestine and the body wall is a second tube (the gonad), constructed of somatic cells and containing germ cells. Adult C. elegans occur in two forms—hermaphrodites and males. For simplicity, hermaphrodites can be viewed as females that produce a limited amount of sperm: a female can reproduce either by self-Fertilization using her own sperm or by cross-fertilization through mating with a male hermaphrodite. Self-fertilization predominantly yields homozygous progeny, making C. elegans an exceptionally advantageous subject for genetic research.

Fig. 16-30. Adult hermaphrodite of Caenorhabditis elegans (lateral view). (J. E. Sulston, H. R. Horvitz, Dev. Biol., 56, 110–156, 1977.)

The comparative simplicity of C. elegans anatomy is matched by an equally simple genetic apparatus. Its 6 pairs of homologous Chromosomes appear to contain only about 3,000 vital genes. The haploid genome contains 80 × 10⁶ nucleotide pairs, which is roughly 17 times larger than that of E. coli and 38 times smaller than that of humans. At present, approximately 800 genes have been identified through mutational analysis. These include genes affecting traits such as worm shape and behavior, genes encoding well-known Proteins like Myosin, and genes controlling the pattern and direction of development. In addition, a genomic library has been constructed in the form of a large set of overlapping DNA fragments (see Section 5.6.3).

16.3.2. Nematode development exhibits striking constancy [28]

Development in C. elegans begins with a single cell—the fertilized egg—which undergoes successive divisions; at the 558-cell stage, a small worm forms beneath the eggshell. Following hatching, cell divisions continue, leading to further growth and sexual maturation as the worm passes through 4 successive larval stages separated by molts. The final molt gives rise to the adult—a hermaphroditic worm that begins laying its own eggs. The complete developmental cycle from egg to egg takes about 3 days.

Because the worm's body is transparent, cell division, migration, and differentiation can be observed live (Fig. 16-31), and the genealogical relationships and behavior of all cells can be charted from the single-celled egg stage to the adult animal. These studies have demonstrated that the animal's somatic structures are generated according to a single invariant lineage scheme. Each of the numerous cell divisions occurs at a strictly predetermined time. This means that every precursor cell and all its descendants divide identically in all individuals (with very few exceptions). That is why, knowing a given cell's position on the lineage tree, one can predict its fate (Fig. 16-32). In contrast, the division of germ-line cells is not as rigidly programmed: after the worm hatches from the egg, two precursor cells of the primordial germ cells undergo several divisions, giving rise to A large number of descendants whose fate is determined by their position within the gonad.

Fig. 16-31. A region of the mid-ventral area of a living C. elegans larva, photographed four times in succession at intervals of a few minutes. A dividing hypodermal cell is visible. (Courtesy of John Sulston and Judith Kimble.)

A complete Description of the C. elegans lineage brings us closer to answering a fundamental question. Nematodes, like most animals, consist of a relatively large number of cells that can be classified into only a few types. Knowing that cells have different origins, it is tempting to yield to the temptation and explain the differences between them by descent from different clones; in other words, all cells of a given type might be descendants of a single "founder" cell and destined to develop along a specific pathway. However, lineage analysis shows that although individual cells become determined as precursors of a specific cell type at Cytology/cytology/16.html">Early stages of development, this rule generally does not hold true for either nematodes or other animals. Thus, in C. elegans (excluding intestinal and germ cells), cells of each differentiated type—such as hypodermal, neural, muscle, and gonadal cells—derive not from a single founder cell, but from several founder cells originating in independent Branches of the lineage tree (see Fig. 16-32). Consequently, cells with similar characteristics do not necessarily have to be closely related. Conversely (though rarely), cells that differ in multiple traits may turn out to be closely related; for instance, in C. elegans, certain Neurons are sister cells to muscle cells.

Therefore, our task is to understand the internal rules operating within each branch of the lineage tree, according to which a series of cell types, represented by a specific number of cells, arises.

16.3.3. Developmental control genes refine the program governing nematode cell lineage [29]

During development, much like a computer, a cell makes choices among various options: to divide or not to divide, to differentiate into a neuron or a muscle cell, to become a precursor of this or that branch of the lineage tree. To understand The Mechanism of such choices, it is necessary to identify the corresponding genes. Mutations in these genes cause developmental disruptions, but this is not the only category of mutations leading to such consequences. For example, certain mutations in vital genes can "truncate" the lineage tree by causing premature embryonic death. Mutations in genes governing the synthesis of specialized proteins result in some differentiated cell types functioning abnormally despite an outwardly normal body structure. In contrast to all the cases mentioned above, mutations in genes affecting the choice of developmental programs lead to defects in the overall body plan: as a result, normal differentiated cells will occupy abnormal positions within the organism. Alterations in the lineage tree can also manifest as the appearance of an abnormal number of certain cells. All these Types of mutations make it possible to identify genes that control development (Fig. 16-33).

Fig. 16-32. Lineage of the cells forming the intestine of C. elegans. The egg (top) is shown at the same scale as the adult (bottom). Note that the intestinal cells form only a single clone (as do the germ-line cells), whereas in most other tissues, cells can give rise to multiple clones.

Mutations in certain control genes lead to the substitution of one cell type for another during the late Stages of the developmental program's execution. In such instances, a choice made at earlier stages is altered, such that one branch of the tree is completely replaced by another. As shown in Fig. 16-34, a particular type of development can be recapitulated with high frequency across several independent branches of the lineage tree. Such a phenomenon implies the existence of a certain standard pattern implemented under various conditions. Overall, one gets the impression that each branch of the lineage tree is controlled by a complex combination of genes, many of which are also involved in controlling other branches.

Fig. 16-33. A simple cell lineage fragment and some types of variations that can occur as a result of mutations in developmental control genes. Analysis of mutant phenotypes makes it possible to determine what the normal function of the mutated genes was. Letter designations indicate the internal state of the cell.

Systems programmers are well aware that even minor modifications to a program can profoundly affect the outcome of its execution. Similarly, a mutation in a single control Gene leads to a drastic distortion of the lineage tree. This principle is well illustrated by so-called heterochronic mutations, as a result of which certain sets of cells behave according to rules operating at a different stage of normal development. For example, a daughter cell may behave like its mother or an even earlier precursor, and its descendants reproduce their characteristic phenotype, and so on. Thus, a fragment of the lineage tree is repeated multiple times, and the Development of the entire organism is disrupted. To explain this phenomenon, Fig. 16-35 illustrates the effects of a series of mutations in the lin-14 gene. Instead of following the normal pattern of Cell Differentiation characterized by the sequential transition through the 1st, 2nd, 3rd, and 4th larval stages followed by Cell Cycle arrest, many lin-14 mutant cells reiterate the pattern characteristic of the 1st larval stage, undergoing 5 to 6 molting cycles and continuing to produce immature-type cuticle. Other mutations in this gene have the opposite effect, forcing cells to reach maturity prematurely, accompanied by the loss of intermediate stages. As a result, the animal reaches the definitive stage possessing an abnormally small number of cells. Such premature development occurs in mutants characterized by a deficiency of normal lin-14 activity, whereas developmental delays are observed in mutants with an abnormally high level of this gene's activity. Thus, The Effect of the lin-14 gene product appears to be the maintenance of cells in a "young" state, and normal development presumably involves the gradual restriction of this product's synthesis as the animals mature.

Fig. 16-34. Pedigree diagrams illustrating how mutations in developmental control genes (in this case, the *lin-22* gene) can lead to identical transformations of one cell lineage variant into another across several distinct branches of the cell lineage tree. The descendant cell lineages are shown for each of the 6 larval hypodermal precursor cells of *C. elegans*; the positions of these cells are indicated at the top in the diagram of a young larva. Crosses mark instances of programmed cell death (apoptosis), a fairly common occurrence during the normal development of *C. elegans* and other species. (Y. R. Horvitz et al., *Cold Spring Harbor Symp. Quant. Biol.*, 48, 453–463, 1983.)

Fig. 16-35. Heterochronic mutations of the *lin-14* gene in *C. elegans* and their effect on one of the cell lineages they influence. A loss-of-function (recessive) mutation in this gene causes cells to divide and differentiate in the same manner as in late-stage larvae; gain-of-function (dominant) mutations of this gene produce the opposite effect. Crosses indicate instances of programmed cell death. (V. Ambros, H. R. Horvitz, *Science*, 226, 409–416, 1984.)

16.3.4. The Differentiation Program Is Coordinated with the Cell Division Program [30]

In studying heterochronic mutations, researchers encountered yet another complex problem. The fact is that during development, The Genome orchestrates the refinement of both the cell division program and the cell differentiation program, and these two manifestations of cellular behavior must be precisely synchronized. In heterochronic mutants, both division and differentiation are disrupted. This suggests that both processes are regulated by a single mechanism that has been impaired as a result of mutation.

One might assume that the cell cycle plays The Role of such this mechanism. However, the facts do not support this hypothesis: the differentiation of early embryonic cells proceeds according to its established pattern even when cell divisions are artificially restricted by chemical agents that inhibit cytokinesis or DNA Synthesis. Cell divisions should not be likened to the swinging pendulum of a biochemical clock that dictates The rate of development; rather, the situation is precisely the reverse—it is the biochemical clock that controls the rate of cell divisions and the duration of the cell cycle in many animal species. A change in the chemical state of a cell simultaneously influences decisions regarding cell division as well as the timing and direction of differentiation. The molecular mechanisms controlling cell divisions in Embryogenesis remain virtually unstudied and represent one of the central PROBLEMS OF MODERN developmental biology. Nematode lineage mutants may play a pivotal role in solving this problem.

16.3.5. Autonomous Cell Behavior and Cell-Cell Interactions Are Interconnected and Shape the Character of Development [31]

Developmental processes are difficult to explain without data on the degree of autonomy in individual cell behavior and THE CONTRIBUTION OF cell-cell interactions. By observing the consequences of cell ablation in nematodes, one can assess The Importance of signaling exchange between cells. To achieve this, a laser beam is focused to a spot 0.5 µm in diameter (the average diameter of a *C. elegans* nucleus is about 2 µm), and repeated pulses are used to destroy The Nucleus of a specific cell. The cell with the destroyed nucleus dies, while its surrounding neighboring cells continue to develop. The obvious Conclusion is that the majority of nematode cells follow their own developmental programs for a prolonged period of their life cycle and are independent of signals from neighboring cells.

Nevertheless, intercellular signals play a vital role in the development of *C. elegans* and other animals. A prime example is the development of the Vagina, or vulva—the opening in the hypodermis (skin) of hermaphrodites used for egg-laying. The vulva is located on the ventral side of the body and is formed from 22 cells derived from three hypodermal precursor cells through a specific lineage pathway. One of the gonadal cells, known as the anchor cell, attaches or "anchors" the overlying gonad (Uterus) to the developing vulva, establishing a pathway for eggs to exit the uterus into the external environment. Results from laser ablation experiments demonstrate that this particular cell induces vulval development from the three nearest hypodermal cells. Following the destruction of the anchor cell, a patch of normal hypodermis develops from these cells instead of a vulva. Thus, the anchor cell induces vulval differentiation in *C. elegans*, much like vegetal blastomeres induce mesoderm differentiation in early *Xenopus* embryos. A single anchor cell is sufficient for this induction: normal vulval development proceeds even when all gonadal cells except the anchor cell are destroyed (see Fig. 16-36A).

Fig. 16-36. A. Experimental design demonstrating that the inductive Influence of the anchor cell is required for vulval development. B. Enlarged view of the six cells of the vulval equivalence group (shown in color) in the ventral hypodermis, with the normal lineage of their descendants shown at the bottom. All 6 of these cells (and only these) are capable of responding to the inductive signal from the vulva, but under normal conditions, only 3 of them actually do so.

The inducing signal from the anchor cell ensures that vulval development is spatially coordinated with THE POSITION OF the gonad, yet this system possesses a certain degree of flexibility. The three hypodermal cells normally involved in vulval formation are adjacent to three other cells that are equally capable of performing this task if they fall under METABOLISM/18.html">The Influence of the anchor cell. When the vulval precursor cells are destroyed by a laser beam, these neighboring cells deviate from their normal developmental path—likely due to a change in position—and form a vulva instead of hypodermis. Other cells are incapable of responding to the inductive signal and cannot participate in vulval formation under any circumstances. We can introduce THE CONCEPT OF an equivalence group, exemplified in the *C. elegans* model (Fig. 16-36B) by the three normal vulval precursor cells and the three neighboring cells that are induced in a similar manner, as these cells are functionally interchangeable. All members of the equivalence group differ from surrounding cells in that they possess a distinct developmental history. It is this circumstance that grants such cells The unique ability to respond to the inducing signal. The cells of this group acquire their differences under the influence of the inducing signal. Here we encounter a striking example of how the cellular developmental program relies on the interplay between cellular memory (an autonomous cellular property) and signals received from other cells (external control).

16.3.6. Experiments Help Clarify the Role of Genes Controlling Development [32]

The experiments described in the previous section help interpret genetic data. Mutations in any of 5 different genes can lead to a *multivulva* phenotype, in which all 6 cells of the equivalence group enter the vulval developmental pathway, resulting in 48 cells instead of the usual 22, and consequently, multiple vulvae. In these mutants, all 6 cells behave as though they are locked in a state of activation by the anchor cell, and the destruction of the latter does not alter the course of vulval development. In this case, the mutations likely affect genes that normally function within the cells of the vulval equivalence group to determine their response to the anchor cell signal. Mutations in another small set of genes result in an opposite, *vulvaless* phenotype. Here too, the ability of the equivalence group cells to respond to induction is presumably altered.

Mutations in the *lin-12* gene (which causes The formation of multiple vulvae) exert a pleiotropic effect: they can result in either the absence or the duplication of the anchor cell, thereby altering the distribution of the inducing signal and the rules by which the cells of the vulval equivalence group respond to that signal. *lin-12* mutations also affect other equivalence groups, suggesting a universal significance for the *lin-12* gene in intercellular interactions through which cells of a given equivalence group are directed down distinct developmental pathways. The nucleotide sequence of the *lin-12* gene was determined, and it turned out that the protein encoded by this gene exhibits Homology with a family of proteins believed to participate in cell-to-cell communication in vertebrates and insects (see Section 12.3.12). This family includes mammalian epidermal growth factor (EGF), certain mammalian cell surface receptors such as the low-density lipoprotein (LDL) receptor (see Section 6.5.8), and the product of the *Drosophila* *Notch* gene, which facilitates the choice between ectodermal developmental pathways: neural versus epidermal differentiation. These findings suggest that the molecular lessons learned from studying nematode development may find application in understanding the development of many other species.

16.3.7. The Developmental Program of Individual Mammalian Cells Can Be Analyzed in Cell Culture: Differentiation of Rat Optic nerve Glial Elements [33]

The most direct way to distinguish between autonomous cell behavior and behavior governed by cell-cell interactions is to investigate how transplanting cells to a different region of the body or isolating them to artificially alter their environment affects their behavior. These questions are difficult to study in *C. elegans* because laser cell ablation does not readily lend itself to this experimental design. The lack of information about cell-cell interactions during the development of this nematode complicates The Study of fundamental cellular control programs, despite the well-characterized cell lineage and the availability of many developmental mutants. The cellular environment is much easier to manipulate in culture. This provides an opportunity for the direct analysis of both the programs governing cell behavior and the intercellular interactions that drive cell division and differentiation during the development of various vertebrate body parts.

The optic nerve is one of the simplest elements of the mammalian Central Nervous system and is widely used as a model system for such studies. It contains long axons of retinal ganglion cells extending from the eye to the brain. Structural and functional support for axons within the nerve is provided by Three types of glial cells (see Section 19.1.6): oligodendrocytes and Two Types of astrocytes, known simply as type 1 and type 2 astrocytes (Fig. 16-37). All three types of glial cells are distinguished using Antibodies; this same method is used to differentiate glial cells from precursor cells. Investigations of cells isolated from the optic nerve at various developmental stages and placed in culture have demonstrated that the three glial cell types (which normally divide rarely, if at all) arise at different times and originate from two distinct branches of the lineage tree. Type 1 astrocytes arise prenatally from a single type of precursor cell, whereas oligodendrocytes and type 2 astrocytes form postnatally from another type of precursor cell, referred to as the "O-2A progenitor cell."

Fig. 16-37. The rat optic nerve contains retinal neuron axons (only one such cell is shown) and 3 types of glial cells. Type 1 astrocytes provide structural support for the nerve, playing a role analogous to that of Connective Tissue cells in a limb. Oligodendrocytes and type 2 astrocytes cooperate to form a sheath around each axon: oligodendrocytes wrap around the axon to form an insulating myelin layer (see Section 19.2.11), while type 2 astrocytes extend fine processes that contact the axons at breaks in the myelin sheath, forming the nodes of Ranvier. As a Nerve Impulse travels along the axon, it propagates from one node of Ranvier to the next (see Section 19.2.4).

Oligodendrocytes normally appear around the time of birth, and type 2 astrocytes emerge a week after birth.

Having elucidated the lineage of these cells and the timing of their appearance, researchers focused on the rules governing cellular behavior. For example, what triggers O-2A progenitor cells to differentiate, and why do they become oligodendrocytes if they differentiate at one time, but type 2 astrocytes if they differentiate at another? To what extent are these events driven by an intrinsic autonomous cellular clock mechanism, and to what extent does differentiation depend on the timely receipt of external signals?

If O-2A progenitor cells are isolated from the embryonic optic nerve and cultured in the absence of type 1 astrocytes (which are their normal neighbors), the O-2A cells cease dividing and prematurely undergo differentiation; however, when type 1 astrocytes are introduced into this culture, O-2A cell division halts, and their differentiation is delayed until a time corresponding to their in vivo birthday. The effect of type 1 astrocytes was found to be mediated by a secreted growth factor. This growth factor is believed to correspond to platelet-derived growth factor (PDGF; see Section 13.3.4). The presence of this factor prevents the premature differentiation of O-2A progenitor cells; however, increasing the concentration of this factor does not delay differentiation beyond the normal schedule. Presumably, embryonic optic nerve O-2A progenitor cells have previously adopted a developmental program whereby, after a certain time elapses or a specific number of cell divisions occur, the cells lose their responsiveness to the growth factor and differentiate regardless of its presence. Individual O-2A progenitor cells adopt this program at different times, and the earliest ones normally begin to differentiate at birth.

What determines the specific pathway of cell differentiation into an oligodendrocyte versus a type 2 astrocyte? When cultured in a serum-free medium, O-2A cells differentiate exclusively into oligodendrocytes: for them, this is the constitutive developmental pathway, or what in information terms is called the "default" pathway. In contrast, the differentiation of type 2 astrocytes appears to be an induced pathway: cells initiate this differentiation in response to a specific protein that can be isolated from the rat optic nerve. This protein is synthesized in significant quantities starting from the second week of development—precisely when type 2 astrocyte differentiation begins. It turns out that this protein is identical to ciliary neurotrophic factor (CNTF; see Section 13.3.4). This suggests that the timing of CNTF synthesis dictates the timing of type 2 astrocyte differentiation in the developing nerve. It remains unclear which specific cells synthesize CNTF in vivo and what factors control the precise timing of the onset of its synthesis, but it is known that in culture, this factor can be synthesized by type 1 astrocytes.

These research findings further support the hypothesis that cell development results from an interplay between autonomous processes and environmentally signaled behavior. Here, we have demonstrated the advantages of the cell culture method for studying the Molecular Basis of these processes. These findings can be summarized as an Information Flow diagram which, in computer science terms, represents the internal program of O-2A progenitor cells—that is, a set of rules by which cells interpret environmental signals to determine the timing and Location of differentiation (Fig. 16-38). Such data on individual cell behavior are crucial if we are truly to understand the principles of embryonic development. While we cannot yet answer the fundamental question of how the spatial distribution of numerous cells arises throughout the entire organism, we will return to this problem in the next section.

Fig. 16-38. Diagram illustrating a simplified model of intercellular control as a computer program. This program is likely involved in regulating cell division and differentiation of oligodendrocyte-type 2 astrocyte (O-2A) progenitor cells in the developing mammalian optic nerve. Broad arrows represent data input and output, i.e., extracellular signals received or emitted by the cell. Incomplete or putative sections of the diagram are indicated by dashed lines; the corresponding broad arrow pointing to the left demonstrates the possibility of feedback from O-2A progenitor cells, which could help modulate signals originating from type 1 astrocyte progenitor cells. O-2A progenitor cells are thought to continuously renew themselves in vivo, beginning roughly in the final week before birth (not shown).

Conclusions

To understand the developmental programs of individual embryonic cells, it is necessary to trace their history through cell lineage analysis. Vertebrate genealogies are characterized by stochastic variability. However, in certain nematodes and a few other invertebrate groups, the pattern of cell division during development is controlled so precisely and predictably that cells occupying specific positions in the animal's body follow the exact same developmental pathway in all individuals of a given species. The normal fate of every cell in the nematode Caenorhabditis elegans has been tracked throughout its entire development, and the consequences of experimental manipulation at the level of individual, identifiable cells have been investigated. Many genes that determine the execution of the cell development program have been identified by studying alterations in cell lineages caused by mutations in these genes. In general, mutations in development-controlling genes have coordinated effects on both cell differentiation and division. These observations suggest that these cellular processes are governed by a common underlying mechanism. Laser ablation experiments revealed that the developmental invariance of C. elegans results from a precise interplay between intercellular interactions and autonomous processes operating within individual cells. For example, observing vulval development allows researchers to identify inductive interactions that may be linked to the action of specific developmental control genes.

Another approach to analyzing the programs controlling individual cell development is the Study of Cellular behavior in culture. This approach has been used to uncover the rules governing cell differentiation and division during the development of mammalian optic nerve glial cells. This work provides further evidence that cell behavior is regulated not only by intercellular signals, but also by autonomous developmental programs intrinsic to individual cells.