Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 108
Scientific Frontiers in Developmental Toxicology and Risk Assessment 6 Recent Advances in Developmental Biology The absence of an incisive understanding of the action of toxicants on development has been in large part attributable to the absence of understanding of development itself. Until a few years ago, there was no understanding of a “developmental mechanism” at the molecular level although there were explanations at the cellular and tissue levels, such as “gastrulation is the mechanism by which the organization of the egg is transformed into the organization of the embryo.” Recent advances in developmental biology have been substantial enough for scientists to be confident for the first time that some aspects of development in some organisms are understood at the molecular level. Protein components are identified, their functions in developmental processes are known, and the time and place in the embryo of expression of the genes encoding them are known. This knowledge greatly benefits elucidating the mechanisms of developmental toxicity. In this chapter, the committee, in response to its charge, evaluates the state of the science for elucidating mechanisms of developmental toxicity and presents insights of developmental biology. It will show the promise of the subject in the next decade for understanding the action of developmental toxicants. A BRIEF HISTORY OF DEVELOPMENTAL BIOLOGY Observations of embryos and embryonic stages were made and recorded in antiquity (e.g., Aristotle, fourth century BC) and with increasing attention in recent centuries (e.g., Malphigi in the 1600s, Wolff in the 1700s, and von Baer in the early 1800s). However, it was only in the late nineteenth century that scientists pursued a detailed description of the embryonic stages of a variety of verte-
OCR for page 109
Scientific Frontiers in Developmental Toxicology and Risk Assessment brates and invertebrates, aided by the then-recent improvements in light microscopy and in staining methods and stimulated by Darwin’s proposals that the study of ontogeny (i.e., the animal’s embryonic development) holds clues to phylogeny (i.e., its evolutionary origin). Among the highlights during the period of 1880-1940 were the detailed anatomical descriptions of developmental stages of embryos, including the first atlas of human embryos, reconstructed from microscopic sections, published by W. His, Sr., in 1880-1885. In vertebrate embryology, these descriptions revealed the organogenesis of the heart, kidney, limbs, central nervous system (CNS), and eyes. Developmental-fate mapping studies revealed the embryonic sites of the origin of cells of the organs and the rearrangements of groups of cells in morphogenesis. The stages of development were found to include, in reverse order, cytodifferentiation, organogenesis, morphogenesis (gastrulation and neurulation), rapid cleavage, fertilization, and gametogenesis. By the 1940s, anatomical descriptions of the embryos of related animals were integrated into coherent evolutionary schemes, taught in comparative embryology classes, revealing, for example, the modification of the gill slits of jawless fish to the jaw of jawed fish and further modification to the middle ear of mammals. Also, by this time, Haeckel’s oversimplified scheme had been abandoned, namely, that ontogeny merely recapitulates phylogeny. Experimental embryology also began in the late 1800s. In experimental studies, which mostly involved techniques of cell and tissue transplantation and removal, the central role of cytoplasmic localizations and cell-lineage-restricted developmental fates was recognized in the development of certain invertebrates by the early 1900s. In vertebrate development, the importance of inductions (also called tissue interactions) was recognized in the 1920s, following the stunning organizer transplantation experiments by Spemann and Mangold (1924) on newt embryos. By the 1950s, inductions had been found in every stage and place in the vertebrate embryo, for example, in all the kinds of organogenesis. Vertebrate development, including that of mammals, had become comprehensible as a branching succession of inductive interactions among neighboring members of an increasingly large number of different cell groups of the embryo. Developmental mechanisms, as understood even in the 1970s, were descriptions of the movements and interactions of cells or groups of cells. They were cellular- or tissue-level mechanisms. The all-important “inducers” were materials of unknown composition released by one cell group and received by another group. Consequently, the recipient cells took a path of development different from the one that would have been taken if they were unexposed. The progression or momentum of development also was recognized: that the individual events of interactions and responses are time-critical, and that certain subsequent aspects of development never occur if one event is prevented. Molecular mechanisms, however, were not understood at that time. Embryologists encountered the limits of the field in the 1940-1970 period, as they tried to discover the chemical nature of inducers and the responses of cells to them.
OCR for page 110
Scientific Frontiers in Developmental Toxicology and Risk Assessment The basic information and methods of biochemistry, molecular biology, cell biology, and genetics were not yet available to analyze cell-cell signaling and transcriptional regulation in embryos. In light of discouraging results, some embryologists considered that the organizer concept was faulty and that inducers were an experimental artifact (see later discussion for recent successes in understanding inductions). Although Morgan and other early geneticists had proposed that inducers and cytoplasmic localizations elicit specific gene expression and that development was in large part a problem of ever-changing patterns of gene expression (Morgan 1934), the means were not at hand to pursue those insights. Roux, Spemann, and Harrison had outlined plausible lines of inquiry into determination and morphogenesis in the early part of the twentieth century; however, the means were also not available to pursue those questions at that time. To many scientists in the 1940-1970 period, the study of development seemed messy and intractable. Researchers turned to more informative subjects such as the new molecular genetics of bacteria and phages (viruses that infect bacteria). From those inquiries came new insights in the 1950-1965 period on the nature of the gene and the code and the processes of replication, transcription, translation, enzyme induction, and enzyme repression. For example, it was only in 1961 that Monod and Jacob described gene regulation in bacteria in terms of promoters, operators, and repressor proteins (Monod and Jacob 1961). Those authors immediately saw the relevance to animal development. All of their insights made possible the invention of techniques for gene isolation and amplification, for in vitro expression of genes, for genome analysis, and, thereafter, for the new developmental biology. With so little molecular information about developmental processes, there was scarcely any understanding of the action of developmental toxicants. For example, Wilson (1973) in his book Environment and Birth Defects could only raise the following possibilities for connections between inductions and developmental defects: It has long been accepted that cell interactions (induction) are an important part of normal embryogenesis, despite the fact that specific “inducer substances” have not been identified. [Failures] of normal interactions which may lead to deviations in development include, for example, lack of usual contact or proximity, as of optic vesicle with presumptive lens ectoderm; or the incompetence of target tissue to be activated in spite of its usual relationship with activator tissue, as in certain mutant limb defects; or the inappropriate timing of the interrelation, even though all parts are potentially competent. That the nature of cell-to-cell contacts and the manner of their adhesion are important determinants in both normal and abnormal development has been demonstrated…. Insufficient or inappropriate cellular interactions usually result in arrested or deviant development in the tissue ordinarily induced or activated by the interaction. This committee will later argue that Wilson’s insight was well directed and is now ready to be pursued.
OCR for page 111
Scientific Frontiers in Developmental Toxicology and Risk Assessment ADVANCES IN DEVELOPMENTAL BIOLOGY In the past 15 years, developmental biology has advanced remarkably, perhaps as at no other time in the field’s history. It is now known that the trillions of cells of a large animal, such as a mammal, have the same genotype, which is the same as that of the single-celled zygote (the fertilized egg) from which the animal develops. That is to say, the genetic content of somatic cells does not change during the development of most animals. The recent clonings of Dolly the lamb (Wilmut et al. 1997), the Cumulina mouse family (Wakayama et al. 1998), and a nonhuman primate (Chan et al. 2000) reaffirm the fact that a specialized cell, such as a mammary or cumulus cell, carries the genes for all other kinds of cells of the animal. The scientific advances that led to these clonings were built on earlier nuclear transplantation successes in frogs, first by Briggs and King (1952), but particularly by Gurdon (1960), which had led to similar conclusions for a nonmammalian vertebrate. Despite the same genes, the cells within the individual organism differ greatly in their appearance and functions, meaning that they have the same genotype and different phenotypes. The cell types differ greatly in the ribonucleic acids (RNAs) and proteins contained within them. They differ in which subset of genes they express from their total genomic repertoire. At least 300 cell types are recognized in humans (e.g., red blood cells, Purkinje nerve cells, and smooth or striated muscle cells). The number of cell subtypes is much larger, perhaps numbering tens of thousands, when further differences are taken into account related to the cell’s stage of development and location in the body, as has been discovered in recent years. Development can be viewed as evolution’s crowning example of complex gene regulation. From the single genome, thousands of different gene combinations must be expressed at specific times and places in the developing organism, and from the developing egg the information for the selective use of combinations must be generated. A major factor in this regulation is the transfer of chemical information (i.e., signals) between cells during development. From recent research, which has built on earlier findings, the following is now realized: Embryonic cells of arthropods and nematodes make many of their developmental decisions based on which chemical signals they receive from other cells just as vertebrate embryonic cells do. Later the embryonic cells of all these organisms will make further decisions based on other signals. The cycles of signaling and responding are repeated over and over as development progresses. With that in mind and the fact that one genotype supports hundreds or thousands of cellular phenotypes, development can be said to rely on “genotype-environment interactions,” where the local environment of each cell is generated by neighboring groups of cells. The genotype and cell’s previous developmental decisions determine its options for responses to the signals currently present (Wolpert 1969). The signaling pathways involved in this information transfer are known to be of 17 types (a few more may remain undiscovered). They are used repeatedly
OCR for page 112
Scientific Frontiers in Developmental Toxicology and Risk Assessment at different times and places in the embryo, from the earliest stages through organogenesis and cytodifferentiation, and even in the various proliferating and renewing tissues of the juvenile and adult (see Appendix C). The signaling pathways are highly conserved across a wide range of phyla of animals (from chordates to arthropods to roundworms), presumably because they were present and already functional in the pre-Cambrian common ancestor of those animals. Many of the kinds of cell responses to signals also are conserved (e.g., responses of selective gene expression, secretion, cell proliferation, or cell migration). The response of developing cells to signals involves activation or repression of the expression of specific genes by transcription factors contained within genetic regulatory circuits. Signaling pathways frequently affect the activity of those factors. Many of the transcription factors and circuits are conserved across a wide range of phyla of animals. Thus, an effective and general approach to the experimental analysis of developmental processes at all stages has been to inquire about the signaling pathways and transcriptional regulatory circuits that operate in the particular instance of development under study. Different organisms, which differ in aspects of their development, nonetheless use the same conserved signaling pathways and regulatory circuits, but in different combinations, times, and places, and have different genes as the targets of their transcriptional regulatory circuits. Processes of development, which seemed to confront scientists with infinite complexity and variety just a few years ago, now seem interpretable as composites of a small number of conserved elemental processes, namely, those of intercellular signaling, intracellular regulatory circuits, and a limited variety of targeted responses. These conclusions, which were reached by the analysis of development in animals as remote as mice, flies, and nematodes, give great validity to the use of model organisms in studying mammalian development, including that of humans, and in the future analysis of the action of developmental toxicants and in their detection. Although the signal-response pathways are highly conserved, evolution has produced an increasing complexity of the “community” of pathways in vertebrates. This complexity is evident both in the increased number of closely related pathway components (diversifed protein family members) and in the increased possibilities for cross-talk among pathways. The redundant function of closely related components was made evident by existence of numerous targeted gene-knockout mutations in the mouse that produced little or no identifiable pheno-types—that is, the mice are normal or nearly normal under laboratory conditions (see Table 6-5, later in this chapter). It must be emphasized, however, that functional redundancy provides two advantages. It protects the organism by ensuring that a fundamental process can proceed even in the absence or reduced presence of a critical gene activity. On an evolutionary scale, the multiplicity of overlap-
OCR for page 113
Scientific Frontiers in Developmental Toxicology and Risk Assessment ping functions provides a basis for generating diversity without losing essential functionality. The Drosophila Breakthrough The recent molecular understanding of developmental processes and components was gained from the experimental analysis of a few model organisms such as Drosophila melanogaster (the fruit fly), Caenorhabditis elegans (a free-living nematode), Danio rerio (the zebrafish), Xenopus laevis (a frog), the chick, and the mouse (see Chapter 7 for proposals about their use in the assessment of developmental toxicities). D. melanogaster and C. elegans were chosen by researchers for their amenability to genetic analysis, afforded by their small size (hence, large populations) and short life cycle (hence, many generations). Nüsslein-Volhard and Wieschaus (1980) began a systematic search for developmental mutants of Drosophila in the mid-1970s. They submitted adults to high-frequency chemical mutagenesis and then inspected large populations of offspring for mutant individuals with strong and early developmental defects (before hatching) at discrete locations and discrete stages in the embryo. They discarded mutants with weak or pleiotropic effects as ones too difficult to analyze at their start. They examined mutagenized flies until the same kinds of mutants began to appear repeatedly in their collections. The recurrence was evidence that they had obtained all the different kinds of zygotic mutants (those affected in genes transcribed after fertilization) that mutagenized flies could yield under the conditions of inspection. This procedure is called “saturation mutagenesis,” in which all the susceptible genes whose encoded products are important in development are thought to be revealed. Several laboratories, including those of Nüsslein-Volhard and Wieschaus, were also collecting maternal-effect mutants (those affected in genes transcribed in female germ cells before fertilization) and pursued this search to saturation. The Drosophila mutants were categorized by phenotype and complementation behavior (putting two mutations together in a heterozygote to see whether they are alike or different) to establish the number of different genes whose mutations give the same phenotypic defect of development. Their categories included those embryos failing to develop the anterior or posterior end, odd or even segments, dorsal or ventral parts, mesoderm, endoderm, or nervous system. Further mutant combinations were made to establish epistasis (the interaction of different gene products, reflected in the dominance of one mutant defect over another) and to deduce plausible developmental pathways in which the actions of the encoded gene products could be related and ordered. By the late 1980s, a solid base of observations of Drosophila mutant phenotypes and gene locations had been built, and ordered pathways of function based on the mutant interactions had been proposed. This information served as the foundation for future molecular genetic analysis. The research was the first systematic and exhaustive approach to under-
OCR for page 114
Scientific Frontiers in Developmental Toxicology and Risk Assessment standing an organism’s development and to identifying components of developmental processes. Synergy with Research Advances in Other Areas Meanwhile, other researchers worldwide made advances in biochemistry, molecular biology, cell biology, and genetics. They learned an enormous amount about the function of proteins in replication, transcription, translation, secretion, uptake, membrane trafficking, cell motility, cell division, the cell cycle, cell adhesion, and apoptosis (programmed cell death), to mention but a few of the cellular processes. Researchers improved the methods to isolate genes, sequence them, manipulate sequences, make transcripts in vitro, detect messenger (m)RNAs in cells by in situ hybridization, translate RNAs to proteins in vitro, and make antibodies to proteins. In situ hybridization, which graphically revealed the time and place of expression of specific genes in the embryo, was to prove particularly important for connecting the new molecular analysis to the older developmental anatomy. Much of the work was initially done with single-celled organisms: bacteria, yeast, or animal cells in culture. Some insights and techniques came from the study of cancer cells in the search for oncogenes. In the course of that work, many of the processes, protein functions, and protein sequences were found to be strongly conserved among organisms as diverse as yeast and humans or even bacteria and humans. Various proteins of different organisms, and also within the same organism, shared “sequence motifs” by which the protein could be recognized as a member of a protein family with a particular function and descended from a common sequence ancestor. Newly discovered proteins could be assigned a function from just their possession of a particular motif. As more motifs were found, it will be easier to categorize newly discovered proteins. For example, receptor tyrosine kinases were recognizable by their transmembrane hydrophobic motifs and adenosine triphosphate (ATP)-binding domains. G-protein-linked receptors could be distinguished by a seven-pass (serpentine) transmembrane motif. Transcription factors could be recognized by the sequence motifs of their deoxyribnucleic acid (DNA) binding domains (e.g., zinc finger, basic helix-loop-helix, homeodomain, or leucine zipper domains). Of the recently sequenced genomes of yeast and C. elegans, for example, about 40% of the open reading frames (ORFs) are recognizable by known motifs (Chervitz et al. 1998). Function can be assigned, at least preliminarily, to the products of those genes. Plans are afoot to define the function of the missing ORFs of yeast and make the functions of all proteins assignable from sequence. At the same time, there are plans to identify a large number of protein-binding sequences in the regulatory regions of genes to be able to predict better the conditions of expression of genes. These plans are among the aims of “functional genomics,” as described in Chapter 5. All the information on sequences, motifs, and function is stored in databases available
OCR for page 115
Scientific Frontiers in Developmental Toxicology and Risk Assessment to researchers worldwide (e.g., the Basic Local Alignment Search Tool (BLAST) <http://www.ncbi.nlm.nih.gov/BLAST/>). Drosophila Development at the Molecular Genetic Level By the time the Drosophila mutants were characterized in the mid-1980s, techniques were well-suited for molecular genetic analysis of affected genes and gene products. This part of the work moved quickly, thanks to gene-cloning techniques, background information about gene sequence motifs and protein function, and databases available to researchers worldwide. The successful isolation of a gene responsible for a developmental phenotype (when the gene was mutated) could be validated by the rescue of the mutant phenotype by transformation with the wild-type gene (usually as DNA included in a P-element transposon). In situ hybridization, coupled with color stains, readily revealed the normal time and place of expression of the specific genes whose mutations had been isolated. Regarding the function of these developmental genes, many were found to encode proteins with familiar motifs, such as those for receptor tyrosine kinases or various transcription factors. In fact, a surprisingly large number turned out to be transcriptional regulators. Function could be rapidly concluded from sequence data. Other Drosophila genes encoded proteins whose specific functions were unknown, yet they were recognizable generally as secreted proteins by their signal sequences or as new transcription factors by the fact they accumulated in nuclei and could bind to DNA. In the course of this analysis, new intercellular signaling pathways were discovered, such as those involving the Decapentaplegic (DPP), Hedgehog (HH), Wingless (WG), and Notch/Delta ligands. (The whimsical names are those given by researchers to mutants based on the phenotypes.) Hundreds of laboratories worldwide joined the work on Drosophila mutants, and the picture of early development took on a satisfying coherence and clarity, especially the steps of generation of segmentation and of the overall body organization in the anteroposterior and dorsoventral dimensions. These steps of early development are known collectively as “axis specification.” The following is a brief summary of that picture to illustrate its completeness at the molecular level. The steps are stage-specific mechanisms of development. The mechanisms are now better understood in Drosophila than in any other organism. It is the kind of information scientists would like to have, but do not yet have, for mammalian development. At the start of Drosophila development, the oocyte is provisioned with hundreds of maternal gene products that are uniformly distributed in the egg during oogenesis. Four gene products are spatially localized in the egg, however, and they provide the initial asymmetries on which the entire anteroposterior and dorsoventral organization of the embryo is built stepwise in development after fertilization. The four gene products include the following:
OCR for page 116
Scientific Frontiers in Developmental Toxicology and Risk Assessment An mRNA located internally at the anterior end (encoding a transcription factor, named Bicoid). An mRNA located internally at the posterior end (encoding an inhibitor of the translation of the mRNA for a transcription factor, named Nanos). An external protein anchored to the egg shell at both ends of the egg (involved in the production of a ligand of a receptor tyrosine kinase in the egg-cell plasma membrane). An external protein also anchored to the egg shell but at the prospective ventral side (involved in the production of a signal ligand of the Toll receptor in the egg-cell plasma membrane). To exemplify the steps of use of those gene products, only one of the dimensions, the anteroposterior, will be described. The two mRNAs are initially at opposite ends of the egg. They are translated after fertilization, and the encoded proteins diffuse from the ends to form opposing gradients reaching to the middle of the egg. These proteins will act in concert to generate a gradient, high at the anterior end and low at the posterior end, of another transcription factor. The nuclear number increases rapidly in the uncleaved cytoplasm. The graded transcription factors, called members of the “coordinate class” or “egg-polarity class” of gene products, activate at least eight gap genes in nuclei along the egg’s length at different positions, each position unique in terms of the local quantity of transcription factors of the coordinate class. (The terms “coordinate,” “egg polarity,” and “gap” also derive from mutant phenotypes.) The encoded gap proteins, which are all transcription factors themselves, accumulate in a pattern of eight broad and partially overlapping stripes along the egg’s length. The proliferating nuclei are not yet separated by cell membranes—that comes later. These proteins in turn activate at least eight pair-rule genes, all of which also encode transcription factors. Complex cis-regulatory regions of the various pair-rule genes define their expression responses to the spatially distributed gap proteins. The pair-rule proteins then activate at least 12 segment-polarity genes, some of which encode transcription factors and some of which encode secreted protein signals. The pair-rule and gap proteins together also activate eight homeobox (Hox) genes to be expressed in broad stripes, as discussed in the next section. Thus, the early steps of development involve cascades of transcription factors distributed in space according to the initial gradients of a few agents and to the expression rules contained in the complex cis-regulatory regions of genes for yet other transcription factors. These key steps are accomplished in the first 3 hours of development, mostly before cell membranes are formed and gastrulation begins, although the final elaboration of the segment-polarity and Hox genes occurs after cells form. Once the segment-polarity genes and Hox genes are activated, they maintain their expression in cells by an auto-activating circuitry, in some cases by the encoded transcription factor activating expression of its own gene. The coordi-
OCR for page 117
Scientific Frontiers in Developmental Toxicology and Risk Assessment nate, gap, and pair-rule proteins are then no longer needed. Their products disappear, and the genes are no longer expressed. Similar conclusions apply to the development of the termini and the dorsoventral dimension, which also rely on initially asymmetric signals. The developmental mechanisms of the termini and dorsoventral dimension are of additional interest, because the signals bind to transmembrane receptors and activate signal transduction pathways, eventually leading to the activation of transcription factors and new gene expression. These inductions are the first to occur in the developing Drosophila egg. Approximately 100 genes and encoded gene products have been identified as necessary to establish the organization of the early gastrula. Hundreds more participate in the accomplishment of these events, but they are less well described at present. In most cases, these genes probably encode proteins required in numerous developmental processes and, hence, were not recovered under the conditions of the mutant inspections used here. As shown in Figure 6-1A-D, a coherent scheme of early development was proposed and well supported by 1992, the first of such complexity and completeness at the molecular level for any organism. FIGURE 6-1A Outline of anteroposterior development in Drosophila and the steps of regulated gene expression (Ingham 1989). Heavy dashed arrows indicate the activation of specific gene expression by transcription factors. Thin solid arrows indicate transcription and translation. Note that Hox genes are activated by both pair-rule and gap proteins, whereas segment-polarity genes are activated by pair-rule proteins alone. In the anteroposterior dimension, segments and HOX domains are formed. Further explanation is given in Figure 6-1B.
OCR for page 118
Scientific Frontiers in Developmental Toxicology and Risk Assessment FIGURE 6-1B Anteroposterior development in Drosophila (Nüsslein-Volhard 1991). Figure 6-1B is shown diagrammatically here, for segment formation and HOX compartment formation. The coordinate proteins Bicoid, Nanos, and Cad are translated from mRNAs localized at the two poles of the egg during oogenesis. Translation generates gradients of proteins. Bicoid and Cad are transcription factors, whereas Nanos protein inhibits the translation of another translation factor (Hunchback) in the posterior half of the egg. The graded transcription factors activate eight gap genes, and different factor concentrations activate different gap genes. The gap proteins are also transcription factors. Each diffuses locally and inhibits other gap genes, setting up eight partially overlapping stripes of gap protein along the egg’s length. The gap proteins activate eight pair-rule genes, each of which has a complex cis-regulatory region and is activated by seven combinations of gap proteins, each making seven evenly spaced stripes of protein. Thus, there are 8 × 7 or 56 stripes of pair rules along the egg’s length, arranged in 7-fold repeats. The pair-rule proteins are all transcription factors. These activate eight segment-polarity genes, each of which has a complex cis-regulatory region activated by at least two combinations of pair-rule proteins, to give 14 stripes of expression each. Thus, there are 14 × 8 or 104 stripes of segment-polarity proteins. The 14-fold repeat is the basis for 14 segments of the posterior head, thorax, and abdomen. The pair-rule and gap proteins together activate Hox genes in eight domains in the posterior head, thorax, and abdomen. Cell outlines are not shown, but cells are present in the two lowest panels.
OCR for page 140
Scientific Frontiers in Developmental Toxicology and Risk Assessment One more kind of molecular-stress and checkpoint pathway should be noted: the apoptosis pathway, which is also a signaling pathway (see Table 6-2). It can be activated by either extracellular or intracellular signals and leads to the “programmed” death and destruction of a cell. It is a tightly controlled process in which a cell is destroyed but neighboring cells are unaffected. Apoptosis is not found in single-celled organisms. It is an invention of metazoa and is used in normal embryonic development as well as in recovery attempts of teratogen-damaged embryos. In normal development, where apoptosis is also known as programmed cell death, it is important in the shaping of tissues and organs (e.g., the elimination of cells from the interdigital spaces of the human hand). Cells undergoing apoptosis are found throughout most embryonic mesenchymal tissues, presumably reflecting the elimination of cells that have not been able to successfully integrate the signals impinging upon them. Some mouse mutants, such as Hammertoe, fail to initiate the normal amount of programmed cell death in normal limb development, and an abnormal limb results (Zakeri et al. 1994). Apoptosis is also the ultimate molecular-stress and checkpoint pathway, for it eliminates cells too damaged to be restored to a normal state by the various repair and checkpoint pathways. For example, if DNA repair is incomplete and the cell attempts to divide, it is killed and autolyzed. It has been proposed that cell death is less detrimental to the multicellular organism than having live cells with highly modified DNA, perhaps proliferating uncontrollably and interacting aberrantly with other cells. Apoptotic cell death is an early response of embryos to many if not all teratogens (Scott 1977; Knudsen 1997). Often, teratogen-induced cell death occurs in the areas of normal programmed cell death but in an expanded area (Alles and Sulik 1989). If cell death is not too extensive, embryos are thought to recover by compensatory cell proliferation (Sugrue and DeSesso 1982). Excessive teratogen-induced cell death, however, is directly linked to abnormal development. For example, eye defects induced by 2-chloro-2'-deoxyadenosine are associated with excessive teratogen-induced cell death (Wubah et al. 1996). The intracellular signals of apoptosis are not yet known. Key components in the execution phase of the apoptotic pathway are the intracellular cysteinyl-aspartate proteases known as caspases, particularly caspase-3 (Colussi and Kumar 1999). These enzymes are normally present in all cells as inactive precursors that become activated by cleavage at specific internal motifs, in response to cytochrome c leaked by mitochondria into the cell’s cytoplasm. Once activated, these caspases function to degrade specific target substrates such as poly(ADP-ribose)-polymerase (PARP), DNA-PKs, and lamins. Thereafter, chromosomal DNA is broken down. Treatment of cells with such developmental toxicants as hyperthermia, cyclophosphamide (an alkylating agent), and sodium arsenite (thiol oxidant) leads to the activation of caspase-3, cleavage of PARP, fragmentation of DNA, and cell death (Mirkes and Little 1998). It is not known how cells in the embryo recognize exposure to a developmental toxicant and initiate the apoptotic
OCR for page 141
Scientific Frontiers in Developmental Toxicology and Risk Assessment response, but perturbation of the redox status of the cell and oxidative stress are often, if not always, involved. As in other pathways, the apoptotic pathway engages in cross-talk, for example, with the nuclear factor-kappaB (NF-kB) and INF and FAS pathways. A recent report demonstrates that heat shock (43°C) can rapidly activate the stress-activated protein kinase pathways mediated by c-JUN terminal kinase (JNK) and p38 (Wilson et al. 1999). As noted for the drug-metabolizing enzymes discussed in Chapter 5, these molecular-stress and checkpoint pathways deserve attention as elements of the organism’s defense against physical and chemical interventions. It remains to be learned whether polymorphisms of defense components exist in humans, compromising their responses to environmental agents. The extent to which the germ line, gametes, and early embryos operate these molecular-stress and checkpoint pathways is also poorly understood. Developmental Differences Although Drosophila and mouse development share more similarities than anyone thought 15 years ago, significant differences do exist. Mice share more aspects of development with other chordates (the chordate phylum includes vertebrates, cephalochordates, and urochordates) than they do with Drosophila, and they share still more aspects with other mammals. There appear to be “nested similarities” of development (i.e., the more recent the common ancestor of two groups, the more shared features of their development). Regarding HOX genes, for example, chordates have four more kinds of genes (HOX 10-13) than do arthropods. These differ slightly in sequence from the others and are located at the 5′ end of each cluster. They are expressed in the postanal tail, which is a chordate structure not shared by arthropods, and also in the developing vertebrate limb. Still, the difference between chordates and arthropods is a modification of a shared feature, namely, the use of HOX genes to divide the anteroposterior dimension of the animal into nonequivalent spatial compartments. Chordates, but not arthropods, share the development of a dorsal hollow nerve cord, a notochord, and a segmentally arranged pharyngo-branchial apparatus, in addition to a postanal tail. They also share a kind of development involving a centralized “organizer” group of cells, the Spemann organizer, which releases inducers important in the placement, orientation, and scaling of later development by surrounding cells. The inducers secreted by the organizer have now been identified. Several inducers are secreted protein antagonists of the TGFβ and WNT signals and are used by surrounding cells to maintain their ventral posterior paths of development. The inducer antagonists disinhibit and hence release the inherent capacity of the surrounding cells to undertake dorsal anterior kinds of development (e.g., to form the neural tube rather than epidermis) (Harland and Gerhart 1997; Smith and Schoenwolf 1998; Weinstein and Hemmati-Brivanlou 1999). Few researchers would have guessed a few years ago
OCR for page 142
Scientific Frontiers in Developmental Toxicology and Risk Assessment that this subtle depression of neural development is the organizer’s function in all chordates. Nonetheless, it should be noted that Holtfreter (1947), building on a discovery of Barth (1939), suggested that neural inducers provide little information except to release the inherent capacity of ectoderm cells to develop as neural tissue. This suggestion came from Barth and Holtfreter’s findings that ectoderm would develop neural tissue if merely shocked briefly by ion imbalances or pH extremes. Even though the organizer mode of development is distinctive to chordates, the components of the process are common to a wide range of other animals. For example, one antagonist, the Chordin protein, exhibits significant homology with the SOG protein of Drosophila. The SOG protein antagonizes a TGFβ inductive signal (called Screw) in Drosophila as part of the development of regions of neural versus epidermal development (Neul and Ferguson 1998). Furthermore, in both Drosophila and frogs, there is a specific metalloproteinase that degrades the signal-antagonist complex, releasing the signal. The chordate and Drosophila inductive processes have deep similarities, though differing in details of time, place, and circumstances of use. As a final example of differences, the dorsoventral dimension of arthropods looks quite different from that of a mouse, but recent analysis has shown that a number of similar genes are expressed in the nerve cords, hearts, body muscle, visceral mesoderm, and gut of both. It is currently accepted that these organs were present in primordial form in a common ancestor, but the arrangement of the organs in chordates is the inverse of that in arthropods. That is, the nerve cord is dorsal in chordates and ventral in arthropods, and the heart is ventral in chordates and dorsal in arthropods. The inversion of the dorsoventral axis is thought to have occurred in the chordate line after hemichordates split off (Nübler-Jung and Arendt 1996). Recognizing the fact that Drosophila does not share all details of early development and organogenesis with vertebrates, researchers have begun a systematic collection of developmental mutants of the zebrafish, a small vertebrate with a short life cycle (see Chapter 7), suitable for the production of a large mutant collection. The organs of embryonic zebrafish, more than the organs of Drosophila, resemble those of mammalian embryos in structure and function. In light of the extensive conservation of developmental processes found thus far, it is expected that in most cases what is true for fish development, as learned from those mutants, will be true for mammalian development, down to the level of molecular details of components and processes. That is not meant to deny differences among organisms (e.g., mammals undergo placental development with extensive extra-embryonic tissues not found in a zebrafish), nor to dismiss the possibility that developmental biologists might be misled in some instances by the study of model organisms. The greater part of mammalian development can be understood, however, by the study of other organisms’ development. Ultimately, mammalian development will have to be understood in all the details of its differ-
OCR for page 143
Scientific Frontiers in Developmental Toxicology and Risk Assessment ences, but even this pursuit will benefit from the context of knowledge of the processes shared with other organisms. For example, unique mammalian processes, such as extra-embryonic tissue formation or more extensive forebrain development, are still expected to entail many of the same signal transduction pathways and genetic regulatory circuits as used elsewhere in development. The Evolutionary Perspective In light of the availability of base sequences for a variety of kinds of genes in a variety of organisms, the place of metazoa (the multicellular animals) among the kingdoms of living organisms has been recently re-evaluated. It now appears that animals share a common ancestor more closely with plants (especially fungi) than with protozoa such as ciliates or amoebae. These three multicellular kingdoms arose from a common eukaryotic ancestor (probably single celled) perhaps 1.2 billion years ago, whereas eukaryotic single cells go back 2.2 to 2.7 billion years and prokaryotic life goes back perhaps 3.5 billion years (Feng et al. 1997; Pace 1997). The conservation of basic biochemical, genetic, and cell biological functions has been surprisingly extensive in that long lineage. At least 3 billion years ago, ancient prokaryotes originated the processes of replication, transcription, translation, energy metabolism, and biosynthesis, and those processes have been carried forward to this day with little change in all life forms, including animals. The comparisons of the whole genomes of bacteria, yeast, and now the nematode, C. elegans, show clearly the conservation of the protein-coding sequences of genes. At least 2 billion years ago, single-celled eukaryotes originated the basic cell biological processes of mitosis, meiosis, a cdk-cyclin-based cell cycle, an actin-based cytoskeleton and myosin-based movements, a tubulin-based cytoskeleton and kinesin-dynein-based movements, membrane-trafficking, and membrane-bounded organelles. These processes and structures have been carried forward by the single-celled eukaryotes and animals with little change to this day. In light of this conservation of ancient processes, what have metazoa added in the past 1.2 billion years? Their innovations include abundant cell-cell signaling, extracellular matrix, cell junctions, and a wide range of responses to intercellular signals based on complex genetic regulatory circuits and protein phosphorylation. The C. elegans genome shows that, compared with yeast, metazoa have greatly expanded the number of genes encoding proteins of signal transduction, the cytoskeleton, and transcriptional regulation and have greatly increased the size and complexity of the cis-regulatory regions of genes. Metazoa seem to have evolved in a regulatory or informational direction, that of determining the time, location, and circumstances within a multicellular population for activating and inhibiting the many conserved biochemical and cell biological processes brought forward from their single-celled ancestors. All these ancient processes have been made contingent on cell-cell signaling.
OCR for page 144
Scientific Frontiers in Developmental Toxicology and Risk Assessment As mentioned above, the importance for developmental toxicology of the discovery of extensive conservation of components and processes among seemingly disparate animals is the conclusion that the study or testing of toxicants in model animals can provide relevant information about humans, as long as the extrapolation is done within conserved responses, of which there are many. Organogenesis Organs are usually defined as containing two or more tissues, each tissue containing differing cell types and cell functions, coordinated in a higher level of organization and function than the independent tissues. Second to the organism’s overall body organization, organs are the most complex level of organization of cells. Organogenesis is the organ-forming phase of embryonic development. It begins once the basic anteroposterior and dorsoventral organization of the embryo is established by gastrulation and neurulation. During organogenesis, cytodifferentiation takes place, and then the organ begins to function. A fundamental question about organogenesis concerns the means by which the different parts of the organ are brought into complex alignment and integrated function. In the first half of the twentieth century, organ formation was described in detail by light microscopy, and the inductive interactions of different cell groups involved in organ formation were revealed by experimental analysis. In general, the different tissues of the organ were not found to form independently and then come together in perfect apposition. Rather, tissues that are nearby as a result of extensive movement during gastrulation and neurulation interact with each other and also with surrounding tissues. Combinations of signals establish positional identity and initiate the progressive delineation of organ-specific gene activations. Thus, it is not necessary that all participants in early organogenesis have position and cell-type specific information. Cell signaling operates throughout organogenesis. Recently, the local signals and responses have been identified in several kinds of organogenesis, the responses often being experimentally proven by using “marker” or “reporter” genes activated at various stages of the process and visualized by staining specific mRNAs by in situ hybridization. Extensive molecular descriptions and cellular and genetic analyses have defined key regulatory pathways that facilitate the development of many vertebrate systems, including the following: Neural tube: regionalization of forebrain, midbrain, hindbrain, and spinal cord (for reviews, see Wassef and Joyner 1997; Brewster and Dahmane 1999; Dasen and Rosenfeld 1999; Veraksa et al. 2000). Neural tube: dorso-ventral organization of brain and spinal cord (for reviews, see Edlund and Jessell 1999; Lee and Jessell 1999). Sensory systems: optic vesicle and eye, otic vesicle and inner ear, and olfactory epithelium (for reviews, see Fekete 1999; Holme and Steel 1999; Kraus and Lufkin 1999; McAvoy et al. 1999).
OCR for page 145
Scientific Frontiers in Developmental Toxicology and Risk Assessment Neural crest: autonomic and sensory ganglia and glia and melanocytes (for reviews, see Francis and Landis 1999; Gershon 1999a,b; LaBonne and Bronner-Fraser 1999). Neural crest: midfacial and branchial connective tissues and teeth (for reviews, see Francis-West et al. 1998; Peters and Balling 1999; Schneider et al. 1999; Tucker and Sharpe 1999; Vaglia and Hall 1999). Paraxial mesoderm: somites, skeletal muscle, vertebrae, and ribs (for reviews, see Brand-Saberi and Christ 1999; Relaix and Buckingham 1999; Burke 2000; Rawls et al. 2000; Summerbell and Rigby 2000). Intermediate mesoderm: kidneys, gonads, reproductive ducts, and sex determination (for reviews, see Sariola and Sainio 1998; Horster et al. 1999; Parker et al. 1999; Swain and Lovell-Badge 1999). Cardiovascular system: heart, angiogenesis, and hematopoiesis (for reviews, see Baldwin and Artman 1998; Mercola 1999; Morales-Alcelay et al. 1998; Tallquist et al. 1999). Limb: growth and specification of axes (for reviews, see Martin 1998; Ng et al. 1999; Vogt and Duboule 1999). Pharyngeal endoderm: thyroid and thymus (for reviews see Bodey et al. 1999; Missero et al. 1998). Gut tube: lungs, liver, pancreas, stomach, and intestines (for reviews, see Gretchen 1999; St-Onge et al. 1999; Warburton and Lee 1999). The familiar conserved signaling pathways are used over and over in many different contexts in organogenesis and other steps of development, as listed in Table 6-3. For example, the Sonic Hedgehog (SHH) signaling pathway is involved in establishing asymmetry in the early gastrula, inducing floor plate and motor neurons, separating the single eye field into paired optic primordia, maintaining proliferation in migrating neural crest cells, establishing patterning of the medial and lateral nasal prominences and tooth induction, inducing sclerotome segregation and epaxial muscle formation in somites, development of the prostate gland, determining left-right asymmetry, establishing the anteroposterior (rostrocaudal) axis of the limbs, delineating the tracheo-esophageal diverticulum, and establishing sites of formation and branching patterns in lung and pancreatic epithelia. Comparable matches could be made for members of the fibroblast growth factor (FGF), TGFβ, BMP, and WNT signaling families. Although the signaling pathways involve the same or closely related signaling molecules, the responses made by cells are distinct because of the genes and gene products they express prior to and in response to the many different combinations of these signaling factors. The Vertebrate Limb: The Best Known Organogenesis Model The limb is by far the most studied organ rudiment of vertebrates, supported by over 50 years of experimental embryology and intensive recent molecular
OCR for page 146
Scientific Frontiers in Developmental Toxicology and Risk Assessment genetic analysis. Cell-cell interactions, based on known signaling pathways, are well understood, as are the patterns of cell proliferation within the developing limb. Surprisingly, many homologous genes and signaling pathways are conserved between the vertebrate limb and the developing leg or wing of Drosophila. The researchers who followed the seemingly remote leads from the early Drosophila work on appendages have made rapid progress on vertebrate limb development in the past few years. Limb development is discussed here to show (1) the importance of precise temporal and spatial signaling for organizing a complex organ rudiment, and (2) the interaction of multiple signaling pathways to establish the organ’s three-dimensional morphology. The developing limb contains three axes: proximodistal, anteroposterior (rostrocaudal), and dorsoventral. Each axis has its own secreted signals and these are integrated in the limb bud, as shown in Figure 6-4 and as summarized by Johnston and Tabin (1997) and Ferretti and Tickle (1997). The bud originates as a locus of rapidly dividing cells in the somatic mesoderm and overlying epidermal ectoderm of the flank of the trunk. Outgrowth of the paired anterior and posterior limbs involves the maintenance of a high rate of cell division within the bud. Local trunk structures, such as the mesonephros, somites, and notochord, secrete FGF10 onto the flank tissues, and this signal initially keeps the bud proliferating. As outgrowth begins, the ectoderm overlying the bud locally thickens to form a ridgelike structure called the apical ectodermal ridge (AER), which then secretes several FGFs, notably FGF8, and takes over for the flank in maintaining cell proliferation in the adjacent bud mesoderm. The bud is then self-sufficient for this signaling and the flank stops serving as an FGF source. (The bud can be transplanted to a remote site, such as the yolk sac, at this stage and will develop autonomously.) The area of rapid division in the bud is called the progress zone (PZ). It lies just beneath the AER. As cells proliferate in it, some are displaced away from the AER and are no longer exposed to FGF. They slow their division rate and stop changing their specification (i.e., their capacity to develop as one
OCR for page 147
Scientific Frontiers in Developmental Toxicology and Risk Assessment FIGURE 6-4 Development of the limb bud in tetrapods (the four-legged vertebrates). (Panel A) Cross section of mouse embryo after ventral closure, at the level of the forelimbs. The limb buds emerge as small protrusions from the left and right flank. They consist of a surface layer of epidermal cells (ectoderm) and an internal mass of mesenchymal cells (mesoderm). The latter engage in rapid proliferation. (Panel B) Close-up view of a bud, cross section, dorsal, ventral. The apical ectodermal ridge (AER) secretes FGF onto the underlying mesenchyme. The zone of polarizing activity (ZPA) of the mesenchyme secretes SHH onto the ridge. A reciprocal activation circuit is completed in which the AER and ZPA keep each other active. In the mesenchyme close to the ridge is the progress zone (PZ), a population of rapidly dividing cells. Their division is kept going by FGF and SHH. Dividing cells keep changing their option for a future developmental path, in a progression, proximal (shoulder) to distal (hand). As they divide, some are displaced toward the flank, out of the PZ, too far away to receive FGF signals. Their division slows, and they adopt the developmental path of the step of the progression at which they were when they were last in the PZ. The SHH from the ZPA spreads in a gradient toward the anterior edge of the bud. The anteroposterior differences of the limb are signaled by this gradient, perhaps through a BMP2,4 signaling coupled to SHH signaling. (Panel C) The limb bud in cross section, at right angles to panel B, to show the dorsoventral plane. The dorsal epidermis secretes WNT7A onto the mesenchyme. That signals it to develop dorsally. The ventral epidermis does not secrete WNT7A, because the cells express the En gene encoding a transcription factor inhibiting the Wnt7a gene from expression. See text for further information and references.
OCR for page 148
Scientific Frontiers in Developmental Toxicology and Risk Assessment part or another of the future limb). As a consequence, cells leave the PZ in a proximal to distal order of specification (i.e., the first to leave have the capacity to form the upper limb, the next to form the lower limb, and the last to form the hand or foot). When cells leave the PZ, they start to differentiate into cartilage, bone, and connective tissue of their specified limb level. Precursor cells of limb muscles migrate into the bud from the adjacent somitic myotomes, and nerves extend in from spinal ganglia and the spinal cord. Interactions between the AER and PZ are reciprocal (i.e., the AER maintains mitosis within the PZ by way of FGF, and the PZ maintains the thickened AER by way of a yet unknown signal). The second axis of the limb, the anteroposterior axis, is also established through cell-cell interactions and secreted signals. The posterior portion of the limb bud contains a specialized region called the zone of polarizing activity (ZPA). This zone was originally recognized as a signaling center, because when it was transplanted to the anterior side of another limb bud, the bud develops a mirror-image duplicated limb. The ZPA secretes SHH, which diffuses across the anteroposterior dimension of the bud, establishing a gradient and setting off the formation of a second gradient of BMP2 and 4, two kinds of TGFβ signals. SHH is both sufficient and necessary to establish the anteroposterior pattern of the limb. Retinoic acid, acting via a nuclear receptor, might also have a role. The third axis of the limb, the dorsoventral axis, is established through interactions between the nonridge dorsal ectoderm of the limb and the underlying bud mesoderm. That was initially shown by experiments in which the dorsal and ventral regions of the ectoderm were rotated with respect to the mesoderm. The dorsal ectoderm releases the WNT7A signal, which induces the expression of the Lmx-1 gene in the underlying mesoderm and which suppresses expression of the engrailed-1 gene, thereby restricting its expression to the ventral ectoderm. Mutants defective in WNT7A signaling develop double ventral limbs. Mutants defective in engrailed-1 expression develop double dorsal limbs with double sets of fingernails. It remains to be learned how the signaling pathways of each axial dimension are coordinated with those of the other dimensions, and how the integration of these pathways leads to the formation of unique skeletal structures in precise locations within the limb. The limb exemplifies the advanced understanding of vertebrate organogenesis at a molecular genetic level (i.e., of the signal pathways and the genetic regulatory circuits involved in changing transcription and regulating cell proliferation). This kind of understanding is prerequisite to understanding the action of toxicants on embryogenesis. On the basis of new information, mechanisms of action of toxicants have been recently proposed, although these have yet to be tested. For example, thalidomide leads to a failure to form proximal parts of the limb (the upper and middle parts of the limb), but the hand or foot is usually formed. That developmental outcome is paradoxical, because other treatments, such as the removal or inactivation of the AER, lead to truncation of the limb in the reverse order; the upper and middle parts are present but the hand
OCR for page 149
Scientific Frontiers in Developmental Toxicology and Risk Assessment or foot is missing. Tabin (1998) recently proposed that thalidomide reduces cell proliferation in the PZ, by means yet unknown, and the few cells that remain there in prolonged contact with FGF8 secreted by the AER are specified as hand or foot cells, the last normally to emerge from the zone. This is an example of an incisive prediction about a long-known toxicant made on the basis of recent knowledge. Yet, the chemical basis for thalidomide’s specific effect on cell proliferation in the PZ escapes even a proposal at this time, and so the hypothesis is incomplete. Stephens and Fillmore (2000) have suggested that thalidomide interferes with integrin gene expression in limb bud mesenchyme cells and, thereby, with their ability to stimulate angiogenesis at the level needed for continued rapid proliferation. The capacity of the limb bud to develop normally after injury has been studied. Large numbers of cells can be removed at early stages, and as long as representatives remain of the AER, ZPA, and PZ, development will be normal. Immigrating muscle cells from any myotome will enter and adapt to the limb bud, and nerves from any spinal cord level will enter and make neuromuscular connections, although the CNS circuitry of that level may not be appropriate for normal limb movement. The robustness of limb development, like that of other parts of the body, is substantial, because each of the interactive cell groups is much larger than minimally necessary and is capable of proliferation. Robustness is not unlimited, however, and total removal of a key signaling or responding group is deleterious. Regenerating limbs, such as those of newts, have surmounted even that limit, but they are the exception among vertebrates. SUMMARY This committee has been asked to evaluate the state-of the-science for elucidating mechanisms of developmental toxicity. It seems self-evident that the knowledge about the basic processes of development provides developmental biologists with an understanding of normal development not even thought possible a decade ago, and also provides developmental toxicologists with improved tools to understand the mechanisms by which chemicals cause abnormal development. In the last decade, great advances have been made in the understanding of developmental processes on a molecular level in model organisms, such as Drosophila and C. elegans, and in several vertebrates, including the mouse. For the first time, molecular components and their functional interactions have been identified. Developmental processes can be described for the first time as organized networks of these components and their functions. The examples examined so far primarily concern early development before organogenesis but also organogenesis in a few cases. Cell-cell interactions by way of intercellular signals are pervasively and repeatedly used. In all aspects of development, including organogenesis and cytodifferentiation, signaling is expected to be of central impor-
OCR for page 150
Scientific Frontiers in Developmental Toxicology and Risk Assessment tance. A small number of signal transduction pathways are used in these interactions. There are approximately 10 kinds in early development and organogenesis. (Seven more are used by differentiated cell types.) The number of allelic variants that exist in these human genes remains to be studied. These pathways are conserved among animal phyla, as are many of the genetic regulatory circuits involved in the responses of cells to signals. In addition, many of the basic cell processes, such as proliferation, secretion, motility, and adhesion, are also highly conserved among animals. This extensive conservation gives strong justification to the use of model animals, including Drosophila, C. elegans, zebrafish, and mouse, to learn about basic aspects of mammalian, even human, development. The understanding of development is far from complete. Although a number of main components have been identified for early processes, their interactions and use in combinations introduce substantial complexity to an inclusive understanding of development. Few of the many types of mammalian organogenesis have been analyzed, and only a few of the 300 types of cytodifferentiation have been studied. As more components and interactions are revealed, it will become important to establish readily accessible databases containing all the information about development.
Representative terms from entire chapter: