SEARCH

SEARCH BY CITATION

Keywords:

  • comparative embryology;
  • orthologs;
  • modularity;
  • model organisms;
  • ectoderm;
  • mesoderm;
  • endoderm

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

More than a century ago, Ernst Haeckel created embryo drawings to illustrate the morphological similarity of vertebrate early embryos. These drawings have been both widely presented and frequently criticized. At the same time that the idea of morphological similarity was recently attacked, there has been a growing realization of molecular similarities in the development of tissues and organs. We have surveyed genes expressed in vertebrate embryos, and we have used them to construct drawings that we call Molecular Haeckels. The Molecular Haeckels emphasize that, based on gene expression, there is a greater similarity among vertebrate embryos than even Haeckel might have imagined. Developmental Dynamics 239:1905–1918, 2010. © 2010 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The most famous drawings of embryos are those that Ernst Haeckel used to illustrate his law of ontogenetic connection of systematically related forms (Elinson,1987). Haeckel argued that vertebrate embryos at a particular point in their development look so similar that they cannot be told apart by a casual examination. Haeckel's drawings have been alternatively praised and damned over the decades, with a renewal of attacks approximately 10 years ago (Richardson,1995; Richardson et al.,1997). Despite counters to the attacks (Sander,2002; Hopwood,2006) and a reassessment (Richardson et al.,1998a; Richardson and Keuck,2002), Haeckel's drawings have been removed from textbooks. Compare Gilbert (1997) and Alberts et al. (1994) to Gilbert (2000) and Alberts et al. (2002).

It is ironic that the recent attacks on Haeckel's concept of a common vertebrate morphological stage came in an era when developmental biologists became increasingly aware of common molecular events underlying development in a wide range of animals from fruit flies to mammals. Anterior–posterior patterning by a complex of homeodomain-containing transcription factors is a prominent example of highly conserved molecular events (Slack et al.,1993). Others are Pax6 control of eye development and the importance of tinman/NKx2.5 in producing a beating heart. Conserved gene expressions and the molecular interactions that produce particular tissues and organs have led to concepts like the molecular tool-box (Carroll et al.,2001; Canestro et al.,2007; Carroll,2008; Holland,2009), gene kernels and subcircuits (Hinman and Davidson,2007; Peter and Davidson,2009), conserved core components (Gerhart and Kirschner,2007), gene regulatory networks (Davidson et al.,2002; Rebeiz et al.,2005; Davidson and Levine,2008; Ettensohn,2009), and core gene network (Woodland and Zorn,2008). To a taxonomist, many of these elements are symplesiomorphies or shared basal characters (Richardson et al.,2001).

Haeckel sensed a commonality in early embryos. Although that sense could have biased his depiction of embryos, Haeckel was more correct than the data of his day allowed. Commonality in early embryos is more apparent from molecules than from morphology. Molecules may provide a better way to depict the common vertebrate embryonic stage that Haeckel sought, and a first generation of that molecular depiction will be presented here.

CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

To create a Molecular Haeckel, we will place regulatory molecules, common among vertebrate embryos, onto a schematic morphology. The first step is to design the morphological framework of an idealized, generic vertebrate embryo at an early developmental stage. There are a large number of common features of vertebrate early embryos that can be incorporated into the framework. This collection of common features has been used to generate concepts such as the basic or primitive vertebrate body plan (e.g., Arey,1924,1954), the pharyngula (Ballard,1981), the phylotypic stage (Sander, 1983, as cited in Sander,2002; Slack et al.,1993), and the developmental hourglass (Elinson,1987; Duboule,1994).

This generic vertebrate embryo has three germ layers arranged in concentric tubes and a set of dorsal axial structures. In the trunk and post-anal tail, the dorsal axial structures consist of a spinal cord, underlain by a notochord and flanked by segmented, mesodermal somites. Completing the mesodermal tube on the lateral and ventral sides are intermediate mesoderm and lateral plate mesoderm, split into an inner splanchnic layer and an outer somatic layer. Four limb buds arise from the somatic layer. The mesodermal tube surrounds the inner endodermal tube, which joins the outer ectoderm at the mouth and the anus. Various evaginations of the endodermal epithelium foreshadow organs such as the liver and pancreas. The anterior end of the embryo is highly modified to form the head. The neural tube is expanded to form the brain; sense organs arise from ectodermal placodes and connect to the brain, and much of the prospective craniofacial cartilage is derived from cranial neural crest. The important roles played by placodes and cranial neural crest led to Gans and Northcutt's (1983; Northcutt,2005) intriguing hypothesis that the evolution of vertebrates involved the addition to an ancestral chordate of a new head from these two embryonic cell types.

This generic embryo looks similar to actual embryos to a greater or lesser degree. There are three major sources of deviations. First, different vertebrate embryos have different amounts of yolk, ranging from no yolk in placental mammals to huge amounts of yolk in birds. Furthermore, yolk may either be included within the embryo's body, as in amphibians, lungfish, teleosts, and sturgeon, or contained in an extraembryonic yolk sac attached by a stalk to the body, as in shark, coelacanth, birds, and reptiles. When images of embryos are modified to remove the yolk, they look much more similar (Richards,2008, p. 307).

The second major source of deviation between the generic embryo and actual embryos is the relative size of parts of the embryo. The size differences can be quantitative or qualitative. The most obvious quantitative difference is the number of somites in the trunk (Richardson et al.,1998b; Woltering et al.,2009; Gomez and Pourquie,2009). The frog X. laevis has only nine trunk somites (Tucker and Slack,1995), whereas animals with elongated bodies, such as eels, caecilian amphibians, and snakes, can have hundreds of somites (Gomez et al.,2008; Woltering et al.,2009). Chick and mouse have 39 and 35 non-tail somites, respectively (Burke et al.,1995).

An example of a qualitative difference is the size of the head relative to the body. Amniote embryos generate anterior structures, including the head and heart, precociously relative to the trunk and tail, and these structures are disproportionately large. Within amniotes, embryos of non mammalian amniotes have much larger eyes proportional to the head, than mammalian embryos (Jeffery et al.,2002). There are many examples of size differences in limb development. Marsupials have very large forelimbs early to enable the embryo to crawl to the teat. Conversely, tadpoles of frogs have practically no limb buds until they begin feeding, and after that, the limbs remain small and inconspicuous.

The third major source of deviation between the generic embryo and actual embryos is the relative timing of development of one part relative to another. For example in some embryos, forelimb buds arise before hindlimb buds, while hindlimb buds appear first in other embryos (Bininda-Emonds et al.,2007; Richardson et al.,2009). These heterochronies are rampant (Richardson,1995), complicating comparisons between different embryos and muddying the concept of a common phylotypic stage.

Analytical schemes have been proposed not only to detect heterochronies between embryos of different animals but also to provide measures of differences when comparing different embryos (Smith,2001; Schlosser,2001; Jeffery et al.,2005; Maxwell and Harrison,2009; Werneburg,2009). Whereas heterochrony makes it difficult to compare whole embryos of different animals, the concept of modularity obviates difficulties in comparisons at lower levels of organization.

Modularity refers to the integrated and autonomous behavior of a molecular pathway, a group of cells in a tissue, or even a developing organ (Schlosser and Wagner,2004). For example, many cellular and molecular interactions occur autonomously within a developing limb bud, without reference to the rest of the embryo. Similarly, a relatively small number of transcription factors serve as master regulatory genes for a tissue, like skeletal muscle, or an organ, like the pancreas. Once these master regulatory genes are turned on, a cascade of molecular and cellular events follows to generate the tissue or organ.

Given modularity, we will place key regulatory molecules that operate within a tissue or organ on our morphological framework. Essentially, the embryo is depicted as a mosaic of gene expression territories, largely independent of each other. In one sense, we are simply shifting the concept of a common embryonic stage from one level of organization, the whole embryo, to a lower level of organization, the individual tissues and organs. By doing so, we suggest that molecular commonality of tissues and organs is stronger than the morphological commonality of the whole embryo at the phylotypic stage. We suggest that our molecular pictures have core truths, as Haeckel claimed for his morphological pictures.

CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The genes that are used to construct the pictures are regulatory genes that are expressed in a specific organ or tissue in embryos across a range of vertebrates.

Range of Vertebrates

Gene expression and function in development are known mainly from embryos of four model animals: zebrafish, Xenopus, chick, and mouse. Chick and mouse represent the monophyletic Amniota; amniotes plus Xenopus represent the monophyletic Tetrapoda, and tetrapods plus zebrafish represent the monophyletic Osteichthyes. If a gene is expressed in both zebrafish and mouse embryos, it suggests that their last common ancestor had that organ specific gene expression and that all animals within Osteichthyes are expected to have that gene expression. Gene expression in both zebrafish and mouse embryos was our usual minimum criterion for inclusion in the Molecular Haeckels.

Of course, if Xenopus and chick also have that gene expression, the more likely it is that it represents a common character. Some embryonic gene expressions have been reported for other teleosts such as medaka, fugu, and stickleback, other amphibians such as axolotl and caecilian, and other amniotes such as alligator, snake, and human. A few embryonic gene expressions have been reported for non-teleost Actinopterygians, such as bichir (Takeuchi et al.,2009) and paddlefish (Metscher et al.,2005; Davis et al.,2007), indicating that these embryos are available to strengthen any conclusions of commonality based on teleosts. Some gene expressions have been reported in embryos of Chondrichthyes, the cartilaginous fish. These include reports on catshark, dogfish, and skate (Neyt et al.,2000; Derobert et al.,2002a,b; Tanaka et al.,2002; Okabe and Graham,2004; Dahn et al.,2007; Bajoghli et al.,2009; Suda et al.,2009). As more genes are analyzed in cartilaginous fishes, the Molecular Haeckels can be drawn to include all jawed vertebrates, the Gnathostomes.

Regulatory Genes

Genes were chosen, primarily based on expression patterns. A more stringent requirement for inclusion would be demonstrations of function in the development of a particular tissue or organ. In many cases, function was shown in one model animal by the most convenient method, i.e., mouse knockouts, zebrafish mutants, or Xenopus overexpression, and expression was reported in other animals. For the gene collection used here (Table 1), gene expression was a sufficient criterion for inclusion.

Table 1. Genes Involved in Development of Vertebrate Embryosa
inline image

The two main classes of regulatory molecules are transcription factors and components of signaling pathways, usually the ligands. It was easier to include the transcription factors compared with ligands, although that may be partially an artifact of naming conventions. Both types of molecules are composed of families, whose members are often numbered. For example, there are nine Pax genes and more than 20 fibroblast growth factor (FGF) genes (Itoh,2007). There are more transcription factor families compared with ligand families, of which there are mainly Wnt, FGF, Hh, Delta-Notch, and transforming growth factor-beta (TGFβ). The TGFβ superfamily is further split into bone morphogenetic proteins (BMPs), nodals, growth differentiation factors (GDFs), and others. Another important ligand is retinoic acid, which is generated by several different enzymes.

For many tissues and organs, activities of several signaling families are involved at different times and in different places. Sorting out whether a particular ligand or a different ortholog is used in different animals for the same activity is a major task, even in well-known cases. For example, Shh is expressed in the notochord, the floor plate, and the zone of polarizing activity of the limb, and there is considerable functional analysis. Complicating its placement in a Molecular Haeckel is the fact that the Xenopus Shh ortholog falls into the Ihh family phylogenetically (Varjosalo and Taipale,2008). Because of the usage of multiple ligands in many tissues, less emphasis was placed on ligands than on transcription factors.

Not only are there more transcription factor families, the conserved expression patterns of differently numbered family members are often well-documented. One example is the expression of the Hox genes in the spinal cord, the somites, and the limb. A second example is the tissue and organ specific expression of Pax and Tbx genes. There are interesting exceptions, however. For example, Pax8 is important for mouse thyroid development, but the related Pax2 and Pax2a are used in Xenopus and zebrafish respectively (Table 2). Different Msx and Gbx members are expressed in zebrafish brain development compared with the tetrapods (Table 2). Other cases of differences of ortholog usage between zebrafish and tetrapods are Tbx in somite segmentation and Cdx in intestine (Table 2). For the most part, this type of ortholog difference has been ignored. A future criterion for inclusion would be whether the differently numbered orthologs have the same activity; that is, can the gene from one animal substitute for a family member in a different species?

Table 2. Ortholog Differences in Transcription Factor Gene Expression
GeneSpeciesaTissueDifference
  • See Table 1 for references.

  • a

    m, mouse; c, chicken; x, Xenopus; z, zebrafish.

  • b

    Khan et al. (2002) report differences in Tbx expression in newt limbs. There are other differences reported for limbs of other urodele amphibians. These include Fgf-8 expression in the mesenchyme of axolotl limb buds, rather than the apical ectodermal ridge (Han et al.,2001), and differences in Shh expression (Stopper & Wagner,2005).

Pax8mThyroidXenopus Pax2, zebrafish Pax2a
Cdx1mIntestineXenopus Xcad2, zebrafish cdx1b
Cdx2 (major)mIntestineXenopus Xcad1, zebrafish cdx1b
Tbx6m, x, cPSM/somite borderZebrafish Tbx24
Msx1m, x, cBrain, spinal cord roof plateZebrafish msxB,C,E
Math1m, cSpinal cordXenopus Ath3
Mash1m, c, zSpinal cordXenopus Xash3
Gbx2m, x, cBrainZebrafish gbx1
Tbx4m, x, c, zHindlimbNewt Tbx5b
Tbx5m, x, c, zForelimbNewt Tbx4b
Hox6–9mThoracic somitesSnake–different order

THE MOLECULAR HAECKELS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Our drawings are shown in Figures 1, 2, 3, 4, 5, 6, 7, 1–7. Traditional colors have been used with blue for epidermis, green for central nervous system, light green for neural crest, red for mesoderm, and yellow/orange for endoderm. Each tissue or organ is constructed using the acronyms of the important gene expressions in the development of the tissue or organ. For some tissues, there are many gene expressions known, so the whole tissue texture is composed of acronyms. This is the case for the central nervous system, neural crest, somites, limb buds, heart, liver, and pancreas. For other tissues, only a few key expressions have been identified. These tissues are represented by a pattern with interspersed acronyms.

thumbnail image

Figure 1. Ectoderm. The skin has been removed from the embryo to reveal the brain and spinal cord (green), the midbrain–hindbrain boundary (turquoise) and the cranial and trunk neural crest (light green).

Download figure to PowerPoint

thumbnail image

Figure 2. Mesoderm. The skin has been removed to reveal the mesoderm. The anterior somites have formed, while somitogenesis continues posteriorly. The somites, intermediate mesoderm, heart, and limb buds are composed of gene acronyms, while the lateral plate mesoderm and the mesenchyme of the head and tail are represented by textures.

Download figure to PowerPoint

thumbnail image

Figure 3. Endoderm. The tube of endoderm has buds for the thymus and parathyroid gland, thyroid gland, lungs, liver, and pancreas, and these organs are well-represented by gene acronyms. There are some common gene expressions within the endoderm, as well as some specific ones for stomach, intestine and cloaca.

Download figure to PowerPoint

thumbnail image

Figure 4. Cross-section at the forelimb level. Dorsoventral and other spatially restricted gene expressions are depicted for the spinal cord, somite, and both the ectoderm and mesoderm of the limb bud.

Download figure to PowerPoint

thumbnail image

Figure 5. Eye. The eye region is presented at full-size to show specific eye expressions as well as surrounding brain expressions. The full-size ectoderm picture, from which this figure was taken, is Supp. Fig. S1.

Download figure to PowerPoint

thumbnail image

Figure 6. Heart. The full-size mesoderm picture, from which this figure was taken, is Supp. Fig. S2.

Download figure to PowerPoint

thumbnail image

Figure 7. Liver and pancreas. The full-size endoderm picture, from which this figure was taken, is Supp. Fig. S3.

Download figure to PowerPoint

For some tissues, not only is gene expression conserved but also the spatial location within the tissue of the cells expressing that gene is conserved. Hox genes are expressed with anterior–posterior polarity in the neural tube and the somites and with proximal–distal polarity in the limb buds. Other examples are the dorsal–ventral pattern of gene expressions in the neural tube, the regional expressions within the somite, the posterior expression of Shh in the limb bud, and the expressions of the limb bud ectoderm. In these cases, the placement of acronyms reflected the conserved spatial locations.

When viewed as reduced drawings (Figs. 1, 2, 3, 4, 1–4), the acronyms are not legible. The acronyms can be seen in the full-sized drawings, and examples of the eye, heart, liver, and pancreas are presented (Figs. 5, 6, 7, 5–7). The complete full-sized drawings are in the Supplementary Material (Supp. Figs. S1–4, which are available online).

FUTURE MOLECULAR HAECKELS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Constructing the Molecular Haeckels provides a glimpse into the difficulties that Haeckel must have encountered in drawing his pictures. Which details should be emphasized and which differences should be ignored? Genes were included if they were expressed during development of the same tissue or organ in a range of animals. A more stringent requirement for inclusion would be a demonstration of an important developmental function of the gene in both a teleost and an amniote. A second example is the decision to included different orthologs within a family of transcription factors, mentioned earlier (Table 2).

The molecular database is much richer than the morphological one used by Haeckel. We have identified more than 160 conserved associations of regulatory genes involved in the early development of specific tissues. Many of the genes are expressed in conserved locations within a tissue, adding to the number of molecular characters than can be used.

Nonetheless, there may be a bias in selecting the genes. Many of the included genes were discovered because they have a strong effect on development in one species. Once that was known, its ortholog was cloned and analyzed in other species. This search process is biased toward commonality, the very feature highlighted by the Molecular Haeckels. Future data, generated by microarrays, should give unbiased pictures of genes expressed in particular developmental events in different animals. The problem then becomes identifying gene expression differences that have significant functional differences between animals.

There are several ways to advance the Molecular Haeckels. One way is to design them using the enlarging data base of gene regulatory networks (GRN). Rather than individual gene expressions, tissues would be constructed with conserved pathways and molecular interactions. Giudice and Onorato (2003) provided a prototype for this type of image. A second way to advance is to include timing and the order of gene expression. That would require computer animations of Molecular Haeckels.

Haeckel's pictures have not been forgotten after a century, because they contain a core truth of commonality among vertebrate embryos. These commonalities are more apparent than ever.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSTRUCTING A MOLECULAR HAECKEL: DESIGNING A MORPHOLOGICAL FRAMEWORK
  5. CONSTRUCTING A MOLECULAR HAECKEL: CHOOSING GENES
  6. THE MOLECULAR HAECKELS
  7. FUTURE MOLECULAR HAECKELS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
DVDY_22337_sm_suppinfo.doc123KSupporting Information.
DVDY_22337_sm_suppfig1.tif9696KSupporting Figure 1.
DVDY_22337_sm_suppfig2.tif11949KSupporting Figure 2.
DVDY_22337_sm_suppfig3.tif4766KSupporting Figure 3.
DVDY_22337_sm_suppfig4.tif6800KSupporting Figure 4.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.