An explanation of evolutionary change in animal form in mechanistic terms rests on a classic syllogism. This goes as follows: If the morphological features of an animal are the product of developmental process, and if the developmental process in each animal is controlled by its genomic regulatory program, then the evolution of animal forms is the consequence of evolution of these genomic regulatory programs. This is scarcely a new thought (Britten and Davidson, 1971), but what is new is that now we can do something to study it experimentally. The fields of developmental molecular biology and evolution increasingly overlap, creating an exciting new domain, in fact a new field of bioscience. This is at present in process of redefining itself as it absorbs influences from new initiatives in other disciplines, including phylogenetics, paleontology, comparative genomics, and the newly resuscitated arts of invertebrate anatomy, amongst others. Most would concur with this description of present events. We shall take a step further, however, and imagine that what we are now seeing is the initial phase of a major intellectual realignment, in which the study of the mechanisms by which animal body plans evolve will ultimately be regarded as a branch of regulatory genomics, a central domain of bioscience. If we knew in functional terms the components of specific genomic control elements that result in different morphological outcomes in two animals of common ancestry, we would see exactly what are the essential causal differences in the DNA of these animals. So, by learning how these programs work, the objectives of both evolutionary and developmental inquiry are attained. In fact, these problems are so intertwined as to become indistinguishable, and the solutions can be seen to lie in the genomes of extant bilaterians.
Regulatory Evolution, and Evolution in General
Darwinian evolution has a long and important history, first because it drove the development of the whole concept of evolution per se, and proved that evolution can, did, and does happen; and second because over the last century it has generated a vast calculus for analyzing adaptive and selective processes. The objective has been to be able to treat the outcomes of confrontations between populations of organisms possessing given adaptive properties and their ecological environments, and to define the rules according to which heritable variations spread through populations. But classical Darwinian evolution could not have provided an explanation, in a mechanistically relevant way, of how the diverse forms of animal life actually arose during evolution, because it matured before molecular biology provided explanations of the developmental process. To be very brief, the evolutionary theory that grew up before the advent of regulatory molecular biology dealt with the problem of the origin of novel organismal structures in two ways. The first has been to treat the mechanisms generating novel morphological structures as a black box. New forms were considered to arise "because" the environment changed. But while changes in Precambrian or Ordovician weather, continental shifts, or temperature may have contributed crucial selective forces, they do not generate heads or appendicular forms; only genes do that. The second mode of classical argument was that organismal evolution is the product of minute changes in genes and gene products, which occur as point mutations and which accumulate little by little, providing the opportunity for selection and ultimately reproductive isolation. The major forms this argument has taken have focused on stepwise, adaptive changes in protein sequence, but this is probably largely irrelevant to the evolution of any salient features of animal morphology (see, e.g., Miklos, 1993). In hindsight we can see that since the families of genes encoding the key regulators of development, viz, transcription factors and signaling pathways, are all panbilaterian, as well as are representatives of most other gene families, the differences between animals displaying different body plans are not likely to be explained in general by differences in these proteins. Many remarkable examples of diverse usage of similar genes and gene families in the development and evolution of diverse organisms are to be found later in this book.
There is another version of this argument on which the jury remains out, however, and this concerns gradual mutational changes in czs-regulatory sequence. It is not at all clear as yet what is the relative importance of stepwise mutational change in czs-regulatory sequence relative to other kinds of alteration in DNA sequence, such as transpositional insertions of regulatory modules or of genes in the vicinity of these modules; sequence deletions; local genomic rearrangements; replication of genes or their czs-regulatory target sites; gene conversion, etc. (for general discussions see Dover, 1982, 1987; Britten, 1997; and the following text).
Interpretation of evolutionary change, it seems apparent, is going to take the form that changes in animal morphology, whether great or small, are generated largely by alterations in developmental regulatory sequence. There is no point here in distinguishing between changes in cis- as opposed to irans-regulatory elements, since a change in the expression of the iraws-regulator of a given set of genes is simply a change in the czs-regulatory module controlling the relevant phase of its own expression. When clades of animal become extinct it is specific developmental gene regulatory networks that have disappeared from the earth; or alternatively, it is the genomic regulatory networks of evolutionarily successful clades that have become dominant.
It is impossible to stress too much the importance of phylogeny for understanding developmental regulatory comparisons between different animals. If comparative regulatory genomics holds the key to evolution of morphologies, it does so only if the evolutionary relationships of the respective animals can be perceived correctly. These relationships must be known in order to deduce the polarity of changes; to determine whether apparent similarities are the consequence of conservation or of convergence; and, in general for any precise conclusions regarding evolutionary regulatory history.
In Fig. 1.6 is shown a current consensus phylogeny of all bilaterian phyla. This is essentially a ribosomal DNA (rDNA) phylogeny (see legend for references to the multiple studies supporting each individual region of the topology). The bilater-ians are divided into three great clades. On the top of Fig. 1.6 (blue) are the ecdysozoans, i.e., animals which molt, including arthropods, nematodes, and many minor phyla. In the center (green) are the lophotrochozoans, i.e., animals displaying spiral embryonic cleavage, such as annelids, molluscs, and flatworms, plus a large number of clades distinguished by tentaculate head structures (lopho-phores) including brachiopods, entoprocts, and many others. On the bottom (red) are the deuterostomes, i.e., echinoderms, hemichordates, and chordates (including ourselves). The arrangement shown in Fig. 1.6 is supported by much evidence in addition to the sophisticated analyses of rDNA sequences on which the tripartite tree is based (Aguinaldo et al., 1997; Adoutte et al., 2000). First, and perhaps most importantly, it is largely consistent with appropriately performed cladistic analyses of morphological characters (Peterson and Eernisse, 2001). Very strong and completely independent support for the assignment of key species within the tripartite arrangement of animal phyla shown in Fig. 1.6 has subsequently emerged from phylogenetic analyses of box gene sequences (de Rosa et al, 1999; Finnerty and Martindale, 1998; Balavoine, 1997). Figure 1.6 is conservative, in that controversial branch placements have been avoided, so that many of the minor phyla are portrayed as equally related to one another (i.e., their relationships are portrayed as coterminal). Some obvious sister groups based on morphology are included in the diagram, such as annelids and echiurans; molluscs and sipunculids; nematodes and nematomorphs; and the panarthropods (i.e., arthropods proper, onycophorans, and tardigrades).
For what follows the most important points to be derived from Fig. 1.6 are its implications with respect to common ancestry. It is clear that the bilaterians as a whole are monophyletic: all bilaterians are more closely related to one another than any is to any other kind of animal. As Fig. 1.6 shows explicitly, they all derive from a common ancestor. This basic conclusion is of course also implied by the fundamental similarity in bilaterian gene repertoires, particularly those gene classes specifically involved in controlling development of the body plan, as discussed above. There is a deep cleavage to the nearest outgroup shown, the cnidarians (i.e., jellyfish, hydra-like creatures, sea anemones, and corals). This is clearly evident in rDNA sequence comparisons, and the differences between cnidarian and bilaterian morphological organization have long been appreciated (see e.g., Hyman, 1940). For example, cnidarians lack the mesodermal layers that are fundamental to the construction of all bilaterian body plans, and they also lack any organ-level morphological and functional structures. Nor do they possess a box gene cluster that includes representatives of all the major groups of box genes found in bilaterians, possessing only linked "anterior" and "posterior" type box genes (Finnerty and Martindale, 1998, 1999; de Rosa et al, 1999).
Each of the three great bilaterian clades (ecdysozoans, lophotrochozoans, deuterostomes) is defined by a unique set of shared characters. This turns out to be extremely important in thinking about the evolution of body plans, because it provides a framework for interpreting similarities and differences in use of regulatory mechanisms during development of various bilaterians. For example, until
Cnidarians Chaetognaths Gastrotrichs Cephalorhynchs i— Nematodes 1— Nematomorphs
-Tardigrades i— Onychophorans '— Arthropods
□ Rotifers Gnathostomulids
— Entoprocts Nemerteans Annelids Echiurans Molluscs Sipunculans Chordates Echinoderms Hemichordates
Deuterostomes Lophotrochozoans 1 Ecdysozoans J Protostomes
FIGURE 1.6 Phylogeny of the bilaterian phyla. This tree is primarily derived from I8S rDNA and is supported by hox gene phylogeny and morphological cladistics. The bilaterians (shown in color) are divided into three major groups: the deuterostomes (red); (for deuterostome phylogeny, Wada and Satoh, 1994; Turbeville et al., 1994; Eernisse, 1997; Cameron et al., 2000); the lophotrochozoans (green); (Halanych et al., 1995); and the ecdysozoans (blue); (Aguinaldo et al., 1997; Giribet and Ribera, 1998). The lophotrochozoans and ecdysozoans are each monophyletic, and together recently it was assumed that annelids and arthropods are evolutionary sister groups, and that therefore their mechanisms of segmentation must be very similar; but now we see that annelids are lophotrochozoans and arthropods are ecdy-sozoans, and both of these large clades include many unsegmented animals. Not surprisingly, when looked at more carefully at a gene regulatory level, segmentation in annelids and arthropods looks a lot less similar than it used to (Shankland and Seaver, 2000). To cite another example, it is apparent from Fig. 1.6 that the deuterostome chordates cannot be considered a sister group of the ecdysozoan arthropods derived from an arthropod/deuterostome common ancestor, contra some earlier speculations (e.g., Arendt and Niibler-Jung, 1994). The phylogenetic positions of arthropods and chordates impinges directly on deductions regarding the evolutionary origins of genomic regulatory programs for many important processes, including development of head structures, CNS, heart, segmentation, etc.
Someday we are going to be able to write down the regulatory network architecture for key features of development that are characteristic of animals representing diverse regions of the phylogenetic diagram. This will take us far beyond the distribution of shared sequence motifs such as now provide the most reliable basis for these diagrams. While phylogenetic sequence analysis provides the quantitative indices of evolutionary relatedness in rDNA sequences that were used to construct the tree in Fig. 1.6, rDNA sequence differences in themselves of course play no causal evolutionary role. Differences in the regulatory network architecture are what cause phylogenetic diversification, and therein resides the "deep structure" that underlies diagrams such as that in Fig. 1.6.
they constitute the traditional protostomes (Giribet and Ribera, 1998). The echino-derm/hemichordate sister grouping is also supported by mitochondrial analyses (Cas-tresana et al., 1998). The relationships shown within the ecdysozoan clade are derived primarily from Schmidt-Rhaesa et al. (1998). Inclusion of the chaetognaths into this clade is from Halanych (1996) and Eernisse (1997); the basal positions of these animals was proposed because although they possess a cuticle with chitin, they lack specific morphological details found in the cuticles of gastrotrichs and the other ecdysozoans. Within the lophotrochozoans the division into apical and basal groups is primarily from Halanych et al. (1995) and Aguinaldo et al. (1997). See Cohen et al. (1998) for the [phoronid + brachiopod] sister grouping. The [annelid + echiuran] and [mollusc + sipunculan] sister groupings are based on unique embryological characters found in each clade (see Ruppert and Barnes, 1994). The rotifer clade (which includes [acantho-cephalans]; Garey et al., 1996) + cycliophorans is supported by I8S rDNA (Winne-penninckx et al., 1998); the [rotifer + gnathostomulid] clade is based on several unique morphological characters (e.g., Rieger and Tyler, 1995). Other relationships are left undefined, so that the various phyla are presented as equally related to a common ancestor. [From K. J. Peterson and E. H. Davidson, unpublished.]
Was this article helpful?