FIGURE 5.8 Regulatory transactivations of Pax6 in vertebrate and Droso-phila eyes. (A) Modular organization of c/s-regulatory elements controlling mouse pax6 expression during development. Two modules control expression in the eye, an upstream element that mediates expression in lens, cornea, tear gland, and conjunctiva (tan); and an intron enhancer responsible for expression in the retina (red). A complex pattern of expression in the brain and spinal cord is activated by elements in another intron (telencephalon enhancer; blue); and expression in the pancreas is controlled by an upstream element (green) near the cornea/lens enhancer module. (B) Function of mouse and fish pax6 eye enhancers: (Bl) embryonic expression of lacz construct driven by mouse retina enhancer, 13.5 dpc. (B2) Same, except the enhancer driving the construct is a similar DNA sequence from the zebrafish pax6 gene (the curved arrow in BI and the two arrows in B2 denote the dorsal edge of the retinal domain where these constructs do not express as robustly as does the endogenous gene). [(A, Bl, B2) From Kammandel et al. (1999) Dev. Biol. 205, 79-97.] (B3) Expression of lacz construct driven by mouse pax6 lens/cornea enhancer, 12.5 dpc. Lacz staining in section is shown: le, lens; plfc, primary lens fiber cells; cor, cornea; pr, prospective retina; rpe, presumptive retinal pigmented epithilium. This module produces expression in cornea plus lens, not in retina. [(B3) From Williams et al. (1998) Mech. Dev. 73, 225-229, copyright Elsevier Science.] (C) c/s-Regulatory elements of some lens crystal-lin genes. Known transcription factors are indicated, including pax6 (dark blue, regulatory significance of binding interactions shown experimentally; light blue, observation that Pax6 paired domain binds in vitro). aA(c), chick aA crystallin; aA(m), mouse aA crystallin; 81(c), chick 81 crystallin. Transcription factors which serve as activators are shown above line representing the DNA; those below the line function as repressors. [(C) Simplified from Cvekl and Piatigorsky (1996) BioEssays 18,621-630, copyright Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc.] with information from llagan et al. (1999) added (see these sources for further details and additional factors which regulate these and many other lens crystallin genes). (D) Control of rhodopsinl gene of Drosophila by Pax6, the product of the eyeless (ey) gene. (Dl) Expression of rhodopsin.lacz construct in section of fly head; arrow indicates lamina. Staining is confined to photoreceptor cells (RI-R6) and their projections to the lamina. (D2) This construct is not expressed when the Pax6 binding region is mutated. [(D)
From Sheng et al. (1997) Genes Dev. 11,1122-1131.] (E) Expression of retinal markers in ectopic eye-like structures of Xenopus tadpoles, produced by injection of Pax6 RNA into animal blastomeres at 16-cell stage. (El) Normal eye, to display markers, here revealed by immunocytology: red, Islet-1 (transcription factor), present in retinal cell nuclei; green, rhodopsin of photoreceptors; blue, Müller cells (identified by an antibody to glutamine synthetase) which are closely associated with neural cells of the inner retinal layer. (E2) Same markers in ectopic eye. [(E) From Chow et al. (1999) Development 126, 4213—4222 and The Company of Biologists Ltd.] (F) Interactions among Drosophila genes for which evidence is shown in (G). The four genes included in the diagram are all involved in pattern formation for eye development (reviewed by Czerny et al., 1999): twin of eyeless (toy), and ey, encode largely similar though not identical Pax6 factors, but have different c/s-regulatory systems; sine oculis (so), encodes a homeodo-main nuclear factor required for eye development; and eyes absent (eya), encodes a nuclear protein which physically interacts with the so gene product (Pignoni et al. 1997). (G) Epistasis experiments. (GI, G2) toy is upstream of ey. (GI) Ectopic eye (indicated by red eye pigment) appearing on a malformed leg of a fly in which toy expression was forced to occur in the leg imaginal disc under control of a dpp-Gal4/ UAS-toy expression system; (G2) same experiment, but carried out in an ey mutant fly. Though the leg is still malformed, no ectopic eye forms, because ey is required for eye development and toy works through ey [cf.(F)]. Other evidence thit toy is upstream of ey (Czerny et al., I999) is that the toy gene is expressed normally in ey mutants; that ectopic toy expression causes ectopic ey expression but not vice versa; that ectopic ey expression causes appearance of ectopic eyes (Halder et al., 1995) in toy mutants, but (as illustrated here) not vice versa; and that the Toy protein binds directly to ey c/'s-regulatory sites essential for ey expression. [(Gl, G2) From Czerny et al. (1999) Mol. Cell 3, 297-307, copyright Cell Press.] (G3, G4) ey is upstream of so and eya: (G3) third instar wing disc of so mutant, ectopically expressing Eya, as detected by immunocytology (orange fluorescence). Expression occurs in a region along the A/P boundary where ey expression has been forced by use of dpp-Gal4/UAS-ey. This does not cause ectopic eye formation, because the so gene product is required for eye development. (G4) Converse experiment: expression of ey forced along A/P axis of wing disc just as in (G3), but in an eya mutant; expression of so gene product is here displayed (orange fluorescence). Ectopic eyes do not form because the eya gene product is also required for eye development. [(G3, G4) From Halder et al. (1998) Development 125, 21812191 and The Company of Biologists Ltd.] Other evidence for the relation between ey and so shown in (F) is that the Ey protein binds directly to so cis-regulatory elements and activates this gene (Niimi et al., 1999). Furthermore, ey is expressed earlier than are so and eya, and ey expression is required for expression of both eya and so in the eye disc, but not vice versa. Ectopic expression of eya, so, and also dachshund (another gene downstream of eya and so in the eye patterning network) produces ectopic eyes, more so when they are expressed in combination (Pignoni et al., 1997; Chen et al., 1997; Czerny et ai, 1999).
Gehring and Ikeo, 1999). This looks to be a canonical example for the kind of evolutionary process envisioned in Fig. 5.7.
Other cases come to mind as well. Heart development is one: tinman (tin) and its vertebrate orthologues, the nkx2.5 and nkx2.3 genes, are alike required for upstream pattern formation processes in the development of the heart in Droso-phila and in vertebrates, respectively (Chapter 4). But these factors also exert distinct cfe-regulatory control over terminal heart cell differentiation genes, e.g., in amniotes, genes encoding cardiac atrial natriuretic peptide (Lee et al, 1998; Durocher and Nemer, 1998) and cardiac a-actin (Sepulveda et al., 1998). Furthermore, activation of these differentiation genes requires the participation of Gata factors, and Gata regulators control several other terminal differentiation genes in the heart in addition to the above (Wang et al, 1998; Grepin et al., 1995; Molkentin et al., 1998). On the other hand, gata genes too are required in various upstream pattern formation processes in heart development, in both mouse and Drosophila (Chapter 4). The morphogenetic features of Drosophila and vertebrate hearts are (of course) different: the nkx2.5 gene participates in multiple vertebrate-specific functions beyond initial specification of the progenitor field, viz, formation of the heart tube, specification of its asymmetric chamber domains, and completion of its complex morphology (Lin et al., 1997; see above). Different interactions are required of Nkx2.5 than of the Tin factor and if introduced into the genome nkx2.5 cannot rescue heart formation in Drosophila tin mutants, unless it is equipped with a piece of the tin coding sequence (Ranganayakulu et al., 1998). Here again control of terminal heart cell differentiation processes may be the shared ancestral property, and the pattern formation processes mediated by both gata and tin/nkx2.5 genes may have been inserted stepwise into developmental networks for heart formation, differently in the evolution of different animal clades.
Perhaps these steps are indicated by the modular cz's-regulatory organization of the Drosophila tin (Fig. 4.8E) and the mouse nkx2.5genes (reviewed by Schwartz and Olson, 1999). The vertebrate heart has a modular anatomical construction (Fishman and Olson, 1997), and expression of the nkx2.5 gene in the various subdomains of the developing heart turns out to be controlled by many distinct czs-regulatory elements. These operate in the different spatial domains of the forming heart, at different stages, e.g., when the heart is a simple linear tube, or when it is forming left and right ventricles, or later in development. Each enhancer responds to different inputs, so that lacz constructs which are made from them are each expressed differently (for references, see Schwartz and Olson, 1999). As suggested in Fig. 5.7, modularity in the control systems of genes that function during pattern formation for a body part may provide a living trail of the evolutionary stages through which the patterning network was assembled.
How generally could the concept of body part evolution here on the table be applied? There are a great many examples of the paradox with which this section begins. Some are more trivial than others. For example, it is improbable that any particular use of BMP signaling pathways could reliably indicate survival of a dedicated ancestral bilaterian pathway for making a given structure, since BMP family members are deployed at some point in practically every structure in the body (reviewed by Kingsley, 1994). An interesting case, on the other hand, is that of the caudal (.cad) genes. These encode a family of homeodomain regulators that throughout Bilateria seem to be used in the embryonic specification of the hind end of the animal, including the posterior gut (Marom et al., 1997; Gamer and Wright, 1993; Katsuyama et al., 1999)- The cad genes clearly execute upstream pattern formation functions. For example, in Drosophila cad activates forkhead and wg genes needed for posterior gut formation in the embryo (Wu and Lengyel, 1998). The cad gene is necessary and sufficient for specification of the most posterior adult body segment, in which process it activates the brachyen-teron and evenskipped genes in the hindgut, and the distal-less gene in the forming anal plates (Moreno and Morata, 1999). In mouse the cdx caudal orthologues provide direct transcriptional input into box genes (Charité et al., 1998) another upstream function; and in Xenopus cad overexpression or loss of function disturbs embryonic A/P patterning (Epstein et al., 1997). But once again cad/cdx genes may have terminal differentiation roles as well. The Drosophila cad gene controls expression of the short gastrulation gene, which encodes a secreted molecule required for cell movement during hindgut invagination (Wu and Lengyel, 1998). In chick and mouse the cdx genes are expressed even in adult intestine (James et al., 1994; Marom et al., 1997). The terminal roles of cad genes, perhaps control of differentiation gene batteries in the posterior gut, may have been their ancestral evolutionary function. Because these genes were activated at the posterior end, cooption could have led during evolution to their inclusion in morphogenetic patterning networks for posterior structures. A similar situation obtains at the opposite end of the animal. Transcription factors encoded by the orthodenticle gene of Drosophila and the otx genes of vertebrates participate in the pattern formation networks underlying brain development in both (for Drosophila, reviewed by Cohen and Jurgens, 1991; Finkelstein and Boncinelli, 1994; You-nossi-Hartenstein et al., 1997; for Xenopus, Pannese et al., 1995; for mouse, Fig. 4.1 and references in Chapter 4 related thereto). But the otx genes are also expressed in specific neurons, and in specific cell types of olfactory, ocular, and acoustic sense organs, which are affected in otx knockouts (Acampora et al., 1996). Perhaps the original role of otx genes was control of sensory neuron differentiation, which necessarily was installed at the anterior end of the bilaterian ancestor. These genes could then have been coopted by inclusion in diverse patterning gene networks during the evolution of brains in the different bilaterian clades.
The argument made in this section boils down to the proposition that what is actually shared amongst all bilaterians in respect to particular body parts is only the basal repertoire of differentiation gene batteries required in each, including their specific controllers. The pattern formation process underlying the morphogenesis of analogous body parts in different clades are the outcome of a long series of cooptions and reorganizations, that have retained selective value at each step due to expression of their cell type-specific functions. This is a rather different view of the evolutionary process from that in which the shared use of upstream regulators is held as evidence for conservation of a genomic program for pattern formation. That view leads to the position that seeing is not believing; that though analogous adult body parts of diverse clades look different, beneath the surface there lies hidden a common, unchanging morphogenetic deep structure inherited from the bilaterian common ancestor. But if there is one impression that all the figures and examples in this chapter should convey, it would be of the mutability of pattern formation networks, and the ubiquity and prevalence of evolutionary cooption of regulatory genes in these networks.
CONCLUDING COMMENT: CONCEIVING EVOLUTION AS A PROCESS OF CHANGE IN REGULATORY GENE NETWORKS
For anyone interested in mechanism, there is in fact no other way to conceive of the basis of evolutionary change in bilaterian form than by change in the underlying developmental gene regulatory networks. This of course means change in the czs-regulatory DNA linkages that determine the functional architecture of all such networks. A glance toward the horizon suggests that some of the most intriguing concepts and problems in evolutionary biology will in some measure undergo redefinition as comparative knowledge of gene regulatory networks expands.
Take for example the concept of homology as applied to usage of genes. Whole symposia, treatises, and a good deal of argument are devoted to the meaning and implications of homology in evolution (Meyer, 1999). But with reference to the genetic basis of evolutionary change, we can ask the precise question "what is a homologous use of a given gene in a comparison of diverse organisms," and we can look forward to a precise answer. This emerges from the structure of gene networks. A gene performs a homologous function in two animals if at least some of its upstream or its downstream linkages (or both) remain the same in the two genomes, and the function it performs is descendant from their common ancestor. Otherwise homologous use cannot be supported.
Or consider the appearance of sudden "bursts" of evolutionary change in morphology, followed by periods of stability, which is suggested by some aspects of the bilaterian fossil record. This is somewhat a fuzzy proposition from the outset, since everything depends on what level of evolutionary change is meant. Considered in terms of regulatory gene networks, small changes at the level of development of terminal differentiation gene batteries are likely to be going on continuously (as in the bristle patterns of Fig. 5.6). These changes affect species-specific and intraspecific characters, and they are fixed according to population parameters, selective factors and so forth. But the architecture of the gene networks that control more momentous aspects of form requires that some changes will have major effects, and others smaller effects. For instance, large morphological effects will follow relocation or duplication of progenitor fields for given body parts (Chapter 4), as clearly occurred in the evolution of vertebrate appendages (Fig. 5.4C). The result of a successful upstream change of this kind is to create a new set of possibilities downstream, a "morphospace" to fill: this indeed happened, for example, when paired limbs evolved in diverse ways in different gnathostome clades. Given that gene networks have upstream and downstream polarities, major evolutionary change in developmental process must be discontinuous in rate.
The fields of development and evolution are convergent because both subjects are rooted in the genomic regulatory programs for body part formation. And the convergence is going to go much further than it has. Very soon we will be able to understand exactly how the cis-regulatory systems of given gene networks function to generate the spatial aspects of development. By comparative reference to other phylogenetically informative genomes, we will thereby learn just what happened at the DNA level which led in evolution to various forms of the body part. This will perhaps require something like what comparative linguists do when they reconstruct extinct proto-languages from a family of extant descendants: i.e., reconstruction of ancestral-gene regulatory networks from those of related modern animals. But first we need to obtain and to understand in detail some of the important developmental regulatory networks in the genome.
This leads to a final comment. Though the ancestors of modern animals are extinct the evidence of how they worked is not. The evidence is swimming, walking, and flying around outside, in the DNA of modern bilaterians. What happened in evolution will emerge from knowledge of the regulatory pathways of development, the subject of most of this book; and from application of computational tools for tracing, analyzing, and conceiving regulatory network architecture. Comparative sets of genomic network "wiring" diagrams will emerge. Strange as it may seem, it is these that will tell us what the Bilateria are and where they came from.
Was this article helpful?