Castresana et al., I998; Turbeville et al., 1994; Wada and Satoh, 1994). The arthropods, which are ecdysozoans (see Fig. 1.6), serve as an outgroup. Numbers represent acquisition of characters relevant to the following figures: I, bilateral organization; 2, paired anterior, middle, and posterior larval stage coeloms which give rise to the major mesodermal components of the adult body plan; 3, fivefold radially symmetric adult body plan of echinoderms; 4, calcite endoskeleton of echinoderms; 5, hemichordate proboscis, incorporating anterior coelom(s); 6, chordate notochord. (B-F) Cooptions of brachyury transcriptional regulators in deuterostome evolution. (B) Distribution of Strongylocentrotus purpuratus brachyury transcripts visualized by in situ hybridization in larva of S. purpuratus. Expression is observed in the invaginating vestibule (arrow) and the middle coelom (hydrocoel; arrowhead) with which the vestibule makes contact on the left side of the larva to form the rudiment of the adult. Expression is also observed in the right axohydrocoel (arrowhead; the anterior coelom or axocoel, and the middle coelom, the right hydrocoel, are fused on this side). E, esophagus; S, stomach; I, intestine. [(B) From Peterson et al. (1999b) Dev. Biol. 207, 419-431.] (C) Expression of brachyury gene in metamorphosing larva of an enteropneust hemichordate, Ptychodera flava, by in situ hybridization. (CI-C3) Dorsal views, anterior up; (C4, C5) sections at planes indicated in (CI) and (C2). (CI) Expression in anterior coelom of the proboscis in a competent larva. (C2) Metamorphosing larva (one day after stimulus). Expression of brachyury gene is now also seen in middle coelom, and in caudal region. (C3) Three-day juvenile; expression of brachyury in coelomic mesoderm of all three body regions. (C4) Transverse section through proboscis, showing expression in mesodermal cells of protocoel (Pr; these cells are the progenitors of the proboscis muscle), but not in ectoderm; (C5) Transverse section through hindgut: both endoderm and posterior coelomic mesoderm express brachyury but not ectoderm as seen in inset (Mt, metacoel, posterior coelomic lining of body wall; I, intestine). [(C) From Peterson et al. (1999a) Development 126, 85-95 and The Company of Biologists, Ltd.] (D) Expression of brachyury gene in presumptive notochord of a larvacean urochordate (Oikopleura dioica), visualized by in situ hybridization, lateral view, anterior left. Larvaceans retain the swimming tail in the adult form. An early hatched stage is shown. The brachyury gene is also expressed in the developing hindgut


additional brachyury functions. In both sea urchin and hemichordate embryos the brachyury gene is expressed at the blastoporal (prospective anal) and oral ends of the gut (as well as in the secondary mesenchyme in sea urchin embryos; for sea urchins, Harada et al., 1995; Peterson et al., 1999b; and unpublished data; for hemichordates, Tagawa et al., 1998; Peterson etal., 1999a). However, the embryonic phase of brachyury expression in the oral domains of these embryos has nothing to do with adult body plan formation, since the embryonic expressions are extinguished before the larval stage when the adult body is formulated, and since they occur in regions that are not incorporated in the adult body plan. The use of the brachyury gene in the oral regions of these embryos is a cooption particular to the echinoderm plus hemichordate clade (cf. Fig. 5.1A). However, each group has coopted the gene for different developmental purposes in respect to adult body plan formation.

(not visible here). [(D) From Bassham and Postlethwait (2000) Dev. Biol. 220, 322-332.] (E) Expression of brachyury genes in amphioxus (Branchiostoma floridae); there are two similar brachyury genes in this species and the probe detects both. Expression shown by in situ hybridization in neurula stage embryo, dorsal view, anterior left, and in presumptive notochord (pn) plus caudal mesoderm. [(E) from Holland et al. (1995) Development 121, 4283—4291 and The Company of Biologists, Ltd.] (F) Expression of brachyury (T) gene product, visualized by immunocytology in 8.5 dpc mouse embryo. Expression is seen in notochord (nc), notochordal plate (npl), and caudal primitive streak (ps) mesoderm; al, allantois; so, somites. [(F) From Kispert and Herrmann (1994) Dev. Biol. 161, 179-193.] (G) Expression of orthologous brachyury gene of Drosophila, the brachyenteron gene. (Gl) brachyenteron transcripts visualized by in situ hybridization at late syncytial blastoderm stage (cycle 14), anterior left. The gene is expressed in the region from which hindgut, anal pads, and posterior mesoderm will develop. [(Gl) From Kispert et al. (1994) Genes Dev. 8, 2137-2150.] (G2) Brachyenteron protein at gastrula stage visualized by immunocytology, same orientation as in (Gl). The hindgut and the primordium of the posterior visceral mesoderm (arrowhead) display the protein. [(G2) From Kusch and Reuter (1999) Development 126, 3991-4003 and The Company of Biologists, Ltd.] (H) Expression of Sphox3 gene in larval rudiment of adult S. purpuratus, visualized by in situ hybridization; view from future oral surface. Expression occurs in a fivefold radially symmetric pattern, in mesodermal structures related to the dental sacs. [(H) From Arenas-Mena et al. (1998) Proc. Natl. Acad. Sci. USA 95, 1306213067, copyright National Academy of Sciences, USA]. (I) Detection of Engrailed protein in a juvenile brittle star (an ophiuroid, Amphipholis squamata) by immunocytological staining. Expression is observed in a fivefold radially symmetric pattern, in regions bounding skeletogenic domains in which the endoskeletal plates are at an early stage of formation. (J) Distribution of Distal-less protein observed by immunocytology in the invaginating vestibule (between arrowheads in inset), the neuroectodermal anlage of the adult body plan in larva of the echinoid S. drobachiensis. [(I-J) From Lowe and Wray (1997) Nature 389, 718-721, copyright Macmillan Magazines, Ltd.]

A specific feature of echinoid echinoderms (i.e., sea urchins) is their mode of indirect development. The adult body plan forms within a feeding larva. The oral surface of the adult, including its radially symmetric water vascular and central nervous systems, arise within a multilayered imaginal structure, the "rudiment." This imaginal structure is constituted initially from the union of an invaginating neuroectodermal pouch, the "vestibule," and the left middle coelom of the larva, the "hydrocoel." As shown in Fig. 5.IB, in sea urchins the brachyury gene is expressed in both the invaginating vestibule and the hydrocoel to which it is apposed at the beginning of adult rudiment development. The hydrocoel is equivalent to the middle coelom of the hemichordate. But Fig. 5.1C shows that all three of the coeloms of the adult body plan of Ptychodera flava express the brachyury gene during metamorphosis. Expression in the posterior coelom is the ancestral bilaterian usage; expression in the middle coelom is an echinoderm plus hemichordate character; but expression in the mesoderm of the anterior coelom, which forms the hemichordate proboscis, is a cooption particular to this group, just as the proboscis is an anatomical feature particular to hemichordates.

The brachyury gene assumes its most famous function in chordates, as a transcription factor required to specify notochord (reviews op. cit.). Expression of brachyury genes in the notochord of a larvacean ascidian is shown in Fig. 5. ID; in the notochord of amphioxus in Fig. 5.IE; and in the notochord and notochord progenitors of a mouse embryo in Fig. 5.IF. In all of these examples expression can also be seen at the caudal end of the embryo (see legend for details and references). But the phylogeny in Fig. 5.1 shows that the famous notochordal role of brachyury is also a cooption, peculiar to the chordates, and distinct from its ancestral panbilaterian function at the caudal end of the embryo.

In summary, we see in Fig. 5.1 unique brachyury cooptions for every one of the deuterostome clades included in Fig. 5.1 A. Some other examples, which unmistakably indicate cooptions to the unique developmental processes of echinoderms, are shown in Fig. 5.1H-J. Figure 5.1H displays transcripts of the hox3 gene in the imaginal structure from which the radial body plan of the sea urchin arises. The expression domains are associated with the five developing dental sacs (Arenas-Mena et al., 1998). In Fig. 5.11 the radially symmetric distribution of the engrailed gene product is seen in a developing brittle star (Lowe and Wray, 1997). Expression of this gene occurs at the boundaries of the skeletogenic domains within which the endoskeletal elements of the juvenile are being laid down, near the bases of the future arms. Since the radially organized body parts to which the patterns of Fig. 5.1H and I pertain are echinoderm-specific (characters 3 and 4 in Fig.5.1A) with respect to the bilateral ancestor (character 1), these transcriptional regulatory genes also must have been coopted to new patterning functions. In Fig. 5.1J the panbilaterian distal-less gene which we last met in Drosophila imaginal discs, is seen to be expressed in the invaginating vestibule of the sea urchin larva. This is an echinoid-specific cooption of the distal-less gene (Lowe and Wray, 1997), as only the echinoids (i.e., sea urchins) produce a vestibule. The cooptions illustrated in Fig. 5.1H-J may indeed have occurred during the evolution of the particular echinoderm classes represented in these examples (Lowe and Wray, 1997).

The examples here are mainly cases of evolutionarily novel gene use early in the transcriptional patterning of important body parts: the notochord in chordates; the proboscis in the hemichordate; and the major components of the imaginal rudiment in the sea urchin. But cooption happens at every level and affects all of the stages of body part formation sketched in the last chapter. It is the fundamental process by which evolutionary change in bilaterian form has occurred, from the major changes in body plan that define high level taxonomic groups (e.g., those of Fig. 5.1A) to the detailed changes that occur in speciation.


The A/P hox gene patterning system is widely regarded as the definitive shared developmental mechanism of the bilaterians. This is because during development arthropods and chordates, about which we know by far the most, generally express genes of the hox cluster(s) in the same sequence along their A/P axes. There are relatively few observations on animals belonging to other groups, except for Caenorhabditis elegans, which has a degenerate and rearranged hox complex and in which only parts of the general A/P expression pattern can be discerned (Ruvkun and Hobert, 1998; Van Auken et al., 2000; Brunschwig et al., 1999-)- The most important and unusual aspect of the A/P expression patterns of hox cluster genes is that they are "colinear" with the order in which these genes occur in the chromosome; that is, at least the anterior boundaries of the hox gene expression domains lie in the same posterior-to-anterior morphological sequence as the 5' to 3' sequence of the genes within the cluster(s) (reviewed by McGinnis and Krumlauf, 1992). Some exceptions are known, due to the dynamic quality of hox gene expression patterns, particularly in mammals where there are four clusters and in which individual hox genes have acquired special functions. Some genes are expressed a bit out of order at given stages (e.g., hoxbl in r4 of the mouse hindbrain at day 8.5; e.g., Fig. 4.9G). But these special situations are irrelevant to the generality of colinear A/P expression. Colinearity means that some aspect of the chromosomal gene arrangement is utilized in the spatial control mechanisms that determine the domains of hox gene transcription. The organization of the cluster could be important either in respect to setting up the initial domains of hox gene expression or in maintaining stable patterns that are initially installed by specialized czs-regulatory modules (as in the examples in Fig. 4.9). How colinearity works remains a subject of much interesting debate, which, however, does not concern us here. We shall simply take for granted that it does work (see, Gellon and McGinnis, 1998; Kmita et al., 2000). The essential point is that the hox gene clusters are to be thought of as vectorial transcriptional patterning systems: they produce an ordered sequence of expression domains in embryonic space. Such systems are potentially very useful anywhere a transcriptional specification sequence is required to be arranged directionally along some kind of an axis.

Numerous experiments in both Drosophila and mouse have demonstrated regional developmental failures in specification along the A/P axis if given box genes are knocked out or mutated (for general review of the observations that provide this foundation see McGinnis and Krumlauf, 1992). However, box genes do not "make" the anterior or middle or posterior domains of the forming body plan. The genes of the box clusters are activated after the primary spatial regulatory systems of the embryo have set up transcriptional domains that denote its major future components, such as its head, tail, gut, central nervous system, body wall, and so forth. In indirectly developing animals the box gene cluster is not used at all for formation of the completed embryo, and in postembryonic larval development the genes of the cluster are all mobilized only in tissues from which elements of the adult body will form (e.g., in a sea urchin, Arenas-Mena et al., 1998; in an indirectly developing polychaete annelid, Peterson et al., 2000a). The general view is not that individual box genes directly act to produce heads or tails, but that the vectorial expression pattern of the box gene cluster is essential for installing the correct developmental pattern along the A/P axis of the body plan. And that A/P axial patterning by genes of the box cluster is a deeply conserved and fundamental bilaterian feature.

But when we come to mechanism there is something paradoxical about this last conclusion. Consider the example of rhombomere specification, on which we expended some attention in Chapter 4. This is a specific and clear case of axial A/P patterning (cf. Figs. 4.5 and 4.9), in which the rhombomeric expression domains are in general colinear with gene order. However, this cannot be an example of an ancient, conserved patterning mechanism: animals that are not vertebrates do not generate rhombomeres, so the detailed genetic regulatory network that sets up rhombomeric expression patterns cannot be conserved, except within the vertebrates. We saw parts of that network in Fig. 4.9H: its architecture, and the specific connections indicated, must (at least) in part represent the genomic anatomy of a vertebrate cooption. At minimum, key components of this network must be novel with respect to the antecedents of the vertebrates. Furthermore, many of the network connections link genes of different box clusters in specific ways, and lower deuterostomes, i.e., amphioxus (Garcia-Fernandez and Holland, 1994) and echinoderms (Martinez et al., 1999) have only single box gene clusters. Therefore all intercluster regulatory linkages are necessarily cooptions. What we need to understand are the evolutionary pathways that have led mechanistically to novel box cluster gene functions, which nonetheless still mediate axial A/P patterning. In the real terms of specific regulatory linkages these specific functions are not in detail "old," let alone panbilaterian.

Hox gene products act at many different levels of the developmental regulatory apparatus. Some even operate in single differentiation pathways, e.g., box7 in aboral ectoderm of sea urchin embryos (Angerer et al., 1989); or labial in copper cells of Drosophila (Hoppler and Bienz, 1994; see below). However, by far the most important functions of box genes are interventions in the os-regulatory networks that effect transcriptional pattern formation. They do this by altering expression of genes that encode signaling components and transcription factors (for reviews see Carroll, 1995; Weatherbee and Carroll, 1999; Gellon and McGin-nis, 1998; Bilder et al., 1998; Bienz, 1994). The following example shows this in detail.

A Specific Case of A/P Patterning: How a box Gene Does its Job

Among the now classic examples of A/P patterning mediated by box genes is the diversification of segmental appendages in Drosophila (Lewis, 1978). Specific box genes affect the identity of halteres, wings, and legs, and repress antennal fate, acting on developmental processes within the imaginal disc from which these structures derive (reviewed by Weatherbee and Carroll, 1999). The halteres form from the dorsal imaginal discs on the third thoracic segment (T3), and the wings from the corresponding discs on the second thoracic segment (T2). The halteres, which are small balancing organs, lack many of the specific pattern features of the wing (for latter, Fig. 4.7). One of the most impressive definitions of the "power of the box genes" was the demonstration (Bender etal, 1983) that a complete loss of the ubx gene product causes the haltere imaginal disc to give rise instead to a wing. In these mutants execution of the normal T2 imaginal disc program for wing development remains unaffected (ubx is of the same subclass as the vertebrate hox6-8 genes; de Rosa et al., 1999). The result is a four-winged fly, illustrated in Fig. 5.2A. The direct implication is that wing patterning functions in the haltere imaginal discs are repressed by the Ubx transcription factor, which at the key time is present in the T3 disc but not the T2 disc.

We now know a good deal about exactly what the Ubx factor does in the haltere disc to prevent wing patterning from taking place: it acts as a repressor for at least six different wing-specific patterning functions (Weatherbee et al, 1998). Some examples are shown in Fig. 5.2B-E, and a summary of known Ubx effects on the wing patterning network is reproduced in Fig. 5.2F. Figure 5.2B concerns expression of one of the transcriptional regulators of wing disc patterning that we encountered earlier, spalt-related (salt), which acts as a controller of vein positioning (Fig. 4.7C6, E2). The salr gene is turned off by Ubx in the haltere, and is repressed in the wing if ectopic expression of the ubx gene is forced to occur there. Similarly, expression of the gene encoding DSRF, a transcription factor that is required to set in train the differentiation of the collapsed intervein epithelium of the wing surface, is repressed by Ubx (Fig. 5.2E). So also is expression of Wg in the posterior region of the haltere disc (Fig. 5.2D). These effects are all autonomous and they are likely to be direct effects on the respective c/s-regulatory elements; and this is even more clearly so for the example in Fig. 5.2C. The vestigial (vg) gene quadrant enhancer (for which see Fig. 2.7), is normally silent in the pouch region of the haltere disc. But as illustrated in Fig. 5.2C, a quadrant enhancer vg.laczconstruct springs to life within a Ubx^ clone made in a haltere disc (Weatherbee et al., 1998). The summary diagram in Fig. 5-2F shows that Ubx represses regulatory functions needed to set up a number of different transcriptional patterning domains in the developing wing: it blocks expression of a factor that mediates growth in the wing blade (i.e., Vg); of a factor that controls vein designation (i.e., Salr); of factors that are necessary for peripheral nervous system development (i.e., the AS-Cfactors); of a differentiation factor for intervein epithelium (i.e., DSRF); and upstream of this it interferes with expression of signaling ligands (i.e., Ser and Wg) which are involved in the initial spatial specification of some of these transcriptional domains.

So we have here a clear explanation, at the level of the regulatory pattern formation network, of how an A/P specification function of a box gene actually works. But this function is like the rhombomere patterning example discussed above, in that it also has to be considered an evolutionary cooption, and not a very ancient one at that. For though the ubx gene is expressed similarly throughout insects (for references, Abzhanov and Kaufman, 2000), less derived orders do not differentiate their hindwing and forewing appendages; in dragonflies, for example, the T2 and T3 appendages are similar and the adult has four wings. The most basal insects do not even have wings. Therefore the cooption of the ubx gene to repression of wing patterning in the haltere disc necessarily postdated the origin of the wing patterning system in which it intervenes. It must have occurred by steps, in which the precursor of the haltere progressively lost wing size and pattern elements. This would account for both the parallel sites of Ubx intervention, and the ultimate effectiveness of the patterning blockade (Weatherbee et al., 1998).

Evolutionary Changes in box Gene Expression in the Arthropods

The arthropods include among living animals insects plus crustaceans, myriapods (i.e., millipedes and centipedes), and chelicerates (spiders, mites, and horseshoe crabs). The basic anatomical similarities in the segmental organization of these animals provides a leg up in considering evolutionary homologies among their diverse body plans. In insects many of the segments are distinguished from one another by special sensory, feeding, and locomotory appendages and other structures; while at the opposite extreme, in myriapods most segments are alike. The arthropods offer illuminating comparisons of box cluster functions in patterning segmental appendages along the A/P axis. In terms of box cluster genes the major arthropod groups appear to share the same genomic equipment, so the differences in their body plans cannot be regarded as the consequence of differences in their box gene complements (Akam, 1995; de Rosa et al., 1999; Grenier et al., 1997). To the extent that the segmental morphologies of these animals depend on box gene function, the diversity of these morphologies is due to change in the regulatory linkages which affect box gene targets, and by which their own domains of expression are controlled.

FIGURE 5.2 How Ubx blocks the wing transcriptional patterning system in the haltere disc of Drosophila. (A) Four-winged fly, resulting from total loss of ubx gene function. In this animal the imaginal discs of the third thoracic segment have executed the same spatial patterning functions as the wing discs of the second thoracic segment. [(A) From Bender et al. (1983) Science 221, 23-29, copyright American Association for the Advancement of Science.] (B-E) Examples demonstrating Ubx repression of specific genes that participate in wing patterning, displayed by immuno-cytology. The initial A/P patterning systems of wing and haltere discs are the same, as marked by expression of engrailed in the posterior domains and by the stripe of Dpp along the A/P boundary of the distal pouch of the discs (not shown): Ubx does not affect this early A/P patterning system (Weatherbee et al., 1998). (B) Effects of Ubx on salr expression (green). (BI) The sair gene is expressed in the central portion of the wing blade where it is necessary for spatial specification of certain veins (see Fig. 4.7E), but it is not normally expressed in this region in the haltere disc (right side of figure). (B2) Ectopic expression of salr in a ubx~ clone (arrowhead) in a haltere disc: Ubx protein is shown in red. (B3) The converse experiment: ectopic expression of the ubx gene in the blade region of the wing disc (red; white arrows) blocks salr expression in the ubx+ cells. (C) Repression of the quadrant enhancer of the vestigial gene (vg*3). The figure shows a haltere disc from an animal carrying a vg^.lacz construct. Ubx protein, shown in red, is present throughout, except in clones (arrow) of ubx~ cells, and there the vgQ.lacz construct is expressed; the lacz product is displayed immunocytologically in green. (D) Ubx effect on posterior expression of the wg gene, shown in green. (DI) Wg protein is seen along the D/V boundary of the wing blade region of the wing disc, and the Apterous transcription factor (cf. Fig. 4.6D) in the dorsal domain (purple). (D2) Same for the haltere disc, but wg expression does not extend to the posterior margin (white bracket). Since Wg is required for growth, this may account in part for the smaller size of the haltere (Weatherbee et al., 1998). (D3) Haltere disc with several ubx" clones (dark holes; Ubx in red). One of these clones lies at the posterior D/V boundary, and here ectopic Wg is detected (white arrow). (E) Ubx effects on expression of the DSRF transcription factor (green), which is required for the intervein epidermal fate. (El) DSRF is expressed in the center of the wing blade region of the wing disc, left, between the vein domains (dark stripes), but not in the equivalent region of the haltere disc (right). (E2) Haltere disc bearing a large clone of ubx~ cells in center of pouch region: here DSRF is expressed and the vein domains are formed as in the wing disc (Ubx in red). (E3) Wing disc, with ectopic stripe of Ubx expression (red). The arrow indicates a region of the normal intervein DSRF domain where, in consequence of Ubx expression, DSRF is absent. (F) Summary of Ubx effects on wing patterning system; post, posterior. These effects occur at genes shown in red boxes; solid red box denotes the Ubx effect on the vg® enhancer, assayed directly with a c/s-regulatory construct. [(B-F) From Weatherbee et al. (1998) Genes Dev. 12, 1474-1482.]

Was this article helpful?

0 0

Post a comment