L2 L3 L4 L5

En Dpp Sal/Salr Kni Iro

of the enhancer modules were determined from the phenotypic effects of chromosomal breakpoints, insertions and deletions, and by ris-regulatory analysis using lacz constructs, as illustrated for two examples below. [(A) From Gomez-Skarmeta et al. (1995) Genes Dev. 9, 1869-1882.] (B) Wing imaginal disc stained for Lacz expression (green-blue stain) in an animal bearing a Lacz expressing insertion near the enhancer for the sensory organ precursor cells. Lacz expression reveals already specified SOPs. The disc was also hybridized with an sc probe (purple) to reveal activation of this gene in the proneural clusters. This disc was taken just prior to puparium formation; many of the prospective macrochaete SOP cells have been singled out from the clusters and have begun expressing Lacz. The ps cluster is an example of a proneural cluster which is expressing the proneural genes but has not yet activated the SOP lacz construct. The orientation of the disc is 90° rotated with respect to that in (Al), but most of the proneural clusters in (A2) can be recognized in (B). [(B) From Cubas etal. (1991) Genes Dev. 5, 996-1008.] (C) Control of expression of arucan (ara) and caupolican (cau) genes. These genes encode homeodomain transcription factors which provide spatial inputs into the L3 cis-regulatory module of the AS-C. (CI) Expression of 13—lacz construct from the AS-C (cf. A2; initial pattern); (C2) same but after mutation of binding site for Ara/Caup factors. (C3) Expression of ara gene by in situ hybridization in various regions of the 3rd instar wing disc, including L3. [(CI-C3) From Gomez-Skarmeta et al. (1996) Cell 85, 95-105; copyright Cell Press.] (C4) The L3 domain of ara and caup expression, visualized immunocytologically by use of an L3 ara<aup enhancer trap which expresses Lacz (red), shown together with domains of expression of cubitus interruptus (ci) gene (green). Ci encodes a transcription factor that mediates Hedgehog (Hh) signaling, which occurs throughout the anterior domain of the wing blade (see Fig. 4.6D); here anterior is left, and the black region is the posterior domain of engrailed (en) expression. The white arrows indicate the vertical stripe within which the highest level of ci expression occurs, centered on the Dpp stripe along the A/P boundary. Expression of ara and caup overlaps this stripe above and below the D/V boundary domain, and is bounded on the posterior side by the en expression domain. (C5) Relation between the transcriptional domains that set up the axes of the wing blade, and the L3 sensory organ progenitor field set up by expression of ara and caup: Iro, Iroquois, denotes the products of the ara-caup gene complex, red striped region; Dpp is secreted from the A/ P boundary stripe and diffuses outward (blue striping); Ci domain, yellow; En domain (green striping); the wingless (Wg) signaling molecule is expressed along the D/V boundary (magenta striping). [(C4, C5 From Gomez-Skarmeta and Modolell (1996) Genes Dev. 10, 2935-2945.] (C6) Summary of inputs into the c/s-regulatory system controlling ara and caup expression in AS-C (L3), based on analysis of ara and caup activity in clones expressing or failing to express the ci, dpp, sal/salr, wg, hh, and Hh signal transduction intermediate genes (Gomez-Skarmeta and Modolell, 1996; Biehs et al., 1998; de Celis and Barrio, 2000). Transcription factor inputs are shown as solid arrows; signal system inputs are dashed lines. The actual order of target sites in the L3 c/s-regulatory domain of the Iro-C is not known, and the diagram is oriented A (left) to P, as is the disc. (D) The DC enhancer of the AS-C. (Dl) Expression of a DQ.lacz enhancer construct (cf. A). (D2) Diagram of inputs

The individual pattern formation modules produce very sharply defined subsets of the overall pattern of AS-C expression, just as the diagrams in Fig. 4.7A indicate. The activity of the L3 module (green in Fig. 4.7A1) is demonstrated by expression of a Lacz reporter in Fig. 4.7C1. This enhancer is responsible for AS-C expression in a patch of cells that later separates into the several clusters along vein L3 from which arise the sensilla campaniformia associated with this vein (Fig. 4.7A3). The L3 enhancer also directs expression in a proneural cluster that produces the twin sensilla (TSM) of the wing margin. The key positive input into the L3 module is provided by homeodomain transcription factors encoded by the Iroquois Complex (Iro-C ). This contains three genes, araucan (ara), caupolican (caup), and mirror (Gomez-Skarmeta et al., 1995, 1996; Kehl et al., 1998), of which the first two, which are coregulated, are relevant to the L3 enhancer. If the ara/caup target sites within the L3 enhancer are destroyed, its activity is lost, as illustrated in Fig. 4.7C2 (Gomez-Skarmeta et al., 1996). Iro-C expression (Fig. 4.7C3) includes the L3 domain at the stage shown. But institution of even this relatively simple-looking L3 pattern is in fact quite expensive in terms of regulatory information, as summarized in Fig. 4.7C4-6. Figure 4.7C4 displays the locations of some of the inputs which determine ara and caup expression in the L3 domain (see legend) and these and others are indicated spatially in Fig. 4.7C5 (Gomez-Skarmeta and Modolell, 1996) and more abstractly in Fig. 4.7C6.

into this enhancer. Direct evidence demonstrates activation at Gata sites where the product of the pannier (pnr) gene binds and for a negatively acting complex between the Pnr factor and a cofactor encoded by the u-shaped gene (ush); these two factors determine the dorsolateral boundaries of DC enhancer expression (Cubadda et al., 1997; Haenlin et al., 1997; Garcia-Garcia et al., 1999). There is also indirect genetic evidence for DC enhancer repression by Dpp at the anterior boundary of the DC domain. Wg is necessary for DC enhancer expression but does not determine boundaries of expression. [(D) Adapted from Garcia-Garcia et al. (1999) Development 126, 3523-3532 and The Company of Biologists Ltd.] (E) Transcription control of patterning in specification of longitudinal veins L2-L5 of the wing (see A3 for disposition of these veins). (EI) Schematic representation of a transverse section of wing, veins indicated as magenta circles. The L2 vein field expresses knirps and knirps-related (blue), and L3 and L5 express ara and caup (green), (E2) Summary of spatial domains of gene products listed on left. Vertical orange line represents A/P boundary from which Dpp ligand diffuses; En is expressed only in posterior domain, except for region between L4 and L3. Within each vein progenitor field EGFR is activated in the central and flanking cells and N in adjacent cells, A and P (Sturtevant et al., 1993). For L3, which contains peripheral nervous system elements, the key regulators are those of the Iro-C [as shown in (C); for the controlling inputs see (C5)]. This diagram suggests that there exists a distinct set of tis-regulatory inputs for each vein. The veins form at the borders of spatial transcriptional domains which specify the intervein regions (Bier, 2000; Mohler et al., 2000; Sturtevant et al., 1997). [(E) From de Celis and Barrio (2000) Mech. Dev. 91,31-41; copyright Elsevier Science.]

A very important aspect of the discovery of the role played by the Iro-C in positioning the proneural clusters is that as indicated in Fig. 4.7C, it links AS-C expression with the prior state of spatial specification. Thereby we can now perceive the depth of the genomic regulatory network which underlies wing patterning. Following initial establishment of the imaginal discs (Fig. 4.IE) a new set of interactions sets up the coordinate system of the wing disc, defining as transcriptional domains the wing blade as opposed to the hinge and more proximal (notum) structures (Casares and Mann, 2000); and within the wing blade, the A/P and D/V axes. The four quadrants thus formed within the wing blade, and the orthogonal axial stripes, are also defined as transcriptional domains in this process (see Fig. 4.6D; for reviews that summarize this layer of the patterning network and the signaling interactions by which the coordinates are installed, see Lawrence and Struhl, 1996; Williams and Carroll, 1993; Cohen, 1993; de Celis, 1998; Modolell and Campuzano, 1998). The next layer of the network is that determining expression of ara and caup in the L3 domain considered in Fig. 4.7C3-C6. As summarized in Fig. 4.7C6, the inputs are those generated by the preceding coordinate-making system: the boundaries are set negatively by the en domain (posterior), the wg domain (D/V boundary), and by the spalt (sal)/spalt related (salr) domain (anterior; de Celis and Barrio, 2000). Expression occurs within the A/P boundary domain of dpp gene expression, and also requires high levels of the cubitis interruptus (ci) gene product (expressed in consequence of the Hedgehog signaling set up with respect to the original A/P coordinates of the wing blade; Fig. 4.6D). The next level, after regulation of ara and caup expression, is that displayed in Fig. 4.7C1, i.e., the L3 c/s-regulatory module of the AS-C, which is driven by the Ara/Caup factors. Following this are further levels where the downstream systems that control SOP specification within the proneural clusters operate. There follows the institution of cell type-specific sensory organ differentiation programs, as discussed in the next section of this chapter. The wing disc provides a remarkable illustration of the concept of regulatory network depth.

Furthermore, distinct sets of regulatory transactions take place at given levels in different regions of the disc, as represented by the Iro-C cz's-regulatory inputs and the L3 AS-C enhancer. The operation of the DC enhancer of the AS-C, which is responsible for certain large sensory bristles, or macrochaetes, of the notum (Fig. 4.7A2, A3), is illustrated in Fig. 4.7D1. Spatial inputs which control this enhancer are indicated in Fig. 4.7D2 (see legend for brief description). Entirely different transcription factors are involved than are required for L3 (Garcia-Garcia et al., 1999; Cubadda et al., 1997). These are a Gata factor, the product of the pannier (pnr) gene, and a repressive cofactor of the Pnr protein, the product of the u-shaped{ush) gene (Haenlin et al., 1997).

Transcriptional inputs required for the patterning system that sets up the veins illustrate the same general theme: each vein needs a different set of spatial inputs, which again represent prior pattern information. This is summarized in the diagrams of Fig. 4.7E (de Celis and Barrio, 2000), which indicate some of the transcriptional regulators and signaling systems necessary for the specification of veins 2-5 (cf. Fig. 4.7A3). Vein formation depends directly on a complex set of signaling interactions that define the positions where each vein will arise (for reviews and data Garcia-Bellido and de Celis, 1992, 1998; Biehs et al., 1998; Sturtevant et al, 1997; Mohler et al., 2000; Milán and Cohen, 2000; Bier, 2000; Baonza et al, 2000). Genes encoding transcription factors, viz, the Iro-C genes, knirps/knirps-related, knot and sal/salr, all play key roles, as does the original posterior expression of en. The outcome is a combinatorial specification of each vein position, defined in respect to the pattern of transcriptional domains in the wing disc.

Though knowledge is yet anything but complete, the regulatory mechanisms that drive pattern specification in the wing disc illustrate some very important, and very general points. This is an increasingly beautiful story: we can see how wing patterning is likely to be genetically organized at the cis-regulatory level, and can begin to think about how it evolved. Several of the cardinal features of bilaterian patterning networks are here clearly illustrated. These include the great depth of the network architecture, which layer by layer, controls a progression of transcriptional patterning states; the need for extensive czs-regulatory information processing which allows integration of the multiple spatial inputs presented at each stage; and the importance of c/s-regulatory modularity. This last continues to surprise, as more and more examples are discovered in different systems. But given the complexity of the information processing job required for each pattern element, it is a pleasing outcome to find that each such job is done by a separate piece of regulatory DNA. In addition, though not considered explicitly here, note the direct inputs to the key transcriptional regulatory genes from signal systems; for example the wg and dpp genes are major players in the initial specification of the imaginal discs (Fig. 4.IE); in the organization of the wing blade coordinate system (Fig. 4.6D); and in the specification of both the Iro-C L3 domain (Fig. 4.7C5, C6) and the AS-CDC domain (Fig. 4.7D2).

Patterning the Heart Progenitor Field in Drosophila

The tubular heart of Drosophila forms by fusion of two bilateral, metameric arrays of precursor cells that arise in the dorsal mesoderm of the postgastrular embryo (reviewed by Bodmer and Frasch, 1999). A key transcriptional regulator required for this specification function is Tinman, a homeodomain transcription factor orthologous to Nkx2.5, which as we have seen (Fig. 4.1A, B) is also utilized for specification and subsequent development of the heart progenitor field in vertebrates. However, the tinman (tin) gene in Drosophila is expressed more broadly in the mesoderm than the heart progenitor field, and is required for development of all dorsal mesoderm derivatives, viz. dorsal muscles, visceral mesoderm, and heart, as well as for certain ventral muscles (Azpiazu and Frasch, 1993; Bodmer, 1993). The genomic apparatus controlling tin expression in the postgastrular embryo, the inputs to this apparatus, and its position in the overall network that controls heart formation provide another excellent example with which to illustrate the major themes in this discussion. Several aspects are summarized in Fig. 4.8 (see legend for details).

The tin gene is initially activated in the invaginated mesoderm and its expression thereafter goes through several developmental phases (Fig. 4.8A, C). Its initial transcription is mediated by the mesodermal twist (tun) bHLH factor (stage 9). The tin expression pattern becomes metameric in stage 10, as its transcription is turned down within the domains defined by the evenskipped {eve) stripes, and here the bagpipe regulatory gene is instead activated (Fig. 4.8A). The clusters of cells occupying these regions of bagpipe expression become visceral rather than cardiac mesoderm. In the adjacent stripes where the sloppy-paired (sip) forkhead family gene is expressed, tin transcription is stimulated and these regions give rise to cardiac cells. Expression of sip promotes wg gene expression, which is essential for heart development (Park et al., 1996; Wu et al., 1995); a role of Wg factor secreted from the ectodermal cells could be to stimulate high levels of tun expression in the prospective cardiac cells (Reichmann et al., 1997). As shown in Fig. 4.8B, the originally metameric units of cardiac and visceral mesoderm soon generate continuous columns. After gastrulation expression of tin becomes dependent on Dpp signaling (Fig. 4.8A, stage 9), and on its own product. Later in development tin is expressed under Wg control, in the two rows of cardiac cells from which the heart forms. These successive phases of tin expression are illustrated in Fig. 4.8C1-C3. Still further on, cardiac cell types are specified. A new set of transcription factors now become involved, particularly Mef2, a Mads box factor; and the homeodomain factors Zfhl, Ladybird, and Eve (Lilly et al., 1995; Jagla et al., 1997; Gajewski et al., 1997; Su et al., 1999). Just as in the wing disc patterning process, the heart patterning regulatory network is multilayered. At each stage the spatial information presented in the form of regional transcriptional states is utilized by the czs-regulatory elements then active to generate a finer pattern. Signaling interactions provide essential spatial input in every phase of the process. As indicated in Fig. 4.8A, both Dpp and Wg are required from early on, and at the subsequent stage of cardiac cell type specification both of these same signal systems as well as Notch are utilized (Su etal., 1999).

The stepwise regulation of the tin gene is an essential driver of the heart patterning process. As shown in Fig. 4.8E the tin «s-regulatory system is modular in organization. It is particularly revealing that each of the phases of tin expression is controlled by a specific enhancer element. Four such elements have been found (Yin et al, 1997), and the pattern of expression that each generates in isolation is illustrated in Fig. 4.8D1-D4. These correspond remarkably to the successive patterns shown in Fig. 4.8C. Module D operates at stage 11 (Fig. 4.8D2) and confers control of tin expression to Dpp signaling and to its own product. Thereby it integrates the result of the previous tin expression pattern set up by Module B, the early acting tin czs-regulatory element which activates the tin gene in the mesoderm (Fig. 4.8D1). The extent of mesodermal invagination can be decreased by narrowing the domain of snail expression in the prospective mesoderm just before gastrulation, with the result that contact between the migrating mesoderm and the dorsal ectodermal Dpp domain is restricted (Maggert et al., 1995). In consequence Dpp-driven tin expression is severely depressed, and heart formation is much reduced. Module C functions later, exclusively in cardial cells (Yin et al., 1997). Some inputs into these modules are shown in Fig. 4.8E. Module B has been studied in most detail. Like many of the elements that carry out specification jobs discussed in Chapter 2 it utilizes both positive and negative inputs: Twi turns it on, and Eve, directly or indirectly prevents its expression in the eve stripe domains, while (probably indirectly) the buttonheadibth) gene product represses Module B in the domain of the head where hemocytes form. In Fig. 4.7A we saw a strikingly modular czs-regulatory organization the function of which was to organize expression in the many individual proneural spatial domains; here we see one the function of which is to organize successive phases of expression. What the two strategies have in common is that in both, every individual module has to process different sets of inputs. In order for these alternative transcriptional states to be imposed on a given gene a rigidly effective "traffic control system" must exist, so that in any given cell at any given time only the relevant module communicates with the basal transcription apparatus.

At the stage when differentiation programs are called in, particular czs-regula-tory modules control expression of eve (Su et al., 1999) and of the mef2 gene (Nguyen and Xu, 1998; Cripps et al., 1999)- These genes drive the specification of particular cardiac cell types. Figure 4.8F shows an example of a cardiac cis-regulatory element, from the mef2 gene. This element integrates a spatial input from the preceding stage, i.e., Tin, with a different input, viz. the Gata factor Pnr (Gajewski et al., 1998, 1999). The element uses "and" logic, and together these two factors suffice to activate the mef2 enhancer and produce cardial cell fates even in ectopic locations (Gajewski et al., 1999).

Though they operate in a very different context from the imaginal disc, the mechanisms of transcriptional control in embryonic heart formation reveal organizational similarities. We again encounter modular as-regulatory programming that provides the execution of diverse information processing jobs; depth in the regulatory network, with multiple layers prior to cell type specification; and the ubiquitous use of signaling inputs to provide spatial cues at each layer of the regulatory system.

Encoding Hindbrain Regulatory Patterns

We move now to rhombomeric transcriptional patterning in the hindbrain of the mouse embryo, briefly touched on earlier (Fig. 4.5). Our subject is the genomic apparatus of which the function is to distinguish each rhombomere as a unique domain. This example is of particular value because many of the linkages in the fragments of the control network which are known have been established by direct czs-regulatory analysis.

Embryo Devolopment

Stage 10

Stage 11

Stage 10

Stage 11

Was this article helpful?

0 0

Post a comment