ABSTRACT
The Distal-less gene is known for its role in proximodistal patterning of Drosophila limbs. However, Distal-less has a second critical function during Drosophila limb development, that of distinguishing the antenna from the leg. The antenna-specifying activity of Distal-less is genetically separable from the proximodistal patterning function in that certain Distal-less allelic combinations exhibit antenna-to-leg transformations without proximodistal truncations. Here, we show that Distal-less acts in parallel with homothorax, a previously identified antennal selector gene, to induce antennal differentiation. While mutations in either Distal-less or homothorax cause antenna-to-leg transformations, neither gene is required for the others expression, and both genes are required for antennal expression of spalt. Coexpression of Distal-less and homothorax activates ectopic spalt expression and can induce the formation of ectopic antennae at novel locations in the body, including the head, the legs, the wings and the genital disc derivatives. Ectopic expression of homothorax alone is insufficient to induce antennal differentiation from most limb fields, including that of the wing. Distal-less therefore is required for more than induction of a proximodistal axis upon which homothorax superimposes antennal identity. Based on their genetic and biochemical properties, we propose that Homothorax and Extradenticle may serve as antenna-specific cofactors for Distal-less.
INTRODUCTION
The similar developmental potentials of the Drosophila antenna and leg primordia are evidenced by the number of mutations that result in transformation of one tissue into the other. Genes in which loss-of-function mutations lead to partial antenna-to-leg transformations include homothorax (hth) (Casares and Mann, 1998; Pai et al., 1998), Distal-less (Dll) (Sunkel and Whittle, 1987), and spineless (ss) (Balkaschina, 1929; Struhl, 1982). Complete transformations of the antenna into leg are observed with gain-of-function alleles of the Drosophila trunk Hox gene Antennapedia (Antp) (Gehring, 1966). Transformations of leg into antenna are much less common and have been observed with loss of Antp function (Struhl, 1981). Analysis of the genetic hierarchies among genes required to distinguish the antenna and the leg is likely to provide insights into both limb development and the generation of morphological diversity. Dll and Antp previously were found to repress hth in the developing leg (Casares and Mann, 1998; Gonzalez-Crespo et al., 1998; Abu-Shaar and Mann, 1998), and Dll shown to be required for ss expression in both leg and antenna (Duncan et al., 1998). Here, we analyze the relationship between Dll and hth in the antenna. Drosophila heterozygous for intermediate and strong loss of function of Dll alleles exhibit truncations of the Drosophila antenna and leg, indicating that Dll plays an essential role in forming the proximodistal (PD) axis in both limb types (Sunkel and Whittle, 1987; Cohen and Jurgens, 1989; this work). Ectopic expression of Dll can induce the formation of new PD axes in various positions of the body (Gorfinkiel et al., 1997). These ectopic limbs take on identities appropriate to their anteroposterior position along the main body axis, e.g. antennal elements on the head and leg elements on the wing (Gorfinkiel et al., 1997). This is consistent with the idea that Dll plays a single role during limb development, that of inducing the formation of PD axes upon which selector genes, such as the Hox genes, superimpose information regulating limb identity. However, in Drosophila carrying only one functional copy of the Dll gene or heterozygous for combinations of hypomorphic Dll alleles, the antenna is partially transformed toward leg (Sunkel and Whittle, 1987; Cohen and Jurgens, 1989; this work). The transformation can occur without any concomitant loss of PD information. This indicates that Dll has a second function during limb development, specifying antennal cell fates. How Dll effects the differentiation of distinct limb types is not immediately apparent since Dll is expressed in similar patterns in the presumptive distal and medial cells of both the antenna and the leg (Cohen, 1990; Diaz-Benjumea et al., 1994). One plausible mechanism would be via interaction with other factors whose expression is limb specific.
hth meets three important criteria for a factor that could cooperate with Dll to regulate antennal identity: (1) it is expressed throughout the antennal primordium for much of development, (2) it is required for the differentiation of proximal, medial and distal antennal elements, and (3) it is differentially expressed between the antenna and the leg (Casares and Mann, 1998; Pai et al., 1998). In developing legs, hth expression is restricted during embryogenesis to presumptive proximal cells. hth and Dll expression therefore do not overlap in the leg throughout most of development (Casares and Mann, 1998). hth encodes a TALE-class homeodomain protein required for the nuclear localization of a PBC-class homeodomain protein encoded by extradenticle (exd) (Pai et al., 1998; Rieckhof et al., 1997; Kurant et al., 1998). Exd is a transcriptional cofactor for a variety of homeodomain proteins, including several Hox gene products (reviewed in Mann and Chan, 1996). Genetic studies indicate that both exd and hth are needed for normal development of the entire antenna and of the proximal leg. Loss of either exd or hth function in the developing antenna causes cell-autonomous transformation of medial and distal antennal structures into medial and distal leg structures (Casares and Mann, 1998; Pai et al., 1998; Gonzalez-Crespo and Morata, 1995). Ectopic expression of Meis-1, a murine homolog of hth, can induce the formation of antennae in the genital disc derivatives (Casares and Mann, 1998). However, ectopic expression of either Meis-1 or hth in the developing leg, wing or head does not lead to antennal differentiation (Casares and Mann, 1998; this work), therefore hth is insufficient to induce antennal development in most contexts.
In this study, we demonstrate that Dll and hth are independently regulated and expressed in partially overlapping domains in the developing antenna. Reducing Dll activity or eliminating hth activity transforms the antenna to leg and results in reduction or loss of spalt (sal) expression. Ectopic expression of Dll in domains of endogenous hth expression or coexpression of Dll and hth using the GAL4/UAS system (Brand and Perrimon, 1993) induces expression of sal, and leads to the differentiation of antennal cuticular structures from the eye, leg, wing and anal/genital primordia. Based on these results, we propose that Dll and hth act at the same level of the genetic hierarchy to coordinately activate the antennal developmental program.
MATERIALS AND METHODS
Immunohistochemistry
Antibody and X-gal stainings were carried out as described (Halder et al., 1998). Antibodies used were: chicken anti-Hth (Casares and Mann, 1998), rabbit anti-Hth (Pai et al., 1998), rabbit anti-Dll (Panganiban et al., 1995), mouse anti-Dll (Vachon et al., 1992), and rabbit anti-Sal (Kuhnlein et al., 1994). Secondary antibodies coupled to Cy2, Cy3 and Cy5 were obtained from Jackson ImmunoResearch. Imaging was carried out on BioRad MRC1024 confocal and Zeiss Axiophot microscopes.
Fly strains
The following fly strains were employed: (1) dpp-GAL4 (4A.3)/TM6B (Morimura et al., 1996), (2) act>CD2>GAL4 (= actin promoter-FRT-CD2-FRT-GAL4) (Pignoni and Zipursky, 1997), (3) w; UAS-Dll/ In (2LR) Gla, Gla, Bc, Elp (Konrad Basler), (4) w; UAS-GFP-hth8/TM6B, Tb, Hu (Casares and Mann, 1998), (5) w; FRT82B hthp1 (Pai et al., 1998), (6) w; FRT82B AntpRC3 (Gary Struhl), (7) hth-lacz (= hth05745; Bloomington), and (8) sal-lacz (= salm03602; Bloomington). Stocks constructed by us for these experiments were: (1) Dll1/CyO wg-lacz, (2) Dll3/CyO wg-lacz, (3) Dll7/CyO wg-lacz, (4) DllSA1/ CyO wg-lacz, (5) y, hs-FLPase; FRT82B, 2piM, (6) w; UAS-Dll/In (2LR) Gla, Gla, Bc, Elp; UAS-GFP-hth8/TM6B, Tb, Hu, (7) y, hs-FLPase; UAS-Dll/TM6B, Tb, Hu, (8) y, hs-FLPase; FRT43D, 2piM, (9) w; FRT43D DllSA1, and (10) sal-lacz/CyO; dpp-GAL4/TM6B, Tb, Hu.
Genetic manipulations
Dll hypomorphic larval imaginal discs were generated by crossing heterozygous Dll mutant animals in which each Dll mutant chromosome was balanced over CyO, wg-lacz. Mutant animals were identified by the absence of X-gal staining in the larval tails. Ectopic expression of Dll and hth was induced using the GAL4-UAS binary system (Brand and Perrimon, 1993). dpp-GAL4 was used to activate UAS-hth and/or UAS-Dll along the anteroposterior compartment boundary of the developing imaginal discs. Clones of cells ectopically expressing Dll were generated using a modified GAL4/UAS system (Pignoni and Zipursky, 1997) in which y, hs-FLPase; UAS-Dll/TM6B, Tb, Hu flies were crossed to act>CD2>GAL4 and the resulting larvae heatshocked at 37°C for 10 minutes at 72-96 hours after egg laying (AEL) to induce site-specific recombination between the FRT sites, which in turn results in constitutive GAL4 expression in the clones.
Dll, hth and Antp null clones were generated using the FLP/FRT system (Xu and Rubin, 1993). Animals of the genotypes: (1) y hs-FLPase; FRT43D, 2piM/FRT43D DllSA1, (2) y, hs-FLPase; FRT82B, 2piM/ FRT82B hthP1, and (3) y, hs-FLPase; FRT82B, 2piM/ FRT82B AntpRC3 were heat shocked at 37°C for 1 hour at 48-72 hours AEL and examined in mid- to late-third instar.
RESULTS
Dll is required for antennal identity
Animals heterozygous for Dll null alleles exhibit partial antenna-to-leg transformations (Sunkel and Whittle, 1987; Cohen and Jurgens, 1989), indicating that Dll levels may be important for antennal determination. Weak hypomorphic combinations of Dll alleles also lead to partial transformation of the third antennal segment (a3) and the arista into leg-like structures (Fig. 1A,B). Intermediate hypomorphic combinations of Dll alleles transform the medial antenna toward leg and exhibit distal truncations (Fig. 1C). Strong combinations of Dll alleles exhibit more severe antennal truncations (Fig. 1D). These same allelic combinations result in progressively more severe truncations of the distal leg (Fig. 1E-H). Notably, the antenna-to-leg transformations are not a property of a specific subset of Dll alleles, but are observed with all Dll alleles when assayed in appropriate combinations. For the transformation phenotype to be apparent, Dll PD function must be largely intact. This is likely due to the necessity of having a PD axis for either antennal or leg identity to be manifested. That we observe transformation without limb truncation indicates that the antennal selector function is more sensitive to Dll dosage than its PD function. We emphasize that transformation is a loss-of-function phenotype of Dll, not a neomorphic or hypermorphic one. Together, the Dll phenotypic analysis indicates that Dll is required for antennal identity, as well as for limb outgrowth.
Dll and hth are not required for each others expression in the larval antenna
Both loss-of-function Dll alleles and loss-of-function hth alleles lead to antenna-to-leg transformations. It therefore was possible that Dll might activate hth expression in the antenna or vice versa. To test whether this is the case, we examined expression of each gene in animals mutant for the other. At late third instar, Dll is expressed in presumptive a2, a3 and arista (Fig. 2A,B), while Hth is expressed in presumptive a1, a2 and, weakly, a3 (Fig. 2A,C). Thus the expression of Dll and Hth overlaps in a2 and a3 (Fig. 2A). In Dll hypomorphic antennal discs, both the pattern and level of Hth appear normal (Fig. 3F). In Dll null clones in a2 or a3, Hth levels appear normal (Fig. 2D-F). In hth null clones in presumptive a2 and a3, Dll levels appear normal (Fig. 2G-I). Thus Dll and Hth are not required during larval stages for each others expression in the antenna.
Dll and hth are required together for sal expression in the antenna
The experiments described above establish that Dll and hth act in parallel as antennal selectors. An obvious next question was whether mutations in both genes affect the expression of antennal markers in the same way. To test this, we made use of the antennal marker spalt (sal). sal is one of few genes known to be expressed in the antenna but not in the leg (Wagner-Bernholz et al., 1991). Consistent with the idea that sal lies genetically downstream of both Dll and hth, we found that sal expression is restricted to the presumptive medial cells of the antenna (a2 and a3), precisely where the domains of Dll and Hth expression overlap (Fig. 3A-C). To test whether both Dll and Hth are required for sal expression in the antenna, the expression of sal was examined in Dll hypomorphs and in clones null for either hth or Dll. In Dll hypomorphic combinations exhibiting antenna-to-leg transformations, e.g. Dll3/Dll7 (Fig. 3D,E) and Dll1/Dll3 (not shown), Sal is significantly reduced (Fig. 3E). In Dll null clones, Sal is undetectable (Fig. 3G-I). In hth null clones, Sal is also undetectable (Fig. 3J-L). Together, these results demonstrate that sal expression in the antenna requires both Dll and hth functions.
Coexpression of Dll and Hth in the antenna activates sal
The results described above indicate that Dll and Hth are both necessary for sal expression in the antenna. If Dll and Hth were sufficient to activate sal, we would predict that ectopic expression of Hth in the Dll domain or ectopic expression of Dll in the Hth domain would induce the expression of sal. To test this, dpp-GAL4 (Morimura et al., 1996) was used to drive expression of a UAS-GFP-hth construct (Casares and Mann, 1998) along the anteroposterior compartment boundary of the antennal disc. This resulted in the expression of Hth in a subset of the presumptive distal a3 and aristal cells that normally express Dll, but do not normally express Hth during the third instar (Fig. 4A). Under these conditions, Sal expression is induced (Fig. 4A,B).
To test the consequences of expressing Dll in the Hth domain, a combined ‘flip-out’ GAL4/UAS system was used. In this case, clones of cells expressing Dll were generated in the proximal part of the Hth domain where Dll is normally not expressed (Fig. 4C-F). Only small clones could be recovered (Fig. 4C,D), suggesting that high levels of Dll may be cell lethal. However, cells in these clones frequently express Sal (Fig. 4C,F). We therefore conclude that antennal cells both proximal and distal to the normal Sal domain are competent to express Sal if provided with Dll and Hth, i.e. within the context of the antennal disc, neither Dll nor Hth alone can activate sal, but that together they are sufficient to do so.
Ectopic Dll induces sal expression and antennal differentiation where there is endogenous Hth
To test whether coexpression of Dll and Hth induces antennal differentiation, we examined the phenotypic consequences of ectopically expressing Dll in Hth domains in areas outside of the antenna. As previously reported (Gorfinkiel et al., 1997), we find that ectopic Dll can induce the differentiation of antennal tissue elsewhere in the head (Fig. 5E-G) and ectopic leg tissue on the wing (not shown). However, we also find that ectopic expression of Dll induces the differentiation of antennal structures, primarily arista, in the proximal wing and, less frequently, in the proximal leg (Fig. 6A-D).
The locations where ectopic antennal structures can be induced are correlated with both endogenous Hth expression and ectopic Sal expression. In the eye-antennal imaginal disc, Hth is expressed in the presumptive head capsule and behind the morphogenetic furrow (Fig. 5A,C). Ectopic Dll expression in the Dpp domain of the eye disc creates a region of overlap of Dll with the presumptive head capsule domain of Hth expression (Fig. 5A-C). Ectopic Sal can be detected in the region of overlap, but not in adjacent cells that express Dll but lack endogenous Hth (Fig. 5A,D). Using X-gal stainings of hth-lacz pupae, we observe that the sites where differentiation of antennal structures can be induced by ectopic Dll express endogenous hth (Fig. 6E,F). Together, these results are consistent with Dll being able to induce antennal structures and sal expression only where Hth is present.
Ectopic Dll and Hth together can induce sal expression and antennal differentiation in the leg, head and genital discs
hth expression is restricted to presumptive proximal leg cells and Dll expression is restricted to presumptive distal leg cells during embryogenesis (Casares and Mann, 1998; Goto and Hayashi, 1997). Thus their expression does not overlap for most of leg development. If Dll and Hth cooperate to induce antennal differentiation, then leg tissue may express sal and differentiate antennal structures if provided with Dll and Hth simultaneously. To test this, Dll and Hth were ectopically expressed together using the dpp-GAL4 driver. Under these conditions, Sal expression was induced in the medial to distal leg disc (Fig. 7A,B) and ectopic antennal structures could be detected on the adult legs (Fig. 7C,D).
Dll and Hth expression also do not normally overlap in the eye, the presumptive head capsule, or the genital disc. When dpp-GAL4 is used to coexpress Dll and Hth in these tissues, ectopic Sal expression (Figs 8A-C, 9A,C,E) and the formation of ectopic antennal cuticular structures result (Figs 7C,D, 8D, 9B,D). X-gal staining of UAS-Dll; UAS-GFP-hth/dpp-GAL4 pupae harboring a sal-lacz enhancer trap, results in specific stainings of the ectopic antennal structures (not shown), indicating that sal expression can be correlated directly with antenna formation.
We note that ectopic Dll and Sal are sometimes found where GFP-Hth is very low or not detected (Figs 4A, 5A, 7A, 8A, 9C). This probably reflects our detection methods and not an absence of Hth in these cells. Both Dll and Sal proteins were visualized using antibody reagents that amplified the signals. Hth was visualized by means of the GFP tag in the UAS construct, thus the sensitivity was not as high. The cells in which ectopic Sal can be detected therefore probably contain Hth as well as Dll.
We have compared the frequencies of ectopic antenna formation induced by ectopic coexpression of Dll and Hth in the legs and genitalia with the frequencies of ectopic antenna formation induced by either ectopic Dll or Hth alone. With ectopic Dll alone, we have observed only one recognizable arista on a leg in more than 20 animals, i.e. 120 legs, examined, a frequency of less than 1%. The reason for this low frequency may be due to the fact that there is little overlap of the ectopic Dll with endogenous Hth in the leg disc when UAS-Dll is expressed using the dpp-GAL4 driver. A second possibility is that antennal development is impeded when Antp (or another trunk Hox gene product) is present. As for why ectopic expression of Hth in the Dll domain of the leg does not induce antennal differentiation, when Hth is expressed ectopically there using dpp-GAL4, Dll is repressed (not shown). Thus if both Dll and Hth are required for antennal differentiation, these conditions would not lead to ectopic antennae. Indeed, no recognizable antennal structures were found in the legs of the more than 20 animals examined with ectopic Hth alone (not shown). We also have observed no ectopic aristae in the genital disc derivatives of animals ectopically expressing either Dll or Hth alone. In each case, at least 20 animals were examined. In contrast, with ectopic Dll and Hth together, we found 8 aristae on the legs of 7 animals, or 42 legs, a frequency of 19%. These aristae were distributed among T1, T2 and T3 legs, could be found in both males and females, and were often associated with a3-like tissue. We also found aristae and a3-like tissue on both male and female genital disc derivatives of 3 of these 7 animals, a frequency of 43%. These results support the conclusion that Dll and Hth synergize to initiate the antennal developmental program.
Antp may repress sal indirectly via hth
Antp represses hth, thereby restricting hth expression to the proximal region of the leg (Casares and Mann, 1998). Antp also represses sal in the leg (Wagner-Bernholz et al., 1991). Because antennal sal expression is dependent upon both Dll and Hth, we hypothesized that Antp repression of sal might be mediated via Antp repression of hth, which in turn prevents the overlap of Dll and Hth in the distal leg. Consistent with this possibility, Sal is expressed in Antp null clones in the Dll domain where hth is derepressed (Fig. 10A-D). We therefore think it likely that Antp may be repressing sal expression indirectly by preventing hth from being expressed in the Dll domain of the leg. Since both Dll and Hth are required for antennal differentiation, by preventing the coexpression of Dll and Hth, Antp can preclude antennal development.
DISCUSSION
Dll and hth act in parallel as antennal selectors
Dll is required for the formation of distal elements in all ventral appendages, including the antenna and the leg. hth is required for the formation of proximal elements in the antenna and the leg. For their roles in PD patterning, Dll and hth act independently. However, Dll (Sunkel and Whittle, 1987; Cohen and Jurgens, 1989) and hth (Pai et al., 1998; Rieckhof et al., 1997) also function as selector genes in the antenna. Via both loss- and gain-of-function experiments, we have demonstrated here that they cooperate to regulate antennal differentiation (Fig. 11).
Not only are Dll and hth required in parallel for normal antennal development and antennal sal expression, but coexpression of Dll and Hth activates sal expression and can induce the formation of antennal structures in many different areas in the body, including the head, wings, legs and anal plates. It has been reported that ectopic expression of either Dll or Hth alone can result in the formation of ectopic antennal structures. This includes ectopic expression of Dll in the head capsule (Gorfinkiel et al., 1997) and ectopic Hth or its vertebrate homolog, Meis-1, in the anal plates (Casares and Mann, 1998). Both cases are consistent with the results and model presented here (Fig. 11), in which coexpression of Dll and Hth is required to determine antennal fates. For instance, we find that the ectopic antennal tissue induced by ectopic Dll alone in the head capsule, proximal leg and proximal wing correlates with sites of endogenous hth expression. We also demonstrate that ectopic Dll alone in the presumptive head capsule region of the eye imaginal disc induces sal expression only when Dll overlaps endogenous Hth. Together, these results suggest that ectopic expression of Dll alone can only lead to formation of antennal tissue from cells with endogenous Hth.
If Dll and Hth collaborate to regulate antennal development, then the converse should also be true, i.e. ectopic expression of Hth should induce the differentiation of antennal structures wherever Dll is expressed endogenously. Indeed, in the reported instances of arista induction with ectopic Meis-1 or Hth in the anal plates, the ectopic expression was driven by a Dll-GAL4 line (Casares and Mann, 1998). This produced coincident expression of Dll and Meis-1 or Hth, again consistent with the proposed model (Fig. 11).
However, Hth or Meis-1 alone, when ectopically expressed in Dll domains elsewhere in the body does not lead to ectopic Sal expression (our observations) or to the differentiation of recognizable antennal structures (Casares and Mann, 1998). The explanation for this that we favor is that when Hth is ectopically expressed using the GAL4/UAS system, Dll expression is downregulated in the cells producing Hth. These cells would then have Hth, but insufficient Dll. If both are required for antennal differentiation, antennal differentiation would not be possible. Consistent with this idea, we and others (Gonzalez-Crespo et al., 1998) have observed a decrease in Dll expression in leg cells ectopically expressing Hth.
Hth and arista formation
Coexpression of Dll and Hth in the leg, the wing, and the genital discs using the dpp-GAL4 driver frequently leads to the formation of ectopic aristae. However, while Hth is required cell autonomously for arista development, Hth expression is not detected in the presumptive arista of third instar antennal discs, and ectopic expression of either Hth or Meis-1 in the presumptive arista late in development prevents arista formation (P. D. S. D., unpublished results). Since Dll has functions in PD outgrowth independent of Hth, a plausible explanation is that the ectopic Hth is titrating out the Dll needed for PD outgrowth. If Hth must be lost or reduced to permit arista differentiation, why then does coexpression of Dll and Hth lead to arista development? The explanation that we favor is that it reflects both the ability of Dll to autoregulate and the dynamics of dpp-GAL4 expression. The width of the stripe of dpp expression remains fairly constant as the discs grow (Masucci et al., 1990; Weigmann and Cohen, 1999). As a result, the anteriormost cells that express the dpp-GAL4 driver early, give rise to cells that do not express dpp-GAL4 later in development. Thus some cells in the anterior compartment will express Hth and Dll from the UAS elements only transiently. Because Dll can autoregulate in imaginal discs (Gorfinkiel et al., 1997), transient expression of Dll from the UAS element activates the endogenous Dll gene. By late third instar, portions of the imaginal discs resemble normal presumptive aristal cells in expressing Dll and having expressed, but lost, hth expression. It may be that these cells differentiate as arista.
Could Dll form a functional complex with Hth and Exd in the antenna?
Given that Dll and Hth cooperate to regulate antennal differentiation, it is of interest to elucidate the molecular basis of this synergy. Exd and its vertebrate counterpart, Pbx, are known cofactors for a variety of homeodomain proteins, including Labial, Engrailed and Ultrabithorax (reviewed in Mann and Chan, 1996). Hth is required for retention of Exd in the nucleus (Pai et al., 1998; Rieckhof et al., 1997; Kurant et al., 1998) and may form part of the functional Exd/Hox complex (Rieckhof et al., 1997). Vertebrate homologs of Hth, the Meis and Prep proteins, have been shown to form trimeric complexes with Hox and Pbx proteins (Berthelsen et al., 1998; Swift et al., 1998; Goudet et al., 1999; Shen et al., 1999). Several lines of evidence now support the idea that Exd and Hth are cofactors for the Dll homeodomain protein in the developing Drosophila antenna. These include: (1) the similar antenna-to-leg transformation phenotypes of Dll, hth and exd mutants, (2) the known physical interactions of Exd and Hth with other homeodomain proteins, (3) the fact that Dll and hth function in parallel to regulate antennal development, and (4) the fact that ectopically expressing Hth can mimic loss of Dll function in the antenna. Testing whether Dll, Hth and Exd interact physically and whether such a complex activates antennal enhancers will be important steps toward understanding limb development and tissue-specific gene regulation.
Acknowledgments
We appreciate the comments of Reese Bolinger, Sean Carroll, Karen Downs, Kirsten Guss, Georg Halder and Al Laughon on the manuscript. We are grateful to Jen Scholz for technical support, Sean Carroll for access to his confocal microscope, Richard Mann for the chicken Hth antibody and UAS-GFP-hth Drosophila lines, Konrad Basler for UAS-Dll Drosophila lines, Henry Sun for the FRT-hth Drosophila line and rabbit Hth antibody, Gary Struhl for the FRT-Antp Drosophila line, Reinhard Schuh for the Sal antibody, and the Bloomington Drosophila Stock Center for many of the Drosophila lines used in these experiments. P. D. S. D. was the recipient of a fellowship from the University of Wisconsin Graduate School. J. C. was supported by NIH Predoctoral Training Grant #T32-HD07477. G. P. is the recipient of an award from the Howard Hughes Medical Institute and the University of Wisconsin Medical School.