Report

The Ground State of the Ventral Appendage in Drosophila

See allHide authors and affiliations

Science  24 Aug 2001:
Vol. 293, Issue 5534, pp. 1477-1480
DOI: 10.1126/science.1062542

Abstract

In Drosophila melanogaster, the antennae, legs, genitalia, and analia make up a serially homologous set of ventral appendages that depend on different selector genes for their unique identities. The diversity among these structures implies that there is a common ground state that selector genes modify to generate these different appendage morphologies. Here we show that the ventral appendage that forms in the absence of selector gene activity is leglike but consists of only two segments along its proximo-distal axis: a proximal segment and a distal tarsus. These results raise the possibility that, during evolution, leglike appendages could have developed without selector gene activity.

Selector genes encode transcription factors that specify the identity of segments and appendages in insects and vertebrates (1, 2). The Hox genes are a subset of selector genes that are required for generating morphological differences along the antero-posterior axis of most animals. Studies in the fruit fly,Drosophila melanogaster, demonstrate that altering Hox function can cause one body part to be transformed into another. Perhaps in large part because they govern the development of entire body parts, changes in how Hox genes, and selector genes in general, were used during evolution have led to modifications in animal body plans throughout the animal kingdom (2, 3).

The choice between leg and antenna development in the fruit fly is a classic example of a selector gene function. Eliminating the function of the Hox gene Antennapedia (Antp) causes the second thoracic (T2) leg to develop as antenna (4,5). Thus, one role of Antp in T2 leg development is to repress genes required for antennal development (4). Analogous leg-to-antenna transformations have been observed in the absence of Antp function in other arthropods, suggesting that this Hox function is ancient (6–8). We previously showed that Antp is a repressor of homothorax (hth) in the T2 leg and that hth fulfills the criteria for being an antennal selector gene: hth function is required for antennal development, and ectopic expression of hth in the analia can transform them into antennae [(9); see also (10)]. Selector genes for the analia and genitalia have also been described (11, 12).

Implicit in the idea that selector genes determine the identities of ventral appendages is that there exists an underlying developmental ground state that selector genes act upon. The ground state appendage would therefore be the structure that forms in the absence of selector gene activity (13). What does the ground state appendage look like? What developmental pathways are modified by selector genes to generate different appendage types? And what relation, if any, does the developmental ground state have to the evolutionary predecessor of the arthropod leg?

To begin to address these questions, we studied the morphology of the ventral thoracic appendage that forms in the absence of both Hox andhth function. We generated mosaic animals that had large clones of tissue mutant for both Antp and hth(14). The appendages that formed in the T2 segment were leglike throughout their entire proximo-distal (P-D) axis (Fig. 1C). However, these legs were not wild type. Instead of having five distinct segments along the P-D axis (15) (Fig. 1A), these appendages had only two: a complete tarsus (with five subsegments and claw) and a single proximal segment, Pr* (Fig. 1C). Because the Pr* has features that derive from several proximal leg segments (see below), it likely results from a fusion of the four proximal-most segments of a wild-type leg.

Figure 1

The ground state ventral appendage is a leglike appendage with two segments. (A) A wild-type (WT) T2 leg has five segments from proximal to distal: coxa (co), trocanter (tr), femur (fe), tibia (ti), and tarsus, which is subdivided into tarsal subsegments 1 to 5 (t1 to t5) and a distal claw (c) (15). Five bristle types are indicated: bracted (green arrows), unbracted (red arrows), curved, spurs (sp), and apical (ap). The inset shows a closeup of the proximal femur where both bracted and unbracted bristles are present. The inset comes from a different wild-type leg. (B) A wild-type antenna consists of four segments, from proximal to distal: antennal segments 1 to 3 (a1 to a3) and arista (ar). (C) Antp hth T2 leg. Most of this appendage is mutant (y ). The recovered tarsal segments (t1* to t5*) and single proximal segment (Pr*) are indicated. The inset shows a region of a similar appendage with bracted and unbracted bristles. The asterisk [also in (D)] indicates a proximal plate with unbracted bristles that is typically associated with the ground state. (D) An hth antenna results in an indistinguishable appendage morphology as seen in (C). Most of this appendage is mutant (y ). (E) A high-magnification view of part of the t1* and Pr* segments of an hth y antenna. The same bristle types are observed inAntp hth T2 legs. (F) An hth T1 leg with proximal fusions. Transverse row bristles (arrow), which are indicative of a first leg identity, are observed.

If the two-segment appendage described here represents the ground state, a nearly identical appendage should be generated when selector genes are removed from other ventral appendages. Consistent with this prediction, when hth function is removed from the antenna (where no other identity selector gene is expressed) (13), the resulting appendage is indistinguishable fromAntp hth T2 legs (Fig. 1, C and D) (14). The two-segment appendage also forms in first thoracic (T1) segments that are mutant forAntp, hth, and Sex combs reduced(Scr), the Hox gene normally required for T1 leg identity (5, 14). The similarity between these structures supports the idea that it represents the developmental ground state.

As in wild-type legs, the ground state appendage exhibits polarity along its P-D axis. The bristles and trichomes usually point distally, and distinct bristle types can be observed at different positions along its P-D axis (Fig. 1, C and E) (16). For example, the bristles closest to the body wall usually do not have bracts (small triangular structures at the base of the bristle), whereas the more distal bristles are bracted (Fig. 1, A and C, insets). Distally in Pr* segments, there are usually (>90%) two to three spurs, a thick bristle found in distal tibia (16) (Fig. 1, A, C, and E). A single apical bristle, also in distal tibia of T2 legs, is observed in some (∼10%) Pr* segments (Fig. 1E). Consistent with these morphological landmarks, Distal-less(Dll) and dachshund (dac), which are markers for distal and intermediate positions of the P-D axis in legs (17–19), are expressed normally in the imaginal discs that give rise to these appendages. Thus, the underlying positional information along the P-D axis is intact in the absence of selector gene function.

Our observation that simultaneously removing both Antp andhth from the T2 appendage results in a leglike structure is consistent with the idea that Antp functions primarily by repressing hth. If the only role of Antp is to repress hth, Antp hth legs should be identical tohth legs. We tested this prediction by examining Antp + hth legs. In T2, these legs are indistinguishable fromAntp hth legs (14). In fact, removing hth function from the first (T1) or third (T3) thoracic legs also results in a similar overall morphology: a single proximal segment similar to Pr* and a complete tarsus (Fig. 1F) (14). The only apparent difference between hth first, second, and third legs is that these appendages retain leg type–specific bristle patterns such as transverse row bristles in T1 (Fig. 1F, arrow). This is expected becausehth legs still express Hox genes (in this example, Scr), which specify leg type–specific bristle patterns (5).

These results demonstrate that for ventral appendage development,hth is epistatic to Antp; that is, in the absence of hth, Antp + activity is irrelevant. This conclusion is consistent with the idea that Antp's only function in the ventral appendage is to repress hth. However, two observations suggest that Antp has additional functions. The first is that the coxa of the T2 leg and first antennal segment both express hth but have different morphologies that depend on Antp + (4,20). The second is that Antp hth + T1 or T3 legs occasionally exhibit fusions between the femur and tibia in the absence of hthderepression (4, 5, 19,21). Although these phenotypes are less severe than those observed in hth legs, they suggest thatAntp + contributes to the growth and segmentation of the proximal leg and that these functions are not mediated by repressing hth.

One of the most striking features of the putative ground state appendage is the absence of segmentation in Pr*. In wild-type legs, the joints that form between leg segments require the spatially restricted activation of the Notch pathway (22–24). Thus, the lack of segmentation in the ground state could be due to a difference in Notch signaling. To test this, we examined the expression patterns of Notch, fringe [a modifier of the Notch pathway (25)], and the two Notch ligands, Delta (Dl) and Serrate (Ser), in legs, antennae, and ground state appendages. No differences between the patterns of Notch or fringe expression were detected between these three appendage types. In contrast, the patterns of Dl and Ser expression were very different in legs, antennae, and ground state appendages. In pupal leg discs (when Dl and Ser expression along the P-D axis can be more easily visualized), peak levels of these proteins were detected in rings proximal to the joints that separate each leg segment (Fig. 2, A, D, and G) (19, 22–24). In antennal discs, Ser was detected in a small arc of cells in the first antennal segment (Fig. 2B), and Dl was observed in a broad domain of the P-D axis, including cells that will give rise to both the third and second antennal segments (Fig. 2, E and H) (19). Inhth antennae or Antp hth T2 legs, coexpression of Ser and Dl was observed in the six distal-most tarsal rings (Fig. 2, C, E, F, and I) (19). In Pr*, Ser and Dl were detected at low levels but did not show a consistent pattern. Thus, the lack of segmentation in Pr* correlates with a lack of Dl and Ser expression in rings.

Figure 2

Dl and Ser are expressed differentially in leg, antenna, and ground state. (A) Wild-type pupal leg disc stained for Ser (red), Hth (blue), and Fj (yellow). Tarsal segments are numbered and the ti and fe are indicated. (B) Wild-type pupal antenna stained for Ser (red) and Hth (blue). Ser is only detected in a small region of a1 (arrow). (C)Antp hth T2 leg (ground state) stained for Ser (red) and Dl (green). Absence of CD2 (purple) staining marks mutant tissue. Tarsal segments and Pr* are indicated. (D) Wild-type pupal leg stained for Ser (red) and Dl (green). Peak levels of Ser and Dl are seen in rings proximal to the joints that separate the leg segments: bw (body wall), co, tr, fe, ti, and tarsal segments t1 to t5. (E) A third instar antennal disc with a hth clone (outlined with white line, marked by the absence of CD2 staining) stained for Dl (green; single channel in inset) and CD2 (purple). (F) hth pupal antenna stained for Ser (red), Fj (yellow), and CD2 (purple). Tarsal segments and Pr* are indicated. (G to I) Schematic drawings of pupal leg (G), antenna (H), and ground state (I) summarizing the expression of Hth (blue outline), Ser, Dl, and Fj. The Pr* (I), where we observe weak expression of multiple markers, is indicated in gray.

We also examined the expression of two targets of Notch signaling in the leg: four-jointed (fj) andoddskipped (odd) (23, 26,27). fj is expressed in three rings along the P-D axis of legs but is only weakly expressed in the antenna (Fig. 2, A, G, and H) (19). In the ground state, fj is derepressed in most cells in the Pr* but was not detected in the tarsus (Fig. 2, F and I). In wild-type leg discs, odd is expressed in six concentric rings along the P-D axis (Fig. 3, A and B) (19). In wild-type antennal discs, odd is expressed in only two rings, consistent with the fewer number of joints in wild-type antennae (Fig. 3, D and E). In the ground state, the distal-most odd ring is recovered, and, proximally, most cells weakly express this gene (Fig. 3, C and F). These data support the idea that the activity of the Notch pathway is different among these appendages.

Figure 3

The Notch target odd is expressed differently in leg, antenna, and ground state imaginal discs. (A and B) Wild-type leg discs stained for odd-lacZ (red); (A) is also stained for Hth (blue). Six odd rings are numbered from proximal to distal, 1 to 6, and the P-D axis is indicated. Rings 1 to 3 overlap with Hth. (C) A T2 leg disc with an Scr Antp hth clone, marked by the absence of CD2 staining, stained for odd-lacZ (red). Additional odd-Z staining can be seen between rings 1 and 2 because of proximal fusions (arrow). Rings 5 and 6 are observed. (D and E) Wild-type antenna discs stained for odd-lacZ (red); (D) is also stained for Hth (blue). Two odd rings are numbered from proximal to distal and the P-D axis is indicated. (F) Everting hth antennal disc stained for odd-lacZ (absence of CD2 marks the mutant tissue; most of this appendage is mutant). The sixth odd ring is recovered (arrow) in the ground state.

In addition to defining the developmental ground state for the ventral appendage, these results clarify the roles of hth andAntp in leg development. The main points can be summarized as follows (Fig. 4): (i) An apparently normal tarsus can develop in the absence of all Hox and hthfunction. (ii) In the distal and medial leg, Antp's only function is to repress hth. (iii) In the proximal leg,hth is required for growth and segmentation in all three thoracic legs. (iv) Antp also contributes to proximal leg development where it does not function by repressing hth. (v) Selector genes modulate Notch signaling in the ventral appendage by altering the expression of the Notch ligands, Dl and Ser. As Notch signaling is required for both joint formation and growth of the leg, controlling Notch ligand expression provides a plausible mechanism for varying the size and number of segments among the different ventral appendages.

Figure 4

The relationship between the ground state, the antenna, and the T2 leg. The ground state appendage has a single Pr* segment (light blue box) and five tarsal subsegments (smaller blue boxes). Addition of hth generates antennae, which have three segments and the arista (green boxes). Addition of Antp to the antenna results in a T2 leg, which have five tarsal subsegments and four proximal segments. The tarsus of the leg is similar to the tarsus of the ground state. In the medial leg (darker blue boxes),Antp is required to repress hth. In the proximal leg (blue-green boxes), Antp and hth are both required for its proper specification. Dl and Ser expression domains are indicated by orange bars proximal to each leg joint. Dl expression in the antenna is shown by the darker green color. The delineation between the appendage (ap), proper, and the body wall (bw) is based on the proximal limit of the Dac and Dll expression patterns and is indicated by the dashed lines.

Finally, these results raise the possibility that there may be a relationship between the developmental ground state appendage defined here and the ancestral leg (evolutionary ground state) that predates the arthropods. Although all arthropods have highly segmented legs, it is likely that the predecessor to the arthropods had simpler, unsegmented legs (28, 29). Because the developmental ground state appendage is a simple leglike appendage, it is plausible that at some time during evolution, legs developed without Hox or hth inputs. If nonarthropod phyla are surveyed for Hox and hth expression patterns, it may be possible to find an extant example of such an appendage.

  • * Present address: Instituto de Biologia Molecular e Celular, Rua do Campo Alegre 823, Porto-4150, Portugal.

  • To whom correspondence should be addressed. E-mail: rsm10{at}columbia.edu

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article