Initiation of the proximodistal axis in insect legs

Transcription

1 Development 121, (1995) Printed in Great Britain The Company of Biologists Limited 1995 Review article 619 Initiation of the proximodistal axis in insect legs Gerard Campbell and Andrew Tomlinson Center for Neurobiology and Behavior, and Department of Genetics and Development, Room 1120, 701 West 168th Street, New York, NY 10032, USA SUMMARY Much of the cell-cell communication that controls assignment of cell fates during animal development appears to be mediated by extracellular signaling molecules. The formation of the proximodistal (P/D) axis of the legs of flies is controlled by at least two such molecules, a Wnt and a TGFβ, encoded by the wingless (wg) and decapentaplegic (dpp) genes, respectively. The P/D axis appears to be initiated from the site where cells expressing wg are in close association with those expressing dpp. Support for this hypothesis comes from two sources: classical grafting experiments in cockroaches and ectopic protein expression in Drosophila. Key words: axis formation, insect leg, cell-cell communication, pattern formation, cell signaling INTRODUCTION Cell-cell communication is a primary mechanism by which spatial pattern is generated in developing animals. The famous grafting experiments of Spemann and Mangold (1924) with the amphibian embryo showed that an ectopically positioned dorsal blastopore lip could direct the fates of surrounding cells and organize them into a supernumerary gastrulation resulting in a secondary body axis. These and many other experiments demonstrated the activity of cell signaling but gave little indication of its biochemical basis. More recent molecular studies have shown the ability of extracellular polypeptides secreted by one group of cells to direct the fate of others. Several classes of these secreted signaling molecules have been identifie d including the Wnt and TGFβ families of proteins (reviewed in Nusse and Varmus (1992) and Kingsley (1994)). In this review, we will discuss the formation of the proximodistal (P/D) axis in insect legs highlighting experiments that have demonstrated how the cells of the presumptive leg communicate with each other to establish this axis. The major indications of how the P/D axis is established come from experiments that result in the formation of supernumerary legs (secondary P/D axes). These experiments fall into two classes: surgical grafting experiments performed predominantly on cockroaches, and ectopic protein expression induced in D r o s o p h i l a (Fig. 1). We will show how the results of these experiments are consistent with a model (Campbell et al., 1993) whereby the P/D axis is organized from a point in the developing leg where cells expressing one secreted signaling molecule, a Wnt encoded by the w i n g l e s s (w g) gene, abut those expressing another, a TGFβ encoded by the d e c a p e n t a p l e g i c (d p p) gene. THE INSECT LEG AND THE IMAGINAL DISCS OF DROSOPHILA The insect leg consists mainly of a shell of integument that is composed of an outer cuticle, which forms the exoskeleton, and an underlying epidermis, which is responsible for secreting the cuticle. The epidermis is a simple monolayer bounded apically by the cuticle and basally by a basement membrane (Wigglesworth, 1959). Unlike the vertebrate appendage where mesoderm/ectoderm interactions are a crucial part of limb patterning (Bryant et al., 1987), in the insect limb there is little or no input from mesoderm so that patterning occurs two dimensionally, confined to the plane of the epithelium (French et al., 1976). Pattern variations within the leg are manifest in the cuticular secretions of the epidermis and by structures such as bristles associated with sensory neurons also of epidermal origin (Locke, 1984). The regional specialisation of the leg epidermal cells is evident along the three major body axes present within the limbs. There are the anteroposterior (A/P) and dorsoventral (D/V) axes of the body proper and the proximodistal (P/D) axis specific to the appendages. The P/D axis relates to the distance from the body trunk (the proximal extreme of an arm is the shoulder, the distal extreme is the fingers), while the A/P and D/V axes unite to form the single circumferential axis (Fig. 2C). Thus, with regard to pattern formation, the leg can be viewed as a hollow cone of epidermal cells (Fig. 2E), with the circumferential axis representing the pattern elements around the circular cross-section of the cone and the P/D axis representing those from tip to base. Hemimetabolous insects such as cockroaches develop legs during embryogenesis so that both the larval and adult stages possess functional legs. The larval stages of many holometabolous insects such as flies, however, are without legs; the legs of the adult fly grow and become patterned inside the developing larva in ectodermal invaginations known as imaginal discs (Bryant, 1978). The leg disc is essentially a circular, flattened, monolayer, columnar epithelium with the centre of the disc corresponding to the presumptive distal tip of the leg and the periphery to its base (Fig. 2; Schubiger, 1968;

2 620 G. Campbell and A. Tomlinson Fig. 1. The growth of supernumerary legs (duplications of the P/D axis) in insects can be induced by grafting experiments and by ectopic expression of secreted signaling molecules. (A) Cockroach leg showing two supernumeraries (s) that have developed following a grafting operation explained in detail in Fig. 3 (the tip of the graft (g) is hidden behind one of the supernumeraries). h, host (courtesy of Peter Shelton and Paul Truby). (B) Drosophila leg showing a duplication (s) induced by a small clone of cells constitutively expressing wg in the dorsal region of the leg. (C) Drosophila leg showing a duplication (s) induced by two small clones of cells constitutively expressing hh in the anterior region of the leg. The supernumerary contains only anterior compartment tissue. Abbreviations: g, graft; h, host; s, supernumerary leg. Bryant, 1978). Accordingly, the radius of the disc corresponds to the P/D axis of the leg (Fig. 2). The P/D axis is fully established by the end of the third, last, larval instar, but the elongated form of the leg is not manifest until later during the prepupal period (Schubiger, 1968). At this time, disc eversion telescopes out the three-dimensional leg from the flattened disc. For geometrical simplicity, we view this as the morphogenetic elaboration of a hollow three-dimensional cone from a flattened disc (Fig. 2D,E). GRAFTING EXPERIMENTS IN COCKROACHES Larval cockroaches, in common with several other insects, possess the ability to regenerate their legs (Bodenstein, 1933). Amputation results in the formation of a blastema, a proliferative mass of dedifferentiated cells derived from the regions surrounding the wound, that regenerates the missing part (Truby, 1983). This ability to generate new P/D axes during larval life is demonstrated more dramatically following several grafting experiments that result in the development of supernumerary legs (reviewed in French et al., 1976). The manipulations that we will describe involve the amputation of a leg which is then grafted onto the stump of the contralateral counterpart (at the same P/D level) with or without a rotation of 180 (Bodenstein, 1937; Bohn, 1965; Bulliere, 1970). When the distal part of a leg is grafted onto the corresponding stump of its contralateral counterpart (without rotating the graft) then the anterior/posterior (A/P) regions remain aligned, but the dorsal part of the graft juxtaposes the ventral part of the stump and the dorsal stump tissue now lies face to face with the ventral region of the graft (Fig. 3Aiii, upper level). At these positions of maximal circumferential misalignment (the two points where the dorsal and ventral tissues are juxtaposed) supernumerary legs develop (Fig. 3Aiv). These supernumeraries are not complete appendages, but are duplications of only those leg structures distal to the site of the graft. The resulting structure is a single proximal stump bearing three distal parts (Fig. 1A) and each of the supernumeraries has a characteristic chirality (left or right) and orientation. (Fig. 3Aiv). Similarly, if the graft is rotated through 180 before joining, then the A/P circumferential values are misaligned and D/V are in register (Fig. 3B, upper level). Here again appendage duplications arise at the positions of maximal circumferential misalignment and again they have a characteristic chirality and orientation (Fig. 3Bv). Misalignment of both A/P and D/V by amputation and 180 rotation back onto the same stump can produce supernumerary appendages but grafts tend to rotate back to their original position (Bohn, 1965; Bulliere, 1970) making interpretation difficult and so we will not discuss these particular operations. The clear message from these grafting operations was that juxtaposing cells from different regions around the circumference of the leg (dorsal with ventral or anterior with posterior) can result in the generation of supernumerary P/D axes. The results of these experiments were used to formulate theoretical models that attempt to explain how the P/D axis may be initiated. We will describe two of these, the polar coordinate model and the boundary model. THEORETICAL MODELS OF P/D AXIS INITIATION The polar coordinate model of French, Bryant and Bryant (1976) suggested that positional information is specified in terms of polar coordinates so that a cell has a positional value along the P/D axis and another around the circumference (Fig.

3 Initiation of the proximodistal axis in insect legs 621 4). Around the circumference, positional value is a continuum with no points of discontinuity. This is a two-dimensional model where the three-dimensional cone of the leg is transformed into a two-dimensional circle (Fig. 4; as described above, such a transformation is justified in terms of the insect leg). The distal tip is the centre of the circle and the P/D axis corresponds to the radius. The model has two rules. Firstly, the rule of shortest intercalation states that the confrontation of cells with different positional values would result in the intercalation of intermediate values via the shortest route. Secondly, the full circle rule stated that leg outgrowth and ultimately a P/D axis would be generated at the site where there was a full complement of circumferential positional values. This second rule was later modified to take into account the observations that a full complement of circumferential values does not appear to be required to get some outgrowth, but even the modified version requires a full or almost full complement of circumferential values to generate a complete P/D axis (Bryant et al., 1981). This model can account for the formation of supernumerary legs following the grafting experiments because a full circumference of positional values will be intercalated at the two positions of maximal circumferential misalignment (French et al., 1976). The second model is the boundary model of Meinhardt (1982) and is quite different from the polar coordinate model. It was based in part on lineage studies in Drosophila where it was shown that the leg is divided into two lineage units the anterior and posterior compartments and that the anterior compartment may be divided into a dorsal and a ventral subcompartment (Steiner, 1976; Lawrence and Morata, 1977). Accordingly, it was proposed that the leg is divided into three circumferential cell populations or compartments the posterior, dorsal/anterior and ventral/anterior (Fig. 4) Further, it was suggested that the site where the boundaries between these cell populations intersect defines the presumptive distal tip of the leg and is the source of a diffusible morphogen that induces outgrowth and specifies cell fate along the P/D axis. This model can explain the supernumerary legs because the grafting operations create two new sites where the boundaries between the three cell compartments will intersect. Although molecular studies in Drosophila have been used to support the polar coordinate model (Couso et al., 1993; Bryant, 1993; French and Daniels, 1994), we suggest that the recent findings particularly from ectopic expression of signaling factors in Drosophila lend greater support to the boundary model. We will describe the relevant molecular biology of cell signaling in the Drosophila leg and show how it can explain the grafting experiments described above. Fig. 2. The leg imaginal disc of Drosophila. (A) Diagram of a late third instar leg disc. (B) Cross section through this disc (running dorsal to ventral). (C) Diagram of an adult leg. The leg imaginal disc consists of a flattened monolayer of epidermis that is thrown into concentric circular folds. The centre of the disc corresponds to the presumptive distal tip of the leg and the periphery to the base of the leg and body wall so that the radius of the disc corresponds to the proximodistal axis of the leg. During the prepupal period, the disc everts to form the elongated morphology of the leg. Because the epidermis is a monolayer, pattern is specified in only two dimensions along the P/D axis and the circumferential axis (the latter corresponds to the combined dorsoventral and anteroposterior axes). (D) The third instar disc is represented as a 2-D disc showing the expression pattern of several genes. hedgehog (hh) and engrailed (en, green) are expressed in the posterior compartment. Cells immediately anterior to the compartment border express decapentaplegic (dpp, blue) in the dorsal half and wingless (wg, orange) in the ventral half. The cells in the centre, the presumptive distal tip, express aristaless (al, black). (E) The adult leg represented as a cone showing how the essentially 2-D circular sheet of epidermis in the disc in (D) telescopes out into the 3-D hollow cone of the leg. Abbreviations: A, anterior; D, dorsal; P, posterior; P/D, proximodistal; V, ventral.

4 622 G. Campbell and A. Tomlinson SECRETED FACTORS IDENTIFIED IN INSECT APPENDAGES The leg disc is divided into two lineage compartments (Steiner, 1976; Lawrence and Morata, 1977) which abut along an extended interface. The cells of the smaller posterior compartment are defined by their expression of the Engrailed homeodomain protein (Fig. 5A; Kornberg et al., 1985) and secrete Hedgehog (Hh), a signaling peptide (Lee et al., 1992; Mohler and Vani, 1992; Tabata et al., 1992). Dorsal cells in the anterior compartment close to and abutting the posterior compartment express d p p (Figs 2D, 5B,D; Masucci et al., 1990; Raftery et al., 1991; Padgett et al., 1987), a member of the TGFβ g r o u p of secreted growth factors, and in the corresponding ventral position cells express w g, a member of the Wnt family of secreted factors (Figs 2D, 5C,D; Struhl and Basler, 1993; Baker, 1988a; Couso et al., 1992; Rijsewijk et al., 1987). It is important to note that there is only a single site where cells expressing

5 Initiation of the proximodistal axis in insect legs 623 Fig. 3. Diagrammatic representation of the transplantation studies performed on cockroach legs and the cell-cell interactions and responses inferred from molecular studies in D r o s o p h i l a. (A) The upper series represents the cockroach leg as a 3-D cone-shaped epithelium (see Fig. 2E), showing the transplantation of a right stump to a left graft inducing two supernumerary appendages at the positions shown. The lower series represents the D r o s o p h i l a imaginal leg as a 2-D disc. The radius of the disc corresponds to the P/D axis. The lower disc series is coloured to display the relevant expression patterns and the inferred molecular consequences of the transplantations. The leg cones are also coloured to display how the molecular expressions shown in the disc correspond on the 3-D structure. The molecular expression pattern colours are: green is e n g r a i l e d (e n), orange is w i n g l e s s (w g), blue is d e c a p e n t a p l e g i c (d p p) and the black circle is a r i s t a l e s s (a l). i and i i represent right and left legs, respectively. Both legs are cut at the same P/D level which is represented by the inner circle in the discs. i i i, The right distal graft is transplanted onto the left stump, which keeps the A/P axis in register but misaligns D/V. w g-expressing cells now juxtapose d p p- e x p r e s s i n g cells at the two positions where the supernumerary appendages grow out in i v. Note the two supernumeraries are of the left chiral form which corresponds with the spatial arrangement of the anterior, posterior, dorsal and ventral domains brought together by the transplantation. (B) Details are as given in A, except the viewer s perspective is rotated by 90. When the right graft is rotated through 180 before transplantation, the D/V axis is now held in register but A/P is misaligned. i i i, e n-expressing cells now confront non-e n- expressing cells (white) at two new interfaces. e n-expressing cells express h e d g e h o g (h h) a secreted factor represented by the arrows. The anterior cells (white) respond differently to this signal in the dorsal and ventral regions and the horizontal line through the anterior cells represents this demarcation. i v, h h signaling elicits d p p expression along the new A/P interface in the dorsal anterior and w g expression is induced in the corresponding ventral anterior position. w g-expressing cells now juxtapose d p p-expressing cells at the two positions where the supernumerary appendages grow out in v. N o t e again the spatial arrangement of the anterior, posterior, dorsal and ventral domains brought about by the transplantation corresponds to the chiral form of supernumerary appendage. high levels of d p p* come into close proximity with those expressing w g and this is in the centre of the disc, the presumptive distal tip, where the homeobox gene a r i s t a l e s s (a l) is expressed (Fig. 5D; Campbell et al., 1993). Genetic analyses have shown that w g and d p p are required to establish the P/D axis (Sharma and Chopra, 1976; Baker, 1988b; Spencer et al., 1982). Further to this Gelbart (1989) suggested a direct role for d p p in demarcating the centre of the disc as a focus (organizing centre) of the P/D axis. However, the real clues as to how w g and d p p might be controlling initiation of the P/D axis have come from misexpression of w g and d p p, either directly in the case of w g or indirectly by ectopic expression of h h. MISEXPRESSION OF WG AND THE INTERACTION BETWEEN DORSAL AND VENTRAL CELLS wg is normally expressed only in ventral regions but it has been *dpp is also expressed at lower levels in the ventral region of the leg disc (Fig. 5B). This region of expression appears to have no influence on the establishment of the P/D axis: for example, in the ectopic wg expression experiments (Fig. 5E,I) supernumerary legs and ectopic al expression are only induced in the dorsal regions where there are high levels of dpp expression. Therefore, in this review on the P/D axis, where we refer to dpp expression, this is the high levels of dpp expression characteristic of the dorsal regions of the disc. misexpressed in lateral and dorsal regions by the flip technique developed by Struhl and Basler (1993), who showed that ectopic expression of wg in small clones could induce the development of supernumerary legs. These secondary P/D axes are induced only when the ectopic patches of wg-expressing cells are in close proximity to the endogenous dpp-expressing cells in the dorsal region of the leg disc, and al expression is transcriptionally activated at this position (Figs 5E, 6A; Campbell et al., 1993). Ubiquitous wg expression changes the shape of the leg from a disc to an oval with al expression (marking the tip) now extending as a line along the dorsally positioned dpp stripe (Figs 5I, 6C). These results suggested that a P/D axis is initiated at the site where dpp-expressing cells come into close association with those expressing wg (Fig. 6A; Campbell et al., 1993). In addition, we proposed that this site, termed the distal organizer (because it corresponds to the presumptive distal tip of the leg), has properties analogous to the organizer of vertebrate embryology and controls outgrowth and specification of cell fate along the P/D axis (Fig. 7). The two cogent pieces of evidence indicating the presence of the distal organizer are; first, as described above with small ectopic wg patches, the establishment of supernumerary sites triggers supernumerary appendages, and second, (also described above) changing the shape of the site, as achieved by ubiquitous expression of wg, changes the shape of the leg (Fig. 6). On the assumption that insects use common patterning mechanisms in their appendages, the molecular expression domains and the cellular interactions that trigger the distal organizer in Drosophila can now be used to explain the supernumerary appendages resulting from the grafting experiments in cockroaches. Grafting without rotation (Fig. 3A), brings ventral wg-expressing cells into direct contact with dorsal dppexpressing cells at the sites where the two new appendages grow out. As predicted from the Drosophila studies the ectopic association of wg- and dpp-expressing cells triggers new P/D axes (Fig. 3A). MISEXPRESSION OF HH AND THE INTERACTION BETWEEN ANTERIOR AND POSTERIOR CELLS The postulated wg/dpp interaction, however, fails to explain the results from the second class of operation (Fig. 3B) where anterior and posterior tissues are inappropriately juxtaposed but the dorsal (dpp-expressing) and ventral (wg-expressing) cells are correctly aligned. However, recent work of Basler and Struhl (1994) provides cogent support for this model by demonstrating the regulative molecular responses that occur when anterior and posterior tissues are juxtaposed. This study investigated the role of Hedgehog which is normally expressed only in the posterior compartment (en-expressing) cells and we highlight three salient results. First, ectopic expression of hh in small clonal patches can induce duplications of the P/D axis (Fig. 1C). These duplications contain only anterior compartment cells, providing further evidence that a full complement of circumferential positional values is not required for the formation of a normal P/D axis. Second, these duplications appear to be induced only when clonal patches ectopically expressing hh are present in both dorsal/anterior and ventral/anterior regions. Third, ectopic hh induces ectopic

6 624 G. Campbell and A. Tomlinson expression of dpp in the dorsal half of the anterior compartment and ectopic wg expression in the ventral half. This observation divides the anterior cells into two cell populations on the basis of their response to hh, a division predicted in the boundary model. The latter two observations and the results from ectopic wg Fig. 4. Theoretical models proposed to explain how the P/D axis is initiated. According to the polar coordinate model, positional information is specified in terms of polar coordinates so that a cell has a positional value along the P/D axis and around the circumference. The original version proposed that leg outgrowth and ultimately a P/D axis is generated wherever there is a full complement of circumferential positional values. The supernumerary legs generated following grafting experiments are explained because these operations will result in the intercalation of two new full circles of circumferential values. According to the boundary model, the leg is divided into three cell populations around the circumference: posterior (P), anterior/dorsal (A/P), and anterior/ventral (A/V). It was proposed that a P/D axis is generated wherever the boundaries between these cell populations intersect and that this site is the source of a morphogen that controls outgrowth and specification of cell fate along the P/D axis. The supernumerary legs generated following grafting experiments are explained because two new sites are created where the boundaries between these cell populations intersect. Fig. 5. Leg imaginal discs from D r o s o p h i l a, stained for e n, d p p, w g or a l expression. All are whole mounts of late third instar discs, anterior to the left, dorsal at the top. (A-D) Normal leg discs showing the characteristic concentric rings of folding, one fold probably marks the same P/D level around the circumference. (A) e n expression in brown (enhancer trap) and a l expression in blue (antibody). The leg is divided into anterior and posterior lineage compartments. The posterior compartment is characterized by the expression of the e n homeobox gene; these cells also express h h. a l is expressed in the centre of the disc, the presumptive distal tip of the leg and also in a more proximal location in the dorsal region. (B) d p p expression in brown (reporter gene) and a l expression in blue (antibody). d p p is expressed at high levels in the dorsal region of the disc just anterior to the compartment border and at much lower levels in the ventral region. (C) w g expression in brown (enhancer trap) and a l in blue (antibody). w g is expressed in a ventral sector anterior to the compartment border. (D) d p p expression in blue (antibody) and w g expression in brown (enhancer trap). The domains of ventral w g expression and high d p p expression in the dorsal region (the d p p antibody does not appear to be sensitive enough to detect the low level expression in the ventral region) do not overlap, but meet in the centre of the disc, the presumptive distal tip of the leg; a l is expressed at this site. (E-H) Leg discs containing a supernumerary induced by small clones of cells ectopically expressing w g in E and h h in F-H. The supernumerary (s) is obvious from the change in the folding pattern of the disc (instead of only concentric circles there are adjacent circles) and in E, F and G by ectopic a l expression (blue) at the presumptive tip. (E) Leg disc containing a supernumerary (s) induced by a small clone of cells constitutively expressing w g (the clone can not be identified) stained for expression of d p p (brown) and a l (blue). Both legs have a l expression in the centre. The supernumerary lies on the dorsal stripe of d p p expression. (F-H) Leg discs containing a supernumerary (s) in the anterior compartment induced by small clones of cells constitutively expressing h h. The clones cannot be i d e n t i fied directly but their presence is indicated by ectopic w g and ectopic d p p expression (see G and H). (F) Stained for expression of d p p (brown) and a l (blue). The ectopic d p p expression in the dorsal region associated with the duplication (s) is arrowed. (G) Stained for w g ( b r o w n ) and a l (blue). The ectopic w g associated with the supernumerary (s) is arrowed. (H) Stained for d p p (blue, antibody) and w g (brown). The supernumeraries are associated with ectopic w g and ectopic d p p expression, which, as in the normal leg, meet in the centre of the leg, the presumptive distal tip. (I-K) Ubiquitous expression of w g or h h transforms the presumptive leg from a disc into an oval. (I) Leg disc from third instar larva in which ubiquitous w g expression was induced when it was a first instar. Stained for d p p (brown) and a l (blue). The morphology of the disc is modified into an oval and the central spot of a l expression has extended as a line along the dorsal d p p stripe (there is no ectopic a l expression in the ventral region suggesting that the ventral d p p expression is too low to function in P/D axis initiation). (J,K) Leg discs from third instar larva in which ubiquitous h h expression was induced when it was a first instar. Stained for a l (blue) and in G, d p p (brown) and in H, w g (brown). The morphology of the disc is modified into an oval and the central spot of a l expression has extended as a line. This line marks the division of the anterior compartment into dorsal and ventral regions. In the dorsal region ubiquitous h h expression induces ubiquitous d p p expression; in the ventral region ubiquitous h h expression induces ubiquitous w g expression. (The morphology of the leg discs in G and H is different from those in Basler and Struhl, 1994, Fig. 4. because in that study ubiquitous h h expression was induced at a later stage. Ectopic h h h a s no influence on the P/D axis in mid-late third instars). METHODS. The ectopic w g discs are from Campbell et al. (1993). The ectopic h h discs are previously unpublished. Ectopic h h expression was performed as in Struhl and Basler (1994). Antibody staining is described in Campbell et al. (1993) along with antibodies used, except the dpp antibody which was a kind gift of Mike Hoffman (unpublished). For F-H over 40 imaginal disc containing supernumeraries were examined in this study and all the supernumeraries were associated with ectopic d p p or w g expression at the presumptive tip.

7 Initiation of the proximodistal axis in insect legs 625 expression described previously led Basler and Struhl, (1994) to suggest that the leg duplications induced by ectopic hh are the result of juxtaposing cells ectopically expressing wg with those ectopically expressing dpp (Fig. 6B). We have tested this by examining dpp, wg and al expression in discs containing clones of cells ectopically expressing hh, or discs subject to prolonged ubiquitous hh expression (Fig. 5F,G,H,J,K). As expected, supernumerary legs induced by ectopic hh are always associated with clones of cells ectopically expressing wg in the ventral region and dpp in the dorsal region (Fig. 5F- H). Ubiquitous expression of hh throughout most of larval life results in overgrowth of leg anterior compartments that stretch the typical circular disc into an oval (Fig. 5J,K). By appearance, wg is expressed in all cells of the ventral half of the enlarged anterior compartment of these leg (Fig. 5K) and dpp appears in all cells of the dorsal half (Fig. 5J). These two domains meet along an extended front running centrally through the anterior compartment and al is expressed along the length of this interface (Fig. 5J,K). al marks the presumptive tip of the growing appendage and thus, instead of the normal spot of al expression positioned centrally in the circular disc, there is now a line of al running through the oval. Again, as with ubiquitous wg expression described above, changing the site of the wg/dpp interface from a point to a line changes the shape of the leg from a disc to an oval (Fig. 6B). The above observations provide further evidence for the potency of dpp and wg to trigger a P/D axis and provide a molecular explanation for why misalignment of anterior and posterior tissues cause duplications in the transplantation experiments. When the transplantation with 180 rotation is performed (Fig. 3B), the two new sites of interface between posterior and anterior tissues will behave similarly. Hence, two new stripes of dpp will be induced in the dorsal regions of the two new A/P interfaces and corresponding wg expression will

8 626 G. Campbell and A. Tomlinson occur ventrally (Fig. 3Biv). This now provides two supernumerary juxtapositions of tissues expressing the secreted factors wg and dpp (Fig. 3Bv) the required interaction for the initiation of a duplicated P/D axis. THE CHIRALITY OF SUPERNUMERARY LEGS When a right graft is transplanted and misaligned onto a left stump, the two supernumeraries are always of the opposite (left) chiral form to the graft. The chirality of a leg is determined by the spatial arrangement of the A/P and D/V axes the left arrangement being the mirror image of the right. In Fig. 3Aiv, the graft is of right chirality (posterior (green) is to the bottom, dorsal (blue) is to the right and ventral (red) is to the left). The two supernumeraries (characterized by the ectopic juxtaposition of wg- and dppexpressing cells), however, have the opposite (left) chirality, posterior (as in the graft) is still to the bottom but now the dorsal tissue (dpp-expressing) is to the left and ventral tissue (wg-expressing) is to the right. Similarly, in Fig. 3Bv the graft is of right chirality (posterior to the left and dorsal below) but the duplications are the mirror image (posterior to the right and dorsal below). Hence, the very manipulations that result in the duplicated outgrowths supply the circumferential (chiral) identity of the supernumeraries suggesting that the interaction of dpp-expressing cells with wgexpressing cells does not redefine the axial information of a complete new appendage. Rather the A/P and D/V information is provided by the manipulations and it is the initiation of a new P/D axis that is the key event. PROXIMODISTAL AXIS INITIATION AND THE DISTAL ORGANIZER We have presented extensive evidence that, in the insect leg, a P/D axis is initiated at the site where cells expressing wg abut those expressing dpp. We propose that here a proximodistal organizing center is triggered; this organizer is central to the two dimensional disc (corresponding to the presumptive tip of the three dimensional appendage) and controls outgrowth and the specification of cell fates along the P/D axis by a mechanism still to be determined. Normal legs contain a single P/D axis because there is only one site at which wg-expressing cells abut dpp-expressing cells; this is in the Fig. 6. The effect of ectopic expression of w g and h h on the P/D axis can be explained if the P/D axis is initiated from the site where cells expressing w g abut those expressing d p p. (A) A small clone of cells ectopically expressing w g (orange) will induce a secondary P/D axis only if it is situated in the dorsal region next to the endogenous stripe of d p p (blue) expression. Ectopic a l expression is induced where cells expressing d p p juxtapose those expressing w g and this marks the presumptive tip of the supernumerary. (B) A small clone of cells ectopically expressing h h in the dorsal/anterior region of the leg disc will induce ectopic expression of d p p. Similarly such a clone in the ventral/anterior will induce ectopic expression of w g. A supernumerary leg will be generated if two such clones are closely situated in the same leg disc. (C) Ubiquitous expression of w g or h h transform the disc into an oval. Normally, the P/D axis is initiated at only a single point, the centre of the disc, because this is the only site where cells expressing w g come into contact with those expressing d p p. When w g is expressed ubiquitously this single point becomes a line that corresponds to the dorsal stripe of d p p stretching the disc dorsally. When h h is expressed ubiquitously, d p p is induced in all of the cells of the dorsal/anterior and w g in the ventral/ anterior so that, again, the P/D axis is initiated as a line, in this case corresponding to the interface between dorsal/anterior and ventral/anterior thus stretching the disc anteriorly.

9 Initiation of the proximodistal axis in insect legs 627 Fig. 7. Molecular interactions that maintain a single, central distal organizer in the leg. These diagrams represent the leg imaginal disc of Drosophila and show that normally there is only one site at which cells expressing dpp abut those expressing wg and this is in the centre of the disc. (A) The leg is divided into three cell populations - posterior, dorsal/anterior and ventral/anterior (the posterior compartment cells are characterized by the expression of the en homeobox gene, but the molecular basis for the subdivision of the anterior compartment is not known). (B) The posterior cells express the hh gene (green) which encodes a secreted signaling molecule that is able to diffuse into the anterior compartment (arrows). (C) Hh has no effect on posterior cells, but in the dorsal/anterior it induces dpp expression (blue) and in the ventral/anterior it induces wg expression (orange). Because Hh only has a limited effective range of diffusion, wg and dpp are only expressed in a thin strip of cells at the compartment border. (D) The only site at which cells expressing dpp abut those expressing wg is in the center of the disc and is characterized by expression of the al homeobox gene (black). At this site, a proximodistal organizing center is triggered and controls outgrowth and the specification of cell fates along the P/D axis. centre of the imaginal disc, the presumptive distal tip. To understand why there is only one site requires an understanding of the control of wg and dpp expression in the leg. In hemimetabolous insects the legs grow directly from the embryonic ectoderm but in holometabolous insects such as flies the presumptive leg epithelium invaginates as an imaginal disc and the elaboration of a leg structure is suppressed for a number of days. In Drosophila the cells forming the leg discs invaginate in the ventrolateral region of the thoracic segments where a ventral stripe of wg expression intersects perpendicularly with a ventrolateral/anteroposterior line of dpp expression (Cohen et al., 1993). This probably marks the site where the proposed distal organizer would be initiated in hemimetabolous development. It is not clear if this is established at this time in Drosophila, but this may be the case because both aristaless and another homeobox gene, Distaless, which mark distal leg structures, are transcriptionally activated in cells of the presumptive leg disc (although aristaless is turned off soon after invagination and reappears later; Cohen et al., 1993; Campbell et al., 1993). The initial expression domains of wg and dpp in the embryonic ectoderm appear to be independent of hh (Martinez Arias, 1993) and so the distal organizer may be initiated independent of a Hh signal. wg expression soon becomes hh-dependent (around the same time as disc invagination; Ingham and Hidalgo, 1993), but it is not clear exactly when dpp expression comes under the control of Hh. During larval life, wg and dpp expression in the leg imaginal discs appear to be controlled by Hh. As in the boundary model (Meinhardt, 1982), the leg can be divided into three populations of cells: (a) posterior cells, characterized by engrailed expression and the by the secretion of Hh protein; (b) anterior/dorsal cells, which respond to Hh by expressing dpp; (c) anterior ventral cells, which respond to Hh by expressing wg (Fig. 7A). The molecular or developmental basis for the subdivision of the anterior compartment is unknown. Hh protein diffusing from the posterior compartment elicits wg and dpp expression only in the nearby anterior cells (Fig. 7B, C; Struhl and Basler, 1994). As a consequence, a thin wedge of ventral wg expression and a thin strip of dorsal dpp expression lie tightly up against the compartment border and meet each other at the disc s centre. Thus, there is only a single position where dpp and wg expression domains meet (Fig. 7D) defining the position of the distal organizer. At present there is no evidence regarding the molecular nature of the distal organizer; Meinhardt s original boundary model suggested that this site is the source of another signaling molecule or morphogen but there is no evidence for or against this. At present the model remains at the level of axis initiation: where wg-expressing cells abut dpp-expressing cells, a P/D axis will be triggered. THE COMPARISON WITH VERTEBRATES Although insect and vertebrate legs have evolved independently and are anatomically different there are some important similarities in their developmental biology. In common with many insects, some amphibians, such as salamanders, possess the ability to regenerate their legs. Additionally, salamander legs behave almost identically to those of insects in the formation of supernumeraries following grafting experiments that inappropriately juxtapose anterior and posterior or dorsal and ventral tissues (Harrison, 1921; Bryant and Iten, 1976). This poses the question of whether there is a similar mechanism for P/D axis initiation in vertebrate and insect legs. Patterning in the insect leg occurs in the monlayered epidermis which forms a hollow cone; in vertebrate legs, the epithelial cone is filled with mesoderm and this appears to be the tissue that controls axis initiation (reviewed in Bryant et al., 1987). However, in regenerating amphibian legs, the mesodermal cells controlling patterning may be limited to the fibroblasts which form a two-dimensional cone within the leg, reminiscent of the epidermal cone of the insect leg (Bryant et al., 1987). Thus, the underlying geometry of pattern formation in the vertebrate and invertebrate legs may be similar. The three classes of secreted factors that interact in the control of P/D axis initiation in insects, TGFβ, Wnt and Hedgehog, are all present in the developing vertebrate appendage and, as in the insect leg, show regional circumferential expression (Lyons et al., 1989; Gavin et al., 1990; Parr et al., 1993; Riddle et al., 1993; Echelard et al., 1993; Francis et al., 1994). We await further investigations that will reveal if interaction between groups of cells expressing different signaling molecules is required to initiate the P/D axis in vertebrates in a similar fashion to the mechanism that we have described for the insect. We thank Gary Struhl and Konrad Basler for the hh flies, Mike Hoffman for the dpp antibody, Peter Shelton and Paul Truby for the

10 628 G. Campbell and A. Tomlinson cockroach leg, Richard Axel, Tom Jessell, Ariel Ruiz i Altaba, Claudio Stern, Gary Struhl and Marcel Wherli for their comment on the manuscript and helpful discussions and also to Alfonso Martinez Arias and an anonymous reviewer for helpful comments. REFERENCES Baker, N. E. (1988a). Transcription of the segment-polarity gene wingless in the imaginal discs of Drosophila, and the phenotype of a pupal-lethal wg mutation. Development 102, Baker, N. E.(1988b). Embryonic and imaginal requirements for wg, a segment polarity gene in Drosophila. Dev. Biol. 125, Basler K. and Struhl, G. (1994). Compartment boundaries and the control of Drosophila limb pattern by hedgehog protein. Nature 368, Bodenstein, D. (1933). Beintransplantationen an Lepidopterenraupen. II. Zur Analyses der Regeneration der Brustbeine von Vanessa urticae-raupen. Wilhelm Roux Arch. Entwickl. Mech. Org. 130, Bodenstein, D. (1937). Beintransplantationen an Lepidopterenraupen. IV. Zur Analyse experimentell erzeugter Beinmehrfachbildungen. Wilhelm Roux Arch. Entwickl. Mech. Org.136, Bohn, H. (1965). Analyse der Regenerationsfahigkeit der Insekten-extremitat durch Amputations- und Transplationsversuche an Larven der afrikanschen Schabe Leucophaea maderae Fabr. (Blattaria). II. Mitt. Achsendetermination. Wilhelm Roux Arch. EntwMech. Org. 156, Bryant, P. (1978). Pattern formation in imaginal discs. In: The Genetics and Biology of Drosophila, Volume 2c (eds. M. Ashburner and T. R. F. Wright). New York: Academic Press. Bryant, P. (1993). The polar coordinate model goes molecular. Science 259, Bryant, S. V., French, V. and Bryant, P. J. (1981). Distal regeneration and symmetry. Science 212, Bryant, S. V., Gardiner, D. M. and Muneoka, K. (1987). Limb development and regeneration. Amer. Zool. 27, Bryant, S. V. and Iten, L. E. (1976). Supernumerary limbs in Amphibians: Experimental production in Notophthalmus viridescens and a new interpretation of their formation. Dev. Biol. 50, Bulliere, D. (1970). Interpretation des regenerats multiples chez les insectes. J. Embryol. Exp. Morph. 23, Campbell, G., Weaver, T. and Tomlinson, A. (1993). Axis specification in the developing Drosophila appendage: the role of wingless,decapentaplegic, and the homeobox gene aristaless. Cell 74, Cohen, B., Simcox, A. A. and Cohen, S. M. (1993). Allocation of the thoracic imaginal primordia in the Drosophila embryo. Development 117, Couso, J. P., Bate, M. and Martinez-Arias, A. (1993). A wingless-dependent polar coordinate system in Drosophila imaginal discs. Science 259, Echelard, Y., Epstein, D. J., St-Jaques, B., Shen, L., Mohler, J., McMahon, J. A. and McMahon, A. P. (1993). Sonic hedgehog, a member of a family of putative signaling molecules is implicated in the regulation of CNS polarity. Cell 75, Francis, P. H., Richardson, M. K., Brickell, P. M. and Tickle, C. (1994). Bone morphogenetic proteins and a signalling pathway that controls patterning in the developing chick limb. Development 120, French, V., Bryant, P. J. and Bryant, S. V. (1976). Pattern regulation in epimorphic fields. Science 193, French, V. and Daniels, G. (1994). The beginning and the end of insect limbs. Current Biology 4, Gavin, B. J., McMahon, J. A. and McMahon, A. P. (1990). Expression of multiple novel Wnt-1/int-1-related genes during fetal and adult mouse development. Genes Dev. 4, Gelbart, W. M. (1989). The decapentaplegic gene: a TGF-β homologue controlling pattern formation in Drosophila. Development 107 Supplement, Harrison, R. G. (1921). On relations of symmetry in transplanted limbs. J. Exp. Zool. 9, Ingham, P. W. and Hidalgo, A. (1993). Regulation of wingless transcription in the Drosophila embryo. Development 117, Kingsley, D. M. (1994). The TGF-β superfamily: new members, new receptors, and new genetic tests of function in different organisms. Genes Dev 8, Kornberg, T, Siden, I, O Farrell, P. and Simon, M. (1985). The engrailed locus of Drosophila; in situ localization of transcripts reveals compartment specific expression. Cell 40, Lawrence, P. A. and Morata, G. (1977). The early development of mesothoracic compartments in Drosophila; An analysis of cell lineage and fate mapping and an assessment of methods. Dev. Biol. 56, Lee, J. J., von Kessler, D., Parks, S. and Beachy, P. A. (1992). Secretion and localized transcription suggests a role in positional signaling for products of the segmentation gene hedgehog. Cell 71, Locke, M. (1984) Epidermal cells. In: Biology of the Integument. I Invertebrates (ed. J. Bereiter-Hahn, A.G. Matoltsy and K. Sylvia Richards). Berlin: Springer-Verlag. Lyons, K. M, Pelton, R. W and Hogan, L. M. (1989). Patterns of murine Vgr- 1 and BMP-2a RNA suggest that transforming growth factor-β-like genes coordinately regulate aspects of embryonic development. Genes Dev. 3, Martinez Arias, A. (1993). Larval epidermis of Drosophila. In The Development of Drosophila melanogaster (ed. M. Bate and A. Martinez Arias). New York: Cold Spring Harbor Laboratory Press. Masucci, J. D., Miltenberger, R. J., and Hoffmann, F. M. (1990). Patternspecific expression of the Drosophila decapentaplegic gene in imaginal disks is regulated by 3 cis-regulatory elements. Genes Dev. 4, Meinhardt, H. (1982). Generation of structures in a developing organism. In: Developmental Order: Its Origin and Regulation (ed. S. Subtelny and P.B Green) pp New York: Alan Liss Mohler, J. and Vani, K. (1992). Molecular organization and embryonic expression of the hedgehog gene involved in cell-cell communication in segmental patterning of Drosophila. Development 115, Nusse, R. and Varmus, H. E. (1992) Wnt-genes. Cell 69, Padgett, R. W., St. Johnston, R. D., and Gelbart, W. M. (1987). A transcript from a Drosophila pattern gene predicts a protein homologous to the transforming growth factor-β family. Nature 325, Parr, B. A., Shea, M. J., Vassileva, G., and McMahon, A. P.(1993). Mouse Wnt genes exhibit discrete domains of expression in the early embryonic CNS and limb buds. Development 119, Raftery, L. A., Sanicola, M., Blackman, R. K., and Gelbart, W. M. (1991). The relationship of decapentaplegic and engrailed expression in Drosophila imaginal disks: do these genes mark the anterior-posterior compartment boundary. Development 113, Riddle, R. D., Johnson, R. L., Laufer, E. and Tabin, C. (1993). Sonic hedgehog mediates the polarizing activity of the ZPA. Cell 75, Rijsewijk, F., Schuermann, M., Wagenaar, E., Parren, P., Weigel, D. and Nusse, R. (1987). The Drosophila homologue of the mouse mammary oncogene int-1 is identical to the segment polarity gene wingless. Cell 50, Roelink, H., Augsberger, A. M., Heemskerk, J., Korzh, V., Norlin, S., Ruiz i Altaba, A., Tanabe, Y., Placzek, M., Edlund, T., Jessell, T. M. and Dodd, J. (1994). Floor plate and motor neuron induction by vhh-1, a vertebrate homolog of hedgehog expressed by the notochord. C e l l 7 6, Sharma, R. P., and Chopra, V. L. (1976). Effect of the wingless (wg 1 ) mutation on wing and haltere development in Drosophila melanogaster. Dev. Biol. 48, Schubiger, G. (1968). Anlageplan, Determinationzustand und Transdeterminationleistungen der mannlichen Vorderbeinscheibe von Drosophila melanogaster. Roux s Arch. EntwMech. Org.160, Schubiger, G. and Schubiger, M. (1978). Distal transformation in Drosophila leg imaginal disc fragments. Dev. Biol. 67, Spemann, H., and Mangold, H. (1924). Über Induktion von Embryonenanlagen durch Implantation artfremder Organisatoren. Arch. Microsk. Anat. Entwicklungsmech.100, Spencer, F. A., Hoffman, F. M. and Gelbart, W. M. (1982). Decapentaplegic: a gene complex affecting morphogenesis in Drosophila melanogaster. Cell 28, Steiner, E. (1976). Establishment of compartments in the developing leg imaginal discs of Drosophila melanogaster. Wilhelm Roux Arch. Entwicklungsmech. Organismen 180, Struhl, G., and Basler, K. (1993). Organizing activity of wingless protein in Drosophila. Cell 72, Tabata, T., Eaton, S. E and Kornberg, T. B. (1992). The Drosophila hedgehog gene is expressed specifically in posterior compartment cells and is a target of engrailed expression. Genes Dev. 6, Truby, P. (1983). Blastema formation and cell division during cockroach limb regeneration. J. Embryol. Exp. Morph.85, Wigglesworth, V. B. W. (1959) The Control of Growth and Form: a Study of the Epidermal Cell in an Insect. New York: Cornell University Press. (Accepted 17 November 1994)