defective proventriculus is required for pattern formation along the proximodistal axis, cell proliferation and formation of veins in the Drosophila wing

The development of the Drosophila wing has become an important system to study the patterning and morphogenesis of an animal appendage (for a review, see Klein, 2001). The wing is formed by one of the imaginal discs, which are sheets of epithelial cells defined during embryogenesis that proliferate during larval development and form most of the adult fly.

Most work has concentrated on the patterning events that occur along the two existing axes, the anteroposterior (AP) and dorsoventral (DV) axes. Two patterning centres located at the DV and AP compartment boundaries provide positional information for the cells of the wing. At the AP boundary, a band of anterior cells along the boundary, defined by the Hedgehog (Hh) signal from posterior cells, express the secreted factor Decapentaplegic (Dpp). Dpp diffuses from these cells to both sides and generates a gradient, which supplies the wing cells with positional information along the AP axis(reviewed by Basler, 2000; Klein, 2001). Likewise, the Wg protein is produced in cells at the DV boundary under control of the Notch pathway and the nuclear factor Vestigial (Vg), and forms a bipartite gradient on each side of the boundary. This gradient is required to maintain the expression of vg in the cells of the wing pouch and to stabilize the expression of genes in the domain of Vg(Basler, 2000; Klein, 2001).

As the wing imaginal disc is a two-dimensional structure, the third axis,the PD axis, must be generated and patterned with help from the two existing axes. In the adult wing three regions of the PD axis are easily distinguishable. From proximal to distal, these are the hinge, the proximal wing (PW) and the wing blade (see Fig. 1A,C). Little is known about the genes and molecular strategies that establish and pattern this axis. It is known that the activity of the vg gene is required for the establishment of all distal wing fates(Klein and Martinez-Arias, 1998a; Klein and Martinez-Arias, 1999; Liu et al., 2000). Vg is a nuclear protein that associates with Scalloped (Sd) to form a bipartite transcription factor(Halder et al., 1998; Simmonds et al., 1998). The expression of vg is initiated at the DV boundary through the Notch signalling pathway (Kim et al., 1996). The descendants of the cells of the DV boundary will form the wing pouch (Klein and Martinez-Arias, 1999). Recent work has shown that Vg does not only determine the fate of cells within its domain of expression, but also in cells outside in the PW. Vg seems to activate an unidentified signal that induces the expression of rotund (rn) and nubbin(nub) in larger domains. Rn is a Zinc-finger containing transcription factor that is required, together with the POU domain transcription factor Nub(Nub), to induce the expression of Wg in a ring-like domain that frames the wing pouch (St Pierre et al.,2002; del Alamo Rodriguez et al., 2002). The induction of wg is a crucial event for the formation of the medial region of the PW(Neumann and Cohen, 1996). To get further insight in the patterning mechanisms that operate along the PD axis, it is important to identify new genes that are involved.

We report that the gene, defective proventriculus (dve)is required for the establishment of the region of the PD axis that comprises the distal part of the PW and a small part of the adjacent wing pouch. The function of dve is further required for other aspects of wing development, such as the formation of wing veins 2 and 5, and the proliferation of cells of the wing pouch. dve encodes a novel type of homeobox protein, which was originally shown to be required for the proper development of the larval proventriculus(Fuss and Hoch, 1998; Nakagoshi et al., 1998). The expression of dve in the wing imaginal disc occurs in a disc-like domain that is smaller than that of Nub and Rn but larger than that of Vg. It is initiated by Vg in a non-autonomous manner, and suppressed at the DV boundary by the combined activity of Nub and Wg. Our work reveals that Vg patterns the PW through the induction of expression of genes in disc-like domains of different sizes. The size difference of the expression domains creates concentric regions with differential gene activity, which directs the formation of the three regions of the PW. The regulation of dveexpression by Vg is also the first obvious connection between Vg and the regulation of cell proliferation, which is clearly distinguishable from its role in pattern formation.

The dve^P1738-FRTG13 and UAS-dve lines, as well as Df(2R) 58-5, have been described previously(Fuss and Hoch, 1998; Nakagoshi et al., 1998). dveGal4 (P(GT1)dve^BG02382) was obtained from the Bloomington Stock Centre. UAS-vg is described by Kim et al.(Kim et al., 1996). nub¹, spd^flg were provided by Steve Russell. fj-lacZ(Villano and Katz, 1995) was a gift of F. Katz. rn-lacZ was a gift of J.-P. Couso and is described by St. Pierre et al. (St. Pierre et al., 2002). vg^83b27R is a null mutation of vg, and together with the UAS-vg, vg-QE and vg-BE,was provided by S. Carroll (Kim et al.,1996). The arr²-FRTG13 chromosome(Wehrli et al., 2000) was a gift of S. DiNardo. dppGal4, UAS-Nintra and UAS-wgstocks are described by Klein and Martinez-Arias(Klein and Martinez-Arias,1998), and UAS-Flp(Duffy et al., 1998) was provided by N. Perrimon.

The arr²-FRTG13 chromosome was used to induce arr-mutant clones with help of the FLP/FRT system. The mutant clones were induced using an UAS-FLP construct activated by vgGAL4. dve^P1738-FRTG13 clones were induced using an hsFlp construct. Wing imaginal discs were prepared at the late third larval instar stage, 48 hours after heat shock. Flip-out clones were induced with help of the AyGal4-UAS-GFP chromosome, kindly provided by K. Ito(Ito et al., 1997).

The following antibodies were used: anti-Wg, anti-Dve, anti-β-Gal and anti-Nub. The anti-Wg antibody was obtained from the Developmental Studies Hybridoma Bank, developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242, USA. Anti-Nub was a gift of M. Averof and anti-Dve(Nakagoshi et al., 1998) was a gift of F. Matsuzaki.

Staining was performed according to standard protocols. The FITC- and Texas Red-conjugated secondary antibodies were purchased from Jackson Immuno Research.

The dve^P1738 mutation (also called dve¹) is an insertion of a P-lacZ transposable element in the second intron of the dve gene. The insertion causes a severe truncation of the mRNA of the large transcript encoded by the dve gene (Fuss and Hoch,1998). Homozygous dve^P1738 animals are reported to die during the first larval instar stage as a result of a failure of the formation of the proventriculus(Fuss and Hoch, 1998; Nakagoshi et al., 1998). Previous work has shown that the mutant phenotype is caused by the P-element insertion, and can be reverted by precise excision of the P-element(Fuss and Hoch, 1998; Nakagoshi et al., 1998).

We found that, although the majority of the mutant animals die as first instar larvae, a small percentage of the animals develop until adulthood, and some even hatch. However, these flies displayed defects in several adult structures, such as the wing, the haltere, the leg and the head. We have identified another insertion of a P-Gal4 construct: P(GT1)dve^BG02382 (Gene Disruption Project members,2001.1.29), inserted in the second intron of dve, which will now be referred to as dve-Gal4. dve^P1738 is lethal in trans-heterozygousity to the deficiency Df(2R) 58-5 or dve-Gal4, indicating that the dve^P1738 is not a null allele in general. However, we could not detect any protein in wing imaginal discs of homozygous-mutant animals (see below), indicating that dve^P1738 is a strong allele for wing development. The lethality of the dve-Gal4/dve^P1738-heterozygous animals can be rescued by the presence of a UAS-dve construct. These`rescued' animals exhibit a slightly weaker wing phenotype than dve^P1738-homozygous mutants. We could not detect any activity of dve-Gal4 in any imaginal disc using a UAS-GFP reporter construct. Thus, it appears that dve-Gal4 expression is restricted to the time of embryogenesis and that this expression is sufficient to let the dve-Gal4/dve^P1738 animals survive in the presence of UAS-dve. Nevertheless, the adult phenotype displayed by this allelic combination indicates that the loss of dve function is the cause of the observed wing defect. In this work, we have concentrated on the analysis of the function of dve during wing development.

The adult Drosophila wing is subdivided into several domains along the proximodistal axis. Proximal-most is the hinge, which connects the proximal wing and the wing blade to the body wall(Fig. 1A). The proximal wing consists of several regions, among them the costa at the anterior margin. The costa can be further subdivided into three easily distinguishable parts: a proximal, a medial and a distal part (Fig. 1A,C). We found that wings of dve-mutant flies were smaller and shorter than wild-type flies(Fig. 1B). A detailed analysis showed that a region that encompasses most of the distal costa and a small part of the adjacent wing blade is missing in dve-mutant flies(Fig. 1C,D). As a result of the deletion, the distance between the end of the medial costa and the first cross-vein of the wing blade is reduced (compare distance between the arrowheads in Fig. 1A and B).

The comparison of wild-type and mutant wings further reveals that dve-mutant wings are also reduced in size along the anteroposterior(AP) axis (compare Fig. 1A and B; see Fig. 3A). Furthermore, wing vein 2 is interrupted in the proximal part, and the distal part of vein 5 is lost (arrows in Fig. 1A,B).

These phenotypes indicate that Dve is involved in the formation of the distal part of the PW, and in the correct development of wing veins 2 and 5,and that it is required for the regulation of the size of the wing.

dve mutants show recognizable abnormalities by the late third larval instar wing imaginal discs (Fig. 2). In mutant discs, the anlage of the dorsal part of the wing pouch is shorter, as revealed by the distance between the DV boundary and the inner ring-like expression domain of Wg expression in the proximal wing(Fig. 2A,C). The defects are more easily recognized during the early pupal phase, when the wing has evaginated (Fig. 2B,D). The wing pouch of dve mutants is smaller than in wild type and has a small indentation at the anterior wing margin (arrow in Fig. 2B,D). Furthermore, the fold adjacent to the wing blade (asterisk in Fig. 2B,D) appears to be reduced in the mutant wings (see arrowheads in Fig. 2B,D). These observations indicate that the defect in dve-mutant wing imaginal discs occurs before the late third larval instar stage. As we do not find any abnormal cell death in dve-mutant wing imaginal discs (data not shown), it is probable that the anlage of the proximal part of the PW, as well as the adjacent area of the blade, is not established in the absence of Dve function.

Nakagoshi et al. reported that dve^P1738-mutant cell clones, including the wing margin, cause the formation of ectopic bristles as well as nicks in the wing margin (Nakagoshi et al., 2002). We could not observe such perturbations in animals homozygous for this allele. (For an explanation of this discrepancy, see Discussion.)

The reduction in size of the dve-mutant wings along the AP axis could simply be due to lack of cell growth. This conclusion is supported by the feeding defect reported for dve mutants(Fuss and Hoch, 1998), as the development of starved flies is slower and their cells are smaller. However,the size of the area outlined by the wing veins 3 and 4, and the anterior cross-vein and wing margin is of similar size in wild type and mutant(Fig. 3A). This suggests that at least in this area the cells are of similar size. The size reduction of the mutant wings could also be a result of increased cell death. However, we did not observe any enhanced cell death in dve-mutant wing imaginal discs of the early and late third larval instar stage (see above).

A further possibility is that the cells proliferate less in the mutant wings. To test this possibility, we compared distances and cell densities in several regions of the mutant and wild type wing(Fig. 3A). The cell density of both types of wings was very similar in the area between wing veins 3 and 4. Furthermore, the distance between wing vein 3 and wing vein 4, measured by the numbers of cells between them, is the same. Hence, no differences exist between mutant and wild-type wings in this region. These data are consistent with our observation that this region is of similar size in both genotypes.

However, we found differences between wild-type and dve-mutant wings in more anterior and posterior regions. Although the cell density in the region anterior to vein 3 was similar, the distance between vein 3 and the anterior wing margin was reduced in the dve-mutant wing, indicating that this area consists of fewer cells(Fig. 3A, measurements A and D). A similar difference was observed in the area between wing vein 4 and the posterior margin. We found that the distance from vein 4 to the posterior margin is again reduced. In addition, the cell density in this region is lower in the mutant (Fig. 3A,measurements C and F), indicating that the mutant cells are larger. Thus,there are fewer cells and the cell size is increased in the posterior area of dve-mutant wings. Enlargement of cells is a typical reaction of wing cells if their proliferation is inhibited and it is interpreted as a compensatory mechanism in order to achieve a normal organ size(Weigmann et al., 1997; Neufeld et al., 1998). The data suggest that the observed size reduction of dve-mutant wings is the result of a lower proliferation rate of cells located within the regions anterior of vein 3 and posterior of vein 4. Thus, Dve is required for the correct proliferation of wing pouch cells in these regions.

To further explore whether dve-mutant cells proliferate less than wild-type cells, we examined the behaviour of dve^P1738-mutant cell clones(Fig. 3B-F; Table 1). The clones were induced with help of the Flp/FRT system and were analysed 48 hours after clone induction by hsFlp. Because our analysis of the adult wing suggests that the behaviour of dve-mutant cells is dependent on the region of the wing,we subdivided the wing blade along the AP axis into three areas. These areas were defined by the expression pattern of Delta (Dl), which is expressed in the anlagen of wing veins 3, 4 and 5 (Fig. 3E). Area 1 (A1) extended from the anterior wing margin to vein 3,area 2 (A2) extended from vein 3 to vein 4, and area 3 (A3) extended from vein 4 to the posterior margin (Fig. 3E). Three effects were observed(Fig. 3B-F; Table 1). First, we found that mutant clones had fewer cells than their wild-type twin clones (see Table 1). However, the number of cells in mutant clones was dependent on its location: whereas the number of cells in mutant clones in A1 and A3 was roughly half that of their wild-type counterparts, the mutant clones in A2 had around two thirds of the cells that their wild-type twins did (see Table 1). The results indicate that after 48 hours, the mutant cells in A1 and A3 have gone through one cell cycle less than their wild-type counterparts.

Second, we found that in A1 and A3 (but not in A2), nearly half of the wild-type clones did not have a mutant counterpart (47% and 43%,respectively), which suggested that the mutant cells had died. As the mutant cells do not die in homozygous animals, we believe that cells of mutant clones undergo apoptosis because of a disadvantage in competing with wild-type neighbours. Cells defective in proliferation typically apoptose(Neufeld et al., 1998; Moreno et al., 2002). Altogether, these results further support our conclusion that dve-mutant cells are defective in cell proliferation.

A third interesting aspect is that, in some cases, a wild-type clone is separated from its mutant twin clone by a band of heterozygous cells (see arrows in Fig. 3F). This suggests that the two types of clones might have different adhesive properties. Alternatively, the heterozygous cells could have migrated into the area between the two clone types because some of the mutant cells died.

To gain further insight in the function of Dve, we monitored its expression pattern during wing development by use of an anti-Dve antibody. We compared the expression pattern of Dve with that of Wg, which is expressed throughout wing development in a pattern that reveals the organization of the developing wing (Fig. 4B,E,H,K). Wg is initially expressed in a ventral domain during the second larval instar stage and defines the wing area or wing field(Fig. 4B,C) (reviewed by Klein, 2001). At this time,Dve is not expressed in the wing imaginal disc(Fig. 4A,C). At the beginning of the third larval instar stage, Wg expression resolves into a stripe along the future DV compartment boundary and a proximal ring-like domain(Fig. 4E). In the middle of third larval instar stage a second ring-like domain in the proximal region of the anlage is added. The two ring-like domains of Wg expression highlight the anlagen of the proximal and medial regions of the proximal wing, as deduced from X-Gal staining of adults carrying a wg-lacZ construct(see Fig. 5A). Dve expression is initiated at the time when Wg resolves into a ring-like domain in the periphery and a domain along the DV boundary, and it becomes expressed in all cells inside the region framed by the ring-like domain of Wg(Fig. 4D,F). Dve continues to be expressed in a disc-like domain that fills the inside of the inner ring-like expression domain of Wg until the late third larval instar stage(Fig. 4G,I).

The anlage of the distal region of the PW, is located outside the wing pouch and inside the inner ring-like domain of Wg expression(Fig. 4I; see also Fig. 5A). Dve is expressed continuously in this region, and is present at the right place and time to control the development of this structure, which is absent in the mutants.

At the DV boundary, Dve is initially expressed(Fig. 4D,F), but it becomes downregulated soon after its initiation (arrowhead in Fig. 4G,I), with the exception of a short stretch at the anterior side (arrowhead in Fig. 4G). During the late third larval instar stage, it is also downregulated in the primordia of wing veins 3 and 4 (arrows in Fig. 4G).

We failed to detect any Dve protein in wing imaginal discs of homozygous-dve^P1738 larvae(Fig. 4J,K), which suggests that this allele is probably a null allele for wing development.

We have mapped the expression domain of dve in relation to that of other genes known to be involved in PD patterning of the wing, and in relation to the ring-like domains of wg. The ring-like domains label the region of the proximal and medial costa, as revealed by the X-Gal staining of adult wings bearing a wg-lacZ insertion(Fig. 5A).

vestigial (vg) is required for all distal fates from the medial costa distalwards. It is initially expressed in all pouch cells(Kim et al., 1996; Klein and Martinez-Arias, 1998a; Liu et al.,2000) and its expression is controlled through the vg-Quadrant enhancer (vg-QE)(Kim et al., 1996). We found that the expression domain of dve is larger earlier than that of the vg-QE. In addition, dve expression is initiated before the vg- QE is activated, which indicates that dve expression is initiated before the wing pouch forms (Fig. 5C-E; data not shown).

Nub is involved in patterning the wing from the medial costa distalwards(Ng et al., 1995). The nub gene is expressed in a disc-like domain that is slightly larger than that of dve (Fig. 5F-H) and that extends to the area between the two ring-like domains of wg expression (Fig. 5G,H). Examination of wing discs of early third instar larvae revealed that nub expression is initiated earlier than dve,and is always expressed in a larger domain than dve (data not shown).

The boundary of the expression domain of rotund (rn)falls between that of dve and nub. Its domain reaches the proximal boundary of the inner ring-like domain of wg expression(Fig. 5J).

By contrast, the expression domain of dve is larger than that of the four-jointed (fj) gene, which is expressed in a similar pattern to vg (Fig. 5I). The results of the comparison of the expression domains are schematically summarized in Fig. 5K. The cartoon reveals that the cells of the different regions of the proximal wing contain different combinations of gene activities. These specific combinations appear to trigger the region-specific differentiation in these cells.

Vg is required for the establishment of distal wing fates, including the medial and distal areas of the proximal wing (Klein and Martinez-Arias, 1998a; Liu et al., 2000; del Alamo Rodriguez et al.,2002). This raises the possibility that Vg might activate the expression of dve. To test this possibility, we first monitored the expression of dve in vg-mutant wing imaginal discs. We found that in vg^83b27R mutants, the expression of dveis lost (Fig. 6A), which indicates that Vg activity is required for its expression. Note that the expression domain of vg is always smaller than that of dve(see above), which suggests that Vg regulates the expression of dvein a non-autonomous manner. We next investigated whether Vg is sufficient to activate dve expression. To address this question, we generated clones of vg-expressing cells in the wing imaginal disc with help of the Flip-out technique (Ito et al.,1997). Clones of vg-expressing cells were indeed able to induce ectopic expression of Dve (Fig. 6B,C). This result indicates that Vg is sufficient to activate expression of dve. The ectopic expression of dve was not restricted to the clones, but also occurred in cells surrounding the clones(arrows in Fig. 6B,C). The result confirms the conclusion that Vg induces dve expression in a non-autonomous manner. Hence, the induction of dve expression by Vg is indirect and probably mediated by a diffusible factor, the expression of which is controlled by Vg. Note, that the ability of Vg to ectopically induce dve expression is restricted to the wing and pleural regions,indicating that it requires the activity of other factors in other regions of the disc. As co-expression of vg and wg can induce wing fates in the notum (Klein and Martinez-Arias, 1998), we tested whether this combination is also sufficient to activate expression of dve. Indeed, we found that the combination of UAS-vg and UAS-wg activated by dpp-Gal4 was able to induce expression of dve in the notum(data not shown).

The study of dve expression during wing development revealed that it is downregulated at the DV boundary (see Fig. 4). The Notchpathway is active at the DV boundary and regulates the expression of genes such as wg and vg. We therefore wondered whether it is the activity of Notch that suppresses the expression of dve at the DV boundary. The ectopic expression of the activated intracellular domain of Notch, UAS-Nintra, in the wing pouch with dpp-Gal4 results in the loss of dve expression (Fig. 6D), which indicates that activation of the pathway suppresses the expression of dve in pouch cells. However, the suppression of dve expression in normal wing imaginal discs occurs gradually and reaches several cell diameters from the DV boundary into the wing pouch, where the Notch pathway is not active. This suggests that the influence of Notch on the expression of dve is mediated by a diffusible factor that is controlled by Notch signalling at the DV boundary. Wg is such a factor and we tested whether it can suppress the expression of dve. We found that UAS-wg, expressed with dpp-Gal4, can suppress dve expression ectopically in pouch cells in the same manner as Nintra (Fig. 6E). Furthermore,pouch cells that lack the Wg co-receptor Arrow and cannot receive the Wg signal (Wehrli et al., 2000),express higher levels of Dve at the DV boundary, and even further away from the DV boundary (Fig. 6F-I). Both results suggest that Wg mediates the suppressive effect of the Notch pathway on dve expression in pouch cells near the DV boundary.

The spade^flag (spd^flg) mutation of wg is lacking the regulatory region that directs expression of wg in the inner ring-like domain and causes the loss of the medial block of the proximal wing. We found that in spd^flgmutants, the expression of dve is not affected(Fig. 6M). In agreement with this is that the distal part of the PW forms normal in spade^flag mutants.

We also found that dve is still expressed in nub mutants(Fig. 6J-L), indicating that Nub is not required for the expression of dve. However, dveexpression was not suppressed at the DV boundary(Fig. 6K,L). Hence, in addition to Wg, Nub seems to be required to suppress dve expression at the DV boundary. As expected, Nub expression is not altered in dve-mutant wing discs, indicating that Dve is not required for its expression in the wing region (Fig. 6N).

To further explore the function of Dve during the development of the wing,we studied the effects of ectopic expression of Dve in the wing imaginal disc. We used the Flip-out technique to ectopically express UAS-dve in clones of cells.

We observed three effects caused by clones of dve-expressing cells(Fig. 7A-G). dve-expressing clones induced the formation of folds around the clone if they were located in the region of the anlage of the distal part of the PW(arrowhead in Fig. 7A-C); this is a region where it is normally expressed. Thus the formation of these ectopic folds suggests that elevation of Dve levels can induce the proliferation in cells of the distal part of the PW. This is probably mediated by a diffusible factor.

The second effect was that clones at the DV boundary suppressed the expression of Wg (Fig. 7F,G). In accordance with this result, we found that expression of UAS-dvewith sd- Gal4 or vg-BE-Gal4, which are strongly expressed at the DV boundary throughout wing development, results in a loss of wing structures as indicated by the absence of wg expression in the wing region (Fig. 7H; data not shown). These results suggest that expression of Dve at the DV boundary is deleterious for wing development, probably because it suppresses the expression of wg.

In addition, we found a third, low frequency phenotype. Clones of dve-expressing cells located outside the normal expression domain were able to non-autonomously expand the expression of nub, wg and dve itself (Fig. 7A-E). The ability to expand the expression of these genes was restricted to clones located in the region of the PW and the hinge. As in the wild-type discs, the ectopic expression domain of nub was larger than, and included, that of dve(Fig. 7A-C; highlighted by the arrows). wg expression was correspondingly expanded(Fig. 7D). As the functions of Nub, Wg and Dve are necessary to specify the medial and distal hinge, the results suggest that Dve can induce these more distal fates in the proximal region of the PW, albeit with a low frequency.

However, the low frequency of this effect and the observation that dve-expressing clones located in the notum had no detectable effects,suggests that during normal development Dve is probably not sufficient to establish the distal part of the PW. In agreement with this conclusion is the observation that ectopic expression of UAS-dve with dpp-Gal4 deletes most of the PW. This can be seen in the wing imaginal disc depicted in Fig. 7E, where both ring-like domains of Wg are interrupted in the dpp domain. This interruption is not caused by an extension of the anlage of the distal area of the PW, because in the corresponding adult wings this part, as well as the other regions of the PW, is severely reduced or deleted (data not shown). These data suggest that in the majority of cases the ectopic and overexpression of Dve is deleterious for wing development. Thus expression of Dve has to be tightly coordinated with that of other factors.

Relatively little is known about the genes and mechanisms that pattern the PD axis. We have identified dve as a gene that is involved in pattern formation along this axis, i.e. for the specification of most of the distal region of the proximal wing and the adjacent proximal region of the pouch. In dve-mutant flies this region is deleted, whereas all other pattern elements of the PD axis are not affected. Expression of dve is initiated at the beginning of wing development, and it is expressed in the region from which the distal part of the PW and a small stripe of the adjacent wing blade form. We could observe defects in the morphology of mutant-wing discs by the late third larval instar stage. Because abnormal cell death was not observed in dve-mutant wings at earlier stages, the lack of the distal part of the PW in dve mutants could be caused by a failure in establishment of this region. However, we found that overexpression of Dve achieved through the Flp-out technique results in excessive proliferation of cells in the region of the distal part of the PW. This suggests that Dve might be required for the correct proliferation of the cells in this region. Hence,the loss of the distal part of the PW in dve mutants could also be explained by a failure in proliferation of the cells in the anlage of the distal region of the PW.

We found that ectopic expression of Dve does not cause the more proximal regions of the PW to become more distal, which indicates that other factors are required in addition to Dve to establish the distal part of the PW. One of these factors is Nub, which is involved in the establishment of the medial as well as the distal area of the PW (Ng et al., 1995; Rodriguez et al., 2002). However, neither ectopic expression of Nub (Neumann and Cohen,1998) (T.K., unpublished), nor a combination of Nub and Dve (T.K.,unpublished), consistently induces ectopic structures characteristic of the PW. Therefore, it is likely that a combination of Dve, Nub and other factors is required for the establishment of the distal area of the PW and the adjacent blade region.

Recent work has revealed that Nub seems to act in combination with Rn to establish the medial part of the PW. Both factors cooperate to establish the inner ring-like domain of wg expression(del Alamo Rodriguez et al.,2002). Thus, it appears that separate regions of the PW are established independently through different combinations of transcription factors.

Nakagoshi et al. reported that dve^P1738-mutant cell clones near the DV boundary of the wing lead to the formation of ectopic bristles characteristic for the wing margin (Nakagoshi et al., 2002). Concomitant with these pattern disturbances, the authors found ectopic expression of wg in the mutant cell clones. Based on these observations, they proposed that Dve is required for the refinement of wg expression. However, we do not find any defects in the bristle pattern of flies, homozygous for the same allele, or in other dve-mutant situations. Therefore, we believe that the disturbances in the bristle pattern caused by the mutant clones are a result of the artificial apposition of Dve-expressing and non-expressing cells near the DV boundary,created by the induction of clones. We think that these disturbances do not reveal the biological function of Dve. In accordance with this conclusion is the observation that expression of Dve is suppressed along the DV boundary.

In addition to its function in pattern formation along the PD axis, our work showed that Dve is required for the proper proliferation of the wing pouch cells. Interestingly, the requirement for Dve differs along the PD axis. In the area anterior to wing vein 3 or posterior to wing vein 4 (areas A1 and A3 in Fig. 3E), dve-mutant cell clones contained only half as many cells as their wild-type counterpart. Hence, the mutant cells trailed their wild-type counterpart by one cell cycle after 48 hours. In addition, in many cases orphan wild-type clones without a mutant twin were found, which suggested that the mutant cells had died. Cell death is a typical reaction for cells that are impaired in cell proliferation (Weigmann et al., 1997; Neufeld et al.,1998). Both observations indicate that dve-mutant cells have a slower proliferation rate than wild-type cells. It is likely it is the slower rate of proliferation that causes the size reduction we observed in regions A1 and A3 of the dve-mutant wings. Proliferation of dve-mutant cells in the area A2 is also reduced, albeit to a lesser degree. The mutant clones contained 66% of the number of cells that their wild-type counterparts did. More importantly, we did not observe orphan wild-type clones, which indicates that the mutant cells do not undergo apoptosis in this region. Furthermore, the A2 area is of the same size in dve-mutant and wild-type wings. Hence, it appears that proliferation of dve-mutant cells is not as severely affected in A2 as it is in the other regions. This milder defect in proliferation of mutant cells in A2 seems to be compensated during later development. Altogether, our data suggest that Dve is required for the proliferation of all wing pouch cells, but the requirement for its activity varies along the AP axis.

Why do dve-mutant cells proliferate less? The observed cell death of mutant cells in A1 and A3 gives a hint to the answer. Cell death is probably not caused by a defect in the cell cycle machinery itself, as no increased cell death was found in homozygous dve-mutant animals. Furthermore, overexpression of Dve using the Flp-out technique does not lead to an over-proliferation of pouch cells. Hence, it is probable that the mutant cells die as a result of being disadvantaged when in competition with normal cells for survival factors, as has been recently shown for cells heterozygous for Minute mutations (Moreno et al., 2002). In the case of the Minute mutations, the survival factor is Dpp, which is also responsible for pattern formation along the AP axis (Moreno et al.,2002). The differential requirement of Dve along the AP axis suggests that it might be required for the reception of Dpp in pouch cells. However, one result argues against this possibility: Dve is required most in cells that are far away from the source of Dpp (which is at the AP boundary). However, these cells are not, or are only weakly, dependent on Dpp for their survival. Hence, it is unlikely that dve-mutant cells cannot properly receive Dpp.

We found that dve expression is initiated shortly after the start of wing development, during the early phase of the third larval instar stage. It is expressed in a disc-like domain that fills the region inside the inner ring-like domain of wg expression. We found that Vg is required, and is sufficient, for dve expression in the wing region. Importantly,our data show that Vg activates the expression of dvenon-autonomously, which indicates that it must be mediated by a secreted factor that is regulated by Vg.

Nakagoshi et al. presented results that show that the expression of dve is dependent on Dpp and Wg signals (Nakagoshi et al., 2002). As vg is itself regulated by these signals(Williams et al., 1994; Kim et al., 1996; Kim et al., 1997), we think that Vg mediates the effect of these signals on the expression of dve.

We also found that expression of dve at the DV boundary is suppressed shortly after its initiation. We confirm the findings of Nakagoshi et al. (Nakagoshi et al., 2002) that Wg is required for this repression. In addition, we identify Nub as another factor required for the repression of dve expression. Our data suggest that this suppression is important,because we show that forced expression of dve along the DV boundary is deleterious for wing development. One gene affected by the forced expression of dve is wg, which is required for the development of the wing through maintenance of the expression of Vg in pouch cells (Klein and Martinez-Arias,1999). Although the expression of other genes might be also affected, the loss of the expression of Wg is already sufficient to explain the loss of wing development upon forced expression of dve.

The wing imaginal disc is a single-cell layered epithelium and, thus, is a two-dimensional structure. Therefore, establishment and patterning of the PD axis must occur with the help of the existing AP and DV axes. The vggene is an important translator of the positional values of these axes in corresponding PD values. Previous work showed that vg is required for the establishment of distal wing fates (reviewed by Klein, 2001). This work,together with that previously reported, gives insight into how Vg organizes the PD axis.

Previously, it has been shown that Vg is required for the establishment of the medial part of the PW (Liu et al.,2000; del Alamo Rodriguez et al., 2002). During this process Vg induces the expression of rn. Expression of rn is in turn required to set up the inner ring-like expression domain of Wg, which subsequently organizes the formation of the medial part of the PW (del Alamo Rodriguez et al., 2002; Neumann and Cohen, 1996). Our work shows that Vg is further required for the establishment of the distal part of the PW. It shows that one crucial event during this process is the establishment of the expression of dve by Vg. Importantly, Vg induces both parts of the PW in a non-autonomous manner. This indicates that Vg controls the expression of a diffusible factor that induces the expression of genes, such as dve and rn, in cells inside and outside of its expression domain, in order to establish the corresponding regions of the PW. Furthermore, we find that the induction of expression of rn and dve occurs independently from each other. The expression domain of rn is larger than that of dve. Taking for granted that expression of both genes is induced by the same diffusible factor, this observation suggests that it might act in a concentration dependent manner. In this scenario the induction of rn expression would require less activity than the induction of dve.

del Alamo Rodriguez et al. reported evidence that expression of nub is lost in vg-mutant wing imaginal discs(del Alamo Rodriguez et al.,2002), suggesting that Vg is also required non-autonomously for the activation of nub, in a yet larger domain than dve and rn. However, these results are in conflict with earlier work that reports that nub expression is not dependent on Vg function(Klein and Martinez-Arias,1998; Ng et al.,1996). This showed that Wg, but not Vg, is able to induce ectopic expression of nub in the notum of the wing imaginal disc. Furthermore, expression of nub RNA was observed in vg-null mutant wing imaginal discs. These data strongly suggest that Wg is required to activate expression of nub. Hence, further work is necessary to resolve the contradictions, and to determine whether Vg also plays a role during activation of the expression of nub. Despite this uncertainty,all of the mentioned genes are expressed in disc-like domains of different sizes. Their expression leads to concentric areas with different combinations of gene activities. It seems likely that a particular combination of these genes establishes a specific part of the PW (see Fig. 5K).

Our data provide evidence that Vg controls the expression of fj,within an expression domain that corresponds to the wing pouch. Fj is required for the establishment of a proximal region of the wing pouch and also for planar polarity of the wing (Villano and Katz, 1995; Zeidler et al.,2000). Furthermore, Vg regulates the expression of Distal-less (Dll), which is required to pattern the wing margin (Klein and Martinez-Arias,1999). Thus, Vg is involved in the patterning of the PD axis inside as well as outside its expression domain.

It is widely accepted that pattern formation and cell proliferation are closely connected during wing development. However, it has not been clear how these processes are connected. The fact that expression of dve is initiated by one of the central patterning factors, Vg, provides a possible link.

The data presented here, together with recently published work, reveal how patterning along the PD axis might occur with help of the two other existing axes (Fig. 8). Previous work has established that wing development starts at the cross-point of the expression domains of Dpp and Wg in the ventral part of the wing disc(Fig. 8, Step 1)(Klein and Martinez-Arias,1999; Wu and Cohen,2002) (for a review, see Klein, 2002). It appears that the combined activity of the two signals define the wing field. Although the activity of Wg is sufficient to establish the proximal-most pattern elements,the hinge and the proximal region, of the PW(Ng et al., 1996; Klein and Martinez-Arias, 1998a or b?), the establishment of all distal regions requires the additional activity of vg(Klein and Martinez-Arias,1998; Klein and Martinez-Arias, 1999). In the wing field, the Notchsignalling pathway activates the expression of vg in cells at the future compartment boundary (Fig. 8A). In addition, Wg, perhaps in collaboration with Vg/Sd,activates the expression of nub.

In the next step Vg induces the expression of wg in cells at the DV boundary, in collaboration with the Notch pathway(Fig. 8B). In addition, it activates an unknown diffusible factor that induces the expression of dve and rn in disc-like domains of different sizes(Fig. 8B). All these domains are larger than that of Vg, and expression of the three genes is established independently from each other. This fact suggests that the diffusible factor might act in a concentration-dependent manner, as is typical for morphogens. Dve and Rn act in collaboration with Nub to establish the medial and distal parts of the PW.

When the expression of nub, rn and dve is initiated, Vg is expressed in cells at the DV boundary(Fig. 8B). These cells will later form the distal-most structure, the wing margin. The wing pouch is formed by the progenies of cells at the DV boundary, and is therefore intercalated between the margin and the anlagen of the PW(Fig. 8C)(Klein and Martinez-Arias,1999). During its formation, the pouch will be further subdivided through the combined activity of Vg and Wg. Both proteins generate gradients that further subdivide the pouch along the DV axis.

In summary, the data suggest that pattern formation along the PD axis occurs in several steps and uses a similar strategy to that observed during leg development (Galindo et al.,2002; Campbell,2002; Goto and Hayashi,1999). It is initiated by the definition of the proximal (hinge and the distal part of the PW) and the distal-most point (wing margin), with help of the existing AP and DV axes. During development, the intermediate pattern elements (first the anlagen of the medial and distal part of the PW,then the wing blade) are intercalated stepwise with respect to these reference points.

Thanks to S. Lüschnig, J. P. Couso, S. Cohen, F. Katz, S. Carroll, M. Muskavitch and K. Ito for sending stocks and reagents. We would further like to thank M. Kaspar and Robert Wilson for critical comments on the manuscript. Thanks to M. Leptin for various help during the course of the work. The work of T.K. and M.H. is supported by the Deutsche Forschungsgemeinschaft (DFG)through SFB 572.