The klumpfuss (klu) transcription unit in Drosophila gives rise to two different transcripts of 4.5 and 4.9 kb, both of which encode a putative transcription factor with four zinc-finger motifs of the C2H2 class. Zinc-finger 2-4 are homologous to those of the proteins of the EGR transcription factor family. As in the case of the most divergent member of the family, the Wilms’ tumor suppressor gene (WT-1), klu contains an additional zinc finger, which is only distantly related. Loss of klumpfuss function is semilethal and causes a variety of defects in bristles and legs of adults, as well as in mouth hooks and brains of larvae. Analysis of the mutants indicates that klumpfuss is required for proper specification and differentiation of a variety of cells, including the sensory organ mother cells and those of the distal parts of tarsal segments.
In Drosophila, selection of sensory organ progenitor cells (SOP) within proneural clusters (Ghysen and Dambly-Chaudiere, 1989; Romani et al., 1989) depends on a signal network formed by products of the proneural and neurogenic genes (see Campuzano and Modolell, 1992; Ghysen et al., 1993, for reviews). Proneural genes, especially the genes achaete (ac) and scute (sc) of the achaete-scute complex (AS-C), initially confer on the proneural clusters neurogenic capabilities (Cubas et al., 1991); lateral inhibition mediated by the neurogenic genes restricts proneural gene activity to one or a few cells in each cluster, thus determining which cells will eventually follow the neural pathway. Additional proneural genes act in internal sensory organs [atonal (ato; Jarman et al., 1993) and within the anterior margin of the wing asense (ase, Dominguez and Campuzano, 1993)]. The proneural genes encode bHLH proteins (Villares und Cabrera, 1987; Gonzalez et al., 1989; Jarman et al., 1993) which, as ectopic and overexpression experiments show, are functionally redundant (Rodriguez et al., 1990; Dominguez und Campuzano, 1993; Jarman et al., 1993; Hinz et al., 1994).
During the lifetime of a proneural cluster, there is a fixed order of events leading to the selection and specification of SOPs (Cubas et al., 1991). Within a cluster of about thirty cells, approximately six accumulate higher levels of ACHAETE and SCUTE proteins. One of these cells is finally selected to become the SOP and continues to accumulate proneural proteins, whereas the concentration of these proteins in the adjacent cells decreases. Apart from the participation of achaete and scute, very little is so far known about the mechanisms of SOP specification within a proneural cluster. The activity of proneural genes is followed by that of so-called neural precursor genes (Brand et al., 1993), such as deadpan (Bier et al., 1992), couch potato (Bellen et al., 1992) and asense (Brand et al., 1993), which are expressed in most or all neural progenitor cells and control differentiation steps of these cells.
The EGR family of transcription factors is defined by a set of three C2H2 zinc finger DNA-binding domains. At present the family is represented by six members, identified in various species: EGR1-3, hpath133 (also named EGR4), the Drosophila gene stripe (sr) (Lee et al., 1995; Frommer et al., 1996) and the Wilms’ tumor-associated protein WT-1 (Mueller et al., 1991; see Madden and Rauscher, 1993, and Miyagawa et al., 1994 for reviews). WT-1 differs from the other members of the family by having an additional, rather divergent zinc finger domain. Moreover, unlike the other members of the family, WT-1 encodes four proteins by alternative splicing (Haber et al., 1991). Two of the WT-1 proteins bind to the same DNA target sequence as the EGR proteins and, unlikely these, seem to suppress transcription of target genes (Madden and Rauscher, 1993).
Here we describe a new Drosophila gene, which we have called klumpfuss (klu) meaning club-foot, on the basis of leg defects found in mutants. klumpfuss encodes a protein with four zinc finger motifs of the C2H2 type, three of which are homologous to those of the proteins of the EGR transcription factor family while the fourth resembles the divergent zinc finger of WT-1. Mutations in klumpfuss cause a variety of defects, among them loss of bristles and tarsal segment fusion. During bristle development, klumpfuss is required to specify an epidermal cell as an SOP, as well as for bristle differentiation. We show that the leg defects arise because the cells in the distal part of the tarsal segment enter apoptosis, probably due to a failure to correctly specify their fate. The analysis of the bristle and leg mutant phenotypes suggests that like WT-1, klumpfuss also plays a role at the onset of differentation processes.
MATERIALS AND METHODS
Mutants and markers
Table 1 lists the klu alleles and klumpfuss deficiencies used in this study. In addition, the following mutations and deficiencies were used: ase1, sc10.1, ac3 and Hw49c all maintained over an FM7cftz-lacZ balancer. Mutations on the third chromosome were balanced over TM6b, which carries Tubby, allowing one to recognize homozygous larvae and pupae. Double mutant larvae were scored using the markers y and Tubby. P764 (kindly provided by Marc Haenlin) is a P-lacZ insertion in the genomic region of klumpfuss.
To analyse the bristle defects, we used A293.1M3 (Blair et al., 1992), an enhancer trap line that expresses lacZ in all cells of the sensilla; A101.F3, which drives lacZ expression in SOPs shortly after their generation (Boulianne et al., 1991; Huang et al., 1991); and A37, in which the socket cells show selective expression (Huang et al., 1991; Blair et al., 1992). These insertions were each placed in a klu− background carrying the amorphic allele klu212lR51C and the deletion kluXR19, and the wings and thorax of the mutant adults were subjected to X-gal staining. We also used A109, which carries an achaete promoter-lacZ fusion (Martinez et al., 1993). For the analysis of the leg defects, we used as markers P-(odd), an enhancer trap insertion in odd-skipped, which is expressed in the 1st and 5th tarsal segments (Cohen, 1993), and a P-lacZ insertion in the gene disconnected (disco), expressed in concentric stripes in the distal region of the tarsal segments and the tibia (Cohen, 1993).
UAS-l’sc, patched-Gal4 and kluG410, a P-Gal4 insertion in klu, were kindly provided by Uwe Hinz (Hinz et al., 1994). The UAS-GFP insertion was kindly provided by G. Boulianne (Yeh et al., 1995) and recombined onto the kluG410 chromosome. Other genes used in this study are described in Lindsley and Zimm (1992). Flies were cultured, and egg and pupae collections made, under standard conditions.
Antibody and X-gal staining was done according to standard protocols. We used anti-ACHAETE (Skeath and Carroll, 1991), 22C10 (Fujita et al., 1982), 44C11 (Bier et al., 1988), BP104 (Hortsch et al., 1990) and anti-β-galactosidase (Cappel, USA) antibodies. For detection of Green Fluorescent Protein, the FITC filterset was used on the Zeiss Axiophot microscope. The FITC-conjugated secondary antibody were purchased by Jackson Immuno Research Laboratory Inc.). Acridine-orange staining was performed according to Masucci et al. (1990). In situ hybridizations were conducted on whole mounts according to the protocol of Tautz and Pfeifle (1989).
Southern and northern blot analyses were conducted according to standard procedures (Sambrook et al., 1989). For isolation of the genomic clones, two EMBL-4 libraries were screened (Pirotta et al., 1983; M. Noll, for reference, see C. Klambt, 1993). For isolation of cDNA clones, we used the pNB40 embryonic library (4-8 hours) of Brown and Kafatos (1988). Colony screening followed standard protocols (Sambrook et al., 1989). For sequencing, the USB Sequenase kit was used according to the protocol (USB).
To obtain the UAS-klumyctag11B constructs, a 5′ EcoRI site was introduced into CNB4 by PCR of the region extending from the 5′ end of klu up to the unique BglII site with the following primers: 5′GAATTCAAGCTTGAATTCCAATAACGATCGGCGCGT and 3′ATCGCTGCAGATCTGGCA. The amplified fragment was digested with EcoRI and BglII. To introduce the myc tag at the 3′ end, PCR was conducted from the 3′ end of the translated region up to the unique NsiI site in CNB4, with the following primers: 5′AGCAGATGCATCATTC; 3′GAATTGCGGCCGCTTCAGGTCTTCCTCGCTGATCAGCTTCT-GCTCTGTCCAGCAGCGAATGGGAAC (the sequence enconding the myc-epitope is underlined). In addition, the last primer includes a NotI site for convenient cloning. The amplified PCR fragments, together with the central part of CNB4 (BglII – NsiI fragment), were ligated into the EcoRI/NotI-digested pUAST vector (Brand and Perrimon, 1993), the structures were verified by sequence analysis, and the constructs were injected into flies. Several transformants were recovered and tested with various Gal4 activators. Based on this characterization, UAS-klumyctag11B lines were chosen for the rescue experiments.
klumpfuss was first identified by the β-galactosidase expression pattern associated with two lacZ-P-element insertions, kluP212 und kluP819 (A. Beermann, C. Schulz and J. A. Campos-Ortega, unpublished data). In these lines, β-galactosidase is strongly expressed in neuroblasts of the procephalon and of the trunk from stage 9 onwards, as well as in imaginal discs. Homozygotes for kluP212 show a characteristic fusion of tarsal segments (see below), which gives the locus its name, and various defects in the development of the imaginal sensory organs.
Molecular analysis of klumpfuss
The P-element insertion site in kluP212 was cloned by standard plasmid rescue of the P-lacW plasmid (Bier et al., 1989). The 2.8 kb of flanking DNA obtained hybridised to the 68A region and was used to screen two genomic EMBL-4 phage libraries (see materials and methods). A total of 52 kb of genomic DNA around the insertion sites was cloned (Fig. 1A). Southern-blot analysis of genomic DNA from kluP212 and kluP819 suggests that the P-element inserted at the same site in both cases. An 9.6 kb genomic EcoRI fragment including the insertion site was labelled with digoxigenin and used for in situ hybridization to embryonic wholemounts. Since the observed pattern corresponds to the β-galactosidase pattern seen in the insertion lines (see below), this fragment was used to screen the pNB40 cDNA library (Brown and Kafatos, 1988). Three overlapping cDNAs (CNB4, of about 3.8 kb, CNB8, of 2.7 kb and CNB2, of 1.7 kb) were obtained from a total of 400 000 clones screened. The largest, CNB4, hybridised at four different sites distributed over about 40 kb of the genomic walk and revealed that the 5′ part of the transcription unit is missing. The smaller cDNAs (CNB2 and CNB8) have an additional Sst1 site, relative to CNB4. The significance of this polymorphism is not clear at the moment. Sequencing showed that CNB4 includes a large open reading frame (see below). Various genomic fragments and the CNB4 cDNA were radioactively labelled and hybridized to a northern blot prepared with total RNA from the various embryonic stages. Two major transcripts, of 4.5 and 4.9 kb, were detected from early stages on, peaking at about 10-12 hours.
Attempts were made to map restriction enzyme polymorphisms in genomic DNA of kluP212 and klu212lR25, klu819lR5, klu212lR14 and klu212lR40, klu212lR63, klu212lR51C and the cytologically visible deficiency kluXR19. In addition, we also mapped P764, a P-insertion in the klumpfuss region that has no obvious abnormal phenotype when homozygous. The strongest hypomorph klu819lR5 carries a 4.5 kb deletion, which removes the region containing the last, non-coding exon. In situ hybridisations with CNB4 revealed that, while the deletion does not affect the translated region of the transcript, the expression of the gene is strongly reduced (Fig. 1A). This suggests that, in addition to the one exon, regulatory regions are also affected. The amorphic alleles klu212lR63 and klu212lR51C carry deletions with one breakpoint at the insertion site; the position of the second breakpoint could not be determined owing to the presence of repetitive sequences within the cloned genomic region (Fig. 1A). Since no transcripts could be detected with the longest cDNA in hybridizations to klu212lR51C mutant discs, the complete transcription unit must be deleted in this allele, the deletion extending beyond the limits of the cloned region. All the other alleles tested exhibit polymorphisms in the region of the P-element insertion similar to those associated with the the P-insertions in kluP212 and kluP819. However, the size of the fragments used for Southern blot analysis did not allow us precisely to determine whether genomic deletions were produced by the reversion event.
klumpfuss encodes a member of the EGR family of transcription factors
Sequence analysis of the longest cDNA, CNB4, predicts a single open reading frame encoding a protein of 750 amino acids (Fig. 1B). The KLUMPFUSS sequence is characterized by four zinc-finger domains of the C2H2 class; part of the N terminus is negatively charged; the C terminus, including the zinc-fingers, is positively charged. Moreover, the N-terminal region contains glutamine-, histidine- and proline-rich stretches, features found in transcriptional activation and repression domains (Mitchell and Tjian, 1989; Han and Manley, 1993). There are three poly-alanine stretches, two in the N- and one in the C-terminal region. Such stretches are implicated in transcriptional repression (Licht et al., 1990). In addition, there are three poly-asparagine stretches in the N-terminal half and, at its very end, four repeats of a collagen-like triplet GXY, where X is arginine or lysine and Y is always glutamine. The significance of these repeats is unknown. Three putative nuclear localisation sites are found in the protein.
A comparison with the proteins in the SWISS-PROT, PIR(R) and GenPept databases shows that the zinc fingers 2-4 of KLUMPFUSS have a high degree of similarity to the zinc-finger domains of the members of the EGR family (Fig. 2A). As in WT-1, KLUMPFUSS contains an additional zinc finger 1 which is only distantly related to those of the EGR proteins. Besides WT-1, KLU is the only other member of the Family with four zinc fingers. Closer comparison of fingers 2-4 of KLUMPFUSS with those of the other EGR-like proteins revealed complete conservation of the amino acids that contact the DNA-binding consensus sequence (arrows in Fig. 2A; Pelletier et al., 1991; Pavletich and Pabo, 1991; Nardelli et al., 1992). Furthermore, the aspartic acid that follows the first arginine contacting the target sequence, which is conserved among EGR proteins, is also conserved in KLUMPFUSS (squares in Fig. 2B). This aspartic acid in the third finger of WT-1 is crucial for its binding capacity; it is thought to stabilize the binding of the preceding arginine to a guanine base in the target sequence (Pelletier et al., 1991).
Expression pattern of klumpfuss during imaginal disc development
Hybridization of klumpfuss cDNAs to Drosophila embryos and imaginal discs reveals a complex and dynamic expression pattern throughout much of larval and imaginal development. Here, we focus solely on the klumpfuss expression pattern in the imaginal discs, since it is relevant to the phenotypic defects described further below. We found klumpfuss transcripts in every disc examined in a pattern that is identical to that of β-galactosidase expression of kluP212 and kluP819 and that produced by kluG410-driven expression of an UAS-GFP fusion (Fig. 3).
Expression starts in the wing imaginal disc within the prospective wing area early in the third larval instar. Shortly thereafter, expression becomes restricted to the prospective margin and the hinge of the wing and, at about the same time, transcripts appear in the anlagen of notum and scutellum (Figs 3A, 4B). Double staining with ACHAETE antibody reveals that klumpfuss is expressed in most proneural clusters at, or shortly after, the onset of ACHAETE expression (Fig. 4A-C); furthermore, simultaneous staining of SOPs with A101 (Boulianne et al., 1991) reveals that klumpfuss expression in these regions precedes the appearance of SOPs (Fig 3H; Fig. 4E,F). lacZ expression in kluP212 is spotty in the wing (and leg, see below) imaginal discs, with zones of higher density surrounding areas lacking expression (Fig. 3G,I). These areas correspond to the SOPs as shown by double stainings with A101 (Fig. 3H), and the spotty pattern is not seen in imaginal discs of In(1)sc10-1 flies, which lack SOPs (Fig. 4G). We also do not detect expression of lacZ until the onset of the differentiation of the sensillum cells (Fig. 4I-J). Thus, while cells of the proneural clusters express klumpfuss, SOPs themselves do not. Since klumpfuss expression is rather uniform before the appearance of SOPs, the gene must be switched off in the cells that initiate neural development.
Expression in the leg discs starts early during the third larval instar. At this time, the klumpfuss expression domain occupies a wedge-like sector encompassing roughly one third of the circumference of the leg disc. Rings of expressing cells successively become visible underneath a knob-like central structure during the third larval stage (data not shown). The rings correspond to the anlagen of the leg segments and the order of their appearence reflects the developmental pattern of the leg disc (Schubiger, 1974; Fristrom and Fristrom, 1975; Norbeck and Denburg, 1991). Around the time of puparium formation, klumpfuss expression seems to be restricted to the distal half of each leg segment in concentric domains (Figs 3D, 4D), spreading later over the whole leg (6 hours after puparium formation). Expression in the antennal and dorsal prothoracic discs also occurs in concentric domains (Fig. 3B). In the eye disc, expression starts behind the morphogenetic furrow and extends through the whole anlage (Fig. 3B). klumpfuss is further expressed in the parts that form the head capsule (Fig. 3B). Expression in the larval brain is restricted to the neuroblasts and the proliferation zone of the optic lobes (Fig. 3C).
klumpfuss is involved in a variety of developmental processes
Flies homozygous for the insertions kluP212 and kluP819 are viable, but show loss of bristles at some positions and fusion of tarsal segments. Excision of the P-element insertions in both lines was attempted using either the Δ2-3 transposase (Robertson et al., 1988) or X-rays. In the first case, a total of 200 revertants were analysed, 170 of which were wild-type, indicating precise excision of the P-elements. In the remaining 30, phenotypic defects were observed in homozygous flies or in combination with the original insertion; all 30 mutants fell into one complementation group. Twelve of these latter excisions were analysed in detail (see M&M). For the X-ray reversion screen, an isogenic strain of kluP212 was used. Several alleles were recovered, among them the two deficiencies Df(3L)kluXR17 (67E; 68B) and Df(3L)kluXR19 (68A). The cytogenetic analysis of all these rearrangements, as well as Df(3L)vin2 (67F; 68D6), Df(3L)vin5 (68A2-3; 68F3-6), Df(3L)lxd2 (68A2-3; 68C5-7) and Df(3L)lxd8, together with the results of complementation crosses between the deficiencies and the amorphic allele klu212lR63, localized the gene at 68A1-2.
Even the strongest alleles are semilethal when homozygous, some animals developing to adulthood but dying shortly after hatching. However, most of the homozygotes die at the end of the third instar. No defects were detected in either cuticle preparations or antibody staining with 22C10, 44C11 or BP-104 of embryos homozygous for the smallest deficiency kluXR19 or for other strong alleles. Among the mutant larvae, defects were detected in the mouth-hooks, where some teeth were missing, and in the larval brain, the morphology of which is obviously abnormal (not shown).
klumpfuss is required to specify epidermal cells as SOPs and for bristle differentiation
We found that a number of macrochaetae are missing in head and thorax, particularly from the anterior margin of the wing, the wing veins, antennae and legs of homozygotes for all of the alleles (Figs 5, 6, 8). Non-innervated bristles at the margin of the alula are also affected. In addition, differentiation defects with incomplete penetrance were observed at some bristle positions (Fig. 6). To assess the phenotypic variability, 19 machrochaete positions on head and notum were studied in about 50 flies from each of the crosses between 13 different klumpfuss alleles and Df(3L)vin2. We found that weak alleles are more variable in their phenotype than strong alleles; moreover, penetrance is position dependent: only one of the 19 bristles considered in this analysis, the anterior sternopleural bristle, is consistently missing in these individuals. However, in each mutant individual, a significant number of bristles is missing. The proximal costa is most frequently affected, the number of bristles there always being severely reduced (Fig. 6A-D). Therefore, due to the characteristic location and penetrance of the phenotypic defects, we concentrated our analysis on this position.
Using various markers (see Fig. 6A-D and Materials and methods), two different defects can be distinguished. In most cases, the lack of bristle apparatus is correlated with a lack of all the bristle cells (Fig. 6A-D). As mentioned, in strong klumpfuss mutants, the number of bristles in the proximal costa is strongly reduced; we found a strong reduction in the number of A101-positive cells in this region at 26 hours apf, i.e. around the time when, in wild type, most SOPs are detectable (Fig. 6E,F). In the wild type, three recurved bristles are present in the proximal region of the costa; in all klu212lR51C / kluXR19 animals, there are only two of these bristles and only two A101-positive cells are visible at this position. Some 70% of the klu212lR51C / kluXR19 animals lack the presutural bristle; the corresponding A101 cells are absent in the majority of the imaginal discs stained (not shown). The same phenotype is detected in kluG410/ klu819lR5 and kluG410/ klu212lR51C heterozygotes.
In some cases, e. g., at the dorsocentrals and in the proximal region of the costa, remnants of the bristle apparatus are still visible at higher magnification; in these positions, bristle cells are detectable with various markers (see Fig. 6G-I). This indicates defects in cyto-differentiation of the cells of the sensillum. The results therefore suggest that, in some positions, klumpfuss is required in order for epidermal cells to develop as SOPs and, in other positions, for proper differentiation of the progeny cells.
klumpfuss is required for cell differentiation in the leg imaginal discs
In klumpfuss mutants, the distal regions of the leg segments are preferentially affected. In homozygotes for hypomorphic alleles, tarsal segments 3-5, on the one hand, and trochanter and femur, on the other, are fused in all three leg pairs. In homozygotes for severe alleles, the defects are stronger and partial fusions of the first tarsal segment to the tibia and/or the second tarsal segment occur occasionally (Fig. 5A-F). There are fewer sex comb bristles and first tarsal segment bristles in homozygotes for weak alleles; in those bearing severe alleles, the first tarsal segments, including the sex combs, are lost (Fig. 5E-F). Bristles characteristic of the proximal part of other segments are still present, although reduced in number and sometimes showing cytodifferentiation defects (Fig. 5C-F).
We detected the first visible leg defects shortly after evagination of the discs (0-1 hour apf), i.e., when tarsal segments become normally recognisable (Fristrom and Fristrom, 1975). The use of various P-lacZ insertions as markers, helps to analyse the occuring defects: kluP212, odd-skipped, expressed in the 1st and 5th tarsal segments, and disconnected (disco), expressed in concentric stripes in the distal region of all tarsal segments and the tibia (Cohen, 1993), allow one to distinguish the divisions of the leg disc (see Fig. 7). In leg discs of homozygous larvae, one can still discriminate remnants of the tarsal segments with those markers (Fig. 7B,C,E), indicating that the initial subdivision of the leg segments takes place. Especially the disco expression pattern is of particular use to monitor the defects, since it is expressed in the distal region of all tarsal segments and the tibia (Cohen, 1993; see Fig. 7A). In late third instar, the tarsal segments seem to be present and one can detect all concentric rings of disco-expression (Fig. 7C). However, slight defects in the expression pattern are detectable (arrow in Fig. 7C). Later, approximately 1 hour apf, the expression pattern is severely disturbed and the morphology of the tarsal region becomes very abnormal. At this time, we observe massive cell death in this region (Fig. 7F,G).
All these data indicate that klumpfuss is not involved in the proximal-distal pattern formation of the leg disc, rather it is required for the differentiation of the distal tarsal region. In addition to the leg defects, the size of the basal cylinder of the antenna, which is homologous to the tarsal segments (Postlethwait and Schneidermann, 1971), is also reduced (Fig. 5I-J). However, we did not analyse the development of the defects in the antenna.
Most traits of the klumpfuss loss-of-function phenotype is rescued by the CNB4 cDNA
We conducted a rescue experiment using the Gal4 system (Brand and Perrimon, 1993) to verify that the cloned transcription unit is klumpfuss. As activator, we used the Gal4 insertion kluG410, which drives lacZ expression in a pattern essentially identical to that of β-galactosidase in kluP212, although somewhat patchy. It induces a strong klumpfuss phenotype at 17° C, which is not complemented by any of several klu alleles tested. Activation of UAS-klumyctag11B in either a kluG410 / klu819lR5 or a kluG410 / klu212lR51C background leads to complete rescue of the bristle phenotype and partial rescue of the leg phenotype (Fig. 8A-F). The incomplete rescue of the leg phenotype is most probably due to the patchy activation of UAS-klumpfuss by kluG410 (Fig. 3F). The effects of ectopic expression of klumpfuss will be described elsewhere.
Phenotypic interactions reveal functional relationships of klumpfuss with the proneural genes
To detect possible functional interrelationships with the proneural genes, we combined klumpfuss mutations with mutations of the achaete-scute complex (AS-C, Campuzano and Modolell, 1992). In(1)ac3 animals lack most microchaetae and several macrochaetae in notum and legs (Fig. 9A; Lindsley und Zimm, 1992). This phenotype is strongly enhanced in flies homozygous for the severe allele klu212lR63, and less so in homozygotes for the hypomorph kluP212. Synergistic effects are obvious in the leg of these animals (Fig. 9C). The mutation Hw49C causes strong ectopic expression of ACHAETE and SCUTE (Skeath und Carroll, 1991), which leads to development of ectopic bristles. Thus, in Hw49C mutants, multiple bristles develop instead of the single anterior sternopleural bristle. In Hw49C;kluP212 double mutants no bristles develop at the anterior sternopleural position. Hence, kluP212 is epistatic to Hw49C.
ase1 is a 19 kb deletion that removes the asense gene (Brand et al., 1993; Dominguez und Campuzano, 1993). ase1 flies have a reduced number of bristles on anterior wing margin and abdomen; in addition, some of the wing margin bristles exhibit differentiation defects (Fig. 9E). In klu mutants, we observe a slight reduction in the number of bristles at the wing margin (Fig. 9D), which is strongly enhanced in double mutant combinations (Fig. 9F and G). Thus, ase1;klu212lR51C flies lack almost all bristles at the anterior wing margin and abdominal microchaetae are missing (Fig. 9). Since these defects have never been observed in the single mutants, a close functional relationship between klumpfuss and asense is probable. Interestingly, asense has a proneural function in the anterior wing margin but not in other regions of the body.
The phenotypic interactions described suggest functional interrelationships between klumpfuss and the AS-C. The epistatic relations between Hw49c and kluP212 suggest that klumpfuss requires the activity of achaete and scute for its role in bristle development. Therefore, we examined the β-galactosidase expression of kluP212 in achaete and scute mutants, and vice versa.
Ectopic expression of ACHAETE and SCUTE in Hw49c is particularly strong along the second wing vein (Skeath und Carroll, 1991). Ectopic β-galactosidase expression was found in the territory of the second wing vein of Hw49c;kluP212 imaginal discs. We next expressed UAS-lethal of scute driven by patched-Gal4 (Hinz et al., 1994) and tested whether klumpfuss is ectopically expressed. Under these conditions, expression of lethal of scute in wild-type background induces bristle development within the patched domain (Hinz et al., 1994). We found ectopic activation of klumpfuss to be restricted to the patched expression domain in the wing disc and absent from the other regions. This indicates that lethal of scute is able to activate klumpfuss expression, but apparently only in the wing region.
In In(1)sc10-1 flies, where no ACHAETE or SCUTE protein is detectable (Skeath and Carroll, 1991) and all innervated bristles of head and thorax are missing (Garcia-Bellido, 1979), the overall klumpfuss expression pattern in imaginal discs is normal. However, we found that a few expression domains are weaker or absent in the wing region (arrows in Fig. 5G). Conversely, no abnormality in the ACHAETE expression pattern could be detected in klu mutants (data not shown).
These results indicate a weak influence of AS-C genes on klumpfuss expression, which is restricted to the wing area of the wing disc. However, the overall expression pattern of klumpfuss is largely independent of proneural genes. Consequently, the functional relationship revealed by the genetic interactions, is likely to occur at a post-transcriptional level.
klumpfuss is a member of the EGR family of transcription factors
Sequence analysis predicts a protein of 750 amino acids with four zinc-finger domains. Sequence comparison revealed a high homology of zinc fingers 2-4 with those of the members of the EGR family. All members of this family bind the DNA consensus sequence GCGC(G/T)GGGCG. Although DNA-binding data are still lacking, it is likely that KLUMPFUSS also binds to this sequence, since all amino acids that make contacts with the EGR target sequence, as well as the aspartic acid following the first arginine contacting the target sequence of zinc finger 3 of WT-1, are conserved in KLUMPFUSS. This residue is essential since exchanges at the aspartic acid position abolish the DNA-binding capacity of WT-1 (Pelletier et al., 1991). Interestingly, an aspartic acid at the corresponding position in the other two zinc fingers is completely conserved among the EGR proteins, WT-1 and KLUMPFUSS (see Fig. 2A). Like WT-1, KLUMPFUSS has an additional, fourth zinc finger which is unrelated to those of other members of the family (Call et al., 1990; Gessler et al., 1990). However, it seems improbable that KLUMPFUSS represents a true orthologue of WT-1, since WT-1 homologues in vertebrates (Larson et al., 1995) and KLUMPFUSS share no other significant homology outside the zinc finger region.
So far only two other proteins are described as EGR-like in invertebrates the Drosophila genes stripe (sr) and huckebein (hkb, Broenner et al., 1994). However, whereas the zinc finger of STRIPE have the characteristic amino acids for the binding of the DNA consensus sequence, in the zinc fingers of HUCKEBEIN four out of the six amino acids are not conserved. Therefore, from these data, it appears improbable that HUCKEBEIN binds to the same target sequence as the EGR proteins and is probably a more distantly related protein. Outside the zinc-finger region all proteins of the EGR family are different; thus, on the basis of the sequence, it is not possible to decide about whether KLUMPFUSS acts as a transcriptional activator, like the EGR proteins, or as a repressor.
klumpfuss is involved in initial steps of differentiation processes
Certain bristles and the corresponding A101-positive (SOP) cells were frequently missing in klumpfuss mutants. One may of course ask whether lack of A101 expression indeed signals the lack of the SOP, and whether this is due to functional deficits of the genes of the AS-C in klumpfuss mutants. If loss of klumpfuss function were to lead to functional defects of AS-C gene function, commitment of epidermal cells as SOPs would not occur. However, single cells expressing ACHAETE, i.e., with the characteristics of SOPs, were indeed present in the proneural clusters of klumpfuss mutants. Therefore, SOP development appears to be initiated correctly. The data on klumpfuss expression in flies mutant for the AS-C point to a role for klumpfuss in early steps of bristle development, but downstream of the proneural proteins. We assume that the cells that would normally have become SOPs switch fate before they activate A101 expression and probably die. The assumption that SOPs enter apoptosis is supported by the observation of abundant cell death in other developing organs of klu mutants, like the legs. All these data thus suggest that at certain bristle positions, such as that of the anterior sternopleural, klumpfuss is required during early bristle development immediately after proneural gene function, in order to allow a particular epidermal cell to develop as a SOP.
Our data imply that klumpfuss activity is restricted to the time of SOP specification, expression being switched off when the cell initiates SOP development. Since klumpfuss is apparently not expressed in bristle cells later in development, the gene has an influence on cells in which it is apparently not expressed. Such a delayed effect is also found for proneural genes. Dominguez and Campuzano (1993) showed that bristle differentiation defects at the anterior wing margin in asense mutants can be rescued by duplications of the achaete and scute genes. These latter genes are, however, switched off in the nascent SOP (Cubas et al., 1991); consequently, the effect on asense has to be indirect. This implies that a differentiation program of which klumpfuss takes part is initiated during specification of the SOP, and is needed for the proper differentiation of the SOP and the cells that it generates. In certain bristle positions, e.g. the dorsocentrals, all bristle cells are still detectable with specific markers, but only remnants of the bristle are present. Granted that klumpfuss expression probably occurs in all proneural clusters that give rise to bristles, its role obviously varies in importance depending on the site. It might nevertheless act as a general factor in bristle development, both at intitial steps of SOP specification and during differentiation of its progeny cells.
Following overexpression of proneural genes, supernumerary bristles arise within the regions in which bristles normally develop in the wild type. Rodriguez et al. (1991) interpreted this observation as a manifestation of spatially restricted competence of the imaginal discs to develop sensory organs. That is to say, imaginal disc cells require a certain competence in order to respond to proneural gene activity. klumpfuss expression domains include most, or all, proneural clusters that give rise to bristles in the imaginal discs; however, with the exception of the wing area, its expression seems to be largely independent of proneural gene activity in the imaginal discs. It is tempting to speculate that klumpfuss may act to confer on the imaginal neurogenic regions the competence to respond to proneural gene activity.
With respect to the leg defects, similar conclusions can be drawn. Prior to, and at the time when the defects arise in the mutants (0-1 hpf), expression of klumpfuss is confined to the distal region of each leg segment. klu mutants lack distal structures in tarsal segments, which are frequently fused. These defects are visible as early as the time at which the tarsal segments become recognizable and, at this time, clusters of apoptotic cells are seen. Therefore, although the initial subdivision of leg segments takes place, the cells in the distal part of the tarsal segments eventually enter apoptosis. On this basis, we propose that klumpfuss is required for the onset of differentiation of the distal part of the tarsal segments. In addition, klumpfuss is also required for bristle development within proximal regions of the leg discs.
We thank Bill Chia and Samy Bahri for help with sequencing, and Susan Rolfe for excellent technical assistance in some experiments. Further, T. K. would like to thank Dr Alfonso Martinez-Arias for help during the end of the work. This work was supported by the Deutsche Forschungsgemeinschaft (SFB 243), a grant of the European Union (No. CI1*-CT92-0014), and the Fonds der Chemischen Industrie.