The bric à brac (bab) locus acts as a homeotic and morphogenetic regulator in the development of ovaries, appendages and the abdomen. It consists of two structurally and functionally related genes, bab1 and bab2, each of which encodes a single nuclear protein. Bab1 and Bab2 have two conserved domains in common, a BTB/POZ domain and a Psq domain, a motif that characterizes a subfamily of BTB/POZ domain proteins in Drosophila. The tissue distribution of Bab1 and Bab2 overlaps, with Bab1 being expressed in a subpattern of Bab2. Analysis of a series of mutations indicates that the two bab genes have synergistic, distinct and redundant functions during imaginal development. Interestingly, several reproduction-related traits that are sexually dimorphic or show diversity among Drosophila species are highly sensitive to changes in the bab gene dose, suggesting that alterations in bab activity may contribute to evolutionary modification of sex-related morphology.
As every newcomer to a fly laboratory learns, the easiest way to distinguish between Drosophila melanogaster (Dm) females and males is to look for the sex-specific pigmentation pattern of the posterior abdomen, the male-specific sex combs on the forelegs and/or the genital structures. These morphological features not only constitute sexually dimorphic traits but show considerable interspecific variation among members of the Dm species group and subgroup, and some of these features also show intraspecific variation. Concerning sex combs, variation is observed in the number of tarsal segments that display sex combs, their number per segment, and the number of sex comb bristles (Bock and Wheeler, 1972; True et al., 1997; Nuzhdin and Reiwitch, 2000; Macdonald and Goldstein, 1999). In Dm males, a sex comb is only found on the most proximal or first tarsal segment (TS1), and consists of a single row of approximately ten sex comb bristles.
Another sexually dimorphic trait in Dm is the pigmentation of the tergites of the posterior abdomen. Whereas an alternating dark-light stripe pattern is characteristic of all abdominal tergites in females, the posterior-most two tergites in males are darkly pigmented throughout. Sexually dimorphic abdominal pigmentation is not characteristic for all Drosophila species, however (Kopp et al., 2000), and a large diversity in abdominal pigmentation is seen within the genus Drosophila (Ashburner, 1989; Hollocher et al., 2000; Eisses and Santos, 1997). Even within the Dm species subgroup, in which the abdominal pigmentation pattern is rather homogeneous, a newly discovered species, D. santomea lacks any abdominal pigmentation (Lachaise et al., 2000). In addition, considerable intraspecific variation in the pigmentation of posterior tergites in females has been described for different wild-type populations of Dm (Robertson et al., 1977; Eisses and Santos, 1997). Together, these findings indicate a rapid phylogenetic divergence of sex comb and abdominal pigmentation patterns.
The genetic and developmental mechanisms that allow for fast evolutionary modifications of characteristics that are connected to reproduction/mating (secondary sexual traits) and may be involved in speciation are largely unknown. A strong modifier locus of abdominal pigmentation in wild-type populations of Dm was mapped to the distal arm of the third chromosome, and is called fap (female abdominal pattern) (Robertson and Riviera, 1972; Robertson et al., 1977). fap is very likely to correspond to the bab locus that maps to the same chromosomal position, and was recently shown to be involved in the regulation of abdominal pigmentation and morphology (Kopp et al., 2000). The correlation between the expression pattern of bab and the abdominal pigmentation pattern in species of the Dm species group suggests that divergence in the regulation of bab expression may have contributed to the variation of abdominal pigmentation in this species group (Kopp et al., 2000).
The bab locus is also a homeotic regulator of the sex comb pattern (Godt et al., 1993). Mutations in bab cause a transformation of the bristle pattern in the tarsal segments, leading to the appearance of ectopic sex combs in distal tarsal segments. The graded requirement and graded expression pattern of bab along the proximodistal axis of the tarsus suggest that the concentration of Bab specifies the identity of the bristle pattern in tarsal segments in a dose-dependent manner. Moreover, bab plays a second role in leg development, which appears to be independent of its function in the specification of the bristle pattern. bab mutations cause a failure in joint formation, resulting in the fusion of tarsal and antennal segments (Godt et al., 1993). The phenotype of bab mutants shows that this locus is also an important morphogenetic regulator of ovarian development. During ovarian development, bab mediates the formation of the terminal filaments, stacks of cells that are required for the formation of ovarioles (King, 1970; Godt and Laski, 1995; Sahut-Barnola et al., 1995) and contribute to the regulation of germline and follicle stem cell divisions during oogenesis (Lin and Spradling, 1993; Forbes et al., 1996).
We show that the bab locus is composed of two paralogous genes, bab1 and bab2. The corresponding proteins Bab1 and Bab2 belong to a subfamily of BTB/POZ domain (BTB domain)-containing nuclear proteins that have a Psq-domain but no Zn-finger motif. Although partially redundant in function, our data indicate that both genes are required for normal development. bab2 appears to play a predominant role in ovarian and particularly in leg development, which points to a functional divergence of the two Bab genes.
MATERIALS AND METHODS
The bab alleles babE1, babP, babPR72, babPRDS, babPR24, babA128 and Df(3L)babPG are described elsewhere (Godt et al., 1993). babAR07 was isolated in a P-element-mediated excision mutagenesis that was conducted for babA128 (=P[ry+, lacZ]A128) similar to that described for babP (Godt et al., 1993). The mutations babE3, babE4, babE5 and babE6 were isolated in an EMS mutagenesis using an ri e mutant background similar to that described by Godt et al. (Godt et al., 1993). Lethal hits on the chromosomes that carried the bab mutations were removed using an ecd st red e ca chromosome for recombination. The deficiencies Df(3L)Fpa1 (61D1;61F1-2) and Df(3L)Fpa2 (61F1;61F4) are γ-ray induced. The allele babD1 was a gift from K. Burtis. TM6B, Tb and genetic markers are described elsewhere (Lindsley and Zimm, 1992). Oregon-R was used as a wild-type stock.
Flies used for phenotypic analysis of the abdomen were free of genetic markers that affect the body color, such as y or e. In flies that carry the babE1, babE5 and babP allele, respectively, the e marker was replaced by an e+ allele using an Oregon-R chromosome for recombination. Cuticles of abdomina and legs were mounted in Hoyer’s medium (Ashburner, 1989). Flies were raised at 25°C if not indicated otherwise.
Molecular analysis of the bab locus
Genomic DNA from the babP line was used to clone a TaqI fragment adjacent of the PlacZ insertion site by inverse PCR (Godt et al., 1993). This fragment was used to establish a genomic walk of 140 kb of DNA that covers both bab genes, using genomic lambda phage (Maniatis et al., 1978) and cosmid libraries (Tamkun et al., 1992). The orientation of the bab walk and the molecular breakpoints of deficiencies were defined by chromosome in situ hybridization using genomic DNA probes. The structures of the bab1 gene and transcript were deduced from genomic sequence in combination with an RT-PCR analysis. The 5′ and 3′ ends of the terminal exons were identified by RACE-PCR analysis using polyA+ RNA from third instar larvae. Overlapping RT-PCR fragments were used to reconstruct a complete bab1 cDNA (AJ252082). The BTB-II domain identified previously (Zollman et al., 1994) was shown to be part of bab2. Sequence and Southern blot analysis, using bab2 genomic DNA and cDNAs led to the determination of the structure of the bab2 gene and transcript (AJ252173). Close to full-length bab2 cDNAs were isolated from an 8- to 12-hour-old embryo library (provided by N. Brown). The longest bab2 cDNA isolated, 5.1 kb in size and named cDNA2, corresponds to the predicted bab2 transcript, except that it retains the 70 bp intron 3. This intron is not present in a second cDNA (cDNA4) analyzed, nor was it found when mRNA from embryos and adult ovaries was analyzed by RT-PCR using primers that flank the intron, suggesting that the retention of intron 3 in cDNA2 is an anomaly, and the 70 bp intron is normally removed by splicing. For Northern blot analysis, total RNA was prepared from third instar larvae with the SDS-hot phenol-chloroform technique (Palmiter, 1974), and PolyA+-RNA was selected via oligo-dT.
New cDNAs of the tkr gene were isolated from a 0- to 24-hour-old embryonic cDNA library (Palazzolo et al., 1990).
Generation of UAS-bab2 and UAS-bab1 transgenic flies
The bab2 cDNA2 and cDNA4, both of which contain the entire open reading frame of bab2 (5′ and 3′ untranslated regions are incomplete in cDNA4) were subcloned into the pUAST vector (Brand and Perrimon, 1993). More than 20 independent transgenic lines were established for each construct. UAS-bab2 cDNA2 and cDNA4 transgenic lines produced comparable mutant phenotypes. A Hsp70-Gal4 driver line, carrying the P[Gal4-Hsp70.PB] construct on the second chromosome (Brand et al., 1994), was used for ubiquitous expression of bab2.
To generate a UAS-bab1 transgene, a bab1 minigene was constructed as follows. A 2.7 kb PstI-SalI genomic fragment, starting at a PstI site 6 bp upstream of the ATG and ending at a SalI site 422 bp after the 5′ splice site of intron 1, and a 4.5 kb BamHI genomic fragment that contains all of the other exons from bab1, starting 146 bp upstream the 3′ splice site of exon 1 and ending 273 bp after the end of the bab1 cDNA, were fused via blunt-ended SalI and BamHI sites and cloned into the pUAST vector. Five independent transgenic lines were established.
Production of polyclonal anti-Bab2 antibodies
A 1.3 kb BamHI fragment (called 1.3B), which encodes amino acids 126 to 558 of Bab2, and a 1.6 kb XmaI fragment (called 1.6X; amino acids 470 to 1014 of Bab2) were subcloned into the pGEX-1N and pGEX-3X protein expression vectors (Pharmacia-Amersham), respectively. Fusion protein 1.3B-GST was soluble and extracted as described by Godt et al. (Godt et al., 1993). Fusion protein 1.6X-GST was insoluble, and was purified from inclusion bodies using urea (Grisshammer and Nagai, 1995). The fusion proteins were used for immunization of rats at Pocono Rabbit Farm and Laboratories.
The Bab2 antibodies R6 and R7 are directed against the 1.3B-peptide that includes the BTB domain of Bab2. To test whether these antibodies are specific for Bab2, their ability to recognize the BTB domain of Bab2 and Bab1 was tested by immunoblot analysis (data not shown). Both antibodies, diluted 1:4000 recognized the BTB domain of Bab2 (0.1 μg of protein; a gift of G. Privé) but did not recognize the BTB domain of Bab1 (0.5 μg of protein). As the BTB domain is the only conserved domain between Bab1 and Bab2 in this peptide, we believe that the R6 and R7 antibodies are Bab2-specific. The Bab2 antibody R10 is directed against the 1.6X peptide that includes the BabCD with the Psq motif (Fig. 2C) The Bab2 specificity of this antibody was demonstrated indirectly by showing that the R10 antibody does not recognize a protein in bab2 mutant tissues that are homozygous for babE1 (data not shown).
Immunoblotting and tissue staining
Imaginal discs and brains were dissected from white prepupae in phosphate-buffered saline (PBS), pH7, transferred to 2× Laemmli buffer (Sambrook et al., 1989) on ice and homogenized. Protein samples containing tissues from two prepupae were boiled for 5 minutes and loaded onto a 6% polyacrylamide gel (20 μl/lane). To identify the Bab proteins on western blots, α-Bab2-R6 (1:20,000) and R7 (1:30,000) were used as primary antibodies, and α-rat HRP-conjugated antibodies (1:1000; Jackson Laboratories) and the ECL-solution (Pharmacia-Amersham) were used for signal detection. β-Tubulin was detected with mAB-E7 (1:250; DSHB).
Dissection of larval tissues, tissue in situ hybridization and immunostaining were done as described by Godt et al. (Godt et al., 1993), with some modifications. For in situ hybridization, the first fixation was in 10% formaldehyde in PBS for 20 minutes, and hybridization took place at 50°C for 48 hours. For immunostaining, larval tissues were fixed in 5% formaldehyde in PBS for 10 minutes. Primary antibodies were α-Bab1-r2 (1:1000) (Godt et al., 1993) and α-Bab2-R7 or R10 (1:4000). α-rat secondary antibodies were Biotin-conjugated (1:400; Jackson Laboratories).
The bab locus consists of two structurally related genes
bab was originally identified by the babP allele, a P-element insertion mutation that maps to polytene band 61F1-2 (Godt et al., 1993). An open reading frame (ORF) adjacent to the babP insertion identified a BTB domain, a domain found primarily but not exclusively in Zn-finger containing transcription factors (Collins et al., 2001). In addition to bab, a second gene encoding a BTB domain (called BTB-II) was identified at 61F1-2 and found to have an expression pattern similar to bab (Zollman et al., 1994). We show that bab and BTB-II constitute a gene complex. For clarity, the gene previously called bab will be renamed bab1, and BTB-II will be renamed bab2. The bab locus or bab will refer to both genes together. The structure of the bab locus is shown in Fig. 1, and the structure of the Bab1 and Bab2 proteins is shown in Fig. 2.
The bab1 and bab2 genes have the same orientation and show several structural similarities (Fig. 1), suggesting that they are the result of a chromosomal duplication. bab1 and bab2 span approximately 60 kb and 25 kb of genomic DNA, respectively. The insertion point of babP is located in intron 1 of bab1, 236 bp downstream of the 5′ splice site. Both genes have four introns, three of which are at homologous positions in the coding region. Sequence analysis of the bab1 and bab2 transcript predicts proteins of 967 and 1067 amino acids, respectively (Fig. 2A). Bab1 and Bab2 have two evolutionarily conserved domains in common. Outside of these domains Bab1 and Bab2 show only low sequence similarity to each other and no significant similarity to other proteins. A BTB domain, a conserved domain of 115 amino acids, is found in the N-terminal region of both proteins (Fig. 2A). In contrast to most known BTB domain-containing proteins that are transcriptional regulators, the BTB domains of Bab1 and Bab2 do not start within the first few amino acids of the predicted proteins but at amino acids 90 and 195, respectively. The Bab1 and Bab2 BTB domains are the most closely conserved domains within the BTB family (Fig. 2B) (Zollman et al., 1994).
In contrast to many other known BTB domain-containing proteins that are transcription factors, neither Bab1 nor Bab2 contains a Zn-finger motif. However, these proteins have a second domain in common that we call BabCD for Bab-conserved domain (Fig. 2C). In both proteins, this domain is encoded by three exons with the splice sites at homologous positions. The BabCD contains two known motifs, a Psq domain and an AT-hook like motif. The Psq domain can mediate DNA binding, and is named after pipsqueak (psq), another BTB domain-encoding gene (Weber et al., 1995; Horowitz and Berg, 1996; Lehmann et al., 1998). The Psq domain, which is 48 amino acids in length, is 97% identical between Bab1 and Bab2, and is the region of highest similarity within the BabCD (Fig. 2C). A Psq domain has also been identified in the Tkr protein (Haller et al., 1987) and in Piefke (CG15812) (Schwendemann et al., 2001). Piefke also has a BTB domain. The previously published sequence of Tkr lacks a BTB domain, however newly isolated cDNAs for the tkr gene show that it contains the BTB-III domain (Accession Number AJ252174), which we previously found to map to the same chromosomal location and to display the same expression pattern as tkr (Zollman et al., 1994). Therefore, all five genes known to encode proteins with a Psq domain in Drosophila also have a BTB domain (Fig. 2D).
The C-terminal region of the BabCD contains an AT-hook-like motif (Fig. 2C). It has the invariant peptide core motif R-G-R-P flanked on either side by other positively charged amino acids but does not correspond to any known AT-hook variant identified so far. AT-hook motifs have been shown to mediate binding to the minor groove of stretches of AT-rich DNA (Reeves and Nissen, 1990). Whether the AT-hook-like motif in Bab1 and Bab2 is involved in DNA-binding remains to be investigated. Both Bab proteins also have a short motif that is very rich in glutamine and histidine residues (Fig. 2E), which could aid transactivation (Courey and Tjian, 1988; Gerber et al., 1994). The domain architecture of Bab1 and Bab2, together with their nuclear localization (Godt et al., 1993; Godt and Laski, 1995), suggest that these proteins function as transcriptional regulators.
bab1 and bab2 have overlapping expression patterns
During embryogenesis, bab2 is zygotically expressed in a complex pattern that has been described previously (Zollman et al., 1994), whereas bab1 is not expressed at a detectable level. bab seems to have no essential function during embryonic development as even mutants that lack both bab genes are not embryonic lethal.
During post-embryonic stages, bab2 is expressed in a broader range of tissues than bab1 and generally shows a higher level of expression. In larval and prepupal ovaries, bab1 transcript and protein are only detected in cells that form the terminal filaments (Fig. 3B,C, Fig. 4D). The expression of bab2 is more complex (Fig. 3A, Fig. 4A-C). At early to mid third larval instar, prominent bab2 expression is seen in the developing terminal filaments and in a cell population that we named ‘swarm cells’. Swarm cells migrate from anterior to posterior past the cluster of germ cells during third larval instar (Fig. 4A,B). They produce the basal stalks, a pupal-specific tissue, and may also contribute to tissues of the adult ovary (M. L. and D. G., unpublished). The highest level of bab2 expression in the swarm cells is seen during their migration. bab2 is also expressed in the apical cells of the larval ovary. After terminal filaments have formed, apical cells migrate between the terminal filaments posteriorly and form the outer sheaths of the egg tubes. The level of expression in these cells increases during the third larval instar and is highest at the time the cells begin their posterior migration (Fig. 4C). bab2 expression is also seen in the interstitial cells that intermingle with the germ cells. bab mutant ovaries not only display defects in terminal filament formation but also in other cell populations of the ovary, such as the apical cells and the basal stalk primordium (Godt and Laski, 1995). If the development of these cell populations depends on the presence of terminal filaments, then the observed defects in these cell populations could be a secondary effect of bab mutations. Alternatively, bab may be directly required for the development of these cell populations as the apical cells and swarm cells express bab2.
bab1 and bab2 transcripts are expressed in a similar pattern in the tarsal primordium of leg imaginal discs (Fig. 3D,E) and in a subdistal region of the antennal disc (data not shown) (Godt et al., 1993). Similar to the protein distribution of Bab1 (Fig. 4J) (Godt et al., 1993), Bab2 protein is expressed in a graded manner in the tarsal primordium, with the concentration of Bab2 highest in tarsal segments TS3 and TS4, lower in TS2 and even lower in TS1 (Fig. 4H,I). However, the differences in the level of expression between the tarsal segments are not as pronounced as with Bab1. Both Bab proteins are enriched in the ridges compared to the furrows of the tarsal folds. In contrast to Bab1, Bab2 expression is not restricted to TS1-4 but is also found in the proximal region of TS5 (Fig. 4H), in the peripodial membrane, and in the periphery of the leg imaginal disc that gives rise to thorax structures (Fig. 4I). No morphological defects have been observed in derivatives of leg imaginal discs outside the tarsus.
Both bab genes are expressed in the female and male genital discs. The genital discs give rise to the internal and external structures of the genitalia, the A8 and A9 tergites and the anal plates (Bryant, 1978; Casares et al., 1997). The strongest expression of Bab proteins in the female genital disc is found in the primordium of the vaginal plates and the A8 tergite (Fig. 4F), structures that are affected in bab mutants. In the male genital disc, bab expression is mainly seen in a region of the male genital primordium (Fig. 4G). In addition, in the central nervous system (CNS), Bab1 and Bab2 are distributed in a similar pattern. bab-expressing cells are found in the central brain hemispheres and the thoracic ganglia of late 3rd instar larvae and prepupae (Fig. 4E). In bab mutants, no gross morphological defects were observed in histological sections of the prepupal CNS, and it remains uncertain whether bab has a function in the brain.
In summary, the expression pattern of bab1 during imaginal development can be described as a subpattern of the bab2 expression pattern. In tissues that require bab function for development, bab1 and bab2 are usually co-expressed.
Molecular analysis of bab mutations reveals the functional importance of both bab genes
To determine the roles of bab1 and bab2 in mediating bab function, we conducted a molecular and phenotypic analysis of mutations in the bab locus. The P-element insertion of babP maps close to the 5′ end of the first intron of bab1 (Fig. 1). babP does not affect the transcription of bab2 but results in the loss of the bab1 5.4 kb transcript and the appearance of an abundant 2.6 kb transcript (Fig. 5A). This shorter bab1 transcript is detected in babP heterozygotes and homozygotes, and is detected only with probes located upstream of the insertion. Characterization of the 3′ end of this 2.6 kb mRNA by 3′ RACE-PCR showed that this transcript is a hybrid of the first bab1 exon and a region of the P[ry+, lacZ] construct. The 5′ splice site of the bab1 transcript that normally functions to splice out the large first intron of bab1 and the 3′ splice site of the l(3)S12 gene, contained in the P[ry+, lacZ] construct just upstream of the rosy gene (Dutton and Chovnick, 1991), are spliced together. This demonstrates that the P[ry+, lacZ] insertion causes aberrant splicing and transcription termination of the bab1 transcript. A similar event has been reported for a PlacZ allele of the psq gene (Horowitz and Berg, 1995). The 2.6 kb truncated bab1 transcript is more abundant than the 5.4 kb RNA (Fig. 5A) suggesting that it might be more stable than the wild-type transcript. Translation of this transcript would produce a protein that contains the BTB domain of Bab1 but not the BabCD. A Bab1-specific antibody directed to a domain of the Bab1 protein that is encoded by the truncated transcript, however, did not detect any protein in babP homozygous flies (data not shown). These results suggest that the babP P-element insertion severely disrupts or abolishes the function of the bab1 gene. Nevertheless, babP homozygous flies display ovary defects of only intermediate strength and reveal no leg defects (Table 1) (Godt et al., 1993; Sahut-Barnola et al., 1995). This indicates that a loss of bab1 does not produce a bab null mutant phenotype and suggests that a second gene is involved in bab function.
This hypothesis is corroborated by the analysis of another P-element insertion, babA128 (Godt et al., 1993) that maps 57 bp from the 5′ end of the bab1 transcript. Only one phenotypic trait of bab mutants, an abdominal pigmentation defect in females is associated with this insertion. Because homozygous babA128 flies have no ovary or leg defects, it was surprising to find that Bab1 protein is not detectable in babA128 mutant tissues (data not shown). A strong reduction in the amount of bab1 transcript, as seen by in situ hybridization and RNA blot analysis (data not shown), indicates that the babA128 insertion interferes with the transcription of bab1, reducing Bab1 to undetectable levels. By contrast, flies that are heterozygous for strong bab mutations, such as babPR72, or deletions of the bab locus, such as Df(3L)Fpa1, have reduced but clearly detectable levels of Bab1 (data not shown), and nevertheless show leg defects in addition to defects in abdominal pigmentation (Table 1). Taken together, analysis of the mutations babP and babA128 strongly suggests that bab1 is not the only gene involved in bab function.
That bab2 is involved in bab function was confirmed by a protein analysis of bab mutants. The wild-type Bab2 protein is detected in immunoblots as a band of approximately 145 kDa using different anti-Bab2 antibodies that recognize either the N- or C-terminal region of Bab2 (Fig. 5B-D, data not shown). An analysis of bab mutants revealed that the alleles babE1, babE4 and babE5 affect the Bab2 protein (Fig. 5B). In a homozygous babE1 mutant, Bab2 protein is reduced to barely detectable levels. babE4 and babE5 mutants produce truncated Bab2 proteins. By contrast, Bab1 expression appears normal in these three mutants, shown by tissue immunostaining as the anti-Bab1 antibody does not produce a signal in immunoblots (Table 1). No change in the size of Bab2 and the expression level of either Bab1 or Bab2 were detected in babE3 and babE6 mutants. babE1, babE4, and babE5 mutants display developmental defects in ovaries, legs and the abdomen (Table 1), demonstrating that the bab2 gene plays an essential role in development and that it is functionally related to bab1.
The bab null phenotype requires a lack of both bab1 and bab2 activity
The strongest bab mutations previously published have both a strong ovary and leg phenotype (Godt et al., 1993; Godt and Laski, 1995). Further analysis of two of these mutations, babPRDS and babPR72, revealed that they affect the expression of both bab1 and bab2. They each lack detectable amounts of Bab1 and have reduced levels of Bab2 (Fig. 5C). The mutant phenotypes caused by babPRDS or babPR72 are slightly enhanced in trans to a large deletion (Df(3L)babPG), indicating that these mutations are not null for the bab locus (Godt et al., 1993).
Additional bab alleles were isolated and studied to find one that completely lacks bab activity. Two deletions, Df(3L)Fpa1 and Df(3L)Fpa2, that were isolated based on a dominant female pigmentation defect (Ken Burtis, personal communication), extend into the bab locus from opposite sides, each having a breakpoint in the bab locus (Fig. 1). Df(3L)Fpa2 deletes bab2 completely, and deletes the 5′ region of bab1, including the BTB domain. Df(3L)Fpa1 deletes bab1, and has a breakpoint in the second intron of bab2, deleting everything downstream of the BTB domain. In Df(3L)Fpa1/Df(3L)Fpa2 transheterozygotes and in flies homozygous for the mutation babAR07, neither Bab1 nor Bab2 are detected (Fig. 5C,D). The phenotype of these genotypes is stronger than that of previously described mutations (Table 1). As these flies lack both bab1 and bab2 function, this phenotype corresponds to the null phenotype of the bab locus.
The bab locus regulates the development of ovaries, limbs and the abdomen
To further analyze the function of the two bab genes, we studied the phenotypic series of bab mutations and the bab null mutant phenotype. bab null mutants are semi-viable. They develop into pharate adults but often have difficulties in eclosing from the pupal case, which may be a result of their leg defects. bab null mutants display defects in ovaries, tarsal segments, antennae, abdominal segments and female genital disc derivatives.
Based on the phenotypic series of bab alleles of varying strength, we define four phenotypic classes of bab mutant adult ovaries (Fig. 6).
(1) Females with a weak bab mutant ovary phenotype are fertile but have ovaries that are somewhat smaller than wild-type ovaries, slightly irregular, and rounded at the anterior end, owing to defects in terminal filament formation (Fig. 6B). The ovarioles contain normal-looking follicles and mature oocytes.
(2) Female flies with an intermediate bab mutant ovary phenotype are semi-sterile to sterile. The ovaries have a very irregular shape and are substantially smaller than wild-type ovaries (Fig. 6C,D). They contain a reduced number of ovarioles that are abnormally oriented with the germaria often not located at the anterior end but inside the ovary (Sahut-Barnola et al., 1995).
(3) Females with a strong bab mutant ovary phenotype do not lay eggs. The ovaries are very small and contain only one to two ovarioles of very abnormal structure and orientation (Fig. 6E,F) (Godt and Laski, 1995). Only very few and defective follicles are found in these ovarioles.
(4) In bab null mutants, the ovaries are even smaller and no developing follicles have been observed.
The bab mutant leg phenotype that involves all three leg pairs in both females and males has two characteristics: (1) a fusion of tarsal segments, characterized by a shortening of tarsal segments and a loss of tarsal joints; and (2) a transformation of the bristle pattern of distal tarsal segments toward the bristle pattern of the first tarsal segment (Fig. 7) (Godt et al., 1993). Most sensitive to a fusion are tarsal segments TS5 and TS4 (Fig. 7C,D). The stronger the bab mutation, the further proximal the fusion extends. In bab null mutants, TS5 to TS2 are frequently fused into a single segment (Fig. 7E-H). Sensitivity to a transformation of the bristle pattern of tarsal segments decreases from proximal to distal, involving only TS2 in weak bab mutants and TS2-4 in strong bab mutants (Fig. 7) (Godt et al., 1993). This can best be seen using the prominent sex comb bristles of the prothoracic legs of males as a marker (Fig. 7C,E), and the transverse bristle rows of the pro- and metathoracic legs of both sexes (Fig. 7C-H). In a bab null mutant, the bristle pattern of TS2-4 is transformed; however, the sex combs are often eliminated, owing to the shortening and fusing of the tarsal segments (Fig. 7G). A thickening of the distal tarsal segments seen in bab mutant legs is an additional indication that the distal tarsal segments are transformed towards the identity of the first tarsal segment, which in wild type is much thicker than the distal tarsal segments (Fig. 7A,B,E-H).
bab loss-of-function defects in the abdomen
Wild-type females have eight tergites formed by abdominal segments A1-A8 (Fig. 8A), whereas wild-type males have seven tergites corresponding to A1-A6 + A9 (Fig. 8G) (Jürgens and Hartenstein, 1993). In females, the tergites of A1-A6 each show a darkly pigmented posterior and lightly pigmented anterior band. The two tergite plates of A7 are variably pigmented and A8 has a light coloration. The tergites of A1-A4 in males are similarly pigmented as in females, whereas the tergites of A5-A6 + A9 are darkly pigmented throughout. bab null mutants display a change in the pigmentation pattern of both sexes. A thorough dark pigmentation is found in A3-A8/A9 and is seen with low penetrance also in A2 (Fig. 8D,F,H). Ectopic dark pigmentation in A2 and A3 is usually patchy and restricted to the anterior margin (Fig. 8D,H). The phenotypic series of bab mutants shows that the sensitivity towards a change in pigmentation decreases from posterior to anterior, with A6 being more sensitive than A5, and A5 more sensitive than more anterior segments (Fig. 8B-D). In weak bab mutants, a change in pigmentation is therefore seen only in females. In summary, this indicates that loss of bab function leads to a transformation of the pigmentation from a female to a male-like pattern as well as from an anterior to a posterior-like pattern. The bab locus regulates the pattern and amount of pigmentation in all abdominal segments (except for A1), and suppresses dark pigmentation in the anterior region of abdominal segments in both sexes with the exception of A5 + A6 in males.
Except for a change in the pigmentation pattern, the morphology of A2-A5 appears to be normal in both sexes of bab null mutants. However, the posterior segments A6-A8 show additional morphological abnormalities, most of which are restricted to females. The trichome pattern of A6 is affected in both sexes. In bab null mutants, trichomes are not restricted to the anterior and lateral margin of the A6 tergite as in wild type, but are found at a low density throughout the tergite, similar to the normal trichome pattern of A5 (data not shown). This suggests a posterior-to-anterior transformation of the trichome pattern. Furthermore, the A6 tergite of bab mutant females is broader (anteroposterior) than the more anterior tergites (Fig. 8F, double-arrow) in contrast to wild type (Fig. 8E), which together with the heavy pigmentation gives this tergite a male-like appearance.
In contrast to A1-A6, in which the two primordia of each tergite fuse into a single plate, the A7 tergite consists of two loosely connected triangular plates in wild-type females that show small slightly twisted bristles and often two to three larger bristles (Fig. 8E). In bab null mutants, the A7 plates are fused into a continuous plate, are considerably broader (anteroposterior) than in wild type, and display an increased number of large bristles (Fig. 8F). These morphological changes suggest a transformation of A7 towards a more anterior segment fate. Furthermore, instead of the small pale bristles, which are characteristic of an A8 tergite in wild-type females, larger pigmented and slightly twisted bristles are found in a bab mutant A8 tergite (data not shown). Such bristles are similar to those normally found in A7 of females or in A9 of males, again suggesting homeotic transformations. In addition, the two rows of thorn bristles seen on the vaginal plates of wild-type females (Fig. 8I) are replaced in bab mutants by a different type of bristles, which are longer and twisted (Fig. 8J,K).
bab mutants also display defects in the sternites of the abdominal segments. Shape, pigmentation, and bristle pattern of the A6 and A7 sternites in females are different from wild type and show similarities to the A6 sternite in males (Fig. 8I-L). Both sternites are more strongly pigmented, and the number of bristles is considerably decreased compared with wild type. Taken together, the alterations in the shape and the bristle and trichome patterns of posterior segments indicate that loss of bab function causes posterior-to-anterior transformations of some abdominal features (opposite to the change of the pigmentation pattern), and also transformations from a female to a male-like morphology.
bab gain-of-function defects in the abdomen
Ubiquitous overexpression of a UAS-bab2 transgene under control of Hsp70-Gal4 causes reduced viability, a general reduction in the pigmentation of the cuticle and bristles, and defective macrochaetae when flies are raised at a constant temperature of 25°C. In the abdomen of both sexes, the posterior dark pigmentation of A6 is reduced or missing, and little of the dark pigmentation is left in A5 and A4 (Fig. 8M-P). Tergites anterior to A4 are less affected than the posterior tergites. In both, bab loss- and gain-of-function experiments, pigmentation in posterior segments is more strongly affected than in anterior segments, indicating a graded requirement for bab along the anteroposterior axis. Similar phenotypic effects were observed in bab1 overexpression experiments (data not shown). Together, our loss- and gain-of-function studies show that the bab locus is a suppressor of dark cuticle pigmentation in the fly.
When flies, carrying UAS-bab2 under the control of Hsp70-Gal4, were shifted to 32°C during the late 3rd instar/pupal stages, they showed a ‘split-tergite’-phenotype in addition to the loss of pigmentation (Fig. 8N). Here, the tergite primordia of all abdominal segments do not fuse, a trait normally only found in A7. The split-tergite phenotype is also seen when UAS-bab2 expression is driven by a bab1-Gal4 transgene (F. L., unpublished transgenic line), and is therefore likely not an artifact of the heat shock but a consequence of bab2 overexpression. These data suggest that bab plays a role in tergite morphogenesis and is required to prevent a fusion of the A7 tergite primordia.
The functional relationship between Bab1 and Bab2: synergistic and distinct effects
To analyze the relative functions of bab1 and bab2, and to look for possible interactions, we compared the phenotypic effects of bab mutations in ovaries and legs, and studied their complementation behavior. This study involved mutations that affect bab1 (babP and babA128), bab2 (babE1, babE4, and babE5), bab1 and bab2 (babPRDS and babPR72), or are null for bab1 and bab2 (babAR07 and deletions of the bab locus), and some molecularly uncharacterized bab alleles.
All five EMS alleles (babE series of alleles) were isolated based on dominant leg defects. babE1 is the strongest EMS allele, causing a strong recessive phenotype in ovaries and legs; babE4 and babE5 produce an intermediate, and babE6 and babE3 a weak recessive phenotype in those organs (Tables 1,Table 2-3, Fig. 6). The phenotype seen in flies transheterozygous for any two of the EMS alleles is intermediate in strength to the phenotypes displayed by the homozygotes, indicating that the phenotypic effect of the EMS alleles is additive (Tables 1,Table 2-3). The EMS alleles in trans to the strongest allele babE1 produce a mutant phenotypic series that is comparable with, but less severe than, each EMS allele in trans to a deletion of the bab locus (Tables 2, 3). As babE3 and babE6 do not complement babE1 but complement babP they should represent bab2 mutations like the other EMS alleles (Table 2). Therefore, all bab alleles isolated on the basis of a dominant mutant leg phenotype are mutations in bab2.
All bab alleles that were isolated as excision derivatives of babP (the babPR alleles) show non-complementation in trans to each other or in trans to a deletion of the bab locus, and display a normal phenotypic series (Tables 2, 3). The babPR alleles that cause a strong mutant phenotype, such as babPR72 and babPRDS, not only reduce the expression level of bab1, but also of bab2. Mutations in bab that affect bab1 but have no detectable effect on the expression of bab2, such as babP and babA128, have a considerably weaker mutant phenotype. No bab mutation has been identified that affects only bab1 and causes strong mutant defects in ovaries and/or legs. Therefore, we propose that bab2 plays a predominant role in exerting bab function in ovarian and particularly in leg development.
We also compared the effects of bab1 and bab2 mutations on abdominal pigmentation. Mutations in either bab gene cause dominant and more pronounced recessive pigmentation defects. Females homozygous for the strong bab2 allele babE1 display an uniformly dark pigmentation of tergites in A5 and A6, and females homozygous for the intermediate allele babE5 show uniformly dark pigmentation in A6 and partial ectopic pigmentation in A5 (data not shown). By contrast, bab1 mutant females that are homozygous for babP or babA128 show ectopic dark pigmentation in the tergites of A6 and A7 (Fig. 8B). These observations suggest that there is an overlapping and differential requirement for bab1 and bab2 in abdominal segments.
To gain a better understanding of the relationship of the two bab genes, mutations in bab1 and bab2 were tested for complementation. Partial non-complementation is observed between bab1 and bab2 alleles in ovaries (Table 2) and legs (Table 3). First, flies carrying a bab1 and a bab2 mutation in trans display a mutant ovary phenotype, although it is weaker than the one observed in flies homozygous for either the bab1 or the bab2 mutation. This may be caused by an interaction of these alleles with the wild-type copy of bab1 and/or bab2, as flies heterozygous for a deletion of the bab locus do not have ovary defects. Second, although bab1 mutations produce neither dominant nor recessive leg defects in a background that is wild-type for bab2, flies carrying a bab1 mutation in trans to a bab2 mutation show leg defects that are stronger than the dominant leg defects caused by the bab2 mutation in trans to a wild-type chromosome. This suggest that bab1 is functionally active although not essential in leg development. Our complementation data point to functional dependency between mutations in bab1 and bab2, the nature of which remains to be explored.
In summary, our genetic analysis of the bab locus indicates that bab1 and bab2 are partially but not fully redundant in function, and that both bab genes are required for normal bab function.
Bab1 and Bab2 belong to a distinct subfamily of BTB domain proteins in Drosophila
The superfamily of BTB domain proteins, to which Bab1 and Bab2 belong, constitutes one of the largest protein families in C. elegans, Dm and humans, with 58 representatives found in the Dm genome (Zollman et al., 1994; Lander et al., 2001) (Pointud and J.-L. C., unpublished). The majority of these proteins in Dm and humans also contain Zn-finger domains and are thought to be transcriptional regulators. The BTB domain is likely to contribute to this role by mediating protein oligomerization (Bardwell and Treisman, 1994). BTB domains can form homo-oligomers and hetero-oligomers (Dhordain et al., 1995; Dong et al., 1996; Kobayashi et al., 2000). Crystallographic data suggest that the BTB domain of PLZF (promyelocytic leukemia Zn-finger) can form homodimers that show an extended hydrophobic interaction between the BTB domains, indicating a high degree of stability (Ahmad et al., 1998; Li et al., 1997). We have shown that the BTB domain of Bab1 can interact with itself and with the BTB domain of Bab2 in a yeast two-hybrid system suggesting that the Bab proteins can homomerize and/or heteromerize via this domain (Pointud et al., 2001).
It has also been shown that BTB domains can interact with non-BTB domain-containing proteins. For example, the transcriptional repressors PLZF or BCL6 interact via their BTB domains with the co-repressors SMRT, N-CoR and mSin3A (Hong et al., 1997; Dhordain et al., 1997; Guidez et al., 1998; Huynh and Bardwell, 1998; Lin et al., 1998). These co-repressors bind in turn to a histone deacetylase that can modify the local chromatin and inhibit transcription (David et al., 1998; Dhordain et al., 1998; Lin et al., 1998). This mechanism seems not to be universal for all BTB domain-containing transcriptional repressors, however, because the HIC-1 protein is not able to bind to N-CoR, SMRT or mSin3, and its repression activity is not affected by inhibitors of histone deacetylases (Deltour et al., 1999). We have recently shown that the BTB domain of the Bab proteins binds to the two nuclear proteins BIP-1 and BIP-2 (Bab Interacting Protein) (Pointud et al., 2001). BIP-2/TAFII155 is a TFIID component, and is homologous to yTAFII47, a histone fold containing TAFII (Gangloff et al., 2001). This suggests that the Bab proteins may regulate transcription of their target genes by directly interacting with a component of the general transcription machinery via their BTB domains.
The Bab proteins belong to a subfamily of nuclear BTB domain proteins that have no Zn fingers for DNA binding. However, the BabCD in Bab1 and Bab2 contains a Psq domain and a motif that may correspond to an AT-hook. As both Psq domains and AT-hooks are known DNA-binding domains (Lehmann et al., 1998; Reeves and Nissen, 1990), we propose that the BabCD mediates DNA binding. It is noteworthy that the protein di/oligomerization domain is always found in the N-terminal region and widely separated from the DNA-binding domain in BTB domain proteins, including Bab1 and Bab2. Such an organization, unlike many other transcription factors, should allow for high flexibility of these proteins when bound to DNA. This property and the capacity of the BTB domain to form oligomers suggest a model for how BTB domain proteins can regulate gene expression at the chromatin level by binding to distant DNA-binding sites and forming a large complex that brings together regions that are far apart on the DNA, moving enhancers or silencers closer to the basal transcriptional machinery.
The functional relationship of the paralogues bab1 and bab2
Gene duplications are thought to play a crucial role in the evolution of new morphologies and the generation of new species. Increase of gene copies through duplication can lead directly to morphological changes if the gene function is dose sensitive. Furthermore, duplications provide the raw material for the development of new gene functions (for reviews, see Sidow, 1996; Wagner, 1998), as seen, for example, in the Hox gene cluster (for reviews, see Duncan, 1987; Carroll, 1995; Gellon and McGinnis, 1998). It has been also proposed that the inactivation or loss of duplicated genes can efficiently induce reproductive isolation, a basis for divergence and speciation (Lynch and Conery, 2000). The functional study of duplicated genes can therefore illuminate evolutionary processes and can help to determine how molecular divergence translates into divergence of gene function. Our characterization of the paralogous genes bab1 and bab2 suggests (1) that both genes are required for normal development, (2) that their functions show redundant and divergent aspects, and (3) that bab2 plays a predominant role in exerting bab function.
Divergence in function of two genes that arose by tandem duplication might be due to qualitative or quantitative differences in cis-regulation of gene expression or due to differences in protein structure. In case of the bab paralogues, we document evidence for both as discussed below. bab1 is expressed in a subpattern of bab2 and shows a generally lower expression level than bab2 at the RNA and protein level. In most tissues that have been found to display morphological defects in bab mutants, bab1 and bab2 are co-expressed, raising the possibility that bab1 and bab2 are co-required for normal bab function. They could function as heterodimers that are formed through the interaction of their BTB domains. However, if Bab1/Bab2 heterodimers would be the only functional unit of bab action, we would expect that a knock out of either bab1 or bab2 would yield a bab null phenotype, which is not the case. In addition, ectopic expression of either bab1 or bab2 alone has a phenotypic effect suggesting that each bab gene is able to exert a function on its own. Whether Bab1 and Bab2 interact molecularly in the regulation of downstream genes awaits further analysis.
The two bab genes seem partially redundant as the strongest developmental defects in ovaries, legs and the abdomen associated with the bab locus have been observed only in mutants that are null for bab1 and bab2. The two bab genes are not functionally equivalent, however. First, there is an overlapping but also differential requirement for bab1 and bab2 in the pigmentation of different abdominal segments, with A7 being more dependent on bab1 and A5 on bab2 activity. Second, ovarian defects are seen with mutations affecting either bab1 or bab2, but loss-of-function of bab2 causes a more severe phenotype. As the function of the bab locus is strongly dose/concentration dependent, the predominance of bab2 in regulating ovarian development may be a result of the higher expression level of bab2. Furthermore, the differences in the ovarian expression patterns may have functional significance. Cis-regulatory differences, however, cannot sufficiently explain the differential requirement of bab1 and bab2 in leg development. Although both genes are similarly expressed in leg imaginal discs, a bab1 knockout does not cause a mutant leg phenotype, whereas even weak bab2 mutants display dominant leg defects. This indicates that only bab2 plays an essential role in leg development and suggests a qualitative divergence in the function of Bab1 and Bab2 proteins. Taken together, we propose that bab1 and bab2 have not only developed differences in transcriptional regulation but also differences in protein function that could be responsible for changes in the interaction with other transcription factors and/or DNA-binding sites.
Function of bab as a homeotic and morphogenetic regulator of imaginal development
bab acts as a homeotic regulator, as bab mutations cause homeotic transformations in the legs and the abdomen (this work) (Godt et al., 1993; Kopp et al., 2000). Here, we show that the homeotic transformations in the abdomen of bab mutants are complex. bab loss-of-function mutants display a combination of anterior-to-posterior transformations (pigmentation), posterior-to-anterior transformations (bristles, trichomes and segment shape and size), and female to male transformations (pigmentation, bristles and segment shape and size). bab seems to be mainly required in the posterior segments A5-A8. This domain that is mostly controlled by the Hox gene Abdominal-B (Abd-B), whose loss-of-function causes posterior to anterior transformations of segment identity (Karch et al., 1985; Sánchez-Herrero et al., 1985). It has been demonstrated that bab expression is repressed by Abd-B, either directly or indirectly, in posterior abdominal segments at the late pupal stage (Kopp et al., 2000). As bab acts as a suppressor of pigmentation, the repression of bab expression by Abd-B function leads to the complete pigmentation of the A5 and A6 tergites in wild-type males. In females, the repression of bab by Abd-B is counteracted by the female specific doublesex (dsxF) gene product (Kopp et al., 2000). It is unlikely, however, that Abd-B is a general repressor of bab activity, as bab mutants show not only anterior-to-posterior but also posterior-to-anterior transformations in the abdomen. This indicates that the regulation of bab activity is complex. Abd-B in conjunction with co-regulators might repress or activate bab function, dependent on the cell type and on the developmental time at which specific morphological features are specified. We propose that the differential, fine-tuned spatial and temporal regulation of bab expression plays a crucial role in providing morphological diversity between the abdominal segments along the anteroposterior axis and between the sexes. Similar to the abdomen, bab plays a role in the generation of morphological diversity between distal segments in the leg. bab is part of a network of transcription factors that divide the proximodistal axis into successively smaller domains, leading to the formation and specification of the different leg segments (Godt et al., 1993; Duncan et al., 1998; Campbell and Tomlinson, 1998; Couso and Bishop, 1998; Galindo and Couso, 2000).
bab also plays a role as a morphogenetic regulator of development. Previous studies have indicated that bab controls cell rearrangements during terminal filament formation in the ovary (Godt and Laski, 1995; Sahut-Barnola et al., 1995; Sahut-Barnola et al., 1996). bab is also required for the proper folding of leg imaginal discs, which may be important for tarsus segmentation (D. G., unpublished). Furthermore, bab negatively regulates the fusion of the tergite primordia in the abdomen, a process that is also controlled by the Hox genes (Karch et al., 1985). This suggests that the Bab transcription factors control the morphogenetic behavior of cells in different developmental processes. It will be a future challenge to determine whether bab directly regulates expression of proteins that mediate cell shape changes and cell movements.
Does bab play a role in variation and divergence of reproduction-related traits in the Drosophila family?
Flies of the Drosophila family show substantial intraspecific and interspecific variation in sex-related traits, including sex combs and abdominal pigmentation as described before, as well as male genital structures (Bock and Wheeler, 1972; True et al., 1997) and number of ovarioles (Mahowald and Kambysellis, 1980; Hodin and Riddiford, 2000). Variation in these traits can affect mate choice and fertility, and thus reproductive success (Spieth, 1952; Cook, 1977; Coyne, 1985; Markow et al., 1996; Arnqvist, 1998; Kopp et al., 2000; Hodin and Riddiford, 2000). Furthermore, there is evidence that divergence of phenotypic traits related to reproduction in combination with ecologically adaptive divergence in sexual selection can lead to reproductive isolation and speciation (Boughman, 2001; Schneider, 2000).
Interestingly, bab controls the morphology of several traits that are involved in reproduction and show rapid evolutionary divergence. bab regulates the formation of the reproductive organ in females, as it is required for terminal filament formation and consequently for the development of ovarioles in the ovary (Godt and Laski, 1995; Sahut-Barnola et al., 1995). bab mutations of increasing strength cause a decrease in the number of ovarioles, raising the possibility that bab might be involved in determining ovariole number in Dm. Moreover, bab controls several secondary sexual traits. bab activity suppresses sex combs on tarsal segments distal to TS1 (this work) (Godt et al., 1993). bab may also be involved in determining the number of sex comb bristles in TS1, as overexpression of Bab2 in TS1 causes a reduction in the number of sex comb bristles compared with wild type (D. Godt, unpublished). Furthermore, bab regulates sexually dimorphic bristle and trichome patterns and the pigmentation of posterior abdominal segments (this work) (Kopp et al., 2000). A comparison of abdominal pigmentation and bab expression pattern between the two sexes of different members of the Drosophila species group demonstrates a striking correlation between phenotypic differences and bab expression patterns, suggesting a causal relationship (Kopp et al., 2000). That changes in cis-regulatory elements can indeed lead to variation in morphology between different Drosophila species was documented for the ovo/shavenbaby locus that controls the segmental trichome pattern in larvae (Sucena and Stern, 2000). As bab loss- and gain-of-function mutations of bab have pleiotropic effects on the development of reproduction-related characteristics, evolutionary alterations in bab function could lead to a diversification of multiple sex traits.
The bab locus appears to have two important properties that make it suitable to cause variation in the development of morphological traits. First, because the bab locus represents a tandem duplication, redundancy between bab1 and bab2 may have facilitated fast molecular modifications, resulting in the observed alterations of the expression level and pattern of bab1 and bab2 and their functional diversification. One potential consequence, for example, would be that abdominal pigmentation could change independently of the leg pattern through mutations in bab1, as this gene is no longer essential for leg development. In fact, many alleles of the fap locus, which affects female abdominal pigmentation and is presumably identical to bab, have been proposed to occur in wild-type populations of Dm (Robertson et al., 1977).
Second, bab function is highly dose dependent. bab is haploinsufficient, and bab mutations cause dominant homeotic transformations of adult characteristics that do not interfere with viability in laboratory cultures. The expression profile of bab in imaginal discs and the abdomen is graded, and differences in bab concentration determine morphology of legs and abdomina (Godt et al., 1993; Kopp et al., 2000). As concentration matters, small variations in the expression level or shape of the bab gradient could lead to morphological diversification. Such a mechanism has also been proposed for Ubx in establishing the trichome pattern in one of the leg segments (Stern, 1998). Differences in Ubx expression between Drosophila species are responsible for the difference in their trichome morphology as shown by interspecies crosses involving Ubx mutations of these species (Stern, 1998). Taken together, we propose that bab is an important regulator of reproduction-related characteristics in Dm, and therefore may play an active role in the variation, divergence and speciation in the genus Drosophila.
We thank Ken Burtis for sharing unpublished information and babD1 flies; Gail Flores and Silvia Yu for help with mutagenesis experiments; Gil Privé, Andrea Brand, L. Kalfayan, Carl Thummel, the DSHB and Bloomington stock center for flies and reagents; and Ulrich Tepass, Joel Atallah and Corinne Lours for helpful comments on the manuscript. Funding was provided by the INSERM and grant GREG 59/95 (to J. L. C.), by the NSERC (to D. G.), and by the NIH (#GM40451 to F. A. L.).