The Drosophila gypsy retrotransposon disrupts gene activity by blocking the interactions of distal enhancers with target promoters. This enhancer‐blocking activity is mediated by a 340 bp insulator DNA within gypsy. The insulator contains a cluster of binding sites for a zinc finger protein, suppressor of Hairy wing [su(Hw)]. Recent studies have shown that a second protein, mod(mdg4), is also important for normal insulator function. Mutations in mod(mdg4) exert paradoxical effects on different gypsy‐induced phenotypes. For example, it enhances yellow2 but suppresses cut6. Here, we employ a stripe expression assay in transgenic embryos to investigate the role of mod(mdg4) in gypsy insulator activity. The insulator was inserted between defined enhancers and placed among divergently transcribed reporter genes (white and lacZ) containing distinct core promoter sequences. These assays indicate that mod(mdg4) is essential for the enhancer‐blocking activity of the insulator DNA. Moreover, reductions in mod(mdg4)+ activity cause the insulator to function as a promoter‐specific silencer that selectively represses white, but not lacZ. The repression of white does not affect the expression of the closely linked lacZ gene, suggesting that the insulator does not propagate changes in chromatin structure. These results provide an explanation for why mod(mdg4) exerts differential effects on different gypsy‐induced mutations.
Insulator DNAs are thought to isolate genetic loci by blocking interactions between cis regulatory elements and inappropriate target promoters in neighboring transcription units (Kellum and Schedl, 1991; Chung et al., 1993). The best characterized ‘authentic’ insulators correspond to scs and scs‘, which flank the Drosophila hsp70 locus (Kellum and Schedl, 1991). In addition, a variety of studies suggest that a 340 bp DNA element contained within the gypsy retrotransposon also functions as an insulator (Geyer et al., 1986; Geyer and Corces, 1992). Although this insulator has no known cellular counterpart, it behaves in a fashion that is indistinguishable from scs and scs’ in enhancer‐blocking assays (see below).
We recently devised a stripe expression assay in transgenic embryos to characterize insulator DNAs (Cai and Levine, 1995). This assay was used to show that the scs, scs' and gypsy insulators selectively block the expression of distal, not proximal, even‐skipped (eve) stripe enhancers when positioned between divergently transcribed reporter genes. Enhancers barred from acting on one basal promoter are able to activate the other promoter, suggesting that insulators do not propagate changes in chromatin structure. Insulator DNAs appear to lack an intrinsic polarity, and are equally effective in blocking distal enhancers whether they are positioned in the same or opposite orientation relative to the target transcription unit. Insulators can quantitatively modulate enhancer–promoter interactions, in that a portion of the scs element only partially blocks distal stripe enhancers. Finally, insulator DNAs may possess some regulatory specificity. For example, the gypsy insulator selectively blocks the activators, but not silencers, contained within the composite ventral repression element from the zerknullt (zen) promoter region (Jiang et al., 1992, 1993; Cai et al., 1996).
It has been proposed that insulators segregate DNA into distinct chromatin loop domains (Galloni et al., 1993; Udvardy and Schedl, 1993; Corces, 1995). Evidence for this type of model stems from the demonstration that scs, scs‘ and the gypsy insulator element contain a series of DNase I hypersensitive sites (Udvardy et al., 1985; Udvardy and Schedl, 1993; Vazquez and Schedl, 1994). However, it is equally plausible that insulators function as autonomous, specialized DNA elements which attenuate or block distal enhancers. Insulators might block the tracking of distal enhancers by bending DNA, and mimicking the structural changes mediated by the binding of the TFIID complex to the promoter (Dunaway and Droge, 1989; Shen et al., 1994; Oelgeschlager et al., 1996). Alternatively, insulators might contain binding sites for ’pairing proteins' that are required for the looping of distal enhancers to the transcription complex (Georgiev and Corces, 1995; Zhou et al., 1996).
As a first step towards determining how insulators block distal enhancers, we have analyzed the expression of synthetic gene complexes containing the gypsy insulator in various mutant embryos. The minimal, 340 bp element contains a cluster of 12 tightly linked binding sites for a Drosophila zinc finger protein, suppressor of Hairy wing [su(Hw)] (Geyer et al., 1988; Parkhurst et al., 1988; Spana et al., 1988; Harrison et al., 1993; Kim et al., 1996). The binding of the su(Hw) protein to the insulator is required for the blocking of distal enhancers, as well as the mutagenic effects of the gypsy retrotransposon (Harrison et al., 1989; Hoover et al., 1992; Smith and Corces, 1992; Dorsett, 1993).
Recent studies have shown that su(Hw) interacts with a protein encoded by the mod(mdg4) locus, which was first identified as an enhancer of position effect variegation (Georgiev and Gerasimova, 1989; Dorn et al., 1993; Gerasimova et al., 1995; Georgiev and Kozycina, 1996). The functional significance of this protein–protein interaction is unclear. In some cases it appears that mod(mdg4) is simply required for gypsy insulator activity. For example, the phenotype caused by a gypsy insertion in the promoter region of the cut gene is suppressed by a mutation in mod(mdg4) (Georgiev and Gerasimova, 1989; Georgiev and Kozycina, 1996). This observation suggests that the loss of mod(mdg4) function permits the distal wing margin enhancer, located upstream of the gypsy insertion, to interact with the cut promoter (see Figure 1). In contrast, the same mod(mdg4) mutation enhances the phenotype caused by a gypsy insertion in the promoter region of the yellow locus (Georgiev and Gerasimova, 1989; Georgiev and Kozycina, 1996). In this case, tissue‐specific enhancers located both upstream and downstream of gypsy appear to be repressed (Gerasimova et al., 1995; see Figure 1).
To investigate the seemingly paradoxical effects of mod(mdg4) mutants on gypsy insulator function, we examined transgenic embryos expressing fusion promoters with defined enhancers and divergently transcribed test promoters. Mutations in mod(mdg4) cause distal enhancers to be derepressed on both the white and eve/lacZ test promoters, suggesting that mod(mdg4) is essential for gypsy insulator function. Moreover, there is a general repression of white, suggesting that the gypsy insulator functions as a promoter‐specific silencer in mod(mdg4) mutants. This selective repression is observed even when the gypsy insulator is located far from the white promoter. In contrast, the closely linked eve/lacZ gene, which contains a distinct basal promoter sequence, is not repressed in mod(mdg4) mutants, suggesting that silencing of the white gene does not involve long‐range changes in chromatin structure. These studies provide the first evidence for promoter‐specific silencing, and offer an explanation for why mod(mdg4) mutants suppress some gypsy‐induced mutations, but enhance others.
A hypomorphic mutation in mod(mdg4), mod(mdg4)u1 (Georgiev and Gerasimova, 1989; Gerasimova et al., 1995; Georgiev and Kozycina, 1996), exerts paradoxical effects on two different gypsy‐induced mutations (Figure 1). It enhances the y2 phenotype, so that there is a loss of pigmentation in mesothoracic bristles (Figure 1B; compare with A). Conversely, mod(mdg4)u1 suppresses the ct6 phenotype, resulting in the restoration of the wing margin (Figure 1D; compare with C).
Both the yellow and the cut promoter regions contain a series of tissue‐specific enhancers (Jack, 1985; Geyer et al., 1986; Geyer and Corces, 1987; Jack and DeLotto, 1995). In the y2 mutant, gypsy is inserted between an upstream body cuticle enhancer and a downstream bristle enhancer. In the y2 mutant (Figure 1A), the pigmentation of the body cuticle is yellow but the bristles are dark, indicating that the enhancers located upstream of gypsy are blocked while the proximal bristle‐specific enhancer is active. The mutant phenotype is enhanced in flies doubly mutant for y2 and mod(mdg4)u1, in that both the cuticle and the bristles are yellow (Figure 1B, compare with A), indicating that the proximal bristle‐specific enhancer is inactivated. It has been suggested that this loss of function of the proximal enhancer is caused by the conversion of the gypsy insulator into a bidirectional repressor (Gerasimova et al., 1995). The ct6 mutant contains a gypsy insertion between the cut promoter and a distal wing‐margin enhancer that maps nearly 90 kb upstream of the promoter. This enhancer is blocked in the ct6 mutant, resulting in the loss of wing margin tissue (Figure 1C). In contrast, the enhancer appears to be reactivated in the double mutant, ct6 and mod(mdg4)u1, resulting in the restoration of the wing margin (Figure 1D, compare with C).
The enhancer‐blocking assay (Cai and Levine, 1995) was used to investigate how mod(mdg4) mutants exert these opposing effects on yellow and cut gene activity. The 340 bp gypsy inuslator was placed between divergently transcribed white and lacZ reporter genes. The promoter‐proximal sequences associated with these reporters are distinct. The leftward white gene lacks a canonical TATA (O'Hare et al., 1984), while the rightward lacZ gene is driven by a small segment of the eve promoter, which contains an optimal TATA element (Frasch and Levine, 1987). Transgenic embryos were hybridized with either digoxigenin‐labeled white or lacZ antisense RNA probes to visualize the expression of the reporter genes.
eve stripe assay
The insulator was placed between the 500 bp eve stripe 3 enhancer (‘E3’) and the 480 bp eve stripe 2 enhancer (‘E2’). As shown previously, the insulator selectively blocks interactions between the distal enhancer and target promoter (Cai and Levine, 1995). The proximal E3 directs the expression of the leftward white reporter gene within the limits of stripe 3 (Figure 2B), but the distal E2 is blocked. Conversely, the proximal E2 directs the expression of the rightward lacZ reporter within the limits of stripe 2, while the distal E3 is severely attenuated (Figure 2C). The head stripe observed with the lacZ probe is due to cryptic regulatory sequences in the P‐element transformation vector (Small et al., 1992).
Distinct patterns of expression are observed when this fusion promoter is expressed in embryos derived from mod(mdg4)u1 homozgyous females (Figure 2A and D). There is a severe reduction in white expression (Figure 2A), suggesting that both the distal E2 and proximal E3 are ineffective in activating the white promoter (compare with Figure 2B). Moreover, there is residual expression of the white reporter gene within the limits of both stripes 2 and 3, suggesting that the distal E2 is derepressed. These results suggest both a loss of gypsy insulator function, and a general repression of white.
Embryos derived from mod(mdg4)u1 homozygous mothers exhibit intense lacZ expression (Figure 2D), even though the closely linked white gene is repressed. Staining is detected within the limits of stripes 2, 3 and 7, although stripes 3 and 7 are somewhat weaker than stripe 2 (stripe 7 is regulated by sequences within E3; Small et al., 1996). This result indicates that the distal E3 is severely derepressed (Figure 2D; compare with Figure 2C). A similar derepression of the lacZ reporter gene is observed in hypomorphic mutations in su(Hw) (Cai and Levine, 1995; data not shown). Thus, both mod(mdg4) and su(Hw) are required for gypsy insulator function.
Control experiments were done to determine whether the repression of the white promoter seen in mod(mdg4)u1 mutants (Figure 2A) depends on the presence of the gypsy insulator DNA. For this purpose, we examined a fusion promoter that contains E3 and E2 separated by a 400 bp spacer sequence (‘T0.4’; see diagram below Figure 2E–H). This spacer permits both enhancers to interact with both the leftward white promoter (Figure 2F) and the rightward eve/lacZ gene (Figure 2G) in wild‐type embryos. These staining patterns are not altered in embryos derived from females homozygous for the mod(mdg4)u1 mutation (Figure 2E and H). In particular, the white reporter is expressed within the limits of stripes 2 and 3, indicating that the mod(mdg4)u1 mutation does not cause a general silencing of white when the gypsy insulator is replaced by a random spacer sequence (Figure 2E).
Analysis of additional embryonic enhancers
The gypsy insulator was placed between additional combinations of embryonic enhancers. The results obtained with a fusion promoter containing a mesoderm‐specific enhancer (‘2PE’) and E3 are presented in Figure 3. These enhancers normally direct a composite staining pattern consisting of a central segmentation stripe and a band of staining along the ventral surface (Figure 3F and G). As seen for E2–E3 fusion genes, the gypsy insulator selectively blocks distal, but not proximal, enhancers (Figure 3B and C). The insulator blocks interactions between the distal E3 and white promoter, so that expression is mediated only by the proximal 2PE enhancer in the ventral mesoderm (Figure 3B). Conversely, the eve/lacZ reporter gene is expressed within the limits of stripe 3, while the distal 2PE is blocked (Figure 3C).
The activities of this fusion promoter are altered in mod(mdg4)u1 mutants (Figure 3A and D). As before, there is a general repression of white, so that both the distal E3 and proximal 2PE fail to direct normal levels of expression (Figure 3A). The weak stripe 3 pattern suggests that the distal E3 is derepressed, although it is difficult to assess E3–white interactions since E3 is a weak enhancer that is quite sensitive to general spacing effects (see below). Nonetheless, reduced expression of the proximal 2PE enhancer indicates a general repression of the white promoter. In contrast, the rightward lacZ reporter gene is fully active, so that both the proximal E3 and the distal 2PE direct a composite staining pattern consisting of a central stripe and a ventral band (Figure 3D). The mod(mdg4)u1 mutation causes a nearly complete derepression of the distal 2PE enhancer, in that staining in ventral regions is nearly as intense as that observed for fusion promoters lacking the insulator (Figure 3D, compare with H; see below).
Control experiments involved the use of a large, 1.4 kb, spacer DNA from λ (see diagram below Figure 3E–H). Unfortunately, this spacer causes a reduction in E3–white interactions, leading to a relatively weak stripe 3 pattern (Figure 3F). This spacer does not reduce 2PE–eve/lacZ interactions on the lacZ reporter gene (Figure 3G). Moreover, the same spacer does not impede interactions between white and a different enhancer, hairy H1 (see below). These results, as well as previous studies (e.g. Cai and Levine, 1995), suggest that E3 is inherently ‘weak’ and easily blocked. There is no significant change in the composite staining pattern when this fusion gene is expressed in embryos derived from mod(mdg4)u1 females (Figure 3E and H; compare with F and G, respectively). These observations suggest that the selective repression of white observed in mod(mdg4)u1 mutants depends on the presence of the gypsy insulator.
We also analyzed a fusion promoter that contains the 330 bp rhomboid lateral stripe enhancer (‘NEE’; Ip et al., 1992) and a 200 bp hairy stripe 1 enhancer (‘H1’; see Zhou et al., 1996). The two enhancers direct a composite pattern of expression, consisting of a head stripe (H1) and lateral stripes (NEE; e.g. Figure 4G). The gypsy insulator selectively blocks or attenuates the distal enhancer so that the NEE directs lateral stripes of white expression (Figure 4B), while H1 directs an intense head stripe of lacZ; the distal NEE is attenuated, but not completely blocked (Figure 4C). In this regard, we note that the NEE is most resistant to inhibition by the insulator among the various enhancers that have been tested in this assay.
As before, the mod(mdg4)u1 mutation exerts differential effects on the white and lacZ staining patterns. Both proximal and distal enhancers interact with the eve/lacZ gene, thereby resulting in a composite lacZ staining pattern which includes intense NEE lateral stripes (Figure 4D). In contrast, there is a general repression of white, so that the proximal NEE directs weaker lateral stripes in the mutant background as compared with normal embryos (Figure 4A, compare with B). The use of the l.4 kb spacer sequence indicates that the repression of white depends on the presence of the gypsy insulator (Figure 4A and E).
gypsy can function as a long‐range repressor
Additional experiments were conducted to determine whether the gypsy insulator can work over long distances to repress white. For this purpose, we examined a fusion gene containing E3 positioned between the leftward white gene and rightward lacZ gene. The insulator was placed downstream of the lacZ transcription unit, >4 kb away from the eve/lacZ transcription start site, and even farther from the white promoter (see diagram below Figure 5). The insulator does not interfere with the activities of E3 in wild‐type embryos (Figure 5A and B), so that both reporter genes are expressed within the limits of stripe 3. The weak stripe 7 pattern is mediated by regulatory sequences contained within E3 (Small et al., 1996; see Figure 2).
The lacZ staining pattern is not altered when this fusion gene is expressed in embryos derived from mod(mdg4)u1 homozygous females (Figure 5D; compare with B). In contrast, there is a subtle, but reproducible reduction in white expression (Figure 5C; compare with A). Control stainings were done with a comparable fusion gene lacking the insulator DNA; no reduction in stripe 3 expression was observed (data not shown). While it appears that the insulator can silence white over long distances, we note that this repression is far less efficient than that obtained when the insulator is located near white (e.g. Figure 2A).
Analysis of a mod(mdg4) null mutant
mod(mdg4) encodes a nuclear protein with a BTB domain that mediates protein–protein interactions, including interactions with su(Hw) (Dorn et al., 1993; Gerasimova et al., 1995). The mod(mdg4)u1 allele is caused by the insertion of a Stalker transposon into C‐terminal sequences of the coding region (Gerasimova et al., 1995). This insertion could result in the synthesis of truncated proteins, which might function in a dominant‐negative or neomorphic fashion. To investigate this possibility, we examined a null mutation that lacks the promoter and first exon of the coding region.
The insulator selectively blocks the interaction of the distal E2 with the leftward white reporter gene (Figure 6A), and attenuates the interaction of the distal E3 with the rightward lacZ reporter (Figure 6B). The insulator works equally well in wild‐type embryos and embryos derived from mod(mdg4)u1/+ heterozygous females (Figure 6A; compare with Figure 2B). As shown previously, the lacZ staining pattern is derepressed and white expression attenuated in embryos derived from mod(mdg4)u1 homozygous females (Figure 6C and D). Very similar staining patterns are observed in embryos derived from transheterozygous females, which contain one copy of the mod(mdg4)u1 allele and one copy of the null allele (Figure 6E and F). Once again, there is a derepression of the distal E3 on the lacZ gene (Figure 6F) and a general repression of white (Figure 6E). Similar staining patterns were observed in embryos derived from null homozygous oocytes (data not shown), which were produced as germ line clones using the flip–frt method (Chou and Perrimon, 1992). These results suggest that it is the reduction in the levels of the mod(mdg4) protein, and not a change in the nature of the protein, which alters the behavior of the gypsy insulator in mod(mdg4)u1 mutants.
It would appear that mod(mdg4) is required for the normal function of the gypsy insulator DNA since mod(mdg4) mutants result in a derepression of distal enhancers on the eve/lacZ promoter (e.g. Figure 6D and F). A similar loss of insulator function is observed when the same fusion gene is expressed in embryos carrying a transheterozygous combination of different su(Hw) hypomorphic alleles. In these mutants, both the E3 and E2 enhancers are active on both the white and lacZ promoters (Cai and Levine, 1995; data not shown). Thus, both su(Hw) and mod(mdg4) are required for normal insulator activity; however, mod(mdg4) mutants result in the repression of the white promoter, while su(Hw) mutants do not.
The gypsy retrotransposon disrupts gene function by blocking the interactions of distal enhancers with target promoters. Extragenic modifiers of gypsy‐induced mutations have been identified, and among these the mod(mdg4) gene is the most paradoxical since it suppresses the effects of some gypsy insertions, but enhances others. We have presented evidence that mod(mdg4) is essential for normal insulator activity. Moreover, the gypsy insulator functions as a promoter‐specific silencer in mod(mdg4)u1 mutant embryos. Selective repression of white, but not lacZ, is observed even when the insulator is positioned far from both promoters. Repression of white does not influence the expression of a closely linked eve/lacZ gene, suggesting that the insulator does not occlude binding of upstream activators through general changes in chromatin structure such as chromosome condensation. Instead, promoter‐specific repression can explain why mod(mdg4)u1 suppresses the ct6 mutation but enhances y2.
Mechanism of insulator function
A model summarizing the results of this study is presented in Figure 7. As shown previously, the zinc finger su(Hw) protein binds to multiple sites within the 340 bp gypsy insulator DNA (Spana et al., 1988; Harrison et al., 1993; Kim et al., 1996). It would appear that mod(mdg4) lacks a DNA‐binding domain and requires specific protein–protein interactions with su(Hw) in order to interact with the insulator (Gerasimova et al., 1995). The su(Hw)–mod(mdg4) protein complex divides the bidirectional white–lacZ gene complex into two domains. Enhancers to the left of the insulator (e.g.‘E1’; see Figure 7) interact with the leftward target promoter, while enhancers to the right of the insulator complex (‘E2’) interact with the rightward promoter. As discussed earlier, it is unclear whether this separation of the leftward and rightward transcription units into separate domains involves the formation of chromatin loops. It has been proposed that endogenous insulators located near the site of integration would interact with the su(Hw)–mod(mdg4) complex to separate the two genes into distinct nuclear compartments (Corces, 1995). An alternative view is that the su(Hw)–mod(mdg4) complex mimics the basal promoter region, for example by bending the DNA, similar to the structural changes induced by the binding of the TFIID complex to TATA (Oelgeschlager et al., 1996).
Reduction or loss of mod(mdg4) activity causes the su(Hw) insulator to mediate promoter‐specific repression of the white gene. In the absence of mod(mdg4) function, it is possible that su(Hw) interacts with a different, unknown protein (‘?’), which selectively represses white. Moreover, because mod(mdg4) is absent, the insulator no longer blocks distal enhancers, so both the E1 and E2 enhancers can interact with both promoters, particularly the rightward eve/lacZ reporter gene.
The promoter‐specific repression of white might explain why the mod(mdg4)u1 mutation enhances y2, but suppresses ct6. Perhaps the y and white promoters are repressed, while cut and eve/lacZ are refractory to gypsy‐mediated silencing. It is possible that proximal sequences in the white and y promoters are responsible for mediating this repression.
Modes of repression
Previous studies suggest that there may be two basic forms of repression in the early Drosophila embryo, short‐range and long‐range. Short‐range repressors function over distances of just 50–100 bp to inhibit, or quench, neighboring activators bound within an enhancer (Gray et al., 1994; Arnosti et al., 1996; Gray and Levine, 1996). This form of repression permits enhancer autonomy in modular promoters, so that a short‐range repressor bound to a given enhancer does not interfere with the activities of other enhancers contained within the same promoter region. Several spatially localized repressors have been shown to function in this manner, including the zinc finger repressor snail, which is a mesoderm determinant, and the segmentation repressor Kruppel, which defines the posterior border of eve stripe 2. Short‐range repressors do not exhibit promoter specificity. For example, Kruppel represses the stripe 2 enhancer and establishes the posterior stripe 2 border on both the white and eve/lacZ reporter genes (e.g. Gray and Levine, 1996).
Long‐range repressors can function over distances of several kilobases to block the assembly or function of the basal transcription complex (for review, see Cai et al., 1996). Such repressors function in a dominant fashion, and block the activities of multiple enhancers in modular promoters (Jiang et al., 1992, 1993). The dorsal protein can mediate long‐range repression through a 600 bp silencer element, the VRE, located within the promoter region of zen. The VRE contains several dorsal binding sites and closely linked negative response elements (NREs). Dorsal is inherently an activator, but mediates repression by recruiting ‘co‐repressors’ to the NREs (Lehming et al., 1994). The VRE can function over distances of several kilobases to repress the ventral expression of the eve stripe 2 and stripe 3 enhancers (Cai et al., 1996). The long‐range, dominant repression activity of the VRE does not appear to exhibit promoter specificity. Thus, the VRE represses the ventral expression of both eve stripes when monitoring either the white reporter gene or lacZ.
The repression mediated by the gypsy insulator seems to be distinct from previously characterized short‐range and long‐range repressors. It can function over long distances, >4 kb, to attenuate E3–white interactions (see Figure 6). However, unlike the zen VRE, the gypsy insulator appears to function in a promoter‐specific fashion, and does not interfere with lacZ expression.
Mechanisms of repression
The gypsy‐mediated silencing observed in mod(mdg4)u1 does not appear to involve general changes in chromatin structure such as chromosome condensation. For example, the lacZ reporter gene is fully active in situations where the nearby white promoter is repressed. This result suggests that the insulator does not interfere with the binding of upstream activators to neighboring enhancers. Previous genetic studies have identified a number of suppressors of mod(mdg4) mutants, including su(var)205 (Dorn et al., 1993), which encodes HP1, a constitutent of heterochromatin (James et al., 1989). Gerasimova et al. (1995) have proposed that HP1 mediates general silencing of the y promoter region in the absence of mod(mdg4) gene activity. Genetic tests are consistent with the notion that this repression of y involves a heterochromatin‐mediated process. Our studies do not address the issue of heterochromatin, but strongly suggest that the conversion of the gypsy insulator DNA into a silencer does not involve long‐range changes in chromatin structure.
Materials and methods
Drosophila strains and genetic crosses
yw67 flies were used for all P‐transformation injections. This strain was also used as the wild‐type stock for control stainings. Females of genotype y2 sc1 waG ct6; mod(mdg4)u1 (kindly provided by Tatiana Gerasimova and Victor Corces, see also Gerasimova et al., 1995) were mated with yw67 males to generate mod(mdg4)u1 heterozygous and mod(mdg4)+ flies carring gypsy insertional mutations on the X chromosome. To generate a transheterozygous strain for mod(mdg4)u1 and the E(var)93D deficiency, the above stock was crossed with E(var)93D‐P142.4D15 TM3 Sb/TM6 (kindly provided by Manfred Frasch, see Azpiazu and Frasch, 1993) which contains a 1 kb deletion removing the promoter and first exon of the gene. Virgin females of the indicated genotype were mated with wild‐type males carring the P‐transposon. Embryos were collected and fixed as described previously (Jiang et al., 1992).
Wings were dissected from wild‐type and mutant flies, and mounted in a drop of glycerol. Notums were boiled in 10% KOH for 10 min and washed in H2O (Ashburner, 1989). Following serial dehydation in 50, 70 and 100% ethanol, they were rinsed in xylene and mounted in Permount (Fisher, SP15‐100). The mounted cuticles were photographed using Nomarski optics.
P‐element transformaiton and whole mount in situ hybridization
Fusion genes were made by inserting test enhancers into the unique EcoRI site of a P‐element transformation vector that contains the eve basal promoter (from −42 bp), a 100 bp untranslated leader sequence and the first 22 codons of the coding region fused in‐frame to the lacZ gene (pEb vector; Small et al., 1992; S.Small, unpublished data). This CaSpeR derivative (Thummel et al., 1988) contains the mini‐white gene transcribed in the opposite orientation relative to the linked eve/lacZ fusion gene. P‐element fusion genes were introducted into the Drosophila germ line as described in Small et al. (1992). Multiple transformant lines were generated for each construct, and at least three independent lines were tested. Hybridizations using digoxigenin‐UTP‐labeled antisense RNA probes to lacZ or white were performed exactly as described by Jiang et al. (1991).
Construction of eve/lacZ P‐element transposons
The E3–SU–E2 fusion gene shown in Figures 2 and 6 is described in Cai and Levine (1995). The E3–T0.4–E2 gene containing a 400 bp DNA fragment from the twist promoter region (located between the DE and PE cis regulatory elements; see Jiang et al., 1991) was kindly provided by Steve Small. The 2PE–SU–E3 gene (Figure 3) was prepared with a 500 bp eve stripe 3 enhancer (Small et al., 1993) and a 560 bp DNA fragment containing two tandem copies of the twist PE (‘proximal element’; see Jiang et al., 1991; Jiang and Levine, 1993). These DNAs were co‐ligated into the unique EcoRI site in the pEb vector, creating a new NotI site between the two enhancers. The correct orientation of the enhancers was confirmed by DNA sequencing. Either the 340 bp gypsy insulator DNA or a 1.4 kb EcoRI fragment from λ was inserted into the NotI site using a NotI linker. The NEE–SU–H1 and NEE–λ1.4–H1 fusion genes (Figure 4) were prepared with the 330 bp rhomboid NEE (Ip et al., 1992; Gray et al., 1994; Gray and Levine, 1996) and a 200 bp hairy stripe 1 enhancer (kindly provided by Gary Struhl; see Zhou et al., 1996). These DNAs were co‐ligated into the unique EcoRI site of the pEb vector, creating a new NotI site between the two enhancers. The position and orientation of the enhancers were confirmed by DNA sequencing. The 340 bp gypsy insulator or 1.4 kb λ spacer was inserted into the NotI site. The E3–SU fusion gene shown in Figure 5 was prepared wtih a 700 bp DNA fragment containing both the 340 bp gypsy insulator and the 330 bp NEE. This DNA was ligated into the unique XbaI site located just downstream of the lacZ transcription unit. The NEE is not active on either the white or lacZ genes from this distant location.
We thank Tatiana Gerasimova and Victor Corces for providing the mod(mdg4)ul stock. We thank Marietta Dunaway and Don Rio for critically reading the manuscript. This study was funded by a grant from the NIH (GM 34431).
- Copyright © 1997 European Molecular Biology Organization