The su(Hw) Insulator Can Disrupt Enhancer-Promoter Interactions When Located More than 20 Kilobases Away from the Drosophila achaete-scute Complex

Abstract

Here we report that the su(Hw) insulator may not necessarily separate promoters from enhancers to allow inhibition of transcription by the su(Hw) protein. For this purpose we used the strains of Drosophila melanogaster which carry inversion of the region containing the yellow gene and the achaete-scute complex (AS-C). Despite the reverse orientation of the region, the AS-C enhancers continue to activate achaete and scute gene expression. The su(Hw) insulator, located more than 20 kb away from the inversion, facilitates strong suppression of achaete and scute gene expression, although is does not separate the promoters from the AS-C enhancers.

Enhancers exert long-distance effects, which raises a question as to how an enhancer specifically activates its target gene without affecting adjacent genes. Recent experimental evidence suggests two different models which help clarify this question. According to the first model, some structural features of chromatin divide the chromosome into distinct domains of gene action, and a given enhancer can interact with a promoter only if they reside in the same domain (10, 31–33). According to the second or promoter specificity model, the inherent properties of the promoters and enhancers allow only some combinations to interact, while other combinations are inefficient (34, 37).

Several sequences, referred to as insulators, have been found to prevent activation or repression from extending across them to a promoter. To date, only a few insulators have been well characterized: the insulator of the chicken β-globin gene cluster (7), the scs and scs′ elements of the Drosophila heat shock gene (28, 29, 53, 55), and the regulatory region of the gypsy (mdg4) retrotransposon, su(Hw)-binding region (3, 22, 26, 27, 45, 49). The su(Hw)-binding region contains 12 binding sites for the su(Hw) protein (36, 52). This protein plays a pivotal role in the insulation function, since mutations in the su(Hw) gene eliminate the enhancer blocking (8). Mutations in another gene, modifier of mdg4 [mod(mdg4)], alter the phenotypes of several gypsy-induced mutations, indicating that this gene encodes a protein which is also involved in the function of the su(Hw) insulator (4, 14, 16, 18, 50).

In this study, we have examined enhancer-promoter interactions in the presence of the su(Hw)-binding region in the genome area containing the yellow, achaete (ac), and scute (sc) genes. The yellow gene determines the proper pigmentation of the cuticle structures, and its expression in different tissues is controlled by enhancers located in the 5′ upstream region and intron of the gene (21, 40). In the y² mutation, the retrotransposon gypsy is inserted between the enhancers controlling the yellow expression in the wings and body cuticle and the yellow promoter (19, 20, 41). After gypsy insertion, enhancers active in the body and wing are blocked due to insulation.

The ac and sc proneural genes, two members of the achaete-scute complex (AS-C), are located in the vicinity of the yellow gene (5) and differ by spatial and temporal patterns of expression. The proteins encoded by these genes are the most essential for the formation of bristles (macrochaetae) (6, 12). The expression of the ac and sc genes is confined to the proneural clusters that determine the precise positions of macrochaetae (9, 44, 51). A very complex pattern of ac and sc expression is mediated by the action of site-specific, enhancer-like elements distributed over about 90 kb of the AS-C (6, 24, 44, 46, 47) cluster. The ac and sc genes are both expressed in the same cells as a consequence of the activation of both genes by the same set of enhancers (24).

To address whether the transcription of the yellow gene and AS-C depends on the structure of the region, we used a derivative of the previously described y^+ns mutation in the yellow locus generated by insertion of a chimeric element (17). The chimeric element consists of two identical copies of the P element with deleted central portions and a 19.7-kb duplicated genomic sequence inserted between them. We obtained the y^2ns1 sc^mes1 derivative by mobilization of the P element in the y^+ns strain. The mutant was generated by an inversion of a region containing the yellow, ac, and sc genes. One of the inversion boundaries in the obtained mutant is located in the same region as in the previously described sc⁴ and sc^260-15 strains with a profound sc phenotype, where all cis-regulatory elements of AS-C are displaced. However, y^2ns1 sc^mes1 flies exhibited a moderate sc mutant phenotype, suggesting that some enhancers were still capable of activation of the ac and sc transcription. Surprisingly, we found that the su(Hw) insulator, located within the yellow gene at a distance of more than 20 kb from the closest inversion boundary, is responsible for the sc mutant phenotype, although it does not separate the AS-C enhancers from the ac and sc promoters. Here we also demonstrated that the leucine zipper and at least one of the acidic domains of the su(Hw) protein are required for the observed repression. The same domains of the su(Hw) protein had been found to be essential for the insulator function (16, 25, 30). The 22.1-kb chimeric element does not influence the su(Hw)-mediated inhibition of the ac and sc expression. These data suggest that localization of an insulator between a promoter and the respective enhancers is not a necessary prerequisite for its interference with the enhancer activity.

MATERIALS AND METHODS

Drosophila strains.

All flies were maintained at 25°C on a standard yeast medium. The w; Sb P[ry⁺ Δ2-3]e/TM6,e stock providing a stable source of transposase (43) was obtained from the Bloomington stock center. In what follows, the P[ry⁺ Δ2-3]99B construct is referred to as Δ2-3. Highly unstable mutations at the yellow locus were described previously (17). The mutations and constructs of the su(Hw) gene were obtained from V. Corces, their origin and structure having been described by Harrison et al. (25). All other mutant alleles and chromosomes used in this work and all balancer chromosomes were described by Lindsley and Zimm (35).

Genetic crosses.

To induce mutagenesis, females from a strain with certain y* and sc* alleles were crossed with w; Sb Δ2-3 e/TM6,e males to produce dysgenic males of the y*sc*/Y; Sb Δ2-3 e/+ genotype, where y* and sc* were any y or sc mutations. Three to ten F₁ males from each bottle were then individually crossed with 10 to 12 C(1)RM,yf females with attached X chromosomes (/Y). The F₂ progeny were analyzed for mutants. All males with a new y or sc phenotype were individually mated to virgin C(1)RM,yf females, and their phenotype was examined in the next generation.

The mutations and constructions in the su(Hw) gene and the mod(mdg4)^1u1 mutation were combined with the y*sc* mutations as described previously (16).

For determination of the yellow phenotype, the extent of pigmentation in different tissues of adult flies was estimated visually in 3- to 5-day-old male flies developing at 25°C. The wild-type expression was ranked as 5, while the absence of yellow expression was ranked as 0. Flies with the previously characterized y alleles (15) were used as a reference to determine the levels of pigmentation.

Mutant sc phenotypes were analyzed according to the method of Garcia-Bellido (12) as follows: ♀/Y × ♂X*/Y → analysis of X*/Y male flies, where X* is the X chromosome of interest. In the case of each sc allele, 70 to 100 male flies were examined at 25°C to determine the sc phenotype.

Molecular methods.

For Southern blot hybridization, DNA from adult flies was isolated as described by Ashburner (1). Treatment of DNA with restriction endonucleases, blotting, fixation, and hybridization with radioactive probes prepared by random primer extension was performed as described in the protocols for Hybond-N nylon membrane (Amersham) and by Sambrook et al. (48). Phages with cloned regions of the yellow locus were obtained from J. Modolell and V. Corces. The probes were prepared from gel-isolated fragments obtained after restriction endonuclease digestion of the plasmid subclones.

Genomic DNA libraries were constructed by using DNA isolated from flies with a definite genotype and were partially digested with Sau3A endonuclease. The digested DNA was ligated in the λgem11/BamHI phage vector (Promega). The recombinant DNA was packaged in vitro by using a packaging extract from Promega, and the material was plated on petri dishes by using E. coli LE392 at a density of 3,000 PFU/plate. The plaques were blotted onto Hybond-N⁺ nylon membranes according to the supplied protocol (Amersham). These membranes were hybridized with ³²P-labelled DNA probes to select the desired plaques; 30,000 to 40,000 plaques from each recombinant DNA library were screened. Positive plaques were cored from the plates and rescreened to obtain pure clones.

DNA sequence analysis, subcloning, and purification of the plasmid DNA and mapping of the restriction sites were performed by standard techniques (48).

The regions of interest in the y mutations were cloned via DNA amplification by standard PCR techniques (11). The following primers in DNA amplification were used with the yellow gene (y1, TCTGTGGACCGTGGCGCGGTAAC; y2, TTGAACTGACAGCTAATCGTCGG; and y3, CTAACATTGCCGTGGATATAGGC), the P element sequences (p1, TCGGTAAGCTTCGGCTTTCGAC; p2, CGTCCGCACACAACCTTTCCTCTC; and p3, AATAAGTCCGCCGTGAGACACCTC), and the AS-C (sc1, GACTTTAAGATGCTTTCAGAGATCCC; and sc2, GGCGTGTGCTACTTGTCTTAGG). The products of amplification were fractionated by electrophoresis in a 1.5% agarose gel in Tris-agarose-EDTA buffer. The successfully amplified products were directly sequenced with an Amersham sequencing kit for PCR products by using the same or internal primers.

Genetic system.

The results here were obtained by comparing phenotypes with molecular organization in a series of double mutants in the yellow locus and AS-C. Such mutations were obtained in a previously described system of highly unstable mutations in the yellow locus (15, 17).

The original highly unstable yellow mutation, y^+ns, was induced by the insertion of chimeric mobile elements consisting of duplicated genomic sequences framed with P elements (15, 17). The main double mutant, y^2ns1 sc^mes1, was obtained after mobilization of the P element in the y^+ns strain. High instability can be induced by supplying the mutant strains with a source of active P element transposase, Δ2-3 strain (43). This leads to the appearance of a large number of derivatives with different y and sc phenotypes. Figure 1 describes the origin of the y and sc mutant phenotypes obtained during this work.

FIG. 1.

The lineage of mutations described in Fig. 2 to 7. All alleles are indicated by boldface letters. The designations in brackets are the individual representatives of the group shown above them. The values in parentheses near the allele designations show the total number of scored flies, and the specific figure where the maps of corresponding alleles are presented is given in brackets. The values in parentheses near the arrows show the frequency of appearance of the corresponding alleles (the number of independent events with similar y and sc phenotypes/total number of scored flies).

Further strategy of the study was as follows. The mutants of interest were selected by the analysis of phenotypes and by the preliminary Southern blot analysis, which included the double digestion of DNA with BamHI and BglII endonucleases not cutting P element sequences and hybridization with probes from the yellow (HindIII-BamHI), AS-C (EcoRI-BglII), and chimeric insertion (HindIII-EcoRI) (see below and Fig. 3). Southern blot analysis of digested DNA isolated from the original and derivative strains gives information about the position of changes and can identify inversion between the yellow gene and AS-C or between AS-C and some other region of the genome. Thereafter, the yellow and sc regions of the selected derivatives were studied in detail by Southern blot hybridization with different combinations of restriction enzymes and probes from the yellow gene, AS-C, and the chimeric element as shown in Table 1. In some derivatives, the region of change was cloned by PCR amplification and directly sequenced.

FIG. 3.

Structure and phenotypes of y^2ns1 sc^mes1 derivatives. (A) Structure of y^2ns1 sc^mes1 mutation and its X-ray-induced derivative y^2ns14 sc^1x1. Restriction enzymes: R, EcoRI; H, HindIII; G, BglII; X, XhoI; B, BamHI; and S, SalI. The genomic DNA fragments HindIII-BamHI (yellow), EcoRI-BglII (AS-C), and HindIII-EcoRI (chimeric element) used for Southern blot analysis are indicated by thick lines. The P-element sequences are shown by boxes. Relative orientations of the gypsy LTR and the P element are indicated by arrows. (B) Structure of y^+ns sc^+s derivatives of y^2ns1 sc^mes1. (C) Phenotypes of the indicated sc mutations in male flies. All designations are as described in Fig. 2D.

TABLE 1.

Description of Southern blot hybridization experiments

Purpose of expt	Restriction enzyme combinations	Hybridization probesa
To determine the structure of P elements at the breakpoints of inversion/reinversion at:
The border with the yellow gene	BamHI-HindIII, EcoRI, XhoI, PstI, SalI, or KpnI	HindIII-BamHI (yellow)
The border with the sc gene	BglII-HindIII, EcoRI, XhoI, PstI, SalI, or KpnI	EcoRI-BglII (AS-C)
The border with the chimeric element	BglII-XhoI, PstI, SalI, or KpnI	HindIII-EcoRI (chimeric element)
To determine the presence of an inversion between yellow and AS-C or between AS-C and another region of the genome	BamHI-BglII, BglII, BamHI-XbaI, or BglII-XbaI	HindIII-BamHI (yellow) and EcoRI-BglII (AS-C)
To determine the no. of P elements at the breakpoints of inversion	BamHI-BglII or BglII	HindIII-BamHI (yellow), EcoRI-BglII (AS-C), and HindIII-EcoRI (chimeric element)
To determine deletions in the sequences flanking the P elements	BamHI-BglII, BglII, BamHI-XbaI, or BglII-XbaI	HindIII-BamHI (yellow), HindIII-HindIII (yellow), EcoRI-BglII (AS-C), HindIII-EcoRI (chimeric element), and λsc phages

^aThe region from which the probe was taken is indicated in parentheses. 

RESULTS

The su(Hw) insulator inhibits the activity of AS-C enhancers in the y^2ns1 sc^mes1 mutant which carries inversion of the yellow-ac-sc region.

Here we analyzed the action of the su(Hw) insulator on various combinations of mutations in the yellow gene and in the AS-C.

The Drosophila strains with these mutations were obtained by mobilization of P elements in the parental y^+ns strain as a result of crosses with the flies carrying Δ2-3 construct as an autonomous source of transposase (43). The y^+ns allele (Fig. 2B) has two insertions within the regulatory region: the gypsy mobile element at the position −700 bp and a chimeric element at the position −69 bp from the transcription start site of the yellow gene. In the chimeric element, a 19.7-kb sequence combined from three different X-chromosomal regions is flanked by two identical copies of a defective 1.2-kb P element designated P1 and P2 oriented in the same way (17). The P1 element is distal and the P2 element is proximal to the yellow promoter. y^+ns flies have the wild-type pigmentation of the body and wings. This pigmentation pattern relies on the transcription of the yellow gene activated by the enhancers within the chimeric element (17). By mobilization of P elements in the y^+ns strain (Fig. 1), we obtained a mutant devoid of pigmentation of the wing blade, the body cuticle, the notum, the legs, and partially the wing bristles (the y phenotype). Moreover, several types of bristles were missing (the sc phenotype). The modified yellow and sc alleles were designated y^2ns1 and sc^mes1, and the corresponding Drosophila strain was designated y^2ns1 sc^mes1.

FIG. 2.

The nature and properties of original mutations in the yellow, ac, and sc loci. (A) Schematic presentation of the yellow-ac-sc region in the previously described y, ac, and sc mutants (5). The coordinates in the AS-C region are as defined by Campuzano et al. (5). Vertical arrows indicate the positions of chromosomal breakpoints associated with corresponding mutations. Arrows with a triangle show insertions of gypsy and P elements associated with certain mutations. The size of insertions corresponds to the length of the base of the triangle. Relative orientations of P elements are indicated by arrows. Thick horizontal arrows show the positions and approximate sizes of transcribed regions and the direction of transcription. The black circle indicates the su(Hw)-binding region of gypsy. The enhancer elements are shown by boxes with a definite number as follows: 1, the body and wing enhancers of the yellow gene (21); 2, the regulatory region of the chimeric element (17); 3, the yellow bristle enhancer (21); 4, the enhancer responsible for the formation of anterior and posterior dorsocentral macrochaetae; 5, the anterior supra-alar and posterior postalar enhancer; 6, the anterior postalar enhancer; 7, the anterior notopleural enhancer; 8, the scutellar enhancer (24). The size of the sc⁶ deletion is indicated by an elongated open box (5). This deletion includes the enhancers responsible for the formation of postvertical, anterior orbital, and ocellar macrochaetae. (B) Schematic presentation of the yellow-ac-sc region in the y^+ns strain. Designations are the same as in Fig. 2A. (C) Schematic presentation of the yellow-ac-sc region in the y^2ns1 sc^mes1 strain. Designations are the same as in Fig. 2A. (D) Phenotypes of the indicated sc mutations in males. The standard nomenclature for each bristle is indicated as follows (35): HU, humeral; ASA, anterior supra-alar; PSA, posterior supra-alar; APA, anterior postalar; PPA, posterior postalar; PS, presutural; AOR, anterior orbital; OC, ocellar; PV, postvertical; ANP, anterior notopleural; PNP, posterior notopleural; and SC, scutellar. Only affected bristles in sc mutations are shown. Empty boxes indicate that the corresponding bristles are present (wild-type phenotype). One-quarter-full, one-half-full, and completely full boxes mean that the corresponding bristle(s) is(are) absent in more than 10, 50, or 90% of the flies, respectively. For scutellars, boxes that are one-quarter full, half full, or completely full mean that 3 to 4, 2 to 3, or 0 to 1 scutellar bristles, respectively, are present. Their number was calculated as an average among ca. 100 scored flies. The phenotypes of sc⁶, sc⁴, sc^260-15, and sc^s1 are as published previously (5).

To study the molecular structure of the yellow-ac-sc region in y^2ns1 sc^mes1 flies, we isolated the DNA fragment containing the yellow gene and AS-C. A genomic library was prepared from y^2ns1 sc^mes1 flies. The HindIII-BamHI fragment of the yellow locus and the BglII-EcoRI fragment of AS-C (see Fig. 3B) were used as probes to screen the library. Several recombinant phages hybridized with both probes, thus indicating inversion of the region containing the yellow gene and AS-C. This was confirmed by restriction mapping and sequencing of the cloned DNA fragments. The structure of the yellow-ac-sc region in y^2ns1 sc^mes1 flies is presented in Fig. 2C and 3A.

The complex mutation y^2ns1 sc^mes1 was generated by an inversion of the region between the P2 element and the P4 element located at the position 24.5 kb according to the physical map of the AS-C (5). As a result, at one boundary of the inversion the P2 element became linked to the P3 element (2.9 kb) and was oriented opposite to P1 and P2. The other end of the inversion is made up of the regulatory portion of the yellow gene interrupted at the position −69 bp. At this boundary, the inversion is flanked by the P4 element (2.2 kb) (Fig. 2C and 3A). Thus, the inversion (47 kb) reverses the orientations of the yellow, ac, and sc genes (see Fig. 8A and B).

FIG. 8.

The schematic presentation of the yellow-ac-sc region in the described yellow, ac, and sc mutants. The P elements, the genomic sequence of the chimeric element, the enhancers (yellow and AS-C enh), and the promoters (small boxes), and the su(Hw)-binding region of gypsy are indicated. They are not presented in scale. The arrows show the direction of transcription of the yellow, ac, and sc genes. The number of “+” signs indicates the approximate level of AS-C expression in the presence or absence of the su(Hw) protein. Range: ++++, wild-type sc⁺ phenotype; −, extreme mutant sc phenotype.

One of the inversion boundaries maps to the same region of AS-C, as in the case of previously described inversions underlying a strong sc phenotype: sc⁴, sc^260-15, and sc^s1 (5). In these mutants the ac and sc gene expression is impaired since all AS-C enhancers are displaced to a distant location. This is believed to be a molecular basis for severe phenotypes (24 [see also Fig. 2A]). On the contrary, y^2ns1 sc^mes1 flies exhibit moderate mutant sc phenotype, suggesting that ac and sc genes are repressed incompletely (Fig. 2D). The AS-C enhancers seem to retain their activity toward the ac and sc genes, although the inversion displaces them from the promoters. On the other hand, these flies are devoid of pigmentation of the wings, the body cuticle, the bristles of the notum, the legs, and partially the wings. This suggests a strong inhibition of the yellow transcription in the new vicinity.

Both y^+ns and y^2ns1 sc^mes1 carry the gypsy mobile element in the regulatory region of the yellow locus. The su(Hw)-binding region located in the 5′ regulatory portion of gypsy is the typical, best-characterized insulator of Drosophila melanogaster (3, 22, 26, 27, 45, 49). In y^2ns1 sc^mes1 mutants, the gypsy element is excluded from the inverted region (Fig. 2C and 3A). To determine the contribution of the su(Hw) insulator to the sc phenotype, we combined the y^2ns1 sc^mes1 background with the su(Hw)² and su(Hw)^v mutations where the su(Hw) gene is inactivated (25). Unexpectedly, the su(Hw)²/su(Hw)^v transheterozygous flies exhibited a strongly suppressed mutant sc phenotype: only humeral (HU), partially anterior orbital (AOR), and anterior notopleural (ANP) bristles were missing. All other groups of bristles were completely restored (Fig. 2D). This suggests that su(Hw) insulator, which does not physically separate the AS-C enhancers from the respective promoters, facilitates the suppression of transcription.

We also combined a mutation in the mod(mdg4) gene, mod(mdg4)^1u1, with the y^2ns1 sc^mes1 background. The product of the mod(mdg4) gene mediates the action of su(Hw) protein (4, 14, 16, 18, 50). The homozygous mod(mdg4)^1u1 mutation produced a similar, although slightly milder suppressive effect on the sc phenotype of y^2ns1 sc^mes1 flies than did su(Hw)²/su(Hw)^v (Fig. 2D). When su(Hw) and mod(mdg4) mutations were tested in combination, they suppressed the mutant y^2ns1 sc^mes1 phenotype to the same extent as did the su(Hw)²/su(Hw)^v mutation alone (Fig. 2D).

The effects of su(Hw) and mod(mdg4) mutations are similar to those described previously for the sc^D1 mutation (16). This mutation was generated by the insertion of the gypsy retrotransposon into AS-C at position 0 according to the physical map (5) (Fig. 2A). In this case, the su(Hw) protein blocks only those AS-C enhancers which are separated from the promoter by gypsy in the sc^D1 allele. The insertion of gypsy between the yellow and ac genes in the sc^3B allele does not lead to the sc phenotype (5, 16). On the contrary, the y^2ns1 sc^mes1 mutant, in which the su(Hw)-binding region is located much farther from the AS-C genes and their enhancers, does display the sc phenotype.

The data obtained here suggest that the su(Hw) insulator disrupts the effect of AS-C enhancers even without physical interference between these enhancers and the promoters of the inverted genes in the y^2ns1 sc^mes1 mutant.

Structural and functional analysis of the domains of the su(Hw) protein in relation to the insulator function.

The amino- and the carboxy-terminal acidic domains and the region homologous to the leucine zipper motif have been shown to be essential for the insulator function of the su(Hw) protein (25). To determine whether the su(Hw) protein uses the same domains to suppress the ac and sc gene expression in the y^2ns1 sc^mes1 mutant, we analyzed the correlation between mutations in the su(Hw) protein and the phenotypes of y^2ns1 sc^mes1 flies (Fig. 2D).

For this purpose we introduced a construct containing the su(Hw) gene lacking both acidic domains into y^2ns1 sc^mes1 flies. Such a protein is believed to lose its insulator function (25). As expected, su(Hw)^NoAD relieves the y^2ns1 sc^mes1 mutant phenotype, as in the case of the su(Hw)²/su(Hw)^v transheterozygotes, where the su(Hw) gene was completely inactivated. On the contrary, deletion of either the N-terminal [su(Hw)^Δ100] or C-terminal [su(Hw)^j] acidic domain only slightly affects the y^2ns1 sc^mes1 mutant phenotype. These data support the previous observation that the acidic domains are redundant in facilitating the insulator function of su(Hw) protein. The su(Hw)^Δ283 construct carries the su(Hw) gene in which the leucine zipper domain is deleted. This deletion leads to moderate suppression of the sc phenotype of y^2ns1 sc^mes1 flies. Similar results have previously been obtained for the sc^D1 mutation (16). This suggests that the same domains of the su(Hw) protein are required for the repression of the ac and sc gene expression in the y^2ns1 sc^mes1 strain as was described for other known mutations generated by the gypsy insertion.

The su(Hw) insulator located within the yellow locus is the only prerequisite for suppression of the ac and sc genes in the y^2ns1 sc^mes1 mutant.

The available data demonstrate that the su(Hw) protein acts in a directional fashion: only those enhancers which are separated from the respective promoter by the su(Hw)-binding region are affected (3, 8, 22, 27, 45, 49). Our data suggesting that inactivation of su(Hw) protein leads to suppression of the sc phenotype in y^2ns1 sc^mes1 flies contradict the previous observations. In order to rule out the possibility that some additional su(Hw) insulators may be present within AS-C, we examined the region by genomic Southern analysis.

An 80-kb region of the genome, including the yellow gene and AS-C, was probed with the inserts contained in the λsc phages λsc133, λsc112, λsc101, λsc94, λsc64, λsc22, and λsc17 (5). We observed no changes in the restriction pattern. This finding confirms that no other insertions containing su(Hw) insulator were present in the AS-C region of the y^2ns1 sc^mes1 strain.

We also analyzed nine y^+ns sc^+s derivatives of y^2ns1 sc^mes1 (Fig. 1) which showed complete reversion of the sc mutant phenotype and a pigmentation pattern similar to that of the original y^+ns flies (Fig. 3C). The structure of the yellow-ac-sc region in these flies was determined by Southern blot analysis. We found that reversion of the region between the P elements had occurred (Fig. 3B). The AS-C inversion boundary of six y^+ns sc^+s derivatives was made up of two P elements (P3 and P4) oriented in a head-to-head fashion. Only one P element (P4) was found in the other three y^+ns sc^+s derivatives in this location. Introduction of the su(Hw)²/su(Hw)^v transheterozygous and homozygous mod(mdg4)^1u1 mutations had no effect on the sc phenotype of the y^+ns sc^+s flies (Fig. 3C). All of these data indicate that no additional su(Hw)-binding region is present within the AS-C. Therefore, suppression of AS-C enhancers by the su(Hw) protein may be attributed only to the inversion which disrupts normal structure of AS-C.

To obtain the direct evidence that the su(Hw) insulator located in the control region of the yellow gene determines the sc mutant phenotype, we selectively removed the su(Hw)-binding region from the yellow locus. It was deleted by X-ray irradiation of y^2ns1 sc^mes1 males. Approximately 27,000 flies from the progeny were analyzed. Among these we found one, y^2ns14 sc^1x1 (Fig. 1), which exhibited a strongly suppressed sc phenotype. To examine the structure of the yellow-ac-sc region, we used Southern blot analysis, which revealed that the probed region of y^2ns14 sc^1x1 flies differed from the parental genotype by the deletion of gypsy as a result of recombination between the long terminal repeats (LTRs) (Fig. 3A; see also Fig. 8C). These mutants displayed an sc phenotype identical to that observed when the su(Hw)²/su(Hw)^v mutation was introduced into y^2ns1 sc^mes1 flies. When the su(Hw)²/su(Hw)^v transheterozygous mutation was combined with the y^2ns14 sc^1x1 background lacking the insulator, the sc phenotype remained unchanged (Fig. 3C; see also Fig. 8C). These data provide direct evidence that the su(Hw)-binding region located in the yellow locus is responsible for the repression of the ac and sc genes in the y^2ns1 sc^mes1 strain.

The 22.1-kb chimeric element inserted next to gypsy has no influence on su(Hw)-mediated repression of AS-C enhancers.

As mentioned above, the y^2ns1 sc^mes1 mutant contains a chimeric element between the su(Hw)-binding region and the closest inversion boundary. The element is constituted by a 19.7-kb genomic sequence flanked by P elements. The genomic sequence contains the regulatory region which activates the yellow gene expression. To show that the chimeric element does not interfere with the suppressive effect of su(Hw) on ac and sc gene expression, we removed the chimeric element from the locus. For this purpose, we used one of the y^2s sc^+s mutants (Fig. 1). This mutant carries gypsy, a 1.2-kb P element in the yellow locus (P1), and two P elements (P3 and P4) in AS-C (Fig. 4A). y^2s sc^+s flies have cuticle pigmentation identical to that of y² flies, since the su(Hw)-binding region blocks the enhancers of the yellow gene responsible for pigmentation of the body and wings. The flies display no sc phenotype, which confirms that two P elements of AS-C do not affect the ac and sc gene expression.

FIG. 4.

Structure and phenotypes of y^2s sc^+s and its derivatives. (A) Structure of y^2s sc^+s and its derivatives y^2ns21 sc^mes21 and y^2ns26 sc^mes26. Designations are as described in Fig. 3A. (B) Phenotypes of the indicated sc mutations in male flies and their interaction with mutations in the su(Hw) and mod(mdg4) genes. All designations are as described in Fig. 2D.

To induce inversion of the yellow-ac-sc region in these flies, we mobilized P elements and obtained eight y^2ns sc^mes derivatives (Fig. 1) which had the same phenotype as the y^2ns1 sc^mes1 flies. To determine the structure of the inverted region, we used Southern blot analysis. This revealed that five of the derivatives harbored the inverted region between P1 and either P3 (y^2ns21 sc^mes21) or P4 elements (Fig. 4A). In the other three, the inversion occurred together with deletion of either one (data not shown) or both P2 and P3 elements (y^2ns26 sc^mes26) from the region between gypsy and AS-C (Fig. 4A).

Transheterozygous su(Hw)²/su(Hw)^v and homozygous mod(mdg4)^1u1 mutations were introduced into y^2ns21 sc^mes21 and y^2ns26 sc^mes26 flies carrying the inversion. The progeny displayed a suppressed sc mutant phenotype similar to that of the original y^2ns1 sc^mes1 flies containing the same su(Hw) mutations (Fig. 4B; see also Fig. 8D).

All of these results suggest that the 22.1-kb chimeric element inserted next to gypsy does not interfere with the suppressive effect of su(Hw) insulator on ac and sc gene expression.

Suppression of ac and sc gene expression is not mediated by insulation of yellow enhancers.

To rule out the possibility that the su(Hw) insulator suppresses ac and sc gene expression by blocking yellow rather than AS-C enhancers, we displaced the latter by inverting the AS-C regulatory region (see Fig. 8E). Flies with displaced AS-C enhancers were obtained as a two-step process. First, we used three independent y^+ns sc^+s lines derived from y^2ns1 sc^mes1. These mutants displayed a reversed sc phenotype and a pigmentation similar to that of the original y^+ns flies (Fig. 3C). Then we mobilized the P elements in these lines. Among the progeny, we selected 11 mutants (Fig. 1) with a prominent sc phenotype. According to the phenotype, these mutants were divided into two groups: sc^eas (less-prominent sc phenotype) and sc^ebs (more-prominent sc phenotype) (Fig. 5A). The phenotype of these mutants (Fig. 5D) was identical to that of sc⁴, sc^260-15, and sc^s1 flies, where AS-C enhancers were displaced due to inversion of the AS-C regulatory region (Fig. 2A and 5D). We analyzed the structure of AS-C in the sc^eas and sc^ebs mutants by Southern hybridization. We found that inversion of the region between AS-C P elements and another locus of the X chromosome underlies the extreme sc phenotype of the y^+ns sc^eas and y^+ns sc^ebs mutants. Both boundaries of the inversion are constituted by either one or two P elements (Fig. 5A).

FIG. 5.

Schematic presentation and phenotypes of mutants induced by the removal of the AS-C regulatory elements. (A and B) Structure of y^+ns1 sc^ebs1 and y^+ns4 sc^eas4 (A) and their derivatives (B). (C) Structure of y^+s211 sc^eas211 derivative. Only P elements indicated by thick arrows or by boxes are given in the scale, and their restriction maps are presented. (D) Phenotypes of the indicated sc mutations in male flies and their combinations with mutations in the su(Hw) and mod(mdg4) genes. All designations are as described in Fig. 2D. BR means PPA, posterior orbital, OC, ANP, PV, HU, AOR, PS, APA, ASA, and SC bristles. Abbreviations are as defined for Fig. 2D.

Localization of the boundaries of the inverted region was determined by in situ hybridization on polytene chromosomes from heterozygous y^+ns sc^eas/y⁺ or y^+ns sc^ebs/y⁺ females. The complete P element sequence was used as a probe. In both sc^eas and sc^ebs mutants, AS-C enhancers became separated from the respective promoters as a result of large inversions. In all of the mutants analyzed, the inversion boundaries within AS-C were identical. Two mutants with the other boundary mapped to positions 9A (y^+ns1 sc^ebs1) and 4D (y^+ns4 sc^eas4) were chosen for further mutagenesis. To induce inversion of the region between yellow and AS-C, another round of P-element mobilization was done. The progeny with the inversion were selected based on cuticular pigmentation, since impaired transcription of the yellow gene underlies the yellow coat color. As a result, nine derivatives (y^2ns sc^eas and y^2ns sc^ebs) were selected for further analysis (Fig. 1). The inversion of the region between yellow and AS-C was confirmed by Southern blot hybridization (Fig. 5B). All of the flies retained a prominent sc phenotype, indicating that the ac and sc genes remained inactive.

The obtained inversions as well as the parental backgrounds y^+ns1 sc^ebs1 and y^+ns4 sc^eas4 were combined with mutations in the su(Hw) and mod(mdg4) genes (Fig. 5D). When the su(Hw)²/su(Hw)^v or mod(mdg4)^1u1/mod(mdg4)^1u1 mutations were introduced into parental y^+ns1 sc^ebs1 flies or the derivatives y^2ns11 sc^ebs11 and y^2ns12 sc^ebs12 carrying inversions, no changes in the sc phenotype were observed. Combinations of mutant su(Hw) and mod(mdg4) genes with the parental y^+ns4 sc^eas4 backgrounds or the derivatives y^2ns41 sc^eas41 and y^2ns42 sc^eas42 caused the sc phenotype to become more prominent (Fig. 5D). The latter may be explained by su(Hw)-mediated repression of putative negative regulatory elements which were possibly located at the 4D inversion boundary. The results obtained here confirm that the expression of ac and sc genes is independent of the yellow enhancers and fully relies on those of AS-C.

In addition, we found a mutant which displayed the wild-type pigmentation typical of high level yellow expression, along with the extreme sc phenotype. This mutant (y^+s211 sc^eas211) was derived from y^2ns21 sc^mes21 flies (Fig. 1). The structure of the yellow-ac-sc region was determined by using Southern blot analysis. It was revealed that the y^+s211 sc^eas211 mutant carried two inversions: one of the y-ac-sc region, as in the parental y^2ns21 sc^mes21 line, and the other of the region between the P element flanking yellow and another P element located within the 1A region closer to the telomere (Fig. 5C). The latter brings putative enhancers of the 1A region (17) to the vicinity of the yellow gene. As a result, yellow, ac, and sc genes become surrounded by two sets of enhancers (Fig. 5C). However, ac and sc genes remain inactive as judged by the phenotype. When su(Hw)²/su(Hw)^v mutations were introduced into these flies, the y and sc phenotypes remained unaffected (Fig. 5D). These data provide more grounds for ruling out the role of any other regulatory elements except for the AS-C enhancers in activation of ac and sc gene expression.

The yellow promoter is responsible for inhibition of HU, ANP, and AOR enhancers of AS-C.

As shown above, inactivation of su(Hw) protein in the su(Hw)²/su(Hw)^v transheterozygote did not completely suppress the sc mutant phenotype of y^2ns1 sc^mes1 flies: the HU bristles and partially the AOR and ANP bristles were missing (Fig. 2D). y^1s sc^ms is the most frequently encountered derivative of the y^2ns1 sc^mes1 background (Fig. 1). In the y^1s sc^ms flies the yellow gene is completely inactivated (y-null mutation), while the sc mutant phenotype is slightly suppressed. In contrast to the y^2ns1 sc^mes1 flies, introduction of either su(Hw)²/su(Hw)^v transheterozygous or homozygous mod(mdg4)^1u1 mutations into y^1s sc^ms flies completely suppressed the mutant sc phenotype (Fig. 6B).

FIG. 6.

Role of the yellow promoter and the P-element sequences in the sc mutant phenotype in the case of an inversion. (A) The structure of y^1s sc^ms mutations. The figures indicate the number of P-element nucleotides at the breakpoints of internal P-element deletions according to the P-element sequence (41) and at the breakpoints of deletions in the yellow gene according to the yellow sequence (20). The localization and direction of the primers in the yellow gene (y), AS-C (sc), and P element (p) used for PCR are shown by arrowheads. The nucleotides remaining after P-element excision in five sequences y^1s sc^ms derivatives of y^2ns1 sc^mes1 are shown in brackets. The DNA sequence at the junction is presented in such a way that a slash separates the sequences assigned to both sides of the breakpoint. The uppercase letters represent the known P-element sequences, and the lowercase letters represent the filler sequences present at the breakpoint. Designations are as described in Fig. 3A. (B) Phenotypes of the indicated sc mutations in male flies and their combinations with mutations in the su(Hw) and mod(mdg4) genes. All designations are as described in Fig. 2D.

To reveal the differences between the parental y^2ns1 sc^mes1 background and the y^1s sc^ms derivative, we used Southern and PCR analyses. Five independent y^1s sc^ms strains were analyzed. We found that internal deletions in the P4 element flanking the yellow promoter underlie the y^1s sc^ms phenotype. The deletions cover practically the whole sequence of the P element except for the inverted terminal repeats (Fig. 6A). All of these results suggest that the P4 element is responsible for the suppression of the AS-C genes in HU, AOR, and ANP bristles. This element also seems to be essential for the expression of yellow.

To clarify the role of the P4 element, we studied 31 y^1s sc^ms mutants. Among these, 30 strains lacked the P4 element. In one strain, y^1s16 sc^ms16, P4 was retained, but the adjacent 637-bp region of the yellow locus was missing, including the yellow promoter (Fig. 6A). When su(Hw)²/su(Hw)^v and mod(mdg4)^1u1 mutations were introduced into y^1s16 sc^ms16 flies, complete suppression of the sc mutant phenotype was achieved (Fig. 6B; see also Fig. 8F). This suggests that the yellow promoter is responsible for the residual sc mutant phenotype in y^2ns1 sc^mes1, su(Hw)²/su(Hw)^v flies.

It should be pointed out that all highly unstable yellow alleles contain the promoter where the portion from −70 to −146 bp is deleted. This underlies a strong decrease in yellow expression. The regulatory region of the inserted P element compensates for the deletion and restores proper activation of the yellow promoter by yellow enhancers (2). This explains why the removal of either the residual yellow promoter or the P4 element has a similar effect.

Similar results were obtained when y^1s sc^ms derivatives of y^2ns21 sc^mes21 (Fig. 1) were analyzed. As mentioned above, y^2ns21 sc^mes21 flies lack the chimeric element (Fig. 6A). Among the y^1s sc^ms derivatives, we found variations in the P4 element structure. The y^1s212 sc^ms212 derivative carried duplicated P4 element with a deleted 5′ end of each copy. The y^1s217 sc^ms217 derivative had a small deletion (from 14 to 201 bp of the sequence) in the 5′ regulatory region of the P4 element (41). Despite the presence of an almost complete P4 sequence, the su(Hw)²/su(Hw)^v and mod(mdg4)^1u1 mutations completely suppressed the mutant sc phenotype (Fig. 6B) in these flies. This indicates that the 5′ regulatory region is the portion of the P4 element responsible for the activation of yellow and the repression of the ac and sc genes.

To address the role of P element sequences and the yellow promoter in the case of uninverted AS-C, we used y^2s34 sc^1s4 derivatives of the y^2s sc^+s strain (Fig. 1 and 4A). The y^2s34 sc^1s4 flies carry a duplication of the yellow sequence from position −69 bp throughout the whole yellow gene oriented from P3 toward P4 (Fig. 7A). These flies display a weak mutant sc phenotype, showing impaired formation of HU and partial formation of AOR bristles (Fig. 7B). Introduction of the su(Hw)²/su(Hw)^v transheterozygous mutation did not alter the sc phenotype. This allows a conclusion that the su(Hw)-binding region of gypsy is not responsible for the sc mutant phenotype in flies where the AS-C inversion did not occur.

FIG. 7.

The role of the yellow promoter and the P4-element sequences in the sc mutant phenotype without an inversion. (A) Structure of the y^2s34 sc^ls4 mutation and its derivatives. The localization and direction of primers in the yellow gene (y), AS-C (sc), and P element (p) used for PCR are shown by arrowheads. Designations are as in Fig. 3A and Fig. 6A. (B) Phenotypes of the indicated sc mutations in male flies and their combinations with mutations in the su(Hw) and mod(mdg4) genes. Designations are as described in Fig. 2D.

To appreciate the contribution of the yellow promoter to the sc mutant phenotype of y^2s34 sc^1s4 derivatives, we analyzed the sc⁺ revertants selected among the offspring of y^2s34 sc^1s4 flies after P element mobilization (Fig. 1). We used Southern blot and PCR analyses to study the genomic structure of the revertants. It was revealed that one of these lacked the duplicated copy of yellow which separated P3 and P4 elements in y^2s34 sc^+s43 flies (Fig. 7A). Another sc⁺ revertant, y^2s34 sc^+s41, had a deletion of the P3 element flanking the yellow promoter. To rule out the contribution of the P4 element to the sc phenotype, we also analyzed the y^2s34 sc^1s4 offspring with an unaltered sc phenotype. We found a strain in which the P4 element was lost (y^2s34 sc^1s42); however, the sc mutant phenotype was unaffected (Fig. 7A). All of these data allow a conclusion that residual yellow promoter and the 5′ regulatory region of the P element adjacent to the yellow gene are responsible for the inhibition of the HU, ANP, and AOR enhancers of AS-C.

DISCUSSION

Specificity of enhancer-promoter interactions in the yellow-ac-sc gene region.

Complex patterns of ac and sc expression are constructed by separable cis-controlling elements present within a large (ca. 90-kb) region (24). The yellow gene located 10 kb from ac has completely different expression patterns and is activated by different enhancers (5, 20, 21). Therefore, these genes may serve as a good model system for the analysis of proper enhancer-promoter recognition. The recognition may depend on the existence of an interdomain boundary between AS-C and the yellow locus, or it may be determined by the specificity of the proteins assembled on a certain enhancer and promoter.

We describe here an inversion which puts the yellow gene between the ac and sc genes and almost all of their cis-regulatory elements (Fig. 8A, B, and C). This inversion shows only weak interference with the expression of the ac and sc genes. When the su(Hw)-binding region is deleted (Fig. 8C) or inactivated by the su(Hw) mutation, the sc phenotype of the flies is practically indistinguishable from that of the wild type. The presence of the yellow gene between the AS-C enhancers and the promoters of the ac and sc genes does not interfere with ac and sc expression in most areas. Similar results were obtained for the ANP enhancer that can drive expression of the ac and sc genes even when it is located in the regulatory region of the yellow gene (13). A foreign 1A enhancer (Fig. 5D) able to activate yellow expression cannot maintain the ac and sc expression. On the other hand, the yellow gene located in close proximity to the AS-C enhancers is not activated by them. One possible explanation for these results is that the inversion responsible for the studied mutation has a breakpoint very close to the TATA promoter of the yellow gene that impairs the function of the promoter. However, we have found that the P-element insertion and the yellow regulatory sequences from −1 to −69 bp are sufficient for complete and proper activation of the yellow promoter by the yellow enhancers (2). Thus, the obtained results may be better explained by the “promoter specificity” model in the case of promoter-enhancer interactions in the yellow-ac-sc region.

We also found a weak inhibition of ac and sc expression in the case of an inversion or the yellow promoter insertion between the ac and sc promoters and their cis-regulatory elements (Fig. 8B and F). The inhibition depends on the presence of the active yellow promoter that seems to block the corresponding AS-C enhancers. The effect of the deletion of P-element 5′-terminal sequences may also be explained by the above-mentioned data on the functioning of the yellow promoter. The yellow gene in the y^2ns1 sc^mes1 and derivative strains has a deletion spreading from −146 to −70 bp relative to the transcription start site (17). This sequence is needed for the functioning of the yellow promoter. However, the 5′ region of P element between 23 and 108 bp compensates for the effect of the deletion and restores yellow expression (2). This means that the P element acts by reconstituting the yellow promoter activity. One possible conclusion is that most AS-C enhancers may discriminate between their own promoters and the promoter of yellow, while the HU enhancer and, to lesser extent, the AOR and ANP enhancers may also interact with the yellow promoter. Thus, promoter specificity in the yellow-ac-sc region has some minor limitations.

Novel properties of the su(Hw) insulator.

We describe here a novel feature of the su(Hw) insulator in directly blocking the enhancers that are not physically separated by the insulator from their promoters. Numerous control experiments described above show that other sequences are not involved in this process. This observation contradicts the previous results of other authors (22, 45, 49) regarding the strictly directional action of insulators.

The effects of the su(Hw)-binding region on gene expression were explained by assuming that the su(Hw) protein established a domain boundary which limited the activity of the enhancers (18, 19, 45). A domain surrounded by boundaries may prevent the interactions between regulatory elements by promoting the folding of higher-order chromatin in a manner that effectively minimizes the interactions of proteins assembled on an enhancer with proteins assembled on a promoter (19, 23).

Geyer (23) proposed the decoy model that presents insulators as assembling complexes. These complexes may catch an enhancer into a nonproductive interaction because the insulator lacks the promoter function (23, 50). No transcription occurs as a result. The interactions between the insulator decoy and the enhancer are reversible given that the enhancer remains active and may be similar to the normal dynamic interactions between enhancers and promoters (54). An alternative model postulates that the insulator interferes with the activity of proteins that facilitate the enhancer-promoter interactions (38, 39). The authors of that model proposed that there is a special class of proteins, enhancer-facilitators, whose function is to help to form chromatin structures that bring enhancers and promoters closer together.

The ability of the su(Hw)-binding region to directly block the activity of enhancers not separated by the insulator from their promoters is rather difficult to explain in terms of the domain boundary hypothesis. First, su(Hw) insulator does not create any boundary between enhancers and promoters in this case. Moreover, it is separated from the target enhancers and promoters by several other genes with their own enhancers present in the chimeric element and in the yellow gene. It should also be pointed out that our results indicate the absence of any other insulator-like element in the analyzed region: even in the inversion, AS-C enhancers normally activate ac and sc promoters if the gypsy su(Hw)-binding region is eliminated or inactivated. In addition, the ANP enhancer can activate the AS-C promoters transferred to the yellow locus (13), indicating the absence of an insulator between the yellow and scute loci.

The decoy model does not contradict the above results, although neither is it proved by them. The model can merely explain the loss of directionality by the formation of insulator-enhancer (or promoter) contacts due to the appearance of an inversion. It was shown previously that the su(Hw)-binding region in the gypsy-induced sc^D1 and sc^e mutations completely blocked only the enhancers separated by the su(Hw)-binding region from the ac and sc genes (5). In the sc^3B mutation, the gypsy element was located between the yellow and ac genes, and in this mutation the su(Hw)-binding region did not influence the sc phenotype (16). Thus, the su(Hw)-binding region without an inversion does not affect the expression of the ac and sc genes (Fig. 8A). In contrast, the su(Hw)-binding region in y^2ns1 sc^mes1 (Fig. 8B) partially blocks (low probability of bristle formation) even more enhancers than in sc^D1. All of the affected enhancers change their original localization relative to the promoter elements of the ac and sc genes. Therefore, the normal interaction between enhancers and promoters in the AS-C may be partially disrupted and become sensitive to external factors such as the su(Hw) insulator located in the yellow locus.

The mechanism of direct interaction between AS-C enhancers and su(Hw) insulator is not yet clear. One possibility is that the pairing between the P elements located at the breakpoints of the inversion facilitates such interaction. However, the deletions of the P elements on both sides of the inversion as in the y^2ns26 sc^mes26 derivative (Fig. 4A) fail to influence the repression mediated by the su(Hw) insulator. Another possibility is that the inversion brings the su(Hw)-binding region into a close contact with the AS-C cis-regulatory elements due to changes in chromatin folding, which then leads to new long-range contacts between certain chromatin regions. As a result, the su(Hw)-mod(mdg4) complex formed on the su(Hw) insulator becomes capable of interacting directly with enhancer-bound transcription activators or with proteins responsible for enhancer-promoter interactions. The fact that the inactivation of AS-C control elements by the su(Hw)-binding region in the inversion is only partial may be explained by reversible interactions between the insulator and enhancers similar to the normal dynamic interactions observed between enhancers and promoters (54).