Enhancer Blocking by the Drosophila gypsy Insulator Depends Upon Insulator Anatomy and Enhancer Strength

Corresponding author: Pamela K. Geyer, Department of Biochemistry, The University of Iowa, Iowa City, IA 52242., pamela-geyer{at}uiowa.edu (E-mail)

Communicating editor: M. J. SIMMONS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Insulators are specialized DNA sequences that prevent enhancer-activated transcription only when interposed between an enhancer and its target promoter. The Drosophila gypsy retrotransposon contains an insulator composed of 12 degenerate binding sites for the Suppressor of Hairy-wing [Su(Hw)] protein that are separated by AT-rich DNA possessing sequence motifs common to matrix/scaffold attachment regions (MARs/SARs). To further understand mechanisms of insulator function, the parameters required for the gypsy insulator to prevent enhancer-activated transcription were examined. Synthetic binding regions were created by reiteration of a single Su(Hw) binding site that lacked the MAR/SAR motifs. These synthetic binding regions reconstituted insulator activity, suggesting that the property of enhancer blocking may be distinct from matrix association. We found that the number and spacing of Su(Hw) binding sites within the gypsy insulator, as well as the strength of the enhancer to be blocked, were important determinants of insulator function. These results provide a link between transcription and insulation, suggesting that these processes may be mechanistically interconnected.

IN eukaryotes, gene expression often depends upon the regulatory activity conferred by complexes of transcription factors bound to enhancer elements. Enhancers act in a distance- and orientation-independent manner to direct spatial and temporal patterns of transcription. Most enhancers can activate transcription from a wide variety of promoters. This is evidenced by the similar expression patterns of neighboring genes, reflecting the action of shared enhancers (COLEMAN et al. 1987 ; RUSHLOW et al. 1987 ; LOGAN et al. 1989 ), and by the successful application of the enhancer trap technique, which identifies tissue-specific enhancers through the activation of a heterologous promoter (O'KANE and GEHRING 1987 ). Enhancers can exert effects over large distances—in some cases acting over ranges of 80 kb or more (JACK 1985 ; JACK et al. 1991 ; BLACKMAN et al. 1991 ), or across homologous chromosomes, as observed in transvection (WU and GOLDBERG 1989 ; GEYER et al. 1990 ; MORRIS et al. 1998 ). The promiscuous nature of enhancer activity, coupled with an ability to affect promoters from remote locations, indicates that undesirable trans-interactions and cross-regulation of gene expression are possible. Thus, it is likely that constraints on enhancer function exist (EISSENBERG and ELGIN 1991 ).

Insulators are specialized DNA sequences that have properties consistent with a role in limiting enhancer activity (reviewed in EISSENBERG and ELGIN 1991 ; CORCES 1995 ; GEYER 1997 ). First, insulators block enhancer-activated transcription in a position-specific manner. An insulator prevents enhancer function only when interposed between an enhancer and promoter, thereby distinguishing insulators from silencers. Second, insulators protect a gene and its regulatory elements from both positive and negative influences of nearby chromatin. Transgenes flanked by insulators display a similar level of expression, independent of genomic insertion site. Third, some insulators are associated with defined limits of similar chromatin structure, suggesting that functional isolation of gene expression may be related to changes in global chromatin organization. Recent studies indicate that sequences with insulator properties reside within large, complex control regions and are required for appropriate levels of gene expression. For example, several insulators have been identified within the Drosophila bithorax complex that are required to confine both activator and repressor signals to specific regulatory domains, thereby maintaining correct gene expression along the anterior-posterior axis of the developing embryo (GYURKOVICS et al. 1990 ; GALLONI et al. 1993 ; KARCH et al. 1994 ; HAGSTROM et al. 1996 ; ZHOU et al. 1996 ).

Insulators have been identified in several organisms. Several well-characterized insulators in Drosophila include the gypsy insulator (HOLDRIDGE and DORSETT 1991 ; GEYER and CORCES 1992 ; DORSETT 1993 ; ROSEMAN et al. 1993 ; CAI and LEVINE 1995 ; SCOTT and GEYER 1995 ; SIGRIST and PIRROTTA 1997 ; MALLIN et al. 1998 ; GERASIMOVA and CORCES 1998 ); the specialized chromatin structures scs and scs' surrounding the Drosophila 87A7 heat shock genes (UDVARDY et al. 1985 ; KELLUM and SCHEDL 1991 , KELLUM and SCHEDL 1992 ; VAZQUEZ and SCHEDL 1994 ); and the Fab-7 and Mcp elements of the Drosophila bithorax complex (GYURKOVICS et al. 1990 ; GALLONI et al. 1993 ; KARCH et al. 1994 ; HAGSTROM et al. 1996 ; ZHOU et al. 1996 ). In vertebrates, the identified insulators include the human T cell receptor / locus BEAD element (ZHONG and KRANGEL 1997 ); the human apolipoprotein B-100 (apoB) matrix attachment region (MAR; NAMCIU et al. 1998 ); and several mouse and human locus control regions (LCRs) such as the ß-globin LCR (GROSVELD et al. 1987 ; ELLIS et al. 1993 ; MILOT et al. 1996 ), the human CD2 LCR (FESTENSTEIN et al. 1996 ), the human adenosine deaminase LCR (ARONOW et al. 1995 ), the mouse tyrosine LCR (MONTOLIU et al. 1996 ), and the constitutive 5' HS4 site within the chicken ß-globin LCR (CHUNG et al. 1993 ). Interestingly, some insulators function in different species, showing that insulator activity is conserved. For example, both the chicken HS4 insulator and human apoB MAR block chromosomal position effects in Drosophila (CHUNG et al. 1993 ; NAMCIU et al. 1998 ). The chicken HS4 insulator also functions in human erythroid cells (CHUNG et al. 1993 ). Similarly, the Drosophila scs insulator obstructs enhancer-activated transcription in a Xenopus oocyte microinjection assay (DUNAWAY et al. 1997 ) and in human Jurkat cells (ZHONG and KRANGEL 1997 ).

The gypsy insulator was identified through studies focused on understanding the mutagenic effects of the gypsy retrotransposon. A large number of tissue-specific mutations in Drosophila are caused by insertion of gypsy into a gene (MODOLELL et al. 1983 ). In most cases, the mutant phenotypes caused by these insertions are reversed by mutations in the second site modifier gene, suppressor of Hairy-wing [su(Hw); MODOLELL et al. 1983 ; RUTLEDGE et al. 1988 ]. This gene encodes a ubiquitously expressed protein with several functional motifs characteristic of transcription factors, including a zinc-finger DNA binding domain, a leucine zipper-like motif, and amino- and carboxy-terminal acidic domains (HARRISON et al. 1993 ; KIM et al. 1996 ). The Su(Hw) protein binds to a region of the gypsy retrotransposon just 3' of the 5' long terminal repeat (LTR), which contains a cluster of 12 degenerate binding sites (PARKHURST et al. 1988 ; SPANA et al. 1988 ; MAZO et al. 1989 ; DORSETT 1993 ; SHEN et al. 1994 ). Analysis of complete and partial revertants of gypsy-induced mutations indicates that the Su(Hw) binding sites are critical for gypsy mutagenesis (GEYER et al. 1988 ; PEIFER and BENDER 1988 ; SMITH and CORCES 1992 ). Furthermore, the Su(Hw) binding sites alone reproduce the mutagenic effects of the gypsy retrotransposon, demonstrating that this region contains insulator function (HOLDRIDGE and DORSETT 1991 ; GEYER and CORCES 1992 ).

The connection between the degree of enhancer blocking and the number of Su(Hw) binding sites in the gypsy insulator is unclear. Su(Hw) binding regions with as few as four or five binding sites can confer a substantial block of enhancer function (SMITH and CORCES 1992 ; HAGSTROM et al. 1996 ), whereas in another case, a binding region with seven nonoverlapping sites fails to impede enhancer function (FLAVELL et al. 1990 ). These studies suggest that the number of binding sites does not determine effectiveness of an enhancer block and raise the question of what factors influence gypsy insulator function. To address these issues, synthetic Su(Hw) binding regions were generated and tested for their ability to impede communication between the yolk protein fat body enhancer (FBE1) and the yp2 promoter. We found that both the number and arrangement of binding sites impacted the reconstitution of a functional insulator. These experiments provided a context to determine whether the requirements for insulation varied when transcription was directed by an augmented version of FBE1. We found that the activity of a strengthened enhancer, responsible for directing higher levels of transcription, is more difficult to block, indicating that the nature of regulatory interactions within a gene impacts the effectiveness of an insulator. Our results suggest a direct link between insulator function and transcriptional processes.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Synthetic Su(Hw) binding region construction:
The synthetic Su(Hw) binding regions were created by concatamerization of oligonucleotides containing binding site 3 of the natural gypsy insulator (nucleotides 732–759; MARLOR et al. 1986 ). Two pairs of single-stranded, 31-bp oligonucleotides (corresponding to the sense and antisense strands) were synthesized to contain overhangs for either the SalI or BamHI (B) and BglII (G) restriction sites (Integrated DNA Technologies). The sequence of the sense strand for the complementary pair of SalI oligonucleotides was 5' TGCAAAAAAATAAGTGCTGCATACTTTTTAG 3'; and the antisense strand was 5' TGCACTAAAAAGTATGCAGCACTTATTTTTT 3'. The sequence of the sense strand of the complementary pair of B/G oligonucleotides was 5' GATCAAAAAATAAGTGCTGCATACTTTTTAG 3'; and the antisense strand was 5' GATCCTAAAAAGTATGCAGCACTTATTTTTT 3'.

Oligonucleotides were resuspended in STE buffer (50 mM NaCl, 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, pH 8.0) at a concentration of 0.1 OD units/µl and equal molar amounts of each were mixed, heated to 100° for 5 min, and cooled slowly to room temperature to promote annealing. Annealed oligonucleotides (2.7 OD units) were ligated in a 30-µl reaction. During the ligation of the BamHI/BglII oligonucleotides, 50 units each of the BamHI and BglII restriction endonucleases were included to force directional concatamerization. After ligation, the reaction mixture was fractionated by electrophoresis in a 2% agarose gel. The desired concatamers were isolated, purified through filtered pipette tips, phenol/chloroform extracted, and concentrated by ethanol precipitation.

The 3:R1 and 3:R3 synthetic binding regions were made using oligonucleotides with SalI overhangs. Concatamers containing one or three Su(Hw) binding sites were subcloned into a modified pUC18 plasmid vector in which the XbaI site was changed to HindIII. The 3:R4 and 3:R6 synthetic binding regions were made using oligonucleotides with BamHI/BglII overhangs. To make R3sR3, the R3 HindIII fragment was isolated and dimerized. The R3ssR3 and R3sssR3 synthetic Su(Hw) binding regions were made by PCR amplification of 85- and 147-bp fragments, respectively, from pBR322 using primers containing PstI sites: 5'pBR322, 5' GAACTGCAGGGTTATTGTCTCATGAGC 3' (anneals at 4188–4209 in pBR322); 3'pBR322#1, 5' GAACTGCAGTTTCGGGGAAATGTGC 3' (anneals at 4257–4273 in pBR322); 3' PBR322#2, 5' GAACTGCAGGATACGCCTATT 3' (anneals at 4325–4337 in pBR322). PCR products were digested with PstI, subcloned into the PstI site present between R3 subunits of the R3sR3 synthetic Su(Hw) binding region, and inserted into the FBE1 cassette plasmid. In this way the distance between centers of core Su(Hw) binding sites in R3sR3 was increased from 62 to 147 bp in R3ssR3 and to 211 bp in R3sssR3.

Construction of yp2 reporter genes:
The plasmid vector p[CR2] contains 1.3 kb of yolk protein DNA representing the yp intergenic region and includes the promoter region of the yp1 and yp2 genes (LOGAN and WENSINK 1990 ). The effects of the synthetic Su(Hw) binding regions were tested using a derivative of p[CR2] that contained only FBE1 and the yp2-lacZ fusion gene. This transgene was made by digesting p[CR2] with restriction enzymes EcoRI and HindIII and isolating a 4.4-kb fragment that contains 335-bp upstream yp2 promoter sequences, the yp2 promoter, and the yp2 5' untranslated region fused to the bacterial lacZ gene. This fragment was cloned into a modified pBluescript vector in which the KpnI site was changed to an XbaI site, thereby creating the plasmid yp2-lacZ-pBlue. The FBE1 enhancer was isolated from p[CR2] as an EcoO109I and BglII fragment and blunt-end ligated into the Klenow-repaired XhoI site of yp2-lacZ-pBlue plasmid, creating a plasmid called FBE1 cassette. This places FBE1 365 bp upstream of the yp2 transcription start site. Each Su(Hw) binding region was subcloned into a HindIII site, positioned between FBE1 and the yp2 promoter, 335 bp upstream of the yp2 transcription start site. In some constructs, the HindIII site was changed to a BglII site to facilitate insertion of the synthetic binding regions. The FBE1-REP cassette plasmid was made by blunt-end ligation of the 12 Su(Hw) binding sites into the HindIII site of FBE1 cassette. These binding sites were isolated from the REP 1 plasmid, which contains gypsy sequences between nucleotides 647 and 1077 (as numbered in MARLOR et al. 1986 ) inserted into pUC18.

Construction of the plasmid called Mega cassette was done in the following manner. The EcoO109I-BglII fragment containing the 140-bp FBE1 enhancer was blunt-end ligated into the HindIII site of the pSP73 plasmid (Promega, Madison, WI) and subsequently removed as a SalI and XhoI fragment. This fragment was cloned into the XhoI site of yp2-lacZ-pBlue under conditions of excess FBE1. One clone containing four direct repeats of the FBE1 enhancer was selected and the plasmid was designated Mega cassette.

Synthetic Su(Hw) binding regions containing one or three Su(Hw) binding sites were removed from the modified pUC18 plasmid by digestion with HindIII and cloned into the HindIII site of the FBE1 cassette or Mega cassette plasmid. Concatamers containing 3:R4 and 3:R6 were cloned directly into the BglII site in the FBE1 cassette or Mega cassette plasmids. To ensure stability of the synthetic binding regions, plasmids containing 3:R4 and 3:R6 were grown in the STBL2 bacteria host (GIBCO BRL, Gaithersburg, MD), which is recombination deficient. All synthetic binding regions were sequenced to verify their integrity.

The yp2-lacZ fusion genes carrying the natural or synthetic Su(Hw) binding regions were removed as XbaI fragments from the pBluescript vector and subcloned into the XbaI site of the P-element transformation vector pCaSpeR (FBE1 cassette derivatives) or a modified pCaSpeR vector, RR3, which contains a gypsy insulator at the 3' end of the white gene (Mega cassette derivatives). In all cases, clones were selected in which the yp2 transcription unit was in the divergent orientation relative to the white gene in the transformation vector. Before injection, each plasmid was sequenced to ensure that the synthetic Su(Hw) binding region was intact using a yp primer that annealed to sequences located -318 bp relative to the yp2 transcription start site (5' CATGCACAGGTCAAG 3'). Plasmid DNA isolation and DNA manipulations were carried out by standard procedures (AUSUBEL et al. 1994 ).

Genetic manipulations:
Flies were raised at 25°, 70% humidity on standard corn meal/agar medium. Germline transformation was carried out as described by RUBIN and SPRADLING 1982 . Two host strains were used in these experiments. The first was y^-ac^-w^-. These flies carry a deletion of the yellow and achaete loci and the w¹¹¹⁸ deletion (HAZELRIGG et al. 1984 ). The second host strain was y^-ac^-w^-; Sb ry⁵⁰⁶ P[ry⁺2-3](99B)/TM6, where P[ry⁺2-3](99B) provided a chromosomal source of P transposase. DNA concentrations used in these experiments were 400 µg/ml of the FBE1 cassette or Mega cassette derivative construct and 200 µg/ml of the "wings clipped" helper plasmid p25.7 (KARESS and RUBIN 1984 ). Transformants, recognized by a dark eye color phenotype, were used to establish stocks. Additional independent insertion lines were obtained by mobilizing the transposon in one line with the Sb ry⁵⁰⁶ P[ry⁺2-3](99B) chromosome (ROBERTSON et al. 1988 ). For each independent line, the number of insertions and the integrity of the transposon were determined by DNA Southern analysis. Only lines with single insertions were analyzed further and at least three lines per construct were studied.

To determine the effects of the Su(Hw) protein on yp gene expression, lines containing the FBE1-cassette or Mega-cassette derivatives were crossed into a su(Hw)^v/su(Hw)^f mutant background. This combination of su(Hw) alleles reverses the phenotypes associated with gypsy insertions and is female fertile. The su(Hw)^v mutation is a small deletion that encompasses a portion of the su(Hw) gene (HARRISON et al. 1992 ), whereas su(Hw)^f is a point mutation in one of the Zn fingers, which retains some ability to bind DNA (HARRISON et al. 1993 ). Transgenic flies carrying either first or second chromosome insertions were crossed into a su(Hw) mutant background as described previously (ROSEMAN et al. 1995 ).

ß-Galactosidase spectrophotometric assay:
The level of yp2 promoter activity was assessed using quantitative ß-galactosidase assays, performed essentially as previously described (SIMON and LIS 1987 ; SCOTT and GEYER 1995 ). Extracts were isolated from flies grown under uncrowded conditions to ensure similar body size. This was achieved by mating two male flies carrying the transgene of interest (either p[CR2] control or various FBE1 or Mega experimental transgenes) to three female flies of the host strain y^- ac^- w^-. After 3 days, the parents were transferred to a clean vial and discarded on day 6. Carcass extracts were isolated in the following manner. Females were collected within 24 hr of eclosion and aged for 2 days on food supplemented with yeast paste. Healthy females of equivalent size were assayed. For each experiment, 10–20 carcasses were used, with amounts of extract assayed depending upon the transgenic fly line tested. Quantities were increased for lines carrying a transgene with low-level yp2 expression and decreased for lines carrying transgenes with a high level of yp2 expression. Preliminary studies were undertaken to determine an appropriate dilution for each experimental transgene such that the values obtained were in the linear range of determinations as established for the control p[CR2]. ß-Galactosidase activity assay was started by adding 180 µl of a 1 µg/ml solution of the substrate chlorophenol red-ß-D-galactopyranoside (CPRG) in assay buffer. Optical density was taken every 15 min for 1 hr at 595 nm in a Titertek Multiscan MCC/340 microtiter plate reader. Activity was expressed as the change in OD/min/fly and normalized to extract from the control, p[CR2], which was arbitrarily assigned a value of 100 activity units (au; SCOTT and GEYER 1995 ). At least three transformed lines were analyzed for each construct. Each transformed line was assayed using extracts isolated at least three different times. Each extract was assayed in triplicate, and the error between these samples was <10%. Average promoter activity and standard deviation were determined using the statistical analysis feature of the Microsoft Excel program.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The natural gypsy insulator contains 12 degenerate binding sites for the Su(Hw) protein (Fig 1). Each binding site corresponds to a 12-bp core sequence (consensus, 5' PyPuTTGCATACCPy 3') that is separated from the next by a short, variable AT-rich sequence. The distance between centers of each core binding site is variable, ranging between 26 and 28 bp for most sites, and 35 bp between the centers of sites 3 and 4 (Fig 1; MARLOR et al. 1986 ; SPANA and CORCES 1990 ; DORSETT 1993 ). To understand the underlying requirements for formation of an insulator, we created synthetic Su(Hw) binding regions and tested their effects on enhancer-promoter communication. The synthetic binding regions were generated by concatamerization of a 31-bp oligonucleotide corresponding to the third Su(Hw) binding site (BS3, Fig 1B). We selected BS3 because the interaction of Su(Hw) protein with these sequences was well characterized (SPANA and CORCES 1990 ). The BS3 oligonucleotide included all of the sequences necessary for the association of the Su(Hw) protein, specifically the core binding site and L and R bends.

View larger version (38K): In this window In a new window Download PPT slide   Figure 1. The gypsy insulator and synthetic Su(Hw) binding regions. (A) The sequence of the 340-bp gypsy insulator isolated from the gypsy retrotransposon is shown (MARLOR et al. 1986 ). The 12 core binding sites are boxed. (B) The sequence of the oligonucleotide used to produce the synthetic Su(Hw) binding regions. This oligonucleotide contains gypsy insulator sequences from 732 to 759 (MARLOR et al. 1986 ) corresponding to binding site 3 and includes the core sequence and R and L bends (SPANA and CORCES 1990 ). (C) The structure of the yolk protein fusion reporter gene that contains one or three synthetic Su(Hw) binding sites (solid rectangles) inserted 335 bp upstream from the start site of yp2 transcription is shown, at a position between FBE1 (oval) and the yp2 promoter (bent arrow). The yp2 promoter is fused to the bacterial lacZ gene (open rectangle).

The function of each synthetic binding region was assayed using the yolk protein 2-LacZ fusion gene as a reporter (LOGAN and WENSINK 1990 ; SCOTT and GEYER 1995 ). This gene is a useful experimental system for several reasons. First, the transcriptional activity of the yp2 promoter can be measured indirectly using a quantitative enzymatic assay for ß-galactosidase. Second, sex-and tissue-specific expression of the reporter gene can be monitored histochemically, providing an internal control for appropriate gene expression. Third, enhancers that control yp gene expression are well characterized, allowing an evaluation of the effects of the gypsy insulator in the context of a homologous enhancer-promoter interaction. We studied the effects of synthetic binding regions on the major fat body enhancer, FBE1 (GARABEDIAN et al. 1985 , GARABEDIAN et al. 1986 ; ABRAHAMSEN et al. 1993 ; SCOTT and GEYER 1995 ). Previous experiments demonstrated that the gypsy insulator effectively blocks FBE1 activity without altering the stage- or tissue-specific yp2 expression (SCOTT and GEYER 1995 ).

Synthetic Su(Hw) binding regions block the yolk protein fat body enhancer:
We first determined the threshold number of binding sites required to impede FBE1 activity. To accomplish this goal, synthetic binding regions with one, three, four, or six copies of BS3 were cloned between FBE1 and the yp2 promoter (Fig 1C). The resulting yp2 genes were inserted into a CaSpeR transformation vector and transformed into flies by P-element-mediated germline transformation (RUBIN and SPRADLING 1982 ). Transgenic lines were established and those containing single insertions were characterized. In all experiments, the level of yp2 promoter activity was quantitated using ß-galactosidase assays on extracts obtained from ovariectomized transgenic females to examine yp2 expression only in the fat body.

Two control transgenes were constructed to provide an assessment of the level of yp2 promoter activity associated with enhancer-activated (FBE1-ctrl) and enhancer-blocked (FBE1-REP) transcription. The FBE1-ctrl transgene contained FBE1 positioned 365 bp upstream of the yp2 transcription start site. Four independent transformed lines carrying the FBE1-ctrl transgene were established. The average level of yp2 activity was determined to be 144 au (Fig 2), which was not significantly affected by the Su(Hw) protein. This was demonstrated by crossing the FBE1-ctrl transgene of one line into a su(Hw) mutant background. The su(Hw) mutant flies had an activity of 121 au relative to 140 au found in su(Hw)⁺ flies (Fig 3). The value of 144 au was considered to represent a wild-type level of FBE1 function. The second control, FBE1-REP, contained the natural gypsy insulator inserted 335 bp upstream of the yp2 transcription start site in a position between FBE1 and the yp2 promoter. The average activity determined from three independent lines was 6 au. In a su(Hw) mutant background, the level of yp2 promoter activity in two FBE1-REP transgenic lines increased to a value similar to that observed in FBE1-ctrl (137 au or 95% FBE1-ctrl; Fig 3). These results indicate that the block of FBE1 activity is dependent upon the presence of the Su(Hw) protein and that reduced promoter activity in FBE1-REP is not an indirect effect of increased distance between yp control elements or the result of a genomic position effect. The level of yp2 expression remaining in FBE1-REP flies is similar to that observed previously (SCOTT and GEYER 1995 ) and may reflect either an incomplete block of enhancer function or residual, nonenhancer-activated transcription corresponding to basal transcription.

View larger version (49K): In this window In a new window Download PPT slide   Figure 2. The enhancer blocking effects of synthetic Su(Hw) binding regions. (Left) The structures of the yp2-lacZ fusion genes carried by transgenic flies that were assayed for yp2 promoter activity are shown. (Right) The average ß-galactosidase activity units (au) are shown for each transgene. An activity unit corresponds to the rate of change in OD595 per minute per fly and was expressed relative to the activity of the yp2 promoter in the p[CR2] reference construct, which was arbitrarily assigned a value of 100 au. Each independent line contained a single copy of the integrated transgene and was assayed at least three times. The average of these independent experiments is represented by the shaded bars. Standard deviations between independent determinations are shown. The vertical dotted line represents the average activity of the four FBE1-ctrl lines (144 au). The symbols are described in the legend to Fig 1, except that the solid triangle represents the natural gypsy insulator isolated from the gypsy retrotransposon.

View larger version (47K): In this window In a new window Download PPT slide   Figure 3. Effects of a su(Hw) mutant background on the enhancer blocking activity of synthetic Su(Hw) binding regions. Several independent lines containing a single copy of each transgene indicated on the left were selected and crossed into a su(Hw)v/su(Hw)f mutant background. ß-Galactosidase activity was determined for each independent line and is represented by the solid bars. For comparison, the ß-galactosidase activity in a su(Hw)+ background is shown as shaded bars.

Insertion of one or three Su(Hw) binding sites between FBE1 and the yp2 promoter [FBE1: binding site 3-Repeated 1 (FBE1:3-R1) or FBE1:3-R3] had no effect on FBE1-activated transcription (Fig 2). The average activity obtained for four independent transformed lines carrying the FBE1:3-R1 transgene was 158 au, approximately the same as the FBE1-ctrl (144 au; Fig 2), which was unchanged in a su(Hw) mutant background (Fig 3). Four independent lines carrying the FBE1:3-R3 transgene had a slightly lower average yp2 promoter activity (90% FBE1-ctrl, 129 au; Fig 2). However, this decreased level of yp2 expression did not represent a partial block of the enhancer because, in the absence of functional Su(Hw) protein, the average level of yp2 promoter activity remained slightly lower than the FBE1-ctrl (128 au or 90% of FBE1-ctrl).

Addition of one more Su(Hw) binding site dramatically changed the activity of the yp2 promoter (Fig 2). Analysis of four FBE1:3-R4-transformed lines showed that insertion of four Su(Hw) binding sites between yp control elements reduced yp2 expression to an average of 6 au (4% of the FBE1-ctrl). In a su(Hw) mutant background, yp2 promoter activity in these transgenic lines increased and was comparable to that of FBE1-ctrl (121 au or 84% FBE1-ctrl), demonstrating that the block of enhancer activity was dependent upon the presence of the Su(Hw) protein (Fig 3). The comparable level of yp2 promoter activity in FBE1:3-R4 and FBE1-REP indicates that the synthetic R4 binding region conferred a complete block of FBE1 activity and was as effective as the natural gypsy insulator. This conclusion was supported by examining the effects of a synthetic binding region containing two more Su(Hw) binding sites (3:R6). Three independent FBE:3-R6 lines had an average level of yp2 promoter activity similar to FBE1:3-R4 (7 au or 5% FBE1-ctrl vs. 6 au or 5% FBE1-ctrl; Fig 2). In two FBE:3-R6 lines tested, yp2 activity increased to that of the FBE1-ctrl in a su(Hw) mutant background (Fig 3). These findings support the conclusion that a binding region containing only four Su(Hw) binding sites completely blocks FBE1-activated transcription.

Altered synthetic Su(Hw) binding regions have reduced enhancer blocking activity:
Our determination that four but not three Su(Hw) binding sites blocked FBE1 function provided an opportunity to address the role of the arrangement of binding sites on insulator function. The natural gypsy insulator is composed of a closely spaced array of binding sites, such that the centers of core sites are usually spaced by 26 to 28 bp. This close clustering was maintained in the synthetic Su(Hw) binding regions, which had 31 bp between centers of core sites. To test whether a close arrangement of binding sites was critical for insulator function, we evaluated the effect of increasing the distance between sites. We reasoned that this question could be addressed by inserting spacer DNA between binding sites 3 and 4 of the synthetic R6 insulator. If the separation of core binding sites was detrimental to insulator activity, then the R6 insulator would be divided into two clusters of three binding sites, which should fail to block FBE1 activity (Fig 2). We modified the R6 synthetic Su(Hw) binding region by inserting spacer DNA between sites 3 and 4, while maintaining the flanking sequences necessary for Su(Hw) association. The resulting synthetic binding regions contained 62 bp (FBE1:3-R3sR3, where "s" represents "spacer"), 147 bp (FBE1:3-R3ssR3), or 211 bp (FBE1:3-R3sssR3) between the centers of Su(Hw) binding sites 3 and 4.

Five independent lines of flies carrying the FBE1:3-R3sR3 transgene were established and analyzed. The average level of yp2 promoter activity in these lines was higher than the FBE1:3-R6 transgene (21% of the FBE1-ctrl, 30 au; Fig 4). These results suggest that the R3sR3 synthetic binding region only partially attenuated FBE1 activity, even though it contained more than a sufficient number of Su(Hw) binding sites to block the enhancer completely. The partial block of FBE1 depended on the Su(Hw) protein because in the two FBE1:3-R3sR3 lines tested, the level of yp2 expression increased to a value similar to the FBE1-ctrl in a su(Hw) mutant background (Fig 4).

View larger version (56K): In this window In a new window Download PPT slide   Figure 4. Effect of increased spacing between binding sites on insulator function. (Left) The structures of the yp2-LacZ experimental transgenes are shown. The number within each triangle indicates the number of nucleotides inserted between the third and fourth core Su(Hw) binding sites. (Right) The ß-galactosidase activity that was determined for each independent line is shown, represented as a shaded bar. Several independent lines were selected and crossed into a su(Hw)v/su(Hw)f mutant background. Each solid bar represents the average ß-galactosidase activity of a single independent line assayed at least three times.

Enhancer blocking activity was completely compromised by further increasing the spacing between core Su(Hw) binding sites. The average yp2 promoter activity in five transgenic lines carrying the FBE1:3-R3ssR3 (138 au or 96% FBE1-ctrl) and four lines carrying the FBE1:3-R3sssR3 (162 au or 113% FBE1-ctrl) transgene was similar to the activity of the FBE1-ctrl, indicating that these synthetic Su(Hw) binding regions failed to block FBE1 (Fig 4). These findings demonstrate that increasing the distance between core centers by as few as 31 bp of DNA (from 31 bp in R6 to 62 bp in R3sR3) interferes with insulator activity. Thus, in addition to the number of Su(Hw) binding sites between an enhancer and a promoter, potent insulator activity depends upon binding site arrangement.

Insulator activity is dependent upon the nature of the enhancer:
Having determined the threshold number of Su(Hw) binding sites needed to block FBE1, we were poised to assess the relationship between enhancer strength and insulator function. To provide an internally controlled system, an augmented version of FBE1 was developed. We reasoned that a stronger FBE1 enhancer could be achieved by increasing the number of transcription factor binding sites. A strengthened enhancer would be evidenced by an increase in the level of yp2 promoter activity. To test this idea, a transgene (Mega-ctrl) was created that contained four copies of FBE1 (Mega enhancer) inserted upstream of the yp2 promoter. Following germline transformation, four independent transformed lines were established and the level of yp2 promoter activity was determined. Consistent with our prediction, flies carrying Mega-ctrl had an average yp2 promoter activity of 305 au (212% of the FBE1-ctrl; Fig 5), indicating that concatamerization of FBE1 strengthened the enhancer. The increase in promoter activity was less than fourfold, as would be predicted if each enhancer was capable of stimulating expression to the same degree as a single copy of FBE1, implying that the yp2 promoter in Mega-ctrl may have reached a maximum level of activity, after which additional enhancers could not elevate transcription. Nonetheless, the Mega enhancer significantly increased the level of yp2 promoter activity; thus it is stronger than FBE1 alone.

View larger version (51K): In this window In a new window Download PPT slide   Figure 5. The effect of enhancer strength on an enhancer block conferred by the synthetic Su(Hw) binding regions. The general structures of the yp2-LacZ fusion genes are shown at the top [FBE1 (left); Mega (right)]. The identity of the specific Su(Hw) binding region inserted between the FBE enhancers and yp2 promoter is shown between the graphs. Activity of the yp2 promoter in a wild-type su(Hw) background is indicated by the shaded bars. Activity in a su(Hw)v/su(Hw)f mutant background is indicated by the solid bars. Each vertical dotted line represents the average activity of either the FBE1-ctrl or Mega-ctrl lines.

Histochemical staining was carried out on adult female and male flies from several Mega-ctrl lines to ensure that the stronger enhancer did not alter the sex- and tissue-specific expression of the yp2 promoter. In adult females, the yp2 gene was expressed in a tissue-specific pattern that was indistinguishable from FBE1-ctrl lines (data not shown). ß-Galactosidase activity was detected in the thoracic and abdominal fat body tissue, as well as ovarian follicle cells (LOGAN et al. 1989 ; LOGAN and WENSINK 1990 ; SCOTT and GEYER 1995 ). In males carrying Mega-ctrl, there was no ß-galactosidase activity other than nonspecific background staining in the gut, which occurs in both sexes (data not shown; GARABEDIAN et al. 1985 , GARABEDIAN et al. 1986 ; SCOTT and GEYER 1995 ). Therefore, the Mega-enhancer did not alter yp2 expression.

To investigate the effects of enhancer strength on insulator function, we tested whether the natural gypsy insulator impeded the activity of the Mega enhancer. Three independent lines containing the Mega-REP transgene were established. We found that flies carrying Mega-REP had an average yp2 activity of 25 au (Fig 5), corresponding to 8% of the Mega-ctrl. This level of yp2 activity represents a significant increase relative to that found in FBE1-Rep lines (average 6 au; P < 0.0015). This surprising fourfold increase suggests that while the natural insulator substantially interferes with the activity of the Mega enhancer, more residual enhancer activity may pass by the natural insulator than occurred in the FBE1-REP transgene, leading to a higher level of yp2 transcription. Thus, the natural insulator may be slightly less effective at blocking the Mega enhancer than it was at blocking the single FBE1. To verify that the reduced yp2 transcription in Mega-REP flies depended on the Su(Hw) protein, one line was selected and crossed into a su(Hw) mutant background. The yp2 activity in a su(Hw) mutant was restored to 300 au (98% of Mega-REP; Fig 5), demonstrating that the observed decrease in expression did not result from a position effect.

We next examined whether synthetic binding regions were capable of blocking the Mega enhancer. Three independent transformed lines were established that carried the R6 synthetic binding region inserted between the Mega enhancer and the yp2 promoter. Flies carrying Mega:R6 had an average activity of 26 au (8% of the Mega-ctrl; Fig 5). The Mega:R6 transgene from one line was crossed into a su(Hw) mutant background, which verified that the decreased expression was dependent upon the Su(Hw) protein (Fig 5).

The effects of the improperly spaced 6-mer (R3sR3) were very different from those found for 3:R6. Transgenic flies carrying Mega:3-R3sR3 had an average level of yp2 promoter activity similar to the value observed for the Mega-ctrl transgene (263 au or 87% Mega-ctrl; Fig 5), demonstrating that R3sR3 is incapable of blocking the Mega enhancer. This result contrasts with the effect of R3sR3 on the single FBE1, where this synthetic insulator provided a substantial impediment to enhancer function. We conclude that the effectiveness of an insulator block is dependent upon the strength of the enhancer-promoter interaction.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Reiteration of a single Su(Hw) binding site constituted enhancer blocking activity:
We sought to determine what factors impact the ability of the gypsy insulator to prevent enhancer-promoter communication. This question was addressed in a systematic way by first defining the features of the gypsy insulator required to impede the function of the well-characterized yolk protein enhancer, FBE1, and then determining whether similar requirements applied to an augmented version of FBE1, the Mega enhancer. Our results provide an understanding of the requirements for enhancer blocking by the gypsy insulator and help explain inconsistencies concerning effects of defective gypsy retrotransposons on gene expression.

A functional insulator was recreated by multimerization of a single gypsy Su(Hw) binding site (site 3). Synthetic binding regions with one or three copies of site 3 had no effect on the FBE1 enhancer, while a binding region with four sites conferred a block of the FBE1 enhancer that was quantitatively the same as that imparted by the natural gypsy insulator (Fig 2). This sharp threshold was unexpected. Previous studies of the effects of gypsy retrotransposons carrying four or five binding sites showed that they produced only a partial block of enhancer function (SMITH and CORCES 1992 ; HAGSTROM et al. 1996 ). There may be several factors responsible for this effect. First, the ability of binding site 3 concatamers to reconstitute insulator function may reflect a distinct property of these sequences. Individual sites within the gypsy insulator possess different affinities for the Su(Hw) protein (KIM et al. 1996 ). Synthetic binding regions containing only high affinity sites would be predicted to recruit the Su(Hw) protein more effectively, thereby increasing insulator potency. The strength of the 3:R4 insulator is consistent with the observation that binding site 3 may have a high Su(Hw) affinity (KIM et al. 1996 ). However, binding site 3 is not the only site within the natural gypsy insulator that participates in Su(Hw) protein association and enhancer blocking. A gypsy insulator deleted for Su(Hw) binding sites 2–6, as found in the ct^K allele, maintains the ability to obstruct enhancer-activated transcription substantially (HOOVER et al. 1992 ). A second factor that may contribute to the sharp threshold may be that FBE1 is a relatively weak enhancer, allowing for its substantial block with a only few binding sites (see below).

The natural gypsy insulator contains several sequences found in previously characterized matrix or scaffold attachment regions (MARs/SARs), including four near-perfect matches to the ATC regions [(T/C)TTTTAATAAA(T/A)A(T/C)ATT and in vitro topoisomerase II cleavage sites [(A/T)A(C/T)ATT; NABIROCHKIN et al. 1998]. The functional significance of these sequences was suggested by in vitro assays that demonstrated nuclear matrix binding and topoisomerase II association of the gypsy insulator (NABIROCHKIN et al. 1998 ). These MAR/SAR motifs are located in the spacer DNA and are not coincident with the core binding sites for the Su(Hw) protein. Synthetic binding regions composed of site 3 lacked these motifs, yet were able to block enhancer-activated transcription. These results suggest that either sequences other than the ATC and topoisomerase II cleavage sites are responsible for matrix localization of the insulator or that matrix attachment and enhancer blocking are separable properties.

The tight threshold for insulation of FBE1 provided an opportunity to examine the requirement for a closely spaced array of Su(Hw) binding sites in enhancer blocking. We found that increasing the space between binding sites strongly compromised insulator function (Fig 4 and Fig 5). Interestingly, insertion of as little as 31 bp (R3sR3) disrupted the strength of the insulator. We believe that it is unlikely that this effect was caused by a change in the positioning of Su(Hw) binding sites along the helical face of DNA. In the natural gypsy insulator, the distance between the centers of the core binding sites ranges, with most binding sites spaced 26 or 28 bp apart, causing every other Su(Hw) binding site to align on the same face of the helix. In contrast, in the 3:R6 insulator binding sites are centered 31 bp apart, placing every Su(Hw) site on the same face of the helix (Fig 1). As both the natural and 3:R6 insulators conferred equivalent blocks (Fig 2 and Fig 5), this would suggest that the relative positioning of binding sites on the DNA helix does not impact the degree of insulation conferred. Additionally, it is unlikely that an altered nucleotide composition in the spacer DNA was responsible for the diminished insulator activity. In the natural and synthetic R6 insulators the spacing is AT rich, ranging from 71 to 93% AT. In the R3sR3, R3ssR3, and R3sssR3 insulators, the AT content was reduced to 57, 45, and 50%, respectively. However, the level of the enhancer block did not correlate with AT content [compare FBE1:3-R3sR3 (57% AT), 30 au to FBE1:3-R3sssR3 (50% AT), 162 au; Fig 4]. Therefore, changes in the composition of the spacer DNA may not significantly contribute to observed reduction in insulation. We favor a model whereby altered spacing between Su(Hw) binding sites may interfere with cooperative interactions between Su(Hw) molecules that are required for insulation. Although previous studies failed to identify association between Su(Hw) proteins in vitro (HARRISON et al. 1993 ; KIM et al. 1996 ), it is possible that these interactions occur in vivo. For example, a tightly spaced array of Su(Hw) binding sites may facilitate protein assembly on the DNA, perhaps by increasing protein accessibility to the gypsy insulator assembled into a nucleosome (ADAMS and WORKMAN 1993 , ADAMS and WORKMAN 1995 ). Alternatively, the requirement for closely spaced binding sites may reflect an overall structural arrangement of the Su(Hw) binding region, which is required to recruit other insulator proteins.

The determination that the organization of Su(Hw) binding sites is important for gypsy insulator function provides predictive information concerning the effects of Su(Hw) protein bound to non-gypsy sites. The Su(Hw) protein associates with 100–200 euchromatic sites throughout the genome (SPANA et al. 1988 ; GERASIMOVA and CORCES 1998 ) that are distinct from those present in the gypsy retrotransposon (R. ROSEMAN and P. K. GEYER, unpublished observations). This raises the question of the role of Su(Hw) protein at the non-gypsy Su(Hw) binding sites. It has been assumed these endogenous Su(Hw) sites correspond to positions of cellular insulators within the genome (GEYER 1997 ; GERASIMOVA and CORCES 1998 ). However, it remains possible that the Su(Hw) protein associated with non-gypsy sites has a distinct function (KIM et al. 1996 ). Our studies suggest that if the endogenous Su(Hw) binding sites have insulator activity, then the genomic Su(Hw) binding sites should be tightly clustered in arrays of binding sites (four or more), because we observed little or no enhancer blocking activity for synthetic binding regions that lacked this organization (Fig 4 and Fig 5). Alternatively, it may be that endogenous binding sites for the Su(Hw) protein occur as single sites, which are nearby recognition sites for other insulator proteins that together constitute a functional non-gypsy insulator. Recent studies support the proposal that insulator function may be imparted by the cooperation of multiple components (GEYER 1997 ; UDVARDY 1999 ). For example, analysis of the 1.7-kb scs region demonstrates that this insulator has two separable enhancer blocking components (VAZQUEZ and SCHEDL 1994 ). Deletion of either region from an otherwise intact scs insulator produces an intermediate enhancer blocking capability, whereas multimerization of either element reconstitutes blocking activity. These studies suggest that the scs insulator is assembled from independent, functionally redundant components that collaborate to confer enhancer blocking. Determination of whether the endogenous, non-gypsy Su(Hw) binding sites correspond to sites of functional insulators requires the identification and characterization of these sites.

A third significant finding of these studies was that the effectiveness of an insulator depends upon the strength of the upstream enhancer. This is best illustrated by the effects of the R3sR3 insulator (Fig 5). While R3sR3 produced a significant block of activated transcription by the single FBE1, it was incapable of impeding the robust Mega enhancer. These studies indicate that enhancer strength influences the degree to which an insulator interferes with enhancer-activated transcription. Interestingly, we found that placement of both the natural and the R6 insulator downstream of the Mega enhancer produced a fourfold higher level of yp2 promoter activity than observed when these insulators were downstream of FBE1. It is likely that this increased promoter activity reflects a decreased ability of these two insulators to block the transcriptional activating capacity of the Mega enhancer. Thus, this increased yp2 promoter activity may indicate that in addition to R3sR3, neither the R6 or natural gypsy insulator completely blocked the stronger Mega enhancer. A similar relationship between enhancer strength and insulator function was observed in the analysis of the effects of the Drosophila scs insulator on white gene expression (VAZQUEZ and SCHEDL 1994 ). However, our studies provide the first quantitative analysis of these effects.

Taken together, these experiments explain the inconsistencies observed for the effects of defective gypsy retrotransposons on gene expression (PEIFER and BENDER 1986 , PEIFER and BENDER 1988 ; FLAVELL et al. 1990 ; SMITH and CORCES 1992 ; HAGSTROM et al. 1996 ). Our data demonstrate that consideration of the number of binding sites positioned between an enhancer and promoter by itself provides poor predictive information concerning the strength of the insulation. Additional contributing factors include which binding sites are included and their relative organization. Finally, the nature of the regulatory interactions within the target gene is important. Our studies show that there is a link between transcriptional competency and degree of an enhancer block.

Models for insulator function:
Several models have been proposed to explain the mechanism by which insulators interfere with enhancer-activated transcription. The domain boundary model suggests that insulators are boundary elements that assemble specialized nucleoprotein complexes that demarcate distinct domains of chromatin structure. These domains are proposed to be assembled into higher-order structures, which prevent interactions between proteins bound in different domains (KELLUM and SCHEDL 1991 ; ROSEMAN et al. 1993 ; VAZQUEZ and SCHEDL 1994 ; CORCES 1995 ; GERASIMOVA et al. 1995 ; GERASIMOVA and CORCES 1998 ). This model directly links physical changes in chromatin structure with a block of enhancer activity and is supported by recent studies indicating that the Su(Hw) and Mod(mdg4) proteins are required for chromatin organization (GERASIMOVA and CORCES 1998 ).

A second model suggests that insulators function as transcriptional attenuators (CAI and LEVINE 1997 ) or decoys (GEYER 1997 ), which more directly interfere with transcriptional processes. This alternative model proposes that an insulator assembles a protein complex that intercepts the enhancer signal before it reaches the promoter. The decoy model is distinguished from the domain boundary model by the proposal that an insulator provides a functional, not structural, isolation of gene expression. Support for a functional model comes from the demonstration that changes in higher-order chromatin structure are not essential for insulator activity because insulators can block enhancer activity on a plasmid template prior to the assembly of mature chromatin in Xenopus oocytes (KREBBS and DUNAWAY 1998 ). Furthermore, sequences in close proximity to, but separable from, core promoter elements have enhancer blocking activity, linking transcriptional elements and insulators (OHTSUKI and LEVINE 1998 ). Our observations provide an additional connection between transcription and insulation, as we found that enhancer strength impacts insulator effectiveness. Insulator bypass by the Mega enhancer may result because the insulator cannot completely capture the enhancer-generated signal. The process by which an insulator captures an enhancer signal is unknown and depends upon the nature of the signal. For example, enhancers may assemble protein complexes that help recruit RNA polymerase to a promoter (reviewed in PTASHNE and GANN 1997 ). An insulator may disrupt these efforts, perhaps by directly interacting with the proteins required for RNA polymerase procurement. Alternatively, insulators may trap enhancer-bound proteins by creating an open chromatin conformation, which inappropriately directs enhancers to the insulator at the expense of interaction with general transcription machinery. Recent studies support a connection between states of increased histone acetylation and insulator function (PIKAART et al. 1998 ).

Further experimentation is required to distinguish between models of insulator function. Considering the large number of sequences that function as insulators, it is likely that there are different classes of elements that utilize distinct mechanisms to interfere with enhancer-activated transcription. A more complete characterization of the individual components involved in the function of various insulators and a better understanding of the physical organization of chromosomes will provide insights into the possible models.

	FOOTNOTES

¹ Present address: MRC Human Genetics Unit, Western General Hospital, Crewe Rd., Edinburgh EH4 2XU, Scotland, United Kingdom.

	ACKNOWLEDGMENTS

We thank Pieter Wensink for kindly providing plasmids and control fly stains that were critical for these experiments. We thank Lori Wallrath, Kate Huisinga, Emily Kuhn, Robin Roseman, and Tim Parnell for critically reading this manuscript. This work was supported by a National Institutes of Health grant GM42539 to P.K.G.

Manuscript received January 21, 1999; Accepted for publication June 21, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ABRAHAMSEN, N., A. MARTINEX, T. KJAER, L. SONDERGAARD, and M. BOWNES, 1993  Cis-regulatory sequences leading to female-specific expression of yolk protein genes 1 and 2 in the fat body of Drosophila melanogaster.. Mol. Gen. Genet. 237:41-48[Medline].

ADAMS, C. C. and J. L. WORKMAN, 1993  Nucleosome displacement in transcription. Cell 72:305-308[Medline].

ADAMS, C. C. and J. L. WORKMAN, 1995  Binding of disparate transcriptional activators to nucleosomal DNA is inherently cooperative. Mol. Cell. Biol. 15:1405-1421[Abstract].

ARONOW, B. J., C. A. EBERT, M. T. VALERIUS, S. S. POTTER, and D. A. WIGINTON et al., 1995  Dissecting a locus control region: facilitation of enhancer function by extended enhancer-flanking sequences. Mol. Cell. Biol. 15:1123-1135[Abstract].

AUSUBEL, F. M., R. BRENT, R. E. KINGSTON, D. D. MOORE, J. G. SEIDEMAN et al., 1994 Current Protocols in Molecular Biology. Green Publishing Associates and John Wiley and Sons, New York.

BLACKMAN, R. K., M. SANICOLA, L. A. RAFTERY, T. GILLEVET, and W. M. GELBART, 1991  An extensive 3' cis-regulatory region directs the imaginal disk expression of decapentaplegic, a member of the TGF-beta family in Drosophila. Development 111:657-666[Abstract].

CAI, H. and M. LEVINE, 1995  Modulation of enhancer-promoter interactions by insulators in the Drosophila embryo. Nature 376:533-536[Medline].

CAI, H. and M. LEVINE, 1997  The gypsy insulator can function as a promoter-specific silencer in the Drosophila embryo. EMBO J. 16:1732-1741[Medline].

CHUNG, J. H., M. WHITELEY, and G. FELSENFELD, 1993  A 5' element of the chicken ß-globin domain serves as an insulator in human erythroid cells and protects against position effects in Drosophila.. Cell 74:505-514[Medline].

COLEMAN, K. G., S. J. POOLE, M. P. WEIR, W. C. SOELLER, and T. KORNBERG, 1987  The invected gene of Drosophila: sequence analysis and expression studies reveal a close kinship to the engrailed gene. Genes Dev. 1:19-28[Abstract/Free Full Text].

CORCES, V. G., 1995  Chromatin insulators. Keeping enhancers under control. Nature 376:462-463[Medline].

DORSETT, D., 1993  Distance-independent inactivation of an enhancer by the suppressor of Hairy-wing DNA-binding protein of Drosophila.. Genetics 134:1135-1144[Abstract].

DORSETT, D., 1996  The suppressor of Hairy-wing protein and long distance enhancer-promoter interactions. Mol. Cell 6:381-387.

DUNAWAY, M., J. Y. HWANG, M. XIONG, and H. L. YUEN, 1997  The activity of the scs and scs' insulator elements is not dependent on chromosomal context. Mol. Cell. Biol. 17:182-189[Abstract].

EISSENBERG, J. C. and S. C. R. ELGIN, 1991  Boundary functions in the control of gene expression. Trends Genet. 7:335-340[Medline].

ELLIS, J., D. TALBOT, N. DILLON, and F. GROSVELD, 1993  Synthetic human ß-globin 5' HS2 contructs function as locus control regions only in multicopy transgene concatamers. EMBO J. 12:127-134[Medline].

FESTENSTEIN, R., M. TOLAINI, P. CORBELLA, C. MAMALAKI, and J. PARRINGTON et al., 1996  Locus control region function and heterochromatin-induced position effect variegation. Science 271:1123-1125[Abstract].

FLAVELL, A. J., L. S. ALPHEY, S. J. ROSS, and A. LEIGH-BROWN, 1990  Complete reversions of a gypsy retrotransposon-induced cut locus mutation in Drosophila melanogaster involving jockey transposon insertions and flanking gypsy sequence deletions. Mol. Gen. Genet. 220:181-185[Medline].

GALLONI, M., H. GYURKOVICS, P. SCHEDL, and F. KARCH, 1993  The bluetail transposon: evidence for independent cis-regulatory domains and domain boundaries in the bithorax complex. EMBO J. 12:1087-1097[Medline].

GARABEDIAN, M. J., M. C. HUNG, and P. C. WENSINK, 1985  Independent control elements that determine yolk protein gene expression in alternative Drosophila tissues. Proc. Natl. Acad. Sci. USA 82:1396-1400[Abstract/Free Full Text].

GARABEDIAN, M. J., B. M. SHEPHERD, and P. C. WENSINK, 1986  A tissue-specific transcription enhancer from the Drosophila yolk protein 1 gene. Cell 45:859-867[Medline].

GERASIMOVA, T. I. and V. G. CORCES, 1998  Polycomb and trithorax group proteins mediate the function of a chromatin insulator. Cell 92:511-521[Medline].

GERASIMOVA, T. I., D. A. GDULA, D. V. GERASIMOVA, O. SIMONOVA, and V. G. CORCES, 1995  A Drosophila protein that imparts directionality on a chromatin insulator is an enhancer of position-effect variegation. Cell 82:587-597[Medline].

GEYER, P. K., 1997  The role of insulator elements in defining domains of gene expression. Curr. Opin. Genet. Dev. 7:242-248[Medline].

GEYER, P. K. and V. G. CORCES, 1992  DNA position-specific repression of transcription by a Drosophila zinc finger protein. Genes Dev. 6:1865-1873[Abstract/Free Full Text].

GEYER, P. K., M. M. GREEN, and V. G. CORCES, 1988  Mutant gene phenotypes mediated by a Drosophila melanogaster retrotransposon require sequences homologous to mammalian enhancers. Proc. Natl. Acad. Sci. USA 85:8593-8597[Abstract/Free Full Text].

GEYER, P. K., M. M. GREEN, and V. G. CORCES, 1990  Tissue-specific transcriptional enhancers can act in trans on the gene located in the homologous chromosome: the molecular basis of transvection in Drosophila.. EMBO J. 9:2247-2256[Medline].

GROSVELD, F., G. BLOM VAN ASSENDELFT, D. R. GREAVES, and G. KOLLIAS, 1987  Position-independent, high-level expression of the human ß-globin gene in transgenic mice. Cell 51:975-985[Medline].

GYURKOVICS, H., J. GUASZ, J. KUMMER, and F. KARCH, 1990  A new homeotic mutation in the Drosophila bithorax complex removes a boundary separating two domains of regulation. EMBO J. 9:2579-2585[Medline].

HAGSTROM, K., M. MULLER, and P. SCHEDL, 1996  Fab-7 functions as a chromatin domain boundary to ensure proper segment specification by the Drosophila bithorax complex. Genes Dev. 10:3202-3215[Abstract/Free Full Text].

HARRISON, D. A., M. A. MORTIN, and V. G. CORCES, 1992  The RNA polymerase II 15-kilodalton subunit is essential for viability in Drosophila melanogaster.. Mol. Cell. Biol. 12:928-935[Abstract/Free Full Text].

HARRISON, D. A., D. A. GDULA, R. S. COYNE, and V. G. CORCES, 1993  A leucine zipper domain of the suppressor of Hairy-wing protein mediates its repressive effect on enhancer function. Genes Dev. 7:1966-1978[Abstract/Free Full Text].

HAZELRIGG, T., R. LEVIS, and G. M. RUBIN, 1984  Transformation of white locus DNA in Drosophila: dosage compensation, zeste interaction, and position effects. Cell 36:469-481[Medline].

HOLDRIDGE, C. and D. DORSETT, 1991  Repression of hsp70 heat shock gene transcription by the suppressor of Hairy-wing protein of Drosophila melanogaster.. Mol.