The Mcp Element From the Drosophila melanogaster Bithorax Complex Mediates Long-Distance Regulatory Interactions
Martin Muller, Kirsten Hagstrom, Henrik Gyurkovics, Vincenzo Pirrotta, Paul Schedl

Abstract

In the studies reported here, we have examined the properties of the Mcp element from the Drosophila melanogaster bithorax complex (BX-C). We have found that sequences from the Mcp region of BX-C have properties characteristic of Polycomb response elements (PREs), and that they silence adjacent reporters by a mechanism that requires trans-interactions between two copies of the transgene. However, Mcp trans-regulatory interactions have several novel features. In contrast to classical transvection, homolog pairing does not seem to be required. Thus, trans-regulatory interactions can be observed not only between Mcp transgenes inserted at the same site, but also between Mcp transgenes inserted at distant sites on the same chromosomal arm, or even on different arms. Trans-regulation can even be observed between transgenes inserted on different chromosomes. A small 800-bp Mcp sequence is sufficient to mediate these long-distance trans-regulatory interactions. This small fragment has little silencing activity on its own and must be combined with other Polycomb-Group-responsive elements to function as a “pairing-sensitive” silencer. Finally, this pairing element can also mediate long-distance interactions between enhancers and promoters, activating mini-white expression.

THE trans-regulatory interactions (often called transvection) that depend upon homolog pairing have been documented extensively in Drosophila. One of the first examples of transvection was described by Lewis (1954) for the Ultrabithorax (Ubx) gene of the bithorax complex (BX-C). He found that the mutant phenotype produced when mutations in Ubx and its regulatory region are combined in trans can be altered by rearrangements that suppress pairing (and therefore transvection) in the vicinity of BX-C. Since then, many other instances of regulatory interactions that require homolog pairing have been described (for review see Tartof and Henikoff 1991; Henikoff 1997). It is generally thought that trans-regulation involves the same type of interactions that normally control gene expression in cis; however, these interactions occur between regulatory elements on one homolog, such as enhancers, and their target sequences, the promoters, on the other homolog. In the case of the yellow (y) gene, for example, Geyer et al. (1990) showed that y cuticle enhancers on one chromosome can trans-activate the y promoter on the other chromosome when the two y genes are paired. Although pairing-dependent regulatory interactions like those observed in Ubx and y have usually been detected in special mutant backgrounds, there are reasons to believe that trans-regulation may play a role in the expression of completely wild-type Drosophila genes. Goldsborough and Kornberg (1996) showed that the wild-type Ubx gene requires pairing to achieve normal levels of transcription.

An important criterion for establishing that transvection is responsible for a given intergenic interaction is the demonstration that this interaction is affected by chromosomal rearrangements that disrupt pairing between the two copies of the gene. In the case of Ubx, rearrangements with breakpoints anywhere centromere proximal to the BX-C locus at 89E on the right arm of the third chromosome interfere with pairing in BX-C and suppress transvection (Lewis 1955, 1985). Most other examples of trans-regulatory interactions in flies are also sensitive to perturbations in homolog pairing induced by chromosomal rearrangements (e.g., Gelbart 1982; Pattatucci and Kaufman 1991). However, there have been several recent reports describing trans-regulatory interactions involving the Abdominal-B (Abd-B) gene of BX-C that are remarkably resistant to the disruptive effects of chromosomal rearrangements (Hendrickson and Sakonju 1995; Hoppmanet al. 1995; Siposet al. 1998). These interactions occur between the Abd-B promoter and the cis-regulatory domains (iab-5, iab-6, iab-7, and iab-8), which control its activity in the posterior parasegments of the fly. The cis-regulatory domains were found to weakly activate the Abd-B promoter in rearrangements in which the interacting partners are located on completely different chromosomes (Hendrickson and Sakonju 1995; Hoppmanet al. 1995). In addition, a much stronger tissue-specific regulatory interaction was detected in inversion chromosomes where the cis-regulatory domains of the Abd-B gene were separated from the Abd-B gene and its promoter by thousands of kilobases (Siposet al. 1998).

Trans-regulatory interactions do not always involve the activation of a promoter by an enhancer located on the other chromosome; they can also involve trans-interactions between negative elements or silencers. The classic example of a pairing-dependent negative regulatory interaction is that induced by the neomorphic zeste mutation z1 on the X-linked white (w) gene (Zacharet al. 1985; Smolik-Utlaut and Gelbart 1987; Pirrotta and Rastelli 1994). The mutant Z1 protein greatly reduces w expression in females, where there are normally two paired copies of the w gene. Wild-type levels of w expression can be restored in z1 mutant females by rearrangements that prevent pairing between the two w genes. Wild-type levels of w expression are also observed in z1 mutant males that have only a single copy of the X-linked gene. Although it is not fully understood how the Z1 protein silences paired w genes, silencing appears to depend upon the recruitment of Polycomb-Group (Pc-G) proteins (Wuet al. 1989; Pirrotta and Rastelli 1994).

The Pc-G genes encode a heterogeneous group of proteins that play a key role in the regulation of the homeotic genes in the Antennapedia and Bithorax complexes, as well as other genes, like engrailed (en), that control cell fate decisions (Paro 1993; Simon 1995; Pirrotta 1997). In the case of BX-C, the Pc-G proteins function to maintain parasegmentally restricted patterns of expression of the three homeotic genes: Ubx, abdominal-A (abd-A), and Abd-B. The homeotic genes in BX-C are regulated by a large cis-regulatory region (>300 kb) that is subdivided into nine domains (abx/bx, bxd/pbx, iab-2, iab-3, iab-4, iab-5, iab-6, iab-7, and iab-8). Each domain is responsible for directing the appropriate parasegmental transcription of one of the three homeotic genes (Duncan 1987). For example, Abd-A expression in PS8 is controlled by iab-3, while it is controlled by iab-4 in PS9 (Karch et al. 1985, 1990; Celnikeret al. 1990; Maciaset al. 1990; Bouletet al. 1991; Sanchez-Herrero 1991). In PS10, segmental identity is determined by Abd-B, which is under the control of the iab-5 cis-regulatory domain. The parasegmental patterns of expression are set early in development by the gap and pair-rule proteins, which interact with target sequences in the nine cis-regulatory domains (e.g., White and Lehmann 1986; Shimellet al. 1994; Casares and Sanchez-Herrero 1995). When the products of these segmentation genes disappear, regulation of BX-C switches to a maintenance mode that recognizes and propagates the initial pattern through the remainder of development. Maintenance requires not only the Pc-G, but also a second group of genes called the trithorax-Group (trx-G) genes (reviewed in Kennison 1993; Paro 1993; Simon 1995).

Experiments with homeotic reporter constructs have identified elements in several of the BX-C cis-regulatory domains that appear to be targets for Pc-G (and perhaps also trx-G) action. When these elements (called Polycomb response elements or PREs) are combined with parasegment-specific initiation elements, they are able to maintain segmentally restricted patterns of expression conferred on the reporter by the initiation elements (Simon et al. 1990, 1993; Müller and Bienz 1991; Busturia and Bienz 1993; Chanet al. 1994; Chianget al. 1995; Gindhart and Kaufman 1995; Hagstromet al. 1997). In addition to this maintenance function, PREs will also silence a heterologous gene when combined in the same construct. In the case of the mini-white reporter, PRE-mediated silencing has the unusual feature that it is pairing sensitive; silencing is observed in animals homozygous for a particular PRE mini-white insert, while little or much less silencing is evident in animals heterozygous for the same PRE mini-white insert (Kassiset al. 1991; Fauvarque and Dura 1993; Chanet al. 1994; Kassis 1994; Gindhart and Kaufman 1995; Hagstromet al. 1997). Like most other trans-regulatory interactions in the fly, mini-white silencing by the PRE is suppressed by “rearrangements” that disrupt pairing interactions between transgene inserts; thus, silencing is usually not observed when PRE mini-white transgenes inserted at different chromosomal sites are combined in trans.

In the studies described here, we have investigated the trans-regulatory activities of the Mcp element from BX-C. Mcp was initially identified because deletions of the element cause a dominant gain-of-function transformation of PS9 into PS10 (Lewis 1978; Karch et al. 1985, 1994). This transformation in parasegmental identity is due to the inappropriate activation of the Abd-B gene in PS9 (a parasegment in which Abd-B is normally off) by the iab-5 cis-regulatory domain, which specifies PS10 identity (Celnikeret al. 1990; Sanchez-Herrero 1991). Two models have been proposed to explain the gain-of-function phenotypes associated with Mcp deletions (reviewed in Schedl and Grosveld 1995). In the first, the Mcp deletions remove a PS10 silencer that functions to keep the iab-5 cis-regulatory domain off in PS9 (Busturia and Bienz 1993; Busturiaet al. 1997). When this silencer is removed, iab-5 is activated in PS9, turning on Abd-B. In the second, Mcp corresponds to a boundary element that functions to preserve the functional autonomy of the iab-4 and iab-5 cis-regulatory domains (Karch et al. 1985, 1994; Gyurkovicset al. 1990).

With the aim of distinguishing between these models, we tested fragments from the Mcp region of BX-C in a number of different transgene assays. While we have not resolved the question of whether Mcp corresponds to a silencer, a boundary, or both (as is the case for the element deleted by another BX-C gain-of-function mutation, Fab-71; Hagstrom et al. 1996, 1997; Mihaly et al. 1997, 1998a), we have uncovered a novel activity. As predicted by the silencer model, we found that fragments spanning the original Mcp deletion behave like many other PREs from BX-C and elsewhere in the Drosophila genome. When introduced into mini-white transgenes, they silence mini-white expression by a mechanism that requires two copies of the transgene. However, Mcp-mediated trans-regulatory interactions are unusual in several respects. Unlike other PREs or classical BX-C transvection, the Mcp-mediated trans-regulatory interactions do not seem to depend upon homolog pairing. In fact, there appear to be no requirements for homology or homolog pairing in the immediate vicinity of the Mcp transgene inserts. Thus, trans-regulatory interactions are observed not only between Mcp transgenes inserted at the same site, but also between Mcp transgenes inserted at distant sites on the same chromosomal arm or on different arms. Under special circumstances, trans-regulation can even be observed between transgenes inserted on different chromosomes. Deletion analysis indicates that a small ~800-bp Mcp sequence is sufficient to mediate long-distance trans-regulatory interactions. This small fragment has little or no silencing activity on its own and must be combined with other Pc-G-responsive elements to function as a “pairing-dependent” silencer. Finally, the long-distance trans-regulatory interactions mediated by the Mcp pairing element are not restricted to silencing. When this Mcp fragment is combined with a white enhancer, it can mediate the long-distance interaction of this enhancer with the mini-white promoter of another transgene carrying the Mcp fragment, activating mini-white expression.

MATERIALS AND METHODS

Plasmids and P-element constructs: Mcp element: A BX-C subclone, Φ3908, spanning the Mcp region of BX-C was obtained from Dr. S. Sakonju (Cumberledgeet al. 1990). The position of the 3908 subfragment relative to the DNase I-hypersensitive site mapped by Karch et al. (1994) is shown in Figure 1A. This Mcp element (which extends from 1 to 2865 bp in the sequence of Karchet al. 1994) is present in P-element constructs w#11, w#12, w#13, w#14, w#15, wy#2, ff#10, and ff#11. An 817-bp SalI/XbaI (1154–1971 bp) fragment spanning the major chromatin-specific DNase I-hypersensitive region is present in P-element constructs w#19, w#21, yw#1, ff#12, ff#13, ff#14, and ff#15. w#20 has a 464-bp AflII-XbaI fragment (1507–1971 bp) containing the DNase I-hypersensitive region. In w#18, a 754-bp PstI fragment (1093–1847 bp) spanning the DNase I-hypersensitive region was deleted from the 3908 subfragment. Note that this deletion is shifted slightly proximally compared to the 0.81-kb Mcp SalI/XbaI fragment. The Mcp fragment in w#16 lacks sequences proximal to the SalI restriction site. The Mcp fragment in w#17 lacks sequences distal to the XbaI restriction site. The scs element present in w#15 is the 1.7-kb BamHI/BglII fragment used by Vazquez and Schedl (1994).

The P-elements: w#1 and w#2 transgenes: These two mini-white reporter constructs are identical to the P elements used by Hagstrom et al. (1996) and were derived from a similar P element (pRW) described in Vazquez and Schedl (1994). ftz:ftz promoter/LacZ transposon: ff#1 has been described previously by Hagstrom et al. (1996). It contains unique XhoI and NotI sites between the Zebra and the Neurogenic enhancer elements originating from the fushi tarazu (ftz) cis-regulatory region (Hiromiet al. 1985). wy#1 and wy#2 transposon: The yellow gene contained in these two P elements corresponds to the D-1868 BglII fragment in Geyer and Corces (1987) except that the yellow gene is the intronless yellow gene (Mallinet al. 1998; DNA obtained from P. Geyer). It contains the body color enhancer element (bce) that directs yellow gene expression in the cuticle of the adult fly. yw#1 transposon: This construct contains a fragment from the yellow gene including the whole yellow intron.

P-element transformation: From each P-element construct, 0.5 mg/ml DNA was injected along with P-turbo helper plasmid (pUChspD2-3wc) into w1 or Df(1)w67c23,y, respectively. Transformants were detected by rescue of the white eye-color phenotype or the rescue of the yellow body-color phenotype, respectively. Crosses to marked balancer chromosomes were performed to generate balanced stocks and to determine the chromosome of insertion for each line.

Isolation of transvection-suppressing rearrangements R(ff# 10.5): Transvection suppressing rearrangements on the ff#10.5 chromosome were isolated essentially as described by Lewis (1985). ff#10.5 males were irradiated for 6 min at 5 mA/145 kV with a 1-mm aluminum filter. They were then crossed en masse with Cbx1Ubx1gl3/TM6B,DrMio virgins (stock obtained from Kathy Matthews, Bloomington Drosophila Stock Center). Among the ff#10.5/Cbx1Ubx1gl3 progeny, individuals were identified in which the weak Cbx phenotype characteristic of Cbx1Ubx1gl3/+++ flies was largely suppressed. Approximately 9000 mutagenized chromosomes were screened. The four strongest transvection-suppressing rearrangements [R(ff#10.5)] were analyzed cytologically.

Chromatin studies: DNase I digests of embryonic nuclei isolated from wild-type and transgenic lines were preformed and analyzed as in Karch et al. (1994).

Scoring eye colors: The pigmentation of the fly eye as a consequence of mini-white gene expression depends strongly on the age and the sex of the fly (Qian and Pirrotta 1995). In addition, we have found that balancer chromosomes can also have an effect on the eye color of flies containing {Mcp, mini-white} transgenes. Therefore, care was taken to only compare and score the eye color of flies of similar age and sex in the absence of balancer chromosomes. Flies were grown on standard cornmeal medium at 22 ± 1°.

Testing for pairing sensitivity: A particular Mcp, mini-white insert was considered to be “pairing sensitive” if the eye color of the homozygous insert was either lighter or indistinguishable from the eye color of the heterozygous insert. The latter rule was not applied when flies with darkly pigmented eyes were compared. Such lines were considered as non-pairing-sensitive lines. A few of the transgenic lines obtained displayed nonuniform or mottled pigmentation patterns; however, the mini-white expression in these lines changed appropriately with dose. Such lines were not scored as pairing sensitive, though we observed that the pattern could be modified in certain Pc-G mutant backgrounds.

Testing for nonhomologous pairing-sensitive repression: Flies of the genotype P1+/+P2 were compared to P1/+ and P2/+ control flies. Two transgene inserts, P1 and P2, were considered to interact strongly if the eye color of P1+/+P2 flies was lighter than both heterozygous controls. An interaction was scored as moderate if the P1+/+P2 flies had lighter eye color than the darker of the two controls. All test crosses performed for the third chromosome were done reciprocally, and the same results were obtained.

P1P2/++ recombinants were established from interacting pairs for which the P1+/+P2 combination had a clearly distinct eye-color phenotype compared to P1/+ and P2/+ parental flies. P1+/+P2 virgins were crossed to w1 males, and the progeny was screened for males with the expected eye phenotype. Balanced stocks were established for at least two individual recombinant males recovered for a particular P1P2/++ combination, and the same results were obtained for them in the P1P2/++ vs. P1+/+P2 test. Some of the recombinants obtained were verified by crossing P1P2/++ virgins with w1 males. In all cases, P1/+ and P2/+ eye colors reappeared in the F1 generation. The w#14.17,w#11.102 recombinant was also verified by in situ hybridization to salivary gland chromosomes.

The eye color of P1P2/P1+ flies was compared to that of P2/+ flies for combinations where the homozygous P1 insert on its own had essentially white eyes due to pairing-sensitive inactivation of the mini-white gene. The P2 insert in the P1P2/P1+ combination was scored as interacting with P1/P1 if the eye color of P1P2/P1+ flies was lighter than that of P2/+ controls. The same rules apply for inserts located on different chromosomes.

Testing for trans-activation: All possible intrachromosomal crosses between {w#21 and wy#1} and {yw#1 and wy#2} lines were established, and the eye color of the resulting trans-heterozygous flies was compared to the eye color of the parental controls (w#21/+ and yw#1/+, respectively).

Mutant strains: The following mutant chromosomes were used to study the effect of Pc-G alleles on Mcp-mediated repression and trans-activation: Pc106 has a point mutation in the chromodomain (Messmeret al. 1992). It was obtained from Renato Paro. This Pc106 stock has a rather strong Mcp phenotype that is separable from the mutation in the Pc gene by recombination (M. Müller, unpublished observation). Pc2 is an X-ray-induced, antimorphic allele; it was obtained from F. Karch. Df(3L)PcT7 is deficient for Pc and was obtained from S. Tiong. Sce1 is the single allele for this locus (Breen and Duncan 1986); it was obtained from I. Duncan. PclT1 was obtained from J. Kassis. ScmET50 and ScmR5-13B are hypomorphic Scm alleles of very similar character (Bornemannet al. 1998). ScmET50 was obtained from T. Wu, and ScmR5-13B was induced in our lab. Df(3R)GB104 is deficient for Scm and was obtained from K. Matthews at the Bloomington Drosophila Stock Center. AsxIIF51 (also known as Asx1) appears to be a gain-of-function allele (Chenget al. 1994); it was obtained from Y. Hiromi. AsxXF23 behaves genetically as a null (Simonet al. 1992); it was obtained from R. Paro. pho1 and phocv are alleles of the only DNA-binding Pc-G gene identified so far, pleiohomeotic (pho; Brownet al. 1998). These appear to be hypomorphic mutations. These stocks were obtained from J. Kassis. Psc1 is a strong hypomorph with respect to homeotic function; it was obtained from T. Wu. Su(z)21 and Su(z)25 are dominant suppressors of the zeste1-white interaction. Su(z)21 is an antimorphic allele of Su(z)2, whereas Su(z)25 deletes the Psc and the Su(z)2 genes (Wu and Howe 1995). Both stocks were obtained from T. Wu. zaw11E4: za is a loss-of-function allele of z, while w11E4 deletes the w gene. z1w11E4: z1 is of neomorphic nature (Wuet al. 1989). The z1 mutation has been shown to enhance Zeste protein aggregation. The degree of Zeste protein aggregation correlates well with white gene repression (Chen and Pirrotta 1993). The za and z1 stocks were obtained from J. Kassis. zop6w: zop6 was induced on the z1 chromosome, and it has been shown to contain the z1 and an additional point mutation in the zeste open reading frame (ORF; Pirrottaet al. 1987). While z1 requires two synapsed white genes to mediate the zeste effect, zop6 only requires a single copy of the white gene.

Immunohistochemical staining of embryos: The method used was that described in Hagstrom et al. (1997).

RESULTS

Mcp mediates the pairing-dependent inactivation of the mini white gene: Three Mcp alleles, Mcp1, McpH27, and McpB116, have been characterized in detail (Karchet al. 1994). All three are small overlapping deletions located nearly 60 kb downstream of the Abd-B transcription unit in the region between the iab-4 and iab-5 cis-regulatory domains (Figure 1A). In chromatin digests, this region of BX-C is punctuated by a large 0.4-kb nuclease-hypersensitive site (Karchet al. 1994). There are also several minor hypersensitive sites spaced at roughly nucleosome-length intervals just proximal to the major hypersensitive site. Although the three Mcp alleles delete slightly different DNA sequences, they all remove the 0.4-kb hypersensitive site, suggesting that it may contain elements critical for Mcp function.

To test the activity of Mcp isolated from its normal environment in BX-C, we introduced a 2.9-kb restriction fragment spanning the prominent hypersensitive site into the mini-white reporter construct w#2 (Figure 1B). w#2 has the white enhancer upstream of the mini-white gene, and, in the absence of negative chromosomal position effects, this construct is expected to give transgenic animals that have near wild-type eye color (Hagstrom et al. 1996, 1997). To protect against position effects arising from regulatory elements located downstream of the transgene, the construct has an scs′ fragment at the 3′ end of the mini-white transcription unit (Kellum and Schedl 1991). Since one of the proposed activities of Mcp is a boundary function, we inserted the 2.9-kb Mcp fragment at two different sites in the w#2 construct, either between the promoter and the enhancer (w#12, w#13) or upstream of the enhancer-(w#14). If the Mcp fragment has boundary activity, it should reduce mini-white expression when interposed between the enhancer and promoter; while in the upstream position, it should not interfere with enhancer-driven expression of the mini-white reporter.

The eye-color phenotypes of the transgenic lines for the control constructs fell in the expected range and changed appropriately with transgene dose: the w#2 transgenic lines typically had dark-orange to red eye pigmentation, while flies homozygous for the transgene had noticeably darker eye colors than their heterozygous sibs. When the Mcp fragment is located in the blocking position, as in w#12 or w#13, the eye pigmentation of flies carrying a single copy of the construct was reduced compared to the w#2 control (data not shown). The effects of the Mcp fragment on mini-white expression were most clear cut for w#13, where >90% of the lines were less pigmented than the w#2 lines. Conversely, when the Mcp fragment was located in the upstream positions, as in w#14, the eye color of the transgenic lines was the same or darker than w#2. By way of comparison, when the Fab-7 boundary element was placed in the blocking position in the mini-white transgene, a reduction in eye color was observed in ~50% of the lines (Hagstromet al. 1996).

Figure 1.

Sequence organization of Mcp transgenes. This figure shows diagrammatic representations of the different Mcp transgenes. (A) A restriction map of the Mcp region of the bithorax complex, including the approximate limits of the three Mcp deletions and the position of the nuclease-hypersensitive sites seen in chromatin digests. (B) Maps of the two control mini-white transgenes, one without the white enhancer and one with the white enhancer (WE). The stippled box on right side of both transgenes is a 0.5-kb scs′ fragment. Also shown are five mini-white transgenes that contain the 2.8-kb EcoRI Mcp fragment (see materials and methods) inserted at different sites. Indicated on the far right of each transgene are the number of lines that are pairing sensitive and the total number of lines. For example, 10 out of a total of 15 w#12 lines are pairing sensitive. In parentheses are the percentages of lines that show pairing-sensitive silencing (66% in this example). Note that the restriction maps of the transgenes are not drawn to scale. (C) The structure of two yellow transgenes (see materials and methods). These transgenes contain the yellow gene and bce. The Mcp fragments in the two transgenes are either the 2.8-kb EcoRI fragment or a smaller ~800-bp XbaI-SalI fragment that spans the major nuclease-hypersensitive region. In addition, the constructs contain the white enhancer (WE). Note that these transgenes do not contain scs′. (D) mini-white transgenes with different Mcp deletions. The construction of these deletions is described in materials and methods. All but w#21 contain the white enhancer. (E) The structure of the ftz lacZ transgene without and with Mcp. As described in the text and materials and methods, the LacZ is fused in frame to ftz protein coding sequences. Upstream of the ftz promoter are the Zebra (Z), Neurogenic (NE), and UPS enhancer elements. The transformation reporter is mini-white (MW). (F) The structure of several derivatives of the ftz LacZ transgene in which we have deleted sequences from the ftz upstream regulatory region or from Mcp. SX corresponds to the ~800-bp SalI-XbaI Mcp fragment. (G) The structure of a mini-white transgene that contains the ~800-bp SalI-XbaI Mcp fragment, the tce and ble enhancers from the yellow gene (see materials and methods), and the mini-white gene.

While these findings would seem to be consistent with an enhancer blocking activity, this interpretation is complicated by other observations. First, a significant fraction (from 5 to 25%, depending on the transgene) of the Mcp lines have variegating eye pigmentation as heterozygotes (see P/+ in Figure 2, A and B). Second, while mini-white expression for the control construct is typically proportional to the dose of the transgene, most of the w#12, w#13, and w#14 transgenic lines do not become darker when there are two copies of the transgene. Instead, the eye color of the homozygous flies is either the same (Figure 2C) or lighter (Figure 2, A and B) than their sibs that have only a single copy of the transgene. In some cases, mini-white expression in homozygous animals is virtually undetectable (Figure 2E), while in other cases, mini-white is expressed at high levels in only part of the eye (Figure 2D). This phenomenon is known as pairing-sensitive silencing (PS) and is a characteristic feature of DNA segments containing PREs.

As indicated in Figure 1A, there is little difference in the frequency of PS lines for constructs w#12 and w#13. Since PS activity does not seem to depend upon the orientation of the 2.9-kb Mcp fragment, this would argue against a directional silencing mechanism. Similarly, the position of the Mcp fragment relative to the enhancer and promoter does not seem to have a major effect on silencing activity. To determine whether the presence of the white enhancer might reduce the apparent silencing activity, we cloned the Mcp fragment into a mini-white construct, w#1, which lacks the enhancer, to give construct w#11. Pairing-sensitive w#11 lines were recovered at the same frequency as that observed for the transgenes carrying the white enhancer (see Figure 1A).

Mcp silences yellow but not fushi tarazu: We wondered whether Mcp would silence genes other than mini-white. To address this question, we introduced the 2.9-kb Mcp fragment into transgenes containing ftz and y. ftz is expressed in the early embryo (Hiromiet al. 1985), while y, like white, is expressed in the pupal stages (Geyeret al. 1990). The ftz transgene, ff#1, was generated by fusing the sequences encoding the entire N-terminal activation domain of ftz to Escherichia coli Lac-Z. To drive expression of the Ftz-β-galactosidase protein in embryos, the ftz promoter was joined to the three known upstream cis-regulatory elements, the Z, NE, and UPS enhancers (Hiromiet al. 1985). This ftz-Lac-Z fusion gene was then introduced into a mini-white vector, and the 2.9-kb Mcp fragment was inserted in either orientation between the Z and NE enhancers (ff#10 and ff#11 in Figure 1E). The y transgene, wy#2, has a y cDNA joined to the y promoter and the upstream y bce (Geyeret al. 1990). As indicated in Figure 1C, the wy#2 construct includes both the 2.9-kb Mcp fragment and the white enhancer (see below).

Figure 2.

Mcp silences mini-white and yellow, but not ftz. (A–E) Photographs of eyes from different mini-white transgene lines when the transgene is heterozygous (P/+) or homozygous (P/P). The particular transgene (e.g., w#14) and line number (e.g., 0.58) are indicated in each panel. The structure of the transgene (e.g., w#14) can be seen in Figure 1. (F) The male abdomen of the wy#2 transgenic line (see Figure 1) wy#2.63 when the transgene is heterozygous(P/+) or homozygous (P/P). (G and H) The pattern of Ftz-LacZ expression in two homozygous ff#10 lines (see construct in Figure 1): one that is pairing sensitive for mini-white expression (G) and one that is not (H). G is line ff#10.6, and H is ff#10.19.2. Note that there is no apparent difference between the two transgenes in Ftz-LacZ protein expression, either in the stripes or in the central nervous system. Also, no differences were evident when the lines were heterozygous. Ftz-LacZ expression was visualized with antibody against LacZ.

ftz transgene: Even though the 2.9-kb Mcp fragment is located farther away from the mini-white gene in the ftz-LacZ construct, the frequency (~45%) of mini-white silencing in the ftz construct is close to that observed in mini-white vectors lacking ftz DNA. While Mcp silences mini-white in this transgene, we were unable to detect any silencing of the ftz gene, either in early embryos, when it is expressed in a stripe pattern, or in older embryos, when it is expressed in the central nervous system. Shown in Figure 2, G and H, is the expression pattern of the Ftz-β-galactosidase fusion protein in embryos from two ff#10 lines: one that shows PS silencing of mini-white (Figure 2G) and one that does not (Figure 2H). The expression patterns are indistinguishable and similar to those seen in control ftz constructs.

y transgene: While Mcp is apparently unable to silence the ftz gene in embryos, it can silence y. Of the 10 wy#2 lines, 2 show evidence of silencing. As illustrated for one of the PS lines in Figure 2F, homozygous males display patchy pigmentation in abdominal segments A5 and A6 due to pairing-dependent variegation of the y gene. It is unclear why the frequency of pairing-sensitive lines for mini-white and y differ. It is possible that the y promoter is more resistant to Mcp-induced repression than the mini-white promoter. On the other hand, the difference may simply reflect the fact that it is easier to detect small changes in eye pigmentation than it is in body pigmentation.

A 0.81-kb Mcp fragment containing the major DNase I-hypersensitive site is necessary but not sufficient for silencing: To more precisely define the sequences required for pairing-sensitive silencing, we subdivided the 2.9-kb Mcp fragment into three different regions: a proximal 1.2-kb EcoRI-SalI fragment, a central ~800-bp SalI-XbaI fragment that contains the prominent nuclease-hypersensitive site, and a distal ~900-bp XbaI-EcoRI fragment. Different combinations of these subfragments were then tested for PS activity in the w enhancer mini-white vector. As summarized in Figure 1D, nearly full silencing activity could be reconstituted by combining the central ~800-bp SalI-XbaI fragment with either the proximal (w#17) or the distal (w#16) subfragment. The central subfragment seems to be a critical component since strong silencing activity could not be regenerated by combining just the proximal and distal fragments (w#18). While the central fragment is required for silencing, it is not in itself sufficient to induce silencing of the mini-white reporter either with (w#19) or without (w#21) the w enhancer.

These findings are consistent with the idea that Mcp silencing requires two distinct activities: one provided by the central subfragment that contains the 0.4-kb Mcp nuclease-hypersensitive site, and a second by subsidiary elements contained either in the proximal or distal subfragment. This possibility is supported by experiments in which we introduced the central subfragment into different versions of the ftz-LacZ transgene. When the central ~800-bp fragment is incorporated into the starting ftz-LacZ transgene (ff#13), efficient silencing of mini-white is observed, and about three-quarters of the lines are PS (Figure 1F). A high frequency of PS is also observed when this subfragment is included in ftz-LacZ transgenes that lack either the UPS (ff#14) or the UPS and NE upstream (ff#15) enhancers. The high frequency of PS lines obtained with these three ftz-LacZ transgenes indicates that some sequence in the ftz fusion gene can effectively substitute for subsidiary element(s) in the proximal or distal Mcp subfragments. This subsidiary element presumably lies within either the ftz promoter or the Zebra element. Like ftz, sequences from the y gene can also restore the silencing activity of the central ~800-bp fragment (see Figure 1G).

Mcp repression of mini-white depends upon Pc-G group genes: PREs located elsewhere in BX-C require proteins in the Polycomb group for their silencing activity (Pirrotta 1997). To confirm that Mcp silencing also depends upon Pc-G genes, we tested whether it is affected by a reduction in the dose of Pc-G genes. As can be seen in Table 1, mutations in Pc-G genes generally suppress Mcp silencing of mini-white, giving a darker eye color in homozygous transgene animals. In contrast, the mutations have little or no effect on the eye color of a transgene line, ff#10.19, which is not pairing sensitive. Each pairing-sensitive line responds somewhat differently to changes in the dose of Pc-G genes. Similar findings have been reported for other PREs (e.g., iab-7 PRE; Hagstromet al. 1997), and it is thought that this variability arises from local position effects. In addition to suppressing pairing-dependent silencing, we found that Pc-G mutations suppress the variegated eye-color phenotypes observed in all lines tested when there is only a single copy of the Mcp transgene (data not shown).

Figure 3.

Mutations in pleiohomeotic suppress Mcp-mediated pairing-sensitive silencing. This figure shows photographs of eyes from two homozygous w#11 lines, w#11.16 and w#11.102, in a wild-type background (pho+) and in a pho mutant background (either pho or pho/pho+). Note that mini-white expression increases in the pho mutant backgrounds.

Of the Pc-G proteins characterized to date, only one, the product of the pleiohomeotic (pho) gene, is known to bind directly to DNA. The pho gene encodes a zinc-finger protein and appears to be the fly homolog of the mammalian YY1 transcription factor (Brownet al. 1998). Mcp has a single consensus sequence for pho, which is located in the large nuclease-hypersensitive region (Karchet al. 1994; Mihalyet al. 1998b). As a result, it was of interest to determine if Mcp-mediated silencing depends on pho gene function. As can be seen in Figure 3 and in Table 1, a viable combination of two hypomorphic pho mutations suppresses Mcp-mediated silencing, significantly darkening eye color. In some lines, suppression is observed in flies heterozygous for a pho mutation (see Figure 3).

Chromatin structure of PS and non-PS transgenes: The experiments described in the previous section indicate that sequences in the central Mcp subfragment are critical for silencing activity. When located in the context of BX-C, this particular DNA segment contains a prominent nuclease-hypersensitive region. We wondered whether this nuclease-hypersensitive region is reformed when the Mcp fragment is included in transgenes, and whether there are any differences in the chromatin structure of PS and non-PS lines. To address these questions, we DNase I digested 12- to 24-hr embryonic nuclei prepared from two different ff#10 transgenic lines, one of which was pairing sensitive (ff#10.5) and the other not (ff#10.19.1). Indirect end labeling was then used to examine the DNase I cleavage pattern in the Mcp element of each transgene and as a control in the endogenous BX-C element (not shown).

View this table:
TABLE 1

Modifiers of Mcp-mediated pairing-dependent silencing

The DNase I cleavage pattern in the ff#10 transgene chromatin was examined by probing Asp718 restriction digests with a 32P-labeled fragment derived from the ftz UPS element. Asp718 generates a restriction fragment of ~10 kb from the endogenous ftz gene and a much larger (>20 kb) fragment from each of the ff#10 transgene inserts. Both of these full-length restriction fragments are observed in naked ff#10 genomic DNA (not shown) and in DNA prepared from transgenic nuclei that have been lightly digested with DNase I (see Figure 4, left, top two bands). In addition to the two restriction fragments, we observe a less strongly labeled band of ~6 kb (see arrow in Figure 4) in the DNase I chromatin digests of both the pairing-sensitive and the non-pairing-sensitive lines. The yield of this ~6-kb fragment appears to be slightly greater from the pairing-sensitive line.

Significantly, a ~6-kb DNase I cleavage product is not observed when DNase I digests of 12- to 24-hr wild-type nuclei (that lack the ff#10 transgene) are hybridized with the ftz UPS probe (Figure 4). Thus, it would appear that this subfragment is derived from DNase I cleavage in the chromatin of the ff#10 transgene and not in the chromatin of the endogenous ftz gene. To confirm this suggestion, we restricted DNA isolated from the ff#10.5 and ff#10.19.1 DNase I digests with Asp718 and BalI. In the endogenous ftz gene, BalI cleaves ~5 kb from the Asp718 site. In the ff#10 transgene, BalI cleaves ~7.5 kb from the Asp718 restriction site. If the ~6-kb DNase I cleavage product is derived from the endogenous gene, it should disappear after BalI restriction, while it should remain if it is derived from the ff#10 transgene. As can be seen in Figure 4, the 6.0-kb DNase I cleavage product is still observed after BalI digestion, indicating that it must be derived from cleavage in ff#10 transgene chromatin. On the basis of the size of the 6-kb fragment, the DNase I cleavage site in the transgene chromatin would map to the center of the Mcp fragment (see Figure 1) at approximately the same position as the DNase I cleavage site in the Mcp element of BX-C.

Chromosomal rearrangements that suppress transvection in BX-C do not disrupt Mcp-mediated silencing of mini-white: Gindhart and Kaufman (1995) showed that chromosomal rearrangements that interfered with homolog pairing in the vicinity of a transgene insert carrying a Sex combs reduced PRE abolished the pairing-dependent silencing of the linked mini-white reporter. We wondered whether Mcp-mediated silencing also depends upon homolog pairing. To address this question, we took advantage of the Cbx mutation in BX-C (Lewis 1955, 1985). The Cbx mutation inappropriately activates Ubx in PS5, transforming the wing toward haltere. While this transformation can be suppressed by introducing a Ubx mutation in cis to Cbx, suppression is incomplete because Cbx trans-activates the Ubx gene on the other homolog. This trans-activation is prevented by chromosomal rearrangements that disrupt BX-C pairing (Lewis 1955, 1985). Lewis showed that transvection-suppressing rearrangements must have at least one breakpoint in a “critical region” that extends from the BX-C locus at 89E to the base of the right arm of the third chromosome. In polytene chromosomes, these rearrangements perturb homolog pairing in BX-C and more distal sequences. Since breakpoints distal to BX-C do not suppress transvection, this has led to the idea that the homologs pair by zippering distally from the centromere.

We chose a pairing-sensitive Mcp transposon ff#10.5, which is inserted close to BX-C at 91A. As in wild type, the Ubx gene on the ff#10.5 homolog is activated by Cbx when trans to a Cbx Ubx chromosome, giving a wing with an abnormal shape and an alula that is greatly reduced in size. We selected for X-ray-induced rearrangements in the ff#10.5 chromosome that suppress these Cbx gain-of-function phenotypes when trans to the Cbx Ubx chromosome. The four strongest transvection-suppressing rearrangements [R(ff#10.5)] were analyzed cytologically. As indicated in Figure 5A, they have at least one breakpoint in the critical region, which Lewis showed disrupts pairing in BX-C.

We then combined each of the four rearranged chromosomes [R(ff#10.5)] with the original ff#10.5 chromosome and examined the eye-color phenotype. Since these four rearrangements suppress transvection at BX-C, we anticipated that they would also suppress the pairing-sensitive silencing of mini-white in the nearby Mcp ff#10.5 transgene. In all four cases, however, the eye color of the R(ff#10.5)/ff#10.5 animals was indistinguishable from that of the control ff#10.5 homozygotes. This can be seen for one of the R(ff#10.5)ff#10.5 combinations in Figure 5A. Moreover, precisely the same result was obtained when we combined the R(ff#10.5) chromosomes in all possible pairwise combinations; in every case, the pairing-sensitive silencing of mini-white was equivalent to that observed in animals homozygous for the starting ff#10.5 transgene (not shown). These findings indicate that rearrangements that suppress transvection in the Ubx locus are incapable of interfering with the pairing-sensitive silencing of mini-white by the Mcp element in the ff#10.5 transgene.

Silencing interactions between transgenes inserted at different chromosomal sites: The enhancer-promoter interactions involved in CbxUbx/++ transvection are quite sensitive to perturbations in homolog pairing and can be disrupted by DNA rearrangements located at a considerable distance from BX-C. Hence, it is conceivable that the breakpoints we recovered in our screen are too far from 91A to effectively perturb homolog pairing in the vicinity of the ff#10.5 transgene insertion. Since the closest proximal breakpoint in the four rearrangements is at 88C, this would argue that the critical region for silencing by the Mcp element in the ff#10.5 transgene insert might be considerably smaller than the critical region for breakpoints that can disrupt the CbxUbx/++ transvection. To test the most extreme version of this hypothesis, we asked whether the ff#10.5 transgene can participate in silencing interactions with Mcp transgenes inserted at other sites on the third chromosome. For this purpose, we crossed the ff#10.5 line to nine pairing-sensitive Mcp transgene lines located at sites on both arms of the third chromosome and scored the eye-color phenotype of the trans-heterozygotes for mini-white silencing.

Figure 4.

Chromatin structure of the Mcp element in PS and non-PS lines. Nuclei prepared from wild-type embryos and from embryos carrying the ff#10 transgene were digested with DNase I (Karchet al. 1994). Two different transgenic lines were used to prepare nuclei: ff#10.5 and ff#10.19.1. ff#10.5 is a pairing-sensitive line (PS), while ff10.19.1 is a non-pairing-sensitive line (NPS). After isolating the DNase I-digested DNA, the pattern of DNase I cleavage in the endogenous Mcp element (not shown) and in the two transgenes was analyzed using the indirect end-labeling technique. In the experiment shown in this figure, DNase I-digested chromatin samples were restricted with Asp718. After gel electrophoresis and blotting, the chromatin digests were probed with a 1.17-kb EcoRI fragment located in the UPS enhancer element 330 bp from the Asp718 restriction site (bar in the UPS region). This combination of restriction enzyme and probe permits the visualization of the DNase I cleavage pattern in the endogenous ftz gene in all three chromatin samples and in the ff#10 transgene in the chromatin samples prepared from the two ff#10 transgene lines. Note that only the full-length, ~10-kb ftz Asp718 restriction fragment is observed in the DNase I-digested chromatin sample from wild-type embryos, and there are few, if any, DNase I cleavage products (panel on far right). An Asp718 band of the same size is seen in the samples from the transgenic embryos. There is also a second larger band (~14 kb) that corresponds to the Asp718 fragment generated from restriction enzyme cleavage in the ff#10 transgene. In addition to these two restriction fragments, there is a more rapidly migrating band of ~6 kb (arrow) that is present in lower yield than either of the Asp718 restriction fragments. This 6-kb band is not observed in chromatin digests of wild-type embryos and, hence, is likely to be derived from DNase I cleavage in the chromatin of the ff#10 transgene. The 6-kb band is also not seen in Asp718 digests of naked DNA prepared from ff#10 transgenic embryos (not shown). To confirm that this fragment is derived from DNase I cleavage in transgene chromatin, we restricted the chromatin samples with Asp718 and BalI. BalI cleaves in the endogenous ftz gene at a site located ~5 kb from the Asp718 restriction site. If the ~6-kb DNase I cleavage product is derived from DNase I cleavage in the endogenous ftz gene, it should disappear in the Asp718 BalI double digest. BalI also cleaves in the ff#10 transgene; however, the BalI restriction site in the transgene is ~7.5 kb from the Asp718 restriction site. If DNase I cleaves in the transgene chromatin, then the ~6-kb fragment should still be observed in the Asp718 BalI double digest. As can be seen in the central panel, the second prediction is correct; the 6-kb DNase I fragment is still observed in the Asp718-BalI double digests. Hence, this DNase I fragment must be derived from cleavage in the chromatin of the ff#10 transgene. On the basis of the size of this fragment, the DNase I cleavage site must be located in the Mcp sequence of the transgene. Note that the yield of the DNase I cleavage product is slightly greater in the PS line than in the non-PS line.

Remarkably, we observed long-distance trans-silencing interactions with five of the nine heteropartners (see Table 2A and Figure 5B). Although the ff#10.5 transgene only interacts with partners inserted on the right arm of 3R (see Figure 6 and Table 2A), the five interacting transgenes are located at a considerable distance from ff#10.5. The two closest are more than two numbered divisions away (at 87D and 93D), while the other three are between five and eight numbered divisions away. These findings indicate that unlike CbxUbx/++ transvection, uninterrupted homolog pairing in the vicinity of the transgene is not required for the silencing activity of the Mcp element in the ff#10.5 transgene. Moreover, the critical region would appear to be no larger than the transgene itself. Another curious feature of this long-distance interaction is that the eye color of the trans-combination is usually rather homogenous (see Figure 5). Moreover, similar levels of pigmentation are seen in different individuals. Thus, this trans-interaction appears to occur with a rather high fidelity.

Figure 5.

Mcp mediates long-distance trans-regulatory interactions. (A) Listed in the first part of this panel are four third-chromosome rearrangements that were isolated on the basis of their ability to suppress Cbx/Ubx transvection in BX-C (see text and materials and methods for a full description of the screen). The cytology of each third chromosome rearrangement is indicated. Shown in the second part of A is a photo of eyes from different combinations of ff#10.5-derived flies. The yellow eye on the left is a fly heterozygous for one of the rearranged R(ff#10.5) chromosomes (4-12A). The white eye on the top right is a fly homozygous for the original ff#10.5 transgene. The third chromosome of this fly is not rearranged. The white eye on the bottom is the trans-combination between the rearranged R(ff#10.5) chromosome and the original ff#10.5. Note that the eye color is similar to that of the ff#10.5 homozygous fly. Precisely the same results were obtained when ff#10.5 was combined with the other rearranged chromosomes or when the rearranged chromosomes were combined with each other. (B) trans-combinations between ff#10.5 and other pairing-sensitive Mcp transgenes inserted on the third chromosome. In both cases, the trans-combination has a lighter eye color than a fly heterozygous for at least one of the parental transgenes. (C) trans-combinations between different pairing-sensitive Mcp transgenes inserted on the third chromosome. As in B, the trans-combination has a lighter eye color than a fly heterozygous for at least one of the parental transgenes.

We wondered whether the long-distance trans-silencing activity of the ff#10.5 transgene was unique to this particular insert or a general property of transgenes containing the Mcp element. To answer this question, we crossed the nine third chromosome pairing-sensitive Mcp transgene lines in all possible pairwise combinations and examined the eye-color phenotype. Trans-silencing interactions were scored as “moderate” when the eye color of the trans-heterozygous combination (P1 +/+ P2) was lighter than one of the heterozygous parents P1/+ or P2/+, while the interaction was scored as strong when the trans-heterozygous combination (P1 +/+ P2) was lighter than both parents. The w#14.29/w#11.102 combination shown in Figure 5C is an example of a moderate long-distance trans-silencing interaction, while the w#11.16/w#11.102 combination is an example of a strong trans-silencing interaction.

View this table:
TABLE 2

Interactions between Mcp transgenes on the third chromosome

As can be seen in Table 2B, long-distance trans-silencing interactions were observed in half (18 of 36) of the possible trans-heterozygote combinations. In general, long-distance silencing interactions were strongest between the nearest neighbors and weakened as the distance between the partners increased; however, some of the strongly interacting partners were located at considerable distances (8–10 numbered divisions; see Figure 6). While the ff#10.5 transgene did not show any trans-silencing interactions with inserts on the other chromosomal arm (3L), six of the nine transgenes in this group interacted with at least one heteropartner on the opposite arm. In all cases, these interactions fell into the moderate category. As was true for the silencing activity of the Mcp transgene in animals homozygous for a particular insert (homopartners), long-distance trans-silencing between heteropartners requires Pc-G genes (see Table 1).

Figure 6.

Trans-regulatory interactions across the third chromosome. This figure shows the site of insertion of 10 pairing-sensitive Mcp transgene lines: 4 on the left arm of the third chromosome and 6 on the right arm. The black dot in the center represents the centromere. Transgenes connected with heavy lines show strong trans-regulatory interactions. In this case, the trans-heterozygote has a lighter eye color than flies heterozygous for either of the parental transgenes. Transgenes connected by the lighter lines show moderate trans-regulatory interactions. In this case, the trans-heterozygote has a lighter eye color than flies heterozygous for one of the parental transgenes. Not shown are weak interactions. Weak interactions are those in which the eye color of the trans-heterozygote is not additive and is similar to the darkest heterozygous parental transgene.

Long-distance silencing interactions between Mcp transgenes are not restricted to inserts on the third chromosome. We generated all possible pairwise combinations between 14 pairing-sensitive Mcp lines on the second chromosome and 8 pairing-sensitive lines on the X chromosome and then scored the eye-color phenotype as described above. Long-distance trans-silencing was observed in 25% of the second-chromosome trans-heterozygous combinations and 39% of the X-chromosome trans-heterozygous combinations.

Long-distance silencing in cis: Since trans-silencing interactions are observed between Mcp transgenes located at sites many thousands of kilobases apart, it is clear that the two transgenes do not have to be imbedded in a large region of uninterrupted homolog pairing to establish functional silencing complexes. If homolog pairing is not required for silencing, it would seem possible that long-distance silencing interactions might occur not only in trans, but also in cis. To test this possibility, we generated recombinants between 11 Mcp transgene pairs on the third chromosome that exhibited long-distance trans-silencing interactions. In all cases, the eye color of the cis-combination was either the same or slightly lighter than the trans-combination. The cis-trans comparison for one pair, w#14.29 and w#13.101, is shown in Figure 7A. These findings indicate that the Mcp element can mediate long-distance silencing interactions not only in trans, but also in cis.

Rescuing the silencing-defective w#21 transgene by long-distance trans-interactions: Since only 1 of the 34 w#21 lines showed mini-white silencing as a homozygote, we concluded that the ~800-bp Mcp SalI-XbaI fragment in this transgene is not able to promote the assembly of a functional silencing complex on its own. However, we found that the silencing activity of the ~800-bp SalI-XbaI fragment could be “rescued” when it was combined, in the same transgene, with a “subsidiary” element. Complementing subsidiary elements included adjacent proximal or distal Mcp sequences, as well as heterologous sequences from either the ftz or y genes. We wondered whether the silencing activity of the defective ~800-bp SalI-XbaI fragment could be rescued by long-distance interactions with other pairing-sensitive Mcp transgenes. To answer this question, we tested pairwise combinations between non-pairing-sensitive w#21 insertions (10 on the second and 10 on the third chromosome) and pairing-sensitive Mcp transgene inserts on the same chromosome.

Only 1 out of the 40 pairwise combinations on the second chromosome showed evidence of complementation (Table 3); however, this interaction could be classified as strong because the eye color of the P1+/+P2 flies was lighter than flies heterozygous for either of the parental inserts. On the third chromosome, the frequency of complementation was much higher, and 8 out of 40 pairs were scored as interacting. Two examples of complementing combinations are shown in Figure 7B. These findings indicate that the silencing-defective ~800-bp Mcp fragment can be rescued by long-distance interactions with pairing-sensitive Mcp transgenes. It is interesting to note that one of the complementing partners, ff#13.101, has the same ~800-bp Mcp fragment as w#21; however, this 800-bp fragment is inserted into the ftz reporter. As described above, the ftz reporter has cis-acting elements that can substitute for the subsidiary elements present in larger Mcp fragments.

Long-distance silencing interactions are enhanced when the transgene is flanked by scs and scs: The long-distance regulatory activity of the Mcp element can be enhanced by placing it in a mini-white transgene, w#15 (see Figure 1), which is flanked by scs and scs′. We recovered 12 independent w#15 lines, 7 homozygous viable and 5 homozygous lethal. Five of the 7 homozygous viable lines were pairing sensitive, and the differences in eye color of the hetero- and homozygous animals can be quite dramatic in these lines (see Figure 8). Of the 5 homozygous lethal lines, 3 had a variegated eye-color phenotype that was alleviated by Pc-G mutations. To test for trans-silencing, we generated all possible pairwise combinations between the 5 pairing-sensitive lines and the 3 variegating lethal lines. Analysis of these pairwise combinations reveals that the long-distance silencing activity of Mcp in the w#15 transgene is more robust than in the other Mcp transgenes we tested.

As indicated in Table 4, long-distance silencing was observed in each combination of the w#15 inserts on the second chromosome and on the third chromosome. By comparison, the other Mcp transgenes interacted in trans in at most half of the possible pairwise combinations (see Table 2). Since there was only one w#15 insert on the X chromosome, we crossed this line, w#15.60, to the eight X-linked, pairing-sensitive lines tested previously (see above) for trans-silencing interactions. Long-distance trans-silencing interactions were observed with all eight X-linked partners. Again, by comparison, when these eight X-linked transgenes were combined with each other, trans-silencing was observed in less than half of the 28 possible combinations.

Long-distance silencing interactions were also observed between w#15 transgenes inserted on different chromosomes. For example, the X-linked transgene w#15.60 interacted with three of the four w#15 inserts on the second chromosome (Figure 8) and with one of the three w#15 inserts on the third chromosome (see Table 4). In fact, all the w#15 transgenes tested interacted with at least one partner on another chromosome. The most potent of these is w#15.48 (Table 4 and Figure 8), which interacts with all but one w#15 transgene on the second chromosome and with the X-linked insert. (Although a systematic study has not been done, we tested whether some of the other Mcp transgenes can establish interchromosomal silencing complexes. We found several combinations in which inserts on different chromosomes interacted.)

Figure 7.

Mcp silencing: cis/trans and complementation. (A) Comparison of the eye color when the interacting transgenes are in cis and in trans. The fly on the left of the photo is the trans-combination of w#14.29 and w#13.101, while the fly on the right is the cis-combination. The long-distance regulatory interactions between Mcp transgenes are either the same or slightly stronger when the transgenes are in the cis-configuration (see diagram on right) rather than in the trans-configuration. (B) The examples in B show that pairing-sensitive Mcp transgenes, such as w#14 or w#12, are able to complement the defective Mcp fragment in w#21.

Long-distance silencing with more than two partners: In the experiments described thus far, we examined interactions between two Mcp-containing transgenes, either inserted at precisely the same site (homo) or inserted at different sites (hetero). We wondered whether the establishment of functional silencing complexes between homopartners would be perturbed or enhanced by the presence of potential heteropartners. To address this question, we took advantage of the recombinant chromosomes that were generated to test for long-distance cis-silencing interactions. We used these recombinant chromosomes to examine silencing interactions between three or four possible partners.

The simplest cases are the homozygous recombinants P1P2/P1P2, in which both of the starting homopartners, P1/P1 and P2/P2, have white eyes. Six of the recombinants fell into this category. In all six cases, the silencing activity of the heteropartners (either as P1+/+P2 or as P1P2/++) was weaker than the homopartners (P1/P1 and P2/P2), and the eyes were pigmented (though less pigmented than at least one of the two “parental” transgenes, P1/+ or P2/+). If interactions between heteropartners interfere with homopartner interactions, weakening the silencing activity, the eyes of the P1P2/P1P2 homozygotes should become pigmented. In all six cases, however, the eye color of the P1P2/P1P2 combination is white. In the five remaining P1P2 recombinants, at least one of the homopartner interactions (P1-P1 or P2-P2) does not completely silence mini-white, and the flies have pigmented eyes. In these recombinants, it should be possible to detect both a weakening (more pigmentation than the homopartner) or a strengthening (less pigmentation than the homopartner) of the silencing interactions. In all five cases, the eye color of the P1P2/P1/P2 animals was that expected for the “sum” of the eye colors observed in the P1/P1 and P2/P2 animals. These results are consistent with the idea that homopartner interactions are the preferred configuration.

View this table:
TABLE 3

Complementation of the “defective” Mcp element in the w#21 transgenes by long-distance trans-regulating interactions

Figure 8.

Insulation enhances trans-regulation. The Mcp element, the white enhancer, and the mini-white reporter in the w#15 transgene are flanked by scs and scs′. The silencing activity of the Mcp element is increased by the insulators. Many of the w#15 lines express high levels of white when heterozygous and show dramatic reductions in mini-white expression when homozygous. This is illustrated for w#15.61 in the first photo. (P/+, heterozygous; P/P, homozygous). The insulators also appear to promote long-distance regulatory interactions, even between transgenes inserted on different chromosomes. The middle photograph shows trans-regulatory interactions between a w#15 transgene inserted on the X chromosome (w#15.60) and a w#15 transgene inserted on the second chromosome (w#15.102). The photograph on the right shows trans-regulatory interactions between a w#15 transgene inserted on the second chromosome (w#15.102) and a w#15 transgene inserted on the third chromosome (w#15.48).

If silencing complexes are preferentially established between homopartners, do these interactions inhibit the formation of silencing complexes with a heteropartner? To address this question, we examined the eye color of animals that have three interacting Mcp transgenes, P1P2/P1+. For these experiments, we specifically chose combinations in which the eye color of the homopartner P1/P1 was white. In this case, if the P2 transgene can interact with the P1/P1 homopartners, the eye color of the P1P2/P1+ animals should be lighter than that of the control P2/+ animals. As can be seen in Table 5, homopartner interactions can weaken heteropartner interactions; several transgenes (e.g., ff13.101/w#14.29 or ff#10.5/w#12.43) that establish silencing complexes in pairs do not interact when a homopartner is available. However, the presence of a homopartner does not exclude heteropartner interactions (e.g., w#12.43/w#11.102 or w#14.17/w#11.102 in Table 5 and Figure 9). In fact, as illustrated in the bottom panels of Figure 9, tripartite silencing complexes can be established even between w#15 transgenes located on different chromosomes (e.g., w#15.60 and w#15.102 on the X and second chromosomes or w#15.60 and w#15.48 on the X and third chromosomes, respectively). In general, when a tripartite silencing complex can be established between a single P2 heteropartner and a P1 homopartner pair, a silencing complex can also be established in the reciprocal combination, between a P2 homopartner pair and a P1 heteropartner (see Figure 9). In the examples shown in Figure 9, the eye color of the reciprocal tripartite combinations are quite similar; however, this is not always the case.

View this table:
TABLE 4

Silencing interactions between w#15 lines

View this table:
TABLE 5

Interactions between an unpaired insert, P2, and a homozygous insert, P1

The 0.81-kb Mcp fragment is sufficient to mediate long-distance “pairing” interactions: One supposition we have made in the preceding sections is that long-distance silencing is mediated by sequences in the Mcp element and not by sequences elsewhere in the various transgenes. However, all of the transgenes tested for long-distance silencing have in common not only the Mcp element, but also sequences from the mini-white gene and scs′. Hence, it is conceivable that these other sequences could either mediate or contribute to “homologous pairing” between transgenes inserted at distant sites. To ascertain if the Mcp element alone is sufficient to promote these long-distance regulatory interactions, we asked whether functional silencing complexes could be established between Mcp transgenes sharing as few sequences as possible. For this purpose, we took advantage of the second chromosome pairing-sensitive wy#2.63 transgene line, which contains yellow instead of mini-white as the transformation marker (plus the white enhancer). We crossed this yellow transgene line to 16 pairing-sensitive mini-white transgenes on the second chromosomes. Long-distance silencing interactions were scored if the wy#2.63 +/+ P{Mcp, mini-white} trans-heterozygotes had a lighter eye pigmentation than the P{Mcp, mini-white}/+ controls.

Mini-white silencing was observed in half (or 8/16) of the trans-heterozygotes. This frequency is equivalent if not greater than that observed in crosses between Mcp transgenes containing additional homologous sequences, such as mini-white or ftz (for the second chromosome the frequency was 25%). In three cases, long-distance silencing was observed when the wy#2.63 line was combined with an ff#15 transgene. As can be seen from Figure 1, the only sequence in common between wy#2 and ff#15 is the central 0.81-kb Mcp fragment, which contains the DNase I-hypersensitive site (and the P vector). This finding argues that the central 0.81-kb Mcp sequence is sufficient to mediate long-distance silencing interactions.

While mini-white silencing was observed in many of the wy#2/P{Mcp, mini-white} combinations, there was little obvious silencing of the y gene. The one exception is a combination between wy#2.63 and the transgene line ff#15.3, in which y expression is clearly variegated (not shown). This combination also gave the strongest reduction in mini-white expression among the eight wy#2/P{Mcp, mini-white} interacting pairs.

Long-distance interactions with other PREs: Silencing interactions between homo- and, in some cases, heteropartners have been described for, among others, the engrailed, escargot (esg), iab-7, and bxd PREs (Kassis 1994; Hagstromet al. 1997; Sigrist and Pirrotta 1997). Since the pairing-sensitive silencing activity of these other elements depends upon many of the same Pc-G proteins as Mcp, it seemed possible that these PREs might be able to cooperate with Mcp in the long-distance silencing of mini-white. To test this possibility, we crossed pairing-sensitive escargot (esg2 and esg1-8) and iab-7 PRE (iab-7 PRE.82B and iab-7 PRE.84E) lines to the collection of 10 Mcp lines tested in Table 2. Of the 20 escargot-Mcp pairwise combinations, only 1 between esg2 and ff#10.5 showed evidence of silencing. These two inserts are quite close, at 91B and 91A. Each of the iab-7 PRE lines interacted with one nearby Mcp transgene, but they did not interact with any of the other Mcp transgenes. Sigrist and Pirrotta (1997) found that the bxd PRE mediates long-distance silencing interactions when the bxd PRE mini-white transgene also contains the gypsy insulator (Sigrist and Pirrotta 1997). In an extensive set of pairwise crosses between these bxd PRE transgenes and our Mcp transgenes, no interactions were observed (C. Sigrist and V. Pirrotta, unpublished results). Thus, even though these other PREs are likely to share many Pc-G proteins with Mcp, these common factors are not sufficient to promote long-distance regulatory interactions with the Mcp element. This finding suggests that there may be factors that generate specificity in PRE-silencing interactions.

Figure 9.

Three is not a crowd. In the experiments shown in this figure, we examined long-distance regulatory interactions between three partners, a homopartner pair and a heteropartner. In the top panel, all three transgenes are on the same chromosome (the third). There are two copies of w#11.102 and only a single copy of w#14.17. Note that the eye pigmentation of the tripartite combination is reduced compared to w#14.17/+. In the middle and bottom panels, the interacting homo- and heteropartners are on different chromosomes. In the two middle panels, the interacting transgenes are w#15.60, which is on the X chromosome, and w#15.102, which is on the second chromosome. We tested both types of tripartite combinations [in which the heteropartner is either w#15.60 (photo on left) or w#15.102 (photo on right)]. In both cases, white expression in the tripartite combination is reduced compared to a fly heterozygous for the heteropartner. The same type of experiment is shown in the two panels on the bottom. In this case, one of the partners is on the X (w#15.60) and the other is on the third (w#15.48) chromosome. Again, both tripartite combinations show silencing.

Mcp can mediate long-distance enhancer-promoter interactions: A second supposition is that long-distance silencing involves some type of “pairing” between Mcp elements, which brings transgenes inserted at different sites (or on different chromosomes) in close proximity. If this is correct, then this pairing might bring an enhancer located on one Mcp transgene in close enough proximity to a promoter located on another Mcp transgene to trans-activate expression. To test for Mcp-mediated, long-distance trans-activation, we combined two different pairs of transgenes, w#21 and wy#1 and yw#1 and wy#2 (see Figure 1). In the first combination, the w enhancer in the wy#1 transgene could potentially trans-activate the mini-white reporter in the w#21 transgene. In the second combination, we could observe not only trans-activation of mini-white, but also trans-activation of yellow. To trans-activate, the enhancer and promoter must be brought together by the pairing of the Mcp elements, and, at the same time, must overcome the inhibitory effects of any Pc-G complexes assembled on the paired Mcp elements. We only tested transgene pairs that were inserted on the same chromosome; however, many inserts that showed no silencing activity when homozygous were included in this collection.

View this table:
TABLE 6

Modifiers of Mcp-mediated transactivation

We found that the white enhancer in one insert can trans-activate the mini-white reporter in another insert. However, the frequency of long-distance trans-activation is quite low, much less than the frequency of long-distance silencing that is observed when two different pairing-sensitive inserts are combined (>30%). In the first combination, w#21+/+wy#1, the white enhancer stimulated the mini-white reporter in only 1 out of >200 pairs. This pair consists of w#21.124 and wy#1.83. By themselves, neither of these transgenes shows variegation (as a heterozygote) or pairing-sensitive silencing as a homozygote. (The silencing activity of the defective 800-bp Mcp element in w#21.124 can, however, be complemented by pairing-sensitive Mcp transgenes. Additionally, the eye color of w#21.124/+ animals gets darker in a Psc1/+ and a su(z)21/+ background; Table 6.) The eye color of the w#21.124/wy#1.83 is uniform, clearly darker yellow than the light yellow eye color of the w#21.124/+ control (data not shown).

In the second combination, yw#1 and wy#2, trans-activation was observed in only 1 out of 66 pairs. In this case, the white enhancer in the wy#2.50 transgene trans-activates the mini-white gene in the yw#1.103 transgene. As can be seen in Figure 10, the eye color of the control yw#1.103 line is yellow to dark yellow and typically shows some evidence of variegation. This line is pairing sensitive, and in homozygous yw#1.103 flies, mini-white expression is repressed. When trans to the wy#2.50 transgene, mini-white expression is activated and the eye pigmentation is orange or light red, though still variegated (see Figure 10). No trans-activation of the y gene was observed.

As was found for long-distance silencing, long-distance activation can occur not only in trans, but also in cis. We recombined each pair of interacting transgenes, w#21.124 and wy#1.83 and yw#1.103 and wy#2.50, to generate the cis-configuration. In both cases, the long-distance activation in cis was equivalent to, if not slightly stronger than, that observed in trans. Though we have not determined the cytological position of the two pairs of inserts in polytene chromosomes, our recombination data indicate that the yw#1.103 and wy#2.50 pair are separated by >9 cM. The other pair also appears to be separated by a fairly large distance.

Figure 10.

Mcp can mediate long-distance activation. This experiment shows that Mcp can mediate long-distance promoter-enhancer interactions. The yw#1 transgene (see Figure 1) has the Mcp element, the mini-white gene, but no white enhancer. As a heterozygote, it expresses mini-white at a low level, giving a yellow eye-color phenotype. White expression can be activated by combining this transgene insert (either in trans, top panel, or in cis, bottom panel) with another transgene insert that contains Mcp, the white enhancer, and a yellow gene transformation marker. The activation can be seen in the darker eye color of the yw#1/wy#2 fly on the right. In the photo shown in the bottom panel, we examined the effect of z1 on activated mini-white expression. For this purpose, we used a cis combination of the yw#1 and wy#2 transgenes. As can be seen by comparing the two eyes in this photo, the z mutation reduces mini-white expression. This effect on mini-white expression depends upon the presence of the white enhancer in wy#2. Thus no effects on mini-white expression are observed in flies carrying only the yw#1 transgene, which lacks the white enhancer.

We wondered whether mutations in Pc-G group genes (which are known to suppress silencing) and in zeste (which are known to influence w expression) had any effect on trans-activation. As might be expected for a pairing-sensitive line, mutations in several Pc-G genes were found to suppress variegation and darken the eye color of yw#1.103 heterozygotes (see Table 6) and homozygotes (not shown). These mutations also darken the eye color of yw#1.103 wy#2.50/++ flies, indicating that Pc-G-mediated silencing may be counteracting the long-distance activation of the yw#1.103 mini-white by the wy#2.50 white enhancer. Some of the Pc-G mutations also have a slight but discernible effect on mini-white expression in heterozygous w#21.124 flies and when w#21.124 is combined with wy#1.83 (Table 6).

For zeste, we tested the effects of three different alleles, the neomorphic mutations z1 and zop6 and the loss-of-function mutation za. None of the z mutations have any effect on the mini-white transgene alone, and the eye color of w#21.124/+ or yw#1.103/+ flies carrying the z mutations is the same as in a wild-type z background. This is expected, since neither transgene carries the white enhancer and critical sites for Zeste protein action are located in the enhancer. However, as shown in Figure 10, the z1 mutation reduces the eye pigmentation of yw#1.103, wy#2.50/++ flies, indicating that it interferes with the long-distance activation of the yw#1.103 mini-white gene by the wy#2.50 white enhancer. Similar results were obtained for zop and za and for the other interacting transgene pair (see Table 6). These findings support the idea that the Mcp element mediates direct interactions between the white enhancer in one transgene and the promoter of the mini-white reporter in the other transgene.

DISCUSSION

Mcp functions as a silencer: In the studies described here, we have examined the biological activities of the BX-C Mcp element using transgene assays. The Mcp element in BX-C is defined by three overlapping deletions (Karchet al. 1994). Although these deletions differ in size and location, all three have indistinguishable, dominant gain-of-function phenotypes: they transform PS9 into PS10. This transformation in parasegmental identity is due to the ectopic activation of Abd-B in PS9, a parasegment in which Abd-B is normally off. The three deletions (see Figure 1) remove a common region of ~450 bp in length. This common region spans a major ~400-bp chromatin-specific, nuclease-hypersensitive site that is present throughout embryogenesis and in tissue culture cells (Karchet al. 1994). The smallest deletion, McpB116, is slightly larger than this common region, and it removes an additional ~350 bp proximal to the major nuclease-hypersensitive region. As reported previously (Busturia and Bienz 1993; Vazquezet al. 1993; Busturiaet al. 1997), we have found that DNA fragments extending to either side of the small Mcp deletion have silencing activity when linked to either a mini-white or y reporter. Like other silencers in the PRE class (Pirrotta 1997), silencing activity depends upon Pc-G proteins. Included in the Pc-G group is pleiohomeotic, a gene that encodes a DNA-binding protein that appears to be closely related to the mammalian YY1 transcription factor (Brownet al. 1998). As has been found for several other PREs in BX-C (Karchet al. 1994; Mihalyet al. 1998b), the major Mcp nuclease-hypersensitive region has a consensus Pho/YY1-binding sequence. The presence of this sequence, together with the fact that silencing activity depends upon the pho gene, would argue that this DNA-binding protein may play a key role in the assembly of Pc-G-silencing complexes by the Mcp element.

Busturia et al. (1997) have tested Mcp fragments in a pbx-UbxLacZ maintenance assay. In this assay system, the pbx initiation element activates UbxLacZ in PS6 and more posterior parasegments; however, in the absence of a PRE, the spatially restricted pattern of expression is not maintained, and the UbxLacZ reporter becomes activated in more anterior parasegments. Busturia et al. (1997) found that the ~800-bp SalI-XbaI Mcp fragment spanning the nuclease-hypersensitive region is able to maintain UbxLacZ expression in a spatially restricted pattern through the larval stages. They also showed that silencing requires the continued presence of this small Mcp element as late as the third instar larval stage.

We tested precisely the same 800-bp Mcp fragment in mini-white- and y-silencing assays. In our experiments, this Mcp fragment was unable to establish a functional silencing complex on its own. It could, however, promote the formation of functional silencing complexes when combined (in the same transgene) with additional Mcp fragments located either proximally or distally to the 800-bp fragment. These nearby Mcp sequences were not the only sequences that could complement the ~800-bp fragment; silencing activity could also be rescued by sequences from the ftz and y genes. Like the proximal and distal Mcp fragments, the complementing heterologous sequences have little, if any, silencing activity on their own and, hence, might serve a somewhat different function from that of the central ~800-bp Mcp fragment. Pirrotta and Rastelli (1994) have argued that some PREs may require secondary or “subsidiary” PREs to promote the efficient assembly or propagation of silencing complexes. By this model, sequences that complement the 800-bp Mcp fragment in our assays would have subsidiary PRE activity. The discrepancy between our findings and those of Busturia et al. (1997) could then be explained if the pbx UbxLacZ reporter used in their maintenance assay contains sequences that can function as subsidiary PREs. Consistent with this explanation, Christen and Bienz (1994) found that the pbx enhancer used by Busturia et al. (1997) to drive UbxLacZ expression has a weak silencing activity that is dependent upon genes in the Pc-G group.

In other studies, we have characterized another Abd-B PRE, the iab-7 PRE (Mihaly et al. 1997, 1998a; K. Hagstrom and S. Schweinsberg, unpublished data). Like Mcp, the iab-7 PRE is associated with a prominent nuclease-hypersensitive site that contains consensus pho-binding sequences. Moreover, silencing activity requires the pho gene product. However, the iab-7 PRE differs in several respects from Mcp. First, the small iab-7 PRE-hypersensitive region is by itself sufficient to establish functional silencing complexes in the mini-white assay, and no subsidiary elements are required (K. Hagstrom and S. Schweinsberg, unpublished data). Second, deletions that remove the iab-7 PRE in BX-C have no detectable phenotypic consequences as heterozygotes. When the PRE deletion is homozygous, only a few percent of the animals show the gain-of-function transformations expected from a failure to properly silence the iab-7 cis-regulatory domain in PS11. Moreover, even in these animals, iab-7 is ectopically activated in only a limited number of the PS11 cells. To generate dominant gain-of-function transformations of PS11 equivalent (in penetrance and expressivity) to those produced in PS9 by the Mcp deletions, both the Fab-7 boundary, which separates the iab-6 and iab-7 cis-regulatory domains, and the adjacent iab-7 PRE must be deleted. Deletions that remove the Fab-7 boundary, but not the iab-7 PRE, also exhibit the same dominant gain-of-function transformation; however, the phenotype is complicated because the intact iab-7 PRE occasionally silences the iab-6 cis-regulatory domain in PS11 (giving a loss-of-function phenotype; Mihalyet al. 1997).

These observations raise the question of whether the Mcp element is simply a PRE or has additional functions that are perhaps analogous to those of the Fab-7 boundary. In the case of Fab-7, it was possible to dissect the region using both transgene assays (Hagstrom et al. 1996, 1997; Zhouet al. 1996) and deletions generated in the context of BX-C (Mihalyet al. 1997) into two elements: one with PRE activity and one with boundary activity. This has not been true for Mcp. While there are some hints of boundary activity in our Mcp transgene assays, this activity, if it exists, cannot be separated from silencing. Of course, equally perplexing, sequences clearly critical for Mcp function in the context of BX-C do not in themselves have robust silencing activity in transgene assays. An additional property of the Mcp element (which is discussed further below) is its ability to mediate long-distance regulatory interactions. It is possible that it is this activity that is critical to the functioning of the Mcp element in its normal context in BX-C. In this view, the gain-of-function phenotypes of the Mcp deletions would arise, at least in part, from a failure in communication between the distant iab-5 cis-regulatory domain and the Abd-B promoter in PS9 cells (Vazquezet al. 1993).

Long-distance interactions: The most unusual feature of the Mcp element is its ability to promote long-distance interactions. Regulatory interactions are observed between Mcp transgenes inserted at different sites on the same chromosomal arm, on different chromosomal arms, and even between transgenes inserted on different chromosomes. The long-distance regulatory activity of the Mcp element is unusual. Among the previously characterized PREs from BX-C and other genetic loci, pairing-sensitive silencing is generally observed only between interacting partners inserted at the same site (Henikoff 1997; Pirrotta 1997). Only in a few instances have interactions been observed between partners inserted at different sites, and these usually involved PRE transgenes located in quite close proximity.

An exception to this general rule has been reported by Sigrist and Pirrotta (1997), who found that the gypsy insulator element substantially enhances long-distance silencing interactions of a bxd PRE. When their bxd-PRE:mini-white reporter was flanked by the gypsy insulator, long-distance silencing interactions were observed in about one-third of the pairwise combinations between transgenes inserted on the same chromosome. Moreover, the insulating activity of the gypsy element appeared to be important, as mutations in the gene encoding the Su(hw) protein suppress silencing interactions between hetero- but not homopartners. Although we did not test the effects of the gypsy insulator on the Mcp PRE, we did find that flanking the Mcp mini-white reporter with scs and scs′ enhanced both the frequency and strength of the long-distance interactions. Thus, silencing interactions were observed in all heteropartner combinations between scs/scs′ transgenes inserted on the same chromosome, while the frequency of interactions between unflanked transgenes was at most 50%. The presence of scs and scs′ also promoted silencing interactions, in some cases quite strong, between transgenes inserted on different chromosomes. In this context, it should be noted that a quite different result is obtained when the gypsy or scs elements are placed between the PRE and the mini-white reporter; in this case, the elements protect against silencing, and most transgene lines do not even show homopartner silencing interactions (Sigrist and Pirrotta 1997; Mallinet al. 1998; M. Muller, unpublished data).

Sigrist and Pirrotta (1997) also reported that a single gypsy insulator located upstream of the bxd PRE stimulated heteropartner interactions, though the effects were reduced compared to that observed when the bxd PRE mini-white reporter was flanked by the gypsy insulator. Since all the mini-white transgenes used in our experiments have a downstream scs′ element, it is conceivable that this element contributes to the long-distance regulatory activity of the Mcp element. While this possibility cannot be excluded, we believe that the presence of scs′ is unlikely to be the critical factor. First, we have used the same mini-white scs′ vectors to analyze the silencing activity of the iab-7 PRE (Hagstromet al. 1997); even though it silenced the mini-white reporter as a homopartner pair, this PRE was generally incapable of long-distance interactions in heteropartner pairs. Second, Mcp transgenes lacking scs′ can participate in heteropartner interactions. Moreover, long-distance interactions can be established between transgenes that share only the Mcp element (and sequences from the P-transposon).

How does Mcp promote regulatory interactions over long distances? One model is suggested by studies on the trans-inactivation of a wild-type brown (bw) gene by brownDominant (bwD), an insertional mutation that has a large block of heterochromatic satellite repeats (Henikoff 1997). The bwD allele appears to silence the paired wild-type gene by dragging it into a heterochromatin nuclear compartment (Henikoff and Dreesen 1989; Dreesenet al. 1991; Henikoffet al. 1995; Dernberg et al. 1996). One could imagine that the Mcp element silences mini-white by a similar mechanism, targeting the reporter to a repressive “Pc-G” compartment (cf. Paro 1993). Consistent with a compartmentalization mechanism, Pc-G proteins are concentrated in a limited number foci in interphase nuclei (Messmeret al. 1992). Of course, Mcp silencing would differ from bwD in that two (or more) copies of the Mcp element would be required for the efficient localization of mini-white to the “Pc-G” subnuclear compartment. By this model, Mcp-containing transgenes at distant sites would be able to interact because they happen to localize to the same Pc-G compartment.

However, there are a number of reasons to believe that the compartmentalization model does not apply to Mcp-mediated pairing. First, Mcp transgenes interact poorly with transgenes containing other PREs, including the gypsy bxd PRE transgenes described above (C. Sigrist and V. Pirrotta, unpublished data). Unless each PRE has its own special subnuclear compartment, these other PREs should be as effective as an Mcp partner in the targeting process. Second, the “defective” Mcp element (the 0.8-kb fragment) in the w#21 transgene can be rescued by long-distance interactions with a functional Mcp partner. In the compartmentalization model, the “defective” Mcp element would not be expected to enter the Pc-G compartment and, hence, should not be rescued by the functional partner. Third, contrary to the expectations of the compartmentalization model, we have found that the Mcp element can mediate not only long-distance silencing, but also long-distance activation.

These and other findings argue that the Mcp element facilitates long-distance regulatory interactions because it is able to locate and then pair with Mcp elements at other sites. All that appears to be required for this activity is the central 800-bp Mcp sequence; this sequence can mediate both long-distance silencing and activation. How does such a short DNA sequence interact specifically with an Mcp partner over large distances, and why does it differ from other DNAs? The first step in pairing must be finding the appropriate partner. Since pairing interactions should be disrupted by mitosis, the Mcp element would have to search for a partner at the beginning of each cell cycle. Unless Mcp targets the transgene to a special subnuclear environment (however, see above), it is difficult to imagine how it could specifically accelerate the search process over that of any other DNA sequence. If this is the case, then searching most probably occurs by a mechanism similar to that proposed for pairing between homologs—a constrained random walk (Funget al. 1998). Homolog pairing appears to proceed by random diffusional contacts between multiple independent sites rather than by a zippering mechanism that begins at defined points, such as the centromere (Funget al. 1998; Gemkowet al. 1998). However, this random walk is constrained (Marshallet al. 1997). First, as the first sequences pair, this reduces the volume of the search space for nearby unpaired sequences. Second, each chromosomal arm tends to occupy more or less discrete domains in interphase nuclei (Marshallet al. 1996). The interaction pattern seen for Mcp transgenes inserted on the third chromosome (see Figure 6) would be generally consistent with the expectations of a constrained random walk. The strongest interactions are between pairs of transgenes inserted on the same chromosomal arm, especially those located near the distal ends of each arm. Much weaker interactions are observed between transgenes on different arms or between proximal and distal transgenes on the same arm. If the search process is a random walk, it is probably not the limiting step in the trans-interaction. This is suggested by the fact that eye pigmentation in many of the trans-combinations is quite homogeneous and there is little variation between different individuals.

The second step would be the initial trans-interaction between the two Mcp elements, while the third step would be the formation and spread of a functional silencing complex around each element. It is at one of these steps that Mcp is most likely to differ from other PREs. We presume that proteins associated with the 800-bp Mcp sequence mediate the initial interactions and promote complex formation. The fact that Mcp can (when closely linked) establish silencing complexes with heterologous PREs (esg and iab-7 PRE) argues that proteins common to different PREs (e.g., the Pc-G proteins) contribute to the establishment of a functional silencing complex. However, since long-distance interactions are generally restricted to Mcp elements, there must be mechanisms for generating specificity. Such specificity mechanisms could be key to the novel long-distance regulatory activities of the Mcp element. One possibility is that specificity is conferred by a special three-dimensional configuration of a set of Pc-G proteins that are common to many different PREs. An alternative, and probably a more likely hypothesis, is that specificity is generated by a combination of proteins, some that are found in most PREs and some that are unique to the 800-bp Mcp element. That non-Pc-G proteins might also play a role in the pairing process is supported by the fact that Mcp can facilitate the activation of a promoter by a distant enhancer without apparently establishing a fully functional silencing complex.

Left unexplained by this model are the effects of insulating elements, such as gypsy or scs/scs′. When interposed between the PRE and the mini-white reporter, these elements block the formation of pairing-sensitive silencing complexes. On the other hand, when these elements flank the PRE and the mini-white reporter, they somehow enhance long-distance interactions, even promoting the formation of silencing complexes between transgenes inserted on different chromosomes. In the former case, these elements might protect the mini-white reporter by blocking the spread of a Pc-G “chromatin structure” along the chromatin fiber from the PRE to the mini-white reporter. The blocking mechanism would be entirely cis-dependent and have no effect on long-distance interactions that can occur in “trans.” In the latter case, the insulators might promote long-distance interactions by two different mechanisms. They could help nucleate or stabilize the initial pairing by providing additional protein-protein contacts. Alternatively, the insulators might function after the initial interaction, helping to ensure that pairing results in the formation of a functional silencing complex on each element. The insulators could promote this step by acting in cis to block the action of activators in neighboring sequences, or by ensuring the establishment of a homogenous chromatin environment.

Acknowledgments

The authors are indebted to many individuals for providing reagents and stocks. We thank Kathy Mathews, Shige Sakonju, Yash Hiromi, Renato Paro, Ting Wu, Judy Kassis, Julio Vazquez, Francois Karch, Joszi Mihaly, Pam Geyer, Stanley Tiong, and Ian Duncan for fly stocks and/or plasmids. Gretchen Calhoun is acknowledged for sequencing several crucial junctions of the P-element constructs. This work also benefited from discussions with the following people: Julio Vazquez, Shuba Govind, Yash Hiromi, Francois Karch, Joszi Mihaly, Rakesh Mishra, Daniel Pauli, Christian Sigrist, and Judy Kassis. The authors also thank the anonymous reviewers for carefully reading the manuscript and making many useful suggestions. The authors especially acknowledge the Human Frontier Science Program, whose support made this collaborative project possible. In the lab of P.S., M.M. was supported by a European Molecular Biology Organization Long Term Fellowship and by an Advanced Fellowship from the Swiss National Science Foundation. The research of M.M., K.H., and P.S. was supported by a grant from the National Institutes of Health. M.M. and V.P. acknowledge support from the Swiss National Science Foundation. H.G. was supported by the Hungarian National Science Foundation (OTKA T 021051).

Footnotes

  • Communicating editor: S. Henikoff

  • Received April 21, 1999.
  • Accepted July 22, 1999.

LITERATURE CITED

View Abstract