Abstract
How enhancers are able to activate promoters located several kilobases away is unknown. Activation by the wing margin enhancer in the cut gene, located 85 kb from the promoter, requires several genes that participate in the Notch receptor pathway in the wing margin, including scalloped, vestigial, mastermind, Chip, and the Nipped locus. Here we show that Nipped mutations disrupt one or more of four essential complementation groups: l(2)41Ae, l(2)41Af, Nipped-A, and Nipped-B. Heterozygous Nipped mutations modify Notch mutant phenotypes in the wing margin and other tissues, and magnify the effects that mutations in the cis regulatory region of cut have on cut expression. Nipped-A and l(2)41Af mutations further diminish activation by a wing margin enhancer partly impaired by a small deletion. In contrast, Nipped-B mutations do not diminish activation by the impaired enhancer, but increase the inhibitory effect of a gypsy transposon insertion between the enhancer and promoter. Nipped-B mutations also magnify the effect of a gypsy insertion in the Ultrabithorax gene. Gypsy binds the Suppressor of Hairy-wing insulator protein [Su(Hw)] that blocks enhancer-promoter communication. Increased insulation by Su(Hw) in Nipped-B mutants suggests that Nipped-B products structurally facilitate enhancer-promoter communication. Compatible with this idea, Nipped-B protein is homologous to a family of chromosomal adherins with broad roles in sister chromatid cohesion, chromosome condensation, and DNA repair.
INTERACTIONS between transcription activators and promoters can be accommodated by DNA looping when the activator and promoter are within several hundred base pairs of each other (reviewed in Ptashne 1986, 1988). Passive DNA looping is not always sufficient. For example, activation of the Klebsiella pneumoniae nifH promoter by the NifA protein requires binding of integration host factor (IHF) between NifA and the promoter (Santeroet al. 1992). The NifA binding site is ∼130 bp upstream of the transcription start site, and IHF bends the DNA to bring the activator into proximity of the promoter. Similarly, interactions between different proteins binding to the same eukaryotic enhancer are facilitated by formation or deformation of DNA bends by high mobility group (HMG) proteins such as LEF-1 (Giese et al. 1992, 1995) and HMG I(Y) (Falvoet al. 1995; Thanos and Maniatis 1995). Thus, IHF, LEF-1, and HMG I(Y) play architectural roles and help form structures that facilitate interactions between other proteins.
Many metazoan genes contain remote enhancers located several kilobases from the promoter. This implies that in addition to architectural factors such as HMG proteins that facilitate interactions over short distances, higher eukaryotes also have factors that act between enhancers and promoters to facilitate communication over many kilobases.
The Su(Hw) insulator protein encoded by the suppressor of Hairy-wing [su(Hw)] gene of Drosophila interferes with enhancer-promoter communication. Su(Hw) binds a DNA sequence in the gypsy transposon (Dorsett 1990, Spana and Corces 1990). When gypsy inserts into a gene, enhancer-promoter interactions are blocked in a Su(Hw)-dependent manner (Geyeret al. 1990; Holdridge and Dorsett 1991; Jacket al. 1991; Geyer and Corces 1992; Dorsett 1993; Cai and Levine 1995, Scott and Geyer 1995). Only the Su(Hw)-binding region of gypsy is required to block enhancers (Holdridge and Dorsett 1991; Geyer and Corces 1992). Enhancers located promoter-distal to Su(Hw) do not activate, while enhancers promoter-proximal to Su(Hw) function normally. Enhancers blocked by Su(Hw) can still activate a second promoter in the other direction (Cai and Levine 1995; Scott and Geyer 1995), indicating that Su(Hw) does not inactivate enhancers but interferes with their ability to communicate with the promoter.
Su(Hw) blocks virtually all enhancers. Where examined, the same region in Su(Hw) is required, despite a wide diversity in genes and enhancers (Harrisonet al. 1993; Kimet al. 1996b). This implies that Su(Hw) blocks enhancers in different genes by the same mechanism, and therefore that different genes use related mechanisms to facilitate enhancer-promoter communication.
Gypsy insertions in cut block a remote wing margin enhancer located 85 kb upstream of the promoter (Jacket al. 1991; Dorsett 1993). Failure of this enhancer to activate cut results in a cut wing phenotype in which most of the cells that form the adult wing margin are missing. The severity of this phenotype is sensitive to small differences in Su(Hw) insulator activity (Dorsett 1993). This sensitivity was exploited to find mutations that reduce activation by the wing margin enhancer in the presence of a gypsy insertion (Morcillo et al. 1996, 1997). In addition to enhancer-binding activators, these screens could identify architectural factors that act between enhancers and promoters to facilitate communication.
Previously these screens have identified mutations in two known genes, scalloped (sd) and mastermind (mam), and a novel gene named Chip (Morcillo et al. 1996, 1997). Genetic and biochemical evidence indicates that Sd and Mam are enhancer-binding factors (Morcilloet al. 1996). Chip, consistent with a potential role in enhancer-promoter communication, is a ubiquitous chromosomal protein that supports activation by several enhancers (Morcilloet al. 1997).
Here we characterize Nipped, an essential locus isolated in the same screens that identified the other cut regulators. We find that Nipped includes multiple essential complementation groups that play distinct roles both in regulating cut and in Notch receptor signaling. The Nipped-B complementation group is particularly antagonistic to the gypsy insulator in cut and Ultrabithorax. Strikingly, Nipped-B protein is homologous to a family of chromosomal adherins that participate diversely in DNA repair, chromosome compaction, and sister chromatid cohesion. We postulate that Nipped-B protein functions architecturally between enhancers and promoters to facilitate enhancer-promoter interactions.
MATERIALS AND METHODS
Drosophila culture: Flies were raised on cornmeal, yeast, and molasses medium (Wirtz and Semey 1982) at 25°. Crosses were performed in glass shell vials with 5-10 males and 10-15 females. Parents were transferred every 3-4 days, and progeny were scored for 10 days of eclosion.
Genetic screens: The screens for mutations that enhance the cut wing phenotype of ctL-32; su(Hw)e2 flies were previously described (Morcillo et al. 1996, 1997). After backcrossing to the parental ctL-32; su(Hw)e2 stocks for several generations, all homozygous lethal mutations on chromosome 2 were balanced over In(2LR)CyO, Cy, Df(2R)Kr4, KrB80, Dp(1;2)y+ in a y w mutant background.
Complementation tests and mapping of mutations: Lethal complementation tests were performed by crossing balanced mutants to each other and scoring for progeny lacking the balancer. Arthur Hilliker (University of Guelph) provided l(2)41Ae, l(2)41Af, and l(2)41Ah mutants and deficiencies in the 41A region. Complementation tests with known wing development mutations were performed by crossing the balanced lethal mutants to homozygous or balanced known mutants and scoring for progeny lacking balancers. All known mutants were previously described (Dorsett 1993; Morcillo et al. 1996, 1997) or were obtained from the Bloomington stock center at Indiana University. Nipped mutations were mapped by recombination using their lethal phenotypes and P-element markers as previously described (Morcilloet al. 1996).
Determination of Nipped mutant lethal phases: Nipped mutant lethal phases were determined by scoring larval mouthparts in the balanced stocks for the yellow (y) marker as previously described (Morcilloet al. 1996). Mutant mouthparts indicate that the larvae are homozygous or heteroallelic for the Nipped mutations, and wild-type mouthparts indicate presence of the balancer. Approximately 100 each of first, second, and third instar larvae were scored.
Quantification of the effects of Nipped mutations on cut wing and bithorax mutant phenotypes: y w ctL-32; su(Hw)e2 bx34e flies heterozygous for Nipped mutations were generated by crossing balanced Nipped mutant chromosomes into a y w ctL-32; su(Hw)e2 bx34e stock. The Nipped mutant chromosomes were marked with a P element containing a mini-white gene, allowing Nipped mutant progeny to be distinguished in the y w background (Morcilloet al. 1997). Effects of Nipped mutations on ct53d were tested by crossing balanced Nipped mutant males to ct53d females and scoring male progeny wing margins. Effects of Nipped mutations on heterozygous ct2s were quantitated by crossing balanced Nipped mutant males to ct2s females and scoring female progeny. Control progeny were generated by conducting the same crosses with y w, Oregon R, and Nipped parental stocks.
Cut wing margins were quantitated as previously described (Dorsett 1993). The scale ranged from 0 to 31 nicks, with 31 nicks per fly being given to any phenotype stronger than 30 nicks per fly. Bithorax mutant phenotypes were scored as described previously (Morcilloet al. 1996). The scores range from 0, which is wild-type phenotype, to 10, which is the phenotype displayed by homozygous bx34e flies wild type for su(Hw).
Genetic interaction experiments: Flies transheterozygous for a Nipped mutation and a mutation in another gene were generated by crossing flies with a balanced Nipped allele to flies with a balanced mutation in the other gene. Progeny lacking balancers were scored for margin nicks, wing size, wing vein, eye morphology, or bristle defects. Controls were generated by crossing balanced Nipped mutants to the Nipped mutant parental stock, a y w stock, or Oregon-R wild-type flies. Su(H) hypermorphs and Abruptex mutants were provided by Mark Fortini (University of Pennsylvania). Other mutant Notch alleles were provided by Michael Young (Rockefeller University).
For scanning electron microscopy, live flies were mounted on stubs using superglue, and the area surrounding the flies covered with conductive carbon paint. After air-drying overnight, the samples were dried in a vacuum dessicator, sputter coated with gold/platinum, and photographed in a scanning electron microscope at ×180 magnification for eyes and ×78 for bristles.
Reversion of the l(2)02047 P-element insertion: l(2)02047/ CyO; ry506 females (stock obtained from the Bloomington stock center, Indiana University) were crossed to CyO, HOP2/Bc Elp males (stock obtained from William Gelbart, Harvard University). F1 CyO, HOP2/l(2)02047 males were backcrossed to l(2)02047/CyO; ry506 females. Excision events were recovered as ry mutant progeny with Cy wings, and l(2)02047 revertant chromosomes were recovered from Cy+ progeny. Excision and revertant chromosomes were tested for the ability to complement Nipped-B mutations.
Rescue of the l(2)02047 P element from genomic DNA: Genomic DNA from homozygous l(2)02047 second instar larvae was digested with XbaI, religated, and used to transform Escherichia coli using the kanamycin resistance gene in the P{RZ} transposon (Mlodzik and Hiromi 1992).
Isolation of Nipped genomic DNA: A 2.5-kb XbaI-HindIII fragment of genomic DNA flanking the rescued l(2)02047 P element was used to probe Southern blots of an EcoRI digest of the DS08617 P1 phage (obtained from the University of Wisconsin collection) using the procedures previously described (Morcilloet al. 1997). Two neighboring EcoRI fragments of 8 and 5 kb in length hybridizing to the probe were subcloned into a pBluescript (SK+) plasmid vector.
RNA preparation and Northern blot hybridization: RNA isolation and Northern blot hybridization were performed as previously described (Dorsettet al. 1989). Single-stranded [32P]RNA probes were prepared from several restriction fragments spanning the length of the 13-kb region containing the l(2)02047 P-insertion site. Northern blots were stripped as previously described (Dorsettet al. 1989) and reprobed with rp49 antisense probes as a loading control.
Nipped-B cDNA cloning: A third instar imaginal disc cDNA library in λgt10 (provided by Jaeseob Kim, University of Wisconsin) was screened as previously described (Morcilloet al. 1997) using the 2.5-kb XbaI-EcoRI fragment located ∼4 kb from the l(2)02047 P-element insertion site as a probe. Six hybridizing plaques were plaque purified, DNA was prepared, and the EcoRI phage inserts were cloned into pBluescript (SK+) plasmid vector. Restriction maps revealed that five of the phage inserts are overlapping. The largest insert (clone 6-1) is 6.3 kb in length and contains the 3′ end of the open reading frame (ORF). It was sequenced in both directions by the DNA Sequencing Facility at Cornell University. An overlapping 3-kb insert (clone 3-1) containing the 5′ end of the ORF was sequenced using Sequenase (United States Biochemicals, Cleveland) according to the manufacturer’s recommendations. Database searches were performed using NCBI Blast programs (Altschulet al. 1997), and other sequence analysis was performed using MacVector software. The Nipped-B cDNA sequence has been deposited in GenBank under accession no. AF114160.
RESULTS
To identify genes that may encode architectural factors that support activation by the remote wing margin enhancer in cut, we screened for mutations that diminish cut expression. Most wing margin cells are lost when the wing margin enhancer is blocked by a gypsy insertion. Intermediate phenotypes with nicks in the wing margin occur when the wing margin enhancer is partially blocked by gypsy. The screens exploited the intermediate phenotype produced by the ctL-32 gypsy insertion (Figure 1) when partially suppressed by the leaky su(Hw)e2 mutation (Morcillo et al. 1996, 1997). These flies display ∼0.01 wing margin nicks per fly, indicating that cut gene activity in the wing margin is less than half wild type. We reasoned that partial loss of enhancer-binding proteins or architectural factors that facilitate communication with the promoter should decrease activation by the enhancer and increase the number of wing margin nicks. In the screens for dosage-sensitive modifiers, which are described elsewhere (Morcillo et al. 1996, 1997), mutagenized ctL-32; su(Hw)e2 males were mated to ctL-32; su(Hw)e2 females, and progeny with two or more wing margin nicks were tested for the presence of heritable mutations.
—Lesions affecting the remote wing margin enhancer in the cis regulatory region of the Drosophila cut locus. The 2.7-kb fragment containing the cut wing margin enhancer (hatched box labeled “wm”; Jacket al. 1991) is ∼85 kb upstream of the promoter (angled arrow) in the absence of a gypsy insertion (7.5 kb). The gypsy long terminal repeats (LTRs) are indicated by open boxes, and the Su(Hw) insulator protein-binding region by a filled circle. The extents of the ct2s and ct53d deletions are shown underneath by thick lines. ct2s displays an extreme recessive cut wing phenotype (Mogilaet al. 1992). ct53d deletes ∼0.5 kb and displays a weak recessive cut wing phenotype (Jack 1985). The ctL-32 gypsy insertion blocks the wing margin enhancer and displays a strong recessive cut wing phenotype in the presence of wild-type su(Hw).
Most mutations recovered in these screens are recessive lethal. Screening of ∼30,000 progeny (∼8,000 by EMS and ∼22,000 by γ-ray mutagenesis) identified the sd, mam, and Chip genes (Morcilloet al. 1996). An additional screen of ∼220,000 progeny by γ-ray mutagenesis was used to isolate additional Chip alleles (Morcilloet al. 1997). The larger screen also identified several additional mam alleles, three vestigial (vg) mutations, and ∼30 alleles of a complex locus on chromosome 2 that we named Nipped. To understand the functions of Nipped, we characterized its multiple lethal complementation groups and their genetic interactions with cut.
Nipped mutations affect four lethal complementation groups near the chromosome 2 centromere: The abilities of the Nipped mutant chromosomes to enhance the ctL-32; su(Hw)e2 cut wing phenotype are tightly linked with recessive lethal mutations. Dominant enhancement of the cut wing phenotype could not be separated from recessive lethality after multiple backcrosses to the parental ctL-32; su(Hw)e2 stock. As determined by segregation, several Nipped alleles are also translocations between chromosomes 2 and 3 (Table 1). As described below, where tested, the dominant effects of Nipped mutations on cut expression are mimicked by a deficiency, implying that the Nipped mutations are loss-of-function alleles.
Multiple Nipped mutations were mapped by recombination based on the recessive lethality to a position near the centromere on chromosome 2. Complementation tests with known deficiencies confirmed this location (Figure 2). All Nipped alleles are lethal over Df(2R)M41A10 and Df(2R)M41A8 and viable over Df(2R)A″. This places Nipped in the distal portion of 41A, near the heterochromatin-euchromatin boundary.
—Genetic map of the 41A region and the Nipped locus. The chromosome 2 centromere is indicated by an open circle. All Nipped mutations are lethal over Df(2R)M41A10 and Df(2R)M41A8, and viable over the Df(2R)A2 deficiency, placing them near the heterochromatin-euchromatin boundary. Nipped-A mutations complement all lethal mutations except l(2)41Ah (Hilliker 1976) and Nipped-C, D, and E mutations. Nipped-B mutations complement all except Nipped-C, D, and E mutations. Nipped-C alleles complement l(2)41Ae and l(2)41Af mutations (Hilliker 1976), and fail to complement Nipped-A [l(2)41Ah] and Nipped-B mutations. Nipped-D alleles complement l(2)41Ae mutations, and fail to complement l(2)41Af, Nipped-A, and Nipped-B mutations. Nipped-E mutations are lethal over l(2)41Ae, l(2)41Af, Nipped-A, and Nipped-B mutations.
Three lethal complementation groups, l(2)41Ae, l(2)41Af, and l(2)41Ah, were previously identified in the portion of 41A containing Nipped (Figure 2; Hilliker 1976). The Nipped mutations were tested for which lethal complementation groups they disrupt by crossing them to each other and to representative alleles of the 41A lethals. Most Nipped mutations are lethal over each other. However, two smaller groups, Nipped-A and Nipped-B, complement each other and produce viable progeny. All other Nipped alleles are mutant for both Nipped-A and Nipped-B. Nipped-B mutations complement all three previously known 41A lethal mutations, and Nipped-A mutations complement all except l(2)41Ah. We conclude that Nipped-B is a unique complementation group and that Nipped-A is identical to l(2)41Ah.
Characterized mutant alleles of the Nipped locus
The complementation tests divided several of the Nipped alleles mutant for both Nipped-A and Nipped-B into three classes: Nipped-C, Nipped-D, and Nipped-E. Nipped-C alleles fail to complement Nipped-A and Nipped-B mutations, but complement the l(2)41Af and l(2)41Ae mutations. Nipped-D mutations fail to complement Nipped-A, Nipped-B, and l(2)41Af mutations, but complement the l(2)41Ae mutation. Nipped-E mutations fail to complement all four lethal groups (Figure 2). Table 1 lists all the Nipped-A and Nipped-B mutations isolated in the screens and the Nipped alleles tested for complementation of all four lethal groups.
Only one of the characterized Nipped alleles displays ambiguous complementation behavior. Nipped25 fails to complement all Nipped-B mutations but is lethal over some, but not all, Nipped-A alleles. Nipped25 is lethal over T(2;3)Nipped-A394.2 and Nipped-A222.3, semilethal over Nipped-A34-12, and viable over l(2)41Af 45-72 and l(2)41Ae34-14. Because Nipped25 is lethal over more than one Nipped-A allele, we classify it as a Nipped-C allele. Because it may be only weakly mutant for Nipped-A, we avoided use of Nipped25.
Nipped mutant chromosomes do not contain mutations in known wing development genes: It was important to confirm that the Nipped mutant chromosomes do not contain other mutations that affect wing development. Therefore, we tested all Nipped alleles for complementation of mutations in the known wing development genes on chromosome 2, including apterous (ap), vg, wingless (wg), Suppressor of Hairless [Su(H)], mam, and Chip. Only one Nipped mutant chromosome has a second mutation in a known gene. Nipped-A226.1 is lethal over the Su(H)8 and Su(H)2 null alleles, and the Su(H)16 gain-of-function allele. Like gain-of-function Su(H) alleles and unlike Su(H) loss-of-function mutations (Fortini and Artavanis-Tsakonas 1994), Nipped-A226.1 suppresses the lethality caused by negative complementation between two Abruptex (Axe2 and Ax9B2) alleles of Notch (not shown). Other Nipped mutations do not suppress the Ax negative complementation (not shown), leading us to conclude that the Nipped-A226.1 chromosome also contains a Su(H) gain-of-function mutation. We therefore avoided use of Nipped-A226.1 in genetic experiments.
Nipped products are essential during embryonic and larval development: The complementation tests do not distinguish whether the Nipped locus consists of multiple genes or whether it is a single transcription unit that produces multiple products. However, they demonstrate that Nipped has multiple individual functions essential for viability. Nipped is required for viability prior to expression of cut in the wing margin, which begins late in third instar larval development.
Three of the five Nipped-A alleles, including the 2;3 translocation, are lethal at the second to third instar larval transition (Table 1). Nipped-A222.3/Nipped-E338 heterozygotes also die at this stage. Two Nipped-A alleles, Nipped-A357.2 and Nipped-A226.1 (which also contains a Su(H) mutation), are primarily embryonic lethal, but produce a few larval escapers. The Nipped-A34-12 allele is embryonic lethal, but this chromosome has been balanced for many years, and it may have acquired other lethals.
All four Nipped-B alleles, including the two translocations, are lethal at the second to third instar molt (Table 1). Two of the three Nipped-C alleles, including the translocation, are lethal at the same stage, as are Nipped-C160.1/ Nipped-E43 heterozygotes. Because Nipped-C alleles are mutant for both Nipped-A and Nipped-B, this confirms that the second to third instar molt is the primary lethal phase for both Nipped-A and Nipped-B. As shown below for Nipped-B, it is possible that maternally supplied product allows survival to this stage.
All Nipped-D and Nipped-E alleles are homozygous embryonic lethal as are the two Nipped-D/Nipped-E combinations tested (Table 1). Because both Nipped-D and Nipped-E alleles are mutant for l(2)41Af, this indicates that embryogenesis is the lethal phase for the l(2)41Af complementation group. We are uncertain of the l(2)41Ae lethal phase because the only Nipped alleles mutant for l(2)41Ae are the Nipped-E alleles, which are also mutant for l(2)41Af. Although the l(2)41Ae allele is embryonic lethal, this chromosome has been balanced for many years and may have acquired additional lethals.
Nipped mutations magnify the effects of a gypsy transposon insertion in cut: To ascertain the roles of the Nipped locus in regulating cut, heterozygous Nipped alleles were compared for their abilities to alter expression of different mutations in the cis regulatory region of cut (Figure 1; Table 2). The goal was to determine if any of the lethal groups in Nipped specifically magnify insulation by gypsy and Su(Hw). We quantitatively compared the abilities of several Nipped mutations to magnify the partially suppressed ctL-32; su(Hw)e2 cut wing phenotype. To ensure accuracy, we avoided the Nipped alleles that have a second mutation in the same chromosome (Nipped-A226.1), an ambiguous complementation pattern (Nipped-C25), or an atypical lethal phase (Nipped-A357.2 and Nipped-A34-12). Because the su(Hw)e2 mutation could be lost from chromosome 3 during the balancing crosses, we also could not unambiguously test the translocation alleles [T(2;3)Nipped-A394.2, T(2;3)Nipped-B4, T(2;3)Nipped-B359.1, and T(2;3)Nipped-C138.2] with the gypsy insertion. These constraints allowed us to compare two of the five Nipped-A alleles (Nipped-A323 and Nipped-A222.3) and two of the four Nipped-B alleles (Nipped-B292.1 and Nipped-B407). Of the Nipped mutations that affect multiple lethal groups, we were able to test one of the three Nipped-C alleles (Nipped-C160.1), both of the Nipped-D alleles (Nipped-D341.1 and Nipped-D263.3), and all three Nipped-E alleles (Nipped-E299.1, Nipped-E43, and Nipped-E338).
As expected, all Nipped alleles isolated in the screens dominantly increase the severity of the ctL-32; su(Hw)e2 cut wing phenotype (Table 2). However, the two Nipped-B mutations give ∼3- to 12-fold more wing margin nicks (1.2 and 4.8 nicks per fly) than the strongest Nipped-A mutation (0.4 nicks per fly). It is unlikely that the Nipped-A alleles are weaker mutations than the Nipped-B alleles because, as described below, these Nipped-A alleles have stronger effects than the Nipped-B alleles on other cut mutations. The l(2)41Ae34-14 and l(2)41Af 45-72 mutations have no detectable effects on the ctL-32; su(Hw)e2 phenotype, which explains why mutations disrupting only these lethal groups were not isolated in our screens. None of the Nipped-C, D, and E alleles, which disrupt multiple lethal groups, magnify the ctL-32; su(Hw)e2 cut wing phenotype more than the strongest Nipped-B allele. Indeed, only one, Nipped-E299.1 (1.9 nicks per fly), has a slightly larger effect than the weaker Nipped-B allele (1.2 nicks per fly). We deduce, therefore, that disruption of the Nipped-B lethal group causes most of the magnification of the gypsy insertion phenotype by the Nipped-C, D, and E alleles. The weaker effects of the Nipped-C160.1 and Nipped-E338 alleles relative to both Nipped-B mutations suggest that they may not be fully mutant for Nipped-B. Because the Nipped mutations were recently isolated in the same genetic background, do not contain mutations in the known wing development genes, and were crossed to the same ctL-32; su(Hw)e2 stock, the differences between the Nipped alleles are unlikely to be genetic background effects.
Interactions between cut mutations and heterozygous Nipped mutations
Nipped-B mutations amplify the effect of a gypsy insertion in Ultrabithorax: To examine the possibility that Nipped mutations may also magnify the effect that gypsy insertions have on other genes, we tested to see if heterozygous Nipped mutations increase the effect of a gypsy insertion in Ultrabithorax (Ubx). The bx34e gypsy insertion is in a transcribed region (Peifer and Bender 1986), but blocks activation by remote enhancers located in the abx/bx region, ∼50 kb downstream of the Ubx promoter (Simonet al. 1990; Qianet al. 1993). Like gypsy insertions in cut, bx34e is partially suppressed by the su(Hw)e2 mutation (Dorsett 1993), providing a sensitive intermediate phenotype made more severe by partial loss of Chip activity (Morcilloet al. 1996).
The effects of Nipped mutations on su(Hw)e2 bx34e were compared with their Nipped+ siblings. Neither of the Nipped-A mutations tested significantly alters the bithorax phenotype (Table 3; Figure 3). In contrast, both Nipped-B alleles dramatically increase the severity of the mutant phenotype three- to fourfold. Furthermore, the Nipped-C allele, both of the Nipped-D alleles, and two of the three Nipped-E alleles significantly amplify the mutant phenotype. Although the Nipped-E338 allele has little effect, this Nipped-E allele also has the weakest effect on the gypsy insertion in cut (Table 2). None of the Nipped-C, D, or E alleles is more effective than the Nipped-B alleles, indicating that Nipped-B is responsible for the increased severity of the bithorax phenotype.
We conclude that relative to the other Nipped lethal complementation groups, mutations in Nipped-B more strongly intensify the effects of the gypsy insertions in both cut and Ubx. However, we do not think that Nipped-B regulates expression of su(Hw) or gypsy. As described below, Nipped-B mutants display weak cut wing phenotypes in the absence of gypsy insertions.
Nipped mutations amplify the effects of a deletion in the cut wing margin enhancer: Although Nipped-B has greater effects on the gypsy insertions in cut and Ubx than other Nipped lethal groups, it was feasible that Nipped-B products might simply be more limiting for cut expression than other Nipped products. If so, then Nipped-B mutations should also have stronger effects on other types of cut mutants. To test this we quantitatively compared heterozygous Nipped alleles for their ability to magnify the severity of the partial cut wing phenotype of ct53d, a 0.5-kb deletion in the wing margin enhancer (Figure 1). Hemizygous ct53d males display ∼7 nicks per fly in Nipped+ backgrounds (Table 2; Figure 4). This partial phenotype presumably results from changes in the quantity or composition of activation complexes that form on the enhancer.
Interactions between the bx34e gypsy insertion in Ultrabithorax and heterozygous Nipped mutations
In contrast to their strong effects on the cut gypsy insertion, the Nipped-B292.1 and Nipped-B407 mutations do not magnify the ct53d mutant phenotype (Table 2). The two Nipped-B translocations also do not magnify the ct53d mutant phenotype. In contrast, the l(2)41Af 45-72 mutation and three of four Nipped-A mutations increase the severity of the ct53d cut wing phenotype (Table 2; Figure 4). The l(2)41Ae34-14 mutation has little effect on ct53d (Table 2; Figure 4). These results indicate that opposite to what is observed with the ctL-32 gypsy insertion, Nipped-A or l(2)41Af products are more limiting than Nipped-B products for cut expression in ct53d mutants.
—Dominant enhancement of the su(Hw)e2 bx34e bithorax phenotype by some Nipped mutations. Dorsal views of representative flies with the indicated genotypes are shown. Nipped-A222.3/+; su(Hw)e2 bx34e (left) displays a weak bithorax phenotype indistinguishable from Nipped+ controls (not shown). The arrow points to extra dorsal bristles. Nipped-D341.1/+ (middle) and Nipped-B407/+ (right) are strongly enhanced, with increases in thoracic cuticle and bristles between the thorax and abdomen (arrows). See Table 3 for quantitated phenotypes with these and other Nipped alleles.
Comparison of the strongest Nipped-C, D, and E alleles suggests that the effects of the individual Nipped lethal complementation groups on ct53d are additive and confirms that l(2)41Ae has little or no effect (Table 2; Figure 4). We postulate that the Nipped mutations are hypomorphic and that the alleles with the strongest effects are the most mutant. Confirming this idea, Df(2R)M41A8, which deletes all four lethal groups (Figure 2), has the strongest effect, increasing the number of nicks more than 4-fold (Table 2). T(2;3)Nipped-A394.2 is the strongest Nipped-A allele, giving a 1.8-fold increase in the number of wing margin nicks over the controls. T(2;3)Nipped-C138.2, which disrupts both Nipped-A and Nipped-B, increases the number of nicks ∼2.5-fold over the controls. Nipped-D263.3, which is mutant for l(2)41Af, Nipped-A, and Nipped-B, has a stronger effect, increasing the number of nicks ∼3.5-fold. Nipped-E338, which is mutant for all four lethal groups, has a similar effect as Nipped-D263.3, confirming that l(2)41Ae has little or no effect on ct53d. We conclude, therefore, that the effects of Nipped-A and l(2)41Af mutations on the ct53d enhancer deletion are additive.
It was possible that the effects of Nipped mutations on ct53d may depend on the particular sequences deleted from the wing margin enhancer. ct2s is a larger deletion that removes virtually all of the enhancer (Figure 1, Mogilaet al. 1992). Homozygous ct2s females display an extreme cut wing phenotype, while heterozygous females have a wild-type phenotype. We previously observed that loss-of-function mutations in sd, which encodes a protein that binds the enhancer (Morcilloet al. 1996), and null alleles of mam, which encodes a protein that binds chromosomes in the vicinity of cut (Bettleret al. 1996), both dominantly enhance heterozygous ct2s females to produce cut wing phenotypes (Morcilloet al. 1996). Presumably, these phenotypes occur because reducing the amount of enhancer-binding protein reduces activation of cut transcription by the solo wild-type enhancer present in ct2s heterozygotes.
—Dominant enhancement of the ct53d partial cut wing phenotype by some Nipped mutations. Representative wings from ct53d males with the indicated genotypes are shown. A control ct53d wing with a few margin nicks is shown in the upper left (+/+). Little or no effect is observed with Nipped-B407 (middle left), while the number of margin nicks is increased with Nipped-A222.3 (top right), l(2)41Af 45-72 (middle right), Nipped-D341.1 (bottom left), and Nipped-E338 (bottom right). See Table 2 for quantitated phenotypes with these and other Nipped alleles.
All four Nipped-B alleles exhibit weak but significant cut wing phenotypes in the presence of heterozygous ct2s (0.02-0.09 nicks per fly; Table 2). Importantly, this verifies that Nipped-B regulates cut expression in the absence of a gypsy insertion. However, two of the three heterozygous Nipped-A mutations tested display more severe cut wing phenotypes (0.25 and 0.94 nicks per fly, Table 2) in the presence of heterozygous ct2s, indicating that Nipped-A is more limiting for cut expression than Nipped-B. Nipped-A323 does not show a phenotype with ct2s, but this allele also has a weaker effect on ct53d, suggesting that it is a hypomorph. The Nipped-C160.1 mutation, which has no detectable effect on ct53d, also displays a weak cut wing phenotype with heterozygous ct2s (0.08 nicks per fly; Table 2), suggesting that it is weakly mutant for Nipped-A. Surprisingly, T(2;3)Nipped-C138.2, which has a strong effect on ct53d, displays a weak cut wing phenotype in combination with ct2s (0.06 nicks per fly; Table 2). This is the only Nipped allele affecting multiple lethal groups, however, in which the effect on ct53d does not correlate with the effect on ct2s. Thus, both Nipped-D and all three Nipped-E alleles magnify the effect of the ct53d lesion and display strong cut wing phenotypes with heterozygous ct2s (0.43-0.76 nicks per fly; Table 2). None of the Nipped-D and E alleles has stronger effects on heterozygous ct2s than the strongest Nipped-A alleles, suggesting that most of the effect is the result of disrupting Nipped-A. Confirming this idea, l(2)41Af 45-72 has only a weak effect on ct2s (Table 2). Combined, the results with ct53d and ct2s confirm that Nipped-B is more limiting for cut expression than the other Nipped products only when there is a gypsy insulator insertion between the enhancer and promoter.
Nipped mutations do not cause bithorax phenotypes with heterozygous deletions in Ubx: The observation that several Nipped mutations cause cut wing phenotypes in combination with heterozygous ct2s led us to consider the possibility that some Nipped mutations might also cause bithorax phenotypes in combination with heterozygous deletions in Ubx. However, none of the Nipped mutations tested, including Nipped-A222.3, Nipped-A323, Nipped-B292.1, Nipped-B407, Nipped-D341.1, and Nipped-E338, resulted in mutant phenotypes in combination with heterozygous bx34ePartRev and pbx2, both of which cause strong bithorax phenotypes (8 to 10) as homozygotes.
Nipped displays dosage-sensitive interactions with other cut regulators: To further define the roles of the different Nipped lethal groups in regulating cut, we compared dosage-sensitive interactions between Nipped mutations and other wing development mutations. These include mutations in ap, which defines the dorsal-ventral boundary at which the margin will form (Diaz-Benjumea and Cohen 1993; Blairet al. 1994), and Su(H), vg, and wg, which act at the margin prior to and during the time of cut expression (reviewed in Cohen 1996).
None of the Nipped mutations displays a mutant phenotype when combined with heterozygous loss-of-function alleles of Su(H) (Su(H)2 and Su(H)8 (not shown), or ap or wg (Table 4). In contrast, l(2)41Af45-72 and certain Nipped-D and Nipped-E alleles exhibit significant mutant phenotypes when transheterozygous with vg1, with up to ∼1 wing margin nick per fly (Table 4). Nipped-B alleles display little or no mutant phenotype with vg1, while some Nipped-A and Nipped-C alleles display a weak cut wing phenotype (∼0.1-0.2 nicks per fly). Therefore, l(2)41Af and Nipped-A display dosage-sensitive interactions with vg and Nipped-B does not.
Several Nipped mutations also display dosage-sensitive interactions with mam and Chip mutations. The strongest Nipped-A mutations result in 2.5-3 nicks per fly when transheterozygous with mamg2.1, and 1-2 nicks per fly when transheterozygous with Chipe5.5 (Table 4). In contrast, Nipped-B mutations display fewer nicks, from 0.2 to 0.5 nicks per fly with mamg2.1, and <0.2 nicks per fly with Chipe5.5. Nipped-C mutations, which are mutant for both Nipped-A and Nipped-B, display interactions with mam and Chip mutations similar to those displayed by Nipped-A mutations. Nipped-D and Nipped-E mutations have stronger interactions than the Nipped-A mutations, showing 3.5 to nearly 6 nicks per fly with heterozygous mamg2.1, and 2 to nearly 7 nicks per fly with heterozygous Chipe5.5 (Table 4). The stronger effects of Nipped-D and Nipped-E mutations are likely the result of disrupting both Nipped-A and l(2)41Af. The l(2)41Af 45-72 mutation displays ∼1 nick per fly with mamg2.1 and ∼0.4 nicks per fly with Chipe5.5.
Interactions between heterozygous Nipped and wing development mutations
The dosage-sensitive interactions between Nipped mutations and the vg, mam, and Chip mutations have noteworthy parallels in the interactions between Nipped mutations and the ct53d enhancer deletion. In both cases, l(2)41Af and Nipped-A show stronger interactions than Nipped-B mutations, and the effects of Nipped-A and l(2)41Af are additive. The cut wing phenotypes exhibited by flies heterozygous for both Nipped and mam mutations, or by Nipped and Chip transheterozygotes, are similar in strength to those displayed by flies transheterozygous for Chip and mam, Chip, and sd, or sd and mam mutations (Morcilloet al. 1996). These results indicate that l(2)41Af and Nipped-A cooperate closely with Chip, mam, and vg to regulate cut expression in the wing margin, and further confirm that Nipped-B plays a unique role.
Nipped mutations modify Notch mutant phenotypes: The genes that display dosage-sensitive interactions with Nipped function downstream of Notch in the wing margin, leading us to consider the possibility that Nipped may function widely in Notch receptor signaling. We therefore tested the ability of Nipped mutations to modify the phenotypes displayed by various Notch receptor mutants.
The hypomorphic nd1 mutation alters the intracellular domain of the Notch receptor (Xuet al. 1990) and displays a wing margin phenotype similar to that displayed by ct53d, with a few nicks per wing (Figure 5). Heterozygous Nipped-A222.3 (Figure 5) and Nipped-A323 (not shown) enhance this phenotype to give several more nicks. Nipped-B407 has no effect (Figure 5), while Nipped-B292.1 may slightly reduce the number of nicks (not shown). Strikingly, heterozygous Nipped-D341.1 and Nipped-E338 strongly enhance the nd1 phenotype to give strap-like wings (Figure 5). The strong effect of Nipped-D and Nipped-E mutations is likely the result of disrupting l(2)41Af because l(2)41Af 45-72 exhibits a similar phenotype, while l(2)41Ae34-14 has no effect (Figure 5). Because loss of cut expression does not cause loss of wing blade, we conclude that l(2)41Af has functions in wing development beyond regulation of cut.
The ability of heterozygous l(2)41Af mutations to give strap-like wings with nd1 is similar to the effects of heterozygous vg (Rabinow and Birchler 1990; Figure 5), Chipe5.5 (Figure 5), and Su(H) gain-of-function mutations (Fortini and Artavanis-Tsakonas 1994; Figure 5). Chip protein interacts with Apterous protein and is required for Apterous activity in the wing (Morcilloet al. 1997; Milanet al. 1998; Shoreshet al. 1998; P. Morcillo and D. Dorsett, unpublished results). Apterous activity is required for expression of Serrate protein (Diaz-Benjumea and Cohen 1995), which serves as the dorsal ligand for Notch at the wing margin (Cousoet al. 1995; Diaz-Benjumea and Cohen 1995; Kimet al. 1995; De Celiset al. 1996; Neumann and Cohen 1996). However, heterozygous ap56f, in contrast to Chipe5.5, has no detectable effect on the nd1 phenotype (Figure 5), indicating that Chip has roles in wing development beyond mediating Apterous activity and regulation of cut.
—Dominant effects of Nipped mutations on the nd1 Notch mutant phenotype. Representative wings from nd1 males with the indicated genotypes are shown. A control nd1 wing with margin nicks is shown on the left in the top row (+/+). Little or no effect is observed with l(2)41Ae34-14 (third row, left), ap56f (bottom left), and the Su(H)8 loss-of-function allele (bottom row, middle). Slight suppression of the mutant phenotype is observed with Nipped-B407 (top right). Enhancement of the mutant phenotype is observed with Nipped-A222.3 (top center), Nipped-D 341.1 (second row, left), Nipped-E338 (second row, center), l(2)41Af 45-72, Chipe5.5 (third row, center), vg1 (third row, right; Rabinow and Birchler 1990), and the Su(H)16 gain-of-function allele (bottom right; Fortini and Artavanis-Tsakonas 1994).
Nipped mutations also modify the vein-shortening phenotype of the AxE2 gain-of-function allele of Notch, which encodes a Notch receptor with a lesion in epidermal growth factor (EGF)-like repeat 29 in the extracellular portion (Hartleyet al. 1987; Kelleyet al. 1987). Heterozygous Nipped-A323 (not shown) and Nipped-A222.3 (Figure 6) weakly and moderately suppress the vein phenotype, while Nipped-B407 (Figure 6) and Nipped-B292.1 (not shown) have no detectable effect. Nipped-D341.1 (Figure 6), Nipped-E43 (not shown), and Nipped-E338 (Figure 6) strongly suppress the vein-shortening phenotype. The strong suppression is presumably the result of disrupting l(2)41Af because l(2)41Af 45-72 strongly suppresses the AxE2 mutant phenotype, while l(2)41Ae34-14 has no effect (Figure 6). The effects of Nipped-A and l(2)41Af mutations on AxE2 are similar to those of Su(H) gain-of-function mutations (Figure 6). vg (Rabinow and Birchler 1990) and Chip mutations have no visible effects on the AxE2 vein phenotype (Figure 6).
The phenotypes displayed by the split (spl) allele of Notch, which encodes an amino acid substitution in EGF-like repeat number 14 (Hartleyet al. 1987; Kelleyet al. 1987), are also affected by Nipped mutations. spl displays small rough eyes, occasional twinned bristles, and missing bristles (Figures 7 and 8). l(2)41Ae34-14 has no detectable effects on the spl eye phenotype (Figure 7). T(2;3)Nipped-A394.2 (not shown) and Nipped-A222.3 (Figure 7) also have no detectable effect on the rough eye phenotype, while Nipped-A323 slightly suppresses (not shown), making the eyes slightly larger and less rough. Nipped-B 407 (not shown), Nipped-B292.1, and l(2)41Af 45-72 also suppress the rough eye phenotype (Figure 7). Nipped-E338, Nipped-D263.3 (not shown), and Nipped-D341.1, which disrupt both Nipped-B and l(2)41Af, have similar effects to the Nipped-B alleles (Figure 7). The Nipped-A and Nipped-B alleles have little or no effect on the thoracic bristle phenotypes of spl (not shown and Figure 8), but Nipped-D 263.3 (not shown), Nipped-D341.1 (Figure 8), Nipped-E43 (not shown), and Nipped-E338 (Figure 8), all of which disrupt l(2)41Af, make the thoracic bristles sparser, thinner, and shorter. Because the l(2)41Af 45-72 chromosome contains a Pin mutation that affects bristle morphology, we could not evaluate the effect of this allele on bristles. We conclude that Nipped-B and l(2)41Af mutations suppress the spl eye phenotype, and infer that l(2)41Af mutations enhance the thoracic bristle phenotype of the spl allele of Notch.
—Dominant suppression of the AxE2 Notch mutant phenotype by some Nipped mutations. Representative wings from AxE2 males with the indicated mutations are shown. An AxE2 control wing with shortened veins is shown on the left in the top row (+/+). Little or no effect is observed with Nipped-B407 (top row, right), l(2)41Ae34-14 (third row, left), Chipe5.5 (bottom row, left), and vg1 (bottom row, center; Rabinow and Birchler 1990). Lengthening of the veins (suppression) is observed with Nipped-A 222.3 (top row, center), Nipped-D 341.1 (second row, left), Nipped-E338 (second row, center), l(2)41Af 45-72 (second row, right), and the Su(H)16 gain-of-function mutation (third row, right). Shortening of the veins (enhancement) is observed with the Su(H)2 loss-of-function allele (third row, center).
The dominant effects of Nipped mutations on the nd1, AxE2, and spl Notch mutant phenotypes indicate that l(2)41Af influences Notch receptor signaling, or Notch expression during development of the wing margin, wing veins, eye, and thoracic bristles. Nipped-A influences Notch phenotypes in both the wing veins and margin, while Nipped-B has influences primarily in the eye. The effects of l(2)41Af mutations in the wing margin, wing vein, and eye are similar to those gain-of-function mutations in Su(H). Similarly, the effects of Nipped-A mutations in the wing veins and wing margin, and the effect of Nipped-B mutations in the eye also mimic the effects of Su(H) gain-of-function mutations. Su(H) protein is a direct mediator of Notch signaling (Fortini and Artavanis-Tsakonas 1994) and a direct activator of vg (Kimet al. 1996a). We postulate that l(2)41Af mutations, Nipped-A mutations, and Su(H) gain-of-function mutant proteins downregulate Notch signaling in the wing margin. Su(H) gain-of-function mutant proteins may recruit transcription repressors, similar to the wild-type mammalian homologues of Su(H) (Kaoet al. 1998; Taniguchiet al. 1998).
Nipped-B protein is homologous to chromosomal adherins, which have cohesin-like activities: The in vivo observations indicate that the role of Nipped-B in regulation of cut and in Notch signaling differs significantly from those of the other Nipped lethal groups. Most importantly, Nipped-B is particularly antagonistic to the insulator activity of Su(Hw). We hypothesized, therefore, that Nipped-B products may participate in enhancer-promoter communication in cut and Ubx. To explore this idea we cloned and sequenced Nipped-B cDNAs.
Complementation tests revealed that the l(2)02047 P-element insertion (Berkeley Drosophila Genome Project) is allelic to Nipped-B. l(2)02047 flies are lethal over all Nipped-B alleles, and viable over l(2)41Ae34-14, l(2)41Af 45-72, and Nipped-A mutations. Nine of 13 induced excisions of the P element reverted the Nipped-B mutation in l(2)02047, confirming that the P insertion is the Nipped-B mutation.
—Dominant effects of some Nipped mutations on the spl Notch mutant eye phenotype. Representative eyes of spl mutant males with the indicated mutations are shown. All photographs are at the same magnification. In a wild-type background (+/+), spl mutant eyes are small and rough (top row, left). Little or no effect is observed with Nipped-A222.3 (top row, middle), l(2)41Ae34-14 (third row, left), Chipe5.5 (third row, center), vg1 (third row, right), and the Su(H)8 loss-of-function allele (bottom left). Larger, less rough eyes are observed with Nipped-B292.1 (top right), Nipped-D341.1 (second row, left), Nipped-E 338 (second row, center), l(2)41Af 45-72 (second row, right), and the Su(H)16 gain-of-function allele (bottom row, center).
We rescued the P insertion from genomic DNA and cloned a 13-kb genomic region surrounding the insertion site (Figure 9A). In this region, only a 2.5-kb EcoRI-XbaI fragment located ∼4 kb from the insertion site (Figure 9A) detects transcripts in Northern blots. This probe hybridizes to 7- and 4-kb transcripts oriented toward the P-insertion site (Figure 9, B and C). The 7-kb transcript is undetectable in homozygous l(2)02047 second instar larvae, while the relative levels of the 4-kb transcript appear to increase (Figure 9B). The 7-kb transcript is reduced in size in homozygous T(2;3)Nipped-B359.1 second instar larvae, while the 4-kb transcript is unaffected (Figure 9B). The 7-kb transcript is not affected by two Nipped-A mutants, but the levels are reduced to ∼50 and 30% wild-type levels in homozygous Nipped-C160.1 and Nipped-B407 mutants (Figure 9B). Alterations in the size or reductions in the levels of the 7-kb transcript in multiple Nipped-B mutants demonstrate that the 7-kb transcript is a Nipped-B mRNA. These results also confirm that Nipped-B407 is a hypomorph. Nipped-B mRNA is expressed at all stages of development, but the highest levels are present in newly laid embryos, indicating that it is maternally loaded (Figure 9C).
—Dominant effects of some Nipped mutations on the spl Notch mutant bristle phenotype. Dorsal views of representative thoraces from spl mutant males with the indicated mutations are shown. In a wild-type background (+/+, top left), spl displays missing and occasional twinned bristles. Little or no effect is observed with Nipped-A222.3 (top middle), Nipped-B 292.1 (top right), Chipe5.5 (middle row, right), vg1 (bottom left), the Su(H)8 loss-of-function allele (bottom center), and the Su(H)16 gain-of-function (bottom right). Loss of scutellar bristles and shorter, thinner bristles are observed with Nipped-D341.1 (middle row, left) and Nipped-E338 (middle row, center).
The probe detecting the Nipped-B mRNA was used to isolate overlapping cDNA clones from an imaginal disc library. The probe sequence hybridizes near the middle of the cDNAs (Figure 9A). One clone (6-1) contains a long ORF starting at the 5′ end, while an overlapping clone (3-1) contains the putative initiation codon. Combined, the clones give a complete ORF of 6159 nucleotides, encoding a protein of 2053 amino acids. Because the 2.5-kb EcoRI-XbaI genomic probe fragment is the only fragment in the cloned genomic region that hybridizes to the cDNA, and because it only hybridizes to a fragment near the middle of the cDNA, we can deduce that the 5′ end of the cDNA comes from an uncloned genomic region on one side of the P-insertion site in l(2)02047 and that the 3′ end of the cDNA comes from an uncloned region on the other side of the P-insertion site. Thus, we conclude that the P-insertion site is in an intron.
—The Nipped-B02047 P-element insertion [l(2)02047] and Nipped-B transcript. (A) Restriction maps of the l(2)02047 P-insertion site and Nipped-B cDNA. As described in the text, excisions of the P{RZ} element revert the Nipped-B mutation. At the top is a restriction map of the 13-kb genomic region surrounding the P-insertion site (E, EcoRI; X, XbaI; Sp, SpeI; S, SalI; B, BamHI; Sc, SacI; P, PstI). The 2.5-kb EcoRI-XbaI genomic fragment, marked with an arrow indicating the direction of transcription, is the only fragment in the 13-kb region that detects transcripts (7 and 4 kb) in Northern blots. None of the other fragments in the 13-kb genomic region hybridizes to transcripts. The EcoRI-XbaI genomic fragment was therefore used as a probe to isolate cDNAs. As indicated by the dashed line, the SalI-EcoRI fragment in the cDNA, marked with an arrow showing the direction of transcription, is the only fragment in the cDNA that hybridizes to the EcoRI-XbaI genomic fragment. Because the genomic DNA has not been sequenced, the extent of sequence overlap between the EcoRI-XbaI genomic fragment and the SalI-EcoRI cDNA fragment is unknown, but the EcoRI site at the upstream end of the EcoRI-XbaI genomic fragment is not the same EcoRI site present at the downstream end of the SalI-EcoRI cDNA fragment. Indeed, none of the restriction sites shown for the cDNA are present in the 13-kb cloned genomic DNA. Because the EcoRI-XbaI genomic fragment is the only genomic fragment that hybridizes to the cDNA, and because it only hybridizes to the SalI-EcoRI fragment in the middle of the cDNA, it is clear that the 5′ end of the cDNA comes from an uncloned genomic region to the left of the P-insertion site, and that the 3′ end of the cDNA comes from an uncloned genomic position to the right of the P-insertion site. Therefore, we conclude that the P-insertion site is in an intron. The filled region of the cDNA indicates the ORF. (B) Northern blots of total RNA isolated from second (L2) and third (L3) instar larvae. The blots were hybridized with antisense [32P]RNA probe prepared from the 2.5-kb genomic EcoRI-XbaI fragment (top) and then stripped and probed with antisense rp49 probe as a loading control (bottom). On the left, the lanes contained 10 μg of RNA from wild-type second and third instar larvae (WT), homozygous l(2)024047 second instar larvae (Nip-B 02047), and heterozygous l(2)02047 third instar larvae (Nip-B02047/+). On the right, the lanes contained 10 μg of RNA from wild-type second instar larvae and second instar larvae homozygous for the indicated Nipped alleles: Nip-C 160.1, Nip-A222.3, Nip-A323, T(2;3)Nip-B359.1, and Nip-B407. The 7-kb transcript is absent or altered in size in Nip-B02047 mutants and reduced in size in T(2;3)Nip-B Nipped-B mRNA. Probes were as described above. The lanes contained 10 μg of total cellular RNA isolated from wild-type flies at different developmental stages: EE, 0 to 30 min after egg laying; LE, 30 min to 16 hr after egg laying; L1-L3, first to third instar larvae; P1-P4, first to fourth day of pupation; M, 0- to 1-day-old adult males; F, 0- to 1-day-old females.
Database searches reveal homologues of the Nipped-B protein in fungi, worms, and mammals. Only short expressed-sequence tags (ESTs) of Caenorhabditis elegans, mouse, and humans were identified. The human ESTs are from a variety of tissue-specific libraries, suggesting that the human homologues are widely expressed. The combined human ESTs, which do not represent a complete sequence, encode 411 amino acids. Residues 2-232 of the partial human protein overlap Nipped-B residues 1744-1994 with 34% identity and 52% similarity. In order of decreasing homology, the 2157-amino acid Rad9 protein of Coprinus cinereus, the 1583-amino acid Mis4 protein of Schizosaccharomyces pombe, and the 1493-amino acid Scc2 protein of Saccharomyces cerevisiae, are more distantly related. Rad9 residues 669-2071 display 21% identity and 41% similarity to Nipped-B residues 576-1887, Mis4 residues 780-1492 have 19% identity and 41% similarity to Nipped-B residues 1110-1818, and Scc2 residues 697-1291 display 19% identity with 39% similarity to Nipped-B residues 1103-1704. The fungal homologues show similar levels of homology between themselves, but it is evident that there is a large conserved domain among all the proteins (Figure 10).
Consistent with the idea that Nipped-B plays an architectural role in enhancer-promoter communication, the fungal homologues of Nipped-B all participate in regulating chromosome structure, with roles in DNA repair (Valentineet al. 1995; Furuyaet al. 1998), meiotic chromosome condensation (Seitzet al. 1996), or sister chromatid cohesion (Michaeliset al. 1997; Furuyaet al. 1998). It has been proposed that these three fungal proteins define a new class of chromosomal proteins and have been named adherins to distinguish them from the cohesins that have similar functions (Furuyaet al. 1998).
—The Nipped-B protein shares a central core of homology with fungal chromosomal adherins. The proteins are diagrammed with the regions of homology in gray and the lengths of the protein on the right. Comparisons were made using the Blastp program (www.ncbi.nlm.nih.gov/gorf/wblast2.cgi/; Altschulet al. 1997) with the BLOSUM62 matrix, a gap open penalty of 11, a gap extension penalty of 1, gap x_dropoff of 50, expect value of 10.0, a wordsize of 3, and no filtering. Nipped-B protein is compared with the Coprinus cinereus (C.c.) Rad9 protein, which is compared with the Schizosaccharomyces pombe (S.p.) Mis4 protein, which is compared with the Saccharomyces cerevisiae (S.c.) Scc2 protein. The numbers near the ends of the lines that connect the homologous regions indicate the amino acid endpoints of the regions of homology detected by the alignment program, and the numbers between the proteins indicate the percentage of identical residues in the homologous regions and the percentage of residues that are either identical or similar (positives).
DISCUSSION
The Su(Hw) insulator protein that binds gypsy transposon DNA only blocks activation by an enhancer when the gypsy insertion is between the enhancer and promoter, suggesting that Su(Hw) is antagonistic to architectural factors that act between enhancers and promoters to facilitate enhancer-promoter communication. To identify putative architectural factors, we screened for mutations that reduce activation by a remote wing margin enhancer in the Drosophila cut gene partially blocked by a mutant Su(Hw) protein. To date this screen has identified five genetic loci that participate in the Notch receptor signaling pathway and promote cut expression in the wing margin: sd, mam, Chip, vg, and Nipped.
We have previously provided evidence that some genes identified in the screen directly regulate cut. Sd protein is a wing margin enhancer-binding protein (Morcilloet al. 1996), and Mam binds chromosomes in the vicinity of cut, suggesting that it is also a direct regulator (Bettleret al. 1996). Recently, it has been found that Vg protein interacts with Sd to regulate gene expression in the wing (Simmondset al. 1998). Thus, Sd likely recruits Vg to the cut wing margin enhancer.
The screen for cut regulators can also identify factors that act broadly to regulate gene expression. Chip is a ubiquitously expressed chromosomal protein required for maximal activation by several remote enhancers in diverse genes, including Ubx and even-skipped (Morcillo et al. 1996, 1997). In addition to its broad activities, the in vivo antagonism of Chip to the Su(Hw) insulator suggests that it may play a role in enhancer-promoter communication (Morcilloet al. 1996). Chip and its mammalian homologues, Nli (Ldb1/Clim-2) and Ldb2 (Clim-1), interact with themselves and various LIM and homeodomain proteins to promote formation of homo- and heterotypic transcription factor complexes (Agulnicket al. 1996; Jurata et al. 1996, 1998; Bachet al. 1997; Jurata and Gill 1997; Morcilloet al. 1997; Wadmanet al. 1997; Breenet al. 1998; E. Torigoi, C. Rosen, and D. Dorsett, unpublished observations). Thus, Chip and its homologues can act as cross-linking proteins that mediate interactions between diverse DNA-binding proteins. We speculate that in addition to helping form complexes on enhancers, promoting interactions between proteins that bind to DNA between enhancers and promoters could create chromatin loops that bring enhancers and promoters closer together.
Here we present genetic characterization of the Nipped mutations isolated in the screen for cut regulators and identification of a Nipped gene product. Although the data do not yet distinguish whether the heterochromatic Nipped locus is a single complex transcription unit or a cluster of distinct genes, we can draw several conclusions about Nipped functions and their roles relative to the other cut regulators. To summarize, Nipped mutations define three separable essential functions that regulate cut in the wing margin, provided by the Nipped-A, Nipped-B, and l(2)41Af lethal complementation groups. Dosage-sensitive genetic interactions indicate that Nipped-A and l(2)41Af cooperate closely with mam and vg in the regulation of cut. Similar to mam and unlike sd and vg, Nipped-A and l(2)41Af also modulate Notch receptor signaling or expression in multiple tissues. Nipped-B has the most unique function. Like Chip, Nipped-B regulates both cut and Ubx and is antagonistic to insulation by Su(Hw). Together, the antagonism to Su(Hw) and the homology to chromosomal adherins lead us to propose that Nipped-B protein performs an architectural role in enhancer-promoter communication.
Nipped-B is antagonistic to Su(Hw) insulator activity: The primary evidence that Nipped-B is antagonistic to Su(Hw) insulator activity is that Nipped-B activity is only strongly limiting for cut expression when there is a gypsy insertion between the wing margin enhancer and promoter. Strikingly, in contrast to mutations disrupting any of the other cut regulators (Jack and Delotto 1992, Morcilloet al. 1996; R. A. Rollins and D. Dorsett, unpublished observations), including sd, mam, Chip, vg, Nipped-A, and l(2)41Af, heterozygous Nipped-B mutations do not detectably reduce activation by the partially crippled wing margin enhancer in ct53d. Compared with sd, mam (Morcilloet al. 1996), or Nipped-A mutations, heterozygous Nipped-B mutations also only slightly reduce activation of cut expression by the solo wild-type wing margin enhancer present in ct2s heterozygotes. Therefore, with both ct53d and ct2s, Nipped-B products are less limiting for wing margin enhancer activity than are Nipped-A products. Remarkably, the opposite is true when there is a gypsy insulator insertion in cut. Heterozygous Nipped-B mutations are severalfold more effective than Nipped-A mutations in magnifying the effect of the Su(Hw) insulator in ctL-32; su(Hw)e2 flies. Furthermore, of the known cut regulators, only Chip (Morcilloet al. 1996) and Nipped-B mutations magnify the effect of the Su(Hw) insulator in su(Hw)e2 bx34e flies. The antagonism between Nipped-B and Su(Hw) is unlikely to be specific to the Su(Hw)e2 protein. Su(Hw)e2 has an amino acid substitution in a zinc finger that reduces DNA-binding activity but contains a wild-type enhancer-blocking domain (Harrisonet al. 1993; Kimet al. 1996b). Moreover, Nipped-B mutations also reduce cut expression in the absence of a gypsy insertion, indicating that the increased effectiveness of Su(Hw)e2 in Nipped-B mutants reflects a change in cut regulation rather than a change in Su(Hw)e2 protein activity.
Does Nipped-B directly regulate cut? The available data are insufficient to determine with absolute certainty whether or not Nipped-B directly regulates cut. However, direct regulation provides the simplest explanation for several observations. First, as shown above, the ability of Nipped-B mutations to exacerbate different cut mutant phenotypes differs from all other cut regulators such as sd, vg, and mam. Therefore, Nipped-B does not regulate cut indirectly by altering expression of any of the other known cut regulators. Moreover, the effects of the Nipped-B407 mutation on cut and Ubx mutant phenotypes are dominant, although Nipped-B407 only partially reduces Nipped-B mRNA levels. A partial loss of Nipped-B activity is unlikely to cause a similar or greater loss of activity of another cut regulator. Therefore, in light of the observation that Nipped-B mutations magnify insulation by gypsy insertions in both cut and Ubx, we strongly favor the idea that Nipped-B products directly support enhancer-promoter communication in cut and Ubx. Because Nipped-B is essential and Nipped-B mRNA is expressed at all developmental stages, it may play a similar role in other genes.
Nipped-B protein homologues have diverse roles in chromosome structure: The hypothesis that Nipped-B protein plays an architectural role to facilitate enhancer-promoter interactions in cut and Ubx is supported by the diverse effects that the fungal adherin homologues of the Nipped-B protein have on chromosome structure and function. The Rad9 protein of Coprinus was identified in a screen for radiation-sensitive mutants (Valentineet al. 1995). rad9 mutants were subsequently observed to display defects in synaptonemal complex formation and chromosome condensation during meiosis (Valentineet al. 1995; Seitzet al. 1996). Mutations in the Scc2 gene of budding yeast were identified as lethal temperature-sensitive mutants that display defects in sister chromatid cohesion during mitosis (Michaeliset al. 1997). In scc2 mutants, sister chromatids separate prematurely, just after formation of the bipolar spindle. Mutations in the Mis4 gene of fission yeast were identified as temperature-sensitive lethal mutants that missegregate minichromosomes (Takahashiet al. 1994). mis4 mutants also missegregate regular chromosomes and are radiation sensitive (Furuyaet al. 1998). The Mis4 protein is required during S phase and associates with chromosomes during the entire cell cycle (Furuyaet al. 1998). These diverse mutant phenotypes indicate that adherins play fundamental roles in chromosome structure.
Although we do not yet know if Nipped-B also participates in mitotic or meiotic chromosome structure, its homology to adherins suggests explanations for how Nipped-B could architecturally facilitate enhancer-promoter communication. It is tempting to speculate, for example, that the biochemical activity of Nipped-B is to recognize and stabilize chromatin loops that hold distant chromosomal sites closer together. The chromatin loops could be created by other factors involved in enhancer-promoter interactions.
Acknowledgments
We thank Ethan Ubell and Woo Sung Choi for assistance with complementation tests, Nina Lampen for scanning electron microscopy, Arthur Hilliker, Mark Fortini, Kathy Matthews, and Michael Young for providing fly stocks, Mary K. Baylies, Mark Ptashne, Vince Pirrotta, and Christina Rosen for helpful discussions and comments on the manuscript, and Jaeseob Kim, Andrew Simmonds, and John Bell for helpful discussions and sharing unpublished results. This work was supported by National Science Foundation research grant 9404771 to D.D., and National Institutes of Health Cancer Center Support Grant NCI-P30-CA-08748 to Memorial Sloan-Kettering Cancer Center. R.A.R. is a Jack and Susan Rudin Scholar.
Footnotes
-
Communicating editor: J. A. Birchler
- Received December 24, 1998.
- Accepted March 1, 1999.
- Copyright © 1999 by the Genetics Society of America