The NURD and Sin3 histone deacetylase complexes are involved in transcriptional repression through global deacetylation of chromatin. Both complexes contain many different components that may control how histone deacetylase complexes are regulated and interact with other transcription factors. In a genetic screen for modifiers of wingless signaling in the Drosophila eye, we isolated mutations in the Drosophila homolog of p66, a protein previously purified as part of the Xenopus NURD/Mi-2 complex. p66 encodes a highly conserved nuclear zinc-finger protein that is required for development and we propose that the p66 protein acts as a regulatory component of the NURD complex. Animals homozygous mutant for p66 display defects during metamorphosis possibly caused by misregulation of ecdysone-regulated expression. Although heterozygosity for p66 enhances a wingless phenotype in the eye, loss-of-function clones in the wing and the eye discs do not have any detectable phenotype, possibly due to redundancy with the Sin3 complex. Overexpression of p66, on the other hand, can repress wingless-dependent phenotypes. Furthermore, p66 expression can repress multiple reporters in a cell culture assay, including a Wnt-responsive TCF reporter construct, implicating the NURD complex in repression of Wnt target genes. By co-immunoprecipitation, p66 associates with dMi-2, a known NURD complex member.
A key event in most signal transduction pathways is the activation or repression of target genes in the nucleus by transcriptional regulators. In recent years, it has become evident that these transcription factors interact with chromatin and that regulation of chromatin structure plays an important role in controlling gene expression. One important mechanism for regulating chromatin structure involves histone acetylation/deacetylation. Histone acetylases are implicated in transcriptional activation while histone deacetylases are involved in repression (Kadosh and Struhl 1998; Kuo et al. 1998; Wang et al. 1998).
One of the major histone deacetylase (HDAC) complexes is the NURD complex (reviewed in Ahringer 2000). The complex has been purified from mammalian and Xenopus cells and homologs have been identified in Caenorhabditis elegans, Arabidopsis, and Drosophila, suggesting that the complex is conserved throughout plants and animals (Tong et al. 1998; Wade et al. 1998; Xue et al. 1998; Zhang et al. 1998, 1999; Ahringer 2000). The NURD complex is ∼2 MD in size and is composed of eight proteins (reviewed in Ahringer 2000; Ng and Bird 2000). The histone deacetylases HDAC1 and HDAC2 and two histone-binding proteins, RbAp48 and RbAp46, are found in the NURD complex as well as in the other predominant cellular HDAC complex, the Sin3 complex. In addition to these proteins, the other proposed members of the NURD complex are Mi-2, MBD3, MTA1, and p66 (Tong et al. 1998; Wade et al. 1998, 1999; Xue et al. 1998; Zhang et al. 1998, 1999; Brackertz et al. 2002; Feng et al. 2002).
Recently, there has been evidence that the NURD complex is involved in a variety of developmental functions. These include embryonic patterning in Drosophila, C. elegans, and mice, repression by Polycomb proteins in Drosophila, and repression in mouse T-cell development (Ahringer 2000). The NURD complex as well as the Sin3 complex has also been implicated in repression by unliganded nuclear hormone receptors (Heinzel et al. 1997; Xue et al. 1998). In chromatin immunoprecipitation assays, both the Sin3 and the NURD complex were found to be associated with the Xenopus thyroid hormone receptor (TR)β A gene promoter region, suggesting that they could function redundantly to repress transcription (Li et al. 2002). In Drosophila, dSin3A regulates transcription mediated by the ecdysone nuclear hormone receptor (EcR), but there are no data implicating the NURD complex in regulation of ecdysone responses (Tsai et al. 1999).
The NURD complex has also been implicated in Wnt signaling in Drosophila where it may be involved in repression by TCF. dMi-2 is found in a complex with the histone deacetylase Rpd3, which in turn interacts with Groucho, a corepressor for many proteins, including the TCF/LEF-1 family of proteins (Cavallo et al. 1998; Chen et al. 1999; Brehm et al. 2000). In the absence of a Wnt signal, TCF is associated with the corepressors Groucho and CtBP and represses transcription (Cavallo et al. 1998; Valenta et al. 2003). When the Wnt signal is received, the downstream effect is the stabilization of β-catenin protein, which can then translocate into the nucleus where it forms a complex with TCF to activate transcription of target genes (Brunner et al. 1997; van de Wetering et al. 1997). Presumably, β-catenin disrupts the association of TCF with its corepressors so that it can now activate transcription. The mechanism of activation by TCF and β-catenin is not well understood, but recent evidence suggests that histone acetylation is important for this process (Wade et al. 1999; Billin et al. 2000; Hecht et al. 2000; Takemaru and Moon 2000). In addition, mutations in the C. elegans NURD component egl-27 cause Wnt-like phenotypes in C. elegans (Herman et al. 1999), providing another link between Wnt signaling and the NURD complex.
In this study we report the identification and characterization of the Drosophila homolog of p66, a protein associated with the NURD complex. p66 is a nuclear protein that represses wingless (Wg) target genes and plays an essential role in development through the regulation of ecdysone-responsive genes. These results highlight the requirements for p66 and the NURD complex in regulating developmental decisions.
MATERIALS AND METHODS
The UAS-p66 construct was made by inserting the ApaI (blunted)-SmaI insert from LD18074 into the BglII (blunted) site in the UAS vector. Transgenic lines were established using standard methods. For the screen, the following stocks were used: P[sev-wgts] (Cadigan et al. 2002), P[sev-ras1V12] (Karim et al. 1996), and P[sev-ras1N17] (Allard et al. 1996). The following lines were used for the rescue experiment and for overexpression studies: arm-GAL4 (Sanson et al. 1996), hs-GAL4 (Kraus and Lis 1994), da-GAL4 (Wodarz et al. 1995), ptc-GAL4 (Ingham and Fietz 1995), TM3 P[Kr-GFP] (Casso et al. 1999), UAS-p66, and UAS-lacZ (Brand and Perrimon 1993). For germline clone mosaics, mutants were recombined onto P[FRT]2A chromosomes. For somatic mosaic clones, mutants were recombined onto either P[FRT]2A chromosomes or P[FRT]80B chromosomes. Imprecise excisions were generated from P l(3)01814[j4A5] (from L. and Y. Jan collection) using standard methods.
To test the phase during which p66 absence is lethal, we crossed p66 alleles A and L/TM3, sb to sb/TM3, Ser, actin-GFP. Sibling crosses were made from adults carrying the Ser marker and expanded into stocks. GFP-positive or -negative second instar larvae were collected from embryo collections of p66/TM3, Ser, and actin-GFP. Forty larvae were placed into each of three small vials containing food and allowed to develop. The number of pupae in each vial was counted ∼9 days later. Dead pupae were dissected from their pupal cases and examined under a dissecting microscope.
Genetic screen and scanning electron microscopy pictures:
The genetic screen was performed as previously described (Cadigan et al. 2002). Flies were prepared for scanning electron microscopy as previously described (Cadigan and Nusse 1996). The samples were viewed with an AMR1000 SEM and photographed using Polapan 400 film (Kodak).
Germline mosaics were generated using the autosomal FLP-dominant female sterile technique as previously described (Chou and Perrimon 1996). w; p66A P[FRT]2A or w; p66J P[FRT]2A females were crossed to ywP[FLP]22; P[ovoD1]3L P[FRT]2A males. The progeny from this cross were heat-shocked at late third instar for 1 hr at 37°. Mosaic mothers were crossed to w; p667/TM3 P[Kr-GFP] males.
Generation and staining of somatic mosaic clones:
These mosaics were generated with the FLP-FRT system as previously described (Xu and Rubin 1993). p66− P[FRT]2A or p66− P[FRT]80 flies were crossed to hs-FLP; P[pi-myc], P[FRT80B], or hs-FLP; P[pi-myc] P [FRT]2A. Clones were induced by heat shock of 24- to 48-hr after-egg-laying progeny at 37° for 1 hr. For Ac and Dll staining, p6616 P[FRT]80B stock was used. For examination of phenotypes in adult wings and legs, the stocks used were p66J P[FRT]2A, p66A P[FRT]2A, and p6616 P[FRT]80B.
Overexpression of p66:
UAS-p6618 were crossed to Ptc-GAL4 flies. Discs were dissected and fixed as above. Adult flies were raised at 25° and 29°.
In situ hybridization and whole-mount immunostaining:
p66 in situ hybridization and staining was done as previously described (Cadigan and Nusse 1996), as was staining of imaginal discs (Cadigan et al. 1998). The primary antibodies were used at the following dilutions: rat anti-Wg, 1:200; rabbit anti-Wg, 1:8; rabbit anti-lacZ (Cappel), 1:250; and anti-myc, 1:5 (S. Blair, University of Wisconsin); anti-Ac, 1:4 (Developmental Studies Hybridoma Bank); anti-Dll, 1:150 (G. Panganiban); anti-rat p66, 1:50; and anti-rabbit p66, 1:50. For fluorescence microscopy, antibodies were used at the following dilutions: donkey Alexa anti-mouse, 1:1000 (Molecular Probes, Eugene, OR), and donkey Cy3 anti-rabbit, 1:200 (Jackson Immunochemicals). Confocal images were collected with a Bio-Rad (Richmond, CA) MRC 1000 confocal laser setup attached to a Zeiss Axioscope microscope.
Generation of anti-p66 antibody:
p66-C protein (amino acids 873–917) was generated as follows: LD18074 was digested with XhoI and RI. The 250-bp fragment corresponding to amino acids 873–917 was ligated into the XhoI/RI sites of PGEX4-T3 (Pharmacia). p66-N protein (amino acids 116–303) was generated by digesting LD18074 with BamHI and NotI. The 0.5-kb fragment corresponding to amino acids 116–303 was ligated into the BamHI/NotI sites of PGEX4-T3 (Pharmacia). GST-fusion proteins were expressed in BL-21 bacteria and bound to glutathione sepharose beads. After washing, proteins were eluted in elution buffer (0.1% Triton, 200 mm NaCl, 50 mm Tris, pH 9.0). The p66-C protein was injected into rabbits and the p66-N protein was injected into rats (Josman Labs).
Cloning of the p66 gene:
The analysis of the p66 gene and the subsequent isolation of its cDNA were performed using standard molecular biological techniques. The p66 cDNA was isolated from a 0 to 5-hr cDNA library from wild-type embryos (Brown and Kafatos 1988) and is identical in sequence to EST LD18074 (Berkeley Drosophila Genome Project).
Rescue of the larval lethal phenotype:
w; P[UAS-p66]/P[UAS-p66]; p6610/TM6, Tb was crossed to w; arm-gal4/arm-gal4; p667/TM6, Tb. The progeny were reared at 18° and scored for the presence of non-Tubby pupae. As a control, w; +/+; p6610/TM6, Tb was crossed to w; arm-gal4/arm-gal4; p667/TM6, Tb. Two independent transgenic lines were used. Significancy was tested by means of the χ2 test.
RNA isolation and Northern blot:
Larvae were staged as previously described (Andres and Thummel 1994). RNA was collected using Trizol following the manufacturer's protocol (Invitrogen, San Diego). Northern blot was performed using standard methods. The presence of equal amounts of ribosomal RNA, visualized by ethidium bromide staining, served as loading control.
An overnight collection of embryos was dechorionated then resuspended in 10× volume of TNT buffer (1% Triton, 50 mm Tris, pH 7.5, 150 mm NaCl). Embryos were then homogenized using a glass douncer and centrifuged at 14,000 rpm for 5 min. A total of 100 μl of the supernatant was first precleared by the addition of protein-G sepharose for 1 hr. Subsequently, 3 μl of affinity-purified anti-rabbit antibody (p66-C) was added to the precleared supernatant and rotated at 4° for 4 hr. A total of 20 μl of a 1:1 protein G sepharose/TNT slurry was added and rotated for an additional 2 hr. The complex was spun down and washed three times in TNT. Immunoprecipitates were then analyzed by Western blotting. For Western, anti-dMi-2N was used at a concentration of 1:10,000 and anti-p66 rat (p66-N) was used at 1:1000.
pp66myc was generated by cloning the p66 gene in frame with myc in pBSmyc. LD18074 was digested with XbaI and StuI. The fragment containing the coding region was ligated into pBSmyc, which had been digested with EcoRI, treated with Klenow, and then digested with XbaI. The resulting plasmid, pp66myc, was then digested with XbaI and HindIII. The fragment containing the p66 gene was then ligated into the HindIII and XbaI sites of prk5SK to generate p66mycprk5.
Cell culture and luciferase assays:
A total 293 cells were maintained in Dulbecco minimal essential medium + 10% fetal bovine serum (Gemini) supplemented with penicillin and streptomycin. Cells were passaged every 2–3 days. For luciferase assays, 3 × 105 cells were plated in 60-mm wells and 1 μg of total DNA was transfected using Superfect (QIAGEN, Chatsworth, CA) according to protocol. For β-catenin activation assays, transfections consisted of 0.1 μg of optimized top flash reporter (OT) (Klymkowsky et al. 1999) and 0.1 μg of EF1α-βgal (gift from G. Crabtree) and may have included 0.2 μg of EF1α-β-catenin, as well as varying amounts of p66mycprk5. CMV-gfp was added as necessary to keep the total amount of DNA in each transfection constant. Cells were lysed 48 hr after transfection, and luciferase and β-galactosidase assays were carried out according to protocol (Tropix, Promega, Madison, WI). For SRE reporter assays, transfections consisted of 0.1 μg of the SRE reporter, 0.2 μg of M1 receptor, 0.1 μg of EF1α-βgal, and if p66 was included, 0.6 μg of p66myprk5. CMV-gfp was added as necessary to keep the total amount of DNA in each transfection constant. Following transfection using Superfect (QIAGEN), cells were kept in DMEM + 0.5% FBS. Twenty-four hours post-transfection, cells were stimulated with carbachol (5 μm, as indicated) for 6 hr and lysed according to protocol for assays. For the NF-AT reporter assay, cells were transfected with 0.2 μg of NF-AT luciferase reporter, 0.1 μg of EF1α-βgal, and if p66 was included, 0.6 μg of p66mycprk5. CMV-gfp was added as necessary to keep the total amount of DNA in each transfection constant. Twenty-four hours after transfection, cells were stimulated with ionomycin (1 μm), phorbol 12-myristate 13-acetate (PMA, 25 ng/ml), and calcium chloride (10 mm). After 24 hr, cells were lysed for assays. Data shown are from one representative experiment. All measurements were done in duplicate.
p66 enhances the P[sev-wgts] phenotype in the adult eye:
wg is required during many steps in Drosophila development and controls a wide range of patterning events (reviewed in Klingensmith and Nusse 1994). To search for novel genes interacting with wg, we performed a genetic screen to isolate dominant modifiers of a bristle phenotype caused by ectopic expression of wg in the eye. This screen has been described in more detail elsewhere (Cadigan et al. 2002), but briefly, we expressed a temperature-sensitive allele of wg, wgIL114, under the control of the sevenless promoter (sev-wgts) at an intermediate temperature (17.6°) that allows partial wg function (Figure 1, A and B). Whereas a wild-type wg allele expressed using this promoter almost completely blocks bristle formation, expression of the temperature-sensitive allele at an intermediate temperature allows formation of approximately one-third of the wild-type number of bristles (Figure 1B) (Cadigan et al. 2002).
We identified two dominant enhancers on the third chromosome that formed a lethal complementation group, which we named p66, due to its homology with p66 of the Mi-2/NURD complex (see below). The majority of P[sev-wgts]/+ eyes contained 100–150 bristles. In contrast, flies that are additionally heterozygous for either of two independent p66 mutations, p6610 and p667, have a reduced number of bristles (<100). These eyes otherwise appeared normal (Figure 1, C–E).
To determine if the mutations affected the sevenless promoter rather than wg function, we tested whether the loss of one copy of p66 modified two other transgenes under the control of the sevenless promoter, a dominant negative ras (sev-Ras1N17) and an activated ras (sev-Ras1V12). sev-Ras1V12 expression results in a rough eye caused by extra R7 cells (Karim et al. 1996). Similarly, sev-Ras1N17 also has a rough eye phenotype; however, it is due to loss of R7 cells (Allard et al. 1996). Removal of one copy of p66 did not have an effect on either the sev-Ras1V12 or the sev-Ras1N17 phenotype (data not shown). This suggests that p66 does not affect the expression of the transgene through the sevenless promoter, but rather affects Wg signaling or Wg targets. Because p66 was identified as a copurifying protein with the Mi-2/NURD complex (see below), we also tested whether dMi-2 (Kehle et al. 1998) has an effect on the sev-wgts phenotype. We found that loss of one copy of dMi-2 (dMi-26) also causes a reduction in the number of bristles in a sev-wgts (Figure 1E).
More alleles of p66 and phenotypes in homozygous mutant flies:
We found that the deficiency Df(3L)Lxd6 did not complement the p6610 or the p667 mutation. In addition, P element [j4A5], which maps within this deficiency, also failed to complement either p66 allele. Precise excision of the P element reverted the lethality, demonstrating that the P element was responsible for the phenotype. Using imprecise excision, we created a new allele, p6616. In addition, by screening for noncomplementing lethality over existing p66 alleles, we isolated five new alleles using EMS, p66A, p66H, p66J, p66L, and p66S. To determine the strength of each allele, we conducted complementation crosses and ordered the alleles on the basis of progression of development. On the basis of these results, the order of alleles is (from weak to strong): p6616, p667, p66L, p66S, p66H, p66J = p6610 = p66A = p66Lxd6.
In all heteroallelic crosses, we observed survival through embryogenesis into larval stages. Strong allele combinations, p66J/p66A or p66J/p66Lxd6, died during second or third instar larval stages. In weaker allele combinations, mutants developed to the prepupal stage, but did not progress to the pupal stage. Numbers ranged from 30 of 124 (total) larvae reaching pupal stages for weak alleles (using GFP balancers and counting larvae that crawled out of the food) to 0 of 101 (total) for strong alleles. Trans-heterozygotes of weak alleles, such as p6616/p667 or p66L/p6616, survived to late pupal stages. Upon examination, the pupae appeared almost wild type but displayed a variety of phenotypes. Mutant prepupae develop to stage P3 where the dorsal air pocket forms; however, the prepupae do not progress to stage P4 as no withdrawal of pupae to the posterior end is evident. These mutant pupae have defects including shortened or bent legs, bristle defects, a split notum, and wing abnormalities (Figure 2, A–E). Some mutants exhibit necrosis near the head region (Figure 2, G and H). Other mutants lacked pigmentation (Figure 2, C, E, and G). In addition, the cases surrounding the prepupae were filled with white fluid, possibly due to incomplete histolysis of the fat bodies (Figure 2I). In some mutants, accumulation of white fluid caused the pharate adult to stick to the pupal case, possibly preventing the otherwise normal adult from eclosing (not shown).
Some of the p66 mutants displayed phenotypes similar to defects in ecdysone responses such as ultraspiracle and ftz-f1 (Henrich et al. 1994; Broadus et al. 1999). Moreover, mutations in bonus, a gene encoding a homolog of the vertebrate TIF1 transcriptional cofactor, lead to defects in leg elongation, bristle development, and pigmentation. The bonus protein can bind to nuclear receptor proteins. Likewise, mutations in crooked leg, a gene induced by ecdysone, lead to leg defects similar to those seen in p66 (D'Avino and Thummel 1998). To follow up on a possible connection between p66 and ecdysone signaling, we examined ecdysone-regulated gene expression in the mutants.
Metamorphosis is initiated by a pulse of ecdysone, which activates a hierarchy of gene expression (reviewed in Riddiford 1993). One of the genes induced in the early puff is E74, an ETS domain transcription factor (Ashburner et al. 1974; Burtis et al. 1990). In wild-type animals, E74 was induced normally at the larval/prepupal transition. In contrast, we did not detect any E74 expression in the mutant animals (Figure 3). We also examined the expression of DHR3, an orphan receptor that is also induced directly by ecdysone (Koelle et al. 1992; Horner et al. 1995). DHR3 expression peaks at the prepupal stage just as early genes such as E74 are repressed (Koelle et al. 1992). Similar to E74 expression, DHR3 was expressed at lower levels in mutant animals, indicating a defect in activation of genes in response to ecdysone (Figure 3).
p66 appears to be maternally contributed because a p66 transcript is detected in 0- to 2-hr embryos on a Northern blot and in early embryos by in situ hybridization (Northern blot not shown; in situ hybridization in Figure 7A). However, p66 does not appear to be required for embryogenesis, because p66 mutant germline clone embryos that lack both maternal and zygotic p66 function survive to larval or pupal stages. To examine adult phenotypes of p66, we made homozygous loss-of-function clones in imaginal discs. Mutant clones for all alleles appeared to be roughly equal in size to their twin spots, indicating that p66 is not essential for cell viability or cell growth. This is shown for two alleles, p6616, a weak allele, and p6610, a strong allele (Figure 4, C and F). Figure 4G also shows that clones mutant for p6610 lack staining for P66 protein.
Because we hypothesized that p66 repressed downstream targets of wingless (Wg) signaling, we examined two targets of Wg in the wing disc, Distal-less (Dll) and Achaete (Ac) in p6616 clones. Removal of p66 did not affect expression of either gene (shown for Dll in Figure 4, A–E). In addition, mutant clones of the strong alleles p66J and p66A did not display any phenotype in adult wings or eyes (data not shown). Thus, we have been unable to find a loss-of-function phenotype in discs, which we attribute to possible redundancy with other histone deacetylase complexes (see discussion).
Cloning of the p66 gene:
The p6610 and p667 mutations were mapped by recombination and deficiency complementation to 67E-F. Three lethal P-element lines also mapped to 67E-F and did not complement p6610 (Figure 5A). Using the genomic DNA flanking these P elements, we detected 6.4- and 4.7-kb transcripts on a Northern blot (data not shown) and isolated a 3.5-kb cDNA that contained an open reading frame of 2751 nucleotides (CG32067). We sequenced the gene corresponding to the cDNA in the p66L mutant and detected a mutation (G > A) in a conserved exon/intron boundary. The mutant transcript would result in a truncated protein that lacks the last 80 amino acids (Figure 5, A and B).
The cDNA showed significant sequence similarity to Xenopus p66, a protein associated with the NURD complex (Wade et al. 1999). Homologs are also present in human, Xenopus, mouse, and C. elegans (Figure 5B) (Feng et al. 2002). All of the p66 proteins have a conserved zinc-finger motif that is similar to the zinc fingers of GATA transcription factors (Orkin 1992). Immediately adjacent to the zinc finger are a conserved proline-rich region and a conserved basic region (Figure 5B).
To demonstrate that this cDNA encodes the functional p66 protein, we tested whether its expression could rescue p66 mutants using the UAS/GAL4 system (Brand and Perrimon 1993). We expressed p66 in heteroallelic homozygous p6610/p667 mutants and scored for survival to pupal stages. In the absence of transgenic p66 expression, ∼5% of p6610/p667 mutants survived to the pupal stage. However, expression of p66 almost completely rescued this mutant combination (Figure 5C). This result was confirmed using two independent transgenic lines, providing additional evidence that the p66 gene corresponds to the cDNA that we cloned.
Overexpression of p66:
To explore p66 function further, we overexpressed the gene using the UAS-GAL4 system. When we overexpressed p66 using Daughterless-GAL4, a strong embryonic driver, the animals survived embryogenesis, but died during first instar larval stage. We also overexpressed p66 using the Ptc-GAL4 driver, which is expressed in the wing, leg, and notum (Figure 6, B, D, H, and I). Ptc-GAL4; UAS-p66 flies lacked scutellar notum bristles (Figure 6, A and C). The formation of these bristles is wg dependent (Phillips and Whittle 1993); however, wg expression in these animals is similar to the wild-type pattern (Figure 6, B, D, H, and I; we do, however, note that the high intensity of staining for P66 protein in the red channel leads to signal “bleeding” into the green channel used to detect Wg protein). Therefore, p66 does not repress wg itself, but likely represses downstream wg targets in the notum. The adults resulting from overexpression of p66 in imaginal discs had normally patterned wings, but the wings were smaller in size (data not shown). In addition, the legs had elongation defects and often resembled stumps (Figure 6, E–G).
p66 is a ubiquitously expressed nuclear protein:
At the RNA level, p66 is ubiquitously expressed throughout embryogenesis. The transcripts were present by Northern blot in 0- to 2-hr embryos, suggesting that p66 is maternally provided (data not shown). In early syncytial stages, we detected the transcript by in situ hybridization in agreement with our Northern data (Figure 7A). During stage 10, higher levels of the transcript are observed in the presumptive neuroectoderm (Figure 7B). After germband extension, p66 is highly expressed in the central nervous system (Figure 7C).
In addition, we examined p66 protein distribution using antibodies generated to both the C terminus and the N terminus of p66. The distribution of p66 in the embryo, as detected by both antibodies, was ubiquitous, similar to the transcript pattern (Figure 7, D–F). p66 was also expressed in both follicle and nurse cells of the ovary and clearly localized to the nucleus, which is consistent with a role in transcriptional regulation (Figure 7G). Salivary gland cells also displayed p66 staining in the nucleus (Figure 7H). No staining was observed in homozygous p66 mutant clones in the disc, providing further evidence that the p66 cDNA corresponds to the p66 gene as well as demonstrating specificity of the antibody (Figure 4, F and G).
p66 is associated with the NURD complex:
To determine whether Drosophila p66 is associated with the Drosophila NURD complex, we tested for interaction between p66 and dMi-2. Mi-2 is a component that is specific to the NURD complex and not found in the Sin3 complex (reviewed in Ahringer 2000).
We made Drosophila extracts of 0- to 24-hr embryos. We used either the p66 C-terminal antibody (p66-C) or the dMi-2N antibody for immunoprecipitation from these embryonic extracts and analyzed the components by Western blot (Figure 8). The p66-C antibody immunoprecipitated both the p66 protein and dMi-2, while the preimmune serum failed to precipitate either protein. Similarly, using the dMi-2 antibody, we were able to immunoprecipitate p66 (Figure 8). From these experiments we conclude that p66 is in a complex with the Drosophila NURD complex.
p66 can repress a TCF-responsive reporter:
Billin et al. (2000) have demonstrated that inhibition of histone deacetylases causes derepression of Wnt-responsive promoters. In addition, they found that under repressive conditions, LEF-1 interacts with HDAC1 independently of mSin3A, suggesting that LEF-1 mediates repression through the NURD histone deacetylase complex. In agreement, we found that removal of p66 enhanced the sev-wgts phenotype and overexpression repressed wg-dependent bristles without affecting wg expression. In combination with our finding that p66 associates with the NURD complex, these experiments suggest that p66 could be involved in repression by TCF/LEF-1 family members.
To further investigate this possibility, we examined the effects of p66 expression on the activation of the TCF-responsive OT (Klymkowsky et al. 1999). We transfected 293 cells with the OT reporter and β-catenin. As expected, β-catenin efficiently activated the OT reporter. Coexpression of p66 inhibited β-catenin activation of the OT reporter in a dose-dependent manner, maximally inhibiting activation by 70% (Figure 9A). The ability of p66 to repress the OT plasmid was not restricted to 293 cells as we also observed repression of the OT reporter in NIH3T3 cells (data not shown).
To determine if p66 is a specific repressor of the Wnt pathway or a more general repressor, we tested whether p66 could repress other reporter genes. We found that p66 could repress activation of the SRE reporter by the M1 receptor (Figure 9C). Additionally, coexpression of p66 also inhibited activation of the NF-AT reporter by calcium signaling (Figure 9D). However, p66 did not repress an albumin luciferase reporter, demonstrating that there is some specificity in its action (Figure 9B).
NURD complex in Wnt signaling:
Through a genetic screen for modifiers of wingless signaling, we identified mutations in p66, the Drosophila homolog of p66, a protein that had been previously identified as part of the NURD chromatin-remodeling/histone deacetylase complex (Wade et al. 1999). Loss of p66 enhanced the sev-wgts phenotype in the eye; i.e., the resulting phenotype resembled increased wingless signaling. Since loss of p66 did not affect other transgenes expressed using the same promoter as sev-wgts, we believe that removal of p66 does not affect wingless expression through the sevenless promoter, but instead represses wingless target genes involved in bristle formation. Our results are in agreement with previous experiments that have implicated the NURD complex in Wnt signaling. It has been reported that Lef-1 repression involves HDAC-1 function. This interaction takes place in the absence of mSin3A, leading to the hypothesis that Lef-1 repression involves recruitment of the NURD complex (Billin et al. 2000). Here, we present three pieces of evidence to further support this hypothesis. First, loss of p66 enhances a wg overexpression phenotype. Second, overexpression of p66 can repress activation of a TCF reporter by β-catenin in tissue culture cells. Finally, overexpression of p66 in vivo represses the formation of wingless-dependent scutellar bristles without repressing wingless expression. Together, these experiments provide additional evidence that the NURD complex is involved in repression of Wnt target genes.
p66 and the ecdysone pathway:
The lethality of p66 mutants is caused by misregulation of ecdysone-regulated genes during larval stages. If p66 was involved in repression of ecdysone-induced targets such as E74 and DHR3, then loss of p66 should lead to ectopic gene expression. In contrast, we have found that ecdysone-induced genes are not activated, suggesting that the role of p66 is more complex and indirect. In microarray studies of ecdysone response, 44% of genes that changed expression were repressed (White et al. 1999). We speculate that repression of one or more of these genes is required for ecdysone-induced expression and that p66 is required for this process.
An additional connection between the NURD complex and ecydsone response is made through Bonus (bon), the Drosophila homolog of TIF1. TIF1 was identified as a protein that interacts with HP1, a heterochromatin-associated protein (Le Douarin et al. 1996). The NURD complex may be involved in histone modification to allow HP1 binding (Nishioka et al. 2002; Zegerman et al. 2002). Similar to p66 mutants, bon mutants die during pupal stages due to misregulation of ecdysone-induced genes. Furthermore, E74 expression is also reduced in bon mutants (Beckstead et al. 2001). Thus, we speculate that p66 and the NURD complex may be involved in regulation of ecdysone response through HP1-mediated repression.
p66 functions as a repressor:
p66 mutations can affect both wg and ecdysone-induced gene expression. Furthermore, in cell culture reporter assays, we found that expression of p66 inhibits activation of a TCF reporter by β-catenin, of a SRE reporter by the M1 receptor, and of an NF-AT reporter by high levels of intracellular calcium. In addition, two human p66 homologs, hp66α and hp66β, function as transcriptional repressors when tethered to a promoter, suggesting that transcriptional repression is a shared activity of p66 proteins (Brackertz et al. 2002; Feng et al. 2002). However, p66 did not repress an albumin luciferase reporter, implying that expression of p66 does not cause a defect in general transcription. Therefore, we conclude that p66 can repress multiple signaling pathways, and we hypothesize that this repression is mediated by recruitment of the NURD histone deacetylase complex. This conclusion is also supported by previous reports that expression of p66 can change the localization of MBD3, a component of the NURD complex (Feng et al. 2002).
Although overexpression of p66 can repress multiple signaling pathways, we have been unable to detect loss-of-function phenotypes that would be consistent with this hypothesis. In Drosophila, relatively mild and tissue-specific phenotypes for repressors have also been found for Pangolin/dTCF (Schweizer et al. 2003) and naked cuticle (Rousset et al. 2001). With respect to Pangolin/dTCF, this could be due to its dual role as both a repressor and an activator (Schweizer et al. 2003), but in other cases, a lack of phenotypes is possibly due to redundancy among parallel pathways (not necessarily among related genes). This suggestion is in analogy to the SynMuv genes in C. elegans. Animals mutant in either a synMuvA or a synMuvB gene alone have a normal vulva; animals mutant for both a synMuvA gene and a synMuvB gene have a multi-vulval (Muv) phenotype (Horvitz and Sulston 1980; Ferguson and Horvitz 1989). Indeed, in the C. elegans vulva, p66 functions in a redundant pathway (G. Poulin, F. Solari and J. Ahringer, personal communication).
p66 and other components of the NURD complex:
Initial preparations of the NURD complex from mammalian cells did not contain p66 (Tong et al. 1998; Xue et al. 1998; Zhang et al. 1998), but p66 was present when the complex was purified from Xenopus oocytes (Wade et al. 1999). Subsequently, p66 was found to be associated with the NURD complex in mammalian cells in association with the methyl DNA-binding protein MBD2, as part of the MeCP1 complex. Is p66 a component of the NURD complex or an accessory factor? We co-immunoprecipitated dMi-2 and p66, suggesting that they form a complex. However, if all of the complex members participate in the same processes, then the corresponding mutants should also have the same phenotypes. To ascertain whether p66 and rpd3 (HDAC homolog) function together is difficult since rpd3 also participates in the Sin3 complex (reviewed in Ahringer 2000). Thus, it is likely that rpd3 will display a wider range of phenotypes than a mutant of the NURD complex alone.
However, we can compare the p66 mutant to the other NURD complex mutant characterized in Drosophila, dMi-2. dMi-2 is required for oogenesis and for cell viability (Kehle et al. 1998). rpd3 is also likely important for oogenesis, as a hypomorphic rpd3 allele produces very few eggs (Mannervik and Levine 1999). In contrast, p66 mutant germlines produce normal embryos, and p66 mutant clones survive in third instar imaginal discs. dMi-2 zygotic mutants die during larval stages, which coincides with the lethality of strong allele combination of p66 mutants (Kehle et al. 1998). It is possible that dMi-2 may also be required for ecdysone response in larval stages. Since the p66 mutant phenotype does not completely mimic Mi-2 mutant phenotypes, we suggest that p66 is not a core component of the NURD complex, but could have a regulatory function. However, it is equally possible that differences in allelic strength or perdurance of maternal contributions obscures the full range of phenotypes on the several components.
Our results are further supported by experiments demonstrating that dMi-2 and the Drosophila MBD2/3 protein do not colocalize in nuclei. dMi-2 is distributed ubiquitously in embryonic nuclei, while Drosophila MBD2/3 is localized in a speckled pattern (Marhold et al. 2004). This result suggests that Drosophila MBD2/3 is not an integral component of all dMi-2 complexes.
How might p66 function? In mammalian cells, human p66 protein is associated only with the NURD complex as part of the MeCP1 complex, which additionally contains methylated DNA-binding activity through the MBD2 protein (Hendrich and Bird 1998; Feng and Zhang 2001; Feng et al. 2002). Similarly, the Xenopus NURD complex, which copurifies with p66, in contrast to mammalian NURD complexes, also has methylated DNA-binding activity (Wade et al. 1999). The NURD complex, through Mi-2, interacts with the zinc-finger proteins Hunchback, Tramtrak, and Ikaros (Kehle et al. 1998; Kim et al. 1999; Murawsky et al. 2001). We hypothesize that p66, also a zinc-finger-containing protein, functions similarly to these transcription factors to recruit the NURD complex to methylated DNA.
This hypothesis is supported by experiments that demonstrate that human p66 interacts with MBD2 and MBD3 in vitro. Furthermore, overexpression of p66 can change localization of MBD3 (Feng and Zhang 2001; Brackertz et al. 2002). On the basis of these results, we hypothesize that p66 may function as a link between the NURD complex and the methylated DNA-binding proteins MBD2 and MBD3. We propose that p66 recruits the NURD complex to mediate methylation-mediated silencing. Since both DNA methyl transferases and DNA-methylated binding proteins exist in Drosophila, it is likely that there is some vestige of a methylation system in Drosophila, although the function is unknown at this time (Tweedie et al. 1999; Gowher et al. 2000; Lyko et al. 2000).
We thank Matt Fish for making transgenic Drosophila lines and other experimental assistance. Alexander Brehm (Adolf-Butenandt-Institut Munich) kindly provided the Dmi-2 antibody and Carl Thummel kindly provided the cDNAs for E74 and DHR3. We also thank Julie Ahringer (Cambridge) for exchange of information on the C. elegans p66 homolog. These studies were supported by a grant from the U. S. Army Medical Research and Material Command (DAMD17-99-1-9386). C.K. was supported by the National Science Foundation and the Howard Hughes Medical Institute. S.L.S. was supported by the Human Frontier Science Program. R.N. is an investigator of the Howard Hughes Medical Institute.
Communicating editor: A. J. Lopez
- Received August 11, 2004.
- Accepted January 6, 2005.
- Genetics Society of America