Abstract
The products of trithorax group (trxG) genes maintain active transcription of many important developmental regulatory genes, including homeotic genes. Several trxG proteins have been shown to act in multimeric protein complexes that modify chromatin structure. ASH2, the product of the Drosophila trxG gene absent, small, or homeotic discs 2 (ash2) is a component of a 500-kD complex. In this article, we provide biochemical evidence that ASH2 binds directly to Skittles (SKTL), a predicted phosphatidylinositol 4-phosphate 5-kinase, and genetic evidence that the association of these proteins is functionally significant. We also show that histone H1 hyperphosphorylation is dramatically increased in both ash2 and sktl mutant polytene chromosomes. These results suggest that ASH2 maintains active transcription by binding a producer of nuclear phosphoinositides and downregulating histone H1 hyperphosphorylation.
THE homeotic genes of the Antennapedia complex and bithorax complex specify the segmental identities of the fruit fly Drosophila melanogaster (Duncan 1987; Kaufman et al. 1990). During early embryogenesis, the initial expression patterns of homeotic genes are established by the gap and pair rule genes (Akam 1987). After gastrulation, expression of the gap and pair rule genes subsides, and the products of the Polycomb group (PcG) and trithorax group (trxG) genes maintain the proper expression patterns of the homeotic genes. PcG proteins are required for the maintenance of homeotic gene repression, while trxG proteins are required for the maintenance of homeotic gene activation (Kennison 1995; Simon 1995). PcG and trxG proteins are thought to act in large multimeric protein complexes that modify chromatin structure, organizing it into either a “closed” or an “open” conformation (Mahmoudi and Verrijzer 2001). In addition to regulating homeotic genes, PcG and trxG proteins regulate many other genes, implying that the faithful maintenance of active or repressed states may be a general mechanism of development (Francis and Kingston 2001).
The trxG gene absent, small, or homeotic discs 2 (ash2) was found in a screen for late larval/early pupal lethals with imaginal disc abnormalities (Shearn et al. 1971). Homozygous ash2 mutants exhibit homeotic transformations characteristic of loss-of-function mutants in homeotic genes (Shearn et al. 1987). For example, the haltere and third-leg imaginal discs of ash2 trans-heterozygotes show reduced accumulation of the homeotic gene product, Ultrabithorax, and the first leg discs show complete loss of Sex combs reduced (LaJeunesse and Shearn 1995). ash2 mutants display intergenic noncomplementation with other trxG gene mutants (Shearn 1989). In addition, ash2 mutants display pattern formation abnormalities of the legs and wings (Adamson and Shearn 1996). The predicted ASH2 protein contains a target sequence for early degradation (PEST sequence; Rogers et al. 1986), a domain of unknown function (SPRY domain; Ponting et al. 1997), and a bipartite nuclear localization signal (NLS-BP; Robbins et al. 1991). The nuclear localization signal is functional as ASH2 is localized in nuclei of salivary gland cells (Adamson and Shearn 1996). Most significantly for this report, ASH2 contains a PHD finger, a putative double zinc finger involved in mediating protein-protein interactions and modifying chromatin structure (Aasland et al. 1995) and implicated in functioning as a nuclear phosphoinositide receptor (Gozani et al. 2003). Biochemical studies reveal that ASH2 is a subunit of a 500-kD multiprotein complex (Papoulas et al. 1998). Additional components of the complex have yet to be determined in Drosophila. However, biochemical purification of the SET1 protein complex in Saccharomyces cerevisiae reveals that one of the proposed subunits, Bre2p, contains a SPRY domain and another subunit, Saf41p, contains a PHD finger. It has been proposed that Bre2p and Saf41p together constitute a bipartite functional homolog of ASH2 (Nagy et al. 2002). The SET1 complex contains SET1, a SET domain-containing protein that has been shown to methylate lysine 4 of histone H3 (Miller et al. 2001; Roguev et al. 2001; Nagy et al. 2002). Human and Schizosaccharomyces pombe versions of this complex have ASH2 homologs that contain both a SPRY domain and a PHD finger; these complexes have also been shown to methylate lysine 4 of histone H3 (Roguev et al. 2003; Wysocka et al. 2003). The human ASH2 homolog is 47% identical to Drosophila ASH2 (Ikegawa et al. 1999).
Phosphoinositol lipids in the cytoplasm play important roles in growth, differentiation, and vesicular secretion. Phosphatidylinositol 4-phosphate (PtdIns[4]P or PIP) is phosphorylated by phosphatidylinositol 4-phosphate 5-kinase (PIP5K) to become phosphatidylinositol 4,5-bisphosphate (Ptd[4,5]P2 or PIP2; Boronenkov and Anderson 1995; Ishihara et al. 1996). PIP2 is a second messenger thought to modulate the functions of cytoskeletal regulatory proteins such as profilin, coilin, fascin, and gelsolin (Janmey 1994). PIP2 also regulates vesicular trafficking and platelet activation (Toker 1998). Phospholipase C hydrolyzes PIP2 to produce the second messengers diacylglycerol (DAG) and inositol trisphosphate (IP3). DAG activates protein kinase C (PKC), while IP3 releases calcium from intracellular stores (Rana and Hokin 1990). In addition, PIP2 is converted into phosphatidylinositol 3,4,5-trisphosphate, which activates some PKC isoforms (Toker et al. 1994). Phosphoinositide metabolism is also involved in signal transduction and cytoskeleton regulation via interaction with the Rho family of small G proteins (Chong et al. 1994; Ren et al. 1996). An interaction has also been implicated between phosphoinositides and receptor tyrosine kinases (Cochet et al. 1991). There are two types of PIP5Ks, PIP5KI and PIP5KII, with distinct biochemical properties (Anderson et al. 1999). PIPKIs prefer to phosphorylate PI-4-P to PI-4,5-P2 (Anderson et al. 1999), while the preferred substrate of PIPKII is PI-5-P rather than PI-4-P (Rameh et al. 1997). PIP5KI has been shown to be required for vesicular secretion in PC12 cells (Hay et al. 1995), while PIP5KII may be involved in vesicular trafficking in budding yeast (Yamamoto et al. 1995).
Phosphoinositides are also present in the nucleus (Divecha et al. 1993). There is growing evidence that members of the phosphoinositide pathways are involved in post-transcriptional modification and chromatin-mediated gene regulation. Biochemical experiments have shown the association of phosphatidylinositol (PI), phosphatidylinositol phosphate kinase (PIPK), and DAG activities with the nuclear matrix (Payrastre et al. 1992). Multiple isoforms of PIPKs localize to the nucleoplasm and are concentrated at nuclear speckles containing mRNA-processing components (Boronenkov et al. 1998). PIP2 was also detected at these speckles, consistent with its production by PIPKs localized to these sites (Boronenkov et al. 1998). Genetic evidence has implicated nuclear phosphoinositides and their hydrolysis products in the export of mRNA via the nuclear pore complex (York et al. 1999). PIP2 is a necessary component of the pre-mRNA splicing machinery (Osborne et al. 2001). A Dictyostelium nuclear phosphatidylinositol phosphate kinase is required for developmental gene expression (Guo et al. 2001). Nuclear inositol 1,4,5-trisphosphate kinase in yeast also has a role in transcriptional control (Odom et al. 2000). Phospholipids are able to bind histones and nonhistone chromosomal associated proteins (Manzoli et al. 1977). PtdIns[3]P and PtdIns[5]P were shown to bind to the PHD fingers of the chromatin-associated protein ING2 and several other proteins (Gozani et al. 2003). PIP2 stabilizes the association of the SWI/SNF-like BAF complex with chromatin and the nuclear matrix (Zhao et al. 1998), and it binds histone H1, which leads to the inhibition of histone-H1-mediated repression on RNA polymerase II activity (Yu et al. 1998).
The Drosophila gene skittles (sktl) encodes a putative PIP5KI, which is required for cell viability and germline and bristle development; sktl mutations affect the ovary, dorsal appendage, egg, and wing (Knirr et al. 1997; Hassan et al. 1998). In this article, we show that ASH2, a trxG protein, and SKTL bind directly to each other in vitro and in vivo and that sktl mutations enhance the homeotic transformation phenotype of ash2 mutations. We also report that histone H1 hyperphosphorylation within euchromatin is dramatically increased on ash2 and sktl mutant polytene chromosomes. These results support a model in which PIP2 plays a role in maintaining transcriptionally active chromatin via histone H1 modification.
MATERIALS AND METHODS
Yeast two-hybrid screen:
Full-length ash2 was PCR amplified from cDNA LD31680 (Research Genetics, Huntsville, AL) and subcloned into the TA cloning vector (Invitrogen, San Diego). A construct with ASH2 fused to the DNA-binding domain of GAL4 was generated by subcloning a NdeI-SalI fragment into pAS1-CYH2. pAS1-CYH2:ASH2 was transformed into the yeast strain Y190 (MATa gal4 gal80 his3 trp1-901 ade2-101 ura3-52 leu2-3,-112 + URA3:GAL → lacZ, LYS2:GAL → HIS3 cyhr) and transformants were selected with SC-Trp plates. Yeast transformants containing pAS1-CYH2:ASH2 were transformed with a Drosophila third instar larval cDNA library inserted in the pACTII vector containing the activation domain of GAL4. Transformants were selected with SC-Leu, His, Trp, +3-AT plates. Clones positive for β-galactosidase activity were isolated and sequenced.
GST pull-down:
Regions of ash2 were PCR amplified from cDNA LD31680 (Research Genetics) and subcloned into the TA cloning vector (Invitrogen). Constructs encoding in-frame fusions of ASH2 regions to glutathione S-transferase (GST) were generated by subcloning EcoRI fragments into pGex-2TK (Amersham, Buckinghamshire, UK). Full-length sktl was PCR amplified from cDNA LP06742 (Research Genetics) and subcloned into the TA cloning vector (Invitrogen). 6His-SKTL was generated by subcloning a KpnI fragment into pRSETB (Invitrogen). Fusion proteins were expressed in 200-ml cultures of Escherichia coli DH5α strain and solubilized in 1.5% sarkosyl, 1 mm dithiothreitol (DTT), and 0.1 mg/ml lysozyme by sonication. Insoluble proteins were removed by centrifugation. A total of 50 μl of glutathione sepharose beads (Pharmacia) was added to the cleared supernatant to purify the GST fusion proteins. Bead-bound fusion proteins were washed three times in binding buffer (20 mm Hepes-KOH, 2.5 mm MgCl2, 10% glycerol, 1 mm phenylmethylsulfonyl fluoride, 1 mm DTT, 200 mm KCl, 0.08% NP-40). Supernatant of 6His-SKTL was added to the beads and incubated at 4° overnight. After washing the beads three times in binding buffer, SDS-PAGE sample buffer was added, and samples were incubated at 95° for 5 min. GST-ASH2 bound proteins were analyzed by SDS-PAGE and immunoblotted with antibody against 6His (Upstate Biotechnology, Lake Placid, NY).
Generation of P-element transformants expressing FLAG epitope-tagged SKTL:
sktl was PCR amplified from cDNA LP06742 (Research Genetics) and subcloned into the TA cloning vector (Invitrogen). A NotI-KpnI restriction fragment was inserted into the pUAST-FLAG vector, which produces a FLAG epitope-tagged SKTL protein under the control of GAL4. This construct was injected into yw; Dr/TMS, Sb Δ2-3 embryos.
Immunoprecipitation:
Embryos containing both upstream activator sequence (UAS)-FLAG-SKTL and tubulin-GAL4 transgenes were collected over 24-hr periods. Collected embryos were dechorionated with 50% bleach and homogenized in RIPA(−)1 buffer (50 mm Tris-HCl pH 7.5, 1% NP40, 0.1% sodium deoxycholate, 150 mm NaCl) with aprotinin. Embryo extracts were spun to pellet insoluble material. Supernatants were incubated with M5 FLAG monoclonal antibody bound to protein G-sepharose beads (Amersham) with rocking at 4° overnight. After washing the beads three times in RIPA(−)1, SDS-PAGE sample buffer was added, and the samples were incubated at 95° for 5 min. Immunoprecipitates were fractionated by SDS-PAGE and immunoblotted with antibody against ASH2 (Adamson and Shearn 1996).
Immunofluorescence of polytene chromosomes:
Third instar larvae were dissected in PBS + 0.1% Triton X-100. Salivary glands were fixed for 2 min in acetic acid fix (50% glacial acetic acid, 3.7% formaldehyde, 0.1% Triton X-100) and then washed in 45% acetic acid. Fixed salivary glands were pipetted onto poly-l-lysine-coated slides and covered with siliconized coverslips. Polytene chromosomes were spread out by tapping the coverslip with a rubber-tipped hammer and squashed with thumb pressure. Slides were dipped into liquid N2. Chromosomes were blocked in antibody dilution buffer (PBS + 1% BSA + 0.1% Triton X-100) three times for 30 min and incubated with a 1:35 dilution of primary antibody at 4° overnight. Chromosomes were washed three times for 15 min and incubated with a 1:250 dilution of fluorescence-conjugated secondary antibody at 37° for 30 min. Chromosomes were washed three times for 15 min and mounted with Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole. All washes and dilutions were performed in antibody dilution buffer. ASH2 rabbit polyclonal antibody (Adamson and Shearn 1996) was protein A column purified. M5 FLAG mouse monoclonal antibody was from Sigma (St. Louis). Histone H1 mouse monoclonal antibody (Upstate Biotechnology) was generated against nuclei from myeloid leukemia biopsy cells (clone AE-4). Phospho-histone H1 rabbit polyclonal antibody (Upstate Biotechnology) was generated against the hyperphosphorylated isoform of Tetrahymena histone H1 purified by cation-exchange HPLC. Dimethyl-histone H3 (Lys4) rabbit polyclonal antibody (Upstate Biotechnology) was generated against a BSA-conjugated synthetic peptide corresponding to amino acids 1–8 of human histone H3 in which Lys4 was dimethylated.
Immunohistochemistry of imaginal discs:
Third instar larvae were dissected, fixed, and immunostained as described previously (LaJeunesse and Shearn 1995). The DAB reaction consisting of 0.5 μg/ml solution of 3,3′ diaminobenzidine and 0.02% hydrogen peroxide resulted in an orangish-brown stain. Samples were developed simultaneously for the same amount of time and then rinsed with PBS several times. Mouse monoclonal Ultrabithorax (UBX) antibody (described by White and Wilcox 1984) was used at a dilution of 1:50.
Immunoblot of salivary glands:
Salivary glands from third instar larvae were dissected, homogenized in SDS-PAGE sample buffer, incubated at 95° for 5 min, fractionated by SDS-PAGE, and immunoblotted with histone H1 mouse monoclonal antibody (Upstate Biotechnology). The immunoblot was analyzed by enhanced chemiluminescence (Pierce, Rockford, IL). After analysis with histone H1 antibody, the blot was stripped with 0.5 n NaOH for 5 min, rinsed with water for 5 min, and then immunoblotted with tubulin antibody. Exposures of immunoblots were scanned and quantitated by densitometry.
Drosophila stocks:
Drosophila stocks were maintained at 25° on standard cornmeal, molasses, yeast, and agar food containing tegosept and proprionic acid as mold inhibitors. The wild-type strain used was Canton-S. ash21803 (ash22) is a proline-to-threonine point mutation of amino acid 297, ash2703 (ash21) is an inversion with one of the breakpoints within the ASH2 coding region, and ash2X2 (ash218) is a complete deletion of the ASH2 coding region (Adamson and Shearn 1996; M. K. Cheng and A. Shearn, unpublished results). Heteroallelic ash2 mutant larvae were generated by crossing ash2703 to ash2X2. sktlp1409 is a P element inserted into an intron of sktl, sktlΔ15 is a small deletion resulting from the excision of p1409, and sktlΔ20 is a larger deletion resulting from the excision of p1409 (Hassan et al. 1998). Heteroallelic sktl mutant larvae were generated by crossing sktlΔ15 to sktlp1409. ash1RE418 (ash14) has an early stop codon resulting in a protein containing only the first 874 amino acids (of 2144), and ash1VV183 (ash122) has an early stop codon resulting in a protein with only the first 46 amino acids (Tripoulas et al. 1996).
RESULTS
SKTL is a nuclear phosphatidylinositol 4-phosphate 5-kinase:
To identify potential components of an ASH2-containing complex, a Drosophila third instar larval cDNA library was screened for ASH2 binding partners using the yeast two-hybrid assay. A fusion of ASH2 to the GAL4 DNA-binding domain by itself did not activate transcription of a lacZ reporter gene. Two of the cDNAs that did activate transcription of the reporter gene when combined with the ASH2 fusion were sequenced and found to be identical to a portion of the known sequence of sktl. These cDNAs corresponded to amino acids 503–700 of the predicted SKTL protein. The fusion protein encoded by the sktl cDNA fragment fused to the GAL4 activation domain also did not activate transcription of the lacZ reporter by itself. The gene sktl encodes a protein that is 59% identical to a human PIP5KI and 58% identical to a mouse PIP5KI (Hassan et al. 1998). SKTL is required for cell and organism viability, as well as for cytoskeletal regulation during sensory structure development and germline development (Hassan et al. 1998). The Drosophila genome contains nine genes that are predicted to encode proteins with sequences similar to PIPKs (Adams et al. 2000). Four of these gene products, including SKTL, contain a PIP5K domain. A sequence alignment and comparison of three of these gene products (PIP5K59B, SKTL, and CG17471) is shown in Figure 1. The fourth (EG:52C10.5) does not align with the other three. Of these three, only SKTL has a nuclear localization signal (NLS) according to the PSORTII computer program for the prediction of protein localization sites in cells (http://psort.nibb.ac.jp). SKTL has a PSORTII NLS score of 0.70, while PIP5K59B has a score of −0.22 and CG17471 has a score of 0.47. For comparison, ASH2, which is known to localize to nuclei by immunohistochemistry (Adamson and Shearn 1996), has a PSORTII NLS score of 0.77.
Sequence alignment and comparison of Drosophila PIPKs. The sequences used in this alignment include PIP5K59B, SKTL, and CG17471. The alignment was performed using ClustalW and shaded using Boxshade. Identical amino acids are shown in purple. Similar amino acids are shown in pink. The PIP5K domain is underlined in purple and the NLS of SKTL is underlined in pink.
ASH2 binds directly to SKTL in vitro:
Direct binding between ASH2 and SKTL was confirmed in vitro by GST pull-down assays. A series of bacterially expressed fusion proteins containing either full-length ASH2 or regions of ASH2 fused to GST were generated, partially purified, and incubated with a bacterially expressed full-length SKTL tagged with hexahistidine (6His-SKTL). 6His-SKTL protein bound to both the full-length ASH2 fusion (Figure 2B, lane 3) and the fusion containing amino acids 260–404 of ASH2 (Figure 2B, lane 6), as detected by immunoblotting with 6His antibody. 6-His-SKTL did not bind to GST alone (Figure 2B, lane 2) or with the other ASH2 fragments fused to GST (Figure 2B, lanes 4, 5, and 7).
GST fusion constructs and pull-down assay. ASH2 binds to SKTL in vitro. (A) Schematic of GST-ASH2 fusion constructs. All constructs consist of glutathione-S-transferase (gray box) fused to the N terminus of a region of ASH2. The full-length ASH2 construct contains amino acids 5–557 of ASH2, GST-ASH2.N contains amino acids 1–130; GST-ASH2.NM contains amino acids 125–270; GST-ASH2.MC contains amino acids 260–404; and GST-ASH2.C contains amino acids 402–557. PEST domain, checkered box; PHD domain, oval; NLS, dark gray box; SPRY domain, striped box. Arrow shows approximate location of ash21803 mutation (P297T). (B) GST pull-down assay. Equal amounts of GST and GST-ASH2 constructs were purified and bound to glutathione sepharose beads. The beads were incubated with bacterial supernatant expressing 6His-SKTL. The beads were pelleted, fractionated by SDS-PAGE, and probed with antibody to 6His. A total of 10% of the 6His-SKTL supernatant was loaded in the input lane (I).
ASH2 physically associates with SKTL in vivo:
The yeast two-hybrid and GST pull-down results indicate that ASH2 and SKTL can bind directly to each other in yeast and in vitro. To address the issue of whether they are associated with each other in vivo, co-immunoprecipitation experiments from embryo extracts were performed. To perform these experiments, we constructed a transgenic line containing a UAS promoter fused to FLAG epitope-tagged SKTL (FLAG-SKTL). This construct is able to rescue the lethality of sktl mutants when expressed ubiquitously under the control of tubulin-GAL4 (data not shown). This construct leads to accumulation of FLAG-SKTL protein when expressed ubiquitously under the control of a heat-shock protein 70 promoter fused to GAL4 (hsGAL4) if larvae are heat-shocked at 37° for 1 hr and allowed to recover at room temperature for 0.5 hr (data not shown). When FLAG-SKTL-expressing embryo extracts were incubated with the M5 FLAG monoclonal antibody bound to protein G-sepharose, ASH2 was immunoprecipitated, as detected by immunoblotting with ASH2 polyclonal antibody (Figure 3, lane 3). Two ASH2 proteins of 48 and 94 kD are detected in embryos (Figure 3, lane 1). It is of note that only the smaller ASH2 protein was immunoprecipitated. ASH2 was not immunoprecipitated when the identical procedure was performed with wild-type embryo extracts, which contain endogenous SKTL but no FLAG-SKTL (Figure 3, lane 2). ASH2 was also not detected after FLAG-SKTL-containing extracts were incubated with protein G-sepharose only (data not shown).
Co-immunoprecipitation of ASH2 from embryo extracts. An extract of either Canton-S wild-type (wt) or transgenic embryos expressing FLAG-tagged SKTL (FLAG-SKTL) under the control of tubulin-GAL4 was incubated with M5 FLAG monoclonal antibody bound to protein G-sepharose beads. Input lane (I) shows 10% volume of wt extract used as starting material for immunoprecipitation; the two ASH2 bands are marked with an asterisk (*). An immunoblot of the immunoprecipitates probed with ASH2 antibody is shown.
SKTL does not bind to ash2 mutant chromosomes:
The localization of ASH2 on polytene chromosomes from wild-type larval salivary glands is essential for function. This was shown by examining the chromosomal localization of ASH2 in temperature-sensitive ash2 mutants. Mutant proteins accumulate at equivalent levels when temperature-sensitive ash2 mutants are raised at either permissive or restrictive temperatures. However, mutant ASH2 protein localizes only to polytene chromosomes in larvae raised at permissive temperature (M. K. Cheng and A. Shearn, unpublished results). Immunofluorescence of ASH2 reveals a widespread pattern of accumulation along wild-type chromosome arms (Figure 4A). The chromosome-wide distribution of ASH2 is not reduced and may even be increased on sktl mutant (sktlΔ15/sktlp1409) chromosomes (Figure 4B), despite the fact that the sktl mutant chromosomes have somewhat abnormal morphology. The appearance of the mutant chromosomes is probably due to the fact that the salivary glands were dissected from mutant larvae near their lethal phase. Alternatively, the abnormal morphology may be a real sktl mutant phenotype. When third instar larvae with the genotype yw[UAS-Flag-sktl]/+;hsGAL4/+ were heat-shocked at 37° for 1 hr and allowed to recover at room temperature for 0.5 hr, FLAG-SKTL accumulated in the nucleolus (data not shown) and on polytene chromosomes in a diffuse and speckled manner (Figure 4C). No FLAG signal was detected in yw[UAS-Flag-sktl]/+;hsGAL4/+ larvae that were not heat-shocked (data not shown). However, when ash2 mutant larvae (yw[UAS-Flag-sktl]/+;hsGAL4/+;ash2703/ash2X2) were heat-shocked, there was a complete absence of FLAG-SKTL accumulation on the ash2 mutant chromosomes (Figure 4D). These results suggest that although ASH2 does not require SKTL protein to bind to chromosomes, SKTL does require ASH2 protein and/or function to bind to chromosomes.
Requirement of ASH2 for SKTL to bind chromosomes. (A) ASH2 localization on Canton-S wild-type salivary gland polytene chromosomes. (B) ASH2 localization on sktlΔ15/sktlp1409 chromosomes is not affected. (C) FLAG-SKTL localization on yw[UAS-Flag-sktl]/+;hsGAL4 chromosomes. (D) FLAG-SKTL localization on yw[UAS-Flag-sktl]/+;hsGAL4/+;ash2703/ash2X2 chromosomes is greatly affected. ASH2 antibody was detected with a Texas Red-labeled secondary antibody. FLAG monoclonal antibody was detected with a FITC-labeled secondary antibody.
ash2 and sktl display intergenic noncomplementation:
Since we demonstrated that ASH2 and SKTL bind each other both in vitro and in vivo, we wanted to further investigate the functional significance of this interaction. Three kinds of genetic evidence support the idea that the binding of these proteins to each other is of functional significance. ash2 heterozygotes, but not sktl heterozygotes, show a low penetrance of homeotic transformations. sktl/+;ash2/+ double heterozygotes were scored for homeotic transformations and were found to display an increased penetrance of third-leg to second-leg transformations (Figure 5). The penetrance of this phenotype varied depending on which alleles were examined (Table 1). Another trxG mutant, ash1, displayed the same homeotic phenotype in combination with sktl mutations (Table 1).
Homeotic transformation of ash2 sktl double heterozygotes. (A) Wild-type second leg with characteristic preapical (arrow) and apical (arrowhead) bristles. (B) sktlΔ15/+;ash2X2/+ third leg showing second-leg transformation by presence of ectopic preapical bristle (arrow). (C) Wild-type third leg with characteristic absence of the preapical and apical bristles.
Intergenic noncomplementation among ash1, ash2, and sktl mutations
In addition to the intergenic noncomplementation seen in the double heterozygotes, larvae heterozygous for sktl and trans-heterozygous for ash2 (sktlΔ15/+;ash2703/ash2X2) were found to have delayed development and died at an earlier stage than ash2 trans-heterozygotes (data not shown). Also, the haltere and third-leg discs of sktl trans-heterozygotes (sktlΔ15/sktlp1409) have decreased accumulation of UBX (Figure 6), similar to the decreased UBX accumulation seen in the haltere and third-leg discs of ash2 trans-heterozygotes (LaJeunesse and Shearn 1995).
Reduced UBX accumulation in sktl mutants. (A) Wild-type third instar haltere (H) and third-leg (L) discs. UBX accumulates in the haltere and third-leg discs. (B) sktlΔ15/sktlp1409 third instar imaginal haltere and third-leg discs have reduced UBX accumulation. UBX antibody was detected with an HRP-conjugated mouse secondary antibody and developed with a DAB reaction. All samples shown were dissected, immunostained, developed, and photographed simultaneously.
Histone H1 hyperphosphorylation is increased in ash2 and sktl mutants:
PIP2, the product of PIPK activity, has been shown to bind histone H1 in vitro, leading to transcriptional derepression (Yu et al. 1998). Also, protein kinase C, which is activated by the PIP2 metabolites DAG and PIP3, phosphorylates histone H1, which leads to chromatin decondensation. Because of these findings that implicate a relationship between PIP2 and its metabolites and histone H1, we wanted to determine if there is any change in histone H1 chromosomal localization or phosphorylation in ash2 and/or sktl mutants. Immunofluorescence reveals a widespread pattern of histone H1 accumulation along wild-type chromosome arms (Figure 7A). This pattern of accumulation is unchanged on chromosomes from ash2703/ash2X2 (Figure 7B) and sktlΔ15/sktlp1409 (Figure 7C). However, the appearance of the fluorescence is reproducibly altered. Immunofluorescence reveals accumulation of hyperphosphorylated histone H1 at a limited number of bands along wild-type chromosome arms and extensive accumulation in the nucleolus (Figure 7D). Both ash2703/ash2X2 (Figure 7E) and sktlΔ15/sktlp1409 (Figure 7F) chromosomes show a dramatic increase in hyperphosphorylated histone H1. Many more bands are present on the mutant chromosomes, in addition to the strong nucleolar signal. It is likely that this extensive hyperphosphorylation of histone H1 in the mutants is responsible for the altered appearance of immunofluorescence with antihistone H1. Immunoblot analysis of wild-type and sktlΔ15/sktlp1409 third instar larval salivary glands (Figure 8) confirmed the immunofluorescence results. An antihistone H1 immunoblot (Figure 8, top) shows similar levels of accumulation of total histone H1 in sktlΔ15/sktlp1409 larvae as compared to wild type. The same blot was probed with antitubulin as a control and to correct for the amount of protein loaded (Figure 8, bottom). The amount of protein in the hyperphosphorylated and nonhyperphosphorylated histone H1 bands from the wild-type sample together equals the amount in the one hyperphosphorylated histone H1 band from sktl mutant salivary glands when corrected for amount of protein loaded. The striking result is that essentially all of the histone H1 in sktl mutant salivary glands is hyperphosphorylated. These results demonstrating that lack of ASH2 or SKTL leads to an increase in histone H1 hyperphosphorylation imply that the normal functions of ASH2 and SKTL involve the downregulation of histone H1 hyperphosphorylation.
Histone H1 localization on ash2 and sktl mutant chromosomes. Accumulation of histone H1 was detected by immunofluorescence with FITC-labeled secondary antibody (green). Wild-type salivary gland polytene chromosomes (A) have a different appearance than ash2703/ash2X2 (B) and sktlΔ15/sktlp1409 (C) chromosomes. Accumulation of hyperphosphorylated histone H1 was detected by immunofluorescence with a Texas Red-labeled secondary antibody (red). Wild-type chromosomes (D) have dramatically less accumulation of hyperphosphorylated histone H1 than ash2703/ash2X2 (E) and sktlΔ15/sktlp1409 (F) chromosomes have. All images were taken under identical conditions.
Histone H1 accumulation in sktl salivary glands. Salivary glands from Canton-S wild-type (wt) and sktlΔ15/sktlp1409 (sktl−) third instar larvae were dissected, analyzed by SDS-PAGE, and immunoblotted with histone H1 and tubulin antibodies. (Top) Antihistone H1 immunoblot. Histone H1 antibody recognizes both the phosphorylated (32 kD) and nonphosphorylated (31 kD) forms of histone H1. Nonphosphorylated histone H1 is completely absent in sktl mutant salivary glands. (Bottom) Same blot probed with tubulin antibody as a loading control.
DISCUSSION
ASH2 physically associates with SKTL:
SKTL was identified in our yeast two-hybrid screen as a protein that binds to ASH2. The direct physical association between ASH2 and SKTL, as indicated by the yeast two-hybrid assay, was confirmed in vitro by GST pull-down (Figure 2B). Full-length ASH2 (lane 3) and ASH2.MC (lane 6), which contains the entire SPRY domain of ASH2, were sufficient to pull down SKTL. The SPRY domain was named after the two proteins that have this domain, yeast SPla and ryanodine receptor (Ponting et al. 1997). Although the function of the SPRY domain is unknown, our GST pull-down results suggest that one of its possible functions is mediating protein-protein interactions.
The physical association between ASH2 and SKTL, as indicated by the yeast two-hybrid assay and GST pull-down, also occurs in vivo in Drosophila embryos as shown by co-immunoprecipitation (Figure 3). Antibody against the FLAG epitope was able to immunoprecipitate ASH2 from embryos expressing FLAG-SKTL. This result shows that ASH2 and SKTL are physically associated in a complex. In embryos, the ASH2 antibody recognizes two proteins of different sizes, one ∼48 kD and one ∼94 kD (Figure 3, lane 1). At other developmental stages (larval to adult), the 48-kD protein is also present but the larger ASH2 protein is ∼65 kD (M. K. Cheng and A. Shearn, unpublished results). The 48- and 65-kD proteins are the sizes expected from the translation of two ash2 transcripts of 1.4 and 2 kb, respectively, which are detectable by RNA blotting (Beltran et al. 2003). The 94-kD protein in embryos may result from post-translational modification of the 65-kD protein. Despite the fact that the 94-kD protein is more abundant, only the 48-kD protein is immunoprecipitated by SKTL from embryonic nuclear extracts. Assuming that the 48-kD protein is translated from the 1.4-kb transcript, it would lack the PEST sequence and PHD finger of the full-length ASH2 protein, but would still contain the NLS-BP and SPRY domains. The SPRY domain was found in GST pull-down experiments to be sufficient to bind to SKTL in vitro. The 94-kD protein presumably would also contain the SPRY domain, yet it is not immunoprecipitated by SKTL. Perhaps its SPRY domain is modified in some way to prevent physical association with SKTL. ASH2 antibody was not able to immunoprecipitate FLAG-SKTL from embryos expressing FLAG-SKTL (data not shown). This may be because the ASH2 antibody recognizes the same site on ASH2 that mediates binding with SKTL. The 94-kD ASH2 protein was found to associate with a 500-kD multimeric protein complex (Papoulas et al. 1998). SKTL could be a component of this complex. However, it is more likely that SKTL is in a distinct complex with the 48-kD ASH2 protein, because SKTL seems to immunoprecitate only with the 48-kD ASH2 and not with the 94-kD ASH2 found in the 500-kD complex. In S. pombe, the ASH2 homolog is present in two distinct complexes (Roguev et al. 2003). One complex contains Set1 and the other contains a homolog of Drosophila LID (Gildea et al. 2000).
Both ASH2 and SKTL accumulate on polytene chromosomes (Figure 4) and in the nucleolus (data not shown). These results suggest that their physical association and functions involve chromatin and, perhaps, ribosomal DNA transcription. ASH2 accumulates normally on sktl mutant chromosomes while SKTL does not accumulate on ash2 mutant chromosomes (Figure 4). There are two possible explanations for this result. Either SKTL is not made in ash2 mutants or SKTL requires ASH2 protein and/or function to bind to polytene chromosomes. We have shown that SKTL accumulates to normal levels in ash2 mutants that express FLAG-SKTL by immunoblot analysis of embryos and larvae (data not shown), so SKTL needs ASH2 protein and/or function to bind to chromosomes. The ash2 mutant combination we used to represent the null mutant state is a trans-heterozygote of an inversion, ash2703, and a deletion, ash2X2; it does not accumulate either normal-sized ash2 transcript (Adamson and Shearn 1996) or ASH2 protein (data not shown). Since this mutant lacks ASH2 protein, it necessarily lacks ASH2 function; we cannot distinguish with certainty whether SKTL requires ASH2 protein or function to bind to chromosomes. However, since SKTL binds directly to ASH2, we favor the idea that SKTL requires the ASH2 protein to bind to chromosomes.
It was also found that ASH2 does not accumulate on ash1 mutant chromosomes, but ASH1 still accumulates on ash2 mutant chromosomes (M. K. Cheng and A. Shearn, unpublished results). This suggests that ASH2 needs ASH1 to bind chromosomes. We predict that SKTL will not be able to bind to ash1 mutant chromosomes, since SKTL requires ASH2 and ASH2 requires ASH1 to bind to chromosomes. This could also explain why we see intergenic noncomplementation between ash1 and sktl mutations (Table 1).
SKTL is a nuclear Drosophila PIP5K:
Since SKTL was the only Drosophila PIP5K found to have a NLS (Figure 1) and was shown to accumulate in nuclei (Figure 4), we think that SKTL serves a function distinct from those of other Drosophila PIP5Ks. The cytoplasmic PIP5Ks may be able to bind ASH2 in vitro, because they contain protein domains similar to SKTL that ASH2 recognizes. However, since ASH2 and the cytoplasmic PIP5Ks are not localized to the same cellular compartment, we do not expect the other Drosophila PIP5Ks to have a functional physical association with ASH2. The PHD finger of ING2 was shown to function as a nuclear phosphoinositide receptor (Gozani et al. 2003), so it is attractive to speculate that the PHD finger of ASH2 can also bind phosphoinositides that can be processed into PIP2 via its interaction with SKTL. It is likely that other members of the nuclear phosphoinositide signaling pathway will have functional interactions with ASH2. As examples, if a sktl transcription factor were mutated, then less SKTL would be made or if genes that encoded enzymes that generated SKTL substrates such as PIP or metabolites such as IP3 were mutated, then fewer of these metabolites would be generated. ash2 mutants might be expected to display intergenic noncomplementation with these mutants as well.
Functional significance of ASH2-SKTL physical association:
Several lines of evidence show that the physical association between ASH2 and SKTL is functionally significant. First of all, there is the complete loss of SKTL accumulation on ash2 mutant chromosomes (Figure 4), suggesting that SKTL requires ASH2 to bind to chromosomes.
The intergenic noncomplementation between sktl and ash2 mutants also shows that the physical association between the two gene products has functional significance. ASH1 and ASH2, like other trxG proteins, play a role in maintaining transcription activation. Reduced UBX accumulation in sktl mutants (Figure 6), as well as the intergenic noncomplementation seen with sktl mutants in combination with ash1 or ash2 (Table 1), suggests a similar role for SKTL in transcription regulation. The PIP2 generated in the nucleus by SKTL activity could be hydrolyzed to DAG and IP3 (Irvine 2000). IP3 can be further phosphorylated to IP4 and IP5, both of which have been shown to activate transcription (Shen et al. 2003; Steger et al. 2003).
Another result that shows functional significance of the physical association between ASH2 and SKTL is a similar dramatic increase in histone H1 hyperphosphorylation on both ash2 and sktl mutant chromosomes compared to wild-type chromosomes (Figure 8). Histone H1 is thought to be a general repressor of transcription by RNA polymerase II (Croston et al. 1991). The presence of histone H1 affects the ability of transcription factors to interact with DNA and is associated with transcription repression, while the removal of histone H1 is associated with transcriptional activation (Laybourn and Kadonaga 1991; Bresnick et al. 1992; Juan et al. 1994; Schultz et al. 1996). Studies in mammals and Tetrahymena have found a correlation between transcriptional activation and increased histone H1 phosphorylation (Dedon et al. 1991; Bresnick et al. 1992; Dou et al. 1999). Dephosphorylated histone H1 bound to chromatin over the mouse mammary tumor virus promoter is thought to restrict chromatin remodeling and transcription factor access (Banks et al. 2001). Phosphorylation of histone H1 has also been shown to regulate ATP-dependent chromatin-remodeling enzymes (Horn et al. 2002). The effect of phosphorylation is to create a region of negative charge, which may displace histone H1 from chromatin, allowing the binding of specific regulating factors (Dou and Gorovsky 2000). Alternatively, proteins that regulate transcription may recognize the phosphorylated residues (Dou and Gorovsky 2000).
However, histone H1 hyperphosphorylation has the opposite effect and is linked to high chromatin condensation, possibly by allowing the binding of accessory factors (Roth and Allis 1992). During mitosis, histone H1 becomes hyperphosphorylated, which may facilitate the interaction with the DNA minor groove and factors involved in metaphase chromosome condensation (Hohmann 1983; Halmer and Gruss 1996). Therefore, increased histone H1 hyperphosphorylation as observed in ash2 and sktl mutants implies increased chromosome condensation and reduced transcription.
Regulation of transcription activation by the ASH2-SKTL complex:
ASH1 has been shown to be able to methylate K4 of histone H3 (Beisel et al. 2002; Byrd and Shearn 2003) and ash1 mutant chromosomes show complete loss of histone H3 K4 methylation (Byrd and Shearn 2003). This result suggests that ASH1 is required for all of the histone H3 K4 methylation that occurs in vivo. The S. cerevisiae SET1 complex, which contains two subunits that are thought to represent a bipartite functional homolog of ASH2, has also been shown to methylate K4 of histone H3 (Miller et al. 2001; Roguev et al. 2001; Nagy et al. 2002). In Drosophila, if ASH2 were also in a complex that can methylate histone H3 K4, then it would be predicted that ash2 mutant chromosomes would show a decrease in histone H3 K4 methylation. We did indeed observe a decrease in histone H3 K4 methylation on ash2 mutant chromosomes (M. K. Cheng and A. Shearn, unpublished results).
During the assembly of nucleosomes, histone acetylation regulates the binding of histone H1 and chromatin condensation (Perry and Annunziato 1989; Ridsdale et al. 1990). Displacement of histone H1 is required prior to acetylation of target genes and activation of transcription, because histone H1 inhibits histone H3 acetylation by hindering the access of histone acetyltransferases to the histone H3 tail (Herrera et al. 2000). Herrera et al. (2000) predicted that chromatin-remodeling complexes would contain components that modify the interaction of histone H1 with chromatin. ASH2 and SKTL may represent such components. Our results suggest that ASH2 and SKTL are direct binding partners that are associated in a complex. When the ASH2-SKTL complex binds to chromatin, a source of PIP2 (SKTL) is brought to the chromatin. PIP2 can bind to and displace histone H1 and/or be metabolized to IP3 and phosphorylated derivatives. The displacement of histone H1 would prevent its hyperphosphorylation and allow for chromatin decondensation, histone acetylation, and eventually, transcription activation. The presence of IP4 and IP5 would also stimulate transcription.
Acknowledgments
We thank Stephen Elledge for strains and stocks used in the yeast two-hybrid screen, Hugo Bellen for the sktl alleles, Victor Corces and his lab for use of their fluorescence microscope, Evelyn Hersperger for generating the transgenic lines, Grafton Hersperger for technical support and generating the ash2 alleles, and Bethany Simmons and Nick Bongio for setting up and maintaining certain crosses. We also thank Michelle Beaucher, Charles Na, and Kristin Byrd for their helpful comments on the manuscript. This work was supported by a grant from the National Institute of General Medical Sciences.
Footnotes
Communicating editor: J. Tamkun
- Received June 10, 2003.
- Accepted March 19, 2004.
- Genetics Society of America
