Drosophila ptip is essential for anterior/posterior patterning in development and interacts with the PcG and trxG pathways

The establishment and maintenance of gene expression patterns in development is regulated in part at the level of chromatin modification through the concerted actions of the Polycomb and trithorax family of genes(PcG/trxG). In Drosophila, Polycomb and Trithorax response elements(PRE/TREs) are cis-acting DNA sequences that bind to Trithorax or Polycomb protein complexes and maintain active or silent states, presumably in a heritable manner (Grimaud et al.,2006; Ringrose and Paro,2007). In mammalian cells however, such PRE/TREs have not been conclusively identified. Polycomb and Trithorax gene products function by methylating specific histone lysine residues, yet how these complexes recognize individual loci in a temporal and tissue specific manner during development is unclear. Recently, we identified a novel protein, PTIP (also known as PAXIP1), that is part of a histone H3K4 methyltransferase complex and binds to the Pax family of DNA-binding proteins(Patel et al., 2007). PTIP is essential for assembly of the histone methyltransferase (HMT) complex at a Pax DNA-binding site. These data suggest that Pax proteins, and other similar DNA-binding proteins (Shimizu et al.,2001), can provide the locus and tissue specificity for HMT complexes during mammalian development.

In mammals, the PTIP protein is found within an HMT complex that includes the SET domain proteins ALR (GFER - Mouse Genome Informatics) and MLL3, and the accessory proteins WDR5, RBBP5 and ASH2(Cho et al., 2007; Issaeva et al., 2007; Patel et al., 2007). This PTIP containing complex can methylate lysine 4 (K4) of histone H3, a modification implicated in epigenetic activation and maintenance of gene expression patterns. Furthermore, conventional Ptip^-/- mouse embryos and conditionally inactivated Ptip^-/- neural stem cell derivatives show a marked decrease in the levels of global H3K4 methylation,suggesting that PTIP is required for some subset of H3K4 methylation events(Patel et al., 2007). The PTIP protein contains six BRCT (BRCA1 carboxy terminal) domains that can bind to phosphorylated serine residues (Manke et al., 2003). This is consistent with the observation that PAX2 is serine-phosphorylated in response to inductive signals(Cai et al., 2003; Cai et al., 2002). In mammals,PAX2 specifies a region of mesoderm fated to become urogenital epithelia at a time when the mesoderm becomes compartmentalized into axial, intermediate and lateral plate (Bouchard et al.,2002; Torres et al.,1995). These data suggest that PTIP provides a link between tissue specific DNA-binding proteins that specify cell lineages and the H3K4 methylation machinery.

To extend our finding to a non-mammalian organism and address the evolutionary conservation of Ptip, we asked whether a Drosophila ptip homolog could be identified and if so, whether it is also an essential developmental regulator and part of the epigenetic machinery. The mammalian Ptip gene encodes a novel nuclear protein with two amino-terminal and four carboxy-terminal BRCT domains, flanking a glutamine-rich sequence (Lechner et al.,2000). Based on the number and position of the BRCT domains and the glutamine-rich domain, the Drosophila genome contains a single ptip homolog. To understand the function of Drosophila ptipin development, we characterized a ptip mutant allele that contained a piggyBac transposon insertion between BRCT domains three and four. Maternal and zygotic ptip mutant embryos exhibited severe patterning defects and developmental arrest, whereas zygotic null mutants developed to the third instar larval stage but also exhibited anterior/posterior (A/P)patterning defects. In cell culture, depletion of Polycomb-mediated repression activates developmental regulatory genes, such as the homeotic gene Ultrabithorax (Ubx). This derepression is dependent on trxG activity and also requires PTIP. Microarray analyses in cell culture of Polycomb and polyhomeotic target genes indicate that many,but not all, require PTIP for activation once repression is removed. The activation of PcG target genes is coincident with PTIP binding to promoter sequences and increased H3K4 trimethylation. These data argue for a conserved role for PTIP in Trithorax-mediated epigenetic imprinting during development.

Stocks Df(3L)fz-GF3b, P{wAR}66E and Df(3L)ED4543,P{3′.RS5+ 3.3′}ED4543 were obtained from Bloomington stock center. According to Flybase, the two lines have deleted cytological segments 70C2-70D4 and 70C6-70F4, and are referred to as Df(ptip)¹ and Df(ptip)², respectively. PB^c04574 (PBac^ptip) was kindly provided by the Exelixis collection at Harvard Medical School(http://drosophila.med.harvard.edu/).

In situ hybridization of the embryos was performed as described(Cadigan et al., 1998), using digoxigenin-labeled antisense RNA probes. For clonal analysis of the PBac^ptip allele, it was recombined onto an FRT^2A chromosome using standard methods(Xu and Rubin, 1993) and the germline clones were generated using FRT^2AP[ovoD] as described (Chou and Perrimon, 1996), with heat shocks during larval development. The resulting female virgin flies with the genotype FRT^2APBac^ptip/FRT^2A P[ovoD] were then crossed to Df(ptip)¹/TM6 evelacZ males. Embryos lacking zygotic ptip were identified by the absence of evelacZ. Cuticles were prepared and photographed as previously described (Bhanot et al.,1999). Adult flies were frozen overnight and photographed with a Leica M10 microscope.

Zygotic ptip mutant wing imaginal discs and salivary glands were prepared from PBac^ptip/Df(ptip)¹ third instar larvae and immunostained as described(Guan et al., 2007; Parker et al., 2002). Rabbit anti-DLL (1:200) was from J. Kohtz (Northwestern University, Chicago, IL,USA), monoclonal anti-EN (1:20) and anti-WG (1:100) were from the Developmental Studies Hybridoma Bank (Iowa City, IA, USA), rabbit anti-Histone H3 (trimethyl K4, 1:200) and anti-Histone H3 (trimethyl K27, 1:200) were ChIP grade and purchased from Abcam (Cambridge, UK). Cy3- and Alexa Fluor 488-conjugated secondary antibodies were from Jackson Immunochemicals (West Grove, PA, USA) and Molecular Probes (Eugene, OR, USA), respectively. The discs and salivary glands were then digitally photographed with a Nikon Eclipse 800 compound microscope and the images were processed as Abode Photoshop files.

For rescue experiments shown in Fig. 6, P[UAS-mPtip] lines were constructed based on the mouse Ptip cDNA sequence (mPtip) as described(Lechner et al., 2000) and introduced into w¹¹¹⁸ hosts by standard P element-mediated transformation. Five independent lines were obtained, and the line mapping to the second chromosome was used for further studies. Flies with a combination of four transgenes were examined. P[UAS-mPtip], along with P[hs-Gal4], were crossed into a PBac^ptip/Df(ptip)¹ background, and were subjected to 37°C heat shock for 1 hour/day for 5 days after egg laying.

Kc167 (Kc) cells were routinely cultured at room temperature in Schneider's Drosophila media (Invitrogen) containing 5% FBS. RNAi-mediated gene knockdowns were performed essentially as described(Clemens et al., 2000). Briefly, cells were seeded at 10⁶ cells/well and a total of 15μg dsRNA was added. Cells were harvested on the day 7 for total RNA extraction. Templates for dsRNA synthesis were obtained by RT-PCR. The primer sequences are available upon request.

For chromatin immunoprecipitation (ChIP) experiments, Kc cells were seeded at 2×10⁶ cells/60 mm dish and control or PcG dsRNAs were added. The next day, cells were transfected with pUAS-mPtipand a Drosophila β-actin promoter VP16-GAL4 activator using FuGENE-mediated lipofection. Chromatin was prepared 72 hours post transfection as described (Patel et al.,2007). For ChIP experiments, 5 μg chromatin was incubated with antibodies against mPTIP, anti-H3K4me3 (Abcam) or rabbit IgG controls overnight at 4°C and processed as described previously. DNA isolated from the ChIP samples was analyzed with primer pairs that recognize the DNA sequences around the transcription start sites of the genes listed. All analysis was done in triplicate using real time quantitative PCR. Primer sequences (shown 5′ to 3′; F, forward; R, reverse) and position relative to the transcription start sites were as follows: Cad-F TGAGCCTCTCTGCCTGTTTATCTG (-73) and Cad-R TGAAATGTTATTCCGTGCGTGG(+57); Esp-F TGAAGAGCCAAGCGACCAGAAC (+25) and Esp-R CACTGCCCACAAAACACAAGTTG (+174); rib-F TCTCTCGCTCTCACCCTCTGTG (-93)and rib-R GTTTTCCGCTCGCCTTGTG (+102); pnr-F TCTCTCTTGCTCTTTCGCTCACC (-246) and pnr-R CTCACACACACTCACACTCGCTATC(-56); Ubx-F TGGTGCGTATCTGTGTTTTGTGC (-1107) and Ubx-R TTCTACTTTTCCAACCCCTGCTC (-988); αTub-F CAAGCAAAGATTCACGCCCTG(-117) and αTub-R CGCCGCATAACCGATAACTG (-2).

Total RNA was extracted from ∼100 μg fly embryos, larvae and adults or 5×10⁶ Kc cells using Trizol Reagent (Invitrogen). An aliquot (10 μg) of total RNA was electrophoresed in 1% agarose gel containing formaldehyde, blotted to nylon membrane and probed with ³²P-labeled DNA fragments generated by RT-PCR. The primer sequences are available upon request.

Microarray experiments were performed using Affymetrix GeneChip Drosophila Genome 2.0 Array in the UMCCC Affymetrix and cDNA Microarray Core Facility at University of Michigan, Ann Arbor. Statistical analyses were done using previous published methods(Irizarry et al., 2006; Smyth, 2004; Yuan et al., 2005). Gene expression patterns were visualized as colorgrams using TreeView(Eisen et al., 1998). The original and normalized data can be accessed at the Gene Expression Omnibus(http://www.ncbi.nlm.nih.gov/geo)accession number GSE15092.

All vertebrates have a single, conserved Ptip gene, as based on the position and number of BRCT domains and a glutamine-rich central region located between BRCT domains two and three. A BLAST search of invertebrate genomes also uncovered a ptip-related gene in sea urchins, insects and C. elegans. An evolutionary distance tree shows the relationship among ptip homologs from select invertebrates and vertebrates(Fig. 1A). Note that human ptip is more closely related to Drosophila(3×10^-60) and honey bee (5×10^-81) ptip than it is to the next most homologous mammalian BRCT domain encoding gene Mdc1 (2×10^-16). The Drosophila genome contains a single homologous gene (Flybase ID CG32133), which we believe is the fly ptip ortholog based on its markedly similar domain structure to mouse and human Ptip(Lechner et al., 2000). The Drosophila hypothetical PTIP protein contains two amino-terminal and four carboxy-terminal BRCT domains and an expanded glutamine-rich region(Fig. 1B). Northern analysis reveals that fly ptip encodes a single mRNA product of ∼7 kb(Fig. 1D), which we have confirmed by RT-PCR and sequencing (data not shown). Fly ptip is expressed ubiquitously in the embryo as judged by in situ hybridization (data not shown) and in all developmental stages shown by Northern analysis(Fig. 1D).

Three ptip alleles are used in this study. Two are deficiencies,referred to as Df(ptip)¹ and Df(ptip)², that have the entire ptip locus deleted (see Materials and methods for details). The third allele is a lethal piggyBac insertion line, called PB^c04574(PBac^ptip) from the Exelixis collection(Thibault et al., 2004). We used PCR and sequencing to confirm that the PBac^ptipinsertion mapped to the fifth exon of the ptip gene (data not shown)and noted that it is predicted to delete its three carboxy-terminal BRCT domains (Fig. 1C). The PBac^ptip allele is homozygous pre-pupal lethal and is also lethal over Df(ptip)¹ and Df(ptip)²,indicating that the disruption of ptip is responsible for the lethality.

Since ptip mRNA is expressed at particularly high levels in early embryos as well as in female adults, we suspected that it is a maternally expressed gene (Fig. 1D). To remove maternal ptip contributions, we recombined PBac^ptip onto the FRT^2A chromosome and generated germline mutant clones using ovoD for selection(Chou and Perrimon, 1992; Chou and Perrimon, 1996). Indeed, full length ptip mRNA was completely removed in early PBac^ptip germline clone embryos(Fig. 1D), thus confirming a maternal contribution of ptip. Instead of the full length ptip mRNA, these embryos express a truncated form of ptipmRNA, which is consistent with the sequence analysis and indicates a carboxy-terminal disruption of the ptip gene(Fig. 1D). We were not able to determine whether a truncated protein product is produced from this transcript but, even if this were the case, the deleted three carboxy-terminal BRCT domains would be likely to abrogate the function of PTIP in flies, since the carboxy-terminal BRCT pairs are crucial for PTIP function(Manke et al., 2003; Patel et al., 2007). Given the embryonic lethality and gene structure, we believe PBac^ptip is a strong hypomorphic allele, if not a complete null.

As described above, PBac^ptip homozygotes have a pre-pupal lethal phenotype, as do PBac^ptip/Df(ptip)¹ and PBac^ptip/Df(ptip)² transheterozygotes. To test the possible roles of maternal and zygotic ptip in fly embryo development, we examined the cuticles of embryos derived from PBac^ptip germline clones. Wild-type embryos have a distinctive patterning of denticles on their ventral cuticle, with each denticle belt arranged in a trapezoidal pattern with intermittent naked cuticle (Fig. 2A). When mothers producing PBac^ptip mutant eggs are crossed with Df(ptip)¹/TM6B heterozygotes, two classes of mutant phenotype are observed. One mutant displays severely disrupted/complete loss of denticle belts (Fig. 2B), which might indicate early embryonic lethality. In fact, the number of embryos in this class is obviously less than half of the total, but the presence of many empty egg cases indicates that some embryos might die even before they begin to secret cuticle. This phenotypic class was not observed when fathers were wild type, indicating that they are maternal and zygotic ptip mutants. By sharp contrast, the other class of embryos exhibit fairly normal cuticle patterns, although frequently with denticle fusions (Fig. 2C). These embryos are maternal ptip mutants, as they are similar to what we observed when fathers are wild type (data not shown). Therefore, zygotic ptip is sufficient to rescue the cuticle defects caused by the removal of the maternal contribution of ptip. We conclude that maternal and the zygotic ptip expression are both important for normal fly embryo development.

These cuticle defects in ptip mutant flies are consistent with the hypothesis that ptip plays a role in early patterning. To characterize the germline null phenotype more precisely, we examined the expression patterns of maternal effect, pair rule and segment polarity genes(Fig. 3). Localization of maternal bicoid and oskar mRNAs appear unaffected in ptip mutant embryos (Fig. 3A,B). However, the expression of the pair rule genes fushi tarazu (ftz) and even skipped (eve), which both exhibit a characteristic seven-stripe pattern at the cellular blastoderm stage, was severely affected (Fig. 3C,D). The ptip mutants exhibited missing stripes and reduced levels of expression, suggesting failure to activate or maintain transcription from specific enhancer regions of the eve and ftz genes. Although few apparent ptip mutant embryos developed to the extended germband stage, the segment polarity gene engrailed (en) was not detected in the characteristic 14-stripe pattern in ptip mutants(Fig. 3E).

We also examined expression in transheterozygotic ptip flies[PBac^ptip/Df(ptip)¹], which lack zygotic ptip. These mutants develop further, presumably as a result of maternal rescue. Imaginal discs from late third instar larvae were examined for expression of developmental regulators(Fig. 4). The mutant third instar larvae have no difference in size and locomotion compared with wild type, however, their imaginal discs are always morphologically smaller and are distorted (Fig. 4A,B). en is normally only expressed in the posterior compartment of the wing disc and is a widely used molecular marker to examine A/P patterning in the wing disc. However, in the ptip mutant discs EN protein is expressed broadly over the entire disc(Fig. 4B), as a result of derepression of en in the anterior compartment. This altered EN expression pattern is also observed in third instar leg discs(Fig. 4C,D). The EN protein can repress the Hedgehog effector cubitus interuptus (ci)through a PcG-mediated mechanism (Dahmann and Basler, 2000; Maschat et al., 1998). Consistent with the ectopic expression of en,the expression of ci is lost in the anterior half of ptipmutants (Fig. 4G,H). By sharp contrast, wingless (wg), a known signaling determinant of the dorsal/ventral (D/V) axis (Campbell et al., 1993; Galindo et al.,2002), appears to be expressed within normal boundaries(Fig. 4E,F), as is the wg target, Distal-less (Dll). These data suggest that in the absence of ptip there is a disruption of A/P patterning in the imaginal disc, although PcG-mediated repression appears intact.

Given the association of PTIP with a histone H3K4 methyltransferase complex, we examined the levels of H3K4 trimethylation in the nuclei of salivary glands isolated from third instar larvae(Fig. 5A-L). Antibodies against trimethyl histone H3K4 (H3K4me3) exhibit strong nuclear staining in wild-type nuclei but show a marked reduction in staining intensity in ptipmutants. The reduction in staining intensity observed with H3K4me3 antibodies in ptip mutants correlates with an increase in the staining intensity using anti-trimethyl histone H3K27 (H3K27me3) antibodies. To more precisely quantitate these differences, we acid-extracted total histones from wild-type and zygotic ptip mutant larvae and probed western blots with methyl-lysine specific antibodies (Fig. 5M). Image analysis revealed a ∼50% decrease in total levels of trimethyl H3K4 when normalized to total histone H3, whereas levels of trimethyl H3K27 increased more than 80%. These data are consistent with a role for Drosophila ptip as a cofactor for trxG-mediated histone methylation.

Although the structure and number of BRCT domains is similar between the Drosophila and mammalian PTIP proteins, a transgenic rescue approach would be a better test of functional equivalence(Fig. 6). Thus, flies were generated that carried the mouse Ptip cDNA (mPtip) under the control of the UAS-GAL4 promoter, such that mouse PTIP protein could be expressed after heat shock with the hs-Gal4 driver strain(Fig. 6A). The mouse transgene was crossed into the heat shock Gal4 line carrying the PBac^ptip allele to generate w;P[UAS-mPtip]; PBac^ptip/TM6Bmales. These were crossed to females w; P[hsGal4]/CyO; Df(ptip)¹/TM6B to generate P[hsGal4]/P[UAS-mPtip]; Df(ptip)¹/PBac^ptip flies, which express mPTIP, and control w;P[UAS-mPtip]/CyO; Df(ptip)¹/PBac^ptip flies without the GAL4 activator. Although there were some survivors even in the absence of heat shock, suggesting leaky expression of mPTIP, activation by heat shock resulted in viable adult flies that carried both Drosophila mutant ptip alleles (Fig. 6B,C). These rescued flies represented ∼4.1% (23/564) of offspring, close to the expected 6.25%. Notably, no adult flies were observed with the genotype P[UAS-mPtip]/CyO; Df(ptip)¹/PBac^ptip in these crosses,again confirming the lethal nature of the ptip homozygotes. Thus, the expression of the mouse Ptip cDNA was sufficient to rescue the zygotic null phenotype observed in the PBac^ptip/Df(ptip)¹ offspring.

Two lines of evidence led us to propose that ptip might be involved in the epigenetic control of gene expression by the PcG/trxG pathways. First, mammalian PTIP protein is associated with ALR and MLL3, two SET domain H3K4 methyltransferases similar to Drosophila Trithorax(Cho et al., 2007; Issaeva et al., 2007; Patel et al., 2007). Second,analyses of histone methylation in ptip mutants revealed global and reciprocal changes in H3K4 and H3K27 methylation levels. To further examine the function of ptip in Drosophila, we utilized the Kc cell line to assay for gene expression in the presence or absence of Polycomb-mediated repression and ptip using double stranded RNA(dsRNA) mediated knockdown. Since Ultrabithorax (Ubx) is a known PcG target (Castelli-Gair and Garcia-Bellido, 1990; Tillib et al., 1999), we examined whether ptip has an impact on the expression of this gene (Fig. 7). The Ubx gene is expressed at very low levels in Kc cells, as observed in other Drosophila cell lines derived from embryos (Breiling et al., 2001; Schwartz et al., 2006). No significant change in Ubx expression is observed when cells are cultured with 3′ or a mixture of 3′ and 5′ ptipdsRNAs. However, a slight increase in Deformed (Dfd) mRNA is observed in ptip knockdowns (Fig. 7A). Using microarray analyses to compare untreated Kc cells with ptip knockdowns after 7 days of culture revealed 266 genes with significant changes in expression levels(Fig. 8). Of these, 52 were upregulated in the ptip knockdowns and 214 were downregulated. The potential ptip targets corresponded to genes associated with metabolic processes, transcription, translation and signaling, with no specific pattern observed (see Table S1 in the supplementary material for the normalized dataset). The level of ptip knockdown with 3′ and 5′ dsRNAs was consistent at ∼70% as measured by mRNA Northern blotting and phosphoimage analysis. Whether this reflects heterogeneity among cells owing to variable uptake of dsRNAs, or the inability of cells to completely eliminate the ptip mRNA, could not be determined.

A different pattern emerges when ptip was knocked down together with PcG genes. Suppression of the PcG genes Polycomb (Pc)and polyhomeotic (ph) by dsRNA in Kc cells resulted in the upregulation of Ubx (Fig. 7), indicating that Ubx is a target of PcG-mediated repression, as reported previously (Wang et al., 2004). The activation of Ubx expression requires the trxG genes, since the activation of Ubx is low when trxand ash1 are knocked down along with Pc and ph(Fig. 7). These data suggest that the derepression of Ubx in Kc cells is PcG/trxG pathway-specific. Interestingly, when ptip was knocked down together with PcG genes, we found that the activation of Ubx was also decreased compared with the PcG knockdown alone. This suggests a role for ptip in trx-mediated activation of Ubx once PcG repression is removed. After relieving PcG repression, the levels of Ubx were similarly reduced when cells were cultured either with ptip or with trx and ash1 dsRNAs. The homeotic gene Dfd was also activated upon PcG repression in a trx and ash1 dependent manner. However Dfd expression was not ptip dependent, suggesting that ptip is required only in a subset of potential PcG/trxG targets. In fact, Dfd expression was slightly activated upon ptip knockdown and was highest when all three genes, Pc, ph and ptip, were knocked down together(Fig. 7A).

To determine more precisely which genes were ptip-dependent and to extend our findings that suggest ptip might be a component of the PcG/trxG pathway, we performed Affymetrix microarray analysis to compare gene expression patterns in Kc cells under normal conditions, after ptipknockdown, after PcG knockdown and after ptip plus PcG knockdown(Fig. 9; see Table S2 in the supplementary material). We obtained a list of PcG targets that are derepressed by PcG knockdown, as judged by statistically significant differences. Many of the genes are known PcG targets and are developmental regulators. However, if both ptip and PcG are knocked down,activation of many PcG target genes is significantly reduced when compared with expression levels observed in the PcG gene knockdowns alone(Fig. 9A). This is even more evident when the changes in expression levels are averaged among the top 74 PcG target genes (Fig. 9B). Table 1 outlines a select group of PcG targets and known regulators of development that are derepressed upon treatment with dsRNA against Pc and ph and require ptip for activation upon derepression. These PcG targets are generally unaffected by the ptip knockdown alone. In the same experiment, expression levels of the housekeeping genes, such as the tubulins and genes encoding RNA-associated proteins, are minimal(Fig. 9A,B). In addition, the levels of ptip are unaffected in the Pc knockdowns and the level of Pc is unaffected in the ptip knockdowns, as judged by Northern analysis (Fig. 9C). These results indicate that the ptip mutation might be a suppressor of PcG mutations and might be generally required for the trxG-dependent activation of many genes after removal of PcG repression.

Given the effects of ptip knockdowns on PcG target gene expression in Kc cells, we asked whether PTIP can interact directly with the genes whose expression requires PTIP. At present, we do not have good antibodies against Drosophila PTIP nor do we have a full length Drosophila ptipcDNA clone. Cloning of the fly cDNA has proved difficult, in part because of the extreme repetitive nature of the glutamine-rich region. Thus, we expressed mPTIP in the fly Kc cells in concert with dsRNAs against Pc and ph and used these cells for chromatin immunoprecipitation (ChIP; Fig. 10). Cells were transfected with mPTIP and cultured for 3 days with control dsRNAs or dsRNAs against Pc and ph. The mPTIP protein could be readily detected in chromatin preparations after reverse-crosslinking and western blotting (Fig. 10A). Antibodies against mPTIP and H3K4me3 were used for ChIP experiments to determine if PTIP localized to any of the PcG target genes identified by the microarrays (Table 1). Strikingly, mPTIP could localize near the transcription start sites of all of the PcG genes examined but only when cells were also cultured with PcG dsRNAs(Fig. 10B). Furthermore,trimethylation at H3K4 increased significantly after PcG dsRNA treatment(Fig. 10C). This increase in H3K4me3 was already evident after PcG dsRNA treatment, although the added expression of mPTIP increased H3K4me3 levels even further. Without PcG dsRNAs,mPTIP expression did not increase H3K4me3 levels nor did the mPTIP protein localize to the target genes. As a negative control, a tubulin promoter region did not show changes in H3K4me3 nor did it bind to mPTIP. These data strongly argue for a direct interaction between PTIP and the promoter regions of PcG target genes upon derepression.

Embryonic development requires epigenetic imprinting of active and inactive chromatin in a spatially and temporally regulated manner, such that correct gene expression patterns are established and maintained. In this report, we show that Drosophila ptip is essential for early embryonic development. We further show that in larval development, ptipcoordinately regulates the methylation of histone H3K4 and demethylation of H3K27, consistent with the reports that mammalian PTIP complexes with HMT proteins ALR and MLL3, and the histone demethylase UTX(Cho et al., 2007; Issaeva et al., 2007). In wing discs, ptip is required for appropriate A/P patterning by affecting morphogenesis determinant genes, such as en and ci. These data demonstrate in vivo that dynamic histone modifications play crucial roles in animal development and PTIP might be necessary for coherent histone coding. In addition, ptip is required for the activation of a broad array of PcG target genes in response to derepression in cultured fly cells. These data are consistent with a role for ptip in trxG-mediated activation of gene expression patterns.

Early development requires ptip for the appropriate expression of the pair rule genes eve and ftz. The characteristic seven-stripe eve expression pattern is regulated by separate enhancer sequences (Jiang et al., 1991; Small et al., 1993), which are not all equally affected by the loss of ptip. The complete absence of en expression at the extended germband stage also indicates the dramatic effect of ptip mutations on transcription. The characteristic 14 stripes of en expression depends on the correct expression of pair rule genes (DiNardo and O'Farrell, 1987), which are clearly affected in our ptipmutants. However, the maintenance of en expression at later stages and in imaginal discs is regulated by PREs and PcG proteins(Breen et al., 1995; Brizuela and Kennison, 1997; DeVido et al., 2008; Maschat et al., 1998; Moazed and O'Farrell, 1992). If ptip functions as a trxG cofactor, then expression of enalong the entire A/P axis in the imaginal discs of ptip mutants might be due to the absence of a repressor. This might explain the surprising presence of ectopic en in the anterior halves of imaginal discs from zygotic ptip mutants. This ectopic en expression is likely to result in suppression of ci through a PcG-mediated mechanism. Yet,it is not clear how en is normally repressed in the anterior half,nor which genes are responsible for derepression of en in the ptip mutant wing and leg discs.

The direct interaction of PTIP protein with developmental regulatory genes is supported by ChIP studies in cell culture. Given the structural and functional conservation of mouse and fly PTIP, we expressed mPTIP in fly cells and showed that it can localize to the 5′ regulatory regions of many PcG target genes that are activated upon loss of PC and PH activity. Consistent with the interpretation that a PTIP trxG complex is necessary for activation of repressed genes, mPTIP only bound to DNA upon loss of Pc and ph function. In the Kc cells, suppression of both Pc and ph results in the activation of many important developmental regulators, including homeotic genes. A recent report details the genome-wide binding of PcG complexes at different developmental stages in Drosophila and reveals hundreds of PREs located near transcription start sites(Oktaba et al., 2008). Strikingly, most of the genes we found activated in the Kc cells after PcG knockdown also contain PRE elements near the transcription start site.

In vertebrates, PTIP interacts with the Trithorax homologs ALR/MLL3 to promote assembly of an H3K4 methyltransferase complex. The tissue and locus specificity for assembly may be mediated by DNA-binding proteins such as PAX2(Patel et al., 2007) or SMAD2(Shimizu et al., 2001), which regulate cell fate and cell lineages in response to positional information in the embryo. In flies, recruitment of PcG or trxG complexes to specific sites also can require DNA-binding proteins such as Zeste(Dejardin and Cavalli, 2004),DSP1 (Dejardin et al., 2005),Pleiohomeotic (Brown et al.,2003; Wang et al.,2004) and Pipsqueak (Huang et al., 2002). Whereas PcG complexes have been purified and described in detail (Ringrose and Paro,2007; Schuettengruber et al.,2007; Shao et al.,1999), much less is known about the Drosophila trxG complexes. Purification of a trxG complex capable of histone acetylation(TAC1) revealed the proteins CBP and SBF1 in addition to TRX(Petruk et al., 2001; Petruk et al., 2004). By contrast, the mammalian MLL/ALL proteins are components of large multi-protein complexes capable of histone H3K4 methylation(Cho et al., 2007; Hughes et al., 2004; Nakamura et al., 2002). Although the mutant analysis, the reduction of H3K4 methylation and the dsRNA knockdowns in Kc cells all suggest that Drosophila ptip has trxG-like activity and hence might be a suppressor of PcG proteins, a more definitve biochemical analysis awaits the generation of antibodies and the delineation of in vivo DNA-binding sites for PTIP and its associated proteins at specific target genes.

Mammalian PTIP is also thought to play a role in the DNA damage response based on its ability to bind to phosphorylated p53BP1(Munoz et al., 2007). PTIP also binds preferentially to the P-SQ motif, which is a good substrate for the ATR/ATM cell cycle checkpoint regulating kinases(Manke et al., 2003). Several reports demonstrate that PTIP is part of a RAD50/p53BP1 DNA damage response complex, which can be separated from the MLL2 histone H3K4 methyltransferase complex (Cho et al., 2007; Patel et al., 2007). Both budding and fission yeast contain multiple BRCT domain proteins that are involved in the DNA damage response, including Esc4(Rouse, 2004), Crb2(Saka et al., 1997), Rad9(Hammet et al., 2007) and Cut5(Saka and Yanagida, 1993). All of these yeast proteins have mammalian counterparts. However, neither the fission nor budding yeast genomes encodes a protein with six BRCT domains and a glutamine-rich region between domains two and three, whereas such characteristic PTIP proteins are found in Drosophila, the honey bee, C. elegans and all vertebrate genomes. These comparative genome analyses suggest that ptip evolved in metazoans, consistent with an important role in development and differentiation.

In summary, Drosophila ptip is an essential gene for early embryonic development and pattern formation. Maternal ptip null embryos show early patterning defects including altered and reduced levels of pair rule gene expression prior to gastrulation. In cultured cells PTIP activity is required for the activation of Polycomb target genes upon derepression, suggesting an important role for the PTIP protein in trxG-mediated activation of developmental regulatory genes. The conservation of gene structure and function, from flies to mammals, suggests an essential epigenetic role for ptip in metazoans that has remained unchanged.

Supplementary material available online at http://dev.biologists.org/cgi/content/full/136/11/1929/DC1