The catalytic subunit of type 1 serine/threonine protein phosphatase (PP1c) was shown to bind trithorax (TRX) in the yeast two‐hybrid system. Interaction between PP1c and TRX was confirmed in vivo by co‐immunoprecipitation from Drosophila extracts. An amino‐terminal fragment of TRX, containing a putative PP1c‐binding motif, was shown to be sufficient for binding to PP1c by in vitro glutathione S‐transferase pull‐down assays using recombinant protein and fly extracts expressing epitope tagged PP1c. Disruption of the PP1c‐binding motif abolished binding, indicating that this motif is necessary for interaction with PP1. On polytene chromosomes, PP1c is found at many discrete bands, which are widely distributed along the chromosomes. Many of the sites that stain strongly for PP1c correspond to sites of TRX, consistent with a physical association of PP1c with chromatin‐bound TRX. Homeotic transformations of haltere to wing in flies mutant for trx are dominantly suppressed by PP1c mutants, indicating that PP1c not only binds TRX, but is a physiologically relevant regulator of TRX function in vivo.
In Drosophila, the trithorax group (TrxG) of epigenetic regulators is required for the maintenance of normal expression of homeotic genes of the Bithorax (BX‐C) and Antennapedia (ANT‐C) complexes, as well as other genes, after their initial activation. By contrast, the Polycomb group (PcG) is required for the maintenance of genes in an appropriate inactive state. The consequence of these two activities is ‘transcriptional memory’, by which a cell ‘remembers’ what type of cell it is meant to be, after its initial specification (reviewed by Farkas et al., 2000; Francis & Kingston, 2001; Simon & Tamkun, 2002). Disruption of this transcriptional memory can cause a remarkable respecification of segmental identity, leading to the transformation of one body segment into a different one. Mutations in trithorax (trx), the archetypal member of the TrxG family, for instance, result in phenotypes resembling those of loss‐of‐function mutations in multiple homeotic genes (Ingham & Whittle, 1980; Breen, 1999). The TrxG/PcG maintenance system is conserved between flies and mammals, implying an ancient origin and an essential function. The human homologue of trithorax (TRX), ALL‐1/HRX/MLL, is affected by over 30 different chromosomal fusions that lead to leukaemia (van Lohuizen, 1999), indicating that transcriptional control by TrxG genes is also crucial for normal development in humans.
Recently, much progress has been made in identifying TrxG proteins and their mode of action. TRX itself is a component of a histone acetylation complex (trithorax acetylation complex; TAC1), which acetylates core histones in nucleosomes and contains TRX, Drosophila CREB‐binding protein (dCBP) and SET binding factor 1 (Sbf1) (Petruk et al., 2001). Several other TrxG proteins have been directly linked to transcriptional activation, and several large protein complexes containing one or more TrxG proteins have been identified. For example, TRX also interacts with ASH1 (Rozovskaia et al., 1999), a component of a distinct TrxG complex (Papoulas et al., 1998) which might include dCBP (Bantignies et al., 2000). TRX also binds SNR1, a component of the Brahma (BRM, Drosophila SWI/SNF) chromatin remodelling complex (Rozenblatt‐Rosen et al., 1998). This indicates that different TRX complexes might cooperate in changing chromatin structure by a combination of different mechanisms of nucleosome modification. However, little is known about the regulation of these complexes, which must be differentially recruited and used in different cells; this specification must then be ‘remembered’ in different lineages through DNA replication, cell division and differentiation. We show here that TRX both interacts genetically with and binds directly to the catalytic subunit of type 1 serine/threonine protein phosphatase (PP1c), implicating reversible phosphorylation in the control of TAC1 and transcriptional memory.
Isolation of TRX from a screen for PP1c‐binding proteins
A yeast two‐hybrid interaction screen was used to identify potential PP1c‐binding proteins by using the PP1α87B catalytic subunit as ‘bait’. Twenty‐five independent complementary DNAs from 16 independent genes, which interacted with PP1α87B but not with control baits, were isolated from a Drosophila third‐instar larval library (Alphey et al., 1997). One of these cDNAs corresponded to trx. A subsequent screen for PP1β9C‐binding proteins (Bennett et al., 1999) identified 236 positive clones, of which 8 were derived from trx. Further two‐hybrid analysis of a representative trx clone showed that it was capable of binding all four Drosophila PP1c isoforms (PP1β9C, PP1α87B, PP1α13C and PP1α96A; data not shown).
trx is predicted to encode two protein isoforms (of 3,358 and 3,726 amino acids) that differ only at the amino terminus (Breen & Harte, 1991; Mazo et al., 1990). We sequenced the ends of all the trx two‐hybrid positives; all shared the region between residues 379 and 479 of the longer isoform (between residues 11 and 111 of the smaller isoform). This suggests that residues 379–479 contain all the sequences necessary for binding to PP1c in the two‐hybrid system. This region contains a consensus PP1c‐binding motif (KTVTF, residues 431–435) (Egloff et al., 1997), which is found in both protein variants (Fig. 1A).
PP1 binds TRX in vitro and in vivo
The interaction of PP1α87B with TRX was confirmed in vitro through glutathione S‐transferase (GST) pull‐down assays. For these binding experiments the shortest interacting fragment of TRX was expressed as a GST fusion protein in Escherichia coli. GST–TRX[302–479] fusion protein interacted efficiently with haemagglutinin (HA)‐tagged PP1α87B from crude Drosophila extracts (Fig. 1B), whereas cells without the construct or cells expressing GST alone did not bind (Fig. 1B), demonstrating that residues 302–479 of TRX are sufficient for binding to PP1. To examine the importance of the putative PP1c‐binding motif for interaction with PP1, a mutant form of TRX[302–479] in which Phe 435 was replaced with alanine (TRX[302–479]FA), was tested for its ability to bind PP1α87B in our pull‐down assay. TRX[302–479]FA failed to bind PP1, indicating that Phe 435 is crucial for interaction with PP1. To examine whether PP1 and TRX interact in vivo we precipitated endogenous TRX from embryonic nuclear extracts expressing HA‐tagged PP1α87B and examined precipitates by blotting with anti‐HA antibodies. We found that HA–PP1α87B clearly precipitated with TRX (Fig. 1C), indicating that PP1 is indeed found complexed to TRX in vivo.
PP1 and TRX co‐localize on polytene chromosomes
To explore further the association between PP1 and TRX in vivo, we examined the distribution of the two proteins on polytene chromosomes of the larval salivary gland. Immunostaining experiments have shown that TRX associates with multiple sites on polytene chromosomes. The exact number of loci varies depending on the antibody used to detect TRX (Kuzin et al., 1994; Chinwalla et al., 1995). Using the antibody described by Chinwalla et al. (1995) (a gift from P. Harte), we observed approximately 60 TRX sites (Fig. 2), as reported previously. A large proportion of PP1 is found in the nucleus and is reported to associate with chromatin (Bollen & Beullens, 2002); however, the distribution of PP1 on chromosomes is not known. Anti‐HA staining of chromosomes from flies expressing HA–PP1α87B showed that HA–PP1α87B is localized at many discrete sites, which are widely dispersed along the chromosomes, although the intensity of signals observed at different sites varied significantly (Fig. 2). Double staining with anti‐HA and anti‐TRX antibodies showed that virtually all sites staining for TRX also stain strongly for PP1, implying that PP1 is a component of the majority of TRX complexes (Fig. 2). Similar results were obtained with different TRX antibodies (a gift from A. Mazo, data not shown).
PP1 antagonizes trx function in Drosophila
Flies mutant for trx show transformations of segmental identity, in which posterior identity is transformed to a more anterior identity, similar to the transformations caused by loss‐of‐function mutations in homeotic genes (Breen, 1999). We investigated the functional interaction between TRX and PP1 by examining the effect of PP1 levels on homeotic transformations exhibited by trx mutants. Heteroallelic combinations of amorphic trx mutations trxE2 or trxB11 with trx1, a viable hypomorphic mutation that produces reduced levels of TRX (Chinwalla et al., 1995, a gift from P. Harte), result in the partial transformation of wing to haltere structures and of pigmented to non‐pigmented abdominal segments. Although the penetrance of abdominal transformations in these transallelic combinations is variable, the penetrance of wing to haltere transformations is reproducible and can be classified into three categories: no transformation, weak transformation and strong transformation (see Fig. 3 and Table 1). About 50–55% of trx1/trxE2 and trx1/trxB11 flies show strong transformation, 35% show weak transformation and 10–15% show no transformation (Fig. 3, Table 1).
To investigate the role of PP1c we examined the frequency of transformations in trx1/trxE2 and trx1/trxB11 flies that were also heterozygous for mutations in PP1α87B. PP1α87B is the major PP1c isoform in Drosophila, contributing about 80% of the total PP1 activity (Dombrádi et al., 1990). For these experiments we used two different mutant alleles of PP1α87B: PP1α87B87Bg‐6, a protein‐null mutation, and PP1α87B1, a point mutation with reduced PP1 activity. Both PP1α87B87Bg‐6 and PP1α87B1 dominantly suppressed trx‐dependent homeotic transformations: there were more than twice the number of trx flies with no visible transformations when one copy of PP1α87B was mutated (Fig. 3, Table 1). We also examined the effect of PP1α87B overexpression on the frequency of homeotic transformations by using the Gal4–UAS system (Parker et al., 2001). Modest overexpression of UAS PP1α87B by using arm‐GAL4 in either a wild‐type or trx mutant background had no phenotypic effect, indicating that PP1 levels are not limiting for TRX:PP1 function.
Our studies showing genetic interactions between PP1 and TRX implicate reversible phosphorylation, regulated by PP1 and one or more unknown kinases, in the control of TRX function and transcriptional memory. We have shown that PP1 binds TRX in vitro and in vivo and that the two proteins co‐localize on polytene chromosomes. We therefore propose that PP1 is part of one or more TRX chromatin‐associated complexes, in which PP1 exerts an effect on TRX‐dependent transcriptional control. One such complex might be TAC1, which is reported to contain about 90% of total TRX protein (Petruk et al., 2001). However, PP1 might also be part of other complexes, containing dCBP or SNF1 for instance. PP1 is found at other discrete sites that do not contain TRX, indicating that there are other PP1c‐binding proteins that can recruit PP1 to different chromosomal regions.
The exact role of TRX‐bound PP1 is not yet known. TRX, which contains four serine‐rich domains, could itself be a direct target of PP1. However, it is not known whether TRX is a phosphoprotein, and because the exact role of TRX is unclear it is difficult to predict what the effect of phosphorylation might be. It is possible that TRX is an anchor for other chromatin‐associated proteins and that it targets PP1 to some other member of this complex, such as dCBP. Mammalian CBP has been found to be phosphorylated in both quiescent and dividing cells (Yaciuk & Moran, 1991), but surprisingly few studies have addressed the role of phosphorylation in controlling CBP activity. Phosphorylation of the carboxy terminus of CBP has been reported to repress its histone acetylation activity (Ait‐Si‐Ali et al., 1998; Impey et al., 2002), whereas phosphorylation at the N terminus by protein kinase C (at Ser 89) has the opposite effect (Yuan & Gambee, 2000). Further studies will need to examine the phosphorylation state of dCBP in TRX complexes, but one possibility is that PP1 dephosphorylates CBP to allow patterns of transcription to be reprogrammed.
A role for phosphorylation in the control of TAC1 activity is implied by the presence of Sbf1 in the TAC1 complex. Sbf1 belongs to a subset of myotubularin‐related dual‐specificity phosphatases that are catalytically inactive and have been proposed to act as ‘anti‐phosphatases’ by binding to and protecting substrates from dephosphorylation (Laporte et al., 2001). Myotubularin phosphatases seem primarily to dephosphorylate phosphatidylinositol 3‐OH monophosphatase but can also dephosphorylate phosphoserine and phosphothreonine residues (Laporte et al., 2001), indicating that Sbf1 might protect against the effects of PP1. Identification of targets of Sbf1 might therefore help to identify relevant PP1 substrates.
Yeast two‐hybrid screen.
GST pull‐down assays.
The shortest fragment of trx (nucleotides 1745–2275) that bound PP1 in the yeast two‐hybrid system was subcloned from pACT as a Bgl II fragment into the Bam HI site of pGEX‐5X‐1 in frame with an N‐terminal GST tag to generate TRX[302–479]. TRX[302–479]FA was made by site‐directed mutagenesis based on the polymerase chain reaction. Bacterial cell lysates expressing GST‐tagged TRX[302–479] or TRX[302–479]FA were incubated with either arm‐GAL4 UAS‐HA–PP1α87B flies or wild‐type (Oregon R) flies and GST‐labelled protein was precipitated with glutathione–Sepharose 4B resin (Amersham Biosciences). Precipitates were examined by SDS–PAGE followed by immunoblotting with anti‐HA 12CA5 (Roche Diagnostics) and anti‐GST (a gift from I. Farkas) antibodies.
Immunoprecipitation from Drosophila extracts.
Immunoprecipitation from 2–18‐h‐old arm‐GAL4 UAS‐HA–PP1α87B and Oregon R Drosophila embryonic nuclear extracts was performed as described by Rozenblatt‐Rosen et al. (1998), with minor modifications. Nuclear extracts were prepared from ∼200 μl of dechorionated embryos of each genotype and incubated for 2 h with anti‐TRX antibodies (Chinwalla et al., 1995) diluted 1:40, followed by protein G bound to GammaBind Plus Sepharose (Amersham Biosciences) for 2 h. The presence of HA–PP1α87B in precipitates was tested by immunoblotting with anti‐HA 12CA5 antibodies (Roche Diagnostics).
The salivary glands of arm‐GAL4 UAS‐HA–PP1α87B third‐instar larvae raised at 25 °C were dissected as described by Kennison (2000). Immunostaining of polytene chromosomes was performed as described by Zink & Paro (1989), with minor modifications. Double labelling of polytene chromosomes was performed with rabbit anti‐TRX antibody (Chinwalla et al., 1995) at 1:50 dilution and anti‐HA 12CA5 antibody (Roche Diagnostics) diluted 1:100. Identical results were obtained with anti‐TRX antibody (a gift from A. Mazo) at 1:20 dilution. Primary antibodies were detected with preadsorbed Cy3‐conjugated anti‐rabbit and fluorescein‐isothiocyanate‐conjugated anti‐mouse antibodies (Jackson ImmunoResearch) at 1:500 and 1: 200 dilution, respectively.
Fly strains and genetic crosses.
Flies were raised at 25 °C on standard agar–cornmeal–molasses medium. Mutant alleles of PP1α87B and trithorax are described in FlyBase (http://www.flybase.org). The line arm‐GAL4 UAS HA–PP1α87B is essentially identical to arm‐GAL4 UAS HA–PP1β9C (Parker et al., 2001) but with the PP1α87B coding region in place of that of PP1β9C. trx1 PP1α87B87Bg‐6/TM6B and trx1 PP1α87B1/TM6B lines were generated by recombination and crossed to trxB11/TM6B and trxE2/TM6B flies. Penetrance and expressivity of haltere transformation phenotypes were quantified in the progeny: trx1 PP1α87B87Bg‐6/trxB11, trx1 PP1α87B1/trxB11, trx1 PP1α87B87Bg‐6/trxE2 and trx1 PP1α87B1/trxE2 lines compared with trx1/trxB11 and trx1/trxE2 controls. trxB11 UAS HA–PP1α87B/TM6B and trxE2 UAS HA–PP1α87B lines were generated by recombination and crossed to arm‐GAL4/arm‐GAL4; trx1/TM6B flies. The penetrance and expressivity of haltere transformation phenotypes were quantified for arm‐GAL4/+; trxB11 UAS HA–PP1α87B/trx1 and arm‐GAL4/+; trxE2 UAS HA–PP1α87B/trx1 flies compared with trxB11/trx1 and trxE2/trx1 controls.
- Copyright © 2003 European Molecular Biology Organization