|Home | About | Journals | Submit | Contact Us | Français|
In mammals, MYST family histone acetyltransferase MOF plays important roles in transcription activation by acetylating histone H4 on K16, a prevalent mark associated with chromatin decondensation, and transcription factor p53 on K120, which is important for activation of pro-apoptotic genes. However, little is known about MOF regulation in higher eukaryotes. Here, we report that the acetyltransferase activity of MOF is tightly regulated in two different but evolutionarily conserved complexes, MSL and MOF-MSL1v1. Importantly, we demonstrate that while the two MOF complexes have indistinguishable activity on histone H4 K16, they differ dramatically in acetylating non-histone substrate p53. We further demonstrate that MOF-MSL1v1 is specifically required for optimal transcription activation of p53 target genes both in vitro and in vivo. Our results support a model that these two MOF complexes regulate distinct stages of transcription activation in cooperation with other histone modifying activities.
Plasticity in transcription regulation can be achieved by dynamic regulation of histone modifications, particularly on the amino terminal histone tails (Strahl and Allis, 2000). It is well established that transcriptionally active, euchromatic regions of the eukaryotic genomes are marked by hyper-acetylation of all four core histones, while gene-poor, transcriptionally inactive heterochromatin regions exhibit hypoacetylation (Vaquero et al., 2003). Among the lysines that can be acetylated, acetylated H4 K16 is known to play a pivotal role in determining the potential of coding DNA for expression. In support, (i) H4 K16 mutation has specific transcription consequences independent of the mutational state of the other lysines on the H4 tail (Dion et al., 2005), (ii) H4 K16 is the only residue whose acetylation can, on its own, prevent silencing of the mating type genes in yeast (Johnson et al., 1990; Megee et al., 1990), (iii) studies of telomeric silencing have shown that lysine 16-specific acetyltransferase and deacetylase activities determine the boundaries of silenced chromatin (Kimura et al., 2002; Suka et al., 2002), (iv) introduction of H4 K16 acetylation inhibits the compaction of 30 nm chromatin fiber to a greater degree than deletion of the H4 N-terminal tail (Robinson et al., 2008; Shogren-Knaak et al., 2006), and (v) tethering H4 K16 acetyltransferase to a promoter via a DNA-binding domain leads to transcription activation, suggesting that a rather local acetylation may lead to a significant stimulation of transcription (Akhtar and Becker, 2000). In addition, accumulating evidence suggest that H4 K16 acetylation may be the founding acetylation event on histone H4 (Smith et al., 2003; Turner et al., 1992; Zhang et al., 2002). Recently, loss of acetylation at lysine 16 of histone H4 has been identified as a common hallmark of human cancer (Fraga et al., 2005).
H4 K16 acetylation in higher eukaryotes is mainly carried out by histone acetyltransferase (HAT) MOF (male on the first), also called MYST1 or KAT8 (Dou et al., 2005). MOF is one of the MYST (MOZ, YBF2, SAS2, and Tip60) family HATs (Avvakumov and Cote, 2007; Lucchesi, 1998). It was originally described as an essential component of the × chromosome dosage compensation Male Specific Lethal (MSL) complex in Drosophila (Hilfiker et al., 1997), causing a two-fold increase in the expression of X-linked genes in male flies. The function of MOF in dosage compensation is mediated by its acetyltransferase activity, which is tightly regulated by the MSL proteins (Bone et al., 1994; Hilfiker et al., 1997). In vitro biochemical studies revealed that MOF protein alone does not acetylate nucleosomal H4 (Morales et al., 2004). Instead, the specific and efficient acetylation by MOF depends on its interaction with the coiled-coil domain containing protein MSL1 and the chromo-domain containing protein MSL3 (Morales et al., 2004). Both components are highly conserved among eukaryotes. It has been shown that MSL1 functions to increase the HAT activity of MOF and to bridge the interaction between MOF and MSL3. MSL3 plays an important role in increasing the substrate specificity of MOF, directing its activity to nucleosomal H4 (Morales et al., 2004). Other components (i.e. MSL2, MLE and RNA components roX1 and roX2) are required to specifically recruit MOF to a distinct collection of chromatin entry sites on the Drosophila male × chromosome (Bashaw and Baker, 1997; Kelley et al., 1999; Stuckenholz et al., 1999).
Comparing to studies in Drosophila dosage compensation, the functions of MOF in mammals are less well characterized. In mammals, MOF is ubiquitously expressed and is clearly targeted to all chromosomes. Loss of MOF gene in mice causes peri-implantation lethality as a result of massive disruptions of chromatin architecture in a wide range of cells (Gupta et al., 2008; Thomas et al., 2008). In addition to maintain normal chromatin structure, MOF is important for ATM-dependent cell cycle checkpoint control (Gupta et al., 2005) as well as in transcription activation of Hox genes in coordination with the H3 K4 methyltransferase MLL (Dou et al., 2005). MOF was also reported to acetylate non-histone substrate p53 at K120, which occurs rapidly after DNA damage and promotes gene-specific responses (Murray-Zmijewski et al., 2008; Sykes et al., 2006; Tang et al., 2006; Tang et al., 2008).
Several studies suggest that MOF exists in multiple complexes in mammalian cells (Dou et al., 2005; Gupta et al., 2005; Mendjan et al., 2006; Pardo et al., 2002; Smith et al., 2005). In addition to the MSL complex, which is the highly conserved counterpart to the Drosophila MSL complex, MOF was reported to interact with several proteins including WDR5, a key component of the MLL family H3 K4 methyltransferase complexes (Dou et al., 2006; Dou et al., 2005). It was further demonstrated that MOF forms a stable complex with MLL and supports p53-dependent transcription activation in vitro (Dou et al., 2005). Comparing the MLL-MOF complex we previously purified through Flag-WDR5 with MOF interacting proteins purified by other approaches (Smith et al., 2005), six common proteins were identified including MSL1v1 (previously labeled as LOC284058 or KIAA1267), a homolog of MSL1. The MLL-MOF complex does not contain the components essential for the nucleosomal activity of the MSL complex (i.e. MSL1 or MSL3), which raises the questions of how MOF activity is regulated and what are the functions of this new MOF complex in mammalian cells.
To answer these questions, we set out to biochemically reconstitute the MOF activities for the mammalian MSL and MLL-MOF complexes. We found that MSL1v1, the only MSL homolog present in the MLL-MOF complex, is sufficient for regulating MOF acetyltransferase activity on nucleosomes. We decided to focus on MSL1v1 and its role in regulating MOF activity in the new complex. Given the emphasis of this study, we refer to the new MOF complex as MOF-MSL1v1 as opposite to the MSL complex. Interestingly, we found that the activity of the MOF-MSL1v1 complex is significantly different from that of the MSL complex. While the two MOF complexes have indistinguishable activity on histone H4 K16, they differ dramatically in acetylating non-histone substrates. We demonstrate that the MOF-MSL1v1 complex is exclusively required for acetylating transcription factor p53 and for the optimal transcription activation of p53 target genes both in vitro and in vivo. Functional distinction of these two MOF complexes is further supported by their different binding patterns along the transcribed genes as well as their functional and/or physical interactions with different histone methyltransferases.
In our previous studies, we showed that MOF co-purified with WDR5 was able to acetylate nucleosomal H4 K16, an activity indistinguishable from that of the MSL complex (Dou et al., 2005). In order to examine how this activity is regulated in the new context, we first confirmed that MOF is indeed in a distinct complex different from the MSL complex. To this end, we transfected 293T cells with Myc-MSL1v1, a protein co-purified with MOF and WDR5, and performed immunoaffinity purification using beads conjugated with anti-Myc antibody. As shown in Figure 1A, MOF and WDR5, but not the MSL1, 2 or 3 proteins, were immunoprecipitated by Myc-MSL1v1 purification. Furthermore, size-fractionation of HeLa nuclear extract also demonstrated that endogenous MOF-MSL1v1 and MOF-MSL are distinct complexes. As shown in Figure 1B, MSL1v1 resided in a complex of ~1.6 KDa and co-migrated with the H3 K4 methyltransferase MLL, consistent with our previous result (Dou et al., 2005). In contrast, MSL1 resided in a distinct complex of ~600 KD (Figure 1B). Notably, MOF protein was evenly distributed among the MSL1v1 and the MSL1 fractions, suggesting that MOF-MSL1v1 and the MSL complex are equally abundant in HeLa cells. We also observed MOF in lower molecular weight fractions corresponding to the free form (Figure 1B). Together, these results strongly argue that the MOF-MSL1v1 and the MSL complexes are two distinct MOF complexes in mammals.
Comparing proteins in the MOF-MSL1v1 complex with those of the MSL complex, we found that MSL1v1, an evolutionarily conserved homolog of MSL1, is the only component sharing sequence homology to the MSL proteins. Comparison of the sequences of MSL1v1 (1105 amino acid) and MSL1 (614 amino acid) shows that these two proteins share a conserved N-terminal coiled-coil domain and a C-terminal PEHE domain, which interacts with MOF (Marin, 2003; Smith et al., 2005) (Figure 1C). The coiled-coil domain of MSL1 interacts with the ring-finger domain of MSL2 (Li et al., 2005). The function of the coiled-coil domain in MSL1v1 remains unclear. In contrast to MSL1, MSL1v1 does not have the MSL3 interacting domain at the C-terminus (Smith et al., 2005). Instead, it has a much larger and structurally undefined region proceeding to the PEHE domain (Figure 1C).
Given the essential roles of MSL1 and MSL3 in the Drosophila MSL complex for regulating MOF nucleosomal activity, we decided to test whether MSL1v1 is the equivalent in regulating MOF activity in the new complex. To this end, we reconstituted and purified MOF-MSL1v1 by co-infection of insect cells with baculoviruses expressing MOF and Flag-MSL1v1 respectively (Figure 1D). As controls, we also expressed and purified MOF alone and the mammalian MSL complex using a similar approach (Figure 1D). Comparing the HAT activities of MOF alone and two MOF complexes, we found that MSL1v1 is necessary and sufficient to support K16 specific activity on nucleosomal H4 (Figure 1E). In fact, the activity of MOF-MSL1v1 on nucleosomal H4 was indistinguishable from that of the MSL complex when equivalent amount of MOF was used in the assay (Figure 1E). In contrast, MOF alone, which was purified through the same procedure, acetylates free histones but not the nucleosomal H4 (Figure 1E). Thus, the activity for the reconstituted mammalian MOF-MSL1v1 and MSL complexes share the same characteristic as the Drosophila MSL complex (Morales et al., 2004). They are functionally equivalent in terms of acetylating nucleosomal H4 in vitro. Furthermore, comparing to the activity of MOF-MSL1v1 with the holo-complex, MSL1v1 alone is sufficient to recapitulate most of the complex activity on nucleosomes (data not shown).
In addition to histone H4, tumor suppressor p53 is anther important substrate for MOF. It was reported that MOF acetylates p53 on K120 (Sykes et al., 2006). We next compared the activities of MOF alone with the reconstituted MOF-MSL1v1 and the MSL complex on recombinant p53 by in vitro HAT assay (Figure 2B). As in the case of nucleosomal substrates, MOF-MSL1v1 acetylates p53 much more efficiently than MOF alone when limited amount of enzymes were used (Figure 2B, lane 1 versus 2). The activity of MOF-MSL1v1 is specific for p53 K120 as the acetylation was abolished when K120 was mutated to alanine (Figure 2B, lane 3). Strikingly, we found that while the MSL complex can efficiently acetylate nucleosomal H4 in vitro (Figure 1E), its activity on p53 is minimal (Figure 2B, lane 5). The different activities of MOF-MSL1v1 and the MSL complexes on p53 suggest that acetylating p53 K120 is a unique function for the MOF-MSL1v1 complex.
To study the mechanism underlying the differential activity of two MOF complexes on p53, we decided to dissect MSL1v1 to identify the essential domains. To this end, we made two deletion mutants of MSL1v1: Δ429 amino acid (aa) that deletes the N-terminal coiled-coil domain and 800aa that deletes most of the N-terminus (Figure 3A). The MOF-MSL1v1 mutant complexes were purified through Flag-tagged MSL1v1 after co-expression with MOF in insect cells (Supplementary Figure S1A). Several bands corresponding to MSL1v1 Δ800aa were detected (Supplementary Figure S1A), which are due to phosphorylation (unpublished observation). The activities of these two mutant complexes were compared to that of wild-type MOF-MSL1v1. As controls, HAT assays using MOF and the MSL complex were included. Equivalent amounts of MOF were used in all reactions. As shown in Figure 3B, both Δ429aa and Δ800aa MOF-MSL1v1 complexes acetylate nucleosomal H4 with the same efficiency as the wild-type complex. However, only MOF-MSL1v1 Δ429aa is capable of acetylating p53 (Figure 3B). This result suggests that the regulation of p53 acetylation by MSL1v1 is different from that of H4 acetylation.
To further define the essential domains in MSL1v1 for p53 acetylation, we expressed and purified five MSL1v1 polypeptides containing 429-800aa, 480-982aa, 560-982aa, and 705-982aa of MSL1v1 respectively (Figure 3C and Supplementary Figure 1B). These MSL1v1 truncation mutants were tested for MOF-mediated p53 acetylation in vitro. As shown in Figure 3D, we found that the regulation of MOF activity by MSL1v1 requires direct interactions between the MSL1v1 PEHE domain and MOF. The MSL1v1 429-800aa fragment that does not contain the PEHE domain (848-982aa) failed to support p53 acetylation (Figure 3D, lane 5). In contrast, deleting the N-terminal 705aa of MSL1v1 does not affect p53 acetylation (Figure 3D). This result and results from previous experiments (Figure 3B) suggest that the region between 705aa and 800aa of MSL1v1 contains essential features that support p53 acetylation but is dispensable for nucleosomal acetylation. In this experiment, we also noticed that all MSL1v1 polypeptides were acetylated by MOF in vitro (Figure 3D, indicated by arrow). Since MSL1v1 560-982aa migrates at almost the same position as p53, we included a sample of MOF-MSL1v1 560-982aa without p53 as a control to differentiate the acetylation signals of MSL1v1 560-982aa and p53 (Figure 3D, lane 6 versus lane 3). The result showed that most of the signal came from p53 acetylation.
To understand the mechanism for the specific requirements for MOF-MSL1v1 in p53 K120 acetylation, we performed GST-pull down assays using GST-p53 and purified Flag-MSL1v1 705-982, Flag-MSL1v1 Δ800 or Flag-MSL1 proteins respectively. We found that GST-p53, but not GST alone, efficiently bound to MSL1v1 705-982. In contrast, MSL1v1 Δ800 or MSL1 did not interact with p53 under the same condition (Figure 3E). It is likely that MSL1v1 stimulates MOF activity on p53 by, at least in part, stabilizing its interaction with the substrate.
To test the function of the p53-specific HAT activity of the MOF-MSL1v1 complex in transcription regulation, we turned to the established in vitro transcription system (Figure 4) (Dou et al., 2005). As described previously, the DNA template used for the in vitro transcription assay contains p53 consensus sites upstream of the major late promoter that drives the expression of a G-less cassette (An et al., 2004). In this assay, the transcription activation is strictly p53-dependent, since no activity was detected in the absence of p53 (Figure 4B, lane 1 versus lane 3). Using this assay, we found that p53 K120 is essential for p53 transcription activity in vitro. Mutating p53 K120 to alanine led to a significant reduction of transcription activity (Figure 4B, lane 2). Furthermore, acetylation of p53 K120 by the MOF-MSL1v1 complex greatly enhanced the transcription activity of p53 (Figure 4C). This activation was a direct result of HAT activity of the MOF-MSL1v1 complex on p53 K120 since adding the MOF-MSL1v1 complex without acetyl-CoA or adding acetyl-CoA without the MOF-MSL1v1 complex had no effect (Figure 4C). Given that MOF-MSL1v1, but not the MSL complex, acetylated p53 K120 in the in vitro HAT assays, we next tested whether the transcription activation is specific for MOF-MSL1v1 (Figure 4D). We found that the MSL complex had minimal effect on p53 dependent transcription activation in vitro using the DNA template based assay (Figure 4D, left panel). Consistently, MOF-MSL1v1 705-982, which is capable of acetylating p53 K120, activates transcription as efficiently as the full length MSL1v1 while MOF-MSL1v1 Δ800, which does not acetylate p53 K120, had minimal effect (Figure 4D, right panel). Together, these results suggest that MOF-MSL1v1 can greatly enhance p53-dependent transcription activation through targeted acetylation at p53 K120.
As shown by in vitro HAT assays, MSL1v1 plays an essential role for regulating MOF activity on both nucleosomal H4 and tumor suppressor p53. We also demonstrated that MOF-MSL1v1 mediated p53 acetylation enhanced the p53 transcription activity on DNA template in vitro. In order to study the distinct functions of both MOF complexes in vivo, we performed shRNA or siRNA mediated knockdown experiments for MSL1v1 and MSL1 in 293T cells (Figure 5A). As controls, we also performed a knockdown experiment for MOF (Figure 5A). Consistent with previous reports (Smith et al., 2005), knocking down MSL1, the key component of the MSL complex, leads to global reduction of H4 K16 acetylation (Figure 5B). More importantly, we found that knocking down MSL1v1 also leads to global reduction of H4 K16 acetylation in 293T cells (Figure 5B). Immunoblot using anti-H3 antibody was included as a loading control (Figure 5A and 5B). This result suggests that MOF-MSL1v1 is equally important as the MSL complex for acetylating H4 K16 in vivo. The reduction of H4 K16 acetylation in MSL1v1 knockdown cells was not likely a result of reduced expression of MOF or MSL1 as the protein levels of MOF and MSL1 were the same in the MSL1v1 knockdown cells (data not shown). In addition to H4 K16 acetylation, we also tested the global p53 K120 acetylation in all the knockdown samples. Since the anti-p53 K120ac antibody cannot be used for immunoblot (Sykes et al., 2006), we first used anti-p53 K120ac antibody to enrich the acetylated p53, which was then detected by immunoblot using an anti-p53 antibody. As shown in Figure 5C, knocking down MOF and MSL1v1 significantly reduced global p53 K120 acetylation. In contrast, MSL1 knockdown has much less effect (Figure 5C). The background of p53 K120 acetylation in the MOF or MSL1v1 knockdown samples is probably due to acetylation by endogenous Tip60, another p53 K120 specific acetyltransferase in vivo (Sykes et al., 2006; Tang et al., 2006).
Since p53 K120 acetylation by MOF leads to increased expression of pro-apoptotic genes such as PUMA and BAX (Sykes et al., 2006), we next test whether MOF-MSL1v1 regulates p53 K120 acetylation-mediated gene activation in vivo. We decide to first study the function of MOF-MSL1v1 using an over-expression approach. Specifically, p53 null human lung carcinoma H1299 cells were transfected with p53, MOF and MSL1v1 480-982aa and examined for p53-mediated expression of PUMA and BAX. H1299 cells were previously used to demonstrate the functions of p53 K120 acetylation in vivo (Sykes et al., 2006; Tang et al., 2006). MSL1v1 480-982aa, which is fully active on both histone and non-histone substrates in vitro, was used instead of full length MSL1v1 as it was expressed at a similar level as MSL1 and MSL3 in H1299 cells (Supplementary Figure S2). Despite the high basal transcription activity for p53 alone, which is probably due to the endogenous MOF and Tip60, we found a specific increase of PUMA and BAX expression in cells over-expressing MSL1v1 480-982aa together with p53 and MOF. In contrast, over-expressing MSL1/3 has no such effect (Figure 6A, left and middle panels). The expression of p21, which is not regulated by K120 acetylation (Sykes et al., 2006), does not show differential regulation by MSL1v1 (Figure 6A, right panel). Of note, the basal expression of all p53 target genes is higher in cells transfected with wild-type p53 than with p53 K120A (Figure 6A). Similar reduction of gene expression in cells expressing p53 K120R mutant was previously reported (Tang et al., 2006). This reduction is also consistent with our previous results that transcription activity of p53 K120A mutant is much lower than that of wild-type p53 in vitro (Figure 4B).
To further confirm that the PUMA and BAX activation depends on MOF-MSL1v1 mediated p53 K120 acetylation, we set up a control experiment using MSL1v1 Δ800aa to co-transfected H1299 cells with p53, MOF. As shown in the in vitro HAT assay, MSL1v1 Δ800aa acetylates H4 K16 but not p53 K120 (Figure 3B). Consistently, we found that MSL1v1 Δ800aa did not further stimulate p53-mediated expression of PUMA and BAX compared to p53 alone or to the MSL complex (Figure 6B). Together, these results strongly argue that while both MOF complexes activate p53 targets by promoting H4 K16 acetylation, the optimal activation of PUMA and BAX requires MSL1v1 mediated p53 K120 acetylation.
In addition to over-expression studies, we also knocked down MOF, MSL1v1 or MSL1 in 293T cells and tested the expression of BAX, PUMA and p21. Consistently, moderate reduction of BAX and PUMA expression was observed in cells knocked down for MOF and MSL1v1 but not for MSL1 (Figure 6C).
To demonstrate that MOF-MSL1v1 activates PUMA and BAX expression directly through p53 K120 acetylation, chromatin immunoprecipitation (ChIP) assays using the antibody against acetylated p53 K120 were performed at transcription start sites of PUMA, BAX and p21 genes (Figure 6D and 6E). Consistent with the in vitro HAT assays, p53 K120 acetylation at the promoters of PUMA and BAX was much higher in cells co-transfected with MOF and MSL1v1 in comparison to cells co-transfected with either p53 alone or with MOF and MSL1/3 (Figure 6D, top). In contrast, p53 K120 acetylation at the p21 promoter was extremely low and no MSL1v1 dependent increase in p53 K120 acetylation was observed (Figure 6D, top). Transfection of p53 K120A mutant was included to demonstrate the specificity of the p53 K120ac antibody (Figure 6D, top). Consistently, MOF and MSL1v1 dependent decrease in p53 K120 acetylation was observed in cells knocked down for either protein (Figure 6E, top). Contrary to p53 K120 acetylation, p53-dependent H4 K16 acetylation dramatically increased at the promoters of PUMA and BAX genes upon over expressing p53, MOF and either MSL1v1 or MSL1/3 (Figure 6D, bottom). Similarly, knocking down MOF, MSL1v1, or MSL1 resulted in the same reduction of H4 K16 acetylation at the gene loci of BAX and PUMA (Figure 6E, bottom). This is consistent with the global reduction of H4 K16 acetylation observed in these cells (Figure 5B). In addition to examine the modifications, we also tested the direct binding of MSL1v1 and MSL1 along the genes of BAX and PUMA upon over-expression. Four sets of primers were used as indicated for distal promoters, transcription start sites, 3′ transcribed region near stop codon and downstream genomic loci respectively. Comparing to MSL1, MSL1v1 showed relatively higher binding at the transcription start sites (Figure 7A), in line with its functional interaction with transcription factor p53 as well as its association with the H3 K4 histone methyltransferase MLL (Dou et al., 2005). In contrast, MSL1 showed enrichment at 3′ UTR of PUMA (Figure 7A, bottom), consistent with higher H4 K16 acetylation observed at this locus (Figure 6D, bottom). As a negative control, ChIP assays for p53 K120ac, H4 K16ac and both MSL proteins were performed at HPRT genomic locus, which is p53-independent (Wu et al., 2005). No signals were detected for p53 K120 acetylation and the binding of either MSL protein (Figure 6D, 6E and and7A7A).
Several studies provide early indications that human MOF associates with more than one complex (Dou et al., 2005; Gupta et al., 2005; Mendjan et al., 2006; Pardo et al., 2002; Smith et al., 2005). Our previous studies showed that MOF resided in a complex that is vastly different from the well-characterized Drosophila dosage compensation MSL complex (Dou et al., 2005). This new MOF complex co-elutes with the H3 K4 methyltransferase MLL in a superose 6 column. The peak fractions for MSL1v1 are distinct from MSL1, a key component of the evolutionarily conserved MSL complex (Figure 1B). By biochemical reconstitution, we demonstrate that MSL1v1 is necessary and sufficient to regulate MOF activity on nucleosomes (Figure 1E). This is in contrast to the MSL complex, where specific and efficient acetylation for nucleosomal H4 requires both MSL1 and MSL3 (Morale et al., 2004). In addition to fulfill the function of both MSL1 and MSL3, MSL1v1 is also capable of stimulating MOF activity towards non-histone substrates (Figure 2), significantly expanding the substrate spectrum for MOF. Intriguingly, the essential domain of MSL1v1 for acetylating H4 and p53 were mapped to different regions of the protein. While deletion of the N-terminal 800 amino acids of MSL1v1 does not affect the acetylation of nucleosomal H4, it completely abolishes the acetylation of p53 (Figure 3B). This result suggests that different mechanisms may be involved in regulating MOF activity on different substrates. Sequence alignment of MSL1v1 and MSL1 shows that although PEHE domain (amino acids 848-982) of MSL1v1 is conserved in MSL1, the region between 705aa and 800aa of MSL1v1, which is highly conserved among MSL1v across species, varies dramatically from MSL1 (Smith et al., 2005). Although the exact mechanism by which MSL1v1 modulates the MOF activity remains to be decided, it is likely that this region of MSL1v1 helps to stabilize the interaction of MOF with its non-histone substrate and thus facilitates the retention of non-histone substrates at the MOF catalytic site. In support of this notion, we found that MSL1v1 705-982, but not MSL1v1 Δ800 or MSL1, is able to interact with p53 (Figure 3E). To our knowledge, MOF is quite unique as a histone acetyltransferase to have distinct activities towards histone versus non-histone substrates depending on which complex it resides in.
Different substrate spectrums of MOF-MSL1v1 and the MSL complexes are indicative of functional specifications of these two MOF complexes. Indeed, this was clearly demonstrated using a defined DNA template based in vitro transcription assay (Figure 4). Using this assay, we showed a specific requirement for MOF-MSL1v1 in p53-dependent transcription activation. This function of MOF-MSL1v1 was based on its unique (compared to MOF-MSL) ability to acetylate p53 K120. Neither MOF alone nor the MOF-MSL complex was able to stimulate p53-dependent transcription activation from the DNA template. Consistent with the in vitro assays, the MOF-MSL1v1-specific transcription stimulation was also observed at several p53 target genes in vivo (Figure 6A). ChIP analyses of p53 K120 acetylation demonstrated that it correlated well with the expression of MSL1v1 and with the transcription activation. In contrast, similar changes of H4 K16 acetylation were observed in both over-expression and knockdown studies for MSL1v1 and MSL1. Thus, although MOF-MSL1v1 acetylates both H4 K16 and p53 K120 in vitro (Figure 3B) and in vivo (Figure 6D), it is likely that MOF-MSL1v1-mediated p53 K120 acetylation plays a predominant role for p53 target gene (e.g. PUMA and BAX) activation in vivo. Future mechanistic studies are needed to further dissect the contribution of p53 K120 and H4 K16 acetylation to p53 dependent transcription activation. Interestingly, consistent with its interaction with p53, MSL1v1 is relatively enriched at the gene promoters by ChIP assays (Figure 7A). In contrast, while the binding of MSL1 is consistently lower than that of MSL1v1 at the promoters, it binds at relatively higher level at the 3′ end of PUMA gene (Figure 7A). This result is reminiscent of a recent report mapping the binding sites of MSL3 in the Drosophila genome (Kind et al., 2008). It was proposed that MSL3 binds preferably at the coding region as well as the 3′UTR of transcribed genes.
The difference in the ability of two MOF complexes to interact with p53 supports distinct recruitment mechanisms for the two MOF complexes to targets. Given that MOF-MSL1v1 stably associates with WDR5 (Dou et al., 2005, Smith et al., 2005), a key component of the MLL family H3 K4 methyltransferase (Dou et al., 2006), it is likely that the MOF-MSL1v1 complex functions in coordination with the H3 K4 methyltransferases at the gene promoters. We envision that MOF-MSL1v1 and MLL, interfaced by WDR5, facilitate the recruitments of each other and promote transcription initiation at target genes (Figure 7B). Once recruited, MOF-MSL1v1 plays two functions at the gene promoter: acetylating transcription factor p53 and acetylating histone H4 K16 to establish open chromatin structures (Figure 7B). This model is supported by our previous observation that both MLL and MOF are required for optimal p53-dependent transcription activation in vitro (Dou et al., 2005). In contrast, the MSL complex is likely to be recruited by a transcription factor-independent mechanism that involves SET2-mediated H3K36 tri-methylation (Larschan et al., 2007; Strahl et al., 2002; Sural et al., 2008), a histone mark enriched at the coding region as well as 3′ UTR of genes (Figure 7B) (Barski et al., 2007). In Drosophila, it was shown that mutations in either histone methyltransferase SET2 or in the chromo-domain of MSL3, which is essential for binding to tri-methylated H3K36, greatly compromise the recruitment of the MSL complex to chromatin (Bell et al., 2007; Larschan et al., 2007; Sural et al., 2008). Given the conservation of the MSL complex in Drosophila and mammals, it is likely that the mammalian MSL complex is recruited to transcribed genes through similar mechanisms. Once recruited, the MSL complex can acetylate H4 K16 and function coordinately with SET2 to play roles at later stages of transcription activation (Figure 7B). Future mechanistic analysis for the function of two MOF complexes in transcription activation bears significance not only for elucidating the roles of MOF in the process but also for studying the functional coordination of multiple histone modifying enzymes in gene regulation.
The finding that MOF-MSL1v1 is specifically required for acetylating transcription factor p53 makes it highly likely that MOF has a much broader substrate spectrum than previously described (i.e. H4 K16 and p53 K120). Indeed, we found that MOF-MSL1v1 is also able to acetylate several other previously uncharacterized substrates for MOF (unpublished observation). Given the importance of MOF-mediated p53 K120 acetylation in transcription activation, it is likely that MOF is involved in many important biological processes through acetylating non-histone substrates. As more is learned about the substrates and target genes for the MOF complexes, we anticipate and look forward to a better understanding of this important acetyltransferase in higher eukaryotes.
Baculovirus expression vectors were created by insertion of cDNAs into pFastBac HTB or pFastBac1 vectors (Bacvector 3000, Novagen): Flag-MSL3, FLAG-MSL1v1 wild-type and serial deletion mutants were cloned into EcoRI and XhoI sites; MOF, was cloned into BamHI and XhoI sites of pFastBac HTB; HA-MSL1 and MSL2 were cloned into XhoI and HindIII sites. Baculoviruses were generated by following the manufacturer’s protocols. Reconstitution was achieved by co-infecting Sf9 cells with different baculoviruses of similar titers and proteins were harvested 60 hour post infection. For mammalian expression vectors, p53 and p53 K120A were cloned into EcoRV and NotI sites of pIRES-neo (Clontech); Flag-MOF, Myc-MSL1v1, Myc-MSL1v1 Δ800aa, Myc-MSL1 and MSL3 were cloned into BamHI and XhoI sites in pcDNA3 vector (Invitrogen). Bacterial expression vectors for p53 and p53 K120A were previously described (Dou et al., 2005).
Nuclear extracts were obtained from HeLa cells by a modified Dignam procedure (Dignam et al., 1983). For fractionation, 500μl nuclear extract was loaded onto a superose-6 size exclusion column and 0.5ml fractions were collected. The fractions were subject to immunoblot.
The whole cell lysate was incubated with M2 agarose for Flag-MOF, Flag-MSL1v1 (wild type and deletion mutants) or Flag-MSL3 in BC500, 0.05% (v/v) NP40 at 4 °C for 4 hours and extensively washed with BC500, 0.05% (v/v) NP40. The bound proteins were eluted with 0.25 mg/ml FLAG peptide in BC100. Eluted complexes were resolved by SDS-PAGE and visualized by Coomassie stain or by immunoblot.
Antibodies were obtained commercially as follows: p53 (Santa Cruz, DO1) and p53 K120ac (Abcam); M2 and M2 agarose (Sigma); H4 K16ac antibody (Millipore); Myc antibody clone 4A6 (Millipore); and HA (Sigma). Anti-MOF and Anti-MSL1v1 antibody were raised against the full-length recombinant proteins.
Recombinant histone octamers or HeLa nucleosomes were prepared as described (Dou et al., 2005). For each HMT assay, 2 μg of nucleosomes or recombinant histone octamers were used. For all reactions, equivalent amount of MOF protein (less than 100ng) was used in the form of MOF alone, MOF-MSL1v1 or MOF-MSL1-3. Reactions were carried out at 30 °C for 1 hour in the presence of [3H]-acetyl-CoA.
For GST pull-down assays, 10 μg GST-tagged protein and 10 μg purified recombinant MSL1v1 705-982aa, MSL1v1 Δ800aa or MSL1 were used in each binding assay. Reactions were carried out in BC300, 0.05% (v/v) NP40 with 10ug BSA at 4 °C for 4 hours.
The assays were performed as described previously (Dou et al., 2005). Briefly, 40 ng p53 and about 100 ng purified MOF, MOF-MSL1v1 or MOF-MSL complexes were used for each reaction. The complexes were normalized to the same amount of MOF protein. 32P-CTP was used to detect the RNA transcripts.
H1299 or 293T cells were transfected with siRNA duplexes (200 pmol, Dharmacon) or shRNA plasmids using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Cells were harvested 24 hours after transfection for RT-PCR or 48 hours after transfection for western blot. The sequences for small interfering RNAs (siRNAs) used in the assays are available upon request.
Cells were fixed with 1% (w/v) paraformaldehyde for 10min at 37 °C. Chromatin immunoprecipitation was performed using the Chromatin Immunoprecipitation Assay Kit (Millipore) and protocols recommended by the manufacturer. ChIP DNA was quantified by real-Time PCR as outlined (Milne et al., 2002). Real-time quantitation of PUMA, BAX and p21 transcripts in RT-PCR experiments was normalized against the control 5s-rRNA expression. Primers for ChIP and Q-PCR experiments are available upon request. In all assays, the Monte-Carlo method was used to calculate the statistic significance (see supplementary material).
We are grateful to Drs. E. Smith and J. Lucchesi for cDNAs for MSL1, 2 and 3 as well as anti-MSL1 and anti-MOF antibodies. We are grateful to Drs. M. Bedford and Y. Shi for reagents. We are grateful to Dr. Y. Liu for critical reading of the manuscript. This work was supported by NIH grant (R01GM082856) and by the John S. and Suzanne C. Munn Cancer Research Fund of University of Michigan Comprehensive Cancer Center to Y.D.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.