The MUT9p kinase phosphorylates histone H3 threonine 3 and is necessary for heritable epigenetic silencing in Chlamydomonas

Abstract

Changes in chromatin organization are emerging as key regulators in nearly every aspect of DNA-templated metabolism in eukaryotes. Histones undergo many, largely reversible, posttranslational modifications that affect chromatin structure. Some modifications, such as trimethylation of histone H3 on Lys 4 (H3K4me3), correlate with transcriptional activation, whereas others, such as methylation of histone H3 on Lys 27 (H3K27me), are associated with silent chromatin. Posttranslational histone modifications may also be involved in the inheritance of chromatin states. Histone phosphorylation has been implicated in a variety of cellular processes but, because of the dynamic nature of this modification, its potential role in long-term gene silencing has remained relatively unexplored. We report here that a Chlamydomonas reinhardtii mutant defective in a Ser/Thr protein kinase (MUT9p), which phosphorylates histones H3 and H2A, shows deficiencies in the heritable repression of transgenes and transposons. Moreover, based on chromatin immunoprecipitation analyses, phosphorylated H3T3 (H3T3ph) and monomethylated H3K4 (H3K4me1) are inversely correlated with di/trimethylated H3K4 and associate preferentially with silenced transcription units. Conversely, the loss of those marks in mutant strains correlates with the transcriptional reactivation of transgenes and transposons. Our results suggest that H3T3ph and H3K4me1 function as reinforcing epigenetic marks for the silencing of euchromatic loci in Chlamydomonas.

The eukaryotic nuclear genome is organized as a large polymeric DNA/protein/RNA assembly known as chromatin. The basic repeating unit of this structure is the nucleosome, consisting of ≈150 bp of DNA wrapped around an octamer composed of two copies each of histones H2A, H2B, H3, and H4 (1, 2). The terminal tails of histones, as well as some residues in their globular domains, can carry a variety of posttranslational modifications such as phosphorylation, acetylation, methylation, ubiquitination, sumoylation, or biotinylation (1–4). These modifications can alter DNA–histone interactions within and between nucleosomes as well as the binding of effector molecules that determine higher-order chromatin structures (1–3). Thus, histone modifications can collaborate to influence a multitude of DNA-mediated cellular processes, including transcription, replication, recombination, DNA repair, and chromosome condensation. Certain histone posttranslational modifications may also play a role in the inheritance of chromatin states (1, 5–7).

Phosphorylation of a number of Ser or Thr histone residues has been implicated in gene activation as well as in cell cycle-dependent chromosome condensation, DNA repair, and apoptosis-induced chromatin compaction (1, 8–10). In several plant species, similarly to metazoans, histone H3 is hyperphosphorylated at Ser-10/28 during mitosis and meiosis, and these modifications appear to be required for proper chromosome segregation and cell cycle progression (9, 10, 11). H3S10 phosphorylation (H3S10ph) is also involved in the transcriptional activation of genes responding to stress or mitogen-stimulated signaling pathways in animals (10, 12), and it has been found to increase transiently in plant cells subject to abiotic stresses (9, 13). Indeed, H3S10ph facilitates RNA polymerase II release from promoter-proximal pausing in Drosophila (12) and binding of heterochromatin protein 1 (HP1), a polypeptide involved in heterochromatin formation, to H3K9me3 is disrupted by the presence of a phosphate group on Ser-10 (14, 15). However, H3S10ph has also been implicated in the marking of silent chromatin in postmitotic mammalian cells (14), and it is associated with heterochromatin in interphase cells of Arabidopsis (16). The puzzle posed by the correlation of H3S10ph with two apparently opposed chromatin states (active euchromatin and silent heterochromatin) is not clearly understood and emphasizes that the meaning of posttranslational histone modifications may depend on genomic context.

Histone phosphorylation is also involved in DNA-repair processes (1, 8, 17). One of the best characterized DNA damage-responsive chromatin modifications is the phosphorylation of the SQ motif found in the C-terminal tail of histone H2A or the H2AX variant in higher eukaryotes (17, 18). In addition, H4S1 phosphorylation seems to have an evolutionarily conserved role in chromatin compaction during the later stages of gametogenesis (1, 19). Thus, phosphorylation of specific histone residues appears to function in a number of chromosomal transactions (1, 8–10), but its potential role in epigenetic gene silencing remains uncertain. Interestingly, we have now identified a Chlamydomonas mutant, disrupted in the gene encoding a histone kinase that is defective in heritable gene silencing. This mutant shows deficiencies in both H3 and H2A phosphorylation. Moreover, our findings implicate H3T3ph as an epigenetic mark associated with repressed transgenes and transposons.

Results

The Mut-9 Mutant Is Deficient in Epigenetic Transgene Silencing.

By insertional mutagenesis of a strain (11-P[300]) containing a transcriptionally repressed copy of the RbcS2:aadA:RbcS2 transgene (conferring resistance to spectinomycin), we have isolated silencing-defective mutants in the green alga Chlamydomonas (20). One of the mutations, identified in the Mut-9 strain, reactivates the expression of both unmethylated, single-copy (Fig. 1 A and B) and partly methylated, multiple-copy transgenes [supporting information (SI) Fig. S1 A and B]. Interestingly, multiple-copy transgenes often show various degrees of enhanced cytosine methylation in the mutant background (Fig. S1C), and in some cases this increased DNA methylation appears to counteract the reactivation of expression imposed by the mut9 mutation (Fig. S1, Tetrad 12, product 2). Indeed, in the case of cis-repeated transgenes, the loss of the MUT9p repression mechanism appears to be partly compensated by an alternative silencing pathway involving DNA methylation. As previously reported (20), Mut-9 also shows derepression of several transposons and hypersensitivity to genotoxic agents causing DNA double strand breaks (Fig. 1C).

Fig. 1.

Effect of the mut9 mutation on transcriptional gene silencing, sensitivity to DNA-damaging agents, and meiotic heritability of the repressed states. (A) Growth and survival on Tris-acetate-phosphate (TAP) medium or on TAP medium containing spectinomycin of the wild-type untransformed strain (CC-124), the silenced parental strain (11-P[300]), the mutant strain (Mut-9), and two strains complemented with a MUT9 wild-type copy [Mut-9(MUT9)-20 and Mut-9(MUT9)-28]. (B) Northern blot analysis of the indicated strains. The levels of the wild-type MUT9 mRNA, the truncated transcript resulting from the insertion mutation (mut9), and the RbcS2:aadA:RbcS2 transgenic RNA (aadA) are shown. The RBCS2 transcript was used as a control for equivalent loading of the lanes. (C) Survival of 11-P[300] (filled squares), Mut-9 (filled diamonds), and Mut-9(MUT9)-20 (open triangles) on TAP medium containing increasing concentrations of bleomycin. (D) A meiotic tetrad product (Mut-9-9-1), containing exclusively the mut9 mutation, was crossed to 11-P[300], and reactivation of transgene expression was tested in the progeny by their ability to grow on spectinomycin-containing medium. The segregation of the mut9 mutation and the RbcS2:aadA:RbcS2 transgene (aadA) was assessed by Southern blotting (Lower). (E) Mut-9 was crossed to the untransformed wild-type strain of opposite mating type (CC-125), and transgene expression as well as meiotic segregation was evaluated as described above.

In Mut-9 the mutagenic plasmid integrated into intron 8 of a gene that we have named MUT9 (Fig. S2A), causing the production of a truncated MUT9 mRNA (Fig. 1B). Transformation of Mut-9 with a genomic DNA fragment encompassing MUT9 (Fig. S2 B and C) resulted in full complementation of the hypersensitivity to bleomycin (Fig. 1C) but only partial resilencing of the activated RbcS2:aadA:RbcS2 transgene (Fig. 1 A and B). Crossing the 11-P[300] strain, containing a silenced transgene, with a tetrad product, containing exclusively the mut9 mutation (Mut-9-9-1), gave rise to reactivated transgene expression, albeit to different extents, only in meiotic tetrad products where the transgene and the mutation cosegregated (Fig. 1D, Tetrad 1 products 3 and 4). However, once a transgene was reactivated by a defect in MUT9, crossing it back into the wild-type background often resulted in incomplete resilencing (Fig. 1E, Tetrad 1, products 1 and 4). This is in contrast to another silencing-defective mutant isolated in the same genetic background, Mut-11 (see below), whose deficiency in transgene repression can be fully reverted by transformation with a MUT11 genomic copy (21) or by crossing the activated transgene back into the wild type (Fig. S3). These observations suggested that the loss of MUT9p affects the reestablishment of silencing, implying that MUT9p itself and/or the product (or products) of its activity may have a role in the inheritance of the silent state. Moreover, we have been unable to detect small RNAs corresponding to the RbcS2:aadA:RbcS2 transgene in the 11-P[300] strain (data not shown) and thus have not found any evidence that RNA interference mechanisms are involved in the silencing of this transgene.

MUT9p Is a Ser/Thr Protein Kinase That Phosphorylates Histones H3 and H2A.

The MUT9 gene codes for a protein kinase related to casein kinase 1 (CK1) (Figs. S4 and S5). However, the region of homology is largely limited to the catalytic motif (Fig. 2A), which contains all 12 subdomains typical of eukaryotic protein kinases (22) (Fig. S4). MUT9p also contains long amino- and carboxyl-terminal regions that are present only in homologous proteins in the land plant (23) and green algae lineages (Fig. 2A and Fig. S5). Consistent with its similarity to CK1 (24), recombinant MUT9p phosphorylates casein in vitro, and its activity is partly inhibited by the isoquinolinesulfonamide CKI-7 (Fig. S6). Yet, there are considerable differences among MUT9p orthologs and CK1 orthologs in the amino acid residues within kinase subdomains III, VIII, and IX (Fig. S4), which play a major role in substrate recognition by eukaryotic protein kinases (22).

Fig. 2.

MUT9p is a Ser/Thr protein kinase that phosphorylates histone H3 Thr-3 and histone H2A. (A) Schematic representation of MUT9p and related Ser/Thr protein kinases. The catalytic domain and a nuclear localization signal (NLS) are indicated on the MUT9p diagram. The brown or pink regions in related proteins represent aligned sequences with >55% or >35% identity, respectively. At, Arabidopsis thaliana; Hs, Homo sapiens; Sc, Saccharomyces cerevisiae; Sp, Schizosaccharomyces pombe. (B) Phosphorylation activity of MUT9p on calf thymus histones. An RNA helicase (MUT6p) and a MUT9p point mutant (MUT9p K174R) were used as negative controls. (C) Recombinant MUT9p (rM9) activity on itself and on Chlamydomonas (CrH3) or calf thymus (cH3) histone H3. (D) Phosphorylation activity of MUT9p on HPLC-purified Chlamydomonas histones, with the specific histones indicated on the left. (E) Cauliflower mononucleosomes as the substrate for MUT9p. (F) Mononucleosomes, predigested with trypsin, as the substrate for MUT9p. The globular histone cores are indicated by asterisks. (G) Identification of the MUT9p phosphorylation site on H3 by N-terminal sequencing of ³²P-labeled protein. The MUT9p activity on recombinant H3 was also examined by immunoblotting with an anti-H3T3ph antibody (Inset). (H) In vivo phosphorylation levels, detected with an anti-phospho-Thr antibody, in partly purified histones from the indicated strains. Calf thymus histones were used as a positive control (Histones). (I) Immunoblot analysis of in vivo H3T3ph amounts with the modification specific antibody.

Because a defect in MUT9 resulted in reactivation of transcriptionally silent loci (20) and a MUT9p:β-Glucuronidase fusion protein localized almost exclusively in the nucleus (Fig. S7), we tested whether MUT9p could modify histones, conceivably affecting chromatin states. In vitro, recombinant MUT9p phosphorylates histones H3, H2A, and, to a lesser extent, H4 (Fig. 2 B–D). However, this activity is virtually abolished by mutation of an invariant Lys residue (K174R) in subdomain II (Fig. 2B and Fig. S4), implicated in anchoring and orienting the ATP phosphate donor (22). Interestingly, MUT9p displayed greater specificity in the presence of mononucleosomes as the substrate, modifying exclusively H3 and several H2A variants (Fig. 2E). Moreover, the phosphorylation sites appear to be located in the histone tails because partial digestion of mononucleosomes with trypsin, which leaves the globular cores intact, significantly decreased their capacity to serve as a substrate (Fig. 2F).

By N-terminal Edman sequencing of radioactively phosphorylated protein, we determined that MUT9p predominantly modifies histone H3 on Thr-3 (Fig. 2G). This was also supported by immunoblot analysis of MUT9p-modified recombinant H3 with an antibody that specifically recognizes H3T3ph (Fig. 2G Inset). Moreover, in vivo, the Mut-9 strain shows defective phosphorylation of histones H3 and H2A, as detected by immunoblotting with an anti-phospho-Thr antibody (Fig. 2H). The anti-H3T3ph antibody also revealed a reduction in the level of this modification in the mutant, which was partly restored in MUT9-complemented strains (Fig. 2I). However, H3T3ph is not entirely lost in Mut-9, likely because of the redundant activity of a MUT9p paralog (Fig. S5, Cr SRK). Thus, the MUT9p kinase phosphorylates histones H3 and H2A both in vitro and in vivo, although the specific residue (or residues) modified in H2A remains to be examined.

Histone H3 Phosphorylated at Thr-3 Is Associated with Transcriptionally Repressed Loci.

The Mut-9 phenotypes are consistent with a role of H3T3ph (and, presumably, phosphorylated H2A) in transcriptional silencing. Fischle et al. (25) have hypothesized that the Thr-3/Lys-4 residue pair in the H3 tail may represent a “methyl/phospho switch,” in which the activating function of H3K4me3 may be counteracted by H3T3ph, perhaps by a mechanism analogous to the H3S10ph-mediated abolishment of HP1 binding to H3K9me3 (14, 15). In addition, although H3K4me2 and H3K4me3 have been firmly linked to transcriptionally competent chromatin (1–3), our analysis of another silencing-defective Chlamydomonas mutant, Mut-11, has implicated H3K4me1 in euchromatic silencing (26). Mut-11 lacks a WD40 repeat protein homologous to mammalian WDR5 (26), a core subunit of H3K4 methyltransferase complexes (27, 28). Thus, we decided to examine the relationship between phosphorylated H3T3 and methylated H3K4 in chromatin-mediated gene silencing.

By immunoblotting with histone modification-specific antibodies, Mut-9 showed a global decrease in both H3T3ph and H3K4me1 when compared with the wild type (Fig. 3A). Mut-11 displayed reduction in H3T3ph, an almost complete lack of H3K4me1, enhanced levels of H3K4me2, and a small defect in H3K4me3 (Fig. 3A). The Mut-9 Mut-11 double mutant, which reactivates transgenes and transposons to a higher degree than the individual mutants (20), behaved similarly to Mut-11 except for a somewhat greater deficiency in H3T3ph (Fig. 3A). Thus, both single mutants were defective, to different extents, in H3K4me1 and H3T3ph; possibly implying crosstalk between the machineries responsible for these modifications. Alternatively, the loss of one epigenetic mark in a single mutant might lead to partial reactivation of expression and the transcription-dependent subsequent erasure of the neighboring modification by histone exchange or enzymatic reversal (1–3).

Fig. 3.

Phosphorylated H3T3 and monomethylated H3K4 are preferentially associated with transcriptionally silent loci. (A) Immunoblot analysis of global H3T3ph and H3K4 methylation states in the indicated strains. The asterisk denotes a nonspecifically cross-reacting protein. Mut-11, mutant defective in a core subunit of H3K4 methyltransferase complexes; Mut-9, Mut-11, mut9 and mut11 double mutant. (B) Chromatin immunoprecipitation assay of H3T3ph levels (in relative units) associated with the promoter regions of the RbcS2:aadA:RbcS2 transgene (aadA), the TOC1 retrotransposon (TOC1-LTR), and the constitutively expressed RPS3 gene in the described strains. (C) ChIP analysis of H3T3ph and H3K4 methylation states associated with the transcription units described above and with the TOC1 coding region.

To examine whether changes in H3T3ph were directly involved in the transcriptional reactivation of the RbcS2:aadA:RbcS2 transgene and the TOC1 retrotransposon, we analyzed their chromatin environment by chromatin immunoprecipitation (ChIP) assays. H3T3ph was enriched in the chromatin associated with the promoter of these silenced loci, whereas it was present at low levels in the regulatory region of the constitutively expressed RPS3 gene (encoding ribosomal protein S3) (Fig. 3B, 11-P[300]). Moreover, consistent with a role of H3T3ph in gene repression, this modification was considerably reduced in the reactivated transgene and transposon units in Mut-9, but it was partly restored in a MUT9-complemented strain (Fig. 3B). Interestingly, in Chlamydomonas, H3T3ph and H3K4me1 appear to operate antagonistically to H3K4me2 and H3K4me3, with the first two modifications being enriched in transcriptionally silent units (Fig. 3C, 11-P[300] aadA) and the last two marks occurring predominantly in transcriptionally active units (Fig. 3C, 11-P[300] RPS3). However, this dualism is less pronounced in the case of dispersed TOC1 transposon repeats, in which individual sequences may be subject to different degrees of repression. ChIP analyses in the mutant backgrounds supported this inference. In Mut-9, the loss of H3T3ph from the chromatin associated with RbcS2:aadA:RbcS2 and TOC1 was accompanied by enhanced levels of H3K4me2 and, at least in the case of the transgene, H3K4me3 (Fig. 3C). In Mut-11, as well as in the double mutant, these same transcription units showed reduced levels of both H3T3ph and H3K4me1, whereas H3K4me2 was elevated (Fig. 3C). However, an increase in H3K4me3 in the reactivated transgene and transposon sequences was less obvious in the mut11 mutant backgrounds because these strains are also deficient in H3K4 trimethylation (26), as reflected by the lower H3K4me3 levels associated with the RPS3 chromatin (Fig. 3C).

Discussion

Our results indicate that the MUT9p kinase is required for H3T3 and H2A phosphorylation, and for long-term, heritable gene silencing, in Chlamydomonas. Thr-3 is an evolutionarily conserved residue in histone H3, but the specific function of its phosphorylated form is unclear and may differ between animals and plants. In mammals, H3T3ph is abundant in mitotic chromosome centromeres but virtually undetectable in interphase cells (11, 29). Moreover, phosphorylation of H3T3 by the kinase Haspin (unrelated to MUT9p) has been implicated in sister chromatid cohesion and chromosome segregation (11). In contrast, in higher plants, H3T3ph occurs along entire chromosome arms and correlates with their condensation during mitosis (9). Interestingly, this mark is also detectable in asynchronously cultured cells and enhanced by osmotic stress (9). Our findings suggest that in photosynthetic eukaryotes, H3T3ph plays a role in gene repression.

The specific residue (or residues) modified by MUT9p in histone H2A was not examined. However, mass spectrometry analysis of posttranslational histone modifications in Arabidopsis revealed that the C-terminal tail of H2A is a major site for unique phosphorylations (30). Zhang et al. (30) have detected phosphorylation at S129 and S141 of H2A.7 and at S145 of H2A.5 and H2A.7. Intriguingly, these are some of the larger H2A variants consistent with those detected as phosphorylated by MUT9p in our nucleosome assays (Fig. 2 E and F). Moreover, homology-based structural modeling suggests that MUT9p, unlike CK1, which prefers acidic substrates (31), interacts with peptides in which basic residues flank potential phosphorylation sites (Fig. S8 and Table S1). Such peptides are common in the carboxyl tail of several H2A variants (17, 30), but their putative role as MUT9p substrates awaits further studies.

The MUT9p kinase plays an essential role in epigenetic gene silencing in Chlamydomonas. Although we cannot rule out that this effect might be partly due to a direct function of the protein in chromatin organization and/or to the phosphorylation of polypeptides modulating chromatin structure, the combined analysis of the Mut-9 and Mut-11 mutants supports a role for H3T3ph in the maintenance of silent euchromatin. Both H3T3ph and H3K4me1 are preferentially associated with repressed transcription units (this work and ref. 26). Reactivation of transcription in Mut-11, lacking a H3K4 methyltransferase complex subunit (26), correlates not only with a substantial defect in H3K4me1 but also with the partial loss of H3T3ph. Similarly a MUT9p deficiency results in reduced levels of both H3T3ph and H3K4me1. These effects are further enhanced in a Mut-9 Mut-11 double mutant. Moreover, the presence of H3T3ph appears to antagonize that of the activating di/trimethyl forms of H3K4. Interestingly, a mutually exclusive relationship between H3K4me3 and asymmetric dimethylation of histone H3 Arg-2 (H3R2me2a), the residue adjacent to Thr-3, has recently been reported in budding yeast and mammals (32, 33).

H3K4me3 and H3K4me2 contribute to transcriptional activation by recruiting chromatin-remodeling enzymes, basal transcription factors such as TFIID, and a H3K4 methyltransferase complex (1, 3, 32, 33). As demonstrated for H3R2me2a (32, 33), H3T3ph may play a role in gene repression by interfering with the binding of some of these enzymes to the chromatin of promoter regions. Moreover, in budding yeast, H3K4me1 and H3R2me2a are also enriched toward the 3′ end of genes (32, 34, 35) and, like di/trimethylated H3K36, may help to suppress spurious intragenic transcription by maintaining a repressive chromatin structure after the passage of RNA polymerase II (2, 3, 36, 37). We hypothesize that in Chlamydomonas, H3K4me1 and H3T3ph may operate in similar fashion. In addition, based on evidence from other organisms, phosphorylated H2A may also have a role in transcriptional silencing (18, 38).

MUT9p also appears to be involved in the inheritance of silent chromatin states. A transgene reactivated by a defect in MUT9 commonly underwent deficient resilencing when crossed back into the wild-type background. By contrast, when the same transgene was activated by disruption of MUT11 (21, 26), or by two other unlinked mutations (data not shown), it was fully repressed upon its transfer back into the wild type. Given the hypersensitivity of Mut-9 to genotoxic agents such as bleomycin (this work and ref. 20), the MUT9p-mediated phosphorylation of histones may initially be associated with the repair of DNA double-strand breaks (also coupled to the genomic integration of transgenes and certain transposons). Once established, this chromatin structure might, in the absence of activating signals, self-perpetuate, resulting in the sequence-independent repression of integrated foreign DNA. Our results strongly implicate H3T3ph (and H3K4me1) as an epigenetic mark involved in the maintenance of this gene silencing. However, the inheritance of the repressed state does not appear to depend on H3T3ph because this mark is also lost to a considerable degree in Mut-11 without any consequence in subsequent transgene resilencing. Indeed, the potential role of MUT9p in the re-establishment of gene silencing through cell divisions remains to be explored.

Materials and Methods

Chlamydomonas reinhardtii Strains, Culture Conditions, and Genetic Analyses.

The 11-P[300], 1-P[300], and Mut-11 strains have been previously described (20, 26, 39). Chlamydomonas cells were routinely grown in TAP medium under moderate light conditions (20, 26, 40). Reactivation of expression of the RbcS2:aadA:RbcS2 transgene was examined by spotting dilutions of cells on TAP plates containing spectinomycin (39). To test whether the insertional mutagen cosegregated with reactivation of transgenic expression, we crossed Mut-9 and Mut-11 to the wild-type strain of opposite mating type, CC-125, and dissected tetrads as previously described (40). A meiotic tetrad product (Mut-9-9-1), containing exclusively the mut9 mutation, was then crossed to the 11-P[300] and 1-P[300] strains, and transgenic expression in the meiotic progeny was evaluated by spot tests. Chlamydomonas sequences flanking the tagging plasmid in Mut-9 were recovered by constructing a cosmid library from mutant genomic DNA and screening, by hybridization, for cosmids containing the inserted vector DNA (41). These sequences were then used as probes to identify hybridizing cosmids from a Chlamydomonas wild-type genomic library.

DNA and RNA Analyses.

Standard protocols were used for nucleic acid isolation, fractionation by gel electrophoresis, and detection (20, 39, 41). DNA methylation was assayed as previously described (39, 42).

Phylogenetic Analyses.

Sequences corresponding to the Ser/Thr kinase domain were identified by using the SMART database and aligned with ClustalX v1.81 (43). The Neighbor-joining method (44) and the MEGA program v3.1 (45) were used to obtain phylogenetic trees with Poisson-corrected amino acid distances.

Protein Kinase Assays.

MUT9p was expressed as a His-tagged protein in Escherichia coli and purified with a Ni-NTA His binding resin (Novagen). A recombinant MUT9p protein with a point mutation in the Ser/Thr protein kinase domain (K174R) was generated with the QuikChange Site-Directed Mutagenesis kit (Stratagene). Protein kinase activity of the recombinant polypeptides was assayed as previously described (24). Some reactions contained the CK1-specific inhibitor CKI-7 (Toronto Research Chemicals). Protein substrates included α-casein (Sigma), calf thymus histone H1 (Sigma), calf thymus core histones (Roche), HPLC-purified Chlamydomonas histones (26), and cauliflower mononucleosomes, prepared as previously described (46).

Immunoblot Analyses.

Histone phosphorylation or methylation status in vivo was examined by Western blotting (26, 42). H3K4 methylated states were detected with antibodies against H3K4me1 (ab8895; Abcam), H3K4me2 (07-030; Upstate), or H3K4me3 (ab8580; Abcam). H3T3 phosphorylation was examined with two antibodies: 07-424 (Upstate) for the immunoblots presented in Fig. 2 G and I and 05-746 (Upstate) for all remaining work. Sample loading was adjusted relative to the signal obtained with a modification-insensitive anti-H3 antibody (ab1791; Abcam). Histones purified by differential centrifugation and acid extraction (42), in the presence of phosphatase inhibitors, were used to analyze their degree of phosphorylation with an anti-phospho-Thr antibody (catalogue no. 9381; Cell Signaling Technology).

Chromatin Immunoprecipitation Assays.

The methylation/phosphorylation status of histone H3 at specific chromosomal loci was examined by using ChIP, as previously described (26, 42). Because the antibody against H3K4me3 (ab8580; Abcam) cross-reacts with several nonhistone proteins, sequential ChIP was used to immunoprecipitate DNA specifically associated with this modification (42). Human IgG (I-2511; Sigma) was used as a negative control. Immunoprecipitated DNA was quantified by real-time PCR on a Bio-Rad iCycler with SYBR Green (42). The primers used for amplification of target loci have been previously reported (26, 42).

Analysis of MUT9p-GUS Subcellular Localization by Transient Expression in Onion Cells.

A fusion between MUT9p and E. coli β-glucuronidase (GUS), under the control of the cauliflower mosaic virus 35S promoter, was constructed in plasmid pPTN134 (47). Onion epidermal cells were transformed by microprojectile bombardment, and GUS activity was detected as previously described (47).

Homology-Based Structural Modeling of MUT9p.

Homology modeling was performed on the SWISS MODEL web server (48) by using the catalytic subunit of cAMP-dependent protein kinase A (PDB 1JBP) as the template. Structural mining analyses were done with the Swiss-Pdb Viewer (49), and graphical representations were prepared with PyMOL (DeLano Scientific).