A phosphorylated subpopulation of the histone variant macroH2A1 is excluded from the inactive X chromosome and enriched during mitosis

mH2A Is Phosphorylated in Its Basic Hinge Domain.

To enrich for endogenous mH2A, we used HEK293 cells, a female cell line that contains multiple inactive X chromosomes as a result of aneuploidy. Because mH2A is enriched on the Xi, we rationalized that this cell line would be a potentially enriched source of this minor variant. Isolation of histone variants from endogenous, nonoverexpressed sources can be challenging because of their low quantity in chromatin (<10% of total histone pool). To this end, mH2A was acid-extracted along with other core histones from HEK293 cells under asynchronous growth conditions. Total acid-extracted histones were fractionated by reverse-phase HPLC (Fig. 1a), and fractions containing mH2A were determined by using an antibody against mH2A1, which recognizes both mH2A1.1 and mH2A1.2 (Fig. 1b). mH2A1 elutes late in the HPLC chromatogram and is significantly less abundant than conventional histones (Fig. 1a). A slower-migrating, cross-reacting band above the mH2A1 signal was repeatedly observed by immunoblotting (IB) and is consistent with the monoubiquitylated form of mH2A given its size (an ≈8-kDa shift; Fig. 1b, asterisk) (28, 29). Multiple HPLC runs were carried out, and peak fractions from all runs were pooled for MS; a small portion of the total pool was examined by SDS/PAGE to determine the major constituents of this sample (Fig. 1c). MS analysis identified mH2A as the major component, and HSP60 as a likely contaminant (Fig. 1c). Both in-gel and in-solution enzymatic digestion was performed for MS to identify PTMs of mH2A1 [see supporting information (SI) Materials and Methods].

In the present work, particular attention was paid to the hinge region of mH2A1, a very basic stretch of amino acids that contains multiple potential phosphate acceptor residues (Fig. 2a, Upper). Because this region is lysine-rich, it is not amenable to enzymatic digestion with trypsin and therefore was targeted with endoproteinase GluC, generating a peptide of ≈40 aa (Fig. 2a, Upper and SI Fig. 5 a and b). Through our MS analysis, two phosphorylation sites were detected within the hinge of endogenous mH2A1: T128ph and S137ph, the latter of which is a previously unidentified modification of mH2A (for spectra, see SI Fig. 5 a and b). Intriguingly, a significant fraction (≈5%) of mH2A was phosphorylated at S137 in asynchronous cultures, a considerable amount given that there was no prior enrichment for phosphopeptides. Of note, S137ph was detected on peptides that contained either KK or KKK C-terminal to S137, which are forms of alternate splice selection (Fig. 2a, underlined) (6). The significance of this splice selection is unknown.

To investigate the biological implications of mH2AS137ph, a highly specific affinity-purified polyclonal antibody was generated. Peptide competitions demonstrated that a peptide phosphorylated at S137 could compete antibody reactivity when HEK293 histones were analyzed by IB, whereas the unmodified peptide could not (Fig. 2b). Furthermore, peptide competitions with various phosphorylated peptides from other histones such as H3S10ph, H3S28ph, and H3.3S31ph (SI Fig. 5c) and peptide dot blots and Luminex assays (data not shown) demonstrated specificity for mH2A phosphorylation.

mH2AS137ph Is a Prevalent PTM.

The hinge region of mH1A1.1 and 1.2 is identical, and we questioned whether both splice forms are phosphorylated on S137. However, because our MS analyses generated an identical peptide from mH2A1.1 and 1.2, we were unable to confirm whether the phosphorylation is present on one or both splice variants. Thus, we used SDS/PAGE to separate mH2A1 isoforms from adult mouse liver nuclei, a rich source of mH2A. Interestingly, both mH2A1.1 and 1.2 are phosphorylated at S137 in male and female liver (Fig. 2c). Adult mouse liver is largely devoid of dividing cells, and the presence of S137ph in this tissue may represent a potential role of mH2A in senescence because mH2A is also enriched in senescence-associated heterochromatin foci (11). Of note, S137ph signal is absent from mH2A knockout liver in IB, again demonstrating specificity of this antibody (Fig. 2c).

Although mH2A2 is not highly conserved with mH2A1 in the hinge region, it is also lysine-rich. Sequence alignments of the three mammalian mH2A isoforms show that T128 is replaced by a serine in mH2A2, a potentially phosphorylated residue (Fig. 2a, Lower). However, in the case of S137, mH2A2 contains a lysine residue (Fig. 2a, Lower). Thus, the ability of only mH2A1 to be phosphorylated at S137 may provide a unique role to this variant.

We next used the S137ph antibody to screen cell lines of different origin for its presence, and all cells tested contain this PTM (Fig. 2d). Interestingly, differentiated mouse ES cells (and 3T3s) contain more mH2AS137ph than undifferentiated ES cells, suggesting a potential role for this modification in chromatin changes throughout differentiation. S137ph is not exclusive to female cells because male ES cells also have significant amounts this modification (Fig. 2d, Right); it is also observed in male mouse liver (Fig. 2c). We note that although female liver has higher levels of S137ph than males, and the opposite trend occurs in ES cells, this difference may reflect variation in cell/tissue types. Taken together, these data suggested that the subpopulation of mH2A containing S137ph was not exclusive to the Xi, but rather that this modified form of mH2A has a distinct function within the chromatin template (see below).

mH2AS137ph Is Excluded from the Xi.

To test the above hypothesis directly, we performed immunofluorescence (IF) analyses in HEK293 cells with antibodies against S137ph and mH2A (Fig. 3a). Although mH2A staining showed two or three intensely stained foci (Xis), S137ph staining did not produce similar results. We next generated a HEK293 cell line stably expressing mH2A1.2 fused to GFP, allowing direct visualization of Xis. Again, IF analysis, by using antibodies in direct comparison with the mH2A-GFP fusion, showed that, whereas mH2A was enriched on the Xi, the phosphorylated form was not (Fig. 3b). This lack of enrichment is not caused by the inability of the GFP fusion protein to be phosphorylated on S137 (SI Fig. 5d). Although the localization of S137ph is not completely excluded from the Xi in all cases (≈75% Xis show no enrichment, with 148 Xis counted), this modification is not enriched on the Xi as seen for general mH2A antibody (Fig. 3b). Furthermore, S137ph is predominantly chromatin-associated as demonstrated by a chromatin fractionation assay and is released from chromatin by micrococcal nuclease digestion (Fig. 3c).

To test the localization of S137ph mH2A during ES cell differentiation, we used a female mouse ES cell line (LF2) that inactivates one of its X chromosomes upon retinoic acid treatment (30). Because mH2A associates with the Xi late in the differentiation process [≈day 5 or 6 in LF2 cells (31)], we examined S137ph localization before (day 2) and after (day 9) mH2A associates with the Xi (Fig. 3d). In both cases, mH2AS137ph is absent from the Xi (marked by Xist RNA FISH; Fig. 3d), indicating that a unique pool of phosphorylated mH2A is excluded from this important epigenetically regulated heterochromatic structure. Interestingly, S137ph stains interphase nuclei in a highly punctate pattern, the significance of which is unclear; however, we observed some overlap with regions of active transcription in LF2 cells (visualized by RNA polymerase II staining; SI Fig. 6).

S137 Phosphorylation of mH2A Is Enriched During Mitosis.

The absence of S137ph on the Xi raises the possibility that phosphorylation of mH2A in the hinge region might prevent its association with the Xi or certain forms of heterochromatin. To dissect the role of this PTM further, we examined three key cellular mechanisms linked to phosphorylation of other histones: (i) DNA damage response, S139ph on H2A.X; (ii) apoptotic chromatin, H2BS14ph; and (iii) chromatin condensation during mitosis, S10ph and S28ph on H3 (22–26, 32). We examined the levels of S137ph after induction of apoptosis by etoposide treatment and DNA damage induced by ionizing irradiation, and we observed no change in S137ph levels after either treatment (SI Fig. 7 a and b). We did, however, observe a consistent change in the levels of S137ph during cell cycle progression. Although asynchronous cell populations contain S137ph (Fig. 2d), the levels increase significantly upon nocodazole treatment, whereas an S phase block with hydroxyurea results in very low levels (Fig. 4a). Analyses of synchronized HeLa cells induced by a double thymidine block procedure confirmed that the levels of mH2AS137ph increase during mitosis (Fig. 4b). These IB results were confirmed by IF studies, which showed strong staining of mitotic chromosomes in U2OS cells compared with interphase nuclei (Fig. 4c); similar results were seen with HeLa cells (SI Fig. 7c).

Phosphorylation and dephosphorylation of many substrates are well documented to mediate large-scale changes in cellular organization during the cell cycle. Because mH2AS137ph is observed by IB throughout the cell cycle and in IF staining during interphase, we suggest that the role of this mark is not limited to mitosis. In keeping, other well characterized histone phosphorylation events exhibit complex relationships during the cell cycle. For example, H3S10 and S28 are phosphorylated during periods of immediate–early gene induction or mitosis (23, 27). The timing of mH2AS137 phosphorylation resembles that of linker histone H1, which is present during interphase and becomes hyperphosphorylated during mitosis (22). Because S137ph is next to a proline residue (Fig. 2a) and appears similar in timing to cyclin-dependent kinase (Cdk)-mediated H1 phosphorylation, we suspected that S137 phosphorylation is also catalyzed by Cdks. In vitro kinase assays demonstrate that recombinant Cdk complexes, including Cdk2/cyclin E and Cdk1/cyclin B (S phase and mitotic regulators, respectively) can phosphorylate an mH2A peptide containing an unmodified S137 residue (Fig. 4d). Interestingly, phosphorylation of nearby S139 (a currently unidentified mH2A PTM) prevents S137 phosphorylation (Fig. 4d) and may represent a switching mechanism for potential binding effectors. This effect is not caused by epitope occlusion by S139ph because S137 phosphorylation is also reduced on this peptide in radioactive kinase assays (data not shown). Other Cdk/cyclin combinations can also phosphorylate the unmodified peptide, whereas S/T kinases such as casein kinase II and tyrosine kinases such as Wee1 are incapable of phosphorylating the same peptide in parallel assays (data not shown). Because Cdks can compensate for the loss of other Cdks during the cell cycle, it has been difficult to assess which Cdk is responsible for S137 phosphorylation in vivo, and it remains an important challenge for the future.

Histone Isolation and Purification.

Histone isolation was performed essentially as described in ref. 44 and SI Materials and Methods. Histones were purified by RP-HPLC and screened for mH2A by IB, and positive fractions were pooled for MS analysis (see SI Materials and Methods).

IB and IF.

For antibodies used in this work, see SI Materials and Methods. Peptide competitions for IB were performed by incubating S137ph antibody with each peptide at 2 μg/ml for 2 h at room temperature before overnight incubation with the membrane. IF analysis on HEK293 cells was carried out by using poly-l-lysine (Sigma)-covered slips, fixed in 2% formaldehyde, permeabilized, and stained. Cells were analyzed on Axioskop 2 plus and Axioplan 2 microscopes (Zeiss). IF and RNA FISH on female ES cells (LF2) were performed as described in ref. 31, and cells were analyzed on a Deltavision microscopy system (Applied Precision).

Chromatin Fractionation and Separation of mH2A Isoforms.

Chromatin fractionation was performed as described in ref. 45. Mouse liver mH2A isoforms were separated as described in ref. 35.

Cell Culture and Cell Cycle Arrest.

Cell lines were grown under standard conditions (SI Materials and Methods). Rat mH2A1.2 (95% similar to human; GenBank accession no. U79139) was cloned into EGFP-N1 (Clontech), generating a C-terminal EGFP tag. Stable HEK293 cell lines were created by neomycin selection. Nocodazole-arrested HeLa cells were treated with 200 ng/ml inocodazol (Sigma) for 16 h and harvested by mitotic shake off. HeLa cells were arrested in G₁/S by double-thymidine block with 2 mM thymidine for 16 h, released for 9 h, and blocked for 16 h. Hydroxyurea block was used to synchronize cells in S phase by treating for 20 h with 2 mM hydroxyurea.

Kinase Assays.

Unmodified and phosphorylated mH2A peptides (amino acid 132–142) were incubated with 10–40 ng of recombinant Cdk/cyclin complexes (Millipore) and ATP (500 μM) for 1 h at 30°C. Recombinant H1 was used as a positive control for Cdk reactions (data not shown). Detailed reaction conditions can be found at the distributor's website.