|Home | About | Journals | Submit | Contact Us | Français|
Defects in the nuclear envelope or nuclear ‘lamina’ networks cause disease and can perturb histone posttranslational (epigenetic) regulation. Barrier-to-Autointegration Factor (BAF) is an essential but enigmatic lamina component that binds lamins, LEM-domain proteins, DNA and histone H3 directly. We report that BAF copurified with nuclease-digested mononucleosomes and associated with modified histones in vivo. BAF overexpression significantly reduced global histone H3 acetylation by 18%. In cells that stably overexpressed BAF 3-fold, silencing mark H3-K27-Me1/3 and active marks H4-K16-Ac and H4-Ac5 decreased significantly. Significant increases were also seen for silencing mark H3-K9-Me3, active marks H3-K4-Me2, H3-K9/K14-Ac and H4-K5-Ac and a mark (H3-K79-Me2) associated with both active and silent chromatin. Other increases (H3-S10-P, H3-S28-P and silencing mark H3-K9-Me2) did not reach statistical significance. BAF overexpression also significantly influenced cell cycle distribution. Moreover, BAF associated in vivo with SET/I2PP2A (protein phosphatase 2A inhibitor; blocks H3 dephosphorylation) and G9a (H3-K9 methyltransferase), but showed no detectable association with HDAC1 or HATs. These findings reveal BAF as a novel epigenetic regulator and are discussed in relation to BAF deficiency phenotypes, which include a hereditary progeria syndrome and loss of pluripotency in embryonic stem cells.
Nucleosomes, the basic unit of chromatin structure, are regulated by mobile chromatin-associated proteins1,2 and enzymes that reversibly (‘epigenetically’) modify DNA or specific core histone residues.3 Such modifications generate two general transcriptional states: active, where chromatin is typically more acetylated and silent, where chromatin is more methylated.4 However, the functions and relationships of histone ‘marks’ are complex and dynamic; some marks recruit further regulators (e.g., methyl H3-K9 recruits heterochromatin protein 1 [HP1]),5 and some are also important for cell cycle control (e.g., H3-S10-phosphorylation during mitosis).6
Genome organization, gene position and epigenetic control are further influenced by nuclear intermediate filaments formed by A- or B-type lamins (reviewed in refs. 7–9) and by promoter proximity to the nuclear envelope.10-13 Lamin filaments concentrate near the nuclear envelope (NE) and bind most NE inner membrane proteins directly,14 forming ‘lamina’ networks that support transcription, replication, genome organization, development and DNA repair (reviewed in refs. 7 and 15). Lamina defects cause a myriad of human diseases (laminopathies) including muscular dystrophy, cerebellar disorders and Hutchinson-Gilford Progeria Syndrome (HGPS).16 Progeria-associated and other mutations in LMNA perturb not only nuclear structure, but also heterochromatin and epigenetic regulation.17-22 How the genome depends on lamin networks is a central question.
Lamina networks have three major types of components: lamins, LEM-domain proteins (e.g., LAP2, emerin, MAN1, Lem2/NET25; most located at the NE inner membrane)14 and an enigmatic soluble protein named Barrier-to-Autointegration Factor (BAF).23 These components are mutual interactors that are collectively required to rebuild the nucleus during anaphase and telophase.24-29 The timing of nuclear assembly is regulated, at least in part, by a conserved kinase, Vaccinia-related kinase 1 (VRK1), which directly phosphorylates BAF during mitosis to inhibit its chromatin association and consequently also the chromatin association of LEM-domain proteins and nuclear membranes.28 After assembly, lamins and LEM-domain proteins are in molecular contact with and are proposed to regulate and/or maintain silenced chromatin.8,11,30-32 Whether BAF influenced genome silencing was previously unknown.
BAF is abundant in mammalian cells (~9 μM near NE, ~2 μM in nucleoplasm),33 consistent with its role as an essential component of nuclear lamina networks. Perhaps contrary to one’s expectations, BAF also has high diffusional mobility during interphase.34,35 BAF is a functional homodimer with two binding sites for dsDNA36,37 that can condense long DNA by ‘looping’ in vitro.38 BAF profoundly influences higher-order chromatin structure,26,27,29,39 and represses transcription at specific promoters.35,40 BAF also associates in vivo with specific linker histones,41 transcription factors (including Sox2, a master regulator of pluripotency),40,42,43 poly(ADP) ribose polymerase 1 (PARP1) and retinoblastoma binding protein 4 (RBBP4).42 Notably, BAF binds the tail plus helix-αN of histone H3 with high affinity in vitro,41 suggesting it associates with nucleosomes. New findings, presented here, support this model and reveal BAF as an epigenetic regulator. These findings are discussed in relation to the recently discovered roles of BAF in stem cell self-renewal,43 the epigenetic defects caused by progeria-associated lamin A mutations and the discovery that BAF insufficiency causes a hereditary progeroid syndrome.44
Previous studies using 35S-Met-labeled BAF to probe immobilized core histones41 suggested weak but detectable direct binding to H4, in addition to H3. We confirmed direct binding between recombinant purified Xenopus histone H4 (xH4) and recombinant purified His-tagged human BAF dimers (hereafter, ‘H6BAF’). xH4 pelleted in the presence but not in the absence of H6BAF (Fig. 1A, lanes 4 vs. 2; blot shown is typical of three independent experiments [n = 3]). Direct binding to H4, in addition to H3 and DNA, supported potential association(s) with nucleosomes, non-nucleosomal H3-H4,45 or both.
To test potential BAF association with nucleosomes in vivo, we isolated chromatin from formaldehyde-crosslinked HeLa cells and digested extensively with micrococcal nuclease to generate mononucleosomes, as described by Kim et al.46 Digestion was complete, yielding primarily ~80–180-bp DNA fragments as assayed by agarose gel electrophoresis (Fig. 1B, arrow). The presence of core histones was verified by SDS-PAGE with purified chicken core histones as markers (Fig. 1C, lanes 1 and 2, respectively). Endogenous BAF co-purified with these isolated mononucleosomes, shown by direct protein gel blotting with antibodies against BAF (Fig. 1D, input (I), lanes 1 and 4; n = 3) and by immunoprecipitating BAF from isolated mononucleosomes (Fig. 1D, lane 5 vs. 2; n = 3). These results supported the hypothesis that BAF associates with nucleosomes in vivo. Strengthening our results, an independent laboratory previously identified BAF by mass spectrometry from purified H3.1-nucleosome complexes.47 As ‘food for thought’, the atomic structures of the nucleosome48 and BAF,49 determined separately, are shown at the same scale in Figure 1E.
To determine if BAF associated with endogenous posttranslationally modified histones, antibodies specific for each of eight posttranslational modifications on H3 or H4, or control antibodies against the C-terminus of H3 (‘bulk H3′), were used to immunoprecipitate histones from HeLa whole cell lysates. An example of an original protein gel blot is shown in Fig. S1; blots were quantified as described in Methods. BAF co-precipitated to varying extents with all tested active (dimethylated H3-K4, phosphorylated H3-S10, hyperacetylated H4 [H4-Ac5; recognizes tri- and tetra-acetylated H4], acetylated H4-K5 and acetylated H4-K16) or silenced (dimethylated H3-K9, mono/trimethylated H3-K27 and monomethylated H4-K20) marks on histones H3 and H4 (Fig. 2; n = 3). Our results revealed no obvious preference for silenced vs. active marks, but supported an association between BAF and endogenously modified chromatin.
To test if BAF influences histone posttranslational modifications (PTMs) in vivo, we used metabolic labeling with sodium [14C]-acetate to quantify core histone acetylation in HeLa cells that transiently overexpressed BAF. Cells that overexpressed BAF 2-to-10-fold (data not shown) incorporated significantly (~20%) less [14C]-acetate into H3 (Fig. 3A; p = 0.03, n = 5). Transient BAF overexpression did not detectably alter histone protein levels (data not shown). The same trend (~20% less acetylation) was seen for H4 and H2A/H2B (which co-migrated in our gels), but did not reach statistical significance (Fig. 3A; p = 0.08 each, n = 5), possibly due to variable BAF expression levels. Reduced bulk H3 acetylation suggested excess BAF influenced the activity or access of histone-modifying enzymes.
To achieve consistent BAF overexpression levels, we used previously-described50 stable cell lines, derived from human HEK293 cells with tetracycline (Tet)-inducible expression of a single-copy insert of FLAG-tagged BAF (293:BAF cells) and corresponding control (293:CAT) cells. Total BAF levels in 293:BAF cells increased ~3-fold within 8.5 h after Tet-induction (Fig. S2A) and remained near this level for at least three days (Fig. S2B).
We Tet-induced BAF expression for 32–52 h and protein gel blotted whole cell lysates to determine if excess BAF altered the protein levels of specific histone-modifying enzymes. Relative to γ-tubulin, total BAF (endogenous plus FLAG-BAF) levels were ~3-fold higher in the 293:BAF cells (Fig. 3B; n = 4). In contrast, no significant changes were seen in the protein levels of histone deacetylase 2 (HDAC2) or HDAC4 (class I and class IIa HDACs, respectively)51 (Fig. 3B), emerin or the histone methyltransferase G9a (Fig. S2A). To quantify potential changes in specific histone PTMs, histones were acid-extracted from the same 293:BAF and 293:CAT samples, resolved by SDS-PAGE and protein gel blotted using antibodies specific for each of eight marks on H3 (Fig. 3C) or six marks on H4 (Fig. 3D). Examples of original protein gel blots are shown in Figure S3 and were quantified as described in Materials and Methods. Several marks were affected in BAF-overexpressing cells. Three marks decreased significantly: silencing mark H3-K27-Me1/3 (by ~15%; Fig. 3C; p = 0.02, n = 4) and active marks H4-K16-Ac (by ~10%; Fig. 3D; p = 0.03, n = 4) and H4-Ac5 (by ~30%; Fig. 3D; p = 0.04, n = 4). Five marks increased significantly: silencing marks H3-K9-Me3 (by ~20%; Fig. 3C; p = 0.002, n = 4) and H3-K79-Me2 (by ~10%; Fig. 3C; p = 0.01, n = 4) and active marks H3-K4-Me2 (by ~15%; Fig. 3C; p = 0.01, n = 4), H3-K9/K14-Ac (by ~15%; Fig. 3C; p = 0.02, n = 4) and H4-K5-Ac (by ~15%; Fig. 3D; p = 0.02, n = 4). Other increases that did not reach statistical significance were seen for silencing mark H3-K9-Me2 (increased ~30%; Fig. 3C; n = 4) and phosphorylation at H3-S10 and H3-S28 (increased ~30% and ~20, respectively; Fig. 3C; n = 4). Two factors might have influenced or obscured detection of significant BAF-induced changes in H3 phosphorylation: H3 hyperphosphorylation during mitosis and potential effects of BAF overexpression on cell cycle progression. For example BAF-null Drosophila and C. elegans show cell cycle arrest phenotypes,26,27 and BAF downregulation prolongs S phase in somatic mammalian cells,52 and increases the proportion of cells in G2/M in embryonic stem cells.43
To test for potential cell cycle phenotypes, we examined the cell cycle distribution of unsynchronized 293:CAT and 293:BAF cells Tet-treated for 35 h to induce FLAG-BAF expression. Based on representative histograms of DNA content obtained by flow cytometry (Fig. 4A, ‘DNA content’ panels), we determined the distribution of cells in specific cell cycle phases using ModFit software (Fig. 4A, right). The percentage of G0/G1 cells, averaged from duplicates of three independent experiments, was significantly higher in BAF-overexpressing populations (42.6 +/− 4.1%) than in 293:CAT controls (39.0 +/− 4.2%; Fig. 4A, right; p = 0.04). BAF-overexpressing populations appeared to have correspondingly fewer cells in S phase, but this difference was not statistically significant by this assay (Fig. 4A, right; p = 0.08). To study S phase more precisely, we did parallel studies in which asynchronous 293:BAF and 293:CAT cells were Tet-treated 35 h, then pulse-labeled 30 min with 5-Bromo-2’-Deoxyuridine (BrdU). In three independent duplicate experiments, significantly fewer 293:BAF cells (48.55 +/− 3.59%) than 293:CAT controls (54.32 +/− 4.47%) were in S phase, as assessed by BrdU incorporation (Fig. 4B, right; p = 0.02). Thus, G0/G1 phase was significantly longer and S phase was significantly shorter in asynchronous BAF-overexpressing cells.
To specifically examine progression to mitosis, asynchronous Tet-induced cells were assayed by flow cytometry using the M-phase antibody MPM-2,53 which recognizes mitosis-specific phospho-epitopes.54 In three independent experiments a significantly higher percentage of BAF-overexpressing 4C cells (G2/M) were MPM-2-positive (2.3 +/− 0.08%) compared with 293:CAT controls (1.44 +/− 0.21%; Fig. 4C, right). This significant (p = 0.02) difference suggested BAF-overexpression reduced G2 phase, or prolonged M phase, or both, in unsynchronized cells. Neither population accumulated polyploid cells (> 4C DNA content; Fig. 4A), suggesting cytokinesis was unaffected.
As controls, whole cell lysates were protein gel blotted for BAF and other proteins: G9a, SET/I2PP2A (inhibitor of protein phosphatase 2A), HDAC2, HDAC4, lamins A/C and cyclin E. Tet-induced 293:BAF cells showed a significant ~2.5-fold increase in total BAF (endogenous plus FLAG-BAF) relative to 293:CAT controls (Fig. 4D; p = 0.04; n = 3), as expected. Other tested proteins showed no significant changes (Fig. 4D). Overall, asynchronous BAF-overexpressing populations showed a significant (~10%) increase in the proportion of G0/G1 cells, a significant (~10%) decrease in S-phase cells and a significant (~50%) increase in the percentage of M-phase cells.
To explain some of the changes in histone posttranslational modification in BAF-overexpressing cells, we hypothesized BAF might interact with and potentially recruit, specific chromatin regulators. Potential interactors from our BAF proteome42 included HDAC1 and/or HDAC2 (distinguishing peptides not found), SET/I2PP2A (blocks H3-S10 dephosphorylation)55 and G9a, an H3-K9 methyltransferase that generates the H3-K9-Me2 mark.56 We therefore tested potential association with HDAC1, SET/I2PP2A or G9a and also tested for potential associations with histone acetyltransferases (HATs) or Aurora B (phosphorylates H3-S10 and H3-S28 during mitosis),6 as controls for the observed changes in histone acetylation and phosphorylation.
We transiently expressed in HeLa cells FLAG-tagged versions of three different HATs (PCAF, p300 or Tip60), HDAC1, G9a or SET/I2PP2A, with FLAG-H3.1 and empty FLAG vector as positive and negative controls, respectively. Cell lysates were immunoprecipitated with FLAG antibodies. Endogenous BAF co-immunoprecipitated with histone H3.1 as expected41 (Fig. 5A, lane 4, αBAF) but showed no detectable association with any tested HAT (Fig. 5A, lanes 6, 8, 10; n ≥ 3) or HDAC1 (Fig. 5A, lane 12; n = 3) in vivo. Endogenous BAF consistently and robustly co-immunoprecipitated with FLAG-G9a (Fig. 5B, lane 4, αBAF; n = 3) and FLAG-SET/I2PP2A (Fig. 5C, lane 4, αBAF; n = 3) from cells. Reciprocal experiments gave similar results: FLAG-BAF immunoprecipitated both endogenous G9a (Fig. 5D, lane 4, αG9a; n > 3) and endogenous SET/I2PP2A (Fig. 5D, lane 4, αSET; n = 3). No association with FLAG-Aurora B was detected in HeLa cells, where positive controls verified BAF association with FLAG-SET/I2PP2A (Fig. 5E, lanes 2 vs. 4, αBAF; n = 3). These results independently validated G9a and SET/I2PP2A as BAF-associated (directly or indirectly) in vivo, confirming our previous proteomic results.42 We speculate BAF might favor H3-S10 phosphorylation and H3-K9 methylation in vivo by promoting the association of G9a and SET/I2PP2A with chromatin. On the other hand, despite its significant effects on histone acetylation (decreased global H3 acetylation; increases or decreases in four specific acetylation marks), we found no evidence that BAF associates with any tested HDAC or HAT in vivo. Further work is needed to understand the mechanism(s) by which BAF affects histone acetylation. Notably, as an essential-yet-mobile component of the ‘lamina’ network, BAF might influence epigenetic regulation by a mechanism that does not involve the direct recruitment of HDACs, HATs or other chromatin-associated factors to chromatin.
BAF, an essential and highly mobile component of the nuclear lamina network,23 is shown here to associate with nucleosomes and broadly influence the epigenetic landscape. Eight histone marks increased or decreased significantly in cells that overexpressed BAF ~3-fold, with six other marks also potentially affected. BAF-overexpression caused histones to be underacetylated (reduced bulk acetylation and four of six site-specific marks decreased) and hypermethylated (five of six marks increased). These changes are unlikely to reflect a single underlying mechanism, given the complex functional hierarchies and dynamics of histone marks and their regulators,57 and huge gaps in knowledge about how chromatin is influenced by nuclear structure.8,9 Our findings begin to fill this gap, by revealing BAF as a novel epigenetic regulator.
Significantly fewer BAF-overexpressing cells were in S phase, suggesting S phase was shorter. This result was consistent with and strongly supports the proposal by Haraguchi and colleagues that BAF promotes DNA replication,52 based on their finding that S phase is four hours longer in BAF-downregulated HeLa cells. Whether S-phase roles for BAF include lamins15 or LEM-domain proteins58 is still unknown. BAF localizes exclusively in the nucleus during S phase in mortal cells.52 Our findings suggest potential epigenetic mechanism(s) by which BAF might influence replication, since histones are specifically modified at replication origins and during elongation.59,60 For example, H4-K20-Me, which increased without reaching statistical significance in asynchronous BAF-overexpressing cells, has positive roles in replication. Cells downregulated for SET8 (PR-Set7), the methyltransferase responsible for this mark, accumulate in S phase due to slowed replication forks and origin firing, dsDNA breaks and DNA damage response signaling.61,62 Alternatively, BAF might facilitate replication-induced DNA damage repair,42 or the release of late-replicating heterochromatin, which relocates to the nuclear interior for replication and is subsequently retethered to the NE.63 Interestingly, specific regions (about 1%) of the human genome replicate after canonical S phase, late in G2/M phase.64 Moreover, BAF-downregulated embryonic stem cells (mouse and human) showed significantly fewer cells in S phase and correspondingly more G2/M-phase cells.43 We speculate G2 is prolonged in BAF-deficient embryonic stem cells because BAF is also needed during the final (1%) phase64 of DNA replication. Not surprisingly, BAF-downregulated embryonic stem cells also suffer high rates of apoptosis.43
BAF showed no detectable association with any tested HAT or HDAC either in vivo or when tested directly in vitro (unpublished observations). However, purified BAF competitively inhibited HAT-mediated acetylation of recombinant purified H3 and H4 in vitro (unpublished observations), consistent with its direct binding to H341 and H4 (reported here). We therefore propose that BAF (like linker histones H1 and H5)65 might sterically and nonspecifically inhibit access of chromatin regulators to nucleosomes. We further propose BAF also influences the epigenome selectively, by recruiting specific regulators to chromatin in vivo. As precedent, we note that HMGN1, a mobile nucleosome-binding protein, promotes H3-K14 acetylation by helping recruit the histone acetylase PCAF.66,67 This ‘recruitment’ model is based on evidence that BAF associates in vivo with several specific regulators: G9a and SET/I2PP2A (shown here) and PARP1, RBBP4, DDB1/2 and CUL4.42 Our proposal that BAF recruits SET/I2PP2A is consistent with a previous study showing H3-S10 phosphorylation decreased in BAF-downregulated cells.52
The 30% increase in the silencing mark H3-K9-Me2 seen in BAF-overexpressing cells did not reach statistical significance, but is noteworthy since we show here BAF associates directly or indirectly with G9a, one of the enzymes responsible for this mark. This mark helps retain G9a68 and further recruits heterochromatin protein 1 to maintain silencing.5 G9a also cooperates with a specific DNA methyltransferase (Dnmt3) to regulate gene expression during retinal development,69 a process also regulated by BAF.40,70 The developing retina might be an excellent system in which to test for potential BAF-mediated recruitment of G9a as a silencing mechanism.
BAF can repress specific developmentally-regulated promoters in C. elegans.35 Since BAF binds DNA nonspecifically, this promoter-specificity is attributed to protein-protein interactions. For example, BAF directly binds Crx and other ‘paired rule’ homeodomain transcription factors and thereby represses Crx-dependent genes.40 Interestingly, BAF also associates with Sox2, one of three master regulators (Oct4, Nanog, Sox2) that maintain pluripotency in embryonic stem cells.43 Downregulating BAF promoted differentiation of mouse embryonic stem cells and appeared to bias differentiation toward mesodermal or trophectodermal fates.43 This would be a fascinating system in which to explore BAF’s proposed roles as an epigenetic regulator and whether these roles depend on B-type lamins, A-type lamins, LEM-domain proteins or other BAF-binding proteins (e.g., NEMP1).70 Lamin B1 and emerin contact large specific regions of silent chromatin (Lamina-Associated Domains, LADs).11 These regions of the human genome show ~80% overlap with large blocks of G9a-dependent H3-K9-Me2-enriched silent chromatin, known as Large Organized Chromatin K9 modifications (LOCKs).71 Our discovery that BAF associates with nucleosomes and specific chromatin regulators including G9a, raises the hypothesis that BAF may promote chromatin silencing in the context of tethering to the lamina network (lamin filaments, LEM-domain proteins); for example, by inhibiting activation of certain genes, as was observed in mESC where BAF downregulation increased the expression of specific cell lineage genes promoting differentiation.43
Three methyl marks (H3-K4-Me2, H3-K9-Me3, H3-K79-Me2) increased significantly in BAF-overexpressing cells. This ‘pro-methylation’ activity of BAF is intriguing, but defies simple interpretation. For example, H3-K4-Me2 is generally associated with active chromatin, while H3-K9-Me3 correlates with silent chromatin (i.e., constitutive heterochromatin) and H3-K79-Me2, is enriched in both silent and active chromatin loci.72,73 Conversely, the ‘silent’ mark H3-K27-Me1/3 (i.e., facultative heterochromatin) decreased significantly in BAF-overexpressing cells. We note that lamin-network-disrupted HGPS patient cells also show complex epigenetic changes, which include reduced H3-K27-Me3 and increased H4-K20-Me3 (both silent/heterochromatin marks),22 similar to BAF-overexpressing cells. By contrast, silent mark H3-K9-Me3 was reduced in HGPS cells19,20,22 but increased in BAF-overexpressing cells. We suggest epigenetic control in HGPS is perturbed by at least two mechanisms: via loss of BAF epigenetic functions that require intact lamin filament networks and (as lamin networks collapse) via increased levels of ‘free’ BAF that mimic BAF-overexpression. Of note, BAF and lamin A can each associate with the histone chaperone RBBP4,42 which is part of several chromatin remodeling complexes and expression of which is reduced in both premature and normal aging.74
Untangling the epigenetic roles of BAF vs. A- and B-type lamins will be challenging, particularly in HGPS cells where the entire nucleoskeleton is disrupted.20,75 Even though BAF is a small and rapidly-diffusing ‘non-structural’ protein, it is also an essential component of the nuclear lamina network.24,27,29 BAF’s importance to the nucleoskeleton is underscored by the discovery that a recessive BAF mutation causes a hereditary progeroid syndrome.44 Individuals homozygous for the BAF missense mutation A12T express BANF1 mRNA at normal levels, but have greatly reduced levels of BAF protein (~5–10% of normal). Patients with this ‘reduced-BAF’ syndrome partially resemble those with typical HGPS, but in contrast present a longer lifespan, profound skeletal abnormalities and lack of cardiovascular problems. This has important medical and therapeutic implications. At the cellular level, BAF A12T fibroblasts showed inefficient emerin localization at the NE and profoundly abnormal nuclear morphology. Both phenotypes were rescued by ectopic expression of wildtype GFP-BAF,44 consistent with evidence that BAF stabilizes binding between emerin and lamin A.33,76 Our new findings suggest BAF is also an epigenetic regulator. This adds an important new piece to the laminopathy and aging ‘puzzles’ that may help to untangle and distinguish how BAF and its nucleoskeletal partners each contribute to nuclear structure and genome function in specific tissues.
Purified recombinant His-tagged BAF dimers (10 μg; final concentration 1.78 μM) and purified recombinant Xenopus histone H4 (xH4, 100% identical to human H4; 20 μg, final concentration 7.6 μM; kind gift from J. C. Hansen, Colorado State University) were each centrifuged (14,000 rpm, 10 min, 4°C) to remove aggregates, then mixed and gently rotated (1 h, 25°C) in binding buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 10 mM imidazole, 0.1% Triton X-100, 1 mM PMSF and benzonase (EMD Chemicals Inc.) in a final volume of 200 μl. Reactions were supplemented with 15 μl Ni-NTA agarose beads (Qiagen), incubated overnight (rotating, 4°C) and washed extensively with binding buffer. Bound proteins were eluted with 2X-Sample Buffer (2X-SB), resolved by SDS-PAGE and stained with Coomassie Brilliant Blue (CBB).
Mononucleosomes were isolated as described,46 and then immunoprecipitated using antibodies against human BAF (15 μl each of rabbit serum 327324 and rabbit serum 5045,41 per 300 μl reaction) or no antibody as control, overnight at 4°C. Precipitates were resolved by SDS-PAGE (4–12% NuPAGE™ gels; Invitrogen), transferred to nitrocellulose (Schleicher and Schuell Bioscience) and protein gel blotted using BAF serum 3273 (1:10,000 dilution).
HeLa cells were cultured in Dulbecco modified Eagle medium (DMEM) containing 10% fetal bovine serum (GIBCO; Invitrogen). Cells were lysed in immunoprecipitation (IP) buffer (20 mM Hepes pH 8, 150 mM NaCl, 0.1% Triton X-100, 10 mM EDTA, 2 mM EGTA, 2 mM DTT, 1 mM PMSF and benzonase [300 U/μl; EMD Chemicals Inc.]) on ice, 20 min. Lysates were sonicated 3 times (15 sec each) and centrifuged (20 min, 50,000 rpm in a Beckman TL-100 ultracentrifuge, 4°C) to remove insoluble material. Supernatants were incubated (30 min, 4°C) with GammaBind G Sepharose (Amersham Biosciences Corp.), then centrifuged (1,500 rpm, Eppendorf 5415C Microcentrifuge, 4°C) to obtain cleared cell lysates. For each immunoprecipitation, 200 μl cleared lysate was incubated 1 h (4°C) with each indicated antibody: 5–10 μl anti-histone H3 #1791, 10 μl anti-H3-K27-Me1/3 #14222, 4 μl anti-H4-K20-Me #9051 (all from Abcam); 2–4 μl anti-H3-K4-Me2 #07–030, 5 μl anti-H3-K9-Me2 #07–441, 15 μl anti-H3-S10-P #05–817, 1–2 μl anti-H4-Ac5 #06–946, 1–2 μl anti-H4-K5-Ac #07–327 or 4 μl anti-H4-K16-Ac #07–329 (all from Millipore; Billerica, MA) or no serum as control, then supplemented with 15 μl GammaBind G Sepharose (Amersham Biosciences Corp.) and rotated overnight (4°C), pelleted (1,500 rpm, Eppendorf 5415C Microcentrifuge, 4°C) and washed three times with IP buffer. Nonspecifically-associated proteins were competed by two 15 min (37°C) incubations with non-specific peptide (0.1 mM p53 peptide; C. Wolberger, Johns Hopkins University) and washed once with IP buffer. Bound proteins were eluted using 2X-SB, resolved by SDS-PAGE (4–12% NuPAGE™ gels; Invitrogen), transferred and protein gel blotted using rabbit serum 3273 against BAF (1:10,000 dilution) or rabbit anti-histone H3 (1:2,500; Abcam #1791). Images were acquired using VersaDoc 5000 (BioRad) and quantified by densitometry using Quantity One ® Software (BioRad).
Constructs encoding FLAG-PCAF (from P. Puigserver, Dana Farber Cancer Institute), FLAG-p300 and FLAG-Tip60 (from W. Gu, Columbia University), FLAG-HDAC1 (from R. Roeder, Rockefeller University), FLAG-G9a (from M. Stallcup, University of Southern California), FLAG-SET/I2PP2A (from T. Papamarcaki, University of Ioannina, Greece), FLAG-Aurora B (from H. Wang, University Alabama, Birmingham), FLAG-BAF (from P. Traktman, Medical College of Wisconsin), FLAG-H3.1 or the empty FLAG vector (described in ref. 41) were transiently transfected into HeLa cells using LT1 (Mirus) per manufacturer instructions and expressed for 24–48 h. Cells were then lysed by incubating 10 min on ice in lysis buffer (20 mM Hepes pH 8, 300 mM NaCl, 0.3% Triton X-100, 0.2 mM EDTA, 0.2 mM EGTA, 1.5 mM MgCl2, 2 mM DTT, 10% glycerol, 1 mM PMSF and 300 U/μl benzonase [EMD Chemicals Inc.]), then diluted 1:1 v/v with lysis buffer to achieve final concentrations of 150 mM NaCl, 0.15% TX-100 and 5% glycerol, incubated 10 more min on ice, sonicated four times (15 sec each) and centrifuged (50,000 rpm, 30 min, 4°C) to recover soluble proteins for subsequent immunoprecipitation. For each immunoprecipitation reaction 250 μl lysate was incubated 3–4 h (4°C) with 3 μl FLAG-M2 agarose beads (Sigma-Aldrich), then pelleted and washed four times with lysis buffer (150 mM NaCl, 0.15% Triton-X). Nonspecifically-associated proteins were competed by two 15 min (37°C) incubations with a non-specific peptide (0.1 mM Sir2 peptide, from C. Wolberger, Johns Hopkins University) and then washed once with lysis buffer. Bound proteins were eluted with 2X-SB, resolved by SDS-PAGE (4–12% NuPAGE™ gels; Invitrogen) and protein gel blotted with rabbit anti-FLAG (1:7,000; Sigma #F7425), rabbit anti-G9a (1:500; Cell Signaling #3306), rabbit anti-SET (1:300; Abcam #1183), rabbit anti-histone H3 (1:2,500; Abcam #1791) or rabbit serum 3273 against BAF (1:10,000).
An N-terminally His-tagged human BAF (H6BAF) construct76 or corresponding control empty vector, were transiently transfected into HeLa cells using LT1 reagent (Mirus) per manufacturer instructions and expressed 24–48 h. Then new media containing [14C]-sodium acetate (~90–150 μCi final concentration; 57.5 mCi/mmol; Sigma) or an equal volume of solvent (water), plus deacetylase inhibitor (1 μM TSA)77 were added to cells for 4 h. Cells were then harvested, pelleted and 20% of each pellet was lysed and protein gel blotted to quantify BAF expression levels. The remaining cell pellet (80%) was fractionated to generate nuclei; histones were isolated by acid-extraction as described,78 resolved by SDS-PAGE (4–12% NuPAGE™ gels; Invitrogen) and stained with CBB to quantify protein levels of H3, H4 and H2A/H2B (which co-migrated in these gels). Incorporated [14C]-acetate was measured by fluorography, quantified by densitometry (Quantity One ® Software; BioRad) and plotted as the percentage of [14C]-incorporated per total histone, with control cells set to 100%. Significance was determined using a two-tailed Student t-test.
HEK293 cell lines with stable Tet-inducible expression of FLAG-BAF (293:BAF cells) and corresponding control 293:CAT cells were from P. Traktman.50 Cells were cultured in DMEM containing 10% Fetal Bovine Serum (GIBCO; Invitrogen), 100 μg/ml hygromycin (Invitrogen) and 15 μg/ml blasticidin (Invitrogen). Cells were treated 32–52 h with Tet (1.5 μg/ml; Sigma) to induce FLAG-BAF expression and harvested: 20% was used to generate whole cell lysates and 80% was used to acid-extract histones (described above). Whole cell extracts were resolved by SDS-PAGE and protein gel blotted with antibodies specific for BAF (serum 3273; 1:10,000), γ-tubulin (loading control; 1:500,000; Sigma #T6557), α-actin (loading control; 1:10,000; Millipore), HDAC4 (1:1,000; BioLegend), HDAC2 (1:2,000; sc-7899, Santa Cruz Biotechnology, Inc.), G9a (1:500; Cell Signaling #3306), SET (1:300; Abcam #1183), lamins A/C (1:1,000; Millipore #MAB3211), emerin (1:2,000; Santa Cruz #sc-15378) or cyclin E (1:500; Santa Cruz #sc-248). Protein levels were quantified by densitometry (described above) and normalized to γ-tubulin or α-actin. Acid-extracted histones were resolved by SDS-PAGE and protein gel blotted using antibodies specific for the following marks: H3-K4-Me2 (1:10,000; Millipore #07–030), H3-K9-Me2 (1:30,000; Millipore #07–441), H3-K9-Me3 (1:20,000; Abcam #8898), H3-K9/K14-Ac (1:30,000; Millipore #06–599), H3-S10-P (1:5,000; Millipore #05–817), H3-K27-Me1/Me3 (1:20,000; Abcam #14222), H3-S28-P (1:5,000; Millipore #07–145), H3-K79-Me2 (1:30,000; Abcam #3594), H4-K5-Ac (1:10,000; Millipore #06–759), H4-K8-Ac (1:10,000; Millipore #06–760), H4-K12-Ac (1:10,000; Millipore #06–761), H4-K16-Ac (1:10,000; Millipore #06–762), H4-K20-Me (1:30,000; Abcam #9051) and H4-Ac5 (1:30,000; Millipore #06–946). Results for each mark specific antibody were quantified by densitometry as described above, normalized to each Ponceau-stained H3 or H4 band and plotted as a percentage of the corresponding 293:CAT signal. Significance was determined using a two-tailed Student t-test.
Asynchronous 293:BAF or 293:CAT cells were treated with Tet (1.5 μg/ml) to induce FLAG-BAF expression for 35 h. The distribution of cells in G0/G1, S and G2/M phases was estimated for each sample from histograms of DNA content using the ModFit LT software (Becton Dickinson and Co.). This program fit a binomial curve to the G0/G1 and G2/M peaks and determined the percentage of cells in S phase by subtracting the percentage of cells in G0/G1 and G2/M. To analyze DNA content, cells were stained with 50 μg/ml propidium iodide (Sigma #P4170) plus 30 U/ml RNase A. Data was collected using a FACScan or FACScalibur instrument (Becton Dickinson and Co.). For each sample, cell aggregates were gated out and 10,000 cell events were analyzed.
To specifically measure the percentage of cells in S phase, samples were pulsed with 30 μM 5-Bromo-2’-Deoxyuridine (Sigma #B5002) for 30 min, then washed, harvested, fixed, labeled with anti-BrdU and analyzed by flow cytometry. Briefly, we resuspended 106 cells in denaturating buffer (2 M HCl in PBS/0.5% TritonX-100) for 30 min (25°C), then pelleted and resuspended in neutralizing buffer (0.1 M sodium tetraborate in PBS, pH 8.5), washed once with PBS and incubated ~1 h (37°C) with FITC-conjugated mouse anti-BrdU (#556028; BD Biosciences) in PBS/3% BSA/0.05% Tween. DNA was propidium iodide stained as above.
Harvest of cells for incubation with mouse anti-MPM-2 (1:200; #05–368) primary antibody from Millipore and analysis by flow cytometry was as previously described.79 FITC-conjugated goat anti-mouse (1:1,000; #115–095–146) secondary antibody was from Jackson Laboratories. The percentage of MPM-2 positive cells was determined using CellQuest version 3.3 software (Becton Dickinson and Co.).
Supplementary PDF file supplied by authors.
We are grateful to P. Traktman (Medical College of Wisconsin) for the 293:BAF and 293:CAT cell lines; R. Stolle for purified BAF; M. J. Eddins and K. Fahie (Johns Hopkins School of Medicine) for chromatography help and M. J. Eddins for Figure 1E. We thank C. Lerin and P. Puigserver (Dana Farber Cancer Institute), I. Celic and J. Boeke (Johns Hopkins School ofMedicine), W. L. Kraus (Cornell), J. Th’ng (Northwestern Ontario Regional Cancer Centre) and C. Slawson and L. Blosser (Johns Hopkins Flow Cytometry Facility) for reagents and advice. We thank W. Gu (Columbia), R. Roeder (Rockefeller), M. R. Stallcup (Univ. Southern California), T. Papamarcaki (Univ. Ioannina) and H. Wang (Univ. Alabama) for constructs and S. Taverna and the Wilson lab for insightful discussions. This work was funded by American Heart Association Predoctoral fellowship 0615601U (R.M.) and National Institutes of Health RO1 GM48646 (K.L.W.).
The authors have declared that no competing interests exist.
Previously published online: www.landesbioscience.com/journals/nucleus/article/17960