Report

Acetylation by Tip60 Is Required for Selective Histone Variant Exchange at DNA Lesions

See allHide authors and affiliations

Science  17 Dec 2004:
Vol. 306, Issue 5704, pp. 2084-2087
DOI: 10.1126/science.1103455

Abstract

Phosphorylation of the human histone variant H2A.X and H2Av, its homolog in Drosophila melanogaster, occurs rapidly at sites of DNA double-strand breaks. Little is known about the function of this phosphorylation or its removal during DNA repair. Here, we demonstrate that the Drosophila Tip60 (dTip60) chromatin-remodeling complex acetylates nucleosomal phospho-H2Av and exchanges it with an unmodified H2Av. Both the histone acetyltransferase dTip60 as well as the adenosine triphosphatase Domino/p400 catalyze the exchange of phospho-H2Av. Thus, these data reveal a previously unknown mechanism for selective histone exchange that uses the concerted action of two distinct chromatin-remodeling enzymes within the same multiprotein complex.

DNA double-strand breaks (DSBs) are a deleterious type of DNA damage leading to chromosomal breakage. Cells have developed mechanisms to detect and repair DSBs (1, 2), which must access nucleosomal DNA. Two classes of activities regulate the accessibility of DNA by either covalently modifying histones or using adenosine triphosphate (ATP) hydrolysis to catalyze histone mobilization (3, 4). Current knowledge suggests that covalently modified histones can create specific interaction sites for regulatory proteins and complexes (5, 6).

Incorporation of histone variants into nucleosomes provides another mechanism for altering chromatin structure (7). Whereas the major histones are assembled into nucleosomes during DNA replication, histone variants can be incorporated into chromatin in a replication-independent manner (810). An example of such an activity is the yeast Swr1p ATPase complex, which catalyzes the exchange of H2A for the variant H2A.Z in nucleosomes (911).

Histone modifications can mark distinct chromatin locations. H2A.X, an essential mammalian histone variant required for genomic stability, becomes phosphorylated at sites of DSBs by conserved DNA damage–recognizing factors (12, 13). Like H2A.X, H2A and H2Av become phosphorylated at DSBs in yeast and flies, respectively (14, 15). Because repair requires access to DNA, it has been suggested that this phosphorylation might attract chromatin-remodeling complexes to DSBs (16). The removal of phospho-H2A.X is replication-independent and could be catalyzed by the same complexes. DSBs accumulate upon inactivation of the human Tip60 complex, implicating it as one candidate for a chromatin-remodeling complex with a role in DNA repair (17).

We demonstrate that the Drosophila dTip60 multiprotein complex catalyzes exchange of phospho-H2Av with unmodified H2Av. This reaction is catalyzed by two chromatin-dependent enzymes within the dTip60 complex: the histone acetyltransferase dTip60 and the ATPase Domino. These factors sequentially acetylate and then replace nucleosomal phospho-H2Av with H2Av from within the dTip60 complex.

The dTip60 complex was purified from Drosophila S2 cells. dPontin, the fly homolog of a subunit of the human Tip60 complex (17), was epitope-tagged with a hemagglutin (HA)-Flag tag at the C terminus (fig. S1). The dPontinHAFlag-associated proteins were isolated from nuclear extracts by sequential Flag- and HA-affinity purification followed by a glycerol gradient (fig. S1). Peak fractions of dPontin-HAFlag, dTip60, and Domino were identified by immunoblotting (Fig. 1A) and assayed for histone acetyltransferase activity (fig. S2). Several polypeptides that copurified with dPontinHAFlag (fig. S1) were identified by multidimensional protein identification technology (MudPIT) (18). We identified polypeptides with homology to all 16 subunits of the human Tip60 complex (table S3). This analysis also revealed a substantial number of tryptic peptides from histones H2Av and H2B but not from other histones.

Fig. 1.

Characterization of the dTip60 complex (also fig. S1). (A) Western blots of 10% to 30% glycerol gradient fractions. i; input of HA eluates. Top numbers indicate fraction numbers. Higher numbers represent lower molecular masses. (B) Western analysis of dTip60 peak fractions from glycerol gradients. (C) Western analysis of polypeptides co-immunoprecipitated with dTip60 from S2 nuclear extracts. dPon:hf and fh:dRep are cells that expressed dual HA- and Flag-tagged dPontin and dReptin. The proteins were detected with antibodies against the HA epitope. pre; pre-immunesera, is; immunesera against dTip60.

We raised antibodies against dTip60, dMrg15, dTra1 (19), dGas41, dIng3, and E(Pc). These antisera and antibodies against Domino, H2Av, and H2B were used in immunoblotting of gradient peak fractions and anti-dTip60 immunoprecipitates from nuclear extracts to confirm that these proteins are part of the dTip60 complex (Fig. 1, B and C). dPontin-HAFlag stably associated with all dTip60 complex subunits examined, including dReptin, the fly homolog of the human Tip60 complex component Tip49b (Fig. 1C) (17). Histones H2Av and H2B stably associated with the dTip60 complex, whereas histone H2A and other histones were not detected (Fig. 1, B and C) (20).

Tip60 complexes function in DSB repair and contain the ATPase Domino/P400 and H2Av/H2B heterodimers. Because H2Av becomes phosphorylated at sites of DSBs (15), we tested whether dTip60 complex remodeled nucleosomes containing phospho-H2Av. We assembled recombinant Drosophila nucleosomes containing H2Av with a point mutation that mimicked phosphorylation at Ser137 (Ser137 to Glu137; H2AvE) (Fig. 2A and fig. S4) (21). Upon incubation with the dTip60 complex, recombinant H2AvFlag/H2B heterodimers, acetyl-coenzyme A (acetyl-CoA), and ATP, a transfer of H2AvFlag to the nucleosomal arrays was observed (Fig. 2B, lanes 7 to 9, and figs. S5 and S6). The transfer reaction proceeded rapidly (notable amounts of H2AvFlag were incorporated within 5 min; lane 7) and depended on the presence of nucleosomes (no transfer onto free DNA occurred; lane 2). Although relatively small amounts of H2AvFlag were transferred in the absence of ATP and/or acetyl-CoA (lanes 4 to 6), it was about seven times more efficient in the presence of both cofactors (lane 9). Addition of a nonhydrolyzable ATP analog (γS-ATP; lane 10) reduced the background activity of the complex. The dTip60 complex was highly selective for incorporation of H2Av into H2AvE-containing nucleosomal arrays (Fig. 2C). No H2AvEFlag was incorporated into nucleosomes containing H2Av (lanes 1 and 2), and no significant release of H2AvFlag was observed from nucleosomal arrays in the presence of H2AvEFlag/H2B heterodimers (lanes 3 and 4; also fig. S7). Time course experiments revealed that the presence of acetyl-CoA enhanced the transfer speed and the quantity of H2Av incorporation (Fig. 2D, compare lanes 6 and 10). The incorporation rate of H2AvFlag into the nucleosomal arrays was unchanged when acetyl-CoA only was temporarily added to the exchange reactions and removed before the addition of heterodimers (Fig. 2E, lanes 3 and 4). This strongly suggests that the acetylation of the nucleosomal arrays by the dTip60 complex, but not of heterodimers, is crucial for optimal H2Av exchange.

Fig. 2.

The dTip60 complex exchanges phospho-H2Av from nucleosomal arrays with H2Av. (A) Experimental scheme. (B) Western analysis of nucleosomal arrays probed for incorporated H2AvFlag (H2Av:f) with antibodies against Flag. Input is 20% of the H2Av:f/H2B heterodimers used in each reaction. Added substrates and incubation times are indicated at the right. γS-ATP, adenosine-5′-O-(3-thiotriphosphate). (C) The dTip60 complex preferentially incorporates H2Av into phospho-H2Av-containing nucleosomes but not vice versa. Arrays, nucleosomal arrays; sup, supernatants. Roman numerals indicate the combinations of heterodimers and arrays used in the reaction as indicated to the right [the exposure shown is part of the same exposure and blot shown in (B)]. (D) Histone exchange by the dTip60 complex is enhanced by the presence of acetyl-CoA. For details, see (B). (E) The acetylation of nucleosomes is essential for optimal histone exchange. H2Av/H2B heterodimers were added either after preacetylating of nucleosomal arrays containing H2AvE for 30 min (lane 3) or simultaneously with AcCoA (lane 4). No significant difference in amounts of H2Av:f incorporation was observed.

To examine the acetyltransferase specificity of the dTip60 complex, we used different combinations of recombinant histones as substrates in histone acetyltransferase (HAT) assays. In the presence of core histones, H2A, H2Av, and H2AvE were acetylated at equally low levels (Fig. 3A). However, in a nucleosomal context, acetylation of H2AvE was significantly increased over that observed for all other histones (Fig. 3B). This confirms that the dTip60 complex preferentially targets and acetylates phospho-H2Av in nucleosomes. In fact, Lys5 of histone H2Av is acetylated by the dTip60 complex (Fig. 3C). As individual monomeric histones, H2A, but not H2Av or H2AvE, was the preferred substrate of the dTip60 complex (fig. S7). By contrast, acetylation was about equal between H2A and H2Av when heterodimers with H2B were assayed, whereas acetylation of H2AvE was unchanged (fig. S7). Thus, dTip60 complex prefers H2Av-containing heterodimers over those containing H2AvE.

Fig. 3.

The dTip60 complex preferentially acetylates phospho-H2Av in nucleosomes. Top panels show fluorographies (F) indicating incorporation of tritiated acetyl-CoA. Bottom panels show Coomassie (C) Blue R250–stained SDS–polyacrylamide gel electrophoresis. (A) HAT activity of dTip60 complex using core histones. Labels on top indicate the various forms of H2Av incorporated into core histones. (B) H2AvE is the preferred substrate of the dTip60 complex when incorporated into nucleosomes. Labels on top indicate the H2A forms used for nucleosome reconstitutions. Note that histone H4 is not the preferred substrate of Tip60-type complexes in recombinant nucleosomes. (C) Antibodies against H2A(acK5) recognize acetyl-K5 of H2AvE. Recombinant nucleosomal arrays were acetylated by the dTip60 complex. Antibodies against H2A(acK5) fail to recognize acetylated arrays containing H2AvE:f (K5->A). Membrane was probed with antibodies against Flag as loading control.

Upon induction of DSBs, phospho-H2Av rapidly accumulates on chromatin with peak amounts after 10 to 15 min (15). During the course of DNA repair, this phosphorylation becomes undetectable within 180 min. The dTip60 complex acetylates and removes phospho-H2Av from nucleosomes in vitro. Thus, we tested whether removal of phospho-H2Av during repair was dependent on dTip60 complex in vivo. We depleted dTip60 or dMrg15 from S2 cells by RNA interference (RNAi) (Fig. 4A) (18). These cells were exposed to γ irradiation to induce DSBs, and the nucleosomal histones were extracted after 0, 15, and 180 min. The amounts of H2Av and phospho-H2Av were compared by immunoblotting (Fig. 4B). In mock-treated cells, phospho-H2Av levels peaked after 15 min and were undetectable after 180 min (Fig. 4B). By contrast, phospho-H2Av levels remained high in cells depleted for either dTip60 or dMrg15. To confirm these findings in embryos, we generated a null allele of dMrg15 (18) and tested phospho-H2Av levels after γ irradiation. Again, the levels of phospho-H2Av remained higher in dMrg15 mutants than in wild-type embryos (Fig. 4C and fig. S8).

Fig. 4.

Loss of dTip60 complex leads to the accumulation of phospho-H2Av upon DNA damage and abolishes DSB-dependent transient acetylation of phospho-H2Av. (A) RNAi assays for lacZ, dTip60, and dMrg15. Western blots of extracts from dsRNA-transfected cells. (B) Western blots of chromatin extracts from dsRNA-treated cells. After RNAi-treatment, the cells were γ-irradiated [50 gray (Gy)] and harvested at given time points. The blots were probed for H2Av and phospho-H2Av (H2AvS137P). (C) Close-up of whole-mount embryos immunolabeled for phospho-H2Av (red) 0, 15, and 180 min after γ irradiation. The DNA was counterstained with 4′,6′-diamidino-2-phenylindole (blue). Left images are wild-type embryos; right images are dMrg15 mutants (also fig. S8). (D) Chromatin extracts of dsRNA-treated and γ-irradiated cells were probed with antibodies against H2A(acK5) and H2Av. (E) Magnification of whole-mount embryos stained with antibodies against H2A(acK5) (red; DNA in blue; also fig. S9). For details, see (C). White bars indicate 10 μm.

Because the dTip60 complex acetylated nucleosomal phospho-H2Av in vitro, we tested dependence of H2Av acetylation on dTip60 complex components in vivo. We probed chromatin extracts from γ-irradiated double-stranded RNA (dsRNA)–treated S2 cells as well as dMrg15 mutant embryos with antibodies against H2A(acK5), which recognized H2Av(acK5) (Fig. 3C). We detected transient acetylation of a protein band that exhibits the migratory properties of phospho-H2Av (Fig. 4D). This acetylation was most prominent 15 min after γ irradiation and was not detected in extracts of cells lacking dTip60 or dMrg15. Similar observations were made by immunolabeling dMrg15 mutant embryos (Fig. 4E and fig. S9). We conclude that the dTip60 complex acetylates nucleosomal phospho-H2Av at Lys5 in a DSB-dependent manner.

The Drosophila dTip60 complex is structurally homologous to its human counterpart (Fig. 1 and table S3) (17, 22, 23). Both complexes share factors that are linked to cancer, transcription, and DNA repair, including Pontin, Reptin, Mrg15, Tra1, E(Pc), Gas41, and Tip60. We also identified the histone variant H2Av within the Drosophila dTip60 complex. The human Tip60 complex is essential for DSB repair and regulation of apoptosis, two processes that have been linked to histone H2Av in flies (15, 17). Also the yeast NuA4 complex appears to accumulate at DSBs (24).

We demonstrated that the Drosophila dTip60 complex acetylates nucleosomal phospho-H2Av and exchanges it with an unmodified H2Av (Figs. 2 and 3 and figs. S4 to S6). The histone-exchange reaction catalyzed by the ATPase Domino is enhanced by dTip60-mediated acetylation of nucleosomal phospho-H2Av. It appears likely that phospho-H2Av recruits the dTip60 complex to DSBs to facilitate chromatin remodeling during DNA repair. In yeast, the DNA damage–dependent H2A kinase Mec1 genetically interacts with subunits of the NuA4 complex (21, 25), and cells missing NuA4 subunits are sensitive to DSB-inducing agents (25, 26). The physiological roles of the dTip60-mediated phospho-H2Av removal at sites of DSBs could not be clearly separated from a potential function of this complex in DSB repair because of the intimate temporal link between DSB repair and phospho-H2Av clearance (20). However, the overexpression of phospho-H2Av did not induce G2/M arrest or affect DSB-dependent G2/M arrest (fig. S10) (14, 21), suggesting that this signal is not sufficient for damage checkpoint control.

The loss of human Tip60 leads to the accumulation of DSBs and is linked to a growing number of cancer types (26, 27). The histone variant H2A.X is essential for genomic stability and a candidate tumor suppressor (13, 28, 29). Thus, our findings help to understand the functional link between DNA damage–dependent H2A.X phosphorylation and the role of Tip60-type complexes during DSB repair in chromatin.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1103455/DC1

Materials and Methods

Figs. S1, S2, S4 to S10

Table S3

References and Notes

References and Notes

View Abstract

Navigate This Article