Deacetylase Activity Associates with Topoisomerase II and Is Necessary for Etoposide-induced Apoptosis*

DNA topoisomerase II (topo II) is a ubiquitous nuclear enzyme that is involved in DNA replication, transcription, chromosome segregation, and apoptosis. Here we show by immunoprecipitation, pull down with glutathioneS-transferase fusion proteins, and yeast two-hybrid analysis that both topo IIα and -β physically interact with the histone deacetylase HDAC1. The in vitro DNA decatenation activity of recombinant topo IIα and -β is inhibited by association with catalytically inactive, recombinant HDAC1. We provide evidence for the in vivo significance of the topo II-HDAC1 association, showing that inhibition of HDAC activity with trichostatin A suppresses apoptosis induced by the topo II poison etoposide, but not by the topoisomerase I inhibitor camptothecin. We suggest that chromatin remodeling by an HDAC-containing complex facilitates both topo II-catalyzed DNA rearrangement and etoposide-induced DNA damage in vivo.

For completion of cell division, the DNA of replicated chromosomes must be disentangled to allow the segregation of sister chromatids. In humans, this is achieved by the unique decatenation activity of DNA topoisomerase II (topo 1 II). Topo II is essential for normal and neoplastic cellular proliferation, and several common anti-cancer drugs exert their cytotoxic effects through this enzyme (1,2).
Topoisomerase II activity in mammalian cells has been attributed to at least two isoforms. Topo II␣ (p170) associates with chromosomes during prophase and throughout mitosis and is thought to be a major component of the nuclear scaffold (3,4). It has a peak of expression during G 2 /M of the cell cycle (5). In contrast, the closely related topo II␤ (p180) isoform is thought to have a more general role in DNA metabolism, with expression levels that remain relatively constant during cell and growth cycles (5). Both isoforms interact with the C-terminal region of the tumor suppressor protein, p53 (6). p53 is a component of a multiprotein complex that contains the histone deacetylase HDAC1 and the corepressor Sin3a (7)(8)(9)(10)(11).
HDAC1, and the closely related HDAC2, are both components of two separate multiprotein complexes. The NuRD/Mi-2 repression complex contains both nucleosome remodeling and histone deacetylase activities (12), whereas the Sin3 complex contains only the latter (9). Both complexes contain the Rbassociated proteins RbAp46 and RbAp48 and associate with various, sometimes DNA-binding, transcriptional repressor and corepressor proteins (11). The Xenopus NuRD complex (which contains homologues of mammalian HDAC1, RbAp48, and the methyl-CpG-binding protein MBD3) copurifies with DNA topoisomerase II (13), raising the possibility that mammalian topo II isoforms and HDAC1 may interact in a multiprotein complex.
Here we show that HDAC1 and DNA topoisomerase II isoforms physically interact both in vivo and in vitro. We also show that the HDAC inhibitor, TSA, suppresses apoptosis induced by the topo II poison etoposide, but not by the topo I inhibitor camptothecin. Our results raise the interesting possibility that chromatin remodeling by a topo II-HDAC-containing complex is involved in topo II-catalyzed DNA rearrangements and/or generation of etoposide-induced DNA strand breaks in vivo.

MATERIALS AND METHODS
Cells, Reagents, and Materials-The human cell lines HL-60 (promyelocytic leukemia; p53 null) and HeLa were grown in RPMI 1640 medium containing 8% fetal calf serum. Regions of HDAC1 cDNA were subcloned into the pGEX3T-4 family of vectors (Amersham Pharmacia Biotech) and verified by sequencing. GST fusion proteins were purified essentially as described previously (6). Recombinant human DNA topoisomerase II␣ and -␤ were made in a yeast system and purified as described previously (14). Characterization and use of rabbit polyclonal antibodies against topo II␣ (18511␣) and topo II␤ (18513␤) are described elsewhere (15). A polyclonal rabbit antibody against mammalian HDAC1 was raised against a synthetic peptide corresponding to amino acid residues 467-482 and affinity-purified as described previously (16). Antibody against topo I was obtained commercially (Topo-Gen, number 2012).
Immunoprecipitations, in Vitro Binding Assays, and Western Blot Analysis-HeLa whole cell extract was prepared by lysing cells in incubation buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, 5 mM EGTA, 10% (v/v) glycerol) containing 1.0% (v/v) Nonidet P-40, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride and "Complete Mini TM " tablets, Roche Molecular Biochemicals) and 50 units of DNase I (Amersham Pharmacia Biotech) per 10 8 cells. The lysate was incubated in ice for 10 min, and the clarified supernatant was used in standard immunoprecipitations as described previously (6,17). To confirm specificity, cognate blocking peptide (10 g) was incubated with the antibody for 30 min before the addition of extract. Preimmune serum and irrelevant antisera were used as controls. GST pull down experiments used equivalent amounts of GST fusion proteins prebound to glutathione-Sepharose beads (Amersham Pharmacia Biotech) as described previously (6). Interactions with recombinant topo II␣ were performed in incubation buffer containing 0.1% (v/v) Nonidet P-40.
Yeast Two-hybrid Assays-Yeast strains CG-1945 from a Matchmaker Two-Hybrid System II kit (CLONTECH) were transformed with appropriate binary combinations of constructs containing the GAL4 DNA-binding domain and the GAL4 activation domain, as recommended by the manufacturers. HIS3 reporter gene expression was assayed on plates (6), in the presence of 25 mM 3-amino-1,2,4-triazole to suppress background growth (18).
Detection of Apoptosis-HL-60 cells were grown until in mid-log phase, then treated with 100 nM (30 ng/ml) TSA for 0.5 h before additional treatments with either 100 M (59 g/ml) etoposide or 5.8 M (2 g/ml) camptothecin for 1.5 h. Control samples were treated with the dilution vehicles (0.1% Me 2 SO and 0.1% ethanol). All cells were observed in situ with phase-contrast microscopy to count cells with an apoptotic morphology, after staining with 10 M Hoechst 33342 with the addition of 0.1 M propidium iodide to visualize necrotic cells. Cells were also labeled with either FITC-annexin V conjugate (PharMingen) or with the FAM-VAD-FMK reagent provided in the CaspaTag TM fluorescein caspase activity kit (Intergen). Labeled cells were detected by indirect fluorescence microscopy (all cells) or by FACS analysis on a Coulter Epics flow cytometer.

RESULTS
Interaction of HDAC1 and DNA Topoisomerase II-HeLa whole cell extract was immunoprecipitated with affinity-purified antibody against mammalian HDAC1 and precipitated material tested for the presence of topo II␣ by Western blotting (Fig. 1A). Anti-HDAC1 brought down easily detectable amounts of topo II␣. There was no detectable immunoprecipitation of topo II␣ with preimmune serum, and immunoprecipitation was completely abolished by inclusion in the incubation mix of the peptide used to raise the anti-HDAC1 antibody (Fig.  1A). The anti-HDAC1 antibody did not immunoprecipitate detectable levels of topo II␤ (data not shown), and antibody to topo II␤ brought down only a comparatively small amount of the ␣ isoform (Fig. 1A). However, antisera against both topo II␣ and topo II␤ immunoprecipitate 6 -9% of total deacetylase activity from HeLa whole cell extract (Fig. 1B). The activity is fully inhibited by TSA. Negative control immunoprecipitations with preimmune serum, an irrelevant antibody (anti-CDK7), or an antibody against DNA topoisomerase I (topo I) did not bring down activity above that of the no-antibody control.
We performed in vitro pull down experiments of endogenous protein with GST fusion proteins. Full-length mammalian HDAC1, tagged with a GST moiety, but not GST itself, bound endogenous topo II␣ in whole cell extract ( Fig. 2A). In the converse experiment, a GST fusion protein containing the Cterminal domain (CTD) of topo II␣ was able to pull down endogenous HDAC1 (Fig. 2B). The fusion protein of the CTD of topo II␤ was also able to pull down small amounts of HDAC1 (Fig. 2B). Fusion proteins of the CTD of topo II␣ and topo II␤ were able to pull down between 9 and 11% of deacetylase activity (Fig. 2C), comparable with the amounts brought down by immunoprecipitation (Fig. 1B). Whereas GST-topo II␤ pulls down less HDAC1, as detected on Western blots, than comparable amounts of GST-topo II␣ (Fig. 2B), the two different fusion proteins bring down similar amounts of deacetylase activity (Fig. 2C). A possible explanation for this quantitative discrepancy is that other deacetylases, in addition to HDAC1, are preferentially associated with topo II␤.
GST fusion proteins containing the C-terminal domain of HDAC1 interact with recombinant topo II␣ (Fig. 3). This domain has previously been shown to contain the LXCXE motif (residues 414 -418), that appears to mediate interactions with the retinoblastoma protein pRb (19). In contrast, an N-terminal HDAC1 fusion protein, containing the catalytic site, showed minimal interaction with recombinant topo II␣ (Fig. 3).
A yeast two-hybrid system (18) was used to test for direct in vivo interaction between topo II and HDAC1. Inserts were constructed to express the topo II␣ and topo II␤ C-terminal domains (6) and the HDAC1 region 220 -482 (Fig. 3). Expression of the integrated, GAL4-dependent HIS3 reporter gene was used to detect interactions between "bait" and "prey" proteins in vivo. Topo II␣ CTD or topo II␤ CTD as bait, together with HDAC1 as prey, allowed growth of large colonies (over 2 mm diameter) on His-selective medium. All three proteins were ineffective when expressed individually (Fig. 4).
To explore the biological significance of the topo II-HDAC1 interaction, we tested the ability of full-length recombinant HDAC1 to modulate the functional properties of recombinant topo II␣ and -␤. Both of these enzymes can decatenate kinetoplast DNA (kDNA) to minicircle monomers, a process that requires a double-stranded break in the kDNA to allow strand passage. The addition of increasing amounts of HDAC1 to the reaction decreases the decatenation of kDNA by topo II␣ and -␤ (Fig. 5). Addition of GST alone did not affect decatenation by either topo II␣ and -␤.
Suppression of Etoposide-mediated Apoptosis by the HDAC Inhibitor Trichostatin A-We tested the effect of the HDAC inhibitor trichostatin A (20) on apoptosis induced by the chemotherapeutic agent etoposide (VP-16). Etoposide causes topoisomerase II-mediated DNA damage by increasing the steady-state concentration of covalent DNA cleavage complexes (1,2,4). Cells treated with etoposide acquire an apoptotic morphology, notably the condensation of chromatin at the nuclear periphery and blebbing of the plasma membrane (2,21). HL-60 cells displayed apoptotic chromatin condensation after only 1.5-h treatment with either 100 M etoposide or 5.8 M camptothecin, an inhibitor of topo I (Fig. 6A). Plasma membrane changes during early apoptosis include the exposure of phosphatidylserine to the external cellular environment (22). This change was measured by binding of FITC-conjungated annexin V and counting of labeled cells by fluorescence microscopy (Fig. 6B). Activation of cysteine aspartyl proteases (caspases) (21) during the apoptosis of HL-60 cells was assayed with a fluorescent substrate and FACS analysis of viable cells (Fig. 6C). Chromatin condensation, membrane changes, and caspase activation all demonstrated that prior treatment with 100 nM TSA suppresses the apoptotic effect of etoposide (Fig. 6, A-C). In contrast, TSA did not affect apoptosis induced by the topo I inhibitor camptothecin (Fig. 6, A-C). Note that topo I does not associate with detectable amounts of deacetylase activity (Fig. 1B). An identical anti-apoptotic effect of TSA treatment was also observed for the human lung adenocarcinoma cell line H1299 and HeLa cells (data not shown). DISCUSSION The results presented show that the histone deacetylase HDAC1 is physically associated with each of the two isoforms of human topoisomerase II, topo II␣ and topo II␤. The association occurs in vivo, being detectable by coimmunoprecipitation from human cell extracts and by yeast two-hybrid assay. It also FIG. 4. Yeast two-hybrid assay showing that the CTDs of topo II␣ and topo II␤ can interact with HDAC1 CTD (amino acids 220 -482) in vivo. Expression of the reporter gene HIS3 in yeast stain CG-1945 (CLONTECH) was determined by two parallel series of spot assays on selective medium plates lacking tryptophan, leucine, and histidine (ϪTrp, ϪLeu, ϪHis), but in the presence of 25 mM 3-amino-1,2,4-triazole to suppress background growth (6,18). Colony size was compared with that on plates lacking tryptophan and leucine (ϪTrp ϪLeu) as control. Strong growth in ϪHis medium occurs only in cells in which the bait and prey proteins physically interact. occurs in vitro. GST-coupled recombinant topo II␣ and topo II␤ pull down significant amounts of HDAC activity from cell extracts, while recombinant HDAC1 inhibits the in vitro decatenation activity of recombinant topo II␣. Since completion of the work reported here, Tsai et al. (23) have reported essentially the same findings for the two very similar deacetylases HDAC1 and HDAC2. Interestingly, whereas Tsai et al. (23) find evidence for an interaction between topo II␣ and various regions of HDAC2, including N-terminal residues 1-57, we find that only the C-terminal region of HDAC1 (residues 220 -482) interacts with topo II in vitro. These two deacetylases seem to differ in their mode of interaction with topo II.
In experiments to assess the biological significance of the topo II-HDAC interaction, we analyzed the effect of the deacetylase inhibitor TSA on processes known to require topo II activity. The most striking effect so far has been on the ability of the topo II poison etoposide to drive cells into apoptosis. We show that treatment with TSA prior to the addition of etoposide suppresses apoptosis in a variety of cell lines. The effect is seen even with HL60 cells, in which apoptosis is detectable within less than 1 h, a finding that minimizes the probability that inhibition of apoptosis is due to pleiotropic effects of TSA, such as its ability to alter cell cycle progression. The inhibitory effect of TSA was detected in several p53-null cell lines, so the interaction between HDAC1 and p53 (7) cannot be responsible. Microscopically detectable chromatin remodeling is a diagnostic characteristic of cells in the later stages of apoptosis, and recent reports indicate that both topo II and histone acetylation play a role in this process (24). However, our results indicate that this is not the stage at which TSA exerts the inhibitory effect reported here. We have shown that TSA inhibition is detectable even when using an assay that measures one of the earliest changes of apoptosis, namely the alteration in membrane phospholipids detected by binding of annexin V (22). These findings argue that TSA is acting at a relatively early stage in apoptosis, prior to the onset of major changes in nuclear ultrastructure. It remains possible that TSA also effects more subtle chromatin changes, possibly those determining expression of genes required for progression through apoptosis (25). These effects are not mutually exclusive. Indeed, recent results indicate that both topo II and changes in acetylation act at various stages in the pathways by which cells progress through apoptosis (24, 26 -28).
In attempting to explain the effect of TSA on etoposideinduced apoptosis, it is important to note that etoposide is a topo II poison that blocks the enzyme after DNA cleavage but prior to strand passage (4). Covalent topoisomerase-DNA cleavage complexes accumulate in the presence of such poisons. DNA replication, transcription, or helicase activity all disrupt these complexes, releasing the DNA double-strand breaks that can precipitate apoptosis (1,4). Reducing either the accumulation of topo II-DNA complexes, or their breakdown, will both reduce DNA damage and hence apoptosis. A possible explanation for the results presented here is that HDAC-dependent chromatin remodeling is necessary for the initiation of topo II-catalyzed DNA rearrangement or for dissociation of topo II-DNA complexes, or both. If this were the case, then HDAC inhibitors such as TSA would be expected to prevent the appearance of DNA damage in the presence of topo II poisons, and consequent progression into apoptosis, exactly as we have found. Crucially, TSA has no effect on apoptosis induced by the topo I inhibitor camptothecin. We show here that topo I is not associated with HDAC1. Further support for these ideas comes from the recent results of Tsai et al. (23), who show that topo II is associated not only with HDAC1/2, but also with MTA2, a protein that is part of the NuRD chromatin remodeling complex. The NuRD complex contains both HDAC1/2 and Mi-2, a protein with ATPase/helicase activity (12).
We and others (23) find that HDAC1 can inhibit the catalytic activity of topo II in vitro. This is an important confirmation of the ability of topo II and HDAC1 to interact, but is not, at first sight, consistent with the proposition outlined above that HDAC activity facilitates topo II catalysis, or its consequences. This can be resolved by noting that the GST-HDAC1 construct used to inhibit topo II in vitro is catalytically inactive, presumably because it lacks essential protein partners such as RbAp46/48 (8,11). It would be wrong to assume that catalytically active HDAC1, in the context of a multiprotein complex that includes both HDAC and chromatin remodeling activities (23), is also inhibitory. It might even be the case that, in vivo, topo II activity is inhibited only by association with HDAC rendered catalytically inactive by inhibitors such as TSA. Such an effect would complement the suppression of topo II activity brought about by inhibition of chromatin remodeling.