Resveratrol: A novel type of topoisomerase II inhibitor

Resveratrol, a polyphenol found in various plant sources, has gained attention as a possible agent responsible for the purported health benefits of certain foods, such as red wine. Despite annual multi-million dollar market sales as a nutriceutical, there is little consensus about the physiological roles of resveratrol. One suggested molecular target of resveratrol is eukaryotic topoisomerase II (topo II), an enzyme essential for chromosome segregation and DNA supercoiling homeostasis. Interestingly, resveratrol is chemically similar to ICRF-187, a clinically approved chemotherapeutic that stabilizes an ATP-dependent dimerization interface in topo II to block enzyme activity. Based on this similarity, we hypothesized that resveratrol may antagonize topo II by a similar mechanism. Using a variety of biochemical assays, we find that resveratrol indeed acts through the ICRF-187 binding locus, but that it inhibits topo II by preventing ATPase domain dimerization rather than stabilizing it. This work presents the first comprehensive analysis of the biochemical effects of both ICRF-187 and resveratrol on the human isoforms of topo II, and reveals a new mode for the allosteric regulation of topo II through modulation of ATPase status. Natural polyphenols related to resveratrol that have been shown to impact topo II function may operate in a similar manner.

The rigorous study of natural products has been tremendously successful in providing leads for drug discovery, with 34% of small-molecule therapies approved by the United States Food and Drug Administration (FDA) between 1981 and 2010 consisting of natural products or natural-product derivatives (1). However, the use of natural products in the form of dietary supplements, which are less stringently regulated by the FDA, can pose a public health risk (2). Despite a lack of scientific consensus on the safety and efficacy of many available dietary supplements, the multi-billion dollar supplement industry continues to expand at a significant rate (3).
Resveratrol, a polyphenol found in berries, nuts, traditional Asian medicines, and red wine, is one example of a popular natural supplement that is widely consumed in the face of ongoing debate around its reported health benefits and possible adverse side effects (4). Interest in resveratrol first arose after it was hypothesized to be the causative agent of the apparent health benefits of red wine (5). Since then, research has shown resveratrol to have pleiotropic effects, including potential anticancer, anti-inflammatory, cardioprotective, and neuroprotective activity (6 -10). These claims have led to the rapid expansion of resveratrol supplementation; however, there is limited clinical data on the efficacy of resveratrol and there is no scientific consensus on the underlying metabolic and molecular mechanisms of resveratrol activity (4,(11)(12)(13). To date, resveratrol has been reported to act on many different molecular targets, including sirtuins (14 -16), p53 (17), PI3K/Akt (18), and NF-B (19).
In the last 10 years, eukaryotic DNA topoisomerase II (topo II) 2 has been added to the list of potential resveratrol targets as a possible explanation for the anticancer effects of the agent (20 -22). Topo II is a member of the type II topoisomerase superfamily, a group of essential enzymes that manages supercoils formed during cellular processes such as replication and transcription, and it also unlinks entangled double-stranded segments of DNA to promote chromosome segregation (23). Type II topoisomerases utilize a so-called DNA "strand-passage" mechanism, which couples ATP hydrolysis to the transient, enzyme-mediated formation of a double strand break (DSB) to physically move one DNA duplex through another (24 -27). In humans, topo II is a validated target of anticancer agents, as it is required for actively dividing cells, and disruption of enzyme function can induce the formation of cytotoxic DNA damage (28,29).
Topoisomerase inhibitors are broadly categorized into two classes: poisons and general catalytic inhibitors. Topoisomerase poisons induce DSB formation, whereas general catalytic inhibitors reduce enzyme activity without generating DNA damage directly (30). Topo II poisons such as epipodophyllotoxins (etoposide) and anthracyclines (doxorubicin) work by intercalating between DNA bases at the site of strand cleavage, stabilizing the cleaved complex, and preventing religation of the double strand break (31,32). By comparison, catalytic inhibitors have more varied mechanisms of action, and can impact enzyme activity by affecting substrate binding, DNA cleavage, or enzyme turnover (33). In the clinic, catalytic inhibitors can be used to reduce the toxicity of topoisomerase poi-sons during chemotherapy. For example, ICRF-187 (commonly known as dexrazoxane), is used to mitigate the cardiotoxicity of anthracycline-based treatments (34 -36).
Based on increased DNA damage and cell-cycle delay following treatment in mammalian cells, resveratrol has been claimed to act as a topo II poison (22,37,38). Additionally, resveratrol has been shown to inhibit topo II decatenation activity in vitro (22,38). We noted that resveratrol bears some intriguing chemical similarities to ICRF-187 (Fig. 1A). In particular, both compounds are composed of two 6-membered rings separated by a 2-carbon linker, and one of the rings on resveratrol has two, meta-positioned exocyclic hydroxyls that resemble the orientation of the oxygens on ICRF-187. Biochemical, structural, and mutational analyses of ICRF-187 inhibition of topo II have shown that ICRF-187 binds to the dimerization interface formed between the enzyme's ATPase domains and, following nucleotide binding, stabilizes their interaction to prevent subunit re-opening (39 -43). The action of ICRF-187 does not directly promote the formation of topo II-dependent cleaved DNA complexes; however, studies in yeast and mammalian cells have shown that ICRF-187, like resveratrol, can increase indicators of DNA damage (44,45), possibly as a result of topo II inhibition and the build-up of problematic DNA topologies. ICRF-187, a synthetic molecule, is currently the only known ligand of a large, solvent-filled pocket that forms between nucleotide-dimerized ATPase domains of topo II (Fig. 1B) (43). Interestingly, the ICRF-187-binding site is highly conserved in eukaryotic topo IIs (Fig. 1C), but absent in other related GHKLfamily ATPases (including prokaryotic type II topoisomerase homologs), suggesting that this locus may constitute a binding site for cellular effector molecules.
To define the mechanism by which resveratrol inhibits eukaryotic topo II, we carried out a comprehensive series of assays comparing Saccharomyces cerevisiae topo II (ScTop2) and both isoforms of human topo II (HsTop2␣ and HsTop2␤). We confirm that resveratrol is indeed a topo II inhibitor but demonstrate that it is not a poison as originally suggested.
Interestingly, we find that both ICRF-187 and resveratrol show preferred selectivity for the human enzymes as compared with budding yeast topo II. A single resistance mutation in the ICRF-187-binding locus is shown to impart cross-resistance to resveratrol, indicating that both agents act at this site; however, using a FRET-based assay to monitor enzyme conformational dynamics, we find that unlike ICRF-187, which stabilizes topo II ATPase domain dimerization, resveratrol impedes this interaction. Together, these results describe a novel mechanism of topo II inhibition and implicate resveratrol as the first example of a natural molecule that can associate with the highly conserved ICRF-187-binding site.

Like ICRF-187, resveratrol is not a topo II poison
To begin to understand how resveratrol acts on topo II, we first performed a series of assays comparing the biochemical effects of the agent to that of ICRF-187. We conducted DNA supercoil relaxation assays with both S. cerevisiae topo II (ScTop2) and the two isoforms of Homo sapiens topo II (HsTop2␣ and HsTop2␤). Purified enzymes were incubated with negatively supercoiled plasmid substrate and ATP to catalyze relaxation. Reactions were then quenched at various time points (supplemental Fig. S1), and the topoisomer distributions were resolved by native agarose gel electrophoresis. Resveratrol and ICRF-187 were able to inhibit the relaxation activity of both human isoforms at similar compound concentrations, as indicated by increased supercoiled substrate retention at higher inhibitor concentrations after a 20-min reaction incubation (Fig. 2). Parallel studies assessing the ability of HsTop2␣ to decatenate kinetoplast DNA (kDNA) yielded similar results (supplemental Fig. S2). Notably, both compounds were less efficient at inhibiting ScTop2 as compared with the human enzymes ( Fig. 2 and supplemental Fig. S1).
ICRF-187 is most often described as a topo II catalytic inhibitor, but it has been likened to a weak poison based on the Residue Thr 27 , which is homologous to Thr 49 in HsTop2␣ and constitutes a known ICRF resistance locus (43,57), is shown in red. C, conservation of the ICRF-187/resveratrol-binding site among eukaryotic type II topoisomerases. Conservation scores of residues from a multiple sequence alignment of 106 eukaryotic topo II sequences (70) are mapped onto a surface view of one protomer of the ScTop2 ATPase domain bound to ICRF-187(green) and AMPPNP (yellow) (PDB code 1QZR (43) using CONSURF (71). The other monomer is shown as a cream ribbon.

Resveratrol: A novel type of topoisomerase II inhibitor
observation that bis-dioxopiperazine cytotoxicity requires enzyme activity (44) and that topo II-DNA covalent complexes have been seen in budding yeast cells following drug treatment (45). Similarly, the accumulation of DNA damage markers in human cells upon resveratrol exposure has led to the suggestion that resveratrol is also a topo II poison (22,38). To further investigate the capacity of ICRF-187 and resveratrol to act as poisons, we monitored the ability of both agents to induce the linearization of plasmid DNA in the presence of topo II. Etoposide was included as a positive control for topo II poisoning. To observe the formation of single-or double-stranded DNA breaks on a supercoiled plasmid substrate, we incubated negatively supercoiled plasmid with purified enzyme, ATP, and inhibitor for 20 min. The products were then resolved on an ethidium bromide-containing agarose gel to optimally separate the cleavage products from closed plasmid topoisomers. As expected, etoposide strongly induced DNA cleavage in a dosedependent manner, especially with the yeast enzyme (Fig. 3). By contrast, neither resveratrol nor ICRF-187 strongly induced DNA cleavage by any of the topo II isoforms beyond levels seen with the drug-free enzymes (some DNA cleavage products did occasionally appear when HsTop2␣ was incubated with ICRF-187; however, the amount of cleavage was considerably less than what was formed upon etoposide treatment and, importantly, did not increase in a dose-dependent manner, indicating that it was not drug dependent) (Fig. 3). Overall, these results demonstrate that resveratrol and ICRF-187 have little to no capacity to poison topo II, and that their ability to inhibit supercoil relaxation derives from a different mode of action.

Resveratrol inhibits topo II ATPase activity, but in a manner distinct from ICRF-187
The topo II reaction relies on the hydrolysis of two molecules of ATP to drive strand passage (27,46,47). Having eliminated cleavage-complex stabilization as a possible mechanism of resveratrol inhibition, we assessed the compound's effect on the ATPase activity of topo II. ICRF-187, an uncompetitive inhibitor of ATP hydrolysis (41), was again used as a comparative control. The relative rates of ATP hydrolysis were measured in a NADH-coupled ATPase assay (48,49) at different ATP and inhibitor concentrations and plotted as Michaelis-Menten curves (Fig. 4). In contrast to the supercoil relaxation assays ( Fig. 2 and supplemental Fig. S1), resveratrol's effect on ATPase activity was not as robust as seen with ICRF-187 (Fig. 4). Moreover, resveratrol and ICRF-187 displayed different modes of inhibition (supplemental Table S1). For example, ICRF-187 treatment led to an apparent decrease in both V max and K m , as expected for its previously reported, uncompetitive mode of inhibition (41). By comparison, the addition of resveratrol led to an increase in K m and a less than 2-fold decrease in V max , a behavior consistent for a mixed-type inhibitor with more competitive character (supplemental Table S1). Despite the notable differences in modes of inhibition, both ICRF-187 and resveratrol proved to be more potent against both human topo IIs. In the case of resveratrol inhibition, HsTop2␣ experienced a

Resveratrol: A novel type of topoisomerase II inhibitor
stronger V max effect and HsTop2␤ experienced a stronger K m effect as compared with ScTop2.

Resveratrol antagonizes topo II ATPase domain dimerization
During type II topoisomerase activity, nucleotide binding induces dimerization of the ATPase domains to allow ATP hydrolysis to occur (50). Because of its chemical dissimilarity to ATP, we reasoned that rather than directly competing for nucleotide binding per se, resveratrol might act as a "pseudocompetitive" (51) inhibitor that interferes with ATPase domain dimerization, which is necessary for the productive binding of nucleotide. To test this idea, we developed a FRET-based assay that monitors the oligomeric status of the topo II ATPase elements. We first generated a C-terminally truncated construct of budding yeast topo II (ScTop2 ⌬CTD , residues 1-1177) that removes an unstructured region not required for activity (47,52). We then mutated several surface-exposed cysteines (C48A, C381A, C471V, and C731A) to eliminate all reactive sites other than Cys 180 (which already resides on the topo II ATPase domain) for labeling with a fluorophore. This "Cys-lite" holoenzyme was labeled with Alexa Fluor 555-maleimide and Alexa Fluor 647maleimide to introduce FRET pairs on ATPase region ( Fig.  5A and "Experimental procedures"). The labeled enzyme showed no apparent loss in supercoil relaxation activity compared with full-length, native ScTop2 (supplemental Fig. S3).
Using the FRET-labeled topo II construct, we first assessed the effects of ICRF-187 and resveratrol on ATPase status in the presence and absence of the turnover product, ADP (Fig. 5B). The nonhydrolyzable ATP analog AMPPNP, which induces the formation of a stable ATPase domain dimer state (27,50,53,54), was also included as a positive control. The addition of AMPPNP produced a significant increase in FRET efficiency from the dual dye-labeled construct as compared with the nucleotide-free enzyme control, indicating a shift in equilibrium toward the gate-closed state. ADP alone led to a less dramatic, but still detectable, increase in FRET efficiency (p Ͻ 0.01), suggesting that product binding leads either to partial gate closure or to a mixture of open and closed states (at very high concentrations, greater than those used here, ADP can stabilize ATPase domain dimerization (55,56)). ICRF-187 by itself had no significant effect on ATPase status, and did not appreciably alter FRET values compared with ADP alone. By contrast, resveratrol produced a significant decrease in FRET when incubated with dual-labeled topo II alone or when ADP was present. This result indicates that topo II bound to resveratrol favors an open ATPase domain conformation.
Based on the outcome of the single time point measurements, we next conducted a time course experiment to assess

Resveratrol: A novel type of topoisomerase II inhibitor
the effects of ICRF-187 and resveratrol on ATPase domain dimerization. Low concentrations of AMPPNP, below the established K m values we measured for ATP (supplemental Table S1), were used to slow down the rate of ATPase dimerization to permit analysis without resorting to rapid-mix experiments (Fig. 5C). Dual-labeled topo II was preincubated with either ICRF-187 or resveratrol on ice, and a 0-min time point was taken prior to adding AMPPNP. Following the addition of AMPPNP, reactions were shifted to room temperature and changes in FRET were monitored as a function of time. As before, the presence of ICRF-187 yielded no change in FRET in comparison to the AMPPNP-only control sample. This behavior indicates that ICRF-187 does not stimulate ATPase domain association, but rather serves to disfavor their separation, a model consistent with prior pre-steady state studies of the inhibitor (41). By contrast, resveratrol led to increasingly lower FRET efficiencies over time compared with the nucleotide-free sample, providing further evidence that resveratrol inhibits ATPase domain dimerization.

Resveratrol does not affect topo II ATPase status if the domains are pre-dimerized
Given that resveratrol appears to inhibit ATPase domain dimerization, we next set out to test whether the agent might act at the ICRF-binding site, or at a different site that would allosterically trigger ATPase domain monomerization. Because the ICRF-binding pocket is masked and solvent inaccessible following ATPase domain association (43), we reasoned that if this locus were to bind resveratrol, then the compound should not be able to interact with an enzyme whose ATPase regions have been pre-dimerized. To compare the effect of resveratrol on an "open" versus "closed" pair of ATPase domains, we performed an order-of-addition experiment with the inhibitor and AMPPNP ( Fig. 5D and "Experimental procedures"). When AMPPNP was preincubated with topo II for 15 min to promote ATPase domain dimerization prior to adding resveratrol, the sample initially displayed a high level of FRET efficiency compared with the apoenzyme that stayed relatively constant over time. By contrast, when resveratrol was added first, the observed FRET efficiencies were initially below that of that the apoenzyme, and stayed low throughout the 30-min reaction time. Collectively, these data demonstrate that resveratrol affects ATPase domain dimerization if introduced prior to gate closure, and are consistent with a model in which resveratrol binds to the ICRF-187-binding site at the ATPase domain dimerization interface.

Resveratrol is likely to interact with topo II using the same binding pocket as ICRF-187
The chemical structures of ICRF-187 and resveratrol show certain similarities with respect to size, shape, and functionality: both possess a pair of six-membered rings separated by a two-carbon linker, and both retain a pair of exocyclic, metapositioned hydrogen-bond acceptor oxygens on at least one of the two rings (Fig. 1A). Our results show that both compounds also have enhanced activity against the human topo II isoforms compared with yeast, and that both agents interfere with the normal ATPase domain dimerization cycle. Together, these observations indicate that ICRF-187 and resveratrol share a binding locus on topo II.
To more specifically probe the prospective binding site for resveratrol, we examined the ability of the compound to inhibit a topo II mutant resistant to ICRF-187. A T49I mutation in HsTop2␣ has been previously isolated for ICRF-187 resistance (43,57). The homologous residue in ScTop2 (Thr-27) maps to the ICRF-187-binding site in a crystal structure of the isolated ATPase region with the drug (43), suggesting that the T49I mutation likely interferes with ICRF-187 binding (Fig. 1B). Because HsTop2␣ is more sensitive than ScTop2 to both ICRF-187 and resveratrol (Figs. 2 and 4), we elected to use the equivalent HsTop2␣ T49I mutant for our comparative studies in vitro. We first examined the activity of HsTop2␣ T49I compared with native HsTop2␣. Although the HsTop2␣ T49I was ϳ5-fold slower in both supercoil relaxation and ATP hydrolysis assays in the absence of inhibitor (supplemental Fig. S4, Fig. 6, and supplemental Table S1), the mutant nonetheless had sufficient activity to assess inhibition by ICRF-187 and resveratrol.
To examine the effects of the T49I substitution on inhibition by ICRF-187 and resveratrol, time course DNA supercoil relaxation assays were conducted for both wild-type and mutant HsTop2␣. Upon addition of ICRF-187, HsTop2␣ T49I showed no change in relaxation activity, in contrast with the wild-type enzyme, which was effectively inhibited (Fig. 7A). Significantly, HsTop2␣ T49I was also resistant to resveratrol inhibition of supercoil relaxation as compared with HsTop2␣ WT (Fig. 7B). ATPase assays conducted with HsTop2␣ WT and HsTop2␣ T49I corroborated these findings, showing that the T49I mutation

Resveratrol: A novel type of topoisomerase II inhibitor
imparts resistance to both ICRF-187 and resveratrol ( Fig. 6 and supplemental Table S1). Overall, these results strongly indicate that resveratrol binds to the same conserved pocket as ICRF-187 on the topo II ATPase domain.

Discussion
Although there have been many claims of the potential health benefits of resveratrol (4, 6 -10), the actual molecular targets of this agent are still disputed (4,(11)(12)(13). Among its actions, resveratrol has been suggested to be a topo II poison based on the observations that mammalian cancer cell lines treated with resveratrol accumulate DSBs, and that resveratrol inhibits topo II-dependent decatenation activity in vitro (22,38). Interestingly, resveratrol has a degree of chemically similarity to ICRF-187 (Fig. 1A), a topo II inhibitor that does not induce DSB formation in vitro (Fig. 3). ICRF-187 is used clini-cally to attenuate the cardiotoxicity of doxorubicin, a known topo II poison (34 -36), and there is evidence that resveratrol may similarly modulate doxorubicin toxicity (58). Therefore, we hypothesized that resveratrol may behave more akin to ICRF-187 in its action than to a classic topo II poison.
Our biochemical data establishes that resveratrol is a topo II inhibitor and not a poison. Resveratrol inhibits the DNA supercoil relaxation activity (Fig. 2), decatenation activity (supplemental Fig. S2), and ATPase activity of the three topo II homologs tested here (Fig. 4). Like ICRF-187, resveratrol is also unable to induce topo II to form DSBs in a dose-dependent manner (Fig. 3). Interestingly, ICRF-187 and resveratrol are both more active against the human topo II isoforms (Top2␣ and Top2␤) than against budding yeast topo II (Figs. 2 and 4). This differential sensitivity suggests that ScTop2 may not be an The graph to the right shows example emission spectra. Normalized intensity values represent the emission intensity measured at each wavelength divided by the total emission intensity measured across the entire spectrum. The ratiometric FRET efficiency (E) was calculated from the maximum donor emission (F D ) and maximum acceptor emission (F A ) as shown in the equation. The change in FRET (⌬FRET) was calculated with respect to the FRET efficiency of the apoenzyme sample (E apo ). B, FRET-based monitoring of ATPase domain closure in the presence of different ligands. Emission spectra were collected from solutions containing 200 nM FRET-labeled topo II and the ligands indicated on the x axis after equilibrating for 30 min at room temperature. Asterisks indicate significant differences from the apoenzyme signal (⌬FRET ϭ 0) as determined by Student's t test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). C, time-dependent ATPase closure assay comparing AMPPNP and resveratrol order of addition. Emission spectra were collected from solutions containing 200 nM FRET-labeled enzyme at indicated time points immediately after AMPPNP addition. Error bars indicate the average of three independent runs. D, quantification of AMPPNP and resveratrol order of addition assays. 200 nM FRET-labeled enzyme was preincubated with ligand (1) at room temperature for 15 min before addition of ligand (2). Emission spectra were taken 0 and 30 min after addition of second ligand. Concentrations of ligands used were 0.03 mM AMPPNP and 0.2 mM resveratrol (RSV). Asterisks indicate significant differences as determined by Student's t test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).

Resveratrol: A novel type of topoisomerase II inhibitor
ideal model to study drugs aimed at targeting the ATPase domain of human topo II. Our findings also indicate the DNA damage products formed in cells upon treatment with resveratrol (22, 38) (or ICRF-187 (44, 45)) likely do not arise from any poisoning activity per se, but rather from the loss of essential topo II activity, such as chromosome disentanglement during replication and cell division.
Although resveratrol and ICRF-187 appear similar in their ability to inhibit supercoil relaxation activity by topo II, ATPase activity assays demonstrate that the mechanism of resveratrol inhibition differs from that of ICRF-187 ( Fig. 4 and supplemental Table S1). Consistent with prior studies using ScTop2 (39,41,43) we find that ICRF-187 acts as a mixed (uncompetitive) inhibitor of ATPase activity of not only the yeast enzyme, but of both human topo IIs as well. By contrast, resveratrol appears to act more like a competitive inhibitor. A traditional interpretation of the ATPase inhibition data would suggest that resveratrol competes with nucleotide binding to the catalytic center. However, this mechanism seems highly unlikely due to the disparity in chemical shape and properties between ATP and resveratrol. Moreover, our analysis of the effects of resveratrol on ATPase domain status reveals that resveratrol interferes with the nucleotide-dependent dimerization of this region (Fig. 5); classic competitive inhibitors of type II topoisomerases (e.g. AMPPNP (27,50), or novobiocin with bacterial homologs (59,60)) stabilize ATPase domain closure. These findings collectively suggest that resveratrol acts as a pseudo-competitive inhibitor of topo II ATPase activity. In pseudo-competitive inhibition, the binding of substrate to a catalytic site is perturbed by the binding of the inhibitor to a second, distal site, but the two sites are linked in such a way that only substrate bind-ing, and not k cat , is affected (51). The strict reciprocity that we observe between substrate and inhibitor action on ATPase dimerization status, coupled with the lack of changes seen for V max calculated from ATP turnover studies, argues against the existence of a resveratrol-binding site that allosterically modulates ATP binding at the expense of catalytic prowess (as occurs in noncompetitive inhibition), and is instead consistent with classic pseudo-competitive models.
One of the most compelling pieces of data that argue against direct competition between resveratrol and ATP for a common binding site is our discovery that a mutation which confers ICRF-187 resistance also confers resistance to resveratrol (Figs. 1B, 6, and 7). However, resveratrol and ICRF-187 also appear to inhibit topo II through distinct mechanisms (Fig. 8). The basis for these distinct modes of action likely arises from chemical differences between the two agents. ICRF-187 is a flattened, pseudo-2-fold symmetric compound that stabilizes ATPase domain interactions by binding to and bridging two protomers across the dyad axis of an ATP-bound dimer (43). By contrast, resveratrol, which inhibits dimerization, is asymmetrical, with only one six-membered ring resembling the ICRF-187 dioxopiperazine rings. Given these features, we posit that each topo II ATPase domain can potentially interact with one resveratrol molecule, compared with two domains sharing a single ICRF-187 moiety. Even if both resveratrol-binding sites are not occupied, the binding of one molecule may be enough to block ATPase domain dimerization because the rigid, planar structure of resveratrol and the para-hydroxyl group on the second ring are likely to incur steric clashes upon gate closure.
Our findings identify resveratrol as a natural compound, the first that we are aware of, which can alter topo II activity

Resveratrol: A novel type of topoisomerase II inhibitor
through the ICRF-187 binding locus (ICRF-187 is a synthetic molecule (61)). Interestingly, evolutionary conservation analysis of eukaryotic type II topoisomerases reveals that the ICRF-187 interaction site, which comprises a large (350 Å 3 ) solventfilled cavity, is as highly conserved as the ATP-binding site itself (Fig. 1, B and C). This observation suggests that the ICRF-187binding pocket may serve as an effector site for modulating topo II activity. Resveratrol is one of many natural polyphenols that are widely consumed by humans as part of our diet, and various classes of polyphenols (including resveratrol) are also often sold as dietary supplements for purported health benefits (62,63). Like resveratrol, the actual physiological effects of consuming these compounds (particularly in excess, as is the case with supplementation) are unknown (64). Many polyphenols, such as flavonoids, have dihydroxyphenyl groups that resemble the six-membered ring on resveratrol; we speculate that this moiety may underlie the reported activity of polyphenol agents as topo II antagonists (62). In this view, resveratrol would constitute just one of many natural products that could allosterically affect topo II function through the highly conserved binding site at the ATPase domain dimerization interface. Probing whether there exist other natural agents that bind to this locus is a direction for further study. Moreover, the ability of ICRF-187 and resveratrol to show differential potency against the human isoforms of topo II as compared with ScTop2, as well ICRF-187's differential effects against HsTop2␣ versus HsTop2␤ (Fig. 2 and 4), suggest that modulators acting through this site may be tunable in their efficacy. Given the proven utility of ICRF-187 in certain chemotherapeutic regimens, the identification of such modulators may provide a new path to the development of new classes of topo II inhibitors of clinical value.

Cloning and mutagenesis
Expression constructs for full-length, wild-type topo II enzymes were generated by PCR amplification of the genes, followed by insertion into the 12UraB plasmid (Addgene) for ScTop2 and HsTop2␣ or the 12UraC plasmid (Addgene) for HsTop2␤ by ligation-independent cloning (65). Both plasmids contain a galactose-inducible promoter for expression of an N terminally His 6 -tagged (12UraB) or His 6 MBP-tagged (12UraC) fusion protein, with a downstream tobacco etch virus (TEV) protease cleavage site for removal of the tag.
The expression construct for HsTop2␣ T49I was generated by site-directed mutagenesis of the wild-type expression plasmid. The expression construct for ScTop2 (⌬CTD,Cys-lite) was similarly generated by site-directed mutagenesis of a previously described ScTop2 (⌬CTD) expression construct (47) to mutate surface-exposed cysteines as follows: C48A, C381A, C471V, and C731A.

Protein expression and purification
All proteins were expressed in the S. cerevisiae strain BCY123 (66). Starter cultures were grown to saturation overnight in CSM-Ura Ϫ media supplemented with 2% dextrose, 2% lactic acid, and 1.5% glycerol at 30°C. Starter cultures were then diluted 10-fold in YP media with 2% lactic acid and 1.5% glycerol, and grown to an OD of 1.0 -1.3 at 30°C (12-15 h), at which point protein expression was induced by the addition of 2% galactose for 6 h at 30°C. Cells were harvested by centrifugation, resuspended in 1 ml of 1 mM EDTA and 250 mM NaCl per liter of culture, and flash frozen dropwise in liquid nitrogen for storage at Ϫ80°C.
For purification, frozen pellets were lysed by cryogenic grinding in a 6870 Freezer Mill (SPEX SamplePrep). The cell powder was resuspended in 20 mM Tris-HCl (pH 8.5), 300 mM KCl, 20 mM imidazole (pH 8.0), 10% glycerol, 1 mM PMSF, 2.34 M leupeptin, 1.45 M pepstatin, and 0.5 mM TCEP (TCEP was excluded for preparation of proteins for fluorophore labeling). Lysate was clarified by centrifugation at 15,000 rpm in a JA 25.50 rotor for 45 min and loaded onto a HisTrap FF (GE) nickel-chelating Sepharose column. The HisTrap FF column was first washed with resuspension buffer and then a low salt version of the resuspension buffer containing 150 mM KCl. Protein was directly eluted onto a HiTrap SP (GE Healthcare) cation exchange column with 200 mM imidazole in the low salt resuspension buffer. Following a salt gradient elution (150 to 500 mM KCl in 20 mM Tris-HCl (pH 8.5) and 10% glycerol), the His 6 or His 6 MBP tags were removed by the addition of His 6tagged TEV protease (QB3 MacroLab) (67,68) at 4°C for 12 h, followed by a second nickel-affinity step to remove the cleaved tag, uncleaved protein, and TEV protease. Protein was further purified on a gel-filtration column (S-400, GE Healthcare) in 500 mM KCl, 20 mM Tris-HCl (pH 7.9), 10% glycerol, and 0.5 mM TCEP. The purity of peak fractions was assessed by SDS-PAGE. The purified protein was concentrated using a 30-kDa MWCO filter (Amicon) and flash frozen in gel filtration buffer supplemented with 30% glycerol (concentrated protein was diluted with gel filtration buffer containing 70% glycerol to increase final glycerol concentration to 30% prior to freezing) for storage at Ϫ80°C.

Preparation of substrates and reagents for DNA relaxation and cleavage assays
Negatively supercoiled pSG483, a derivative of pBlue-Script SK, was prepared from the XL1-Blue Escherichia coli strain using NucleoBond Xtra Maxi columns (Macherey-Nagel). The purity and topology of the final substrate were assessed by native agarose gel electrophoresis. For storage, plasmid DNA was diluted to 250 ng/l (140 nM) in water, flash frozen in 25-l aliquots, and stored at Ϫ20°C. Etoposide, ICRF-187, and resveratrol (Sigma) were dissolved in fresh DMSO (Sigma) and stored at Ϫ20°C. ATP stocks were prepared in water and neutralized by adding 10 M NaOH to pH 7.0 before flash freezing.

DNA relaxation, kDNA decatenation, and DNA cleavage assays
Freshly thawed topo II aliquots were diluted on ice in 2-fold increments to a final concentration in 30 mM Tris-HCl (pH 7.9), 500 mM potassium acetate, 10% glycerol, and 0.5 mM TCEP. DNA substrate and inhibitor were then added to the enzyme on ice. Negatively supercoiled pSG483 was the DNA substrate for relaxation and cleavage assays, and kDNA (Inspiralis) was the DNA substrate for decatenation assays. Reactions were initi-Resveratrol: A novel type of topoisomerase II inhibitor ated by adding freshly thawed ATP and shifting to 30°C for ScTop2 assays and 37°C for HsTop2␣/␤ assays. The final reaction solution contained 32 mM Tris-HCl (pH 7.9), 100 mM potassium acetate, 10 mM magnesium acetate, 0.05 mg/ml of BSA, 0.6 mM TCEP, 10% glycerol, 2.5% DMSO, and 1 mM ATP. Reactions were stopped with 20 mM EDTA and 1% SDS. Quenched reactions were treated with 0.2 mg/ml of Proteinase K at 37°C for 30 min to remove any protein still bound to DNA. Relaxation assays and decatenation assays were resolved on native 1.4% agarose, 1ϫ TAE gels. Cleavage assays were resolved on 1.4% agarose, 1ϫ TAE, and 0.4 g/ml of ethidium bromide gels to optimally resolve linear products. (The intercalation of ethidium bromide into DNA generates a local decrease in the twist of the double helix at the site of binding, which in turn induces compensatory positive writhe in a closed plasmid. As a result, at moderate concentrations of ethidium, relaxed plasmids become positively supercoiled and run faster than negatively supercoiled plasmids, which adopt a more relaxed state. Nicked and linear plasmids do not experience any appreciable change in writhe upon ethidium bromide intercalation, and therefore migrate similarly as compared with native (ethidium-free) conditions.) Gels were run at 2-2.5 V/cm for 15-20 h at room temperature, stained with 1 g/ml of ethidium bromide for 30 min, destained in water for 1.5 h, and visualized by UV transillumination.
For quantification of unreacted substrate in relaxation assays, supercoiled band intensities were calculated in ImageJ (69) from unaltered images. Supercoiled band intensities were then normalized to either the fully retained substrate band in the negative control lane or the maximally depleted substrate band in the no-inhibitor lane as appropriate for each experiment. To quantitate enzyme activity in decatenation assays, we similarly calculated the intensities of the minicircle bands as a measure of product formation. The minicircle band intensities were then normalized to the intensity of the minicircle band in the no-inhibitor lane.

ATP-gate labeling and FRET assays
Purified ScTop2 (⌬CTD,Cys-lite) was simultaneously labeled with a 10-fold molar excess of Alexa Fluor 555 C 2 maleimide and Alexa Fluor 647 C 2 maleimide (Sigma) in 20 mM Tris-HCl (pH 7.5), 500 mM KCl, 15 M TCEP, and 10% glycerol at 4°C for 12-15 h. Because the enzyme purifies as a homodimer, this labeling strategy results in ϳ50% of holoenzymes labeled on both sites with the same dye and 50% of holoenzymes labeled with one of each dye. Labeling reactions were quenched with 5 mM DTT and run over a gel-filtration column (S400, GE) equilibrated in 20 mM Tris-HCl (pH 7.9), 500 mM KCl, and 10% glycerol to removed unreacted dye and any aggregated protein that might have accumulated during the labeling process. Labeling efficiency was quantified by measuring the protein concentration by light absorbance at 280 nm and dye concentration by absorption at the excitation wavelength for each label. The specific activity of labeled proteins was assessed in supercoil relaxation assays as described.
Samples for FRET measurement were prepared on ice with 200 nM dual-labeled ScTop2 (⌬CTD,Cys-lite) , 10 mM Tris-HCl (pH 7.9), 100 mM KCl, 5 mM MgCl 2 , 2% glycerol, 2.5% DMSO, and any other additional ligands (e.g. resveratrol, ADP, AMPPNP, and ICRF-187) specific to a particular given reaction scheme. Reactions were then incubated at room temperature as specified for each experiment. Samples were illuminated with 530 nm light and emission spectra were collected from 545 to 700 nm in a Fluoromax fluorometer 4 (HORIBA Jobin Yvon, Edison, NJ). An apoenzyme sample was used as a baseline for calculating changes in FRET efficiencies upon ligand binding, and each comparative set of experiments was performed with protein from the same labeling reaction to account for variation in the labeling efficiency and population distribution of duallabeled enzymes. In experiments where two ligands were added sequentially, dual-labeled ScTop2 ⌬CTD,Cys-lite was allowed to incubate with the first ligand at room temperature for 15 min prior to addition of the second ligand. A 0-min spectrum was taken immediately after adding the second ligand, followed by second spectrum after incubating the reaction at room temperature for 30 min. In the control samples containing only one ligand, the compound was added immediately prior to the initial spectrum after a 15-min ligand-free preincubation of the apoenzyme at room temperature. FRET spectra were normalized by the total integral across each spectrum. To determine ratiometric FRET efficiencies, the maximum of the acceptor emission peak was divided by the sum of the maximums of the donor and acceptor emission peaks (Fig. 4).

Evolutionary conservation analysis of ICRF-187-binding locus in eukaryotic type II topoisomerases
A multiple sequence alignment of 106 eukaryotic type II topoisomerase sequences, ranging from budding yeast to humans, was generated through MAFFT (70). The resulting multiple sequence alignment file was then uploaded to the CONSURF server (71) with the structure of the ScTop2 ATPase domain bound to ICRF-187 and AMPPNP (PDB code 1QZR (43)) as a reference to generate a PDB file in which the residues Resveratrol: A novel type of topoisomerase II inhibitor are colored based on their conservation scores. Highly conserved regions in the ATPase domain of topo II were visualized and analyzed in CHIMERA (72).