Chromatin folding and DNA replication inhibition mediated by a highly antitumor-active tetrazolato-bridged dinuclear platinum(II) complex

Chromatin DNA must be read out for various cellular functions, and copied for the next cell division. These processes are targets of many anticancer agents. Platinum-based drugs, such as cisplatin, have been used extensively in cancer chemotherapy. The drug–DNA interaction causes DNA crosslinks and subsequent cytotoxicity. Recently, it was reported that an azolato-bridged dinuclear platinum(II) complex, 5-H-Y, exhibits a different anticancer spectrum from cisplatin. Here, using an interdisciplinary approach, we reveal that the cytotoxic mechanism of 5-H-Y is distinct from that of cisplatin. 5-H-Y inhibits DNA replication and also RNA transcription, arresting cells in the S/G2 phase, and are effective against cisplatin-resistant cancer cells. Moreover, it causes much less DNA crosslinking than cisplatin, and induces chromatin folding. 5-H-Y will expand the clinical applications for the treatment of chemotherapy-insensitive cancers.

platinum-based drugs, such as carboplatin and oxaliplatin 6 , are considered to work in a similar fashion 7 . The platinum-DNA interactions make both inter-and intrastrand crosslinks in DNA, suppressing DNA replication 7 and also RNA transcription 8 . For DNA replication process, intra-strand DNA crosslinks can be bypassed by some translesion synthesis (TLS) polymerases 7 . To deal with interstrand DNA crosslinks (ICLs), mammalian cells have evolved the Fanconi anemia (FA)/BRCA pathway, which is coupled with DNA replication 9 . FA is a rare genetic disorder characterized by progressive bone marrow failure and a highly elevated risk of hematological and squamous cancers 10 . To date, 19 FANC genes have been identified from FA patients, whose cells are highly sensitive to ICL-inducing agents, including cisplatin. Although the precise mechanism of ICL repair by the FA/BRCA pathway is not yet fully understood, it is clear that complex actions of FA proteins, nucleases, TLS polymerases, and homologous recombination proteins are involved. Importantly, loss of any protein involved in the FA/BRCA pathway ultimately leads to hypersensitivity to cisplatin.
A common problem with cisplatin and its derivatives is that prolonged treatment generates resistant cancer cells (e.g. ref. 11). Thus, it is important to develop new drugs that can kill cisplatin-resistant cancer cells. Ideally, a next-generation platinum-based drug should show high therapeutic efficiency and a wide anticancer spectrum, that is, a cytotoxicity profile with many cancer cell lines. Such a new drug is also likely to be effective against chemotherapy-insensitive cancers, such as pancreatic cancer. However, conventional platinum-based drugs usually have similar anticancer spectra and the clinical platinum-based drugs show cross-resistance 12,13 . Thus, a significant structural modification appears to be required to design candidate next-generation, platinum-based drugs [14][15][16][17][18] .
In the present study, using combined techniques of cell biology, structural biology, and biophysics, we investigated the cytotoxic mechanism of 5-H-Y. We found that the compound inhibits DNA replication and RNA transcription, and arrests treated cells in the S/G2 phase, causing great cytotoxicity. 5-H-Y has much less DNA crosslinking ability than cisplatin, and binds to DNA very tightly, inducing chromatin folding. We also found that DNA damage by 5-H-Y is repaired differently from ICL generated by cisplatin, and 5-H-Y is effective against cisplatin-resistant cancer cells. Our study provides a mechanistic insight into the cytotoxicity of 5-H-Y.

Results
The novel platinum complex 5-H-Y inhibits cell proliferation. To evaluate the effects of 5-H-Y and cisplatin on cell growth inhibition, we first performed cell proliferation assays using four human cell lines (PC9, HeLa, U2OS, and TIG-1) (Figs 1B and S1A). PC9, HeLa, and U2OS cells are cancer cell lines and TIG-1 is a 'normal' human fibroblast line. Cell numbers were examined over time under various concentrations of 5-H-Y and cisplatin, from 0 to 96 h. Both drugs inhibited the growth of all cell lines tested in a similar manner (Figs 1B and S1A), consistent with a previous report 21 . These results suggest that 5-H-Y and cisplatin show comparable inhibitory effects on the proliferation of these cell lines.

5-H-Y is incorporated into cell nuclei.
To gain clues into the mechanism of 5-H-Y cytotoxicity, we next investigated the intracellular localization of 5-H-Y and cisplatin in PC9 cells. It is normally not possible to examine drug localization by conventional cell biological methods. For this purpose, we used scanning X-ray fluorescence microscopy (SXFM) 29,30 (Fig. 1C). This method enables the detection of the target elements at a single-cell level and gives a cellular localization profile of these elements. We examined various element localizations in both 5-H-Y-and cisplatin-treated PC9 cells. We detected many elements, including phosphorus, sulfur, zinc, and platinum (Fig. 1D). Signals of phosphorus, sulfur, and zinc mainly reflect on localizations of nucleic acids, proteins, and DNA-binding proteins, respectively 29,30 . In 5-H-Y-treated cells, platinum was observed throughout the cells, including in the nuclei. The cisplatin-treated cells also showed platinum signals, consistent with our previous study [29][30][31] .
To further confirm these findings, we fractionated the drug-treated PC9 cells as whole cells, nuclei (detergent-treated), and naked DNA fractions, and analyzed the amount of platinum in each by inductively coupled plasma-mass spectrometry (ICP-MS). Considerable amounts of platinum were present in all fractions from both cell groups (Fig. 1E), suggesting that 5-H-Y and cisplatin are incorporated into nuclei and some of the drug interacts tightly with DNA. Because 5-H-Y was detected in nuclei and found even in the DNA fraction, similar to the case of cisplatin, we next paid attention to DNA replication and RNA transcription.  U2OS, PC9, and TIG-1 cells, using flow cytometry (FCM) (Figs 2A,B and S2). Cisplatin inhibited the incorporation of a thymidine analog, 5-ethynyl-2′ -deoxyuridine (EdU), into newly synthesized DNA, suggesting that DNA replication was inhibited (Figs 2A,B and S2). Consistently, cisplatin-treated cells were arrested in the S/G2 phase (Figs 2A and S2A). In 5-H-Y-treated cells, a 3-to 10-fold reduction of EdU incorporation and cell cycle arrest in S/G2 were observed, similar to the effects of cisplatin (Figs 2A and S2A). Furthermore, we synchronized HeLa cells at the G1/S phase boundary before treatment with 5-H-Y for 15 h (Experimental scheme is shown in Fig. S3). EdU incorporation was almost completely inhibited after release from the G1/S block (Fig. 2C). These results suggest that 5-H-Y has an inhibitory effect on DNA replication, as does cisplatin.

5-H-Y inhibits
While we observed a similar inhibition of DNA replication between the cells treated with 5-H-Y and cisplatin for 15 h (Fig. 2C) or 24 h ( Fig. 2A,B), we found that 5-H-Y has a more rapid effect on DNA replication than cisplatin (Fig. S4). When we treated HeLa cells with the drugs for 2 h, 5-H-Y inhibited DNA replication more severely than cisplatin (Fig. S4). This effect was also observed in other human cell lines (Fig. S4B).

5-H-Y reduces RNA transcription.
Since it was reported that cisplatin could inhibit RNA transcription e.g. ref. 8, we examined effect of 5-H-Y on RNA transcription by incorporation of 5-Ethynyl uridine (EU). 5-H-Y treatment decreased the EU incorporation into newly synthesized RNA in the cells (Fig. 2D), suggesting that global RNA transcription was reduced in the treated cells. Consistent with the previous reports, the EU incorporation in cisplatin-treated cells was also reduced (Fig. 2D).

5-H-Y induces fewer γH2AX foci than cisplatin.
Next, we examined foci formation of phospho-H2AX (γ H2AX) in the 5-H-Y-treated cells, which are often associated with DNA double-strand breaks (DSBs) 33,34 (Fig. 3A). We observed γ H2AX foci in various 5-H-Y-treated cells, such as HeLa, PC9, and TIG-1 cells, but the foci were significantly fewer and weaker than those observed in cisplatin-treated cells (Figs 3A and S5). In addition, in PC9 cells, the γ H2AX-foci localization seemed to differ between 5-H-Y-and cisplatin-treated cells: the foci with cisplatin were enriched in the nuclear rim, while those with 5-H-Y were localized more uniformly in the nucleoplasm (Fig. S5). Furthermore, when we examined checkpoint activation by hyperphosphorylation of the checkpoint mediator Chk1 in the 5-H-Y treated cells, significantly lower levels of Chk1 phosphorylation were seen than in cisplatin-or mitomycin C-treated cells (Fig. 3B). These results suggest that DNA damages induced by 5-H-Y are somehow distinct from those by cisplatin.

5-H-Y provides less amount of DNA crosslinks.
To investigate the DNA crosslinking ability of 5-H-Y, DNA purified from calf thymus was incubated with 5-H-Y or cisplatin for various periods of time. Quantification analysis showed that ~5-fold less 5-H-Y than cisplatin was bound covalently to DNA (Fig. 4A).
Next, we examined the frequency of interstrand DNA crosslink (ICL) formation directly using drug-treated plasmid DNAs (pUC19 and pBluescript II) separated in alkaline agarose gel (Fig. 4B). In both cisplatin-treated plasmid DNAs, there was much more dsDNA (representing ICLs; arrowheads in Fig. 4B) than in 5-H-Y-treated DNAs, suggesting that 5-H-Y induces 2.2-to 5.9-fold less ICL than cisplatin.
Furthermore, we performed semi-quantitative PCR using 5-H-Y or cisplatin-treated DNAs (pUC19 and pBluescript II) as the template (Fig. 4C). The plasmid DNAs were treated with 5-H-Y or cisplatin, purified and used as template DNAs for PCR. In the PCR with the cisplatin-treated template, ~10-fold less PCR products were detected than with 5-H-Y-treated one (Brackets in Figs 4D and S6). Because PCR using mixed DNA templates of cisplatin-treated and untreated plasmids showed successful amplification (the "Cisplatin + Control" lanes in Fig. 4D), DNA crosslinks, not DNA polymerase inhibition by cisplatin, suppressed the PCR reaction. The two DNA templates, pUC19 and pBluescript II, produced similar results, showing that the result had no DNA sequence dependency (Fig. 4D). We concluded that 5-H-Y generates intra-and interstrand crosslinks with much lower frequency compared to cisplatin.

5-H-Y binds tightly to chromatin DNA and folds chromatin in vitro and in vivo. How does 5-H-Y
inhibit DNA replication and RNA transcription? Because 5-H-Y is positively charged and induces compaction of naked DNA 26 , we examined the effects of 5-H-Y on higher-order chromatin structure. To quantitate chromatin structure in solution in vitro, arrays of 12 positioned nucleosomes were reconstituted from pure histones and DNA as a model chromatin (Fig. 5A, top left), followed by sedimentation velocity experiments in an analytical ultracentrifuge (Fig. 5A, bottom left). The degree of folding of the 12-mer nucleosomal arrays was described quantitatively by the sedimentation coefficient (S) 35 . The extended beads-on-a-string conformation sediments at ~29 S, whereas folding causes the nucleosomal arrays to become compact and increases the sedimentation coefficient to ~40-55 S 35 . When the nucleosomal arrays were exposed to 5-H-Y, the sedimentation coefficient increased, from 27 S to 40-55 S, in a dose-dependent manner. In contrast, cisplatin did not affect the sedimentation of the nucleosomal arrays (Fig. 5B). These results indicate that 5-H-Y, but not cisplatin, induced folding of nucleosomal arrays in vitro.
To further investigate nuclear chromatin folding by 5-H-Y, we used permeabilized human cell nuclei attached to glass surfaces (Fig. 5A, right) 31 . Because chromatin is negatively charged, the compaction states of nuclei and their chromatin depend on the cation concentration in the environment 31,36 . For example, in low cation environments (e.g., low Mg 2+ concentration), nuclear chromatin unfolds, leading to an expansion of nuclear volume 31 . However, nuclear chromatin in the presence of a cation (e.g., 5 mM Mg 2+ ) becomes highly condensed and the nuclear volume decreases 31 . As shown in Fig. 5C, nuclear volume, measured with a confocal laser scanning microscope, decreased with the addition of 5-H-Y in a concentration-dependent manner. Our results indicate that 5-H-Y can induce the folding of nuclear chromatin. Notably, permeabilized nuclei pre-treated with 5-H-Y did not increase in volume even after washing with low-salt buffer, while nuclei pre-treated with 5 mM Mg 2+ unfolded greatly after washing (Fig. 5D). This suggests that 5-H-Y binding to nuclear chromatin DNA is quite tight and not a simple electrostatic attraction.
Next, we tested whether 5-H-Y could condense chromatin in vivo (Fig. 5E). To clearly visualize the chromatin condensation in vivo, chromatin in HeLa cells was decondensed by treatment with the HDAC inhibitor trichostatin A (TSA) 37 . When treated with 5-H-Y, the HeLa cells showed prominent chromatin condensation, especially around the nuclear periphery and nucleoli in the cells (Fig. 5E)   Values below the gel indicate the fluorescent intensities of the PCR product normalized by that of control. In the "Cisplatin + Control" lanes, PCR was performed using mixed templates of cisplatin-treated and no-treated plasmids. Note that PCR using cisplatin-treated template produced much less product. deficient in the FANC genes, the products of which are involved in ICL repair. Thus, we examined whether 5-H-Y had a different reaction in such cells. For this purpose, we took a genetics approach using chicken DT40 cells, the genes of which can be modified efficiently using homologous recombination-mediated targeting 38 .
5-H-Y and cisplatin inhibited wild-type DT40 cell growth in a similar manner (Fig. S1B). Then we examined the cell viability of DT40 cells lacking one of the FANC genes, FANCD2, by colony formation assays in the presence of 5-H-Y or cisplatin 39 . FANCD2-KO cells showed no hypersensitivity to 5-H-Y while just 2 μM cisplatin was enough to completely inhibit colony formation (Fig. 6A). A similar tendency was also observed using the FANCC-and FANCJ-KO DT40 cells although they seem to be more sick and more sensitive to any perturbations than FANCD2-KO cells 40,41 (Fig. S7). Taken together with the in vitro data, our results demonstrated that 5-H-Y has a different cytotoxic mechanism than cisplatin: DNA damage by 5-H-Y is repaired by different pathways from ICL repair. Furthermore, 5-H-Y can be effective in cells with acquired resistance to cisplatin (Fig. 6B). Generally, tumor cells with the BRCA2 mutation show hypersensitivity to ICL-inducing agents, such as cisplatin 42 . However, such tumor cells ultimately develop cisplatin resistance 11 . For example, a BRCA2-mutated breast cancer cell line, HCC1428, partially acquired resistance to cisplatin by a secondary genetic change in BRCA2 that rescued BRCA2 function 11 . We found that HCC1428 cells still had higher sensitivity to 5-H-Y than cisplatin (Fig. 6B). Consistently, a previous report showed that 5-H-Y killed cisplatin-resistant types of PC-9 and PC-14 cells more efficiently than cisplatin 27 . These findings suggest that 5-H-Y can effectively suppress proliferation of cisplatin-resistant cancer cells.

Discussion
Using various techniques, we demonstrated that the azolato-bridged complex 5-H-Y is incorporated into nuclei (Fig. 1) and inhibits DNA replication and RNA transcription, arresting the treated cells in the S/G2 phase (Figs 2 and 6C). 5-H-Y binds tightly to chromatin DNA and clearly induces chromatin folding in vitro and in vivo (Figs 5 and 6C). In addition, 5-H-Y has much less intra-and interstrand crosslinking ability than the commonly used anti-cancer drug cisplatin (Fig. 4). These results are consistent with genetic data that have shown that DNA damage induced by 5-H-Y is not processed by the FA/BRCA pathway, which plays an important role in the repair of cisplatin-induced ICL (Figs 6A and S7). Moreover, 5-H-Y can suppress proliferation of cisplatin-resistant cancer cells (Fig. 6B) 27 . Our study provides a mechanistic insight into the differences between 5-H-Y and cisplatin. 5-H-Y may be effective against chemotherapy-insensitive cancers, especially against platinum-refractory cancers, and could be a promising alternative to platinum-based drugs.
Regarding the inhibition mechanisms of DNA replication and RNA transcription, chromatin folding by 5-H-Y could contribute to the processes. Although the higher-order chromatin structure is not fully understood, recent evidence, including our own, suggests that interphase chromatin forms numerous condensed chromatin domains 43,44 , consisting of irregularly folded nucleosome fibers 36,[45][46][47] . Because DNA replication and RNA transcription might occur at opened chromatin at the surface or outside of such compact domains 48-50 , we propose that 5-H-Y can inhibit the opening of chromatin and subsequent initiation processes in treated cells.
Another possibility is that the tight binding of 5-H-Y to chromatin DNA and stabilization of the DNA duplex inhibit the DNA replication and RNA transcription processes directly. Recently, one azolato-bridged complex, [{cis-Pt(NH 3 ) 2 } 2 (μ-OH)(μ-pyrazolato)] 2+ , was shown to stay in the AT-tract minor groove of DNA by non-covalent interactions (unpublished result, Komeda et al.). 5-H-Y may also be a minor-groove binding agent and may act in a similar way to minor-groove binders such as netropsin and distamycin A 51 , which can stabilize the DNA duplex to suppress the unwinding of DNA, a critical first step in the DNA replication and RNA transcription.
Inhibition of DNA replication by non-covalent DNA binding could be advantageous over other cancer chemotherapy agents, because covalent modification of DNA may alter genomic information (the DNA sequence) during the DNA repair process in an irreversible way, leading to the production of abnormal proteins and also drug-induced tumorigenesis. Given that cytotoxicity by 5-H-Y is assumed to change less genome DNA sequences in non-cancer cells, genome integrity could be maintained better in such cells. Efficient PCR amplification using 5-H-Y-treated template DNA (Figs 4D and S6) supports this notion.
To date, most studies on platinum anticancer agents have focused on the binding modes and geometries of covalently bound platinum-DNA adducts: DNA intra-or interstrand crosslinks. However, it has also been suggested that non-covalent interactions between cationic platinum(II) complexes and DNA are sufficient to exhibit cytotoxic effects 52,53 . Accordingly, we may have found a new "pharmacophore" in non-covalent platinum-DNA interactions, with which 5-H-Y binds tightly to DNA and induces a change in chromatin structure. On the other Nuclei treated with 50 μM 5-H-Y showed a 12-fold decrease in the volume. This indicates that 5-H-Y induces chromatin folding. The nuclei treated with 5 mM Mg 2+ were prepared as a control for the nuclei with highly folded chromatin. The error bars represent the standard deviation. For each point, n = ~100. (D) Volumes between Mg 2+ -pretreated and 5-H-Y-pretreated nuclei after buffer washing. When the volume was normalized by Mg 2+ -pretreated nuclei, although Mg 2+ -pretreated nuclei became large after the washing (relative nuclear volume = 1), 5-H-Y-pretreated nuclei did not change (~0.1). 5-H-Y seems to bind tightly to chromatin DNA, in contrast to Mg 2+ . The error bars represent the standard deviation. (E) 5-H-Y induces chromatin condensation in vivo. HeLa cells were treated with TSA to decondense chromatin and then with 5-H-Y. 5-H-Y induced enrichment of chromatin at nuclear periphery and nucleoli (left) although we cannot exclude the possibility that condensation by 5-H-Y only occur around nucleoli and nuclear periphery. Right plot shows the intensity quantification of nuclear periphery chromatin. **p < 0.01, Chi-square test.
Scientific RepoRts | 6:24712 | DOI: 10.1038/srep24712 hand, the possibility of contribution of covalent platinum-DNA adducts by 5-H-Y might also not be excluded, because 1,2-intrastrand crosslinks, probable covalent DNA adducts of one of the azolato-bridged complexes, are recognizable by DNA repair systems less efficiently than those of cisplatin 25 . Our observations of fewer and weaker γ H2AX foci and a lower level of activated Chk1 in the 5-H-Y-treated cells versus cisplatin-treated cells might support this possibility.
Our study has provided a mechanistic insight into the actions of 5-H-Y, which are directly related to its effects on cisplatin-resistant cancer cells and in vivo antitumor efficacy against chemotherapy resistant cancers, such as pancreatic cancer. Azolato-bridged complexes are among the most promising anticancer drug candidates.

Methods
Chemicals. 5-H-Y was prepared as reported previously 20 . Cisplatin was purchased from Bristol-Myers Squibb.
Cell proliferation assay. Cells were seeded in 6-well plates (1 × 10 5 or 2 × 10 4 cells/mL) with various concentrations (0-4 μM) of 5-H-Y or cisplatin. The numbers of proliferated viable cells were examined microscopically at several time points, as indicated in each figure.
Cell viability assay. Serially diluted cells were plated in medium containing 1.5% methylcellulose. To measure sensitivity to 5-H-Y or cisplatin, exponentially growing cells were incubated in methylcellulose medium with the drugs. Colonies were counted after incubation for 1-2 weeks. Immunofluorescence staining and immunoblotting. Immunofluorescence staining was performed as described previously ref. 65. The primary antibody, anti-phospho H2AX (Ser139) mouse monoclonal (Upstate), and the secondary antibody, Alexa-Fluor-594-conjugated goat anti-mouse IgG (Invitrogen), were used at dilutions of 1:3000 and 1:1000, respectively. The samples were analyzed under a DeltaVision microscope (Applied Precision). Images were analyzed using the ImageJ software 56 .

Sedimentation Velocity of Nucleosomal Arrays and 5-H-Y.
Nucleosomal arrays were assembled as described 57 , using a 12-mer 601 58 DNA template and native chicken core histone octamers 59 . Samples were prepared for the analytical ultracentrifuge by diluting to an absorbance of approximately 0.6 at 260 nm, and the 5-H-Y or Cisplatin added to the appropriate concentration.
Sedimentation velocity experiments were conducted in a Beckman XL-A/I analytical ultracentrifuge at 17,000 RPM using absorbance optics as described 60 . The scans were analyzed using the enhanced van Holde-Weischet method 61 implemented in the Ultrascan II data analysis software 62 to yield an integral distribution of diffusion-corrected sedimentation coefficients.
Chromatin compaction assay by measurement of nuclear volume. For condensed chromatin, isolated nuclei (~1 × 10 7 ) were suspended in HM buffer (10 mM HEPES-KOH, pH 7.4, and 5 mM MgCl 2 ) and attached to poly L-lysine-coated coverslips by centrifugation (400 × g, 5 min) 31 . For decondensed chromatin, the nuclei on the coverslips were gently transferred to HM buffer or 1 mM EDTA buffer (pH 8.0). The nuclei were treated with 5-H-Y overnight at room temperature in the dark. Hereinafter, all solutions included 5-H-Y. After fixation with 1% formaldehyde, the nuclei were washed with 50 mM glycine and stained with 2 μM TO-PRO-3 solution (Invitrogen) at 37 °C for 30 min. After washing, z-stack images were acquired using an LSM510 META laser scanning confocal microscope (Carl Zeiss, Wetzlar, Germany) with a 100 × objective at 0.48 μm intervals. The images were processed using the LSM Image Browser (Carl Zeiss) and ImageJ software 56 .
To examine whether buffer washing removed 5-H-Y from chromatin, 5-H-Y-pretreated chromatin or 5 mM Mg 2+ -pretreated chromatin were further washed with 1 mM EDTA without 5-H-Y three times. Then, nuclear volumes were measured as described above and normalized by the average volume of 5 mM Mg 2+ -pretreated nuclei. At  chromatin in the cells (e.g. Ref. 37). Then 10 μM 5-H-Y or cisplatin was added to the TSA-treated cells and further incubated for 1 h. The cells were observed by live cell imaging with a fluorescent microscope (Nikon Eclipse Ti2000-E). We used oblique illumination microscopy 63,64 . For the quantification of condensation, fluorescent intensity of the nuclear rim (average width of 5 pixels, 320 nm) and the nucleoplasm (average width of 10 pixels, 640 nm) were measured by line scan method. The induced condensation was evaluated by the ratio of nuclear rim intensity to nucleoplasm intensity. These analyses were performed with the ImageJ software 56 . The statistical significance was evaluated by Chi-square test. PCR using drug-treated plasmid DNA. pUC19 and pBluescript II were linearized by EcoRI digestion, recovered by ethanol precipitation, and treated with 2.5 μM cisplatin or 5-H-Y (DNA base:drug = 30:1) for 24 h or 48 h at 37 °C. After purification using Wizard SV Gel and the PCR Clean-Up System (Promega), 50 ng purified plasmid was used as templates for PCR with the following set of primers: primer F, AGCAAAAACAGGAAGGCAAA and primer R, ACTGGCCGTCGTTTTAC. PCR was performed with the KOD-Plus kit (Toyobo) according to the manufacturer's protocol. The cycle numbers used were 4, 7, and 10 cycles. The PCR products were electrophoresed on 0.8% agarose gels and stained with EtBr to visualize DNA.

Observation of chromatin compaction in vivo.
Alkaline agarose electrophoresis. To detect interstrand crosslinks in drug-treated DNAs, alkaline agarose gel electrophoresis was carried out. pUC19 and pBluescript II were linearized by EcoRI digestion, recovered by ethanol precipitation, and treated with 2.5 μM cisplatin or 5-H-Y (DNA base:drug = 30:1) for 24 h or 48 h at 37 °C. After purification using Wizard SV Gel and the PCR Clean-Up System (Promega), 1 μg of each purified plasmid was mixed in a buffer containing 50 mM NaOH, 1 mM EDTA (pH 8.0), 3% (w/v) Ficoll, and 0.0425% (w/v) xylene cyanol. The plasmid samples were electrophoresed on 0.8% alkaline agarose gel in 50 mM NaOH and 1 mM EDTA (pH 8.0). After electrophoresis, the gel was neutralized in a buffer containing 1 M Tris-HCl (pH 7.6) and 1.5 M NaCl for 45 min and stained with 0.3 μg/mL EtBr in 1 × TAE buffer (40 mM Tris, 20 mM sodium acetate, and 1 mM EDTA, pH 8.0).

The methods of the following 4 issues are described in the Supplementary Methods.
• Scanning X-ray fluorescence microscopy (SXFM).
• Measurement of cellular platinum by ICP-MS.
• In vivo EdU labeling.