Interaction of FACT, SSRP1, and the High Mobility Group (HMG) Domain of SSRP1 with DNA Damaged by the Anticancer Drug Cisplatin*

The structure-specific recognition protein SSRP1, initially isolated from expression screening of a human B-cell cDNA library for proteins that bind to cisplatin (cis -diamminedichloroplatinum(II))-modified DNA, contains a single DNA-binding high mobility group (HMG) domain. Human SSRP1 purifies as a heterodimer of SSRP1 and Spt16 (FACT) that alleviates the nucleosomal block to transcription elongation by RNAPII in vitro . The affinity and specificity of FACT, SSRP1, and the isolated HMG domain of SSRP1 for cisplatin-dam-aged DNA were investigated by gel mobility shift assays. FACT exhibits both affinity and specificity for DNA damaged globally with cisplatin compared with unmodified DNA or DNA damaged globally with the clinically ineffective trans -DDP isomer. FACT binds the major 1,2-d(GpG) intrastrand cisplatin adduct, but its isolated SSRP1 subunit fails to form discrete, high affinity complexes with cisplatin-modified DNA under similar conditions. These results suggest that Spt16 primes SSRP1 for cisplatin-damaged DNA recognition by unveiling its HMG domain. As expected, the isolated HMG domain of SSRP1 is sufficient for specific binding to cisplatin-dam-aged domain and were performed as follows. To assay for protein recognition of cisplatin-damaged DNA, cisplatin-modified or unmodified 127-base pair duplexes (0.2 n M , 40,000 cpm) were titrated with protein in buffer PP250 (10 m M Tris, pH 7.5, 250 m M NaCl, 10 m M MgCl 2 , 0.5 m M EDTA, 5% glycerol, 1 m M dithiothreitol) containing 0.2 erythrocyte onto or m M m M was fo Toassay for binding of the major 1,2-d(GpG) intrastrand cisplatin-DNA cross- link, AG*G*A and AGGA 15-bp oligonucleotide duplexes cpm) were with (10 m M m M , m M LiCl , m M NaCl, 1 m M spermidine, 0.2 bovine m amount bound and free oligonu- cleotide GS-525 phosphorimaging To examine the affinity and specificity of FACT for cisplatin-dam- aged DNA, 127-bp duplexes (0.02–2 n M , 20,000 cpm) undamaged or globally damaged with cisplatin or trans- DDP, or 156-bp site-specific modified or unmodified duplexes (0.1 n M , 20,000 cpm) were combined with hFACT, rFACT, rSpt16, or rSSRP1 in binding buffer PP50 (10 m M Tris, pH 7.5, 50 m M NaCl, 10 m M MgCl 2 , 0.5 m M EDTA, 5% glycerol, and m M dithiothreitol). Chicken erythrocyte genomic DNA competitor (0.2–2.0 mg/ml) was included in reactions containing globally modified probes incubated on h prior to loading onto running, pre-equilibrated native 4% polyacrylamide gels acrylamide:bisacrylamide, phosphorimaging GS-525 phosphorimaging device

Eukaryotic cells package DNA into chromatin, the structure of which impedes essential cellular processes that require DNA for function. Such processes include replication, recombination, and transcription. The regulation of gene expression is therefore intimately tied to chromatin structure.
DNA transcription in vitro can be reconstituted by a minimal set of general transcription factors and RNA polymerase II (1). These minimal components are insufficient for transcription from reconstituted nucleosomal templates, however, implying the existence of cellular mechanisms for chromatin remodeling to facilitate access of the transcription machinery to DNA (2). Two classes of nuclear factors, ATP-dependent chromatin remodeling enzymes and histone acetyltransferases, allow the transcription machinery to assemble and initiate transcription from chromatin templates (3,4). Recently, a novel factor, FACT 1 (Facilitates Chromatin Transcription), which enables transcription elongation past nucleosomes, was isolated from HeLa nuclear extracts (5,6). This factor is a heterodimer of human Spt16/Cdc68 and human SSRP1 proteins (7). FACT is inactivated by chemical cross-linking of the histone octamer, suggesting that it may function to unravel H2A/H2B histone dimers from nucleosome cores (7). The remaining H3/H4 tetramer is itself insufficient to repress the elongating RNA polymerase (8,9).
FACT is a very abundant nuclear protein complex with an estimated 100,000 copies/HeLa cell (7). The complex is conserved across a diverse range of organisms, analogous complexes of SPT16/SSRP1 homologs having been isolated from Xenopus and Saccharomyces cerevisiae (10 -12). Although distinct roles for the two protein components of FACT have yet to be elucidated, the DNA-binding high-mobility group (HMG) domain of SSRP1 (13) may target the complex to nucleosomes (7). Circumstantial evidence suggests that HMG domain proteins bind to DNA as it enters and exits the nucleosome (14). Consistent with the hypothesis that the HMG domain of SSRP1 is responsible for FACT binding to DNA, FACT activity is abolished by addition of a 5-fold excess of superhelical plasmid DNA competitor (5). The mechanistic details of derepression of transcription elongation by FACT remain to be elucidated.
Human Cdc68/Spt16 is a 119.9-kDa nuclear protein with 36% identity to its yeast homolog. Previous genetic studies with yeast Cdc68 suggested a role in modulating chromatin structure to effect both gene repression and activation (15)(16)(17). The domain features of the 81-kDa human SSRP1 protein are shown in Fig. 1 (18). Initially isolated from expression screening of a human B-cell cDNA library for proteins that bind DNA modified by the antitumor agent cisplatin (cis-diamminedichloroplatinum(II)), SSRP1 is expected to bind distorted DNA structures through its DNA-binding HMG domain (18). The ability of SSRP1 to recognize DNA modified by cisplatin and the deleterious effect of cisplatin on transcription suggest a possible role for SSRP1 and its physiologically relevant complex FACT in the cisplatin anticancer mechanism (19 -21). Although the success of cisplatin therapy on a variety of cancers, including testicular, ovarian, and head and neck, has been remarkable since its introduction in 1979, there are several drawbacks to cisplatin treatment. These include toxic side effects, inherent or acquired resistance, and efficacy in only a handful of tumor types (22). A detailed understanding of its mechanism of action may allow for the rational design of new antitumor therapies.
Cisplatin reacts with a number of cellular components, but it is generally accepted that the biologically relevant target is DNA. After loss of the chloride ligands due to the relatively low concentration of Cl Ϫ ions in the cell, cisplatin coordinates preferentially to the N-7 atoms of two adjacent purine nucleotides by forming mainly intrastrand d(GpG) and d(ApG) DNA crosslinks (23,24). Such adducts bend the duplex toward the major groove and unwind the DNA helix. The minor groove is concomitantly flattened and widened. Delineation of the cellular processing of cisplatin-DNA adducts is of great importance to unraveling its mechanism of cytotoxicity. Therefore, much work has focused on identifying and characterizing proteins that bind 1,2-intrastrand cross-links with high affinity and specificity (25). Structural distortions imposed on DNA by cisplatin 1,2-intrastrand cross-links are recognized by members of the HMG domain protein family including SSRP1 with notable specificity over unmodified double-stranded DNA.
Like other members of the HMG-domain protein family, SSRP1 is expected to recognize cisplatin-damaged DNA through this DNA-binding component. Here we demonstrate the affinity and specificity of FACT for cisplatin-damaged DNA and the major 1,2-intrastrand cisplatin-DNA lesion in particular. Electrophoretic mobility shift assays reveal that both the SSRP1 and Spt16 subunits of FACT are necessary for high affinity binding to cisplatin-modified DNA. The isolated HMG domain of SSRP1, however, is sufficient for binding the major 1,2-d(GpG) intrastrand cisplatin-DNA cross-link. The affinity and specificity of FACT for cisplatin-modified DNA and its role in modulating chromatin structure during transcription suggest that the interaction of FACT and cisplatin-damaged DNA may be important in the cisplatin anticancer mechanism.

EXPERIMENTAL PROCEDURES
Oligonucleotide Probes-A 127-base pair probe was constructed by digesting a commercially available 123-base pair ladder (Life Technologies, Inc.) with AvaI restriction enzyme (New England Biolabs). The digested 123-base pair fragment containing 4-bp 5Ј-overhangs was purified on a 10% native polyacrylamide gel, ethanol-precipitated, and quantitated by UV-vis spectroscopy (Hewlett Packard 8453). The purified probe was modified with cisplatin or trans-DDP at various platinum:nucleotide ratios (r f , added platinum:nucleotide ratio) following published procedures (26). The extent of platination (r b , bound platinum:nucleotide ratio) was determined by flameless atomic absorption spectroscopy (PerkinElmer Life Sciences HGA-800 AAnalyst 300) and UV-vis spectroscopy. Platinated and unplatinated 123-base pair probes having recessed 3Ј-ends were fill-in labeled with Klenow fragment of DNA polymerase (New England Biolabs) and [␣-32 P]dCTP (PerkinElmer Life Sciences), cold dATP, dTTP, and dGTP to give 127base pair oligonucleotide duplexes. Radiolabeled probes were purified with G25 Quick Spin columns (Roche Molecular Biochemicals), ethanolprecipitated, and quantitated by scintillation counting (Beckman LS 6500). The radiolabeled 15-bp oligonucleotide duplexes AG*G*A (d(CCTCTCAG*G*ATCTTC)/d(GAAGATCCTGAGAGG), where asterisks denote the sites of cis-diammineplatinum(II) cross-linking via the N-7 positions of adjacent guanines and AGGA, the unplatinated analog) were prepared according to published procedures (27).
Plasmids-cDNA encoding the HMG (residues 539 -614) domain of SSRP1 was amplified by PCR using human SSRP1 cDNA and the primers 5Ј-GCTCTAGAAAGGAGGTGGAGATGAAGAAGCGCAAAGA-C-3Ј and 5Ј-CTCGCCTCGGCATATGTTAATATTCTTTCATGGCTTT-3Ј. The PCR primers contained restriction sites for XbaI and NdeI (italics) as well as initiation, termination, and ribosome binding sites (underlined); the last, 5Ј-AAGGAG, was positioned nine base pairs upstream of the initiation codon. PCR was performed with Taq DNA polymerase (Life Technologies, Inc.) under the following thermal cycler conditions: 5 min at 94°C (denaturation cycle); 1 min at 94°C, 2 min at 50°C, and 3 min at 72°C (25 cycles); 10 min at 72°C (extension cycle). The resulting fragments were digested with XbaI and NdeI (New England Biolabs), purified by using a PCR purification kit (Qiagen), and ligated into the XbaI and NdeI sites of pET3a (Novagen) to give p3SSRP.d1 for expression of the HMG domain. Correct insertion of the PCR fragment was confirmed by restriction enzyme mapping and DNA sequencing.
Expression and Purification of Recombinant SSRP1 HMG-domain Peptides-BL21(DE3) Codon Plus RIL (Stratagene) cells harboring p3SSRP.d1 were grown at 37°C in LB containing 100 g/ml ampicillin and 20 mM methionine. Protein production was induced at an A 600 of 0.7 by addition of isopropyl ␤-D-thiogalactopyranoside to a final concentration of 1 mM. After an induction period of 5 h, cells were collected by centrifugation, resuspended in cold lysis buffer (50 mM Tris, pH 7.0, 20 mM NaCl, 10 mM EDTA, 10 mM EGTA, 10 mM MgCl 2 , 0.035% ␤-mercaptoethanol, 1 mM Pefabloc, 1 mM DNase I, 2 mM dithiothreitol, 20 mM methionine, 2 mM NaS 2 O 4 ), and lysed by sonication. Debris was removed by centrifugation, and cellular proteins were precipitated with 55% saturated ammonium sulfate. Recombinant protein was precipitated with 98% saturated ammonium sulfate, collected by centrifugation, resuspended in buffer A (20 mM Tricine, pH 8.3, 5 mM dithiothreitol, 1 mM EDTA, 20 mM methionine, 2 mM NaS 2 O 4 ), and dialyzed against the same buffer. The desalted protein solution was loaded onto a cation-exchange SP-Sepharose Fast Flow (Amersham Pharmacia Biotech) column washed with buffer A. Proteins were eluted with a linear gradient of 0 -1 M NaCl in buffer A. Eluted proteins were detected by SDS-PAGE followed by Coomassie staining. Fractions containing SSRP1 HMG domain were pooled, dialyzed, concentrated, and applied to a Superdex 75 column (Amersham Pharmacia Biotech) washed with buffer B (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 , 1.4 mM KH 2 PO 4 , 20 mM methionine, 2 mM NaS 2 O 4 , pH 7.3). Proteins were eluted with column buffer, and fractions containing pure SSRP1 HMG domain were pooled, concentrated, frozen in liquid nitrogen, and stored at Ϫ80°C. SSRP1 HMG domain concentrations were determined from A 280 values by using an extinction coefficient (⑀ 280 ϭ 18,600 M Ϫ1 cm Ϫ1 ) calculated by quantitative amino acid analysis.
Isolation of FACT from HeLa Cells-FACT was isolated from HeLa cells as described (5). The purity of native FACT (hFACT) was assessed by denaturing and native PAGE.
Production of Recombinant SSRP1, Spt16, and FACT-Methods for recombinant SSRP1 (rSSRP1) and Spt16 (rSpt16) production and purification will be reported elsewhere. 3 Recombinant FACT (rFACT) was obtained by mixing equimolar amounts of pure rSSRP1 and pure rSpt16, followed by gel filtration chromatography.
Gel . Electrophoresis was continued for 1.5-2 h at 300 V and 4°C. Following electrophoresis, gels were dried and exposed to a phosphorimaging plate for 12-24 h. The amount of bound and free oligonucleotide was assessed with a Bio-Rad GS-525 phosphorimaging device.

Expression and Purification of SSRP1 HMG Domain-Full-
length SSRP1 contains a single DNA-binding HMG domain flanked by two short basic regions (Fig. 1). The minimal HMG domain was recombinantly expressed and purified. As shown in Fig. 2, the SSRP1 HMG domain was obtained with Ͼ95% purity. Western blots with polyclonal antibodies raised against the full-length hSSRP1 protein confirmed the identity of the recombinant protein. Electrospray ionization mass spectrometry of the protein confirmed the absence of oxidized contaminants. N-terminal sequencing of each protein yielded the predicted amino acid sequence with retention of the initial methionine.
Affinity and Specificity of the HMG Domain of SSRP1 for Cisplatin-damaged DNA-The HMG domain of SSRP1 is expected to confer recognition of cisplatin-modified DNA to the full-length SSRP1 protein. To examine the affinity and specificity of the hSSRP1 HMG domain for cisplatin-modified DNA, a series of 127-bp probes, unmodified or containing various levels of global cisplatin-or trans DDP damage were used in bandshift assays with recombinant hSSRP1 HMG domain. All bandshifts were performed in the presence of a large excess of nonspecific DNA competitor. The recombinant hSSRP1 HMG domain retains affinity for cisplatin-modified DNA, as evi-denced by gel-mobility shifts of the domain with a 127-bp probe damaged globally with cisplatin (Fig. 3A). The interaction between the domain and cisplatin-modified DNA is not a result of general DNA affinity for the probe because the domain fails to bind the unmodified 127-mer under identical conditions (Fig.  3A). The domain does exhibit very low affinity for probes damaged with the clinically ineffective trans-DDP isomer (data not shown); however, protein affinity for the cisplatin-damaged probe is significantly higher than that for the corresponding trans-DDP-damaged DNA.
The mobility of protein-DNA complexes formed upon incubation of hSSRP1 HMG domain with the cisplatin-modified 127-bp probe decreases with increasing protein concentration (Fig. 3A). A similar pattern of mobility shifts occurs when probes with increasing damage levels (0.003-0.044 platinum atoms/nucleotide, 1-12 platinum atoms/duplex DNA) are used (data not shown). Both of these results are consistent with multiple proteins binding to these long probes at high platination or protein levels.
Recognition of the Major 1,2-d(GpG) Intrastrand Cisplatin-DNA Adduct by hSSRP1 HMG Domain-To confirm that the minimal HMG domain of hSSRP1 can bind the major 1,2d(GpG) intrastrand cisplatin-DNA cross-link, the affinity of the domain for the oligonucleotide duplex AG*G*A, 5Ј-CCTCTCAG*G*ATCTTC-3Ј, where asterisks denote the sites of platinum coordination to the N-7 positions of adjacent guanines, was investigated. A representative gel mobility shift with AG*G*A and hSSRP1 HMG domain is shown in Fig. 3B. The domain fails to shift the corresponding unplatinated duplex under identical conditions (data not shown).
Isolation of hFACT from HeLa Cells-Human FACT, the heterodimer of hSpt16 and hSSRP1, was isolated from HeLa cells as described (5). The complex contains hSpt16 and hSSRP1, as well as a minor contaminant at ϳ40 kDa as judged by SDS-PAGE followed by silver staining. This contaminant is not necessary for FACT activity (7). Native PAGE demonstrates that hFACT isolated in this manner contains no free hSSRP1 or hSpt16.
Gel Mobility Shift Assays of hFACT with DNA Damaged with Cisplatin-To assess the ability of hFACT to bind specifically to cisplatin-modified DNA, gel mobility shift assays were performed with a series of platinated or unplatinated 127-bp probes (Fig. 4A). Cisplatin-and trans-DDP-damaged probes used in these experiments were modified at similar levels. absence of competitor; however, the band of lowest mobility observed in lanes 2, 5, and 8 corresponds to the position of free protein (data not shown) and disappears in the presence of excess competitor (Fig. 4A). A specific protein-DNA complex is formed upon incubation of FACT with the 127-bp probe modified globally with cisplatin. This complex persists even in the presence of a 3.6 ϫ 10 5 -fold excess of nonspecific competitor DNA (Fig. 4A, lanes 5 and 6) and remains robust in NaCl concentrations of at least 250 mM (data not shown). The mobility of the putative FACT-cis127-bp complex is unchanged by increasing platinum damage levels within the range of 1-8 platinum atoms/127-bp duplex (data not shown).
Binding of FACT to the Major 1,2-d(GpG) Intrastrand Cisplatin-DNA Cross-link-Native hFACT forms a specific complex with a 156-bp probe containing a single, centered 1,2d(GpG) intrastrand cisplatin-DNA cross-link but not with the corresponding unmodified probe (Fig. 4B).
Binding of Individual FACT Subunits to Cisplatin-modified DNA-Recombinantly produced rSSRP1, as well as rSpt16 and rFACT, were used to confirm that the SSRP1 subunit confers cisplatin-modified DNA binding activity on the complex. Proteins obtained as described under "Experimental Procedures" are pure as judged by SDS-PAGE followed by Coomassie staining (Fig. 5A). Silver-stained native PAGE confirms that both rSpt16 and rSSRP1 exist as monomers (data not shown) and demonstrates that the rFACT preparation contains no free rSSRP1 or rSpt16 (Fig. 5B). Furthermore, rFACT produced in this manner is competent for in vitro RNA polymerase transcript elongation on chromatin templates. 3 The recombinant complex yields a bandshift pattern with the cis127-bp probe identical to that of native FACT (compare Figs. 6A and 4A). As with native FACT, rFACT fails to bind the trans-DDP-modified probe under these conditions (data not shown). In contrast, gel mobility assays with each of the recombinant FACT subunits indicate that the Spt16 subunit fails to bind the cis127-bp probe (Fig. 6B) and that the SSRP1 subunit gives rise to only a very weak shift that disappears in the presence of excess competitor (Fig. 6C). Although native and recombinant FACT form specific protein-cis127-bp complexes at protein concentrations as low as 3 nM (data not shown), rSSRP1 fails to yield a bandshift pattern similar to that of rFACT or native FACT even at protein concentrations of 900 nM (Fig. 6C). DNA damaged globally with cisplatin in filter binding assays (18). This interaction is expected to be mediated by the DNAbinding HMG domain of SSRP1 (Fig. 1). Several proteins containing one or more HMG domains bind to cisplatin-damaged DNA, including HMG1 (32,33), HMG2 (33,34), SSRP1 (18), Ixr1 (35,36), hUBF (37), LEF1 (38), and SRY (39). In particular, full-length HMG1 binds DNA damaged globally with cisplatin but not DNA damaged with its clinically inactive isomer trans-DDP (32). The HMG domains of HMG1, HMG2, SSRP1, and other HMG-domain proteins that bind cisplatin-damaged DNA are likely to be the key elements in protein recognition of cisplatin-DNA lesions. Evidence supporting this conclusion has been described for both HMG domains of HMG1 (27,40). To determine whether the HMG domain of SSRP1 is sufficient for cisplatin-modified DNA binding, a fragment corresponding to residues 539 -614 was expressed and purified from Escherichia coli. This recombinant HMG domain was used in gel mobility shift experiments with globally cisplatin-modified and unmodified probes. The isolated domain is sufficient for specific binding to DNA damaged globally with cisplatin. The domain is selective for cisplatin-modified DNA with respect to both unmodified DNA and DNA containing trans-DDP adducts. In addition, like other isolated HMG domains (27,40), the HMG domain of SSRP1 binds oligonucleotide probes containing the major 1,2-d(GpG) intrastrand cisplatin-DNA cross-link. The affinity of this interaction depends on the DNA sequence flanking the drug-DNA lesion as seen for HMG domains A and B of HMG1 (27).

Interaction of the
Interaction of FACT and SSRP1 with Cisplatin-damaged DNA-Both SSRP1 and its physiologically relevant complex with Spt16, FACT, are expected to bind distorted DNA structures including cisplatin-DNA cross-links by means of the HMG domain of SSRP1. Here we show that FACT has both affinity and specificity for DNA damaged globally with cisplatin with respect to undamaged DNA. A tightly bound complex results from incubation of FACT with cisplatin-modified DNA as evidenced by its resilience in the presence of 3.6 ϫ 10 5 -fold excess of nonspecific DNA or Ͼ50-fold excess of unplatinated competitor DNA. Like other HMG-domain proteins, FACT binds the major 1,2-intrastrand cross-links of cisplatin-modified DNA. The 1,2-intrastrand cisplatin-DNA cross-links are the most abundant DNA adducts formed following cisplatin treatment (41). These adducts are thought to be crucial to cisplatin toxicity because geometric constraints prohibit the clinically inactive trans-DDP from forming 1,2-intrastrand adducts, although trans-DDP can form DNA cross-links similar to the less abundant cisplatin-DNA lesions (21). Notably, FACT fails to bind DNA damaged with the clinically ineffective trans-DDP isomer.
We then sought to confirm that the SSRP1 subunit confers the ability to recognize cisplatin-modified DNA on the FACT complex. As anticipated, Spt16, which contains no putative DNA-binding domains, has no affinity for either cisplatin-damaged or undamaged probes. Unexpectedly, SSRP1 fails to form a high-affinity complex with the same cisplatin-modified probe in gel mobility shift assays under conditions sufficient for hFACT binding. The complex of rSSRP1 and rSpt16, however, is capable of binding cisplatin-damaged DNA with affinity and specificity comparable with hFACT in identical bandshift assays. A Ͼ300-fold excess of rSSRP1 fails to afford a protein-DNA complex of affinity comparable with that of hFACT or rFACT, suggesting that protein concentration discrepancies are unlikely to explain the lack of a bandshift with rSSRP1. The observed differences in affinity between the protein-DNA complex formed by hFACT or rFACT and that formed by SSRP1 suggest that Spt16 primes SSRP1 for cisplatin-modified DNA recognition. We suggest that, in the absence of Spt16, SSRP1 adopts a fold that renders its HMG domain inaccessible to cisplatin-modified DNA. When Spt16 is present, a conformational change occurs unveiling the HMG domain of SSRP1 and leading to the observed binding interaction.
Functional Implications-Because modulation of chromatin structure is essential to many cellular processes that use DNA as a substrate, including replication and transcription, FACT may effect such processes. The specific interaction of SSRP1 and FACT with DNA damaged with the anticancer drug cisplatin as judged by electrophoretic mobility shift assays implicates both SSRP1 and FACT in the mechanism of cisplatin cytotoxicity. The binding of SSRP1 or FACT to cisplatin-damaged DNA could mediate cellular sensitivity to the drug by shielding its DNA cross-links from repair, allowing the drug-DNA lesions to persist and eventually leading to cell death (28,29,35,39,(42)(43)(44). Moreover, presence of cisplatin-DNA adducts could titrate SSRP1 or FACT from its normal binding sites, thereby disrupting SSRP1/FACT function(s) (45,46). Finally, the formation of stable FACT/SSRP1 complexes at cisplatin-DNA cross-links could lead to stalling of the RNAPII transcription machinery, ubiquitination of RNAPII, and proteolysis resulting ultimately in cell death (30,31). Any or all of these activities may contribute to the mechanism of cisplatinmediated cytotoxicity.