Cisplatin Sensitivity in Hmgb1−/− and Hmgb1+/+ Mouse Cells*

The study presented here investigates the effect of HMGB1 knockout on the sensitivity of mouse embryonic fibroblasts treated with the anticancer drug cisplatin. We evaluated both the growth inhibition by cisplatin and cisplatin-induced cell death in the Hmgb1−/− cells and its wild-type counterpart. No significant differences were observed in the responses of these cells to cisplatin, indicating that HMGB1 does not play a significant role in modulating the cellular responses to cisplatin in this context. Since HMGB1 significantly enhances the cytotoxicity of cisplatin in other cells, these results illustrate the importance of cell type in determining the ability of this and probably other cisplatin-DNA-binding proteins to influence the efficacy of the drug.

lesions and restoration of genomic integrity. Studies have suggested that increase in the repair of platinum-DNA adducts is key to cisplatin resistance (3). Several proteins not involved in repair, including HMGB1, TATA-binding protein, and other structure-specific recognition proteins, also bind tightly to the major platinum-DNA adducts (1). The role that these proteins might play in mediating the cytotoxicity of cisplatin is a subject of much current interest. One hypothesis, which has been proposed in various studies, suggests that the binding of these proteins to platinum-DNA adducts blocks the removal of DNA lesions, thereby enhancing the sensitivity of cells to cisplatin (8 -14). This model is termed repair shielding. HMGB1, an abundant chromosomal protein in mammalian cells, interacts with platinum-DNA intrastrand d(GpG) and d(ApG) crosslinks and interferes with their repair in vitro. It is thus one of the candidate proteins for participation in the repair shielding mechanism (11,15). HMGB1 belongs to the family of proteins that contain at least one HMG box domain, an 80-amino acid DNA-binding motif that recognizes bent DNA structures (16,17). HMGB1 is highly conserved in mammals with Ͼ95% amino acid identity between rodent and human forms (16). HMGB1 is an extremely versatile protein. It increases transcription activation involving steroid hormone receptors, p53, Hox, and Pou protein (18 -20). The regulation of transcription activation by HMGB1 is attributed to its DNA binding and bending activity. Proteinprotein interactions between HMGB1 and the transcription factors presumably facilitate the formation of high order nucleoprotein complexes required for transcription initiation (for a review, see Ref. 21). Hmgb1 knockout mice die shortly after birth due to hypoglycemia resulting from a defect in transcription activation by the glucocorticoid receptor (22). Beside its intranuclear roles, HMGB1 was also discovered recently to function as an extracellular signaling molecule during inflammation, cell differentiation, cell migration, and tumor metastasis (23)(24)(25). HMGB1 is secreted by certain cells, including macrophages and monocytes (21). The secretion and signaling mechanism by which HMGB1 activates cells during these processes is largely unknown.
HMGB1 has been linked to cisplatin activity in a number of studies. HMGB1 binds specifically to cisplatin-modified 1,2intrastrand d(GpG) and d(ApG) cross-links but not to the DNA adducts formed by the clinically inactive isomer, transdiamminedichloroplatinum(II) (15,26). Addition of HMGB1 to an in vitro nucleotide excision repair assay inhibited the repair of cisplatin-DNA damage (11,14). Given the abundance of HMGB1 in nuclei, one copy per kb of the human genome, and its roles in various biological processes, it was hypothesized that the cellular HMGB1 level might modulate the cisplatin sensitivity of cancer cells. Consistent with this notion, hormone-induced HMGB1 up-regulation in MCF-7 breast cancer cells correlates with enhanced cisplatin sensitivity (27). Increased cisplatin sensitivity was also observed in a lung adenocarcinoma cell line transfected with a plasmid expressing HMGB2, a protein with more than 80% identity to HMGB1 (28). A recent study, however, showed that loss of NHP6A, an abundant HMG box protein in Saccharomyces cerevisiae, sensitized yeast to cisplatin, suggesting that a mechanism other than repair shielding was in effect in the yeast system (29).
To investigate further how HMGB1 might affect cisplatin sensitivity in mammalian cells, we studied the cisplatin sensitivity and cisplatin-induced apoptosis in the mouse Hmgb1 Ϫ/Ϫ cell line. Here, we report the results of this work and discuss its implications. Our studies reveal little difference in the cisplatin sensitivity and cisplatin-induced apoptosis between the Hmgb1 Ϫ/Ϫ and its parental cell lines, suggesting that in this mouse model, HMGB1 does not modulate cellular sensitivity to cisplatin. To our knowledge, this is the first direct examination of the impact of HMGB1 loss on cisplatin sensitivity in mammalian cells.
Western Blotting-Cell extracts were prepared from near confluent cells on 100-mm plates. Cells were washed with phosphate-buffered saline and lysed in 250 l of buffer containing 2% SDS, 10 g/ml of pepstatin, leupeptin and aprotinin, and 0.5 mg/ml Pefabloc in phosphate-buffered saline. The lysate was pulled through a 25G 5/8 syringe needle to shear genomic DNA and was cleared by centrifugation. The protein concentrations of extracts were determined by a bicinchoninic acid protein assay kit (Sigma). The proteins in cell extracts were resolved on 12% SDS-PAGE gels and then electroblotted onto polyvinylidene difluoride membranes (Bio-Rad). Following blocking with a 5% bovine serum albumin and 0.01% Tween 20 solution in Tris-buffered saline at pH 7.4, membranes were incubated in the primary antibody, rabbit ␣-HMGB1 (1:2,000 dilution) (Pharmingen), in 0.01% Tween 20 at pH 7.4 with Tris-buffered saline. Finally, blots were incubated in the secondary antibody, 1:2,000 dilution of ␣-rabbit whole antibody conjugated to horseradish peroxidase (Amersham Biosciences). After washing, blots were soaked for 1 min in chemiluminescent reagents (Sigma) and then exposed to BioMax film (Eastman Kodak Co.).
Growth Inhibition Assay-Cells were seeded on 96-well plates at a density of 1,000 cells/100 l/well. Treatment with cisplatin began the next day after cells had attached to the plates. For continuous treatment, cells were incubated in growth medium containing various concentrations of cisplatin for 72 h. For short time treatment, cells were treated with cisplatin-containing medium for 4 h followed by incubation in drug-free medium for an additional 90 h. After incubation, the cell density of each sample was determined by using WST-1, a tetrazolium salt (Chemicon International Inc.). The assay was carried out according to the manufacturer's protocol. Briefly, 10 l of WST-1 dissolved in Electro Coupling Solution (Chemicon International Inc.) was added to each well of 96-well plates that contained cells at the end of the incubation. Plates were incubated under standard growth conditions for 1.5 h, and the absorbance at 440 nm was recorded by using an absorbance microplate reader, SpectroMax 340pc (Molecular Devices). Percentages of surviving cells were calculated by the ratio of absorbance of cisplatin-treated cultures over that of untreated control culture. Cisplatin kill curves were obtained by plotting the percentage of survival against the concentration of cisplatin. In addition, a second set of cultures was analyzed by using sulforhodamine B (SRB). First, 10% cold trichloroacetate solution was added to the cells. Plates were incubated for 30 min at 4°C to allow fixation to come to completion and then washed five times with distilled water. Plates were air-dried at room temperature in the hood. Trichloroacetate-fixed cells in each well were then stained with 100 l of 0.4% (w/v) SRB in 1% acetic acid for 30 min. Excess SRB was removed by washing the plates four times with 1% acetic acid. Plates were air-dried until no remaining moisture was visible. The bound SRB was solubilized by adding 100 l of 10 mM Tris base (pH 10.5) to each well and shaking for 5 min on a shaker platform. The absorbance at 564 nm was recorded by using the microplate reader. Cisplatin kill curves were obtained as described above.
Annexin V Assay-A portion (1.5 ϫ 10 5 ) of cells was plated on 60-mm plates and allowed to attach overnight. Cells were then incubated in cisplatin-containing media for 34 h. After incubation, the attached cells in each plate were collected by trypsinization and combined with the detached cells of the same sample. Cells were washed twice with cold phosphate-buffered saline and resuspended in the binding buffer (Pharmingen). Cell suspensions were counted with a hemocytometer and diluted to a final concentration of 1 ϫ 10 6 cells/ml. An aliquot of cells (100 l) was labeled with propidium iodide (PI) and fluorescein isothiocyanate-conjugated annexin V (annexin V-FITC) according to the manufacturer's instructions. The labeled cell suspensions were analyzed by a flow cytometer, FACScan, equipped with a 488-nm argon laser light source (BD Biosciences). The emission filter used was 515-545 nm for FITC and 563-607 nm for DNA⅐PI complexes. Data were analyzed with CellQuest software. Cell debris was excluded from data analysis.

RESULTS
Cisplatin Sensitivity of Hmgb1 ϩ/ϩ and Hmgb1 Ϫ/Ϫ Cells-Two independent methods were used to compare the cisplatin sensitivity of the Hmgb1 Ϫ/Ϫ and wild-type mouse cells. The WST-1 assay measures the live cell population because WST-1 is quantitatively reduced to a colored formazan in viable cells (30,31). The SRB assay measures cell density by quantitating colored sulforhodamine B bound to cells fixed to the plates by trichloroacetate (32,33). Similar results were obtained by using the two methods (Fig. 1). Cisplatin inhibited the growth of Hmgb1 ϩ/ϩ and Hmgb1 Ϫ/Ϫ cells to a similar extent when added to growth medium for 72 h (Fig. 1, A and B). Likewise, a 4-h cisplatin treatment followed by an 84-h incubation in drug-free medium equally inhibited the growth of both cell lines (Fig. 1, C and D). No significant difference between the cisplatin sensitivity of Hmgb1 ϩ/ϩ and Hmgb1 Ϫ/Ϫ cells was observed in repeated trials.
Cisplatin-Induced Apoptosis in Hmgb1 ϩ/ϩ and Hmgb1 Ϫ/Ϫ Cells-A common cellular response to cisplatin exposure is activation of the apoptotic pathway (34). In the present study, the apoptosis of Hmgb1 ϩ/ϩ and Hmgb1 Ϫ/Ϫ cells induced by cisplatin was probed by labeling with PI and annexin V-FITC. Early apoptotic cells lose membrane phospholipid asymmetry and expose phosphatidylserine on the outer leaflet of the plasma membrane. Previous work has demonstrated that phosphatidylserine externalization in apoptotic cells can be labeled with annexin V-FITC and quantitated by flow cytometry (35)(36)(37). The late apoptotic cells and dead cells can also be labeled with PI in addition to the annexin V labeling because of damaged cell membrane. The annexin V-FITC/PI assay has been successfully applied to study cisplatin-induced apoptosis in a number of cell lines (35,38,39). Here, we used Western blotting to compare the levels of HMGB1 in the mouse fibroblasts with those in MCF-7 cells, a human breast cancer line. The same total amount of cellular proteins was subjected to SDS-PAGE separation and Western analysis using the HMGB1 antibody. The Western blot showed similar levels of HMGB1 expression in the mouse fibroblasts and MCF-7 cells (Fig. 4). DISCUSSION The goal of this study was to explore the connection between HMGB1 levels and the sensitivity of mammalian cells to cisplatin. We addressed this question by comparing the response of wild-type cells with those of Hmgb1 knockout mouse cells established previously (22). Hmgb1 Ϫ/Ϫ mice die shortly after birth due to a defect in the transcriptional activation of the glucocorticoid receptor, consistent with the notion that HMGB1 functions as a regulator of transcription involving steroid hormone receptors. The HMGB1 deficiency is also manifested at the cellular level. The cultured Hmgb1 Ϫ/Ϫ embryonic fibroblasts are defective in transcription activation by the glucocorticoid and progesterone receptors (22). The Hmgb1 Ϫ/Ϫ cells have a morphology and growth rate similar to those of the Hmgb1 ϩ/ϩ cells, and the levels of Hmgb2 and Hmgb3 mRNA are comparable in both cell lines (22). The Hmgb1 Ϫ/Ϫ and the wild-type cell lines thus provide an adequate system for investigating the involvement of HMGB1 in cellular responses to cisplatin.
Both the growth inhibition by cisplatin and cisplatin-induced apoptosis in the Hmgb1 Ϫ/Ϫ and the wild-type cells were examined. Two independent methods (see "Experimental Procedures") revealed that continuous treatment with cisplatin inhibited the growth of Hmgb1 Ϫ/Ϫ cells to the same extent as that of Hmgb1 ϩ/ϩ cells (Fig. 1). When cells were treated with cisplatin for a short time (4 h), the survival rates of both cell lines were also similar (Fig. 1). Furthermore, the cisplatin-induced apoptosis in these cells was investigated by using annexin V-FITC/PI labeling. In cisplatin-treated cell cultures of both lines, there was a significant apoptotic (annexin V-positive) but very little necrotic population (annexin V-negative/PI-positive), suggesting that the cells die via the former pathway (Fig. 2). Increasing cisplatin concentration in the cultures resulted in increasing apoptosis in both cell lines, confirming that apoptosis of the cells was induced by cisplatin ( Figs. 2 and 3). Comparison of the apoptotic population in the Hmgb1 Ϫ/Ϫ and Hmgb1 ϩ/ϩ cultures treated with the same concentrations of cisplatin showed essentially the same degree of cell death in both cell lines (Fig. 3). Collectively, these results indicate that the Hmgb1 Ϫ/Ϫ cells are as sensitive to cisplatin as the Hmgb1 ϩ/ϩ cells.
HMGB1 has been implicated in the cytotoxicity of cisplatin since the discovery of its ability to bind cisplatin-modified DNA in vitro. A number of experiments (42)(43)(44)(45)(46)(47) have discerned the molecular interactions between HMGB1 and cisplatin-modified DNA. Few studies, however, investigated the effect of HMGB1 on cisplatin sensitivity at the cellular level. The present results suggest that HMGB1 does not play a significant role in the mechanism of cisplatin-induced cytotoxicity in the mouse embryonic cells. Given the high binding affinity of HMGB1 for cisplatin-modified DNA and the abundance of HMGB1 in mammalian cells, this observation seems surprising. There are, however, a number of scenarios that might prevent HMGB1 from interacting with cisplatin-modified DNA in the cells and modulating their responses to cisplatin. One possibility is that the HMGB1 level in the mouse embryonic fibroblasts is considerably lower than that of cells in other tissues or animals, such as MCF-7 human breast cancer cells, the subject of a previous study (27). A Western analysis for HMGB1 levels in both cell lines excluded this possibility. More likely is that HMGB1 proteins engage in tight interactions with other targets in the cells. HMGB1 contains two tandem HMG boxes and an extremely acidic C-terminal domain (21), through which it interacts with several other proteins (19,20,41). In the mouse fibroblasts, HMGB1 proteins may be engaged in tight nucleoprotein complexes and may not be available for ready binding to cisplatin-modified DNA. A recent study discovered that during apoptosis induced by tumor necrosis factor ␣, HMGB1 became firmly attached to the hypoacetylated chromatin. The binding of HMGB1 to apoptotic chromatin may similarly block the interaction of HMGB1 to platinum-DNA adducts (23). Finally, as yet unidentified proteins in these embryonic cells may dominate the interaction with cisplatin-modified DNA. A comprehensive analysis of the correlation between cisplatin sensitivity and protein expression levels in cells may lead to the discovery of such protein candidates. In any event, our results stress the importance of cell type in determining the ability of this and possibly other platinated DNAbinding proteins to influence the effectiveness of cisplatin as a cytotoxic agent. CONCLUSION HMGB1 has been postulated to influence cellular sensitivity to cisplatin through specific binding to cisplatin-DNA 1,2-intrastrand cross-links. Our results reveal little difference in the cisplatin sensitivity and cisplatin-induced apoptosis in embryonic native and Hmgb1 Ϫ/Ϫ cell lines, suggesting that in this mouse model, HMGB1 does not modulate the cellular sensitivity to cisplatin.