Thiol-Modulated Mechanisms of the Cytotoxicity of ... - ACS Publications

Faculty of Pharmacy, UniVersity of Manitoba, 50 Sifton Road, Winnipeg, Manitoba, ... Department of Pharmacology, UniVersity of Pittsburgh School of Me...
0 downloads 0 Views 424KB Size
Chem. Res. Toxicol. 2008, 21, 483–493

483

Thiol-Modulated Mechanisms of the Cytotoxicity of Thimerosal and Inhibition of DNA Topoisomerase IIr Xing Wu,† Hong Liang,† Kimberley A. O’Hara,† Jack C. Yalowich,‡ and Brian B. Hasinoff*,† Faculty of Pharmacy, UniVersity of Manitoba, 50 Sifton Road, Winnipeg, Manitoba, R3T 2N2, Canada, and Department of Pharmacology, UniVersity of Pittsburgh School of Medicine, Pittsburgh, PennsylVania 15261 ReceiVed September 18, 2007

Thimerosal is an organic mercury compound that is widely used as a preservative in vaccines and other solution formulations. The use of thimerosal has caused concern about its ability to cause neurological abnormalities due to mercury accumulation during a normal schedule of childhood vaccinations. While the chemistry and the biological effects of methylmercury have been well-studied, those of thimerosal have not. Thimerosal reacted rapidly with cysteine, GSH, human serum albumin, and single-stranded DNA to form ethylmercury adducts that were detectable by mass spectrometry. These results indicated that thimerosal would be quickly metabolized in vivo because of its reactions with protein and nonprotein thiols. Thimerosal also potently inhibited the decatenation activity of DNA topoisomerase IIR, likely through reaction with critical free cysteine thiol groups. Thimerosal, however, did not act as a topoisomerase II poison and the lack of cross-resistance with a K562 cell line with a decreased level of topoisomerase IIR (K/VP.5 cells) suggested that inhibition of topoisomerase IIR was not a significant mechanism for the inhibition of cell growth. Depletion of intracellular GSH with buthionine sulfoximine treatment greatly increased the K562 cell growth inhibitory effects of thimerosal, which showed that intracellular glutathione had a major role in protecting cells from thimerosal. Pretreatment of thimerosal with glutathione did not, however, change its K562 cell growth inhibitory effects, a result consistent with the rapid exchange of the ethylmercury adduct among various thiol-containing cellular reactants. Thimerosal-induced single and double strand breaks in K562 cells were consistent with a rapid induction of apoptosis. In conclusion, these studies have elucidated some of the chemistry and biological activities of the interaction of thimerosal with topoisomerase IIR and protein and nonprotein thiols and with DNA. Introduction Thimerosal (Figure 1) is an organic mercury compound with bactericidal and fungicidal properties that is widely used as a preservative in multiuse vials of vaccines, ophthalmic, otic, nasal, and topical products (1–3). There has been a public perception that thimerosal use in vaccines is unsafe after suggestions that it caused a predisposition to autism in children (1, 4). However, recent epidemiological studies have not supported this hypothesis (4). On the basis of the risk assessment assumption that the dose–effect and dose–response relationships of ethylmercury, the presumed metabolite of thimerosal, and methylmercury were the same, thimerosal was removed from most pediatric vaccines in the United States in 2001 (1, 3). Prior to 2001, by 18 months of age, a child in the United States undergoing a routine schedule of immunizations would have received a cumulative dose of 200 µg of mercury (3). The fact that the cumulative exposure to mercury from thimerosal in infants undergoing immunization during the first 6 months of life could exceed U.S. Environmental Protection Agency guidelines provided impetus for the removal of thimerosal from pediatric vaccines (3). Much of what is known about chronic low-dose human methymercury toxicity causing neurologic abnormalities comes from poisoning episodes and environmental exposure (1, 3). Far * To whom correspondence should be addressed. Tel: 204-474-8325. Fax: 204-474-7617. E-mail: [email protected]. † University of Manitoba. ‡ University of Pittsburgh School of Medicine.

less is known about the effects of thimerosal or its presumed metabolite, ethylmercury (1, 3, 5, 6). The initial distribution of ethylmercury in neonatal mice is similar to that of methylmercury, but they differ sharply in their tissue deposition and their metabolism to Hg2+ (4). This suggests that the data on methylmercury may not be suitable for risk assessment for thimerosal (1, 5). Methylmercury reacts rapidly with and has a very high affinity for protein and nonprotein thiols (1, 78), and ethylmercury is likely similar in this regard. Thus, to elucidate some of the basic chemistry and biochemistry of thimerosal, the reactions of thimerosal with nonprotein and protein thiols and the cellular effects of thimerosal have been studied. While the reaction of thimerosal with thiols has been assumed to be an exchange reaction to yield an ethylmercury-thiol adduct (Figure 1), this does not seem to have been shown. In this study, we showed by MS that thimerosal undergoes an exchange reaction with cysteine, GSH, and human serum albumin (HSA)1 (Figure 1) and forms an ethylmercury adduct with single-stranded DNA. 1 Abbreviations: Annexin V-FITC, annexin V-fluorescein isothiocyanate conjugate; 6-mer DNA, DNA with the sequence 5′-CACGTG-3′; 20-mer DNA, self-complementary hairpin DNA with the sequence 5′-TATGATATTTTTATATCATA-3′; BSO, buthionine sulfoximine; DTT, dithiothreitol; ESI-MS, electrospray ionization mass spectrometry; FCS, fetal calf serum; HBSS, Hank’s balanced salt solution; HSA, human serum albumin; IC50, 50% inhibitory concentration; kDNA, kinetoplast DNA; thimerosalDNA, thimerosal-treated and washed kDNA; MTS, 3-(4,5-dimethylthiazol2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium; OTC, (-)-2-oxo-4-thiazolidinecarboxylic acid.

10.1021/tx700341n CCC: $40.75  2008 American Chemical Society Published on Web 01/16/2008

484 Chem. Res. Toxicol., Vol. 21, No. 2, 2008

Figure 1. Structure of thimerosal and its exchange reaction with thiols to form an ethylmercury-thiol adduct and thiosalicylic acid.

Topoisomerase II is a nuclear enzyme that is critical for cell division and is mostly highly expressed in cells undergoing division. It allows chromosome separation by its ability to effect temporary double strand breaks in DNA through which DNA is passed. Studies by us and others have shown that topoisomerase II is highly sensitive to thiol-reactive agents (9–14). Given that a course of thimerosal-containing vaccines is typically given to young children, we hypothesized that thimerosalinduced poisoning of topoisomerase II might be a mechanism by which it exerts its cytotoxicity on dividing cells. In this study, we showed that thimerosal potently inhibited the decatenation activity of DNA topoisomerase IIR but did not act as a topoisomerase IIR poison. The importance of GSH in modulating the activity of thimerosal was shown in experiments in which K562 cells were depleted of GSH with buthionine sulfoximine (BSO). We also show that induction of apoptosis by thimerosal was accompanied by both single and double strand DNA breaks.

Materials and Methods Materials. The oligodeoxynucleotides 5′-CACGTG-3′ (6-mer DNA) and self-complementary 20-mer hairpin 5′-TATGATATTTTTATATCATA-3′ (20-mer DNA) were from Integrated DNA Technologies (Coralville, IA). Kinetoplast DNA (kDNA) was obtained from TopoGEN (Columbus, OH), and pBR322 plasmid DNA was obtained from MBI Fermentas (Burlington, ON, Canada). HSA (catalog number A1887, essentially fatty acid free), thimerosal (Figure 1), and all other chemicals, unless otherwise indicated, were from Sigma-Aldrich (Oakville, ON, Canada). The cell growth inhibition curves were fit to a four-parameter logistic equation using (SigmaPlot, SyStat, Point Richmond, CA) as described (15). The errors quoted for the IC50 values are standard errors from the nonlinear least-squares analyses. In the DNA damage assay, a twotailed t test (SigmaStat, SyStat) was used to test for significance between thimerosal-treated cells and untreated cells. The level of significance used was p < 0.05. Electrospray Ionization Mass Spectrometry (ESI-MS) Studies on Thimerosal Adducts of HSA and DNA Oligodeoxynucleotides. The 20-mer DNA was annealed by heating to 75 °C for 20 min and cooling slowly to room temperature. The MS thimerosal binding studies were carried out using an Applied Biosystems API 2000 Triple Quadrupole mass spectrometer (Thornhill, Canada) equipped with a syringe pump (Harvard Apparatus, Holliston, MA) at a flow rate of 5-10 µL/min. The Analyst software (version 1.4) from Applied Biosystems was used for system control and data acquisition. MagTran (16) (version 1.02, http://www.geocities.com/ SiliconValley/Hills/2679/magtran.html) was used for charge state deconvolution of HSA and its adducts. The ESI source was operated in the positive or negative ion mode as indicated for the reactions with thimerosal with an electrospray voltage of +4.5 (cysteine), -4.5 (GSH), +4.4 (HSA), or -3.0 (DNA) kV, without capillary heating. The reaction of thimerosal and cysteine was carried out by mixing thimerosal (500 µL, 100 µM) and cysteine (500 µL, 100 µM) in methanol (50% v/v) for 5 min at room temperature. Acetic acid (100 µL, 6% v/v) was then added. Scanning was 100–1000 m/z units every 2 s with a step size of 0.10 amu. The reaction of thimerosal and GSH was carried out as with cysteine except that no acetic acid was added. The reaction of thimerosal and HSA was carried out by mixing thimerosal (25 µL, 100 µM) with HSA (25 µL, 100 µM) for 5 min at room temperature. This mixture was

Wu et al. diluted with water (87 µL) and methanol (38 µL) and formic acid (8.5 µL, 6% v/v). Scanning was 1000–1800 m/z units every 4 s with a step size of 0.10 amu. The reaction of thimerosal and DNA was carried out by mixing thimerosal (20 µL, 200 µM) and the DNA (20 µL, 200 µM) at room temperature for 30 min. Acetonitrile (40 µL) was then added. Scanning was 100–1800 m/z units every 6.5 s with a step size of 0.10 amu. Topoisomerase IIr kDNA Decatenation Inhibition Assay. The ability of thimerosal to inhibit topoisomerase IIR was determined via a spectrofluorometric decatenation assay as we have described (13). Topoisomerase IIR is able to decatenate the highly knotted circular kDNA resulting in smaller circles of DNA in an ATPdependent reaction. Each 20 µL reaction contained 0.5 mM ATP, 50 mM Tris (pH 8.0), 120 mM KCl, 10 mM MgCl2, 30 µg/mL bovine serum albumin, 40 ng of kDNA, thimerosal (0.5 µL in water), and 15 ng of topoisomerase IIR protein (the amount that gave approximately 80% decatenation). Full length human topoisomerase IIR was obtained as we previously described (9). The reactions were incubated at 37 °C for 20 min, after which time the reactions were terminated by the addition of 12 µL of 250 mM disodium EDTA. Samples were centrifuged at 8000g for 15 min at 25 °C, and 20 µL of supernatant was mixed with 180 µL of 600-fold diluted PicoGreen dye (Molecular Probes, Eugene, OR) in a 96 well plate. Fluorescence, which was proportional to the amount of kDNA present, was measured in a fluorescence plate reader using an excitation wavelength of 485 nm and an emission wavelength of 520 nm. For some experiments, the effect of thimerosal on the decatenation activity of topoisomerase IIR was also followed using a gel assay as we previously described (9) with the reaction conditions as described above. The decatenated kDNA was separated by electrophoresis (2 h at 8 V/cm) on a Tris-acetateEDTA agarose gel (1.0%, w/v) to which ethidium bromide (0.5 µg/mL) had been added. The kDNA in the gel was imaged by its fluorescence on an Alpha Innotech Fluorochem 8900 (San Leandro, CA) imaging system equipped with a 365 nm UV illuminator and a charged-coupled device camera. pBR322 DNA Relaxation and Cleavage Assays. Topoisomerase II-cleaved DNA complexes produced by anticancer drugs may be trapped by rapidly denaturing the complexed enzyme with SDS (14, 17). The cleavage of double-stranded closed circular pBR322 DNA to form linear DNA was followed by separating the SDStreated reaction products using ethidium bromide gel electrophoresis as described (13, 17). The 20 µL cleavage assay reaction mixture contained either 150 ng of topoisomerase IIR protein as indicated, 50 ng of pBR322 plasmid DNA (MBI Fermentas, Burlington, Canada), 0.5 mM ATP in assay buffer [10 mM Tris-HCl, 50 mM KCl, 50 mM NaCl, 0.1 mM EDTA, 5 mM MgCl2, and 2.5% (v/v) glycerol, pH 8.0], and thimerosal or etoposide as indicated. The order of addition was assay buffer, DNA, drug, and then topoisomerase IIR. To determine whether the order of addition affected the results, experiments were also carried out as described (14) in which the whole reaction mixture except for the drugs were assembled on ice. The drugs were then added, and the temperature was brought up to 37 °C for 10 min to carry out the reaction. In either case, the reaction mixture was quenched with 1% (v/v) SDS/ 25 mM disodium EDTA. The reaction mixture was treated with 0.25 mg/mL proteinase K at 55 °C for 30 min to digest the protein. The linear pBR322 DNA cleaved by topoisomerase IIR was separated by electrophoresis (2 h at 8 V/cm) on a Tris-acetateEDTA buffer (Tris base (4 mM)/glacial acetic acid (0.11% v/v)/ disodium EDTA (2 mM) buffer) ethidium bromide (0.5 µg/mL) agarose gel (1.2% w/v). The DNA in the gel was imaged by its fluorescence on a Alpha Innotech Fluorochem 8900 imaging system. Cell Culture and Growth Inhibition Assays. Human leukemia K562 cells, obtained from the American Type Culture Collection, and K/VP.5 cells (a 26-fold etoposide-resistant K562-derived cell line with decreased levels of topoisomerase IIR protein and mRNA) (18) were maintained as suspension cultures in Dulbecco’s modified Eagle medium (Invitrogen, Burlington, Canada) containing 4 mM L-glutamine and supplemented with 20 mM HEPES, 10% fetal calf

Thimerosal-Protein and DNA Interactions serum (FCS) (Invitrogen), 100 units/mL penicillin G, and 100 µg/ mL streptomycin in an atmosphere of 5% CO2 and 95% air at 37 °C (pH 7.4). For the measurement of growth inhibition, K562 and K/VP.5 cells in exponential growth were harvested and seeded at 6000 cells/ well in 96 well plates (100 µL/well). Twenty-four hours later, cells were treated with vehicle or various concentrations of thimerosal and allowed to grow an additional 72 h. For assessment of the effects of lowering GSH levels on thimerosal cytotoxicity, K562 cells in exponential growth were seeded to yield when thimerosaltreated approximately 12000 cells/well in 96 well plates (100 µL/ well). Forty-eight or twenty-four hours after seeding, cells were treated in the presence or absence of either (-)-2-oxo-4-thiazolidinecarboxylic acid (OTC) (5 mM) or BSO (100 µM), respectively, and were then allowed to grow an additional 24 or 48 h, respectively. The cells were then treated with vehicle or various concentrations of thimerosal and allowed to grow an additional 72 h. Neither the OTC nor the BSO treatment caused any measurable cytotoxicity. After treatment, cells were assayed with the 3-(4,5dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) CellTiter 96 Aqueous One Solution Cell Proliferation assay (Promega, Madison, WI). The spectrophotometric 96 well plate cell growth inhibition assay measured the ability of the cells to enzymatically reduce MTS. Three replicates were measured at each drug concentration, and the IC50 values for growth inhibition were measured by fitting the absorbance-drug concentration data to a four-parameter logistic equation as described (15). In certain experiments, the 10 mM thimerosal stock solution was preincubated with 20 mM GSH or dithiothreitol (DTT) in Hank’s balanced salt solution (HBSS) for 30 min at 37 °C before addition to cells to determine the effect on growth inhibition of pretreating thimerosal with thiols. Fast Micromethod DNA Single Strand Break Assay. The Fast Micromethod DNA Single Strand Break assay, which is based upon the Fluorometric Analysis of DNA Unwinding assay (FADU assay), in which damaged DNA with alkaline-labile sites or single strand breaks undergoes an increased rate of unwinding under alkaline conditions, was performed as previously described (12, 19). Exponentially growing cells (35000) were incubated at 37 °C for 2 h in complete medium in the absence or presence of thimerosal. The cells were centrifuged at 1000g for 6 min, and the supernatant was discarded. Cells were resuspended in 125 µL of Tris-EDTA buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA), and 25 µL of suspension of each treatment (280 cells/µL) were placed into four wells of a 96 well plate. A set of blanks containing 25 µL of TrisEDTA buffer were also analyzed in the same manner. Lysing buffer [25 µL of 9 M urea, 0.2 M EDTA, and 0.1% SDS (pH 10.0)] supplemented with PicoGreen (20 µL of original stock dye/mL of lysing solution) was gently added to each well. Lysing occurred in the dark at room temperature for 1 h. To initiate DNA unwinding, 250 µL of working 0.1 M NaOH solution was added to give a final pH of 12.4. Measurements were immediately started in the fluorescence plate reader using an excitation wavelength of 485 nm and an emission wavelength of 520 nm at 25 °C and continued for 40 min. All values were corrected with blank readings on wells that contained only buffer. γH2AX Assay for DNA Double Strand Breaks in Thimerosal-Treated K562 Cells. The γH2AX assay was carried out as we previously described (13). K562 cells in growth medium (0.5 mL/ well in a 24 well plate, 1 × 106 cells/mL) were incubated with 20 µM thimerosal for either 2 or 20 h. The cells were washed with PBS and lysed with cell lysis buffer containing 30 mM Tris-HCl buffer (pH 7.2), 0.5% Triton-X100, and 1/100 protease inhibitor cocktail (1 mM phenylmethanesulfonyl fluoride, 1 mM sodium orthovanadate, 40 mM β-glycerophosphate, 30 mM NaF, and 5 mM EDTA). The protein concentration was determined with a Bradford assay. After adding sample buffer (60 mM Tris-HCl, pH 6.8, 10% v/v glycerol, 2% w/v SDS, 5% v/v 2-mercaptoethanol, and 0.0025% w/v bromophenol blue), the samples were sonicated for 10 s. Cell lysates (15 µg protein) were subjected to SDSpolyacrylamide gel electrophoresis on a 14% w/v gel. Separated

Chem. Res. Toxicol., Vol. 21, No. 2, 2008 485 proteins were transferred to nitrocellulose membranes, blocked with 5% w/v skimmed milk, and incubated overnight with rabbit antiγH2AX primary antibody diluted 1:2000 (Upstate, Charlottesville, VA). This was followed by incubation for 1 h with peroxidaseconjugated goat anti-rabbit secondary antibody (Sigma) diluted 1:5000. After incubation with ChemiGlow chemiluminescent substrate (Alpha Innotech), chemiluminescence of the γH2AX band was imaged on the Alpha Innotech Fluorochem 8900 imaging system. The β-actin bands on the blot were likewise imaged after reprobing with mouse monoclonal anti-β-actin antibody (Sigma, 1:5000) and peroxidase-conjugated goat anti-mouse secondary antibody (Pierce, Rockford IL, 1:5000). Two-Color Flow Cytometry. The fraction of apoptotic and necrotic cells induced by treatment of K562 cells with thimerosal was quantified by two-color flow cytometry by simultaneously measuring integrated green (annexin V-fluorescein isothiocyanate conjugate, Annexin V-FITC) fluorescence and integrated red (propidium iodide) fluorescence. Binding of annexin V-FITC to phosphatidylserine present on the outer cell membrane was determined using an ApoAlert Annexin V-FITC apoptosis kit (Clontech, Palo Alto, CA). Briefly, K562 cells in suspension were untreated or treated with 5 µM thimerosal at 37 °C for 24 h. The cells were collected by centrifugation at 700g for 5 min, and the pooled cells were washed with the manufacturer-supplied binding buffer. Approximately 5 × 105 cells were resuspended in 50 µL of manufacturer-supplied binding buffer and mixed with 1 µL of Annexin V-FITC at final concentration of 0.4 µg/mL and 1 µL of propidium iodide at a final concentration of 1 µg/mL. After 30 min of incubation in the dark, the cells were analyzed using flow cytometry.

Results ESI-MS Characterization of the Reactions of Thimerosal with Cysteine and GSH. While the reaction of thimerosal with thiol compounds is assumed to be an exchange reaction of the type shown in Figure 1, this reaction does not appear to have been characterized (2). Given the high levels of GSH in the cell and its possible protective role in protecting topoisomerase II and other thiol-sensitive enzymes, the reaction of cysteine and GSH with thimerosal was studied by characterizing the products produced by ESI-MS. The reaction of cysteine with thimerosal produced a cysteine-ethylmercury adduct at a m/z of 351.9 (calculated for C5H12HgNO2S+, 352.0 m/z). The reaction of glutathione and thimerosal produced a GSHethylmercury adduct at a m/z of 536.2 (calculated for C12H20HgN3O6S-, 536.1 m/z). In the negative ion mode, another product with a m/z of 153.1, which corresponded to the thiosalicylate anion (calculated for C7H5O2S-, 153.0 m/z), was also observed. Thus, these studies confirmed that thimerosal reacted with cysteine or GSH to produce a thiol-ethylmercury adduct and thiosalicylate as shown for the reaction in Figure 1. Mercury has seven naturally occurring high-abundance isotopes with a mass range of 8 amu. The most abundant is 202Hg (201.97 amu, 29.7 isotope %), and there are five others in the range of 6.8 to 23.1 isotope %. Because of the abundance of these naturally occurring isotopes, the mass spectrum of a mercury compound yields a characteristic mass spectral fingerprint. The cysteine- and GSH-ethylmercury adducts all had this characteristic fingerprint (data not shown). MS Characterization of the Reaction of Thimerosal with HSA. HSA, which is a major component in blood plasma (46 g/L), contains a single free cysteine-34 residue. The reaction of thimerosal with HSA was studied both as a potential thiolcontaining protein trap for thimerosal in vivo and as a model for its interaction with topoisomerase IIR. The positive ion ESIMS spectra of HSA and HSA after being treated with thimerosal

486 Chem. Res. Toxicol., Vol. 21, No. 2, 2008

Wu et al.

Figure 3. ESI-MS spectra showing that thimerosal forms a 1:1 ethylmercury adduct with 6-mer single-stranded DNA. (A) ESI negative-ion mass spectrum of 100 µM thimerosal and 100 µM DNA in 50% water/acetonitrile (v/v) obtained 30 min after preparation. (B) ESI negative-ion mass spectrum of thimerosal monoanion (calcd 383.0 m/z) in the 375-390 m/z range showing the mass distribution due to multiple mercury isotopes. (C) ESI negative-ion mass spectrum of the DNAthimerosal 4– anion (calcd 504.8 m/z) in the 502-506 m/z range showing a similar mass distribution due to multiple mercury isotopes. (D) ESI negative-ion mass spectrum of the DNA-thimerosal 3– anion (calcd 672.4 m/z) in the 670-674 m/z range showing a similar mass distribution due to the multiple mercury isotopes. The peak at 152.7 m/z is the mercaptobenzoate (calcd 153.0 m/z) reaction product produced from the reaction of thimerosal with DNA. The peak at 339.0 m/z is from decarboxylated thimerosal anion (calcd 339.0 m/z). Peaks at 446.6 m/z and 596.4 m/z are unreacted 4– and 3– anions of DNA, respectively (calcd 447.1 and 596.4 m/z, respectively). Figure 2. ESI-MS spectra showing that thimerosal reacts with HSA giving a 1:1 ethylmercury adduct. (A) ESI positive-ion mass spectrum of 14 µM HSA. The numbers above the peaks identify the multiply charged positive ions of HSA used in the deconvolution of the spectra. (B) Deconvoluted molecular ion region obtained from the spectrum in A identifying unmodified HSA and cysteinylated HSA. (C) ESI positive-ion mass spectrum of 14 µM HSA as in A above but 15 min after the addition of 14 µM thimerosal. (D) Deconvoluted molecular ion region obtained from the spectrum in C identifying cysteinylated HSA and the ethylmercury adduct of HSA. The spectra were obtained in 16% methanol and 84% water containing 0.3% (v/v) formic acid.

are shown in Figure 2A,C, respectively. The deconvoluted mass spectrum of HSA and HSA after being treated with thimerosal that were obtained from these spectra are shown in Figure 2B,D, respectively. HSA is known to be heterogeneous due to several post-translational modifications (20, 21). The best MS-characterized modification is due to cysteinylation at cysteine-34 of HSA to produce a disulfide (20). Other possible modifications include NO adducts, glycation, and N- and C-terminal amino acid deletions (21). Our deconvoluted MS value of 66430 Da for unmodified HSA (Figure 2B) compared very well to the theoretical protonated molecular ion value of 66439 Da from the amino acid sequence (SwissProt P02768) with 17 disulfide bridges (20). The peak at 66542 Da in Figure 2B was most likely due to cysteinylation at cysteine-34 (observed, ∆m +112 amu; calculated for cysteinylation, ∆m +119 amu) as has been previously shown (20). After treatment of HSA with thimerosal, a new peak in the deconvoluted spectrum in Figure 2D appeared at 66651 Da, which was most likely due to formation of an HSA-ethylmercury adduct (observed, ∆m +221 amu; calculated for ethylmercury adduct, ∆m +230 amu). This is in very good agreement, considering the number of high-abundance mercury isotopes that had the effect of broadening the deconvoluted spectrum. Judging by the large decrease in the peak at 66430 Da all, or nearly all, of the unmodified HSA appears to have reacted with thimerosal. The peak at 66558 Da in Figure 2D was likely cysteinylated HSA, and its presence after thimerosal treatment indicates that over the time that this experiment was carried out thimerosal did not measurably displace the cysteine from cysteinylated HSA.

MS Characterization of the Reaction of Thimerosal with DNA. The interactions of methylmercury and ethylmercury with DNA are well-known and have recently been reviewed (22, 23). However, the reaction of thimerosal with DNA does not appear to have been studied. The reaction of thimerosal with DNA was also studied because DNA is a substrate for topoisomerase II (24, 25) and other DNA-processing enzymes. Thus, binding of thimerosal to DNA or induction of DNA damage are potential mechanisms by which thimerosal might exert its cytotoxicity. To determine if thimerosal could react with DNA, the reaction of 6-mer single-stranded DNA and selfcomplementary double-stranded 20-mer DNA was studied by MS. As shown in Figure 3A, the addition of thimerosal to 6-mer DNA (calculated mass, 1792 Da) produced several new peaks that were highly characteristic of an ethylmercury-6-mer DNA adduct (Figure 3C,D). As we previously showed with several anthrapyrazoles (26), several DNA-drug ionization states are typically observed upon binding to DNA. The peaks at 504.2 m/z and 672.4 m/z (Figure 3C,D, respectively) corresponded to an ethylmercury-6-mer DNA adduct with 4– and 3– ionization states (calculated, 504.8 m/z and 673.4 m/z, respectively). Both of these broadened peak envelopes were consistent with the presence of natural abundance mercury isotopes in the adduct. These can be compared to the broadened peak envelope of unreacted thimerosal shown in Figure 3B (1– ionization state). Unreacted 6-mer DNA with 4– and 3– ionization states was found at 446.6 m/z and 596.4 m/z, respectively (calculated, 447.1 and 596.4 m/z, respectively). Under the conditions of this study, the 6-mer DNA would only be present as a single-stranded species. The presence of a 152.7 m/z peak in the spectrum of Figure 3A, which corresponded to the thiosalicylate anion (calculated, 153.0 m/z) product produced upon formation of the ethylmercury adduct also supports the conclusion that thimerosal reacted with 6-mer DNA. To determine if thimerosal could react with double-stranded DNA, a self-complementary 20-mer hairpin DNA (calculated mass, 6095 Da) was studied. This hairpin DNA has a melting temperature of 45 °C (27), which is well above the experimental conditions of this study of 20 °C, and thus, it would be base

Thimerosal-Protein and DNA Interactions

paired. No high-charge state peaks that corresponded to a ethylmercury adduct could be positively identified after a 30 min or an overnight thimerosal (50 µM) treatment of the 20mer DNA (50 µM). As compared to the experiments performed with 6-mer DNA, these experiments were more problematic in that this larger 20-mer DNA showed peaks corresponding to 7– to 13– charge states with one and two sodium ions adducted (and sodium plus acetonitrile adducted) in its mass spectrum (data not shown). This served to complicate and broaden the peak envelopes in the spectra and obscure identification of any potential 20-mer DNA ethylmercury adducts. However, it should be noted that these results do not rule out the possibility that thimerosal reacted with the 20-mer DNA as the adduct may have been produced in such low abundance as to have been undetectable. Effects of Thimerosal on Decatenation of Topoisomerase IIr in Vitro and on Growth in Intact K562 Cell Lines That Vary in Their Levels of Topoisomerase IIr. We and others have shown that the activity of topoisomerase IIR is highly sensitive to thiol-reactive drugs such as cisplatin (9) and quinones (10, 11, 13, 14). Given the high affinity of thimerosal (2) and methylmercury for thiols (1, 22, 28), we decided to investigate whether thimerosal may, in part, be exerting its cytotoxicity by inhibiting topoisomerase IIR through reaction with its free cysteine groups. Given the critical role of topoisomerase IIR in cell division (24, 25), this enzyme is a possible target for thimerosal. We showed in an earlier MS proteomics study that the topoisomerase IIR monomer contains at least five free cysteines (amino acids 170, 216, 300, 392, and 405) (9). The results in Figure 4B show that thimerosal strongly inhibited the decatenation activity of purified topoisomerase IIR with an IC50 of 9.0 µM. To determine whether thimerosal acted as a topoisomerase II poison, the growth inhibitory effects of thimerosal on K562 and K/VP.5 cells were compared. The K/VP.5 cell line contains one-fifth the amount of topoisomerase IIR as compared to the parent K562 cell line (18) and can be a convenient test of whether a compound is a topoisomerase II poison (9). Fewer DNA strand breaks will be produced in cells containing less topoisomerase II, and thus, topoisomerase II poisons will be less potent toward the K/VP.5 cell line. As shown in Figure 4A, the IC50 for growth inhibition of K562 cells was 0.71 µM, while that of K/VP.5 cells was 1.4 µM. Thus, the lack of cross-resistance indicates that, while thimerosal can inhibit the catalytic activity of topoisomerase IIR, it likely did not act as a topoisomerase II poison in a cellular context. GSH Protects Topoisomerase IIr Decatenation Activity from Inhibition by Thimerosal. To determine whether GSH could protect topoisomerase IIR from inhibition by thimerosal, experiments were carried out in which various concentrations of GSH were added to topoisomerase IIR before thimerosal treatment. Topoisomerase II alters DNA topology by catalyzing the passing of an intact DNA double helix through a transient double strand break made in a second helix and is critical for relieving torsional stress that occurs during replication and transcription and for daughter strand separation during mitosis (24). As the results in lanes 10-12 of Figure 4C show, levels of GSH (5 mM) present intracellularly completely protected topoisomerase IIR decatenation activity from 5 to 20 µM thimerosal-induced inhibition. A lower concentration of GSH (0.1 mM) (lanes 7-9) also completely protected topoisomerase IIR from 5 and 20 µM thimerosal. These results suggest that the ethylmercury-GSH adduct did not significantly exchange with the free cysteines on topoisomerase IIR to inhibit it. The

Chem. Res. Toxicol., Vol. 21, No. 2, 2008 487

Figure 4. Effects of the interaction of thimerosal with topoisomerase IIR and growth inhibitory effects on a K562 cell line with reduced levels of topoisomerase II. (A) Inhibition of growth of K562 (O) and K/VP.5 (b) cells by thimerosal. Cells were treated with thimerosal for 72 h prior to the assessment of growth inhibition by an MTS assay. The curved lines are nonlinear least-squares fits to logistic equations and yield IC50 values of 0.71 ( 0.05 and 1.4 ( 0.2 µM, respectively, for K562 and K/VP.5 cells yielding 2.0-fold relative resistance. A replicate experiment on a different day gave a similar result with a relative resistance of 2.2-fold. (B) Inhibitory effects of thimerosal on the catalytic decatenation activity of topoisomerase IIR. The fluorescence measures the amount of decatenated kDNA minicircles in the supernatant of the centrifuged 20 µL assay mixture. The solid line is a nonlinear least-squares fit to a four-parameter logistic equation and yields an IC50 of 9.0 ( 2.6 µM. The reaction mixture contained 0.12 µM DTT from the topoisomerase IIR preparation. A replicate experiment on a different day gave a similar result. (C) This fluorescent image of the ethidium bromide-stained gel shows the effect of the pretreatment of thimerosal with GSH on the decatenation activity of topoisomerase IIR. Lane 1, reaction mixture in the absence of topoisomerase IIR. Lanes 2-4, thimerosal (5-20 µM) completely inhibited the decatenation of kDNA. Lanes 5 and 6, GSH (0.1 and 5 mM) had no effect on decatenation activity. Lanes 7-9, GSH (0.1 mM) completely protected topoisomerase IIR from the inhibitory effects of 5, 10, and 20 µM thimerosal. Lanes 10-12, GSH (5 mM) completely protected topoisomerase II from the inhibitory effects of 5, 10, and 20 µM thimerosal. Lane 13, untreated topoisomerase IIR control showing complete decatenation of kDNA. Where indicated, topoisomerase IIR (12 ng) was treated with thimerosal in the presence or absence of GSH for 10 min at 37 °C. ORI, loading well origin; NOC, nicked open circular kDNA; CC, closed circular decatenated kDNA; and Topo IIR is topoisomerase IIR. The reaction mixture contained 0.12 µM DTT from the topoisomerase IIR preparation. A replicate experiment on a different day gave a similar result. (D) Thimerosal inhibited the topoisomerase IIR-mediated relaxation of supercoiled pBR322 DNA but did not act as a topoisomerase IIR poison to produce linear DNA. This fluorescent image of the ethidium bromide-stained gel shows that topoisomerase IIR relaxed supercoiled pBR322 plasmid DNA (SC) to relaxed DNA (RLX) (lane 2). Lanes 3-5 show that the addition of 1, 10, or 100 µM thimerosal progressively inhibited the relaxation reaction. As shown in lane 6, etoposide treatment produced linear DNA (LIN). Lanes 7-9 show that the addition of 1, 10, or 100 µM thimerosal progressively inhibited the etoposide-mediated formation of linear DNA. A small amount of nicked circular (NC) is normally present in the pBR322 DNA. The reaction mixture (20 µL) contained 150 ng of topoisomerase IIR and 50 ng of pBR322 DNA, and it also contained 1.5 µM DTT from the topoisomerase IIR preparation. The result was typical of a two other replicate experiments.

488 Chem. Res. Toxicol., Vol. 21, No. 2, 2008

fact that GSH was protective is consistent with thimerosal inhibiting topoisomerase IIR by binding to free cysteines. Inhibition of Topoisomerase IIr Activity by Thimerosal Was Not Accompanied by Stabilization of the Covalent Topoisomerase IIr-DNA Cleavable Complex. Several widely used anticancer agents, including etoposide, are thought to be cytotoxic by virtue of their ability to stabilize a covalent topoisomerase II-DNA intermediate (the cleavable complex) (24). Thus, DNA cleavage assay experiments were carried out using etoposide as a positive control to see whether the test compounds stabilized the cleavable complex. As shown in Figure 4D, the addition of 100 µM etoposide to the reaction mixture containing topoisomerase IIR and supercoiled pBR322 DNA induced formation of linear pBR322 DNA (lane 6). Linear DNA was identified by comparison with linear pBR322 DNA produced by action of the restriction enzyme HindIII acting on a single site on pBR322 DNA (data not shown). However, as shown (Figure 4D), the addition of up to 100 µM thimerosal to the reaction mixture induced little or no detectable formation of cleaved linear pBR322 DNA. Using a previously published experimental protocol, all of the reactants except the drugs were assembled on ice (14). The drugs were added, and the reaction mixture was brought up to 37 °C for 10 min. Identical results were obtained using a slightly different protocol (see the Materials and Methods) in which topoisomerase IIR was added last (results not shown). The results in Figure 4D (lanes 3-5) also show that the addition of increasing concentrations of thimerosal to the reaction mixture progressively inhibited the DNA relaxation activity of topoisomerase IIR, as indicated by the progressive loss of relaxed pBR322 DNA relative to supercoiled pBR322. This result was consistent with thimerosalinduced inhibition of decatenation where kDNA was the substrate (Figure 4B,C). Experiments were also carried out to see if thimerosal could inhibit the topoisomerase IIR-mediated etoposide-induced formation of linear DNA. As shown in lanes 7-9, the addition of 1, 10, or 100 µM thimerosal progressively inhibited the etoposide-induced formation of linear DNA. A comparison of the results of Figure 4C,D show that the relaxation activity (Figure 4D) of topoisomerase IIR was not as strongly inhibited by thimerosal as the decatenation activity (Figure 4C). The reason for this was likely due to the higher concentration of DTT in the assay mixture (0.12 µM for decatenation vs 1.5 µM for cleavage/strand passage assays). Because DTT is a dithiol, it would more potently protect topoisomerase IIR than GSH because it can complex with two ethylmercury ions (29). Reducing Intracellular GSH Levels Increases Thimerosal Cytotoxicity. Because the MS studies indicated that thimerosal reacted with GSH, we investigated whether depletion or supplementation of intracellular GSH levels affected the cytotoxicity of thimerosal. GSH levels were increased (by OTC) or decreased (by BSO). OTC is a nonthiol precursor of cysteine that is converted by intracellular 5-oxoprolinase to cysteine, a precursor for the biosynthesis of GSH (30). GSH levels were decreased in K562 cells by a 48 h preincubation with 100 µM BSO, an inhibitor of the rate-limiting GSH-synthesizing enzyme γ-glutamylcysteine synthetase (30). Using a fluorescent Thioglo-1 thiol assay, cellular levels of GSH levels in K562 cells were either increased by 87% or reduced by 71% following OTC or BSO treatment, respectively (12). Because the Thioglo-1 assay measures thiols and is not specific for GSH, the OTCinduced increase in thiol levels likely reflects an increase in both cysteine and GSH levels. As shown in Figure 5A, these concentrations of BSO or OTC did not cause any measurable

Wu et al.

Figure 5. (A) Reduction of intracellular GSH levels in K562 cells increases the growth inhibitory effects of thimerosal. K562 cells were preincubated either without (O) or with 5 mM OTC (1), or with 100 µM BSO (b) for 24 or 48 h, respectively, and then treated for 72 h with thimerosal. The data shown are an average of three replicates. Three other treatments of cells performed on different days yielded similar results. The curved lines are nonlinear least-squares fits to fourparameter logistic equations and yield IC50 values of 2.0 ( 0.3, 0.12 ( 0.02, and 2.6 ( 0.4 µM for K562 cells not treated with BSO, treated with BSO, or treated with OTC, respectively. Error bars are standard errors. (B) Effect of pretreating thimerosal with either GSH or DTT before treating K562 cells. The curved lines are nonlinear least-squares fits to four-parameter logistic equations and yield IC50 values of 1.8 ( 0.2, 1.5 ( 0.1, and 10.7 ( 3.8 µM for K562 cells in which 10 mM thimerosal was not pretreated (O) or was pretreated with either 20 mM GSH (b) or DTT (1), respectively, for 30 min at 37 °C in HBSS. Error bars are standard errors.

cytotoxicity. The BSO or OTC pretreatments were followed by a 72 h exposure to various concentrations of thimerosal, after which cell growth was measured with the MTS assay. As shown in Figure 5A, thimerosal inhibited cell growth with an IC50 value of 2.0 for untreated cells. For the OTC- and BSO-treated cells, this value was either increased 1.3-fold or reduced 16.7-fold, respectively. A t test of three IC50 determinations showed that the small OTC-dependent increase in IC50 was not significant (p ) 0.22). However, the BSO-dependent decrease in IC50 was highly significant (p < 0.0001). These results demonstrate that GSH partially protected these cells from the growth inhibitory effects of thimerosal. Effect of Pretreating Thimerosal with Either GSH or DTT on Its Cytotoxicity Toward K562 Cells. As shown in Figure 5B, when 10 mM thimerosal was pretreated with 20 mM GSH for 30 min at 37 °C in HBSS before it was added to the K562 cell suspension, the IC50 was virtually unchanged (1.8

Thimerosal-Protein and DNA Interactions

Chem. Res. Toxicol., Vol. 21, No. 2, 2008 489

Figure 6. (A) Thimerosal treatment produces DNA single strand breaks in K562 cells. The amount of DNA damage as a result of a 2 h exposure to thimerosal was determined using the Fast Micromethod DNA Single Strand Break assay. Unwinding of DNA in K562 cells was measured from the fluorescence of double-stranded DNA remaining after exposure to various concentrations of thimerosal. Error bars were from four replicate measurements. At thimerosal concentrations of 50 µM and higher, there was a significant difference between untreated cells and thimerosal-treated cells (*p < 0.05, and **p ) 0.01). (B) Thimerosal induced double strand DNA breaks in K562 cells as indicated by the formation of γH2AX. K562 cells were treated with the concentrations of the drugs indicated for the times indicated in growth medium, lysed, and subjected to SDS-PAGE electrophoresis and Western blotting. The blots were probed with antibodies to γH2AX and β-actin (as a loading control) and a chemiluminescentproducing secondary antibody.

µM as compared to 1.5 µM for GSH treatment). However, pretreatment of thimerosal with DTT increased the IC50 6-fold to 10.7 µM. Thus, even though the ethylmercury would be extremely strongly bound to glutathione because of the pretreatment, it was still able to exert its cytotoxicity. Methylmercury is known to very rapidly exchange (