Mechanisms of Aurothiomalate− Cys2His2 Zinc Finger Interactions

DNA binding assays were used to analyze functional interactions between AuTM and two model Cys2His2 zinc finger transcription factors, TFIIIA and Sp1 ...
0 downloads 0 Views 333KB Size
Chem. Res. Toxicol. 2005, 18, 1943-1954

1943

Mechanisms of Aurothiomalate-Cys2His2 Zinc Finger Interactions Jason L. Larabee, James R. Hocker, and Jay S. Hanas* Department of Biochemistry & Molecular Biology, University of Oklahoma Health Science Center, Oklahoma City, Oklahoma 73104 Received June 2, 2005

Zinc finger motifs are present in a wide variety of regulatory proteins and generally function as interaction modules between macromolecules. These functional interactions are controlled by mechanisms of zinc (Zn2+)-binding and release. Besides Zn2+ certain electrophilic metals can potentially react with zinc finger domains and lead to changes in the structure and function of those domains. In these studies, the Cys2His2 zinc finger was chosen as a model for understanding how the gold (I) (Au1+) drug, aurothiomalate (AuTM), interacts mechanistically with the Zn2+ coordination sphere. DNA binding assays were used to analyze functional interactions between AuTM and two model Cys2His2 zinc finger transcription factors, TFIIIA and Sp1; inhibition in the micromolar range of AuTM was observed in both cases. Electrospray ionization mass spectrometry (ESI-MS) was utilized to examine molecular interactions between AuTM and a zinc finger peptide modeled after the third finger of Sp1 (Sp1-3). These experiments demonstrated Au1+ ions can bind the zinc finger structure and trigger the release of the Zn2+ ion. Quantifying the ESI-MS data allowed for a relative affinity value between Zn2+ and Au1+ ions to be calculated and shows Au1+ has a 4-fold higher affinity for Sp1-3 than Zn2+. Mechanistic differences between Zn2+ and Au1+ binding to the model Sp1-3 zinc finger were analyzed at isotopic resolution, and the metal-coordination spheres were probed with small molecules (H+, hydrogen peroxide, glutathione disulfide, and iodoacetamide). Natural isotope cluster analysis suggested the presence of a metal-thiol bond in the Cys2His2 zinc finger structure. Metal exchange reactions between zinc fingers demonstrated Zn2+ ions exchanged more rapidly than Au1+ ions. Circular dichroism (CD) exhibited differences in the secondary structure of the Sp1-3 model peptide when binding Zn2+ or Au1+ ions.

Introduction Chemical compounds containing gold (I) (Au1+) ions (sodium aurothiomalate, aurothioglucose, auranofin.) are thiol-reactive agents and are used for the treatment of rheumatoid arthritis and other inflammatory disorders (1). The mode of action of these chemicals is not known, although reactions with cysteine (Cys) residues in proteins are possibilities. These gold (I) agents alter gene expression, and a current hypothesis for their antiinflammatory activities involves interactions with Cysdependent transcription factors (2-4). The Au1+ ion is proposed to be the active component of these compounds (1). This ion has chemical properties similar to zinc (Zn2+) ions (e.g., filled d-orbital), and a potential mechanism of action is through interactions with Cys-rich Zn2+ coordination spheres. In animal studies mimicking gold (I) therapy, the Au1+ ion was shown to bind metallothionein, suggesting Au1+ions can interact with Cys-rich Zn2+ coordination spheres in vivo (5). Metal binding reactions with Cys-rich domains are important processes for Zn2+dependent DNA binding proteins. The Cys residues of Zn2+ coordination spheres are labile, and changes in the cellular redox state and/or exposure to electrophilic agents (e.g., xenobiotic metals) can alter DNA binding activities of zinc finger transcription factors and overall * To whom correspondence should be addressed at 800 Research Parkway, Room 448, Oklahoma City, OK 73104. Tel, 405-271-2995; fax, 405-271-5440; e-mail, [email protected].

Zn2+ homeostasis. For instance, studies with cadmium (Cd2+) ions have demonstrated DNA repair processes are disrupted by these ions, potentially through interactions with zinc fingers of DNA repair proteins (6). Elucidating how Au1+ ions interact with Cys residues of Zn2+ coordination spheres will shed light on the structure and function of metal binding domains. In addition, such studies will lead to an understanding of the mechanism of action of this drug class and possibly to the design of more specific Cys-reacting compounds. In these current studies, DNA binding assays were used to functionally evaluate interactions of the gold (I) drug, aurothiomalate (AuTM),1 with two model Cys2His2 zinc finger transcription factors, TFIIIA and Sp1. TFIIIA is composed of nine Cys2His2 zinc finger domains and is a positive regulator of 5S ribosomal RNA synthesis in eukaryotic cells (7). Sp1 is an RNA polymerase II specific transcription factor possessing three Cys2His2 zinc fingers in the DNA binding domain of the C-terminal region; this protein plays a prominent role in stimulating cell growth and invasion (8, 9). To elucidate the molecular interactions between AuTM and zinc fingers, electrospray 1 Abbreviations: AuTM, aurothiomalate; CD, circular dichroism; DBD, DNA binding domain; DTT, dithiothreitol; ESI-MS, electrospray ionization mass spectrometry; H2O2, hydrogen peroxide; IA, iodoacetamide; Sp1, specificity protein 1 transcription factor; Sp1-3, the third zinc finger motif from the amino terminal end of the Sp1 transcription factor; TOF, time-of-flight; TFIIIA, transcription factor IIIA.

10.1021/tx0501435 CCC: $30.25 © 2005 American Chemical Society Published on Web 11/10/2005

1944

Chem. Res. Toxicol., Vol. 18, No. 12, 2005

ionization mass spectrometry (ESI-MS) was employed to analyze binding mechanisms between AuTM and the Zn2+ coordination sphere of a Cys2His2 zinc finger peptide (Sp1-3), which is modeled after the third zinc finger of Sp1 (10). These ESI-MS experiments demonstrate that Au1+ ions can dissociate from the thiomalate moiety and bind the zinc finger structure replacing the Zn2+ ion. Quantifying the ESI-MS data allows for a relative affinity value between Zn2+ and Au1+ ions to be calculated for the Sp1-3 motif. To further elucidate differences in the mechanism of Zn2+ and Au1+ binding, the Sp1-3 zinc finger was probed with other small molecules (H+, H2O2, glutathione disulfide, and iodoacetamide) that can react with Zn2+ coordination spheres. Also, rates of metal exchanges between two zinc finger domains were compared, and Au1+ ions were found to exchange more slowly than Zn2+ ions. Circular dichroism (CD) was utilized to elucidate differences in structure and conformation between Zn2+- and Au1+-bound Sp1-3 peptides. These data indicate Au1+ ions can specifically interact with Zn2+ coordination spheres and have different binding mechanisms than Zn2+.

Experimental Procedures Isolation of TFIIIA and Recombinant Sp1 DNA Binding Domain (DBD). Immature ovarian tissue obtained from 4 to 5 cm female Xenopus laevis (Nasco, Fort Atkinson, WI) was homogenized in buffer A (50 mM Tris-HCl, pH 7.6, 50 mM KCl, 5 mM MgCl2, 0.5 mM DTT, and 0.2 mM phenylmethylsulfonylfluoride). The homogenate was centrifuged for 20 min at 10 000g, and aliquots were layered onto 15-30% v/v glycerol gradients in the buffer A. Gradients were centrifuged for 24 h at 34 000 rpm in a Beckman SW41 rotor; all manipulations were performed at 0-4 °C. 7S particles were further purified to 90% homogeneity by DEAE ion-exchange chromatography as described previously (11). 5S RNA was removed from TFIIIA by digestion of the 7S particle (20 µg/mL) with RNase A (10 µg/ mL) in buffer B (20 mM Tris-HCl, pH 7.6, 320 mM KCl, 2 mM MgCl2, 0.2 mM β-mercaptoethanol, and 0.1% v/v NP-40 detergent) for 30 min at room temperature and then placed on ice. The human Sp1 cDNA (kind gift from Dr. Robert Tjian) was cloned into the pUC18 plasmid. The C-terminal DNA binding domain of Sp1 (amino acids 535-696) was cloned into the pProEX HTa (Invitrogen, Carlsbad, CA) vector between BamHI and HindIII restriction sites. The expression vector was transformed into Escherichia coli DH5R competent cells (Invitrogen, Carlsbad, CA) and expressed as a His-tagged protein. The transformed cells were grown in LB and induced with 0.6 mM isopropyl-D-thiogalactoside (ITPG) when the A590 reached 0.5. The induced cells grew for 3 h at 37 °C and were harvested by centrifugation (10 min at 10 000g). The harvested cells were suspended in the following lysis buffer: 5 mM imidazole, 20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 6 M urea, and a protease inhibitor cocktail (Roche Applied Sciences, Indianapolis, IN). Cells were lysed by two passes through a French Pressure cell at 14 000 psi, and the resulting extracts were cleared by centrifugation (10 min at 10 000g). The extract was passed over a column equilibrated with lysis buffer and composed of NiNTA (nickel-nitrilotriacetic acid) resin (Qiagen, Valencia, CA). The His-tagged protein (Sp1-DBD) bound to the column was eluted in the following buffer: 1 M imidazole, 20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 6 M urea, and a protease inhibitor cocktail (Roche Applied Sciences, Indianapolis, IN). After this purification step, Sp1-DBD was >90% pure as determined by SDS-PAGE and Coomassie blue staining. The protein was dialyzed into buffers containing progressively less urea, and the final buffer contained 20 mM MES-NaOH, pH 6.5, 100 mM KCl, 1 mM DTT, 50 µM ZnCl2, and 5% glycerol. Transcription Factor-DNA Binding Reactions. TFIIIAdependent DNase I protection was analyzed on a 303 bp DNA

Larabee et al. insert containing the 120 bp Xenopus borealis somatic 5S ribosomal RNA gene, which was 32P end-labeled on the coding strand as described previously (11). The specific activity of this DNA insert was determined by UV absorption at 260 nm and Cerenkov counting. To study the effects of AuTM and Zn2+ on TFIIIA function, TFIIIA in buffer B was diluted 5-fold in buffer C (20 mM Tris-HCl, pH. 7.6, 70 mM NH4Cl, 7 mM MgCL2, 100 µM DTT, and 0.1% v/v nonionic detergent NP-40) and incubated at room temperature with sodium AuTM or ZnCl2 under the conditions indicated in the figure legends. TFIIIA was then diluted (12-fold) to 10 nM in the same buffer minus sodium AuTM or ZnCl2, end-labeled 5S gene was added to a final concentration of 1 nM (about 104 cpm), and the binding reaction (20 µL) proceeded for 15 min. DNase I was added to the binding reactions to a final concentration of 1-2 µg/mL and incubated for an additional minute at room temperature. The digestion was terminated by addition of 100 µL of stop buffer (20 mM Tris-HCl, pH 7.6, 1 mM EDTA, 0.1% SDS, and 30 µg of sonicated salmon sperm DNA/mL). The DNA was ethanol-precipitated and resuspended in 4 µL formamide solution (20 mM Tris-HCl, pH 7.6, 95% deionized formamide, 1 mM EDTA, 0.01% xylene cyanol and bromophenol blue), heated at 95 °C for 5 min, and then electrophoresed through a 7 M urea-7% w/v polyacrylamide gel until the xylene cyanol marker migrated two-thirds down the gel. The gel was then transferred to blotting paper, dried, and subjected to autoradiography overnight at -70 °C exposed to Kodak XAR-5 film. For gel shift assays the Sp1-DBD was diluted to a concentration of 100 nM into DNA binding buffer (20 mM Tris-HCl, pH 7.6, 90 mM KCl, 1 mM MgCl2, 0.1 mM DTT, 0.05% v/v NP-40, and 8% glycerol), and incubated at room temperature for 30 min with sodium AuTM at the concentrations indicated in the figure legend. The DNA binding reaction proceeded for 15 min after the reaction was initiated with the addition of 1 nM 32P end-labeled Sp1 consensus DNA (5′ATTCGATCGGGCGGGGCGAGC 3′ Promega). The samples were then electrophoresed for 15 min at 300 V on a preelectrophoresed 6% w/v polyacrylamide gel in 0.5× TBE (25 mM Tris base, 25 mM boric acid, and 0.5 mM EDTA). The gel was transferred to blotting paper and subjected to autoradiography overnight at -70 °C exposed to Kodak XAR-5 film. Synthesis and Handling of Sp1-3 Model Peptides. A peptide related to zinc finger three of Sp1 (Sp1-3) was commercially produced with Fmoc-solid-phase peptide synthesis technology (Genemed Synthesis, Inc., San Francisco, CA). The peptide has a monoisotopic mass of 3365.7 Da, which was confirmed by ESI-MS. The amino acid sequence of the peptide is as follows: KKFACPECPKRFMSDHLSKHIKTHQNKK. A molecular mass variant of this peptide (Sp1-3A) was generated to use for the metal exchange and titration experiments. The amino acid sequence of this peptide is KKFACPACPKRFMSDHLSKHIKTHQNK. The peptides were HPLC-purified, lyophilized, and stored in 1 mg aliquots at -20 °C. For experimental analysis, an aliquot of Sp1-3 was suspended in 1-3 mM DTT in deionized water to a final peptide concentration of 1 mg/mL. This solution was then realiquoted and stored at -20 °C. Electrospray Ionization Mass Spectrometry (ESI-MS). Binding events with the Sp1-3 model peptide were analyzed by ESI time-of-flight (TOF) mass spectrometry using an Applied Biosystems Mariner System (Foster City, CA). Typically, the diluted peptide was reacted with zinc acetate and/or sodium AuTM for 10 min at 25 °C and applied to the mass spectrometer utilizing the following buffers in deionized water: 30 µM to 1.5 mM DTT, 10-20 µM peptide, 5 mM ammonium acetate, pH 6.8, and 5% methanol. The pH of the buffer was lowered with acetic acid. Experiments examining oxidation of the metal-bound Sp1-3 peptide were performed by first binding metal ions to the peptide, followed by 30-60 min treatment with H2O2 or glutathione disulfide. Iodoacetamide (IA) modifications were similarly performed by binding the peptide to metal and then incubating the metal-bound peptide overnight with IA (100500 µM). The reactions were infused directly into the electro-

Aurothiomalate-Cys2His2 Zinc Finger Interactions spray ionization source at a flow rate of 1 µL/min and analyzed at isotopic resolution in the positive ion mode. Spectra were an average of 10 s per scan acquired over 5 min. Mass spectra data were examined using the software package, Data Explorer Version 4.0.0.1, supplied by Applied Biosystems. This software package was used to produce theoretical isotope clusters from the formula weight of potential Sp1-3 complexes. The instrument settings for these ESI-MS studies were as follows: spray tip potential, 1788.87; nozzle potential,100.10; skimmer 1 potential, 11.01; quad DC potential, 6.23; deflection voltage, 0.39; einzel lens potential, -27.00; quad RF voltage, 700.2; nozzle temperature, 135.01 °C; push pulse potential, 675-700; pull pulse potential, 249.07; pull bias potential, 9.00; accelerator potential, 3999.94; reflector potential, 1449.97; deflector potential, 2499.9. Calculation of Relative Affinity from Competition Experiments. To determine a relative affinity value between Zn2+ and Au1+, the Zn2+-bound peptide was incubated with increasing concentration of AuTM. Metal binding to the peptide was monitored using ESI-MS. The total signal for each Sp1-3 form was determined by summing the areas under the peaks representing the different species of Sp1-3. The total signal corresponding to both the Zn2+-bound peptides and Au1+-bound peptides was plotted as a function of the ratio of Au1+ ions to Zn2+ ions. These data were fitted to a third-order polynomial, and this function was used to derive a relative affinity (KdZn/ KdAu) value using the following equation:

KdZn/KdAu ) ([Au1+ - peptide][Zn2+])/ ([Zn2+ - peptide][Au1+]) Circular Dichroism (CD) Spectropolarimetry. The CD spectra of the Sp1-3 model peptide were recorded on a JASCO J715 Spectropolarimeter (Jasco, Corp., Tokyo, Japan). The buffer conditions utilized for these CD studies were the same as those used for the ESI-MS experiments, except that the pH was lowered by the addition of HCl. Each spectrum was recorded at a wavelength range of 190-250 nm with 10 accumulations per spectrum in a 0.1-cm cuvette path length at 20 °C.

Results AuTM Inhibits the DNA Binding Activities of the Cys2His2 Zinc Finger Transcription Factors TFIIIA and Sp1. The ability of AuTM to inhibit the specific binding of TFIIIA to the 5S RNA gene was analyzed using the DNase I protection. In this assay, TFIIIA binds the 5S gene and provides protection from DNase I digestion of the 5S RNA gene from nucleotides +43 to +96 relative to the +1 transcription start site (Figure 1A, lane 2). The absence of TFIIIA binding allows DNase I to digest this area as evident by bands indicating 32P DNA fragments in this region (Figure 1A, lane 1). Exposing TFIIIA to 5 µM or higher concentrations of AuTM followed by diluting the AuTM-exposed TFIIIA by 12-fold into the DNA binding reaction resulted in the inhibition of DNA binding as apparent by the loss of DNase I protection (Figure 1A, lanes 5-7). The inhibition of TFIIIA-dependent DNase I protection occurred over the entire binding region, from +43 to +96 on the 5S gene where all nine zinc fingers of TFIIIA interact. The DNA binding activities of the three finger, recombinant DNA binding domain of Sp1 (DBD-Sp1) were evaluated by gel shift analysis in the presence of AuTM. The autoradiogram in Figure 1B demonstrates the ability of the recombinant DBD-Sp1 to bind and retard the mobility of the 32P end-labeled consensus Sp1 DNA binding site (lane 2). Treating Sp1-DBD with 0.2 and 1 µM of AuTM had no affect on DNA binding activities (Figure 1B, lanes

Chem. Res. Toxicol., Vol. 18, No. 12, 2005 1945

Figure 1. Inhibition of DNA binding activities of TFIIIA and Sp1-DBD by AuTM. Production and purification of recombinant Sp1-DBD, isolation of TFIIIA from immature X. laevis ovarian tissue, 32P end-labeling of the X. borealis somatic 5S RNA, DNase I protection and gel shift assays, polyacrylamide gel electrophoresis, and autoradiography were performed as described in Experimental Procedures. (A) DNase I protection autoradiogram of AuTM inhibition of TFIIIA binding to the 5S RNA gene. The nucleotide positions marked on the left margin (+43 and +96) are on the coding strand of the 5S gene and are relative to the +1 transcription start site. TFIIIA-DNA binding and DNase I protection reactions electrophoresed in lanes 1 and 2 were performed with 1 nM end-labeled 5S gene in the absence or presence of 10 nM TFIIIA. TFIIIA at 200 nM was preincubated with 0.1, 1, 5, 10, or 20 µM AuTM (lanes 3-7). These preincubation reactions were diluted 12-fold into DNA binding and DNase I reactions. (B) Gel shift analysis of DNA binding activities of recombinant Sp1-DBD after AuTM treatment. Reactions electrophoresed in lanes 1 and 2 were performed with 1 nM end-labeled Sp1 consensus DNA in the absence or presence of 100 nM Sp1-DBD. Free probe is indicated by the bottom arrow, and Sp1-DBD-bound probe is specified by the top arrow. Lane 3 displays the DNA binding activities of Sp1-DBD when treated with 2.5 mM 1,10-orthophenanthroline. Lane 4 exhibits DNA binding when Sp1-DBD was preincubated with unlabeled probe at a 30-fold excess of the 32P labeled probe. In lanes 5-9, AuTM inhibition of the DNA binding activities of Sp1-DBD was assessed in the presence of 0.2, 1, 5, 10, or 25 µM AuTM, respectively.

5-6). However, a large reduction in the DNA binding activities was observed by increasing AuTM to 5 µM (lane 7); complete inhibition was observed at 25 µM AuTM (lane 9). The DNA binding activities of DBD-Sp1 were inhibited after an exposure to 2.5 mM 1,10-orthophenanthroline, a specific Zn2+ chelator (lane 3). The assay in lane 4 contained unlabeled Sp1-specific oligonucleotides; a reaction that contained 32P labeled nonspecific oligonucleotides gave no band shift (not shown). AuTM is known to interact with thiols, and the addition of excess thiols could reverse the inhibitory effect of AuTM depending upon any preferential specificity for TFIIIA or glutathione. The DNA binding assays were performed in the presence of 100 µM of the dithiol reagent, DTT, and a high degree of AuTM-induced inhibition of DNA binding was still observed. To further demonstrate the specificity of these inhibitory reactions for TFIIIA, DNA binding was assayed with increasing

1946

Chem. Res. Toxicol., Vol. 18, No. 12, 2005

Figure 2. DNase I protection autoradiograms analyzing AuTM inhibition of DNA binding with high levels of glutathione and the protection from AuTM inhibition when TFIIIA binds 5S RNA. Experimental details are described in Experimental Procedures and the legend in Figure 1. (A) TFIIIA was preincubated in lanes 3-5 with 1, 5, or 10 µM AuTM, prior to dilution into DNA binding reactions. In lanes 6-9, DNA binding was assessed when TFIIIA was pretreated with 10 µM AuTM and 0.5, 1, 2, or 3 mM glutathione. (B) 5S RNA was digested from TFIIIA with RNase A (5S RNA designation “-“), and then TFIIIA was pretreated with 0, 5, 10, or 20 µM AuTM (lanes 2-5). In lanes 6-9, DNA binding was evaluated when TFIIIA was pretreated with 0, 5, 10, or 20 µM AuTM while binding 5S RNA; 5S RNA was digested with RNase A in the DNA binding reactions (5S RNA designation “+”).

concentrations of glutathione during the AuTM inhibitory reaction. In Figure 2A, 10 µM AuTM substantially inhibited the DNA binding of TFIIIA at glutathione levels up to 2 mM (Figure 2A, lanes 6-8). The AuTM inhibitory effect on DNA binding was reversed only at a glutathione level of 3 mM (about a 300-fold thiol excess, Figure 2A, lane 9). Even though 75% of the TFIIIA structure is composed of nine zinc finger domains, sites beside zinc fingers could account for the inhibitory effect of AuTM on TFIIIA. TFIIIA zinc fingers bind 5S RNA using a mechanism competitive with 5S gene DNA (12). Previously, selenite ions were shown to inhibit TFIIIA only upon direct exposure to the free protein but not when the protein was bound to 5S RNA in the 7S particle (13). This result is consistent with selenite interacting with zinc finger domains during the inhibition of DNA binding (13). A similar experiment was performed to determine if the AuTM-induced inhibition of TFIIIA was through interactions with zinc finger domains. When bound to 5S RNA, TFIIIA was unable to bind 5S DNA (not shown) unless the 5S RNA was digested with RNase A before the DNA binding reaction (lane 2) or during the DNA binding reaction (lane 6). In Figure 2B, lanes 3-5 depict AuTM inhibiting the DNA binding activities of TFIIIA when 5S RNA was digested prior to AuTM inhibition reactions

Larabee et al.

(Figure 2B). In lanes 7-9, inhibition of the DNA binding activities of TFIIIA is shown when 5S RNA was bound to TFIIIA during the AuTM inhibition reaction and then digested away in the DNA binding reactions. These results demonstrate 5S RNA binding to TFIIIA during the AuTM inhibition reaction (at 5 or 10 µM) provided some protection from AuTM inhibition. This observation indicates zinc finger interactions were critical for the effects of AuTM on DNA binding activities of TFIIIA (Figure 2B). Only at elevated concentrations of AuTM (20 µM) was the inhibitor able to penetrate the 5S RNA protection of the TFIIIA zinc fingers (Figure 2B, lane 9). These experiments also helped discount the possibility that AuTM was directly binding DNA and inhibiting the DNA binding activities of TFIIIA. Au1+ Ions Bind and Replace Zn2+ Ions in a Model Sp1 Zinc Finger Peptide (Sp1-3) As Elucidated by Electrospray Ionization Mass Spectrometry (ESIMS). To help understand molecular interactions of AuTM with the metal coordination sphere of the Cys2His2 zinc finger, a zinc finger domain was modeled after the third finger of Sp1. This 28 amino acid peptide (see Experimental Procedures) was previously utilized to analyze Zn2+ ion binding and redox chemistry of this finger (10). A similar peptide was examined by absorption spectroscopy as well as nuclear magnetic resonance (NMR) (14, 15). The soft ionization process of ESI allows coordination events between metals and peptides to be transferred into the gas phase where accurate mass measurements are made of this complex at isotopic resolution. In Figure 3A, the uncharged mass spectrum was deconvoluted from a series of multiple charged peaks. This mass spectrum exhibits a reaction between the Sp1-3 peptide and AuTM at a molar ratio of 1:1 (10 µM). A high abundance peak was observed indicating a monoisotopic mass of 3561.6 Da, which is equal to the reduced model peptide (3365.7 Da) and the addition of an Au1+ ion minus a proton (196 Da). A very minor peak was also observed, indicating the reduced apo-Sp1-3 peptide (3365.7 Da). This experiment shows the Au1+ ion was released from the thiomalate moiety and bound stoichiometrically to the model zinc finger. As analyzed by ESI-MS, the 1:1 stoichiometry remained constant even when the concentration of Au1+ was at a significant excess of the concentration of the peptide (not shown). To determine if Au1+ ions can displace Zn2+ ions from the Sp1-3 motif, the peptide was treated with excess Zn2+ ions (2 equiv) and then reacted with 1 equiv of AuTM. In Figure 3B, the mass spectrum shows Zn2+ binding the model Sp1-3 peptide relative to the internal standard, neurotensin. The m/z values displayed in the mass spectrum depict the Sp1-3 peptide in the +4 charge state, which is the predominate charge state observed in this buffer system for both metal and nonmetal forms of Sp1-3. Also in this spectrum, neurotensin is displayed in the +2 charge state with an m/z value of 836.9. The Zn2+-bound peptide from Figure 3B was exposed to a concentration of AuTM stoichiometric to the peptide, and the resulting mass spectrum is exhibited in Figure 3C. Au1+ ions were able to replace Zn2+ to a considerable degree in the Sp1-3 motif. The reduction in Zn2+ binding is obvious between panels B and C of Figure 3 when the intensity of the neurotensin peak is compared to the intensity of the Zn2+-Sp1-3 peak. These ESI-MS spectra also demonstrate that only one metal ion bound the Sp1-3 peptide, and no complexes

Aurothiomalate-Cys2His2 Zinc Finger Interactions

Figure 3. ESI-MS analysis of interactions between AuTM and the apo-Sp1-3 model peptide and the displacement Zn2+ from the Sp1-3 peptide by AuTM. The synthesis, handling, and ESIMS analysis of the Sp1-3 model peptide are described in Experimental Procedures. (A) A deconvoluted mass spectrum indicating the products of a reaction between 10 µM of the Sp1-3 model peptide and 10 µM AuTM. The monoisotopic mass of each set of peaks, as well as the peptide complex, is specified in the panel. The reaction contained 10 µM Sp1-3 peptide, 100 µM DTT, 5 mM ammonium acetate, pH 6.8, 5% methanol, and 10 µM AuTM. (B) Reaction between 20 µM of the Sp1-3 model peptide and 40 µM zinc acetate. The reaction contained 20 µM Sp1-3 peptide, 3 µM neurotensin, 180 µM DTT, 5 mM ammonium acetate, pH 6.8, and 5% methanol. (C) AuTM (20 µM) was added to the reaction mixture in panel B. The mass spectra in panels B and C show Sp1-3 in the +4 charge state and neurotensin in the +2 charge state. The monoisotopic m/z values are indicated in panels B and C. The percent of the zinc finger peptide binding metal is specified on each panel, which was determined by calculating the area under the peaks representing the different forms of Sp1-3. * indicates peptide Met oxidation as a 16 Da increase. ** indicates sodium adduct.

were discerned that contained multiple metal ions such as combinations of Zn2+ and Au1+ ions. In the mass spectra from Figure 3, peak areas were summed for the different forms of Sp1-3 as well as for neurotensin. The neurotensin signal response is 30% of the total signal response before and after the addition of AuTM to the Zn2+-bound Sp1-3 peptide (Figure 3B,C). The constant internal standard response confirms that the changes in peak intensities are proportional between the different forms of the Sp1-3 peptide. Therefore, summing the areas of the different forms of Sp1-3 can be used to gain quantitative information. In Figure 3B

Chem. Res. Toxicol., Vol. 18, No. 12, 2005 1947

Figure 4. AuTM titration of a Zn2+-bound Cys2His2 zinc finger peptide and metal binding to this Cys2His2 zinc finger peptide at various acidic pH’s. (A) ESI-MS was used to generate curves fit to data indicating Zn2+ release (O) and Au1+ binding (0). The fraction of the zinc finger peptide binding metal was determined by calculating the area under the peaks representing the different forms of Sp1-3A. These data were fitted to a thirdorder polynomial with an R2 value of 0.99. The relative affinity calculated from this curve is KdZn/KdAu ) 4.2. These titrations were performed at 25 °C in 60 µM DTT, 5 mM ammonium acetate, pH 6.8, and 5% methanol. The concentration of peptide was 20 µM, and the peptide was treated with 40 µM zinc acetate. (B) ESI-MS spectra were obtained for 10 µM of the Sp1-3 peptide reacted with either 20 µM zinc acetate (b) or 20 µM AuTM (9) at various acid pH’s. The fraction of the zinc finger peptide binding metal was determined from peak areas of the different forms of Sp1-3. These reactions were performed at 25 °C in 30 µM DTT, 5 mM ammonium acetate, pH 6.8, and 5% methanol. The pH of this reaction mixture was adjusted with acetic acid.

before the addition of AuTM, 88% of the zinc finger was in the Zn2+-bound form and the remainder was in the apo form. With the addition of AuTM, the resulting spectrum indicates 40% of the zinc finger peptide was in the Zn2+-bound form and 60% was in the Au1+-bound form (Figure 3C). Using data from ESI-MS spectra, a binding curve can be constructed that allows the calculation of the relative affinity between Zn2+ and Au1+ ions for a model Sp1-3 peptide. For other systems, ESI-MS was used to gain quantitative binding information, which was shown to agree with other solution techniques (1621). In Figure 4A, a binding curve is presented from ESIMS data obtained from titrating the Zn2+-saturated Sp13A model peptide with increasing concentrations of AuTM. This Sp1-3A peptide has an Ala substituted for a Glu (see sequence in Experimental Procedures) and has identical metal-binding properties as the Sp1-3 peptide (see Figure 7). From this curve, a relative affinity value

1948

Chem. Res. Toxicol., Vol. 18, No. 12, 2005

Larabee et al.

Figure 5. Mass spectra of reactions of the metal-bound Sp1-3 peptide with glutathione disulfide, H2O2, and iodoacetamide (IA). In these experiments, metal binding was performed by reacting 20 µM of the Sp1-3 peptide with 30 µM zinc acetate and 5 µM AuTM for 10 min followed by the treatment indicated in each panel. The reaction mixture contained 60 µM DTT, 5 mM ammonium acetate, pH 6.8, and 5% methanol. The monoisotopic mass of each set of peaks, as well as the form of the peptide, is indicated on each panel. (A) Metal binding with no other treatment. (B) The metal-bound zinc finger was reacted with 200 µM glutathione disulfide for 30 min. (C) The metal-bound peptide was reacted with 3 mM H2O2 for 30 min. (D) Apo-Sp1-3 was reacted with 100 µM IA overnight. (E) The metal-bound peptide was reacted with 200 µM IA overnight. (F) The metal-bound peptide was reacted with 500 µM IA overnight.

was calculated as KdZn/KdAu ) 4.2 for the Sp1-3A model peptide. This analysis was repeated using 50% less peptide, and the same relative affinity was determined (not shown). As another demonstration of Au1+ ions binding tighter to the Sp1-3 peptide than Zn2+ ions, the metal-binding reactions were performed at various acidic pH’s. Protons can directly compete for the electron density of metalcoordinating amino acids and can be used to estimate metal-binding strengths. In Figure 4B, the graph exhibits the fraction of the Sp1-3 peptide binding Zn2+ or Au1+ ions at various acidic pH’s. The fraction of the zinc finger peptide binding metal was quantified by summing the areas under the peaks representing the different forms of Sp1-3. The ratio of peptide to metal was maintained at a 1:2 for all these binding reactions. At pH 6.8, approximately 90% of the zinc finger peptide was binding Zn2+ or Au1+ ions. When the pH was then decreased to 5.2, the binding of Zn2+ was decreased by about 40%, whereas the binding of Au1+ remained unchanged. At pH 4.3, Zn2+ binding was reduced a total of about 80%, and Au1+ binding was reduced about 50%. At this lower pH, the peptide was likely less ordered possibly favoring Au1+ binding to the two closely spaced Cys residues as opposed to a more ordered structure necessary for Zn2+ coordination between the two Cys residues and the two His residues. Differences in Au1+ and Zn2+ Binding to the Sp1-3 Model Peptide As Analyzed by Reaction with Sulfhydryl-Specific Reagents. In these ESI-MS

experiments, the sulfhydryl agent DTT was included in the buffers to maintain reducing conditions because ESI is an oxidizing process (22-24). Although DTT is known to bind Zn2+ and other metals (24), it does not appear to inordinately affect Zn2+ and Au1+ binding to the Sp1-3 model peptide. For example, raising the DTT concentration from 60 µM to 1.5 mM reduced Zn2+ and Au1+ binding both by only 20% (data not shown). In Figure 5, differences in Zn2+ and Au1+ binding to the Sp1-3 peptide in the presence of sulfhydryl-reacting agents are presented. In these experiments, the Sp1-3 model peptide was reacted with a 6:1 molar ratio of Zn2+/AuTM. When this ratio of Zn2+/AuTM is used, binding levels of both metals can be assessed in the same reaction and spectrum. In Figure 5B, the metal-bound peptide was reacted with 240 µM glutathione disulfide, and the appearance of a peak of 3363.7 Da was observed, which corresponded to a 2 Da decrease in the peptide resulting from the loss of two protons from disulfide formation. This spectrum also shows Zn2+ binding was greatly reduced and Au1+ binding was relatively unaffected. In Figure 5C, the metal-bound peptide was reacted with 3 mM H2O2; this spectrum shows reduction in both Zn2+ and Au1+ binding with Zn2+ binding reduced to a greater degree. Another specific modifier of Cys residues is the alkylating agent, iodoacetamide (IA). IA modification results in a mass increase of 57 Da per Cys residue. Analyzing IA modifications to this zinc finger helped determine if Zn2+ or Au1+ better protects Cys residues of the Zn2+

Aurothiomalate-Cys2His2 Zinc Finger Interactions

coordination sphere and provided further information about how Au1+ was binding the zinc finger structure. In Figure 5D, the apo-Sp1-3 peptide was reacted with 100 µM IA, and the resulting spectrum shows peptide peaks with both one modification (57 Da increase, 1X) and two modifications (114 Da increase, 2X). Increasing the concentration of IA to 200 µM resulted in all of Sp1-3 possessing two modifications (data not shown). When the metal-bound peptide was reacted with 100 µM IA, no IAinduced modifications were observed, suggesting protection of Cys residues by both metal ions at this IA concentration (data not shown). In Figure 5E, the Zn2+or Au1+-bound peptide was treated with 200 µM IA. A reduction in the relative intensity of the apo-peak and the Zn2+ peak was observed, while the relative intensity of the Au1+-bound peak was unaffected. The peaks that merge into the Zn2+-binding peak indicate a single modification by IA. Peaks corresponding to the doublemodified peptide were also observed. In Figure 5F, treating the metal-bound Sp1-3 peptide with 500 µM IA produced a large percentage of doubly modified Sp1-3. With this IA treatment, Zn2+ binding was eliminated, and Au1+ binding was significantly reduced but not abolished. During these IA treatments, metal binding was not observed to a peptide that also contained a single or a double IA modification. Isotope Cluster Analysis of Au1+ and Zn2+ Ion Interactions with the Sp1-3 Model Peptide. The MS figures indicate Au1+ binds one thiolate and one thiol in the Sp1-3 peptide. Previously, isotope cluster analysis of the Zn2+-bound form of the Sp1-3 peptide demonstrated Zn2+ binds not only Cys thiolates but also some amount of Cys thiol depending on reducing conditions (10). In Figure 6, natural isotope clusters of Zn2+ and Au1+ binding to the Sp1-3 peptide were resolved in the same spectrum. The first peak on the left (within each isotope cluster) was the monoisotopic peak and consisted of a combination of the lowest mass isotopes of carbon (12C) and the respective metal ions (64Zn2+; Au1+ has only one natural isotope). The remaining peaks in the isotope cluster were forms of the peptide containing different combinations of 12C and 13C isotopes or different Zn2+ isotopes. Each isotope cluster (black solid line) was compared to the theoretical isotope cluster generated by the MS software (dotted line). For Au1+ binding (Figure 6A), the theoretical isotope cluster (dotted line) was based on the assumption that Au1+ binds to one thiolate and one thiol with a monoisotopic mass of 3561.6 Da. The experimental isotope cluster (black solid line) exactly matched the theoretical isotope cluster, which confirms one proton was released during Au1+ binding and no mixed protonation states were observed. For Zn2+ binding (Figure 6B), two theoretical isotope clusters were calculated. One was based on the assumption Zn2+ binds to two thiolates and has a monoisotopic mass of 3427.7 Da (dotted line). The other isotope cluster was based on Zn2+ binding one thiolate and one thiol with a monoisotopic mass of 3428.7 Da (gray solid line). The experimentally derived isotope cluster (black solid line) resides between the two theoretical isotope clusters. As observed in previous studies, the position of the experimental isotope cluster reveals a small portion of Zn2+ coordination was achieved by a Cys thiol. This experiment also demonstrates this proposed thiolate-metal ion-thiol structure was more readily obtained with Au1+ than Zn2+ under the same reducing conditions.

Chem. Res. Toxicol., Vol. 18, No. 12, 2005 1949

Figure 6. ESI-MS analysis of the isotope cluster for the Zn2+and Au1+-bound Sp1-3 model peptide. (A and B) The Sp1-3 peptide (20 µM) was reacted with 30 µM zinc acetate and 5 µM AuTM for 10 min prior to infusion into the ESI source. In the resulting spectrum, the isotope cluster for Au1+-bound Sp1-3 (panel A) and Zn2+-bound Sp1-3 (panel B) are presented as the black solid lines. In panel A, the dotted line denotes a calculated theoretical spectrum based on the addition of an Au1+ ion and the release of one proton. In panel B, the dotted line represents a calculated theoretical spectrum based on the addition of Zn2+ and the release of two protons. The gray solid line represents a theoretical spectrum based on the addition of Zn2+ and the release of one proton. The structural inset (just of the partial Cys residue) relates the spectrum in each panel to the protonation state of the metal-coordinating Cys residues.

Metal Exchange Reactions of Cys2His2 Zinc Fingers Bound to Zn2+ or Au1+ Ions. The kinetics of metal exchange between zinc fingers likely depends on the chemistry of the metal ion as well as the structure of the metal-bound coordination sphere of the zinc finger. To examine the effects of Zn2+ and Au1+ ions on metal exchange reactions, the transfer of these metals between the Sp1-3 peptide and a second zinc finger peptide was monitored by ESI-MS. The second zinc finger (Sp1-3A) with a slightly smaller molecular mass than Sp1-3 possessed the same sequence as Sp1-3 except the Glu at position 7 was changed to an Ala and the C-terminal Lys was deleted. In these experiments, the zinc finger peptides were infused into the mass spectrometer, and the transfer of metals between the two peptides was examined by monitoring changes over time in the relative intensity of the metal and nonmetal binding zinc finger peaks. If both Sp1-3 and Sp1-3A were treated simultaneously with a Zn2+ concentration that was 50% of the total peptide concentration, the resulting mass spectrum indicates each peptide binds about 50% Zn2+ (data not shown). To analyze Zn2+ exchange between the zinc fingers, the Sp1-3 peptide was incubated first with stoichiometric amounts of Zn2+ for 5 min, followed by the addition of the Sp1-3A peptide. The reaction was then immediately infused into the mass spectrometer, and the resulting peaks demonstrate each peptide was binding

1950

Chem. Res. Toxicol., Vol. 18, No. 12, 2005

Larabee et al.

indicating Au1+ ions were being released from the Sp1-3 peptide and were binding the Sp1-3A peptide (Figure 7). This exchange reaction reached equilibrium at approximately 60 min. If Au1+ was bound to the Sp1-3A peptide first, the same results were obtained (Figure 7). In Figure 7, panel A exhibits metal binding to the second peptide added to the reaction and panel B displays the release of metal from the zinc finger initially loaded with metal.

Figure 7. Metal-exchange reactions of Cys2His2 zinc fingers. Differences in the metal-exchange reactions were monitored between zinc fingers binding Zn2+ or Au1+. These exchange reactions were tracked using ESI-MS to monitor metal binding by examining changes in peak heights as the metal-exchange reaction was continuously infused into the mass spectrometer. These reactions were performed at 25 °C in a reaction mixture containing 60 µM DTT, 5 mM ammonium acetate, pH 6.8, and 5% methanol. In the Zn2+ exchange experiments, 10 µM of the Sp1-3 model peptide was reacted with 10 µM zinc acetate for 5 min followed by the addition of 10 µM Sp1-3A. The order of the experiment was reversed by incubating 10 µM Sp1-3A with 10 µM zinc acetate for 5 min and then adding 10 µM Sp1-3. In the Au1+ experiment, 10 µM of the Sp1-3 model peptide was reacted with 7 µM AuTM for 5 min followed by the addition of 10 µM Sp1-3A. The order was reversed, and 10 µM of the Sp13A model peptide was reacted with 7 µM AuTM for 5 min followed by the addition of 10 µM Sp1-3. These graphs depict the change in the percent metal-bound zinc finger over a time period of 60 min. (A) Binding to the second peptide added to the exchange reaction (∆, Zn2+ binding to Sp1-3A; 0, Au1+ binding to Sp1-3A; 2, Zn2+ binding to in Sp1-3; 9, Au1+ binding to Sp1-3). (B) The release of metal from the first zinc finger added to the reaction (∆, Zn2+ binding to Sp1-3A; 0, Au1+ binding to Sp1-3A; 2, Zn2+ binding to in Sp1-3; 9, Au1+ binding to Sp1-3).

50% Zn2+ within about 2 min (the minimum time needed to mix the peptides, to infuse into the mass spectrometer, and to observe peaks) (Figure 7). If Zn2+ was bound to the Sp1-3A peptide first, the same results were achieved (Figure 7). This result indicates Zn2+ exchange between these zinc fingers was very rapid and was at a rate that is faster than this assay can measure. A similar experiment was repeated analyzing the exchange of Au1+ ions. Zinc fingers may exchange Au1+ ions much slower than Zn2+, which could possibly be measured on the time scale of this assay. For this experiment, the Sp1-3 peptide was incubated first with a concentration of AuTM slightly below stoichiometric amounts. After an incubation of 5 min, the Sp1-3A peptide was added to the reaction, and the reaction was infused into the mass spectrometer. At 2 min post Sp1-3A addition, the Sp1-3 peptide was almost saturated with Au1+ ions, and the Sp1-3A peptide had relatively small amounts of Au1+ ions binding. Over the course of 60 min, the intensity of the peaks changed,

Circular Dichroism (CD) Analysis of Structural Features of the Au1+-Bound Zinc Finger Motif. CD measurements were used to analyze conformational changes of the Sp1-3 peptide under a variety of conditions including Au1+ binding. In Figure 8A, the CD spectrum of the Sp1-3 model peptide in the absence of metal at pH 2.2 shows features indicative of a random coil structure. By raising the pH and adding Zn2+ to the model peptide, we produced a CD spectrum with features indicative of Zn2+ binding such as less negative ellipticity, a red-shifted minimum, and positive ellipticity between the wavelengths of 195 and 190 nm. These features were observed in other CD experiments, and the red shift of the minimum, in particular, is indicative of R-helix formation during Zn2+ binding (25-27). The addition of AuTM to the model peptide produced a CD spectrum showing the peptide folds into an ordered structure with different conformational features than those observed with the Zn2+-bound peptide (Figure 8A). With Au1+ binding, a slight red shift of the minimum was observed but not to the degree of the Zn2+-bound peptide, which indicates the Au1+-bound peptide has less R-helix. Like the Zn2+-bound peptide, the Au1+-bound peptide had a minimum with much less negative ellipticity than the random coil structure. However, the Au1+-bound peptide did not possess positive ellipticity at the low wavelengths like Zn2+ binding. In Figure 8B, the CD spectrum of the Au1+-bound zinc finger was compared to the spectrum of the disulfide form of the peptide; the two spectra looked similar, except the disulfide form of the zinc finger had more negative ellipticity. In Figure 8C, Cd2+ binding to the Sp1-3 model peptide was compared to the Au1+- and Zn2+-bound form of the zinc finger. The CD spectrum of the Cd2+-bound peptide had a minimum that was slightly red-shifted but not to the degree of the Zn2+-bound peptide. The Cd2+ minimum had less negative ellipticity than the random coil but more negative ellipticity than Zn2+or Au1+. These observations suggest the Cd2+-bound peptide had some secondary structure, but the structure was much less ordered than the Zn2+- or Au1+-bound zinc finger. In Figure 8D, the Zn2+-bound form of the peptide was titrated with increasing concentrations of AuTM. In this titration experiment, increasing amounts of AuTM caused the minimum to blue-shift and the ellipticity to become less negative. Also, increasing AuTM caused a decrease in the positive ellipticity between 195 and 190 nm. A significant change in the Zn2+-bound CD spectrum was observed with the addition of 5 µM AuTM, which was 20% of the molar concentration of the zinc finger peptide.

Discussion Gold (I) compounds have important therapeutic roles, and their mechanism of action is hypothesized to be

Aurothiomalate-Cys2His2 Zinc Finger Interactions

Chem. Res. Toxicol., Vol. 18, No. 12, 2005 1951

Figure 8. Comparisons of CD spectra of different metal complexes of the Sp1-3 model zinc finger. The CD spectra were obtained in conditions similar to those used in the ESI-MS experiments and are as follows: 25 µM Sp1-3 peptide, 90 µM DTT, 5 mM ammonium acetate, pH 6.8, and 5% methanol. The CD spectrum of the apo-Sp1-3 peptide was obtained by adding HCl to pH 2.2. For metalbinding CD spectra, Sp1-3 was reacted with either 38 µM zinc acetate, 38 µM AuTM, or 75 µM cadmium acetate for 10 min at 25 °C. The disulfide form of the Sp1-3 peptide was prepared by an incubation of 1 h with 6 mM H2O2. (D) The AuTM titration was performed by first reacting the Sp1-3 peptide with 38 µM zinc acetate for 5 min followed by CD analysis. AuTM was then added to the Zn2+-bound Sp1-3 peptide, incubated for 5 min, and then CD spectra were obtained. The final AuTM concentration of the titration steps were 5, 15, 25, 30, 38, and 48 µM. The arrows in this panel indicate the direction of the shift in the CD spectrum as the concentration of AuTM was increased.

mediated by their direct interaction with protein Cys residues. In the present studies, the DNA binding abilities of the Cys2His2 zinc finger transcription factors, TFIIIA and the Sp1, were inhibited at low micromolar concentrations of AuTM, the first such observation of AuTM affecting this important class of transcription factors (Figure 1). Because serum concentrations of Au1+ ions can reach levels of 30-40 µM during therapeutic conditions, the low micromolar concentrations utilized in these DNA binding assays could easily be achieved in the cell (2). These inhibition kinetics are very rapid (less than a minute), suggesting a direct interaction with the TFIIIA structure (data not shown). Evidence is presented that AuTM is likely directly affecting the Cys2His2 zinc finger structures of TFIIIA and Sp1. This conclusion is drawn from the observation that 5S RNA, when bound to the TFIIIA DNA binding domain, protects the protein from AuTM inhibition (Figure 2) and from the observation that the DNA binding domain of Sp1 is used in the AuTM inhibition assays (Figure 1). Previous studies demonstrated the DNA binding abilities of other Cysdependent transcription factors are inhibited by gold (I) drugs including the Cys2Cys2 progesterone receptor, NFκB, and AP-1, which all have critical Cys residues for function (2-4). These previous studies did not demonstrate a direct interaction between Au1+ and Cys residues in these transcription factors. Other studies analyzed

Au1+ binding to the Cys residues of serum albumin and metallothionein and have begun to analyze mechanisms of Au1+ binding to these proteins (1, 28, 29). Our current studies build upon these previous studies by analyzing a model Cys2His2 domain with ESI-MS and CD. In these studies, a direct interaction between Au1+ ions and the Cys residues of a Sp1-3 zinc finger peptide is demonstrated, and basic mechanistic information is elucidated about zinc finger interactions with Au1+, Zn2+, and redox molecules (described below). In Figure 3, ESI-MS demonstrated that Au1+ interacts with the Sp1-3 peptide with saturation binding at a 1:1 stoichiometry, which is indicative of binding specificity. This figure also demonstrated that Au1+ ions can displace Zn2+ ions from Sp1-3, suggesting a higher affinity for Au1+ binding. In the concentration range applicable to ESI-MS, the affinity of Au1+ ions for the zinc finger structure is too tight to allow for a direct titration of Au1+ ions for the Sp1-3 model peptide. However, a competition experiment between Zn2+ ions and Au1+ ions can yield a relative affinity value. Such a competition was performed (Figure 4A), and the relative affinity between Zn2+ and Au1+ ions was calculated for the Sp1-3A peptide (KdZn/ KdAu ) 4.2). This relative value can be used to derive an approximate affinity of Au1+ ions for the Sp1-3 model peptide using the known KdZn of 6.0 × 10-10 M for the Sp1-3 peptide (14). This calculation results in a KdAu of

1952

Chem. Res. Toxicol., Vol. 18, No. 12, 2005

about 1.4 × 10-10 M for the model Sp1-3 peptide. A relative affinity between two metals can be a more pragmatic approach because the affinity ratio can be compared over a variety of experimental conditions (30). Previous studies have used ESI-MS to quantify noncovalent interactions with macromolecules, and a number of these studies have analyzed interactions between metals and peptides (16-21). Furthermore, affinity constants obtained from ESI-MS data are in good agreement with solution techniques (17-20). Lead (Pb2+) ions are another xenobiotic metal that have a high affinity for protein Cys residues and can specifically alter the DNA binding activities of Cys2His2 transcription factors (31, 32). The relative affinity of Pb2+ for the Cys2His2 consensus zinc finger was previously measured as KdZn/KdPb ) 0.1 (33), which is in contrast to KdZn/ KdAu ) 4.2 measured in these studies. Previous ESI-MS studies analyzing Pb2+ binding to the Sp1-3 peptide demonstrated that Pb2+ does not readily displace Zn2+ in the Sp1-3 peptide (10), which agrees with the trend observed in the Pb2+ studies by Payne et al. (33). Specificity is indicated in this study by the 1:1 stoichiometry observed between Zn2+ or Au1+ ions bound to the Sp1-1 peptide as well as by not binding the neurotensin peptide (Figure 3). The ability of Au1+ and Zn2+ to bind this finger in the presence of 1.5 mM DTT (not shown), which itself has a high affinity for Zn2+ and other soft metals (24), is also an indicator of specificity. In addition, AuTM inhibits the DNA binding activities of TFIIIA in the presence of high thiol concentrations such as 100 µM DTT and 1 mM glutathione (Figure 2). Upon interaction of Au1+ with the Sp1-3 zinc finger, Zn2+ is released in rapid fashion (Figure 3). In general, Zn2+ is released from this peptide in a more rapid fashion than Au1+ (Figure 7), possibly contributing to the overall greater binding affinity of Au1+ to this Cys2His2 zinc finger than Zn2+. If Au1+ interacts with other Cys-rich Zn2+-binding proteins in cells in a similar fashion, Au1+ compounds may disrupt Zn2+ homeostasis, which could contribute to the therapeutic action of Au1+ compounds. Since free Zn2+ levels are very low in a cell, disruption of Zn2+ homeostasis could lead to elevated levels of Zn2+ that could affect many other cellular factors (2, 34-37). One of the characteristics of zinc fingers is the Cysrich nature of their metal-coordinating sites. The experiment analyzing iodoacetamide (IA) modification to the metal-bound peptide demonstrates Au1+ ions protect both the Cys residues of the model zinc finger (Figure 5). Also, the change in mass observed during Au1+ binding suggests the Sp1-3 peptide binds Au1+ through one Cys thiolate (S-) and one Cys thiol (SH). To further analyze this Au1+ coordination structure, natural isotope cluster analysis was utilized with the high resolution and high mass accuracy of the electrospray ionization time-of-flight mass spectrometer. This experiment (Figure 6) revealed an exact correlation between the theoretical and experimental spectra, indicating all the Au1+-bound peptides coordinate Au1+ by one thiolate and one thiol. Depending on redox conditions, Zn2+ can possibly bind this zinc finger with all thiolate coordination or combinations of thiol and thiolate coordination (10). In the experiment in Figure 6, the Zn2+-binding isotope cluster revealed a coordination sphere composed mostly of two thiolates with a minor level of Zn2+ coordination to a thiol and thiolate. Previous quantum mechanical analysis for Zn2+ interacting with a Cys2His(imidazole)2 synthetic coordi-

Larabee et al.

nation sphere indicated the most stable structure is achieved with one thiolate and one thiol binding the metal ion (38). A low dielectric environment, possibly occurring in a hydrophobic environment of a Zn2+-binding site, could favor thiol-metal interactions (39). Other theoretical considerations have also provided evidence for the possibility for Zn2+-thiol interactions (40). Although to our knowledge no solution evidence exists for Zn2+thiol interactions in metal-binding sites, such evidence exists for ferrous (Fe2+)-thiol interactions in heme proteins (41). Fe2+, like Zn2+, is an intermediate ion in the hard-soft classification of metals, whereas Au1+ is a soft ion (42). The ability of Au1+ ions to form thiolate/ thiol coordination spheres could also be a factor contributing to the binding strength of Au1+ for the zinc finger peptide. The Au1+-bound peptide is resistant to reactions with electrophiles such as H2O2, glutathione disulfide, and IA (Figure 5). Because thiolates are more reactive to electrophiles than thiols, the thiol/thiolate coordination sphere of Au1+ could contribute to this greater resistance to electrophiles than the Zn2+ coordination sphere composed of lesser levels of thiol/thiolate binding. Chemical properties of the Au1+ ion could also contribute to binding strength and the resistance of the Au1+-bound peptides to electrophiles. For instance, the Au1+ ion is much larger than the Zn2+ ion, which could allow the Au1+ ion to better shield the Cys residues than the Zn2+ ion (43). Also, the Au1+ ion is more polarizable than the Zn2+ ion, allowing Au1+ ions to have stronger interactions with the highly polarizable sulfur of the Cys residue (42). The ESIMS data in this study as well as in previous studies (10, 44) lend support to the possibility of metal-thiol bonds occurring in Zn2+-binding sites but does not prove their existence. Significant differences in the conformations of the Sp1-3 peptide are observed (as assayed by CD) when bound with either Zn2+ or Au1+ ions (Figure 8). Zn2+ binding induces a pronounced red shift (higher wavelength) of the ellipticity minimum, which is consistent with the stabilization of an R-helix (26). Au1+ binding produces less of a red shift in the minimum, indicating Au1+ does not stabilize an R-helix to the extent observed with Zn2+. Because Au1+ does not appear to stabilize the R-helical structure, the possibility exists that Au1+ ions are not interacting with His residues of this zinc finger structure. Our pH data (Figure 4B) lend some support to this notion, as Au1+ binding is not affected at pH 5.2 (possible pKa range of imidazole groups in proteins, ref 45), whereas Zn2+ binding is significantly reduced at this pH. These CD features of the Au1+-bound peptide are more closely related to the disulfide form of the Sp1-3 peptide than the Zn2+-bound form, which could possibly be a result of Au1+ forming a bidentate chelation complex with two Cys residues. The Cd2+-bound Sp1-3 peptide has a CD spectrum more similar to the Zn2+-bound form of the peptide than the Au1+-bound form. The CD spectrum suggests the Cd2+-bound peptide folds into a conformation that resembles a classic zinc finger, but the Cd2+-bound peptide possesses more disorder than the Zn2+-bound form. A comparison of the CD spectrum of the Zn2+-, Cd2+-, and Au1+-bound forms reveals that slight differences in the chemical properties of metals with filled d-orbitals can have dramatic effects on the conformation of Cys2His2 zinc fingers. We also corroborate our MS experiments using CD to show Au1+ can displace Zn2+ from the finger structure, as evident by the reversal of

Aurothiomalate-Cys2His2 Zinc Finger Interactions

the Zn2+-dependent spectrum into the Au1+-dependent spectrum upon addition of AuTM to the Zn2+-Sp1-3 complex (Figure 8D). These CD experiments provide evidence that Au1+ induces a different conformation in this Cys2His2 zinc finger domain, which is likely responsible for the loss of specific DNA binding by the Cys2His2 transcription factors TFIIIA and Sp1 (Figure 1). Besides disrupting the structures of Cys2His2 zinc fingers, Au1+ binding to Cys-rich Zn2+-binding proteins could generally disrupt Zn2+ homeostasis. Zn2+ homeostasis involves the ability of cellular proteins to acquire and release Zn2+ and is an essential mechanism regulating many cellular processes including inflammation (46, 47). Zn2+ homeostasis in a cell is believed to be controlled kinetically by a pool of labile Zn2+-binding sites (46). These labile sites are Cys-rich and are reactive with many cellular oxidants such as nitric oxide (NO), H2O2, and glutathione disulfide. Reactions with these cellular oxidants are believed to aid in the trafficking of Zn2+ in a cell (46). Potentially, these labile Cys-rich Zn2+-binding sites could become saturated with Au1+ ions leading to the release of Zn2+. Because Au1+ does not readily exchange from Cys-rich binding sites, these Cys-rich Zn2+-binding sites may be unavailable for Zn2+ binding for certain periods of time. With a decrease in exchangeable Zn2+-binding sites, the level of free Zn2+ ions could increase in the cytoplasm. Recent studies have suggested the release of Zn2+ from labile binding sites is involved in inflammation-related signaling pathways (47). This Zn2+ release could have anti-inflammatory effects. For instance, Zn2+ ions can decrease the gene expression of inducible NO synthase (iNOS) (48) and can directly inhibit the activities of NO synthase (49). Therefore, the therapeutic anti-inflammatory activities of gold (I) drugs could involve the ability of Au1+ ions to disrupt the structure of certain regulatory zinc finger domains as well as the ability of Au1+ ions to generally stimulate the release of Zn2+. If Zn2+ ions accumulate at excessive concentrations in a cell, these effects could potentially lead to cellular toxicity and apoptosis (46).

Chem. Res. Toxicol., Vol. 18, No. 12, 2005 1953

(8)

(9)

(10) (11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

Acknowledgment. This work was supported by the Department of Defense Grant Number F49620-01-10452.

References (1) Shaw, C. F., III (1999) Gold-based therapeutic agents. Chem. Rev. 99, 2589-2600. (2) Handel, M. L., DeFazio, A., Watts, C. K., Day, R. O., and Sutherland, R. L. (1991) Inhibition of DNA binding and transcriptional activity of a nuclear receptor transcription factor by aurothiomalate and other metal ions. Mol. Pharmacol. 40, 613618. (3) Handel, M. L., Watts, C. K. W., DeFazio, A., Day, R. O., and Sutherland, R. L. (1995) Inhibition of AP-1 binding and transcription by gold and selenium involving conserved cysteine residues in Jun and Fos. Proc. Natl. Acad. Sci. U.S.A. 92, 4497-4501. (4) Yang, J. P., Merin, J. P., Nakano, T., Kato, T., Kitade, Y., and Okamoto, T. (1995) Inhibition of the DNA-binding activity of NFkappa B by gold compounds in vitro. FEBS Lett. 361, 89-96. (5) Schmitz, G., Minkel, D. T., Gingrich, D., and Shaw, C. F., III (1980) The binding of Gold(I) to metallothionein. J. Inorg. Biochem. 12, 293-306. (6) Jin, Y. H., Clark, A. B., Slebos, R. J. C., Al-Refai, H., Taylor, J. A., Kunkel, T. A., Resnick, M. A., and Gordenin, D. A. (2003) Cadmium is a mutagen that acts by inhibiting mismatch repair. Nat. Genet. 34, 326-329. (7) Shastry, B. S., Honda, B. M., and Roeder, R. G. (1984) Altered levels of a 5S gene-specific transcription factor (TFIIIA) during

(21)

(22)

(23) (24)

(25)

(26)

(27)

(28)

(29)

oogenesis and embryonic development of Xenopus laevis. J. Biol. Chem. 259, 11373-11382. Kadonaga, J. T., Carner, K. R., Masiarz, F. R., and Tjian, R. (1987) Isolation of cDNA encoding transcription factor Sp1 and functional analysis of the DNA binding domain. Cell 51, 1079-1090. Ishibashi, H., Nakagawa, K., Onimaru, M., Castellanous, E. J., Kaneda, Y., Nakashima, Y., Shirasuna, K., and Sueishi, K. (2000) Sp1 decoy transfected to carcinoma cells suppresses the expression of vascular endothelial growth factor, transforming growth factor beta1, and tissue factor and also cell growth and invasion. Cancer Res. 60, 6531-6536. Larabee, J. L., Hocker, J. R., and Hanas, J. S. (2005) Cys redox reactions and metal binding of a Cys2His2 zinc finger. Arch. Biochem. Biophys. 434, 139-149. Hanas, J. S., Hazuda, D. J., Bogenhagen, D. F., Wu, F. Y., and Wu, C.-W. (1983) Xenopus transcription factor A requires zinc for binding to the 5 S RNA gene. J. Biol. Chem. 258, 1412014215. Hanas, J. S., Bogenhagen, D. F., and Wu, C.-W. (1984) Binding of Xenopus transcription factor A to 5S RNA and single-stranded DNA. Nucleic Acids Res. 12, 2745-2758. Larabee, J. L., Hocker, J. R., Hanas, R. J., Kahn, F. M., and Hanas, J. S. (2002) Inhibition of zinc finger protein-DNA interactions by sodium selenite. Biochem. Pharmacol. 64, 17571765. Posewitz, M. C., and Wilcox, W. E. (1995) Properties of the Sp1 zinc finger 3 peptide: coordination chemistry, redox reactions, and metal binding competition with metallothionein. Chem. Res. Toxicol. 8, 1020-1028. Narayan, V. A., Kriwacki, R. W., and Caradonna, J. P. (1997) Structures of zinc finger domains from transcription factor Sp1. Insights into sequence-specific protein-DNA recognition. J. Biol. Chem. 272, 7801-7809. Whittal, R. M., Ball, H. L., Cohen, F. E., Burlingame, A. L., Prusiner, S. B., and Baldwin, M. A. (2000) Copper binding to octarepeat peptides of the prion protein monitored by mass spectrometry. Protein Sci. 9, 332-343. Rosu, F., Gabelica, V., Houssier, C., and De Pauw, E. (2002) Determination of affinity, stoichiometry and sequence selectivity of minor groove binder complexes with double-stranded oligodeoxynucleotides by electrospray ionization mass spectrometry. Nucleic Acids Res. 30, e82. Rosu, F., De Pauw, E., Guittat, L., Alberti, P., Lacroix, L., Mailliet, P., Riou, J. F., and Mergny, J. L. (2003) Selective interaction of ethidium derivatives with quadruplexes: an equilibrium dialysis and electrospray ionization mass spectrometry analysis. Biochemistry 42, 10361-10371. Zhang, S., Van Pelt, C. K., and Wilson, D. B. (2003) Quantitative determination of noncovalent binding interactions using automated nanoelectrospray mass spectrometry. Anal. Chem. 75, 3010-3018. Shields, S. B., and Franklin, S. J. (2004) De novo design of a copper(II)-binding helix-turn-helix chimera: the prion octarepeat motif in a new context. Biochemistry 43, 16086-16091. Palumaa, P., Kangur, L., Voronova, A., and Sillard, R. (2004) Metal-binding mechanism of Cox17, a copper chaperone for cytochrome c oxidase. Biochem. J. 382, 307-314. Johnson, K. A., Shira, B. A., Anderson, J. L., and Amster, I. J. (2001) Chemical and on-line electrochemical reduction of metalloproteins with high-resolution electrospray ionization mass spectrometry detection. Anal. Chem. 73, 803-808. Baldwin, M. A., and Benz, C. C. (2002) Redox control of zinc finger proteins. Methods Enzymol. 353, 54-69. Kre¸ z˘ el, A., Les´niak, W., Jez˘ owska-Bojczuk, M., Młynarz, P., Brasun˜, J., Kozłowski, H., and Bal, W. (2001) Coordination of heavy metals by dithiothreitol, a commonly used thiol group protectant. J. Inorg. Biochem. 84, 77-88. Frankel, A. D., Berg, J. M., and Pabo, C. O. (1987) Metaldependent folding of a single zinc finger from transcription factor IIIA. Proc. Natl. Acad. Sci. U.S.A. 84, 4841-4845. Nomura, A., and Sugiura, Y. (2002) Contribution of individual zinc ligands to metal binding and peptide folding of zinc finger peptides. Inorg. Chem. 41, 3693-3698. Simpson, R. J. Y., Cram, E. D., Czolij, R., Matthews, J. M., Crossley, M., and Mackay, J. P. (2003) CCHX zinc finger derivatives retain the ability to bind Zn(II) and mediate protein-DNA interactions. J. Biol. Chem. 278, 28011-28018. Laib, J. E., Shaw, C. F., III, Petering, D. H., Eidsness, M. K., Elder, R. C., and Garvey, J. S. (1985) Formation and characterization of aurothioneins: Au,Zn,Cd-thionein, Au,Cd-thionein, and (thiomalato-Au)χ-thionein. Biochemistry 24, 1977-1986. Shaw, C. F., III, Schaeffer, N. A., Elder, R. C., Eidsness, M. K., Trooster, J. M., and Calis, G. H. M. (1984) Bovine serum albumin-

1954

(30)

(31)

(32) (33)

(34) (35)

(36)

(37) (38) (39)

Chem. Res. Toxicol., Vol. 18, No. 12, 2005

gold thiomalate complex: gold-197 Moessbauer, EXAFS and XANES, electrophoresis, sulfur-35 radiotracer, and fluorescent probe competition studies. J. Am. Chem. Soc. 106, 3511-3521. Magyar, J. S., and Godwin, H. A. (2003) Spectropotentiometric analysis of metal binding to structural zinc-binding sites accounting quantitatively for pH and metal ion buffering effects. Anal. Biochem. 320, 39-54. Zawia, N. H., Sharan, R., Brydie, M., Oyama, T., and Crumpton, T. (1998) Sp1 as a target site for metal-induced perturbation of transcriptional regulation of developmental brain gene expression. Dev. Brain Res. 107, 291-298. Hanas, J. S., Rodgers, J. S., Bantle, J. A., and Cheng, Y.-G. (1999) Lead inhibition of DNA-binding mechanism of Cys2His2 zinc finger proteins. Mol. Pharmacol. 56, 982-988. Payne, J. C., ter Horst, M. A., and Godwin, H. A. (1999) Lead fingers: Pb2+ binding to structural zinc-binding domains determined directly by monitoring lead-thiolate charge-transfer bands. J. Am. Chem. Soc. 121, 6850-6855. Maret, W., Jacob, C., Vallee, B. L., and Fischer, E. H. (1999) Inhibitory sites in enzymes: zinc removal and reactivation by thionein. Proc. Natl. Acad. Sci. U.S.A. 96, 1936-1940. Maret, W., Yetman, C. A., and Jiang, L.-J. (2001) Enzyme regulation by reversible zinc inhibition: glycerol phosphate dehydrogenase as an example. Chem.-Biol. Interact. 130-132, 891-901. Gazaryan, I. G., Krasnikov, B. F., Ashby, G. A., Thorneley, R. N. F., Kristal, B. S., and Brown, A. M. (2002) Zinc is a potent inhibitor of thiol oxidoreductase activity and stimulates reactive oxygen species production by lipoamide dehydrogenase. J. Biol. Chem. 277, 10064-10072. Haase, H., and Maret, W. (2003) Intracellular zinc fluctuations modulate protein tyrosine phosphatase activity in insulin/insulinlike growth factor-1 signaling. Exp. Cell Res. 291, 289-298. Deerfield, D. W., II, and Pedersen, L. G. (1997) Structure and relative acidity for a zinc finger. THEOCHEM 419, 221-226. Dudev, T., and Lim, C. (2002) Factors governing the protonation state of cysteines in proteins: an Ab initio/CDM study. J. Am. Chem. Soc. 124, 6759-6766.

Larabee et al. (40) Simonson, T., and Calimet, N. (2002) CysxHisy-Zn2+ interactions: thiol vs. thiolate coordination. Proteins 49, 37-48. (41) Perera, R., Sono, M., Sigman, J. A., Pfister, T. D., Lu, Y., and Dawson, J. H. (2003) Bioinorganic chemistry special feature: neutral thiol as a proximal ligand to ferrous heme iron: Implications for heme proteins that lose cysteine thiolate ligation on reduction. Proc. Natl. Acad. Sci. U.S.A. 100, 3641-3646. (42) Lippard, S. J., and Berg, J. M. (1994) Principles of Bioinorganic Chemistry. University Science Books, Mill Valley, CA. (43) Weast, R. C., Selby, S. M., and Hodgman, C. D. (1964) Handbook of Chemistry and Physics 45th Edition. The Chemical Robber Co., Cleveland, OH, p F-89. (44) Fabris, D., Hathout, Y., and Fenselau, C. (1999) Investigation of zinc chelation in zinc-finger arrays by electrospray mass spectrometry. Inorg. Chem. 38, 1322-1325. (45) Wilkins, M. E., Hosie, A. M., and Smart, T. G. (2002) Identification of a beta subunit TM2 residue mediating proton modulation of GABA type A receptors. J. Neurosci. 22, 5328-5333. (46) Maret, W. (2004) Zinc and sulfur: a critical biological partnership. Biochemistry 43, 3301-3309. (47) Spahl, D. U., Berendji-Gru¨n, D., Suschek, C. V., Kolb-Bachofen, V., and Kro¨ncke, K.-D. (2003) Regulation of zinc homeostasis by inducible NO synthase-derived NO: nuclear metallothionein translocation and intranuclear Zn2+ release. Proc. Natl. Acad. Sci. U.S.A. 100, 13952-13957. (48) Yamaoka, J., Kume, T., Akaike, A., and Miyachi, Y. (2000) Suppressive effect of zinc ion on iNOS expression induced by interferon-gamma or tumor necrosis factor-alpha in murine keratinocytes. J. Dermatol. Sci. 23, 27-35. (49) Abou-Mohamed, G., Papapetropoulos, A., Catravas, J. D., and Caldwell, R. W. (1998) Zn2+ inhibits nitric oxide formation in response to lipopolysaccharides: implication in its anti-inflammatory activity. Eur. J. Pharmacol. 341, 265-272.

TX0501435