The Interactions of Zinc, Nickel, and Cadmium with Xenopus Transcription Factor IIIA, Assessed by Equilibrium Dialysis Gregory S. Makowski and F. William Sunderman Jr. Departments of Laboratory Medicine and Pharmacology, University of Connecticut Medical School, Farmington, Connecticut
ABSTRACT Transcription factor IIIA (TFIIIA) was isolated from Xenopus ovary and treated with l,lOphenanthroline to remove zinc. The interactions of apoTFIIIA with Zn*‘, Ni*+, and Cd*’ were studied by equilibrium dialysis under anaerobic conditions (pH 7.0, 25X), using 65ZnC12, 63NiC12, and 109CdC12 as the radioligands. The data for binding of Zn*+, Ni*+, and Cd*+ to apoTFIIIA were best-fitted by a model with two classes of binding sites. For Zn*+, the apparent dissociation constants ( KdlZnand K,,*“’ ) for the high- and low-affinity sites were 1.0 x 10m8 and 2.6 X lo-’ M; the apparent binding capacities of the two classes were 0.8 f 0.5 and 9.6 k 0.3 g-atoms of Zn/mol; the Hill coefficient was 1.18, consistent with positive cooperativity of Zn-binding sites. For Ni*+, the apparent KdINiand K,2Ni values were 2;3 x lo-’ and 5.2 x 10V4 M; the apparent binding capacities were 2.3 +_0.6 and 8.6 * 0.6 g-atoms of Ni/mol; the Hill coefficient was 1.20, consistent with positive cooperativity of Ni-binding sites. For Cd*+, the apparent Kdlcd and Kd2Cd values were 2.8 x lop6 and 1.6 x 10e4 M; the apparent binding capacities were 0.9 + 0.3 and 2.4 f 0.5 g-atoms of Cd/mol; the Hill coefficient was 0.53, consistent with negative cooperativity or heterogeneity of Cd-binding sites. This study has the following significance: First, it hely to resolve a controversy about the zinc content of purified TFIIIA. Second, it shows that the K,, ” of apoTFIIIA is less than the reported KdZ” of thionein, consistent with the hypothesis that thionein modulates gene expression by competing with TFIIIA and other Zn-finger proteins for intracellular Zn*+ stores. Third, it confirms previous indirect evidence that the affinity of apoTFIIIA for Zn*’ is much greater than for Cd*+, and that the affinity for Cd*+ is greater than for Ni*+.
Address reprint request and correspondence Utilities Professor of Toxicology, University Avenue, Famington, CT 06030. Journal ofInorganic Biochemistry, 48,107-119 (1992)
to: F. William Sunderman Jr., M. D., Northeast of Connecticut Medical School, 263 Farmington
107 0 1992Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, NY, NY 10010 0162-0134/92/$5.00
108
G. S. Makowski and F. W Sunder-man Jr.
INTRODUCTION TFIIA is one of several proteins that are required for the transcription of 5S rRNA genes by RNA polymerase III. TFIITA binds to a specific DNA sequence of N 50 base pairs in the intragenic control region of the 5s rRNA gene [l--3]. Pre-vitellogenic oocytes from Xenopus faeuis contain abundant TFIIIA in 7s ribonucleoprotein (RNP) particles, where TFIIIA is stored in stoichiometric association with 5S rRNA [4, 51. TFIIIA consists of a single polypeptide chain of 344 amino acids (38.5 kDa) [6], comprising a C-terminal domain that evidently interacts with other components of the transcriptional machinery (e.g., TFIIIB, TFIIIC, RNA polymerase III) [7], and N-terminal and central domains that contain nine imperfect tandem repeats of a Zn-finger motif of the C2H2 class, which bind to DNA and RNA [3, 6, 8, 91. The structure and functions of Xenopus TFIIIA have been summarized in recent articles IlO-121. The ligand interactions of zinc in the finger motifs of TFIIA have been discussed in several reviews [13-161. The present paper reports the affinities of apoTFIIIA for Zn”, Ni2+, and Cd*+, measured by equilibrium dialysis. EXPERIMENTAL SECTION Experimental
Animals and Chemicals
Xenopus
laevis females (Xenopus I, Inc., Ann Arbor, MI) were housed in polycarbonate aquaria that were filled to 10 cm with dilute NaCl solution (10 mM). The aquaria were kept in a constant temperature room (18 + 1°C) with a 12 hr light/dark cycle. The frogs were fed thrice weekly, once with pelleted frog brittle (Nasco, Inc., Fort Atkinson, WI), and twice with ground beef heart and lung from a local abatoir, supplemented with a vitamin mixture (Polyvisol, Mead Johnson Co., Evansville, IN). 65ZnC1,! (4.0 mCi/mg), ‘“NiC12 (12 mCi/mg), and ‘09CdCl, (151 mCi/mg) were purchased from NEN Division, E.I. DuPont Co. (Wilmington, DE). Ultrapure ZnCl,, NiCl,, and CdCl, were purchased from Johnson Matthey, Inc., (Ward Hill, MA). Media for chromatography were purchased from Pharmacia LKB Biotechnology, Inc. (Piscataway, NJ) or Whatman Co. (Hillsboro, OR). Chemicals (reagent grade) were purchased from Sigma Chemical Co. (St. Louis, MO) or J. T. Baker Chemical Co. (Phillipsburg, PA), unless otherwise specified.
General Procedures Centrifugations and chromatographic fractionations were performed at 4°C. Ultrapure water (resistivity N 18 Ma/cm> was prepared with a “Mega-Pure” system (MP3A, Coming Scientific Co., Coming, NY), using distillation from an all-glass still and demineralization with activated carbon and mixed-bed ion-exchange columns. Stringent precautions were observed to control contamination by trace metals [17]; plasticware was soaked in “Nuclear?’ detergent (National Diagnostics, Inc., Somerville, NJ) and rinsed with ultrapure water; dialysis bags and membranes (Spectrum Medical Industries, Inc., Los Angeles, CA) were heated twice at 70°C for 1 hr in 1 L of ultapure water, rinsed after each heating with 10 L of ultrapure water, and stored at 4°C in a water-filled polyethylene canister 1181; zinc-free buffers were passed through a Chelex-100 (KC-form) resin column, as described by Holmquist [19], and analyzed by atomic absorption
Zn2+, Ni2+, AND Cd 2+ BINDING
TO TFIIIA
109
spectrometry to verify that the zinc concentration was < 0.3 pg/L [20]. Purification of 7s RNP particles and TFIIIA was performed by a modification of the procedure of Han et al. [21]. At each step, centrifugal and chromatographic fractions were analyzed by SDS-polyacrylamide gel electrophoresis [22, 231; the fractions that contained TFIIA were identified by probing Western blots with monoclonal anti-TFIIIA antibody or @Zn2+, as previously described [23,241. Preparation of Ovary CytOsol A total of 46 g (wet wt) of ovary tissue was obtained from 77 juvenile females with body weights (mean f SD) of 17.7 f 4.3 g. The frogs were anesthetized by immersion in Tricaine (20 mM, ethyl m-aminobenzoate methanesulfonate). The ovaries were excised, rinsed, blotted, and weighed (mean f SD = 0.60 f 0.11 g/frog); 12 batches of N 4 g of pooled ovary tissue were homogenized with a motor-driven, Teflon-glass tissue homogenizer in 12 ml of BufferA (HEPES, 50 mM; KCl, 25 mM; MgCl,, 5 mM; ZnSO,, 10 PM; dithiothreitol (DTT), 1 mM; phenylmethylsulfonylfluoride (PMSF), 0.5 mM; pH 7.5). The homogenates were centrifuged at 4°C for 30 min at 16,500 g. To remove debris and lipid droplets, the supematant cytosol (12 samples of N 10 ml) was filtered through small columns (1.5 cm diameter, 1.5 cm height) of Sephacryl S-300, equilibrated with Buffer A.
Preparation of 7s RNP Particles The 12 cytosol samples were fractionated individually on gel filtration columns of Sephacryl S-300 (2.5 cm diameter, 90 cm length) by elution in Buff2r A at a flow-rate of 0.8 ml/min; B-ml fractions were collected. The 7s RNP particles, which eluted in tubes No. 30 to No. 33, were polled in four batches and fractionated on anion exchange columns of DEAE-cellulose (2.5 cm diameter, 10 cm length. Whatman DE-52), equilibrated in Buffer B (HEPES, 20 mM; MgCI,, 1.5 mM; ZnSO,, 10 PM; DTT, 1 mM; pH 7.5). The flow rate was 1 ml/min; lo-ml fractions were collected. The columns were eluted with 200 ml of Buffer B, followed by a linear KC1 gradient (0 to 0.6 M) in 600 ml of Buffer B. The 7s RNP particles were eluted in tubes No. 44 to No. 47, which contained _ 0.25 M KCl. Isolated of TFIIIA After addition of 0.6 mg of bovine pancreatic ribonuclease (protease-free, 70 units/mg, Calbiochem Corp., San Diego, CA) to the four 40-ml samples of purified 7s RNP particles, the samples were incubated for 1 hr at 25°C. Ribonuclease digestion was stopped by adding an equal volume of urea (10 Ml in Buffer C (HEPES, 50 mM; MgCl,, 5 mM; ZnSO,, 10 PM; DTT, 1 mM; glycerol, 200 ml/L; pH 7.5). The four samples were fractionated individually on columns of CM-Sephadex (1.25 cm diameter, 3 cm length), equilibrated with Buffer C containing KC1 (0.1 M); the flow-rate was 1 ml/min; lo-ml fractions were collected. Stepwise elution was performed using (a> 90 ml of Buffer C with 0.1 M KCl, (b) 60 ml of Buffer C with 0.25 M KCI, (c) 30 ml of Buffer C with 0.4 M KCl, and (d) 80 ml of Buffer C with 1 M KCl. During step cd), TFIIIA was eluted in tubes No. 4 to No:7, which contained 1 M KCl.
110
G. S. Makowski and F. W Sunderman Jr
Preparation
of ApoTFIIIA
The four samples of TFIIIA were placed in dialysis bags and dialyzed at 4°C for 24 hr against three changes (1 L) of &fleer D (HEPES, 50 mM; KCI, 50 mM; DIT, 1 mM; pH 7.0). Aliquots (1 ml) were taken for protein and zinc analyses. The bags were resealed, rinsed with Bugler E (HEPES, 50 mM; KCI, 50 mM; DTI, 1 mM; l,lO-phenanthroline, 1 mM; pH 7.01, and dialysis was continued at 4°C for 24 hr against three changes (1 L) of Buffer E. The bags were placed in polypropylene bottles that contained Buffer F (HEPES, 50 mM; KC1 50 mM; pH 7.01, which had been degassed under vacuum and bubbled with ultrapure argon for 20 min, three times, to remove dissolved oxygen. The bottles were flushed with argon and the apoTFIIIA samples were dialyzed anaerobically at 4°C for 24 hr against three changes (1 L) of Buffer F. The bottles were placed in an argon-filled glove bag; the dialysis bags were opened, and aliquots (1 ml) were taken for protein and zinc analyses. The apoTFIIIA samples were stored at 4°C in argon-filled tubes. Protein Assays
Total protein was measured by two methods: a reference biuret procedure [25] and a dot-blot assay based on amido black staining of protein adsorbed on nitrocellulose membranes [26], using a 96-well filtration manifold (“Minifold-1” Schleicher and Schuell, Inc., Keene, NH). Both assays were standardized with bovine serum albumin (SRM No. 927, National Institute of Standards and Technology, Gaithersburg, MD). Zinc Assays
Zinc was measured at 213.9 nm by atomic absorption spectrometry with airacetylene flame and deuterium background correction, using a model 1lOOB spectrometer (Perkin-Elmer Corp., Norwalk, CT), with reference materials and quality assurance procedures recommended by Falchuk et al. [20]. The samples were diluted in “Brij-35” solution (polyoxyethylene-23-lauryl ether, 30 mg/L), as described by Perry [27]. Equilibrium
Dialysis
Equilibrium dialysis was performed as previously described [28-301, using several microdialysis apparatuses (model 935/411, Chemical Rubber Co., Cleveland, OH) that consisted of two matched acrylic blocks, each with five cylindrical half-cells (diameter 1 cm, capacity 1 ml), separated by a dialysis membrane. Each half-cell contained two plastic beads to facilitate mixing. The apparatuses were assembled, placed in a glove bag, and flushed with argon. Into one set of half-cells was placed 1 ml of apoTFIIIA solution; into the complementary set was placed 1 ml of ‘j5ZnCl,, (j3NiC12, or ‘“9CdCl, solutions in degassed Bufleer F. These solutions contained fixed concentrations of the radionuclides ( - 2 x 10h dpm/ml) and graded concentrations of nonradioactive ZnCl,, NiCl,, or CdCl, (2pM to 1.5 mM1. Control d’la 1ysis cells held either apoTFIIIA without the radioligands, or the radioligands without apoTFIIIA. The dialysis apparatuses were placed in a rotatory environmental chamber (SO cycles/min, 25°C New Brunswick Scientific Co., New Brunswick, NJ). After 18 hr, the microdialysis apparatuses were removed and samples were taken from each half-cell for
Zn’+, Ni*+, AND Cd *+ BINDING
TO TFIIIA
111
radioactivity measurements and protein analyses. Trials showed no significant differences between the radioactivity of samples taken after 18 vs 24 hr, indicating that equilibrium was attained under the experimental conditions. Nonspecific binding of 65Zn, 63Ni, or “‘Cd to the dialysis chamber and membrane was negligible. Liquid Scintillation Spectrometry Duplicate aliquots (10 ~1) of each sample were mixed in counting vials with 12 ml of “Optifluor” liquid scintillation fluid (Packard Instruments, Inc., Downers Grove, IL) and counted for 10 min or 1% error, using a “Tri-Garb” spectrometer (model 4530, Packard Instruments, Inc.) set at the optimal discriminator windows for 65Zn, 63Ni, or “‘Cd. Computations
and Statistical Methods
Computations of metal-binding parameters were based on a molecular weight of 38.5 kD for apoTFIIIA [6]. Data analyses and curve-fitting were performed by the “Ligand-4.1” program (Unit of Biostatistical Methodology, NIH, Bethesda, MD) [31], using a Macintosh IIfx computer (Apple Computer Co., Cuppertino, CA). The program adjusted the data from each experiment for differences in protein concentrations and generated binding curves and Scatchard plots, including confidence limits (+2 SD) for points around the fitted curves [32]. The program evaluated the “goodness-of-fit” of parametric and nonparametric models by three tests: (a) F-test of extra sum of squares, (b) root mean square (rms) error, and (cl runs-test of signs of residuals [33]. Hill analysis of the binding data was performed according to Weiland and Molinoff 1341, using computer programs that fitted linear or polynomial equations to the data (“Ultrafit”, Biosoft, Ltd., Ferguson, MO; “Cricket Graph”, Cricket Software, Inc., Malvern PA). Data are reported as mean -&SE, unless other error limits are specified.
RESULTS Analyses by SDS-polyacrylamide gel electrophoresis yielded a single protein band (N 38.5 kD) when gels, heavily loaded with the four batches of TFIIIA, were stained with Coomassie blue; the purity of the TFIIIA samples appeared to be 2 98%. The overall yield of TFIIIA was 22.8 mg/46 g of ovary (0.50 mg/gJ The protein was positively identified as TFIIIA by probing Western blots with anti-TFIIIA antibody (not shown). One sample of TFIIIA, concentrated by lyophilization, gave a protein concentration of 3.25 mg/ml, based on biuret analysis with NIST bovine albumin as the standard. Aliquots of this TFIIIA sample and a comparable solution of NIST bovine albumin were diluted l:lO, 1:20, 1:30, and 1:40 with BufSer F and analyzed by amido black staining in the dot-blot protein assay. Since the two proteins yielded calibration lines with identical slopes and intercepts, bovine albumin was used routinely as the standard for analyses of TFIIIA concentrations by the dot-blot assay. After dialysis against zinc-free buffer, protein concentrations of the four TFIIIA samples averaged 142 f 26 ,ug/ml and the zinc concentrations averaged 5.5 k 0.1 g-atom/mole. After exposure to l,lO-phenanthroline and further dialy-
112
G. S. Makowski and F. W. Sunderman Jr.
sis against zinc-free buffer, the zinc concentrations of the apoTFIIIA samples concentrations averaged 0.20 + 0.03 g-atom/mole. Interactions of Zn’+, Ni2+, and Cd2+ with apoTFIIIA were measured in eight equilibrium dialysis experiments, each testing several concentrations of one metal ion against one apoTFIIIA sample. No significant differences were obtained with the four apoTFIIIA samples, so the binding data for each metal ion were combined for statistical analyses. The experimental results are shown by computer-generated curves (Fig. 11, in which fractional binding of Zn’ ‘, Ni2’ , and Cd2’ to apoTFIIIA is plotted against total metal concentration. Transformations of these data as Scatchard and Hill plots are shown in Figures 2 and 3, and the calculated binding parameters are listed in Table 1. The data for Zn*+-, Ni*+-, and Cd’+-binding to apoTFIIIA were best-fitted by a model with two classes of binding sites, based on the F-tests, rms error terms, and runs-tests of residuals for each set of experiments. For Zn2’, the apparent dissociation constants (K,,“” and K,,““) of the two classes of binding sites were 1.0 x lo- 8 M and 2.6 X 1f.Y” M. The apparent binding capacities of the two classes were 0.8 + 0.5 and 9.6 + 0.3 g-atoms/mole, respectively, based on Scatchard analysis. Zn2+ data in the Hill plot were best-fitted by a second-order polynomial (y = 0.21x2 + 0.60x - 1.21; corr. coef. = 0.97), which gave an overall dissociation constant (K,‘” 1 of 2.5 X lo-” M for
Log[Tl
02 B/T
1
, -
NI
-5
-4
-3
LogPI
-55
-45
‘-ogt-0
-35
FIGURE 1. Equilibrium dialysis data for the binding of Zn” (middle panel), and Cd!‘,Opfb$~?p!%) to apoTFIIIA. The ordinate is B/T (i.e., the ratio of bound to total metal concentration); the abscissa is log(T) (PM) (i.e.. log,,, of total metal concentration (PM)). The computer-generated curves show the estimated mean binding ratios and & 2 SD confidence limits, based on a two-site model and calculated by nonlinear regrcssion analysis with the “Ligand 4.1” program.
Zn2+ 7 Ni*+ > AND Cd2+ BINDING
0
12
El
4 B/P
_ _ Cd _
_ _ 4
TO TFIIIA
113
FIGURE 2. Scatchard plots of the binding of Zn2+ (top panel), Ni2+ (middle panel), and Cd’+ (bottom panel) to apoTFIIIA. The ordinate is B/(F* P) and the abscissa is B/P, where B = bound metal concentration (PM), F = free metal concentration (PM), and P = apoTFIIIA concentration ( PM). The computer-generated curves show the best fits of the data (including k2 SD confidence limits), based on a two-site model and calculated by nonlinear regression with the “Ligand 4.1” program.
Zn’+-binding to apoTFIIIA. The Hill coefficient (nH = 1.18) was significantly greater than 1.0, which suggests positive cooperativity of the Zn2+-binding sites [321. For Ni2+, the apparent dissociation constants (K,,Ni and KdzNi) of the two classes of binding sites were 2.3 X 1O-5 M and 5.2 X low4 M, indicating that the affinity of apoTFIIIA for Ni2+ is much weaker than for Znzf. The apparent binding capacities of the two classes were 2.3 f 0.6 and 8.6 f 0.6 g-atoms/mole, respectively. Ni-‘+ data in the Hill plot were best-fitted by a second-order polpominal (y = 0.39x -‘0.56x - 0.74; corr. coef. = 0.961, which gave an overall K, ’ of 1.9 x 10m4 M for Ni’+-binding to apoTFIIIA. The Hill coefficient (n, = 1.20) was significantly greater than 1.0, which suggest positive cooperativity of the Ni*+-binding sites [32]. The Scatchard and Hill plots for Cd ‘+-binding to apoTFIIIA were distinctly different from those for Zn2+ or Ni’+. The apparent dissociation constants ( K,,Cd and KdzCd) of the two classes of Cd’+-binding sites were 2.8 X 10V6 M and 1.6 X lop4 M, respectively, indicating that the affinity of apoTFIIIA for Cd’+ is less that for Zn’+, but greater than for Ni*+. The apparent binding capacities of the two classes were 0.9 rfi 0.3 and 2.4 & 0.5 g-atoms/mole. Cd’+ data in the Hill plot were best-fitted by linear regression (y = 0.53x - 1.10; corr. coef. = 0.971, which yielded an apparent overall Kdcd of 1.3 X 10m4 M for Cd’+-binding to apoTFIIIA. The Hill coefficient (nH = 0.53) was significantly
2
2
21
27
zI?+
Ni2+
Cd2+
1.0 x 10-a (f0.5 x 10-8) 2.3 x 1O-5 (fO.8 x 10-5) 2.8 x 1O-6 (kO.4 x 10-61
&i, (mol/L)” (t0.3) 8.6 (kO.6) 2.4 (kO.5)
(kO.2 x 10-Y 5.2 x 10m4 (k 1.8 x 1O-4) 1.6 x 1O-4 (+0.2x 10-41
(kO.5)
( >bq3)
&36)
9.6
(g-at./mol)g
R2
2.6 x 10-s
(mold;L)’
K
0.8
R, (g-at./mol>”
Dialysis
0.53
1.20
1.18
nH
1.3 x 10-4
1.9 x 10-4
2.5 x 1O-5
(mol/L)
Kd
Hill Analysis+
*Parameters estimated by Scatchard analysis of equilibrium dialysis data (Fig. 21, using a nonlinear model-fitting program [29]. K,, and K,, are the apparent dissociation constants of the first and second classes of binding sites. R, and R, are the apparent capacities of the first and second classes of binding sites. ’ Parameters estimated by Hill analysis of equilibrium dialysis data (Fig. 3) [32], using the apparent binding capacities obtained by Scatchard analysis. The Hill coefficient (n,) is the slope of the curve at the point (K,) where half of the metal-binding sites are occupied. ’ Mean *SE.
2
30
Metal
Classes of Binding Sites
Scatchard Analysis*
of Zn’+, Nizf, and Cd’+ with apoTFIIiA, Assessed by Equilibrium
Number of Data Points
TABLE 1. Interactions
a
ti
.I.
N
Z _.
N +
P
116 G. S. Makowski and F. W Sunder-man Jr,
depleted the remaining zinc, leaving only 0.20 _t 0.03 g-atoms of Zn/mole of TFIIIA. Free-ion depletion by buffers (e.g., Tris or “Good’s” buffers) is a potential source of error in assays of metal-complexation by proteins [38, 391. HEPES buffer (4-(2-hydroxyethyl)piperazine-1-sulfonic acid), which has negligible affinity for metals [40, 411, was used in several previous studies of metal binding to Zn-finger proteins and peptides [21, 42, 431. Therefore, HEPES buffer was selected for the present study. Zn2+-binding to TFIIIA evidently occurs in the N-terminal and central regions, which contain the finger motifs. Based on x-ray fluorescence spectroscopy and EXAFS (extended x-ray absorption fine structure) patterns. Diakun et al. [441 deduced that TFIIIA isolated in Zn’+-containing media has 8 to 9 g-atoms of Zn/mole, tetrahedrally coordinated to thiol sulfur and imidazole nitrogen atoms in cysteines and histidines, respectively, of the 9 putative Zn-fingers. Hanas et al. [4.5] reported that removal of zinc from TFIIIA by chelation changes the conformation of the N-terminal half of the molecule, increasing its susceptibility to tryptic digestion and enhancing its fluorescence when derivatized with a fluorescent probe. Makowski et al. 1231 showed that 65Zn2+ binds to the cyanogen bromide cleavage fragments from the N-terminus and mid-portion of TFIIIA, but not to the C-terminal CNBr-fragment, which has no finger motifs. Frankel et al. [42] studied a synthetic 30-residue peptide with an amino acid sequence corresponding to the second finger-loop of TFIIIA. Based on changes of the circular dichroism spectrum that occurred when the Co*‘-substituted peptide was titrated with Znzt at pH 7.0 in HEPES buffer, they derived an apparent KdZn of 2.8 X lo-” M [421, which is lower than the present KdlZn value of 1.0 X 10e8 M for the native apoTFIIIA molecule. The present study confirms the observation of Han et al. [21] that TFIIIA has two affinity classes of Zn’+-binding sites, based on the release of zinc after additions of p-hydroxymercuriphenylsulfonate (PMPS) and PAR to TFIIIA in HEPES buffer at pH 7.0. Combination of PMPS and PAR liberated N 9 g-atom of Zn/mole of TFIIIA, of which - 4 g-atom rebound to TFIIIA when PMPS was removed 1211.According to Shang et al. [37], only two firmly bound g-atoms of Zn/mole are needed for TFEIIA to activate transcription of the 5S RNA gene by RNA polymerase III. From their finding that a chelating indicator (i.e., 4-(2-pyridylazohesorcinol, PAR) did not remove the two firmly bound Zn atoms from TFIIIA, Shang et al. [37] estimated the K, for these Zn atoms to be < lo-‘* M. The scatter of experimental points shown in Figures 1 to 3 of the present paper is considerable; the observed values for metal-binding affinities and capacities of TFIIIA should therefore be viewed as approximations that are in general agreement with the findings of Han et al. [21] and Shang et al. [37]. Zeng et al. 1461 showed by in vitro RNA and DNA binding assays that addition of thionein (apometallothionein) suppressed TFIIIA binding to the 5S RNA gene and to 5S RNA. Thionein also abrogated the activating effect of TFIIIA on RNA polymerase III-catalyzed synthesis of 5S RNA. The Kdz” of thionein is N 5 X lo- ” M [47], which is less than the KdlZ” value for apoTFIIIA, as reported herein. The present results support the conclusion of Zeng et al. [461 that thionein competition with TFIIIA for Zn”-binding is energetically feasible, and that an excess of metal-binding equivalents of thionein over the total amount of free and TFIIIA-bound zinc in the reaction mixture could account for suppressed binding of TFIIIA to the 5S RNA gene and to 5S RNA.
Zn’+, Ni*+, AND Cd*+ BINDING
TO TFIIIA
117
Thus, thionein might modulate gene expression by competing with TFIIIA and other Zn-finger proteins for intracellular Zn2+ stores [46]. In the present study, the Hi11 coefficients for Zn2+ or Ni*+ binding to apoTFIIIA were positive, consistent with positive cooperativity of the binding sites, while the Hill coefficient for Cd*+ was strongly negative, suggesting negative cooperatively or heterogeneity of Cd 2f-binding sites 1321.The observed metal-binding capacity of apoTFIIIA for Cd2+ was much less than for Zn2+ or Ni*+. Therefore, the authors suspect that heterogeneous combinations of Cys residues in the various finger motifs may be responsible for Cd*+ binding, rather than the paired Cys and His residues of the nine individual fingers, which probably bind Zn*+ and Ni2”. Sunderman and Barber [48] proposed that the genotoxicity of certain divalent metal ions could involve substitution of the foreign metals for Zn2+ in finger motifs of transcription factors, hormone receptors, and proteins encoded by oncogenes. This hypothesis has not yet been tested in vivo, but some in vitro studies are relevant. Makowski et al. [23] ranked the affinity of TFIIIA for metals as follows: Zn2+ = Cu2+ > Hg2+ > Cd2+ > Co*+ > Ni2+ > Mn*+, based on a Zn-blot competition assay; their findings in regard to Zn2+, Cd’+, and Ni2’ were confirmed by the present study. Hanas et al [36] did not observe specific binding of TFIIIA to the 5s RNA gene after additions of Co2+, Ni2’, Mn2+, or Fe2+ to apoTFIIIA, although Zn2+ restored the binding. Predki and Sarkar [49] substituted metal ions for Zn2+ in finger motifs of the C2C2 class, using an apopolypeptide that contained the DNA-binding domain of the human estrogen receptor. Affinity of the apopolypeptide for metals were ranked as follows: Cu*+ > Cd*+ > Zn2+ > Co2+ > Ni 2+, based on a Zn-blot competition assay. After dialysis against Co2+, Cd2+, or Zn2+, the polypeptide became capable of binding specifically to the estrogen response element (ERE); after dialysis against Cu*+ or Ni*+ binding of the polypeptide to the ERE was not detected. Predki and Sarkar [49i concluded that certain metals might produce toxic effects by altering the DNA-binding properties of hormone receptors. Based on their findings and the present results, Cd2+ is more likely to substitute for Zn2+ in vivo in proteins with the C2C2 class of finger motifs (e.g., the superfamily of hormone and retinoid receptors) than in proteins with the C2H2 class (e.g., TFIIIA); on the other hand, Ni2+ seems unlikely to substitute for Zn2+ in proteins with either class of finger motifs. The authors are grateful to Sidney M. Hopfer, Bonnie L. Be& and Paul A. Zkzmerfor helpful advice, Kevin R. Sweeney and Elizabeth R Sunderman for statistical assistance, Marilyn R. Plowman, Odette Zaharia, and Jennifer Martin for technical assistance, and Robert G. Roeder and Elizabeth Morefield (Rockefeller University, New York) for a generous gift of the monoclonal anti-TFIZZA antibody. This study was supported by research grants from the National Institute of Environmental Health Services, the March of Dimes, and Northeast Utilities,Inc.
REFERENCES 1. D. R. Engelke, S. Y. Ng, B. S. Shastry, and R. G. Roeder, Cell 19, 717 (1980). 2. E. P. Geiduschek and G. P. Tocchini-Valentini, Ann. Rev. Biochem. 57, 873 (1988). 3. A. P. Wolffe and D. D. Brown, Science 241, 1626 (1988).
118
G. S. Makowski and F. W. Sunderman Jr.
4. 5. 6. 7. 8. 9. 10. 11. 12.
13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
26. 27. 28. 29. 30. 31.
32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.
B. Picard and M. Wegnez, Proc. Natl. Acad. Sci. U.S. A. 76, 241 (1977). H. R. B. Pelham and D. D. Brown, Proc. Natl. Acad. Sci. U.S.A. 77, 4170 (1980). A. M. Ginsberg, B. 0. King, and R. G. Roeder, Cell 39, 479 (1984). J. Hayes, T. D. Tulhs, and A. P. Wolffe, .I. Biol. Chem. 264, 6009 (1989). J. Miller, A. D. McLachlan, and A. KIug, EMBO J. 4, 1609 (1985). R. S. Brown, C. Sander, and P. Argos, FEBS Lett. 186, 271 (1985). B. S. Shastry, Prog. Biophys. Mol. Biol. 56, 135 (1991). A. P. Wolffe, Biochem. J. 278, 313 (1991). C. W. Wu, E. E. Sierra, T. J. Daly, G. J. Wu, and F. Y. H. Wu, in The Eukaryotic Nucleus: Molecular Biochemistry and Macromolecular Assemblies, S. H. Wilson, Ed., Raven Press, Caldwell, NJ, 1990, Chap. 17. J. M. Berg, Metal Ions Biol. Syst. 25, 235 (1989). A. Klug and D. Rhodes, Cold Spring Harbor Symp. Quant. Biol. 52, 473 (1987). B. L. Vallee and D. S. Auld, Biochemistry 29, 5647 (1990). F. Y. H. Wu, in Structure and Function of Nucleic Acids and Proteins, F. Y. H. Wu and C. W. Wu, Eds., Raven Press, New York? 1990, Chap. 3. B. L. Vallee and A. Galdes, Adr. Enzymol. Relat. Areas Mol. Biol. 56, 288 (1984). D. S. Auld, Meth. Enzymol. 158, 13 (1988). B. Holmquist, Meth. Enqmol. 158, 6 (1988). K. H. Falchuk, K. L. Huh, and B. L. Vallee, Meth. Enzymol. 158, 422 (1988). M. K. Han, F. P. Cyran, M. T. Fisher, S. H. Kim, and A. Ginsburg, .I. Biol. Chem. 265, 13792 (1990). U. K. Laemmli, Nature 227, 680 (1970). G. S. Makowski, S. M. Lin, S. M. Brennan, H. M. Smilowitz, S. M. Hopfer, and F. W. Sunderman Jr., Biol. Trace Elem. Res. 29, 93 (1991). A. Kramer and R. G. Roeder, J. Biot. Chem. 258, 1915 (1983). T. Peters Jr., G. T. Biamonte, and B. T. Doumas, in Selected Methods of Clinical Chemistry, W. R. Faulkner and S. Meites, Eds. American Association for Clinical Chemistry, Washington, DC, 1982, Vol. 9, p. 317. W. Schaffner and C. Weissmann, Anal. Biochem. 56, 502 (1973). D. F. Perry, J. Assoc. Off. Anal. Chem. 73, 619 (1990). W. C. Callan and F. W. Sunderman Jr., Res. Commun. Chem. Pathol. Pharmacol. 5, 459 (1973). S. L. Guthans and W. T. Morgan, Arch. Biochem. Biophys. 218, 320 (1982). N. Raos and K. S. Kasprzak, Fund. Appt. Toxicol. 13, 816 (1989). P. J. Munson, A User’s Guide to LIGAND, Unit of Biostatistical Methodology, National Institute of Child Health and Human Development, Bethesda, MD, 1990, p. 136. P. J. Munson and D. Rodbard, Anal. Biochem. 107, 220 (1980). P. J. Munson, Meth. Enzymol. 92, 543 (1983). G. A. Weiland and P. B. Molinoff, Lije Sci. 29, 3 13 (1981). N. W. Cornell and K. E. Crivaro, Anal. Biochem. 47, 203 (1972). J. S. Hanas, D. J. Hazuda, D. F. Bogenhagen, F. Y. H. Wu, and C. W. Wu, J. Biol. Chem. 258, 14120 (1983). Z. Shang, Y. D. Liao, F. Y. H. Wu, and C. W. Wu, Biochemistry 28, 9790 (1989). D. E. Allen, D. J. Baker, and R. D. Gillard, Nature 214, 906 (1967). R. Nakon and C. R. Krishnamoorthy, Sciences 221, 749 (1983). T. Scott and M. Eagleson, Concise Encyclopedia of Biochemistry, Walter der Gruyter, Berlin, 1988, 2nd Ed. p. 80. F. W. Wagner, Meth. Enzymol. 158, 21 (1988). A. D. Frankel, J. M. Berg, and C. 0. Pabo, Proc. Natl. Acad. Sci. U.S.A. 84, 4841 (1987). J. M. Berg and D. L. Merkle, J. Am. Chem. Sot. 111,3759(1989).
Zn2+,
Ni*+, AND
Cd*+
BINDING
TO TFIIIA
119
44. G. P. Diakun, L. Fairall, and A. Klug, Nature 324 698 (1986).
45. J. S. Hanas, A. L. Duke, and C. J. Gaskins, Biochemistry 28, 4083 (1989). 46. J. Zeng, B. L. Vallee, and J. H. R. Kiigi, Proc. Natf. Acad. Sci. U.S.A. 88, 9984 (1991). 47. J. H. R. K;igi and Y. Kojima, Experienfia Suppl. 52, 25 (1987). 48. F. W. Sunderman Jr. and A. M. Barber, Ann. Clin. Lab. Sci. 18, 267 (1988). 49. P. F. Predki and B. Sarkar, .I. Biol. Chem. 267, 5842, (1992). Received January 23, 1992; accepted March 31, 1992