Transcription factor IIIA (TFIIIA) was isolated from Xenopus ovary and treated with 1,10-phenanthroline to remove zinc. The interactions of apoTFIIIA with Zn2+, Ni2+, and Cd2+ were studied by equilibrium dialysis under anaerobic conditions (pH 7.0, 25-degrees-C), using (ZnCl2)-Zn-65, (NiCl2)-Ni-63, and (CdCl2)-Cd-109 as the radioligands. The data for binding of Zn2+, Ni2+, and Cd2+ to apoTFIIIA were best-fitted by a model with two classes of binding sites. For Zn2+, the apparent dissociation constants (K(dl)Zn and K(d2)Zn) for the high- and low-affinity sites were 1.0 x 10(-8) and 2.6 x 10(-5) M; the apparent binding capacities of the two classes were 0.8 +/- 0.5 and 9.6 +/- 0.3 g-atoms of Zn/mol; the Hill coefficient was 1.18, consistent with positive cooperativity of Zn-binding sites. For Ni2+, the apparent K(dl)Ni and K(d2)Ni values were 2.3 x 10(-5) and 5.2 x 10(-4) 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 Cd2+, the apparent K(dl)Cd and K(d2)Cd values were 2.8 x 10(-6) and 1.6 x 10(-4) M; the apparent binding capacities were 0.9 +/- 0.3 and 2.4 +/- 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 helps to resolve a controversy about the zinc content of purified TFIIIA. Second, it shows that the K(dl)Zn of apoTFIIIA is less than the reported K(d)Zn of thionein, consistent with the hypothesis that thionein modulates gene expression by competing with TFIIIA and other Zn-finger proteins for intracellular Zn2+ stores. Third, it confirms previous indirect evidence that the affinity of apoTFIIIA for Zn2+ is much greater than for Cd2+, and that the affinity for Cd2+ is greater than for Ni2+.