We analysed the inhibitory effects in vitro and in vivo of several metal ions on aldosterone binding to the rat kidney mineralocorticoid receptor with the purpose of assessing possible toxic effects of those ions on sodium retention, as well as to obtain information on receptor structural requirements for ligand binding. For the assays in vitro, the inhibitory effects of 20 metal ions were analysed on steroid-binding capacity for renal receptor cross-linked to 90-kDa heat-shock protein (hsp90) by pretreatment with dimethyl pimelimidate. Cross-linking prevented the artifactual dissociation of hsp90 (and, consequently, the loss of steroid binding) from the mineralocorticoid receptor due to the presence of high concentrations of salt in the incubation medium. Cross-linked heterocomplex showed no difference in ligand specificity and affinity with respect to native receptor, but increased stability upon thermal- or ionic-strength-induced destabilization was observed. Treatments in vitro with metal ions in the range 10(-8)-10(-1) M resulted in a differential inhibitory effect for each particular ion on aldosterone binding. Using the negative logarithm of metal concentration for 50% inhibition, the ions could be correlated with their Klopman hardness constants. The analysis of this relationship led us to postulate three types of reaction: with thiol, imidazole and carboxyl groups. The essential role played by these residues in steroid binding was confirmed by chemical modification of cysteines with dithionitrobenzoic acid, histidines with diethyl pyrocarbonate and acidic amino acids with Woodward's reagent (N-ethyl-5-phenylisoxazolium-3'-sulphonate). Importantly, the toxic effects of some metal ions were also observed by treatments in vivo of adrenalectomized rats on both steroid-binding capacity and aldosterone-dependent sodium-retaining properties. We suggest that those amino acid residues are involved in the activation process of the mineralocorticoid receptor upon steroid binding. Thus toxic effects observed with these metal ions may be a consequence of modifications of those essential groups. Our results support the notion that toxicity of metals on renal mineralocorticoid function may be predicted according to their chemical hardness.