The cytosolic untransformed molybdate-stabilized glucocorticoid-receptor complex from rat liver was eluted as a heterogenous peak containing two components with Stokes radii (Rs) of 8.3 nm and 7.1 nm when analyzed by size-exclusion HPLC even in the absence of molybdate. In contrast, the highly purified glucocorticoid receptor yielded a sharp symmetrical peak of Rs = 7.1 nm. We demonstrate that the 7.1-nm component could not result from a proteolytic degradation of the 8.3-nm receptor form. The same receptor heterogeneity was observed in thymus cytosol which contains less proteases than liver. After labeling with [3H]dexamethasone 21-mesylate and SDS/PAGE the same 94-kDa receptor band was revealed in both the 8.3-nm and 7.1-nm forms. Immunoblotting experiments showed that both the 94-kDa hormone-binding subunit and the 90-kDa heat-shock protein were present in the two different receptor forms. The 8.3-nm receptor form was converted to the 7.1-nm receptor form after treatment by ribonuclease A in the presence of molybdate and this effect was dose-dependent, being completely prevented by placental ribonuclease inhibitor (RNasin). In contrast, in the presence of molybdate, the 7.1-nm receptor form was ribonuclease-insensitive. Treatment of cytosol with RNase A in the absence of molybdate, partially shifted the untransformed receptor towards the 5.2-nm transformed receptor form. This effect was abolished by placental ribonuclease inhibitor. RNase S protein, an enzymatically inactive proteolytic fragment of RNase A, or S1 nuclease, which is specific for single-stranded nucleic acids, were ineffective when used instead of RNase A. In contrast, cobra venom endonuclease, which preferentially attacks double-stranded regions of small RNAs, caused a complete conversion of the 7-8-nm untransformed receptor to the 5.2-nm transformed receptor form. These results were not observed in the presence of molybdate. Addition of RNasin prior to heating cytosol in the absence of molybdate did not prevent the receptor from dissociating to the 5.2-nm form, suggesting that an endogenous RNase is not involved in the transformation process. The 7.1-nm receptor form was shifted to a 9.2-nm complex when incubated with an excess of GR 49 antireceptor antibody, whereas the 8.3-nm receptor form did not bind to the antibody.(ABSTRACT TRUNCATED AT 400 WORDS)