The nonactivated progesterone receptor is a nuclear heterooligomer.

The discovery of the nuclear localization of estradiol and progesterone receptors in the absence of the steroid hormone has led to reconsideration of the model of cytoplasmic to nuclear translocation of these receptors upon exposure to hormone. Unoccupied nonactivated receptors are thought to be weakly bound to nuclei of target cells from which they are leaking during tissue fractionation and thus found in the cytosol fraction of homogenates in a nontransformed heterooligomeric "8-9 S" form, which includes hsp90. However, no direct biochemical evidence has yet been obtained for the presence of such heterooligomers in the target cell nucleus, possibly because it dissociates in high ionic strength medium used for extraction of the nuclear receptor. We took advantage of the combined stabilizing effects of tungstate ions and antiprogestin RU486 to extract a nuclear non-DNA binding nontransformed 8.5 S-RU486-progesterone receptor complex from estradiol-treated immature rabbit uterine explants incubated with the antagonist. As demonstrated by immunological criteria and by irreversible cross-linking with dimethylpimelimidate, the complex contained, in addition to the hormone binding unit, hsp90, and p59, another nonhormone binding protein. Control experiments carried out with the progestin R5020 yielded the expected nuclear transformed DNA binding 4.5 S-R5020-progesterone receptor complex. These results offer evidence for two distinct forms of steroid receptor in target cell nuclei. Besides the classical "4 S" agonist-receptor complex, tightly bound to the DNA-chromatin structure and in all probability able to trigger the hormonal response, we have observed in the RU486-bound state a non-DNA binding nontransformed 8.5 S form, presumably already present in the nucleus in the absence of hormone and representing the native nonactive form of the receptor.

The discovery of the nuclear localization of estradiol and progesterone receptors in the absence of the steroid hormone has led to reconsideration of the model of cytoplasmic to nuclear translocation of these receptors upon exposure to hormone. Unoccupied nonactivated receptors are thought to be weakly bound to nuclei of target cells from which they are leaking during tissue fractionation and thus found in the cytosol fraction of homogenates in a nontransformed heterooligomeric "8-9 S" form, which includes hsp90. However, no direct biochemical evidence has yet been obtained for the presence of such heterooligomers in the target cell nucleus, possibly because it dissociates in high ionic strength medium used for extraction of the nuclear receptor.
We took advantage of the combined stabilizing effects of tungstate ions and antiprogestin RU486 to extract a nuclear non-DNA binding nontransformed, 8 Biochemical analyses of steroid hormone receptors have been hampered by extraction conditions. In the hormone-free state, receptors are easily extracted in low salt-containing medium and are recovered in the cytosoluble fraction (cytosol) of target tissue homogenates, a result contrasting with immunohistochemical data demonstrating the nuclear localization of estradiol and progesterone receptors (PR),' including * This work was supported by the Institut National de la Santk et de la Recherche Mkdicale and by the United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
' The abbreviations used are: PR, progesterone receptor; GR, glucocorticosteroid receptor; R5020, 17,RU486, hsp90, heat shock protein of Mr 90,000; p59, protein of Mr 59,000. in the absence of steroid (l-3). These unoccupied cytosol receptors are heterooligomers which include hsp90 (4)(5)(6)(7)(8) and, in the case of mammals, another nonhormone binding protein p59 (9). In the presence of hormone, the receptors become difficult to extract in low salt medium, and high ionic strength (~0.4 M KCl) is then required to recover the major part from nuclei (lo), probably because receptors are tightly bound to chromatin components. Unfortunately, ionic strength is well known to also dissociate the large non-DNA binding nontransformed ("8 S") form of steroid receptors found in the cytosol (11,12). Numerous reports have indicated that transition metal oxyanions protect this large receptor form from high ionic strength-or heat-induced dissociation (transformation) and the subsequent ability to bind DNA. Among these agents, molybdate ions have been the most widely used, and tungstate ions have also been described as effective (13,14). In this work,' we used the higher stabilization effect of tungstate ions, as compared with molybdate ions, against the KCl-induced transformation of the -9 S rabbit uterus PR, to identify a nuclear form of PR containing hsp90 and p59. This form was also stabilized by the antiprogestin RU486 which has been previously shown to maintain the association between rabbit PR and hsp90 (15).
' A preliminary report was presented at the 7lst Annual Meeting of the Endocrine Society, Seattle, WA, June 21-24,1989 After incubation, pieces of tissues were rinsed 2 times with ice-cold 20 mM tungstate containing phosphate buffer saline followed by centrifugations (700 x g for 10 min). Homogenization was carried out in 2 volumes of PGW buffer and protease inhibitors, as described previously (7), using a glass/glass Potter-Elvehjem homogenizer. The homogenate was centrifuged at 105,000 X g for 1 h, and the pellet was washed once with 50 ml of Triton X-100 (1%) containing buffer (5 InM

AND DISCUSSION
First, we tested the effect of tungstate ions on the stabilization of rabbit uterus cytosol PR. Experiments depicted in Fig. 1  sediments as a 9.5 S entity on density gradient prepared in PG buffer (Fig. lu) and as 8,5 S when the PG buffer contains 0.4 M KC1 (Fig. lb). In contrast to tungstate ions, molybdate ions were unable to prevent the KCl-induced decrease to 4.5 S of the 9.5 S-PR complexes (Fig. lb). The 9.5 S PR form, stabilized or not by oxyanions, includes in addition to the hormone binding unit, two nonhormone binding proteins, hsp90 (4,7,9) and p59 (9). Upon incubation with EC1, the 9.5 S PR shifts to 11 S (Fig. 1~). On the contrary, no shift of the 8.5 S-PR observed in the 0.4 M KCl-containing gradient occurred after incubation with EC1, suggesting the loss of p59 (Fig. lb). We checked that this result was due neither to a direct action of KC1 on the recognition of the epitope by the antibody nor to a dissociating effect on antigen-antibody complexes. In fact, when the cytosol PR (initially 9.5 S in low salt medium) was cross-linked by dimethylpimelimidate (19), its sedimentation coefficient was 8.5 S in the KCl-containing gradient, and it reacted with EC1, shifting to -10 S (Fig. lc). From these experiments, we conclude that 1) tungstate ions stabilize, better than molybdate ions, receptor-hsp90 interaction against the dissociating effect of high salt during the 16-h ultracentrifugation and 2) in the absence of cross-linking, tungstate ions do not prevent the release of p59 from the rest of the heterooligomer under the same conditions. Identical results have been obtained with other steroid receptors cm RU486 has been shown to stabilize the interaction between hsp90 and GR or PR (15,(23)(24)(25) -30% of radioactivity was found in the nuclear fraction after 60-90 min of incubation.
Western blotting of nuclear extracts with mPR II revealed a major band corresponding to progesterone binding, PR B unit (-120 kDa), and two minor bands, at -85 kDa (likely subunit A) and -75 kDa. In addition, positive signals were observed with AC88 and EC1 at Mr 90,000 and 59,000, respectively. AC88 was described to recognize hsp90 in a large variety of species (17), and we showed recently that EC, recognizes the Mr 59,000 protein in several mammals (22). These findings indicate that hsp90 and p59, respectively, were also present in the nuclear extracts (Fig. 2, u and b, 1). To assess if the three proteins were associated to form the nuclear heterooligomeric PR, immunoadsorption of the nuclear extracts on protein A-Sepharose-mPR II affinity resin was performed, and the eluates were analyzed by Western blotting (Fig. 2, u and b, 2).
In the RU486 experiment (Fig. 2u, 2), in addition to the PR units, both hsp90 and p59 were also detected in the eluate of the anti-PR column, suggesting the association of the three components in a multimeric structure. In contrast, in the R5020 experiment (Fig. 2b, 2), only PR subunits (-120 and 85 kDa) were found, suggesting that dissociation of the preexisting heterooligomer occurred upon agonist binding and, at the same time, indicating the lack of nonspecific interactions during these experimental procedures. In parallel experiments, we studied the nuclear PR by sucrose gradient ultracentrifugation analysis. In the RU486 incubation experiment, only a -8.5-S peak, nondisplaceable by EC, was observed (not shown), suggesting release of p59 and thus apparently in contradiction with the results obtained after immunopurification.
However, if the nuclear extract was treated with dimethylpimelimidate immediately after extraction, then part of 8.5 S PR analyzed after cross-linking in high salt gradient was displaced to -10 S after incubation with EC (Fig. 3~). The complete dissociation of p59 in the absence of cross-linking is best explained by the prolonged incubation of the p59-containing heterooligomer in the 0.4 M KC1 extract with the antibody. The partial dissociation of p59 in the cross-linking experiment is also probably due to the lh exposure to high salt during extraction.
It is likely that the 0.4 M KC1 nuclear extract contains a heterogeneous population of RU486-PR complexes, some including p59 and others lacking p59 that has dissociated in the presence of high salt. Conversely, dissociation does not seem to occur when immunoadsorption is immediately performed (Fig. 2u, 2). on Affi-Gel lo-EC, affinity resin was also performed. Only part of the RU486-PR complexes was bound to the column; the eluted PR migrated as a sharp -9 S peak in density gradient and was completely displaced to -10.5 S by EC1 (Fig.  3b). This result demonstrates that it is possible to isolate a homogeneous population of nuclear heterooligomeric PR containing hsp90 and p59. Unfortunately, similar experiments of immunoadsorption and sucrose gradient analysis are not passible with any anti-hspQ0 antibody that we tested, namely AC88 which does not interact with hsp90 included in the 8.5 S nontransformed form of PR (17). While these results were obtained with RU486 as ligand, in the R5020 experiment and as expected from data of Fig. 2b, 2, no significant 8-9 S large PR form was found in the nuclear extract. Only a -4.5 S PR, neither displaced by EC1 (Fig. 3~) nor retained on Affi-Gel lo-EC1 affinity resin (not shown), was found. This form, unlike the RU486-bound one, was apparently unstable and rapidIy released free radioactive R5020. According to the immunoaffinity data reported above (Fig. 2b, 2), it is likely that this 4.5 S PR did not appear during incubation and ultracentrifugation but rather was initially present in the nuclear extract of agonist-treated explants. Control experiments demonstrated that always cytosol contamination can be ruled out, since lactate dehydrogenase was never found in nuclear extracts, while normal lactate dehydrogenase levels were measured in cytosol supplemented or not with 0.4 M KCl. From these results, we conclude that hsp90 as well as p59 are originally associated to the PR in the nuclei of immature rabbit uterus cells. Tungstate ions do not induce artifactual formation of heterooligomer during homogenization and extraction, as indicated by results with R5020 (Fig. 3~); these ions appear only required to maintain hsp90 bound to the receptor exposed to high salt in the absence of chemical cross-linking.
The latter is still necessary to prevent the release of p59.
We also measured the DNA cellulose binding of the nuclear receptor, a criterion for in uitro activation. After 60 min of incubation of the explants with agonist or antagonist, the DNA cellulose binding of nuclear 8.5 S [3H]RU486-PR, as expected for a nontransformed receptor form, was much lower than that of the 4.5 S yH]R5020-PR (5 uersus 45%). How does the observation of a nuclear stabilized RU486-8 S receptor form fit with data related to receptor function and RU486 mechanism?
Altogether, our results are consistent with the decrease of RU486-PR and -GR complexes tight binding to nuclei, as compared with that of agonist-receptors complexes (26,27), and the lack of down-regulation observed biochemically (28) or immunohistologically (29) after RU486 treatment, in contrast to the effect of progestin (30). The subcellular distribution of GR is still a controversial matter, but there is definitively nuclear GR in the absence of hormone and agonist-induced increased binding to nuclei (31). Stabilization of 8 S-GR may also explain that experimentally the chemical methylation of hormone response element of the glucocorticosteroid-regulated tyrosine aminotransferase gene is possible, in intact cells, in the presence of RU486, contrary to dexamethasone (32). Differently, other observations are not in favor of the hsp90 stabilization hypothesis and its consequence on RU486-induced lack of DNA binding of the receptor. For instance, in uitro binding of RU486-PR and -GR complexes to corresponding HREs (33, 34) was reported, but the receptors which were used were dissociated from hsp90, and this experimental situation may be artifactual. Our data may also appear to contradict reports of competition by RU486-GAL-GR chimera (35) or RU486-PR (36) complexes for the transcriptional activity of GAL,4 or constitu-tively active truncated PR, respectively, suggesting total or partial release of hsp90 from receptors upon binding of RU486 and subsequent binding of RU486-receptor complexes to HRE. Antihormone activity would then be due to the abortive nature of the complexes at the DNA binding level (imperfect interaction) or after it (e.g. steric hindrance precluding binding of transcription factor(s) to the DNA or the receptor itself).
In fact, there is no direct evidence for DNA binding activity of RU486-receptor complexes in intact cells, and the competitive effect in the transcription experiments may simply be due to physically impeded access to DNA of active GAL4 or truncated receptor by the nuclear stabilized RU486-8 S receptor. In any case, it is presently impossible to reason in terms of "simple" receptor-DNA interaction with models not taking into account transcription factors, probably other chromatin proteins, phasing of nucleosomes (as demonstrated in the promoter region of the glucocorticosteroid-inducible mouse mammary tumor virus (37)), and yet undefined features of chromatin structure. Presently no data contradict the hspQO-stabilizing hypothesis, but it may well be that the anithormone effect depends on several, possibly sequential, molecular mechanisms. For instance, there could be a retarding effect on activation, and secondarily, formation of abortive RU486-receptor complexes as described above. In fact, we have observed an increase of DNA cellulose binding (23%) of [3HJRU486-PR complexes isolated from nuclei of explants exposed for 90 min to the antihormone.
In parallel, we noted the appearance of a form sedimenting at -6 S with a concomitant decrease of 8.5 S RU486-PR (not shown). With regard to previous reports on either the estrogen (8) or the glucocorticosteroid (38) receptors, the 6 S form may be a dimeric form of PR, as recently demonstrated (39). It is clear that more work is still needed to elucidate the RU486 mechanism in detail. It remains that the present report strongly suggests that, originally in the absence of hormone as after stabilization by RU486, the nuclear receptor is present in a heterooligomeric non-DNA binding form, presumably inactive.3 Even if a role of nuclear p59 in the heterooligomeric form of steroid receptor has not yet been established, we propose that it may anchor the hspQO-receptor complex within the nucleus until hormone binds to receptor which is then released and can initiate the response.