J. Biochem. 112, 535-540 (1992)

In Vitro Formation of Estrogen Receptor-Heat Shock Protein 90 Complexes1 Koichi Inano,* Takehiko Ishida,** Saburo Ornate,*1** and Tsuneyoshi Horigome*'" 'Department of Biosystem Science, Graduate School of Science and Technology, and "Department of Biochemistry, Faculty of Science, Niigata University, 2-Igarashi, Niigata, Niigata 950-21 Received for publication, April 9, 1992

We previously showed that the 9 S estrogen receptor can be reconstituted from purified vero ER (estradiol binding subunit) and purified hsp 90 (heat shock protein 90) in vitro [Inano, K. et aL (1990) FEBS Lett. 267, 157-159]. In this study, we further characterized our reconstitution system to investigate the mechanism underlying the formation of 9 S ER. When a vero ER preparation stored at 4*C for more than 20 h after affinity chromatography was used for the reconstitution of 9 S ER, 0.5 M NaSCN was essential, but not Na2MoO, or other reagents. When, however, vero ER was used within 3 h after dissociation from an affinity resin, 9 S ER could be reconstituted in a relatively high yield without NaSCN. Moreover, if such a fresh vero ER preparation was used, 9 S ER could be reconstituted in the absence of NaSCN from not only unoccupied vero ER but also the occupied form. From these results it was suggested that the conformation of purified vero ER tends to change quickly in a time dependent manner, and so a chemical perturbant, NaSCN, is generally necessary for the reconstitution of 9 S ER from purified vero ER and purified hsp 90. The concentration of hsp 90 required for the reconstitution was only about 1.0 //M, which was lower than its physiological concentration. Based on these results, the mechanism underlying the formation of 9 S ER was discussed.

In cytosol fractions prepared with a hypotonic buffer solution, estrogen receptors (ERs) are recovered as large complexes with a sedimentation coefficient of 8-9 S, which are stabilized by sodium molybdate {1-3). These untrans-formed estrogen receptor complexes contain the steroid binding subunit {vero ER) and the non-hormone-binding subunit of a 90 kDa protein (4). It was confirmed that this 90 kDa protein was heat shock protein 90 (hsp 90) by using an anti-hsp 90 antibody {4, 5). It has also been confirmed that all other members of the untransformed steroid hormone receptor family commonly contain hsp 90 (5, 6). In the ER system, it was demonstrated that the untransformed 8-9 S ER might contain two vero ER and two hsp 90 molecules (7). In glucocorticoid (8) and progesterone receptor (9, 10) systems, the untransformed 8-9 S receptors contain not only hsp 90, but also heat shock protein 70 and some other proteins, respectively. It was shown in earlier studies that the function of hsp 90 associated with steroid receptors is to mask receptor functions, because untransformed 8-9 S receptors could not bind to nuclei or non-specific DNA, and 4 S receptors could bind to nuclei and DNA after a transformational change {11-13). However, some researchers doubted the biological significance of 9 S ER, because 9 S ER could only be observed at a nonphysiological salt concentration, and the 1 This work was supported in part by a Grant-in-Aid from the Ministry of Education, Science and Culture of Japan. Abbreviations: CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]1-propanesulfonate; Ej, 17/9-estradiol; ER, estrogen receptor; Hsp 90, heat shock protein with a molecular mass of about 90 kDa; PMSF, phenylmethanesulfonyl fluoride; vero ER, estradiol binding subunit of the estrogen receptor.

Vol. 112, No. 4, 1992

535

molybdate ion was necessary to obtain stable 9 S ER {14). But recently it was shown that hsp 90 in the untransformed estrogen receptor played some role in the signal transduction pathway in vivo {15). It has been reported that the transcriptional activation by ER depending on estradiol was observed in the case of the expression of ER in the presence of hsp 90 at the wild type level in the cells, but was greatly reduced in the case of expression of ~ 5% of the wild type level of hsp 90 in the cells. This result constitutes the first biological evidence that the untransformed ER containing hsp 90 plays a significant role in living cells. So the untransformed ER containing hsp 90 seems to be very significant and interesting. But, it is unclear whether or not the untransformed ER contains proteins other than vero ER and hsp 90, how the ER complexes are constructed, how the ER complexes work in living cells, and so on. Therefore, a system for the reconstitution of untransformed estrogen receptors from purified preparations is necessary to investigate these points in more detail. There have been several pioneer studies on the reconstitution of the 8 S estrogen receptor {16, 17) and androgen receptor {18). However, the 8 S ER promoting factors reported {16,17) have not yet been highly purified, and also these factors seemed to be different from hsp 90. Recently, regarding glucocorticoid (79) and progesterone {20) receptor systems, reconstitution systems for untransformed receptors with hsp 90 and/or some other proteins involving the use of a rabbit reticulocyte lysate were reported. But these systems are inconvenient for examining the functions of 8-9 S receptors in detail, because they involve the direct use of a reticulocyte lysate instead of purified hsp 90. On the other hand, we have reported preliminarily that 9

536

K. Inano et aL

S ER could be reconstituted from purified vero ER and purified hsp 90 (21). The reconstituted ER exhibited a sedimentation coefficient of about 9 S, which was very similar to that of the cytosolic untransformed ER. The sedimentation coefficient decreased to about 5 S on treatment with a high salt buffer. Moreover, the sedimentation coefficient of the reconstituted 9 S ER increased on reaction with anti-hsp 90 antiserum prior to the centrifugation (21). These observations suggested that the reconstituted ER complexes actually consisted of purified vero ER and purified hsp 90. In this study, we refined the reconstitution system and found that a chaotropic ion, thiocyanate, was very useful for the reconstitution of 9 S ER. Moreover, we showed that the reconstitution could be accomplished in a relatively high yield without the chaotropic ion when very fresh vero ER was used. Based on these results, the mechanism underlying the complex formation was discussed. EXPERIMENTAL PROCEDURES

Materials—The following compounds were used: a radioactive hormone, 17£-[6,7- 3 H(N)]estradiol (sp. act., 48.3 Ci/mmol), from New England Nuclear; HEPES and CHAPS from Dojin Lab. (Kumamoto); and 17/3-estradiol, Tris, sodium thiocyanate, and other reagents from Wako Chemical (Osaka). Sepharose 6B and PD-10 columns were purchased from Pharmacia (Tokyo); and Whatman DE52 DEAE-cellulose was from Funakoshi (Tokyo). Buffers—The following buffer solutions were used: buffer A: 25 mM HEPES-KOH (pH 7.5) containing 10 mM Na2MoO4, 1 mM EDTA, 0.02% NaN3) 10 mM thioglycerol, 0.1 mM PMSF, and 3 /*g/ml antipain and leupeptin; buffer B: buffer A containing 0.5 M NaSCN, 2 mM CHAPS, 10% dimethylformamide, and 1 //g/ml antipain and leupeptin; buffer C: 18 mM HEPES-KOH (pH 7.5) containing 6 mM Na2MoO4, 0.5 M NaSCN, 2mM CHAPS, 7% dimethylformamide, and 12% glycerol; buffer D: 10 mM H3PO4KOH (pH 7.5) containing 20 mM Na,MoO4, 2 mM MgCl2, 0.1 mM ZnCl 2 ,5% glycerol, and 10 mM thioglycerol; buffer E: 10 mM H3PO4-KOH (pH 7.5) containing 0.1% Tween20, 1.5 mM EDTA, 10% glycerol, and 10 mM thioglycerol. The pHs of all buffers were adjusted at room temperature. Purification of Calf Uterus Vero ER— Vero ER labeled with [ 3 H]Ej (sp. act., 10 Ci/mmol) was purified from calf uteri using an estradiol-linked affinity resin, as previously described (22). After the calf uterus cytosol fraction adjusted to 0.7 M KC1 had been applied to an E2-Sepharose 6B column, the column was washed once with buffer A containing 0.7 M KC1, instead of three cycles of washing with three kinds of buffers, to obtain a large amount of the preparation. Vero ER was usually eluted from the column with buffer B containing 0.3 //M [ 3 H]E 2 , and [3H]E2-i>ero ER complexes thus obtained were designated as purified vero ER. When unoccupied vero ER was necessary, vero ER was eluted from the column with buffer B in the absence of E 2 , and the fraction obtained was designated as purified unoccupied vero ER. About 4,000 and 1,000-fold purification was achieved with these methods, respectively. Purification of Hsp 90—Hsp 90 was purified from rat liver as previously described (23). Reconstitution of 9 S ER—The reconstitution procedures were carried out at 4'C. Standard reconstitution

procedures were performed as previously described (21). Purified vero ER labeled with [3H]E2 or purified unoccupied vero ER, 75 ng, and pure hsp 90, 60 fig, were mixed in 0.4 ml of buffer C. Then the mixture was dialysed against buffer D for 6 h. Glycerol Gradient Ultracentrifugation—Samples of reconstituted vero ER-hsp 90 complexes (150-500//I) were layered on the top of a 10-35% (v/v) or 15-45% (v/v) glycerol gradient (4 ml) prepared with buffer D in the presence or absence of 20 mM Na2MoO4. The gradients were centrifuged at 140,000 x g for 15 h in a HITACHI RPS 40T-2 or RPS 65T rotor and then fractionated into about 25 fractions in scintillation vials, and then [3H]E2 radioactivity was measured. Sedimentation coefficients were determined using [ u C]ovalbumin (3.6 S) and [14C]glucose oxidase (7.9 S) as external standard proteins. Chromatogrophic Analysis of Cytosolic and Reconstituted ERs—Cytosol fractions for DEAE-cellulose chromatography were prepared from a calf uterus. The tissue was homogenized in buffer A with a Polytron PT10-35. The homogenate was centrifuged at 100,000 X g for 1 h, and then the supernatant fraction was incubated with 10 nM [3H]E2(sp. act., 20 Ci/mmol) for 2 h at 4 "C. The labeled cytosol fraction thus obtained was mixed with a charcoaldextran suspension to remove free steroids. For the transformation of the cytosolic receptor, the labeled cytosol fraction was passed through a PD-10 column equilibrated with buffer E to remove Na2MoO4, and then the fraction was incubated at 37'C for 30min. A DEAE-cellulose column (0.7 X 5.0 cm) was equilibrated with 20 ml of buffer E, and then 80 mg of bovine serum albumin was loaded onto it to block irreversible binding sites. The column was washed with 8 ml of buffer E containing 0.4 M KC1, and then reequilibrated with 20 ml of buffer E. A 0.9 ml portion of the cytosol fraction, purified vero ER or reconstituted 9 S ER labeled with [3H]E2 was loaded on the column, and then the column was washed with 20 ml of buffer E. Receptors were eluted with a 0 to 0.4 M KC1 linear gradient in buffer E (total, 50 ml) at 12 ml/h. Fractions of 1.5 ml were collected in plastic tubes containing 0.2 ml of 5 mg/ml bovine serum albumin. The radioactivity of each fraction was measured with a liquid scintillation counter. Buffers containing 20 mM Na2MoO4 were used for the analysis of untransformed receptors. RESULTS Effects of Sodium Thiocyanate on the Reconstitution of Vero ER-Hsp 90 Complexes—We previously showed that 9 S ER could be reconstituted from purified [3H]E2-uero ER and hsp 90 under the standard conditions given under 'EXPERIMENTAL PROCEDURES" (21). Thus, some effective components in the medium for the 9 S ER reconstitution and the mechanism underlying the reconstitution of 9 S ER in vitro were examined in this study. When 0.5 M NaSCN was omitted from the reconstitution medium, the resulting samples contained little reconstituted 9 S ER (Fig. IB). Surprisingly, although it was previously reported that Na2MoO4 stabilizes cytosolic 9 S ER by numerous investigators (1-3), it had no effect on the reconstitution of 9 S ER (Fig. 1C). The magnesium ion, a typical divalent cation in living cells, was also not effective (Fig. ID). Wilson showed previously that zinc chloride was effective for the reconJ. Biochem.

Formation, of ER-Hsp 90 Complexes

537

stitution of the 8 S androgen receptor (24), but it was not effective in the case of our reconstitution systems (Fig. IE). The effects of the omission of CHAPS, dimethylformamide, glycerol, and thioglycerol were examined, and it was shown that the latter three reagents had essentially no effect on the reconstitution yield (data not shown). CHAPS, however, seemed to increase the recovery of the ER a little after the reconstitution procedures. While the potassium phosphate and Tris-HCl buffers could both be used as the dialysis buffer, the phosphate buffer gave a little better yield sometimes, but not significantly (data not shown). These data suggest that the most important component for the reconstitution is NaSCN. Then we investigated the effect of the concentration of NaSCN on the reconstitution of 9 S ER. The reconstitution procedures were performed at NaSCN concentrations of 0.5, 0.25, 0.1, and 0M, respectively, under the standard conditions (Fig> 2). With the decrease in the NaSCN concentration, the amount of reconstituted 9 S ER decreased and it finally disappeared in the absence of NaSCN, furthermore, the tritiated count at the bottom of the gradient (>11 S) increased. In other experiments, a little reconstituted 9 S ER was obtained in the absence of NaSCN (not shown). But, most results of all experiments were the same as shown in Fig. 2. These observations show that NaSCN is a very useful reagent for

A

8

B

A

c-

preventing aggregation and inactivation of the purified vero ER during the reconstitution procedures. Effects of the Hsp 90 Concentration on the Reconstitution of 9 S ER—The reconstitution procedures were performed with various concentrations of hsp 90, and the amounts of reconstituted 9 S ER and free vero ER were determined from the [3H]E2-counts on the sedimentation patterns (Fig. 3). The amount of reconstituted 9 S ER was saturated at about 0.7//M hsp 90. The amount of free vero ER decreased with the increase in the hsp 90 concentration, reaching a constant value of about 200fmol. However, one-third of the vero ER remained as the 5 S type at the hsp 90 concentration giving the maximum amount of reconstituted 9 S ER. This remaining 5 S type vero ER may have lost the ability to reassociate with hsp 90 due to dephosphorylation or misfolding, etc. With 0.1 //M hsp 90, only one-third of the maximum amount of 9 S ER was reconstituted, though the amount of hsp 90 was a 35-fold molar-excess as to vero ER. Therefore, the formation of 9 S ER seemed to depend on the concentration of hsp 90, rather than the molar ratio of hsp 90 and vero ER. DEAE-CeUulose Chromatography of the Reconstituted 9 S ER—The observations described above show that 9 S ER can be reconstituted from vero ER and hsp 90. We have suggested strongly that hsp 90 is included in the reconstituted 9 S ER complexes because of its sedimentation coefficient (21), the large shift of its sedimentation coefficient on treatment with antiserum against hsp 90 (21), and

fO

b6

.1

1 p

It 4

u

6

I

10

D

J

20

1

/I

A

^4

(dpn

r >


20 0 10 Fractions

20

Fig. 6. Time dependent change in the reconstitution ability of purified vero ER. The upper two panels show the sedimentation patterns of 9 S ER reconstituted from hsp 90 and affinity purified unoccupied vero ER, which was immediately labeled with 10 nM [SH]E, for 3 h at 4'C after elution from the affinity column (O), in the presence (left) and absence (right) of 0.5 M NaSCN, respectively. The lower two panels show the sedimentation patterns of 9 S ER reconstituted from hsp 90 and the purified unoccupied vero ER preparation, which was labeled with 10 nM ['H]E, at 4'C for 20 h after elution from the column (•), with (left) and without (right) 0.5 M NaSCN. The arrows indicate the positions of 9 S.

column (22, 27). In our reconstitution system, it could be expected that a ER conformation which could not reassociate with hsp 90 was changed to another one which could bind with hsp 90, as described in the above cases, through the chemical perturbation of the vero ER structure due to 0.5 M NaSCN. As can be seen in Fig. 2, according to the decrease in the NaSCN concentration in the reconstitution medium, the amount of 9 S ER reconstituted decreased and the large aggregate of ER (> 11 S and precipitate) increased. These data suggest that NaSCN also prevents non-specific aggregation of ER during the reconstitution procedures. Indeed, NaSCN is usually used at the concentration of 0.5 M for dissolving protein complexes through hydrophobic interactions in aqueous solutions (28). As described above, NaSCN is a very useful reagent for the reconstitution of 9 S ER from purified vero ER and purified hsp 90 in a high yield. In our preliminary experiments carried out so far, no difference was detected between the 9 S ERs reconstituted with and without NaSCN. Therefore, this reconstitution method involving NaSCN should be useful for investigation of the mechanisms underlying the ER system in more detail. In the target cells of estrogen, 9 S ER may be formed spontaneously from vero ER and hsp 90, as in the case of the

540

K. Inano et al.

reconstitution involving a fresh vero ER preparation in the absence of NaSCN, shown in Fig. 5, because hsp 90 is an abundant protein in the target cells (23, 29), and the concentration necessary for the reconstitution of 9 S ER in vitro is lower than the physiological concentration of hsp 90 (Fig. 4). It was shown that the translated glucocorticoid receptor in vitro immediately formed complexes with hsp 90 (30). Moreover, it was suggested in this study that the conformation of the purified vero ER tended to change, the resulting form still having E2-binding activity but not spontaneous hsp 90-binding activity. We have also additional preliminary data showing that the aged purified receptor cannot bind with the oligonucleotide of the estrogen responsive element of vitellogenin, but that fresh vero ER can bind with it. These observations suggest that the conformation of the purified vero ER, stripped of other protein(s), changes very easily, and that it loses important activities such as the responsive element binding activity, though the change cannot be detected by means of a ligand binding assay. One of the possible roles of hsp 90 in the untransformed ER may be the maintenance of the unstable active conformation of ER, which is an important molecular form for the estrogen signal transduction pathway. Further investigation on the mode of binding of vero ER and hsp 90, and the properties of the reconstituted 9 S ER are necessary to clarify the function of 9 S ER. The authors wish to thank Miss Takako Matsuzawa for her technical assistance. REFERENCES 1. Shyamala, G. & Leonard, L. (1980) J. Biol. Chem. 255, 60286031 2. Redeuilh, G., Secco, C , Baulieu, E.E., & Richard-Foy, H. (1981) J. Biol. Chem. 256, 11496-11502 3. Krozowski, Z.S. & Murphy, L.C. (1981) J. Steroid Biochem. 14, 363-366 4. Joab, I., Radanyi, C, Renoir, M., Buchou, T., Catelli, M.G., Binart, N., Mester, J., & Baulieu, E.E. (1984) Nature 308, 860853 5. Catelli, M.G., Binart, N., Jung-Testas, I., Renoir, J.M., Baulieu, E.E., Feramisco, J.R., & Welch, W.J. (1985) EMBO J. 4, 31313135 6. Nemoto, T., Ohara-Nemoto, Y., & Ota, M. (1987) J. Biochem.

102, 513-523 7. Redeuilh, G., Moncharmont, B., Secco, C, &Baulieu, E.E. (1987) J. Biol. Chem. 282, 6969-6975 8. Sanchex, E.R. (1990) J. BioL Chem. 265, 22067-22070 9. Gasc, J.M., Renoir, J.M., Faber, L.E., Delahaye, F., & Baulieu, E.E. (1990) Exp. CeU Res. 186, 362-367 10. Kost, S.L., Smith, D.F., Sullivan, W.P., Welch, W.J., & Toft, D.O. (1989) MoL CeU. Biol. 9, 3829-3838 11. Jensen, E.V., Suzuki, T., Kawashima, T., Stumpf, W.E., Jungblut, P.W., & DeSombre, E.R. (1968) Proc. Nati. Acad. Sci. USA 59, 632-638 12. Leach, K.L., Darmer, M.K., Hammond, N.D., Sando, J.J., & Pratt, W.B. (1979) J. Biol. Chem. 254, 11884-11890 13. Nishigori, H. & Toft, D.O. (1980) Biochemistry 19, 77-83 14. King, R.J.B. (1986) J. Steroid Biochem 26, 451-454 15. Picard, D., Khursheed, B., Garabedian, M.J., Fortin, M.G., Lindquest, S., & Yamamoto, K.R. (1990) Nature 348, 166-168 16. Murayama, A., Fukai, F., Hazato, T., & Yamamoto, T. (1980) J. Biochem. 88, 963-968 17. Murayama, A., Fukai, F., & Yamamoto, T. (1980) J. Biochem. 88, 969-976 18. Colvard, D.S. & Wilson, E.M. (1981) Endocrinology 109, 496504 19. Scherrer, L.C, Dalman, F.C., Massa, E., Meshinchi, S., & Pratt, W.B. (1990) J. BioL Chem. 265, 21397-21400 20. Smith, D.F., Schowalter, D.B., Kost, S.L., & Toft, D.O. (1990) MoL EndocrinoL 4, 1704-1711 21. Inano, K., Haino, M., Iwasaki, M., Ono, N., Horigome, T., & Sugano, H. (1990) FEBS Lett. 267, 157-159 22. Horigome, T., Golding, T.S., Quarmby, V.E., Lubahn, D.B., McCarty, K., Sr., & Korach, K.S. (1987) Endocrinology 121, 2099-2111 23. Iwasaki, M., Saito, H., Yamamoto, M., Korach, K.S., Horigome, T., & Sugano, H. (1989) Biochim. Biophys. Acta 992, 1-8 24. Wilson, E.M. (1985) J. BioL Chem. 260, 8683-8689 25. Sabbah, M., Redeuilh, G., & Baulieu, E.E. (1989) J. BioL Chem. 264, 2397-2400 26. Sica, V., Puca, G.A., Molinari, A.M., Buonaguro, F.M., & Bresciani, F. (1980) Biochemistry 19, 83-88 27. Lubahn, D.B., McCarty, K.S., Jr., & McCarty, K.S., Sr. (1985) J. BioL Chem. 260, 2515-2526 28. Sawyer, W.H. & Puckridge, J. (1973) J. BioL Chem 248, 84298433 29. Lai, B.T., Chin, N.W., Stanek, A.E., Keh, W., & Lansks, K.W. (1984) MoL CeU. BioL 4, 2802-2810 30. Dalman, F.C., Bresnick, E.H., Patel, P.D., Predew, G.H., Watson, S.J., Jr., & Pratt, W.B. (1989) J. BioL Chem 264, 19815-19821

J. Biochem.