THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 266. No. 36, Issue of December 25, pp. 24601”24605,1991 (0 1991 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U. S. A.

Protein Components of the Nonactivated Glucocorticoid Receptor* (Received for publication, May 28, 1991)

Martin Rexin, Willi Busch, and Ulrich GehringS From the Institut fur Biologische Chemie, Uniuersitat Heidelberg, Im Neuenheimer Feld 501, 0-6900 Heidelberg, Germany

The nonactivated glucocorticoid receptor (M, - 350,000) of WEHI-7 mouse lymphoma cells was inves- tigated with respect to the stoichiometry of protein subunits. Cross-linking patterns obtained by affinity labeling and denaturing gel electrophoresis revealed a heterotetramer consisting of one receptor polypeptide in association with two 90- and one -50-kDa subunits. The receptor stabilized by molybdate, disulfide bond formation, or chemical cross-linking was purified roughly 6000-fold by immunoaffinity chromatography and analyzed by gel electrophoresis and immunoblot- ting. The 90-kDa component was consistently detected in a 2:l ratio with respect to the receptor polypeptide and was identified as the 90-kDa heat shock protein, hsp9O. A 70-kDa heat shock protein was found in both stabilized and nonstabilized receptors and bound to the immunomatrix independent of receptor. The addi- tional receptor subunit was unequivocally identified as the 69-kDa protein previously described (Tai, P.-K. K., Maeda, Y., Nakao, K., Wakim, N. G., Duhring, J. L., and Faber, L. E. (1986) Biochemistry 25, 5269- 5275). This component was found only in complexes cross-linked via amino groups. It was removed from the molybdate-stabilized receptor under our purifica- tion conditions, thus leaving behind a trimer composed of the receptor polypeptide and two molecules of hsp90. In the absence of hormone, the receptor had the same subunit composition as in its presence.

The intracellular receptor specific for glucocorticoid hor- mones (for reviews, see Refs. 1-8) is known to exist in at least two distinct molecular forms. One is of very high molecular mass, roughly 350 kDa, and is a hetero-oligomer. It binds the hormonal ligand but is unable to interact with DNA. The other receptor form is much smaller in size and can bind to specific regulatory DNA sequences, so-called glucocorticoid response elements.

Our interest concentrates on the subunit structure of the high molecular weight glucocorticoid receptor. It contains a single hormone-binding polypeptide (9, lo), the receptor in the strict sense, cloned some years ago from several mam- malian species (11). This receptor polypeptide of M , - 100,000 is associated with other nonsteroid binding proteins. In par-

* This investigation was supported by a grant from the Deutsche Forschungsgemeinschaft and by Fonds der Chemischen Industrie. 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.

562514. $TO whom correspondence should be addressed. Tel.: 06221-

ticular, hsp90l has been identified by several laboratories as a receptor component (1, 7, 12-17), and it is known to be present in two copies. In fact, the receptor of rat hepatocytes has been proposed to be a trimer of two hsp90 molecules and one receptor polypeptide (18, 19). On the other hand, we previously obtained evidence for the receptor of lymphoid cells to be a tetrameric structure containing an additional component of roughly 50 kDa (20). Several reports suggest the association with steroid hormone receptors of two com- pletely independent proteins: a protein of 59 kDa (21-23) and the heat shock protein hsp70 (24-26). We set out to identify the missing subunit of our tetrameric structure and indeed found the 59-kDa protein by cross-linking and extensive purification to be an integral component of the glucocorticoid holoreceptor. By contrast, hsp70 turned out to be a very persistent contaminant of our receptor-specific immunopuri- fication procedure.

EXPERIMENTAL PROCEDURES A N D RESULTS’

components of the Cross-linked Holoreceptor-In order to avoid subunit dissociation of the high molecular weight recep- tor affinity-labeled with [3H]dexamethasone mesylate, we used cross-linking with EGS (16, 31), which is resistant to reductive cleavage. Purification was in two steps: immunoaf- finity chromatography with the receptor-specific monoclonal antibody mab49 followed by continuous SDS-PAGE. The radiolabeled material of molecular mass 300-400 kDa was electroeluted, and the chemical cross-links were subsequently cleaved with hydroxylamine. Analysis was again by standard gel electrophoresis followed by immunoblotting (Fig. 4, lanes 1 ), in comparison with an identical sample in which cleavage with hydroxylamine was carried out right after the immuno- purification, thus omitting the first gel electrophoresis (lanes 2). Although hsp70 was easily detectable in this sample by staining with antibody against hsp70 (panel A ) , it was missing from the sample that had been further purified by SDS- PAGE. In order to show that both samples do contain the receptor polypeptide in equivalent amounts, the same blot was stained with the receptor-specific antibody mab49 (panel B). Thereafter, the membrane was reacted for a third time with antiserum against hsp90 (panel C), and hsp9O was also detected in equal amounts in both samples. This result une-

The abbreviations and trivial names used are: hsp90 and hsp70, heat shock proteins of M, 90,000 and 70,000; dexamethasone mesyl- ate, 9a-fluoro-l6cy-methyl-ll~,17a,21-trihydroxy-1,4-pregnadiene- 3,20-dione-21-mesylate; DMS, dimethyl suberimidate; DSP, di- thiobis(succinimidy1propionate); EGS, ethylene glycolbis(succini- midylsuccinate); PAGE, polyacrylamide gel electrophoresis; p59, pro- tein of M. 59,000; SDS, sodium dodecyl sulfate.

Portions of this paper (including “Experimental Procedures,” part of “Results,” and Figs. 1-3) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

24601

24602 Subunits of the Nonactivated Glucocorticoid Receptor

quivocally proves that hsp70 is not a component of the EGS- stabilized holoreceptor.

The blots of the EGS cross-linked receptor samples were also stained with an antiserum directed against the 59-kDa protein (21). Both samples gave roughly equal signals (Fig. 4, panel D), proving that p59 is a component of the cross-linked receptor.

We also attempted to detect p59 in receptor-steroid com- plexes stabilized either by sodium molybdate, by oxidation with cupric o-phenanthroline, or by cross-linking with DSP (Fig. 5). Upon silver staining, we indeed observed a band a t about 60 kDa in the DSP cross-linked receptor (panel A, lane 4 ) , and the immunoblot with the specific antiserum yielded a strong signal (panel B, lane 4 ) . By contrast, we obtained no p59-specific signal with the molybdate-stabilized receptor and only a very faint band in the material treated with cupric phenanthroline (Fig. 5B, lanes 2 and 3 ) . This suggests that p59 is less tightly integrated in the receptor complex than the

A B C

97000- .receptor - hsp90

66700- - h ~ p 7 0

1 z 1 2 1 2

0

60000- * .? 1 L

FIG. 4. Analysis of EGS cross-linked receptor. Cytosolic receptor complexed with ["H]dexamethasone mesylate was cross- linked with EGS and submitted to immunopurification, and the sample was divided into two. One portion was submitted to continuous SDS-PAGE for 2 h a t 30 mA. The molecular weight range from 300 t o 400 kDa was cut out, and the protein was electroeluted, treated with hydroxylamine, and run on a discontinuous SDS gel (lanes I ) with markers as in Fig. 2. In the other portion (lanes 2 ) the electro- phoretic purification prior to cleavage with hydroxylamine was omit- ted. Following blotting, the portion of the membrane corresponding to molecular masses above 65 kDa was cut out and immunostained in three stages with monoclonal antibody N27 against hsp7O (panel A ) , followed by the receptor-specific monoclonal antibody mab49 (panel B ) , followed by antiserum against hsp90 (panel C ) . The area of 50-65 kDa was stained with antiserum JP-1 against p59 (panel Dl.

hsp90 subunits and is removed from the immunoadsorbed receptor by washing with 0.2% Triton X-100 and 600 mM KC1 in the presence of molybdate. Indeed, when we carried out cross-linking with DMS under such solvent conditions, we obtained a subunit pattern that is in contrast to what we see with the holoreceptor (cf. Fig. 1,panels B-D). This pattern shows a trimeric structure (Fig. 6) with only one intermediate form of M , - 187,000 (receptor + one hsp90) in addition to the fully cross-linked material of M , - 284,000 (receptor + two hsp90). These molecular weights are the averages of three independent experiments.

For the above experiments, we routinely used the receptor complexed with steroid. We now wished to find out whether the unliganded receptor structure has the same subunit com- position. We therefore cross-linked the receptor in the ab- sence of steroid with DSP and carried out a similar analysis. This resulted in a protein pattern comparable with that obtained in the presence of ligand (Fig. 5A, lane 5). The receptor polypeptide, hsp90, and p59 were detected at the expected intensities. The p59 also showed up in the respective immunoblot (Fig. 5B, lane 5 ) .

DISCUSSION

In this, as well as in a previous study (20), we demonstrate that the nonactivated glucocorticoid receptor of lymphatic cells is a heterotetramer composed of one receptor polypeptide associated with two molecules of hsp90 and a protein subunit of roughly 50 kDa. As the molecular mass of this latter component was determined by a subtractive method, it has to be viewed as fairly imprecise. Our present work is concerned with the identification of this novel receptor subunit. A major problem was to find out whether or not this is identical with hsp7O. Indeed, we found hsp70 in association with the receptor even after a roughly 6000-fold purification using immunoaf- finity chromatography. However, hsp7O similarly copurified with the activated receptor (Fig. 2), which is of a much lower molecular mass (30) that excludes a dimer of the receptor polypeptide with a 70-kDa component.

ATP is known to affect the interaction of hsp7O with other proteins and was found to cause the release of hsp7O from the progesterone receptor complex (24). Similarly, we observed that hsp70 was displaced by ATP from the glucocorticoid

39 000 - 37000 -

29000 -

1 2 3 L 5

B 60000- - p59

1 2 3 L 5 FIG. 5. Protein patterns of stabilized receptors. Cytosolic

receptor was either complexed with dexamethasone mesylate (lanes I to 4 ) or left unliganded (lanes 5) . Samples were either untreated (lane I ) , kept in 20 mM sodium molybdate (lanes 2) , or treated with Cu'+/o-phenanthroline (lanes 3 ) or with DSP (lanes 4 and 5). Im- munopurification and discontinuous SDS-PAGE were carried out as described with markers as in Fig. 2. In panel A, detection was by silver staining. Panel R shows immunostaining with antiserum JP-1 against p59 of the 50-65 kDa portion of a corresponding blot.

I I I I ;o LO Qo BO 100 120

I

Slice Number

FIG. 6. Release of p59 from the holoreceptor. Cytosolic recep- tor complexed with ["H]dexamethasone mesylate (radiolabel not di- luted) was incubated with 300 mM KC1 for 1 h in the presence of 20 mM molybdate, diluted 5-fold with buffer containing 0.2% Triton X- 100,600 mM KCI, 20 mM molybdate, and left in the cold for another 60 min prior to cross-linking with DMS for 90 min. Immunopurifi- cation and SDS-PAGE were as in Fig. 1.

Subunits of the Nonactivated GEucocorticoid Receptor 24603

receptor even after chemically stabilizing the complex (Fig. 3). Moreover, the subunit pattern obtained by progressive cross-linking was not changed upon prior treatment with ATP. These, as well as several additional observations, strongly suggest that hsp70 is not specifically associated with the receptor but rather binds to the IgGl immunomatrix. Most significantly, extensive purification of the chemically cross-linked holoreceptor by use of SDS-PAGE resulted in complete removal of hsp70, whereas hsp90 was found to be fully retained (Fig. 4). Similarly, the nonactivated glucocor- ticoid receptor of mouse L cells was recently found to be devoid of hsp7O (26).

We now provide unequivocal evidence for a 59-kDa protein being the fourth component of the tetrameric holoreceptor. When we used the bifunctional N-hydroxysuccinimidylesters DSP and EGS, which react with amino groups of lysine side chains, we detected p59 as a constituent of the highly purified complex. p59 has recently been described to be associated with several steroid hormone receptors (21-23). In particular, when the progesterone receptor from rabbit uterus was sub- mitted to chemical cross-linking, it was stabilized in the high molecular weight form, which contained p59 (22). The stoi- chiometry of the progesterone receptor subunits, however, has not been determined directly so far.

A hierarchical order of subunit arrangement within the glucocorticoid holoreceptor is suggested by our progressive cross-linking experiments. Although p59 is rarely linked to the receptor polypeptide itself (Fig. 1, peak b ) , its association with one hsp90 subunit plus receptor (peak d ) is quite prom- inent. This suggests that in the native structure, it is not in close proximity to the receptor polypeptide but rather asso- ciated with one of the hsp90 subunits. Accordingly, we found that p59 was released from the molybdate-stabilized holore- ceptor by mild detergent in the presence of high salt. Chemical cross-linking under such conditions disclosed a trimeric struc- ture with just two 90-kDa subunits associated with the recep- tor polypeptide (Fig. 6). It thus appears that receptor stabili- zation by oxianions such as molybdate predominantly affects the hsp90 subunits and their association with the receptor. We also conclude from our observations in Fig. 5 that p59 does not participate in receptor cross-linking via disulfide bridges, whereas cross-linking via amino groups is possible and may be facilitated by the fact that the cross-lickers provide a fairly long bridging molecule of more than 12 A.

The p59 protein exists in several isoforms, most notably in a smaller version of 56 kDa (23). This has been found in cell extracts as part of a larger complex with both hsp90 and hsp70 (23, 38, 39) and it may itself be a heat shock protein (38). However, it remains unknown how the biological effects of heat shock or chemical stress on cells relate to the physi- ology of steroid hormone action. Hsp9O at least is expressed in eukaryotic cells independent of stress situations (40, 41).

Highly purified progesterone and estrogen receptors have been reported to be associated with several proteins of molec- ular masses of about 23, 38, 50, and 54 kDa (25, 42). We also observed a series of accessory protein bands (cf. Figs. 2 and 5) which, however, are not bona fide receptor components but rather persistent contaminants. In several early experiments, we detected a prominent 37 kDa band that we identified as the subunit of glyceraldehyde-3-phosphate dehydrogenase, but upon improvement of our immunopurification procedure this disappeared. Similarly, we did not consistently see a 23- kDa component. The 56- and 58-kDa proteins that we rou- tinely obtained as contaminants of our immunopurification are most likely identical with a- and @-tubulins.

The vast majority of biochemical studies on steroid hor-

mone receptors have been carried out with receptor-hormone complexes. We now also investigated the unliganded form of the high molecular weight glucocorticoid receptor and show for the first time that it likewise contains p59 in addition to hsp90 (Fig. 5, lunes 5 ) . It now becomes apparent that the biological function of this large structure lies in the ability to bind the steroid. Loss of hormone-binding activity was found to occur in parallel with dissociation of hsp90 from the ligand- free receptor (43). Similarly, the receptor dissociated into subunits was unable to rebind steroid once it had lost the ligand (44). When glucocorticoid and estrogen receptor con- structs were transfected into yeast, the respective receptor polypeptides were found to associate with hsp90 (45). In this system, it was possible to significantly reduce the intracellular level of the heat shock protein, which resulted in impaired transcriptional enhancement by hormone. This clearly points to a physiological role of hsp90 in steroid receptor activity. The specific function of p59, however, remains unknown. It appears likely that p59 acts in concert with the hsp90 subunits of the holoreceptor in facilitating hormone binding and sub- sequent receptor activation.

Acknowledgments-We are very grateful to Dr. Lee E. Faber for providing the goat antiserum JP-1 directed against rabbit p59. We would like to thank Dr. G. Moldenhauer for a mouse monoclonal anti-idiotype antibody (IgG1) against human IgG.

1. 2. 3.

4. 5. 7. 8.

9. 10.

11. 12.

13.

14.

15. 16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

REFERENCES Pratt, W. B. (1987) J. Cell. Biochem. 35 , 51-68 Gehring, U. (1987) Trends Biochem. Sci. 12, 399-402 Gehring, U. (1988) in Hormones and Their Actions (Cooke, B. A.,

King, R. J. B., and van der Molen, H. J., eds) Part I, pp. 217- 239, Elsevier, Amsterdam

Litwack, G. (1988) Cancer Res. 48, 2636-2640 Beato, M. (1989) Cell 5 6 , 335-344 Pratt, W. B. (1990) Mol. Cell. Endocrinol. 74, C69-C76 Danielsen, M. (1991) in Nuclear Hormone Receptors, Molecular

Mechanisms, Cellular Functions, Clinical Abnormalities (Par- ker, M. G., ed) pp. 39-76, Academic Press, London

Gehring, U., and Arndt, H. (1985) FEBS LettO179, 138-142 Okret, S., Wikstrom, A.-C., and Gustafsson J.-A. (1985) Biochem-

Evans, R. M. (1988) Science 240 , 889-895 Catelli, M. G., Binart, N., Jung-Testas, I., Renoir, J. M., Baulieu,

E.-E., Feramisco, J . R., and Welch, W. J. (1985) EMBO J. 4 ,

Toft, D. O., Sullivan, W. P., McCormick, D. J., and Riehl, R. M.

Howard, K. J., and Distelhorst, C. W. (1988) J. Biol. Chem. 2 6 3 ,

Mendel, D. B., and Orti, E. (1988) J. Biol. Chem. 263,6695-6702 Rexin, M., Busch, W., and Gehring, U. (1988) Biochemistry 2 7 ,

5593-5601 Denis, M., and Gustafsson, J.-A. (1989) Cancer Res. 4 9 , 2275s-

2281s Denis, M., Wikstrom, A.-C., and Gustafsson, J.-A. (1988) J.

Steroid Biochem. 3 0 , 271-276 Lefebvre, P., Sablonniere, B., Tbarka, N., Formstecher, P., and

Dautrevaux, M. (1989) Biochem. Biophys. Res. Commun. 159 ,

Rexin, M., Busch, W., Segnitz, B., and Gehring, U. (1988) FEBS

Tai, P.-K. K., Maeda, Y . , Nakao, K., Wakim, N. G., Duhring, J.

Renoir, J.-M., Radanyi, C., Faber, L. E., andBaulieu, E.-E. (1990)

Sanchez, E. R., Faber, L. E., Henzel, W. J., and Pratt, W. B.

Kost, S. L., Smith, D. F., Sullivan, W. P., Welch, W. J., and Toft,

Smith, D. F., Faber, L. E., and Toft, D. 0. (1990) J. Biol. Chem.

Sanchez, E. R., Hirst, M., Scherrer, L. C., Tang, H.-Y., Welsh, M. J., Harmon, J. M., Simons, S. S., Jr., Ringold, G. M., and

istry 24,6581-6586

3131-3135

(1987) Biochem. Actions Horm. 14,293-316

3474-3481

677-686

Lett. 241 , 234-238

L., and Faber, L. E. (1986) Biochemistry 25 , 5269-5275

J. Biol. Chem. 2 6 5 , 10740-10745

(1990) Biochemistry 2 9 , 5145-5152

D. 0. (1989) Mol. Cell. Biol. 9 , 3829-3838

265,3996-4003

24604 Subunits of the Nonactivated Glucocorticoid Receptor

27.

28. 29.

30.

31.

32.

33. 34.

35.

Pratt, W. B. (1990) J. Biol. Chem. 265, 20123-20130

(1973) J. Immunol. 110,431-437 Harris, A. W., Blankhurst, A. D., Mason, S., and Warner, N. L.

Bourgeois, S., and Newby, R. F. (1977) Cell 11, 423-430 Gehring, U., Mugele, K., and Ulrich, J. (1984) Mol. Cell. Endo-

Gehring, U., Mugele, K., Arndt, H., and Busch, W. (1987) Mol.

Abdella, P. M., Smith, P. K., and Royer, G. P. (1979) Biochem.

Westphal, H. M., Moldenhauer, G., and Beato, M. (1982) EMBO

Laemmli, U. K. (1970) Nature 227, 680-685 Oakley, B. R., Kirsch, D. R., and Morris, N. R. (1980) Anal.

Hunkapiller, M. W., Lujan, E., Ostrander, F., and Hood, L. E.

crinol. 36, 107-113

Cell. Endocrinol. 5 3 , 33-44

Biophys. Res. Commun. 87,734-742

J. 1, 1467-1471

Biochem. 105, 361-363

(1983) Methods Enzymol. 91,227-236

36. Schagger, H., Aquila, H., and von Jagow, G. (1988) Anal. Biochem. 173,201-205

37. Housley, P. R., Grippo, J. F., Dahmer, M. K., and Pratt, W. B.

38. Sanchez, E. R. (1990) J. Biol. Chem. 265,22067-22070 39. Perdew, G. H., and Whitelaw, M. L. (1991) J. Biol. Chem. 266,

40. Pelham, H. R. B. (1986) Cell 46,959-961 41. Schlesinger, M. J. (1990) J. Biol. Chem. 2 6 5 , 12111-12114 42. Ratajczak, T., Hlaing, J., Brockway, M. J., and Hahnel, R. (1990)

J . Steroid Biochem. 35,543-553 43. Bresnick, E. H., Dalman, F. C., Sanchez, E. R., and Pratt, W. B.

(1989) J. Biol. Chem. 264,4992-4997 44. Segnitz, B., and Gehring, U. (1990) J. Biol. Chem. 265 , 2789-

2796 45. Picard, D., Khursheed, B., Garabedian, M. J., Fortin, M. G.,

Lindquist, S., and Yamamoto, K. R. (1990) Nature 348, 166- 168

(1984) Biochem. Actions Horm. 11,347-376

6708-6713

IIXPBHIWENTAL PA0C:EDURES

Subunits of the Nonactivated Glucocorticoid Receptor

F i r . I

A

39000- 37000-

nw0 - ._ . 1 2 . L

- receptor

kz - hsp70

1 2 3 4 5 6 7 8 9