Heat shock protein 90 and the nuclear transport of progesterone receptor

Abstract

Steroid receptors exist as large oligomeric complexes in hypotonic cell extracts. In the present work, we studied the nuclear transport of the 2 major components of the oligomeric complex, the receptor itself and the heat shock protein 90 (Hsp90), by using different in vitro transport systems: digitonin permeabilized cells and purified nuclei. We demonstrate that the stabilized oligomeric complex of progesterone receptor (PR) cannot be transported into the nucleus and that unliganded PR salt dissociated from Hsp90 is transported into the nucleus. When nonstabilized PR oligomer was introduced into the nuclear transport system, the complex dissociated and the PR but not the Hsp90 was transported into the nucleus. If PR exists as an oligomeric form after synthesis, as suggested by the experiments with reticulocyte lysate, the present results suggest that the complex is short-lived and is dissociated before or during nuclear transport. Thus, the role of Hsp90 in PR action is likely to reside in the Hsp90-assisted chaperoning process of PR preceding nuclear transport of the receptor.

INTRODUCTION

Steroid receptors are transcription factors whose function is regulated by ligand binding. Recently, a number of receptor-interacting proteins (coactivators and corepressors) have been identified that are proposed to mediate transcription regulation. Ligand binding regulates their association and dissociation by altering the structure of the receptors and thus affecting the surface properties of the receptor (Renaud et al 1995; Wurtz et al 1996; Moras and Gronemeyer 1998; Torchia et al 1998). In addition to these cofactors, a number of proteins have been shown to associate with receptors in hypotonic cell extracts and in the reticulocyte lysate, forming an oligomeric complex (Dougherty et al 1984; Kost et al 1989; Smith et al 1990; Johnson et al 1994; Catelli et al 1985). The function of the proteins associated with the oligomeric form of the steroid receptors (receptor-associated proteins or RAPs) has been debated (Tuohimaa et al 1993; Pratt and Toft 1997; Ylikomi et al 1998). The first RAP identified, heat shock protein 90 (Hsp90) (Dougherty et al 1984), has been proposed to function as a repressor since its association prevented DNA binding (Kost et al 1989; Onate et al 1991). Function as a repressor was further supported by the notion that deletion of the ligand binding domain (LBD) generates constitutively active receptors that do not form stable oligomeric complex with Hsp90 in vitro, suggesting that the interaction with Hsp90 is responsible for the repressor function of nonliganded LBDs (Scherrer et al 1993). Recent observations do not, however, support this conclusion, since it has been shown that a nonliganded LBD of estrogen receptor (ER) represses the activity of constitutive transactivator VP16-GAL but does not interact with Hsp90 in vitro, and some constitutively active ER mutants interact with Hsp90 in vitro (Lee et al 1996; White et al 1997).

Most RAPs function as chaperones. Experiments with Hsp90-deficient yeast strains have suggested that Hsp90 is required for the activity of steroid receptors (Picard et al 1990; Bohen and Yamamoto 1993). Also, ligand binding to glucocorticoid receptor (GR) is compromised when Hsp90 is dissociated from the receptor complex (Bresnick et al 1989). In both cases, chaperoning could account for these observations. The most direct evidence that RAPs can chaperone steroid receptors comes from the experiments that show that Hsp90 could restore DNA binding of partially denatured ER in vitro (Inano et al 1994). However, intact Hsp90 is required also for the activity of the retinoid receptors, which do not form a complex with Hsp90 in vitro, indicating that oligomeric complex formation in vitro does not correlate with biological activity of Hsp90 in vivo (Dalman et al 1991; Holley and Yamamoto 1995).

Most nuclear proteins enter the nucleus through an energy-requiring process that involves recognition of specific sequences known as nuclear localization signals (NLS) (Richardson et al 1988). The nuclear localization of steroid receptors is composed of multiple protosignals whose cooperation is required for nuclear targeting (Ylikomi et al 1992). Steroid receptors, as with most nuclear proteins, shuttle between cytoplasm and nucleus (Goldstein 1958; Guiochon-Mantel et al 1991, 1994; Perrot-Applanat et al 1992). The biological significance of the nucleocytoplasmic shuttling of nuclear receptors is not clear. It is possible that the cytoplasmic receptor may have a distinct biological function, as proposed by Verdi and Campagnoni (1990) and Migliaccio et al (1996), or repeated chaperoning in the cytoplasm might stabilize some structurally unstable nuclear proteins. The mechanism of nucleocytoplasmic shuttling of steroid receptors appears to use mechanisms different from those previously described, in that it is not nuclear export signal mediated but probably involves the NLS (Tyagi et al 1998; Weis 1998).

There is some evidence that RAPs are involved in nuclear translocation of nuclear proteins. Intracellular injection of antibodies against a negatively charged sequence of rabbit Hsp56 that is electrostatically complementary to a prototypic NLS-impaired dexamethasone-induced nuclear transport of GR (Czar et al 1995). Depleting Hsp70 from the cytosol inhibits nuclear transport of SV40 large T antigen but not GR, indicating Hsp70 specificity in transport of different nuclear proteins (Imamoto et al 1992; Yang and DeFranco 1994). The mechanism by which Hsp70 assists nuclear transport is not known. It has been proposed that Hsp70 induces structural alterations required for the nuclear transport of certain proteins (Jeoung et al 1991). There is no evidence that Hsp90 is directly involved in the nuclear transport. Interestingly, however, molybdate-stabilized oligomeric GR or PR cannot be transported into the nucleus (Yang and DeFranco 1996). In the present work, we studied further the involvement of Hsp90 in the nuclear transport of PR.

MATERIALS AND METHODS

Animals and hormone treatments

Two-week-old white Leghorn chicks were used. They were treated with 17β-estradiol (Sigma) dissolved in propylenglycol, 1 mg/d per animal, for 1 week followed by a withdrawal period for another week. After the withdrawal period, animals received hormone injections for 4 days and they were killed after 24 hours of last injection by cervical dislocation. Magnum parts of oviducts were dissected immediately and stored in liquid nitrogen.

Sample preparation

Cytosolic extract of oviduct was prepared as previously described with the following modifications (Pekki et al 1995). Oviduct samples were homogenized in 3 volumes of in vitro transport buffer (20 mM N-2-hydroxyethylpiperazine-N′-2-ethane-sulfonic acid [pH 7.3], 110 mM potassium acetate, 5 mM sodium acetate, 2 mM magnesium acetate, 1 mM ethylene glycol-bis [β-amino-ethyl ether]–N,N,N′,N′-tetraacetic acid, 2 mM dithiotreitol), both containing a mixture of protease inhibitors (described in Sullivan et al 1988). The cytosolic extracts were either used as such (untreated cytosol) or treated with sodium molybdate (0.02 M) or potassium chloride (KCl) (0.3 M) for 2 hours at 4°C. The KCl-treated cytosols were dialyzed overnight at 4°C against transport buffer.

In vitro assay of nuclear transport with permeabilized HeLa cells

In vitro nuclear transport of progesterone receptor was carried out as described (Adam et al 1990; Yang and DeFranco 1994) with the following modifications. HeLa cells were grown on glass coverslips (coated with poly-L-lysine) in Dulbecco modified Eagle medium/F12 medium with 5% fetal bovine serum for 24 to 48 hours. The final transport mixture contained 50% (vol/vol) either untreated or KCl-treated cytosol, 0.02 M sodium molybdate, 1 mM adenosine triphosphate (ATP), 5 mM creatine phosphate, and 20 U/mL of creatine phosphokinase diluted in transport buffer. In some experiments, progesterone was added as a final concentration of 1 μM to the transport mixture before incubation at 30°C.

Immunohistochemistry

After incubation with transport mixture, the cells were fixed with −20°C methanol at room temperature for 10 minutes and washed twice with ice-cold phosphate-buffered saline. The cells were processed for immunohistochemical staining using the avidin-biotin complex method (Ylikomi et al 1992). Monoclonal PR antibody PR22 (Sullivan et al 1986) and monoclonal Hsp90 antibody 7Dα (Sullivan et al 1985) were used at final concentrations of 1 μg/mL.

In vitro assay of nuclear transport with isolated nuclei of HeLa cells

HeLa cells were grown to confluence and nuclei were prepared as described (Greenberg and Ziff 1984). The nuclei were either used immediately or stored in liquid nitrogen in nuclear storage buffer (50 mM Tris hydrochloride, pH 8.3, 40% glycerol, 5 mM magnesium chloride, 0.1 mM ethylenediamine-tetraacetic acid).

The isolated nuclei (7 × 10⁵ to 1 × 10⁶ nuclei/reaction) were incubated at 30°C with transport mixture (25–35% [vol/vol] cytosol, 5 mM ATP, 5 mM creatine phosphate, and 20 U/mL of creatine phosphokinase diluted in transport buffer). After the transport reaction, the nuclei were harvested by centrifugation and the nuclear membranes were washed in the presence of 0.1% Triton X-100. The nuclear fractions and respective supernatants were resolved in 7.5% sodium dodecyl sulfate–polyacrylamide gel. Immunoblotting was carried out as described (Pekki et al 1995). The final concentration of both primary antibodies PR22 and 7Dα was 1 μg/mL.

RESULTS

Dissociation of the oligomeric complex is prerequisite for nuclear transport of PR

We have studied nuclear uptake of chicken PR and Hsp90 in vitro by digitonin permeabilization technique, which allows the exchange of an endogenous cytosol with an exogenous one (Adam et al 1990). We permeabilized HeLa cells and exchanged the cytosol with chicken oviduct cytosol and studied the nuclear transport of PR using antibody (PR22) recognizing chicken PR but not corresponding mammalian protein. When a cytosol containing a liganded PR was incubated with permeabilized cells, PR was seen to accumulate into the nucleus (Fig 1A). Molybdate is known to stabilize the oligomeric complex, and this stabilized complex resists both ligand- and elevated temperature–induced dissociation (Nishigori and Toft 1980). When molybdate-stabilized PR was allowed to bind ligand and then incubated with permeabilized cells, no nuclear accumulation could be detected (Fig 1B). When molybdate was added after ligand occupation, PR was seen to accumulate into the nucleus, indicating that molybdate per se cannot interfere with nuclear transport (Fig 1C). When the transport reaction was not supplemented with ATP, nuclear transport of PR was substantially inhibited (Fig 1D). These results indicate that the PR in its oligomeric complex with Hsp90 cannot be transported into the nucleus.

Fig 1.

 Stabilization of oligomeric complex inhibits nuclear import of liganded PR. Permeabilized HeLa cells were incubated with complete transport mixture, including 50% (vol/vol) oviduct cytosol at 30°C for 30 minutes with 1 μM progesterone (A) and with 1 μM progesterone and 20 mM sodium molybdate (B). (C) Transport mixture was first preincubated for 10 minutes with 1 μM progesterone after which 20 mM sodium molybdate was added and the mixture was incubated at 30°C for 30 minutes with permeabilized HeLa cells. (D) Transport mixture as in A but without added ATP and ATP regeneration system. PR was detected by immunoperoxidase technique using monoclonal antibody PR22

Dissociation of the oligomeric complex takes place before or during nuclear transport

We then studied nuclear uptake of the nonstabilized PR complex using in vitro nuclear transport assay with purified HeLa cell nuclei. Cytosol fraction (without added molybdate) was incubated with purified nuclei in the presence of an ATP regeneration system at 30°C for 10–30 minutes after which nuclear and cytosolic fractions were separated. Both fractions were run on the sodium dodecyl sulfate–polyacrylamide gels and immunoblotted with PR22 and 7Dα. As with digitonin permeabilized cells, PR was seen being rapidly transported into the nucleus, whereas Hsp90 remained in the cytosol (Fig 2). Depending on the strength of Triton X-100 wash, a signal of varying intensities with the Hsp90 antibody was detected in the nuclear fraction, indicating binding of Hsp90 to the membrane components of the purified nuclear fraction. This result indicates also that if the oligomeric complex exists in the cytoplasm it dissociates before or during nuclear transport.

Fig 2.

 Dissociation of the oligomeric complex takes place before or during nuclear transport. Purified HeLa cell nuclei were incubated with transport mixture, including 35% (vol/vol) untreated oviduct cytosol and ATP and ATP regeneration system for 10 minutes (lane 1), 20 minutes (lane 2), and 30 minutes (lane 3) at 30°C or at 4°C (lane 4). After incubation nuclei were isolated by centrifugation and washed with presence of 0.1% Triton X-100. The resulting nuclear fraction was analyzed by immunoblotting using monoclonal antibody PR22 for PR detection (A) and monoclonal antibody 7Dα for Hsp90 detection (B)

PR salt dissociated from Hsp90 is transported to the nucleus

We also studied nuclear transport of salt-dissociated PR and effect of ligand binding on that. Similarly, as with nonstabilized oligomeric complex, nonliganded receptor salt dissociated from Hsp90 accumulated in nuclei, whereas Hsp90 remained in the cytosol (Fig 3 A,B, lane 1). Nuclear uptake of PR was not markedly affected by ligand binding (Fig 3A, lane 2), but molybdate stabilization of the oligomeric complex inhibited the nuclear accumulation of PR (Fig 3A, lane 4). When the nuclei were incubated on ice in the presence of apyrase, no nuclear accumulation of PR was seen, indicating ATP dependence of the transport (Fig 3A, lane 5).

Fig 3.

 PR predissociated from Hsp90 is transported to the nucleus. Purified HeLa cell nuclei were incubated with transport mixture, including 25% (vol/vol) salt-treated oviduct cytosol and ATP and ATP regeneration system for 20 minutes at 30°C with 1 μM R5020 (lane 1) or without added ligand (lane 2). Controls (lanes 3–5) included transport mixture with no cytosol added (lane 3), transport mixture containing untreated cytosol supplemented with 20 mM sodium molybdate (lane 4), and transport mixture supplemented with apyrase (0.1 U/μL) instead of ATP and ATP regeneration system with incubation on ice (lane 5). After incubation nuclei were isolated by centrifugation and washed with presence of 0.1% Triton X-100. The resulting nuclear fraction was analyzed by immunoblotting using monoclonal antibody PR22 for PR detection (A) and monoclonal antibody 7Dα for Hsp90 detection (B)

DISCUSSION

When purified steroid receptors are incubated in the reticulocyte lysate or when they are in vitro translated, the oligomeric complex is assembled through a series of intermediate complexes. So-called mature complex is composed of the receptor and 2 Hsp90 molecules, p23 and an immunophilin (Smith 1993). This mature complex resembles that seen in the cell extracts and is thought to be a poised form of a receptor ready to bind and to be activated by the ligand (Bohen and Yamamoto 1993). In the reticulocyte lysate, mature complex dissociates with a half-life of 5 minutes, but under steady-state conditions most of the PR exists as the mature complex (Smith 1993). We showed that when the nonstabilized oligomeric form of the receptor was introduced into the nuclear transport system, PR but not Hsp90 was transported into the nucleus and also that nonliganded PR, dissociated from Hsp90 by salt treatment, can enter into the nucleus. These results suggest that if PR forms an oligomeric complex in the cytoplasm after synthesis and during nucleocytoplasmic shuttling, the oligomeric complex is a short-lived intermediate, dissociating quickly before or during the nuclear transport.

Our previous immunohistochemical studies, performed with antibodies that distinguish the oligomeric and dissociated forms of the receptors, are also in concert with this interpretation (Passinen et al 1999). It has been demonstrated that recombinant Hsp90, artificially targeted to the nucleus by a heterologous NLS, cotransported a fraction of cytoplasmic steroid receptor mutants into the nucleus (Kang et al 1994; Passinen et al 1999). Kang et al (1994) reasoned that if NLS-Hsp90 could transport the cytoplasmic receptors into the nucleus, intact steroid receptors should reciprocally cotransport cytoplasmic Hsp90 to the nucleus. However, it had already been demonstrated that this is not the case, since overexpressed wild-type PR was incapable of significantly altering the cytoplasmic location of wild-type Hsp90 (Tuohimaa et al 1993). When the oligomeric complex of the PR was stabilized with molybdate, no nuclear accumulation could be detected. A similar phenomenon has also been demonstrated with intact cells when molybdate has been introduced into cells by liposomes (Yang and DeFranco 1996). Molybdate is known to induce a large conformation change of the Hsp90 and alter its interaction with substrates by directly binding to Hsp90 (Soti et al 1998; Hartson et al 1999). This explains why it does not block nuclear transport of PR when added after ligand occupation. It is not known which transport factors are involved in the nuclear transport of steroid receptors, but Hsp90 might function to inhibit nuclear transport by masking the PR NLSs. This is supported by the notion that the D region of the PR containing the constitutive NLSs is not accessible to the antibodies in the oligomeric form (Weigel et al 1989; Pekki et al 1995). The results suggest that Hsp90 is not required for the nuclear transport per se, but dissociation of the complex is required for the transport. The interacting partners of nonliganded steroid receptors in the nucleus are not known. Interestingly, PR has been shown to interact with chromatin constituents such as the high-mobility group chromatin protein 1 and 2, which might complement Hsp90 function in keeping the receptors in a poised conformation (Boonyaratanakornkit et al 1998).

In cell extracts all of the PR is associated with the Hsp90, which is in a direct contradiction with the present results and questions the origin of the oligomeric complex: Is it present in vivo in intact cells or is it formed in vitro during cell fractionation? LBDs of the steroid receptors are highly hydrophobic and thus good targets for the chaperoning Hsps (Hansen and Gorski 1985; Alnemri and Litwack 1993; Ylikomi et al 1998). Ligand binding decreases the hydrophobicity of the LBD (Hansen and Gorski 1985; Alnemri and Litwack 1993; Ylikomi et al 1998). Overexpression of nonliganded LBDs of PR and GR, but not liganded LBDs, activates endogenous heat shock factor, suggesting that the nonliganded LBDs of steroid receptors can interact with Hsps in vivo (Xiao and DeFranco 1997). Homogenization releases the receptor into an aqueous milieu and thus exposes the hydrophobic regions of the LBD, which is likely to trigger the association of Hsps. In fact, there is ample evidence for the in vitro formation of the oligomeric complex, since oligomeric complexes can be generated with purified steroid receptors in reticulocyte lysate and with several other cell lysates (Smith et al 1990; Stancato et al 1996). The oligomeric complex can also be reconstituted from dissociated PR during homogenization with a tissue that does not contain PR (Tuohimaa et al 1993). The oligomeric complex cannot be seen in vitro with liganded receptor or with retinoic acid receptors, since their LBDs are less hydrophobic and thus they associate with the Hsps with a lower affinity (Ylikomi et al 1998). It is conceivable that the association of PR with Hsp90 in vivo takes places transiently after receptor synthesis. This association could also occur in the cytoplasm during nucleocytoplasmic shuttling of receptor (Chandran and DeFranco 1992). If so, repeated chaperoning of the receptor in the cytoplasmic compartment of the cell apparently plays a crucial role in the maintenance of functional activity of PR.