THE JOURNAL

0 1992 by The American Society for Biochemistry and Molecular Biology,

OF BIOLOGICAL CHEMISTRY
Inc.
Vol. 267, No. 19, Issueof J uly 5, pp. 13728-13734.1992
Printed in U.S.A.
Dual Roles of the 90-kDa Heat Shock Protein hsp90 in Modulating
Functional Activities of the Dioxin Receptor
EVIDENCE THAT THE DIOXIN RECEPTOR FUNCTIONALLY BELONGS TO A SUBCLASS OF NUCLEAR
RECEPTORS WHICH REQUIRE hsp90 BOTH FOR LIGAND BINDING ACTIVITY AND REPRESSION OF
INTRINSIC DNA BINDING ACTIVITY*
(Received for publication, December 23, 1991)
Ingemar Pongratz, Grant G. F. Mason$, and Lorenz Poellingerf
From the Department of Medical Nutrition, Karolinska Institutet, Huddinge University Hospital F-60, Nouum,
S-14186 Huddinge, Sweden
Signal transduction by dioxin (2,3,7,84etrachloro-
dibenzo-p-dioxin) is mediated by the intracellular
dioxin receptor which, in its dioxin-activated state,
regulates transcription of target genes encoding drug
metabolizing enzymes such as cytochrome P-450IA1
and glutathione S-transferase Ya. Upon binding of
dioxin the receptor translocates from the cytoplasm to
the nucleus in vivo and is converted from a latent non-
DNA binding form to a species which binds to dioxin-
responsive positive control elements in vitro. The la-
tent receptor form is associated with an inhibitory
protein (the 90-kDa heat shock protein, hsp90), the
release of which is necessary to unmask the DNA bind-
ing activity of the receptor. Here we have established
a protocol to disrupt the hsp90-receptor complex in the
absence of ligand. We show that it was possible to
covalently cross-link with dioxin only the hsp90-as-
sociated form of dioxin receptor. In contrast, the dis-
rupted hsp90-free form of receptor did not form a
stable complex with dioxin but bound DNA constitu-
tively. Moreover, we could partially reconstitute the
ligand binding activity of the salt-disrupted hsp90-free
dioxin receptor by incubation with hsp90-containing
reticulocyte lysate but not by incubation with wheat
germ lysate which lacks immuno-detectable levels of
hsp90. Thus, we demonstrate that the dioxin receptor
loses its high affinity ligand binding activity following
release of hsp90 and that it is possible to reverse this
process. In conclusion, hsp90 appears to play dual roles
in the modulation of functional activities of the dioxin
receptor: (i) it represses the intrinsic DNA binding
activity of the receptor and (ii) it appears to determine
the ability of the receptor to assume and/or maintain a
ligand binding conformation.
~~ ~ ~
Dioxin and related halogenated aromatic hydrocarbons are
* This work was supported by Grant ESO 3954 from the National
Institutes of Health and by a grant from the Swedish Cancer Society.
The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby
marked advertisement in accordance with 18U.S.C. Section 1734
solely to indicate this fact.
$ Recipient of a research fellowship from the Swedish Medical
Research Council.
5 To whomcorrespondence should be addressed Dept. of Medical
Nutrition, Karolinska Inst., Huddinge University Hospital F-60, S-
141 86 Huddinge, Sweden. Tel.: 46-8-779-7124; Fax: 464-711-6659,
The abbreviations and trivial names used are: dioxin, 2,3,7,8-
tetrachlorodibenzo-p-dioxin; hsp90, the 90-kDa heat shock protein;
XRE, xenobiotic response element; [3H]dioxin, [1,6-3H]2,3,7,8-te-
trachlorodibenzo-p-dioxin; [251[dioxin, 2-azid0-3-[~~1]2-iod0-7,8-di-
bromo-dibenzo-p-dioxin; SDS, sodium dodecyl sulfate; Hepes, 4-(2-
hydroxyethy1)-l-piperazineethanesulfonic acid.
a class of environmental pollutants that give rise to a plethora
of biochemical and toxic responses, including the induction
of transcription of genes encoding the drug-metabolizing en-
zymes cytochrome P-450IA1 and glutathione S-transferase
Ya. This induction response is mediated by the intracellular
dioxin receptor protein (reviewed in Landers and Bunce,
1991). The dioxin receptor is a ubiquitous regulatory factor
that is present in a latent (non-DNA binding) configuration
in the absence of ligand. This latent form of receptor is
presumably located in the cytoplasmic compartment of target
cells (for reviews, see Landers and Bunce, 1991; Poellinger et
al., 1991). Upon ligand binding, however, the receptor is
translocated to the nucleus and converted to a DNA binding
form recognizing in vitro dioxin-inducible positive control
elements (XREs) which confer dioxin responsiveness to target
promoters in uiuo (Fujisawa-Sehara et al., 1987,1988; Denison
et al., 1988; Hapgood et al., 1989; Neuhold et al., 1989; Paulson
et al., 1990).
The mechanism of activation of the receptor to a functional
species is poorly understood. A non-dioxin binding factor
appears to be required for nuclear translocation of the ligand-
activated dioxin receptor (Hoffman et al., 1991). In addition,
preliminary evidence suggests that the DNA binding activity
of the dioxin receptor is regulated by phosphorylation (Pon-
gratz et al., 1991). Finally, weand others (Denis et d., 1988a;
Perdew, 1988; Nemoto et al., 1990a) have shown previously
that the non-ligand-binding 90-kDa heat shock protein hsp90
forms a stable heteromeric complex with the dioxin receptor
i n vitro, thereby preventing the receptor from binding to DNA
recognition elements (Wilhelmsson et al., 1990). Thus, dioxin-
induced activation of the receptor requires release of hsp90.
The specific DNA binding activity of several steroid receptors,
most notably that of the glucocorticoid receptor, has also been
shown to be repressed by association with hsp90 (Sanchez et
al., 1987; Denis et al., 1988b and references therein). In
addition, hsp90 appears to modulate the steroid binding ac-
tivity i n vitro (Bresnick et al., 1989; Nemoto et al., 1990b)
and, presumably as a consequence thereof, hormone respon-
siveness i n vivo (Picard et al., 1990) of the glucocorticoid
receptor.
In the present report, weexamine DNA and ligand binding
activities of hsp90-free or hsp90-associated forms of dioxin
receptor. We show that, concomitant with artificially induced
release of hsp90 in vitro and the unmasking of the XRE
binding activity of the receptor, its ligand binding capacity
was lost. Thus, our results indicate that hsp90 is required to
maintain the ligand-free receptor in a high affinity ligand
binding conformation. Consistent with this hypothesis, we
13728
Dual Roles of hsp90 in Dioxin Receptor Function 13729
demonstrate that, in the absence of hsp90, the receptor be-
haves as a constitutive dioxin-insensitive DNA binding factor.
More importantly, wecan partially reconstitute ligand bind-
ing activity of the salt-disrupted hsp90-free receptor by in-
cubation with hsp90-containing rabbit reticulocyte lysate but
not with wheat germ lysate which does not contain any
immuno-detectable levels of hsp90. In conclusion, the dioxin
receptor together with the glucocorticoid receptor (reviewed
in Dalman et al., 1991) are apparently unique among members
of the nuclear receptor family in that hsp90 is important for
two critical receptor functions in the noninduced cell: (i)
maintaining the receptor in a non-functional state and (ii)
maintaining the receptor in a ligand binding configuration.
EXPERIMENTAL PROCEDURES
Cells and Preparation of Cellular Extracts-Hepa l cl c7 mouse
hepatoma cells (Hankinson, 1983) were generously provided by Dr.
Oliver Hankinson (UCLA) and grown in minimum essential medium
as described previously (Cuthill et al., 1987; Wilhelmsson et al., 1990).
Cells were grown to near confluence in an atmosphere of 6% COZ
until harvested. Cytosol was prepared by homogenization of untreated
cells in 2 volumes of ETG buffer (20 mM Tris-HC1, pH 7.4, 10% (w/
v) glycerol, 1 mM EDTA, and 1 mM dithiothreitol), and centrifugation
at 120,000 X g for 45 min. The resulting supernatant was taken as
the cytosolic fraction and either used immediately or frozen in small
aliquots at -70 "C.
Fractionation of Cytosol by Sucrose Density Gradient Centrifuga-
tion-Hepa lclc7 cell cytosol (500 pl ; about 2 mg of protein/ml of
cytosol) was layered on 10-40% (w/v) linear sucrose gradients pre-
pared in ETG buffer containing 10 mM sodium molybdate and 50-
150 mM NaCl as indicated. The gradients were centrifuged at 300,000
X g to a cumulative centrifugal effect of 1.7 X 10" radians2/s in a
Beckman L8-60 ultracentrifuge. Fractions were collected by gravity
flow starting from the bottom of the gradients. 14C-Labeled IgG (6.6
S) and bovine serum albumin (4.4 S) were used as external sedimen-
tation marker proteins.
I n Vitro DNA Binding Assay-Dioxin receptor-dependent DNA
binding activities were analyzed by a gel mobility shift assay per-
formed essentially as described previously (Hapgood et al., 1989;
Nemoto et al., 1990a). DNA binding reactions were assembled with
the indicated protein fractions in 10 mM Hepes, pH 7.9, 5% (v/v)
glycerol, 0.5 mM dithiothreitol, 2.5 mM MgC12, 1 mM EDTA, 0.08 (w/
v) Ficoll, and 4 mM spermidine at a final concentration of 60 mM
NaCl in a final volume ranging between 20 and 50 pl. A 32P-3'-end-
labeled double-stranded 36-base pair oligonucleotide XRE (Pongratz
et al., 1991) spanning the dioxin-responsive XREl element of the rat
cytochrome P-450IA1 upstream/promoter region (Fujisawa-Sehara et
al., 1987) was added to the reactions as a specific probe in the presence
of nonspecific competitor DNA (100 ng to 1 pg of poly(d1-dC)). After
addition of specific probe, the reactions were incubated for 15 min at
25 "C or at 0-4 'C, as indicated. Protein-DNA complexes were re-
solved on 4% (acrylamidefiisacrylamide ratio of 301) low ionic
strength native polyacrylamide gels at 30 mA at 0-4 'C, using a Tris-
glycine buffer (Hapgood et al., 1989). A double-stranded 38-base pair
oligonucleotide (Pongratz et al., 1991) spanning a glucocorticoid re-
sponse element of the rat tyrosine aminotransferase gene was used
as an XRE-unrelated sequence in DNA binding competition experi-
ments.
Noncoualent and Covalent Labeling of the Dioxin Receptor-The
cytosolic dioxin receptor was noncovalently labeled by incubation of
crude or fractionated cytosol for 3 h at 30 "C with 10 nM [3H]dioxin
(Chemsyn, Lenexa, KS; specific activity 40 Ci/mmol) or 10 nM
nonradioactive dioxin. The affinity ligand [1Z51]dioxin was synthesized
and purified essentially as described (Poland et al., 1986).' Based on
the incorporation of " ' 1 , the overall yield of photoaffinity ligand was
generally 20%, and the specific activity was estimated to be 2176 Ci/
mmol. The dioxin receptor was covalently labeled by incubation for
1 h at 0-4 " C of crude or fractionated cytosol with ['2sI]dioxin and
subsequent ultraviolet irradiation at 330 nm for 15 min as described
(Perdew, 1988; Pongratz et al., 1991). Prior to ultraviolet irradiation,
the sample was treated with dextran-coated charcoal to remove
nonbound ligand. Labeled proteins were analyzed on 10% SDS-
polyacrylamide gels and visualized by autoradiography. Prestained,
G. G. F. Mason, and J . Bergman, unpublished data.
I4C-labeled molecular mass marker proteins were purchased from
Bio-Rad and Amersham, respectively.
Reconstitution of the Ligand Binding Activity of the Salt-disrupted
Dioxin Receptor-Salt-disrupted hsp90-free dioxin receptor was pro-
duced by treatment of crude cytosol from untreated Hepa lclc7 cells
with 0.5 M NaCl for 1 h at 25 "C in the absence of ligand, followed by
fractionation at 0-4 "C on a 10-40% (w/v) sucrose gradient as de-
scribed above. Fractions containing hsp90-free dioxin receptor sedi-
menting in the 4-5 S region of the gradient were identified by gel
mobility shift DNA binding analysis and incubated with either rabbit
reticulocyte lysate (methionine-free; Promega, Madison, WI ), wheat
germ lysate (Promega), or buffer for 1 h at 37 "C. Ligand binding
activity was subsequently analyzed by covalent labeling of the recep-
tor with ['251]-dioxin and SDS-polyacrylamide gel electrophoresis as
described above.
Safety Precautions-In experiments involving the use of dioxin,
special handling procedures were employed (Cuthill and Poellinger,
1988 and references therein), and contaminated materials were dis-
posed of by high temperature incineration.
RESULTS
Experimental Design-Hsp9O-associated and hsp90-free
forms of dioxin receptor can be conveniently monitored by
sucrose gradient sedimentation analysis. The -300-kDa la-
tent hsp90-associated dioxin receptor form sediments around
9 S (Wilhelmsson et al., 1986, 1990; Denis et al., 1988a),
whereas the hsp90-free apparently monomeric form of the
-100-kDa receptor sediments around 4-5 S (Poellinger et al.,
1983). In addition, an hspgo-free, either homodimeric or het-
eromeric, DNA-binding receptor form of -200 kDa is detected
under certain conditions and sediments around 6 S (Prokip-
cak and Okey, 1988; Dougherty, 1989; Wilhelmsson et al.,
1990; Gasiewicz et al., 1991). In the present experiments, we
fractionated crude cytosol from the dioxin-responsive mouse
hepatoma cell line Hepa lclc7 (Hankinson, 1983) on sucrose
gradients to identify and characterize components required
for ligand-dependent activation of the dioxin receptor.
Salt- and Temperature-induced Disruption of the 9 S hsp90-
containing Heteromeric Dioxin Receptor Form-Given the
background that the 9 S hsp9O-associated form of dioxin
receptor recovered in cytosolic extracts from mouse liver or
mouse hepatoma cells has been reported to be very difficult
to disrupt in vitro (Denison et al., 1986; Denison and Vella,
1990), we initially tested and optimized our experimental
conditions to induce subunit dissociation of this form of
receptor. To this end, we used the latent form of dioxin
receptor recovered in cytosolic extracts from untreated Hepa
lclc7 cells (Cuthill et al., 1991 and references therein). To
detect the ligand-binding subunit of the latent 9 S receptor
complex, it was covalently labeled with [1251]dio~in by ultra-
violet irradiation prior to biochemical manipulation.
"51-Photoaffinity labeling of the receptor in crude
Hepalclc7 cytosol produced a 95-kDa labeled band on SDS-
polyacrylamide gels (Fig. 1). Labeling of this band was greatly
reduced in the presence of a 100-fold molar excess of the high
affinity (Gillner et al., 1985 and references therein) dioxin
receptor ligand 2,3,7,8-tetrachlorodibenzofuran but not in the
presence of an identical excess of dexamethasone (Fig. 1,
compare lanes 1-3), a compound that does not bind to the
receptor (Poellinger et al., 1983). Moreover, the molecular
mass of the covalently labeled species was identical to that of
the bona fide murine dioxin receptor (Perdew, 1988; Landers
et al., 1989; Perdew and Hollenback, 1990; Pongratz et al.,
1991). Thus, the photoaffinity ligand selectively labeled the
receptor protein and exhibited very low, if any, levels of
nonspecific binding activity.
The '251-photoaffinity-labeled cytosolic Hepa lclc7 dioxin
receptor was fractionated on sucrose gradients and analyzed
by SDS-polyacrylamide gel electrophoresis and autoradiog-
13730 Dual Roles of hsp90 in Dioxin Receptor Function
LL
o x
Competitor: - S
Mr:
. . ~. . . .- .=.
200
R -
46
30
14.5
M 1 2 3 M
FIG. 1. Photoaffinity labeling of the cytosolic dioxin recep-
tor with ['2'I]dioxin. A crude cytosolic extract (-2 mg of protein/
ml) from nontreated Hepa lclc7 cells was incubated with about 5 nM
["'Ildioxin and irradiated with ultraviolet light as detailed under
"Experimental Procedures" in the absence or presence of a 100-fold
molar excess of either 2,3,7,8-tetrachlorodibenzofuran ( TCDF) or
dexamethasone ( Dex) . The covalently labeled receptor was analyzed
by electrophoresis on a 10% (w/v) SDS-polyacrylamide gel and au-
toradiography. The molecular mass values of the protein standards
(lanes designated M) are indicated in kilodaltons.
A
Ligand-free linking hsp90 Centrifugation
UV-Cross- Releaseof Sucrose Gradient
Receptor 1 I 1
k k
I 1 h: 25% I
4
+lZ51-dioxin +0.5M KC1 Fractions
(SDSPAGE Analysis)
B
6.6 S 4.5 S
+ + kD:
-. ".
"
205
- 117
80
50
d'
M I 5 10 15 20 n M
SDG FRACTIONS
FIG. 2. Salt- and temperature-induced subunit dissociation
of the Hepa 1 dioxin receptor. A, schematic diagram of the
experimental protocol used for the disassembly of the latent dioxin
receptor form. Cytosol (about 2 mg of protein/ml) was covalently
labeled with ['ZSl]dioxin and exposed to an increased ionic strength
and an elevated temperature as indicated. This material (500 pl) was
then fractionated on IO-40% (w/v) sucrose density gradients (SDG)
containing 150 mM NaCI. B, the individual fractions were analyzed
on a 10% (w/v) SDS-polyacrylamide (SDS-PAGE) gel and autora-
diography. The molecular mass values of the protein standards (lanes
M ) are indicated in kilodaltons. On top of the autoradiograph are
indicated the sedimentation positions of sedimentation marker pro-
teins (albumin, 4.5 S and IgG, 6.6 S) run simultaneously on separate
gradients.
ligand-free dioxin receptor, respectively. The DNA binding
activity of the receptor was monitored by a gel mobility shift
assay which used as specific DNA probe an oligonucleotide
spanning the dioxin response element XREl of the rat cyto-
chrome P-450IA1 gene (Fujisawa-Sehara et al., 1987). In the
absence of any in vitro manipulation, the ligand-free dioxin
receptor did not generate any detectable complex with the
labeled XRE probe upon sucrose gradient fractionation (Fig.
3). In parallel experiments, the crude ligand-free cytosolic
receptor was exposed to a combination of high ionic strength
and elevated temperature (diagrammed in Fig. 4A) according
to the very protocol which induced subunit dissociation of the
ligand-occupied receptor (Fig. 2). Subsequently, the material
was fractionated on sucrose gradients. Gel mobility shift
analysis of the individual fractions demonstrated a broad peak
of XRE binding activity which resulted in formation of an
XRE complex of identical relative mobility as compared with
that of the dioxin-induced receptor-dependent complex in
crude cytosol (Fig. 4B, compare sucrose gradient fractions 17-
21 with input (inp) lanes). This peak of DNA binding activity
was centered in the 4-5 S region of the gradient with a small
plateau at -6 S. Note that the levels of a nonspecific DNA
binding activity generating a complex with the labeled XRE
probe that migrates faster than the receptor-dependent com-
plex vary in different cytosolic preparations (Fujisawa-Sehara
et al., 1988; Cuthill et al., 1991; Figs. 3 and 4B, compare input
(inp) lanes). Although broader, the sedimentation position of
the receptor-dependent XRE binding activity appears to co-
incide with that of the ligand-bound hsp90-free form of recep-
tor in nuclear extract (Wilhelmsson et al., 1990). Importantly,
receptor-dependent XRE binding activity was not detected in
the 9 S region (Fig. 4B, fractions 10-1 1 ) of the sucrose gradient
containing the hsp90-associated form of receptor (Wilhelms-
son et al., 1990).
The Salt-disrupted hsp90-free Dioxin Receptor Does Not
Form a Stable Complex with Dioxin-To examine the dioxin
6.6 S 4.5 S
t +
4 - R
raphy. In the absence of any prior in vitro manipulation, this
analysis demonstrated the -95-kDa receptor only in the 9 S
region of the sucrose gradient (data not shown), consistent
with the sedimentation properties of the intact receptor-hsp90
heteromeric complex (Denis et al., 1988a). However, exposure
of the "'I-photoaffinity-labeled receptor to both a high con-
centration of salt and an increased temperature (summarized
schematically in Fig. 2A) resulted in sedimentation of the
labeled 95-kDa protein both in the 9 and 4 S regions of the
gradient (Fig. 2B). Thus, this treatment produced a signifi-
cant, albeit not total, disruption of the hsp90-associated 9 S
form of receptor and generation of its apparently hsp90-free
(Wilhelmsson et al., 1990) form.
The XRE Binding Activity of the Ligand-free Dioxin Recep-
tor Is Unmasked upon Disruption of the hsp90-Receptor Het-
eromeric Complex-We next examined the specific DNA bind-
ing activity of untreated and salt- and temperature-disrupted
$ 1 5 10 15 20 g a .P x x .P
a
?
SDG FRACTIONS
u u
I +
-
inp
FIG. 3. Gel mobility shift analysis of untreated ligand-free
dioxin receptor following sucrose gradient fractionation. Un-
treated Hepa 1 cytosol (500 pl ; about 2 mg of protein/ml) was analyzed
on sucrose gradients containing 150 mM NaCl. The same sedimen-
tation marker proteins were used as in Fig. 1. The individual gradient
fractions were analyzed by gel mobility shift analysis using a :'lP-
labeled XRE oligonucleotide probe. As a control, the XRE binding
activity of unfractionated input (inp) cytosol was also analyzed. To
this end, input cytosol was incubated in the absence (- dior) or
presence (+dior) of 10 nM dioxin at 25 "C for 3 h. The relative
mobility of the dioxin-induced receptor-dependent XRE complex is
indicated ( R) . The mobility of the probe in the absence of any added
protein is shown in the left lane. The same external sedimentation
marker proteins were used as in Fig. 1.
Dual Roles of hsp90 in Dioxin Receptor Function 13731
A
B
Ligand-free I hspgO I
Release of SucroseGradient
Centrifugation
Receptor 1 1 , I I
* 4
+0.5 h KC1 Fractions
a) ONA Binding Assay
b) Ligand Binding Assay
(UV Crosslinking and
SDSPAGE Analysis)
c
2 ' 5 10
e
P SDG FRACTIONS
15 20 n x x
V U
I +
e .P .e
-
InD
C
P "
"
"
L.
6.6 S 4.5 S
* * kD:
""
?05
I17
80
50
M 1 5 10 15
20 P M
SDG FRACTIONS
e
FI G. 4. The salt-disrupted hsp90-free dioxin receptor form
does not generate a stable complex with ligand. A, schematic
diagram of the protocol used for disrupture of the latent dioxin
receptor form. Ligand-free cytosol (500 pl; about 2 mg of protein/ml)
was exposed to high ionic strength and an elevated temperature as
indicated and subsequently fractionated by sucrose density gradient
( SDG) centrifugation. The gradients contained 150 mM NaCI. The
individual sucrose gradient fractions were analyzed for XRE binding
( R ) and dioxin binding (C) activities. The DNA binding activity was
analyzed by gel mobility shift analysis using a labeled XRE probe.
Control lanes included untreated and dioxin-treated crude input
cytosol as outlined in the legend to Fig. 2. The relative mobility of
the receptor-dependent "'P-XRE complex is indicated ( R) . The mo-
bility of the probe in the absence of any added protein is shown in
the left lane in B. The ligand binding activity of the receptor was
monitored by photoaffinity labeling of the individual sucrose gradient
fractions with [1251]dioxin. Covalently labeled material was analyzed
on 10% polyacrylamide gels and autoradiography as detailed in the
legend to Fig. 1. The same external sedimentation marker proteins
were used as in Fig. 1.
binding activity of the in vitro manipulated ligand-free form
of receptor, sucrose gradient fractions containing salt- and
temperature-treated ligand-free receptor were analyzed by
photoaffinity labeling with ['*'I]dioxin. SDS-polyacrylamide
gel electrophoresis of covalently labeled complexes demon-
strated the -95-kDa ['251]dioxin-receptor complex only in a
region of the sucrose gradient (around fractions 10-11) cor-
responding to the sedimentation position of the -9 S dioxin
receptor-hsp90 heteromeric complex (Fig. 4C). Thus, it was
not possible to detect the [1251]dioxin-receptor complex in the
4-5 S region of the gradient containing the DNA-binding
form of dioxin receptor (Fig. 4, compare B and C) . In agree-
ment with these observations, noncovalent dioxin binding
assays using ['Hldioxin as specific ligand only showed detect-
able levels of ligand binding activity in the 9 S region of the
gradient containing the hsp90-associated form of dioxin
receptor (data not shown). Consequently, the apparently
hsp90-free form of dioxin receptor appears to exhibit a signif-
icantly reduced affinity for dioxin, resulting in failure to
produce stable complexes with the photoaffinity ligand ['*'I]
dioxin.
Constitutive XRE Binding Activity of the Salt- and Tem-
perature-disrupted Form of Dioxin Receptor-Activation of
the dioxin receptor is a ligand-dependent event both in vivo
(Fujisawa-Sehara et al., 1988; Denison et al., 1988; Hapgood
et al., 1989; Neuhold et al., 1989) and in reconstituted exper-
iments in cell-free model systems (Fujisawa-Sehara et al.,
1988; Nemoto et al., 1990a; Cuthill et al., 1991). However, salt-
and temperature-induced release of hsp90 produced XRE
binding activity of the 4-5 S hsp90-free form of dioxin recep-
tor even in the absence of ligand (Fig. 4B). Sucrose gradient
fractions containing this form of receptor were pooled and
further characterized. Oligonucleotide competition experi-
ments confirmed that the XRE binding activity in the 4-5 S
region of the gradient was specific. XRE complex formation
was greatly reduced in the presence of an excess of the
unlabeled XRE oligonucleotide, whereas an identical excess
of an oligonucleotide spanning an unrelated sequence motif
did not inhibit formation of the complex (Fig. 5A, compare
lanes 2-4). Thus, the salt-disrupted form of receptor was
indistinguishable from the ligand-activated receptor both in
terms of the relative mobility of the receptor-XRE complex
and of the DNA binding specificity. In further experiments,
the salt-disrupted dioxin receptor form was incubated with
increasing concentrations of dioxin (up to 100 nM) according
to the protocol resulting in dioxin-dependent activation of the
XRE binding activity of the crude cytosolic receptor (Cuthill
A B
Comp.: - - x 0 (nM): , Z 5
Dioxin
I
I"
- R ' R
FIG. 5. Constitutive DNA binding activity of the apparently
hsp90-free dioxin receptor form. Ligand-free crude Hepa 1 cy-
tosol (500 pl; about 2 mg of protein/ml) was treated and fractionated
on sucrose gradients as described in the legend to Fig. 3. The DNA
binding salt- and temperature-disrupted form of dioxin receptor
sedimenting in the 4-5 S region of the gradient was pooled and
further characterized by gel mobility shift experiments using a labeled
XRE probe. A, DNA binding specificity. The pooled fractions were
incubated with the labeled probe in the absence (lane 2) or presence
of a 40-fold molar excess of the unlabeled XRE oligonucleotide (lane
3) or an unrelated competitor (Comp) oligonucleotide (lane 4 ) span-
ning a glucocorticoid response element (GRE). Lane I shows the
mobility of the probe in the absence of any added protein. B, induction
experiments. The pooled material was incubated in the absence (lane
I ) or presence of 10 (lune 2 ) or 100 nM (lane 3) ["Hldioxin at 25 "c
for 3 h prior to gel mobility shift analysis. The relative mobilities of
the receptor-dependent complex (designated R ) and of the probe
(designated Free) are indicated.
13732 Dual Roles of hsp90 in Dioxin Receptor Functi on
et al., 1991). The crude cytosolic dioxin receptor binds dioxin
with a Kd of about 1 nM and is strongly in vitro activated in
the presence of 1-5 nM dioxin (Cuthill et al., 1991 and refer-
ences therein). However, no increase in the DNA binding
activity of the salt-disrupted hsp90-free form of receptor was
observed in the presence of either 10 or 100 nM dioxin,
demonstrating that this form of receptor behaved like a con-
stitutive dioxin-nonresponsive DNA binding factor (Fig. 5B,
compare lanes 1-3).
Reticulocyte Lysate Directs Reconstitution of the Ligand
Binding Activity of the Salt-disrupted Dioxin Receptor-The
dioxin receptor-hsp90 complex was disassembled by exposing
the ligand-free receptor to high concentrations of salt and an
elevated temperature as diagrammed in Fig. 4A. The hsp90-
free receptor form was separated from the hsp90-receptor
heterodimer by sucrose gradient fractionation and incubated
with rabbit reticulocyte lysate or wheat germ lysate, respec-
tively. Reticulocyte lysate contains high levels of hsp90 (Denis
and Gustafsson, 1989) and directs formation of the hsp9O-
glucocorticoid receptor complex following either in vitro
translation of glucocorticoid receptor mRNA (Denis and Gus-
tafsson, 1989; Dalman et al., 1989) or incubation with the
activated hsp90-free receptor protein (Scherrer et al., 1990).
I n contrast, the glucocorticoid receptor-hsp90 complex is not
reconstituted in wheat germ lysate, which lacks immuno-
detectable levels of hsp90 (Dalman et al., 1989). Following
incubation with either reticulocyte or wheat germ lysate the
originally hsp90-free form of dioxin receptor was labeled with
the photoaffinity ligand ['251]dioxin and analyzed by poly-
acrylamide gel electrophoresis and autoradiography. We could
not detect any covalent labeling of proteins by incubating
['"I]dioxin with rabbit reticulocyte lysate alone (data not
shown). In contrast, Fig. 6 shows that incubation of the
sucrose gradient-fractionated hsp90-free form of dioxin recep-
tor with reticulocyte lysate converted the receptor from a
form which showed no detectable ligand binding activity to a
form which, albeit at a low level, could be covalently labeled
with ["'I]dioxin, as assessed by autoradiography of receptor
material analyzed on SDS-polyacrylamide gels. Since the
input amount of receptor can only be assessed by the rather
inaccurate gel mobility shift DNA binding assay, we can only
tentatively estimate the recovery of ligand binding activity to
about 20%. A similar level of reconstitution of the ligand
binding activity of the salt-disrupted receptor was also ob-
served employing a noncovalent hydroxylapatite adsorption
50
1 2
FIG. 6. Reticulocyte lysate directs reconstitution of the li-
gand binding activity of the disrupted hsp90-free dioxin
receptor form. The DNA-binding salt- and temperature-disrupted
form of dioxin receptor was produced as described in Figs. 3 and 4.
The sucrose gradient fraction exhibiting maximal levels of receptor-
dependent XRE binding activity was identified by gel mobility shift
analysis. This fraction (50 pl) was incubated with 20 p1 of either
wheat germ lysate ( WGL) or reticulocyte lysate ( RL) at 37 "C for 1 h
prior to photoaffinity labeling with [''"Ildioxin. The labeled material
was analyzed on 10% (w/v) SDS-polyacrylamide gels and autoradi-
ography. The mobilities of protein standards are indicated. The
covalently labeled receptor protein is also indicated ( R) .
ligand binding assay (data not shown). Importantly, the co-
valently labeled -95-kDa receptor complex was not generated
in wheat germ lysate (Fig. 6, compare lanes 1 and 2), sup-
porting the notion that the observed (at least partial) recon-
stitution of the ligand binding activity of the dioxin receptor
in reticulocyte lysate may be facilitated by hsp90.
DISCUSSION
Hsp90 Appears to Be a Critical Factor for Maintaining the
Dioxin Receptor in a Ligand Binding State-In the present
study we have shown that in vitro manipulation of the origi-
nally latent ligand-free dioxin receptor produces two receptor
forms, only one of which, the 9 S receptor, shows dioxin
binding activity. I n contrast the other form of receptor lacks
ligand binding activity but shows constitutive DNA binding
activity. Clearly, the 9 S form of receptor still retains some
key factor orland configuration that enables it to bind to its
ligand. Since this form of receptor is associated with hsp90
(Denis et al., 1988a; Wilhelmsson et al., 1990), we propose
that hsp90 may represent a critical determinant of the ligand
binding conformation of the dioxin receptor.
Role of hsp90 in Modulating Nuclear Receptor Function-
Many heat shock proteins are believed to function as molec-
ular chaperones associating with newly synthesized proteins
in normal cells or with damaged proteins in stressed cells and
to thereby play a central role in mediating protein folding and
transport (reviewed by Hightower, 1991; Ellis and van der
Vies, 1991). By forming a complex with the dioxin receptor,
hsp90 interferes with its binding to DNA in vitro (Wilhelms-
son et al., 1990; Fig. 4). Salt- and temperature-induced disas-
sembly of the hsp90-receptor complex results in binding of
the receptor to DNA, strongly suggesting that the dioxin
receptor has an intrinsic affinity for DNA, which is repressed
in the absence of ligand. Thus, hsp90 appears to also function
both as an inhibitory protein repressing the intrinsic DNA
binding activity of the dioxin receptor and as a molecular
chaperone maintaining the receptor in a ligand binding con-
figuration. In line with this model, the salt-disrupted ligand-
and hsp9O-free dioxin receptor form did not form a stable
complex with dioxin and showed a constitutive level of DNA
binding activity that was not further increased by dioxin
treatment.
I n the case of the members of the steroid hormone receptor
superfamily, a subclass of receptors including the glucocorti-
coid receptor has been shown to interact with hsp90 (Sanchez
et al., 1987 and references therein); whereas the thyroid hor-
mone and retinoic acid receptors do not appear to form any
complex with hsp9O (Dalman et al., 1990,1991). Importantly,
hsp90-glucocorticoid receptor interaction has been demon-
strated to result in repression of receptor-dependent DNA
binding activity in vitro (Sanchez et al., 1987; Denis et al.,
1988). Moreover, the glucocorticoid receptor domain which
interacts with hsp90 is important for repression of receptor
function in a hormone-free environment in vivo (Picard et al.,
1988; Cadepond et al., 1991 and references therein), lending
support to the model that hsp90 functions as an inhibitor of
the glucocorticoid receptor in vivo. A similar negative regula-
tory strategy has been experimentally established for several
other classes of transcription factors. For instance, the non-
DNA binding inhibitory proteins Id, I-POU, and IKB dimerize
with their respective target transcription factors: MyoD,
POU-domain, and the Rel-related NF-KB factors, respec-
tively, thereby repressing both the DNA binding activity and
function of these distinct classes of factors (Benezra et al.,
1990; Treacy et al., 1991, Davis et al., 1991). Possibly the I P-
1 factor acts by a similar mechanism to repress the activity
Dual Roles of hsp90 in Dioxin Receptor Function 13733
of the Fos-J un complex (Auwerx and Sassone-Corsi, 1991).
In analogy to our results with the dioxin receptor, hsp90
seems to be required for efficient binding of ligand by the
glucocorticoid receptor. Whereas, for instance, the ligand
binding activity of the progesterone receptor has been re-
ported not to be affected by the release of hsp90 (Bagchi et
al., 1990), disassembly of the glucocorticoid-hsp90 complex
appears to induce a receptor conformation which exhibits low
affinity for the hormone in vitro (Bresnick et al., 1989; Nemoto
et al., 1990b). In agreement with this observation, a down-
regulated level of hsp90 expression reduces glucocorticoid
hormone responsiveness in vivo (Picard et al., 1990). In con-
clusion, both the glucocorticoid and dioxin receptors seem to
functionally belong to a distinct subclass of nuclear receptors
which associate with hsp90 and require hsp90 as a molecular
chaperone to stabilize and maintain a high affinity ligand
binding conformation.
Mechanism of Activation of the DNA Binding Activity of the
Dioxin Receptor-In addition to ligand-induced release of
hsp90, activation of the dioxin receptor may require phos-
phorylation of the receptor to produce the DNA binding form
(Pongratz et al., 1991). It will now be interesting to establish
whether the receptor is phosphorylated prior to the assembly
of the hsp90-receptor complex or whether receptor phos-
phorylation occurs following dissociation of the complex.
Moreover, the architecture of the ligand-activated nuclear
form of dioxin receptor is presently not fully understood.
Under nondenaturing conditions this form of receptor is
detected as a 200-kDa protein (Prokipcak and Okey, 1988;
Hapgood et al., 1989; Wilhelmsson et al., 1990). In addition,
DNA cross-linking studies have indicated association between
the 100-kDa ligand-binding receptor and a protein of very
similar molecular mass to form a complex of about 200 kDa
(Elferink et al., 1990; Gasiewicz et al., 1991). It is still an
unresolved issue, however, whether the active 200-kDa dioxin
receptor form represents a homo- or heterodimeric complex
(Poellinger et al., 1992). An obvious candidate for a receptor-
associated factor and component of the nuclear -200-kDa
receptor complex is the 87-kDa Arnt factor which facilitates
nuclear translocation of the dioxin receptor and carries a
putative helix-loop-helix dimerization interface motif (Hoff-
man et al., 1991). In the case of the retinoic acid, thyroid
hormone, and vitamin D receptors, their DNA binding activ-
ity is enhanced by heterodimerization with the structurally
related RXR coregulator (Yu et al., 1991) or with additional,
as yet unidentified cell type-specific coregulators (Glass et al.,
1990). Moreover, it is noteworthy that dimerization of both
the glucocorticoid and progesterone receptors appears to occur
in the absence of DNA (Cairns et al., 1991; DeMarzo et al.,
1991) and that the dimerization process of the progesterone
receptor has been shown to be related to the release of hsp90
(DeMarzo et al., 1991). The role of hsp90, if any, in the
dimerization process of the dioxin receptor remains therefore
to be investigated. In summary, activation of the dioxin
receptor is a multistep process which is initiated by the
binding of ligand and the ligand-promoted release of hsp90.
This release, although necessary, appears not to be sufficient
for maximal activation of the receptor. I t will therefore be
important to further characterize the architecture of both the
unactivated and activated receptor complexes and to study
the roles of combinatorial regulation and phosphorylation in
the receptor activation pathway.
Acknowledgments-We thank D;. Oliver Hankinson (UCLA) for
supplying the Hepalclc7 cell line, Asa Andesson and Kristina Sjob-
lom for expert technical assistance, and Dr. J an-Ake Gustafsson for
a critical review of the manuscript.
REFERENCES
Auwerx, J ., and Sassone-Corsi, P. (1991) Cell 64,983-993
Bagchi, M. K., Tsai, S. Y., Tsai, M.-J ., and O'Malley, B. W. (1990)
Nature 345,547-550
Benezra, R., Davis, R. L., Lockshon, D., Turner, D. L., and Wein-
traub, H. (1990) Cell 61, 49-59
Beato, M. (1989) Cell 56, 335-344
Bresnick, E. H., Dalman, F. C., Sanchez, E. R., Pratt, W. B. (1989)
J. Biol. Chem. 264,4992-4997
Cadepond, F., Schweizer-Groyer, G., Segard-Maurel, I., J ibard, N.,
Hollenberg, S. M., GiguGre, V., Evans, R. M., and Baulieu, E.-E.
(1991) J. Bi d. Chem. 266,5834-5841
Cairns, W., Cairns, C., Pongratz, I., Poellinger, L., and Okret, S.
(1991) J. Biol. Chem. 266,11221-11226
Cuthill, S., and Poellinger, L. (1988) Biochem+try 27, 2978-2982
Cuthill, S., Poellinger, L., and Gustafsson, J .-A. (1987) J. Biol. Chem.
Cuthill, S., Wilhelmsson, A,, and Poellinger, L. (1991) Mol. Cell. Biol.
Dalman, F. C., Bresnick, E, H., Patel, P. D., Perdew, G. H., Watson,
S. J ., J r., and Pratt, W. B. (1989) J. Biol. Chem. 264,19815-19821
Dalman, F. C., Koenig, R. J ., Perdew, G. H., Massa, E., and Pratt,
W. B. (1990) J. Biol. Chem. 265, 3615-3618
Dalman, F. C., Sturzenbecker, L. J ., Levin, A. A., Lucas, D. A.,
Perdew, G. H., Petkovitch, M., Chambon, P., Grippo, J . F., and
Pratt, W. B. (1991) Biochemistry 30,5605-5608
Davis, N., Ghosh, S., Simmons, D. L., Tempst, P., Liou, H.-C.,
Baltimore, D., and Bose, H. R., J r. (1991) Science 253, 1268-1271
DeMarzo, A. M., Beck, C. A., Onate, S. A., and Edwards, D. P. (1991)
Proc. Natl. Acad. Sci. U. S. A. 88, 72-76
Denis, M., and Gustafsson, J.-A. (1989) J. Biol. Chem. 264, 6005-
6008
Denis, M. Cuthill, S., Wikstrom, A.-C., Poellinger, L., and Gustafs-
son, J.-A. (1988a) Biochem. Biophys. Res. Commun. 155,801-807
Denis, M., Poellinger, L., Wikstrom, A.-C., and Gustafsson, J .-k
(1988b) Nature 333, 686-688
Denison, M. S., and Vella, L. M. (1990) Arch. Biochem. Biophys.
Denison, M. S., Vella, L. M., and Okey, A. B. (1986) J. Biol. Chem.
Denison, M. S., Fischer, J . M., and Whitlock, J . P., J r. (1988) Proc.
Dougherty, J . J . (1990) J. Biol. Chem. 264,8786-8790
Elferink, C. J ., Gasiewicz, T. A., and Whitlock, J . P., J r. (1990) J.
Ellis, R. J ., and van der Vies, S. M. (1991) Annu. Reu. Biochem. 60,
Fujisawa-Sehara, A., Sogawa, K., Yamane, M., and Fujii-Kuriyama,
Fujisawa-Sehara, A,, Yamane, M., and Fujii-Kuriyama, Y. (1988)
Gasiewicz, T. A., Elferink, C. J ., and Henry, E. C. (1991) Biochemistry
Gillner, M., Bergman, J ., Cambilleau, C., Fernstrom, B., and Gus-
Glass, C. K., Devary, 0. V., and Rosenfeld, M. G. (1990) Cell 63,
Hankinson, 0. (1983) Somatic Cell Genet. 9 , 497-514
Ha good, J ., Cuthill, S. Denis, M., Poellinger, L., and Gustafsson, J .-
dE: (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 60-64
Hapgood, J., Cuthill, S., Soderkvist, P., Wilhelmsson, A., Pongratz,
I., Tukey, R. H., J ohnson, E. F., Gustafsson, J .-A., and Poellinger,
L. (1991) Mol. Cell. Biol. 11,4314-4323
Hightower, L. E. (1991) Cell 66, 191-197
Hoffman, E. C., Reyes, H., Chu, F.-F., Sander, F., Conley, L., Brooks,
B. A., and Hankinson, 0. (1991) Science 252,954-957
Landers, J . P., Piskorska-Pliszczynska, J ., Zacharewski, T., Bunce,
N. J ., and Safe, S. (1989) J. Bi d. Chem. 264, 18463-18471
Landers, J . P., and Bunce, N. (1991) Biochem. J. 276, 273-287
Nemoto, T., Mason,S. G. F., Wilhelmsson, A,, Cuthill, S., Hapgood,
J.. Gustafsson, J.-A., and Poellinger, L. (1990a) J. Biol. Chem. 265,
2269-2277
Nemoto, T., Ohara-Nemoto, Y., Denis, M., and Gustafsson, J .-A.
(1990b) Biochemistry 29, 1880-1886
Neuhold, L. A., Shirayoshi, Y., Ozato, K., J ones, J . E., and Nebert,
D. W. (1989) Mol. Cell. Bi d. 9 , 2378-2386
262, 3477-3481
11,401-411
277,382-388
261,10189-10195
Natl. Acad. Sci. U. S. A. 85, 2528-2532
Biol. Chem. 265,20708-20712
321-347
Y. (1987) Nucleic Acids Res. 15,4179-4191
Proc. Natl. Acad. Sci. U. S. A. 85, 5859-5863
30,2909-2916
tafsson, J.-A. (1985) Mol. Pharmacol. 28, 357-363
729-738
13734 Dual Roles of hsp90 in Dioxin Receptor Function
Paulson, K. E., Darnell, J . E., Rushmore, T., and Pickett, C. B. (1990)
Perdew, G. H. (1988) J. Biol. Chem. 263, 13802-13805
Perdew, G. H., and Hollenback, C. E. (1990) Biochemistry 29,6210-
Picard, D., Saker, S. J ., and Yamamoto, K. R. (1988) Cell 54, 1073-
Picard, D., Khursheed, B., Garabedian, M. J ., Fortin, M. G., Lind-
Poellinger, L., Gottlicher, M., and Gustafsson, J .-A. (1992) Trends
Poellinger, L., Lund, J ., Gillner, M., Hansson, L.-A., and Gustafsson,
Poellinger, L., Wilhelmsson A., Cuthill, S., Pongratz, I., Mason, G.
Poland, A., Glover, E., Ebetino, F. H., and Kende, A. S. (1986) J.
Mol. Cell. Biol. 10, 1841-1852
6214
1080
quist, S., and Yamamoto, K. R. (1990) Nature 348, 166-168
Pharmacol. Sci., in press
J.-A. (1983) J. Biol. Chem. 258, 13535-13542
G. F., and Gustafsson, J . - k (1991) Banbury Rep. 35, 311-320
Biol. Chem. 261, 6352-6365
Pongratz, I., Stromstedt, P.-E., Mason, G., and Poellinger, L. (1991)
Prokipcak, R. D., and Okey, A. B. (1988) Arch. Biochem. Biophys.
Sanchez, E. R., Meshinchi, S., Tienrungroj, W., Schlesinger, M., Toft,
D. O., and Pratt, W. B. (1987) J. Biol. Chem. 262,6986-6991
Schemer, L. E., Dalman, F. C., Massa, E., Meshinchi, S., and Pratt,
W. B. (1990) J. Biol. Chem. 265, 21397-21400
Treacy, M. N., He X., and Rosenfeld, M. G. (1991) Nature 350,577-
584
Wilhelmsson, A., Wikstrom, A.-C., and Poellinger, L. (1986) J. Biol.
Chem. 261,13456-13463
Wilhelmsson, A., Cuthill, S., Denis, M., Wikstrom, A.-C., Gustafsson,
J .-A., and Poellinger, L. (1990) EMBO J. 9, 69-76
Yu, V. C., Delsert, C., Andersen B., Holloway, J . M., Devary, 0. V.,
Naar, A. M., Kim, S. Y., Boutin, J .-M., Glass, C. K., and Rosenfeld,
M. G. (1991) Cell 67, 1251-1266
J. Biol. Chem. 266, 16813-16817
26 1,6352-6365