Hormone-dependent Transactivation by the Human Androgen Receptor Is Regulated by a dnaJ Protein*

Genetic studies were performed to examine the role of eukaryotic dnaJ protein, Ydjlp, in the regulated activa tion of human androgen receptor (bAR) after heterolo gous expression in Saccharomyces cerevisiae. Hormone dependent activation of bAR was measured as a function of lacZ reporter gene expression, which was defective in ydjl-151 and ydjl_2 4 null mutant strains compared to the wild type. This defect was not due to receptor misfolding, since bAR in both wild type and mutant strains had a similar capacity to bind hormone. The target for Ydjlp action was determined to be the bAR hormone binding domain since an N·terminal frag ment lacking this region was constitutively active in both wild type and ydjl.151 mutant strains. These data correlate hormone dependence of bAR activation with a requirement for Ydjlp function and are consistent with a role for dnaJ proteins in signal transduction by steroid hormone receptors.

Genetic studies were performed to examine the role of eukaryotic dnaJ protein, Ydjlp, in the regulated activation of human androgen receptor (bAR) after heterologous expression in Saccharomyces cerevisiae. Hormonedependent activation of bAR was measured as a function of lacZ reporter gene expression, which was defective in ydjl-151 and ydjl_2 4 null mutant strains compared to the wild type. This defect was not due to receptor misfolding, since bAR in both wild type and mutant strains had a similar capacity to bind hormone. The target for Ydjlp action was determined to be the bAR hormone binding domain since an N·terminal fragment lacking this region was constitutively active in both wild type and ydjl.151 mutant strains. These data correlate hormone dependence of bAR activation with a requirement for Ydjlp function and are consistent with a role for dnaJ proteins in signal transduction by steroid hormone receptors.
Steroid hormones effect profound physiological changes in animal systems by signaling cells to activate or inhibit a variety of genes. These signaling events are mediated by intracellular receptors which become active transcription factors when bound with ligand. The five most studied receptors of this class are androgen, estrogen, glucocorticoid, mineralocorticoid, and progesterone receptors. In addition, there are more than 50 so-called orphan receptors for which no ligand or responsive genes have been characterized (O'Malley and Connely, 1992).
Molecular chaperone proteins have been reported to have a role in the activation of steroid hormone receptors. For example, the inactive form of the glucocorticoid receptor is a 9 S complex that contains two molecules of Hsp90, a heat shock protein whose precise function remains unclear. Other proteins that bind to Hsp90 in this complex include immunophilins (e.g. Hsp56) and p23. The molecular chaperone Hsp70 has also been described as a component, but unlike Hsp90, its role appears to be transitory since its stoichiometry is less than one molecule per complex (Diehl and Schmidt, 1993). Molecular chaperones dissociate from steroid hormone receptors after hormone treatment (see Pratt (1993) for review).
Recent studies indicated that Hsp70 was required for assembly of the rat glucocorticoid receptor with Hsp90. This was demonstrated in reconstitution experiments between immunopurified glucocorticoid receptor and Hsp90 in rabbit reticulocyte lysates; depletion of Hsp70 from the lysates inhibited * 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 18 U.S.C. Section 1734 solely to indicate this fact. § To whom correspondence should be addressed. Tel.: 212-241-6563; Fax: 212-860-1174; E-mail: caplan@msvax.mssm.edu.
reconstitution, but its readdition facilitated complex formation . Using a similar approach, Smith et al. (1992) demonstrated that antibodies specific to Hsp70 inhibited reconstitution of the progesterone receptor with Hsp90. The role of Hsp70 in protein assembly events, however, is not limited to steroid hormone receptor complexes. In Escherichia coli, Hsp70 (dnaK) and its partner, the dnaJ protein, function together in the assembly of pre-primosomal complexes for phage A replication (Georgopoulos et al., 1990), as well as in disassembly of the inactive dimer form of phage PI repA proteins (Wickner et al., 1991).
Together, dnaJ and Hsp70 participate in a variety of cotranslational and post-translational events that mediate the fate of nascent polypeptide chains. Eukaryotic homologues of E. coli dnaJ have only recently been characterized, and they constitute a large protein family with specific members present in different organelles (for reviews, see Caplan et al. (1993) andCyret al. (1994». The events in which dnaJ and Hsp70 proteins function together include transfer of polypeptides to chaperonins or foldases (e.g. GroEL and TriC) (Langer et al., 1992;Frydman et al., 1994) and transport across biological membranes (Chirico et al., 1988;Deshaies et al., 1988;Atencio and Yaffe, 1992;Caplan et al., 1992a). The relationship between these two proteins is based, at least in part, on the ability of dnaJ to interact with Hsp70 and stimulate its ATPase activity (Liberek et al., 1991a;Cyr et al., 1992;Brodsky and Schekman 1993;Scidmore et al., 1993;Cheetham et al., 1994). This affects the conformation of Hsp70, its affinity for polypeptide, and, presumably, its role in protein assembly events (Liberek et al., 1991b).
The Saccharomyces cerevisiae YDJI gene encodes a functional homologue of E. coli dnaJ (Caplan et al., 1992a). Ydjl protein (YdjIp) is localized to the cytosol and post-translationally modified by farnesylation at its C terminus (Caplan and Douglas, 1991;Caplan et al., 1992b). Ydjlp that was purified after overexpression in E. coli interacts specifically with Hsp70 proteins of the SSA subfamily and stimulates their ATPase activity . Theydjl-151 mutant protein has a reduced ability to stimulate this ATPase, and yeast strains expressing this mutant allele are defective for polypeptide translocation across both endoplasmic reticulum and mitochondrial membranes (Caplan et al., 1992a). This report examines the role of Ydjlp in the activation of heterologously expressed human androgen receptor. Like glucocorticoid receptor, human androgen receptor (hAR)1 interacts with Hsp90 (Mariovet et al., 1992;Veldscholte et al., 1992) and is regulated in yeast by hormone (Purvis et al., 1991), suggesting conservation in the cellular machinery responsible for maintaining the apo-receptor inactive. The results shown be-low indicate that Ydj1p performs a regulatory function in the activation ofhAR, specifically via the hormone binding domain.

EXPERIMENTAL PROCEDURES
Genetic Methods-Yeast cells were cultured in minimal medium containing 0.67% yeast nitrogen base, 2% dextrose, and appropriate purine, pyrimidine, and amino acid supplements depending on auxotrophy. Yeast cells were transformed by the lithium acetate procedure essentially as described by Ito et al. (1983). Integrative transformations were performed after linearization of pPGKareLacZI with Ncol and pPGKhARI with EcoRV. Construction of the ydjl deletion mutant and ydjl-151 strain ACY17b have been described (Caplan and Douglas, 1991;Caplan et aI., 1992a). JC2LZ strain was constructed by transforming JC2 with pPGKareLACZI (Purvis et al., 1991). ACY68 is a diploid strain of JC2 and ACY45. Sporulation and dissection of ACY68 tetrads yielded strains ACY72-ACY75.
Plasmid Constructions-pGl-hAR was constructed after ligating a 3-kilobase pair BglII fragment containing the hAR cDNA (from pPGKhARI, Purvis et al., 1991) into BamHI-digested pGl (the gift of Dr. M. G. Douglas). pARVP-16 was constructed by digesting pGl-hAR with TthIII and Sail releasing a 1.2-kilobase pair DNA fragment encoding the hAR hormone binding domain. In its place was ligated a similarly digested 300-base pair fragment encoding the VP-16 activating domain that was amplified from pVP16 (the gift of Dr. J. Licht) using the primers 5'-GCGCGCGACAGTGTCAGCCCCCCCGAC-CGATGTC and 3'-GCGCGCGTCGACCGAACCGGGGACGGGAGG by PCR (20 cycles at 55°C annealing temperature). pABC was constructed by ligating together the blunted overhangs left after TthIIIISall digestion of pG lhAR.
Yeast Extracts and {3-Galactosidase Assays-Yeast cultures were grown in minimal medium with and without the agonist R1881 (a synthetic androgen stored in ethanol at -20°C) to log phase. The cells (typically from 10-ml cultures) were harvested at 3000 x g and washed once in extract buffer (20 mM Hepes-KOH, pH 7.4,150 mM NaCl, 1 mM EDTA, 10% glycerol, 5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 2 /-Lg/ml each aprotinin, chymostatin, leupeptin, and pepstatin). Cells were resuspended in 200 /-Ll of extract buffer in 0.5-ml Eppendorf tubes and broken with glass beads (0.45 /-L) in a mini-bead beater with a single 50-s burst on medium speed at 4°C. The tubes were pierced with a 25-gauge needle and the extract eluted by centrifugation into capless 1.5-ml centrifuge tubes. The protein content was quantitated by the method of Bradford (as described by Stoscheck (1990)). Assays for {3-galactosidase activity were performed using of 1 ml of protein extract diluted (typically to 1 /-Lg/ml) in Z buffer (60 mM Na 2HP04·7H20, 40 mM NAH 2P04·H20, 10 mM KCl, 1 mM MgS0 4·7H20, and 50 mM {3-mercaptoethanol. Reactions were started by the addition of 200 /-Ll of a 4 mg/ml solution of o-nitrophenyl {3-Dgalactopyranoside (in 0.1 Msodium phosphate buffer, pH 7), incubated for 15-60 min at 30°C, and terminated by the addition of 0.5 ml of 1 M Na 2C0 3. {3-Galactosidase activity was quantitated after measuring the absorbance of the reactions (performed in duplicate) at A 4 2 0 in a Milton Roy Spectronic Genesys 5 spectrophotometer. {3-Galactosidase units are defined as nM o-nitrophenol generated/min/mg extract using the molar extinction coefficient of o-nitrophenol (4500 at A 4 2 0 ; described by Miller (1972)).
Extracts for the Western blot shown in Fig. 2 were prepared as described above but using lysis buffer (50 mMTris-HCl, pH 7.5,1% SDS, 1 mMphenylmethylsulfonyl fluoride) instead of extract buffer. Also, the extracts were boiled prior to the quantitation of protein using the BCA assay (Pierce).
R1BBl Binding Assay-The binding ofR1881 to wild type and ydjl-151 yeast cells:': hAR was quantitated in a saturation binding assay as follows. Yeast cells from overnight cultures were adjusted to 2 X 10 6 cells/ml in fresh medium. One ml cultures were incubated at 30°C for 2 h prior to addition of saturating amounts (36-72 nM; n = 5) of [3HJR1881 (DuPont) :': 100-fold excess of unlabeled R1881. After an additional 2-h incubation, the cells were washed three times in water then lysed in 200 /-Ll oflysis buffer (20 mMHepes-KOH, pH 7.4, 1% SDS) using glass beads as described above. Extracts from these cells were quantitated for protein concentration and samples counted in a scintillation counter. The amount of specific [3HJR1881 bound was calculated by subtracting the counts from the samples containing the excess cold hormone (corresponding to the nonspecific binding) and converting them to pmollmg extract protein.
Northern Blot-RNA was extracted from growing yeast cultures according to the method of Schmitt et al. (1990) with the exception that three cycles of heating (at 65°C) and freezing (on dry ice) were per-formed. RNA was resolved on formaldehyde-agarose gels (1.2% according to Sambrook et al. (1989)) and transferred to Amersham Hybond-N membrane. Hybridizations were performed in Denhardt's solution (Sambrook et al., 1989) at 65°C using 32P-labeled probes generated by the random priming method of Feinberg and Vogel stein (1984). Filters were washed two times for 15 min in 2 X SSC, 0.1% SDS and two times for 15 min in 0.1 X SSC, 0.1% SDS at 65°C.
Western Blot-Western blots were performed after resolving protein extracts in SDS-polyacrylamide gels and transferring to nitrocellulose membrane (0.45 urn, Micron Separations) using a semi-dry apparatus (Bio-Rad), Filters were processed by standard methods; briefly, filters were washed in TTBS (20 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 0.05% Tween 20) and blocked for 16 h in TTBS plus 5% nonfat dry milk solution. Primary antibody incubations with AR52 (an IgG fraction at 0.9 JLg/ml) were for 1-2 h in 1 X phosphate-buffered saline, 3% bovine serum albumin, 0.05% Tween 20, 0.1% thimerosal. Filters were washed three times for 10 min in TTBS before incubation in secondary antibody (conjugated with horseradish peroxidase) at 1:10,000 dilution for subsequent detection by chemiluminescence (Renaissance, DuPont). Filters were washed three times for 5 min prior to treating with detection reagent.

RESULTS
In order to examine the role of S. eerevisiae Ydj1p in hAR activation, yeast strains were constructed that constitutively express the hAR gene from a 2-/Lm multi-copy plasmid. In addition, the E. coli laeZ gene, itself under control of androgen response elements was integrated into the yeast genome (at chromosome V, linked to the URA3 gene) and served as a reporter for hAR activation (Purvis et al., 1991). Strains containing multiple copies of the hAR gene and a single copy of the lacZ gene under control of hAR were constructed in both wild type (YDJ1) and mutant (ydjl-151) backgrounds (see Fig. 1 for graphic representation of strains and Table I for strain genotypes). The hAR gene was constitutively expressed using the promoter from the glyceraldehyde-3-phosphate dehydrogenase gene.
Characterization of hAR in Wild Type and ydjl-151 Mutant Yeast-The relative level of expression of the hAR gene in wild type and ydjl-151 backgrounds was determined by Northern blot analysis. In the experiment shown in Fig. 2A, the wild type strain exhibits 1.5-fold more hAR mRNA than the ydjl-151 mutant after normalization to the levels of actin mRNA. The level of hAR mRNA is 4.5-fold greater than was found in wild type cells expressing hAR from a single gene integrated into the yeast genome (data not shown). Note that two bands are observed in the lanes corresponding to hAR mRNA. The smaller of the two (denoted with an asterisk) corresponds in size to actin mRNA and may result from internal priming within hAR eDNA (Simental et al., 1991). Neither band is observed when the hAR probe is hybridized to RNA from similar strains not expressing hAR ( Fig. 2A, lanes 1 and 2).
The hAR protein was detected in both wild type andydj-151 mutant strains using a specific polyc1onal antibody in a Western blot experiment (AR52; Fig. 2B). The hAR protein expressed in yeast comigrates with the 118-kDa recombinant hAR protein expressed in insect cells (see Fig. 2B, lane 5; Wong et al. (1993». The level of hAR in the ydj-151 mutant was usually slightly higher than that found in the wild type (compare lanes 1 and 3 of Fig. 2B), but the steady state levels in both strains decreased when the cells were grown at 30°C with 100 nM R1881, a synthetic androgen. This appears to contrast with the effect of hormone on hAR levels in transfected COS cells, where treatment with hormone stabilized hAR and increased its half-life from 1 to 6 h at 37°C (Kemppainen et al., 1992). The level ofhAR mRNA was unaffected by growing either wild type or mutant cells in the presence of 100 nM R1881 (data not shown). Quantitative hormone binding studies (Fig. 2C) indicated hAR to have a similar capacity for R1881 in both wild type and ydjl-151 mutant strains, consistent with hAR being Roman numerals denote chromosome assignments.ydjl-2::HIS3 denotes a gene deletion allele where the wild type YDJI gene has been replaced by HIS3 on chromosome XIV. The mutantydjl-151 gene is integrated at the LEU2locus on chromosome III (Caplan et al., 1992a). The lacZ gene (under control of androgen response elements; ARE in the figure) was integrated at the URA3locus on chromosome V. Both strains have the hAR gene under control of the constitutive glyceraldehyde-3-phosphate dehydrogenase promoter (G3PDH) on a multi-copy (2 f-Lm) plasmid. folded correctly. The mean 1.6-fold increase in binding seen with the mutant strain (maximum difference observed was 3-fold) correlates with the increased steady state level of protein observed by Western blot (Fig. 2B). There was negligible specific binding of R1881 to cells not expressing bAR (Fig. 2C,  lanes 1 and 3).
Hormone-dependent Activation of hAR Is Defective in ydjl Mutant Strains-The ability of R1881 to activate bAR in wild type and mutant strains was determined as a function of lacZ gene induction by measuring J3-galactosidase present in soluble whole cell extracts. In the experiment shown in Fig. 3, liquid cultures of wild type and ydjl-151 cells were titrated with R1881 and lacZ reporter gene activity measured after 6 ( Fig.  3A) and 16 h (Fig. 3B). The level of J3-galactosidase protein induced in the wild type strain in the presence of 100 nMR1881 varied in range from 6· to over 20-fold above the background (uninduced) levels (average from seven independent experiments, induction ratio of 14.5). These induced levels were typically 2500-5000 J3-galactosidase units (nM o-nitrophenol! min/mg extract). By contrast, induced levels of J3-galactosidase R egulation of Androgen R ecept or Activation by Ydjlp in t he ydjI-I5 1 strai n was not more than 3-fold above backgro und, eve n at 100 mt R1881 after incubation times of6 (Fig.  3A ) or 16 h (Fig. 3B ) und er perm issive gro wth conditio ns for t his st ra in (30°C; Ca pla n et al., 1992a ). Northern blot analysis of la cZ mRNA revealed strong expression withi n 1 h of R1881 treatment in wild typ e bu t not mutant cells (data not shown), confirming th at th e defect obse rved in th e y dj I -I5 1 mutant reflects a n induction phen otype rather tha n labili ty in t he {:l-galactosidase reporter pr otein. Fu rtherm ore, constitu tive expr ession of lac Z resul ts in simila r levels of {:l-galactosidase protein in both wild type a nd mutant cells (see below a nd Fig.  5). Th e defective induction phen otype a lso a ppea rs to be genera l si nce a si mila r defect was obse rve d with th e ra t glucocorticoid recep tor expressed in th e ydj I -I5 1 stra in.f To confir m th at t he low induction ph enotype resulted from mutation in th e YDJ 1 gene, t he st ra in ACY45 was back -crossed to a wild typ e st rai n (JC2LZ, see Tabl e I) . Th e resulting diploid strain (ACY68) was sporu lated, and te tra ds wer e dissect ed to yield four haploids. Sinc e ACY45 carried both th e ydj 1 null a llele (ydjI -2::HI S3, referred to as ydj 1-2.}. hereafter) a nd ydj 1-151 all eles on differen t chromosomes (see Fig. 1), it was possible to isola te hapl oids, from tetra ds segregating as tetratypes, that contai ned one wild type (Y DJ ]), one heterozygous wild type (YDJ I a nd ydjl-15]), one mutant strain equiva lent to ACY45 (i.e. ydjI · I5 1 a nd ydjI -2.}. ), a nd one complete null stra in (ydj l-2 .}. ). Three colonies from ea ch strain (derived from t hree differ en t tetraty pes ) were a na lyzed for t heir ab ility to express th e la cZ reporter gene in t he pr esence of R188!. In each case (see Fig. 4), max imal respon se to hormone was observ ed in the wild typ e (YDJ ]) strain. Th e lowest activity was by th e ydj I -2.}. null st rain « 3-fold activation of the lacZ gene). Interm ediate levels of lacZ gene induction were obse rve d in stra ins expressing th e ydjI -15 1 all ele, eit he r by itself or in th e pr esence of the wild type gene . Thi s heterozygous stra in (i.e. YDJI a nd ydjl -15 ]) had a lower activity t han the wild ty pe strai n (YDJ ]), suggesting th a t the ydjI-I5 1 ph enotype is domin an t nega tive. Note a lso that background (uninduced) levels of {:l-galactos idase wer e on average 2-fold high er in y dj I· I5 1 a nd ydjI -2.}. st ra ins th an was obse rved in wild type stra ins (see also Fig. 5). Whether thi s reflects deregulation of hAR or a not he r aspect of lacZ gene transcription is uncl ear.
Similar re sults were record ed when th e original ACY45 mutant strain was transform ed by a plasmid overexpressin g wild type YDJ1. In these experime nts, overexpression of wild type YDJI in the ydjI-I51 stra in largely su ppressed th e low induction ph enotyp e of the mu tant a lone , stimulating hAR-dependent lacZ gene expression 2.7-fold above va lues ty pica lly obta ine d with the mu ta nt, but at levels that were still 72% of the wild type value (da ta not shown ). Overexpression of wild type YDJ 1 itself has little effect on hAR activation in t he wild type ACY44 st rai n, bu t a simila r st ra in overexpressing th e mutant ydj I· I5 1 gene exhibite d a n average 20% decrea se in {:l-galactosidase levels (data not shown).
Ydjlp Function s via th e hAR Hormone Binding Doma in-Hsp90 form s a compl ex with hAR via th e horm one binding domain a t th e C-te r mina l end of th e protein (Ma riovet et al ., 1992). To determine wheth er Ydjlp a lso function s via th is domain, a n experime nta l st rategy was devised to ta ke adva n-4 + + , lan e 2 ) pG1hAR , which expresses hAR from the constitutive glyceralde hyde-3-phos pha te dehyd rogen a se pr omoter . Duplic ate samples from the sa me gel wer e probed for acti n mRNA as shown. B , Western blot ana lysis. Pr otein extrac ts from strai ns ACY44 tlones 1 a nd 2) and ACY45 ilanes 3 a nd 4) as well as recombinan t hAR from baculovirus (B )-infecte d SF 9 cells tlane 5 ) wer e probed th e AR52 (\gG fracti on a t 0.9 ug/ml ), Strain s ACY44 a nd ACY45 were grown overn igh t in the abse nce Ilanes 1 a nd 3 ) or in the presence of 100 nMR1881 ilanes 2 a nd 4). Molecul ar size markers are indicated at left: 97, 68, a nd 45 kDa. Haploids from three complete tetrads (each a tetratype in A, B, and C) dissected from sporulated ACY68 diploid cells were tested for laeZ gene induction after overnight growth in the presence (lanes 2, 4,6, and 8) or absence (lanes 1,3,5, and 7) of 100 IlM RI88!. The genotypes for each strain are given below the graph and correspond to the strains listed in Table I as follows: YDJI (ACY72), YDJI andydjl-151 (ACY73),ydjl-151 andydjl-2" (ACY74), andydjl-2" (ACY75).
tage of the constitutive activation that occurs in hAR truncation mutants lacking the C terminus (Simental et al., 1991, Zhou et al., 1994. As with other steroid hormone receptors, loss of the steroid binding region relieves the negative regulatory function of this domain in the absence of ligand. For these experiments, truncated versions of hAR were expressed in yeast and their constitutive trans-activating function measured by the steady state levels of J3-galactosidase. If Ydj1p function is via the steroid binding domain, then its removal might also eliminate the need for Ydj1p action. If this occurred, then the defect manifest inydjl-151 might also be suppressed and the hAR truncation mutant would be similarly active in both wild type and mutant strains. If, however, Ydj1p function is via another domain still present in the truncation mutant, then the hAR trans-activation defect in ydjl-151 should still manifest. This would result in greatly different steady state levels of f3-galactosidase in the wild type and mutant strains. Two deletion mutants of bAR were constructed for these experiments: one in which the C-terminal259 amino acids of bAR (including the entire hormone binding domain) was deleted (hAR l -6 6 0 ) and a second where the hormone binding domain was replaced with the 78-amino acid C-terminal activation domain of the viral transcription factor VP-16 (termed ARVP-16). Wild type and mutant strains were transformed with multi-copy plasmids (Table II) encoding hAR 1 -660 and the ARVP-16 chimeric gene under control of the yeast glyceraldehyde-3-phosphate dehydrogenase gene promoter. The results of this experiment, shown in Fig. 5, reveal that hAR l -6 6 0 and ARVP-16 behave as high level constitutive activators in both wild type andydjl-151 strains. The steady state level of f3-galactosidase activity was at least lO-fold greater than was observed in either strain expressing fulllength bAR in the absence of hormone (the data for constitutive levels of hAR in wild type and ydjl-151 strains are comparable to the background levels found in other experiments shown in Figs. 3 and 4). The defect associated with the ydjl-151 mutation, therefore, is relieved when the hAR hormone binding domain is deleted. This is consistent with a specific role for Ydjlp in hAR activation through interaction with the hormone binding domain.

DISCUSSION
The results described in this paper are consistent with the Ydjlp molecular chaperone playing a role in the regulation of hAR expressed in yeast. The hAR protein was barely activated by hormone in theydjl-151 strain (and theydjl-26. null strain), yet was fully active after deletion of the steroid binding domain. Furthermore, since hAR had a similar binding capacity for hormone in both wild type andydjl-151 mutant strains, the role of Ydj1P in receptor activation is apparently independent of any function it may have in folding of nascent polypeptide chains (see below).
Ydjlp thus joins other molecular chaperones such as Hsp90 and Hsp70 that appear to function in hormone-regulated activation of steroid receptors via the hormone binding domain. Unlike these other Hsps, however, a dnaJ protein has not previously been described as a component of the 9 S complex nor involved in steroid-dependent activation. There is evidence, however, for unidentified factors present in rabbit reticulocyte lysates that function in the formation of Hsp70-Hsp90 complexes and Hsp90-hormone receptor complexes. One example is from the recent work by , who propose the existence of an Hsp70-Hsp90 complex forming factor. This is based on the observation that complex formation between Hsp70 and Hsp90 is stimulated by the presence of additional factors in rabbit reticulocyte lysate. Other studies from the laboratory of Pratt suggest that factors in addition to Hsp70 are required for heterocomplex assembly between Hsp90 and glucocorticoid receptor . The dissociation of Hsp90 from the progesterone receptor is energy-dependent and requires factors other than hormone (Kost et al., 1989 andSmith et al., 1992). Although a dnaJ protein was not identified in any of these studies, the known functions of dnaJ correlates well with the activities of these factors, that is, co-operation with Hsp70 in protein assembly and disassembly (Georgopoulos et al., 1990). If the proteins required for these events prove to be dnaJ proteins, then the function ofYdjlp in hAR activation might also involve assembly and/or disassembly of the receptor-Hsp90 complex.

Plasmids used in this study
Little is known of how Hsp90 assembles with steroid hormone receptors except that it appears to involve Hsp70 Smith et al., 1992) in a post-translational event. Several lines of evidence are consistent with this assembly being post-translational rather than co-translational. First, receptors that have been isolated after immunoadsorption from animal cell cytosols are competent for reconstitution with Hsp90 in rabbit reticulocyte lysates (see Pratt (1993) for review). Second, Hsp90 will only bind to full-length glucocorticoid receptors after in-vitro translation (Dalman et al., 1989). Third, Hsp70 and Hsp40 (a dnaJ protein) but not Hsp90 are associated with polysomes (Frydman et al., 1994).
The binding of dnaJ and Hsp70 to nascent polypeptide chains is thought to reflect the first step in a chaperone-mediated protein folding pathway (Langer et al., 1992;Hendrick et al., 1993;Frydman et al., 1994). This binding is followed by transfer of the nascent polypeptide chain to a chaperonin for folding. The sequential binding of molecular chaperones also occurs as polypeptides are imported into mitochondria (Manning-Krieg et al., 1991), and in the endoplasmic reticulum immunoglobulin light chains bind sequentially, first to Bip (an Hsp70 protein located in the lumen) and then to grp94 (a similarly located Hsp90 protein; Melnick et at. (1994)).
Whether Ydjlp participates with Hsp70 in similar events for assembly of hAR-Hsp90 complexes has yet to be addressed. However, a common link between Hsp90 and Ydjlp is via their specific association with Hsp70 sSA but not Hsp70 sSB subfamilies (Chang and Lindquist, 1994;. This specificity provides indirect evidence that Ydjlp could affect Hsp90 function via its interaction with Hsp70. The defect for hAR induction in the ydjl-151 strain might then be explained by the failure of the mutant ydjl-151 protein to assist in Hsp70-dependent assembly (or perhaps disassembly, see below) of the receptor-Hsp90 complex. This is supported by the previous observation that purified ydjl-151 protein was only 16% as effective as wild type Ydjlp for stimulating the ATPase activity of Hsp70 sSA1 (Caplan et al., 1992a).
Previous studies using yeast have established a physiological role for Hsp90 in the hormone-dependent activation of several steroid hormone receptors (Picard et al., 1990;Bohen and Yamamoto, 1993). In the study by Picard et at. (1990), decreasing levels of Hsp90 reduced the hormone inducible activation of the glucocorticoid, estrogen, and mineralcorticoid receptors. In a similar study, Xu and Lindquist (1993) discovered that a yeast mutant having substantially reduced levels of Hsp90 remained viable when pp60 vs r c is expressed, which results in lethality in wild type yeast cells (Brugge et al., 1987). This genetic study revealed a physiological basis for the pp6OV-s r c _ Hsp90 interaction that was previously observed in animal cells (Brugge, 1986). Recent genetic studies also revealed a role for Source Purvis et al. (1991) Purvis et al. (1991 This study This study This study URA3 LacZ TRPlhAR TRPI hAR 2f.L TRPI hAR l -6 6 0 2f.L TRPI AR-VPI6 2f.L Plasmid pPGKareLacZI pPGKhARI pGl-hAR pABC pARVP16 Ydj1p in the activation of pp60 vs r c in yeast cells, since mutation of YDJ1 also suppresses the lethal phenotype resulting from pp60 vs r c expression.i' In immunoprecipitation experiments using the ydjl-151 strain, much higher levels of Hsp90 were coimmunoprecipitated with antibodies specific to pp60 vsrc than were found for the wild type strain after inducible expression. These data appear to confirm that mutation in the YDJl gene affects the interaction ofHsp90 with other proteins. Whether this is true for hAR in the ydjl-151 mutant strain awaits further investigation.
It seems likely that Ydj1p function in hAR activation will be conserved in animal cells since a human counterpart, HDJ2 (47% identity), has recently been described (Chellaiah et al., 1993;Oh et al., 1993). As shown recently by Chang and Lindquist (1994), proteins that form a stable complex with Hsp90 are also conserved in yeast and animal cells. Whether all events in the hAR activation pathway are conserved in yeast, however, is open to question. For example, hormone stabilizes hAR in animal cells (Kemppainen et al., 1992), yet reduces steady state hAR levels in yeast (Fig. 2B). While the basis for this is unclear, such differences warrant a cautious interpretation when considering data obtained using yeast and extrapolating its significance to higher animal systems. The similarities between chaperone components, on the other hand, point to a conservation in the mechanism of activation as it relates to these proteins. This is supported by the finding of a fungus-like water mold, Achyla ambisexualis, which uses steroids as both pheromones and hormones. Significantly, the unliganded receptors for these steroids form 9 S complexes that contain Hsp90 (Riehl et al., 1985, Brunt et al., 1990.