The molecular chaperone Hsp90 mediates heme activation of the yeast transcriptional activator Hap1.

Hsp90 plays critical roles in the proper functioning of a wide array of eukaryotic signal transducers such as steroid receptors and tyrosine kinases. Hap1 is a naturally occurring substrate of Hsp90 in Saccharomyces cerevisiae. Hap1 transcriptional activity is precisely and stringently controlled by heme. Previous biochemical studies suggest that in the absence of heme, Hap1 is bound to Hsp90 and other proteins, forming a higher order complex termed HMC (high molecular weight complex), and is repressed. Heme promotes the disruption of the HMC and activates Hap1, permitting Hap1 to bind to DNA with high affinity and to stimulate transcription. By lowering the expression levels of wild-type Hsp90, using a highly specific Hsp90 inhibitor, and by examining the effects of various Hsp90 mutants on Hap1, we show that Hsp90 is critical for Hap1 activation by heme. Furthermore, we show that many Hsp90 mutants exert differential effects on Hap1 and steroid receptors. Notably, mutant G313N weakens Hsp90 steroid receptor interaction but strongly enhances Hsp90-Hap1 interaction and increases Hap1 resistance to protease digestion. Additionally, we found that a heme-independent Hap1 mutant still depends on Hsp90 for high activity. These experiments together suggest that Hsp90 promotes Hap1 activation by inducing or maintaining Hap1 in a transcriptionally active conformation.

The molecular chaperone Hsp90 is a highly conserved stressinduced protein that is ubiquitously expressed in almost all cells at a high level (1,2). It is required for viability in eukaryotes and plays critical roles under physiological conditions (2). Hsp90 is required for the proper functioning of many signal transducers such as steroid receptors and protein kinases (3). Substrates of Hsp90 are diverse. It is highly probable that Hsp90 promotes the actions of diverse substrates by different mechanisms. Even among nuclear hormone receptors, the modes of Hsp90 action are quite different. Although Hsp90 forms stable complexes with steroid receptors, it does not form complexes with vitamin receptors, but it is important for the activity of vitamin receptors (4,5). Within the family of steroid receptors, Hsp90 mutants also exert differential effects on different receptors (6,7). Nonetheless, previous biochemical and genetic studies (8,9) have suggested that Hsp90 probably plays a common dual role in steroid signaling-maintaining receptors in a repressed state in the absence of ligand and promoting ligand activation by keeping the receptors in a high ligand binding affinity conformation.
Genetic studies (8,9) in yeast have provided valuable insights into the function of Hsp90 in steroid signaling. However, this approach is somewhat compromised by the fact that many of these previously studied proteins, such as steroid receptors and tyrosine kinases, are not naturally occurring in yeast. Thus, the events observed in yeast may not exactly reflect the bona fide functions of Hsp90. Therefore, it is necessary to study the role of Hsp90 in signaling pathways with natural yeast substrates. Recent studies (10,11) have identified a few natural substrates of Hsp90 in yeast including the heme activator protein Hap1 and Ste11 of the pheromone-signaling pathway. In particular, Hap1 is a versatile transcriptional regulator whose activity is precisely and stringently controlled by heme concentrations (12). Hap1 mediates the effect of oxygen on gene expression in yeast. In response to heme, Hap1 binds to upstream activation sequences (UASs) 1 of numerous genes encoding functions required for respiration and for controlling oxidative damage and activates transcription of these genes (13,14). Hap1 also activates the transcription of the ROX1 gene encoding the Rox1 repressor, which represses the transcription of genes encoding anaerobic-specific functions such as ANB1 (15,16).
Hap1 contains 1483 amino acid residues and is composed of multiple functional domains and elements, such as the C6 zinc cluster and the dimerization element at the N terminus, which permit Hap1 to bind to DNA, the repression modules-1-3 (RPM1-3) and the heme-responsive motifs-1-7 (HRM1-7), which mediate heme regulation, and the activation domain at the C terminus (13,14,(17)(18)(19)(20)(21). RPMs and HRMs are two distinct classes of Hap1 elements essential for heme regulation (17,18). RPMs mediate Hap1 repression in the absence of heme; the deletion or disruption of any one of RPMs causes Hap1 to gain high levels of activity even in the absence of heme (17). HRMs, particularly HRM7, bind to heme and mediate the activation of Hap1 in response to heme (17,18). Furthermore, heme regulation of Hap1 appears to involve certain cellular proteins including the molecular chaperone Hsp90 (10). In the absence of heme, Hap1 is bound by Hsp90, Ydj1, and other proteins, forming a high molecular weight complex termed HMC (10,12,22). At 4°C, the HMC is able to bind DNA with low affinity in vitro as detected by electrophoretic mobility shift assays (10,22). When heme binds to Hap1, the HMC is disrupted, and Hap1 binds DNA with high affinity to activate transcription (10,12,19,23).
In this report, to test rigorously the functional importance of Hsp90 in heme regulation of Hap1 and to decipher the molecular mechanism by which Hsp90 promotes heme regulation, we carried out extensive functional and biochemical analyses of the effects of various Hsp90 mutants on Hap1 activity and on Hap1 heme responsiveness. We found that Hsp90 is critical for heme activation of Hap1. Defective Hsp90 function greatly diminishes Hap1 activation by heme. However, individual Hsp90 mutants exert different effects on Hap1 activity and on Hap1-Hsp90 interaction compared with those exerted on the activities of steroid receptors and on Hsp90-steroid receptor interaction. Our data also suggest that a heme-independent LexA-Hap1 mutant protein is still Hsp90-dependent. These results support the idea that Hsp90 promotes Hap1 activation by inducing or maintaining Hap1 in a transcriptionally active conformation.
␤-Galactosidase Assays-To measure ␤-galactosidase levels from the Hap1-driven reporter gene at various heme concentrations, yeast cells were transformed with the UAS1/CYC1-TATA (CYC1)-lacZ reporter plasmid. Cells were grown in synthetic complete medium containing 2% glucose with limiting amounts of heme precursor ␦-aminolevulinate (2 g/ml) to an optical density (A) of approximately 1.0 and diluted 2-fold in medium containing various concentrations of heme analogue, deuteroporphyrin (dpIX). Cells were then grown for 7 h and harvested for determination of ␤-galactosidase activity as described previously (10,18). ␤-Galactosidase levels from the LexA operator-lacZ reporter (29) and the basal reporter pLG178 (30) were similarly measured. ␤-Galactosidase levels from the GR-driven reporter gene at various deoxycorticosterone concentrations were measured as described previously (25).
Preparation of Yeast Extracts and Electrophoretic Mobility Shift Assays-Extracts were prepared according to previously established protocols (10,22). Yeast cells were grown to A ϭ 0.8, harvested, and resuspended in three packed cell volumes of buffer (20 mM Tris, 10 mM MgCl 2 , 1 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 0.3 M NaCl, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml of pepstatin, 1 g/ml of leupeptin). Cells bearing the GAL1-driven HAP1 expression vector (28) were induced by 2% galactose for 5-6 h prior to collection as described previously (17). Cells were then permeabilized by agitation with four packed cell volumes of glass beads, and extracts were collected. Protein concentrations were subsequently determined by Bradford assay.
DNA binding reactions were carried out in a 20-l volume with 5% glycerol, 4 mM Tris, pH 8.0, 40 mM NaCl, 4 mM MgCl 2 , 2 ng/l of heme, 10 mM dithiothreitol, 3 g of salmon sperm DNA, 10 M ZnOAc 2 , and 300 g/ml of BSA as described previously (10,18,22). Approximately 0.01 pmol of labeled DNA and 20 g of protein extracts were used in each reaction. The Hap1 antibody was included in reaction mixtures as described previously (31). The reaction mixtures were incubated at 4°C for 1 h and then loaded onto 4% polyacrylamide gels in one-third of Tris borate/EDTA for gel electrophoresis at 4°C. Radioactivity of the interested bands was visualized and quantified by using the PhosphorImager system (Molecular Dynamics).
Protease Digestion and Pull-down Experiments-Whole cell extracts prepared from cells expressing Hap1 and wild-type Hsp90 or mutants were incubated with elastase or other proteases for 30 min at room temperature. The reactions were then stopped by the addition of SDSpolyacrylamide gel-loading dye and subjected to electrophoresis and Western blotting analysis.
For DNA pull-down assays, extracts prepared from cells expressing Hap1 and wild-type or mutant Hsp82 were incubated with streptavidinconjugated magnetic beads (DYNAL) prebound with the biotinylated Hap1 binding site under the same condition as that for electrophoretic mobility shift assays as described above. The beads were extensively washed and boiled in SDS gel-loading buffer to release the bound proteins (32). Proteins were then analyzed by SDS-PAGE followed by Western blotting analysis.

Defective Hsp90 Function Reduces Hap1 Activation by
Heme-In Saccharomyces cerevisiae, Hsp90 is encoded by two functionally equivalent and independent genes, HSC82, which is constitutively expressed at a high level and induced only 2-3-fold by stress, and HSP82, which is constitutively expressed at a low level and strongly induced by stress (2). Hsp82 and Hsc82 share a 97.6% identity. Previously, we have shown that these Hsp90 proteins are components of the high order Hap1 complex, HMC, formed in the absence of heme (10). To rule out the possibility here that the interaction of Hsp90 with Hap1 is an artifact and to ascertain whether and how this interaction is important for heme regulation, we examined the effects of defective Hsp90 function on Hap1 activity. First, we examined the effect of low levels of Hsp90 on Hap1 activity at a range of heme concentrations in a strain carrying deletions of the wild-type HSC82 and HSP82 genes and expressing Hsp90 from the inducible GAL1 promoter (Fig. 1A). Low Hsp90 expression levels were achieved by culturing cells in the presence of 1% galactose and 1% glucose. Wild-type expression levels were achieved by culturing cells in the presence of 2% galactose (8,10). When Hsp82 and Hsc82 are under the control of their natural promoters, Hap1 activity is not reduced by the presence of glucose (17,30).
At heme concentrations that permit Hap1 activation, Hap1 activity was greatly reduced in cells (1% galactose) expressing low levels of Hsp90 compared with cells (2% galactose) expressing high levels of Hsp90 (Fig. 1A). However, Hap1 remained repressed in the absence of heme. We confirmed Hsp90 protein levels in these cells by Western blotting (see Fig. 1B). Furthermore, in wild-type cells in which Hsc82 and Hsp82 are expressed from their own promoters, Hap1 activity was somewhat higher in cells grown in medium containing 1% glucose and 1% galactose than in cells grown in medium containing 2% galactose (Fig. 1C). This finding shows that the difference in the galactose/glucose ratio of media did not contribute to the reduction of Hap1 activity caused by low Hsp90 expression levels. Similar results were obtained when we measured Hap1 activity in a strain that expresses a low level of Hsp90 in normal medium due to a mutant promoter (24) using the LEP-HSP82 expression plasmid (data not shown).
To provide an independent method for inhibiting Hsp90 function, we employed an Hsp90-specific inhibitor in wild-type cells (Fig. 1D). Macbecin I (33) is one of the ansamycin antibiotics. This family of inhibitors includes geldanamycin, which is more commonly used in mammalian cells, and Macbecin I, which is more effective in yeast. Both bind to Hsp90 with a high degree of specificity and inhibit its activity (34,35). They are frequently used to identify specific substrates of Hsp90 (35,36). Macbecin I greatly reduced Hap1 activity (Fig. 1D) at heme concentrations that normally activate Hap1, but did not affect Hap1 repression in the absence of heme. The effect of Macbecin I on Hap1 is similar to its effect on steroid receptors and on the neuronal nitric-oxide synthase (33,37). These results strongly suggest that Hsp90 function is critical for Hap1 activation by heme.

Hsp90 Mutants Exert Differential Effects on Hap1 and Steroid
Receptors-To further ascertain the role of Hsp90 in heme regulation of Hap1 and to gain insights into the mode of Hsp90 action, we examined the effects of three sets of Hsp90 mutants (6, 7, 25) on Hap1 activity. The first set includes four mutants with point mutations in Hsp90, A576T/R579K, E431K, T525I, and G313N (6). The second set of mutants examined include three deletion mutants, 1-704, 1-685, and ⌬211-259 (7). The third set of mutants includes six mutants with point mutations, G170D, A41V, E381K, A587T, G313S, and T101I (25). G170D is a classical temperature-sensitive mutant that appears to be near wild type at 25°C or at lower temperatures but completely loses its activity at 34°C or at higher temperatures (25).
We examined the effects of these mutants on Hap1 activity at various heme concentrations in cells expressing wild-type or mutant Hsp90 alone (Fig. 2, A-D). The first set of mutants exerts varying effects on glucocorticoid receptor (GR), progesterone receptor, estrogen receptor, and mineralocorticoid receptor, but G313N greatly reduced the activities of these receptors (6), whereas T525I moderately reduced GR activity (38). Our data ( Fig. 2A) show that G313N greatly reduced the extent of Hap1 activation by heme. At heme concentrations that permit Hap1 activation, Hap1 activity in cells expressing G313N was much lower than that in cells expressing wild-type Hsp90. T525I, A576T/R579K, and E431K had no significant effect beyond the variations that can be contributed to standard deviations ( Fig. 2A). The second set of mutants does not have a significant effect on GR, estrogen receptor, and progesterone receptor (7). Curiously, we found that mutant ⌬211-259 enhanced Hap1 activity, whereas mutants 1-685 and 1-704 appeared to slightly reduce Hap1 activity (Fig. 2B).
The difference between GR and Hap1 became more apparent when the third set of Hsp90 mutants was analyzed (Fig. 2, C-E). At permissive temperatures, GR activity is most severely diminished by G313S, whereas it is moderately diminished by T101I and A587T (see Fig. 2C and Table I) (25). By contrast, Hap1 activity was greatly diminished by A587T and T101I and was moderately reduced by G313S ( Fig. 2D and Table I). E381K moderately reduces both Hap1 and GR activities (Fig. 2, C and D). As expected, G170D had little effect on Hap1 activity at 28°C (Fig. 2D) but strongly reduced Hap1 activity at 34°C at all tested heme concentrations, which permit Hap1 activation (Fig. 2E). The differential effects of Hsp90 mutants on Hap1 and GR are summarized in Table I. G313N is also shown for comparison.
To rule out the possibility that the reduction of the Hap1driven reporter activity is not caused by nonspecific effects of Hsp90 mutants on transcription from the reporter system, we showed that mutants that severely reduced Hap1-driven reporter activity did not significantly affect the basal TATA (CYC1)-lacZ reporter activity (Fig. 2F). Likewise, the activity of the Hap2⅐Hap3⅐Hap4⅐Hap5 complex-driven UAS2up1-TATA (CYC1)-lacZ reporter was not considerably affected by Hsp90 mutants that greatly reduce Hap1 activity (data not shown). In addition, the fact that Hsp90 mutants affect Hap1 and GR differentially suggests that their effects on Hap1 are specific.
To determine whether the effects of Hsp90 mutants on Hap1 activity are because of changes in Hap1 expression, we measured the activity of the HAP1 promoter using a reporter gene fusion. The activity of this promoter was unaffected by different Hsp90 mutants (data not shown). We next examined Hap1 DNA binding activity in extracts from wild-type cells or cells expressing various Hsp90 mutants. Unfortunately, Hap1 protein levels in these strains expressed from the chromosomal gene are not high enough for detection by Western blotting analysis. To measure Hap1 accumulation, we took advantage of previous data showing that Hap1 levels detected by electrophoretic mobility shift assays in the presence of heme and excess DNA are consistent with those detected by Western blotting (17,18). The identity of the Hap1⅐DNA complex was verified by supershift with an anti-Hap1 antibody (31) (see Fig.  3A, lanes 1, 3, 5, 7, and 9 for examples). Pre-immune serum did not shift the Hap1-DNA band, and this band is absent in cells with the HAP1 gene deleted (data not shown) (for review see Ref. 31). Clearly, the extracts from cells expressing various mutants (Fig. 3, A and B) showed that levels of Hap1⅐DNA complexes formed in vitro were similar or even higher than those of wild-type cells. These results suggest that the effects of Hsp90 mutations in diminishing Hap1 activation were not caused by lowered Hap1 protein levels or DNA binding capacities. However, the level of Hap1 protein expressed from the chromosomal gene is too low to allow further analysis of the effects of Hsp90 mutants on the HMC by pull-down assay or even by electrophoretic mobility shift assays.
Mutant G313N Enhances Hap1-Hsp90 Interaction and Increases Hap1 Resistance to Elastase Digestion-To carry out biochemical analyses of Hsp90 mutants' effects on Hap1, we produced extracts from cells expressing wild-type or mutant Hsp90 and high levels of Hap1 from the GAL1 promoter (10, 17). Using these extracts, we examined Hap1 protein levels in  ). A, the effect of a low Hsp90 expression level on Hap1 activity. Cells bearing deletions of HSP82 and HSC82 genes and expressing Hsp90 from the GAL1 promoter were grown in the presence of 2% galactose (2% Gal) for high Hsp90 expression levels or 1% galactose and 1% glucose (1% Gal) for low Hsp90 expression levels, and ␤-galactosidase activities were detected. B, Western blot showing Hsp90 protein levels in cells grown in medium containing 2% galactose or 1% galactose and 1% glucose as shown in A. C, Hap1 activity in wild-type cells grown in medium containing 2% galactose or 1% galactose and 1% glucose. D, the effect of the Hsp90-inhibiting drug, Macbecin I, on Hap1 activity. ␤-Galactosidase activities were detected from wild-type cells with intact HSP82 and HSC82 genes, which were grown in the absence (None) or presence of 25 M Macbecin I (Mac I). The plotted data are averages from at least three independent transformants. The standard deviations are not shown for clarity, but they were within 30%. cells expressing wild-type or mutant Hsp90 by Western blotting (Fig. 4). We found that when overexpressed, Hap1 was less stable in extracts prepared from cells expressing certain Hsp90 mutants including T525I, G313S, and A587T (Fig. 4). The difference between the stability of Hap1 under these conditions and that under the conditions used in Figs. 2 and 3 might be attributed to different growth conditions and/or different Hap1 expression levels. Hap1 is a large protein with 1483 amino acid residues and is easily susceptible to degradation (see Refs. 17 and 18 for examples). The bands below the main band were likely caused by degradation (Fig. 4). However, even in these extracts shown in Fig. 4, many Hsp90 mutants did not significantly affect Hap1 protein levels. More importantly, two mutants, G313N and T101I, which most severely affected Hap1 activity (Fig. 2, A and C), had only minor effects on Hap1 protein levels under all tested conditions (Figs. 3 and 4). In cells expressing the hyperactive ⌬211-259, the Hap1 protein level appeared to be slightly higher than that in cells expressing wild-type Hap1 (Figs. 3B and 4B).
To gain insights into how Hsp90 mutants affect Hap1-Hsp90 interaction and thus Hap1 activation by heme, we analyzed the effects of G313N, T101I, and other mutants on Hap1-Hsp90 interaction. We initially examined the effects of Hsp90 mutants on the formation of Hap1 DNA binding complexes. In the absence of heme, Hap1 is bound by Hsp90 and other proteins forming a high molecular weight complex (10,22,39). Unlike the steroid receptor-Hsp90 complexes, the HMC is able to bind DNA with low affinity, although the DNA-bound HMC is detected only at 4°C (10,22). When the heme concentration increases, the HMC is gradually transformed into Hap1-dimeric complexes. Hap1-dimeric complexes may also form in the absence of heme when Hap1 is overexpressed and certain non-Hap1 components of the HMC are titrated out (10, 22, 40).
We found that none of the Hsp90 mutants had a major effect , and T101I, on GR activity. Note that mutant names shown on the right are listed in the same order as the activity curves. D, the effects of six Hsp90 mutants, G170D, A41V, E381K, A587T, G313S, and T101I, on Hap1 activity. Note that mutant names shown on the right are listed in the same order as the activity curves. E, the effect of G170D on Hap1 activity at 34°C. If not indicated, cells were generally grown at 28°C. In A-E, the plotted data are averages from at least three independent transformants. The standard deviations are not shown for clarity, but they were within 30%. F, the effects of G313N, A587T, G170D, and T101I on the basal transcriptional activity. Extracts were prepared from cells bearing the pLG178 TATA (CYC1)-lacZ basal reporter (30), and ␤-galactosidase activities were measured and plotted in nmol/min/mg protein.

TABLE I Summary of the effects of Hsp90 mutants on GR, Hap1, and Ste11
Shown here are the effects of G313N and Hsp90 mutants that exert differential effects on GR and Hap1. ϩϩϩ indicates wild-type levels of Hap1 or GR or Ste11 activity; ϩϩ indicates moderately reduced levels of activity; ϩ indicates severely reduced levels of activity; and ϩϩϩϩ indicates enhanced activity. The effects of G313N and T525I on Ste11 are also shown for comparison. on HMC formation or Hap1 DNA binding (Fig. 5). With the exception of G313S, the intensity of HMC and of dimeric Hap1 complexes formed in the presence of other Hsp90 mutants was similar to that formed in the presence of wild-type Hsp90 (Fig.  5, B and C). Together with functional data (Fig. 2), these results show that mutants G313N and T101I are the most defective in heme activation of Hap1 and that the defect is not attributed to reduced Hap1 protein levels or reduced DNA binding activity. Furthermore, analysis of HMCs and dimeric complexes formed at various heme concentrations showed that none of the Hsp90 mutants affected the heme concentration required for the disruption of the HMC and the formation of the dimeric complex (data not shown and Fig. 5A). This result suggests that these Hsp90 mutants do not affect heme binding by Hap1. Because G313N and T101I greatly reduced heme-activated Hap1 activity but had little effect on Hap1 protein levels under all tested conditions, we further examined their effects on Hap1-Hsp90 interaction by pull-down experiments (Fig. 6). Wild-type Hsp90, 1-704, and ⌬211-259 were also examined for comparison. DNA pull-down experiments were carried out in the absence of heme by incubating a biotinylated Hap1 binding DNA site with extracts prepared from cells expressing high levels of Hap1 and wild-type Hsp90 or an indicated mutant (see "Materials and Methods"). The bands below the top band were most probably Hap1 degradation products (Fig. 6), which accumulated because of extended incubation time during pull-down assays. Note that Hap1 is a large protein and is easily susceptible to degradation. When a mutated Hap1 binding DNA site was used in the pull-down experiments, the amounts of bound Hap1 and Hsp90 proteins were undetectable (data not shown). Fig. 6 shows that similar levels of Hap1 pulled down approximately 10-fold more Hsp90 in extracts from cells expressing G313N (see lanes 2 and 4) than those expressing wild-type Hsp90 (see lanes 1 and 3). Also, similar levels of Hap1 pulled down similar levels of Hsp90 in extracts from cells expressing 1-704 (see lanes 6 and 9) and those expressing wild-type Hsp90 (see lanes 7 and 10). Hap1 pulled down a slightly higher level of Hsp90 in extracts from cells expressing ⌬211-259 (see lanes 5 and 8) than those expressing wild-type Hsp90 (see lanes 7 and 10). However, similar levels of Hap1 pulled down similar levels of Hsp90 in extracts from cells expressing T101I (see lanes 12 and 14) as those expressing wild-type Hsp90 (see lanes  11 and 13).
These results show that G313N strongly enhanced Hap1-Hsp90 interaction, ⌬211-259 slightly enhanced Hap1-Hsp90 interaction, and T101I and 1-704 did not noticeably affect Hap1-Hsp90 interaction. Pull-down experiments using His 6 -tagged Hap1 also yielded the same results (data not shown). Although the HMC is disrupted by heme as detected by electrophoretic mobility shift assays, Hap1 can still interact loosely with Hsp90 even in the presence of heme (for review see Ref. 10), and the enhancement of Hsp90-Hap1 interaction by G313N persisted in the presence of heme (data not shown). Strikingly, the effect of G313N on Hap1-Hsp90 interaction contrasts strongly with its effect on GR-Hsp90 interaction, because G313N significantly weakens GR-Hsp90 interactions (38).
To further dissect the mode of Hsp90 action in heme regulation, we examined the effects of G313N and T101I on Hap1 conformation using protease digestion. Equal amounts of cell   FIG. 3. Hap1-DNA complexes formed in the presence of wild-type or mutant Hsp90. Extracts were prepared from cells expressing wild-type or mutant Hsp90 and Hap1 from the chromosomal gene at a low level. Note that this level is too low to allow the detection of the HMC, because the DNA binding affinity of the HMC is very low (22,40). Thus, all reactions shown here were done in the presence of heme. The denoted DNA⅐Hap1 complex is the Hap1 dimeric complex formed in the presence of heme. In A, extracts prepared from cells expressing wild-type Hsp90 (lanes 9 and 10), A576T/R579K (lanes 7 and 8), T525I (lanes 5 and 6), G313N (lanes 3 and 4), or E431K (lanes 1 and 2) mutant protein were incubated with radiolabeled DNA in the presence (lanes 1, 3, 5, 7, and 9) or absence (lanes 2, 4, 6, 8, and 10) of an anti-Hap1 antibody (10) and with 2 ng/l of heme. In B, extracts prepared from cells expressing wild-type Hsp90 (lane 10) and the indicated mutants in lanes 1-9 were incubated with radiolabeled DNA in the presence of 2 ng/l of heme. These DNA-Hap1 bands were also supershifted by the anti-Hap1 antibody but are not shown for brevity. The reaction mixtures were analyzed on a 4% non-denaturing polyacrylamide gel. The positions of the Hap1⅐DNA and Hap1⅐DNA-antibody complexes are marked.

FIG. 4. Hap1 protein levels in cells expressing Hsp90 mutants.
Extracts were prepared from cells expressing wild-type or the indicated mutant Hsp90 and high levels of Hap1 from the GAL1 promoter as described previously (10). A, Western blots showing Hap1 protein levels in cells expressing wild-type (wt) or one of the Hsp90 mutants as shown in Fig. 2A. B, Western blots showing Hap1 protein levels in cells expressing wild-type Hsp90 or one of the deletion mutants as shown in Fig. 2B. C, Western blots showing Hap1 protein levels in cells expressing wild-type or one of the Hsp90 mutants as shown in Fig. 2C. For each separate blot shown here, extracts containing equal amounts of total protein were loaded in each lane with the exception of A587T. Specifically, 100 g of extracts prepared from cells expressing A587T were loaded, whereas only 30 g extracts prepared from cells expressing wild-type Hsp90 or other mutants were loaded. Note that we included a wild-type control for each blot. The intensity of Hap1 in different blots varied because of different exposure times. These experiments were repeated three times. extracts prepared from cells expressing Hap1 and wild-type Hsp90 or G313N or T101I were treated with varying amounts of elastase. Elastase cleaves proteins preferentially at peptide bonds involving neutral amino acids and thus should fully hydrolyze Hap1 when its amount is sufficient. As shown in Fig.  7A, Hap1 was much more resistant to elastase digestion in extracts from cells expressing G313N than those expressing wild-type Hsp90. The increased Hap1 resistance to elastase digestion by G313N also persisted in the presence of heme. These results together with those of pull-down experiments suggest that G313N interacts with Hap1 strongly and is unable to promote or maintain Hap1 in a conformation necessary for activation. In contrast, Hap1 sensitivity to elastase was largely unaltered by mutant T101I (Fig. 7B), consistent with the result that Hap1-Hsp90 interaction was unaffected by T101I (Fig. 6). These results show that Hsp90 mutants affect Hap1 activity in different manners.
A Heme-independent Hap1 Mutant Still Depends on Hsp90 for High Activity-To further ascertain whether Hsp90 promotes Hap1 heme binding, we decided to examine the effect of defective Hsp90 function on a heme-independent Hap1 mutant. We believe that if Hsp90 promotes Hap1 activation only by facilitating heme binding, then heme-independent Hap1 mutants that cannot bind heme should be Hsp90-independent. Previously, we made a LexA-Hap1 fusion containing LexA residues 1-202 and Hap1 residues 946 -1483 (Fig. 8) (23). The transcriptional activity of this protein is stimulated by heme by 2-3-fold (Fig. 8) as shown previously (23). When the critical Cys residue in the HRM7 motif is mutated to Ala, the fusion protein becomes completely heme-independent (Fig. 8), because the mutation completely abolishes heme binding (23). Curiously, the LexA-Hap1Ala mutant activity is much higher than that of LexA-Hap1 in the wild-type cells, suggesting that HRM7 may play a role in the general heme-independent repression of Hap1. We detected the activities of LexA-Hap1 and LexA-Hap1Ala in wild-type and LEPHsp82 cells in which Hsp90 is expressed at a low level from a mutant promoter (24). We used LEPHsp82 cells, not cells expressing Hsp90 mutants with point mutations (Fig. 2), because a low expression level is more appropriate for examining Hsp90 dependence, whereas point mutants may exert mutation-specific and/or gain-of-function effect on a Hap1 mutant. Because the LEPHsp82 strain with the HEM1 gene deleted is very sick and may cause general reduction of transcriptional activity, we used LexA-GCN4 as a control. Indeed, LexA-GCN4 activity was reduced by 6 -8-fold in LEPHsp82 cells (Fig. 8). As expected, the activity of LexA-Hap1 was also greatly reduced in LEPHsp82 cells (Fig. 8).
Notably, LexA-Hap1Ala activity was, however, reduced by approximately 40-fold in LEPHsp82 cells (Fig. 8). The data suggest that a low Hsp90 expression level causes a specific reduction in LexA-Hap1Ala activity greater than that caused by its general effects on transcription as indicated by the reduction of LexA-GCN4 activity. These results suggest that Hsp90 can affect Hap1 transcriptional activity independently of heme FIG. 5. The effects of Hsp90 mutants on Hap1 DNA binding complexes. Extracts were prepared from cells expressing wild-type or the indicated mutant Hsp90 and high levels of Hap1 from the GAL1 promoter. A, DNA binding complexes formed in extracts prepared from cells expressing the Hsp90 mutant G313N. Extracts prepared from cells expressing wild-type Hsp90 (wt) or G313N were used in DNA binding reactions in the absence (lanes 3 and 6) or presence of 0.5 ng/l (lanes 2 and 5) or 2 ng/l (lanes 1 and 4) of heme. B, DNA binding complexes formed in extracts prepared from cells expressing Hsp90 deletion mutants. Extracts prepared from cells expressing wild-type Hsp90 (lanes 1 and 2), ⌬211-259 (lanes 3 and 4), 1-704 (lanes 5 and 6), or 1-685 (lanes 7 and 8) were used in DNA binding reactions in the absence (lanes 2, 4, 6, and 8) or presence of 2 ng/l (lanes 1, 3, 5, and 7) of heme. C, DNA binding complexes formed in extracts prepared from cells expressing Hsp90 mutants as shown in Fig. 2C. DC, dimeric complex. These experiments were repeated at least three times.
FIG. 6. Analysis of Hap1-Hsp90 interaction by DNA pull-down assay. Extracts prepared from cells expressing wild-type or the indicated mutant Hsp90 and high levels of Hap1 were incubated with streptavidin-conjugated magnetic beads (DYNAL) prebound with the biotinylated wild-type Hap1 binding site. The beads were extensively washed and boiled in SDS gel-loading buffer to release the bound proteins (32). When a mutated Hap1 binding site was used in the pull-down assay, the amounts of bound Hap1 and Hsp90 proteins were undetectable (data not shown). Hap1 in lanes 1 and 2 was detected by an enhanced Western blotting kit. These experiments were repeated at least twice.
binding and support the idea that Hsp90 is critical for promoting Hap1 conformational changes necessary for transcriptional activation or maintaining Hap1 in a transcriptionally active conformation.

DISCUSSION
In this report, by lowering wild-type Hsp90 expression, by using a specific Hsp90 inhibitor, and by analyzing the effects of various Hsp90 mutants on Hap1, we show that Hsp90 is critical for Hap1 activation by heme. Our results from analyses of Hsp90 mutants indicate that Hsp90 mutants affect Hap1 and GR in different manners and suggest that Hsp90 promotes Hap1 activation by inducing Hap1 conformational changes or maintaining an active Hap1 conformation independently of heme binding.
The Modes of Hsp90 Action in Heme and Steroid Signaling Differ-Hsp90 promotes the actions of diverse substrates most probably through diverse mechanisms. Even in the regulation of nuclear hormone receptors, Hsp90 appears to act somewhat differently in the regulation of different receptors. For example, Hsp90 forms stable complexes with steroid receptors but not vitamin receptors, although it is functionally important for the activities of both steroid and vitamin receptors (1,4,5,9,41). Furthermore, at least five Hsp90 co-chaperones exist (42), and different sets of co-chaperones are present in at least some of the different substrate-Hsp90 complexes tested (3,42). Thus, it is not surprising that Hsp90 may promote heme activation of Hap1 by a mechanism distinct from that governing Hsp90 action in steroid signaling.
Three lines of evidence support the idea that the mode of Hsp90 action in promoting Hap1 activation is distinct from that in promoting the activation of steroid receptors. First, many Hsp90 mutants affect the activities of steroid receptors and Hap1 in different manners. As shown in Table I, five  mutants, A597T, G313S, T101I, T525I, and ⌬211-259, exert differential effects on Hap1 and GR. Notably, Ste11 like Hap1 is also severely affected by G313N but not T525I. Second, the effect of G313N on Hap1-Hsp90 interaction contrasts strongly with its effect on GR-Hsp90 interaction. G313N significantly weakens the physical interaction between GR and Hsp90 (38) but strongly enhances the interaction between Hap1 and Hsp90 (Figs. 5 and 6). Third, the heme-independent LexA-Hap1Ala still depends specifically on Hsp90 for high transcriptional activity.
How does Hsp90 promote Hap1 activation? In the case of steroid receptors, previous studies (38,43) have suggested that Hsp90 promotes the activation of steroid receptors by maintaining steroid receptors in a high ligand binding affinity conformation, thereby promoting ligand binding and activation. In the case of Hap1, we cannot measure in vivo Hap1 heme binding affinity because of technical difficulties. However, an analysis of HMCs and dimeric complexes formed at various heme concentrations showed that Hsp90 mutants did not alter the heme concentrations required for the transformation of the HMC to dimeric complexes ( Fig. 5 and data not shown), suggesting that Hsp90 does not affect Hap1 heme binding affinity (17). In addition, the fact that Hsp90 mutants exert different effects on Hap1 and GR (Table I) supports the idea that Hsp90 does not promote heme activation of Hap1 by facilitating heme binding. Furthermore, even the heme-independent LexA-Hap1 mutant is Hsp90 dependent. Together, these results strongly suggest that Hsp90 promotes Hap1 activation in ways that are independent of heme binding. Nonetheless, these results do not exclude the possibility that Hsp90 may also promote heme binding by Hap1.
Mutant G313N may provide insights into how Hsp90 promotes heme activation. G313N enhanced Hap1-Hsp90 interaction and caused an increased resistance of Hap1 to elastase (Fig. 7A), which persisted even in the presence of heme. G313N might adopt an altered conformation, which leads to a stronger interaction with Hap1 whether or not heme is present. Consequently, Hap1 is unable to initiate conformational changes necessary for activation upon heme binding or to maintain FIG. 7. A, G313N renders Hap1 more resistant to elastase digestion. Extracts prepared from cells expressing G313N or wild-type Hsp90 and high levels of Hap1 were treated with increasing amounts (0 -50 ng) of elastase for 30 min. B, T101I does not alter Hap1 sensitivity to elastase digestion. Extracts prepared from cells expressing T101I or wild-type Hsp90 and high levels of Hap1 were treated with increasing amounts (0 -60 ng) of elastase for 30 min. The treated proteins were analyzed on a SDS-polyacrylamide gel, transferred to polyvinylidene difluoride membrane, and probed with an antibody against Hap1 residues 1-171 (10). For controls, reactions with extracts from cells expressing wild-type and mutant Hsp90 were done simultaneously in A or B.