Interaction between the N-terminal and Middle Regions Is Essential for the in Vivo Function of HSP90 Molecular Chaperone*

At the primary structure level, the 90-kDa heat shock protein (HSP90) is composed of three regions: the N-terminal (Met1–Arg400), middle (Glu401–Lys615), and C-terminal (Asp621–Asp732) regions. In the present study, we investigated potential subregion structures of these three regions and their roles. Limited proteolysis revealed that the N-terminal region could be split into two fragments carrying residues Met1 to Lys281 (or Lys283) and Glu282 (or Tyr284) to Arg400. The former is known to carry the ATP-binding domain. The fragments carrying the N-terminal two-thirds (Glu401–Lys546) and C-terminal one-third of the middle region were sufficient for the interactions with the N- and C-terminal regions, respectively. Yeast HSC82 that carried point mutations in the middle region causing deficient binding to the N-terminal region could not support the growth of HSP82-depleted cells at an elevated temperature. Taken together, our data show that the N-terminal and middle regions of the HSP90 family protein are structurally divided into two respective subregions. Moreover, the interaction between the N-terminal and middle regions is essential for the in vivo function of HSP90 in yeast.

This assembly process of the HSP90-substrate protein complex requires ATP (14,15), which induces a conformational change in HSP90 (16 -18). Recently, it was demonstrated that HSP90 is capable of linking substrates for degradation by the ubiquitin-proteasome pathway by cooperating with the E3 ligase carboxyl terminus of HSC70-interacting protein (CHIP) (19 -21). Thus, HSP90 may play a central role in deciding the fate of proteins, refolding or degradation.
HSP90 family proteins are composed of three regions at the primary structure level (22,23). In the present study using human HSP90␣, we denote the N-terminal region, Met 1 -Arg 400 , as Region A; the middle region, Glu 401 -Lys 615 , as Region B; and the C-terminal region, Asp 621 -Asp 732 , as Region C (23). The N-terminal domain (residues 9 -232) defined as the ATP/geldanamycin-binding region (24,25), the tertiary structure of which has been clarified (24 -26), corresponds to the N-terminal half of Region A. Prodromou et al. (25) observed a dimeric crystallographic structure of the N-terminal ATP-binding domain of yeast HSP90. Subsequently, they showed an ATP-dependent dimerization of the N-terminal domain independent of the C-terminal dimeric region (27).
Region B and Region C mediate dimerization of the HSP90 family proteins; Region B of one subunit is associated with Region C of another subunit in an antiparallel fashion (22). Electron microscopy showed that an HSP90 dimer consists of four linearly arranged globules (18), and the N-and C-terminal immunogenic sites (23) were localized in the terminal and interior globules, respectively (28).
To accomplish the molecular function of HSP90, each region may have additional roles that should be unveiled. For instance, although the ATP binding site has been localized toward the amino terminus of HSP90, ATP binding as well as elevated temperature bring about a profound conformational change that is not restricted to the ATP-binding domain (16 -18). When the concentration of HSP90 was lower than 1 M, both ATP binding and elevated temperature induced an equivalent conformational change, converting HSP90 from a linear dimer into an O-ring-shaped structure (18). On the other hand, when the concentration of HSP90 was sufficiently high, HSP90 self-oligomerized instead of formed O-ring-shaped molecules, probably through essentially identical interactions (29). Alteration of the regional interaction may be closely related to these conformational changes. In this connection, we recently proposed that the liberation of the N-terminal client-binding region from the middle suppressor region is the mechanism underlying the temperature-dependent activation of HtpG, an Escherichia coli homologue of mammalian HSP90 (30).
Our previous study on limited proteolysis of human HSP90␣ strongly suggested the existence of subregional structures within the three respective regions (23). Further structural and functional analyses on potential substructures presented difficulty because of limited information on regional functions in those days. We recently reported the regional structures and their interactions of HtpG (28). As a result, several characteristics of HtpG and HSP90 regions emerged, which provided the probes for investigation of the subregional structures of HSP90.
In the present study, we investigated potential subregional structures in the three regions of HSP90 responsible for the regional interactions and further investigated the role of the regional interactions of HSP90 in vivo by using budding yeast, Saccharomyces cerevisiae. We demonstrate that the intramolecular interaction between Regions A and B is indispensable for the in vivo function of HSP90.
Construction of Bacterial Expression Vectors-The DNAs encoding the full-length form of human HSP90␣ (31) and E. coli HtpG (32) were generously provided by Drs. K. Yokoyama (Riken Life Science Center, Tsukuba, Japan) and E. A. Craig (University of Wisconsin Medical School, Madison, WI), respectively. Construction of the plasmids carrying the full-length form of HSP90␣ and HtpG tagged with a histidine hexamer, designated pH 6 HSP90␣ and pH 6 HtpG, respectively, and their regions (A or BC) were described previously (28,33). Y1090(pREP4) was transformed with the plasmids and selected on Luria broth agar plates containing 50 g/ml of ampicillin and 25 g/ml of kanamycin.
Expression and Purification of Recombinant Proteins-H 6 HSP90␣, H 6 HtpG, and their truncated forms were expressed and purified by use of Talon affinity resin according to the manufacturer's protocol, except that 10 mM imidazole was added in the lysis/washing buffer. Bound proteins were eluted with 0.1 M imidazole (pH 8.0) containing 10% (v/v) glycerol.
Substitution of Conserved Amino Acids in Region B-Three amino acids within Region B were substituted with Ala in human HSP90␣, E. coli HtpG, and yeast HSC82 by site-directed mutagenesis in combination with the DpnI degradation elimination of template DNA. The mutations were confirmed by DNA sequencing. Leu 477 , Glu 517 , and Leu 592 of human HSP90␣ (34) are equivalent to the respective Leu 416 , Glu 456 , and Leu 532 in E. coli HtpG (34) and to Leu 453 , Glu 493 , and Leu 567 in yeast HSC82 (35). Throughout the present study, we use the amino acid numbers of human HSP90␣ to refer to these three amino acids for simplicity.
Bacterial Two-hybrid System-Minimal regions responsible for the interaction between Region A and Region B were determined by the bacterial two-hybrid system according the method of Karimova et al. (36). The DNA fragments encoding Region A of HSP90␣ and its truncated forms amplified by PCR were inserted into pKT25 kan and designated pKT25-HSP90␣A-(1-400), pKT25-HSP90␣-(222-400), and so on (see Fig. 3a). The DNA fragments carrying Region B of HSP90␣ and its truncated regions were amplified by PCR and inserted into pUT18C amp , designated pUT18C-HSP90␣B-(401-618), pUT18C-HSP90␣-(401-600), and so on (see Fig. 3b). The complex formation between coexpressed recombinant proteins was quantified by ␤-galactosidase activity as described previously (36). Values were reported as means of three or four samples and expressed as percentages of that activity in the bacteria co-expressing nontruncated Region A and Region B (100%).
Yeast Expression System-Temperature sensitivity of yeast cells ex-pressing mutated forms of yeast HSC82 was examined as described previously (37). 5CG2HIS (MAT␣ ura3-52 lys2-801 amber ade2-101 ochre triple-⌬63 his3-⌬200 leu2-⌬1 hsc82::HIS3 hsp82::GAL1-HSP82::LEU2) is a strain whose endogenous HSC82 gene was disrupted and HSP82 gene was controlled with a GAL1 promoter (37). The strain forms colonies on SGal plates by the expression of HSP82 but not on S.D. plates. A DNA fragment encoding the full-length form of yeast HSC82 cut out with SfcI and blunt-ended was inserted into a blunt-ended BamHI site of the GPD promoter of a multicopy plasmid, pYO326GPD (38), designated pYO326GPD-HSC82. The colony formation was examined on S.D. plates at 25 or 37°C for 3 days and 14°C for 2 weeks following the introduction of pYO326GPD-HSC82. Transformants grew on both S.D. and SGal plates at 14 -37°C. SDS-PAGE and Protein Sequencing-Electrophoresis was performed at a polyacrylamide concentration of 12.5% in the presence of 0.1% SDS. In cases (see Figs. 1 and 2) where fine separation of proteins smaller than 20 kDa was required, the Tris-Tricine system at a polyacrylamide concentration of 10% was employed (39). Separated proteins were stained with Coomassie Brilliant Blue. Low molecular weight and peptide markers were used as references. For determination of the Nterminal sequences of proteolytic fragments, separated proteins in a polyacrylamide gel were electrotransferred to a polyvinylidene difluoride membrane (Bio-Rad). After having been stained with Coomassie Brilliant Blue, the excised bands were directly subjected to sequencing with a model 477A protein sequencer (PE Biosystems).
Immunoblotting Analysis-Yeast cells (pYO326GPD-HSC82 and pYO326GPD-HSC82-L477A/E517A) cultured in SD medium overnight at 30°C were diluted to 0.15 in absorbance at 600 nm in 50 ml of the same medium and then further cultured at 30 or 37°C for 6 h. The cells were lysed with Y-Per at 2.5 ml/g cell precipitate. SDS-PAGE and immunoblotting were performed thereafter as described previously (23). An anti-HSP90 monoclonal antibody, K41220 (10 g/ml), was used as the first antibody. K41220 binds to human HSP90␣ and HSP90␤ with equal efficiency (23). It recognizes one of the most immunogenic regions of human HSP90␣, 291 NKTKPIWTRNPDDI 304 (34), of which two amino acids (underlined) are replaced in human HSP90␤ ( 283 NK-TKPLWTRNPSDI 296 ) (40) and yeast HSC82 ( 267 NKTKPL-WTRNPSDI 280 ) (35) as described previously (23). Alkaline phosphatase-conjugated goat anti-mouse IgG was used as the second antibody at a 1:2500 dilution. Kaleidoscope prestained standard (Bio-Rad) was used as molecular markers.
Polyacrylamide Gel Electrophoresis under Nondenaturing Conditions-In order to estimate molecular configurations, recombinant proteins and their mixtures were subjected to PAGE on a 7.5% polyacrylamide gel under nondenaturing conditions (41). We found that elution buffers (i.e. 0.1 M imidazole (pH 8) containing 10% (v/v) glycerol for H 6 HtpG) did not interfere with the interaction between recombinant proteins. Accordingly, purified proteins were used without any further treatment in this study. Electrophoresis was performed at room temperature unless otherwise described. Separated proteins were stained with Coomassie Brilliant Blue. Ovalbumin (45 kDa), bovine serum albumin (66 kDa as monomer, 132 kDa as dimer, and 198 kDa as trimer), and catalase (240 kDa) were used as references.
Protein Concentration-Protein concentrations were determined by the bicinchoninic acid method (Pierce).

RESULTS
Limited Proteolysis of Region A-We first performed limited proteolytic analysis of HSP90␣. To make the interpretation simple, we here used the regions of HSP90␣ instead of the full-length form. Limited proteolysis of Region A with trypsin produced a limited number of proteolytic fragments (Fig. 1a). At the lowest trypsin concentration, the 52-kDa Region A was split into 15-and 40-kDa fragments (lanes 2-4). At moderate and higher trypsin concentrations, the respective 30-kDa (lanes 4 -7) and 10-kDa (lanes 6 -8) fragments appeared. The 40-and 10-kDa fragments had an N terminus identical to that of the intact form. The 15-kDa fragment had two adjacent N termini starting at Glu 282 and Tyr 284 ( Table I). The 30-and 40-kDa fragments were also produced by limited proteolysis of the full-length HSP90␣ (23), which implies that the proteolytic pattern is identical between the full-length form and Region A. Interestingly, the 40-kDa fragment was still blotted by an anti-HSP90 monoclonal antibody K41102, which recognizes residues 247-257, but the 30-kDa fragment was not blotted by it (23). Therefore, the 40-and 15-kDa fragments seemed to consist of amino acids 1-281/283 and 282/284 -400, respectively, and the 30-kDa fragment was derived from the 40-kDa species by deletion of the highly charged region (Glu 223 -Lys 283 ), as schematically illustrated in Fig. 1b.
Limited Proteolysis of Region B-Next we performed limited proteolysis on Region B. Unfortunately, Region B of HSP90␣ was not quantitatively recovered from the bacterial expression system employed, presumably because the recombinant protein was unstable. Hence, we expressed Glu 401 -Asp 732 , which consisted of Region B and Region C (designated as Regions BC). The 42-kDa Regions BC was split into several fragments including 14-and 27-kDa fragments at low trypsin levels ( Fig.  2a). At moderate trypsin concentrations, 6.5-, 20-, and 24-kDa fragments appeared. At higher concentrations, 8-and 19-kDa fragments accumulated. Among them, the 14-kDa fragment was the most resistant to a wide range of trypsin concentrations (lanes 2-6).
N-terminal sequencing revealed that the 14-kDa fragment started at Asp 621 (Table II), implying that the fragment corresponded to Region C, as expected. The N-terminal sequence of the 27-kDa fragment was identical to that of the original fragment, indicating that the 27-kDa entity represented Region B. The N termini of the 24-, 20-and 19-kDa fragments were Asn 415 , Leu 447 , and Leu 459 , respectively. Thus, the 27-kDa Region B was successively degraded at the N-terminal side into 24-, 20-, and 19-kDa fragments. In addition to the N-terminal truncation, an 8-kDa fragment was obtained by cleavage at the C-terminal side of 27-kDa Region B. Taken together, the 42-kDa Regions BC was initially split into the 27-kDa Region B and the 14-kDa Region C, and the former was further processed into smaller fragments at cleavages of Lys 414 -Asn 415 , Lys 446 -Leu 447 , Lys 458 -Leu 459 , or Lys 546 -Glu 547 (Fig. 2b). It should be noted that a 22-kDa fragment carrying the N-terminal twothirds of Region B was observed upon limited proteolysis of the full-length form (23) but that such a species was not detected on proteolysis of Region B.
Minimal Regions Sufficient for the Interaction between Region A and Region B-Next we determined the minimal regions required for the interaction between Region A and Region B by using a bacterial two-hybrid system. As shown in Fig. 3a, an interaction between Region A and Region B was ascertained. This interaction reflects the intramolecular interaction of an intact molecule, as reported recently (28). HSP90␣-(222-400) and HSP90␣-(289 -400) (i.e. truncations from the N terminus to Phe 221 and to Glu 288 , respectively) still possessed considerable binding activities. A further deletion up to Gly 310 resulted in a loss of the binding. In contrast, even the smallest deletion (28 residues) from the C terminus of Region A caused complete loss of the activity. Thus, the minimal region of Region A required for the binding to Regions BC was defined as residues 289 -400.
We then defined the minimal region of Region B for binding to Region A (Fig. 3b). When the C terminus of Region B was serially deleted, residues 401-546 retained complete binding to Region A, but residues 401-541 (i.e. an additional 5-amino acid deletion) resulted in complete loss. It should be emphasized that one of the tryptic cleavage sites in Region B was Lys 546 -Glu 547 (Table II). To the contrary, even a 20-amino acid deletion of the N terminus resulted in complete loss. Thus, residues 401-546 are the minimum requirement for the binding to Region A. From these two experiments, we conclude that residues 289 -400 of Region A are associated with residues 401-546 of Region B.
Effect of Substitution of Conserved Amino Acids in Region B-The Region A-Region B interaction is maintained in both E. coli HtpG and human HSP90␣, and Region B of one species can   (44), there are 28 identical amino acids, of which 25 and 3 residues are distributed in the residue sequences 401-546 and 547-615, respectively. Therefore, we expressed Regions BC with amino acid substitutions and tested their bindings to Region A; two of them (Leu 477 and Glu 517 ) were arbitrarily chosen from the N-terminal subsite, and one (Leu 592 ) was from the C-terminal one.
The two-hybrid system revealed that the amino acid substitution in Region B of human HSP90␣ (Leu 477 or Glu 517 to Ala) caused this region to lose its ability to bind Region A, whereas the substitution at position 592 from Leu to Ala still allowed 29% of the binding (Table III). Thus, consistent with the results on the truncated forms of Region B (Fig. 3b), the conserved amino acids, Leu 477 and Glu 517 , located in the N-terminal subsite (residues 401-546) of the region were essential for the binding to Region A, whereas Leu 592 , located in the C-terminal subsite (residues 547-615), was not crucial. We also performed the experiment in the hybrid combination of these mutated HSP90␣BC and HtpGA, substituting HSP90␣A, and obtained an identical binding profile (Table III).
Although the two-hybrid analysis provided convincing results with positive data, there might be several interpretations on negative ones. For instance, steric hindrance caused by the fusion with a protein encoded by a vector might prevent the interaction of a protein of interest. Therefore, we further examined the effect of the mutations on the regional interaction    (Fig. 4b, lanes 1-7). The upper shift of H 6 HtpGB-E517A might have been caused by the reduction in the acidity (Fig. 4b, lane 4) 12 and 13).
We recently reported the temperature-dependent dissociation of the interaction between Region A and Region B of HtpG, which appears to be closely related to the function of the molecular chaperone (30). Accordingly, we further examined the effect of temperatures on the interaction between H 6 HtpGA and mutated H 6 HtpGBC. Essentially all H 6 HtpGBC associated with H 6 HtpGA at 4 -37°C (Fig. 4c, lane 4, arrows). In contrast, the complex formation between H 6 HtpGA and H 6 HtpGBC-L477A/E517A was drastically reduced at all temperatures tested (lane 5, arrows). Nevertheless, it should be noted that some extents of the complex formation were consistently observed at 4 and 20°C. The complex was still observed at 30°C but was scarcely present at 37°C. Thus, we conclude that a single substitution at either Leu 477 or Glu 517 to Ala was not sufficient for the complete disruption of the intramolecular interaction when the two regions were feasibly accessible but that the simultaneous replacement of the two amino acids could completely abrogate the interaction, in particular at 37°C.
To investigate the conformational changes of H 6 HtpGBC induced by the amino acid substitutions, H 6 HtpGBC and its mutated forms were subjected to limited proteolysis to trypsin and papain. Fig. 4d demonstrated that H 6 HtpGBC-L477A/ E517A was highly sensitive to trypsin. Moderate susceptibility was observed on H 6 HtpGBC-E517A. H 6 HtpGBC-L477A and an intact form were resistant to trypsin under the conditions employed. Papain treatment revealed the same order of sensitivity (data not shown). Thus, the double amino acid substitutions seemed to significantly affect the tertiary structure of Regions BC. In the final part of our study, we investigated whether these mutations would affect the in vivo function of HSP90 in a yeast expression system.
Region A-Region B Interaction Is Important in Vivo-Yeast 5CG2HIS strain, which expressed yeast HSP82 under the control of galactose-regulated promoter, formed colonies at 14 -37°C on galactose-containing but not on glucose-containing plates (37). After the introduction of pYO326GPD-HSC82, which resulted in constitutively expressed intact HSC82, the yeast strain formed colonies at 25-37°C on glucose plates, on which expression of HSP82 was repressed. Yeast strains that expressed HSC82 with a single amino acid substitution, L477A or E517A, or with a double mutation, L477A/L592A or E517A/ L592A, also formed colonies at both temperatures on glucose plates (Fig. 5, a and b). In contrast, the strain that carried HSC82 with L477A/E517A double mutations grew at 25°C (Fig. 5a) and 30°C (data not shown) but not at 37°C on glucose plates (Fig. 5b). Therefore, mutated HSC82 that was deficient in the Region A-Region B interaction brought the yeast cell to high temperature sensitivity, suggesting that the interaction between Region A and Region B is essential in vivo.
Immunoblotting analysis of yeast HSC82 (Fig. 6) was performed with an anti-HSP90 monoclonal antibody K41220. After cultivation either at 30 or 37°C for 6 h, an 82-kDa band was found in all cases (i.e. yeast cells expressing intact HSC82 or HSC82-L477A/E517A at both temperatures). The intensity was more predominant at 37°C, presumably simply reflecting elevated metabolism at higher temperatures. Therefore, the selective proteolytic break down of HSC82-L477A/E517A protein was unlikely to explain the growth inhibition of the yeast cells, but a functional defect should occur.
Taken together, the defect in the interaction between Region A and Region B H 6 HtpGBC-L477A/E517A especially at lower temperatures (Fig. 4c) indicate the molecular basis that produced the high temperature-sensitive mutant of the yeast. Yeast strains that carried any mutated HSC82 were not coldsensitive (14°C) and had no dominant negative effect (data not shown). DISCUSSION In the present study, we performed analytically limited proteolysis of Region A and Regions BC of HSP90␣ using bacterially expressed protein fragments to predict the subregion structures of these regions. To apply this technique, one should carefully consider whether their proteolytic profile might be altered from that of the full-length form. The tryptic pattern of Region A appeared to be identical to that of the full-length form (i.e. the major 40-and 30-kDa fragments shown in Fig. 1a were also observed in our previous analysis using the full-length HSP90␣) (23).
On the other hand, the tryptic pattern of Region B seemed to be slightly different between the region fragment and the fulllength form. In other words, the 27-kDa Region B produced from the full-length form was further processed to a 22-kDa species by a cleavage at the C-terminal side (23), but this entity was not formed upon proteolysis of the region because of preferential N-terminal breakdown (Table II). The present study on Regions BC demonstrated that the Lys 546 -Glu 547 bond corresponded to this site producing the 22-kDa species (Table II). Therefore, the limited proteolytic analyses using recombinant fragments reproduced, even if but partially, the cleavage profile of the full-length form, which made it possible for us to predict the subregion boundaries.
One might raise a possibility that the proteolysis occurred in exposed surface loops within a domain. If such fragments were formed, they would possess sticky hydrophobic patches, which would allow interaction with various fragments. However, as exemplified in Table II, the tryptic cleavage at Lys 546 -Glu 547 in Regions BC and, as shown in Fig. 3b, the C-terminal deletion of five amino acids from residues 401-546 to 401-541 of Region B caused a complete loss of the binding to Region A. A similar phenomenon was observed on tryptic cleavage sites (Lys 281 -Glu 282 and Lys 283 -Tyr 284 ) of Region A (Table I) and the region (residues 289 -400) responsible for the binding to Region B and (Fig. 3a). These findings strongly suggested the accordance of the structural and functional units on HSP90␣. Moreover, it should be noted the primary tryptic site within Region A was not located within the highly charged region (residues 223-283) but was the edge of it. Region A of HSP90␣ was divided into two fragments, 40 and 15 kDa. Notably, cleavage sites of Region A of HSP90␣ and HtpG were located at close positions (Fig. 7a). From both of those structural and functional analyses, Region A of HSP90␣ should be divided into two subregions (i.e. residues 1-283 and 284 -400). The former comprises the ATP-binding domain (residues 9 -222) and highly charged region (residues 223-283) (Fig. 7). In view of its high charge density and immunogenic properties (23), the highly charged region may be exposed to the outer surface of the molecule. This region appears to be dispensable for viability and signal transduction in yeast (45) but might have a role in modulating the function of the ATPbinding site (46).
The common cleavage site (Lys 546 -Glu 547 ) in both the full-  length form and Regions BC may be located at the boundary between functional entities. In fact, this is likely, because residues 401-546 were sufficient for binding to Region A, but a further five-amino acid deletion resulted in null binding (Fig.  3b). Moreover, our previous studies demonstrated that the C-terminal portion of the region was sufficient for binding to Region C, that Val 542 -Tyr 627 was associated with Region C after chymotryptic proteolysis (22), and that the C-terminal one-third of HtpG (Gln 481 -Lys 552 ), equivalent to residues Val 542 -Ala 618 of HSP90␣ was capable of binding to Region C (28). Therefore, it is reasonable to conclude that the C-terminal portion of Region B interacts with Region C. In Region B of HtpG, trypsin attacked a site almost identical to that of chymotrypsin, and the two sites are relatively close to the tryptic site of HSP90␣ (Fig. 7b) (22). Therefore, Region B can be divided into two subregions: subregion BI (residues 401-546) specific for binding to region A and subregion BII (residues 547-615) for interaction with Region C.
We recently proposed that liberation of Region A from Region B is important for the chaperone activity of HSP90 (30). To verify our hypothesis, we substituted amino acids of subregions BI and BII conserved among HSP90 family members. As a result, the double mutant (L477A/E517A) within the subregion BI, whose product did not associate with Region A, was unable to support the growth of an HSP82-depleted yeast cell at 37°C. Importantly, the yeast cell could grow at 14 -30°C, but not at 30°C, which faithfully reflected the temperature dependence of the complex formation between HtpGBC-L477A/E517A and HtpGA.
Fine resolution of the region and subregion structures of HSP90 and their roles, as presented in Fig. 7c, has provided the molecular basis for systematic analyses of amino acids screened by mutational studies. The present study on a temperature-sensitive yeast strain that carries HSC82-L477A/ E517A is the first example of such analyses. Although the previous studies (37,47,48) reported the mutations of HSP90 that caused yeast cells to be high temperature-sensitive, the underlying mechanism remains unknown. In contrast, the mechanism on the temperature sensitivity on double mutations of yeast HSC82 reported in this study seems to be readily attributed to the defect in the intramolecular interaction between subregions AII and BI (Fig. 4c). Moreover, it is reasonable to assume that potential client proteins are increased at elevated temperatures, which may additionally contribute to the growth defect of the yeast at higher temperatures. In this respect, it is interesting to investigate whether or not the fourand three-point mutations (Fig. 7c, asterisks) occurring in subregions AII and BI, respectively, reported in previous studies (37,47,48) actually cause the functional defect through the disruption of the regional interaction of HSP90.
In conclusion, the present study describes for the first time the mechanism on the high temperature-sensitive mutation in vivo, coupled with the in vitro functional defect (i.e. loss of the Region A-Region B interaction).