ATP-binding Properties of Human Hsp90*

Hsp90 is one of the most abundant proteins in the cytosol of eukaryotic cells. Under physiological conditions Hsp90 has been shown to play a major role in several specific signaling pathways, including maturation of various kinases and maintenance of steroid receptors in an activable state. It is well established that the level of Hsp90 increases severalfold under stress conditions, and it has been shown that the chaperone function of Hsp90 is ATP-independent. Although yeast Hsp90 does not bind ATP, as determined by a number of methods monitoring tight binding, ATP-dependent functions of Hsp90 in the presence of co-factors and elevated temperatures are still under discussion. Here, we have reinvestigated ATP-binding properties and ATPase activity of human Hsp90 under various conditions. We show that human Hsp90 does not bind ATP tightly and does not exhibit detectable ATPase activity. However, using electron spin resonance spectroscopy, weak binding of spin-labeled ATP analogues with half-maximal binding at 400 μm ATP was detected. The functional significance of this weak interaction remains enigmatic.

Hsp90 is one of the most abundant proteins in the cytosol of eukaryotic cells. Under physiological conditions Hsp90 has been shown to play a major role in several specific signaling pathways, including maturation of various kinases and maintenance of steroid receptors in an activable state. It is well established that the level of Hsp90 increases severalfold under stress conditions, and it has been shown that the chaperone function of Hsp90 is ATP-independent. Although yeast Hsp90 does not bind ATP, as determined by a number of methods monitoring tight binding, ATP-dependent functions of Hsp90 in the presence of co-factors and elevated temperatures are still under discussion.
Here, we have reinvestigated ATP-binding properties and ATPase activity of human Hsp90 under various conditions. We show that human Hsp90 does not bind ATP tightly and does not exhibit detectable ATPase activity. However, using electron spin resonance spectroscopy, weak binding of spin-labeled ATP analogues with halfmaximal binding at 400 M ATP was detected. The functional significance of this weak interaction remains enigmatic.
Under stress conditions, e.g. high temperatures, cells overexpress a distinct set of proteins, the so-called heat shock or stress proteins. The major classes of heat shock proteins, Hsp90, 1 Hsp70, Hsp60, and small Hsps, are thought to function as molecular chaperones during protein folding (1)(2)(3). The mechanisms of these chaperone functions are still under intensive investigation. In the case of Hsp70 and Hsp60, ATP binding and hydrolysis is a major requirement in chaperone-mediated folding (1). In vivo experiments suggest that Hsp90, one of the most abundant and conserved heat shock proteins, is a specific chaperone involved in regulating signal transduction pathways by assisting structural changes of certain kinases and steroid receptors (4 -8). In addition, results from in vitro studies highlight the general chaperone activities of Hsp90 (2,3). Like other chaperones, Hsp90 performs at least part of its activity in association with specific partner proteins (9 -14), some of which seem to function as molecular chaperones themselves (15,16). In this context, the involvement of ATP in Hsp90 function is still a controversial subject (11)(12)(13)(17)(18)(19), and ATP binding as well as ATPase activity of Hsp90 have been reported previously (18, 20 -23). In contrast to these findings, we have shown that yeast Hsp90 does not bind ATP (17) by means of assays that are reflecting structural changes of the observed protein in the presence of ligand, detecting binding of the protein to immobilized ATP or binding of fluorescencelabeled ATP analogues to Hsp90. These observation enabled only the investigation of tight interactions between ATP and Hsp90. New results concerning p23, a partner protein of Hsp90 that is thought to play an important role in Hsp90/steroid receptor complexes, readdressed the possibility that human Hsp90 is an ATP-binding protein (18). This brought up the question of whether human Hsp90 differs in its ATP-binding properties from yeast Hsp90, assuming that p23 does not bind ATP (Ref. 15 and see below).
In agreement with our previous findings regarding yeast Hsp90 (17), we could show that human Hsp90 behaves similarly and does not bind ATP strongly. However, the possibility of weak ATP binding could not be ruled out by the methods used. Therefore, ESR spectroscopy using spin-labeled nucleotides was applied to examine potential low binding affinities of Hsp90 and ATP. This method had successfully been used before to investigate nucleotide binding to enzymes like F1-type ATPases, the chaperone DnaK, as well as the proto-oncogene product p21 ras (24 -27, 29). 2 In ESR spectroscopy the signals of the freely tumbling spin-labeled nucleotides are well separated from the signals of the protein-immobilized components and therefore allow direct determination of the free substrate. In addition, ESR spectroscopy enables the evaluation of rather weak interactions of proteins and nucleotides with dissociation constants in the upper micromolar range that are usually not detectable by the more common techniques like fluorescence spectroscopy.
We also investigated the effect of molybdate on Hsp90 structure and its interaction with nucleotides, since efficient complex formation between Hsp90 and p23 seems to require apart from ATP the polyanion molybdate (11)(12)(13). About 40 years ago, molybdate was found to inactivate phosphoprotein phosphatases (30). Subsequently, molybdate was used to stabilize glucocorticoid receptors, which were thought to be destabilized by dephosphorylation (31). After detecting some low molecular weight factors associated with Hsp90, which seem to be important to keep steroid receptors activatable in the absence of hormone, it was suggested that molybdate mimics the effects of these factors (32,33).
Here we show that human Hsp90 binds ATP with a K D in almost millimolar range as detected by ESR spectroscopy. While p23 seems to slightly affect this process, no further effect of molybdate on Hsp90 structure and its interaction with p23 and ATP could be determined.

Proteins-Human
Hsc70 and human p23 were purified as recombinant proteins from Escherichia coli as described previously (15,34). In the case of Hsc70, an additional gel filtration step with Superdex 200 preparatory grade column was introduced. Murine BiP was purified as described previously (35). Recombinant human Hsp90 was purified from baculovirus infected Sf9 cells, kindly provided by the Colorado Cancer Center. 48 h after infection, a crude cell extract was prepared by gentle pottering. The soluble fraction was loaded onto Q-Sepharose and eluted with a 0.1-1 M NaCl gradient. The appropriate fractions were pooled, dialyzed against 200 mM potassium phosphate, pH 6.8, and loaded onto a hydroxyapatite column. Elution of bound Hsp90 was achieved using a 200 -400 mM potassium phosphate gradient at pH 6.8. Finally, gel filtration chromatography with Superdex 200 preparatory grade was performed. The protein concentrations were determined using extinction coefficients calculated from the known contents of aromatic amino acids (36). The extinction coefficients used are 0.74 for human Hsp90, 0.40 for murine BiP, 0.51 for human Hsc70, and 1.96 for human p23 for a 0.1% solution at 280 nm and 1 cm path length.
Analysis of the Nucleotide Content of Human Hsp90 -Nucleotides were analyzed by reverse phase chromatography with a 2-ml C18 column (Bishop) as described previously (17).
ATP-agarose-Binding to immobilized ATP was performed essentially as described previously (17). 1 ml of ATP-agarose was equilibrated in buffer A (40 mM Hepes-KOH, pH 7.5, 20 mM KCl, 5% glycerol, 5 mM MgCl 2 ) in a batch procedure. 200 g of each protein were added to the equilibrated ATP-agarose together with 5 mg of IgG, to prevent unspecific binding, and rotated for 1 h at 37°C or 43°C. The agarose was washed five times with buffer A (wash 1) and another five times with buffer B (buffer A ϩ 0.5 M KCl, wash 2). Elution of specifically bound proteins was initiated by addition of buffer B supplemented with 5 mM ATP. This procedure was repeated five times. Finally the ATPagarose was washed with two bed volumes of 7 M urea in buffer A. The samples were analyzed by 10% SDS-polyacrylamide gel electrophoresis. All gels were silver-stained.
ATPase Activity-ATPase assays were performed as described previously (35,37). Human Hsp90 and BiP were incubated at 37°C or 43°C and BSA at 37°C with 500 M (final concentration) unlabeled ATP and 10 Ci of [␣-32 P]ATP in a total volume of 20 l, containing 40 mM Hepes-KOH, pH 7.5, and 2 mM MgCl 2 . Protein concentrations varied from 0.05 to 0.75 g/l. A possible influence of molybdate on ATPase activities was examined in the presence of 10 mM Na 2 MoO 4 .
Influence of ATP on the Intrinsic Tryptophan Fluorescence of Selected Proteins-The respective proteins were diluted into various concentrations of ATP in buffer C (40 mM Hepes-KOH, pH 7.5, 5 mM MgCl 2 ) at the concentrations given in the figure legends. The samples were incubated for 5 min at 37°C or 43°C to achieve equilibrium. Subsequently, fluorescence measurements were performed in 3-mm quartz cuvettes in a Perkin-Elmer MPF 44A luminescence spectrometer at 37°C or 43°C. The tryptophan fluorescence was measured at an excitation wavelength of 295 nm and an emission wavelength of 325 nm. The spectral bandwidths were set to 5 nm and 10 nm for excitation and emission respectively. Fluorescence intensities are expressed as the percentage of total protein fluorescence obtained after subtracting the background intensity of the buffer. Changes in tryptophan fluorescence were corrected for the absorption of ATP (inner filter effect) at 295 nm excitation using the formula shown in Equation 1.
F corr is the corrected fluorescence, and F obs is the observed fluorescence after subtracting the background light intensity. L T is the total ATP concentration, and ⑀ L is the absorption coefficient of ATP (38).
Temperature-induced Unfolding of Human Hsp90 -To monitor thermal unfolding of human Hsp90, fluorescence measurements were performed in the temperature range from 20°C to 70°C in stirred 1-cm quartz cuvettes in a Perkin-Elmer MPF 44A luminescence spectrometer equipped with a thermostated cell holder connected to a thermoprogrammer. Human Hsp90 was diluted into 100 mM potassium phosphate, pH 6.8, to a final concentration of 150 g/ml. The fluorescence of Hsp90 was measured at an excitation wavelength of 295 nm and an emission wavelength of 320 nm. The spectral bandwidths were set to 10 and 4 nm for emission and excitation, respectively. The heating rate was 0.5°C/min.
Binding of Fluorescence-labeled ATP Analogues to Human Hsp90 and BiP-Fluorescence measurements were performed with an SLM Smart 8000 Photon-Counting spectrofluorimeter (SLM Instruments) at 37°C or 43°C. The buffer used was 50 mM Tris/HCl, pH 7.5 (25°C), 100 mM KCl, 5 mM MgCl 2 , 2 mM EDTA, 2 mM dithioerythritol. The fluorescent nucleotides were added and mixed manually. All data are volumecorrected averages of 10 readings of 1-s intervals of photon counting. Binding of Spin-labeled ATP Analogues to Human Hsp90 -ESR spectra were recorded using a Bruker ESP 300E spectrophotometer operating in the X-band mode. The resonance cavity used was a TE 102 (ER 4102 ST). The experiments were performed at 19°C in micro-flat cells with a total volume of 40 or 50 l. The solutions were mixed using a platinum wire. The spectra were recorded at 6.3-milliwatt microwave power and a peak to peak modulation amplitude of 1 G. Protein and spin-label concentrations for ESR spectra are given in the figure legends. The nucleotide analogue 2Ј,3Ј-SL-ATP was prepared as described previously (39,40). 2Ј,3Ј indicates a rapid equilibrium of the ester bond between the C-2 and C-3 of the ribose moiety. The amount of proteinbound 2Ј,3Ј-SL-ATP was determined as the difference between the known amounts of total spin-labeled nucleotides added and the directly measured free spin-labeled nucleotides (the signal amplitude of the freely tumbling, not enzyme-bound radicals is directly proportional to their concentration). To avoid chemical reduction of the spin labels, thiol components such as dithioerythritol were removed prior to the experiments.

Highly Purified Human Hsp90 Does Not Show Detectable
ATPase Activity-Recombinant human Hsp90 was purified to homogeneity by column chromatography from baculovirus infected SF9 cells. The purified protein was free of bound nucleotides according to HPLC analysis (data not shown). Previous reports suggested that isolated Hsp90 may be an ATP-hydrolyzing protein (22,23). However, our attempts failed to detect ATPase activity in highly purified bovine Hsp90 preparations (41), and the observed low ATPase activities of purified yeast Hsp90 may be due to impurities such as kinases (17) or they may be caused by ATP/GTPases associated with Hsp90 (42). In agreement with our previous results, human Hsp90 did not exhibit detectable ATPase activity at 37°C (Table I). Furthermore, no ATPase activity of its partner protein p23 could be detected (Table I). Next, we checked the influence of heat shock temperatures on ATPase activity. For BiP, a temperature shift from 37°C to 43°C yielded a 3.5-fold increase in ATPase activity (Table I). However, no significant change in the ATPase activity of Hsp90 could be detected upon an increase of the temperature (Table I). Molybdate alone showed a slight in-crease in ATPase activity (data not shown), which might be due to hydrolysis of ATP by molybdate (43). However, in our assays no significant influence of molybdate on the ATPase activities of BiP, and Hsp90, could be observed (Table I). Since Hsp90/ p23 interaction is supposed to be ATP-dependent (12,13,18), we examined the ATPase activity of a mixture of both proteins to test for cooperative effects. Again, we failed to detect any ATPase activity (Table I).
ATP Does Not Bind Tightly to Purified Hsp90 -Previously, we had reported that addition of ATP to yeast Hsp90 does not significantly alter the intrinsic tryptophan fluorescence of the protein (17). We readdressed this question by comparing the influence of ATP on the intrinsic fluorescence of human and yeast Hsp90 with well known ATP-binding proteins like human Hsc70 and murine BiP; BSA, which binds ATP with a K D between 50 and 100 M (44); and the non-ATP-binding protein IgG at physiological and heat shock temperatures. We detected a slight decrease in tryptophan fluorescence of human and yeast Hsp90 with increasing ATP concentrations, but similar effects of ATP were also observed for IgG and a tryptophan solution (Fig. 1). BSA, as a weak ATP-binding control, showed a higher decrease in fluorescence with increasing amounts of ATP compared with Hsp90 or IgG. For the tight ATP-binding proteins Hsc70 and BiP, a significant effect of ATP on the intrinsic protein fluorescence upon binding was observed (Fig.  1). Switching the temperature to 43°C did not change the general fluorescence patterns of all proteins tested (data not shown). These findings led to the conclusion that ATP does not influence the intrinsic fluorescence of Hsp90.
Because binding of ATP does not necessarily lead to changes in the intrinsic fluorescence of proteins, we performed experiments with fluorescence-labeled ATP analogues to monitor binding directly. Here, specific binding results in fluorescence changes of the label, which can be reversed by competing with unlabeled ATP. Increasing amounts of Hsp90 were incubated with a fixed quantity of the ligand and the fluorescence emission was recorded. The experiments performed at 37°C and 43°C demonstrated a slight increase in fluorescence of the nucleotide analogues MABA-ATP, MANT-ATP, and ⑀-ATP, when up to 2.5 M Hsp90 were added ( Fig. 2A and data not  shown). However, this increase was due to an unspecific effect of the labeled ATP analogues, because the reaction could not be competed with unlabeled ATP ( Fig. 2A and data not shown). In contrast, control experiments with murine BiP showed that the fluorescence of MABA-ADP increased significantly compared with the experiment with Hsp90 (Fig. 2, A and B, inset). In this case, bound MABA-ADP could be removed from BiP by competing with increasing amounts of unlabeled ADP (Fig. 2B) with a calculated dissociation constant of 0.38 Ϯ 0.12 M. This value is similar to measurements performed with DnaK, a Hsc70 analogue from E. coli (45). We also tested p23 for ATP binding, but we failed to detect binding to p23 or Hsp90 and p23 together using MABA-ATP (Fig. 2C) or the other analogues (data not shown). The presence of 20 mM sodium molybdate did not change the ATP-binding properties of all proteins tested (data not shown).
Another method to detect the association of proteins with nucleotides is binding to immobilized ATP. In this experiment the addition of IgG was necessary to prevent unspecific interaction of Hsp90 with the agarose matrix. Previously it was shown that IgG has no influence on the specific binding of the control protein Hsc70 (17). Elution of the bound proteins was induced by addition of ATP after extensive salt washes. Specific binding of Hsc70 to immobilized ATP could be observed at 37°C (data not shown) and at 43°C (Fig. 3A), while binding of Hsp90 could not be detected at both temperatures ( Fig. 3B and data not shown). Again, addition of sodium molybdate did not alter the binding properties of Hsp90 and Hsc70 at the selected temperatures ( Fig. 3C and data not shown).
Together with our previous findings (17), we conclude that human Hsp90 does not bind ATP strongly at physiological or heat shock temperatures and this property is not changed in the presence of molybdate.
Weak ATP Binding of Hsp90 -To investigate the possibility that ATP binds weakly but specifically to Hsp90, we tested three different spin-labeled adenine nucleotides using ESR spectroscopy. The compounds tested included N 6 -SL-ATP (N 6 -(2,2,6,6-tetramethylpiperidin-4-yl-1-oxyl-)adenosine-5-triphosphate) and C 8 -SL-ATP (8-(2,2,6,6-tetramethylpiperidin-4-yl-1oxyl-) aminoadenosine-5-triphosphate), where the adenine moiety is modified at different positions, and 2Ј,3Ј-SL-ATP, where the radical is bound to the ribose. ESR spectroscopy enables the evaluation of rather weak interactions of proteins and nucleotides with dissociation constants in the upper micromolar range that are usually not detectable by techniques like fluorescence spectroscopy. From the three compounds tested, Hsp90 only showed binding affinity for the ribose-modified ATP analogue, indicating that the nucleotide binding fold of Hsp90 tolerates only modification in the ribose moiety of the nucleotide, while modification of the adenine moiety was not accepted by the protein. Fig. 4 shows the ESR spectra of Hsp90 in complex with different concentrations of 2Ј,3Ј-SL-ATP. The signals of the freely tumbling, unbound spin-labeled nucleotide dominate the middle part of the spectra. The signals of the protein-immobilized radicals can be visualized upon amplification of the low  and high field regions of the spectra. The 2A zz value of the protein-bound component is 67 G, indicative of rather high immobilization of the radical. The peak areas of the signals of the bound components increase with rising 2Ј,3Ј-SL-ATP concentrations (Fig.4, compare spectra A through D). At 800 M 2Ј,3Ј-SL-ATP, approximately 0.8 mol of SL-nucleotides were bound/mol of Hsp90 dimer. Half-maximal binding was reached at approximately 400 M 2Ј,3Ј-SL-ATP. Addition of 12 mM ATP to the solution (depicted in Fig. 4D) resulted in complete displacement of the protein-bound spin-labeled ATP (Fig. 4E), indicating that 2Ј,3Ј-SL-ATP competes with unlabeled ATP for the binding site.
Influence of p23 on ATP Binding by Hsp90 - Fig. 5 shows ESR spectra of Hsp90 preincubated with an equimolar concentration of p23 in 40 mM Hepes-KOH, 5 mM MgCl 2 , 5% glycerol, pH 7.5. Again, the signals of the free SL-ANP compound dominate the middle part of the ESR spectra. The signals of the bound component can be visualized at higher signal gain. The outermost splitting of the signals (2A zz value) did not differ from the signals described for Hsp90 in complex with 2Ј,3Ј-SL-ATP, indicating that no drastic conformational change had occurred, when Hsp90 had been preincubated with p23. The spectra indicate a second spectral component of the proteinimmobilized radical, best visible in the low field area in Fig. 5B, which might indicate a second type of nucleotide binding site. Increase of the total concentration of 2Ј,3Ј-SL-ATP leads to an increase of the signals of the bound component (Fig. 5, compare spectra A through D). At 800 M 2Ј,3Ј-SL-ATP, maximum binding of 1 mol of 2Ј,3Ј-SL-ATP/mol of Hsp90 dimer was reached. Again, half-maximal binding was at approximately 400 M. Addition of 13 mM ATP led to full displacement of the proteinbound SL-ANP. Importantly, we were not able to detect any binding of spin-labeled nucleotides to p23 alone (data not shown).
Influence of Molybdate on the Stability and ATP-binding Properties of Hsp90 -Since molybdate seems to promote stable associations of Hsp90 with co-factors or substrate proteins significantly (see Introduction), we analyzed its effect on the structure and ATP-binding properties of Hsp90.
First we examined stabilizing effects of ATP on Hsp90. Previously, it was shown that ATP does not stabilize Hsp90 against guanidinium chloride-induced unfolding (17), whereas the stability of DnaK was significantly increased by ATP (46). Here, we used the thermal unfolding of Hsp90 to monitor effects of ATP. While the intrinsic fluorescence decreases with temperature linearly up to 45°C, an increase in signal was observed between 45°C and 55°C, which is due to structural changes in the protein upon unfolding (47). The midpoint of thermal unfolding was determined to be 49°C. Addition of ATP or various concentrations of molybdate did not change the unfolding behavior and the midpoint of thermal unfolding of Hsp90 ( Fig. 6 and data not shown). Thus, both factors do not seem to exert a stabilizing effect on human Hsp90. Furthermore, in ESR experiments performed to analyze ATP binding to Hsp90, molybdate (10 mM) did not change the shape of the signal or the amount of 2Ј,3Ј-SL-ATP bound (data not shown).

DISCUSSION
The finding that Hsp90 binds ATP weakly but specifically reconciles a number of apparently conflicting reports on the ATP-binding properties of Hsp90.
Previous claims of tight ATP binding (18,20,21) and potent ATPase activity (22, 23) could not be confirmed for yeast (17) and human Hsp90 (this study), using a number of independent experimental techniques, which allow to detect strong interactions. Furthermore, ATP did not influence the chaperone function of Hsp90 in different in vitro folding assays (48 -55). Taken together, these findings lead to the suggestion that Hsp90 functions independently of ATP and is, in this respect, set apart from other members of the Hsp family such as Hsp70 and Hsp60. However, the question remained whether there is any, potentially weak, interaction of Hsp90 with ATP, since it was shown previously that association of Hsp90 with its partner protein p23 requires the presence of ATP (11)(12)(13)18). The only technique available that allows direct detection of weak binding of a small ligand to a protein is ESR spectrometry. Using this method and three different spin-labeled ATP analogues, we were able to demonstrate that ATP and Hsp90 associate with half-maximal binding occurring in the high micromolar range (400 M).
The concentration of ATP necessary for half-maximal binding was not significantly changed by the addition of p23, which itself did not bind ATP. However, addition of p23 seems to lead to a slight change in the signal, indicating formation of a different binding environment of the radical. At the moment it is not clear whether these interactions stabilize the p23/Hsp90 complex. In addition to ATP, the association of p23 and Hsp90 was previously shown to require the polyanion molybdate, which had been used before to stabilize Hsp90/steroid receptor and Hsp90/kinase complexes (30, 56 -58). In general, the chemical properties of molybdate, such as conversion of glucose to mannose (59 -61), are well characterized at acidic pH, but at neutral pH little is known about the charge and binding properties of this polyanion complex. Under the conditions used, we could not find any influence of molybdate on Hsp90 stability or its ATP binding properties. Therefore the underlying functional principle for the effects observed for molybdate in context with the Hsp90 complex remains enigmatic.
The apparent K D value of 400 M for ATP binding to Hsp90 is out of range for the tight binding assays previously employed (17,20,21). To put this constant into perspective, it is useful to compare it to K D values previously reported for other members of the Hsp family. For the bacterial chaperones GroEL and DnaK, K D values of 10 M (28) and 0.22 M (45), respectively, were determined, and in this study the K D for murine BiP was found to be 0.38 M. These values are 40 -2000-fold lower and even the K D value of BSA for ATP is about 8 times lower than that of Hsp90 (44). Given this relatively low affinity, the ques- The spectra were recorded at 19°C. The low and the high field regions were re-recorded at higher signal gain for better visualization of the signals of the protein-bound radicals. The signal gain for all amplified spectra was 1 ϫ 10 6 , the time constant was 2.5 s, the sweep time was 2500 s, and the sweep width was 120 G. The spectra were recorded at 19°C. The low and the high field regions were re-recorded at higher signal gain for better visualization of the signals af the protein-bound radicals. The signal gain for all amplified spectra was 1 ϫ 10 6 , the time constant was 2.5 s, the sweep time was 2500 s, and the sweep width was 120 G.
FIG. 6. Influence of ATP and molybdate on the thermal unfolding of Hsp90. Thermal unfolding of 1.7 M human Hsp90 was performed in 100 mM potassium phosphate, pH 6.8, without additives (Ⅺ) or in the presence of 2 mM MgATP (q) and in the presence of 20 mM Na 2 MoO 4 (å). To monitor unfolding of Hsp90, the fluorescence signal at 320 nm was recorded. The heating rate was 0.5°C/min. tion arises whether a K D value of 400 M allows association of Hsp90 with ATP in the cell. If one takes into account that cytosolic ATP concentrations are in the millimolar range and that Hsp90 is present in micromolar concentrations, one can calculate the amount of Hsp90 found in complex with ATP. Using Hsp90 concentrations ranging from 12.5 to 50 M and ATP concentrations of 1-2 mM, we found that the K D value we determined would be sufficient to saturate 70% of the Hsp90 molecules with ATP.
At present, the significance of weak ATP binding to Hsp90 is far from being understood. The best studied example of Hsp90 function is its interaction with steroid receptors, which allows them to be kept in an activable state (4 -8). In this reaction cascade, some steps involving Hsp90 are clearly ATP-independent, while others, which are ATP-dependent, seem to involve Hsp70. A detailed analysis of ATP binding to Hsp90 seems to be required in comparison to Hsp70 to identify potential steps in the interaction with partner proteins and steroid receptors that may require ATP binding to Hsp90.