Cpr6 and Cpr7, Two Closely Related Hsp90-associated Immunophilins from Saccharomyces cerevisiae, Differ in Their Functional Properties*

Hsp90 is an abundant cytosolic molecular chaperone. It controls the folding of target proteins including steroid hormone receptors and kinases in complex with several partner proteins. Prominent members of this protein family are large peptidyl prolyl cis/trans isomerases (PPIases), which catalyze the cis/trans isomerization of prolyl peptide bonds in proteins and possess chaperone activity. In Saccharomyces cerevisiae, two closely related large Hsp90-associated PPIases, Cpr6 and Cpr7, exist. We show here that these homologous proteins bind with comparable affinity to Hsp90 but exhibit significant structural and functional differences. Cpr6 is more stable than Cpr7 against thermal denaturation and displays an up to 100-fold higher PPIase activity. In contrast, the chaperone activity of Cpr6 is much lower than that of Cpr7. Based on these results we suggest that the two immunophilins perform overlapping but not identical tasks in the Hsp90 chaperone cycle.

Hsp90 is an abundant cytosolic molecular chaperone. It controls the folding of target proteins including steroid hormone receptors and kinases in complex with several partner proteins. Prominent members of this protein family are large peptidyl prolyl cis/trans isomerases (PPIases), which catalyze the cis/trans isomerization of prolyl peptide bonds in proteins and possess chaperone activity. In Saccharomyces cerevisiae, two closely related large Hsp90-associated PPIases, Cpr6 and Cpr7, exist. We show here that these homologous proteins bind with comparable affinity to Hsp90 but exhibit significant structural and functional differences. Cpr6 is more stable than Cpr7 against thermal denaturation and displays an up to 100-fold higher PPIase activity. In contrast, the chaperone activity of Cpr6 is much lower than that of Cpr7. Based on these results we suggest that the two immunophilins perform overlapping but not identical tasks in the Hsp90 chaperone cycle.
PPIases 1 are enzymes that are able to catalyze the cis-trans isomerization of Xaa-Pro peptide bonds in short synthetic peptides as well as in proteins (1). The isomerization of these bonds is a slow process (2) and often a rate-determining step in protein folding (3)(4)(5)(6). PPIases have been found in all organisms and subcellular compartments studied so far (1,7,8). Until now three unrelated families of PPIases could be identified: parvulins, cyclophilins, and FKBPs (1,5,9). Because of their inhibition by the immunosuppressive drugs cyclosporin A (CsA) or FK506/rapamycin, the cyclophilins and FKBPs are also termed immunophilins (10 -13). Several large immunophilins with a molecular mass of about 40 -54 kDa are components of the Hsp90 chaperone complex. In higher eukaryotes, FKBP51, FKBP52, and Cyp40 act together with Hsp90; in yeast these are Cpr6 and Cpr7, two cyclophilins. The Hsp90 complex seems to regulate the conformation of its target proteins. These substrates include steroid hormone receptors. Here, the Hsp90 chaperone cycle is well established (reviewed in Refs. 14 -16).
In the case of the progesterone receptor, the receptor is bound to Hsp70 in an early complex (15,17). Then an intermediate complex is formed in which Hsp90, Hsp70, Hip, Hop, and perhaps Hsp40 participate. The last step of this activation cycle is a complex, consisting of the progesterone receptor, Hsp90, p23, and one of the large immunophilins. The specific function of the immunophilins in the chaperone cycle is far from clear. However, it is well established that progesterone receptor has to be bound in this complex to allow binding of the hormone, dimerization, and subsequently, interaction with DNA. In the absence of hormone, the receptor is released from the complex, and the cycle starts again. The large immunophilins interact with Hsp90 via tetratricopeptide repeats (18 -21). The partner site for the tetratricopeptide-repeat motifs is located in a 12-kDa carboxyl-terminal domain of Hsp90 (22,23). The tetratricopeptide-repeat proteins compete with each other for this binding site (18,19,24,25). In vitro large immunophilins are able to recognize and bind nonnative proteins. Cyp40, the human relative of Cpr6 and Cpr7, was found to be able to keep ␤-galactosidase in a folding-competent state (26). FKBP52 is able to suppress the aggregation process of citrate synthase (CS) and to bind unfolding intermediates of CS (27).
Although Cpr6 and Cpr7 share 47% sequence homology and 38% sequence identity (28), in vivo several differences have been found. Cpr6 and Cpr7 are both constitutively expressed, but only Cpr6 is additionally induced by heat shock (29). Depletion of Cpr7 led to a growth defect that could not be rescued by the overexpression of Cpr6 (28,30). In vitro it was shown that Cpr6 is able to catalyze the isomerization of Xaa-Pro peptide bonds (29,31), whereas Cpr7 is assumed to be inactive (32). Here we analyzed and compared the function and stability of Cpr6 and Cpr7. We demonstrate that significant differences exist between the two Hsp90 partner proteins.

MATERIALS AND METHODS
Proteins-Cpr6, Cpr7, and yeast Hsp90 were purified as described (33,34). Mitochondrial pig heart CS was obtained from Roche Molecular Biochemicals. CS was stored in TE buffer (50 mM Tris, 2 mM EDTA, pH 8.0). The reduced and carboxymethylated form of ␣-lactalbumin and casein were obtained from Sigma. The reduced and carboxymethylated form of ␣-lactalbumin (RCM-La) was stored in 100 mM Tris, 2 M NaCl, pH 8.0. Purified Sti1 was a kind gift of Paul Muschler (Technische Universität Mü nchen). The RNase T1(P55)-carrying plasmid was a kind gift of F. X. Schmid (University of Bayreuth). RNase T1(P55) was purified and modified (35) as described previously. The protein concentrations were determined using the extinction coefficients of 0.51, 0.88, 2.1, and 1.9 of Cpr6, Cpr7, RCM-La, and RCM-T1(P55) for a 0.1% solution at 280 nm and a path length of 10 mm, respectively. The extinction coefficients of the proteins were calculated according to Pace et al. (36).
The antibodies against Cpr6, Cpr7, and RCM-La were produced by standard procedures in rabbits. The RNase T1 antibody was a kind gift of U. Hahn (University of Leipzig). The Hsp90 antibody Spa840 was obtained from Stressgen Biotechnologies Corp. (Victoria, Canada). Immunoblots were probed with a 1:1000 dilution of the respective antibody.
Circular Dichroism Measurements-Near and far UV circular dichroism (CD) spectra were recorded in a Jasco J715 spectropolarimeter equipped with a peltier unit. Cpr6 and Cpr7 were dialyzed overnight against 100 mM potassium phosphate, pH 6.8. Near UV CD spectra were recorded from 250 to 350 nm in thermostated 0.5-cm quartz cuvettes at 25°C. Far-UV CD spectra were recorded from 197 to 250 nm in thermostated 0.1-cm quartz cuvettes at the same temperature. All spectra were buffer-corrected. Mean residue ellipticities for near and far UV CD spectra were calculated based on a mean residue weight of 112. Temperature transitions were recorded at a wavelength of 222 nm from 20 to 80°C with a heating rate of 0.5°C/min. To prevent vaporization, the protein solution was covered with mineral oil.
Biacore Measurements-Surface plasmon resonance data were collected on a Biacore X Instrument. Hsp90 was covalently linked to the CM5 chip via amine residues according to the supplier's instructions. Measurements were performed at a flow rate of 20 l/min at 20°C. To calculate the binding constants, two different approaches were used: (i) direct measurements in which different concentrations of the respective partner protein were injected onto the Hsp90-coated chip and (ii) competitive measurements in which increasing amounts of Hsp90 were added to the respective partner protein before injection. After complex formation, these samples were then injected onto the Hsp90 chip, and the amount of free partner protein was measured.
Data analysis for direct binding curves uses the linear relationship between the resonance signal RU and the amount of protein bound to the chip. RU max (Equation 1) gives the maximum signal, at which all Hsp90 molecules on the chip are saturated with partner protein.
For the competitive approach, data analysis was done using Equation 2. This equation can be derived from the equation for the equilibrium constant and the assumption that the total ligand concentration L 0 equals the sum of the free ligand concentration and the concentration of the complex.
RU 0 in this equation represents the initial value of the resonance signal without the addition of Hsp90, and RU min is the value obtained at maximal suppression of the signal with external Hsp90. c, as well as in the equation above, is the concentration of the protein added, whereas the value L 0 represents the concentration of partner protein, which was kept constant throughout the titration. For the different buffer conditions, different concentrations of partner protein were used in the competition experiments because the resonance signal proved to be sensitive to the buffer conditions. In the low salt buffers, partner protein concentrations were between 80 nM (Cpr6 and Cpr7) and 120 nM (Sti1). For the data analysis, these concentrations were used as L 0 values and kept constant during the data fitting. The statistical interpretation of the data sets was done with the program Scientist from MicroMath.
Prolyl Isomerase Activity-The PPIase activity was measured in a Amersham Pharmacia Biotech Biochrom 4060 spectrophotometer equipped with a thermostated cell holder using the protease-coupled assay (37). The X-Pro bond in the N-Succinyl-Ala-Xaa-Pro-Phe-4-nitroanilide peptide is only cleaved by chymotrypsin if the peptide bond is in the trans position. The increasing amount of the cleaved Pro-Phe-(4)-nitroanilide peptide can be observed photometrically at 390 nm. The protease-free assay was performed as described by Janowski et al. (38). Both assays were performed at 10°C with two different peptides, one containing Ala and the other one containing Leu at the P1 position.
RNase T1(P55) Refolding Experiments-The RNase T1(P55) refolding experiments were performed as described previously (39) with a Spex Fluoromax fluorescence spectrometer at 15°C. The spectral bandwidths were set to 1.5 and 3.5 nm for excitation and emission, respectively. For the competition experiments with RCM-La, the bandwidths for excitation and emission were set to 1.5 nm. The excitation wavelength was set to 268 nm, and the emission wavelength was set to 320 nm, respectively. The slow refolding of RNase-T1 under these conditions is a mono-exponential process. The rate constants of the reaction were determined using the program SigmaPlot 4.0 (Jandel Corp., Erkrath, Germany).
The analysis of the Michaelis-Menten kinetic experiments was performed by determining the initial velocities of RCM-T1 refolding from the progress curves of refolding in the presence of 125 nM Cpr6 under the conditions described above. The RCM-T1 concentration was varied between 0.1 and 10 M. The observed refolding rates were corrected for the contribution of the uncatalyzed refolding of RCM-T1. To estimate K m and V max , we made a Lineweaver-Burk plot with the corrected initial velocities. The values obtained from this plot (K m Х 9 M and V max Х 0.027 M s Ϫ1 ) were used for simulations according to the method of Konfron et al. (40), which accounts for both catalyzed and uncatalyzed prolyl isomerization in a peptide. The time course of RCM-T1 refolding in the presence of a PPIase is described by the following differential equation (Equation 3).  (41). These are the molecules responsible for the slow refolding reaction. At a given concentration of the enzyme, the k cat value can be calculated.
CS Assays-CS was thermally denatured by incubation at 43°C in 40 mM Hepes, pH 7.5. Aggregation of CS was measured by light scattering in stirred quartz cuvettes in a PerkinElmer MPF 44A luminescence spectrophotometer with a thermostated cell holder. Excitation and emission wavelengths were set to 500 nm, with a spectral bandwidth of 2 nm. The CS inactivation and reactivation assays were performed at 41°C in 40 mM Hepes, 10 mM KCl, pH 7.5. Assays were performed in a Amersham Pharmacia Biotech Ultrospec 3000 photometer with a thermostated cell holder. All assays were performed according to Buchner et al. (42) and Grallert et al. (43). Acetyl-CoA was from Roche Molecular Biochemicals, and oxaloacetic acid and 5,5Ј-dithiobis(2-nitrobenzoic acid) were purchased from Sigma.
High Performance Liquid Chromatography Size Exclusion Chromatography-The HPLC size exclusion chromatography experiments were performed with a Superdex 200HR column from Amersham Pharmacia Biotech that was equilibrated in 40 mM Hepes, 150 mM KCl, pH 7.5. In the case of the CS experiments, 1 mM dithioerythritol was added. Proteins were detected by fluorescence using a Jasco FP920 fluorescence detector at an excitation wavelength of 280 nm and an emission wavelength of 330 nm.

RESULTS
Structure and Stability of the Cpr6 and Cpr7-To gain insight into the conformation of Cpr6 and Cpr7, the two large Hsp90-associated immunophilins of Saccharomyces cerevisiae, CD spectra were recorded in the far UV range. These measurements showed a maximum below 200 nm and two minima at 208 and 222 nm, characteristic for ␣-helical proteins (Fig. 1A). Using the secondary structure prediction program SOPM (44), the ␣-helical content of Cpr6 was calculated to be 40%, and for Cpr7, it was 35%. Size exclusion chromatography was performed to determine the quaternary structure of Cpr6 and Cpr7. Both Cpr6 and Cpr7 eluted at time points that correspond to the respective monomer (data not shown). Having established that the Cprs are monomeric helical proteins, we next determined their stability. The stability against thermal unfolding was measured by recording the decrease of the far UV CD signal at 222 nm. Although Cpr7 starts to unfold at temperatures above 45°C, Cpr6 is stable up to 52°C (Fig. 1B). Thus, although the midpoints of the thermal denaturation are the same, Cpr7 is markedly less stable than Cpr6. Due to aggregation of both proteins, the transitions were not reversible.
Binding of Cpr6 and Cpr7 to Hsp90 -Cpr6 and Cpr7 form defined complexes with Hsp90 (25, 45-47). We were interested in determining the relative affinities of the immunophilins for Hsp90 by the plasmon resonance technology. In addition to Cpr6 and Cpr7, we included the Hsp90 partner protein Sti1, the yeast homologue of Hop, in this analysis. To directly compare the affinities of Cpr6, Cpr7, and Sti1, Hsp90 was covalently coupled to the chip surface. The potential bias introduced by coupling Hsp90 covalently to the chip surface was avoided by determining the binding constants for the proteins in solution by a competitive approach. This can be achieved by first incubating constant concentrations of a partner protein and increasing concentrations of Hsp90 in solution. In a next step, this solution is added to the chip. Unliganded partner protein will now bind to Hsp90 immobilized on the chip surface.
In a first set of experiments, binding of Cpr6, Cpr7, or Sti1 to immobilized Hsp90 was analyzed in a low salt buffer. Since the His tag could give rise to unspecific binding, 2 we included EDTA in a second set of experiments. We found that these partner proteins bind to Hsp90 with similar affinity in the range of 14 to 57 nM ( Fig. 2 and Table I).
Taken together these results show that Cpr6 and Cpr7 bind with similar affinity to Hsp90. The differences in binding between Cpr6 and Cpr7 were 2-4-fold. In addition, the competition experiments indicated a stoichiometry of one partner protein bound to one Hsp90 monomer.
Prolyl Isomerase Activity of the Large Immunophilins-First we examined the activity of Cpr6 and Cpr7 in the proteasecoupled peptide assay (37). In contrast to recently published results, we could show that Cpr7 is able to catalyze the isomerization of Xaa-Pro peptide bonds, similar to Cpr6 (Table II). The PPIase activity of both Cpr6 and Cpr7 is inhibited by CsA. Both large immunophilins from S. cerevisiae exhibited pseudo first order kinetics, and the apparent rate constants increased linearly with enzyme concentrations (Fig. 3 and data not  shown). The specificity constants k cat /K m could be estimated from the slope in Fig. 3. The catalytic efficiency of Cpr6 of 4.8 ϫ 10 5 M Ϫ1 s Ϫ1 is comparable with the value published by Warth et al. (29). For Cpr7 (7.5 ϫ 10 4 M Ϫ1 s Ϫ1 ), a 6-fold lower activity was determined. Because for some PPIases the catalytic activity strongly depends on the amino acid that precedes the proline residue, we tested the ability of Cpr6 and Cpr7 to isomerase Xaa-Pro bonds in a peptide in which this amino acid was replaced by a leucine (Leu-Pro instead of Ala-Pro). We were not able to detect a significant difference of the catalytic activity of Cpr6 and Cpr7 for the two peptides, suggesting that they do not discriminate strongly between different amino acids. Since Cpr7 is less stable than Cpr6, it was possible that the low activity of Cpr7 was due to degradation in the protease-coupled PPIase assay. Therefore we performed a different, proteasefree prolyl isomerase assay (38). The k cat /K m values obtained here were comparable with those of the protease-coupled assay (data not shown). Thus, proteolytic digestion did not influence  the activity in the PPIase assays. Taken together, the results show that both proteins are active as PPIases and that the catalytic efficiency of Cpr6 is 6-fold higher than that of Cpr7. Catalytic Activity of the Large Immunophilins in the Acceleration of the Slow Refolding of RCM-RNase T1(P55)-Although peptides are a good model for analyzing prolyl isomerase activity, they do not allow conclusions to be drawn on the ability of these enzymes to catalyze slow-refolding reactions in proteins. This is especially important for a prolyl isomerase exhibiting chaperone properties. To test this ability, we determined the influence of Cpr6 and Cpr7 on the refolding of reduced and carboxymethylated RNase T1(P55) (39,48). The folding mechanism of RCM RNase T1(P55) is well described and characterized (41). In the native conformation, this protein contains one cis proline peptide bond, which determines the rate of refolding. The acceleration of refolding could be achieved with different PPIases (49,50). To determine whether the catalytic activity of Cpr6 and Cpr7 is influenced by high salt concentrations, which are required in the RNase T1 refolding assay, we performed the protease-coupled assay in the presence of increasing amounts of NaCl. Our results show clearly that neither the spontaneous reaction itself nor the activity of the proteins tested is affected by salt concentrations up to 2 M NaCl (cf. Table II). The first order rate constant of folding increased in a linear fashion with the concentration of Cpr6 (Fig. 4A). The specificity constants k cat /K m , which could be determined from the slopes in the inset of Fig. 4A, showed that Cpr6 exhibited a much higher k cat /K m value than Cpr7 (cf .  Table II). Interestingly, the specific rate constant of Cpr7 is 100-fold lower than that of Cpr6 (3.6 ϫ10 4 M Ϫ1 s Ϫ1 ).
To gain insight in the kinetic properties of Cpr6, we determined the parameters K m and k cat . In this experiment, the concentration of Cpr6 was kept constant at 125 nM, and the RCM-T1 concentration was increased from 0.1 to 10 M. Because both catalyzed and uncatalyzed folding reactions occur in the presence of a PPIase, the initial rates were corrected for the values of the uncatalyzed reaction. From a Lineweaver-Burk plot, obtained with the corrected initial velocities, we estimated K m and V max . These values were used for a simulation according to Konfron et al. (40). In these simulations we obtained a The activities were determined as described under "Materials and Methods." The k cat /K m values were determined from the slopes of plots as shown in Fig. 2   value of 6.2 M for K m and 0.3 s Ϫ1 for k cat . The ratio of these two values corresponds to the value obtained from the slope in Fig.  4A. The good agreement of the values (3.6 ϫ 10 4 M Ϫ1 s Ϫ1 and 4.9 ϫ 10 4 M Ϫ1 s Ϫ1 ) confirms our analysis of the kinetic data. Because RCM-T1 starts to aggregate at high concentrations (39) and because of the very low k cat /K m value of Cpr7, this assay could not be performed with Cpr7.
Influence of an Unfolded Protein on the Cpr6-mediated Catalysis of RNase T1(P55)-In the RNase T1 assay, competition experiments with another unfolded protein allows further insight into the binding properties of the folding catalyst for its substrate (39). Provided that PPIases bind their substrates in a nonspecific manner, the presence of another unfolded protein should interfere with the binding of the substrate and, thus, slow down the refolding rate. As a competitor, we used the RCM-La, which, under the conditions used, represents a soluble, permanently unfolded protein (39). Small PPIases were not inhibited by RCM-La. However, for the large immunophilin trigger factor, it was shown that RCM-La competitively inhibits its catalytic activity (39). This suggested that RCM-La and RNase T1(P55) bind with similar rate constants to the same binding site on trigger factor. When we performed this assay with Cpr6, we obtained a different picture. For Cpr6, the experiments showed that the refolding rate of RNase T1(P55) is further accelerated when RCM-La is added (Fig. 4B, cf. Table  II). The acceleration followed a saturation curve and reached a plateau at 12 M RCM-La. The addition of CsA resulted in a deceleration of the refolding reaction that is equal to the velocity constant of the uncatalyzed refolding reaction of RNase T1(P55) (data not shown). Neither RCM-La nor CsA alone affected the slow refolding of RNase T1(P55) (data not shown). When the amount of Cpr6 was increased (from 21 nM up to 500 nM), the folding reaction was accelerated up to 1.6-fold (data not shown). Again, because of its very low activity, this assay could not be performed with Cpr7.
Because the interaction of prolyl isomerases with peptides could also be influenced by the addition of an unfolded protein, we checked the effect of RCM-La on the activity of the PPIases in the protease-coupled assay as well as in the protease-free assay (38). We were not able to detect any influence of RCM-La on Cpr6 and Cpr7 (cf . Table II and data not shown). Thus, the active site of the PPIase domain is not directly affected by RCM-La.
Chaperone Properties of Cpr6 and Cpr7-A key feature of molecular chaperones is their ability to suppress the aggregation of proteins under stress conditions (51,52). To analyze the chaperone properties of Cpr6 and Cpr7, we examined their ability to inhibit the thermal aggregation of CS in vitro. Dimeric CS aggregates spontaneously upon incubation at 43°C (42). Both Cpr6 and Cpr7 were able to suppress the aggregation of CS in a concentration-dependent manner (Fig. 5A). In this assay, Cpr7 had significantly higher activity than Cpr6. Cpr7 was able to suppress CS aggregation completely at a ratio of CS to Cpr7 of 1:4. This is in agreement with results by Bose et al. (27) demonstrating that rabbit FKBP52 is able to completely suppress the aggregation of CS at a molar ratio of 1:6. In contrast, Cpr6 showed nearly maximum suppression of aggregation only at a ratio of 1:16 (Fig. 5A). The inhibition of the prolyl isomerase activity by the addition of CsA had no influence on the chaperone properties of the examined proteins (data not shown). Also IgG, used as a control for unspecific protein effects, did not affect the aggregation reaction (Fig. 5A). Having established that Cpr6 and Cpr7 were able to interfere with late steps of the unfolding pathway of CS, we were interested in determining their influences on early steps of this reaction. To this end we examined the inactivation kinetics of CS at 41°C (27,42,43,53). Under the conditions used here, CS loses its activity within 30 min (Fig. 5B). The addition of Cpr7 or Cpr6 led to a deceleration of the inactivation process. This effect is strongly dependent on the molar ratio of chaperone to CS. The addition of a control protein, IgG, had no significant effect on the inactivation process (data not shown). As in the aggregation assay, Cpr7 showed a higher chaperone activity than Cpr6. At equimolar ratios of the chaperones, CS inactivation was two times slower in the presence of Cpr7 compared with Cpr6.
Next, we were interested in determining in which way the formation and population of CS unfolding intermediates was influenced by the presence of Cprs. Previously, the unfolding process of CS was shown to involve inactive, dimeric intermediates (53). The addition of oxaloacetic acid, a substrate known to stabilize CS (53,54), leads to a reactivation of inactive dimeric CS molecules. This allows determination of the number of these intermediates during the unfolding reaction (16,53). Fig. 5C shows the time course of formation of reactivatable intermediates in the absence and presence of Cprs. Both Cpr6 and Cpr7 significantly increased the amount of reactivatable intermediates. Furthermore, these intermediates were present for a longer period of time. Again, Cpr7 was more effective than Cpr6 in maintaining reactivatable intermediates. Taken together, these results suggest that Cpr6 and Cpr7 prevent irreversible aggregation of CS by interacting transiently with specific unfolding intermediates. DISCUSSION S. cerevisiae possesses two large immunophilins of the cyclophilin type that are specific cofactors of Hsp90. Previous studies have shown that the deletion of Cpr6 has no effect on the viability of yeast cells (55,29,31). In contrast, the deletion of Cpr7 caused a growth defect (28,45). Additionally, Cpr7-depleted cells are more sensitive to geldanamycin, a specific Hsp90 inhibitor, and are defective in restoring the full glucocorticoid receptor activity (30). Our results demonstrate that purified Cpr6 and Cpr7 differ in their stability, their ability to isomerase Xaa-Pro peptide bonds, and in their activity as molecular chaperones, although they share 47% sequence homology and 38% sequence identity (28). We show that, based on CD measurements and secondary structure prediction, Cpr6 and Cpr7 are predominantly ␣-helical proteins. The highly conserved PPIase domain consists of an eight-stranded antiparallel ␤-sheet and two ␣-helices. Thus, the putative chaperone domain including the three tetratricopeptide repeats (28) seems to be mainly ␣-helical. Interestingly, only Cpr6 and not Cpr7 is increasingly expressed at elevated temperatures (29,46). In agreement with these results, we find that Cpr6 is significantly more stable against heat denaturation than Cpr7.
Using a plasmon resonance approach we were able to determine binding constants for the two immunophilins and Hsp90. Both immunophilins, as well as Sti1, which was included as a control, bind to Hsp90 with low nanomolar binding constants (14 and 57 nM for Cpr6 and Cpr7, respectively; 32 nM for Sti1). In both this and the previous study (25), the stoichiometry of the complexes is one partner protein binding to one Hsp90 monomer.
The binding constants of Cpr6 or Cpr7 for Hsp90 differed by a factor of 2-4. We conclude from these data that both proteins bind with similar affinity and the same stoichiometry to Hsp90. It is therefore unlikely that one of the immunophilins is preferentially associated with Hsp90 in the absence of additional factors. Despite their binding to Hsp90, the two immunophilins are functionally strikingly different. In contrast to previously published results (32), we were able to measure a low, but specific PPIase activity for Cpr7. However, the efficacy of the catalysis of prolyl isomerization by Cpr7 is about 6-fold lower than that of Cpr6 for peptides. This difference is drastically increased in the isomerization of proline peptide bonds in proteins. In the RNase T1 assay, Cpr7 showed a 100-fold lower activity than Cpr6. The specific catalytic rate constant of Cpr6 in the RNase T1 refolding assay is in the range of that of the small immunophilin FKBP12 (56). The K m of Cpr6 (6.2 M) is 9-fold lower than that of the trigger factor (0.7 M), reflecting the difference in their catalytic activities. A sequence alignment the PPIase domains of hCyp18, hCyp40, Cpr3, Cpr6, and Cpr7 showed that from 73 residues that are conserved in these proteins, 19 residues are different in Cpr7 (data not shown). These exchanges could influence binding and catalysis and, thus, be responsible for the low PPIase activity of Cpr7. One may speculate that the PPIase activity of Cpr7 may have evolved for special substrates. It is interesting to note that a conserved proline of glucocorticoid receptor is required for the stabilization of Hsp90 heterocomplexes and receptor signaling (57).
Although Cpr7 is less efficient than Cpr6 as a prolyl isomerase, it is the more potent chaperone. The efficiency of Cpr7 is comparable with that of FKBP52, which was found to suppress the aggregation of CS completely (27). For Cpr6, suppression of aggregation was observed only at high ratios of Cpr6 to CS. This is reminiscent of p23, another protein that is found in complexes with Hsp90 (27,58). Both Cpr6 and Cpr7 slow down the thermal inactivation kinetics of CS, suggesting that they interact transiently with the substrate. This results in an increasing amount of reactivatable intermediates of CS. In vivo the deletion of Cpr7 led to a decrease of steroid hormone receptor activation that could not be rescued by overexpression of Cpr6 (32,45). Additionally, it was shown that the PPIase domain of Cpr7 is dispensable for the activation of steroid hormone receptors (32). Thus, these effects may be due to the different chaperone properties of the immunophilins. It is well possible that in vivo Cpr6 and Cpr7 are specialized for a certain protein or a group of proteins. This is supported by recent results showing that FKBP52 is able to bind directly to the glucocorticoid receptor (59) as previously suggested (27).
An important point in understanding the function of large immunophilins is the interplay between PPIase activity and the binding of nonnative proteins. Competition experiments with RCM-La on the ability of Cpr6 and Cpr7 to isomerize Xaa-Pro peptide bonds in short peptides suggest that the prolyl isomerase active site and the binding site for polypeptides are located in different parts of the proteins. In proteins, the ability of Cpr6 to catalyze the isomerization of proline-peptide bonds is enhanced by the addition of RCM-La, used as competitive inhibitor. In contrast, RCM-La inhibits the PPIase activity of trigger factor, supposedly due to the competitive binding of substrate and unfolded protein to the same binding site (39). This interpretation does not apply for Cpr6. A plausible explanation for our results is that RNase T1 can bind either to the PPIase domain or to the putative chaperone domain of Cpr6.
Taken together our data show that, although the overall architecture of the proteins and their interaction with Hsp90 is remarkably similar, Cpr6 and Cpr7 exhibit striking functional differences. Although Cpr6 is a strong prolyl isomerase and a weak chaperone, Cpr7 is a weak prolyl isomerase but a potent chaperone. The marked difference in their ability to interact with nonnative proteins may explain why Cpr7 is more important for the activity of Hsp90 complexes in yeast. It remains to be seen whether this reflects the specific requirements for folding different subgroups of Hsp90 substrates.