Domain Structure of the HSC70 Cochaperone, HIP*

The domain structure of the HSC70-interacting protein (HIP), a 43-kDa cytoplasmic cochaperone involved in the regulation of HSC70 chaperone activity and the maturation of progesterone receptor, has been probed by limited proteolysis and biophysical and biochemical approaches. HIP proteolysis by thrombin and chymotrypsin generates essentially two fragments, an NH 2 -terminal fragment of 25 kDa (N25) and a COOH- terminal fragment of 18 kDa (C18) that appear to be well folded and stable as indicated by circular dichroism and recombinant expression in Escherichia coli . NH 2 -terminal amino acid sequencing of the respective fragments indicates that both proteases cleave HIP within a predicted (cid:1) -helix following the tetratricopeptide repeat (TPR) region, despite their different specificities and the presence of several potential cleavage sites scattered throughout the sequence, thus suggesting that this region is particularly accessible and may constitute a linker between two structural domains. After size exclusion chromatography, N25 and C18 elute as two distinct and homogeneous species having a Stokes radius of 49 and 24 Å, respectively. Equilibrium sedimentation and sedimentation velocity indicate that N25 is a stable dimer, whereas C18 is monomeric in solution, with sedimentation coefficients of 3.2 and

Molecular chaperones of the 70-kDa heat shock protein (HSP70) 1 family play an essential role in protein biogenesis (1-3) through cycles of binding and hydrolysis of ATP coupled to cycles of binding and release of polypeptide substrates (4,5). During this process, HSP70 switches between two forms, an ADP-bound state and ATP-bound state, depending on the efficiency of ATP hydrolysis and that of ADP to ATP exchange and on the modulation by regulatory proteins called cochaperones. In Escherichia coli HSP70 (DnaK), the cochaperone DnaJ activates the hydrolysis of ATP by DnaK, thereby stabilizing the protein in its high affinity ADP form, while the cochaperone GrpE stimulates the exchange of ADP by ATP, thus helping the protein to switch back to its low affinity ATP form (6 -9). In mammals, however, the situation is less clear, and only homologues of the DnaK-ATPase stimulation factor DnaJ have been found in all cell compartments (10), the eukaryotic homologues of the ADP/ATP exchange factor GrpE being restricted to mitochondria and chloroplast, two organelles of prokaryotic origin (11). Current models of the regulation of the mammalian HSP70 ATPase cycle involve at least three factors, HSP40, Bag-1, and HIP (12)(13)(14)(15)(16)(17).
HIP, a protein of 368 residues, regulates HSC70 chaperone activity by binding to its NH 2 -terminal domain and stabilizing the protein in its ADP form, which has a high affinity for the polypeptide substrates (18). It also participates in the maturation of the progesterone receptor by cooperating with HSP70, HSP90, and in the maturation complex (19). High resolution structure of HIP has not been determined, but the polypeptide is assumed to comprise a series of consecutive modules based on sequence alignment criteria and mutagenesis studies (20,21). On this basis, the protein appears to be composed of an NH 2 -terminal module (residues 1-15) responsible for the protein oligomerization, a central TPR module (residues 113-214) involved in HSC70 binding followed by a highly charged region (residues 229 -271) and a GGMP-rich module (residues 282-310) with unknown function. Although this structural representation may give some insight into the overall organization of the protein and its possible relation to function, it gives little information on the number, the nature, and the actual boundaries of the domains in the folded protein. Since protein domains are usually made of well structured, independently folded units that are linked together by solvent accessible, unstructured regions (22,23), limited proteolysis has been the classical approach used to define domain organization of a protein (24 -26). This approach takes advantage of the fact that mild proteolysis will affect preferentially unstructured accessible regions rather than well structured and compactly folded domains (27)(28)(29).
In this report, we describe direct studies of the domain organization of HIP based on limited proteolysis of the native protein as well as structural and functional characterization of the isolated domains.

EXPERIMENTAL PROCEDURES
All size exclusion chromatography experiments were performed using an AKTA-FPLC system and columns from Amersham Biosciences, Inc.
Limited Proteolysis-Recombinant HIP was expressed and purified as described by Velten et al. (32). HIP was submitted to proteolysis by thrombin (10 Ϫ4 units/g of HIP/l) or ␣-chymotrypsin (2 ϫ 10 Ϫ6 units/g of HIP/l) at 37°C in buffer A (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA) containing 1 mM ␤-mercaptoethanol and 2.5 mM CaCl 2 . At the indicated times, an aliquot of the reaction mixture was removed, treated by phenylmethylsulfonyl fluoride at a final concentration of 1 mM to stop the reaction, and analyzed by 12% SDS-PAGE.
NH 2 -terminal Sequence Determination of Proteolysis Fragments-Proteolysis fragments (ϳ10 g each) were separated by 12% SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Roche Molecular Biochemicals) using the Transblot cell from Bio-Rad. Proteins corresponding to the bands on the polyvinylidene difluoride membrane were then subjected to NH 2 -terminal automatic sequence determination.
Circular Dichroism-The CD spectra of HIP, N25, and C18 at about 0.5 mg/ml in 20 mM sodium phosphate at pH 8 were acquired with a Jobin-Yvon CD6 spectropolarimeter at room temperature. To optimize the signal/noise ratio, the spectra were recorded on two contiguous regions, one between 182 and 200 nm, using an integration time of 5 s per step, and the second between 200 and 260 using a 1-s integration time per step. A constant step of 0.5 nm was used in both regions. Each spectrum results from averaging five successive scans. The base line was acquired under the same conditions.
The CD intensities were converted in terms of differential molar extinction per residue. The protein concentrations were determined from amino acid analyses of the samples, using alanine, or tyrosine or phenylalanine as reference. The normalized spectra were then deconvoluted using the Varselec program (30).
Purification of N25 and C18 Prepared by HIP Proteolysis-HIP was subjected to proteolysis by thrombin (57 ϫ 10 Ϫ6 units/g of HIP/l) for 40 min at 37°C, and the reaction was stopped by the addition of phenylmethylsulfonyl fluoride at a final concentration of 1 mM. To separate the two domains, the reaction mixture was injected at room temperature onto a Superdex 200 HR10/30 column preequilibrated in buffer A. Fractions of 0.5 ml were collected at a flow rate of 0.5 ml/min, and those containing either N25 or C18 were separately concentrated by ultrafiltration (YM10 membrane; Amicon) and stored at Ϫ80°C. Protein concentration was determined by the method of Lowry using bovine serum albumin as a standard.
Expression and Purification of Recombinant Proteins-Vectors expressing N25 or C18 fused to a histidine tag at the amino-terminal extremity were constructed by subcloning the corresponding polymerase chain reaction-amplified fragments of the rat HIP cDNA into a pET28 plasmid (Novagen). N25 contains residues 1-226, and C18 contains residues 227-368 of rat HIP. The two constructions were controlled by nucleotide sequencing.
Tagged N25 and C18 were expressed and purified by affinity chromatography on a Ni 2ϩ -nitrilotriacetic acid column, according to the protocol of the manufacturer (Qiagen). The plasmid expressing the C15 mutant was kindly provided by David Smith and Viravan Prapapanich (31). The corresponding protein was purified as described for HIP (32).
Electrophoresis-PAGE in denaturing conditions (SDS) was carried out in 0.75-mm-thick 12% acrylamide slab gels. Gels were run using the Mini-Protean II apparatus from Bio-Rad. Western blot was performed by transferring the proteins, after migration on a 15% acrylamide gel, to a nitrocellulose membrane using the Transblot cell from Bio-Rad. The membrane was then incubated with a monoclonal anti-Myc antibody diluted 1:2000 (Invitrogen). This antibody is conjugated to horseradish peroxidase and therefore has been detected by chemiluminescence using an ECL kit (Amersham Biosciences).
Size Exclusion Chromatography-Samples of N25 and C18 in buffer A were loaded at room temperature on a Superdex 200 HR10/30 preequilibrated in buffer A. Fractions of 0.5 ml were collected at a flow rate of 0.5 ml/min, and absorbance was measured at 280 nm. Proteins of known molecular weight and Stokes radius were used for the calibration curve representing the Stokes radius (R S ) as a function of the distribution coefficient K av (33,34). The molecular mass standards used were ferritin (440 kDa, R S ϭ 61 Å), catalase (232 kDa, R S ϭ 52.2 Å), aldolase (158 kDa, R S ϭ 48.1 Å), albumin (67 kDa, R S ϭ 35.5 Å), ovalbumin (43 kDa, R S ϭ 30.5Å), chymotrypsinogen A (25 kDa, R S ϭ 20.9 Å), and RNase A (13.7 kDa, R S ϭ 16.4 Å).
N25 and C18 molecular masses could be estimated by a combination of the Svedberg and Stokes equations, using the experimental Stokes radii and sedimentation coefficients.
Sedimentation Velocity-Sedimentation velocity experiments were performed on a Beckman Optima XL-A analytical ultracentrifuge equipped with an AnTi 60 titanium four-hole rotor with two-channel 12-mm path length centerpieces, as previously described (35). The compartments were filled with 400 l of three different concentrations of N25 (0.3, 0.5, and 1 mg/ml) or C18 (0.1, 0.25, and 0.5 mg/ml) in buffer A and centrifuged at 60,000 rpm at 4°C. Radial scans of the absorbance at 280 or 230 nm were taken at 5-or 8-min intervals, respectively, for N25 and C18. Experimental base lines were measured for each sample at the end of the run. N25 and C18 remained stable during the time of the run as shown by SDS-PAGE before and after the run. Data analysis was performed, and hydrodynamic parameters were evaluated as described in Ref. 32. The molecular mass of the monomer (N25, 25,596 Da; C18, 17,966 Da) and the partial specific volume at 4°C (v N25 ϭ 0.720 cm 3 ⅐g Ϫ1 ; v C18 ϭ 0.714 cm 3 ⅐g Ϫ1 ) and at 20°C (v N25 ϭ 0.726 cm 3 ⅐g Ϫ1 ; v C18 ϭ 0.720 cm 3 ⅐g Ϫ1 ), calculated from the amino acid composition, and the solvent density ( 4°C ϭ 1.00462 g/cm 3 ; 20°C ϭ 1.00294 g/cm 3 ) were obtained as previously described (32). For C18, the amino acid composition takes into account the addition of Myc epitope and the histidine tag at the COOH terminus of the protein. The degree of hydration of the N25 and C18 proteins (0.495 g H 2 /g protein and 0.450 g H 2 /g protein , respectively) was estimated on the basis of the amino acid composition and corrected as described in Ref. 32. Thus, degrees of hydration of 0.346 g H 2 /g protein for N25 and 0.315 g H 2 /g protein for C18 were used for the calculation of the axial ratio a/b of the hydrodynamically equivalent prolate ellipsoid of revolution. The frictional ratio f/f o was calculated using the corrected sedimentation coefficient, s 20,w 0 , and the molecular weight obtained by sedimentation equilibrium.
Sedimentation Equilibrium-Molecular mass of N25 and C18 proteins was determined by sedimentation equilibrium at 4°C using three loading concentrations of N25 (0.3, 0.5, and 1 mg/ml) or C18 (0.075, 0.15, and 0.3 mg/ml) in buffer A and three rotor speeds (9000, 12,000, and 17,000 rpm for N25; 14,000, 20,000, and 28,000 rpm for C18). Radial scans of the absorbance at 280 nm for N25 or 230 nm for C18 were taken at 3-h intervals, and samples were judged to be at equilibrium by the absence of systematic deviations in overlaid successive scans and when a constant average molecular weight was obtained in plots of molecular weight versus centrifugation time. N25 and C18 proteins remained stable during the time of the run as shown by SDS-PAGE before and after the run.
Multiple data sets were analyzed by nonlinear least-squares procedures provided in the Beckman Optima XL-A software package. Base line offset was obtained from meniscus depletion of the sample at 40,000 rpm at the end of the run. Data analysis was performed as previously described (32).
Rhodanese Aggregation Assay-Rhodanese was denatured in 6 M guanidinium HCl, 30 mM Tris-HCl, pH 7.5, at 25°C for 1 h and then diluted 100-fold to a final concentration of 0.5 M into 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, in the presence or absence of 1 M of effector (HIP, N25, or C18) or lysozyme as a control. Turbidity of the samples under constant agitation was measured at 320 nm for 20 min at 25°C.
In Vitro Protein Binding Studies-10 M histidine-tagged protein (HIP, N25, or C18) was incubated with 20 l of Ni 2ϩ -nitrilotriacetic acid bed resin (Quiagen) in a final volume of 40 l in buffer B (20 mM Tris-HCl, pH 8, 20 mM KCl) for 30 min at 4°C. The resin was washed three times with 0.5 ml of buffer B and was then incubated for 90 min at 4°C with a 20 M concentration of the ATPase domain of HSC70 (N-HSC70) or citrate synthase as a control, in buffer B containing 0.1 mM ADP in a final volume of 40 l (N-HSC70 was purified as described (36). The resin was washed four times with 0.5 ml of buffer B containing 0.1 mM ADP. Specifically retained protein was eluted by incubation for 30 min at 37°C with buffer B containing 1 mM ATP in a final volume of 40 l. Tagged protein was then eluted by incubation for 10 min at 4°C with buffer B containing 250 mM imidazole in a final volume of 40 l. The washing fractions were trichloroacetic acid-precipitated. All fractions were analyzed by SDS-PAGE.

Limited Proteolysis of HIP and Analysis by Western Blot and NH 2 -terminal Sequencing of the Fragments
HIP was submitted to a mild proteolysis by thrombin and chymotrypsin for different time periods and subsequently an-alyzed by polyacrylamide gel electrophoresis. As shown in Fig.  1A, proteolysis by thrombin led to the complete disappearance of HIP and the formation of essentially two fragments with apparent molecular masses of 35 and 20 kDa. Longer incubation times with the protease resulted in further cleavage of the 35-kDa fragment, yielding a 30-kDa fragment. Proteolysis by chymotrypsin, a protease having a different specificity, gave two major fragments of 36 and 18 kDa and two minor fragments of 30 and 13 kDa. These 30-and 13-kDa fragments seemed to be formed to the detriment of the 36-and 18-kDa fragments, as indicated by the kinetics of cleavage and the fact that the former fragments accumulated at the expense of the latter.
This was corroborated by NH 2 -terminal sequencing of the different fragments and comparison of the amino acid sequences obtained with that of HIP. As shown in Table I, HIP was cleaved between residues 237 and 238 by chymotrypsin, giving the two major fragments of 36 and 18 kDa apparent molecular mass, which in turn were cleaved at their COOHterminal extremity, resulting in the formation of the 30-and 13-kDa fragments, respectively, as indicated by the nature of their NH 2 termini (Table I). Similarly, thrombin cleaved HIP between residues 226 and 227, within the same main region as that recognized by chymotrypsin. Thus, the 35-kDa fragment (thrombin) and 36-kDa fragment (chymotrypsin) are NH 2 -terminal fragments, while the 20-kDa fragment (thrombin) and 18-kDa fragment (chymotrypsin) are COOH-terminal fragments of HIP.
Confirmation of this came from Western blot analysis of the proteolysis products, since HIP used in this study bears a Myc epitope at its COOH-terminal extremity. As shown in Fig. 1B, only the 20-kDa fragment obtained after thrombin cleavage and the 18-kDa fragment obtained after chymotrypsin digestion are revealed by an anti-Myc antibody.
Altogether, these results are summarized in Fig. 2, where the cleavage sites are mapped onto the primary structure and sequence modules of HIP as determined previously by sequence alignment, structural modeling, and mutagenesis studies (20,21,32). Thus, both proteases cleave HIP within a highly charged region that is adjacent to the TPR region and predicted to have an ␣-helical structure, yielding an NH 2 -terminal fragment of 226 (thrombin) or 237 (chymotrypsin) residues and a COOH-terminal fragment of 160 (thrombin) or 149 (chymotrypsin) residues. Since the work that follows was performed with thrombin proteolysis fragments, and for clarity, the 35-kDa fragment will be named N25, and the 18-kDa fragment will be named C18 (see also Table I), based on their theoretical molecular mass and their respective positions in the HIP sequence.

Purification and Recombinant Expression of HIP Proteolysis Fragments
To obtain large quantities of N25 and C18 fragments of HIP for subsequent purification and analysis, limited proteolysis by thrombin was performed on a preparative scale, and the resultant reaction mixture was subjected to size exclusion chromatography. As shown in Fig. 3, two well defined peaks are obtained, peaks 1 and 2, having apparent molecular masses of about 170 and 30 kDa, respectively, that correspond to the N25 and C18 as shown by SDS-PAGE (Fig. 3, inset). Thus, the two fragments of HIP eluted as two distinct species, having well defined molecular masses and Stokes radii, indicating either that they do not interact in the entire protein or that their interaction is lost after cleavage. Furthermore, the fact that N25 elutes as a 170-kDa protein and C18 as a 30-kDa protein, while their molecular mass deduced from amino acid composition corresponds to 25 and 18 kDa, respectively, indicates that N25 is oligomeric whereas the C18 is rather monomeric.
Since these results suggested that N25 and C18 fragments may correspond to two independent structural domains, recombinant expression in E. coli of cDNAs corresponding to residues 1-226 (N25) and 227-368 (C18) of HIP fused to a histidine tail have been performed using pET expression vectors. Significant amounts of the respective proteins have been obtained in a soluble form and purified using affinity chromatography (results not shown). Recombinant N25 and C18 were found to be indistinguishable from their proteolytic counterparts in terms of Stokes radius and the ability to inhibit rhodanese aggregation and to bind to the NH 2 -terminal domain of HSC70 (results not shown). Thus, N25 and C18 could be expressed in soluble forms and purified from E. coli as independent structural and functional units.

Secondary Structure Content of HIP Proteolysis Fragments
N25 and C18 proteolysis fragments could not only be produced in E. coli as active independent units, but they also show a high content in secondary structure comparable with that found in authentic proteins. As shown in Fig. 4A, the far-UV CD spectra of N25 and C18 fragments are typical of well structured polypeptides with a large amount of ␣-helices (maximum at 191-192 nm; minima at 207-208 and 222 nm). The deconvolution of these spectra fits the experimental data well. As shown in Table II, the proportions of ␣-helix and ␤-structures of N25 are very close to those of HIP. Moreover, the coincidence between the spectrum of the stoichiometric mixture of N25 and C18 and the sum of the individual spectra of isolated fragments (not shown) is in agreement with the absence of any detectable interaction between the fragments as observed by size exclusion chromatography (Fig. 3). By contrast, and as shown in Fig. 4B, there is a difference between the CD spectrum of HIP and that corresponding to the stoichiometric mixture of N25 and C18. This difference clearly indicates a structural change in the corresponding regions of HIP upon proteolysis, or when the fragments are produced as individual polypeptides. This is corroborated by the fact that the proportion of ␣-helices in C18 is significantly smaller than in HIP.
Since the N25 and C18 show the features of correctly folded structural domains in terms of chromatographic behavior, recombinant expression, and secondary structure content, a study of their structural and functional properties was undertaken.

FIG. 2. Mapping thrombin and chymotrypsin proteolysis sites, as determined by NH 2 -terminal sequencing and Western blot analysis of the fragments, on the primary and modular structure of HIP. A, thrombin cleaves recombinant HIP between Arg-226
and Ala-227. B, the major cleavage site of chymotrypsin is between Tyr-237 and Glu-238. Only primary cleavage sites are shown.

FIG. 3. Analysis of HIP fragments by gel filtration. Recombinant
HIP was digested by thrombin, and the reaction mixture was then injected onto a Superdex 200 HR10/30 column and eluted as described under "Experimental Procedures." The arrows indicate the elution volumes corresponding to the molecular mass (in kDa) standard proteins used for calibration. After elution, peak 1 and peak 2 were analyzed by SDS-PAGE (inset) and found to correspond to N25 and C18, respectively. Lane 1, peak 1; lane 2, peak 2. Molecular weight standards are indicated to the right of the gel. Since N25 contains the three tryptophan residues of the intact HIP, the absorbance of peak 1 is greater than that of peak 2.
FIG. 4. Analysis of secondary structure of HIP and its proteolysis fragments. The spectra were recorded under the conditions described under "Experimental Procedures." A, normalized spectra of isolated N25 (Ⅺ) and C18 (q). B, normalized spectra of HIP (q) and of the stoichiometric mixture of N25 and C18 fragments (Ⅺ). The solid lines correspond to the spectra reconstructed from the best fits with the Varselec analysis (30).

Conformational Properties of HIP Proteolysis Fragments
Size exclusion chromatography indicated that N25 and C18 elute as homogeneous proteins of 49 and 24 Å, respectively (Table III). Although for C18, such a radius was expected, that of N25 is too large for a monomeric protein of 25 kDa. Sedimentation velocity confirmed the monodisperse nature of N25 and C18 as indicated by the decrease of sedimentation coefficients with protein concentration, characteristic of nonassociative particles, and s 0 of 3.2 and 2.3 S were obtained for these two fragments, respectively (Fig. 5). Based on those values and the experimentally determined molecular mass, determined experimentally (see below), frictional ratios of 1.5 and 1.1 could be calculated for N25 and C18, respectively, emphasizing the asymmetric nature of N25 as compared with the C18, which appears to be rather globular (Table III). When N25 and C18 axial ratios, obtained for equivalent hydrated prolate ellipsoid particles, are compared with that of HIP, it appears that the N25 domain possesses, like the entire protein, an elongated shape, whereas the C18 is spherical (Table III).

Quaternary Structure of HIP Proteolysis Fragments
Sedimentation equilibrium data, performed at different concentrations and rotor speeds, could easily be fitted to a dimer model for N25 and monomer for C18 (Fig. 6, bottom) as indicated by the small variation and the random distribution of the residuals representing the variation between experimental and model data (Fig. 6, top). The values returned by the fitting procedure are reported in Table III. While C18 has an experimental molecular mass of 19,322 Da, very close to that of a theoretical monomer (17,966 Da), that of N25 is almost exactly twice that of a theoretical monomer, 51,163 Da compared with 25,582 Da.

Functional Properties of HIP Proteolysis Fragments
Inhibition of Rhodanese Aggregation-As shown in Fig. 7, rhodanese aggregates in a refolding buffer, after being denatured in a guanidinium buffer, as indicated by the increase in light diffusion. This aggregation is reduced by about 50% in the presence of either N25 or C18. Interestingly, this inhibitory effect was greatly increased when these two fragments were used together, becoming comparable with that obtained with the entire protein, suggesting an additive effect (Fig. 7). HIP as well as N25 or C18 activity is specific, since lysozyme, a control protein, had no effect on rhodanese aggregation. Moreover, the additive effect of N25 and C18 could not be obtained when either of these fragments was used with lysozyme, thus indicating that it is not merely due to an increase in protein concentration but truly reflects a cooperation between N25 and C18 to inhibit rhodanese aggregation.
Binding to the ATPase domain of HSC70 -HIP has been identified as an HSC70-interacting protein using the two-hybrid system with the NH 2 -terminal ATPase domain as a bait (18). These results have been reproduced in vitro in binding experiments involving immobilized proteins (20,21). When the same experiments are performed here, using immobilized HIP, N25, or C18, the results of Fig. 8 (left) are obtained. It can be seen that HIP as well as its respective N25 and C18 fragments were able to retain the NH 2 -terminal domain of HSC70. This domain can be eluted from the column in the presence of ATP, thus indicating that it could specifically be released from both the N25 and C18 fragments like the entire HIP. Control experiments using citrate synthase showed no detectable binding (Fig. 8, right). It is interesting to note that direct binding of HIP or its domains to the NH 2 -terminal domain of HSC70 has been obtained in the presence of ADP but in the absence of any other cochaperone, and specifically HSP40, suggesting a direct binding of HIP or its respective domains to HSC70.

DISCUSSION
Although a modular organization of HIP has been previously proposed, based on various sequence alignment criteria followed by mutagenesis analysis (18,20,21), no studies have  5. Analysis of N25 and C18 fragments by sedimentation velocity. Data were obtained at 4°C and at three loading concentrations, resulting in three data sets for each fragment. Since N25 contains the three tryptophan residues of intact HIP, absorbance was measured at 280 and 230 nm for N25 and C18, respectively. The sedimentation coefficients were corrected for the temperature, and the resulting coefficients at 20°C (Ⅺ) were plotted as a function of N25 (A) and C18 (B) loading concentrations and fitted to a linear function as described under "Experimental Procedures." a From Velten et al. (32). b Calculated from amino acid composition. c Determined by sedimentation equilibrium experiments. Numbers between parentheses represent the confidence interval given by the fitting procedure of the nine data sets obtained. d Corrected sedimentation coefficient at 20°C in water, determined by sedimentation velocity experiments. e Stokes radius determined by size exclusion chromatography. The column was calibrated with proteins of known Stokes radii.
f Frictionnal ratio. g Axial ratio for the equivalent particle, hydrated, prolate ellipsoid of revolution for a hydration coefficient of 0. 35. been performed directly on the protein. The work reported here takes advantage of limited proteolysis to probe the actual domain organization of the protein and obtain information on the structural and functional properties of the putative domains.
Limited proteolysis by thrombin and chymotrypsin shows that HIP is cleaved essentially in two fragments of about 25 and 18 kDa that appear to be resistant to further major cleavage. These fragments can be digested further by chymotrypsin, but only marginally and only on the COOH-terminal side, suggesting that they possess a rather compact core and, as expected, more accessible extremities. This was confirmed by the fact that the two proteases cleave only within a narrow region following the tetratricopeptide repeat (TPR), between residues 220 and 240 of HIP, despite the presence of several potential cleavage sites scattered throughout the sequence and despite their different specificity. Most importantly, protein phosphatase 5, another TPR-containing protein (37) on which the TPR region of HIP has been modeled (32), is cleaved primarily in the same region (i.e. beyond the TPR) by trypsin and subtilisin (38), two proteases that exhibit different specificities from those of thrombin and chymotrypsin. Thus, the major cleavage sites of these four proteases are restricted to a 13residue segment on the aligned sequences of HIP and protein phosphatase 5 (Fig. 9), although protein phosphatase 5 is a much larger and distinct protein, having only the TPR in common with HIP. Hence, these results strongly suggest that the N25 and C18 fragments represent authentic structural domains of HIP. Support for this interpretation comes from recombinant expression and biophysical and biochemical characterization of the respective fragments. Indeed, not only these fragments could be expressed in and purified from E. coli as independent folding units, they also exhibit well defined properties in terms of secondary structure content, hydrodynamic parameters, gross conformation, molecular mass, and biological activity. For instance, the sequence of HIP that contains the known remarkable features such as the oligomerization site and the TPR subdomain involved in HSC70 binding (20,21) corresponds to the NH 2 -terminal 30 kDa of the protein, whereas the sequence that shows no peculiar characteristics in terms of primary structure, except a GGMP repeat, maps to the COOHterminal 15 kDa. Interestingly, the same regions that are obtained after limited proteolysis can be expressed in E. coli and exhibit the expected structural and functional properties. The N25 appears to be an elongated dimer, like the entire protein, able to bind HSC70, and the C18 is a globular monomer that appears also to be able to bind HSC70. Thus, a two-domain structure with a linkage between residues 220 and 240 seems to be a plausible model for HIP.
This 220 -240 region, following the TPR in HIP and in which the protease cleavage sites are located, has been found to be an ␣-helix by several secondary structure prediction algorithms FIG. 6. Analysis of N25 and C18 fragments by equilibrium sedimentation. Data were obtained at three loading concentrations and three rotor speeds, resulting in nine data sets for each fragment. Since N25 contains the three tryptophan residues of intact HIP, absorbance was measured at 280 and 230 nm for N25 and C18, respectively. Here are shown the experimental data for N25 (Ⅺ) at 0.5 mg/ml and 12,000 rpm and C18 (E) at 0.15 mg/ml and 20,000 rpm. Data were fitted to a one-species model (solid line) as described under "Experimental Procedures." The residuals representing the variation between the experimental data and those generated by the fit are shown in the upper panels. and three-dimensional molecular modeling (32). In fact, in the TPR crystallographic structures available (39 -44), the region following the TPR is folded in an ␣-helix that packs in part against the TPR domain in a similar arrangement to that of the helices within the TPR domain, suggesting that this region is an integral part of the TPR domain (37,39,43). The remainder of this long helix, where the protease-sensitive sites are found, is weakly conserved in terms of sequence and more accessible to the solvent and thus could provide a link between the TPR and other domains in proteins. Furthermore, although regions connecting domains in proteins are generally believed to be solvent-accessible, flexible loops, helices are often found in crystallographic structures to be domain linkers. This is in fact the case of the HSP70 COOH-terminal domain, which is composed of a ␤-sandwich subdomain and a helical bundle subdomain, connected together by a long helix, in the middle of which protease-sensitive sites are found (45). A structural reorganization or "melting" in solution of such regions from a well defined secondary structure to a flexible loop has been proposed to rationalize this observation (45,46). Together, these data suggest that the ␣-helix following the TPR may constitute a linker between domains whether in HIP or other TPR-containing proteins.
In view of the fact that the N25 domain appears to be responsible for the dimerization and the elongated shape of the whole protein, since the C18 domain is a globular monomer, and that the first 14 residues are necessary for oligomerization (21), it is tempting to speculate on the association mode of the monomers in the dimer. In principle, the two monomers of HIP can associate in different fashions either through the NH 2terminal 14 residues of each monomer or via these residues and another region on the protein. However, it has been reported previously that fusion of the first NH 2 -terminal 124 residues of HIP to the monomeric maltose-binding protein leads to its oligomerization (20), indicating that these residues are sufficient for self-association and that HIP dimerizes through interactions between the NH 2 -terminal residues of the respective monomers. Furthermore, deletion of the 38 NH 2 -terminal amino acids of HIP (20) stabilizes a monomer that behaves as a globular protein on size exclusion chromatography, and preliminary characterization, in our laboratory, of the HIP variant lacking the first 14 residues confirms this observation and indicates that it is globular in shape (results not shown). Therefore, it is reasonable to suggest that the HIP monomer, made of two globular NH 2 -terminal 25-kDa and COOH-terminal 18-kDa domains, is roughly globular in shape but that the observed asymmetry of the whole molecule is induced by dimerization of the two NH 2 -terminal 25-kDa domains via their NH 2 -terminal regions.
As far as functional properties are concerned, it was interesting to find that the C18 domain of HIP is able to bind HSC70. This result indicates that although it does not have a TPR, the C18 participates in HSC70 binding, probably by stabilizing the primary interaction between HSC70 and the N25 domain of HIP, and suggests that, if the TPR is necessary for HSC70 binding, it may not be sufficient. This complementarity between the N25 and C18 domains of HIP in HSC70 binding is also seen in the unfolded rhodanese binding (to inhibit rhodanese aggregation), making it possible that unfolded protein binding and HSC70 binding are mediated by the same site on HIP involving an unidentified region on the C18 domain and the TPR region on the N25 domain. This is supported by the fact that TPRs are not only known to be involved in proteinprotein interactions but have also been shown to bind relatively hydrophobic peptides in an extended conformation (37,39).
On the basis of these hypotheses, a structural scheme for the cochaperone HIP, taking into account the hydrodynamic data and functional properties described in this paper, is presented in Fig. 10. HIP is presented as a dimeric protein in which the two monomers, each one composed of two relatively globular domains, interact through their NH 2 -terminal extremity, thus giving an elongated shape to the protein. This structure provides two binding sites for HSC70 located in the TPR of the NH 2 -terminal 25-kDa domain as suggested previously (32) and confirmed in this work. However, the TPR by itself does not seem to be sufficient for efficient binding, and participation of the C18 domain in the formation of the binding site is proposed, since this domain shows HSC70 and unfolded rhodanese binding activities. Therefore, the HSC70 binding site on HIP, that could overlap the unfolded protein binding site, is presented as shared between the two domains.
Acknowledgments-We are very grateful to David Smith for providing the plasmid expressing the HIP mutant (C15), to Michel Goldberg for discussions and support, to Jean-Pierre le Caer for NH 2 -terminal sequencing of HIP proteolysis fragments, to Noureddine Lazar and Mohamed Rholam for help with the circular dichroism experiments, and to Gérard Batelier for the analytical ultracentrifugation experiments. This model takes into account the hydrodynamic, limited proteolysis, and functional data described in this paper and by Velten et al. (32). HIP, represented as an elongated molecule of axial ratio a/b of 7.5, is composed of two relatively globular monomers interacting through their NH 2 termini. Each monomer is composed of two globular domains, a 25-kDa NH 2 -terminal domain (N25, large sphere) and an 18 kDa COOH-terminal domain (C18, small sphere), linked together via residues 220 -240 (not shown). The dimeric N25 is ellipsoidal like HIP, as indicated by its a/b ratio of 6, while the monomeric C18 is globular (a/b of ϳ1). This structure provides two potential binding sites for HSC70 (indicated by arrows) located at the interface between the TPR of the 25-kDa domain and an unidentified region of the 18-kDa domain. The HSC70 binding site on HIP may overlap the unfolded protein binding site as suggested by data of Fig. 7.