Analysis of the Interaction of Small Heat Shock Proteins with Unfolding Proteins*

The ubiquitous small heat shock proteins (sHsps) are efficient molecular chaperones that interact with nonnative proteins, prevent their aggregation, and support subsequent refolding. No obvious substrate specificity has been detected so far. A striking feature of sHsps is that they form large complexes with nonnative proteins. Here, we used several well established model chaperone substrates, including citrate synthase, α-glucosidase, rhodanese, and insulin, and analyzed their interaction with murine Hsp25 and yeast Hsp26 upon thermal unfolding. The two sHsps differ in their modes of activation. In contrast to Hsp25, Hsp26 undergoes a temperature-dependent dissociation that is required for efficient substrate binding. Our analysis shows that Hsp25 and Hsp26 reacted in a similar manner with the nonnative proteins. For all substrates investigated, complexes of defined size and shape were formed. Interestingly, several different nonnative proteins could be incorporated into defined sHsp-substrate complexes. The first substrate protein bound seems to determine the complex morphology. Thus, despite the differences in quaternary structure and mode of activation, the formation of large uniform sHsp-substrate complexes seems to be a general feature of sHsps, and this unique chaperone mechanism is conserved from yeast to mammals.

In response to environmental stress such as heat shock, which leads to the accumulation of nonnative proteins, cells increase the expression of several classes of proteins (1). The major conserved families of these heat shock proteins (Hsps) 1 have been shown to be involved in protein folding as molecular chaperones (2).
The most divergent of these chaperone classes are the small heat shock proteins (sHsps). sHsps have been found in almost all organisms investigated so far, with the number of members varying from species to species. They share conserved regions mostly in the C-terminal part of the protein, whereas the Nterminal part differs in sequence and length, leading to molec-ular masses of 16 -42 kDa for sHsps in different organisms (3). The conserved C-terminal domain of ϳ100 amino acids shares sequence homology with the major eye lens protein ␣A-crystallin (4). Almost all sHsps assemble into large oligomeric complexes of 9 to Ͼ30 subunits, and complexes in the range of 125 kDa to 2 MDa have been found (5)(6)(7)(8)(9)(10)(11). Some sHsps such as those from plants form assemblies with well defined stoichiometries, whereas other sHsps, including the mammalian proteins, form a range of oligomeric sizes (12,13). This polydispersity has limited the amount of structural information available. The crystal structure of an archaeal sHsp (14) and the cryo-electron microscopy reconstruction of ␣B-crystallin (15) revealed that the overall organization is that of a hollow globular sphere. A variation of this scheme is the three-dimensional structure of wheat Hsp16.9, which assembles into a dodecameric double disk, where each disk is organized as a trimer of dimers (16). Many sHsps have dynamic and variable quaternary structures with subunits that can freely and rapidly exchange between oligomers, such as ␣B-crystallin, Hsp25, Hsp27, and sHsps from Bradyrhizobium japonicum (8,13,15,17,18). Changes in quaternary structure at temperatures that are physiologically relevant for sHsps are important for their chaperone activity (16, 19 -22).
In vitro, sHsps act as molecular chaperones in preventing unfolded proteins from irreversible aggregation and insolubilization (19,(23)(24)(25)(26)(27)(28). Because of their high binding capacity of up to one substrate molecule/sHsp subunit, sHsps are more efficient than other chaperones in this respect (19, 25, 29 -32). The nonnative proteins seem to be surface-exposed in the complexes because bound proteins and peptides are accessible to antibodies (33). The range of substrates recognized covers peptides as well as oligomeric enzymes (25,33,34). No substrate specificity has been observed for sHsps so far, and complex formation with substrates has not been analyzed in detail.
Here, we set out to address this question using Hsp25 from mouse and Hsp26 from yeast, two different well studied members of the sHsp family.
Murine Hsp25 exists as a hexadecamer in solution. These oligomers are in a concentration-dependent equilibrium with tetramers, suggesting tetramers as the basic building block of the hexadecamer (17). It was previously shown that Hsp25 is able to influence the inactivation and subsequent aggregation of the model substrate citrate synthase (CS). Hsp25 binds several nonnative CS molecules and protects them from irreversible aggregation. Under permissive folding conditions, the bound substrates can be released from Hsp25 and, in cooperation with other ATP-dependent chaperones, regain their native structures (34).
Under physiological conditions, yeast Hsp26 forms spherical 24-mers with particle diameters of ϳ15 nm. Elevated temperatures induce the dissociation of the oligomeric complex into dimers. These dimers initiate the interaction with nonnative substrate proteins and reassemble into larger defined sHspsubstrate complexes (19).
To gain insight into the general mode of the interaction of sHsps with nonnative proteins, we used four different model substrate proteins. We show that, for each substrate, morphologically distinct and defined complexes are formed. The stoichiometry of the complexes between sHsps and nonnative protein seems to be dependent on the substrate investigated. Furthermore, we demonstrate that Hsp25 and Hsp26 are able to form mixed complexes with CS and ␣-glucosidase (␣-Gluc) and that the first substrate bound determines the morphology of the resulting sHsp-substrate complexes.

Analysis of Hsp25 and Hsp26 Activities
To induce aggregation, CS and ␣-Gluc were diluted in 40 mM Hepes-KOH, pH 7.5, equilibrated at 43 and 46°C, respectively (24,36,37). Rhodanese (30 M) was diluted 1:100 in 40 mM sodium phosphate, pH 7.7, at 44°C (38). Assays were performed in the absence and presence of Hsp25 or Hsp26. Assays in the presence of IgG served as a control for unspecific protein effects (concentrations given in the figure legends). To monitor the kinetics of thermal aggregation in the presence of Hsp25 and Hsp26, light scattering was measured in a FluoroMax I fluorescence spectrophotometer in a stirred and thermostated quartz cell. During the measurements, both the excitation and emission wavelengths were set to 360 nm with a spectral bandwidth of 4 nm.
The insulin aggregation assay in the presence of Hsp25 or Hsp26 was performed in 40 mM Hepes-KOH, pH 7.5. Insulin (45 M) was equilibrated in the absence and presence of either Hsp25 or Hsp26 at 25 or 43°C, respectively. Assays in the presence of IgG served as a control for unspecific protein effects. The aggregation reaction was started by the addition of dithioerythritol to a final concentration of 20 mM, and the resulting turbidity was monitored at 360 nm in an Amersham Biosciences Ultrospex 4060 UV-visible spectrophotometer equipped with a temperature control unit.

Analysis of sHsp-Substrate Complexes
Size-exclusion Chromatography (SEC)-To analyze complex formation between Hsp25 and the different substrates, SEC was performed using a TosoHaas TSK4000SW column (30 ϫ 0.75 cm, separation range of 20 -7000 kDa). Complex formation between Hsp26 and the different substrates was analyzed on a TosoHaas TSK4000PW column (30 ϫ 0.75 cm; separation range of 10 -1500 kDa). Hsp25 or Hsp26 in the absence and presence of substrate was incubated as described in the figure legends. All samples were centrifuged at 14,000 ϫ g for 5 min at 4°C before application. Chromatography was carried out at 25°C in running buffer (0.1 M Hepes-KOH and 150 mM KCl, pH 7.5) at flow rates of 0.5 ml/min for samples with Hsp25 and 0.75 ml/min for those with Hsp26. The sample volume was 100 l. Hsp25 and Hsp26 were detected by fluorescence at an excitation wavelength of 280 nm and emission wavelengths of 330 and 323 nm, respectively, using a Jasco FP920 fluorescence detector.
Analysis of Binding Stoichiometries-Hsp25-substrate complexes were formed with 4.8 M Hsp25 in the presence of increasing concentrations of CS (0.15-5 M) for 30 min at 43°C. During SEC (described above), the respective complex peaks were collected and precipitated with sodium deoxycholate/trichloroacetic acid. Precipitates were then resuspended in reducing SDS sample buffer (39), separated by SDS-PAGE (10 -20% precast Tricine gradient gels, Novex), and visualized by Coomassie Blue staining. Scanned gels were analyzed using Image-Master-1D software (Amersham Biosciences). Protein amounts and ratios were calculated for monomers and corrected for different staining intensities due to differences in molecular mass.
Electron Microscopy-Complex formation between Hsp25 or Hsp26 and the different substrates was performed following the protocol used for the SEC experiments. The samples were separated on a gel filtration column. The complex peak was collected; and immediately before application, samples were diluted to a protein concentration of 0.1 mg/ml in running buffer. As controls, substrates and Hsp25 or Hsp26 alone were also incubated under the respective experimental conditions. The samples were directly applied to glow-discharged carbon-coated copper grids and negatively stained with 3% uranyl acetate. Electron micrographs were recorded at a nominal magnification of ϫ60,000 using a Philips CM12 electron microscope operating at 120 kV or at a nominal magnification of ϫ33,000 using a Jeol 100CX electron microscope operating at 100 kV.

RESULTS
sHsps Suppress the Aggregation of Different Model Proteins-The most commonly used experiment to investigate chaperone activity is to assay for the ability to suppress aggregation of nonnative proteins (2,41,42). It has been shown that murine Hsp25 and yeast Hsp26 are able to interact with aggregation-sensitive proteins (19,24,33,34). Here, we were interested in whether their modes of interaction differ for various model substrates. The first substrate investigated, dimeric CS (49 kDa/subunit), aggregates rapidly when incubated at heat shock temperatures (Ͼ40°C) (24). This process can be monitored by light scattering (Fig. 1, A and E). The addition of the control protein IgG, even at high excess, did not alter the kinetics of CS aggregation significantly. However, in the presence of increasing amounts of either Hsp25 or Hsp26, the thermal aggregation of CS could be effectively reduced (Fig. 1, A and E). This is in agreement with previous results (19,24,34). Further addition of the respective sHsp (data not shown) did not prevent the small amount of residual aggregation observed, indicating that some CS folding intermediates could not be stably bound to Hsp25 or Hsp26.
Monomeric bovine rhodanese (33 kDa) is widely used as a model for protein folding (38). In the absence and presence of IgG, the enzyme aggregated completely within 25 min when incubated at 44°C (Fig. 1, B and F). Substoichiometric amounts of Hsp25 were already able to suppress rhodanese aggregation significantly (Fig. 1B), whereas a 10-fold excess of Hsp26 was required to obtain a similar effect (Fig. 1F). Similar to the situation with CS, aggregation could not be completely suppressed even at high Hsp25 and Hsp26 concentrations.
Another model substrate that shows the tendency to precipitate at elevated temperatures is the cytoplasmic monomeric ␣-glucosidase P1 from Saccharomyces cerevisiae (68.5 kDa) (36,44). In the absence and presence of IgG, the enzyme aggregated completely within 30 min when incubated at 46°C (Fig. 1, C and G). The addition of Hsp25 resulted in a marked reduction of the light scattering signal. At an enzyme/Hsp25 monomer ratio of ϳ1:3, the thermal aggregation of ␣-Gluc was suppressed to a large extent. In contrast, Hsp26 was less effective in suppressing the thermal aggregation of yeast ␣-Gluc. An excess of Hsp26 was required to reduce the light scattering signal significantly (Fig. 1G). Aggregation could not be completely suppressed, even at high sHsp concentrations. Residual light scattering might indicate that some unfolding intermediates cannot be stably bound or that the large sHsp-substrate complexes scatter light.
To ensure that a broad molecular mass range of substrates was covered, the effect of sHsps on the reduction-induced aggregation of insulin was analyzed. Upon reduction of the interchain disulfide bond, the B-chain of insulin precipitates, whereas the A-chain stays in solution (45). With a molecular mass of 3.3 kDa, the insulin B-chain is significantly smaller than the proteins studied above. After the addition of dithioerythritol, the peptide aggregated completely within 60 min (Fig. 1, D and H). At substoichiometric concentrations of either Hsp25 or Hsp26, however, a concentration-dependent decrease in aggregation at 25 or 43°C, respectively, could be observed (Fig. 1, D and H). The addition of sHsps resulted in a marked reduction of the light scattering signal. At an insulin/Hsp25 monomer ratio of ϳ6:1, the aggregation of insulin was completely suppressed at 25°C. An insulin/Hsp26 monomer ratio of 4:1 and higher temperatures (43°C) were required for complete suppression of insulin aggregation by Hsp26. At 25°C, aggregation could not be completely suppressed, even at high Hsp26 concentrations (data not shown; see Ref. 19).
The insulin assay also allowed us to address whether the chaperone function of Hsp25 is temperature-dependent. For yeast Hsp26, it has been shown previously that the chaperone activity increases with temperature (19). When the insulin assay was performed at 25, 37, or 42°C, the protective activity of Hsp25 did not change (data not shown), whereas for Hsp26, complex dissociation is required for full chaperone activity (19). Thus, the chaperone function of Hsp25 does not seem to be temperature-dependent.
Analysis of sHsp-Substrate Complexes by Electron Microscopy-After having established that mouse Hsp25 and yeast Hsp26 interact with a variety of substrate proteins, we investigated the structures and sizes of sHsp-substrate complexes by electron microscopy. As a control for effects of high temperatures on Hsp25 and Hsp26, the proteins were incubated at 43°C for 15 min. In comparison with a sample applied to the grid after incubation at 25°C, no significant changes in the appearance of the globular Hsp25 particles were observed. In both cases, globular Hsp25 particles with a diameter of ϳ20 nm were detected (data not shown; see also Ref. 34), whereas for Hsp26, significant changes in quaternary structure were detectable after incubation at 43°C. The well defined oligomeric complex dissociated into stable dimers (data not shown; see also Ref. 19). This dissociation seems to expose binding sites for nonnative proteins.
sHsp-substrate complexes (Fig. 2) were formed under experimental conditions comparable to the aggregation experiments ( Fig. 1). When the substrate proteins were incubated at elevated temperatures in the absence of sHsps, in all cases, large amorphous aggregates were detected (Fig. 2, A, D, G, and J).
In the presence of Hsp25 and Hsp26, CS, ␣-Gluc, rhodanese, and insulin formed defined complexes with sHsps (Fig. 2). Although for a particular substrate, except insulin, regular uniform complexes were observed, comparison of the different sHsp-substrate complexes revealed significant differences in size and shape. With CS (Fig. 2B) and rhodanese (Fig. 2E), Hsp25 formed globular or oval structures with mean diameters of up to ϳ70 and ϳ45 nm, respectively. Relatively uniform, globular shaped complexes were detected in the case of Hsp26 and CS with mean diameters of ϳ50 nm (Fig. 2C). In the case of rhodanese and Hsp26, more irregular, smaller, but still globular particles with mean diameters of ϳ45 nm were formed (Fig. 2F).
Globular, regularly shaped particles with a size of ϳ30 nm were found in the case of ␣-Gluc and Hsp25 (Fig. 2H), whereas in the case of ␣-Gluc and Hsp26, the complexes were more irregular, with mean diameters of ϳ30 nm (Fig. 2I). In contrast, binding of insulin to Hsp25 or Hsp26 resulted in the formation of fibrous, network-like structures with diameters of ϳ15-20 nm, corresponding to the size of Hsp26 and Hsp25 oligomers (Fig. 2, K and L). Analysis of sHsp-Substrate Complexes by SEC-After having established that different sHsp-substrate complexes show distinct morphologies, we used gel filtration experiments to gain further insight into complex formation. As already described (34), in the absence of substrate, Hsp25 eluted as a dominant peak of ϳ400 kDa, consistent with a hexadecameric complex. Tetramers represented a minor species. This elution behavior did not change after incubation at elevated temperatures or under reducing conditions (data not shown). At 25°C, Hsp26 eluted as a single distinct peak with an apparent molecular mass of 550 kDa. After incubation under reducing conditions, the elution behavior did not change (data not shown), whereas after incubation at elevated temperatures, the 24-mer complex dissociates reversibly into dimers (19). In agreement with the observation that all of the substrates assayed formed large aggregates after incubation at elevated temperatures (Fig. 2), hardly any soluble protein could be detected in the absence of sHsps (data not shown).
SEC analysis of samples in which Hsp25 had been incubated together with CS or rhodanese at elevated temperatures resulted in the appearance of an additional peak in the void volume of the column consistent with a molecular mass of several megadaltons (Fig. 3, A and B). No intermediate-sized Hsp25-substrate complexes were detected even at low sub-strate concentrations. At all ratios, Hsp25 was also present in its uncomplexed form, indicating that either not all Hsp25 molecules were involved in complex formation or that smaller substrate complexes dissociated during chromatography.
After the incubation of Hsp26 with CS (Fig. 3E) or rhodanese (Fig. 3F) at elevated temperatures, the samples were applied to a gel filtration column running at 25°C. An additional peak appeared that eluted in the void volume of the column (Fig. 3,  E and F). This indicates that binding of denatured CS or rhodanese to Hsp26 at heat shock temperatures resulted in the formation of particles of Ͼ1500 kDa. Similar to Hsp25, fullsized Hsp26-substrate complexes were detected immediately, with no intermediate-sized Hsp26-substrate complexes even at low substrate concentrations.
A different picture emerged for both ␣-Gluc and insulin (Fig.  3, C and D). When these substrates were applied to the column after incubation in the presence of Hsp25, prominent Hsp25substrate peaks appeared, whereas free Hsp25 disappeared. With increasing substrate concentrations, the complexes increased in size until nearly all of the protein with ␣-Gluc and part of the sample with insulin were drawn into the complex peaks that eluted in the void volume of the column.
When Hsp26 was incubated with substoichiometric amounts of ␣-Gluc and applied to the column, only part of the protein was drawn into the complexes that eluted in the void volume of the column (data not shown). Most of the protein was found in small complexes that eluted as peaks in the range of 550 -1500 kDa. The incubation of equimolar amounts of ␣-Gluc and Hsp26 (Fig. 3G) resulted in the formation of a distinct complex peak, but still free Hsp26 and unbound ␣-Gluc could be detected. With increasing substrate concentrations, the complexes increased in size until, at an Hsp26/␣-Gluc ratio of 1:2, all of the protein was drawn into substrate complexes eluting in the void volume (Fig. 3G).
In the case of insulin and Hsp26 (Fig. 3H), complexes larger than Hsp26 alone were detected in the range of 550 kDa to 1 MDa. Unbound protein was still visible and also additional peaks, which may have resulted from interactions of the Hsp26-insulin complexes with the column.
Stoichiometry of Hsp25-CS Complexes-Haslbeck et al. (19) described the formation of the large Hsp26-CS complexes as a highly cooperative process with a maximum binding capacity at a molar ratio of one CS monomer to Hsp26 dimer. To determine the stoichiometry in Hsp25-CS complexes, we incubated Hsp25 at elevated temperatures in the presence of increasing amounts of CS. The samples were separated by SEC, and the complex peaks running in the void volume of the column were collected. Their composition was then analyzed by SDS-PAGE and densitometry (Fig. 4). When the CS concentration was gradually increased, full-sized complexes were present from the start. No intermediate-sized complexes could be detected. These observations were confirmed by electron microscopy.
A maximum of one substrate monomer/Hsp25 monomer was incorporated into the complex (Fig. 4) and one CS monomer/ Hsp26 dimer (19). Densitometric analysis showed that CS was incorporated in a cooperative manner into the substrate complexes.
Mixed Complexes with ␣-Gluc and CS-Because complexes of Hsp25 or Hsp26 with CS or ␣-Gluc showed distinct morphologies, we next investigated whether mixed complexes between sHsps and different substrate proteins can be formed. The two substrates were incubated either sequentially or together with Hsp25 or Hsp26 at elevated temperatures. The resulting complexes were analyzed by SEC/high pressure liquid chromatography and electron microscopy. Representative elution profiles and electron micrographs are shown in Fig. 5. The profiles from the SEC analysis were comparable for the different incubation sequences (Fig. 5, A and B). Whereas in the case of Hsp25, all of the substrate protein was drawn into the Hsp25-substrate complex (Fig. 5A), in the case of Hsp26, free CS, ␣-Gluc, and Hsp26 were still detectable (Fig. 5B). Both substrate proteins were found in complex with the sHsps as analyzed by SDS-PAGE. The analysis of these complexes by electron microscopy (Fig. 5, C-F) revealed that the complexes obtained under one set of conditions were uniform in size and shape. However, if the sHsp was first incubated with CS, the complexes showed marked similarity to the structures that were observed when only CS was incubated with Hsp25 or Hsp26 (Fig. 2, B and C; and Fig. 5, C and D). Preincubation of the sHsp with ␣-Gluc led to typical complexes between the sHsp and ␣-Gluc (Fig. 2, H and I; and Fig. 5, E and F). Interestingly, the first substrate Hsp25 or Hsp26 was preincubated with seems to determine the morphology of the sHspsubstrate complex.
When all three proteins (either Hsp25 or Hsp26 with CS and ␣-Gluc) were incubated together (Fig. 5, G and H), uniform particles were again formed. Most of the particles showed shapes similar to the sHsp-CS complexes to which ␣-Gluc was added (Fig. 5, compare C and D). However, they were significantly smaller, with diameters of ϳ35-40 nm, i.e. between those observed for the two mixed complexes shown in Fig. 5 (C/E and D/F). These experiments show that two different substrates were able to bind simultaneously to sHsps. Binding resulted again in the formation of large distinct complexes. The morphology seems to be determined by the first substrate bound.

DISCUSSION
The overexpression of sHsps has been shown to convey thermotolerance in a number of organisms and cell types (20, 40, 46 -51), indicating a general thermoprotective function of the protein family. The mechanism of sHsp chaperone activity is poorly characterized. Previous work has clearly demonstrated that sHsps recognize misfolded proteins and maintain them in a soluble but inactive state. In contrast to other chaperones, the chaperone function of sHsps appears to be limited to binding and maintaining the solubility of unfolding proteins, without promoting refolding directly (8,19,23,24,29,34).
Haslbeck et al. (19) suggested that, for yeast Hsp26, dissociation is a prerequisite of chaperone function and that the Hsp26 dimer may act as the primary substrate-binding species, followed by reassembly into a larger complex together with the substrate. This hypothesis was further supported by in vitro Hsp25 was incubated with increasing amounts of CS for 30 min at 43°C and applied to a TSK4000SW column. The resulting complex peaks were collected, precipitated, and analyzed by SDS-PAGE. The bands were densitometrically analyzed to determine molar ratios of substrate and Hsp25 in the complex. Ratios refer to monomeric molar concentrations. Band intensities were corrected for differences in staining intensities due to different molecular masses. Inset, representative sections of Coomassie Blue-stained SDS-polyacrylamide gels used in the analysis. and in vivo studies (16,20,26) demonstrating that dynamic changes in oligomerization are a prerequisite for sHsp function in vivo and chaperone activity in vitro. For Hsp25, detectable dissociation of the oligomer does not precede the binding of nonnative protein, but an exchange of subunits has been suggested (34). We used these two members of the sHsp family, mouse Hsp25 and yeast Hsp26, which differ in their quaternary structures and in the mode of activation, to study the general principles of complex formation of sHsps with nonnative proteins.
Formation of Large sHsp-Substrate Complexes-Little is known about the structure of sHsp-substrate complexes. A striking feature of sHsps is that, upon substrate binding, they form very large complexes. For the four different model substrates used in this study, effective interactions with both Hsp25 and Hsp26 were observed. Both sHsps showed comparable promiscuity in substrate binding, and the morphologies of the sHsp-substrate complexes as visualized by negative-stain electron microscopy were similar. Hsp26 and Hsp25 exhibited globular structures with outer diameters of ϳ15 and 20 nm, respectively. These structures increased in size upon substrate binding. Complexes formed with CS showed mean diameters of ϳ50 and ϳ70 nm for Hsp26 and Hsp25, respectively. Complexes formed with rhodanese were smaller, with mean diameters of ϳ45 nm for both sHsps. The ␣-Gluc complexes were smaller (mean diameters of ϳ30 nm) and more irregularly shaped than the CS and rhodanese complexes. The complexes formed with insulin revealed net-like structures with mean diameters of about the same size of the respective sHsp. The differences in complex sizes between Hsp26 and Hsp25 with CS might be due to different binding stoichiometries (see below).
SEC experiments indicated that the process of complex formation of Hsp25 and Hsp26 with nonnative protein is influenced by the substrate used. After incubation of the sHsp with CS or rhodanese, full-sized sHsp-substrate complexes could be detected immediately. Only the amount of particles (and not their size) increased with prolonged incubation. With ␣-Gluc and insulin, intermediate-sized particles could be detected after incubation of substoichiometric amounts of substrate with sHsps.
Surface binding of substrate has been shown for plant Hsp18.1 (29), and the binding of peptides and proteins to Hsp25 (33) also argues for accessibility of bound protein. In the case of Hsp26, the Hsp26-CS complex appears, however, to be a completely new assembly. The original Hsp26 shell is no longer detectable. Instead, an enlarged outer shell and an additional, internal shell of density become visible (19). As the bound substrates are still accessible to proteases, they are at least partially surface-exposed (data not shown). Due to the similarities in complex formation and morphology, it is reasonable to assume that, also in the case of Hsp25 and for sHsps in general, large rearrangements of the complex occur upon binding of nonnative protein.
Taken together, our data suggest that the formation of large defined sHsp-substrate complexes is a general feature of sHsps. The morphology of these complexes is substrate-dependent, but independent of the sHsp used.
Cooperativity of Substrate Protein Binding-The large comexperiment, both substrate proteins were incubated together with either Hsp25 or Hsp26 for 30 min at 44°C. C-H, electron microscopy of mixed complexes. In parallel, the complex peaks were directly applied to copper grids and visualized by electron microscopy (see above). Hsp25 (C) or Hsp26 (D) and CS were incubated first before ␣-Gluc was added. Hsp25 (E) or Hsp26 (F) and ␣-Gluc were incubated first before CS was added. Hsp25 (G) or Hsp26 (H), CS, and ␣-Gluc were incubated together for 30 min at 44°C. Scale bars ϭ 100 nm. rel., relative.
FIG . 5. A and B, formation of mixed complexes between CS, ␣-Gluc, and sHsps. A, 4.8 M Hsp25 was incubated in the presence of 3 M ␣-Gluc for 15 min at 44°C; 4 M CS was added; and the sample was further incubated for 15 min. The resulting complex peak was analyzed by SEC (dashed line). In a second experiment, the order of substrate addition was reversed: Hsp25 and CS were incubated first (solid line). The respective complex peaks were collected, precipitated, and separated by SDS-PAGE, and the resulting bands were analyzed as described above. Insets, representative sections of Coomassie Bluestained SDS-polyacrylamide gels used in the analysis. B, 3 M Hsp26 was incubated in the presence of 1.5 M ␣-Gluc for 15 min at 44°C; 1.5 M CS was added; and the sample was further incubated for 15 min. The resulting complex peak was analyzed by SEC (dashed line). In a second experiment, the order of substrate addition was reversed: Hsp26 and CS were incubated first (solid line). The respective complex peaks were collected, precipitated, and separated by SDS-PAGE. In a third plex structures observed by SEC and electron microscopy indicate binding of several substrate molecules to Hsp25 or Hsp26 oligomers. A maximum binding capacity of one protein/sHsp subunit has repeatedly been suggested (29,31,43). Especially in the case of CS and rhodanese, the particles formed with Hsp25 or Hsp26 showed little variance in size and shape. Only spherical, regularly shaped particles appeared. Partially loaded and therefore smaller sHsp complexes were not detected. This observation suggests coordinated binding of substrate molecules to sHsps. For CS and Hsp26 (19) and for CS and Hsp25 (Fig. 4), this process was found to be highly cooperative.
The simultaneous binding of several nonnative proteins seems to be a prerequisite for efficient and stable complex formation. Interestingly, for all proteins studied, except insulin, binding could be saturated at a defined ratio of nonnative protein to sHsp. The cooperativity and morphology of the complexes indicate that the substrates direct the process of complex formation.
Incorporation of Different Substrate Proteins into sHsp Complexes-The formation of mixed substrate complexes is important for the function of sHsps in the crowded environment of the cell. Here, upon stress, different polypeptides unfold and have to be bound simultaneously. We were interested in whether sHsps are able to bind different substrates in one complex. Interestingly, two different substrates could be incorporated into the sHsp-substrate complex, leading to the formation of morphologically uniform mixed complexes. The analysis of the mixed complexes by electron microscopy revealed significant differences in complex morphologies depending on the order of addition. If either Hsp25 or Hsp26 was first incubated with CS, the complexes showed marked similarity to the structures that were observed upon incubation of only CS and Hsp25 or Hsp26. Preincubation of sHsp with ␣-Gluc led to typical complexes between sHsp and ␣-Gluc. When all three proteins were incubated together, the resulting complexes showed morphologies similar to the sHsp-CS complexes, but they were significantly smaller. Thus, the first substrate bound seems to determine the morphology of the complex.
In conclusion, the analysis of the interaction of different substrate proteins with sHsps provides new insights into the mode of complex formation between sHsps and nonnative substrate proteins. sHsps bind substrates in a range of at least 3-100 kDa. The degree of cooperativity seems to be substratedependent. Taken together, our results show that the formation of large, morphologically distinct complexes with nonnative proteins is a conserved feature of the sHsp family of chaperones.