The Dynamics of Hsp25 Quaternary Structure

Small heat shock proteins (sHsps), including α-crystallin, represent a conserved and ubiquitous family of proteins. They form large oligomers, ranging in size from 140 to more than 800 kDa, which seem to be important for the interaction with non-native proteins as molecular chaperones. Here we analyzed the stability and oligomeric structure of murine Hsp25 and its correlation with function. Upon unfolding, the tertiary and quaternary structure of Hsp25 is rapidly lost, whereas the secondary structure remains remarkably stable. Unfolding is completely reversible, leading to native hexadecameric structures. These oligomers are in a concentration-dependent equilibrium with tetramers and dimers, indicating that tetramers assembled from dimers represent the basic building blocks of Hsp25 oligomers. At high temperatures, the Hsp25 complexes increase in molecular mass, consistent with the appearance of “heat shock granules” in vivo after heat treatment. This high molecular mass “heat shock form” of Hsp25 is in a slow equilibrium with hexadecameric Hsp25. Thus, it does not represent an off-pathway reaction. Interestingly, the heat shock form exhibits unchanged chaperone activity even after incubation at 80 °C. We conclude that Hsp25 is a dynamic tetramer of tetramers with a unique ability to refold and reassemble into its active quaternary structure after denaturation. So-called heat shock granules, which have been reported to appear in response to stress, seem to represent a novel functional species of Hsp25.

Small heat shock proteins (sHsps), 1 exhibiting a monomeric molecular mass of 9 -42 kDa, are expressed in all organisms investigated so far. Because of functional as well as structural homologies, ␣-crystallin, a major mammalian eye lens protein, which is also expressed in non-lenticular tissue, is a member of this protein family (1)(2)(3)(4). Although the overall homology between different sHsps is rather low, they are grouped together based on conserved sequences in the C-terminal half of the protein and short, conserved, phenylalanine-rich stretches near the N terminus of the protein (5,6). Mammalian sHsps are expressed constitutively even under physiological conditions. However, stress factors such as heat shock induce a strong up-regulation of protein levels by 10 -20-fold to maximum concentrations of 0.1% of the cellular protein (7,8). Overexpression of different mammalian sHsps increases cellular thermoresistance significantly (9,10). Furthermore, sHsps have been suggested to function in different, seemingly unrelated processes like RNA stabilization (11), interaction with the cytoskeleton (12,13), or apoptosis (14). Interestingly, sHsps are also overexpressed in several cancers and neurodegenerative diseases like Alzheimer's disease or multiple sclerosis (15)(16)(17). In plants, five different classes of sHsps have been identified, which are partly localized in organella (8,18).
In vitro sHsps act as molecular chaperones in preventing unfolded proteins from irreversible aggregation (3,4,19) and, in cooperation with other factors, e.g. Hsp70 and ATP, facilitate productive refolding of unfolded proteins (20,21). In this context, sHsps are more efficient than the model chaperone GroEL, due to their high binding capacity of up to one substrate molecule per sHsp subunit (21)(22)(23)(24).
Although sHsps are rather heterogeneous both in monomeric molecular weight and amino acid sequence, they all share the striking feature of forming high molecular weight oligomeric complexes of variable size. Although the 16-kDa antigen of Mycobacterium tuberculosis forms a nonamer assembled as a trimer of trimers (25), the well characterized ␣-crystallin has been reported to adopt different oligomeric structures with molecular masses ranging from 125 kDa to 2 MDa (26,27). In electron micrographs, ␣-crystallin and sHsp complexes show typically globular or spherical appearance of 10 -25 nm (20,21,28,29). By cryo-electron microscopy and image analysis, Haley et al. (30) have shown that human, recombinant ␣B-crystallin forms a hollow, globular shell with a molecular mass of 650 kDa (Ϯ 150 kDa) and asymmetric appearance. Several species were observed differing in size and shape. This is interesting in the light of data that demonstrate subunit exchange between ␣-crystallin complexes as well as between Hsp25/27 and ␣-crystallin in vivo and in vitro (19,31,32). It remains to be seen whether this is an indicator of similar oligomeric association of ␣-crystallin and the mammalian Hsp25/27. These hetero-oligomeric complexes dissociate during heat treatment (33), whereas ␣-crystallin and homo-oligomeric sHsp complexes were repeatedly reported to increase significantly in size upon heat shock as so-called "heat shock granules" and to redistribute from the cytosol to the perinuclear space or the nucleus itself (26,28,34). Furthermore, in heat-shocked tomato cell cultures, sHsps were found as large heat shock granules in the nuclear periphery (11). The chloroplast-localized sHsp, Hsp21, stays soluble during an initial heat shock and insolubilizes during additional or prolonged heat treatment (35). This increase in complex size seems to be modulated by the metabolic state (36) and the degree of thermoresistance of the cell (26), e.g. during heat shock, Dm-Hsp27 forms insoluble superoligomers (Ͼ1 MDa), which are redistributed to the nucleus (26), whereas pretreated, thermotolerant cells do not show these effects (37). In contrast, chemical stress such as arsenite treatment or stress caused by serum starvation leads to a decreased level of Dm-Hsp27 oligomerization, shifting the molecular mass to about 200 kDa (34,38).
In all cases described so far, changes in sHsp oligomerization upon stress were reversible, with the complexes relocalizing to their physiological compartment during recovery.
As the quaternary structure of sHsps is essential for their function and regulation of activity, but its basic properties are still rather poorly understood, we investigated changes in the oligomeric structure and stability of murine Hsp25 in detail.

Materials
Recombinant murine Hsp25 was expressed and purified as described previously (4,39). Mitochondrial citrate synthase (CS) from pig heart (EC 4.1.3.7) was obtained from Roche Molecular Biochemicals. The concentrations of CS and Hsp25 were determined using the extinction coefficients of 1.78 and 1.87, respectively, for a 1 mg/ml solution at 280 nm. CS was stored in 50 mM Tris, 2 mM EDTA, pH 8.0; Hsp25 in 40 mM Hepes-KOH, pH 7.5. The concentrations for CS and Hsp25 given in the text refer to a dimer and a hexadecamer, respectively.

Analysis of Hsp25 Activity: Aggregation Assay
15 M CS was diluted 1:200 in 40 mM Hepes-KOH, pH 7.5, equilibrated at 43°C, in the presence and absence of Hsp25 (concentrations given in the figure legends). To monitor the kinetics of thermal aggregation, light scattering was measured in a Perkin-Elmer MPF44A fluorescence spectrophotometer in stirred and thermostated quartz cells. During the measurements, both the excitation and emission wavelength were set to 500 nm with a spectral bandwidth of 2 nm.
Analysis of Hsp25 Particles Electron Microscopy-To determine changes in oligomeric size upon heating, Hsp25 (0.2 M in 40 mM Hepes pH 7.5) was incubated either at 20°C for 60 min or at 80°C for 15 min, and then Hsp25 was cooled to 20°C for another 60 min. Aliquots were applied to glow-discharged carbon-coated copper grids and negatively stained with 3% uranyl acetate. Electron micrographs were recorded at a nominal magnification of 60,000, using a Philips CM12 electron microscope operated at 120 kV.
Native Gel Elecrophoresis-10 g of Hsp25 at concentrations of 2 and 0.25 mg/ml in the suggested native sample buffer were applied to a precast 4 -12% Tris-glycine gel (Novex, San Diego). Buffers and conditions were chosen according to the manufacturer's protocol. As controls, 10 g of ferritin and GroEL were applied according to the same protocol.
Size Exclusion Chromatography-Size exclusion HPLC (SEC) was performed using a TosoHaas TSK 4000 SW column (30 cm x 0, 75 cm; separation range 10 -7,000 kDa). If not otherwise specified, chromatography was carried out in 100 mM Hepes-KOH, pH 7.5, with a flow rate of 0.75 ml/min (Figs. 1, 2, and 4) or 0.5 ml/min (Fig. 6). The sample volume was 100 l. Hsp25 was detected by fluorescence at an excitation wavelength of 280 nm and an emission wavelength of 335 nm, using a Jasco FP 920 fluorescence detector.
For cross-linked samples Hsp25, at a concentration of 0.3 mg/ml was incubated with 10 mM glutaraldehyde for 2 min at 37°C in 40 mM Hepes, pH 7.5. The reaction was stopped with 35 mM Tris/HCl, pH 8.0. The non-cross-linked Hsp25 control was treated accordingly, with the exception that buffer was added instead of glutaraldehyde. Samples were centrifuged for 5 min at 14,000 ϫ g at 20°C. Estimation of molecular masses was achieved by comparison with established marker proteins.
Analytical Ultracentrifugation-Sedimentation velocity analysis was performed in an analytical ultracentrifuge (Beckman Optima XL-A). Double sector cells were used at 20,000 rpm in a rotor AnTi 60 at 20°C. The protein concentration was 0.28 mg/ml. The data were analyzed using the sedimentation time derivative method (40).

Spectroscopy
Temperature-induced Structural Changes in Hsp25-To monitor thermal unfolding, the intrinsic fluorescence and light scattering of the protein solution were measured from 25 to 80°C in stirred quartz cells. The measurements were carried out in a Perkin-Elmer MPF44A fluorescence spectrometer with a thermostated cell holder connected to a thermoprogrammer. Changes in protein structure upon heating, resulting in exposure of tryptophans and tyrosines to the solvent, led to changes in protein fluorescence. The tryptophan fluorescence of Hsp25 was recorded at an excitation wavelength of 295 nm and an emission wavelength of 338 nm. The spectral bandwidth was 5 nm for both excitation and emission. The heating rate was 1°C/min. The Hsp25 concentration was 50 g/ml in 40 mM Hepes, pH 7.5.
Light scattering measurements were performed to determine changes in the size of Hsp25 particles at different temperatures or increasing concentrations of the denaturant urea. Excitation and emission wavelength were set to 360 nm. The spectral bandwidths were 5 and 2.5 nm, respectively. For thermal unfolding and refolding of Hsp25, the heating or cooling rate was 0.5°C/min. The Hsp25 concentration was 50 g/ml in 40 mM Hepes, pH 7.5. To determine the light scattering signal of Hsp25 in increasing concentrations of urea, the protein (50 g/ml) was incubated at the respective urea concentrations overnight at 20°C. Measurements were then performed as described above.
The thermal unfolding transition of Hsp25 in the far UV CD range was recorded in a Jasco J715 CD spectropolarimeter with a PTC 343 Peltier heating unit. The protein concentration was 150 g/ml in 40 mM Hepes, pH 7.5. The cuvette pathlength was 1 mm. The heating rate was 1°C/min, and the CD signal was measured at 225 nm.
Urea-induced Unfolding and Refolding of Hsp25-Urea-induced unfolding of Hsp25 was performed by diluting the protein into increasing concentrations of urea (in 40 mM Hepes, pH 7.5) ranging from 0 to 7 M. The samples were incubated for 20 h at 20°C to achieve equilibrium.
For refolding experiments, Hsp25 was denatured at a concentration of 2 mg/ml by incubation in 8 M urea for 10 h at 20°C. Refolding was started by diluting the protein in decreasing urea concentrations to a final protein concentration of 10 g/ml before incubation at 20°C for 20 h. To study the influence of salt on the stability of Hsp25, ureainduced unfolding and refolding was performed as described in the presence and absence of 1.15 M NaCl.
The unfolding and refolding transitions were monitored by measuring the change in intrinsic fluorescence. Unfolding was furthermore monitored by far UV CD spectroscopy.
The fluorescence measurements were carried out in a Spex Fluoromax fluorospectrometer at 20°C. The individual spectra at each urea concentration were recorded from 295 to 400 nm in 1-cm quartz cells at an excitation wavelength of 280 nm. The spectral bandwidth was 5 nm for both excitation and emission. All spectra were baseline-corrected. Protein fluorescence at 320 nm, the wavelength with the largest signal difference between native and unfolded Hsp25, was plotted against urea concentrations.
To follow the development of secondary structure during urea-induced unfolding, far UV CD spectra of Hsp25 (1 mg/ml) in various urea concentrations, ranging from 0 to 7 M, were recorded in a Jasco J715 CD spectropolarimeter. Spectra were measured from 250 to 200 nm at a pathlength of 0.01 cm at 20°C. All spectra were baseline-corrected. The CD signal at 213 nm was used to evaluate the influence of urea on Hsp25 secondary structure.

Dynamic Hsp25 Oligomers Are Formed from Tetramers-
The analysis of the oligomeric assembly of sHsps is important for understanding the function of this family of chaperones. When applied to a size exclusion chromatography (SEC) column, murine Hsp25 eluted predominantly as a 400-kDa complex as judged by calibration of the column with marker proteins. This peak was consistent with a hexadecamer (Fig. 1A). A second peak corresponding to a tetramer was present in smaller amounts. Apart from a shoulder in the tetramer peak, which represents a dimer, no further species were observed. It is therefore likely that the Hsp25 complex is formed from tetramers as basic building blocks. When the Hsp25 complex was cross-linked with glutaraldehyde prior to chromatography, smaller species were no longer detected, indicating that Hsp25 in solution is solely hexadecameric (Fig. 1A). The slight increase in size of cross-linked Hsp25 was probably due to the addition of glutaraldehyde to the complex during cross-linking. The notion that Hsp25 is present as a defined complex in solution was further confirmed by native gel electrophoresis.
Here again, similar to GroEL and ferritin, only one oligomeric species was detected at different Hsp25 concentrations (Fig.  1B). An estimation of the molecular weight was not possible, as the running behavior of proteins on native gels is not only dependent on their molecular weight but also on the isoelectric point of the proteins. As an independent method to determine the molecular mass of Hsp25 complexes, we used analytical ultracentrifugation. Analysis of sedimentation velocity experi-ments at a Hsp25 concentration of 0.28 mg/ml resulted in a s value of s ϭ 11.9 S, confirming a hexadecameric association of Hsp25 with a calculated molecular mass of 398 kDa. As the data can be described by a fit for a monodispersed system (Fig.  1C) under the given conditions, the association of Hsp25 seems to be a defined process yielding only one oligomeric species.
To further investigate the dynamics of the Hsp25 complex, we applied the protein at concentrations ranging from 37.5 g/ml to 1.2 mg/ml to a TSK4000 SEC column (Fig. 2). At all concentrations, the protein appeared in three oligomeric forms corresponding to a hexadecamer, a tetramer and a dimer (Fig.  2, A-C). The relative ratios of the peaks, however, changed significantly with increasing Hsp25 concentrations. Whereas at the lowest concentration, mainly the dimer was present, at 0.2 mg/ml the dimers shifted to tetramers and the hexa- decamer became the predominant species. This tendency carried on up to a concentration of 1.2 mg/ml, where the hexadecamer was by far the predominant oligomeric species. From these data we conclude that Hsp25 forms an oligomer of 16 subunits, which is in a concentration-dependent equilibrium with a tetrameric form. The smaller amount of dimer present at all concentrations and the fact that no monomer was detected suggest that the tetramer is not formed from four monomers but rather consists of two dimers. To further analyze the concentration-dependent equilibrium between Hsp25 oligomers, we collected the hexadecamer and tetramer peaks (see Fig. 2B, solid and dotted lines) and directly reapplied them to the same column. As expected, the 16-mer peak dissociated to the tetrameric form, whereas the rechromatographed tetramer resisted further dissociation, even at this lower concentration (Fig. 2D). This suggests that the tetramer is a rather stable building unit of the high molecular weight Hsp25 oligomer.
Unfolding and Inactivation of Hsp25 Is a Highly Reversible Process-After having established that Hsp25 forms dynamic oligomeric complexes, we were interested in the stability of those structures against chemical denaturation (Fig. 3A) and the changes in quaternary structure involved in unfolding. Incubation of the protein with increasing concentrations of urea led to a gradual decrease of the fluorescence signal over a concentration range of 0.5-5 M urea. (Fig. 3A). As the maxima of the fluorescence spectra changed in the same urea concentration range, the observed transition reflected a gradual loss of tertiary structure, with tryptophans being successively exposed to the polar solvent (data not shown; see Ref. 19). These unfolding intermediates seemed to expose sticky surfaces, as their analysis on SEC HPLC columns was not possible, due to interaction with the resin (data not shown). To analyze changes in oligomeric structure during the urea transition, the light scattering signal of Hsp25 was measured. At urea concentrations above 1 M, the signal decreased drastically indicating a reduction in particle size. The secondary structure, however, as measured by far UV CD-spectroscopy, did not change below 3.5 M urea, with denaturation being completed at 5 M urea (Fig.  3A). Thus, a first step during urea denaturation of Hsp25 seems to be complex dissociation. The following gradual exposure of hydrophobic amino acids and decrease of fluorescence signal indicated further loss of quaternary and/or tertiary structure, followed by disintegration of secondary structure as a last step of unfolding. The unfolding transition was completely reversible (Fig. 3A), leading to correctly assembled Hsp25 oligomers that were indistinguishable from native Hsp25 in gel filtration experiments (data not shown). To analyze whether the renatured protein had regained its full activity, we measured the ability of Hsp25 to suppress the thermal aggregation of the model substrate CS. When compared with native Hsp25, incubated at the same residual urea concentration, the renatured protein showed unchanged activity in the chaperone assay (Fig. 3B). The same is true for Hsp25 renatured from GdnCl (data not shown).
Using GdnCl, another common denaturant, to unfold Hsp25, we found that the protein loses its oligomeric structure as well as its native fluorescence and CD signal in a cooperative reaction with a midpoint at 1.15 M GdnCl (data not shown). We wondered whether the pronounced difference between the two denaturants was due to the ionic nature of GdnCl. To test this hypothesis, we performed urea transitions in the presence of salt (Fig. 4A). NaCl was added to a concentration of 1.15 M, corresponding to the midpoint of transition of the GdnCl measurement. At this salt concentration, the urea transition of Hsp25 changed markedly, exhibiting a plateau of fluorescence intensity between 1.8 and 4.5 M urea. Secondary structure, as measured by far UV CD, did only change above 3 M urea, coinciding with the decrease in fluorescence signal after the plateau region.
To analyze the quaternary structure of Hsp25 in the plateau region, we performed SEC at 1.15 M NaCl and 3 M urea. The elution profile showed two distinct peaks corresponding to a dimer and a tetramer of Hsp25 (Fig. 4B), the established building units of the complex (see above). The stabilization of small oligomeric species by salt without loss of structure suggests that subunit contacts in these species are predominantly hydrophobic.
Hsp25 Temperature Transition-Having established that Hsp25 shows the remarkable ability to refold completely to the native state after urea and GdnCl denaturation, we wondered whether this reversibility of unfolding and disassembly of the Hsp25 oligomers is restricted to chemical denaturation. As sHsp overexpression and function is often correlated with heat FIG. 3. Urea-induced unfolding, refolding, and dissociation of Hsp25. A, samples were incubated in urea concentrations ranging from 0 to 7 M for 20 h at 20°C to achieve equilibrium. All experiments were performed in 40 mM Hepes, pH 7.5. Fluorescence, unfolding (q) and refolding (E) of Hsp25 at a concentration of 10 g/ml in various urea concentrations was followed by recording the fluorescence signal at 320 nm. Far UV CD, changes in secondary structure with increasing urea concentrations (‚) were determined by monitoring the far UV CD signal of Hsp25 at 213 nm (1 mg/ml) at a pathlength of 0.01 cm. Light scattering, to monitor changes in particle size with increasing urea concentrations light scattering of Hsp25 (50 g/ml) was measured at an excitation and emission wavelength of 360 nm (ࡗ). B, to determine the activity of refolded Hsp25, light scattering of thermally aggregating CS (75 nM) was measured at 43°C. After 8 h of incubation at 20°C in 7 M urea, denatured Hsp25 (1.45 mg/ml) was diluted 1:200 into 40 mM Hepes, pH 7.5. After another 22-h incubation at 20°C, refolded Hsp25 (20 nM) was directly applied to the CS aggregation assay (‚). As a control, native Hsp25 at the same concentration was incubated accordingly in the residual urea concentration (35 mM) before application to the assay (E). Spontaneous aggregation of CS was measured in the absence (q) and presence of 35 mM urea (OE).
shock, we investigated the influence of temperature on the structure and stability of Hsp25. Fig. 5A shows the thermal stability of Hsp25 as measured by fluorescence and far UV CD spectroscopy. In agreement with previous data on ␣-crystallin and Hsp27 (41,42), the far UV CD signal of Hsp25 changed drastically at temperatures above 62°C in a highly cooperative thermal transition. The changes in Hsp25 fluorescence above 60°C are less pronounced than the increase in CD signal, but a change in the slope of the linear decrease in fluorescence confirms a folding transition starting at 60 -63°C. To determine the consequences of the thermal transition on the size of the oligomers, we measured light scattering of Hsp25 between 30 and 80°C (Fig. 5B). The increase in signal around 64°C is indicative of a pronounced increase in Hsp25 particle size. Taken together, between 60°C and 65°C, Hsp25 seems to first lose secondary structure (CD), followed by changes in tertiary structure as observed by fluorescence (Fig. 5A). The tendency of the protein to form large particles increased drastically at temperatures above 65°C and reached a plateau at 75°C. When the sample was subsequently cooled down to room temperature, the light scattering signal decreased only slowly, suggesting that the change in quaternary structure of Hsp25 during heating was not readily reversed. A fluorescence spectrum recorded at 75°C showed an emission maximum of 347 nm, indicative of partial unfolding during the heating period (Fig. 5B, inset). However, the fluorescence spectra of Hsp25 before and after heating were identical, suggesting restoration of tertiary structure.
To analyze changes in complex size during heating and cooling more directly, we performed SEC with a sample that had been heated to 80°C for 15 min and then kept at 20°C for variable times (Fig. 6, A-C). When compared with the elution profile of native, untreated Hsp25 (Fig. 6A), the protein eluted as a prominent peak at a retention time corresponding to about 1.3 MDa, whereas the amount of smaller oligomeric species decreased but retained their sizes. As there were no intermediate size complexes observed between the hexadecamer and the larger complex, unspecific aggregation is unlikely. Rather, hexadecamers seem to associate cooperatively to larger oligomers. As the high molecular mass complexes shifted slowly back to native oligomeric Hsp25 when incubated for extended FIG. 4. Influence of NaCl on the urea-induced denaturation of Hsp25. Samples were incubated in urea concentrations ranging from 0 to 7 M for 20 h at 20°C to achieve equilibrium. All experiments were performed in 40 mM Hepes, pH 7.5 Ϯ NaCl. A, Fluorescence, unfolding of Hsp25 at a concentration of 10 g/ml in various urea concentrations from 0 to 7 M in the absence (q) and presence (छ) of 1.15 M NaCl was recorded by measuring the fluorescence signal at 320 nm; Far UV CD, changes in secondary structure with increasing urea concentrations in the presence of 1.15 M NaCl (‚⌬) were monitored by measuring the far UV CD signal of Hsp25 (1 mg/ml) at 213 nm and a pathlength of 0.01 cm. B, SEC in the plateau region of the urea/NaCl transition. 100 g/ml Hsp25 were incubated in 3 M urea, 1.15 M NaCl, 40 mM Hepes, pH 7.5, for 15 h at 20°C. The sample was then centrifuged to remove aggregates and applied to a TSK 4000 SW gel filtration column equilibrated in the same buffer.

FIG. 5. Temperature-dependent structural changes in Hsp25.
A, thermal unfolding and refolding of Hsp25. The intensity of Hsp25 tryptophan fluorescence at 338 nm (50 g/ml) was monitored at a heating rate of 1°C/min (q). Data points were corrected for the temperaturedependent, linear decrease of tryptophan fluorescence. To monitor changes in secondary structure, the far UV CD signal of Hsp25 (150 g/ml) was recorded at 225 nm with a heating rate of 1°C/min (E). B, changes in light scattering of Hsp25 with increasing temperatures. Hsp25 (50 g/ml) was incubated in a stirred quartz cuvette. Light scattering was monitored at 360 nm both during heating to 78°C (q) and subsequent cooling to 25°C (). The heating rate was 0.5°C/min. Inset, fluorescence spectra of Hsp25 (50 g/ml; excitation wavelength: 295 nm) before heating (solid line) at 75°C (dotted) and after cooling at 25°C (dashed). To compensate for the decrease in fluorescence with temperature, the amplitude of the 75°C spectrum was multiplied by a factor of 2.5.
periods at 20°C, this heat shock complex was in a slow equilibrium with the hexadecamer and the tetramer (Fig. 6, A-C).
Next, we analyzed the size and shape of the Hsp25 particles formed at high temperature by negative stain electron microscopy. Hsp25 kept at 20°C formed the already described spherical particles with a diameter of about 20 nm (20). When the protein was applied to the grid after 15 min of incubation at 75°C, the particles associated into a network-like structure and few isolated, large globular particles. However, when Hsp25 was kept at 20°C for 60 min after heating, the appearance of the complexes changed. In accordance with the gel filtration data, a population of larger Hsp25 particles appeared with diameters ranging from 40 to 70 nm with the original 20-nm species still present to some extend (Fig. 3D).
To investigate whether the observed in vitro "granules" showed chaperone activity, we collected both the high molecu-FIG. 6. Influence of elevated temperature on Hsp25 quaternary structure. A, SEC was performed using a Toso-Haas TSK 4000 SW column as described under "Experimental Proccedures." Hsp25 in 40 mM Hepes, pH 7.5, at a concentration of 0.2 mg/ml was incubated for 30 min at 20°C, centrifuged, and applied to the column. B, to test the influence of temperature on the oligomeric organization of Hsp25, the protein at the same concentrations was incubated at 75°C for 15 min, cooled to 20°C for 30 min, centrifuged, and applied. C, the influence of prolonged incubation at 20°C after heat treatment was investigated by keeping the protein at 20°C for 38 h after the 75°C incubation before application (I and II, see panel E). D, analysis of temperature-induced changes in Hsp25 complex size by electron microscopy. Before application to the grid, Hsp25 (0.2 M, 75 g/ ml) was incubated for 60 min at 20°C (left panel) or was incubated at 75°C for 15 min and then directly applied to the grid (middle panel) or cooled to 20°C for another 60 min (right panel). Magnification was 149,000; 1 cm represents 70 nm. E, activity of the different oligomeric Hsp25 species after heat treatment. Hsp25 at a concentration of 1.2 mg/ml was heat-incubated and cooled as described under panel C. The protein was subsequently applied to the TSK 4000 SW column, and the two predominant peaks were collected. After estimating the protein concentration of the samples according to Bradford (55), the material was directly applied to the CS aggregation assay. The activity of the collected peaks was compared with native, untreated Hsp25 at the same concentrations. Thermal aggregation (43°C) of CS (75 nM) alone (q) and in the presence of 10 g/ml peak I (‚) and native Hsp25 (OE) or in the presence of 28 g/ml peak II (छ) and native Hsp25 (ࡗ). lar mass complex and the hexadecamer eluting from the SEC column (see Fig. 6C, I and II). The activity of the samples in suppressing thermal aggregation of CS was measured in comparison to untreated Hsp25. Hsp25 from both peaks showed undiminished activity in suppressing protein aggregation during heat shock conditions (Fig. 6E). Thus, similar to the situation in vivo, large granular sHsp structures, formed after heat stress, play an active role in protecting proteins from irreversible unfolding processes. DISCUSSION The complex quaternary structure of sHsps is one of their most striking features. Their oligomeric state seems to be a prerequisite for chaperone activity, as, e.g., deletion of a Nterminal region of Caenorhabditis elegans Hsp16-2 led to a loss of both oligomerization and chaperone properties (43). Changes in the oligomeric structure of mammalian sHsp due to stress factors are often correlated with phosphorylation of the proteins. In vitro phosphorylated Hsp27 as well as mutants mimicking phosphorylation appear only in their dissociated tetrameric form, which does not show chaperone activity (44). In contrast to the chaperone properties, which seem to be coupled to large oligomers of sHsps, the influence of Hsp27 on actin polymerization has been mainly connected to small oligomeric and even monomeric forms of Hsp27 (45). In the light of these data, changes in the oligomeric structure of sHsps seem to be one means for regulating their multiple functions.
Recently, the crystal structure of Hsp16.5 from the hyperthermophilic archaeon Methanococcus jannaschii has been solved (46). The monomeric folding unit is a composite ␤-sandwich in which one ␤-strand comes from a neighboring subunit, indicating that dimers are involved in the assembly of the oligomer. This 24-mer forms a hollow spherical complex with octahedral symmetry including eight trigonal and six square "windows." In addition, using cryo-electron microscopy and image analysis, a model for the quaternary structure of human ␣B-crystallin has been proposed (30). ␣-Crystallin 32-mers also form hollow spheres with a large central cavity. An important difference to the archaebacterial complex is that the structure is variable in detail with numerous different conformations being observed. Although the shell-like structure is retained, attachment or dissociation of several subunits is tolerated.
In contrast, we show here that murine Hsp25 exists as a hexadecamer in solution with no significant population of other oligomeric species. Similarly, only one spherical, oligomeric species, in this case a 24-mer, was observed for Hsp16.5 from M. jannaschii, supporting the notion that sHsp quaternary structure can be achieved with different numbers of subunits (47). Behlke et al. (48) reported 32-mers as the predominant oligomeric species for Hsp25. This might be due to the association of two hexadecamers. We did not observe such a "dimer of the 16-mer" under various experimental conditions. In our experiments, the 16-mer is in a concentration-dependent equilibrium with tetramers, suggesting tetramers as the basic building block of the hexadecamer. The notion that the 16-mer is formed from four tetramers, which themselves represent dimers of dimers, was confirmed by the finding that tetrameric and dimeric forms appear during Hsp25 denaturation in urea and can be strongly stabilized by the addition of salt. This is in agreement with data showing that phosphorylation of mammalian sHsp results in the formation of tetramers (45,49) and findings that ␣-crystallin forms tetramers under a number of experimental conditions (49 -51). In addition, it was suggested that the minimum cooperatively melting structure of Hsp25 is a dimer (41).
As sHsps are active at conditions where other proteins are denaturing we investigated the thermal stability of the protein.
The size of Hsp25 particles increased drastically at temperatures above 60°C. This is in agreement with data by Dudich et al. (41). Furthermore, ␣-crystallin seems to unfold in this temperature range (42,52). Defined "heat shock" complexes with a size of about 1.3 MDa were detected both by SEC and EM. These were in a slow equilibrium with the Hsp25 hexadecamer, indicating that they represent a defined form of the sHsp rather than an off-pathway reaction. The emergence of high molecular weight particles is reminiscent of several reports in which sHsp were shown to form large complexes (granules) after heat shock (11,28,35,53), which disappeared after return of the cell to permissive folding conditions. These in vivo heat shock granules were reported to be effective in reactivation of heat-denatured nuclear proteins in mammalian cell lines (53,54). In these studies, overexpression of Hsp27 led to a marked acceleration of recovery after heat shock, which was interpreted to be due to the refolding of thermally unfolded proteins. A similar picture arises for the high molecular mass complex formed by Hsp25 after heat treatment in vitro. The isolated 1.3-MDa peak and the 16-mer peak both showed chaperone activity comparable to untreated Hsp25. However, it is important to note that, although the similarities between the two species are striking, it is not established that the in vitro aggregates correspond directly, both functionally and structurally, to the heat shock granules found in vivo. This remains to be elucidated in future studies.
Interestingly, tetramers, which we defined here as important intermediates in the assembly process of Hsp25, have also been shown to bind unfolded protein (22), indicating that different oligomeric Hsp25 species can posses chaperone activity (see Fig. 7).
From the data presented, we suggest the following, comprehensive model for Hsp25 structure and function (Fig. 7). We could show that unfolding of Hsp25 both after thermal and chemical denaturation is a reversible process, proceeding via dimers and tetramers to the native hexadecameric form of the protein. As no monomer or additional oligomeric form could be detected under any experimental conditions, association/dissociation seems to be a highly ordered process including only a few defined steps. During incubation at elevated temperatures, Hsp25 assembles into even bigger oligomeric complexes, which are in a slow equilibrium with the hexadecamer. Importantly, the large granular form of Hsp25 as well as tetrameric and hexadecameric species are active in preventing heat-induced aggregation of model substrates. Taken together, our data show that Hsp25 oligomers are remarkably dynamic in structure and function, performing chaperone activity at different levels of quaternary structures.