Biochemical Characterization of the Small Heat Shock Protein IbpB from Escherichia coli*

Escherichia coli IbpB was overexpressed in a strain carrying a deletion in the chromosomalibp operon and purified by refolding. Under our experimental conditions, IbpB exhibited pronounced size heterogeneity. Basic oligomers, roughly spherical and approximately 15 nm in diameter, interacted to form larger particles in the 100–200-nm range, which themselves associated to yield loose aggregates of micrometer size. IbpB suppressed the thermal aggregation of model proteins in a concentration-dependent manner, and its CD spectrum was consistent with a mostly β-pleated secondary structure. Incubation at high temperatures led to a partial loss of secondary structure, the progressive exposure of tryptophan residues to the solvent, the dissociation of high molecular mass aggregates into ≈600-kDa oligomers, and an increase in surface hydrophobicity. Structural changes were reversible between 37 and 55 °C, and, up to 55 °C, hydrophobic sites were reburied upon cooling. IbpB exhibited a biphasic unfolding trend upon guanidine hydrochloride (GdnHCl) treatment and underwent comparable conformational changes upon melting and during the first GdnHCl-induced transition. However, hydrophobicity decreased with increasing GdnHCl concentrations, suggesting that efficient exposure of structured hydrophobic sites involves denaturant-sensitive structural features. By contrast, IbpB hydrophobicity rose at high NaCl concentrations and increased further at high temperatures. Our results support a model in which temperature-driven conformational changes lead to the reversible exposure of normally shielded binding sites for nonnative proteins and suggest that both hydrophobicity and charge context may determine substrate binding to IbpB.

Despite their presence in all organisms examined to date, small heat shock proteins (sHsps) 1 remain one of the most disparate family of stress proteins. More than 25 different sHsps have been identified in the cytosol and organelles of heat-shocked plants (1); mammalian and yeast cells encode fewer sHsps (2). The overall homology between sHsps isolated from diverse sources is not as pronounced as that found in other stress protein families. In fact, sHsps are only grouped as a family of heat shock proteins based on a low degree of homol-ogy in a core region of about 85 amino acids (the ␣-crystallin domain), their ability to be induced by cellular stress, and their small protomer molecular mass, which usually ranges between 15 and 30 kDa. Because of structural and functional similarities, the major eye lens proteins ␣Aand ␣B-crystallin are also included in the sHsp family.
sHsps are abundant in unstressed cells (3), and their synthesis is up-regulated by cellular stress as well as a variety of additional signals related to the growth state and oncogenic status of the cell (e.g. Refs. 4 -6). Most sHsps assemble into large oligomers in a process that requires an intact NH 2 terminus (7). In mammalian cells, sHsps form complexes ranging from 300 to 800 kDa (2,8). ␣-Crystallin, which is normally isolated as an 800-kDa oligomer, has been reported to range in size from 280 kDa to 10 MDa (for a review, see Ref. 9). Plant sHsps exhibit a more uniform size distribution (Ϸ200 -240 kDa); pea Hsp18.1 consists of 12 subunits arranged in a globular structure, whereas pea Hsp17.7 is organized in both round and triangular structures (10). The recently crystallized Methanococcus jannaschii Hsp16.5 (MjHsp16.5) is organized as a hollow sphere built from 24 subunits (11). Finally, Mycobacterium tuberculosis Hsp16.3 forms another type of oligomeric structure, a 150-kDa triangular trimer of trimers (12).
sHsps are believed to play a key role in thermoprotection because their overexpression confers a significant degree of thermotolerance to host cells (13)(14)(15)(16)(17). Mammalian sHsps appear to perform this function by stabilizing cytoskeletal elements such as actin (7,18,19). In other systems, sHsps may increase thermotolerance by preventing the aggregation of cellular proteins accumulating as a result of cellular stress. Indeed, a number of sHsps suppress the thermal aggregation of model proteins in vitro (10, 12, 20 -23). Furthermore, murine Hsp25, human Hsp27, and pea Hsp18.1 and Hsp17.7 exhibit a bona fide molecular chaperone function because they partially improve the reactivation of certain chemically unfolded proteins (10,20). In the above cases, the addition of ATP to the refolding buffer is not necessary to enhance refolding. Nevertheless, the reactivation yields of citrate synthase (CS) by human ␣B-crystallin are increased 2-fold in the presence of ATP (17). The formation of large oligomeric structures appears to be necessary for molecular chaperone function because NH 2 -terminally truncated variants of Caenorhabditis elegans Hsp16.2 and the only known monomeric sHsp, C. elegans Hsp12.6, are unable to suppress the aggregation of test substrates (7,24). Purified murine Hsp25 (23) and pea Hsp18.1 (22) have been shown to bind thermally denatured proteins on their surface and maintain them in a folding-competent state. This observation has led to the idea that sHsps act as a reservoir of partially folded proteins that are accessible for refolding by ATP-dependent molecular chaperones once the effect of cellular stress has abated (23).
In contrast to the relatively large amount of information available on eukaryotic sHsps and ␣-crystallins, little is known about their Escherichia coli homologs. The approximately 16-kDa E. coli IbpA and IbpB polypeptides were first discovered as contaminants tightly associated with recombinant protein inclusion bodies (25). These inclusion-body-associated proteins were later found to cosediment with host proteins aggregated intracellularly by heat shock (26). The ibpA and ibpB genes are part of an E 32 -transcribed operon that is the strongest heatinducible operon of E. coli (27). The bacterial sHsps are about 50% identical to each other at the amino acid level and share 20 -27% identity with homologs from plant chloroplasts, Drosophila, C. elegans, and Xenopus (25,27). IbpAB are dispensable in E. coli, and deletion of the ibp operon only leads to modest effects on growth and viability at high temperatures (28). The fact that a ⌬ibp mutation exerts a deleterious effect on the growth of dnaK756 but not groES30 mutants at high temperatures suggests that IbpAB cooperate with DnaK-DnaJ-GrpE but not with GroEL-GroES in vivo (28). This result is in agreement with a recent investigation of the interplay between IbpB and the major chaperone systems showing that purified IbpB forms stable complexes with denatured malate and lactate dehydrogenase and that the bound proteins are specifically delivered to the DnaK-DnaJ-GrpE, but not to the GroEL-GroES system in vitro (29).
To gain further information on the structure-function relationship of bacterial sHsps, we have purified and refolded IbpB from a ⌬ibp strain. We show that despite its pronounced size heterogeneity, IbpB shares many of the traditional features of sHsps. Our results indicate that reversible structural modifications induced by temperature changes in the 32-50°C range drive the exposure of denaturant-sensitive hydrophobic regions that may constitute substrate binding sites for nonnative proteins. Charge effects appear to be important for the optimal exposure of hydrophobic domains to the solvent.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-Plasmid pTb-ibpAB and pTb-ibpA are pT7blue (Novagen) derivatives carrying polymerase chain reactionamplified DNA fragments encoding the complete ibp operon or the ibpA gene without their native promoter. 2 To place the ibp operon (or the ibpA gene) under transcriptional control of the T7 promoter, plasmid pTb-ibpAB (or pTb-ibpA) was digested with XbaI and EcoRI, and the small fragment was ligated into the same sites of pET-22bϩ (Novagen) to yield pJSibpAB (or pJSibpA). Plasmid pJSibpAB was digested with NdeI, and the backbone was religated. In the resulting plasmid, pJSibpB, the T7 promoter is followed by the authentic ibpB ribosome binding site, coding sequence, and transcription terminator. All constructs were verified by restriction analysis and DNA sequencing. The ⌬ibp1::kan null mutation (28) was moved to E. coli BL21(DE3) (Novagen) by P1 transduction to yield JGT14. This strain was used as a host for recombinant protein production.
Purification of IbpB-JGT14(pJSibpB) cultures were grown at 30°C in 500 ml of LB medium supplemented with 0.2% glucose and 50 g/ml carbenicillin. At A 600 Ϸ 0.4, the cultures were induced with 1 mM isopropyl-␤-D-thiogalactopyranoside and incubated at 30°C for an additional 5 h. Cells were harvested by centrifugation at 7,000 ϫ g for 10 min, resuspended in 15 ml of 25 mM NaH 2 PO 4 ⅐H 2 O, pH 7.0 (buffer A), and disrupted by chemical lysis (30). Briefly, the paste was incubated on ice with 400 l of 10 mg/ml lysozyme and 40 l of 9 mg/ml phenylmethylsulfonyl fluoride for 20 min. DNase (200 units) and deoxycholic acid (20 mg) were added, and the mixture was incubated at 37°C for 30 min. Cellular debris was removed by centrifugation at 17,000 ϫ g for 30 min. The clarified lysate was supplemented with 0.8 ml of 200 mM dithiothreitol and 0.4 ml of 5% polyethyleneimine and incubated at room temperature for 10 min to precipitate contaminating nucleotides. After centrifugation at 17,000 ϫ g for 30 min, ammonium sulfate was added to the supernatant to a concentration of 20% (w/v). Precipitated proteins were sedimented by centrifugation at 10,000 ϫ g for 15 min and discarded. Ammonium sulfate was added slowly to 40% (w/v), and insoluble proteins were recovered by centrifugation after a 45-min incubation with stirring on ice. The protein pellet was resuspended in 2 ml of 25 mM NaH 2 PO 4 ⅐H 2 O, pH 7.0, containing 5 M GdnHCl (buffer B) and incubated on ice for 2 h. Denatured proteins were loaded onto a Superose 6 HR column (Amersham Pharmacia Biotech) equilibrated and developed at 0.3 ml/min in buffer B. Fractions eluting between 54.5 and 64.5 min from multiple runs were pooled. Purified IbpB was refolded at 4°C by dialysis against buffer A for 16 h, with buffer exchange at 3 and 12 h, using Spectra/Por dialysis tubing (VWR Scientific) of molecular mass cutoff 12-14 kDa. The protein was concentrated to 0.3-0.5 mg/ml using Centriplus-100 concentrators (Amicon) of molecular mass cutoff 100 kDa. Further concentration favored the formation of insoluble aggregates. This operation also led to the removal of contaminating low molecular mass proteins, and the resulting IbpB was Ϸ95% pure. IbpB samples were centrifuged at 12,000 ϫ g for 5 min before all experiments, and concentrations were determined using the Bradford assay (Bio-Rad). All concentrations are reported in terms of protomers.
Size Aggregation Suppression Experiments-Thermal aggregation suppression experiments were performed using pig heart CS and equine liver alcohol dehydrogenase (Sigma) essentially as described (21,31).  (33). In all cases, stock solutions were stored on ice until mixing and transferred rapidly to thermostated fluorescence cuvettes.
Unfolding Experiments-For chemical denaturation studies, IbpB samples (1 M) were incubated for 1 h at room temperature in buffer A supplemented with the indicated concentrations of GdnHCl. Equilibrium was reached during this time period. The tryptophan red shift was measured in a Hitachi F4500 fluorescence spectrophotometer at an excitation wavelength of 280 nm and slit widths set at 2.5 nm. Emission spectra were plotted between 310 and 380 nm and fitted with a ninthorder polynomial equation. Tryptophan maximum emission wavelengths were obtained from the fit. Samples were next supplemented with 10 M bis-ANS (Molecular Probes), and emission spectra were recorded after a 5-min incubation using an excitation wavelength of 396 nm. Data were plotted between 450 and 550 nm, fitted with a ninthorder polynomial equation, and maximum bis-ANS emission wavelengths and intensities were calculated from the fit. For temperature denaturation studies, single protein samples (1 M) were heated from low to high temperatures in a thermostated spectrophotometer cell with a 1-h incubation at the indicated points. This time period was sufficient to reach equilibrium. Tryptophan max were obtained as above. Reversibility was assessed by cooling the sample down to the indicated temperatures and measuring max after 1 h. Five scans were averaged for all experiments, and the background fluorescence of buffers alone or buffers supplemented with bis-ANS was subtracted from sample spectra.
CD Spectroscopy-IbpB samples at a 0.05 mg/ml (0.62 M) final concentration were incubated at room temperature for 1 h in the presence of the indicated concentrations of GdnHCl or NaCl or held for 1 h at the indicated temperatures in a thermostated cell of 1-mm path length. CD spectra were recorded on a Jasco 720 spectropolarimeter. Spectra were smoothed using the operating software noise reduction filter and represent the average of 16 scans corrected for buffer absorbance.
Transmission Electron Microscopy-IbpB (1 M) in buffer A was stained with 0.5% uranyl acetate. A 1-l aliquot was applied to a Formvar-coated copper grid, vacuum dried, and carbon coated in a 2 J. G. Thomas and F. Baneyx, manuscript in preparation. carbon evaporator. A Philips EM400 transmission electron microscope operating at 100 keV was used to image the sample at magnifications ranging between 6,000 and 280,000.
Dynamic Light Scattering-IbpB (1 M) in 6 ml of buffer A was filtered through a 0.22-m membrane directly into a prewashed light scattering cell. The cell was inserted into the goniometer of a Brookhaven laser spectrometer (Brookhaven Instruments, Holtsville, NY) and data were collected at room temperature at a scattering angle of 45°with sampling time of 5 min and response times between 2 and 200 s using a Brookhaven BI 9000AT autocorrelator (34). Particle size distribution was calculated using the CONTIN method (35).
Bis-ANS Photoincorporation-IbpB (2.8 M) in 22 l of buffer A containing 100 M bis-ANS and supplemented or not with 1 M NaCl was incubated in Eppendorf tubes held at 23, 37, or 45°C for 1 h. Samples were next irradiated for 20 min at the indicated temperatures by placing a hand-held UV lamp (Mineralight UVGL-25 at 254 nm setting) on top of the open tubes. Reducing SDS loading buffer was added, and samples were fractionated on a 15% SDS minigel after a 5-min incubation at 95°C. The gel was photographed with a Kodak DC120 digital camera on a UV transilluminator and later stained using a colloidal Coomassie staining kit (Novex). The intensity of the fluorescent bands was quantified using NIH Image 1.60 for PowerPC.

Expression and Purification of the Bacterial sHsps-Two
pET-22bϩ derivatives encoding the individual ibpA and ibpB genes under transcriptional control of the bacteriophage T7 promoter were constructed by polymerase chain reaction as described under "Experimental Procedures." To avoid the complication of mixed oligomer formation associated with the presence of the chromosomal ibp operon, a BL21(DE3) derivative bearing the ⌬ibp1::kan null mutation (28) was generated by P1 transduction and used as an expression host. Fig. 1A shows that each sHsp was produced at moderate and comparable levels following isopropyl-␤-D-thiogalactopyranoside induction of the transformants. However, whereas overproduced IbpA partitioned approximately equally between soluble and insoluble cellular fractions, IbpB accumulated mostly in a soluble form.
We first attempted to purify native IbpA and IbpB by standard chromatography (e.g. DEAE, Mono Q and hydroxyapatite). Under these conditions IbpA, and to a lesser extent IbpB, smeared over many fractions. The bacterial sHsps were therefore purified by ammonium sulfate precipitation followed by denaturing gel filtration chromatography and refolded by dialysis. Although the refolding yield of IbpB was about 60%, little IbpA (Ͻ5%) could be refolded, and the purified protein had a high tendency to form insoluble aggregates upon concentration and storage. The further characterization of IbpA was therefore not pursued in the present study. Purified IbpB was approximately 95% pure as judged by videodensitometric scanning of overloaded gels (Fig. 1B).
Secondary and Quaternary Structure of IbpB-At 20°C, the far UV CD spectrum of IbpB is virtually identical to that of ␣-crystallins (36 -38) and very similar to that of other sHsps (38). There is a single minimum at 213 Ϯ 1 nm with mean residue ellipticity at 213 nm, [] 213 , of Ϫ3.3 millidegrees cm 2 dmol Ϫ1 (Fig. 2, trace 1), and no fine features are obvious between 200 and 245 nm. The spectrum is consistent with of a mostly ␤-pleated structure that appears to be highly conserved among sHsps (11,39).
sHsps typically form large oligomers that range in size between 300 and 800 kDa. Because most IbpB (Ͼ85%) did not flow through the membrane of 100-kDa molecular mass cutoff concentrators during the purification process, the bacterial sHsps also appeared to assemble into an oligomeric structure. Analysis by size exclusion chromatography was performed on a calibrated BioSep-S4000 column (Fig. 3A). Although a large fraction of the injected material remained irreversibly bound to the gel filtration matrix, the majority of the eluted IbpB was found as a rather broad peak above the 2-MDa marker and near the void volume of the column. In agreement with this result, Veinger et al. recently reported a molecular mass in excess of 2 MDa for IbpB purified under native conditions (29). Inspection of negative stain electron micrographs of IbpB at room temperature revealed pronounced size heterogeneity (Fig. 3B). The smallest particles were roughly spherical and 10 -20 nm in diameter ( Fig. 3B; data not shown). These oligomers interacted to form larger particles in the 100 -200 nm range (Fig. 3C), which themselves appeared to aggregate into m size structures of variable shape (Fig. 3B). The existence of IbpB populations of average sizes 13 Ϯ 3 and 190 Ϯ 24 nm was confirmed by dynamic light scattering after filtration through 0.22-m filters. Larger particles ranging in size between 400 nm and 1.7 m were detected after a 48-h incubation at room temperature. Because IbpB was quantitatively recovered upon filtration through 0.22-m membranes (data not shown), the amorphous, soluble aggregates appear to be loose structures that arise over time from the interaction of smaller particles.
IbpB Reduces the Thermal Aggregation of CS and Alcohol Dehydrogenase-One of the hallmarks of molecular chaperones is their ability to suppress the aggregation of thermally or chemically unfolded substrate polypeptides. Fig. 4A shows that  purified IbpB alleviated the thermal aggregation of CS at 45°C in a concentration-dependent manner, indicating that the refolded protein is functional in a standard test of molecular chaperone function. Aggregation was suppressed by about 50% in the presence of 6 M IbpB, a concentration corresponding to a CS:IbpB ratio of 1:20 based on monomers. Half-suppression of CS thermal aggregation is observed at comparable CS:sHsp ratios (Ϸ1:11-1:15) with pea Hsp17.7 (10) and C. elegans Hsp16.2 (7), and the same result is achieved at the much smaller ratio of 1:3 with pea Hsp18.1 (10) and murine Hsp25 (23). Because we were unable to demonstrate complete suppression of CS aggregation at higher IbpB concentrations due to the tendency of the purified protein to aggregate when stock solutions were concentrated above 0.5 mg/ml, similar experiments were repeated with alcohol dehydrogenase at 41.5°C. Fig. 4B shows that IbpB prevented the thermal aggregation of this enzyme much more efficiently, although residual aggregation was still observed at late time points in the presence of 5 M IbpB (a concentration corresponding to a alcohol dehydrogenase:IbpB ratio of 1:1 based on monomers).
Temperature Affects IbpB Conformation-Temperature-dependent structural changes have been proposed to play a key role in the mechanism of action of sHsps (22). To determine how an increase in temperature would affect the secondary structure of IbpB, far UV CD spectra were recorded after incubating the protein for 1 h at 37 or 55°C. Fig. 2 shows that both [] 213 and [] 220 decreased in a linear fashion upon temperature increase, suggesting a progressive loss of secondary structure and/or extensive structural rearrangements. Despite the appearance of a shoulder centered at Ϸ230 nm in the sample incubated at 55°C (Fig. 2, trace 3), IbpB retained a significant amount of ␤-structure at this temperature, as evidenced by the persistence of a trough at 213 nm and the lack of the strong minimum at 200 -204 nm which is characteristic of completely denatured proteins (40). Upon cooling to 20°C, the CD spectrum of IbpB that had been held at 55°C became virtually identical to that recorded at 37°C (Fig. 2, trace 4). Thus, temperature-driven changes in IbpB secondary structure are either only partially reversible, or reversibility is limited to an upper range of temperature.
IbpB contains two tryptophan residues at positions 13 and 83 (25) which are suitable to probe the conformation of the protein by intrinsic fluorescence spectroscopy. To gain further information on temperature-induced structural rearrangements, the maximum emission wavelength of the tryptophan residues ( max ) after excitation at 280 nm was measured using a protein sample that had been allowed to equilibrate for 1 h at temperatures ranging between 5 and 80°C. Below 20°C, max was approximately 338.5 nm, indicative of the fact that tryptophan residues experience a hydrophobic environment at low temperatures (Fig. 5A). Between 22 and 50°C, a single cooperative transition to longer wavelengths was observed with midpoint at about 32°C. Further heating to temperatures as high as 80°C did not cause any significant change in max , which remained approximately equal to 347 nm (Fig. 5A). The reversibility of the thermally induced conformational change was next studied by heating samples to 32, 50, or 60°C and bringing them to lower temperatures in a successive series of cooling steps. Fig. 5B shows that the structural modifications experienced by IbpB between 10 and 32°C are irreversible because the max of the cooled samples is red shifted by about 5 nm relative to the unheated control. However, structural changes appear to be perfectly reversible between 32 and 50°C (Fig. 5B, circles) and reversible between 32 and 60°C (squares) despite a more pronounced hysteresis. These results are in good agreement with the CD data of Fig. 2 and suggest that reversible conformational changes only occur in an upper range of temperatures.
To investigate the effect of temperature increases on the quaternary structure of IbpB, ice-cold samples of the purified protein were injected on a calibrated BioSep-S4000 size exclusion column immediately before or after a 1-h incubation at either 30 or 50°C, and the column was developed at room temperature. Fig. 6A shows that incubation at high temperatures led to a progressive decrease in the intensity of the high molecular mass peak and the concomitant appearance of a new species comparable in size to typical sHsp oligomers (approximate molecular mass, 605 kDa). To confirm these results, aliquots of ice-cold IbpB were transferred rapidly to buffer held at various temperatures, and the change in right angle scattering intensity was recorded over time (Fig. 6B). In agreement with size exclusion chromatography data, incubation between 23 and 42°C resulted in a progressive decrease in scattering intensity, as would be expected if large aggregates were resolved into smaller particles. At 48°C, however, the initial decrease in light scattering was followed by a slow rise in intensity, and a plateau value was reached after about 1,000 s. Because the magnitude of the light scattering increase was relatively modest (Ϸ20%) and did not change at an IbpB concentration of 4 M (data not shown), this result is best explained by an increase in the number of scattering centers at high temperatures. Finally, we did not observe significant changes in light scattering intensity when IbpB samples heated to a steady-state scattering plateau at 50°C were cooled to 15°C over 45 min (data not shown), indicating that IbpB does not undergo extensive aggregation upon temperature downshift.
High Temperatures Trigger the Exposure of Hidden Hydrophobic Domains-The fluorescent probe bis-ANS exhibits little intrinsic fluorescence but becomes highly fluorescent when bound to hydrophobic domains in proteins. Because molecular chaperones rely on the transient exposure of hydrophobic patches to capture partially folded substrate proteins, bis-ANS has been used extensively in the study of their mode of action. Because the bis-ANS quantum yield varies with temperature, the intensity of emission spectra collected at different temperatures cannot be directly compared. It has, however, been shown that bis-ANS can be incorporated in the vicinity of its binding site by exposing protein samples to shortwave UV light (41,42) and that the efficiency of this process is independent of the temperature (22). To determine if an increase in temperature would lead to the exposure of hidden hydrophobic regions, aliquots of IbpB were incubated in the presence of an excess of bis-ANS at 23, 37, or 45°C, and the probe was photoincorporated into the sHsp by UV irradiation. Fig. 7A shows that approximately 50% more bis-ANS was incorporated into IbpB at 37°C relative to 23°C and that this value raised to Ϸ80% at 45°C. The reversibility of the process was tested by incubating IbpB to various temperatures for 1 h, transferring the samples to room temperature for 5 h, and recording the bis-ANS emission spectra after the addition of 10 M probe. There was essentially no variation in bis-ANS emission intensity or maximum wavelength below 55°C. However, a progressive loss in emission intensity and a slight red shift in max were observed above 55°C (Fig. 7B). Taken together, the above results indicate that the temperature-driven exposure of buried hydrophobic regions in IbpB is reversible in the physiological temperature range for E. coli.
Chemical Unfolding Studies-To compare the effects of temperature and chatropes on IbpB conformation, intrinsic tryptophan fluorescence experiments were repeated after incubation in the presence of up to 5 M GdnHCl. IbpB exhibited a clearly biphasic unfolding behavior (Fig. 8A, ‚) and was highly sensitive to the presence of small concentrations of the denaturant. The tryptophan max changed smoothly from 339 to 348 nm between 0 and 1 M GdnHCl with transition midpoint at Ϸ300 mM GdnHCl. The shape and magnitude of the tryptophan red shift were highly similar to that observed in the melting experiments of Fig. 5 and most likely correspond to the exposure of one of the two tryptophan residues to the solvent. A second cooperative transition centered at 2.3 M was detected between 1.8 and 3 M GdnHCl. Thereafter, max remained unchanged, suggesting that IbpB is denatured completely above 3 M GdnHCl and that both tryptophan residues are solventexposed. The CD spectra of IbpB incubated in the presence of 1.2 M GdnHCl (e.g. after the first transition) and 4 M GdnHCl (after the second transition) were recorded down to 215 nm because of the strong absorption of these denaturants below this wavelength. Traces 1-3 in Fig. 8B indicate that major changes in secondary structure occur between 0 and 1.2 M GdnHCl. At the latter denaturant concentration (trace 2), no feature is apparent between 215 and 245 nm, and the mean ellipticity at 220 nm is Ϫ1.1 millidegrees cm 2 dmol Ϫ1 . This value falls between those obtained when IbpB is incubated at 37 and 55°C ( Fig. 2; [] 220 ϭ Ϫ1.6 and Ϫ0.8 millidegree cm 2 dmol Ϫ1 , respectively). Assuming that [] 220 varies linearly with the incubation temperature between 20 and 55°C (as suggested by the data of Fig. 2), the overall secondary structure of IbpB in the presence of 1.2 M GdnHCl should be similar to that it displays at 48.5°C. Again, this result is in very good agree-  (Fig. 8B, trace 3), the IbpB spectrum exhibits a shoulder at 223 nm and a [] 220 value of Ϫ0.25 millidegree cm 2 dmol Ϫ1 which are consistent with a denatured conformation containing residual secondary structure.
To determine if hydrophobic domains become solvent-exposed during GdnHCl denaturation (and in particular during the first transition), IbpB treated with increasingly high concentrations of GdnHCl was supplemented with a saturating amount of bis-ANS, and emission spectra were recorded after excitation at 396 nm. Although the bis-ANS fluorescence intensity (Fig. 8A, q) increased by about 25% in the presence of 300 mM GdnHCl, this change was transient, and only a slight increase in intensity was detected at similar urea concentrations (data not shown). As the GdnHCl concentration was raised to Ϸ2 M, the bis-ANS intensity decreased while its maximum emission wavelength (E) remained approximately constant. These results indicate that bis-ANS binding sites are progressively lost but that the hydrophobic character of the remaining sites is unchanged. Above 2.5 M GdnHCl, the further decrease in bis-ANS intensity was accompanied by a rise in emission wavelength, a situation consistent with a loss of organized hydrophobic surfaces as IbpB protomers unfold. Thus, although low concentrations of GdnHCl and incubation at high temperatures appear to exert similar effects on the secondary and tertiary structure of IbpB, only temperature increases trigger the progressive exposure of structured hydrophobic regions in the bacterial sHsp.
Ionic Effects-A possible explanation for the fact that low concentrations of GdnHCl but not urea lead to a slight increase in bis-ANS intensity ( Fig. 8A; data not shown) is that the ionic character of GdnHCl, rather than its denaturing ability, is responsible for the transient exposure of hydrophobic domains in the sHsp. To investigate ionic effects in more detail, bis-ANS binding experiments were repeated using IbpB that had been incubated with up to 1 M NaCl at room temperature. Under these conditions, the bis-ANS emission maximum remained constant (Fig. 9, E), whereas the emission intensity increased linearly with the salt concentration (q). Thus, bis-ANS binding sites of comparable hydrophobic character are progressively gained upon salt addition. Because the tryptophan emission max is unaffected by NaCl (Fig. 9, ‚) and the CD spectrum of IbpB recorded in the presence of 1 M NaCl is comparable to that obtained in the absence of salt (traces 1 and 4 in Fig. 8B), high salt concentrations do not appear to induce major secondary or tertiary structural changes. To investigate possible synergism between temperature and ionic effects, the photoincorporation experiment of Fig. 7A was repeated in the presence of 1 M NaCl. Under these conditions, bis-ANS photoincorporation was reduced by 30 -40% relative to unsupplemented samples at all temperatures tested (data not shown). The most likely explanation for this result is that high salt concentrations interfere with the photooxidation process. Nevertheless, the increase in bis-ANS photoincorporation with the incubation temperature was comparable whether the buffer was supplemented or not with NaCl (data not shown), suggesting that buried hydrophobic domains are not fully exposed by ionic perturbations. DISCUSSION Despite a low degree of homology restricted to a core region known as the ␣-crystallin domain, all sHsps characterized to date share common structural and functional features: (i) their protomers are mostly ␤-pleated with little ␣-helical content, as judged from far UV CD spectroscopy and crystallographic data (11); (ii) they form large oligomers ranging in molecular mass between 200 and 800 kDa; and (iii) they can suppress the aggregation of thermally or chemically denatured model proteins by capturing folding intermediates on their surface (22,23). Although E. coli IbpB behaves as a typical sHsp in terms of function and protomer secondary structure (Figs. 2 and 4), it exhibits uncommon size heterogeneity (Fig. 3). The basic oligomer appears to be a roughly spherical particle approximately 15 nm in diameter and 600 kDa in apparent molecular mass, a size and morphology comparable to those of other sHsps (22,23,43,44). Under our experimental conditions, these oligomers interacted to form larger structures in the 100 -200-nm range, which themselves loosely associated to yield amorphous, micrometer size clusters. The eye lens protein ␣-crystallin, which has a dynamic oligomeric structure (44,45), has also been isolated as distinct populations in the 600 -900-kDa, 900-kDa-4-MDa, and greater than 10-MDa ranges (46). However, because concentration, pH, ionic strength, and temperature of isolation have all been reported to affect the oligomeric state of ␣-crystallins (47,48) and because IbpA may modulate the quaternary structure of IbpB, it is not clear whether large IbpB oligomers also exist in vivo.
Temperature increases have profound effects on IbpB conformation which are characterized by: (i) a partial loss of secondary structure (Fig. 2); (ii) the progressive exposure of tryptophan residues to the solvent (Fig. 5A); (iii) the dissociation of large aggregates into constituent oligomers (Fig. 6); and (iv) an increase in protein surface hydrophobicity (Fig. 7). Several lines of evidence suggest that these temperature-driven structural modifications have a functional role. First, IbpB is remarkably thermostable and retains a significant amount of secondary and tertiary structure up to 55°C (Figs. 2 and 5). Second, conformational changes appear to be reversible between the physiological (37°C) and maximum growth temperature (50°C) for E. coli (Figs. 2 and 5B). Finally, after incubation to temperatures as high as 55°C, solvent-exposed hydrophobic patches are reburied upon cooling (Fig. 7B). An increase in bis-ANS labeling with the incubation temperature has been reported previously for ␣-crystallin (49, 50) and pea Hsp18.1 (22), leading to the idea that temperature-induced structural rearrangements promote the exposure of normally hidden hydrophobic regions that allow sHsps to bind and stabilize partially folded proteins as they denature. Our results are fully consistent with this hypothesis. Such a "molecular sponge" function would prevent saturation of the DnaK-DnaJ-GrpE and GroEL-GroES major chaperone systems at high temperatures. As the temperature returns to physiological values, IbpB would act as a "reservoir" of partially folded proteins for the DnaK-DnaJ-GrpE system (22,23,29), thereby eliminating the need for de novo synthesis of thermolabile proteins. Alternatively, bound proteins could be presented to the cellular degradation machinery for recycling.
Although high temperatures and incubation with up to 1.2 M GdnHCl appear to induce comparable structural changes in IbpB (Figs. 2, 5, and 8), bis-ANS binding decreases above 0.3 M GdnHCl while it increases progressively upon temperature upshift (Fig. 7A). This result suggests that efficient exposure of structured hydrophobic patches to the solvent involves fine structural features that are readily disrupted at low denaturant concentrations. The bis-ANS binding sites of pea Hsp18.1 (22) and ␣-crystallin (51) have recently been identified and are located at the NH 2 terminus of the ␣-crystallin domain which is well conserved among all sHsps (39). Based on the observations that bis-ANS photoincorporation decreases when these sHsps are preincubated with denatured proteins (22,51) and that the bis-ANS binding sites of ␣-crystallin correspond to sequences involved in substrate binding as determined by cross-linking experiments (51), it has been proposed that the binding sites for bis-ANS and nonnative proteins are identical in sHsps (51). Pairwise alignments between IbpB and Hsp18.1 indicate that the homologous IbpB domain encompasses residues 50 -67 and has as its sequence ALAGFRQEDLEIQLEGTR. The Hsp18.1 region that binds bis-ANS maps in a surface-exposed loop that links the third and fourth ␤-strands of the composite ␤-sandwich in each folding unit of the MjHsp16.5 crystal structure (11). Whether low concentrations of GdnHCl disrupt an organized hydrophobic site involving this loop or affects IbpB hydrophobicity in a different fashion remains to be determined.
An interesting feature of the melting study of Fig. 5 is that the structural changes that take place between 10 and 32°C are not reversible. Although the physiological significance of this phenomenon remains unclear, it is possible that the cell needs to maintain any IbpB synthesized at low temperatures (e.g. during the Hsp up-regulation period that accompanies recovery from cold shock (52,53)) in a minimally active form. Around 32°C, a thermally induced rearrangement may activate the protein into a fully functional form. In support of this hypothesis, evidence for thermal switching events have been reported between 25 and 30°C in the case of GroEL (54,55) and around 30°C for ␣-crystallin (56).
Our observation that IbpB hydrophobicity increases with NaCl concentration is consistent with the hypothesis of ionic triggering of hydrophobic surface exposure proposed by Horowitz and co-workers for GroEL (57). In this model, hydrophobic stretches in partially folded protein substrates are presented to the chaperonin in a charged context to stimulate and/or lead to the optimal exposure of hidden hydrophobic binding sites to the solvent. In agreement with this idea, we noted that although 1 M NaCl raises the hydrophobicity of IbpB by about 40% at 23°C (Fig. 9), the increase in bis-ANS photoincorporation with the incubation temperature remained comparable whether or not NaCl was present in the buffer (Fig. 7 and data not shown). NMR solution studies have revealed the presence of flexible COOH-terminal extensions 8 -18 residues in length in both ␣-crystallins and mouse Hsp25 (58,59). In MjHsp16.5, the ultimate 6 COOH-terminal residues in one subunit reach out to interact with ␤4 and ␤8 of a neighboring subunit (11), a location that is in close proximity to the bis-ANS photoincorporation site. Because both the COOH-terminal end and putative bis-ANS/substrate binding site of IbpB are rich in charged residues, it is tempting to hypothesize that the bacterial sHsp possesses a flexible COOH terminus that participates in substrate binding through ionic triggering.
It has been shown that the portion of human granulocyte ribonuclease that binds GroEL consists of 9 hydrophobic residues, 4 positively charged arginines and no negatively charged amino acids (60). Similarly, binding of substrates to DnaK is determined by 4 -5 hydrophobic residues flanked by basic residues, and negatively charged residues are excluded throughout (61)(62)(63). It therefore appears that despite large differences in structure, several families of molecular chaperones have conserved a combination of electrostatic and hydrophobic effects as the most effective means to recognize, bind, and possibly exchange partially folded proteins.