Hsp31, the Escherichia coli yedU Gene Product, Is a Molecular Chaperone Whose Activity Is Inhibited by ATP at High Temperatures*

The Escherichia coli chromosome contains several un-characterized heat-inducible loci that may encode novel molecular chaperones or proteases. Here we show that the 31-kDa product of the yedU gene is an efficient homodimeric molecular chaperone that is conserved in a number of pathogenic eubacteria and fungi. Heat shock protein (Hsp) 31 relies on temperature-driven conformational changes to expose structured hydrophobic domains that are likely responsible for substrate binding. Complementing the function of refolding, remodeling, and holding chaperones, Hsp 31 preferentially interacts with early unfolding intermediates and rapidly releases them in an active form after transfer to low temperatures. Although Hsp 31 does not appear to exhibit intrinsic ATPase activity, binding of ATP at high temperatures restricts the size or availability of the substrate binding site, thereby modulating chaperone activity. The possible role of ATP in coordinating the function of the cellular complement of molecular chaperones is discussed. fifth-degree polynomial function to obtain maximum fluorescence intensities and emission wavelengths. ATPase assays were performed essentially as described (13). Briefly, 2 (cid:2) l of Hsp31 (1 mg/ml) was mixed at 23 or 45 °C with 148 (cid:2) l of assay buffer (20 m M HEPES pH 7.0, 5 m M MgCl 2 , and 50 m M KCl) and 10 (cid:2) l of ATP mixture consisting of 4.8 (cid:2) l of 2.5 m M ATP, 8 (cid:2) l of [ (cid:4) - 32 P]ATP (3,000 Ci/mmol; Amersham Biosciences), and 67.2 (cid:2) l of assay buffer. Samples (25 (cid:2) l) were collected immediately after mixing and every 10 min thereafter and subjected to trichloroacetic acid precipitation and molybdate extraction (13). In some experiments, MDH was added at a 3:1 molar excess over Hsp31. Purified GroEL was used as a positive control.

In the cellular environment, the de novo folding of short, single-domain proteins is thought to proceed rapidly and efficiently whereas that of large, multidomain proteins and slow folding polypeptides often requires the assistance of molecular chaperones. In Escherichia coli, nascent and newly synthesized chains that rely on chaperones to reach a proper conformation are engaged by either trigger factor or the DnaK-DnaJ-GrpE system (and in some cases are transferred to the GroEL-GroES team) before being released in a native form. Chaperones interact with their client proteins by binding to solvent-exposed hydrophobic stretches that would normally be buried within the substrate core. By shielding interactive surfaces that give rise to misfolded and aggregated species, chaperones promote on-pathway folding without accelerating folding rates or becoming part of the final structure (for recent reviews, see Refs. [1][2][3][4]. Another role of molecular chaperones is to repair host proteins that have misfolded as a result of temperature increase or other forms of stress. The main players in this process are the DnaK-DnaJ-GrpE and GroEL-GroES teams that actively refold partially folded proteins using conformational changes powered by ATP hydrolysis and the ATPase ClpB that disaggregates thermally aggregated proteins before transferring them to DnaK-DnaJ for refolding (1)(2)(3)(4)(5). The function of other E. coli proteins known to exhibit chaperone activity in vitro is less clear. HtpG has a minor role in de novo folding (6), whereas the ␣-crystallin type IbpA and IbpB proteins are thought to function as holding chaperones that maintain unstructured proteins on their surface until stress has abated and folding chaperones become available (7).
Because temperature increases stimulate cellular protein misfolding, most, but not all (trigger factor and SecB are notable exceptions) chaperones are heat shock proteins (Hsps), 1 the synthesis of which is transiently up-regulated after heat shock. In a genome-wide analysis of the E. coli transcriptome, Blattner and co-workers (8) identified 23 heat-inducible genes of unknown function, raising the possibility that the current inventory of chaperones is incomplete. Here we show that one of these genes, yedU, encodes a homodimeric protein that defines a new family of cytoplasmic molecular chaperone. In a process that is negatively modulated by ATP binding, Hsp31 appears to use temperature-induced exposure of structured hydrophobic domains to capture early unfolding intermediates and rapidly release them in an active form once stress has abated.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-The yedU open reading frame was amplified by colony PCR of E. coli MC4100 by using the primer set 5Ј-CAATAA-GGAATACCATATGACTGTTCA-3Ј and 5Ј-CGGCCTCGAGTGATTAT-GCGCTTACATTCA-3Ј. The amplified product was cloned into pCR2.1 TOPO (Invitrogen), and its identity to GenBank TM accession number g1788278 (8) was confirmed by sequencing. The NdeI-XhoI fragment was subcloned into the same sites of pET-22b(ϩ) (Novagen, Madison, WI) to place yedU under T7 transcriptional control. The resulting plasmid was named pKV111.
Hsp31 Purification-BL21(DE3) transformants grown at 37°C in LB medium supplemented with 100 g/ml carbenicillin were induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at A 600 Ϸ0.4. After 3 h at 37°C, cells were harvested by centrifugation at 2,000 ϫ g, resuspended in 50 mM Tris, pH 7.4, and 100 mM NaCl and lysed by two cycles of French pressing at 10,000 p.s.i. Lysates were clarified by centrifugation at 10,000 ϫ g and glycerol added to a final concentration of 10% (v/v). Samples were diluted 2-fold with 25 mM Tris, pH 7.5, and 50 ml of material was applied to a 15.2 ϫ 1.1-cm column packed with 14.4 ml of Q-Sepharose Fast Flow (Amersham Biosciences). The column was developed at 2.4 ml/min with a 1 M NaCl gradient. Hsp31-rich fractions eluting between 125 and 250 mM NaCl were diluted 2-fold with 5 mM sodium phosphate and loaded onto a 1.6 ϫ 21.9-cm column packed with 44 ml of Macro-prep hydroxyapatite Type 1 (Bio-Rad) and equilibrated with 2.5 mM NaPO 4 , pH 7.0. The column was developed at 3.3 ml/min and Hsp31-rich flow-through fractions were recovered and concentrated (Centricon-10, Millipore, Bedford, MA). This material was applied to a 1.6 ϫ 46.8-cm column packed with 94.1 ml of Superose 12 (Amersham Biosciences). The column was developed in 25 mM Tris, pH 7.4, and 100 mM NaCl at 0.7 ml/min. Hsp31 (Ͼ98% pure) was concentrated, supplemented with 20% (v/v) glycerol, and stored at Ϫ80°C at 1 mg/ml. The molar extinction coefficient of Hsp31 at 280 nm is 3,463 M Ϫ1 cm Ϫ1 .
Aggregation Suppression Experiments-Monomer concentrations of porcine heart CS (Roche Molecular Biochemicals) and equine liver ADH (Sigma) were determined by using molar extinction coefficients of 3.59 ϫ 10 4 M Ϫ1 cm Ϫ1 (9) and 7.75 ϫ 10 4 M Ϫ1 cm Ϫ1 (10), respectively. Hsp31 at the indicated molar excess ratios (all based on monomers) and ADH at 1.6 M final concentration were added successively to 150 mM Tris, pH 7.4, prewarmed to 41.5°C. Samples (1 ml) were transferred to the thermostated cell of a Hitachi F4500 fluorescence spectrophotometer and held at 41.5°C. Right angle light scattering was recorded with excitation and emission wavelengths set at 500 nm and slit widths at 2.5 nm. BSA (Sigma) at a 6-fold molar excess over ADH was substituted to Hsp31 in control experiments. CS (0.4 M final concentration) thermal aggregation experiments were conducted as described previously except that all incubation steps were at 45°C and the buffer was 40 mM HEPES pH 7.8, 20 mM KOH, 50 mM KCl, 10 mM (NH 4 ) 2 SO 4 , and 2 mM CH 3 COOK. Experiments involving adenosine nucleotides were performed as described previously except that the buffers were further supplemented with 2 mM Mg-ATP or ADP and 5 mM MgCl 2 .
Activity Assays-For thermal reactivation experiments, ADH, CS, or porcine heart MDH (Sigma) were mixed at 0.4 M final concentration with no additive or the indicated molar excess of Hsp31 in 100 mM phosphate buffer, pH 7.5 (ADH and MDH), or 150 mM Tris, pH 8.1, and 2 mM EDTA (CS) held at 45°C. Samples (1 ml) were incubated at 45°C for 30 min, transferred to 23°C, and assayed after 30 min (see Fig. 2) or at the indicated times (see Fig. 4). In control experiments, Hsp31 was added at the end of the high temperature incubation step or 30 min after the mixture had cooled to room temperature. ADH, MDH, and CS activity was determined as described (11,12) and normalized to that of samples maintained at 23°C for 60 min with no additive. On heat treatment and in the absence of Hsp31 no CS activity could be detected, whereas 0.1-0.7% of the MDH remained active. For experiments involving adenosine nucleotides, reaction buffers were further supplemented with 5 mM MgCl 2 and either 2 mM Mg-ATP or ADP (Sigma) during the 45°C temperature incubation step. In some cases, nucleotides were added 90 min after transfer to 23°C. For chemical reactivation experiments, MDH was incubated for 30 min in 100 mM Tris, pH 8.1, and 6 M GdnHCl at 23°C. Aliquots were placed in the cap of microcentrifuge tubes and rapidly diluted 100-fold to a 0.4 M final concentration into 1 ml of phosphate buffer, pH 7.5, 5 mM MgCl 2 lacking or containing a 6-fold molar excess of Hsp31, and 2 mM Mg-ATP or ADP at 23°C.
Bis-ANS Binding and Photoincorporation-Temperature-induced exposure of hydrophobic domains was analyzed by mixing 0.8 M Hsp31 and 12 M bis-ANS (Molecular Probes, Eugene, OR) in 30 l of 150 mM Tris, pH 7.5. Samples were incubated at 23, 30, 37, or 45°C for 30 min, and the dye was photoincorporated by 30 min of irradiation at 254 nm by using a handheld UV lamp (Spectroline ENF 240C). Labeled products were fractionated at 1 ml/min on a Biosep S2000 column (Phenomenex, Torrance, CA) developed in 150 mM Tris, pH 7.4. Fluorescence intensity of the effluent was measured at 477 nm by using an online fluorescence detector (Shimadzu RF-10Axl) with excitation wavelength of 340 nm. Absorbance at 280 nm was monitored with a diode array detector. The effect of adenine nucleotides was investigated by mixing 0.8 M Hsp31 or BSA with 12 M bis-ANS in 1 ml of 150 mM Tris, pH 7.5, 50 mM KCl, and 5 mM MgCl 2 at 45°C in the presence or absence of 2 mM Mg-ATP, AMP-PNP, or ADP. Fluorescence intensities were measured on a Hitachi F4500 fluorimeter at 45°C with excitation wavelength of 340 nm and slit widths of 2.5 nm immediately after mixing and 15 and 30 min thereafter.
For the experiment of Fig. 4, 0.8 M Hsp31 was supplemented with 12 M bis-ANS or an identical amount of buffer in 100 l (final volume) of phosphate buffer, pH 7.5. Four samples were incubated at 45°C for 15 min to trigger the exposure of hydrophobic domains, and two of those were irradiated at 254 nm for 30 min at 45°C. Unirradiated controls were held at 45°C for 30 min. All samples were dialyzed against phosphate buffer, pH 7.5, at 4°C for 1 h by using a Microdialyzer system 100 (Pierce) fitted with a 12-kDa cutoff membrane and continuous buffer change. Samples were transferred to ice and used for triplicate activity assays conducted with a 3:1 molar excess ratio of Hsp31 to MDH as described above.
Other Analytical Techniques-CD spectra of Hsp31 (0.3 mg/ml in 10 mM phosphate buffer, pH 7.5) were recorded on a thermostated Aviv 62A DS spectropolarimeter by using a 1-mm pathlength cuvette. For sizing experiments 1.2 g of Hsp31 was injected on a thermostated Biosep S2000 column developed in 150 mM Tris-HCl, pH 7.4. The column was calibrated by using the low weight calibration kit from Amersham Biosciences. Dynamic light-scattering measurements were performed at 20°C with 1 mg/ml Hsp31 in 10 mM NaPO 4 , pH 7.5 by using a DynaPro99 instrument (Protein Solutions, Lakewood, NJ) illuminated at 832.8 nm with a 25-milliwatt solid state laser. Data analysis was performed by using the instrument software. Inverse Laplace transform analysis was used to find the mean and standard deviation (polydispersity) of the protein hydrodynamic radius distribution. For intrinsic tryptophan fluorescence experiments, 0.8 M Hsp31 in 1 ml of 100 mM phosphate buffer, pH 7.5, and 5 mM MgCl 2 was incubated at the indicated temperatures for 30 min. ATP, AMP-PNP, or ADP was added at 2 mM final concentration. Intrinsic fluorescence was measured in a thermostated Hitachi F4500 spectrophotometer 15 min after nucleotide addition by using an excitation wavelength of 295 nm and slit widths at 2.5 nm. Spectra were fitted with a fifth-degree polynomial function to obtain maximum fluorescence intensities and emission wavelengths. ATPase assays were performed essentially as described (13). Briefly, 2 l of Hsp31 (1 mg/ml) was mixed at 23 or 45°C with 148 l of assay buffer (20 mM HEPES pH 7.0, 5 mM MgCl 2 , and 50 mM KCl) and 10 l of ATP mixture consisting of 4.8 l of 2.5 mM ATP, 8 l of [␥-32 P]ATP (3,000 Ci/mmol; Amersham Biosciences), and 67.2 l of assay buffer. Samples (25 l) were collected immediately after mixing and every 10 min thereafter and subjected to trichloroacetic acid precipitation and molybdate extraction (13). In some experiments, MDH was added at a 3:1 molar excess over Hsp31. Purified GroEL was used as a positive control.

Purification and Secondary and Quaternary Structure of
Hsp31-In an effort to assign a function to the putative product of the heat-inducible (14) and H-NS-regulated (15) yedU gene, we amplified the yedU open reading frame and placed it under transcriptional control of the bacteriophage T7 promoter. Induced BL21(DE3) transformants accumulated large amounts of a soluble 31-kDa protein that matched the expected molecular weight of 31,194. The protein was designated Hsp31 and purified to near homogeneity by three chromatography steps (Fig. 1A). Hsp31 was found to assemble as a homodimer by size exclusion chromatography (Fig. 1B). Quaternary structure assignment was confirmed by dynamic light-scattering measurements conducted in triplicate. At 1 mg/ml, the hydrodynamic radius of Hsp31 was 3.28 Ϯ 0.10 nm and the polydispersity was ϳ10%. This corresponds to a molecular mass of 62 Ϯ 5.6 kDa that is fully consistent with a dimeric structure. No appreciable variation was found in elution position when sizing experiments were repeated between 23 and 50°C or in the presence of 100 mM dithiothreitol (data not shown), suggesting that changes in temperature do not affect Hsp31 quaternary structure and that Cys 185 and Cys 207 are unlikely to form an intermonomer disulfide bridge.
Far-UV CD analysis showed that Hsp31 contains a significant amount of ␣-helical structure (54%) and is about 25% ␤-pleated. The protein exhibited little change in secondary structure between 25 and 55°C (Fig. 1C) and appeared remarkably thermostable because the molar ellipticity at 220 nm remained constant at temperatures as high as 90°C (data not shown).
Hsp31 Is a Molecular Chaperone-Most cytoplasmic Hsps characterized to date function either as molecular chaperones or heat shock proteases. Because one of the hallmarks of molecular chaperone function is an ability to suppress the aggregation of unfolding intermediates, we tested whether Hsp31 would prevent the heat-induced misfolding of two model substrates, CS ( Fig. 2A) and ADH (Fig. 2B). In both cases, a concentration-dependent decrease in light scattering was ob-served, and Hsp31 was quite efficient at suppressing CS and ADH aggregation when provided at a 6-fold molar excess (based on monomers) over substrate proteins.
We next investigated whether Hsp31 would promote the reactivation (or prevent the inactivation) of thermally unfolded proteins following stress abatement. For these experiments, ADH, CS, or MDH was incubated at 45°C for 30 min in the presence or absence of Hsp31, and enzymatic activity was measured after 30 min of incubation at 23°C. In all cases, Hsp31 increased the recovery of active proteins relative to controls when present at a 3:1 molar excess, and the yields were further improved when the chaperone was used at a 6:1 molar ratio (Fig. 2C). Addition of Hsp31 at a 3:1 molar excess ratio immediately after the 45°C incubation step or 30 min after the solution cooled to 23°C did not increase the activity of either CS or ADH compared with chaperone-free controls (data not shown). These data suggest that Hsp31 alone is unable to promote the reactivation of preformed thermal aggregates and that it is likely to interact with unfolding intermediates. On the basis of stoichiometric considerations, Hsp31 is approximately as effective a chaperone as Hsp33 (16).
High Temperatures Promote the Exposure of Structured Hydrophobic Domains in Hsp31 That Are Likely Substrate Binding Sites-Because molecular chaperones typically associate with partially folded proteins through hydrophobic-hydrophobic interactions, we investigated the possibility that temperature modulates the degree of Hsp31 surface hydrophobicity. For these experiments, Hsp31 was incubated at four different

Hsp31, an ATP-modulated Chaperone
temperatures and photolabeled with bis-ANS, a molecule that exhibits little intrinsic fluorescence in its free form but becomes highly fluorescent when bound to solvent-exposed structured hydrophobic patches. Although high temperatures do not influence the efficiency of bis-ANS photoincorporation (17), the quantum yield of the dye exhibits significant temperature dependence. The change in Hsp31 hydrophobicity was therefore quantified by injecting the labeled products on a gel filtration column developed at 23°C and by simultaneously monitoring the fluorescence emission of the effluent at 477 nm and its absorbance at 280 nm. Bis-ANS photoincorporation led to slight alterations in the elution profile of Hsp31 at 280 nm (Fig.  3, inset). Although broadening was expected at the leading edge (bis-ANS-Hsp31 covalent complexes have a higher molecular weight than unlabeled Hsp31), the progressive change at the tail of the peak suggests that photoincorporation at high temperatures affects the conformation of a small fraction of Hsp31.
Asymmetric fluorescence peaks were observed over ϳ2 ml, and both peak area and maximum fluorescence intensity increased with the photolabeling temperature (Fig. 3). Samples prepared above 30°C exhibited shoulders at low elution volumes that were found to correspond to small amounts of high molecular weight and high fluorescence cross-linked species via SDS-PAGE analysis (data not shown). At the elution position of Hsp31 (Ϸ7.2 ml), fluorescence intensity increased 160% between 23 and 30°C, 80% between 30 and 37°C, and 60% between 37 and 45°C.
To assess the mechanistic relevance of these data, bis-ANS

FIG. 4. Bis-ANS photoincorporation inhibits Hsp31 chaperone function.
Hsp31 was pretreated (ϩ) or not (Ϫ) at 45°C for 15 min in the presence (ϩ) or absence (Ϫ) of bis-ANS and subjected (ϩ) or not (Ϫ) to UV irradiation for 30 min at the same temperature. All samples were subjected to dialysis and transferred to ice. MDH was incubated for 30 min at 45°C in the presence or absence of the various Hsp31 preparations, and enzymatic activities were assayed after 30 min at 23°C. The control was held for 60 min at 23°C without additive.

FIG. 3. High temperatures lead to the progressive exposure of structured hydrophobic domains in
Hsp31. Bis-ANS was photoincorporated into Hsp31 at 23°C (trace 3), 30°C (trace 4), 37°C (trace 5), or 45°C (trace 6). Trace 1 corresponds to Hsp31 incubated with bis-ANS at 45°C without irradiation, and trace 2 corresponds to Hsp31 alone irradiated at 45°C. Samples were fractionated at 23°C on a BioSep S2000 gel filtration column, and fluorescence emission at 477 nm was monitored online. Inset, sample absorbance at 280 nm. A.U., arbitrary units. was photoincorporated into Hsp31 by UV irradiation at 45°C for 30 min, and the free probe was removed by dialysis. The MDH activity experiment of Fig. 2 was then repeated with the chemically modified chaperone. Fig. 4 shows that pretreatment of Hsp31 at 45°C for 30 min in the presence or absence of UV irradiation had no effect on chaperone function because the recovery levels of active MDH were comparable to the control. However, activity yields were reduced by ϳ10-fold when the assay was conducted with Hsp31 covalently coupled to bis-ANS. When UV irradiation was omitted, the yields of MDH activity were slightly reduced, presumably because a small amount of bis-ANS is not removed on dialysis and remains noncovalently associated with Hsp31. We conclude that the structured hydrophobic domains exposed by Hsp31 at high temperatures and bound by bis-ANS are in close proximity and most likely correspond to the substrate binding site(s) of the chaperone. Taken together, the above results suggest that Hsp31 makes use of temperature-driven exposure of structured hydrophobic domains to capture, and possibly release, nonnative protein substrates.
ATP Inhibits Hsp31 Activity at High Temperatures-Molecular chaperones that actively refold or remodel proteins (e.g. GroEL, DnaK, and the Clp ATPases) use conformational changes driven by ATP hydrolysis to perform their function (1,2,4,5). In the case of Hsp31, we did not detect ATPase activity at 23 or 45°C using [␥-32 P]ATP with or without MDH (at a 1:3 ratio to Hsp31) in the assay mix (data not shown). Thus, Hsp31 has very weak or no ATPase activity. However, when the experiments of Fig. 2 were repeated in the presence of 2 mM ATP, the recovery of active CS or MDH was reduced by approximately half (data not shown).
To gain further information on the role of adenosine nucleotides and the nature of the (un)folding intermediates bound by Hsp31, we monitored the time course of MDH activity recovery after chemical or thermal denaturation. When the refolding of GdnHCl-unfolded MDH was initiated by rapid dilution at 23°C, approximately 10% of the original activity was spontaneously recovered after 180 min of incubation (Fig. 5A, q). Addition of Hsp31 to the refolding buffer increased reactivation yields to Ϸ20% without significantly influencing refolding rates (E) but ATP had no impact on either process (Fig. 5A, f and Ⅺ). The yield improvement in the presence of Hsp31 likely reflects the transient stabilization of MDH folding intermediates via interactions with the substrate binding site of Hsp31 (the socalled buffering effect).
When MDH was incubated for 30 min at 45°C, only traces of enzymatic activity were detected on transfer to 23°C (Fig. 5B,  q). However, if thermal inactivation was carried out in the presence of Hsp31, 20% of the original MDH activity was recovered immediately after temperature downshift and only a small activity gain was observed thereafter (Fig. 5B, E). Identical results were obtained with CS (data not shown). Addition of ATP to the MDH inactivation mixture reduced the beneficial effect of Hsp31 by 50% (Fig. 5B, Ⅺ), whereas ADP had essentially no influence on recovery yields (‚). High temperature conditioning of Hsp31 by ATP was essential to achieve inhibition of chaperone activity because addition of either ATP or ADP 90 min after transfer to 23°C did not affect the yields of enzymatic activity (Fig. 5B, f and OE).
To confirm the above results, aggregation suppression experiments were repeated in the presence of 2 mM ATP or ADP. In agreement with the activity data, addition of ATP to the reaction buffer inhibited the ability of Hsp31 to suppress the heatinduced aggregation of CS by about 50% (Fig. 6A, trace 3) and that of ADH by ϳ35% (Fig. 6B, trace 3). ADP had a smaller but discernible effect (Fig. 6, trace 4). Overall, these results suggest 1) that Hsp31 binds to and stabilizes early unfolding intermediates of MDH and CS at high temperatures and rapidly releases them in an active form upon stress abatement, and 2) that ATP inhibits the chaperone activity of Hsp31 under heat shock conditions by interfering with its ability to interact with partially folded substrate proteins.
ATP Binding Induces Conformational Changes in Hsp31 That Modulate the Exposure of Structured Hydrophobic Domains-The simplest explanation for the results of Figs. 5B and 6 is that the binding of ATP to Hsp31 at high temperatures reduces the availability of binding sites for non-native protein substrates. To test this hypothesis, Hsp31 was incubated with bis-ANS at 45°C in the absence or presence of various nucleotides, and fluorescence emission spectra were recorded at Hsp31, an ATP-modulated Chaperone various time points after excitation at 340 nm. Fig. 7A shows that whereas little difference existed between the bis-ANS emission spectra recorded in the absence or presence of ADP (Fig. 7, traces 1 and 2), ATP reduced the maximum bis-ANS fluorescence intensity by about 30% (Fig. 7, trace 4) with no appreciable time dependence (Fig. 7B). This effect was only apparent at high temperatures, and no statistically significant effect of ATP addition on bis-ANS fluorescence between 23 and 37°C was seen (data not shown). The nonhydrolyzable ATP analog AMP-PNP had an intermediate effect with an about 20% reduction in maximum fluorescence intensity (Fig. 7, trace  3). In contrast, ATP and ADP had no effect on the bis-ANS emission spectrum of the control protein BSA (Fig. 7C).
To determine whether the ATP-mediated quenching of bis-ANS fluorescence at 45°C was the result of a structural rearrangement in Hsp31, we took advantage of the presence of three tryptophan residues at positions 107, 173, and 229 to monitor Hsp31 conformation at various temperatures in the presence or absence of adenosine nucleotides. Fig. 8 shows that the max of Hsp31 tryptophans was about 9 nm lower than that of free tryptophan (reflective of the fact that they experience a hydrophobic environment) and remained essentially unchanged at up to 37°C under all experimental conditions. A small blue shift (Ϸ4 nm) that was unaffected by the presence of nucleotides was observed at 45°C and most likely corresponds to a temperature-induced, fine conformational rearrangement in Hsp31. On the other hand, both nucleotides reduced tryptophan maximum fluorescence intensity, with ATP being approximately three times more effective than ADP at all temperatures (Fig. 8, Ⅺ and ‚). Although these results should be considered with care because Hsp31 dimers contain six tryptophans, they are consistent with the idea that, at high temperatures, the binding of adenosine nucleotides (and particularly that of ATP) induces a conformational change in Hsp31. Because the relative magnitude of intrinsic fluorescence intensity quenching by ATP is comparable at 23 and 45°C but Hsp31 activity is not affected by ATP at the former temperature (Fig.  5A), ATP is likely to function by interfering with the temperature-driven opening of the substrate binding site. In summary, the binding of ATP nucleotides to Hsp31 appears to exert an antagonistic effect on the heat-induced exposure of structured hydrophobic domains by the chaperone and thereby modulates Hsp31 activity.
Hsp31 Is a Representative Member of a New Family of Molecular Chaperones-Similarity searches with the basic local alignment sequence tool (18) were conducted to determine whether Hsp31 orthologs were present in the data bases of nonredundant and unfinished microbial genomes. Fig. 9 shows that Hsp31 is highly conserved (greater than 55% identity at the amino acid levels) in several human pathogens including Vibrio cholerae, the enterohemorrhagic E. coli O157:H7, and the opportunistic bacteria Staphylococcus aureus and Pseudomonas aeruginosa. More divergent orthologs (40 -45% homology) were identified in a number of other eubacteria and fungi including Sinorhizobium melitoli, Agrobacterium tumefaciens, Enterococcus faecalis, Xylella fastidiosa, Coccidioides immitis, Schizosaccharomyces pombe, and Brucella melitensis (data not shown). Of interest was the fact that two of these weak Hsp31 homologs, E. faecalis CAC41347 and B. melitensis NP_541994, have been annotated as proteases on the basis of their homol- ogy to Pyrococcus furiosis protease I (PfpI), a member of the disparate ThiJ (DJ-1)/PfpI family. Although we cannot rule out the possibility that Hsp31 is a protease (particularly if additional cofactors are required for its function), several lines of evidence argue against it. First, the homology between Hsp31 and PfpI is quite low (15% identity) and could not be found by using direct algorithms (analysis with basic local alignment sequence tool 2 failed to produce an alignment). Second, PfpItype proteases are dimers of trimers and only form a catalytic triad in their hexameric state (19,20). In contrast, Hsp31 is a dimer. Third, we were unable to detect any substrate degradation when Hsp31 was incubated with MDH at either high or low temperatures (data not shown). On the other hand, the body of our data is consistent with a molecular chaperone function for Hsp31. DISCUSSION It is now well established that molecular chaperones play an important role in cellular protein folding by promoting the proper isomerization of newly synthesized polypeptides and the remodeling and refolding of misfolded proteins (1)(2)(3)(4)(5)7). Because these tasks become more critical when cells are exposed to heat or other forms of stress, most molecular chaperones residing in the E. coli cytoplasm are Hsps whose genes are transiently transcribed at an elevated level by the E 32 holoenzyme after temperature upshift (21). Genome-wide expression profiling has revealed that 77 genes are induced when E. coli is heat shocked at 50°C (14). Although most of these genes correspond to known members of the cytoplasmic and extracytoplasmic heat shock stimulons, 23 open reading frames of unknown function were also identified, raising the possibility that some of these may encode molecular chaperones of novel function. Bardwell and co-workers (16) capitalized on these findings by demonstrating that Hsp33, the yrfI (hslO) gene product, is a redox-activated chaperone that plays an important role in oxidative stress and by assigning a function to Hsp15 (the yrfH gene product) in the recycling of 50 S ribosomal subunits that still carry a nascent chain (22).
Here we have shown that the yedU open reading frame, now FIG. 8. Intrinsic tryptophan fluorescence of Hsp31 under different conditions. Hsp31 was incubated with no additive (E, q), 2 mM ADP (‚, OE), or 2 mM ATP (Ⅺ, f) at the indicated temperatures. Intrinsic fluorescence spectra were recorded after 15 min following excitation at 295 nm. Maximum fluorescence intensities (open symbols) and maximum emission wavelength (closed symbols) were obtained by fitting of spectra corrected for background fluorescence of buffer and nucleotides. F max /F o max represents the maximum fluorescence intensity ratios of adenine nucleotide-containing samples to that of nucleotide-free samples at the indicated temperature. ⌬ max represents the difference between the maximum emission wavelength of the samples and that of free tryptophan under the same buffer and temperature conditions. FIG. 9. Hsp31 is a representative member of a new family of molecular chaperones. Highly conserved Hsp31 orthologs identified by basic local alignment sequence tool analysis were aligned by using Clustal X (33). Entry numbers and percentages of identity and homology, respectively, to Hsp31 are: S. aureus, NP_371075, 56%, 73%; P. aeruginosa, NP_249826, 55%, 68%; E. coli O157:H7, 99%, 99%; Pseudomonas fluorescens, NC_02716, 61%, 73%; and V. cholerae, NC_002506, 61%, 77%. Identical residues are identified by stars, conservative replacements by colons, and semiconservative replacements by periods. referred to as hchA for heat-inducible chaperone, encodes an ␣-helix-rich homodimeric protein that exhibits chaperone activity in vitro. Homology searches suggest that Hsp31 belongs to a new family of cytoplasmic molecular chaperones that are conserved in eubacteria and fungi. Although the significance of the observation remains unclear, it is interesting to note that many Hsp31-containing organisms are human and plant pathogens.
Like E. coli IbpB (23) and other small Hsps (7), Hsp31 appears to rely on the temperature-driven exposure of structured hydrophobic domains to capture non-native protein substrates (Figs. 3 and 4). However, although small Hsps have been proposed to maintain partially folded proteins on their surface to await active refolding by the Hsp70 system (17,24,25), Hsp31 seems to preferentially bind and stabilize early unfolding intermediates and to rapidly release them in an active form once stress has abated (Fig. 5B). This activity would complement the function of folding chaperones (e.g. DnaK-DnaJ-GrpE and GroEL-GroES) that refold relatively unstructured intermediates, remodeling chaperones (e.g. ClpB) that disentangle early (6) and late (26) protein aggregates before transferring them to the DnaK-DnaJ-GrpE team (27,28), and holding chaperones (e.g. IbpB) that may serve as a reservoir of partially folded polypeptides awaiting refolding.
One of the most intriguing aspects of Hsp31 function is that although it does not appear to exhibit intrinsic ATPase activity, binding of ATP at 45°C induces a conformational change that reduces Hsp31 surface hydrophobicity (Fig. 7) and interferes with its ability to capture substrate proteins (Figs. 5B and 6), possibly by precluding opening of the binding site. The fact that ATP does not affect Hsp31 hydrophobicity at temperatures Ն37°C or the ability of the chaperone to promote the reactivation of chemically denatured MDH at 23°C (Fig. 5A) suggests that this regulatory mechanism preferentially operates under heat shock conditions. Some precedent exists for ATP tuning of molecular chaperone activity in non-ATPase Hsps. For instance, ATP binding induces conformational changes in human ␣-B-crystallin that enhances its chaperone function (29,30). Paradoxically, another small Hsp, tobacco Hsp18, experiences a decrease in chaperone activity in the presence of ATP (31). Hsp31 appears to behave like the latter protein.
Why would such negative modulation of activity be necessary and only involve ATP? A possible explanation is that the cell cytoplasm exhibits heterogeneous unfolding/folding microenvironments after heat shock. In locations where host protein misfolding is acute, ClpB together with the DnaK-DnaJ-GrpE and GroEL-GroES systems may be recruited at high concentrations to repair stress-damaged polypeptides. Because heat shock leads to a rapid decrease in intracellular ATP concentration in some organisms (32), and because all of these Hsps are ATPases, a local depletion of the ATP pool may follow that reduces the ability of folding and remodeling chaperones to turn over their substrates but concomitantly maximizes the capture of early unfolding intermediates by Hsp31. This activity would prevent additional overloading of the refolding systems because Hsp31 substrates would have to be handled by DnaK-DnaJ-GrpE or GroEL-GroES if allowed to undergo significant unfolding. On the other hand, in a microenvironment where protein misfolding is less severe, or if heat shock is mild, ATP may not become limiting. Under these conditions, the rapid repair of early unfolding intermediates by Hsp70/Hsp60, rather than their transient stabilization in an inactive form by Hsp31, would be physiologically more advantageous to the cell.