Resolution, Detection, and Characterization of Redox Conformers of Human HSF1*

We describe here an experimental protocol for the resolution, detection, and quantitation of the reduced and oxidized conformers of human heat shock factor 1 (hHSF1) and report on the effects in vitro and in vivo of redox-active agents on the redox status, structure, and function of hHSF1. We showed that diamide, a reagent that promotes disulfide bond formation, caused a loss of immunorecognition of the monomeric hHSF1 protein in a standard Western blot detection procedure. Modification of the Western blot procedure to include dithiothreitol in the equilibration and transfer buffers after gel electrophoresis allowed for the detection of a compact, intramolecularly disulfide cross-linked oxidized hHSF1 (ox-hHSF1) in the diamide-treated sample. The effect of diamide was blocked by pretreatment with N-ethylmaleimide and was reversed by dithiothreitol added to the sample prior to gel electrophoresis. Incubation with nitrosoglutathione at 42 °C also promoted the conversion of HSF1 to ox-HSF1; at 25 °C, however, nitrosoglutathione was by itself without effect but blocked the formation of ox-hHSF1 in the presence of diamide. The disulfide cross-linked ox-hHSF1 was monomeric and resistant to the in vitro heat-induced trimerization and activation. The possibility that ox-HSF1 may occur in oxidatively stressed cells was evaluated. Treatment of HeLa cells with 2 mm l-buthionine sulfoximine promoted the formation of ox-HSF1 and blocked the heat-induced activation of HSF DNA binding activity. Our result suggests that hHSF1 may have integrated redox chemistry of cysteine sulfhydryl into its functional responses.

We describe here an experimental protocol for the resolution, detection, and quantitation of the reduced and oxidized conformers of human heat shock factor 1 (hHSF1) and report on the effects in vitro and in vivo of redox-active agents on the redox status, structure, and function of hHSF1. We showed that diamide, a reagent that promotes disulfide bond formation, caused a loss of immunorecognition of the monomeric hHSF1 protein in a standard Western blot detection procedure. Modification of the Western blot procedure to include dithiothreitol in the equilibration and transfer buffers after gel electrophoresis allowed for the detection of a compact, intramolecularly disulfide cross-linked oxidized hHSF1 (ox-hHSF1) in the diamide-treated sample. The effect of diamide was blocked by pretreatment with N-ethylmaleimide and was reversed by dithiothreitol added to the sample prior to gel electrophoresis. Incubation with nitrosoglutathione at 42°C also promoted the conversion of HSF1 to ox-HSF1; at 25°C, however, nitrosoglutathione was by itself without effect but blocked the formation of ox-hHSF1 in the presence of diamide. The disulfide cross-linked ox-hHSF1 was monomeric and resistant to the in vitro heat-induced trimerization and activation. The possibility that ox-HSF1 may occur in oxidatively stressed cells was evaluated. Treatment of HeLa cells with 2 mM L-buthionine sulfoximine promoted the formation of ox-HSF1 and blocked the heatinduced activation of HSF DNA binding activity. Our result suggests that hHSF1 may have integrated redox chemistry of cysteine sulfhydryl into its functional responses.
Cysteine, although not among the most common amino acid residues found in proteins, has unique chemical properties that confer upon it important and distinctive roles in protein structure and function. The free thiol group (S Ϫ , thiolate anion) of cysteine is a powerful nucleophile and is the most readily oxidized and nitrosylated of amino acid side chains (1,2). The disulfide-bonded cystine, by comparison, is relatively unreactive but provides a covalent link between different regions of a protein and confers stability to a specific conformation (3,4).
These considerations gave impetus to the suggestion that the redox-dependent thiol-disulfide exchange reaction can provide an important mechanism to regulate protein structure and function (5,6).
The notion that thiol-disulfide exchange may be involved in regulating transcription factor activity has gained much recent interest and support (7). For example, activation of the prokaryotic OxyR transcription factor has been shown to be a two-step process involving first the oxidation of Cys-199 to a sulfenic acid followed by intramolecular disulfide cross-link of Cys-199 and Cys-208 (8). Importantly, OxyR can be activated independently by hydrogen peroxide or by a shift of the cellular redox state as in Escherichia coli mutants lacking components of the thioredoxin and glutaredoxin pathways (9). These observations provided the much needed information to relate changes in cellular redox status to the mechanism of regulation of OxyR and to the oxidative stress response in prokaryotes. Examples of eukaryotic transcription factor regulation in vitro through reversible redox-dependent modification of key cysteine-SH groups include AP-1 (10), Rel/B (11,12), Sp1 (13), and Pax proteins (14). Whereas in vivo evidence for a causal relationship of changes in cellular redox status and function of the candidate eukaryotic transcription factor(s) is lacking, the demonstration of redox-dependent regulation of OxyR in the living prokaryotic cells and the presence in eukaryotic cells of enzymes/proteins that catalyze thiol-disulfide exchange (e.g. thioredoxin, nucleoredoxin, and Ref-1 (15)) would argue for conservation of this regulatory mechanism in evolution.
We are interested in determining if redox may provide a mechanism of regulation of the structure and function of hHSF1, 1 the transcription factor that mediates the heat shock transcriptional response to heat and other environmental stresses (16,17). This interest stems in part from our desire to understand the age-dependent dysfunction of HSF1 in a variety of model systems, including human diploid fibroblasts in culture (18,19), and is based on the generally acknowledged importance of oxidation and oxidative damage of proteins as contributing causes of aging (20). As a first step toward this goal, we developed the necessary reagents and experimental protocols for the resolution, detection, and quantitation of redox conformers of the human heat shock factor 1, and we evaluated the effects in vitro and in vivo of various redox-active compounds on the redox status, structure, and function of hHSF1.

EXPERIMENTAL PROCEDURES
Materials-The polyclonal rabbit anti-hHSF1 antibody was prepared by immunizing a rabbit with histidine-tagged hHSF1 produced in E. coli and affinity-purified using Ni 2ϩ -nitrilotriacetic acid resin from Qiagen, Valencia, CA. The specificity of this antibody was very similar to the antibody previously provided to us (21) from the laboratory of Dr. C. Wu at the NCI, National Institutes of Health (22). The enhanced chemiluminescence Western blot detection kit was from Amersham Pharmacia Biotech. The plasmid, pJC20(HSF1) was from the laboratory of Dr. Carl Wu (23). Restriction enzymes were from New England Biolabs. Inc., Beverly, MA. Transcription in vitro was done using mMessage-mMachine T7-Transcripton kit (Ambion, Inc., Austin, TX), and translation in vitro was done using rabbit reticulocyte lysate from Promega, Inc., Madison, WI. Consensus oligonucleotides of Oct-1, TATA-1, AP-1, NF-B, and SP-1 were from Promega Inc., Madison, WI. Diamide (diazenedicarboxylic acid bis(N,N-dimethylamide)), buthionine sulfoximine (BSO), sodium nitrate, sodium nitroprusside (sodium nitroferricyanide, Na 2 Fe(CN) 5 NO), and other SH-directed reagents were from Sigma or Pierce. Other chemicals were of molecular biology grade or reagent grade.
Cell Culture and Preparation of Cell Extracts-HeLa cells were grown as monolayer cultures at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 50 units/ml penicillin plus 50 g/ml streptomycin. The growth medium was replenished every 3 days until cells reach confluency. Cells were harvested at the end of an experiment by first removing the medium, rinsing the monolayer twice with ice-cold phosphate-buffered saline (150 mM NaCl, 10 mM sodium phosphate, pH 7.4), scraping the cells off with a plastic scraper, and pelleting by centrifugation at 1,800 ϫ g for 4 min.
The primary objective of this study was to determine the effects in vitro of diamide and nitrosoglutathione on the redox status and function of HSF1. For this, the S100 cell extract of control HeLa cells containing the latent HSF1 protein was used. The method of preparation of the S100 extract was as described previously (24). Briefly, the harvested cell pellet was resuspended by light vortexing in 4ϫ packed cell volume of Buffer A (15 mM HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl 2 , 0.5 mM DTT, 0.5 mM PMSF, 1 g/ml each of leupeptin and pepstatin, 0.01 units/ml aprotinin), repelleted by centrifugation, and resuspended in 2ϫ packed cell volume of Buffer A. After a 15-min incubation on ice, the swollen cells were Dounce-homogenized with a B-type pestle (15 strokes). The cell homogenate was then centrifuged at 10,000 ϫ g for 8 min, and the pellet was saved for nuclear extract preparation described below. The supernatant was transferred to a new Eppendorf tube, mixed with 0.11ϫ volume of Buffer B (180 mM HEPES (pH 7.9), 20% glycerol, 1.0 M NaCl, 13.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF, 1 g/ml each of leupeptin and pepstatin, and 0.01 units/ml aprotinin), and centrifuged at 100,000 ϫ g for 60 min. The supernatant was carefully removed and dialyzed against 50ϫ volume of Buffer D (25 mM HEPES (pH 7.9), 22% glycerol, 100 mM KCl, 0.2 mM EDTA, 0.5 mM PMSF, and 0.5 mM DTT) for 4 h at 4°C. The S100 cell extract thus obtained was aliquoted and kept frozen at Ϫ70°C until use.
For studies of the effects in vivo of glutathione depletion on the redox status and function of HSF1, HeLa cells were incubated with 1 mM BSO, a specific inhibitor of ␥-glutamylcysteine synthetase, for 24 h at 37°C, followed by an additional 2-h incubation under control (37°C) or heat shock (42°C) condition. Cells were harvested, and whole cell extracts were prepared essentially according to methods described (24, FIG. 1. Disulfide bond formation caused hHSF1 to evade detection in a standard immuno-Western blot procedure. Aliquots of S100 extract from HeLa cells containing 20 g of protein were incubated with diamide at 25°C for 30 min. To test for reversibility of the effects of diamide, the sample in lane 8 was first incubated with 1 mM diamide at 25°C for 30 min followed by the addition of dithiothreitol to a final concentration of 5 mM and incubation for an additional 30 min. To test for the effects of DTT alone, the sample in lane 9 was incubated with 5 mM DTT at 25°C for 30 min. Samples were mixed with an equal volume of a 2ϫ non-reducing, SDS sample buffer prior to loading onto a 5.5% SDS-polyacrylamide gel. The Western transfer procedure and immuno-probing for the hHSF1 antigen were done according to the standard Western transfer procedure described in the text. In this and our previous studies on immuno-Western blot detection of hHSF1 (21) the specific HSF1 signal (indicated by braces) appeared as a cluster of bands with apparent molecular masses of ϳ85-90 kDa. Duplicate sets of protein samples (control, dithiothreitol-, and diamidetreated) were subjected to polyacrylamide gel electrophoresis. Upon completion of the electrophoresis procedure, the gel was cut in half. For one-half (Standard Western Transfer, SWT), proteins electrophoretically transferred onto a nitrocellulose membrane in buffer without dithiothreitol. For the other half (Reducing Western Transfer, RWT), the gel was equilibrated, and proteins were transferred in buffers containing dithiothreitol. The two identical sets of samples were then probed for the presence of HSF1. The three samples illustrated in the diagram are: C, control; DM, diamide-treated; DTT, dithiothreitol-treated. 25). Briefly, the freshly prepared cell pellet was placed in a liquid-N 2 bath for 10 min. It was then resuspended in 2.5ϫ pellet volume of a modified whole cell lysis Buffer C (20 mM HEPES (pH 7.9), 20% glycerol, 0.42 M NaCl, 1.5 mM MgCl 2 , 0.5 mM iodoacetamide, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 1 g/ml each of leupeptin and pepstatin, and 0.01 units/ml aprotinin) and thawed at 25°C for 10 min with mild vortexing every 2 min. The mixture was then placed on ice for 10 min with periodic pulse vortexing and then centrifuged at 4°C for 10 min at 12,000 ϫ g. The supernatant containing whole cell extract was aliquoted and stored at Ϫ70°C until use.
Preparation of Nitrosothiols and Sulfhydryl Reagents-Nitrosothiols (RSNO; orthionitrites) are unstable esters of thionitrous acid of the general structure R-S-NϭO, and the biological activity of RSNO was likely due to the heterolytic decomposition and release of the nitrosonium cation (NO ϩ ) as oppose to the homolytic decomposition and release of nitric oxide (NO ⅐ ) (26). The RSNO reagents that we have used in this study include S-nitrosoglutathione (GSNO) and S-nitrosoacetylpenicillamine (SNAP). The other NO carrier that we used was sodium nitroprusside from Sigma. GSNO and SNAP were prepared according to procedures described (27). Briefly, sodium nitrate (NaNO 2 ) was mixed with an equimolar concentration of GSH (dissolved in water) or N-acetylpenicillamine (dissolved in 25% methanol). The solutions were adjusted to pH 2.0 and incubated at 37°C for 10 min at which time a characteristic color developed as follows: GSNO gave a deep orange hue, whereas SNAP formation was indicated by a light green color. The samples were then neutralized with NaOH to pH 7.4, aliquoted, frozen, and stored as stock at Ϫ70°C to be used over the next 2 months. For controls, GSH and NAP without sodium nitrate were similarly processed. HeLa cells treated with these control reagents were used to validate the specificity of effects of the S-nitrosothiols.
In Vitro Heat Activation of the Latent HSF1-The in vitro activation of hHSF1 by heat was critically dependent on protein concentration (28,29). We routinely prepared and used S100 cell extracts with a protein concentration Ͼ4.5 g/l in order to get a robust and reliable activation of the HSF1 DNA binding activity. Aliquots of cell extracts were incubated at 42°C for 30 -60 min to activate hHSF1. Extracts incubated at 25°C served as controls.
Gel Electrophoresis and Immuno-Western Blot Detection and Quantitation of HSF1-In order that we may resolve and probe redox con-formers of HSF1 by gel electrophoresis and immuno-Western blot, it was necessary for us to exclude SH-reducing reagents, such as ␤-mercaptoethanol or dithiothreitol, from the sample buffer used for gel electrophoresis. This is an important variation from the standard practice in protein gel electophoresis. Our 2ϫ sample buffer for gel electrophoresis contained 125 mM Tris-HCl (pH 6.8), 20% glycerol, 25 g/ml bromphenol blue, and SDS in concentrations indicated below. For SDS-PAGE, the sample, separation gel, and electrophoresis buffer contained 2, 0.1, and 0.1% SDS, respectively. For native gel electrophoresis, there was no SDS in the separation gel and electrophoresis buffer, although it was necessary to include 0.02% of SDS in the sample buffer. The omission of this 0.02% SDS (0.01 of that present in the normal SDS sample buffer) caused the hHSF1 signal to appear as a smear in immuno-Western blot analysis of proteins resolved by native gel electrophoresis. Perhaps the presence of a trace amount of SDS in the sample buffer had a salutary effect in preventing protein aggregation. Occasionally, when it was necessary for us to determine the total amount of HSF1, we included in the sample buffer 10 mM DTT, and the samples were then processed for gel electrophoresis and immuno-Western blot probing of HSF1.
Aliquots of protein samples with 20 g of protein were mixed with an equal volume of the specified 2ϫ sample buffer and then loaded onto a mini-gel apparatus. Gel electrophoresis was done according to the method of Laemmli as described previously (25) using a 4% spacer gel and, unless indicated otherwise, a 5.5% separation gel. Samples were electrophoresed at 100 V for 70 min or until the tracking dye, bromphenol blue, reached the bottom of the gel.
For assessment of the stoichiometry of hHSF1, samples were incubated with glutaraldehyde (2 mM, 30 min at 25°C) to cross-link protein subunits, followed by quenching of the cross-linking reaction with the addition of 100 mM lysine. These sample was then mixed with an equal volume of a reducing (10 mM DTT) SDS sample buffer and heated at 100°C for 10 min to ensure the complete denaturation of proteins and reduction of disulfide-bonded cystines. Electrophoretic separation of the HSF1 monomer, dimer, and trimer was done using a SDS (4 -10% gradient)-acrylamide gel.
Immuno-Western blot detection of hHSF1 was done under either a standard or reducing condition. For the standard procedure, the gel after electrophoresis was equilibrated for 10 min in 50 ml of 20 mM Tris ; lane 3, sample treated first with 1 mM diamide for 30 min at 25°C followed by 5 mM dithiothreitol and incubation for another 30 min. The treated samples were mixed with an equal volume of the 2ϫ non-reducing sample buffer and then subjected to analysis by gel electrophoresis and immuno-Western blot probing for hHSF1 as specified below. A, standard Western blot detection of hHSF1 in SDS-polyacrylamide gel. Samples were mixed with an equal volume of SDS sample buffer and subjected to analysis by SDS-PAGE. All procedures and buffers used (gel electrophoresis, equilibration, and transfer of proteins from the gel to nitrocellulose membrane) were done without dithiothreitol. B, reducing Western blot detection of hHSF1 in SDS-polyacrylamide gel. Upon completion of the SDS-PAGE procedure, the gel was equilibrated in a buffer containing 5 mM DTT, and proteins were transferred in buffer containing 1 mM DTT. The membrane was then probed for hHSF1 according to methods described in the text. C, reducing Western blot detection of hHSF1 in native polyacrylamide gel. Samples were mixed with an equal volume of the 2ϫ sample buffer for native gel electrophoresis and then loaded onto a native 5.5% polyacrylamide gel. After the electrophoresis procedure, the gel was equilibrated in buffer containing 5 mM DTT; proteins were transferred in a buffer containing 1 mM DTT, and the presence of hHSF1 was probed as in B. The position of the native (and presumably reduced) hHSF1 protein is indicated by a brace. The position of the oxidized hHSF1 is indicated by *. We note that in both SDS-and native PAGE the mobility of the oxidized hHSF1 conformer is faster than the reduced hHSF1, suggesting intramolecular rather then intermolecular disulfide cross-link.
(pH 8.0), 150 mM glycine, and 20% methanol, and proteins were then transferred electrophoretically (50 mA, 60 min at 25°C) onto a nitrocellulose membrane in the same buffer. For the reducing Western transfer procedure, the equilibration and transfer buffers contained 5 and 1 mM DTT, respectively. Nonspecific protein-binding site on the nitrocellulose membrane was blocked by incubation of the membrane with 10% nonfat milk in a Tris-buffered saline/Triton X-100 buffer (10 mM Tris (pH 8.0), 150 mM NaCl, 0.1% Triton X-100). Immunodetection of the hHSF1 protein was done using a 1:2500 dilution of the anti-HSF1 antibody and 1:5000 dilution of the horseradish peroxidase-conjugated goat anti-rabbit IgG antibody. Antigen-antibody complex was detected using the ECL Western Detection Kit from Amersham Pharmacia Biotech. In this as well as our previous studies (22,30) of immuno-Western blot detection of hHSF1, the specific hHSF1 signal appeared as a cluster of bands with apparent molecular masses of ϳ85-90 kDa in SDS-PAGE.
Electrophoretic Gel Mobility Shift Assay of the DNA Binding Activity of HSF1-Electrophoretic gel mobility shift assay of HSF DNA binding activity was performed essentially as described previously (31,32). Aliquots of cell extracts containing 20 -40 g of protein were used to assay for binding to 32 P-HSE. 500 ng of poly(dI/dC) was used to quench nonspecific protein-DNA binding. The reaction was initiated by adding 0.25-1.0 ng of the HSE-oligonucleotide probe (25,000 -30,000 cpm) to the samples and incubation for 20 -30 min at 25°C. Protein-DNA complexes were resolved from the free 32 P-HSE (DNA) probe by electrophoresis in a low ionic strength 4% nondenaturing polyacrylamide gel. A glycerol-bromphenol blue/xylene cyanol dye solution was added to flanking empty wells to monitor the migration of free probes. Samples were electrophoresed at 200 V for about 45 min.

RESULTS
In the initial phase of this study, we noticed that detection of the hHSF1 monomer in HeLa S100 cell extracts using a standard immuno-Western blot detection protocol was markedly affected by cysteine SH-directed reagents. Fig. 1 illustrates this. Thus, treatment of the latent hHSF1 with diamide significantly reduced the immunodetection of hHSF1 in a standard Western blot procedure (lanes 2-7). This effect of diamide was readily and completely reversed by DTT (5 mM) added to the sample prior to gel electrophoresis (lane 8), whereas DTT by itself had no effect (lane 9). This result suggests that the loss of immunoreactivity of the diamide-treated hHSF1 was not because of irreversible degradation of the protein. This cycle of treatment with diamide and then DTT could be repeated at non-reducing SDS sample buffer and then subjected to SDS-polyacrylamide (5.5%) gel electrophoresis. The presence of the hHSF1 protein was probed using the reducing Western blot protocol described. The position on the gel of the "reduced" hHSF1 is indicated by a brace, and the "oxidized" hHSF1 is indicated by an asterisk.
FIG. 5. Temperature-dependent effects of nitrosoglutathione on the redox conformers of hHSF1; analysis by native-polyacrylamide gel electrophoresis. Aliquots of S100 cell extract from HeLa cells containing 20 g of protein were treated with reagents and in the order (1st, 2nd, and 3rd) indicated. The concentrations and/or conditions used for treatment are as follows: Con, control; diamide (DM, 1.0 mM, 10 min, 25°C); DTT (5 mM, 10 min, 25°C); GSNO (1 mM, 10 min, 25°C); GSH (1 mM, 10 min, 25°C); NO 2 (sodium nitrate, 1 mM, 10 min, 25°C); filled triangle (heat-activated, 30 min at 42°C). Samples were mixed with an equal volume of the 2ϫ non-reducing and non-denaturing sample buffer and subjected to native polyacrylamide (5.5%) gel electrophoresis. hHSF1 was probed using the reducing Western transfer protocol described. The position on the gel of the "reduced" hHSF1 conformer is indicated by an open arrow, and the ox-hHSF1 conformer is indicated by a filled arrow.
least two times with little or no change in the signal intensity of hHSF1.
Diamide is an oxidizing reagent known to promote protein disulfide cross-link (33). This being given, the result in Fig. 1 suggested to us that perhaps disulfide bond formation selectively stabilized a form of hHSF1 with a shielded antigenic core that was difficult if not impossible to detect using the anti-hHSF1 polyclonal antibody. Accordingly, we reasoned that breaking the disulfide bond(s) after the completion of gel electrophoresis may allow for the immunorecognition and detection of the oxidized hHSF1 protein. We devised a "reducing" Western blot procedure, in which the gel after electrophoresis was equilibrated in a buffer containing 5 mM DTT, and proteins were transferred from the gel to nitrocellulose membrane in buffer containing 1 mM DTT. A diagrammatic illustration of this experimental strategy is shown in Fig. 2, and the result comparing immuno-Western blot detection of hHSF1 using the standard versus the reducing Western transfer procedure is shown in Fig. 3. We showed that whereas diamide treatment caused the hHSF1 protein to evade immunodetection in a standard Western blot procedure (Fig. 3A), the equilibration and then transfer of proteins after electrophoresis in a buffer containing DTT allowed for the detection of a set of faster moving, and presumably more compact forms of ox-hHSF in both SDS-and native gel electrophoresis (Fig. 3, B and C,  respectively). The fact that the ox-hHSF1 could be resolved from and in fact had a greater electrophoretic mobility than hHSF1, particularly in native gel electrophoresis (Fig. 3C), has two important implications. First, it suggests that the reduced HSF1 protein was metastable and unable to maintain its compact conformation during electrophoresis. Second, the result argues for intramolecular rather than intermolecular disulfide cross-link. Together, the results in Figs. 1 and 3 showed that diamide promoted the formation of ox-hHSF1, an intramolecularly disulfide cross-linked hHSF1 conformer. This effect of diamide could be assessed either by the apparent loss of im-munoreactivity of hHSF1 in a standard Western blot procedure or by the appearance of a more compact form (faster mobility) of the protein in both SDS-and native polyacrylamide gel electrophoresis detectable with a "reducing" Western transfer procedure.
Studies with other cysteine-SH-directed reagents provided additional support that formation of the compact hHSF1 conformer is a result of disulfide cross-link. Fig. 4A showed that the effect of diamide was blocked by prior treatment of the hHSF1 with N-ethylmaleimide, a cysteine-SH-alkylating agent (compare lanes 3 and 6). Similarly, in Fig. 4B, we showed that preincubation of the S100 extract with GSNO, a physiologically important carrier and donor of nitric oxide (34,35), blocked the effect of diamide (compare lanes 3 and 6). NEM (Fig. 4A, lane 5) and GSNO (Fig. 4B, lane 5) were by themselves without effect.
The effects of GSNO were complex and dependent on the temperature of incubation. In Fig. 5 we showed that whereas incubation of the S100 extract with GSNO at 25°C by itself had little or no effect on the mobility of hHSF1 in native PAGE, incubation at 42°C promoted the formation of ox-hHSF1 (com- pare lanes 4 and 7), and this effect of GSNO was readily and completely reversed by DTT (compare lanes 7 and 8). As controls, we showed that incubation at 42°C without (lane 9) and with glutathione (GSH, lane 10) or sodium nitrate (NO 2 , lane 11) had no effect. Our observation of these distinct effects of GSNO on hHSF1 regulation at 25 versus 42°C as presented in Fig. 5 is consistent with the suggested importance of both additive (trans-nitrosylation) and redox (e.g. disulfide bond formation) chemistry in subserving the many and diverse biological effects of GSNO (36,37).
The hHSF1 protein can be induced to undergo monomer to trimer conversion in vitro by heat (42°C) or by treatment with Nonidet P-40 or low pH buffer (28,29). As disulfide bridges are known to increase protein stability by reducing the number of unfolded conformations and decreasing the entropic costs of folding a protein into its single native state (3,4), we decided to evaluate if reagents and conditions that promote disulfide cross-link in hHSF1 would block the global conformation change associated with the trimerization and activation of hHSF1. For the experiment shown in Figs. 6-8, aliquots of S100 cell extract were treated, as specified in the figures, with various NO carriers without or with a subsequent treatment with DTT. The samples were then heat-activated in vitro (42°C, 60 min), and the stoichiometry or DNA binding activity of hHSF1 was determined according to the methods described.
Analysis of the stoichiometry of HSF1 as shown in Fig. 6 demonstrated that whereas preincubation with glutathione had no effect on the in vitro heat-induced trimerization of hHSF1 (compare lanes 2 and 3), incubation with GSNO blocked hHSF1 trimerization (lane 5), and this was reversed by DTT added to the samples prior to the in vitro heat activation procedure (lane 6). Assessment of the DNA binding activity of hHSF1 by electrophoretic mobility shift assay (Fig. 7) showed that the activation of HSE binding activity was blocked by pretreatment of the S100 cell extract with diamide (lane and acetylpenicillamine (AP, lane 8) by themselves had no effect, as did sodium nitrate (NO 2 , lane 9). Furthermore, we showed in Fig. 8 that the effects of GSNO in blocking the in vitro activation of hHSF1 was dose-dependent, with a significant inhibition observed at 0.3 mM GSNO, and was virtually complete at 0.5 mM.
The effects of GSNO on hHSF1 appeared to be specific. In Fig. 9, we showed that incubation of the nuclear extract from control HeLa cells with S-nitrosoglutathione inhibited the FIG. 6. Nitrosoglutathione but not glutathione blocked the in vitro heat-induced trimerization of hHSF1. Aliquots of the S100 cell extract containing 20 g of protein were treated with GSH (1 mM, 10 min, 25°C) or GSNO (1 mM, 10 min, 25°C) as indicated, without or with a subsequent treatment with DTT (5 mM, 10 min, 25°C); Con, control. The samples in lanes 2-6 were then heat-activated in vitro by incubation at 42°C for 60 min. Following cross-linking of proteins with glutaraldehyde, samples were mixed with an equal volume of the 2ϫ reducing (10 mM DTT) SDS sample buffer, incubated at 25°C for 10 min, and then at 100°C for another 10 min to ensure the complete denaturation of protein and removal of all disulfide cross-links. Samples were loaded onto a SDS-polyacrylamide (4 -10%) gradient gel, and hHSF1 was probed using the reducing Western transfer protocol. The positions of the molecular weight standards (220, 118, and 84 kDa) and of the hHSF1 monomer, dimer, and trimer are as indicated.
DNA binding activity of AP-1 and NF-B, and this inhibition was reversed by DTT. This result is consistent with previous observations that the reduction of a highly conserved cysteine residue in the DNA-binding domain of AP-1 and NF-B is essential for the DNA binding activity of the proteins (10 -12). The DNA binding activity of Oct-1, TATA-1, and SP1 appeared to be unaffected under these experimental conditions.
To assess the biological validity of the redox conformers of hHSF1, we decided to evaluate if depletion of intracellular glutathione would promote the formation of ox-hHSF1 in vivo. For this, confluent cultures of HeLa cells were treated with 1 mM BSO, a specific inhibitor of ␥-glutamylcysteine synthetase (38), for 24 h at 37°C followed by an additional 2-h incubation under control and heat shock conditions. Analysis of redox conformers of hHSF1 by native gel electrophoresis in Fig. 10A showed that BSO promoted the formation of ox-hHSF1 under a heat shock condition (Fig. 10A, lane 6), and concomitantly, BSO inhibited the heat-induced activation of HSE binding activity (Fig. 10C). Determination of the total amount of hHSF1 showed that the steady state level of hHSF1 was unaffected by BSO treatment (Fig. 10B). In other experiments, we observed that treatment of HeLa cells with either diamide or GSNO also promoted the formation of ox-hHSF1. However, these reagents (particularly diamide) caused activation of HSF1 DNA binding activity under a non-heat shock condition. In previous studies, diamide has been shown to induce the synthesis of heat shock proteins, presumably by producing oxidatively modified, abnormal proteins (39,40). This suggests that the effects of many of the oxidizing reagents in cells are complex due to their pleiotropic effect on the redox status of many cellular proteins. Our analysis of the role of redox in the regulation of hHSF1 is only one of the first necessary steps in an overall effort to gain a better understanding of the biology of oxidative stress in higher eukaryotic cells. DISCUSSION In this study, we report on the method we developed to resolve, detect, and quantitate the reduced and oxidized Aliquots of S100 extract from HeLa cells containing 20 g of protein were treated with diamide (lane 3), or various concentrations of Snitrosoglutathione (lanes 4 -9). Samples in lanes 2-9 were heat-activated at 42°C for 60 min, and protein binding to 32 P-HSE was determined according to methods described. The positions on the autoradiogram of the specific HSF-HSE complex, nonspecific protein binding to 32 P-HSE (N.S.), and the free 32 P-HSE probe are as indicated. C, control.
FIG. 9. Effects of S-nitrosoglutathione and dithiothreitol on the DNA binding activity of various transcription factors. Aliquots of nuclear extract from HeLa cells containing 10 g of protein were treated with GSNO (1 mM, 10 min, 25°C) either without or with a subsequent treatment with DTT (5 mM, 10 min, 25°C). Samples were used to assay for protein binding to consensus oligonucleotides of Oct-1, TATA-1, AP-1, NF-B, and SP1. hHSF1 conformers. We showed that treatment in vitro of hHSF1 with diamide and various NO carriers as well as treatment in vivo of HeLa cells with buthionine sulfoximine promoted the formation of a compact, intramolecularly disulfide cross-linked ox-hHSF1, and this was readily and completely reversed by dithiothreitol. Importantly, this thiol-disulfide exchange has profound effects on the structure and regulation of hHSF1. Oxidation and intramolecular disulfide cross-link stabilized the compact, monomeric form of the protein (ox-hHSF1), blocked its trimerization and activation, and shielded its antigenic core from immunodetection in a standard Western blot procedure. Our result raises the possibility that, in human and perhaps other mammalian species, HSF1 may have integrated redox chemistry of cysteine sulfhydryl into its functional responses.
Our studies of the effect of S-nitrosothiols on the regulation of hHF1 (that it can block the diamide-induced disulfide crosslink of hHSF1 as well as promote the appearance of disulfidebonded hHSF1) are consistent with the suggestion that these reagents can regulate protein function by both additive and redox chemistry (36,37). Thus, S-nitrosothiols can, by a heterolytic decomposition mechanism, effect the transfer of NO ϩ to a conserved cysteine residue in proteins, whereas redoxbased activity of S-nitrosothiols may capitalize on the capacity of protein-S-NO to accelerate disulfide bond formation particularly in the presence of a vicinal thiol. Most important, as indicated by our results presented in Fig. 5, the S-nitrosothiolinduced disulfide cross-link of hHSF1 requires a higher temperature of incubation (e.g. 42°C) and is strongly favored under conditions used to activate hHSF1. Although at this time we do not completely understand the reaction mechanism of this temperature-dependent effect (41), it would seem likely that disulfide bond formation of the S-nitrosylated hHSF1 and the trimerization of hHSF1 are two competing and mutually exclusive reactions at 42°C. Our results would suggest that disulfide bonding of the S-nitrosylated hHSF1 occurs with a kinetics faster than trimerization.
Clearly, redox-active agents can and do have many and perhaps even pleiotropic effects on cells. Changes in cellular redox status is likely to affect many proteins in different subcellular compartments resulting in either a gain or loss in protein function, and the biological readout is likely to be complex involving cross-talk and integration of the individual responses. Studies on NF-B and AP-1 have provided important insights of the multiple and at times opposing effects of redoxactive reagents in their regulation. Thus, thioredoxin interferes with the signaling events involved in the activation of NF-B in the cytosol, whereas in the nucleus thioredoxin increases NF-B transcriptional activities by enhancing its ability to bind DNA (42). In fact, a highly conserved, cysteine-containing motif RXXRXRXXC has been identified in all Rel/B proteins and is essential for the DNA binding activity of the protein (11,12). Likewise, whereas reactive oxygen and nitrogen species activate AP-1 DNA binding activity in cells (43,44), reduction of a highly conserved cysteine residue in the DNA-binding domain of c-Fos and c-Jun is necessary for DNA binding (10). The effect of phorbol 12-myristate 13-acetate on the activation of AP-1 is attributable largely to the translocation of thioredoxin into the nuclear compartment and the reduction and enhancement of the DNA binding activity of AP-1 by nuclear thioredoxin (15). In the case of hHSF1, while it is clear that oxidation and intramolecular disulfide cross-link would stabilize the compact hHSF1 monomer and block its trimerization and activation, the effects of cysteine-SH oxidation in vivo are much more complex and likely involve multiple targets. For example, we observed that while diamide promoted the formation of ox-hHSF1 in HeLa cells, it also activated hHSF1, a result consistent with published observations (39,40). Furthermore, we showed in a previous study that thiol-reducing reagents inhibit the cellular heat shock response by blocking the activation, trimerization, and nuclear translocation of HSF1 (21). These considerations suggest on the one hand that the cellular signaling events of the heat shock transcriptional response includes an oxidation step and is blocked by SH-reducing reagent, and on the other hand, that the conversion of the HSF1 monomer to trimer involves a global conformation change of the protein and is blocked by intramolecular disulfide cross-link which stabilizes the monomeric conformation of HSF1. Clearly, studying and understanding the role of redox in the regulation of hHSF1 is only one of the many facets of information necessary to better understand the complex biology of oxidative stress in higher eukaryotic cells.
The observation that disulfide cross-link caused the ox-hHSF1 protein to evade immunodetection by a polyclonal antibody is unexpected but perhaps not without precedence (45). Differences in the immunoreactivity of the hHSF1 versus the disulfide cross-linked ox-hHSF1 protein toward a polyclonal antibody may have implications concerning known discrepancies of the cytosolic versus nuclear localization of the HSF1 monomer. For example, it was reported that HSF of Drosophila cultured cells exists as a cytosolic protein that translocates into the nucleus after heat shock and that there is a close correla- FIG. 10. Inhibition of glutathione synthesis in HeLa cells by buthionine sulfoximine promoted the formation of ox-hHSF1 and blocked the heat-induced activation of HSE binding activity. Confluent cultures of HeLa cells were incubated without and with 1 mM BSO at 37°C for 24 h to block ␥-glutamylcysteine synthetase and deplete intracellular glutathione, followed by 2 h of incubation under control (C) (37°C) or heat-shock (HS) (42°C) conditions. Cells were harvested and whole cell extracts prepared according to methods described in the text. A, redox conformers of hHSF1. Aliquots of the cell extract containing 20 g of protein were used for analysis of redox conformers of hHSF1 by native gel electrophoresis and immuno-Western blot probing for hHSF1 using the reducing transfer procedure. Samples in lanes 3 and 4 represent control and heat-shocked HeLa cells without BSO pretreatment, and lanes 5 and 6 were from cells pretreated with 1 mM BSO. For comparison, control (C) and diamide (DM)-treated HeLa S100 extracts were included in the experiment to (lanes 1 and 2, for control and diamide-treated). The positions on the gel of the reduced hHSF1 and ox-hHSF1 are indicated, respectively, by an open and closed arrow. B, total amount of hHSF1. To determine the steady state level of the hHSF1 protein, aliquots of the whole cells extracts corresponding to samples of lanes 3-6 of A were mixed with an SDS sample buffer containing 10 mM DTT. Samples were heated at 100°C for 10 min prior to being subjected to analysis by SDS-PAGE and immuno-Western blot probing for hHSF1. C, electrophoretic gel mobility shift assay of HSE binding activity. Aliquots of whole cell extract from control and heat-shocked HeLa cells without and with BSO pretreatment were used to assay for HSE binding activity according to the methods described. The position on the autoradiogram of the HSF-hHSE complex is as indicated. tion between the subcellular localization, oligomeric state, and DNA binding activity of the protein; drosophila HSF in cytosol is always monomeric and cannot bind DNA and that present in the nucleus is always found in a higher order structure and can bind HSE in a sequence-specific manner (46). This is to be contrasted with the reported strictly nuclear localization of the protein in Drosophila and Xenopus under both non-stressed and stressed conditions (47,48). In studies of the subcellular localization of higher mammalian HSF1, it is generally accepted that hHSF1 exists as a cytosolic protein in non-stressed cells and that upon stress the protein undergoes a multistep activation process that includes nuclear translocation (16,49). A revisit of this issue in a recent study painted a somewhat different picture (50). Thus, whereas biochemical fractionation studies showed HSF1 of several human and monkey cell lines to be distributed evenly or predominantly in the cytosol over the nuclear fraction, immunofluorescence microscopy showed that the HSF1 protein is predominantly a nuclear protein before and after heat stress (50). A review of the literature suggests that the experimental support of a predominantly nuclear localization of HSF1 comes primarily from immunocytochemical staining studies. Given the result of our present findings, it seems possible that tissue fixation, either by protein cross-linking (e.g. glutaraldehyde) or precipitation (e.g. alcohol), and the absence of an SH-reducing reagent in the immunocytochemical staining protocol may have rendered some population of the HSF1 protein (particularly the HSF monomer) undetectable or less detectable by antibodies raised against the native HSF1 protein, thus biasing toward a nuclear staining pattern of the protein. In biochemical and conventional immuno-Western blot studies, samples are routinely treated with denaturing and SH-reducing reagents with heating prior to gel electrophoresis, conditions that would erase differences in protein folding and conformation and may be better suited for the detection of all forms of HSF1. Clearly, the use of antibody generated against the native and presumably activated HSF1 protein versus the SDS-denatured HSF1 also contributed to differences in the results observed (17,51).
These considerations suggest that any detection method will be only as good as the tools are and that a detailed understanding of the limitation of the tools is critical in the proper interpretation of experimental findings. Although the issue of cytosolic versus nuclear localization of the latent HSF1 proteins remains to be resolved once and for all, our study may help to frame the experimental design in order to resolve this important issue.