NFB-dependent Transcriptional Activation during Heat Shock Recovery

Heat shock induces the accumulation of misfolded proteins and results in the preferential expression of heat shock proteins, which help the cell to recover from thermal damage. Heat shock is a well known transcriptional activator of the human immunodeficiency virus type 1 long terminal repeat (LTR). We report here that mutations or deletions of the LTR κB sites impaired the LTR transcriptional activation by heat shock. Further analysis revealed that, during heat shock recovery, the NF-κB p65 and p50 subunits migrated into the nucleus of HeLa cells, bound to DNA, and induced κB-dependent reporter gene expression. This NF-κB activation did not depend on new transcriptional and/or translational events and on the pro-oxidant state generated by heat shock. It was not concomitant with IκBα phosphorylation and was not abolished by the expression of IκB kinase or IκBα dominant-negative mutants. Moreover, NF-κB activation and migration into the nucleus were not concomitant with IκBα/β or p105 degradation. However, during heat shock recovery, NF-κB was dissociated from its complexing partners, allowing its migration into the nucleus. Hence, we describe here a novel mechanism for activation of NF-κB based on the thermolability of the NF-κB·IκB complex.

The human immunodeficiency virus type 1 (HIV-1) 1 long terminal repeat (LTR) contains a complex eukaryotic promoter that regulates the transcription of the provirus (9). The progression of the disease induced by HIV-1 infection is directly correlated with the level of expression of HIV-1 RNA (10 -12). Hence, modulation of HIV-1 LTR activity is a key element in the development of the disease. The HIV-1 promoter contains binding sites for many transcription factors such as NF-B, SP1, upstream stimulatory factor, and AP1 and therefore confers on the virus the possibility of being activated or reactivated by many stimuli (13,14). Among these stimuli are cytokines, phorbol esters, tumor promoters, and protein kinase inhibitors (15), co-infection by other viruses (1), and oxidative (16) or thermal (17)(18)(19) stress. If, for oxidative stress mediated by H 2 O 2 or tumor necrosis factor-␣ (TNF-␣), the intervention of the transcription factor NF-B in the transcriptional activation of the LTR was clearly demonstrated (16), the mechanism regulating the thermal activation of this promoter is still unknown.
Heat shock generates abnormally folded proteins. As a consequence, the expression of most genes is inhibited, whereas a small set of genes (the heat shock genes) are preferentially transcribed. Heat shock genes encode heat shock or stress proteins that can protect cells from thermally induced injuries. Indeed, heat shock proteins act as molecular chaperones that help the cells to cope with aberrant protein folding and, as a consequence, help the cell to recover from thermal damage (20). Several studies suggested that a modification of the redox state homeostasis and particularly of the non-protein thiols such as glutathione could be involved in the heat stress signal transduction pathway that activates heat shock protein synthesis (21)(22)(23)(24). Indeed, hydrogen peroxide treatment can induce, in vitro and in vivo, heat shock gene transcription (25,26) by activating heat shock transcription factor-1 (27,28). Of interest, amino acid analogs, which are powerful agents that disrupt protein folding, are able to induce NF-B activation (2). This observation, together with the fact that NF-B activation is under the control of the intracellular redox state (29,30), prompted us to investigate the mechanism of NF-B activation by heat shock.
NF-B belongs to the Rel/NF-B family of transcription factors that includes many proteins conserved from Drosophila to humans. It controls a variety of physiological aspects of immune, inflammatory, viral, and stress responses (31). NF-B is a heterodimeric (p65/p50) inducible factor whose regulation is centered around nuclear-cytoplasmic shuttling. Indeed, the transcription factor is retained in a latent form in the cytoplasm of unstimulated cells by inhibitory molecules called IB subunits (IB␣, IB␤, and IB⑀ or p50 and p52 precursors called p105 and p100, respectively) (32)(33)(34)(35). IB subunits inhibit NF-B by masking its nuclear localization signal, thereby causing its cytoplasmic retention and blocking both its DNA binding and transactivation ability (36 -39). Migration of NF-B into the nucleus requires cytoplasmic NF-B⅐IB complex disruption. Most NF-B inducers such as inflammatory cytokines (e.g. TNF-␣), phorbol esters (e.g. phorbol 12-myristate 13-acetate), pathogenic agents, and oxidative stress (31) act via a common pathway based on the phosphorylation-induced degradation of IB proteins, which was first described with the best studied and major IB protein, IB␣. Stimuli induce IB␣ phosphorylation at Ser 32 and Ser 36 by the IB kinase (consisting of IKK␣, IKK␤, and IKK␥) (4, 6, 40 -44). This step triggers multi-ubiquitination at Lys 21 and Lys 22 of IB␣, which then signals IB␣ for degradation by the 26 S proteasome (45)(46)(47)(48). Following IB␣ degradation, NF-B migrates into the nucleus as an active factor and induces transcription of B-containing genes such as the IB␣ gene. Newly synthesized IB␣ enters the nucleus, removes NF-B dimers from DNA, and causes their exportin-mediated transport to the cytoplasm (49,50). To date, three exceptions to this universal pathway of NF-B activation have been reported. Activation of NF-B in response to UV radiation or amino acid analog treatment depends on IB␣ degradation, but without its prior phosphorylation (2,51,52). In contrast, anoxia stimulates IB␣ Tyr 42 phosphorylation, a phenomenon that leads to p65/p50 release from the NF-B⅐IB␣ complex (53).
We report here that heat shock induces the transcriptional activation of the HIV-1 LTR through a mechanism that is NF-B-dependent. NF-B activation that occurs during the heat shock recovery period differs from the universal pathway of NF-B activation since it does not involve any prior phosphorylation or degradation step of IB subunits. Moreover, this FIG. 1. B sites are indispensable for HIV-1 LTR transcriptional activation by heat shock. HeLa cells were transiently transfected with the wild-type (wild-type HIV LTR-Cat; black bars) or B-deleted ((⌬B)HIV LTR-Cat; gray bars) HIV-1 LTR-cat reporter construct (A) or with vectors containing the wild-type LTR (pLTR-Cat-wt; B-D, black bars), pLTR-Cat-PstI (B, gray bars), pLTR-Cat-EcoRI (C, gray bars), or pLTR-Cat-NcoI (D, gray bars). Cells were subsequently heat-shocked for 90 min at temperatures ranging from 41 to 44°C. After a 24-h recovery period at 37°C, the level of cytoplasmic CAT enzyme was quantified by the enzyme-linked immunosorbent assay as described under "Experimental Procedures." The results are presented as -fold stimulation, which was calculated as the ratio of the CAT concentration of the different samples to that of the control unstressed cells. The histograms shown are representative of four independent and identical experiments. The DNA sequences of the wild-type and mutated HIV-1 LTRs are shown in E. Mutations are shown in underlined boldface letters. phenomenon does not require new transcriptional and/or translational events and is a redox-independent process. During heat shock recovery, we observed a dissociation of NF-B from its complexing partners, allowing NF-B to migrate into the nucleus. This new mechanism of NF-B activation is characterized by the thermolability of the NF-B⅐IB complex.

EXPERIMENTAL PROCEDURES
Cell Cultures-HeLa cells were grown at 37°C in the presence of 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. For heat shock treatments, the cells were incubated in Dulbecco's modified Eagle's medium and 10% fetal calf serum supplemented with 25 mM HEPES, pH 7.4.
Reagents and Plasmids-Pyrrolidine dithiocarbamate (PDTC), Nacetyl-L-cysteine (NAC), hydrogen peroxide, and type B gelatin were from Sigma (Saint Quentin Fallavier, France). Recombinant human TNF-␣ was purchased from Pepro Tech EC Ltd. (London, United Kingdom). Anti-p50 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) is a goat polyclonal antibody (C-19) reactive with p50 and p105 proteins of human origin. Anti-p52 antibody (Santa Cruz Biotechnology) is a rabbit polyclonal antibody (K-27) specific for p52 and p100 proteins of human origins. Anti-IB␣/MAD-3 antibody (Santa Cruz Biotechnology) is a rabbit polyclonal antibody raised against a full-length recombinant protein of human IB␣. Rabbit anti-p65 polyclonal antibody (Upstate Biotechnology, Inc., Lake Placid, NY) is directed against the 13 Cterminal amino acids of the p65 subunit of human NF-B. The wild-type HIV LTR-Cat plasmid was a kind gift from Dr. F. Arenzana-Seisdedos (Pasteur Institute, Paris, France). It is composed of a 719-base pair XhoI-HindIII fragment containing the HIV-1 LTR (from the pBenn-Cat plasmid (1)) in front of the cat gene of pIBI20 (International Biotechnologies Inc.). The (⌬B)HIV LTR-Cat plasmid was obtained by deleting a 26-base pair fragment containing the two B sites of the wild-type plasmid. The pLTR-Cat-wt and pLTR-PstI plasmids were described elsewhere (2). pLTR-Cat-EcoRI is identical to pLTR-Cat-wt, except that the two B consensus sequences were mutated into two perfect palindromic B sites. These mutants were produced by megaprimer polymerase chain reaction mutagenesis (3). The pLTR-Cat-NcoI plasmid is a mutation of the pLTR-Cat-wt plasmid in which the B sites and the three SP1 sites of the LTR were spaced out by the insertion of a 4-nucleotide sequence (CCAT). The p2xB-37TKcat vector, which is composed of two B sites in front of a cat reporter gene, has already been described (2). pN-FLAG-CHUK(K44A) is a pRK7-S/N vector containing the full-length open reading frame of human IKK␣ cDNA with an alanine substitution of the conserved lysine residue at position 44. The expression of pN-FLAG-CHUK(K44A) leads to a dominant-negative mutant of IKK␣ kinase (4). The pRK5-IKK␤(K44A)-C-FLAG plasmid is the pRK5-C-FLAG vector containing the IKK␤ cDNA encoding amino acids 1-755 with an alanine substitution of the conserved lysine residue at position 44. pRK5-IKK␤(K44A)-C-FLAG encodes a dominant-negative mutant of IKK␤ kinase (4 -6). pLXSN and pLXSN-IB␣M were described elsewhere (7). The expression of pLXSN-IB␣M leads to a dominant-negative mutant of IB␣ in which the N-terminal (Ser 32 and Ser 36 ) and C-terminal (Ser 283 , Ser 288 , Ser 291 , Ser 293 , and Ser 296 ) serine phosphorylation sites are mutated to alanines (7).
Transfection and CAT Assays-HeLa cells were seeded the day before transfection at a density of 2 ϫ 10 6 cells/100-mm dishes. The cells were then transfected with 11 g of the desired plasmids according to the LipofectAMINE TM reagent procedure (Life Technologies, Inc., Cergy Pontoise, France). The lipid⅐DNA complex was applied to the FIG. 2. Heat shock treatment induces p65 migration into the nucleus and binding to B sites. A, HeLa cells were either left untreated or submitted to heat shock treatments at 43°C for 45 min, for 1.5 h, or for 1.5 h followed by a 1-5-h recovery period at 37°C. The cells were then fixed and processed for indirect immunofluorescence analysis using an antibody raised against the p65 subunit of NF-B. B, HeLa cells were either left untreated (control (C)) or submitted to heat shock treatments at 43°C for 90 min (90) or for 90 min followed by a recovery period at 37°C of 1 h (R1), 3 h (R3), or 5 h (R5). These various treatments were performed in the absence of any drug (black bars) or in the presence of 0.5 g/ml actinomycin D (dark-gray bars) or 20 g/ml cycloheximide (light-gray bars) added 5 min before the heat shock treatment. Whole cell extracts were prepared, and NF-B binding to B oligonucleotide was quantified with the Trans-AM TM p65 transcription factor assay kit as described under "Experimental Procedures." The results are presented as -fold stimulation, which was calculated as the ratio of the absorbance of the different samples to that of the control unstressed cells. Specificity of binding was assessed by competition with free oligonucleotide. Twenty pmol of wild-type oligonucleotide prevented the binding of NF-B from the R5 extract to the probe immobilized on the plate (R5 ϩ B). Conversely, 20 pmol of mutated consensus oligonucleotide had no effect on the binding of NF-B from the R5 extract (R5 ϩ mut B). The histogram shown is representative of two identical and independent experiments. cultured cells over 4 h. Two h later, cells were trypsinized and replated onto five or six 60-mm dishes. After 12 h, the cells were submitted to various heat, hydrogen peroxide, or TNF-␣ treatments and harvested after a 24-h recovery period. The transfected cells were then lysed, and 50 g of total cellular proteins were analyzed using the CAT/enzymelinked immunosorbent assay test (Roche Molecular Biochemicals, Meylan, France) according to the manufacturer's instructions.
Indirect Immunofluorescence Analysis-HeLa cells were grown on glass coverslips coated with 0.1% type B gelatin. Twenty h later, the cells were submitted to various heat treatments at 43°C, followed or not by a recovery period at 37°C. Thereafter, the cells were rinsed with phosphate-buffered saline and fixed for 90 s with cold methanol. Anti-p65 antibody was diluted 1:100 in phosphate-buffered saline supplemented with 0.1% bovine serum albumin. Isothiocyanate-coupled goat anti-rabbit immunoglobulin (Organon Teknica-Cappel, Fresnes, France) was used as a second antibody. The stained cells were observed and photographed with a Zeiss Axioskop photomicroscope. Fluorescent images were recorded on Tri-X Pan film (Eastman Kodak Co.).
NF-B p65 Transcription Factor Assay-NF-B binding to B sites was assessed using the Trans-AM TM NF-B p65 transcription factor assay kit (Active Motif Europe, Rixensart, Belgium). In this assay, an oligonucleotide containing the NF-B consensus site is attached to a 96-well plate. The active form of NF-B contained in cell extracts specifically binds to this oligonucleotide and can be revealed by incubation with antibodies using enzyme-linked immunosorbent assay technology with absorbance reading. In our study, HeLa cells were submitted to various heat treatments. Thereafter, whole cell extracts were prepared, and 10 g of total cellular proteins were analyzed for p65 binding to B oligonucleotide according to the manufacturer's instructions. Note that for whole cell extract preparation, a sonication step of 20 s was added after the 10-min incubation time in lysis buffer. Specificity of the assay was monitored by competition with free wild-type B consensus oligonucleotide or mutated B consensus oligonucleotide according to the manufacturer's instructions.
Gel Electrophoresis, Immunoblotting, and Co-immunoprecipitation-Acrylamide (10%) gel electrophoresis and immunoblotting were performed as described (2). For co-immunoprecipitation, 5 ϫ 10 6 cells were resuspended in 500 l of IPP 150 buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.05% Nonidet P-40) and 20 l of cØmplete TM protease inhibitor mixture (Roche Molecular Biochemicals). The cells were lysed by repeated freezing-thawing (three times). After centrifugation for 10 min at 4000 ϫ g, the supernatant was incubated for 3 h on ice with 1 g of nonimmune or anti-IB␣ serum (for p65/IB␣ co-immunoprecipitation) or with 3 g of nonimmune or anti-p65 serum (for p65/p100 co-immunoprecipitation). The immunocomplexes were precipitated with protein A-Sepharose with constant agitation at 4°C for 1 h. Thereafter, the protein A immunocomplexes were centrifuged at 4000 ϫ g for 5 min, washed several times with cold IPP 150 buffer, and boiled in SDS sample buffer. After removal of protein A-Sepharose by centrifugation, samples were analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotted with either anti-p65 or anti-p100 antibody.
In Vivo Fluorescent Measurement of Intracellular Reactive Oxygen Species-ROS detection was performed by ethidium bromide fluorescence as already described (8). Briefly, 1.5 ϫ 10 6 cells pretreated or not with PDTC or NAC were submitted to various heat treatments. Cells were trypsinized and resuspended in phosphate-buffered saline containing 40 g/ml hydroethydine, the sodium borohydride-reduced form of ethidium bromide. Flow cytometric analysis was performed with a FACSCalibur cytometer (Becton Dickinson, Le Pont de Claix, France) using a 488-nm excitation wavelength. The emission filter was 610 nm for oxidized hydroethydine fluorescence.

RESULTS
Heat Shock Induces B-site-dependent Transcriptional Activation of the HIV-1 LTR-We previously reported that the HIV-1 LTR was transcriptionally activated in response to heat shock (19), but the mechanism regulating this thermal activation remained obscure. To unravel the molecular mechanism of this activation, we tested the heat-induced transcriptional activity of the wild-type or B-deleted HIV-1 LTR. HeLa cells were transiently transfected with the wild-type HIV LTR-Cat or (⌬B)HIV LTR-Cat plasmid and thereafter submitted to 90-min heat shock treatments performed at temperatures ranging from 41 to 44°C. In these experiments, the transfection efficiency (determined in a parallel transfection using the ␤-galactosidase gene-bearing plasmid pCMV␤) was ϳ90%. By quantifying the level of CAT polypeptide produced, we observed that up to 43°C, the different heat shock treatments stimulated the transcriptional activity of the wild-type HIV-1 LTR (Fig. 1, A-D). At 44°C, a drastic decrease in CAT polypeptide production was observed. In contrast, deletion of the two B sites of the HIV-1 LTR completely abolished HIV-1 LTR activation by heat shock (Fig. 1A). These results indicate that the HIV-1 LTR transcriptional response to a heat stress is B site-dependent.
We therefore investigated the effect of different nucleotide variations in the sequence of the two B sites present in the HIV-1 LTR. The B consensus sequence (5Ј-GGGA(N/N)YYCC-3Ј) is highly conserved. However, small nucleotide variations have been described to occur preferentially at the level of the 4 central nucleotides (54). The pLTR-Cat-PstI mutant (Fig. 1E) was constructed by modifying the most conserved 5Ј-nucleotides (GG) of the B sites with CA nucleotides. These mutations have been described to impair the transcriptional activity of the HIV-1 LTR after UV irradiation or mitogen treatment (55,56). In the pLTR-Cat-EcoRI mutant, the two B sites were transformed into palindromic B sites (Fig. 1E), a situation that is suspected to improve the activity of NF-B (54,57). These constructions were transfected into HeLa cells, which were submitted thereafter to heat shock treatments. A complete impairment of the HIV-1 LTR response to heat shock was obtained in the case of pLTR-Cat-PstI (Fig. 1B), as was obtained with the B-deleted vector (Fig. 1A), suggesting that B sites are indispensable for HIV-1 LTR activation by heat shock. The pattern of pLTR-Cat-EcoRI activation by heat shock was similar to that obtained with the wild-type LTR (Fig. 1C), except that the maximal degree of activation was increased by 2-fold in comparison with the maximal degree of activation of the wild-type LTR. Therefore, palindromic B sites seem to enhance NF-B transactivation ability. We then assessed whether cooperation between SP1 and NF-B was involved in the HIV-1 LTR response to heat shock. To this end, the pLTR-Cat-NcoI mutant was constructed, in which the B and SP1 sites were spaced out by the insertion of a 4-nucleotide sequence so that NF-B and SP1 are located on opposite sides of the DNA helix (Fig. 1E). We observed that the pLTR-Cat-NcoI response to heat shock was similar to that of the wild-type LTR (Fig. 1D). Therefore, cooperation between the B and SP1 sites is not necessary for HIV-1 LTR activation by heat shock. Taken together, these results indicate an intervention of B sites in HIV-1 LTR activation by heat shock since the different modifications of B sites we have tested modulate the HIV-1 LTR response to heat shock.
Heat Shock Activates the NF-B Transcription Factor Independently of Transcriptional and/or Translational Events during the Recovery Period after Heat Stress-Based on the results presented above, we analyzed whether heat shock was able to activate the NF-B transcription factor itself. To this end, we looked for indices of NF-B activation such as NF-B migration into the nucleus or binding to DNA, transactivation of B-dependent reporter genes, and IB␣ degradation. The nuclear redistribution of NF-B subunits was analyzed by indirect immunofluorescence. We observed that, during heat shock, HeLa cells became rounded and that the cytoplasmic distribution of the p65/RelA subunit was not modified. In contrast, after 1-5 h of heat shock recovery at 37°C, the p65 subunit of NF-B redistributed into the nucleus of the cells (Fig. 2A). The same results were obtained with an antibody raised against the p50 subunit of NF-B (data not shown). This experiment showed that heat shock induces the migration of NF-B into the nucleus, but this event seems to solely occur during the recovery period after a heat stress.
We then quantified NF-B binding to DNA in HeLa cells submitted to heat shock treatments. We observed that p65 binding to B sites was detectable during the first hour of heat shock recovery and increased up to 5 h of recovery at 37°C (Fig.  2B). Specificity of p65 binding was tested by competition with free B oligonucleotide. Wild-type oligonucleotide competed efficiently for p65 binding after 5 h of heat shock recovery (R5 ϩ B), whereas mutated B oligonucleotide had no effect (R5 ϩ mut B). Monitoring of NF-B binding to DNA was also performed in the presence of 0.5 g/ml actinomycin D or 20 g/ml cycloheximide added 5 min before heat shock treatment. These drug concentrations efficiently blocked transcription and translation, respectively (data not shown). We observed that actinomycin D and cycloheximide did not modify the kinetics of NF-B binding to DNA during heat shock recovery. This suggests that no new transcriptional and/or translational events are required during heat shock and heat shock recovery for NF-B activation.
To verify whether NF-B, once in the nucleus, was able to activate B-dependent transcription, we transfected HeLa cells with a plasmid bearing the cat reporter gene under the control FIG. 4. NF-B inhibitory subunits are not degraded during heat shock or during heat shock recovery. A, HeLa cells were either left untreated or submitted to heat shock treatments of different duration (from 10 to 90 min) at 43°C. Cells were also exposed to 43°C for 90 min and allowed to recover at 37°C for 1-5 h. B, HeLa cells were treated with 2000 units/ml TNF-␣ for 3-120 min. Equal amounts of total cellular proteins were separated by 10% SDS-polyacrylamide gel electrophoresis, and the cellular contents of IB␣, IB␤, and p105 were investigated by immunoblot analysis using antibodies specific to these proteins as described under "Experimental Procedures." of two B elements (p2xB-37TKcat). Cells were then submitted to heat shock treatments, followed by a 24-h recovery period at 37°C. A small increase in the level of CAT polypeptide driven by p2xB-37TKcat was observed in cells exposed to 41°C (Fig. 3). This increase was far more pronounced at 42 and 43°C. Indeed, at 43°C, stimulation was 3.9-fold; it was more intense than that observed in cells exposed for 1.5 h to 200 M hydrogen peroxide and represented 65% of that obtained with TNF-␣ treatment. Moreover, the overall kinetics of activation of p2xB-37TKcat was similar to that obtained for the wildtype HIV-1 LTR.
NF-B Activation during Heat Shock Recovery Occurs without Any Prior Phosphorylation and Degradation of IB Subunits-We next determined whether NF-B translocation into the nucleus after heat shock treatment was induced by IB degradation. To this end, the level of several inhibitory subunits (IB␣, IB␤, and p105) was analyzed by immunoblotting. As shown in Fig. 4A, no degradation or phosphorylation of IB␣, IB␤, or p105 could be observed during heat shock treatment performed at 43°C or during the following recovery pe-riod at 37°C. In contrast, control experiments showed that TNF-␣ treatment induced the phosphorylation and/or degradation of these inhibitory subunits (Fig. 4B). Hence, NF-B migration into the nucleus after a heat stress appears to occur independently of degradation and phosphorylation of its inhibitory subunits.
To confirm that NF-B activation during heat shock recovery is not dependent on the classical signal transduction pathway including IB phosphorylation by IKK kinase and subsequent degradation, we used dominant-negative mutants of the two kinases that constitute IKK (IKK␣ and IKK␤). Double transient transfection of the dominant-negative mutant of IKK␣ (pN-FLAG-CHUK(K44A)) or of IKK␤ (pRK5-IKK␤(K44A)) with p2xB-37TKcat was performed with HeLa cells. The cells were then submitted to heat shock treatments, and the level of CAT polypeptide synthesized was analyzed. We observed that neither the IKK␣ nor IKK␤ dominant-negative mutant modified the p2xB-37TKcat activation by heat shock (Fig. 5A). In contrast, the IKK␤ dominant-negative mutant and, to a lesser extent, the IKK␣ mutant (as was expected since IKK␣ is not Presentation of the results is as described in the legend to Fig. 1. B, the same experiment was carried out, but HeLa cells were transiently transfected with the p2xB-37TKcat vector and the pLXSN control plasmid (black bars) or the expression vector of a dominant-negative mutant of IB␣ (pLXSN-IB␣M; gray bars). C, control. absolutely required for IKK activation (44)) inhibited the stimulated expression of the B-dependent reporter gene after TNF-␣ treatment (Fig. 5A). The same observations were made when a dominant-negative mutant of IB␣ in which serines 32, 36, 283, 288, 291, 293, and 296 were mutated to alanines (IB␣M) was transfected with the p2xB-37TKcat vector (Fig.  5B). Indeed, expression of IB␣M did not modify expression of p2xB-37TKcat after a heat stress, whereas it completely abolished its response to TNF-␣ treatment. Therefore, these results show that NF-B activation during heat shock recovery follows a novel signal transduction pathway that does not require phosphorylation and degradation of IB subunits to release the NF-B transcription factor.
The Pro-oxidant State Generated by Heat Shock Is Dispensable for NF-B Activation-Most NF-B inducers have been reported to be inhibited by antioxidants such as PDTC and NAC and detoxifying enzymes such as glutathione peroxidase and catalase (29,30), implying that ROS modulate the signal transduction pathway leading to NF-B activation. Several studies have suggested the involvement of an oxidative stress during heat shock. We therefore tested whether a heat stress could increase the level of ROS and, by using antioxidant drugs such as PDTC and NAC, whether ROS are involved in NF-B activation during heat shock recovery. We performed a kinetic analysis of the in vivo level of ROS during and after heat shock treatment performed at 43°C. As shown in Fig. 6, an increase in the level of intracellular ROS was detectable already after exposing cells for 1 h at 43°C. ROS levels continued to increase up to 2 h of heat shock recovery and then decreased to reach the basal level observed in control cells. In contrast, when cells were pretreated with the antioxidant drug PDTC (Fig. 6) or NAC (data not shown), the level of ROS remained unchanged. PDTC and NAC were then used to evaluate the involvement of ROS in NF-B activation during heat shock recovery.
We first investigated whether these drugs could modulate HIV-1 LTR transcriptional activation or B-dependent gene transcription induced by heat shock. To this end, HeLa cells transiently transfected with the wild-type HIV LTR-Cat (Fig.  7A) or p2xB-37TKcat (Fig. 7B) plasmid were first incubated for 1 h with 100 M PDTC or 10 mM NAC before being exposed or not to heat shock. The level of CAT polypeptide produced was quantified, and we could observe that PDTC drastically inhibited HIV-1 LTR activation by heat shock (Fig. 7A). Moreover, this compound also completely impaired the heat-induced expression of the p2xB-37TKcat plasmid (Fig. 7B). In contrast, pretreatment with the antioxidant drug NAC did not modify the wild-type HIV LTR-Cat or p2xB-37TKcat plasmid response to heat shock. Hence, these contradictory results prevented us from making a conclusion regarding the involvement or not of a pro-oxidant state in NF-B activation during heat shock recovery.
To determine whether the PDTC effect was due to its antioxidant property in the NF-B transduction pathway or to a possible unspecific role in transcription, we analyzed the level of action of this drug and whether PDTC could alter the migration of NF-B into the nucleus during heat shock recovery. Indirect immunofluorescence analyses were performed with HeLa cells pretreated or not for 1 h with 100 M PDTC and thereafter submitted to heat shock treatments. We observed the same kinetics of NF-B nuclear migration in the presence and absence of the antioxidant drug (Fig. 7C). These results therefore suggest that PDTC can affect the overall transcription step in an unspecific way and that ROS produced during heat shock are not involved in NF-B activation after a heat stress, as was suggested by the results obtained using cells treated with NAC (Fig. 7, A and B).
Heat Shock Induces p65 and p50 Dissociation from Their Complexing Partners-To unravel the molecular mechanism responsible for NF-B migration into the nucleus, we analyzed the interactions between NF-B and several of its inhibitors during and after heat shock. To this end, cell extracts from HeLa cells submitted to various heat shock treatments followed or not by a recovery period at 37°C were used for coimmunoprecipitation studies of p65 and IB␣ or p100 subunits. As shown in Fig. 8A, p65 was efficiently immunoprecipitated by anti-IB␣ antibody in control cells. After 90 min of heat shock, decreased co-immunoprecipitation of p65 was observed. This effect was even more pronounced when the experiment was performed with extracts of cells that were allowed to recover for 1 or 3 h at 37°C after heat shock. In contrast, after a 5-h recovery period at 37°C, p65 was again slightly co-immunoprecipitated by anti-IB␣ antibody. These results suggest that the interaction between p65 and IB␣ begins to be altered after 90 min of heat shock treatment. The dissociation of the complex increased after 1-3 h of heat shock recovery at 37°C, but was reversible since co-immunoprecipitation of p65 and IB␣ was again detectable after 5 h of heat shock recovery. The same observations were obtained with p65 and p100 (Fig. 8B) and p65 and p105 (data not shown) co-immunoprecipitations. Hence, heat shock triggers NF-B activation by inducing the dissociation of NF-B from its inhibitory complexing partners during the recovery period that follows a heat stress.
To better document the NF-B⅐IB complex thermolability observed by co-immunoprecipitation assays, we monitored this complex by gel filtration analysis (Fig. 8C). We observed that, in whole cell extracts of control HeLa cells, the p65 protein was eluted in fractions of ϳ100 -250 kDa and that IB␣ was recovered in fractions of 40 -150 kDa. Hence, p65 and IB␣ were both recovered in the 100 -150-kDa fractions that could represent the NF-B⅐IB complex. In contrast, when the same experiment was performed with extracts from HeLa cells submitted to a 90-min heat shock treatment at 43°C followed by a 3-h recovery period at 37°C, the p65 subunit was detected in 66 -200-kDa fractions, whereas the IB␣ subunit was recovered solely in 40-kDa fractions. Therefore, p65 and IB␣ were no longer recovered in the same fractions, suggesting that the FIG. 7. The pro-oxidant state generated by heat shock is dispensable for NF-B activation. A and B, HeLa cells, transiently transfected with either wildtype HIV LTR-Cat or p2xB-37TKcat, respectively, were preincubated for 1 h with either 100 M PDTC (dark-gray bars) or 10 mM NAC (light-gray bars) or were left untreated (black bars). Cells were then submitted to 90-min heat shock treatments performed at 41, 42, 43, and 44°C. After a 24-h recovery period at 37°C, the level of CAT enzyme synthesized was analyzed by enzyme-linked immunosorbent assay. Presentation of the results is as described in the legend to Fig. 1. C, shown are the results for the analysis of p65 cellular localization. HeLa cells pretreated or not for 1 h with 100 M PDTC were either left at 37°C or submitted to heat shock treatments at 43°C for 1.5 h or for 1.5 h followed by a 3-h recovery period at 37°C. Thereafter, the cells were fixed and processed for indirect immunofluorescence analysis using an antibody raised against the p65 subunit of NF-B.
p65⅐IB␣ complex was disrupted. The same results were obtained when HeLa cells were transiently transfected with the IB␣ dominant-negative mutant. In this case, we observed a dissociation of the NF-B⅐IB␣M complex (data not shown). Taken together, these results demonstrate that the NF-B⅐IB complex is thermolabile, leading NF-B to migrate into the nucleus.

DISCUSSION
The HIV-1 LTR is transcriptionally activated by various cellular stresses, including heat hock. However, the mechanisms underlying this thermal activation are still obscure. Indeed, a previous study based on deletion analysis of HIV-1 LTR B sites reported a reduced activation of this promoter in heat-shocked U937 promonocytic cells (18). Another report observed an inhibition of heat shock-induced virus activation by pentoxifylline (an NF-B inhibitor) (58). On the other hand, in other studies, we (19) and others (16) did not observe any specific binding of a protein to B motifs in HeLa or Jurkat cells submitted to heat shock treatments. To answer these contradictory results, we performed studies on mutated HIV-1 LTR activation by heat shock. We constructed mutations in B sites of the LTR or modulated the distance between B and SP1 sites because cooperation between B and SP1 sites has been demonstrated in the case of HIV-1 LTR activation by mitogens (59). We observed that spacing SP1 and B sites does not modify the HIV-1 LTR response to heat shock. In contrast, B sites are indispensable for HIV-1 LTR activation by heat stress. Hence, as heat shock induces the accumulation of misfolded proteins, these results suggest that a prolonged fever or pathologies interfering with protein folding are probably inducers of the HIV-1 LTR by way of NF-B transcription factor activation.
We then analyzed NF-B activation by heat shock and observed by immunofluorescence analysis that heat shock treatment induced the migration of p65 and p50 into the nucleus. However, this effect occurred only during the recovery period after the heat stress. The same conclusions were obtained by analysis of NF-B binding to DNA. Moreover, we observed stimulation of B-dependent reporter gene transcription after heat shock (maximal activation at 43°C for 1.5 h). This stimulation was of similar intensity to that obtained after hydrogen peroxide or TNF-␣ treatment (3.9-fold with heat shock versus 3.1-fold with hydrogen peroxide and 5.8-fold with TNF-␣). Hence, heat shock is a very powerful inducer of B-dependent transcription. These results suggest that, during inflammation processes, a persistent fever could be an important event involved in NF-B recruitment. NF-B activation by heat shock is quite intriguing since several studies have reported that heat shock or hsp70 overexpression decreases subsequent NF-B activation by various stressors (60 -64). In contrast, Jaattela and co-workers (65,66) observed that overexpression of hsp70 in mouse fibrosarcoma cells did not influence NF-B activation by TNF-␣ or UV light. However, these reports studied the influence of heat shock on NF-B activation by others stressors and not the direct influence of heat shock on the NF-B transcription factor. In this regard, we observed that NF-B was activated during the recovery period after and not during the heat stress. These results are consistent with two observations: (i) the fact that heat shock transcription factor-1 activation has FIG. 8. IB␣ or p100 inhibitory subunits dissociate from p65 in HeLa cells recovering from heat shock. HeLa cells were either left untreated (control (C)) or submitted to heat shock treatments at 43°C for 45 min (45), for 90 min (90), or for 90 min with a recovery period at 37°C of 1 h (R1), 3 h (R3), or 5 h (R5). Cells were then processed for immunoprecipitation using nonimmune (A and B), anti-IB␣ (A), or anti-p65 (B) antiserum as described under "Experimental Procedures." Samples were separated by 10% SDS-polyacrylamide gel electrophoresis, and the association of p65 with IB␣ or with p100 was investigated in immunoblots probed with antibodies raised against either the p65 subunit of NF-B (A) or the p100 inhibitory subunit (B). A positive control sample for the presence of p65 (A) or p100 (B), containing total cellular proteins (Tot), was loaded on the gel. C, the gel filtration analysis of p65 and IB␣ association during heat shock recovery. HeLa cells were either left untreated (Control) or submitted to a 43°C treatment for 90 min followed by a 3-h recovery period at 37°C (43°C 1h30 ϩ 37°C 3h). Cell extracts were prepared and applied to gel filtration columns as described under "Experimental Procedures." The fractions eluted from the column were applied to a 10% SDS-polyacrylamide gel, and the fraction content of p65 and IB␣ was investigated by immunoblot analysis.
been associated with NF-B inhibition as if the concomitant activation of both transcription factors was incompatible (67)(68)(69); and (ii) the fact that heat shock pretreatment delays (up to 4 -6 h) NF-B activation by cytokines (62,63). Indeed, this delay could be explained by the fact that, as we show in this study, pretreatment by heat shock induced NF-B activation and migration into the nucleus after 1-5 h of recovery at 37°C. Hence, the transcription factor would not be available for a new induction by other stressors until it is dissociated from DNA and released from the nucleus, thus creating this delay in the NF-B response to a new induction.
Finally, our previous study (19) as well as those of others (16,64) reported the absence of NF-B binding to DNA in cells submitted to heat shock. This discrepancy can be explained by our present results. The previous studies were performed after mild or short heat treatment and, more importantly, without any recovery period after the heat stress. By quantifying NF-B binding during heat shock or during the recovery period, we demonstrated that the previous conditions are not adequate for NF-B activation. Indeed, we showed that NF-B can bind to DNA solely during the recovery period after a heat stress.
We then analyzed NF-B activation in the presence of transcription (actinomycin D) or translation (cycloheximide) inhibitors. The drug concentrations used have been tested beforehand for their ability to efficiently inhibit transcription or translation. We observed the same kinetics of NF-B binding to DNA in cells submitted to heat shock treatments in the presence or absence of these drugs. Hence, de novo transcription and/or translation events during heat shock or during heat recovery are dispensable for NF-B activation. This suggests that NF-B activation during heat shock recovery results from modification of pre-existing factors. In this respect, further studies of NF-B activation in thermotolerant cells submitted to heat stress will probably provide some clues to the sensor responsible for NF-B mobilization. Indeed, one can ask whether accumulation of misfolded protein could be a stimulus that triggers NF-B activation. In a previous study, we found that NF-B was also activated by amino acid analogs (2); these compounds are structural analogs of natural amino acids rapidly incorporated into newly synthesized polypeptides and therefore induce irreversible aberrant protein conformation. Another drug creating abnormal polypeptides in the cell (puromycin, which induces a premature release of polypeptide chains from ribosomes) can induce transcription of B-dependent genes and NF-B activation. 2 But in the case of amino acid analog treatment, the mechanism of NF-B activation is different from the one obtained with heat shock, as it involves IB␣ degradation by the 26 S proteasome. Hence, a still unknown sensor would be responsible for NF-B activation during heat shock recovery.
As most NF-B inducers seem to involve ROS in their signal transduction pathway, we measured the in vivo level of intracellular ROS during and after a heat stress. We observed a transient increase in the level of ROS produced during heat shock and during the first hours of recovery at 37°C. These results confirm earlier studies that observed that heat shock pretreatment can sensitize cells or NF-B activation to hydrogen peroxide treatments (22)(23)(24). Moreover, it was shown that hydrogen peroxide is able to activate heat shock transcription factor-1 (27,28). Other studies revealed that in Saccharomyces cerevisiae, oxidative stress is involved in heat shock-induced death (70) and that ROS are intracellular mediators of hyperthermia-induced apoptosis in human HL-60 cells (71). By using the antioxidant drugs PDTC and NAC, we were able to abolish the heat-induced increased level of ROS. Nevertheless, we obtained contradictory results concerning the impairment or not of NF-B activation by heat shock according to whether PDTC or NAC was used, as PDTC, in contrast to NAC, abolished B-dependent gene expression after heat shock. However, PDTC could not impair NF-B migration into the nucleus during the heat recovery period. Therefore, these results suggest that ROS are not involved in NF-B activation during heat shock recovery and that PDTC and NAC could differently affect transactivation of transcription after heat shock. In this respect, recent studies reported that PDTC can have biphasic effects: one conferred by its antioxidant property and the other by its metal (copper/zinc) chelator property (72). Hence, PDTC could also affect, independently of its potential antioxidant role, transcription factor activity by regulating the intracellular zinc and copper ion levels. Therefore, NF-B activation during heat shock recovery does not appear to be dependent on the pro-oxidant state generated by heat-shock.
The finding of NF-B binding to DNA and B-dependent gene transcription prompted us to determine the events responsible for NF-B activation during heat shock recovery. By using dominant-negative mutants of IKK␣ and IKK␤ kinases, we observed that NF-B activation by heat shock was not dependent on prior phosphorylation of its IB inhibitory subunits. Moreover, immunoblot assays revealed that, at the time point when NF-B migrated into the nucleus after heat shock, its IB␣ inhibitory subunit was not degraded; this remained true for even longer recovery periods. We obtained the same results when other NF-B inhibitory subunits (i.e. IB␤ and p105) were analyzed, demonstrating that NF-B activation by heat shock is independent of IB degradation. These results were confirmed by the use of the IB␣M dominant-negative mutant. IB␣M is a super-repressor that cannot be phosphorylated or degraded, but can still interact with NF-B dimers, being a very potent inhibitor of NF-B by keeping it permanently in the cytoplasm. We observed that overexpression of IB␣M did not abolish the stimulated expression of p2xB-37TKcat after a heat stress, demonstrating that IB␣ phosphorylation and degradation are not indispensable for NF-B activation by heat shock. To explain NF-B migration into the nucleus, we thus co-immunoprecipitated p65 and IB␣ during heat shock and during recovery after heat shock. We observed that p65 and IB␣ dissociated early during heat shock recovery and re-associated 5 h later. The same held true for the major complexing partners of p65 (and p50), which are p100 and p105. Hence, heat shock induces p50 and p65 dissociation from their inhibitors, allowing possible new matching of NF-B subunits, which then can migrate into the nucleus. These results were strengthened by gel filtration analysis showing that, during heat shock recovery, IB␣ and p65 were no longer eluted in the same fractions. Hence, heat treatment dissociates IB from p65/p50 dimers. Moreover, the same dissociation was observed with the dominant-negative mutant of IB␣ (IB␣M). Therefore, these results demonstrate that heat shock activates NF-B by a new mechanism that relies on p65/p50⅐IB complex thermolability. One might suggest that heat shock, known to induce unfolded or denatured proteins, could modify the conformation of NF-B inhibitory subunits and their interactions with p65 and p50 since interactions between p65/p50 and its inhibitors are weak (73). The ankyrin domains shared by IB proteins, p100, and p105 could also be responsible for any destabilization of the NF-B⅐IB complex or precursors associated with p65 or p50 after heat shock. But since this mechanism of IB heat denaturation can hardly explain why the NF-B⅐IB complex is dissociated during the heat recovery period and resists elevated temperatures during heat shock, the involvement of a chaperone-mediated dissociation is one hypothesis that will merit further investigations. In this respect, additional studies on the thermolability of the precursor proteins could be informative. p100 and p105 dissociate from p65 and p50. Whether this dissociation occurs in precursors associated with each other, in precursors involved in heterodimers (precursor and p50 or p65), or in heterotrimers (precursor and p50 and p65) (74, 75) merits further investigations and suggests a mechanism involving specific chaperone recognition of the ankyrin domains of IB proteins and precursors and the specific dissociation of these inhibitors from NF-B.
Hence, we report here a new mechanism of activation of NF-B during heat shock recovery, independent of any de novo transcriptional and/or translational events, with no IB␣ phosphorylation and degradation. This mechanism relies on the thermolability of IB subunits that dissociate from NF-B dimers. Therefore, in addition to physiological stressors (ischemia/reperfusion, liver regeneration, hemorrhagic shock), physical stress, or oxidative stress, heat shock must be considered as a stress inducer of NF-B, making this transcription factor a novel regulator of the cellular stress response. However, it is still not known whether the transcription of B-dependent genes during heat shock recovery is beneficial or not for the recovery of cellular functions injured during heat shock.