Hsp72 and Stress Kinase c-jun N-Terminal Kinase Regulate the Bid-Dependent Pathway in Tumor Necrosis Factor-Induced Apoptosis

The major inducible heat shock protein Hsp72 has been shownto protect cells from certain apoptotic stimuli. Here we investigatedthe mechanism of Hsp72-mediated protection from tumor necrosisfactor (TNF)-induced apoptosis of primary culture of IMR90 humanfibroblasts. Hsp72 temporarily blocked apoptosis in responseto TNF and permanently protected cells from heat shock. An Hsp72mutant (Hsp72 ${Delta}$ EEVD) with a deletion of the four C-terminal aminoacids, which are essential for the chaperone function, blockedTNF-induced apoptosis in a manner similar to that of normalHsp72 but did not inhibit heat shock-induced death. Therefore,the chaperone activity of Hsp72 is dispensable for suppressionof TNF-induced apoptosis but is required for protection fromheat shock. In fibroblasts derived from Bid knockout mice, similartemporal inhibition of TNF-induced apoptosis was seen. In thesecells neither normal Hsp72 nor Hsp72 ${Delta}$ EEVD conferred additionalprotection from apoptosis, suggesting that Hsp72 specificallyaffects Bid-dependent but not Bid-independent apoptotic pathways.Furthermore, both normal Hsp72 and ${Delta}$ Hsp72EEVD inhibited Bid activationand downstream events, including release of cytochrome c, activationof caspase 3, and cleavage of poly-ADP-ribose polymerase. BothHsp72 and ${Delta}$ Hsp72EEVD blocked activation of the stress kinasec-jun N-terminal kinase (JNK) by TNF, and specific inhibitionof JNK similarly temporarily blocked Bid activation and thedownstream apoptotic events. These data strongly suggest thatin TNF-induced apoptosis, Hsp72 specifically interferes withthe Bid-dependent apoptotic pathway via inhibition of JNK.

INTRODUCTION

Top

Introduction

The major inducible heat shock protein Hsp72 protects cellsfrom a variety of stressful conditions, including heat shock,ischemia, UV radiation, tumor necrosis factor (TNF), and anticancerdrugs (for reviews see references 1, 13, and 17). A number ofstudies have recently demonstrated that in the protection ofcells Hsp72 suppresses a stress-induced apoptotic program, butthe mechanism of Hsp72 action in apoptotic signal transductionis still a matter of debate. The key element of the apoptoticprogram is the efflux of cytochrome c from mitochondria to cytosol,where it subsequently forms a complex (apoptosome) with Apaf-1and caspase 9, leading to activation of the latter (3). In turn,active caspase 9 cleaves and activates the major effector caspase,caspase 3, leading to the execution of apoptosis. A number ofstudies from several labs have demonstrated that Hsp72 blocksstress-induced activation of the caspase cascade (2, 5, 12,25, 27, 30) (although it should be noted that there was a reportthat Hsp72 can also regulate apoptosis downstream of caspases[18]).

Experiments with cell lysates showed that Hsp72 may directlyprevent apoptosome formation (2, 30) and caspase 3 activation(19). Other studies, however, indicated that in vivo Hsp72 functionsat an early step, upstream of mitochondria and caspase activation.For instance, Hsp72 overexpression was shown to prevent cytochromec release after heat shock (27) or hydrogen peroxide (7), althoughthe mechanism of this effect of Hsp72 is unknown. It appearsthat Hsp72 can preserve mitochondrial integrity during activationof the apoptotic program by suppression of a signaling pathwaythat leads to mitochondrial damage (e.g., release of cytochromec). Indeed, Hsp72 can block activation of a stress kinase, c-junN-terminal kinase (JNK) (12, 25), an indispensable early elementin stress-induced apoptotic program which controls the releaseof cytochrome c (34). In fact, it has recently been demonstratedthat in murine embryonal fibroblasts derived from a JNK knockoutmouse (a mutant with deletion of the two major isoforms JNK1and JNK2), cytochrome c cannot be released from mitochondriaafter exposure to UV irradiation and some other stressful treatments(8, 35). Hsp72 and JNK also appear to play a general role inthe control of various caspase-independent types of cell death(14, 15).

Hsp72 is a molecular chaperone involved in refolding and degradationof stress-damaged proteins (see reference 9 for a review). Itbinds to denatured polypeptides via its peptide-binding domainand promotes their refolding in an ATP-dependent process (16).Deletion of the ATPase domain locks Hsp72 in a substrate-boundform which inhibits the refolding reaction (4). Whether thechaperone function of Hsp72 is required for its antiapoptoticactivity has been a controversy. Several studies demonstratedthat the protein refolding function of Hsp72 is dispensablefor suppression of JNK activation and apoptosis, since the ATPasedeletion mutant of Hsp72, CTF (C-terminal fragment), was fullyactive in protection of fibroblasts from heat- or UV-inducedcell death (19, 20, 28, 37, 39). On the other hand, a recentstudy with lymphoid cells demonstrated that the chaperone functionof Hsp72 is necessary for protection (27). In this work it wasshown that normal Hsp72 or the CTF mutant could efficientlyblock activation of JNK by heat shock, but only normal Hsp72could protect cells from heat-induced apoptosis (27). Especiallyinteresting were the data with a mutant form of Hsp72 in whichfour critical C-terminal amino acids, EEVD, that are necessaryfor the chaperone function of Hsp72 were lacking or were replacedwith AAAA (11). In contrast to normal Hsp72, this Hsp72 mutantcould not prevent cytochrome c-induced caspase 3 activationin vitro (2) and heat-induced apoptosis in vivo (27); however,it efficiently inhibited JNK activity (27). Since specific inhibitionof JNK was demonstrated to be sufficient to block heat-inducedapoptosis in several cellular models (15, 36, 41), the datawith these mutants appeared puzzling. These data could eithersuggest that Hsp72 can block apoptosis in a JNK-independentmanner or that a latent JNK-independent pathway could be activatedby heat shock when cells express Hsp72 mutants but not normalHsp72. In other words, according to the latter interpretation,when the mutant form of Hsp72 is expressed, protein damage incells exposed to heat shock is so severe that cells start todie in a JNK-independent manner. On the other hand, when normalHsp72 is expressed, JNK is inhibited and proteins become repaired,leading to suppression of apoptosis.

There were two goals of this study: (i) to elucidate whetherthe chaperone function of Hsp72 is necessary for inhibitionof the apoptotic program initiated by agents that do not damageproteins and (ii) to clarify the effect of Hsp72 on the apoptoticprogram.

MATERIALS AND METHODS

Top

Materials and Methods

Cell cultures. IMR90 human lung fibroblasts (American Type Culture Collection)that underwent 10 to 20 population doublings were grown in minimalessential medium with 15% fetal bovine serum and 2 mM glutamine.Bid^-/- and Bid^+/+ simian virus 40-transformed murine embryonalfibroblasts (MEF), kindly provided by S. Korsmeyer (Dana-FarberCancer Institute), were grown in Dulbecco's modified minimalessential medium with 10% serum. Cells were used for experimentsat 50 to 80% confluence. Experiments with TNF (plus emetine)were performed in medium containing 0.5% serum to increase apoptoticsensitivity. A combination of TNF and emetine was essentialfor all the apoptotic experiments, since neither TNF nor emetinealone caused cell toxicity for at least 24 h.

Adenoviral-based expression of Hsp72, ${Delta}$ Hsp72, and SEK (K/R). Recombinant adenovirus vectors expressing Hsp72 and an EEVDdeletion mutant of Hsp72 (Hsp72 ${Delta}$ EEVD) together with green fluorescentprotein (GFP) in a dicistronic transcription unit were describedpreviously (23, 26). This transcription unit is controlled bythe tetracycline-regulated transactivator protein tTA, whichwas expressed from a separate recombinant adenovirus (AdCMVtTA)(23). Administration of 3 x 10⁷ PFU of each virus per 35-mm-diameterdish was sufficient to infect almost 100% of cells. This wasconfirmed each time by observation under a fluorescence microscopeof a proportion of the cells expressing GFP. As a control, adenovirusexpressing GFP under the regulation of tTA was used. Adenovirusexpressing dominant-negative SEK-1 (K/R) is a kinase-inactivemutant of SEK-1 tagged with M2 FLAG epitope at its amino terminus(6). Adenoviruses were propagated in 293 cells, and high-titerstocks were obtained and purified by CsCl₂ density gradientcentrifugation.

Apoptosis quantification. Apoptosis was measured by fluorescence microscopy with Hoechst-33342(5 µM) staining. Rounded cells with condensed or fragmentednuclei were considered apoptotic. About 300 to 500 cells werecounted in each experiment, which was repeated at least twice.Apoptotic caspase activation was assayed either by immunoblottingof total cell lysates with antibodies to caspase 3 (H-277; SantaCruz, Santa Cruz, Calif.); a caspase 3 substrate, poly-ADP-ribosepolymerase (PARP) (Ab-2; Oncogene, Boston, Mass.); or caspase8 (C15; a gift from P. Krammer). Secondary antibodies conjugatedwith peroxidase were visualized with ECL substrates (Amersham,Piscataway, N.J.), and the intensity of the bands was quantifiedby densitometry.

Measurement of JNK activity. Cells were washed twice with phosphate-buffered saline (PBS)on a dish, aspirated, and lysed by being thoroughly scrapedwith a plastic scraper in 200 µl of lysis buffer per 35-mm-diameterdish (40 mM HEPES [pH 7.5], 50 mM KCl, 1% Triton X-100, 2 mMdithiothreitol, 1 mM Na₃VO₄, 50 mM ß-glycerophosphate,50 mM NaF, 5 mM EDTA, 5 mM EGTA, 1 mM phenylmethylsulfonyl fluoride,1 mM benzamedine, and 5-µg/ml concentrations of each leupeptine,pepstatine A, and aprotinin). The lysates were clarified bycentrifugation in a microcentrifuge at 12,000 x g for 5 min.Total protein concentration was measured in the supernatantsby Bio-Rad protein assay reagent, after which they were dilutedwith the lysis buffer to achieve equal protein concentrationsin all samples. All procedures were performed at 4°C. Afterseparation of samples by sodium dodecyl sulfate-polyacrylamidegel electrophoresis and transfer to nitrocellulose membranes,activities of two major isoforms of JNK (46 and 54 kDa) weremeasured with a JNK antibody specific to the activated (phosphorylated)form of JNK (Promega, Madison, Wis.).

Immunolocalization of cytochrome c. For immunostaining, cells on plastic culture dishes were fixedwith 4% formalin for 15 min at room temperature, washed withPBS, frozen in liquid nitrogen, and stored at -80°C. Thecells were thawed in PBS-0.05% Tween 20, postfixed in 2% formalinand 0.5% Triton X-100 in PBS for 5 min at room temperature,and rinsed in PBS-Tween 20 three times before being reactedwith primary antibodies to cytochrome c (BD PharMingen, SanDiego, Calif.) overnight. All antibodies were in a buffer containing2% bovine serum albumin, 0.1% acetylated bovine serum albumin,0.05% gelatin, and 0.03% histone in PBS, and goat immunoglobulinG (50 µg/ml) was added to the buffer during incubationwith the primary antibodies. After overnight incubation withprimary antibodies, cells were washed with PBS-Tween 20 threetimes and then the bound primary antibodies were detected witha AlexorFluor 594 Signal-Amplification kit (Molecular Probes,Eugene, Oreg.). Images were obtained with a Carl Zeiss microscopeAxiovert 200 M and AxioVision software.

RESULTS

Top

Results

Chaperone function of Hsp72 is dispensable for protection from TNF-induced apoptosis. Here we tested the hypothesis that the chaperone activity ofHsp72, while being essential specifically for suppression ofcell death induced by protein damage (e.g., by heat shock),is not necessary for the effect of Hsp72 on the apoptotic machineryin general. To test this possibility, we studied effects ofthe Hsp72 ${Delta}$ EEVD mutant on apoptosis of IMR90 fibroblasts causedby heat shock and TNF, a cytokine that initiates an apoptoticsignal while apparently not causing protein damage. As we reportedpreviously, heat-induced apoptosis in IMR90 cells was dependenton activation of JNK and was inhibited by Hsp72 and by survivalkinases ERK and Akt (15). Interestingly, heat-induced apoptosisin these cells, although morphologically indistinguishable fromTNF-induced apoptosis, was caspase independent, while TNF-induceddeath required caspase activation (15 and see below).

Hsp72 ${Delta}$ EEVD and normal Hsp72 were expressed in IMR90 cells underthe control of a tetracycline-regulated promoter by using adenovirusvectors described previously (15, 27). With this model, neitherHsp72 nor Hsp72 ${Delta}$ EEVD affected growth or were toxic to IMR90 cellsuntil 48 h of expression (Fig. 1A). It should be noted, however,that at the later time points Hsp72, but not Hsp72 ${Delta}$ EEVD, causedsome toxicity. To find the optimal conditions of Hsp72-mediatedprotection, cells were infected with adenovirus encoding Hsp72.After various time periods infected cells were exposed to TNF(10 ng/ml), and expressions of Hsp72 and apoptosis were measured.In this experiment the time-dependent increase of Hsp72 expressioncorrelated with the progressive inhibition of apoptosis (Fig.1B). Maximal protection was seen at 36 h of Hsp72 expression,and these conditions were used in all further experiments.

View larger version (20K):
[in this window]
[in a new window]
FIG. 1. Effects of adenovirus-mediated Hsp72 expression on cell growth and TNF-induced apoptosis. IMR90 human fibroblasts were infected with adenoviruses expressing normal Hsp72, Hsp72 ${Delta}$ EEVD, or GFP (as a control) for various time periods. Cells were exposed to TNF (10 ng/ml) in the presence of the protein synthesis inhibitor emetine (10 µM). (A) The number of live cells per square millimeter (means ± standard deviations) was counted after 24 and 48 h following infection with various adenoviruses and induction of GFP, Hsp72, or Hsp72 ${Delta}$ EEVD. (B) Dependence of inhibition of TNF-induced apoptosis on Hsp72 expression. Different levels of Hsp72 expression were achieved by infection for various time periods. Apoptosis was measured by Hoechst-33342 staining, and the levels of Hsp72 were assessed by immunoblotting with anti-Hsp72 antibody.

Cells were infected with either Hsp72 or Hsp72 ${Delta}$ EEVD adenoviruses,and after 36 h they were exposed to either heat shock (45°Cfor 90 min) or TNF (10 ng/ml). Activation of apoptosis was determinedat different time points after the challenges. The effects ofHsp72 ${Delta}$ EEVD and Hsp72 on TNF- and heat-induced apoptosis differeddramatically. In fact, both Hsp72 ${Delta}$ EEVD and Hsp72 efficientlyinhibited TNF-induced apoptosis, as judged by counting cellswith apoptotic morphology (rounded cells) (Fig. 2A) and Hoechst-33342staining (Fig. 2A and C) and by measuring cleavage of PARP,which is dependent on the activation of caspase 3 (Fig. 2B).By contrast, no protection from heat-induced cell killing wasseen with Hsp72 ${Delta}$ EEVD, while cells expressing normal Hsp72 wereresistant to heat shock (Fig. 2C). The Hsp72-AAAA mutant affectedapoptosis in a manner similar to that of Hsp72 ${Delta}$ EEVD (data notshown). Therefore, in line with a previous report (27), thechaperone function of Hsp72 appeared necessary for suppressionof apoptosis induced by heat shock (a protein-damaging stress).However, the chaperone activity was dispensable for protectionfrom stresses that do not cause protein damage (e.g., TNF).

View larger version (112K):
[in this window]
[in a new window]
FIG.2. Chaperone function of Hsp72 is dispensable for suppression of TNF-induced apoptosis but is required for protection from heat shock. (A to C) Hsp72 and Hsp72 ${Delta}$ EEVD inhibit TNF-induced apoptosis. IMR90 cells were infected with adenoviruses expressing either normal Hsp72, Hsp72 ${Delta}$ EEVD, or GFP (control) for 36 h and then were exposed to TNF and emetine for various time periods. (A) Phase-contrast microscopy (upper panel objective, 10x) and Hoechst-33342 staining (lower panel objective, 40x) of cells exposed to TNF for 8 h. Expression of Hsp72 or Hsp72 ${Delta}$ EEVD prevents the appearance of apoptotic cells. (B) Effects of Hsp72 or Hsp72 ${Delta}$ EEVD on PARP cleavage in cells exposed to TNF for 5 or 7 h. The middle panel shows a loading control (expression of tubulin). The lower panel demonstrates expression of Hsp72 or Hsp72 ${Delta}$ EEVD over the course of the experiment as measured by immunoblotting. (C) Hsp72, but not Hsp72 ${Delta}$ EEVD, can prevent heat-induced cell death. Control fibroblasts or cells expressing Hsp72 or Hsp72 ${Delta}$ EEVD were treated with TNF or heat shock (45°C for 90 min), and their deaths (means ± standard deviations) were assessed by Hoechst-33342 staining after 7 (TNF) or 24 h (heat shock). Hsp72 ${Delta}$ EEVD did not affect heat-induced death at any time point. (D) Protection against TNF-induced apoptosis by Hsp72 or Hsp72 ${Delta}$ EEVD is temporal. Apoptosis was assessed by PARP cleavage.

Hsp72 suppresses the Bid-dependent apoptotic pathway. Protection of IMR90 cells by Hsp72 or Hsp72 ${Delta}$ EEVD from TNF toxicitywas temporal. Indeed, although at 5 to 8 h after TNF additionprotection by Hsp72 or Hsp72 ${Delta}$ EEVD was very remarkable, it diminishedat later time points and disappeared after 16 h of incubation(Fig. 2D). Of note, time courses of death in cells expressingnormal Hsp72 and Hsp72 ${Delta}$ EEVD were similar (Fig. 2D). The temporalprotection of cells by Hsp72 was in sharp contrast to the protectiveeffect of caspase inhibition, since no cell killing was seenin the presence of the irreversible pan-caspase inhibitor z-VAD-fmk(20 µM), even after 16 to 24 h (data not shown; see Discussion).(Of note, 20 µM was the minimal concentration of z-VAD-fmkthat caused complete inhibition of caspase 3 in IMR90 cells,and decreasing the concentration led to progressive reductionof the effects on both caspase 3 and apoptosis.) These datasuggest that the effect of Hsp72 was not directly related toinhibition of caspases.

The temporal suppression of apoptosis by Hsp72 and Hsp72 ${Delta}$ EEVDwas strikingly similar to the effect of a knockout of Bid onTNF-induced apoptosis in mouse fibroblasts (40). Bid is a proapoptoticmember of the Bcl-2 protein family whose activation via cleavageis induced when cells are treated with TNF or FAS (21, 22).The truncated form of Bid translocates to mitochondria causingcytochrome c efflux and activation of the caspase 9/caspase3 cascade (21, 22). As reported previously, lack of Bid didnot completely prevent cell death in response to TNF or FASbut delayed apoptosis by several hours (40), indicating thata Bid-independent apoptotic pathway was activated at later timepoints and destroyed cells. Because of the similarities in timecourses of cell death, we addressed the question of whetherthe effect of Hsp72 on TNF-induced apoptosis was related tothe Bid-dependent apoptotic pathway. Accordingly, we comparedtime courses of TNF-induced apoptosis in Bid^-/- and Bid^+/+ murineembryonal fibroblasts expressing Hsp72 and Hsp72 ${Delta}$ EEVD. MEF cellswere more sensitive to expression of Hsp72 than IMR90 cells,but no inhibition of growth or toxicity was seen until at least40 h after infection. Hsp72 and Hsp72 ${Delta}$ EEVD were expressed ininfected cells for 30 h (Fig. 3A), and then cells were incubatedwith TNF for various time periods and the degree of PARP cleavagewas measured by immunoblotting with an anti-PARP antibody (Fig.3B). In line with what was previously reported (40), TNF-inducedapoptosis in Bid^-/- cells was delayed compared to that of Bid^+/+cells (Fig. 3C and D). Importantly, the time course of apoptosisin Bid^-/- cells was almost identical to that of Bid^+/+ cellsexpressing Hsp72 or Hsp72 ${Delta}$ EEVD (Fig. 3C and D). Moreover, whenHsp72 or Hsp72 ${Delta}$ EEVD was expressed in Bid^-/- cells, no additionalprotective effect was seen (Fig. 3C and D). Similar data wereobtained with Hoechst-33342 staining (data not shown). In otherwords, the protective effects of Hsp72 and knocking out of Bidwere not additive. These data strongly suggest that Hsp72 specificallysuppresses a Bid-dependent but not a Bid-independent pathwayin TNF-induced apoptosis.

View larger version (17K):
[in this window]
[in a new window]
FIG. 3. Suppression of the Bid-dependent apoptotic pathway by Hsp72 or Hsp72 ${Delta}$ EEVD. (A) Expression of Hsp72 in MEF cells. MEF cells were infected with adenovirus expressing Hsp72 for 30 h, and the levels of expression were measured by immunoblotting with anti-Hsp72 antibody. (B) Effects of Bid knockout and Hsp72 on PARP cleavage induced by TNF. (C and D) Hsp72 or Hsp72 ${Delta}$ EEVD delay TNF-induced apoptosis in wild-type (WT), but not in Bid^-/- knockout, fibroblasts. Wild-type (Bid^+/+) and knockout (Bid^-/-) MEF were infected with adenoviruses expressing Hsp72 or Hsp72 ${Delta}$ EEVD for 30 h and were exposed to TNF as for the experiment depicted in Fig. 1, and apoptosis at different time points was assessed by PARP cleavage. As a mock infection we used adenovirus expressing GFP only in tetracycline-regulated systems. The experiment was repeated four times. Data of typical experiments are presented.

Hsp72 suppresses cleavage of Bid and release of cytochrome c from mitochondria. To further investigate the mechanism of suppression of TNF-inducedapoptosis by Hsp72, we addressed the question of what step ofthe Bid-dependent pathway is inhibited by Hsp72. As shown above(Fig. 2B), both normal Hsp72 and Hsp72 ${Delta}$ EEVD inhibited cleavageof PARP, indicating that Hsp72 acts at the level of or upstreamof procaspase 3 activation. Indeed, in Hsp72-expressing IMR90cells caspase 3 could not be activated upon addition of TNF(Fig. 4A). In this experiment we studied the disappearance ofprocaspase 3 by immunoblotting with an anti-caspase 3 antibody.The inhibition of caspase 3 activation was temporal and hada time course similar to that of PARP cleavage or appearanceof cells with apoptotic morphology (data not shown).

View larger version (147K):
[in this window]
[in a new window]
FIG. 4. Hsp72 or Hsp72 ${Delta}$ EEVD block TNF-induced cleavage of Bid, translocation of cytochrome c, and activation of caspase 8 and caspase 3. IMR90 fibroblasts were infected with adenoviruses and exposed to TNF as described for Fig. 2. (A) Effects of Hsp72 and Hsp72 ${Delta}$ EEVD on cleavage of procaspase 8, Bid, and procaspase 3 were assessed by immunoblotting. (B) Effects of Hsp72 and Hsp72 ${Delta}$ EEVD on translocation of cytochrome c were assessed by immunofluorescence. Cells were double stained with anti-cytochrome c antibody and Hoechst-33342. The appearance of distribution of cytochrome c in cells is shown (magnification, x40). (C) Quantification of fractions of cells (means ± standard deviations) with released cytochrome c at 7 h after TNF addition.

Next we studied whether Hsp72 can inhibit a more upstream eventin the apoptotic pathway, the release of cytochrome c from mitochondria.The release of cytochrome c was measured by immunofluorescencewith an antibody against the native form of cytochrome c. Inunstressed IMR90 cells cytochrome c was localized to mitochondria,while after the TNF addition it redistributed to the cytosoland the degree of fluorescence increased (Fig. 4B and C). Uponexpression of Hsp72 or Hsp72 ${Delta}$ EEVD the redistribution of cytochromec to the cytosol of TNF-treated cells was significantly delayed(Fig. 4B and C). These data indicate that Hsp72 somehow regulatesthe release of cytochrome c from mitochondria.

The critical question raised at this point was whether Hsp72can directly or indirectly inhibit activation of Bid. IMR90cells expressing either normal Hsp72 or mutant Hsp72 were exposedto TNF, and Bid cleavage (indicating its activation) was studiedby immunoblotting with an anti-Bid antibody. In line with previouslyreports (21, 22), Bid cleavage (as well as the cytochrome crelease and PARP cleavage) in response to TNF was partiallydependent on caspase 8, since it was inhibited by about 50%by the specific caspase 8 inhibitor z-IETD-fmk (data not shown;see Discussion). Cleavage of Bid in IMR90 cells was initiatedwithin 5 h after the TNF addition and was almost completed by7 h (Fig. 4A). Importantly, expression of Hsp72 or Hsp72 ${Delta}$ EEVDin these cells strongly delayed Bid cleavage and activationof caspase 8 (Fig. 4A). These data indicated that Hsp72 temporallyinhibited Bid activation (probably indirectly; see Discussion),thus delaying cytochrome c release and activation of the downstreamcaspase cascade.

Hsp72 suppresses cleavage of Bid via inhibition of JNK activity. Hsp72-mediated suppression of the Bid-dependent pathway in TNF-inducedapoptosis may be related to the inhibition of JNK. In fact,it has been recently shown that JNK activity is necessary forcytochrome c release and Bid cleavage upon UV-induced apoptosisof fibroblasts (34) (although it should be emphasized that activationof Bid is probably irrelevant to UV-induced apoptosis, sincethis apoptosis was not suppressed in Bid^-/- fibroblasts [38]).Since Hsp72 prevented activation of JNK by various stressfultreatments, we suggested that Hsp72-mediated inhibition of Bidcleavage might be associated with suppression of JNK.

To address this question, we first tested whether Hsp72 ${Delta}$ EEVDand normal Hsp72 suppress TNF-induced JNK activation. Cellsinfected with adenovirus encoding Hsp72 for various time periodswere exposed to TNF (10 ng/ml), and expression of Hsp72 (priorto TNF addition), activation of JNK (1 h after TNF addition),and apoptosis (7 h after TNF addition) were measured. The progressivesuppression of TNF-induced JNK activation upon increase of Hsp72expression correlated with the progressive inhibition of apoptosis(Fig. 5A). Notably, pretreatment of cells with mild heat shockto induce Hsp72 in a more physiologically relevant way alsoled to suppression of the TNF-induced activation of JNK andapoptosis. Effects of mild heat shock pretreatment on TNF-inducedactivation of JNK and apoptosis were similar to that of Hsp72expressed via adenovirus infection for 24 h (Fig. 5A), althoughthe level of Hsp72 induced by mild heat shock was lower (Fig.5A). This difference most likely is due to induction by mildheat shock of Hsp40 and other cofactors that may potentiateHsp72-mediated protection. Importantly, as we have shown previously,the effect of mild heat shock pretreatment on activation ofJNK by TNF was absolutely dependent on induction of Hsp72, sinceit was suppressed by Hsp72 antisense RNA (15). The presentationof data shown in Fig. 5A in a graph of apoptosis versus JNKactivation shows that these two parameters closely correlatewith each other (Fig. 5B).

View larger version (19K):
[in this window]
[in a new window]
FIG. 5. Hsp72 and Hsp72DEEVD suppress TNF-induced JNK activation. (A) TNF-induced activation of JNK and apoptosis in cells expressing different amounts of Hsp72. Either cells were infected with Hsp72-expressing adenovirus for various time periods or Hsp72 was induced by priming cells with heat shock (45°C for 30 min) followed by 16 h of recovery at 37°C. Activation of JNK was measured by immunoblotting with an antibody to the phosphorylated (active) form of JNK. Apoptosis was measured by PARP cleavage. Data of a typical experiment are shown. (B) Presentation of data from panel A in a graph of apoptosis versus JNK activation. (C) Effects of Hsp72 and Hsp72 ${Delta}$ EEVD on TNF-induced activation of JNK. Activation of JNK was measured by immunoblotting with an antibody to the phosphorylated (active) form of JNK. Although various splice isoforms of both JNK1 and JNK2 are present in both 46- and 54-kDa bands, the majority of JNK1 is present in the 46-kDa band while the majority of JNK2 is present in the 54-kDa band (34).

Cells infected with either Hsp72 or Hsp72 ${Delta}$ EEVD adenoviruses for36 h were exposed to TNF, and activation of JNK was determinedat different time points after the challenge. Hsp72 ${Delta}$ EEVD, similarto normal Hsp72, markedly reduced TNF-induced activation ofJNK (Fig. 5C). Of note, neither Hsp72 nor Hsp72 ${Delta}$ EEVD inhibitedTNF-induced degradation of I- ${kappa}$ B (data not shown), indicatingthat the TNF receptor is not affected. Therefore, Hsp72 andHsp72 ${Delta}$ EEVD can suppress TNF-induced JNK activation.

To investigate whether these effects on JNK are sufficient tosuppress TNF-induced apoptosis, we tested if inhibition of JNKby an independent method can prevent Bid cleavage and deathof IMR90 cells upon exposure to TNF. To suppress JNK activation,we employed a dominant-negative mutant of SEK1/MKK4 (dnSEK1),an immediate upstream component of the JNK signaling cascade.We have used this method to study the JNK dependence of heat-inducedapoptosis in IMR90 cells (15). Cells were infected with an adenovirusencoding dnSEK1 for 72 h (which did not affect cell viabilitymeasured either by PARP cleavage or Hoechst-33342 staining),and then TNF was added and activation of JNK and the extentof apoptosis were measured. Expression of dnSEK1 markedly reducedJNK activation after TNF treatment (Fig. 6A and B). Furthermore,expression of dnSEK1 significantly delayed Bid cleavage, caspase8 and caspase 3 activation (Fig. 6C), PARP cleavage (Fig. 6C),cytochrome c efflux, and cell death measured by Hoechst-33342staining (data not shown). The time courses of all these parameterswere similar to those in cells expressing Hsp72 or ${Delta}$ Hsp72EEVD(data not shown). Of note, inhibition of p38 activity by SB203580(10 µM) did not affect TNF-induced apoptosis (data notshown). Thus, it appeared that JNK activation was required forTNF-induced activation of Bid at the initial time period. SinceHsp72 efficiently inhibited activation of JNK by TNF, thesedata indicated that in cells exposed to TNF, Hsp72 most likelyinhibited cleavage and activation of Bid by suppression of JNK(see Fig. 5 and 6). Furthermore, the chaperone activity of Hsp72was not critical for these effects.

View larger version (27K):
[in this window]
[in a new window]
FIG. 6. Suppression of TNF-induced JNK activation prevents cleavage of Bid and caspase activation. IMR90 fibroblasts were infected with dnSEK1 or control adenoviruses for 72 h and exposed to TNF as for Fig. 2. (A and B) Expression of dnSEK1 inhibits TNF-induced JNK activation. JNK activity depicted in panels A and B was measured by immunoblotting as described in the legend to Fig. 5A. Expression of SEK1 was assessed by immunoblotting with an SEK1-specific antibody. Infection of cells with dnSEK1 virus was not toxic for cells for at least 4 days. (C) Expression of dnSEK1 blocks cleavage of Bid, procaspase 8, procaspase 3, and PARP as assessed by immunoblotting.

DISCUSSION

Top

Discussion

Protection of cells from various stresses by Hsp72 was discoveredlong ago, and more recently this protection was attributed toinhibitory effects of Hsp72 on apoptosis and other types ofregulated cell death (1, 13, 17, 31). So far, the exact mechanismof the inhibitory effect of Hsp72 on the apoptotic pathway hasnot been established, although it appears that there are multiplemechanisms, including inhibition of activation of the stresskinase JNK and inhibition of apoptosome assembly (1, 13). Herewe investigated the effects of Hsp72 on cell death by usingTNF-induced apoptosis of primary human fibroblast culture IMR90as a model.

Two lines of evidence indicated that, in the case of TNF-inducedapoptosis, Hsp72 regulates cleavage and, thus, activation ofa proapoptotic Bcl-2 family member, Bid. These include (i) Hsp72inhibited activation of Bid in response to TNF and (ii) Hsp72inhibited apoptotic events downstream of Bid activation, i.e.,cytochrome c efflux from mitochondria, activation of caspase3, and cleavage of PARP. Although in fibroblasts TNF inducedboth Bid-dependent and Bid-independent apoptotic pathways, Hsp72could only inhibit the Bid-dependent pathway. Indeed, eliminatingBid by gene knockout led to a delay of apoptosis similar tothat seen in cells expressing Hsp72 (see Fig. 3). Furthermore,the effects of Bid knockout and expression of Hsp72 were notadditive, indicating that Hsp72 was unable to affect a Bid-independentpathway activated at later time points.

Most likely, Hsp72 affected Bid activation indirectly via suppressionof activation of JNK (Fig. 5 and 6). Indeed, JNK activationwas critical for activation of Bid and other downstream eventsof the apoptotic pathway induced by TNF, since inhibition ofJNK by expression of dnSEK1 temporarily blocked all these events;i.e., it had effects similar to those of expression of Hsp72.Therefore, inhibition of JNK was sufficient to cause an apoptoticdelay. Since Hsp72 efficiently suppressed JNK activation, weargue that Hsp72 affects this apoptotic pathway through inhibitionof JNK. In line with this suggestion was the fact that the timecourses of inhibition of Bid cleavage, activation of caspase8, cytochrome c efflux, and activation of caspase 3 were similarin cells expressing dnSEK1 and cells expressing Hsp72.

How can JNK, and thus Hsp72, regulate Bid cleavage? A knowneffect of TNF on Bid involves procaspase 8, which becomes associatedwith the TNF receptor and is autoactivated in response to TNFbinding and can then directly cleave Bid, thus activating themitochondrial pathway (3). These data were obtained with so-calledtype I cells, in which caspase 8 associated with FAS and TNFreceptors is activated by addition of TNF and then can directlycleave and activate caspase 3 (32). This process does not requireactivation of the mitochondrial pathway, and therefore Bid cleavagein these cells is a side effect of TNF or FAS addition. In contrast,in the majority of other cells (type II cells) the mitochondrialpathway is critical for activation of caspase 3 by TNF (32).In these cells, the major fraction of caspase 8 appears to associatewith mitochondria rather than with the TNF receptor (29). Bidcleavage and activation of the mitochondrial apoptotic pathwayin these cells may be mediated not by caspase 8 but by othercaspases or even proteases other than caspases (34), e.g., cathepsinB (10). TNF-induced cleavage of Bid, cytochrome c release, andactivation of caspase 3 in IMR90 fibroblasts was partially (by40 to 50%) inhibited by a specific inhibitor of caspase 8, suggestingthat in fibroblasts caspase 8 may directly cleave Bid. However,it is clear that other proteolytic enzymes also contribute toBid cleavage, and the nature of these proteases has yet to beidentified. Accordingly, JNK either could regulate cleavageof Bid by active caspase 8 and other proteolytic enzymes orcould regulate activation of these proteases (Fig. 7). In IMR90cells expressing dnSEK1 or Hsp72, activation of caspase 8 byTNF was delayed (Fig. 4 and 6). On the other hand, these dataare difficult to interpret, since the major fraction of caspase8 in these cells is associated with mitochondria, and it hasbeen technically impossible to specifically study the fractionof caspase 8 in these cells that is associated with the TNFreceptor. It should also be noted that JNK was reported to regulatea caspase 8-independent Bid cleavage in UV-irradiated MEF cells(34). Therefore, the exact mechanism of regulation of Bid cleavageby JNK and Hsp72 remains to be studied.

View larger version (18K):
[in this window]
[in a new window]
FIG. 7. Proposed mechanism of Hsp72-mediated inhibition of TNF-induced apoptosis.

What is the mechanism of TNF-induced Bid-independent apoptosisthat takes over at later time points? This apoptosis may beassociated with activation of caspase 3 via direct cleavageby caspase 8 (3) (or some other proteolytic enzyme) (Fig. 7).It is conceivable that in type II cells this process may takeplace, although it takes longer than it does in type I cells.In such a scenario, activation of Bid and release of cytochromec at later time points was probably due to the activation ofa caspase 3-dependent positive feedback loop (Bid cleavage bycaspase 3 was reported previously [33]). Indeed, both Bid cleavageand cytochrome c efflux in TNF-treated IMR90 cells at latertime points were blocked by a caspase 3 inhibitor, DEVD-fmk(data not shown). Importantly, this late caspase 3-dependentBid cleavage and cytochrome c release were not regulated byJNK and Hsp72, since expression of neither dnSEK1 nor Hsp72could affect these processes at later time points after TNFaddition.

Here we also demonstrated that Hsp72 ${Delta}$ EEVD, a mutant that lacksthe four C-terminal amino acids EEVD, can efficiently inhibitTNF-induced apoptosis. Effects of Hsp72 ${Delta}$ EEVD were indistinguishablefrom the effects of normal Hsp72; i.e., both blocked activationof JNK and caused similar delays in Bid cleavage, release ofcytochrome c, activation of caspase 3, cleavage of PARP, andmorphological apoptotic changes in cells. Since the Hsp72 ${Delta}$ EEVDmutant has no chaperone activity, these data indicate that thisactivity of Hsp72 is not critical for inhibition of TNF-inducedapoptosis. Furthermore, since a similar Hsp72 mutant was unableto inhibit apoptosome assembly (2), we conclude that in TNF-inducedapoptosis in IMR90 cells (i) effects of Hsp72 on apoptosomeassembly are not important and (ii) the effect of Hsp72 on JNK-dependentBid cleavage plays a major role in cell protection.

Interestingly, in line with results of a previous report (27),the Hsp72 ${Delta}$ EEVD mutant was unable to protect cells from heat-inducedapoptosis. The strong inhibition of heat-induced JNK activationby Hsp72 ${Delta}$ EEVD (27 and data not shown) indicated that in the presenceof Hsp72 ${Delta}$ EEVD a JNK-independent pathway of cell death was activated.This pathway, however, was dormant in cells that did not expressHsp72 ${Delta}$ EEVD, since inhibition of JNK without expression of Hsp72 ${Delta}$ EEVD(i.e., by dnSEK1) was sufficient to inhibit heat-induced apoptosisin IMR90 cells (15). It is possible that the activity of anendogenous constitutively expressed Hsp72 homolog, Hsc73, iscritical for resistance to heat-induced apoptosis, and Hsp72 ${Delta}$ EEVDmay have a dominant-inhibitory effect on Hsc73. Under theseconditions more protein damage may occur in heat-shocked cells,which may trigger a normally silent JNK-independent apoptoticpathway.

As described previously, inhibition of JNK activity by Hsp72acts mostly through facilitation of JNK dephosphorylation (24),although there seems to be another target of Hsp72 in a JNK-activatingkinase cascade. In contrast to normal Hsp72, Hsp72 ${Delta}$ EEVD was unableto facilitate JNK dephosphorylation (27), since the peptide-bindingdomain which is involved in this regulation (39) is not functionalin this mutant (11). However, Hsp72 ${Delta}$ EEVD inhibited the JNK-activatingkinase cascade (unpublished data), and these effects were strongerthan effects of normal Hsp72. The exact mechanisms of regulationof the JNK-activating cascade by Hsp72 and Hsp72 ${Delta}$ EEVD have yetto be resolved.

In conclusion, this work demonstrated that Hsp72 downregulatesTNF-induced apoptosis via inhibition of JNK, which in turn inhibitsactivation of Bid and subsequent events of the mitochondrialapoptotic pathway. In other types of cell death, especiallyin Bid-independent processes (e.g., UV-induced apoptosis), othermechanisms of Hsp72-mediated protection should be involved.Some of these mechanisms are related to inhibition of JNK. Inother cases, effects of Hsp72 on apoptosome assembly may becritical.

ACKNOWLEDGMENTS

This study was supported by an RO1 grant from the National Institutesof Health to M.Y.S.

We thank S. Korsmeyer for providing Bid knockout cells and P.Krammer for C15 antibody.

FOOTNOTES

* Corresponding author. Mailing address: Department of Biochemistry, Boston University School of Medicine, 715 Albany St., Boston, MA 02118. Phone: (617) 638-5971. Fax: (617) 638-5339. E-mail: sherman{at}biochem.bumc.bu.edu.

REFERENCES

Top