Address correspondence and reprint requests to Dr. R. S. Jope at Department ofPsychiatry and Behavioral Neurobiology, Sparks Center, Room 1057, Universityof Alabama, Birmingham, AL 35294-0017, U.S.A. E-mail:firstname.lastname@example.org
Abstract: Elevated temperatures activate the survival promoters Aktand heat shock factor-1 (HSF-1), a transcription factor that induces theexpression of heat shock proteins (HSPs), such as HSP-70. Because neuronalmechanisms controlling these responses are not known, these were investigatedin human neuroblastoma SH-SY5Y cells. Heat shock (45°C) rapidly activatedAkt, extracellular signal-regulated kinases 1 and 2 (ERK1/2), and p38, butonly Akt was activated in a phosphatidylinositol 3-kinase (PI-3K)-dependentmanner, as the PI-3K inhibitors LY294002 and wortmannin blocked Aktactivation, but not ERK1/2 or p38 activation. Akt activation was not blockedby inhibition of p38 or ERK1/2, indicating the independence of these signalingsystems. Heat shock treatment also caused a rapid increase in HSF-1 DNAbinding activity that was partially dependent on PI-3K activity, as both thePI-3K inhibitors attenuated this response. Because Akt inhibits glycogensynthase kinase-3β (GSK-3β), an enzyme that facilitates cell death,we tested if GSK-3β is a negative regulator of HSF-1 activation.Overexpression of GSK-3β impaired heat shock-induced activation of HSF-1,and also reduced HSP-70 production, which was partially restored by theGSK-3β inhibitor lithium. Thus, heat shock-induced activation of PI-3Kand the inhibitory effect of GSK-3β on HSF-1 activation and HSP-70expression imply that Akt-induced inhibition of GSK-3β contributes to theactivation of HSF-1.
Mobilization of defensive signaling mechanisms in response to stressors is critical for neuronal function and survival. Inadequate responses to stress can result in neuronal death by apoptosis or necrosis, losses that likely contribute to the cognitive decline associated with aging, and to the deficits associated with many neurodegenerative diseases. One of the critical mechanisms by which cells respond to stress is known as “the heat shock response,” one of the most important mechanisms used to bolster cellular defenses to support survival in the face of potentially lethal conditions (Edwards, 1998; Jäättelä, 1999). In addition to elevated temperature, the heat shock response has been reported to be activated by such conditions as oxidative stress (Gomer et al., 1996; Jacquier-Sarlin and Polla, 1996) and osmotic stress (Caruccio et al., 1997), and by agents such as heavy metals (Sarge et al., 1993), alkylating agents (Liu et al., 1996), and protease (Rossi et al., 1998) and proteosome inhibitors (Bush et al., 1997; Kawazoe et al., 1998). As in many other cells, the heat shock response also is activated in cells of neuronal origin (Marini et al., 1990; Nishimura et al., 1991; Mathur et al., 1994; Marcuccilli et al., 1996; Brown and Rush, 1999), but little is known about the neuronal control of this signaling system. The heat shock response culminates in the production of heat shock proteins (HSPs), such as heat shock protein-70 (HSP-70), which chaperone misfolded or damaged proteins (for reviews, see Morimoto, 1998; Morano and Thiele, 1999). The significance of the heat shock response in the nervous system is underscored by evidence that the cytotoxicity of proteinaceous aggregates, which are evident in nervous tissues in certain types of neurodegenerative diseases, can be decreased by the overexpression of HSPs. In two such examples, studies in model systems indicate that overexpression of HSPs can curtail the formation of protein aggregates in trinucleotide-repeat diseases (Cummings et al., 1998; Chai et al., 1999) and aggregates containing amyotrophic lateral sclerosis-associated mutant superoxide dismutase (Bruening et al., 1999). In light of the contributions the heat shock response provides to neuronal survival, it is important to identify signaling systems that modulate its activation.
The expression of HSP-70, as well as other HSPs, is regulated by the transcription factor heat shock factor-1 (HSF-1). Normally, HSF-1 is maintained as an inactive monomer associated with HSPs. Stressors, such as elevated temperatures, induce the dissociation of HSPs from HSF-1, facilitating the formation of homotrimeric HSF-1 complexes that, in the proper phosphorylation state, bind consensus sites on DNA and activate transcription. The mechanisms controlling HSF-1 activity are not completely known, but HSF-1 activity is partially controlled by its regulatory domain, which is located between amino acids 221 and 310 (Green et al., 1995), and is modulated by phosphorylation. For example, HSF-1 activity can be negatively regulated by sequential phosphorylation of Ser303 and Ser307 (Chu et al., 1996; Kline and Morimoto, 1997; Xia and Voellmy, 1997; He et al., 1998). Mutation to alanine of Ser307, which is phosphorylated by the extracellular signal-regulated protein kinases 1 and 2 (ERK1/2), results in constitutive activation of HSF-1 in HeLa cells (Knauf et al., 1996; Kline and Morimoto, 1997). In cultured THP-1 human monocytes and NIH 3T3 mouse fibroblasts, Ser303 of HSF-1 is phosphorylated by glycogen synthase kinase-3β (GSK-3β) subsequent to phosphorylation of Ser307 by the ERKs (Chu et al., 1996), and in HeLa cells overexpression of GSK-3β facilitated HSF-1 inactivation (He et al., 1998). These results indicate that, at least in some types of cells, GSK-3β may be a negative regulator of HSF-1 activity and the subsequent expression of HSPs.
GSK-3β is constitutively active, but is also modulated by phosphorylation. GSK-3β can be inhibited by activation of the Wnt pathway (Cook et al., 1996) or by agents that stimulate signaling cascades involving activation of phosphatidylinositol 3-kinase (PI-3K) (for review, see Coffer et al., 1998). Activated PI-3K catalyzes the production of phosphatidylinositol 3,4-bisphosphate and phosphatidylinositol 3,4,5-trisphosphate, which bind the pleckstrin homology domain of Akt (also known as protein kinase B) to bring it into close proximity with phosphoinositide-dependent kinase-1. This juxtaposition facilitates the phosphorylation and activation of Akt by phosphoinositide-dependent kinase-1 (Alessi et al., 1997). Subsequently, Akt can phosphorylate Ser9 of GSK-3β, which inhibits its activity (Cross et al., 1995). Akt also can be activated by stressors such as oxidative stress, osmotic stress, and heat shock, but contradictory reports (Konishi et al., 1996; Shaw et al., 1998) about the contribution of PI-3K activity to heat shock-induced Akt activation have resulted in ambiguity about the signals that regulate stress-induced Akt activation, and thus its regulation of the downstream HSF-1. In the present study, using the neuronal model system of human neuroblastoma SH-SY5Y cells, we report that heat shock-induced activation of Akt is entirely dependent on PI-3K activity. Furthermore, full activation of HSF-1 was found to be dependent on PI-3K activity, and increased GSK-3β activity was inhibitory for the heat shock response, causing decreased activation of HSF-1 and the associated expression of HSP-70.
MATERIALS AND METHODS
SH-SY5Y human neuroblastoma cells were grown in continuous culture media composed of RPMI 1640 medium (Cellgro, Herndon, VA, U.S.A.) supplemented with 5% FetalClone II (HyClone, Logan, UT, U.S.A.), 10% horse serum, 2 mM L-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (GibcoBRL, Grand Island, NY, U.S.A.). SH-SY5Y cells stably overexpressing hemagglutinin (HA)-tagged GSK-3β (with ∼3.5-fold increased GSK-3β levels) (Bijur et al., 2000) were grown in continuous culture media supplemented with 100 μg/ml G418 (geneticin). Cells were plated at a density of 105 cells/60-mm dish and were cultured for 48 h prior to treatments. Adherent cells were rinsed twice with serum-free media supplemented with 2 mM L-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin and were cultured in serum-free media for 24 h prior to all treatments, unless indicated otherwise. For heat shock treatment, cells cultured in 60-mm tissue culture plates sealed with Parafilm were floated in a 45°C water bath with gentle agitation.
To prepare cell lysates, cells in 60-mm plates were washed twice with phosphate-buffered saline and were lysed with 100 μ1 of lysis buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 2 mM EDTA, 2 mM EGTA, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 100 μM phenylmethylsulfonyl fluoride, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 5 μg/ml pepstatin A, 1 nM okadaic acid, and 0.2% Nonidet P-40). The lysates were collected in centrifuge tubes, sonicated, and centrifuged at 16,000 g for 10 min at 4°C. Protein concentrations were determined using the bicinchoninic acid method (Pierce). Cell lysates were mixed with Laemmli sample buffer (2% sodium dodecyl sulfate) and placed in a boiling water bath for 5 min. Proteins were resolved in 10% sodium dodecyl sulfate-polyacrylamide gels, transferred to nitrocellulose, and incubated with anti-human HSP-70 (Stressgen, Victoria, British Columbia, Canada), anti-GSK-3β (PharMingen/Transduction Laboratories, San Diego, CA, U.S.A.), anti-phospho (Ser473) Akt or an antibody for total Akt, anti-phospho (Thr202/Tyr204) ERK1/2, and anti-phospho (Thr180/Tyr182) p38 (New England BioLabs, Beverly, MA, U.S.A.). Immunoblots were developed using horseradish peroxidase-conjugated goat anti-mouse or goat anti-rabbit IgG, followed by detection with enhanced chemiluminescence.
Electrophoretic mobility shift assay (EMSA)
To prepare nuclear extracts, cells in 60-mm plates were washed twice with phosphate-buffered saline and were lysed with 500 μl of lysis buffer (10 mM Tris-HCl, pH 7.4, 3 mM MgCl2, 10 mM NaCl, 0.5% NP-40). Lysed cells were centrifuged at 4,000 g for 5 min at 4°C, the supernatant was discarded, and the pellet containing the nuclei was retained. Nuclear proteins were extracted with 20 mM HEPES, pH 7.9, 20% glycerol, 0.3 M NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 1 mM dithiothreitol, 0.1 mMβ-glycerophosphate, 0.05 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and 1 μg/ml each of pepstatin A, leupeptin, and aprotinin. After extraction on ice for 30 min, the samples were centrifuged at 16,000 g for 15 min at 4°C. The supernatant containing nuclear proteins was transferred to a microfuge tube, and total protein was determined by the method of Bradford (1976). EMSAs were performed using a double-stranded 20-bp oligonucleotide (5′-CTAGAAGCTTCTAGAAGCTT-3′) containing the consensus sequence for HSF-1. The oligonucleotide (200 pmol) was labeled at 37°C for 1 h in 20 μl containing 10× REACT buffer, 10 μM each of dATP, dGTP, and dTTP, 100 μCi of [α-32P]dCTP, and 40 U of Klenow. After 1 h, the sample was diluted to 100 μl with TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA), and free probe was removed by centrifugation at 4,000 g for 45 s on a Sephadex G-50 column. For the binding reaction, 5 μg of nuclear protein was incubated in a total volume of 20 μl in binding buffer containing 20 mM HEPES, pH 7.9, 4% glycerol, 1 mM MgCl2, 50 mM KCl, 0.5 mM dithiothreitol, 1 μg of poly(dI-dC), and ∼10,000 cpm radiolabeled DNA for 30 min at 4°C. DNA—protein complexes were resolved on a preelectrophoresed 5% nondenaturing polyacrylamide gel in 0.25× TBE (22.3 mM Tris, 22.3 mM boric acid, and 0.5 mM EDTA) at 4°C for 1.5 h at 150 V. The gel was dried under vacuum, and the amount of DNA-protein complex present was analyzed using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA, U.S.A.).
Heat shock activates Akt in a PI-3K-dependent manner
The time-dependent effects of heat shock on the activation of Akt, ERK1/2, and p38 were examined in SH-SY5Y cells by measuring changes in phosphorylation-dependent immunoreactivities that have been shown to reflect changes in enzyme activities. Subjecting cells to heat shock (45°C) resulted in a robust activation of Akt as indicated by increased phospho-Ser473 Akt, whereas total Akt did not increase (Fig. 1A). Maximal activation of Akt occurred within 15 min of heat shock treatment. When cells that had been treated with heat shock for 30 min were returned to a 37°C chamber for recovery incubations, there was a time-dependent decrease in Akt activation, which returned to near basal levels by 120 min after heat shock treatment. Treatment with heat shock also activated ERK1/2 and p38. The time course of heat shock-induced ERK1/2 activation (Fig. 1B) revealed that ERK1/2 were mildly activated within 2.5 min of heat shock treatment, and the increased activation-associated phosphorylation remained stable throughout 30 min of heat shock treatment. Heat shock also rapidly increased the activation-associated phosphorylation of p38 (Fig. 1B), which was stimulated more robustly than ERK1/2, and p38 activation continuously increased throughout 30 min of heat shock treatment.
To test if the heat shock-induced activation of Akt was mediated by PI-3K, cells were pretreated with varying concentrations of the chemically unrelated PI-3K inhibitors LY294002 or wortmannin. The lowest concentration of LY294002 (20 μM) that was tested completely blocked the heat shock-induced activation of Akt (Fig. 2A). The lowest concentration of wortmannin (20 nM) that was tested reduced the heat shock-induced activation of Akt to approximately the basal level, and higher concentrations of wortmannin (40 and 60 nM) reduced phosphorylated Akt further. To test if the activations of Akt, ERK1/2, and p38 were interrelated, cells were pretreated with LY294002 (20 μM) and wortmannin (40 nM), with the p38 inhibitor SB203580, or with the inhibitor of ERK1/2 activation, PD098059. Both PI-3K inhibitors completely blocked the heat shock-induced activation of Akt (Fig. 2). In contrast, neither inhibition of p38 with SB203580 nor inhibition of ERK1/2 activation with PD098059 reduced the heat shock-induced activation of Akt. Additionally, activation of PI-3K was not necessary for the heat shock-induced activation of ERK1/2 or p38, because these responses were not inhibited by either of the PI-3K inhibitors. Taken together, these results demonstrate that the heat shock-induced activation of Akt is completely dependent on the activity of PI-3K and is independent of the activation of p38 and ERK1/2.
Heat shock-induced activation of HSF-1 is partially dependent on PI-3K
The time course and regulation of heat shock-induced HSF-1 DNA binding activity were measured in SH-SY5Y cells. Heat shock treatment induced a large increase in HSF-1 DNA binding activity within 15 min (Fig. 3A), and this increased further following 30 min of heat shock treatment. After 30 min of heat shock, HSF-1 DNA binding activity in cells incubated at 37°C returned to near basal levels within 240 min. To determine if heat shock-induced HSF-1 DNA binding activity is affected by inhibition of PI-3K, ERK1/2, or p38, cells were first treated with either LY294002, SB203580, or PD098059 and then subjected to 30 min of heat shock. Heat shock-induced HSF-1 DNA binding activity was unaffected by either SB203580 or PD098059 (Fig. 3B), but treatment with LY294002 resulted in marked attenuation of heat shock-induced HSF-1 DNA binding activity. Treatment with LY294002 also slightly reduced (15 ± 2%) the basal HSF-1 DNA binding activity (data not shown). Thus, heat shock-induced activation of HSF-1 DNA binding activity, as well as Akt, is dependent on activation of PI-3K, but not ERK1/2 or p38.
As HSF-1 activation was attenuated by LY294002, the concentration-dependent effects of two inhibitors of PI-3K, LY294002 and wortmannin, were measured on heat shock-induced HSF-1 DNA binding activity. Treatment of cells with 10-50 μM LY294002 revealed a concentration-dependent inhibition of heat shock-induced HSF-1 DNA binding activity (Fig. 3C). However, in contrast to the total inhibition of Akt activation, LY294002 did not cause complete inhibition of heat shock-induced HSF-1 DNA binding activity. Similarly, wortmannin treatment also attenuated heat shock-induced HSF-1 DNA binding activity in a concentration-dependent manner (Fig. 3D), but it did not cause complete inhibition even at concentrations that totally blocked Akt activation. Furthermore, simultaneous treatment with LY294002 and either SB203580 or PD098059 did not influence the LY294002-mediated attenuation of heat shock-induced HSF-1 DNA binding activity (data not shown).
A primary function of HSF-1 activation is induction of the expression of HSPs, such as HSP-70. To determine if inhibition of HSF-1 DNA binding activity by LY294002 decreased HSP-70 expression, cells were preincubated for 30 min with or without 20 μM LY294002, subjected to heat shock for 30 min, and then allowed to recover at 37°C for 0-6 h. A heat shock-induced increase in the level of HSP-70 was detected after 3 h of recovery incubation at 37°C (Fig. 3E), and HSP-70 levels continued to increase during the next 3 h. Treatment with LY294002 markedly reduced both the basal and heat shock-induced levels of HSP-70. These results demonstrate that PI-3K is necessary for the full activation of Akt and HSF-1 and the expression of HSP-70 induced by heat shock.
Overexpression of GSK-3β attenuates heat shock-induced HSF-1 DNA binding activity
One of the outcomes of Akt activation is the inhibition of GSK-3β, which raises the possibility that GSK-3β inhibition may be necessary for full activation of HSF-1. To test if GSK-3β modulates the heat shock response, HSF-1 DNA binding activity was measured in SH-SY5Y cells stably overexpressing HA-tagged GSK-3β. Figure 4A shows the levels of GSK-3β and HSF-1 in untransfected control cells, vector-transfected cells, and three HA-GSK-3β-transfected cell lines. Two bands are evident on the GSK-3β immunoblots because HA-GSK-3β migrates more slowly than endogenous GSK-3β due to the HA tag. In these three cell lines, total GSK-3β averaged 330 ± 9% and 362 ± 9% of the GSK-3β in untransfected cells and vector-transfected cells, respectively. In a previous study, the activity of GSK-3β was shown to be increased to approximately the same extent as protein levels in the HA-GSK-3β-transfected cells (Bijur et al., 2000). Overexpression of GSK-3β did not influence the basal levels of HSF-1 protein (Fig. 4A), but significantly attenuated the basal HSF-1 DNA binding activity (Fig. 4B). In these three cell lines, basal HSF-1 DNA binding activity was 57 ± 8% of that in vector-transfected cells. The time-dependent activation of HSF-1 DNA binding activity was examined in control cells, vector-transfected cells, and one clone of HA-GSK-3β-transfected cells (Fig. 5A and B) exposed to continuous heat shock for 0-60 min. In this experiment, the cells were maintained in media containing serum to minimize heat shock toxicity. In control, vector-transfected, and HA-GSK-3β-transfected cells, HSF-1 DNA binding activity increased within 5 min of exposure to heat shock. Maximum HSF-1 DNA binding activity in control, vector-transfected, and HA-GSK-3β-transfected cells was attained following 30 min of continuous exposure to heat shock, but in HA-GSK-3β-transfected cells there was a significant attenuation of HSF-1 DNA binding activity throughout the time course of heat shock treatment compared with control and vector-transfected cells. Heat shock-induced HSF-1 DNA binding activity was also attenuated in the other two clones of HA-GSK-3β-transfected cells (Fig. 5C). The attenuated HSF-1 DNA binding activity due to GSK-3β overexpression corresponded to a marked attenuation of the basal as well as heat shock-induced HSP-70 expression (Fig. 5D), and the inhibition was partially blocked by treatment with 5 mM lithium, a selective GSK-3β inhibitor (Klein and Melton, 1996). These results demonstrate that GSK-3β is an inhibitor of the activation of HSF-1 and the subsequent induction of HSP-70.
During the last few years, much evidence has supported the concept that impaired neuronal function and survival associated with many neurodegenerative diseases and aging are attributable to inadequate survival-supporting responses to cellular stress (Cummings et al., 1998; Bruening et al., 1999; Chai et al., 1999), emphasizing the need to understand the mechanisms mediating these responses. One mechanism by which neurons cope with stress is called the heat shock response, which includes the activation of the HSF-1 transcription factor and the ensuing up-regulation of the production of HSPs. However, relatively little is known about how this critical survival-promoting system is regulated in neurons or neuronal model systems. The results of the present investigation revealed that in SH-SY5Y cells, HSF-1 activation is partially dependent on PI-3K, but is independent of ERK1/2 and p38. Furthermore, GSK-3β was shown to be an inhibitory regulator of the activation of HSF-1, and GSK-3β hyperactivity led to an attenuated production of HSP-70 in response to heat shock.
Much evidence has demonstrated that activation of Akt supports cell survival and counteracts apoptotic signaling (Coffer et al., 1998), a part of which relies on inactivation of GSK-3β (Pap and Cooper, 1998; Bijur et al., 2000), but the signaling mechanisms that regulate stress-induced activation of Akt remain to be clarified. Heat shock is well known to cause the activation of Akt (Konishi et al., 1996; Lin et al., 1997), but contradictory evidence concerning the mechanisms governing this response has been reported, apparently due, at least in part, to cell-specific differences in the signaling pathways that are recruited. For example, heat shock-induced activation of Akt in COS-7 and NIH 3T3 cells was unaffected by treatment with the PI-3K inhibitor wortmannin (Konishi et al., 1996), indicating that a PI-3K independent pathway is involved in the heat shock-induced activation of Akt. Furthermore, Konishi et al. (1996) suggested that Akt may regulate the activity of p38. Contrary to these findings, Shaw et al. (1998) reported that the heat shock-induced activation of Akt in NIH 3T3 and human embryonic kidney 293 cells was dependent on PI-3K. The present study demonstrated that the heat shock-induced activation of Akt was abrogated by inhibition of PI-3K with LY294002 or wortmannin. Furthermore, inhibition of PI-3K and the associated inhibition of Akt did not reduce heat shock-induced activation of p38 or ERK1/2, demonstrating that these kinases are activated by a different signaling mechanism. Taken together, these results demonstrate the dependence on PI-3K in the heat shock-induced activation of Akt without regulatory interactions with p38 and ERK1/2.
Heat shock also activates HSF-1, which induces increases in the expression of chaperone HSPs, such as HSP-70, but the regulation of HSF-1 activation is not entirely understood. HSF-1 activity can be down-regulated by several different means; for example, the association of HSF-1 with HSPs (Shi et al., 1998; Zou et al., 1998) and disruption of HSF-1 homotrimerization (Zuo et al., 1995) both down-regulate HSF-1 activity, demonstrating that HSF-1 is amenable to regulation at different stages during the activation process. In the present study, attenuation of heat shock-induced HSF-1 activity by inhibition of PI-3K and by overexpression of GSK-3β demonstrates two additional modes by which heat shock-induced HSF-1 activity can be regulated. The inhibitory influence of GSK-3β on HSF-1 activity has been demonstrated previously (He et al., 1998), as overexpression of GSK-3β in HeLa cells was reported to result in a more rapid deactivation of HSF-1 during recovery from heat shock compared with cells not overexpressing GSK-3β. However, in HeLa cells, heat shock-induced activation of HSF-1 was independent of PI-3K, as it was unaffected by treatment with wortmannin (He et al., 1998), indicating that there are cell-specific signaling mechanisms that control HSF-1 activity. In contrast to the findings in HeLa cells, the results in the present study demonstrate for the first time that in a neuronal model system perturbations in the PI-3K/Akt signaling pathway can severely impede the robust activation of HSF-1 that occurs in cells subjected to heat shock. However, as the heat shock-induced HSF-1 activity could only be diminished, but not abolished, by inhibitors of PI-3K, it is evident that HSF-1 activation is also partially mediated by PI-3K-independent mechanisms.
In summary, PI-3K was found to be necessary for full heat shock-induced activation of the survival-promoting molecules Akt and HSF-1. Furthermore, GSK-3β was shown to be an inhibitory modulator of the activation of HSF-1 and the associated production of HSP-70, whereas lithium treatment facilitated HSP-70 production. The inhibitory effect of GSK-3β on HSF-1 activation and HSP-70 expression implies that Akt-induced inhibition of GSK-3β is a contributory factor in the heat shock-induced activation of HSF-1. Thus, stimulation of PI-3K and adequate inhibition of GSK-3β are necessary for cells to fully activate the survival-promoting activation of HSF-1 and the increased expression of HSPs. In some conditions, lithium may provide neuroprotection in part by facilitating these cell survival-promoting responses to stress.