Hsp27 binding to the 3′UTR of bim mRNA prevents neuronal death during oxidative stress–induced injury: a novel cytoprotective mechanism

Bim protein mediates neuronal apoptosis under certain stress conditions. In neurons exposed to oxidative stress, Hsp27 prevents neuronal death, repressing bim mRNA translation through its binding to the 3′-untranslated region. This neuroprotective mechanism may provide novel therapeutic approaches in diverse neurodegenerative conditions.

Abstract

Neurons face a changeable microenvironment and therefore need mechanisms that allow rapid switch on/off of their cytoprotective and apoptosis-inducing signaling pathways. Cellular mechanisms that control apoptosis activation include the regulation of pro/antiapoptotic mRNAs through their 3′-untranslated region (UTR). This region holds binding elements for RNA-binding proteins, which can control mRNA translation. Here we demonstrate that heat shock protein 27 (Hsp27) prevents oxidative stress–induced cell death in cerebellar granule neurons by specific regulation of the mRNA for the proapoptotic BH3-only protein, Bim. Hsp27 depletion induced by oxidative stress using hydrogen peroxide (H₂O₂) correlated with bim gene activation and subsequent neuronal death, whereas enhanced Hsp27 expression prevented these. This effect could not be explained by proteasomal degradation of Bim or bim promoter inhibition; however, it was associated with a specific increase in the levels of bim mRNA and with its binding to Hsp27. Finally, we determined that enhanced Hsp27 expression in neurons exposed to H₂O₂ or glutamate prevented the translation of a reporter plasmid where bim-3′UTR mRNA sequence was cloned downstream of a luciferase gene. These results suggest that repression of bim mRNA translation through binding to the 3′UTR constitutes a novel cytoprotective mechanism of Hsp27 during stress in neurons.

INTRODUCTION

Bcl-2 homology domain (BH) 3–containing proteins (BH3-only proteins) couple stress signals to the intrinsic mitochondrial pathway of apoptosis. Accordingly, their levels within cells are tightly regulated to avoid inappropriate activation of the apoptosis program. The BH3-only protein Bim (Bcl-2 interacting mediator of cell death) mediates apoptosis in different neuronal types under diverse stress situations (Gilley et al., 2003; Butts et al., 2005; Zimmermann et al., 2005; Concannon et al., 2010; Xie et al., 2011). Bim promoter activation depends on the simultaneous binding of transcription factors such as FOXO3 and AP1 (Whitfield et al., 2001; Gilley et al., 2003; Biswas et al., 2007). Bim can also be posttranscriptionally regulated by phosphorylation and subsequent degradation via the proteasome (Ley et al., 2003; Luciano et al., 2003; Meller et al., 2006). Furthermore, bim mRNA can be regulated through the 3′-untranslated region (3′UTR; Matsui et al., 2007; Terasawa et al., 2009). This region binds miRNAs and RNA-binding proteins (RBPs), which are able to regulate mRNA stability and/or translation (de Moor et al., 2005; Bolognani et al., 2008).

The heat shock proteins (Hsps) are stress-inducible proteins that exert cytoprotective and antiapoptotic actions (Xanthoudakis and Nicholson, 2000; Lanneau et al., 2008). The Hsp27 isoform (Hsp25 in rodents) has an important neuroprotective role (Latchman et al., 2005), whose exact cellular mechanisms have not yet been fully elucidated. Hsp27 is able to prevent neuronal apoptosis (Wagstaff et al., 1999; Benn et al., 2002; King et al., 2009) by its direct binding to proapoptotic proteins such as cytochrome C (Bruey et al., 2000); however, evidence indicates that Hsp27 can also interfere with apoptotic signaling upstream of mitochondrial cytochrome C release (Paul et al., 2002; Havasi et al., 2008; Stetler et al., 2008). Hsp effects are associated with their chaperone activity, binding misfolded, oxidized, or proapoptotic proteins and facilitating their degradation by the proteasome (Garrido et al., 2006; Arrigo 2007; Kalmar and Greensmith, 2009). Hsps also display the capacity to act as RNA-binding proteins (Henics et al., 1999; Matsui et al., 2007; Sinsimer et al., 2008). Here we describe how enhanced Hsp27 levels during oxidative stress in cerebellar granule neurons (CGNs) prevent Bim protein up-regulation and subsequent neuronal death. Hsp27 effects depend on binding to the bim mRNA transcript and the regulation of its 3′UTR to repress its translation. Together these results identify a novel posttranscriptional mechanism by which Hsp27 opposes neuronal death.

RESULTS

Bim mediates oxidative stress–induced cell death in cerebellar granule neurons

We induced oxidative stress in CGN cultures using hydrogen peroxide (H₂O₂; 25–50 μM) addition. This treatment induces an abrupt increase in reactive oxygen species (ROS) and oxidative stress in CGNs (Davila and Torres-Aleman, 2008). ROS can act as signaling molecules, and their harmful effects include the activation of redox-sensitive proapoptotic pathways (Ueada et al., 2002; Handy and Loscalzo, 2012). H₂O₂ treatment induced cell death in CGNs with typical features of apoptosis, including nuclear pyknosis, caspase activation, and DNA fragmentation (Figure 1A).

FIGURE 1:

Bim mediates oxidative stress–induced cell death in CGNs. (A) CGNs treated with H₂O₂ (37.5 μM), showing typical features of apoptotic cell death 4–6 h after addition. Nuclear pyknosis was detected by Hoechst staining, caspase activation was detected by in situ staining with the fluorescence marker CaspACE FITC-VAD-FMK, and DNA fragmentation was detected by TUNEL staining. (B) Bim mRNA levels were quantified by qPCR. CGNs treated with H₂O₂ (37.5 or 50 μM) showed significant up-regulation of bim (*p < 0.05; n = 3). (C) Western blot analysis revealed a significant increase in Bim protein levels 4–6 h after H₂O₂ (37.5 or 50 μM) addition (*p < 0.05; n = 5). This up-regulation correlated with a decrease of pAKT (Ser-473), pFOXO3 (Thr-32), and Hsp25 (rodent homologue of Hsp27) levels. β-Actin served as loading control. (D) CGNs were transfected with a vector containing a 0.8-kb fragment of the bim promoter. Bim promoter activation was significantly increased after H₂O₂ (37.5 μM) addition (*p < 0.05; n = 3). Mutations of the FOXO3-binding site significantly reduced this activation (*p < 0.05; n = 3). (E) CGNs from bim^−/− mice and wild-type (WT) controls were treated with H₂O₂ (37.5 μM) or sham conditions. At 4–6 h posttreatment, the neurons were stained live with Hoechst. Bar, 2.5 μm. (F) Pyknotic nuclei were scored (*p < 0.05; n = 3).

Four hours after H₂O₂ addition, we detected a significant up-regulation of the proapoptotic BH3-only protein Bim at both mRNA and protein levels (Figure 1, B and C). These results correlated with the activation of the bim promoter. Thus neurons transfected with a luciferase reporter plasmid bearing the bim promoter sequence showed higher luciferase activity after H₂O₂ addition (37.5 μM) compared with control neurons treated with vehicle (Figure 1D). Mutation of the FOXO3-binding sites on the bim promoter prevented its activation (Figure 1D). Moreover, we also found that H₂O₂ addition decreased activation of the kinase AKT, as assessed by Western blot analysis of pAKT (Ser-473) levels, and also the phosphorylation of the AKT target, FOXO3 (Thr-32) (Figure 1C). This step is necessary to allow FOXO3 nuclear accumulation and transcriptional activity (Brunet et al., 1999). Parallel to Bim up-regulation, we also detected a decrease of the total levels of Hsp25 (rodent homologue of Hsp27) induced by H₂O₂ treatment (Figure 1C).

Finally, to determine whether Bim has a causal role in H₂O₂-induced cell death, we compared CGN survival between bim^−/− and wild-type (wt) control cultures treated with H₂O₂ (37.5 μM). Bim-deficient neurons showed resistance to H₂O₂, displaying a lower percentage of cells with nuclear pyknosis compared with wt neurons after treatment (Figure 1, E and F).

Hsp27 regulates Bim protein levels and oxidative stress–induced cell death

Having observed the down-regulation of Hsp25 levels induced by H₂O₂ treatment, we analyzed its possible relation with the expression of Bim. For this purpose, we transfected CGNs with a construct expressing the wild-type form of the human Hsp27 (pNEO-Hsp27). Despite the fact that we also observed a decrease in levels of Hsp27 in H₂O₂-treated CGNs, transfection of Hsp27 prevented the up-regulation of Bim protein levels (Figure 2A) and subsequent apoptotic events such as cytosolic accumulation of cytochrome C (Figure 2B) or accumulation of the active caspase 3 subunit (Supplemental Figure S1A). Hsp27 transfection in CGNs also prevented the nuclear pyknosis induced by H₂O₂ treatment (Figure 2C). On the other hand, depletion of the endogenous Hsp25 levels in CGNs by expression of a specific small interfering RNA (siRNA) targeting Hsp25 (200 pmol) was able to mimic the effect of H₂O₂ treatment, up-regulating significantly Bim protein levels and inducing nuclear pyknosis (Figure 2, D and E).

FIGURE 2:

Hsp27 regulates Bim protein levels and oxidative stress–induced cell death. (A) CGNs were transfected with pNEO-Hsp27 or a control construct before H₂O₂ (37.5 μM) addition. pNEO-Hsp27 neurons displayed significantly lower Bim protein levels than did control neurons after H₂O₂ treatment (*p < 0.05; n = 3). β-Actin served as loading control. (B) At 4 h after H₂O₂ (37.5 μM) treatment, pNEO-Hsp27 neurons displayed lower cytosolic cytochrome C protein levels than in control neurons. Mitochondrial membrane levels of cytochrome C were also higher in pNEO-Hsp27 neurons than in control neurons. IκB-α protein served as marker of cytosolic fraction and LAMP1 as marker of the membrane fraction, including mitochondrial membranes. (C) CGNs transfected with pNEO-Hsp27 or a control construct were treated with H₂O₂ (37.5 μM) or sham conditions and live-stained with Hoechst 4–6 h later. pNEO-Hsp27 neurons showed a significantly lower percentage of cells with pyknotic nuclei than control neurons after H₂O₂ addition (*p < 0.05; n = 3). Bar, 2.5 μm. (D) CGNs were electroporated with Hsp25 siRNA (rodent homologue of Hsp27; 100/200 pmol) or control siRNA (200 pmol), and Bim and Hsp25 protein levels were analyzed after 72 h by Western blot. Hsp25 levels were depleted in Hsp25 siRNA (200 pmol) neurons, which also displayed significantly higher levels of Bim than control siRNA neurons (*p < 0.05; n = 3). β-Actin served as loading control. (E) CGNs electroporated with Hsp25 siRNA or control siRNA were live-stained with Hoechst. Hsp25 siRNA neurons presented a significantly higher percentage of cells with pyknotic nuclei than control neurons (*p < 0.05; n = 3). Bar, 2.5 μm.

Hsp27 effects depend on a posttranscriptional mechanism

We next sought to determine the mechanism used by Hsp27 to prevent Bim induction during oxidative stress–induced cell death. First, we analyzed a possible effect of Hsp27 on bim promoter activation by the AKT/FOXO3 pathway. Overexpression of the pNEO-Hsp27 construct in CGNs did not prevent the down-regulation of pAKT (Ser-473) and pFOXO3 (Thr-32) levels induced by H₂O₂ (37.5 μM) treatment (Figure 3A). We also examined the JNK/AP1 signaling pathway, which is activated by oxidative stress and involved in bim promoter activation (Torres and Forman, 2003; Biswas et al., 2007). Treatment of CGNs with H₂O₂ (37.5 μM) increased JNK activation, as detected by the up-regulation of pJNK (Thr-183/Tyr-185) levels (Figure 3A). However, overexpression of the pNEO-Hsp27 did not prevent this effect. Finally, we cotransfected CGNs with the luciferase reporter plasmid bearing the bim promoter sequence and the pNEO-Hsp27 construct or alternatively its control vector. Hsp27 overexpression did not prevent the up-regulation of the bim promoter after H₂O₂ addition (Figure 3B). These results indicate that the effect of Hsp27 on Bim expression does not depend on the regulation of its promoter.

FIGURE 3:

Hsp27 effects depend on a posttranscriptional mechanism. (A) CGNs were transfected with pNEO-Hsp27 or a control construct before H₂O₂ (37.5 μM) addition. Western blot analysis showed that H₂O₂ treatment decreased pAKT (Ser-473) and pFOXO3 (Thr-32) levels and increased pJNK (Thr-183/Tyr-185) levels both in pNEO-Hsp27 neurons and in control neurons. β-Actin served as loading control. (B) CGNs were cotransfected with the luciferase reporter containing the bim promoter sequence and either pNEO-Hsp27 or a control construct. Bim promoter activation was significantly increased after H₂O₂ (37.5 μM) addition both in pNEO-Hsp27 neurons and in control neurons (*p < 0.05; n = 3). (C) CGNs were pretreated with the proteasome inhibitor epoxomicin (50 nM) or vehicle 45 min before H₂O₂ (37.5/50 μM) addition. Western blot analysis revealed that H₂O₂ treatment down-regulated Hsp25 protein level (*p < 0.05; n = 3) and up-regulated Bim levels. Epoxomicin pretreatment prevented Hsp25 down-regulation and increased Bim protein levels induced by H₂O₂. β-Actin served as loading control. (D) CGNs were transfected with pNEO-Hsp27 or a control construct and pretreated with epoxomicin (50 nM) or vehicle 45 min before H₂O₂ (37.5 μM) addition. pNEO-Hsp27 neurons presented lower Bim protein levels after H₂O₂ addition than control neurons in the presence of epoxomicin or vehicle (*p < 0.05; n = 3). β-Actin served as loading control. (E) CGNs were transfected with pNEO-Hsp27 or a control construct before H₂O₂ (37.5 μM) addition. qPCR analysis showed that H₂O₂ treatment up-regulated bim mRNA levels (*p < 0.05; n = 3). pNEO-Hsp27 neurons treated with H₂O₂ showed significant higher levels of bim mRNA than H₂O₂-treated control neurons (*p < 0.05; n = 3). (F) ATF4 mRNA levels were also quantified by qPCR. H₂O₂ (37.5 μM) addition increased ATF4 mRNA levels (*p < 0.05; n = 3); however pNEO-Hsp27 transfection did not have any significant effect on these levels.

Bim protein levels are tightly controlled by ubiquitin-dependent proteasomal degradation. Pretreatment of CGNs with an inhibitor of the proteasome, epoxomicin (50 nM), 45 min before H₂O₂ addition prevented down-regulation of Hsp25 and increased Bim protein levels. In addition, we also observed an increase in phosphorylation of Bim on Ser-69, a necessary step for its proteasomal degradation (Luciano et al., 2003). These results suggest the partial degradation of both proteins at the proteasome during oxidative stress (Figure 3C). Considering this, we examined whether enhanced levels of Hsp27 could stimulate Bim protein degradation during oxidative stress. To this end, we pretreated CGNs transfected with pNEO-Hsp27 or its control vector with epoxomicin and 45 min later exposed them to H₂O₂ (37.5 μM) or vehicle. Epoxomicin induced a slight increase in Bim protein levels in all the experimental groups; however, this was not statistically significant. Epoxomicin was also not able to prevent the significant decrease in Bim proteins levels induced in H₂O₂-treated cells by pNEO-Hsp27 overexpression (Figure 3D). These results indicate that Hsp27 modulates Bim expression during oxidative stress by a mechanism independent of its proteasomal degradation.

Finally, we analyzed by quantitative real-time reverse transcription (RT)-PCR analysis (qPCR) the effect of Hsp27 on bim mRNA levels during oxidative stress. CGNs transfected with a control vector showed up-regulation of bim mRNA levels after H₂O₂ (37.5 μM) treatment. This up-regulation was significantly higher in CGNs treated with H₂O₂ but transfected with the pNEO-Hsp27 construct (Figure 3E). To determine whether this effect was specific for the bim mRNA transcript or was generic for every mRNA transcript, we also quantified the mRNA levels of ATF4, an oxidative stress–inducible transcription factor in neurons (Lange et al., 2008). H₂O₂ (37.5 μM) treatment also up-regulated ATF4 mRNA levels in CGNs; however, in this case, pNEO-Hsp27 overexpression did not affect ATF4 mRNA levels (Figure 3F). These results suggest that during oxidative stress, Hsp27 overexpression specifically represses bim mRNA translation.

Hsp27 regulates the bim-3′UTR during oxidative stress

To determine whether Hsp27 was able to inhibit bim mRNA translation, we analyzed the amount of bim RNA present at the polysomes (the ribonucleoprotein particles where translation occurs) using the translating ribosome affinity purification (TRAP) methodology (Heiman et al., 2008; Kapeli and Yeo, 2012). Specifically, we determined by coimmunoprecipitation and subsequent qPCR analysis the levels of bim mRNA associated with the ribosomal protein RPL10a present at the polysome. In CGNs treated with H₂O₂ (37.5 μM), overexpression of Hsp27 significantly reduced the levels of bim mRNA associated with RPL10a and therefore present at the polysome (Figure 4A).

FIGURE 4:

Hsp27 regulates bim-3′UTR sequence during oxidative stress. (A) Quantification of bim mRNA levels bound to the ribosomal protein RPL10a present at the polysome. CGNs were treated with H₂O₂ (37.5 μM). Hsp27 overexpression reduced bim mRNA levels coimmunoprecipitated with RPL10a when compared with controls (*p < 0.05; n = 3). (B) CGNs were cotransfected with a luciferase reporter plasmid containing the bim-3′UTR (divided into two sequences, named 1 and 2) and either pNEO-Hsp27 or a control construct. Bim-3′UTR (sequence 1) neurons showed a significant increase in the luciferase activity after H₂O₂ (37.5 μM) addition (*p < 0.05; n = 3). Coexpression of pNEO-Hsp27 prevented this effect, significantly reducing the luciferase activity (*p < 0.05; n = 3). (C) Bim-3′UTR (sequence 2) neurons did not show any modification of the luciferase activity induced by H₂O₂ treatment or pNEO-Hsp27 coexpression. (D) CGNs were coelectroporated with the luciferase reporter plasmid containing the bim-3′UTR (sequence 1) and either Hsp25 siRNA or control siRNA. Hsp25 siRNA neurons displayed significantly higher luciferase activity than control siRNA neurons (*p < 0.05; n = 3). (E) Quantification of bim mRNA levels bound to the endogenous Hsp25 protein and to the wild-type form of the human Hsp27 (pNEO-Hsp27 construct) overexpressed in CGNs. H₂O₂ (37.5 μM) treatment significantly increased bim mRNA levels coimmuno­precipitated with the Hsp25/Hsp27 compared to control neurons (*p < 0.05; n = 4). The coimmunoprecipitated bim mRNA levels were significantly increased in H₂O₂-treated cells when Hsp27 was overexpressed (*p < 0.05; n = 4). As positive and negative controls for the immuno­precipitation, a specific antibody for Argonaute-2 protein and an unspecific serum were used, respectively. Argonaute-2 levels were detected by Western blot. bim mRNA levels were not detected in the negative control.

Next we analyzed whether Hsp27 was able to regulate bim mRNA translation via its 3′UTR. This region contains microRNA sites and cis-acting sequences within AU-rich sequence elements (AREs), which can bind RBPs. These elements are critical for the regulation of bim mRNA stability and/or translation (Matsui et al., 2007; Terasawa et al., 2009). To this end, this region was divided into two overlapping sequences (named here 1 and 2), which were cloned downstream of a luciferase gene in a chimeric reporter plasmid whose expression was controlled by a constitutively active promoter. Therefore the luciferase activity of these reporter plasmids could only be posttranscriptionally regulated by the modulation of the bim-3′UTR. Exposure of CGNs to H₂O₂ (37.5 μM) resulted in up-regulation of the luciferase activity of the reporter plasmid containing sequence 1 (Figure 4B). This effect was abrogated by coexpression of the pNEO-Hsp27 construct (Figure 4B). As a control for the specificity of this effect, we observed that Hsp27 coexpression did not have an effect on the luciferase activity of the reporter plasmid containing sequence 2 (Figure 4C). Next we demonstrated that depletion of endogenous Hsp25 levels in CGNs by expression of a specific siRNA targeting Hsp25 was able to mimic the effect of H₂O₂ treatment, up-regulating the luciferase activity of the reporter plasmid containing sequence 1 (Figure 4D). These results indicated that enhanced levels of Hsp27 during oxidative stress was able to repress bim mRNA translation by a mechanism dependent on the bim-3′UTR.

Finally, we analyzed whether the effect of Hsp27 on bim mRNA translation depended on the regulation of other RBPs or was a direct effect, since several members of the heat-shock protein family, including Hsp27, show affinity for AREs and can act as RBPs (Henics et al., 1999; Sinsimer et al., 2008). To this end, we immunoprecipitated the endogenous Hsp25 protein and the wild-type form of the human Hsp27 (pNEO-Hsp27 construct) overexpressed in CGNs and quantified by qPCR the levels of bound bim mRNA. CGNs treatment with H₂O₂ (37.5 μM) significantly increased bim mRNA levels that were coimmunoprecipitated with endogenous Hsp25 when compared with control neurons (Figure 4E). The coimmunoprecipitated bim mRNA levels were significantly increased in H₂O₂-treated cells when Hsp27 was overexpressed (Figure 4E). These results demonstrated binding of Hsp25/27 to bim mRNA during oxidative stress and suggest that overexpression of Hsp27 stimulates this binding.

Hsp27 also regulates the bim-3′UTR during excitotoxicity

Previous results from our group showed that excitotoxic apoptosis in two different types of neurons—CGNs and cortical neurons—required Bim induction (Concannon et al., 2010). We therefore explored whether the described regulation of the bim-3′UTR sequence by Hsp27 was also evident during excitotoxic cell death. First, we observed that overexpression of the pNEO-Hsp27 construct was able to prevent Bim protein up-regulation in CGNs exposed to glutamate (100 μM)/glycine (10 μM; Figure 5A). Next we explored whether Hsp27 could also modify the luciferase activity of the reporter plasmid containing the bim-3′UTR (sequence 1) during excitotoxicity. CGNs exposed to glutamate (100 μM)/glycine (10 μM) showed a decrease of the luciferase activity compared with control CGNs. This effect was in contrast to the increase in luciferase activity observed in H₂O₂-treated cells, suggesting that additional mechanisms regulate bim-3′UTR during excitotoxicity. However, the decrease was significantly enhanced by overexpression of Hsp27 (Figure 5B). This effect was also observed in cortical neurons exposed to N-methyl-d-aspartate (NMDA; 100 μM)/glycine (10 μM), where Hsp27 overexpression decreased luciferase activity (Figure 5C). These results demonstrate that Hsp27 overexpression also represses bim mRNA translation during excitotoxicity.

FIGURE 5:

Hsp27 also regulates bim-3′UTR sequence during excitotoxicity. (A) CGNs were transfected with pNEO-Hsp27 or a control construct before glutamate (100 μM)/glycine (10 μM) exposure. pNEO-Hsp27–transfected neurons exhibited significantly lower Bim protein levels than control neurons after H₂O₂ treatment (*p < 0.05; n = 3). β-Actin served as loading control. (B) CGNs were cotransfected with a luciferase reporter plasmid containing the bim-3′UTR (sequence 1) and either pNEO-Hsp27 or a control construct. Glutamate/glycine (100/10 μM) exposure decreased the luciferase activity in control neurons (*p < 0.05; n = 3), whereas expression of the pNEO-Hsp27 construct significantly enhanced this effect (*p < 0.05; n = 3). (C) Cortical neurons were cotransfected with a luciferase reporter plasmid containing the bim-3′UTR (sequences 1) and either pNEO-Hsp27 or a control construct. NMDA (100 μM)/glycine (10 μM) exposure decreased the luciferase activity in control neurons (*p < 0.05; n = 3), and expression of the pNEO-Hsp27 construct significantly enhanced this effect (*p < 0.05; n = 3).

DISCUSSION

The present study describes the regulation of the proapoptotic BH3-only protein Bim by Hsp27 in neurons undergoing oxidative stress. In an in vitro model, we observed that H₂O₂ addition to CGNs induced the up-regulation of bim mRNA and protein levels, which was necessary for CGN cell death. In parallel, we also detected a down-regulation of Hsp25 levels induced by H₂O₂. We report that enhancement of Hsp27 levels during oxidative stress is able to prevent Bim protein up-regulation and downstream events such as cytochrome C release, caspase-3 activation, and neuronal death despite the accumulation of bim mRNA. Finally, we show that the main mechanism that mediates the effect of Hsp27 is regulation of the 3′UTR of bim mRNA, repressing its translation. Together these studies reveal a novel posttranscriptional mechanism controlling oxidative stress–induced neuronal death and describe a novel cytoprotective mechanism of Hsp27.

Protein levels of endogenous Hsp25 were down-regulated in CGNs treated with H₂O₂, in spite of its described stress-dependent expression (Xanthoudakis and Nicholson, 2000; Lanneau et al., 2008). Hsp25/27 is able to act as a chaperone of oxidized proteins, facilitating their ubiquitination and proteasomal degradation (Arrigo, 2001, 2007), and this function could be related to its down-regulation after H₂O₂ treatment. Supporting this concept, it has been described that Hsp27 can be ubiquitinated and degraded (Sun et al., 2011). Indeed, our results show that pretreatment with the proteasome inhibitor epoxomicin abrogated H₂O₂-induced down-regulation of Hsp25. Both H₂O₂ treatment and transfection with a siRNA-Hsp25 up-regulated Bim protein levels and induced neuronal apoptosis, which was prevented in bim^−/− CGNs. On the other hand, Hsp27 overexpression prevented Bim protein expression and subsequent cytochrome C release, caspase-3 activation, and neuronal death. These results indicate that Hsp25/27 down-regulation could be a key process to allow Bim expression and neuronal apoptosis during oxidative stress conditions, confirming previous studies that correlate Hsp27 levels with cell survival/death during diverse stress conditions (Wagstaff et al., 1999; Stetler et al., 2008; King et al., 2009; Gibert et al., 2012). Expression of another BH3-only protein, PUMA, has been described as necessary for oxidative stress–induced apoptosis in cortical neurons (Steckley et al., 2007). However, puma-deficient CGNs were not protected against H₂O₂ in our experimental setting (Supplemental Figure S1B), suggesting a cell type– and stimulus-dependent role of BH3-only proteins.

Apoptosis is a terminal cellular decision, and the proteins that control its activation, including Bim, require multiple levels of regulation. We observed that epoxomicin pretreatment significantly increased Bim protein levels and its phosphorylation (Ser-69) in H₂O₂-treated CGNs but did not prevent the effect of Hsp27 overexpression on Bim protein levels. These results confirm the regulation of Bim by proteasomal degradation (Luciano et al., 2003) but also indicate a proteasome-independent effect of Hsp27. Hsp27 overexpression in H₂O₂-treated cells altered neither bim promoter activation nor the signaling pathways involved. However, Hsp27 overexpression increased bim mRNA levels without modifying bim promoter activity. This stabilization and up-regulation of bim mRNA levels could be associated with the repression of mRNA translation (Kawai et al., 2004). This hypothesis was confirmed by determining bim mRNA levels at the polysome using TRAP methodology. Epoxomicin also increased Hsp25 protein levels in H₂O₂-treated cells, but in contrast to Hsp27 overexpression, this increase did not affect Bim protein levels. These results can be explained by Hsp25 proteosomal degradation occurring after bim mRNA translation or nondegraded Hsp25 protein being trapped at the proteasome. On the contrary, Hsp27 overexpression before oxidative stress and bim expression would still exert an effect on Bim translation. Finally, Hsp27 overexpression did not affect other stress response genes, such as ATF4, also up-regulated during oxidative stress (Lange et al., 2008). This result excludes general inhibition of translation as a main mechanism of the effects observed (Cuesta et al., 2000) but instead indicates specific repression of the bim mRNA translation by Hsp27.

Our reporter assays revealed the involvement of the bim-3′UTR in the repression of bim mRNA translation by Hsp25/27. Hsp25 depletion induced by H₂O₂ treatment or siRNA-mediated Hsp25 knockdown allowed the translation of a luciferase reporter plasmid bearing half of the bim-3′UTR (sequence 1), whereas Hsp27 overexpression in H₂O₂-treated cells prevented this translation. Hsp27 overexpression also increased the levels of bim mRNA coimmunoprecipitated in H₂O₂-treated cells, indicating that oxidative stress stimulates Hsp27 binding to the bim-3′UTR. In agreement with this concept, previous studies showed that oxidative stress stimulates the antiapoptotic activity of Hsp27 (Huot et al., 1996). Furthermore, these results also confirm Hsp25/27 as a RBP that binds to 3′UTRs (Sinsimer et al., 2008). The binding of Hsp27 to the 3′UTR of bim could be a result of the affinity of Hsp27 for AREs (Sinsimer et al., 2008), which stimulate mRNA degradation but also stabilization and translational repression (Mukhopadhyay et al., 2003; Barreau et al., 2005). A small sequence of 1 kb that holds phylogenetically conserved AREs (able to bind other heat shock proteins; Matsui et al., 2007) is present in the reporter plasmid bearing the first half of the bim-3′UTR. The presence of these AREs could explain the effect of Hsp25/27 on the reporter plasmid bearing bim-3′UTR (sequence 1) but not on the reporter that contains the second half of the bim-3′UTR (sequence 2). However, multiple AREs are present along the bim-3′UTR, and additional studies are required to map specific interactions with Hsp25/27.

Our results demonstrate how Hsp25/27 levels during oxidative stress situations are crucial in regulating bim mRNA translation and neuronal apoptosis and also underscore the existence of multiple control points between transcription and protein production in the neuroprotective mechanisms of Hsps. The observed bim-mediated, antiapoptotic mechanism would occur before the irreversible step of cytochrome C release and thus may explain the known upstream role of Hsp27 in regulating the mitochondrial pathway of apoptosis (Paul et al., 2002; Havasi et al., 2008). Other mechanisms by which Hsp27 prevents neuronal apoptosis have also been identified, including direct inhibition of apoptosis signal–regulating kinase 1 (Stetler et al., 2008).

Hsp27 overexpression also prevented bim mRNA translation and Bim protein up-regulation during excitotoxic injury in cortical and cerebellar granule neurons. These results suggest the possible importance of Bim regulation by Hsp27 as a broad neuroprotective mechanism, since oxidative stress and excitotoxicity are implicated in multiple neurologic insults and neurodegenerative disorders. Of note, transient and sublethal ischemia (preconditioning) is able to up-regulate Hsps without inducing neuronal death, contributing to tolerance against ischemic brain injury (Liu et al., 2009). The described mechanism by which enhanced Hsp27 levels represses bim mRNA translation potentially expands the number and type of posttranscriptional mechanisms by which ischemic preconditioning reduces Bim protein levels and protects against neuronal death (Meller et al., 2006).

In conclusion, our study describes a new role for Hsp27 levels in the posttranscriptional regulation of Bim during neuronal death induced by oxidative stress (summarized in Figure 6). This involves repression of bim translation by Hsp27 binding to the bim-3′UTR. This mechanism contributes a novel mechanism explaining the neuroprotective activity of Hsp27 and may provide novel therapeutic approaches to prevent neuronal death in diverse neurodegenerative conditions.

FIGURE 6:

Hsp27 posttranscriptionally regulates Bim levels during neuronal death induced by oxidative stress. H₂O₂ treatment induces Bim transcription and neuronal death by the modulation of redox-sensitive pathways such as AKT and JNKs, which leads to the activation of transcription factors that regulate the bim promoter such as FOXO3. H₂O₂ also reduces the levels of the antiapoptotic protein Hsp27. Up-regulation of Hsp27 levels during oxidative stress does not affect bim mRNA transcription or processing; however, it represses its translation by modulation of its 3′UTR. This mechanism contributes a novel mechanism for the neuroprotective activity of Hsp27.

MATERIALS AND METHODS

Materials

Fetal calf serum and MEM were obtained from Invitrogen (Bio Sciences, Dublin, Ireland). Hydrogen peroxide, glutamate, glycine, Hoechst 33258, and propidium iodide (PI) were from Fluka (Sigma-Aldrich, Arklow, Ireland). Epoxomicin was obtained from Calbiochem (Merck Biosciences, Nottingham, United Kingdom). FITC-VAD-FMK was obtained from Promega (Medical Supply, Dublin, Ireland).

Preparation of primary neuron cultures

CGNs were isolated from postnatal day 7 rat (Sprague Dawley) or mouse (C57BL/6) pups as described previously (Ward et al., 2000). CGNs were plated on poly-l-lysine–coated glass coverslips, glass Willco dishes, and six- and 24-well plates at 1 × 10⁶ cells/ml and maintained at 37°C in a humidified atmosphere of 5% CO₂/95% air. CGNs were used for oxidative stress experiments after 6–7 d in culture. Oxidative stress was induced by H₂O₂ treatment (final concentration 25–50 μM). H₂O₂ is an endogenously produced ROS, and previous evidence showed that its exogenous H₂O₂ administration is a model of oxidative stress–induced neuronal death (Davila and Torres-Aleman, 2008). For excitotoxicity experiments, CGNs were used after 9 d in culture and treated for 10 min with glutamate (100 μM)/glycine (10 μM) in experimental buffer (120 mM NaCl, 3.5 mM KCl, 0.4 mM KH₂PO₄, 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid [HEPES], 5 mM NaHCO₃, 1.2 mM Na₂SO₄, 1.2 mM CaCl₂, 1.2 mM MgCl₂, and 15 mM glucose; pH 7.4). As a control, cells were treated with experimental buffer only. Mouse neocortical neurons were used for specific excitotoxicity experiments as outlined in Results and isolated from embryonic day 16 pups as described previously (Concannon et al., 2010). They were used after 9 d in culture and treated 5 min with NMDA (100 μM)/glycine (10 μM) in experimental buffer. All animal work was performed with ethical approval from the Royal College of Surgeons in Ireland under licenses granted to the authors by the Irish Department of Health and Children.

Plasmids, siRNA, and transfection

The firefly luciferase reporter plasmid containing 0.8 kb of the bim promoter sequence along with similar constructs with mutated FOXO3-binding sites were kindly provided by Eric Lam (Imperial College London, London, United Kingdom). The pmax–green fluorescent protein (GFP) construct was obtained from Lonza (Basel, Switzerland). The pRL-TK (TK-Renilla luciferase) was a kind gift of B. M. Burgering (University Medical Center, Utrecht, Netherlands). pNeo-Hsp27 and pNeo-Control were kindly provided by Caoimhin Concannon (Royal College of Surgeons in Ireland, Dublin, Ireland). The bim-3′UTR mRNA sequence was divided into two different luciferase reporter plasmids (termed here 1 and 2). These constructs were obtained from GeneCopoeia (LabOmics, Nivelles, Belgium). The siRNAs targeting the homologue of Hsp27 in rat (Hsp25) and the control sequence were obtained from Sigma-Aldrich. For luciferase assays, CGNs were transfected using the calcium phosphate technique at 7 d in vitro (Weisova et al., 2009). For coimmunoprecipitation and siRNA transfection experiments, CGNs were electroporated before plating using an electroporator (Rat Neuron Nucleofector Kit; Lonza) and treated 5 d after electroporation. Mouse neocortical neurons (Figure 5C) were transfected at 5 d in vitro using Lipofectamine 2000 (Invitrogen) as per manufacturer's instructions.

Western blotting

Western blotting was performed as described (Weisova et al., 2009). Neurons were removed from the plates by washing once with ice-cold phosphate-buffered saline (PBS) and lysed with PIK buffer (1% NP-40, 150 mM NaCl, 20 mM Tris, pH 7.4, 10% glycerol, 1 mM Cl₂Ca, 1 mM Cl₂Mg, 400 μM sodium vanadate, 0.2 mM phenylmethylsulfonyl fluoride, 1 μg/ml leupeptin, 1 μg/ml aprotinin, and 0.1% phosphatase inhibitor cocktails I and II [Sigma-Aldrich]). Blots were probed with rabbit polyclonal antibodies to Bim (H-191; 1:500) and IκB-α (C-21; 1:2000) from Santa Cruz Biotechnology (Dublin, Ireland); phospho (Thr-183/Thr-185) JNK (1:1000), total JNK (1:2000), phospho (Ser-473) AKT (1:2000), total AKT (1:2000), phospho (Thr-32) FOXO3 (1:1000), and cleaved caspase-3 (Asp-175; 1:1000), all from Cell Signaling Technology (Danvers, MA); total FOXO3 (1:2000) from Millipore (Carrigtwohill, Ireland); Hsp25 (1:3000) from Enzo Life Science (Exeter, United Kingdom); and Argonaute-2 (1:1000) and LAMP1 (1:1000) from Abcam (Cambridge, United Kingdom). Blots were probed also with mouse monoclonal antibodies to Hsp27 (1:3000) from Enzo Life Science, β-actin (clone DM 1A; 1:10000) from Sigma-Aldrich, cytochrome C (A-8; 1:2000) and RPL10a (JK-16; 1:1000) from Santa Cruz Biotechnology, and Neu-N (clone A60; 1:2000) from Millipore. Horseradish peroxidase–conjugated secondary antibodies (1:10,000) from Thermo Scientific (Fisher, Dublin, Ireland) were detected using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) and imaged using a FujiFilm LAS-3000 imaging system (FujiFilm, Dublin, Ireland). To normalize for protein load, membranes were reblotted (Re-Blot; Millipore) and incubated with an appropriate control antibody. Levels of the protein under study were expressed relative to protein load in each lane as determined by appropriate control protein content. Different exposures of each blot were collected to ensure linearity and to match control levels for quantification. Densitometric analysis was performed using ImageJ software (National Institutes of Health, Bethesda, MD). A representative blot is shown from a total of at least three independent experiments except when indicated.

Cytochrome C release detection

CGNs cultured on six-well plates were lysated 4 h after H₂O₂ (37.5μM) or vehicle treatment in plasma membrane permeabilization buffer (50 μg/ml digitonin, 80 mM KCl in PBS), and cytochrome C mitochondrial release was detected as described previously (Waterhouse et al., 2004). IκB-α protein was used as marker of the cytosolic fraction, and LAMP1 protein was used as marker of the membrane fraction, including mitochondrial membranes.

Hoechst staining of nuclear chromatin

Neurons cultured on 24-well plates were stained live with Hoechst 33258 and PI at a final concentration of 0.1 μg/ml, respectively. After incubation for 20 min, nuclear morphology was observed in living cells (PI⁻) using an Eclipse TE 300 inverted microscope (Nikon, Amstelveen, Netherlands) and a 20× dry objective. For each treatment, cells were analyzed for apoptotic morphology in three subfields of each culture. Condensed nuclei were scored as pyknotic nuclei and expressed as a percentage of the total population. All experiments were performed in triplicate dishes in at least three independent experiments. Neurons transfected with pNeo-Hsp27 or siRNA-Hsp27 were also cotransfected with a pMax-GFP construct in a 5:1 ratio, with the apoptotic morphology analyzed only in GFP⁺ cells.

Terminal deoxynucleotide transferase–mediated dUTP nick-end labeling and caspase activation

Terminal deoxynucleotide transferase–mediated dUTP nick-end labeling (TUNEL) of CGN cultures was performed using the Dead End Fluorometric TUNEL kit (Promega) as per manufacturer's instructions. Detection of activated caspases in CGN cultures was performed with an in situ fluorescent marker, CaspACE FITC-VAD-FMK (Promega), as per manufacturer's instructions. Photographs were taken using an Eclipse TE 300 inverted microscope (Nikon) and a 20× dry objective.

Luciferase assays

Neurons were transfected with a reporter construct in which the expression of the luciferase firefly was regulated by the wild-type bim promoter or promoter with mutated FOXO3-binding sites (see Plasmids, siRNA, and transfection). In other experiments, neurons were transfected with a reporter plasmid in which the bim-3′UTR mRNA sequence was cloned downstream of the luciferase gene, whose expression is controlled by a constitutively active promoter. For luciferase assays, cells were lysed in passive lysis buffer, and luciferase activity was analyzed using a luminometer and dual luciferase assay kit, according to the manufacturer (Promega). Transfections were performed in triplicate dishes, and luciferase counts (RLUs) were normalized using coexpression of Renilla luciferase. Background luminescence was subtracted, and luciferase activity was expressed as fold increase with respect to the control.

Quantitative real-time RT-PCR analysis

Total RNA was extracted using the RNeasy Mini Kit (Qiagen, West Sussex, United Kingdom). First-strand cDNA synthesis was performed using 2 μg of total RNA as template and reverse transcribed using Superscript II (Invitrogen) primed with 50 pmol of random hexamers. Real-time qPCR was performed using the Light-Cycler 2.0 (Roche, West Sussex, United Kingdom) and the QuantiTech SYBR green PCR kit (Qiagen). Sense and antisense primers were as follows: GCAGTCTCAGGAGGAACCTG and AGTGCCTTCTCCAGACCAGA for Bim; AGAATGGCTGGCTATGGATG and GCCAATTGGGTTCACTGTCT for ATF-4, and GGGAAATCGTGCGTGACATT and TGCCACAGGATTCCATACCC for β-actin. The data were analyzed using the LightCycler software 4.0 (Roche), with all samples normalized to β-actin.

RNA coimmunoprecipitation

Neurons cultured on Petri dishes (100 mm) were transfected or electroporated with the plasmid pNeo-Hsp27 or its control vector and 3 or 4 d later treated with H₂O₂ (30–37.5 μM). After 4 h, cells were homogenized in buffer A (50 mM Tris HCl, pH 7.4, 300 mM NaCl, 5 mM MgCl₂, 0.1% NP-40, and protease and recombinant RNase inhibitors) for Hsp27/Hsp25 immunoprecipitation or buffer B (10 mM HEPES, pH 7.4, 150 mM KCl, 5 mM MgCl₂, 1% NP-40, 0.5 mM dithiothreitol, 100 mg/ml cycloheximide, and protease and recombinant RNase inhibitors) for RPL10a immunoprecipitation and specific buffer for the polysome extraction (Heiman et al., 2008). In both cases, lysates were centrifuged, and 400 μl of the supernatants was incubated overnight at 4°C with 5 μg of the corresponding antibodies against Hsp27/Hsp25 (both from Enzo Life Science) or RPL10. Supernatants were also immunoprecipitated with Argonaute-2 or an unspecific serum as positive and negative controls for the immunoprecipitation, respectively. Protein A agarose beads (Santa Cruz Biotechnology) were added, mixed, and incubated for 1 h at 4°C and then centrifuged, and the supernatant was removed. The pellet was washed and processed with TRIzol (Invitrogen) for RNA and protein extraction. First-strand c-DNA synthesis was performed using 2 μg of the RNA, and real-time qPCR was performed using bim primers (as described previously). bim mRNA levels were expressed as fold increase with respect to the control. bim mRNA levels were below detection levels in the control experiments.

Gene-targeted mice

The generation and genotyping of puma^−/− and bim^−/− mice were previously described. (Bouillet et al., 1999; Villunger et al., 2003). The puma^−/− mice were generated on an inbred C57BL/6 background, using C57BL/6-derived embryonic stem cells. The bim^−/− mice were originally generated on a mixed C57BL/6 × 129SV genetic background, using 129SV-derived embryonic stem cells, but had been backcrossed for >12 generations onto the C57BL/6 background. Wild-type animals of the same genetic background (C57BL/6) were used in this study

Statistical analysis

Data are expressed as mean ± SE. Differences among groups were analyzed by one-way analysis of variance followed by Tukey test. Comparison between two groups was performed with the t test. p < 0.05 was considered statistically significant.

Abbreviations used:

ARE

AU-rich sequence element

BH

Bcl-2 homology domain

BH3-only proteins

BH 3–containing proteins

CGN

cerebellar granule neuron

Hsp27

heat shock protein 27

RBP

RNA-binding protein

ROS

reactive oxygen species

TRAP

translating ribo­some affinity purification

RLU

relative light unit

UTR

untranslated region