Main

An autophagic pathway is a cellular defense process involving the bulk degradation of cellular contents by autophagosomes/lysosomes during starvation or viral infection.1 However, over production of autophagic vesicles interferes with normal membrane trafficking in the cell and causes autophagic cell death with the degradation of cellular organs. These autophagic vesicles are also shown during neurodegeneration such as Huntington disease (HD) with expanded polyglutamine (polyQ) aggregation and Parkinson's disease with α-synuclein aggregation.2, 3

The molecular mechanism of autophagy has been investigated in yeast and mammalian cells.4, 5 The Atg5-dependent conversion of microtubule-associated protein 1 (MAP1) light chain 3 (LC3), the mammalian homolog of Atg8, from LC3-I (free form) to LC3-II (membrane-bound form) is a key step in the induction of autophagy in mammalian cells.5, 6 ProLC3 is processed to LC3-I by cleavage at the C-terminal region7 immediately after synthesis, and then LC3-I is targeted to the isolated membrane (LC3-II), dependent on the Atg5–Atg12 complex throughout the course of membrane elongation.5, 8 Atg16 homo-oligomers crosslink multiple Atg5–Atg12 conjugates to form Atg5–Atg12–Atg16 complex.9 During these processes, LC3-I and Atg12 are activated by Atg7.

Recently, an Atg7-deficient mouse model showed the accumulation of ubiquitinated proteins and abnormal cellular organs in the liver.10 These results strongly suggest that autophagy as well as the ubiquitin/proteasome system is involved in protein degradation in addition to the degradation of cellular organs. Rapamycin (Rap), which stimulates autophagy,11 attenuates polyQ aggregate formation and decreases polyQ toxicity in cell and mouse models of HD.12, 13 This suggests that autophagy seems to contribute to the cellular defense system, by degrading malfolded protein aggregates, as well as cellular dysfunction. However, little is known about the activation of the autophagic pathway in cells expressing malfolded protein aggregates.

The relationship between the intracellular aggregation of malfolded proteins and endoplasmic reticulum (ER) stress has been intensively examined.14 Excess amounts of unfolded or malfolded proteins in the ER are retrotranslocated to the cytoplasm and degraded by the ER-associated ubiquitine/proteasome degradation (ERAD) system.15 However, if the amount of unfolded and malfolded proteins exceeds the capacity of the ERAD system, the proteins start to aggregate in the ER and trigger ER-stress-mediated cell death with caspase-12 activation, a caspase specifically localized at ER.16, 17 The cytoplasmic aggregates of malfolded proteins such as polyQ also stimulate ER stress signals and induce ER-stress-mediated cell death with caspase-12 activation in mouse cells, presumably by the accumulation of unfolded proteins in the ER due to the inhibition of ERAD and retrotranslocation.18, 19

Here, we show that Rap inhibits the polyQ-induced ER-stress-mediated cell death with caspase-12 activation. These results strongly suggest the close relationship between autophagy formation and ER stress signals in cells expressing cytoplasmic polyQ aggregates. In the present study, we used the expanded polyglutamine 72 repeat (polyQ72) and examined the relationship between degradation of intracellular aggregation of malfolded proteins by autophagy and ER stress signals.

Results

Rap inhibits polyQ-induced ER-stress-mediated cell death

PolyQ72 forms cytoplasmic aggregates in C2C5 cells 48 h after transfection of pEGFP (green fluorescent protein)-72CAG and induces ER-stress-mediated cell death with caspase-12 activation during aggregation, while polyQ11 does not form aggregates nor induce the caspase-12 activation.18 To clarify the relation between autophagy and ER-stress-mediated cell death in cells showing polyQ72 aggregates, we examined the effect of Rap, a stimulator of autophagy, and 3-methyladenine (3MA), an inhibitor of autophagy, on polyQ72-induced caspase-12 activation and DNA fragmentation. Rap decreased the amount of insoluble polyQ72 and inhibited caspase-12 activation, about 50% decrease of the amount of the active form fragment, and DNA fragmentation, whereas 3MA increased the amount of insoluble polyQ72 and stimulated caspase-12 activation, about 2.0-fold increase of the amount of the active form fragment, and DNA fragmentation (Figure 1a).

Figure 1
figure 1

Effect of Rap and 3MA on polyQ11 and polyQ72 aggregation and polyQ aggregate-induced ER-stress-mediated cell death. (a) Effect of Rap and 3MA on polyQ11 and polyQ72 aggregate-induced caspase-12 activation and DNA fragmentation. C2C5 cells were transfected with pEGFP-11CAG or pEGFP-72CAG and incubated in the presence or absence of Rap (10 μg/ml) or 3MA (10 mM), and the caspase-12 activation and cell death were examined by immunoblot analysis or DNA fragmentation. lanes 1, 5; untreated cells, lanes 2, 6; polyQ72 or polyQ11 transfected cells, respectively, lanes 3, 7; Rap treated with polyQ72 or polyQ11 transfected cells, respectively, lanes 4, 8; 3MA treated with polyQ72 or polyQ11 transfected cells, respectively. M, DNA size makers. (b) Immunofluorescence images of polyQ11 or polyQ72 aggregation (green) and caspase-12 activation (red) in the presence or absence of Rap or 3MA. Arrowheads indicate polyQ72 aggregates and arrows indicate the cells colocalizing caspase-12 activation and polyQ72 aggregates. Scale bar, 25 μm. (c) Effects of Rap and 3MA on polyQ11 and polyQ72 aggregation, polyQ11- and polyQ72-induced caspase-12 activation, and polyQ11- and polyQ72-induced cell death. Percentages of cells showing polyQ aggregates, caspase-12 activation, or apoptotic feature in cells expressing polyQ11 or polyQ72 were determined in the presence or absence of Rap or 3MA as described in Materials and Methods. Error bars represent S.E.M. of the percentage of cells with polyQ aggregates (left panel), cells with caspase-12 activation (middle panel) or cells showing apoptotic future (right panel) within cells expressing polyQ

Full size image

Rap decreased the number of cells showing polyQ72 aggregates, antiactivated caspase-12 (anti-m12D341) immunoreactivity, and apoptotic feature in cells expressing polyQ72, whereas 3MA increased their cell numbers (Figure 1b and c). However, 3MA did not stimulate polyQ11 aggregation and did not induce caspase-12 activation and DNA fragmentation in cells expressing polyQ11 (Figure 1a–c). These results suggest that autophagy formation is involved in the inhibition of ER-stress-mediated cell death by degrading polyQ72 aggregates.

PolyQ-induced Atg5-dependent LC3 conversion

PolyQ72 induced the conversion of LC3-I (18 kDa) to -II (16 kDa), a key molecule involved in autophagosome formation, in C2C5 cells, but polyQ11 did not (Figure 2a). Deficiency of Atg5, a key molecule in the conversion of LC3-I to -II, inhibited the polyQ72-induced LC3 conversion (Figure 2b). During autophagosome formation, LC3-I was converted to LC3-II, which is tightly associated with the autophagosomal membrane. LC3-I is diffusely stained by anti-LC3 in the cytoplasm of cells with no stress, while LC3-II shows bright vesicular and granular anti-LC3 immunostaining with increased fluorescence intensity.7, 20, 21 Anti-LC3-positive granules were detected in the cytoplasm of cells showing polyQ72 aggregates but not observed in cells expressing polyQ11 (Figure 2c) and in Atg5-deficient mouse embryonic fibroblasts (Atg5-/-MEF) cells showing polyQ72 aggregates (Figure 2d). Thus, granular anti-LC3 immunostaining is available for detection of the polyQ72-induced Atg5-dependent LC3 conversion leading to autophagy. ER stress inducers, such as tunicamycin (Tm), brefeldin A (Bre), and thapsigargin (Thap), as well as starvation, also induced LC3 conversion and anti-LC3 granular immunoreactivities (see Supplemental Figure 1a and b).

Figure 2
figure 2

PolyQ72-induced Atg5-dependent LC3 conversion. (a) PolyQ-induced conversion of LC3-I to -II. After pEGFP-72CAG or pEGFP-11CAG was transfected into C2C5 cells, the conversion of LC3-I to -II was examined in a time-dependent manner by immunoblot analysis. The lower panel indicates the relative amounts of LC3-I (closed bars) an LC3-II (open bar) showing densitometric analyses. Error bars represent S.E.M. of the LC3-I and -II percentage of density. (b) Effect of Atg5-deficiency on polyQ72-induced LC3 conversion. After pEGFP-72CAG was transfected into Atg5+/+ and Atg5−/− MEF cells, the LC3 conversion was examined by immunoblot analysis. The lower panel indicates the relative amounts of LC3-I (closed bars) an LC3-II (open bar) showing densitometric analyses. Error bars represent S.E.M. of the LC3-I and -II percentage of density. (c) Effect of polyQ72 aggregates on the LC3 distribution. LC3 distribution in nontransfected cells, cells expressing EGFP-polyQ72, and -polyQ11 were examined. (Upper panels) EGFP-labeled; green, (middle panels) anti-LC3, rhodamine; red, (lower panels) merged images. Anti-LC3 granular immunoreactivity was detected in cells expressing polyQ72 aggregates. The polyQ aggregate colocalized with anti-LC3. (d) Effect of Atg5-deficiency on the LC3 distribution. Atg5+/+ and Atg5−/− MEF cells were transfected with pEGFP-72CAG (EGFP-labeled: green), and immunofluorescence images of LC3 distribution was detected by immunostaining using anti-LC3 (rhodamine: red; Hoechst 33342: blue). Arrowheads indicate EGFP-polyQ72 aggregates. In contrast with Atg5+/+ MEF cells, the colocalization is not detected in the ATG5−/− MEF cells. (e) Double immunostaining with anti-LC3, anti-c-Jun-p, anti-eIF2α-p, or anti-m12D341 (rhodamine: red) and anti-huntingtin (EM48) (FITC: green) in the brain tissues from transgenic R6/2 mice, a mouse model of HD. Anti-LC3 granular immunoreactivity and anti-m12D341, anti-c-Jun-p, anti-eIF2α-p immunoreactivity were detected in the HD mouse brain (HDtg) but not in the WT mouse brain. Arrows indicate immunoreactive cells. The numbers indicate the population of immunoreactive cells. Scale bars, 25 μm

Full size image

PolyQ aggregates are also detected in the brain of R6/2 mouse, an HD model mouse expressing a truncated form of huntingtin with polyQ144.22 We examined the relationship between LC3 conversion and ER stress in the brain of R6/2 mouse by immunostaining (Figure 2e). Anti-LC3 granular immunoreactivity was detected in 23.2±2.8% of cells expressing polyQ144 aggregates in the brain of R6/2. In addition, caspase-12 activation, a sign of ER stress cell death, was also detected in 5.1±1.2% cells showing polyQ144 aggregates in the R6/2 mouse brain, but undetectable in the wild-type (WT) mouse brain. Furthermore, c-Jun phosphorylation, and eucaryotic translation initiation factor 2 α (eIF2α) phosphorylation, responsive signals to ER stress, were positive in 65.6±8.3, and 29.5±7.7% of cells showing polyQ aggregates, respectively.

PolyQ72 aggregates-induced LC3 conversion with lysosomal turnover of LC3-II

A cellular level of LC3-II does not always show the stimulation of LC3 conversion because autophagosomal LC3-II is degraded by lysosomal hydrolases after formation of autolysosomes; the inhibition of the fusion of autophagosome with lysosome as well as the stimulation of LC3 conversion increases the level of LC3-II. LC3-II significantly accumulates in the presence of E64d, a membrane-permeable inhibitor of cathepsins B, H, and L, and pepstatin A, an inhibitor of cathepsins D and E, under starvation conditions.23 Lysosomal turnover of LC3-II reflects autophagic activity. We examined the effect of E64d and pepstatin A on the level of polyQ72 and the polyQ72-induced LC3-II. The polyQ72, but not polyQ11, accumulated in the insoluble fraction and polyQ72-induced LC3-II also accumulated (Figure 3a). E64d and pepstatin A increased the number of cells showing polyQ72 aggregates, however, did not stimulate polyQ11 aggregation (Figure 3b and c). The increased level of LC3-II induced by polyQ72 was due to the stimulation of the LC3 conversion but not due to the inhibition of the fusion of autophagosome with lysosome. Thus, polyQ72 and polyQ72-induced LC3-II in the autophagosome were degraded in the autolysosome. The cells degradated polyQ72 aggregates in the autolysosome by promoting the LC3 conversion and increasing the autophagic activity.

Figure 3
figure 3

Effect of E64d and pepstatin A on the level of polyQ72 and polyQ72-induced LC3 conversion. (a) C2C5 cells were transfected with pEGFP-11CAG or pEGFP-72CAG and incubated in the presence or absence of E64d (10 μg/ml) and pepstatin A (10 μg/ml), then LC3 conversion and accumulation of polyQ were detected by immunoblot analysis. (b) Immunofluorescence images of polyQ11 or polyQ72 aggregation (green) in the presence or absence of E64d and pepstatin A. Arrows indicate EGFP-polyQ72 aggregates. Scale bar, 25 μm. (c) Effects of E64d and pepstatin A on polyQ11 and polyQ72 aggregation. Percentages of cells showing polyQ aggregates in cells expressing EGFP-polyQ11 or EGFP-polyQ72 were determined in the presence or absence of E64d and pepstatin A as described in Materials and Methods. Error bars represent S.E.M. of the percentage of cells with polyQ aggregates within cells expressing polyQ

Full size image

PolyQ-induced LC3 conversion via PERK/eIF2α-phosphorylation

LC3-I is targeted to the isolated membrane (LC3-II), dependent on the Atg5-Atg12-Atg16 complex, and LC3-I and Atg12 are activated by Atg7 during these process.5, 8, 9 On the other hand, ER stress regulates eIF2α phosphorylation via activation of protein kinase regulated by RNA (PKR)-like ER kinase (PERK).24 We examined whether polyQ72 induces the upregulation of Atg5, Atg7, Atg12, and Atg16 mRNAs via eIF2α phosphorylation. In contrast with Atg5, Atg7, and Atg16 mRNAs, Atg12 as well as CHOP mRNA was upregulated by polyQ72 and their upregulation was inhibited by eIF2αA/A mutation, a knock-in to replace a phosphorylatable Ser51 with Ala51 (Figure 4a).25 Moreover, thapsgargin induced the upregulation of Atg12 mRNA as well as CHOP mRNA and their upregulation was inhibited by eIF2αA/A mutation (see Supplemental Figure 1c). We examined the effect of dominant-negative (DN)-PERK and an eIF2αA/A mutation on polyQ72-induced LC3 conversion (Figure 4b and c). PolyQ72-induced LC3 conversion, C/EBP homologous protein (CHOP), and Atg12 upregulation were inhibited by DN-PERK (Figure 4b) and the eIF2αA/A mutation (Figure 4c). Furthermore, short interfering RNA (siRNA) of Atg12 inhibited polyQ72-induced upregulation of LC3 conversion without inhibiting the eIF2α phosphorylation (Figure 4d). In contrast with polyQ-induced LC3 conversion, starvation- and Rap-induced LC3 conversion were inhibited by the eIF2αA/A mutation (see Supplemental Figure 2b and d), but not inhibited by DN-PERK (see Supplemental Figure 2a and c). Thus, polyQ72 induced the Atg 12-dependent conversion of LC3-I to -II via ER-stress-mediated PERK/eIF2α phosphorylation.

Figure 4
figure 4

Involvement of PERK/eIF2α phosphorylation in polyQ72-induced LC3 conversion. (a) PolyQ72-induced upregulation of mRNA of Atg5, 7, 12, and 16 was examined by RT-PCR and involvement of eIF2α phosphorylation in the up-regulation of Atg genes. C2C5 cells (left panels), eIF2αS/S and eIF2αA/A MEF cells (right panels) were transfected with pEGFP-72CAG and subjected to the RT-PCR analysis. (b–c) Effect of DN-PERK (b) and the eIF2α A/A mutation (c) on polyQ-induced LC3 conversion. C2C5 (WT) cells and DN-PERK cells (b) or eIF2αS/S (WT) and eIF2αA/A (mutant) MEF cells (c) were transfected with pEGFP-72CAG. LC3 conversion, eIF2α phosphorylation were detected by immunoblot analysis. (d) Inhibition of LC3 conversion by silencing of Atg12. C2C5 cells with silencing of control siRNA or Atg12 were transfected with pEGFP-72CAG, and LC3 conversion, eIF2α phosphorylation, expression of Atg 12 were detected by immunoblot analysis

Full size image

Effect of eIF2α dephosphorylation and Atg5 deficiency on polyQ aggregation and polyQ-induced caspase-12 activation

We examined involvement of eIF2α phosphorylation in polyQ72 aggregation and ER-stress-mediated cell death in cells showing polyQ72 aggregates. Caspase-12 was activated at ER of cells showing polyQ72 aggregates and then, activated caspase-12 became punctuated in apoptotic cells as shown previously (Figure 5a).18, 26 The eIF2α A/A mutation increased the population of the antiactivated caspase-12-positive cells in MEF cell expressing polyQ72 (3.4±1.1% of eIF2α S/S cells and 11.1±2.0% of eIF2α A/A cells) (Figure 5a, b) and also increased the number of cells showing polyQ72 aggregates (7.8±2.0% of eIF2α S/S cells and 30.4±3.5% of eIF2α A/A cells) (Figure 5a and c), and apoptotic feature (4.2±1.7% of eIF2α S/S cells and 11.5±0.9% of eIF2α A/A cells) (Figure 5d). Furthermore, the eIF2α A/A mutation increased the amount of insoluble polyQ72 and stimulated the polyQ72-induced caspase-12 activation (Figure 5f). Thus, in addition to polyQ72-induced caspase-12 activation, eIF2α phosphorylation inhibited the polyQ72 aggregation. However, the population of cells with caspase-12 activation was almost the same, about 30%, in eIF2αS/S and eIF2αA/A MEF cells showing polyQ72 aggregates; caspase-12 was activated in 31.6±3.3% of eIF2α S/S cells and 35.3± 2.5% of eIF2α A/A cells showing polyQ72 aggregates (Figure 5e).

Figure 5
figure 5

Involvement of eIF2α phosphorylation in polyQ72 aggregation and polyQ72-induced caspase-12 activation. (a) The effects of eIF2α dephosphorylation on polyQ72 aggregation and polyQ72-induced caspase-12 activation. eIF2αS/S (WT) MEF cells and eIF2αA/A (mutant) MEF cells were transfected with pEGFP-72CAG (EGFP-labeled: green), and immunofluorescence images of polyQ72 aggregation (green) and caspase-12 activation (red) (upper panels) were captured. Arrowheads indicate EGFP-polyQ72 aggregates and arrows indicate the cells colocalizing caspase-12 activation and EGFP-polyQ72 aggregates. Caspase-12 activation (lower panels) was detected by immunostaining using anti-m12D341 (rhodamine: red; Hoechst 33342: blue). Arrowheads indicate EGFP-polyQ72 aggregates. eIF2αA/A(mutant) MEF cells showed the apoptoti feautures with cell shrinkage and caspase-12 activation, but eIF2αS/S (WT) MEF cells did not. Scale bars, 25 μm. (b) Percentages of cells showing caspase-12 activation in cells expressing EGFP-polyQ72. Error bars represent S.E.M. of the percentage of cells with caspase-12 activation within cells expressing polyQ72. (c) Percentages of cells showing polyQ aggregates in cells expressing EGFP-polyQ72. Error bars represent S.E.M. of the percentage of cells with polyQ aggregates within cells expressing polyQ72. (d) Percentages of cells showing apoptotic feature in cells expressing EGFP-polyQ72. Error bars represent S.E.M. of the percentage of cells showing apoptotic feature within cells expressing polyQ72. (e) Percentages of cells showing caspase-12 activation in cells expressing EGFP-polyQ72 aggregate. Error bars represent S.E.M. of the percentage of cells with caspase-12 activation within cells expressing polyQ72 aggregate. The percentages were determined by counting 100–200 cells expressing EGFP-polyQ72 at 48 h after transfection. The values are averages of the percentages of the number of cells obtained in three experiments. (f) The effects of eIF2α dephosphorylation on polyQ72 aggregate-induced caspase-12 activation. eIF2αS/S (WT) MEF cells and eIF2αA/A (mutant) MEF cells were transfected with pEGFP-72CAG, and the caspase-12 activation were examined by immunoblot analysis

Full size image

We examined whether Atg5–Atg12–Atg16 complex-dependent LC3 conversion inhibits polyQ72 aggregate-induced caspase-12 activation (Figure 6). Atg5 deficiency increased the accumulation of the insoluble form of polyQ72 (Figure 6a) and the number of cells showing polyQ72 aggregates (9.4±2.2% of Atg5+/+ cells and 38.6±3.2% of Atg5−/− cells) (Figure 6b). Atg5 deficiency also increased the population of cells with caspase-12 activation (4.9±0.4% of Atg5+/+ cells and 10.9±0.7% of Atg5−/− cells) and apoptotic features (3.9±1.6% of Atg5+/+ cells and 11.6±0.9% of Atg5−/− cells) in cells expressing polyQ72 (Figure 6a, c and d), and increased the amount of insoluble polyQ72 and stimulated caspase-12 activation although it did not affect the amount of eIF2α phosphorylation and CHOP upregulation (Figure 6f). However, the population of cells with caspase-12 was almost the same, about 30%, in the Atg5+/+ cells and Atg5−/− cells showing polyQ72 aggregates; caspase-12 was activated in 32.9±2.1% of Atg5+/+ cells and 29.2±4.3% of Atg5−/− cells showing polyQ72 aggregates (Figure 6e).

Figure 6
figure 6

Effect of Atg5-deficiency on polyQ72 aggregation, and polyQ72- or Thap-induced caspase-12 activation. (a) Effect of Atg5-deficiency on polyQ72 aggregation and polyQ72-induced caspase-12 activation. Atg5+/+ MEF cells and Atg5−/− MEF cells were transfected with pEGFP-72CAG (EGFP-labeled: green), and immunofluorescence images of polyQ72 aggregation (green) and caspase-12 activation (red) (upper panels) were captured. Arrowheads indicate EGFP-polyQ72 aggregates and arrows indicate the cells colocalizing caspase-12 activation and EGFP-polyQ72 aggregates. Caspase-12 activation (lower panels) was detected by immunostaining using anti-m12D341 (rhodamine: red; Hoechst 33342: blue). Arrowheads indicate EGFP-polyQ72 aggregates. Atg5−/− MEF cells showed the apoptotic features with cell shrinkage and caspase-12 activation, but Atg5+/+ MEF cells did not. Scale bars, 25 μm. (b) Percentages of cells showing polyQ aggregates in cells expressing EGFP-polyQ72. Error bars represent S.E.M. of the percentage of cells with polyQ aggregates within cells expressing polyQ72. (c) Percentages of cells showing caspase-12 activation in cells expressing EGFP-polyQ72. Error bars represent S.E.M. of the percentage of cells with caspase-12 activation within cells expressing polyQ72. (d) Percentages of cells showing apoptotic feature in cells expressing EGFP-polyQ72. Error bars represent S.E.M. of the percentage of cells showing apoptotic feature within cells expressing polyQ72. (e) Percentages of cells showing caspase-12 activation in cells expressing EGFP-polyQ72 aggregate. Error bars represent S.E.M. of the percentage of cells with caspase-12 activation within cells expressing polyQ72 aggregate. The percentages were determined by counting 100–200 cells expressing EGFP-polyQ72 at 48 h after transfection. The values are averages of the percentages of the number of cells obtained in three experiments. (f) The effects of Atg5-deficiency on polyQ72 aggregate-induced caspase-12 activation. Atg5+/+ MEF cells and Atg5−/− MEF cells were transfected with pEGFP-72CAG, and the caspase-12 activation and CHOP upregulation, eIF2α phosphorylation were examined by immunoblot analysis

Full size image

Thus, Atg5–Atg12–Atg16 complex-dependent LC3 conversion inhibited ER-stress-mediated cell death with caspase-12 activation in cells expressing polyQ72. These results suggest that Atg5–Atg12–Atg16 complex-dependent LC3 conversion, leading to autophagy, inhibit polyQ72-induced ER-stress-mediated cell death with caspase-12 activation by degrading cytoplasmic polyQ72 aggregates (Figure 7).

Figure 7
figure 7

Involvement of eIF2α phosphorylation in malfolded protein polyQ72-induced LC3 conversion. PolyQ72 aggregates induce ER stress by inhibiting the retrotranslocation and/or ERAD activity.19 The accumulation of malfolded proteins in the ER induces PERK/eIF2α phosphorylation, which induces LC3 conversion via upregulation of Atg12 and probably activation of Atg5–Atg12–Atg16 complex and stimulates autophagy degrading the polyQ72 aggregates. When LC3 conversion is not sufficient to reduce the polyQ72 aggregate, cells may undergo ER-stress-mediated cell death with caspase-12 activation18

Full size image

Discussion

Biological significance of the autophagy formation in cells expressing cytoplasmic polyQ aggregates

In the present study, we showed that Atg5–Atg12–Atg16 complex-dependent LC3 conversion, a key factor for autophagy formation, inhibited the polyQ72-induced ER-stress-mediated cell death (Figures 2, 6). These results suggest that Atg5–Atg12-Atg16 complex-dependent LC3 conversion targets cytoplasmic aggregates of polyQ72. The cytoplasmic polyQ aggregates inhibit retrotranslocation of proteins from the ER and degradation of ER proteins in the ERAD system, resulting in the accumulation of malfolded proteins in the ER,19 and induce ER-stress-mediated cell death.18 Cytoplasmic aggregation of polyQ72 induces the accumulation of dysferlin, a protein involving in the Ca2+-dependent membrane fusion process,27 in the ER, resulting in the ER stress cell death (Fujita et al., submitted elsewhere). Thus, it is likely that autophagy may inhibit the ER-stress-mediated cell death by degrading protein aggregates, including polyQ72 aggregates, which cannot be degraded by the ubiquitine/proteasome system (Figure 7). However, when the protein aggregates are not degraded enough, continuous ER stress may produce excess autophagy, which unselectively degrades not only malfolded protein aggregates but also intracellular contents and induces severe neurodegeneration during excessive autophagic cell death. Autophagy may be protective in initial stages of polyQ-induced ER stress,12 but may be deleterious in later stages.2

Degradation of malfolded protein aggregates by autophagy

Cytoplasmic polyQ aggregates can be degraded.28 As large aggregates become difficult for proteasomes to degrade, involvement of other degradation systems such as autophagy have been suggested to degrade polyQ aggregates. Rap-promoted autophagy prevents polyQ-induced cell death by selectively degrading polyQ aggregates (Figure 1).12, 29 These data and our data showing colocalization of LC3 staining with polyQ72 aggregates (Figure 2c) support the possibility that autophagy degrades polyQ aggregates.

However, it is not yet clear how autophagy recognizes malfolded protein aggregates. Atg7-deficient mice show accumulation of ubiquitinated proteins in the liver,10 strongly suggesting that autophagy as well as the ubiquitin/proteasome system is involved in the degradation of ubiquitinated proteins. PolyQ aggregates show antiubiquitin immunoreactivity30 and that Atg5 deficiency stimulates the accumulation of polyQ72 aggregates. Autophagy may preferentially degrade aggregates of ubiquitinated proteins rather than soluble nonubiquitinated proteins. The ubiquitination state and size of the protein aggregate may be a possible determinant for selective protein degradation by autophagy.

Involvement of eIF2α phosphorylation in LC3 conversion

Thus, LC3 conversion leading to autophagy is a cellular defense mechanism against polyQ72-induced ER-stress-mediated cell death. We examined the polyQ72-induced ER stress response signal stimulating LC3 conversion. ER stress stimulators, including Thap as well as polyQ72, induced LC3 conversion (see Supplemental Figure 1) and eIF2α phosphorylation. The polyQ72-induced LC3 conversion was inhibited in cells containing the eIF2α A/A mutation and DN-PERK (Figure 4b and c), strongly suggesting that the PERK/eIF2α pathway, an ER stress response signal, plays an essential role in polyQ72-induced LC3 conversion.

The relationship between autophagy and eIF2α phosphorylation has been shown during starvation-induced autophagy in Saccharomyces cerevisiae and during starvation- and virus-induced autophagy in mammalian cells.31 In yeast, starvation results in the accumulation of uncharged tRNAs to activate GCN2, the sole eIF2α kinase. In mammals, there are at least four eIF2α kinases including GCN2, PKR, PERK, and HRI, which are activated by amino acid starvation, viral infection, ER stress, and heme depletion, respectively.32 Virus induces autophagy via activation of PKR. Thus, it is possible that various stress conditions, including ER stress, that activate eIF2α kinases may have an ability to induce autophagy in mammalian cells by stimulating LC3 conversion.

Recently, it has been postulated the possibility that inhibition of mTOR, a mammalian target of Rap, induces autophagy and reduces the toxicity of polyQ expansions.13 Rap fails to induce eIF2α phosphorylation in yeast with GCN2 defective in tRNA binding.33 In contrast with polyQ72-induced LC3 conversion, Rap- as well as starvation-induced LC3 conversion was not mediated by PERK/eIF2α pathway (see Supplemental Figure 2). Inhibition of mTOR-mediated eIF2α phosphorylation may be an alternative pathway by which polyQ aggregates induce autophagy. Rap inhibits the ER-stress-mediated cell death (unpublished observation) but stimulates the starvation-induced autophagic cell death.34 Thus, eIF2α phosphorylation does not inhibit the all types of stress-induced cell death.

The possible molecular mechanism of eIF2α phosphorylation inducing LC3 conversion

The molecular mechanism, by which eIF2α phosphorylation regulates LC3 conversion is not yet clear. eIF2α phosphorylation blocks the binding of initiator Met-tRNA to the ribosome by inhibiting the turnover of eIF2B and represses translation of most mRNAs. However, ATF4 mRNA, a member of the cAMP response-element-binding (CREB) family, is selectively translated under these conditions because of its small upstream open-reading frame (uORF) within the 5′ untranslated region, called ‘uORF bypath scanning system’.35, 36 ATF4 upregulates the transcription of downstream genes, including CHOP. Atg12, a component of Atg5–Atg12–Atg16 complex, as well as CHOP mRNA was selectively upregulated by polyQ72 via eIF2α phosphorylation (Figure 4a). Thus, one possible explanation is that the eIF2α phosphorylation-dependent selective translation of transcription factors may increase the upregulation of Atg12, resulting in the stimulation of the Atg5–Atg12–Atg16 complex formation followed by the promotion of the conversion of LC3-I to -II.

eIF2α phosphorylation-induced autophagy and inhibition of ER-stress-mediated cell death

Phosphorylation of eIF2α inhibits ER-stress-induced caspase-12 activation and ER-stress-mediated cells death.37, 38 Cells with phosphorylated eIF2α are resistant to oxidative stress39 and inhibition of eIF2α dephosphorylation promotes survival of stressed cells.40 Atg5 deficiency did not inhibit Thap-induced caspase-12 activation (Supplemental Figure 3). Moreover, Rap inhibits the Thap-induced caspase-12 activation via eIF2α phosphorylation (unpublished observation), suggesting that eIF2α phosphorylation may inhibits Thap-induced cell death via preferentially inhibiting the caspase-12 activation. Thus, it may be possible that Rap inhibits the polyQ72-induced cell death by both pathways; stimulation of autophagy and inhibition of caspase-12 activation.

However, Atg5 deficiency increased the population of cells with caspase-12 activation and apoptotic features in cells expressing polyQ72 although it did not affect the eIF2α phosphorylation and CHOP upregulation (Figure 6). eIF2α A/A mutation increased the number of cells showing polyQ72 aggregates, resulting in increase of the number of cells with caspase-12 activation (Figure 5), but did not increase the population of cells with caspase-12 activation in cells showing polyQ72 aggregates. These results suggest that eIF2α phosphorylation preferentially stimulates the LC3 conversion and the degradation of polyQ72 aggregates to inhibit the polyQ72 aggregates-induced cell death rather than the pathway inhibiting the caspase-12 activation.

Very recently, the accumulated mutant α1-antitrypsin Z inhibitors in the ER has been shown to be degraded by autophagy.41 We have also found that Atg5–Atg12-Atg16 complex-dependent LC3 conversion inhibits the accumulation of the mutated dysferlin in ER (Fujita et al., submitted elsewhere). These results suggest the possibility that autophagye/lysosome as well as ubiquitine/proteasome is involved not only in the degradation of cytoplasmic protein aggregates such as polyQ aggregates but also in the degradation of the misfolded and unfolded ER protein in the ER as one of the ERAD degradation system to inhibit the ER-stress-mediated cell death. This possibility remains to be studied.

Recently, it has been shown that interaction between antiapoptotic protein Bcl-2 and the autophagy protein, Beclin-1, homolog of Atg6, represents a potentially important point of convergence of the apoptotic and autophagic machinery.42, 43, 44 Dissociated Beclin-1 from Bcl-2 mediates the localization of other autophagy proteins to the preautophagosomal membrane and participates in autophagosome formation.45, 46 Thap increased the level of Beclin-1 protein and its dissociation from Bcl-2 (unpublished observation). ER-stress-induced dissociation of the Beclin-1 and Bcl-2 may also play a role in the regulation of the ER-stress-mediated apoptotic cell death and autophagic formation.

Conclusion

The scheme in Figure 7 shows the possible molecular mechanism of polyQ72- and ER-stress-induced LC3 conversion, LC3-I to -II, via activation of the PERK/eIF2α pathway. Atg5–Atg12–Atg16 complex-dependent LC3 conversion, which is stimulated by the activation of the PERK/eIF2α pathway in cells showing polyQ72 aggregates, is the cellular defense system that inhibits ER-stress-mediated cell death by degrading polyQ72 aggregates. Isolated membrane from the ER membrane may be elongated into an autophagosome by LC3-II activated by eIF2α phosphorylation.

Materials and Methods

Cell culture

C2C5 cells were P19 EC cells constitutively expressing c-jun.47 MEFs from Atg5−/− and Atg5+/+ mouse embryos and C2C5 cells were cultured in α-minimum essential medium (α-MEM; Sigma, St Louis, MO, USA) supplemented with 10% fetal calf serum (FCS). MEFs from eIF2αS/S (WT) and eIF2αA/A (mutant) knock-in mouse embryos were cultured in DMEM (GIBCO, Grand Island, NY, USA) supplemented with 10% FCS, 0.1 mM nonessential amino acids solution (GIBCO), 1 × amino acids solution (GIBCO) and 100 U/ml penicillin/streptomycin (GIBCO) at 37°C in a humidified atmosphere of 5% CO2.25

Plasmid construction and establishment of C2C5 cell lines

pEGFP-72CAG and pEGFP-11CAG were prepared as described previously.48 Using restriction enzyme digestion and ligation, we subcloned the myc-tagged PERK(K618A) (DN-PERK)24 into the pcDNA4/TO/myc-His expression vector (Invitrogen, Carlsbad, CA, USA). pcDNA4/TO/myc-His-DN-PERK was transfected into C2C5 cells using the calcium phosphate method,49 and clones expressing DN-PERK were selected by incubation with 50 μg/ml of zeocine and 25 μg/ml blastin (Invitrogen). Cells were cultured in α-MEM supplemented with 10% FCS with 5 μg/ml tetracycline at 37°C in a humidified atmosphere of 5% CO2.

siRNA

siRNAs were designed and created at Nippon EGT Co. (Toyama, Japan). The sequences used were as follows: mouse Atg 12 siRNA, 5′-CCUCGGAACAGUUGUUUAU-3′. Cells were transfected with 10 μg of siRNA using LipofectAMINE per the manufacturer's instructions. In the case of transfectio with complementary DNA and siRNA, DNA were transfected and after 12 h siRNAs were introduced.

Isolation of fragmented DNA

DNA isolation was performed according to the method of Prigent et al.50

Immunoblot analysis

ER stress signals were examined by immunoblot analysis using anti-m12D341, antisera against the cleavage site of mouse caspase-12 at D341,26 anti-CHOP (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA), antitubulin (Sigma), anti-GFP (Boehringer Mannheim, Indianapolis, IN, USA), anti-eIF2α-p (antisera against phosphorylated serine at position 51 of eIF2α; Cell Signaling Technology, Beverly, MA, USA), anti-eIF2α (Cell Signaling Technology), and anti-myc (Santa Cruz Biotechnology Inc.), anti-Atg12 (ZYMED Laboratories, South San Francisco, CA, USA). LC3 conversion from I to II was examined by immunoblot analysis using anti-LC3.20, 23, 51

Cells were treated with Hank's Balanced Salt Solution (HBSS) or incubated with Bre (2 μg/ml), Tm (2 μg/ml), Thap (2 μM), Rap (10 μg/ml) (Sigma), and lysed with phosphate buffered saline (PBS) containing 1% Triton X-100. pEGFP-72CAG and pEGFP-11CAG (5 μg) were transfected using the calcium phosphate method. siRNA transfected cells were transfected with pEGFP-72CAG. Cells were incubated for the indicated period in the presence or absence of Rap (10 μg/ml), 3MA (10 mM), or E64d (10 μg/ml) (Calbiochem, La Jolla, CA, USA) and pepstatin A (10 μg/ml) (ICN Biochemicals Inc., Aurora, OH, USA) then lysed with PBS containing 1% Triton X-100.

After centrifugation at 10 000 × g for 10 min, the cell extracts (50 μg protein) were subjected to SDS polyacrylamide gel (12%) electrophoresis and immunoblot analysis as described previously.18 Insoluble fraction were lysed into SDS sample buffer and subjected to SDS polyacrylamide gel (12%) electrophoresis. Quantification of the images was performed using NIH image (Dr. W Raeband National Institute of Health, Bethesda, MD, USA) programs.

Immunostaining

Cells were transfected with pEGFP-72CAG or pEGFP-11CAG and incubated in the presence or absence of Rap (10 μg/ml), or 3MA (10 mM). After the cells were fixed with 4% paraformaldehyde (PFA) in PBS at the indicated time, they were incubated with anti-LC3 or anti-m12D341 for 24 h at 4°C. They were then incubated with rhodamine-labeled goat anti-rabbit (Leinco Technologies Inc., St Louis, MO, USA) for 1 h at 37°C, and the nuclei were labeled with Hoechst 33342 (Molecular Probes, Eugene, OR, USA) and viewed with a confocal laser-scanning microscope (CSU-10, Yokogawa, Tokyo, Japan). Percentages of cells showing polyQ aggregates, caspase-12 activation, or apoptotic future were determined by counting 100–200 cells expressing EGFP-polyQ72 at 48 h after transfection in the presence or absence of Rap, 3MA, E64d and pepstatin A. The values are averages of the percentages of the number of cells obtained in three experiments.

12-week-old R6/2 transgenic mice22 and B6CBAF1/J WT mice were obtained from Jackson lab. They were anesthetized with diethyl ether and perfused intracardially with 4% PFA in PBS. The brain was removed and incubated overnight in 4% PFA in PBS at 4°C and then soaked in 30% sucrose in PBS at 4°C overnight and embedded in optimal cutting temperature (OCT) compound (Sakura Finetec, Tokyo, Japan) before being frozen. Frozen sections of brain were cut on a cryostat and attached to slides coated with Vectabond reagent (Vector Laboratories, Burlingame, CA, USA) and double stained with antihuntingtin (EM48) (Chemicon International, Temecula, CA, USA) and with anti-eIF2α-p, anti-c-Jun-p (antisera against phosphorylated serine at position 63 of c-Jun; Cell Signaling Technology), anti-LC3 or anti-m12D341. Antihuntingtin reactivity was detected by the incubation with FITC-labeled goat anti-mouse immunoglobulin (Leinco Technologies Inc.) and anti-eIF2α-p, anti-c-Jun-p, anti-LC3 or anti-m12D341 reactivity was detected by he incubation with rhodamine-labeled goat anti-rabbit immunoglobulin for 1 h at 37°C to be viewed with a confocal laser scanning microscope. Cells showing punctuate vesicular anti-LC3 staining were counted as cells showing LC3 conversion.21 Percentages of cells showing caspase-12 activation, c-Jun phosphorylation, eIF2α phosphorylation, and LC3 conversion were determined by counting 100–200 cells. The values are averages of the percentages of the number of cells obtained in three experiments.

RT-PCR

Complementary DNAs were synthesized from total RNA (1 μg) of cells treated with Thap (2 μM) or cells transfected with polyQ72 using reverse transcriptase (Stratagene, La Jolla, CA, USA). cDNAs were subjected to PCR as described in the manual from Perkin-Elmer (Branchburg, NJ, USA) using primers of mouse Atg5; forward primer, 5′-GGAAGAATGACAGATGAC-3′; and reverse primer, 5′-CTTTCAATCTGTTGGCTG-3′, Atg7; forward primer, 5′-ATGGGGGGACCCTGGACTG-3′; and reverse primer, 5′-CAGAACAGTGTCGTCATC-3′, Atg12; forward primer, 5′-ATGTCGGAA GATTCAGAG-3′; and reverse primer, 5′-ATTTCTGGCTCATCCCCA-3′, Atg16; forward primer, 5′-GCGTTCGAGGAGATCATT-3′; and reverse primer, 5′-TTCCTT TGCTGCTTCTGC-3′, CHOP; forward primer, 5′-CAGAAGGAAGTGCATCTT CA-3′; and reverse primer, 5′-TACACTTCCGGAGAGACAGA-3′ and G3PDH (Toyobo, Osaka, Japan). The PCR fragments of the Atg genes were amplified as follows: 1 cycle at 95°C for 2 min, 25 cycles at 95°C for 1 min and 60°C for 1 min, and 1 cycle at 60°C for 7 min. The RT-PCR products were electrophoresed on 2% NuSieve agarose gels (FM Bioproducts, Rockland, ME, USA).