The purpose of this study was to explore the mechanisms of radiation-induced autophagy in caspase-3/7 deficient cells. Our findings revealed that ER stress, mainly through activation of the PERK/eIF2α axis, regulates radiation sensitivity in caspase-3/7 deficient cells and to a much lower extent in WT cells as well. Importantly, we also demonstrate that MFC-7 breast cancer cells, which are relatively resistant to radiation-induced apoptosis because they lack caspase-3, are more sensitive to radiation when an ER stressor, in this case TM, is administered concurrently.
The ability of tumor cells to evade apoptosis is a hallmark of cancer, which relative resistance to apoptosis constitutes an important clinical problem. As an alternative cell death pathway to apoptosis, autophagy is currently an important research target for therapy, mainly because many anti-cancer agents induce autophagy and its implication in tumorigenesis. We previously demonstrated that caspase-3/7 inhibition increased autophagy and subsequently enhanced cancer radiosensitivity in vitro
and in vivo
(Kim et al., 2008c
). Here, we show that ER stress signaling is a mechanism participating in radiation-induced autophagy. Several recent reports have shown that ER stress is associated to autophagy, and mainly as an attempt of cell survival (Bernales et al., 2006
; Ding et al., 2007
; Kouroku et al., 2007
; Ogata et al., 2006
; Yorimitsu et al., 2006
). Although the opposite outcome is not yet well explored, we and others have shown that excessive or prolonged accumulation of stress results in autophagic cell death, especially if apoptosis is unavailable (Cao et al., 2006
; Kim et al., 2008b
; Kim et al., 2008c
; Kim et al., 2006
; Ullman et al., 2008
). Here, we also confirmed that ER stress-induced autophagy would result in radiosensitization in absence of caspase-dependent apoptosis. Indeed, PERK knock-down resulted in more radioresistance in WT and caspase-3/7 DKO cells compared to their respective controls (). Similarly, eIF2α knock-down () or overexpression of eIF2α (S51A) mutant () increased radioresistance in WT and caspase-3/7 DKO cells compared to controls. These data are consistent with a recent study showing that ER stress can result in necrosis-like cell death promoted by autophagy in Bax/Bak DKO cells or Bcl-xL-overexpressing cells that are thus apoptosis defective (Ullman et al., 2008
). We also previously reported that Bax/Bak DKO cells were more radiosensitive through autophagy and that autophagy inhibition using ATG5/Beclin-1 siRNAs induces radioresistance (Kim et al., 2006
). Additionally, our prior studies showed that autophagy modulates radiosensitivity of MDA-MB-231 breast and H460 lung cancer cells (Kim et al., 2008c
; Kim et al., 2006
). Consistently, we here report that, in absence of caspase-3/7, TM increased autophagy by five-fold and enhanced radiosensitivity with a DER of 1.57 (). In scenarios when autophagy leads to cell survival, caspase-3, in addition to its role of apoptosis executioner, may prevent protection by autophagy in order to accelerate cell death. Indeed, one function of autophagy is to eliminate protein aggregates and misfolded proteins in order to limit ER stress response and subsequent apoptosis (Ding et al., 2007
). Thus, one may speculate that absence of caspase-3/7 allows the promotion of autophagy and lead to more survival. However, when stress is excessive or unable to resolve, autophagy becomes a self-destructive process, as illustrated in our study. Nevertheless, autophagy and its cross-talk with apoptosis is complex (Eisenberg-Lerner et al., 2009
) and deserves further investigation for better characterization.
Regarding the signaling of ER stress-induced autophagy, it is still unclear which UPR transducer has a role in mammalian cells. Indeed, Yorimitsu et al. showed that ER stress (induced by dithiothreitol or TM) triggers autophagy in yeast cells, through IRE1-signaling (Yorimitsu et al., 2006
). Consistently, Ogata et al. reported that IRE1 is required for ER stress-induced autophagy in SK-N-SH cells treated with TM or thapsigargin (Ogata et al., 2006
). This is contrasting with our findings, which indicate that mainly PERK knock-down significantly decreased autophagosome formation (), particularly in caspase-3/7 DKO cells. Accordingly, decrease in autophagosome formation correlated with increased radioresistance, suggesting that PERK/eIF2α is required for radiation-induced autophagy ( and ). This discrepancy may suggest a different mechanism for autophagy activation in caspase-3/7 deficient cells as opposed to yeast or SK-N-SH cells. Nevertheless, our findings are consistent with the study by Kouroku et al., which demonstrated that PERK-eIF2α, and not IRE1, induces autophagy by ER stress in MEF and C2C5 cells treated with polyglutamine-aggregate (Kouroku et al., 2007
). Of note, the association between eIF2α- phosphorylation and starvation-induced autophagy was previously reported in yeast and mammalian cells (Talloczy et al., 2002
). As shown in , TM increased P-eIF2α, in WT and caspase-3/7 DKO cells, with a particularly robust induction in the latter. Similarly, P-eIF2α levels were induced by TM in MCF-7/neo cancer cells compared to MCF-7/casp-3 cells (). Therefore, among the two UPR pathways, the specific signaling of ER stress-mediated autophagy appears to vary depending on cell type and stress stimulus.
Interestingly, our study suggests the presence of a spontaneous level of P-PERK and P-eIF2α in caspase-3/7 deficient cells regardless of radiation, as opposed to WT cells (). As we know, PERK mediates its effects by phosphorylating (inactivating) eIF2α (Harding et al., 1999
). In the context of ER stress, inactive P-eIF2α has been shown to reduce ATF4, which mRNA translation is dependent on eIF2α phosphorylation at Ser51 (Scheuner et al., 2001
). In phosphorylated serine absence, ATF4 mRNA is transcribed, but not translated. More recently, Kouroku et al. demonstrated that Ser51-phosphorylation is necessary for LC3 conversion, and thus autophagosome formation (Kouroku et al., 2007
). Consistently, we found decreased autophagosome formation and radiosensitvity using an eIF2α (S51A) mutant in irradiated caspase-3/7 DKO cells (). Conversely, eIF2α overexpression further increased radiation-induced autophagy and radiosensitivity in caspase-3/7 DKO cells. These results support the idea that the enhanced sensitivity of caspase-3/7 deficient cells is mediated by PERK/eIF2α phosphorylation.
It has been previously shown that active caspase-3 cleaves eIF2α at its C-terminal in Saos-2, HeLa, K562, and Jurkat T cells (Marissen et al., 2000
; Satoh et al., 1999
). Therefore, another hypothesis that we tested is whether caspase-3 cleaves eIF2α in our model. To detect the endogenous cleavage of eIF2α, we used two commercially available antibodies: one antibody (Ab-A) recognizing the C-terminus, the other one (Ab-B) recognizing the N-terminus. So, the event of cleavage will eliminate the C-terminus and make the cleaved product be recognized by Ab-B but not Ab-A. We induced active recombinant caspase-3 in WT cells, and found a reduction of uncleaved-eIF2α and total-eIF2α levels (Supplementary Figure 2A
). This is consistent with previous findings that caspase-3 cleaves eIF2α at C-terminus, altering its function (Marissen et al., 2000
; Satoh et al., 1999
). We then found that eIF2α cleavage by caspase-3 is promoted by TM in WT cells and, although to a lesser extent, radiation as well (Supplementary Figure 2B
). Similarly to experiment shown in Supplementary Figure 2B
, we also tested eIF2α cleavage by active caspase-3 after apoptosis induction by radiation and TM in caspase-3/7 DKO cells (Supplementary Figure 2C
). As expected, caspase-3 was not detected in caspase-3/7 DKO cells, and high levels of uncleaved-eIF2α were detected as opposed to WT cells. These results suggest that radiation, similarly to TM, reduces uncleaved-eIF2α levels in WT cells, and not in caspase-3/7 deficient cells. Thus, these data suggest that active caspase-3 is associated with cleavage of eIF2α after exposure to stressors such as radiation and that cleaved-eIF2α could potentially be an additional factor limiting radiation-induced autophagy in WT cells.
Function of ER is disrupted in many diseases, including cancer, thus activating UPR, mainly to resolve cellular stress. In addition, tumor microenvironment is characterized by a “baseline” level of ER stress response that promotes tumor development and metastasis (Lee, 2007
; Romero-Ramirez et al., 2004
). It has previously been shown by immunohistochemistry studies that caspase-7 is downregulated in 85% of colon cancer samples compared with normal mucosa (Palmerini et al., 2001
). Similarly, in a study examining caspase-3 expression in malignant breast tissue samples, it was showed that 75% of breast tumor samples lacked caspase-3 transcript and expression, while remaining samples showed substantial decreases in caspase-3 expression (Devarajan et al., 2002
). These results suggest that defects in caspase-3 or caspase-7 are prevalent in human breast and colon cancer. Therefore, we wanted to translate our findings from caspase-deficient MEFs cells to cancer models that harbor defects in caspase-3. MCF-7 human breast carcinoma cells have been used to study ER stress mechanisms (Delom et al., 2007a
; Delom et al., 2007b
), though not related to autophagy. In our study, we looked at autophagy levels in these cells as well as MCF-7/casp-3 cells following treatment with TM and radiation (). Data showed almost 40% of cells lacking caspase-3 underwent autophagy following combination treatment, as opposed to only 15% of cells expressing caspase-3. As anticipated, absence of caspase-3 also correlated with increased radiosensitization, consistent with our observations in caspase-3/7 DKO cells (). Taken together, these data support the use of an ER stressor, such as TM, to sensitize MCF-7 cancer cells to radiation. Of note, classic agents inducing ER stress such as TM are broadly toxic and subsequently not suitable for use in cancer patients. However, drugs including proteasome inhibitors, selenium, and fatty acid synthase inhibitors were shown to induce ER stress in addition to their specific targeted effects, and, therefore, are more indicated for clinical trials.
The current study provides novel findings that radiation-induced ER stress mediates autophagy. Additionally, our data suggest that caspase-3 has a critical role in modulating the PERK/eIF2α pathway following radiation. To our knowledge, this is the first report of radiation-induced autophagy through UPR mechanisms. Many cancers exhibit multiple deregulations in cell death pathways, allowing for the subsequent promotion of tumor cell survival. As we showed here, there is a potential for novel anticancer strategies to overcome resistant cancer cells with defective apoptosis machinery in order to improve overall therapeutic outcomes.