The activation of PERK following oxygen or glucose restriction, contributes to cell adaptation and are thus implicated in the progression maintenance of tumors (Fels and Koumenis, 2006
). Because cells deficient for PERK are acutely sensitive to ER stress, an understanding of the pro-survival mechanisms initiated by PERK activity may reveal therapeutic targets for cancer as well as other UPR-based pathologies. The pro-survival mechanisms that have been described downstream of PERK are adaptive programs that reduce the cytotoxic consequences of the stress that a cell incurs. The current work demonstrates that in addition to the adaptive pathways downstream of PERK, once a cell has reached a critical level of stress, PERK also mediates cell survival through the selective induction of the IAP family proteins. ER stress-induced expression of cIAPs coincides with reduced activation of caspases, suggesting an inhibitory effect of IAPs on the ER stress-induced apoptotic program. Our results confirm previously reported data that suggests IAPs are transcriptionally induced in an Akt-dependent manner (Hu et al., 2004
). Our data demonstrate that IAP mRNA is transiently induced early during the ER stress response. This induction coincides temporally with ER stress-induced Akt activity and is attenuated by Akt inhibition. While the specific transcription factors downstream of Akt have not been identified, NFκB remains a likely candidate. Akt promotes NFκB activity by phosphorylating and activating IκB kinase, leading to NFκB nuclear translocation and transcription of its target genes, which include cIAP1 and cIAP2 (Bellacosa et al., 2004
; Debatin and Krammer, 2004
). While ER stress can induce NFκB transcriptional activity in a PERK-dependent manner (Deng et al., 2004
; Jiang et al., 2003
), the role of Akt activity in ER stress-induced NFκB activation has not been addressed.
Importantly, while IAP transcription does increase following ER stress, our results indicate that increased expression is not necessary for IAP induction. Inhibition of Akt effectively inhibited IAP mRNA accumulation, but did not inhibit protein accumulation suggesting an alternative mechanism contributes to IAP protein accumulation during ER stress. Indeed, our data demonstrates that IAP mRNA is preferentially translated during the ER stress response in a PERK and eIF2α-dependent manner. While phosphorylation of eIF2α is associated with attenuation of protein synthesis, translation of specific transcripts is induced by eIF2α phosphorylation. The mRNAs of the transcription factor ATF4 (Harding et al., 2000a
) and the cationic amino acid transporter Cat-1 (Fernandez et al., 2002b
) contain upstream open reading frames (uORFs) in their 5’-untranslated regions (5’UTR) that repress translation under basal conditions, but allow for efficient translation when eIF2α is phosphorylated (Schroder and Kaufman, 2005
). The 5’UTRs of both human and mouse cIAP1 and cIAP2 contain uORFs, suggesting a potential regulatory mechanism.
Alternatively, translational regulation of IAPs could reflect the presence of putative internal ribosome entry sites (IRES). IRES elements have been reported for cIAP1 (Warnakulasuriyarachchi et al., 2004
) mRNA, suggesting additional mechanisms by which cells upregulate IAPs during ER stress. IRES elements in the cIAP1 mRNA may explain why, in our hands, inhibition of Akt alone upregulates cIAP1 protein expression.
Although reintroduction of cIAP expression into PERK−/− cells during ER stress only partially rescues these cells from apoptosis, it is important to note that PERK mediates cellular adaptation to ER stress through multiple mechanisms. We thus did not expect to completely rescue PERK−/− cells from ER stress-induced apoptosis by reintroducing only one mechanism by which PERK mediates survival.
Interestingly, the expression of the most potent IAP, XIAP is not induced by ER stress. XIAP directly inhibits caspases-3, -7, and –9 (Deveraux et al., 1997
; Scott et al., 2005
; Shiozaki et al., 2003). Although not direct caspase inhibitors, cIAPs are believed function as modulators of XIAP activity through the sequestration of SMAC (Eckelman and Salvesen, 2006
). Our data is consistent with this hypothesis as ER stress-induced expression of cIAP proteins leads to a reduction of XIAP in SMAC immunoprecipitates. Thus, while XIAP protein levels are not induced by ER stress, there is a net increase XIAP through PERK-dependent cIAP induction. The importance of XIAP activity for survival of ER stress is highlighted by the fact that knockdown of XIAP dramatically sensitizes cells to tunicamycin treatment (Hu et al., 2004
); our unpublished results).
The activation of Akt may also directly affect XIAP activity during ER stress as Akt has been demonstrated to phosphorylate and thus stabilize XIAP protein (Dan et al., 2004
). The transient activation of Akt during ER stress may thus regulate cellular levels of XIAP protein, and possibly contribute the loss of XIAP protein observed at later hours of ER stress.
Another possible and intriguing role for cIAP1 and cIAP2 in ER stress may lie in the modulation of death receptor signaling. Signaling downstream of TNFα can be either pro-survival or pro-apoptotic and cIAPs are crucial for pro-survival signaling through the TNFα receptor (Wang et al., 1998a
). Autocrine TNFα pro-apoptotic signaling can occur during ER stress through IRE1 activity (Hu et al., 2006
). However, the role of cIAP1 and cIAP2 in ER stress-induced TNFα signaling has not been addressed. While we failed to detect a change of TNFα levels in the media of fibroblasts treated with tunicamycin (unpublished data), the critical role of cIAPs in protecting tumor cells from TNFα-induced apoptosis is becoming increasingly clear (Bertrand et al., 2008
; Petersen et al., 2007
; Varfolomeev et al., 2007
; Vince et al., 2007
). Given these observations, PERK inhibition, as a therapeutic strategy may not only eliminate the ability of cells to adapt to the tumor microenvironment, but also induce the extrinsic apoptotic pathway in tumor cells that excrete TNFα. Future experiments will be required to determine the role of PERK in maintaining tumor cIAP levels and viability in vivo
A striking finding is that ER stress-mediated activation of Akt is completely dependent on PERK. The precise mechanism whereby Akt is activated during ER stress is under current investigation. We and others have observed that this activation is dependent on PI3-kinase activity as ER stress-induced Akt phosphorylation is ablated by the presence of LY294002 (Hosoi et al., 2007
; Hu et al., 2004
),our unpublished results. We have also observed induction of PI3K activity in in vitro
kinase assays after prior treatment of cells with tunicamycin (our unpublished data). This activation of PI3K is independent of extracellular growth factor signaling, as Akt becomes activated by ER stress in the absence of serum (Hu et al., 2004
); unpublished data).
A recent report suggests that activation of PI3K occurs downstream of the eIF2α kinase PKR, suggesting that PI3K and Akt activation is a property of all eIF2α kinases (Kazemi et al., 2007
). Elucidation of the mechanism by which eIF2α kinases activate PI3-kinase is of great interest as signaling pathways downstream of PI3K bring to light a myriad of new, unexplored mechanisms by which cells respond to stress.