Endoplasmic reticulum stress-mediated induction of SESTRIN 2 potentiates cell survival

Upregulation of SESTRIN 2 (SESN2) has been reported in response to diverse cellular stresses. In this study we demonstrate SESTRIN 2 induction following endoplasmic reticulum (ER) stress. ER stress-induced increases in SESTRIN 2 expression were dependent on both PERK and IRE1/XBP1 arms of the unfolded protein response (UPR). SESTRIN 2 induction, post ER stress, was responsible for mTORC1 inactivation and contributed to autophagy induction. Conversely, knockdown of SESTRIN 2 prolonged mTORC1 signaling, repressed autophagy and increased ER stress-induced cell death. Unexpectedly, the increase in ER stress-induced cell death was not linked to autophagy inhibition. Analysis of UPR pathways identified prolonged eIF2α, ATF4 and CHOP signaling in SESTRIN 2 knockdown cells following ER stress. SESTRIN 2 regulation enables UPR derived signals to indirectly control mTORC1 activity shutting down protein translation thus preventing further exacerbation of ER stress.


INTRODUCTION
The mTORC1 complex is an important molecular sensor linking the cellular environment to key processes including proliferation and autophagy [1,2]. Careful regulation of mTORC1 signaling is required to maintain cellular homeostasis especially following exposure to stress. Suppression of mTORC1 activity simultaneously reduces protein translation while stimulating pro-survival autophagy. SESTRIN 2 (SESN2), a member of the highly conserved stress-inducible Sestrin family, has been linked to mTORC1 regulation via AMP-activated protein kinase (AMPK/PRKA) [3]. SESTRIN 2 interaction with AMPK both activates and tethers AMPK to TSC2 leading to AMPK-mediated phosphorylation of TSC2 and subsequent inhibition of mTORC1 [3].
Recently endoplasmic reticulum (ER) stress has been linked to SESTRIN 2 regulation [5,6]. ER stress can be induced by various physiological or chemical insults and is characterised by an accumulation of unfolded proteins within the lumen of the ER. Three ER transmembrane receptors IRE1 (ERN1), PERK (EIF2AK3) and ATF6 monitor the health of the ER and upon activation of ER stress act in concert to mediate the unfolded protein response (UPR). The UPR is an integrated signal transduction pathway, which aims to re-establish ER homeostasis through various adaptive mechanisms [7]. IRE1, PERK and ATF6 promote adaptation to proteinfolding stress through the transcriptional upregulation of key ER chaperones, enhanced ER-associated degradation (ERAD) and inhibition of general protein translation. However, when ER stress is persistent, the UPR machinery can activate cell death pathways [8,9]. Recent studies have identified ER stress as a mechanism through which SESTRIN 2 expression can be regulated. The proteasome inhibitor, Bortezomib (Btz), and HIV protease inhibitor, www.impactjournals.com/oncotarget Nelfinavir, have both been demonstrated to induce SESTRIN 2 expression as a consequence of ER stress [6]. Likewise, Park and colleagues reported ER stressmediated regulation of SESTRIN 2 in hepatocytes treated with palmitic acid [5]. How SESTRIN 2 expression is induced by ER stress is not fully understood although the PERK arm of the UPR appears to be required. Downstream targets of PERK, ATF4 and c/EBPβ (CEBPB), have been reported to regulate ER stressinduced SESTRIN 2 expression [5,6].
In this study we demonstrate that induction of ER stress following treatment with pharmacological agents or chemotherapeutics such as Methotrexate (Mtx) lead to a P53-independent increase in SESTRIN 2. Our data indicates both IRE1/XBP1 and PERK branches of the UPR contribute to ER stress-induced SESTRIN 2 expression. Knockdown of SESTRIN 2 reduced stress driven autophagy and enhanced cell death. Intriguingly, our results suggest that enhanced stress-induced cell death observed in SESTRIN 2 knockdown cells is due to an exacerbation of ER stress rather than a decrease in autophagy.

ER stress induces SESTRIN 2 independent of P53
Treatment of MCF7 cells with Thapsigargin (Tg) and Brefeldin A (BFA), two classical inducers of ER stress, was optimized to select doses which would permit Figure 1: Induction of ER stress leads to upregulation of SESTRIN 2 expression independent of P53. MCF7 cells were treated with A. 0.5 μg/ml of BFA or 1 μM of Tg for the indicated time and viability assessed by propidium iodide (PI) uptake. Error bars represent the mean ± SD. MCF7 cells were treated with 0.5 μg/ml of Tg B. or 1 μM of BFA C. for the indicated time and lysates imunoblotted for SESTRIN 2, PERK, XBP1s. ACTIN was used as a loading control. D. MCF7 cells were treated with 1 μM of Tg alone or in combination with Act D (1 μg/ml) for the indicated time and lysates immunoblotted for SESTRIN 2 and ACTIN. E. MCF7 cells were treated with 0.5 μg/ml BFA, 1 μM Tg or 50 μM Etoposide (Etop) for 24 h after which lysates were immunoblotted for SESTRIN 2, Ser15 p-P53 and total P53. F. HCC1806 cells were treated with 1 μM Tg for the indicated time and cell lysates were then immunoblotted for SESTRIN 2 and ACTIN. Results are representative of at least 3 independent experiments. examination of pro-survival and pro-death signaling over time (0.5 µg/ml BFA and 1 µM Tg) ( Figure 1A). Tg and BFA, induced robust expression of SESTRIN 2 and activated the UPR as demonstrated by splicing of XBP1 and PERK phosphorylation (as determined by PERK upshift) ( Figure 1B-1C). Similarly, SESTRIN 2 expression was induced by exposure to the ER stress inducer Dithiothreitol (DTT) (Supplemental Figure 1E). ER stress driven increases in SESTRIN 2 were transcriptionally mediated. Treatment with Tg increased SESTRIN 2 mRNA levels (Supplemental Figure 1A) while, addition of Actinomycin D (Act D) prevented Tg-mediated SESTRIN 2 induction ( Figure 1D). SESTRIN 2 regulation has been demonstrated to occur via P53 dependent and independent mechanisms [3,5,6]. Phosphorylation of P53 (Ser15), while readily detectable in cells treated with Etoposide (Etop), was not observed following either Tg or BFA treatment although a clear increase in the levels of SESTRIN 2 was evident ( Figure 1E). Likewise, HCT116 P53 +/+ and P53 -/cells displayed Tg-induced SESTRIN 2 up-regulation, irrespective of P53 status. While, Etopinduced SESTRIN 2 expression was only observed in  Figure 1B-1C). Increased SESTRIN 2 expression was also detected in cells lacking wild-type P53 (HCC1806, K562) post Tg treatment ( Figure 1F, Supplemental Figure 1D). Collectively, these data demonstrate the ability of ER stress to induce SESTRIN 2 expression in a P53 independent manner.

UPR mediators contribute to ER stress-mediated induction of SESTRIN 2
To understand how ER stress leads to an increase in SESTRIN 2 expression, we determined the contribution of each signaling arm of the UPR. siRNA knockdown of ATF6, while successful, did not inhibit Tg-induced increase in SESTRIN 2 (Supplemental Figure 2A). Addition of a PERK inhibitor alone, IRE1 inhibitor alone or a combination of both reduced Tg and BFA-mediated induction of SESTRIN 2 ( Figure 2A). The functionality of each inhibitor was verified by examining their effect on downstream targets. Treatment with the PERK inhibitor, clearly attenuated ER stress-induced expression of ATF4, while inclusion of the IRE1 inhibitor blocked splicing of XBP1 (Figure 2A). To verify the role of PERK and determine if IRE1-mediated SESTRIN 2 induction is a result of IRE1/XBP1 signaling we examined SESTRIN 2 protein levels following ER stress in Xbp1 -/and Perk -/mouse embryonic fibroblasts (MEF) cells. PERK- mediated translational inhibition is essential for cell survival following ER stress. Consequently, knockout of Perk renders cells exquisitely sensitive to ER stressinduced death [10]. For this reason Perk +/+ and Perk -/-MEF cells were treated with a lower dose (25nM) of Tg as described previously [11]. Perk -/-MEF cells displayed lower SESTRIN 2 induction upon ER stress than their Perk +/+ counterparts ( Figure 2B). In agreement with previous studies, targeting the downstream PERK target ATF4 also reduced Tg-mediated induction of SESTRIN 2 expression ( Figure 2B-2C). Likewise, Xbp1 -/-MEFs and MCF7 cells transfected with XBP1 siRNA displayed a significant reduction in Tg-induced SESTRIN 2 induction underscoring a previously un-described role for XBP1 in SESTRIN 2 regulation ( Figure 2D-2E).

SESTRIN 2 knockdown modulates ER stressinduced autophagy and cell death responses
To determine the relevance of SESTRIN 2 induction, MCF7 cells were transfected with siRNA against SESTRIN 2 and subjected to ER stress. Knockdown was confirmed post Tg or BFA treatment by Western blotting ( Figure 3A-3B). The outcome of SESTRIN 2 knockdown on ER stress-induced cell death was determined by PI uptake. Cells transfected with SESTRIN 2 siRNA displayed significantly higher ER stress-induced cell death compared to their control counterparts ( Figure  3C-3D). This was especially evident at the earlier timepoints, suggesting a role for SESTRIN 2 in the initial protective UPR response. Autophagy, an important prosurvival process, is induced following exposure to a range of stresses including ER stress [12]. Previous work has linked SESTRIN 2 to autophagy induction and cellular survival following genotoxic stress but little is known about its contribution during other types of stress. MTOR dephosphorylation and enhanced LC3-I (MAP1LC3) to LC3-II conversion was evident in cells treated with BFA or Tg suggesting autophagy induction (Supplemental Figure 3A-3B). Examination of GFP-LC3 distribution identified a punctate pattern indicative of autophagy induction upon ER stress (Supplemental Figure 3C). Addition of the autophagy inhibitor Spautin-1 (Spa-1) lowered the number of cells displaying punctate GFP-LC3 distribution following BFA treatment (Supplemental Figure 3C). To determine if SESTRIN 2 contributed to ER stress-induced autophagy, autophagic flux was examined in control and SESTRIN 2 siRNA transfected MCF7 cells. Enhanced LC3-I to II conversion after co-treatment with chloroquine (CQ) was clearly evident in Tg-treated control siRNA cells. However, in SESTRIN 2 siRNA-transfected cells significantly less LC3-I to II conversion was detected in cells treated with both Tg and CQ ( Figure 3E). Likewise, SESTRIN 2 knockdown inhibited autophagy (as determined by puncta formation) in BFA treated GFP-LC3 MCF7 (Supplemental Figure 3D).

The chemotherapeutic Methotrexate induces ER stress and SESTRIN 2 expression via PERK and IRE1 signals
Since our results indicated SESTRIN 2 expression is enhanced by ER stress we sought to determine if commonly used chemotherapeutics also induced SESTRIN 2 expression. Treatment of MCF7 and HCC1806 (triple negative P53 -/breast cancer cell line) cells with a range of chemotherapeutics detected elevated SESTRIN 2 expression following treatment with the antifolate drug Mtx and the proteosomal inhibitor Btz ( Figure  4A-4B). While treatment with Btz clearly induced ER stress (Supplemental Figure 4A) as has previously been reported [13], the induction of ER stress responses by Mtx has not yet been documented. RT-PCR confirmed Mtx, similar to Tg treatment, induced transcriptional upregulation of SESTRIN 2 (Supplemental Figure 1F). Following treatment with Mtx, activation of both the PERK and IRE1 arms of the UPR was observed in MCF7 and HCC1806 cells confirming Mtx treatment, in addition to enhancing SESTRIN 2 expression, also triggered ER stress ( Figure 4C-4D).
To determine the relevance of the UPR to Mtxmediated SESTRIN 2 induction HCC1806 cells were pre-treated with PERK or IRE1 inhibitors prior to Mtx treatment and SESTRIN 2 expression was examined. Addition of either inhibitor reduced SESTRIN 2 expression, with the greatest effect observed in cells pretreated with IRE1 inhibitor ( Figure 4E). Knockdown of XBP1 in HCC1806 cells showed a similar pattern, with reduced Mtx-mediated SESTRIN 2 induction in XBP1 siRNA versus control siRNA-transfected cells ( Figure  4F). Likewise in Btz-treated HCC1806 cells knockdown of XBP1 reduced Btz driven expression of SESTRIN 2 (Supplemental Figure 4B). Collectively, our data suggest a model where treatment with chemotherapeutics such as Mtx and Btz induce ER stress, which in turn, via PERK and IRE1/XBP1 signals, drives SESTRIN 2 expression

SESTRIN 2 knockdown enhances Mtx-induced death in P53 -/-HCC1806 cells
Since our earlier results, using specific ER stressinducing agents, indicated knockdown of SESTRIN 2 expression enhanced ER stress-induced cell death, we examined the outcome of SESTRIN 2 knockdown on Mtx-induced death in HCC1806 cells. Examination of SESTRIN 2 expression following Mtx treatment confirmed knockdown over the 72 h time-course ( Figure 5A). Similar to Tg and BFA, reduction of SESTRIN 2 expression via siRNA increased cell death in response to Mtx ( Figure  5B). Owing to the lack of P53, HCC1806 cells are relatively resistant to Mtx-induced death. To examine the long term outcome of SESTRIN 2 knockdown, HCC1806 cells were transfected with SESTRIN 2 siRNA, followed by Mtx treatment for 72 h. After drug removal cells were left for an additional 10 days and colony formation was evaluated by crystal violet staining. While colonies were abundant in cells transfected with the control siRNA, the same was not true for cells transfected with the SESTRIN 2 siRNA, where few if any colonies were visible ( Figure  5C). This data indicates that SESTRIN 2 knockdown enhances Mtx-induced death in HCC1806 cells in both the short and long term.

SESTRIN 2 knockdown reduces Mtx-induced autophagy independently of its pro-survival function
To understand how SESTRIN 2 knockdown enhanced Mtx-induced cell death we examined the contribution of autophagy. Following Mtx treatment, both MTOR dephosphorylation and enhanced LC3-I to II conversion was evident in HCC1806 cells ( Figure 6A-6B). Knockdown of SESTRIN 2, while enhancing Mtxinduced cell death ( Figure 5A-5B), also decreased Mtxinduced autophagy ( Figure 6C). Likewise, SESTRIN 2 induction in Btz-treated HCC1806 cells was associated with enhanced autophagy (Supplemental Figure 4C). Similarly to Mtx-treated cells knockdown of SESTRIN 2 both inhibited Btz-induced autophagy and enhanced cell death (Supplemental Figure 4D-4E). To determine if the reduced autophagy observed in Mtx-treated SESTRIN 2 knockdown cells directly impacted upon cell death, an inhibitor of autophagy, Spa-1, was employed. Pretreatment of cells with Spa-1, while clearly reducing autophagy, did not increase Mtx-induced cell death ( Figure 6D-6E). In addition, knockdown of BECLIN1 (BECN1) failed to sensitize cells to Mtx-induced death ( Figure 6F-6G). Collectively, these observations suggest inhibition of autophagy is unlikely to contribute to the enhanced death observed in Mtx-treated HCC1806 cells following SESTRIN 2 knockdown.

SESTRIN 2 controls ER homeostasis by regulating MTOR activation in response to Mtx
Recently it has been reported that SESTRIN 2 expression is critical for inhibition of protein translation during ER stress [5]. It was demonstrated that SESTRIN 2 deficiency resulted in persistent protein synthesis under ER stress conditions resulting in enhanced ER stress-induced death [5]. To examine if this was also the case in our model, levels of MTOR phosphorylation were examined in HCC1806 cells following Mtx treatment. Clear inactivation, as determined by MTOR dephosphorylation, was observed in Mtx and Btztreated HCC1806 cells 24 h post treatment ( Figure 6A, Supplemental Figure 4F). Knockdown of SESTRIN 2 abrogated MTOR dephosphorylation and prolonged P70 S6K (RPS6KB1) phosphorylation following Mtx treatment ( Figure 7A). Lower SESTRIN 2 expression has been previously linked to an increase in UPR signals and in particular enhanced activation of the PERK arm of the UPR [5]. To determine whether SESTRIN 2 knockdown potentiates PERK-mediated signaling following Mtxinduced ER stress we examined ATF4, CHOP (DDIT3) and EIF2α (EIF2S1) phosphorylation in Mtx-treated control versus SESTRIN 2 siRNA transfected cells. Prolonged activation of PERK signaling was observed in Mtx-treated cells transfected with SESTRIN 2 siRNA, suggesting SESTRIN 2 knockdown potentiates Mtxinduced ER stress in HCC1806 cells ( Figure 7A-7B). Similar results were also observed with Tg demonstrating this is due to increased ER stress per se and not specifically Mtx treatment (Supplemental Figure 5A). To ascertain if the enhancement in ER stress-induced markers observed in SESTRIN 2 siRNA cells was a direct result of sustained MTOR activity we utilised an inhibitor of MTOR, rapamycin. Treatment of SESTRIN 2 siRNA HCC1806 cells with rapamycin reduced both Mtx-induced mTORC1 activity, as determined by phospho-P70 S6K, and also reduced SESTRIN 2 knockdown-mediated enhancement of PERK signaling ( Figure 7B).

DISCUSSION
Microarray studies within our laboratory had previously identified SESTRIN 2 as an ER stress regulated gene [14]. In this study, we demonstrate ER stress regulation of SESTRIN 2 expression independent of P53. How SESTRIN 2 is induced following ER stress is not entirely understood. Park and colleagues reported PERK-mediated SESTRIN 2 induction via c/EBPβ, while Bruning et al found SESTRIN 2 induction to be ATF4 dependent [5,6]. In our system ER stress-mediated upregulation of SESTRIN 2 was dependent on both IRE1 and PERK signals. Reduced SESTRIN 2 expression was evident in both Perk and Xbp1 deficient MEF cells upon exposure to ER stress. Knockdown of XBP1 in MCF7 and HCC1806 cells attenuated Tg and Mtx-mediated induction of SESTRIN 2. This data suggests enhancement of SESTRIN 2 expression upon ER stress requires both IRE1/XBP1 and PERK-mediated signals. Such a cooperative induction is unsurprising as ER stress sensors rarely function in a singular distinct way but rather in a highly co-ordinated manner. For example, we observed decreased XBP1 splicing in both Perk -/-MEFs and cells treated with PERK inhibitor following ER stress, indicating PERK signaling is required for robust activation of the XBP1. Indeed reduced Xbp1 mRNA expression has been previously reported in Perk -/-MEFs following tunicamycin treatment [15]. Based on these observations it seems likely that reducing/blocking one branch of the UPR can have far reaching effects on other branches of this signaling pathway.
To examine the therapeutic relevance of targeting SESTRIN 2 we examined the ability of a panel of commonly used chemotherapeutics to induce SESTRIN 2 expression in MCF7 cells and the P53 null triple negative breast cancer cell line HCC1806. Treatment with Mtx and or Btz potently induced SESTRIN 2 expression in both cell lines. As the triple negative HCC1806 cell line is devoid of P53 we chose to use this cell line for all further experiments. Examination of both Mtx and Btz-treated cells revealed, that in addition to inducing SESTRIN 2, both drugs triggered ER stress. It has been well established that Btz is a potent inducer of ER stress [13]. However, a role for Mtx as a novel inducer of ER stress has not previously been characterised. Again, as observed with Tg and BFA treatment of MCF7 cells, knockdown of SESTRIN 2 expression decreased Mtxinduced autophagy and enhanced cell death in HCC1806 cells. Since autophagy is predominantly considered a prosurvival process we reasoned enhanced Mtx-induced death in SESTRIN 2 knockdown cells is a direct consequence of lowered autophagy. However, studies using the autophagy inhibitor Spautin-1 or knockdown of BECLIN1 argued against this. Following siRNA knockdown of SESTRIN 2 Mtx-induced dephosphorylation of MTOR was reduced. Associated with this we observed an enhancement of PERK-mediated signaling indicating a potentiation of ER stress. Inhibition of MTOR signaling in Mtx-treated SESTRIN 2 siRNA cells, by addition of rapamycin, lowered PERK driven signals as determined by attenuated EIF2α phosphorylation and CHOP induction. Unexpectedly, addition of rapamycin to control and SESTRIN 2 siRNA cells treated with Mtx also lowered SESTRIN 2 expression ( Figure 7B). This reduction may be a consequence of reduced protein translation by rapamycin-mediated inhibition of mTORC1.
In agreement with our findings, exacerbation of ER stress as a consequence of SESTRIN 2 deficiency has recently been reported. Park and colleagues demonstrated sustained mTORC1 activity leading to elevated ATF4 and CHOP expression in Sestrin 2 deficient cells upon induction of ER stress [5]. Increased ATF4 and CHOP expression has been demonstrated to antagonize translational inhibition by inducing genes involved in protein synthesis triggering oxidative stress and ultimately cell death [16]. Enhancement of SESTRIN 2 expression, by ER stress, blocks protein translation via mTORC1 inactivation promoting cellular adaption and restoration of ER homeostasis. This supports the hypothesis that SESTRIN 2 expression protects breast cancer cells from Mtx-induced cell death by means of inhibiting protein synthesis through inhibition of mTORC1 rather than autophagy induction (Figure 8).
This study highlights the pro-survival role of SESTRIN 2 and demonstrates targeting of SESTRIN 2 in breast cancer cells represents a mechanism to potentiate ER stress-induced cell death. SESTRIN 2 induction has been reported in other cancer models including skin cancer where it was also linked cell survival [17]. These results, together with the results of our study, highlight the importance SESTRIN 2 induction during cell stress.

Flow cytometry
Cells were collected by trypsinization and subsequent centrifugation, resuspended in ice-cold PBS containing 0.7 µg/ml of propidium iodide (Sigma #P4170) and analysed using FACSCalibur flow cytometer (Becton Dickinson).

RNA extraction and cDNA synthesis
Total RNA was isolated from cells using TRI Reagent (Invitrogen) according to the manufactures protocol. To synthesize cDNA, 2 μg of RNA was subjected to DNase treatment and incubated for 15 min at room temperature. The RNA was then reverse transcribed into cDNA using Superscript III first strand RT-PCR system and random hexamers (Invitrogen). The cDNA product was subjected to PCR using the forward primer 5′-CAGAGGGCACAGGAAAGAAG-3′ and the reverse primer 5′-GAACTAGGATTCGGGCAACA-3′ for the detection of human SESTRIN 2. GAPDH was used as an endogenous control using the forward primer 5′-ACCACAGTCCATGCCATC-3′ and reverse 5′-TCCACCACCTGTTGCTG-3′.

Clonogenic assay
HCC1806 cells were transfected with SESTRIN 2 and non-coding control siRNAs as described above. 36 h later they were treated with 20 µM of Mtx for 72 h. After the treatment, media was removed and cells were left to recover for 10 days. Colonies were stained with 2% crystal violet in 20% methanol.

Statistical analysis
Cell death and data is expressed as mean ± SD for three independent experiments. Real Time data is expressed as mean ± SEM for three independent experiments. Differences between the treatment groups were assessed using Graphpad's Two-tailed unpaired student's t-tests. The values with * p < 0.05 is considered statistically significant, ** p < 0.01, *** p < 0.001.