Liraglutide restores chronic ER stress, autophagy impairments and apoptotic signalling in SH-SY5Y cells

Growing evidence suggests that agonists of glucagon-like peptide (GLP-1) receptor exert neuroprotective and neurorestorative effects across a range of experimental models of neuronal degeneration, and, recently, a pilot clinical trial of Liraglutide in Alzheimer’s disease patients showed improvements in cerebral glucose consumption that signifies disease progression. However, the exact underlying mechanism of action remains unclear. Chronic endoplasmic reticulum (ER) stress has recently emerged as a mechanism for neuronal injury, rendering it a potent therapeutic target for acute and chronic neurodegenerative disorders. Here, we investigate the neuroprotective effects of Liraglutide along with the signalling network against prolong ER stress and autophagy impairments induced by the non-competitive inhibitor of sarco/ER Ca2+-ATPase, thapsigargin. We show that Liraglutide modulates the ER stress response and elicits ER proteostasis and autophagy machinery homeostasis in human SH-SY5Y neuroblastoma cell line. These effects correlate with resolution of hyper-activity of the antioxidant Nrf2 factor and restoration of the impaired cell viability and proliferation. Mechanistically, Liraglutide engages Akt and signal transducer and activator of transcription 3 (STAT3) signalling to favour adaptive responses and shift cell fate from apoptosis to survival under chronic stress conditions in SH-SY5Y cells.


Results
Liraglutide rescues thapsigargin-induced cytotoxicity and cell-growth arrest. First, we elucidated the cytoprotective effect of Liraglutide in the SH-SY5Y human neuroblastoma cells from chronic ER stress by thapsigargin. Thapsigargin is a naturally occurring sesquiterpene lactone that selectively inhibits sarcoplasmic/ER Ca 2+ -ATPase (SERCA), triggering a transient increase in the cytosolic calcium and depleting ER calcium stores 28 . Cells were treated with 0, 10, 100 and 1000 nM of thapsigargin in the presence or absence of 100 nM Liraglutide for 16 h, and processed for XTT, BrdU and LDH assays to assess cell viability, proliferation and cytotoxicity, respectively. One-way ANOVA analysis reveals overall significant differences in cell viability (F (7,442) = 79.59, p ≤ 0.001), proliferation (F (7,355) = 48.98, p ≤ 0.001) and cytotoxicity (F (7,436) = 255.2, p ≤ 0.001).
Scientific REPoRTS | 7: 16158 | DOI:10.1038/s41598-017-16488-x Liraglutide restores impaired STAT3 activity and activates Akt signalling to ameliorate thapsigargin-induced apoptosis. Our findings motivated us to scan for the signalling network that underlies the cross-talk among the different sub-cellular compartments and mediates the neuroprotective/restorative effects of Liraglutide [Figs 6 and 7, Table S1]. Sandwich-based antibody array has revealed that persistent disturbance of ER calcium homeostasis significantly precludes the activating phosphorylation of the extracellular signal-regulated kinase 1/2 (ERK1/2) at threonine 202 (Thr202) and tyrosine 204 (Tyr204) residues (p ≤ 0.05) and of Stat3 at Tyr705 (p ≤ 0.01), as shown in Fig. 6 It is well documented that the majority of pro-death signals emerging from UPR regulate the expression and activity of pro-and anti-apoptotic proteins of BCL-2 family 6,8 . Consistently, we report that persistent ER-localised blocking of calcium flux diminishes the inhibitory phosphorylation of the pro-apoptotic BAD at Ser112 (p ≤ 0.05)

Discussion
Neuronal injury owing to chronic and irremediable ER stress has been increasingly correlated with a range of neurodegenerative disorders while UPR activation and deregulation has been repeatedly found in postmortem brain samples from affected patients and animals of AD, PD, ALS, HD, and experimental stroke [1][2][3][4][5] . Pharmacological and genetic manipulation approaches have unravelled promising mechanisms that link ER stress to neurodegenerative processes. In particular, AD features the accumulation of amyloid-β (A β) peptides in the brain, which underlies neuronal dysfunction and cognitive decline. Aβ primarily perturbs the cellular redox status and abnormally increases the amount of calcium that can be released by the ER, triggering ER stress, mitochondrial dysfunction and thereby neuronal toxicity and astrogliosis in vivo 1,4 . Accordingly, α-Synuclein, the molecular determinant of PD pathobiology, alters the interactions between ER and mitochondria, which triggers an aberrant increase in the mitochondrial calcium content and compromises mitochondrial membrane potential, autophagic function and cellular bioenergetics 35 . Recent studies have further revealed that α-synuclein preferentially accumulates within the ER\microsomes, where it aggregates in toxic oligomeric formations in mouse and human brain with α-synucleinopathy. Those oligomeric forms have been further associated with the onset of chronic ER stress and disease progression 36,37 . UPR and ER calcium dynamics seem to additionally govern neuronal injury progression in HD 38,39 , TBI and following stroke 3 . Therefore, resolution of chronic ER stress can benefit acute and chronic neurodegenerative disorders that feature diverse aetiologies and clinical manifestations 3-5 , whilst Figure 5. Autopphagy impairments along with UPR deregulation up-regulate the nuclear content of the nuclear factor erythroid 2-related factor 2 (Nrf2). Liraglutide restores Nrf2 signalling in the neuroblastoma SH-SY5Y cell line upon irremediable ER stress. Twenty-four hours post seeding, SH-SY5Y cells were serum starved for 8 h and treated with 0 or 100 nM of thapsigargin (TG) for 16 h, in the presence or absence of 100 nM Liraglutide (LIRA). Cells were paraformaldehyde-fixed, immunolabelled for Nrf2 and processed for confocal imaging at 40X magnification, as shown in the representative images [(a)]. Eight pictures were captured per experimental group per experiment for quantification. Image J was used to quantify corrected total cell fluorescence (CTCF) of the nuclear Nrf2 staining [(b)]. Each bar represents mean ± SEM from three independent experiments. Data was analysed by one-and two-way ANOVA, followed by post hoc Bonferroni's multiple comparison t-test ( * p ≤ 0.05 compared to CNTRL; # p ≤ 0.05 compared to the corresponding thapsigargin-treated cells). Scale bars: 50 μm. Figure 6. Chronic disturbance of ER calcium homeostasis induces PARP cleavage, diminishes inhibitory phosphorylation of the pro-apoptotic BAD protein and of the multifunctional GSK3b kinase, and impairs ERK and STAT3 signalling to promote genotoxicity and apoptosis. Liraglutide rescues PARP cleavage and activation of STAT3 and p53 kinases. It additionally induces Akt phosphorylation that relieves BAD and GSK3b activity to promote cell survival in the neuroblastoma SH-SY5Y cell line upon persistent ER stress. Twenty-four hours post seeding, SH-SY5Y cells were serum starved for 8 h and treated with 0 or 100 nM of thapsigargin for 16 h, in the presence or absence of 100 nM Liraglutide. Cells were harvested, and Bcl-2 phosphorylation [(a)] and BID expression [(b)] were determined by western blotting. β-Actin was used as the loading control to all western blot analyses. Each bar represents mean ± SEM from four independent experiments. Data is expressed as fold change to the control (CNTRL; unstressed/untreated conditions). Data was analysed by one-and two-way ANOVA, followed by post hoc Bonferroni's multiple comparison t-test ( * p ≤ 0.05, ** p ≤ 0.01 & *** p ≤ 0.001 compared to CNTRL; # p ≤ 0.05 & ## p ≤ 0.01 compared to the corresponding thapsigargin-treated cells).
prolonged perturbation of ER calcium homeostasis may offer an integrated cellular model to simulate neurodegenerative processes for drug discovery.
GLP-1 analogues, which are currently approved for T2DM treatment, have been repeatedly shown to exert neurotrophic/restorative effects in a range of animal models of AD, PD, ALS, TBI and experimental stroke [11][12][13][14][15][16][17][18][40][41][42][43][44][45][46][47][48] . Importantly, GLP-1 mimetics, such as Liraglutide has rescued the AD-related reduction in cortical activity and energy utilisation 22 , and a recently phase II clinical trial testing Exenatide has impeded PD progression 25 . The underlying biochemical processes are manifold. Incretin mimetics have prevented aberrant apoptosis of (hippocampal and primary cortical and dopaminergic) neurons and SH-SY5Y neuroblastoma cells exposed to hypoxia, excitotoxic insults, neurotoxins (e.g., hydrogen peroxide and oxidopamine) and thapsigargin-induced ER stress 17,18,49,50 . Liraglutide pre-treatment favours cell survival over apoptotic signalling to promote cytoprotection from persistent mitochondria dysfunction in SH-SY5Y cells 21 . Similarly, post-treatment with GLP-1R agonists rescues aberrant cytotoxicity and impaired viability, and further enhances cell survival signalling to protect SH-SY5Y neuroblastoma cells from chronic rotenone-induces oxidative stress 19 . In the present study, we have addressed the neuroprotective/restorative effects of Liraglutide, along with the underlying molecular mechanisms and signalling network after prolong perturbation of ER calcium homeostasis. Consistently, we report that Liraglutide impedes the increase in the number of SH-SY5Y cells with compromised plasma membrane and mitochondrial dysfunction induced by thapsigargin, and promotes cell proliferation in a stressor dose-dependent manner.
Thapsigargin is a specific, almost irreversible inhibitor of the SERCA channel, triggering a transient increase in the cytosolic calcium and depleting ER calcium stores 28 . Calcium depletion in the ER precludes the activity of calcium-dependent chaperones to potentiate the accumulation of unfolded/misfolded proteins within the organelle lumen and thereby ER stress 7 . In response, the cell activates the UPR network that integrates signals about the chronicity and severity of the stress stimuli and culminates into disproportionate activation of PERK, ATF6 and IRE1α signalling to determine cell fate 6,8,51 . Particularly, chronic ER stress augments PERK arm to amplify the transcription and translation of the pro-apoptotic transcription factor CHOP 51 , whilst it suppresses IRE1α signalling 51,52 to possibly attenuate the survival and neurotrophic effects of the downstream spliced X-box binding protein 1(XBP-1) and sensitise the cells to ER stress 8 . It can additionally enhance ATF6 activity to reinforce CHOP expression and thus amplify the apoptotic component of the UPR 8,51 . Consistently, we have demonstrated that chronic thapsigargin treatment up-regulates ATF6 which accumulates in the cell nuclei and signifies its activation, whilst almost abolishes the activating phosphorylation of IRE1α at Ser724. We additionally show that chronic thapsigargin treatment triggers an ectopic expression of CHOP that correlates with calnexin deficiency and ERO1α excess in SH-SY5Y neuroblastoma cells. Calnexin silencing has been previously reported to sensitise cardiomyocytes to ER stress and favour apoptosis through CHOP up-regulation and calcium deregulation 53 . Furthermore, induction of the oxidoreductase ERO1α downstream of CHOP perturbs the ER redox state 54 that in turn stimulates inositol-1,4,5-trisphosphate receptor (IP 3 R)-mediated calcium efflux into cytosol 55 . The latter could influence multiple pathways upstream of the core apoptosis machinery and, most importantly, mitochondrial function 56,57 . In addition to CHOP, persistent ER stress can activate the ER-resident CASP12 to further promote cell suicide. CASP12 provokes a downstream caspase cascade that leads to PARP degradation and therefore programmed cell death initiation [58][59][60] , as reflected in our results too. Intriguingly, Liraglutide co-treatment normalises BiP induction along with ATF6 and IRE1α signalling in thapsigargin-treated SH-SY5Y cells. It additionally mitigates abnormal CHOP expression and CASP12 activity, restores calnexin and ERO1α expression, and alleviates PARP degradation. These biochemical traits further relate to the restoration of SH-SY5Y cell viability and proliferation. Collectively, our findings suggest that GLP-1R activation resolves the induction of UPR effectors and pro-apoptotic mediators and promotes chaperone homeostasis in the ER lumen, which signifies cell proteostasis upon persistent, neuronal ER stress.
Acute SERCA channel inhibition has been shown to preclude autophagosome formation 30 and fusion with lysosomes 31 , resulting into autophagy arrest 30,31 . Autophagy fail along with deregulated UPR seem to drive the imbalance between protein generation and degradation that underlies the onset and progression of neuronal degeneration 1,5,32 . Autophagy is a tightly regulated pathway that allows cells to eliminate harmful or damaged components through catabolism and recycling to maintain nutrient and energy homeostasis. As such, autophagy constitutes a crucial mechanism for preserving structures and functioning of subcellular organelles, including ER and mitochondria, when operates at basal levels, and for cell survival in response to stress 1,5,32 . Several studies have shown that Liraglutide promotes autophagy and thereby cell survival in liver, pancreas and SH-SY5Y cells 19 67 . Similarly, Liraglutide has prevented oxidative stress-induced axonal injury by halting excessive autophagy in retinal ganglion cells 68 . Excessive autophagic flux can exert detrimental effects by aberrantly degrading endogenous inhibitors of apoptosis and Atg components, and lead to cell death. That dual role has been further attributed to autophagy under ER stress conditions 32 . Here, we show that apoptotic ER stress correlates with a substantial decrease in the protein levels of beclin-1, which determines the initiation and formation of phagophore. Suppressed IRE1α and Bcl-2 phosphorylation may underlie beclin-1 impairments following prolonged thapsigargin-treatment in SH-SY5Y cells. Previous studies have revealed that, among the three UPR arms, the induction of pro-survival autophagy after ER stress requires IRE1 signalling 69 . The latter mediates the phosphorylation of Bcl-2, which results in its dissociation from and to the release of beclin-1 70 . In addition, spliced XBP1, lying downstream of IRE1 arm, has been previously shown to bound to the promoter of beclin-1 and induce its transcription 71 . Furthermore, we demonstrate that chronic ER calcium dyshomeostasis culminates into Atg3, Atg7 and LC3 protein deficiency that may result from the ectopic CHOP expression. Indeed, CHOP has been previously shown to limit autophagy through the transcriptional control of a dozen of Atg genes involved in phagophore elongation and maturation into the autophagosome, and thereby to stimulate apoptosis upon persistent ER stress 72,73 . Autophagy dysfunction along with persistent ER stress can further trigger the excess accumulation of the autophagy adaptor protein p62 74 , which contains a KEAP1 binding motif similar to Nrf2 75 . Accumulation of p62 leads to KEAP1 sequestration and inactivation, which, in turn, promotes aberrant nuclear Nrf2 localisation and transcription of Nrf2 target genes 75 . Although the Nrf2 transcription program has been recognised as one of the main antioxidant defensive mechanisms for cytoprotection 33 , that constitutive Nrf2 activation along with autophagy preclusion has been previously shown to drive the hepatic injury observed in Atg7-knockout mice 75 . Accordingly, we report that the deficiency in the "core" Atg components, lying downstream of prolong ER calcium perturbation, correlates with Nrf2 hyperactivaiton, suggesting that regulated Nrf2 activation can benefit cellular defensive responses to stress. Strikingly, Liraglutide co-treatment ameliorates Atg3, Atg7, beclin-1, and LC3 expression impairments and further normalises aberrant Nrf2 and Bcl-2 activity in thapsigargin-treated SH-SY5Y cells. Taken together, our findings highlight that GLP-1R activation promotes ER and autophagic machinery homeostasis that signifies about cell adaptation and survival upon unmitigated, neuronal ER stress.
Mechanistically, Liraglutide restores impaired STAT3 signalling and engages Akt pathway to exert its neuroprotective/restorative effects upon persistent ER stress. STAT3 is a latent transcription factor that is primarily activated by the phosphorylation of a single tyrosine residue, Tyr705, in response to the stimulation of cytokine and growth factor receptors 76,77 , including GLP-1R 78 . Once activated, STAT3 dimerises and enters the nucleus to orchestrate -alone or in cooperation with other factors -transcriptional programs for neuronal/glial survival, proliferation and differentiation 76 . Suppressed STAT3 activity has been previously found in the hippocampus of postmortem brain samples from AD patients and transgenic mice 79 . Chiba et al. have further demonstrated that Aβ inactivates hippocampal JAK2 (Janus kinase 2)/STAT3 axis to provoke the basal forebrain cholinergic dysfunction and spatial working memory deficits in vivo, linking STAT3 pathway to neurodegenerative processes 79 . It has been additionally recognised that ER stress perturbs JAK-signal transducers and STAT3 signalling to mediate acute-phase neuroinflammatory responses 80 and leptin resistance in the brain 81 or to drive apoptosis that underlies suppressed hepatic gluconeogenesis 82 in periphery. Although the mechanistic interplay among UPR, STAT3 signalling and GLP-1R in the context of cell fate and neuronal degeneration remains unknown, STAT3 has been recently emerged as a multifaceted determinant for autophagy 77 . For instance, nuclear STAT3 can transcriptionally induce the anti-apoptotic Bcl-2 expression and consequently inhibit autophagy 83 . However, this phenotype is not prominent on our findings as no changes in total Bcl-2 levels occur following chronic thapsigargin and/ or Liraglutide treatments in SH-SY5Y cells [ Figure S1]. Intriguingly, Atg3 promoter has been recently shown to contain a STAT3-responsive element 84 , which may explain the restorative effects of Liraglutide against the thapsigargin-induced Atg3 deficiency, though this hypothesis requires further investigation.
Akt (also known as protein kinase B -PKB) is a serine/threonine kinase member of the AGC protein kinase family with a profound function in growth, proliferation, intermediate metabolism and cell survival 85 , and a pivotal effector of the anti-apoptotic GLP-1R signalling 21 . In response to the phosphoinositide 3-kinase (PI3K) stimulation, Akt is recruited to the cell membrane 86 where it can undergo phosphorylation at the threonine 308 (Thr308) residue by PDK1 87 and at the serine 473 (Ser473) residue by mTORC2 88 . Phosphorylation of Thr308 site critically determines Akt activation whilst phosphorylation of both aforementioned sites is required for the maximal kinase activity 86,89 . In our study, Liraglutide has significantly increased the phosphorylating levels of Akt at Thr308 that signifies kinase activation. Activated Akt phosphorylates multiple targets in the cytoplasm, nucleus, mitochondria and ER membrane to regulate adaptive responses and cell fate under diverse insults, including ER and oxidative stress and DNA damage 85,[89][90][91] . Among others, Akt phosphorylates and inhibits the death-agonist BAD that becomes rapidly de-phosphorylated upon apoptotic stimuli 89 , as prominent in our findings too. Active (de-phosphorylated) BAD binds to the survival-agonist Bcl-x L or Bcl-2 at the mitochondria that provokes Bax and Bak oligomerisation and perturbs mitochondrial membrane permeabilisation to favour the point-of-no-return of apoptotic cell death 92-94 . However, the restoration of BAD phosphorylation downstream of growth factor signalling raises the mitochondrial threshold for apoptosis that renders the cells less vulnerable to death signals 95 , as evident in our results too.
In addition to BAD, Akt phosphorylates and inhibits the GSK3β, a major protein kinase that drives neurodegenerative processes in AD 96 and neuronal apoptosis following ER stress [97][98][99][100] . Indeed, accumulating evidence from diverse neuronal cell lines, primary neuronal cultures, and ER insults has demonstrated that the UPR abolishes the inhibitory phosphorylation of GSK3β at Ser9 97-100 to promote CHOP expression and switch from pro-survival to pro-death signalling during ER stress 98 , as reflected in our results too. Moreover, it has been previously reported that PERK engages GSK3β to phosphorylate the p53 tumour suppressor protein at Ser315 and Ser376 and favour nuclear export and proteosomal degradation of p53 upon ER stress 101,102 . p53 is transcription factor of which activation serves to organise cellular responses with apoptosis, cell cycle arrest, senescence, DNA repair, cell metabolism, or autophagy depending on the nature and degree of stress insult, environmental context, and cell type. The regulation of p53 is complex and involves post-translational modifications -e.g., phosphorylation and acetylation -at multiple sites that impact its cellular localisation, stability and transcriptional activity 103 . We assessed the phosphorylated levels of p53 at Ser15 that facilitates p53 nuclear accumulation and stabilisation by halting the ability of the E3 ubiquitin-protein ligase Mdm2 to interact with and target p53 for proteosomal degradation 104,105 . Our findings indicate that chronic ER stress precludes the phosphorylation of p53 at Ser15 and further confirm that ER stress leads to p53 destabilisation 101,102 , though different phosphorylation sites were examined among the studies. In our study, the p53 destabilisation seems to lie downstream of the decreased activity of ERK1/2, which has been previously shown to regulate the phosphorylation of this transcription factor at Ser15 in SH-SY5Y cells 106 . Although Liraglutide does not alleviate the impaired ERK1/2 phosphorylation, it potentiates Akt signalling which has been previously shown to potentiate the atypical p53-related protein kinase and phosphorylate p53 at Ser15 in human cell lines 107 , as well as to restore p53 stabilisation following cellular stress 108 . Intriguingly, it has been recently reported that Akt phosphorylates and inhibits PERK 109 , which may offer an additional mechanistic link on how Liraglutide confers its restorative effects on p53, though necessitates further experimentation.
Though not examined in the present study, Liraglutide may restore calcium homeostasis to elicit neuroprotection upon chronic thapsigargin treatment. Previous studies have demonstrated that GLP-1R induction potentiates cyclic AMP (cAMP) production 20 and regulates calcium responses 21,49 , which underlie protection of hippocampal neurons and SH-SY5Y neuroblastoma cells from excitotoxicity 49 and oxidative stress-induced 21 apoptosis, respectively. Furthermore, cAMP increase by GLP-1R stimulation potentiates protein kinase A (PKA) that downstream induces SERCA function to promote cytoprotection in insulin-resistant macrophages 110 and high glucose-treated cardiomyocytes 111 . Accordingly, PKA pharmacological inhibition blocks the GLP-1R-mediated anti-apoptotic effects, whilst the adenylate cyclase activator, Forskolin mimics the GLP-1R-induced cardioprotection upon hyperglycaemia 111 . cAMP has been additionally shown to potentiate cAMP-regulated guanine nucleotide exchange factors (also known as Epac) that regulate calcium dynamics in response to GLP-1R stimulation 112 .
In conclusion, our study demonstrates the neuroprotective/restorative effects of Liraglutide upon unmitigated neuronal ER stress. It further unravels a complex signalling network through which Liraglutide regulates UPR outcome, elicits autophagy machinery homeostasis and shifts cell fate from apoptosis to survival, providing additional evidence for the beneficial effects of GLP-1R stimulation and signalling in neurodegenerative disorders and deepening our understanding of the underlying mechanism. Cell culture. The human neuroblastoma SH-SY5Y cell line (ATCC ® CRL2266 ™ ) was obtained from

Materials. Cell proliferation kit II
LGC Standards (Middlesex, UK) and cultivated in Dulbecco's modified eagle medium/nutrient mixture F-12 (DMEM/F-12, 1:1; 1X) Glutamax ™ supplemented with 10% heat-inactivated foetal bovine serum (FBS), 100 IU ml −1 of Penicillin and 100 μg ml −1 of Streptomycin. Cells were maintained at 37 °C in a humidified incubator with 5% CO 2 and 95% air. Cells were subcultured when 80-90% confluent and seeded at 1:10 ratio. When passaged, viable cells were counted and seeded at the desired cell density for the assays using the Countess ™ Automated Cell Counter (Thermo Fisher Scientific, Inchinnan Business Park, Paisley, UK). The latter is based on the standard trypan blue exclusion technique, in which dead cells are selectively permeable to the dye and stained blue. Culture medium was renewed every 3 to 4 days.
Cell treatments. Thapsigargin was received as a colourless solid film, solubilised in 100% dimethyl sulfoxide (DMSO) at a concentration of 1 mM, aliquoted and stored at −20 °C until used. For the experiments, thapsigargin stock preparations were serially diluted in serum-free culture medium at final working concentrations of 10 to 1000 nM, containing ≤0.1% DMSO; DMSO (≤0.1%) did not affect cell viability and proliferation as assessed in preliminary experiments (data not shown).
Liraglutide was purchased from GL Biochem Ltd (Shanghai, China). The purity of the peptide was analysed by reverse-phase HPLC and characterised using matrix-assisted laser desorption ionisation time-of-flight (MALDI -TOF) mass spectrometry, as previously described 113 . The peptide was reconstituted in Gibco Water for Injection for Cell Culture to a concentration of 1, aliquoted and stored at −20 °C until used. For the experiments, Liraglutide stock preparations were diluted in serum-free culture medium to a final working concentration of 100 nM. The concentration was selected on the basis of previous experiments in which our group has established optimal working concentrations for the neuroprotective and anti-apoptotic effects of Liraglutide 19,21 . Cell viability, proliferation and cytotoxicity assessment. Cell viability, proliferation and cytotoxicity were determined using Cell proliferation kit II (XTT), cell proliferation ELISA BrdU (colorimetric) kit and Cytotoxicity Detection Kit PLUS (LDH), respectively. The assays were formatted in Nunc ™ MicroWell ™ flat-bottomed 96-well plates. SH-SY5Y cells were seeded at a density of 2 × 10 4 cells per well for 24 h. The cells were subsequently serum starved in serum-free medium for 8 h and stressed with different concentrations of thapsigargin for 16 h, in the presence or absence of 100 nM of Liraglutide. All cell treatments were performed in sextuplicate per plate per experiment. Following cell treatments, 50 μL of XTT labelling reagent and 1 μL of electron coupling reagent was added to each well, yielding final XTT concentration of 0.3 mg mL −1 . Cells were incubated with the XTT labelling mixture for 6 h at 37 °C in a humidified incubator with 5% CO 2 and 95% air. Plates were then gently shaken for 5 min on a Microtitre plate shaker (Stuart, Staffordshire, UK) and absorbance was measured at 492 and 690 nm (reference wavelength) in a Infinite ® 200 PRO microplate reader (Tecan, Berkshire, UK). The assay rests on the cleavage of the yellow tetrazolium salt XTT into an orange, water-soluble formazan product by metabolically active cells (mitochondrial dehydrogenase enzyme activity). Thus, absorbance values are proportional to the number of viable cells in the respective microcultures.
Alternatively, SH-SY5Y cells were incubated with 10 μM BrdU labelling solution for 6 h at 37 °C in a humidified incubator with 5% CO 2 and 95% air. The pyridine analogue BrdU was incorporated, in place of thymidine, into the newly synthesised DNA strands of proliferating cells. BrdU incorporation was detected by immunoperoxidase staining and a subsequent colorimetric substrate reaction as per manufacturer's protocol. Plates were read at 450 and 690 nm (reference wavelength) using Infinite ® 200 PRO microplate reader. Developed colour and absorbance values reflect the amount of DNA synthesis, and hereby the number of proliferating cells in the respective microcultures.
Cytotoxicity Detection Kit PLUS was used to quantify LDH activity; LDH is a stable cytoplasmic enzyme present in all cells and rapidly released into the culture supernatant upon plasma membrane damage. The assay involves a coupled enzymatic reaction that results in the formation of a red, water-soluble formazan product during a limited time period. The amount of colour developed is proportional to the number of damaged/lysed cells in the respective microcultures. As per manufacturer's instructions, following cell treatments, 50 μL of cell supernatant was collected and incubated with the provided reaction mixture for 30 min at room temperature. The reaction was terminated by adding the Stop Solution from the kit and absorbance was measured at 490 and 650 nm (reference wavelength).
Sample collection and protein extraction. 2×10 6 cells were grown at 75 cm 2 Nunc ™ Cell Culture Treated EasYFlasks ™ for 24 h. Following serum starvation for 8 h, cells were stressed with 100 nM of thapsigargin for 16 h, in the presence or absence of 100 nM of Liraglutide. Thereafter, cells were washed once with ice-cold 1X phosphate-buffered saline (PBS) and harvested in 1X cell lysis buffer containing protease/phosphatase inhibitor cocktail (1X). After two freeze/thaw cycles, whole-cell lysate was collected and total protein was extracted by centrifugation at ~16000×g at 4 °C for 15 min. Quick start ™ Bradford protein assay was conducted to estimate the protein concentration of the samples, as previously described 19,21 . Western blotting. Protein of whole-cell lysate (4 μg) was reduced and denaturated by boiling in lithium dodecyl sulfate (LDS) sample buffer containing 50 mM dithiothreitol (DTT) at 95 °C for 5 min. Replicate protein samples were separated on Bolt ™ 4-12% gradient Bis-Tris gel and blotted onto nitrocellulose membranes using iBlot ® 2 Dry Blotting System. Blots were washed once in 1X TBS for 5 min, blocked in 5% w/v skimmed milk for 1 h at room temperature, and probed with the primary antibodies against BiP, calnexin, Ero1α, PDI, IRE1α, CHOP, ATF6 (1:500), CASP12 (1:2000), BID, LC3, ATG3, ATG7, beclin, and β-actin (1:10 4 ) overnight at 4 °C. Alternatively, blots were blocked in 5% BSA and probed with the primary antibodies against phospho-IRE1 (Ser724) and phospho-Bcl-2 (Ser70) overnight at 4 °C. All the primary antibodies used were diluted in 5% BSA in 1X TBS with 0.05% Tween ® 20 (TBS-T; pH 8) at 1:1000 ratio, unless otherwise specified. All the primary antibodies used were generated in rabbit, except for the antibodies against CHOP, ATF6 and β-actin which were raised in mice. Following primary antibody incubation, blots were washed three times in 1X TBS-T for 5 min each and incubated with the HRP-linked secondary antibodies against the corresponding species IgG (1:2000) for 1 h at room temperature. Blots were developed using Amersham ECL Prime western blotting detection reagent kit as per manufacturer's instructions. ChemiDoc ™ MP Imaging System with Image Lab ™ software (BIO-RAD Laboratories Ltd, Hertfordshire, UK) used to image chemiluminescent bands and perform densitometric analysis. β-Actin protein was served as loading control to which relative peak intensities of the examined markers were normalised. To reprobe, blots were incubated in Restore ™ PLUS stripping buffer for 20 min at 37 °C with gentle agitation and subsequently washed three times in TBS for 5 min each. Chemiluminescent detection to ensure the removal of the original signal preceded blot re-incubation with another primary antibody of interest.
PathScan ® sandwich immunoassay. The PathScan ® Intracellular Signaling Array kit was used in accordance to manufacturer's protocol. Briefly, 100 μL Array Blocking Buffer was added to each well of the the antibody array slide and incubated for 15 min at room temperature. Thereafter, protein of whole-cell lysate (0.3 mg mL −1 ) was placed onto each well and incubated overnight at 4 °C with gentle agitation. All samples per experiment were processed in duplicate. The following day, all the wells were washed three times in 1X Array Wash Buffer for 5 min each, and probed with 1X Detection Antibody Cocktail for 1 h at room temperature with gentle agitation. Streptavidin-conjugated HRP along with LumiGLO ® /Peroxide Reagent were used to visualise the bound antibody cocktail by chemiluminescence. ChemiDoc ™ MP Imaging System with Image Lab ™ software was used to capture images of the slide. The Protein Array Analyzer 114 for Image J (National Institutes of Health, Bethesda, Maryland, USA) was used for the densitometric analysis. The relative peak intensities of the examined signalling molecules were normalised to the corresponding values of the positive and negative controls of the array.
Immunocytochemistry. 1×10 5 cells per well were grown on the Millicell EZ SLIDE eight-well glass chamber slides for 24 h. Following serum starvation for 8 h, cells were stressed with 100 nM of thapsigargin for 16 h, in the presence or absence of 100nM of Liraglutide. Thereafter, cells were washed once with 1X PBS, fixed with 4% paraformaldehyde for 10 min and permeabilised in 0.3% Triton-X-100 for 5 min at room temperature. Cells were blocked in 10% normal goat serum and incubated with the primary antibody against Nrf2 (1:500) or ATF6 (1:200) overnight at 4 C. Primary antibody was diluted in 1% BSA in 1X PBS supplemented with 0.3% Triton-X-100. Following primary antibody incubation, cells were washed three times in 1X PBS for 5 min each and incubated with goat anti-rabbit IgG H & L Alexa Fluor ® 488 secondary antibody (1:1000) for 1 h at room temperature. Specimens were cover-slipped with Vectashield Antifade Mounting Medium with DAPI (Vector Laboratories Ltd, Cambridgeshire, UK) and sealed with nail polish. Zeiss LSM510 Meta Laser Scanning confocal microscope was used for cell imaging and eight pictures were captured per experimental group per experiment for quantification. Image J was used to quantify Nrf2 corrected total cell fluorescence (CTCF) in the cell nucleus. Image acquisition and processing were performed in a blinded fashion.

Statistics.
All the results were expressed as mean ± standard error (SEM) of at least three independent experiments for each group. Differences among means were considered significant if p ≤ 0.05. Data was processed with one way ANOVA analysis, followed by post hoc Bonferroni's multiple-comparison t-tests to identify differences among groups of unstressed and stressed conditions. The effects of Liraglutide were studied by two way ANOVA, followed by post hoc Bonferroni's multiple-comparison t-tests. Statistical calculations were performed in GraphPad Prism 5 (GraphPad Software Inc., San Diego, USA) for Mac OS X software. Data availability. The data supporting the findings of this study are included in this published article and its supplementary information file. All datasets generated during the current study are available from the corresponding author (C.H.) on reasonable request.