INVESTIGATING THE ROLE OF ATF6β IN THE ER STRESS RESPONSE OF PANCREATIC β-cells

Transcription

1 INVESTIGATING THE ROLE OF ATF6β IN THE ER STRESS RESPONSE OF PANCREATIC β-cells by Tanya Odisho A thesis submitted in conformity with the requirements for the degree of Master of Science Graduate Department of Physiology University of Toronto Copyright by Tanya Odisho 2013 i

2 Investigating the Role of ATF6β in the ER Stress Response of Pancreatic β-cells Tanya Odisho Master of Science Graduate Department of Physiology University of Toronto 2013 ABSTRACT Endoplasmic reticulum (ER) stress has been implicated as a causative factor in the development of pancreatic β-cell dysfunction and death resulting in type 2 diabetes. This thesis examined the role of ATF6β in the ER stress response of β-cells. Using an ATF6βspecific antibody, expression of full-length ATF6β was detected in various insulinoma cell lines and rodent islets and the induction of the active form (ATF6βp60) under ER stress conditions. Knock-down of ATF6β in INS-1 832/13 cells did not affect mrna induction of known ER stress response genes in response to tunicamycin-induced ER stress, however it increased the susceptibility of β-cells to apoptosis. Conversely, overexpression of ATF6βp60 reduced the apoptotic phenotype. Microarray results suggest ATF6β functions to induce expression of adaptive genes also regulated by ATF6α, but also several specific targets genes. These findings have increased our understanding of the role of ATF6β in the ER stress response of β-cells. ii

3 ACKNOWLEDGEMENTS I would like to express my sincere gratitude and acknowledge numerous individuals without whom this thesis could not have been made possible. First and foremost, I would like to thank my supervisor, Dr. Allen Volchuk, for giving me the chance to be a part of his lab, which has greatly enhanced my interest in the study of diabetes. His profound scientific insight and his guidance in experimental planning have been indispensable to the completion of this thesis. I also want to thank Allen for supporting my attendance and giving me the opportunity to present my work at the 73 rd American Diabetes Association Scientific Sessions in Chicago, Illinois. Additionally, I would also like to thank my committee members, Drs. Giacca and Brubaker, for their valuable advice and constructive feedback on written work as well as presentations. I would like to express my gratitude to past and present members of the Volchuk lab for their companionship and for making my experience in the lab a pleasure: Liling Zhang, Irmgard Schuiki, Tracy Teodoro-Morrison, Peter Sollazzo, Courtney Nozak and Wilson Kuwabara. I would like to especially acknowledge Liling for her expertise and technical guidance, which have allowed me to successfully produce the results presented in this thesis. I also want to thank all members of the 10 th floor of MaRS that have contributed to a rewarding graduate school experience. I am especially grateful to Wilfred for always challenging me and being a constant support. I would also like to thank Lynn Prefontaine for her administrative support and assistance. I would also like to thank my friends and family for their constant love and support, for keeping me grounded and for always encouraging me to push myself beyond my limits. Last but not least, I would like to acknowledge Banting and Best Diabetes Centre for funding my work and travel expenses. iii

4 TABLE OF CONTENTS ABSTRACT... ii ACKNOWLEDGEMENTS... iii TABLE OF CONTENTS... iv LIST OF TABLES... vi LIST OF FIGURES... vii LIST OF ABBREVIATIONS... viii CHAPTER 1: INTRODUCTION Pancreatic β-cell physiology Glucose-stimulated insulin secretion (GSIS) Mechanism of insulin release Insulin biosynthesis Type 2 diabetes ER stress and pancreatic β-cell failure The endoplasmic reticulum The unfolded protein response The PERK signalling pathway The IRE1 signalling pathway The ATF6 signalling pathway ATF6α ATF6β ER stress-induced apoptosis CHOP/GADD Activation of JNK Activation of Caspases ER stress as causative factor in β-cell dysfunction and death in diabetes Rationale and Hypothesis CHAPTER 2: MATERIALS AND METHODS Cell culture Cell Treatments and Lysis Western Blot Analysis Short-interfering RNA-mediated Knock-down RNA Isolation Reverse Transcription PCR Real-Time Quantitative PCR XBP1 Splicing Assay Apoptosis Assay iv

5 2.10 Generation of ptet-on doxycycline inducible cell line Cloning of FLAG-ATF6βp60 into pshuttle- IRES-hrGFP-2 vector and Adenovirus production INS-1 832/13 Cell Infection with Adenovirus Immunofluorescence Staining Microarray analysis Data analysis CHAPTER 3: EXPRESSION AND ACTIVATION OF ATF6β IN PANCREATIC β-cells AND THE EFFECT OF ATF6β DEPLETION ON THE ER STRESS RESPONSE AND SURVIVAL OF PANCREATIC β-cells Introduction Results ATF6β is expressed in pancreatic β-cell lines and islets and is activated by ER stress Knock-down of ATF6β has no effect on the mrna expression of well-known UPR genes Effect of ATF6β depletion on β-cell susceptibility to ER Stress ATF6β depletion increases susceptibility of pancreatic β-cells to ER-stressinduced apoptosis ATF6βp60 overexpression protects pancreatic β-cells against ER stress-induced apoptosis Discussion Summary and Future directions CHAPTER 4: MICROARRAY ANALYSIS FOR IDENTIFICATION OF ATF6β TARGET GENES Introduction Results Generation of a stable cell line with inducible expression of ATF6βp Ad-FLAG-ATF6βp60 is functional and localizes to the nucleus Identification of potential ATF6β-specific target genes using microarray analysis Discussion Summary and Future directions Overall Conclusion REFERENCES v

6 LIST OF TABLES CHAPTER 4: MICROARRAY ANALYSIS FOR IDENTIFICATION OF ATF6β TARGET GENES Table 4. 1 Attempts at generating a stable cell line with Doxycycline inducible expression of ATF6βp Table 4. 2 Biological processes identified for enriched genes upregulated 2-fold by ATF6βp60 by microarray analysis Table 4. 3 Comparison of enriched genes identified to be upregulated 2-fold by Ad- ATF6αp50 and Ad-ATF6βp60 in comparison to Ad-GFP vi

7 LIST OF FIGURES CHAPTER 1: INTRODUCTION Figure 1. 1 Glucose-stimulated insulin secretion Figure 1. 2 The unfolded protein response in mammalian cells Figure 1. 3 Schematic structures of each region in ATF6α (the ATF6 gene product) and ATF6β (the CREB-RP/G13 gene product) Figure 1. 4 ER stress pathways implicated in mediating cell apoptosis CHAPTER 3: EXPRESSION AND ACTIVATION OF ATF6β IN PANCREATIC β- CELLS AND THE EFFECT OF ATF6β DEPLETION ON THE ER STRESS RESPONSE AND SURVIVAL OF PANCREATIC β-cells Figure 3. 1 ATF6β mrna expression in pancreatic β-cells and islets Figure 3. 2 ATF6β is expressed at the protein level and is activated by ER stress in pancreatic β-cells Figure 3. 3 Knock-down efficiency of the ATF6β protein Figure 3. 4 Effect of ATF6β knockdown on ER stress response genes Figure 3. 5 ATF6β depletion increases susceptibility of pancreatic β-cells to ER stress Figure 3. 6 ATF6β depletion does not affect levels of spliced Xbp Figure 3. 7 ATF6β depletion increases susceptibility of pancreatic β-cells to apoptosis Figure 3. 8 ATF6β knock-down sensitizes pancreatic β-cells to apoptosis Figure 3. 9 ATF6β depletion does not activate JNK and p Figure ATF6βp60 overexpression protects cells against ER stress-induced apoptosis. 48 CHAPTER 4: MICROARRAY ANALYSIS FOR IDENTIFICATION OF ATF6β TARGET GENES Figure 4. 1 Transient transfection of ptet-on INS-1 #46 cells with ptre-flag-atf6βp60 plasmid Figure 4. 2 ATF6βp60 overexpression and localization in pancreatic β-cells Figure 4. 3 Effect of ATF6αp50 and ATF6βp60 on Grp78 mrna Figure 4. 4 Validation of microarray genes upregulated by ATF6βp60 by real-time PCR Figure 4. 5 A hypothetical model for the role of ATF6β in the ER stress response of pancreatic β-cells vii

8 LIST OF ABBREVIATIONS ABI Applied Biosystems ADP adenosine diphosphate ANOVA analysis of variance ASK1 apoptosis signal-regulating kinase 1 ATF3 activating transcription factor 3 ATF4 activating transcription factor 4 ATF6 activating transcription factor 6 ATP adenosine triphosphate BCA bicinchoninic acid Bcl-2 B-cell lymphoma 2 Bak Bcl-2 Homologous Antagonist/ Killer Bax Bcl-2 Associated X-Protein BBF2H7 box B-binding factor 2 human homolog on chromosome 7 BiP immunoglobulin heavy chain binding protein bzip basic leucine zipper Ca 2+ cdna calcium complementary deoxyribonucleic acid CC3 cleaved caspase 3 C/EBP CCAAT/enhancer binding protein CHO Chinese hamster ovary CHOP CAAT/enhancer binding protein (C/EBP) homologous protein COPII coat protein complex II CREB3/LUMAN camp responsive element-binding protein 3 CREB4 camp responsive element-binding protein 4 CREBH camp responsive element-binding protein, hepatocyte specific CREB3L2 CAMP responsive element binding protein 3-like 2 CREB-RP cyclic AMP-response-element-binding protein-related protein CSSR core stress sensing region DAPI 4',6-diamidino-2-phenylindole viii

9 Ddit3 DNA-damage inducible transcript 3 Derlin-1 degradation in endoplasmic reticulum protein 1 DMEM Dulbecco s modified Eagle s medium Dox doxycycline DTT dithiothreitol EGFP enhanced green fluorescent protein EDEM ER degradation enhancing α-mannosidase-like protein eif2α eukaryotic initiation factor 2α ELISA enzyme-linked immunosorbent assay ER endoplasmic reticulum ERAD ER-associated degradation ERO1 ER oxidoreductin 1 ERSE ER stress-response element FBS fetal bovine serum FFA free fatty acids G418 Geneticin GADD34 growth arrest and DNA damage-inducible gene 34 GADD153 growth arrest and DNA damage-inducible gene 153 GFP green fluorescent protein GRP78 glucose-regulated protein of 78-kDa GRP94 glucose regulatory protein of 94-kDa HeLa Henrietta Lacks (HeLa) Immortal Cells HEPES 4-(2-Hydroxyethyl)-1-Piperazineethanesulfonic Acid HERP homocysteine inducible ER protein Hmox1 heme oxygenase (decycling) 1 HRD1 hypoxia responsive domain 1 HRP horseradish peroxidase Hsp70 heat shock protein 70 IL-1β interleukin 1 beta IRE1 inositol-requiring enzyme 1 JNK c-jun N-terminal kinase ix

10 K ATP MHC mrna MEFs NF-Y NuPAGE OASIS Opu PBS ATP-sensitive K + channels major histocompatibility complex messenger RNA mouse embryonic fibroblasts nuclear factor Y Novex high performance polyacrylamide gel electrophoresis old astrocyte specifically-induced substance optical particle units phosphate buffered saline PC1 proprotein convertase 1 PC2 proprotein convertase 2 PCR polymerase chain reaction PDI protein disulfide isomerase PDX-1 pancreatic/duodenal homeobox-1 PERK double-stranded RNA-activated protein kinase (PKR)-like ER kinase PFA paraformaldehyde PMSF phenylmethanesulfonylfluoride PP pancreatic polypeptide cells PP1 phosphoprotein phosphatase 1 qpcr quantitative real-time PCR RIP regulated intramembrane proteolysis RNA ribonucleic acid ROS reactive oxygen species RP reserve pool RPMI Roswell Park Memorial Institute medium RRP ready releasable pool RT-PCR reverse transcription polymerase chain reaction rtta reverse tetracycline-controlled transactivator SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis SDF2L1 stromal cell derived factor 2 like 1 x

11 SERCA sarco-endoplasmic reticulum Ca 2+ -ATPase pump sirna short interfering RNA S1P site-1 protease S2P site-2 protease TET-ON tetracycline ON Tg thapsigargin Tn tunicamycin TNF tumor necrosis factor TRAF2 TNF receptor-associated factor 2 TRB3 tribbles 3 UPR unfolded protein response UPRE unfolded protein response promoter element WFS1 wolframin (Wolfram syndrome 1) XBP1 x-box-binding protein 1 xi

12 CHAPTER 1: INTRODUCTION 1

13 1.1 Pancreatic β-cell physiology The pancreas is a dual-function organ, consisting of both exocrine as well as endocrine glands. The endocrine gland consists of clusters of islets of Langerhans which make up 1-2% of the entire organ and is comprised of four major cell types (1). The pancreatic β-cells are the predominant cell type in the islets, comprising about 60-80% of the islet content, with the remaining cells consisting of the glucagon-releasing α-cells, the somatostatin-producing δ-cells and pancreatic polypeptide (PP) cells (1). As specialized secretory cells, pancreatic β-cells are the only cell type responsible for synthesizing and secreting the hormone insulin post-prandially in order to maintain whole body glucose homeostasis Glucose-stimulated insulin secretion (GSIS) As blood glucose increases in response to food intake, the body s β-cells sense and uptake glucose molecules via facilitated diffusion through the GLUT-2 transporter. Inside the cell, the glucose is converted to glucose-6-phosphate by the rate-limiting enzyme glucokinase. The modified glucose is then metabolized by glycolysis in the cytosol and later in the mitochondria by the Krebs cycle and oxidative phosphorylation to produce energy in the form of ATP. In resting cells in the absence of glucose, the potential difference across the β-cell membrane is negative and maintained at -70mV. However, as glucose is metabolized and the ATP:ADP ratio increases, the ATP-sensitive potassium channels (K ATP ) on the cell membrane close, preventing K + ions from diffusing out of the cell. As K + ions accumulate inside the cell, the cell becomes depolarized, causing voltage-gated calcium channels to open, which allow Ca 2+ ions from the outside to move inside the cell, down their concentration gradient (2). As calcium enters the cell, it stimulates insulin storage granules to fuse with the plasma membrane to release insulin by exocytosis into nearby blood vessels. Although glucose serves as the main secretagogue that stimulates insulin secretion from β-cells, other metabolites such as free fatty acids and certain amino acids can also stimulate insulin secretion (3-5) (Figure 1). 2

14 Figure 1. 1 Glucose-stimulated insulin secretion. Glucose enters the cells via the GLUT2 transporter (1) and undergoes glycolytic and mitochondrial metabolism (2), which ultimately has the effect of increasing the ATP: ADP ratio (3). An increased ATP: ADP ratio leads to the closure of ATP-sensitive K ATP channels (4) and to membrane depolarization (5), which triggers the opening of voltage-dependent Ca 2+ channels (6). The resulting influx of Ca 2+ (7) induces the fusion of insulin-containing granules with the plasma membrane and insulin release from the cell (8). PM, plasma membrane. Adapted from Wang et al., 2009 (6) Mechanism of insulin release Insulin release from pancreatic β-cells is biphasic (7). The biphasic process consists of an initial rapid and brief release of insulin from the readily releasable pool (RRP), and the second phase entails a more sustained release from the reserve pool (RP) when the available insulin granules are depleted (1). Upon stimulation, the β-cell immediately releases insulin from the RRP, which is pre-docked at the plasma membrane ready to fuse. As the β-cell continues to respond to elevated glucose levels, it induces the RP, which is initially farther from the plasma membrane, to move towards the cell membrane before releasing its contents. The first phase of insulin release is maximal at 4-6 min, and this is followed by the more enduring second phase which plateaus at 30 min (7). In order to regenerate the depleted insulin stores during the secretion events, pancreatic β-cells also stimulate both transcription and translation of insulin (8) Insulin biosynthesis In rodents, insulin is expressed from two independent genes, Ins1 and Ins2, while in humans there is only one gene for insulin which is an orthologue of the rodent Ins2 gene (9). 3

15 The Ins1 gene in rodents is thought to have originated from a reverse-transcribed partially processed mrna of Ins2. The insulin gene in humans and in rodents is transcribed to mrna and translated to a preprohormone (preproinsulin) that undergoes extensive post-translational modifications. The preproinsulin contains an amino-terminal signal sequence, which directs the precursor hormone to the ER lumen for post-translational processing. Here, the signal sequence from preproinsulin is cleaved and proinsulin is generated. The proinsulin protein consists of an A chain, a B chain and C-peptide, which are linked by three disulfide bonds: two joining the A chain and B chain, and one intra-chain bond within the A chain (10). The oxidizing environment of the ER allows the formation of the disulfide bonds. The proinsulin is then properly folded in the ER with the assistance of molecular chaperones and then trafficked to the Golgi apparatus. In the Golgi, the proinsulin is proteolytically processed by the proprotein convertase (PC) 1, PC2 and carboxypeptidase E which cleave off the C-peptide from the protein (11). The resulting insulin protein consists of the A and B chains held together by disulfide bonds. The mature insulin is then packaged into secretory vesicles at the trans-golgi network. These granules are rich in zinc and calcium and store insulin in the form of microcrystals in order to concentrate the insulin to maximize the amount of insulin released upon stimulation of the β-cell (12). Thus, the β-cell is ready to trigger release of the insulin granules in response to a glucose stimulus. 1.2 Type 2 diabetes The process of insulin synthesis and secretion by pancreatic β-cells is crucial to maintaining whole body glucose homeostasis. An insult to the β-cell that can compromise the cell s function and result in apoptosis can lead to diabetes mellitus, a complex disorder characterized by high blood glucose levels as a result of insufficient insulin secretion to meet the body s functional requirements (13). Diabetes mellitus is broadly divided into two main types: type 1 diabetes and type 2 diabetes. Type 1 diabetes (T1D) arises as a result of an autoimmune attack on the body s β-cells, resulting in β-cell destruction. Type 2 diabetes (T2D) represents over 90% of all diabetes cases, and is influenced by both genetic and environmental factors that result in pancreatic β-cell failure and peripheral insulin resistance in the liver, muscle and adipose tissues (14). The β-cell s response is to compensate for the insulin resistance by increasing its capacity to secrete insulin. Over time, pancreatic β-cells are thought to become 4

16 exhausted from over-production and in combination with glucotoxicity and lipotoxicity, β-cells become dysfunctional and apoptotic, resulting in T2D. 1.3 ER stress and pancreatic β-cell failure Upon glucose stimulation, there is a 25-fold increase in proinsulin synthesis, and approximately 1x10 6 proinsulin molecules are synthesized per minute (15), which imposes a heavy biosynthetic burden on the β-cell. Thus, one of the characteristic features of β-cells is a highly developed ER and a significant production of proteins involved in ER function, essential for efficient insulin synthesis and secretion (16, 17) The endoplasmic reticulum The ER is a membrane enclosed compartment with a luminal space that provides a specialized environment for the post-translational processing and folding of secreted, transmembrane and resident proteins (18, 19). While a number of important post-translational modifications including lipid biosynthesis, hydroxylation, oligomerization, etc., occur in the ER, disulfide oxidation and N-linked glycosylation are among the general modifications that are common to the majority of secreted proteins. To ensure proper folding of proteins into their functional conformations, the ER houses an elaborate machinery of protein-folding and proteinprocessing enzymes including molecular chaperones, glycosylation enzymes and oxidoreductases (20). Through ATP-hydrolytic cycles of binding and release, ER molecular chaperones (e.g. GRP78) mediate proper protein folding. Glycosylation enzymes modify newly synthesized proteins by covalently attaching glycans. Furthermore, the ER lumen has a highly oxidizing environment, which facilitates proper oxidative protein folding by promoting disulfide bond formation, a process catalyzed by oxidoreductases and protein disulfide isomerases (PDIs). The ER also has a Quality control monitoring system. While properly folded and assembled proteins are cleared to exit the ER and progress down the secretory pathway via COPII vesicles destined for the Golgi compartment, incompletely folded or misfolded proteins are recognized and retained to complete the folding process or retrotranslocated out of the ER into the cytoplasm where they are targeted for degradation in a process named ER-associated degradation (ERAD) (19, 21). In addition to its role in protein folding, the ER serves as a cellular Ca 2+ store. The concentration of Ca 2+ is 3-4 orders of magnitude higher in the ER than in the cytoplasm (22). This gradient is generated by the sarco/endoplasmic reticulum Ca 2+ 5

17 ATPase (SERCA) pump, which pumps Ca 2+ into the ER lumen (23). The Ca 2+ storage in the ER is required for chaperone function and serves as a signalling molecule following its release into the cytosol, which can be used in various molecular pathways (24). A defect in any of the stages of protein synthesis and processing in the ER can affect the protein folding function of the ER and there are various conditions that can disturb ER function that can result in the accumulation of unfolded proteins. These include inhibition of protein glycosylation, reduction of formation of disulfide bonds, calcium depletion from the ER lumen, impairment of protein transport from the ER to the Golgi, expression of malfolded proteins, hypoxia, etc. Such ER dysfunction causes proteotoxicity in the ER, collectively termed ER stress (25, 26), which can compromise cell function and viability. In order to adapt and survive these conditions of ER stress, cells activate a defense mechanism known as the unfolded protein response (UPR) The unfolded protein response The ability to adapt to physiological levels of ER stress is important to cells, especially to specialized secretory cells like the insulin producing β-cells. The UPR is an adaptive program that aims to restore normal ER function by alleviating ER stress. Ultimately, activation of the UPR is brought about by disequilibrium between the protein folding capacity and the protein folding demands of the cell. Transduction of the UPR signal is mediated by three transmembrane proteins, PERK (double stranded RNA-activated protein kinase (PKR)-like endoplasmic reticulum kinase), IRE1 (inositol requiring 1), and ATF6 (activating transcription factor 6). Activation of these ER stress sensor proteins is regulated by the molecular chaperone GRP78, which under basal conditions, binds to their luminal domain to keep them in an inactive state. Upon ER stress, the large excess of unfolded proteins in the ER lumen necessitates GRP78 dissociation and the resultant activation of these proteins (27-30). In general, the UPR involves four distinct processes: (i) downregulation of the biosynthetic load by translational attenuation, (ii) increase in the protein folding capacity through transcriptional transduction of ER chaperones, (iii) degradation of terminally misfolded proteins by ERAD, and (iv) apoptosis, as a last resort, to eliminate damaged cells and maintain homeostasis (31-33) (Figure 2). Thus, the proper functioning of the ER and UPR are crucial for the cell s survival, and this becomes a 6

18 greater necessity for specialized secretory cells, such as pancreatic β-cells, for which the efficiency and fidelity of protein folding needs to be constantly adjusted (26). Figure 1. 2 The unfolded protein response in mammalian cells. Accumulation of unfolded proteins in the ER activates four distinct cellular responses. 1. Translational attenuation reduces the load of new protein synthesis and prevents further accumulation of unfolded proteins, 2. Transcriptional induction of ER chaperones increases protein folding activity and prevents protein aggregation, 3. The ER-associated degradation (ERAD) pathway eliminates misfolded proteins by the ubiquitin-proteasome system, and 4. Apoptosis is induced when the function of the ER is severely impaired and ER stress persists. These responses are mediated by the three ER stress sensor proteins: PERK, IRE1, and ATF6. Activation of the PERK pathway leads to phosphorylation of eif2α, resulting in global translation attenuation. Phosphorylated eif2α selectively translates the ATF4 transcription factor that induces expression of its target genes. Activation of IRE1 leads to splicing of the Xbp1 mrna, translation of which produces an active transcription factor that is response for inducing the transcription of chaperones and ERAD components. ATF6 activation leads to proteolysis to an active form that translocates into the nucleus to and bind the ER stress elements (ERSE) of UPR target genes. Adapted from Araki et al., 2003 (34) The PERK signalling pathway As part of the adaptive response, the PERK arm of the UPR mediates the first response to ER stress by a transient attenuation of global protein translation in order to reduce the client protein load on the ER. PERK is a type I ER-resident transmembrane protein, composed of an 7

19 ER luminal stress sensor and a cytosolic protein kinase domain (35). Under conditions of ER stress, GRP78 dissociates from PERK s N-terminal luminal domain, allowing it to dimerize and trans-autophosphorylate (36, 37). Activation of PERK leads to phosphorylation of the alpha subunit of eukaryotic initiation factor 2 (eif2), which normally recruits charged initiator methionyl trna to the 40S ribosomal unit. The phosphorylation of eif2α on serine 51 blocks the exchange of GDP bound to eif2α, hence inhibiting 80S ribosome assembly to effectively attenuate global translation initiation and protein synthesis (35, 38). Decreased translation reduces the influx of protein into the ER, hence diminishing the folding load. Furthermore, activation of PERK leads to selective stress-induced translation of certain mrnas that are otherwise basally repressed. Genes with internal ribosome entry site (IRES) sequences in the 5 untranslated regions bypass the eif2α translational block (16). This includes activating transcription factor 4 (ATF4), which encodes a camp response element-binding transcription factor (C/EBP) (39). Translation of ATF4 leads to expression of several pro-survival genes including those involved in amino acid transport and synthesis, resistance to oxidative stress (antioxidant genes), and glutathione biosynthesis (40). However, ATF4 also stimulate expression of transcription factor C/EBP homologous protein (CHOP), an important proapoptotic gene of ER stress-mediated β-cell death (41, 42). Translational recovery is mediated by a negative feedback loop, in which ATF4 induces the transcription of growth arrest and DNA damage 34 (GADD34), which interacts with protein phosphatase 1 (PP1) to dephosphorylate eif2α (38) The IRE1 signalling pathway The only ER stress response sensor in yeast, the IRE1 protein is activated by ER stress to induce expression of chaperone proteins and components of ER-associated degradation (43). There are two mammalian IRE1 homologs and both participate in ER stress signalling: IRE1α, which is ubiquitously expressed with relatively higher expression in the pancreas and placenta, and IRE1β, which is selectively expressed in intestinal epithelium cells (44, 45). A type I transmembrane protein, IRE1 consists of an ER luminal dimerization domain and cytosolic serine-threonine kinase and endoribonuclease domains (46). Under non-stress conditions, IRE1 remains in an inactive monomeric state in association with GRP78 (47). As unfolded proteins accumulate in the ER lumen, IRE1 is activated via dissociation from GRP78, which allows 8

20 IRE1 to homo-oligomerize and activate its site-specific endoribonuclease (RNase) domain by autophosphorylation of the C-terminal kinase domain (48). Structural and mutation studies have also proposed, however, that IRE1 can sense unfolded proteins in a GRP78-independent manner, causing conformational changes that permit a similar dimerization and autophosphorylation (49). In yeast, it has been shown that GRP78 release from the IRE1α luminal domain leads to its partial activation, while the binding of misfolded protein to the core stress sensing region (CSSR) of IRE1α allows a conformational change in its cytosolic domains leading to its complete activation (50). Activation of IRE1 enables both its kinase and endoribonuclease activities. The endoribonuclease activity splices a 26-base intron from the X- box binding protein 1 (XBP1) mrna. The spliced mrna is subsequently translated to a potent transcription factor (51, 52), that translocates into the nucleus to bind ER stress response elements (ERSE), an ER stress-specific consensus sequence within the promoter sequence of target genes, in the presence of the co-regulator NF-Y (53-55). XBP1 activates transcription of a variety of UPR genes encoding chaperones, ER-Golgi transport and ERAD degradation machinery components (EDEM1, EDEM2, EDEM3, and Derlins1-3) (27, 53, 56, 57). In addition to splicing XBP1, the endoribonuclease activity of mammalian IRE1 has also been shown to mediate cleavage of additional mrnas, targeted to the ER, including proinsulin mrna (58-60), as well as cleavage of the 28S ribosomal subunit (61). This suggests that IRE1 has a role in translation attenuation by degrading ER targeted mrna transcripts and/or the ribosomal subunits that mediate translation (24, 58, 61). Similar to PERK, IRE1 promotes cell survival mechanisms as well as apoptotic pathways. Under chronic ER stress, IRE1 leads to activation of the pro-apoptotic c-jun N-terminal kinase (JNK) signalling pathway, which causes cellular dysfunction and apoptosis (31) (see section ). The importance of the IRE1/XBP1 pathway to cell viability is demonstrated by studies that show that both IRE1α and XBP1 are essential for mouse development (53, 62, 63) The ATF6 signalling pathway ATF6α A type II ER transmembrane protein, ATF6 in mammalian cells consists of two isoforms, ATF6α (64, 65) and ATF6β (66-68), both of which are expressed ubiquitously (66). Under basal conditions, the ATF6 protein is retained in an inactive state in the ER membrane. 9

21 Similar to the mechanism proposed for PERK and IRE1, the ER stress-sensing mechanism for ATF6 involves dissociation of GRP78 from its luminal domain (28, 69). For the ATF6 protein, however, dissociation from GRP78 exposes redundant Golgi localization sequences, GLS1 and GLS2 (69). This allows ATF6 to translocate to the Golgi apparatus via COPII vesicles, where the protein undergoes regulated intramembrane proteolysis (RIP), whereby the luminal domain is first cleaved by serine protease site-1 protease (S1P) followed by metalloprotease site-2 protease (S2P cleavage (28, 65, 70, 71). This releases the N-terminal active cytosolic portion (ATF6αp50), a basic leucine zipper (bzip) transcription factor (70, 71). Activation of ATF6 has also been suggested to depend upon the state of the disulfide bonds in the protein s luminal domain. Nadanaka and colleagues (72) have demonstrated that under resting conditions, the disulfide bonds keep ATF6 in an inactive state. As the demand for protein synthesis increases, these bonds are reduced, resulting in an increased ability of ATF6 to exit the ER. Although disulfide bond reduction is required for ATF6 activation, it is not sufficient, suggesting that both the ATF6 redox state and GRP78 binding are involved in sensing ER stress and activating ATF6 (72). The cleaved active form of ATF6α (ATF6αp50) migrates to the nucleus where it regulates expression of UPR genes with ATF/cAMP response elements (CREs) and ER stress response elements (ERSE) (54, 65, 73, 74). The induction of UPR genes on some promoters by ATF6 requires interaction with other bzip transcription factors and co-regulators, such as NF-Y, a general transcription factor that binds to a sequence upstream of the ATF6 protein within the ERSE (75). Furthermore, studies show that ATF6αp50 can heterodimerize with sxbp1 to bind both ERSEs and UPR elements (UPRE) conserved in the promoters of UPR genes, resulting in significant activation of genes required to restore proper ER function, protein folding, and ERAD (76). A number of tissue-specific ATF6 homologs located in the ER membrane have been identified, including OASIS, CREBH, LUMAN/LZIP/CREB3, CREB4, and BBF2H7/CREB3L2 which once activated, are processed in a similar manner as ATF6 (31). In vivo studies utilizing double knockout mice of ATF6α and ATF6β caused embryonic lethality whereas single isoform deficient mice were dispensable for embryonic and postnatal development, suggesting that ATF6α and ATF6β possess at least some overlapping function, which is essential for mouse development (76, 77). Microarray profiling analysis of mouse embryonic fibroblasts (MEFs) deficient in ATF6α or ATF6β revealed that ATF6α, but not ATF6β, is required for the induction of ER chaperones, including GRP78, and ERAD 10

22 components (HERP, SEL1L, HRD1 and EDEM1) (78). Additionally, ATF6α induces the expression of the X box-binding protein 1 (XBP1) (52). The transcriptional factor CHOP is known to be regulated by both ATF6 and PERK, therefore induction of CHOP was reduced in ATF6α-/- MEFs (78). The role of ATF6α in pancreatic β-cells has also been studied by our group (79). Comparing INS-1 832/13 cells depleted of ATF6α with control cells, we demonstrated that ATF6αp50 is required for the induction of chaperones (e.g. Grp78, Grp94) and ERAD components (e.g. Herp), whereas mrna levels of the pro-apoptotic gene Chop remained unchanged in ATF6α knock-down cells compared to control cells ATF6β Although the majority of studies have focused on the ATF6α isoform, studies suggest that the activation mode of ATF6β during ER stress is the same as that for ATF6α (75). The ATF6β isoform has been shown to be very similar to ATF6α structurally, with several conserved regions, such as the bzip domain (82% similarity), which allows them to bind to ERSEs as homo or heterodimers (Figure 3) (66, 75). However, despite ATF6β s high degree of homology to ATF6α, the structure of the N-terminal region of ATF6αp50, which possesses the transcriptional activation domain (TAD), differs greatly from that of the active cleaved form of ATF6β (ATF6βp60). The VN8 sequence found in most viral genomes (VP16) and in the ATF6α gene (75% identity), is only 25% conserved in the ATF6β gene (80). This 8 amino acid domain (VN8) confers maximal transcriptional activation and degradation, and mutations in VN8 that reduce VP16 activity decrease degradation. Based on these findings, it was suggested early on that ATF6α has potent transcriptional activity, while ATF6β displays much lower transcriptional activity, which also attested to the shorter (~2 h) and longer (~5 h) half-lives of the two isoforms, respectively (81). Furthermore, based on these and previous findings, it was suggested that since ATF6β is a poor transcriptional activator and can bind ERSEs with or in place of ATF6α, then it is probable that ATF6β might serve as an endogenous repressor of the transcriptional induction effects of ATF6α (80, 81). In fact, the same group had demonstrated that co-transfection of HeLa cells with plasmids expressing the active form of ATF6α (ATF6αp50) and ATF6βp60 led to a 20% reduction in the induction of the Grp78 gene in comparison with cells transfected only with the ATF6αp50-expressing plasmid (81). However, studies using ATF6β-/- MEFs (78) revealed that ATF6β deletion neither compromised nor 11

23 significantly accentuated stress-dependent induction of ERSE-target genes such as Grp78, suggesting that ATF6β does not function as an ATF6α antagonist as previously thought (80, 81). To date, little is known about the role of ATF6β in general or its role in pancreatic β- cells. However, in collaboration with our lab, the Poitout group (82) showed that both ATF6α and ATF6β can bind to the A5/Core region of the rat insulin 2 promoter; however, only overexpression of ATF6α inhibited the activity of the insulin promoter-reporter. Furthermore, a study by Seo et al. (83) showed that hyperactivation of ATF6α decreases insulin gene expression via upregulation of the orphan nuclear receptor small heterodimer partner (SHP), which may be advantageous under non-stimulatory conditions (82, 83). Figure 1. 3 Schematic structures of each region in ATF6α (the ATF6 gene product) and ATF6β (the CREB-RP/G13 gene product). The conserved positions of the basic region, leucine zipper (Leu-Zip) and transmembrane (TM) domain are shown by boxes with percentages of identity indicated. In the basic-leucine zipper domain ATF6α has 61% identity and 82% similarity to human ATF6β (84). The divergent N-terminal transcriptional activation domain containing the VN8 region which is required for optimal transcriptional activity and degradation is also indicated. ER stress stimulates the regulated intramembranous proteolysis (RIP) of both ATF6α and -β near and in the ER-transmembrane domains by the Golgi-associated proteases, S1P and S2P. aa, amino acids. Adapted from Thuerauf et al., 2007 (80). 12

24 1.3.3 ER stress-induced apoptosis When ER stress is severe and chronic and the protective mechanisms of the UPR fail to reverse the ER stress to regain homeostasis, apoptotic pathways are preferentially initiated to eliminate the damaged cell (85). Apoptosis is characterized by nuclear and cytoplasmic condensation, blebbing of the plasma membrane and DNA fragmentation (86). As the process of apoptosis progresses, the dying cell eventually disintegrates into membrane-enclosed apoptotic bodies that are then destroyed by phagocytes or neighbouring cells. The apoptotic pathways triggered by irremediable ER stress fall into three main categories: i) CHOP induction, ii) activation of JNK, and iii) activation of the caspases (Figure 4). 13

25 Figure 1. 4 ER stress pathways implicated in mediating cell apoptosis. I: activation of the PERK and ATF6 pathways leads to the induction of CHOP, which downregulates the expression of the anti-apoptotic protein Bcl-2 and induces GADD34 and ERO1α. The latter promote ER stress by increasing ER protein load (via translational recovery by eif2α dephosphorylation) and ERO1α alters ER redox conditions. II: activated IRE1 binds JIK and recruits TRAF2, which leads to the activation of ASK1 and JNK. JNK phosphorylates Bcl-2 and BH3-only protein (Bim), initiating mitochondria-mediated apoptosis (not shown). III: the recruitment of TRAF2 to IRE1 also permits TRAF2 to dissociate from procaspase-12 (pcp12) residing on the cytoplasmic side of ER membrane, allowing pcp12 activation. During ER stress, Bax and Bak in the ER membrane oligomerize and allow the release of Ca 2+ from the ER to the cytosol, which activates m-calpain, which subsequently cleaves and activates pcp12. Active caspase-12 (CP12) cleaves and activates procaspase-9, which in turn activates downstream caspases, including caspase-3. In addition, Ca 2+ released from the ER is taken up by the mitochondria, causing mitochondrial inner membrane depolarization and cytochrome c release into the cytoplasm. This allows the formation of the apoptosome (consisting of Apaf-1, cytochrome c, ATP, and procaspase-9), activation of procaspase-9, and subsequent downstream caspases leading to cell apoptosis. Adapted from Lai et al., 2008 (32). 14

26 CHOP/GADD153 CHOP (C/EBP homologous protein) (also known as GADD153) is a 29kDa protein member of the C/EBP family of transcription factors (87). As previously mentioned, Chop mrna is induced mainly via activation of the PERK pathway of the UPR; however, the IRE1 and ATF6 pathways have also been shown to regulate Chop induction (37, 76, 88, 89). While levels of Chop are barely detectable under physiological conditions, it is strongly induced in response to ER stress, more than under conditions of growth arrest or DNA damage (90, 91). Several studies have implicated CHOP in ER stress-induced apoptosis. Overexpression and targeted disruption of the Chop gene has demonstrated that CHOP promotes apoptosis in response to ER stress. CHOP -/- MEFs are partially resistant to ER stress and have reduced ER stress-induced apoptosis (41, 91, 92), whereas overexpression of CHOP promotes apoptosis in response to thapsigargin (an inhibitor of the ER-localized Ca 2+ -dependent ATPase) and tunicamycin (an inhibitor of N-linked glycosylation) -induced ER stress (93, 94). As a transcription factor, CHOP upregulates GADD34, a phosphatase which in conjunction with protein phosphatase 1 (PP1), dephosphorylates eif2α, leading to recovery of mrna translation. This premature translational recovery may be detrimental to the cell through the generation of reactive oxygen species (ROS) and increasing ER protein folding load (95). Furthermore, studies have shown that CHOP upregulates ERO1α expression, an ER oxidase that provides oxidizing equivalents to members of the PDI family (95). Marciniak et al. (108) have suggested that CHOP-mediated ERO1α activation causes hyper-oxidizing conditions in the ER, which may increase the levels of misfolded proteins. Additionally, CHOP induction downregulates the expression of the anti-apoptotic protein Bcl-2, which leads to depletion of cellular glutathione and increases cellular ROS, contributing to ER stress-associated cell death (40, 94). Studies have shown that disruption of the Chop gene does not prevent pancreatic β-cell apoptosis, but only delays it in the heterozygous Akita mice, a model of diabetes induced by a missense mutation in the Ins2 gene (41). Furthermore, CHOP is not upregulated in Perk -/- and eif2α (S51A) mutant cells in response to ER stress and yet these cells still undergo ER stressassociated apoptosis (40, 96). These studies suggest that CHOP induction is not the only pathway involved in mediating ER stress-induced apoptosis. 15

27 Activation of JNK The pro-apoptotic effects of the IRE1α pathway are mediated through its activation of the stress-induced c-jun N-terminal kinase (JNK) protein. Under ER stress conditions, IRE1α senses ER-stress via its luminal domain and recruits the adaptor protein tumour necrosis factor (TNF) receptor-associated factor 2 (TRAF2) to the ER membrane (63, 97). The IRE1α-TRAF2 complex recruits and activates apoptosis signal regulating kinase-1 (ASK1), which leads to c- Jun NH 2 -terminal kinase (JNK) and p38 MAPK activation (63, 97). The importance of ASK1 in mediating ER stress-induced apoptosis has been demonstrated in ASK1 -/- primary neurons and MEFs, which are resistant to ER stress inducers and are defective in JNK activation and apoptosis (98). Upon activation, JNK phosphorylates transcription factors c-jun or activation protein-1 (AP1) family and performs a number of functions including activation of the proapoptotic Bim protein (99) and inhibition of the survival protein Bcl-2 (100, 101). Phosphorylation and activation of Bim by JNK allows Bim to translocate to the mitochondrial outer membrane, where it promotes cytochrome c release and caspase activation (102). Inhibition of Bcl-2 by activated JNK leads to oligomerization of pro-apoptotic proteins Bax and Bak, which under basal conditions are found sequestered at the mitochondrial outer membrane by Bcl-2. Once Bax and Bak oligomerize, they insert into the mitochondrial membrane and breach membrane integrity, which results in a net efflux of cytochrome c from the mitochondria to the cytosol (103). A number of cytochrome c molecules bound to Apaf-1 form apoptosomes, which in turn bind and activate caspase-9 and subsequently lead to apoptosis. The IRE1-TRAF2 complex also activates ATF3, a transcriptional repressor, through JNK (104, 105). Dominant negative TRAF2 as well as antisense ATF3 cdna inhibited cell death induced by homocysteine, which alters cellular redox potential causing ER stress (105). This result implicates the JNK-ATF3 pathway in ER stress-induced cell death. Alternatively, in response to ER stress, the IRE1-TRAF2 complex also interacts with procaspase-12, and allows their clustering and activation (106, 107). Cleaved caspase-12 in turn activates caspase-9, and consequently activates a caspase cascade resulting in apoptosis (31, 108). Furthermore, IRE1α has been shown to mediate degradation of mrnas under chronic ER stress conditions (109). This is also thought to contribute to the switch of protective UPR to ER stress induced apoptosis by the IRE1α protein. The endoribonuclease activity of IRE1α becomes non-specific under conditions of ER stress which causes endonucleolytic decay of insulin mrna (110, 111) and 16

28 also the mrnas of various proteins targeted to the secretory pathway (58, 60, 112). Thus, depending on the forms of ER stress, β-cells may generate binary signalling of life and death through IRE1α Activation of Caspases Activation of caspases, or cysteine-aspartic proteases, is one of the most common biomarkers of apoptosis (113). Caspases rapidly dismantle cell cycle, cytoskeletal, and organelle proteins by proteolytic cleavage. The signalling cascades that control caspasedependent apoptosis can be classified into two pathways: the intrinsic (or mitochondrial) death pathway and the extrinsic cell death (or death receptor) pathway. The death-receptor pathway is induced by ligand-mediated activation of death receptors on the plasma membrane. The alternative pathway for caspases activation is mediated by cellular stress, for example, ER stress. In mice, the procaspase-12 is localized on the cytoplasmic side of the ER and is cleaved and activated specifically by ER stress, but not by death receptor- or mitochondria-mediated apoptotic signals (107, 114). Activation of caspase-12 occurs in response to perturbation in ER Ca 2+ pools, which activates calpains, a family of Ca 2+ -dependent cysteine proteases, in the cytosol, which in turn converts procaspase-12 to its active form, caspase-12 (115). Active caspase-12 then initiates a caspase cascade that results in the release of cytochrome c from the mitochondria into the cytoplasm. Cytochrome c induces the formation of the apoptosome complex, which recruits and activates procaspase-9. Activated caspase 9 cleaves and activates caspase-3, which in turn activates downstream death events, such as poly (ADP-ribose) polymerase (PARP), culminating in apoptosis of the cell. Caspase-7, which translocates from the cytosol to the cytoplasmic side of the ER membrane in response to ER stress, has been reported to interact with and cleave caspase-12, leading to its activation (116). This study showed that a dominant negative catalytic mutant of caspase-7 inhibited caspase-12 activation and cell death (116). In addition to the role of calpains and caspase-7 in the activation of caspase-12, TRAF2 has been shown to promote the clustering of procaspase-12 at the ER membrane (101). The interaction between TRAF2 and procaspase-12 is inhibited by ER stress conditions or by overexpressing IRE1. Therefore, it has been proposed that during ER stress, caspase-12 activation requires the dissociation of procaspase-12 from TRAF2, which may subsequently be recruited to IRE1 (101). Although caspase-12 is activated during ER stress, the 17

29 involvement of caspase-12 in ER stress-induced apoptosis is still controversial. Initial reports on caspase-12 -/- mice and MEFs showed resistance to apoptosis in response to ER stress (107). In contrast, a recent study by Saleh et al. (117) reports that caspase-12 -/- mice are not protected from cell death induced by ER stress. Instead, caspase-12 has been suggested to play a role in inflammation. Furthermore, Sanges et al. (118) have proposed that ER stress-induced apoptosis is mediated by calpains, but not by caspases, based on the observation that calpain inhibitors, but not a pan-caspase inhibitor, block tunicamycin- and thapsigargin-induced apoptosis. Finally, humans lack functional caspase-12 due to the presence of a frameshift mutation that results in premature stop codons in the gene (119). However, Hitomi et al. (120) have proposed that caspase-4, which is homologous to mouse caspase-12, performs the function of caspase-12 in humans. Caspase-4 cleavage is specifically induced by ER stress, but not by other apoptotic signals and knock-down of caspase-4 decreases ER stress-induced apoptosis. Although studies have shown that caspase-12 can activate caspase-9 (121), the precise function and downstream targets of caspase-12 requires further study. 1.4 ER stress as causative factor in β-cell dysfunction and death in diabetes Due to several important functions served by the ER, its proper functioning in β-cells is essential for survival, and if this operation is disrupted, β-cell viability is compromised. There is accumulating evidence that indicates that β-cell loss in T2D results from intertwined stress responses of gluco-/lipotoxicity, oxidative stress, and ER stress (33, ). The first clue to the involvement of ER stress in diabetes comes from studies on the Akita mutant in both mouse and human. A spontaneous diabetic model, the Akita mouse is characterized by progressive hyperglycemia with reduced β-cell mass without insulitis or obesity (125, 126). This diabetic phenotype is caused by a missense mutation in the insulin 2 (Ins2) gene (Cys96Tyr), which disrupts a disulfide bond between the A and B chains of proinsulin, and therefore causing incorrect folding of the proinsulin 2 protein in the ER. As a result of this mutation, the proinsulin is retained in the ER and gets degraded by ERAD. Since proinsulin is a major ER client protein in the pancreatic β-cells, the accumulation of the mutant insulin causes β-cell dysfunction and death mainly through ER stress as evidenced by upregulated expression of ER stress markers such as GRP78, sxbp1, ATF6αp50 and CHOP in the Akita mutant in islets and β-cell lines (125, 127, 128). 18

30 Another example of ER stress serving as the cause to a diabetic phenotype is the Wolcott-Rallison syndrome, which is characterized by early infancy diabetes, with endocrine and exocrine insufficiency and pancreatic atrophy with reduced number of β-cells ( ). This syndrome is caused by loss of function mutations in the Perk gene (96, 132), in which Perk -/- cells are unable to phosphorylate eif2α and attenuate translation. As a consequence, when cells are challenged with high glucose, they produce considerably larger amounts of insulin and there is accumulation of unfolded client protein (e.g. proinsulin) in excess of chaperone capacity, resulting in increased oxidative stress and apoptosis ( ). Further studies outlining the importance of translational control to β-cell viability come from murine studies that demonstrate that mice with a homozygous knock-in mutation at the PERK phosphorylation site in eif2α, whereby an alanine is substituted for the serine residue at position 51, have defects in β-cell survival, liver glycogen storage and inhibition of translation under ER stress conditions and as a consequence die from post-natal hyperglycemia (96, 133, 136). Alternatively, enhanced eif2α phosphorylation by pharmaceutical agents such as Salubrinal that inhibit eif2α dephosphorylation leading to reduced translational rate, also results in β-cell dysfunction (137). Moreover, whole body Perk null mice develop normally but present with a diabetic phenotype soon after birth due to β-cell death. Due to uninhibited translation in these mice given a glucose challenge, there was more proinsulin production, causing them to experience higher translational loads and therefore higher levels of ER stress. These studies demonstrate the relationship between ER stress and β-cell function and suggest that ER overload and unresolved ER stress may cause β-cell death. These results suggested that translational regulation through eif2α phosphorylation is required to maintain functional integrity of the ER. Furthermore, the Wolfram syndrome (WFS) is a rare autosomal recessive neurodegenerative disorder characterized by early onset diabetes, optic atrophy and hearing impairment (138). This disorder is caused by loss-of-function mutations in the Wfs1 gene that encodes the Wolframin protein (139, 140). Although the Wfs1 gene is not a direct sensor of UPR, mice deficient is the Wfs1 gene develop glucose intolerance and overt diabetes due to insufficient insulin secretion. Furthermore, these mice experience higher levels of ER stress as monitored by phosphorylation of eif2α and Xbp1 splicing. Additionally, hyperactivation of ATF6α in β-cells may be suppressed by Wolframin by stabilizing HRD1, which brings ATF6α 19

31 to the proteasome (141). Therefore, WFS1 may have a role as a negative regulator of chronic or unresolvable ER stress. In addition to being prone to ER stress as a result of the demands of insulin production, β-cells are also particularly susceptibility to oxidative stress, which can also cause ER stress. Pancreatic β-cells express relatively low levels of antioxidant enzymes (copper-zinc superoxide dismutase, manganese superoxide dismutase, catalase, and glutathione peroxidase), which may increase their susceptibility to oxidative stress and ER stress (142). Moreover, high glucose exposure leads to increased reactive oxygen species (ROS) levels, further reducing β-cell function and enhancing ER stress (123, 143). The high glucose levels lead to high insulin demand, which along with other environmental stressors leads to ER stress. This limited capacity to deal with oxidative stress and excessive ER stress can lead to premature β-cell aging and dysfunction, thereby resulting in the onset of type 2 diabetes. Additionally, our lab has recently demonstrated that HFD-induced ER stress induces β- cell dysfunction and glucose intolerance in control mice, while transgenic mice overexpressing the chaperone GRP78 in pancreatic β-cells were protected against this insult (144). Ultimately, the study implicated ER stress in obesity-induced β-cell dysfunction and showed that increased chaperone capacity in β-cells provides protection against the pathogenesis of obesity-induced type 2 diabetes. 1.5 Rationale and Hypothesis For pancreatic β-cells, the physiological demand for insulin synthesis in response to nutrients constitutes a constant source of stress on the ER, making these cells especially sensitive to disruptions in ER homeostasis. The ER of β-cells is further challenged under pathophysiological conditions, such as obesity and diabetes, in which elevated free fatty acids and hyperglycemia stimulate the β-cell to greatly increase insulin protein production. Thus, improper functioning of the ER can compromise β-cell function and viability, and an understanding of the mechanisms used by the β-cell as a defense against the stress is essential. The unfolded protein response has been studied extensively in the context of diabetes as well as other diseases (145). Although the ATF6α protein has been shown to be a critical component of the UPR that is required by the β-cell to elicit a full ER stress response and to maintain cell survival (79), very little is known about the ATF6β protein. Thus, the objective of this thesis 20

32 was to identify the role of the ATF6β protein in the UPR of pancreatic β-cells. Previous studies have demonstrated that both ATF6α and ATF6β are constitutively synthesized as type II transmembrane glycoproteins embedded in the ER that are proteolytically processed in response to ER stress, allowing their translocation into the nucleus. Furthermore, the N-terminal regions of both isoforms are structurally conserved and contain all the features required for an active transcription factor i.e., DNA binding, dimerization, and activation domains. Therefore, I hypothesized that ATF6β would be expressed in pancreatic β-cells and would play a role in the UPR as an ER stress sensor protein. Consequently, I further hypothesized that depletion of ATF6β would compromise the β-cell s ability to respond to ER stress and decrease cell viability, while ATF6β overexpression would improve β-cell survival under ER stress conditions. 21

33 CHAPTER 2: MATERIALS AND METHODS 22

34 2.1 Cell culture Rat INS-1 (obtained from Dr. Claus Wollheim, University of Geneva, Switzerland (146)) and INS-1 832/13 (obtained from Dr. Chris Newgard, Duke University, Durham, NC, (147)) insulinoma cell lines were maintained in RPMI 1640 (11.1 mm glucose, 1 mm sodium pyruvate, and 10 mm HEPES) supplemented with 10% FBS, 2 mm L-glutamine, and 55 µm β- mercaptoethanol, containing antibiotics (100 U/ml penicillin and 100 µg/ml streptomycin) at 37 C and 5% CO 2. Mouse MIN6 insulinoma cells (obtained from Dr. Michael Wheeler, University of Toronto, Toronto, ON, Canada) and HeLa cells were cultured in DMEM (25 mm glucose, 2 mm L-glutamine, and 1 mm sodium pyruvate) supplemented with 10% FBS and 55 µm β-mercaptoethanol at 37 C and 5% CO 2. Media was replaced every 3-4 days and cells were passaged once they reached ~70% confluence. Human islet samples were obtained from Dr. Patrick Macdonald (Alberta Diabetes Institute, University of Alberta). Mouse and rat pancreatic islets were isolated by Liling Zhang (lab technician). Briefly, the pancreas was initially chopped into small pieces using scissors in Hanks Balanced Salt Solution (HBSS) on ice, then placed in an HBSS solution with 3 mg/ml collagenase P on a rotor in a 37 o C water bath for 12 min and shaken for an additional 3 min by hand at 37 o C. The digested pancreas was then placed on ice and centrifuged at rpm for 10 sec three times to clear the supernatant. The pellet of islets was resuspended in cold HBSS and islets were hand-picked for experiments. Isolated islets were cultured in RPMI 1640 media (7.5% FBS, 10 mm HEPES, 5.6 mm glucose) at 37 o C/5% CO 2 and treated as described in figure legends. All animal procedures were approved and were performed in accordance to the Animal Use Protocols at the Toronto Centre for Phenogenomics. 2.2 Cell Treatments and Lysis For cell lysate preparation, cells were grown in 6-well plates and treated as indicated in the figure legends. After the appropriate cell treatments and incubations, cells were placed on ice and washed once with ice-cold phosphate-buffered saline (PBS). Cells were then incubated in lysis buffer (1% Triton X-100, 100 mm KCl, 20 mm HEPES, 2 mm EDTA, ph 7.3) containing 0.5 mm PMSF and protease inhibitors (Roche) on ice for 20 min, with occasional shaking. Lysates were collected and centrifuged at 13, 200 rpm for 10 min at 4 C. The supernatant was collected and protein concentration was determined using the bicinchoninic acid (BCA) protein assay (Pierce Inc.). In the case of cell lysate preparation for blotting with anti-atf6β antibody, 23

35 after washing with PBS, cells were scraped in 500 µl of ice-cold PBS (containing protease inhibitor cocktail and 10 µm MG132) and collected by centrifugation at 5, 000 rpm for 2 min. Cell pellets were then lysed in 100 µl SDS-sample buffer (50 mm Tris/HCl, ph 6.8, containing 2% SDS, 10% glycerol, protease inhibitor cocktail and 10 µm MG132). Cells were mixed by vortexing, boiled for 5 min and vortexed again. Lysates were then centrifuged at 13, 200 rpm for 2 min and the supernatant was collected and used to determine protein concentration using the BCA protein assay (Pierce Inc.). 2.3 Western Blot Analysis Cells were lysed as described above and lysates were used to determine protein concentration using the BCA protein assay (Pierce). Equal amounts of protein were boiled in 2x SDS sample buffer supplemented with 10% β-mercaptoethanol. Proteins were resolved on 10% SDS-PAGE gels and transferred to nitrocellulose membranes. Alternatively, proteins were resolved using 4 12% NuPAGE gels (Invitrogen) and transferred to Hybond-ECL nitrocellulose membranes (GE Healthcare). Membranes were incubated with blocking solution consisting of 3% skim milk powder in wash buffer (0.05% Nonidet P-40, 0.05% Tween 20 in PBS, WB) for 1 h at room temperature with shaking. Subsequently, blots were incubated overnight at 4 C on a shaker in primary antibodies. The following primary antibodies were used: α-atf6β (obtained from Dr. Kazutoshi Mori, Kyoto University, Kyoto, Japan), α-flag (1:1,000, no. F3165, Sigma), α- PARP (1:500, no. 9542P, Cell Signaling), α-cleaved caspase 3 (1:500, no. 9661S, Cell Signaling), α-γ-tubulin (1:1,000, no. T6557, Sigma), α-phospho-eif2α (1:250, no. 9721, Cell Signaling), GM130 (1:500, no. G65120, BD Transduction Laboratories), phospho-p38 (1:500, no. 9215L, Cell Signaling), total-p38 (1:1000 no. 9212, Cell Signaling), phospho-jnk (1:500, no. 9251, Cell Signaling), total-jnk (1:1000, no. 9252, Cell Signaling). The next day, primary antibodies were removed and the membranes were washed 3 times (15 min each) with WB on a shaker at room temperature. Secondary antibody conjugated to HRP was then added for 1 h at room temperature on a shaker, and then blots were washed again 3 times (15 min each) with WB. The enhanced chemiluminescence detection kit (RPN2106, Amersham Bioscience) was used for detection. Densitometry was performed to quantitate band intensities relative to the loading control. The photographic films were scanned and the relative band darkness was measured using Image J software. 24

36 2.4 Short-interfering RNA-mediated Knock-down Knock-down was achieved using a transfection protocol with RNAiMAX (Invitrogen) according to the manufacturer s instructions. Briefly, 24 pmol sirna was diluted in 400 µl OptiMEM directly into each well of a 6-well plate, mixed gently and incubated at room temperature for 5 min. After incubation, 4 µl of Lipofectamine RNAiMAX reagent was added to each well containing the diluted sirnas, mixed gently and incubated at room temperature for 30 min. INS-1 832/13 cells trypsinized and resuspended in RPMI growth media, and counted to a final concentration of 300, 000 cells/ml. Two milliliters of cell-containing media was added to each well (i.e. 600, 000 cells/well) on top of the sirna complexes to obtain a final sirna concentration of 10 nm. The cells were mixed gently and incubated at 37 C for 48 h or 72 h. GFP sirna was used as control. Short interfering RNAs targeted to ATF6β (sirna#1, #2 and #3) and GFP were obtained from Invitrogen. sirna targeting GFP was used as control. Knockdown was confirmed using western blotting for ATF6β using an ATF6β specific antibody. 2.5 RNA Isolation After cell treatments, media was removed and cells were washed once with PBS at room temperature. Total RNA was isolated by adding 1 ml TRIzol reagent (Invitrogen) to each well. After incubating cells for 5 min at room temperature, lysates were collected into Eppendorf tubes and 200 µl chloroform was added. Tubes were shaken vigorously for 15 s and allowed to sit at room temperature for 2-3 min. Tubes were then centrifuged at 11, 400 rpm for 15 min at 4 C. Isolated RNA was then extracted using the RNeasy RNA Isolation Kit (Qiagen) as per the manufacturer s instructions. RNA concentration and integrity were measured using a Nanodrop machine. 2.6 Reverse Transcription PCR To determine mrna expression of ATF6β, reverse transcription PCR (RT-PCR) was employed following total RNA isolation from MIN6 and INS-1 832/13 insulinoma cells and mouse islets. Human islet RNA samples were provided by Dr. Patrick Macdonald (Alberta Diabetes Institute, University of Alberta). RT-PCR (Qiagen OneStep RT-PCR kit) was used to amplify ATF6β cdna using primers specific for ATF6β: 25

37 For rat and mouse samples: MIN6, INS-1 832/13 cells and mouse islets: ATF6β Forward: 5 atg gcg gag ctg atg ctc ctct 3 ATF6β Reverse: 5 tcc tgt ttc cag acc cca gct 3 For human samples: HeLa cells and human islets: ATF6β Forward: 5 atg gcg gag ctg atg ctg ctc 3 ATF6β Reverse: 5 cag acc cta act tga gct cgc 3 The following experimental conditions were used for the RT-PCR: 50 C (30 min); 95 C (15 min); 30 cycles of 94 C (30 s), 53 C (30 s), 72 C (1.5 min); 72 C (10 min). RT-PCR products were resolved on a 1% agarose gel and visualized using ethidium bromide. The RNA concentration used for cell line samples was 1 µg and 500 ng for islet samples. 2.7 Real-Time Quantitative PCR Total RNA was reverse transcribed to single-stranded cdna using the High-Capacity cdna Reverse Transcription kit (Applied Biosystems). The resulting cdna was used for real-time PCR analysis using the TaqMan Gene Expression system (Applied Biosystems) for the indicated genes. Gene specific primers were obtained from Applied Biosystems: rat Grp78 (Rn _gl), Grp94 (Rn _ml), Herp (Rn _ml), Hrd1 (Rn _gl), Chop (Rn _gl), Txnip (Rn _gl), Atf6α (Rn _ml), and Ins2 (Rn _gl). Similarly, to quantitate genes upregulated in response to overexpression of ATF6βp60 by Ad-ATF6βp60, the following primers were used from Applied Biosystems: rat Hmox1 (Rn _m1), Atf3 (Rn _m1), Chop (Rn _gl), Herp (Rn _ml), Sdf2l1 (Rn _m1), and Dnajb9 (Rn _m1). The β-actin specific primers (rat β-actin, E) and TaqMan probes were also obtained from Applied Biosystems. A standard curve was generated using serial dilutions of the GFP sirna with no treatment cdna. A complete real time PCR reaction (25 µl) consisted of 10 µl of cdna, 1.25 µl of TaqMan gene expression primers (20x), 1.25 µl of double-distilled H 2 O, and 12.5 µl of TaqMan Universal PCR Master Mix (2x). Samples were loaded into an ABI PRISM 96-well optical reaction plate. Reactions were run on an ABI Prism 7900HT Sequence Detection System using the following protocol: 10 min at 95 o C, 40 cycles of 15 s at 95 o C, and 1 min at 26

38 60 o C. The standard curve and corresponding values from each sample were determined by the SDS 2.1 software of the ABI Prism 7900HT instrument. Values were normalized to expression of β-actin mrna and presented as ±S.E. of a minimum of three independent experiments. 2.8 XBP1 Splicing Assay Total RNA was first isolated using Trizol reagent (Invitrogen) followed by isolation using the RNeasy Mini Kit (Qiagen). Rat XBP1 cdna was amplified by RT-PCR (Qiagen OneStep RT- PCR kit, no ) using primers that flank the intron excised by IRE1 endoribonuclease activity as described previously (148). RT-PCR was conducted using the forward XBP1 Primer 5 - AAA CAG AGT AGC AGC ACA GAC TGC-3 and reverse XBP1 primer 5 -TCC TTC TGG GTA GAC CTC TGG GAG-3. The following RT-PCR protocol was used: 50 C (30 min); 95 C (15 min); 30 cycles of 94 C (1 min), 62 C (1 min), 72 C (1 min); 72 C (10 min). The RT-PCR products were resolved on a 3% agarose gel and visualized using ethidium bromide. 2.9 Apoptosis Assay To quantitate apoptotic cell death caused by ATF6β depletion, sirna targeting ATF6β or GFP were used to knock-down the proteins as described above (sirna-mediated knock-down). Approximately 200, 000 cells were seeded per well on 12-well plates and treated with 2 µg/ml tunicamycin for the last 16 h of the experiment. Cell death was measured using the Roche Cell Death Detection ELISA PLUS assay according to the manufacturer s instructions (Roche). Briefly, following treatments, cells were lysed and oligonucleosomes in the cytoplasm (indicative of apoptosis-associated DNA fragmentation) were quantified via antibodies directed against DNA and histones. Samples were normalized to the sample from cells transfected with control GFP sirna Generation of ptet-on doxycycline inducible cell line The FLAG-ATF6βp60 cdna was excised from the pcdna3.1 (-) plasmid (Invitrogen) by restriction digest at the NheI and HindIII (New England Biolabs) sites and ligated into the ptre- Tight vector using the DNA ligation kit (TaKaRa, TAK6021). To generate a double stable pancreatic β-cell line with doxycycline inducible expression, the ptre-tight FLAG- ATF6βp60 plasmid and a hygromycin resistance plasmid (ptk-hygro) (BD Biosciences 27

39 Clontech; Cat. No.: ) were co-transfected into ptet-on INS-1 #46 (previously generated in our lab) cells using electroporation in a 0.4 cm cuvette (BioRad) in a 20:1 ratio. After electroporation, the cells were kept at room temperature for 10 min and then seeded into four 10 cm dishes containing RPMI 1640 complete media. After incubating at 37 C 5% CO 2 for 72 h, the cells were replaced with fresh media containing selective antibiotics (200 µg/ml Geneticin (G418) (Invitrogen, ) and 50 µg/ml Hygromycin B (Invitrogen, ) and 0.25 µg/ml fungizone (Invitrogen, )). The cells were allowed to grow in the presence of G418 and Hygromycin B in media that was changed every 3 to 4 days until individual stable clones were obtained. Fifteen resistant clones were isolated into separate wells of a 12-well plate using small pieces of filter paper soaked with trypsin (Invitrogen) and allowed to grow in the presence of G418 and Hygromycin B. When a few cell clumps were visible under the light microscope, the clones were transferred to a 25 cm 2 flask and continued to grow under G418 and Hygromycin B selection. When cell density reached approximately 70%, the cells were analyzed for doxycycline regulated FLAG-ATF6βp60 expression by Western blotting and immunofluorescence before and after induction with doxycycline (2 µg /ml) for 24 h Cloning of FLAG-ATF6βp60 into pshuttle- IRES-hrGFP-2 vector and Adenovirus production To produce a recombinant adenovirus expressing the active form of ATF6β (Ad-ATF6βp60), the cdna sequence of the 3X-FLAG-ATF6β, encompassing amino acids 1 to 392 (active ATF6βp60) of human ATF6β in the pcdna3.1 (-) plasmid (Invitrogen) was digested using NheI and HindIII restriction enzymes. The insert was then ligated into the pegfp-n1 plasmid after digestion at the same restriction sites. The pegfp-n1-flag-atf6βp60 plasmid and the pshuttle-ires-hrgfp-2 plasmid were single digested twice at restriction sites SalI and NheI. Subsequently, the FLAG-ATF6βp60 insert was ligated into the pshuttle-ires-hrgfp-2 plasmid using the Takara Solution I. This construct was transformed into DH5α cells and plated on kanamycin plates. The following day, resistant clones were picked and amplified overnight. The QIampDNA Mini Kit was used to isolate plasmid DNA and two plasmids were selected for sequencing (ACTG Corp., Toronto, ON). Once the correct sequences were confirmed, the two plasmids were amplified using the Midiprep Qiagen kit. The pshuttle-ires-hrgfp-2-flag- ATF6βp60 plasmid was then linearized with the PmeI restriction enzyme; prior to this step, a 28

40 second PmeI site was identified in the plasmid sequence and was mutated using the QuikChange II XL Site-Directed Mutagenesis kit as per the manufacturer s instructions (Stratagene, no ). Once the plasmid was linearized, it was transformed via electroporation into the BJ1853 cells (Stratagene) in order to induce homologous recombination with the stably transformed viral plasmid (Ad-plasmid). The new recombinant pshuttle-ires-hrgfp-2-flag- ATF6βp60 plasmid was then amplified in the ultracompetent XL10-Gold bacteria cells. The amplified plasmid was subsequently linearized with the PacI restriction enzyme and transfected into the AD-293 cells that express the early enzyme E1 that is required for viral production. The Luo et al. protocol was followed for infection of AD-293 cells with the recombinant adenovirus using lipofectamine (149). After the primary viral stock was prepared by freeze-thawing AD- 293 cells, viral particles were subjected to four rounds of amplification. The amplified adenovirus was purified using the Vivapure AdenoPACK 100 kit (Vivascience) as per the manufacturer s instructions. The purified virus was resuspended in buffer (20 mm Tris/HCl, 25 mm NaCl, 2.5% glycerol, ph 8.0), and aliquots were stored at -80 C. The purified viral titer was determined by measuring the absorbance at 260 nm (OD260) and determined to be 1.5x10 12 opu/ml (optical particle units/ml). Ad-GFP was used as a control for all adenovirus infection experiments. Production of Ad-GFP is previously described (150) INS-1 832/13 Cell Infection with Adenovirus Rat INS-1 832/13 cells were seeded (600, 000 cells/well) in a 6-well plate 24 h prior to infection with adenovirus. In order to optimize adenovirus infection conditions, the next day INS-1 832/13 cells were infected with increasing concentrations (1x10 6 1x10 10 opu/ml) of Ad-GFP and Ad-ATF6βp60 and incubated at 37 C in 5% CO 2 for 2 h with gentle shaking every 30 min. After 2 h, adenovirus-containing media was removed and cells were washed once with PBS and replaced with fresh RPMI 1640 medium. Subsequently, the 1x10 8 and 1x10 9 opu/ml virus concentrations were used to determine the optimal infection time point at (0, 4, 6, 8, 12, 24 h) for Ad-GFP, Ad-ATF6αp50 and Ad-ATF6βp60. Immunofluorescence staining and Western blotting experiments suggested that the 1x10 8 opu/ml concentration was optimal for the Ad- ATF6βp60 infection, while the 1x10 9 opu/ml concentration was adequate for Ad-ATF6αp50 expression at the 24 h infection time point. Thus, these viral concentrations were employed for infection of cells for the initial microarray analysis experiment. For the Western blotting 29

41 experiments used to determine the effect of overexpression of ATF6βp60 in preventing cell death under ER stress conditions, after culturing for 24 h, the cells were left untreated or treated with 2 µg/ml tunicamycin for the last 8 or 16 h, or with 1 µm thapsigargin for the last 6 h, as indicated in the figure legends Immunofluorescence Staining INS-1 832/13 cells were seeded on glass coverslips in 12-well dishes (400, 000 cells/well). After an overnight incubation, the cells were infected with adenovirus as indicated in the figure legends. After treatments, the cells were washed twice with PBS and fixed in 4% paraformaldehyde (PFA)-PBS for 20 min at room temperature. For GFP detection, after incubation in PFA, cells were washed twice and then mounted onto glass slides using Fluoromount G mounting medium (Electron Microscopy Sciences, Inc., Hatfield, PA), and allowed to dry overnight before imaging using a confocal fluorescence microscope (Olympus, IX71). For immunofluorescence, the PFA was removed and cells were washed 3 times with PBS, followed by incubation in 100 mm glycine in PBS for 15 min at room temperature. The cells were washed once and subsequently permealized with 0.1 % Triton X % BSA in PBS at room temperature for 15 min. The cells were then washed 3 times with PBS and blocked with solution containing 2 % non-fat dry milk/2 % BSA in PBS for 1 h at room temperature. Primary antibody was added in blocking solution at 1:500 for α-flag (no. F3165, Sigma). Following primary antibody incubation, cells were washed 3 times at 5 min each with PBS. Secondary antibody to Alexa Fluor 594 labeled anti-mouse (1:1000, Molecular Probes, Inc., Eugene, OR, USA) was then added in blocking solution for 1 h in the dark. Cells were washed 3 times (5 min each) with PBS in the dark. DAPI was added for 5 min in the dark prior to mounting and imaging for nuclear staining. Selected images were pseudo-colored for presentation in Adobe Photoshop software Microarray analysis Microarray expression profiling was used to assess the global transcriptional changes in response to ATF6βp60 expression in INS-1 832/13 cells. Cells were infected with adenovirus expressing GFP as control or adenovirus expressing FLAG-ATF6βp60 or FLAG-ATF6αp50 (79) for 24 h as described (Section 2.12). Total RNA was isolated using Trizol Reagent 30

42 (Invitrogen) followed by isolation using an RNeasy Mini Kit (Qiagen). Assessment of RNA quality and microarray analysis was performed at the University Health Network (Toronto) Microarray Centre ( Briefly, following RNA quality assessment with an Agilent BioAnalyzer, samples were reverse transcribed to complementary DNA (cdna). cdna was purified with a cdna purification module from Affymetrix. Biotin was incorporated during in vitro transcription and purified complementary RNA (crna) was then fragmented with a chemical reaction involving zinc acetate. Labeled and fragmented crna (15 µg) was hybridized to Rat Genome arrays (Affymetrix Genechip) for 17 h at 45 C at 60 rpm speed. The arrays were stained and washed using GeneChip Operating Software (GCOS) and fluidic stations, which stain the GeneChips with phycoerythrin-labeled Streptavidin (SAPE), washes them, and then incubates the GeneChips with biotinylated anti-streptavidin antibody solution. All arrays are scanned using Affymetrix's G scanner. The data sets were analyzed using GeneSpring software (Version 7.1) (Agilent Technologies) and statistical analysis was performed using GeneSpring software according to the manufacturer s specifications. Genes with a minimum 2-fold difference between control and Ad-ATF6αp50 or -ATF6βp60 infected cells are listed in Table 1 (Supplementary Data) Data analysis Results are presented as mean ±SE. Statistical significance between two experimental conditions was analyzed using a two-sample t-test assuming equal variance and between multiple conditions by ANOVA, followed by Tukey s test. p<0.05 was considered statistically significant. 31

43 CHAPTER 3: EXPRESSION AND ACTIVATION OF ATF6β IN PANCREATIC β-cells AND THE EFFECT OF ATF6β DEPLETION ON THE ER STRESS RESPONSE AND SURVIVAL OF PANCREATIC β- CELLS 32

44 3.1 Introduction It is becoming increasingly apparent that overstimulation of pancreatic β-cells in conditions of hyperglycemia and hyperlipidemia disrupts ER function which contributes to β- cell dysfunction and death, resulting in diabetes (151). As such, it is crucial to identify adaptive mechanisms that will allow pancreatic β-cells to survive under conditions of ER stress and improve their function and survival. As previously mentioned, the ATF6α protein contributes greatly to the UPR, increasing the chaperone capacity of cells to allow for adaptation to ER stress and thus preventing cell death. While ATF6α is thoroughly characterized in a number of cell lines, including in pancreatic β-cells by our group (79), very little is known about the ATF6β protein, especially in the context of pancreatic β-cells. Thus, in this study I first set out to identify whether ATF6β is expressed in pancreatic β-cells (various cell lines and islets) and whether it is activated under ER stress conditions induced by pharmacological ER stressors (e.g. thapsigargin, tunicamycin). As mentioned, due to the specialized secretory nature of pancreatic β-cells and the important role that the ER plays in maintaining a homeostatic environment in these cells, I hypothesized that ATF6β would be expressed under basal conditions and activated by ER stress as a component of the UPR. Furthermore, to characterize the role of ATF6β in the ER stress response of pancreatic β-cells, I knocked down endogenous ATF6β using sirna and measured the mrna expression of well-known UPR genes. In this manner, I anticipated that ATF6β would play an important functional role as a component of the ER stress response and its depletion would compromise the full induction of genes involved in the UPR. Additionally, I examined the effect of ATF6β depletion on pancreatic β-cell survival and predicted that depleting pancreatic β-cells of this isoform would sensitize the β-cells to ER stress-induced apoptosis. 3.2 Results ATF6β is expressed in pancreatic β-cell lines and islets and is activated by ER stress Previous studies have demonstrated that ATF6β is expressed ubiquitously at the mrna level, including abundant expression in the pancreas (66). However, these studies did not identify whether ATF6β is expressed in the insulin producing β-cells specifically. Thus, in order to elucidate this, expression of ATF6β was examined at the mrna level using RT-PCR 33

45 and at the protein level using Western blot analysis. ATF6β specific primers were used to examine the mrna expression of ATF6β in total RNA extracted from insulinoma cell lines (MIN6, INS-1 832/13) and mouse and human islets. As shown in Figure 3.1, ATF6β is expressed at the mrna level in both mouse and rat pancreatic β-cell lines (Figure 3.1A and B), in mouse islets (Figure 3.1B) and in the human islet sample (Figure 3.1C). Figure 3. 1 ATF6β mrna expression in pancreatic β-cells and islets. Total RNA was isolated from the indicated cell lines or islets and RT-PCR was performed to examine ATF6β mrna expression in HeLa cells and INS-1 832/13 (A) and in MIN6 cells and in mouse islets (B). Gels are representative of n=2 independent samples. RNA from a human islet sample was obtained from Dr. Patrick Macdonald (Alberta Diabetes Institute, University of Alberta) and ATF6β mrna expression was analyzed by RT-PCR, (n=1, C). I obtained an ATF6β specific antibody from Dr. Kazutoshi Mori (Kyoto University, Kyoto, Japan) in order to examine endogenous protein expression of ATF6β in pancreatic β-cell lines and isolated rodent islets. In HeLa cells, under normal conditions, the ATF6β protein is embedded in the ER membrane, existing in its full-length form (ATF6βp110). ATF6β is posttranslationally modified with five glycosylation sites, and the completely glycosylated fulllength form has an electrophoretic mobility of 110-kDa. Treatment of cells with tunicamycin, an inhibitor of N-linked glycosylation, leads to size reduction of the protein that exhibits an electrophoretic mobility faster than the fully glycosylated form (152). Under ER stress conditions induced by tunicamycin or thapsigargin, the full-length form is cleaved to release the N-terminal active form (ATF6βp60) (66). Protein expression of ATF6β and its activation in 34

46 response to ER stress in pancreatic β-cells has not been previously demonstrated. I hypothesized that similar to activation of ATF6β in HeLa cells, ER stress would induce the cleavage of full-length ATF6β to generate the active form (ATF6βp60) in pancreatic β-cells. To induce ER stress and activation of ATF6β, insulinoma cells (INS-1, INS-1 832/13, MIN6) and rat islets were treated with 300 nm thapsigargin (Tg) for 4 h or 2 µg/ml tunicamycin (Tn) for 16 h. HeLa cell lysates were used as a reference since these cells have been previously shown to express ATF6β and activation of ATF6β by tunicamycin and thapsigargin (66). The results show that under control conditions, ATF6β was synthesized as a 110-kDa precursor protein in all three pancreatic β-cell lines. As hypothesized, when cells were treated with thapsigargin or tunicamycin, the full-length was cleaved to generate the 60-kDa form (ATF6βp60) (Figure 3.2A), suggesting activation of the protein. I also examined ATF6β expression and activation in a more physiologically relevant model using isolated rat islets, which were either left untreated or treated with thapsigargin or tunicamycin to induce ER stress. In this preliminary experiment, the expression of ATF6β in the islets was detected under control conditions and induction of the active form under conditions of ER stress (Figure 3.2B). Furthermore, I wanted to examine whether pathophysiological conditions associated with diabetes such as high glucose or high saturated FFAs that are known to cause ER stress, would also induce activation of ATF6β. In preliminary experiments, INS-1 832/13 cells were treated with 5 mm or 25 mm glucose for 24 or 48 h to examine the effect of high glucose on ATF6β activation. The results in Figure 3.2C show that treating cells with high glucose (25 mm) for 24 or 48 h led to a reduction in protein levels of the full-length ATF6βp110, and presumably conversion to the active ATF6βp60 form, although detection of the cleaved 60-kDa product (ATF6βp60) was obscured by non-specific bands. Further experiments and improved reagents will be required to critically judge the presence and importance of the 60- kda ATF6β cleavage product under conditions of high glucose and saturated free fatty acids. Overall, these results demonstrate that ATF6β is expressed in pancreatic β-cell lines and rodent and human islets and is activated under ER stress conditions. 35

47 Figure 3. 2 ATF6β is expressed at the protein level and is activated by ER stress in pancreatic β- cells. Cell lysates were prepared from the indicated cell lines (n=2, A) or rat islets (n=1, B) either left untreated (control C) or treated with 300 nm thapsigargin for 4 h or 2 µg/ml tunicamycin for 16 h to induce ER stress. INS-1 832/13 cells were also treated with low (5 mm) and high (25 mm) glucose for 24 and 48 h (n=1, C). Expression of full-length ATF6βp110 was detected under control conditions and induction of the active form (ATF6βp60) under ER stress conditions using an ATF6β specific antibody Knock-down of ATF6β has no effect on the mrna expression of well-known UPR genes Having confirmed expression of ATF6β in pancreatic β-cell lines and activation under ER stress conditions, the next objective of the study was to determine whether active ATF6β regulates the expression of any known UPR genes. As previously mentioned, upon activation by ER stress, ATF6 translocates to the nucleus and binds ERSEs and UPREs to upregulate UPR genes such as chaperones and ERAD components. ATF6α and ATF6β are 82% similar in their 36

48 basic-leucine zipper domain, the region required for nuclear translocation for DNA binding and dimerization. Experiments by Haze et al. (66) have shown that similar to ATF6α, ATF6βp60 in HeLa cells is liberated from the ER membrane and translocates into the nucleus. Furthermore, once in the nucleus, ATF6βp60 was shown to bind directly to ERSE in the presence of the general transcription factor, NF-Y. As such, I hypothesized that ATF6β would alter its subcellular localization from the ER to the nucleus when processed in response to ER stress and induce expression of UPR genes. In order to test whether ATF6β is required for induction of UPR genes, I used a loss-of-function approach using sirna targeting ATF6β to investigate the effect of endogenous ATF6β depletion on UPR genes. The efficiency of inhibition of ATF6β expression at the protein level was tested after 24, 48 or 72 h of sirna treatment in each experiment by Western blot analysis of ATF6β expression. All three different sirnas (sirna#1, #2 and #3) were effective (slightly higher knock-down efficiency with sirna#2) in knocking down ATF6β when transfected for a minimum of 48 h, resulting in a 90% knock-down efficiency in INS-1 832/13 cells (Figure 3.3). Thus, INS-1 832/13 cells were transfected with control sirna (targeted to GFP) or ATF6β sirna#2 for 48 h and cells were either left untreated or treated with the ER stressor tunicamycin for 16 h. We have previously demonstrated that the ATF6α protein is a major contributor for maintaining Grp78 mrna levels in pancreatic β-cells under basal conditions and for induction of Grp78 mrna under ER stress conditions (79). Thus, I hypothesized that ATF6β also contributes to the induction of Grp78, since upregulation of the chaperone was not completely inhibited in cells transfected with ATF6α sirna, suggesting that other factors may also contribute to Grp78 expression. However, examination of Grp78 mrna expression by real-time PCR in cells transfected with ATF6β sirna revealed that ATF6β is not required for induction of this gene under basal or ER stress conditions (Figure 3.4A). Examination of another chaperone gene, Grp94, revealed the same trend, whereby ATF6β depletion had no effect on Grp94 mrna under basal or ER stress conditions in comparison to control cells (Figure 3.4B). The ATF6α isoform is also responsible for regulating the expression of ERAD components such as Herp (76, 78). We have also previously shown that ATF6α depletion results in a significant reduction in the mrna expression of Herp in pancreatic β-cells 37

49 undergoing ER stress (79). Thus, I wanted to examine whether knock-down of ATF6β would compromise Herp gene expression. The results in Figure 3.4C show that unlike the ATF6α isoform, when pancreatic β-cells are depleted of ATF6β, Herp mrna levels remained unchanged under control or ER stress induced by tunicamycin. I also examined another ERAD component, Hrd1. Similarly, ATF6β depletion did not significantly affect Hrd1 expression as there was only a trend towards reduced Hrd1 mrna levels in comparison to cells transfected with GFP sirna (Figure 3.4D). In addition to chaperones and ERAD genes, I also examined pro-apoptotic genes known to be induced in response to ER stress, including Chop and Txnip. The results show that ATF6β depletion has no effect on Chop mrna levels under control or ER stress conditions (Figure 3.4E). The Txnip gene has recently been suggested to be the critical signaling node in cells undergoing irremediable ER stress, whereby its activation by the PERK and IRE1 pathways causes procaspase-1 cleavage and interleukin 1β (IL-1β) secretion, thus resulting in pancreatic β-cell death (153, 154). Although these studies demonstrated that Txnip mrna remains unperturbed in ATF6α -/- MEFs, they did not examine whether Txnip expression is altered in ATF6β knockout MEFs. To this end, I depleted ATF6β from pancreatic β-cells and examined Txnip gene expression and its downstream target, Il1b. The results indicate that ATF6β does not affect the mrna levels of Txnip as there was no significant difference between cells transfected with control sirna and ATF6β sirna (Figure 3.4F). The pro-inflammatory cytokines (Il-1β, Il-1, Il-6) are primarily released by macrophages as part of the innate immune response. However, cells outside of the immune system also express Il-1β, including pancreatic β-cells (154), but Il1b mrna was not detected by real-time PCR. Using whole body knockout mice, Yamamoto and colleagues (76) demonstrated that mice lacking either ATF6α or ATF6β develop normally, suggesting that one isoform can compensate for the lack of the other. Thus, I knocked down ATF6β in INS-1 832/13 cells and monitored ATF6α mrna levels. An increase in ATF6α in the absence of ER stress when ATF6β is depleted from the cell was observed, although this did not reach statistical significance (Figure 3.4G). It is possible that longer periods of ATF6β depletion (e.g. 72 h) may be necessary for the ATF6α isoform to be upregulated to maintain a level of function in the cell that is equal to that of both isoforms. I also examined the effect of ATF6β depletion on the 38

50 ins2 gene since previous studies have suggested that the ATF6α isoform downregulates Ins2 gene expression under ER stress conditions in INS-1 cells (83) and both isoforms can bind to the Ins2 promoter (82). Although there appears to be a decrease in Ins2 mrna levels under control and ER stress conditions when ATF6β is knocked down, the results were not statistically significant (Figure 3.4H). Figure 3. 3 Knock-down efficiency of the ATF6β protein. INS-1 832/13 cells were transfected with three short interfering RNAs (sirnas) directed to rat ATF6β (#1, #2, #3) or with lipofectamine reagent (Lipo only), or control GFP sirna as indicated for 48 h. After 48 h, cells were lysed and 10 µg of protein was resolved on a 10% SDS-PAGE gel and immunoblotted with an ATF6β specific antibody and with γ-tubulin as a loading control. Representative of three independent experiments. 39

51 40

52 Figure 3. 4 Effect of ATF6β knockdown on ER stress response genes. INS-1 832/13 cells were transfected with control (GFP) or ATF6β sirna#2 for 48 h and treated with or without 2 µg/ml tunicamycin (Tn) for 16 h. Total RNA was isolated and real time PCR was performed to examine expression of the indicated genes (A-H, n=6, n=3 for Chop and Txnip, * denotes significance at p<0.05, One-way ANOVA, followed by Tukey s test) Effect of ATF6β depletion on β-cell susceptibility to ER Stress It has previously been shown that ER stress-induced translational attenuation via PERKmediated phosphorylation of eif2α is critical to protect cells from ER stress (37). I hypothesized that ATF6β depletion would lead to activation of the PERK and IRE1α pathways as a result of cells undergoing ER stress. To test this hypothesis, I monitored expression of downstream targets of PERK and IRE1, such as phosphorylated eif2α and spliced Xbp1, respectively. Under ER stress conditions PERK activation leads to eif2α phosphorylation, while IRE1α activation leads to splicing of its target, Xbp1. When comparing cells depleted of ATF6β and control cells, I found that cells lacking ATF6β were slightly more susceptible to ER stress, as indicated by increased levels of phospho-eif2α even under basal conditions, which were further enhanced under ER stress conditions induced by thapsigargin (Figure 3.5A). Levels of the loading control γ-tubulin were reduced in cells treated with thapsigargin since this ER stressor caused significant cell death. In contrast to the results obtained for phospho-eif2α, thapsigargin-induced splicing of Xbp1 occurs similarly in control and ATF6β depleted cells (Figure 3.6). Although preliminary, these results suggest that pancreatic β-cells lacking endogenous ATF6β are unable to induce a full ER stress response, and thus experience greater 41

53 ER stress levels. As a consequence, I hypothesized that these cells would be more susceptible to ER stress-induced apoptosis. Figure 3. 5 ATF6β depletion increases susceptibility of pancreatic β-cells to ER stress. INS-1 832/13 cells were transfected with control (GFP) or ATF6β sirna#2 for 72 h and treated with or without 1 µm thapsigargin for 6 h. Cell lysates were prepared and 10 µg of proteins were subjected to 10% SDS-PAGE. Immunoblots were performed with anti-phospho-eif2α and anti-γ-tubulin as a loading control (A). Blots are quantified in (B, n=3). 42

54 Figure 3. 6 ATF6β depletion does not affect levels of spliced Xbp1. INS-1 832/13 cells were transfected with control (GFP) or ATF6β sirna#2 for 72 h and treated with or without 1 µm thapsigargin for 6 h. Total RNA was isolated and RT-PCR was used to detect unspliced and spliced to Xbp1 mrna (n=3) ATF6β depletion increases susceptibility of pancreatic β-cells to ER-stress-induced apoptosis The UPR acts as a binary switch between cell survival and death and when ER stress becomes chronic and unresolvable, apoptotic pathways are favored (20). During prolonged ATF6β knock-down experiments (e.g. 72 h), I noticed that a large number of cells were dying in comparison to control cells. I hypothesized that ATF6β is required to maintain cell viability, and depletion of this factor from the cell would result in increased susceptibility to apoptosis. To test this hypothesis, I knocked down ATF6β and treated cells with the ER stressors thapsigargin (1 µm, 6 h) or tunicamycin (2 µg/ml, 16 h) and measured apoptotic markers such as cleaved-caspase-3 (CC3) and cleavage of its downstream target, poly (ADP-ribose) polymerase (cleaved-parp). Caspase-3 is the main effector caspase that cleaves the majority of the cellular substrates (e.g., PARP) in apoptotic cells (155). Procaspase-3, which is an inactive 35-kDa proenzyme, can be cleaved at a specific residue to produce active 12-kDa and 17-kDa subunits. A protein kinase inhibitor, staurosporine was used as a positive control for caspases-3 and PARP cleavage. The results show that activated caspase-3 was detected in the ATF6βdepleted cells at 72 h post-transfection in control conditions in comparison to GFP sirnatransfected cells. PARP is one of the main substrates of activated caspase 3 (156). The 113- kda uncleaved PARP protein can be proteolytically cleaved to produce an 85-kDa fragment in apoptotic cells. As expected, there was increased cleavage of PARP in ATF6β knock-down cells in comparison to control cells, indicating that ATF6β is required for maintaining cell viability even under non-stress conditions. The apoptotic phenotype in cells depleted of ATF6β 43

55 was more pronounced when cells were stressed with either thapsigargin (Figure 3.7A) or tunicamycin (Figure 3.7B). These data support the hypothesis that ATF6β deficiency in pancreatic β-cells disrupts cell homeostasis and leads to increased susceptibility to apoptosis, suggesting that ATF6β has a pro-survival role in pancreatic β-cells. To further confirm the protective effect of ATF6β on pancreatic β-cell viability, I used an independent apoptosis assay, the Cell Death ELISA kit that measures the relative amount of cytosolic DNA-associated histone complexes, which is indicative of cells undergoing apoptosis. In line with the Western blot results, there was a trend towards increased β-cell apoptosis with ATF6β knock-down relative to control in both basal and ER stress conditions induced by tunicamycin (Figure 3.8). In summary, these data demonstrated that ATF6β is required to maintain cell viability of pancreatic β-cells, in both basal and ER stress conditions. 44

56 Figure 3. 7 ATF6β depletion increases susceptibility of pancreatic β-cells to apoptosis. INS-1 832/13 cells were transfected with control (GFP) or ATF6β sirna#2 for 72 h and treated with or without 1 µm thapsigargin (A) for 6 h or 2 µg/ml tunicamycin (B) for 16 h. Lysates were prepared and 10 µg protein were subjected to 10% SDS-PAGE. Immunoblots were performed with anti-cleaved caspase-3 and anti-parp antibodies and anti-γ-tubulin for loading control. Lysates from staurosporine (1 µm, 2 h) treated cells were used as a positive control. Cleaved caspase-3 and cleaved-parp band intensities at basal conditions are quantified in (C) and (D) respectively, with relative expression normalized to GFP sirna (n=7, One-Way ANOVA, post hoc Tukey s test, **p<0.01). Band intensities for cleaved caspase-3 and cleaved-parp in cells treated with thapsigargin are quantified in (E) and (F), respectively with relative expression normalized to GFP sirna (n=7, One-Way ANOVA, post hoc Tukey s test, *p<0.05). Band intensities for cleaved caspase-3 and cleaved-parp for cells treated with tunicamycin are quantified in (G) and (H), respectively with relative expression normalized to GFP sirna (n=7). 45

57 Figure 3. 8 ATF6β knock-down sensitizes pancreatic β-cells to apoptosis. INS-1 832/13 cells were transfected with control (GFP) or ATF6β sirna#2 for 72 h and treated with or without 2 µg/ml tunicamycin for 16 h. The Cell Death ELISA assay was used to measure cytosolic DNA-associated histone complexes as an indicator of cell apoptosis (n=3). To examine the molecular mechanism by which ATF6β depletion results in increased susceptibility to apoptosis, I measured levels of apoptotic markers upstream of the terminal cleaved caspase-3 and PARP, including the JNK and p38 stress pathways. Western blot analysis was employed to monitor levels of phosphorylated JNK and p38. The results show that ATF6β depletion has no effect on activation of either JNK (Figure 3.9A) or p38 (Figure 3.9B), as levels of the phosphorylated proteins were unchanged in comparison to control cells in either basal or ER stress conditions. Other pathways need to be investigated to elucidate the mechanism by which ATF6β depletion promotes cell death (See Discussion). 46

58 Figure 3. 9 ATF6β depletion does not activate JNK and p38. INS-1 832/13 cells were transfected with control (GFP) or ATF6β sirna#2 for 72 h and treated with or without 1 µm thapsigargin for 6 h or with or with 2 µg/ml tunicamycin for 16 h. Cells treated with UV for 15 min served as positive control for activation of JNK and p38. Cell lysates were prepared and µg of proteins were subjected to 10% SDS-PAGE. Immunoblots were performed with anti-phospho- JNK and total-jnk (A, n=5) and anti-phospho-p38 and total-p38 (B, n=1 for Tg treated cells, n=2 for Tn treated cells) antibodies. 47

59 3.2.5 ATF6βp60 overexpression protects pancreatic β-cells against ER stress-induced apoptosis Since depletion of endogenous ATF6β results in cell death, I hypothesized that overexpression of ATF6β would have a protective effect, which would rescue the apoptotic phenotype. Thus, I infected INS-1 832/13 cells with an adenovirus expressing the active form of ATF6β (Ad-FLAG-ATF6βp60) and treated cells with the ER stressors, thapsigargin (1 µm, 6 h) or tunicamycin (2 µg/ml, 8 h) and measured levels of cleaved caspase-3 and cleaved- PARP. In support of my hypothesis, levels of cleaved caspase-3 and cleaved-parp were significantly reduced in cells overexpressing ATF6βp60 compared to control cells infected with Ad-GFP (Figure 3.10A, quantified in B and C). Figure ATF6βp60 overexpression protects cells against ER stress-induced apoptosis. INS-1 832/13 cells were infected with an adenovirus expressing GFP as control or ATF6βp60 for 24 h. The cells were treated with either 2 µg/ml tunicamycin (Tn) for 8 h or 1 µm thapsigargin (Tg) for 6 h to induce ER stress (A). Cell lysates (10 µg) were resolved on a 4-12% NUPAGE gel and immunoblots were performed using antibodies for anti-parp, anti-cleaved caspase 3 and anti-γ-tubulin as a loading control. Blots were quantified (B, C) using one-way ANOVA, followed by Tukey s test, **p<0.004 (n=4). FL, full-length. 48

60 3.3 Discussion Pancreatic β-cells are specialized secretory cells that produce a significant amount of insulin that must be efficiently regulated in order to maintain whole body homeostasis. The UPR serves as a defense mechanism against stress that the β-cell might experience as a result of insulin production. Thus, β-cells have acquired a highly developed and active ER, reflecting their role in folding, export and processing of newly synthesized insulin. Studies of mammalian Atf6α -/- and Atf6β -/- cells have shown that ATF6α, but not ATF6β, is primarily responsible for transcriptional induction of a cohort of ER proteins including chaperones, folding enzymes and ERAD components (76-78). As a result of ATF6α s potent transcriptional activity, the majority of studies have focused on ATF6α, although the role of ATF6β in the UPR has been studied in the human HeLa cell line (66, 80, 81), in MEFs (53, 76, 77) and in CHO cells (157). Thus, due to very limited studies on ATF6β, a number of questions remain regarding its activation by cellular stress and its role in the ER stress response. Thus, the objective of this thesis was to investigate the function of ATF6β in the pancreatic β-cells. I first examined ATF6β expression in the commonly used rodent pancreatic β-cell lines (MIN6, INS-1, and INS-1 832/13) and in human and rat islets. Haze et al. (66) have previously shown ATF6β mrna is ubiquitously expressed, including in the pancreas; however, expression within islets or the insulin-producing β-cells was not demonstrated. Here, I show that ATF6β is indeed expressed at the mrna (Figure 3.1) and protein level (Figure 3.2) in pancreatic β-cells and islets. Moreover, in line with our hypothesis that ATF6β participates in the UPR in pancreatic β-cells, I show that the active form of ATF6β (ATF6βp60) is induced under conditions of ER stress caused by the pharmacological agents tunicamycin and thapsigargin (Figure 3.2), a findings that supports the results obtained by Haze et al. (66) in HeLa cells. As shown in Figure 3.2A, the extent of ER stress-induced cleavage of ATF6β varies from the HeLa cell samples and the rodent cell samples. This is likely due to the fact that the ATF6β antiserum is specific to the human ATF6β protein (66) and the epitope used to generate the antibody has only a 74% identity with murine ATF6β. Furthermore, other studies have shown variable induction of cleavage of ATF6 depending on the cell line used as well as the ER stress inducer employed. Generally, cleavage of ATF6 is much more extensive in cells treated with DTT (a reducing agent) than in cells treated with thapsigargin or tunicamycin. In line with the findings of Nadanaka et al. (72) using Chinese hamster ovary (CHO) cells, I found that 49

61 thapsigargin was unexpectedly a weaker inducer of ATF6β cleavage than tunicamycin in pancreatic β-cells, unlike their similar effects in HeLa cells. This is due to the fact that the degree of activation is correlated with the extent of reduction. Disulfide bonded ATF6 is reduced on treatment of cells with not only DTT, but also the glycosylation inhibitor, tunicamycin. Since only the reduced monomer ATF6β reaches the Golgi in tunicamycin-treated β-cells, this reduced monomer is a better substrate for SIP, and thus we observe a greater cleavage product in β-cells treated with tunicamycin. Although ER stress induced by pharmacological agents such as thapsigargin and tunicamycin serves to impose stress specific to the ER, these treatments are not physiologically relevant. Our results in Figure 3.2 may suggest proteolytic processing of ATF6β in response to high glucose, as levels of the full-length ATF6βp110 were reduced under these conditions at both the 24 and 48 h. Evidence for ATF6 activation by physiological stimuli has also been shown by Nadanaka and coworkers (158), who observed reduction of ATF6 by glucose starvation. This type of stress occurs under a variety of circumstances in a living organism and serves to demonstrate ATF6 is indeed activated by physiological stress. The finding that ATF6β is expressed in islets and insulinoma cells and is activated by ER stress suggests that the induction of ATF6β serves a role in the UPR. Cleaved ATF6β is expected to translocate into the nucleus and potentially bind to ER stress elements (ERSE) within the promoters of target UPR genes such as Grp78, Chop, Herp and others, leading to their induction (75). Furthermore, ATF6βp60 has also been shown to form a heterodimer with sxbp1 and bind to the UPRE to induce transcription of ERAD components (76). To examine this possibility, I first employed a loss-of-function approach using sirna to knock-down endogenous ATF6β to analyze selected canonical target genes as well as β-cell viability under control and ER stress conditions. One of the first genes I examined was Grp78, which is a major target of the ATF6α and the IRE1α/sXBP1 pathways (76, 159). When comparing the ability of ATF6α +/+ MEFs to induce expression of Grp78 with ATF6α -/- MEFs, Adachi and colleagues (78) found that Grp78 mrna levels were induced 11-fold in wild-type MEFs, and mitigated to less than 5-fold in knock-out MEFs. Furthermore, we have previously demonstrated in pancreatic β-cells that ATF6α knock-down does not prevent Grp78 mrna induction in response to ER stress (79), suggesting other factors may also be responsible for its induction. Thus, I hypothesized that ATF6β could also mediate induction of Grp78 as was 50

62 previously shown in HeLa cells (66, 81). However, similar to results obtained by other groups using MEFs (53, 76, 77) which showed that the expression of chaperones or ERAD components were not dependent on ATF6β, the qpcr results demonstrate that ATF6β is not required for induction of Grp78 expression (Figure 3.4A). It is possible that the partial induction of Grp78 expression in cells transfected with ATF6α sirna is due to XBP1 (53). Thuerauf and colleagues (81) showed that ATF6β has considerably weaker transcriptional activity in comparison to ATF6α, and thus suggested that ATF6β acts as transcriptional repressor of ATF6α s transcriptional activity in a dominant-negative manner. Another mechanism that has been identified as a means of regulating ATF6α activity is via the WFS1 protein, which negatively regulates ATF6α by stabilizing the HRD1 protein, thus bringing ATF6α to the proteasome to be degraded to prevent its hyperactivation (141). Based on these studies, I hypothesized that perhaps in pancreatic β-cells, ATF6β may regulate ATF6α s transcriptional activity by regulating expression of Hrd1. Thus, I examined whether Hrd1 gene expression is altered in cells depleted of ATF6β in comparison to control cells. I found that although ATF6β depletion leads to a trend towards reduced levels of Hrd1 mrna, the results were not significant in comparison to cells transfected with GFP sirna (Figure 3.4D). These findings are in contrast with the results obtained by Thuerauf et al. (81), whereby sirna-mediated knock-down of ATF6β in HeLa cells led to an increase in tunicamycinmediated ERSR (ER Stress Response) gene induction. Furthermore, in line with the findings of Yamamoto et al. (76) using ATF6β -/- MEFs, the induction of Grp78 and other UPR genes was unaltered or not enhanced in our ATF6β knock-down β-cells. This indicates that ATF6β does not serve as a negative regulator of ATF6α, as previously suggested (81). It is possible that the effects of ATF6β knock-down are cell specific. Although I reported a 90% silencing efficiency of the ATF6β protein, it is also possible that residual ATF6β in these ATF6β knock-down experiments was sufficient to induce expression of target genes and maintain the phenotype to a level that cannot be distinguished from control cells. It is possible that pancreatic β-cells rely primarily on the ATF6α isoform for induction of chaperones and ERAD proteins as we previously have shown (79) and the role of ATF6β is dispensable as long as ATF6α is present. Furthermore, the genes that were examined are wellknown targets of ATF6α, thus, it is possible that compensation by ATF6α prevented us from observing a difference in gene expressions in cells depleted of ATF6β. This hypothesis would 51

63 be in line with the work of other groups which suggested a redundancy in the functions of these two isoforms using single isoform knock-out mice (76, 77). I hypothesized that ATF6β depletion might lead to a compensatory increase in the expression of the ATF6α isoform, especially under stress conditions. The results showed a slight increase in ATF6α mrna levels in basal conditions; however, statistical significance was not reached when compared to control cells (Figure 3.4G). It is possible that longer periods of ATF6β depletion (e.g. 72 h) will be necessary for the ATF6α isoform to be upregulated to maintain a level of function in the cell that is equal to that of both isoforms. It would be important to examine protein levels of fulllength and cleaved ATF6α in cells depleted of ATF6β. In collaboration with our lab, the Poitout group has shown that both ATF6α and ATF6β bind to the Ins2 promoter (82); however, only ATF6α is able to regulate its expression, whereby Ins2 is downregulated under ER stress conditions by ATF6α (83). Examination of the Ins2 gene in cells depleted of ATF6β demonstrated a trend towards a decrease in Ins2 expression under control and ER stress conditions (Figure 3.4H). Since both isoforms can bind the Ins2 promoter, it is possible that under ER stress conditions, competition for the site might exist between ATF6α and ATF6β, whereby under acute ER stress conditions, ATF6α is able to bind the site more readily since it is activated before ATF6β (3-4 h vs. 8 h, respectively) (81), and under chronic ER stress, ATF6β is produced to dilute the effects of ATF6α on the Ins2 promoter. Thus, the greater degree of downregulation in Ins2 expression observed in ATF6βdepleted β-cells could be due to the fact that lack of ATF6β provides no opposition to ATF6α s binding affinity to the promoter and as such, ATF6α is able to further downregulate Ins2 expression in ATF6β knock-down cells compared to control cells. This result indicates that expression of ATF6β may be necessary for strict quality control of secretory proteins, including insulin, in β-cells; however, further experiments will need to be performed to confirm this effect. Susceptibility toward ER stress differs from cell to cell and studies (17, 132, 160) have revealed that β-cells represent one of the most susceptible cell types. Therefore, I examined susceptibility of β-cells to ER stress in cells depleted of ATF6β. I found that insulinoma cells deficient in ATF6β are more susceptible to ER stress under control and stress conditions (induced by thapsigargin) as demonstrated by increased levels of phospho-eif2α (Figure 3.5). 52

64 The fact that eif2α phosphorylated levels are increased by ATF6β knock-down suggests that ATF6β is required as part of the mechanism that is activated to alleviate the ER stress in pancreatic β cells. I also examined the activation of the IRE1α pathway by assessing Xbp1 splicing. XBP1 is a key transcription factor in the induction of ER chaperones and components of the ERAD pathway and splicing of the Xbp1 mrna is a widely used marker for detection of ER stress (161). In contrast to the results obtained with eif2α, IRE1α pathway activation in cells depleted of ATF6β was similar to control cells as monitored by levels of spliced XBP1 mrna (Figure 3.6). Our cells were treated with thapsigargin, which induces ER stress by depleting the ER of Ca 2+ (162). Thus, the reason for the differential effect on phospho-eif2α and XBP1 levels may relate to the fact that the PERK pathway is more sensitive to ER stress caused by calcium depletion than the IRE1α pathway (163). Consequently, the reduction in ATF6β expression is enough to induce PERK activation, but less sufficient in activating the IRE1α pathway. Although the expression of UPR chaperones and ERAD components were unchanged in the absence of ATF6β, these preliminary results suggest that pancreatic β-cells lacking endogenous ATF6β are mildly more susceptible to ER stress in comparison to control cells. Although induction of ER chaperones and ERAD components in ATF6β-depleted insulinoma cells was not impaired, I found that ATF6β knock-down increased the susceptibility of β-cells to apoptosis, as levels of the pro-apoptotic markers cleaved-caspase 3 and cleaved- PARP were increased in cells depleted of ATF6β under both basal and ER stress conditions (Figure 3.7). This finding suggests that β-cells require ATF6β expression for their viability. However, this increase in apoptosis in cells depleted of ATF6β is not dependent on upregulation of pro-apoptotic genes as expression of Chop and Txnip were unchanged in cells depleted of ATF6βp60. This also suggests that other factors other than ATF6β regulate the expression of these genes under ER stress conditions to induce apoptosis. Studies have shown that ATF6β depletion has variable effects on the viability of different cell types. Viability of HeLa cells (80) depleted of ATF6β was not compromised under ER stress induced by tunicamycin, while ATF6β -/- MEFs were slightly more sensitive to tunicamycin treatment than ATF6β +/+ MEFs (76). In combination with our results, this suggests that as cells become more specialized, there is a greater requirement for ATF6β to maintain cell viability. Since I obtained some evidence that cells depleted of ATF6β are more susceptible to ER stress than control cells as the PERK 53

65 pathway was activated, I hypothesized that perhaps caspase 3 activation was resulting from activation of the JNK pathway. JNK is one of the mediators that connect excessive ER stress with apoptosis induction (31, 97). Thus, I examined levels of phospho-jnk in cells depleted of ATF6β. However, our results show that knock-down of ATF6β does not induce apoptosis of β- cells via the JNK pathway as levels of phospho-jnk were similar between control cells and ATF6β depleted cells. Since JNK activation is induced by the IRE1 pathway under chronic ER stress, this finding is in line with our result showing that activation of the IRE1 pathway is similar in control and ATF6β sirna-treated cells. Therefore, although ATF6β depletion can induce some ER stress, ER stress is unlikely to be the primary contributor to apoptosis induction. In this chapter, the role of ATF6β as a pro-survival factor in pancreatic β-cells was also confirmed. Due to the resultant apoptosis observed in cells depleted of ATF6β, I hypothesized that ATF6β overexpression would lead to a decrease in apoptosis. Our data supports this hypothesis, whereby adenoviral overexpression of ATF6βp60 in cells treated with tunicamycin or thapsigargin led to decreased levels of cleaved-caspase 3 and its downstream target, cleaved- PARP (Figure 3.11). We previously demonstrated that adenoviral overexpression of the active form of ATF6α (ATF6αp50) was detrimental to β-cell viability (79), likely due to the fact that hyperactivation of ATF6α leads to induction of the pro-apoptotic gene, Chop, and thus needs to be stringently controlled via HRD1 (141). It would be interesting to examine whether overexpression of ATF6βp60 can rescue the apoptotic phenotype in insulinoma cells depleted of ATF6α or overexpressing ATF6α. Overall, these results show that ATF6β is activated by ER stress and is required to maintain viability of pancreatic β-cells. The mechanism by which increased apoptosis occurs in response to ATF6β depletion, however, requires further study (See Summary and Future directions). 3.4 Summary and Future directions In summary, the results in this chapter demonstrate that ATF6β is expressed in pancreatic β-cell lines and islets and is activated by ER stress conditions induced by the pharmacological agents tunicamycin and thapsigargin. Although cleavage of ATF6β suggests translocation into the nucleus and thus involvement in the UPR, I was unable to identify any 54

66 ATF6β-specific target genes using the loss-of-function approach. Although ATF6β depletion did not affect the induction of the well-known UPR genes examined, it is possible that ATF6β activates UPR genes in other cell types. Thus, it would be important to identify whether ATF6β is required for induction of UPR genes in response to ER stress in other cell types, similar to other ATF6 orthologues which display cell specific functions (e.g. OASIS in astrocytes, CREBH in hepatocytes etc.). Furthermore, although ATF6β cleavage by tunicamycin and thapsigargin was demonstrated, using pathophysiological treatments associated with diabetes such as high glucose or saturated free fatty acid treatments would be more physiologically relevant. Although I have initiated these experiments, the lack of ATF6β-specific antibody has limited the progress of such experiments. Therefore, the transcriptional regulation of ER stress signaling might be different depending on the cell types or stimulus condition. Further studies will be required to understand the detailed molecular mechanisms for these differential effects. The lack of change in the expression of UPR genes in insulinoma cells depleted of ATF6β could possibly be explained by a compensatory mechanism by XBP1. Since ATF6β also shares similar DNA-binding specificities with XBP1 (53) and XBP1 is induced by ATF6α under ER stress conditions (52), it is possible that XBP1 may compensate for lack of ATF6β. It would be important to examine protein expression of this transcription factor under conditions of ATF6β depletion. This could also be tested by comparing induction of UPR genes in cells doubly depleted of ATF6β and XBP1 with cells depleted only of ATF6β to determine the degree of contribution of each protein. Furthermore, ATF6β may be post-translationally modulated by stress kinases, a modification that may enhance or depress its transcriptional activity. Studies have shown that ATF6α can be post-translationally modified by the stress kinase p38, which phosphorylates ATF6α and activates its transcriptional activity in myocytes (164). Thus, it would be important to investigate whether similar modifications are required for full activation of ATF6β. In the case of ATF6α depletion, we previously showed a correlation between β-cell death and lower chaperone expression (e.g. GRP78) and increased stress kinase activity (e.g. JNK, p38) (79). However, I was unable to detect altered levels of chaperones, ERAD components, pro-apoptotic genes (Chop or Txnip) or MAP kinases in the absence of ATF6β. Therefore, ATF6β depletion-induced apoptosis must be dependent on other downstream factor(s) or mechanisms that are upregulated in the absence of ATF6β. Thus, it would important 55

67 to examine other mediators of apoptosis in response to ER stress, such as expression of the antiapoptotic Bcl-2 protein, which is downregulated by CHOP, a member of the C/EBP family of transcription factors. ATF/CREB proteins, such as ATF6, form selective heterodimers with each other and with other bzip proteins such as AP-1 (Fos and Jun proteins) and C/EBP families of proteins (165). Thus, it is possible that under ER stress conditions, ATF6β heterodimerizes with CHOP to depress its negative effect on Bcl-2 expression, thus promoting survival of pancreatic β-cells. Moreover, under stress conditions CHOP upregulates GADD34, a subunit of the protein phosphatase 1 (PP1), which dephosphorylates eif2α, leading to recovery of mrna translation (95). This premature translational recovery may be detrimental to the cell through the generation of reactive oxygen species (ROS) and increasing ER protein folding load. Thus, a possible interaction between ATF6β and CHOP may also negate CHOP s effect on GADD34, and therefore maintain translational attenuation and reduce the protein load on the ER to promote survival of the cell. This can be examined by measuring levels of GADD34 or ROS. This possibility is less likely however, since levels of phospho-eif2α were increased in cells depleted of ATF6β. Furthermore, it is also possible that ATF6β s anti-apoptotic effect may be mediated downstream of JNK activation. Upon activation by IRE1, JNK activation leads to phosphorylation and activation of the transcription factors c-jun or activation protein-1 (AP-1) (101) which contribute to apoptosis under cellular stress conditions (166). Thus, it is possible that under ER stress conditions ATF6β may heterodimerize with AP-1 proteins (165) and prevent their pro-apoptotic effect. Experiments examining potential ATF6β-interacting partners will be crucial to elucidate the role of ATF6β in maintaining cell survival of β-cells. Thus, a yeast two-hybrid screen using ATF6β as bait may be utilized to probe for potential ATF6β interacting proteins. Furthermore, as previously mentioned (Section ), activation of caspases is also mediated via recruitment of TRAF2 by IRE1 under ER stress conditions, which releases procaspase-12, allowing its activation. Thus, it is possible that depletion of ATF6β promotes the dissociation of TRAF2 and procaspase-12, thus leading to increased cleaved caspase 3, as demonstrated in our results. Co-immunoprecipitation experiments can be employed to examine whether the association between IRE1 and TRAF2 is increased in the absence of ATF6β. 56

68 One question that arose was whether the induction of ATF6β serves as a protective response for the cells to cope with stress, or a part of the cellular response that leads to detrimental outcomes? Our gain-of-function approach using adenoviral overexpression of ATF6βp60 clearly demonstrated that pancreatic β-cells were protected from ER stress-induced apoptosis, as evidenced by the reduced levels of cleaved-caspase 3 and cleaved-parp. It would be interesting to examine whether ATF6β overexpression in the pancreatic β-cells of mice would protect these mice against pathophysiologically induced apoptosis. Furthermore, in vivo models with β-cell specific ATF6β knock-out mice or whole body ATF6β knock-out mice challenged with a high fat diet or hyperglycemia would further our understanding of ATF6β s pro-survival role in a pathophysiological context. 57

69 CHAPTER 4: MICROARRAY ANALYSIS FOR IDENTIFICATION OF ATF6β TARGET GENES 58

70 4.1 Introduction As shown in the previous chapter, ATF6β is expressed in insulinoma cell lines and in islets and is induced by ER stress, indicating a role in the UPR. Furthermore, ATF6β seems to have a cell protective role as depletion of ATF6β sensitizes cells to ER stress-induced apoptosis and overexpression of ATF6βp60 enhances survival under ER stress conditions. To identify how ATF6β produces these effects, I sought to identify what genes ATF6β regulates in insulinoma cells. Two methods were attempted to do this; I) production of an inducible INS-1 cell line where ATF6βp60 expression is under doxycycline control, and II) production of an adenovirus expressing ATF6βp60 followed by microarray analysis. 4.2 Results Generation of a stable cell line with inducible expression of ATF6βp60 In the first approach, I attempted to generate a double-stable cell line with doxycyclineinducible expression of the active form of ATF6β (ATF6βp60) using the ptet-on gene expression system (see Methods). Following cloning of FLAG-ATF6βp60 into the ptre-tight vector, the plasmid was tested for FLAG-ATF6βp60 expression by transient transfection of the ptet-on INS-1 #46 cells (previously generated in our lab). As shown in Figure 4.1, ATF6βp60 expression was inducible by doxycycline in cells transfected with the plasmid. Thus, I proceeded with generating the double-stable cell line. The ptre-tight-flag- ATF6βp60 and the ptk-hygro plasmids (with 20:1 ratio) were co-transfected into the ptet-on INS-1 #46 cell line by electroporation, followed by selection for positive clones for G418 and hygromycin resistance. After multiple attempts, only a small number of clones were generated (Table 4.1), and these were tested for doxycycline inducible expression of ATF6βp60 by Western blot analysis and immunofluorescence microscopy. None of the clones tested responded positively to doxycycline treatment as assessed by western blotting for anti-flag (no band detected) and immunofluorescence (no nuclear localization detected). Thus, I was unsuccessful in generating a pancreatic double stable β-cell line with doxycycline-inducible expression of active ATF6βp60. 59

71 Figure 4. 1 Transient transfection of ptet-on INS-1 #46 cells with ptre-flag-atf6βp60 plasmid. Doxycycline-inducible expression of FLAG-ATF6βp60 from the ptre-tight plasmid was tested by transient transfection of ptet-on INS-1 #46 cells for 24 h. Lipofectamine (Lipo) treated cells served as control. Cells were either left untreated or treated with 2 µg/ml Doxycycline (Dox) for 24 h. ATF6βp60 protein expression was monitored by Western blotting analysis using anti-flag antibody and γ-tubulin as loading control. Table 4. 1 Attempts at generating a stable cell line with Doxycycline inducible expression of ATF6βp60. All clones screened were negative for expression of FLAG-ATF6βp60 in response to doxycycline by Western blot or immunofluorescence. Number of clones screened ptet-on INS-1-FLAG- ATF6βp60 cell line ~ Ad-FLAG-ATF6βp60 is functional and localizes to the nucleus In the second method utilized to identify specific targets of ATF6β, I generated an adenovirus expressing the active form of the ATF6β protein (e.g. Ad-FLAG-ATF6βp60). The purpose of generating the virus to express the active form of ATF6β is so as to preclude the use of an ER stressor to induce expression of ATF6βp60, which would also cause activation of the other arms of the UPR. An adenovirus expressing the green fluorescent protein (Ad-GFP), which was previously generated in the lab, was used as a control. Furthermore, due to 60

72 limitations of the ATF6β specific antibody, the initial construct was generated to contain a 3X- FLAG-tag on the N-terminal of the ATF6βp60 sequence for convenient identification of ATF6β expression from the adenovirus. First, I confirmed expression of ATF6βp60 from the adenovirus using Western blot analysis. INS-1 832/13 cells were infected with increasing concentrations of Ad-GFP or Ad-FLAG-ATF6βp60 for 24 h. Cell lysates were used to blot for α-flag (indicative of FLAG-ATF6βp60) protein expression. As shown in Figure 4.2A, increasing the concentration of the adenovirus led to a linear dose-dependent upregulation of FLAG-ATF6βp60, demonstrating a significant dynamic range for expression of ATF6βp60. As the adenovirus expresses the active form of ATF6β (ATF6βp60), the expressed protein is expected to localize to the nucleus. In order to confirm this, I used immunofluorescence staining for α-flag to monitor the localization of ATF6βp60. As expected, the protein was localized in the nucleus, which also indicates that the FLAG tag did not impair ATF6βp60 nuclear localization (Figure 4.2B). Next, the effect of ATF6βp60 overexpression and that of the active form of ATF6α (Ad- ATF6αp50), which was previously generated in the lab (79), was tested on induction of Grp78 mrna. As expected, there was a robust increase in Grp78 mrna levels when INS-1 832/13 cells were infected with Ad-ATF6αp50, whereas Ad-ATF6βp60 had no effect in comparison to control (Ad-GFP) (Figure 4.3). Thus, these initial experiments confirmed that our Ad-FLAG- ATF6βp60 virus expressed ATF6βp60 and the protein correctly localized to the nucleus. Identification of global transcriptional changes in response to ATF6βp60 expression was then performed with microarray gene profiling. 61

73 Figure 4. 2 ATF6βp60 overexpression and localization in pancreatic β-cells. INS-1 832/13 cells were infected with varying titers of a recombinant adenovirus expressing the active form of ATF6β (Ad-FLAG-ATF6βp60) for 24 h. Lysates (10 µg) were resolved by SDS-PAGE and immunoblotted for anti-flag to detect ATF6βp60 expression and γ-tubulin as a loading control (A). pfu, plaque forming units. INS-1 832/13 cells were infected with 1x10 9 opu/ml Ad-ATF6βp60 for 6 h. The cells were then fixed and immunostained with an anti-flag antibody and DAPI to stain cell nuclei (B). Nuclear fluorescence for ATF6βp60 was apparent for cells stained for FLAG. Figure 4. 3 Effect of ATF6αp50 and ATF6βp60 on Grp78 mrna. INS-1 832/13 cells were infected with 1x10 8 opu/ml Ad-GFP or Ad-ATF6βp60 or 1x10 9 opu/ml Ad- ATF6αp50 for 24 h. Total RNA was isolated and real-time PCR was used to determine relative amounts of Grp78 mrna in each sample (One-Way ANOVA, post hoc Tukey s test, ***p<0.0001, n=3). 62

74 4.2.3 Identification of potential ATF6β-specific target genes using microarray analysis To examine the global transcriptional response to ATF6βp60 expression, microarray analysis was performed for the gene expression changes following Ad-FLAG-ATF6βp60 infection of INS-1 832/13 cells. Furthermore, I wanted to compare transcriptional regulation of UPR genes between ATF6α and ATF6β. Control cells were infected with the Ad-GFP alone. Thus, the adenoviruses (Ad-GFP, -ATF6αp50, -ATF6βp60) were used to infect INS-1 832/13 cells and RNA was isolated for microarray expression profiling which was performed by the University Health Network (UHN) Genomic Centre (Toronto). For this experiment, I used a concentration of 1x10 8 opu/ml for Ad-GFP and Ad-FLAG-ATF6βp60, and 1x10 9 opu/ml for Ad-FLAG-ATF6αp50, as these concentrations were observed to express equal protein expression of each isoform by Western blotting for anti-flag. Cells were infected for 24 h, then total RNA was isolated and sent to the UHN Genomic Centre for analysis (see Methods - INS-1 832/13 Cell Infection). The microarray experiment identified 49 enriched genes (genes categorized into biological processes) that were upregulated by ATF6βp60 2-fold. Upon grouping of the enriched genes into biological processes, most of the categorized genes were ER and UPR related (Table 4.2), such as chaperones (Pdia4, calr, etc.) and ERAD components (e.g. Herp, Edem, Hrd1). A complete list of genes upregulated 2-fold is included in the Supplemental Data section. When comparing these ER-related genes upregulated by ATF6βp60 with genes that were upregulated by ATF6αp50, 22 of 29 ER-related enriched genes upregulated by ATF6αp50 were found to be also upregulated by ATF6βp60 (Table 4.3). These results are in line with literature that suggests overlapping function between the two isoforms in mice (76, 77). 63

75 Table 4. 2 Biological processes identified for enriched genes upregulated 2-fold by ATF6βp60 by microarray analysis. There are 49 distinct genes upregulated by ATF6βp60 that are ER- and UPR-related. Biological Process # of genes # of distinct genes endoplasmic reticulum 50 endoplasmic reticulum lumen 9 response to unfolded protein/ response to topologically incorrect protein 9 Total Biological Process Endoplasmic reticulum RefSeq Transcript ID Gene Symbol Fold Change NM_ Calr NM_ Cat NM_ Cd NM_ /// XM_ /// XM_ Cisd NM_ Ckap NM_ Creld NM_ Cyp26b NM_ Cyp2f NM_ Cyp2u NM_ Dd NM_ Dnajb NM_ Dnajb NM_ Dnajc XM_ /// XM_ Edem NM_ Ehd NM_ Erp NM_ Fa2h NM_ Herpud NM_ Hmox NM_ Hsd11b NM_ Hsp90b NM_ Hspa NM_ /// NM_ Hyou NM_ Kdelr NM_ Kdelr NM_ Lmf NM_ Manf NM_ MGC

76 NM_ Ormdl NM_ Os NM_ P4ha NM_ Pdia NM_ /// XM_ /// XM_ Phtf NM_ Piga NM_ Plod NM_ Ppm1l NM_ RGD NM_ Sel1l NM_ Sels NM_ Slc7a NM_ Syvn NM_ Tap NM_ Tmem50b NM_ Tmem NM_ Wfs NM_ Yipf Endoplasmic reticulum lumen RefSeq Transcript ID Gene Symbol Fold Change NM_ Calr NM_ Dnajb NM_ Dnajc NM_ Erp NM_ Hsp90b NM_ Hspa NM_ /// NM_ Hyou NM_ Os NM_ Pdia Response to unfolded protein/response to topologically incorrect protein RefSeq Transcript ID Gene Symbol Fold Change NM_ /// NM_ Ddit NM_ Dnajb NM_ Dnajc NM_ Erp NM_ Herpud NM_ Hspa1a NM_ Hspa NM_ Manf NM_ Sels

77 Table 4. 3 Comparison of enriched genes identified to be upregulated 2-fold by Ad-ATF6αp50 and Ad-ATF6βp60 in comparison to Ad-GFP. Genes common to both Ad-ATF6αp50 and Ad-ATF6βp60 are highlighted in red. ATF6α genes UP 2- fold Calr Creld2 Dd25 Ddit3 Dnajb11 Dnajc3 Edem1 Ehd4 Fa2h Herpud1 Hsd11b2 Hyou1 Kdelr3 Manf MGC Ormdl2 Os9 Pdia4 Piga Sel1l Slc7a11 Tmem50b Atp2a2 Eif2ak3 Emd Hmgcs1 Hrc Icmt Nr4a2 ATF6β genes UP 2- fold Calr Creld2 Dd25 Ddit3 Dnajb11 Dnajc3 Edem1 Ehd4 Fa2h Herpud1 Hsd11b2 Hyou1 Kdelr3 Manf MGC Ormdl2 Os9 Pdia4 Piga Sel1l Slc7a11 Tmem50b Cat Cd74 Cisd2 Ckap4 Cyp26b1 Cyp2f4 Cyp2u1 Dnajb5 Dnajb9 Erp44 Hmox2 Hsp90b1 Hspa1a Hspa5 66

78 Kdelr2 Lmf2 P4ha2 Phtf1 Plod1 Ppm1l RGD Sdf2l1 Sels Syvn1 Tap1 Tmem97 Wfs1 Yipf4 As microarray expression profiling can lead to false positives and as this experiment consisted of only one set of samples, it was essential to validate the microarray results by realtime PCR. Thus, INS-1 832/13 cells were infected with 1x10 8 opu/ml of either Ad-GFP as control or Ad-FLAG-ATF6βp60 for 24 h. Total RNA was isolated and gene specific primers were used to determine whether ATF6βp60 overexpression leads to increased levels of the genes uncovered by the microarray data. For our initial experiments, we chose to validate two genes that were upregulated by both ATF6αp50 and ATF6βp60, and two genes that were upregulated by ATF6βp60 only. For genes common to both ATF6 isoforms, I picked the Herp (Herpud1) and the Chop (Ddit3) genes, while for the ATF6β-specific genes, I chose Sdf2l1 and Dnajb9. Herp induction by ATF6α has been well documented in a number of cell lines (76, 77), including by our lab in pancreatic β-cells (79). The effect of ATF6α on Chop induction is controversial, as only a small number of studies have demonstrated this (141) and we previously shown that Chop mrna was unaltered in ATF6α-depleted β-cells (79), although only a 60% knock-down efficiency was achieved in this study. To our knowledge, the effect of ATF6βp60 on these genes has not been demonstrated in β-cells. In line with our microarray results, our real-time PCR data demonstrate a trend towards increased expression of both Herp and Chop in β-cells overexpressing ATF6βp60 in comparison to control (Figure 4.3A and B, respectively). The microarray analysis revealed a 15-fold increase in Sdf2l1 mrna in cells overexpressing ATF6βp60. When I analyzed the induction of this gene by real-time PCR, there was a ~11-fold increase in its expression in comparison to control (Figure 4.3C), although statistical 67

79 significance was not reached when analyzed by the student t-test, likely due to differences in infection efficiency of individual experiments. Validation of the co-chaperone Dnajb9 also showed a positive trend towards upregulation in cells overexpressing ATF6βp60 (Figure 4.3D). These results suggest that ATF6β, similar to ATF6α, contributes to the UPR response by inducing the expression of genes that are also regulated by ATF6α, as well as additional genes that are unique to ATF6β regulation. Validation of additional genes will be performed to further elucidate the role of ATF6β in the ER stress response of pancreatic β-cells. 68

80 Figure 4. 4 Validation of microarray genes upregulated by ATF6βp60 by real-time PCR. INS-1 832/13 cells were infected with either 1x10 8 opu/ml Ad-GFP as control or Ad-FLAG-ATF6βp60 for 24 h. Total RNA was isolated and real-time PCR was used to determine relative mrna levels of the Herp (A, n=4), Chop (B, n=4), Sdf2l1 (C, n=4), Dnajb9 (D, n=4), Hmox1 (E, n=4) and Atf3 (F, n=4) genes. Statistical significance between two experimental conditions was analyzed using a two-sample t- test assuming equal variance. 69

81 4.3 Discussion In accordance with the complexity of their biological systems, higher mammals have evolved two ATF6 genes, ATF6α and ATF6β. However, although this study as well as the work of others has demonstrated ATF6β activation in response to ER stress, a role for this protein has yet to be identified. I attempted to address this problem by knocking down endogenous ATF6β and examining the ability of pancreatic β-cells to induce the expression of a number of UPR genes. I found that ATF6β is not involved in inducing the canonical UPR genes, suggesting possibly a distinct function from the ATF6α isoform. Furthermore, knockdown experiments identified a cell protective role for ATF6β under both control and ER stress conditions. Thus, identification of target genes regulated by ATF6β would help explain its function. In this chapter, I employed a second approach to identify ATF6β-target genes by attempting to generate a double-stable cell line with inducible expression of ATF6βp60 using the Tet-ON system (See Methods). The advantage of using this system would be to allow for doxycycline-dependent control of ATF6βp60 levels in the physiologic range. However, I was unsuccessful in obtaining a positive clone with doxycycline-inducible expression. The primary reason that this may have been the case is that clonal selection proved difficult as very few colonies were generated in response to selection media that could be tested for ATF6βp60 expression. To overcome this obstacle, I generated an adenovirus that overexpressed the active form of ATF6β (Ad-ATF6βp60). To identify genes that were up-regulated by ATF6βp60, whole-genome microarray analysis was performed for RNA samples prepared from INS-1 832/13 cells that had been infected with Ad-FLAG-ATF6βp60. Pancreatic INS-1 832/13 cells infected with an adenovirus expressing ATF6βp60 (Ad-ATF6βp60) were compared with cells infected with Ad-ATF6αp50 or Ad-GFP as a control. In contrast to the gene array results obtained by Lee et al. (53), our gene array identified a large number of genes that were induced in the presence of ATF6βp60 (353 genes upregulated 2-fold) and ~588 genes that were downregulated by 2-fold. The inconsistencies between these data likely reflects the experimental approaches used and the cell system studied. While Lee et al. (53) used the MEF cell line and transient transfection of a plasmid expressing ATF6βp60, I used a more specialized cell type, β-cells, and a more efficient protein expression method using adenoviral infection. Among the 20, 000+ rat genes, 49 genes unique to ATF6β were induced by more than 2-fold in 70

82 the presence of ATF6βp60 that were enriched into biological processes. Gene sets transcriptionally induced by ATF6βp60 were highly enriched for proteins localized to the ER and involved in proteostasis (Table 4.2). In line with their similar activation mechanism and considerable homology in their structures, these microarray results suggest a role for ATF6β that is similar to that of ATF6α. Furthermore, a significant number of genes induced by ATF6αp50 were found to be also induced by ATF6βp60 (Table 4.3), suggesting overlapping functions of the two isoforms. If validated positively, these results will be in line with the results of Yamamoto and colleagues (76) whereby double isoform knock-out mice were embryonic lethal, while single isoform knock-outs were developmentally normal, suggesting compensation by either isoform. The microarray results would also support our knock-down results, in which ATF6β sirna-treated cells induced UPR genes to the same level as control cells, likely reflecting compensation by ATF6α since the tested genes (Grp78, Grp94, Herp, Hrd1) are known to be also regulated by ATF6α as well as by PERK and IRE1 pathways (Chop, Txnip). As this experiment represented a single set of samples and microarray expression profiling can lead to both false-positive and false-negative errors, I validated some upregulated genes of interest using real-time PCR. Importantly, ATF6βp60 overexpression resulted in a trend towards an increase in genes detected by microarray (Herp, Chop, Sdf2l1, Dnajb9), but not two control genes (Atf3, Hmox1). Further validation is required to confirm that induction of these genes is significant as well as additional genes identified by the microarray. Overall these results suggest that ATF6α and ATF6β share some targets, but also activate unique sets of genes. Thus, selective activation and/or kinetic differences in the activation of specific genes may allow the ER to elicit the most appropriate response to prevailing ER stress conditions via these two isoforms. Our results from Chapter 3 indicated that ATF6β depletion compromised β-cell viability, however the mechanism as to how this occurs was not elucidated. The microarray analysis results from Chapter 4 may shed light on how this may occur. Upon examination of the genes regulated by ATF6βp60, I identified certain pro-apoptotic genes that were downregulated, such as Casp3 (Caspase 3, <3.2-fold), Bik (Bcl-2-Interacting Killer, <1.6-fold), Perp (p53 apoptosis effector related to PMP-22, <1.7-fold), and Wbp1 (WW Domain Binding Protein 1, <1.7-fold) among others. Thus, it is possible that ATF6β s prosurvival effects in the face of chronic ER 71

83 stress may be mediated by downregulation of pro-apoptotic genes. For example, Morishima et al. found that ATF6αp50 overexpression in the C2C12 myoblast cell line leads to apoptosis, an effect mediated via the ER stress-inducible protein WBP1 (167) that leads to processing of caspases-12, -9, and -3 for activation. Examining the genes upregulated by Ad-ATF6αp50 in INS-1 832/13 cells, I found that Wbp1 was indeed induced by ATF6αp50 (~1.9-fold), while it was reduced by 1.7-fold in cells overexpressing ATF6βp60. Validation of these results may identify novel mechanism through which ATF6β activation prevents apoptosis under ER stress conditions. Interestingly, two of the genes identified to be upregulated by ATF6βp60 were Wfs1 and Hrd1. As previously mentioned (see Ch. 3 Discussion), Wfs1 regulates ATF6α activity to prevent its hyperactivation under chronic ER stress conditions by mediating its degradation via the E3 ligase Hrd1. It is possible that ATF6β modulates ATF6α s transcriptional activity by upregulating both Wfs1 and Hrd1 expression, as suggested by our microarray results (discussed in the Overall Conclusion). 4.4 Summary and Future directions I attempted to identify ATF6β-specific target genes by infecting INS-1 832/13 cells with an adenovirus overexpressing the active form of ATF6β (Ad-ATF6βp60) and performing global transcriptional profiling using microarray analysis. Validation of selected genes revealed that ATF6β responds to ER stress in a similar manner as ATF6α. Additional genes will be selected for validation by real-time PCR that will include those that are induced by both ATF6α and ATF6β as well as genes that are regulated by ATF6β alone. Furthermore, I will examine whether these genes are induced in response to ER stress by tunicamycin (or thapsigargin) in ATF6β knock-down cells to further confirm the dependence of ATF6β-target genes on ATF6β. I predict that genes common to both ATF6α and ATF6β will be induced by tunicamycin in ATF6β-depleted β-cells to a similar level as control cells, while those primarily induced by ATF6β will be blunted in ATF6β-depleted β-cells treated with tunicamycin. I am currently validating the expression of Wfs1 and Hrd1 in ATF6βp60-overexpressing cells. We can further confirm ATF6β s effect of ER stress response genes by examining whether the activity of UPRE and ERSE reporters in pancreatic β-cells is completely absent or significantly diminished in ATF6β knock-down cells. Conversely, UPRE- and ERSE-reporter 72

84 activity can also be tested in ATF6β-overexpressing cells. Furthermore, the possibility of interacting proteins being required for ATF6β activity can be tested by screening for ATF6β partners using the yeast two-hybrid assay. Future microarray experiments can also be performed with pancreatic β-cells overexpressing ATF6βp60 as well as possible interacting proteins (e.g. ATF6αp50 or sxbp1). ATF6β might act as a transcriptional activator or repressor depending on the setting. For example, ATF3 homodimer is a transcriptional repressor, however heterodimeric complex of ATF3 with c-jun has been demonstrated to function as a transcriptional activator (165). Thus, it is possible that as a homodimer ATF6β acts as a transcriptional repressor, while heterodimerization with other transcription factors can allow ATF6β to act as an activator. It is also possible that ATF6β interacts with co-activators and that overexpression of ATF6β sequesters these factors away from the promoter, resulting in transcriptional repression. Thus, it is essential to investigate whether any post-translational modification or interaction with other proteins alters the transcriptional activity of ATF6β. 4.5 Overall Conclusion The results presented in this thesis contribute to our knowledge of the role of the ATF6β protein in the UPR, specifically in pancreatic β-cells. In chapter 3, I demonstrate that ATF6β is expressed in pancreatic β-cells and is activated by ER stress as evidenced by the expression of the active form of ATF6βp60. Although knock-down of ATF6β does not affect gene expression of canonical UPR genes, ATF6β may play a supporting role secondary to ATF6α, and its effect may be temporal in response to ER stress (80). In fact, our microarray and validation experiments revealed that ATF6β might share some of the responsibility of inducing adaptive UPR genes with ATF6α, as a large number of ER stress responsive genes that were upregulated were shared by the two isoforms. This finding supports the results of other studies (76, 77) demonstrating that the combined ATF6α and ATF6β double knock-out mouse model is embryonic lethal, while the single ATF6α -/- or ATF6β -/- knock-out mice are viable and healthy, suggesting compensation by each isoform for the lack of the other. Furthermore, I have identified ATF6β as a pro-survival factor employed by pancreatic β-cells as a defense against ER stress-induced apoptosis, whereby depletion of ATF6β led to apoptosis, whereas overexpression of the active form reduced the apoptotic effect under ER stress conditions. The mechanism through which ATF6β depletion leads to susceptibility of β-cells to apoptosis 73

85 remains elusive, as I was unable to identity changes in pro-apoptotic pathways that may contribute to this effect. However, our microarray results show that ATF6βp60 upregulates the Wfs1 gene by more than 2-fold and the Syvn1 (Hrd1) gene by ~2.5-fold. WFS1 is a transmembrane protein localized to the ER that has been shown to be a UPR component that mitigates ER stress response in cells (168). Importantly, mutations in the Wfs1 gene cause Wolfram syndrome, a rare neurodegenerative disorder characterized by early onset diabetes, optic atrophy and deafness (138). Furthermore, the diabetic phenotype in Wolfram syndrome patients has been attributed to high levels of ER stress signaling in β-cells. Activation of ER stress signalling must be tightly regulated because hyperactivation or chronic activation of the UPR can cause cell death. We and others have shown that overexpression of ATF6αp50 leads to apoptosis (79, 167); thus ATF6α mediates the transition from self-defense to self-destruction of cells during ER stress, whereby at relatively low levels, ATF6αp50 activates the UPR for self-defense via upregulation of adaptive genes such as chaperones, while at higher levels, ATF6αp50 mediates apoptosis. A recent study has demonstrated the importance of stringent regulation of ATF6α expression in the induction of apoptosis under pathological conditions. Fonseca and colleagues (141) have demonstrated that WFS1 negatively controls ER stress signaling induced by ATF6α by promoting its association with HRD1 and therefore its degradation to prevent its hyperactivation. Considering these findings and the possible induction of Wfs1 and Hrd1 by ATF6β as suggested by our microarray results, a hypothetical model for the role of ATF6β in the ER stress response and viability of β-cells is proposed. I speculate that under acute ER stress conditions, ATF6α is activated to induce expression of its target adaptive genes (chaperones, ERAD components). However, as hyperactivation of ATF6α has been shown to cause apoptosis (79, 80, 167), WFS1 is induced under chronic ER stress conditions to mediate ATF6α s degradation via HRD1, to prevent apoptosis (Figure 4.4A). It is under these chronic ER stress that ATF6β is activated to maintain the expression of adaptive genes and to induce expression of both Wfs1 and Hrd1 in order to promote destruction of ATF6α to prevent its hyperactivation and thereby prevent apoptosis (Figure 4.4B). Thus, ATF6β-depleted cells may be undergoing increased apoptosis in response to ER stress because of dysregulation of ATF6α UPR signaling, whereby a loss of ATF6β would lead to enhanced ATF6α signaling via decreased levels of Hrd1 and Wfs1 (Figure 4.4C). Reduced HRD1 activity would not only increase the availability of 74

86 ATF6α for cleavage, but also contribute to ER stress by promoting the buildup of unfolded/misfolded proteins in the ER. Our finding that ATF6β depletion leads to mild ER stress as indicated by increased levels of phospho-eif2α supports this hypothesis. This model is supported by the findings of Thuerauf et al. (80) which demonstrated that ATF6α has a half-life of ~2 h, while ATF6β has a half-life of ~5 h, and that ATF6α is cleaved 3-4 h after ER stress (by tunicamycin) while ATF6β is activated at ~8 h after ER stress, suggesting temporal regulation of target genes by these two isoforms. Furthermore, Yamamoto et al. (169) showed that HRD1 is regulated by the IRE1-XBP1 pathway and is activated at a later time point during ER stress. This would also suggest that ATF6β serves to turn off signaling by ATF6α via Wfs1 and allow the UPR to switch to IRE1 activation. In chapter 3 I found that ATF6β depletion did not lead to changes in ATF6α levels at the transcriptional level, which may support our model whereby ATF6β may suppress ATF6α by posttranslational regulation; e.g. ubiquitin-mediated degradation. Thus, based on this model, expression of ATF6β in pancreatic β-cells may serve to induce expression of adaptive genes under chronic ER stress conditions and to protect β-cells from apoptosis by preventing hyperactivation of UPR signaling mediated by ATF6α. In order to test this hypothesis, firstly gene induction of Wfs1 and Hrd1 in the presence of ATF6βp60 needs to be validated by real-time PCR. Secondly, if ATF6βp60 does indeed promote the degradation of ATF6α under chronic ER stress conditions, then levels of ATF6α should be reduced in cells overexpressing ATF6βp60. Furthermore, ubiquination of ATF6α can be examined under conditions of ATF6βp60 overexpression by treating cells with the proteasome inhibitor MG132 and immunoprecipitating ATF6α followed by immunoblot with an anti-ubiquitin antibody. Although our in vitro findings may have significant implications if our model is proven to be true, the relevance of ATF6β-dependent cell viability in β-cells in vivo has yet to be investigated. Although the ATF6β -/- mouse is viable and appears to not develop spontaneous diabetes as a result of β-cell dysfunction or death, no studies to date have analyzed β-cell function in ATF6β -/- mice challenged with hyperglycemia or obesity, mediators of ER stress and β-cell dysfunction. Thus, it would be interesting to analyze the role of ATF6β in β-cells in an in vivo model with pancreas-specific ATF6β knock-out. Overall, the findings from this project have significant implications because elucidating the mechanism of ER stress-induced apoptosis in ATF6β depleted cells may be important for 75

87 understanding physiological apoptosis and cell death under pathological conditions, including diabetes mellitus. 76

88 Figure 4. 5 A hypothetical model for the role of ATF6β in the ER stress response of pancreatic β- cells. (A) Under acute ER stress, ATF6α is activated to mediate induction of UPR target genes including chaperones and ERAD components. Hyperactivation of ATF6α, which can lead to cell apoptosis, is prevented by the WFS1 protein, which recruits ATF6α to the E3 ligase HRD1. HRD1 marks ATF6α with ubiquitin for proteasomal degradation. (B) Under chronic ER stress conditions, ATF6β is activated to take over the role of ATF6α in induction of adaptive proteins. ATF6β also induces expression of Wfs1, which leads to degradation of ATF6α and prevention of apoptosis. (C) Depletion of ATF6β from β-cells leads to reduced effects of WFS1 on ATF6α, thereby resulting in increased susceptibility to apoptosis. 77