The eukaryotic 2, a hero turned villain in β cells

By

Baroj Abdulkarim

Université libre de Bruxelles

Faculty of Medicine

ULB Center for Diabetes Research

Academic year 2016-2017

Jury Members: Dr. Ingrid Langer (President) Dr. Miriam Cnop ( and secretary) Dr. Mariana Igoillo Esteve (Co-Promoter) Dr. Daniel Christophe Dr. Christophe Erneux Dr. Claudine Heinrichs Dr. Amar Abderrahmani Dr. Patrick Gilon

Dedicated to my daughter Elîn

2

Contents Papers constituting this thesis ...... 4 Abbreviations ...... 5 Abstract ...... 8 Résumé ...... 9 Introduction ...... 10 Diabetes mellitus ...... 10 How β cells work ...... 11 Type 2 and monogenic diabetes ...... 12 Free fatty acids and diabetes ...... 14 Acute exposure to FFAs: enhanced β cell function ...... 15 Prolonged exposure to saturated FFAs: Lipotoxicity-induced ER stress ...... 17 The unfolded response ...... 17 Dysregulation of PERK/eIF2α pathway in diabetes ...... 22 β cell apoptosis ...... 24 The extrinsic pathway ...... 26 The intrinsic pathway ...... 26 Aims of this thesis ...... 29 Results...... 30 PAPER I ...... 30 PAPER II ...... 48 PAPER III ...... 62 Discussion ...... 84 Diseases caused by dysregulated ER stress signaling ...... 88 Not only β cells ...... 88 Models of PERK/eIF2α pathway ...... 90 Treating ER stress ...... 91 Conclusions and perspectives ...... 92 Acknowledgements ...... 95 References ...... 96 Supplementary data ...... 111

3

Papers constituting this thesis

1. Cnop M, Abdulkarim B, Bottu G, Cunha DA, Masini M, Turatsinze JV, Griebel T, Igoillo-Esteve M, Bugliani M, Villate O, Ladriere L, Marselli L, Marchetti P, McCarthy MI, Sammeth M, Eizirik DL; RNA-sequencing identifies dysregulation of the human pancreatic islet transcriptome by the saturated fatty acid palmitate. Diabetes, 2013, 63, 6, 1978-1993

2. Abdulkarim B, Nicolino M, Igoillo-Esteve M, Daures M, Romero S, Philippi A, Senée V, Lopes M, Cunha DA, Harding HP, Derbois C, Bendelac N, Hattersley AT, Eizirik DL, Ron D, Cnop M, Julier C; A Missense Mutation in PPP1R15B Causes a Syndrome Including Diabetes, Short Stature, and Microcephaly. Diabetes. 2015, 64, 11, 3951-3962.

3. Abdulkarim B,Hernangomez M, Igoillo-Esteve M, Ladriere L, Cunha DA, Marselli L, Marchetti P, Eizirik DL, Cnop M; Guanabenz sensitizes β-cells to stress-induced apoptosis. , 2017, Epub ahead of print.

4

Abbreviations APAF1 Apoptotic protease activating factor 1 ATF Activating factor BH Bcl2 homology Bad Bcl2 associated agonist of cell death Bak Bcl2 antagonist/killer Bax Bcl2 associated X protein Bcl2 B-Cell lymphoma 2 Bid BH3 interacting domain death agonist Bim Bcl2 interacting mediator of cell death BiP Immunoglobulin heavy chain binding protein cAMP Cyclic AMP CHOP CCAAT/enhancer binding protein homologous protein CPE carboxypeptidase E CReP Constitutive repressor of eIF2α DP5 Death protein 5 eIF initiation factor EPAC Exchange factor directly activated by cAMP ER Endoplasmic reticulum ERAD ER associated degradation ERO1 ER 1 FACS Fluorescent activated cell sorting FADD Fas-associated death domain protein FFA Free fatty acid GADD34 Growth arrest DNA damage inducible 34 GATA6 GATA binding protein 6 GCK Glucokinase GCN2 General control nonderepressible 2 GLP-1 Glucagon like peptide 1 Glut Glucose transporter HNF1A Hepatocyte nuclear factor 1 alpha HRI Heme regulated initiation IDF International diabetes federation 5

INS Insulin Ini-Met Initiator Methionine IRE1 Inositol requiring 1 ISR Integrated stress response ISRIB ISR inhibitor JNK c-Jun N-terminal kinase + KATP ATP sensitive K channel KCNJ11 Potassium voltage gated channel subfamily J member 11 LC8 Light chain 8 Mcl1 Myeloid cell leukemia sequence 1 MODY Maturity onset diabetes of the young NeuroD1 Neuronal differentiation 1 NDM Neonatal diabetes mellitus NO Nitric Oxide Noxa phorbol-12-myristate-13-acetate-induced protein 1 Nrf2 nuclear factor erythroid-2-related factor-2 NRSF Neuronal-restrictive silencer factor ORF Open P58IPK 58 kDa inhibitor of PKR PDX1 Pancreatic and duodenal homeobox 1 PERK PKR-like endoplasmic reticulum kinase PKA Protein kinase A PKR Protein kinase R PP1 Protein phosphatase 1 PPAR Peroxisome proliferator-activated receptor PUMA P53 upregulated modulator of apoptosis Rap1 Ras-related protein 1 RNA-seq RNA sequencing RRP Readily releasable pool SERCA Sarcoendoplasmic reticulum Ca2+ ATPase SNAP25 Synaptosomal-associated protein 25 SNARE SNAP receptor TIRF Total internal reflection fluorescence TRAF2 Tumor necrosis factor receptor- associated factor 2

6

Trib3 Tribblespseudokinase 3 uORF Upstream ORF UPR Unfolded protein response VDCC Voltage dependent Ca2+ channel VAMP Vesicle associated membrane protein WFS1 Wolfram syndrome 1 XBP1 X-box binding protein 1

7

Abstract The prevalence of type 2 diabetes is increasing dramatically worldwide. Type 2 diabetes is a major health and socio-economic burden. Genetic predisposition and the obesity epidemic, due to sedentary style and high caloric food intake, are associated with development of type 2 diabetes. Circulating free fatty acids (FFAs), in particular saturated FFAs, are linked with insulin resistance and β cell dysfunction. Following this background we performed RNA sequencing of human pancreatic islets treated with the saturated FFA palmitate to acquire a global image of the islet response to this insult. We identified several stress pathways induced by palmitate with a major induction of the endoplasmic reticulum (ER) stress response.

The ER stress response, in particular the PKR-like ER kinase (PERK) branch, has been shown to be induced by saturated FFA. It leads to increased β cell apoptosis both in fluorescence activated cell sorter (FACS) purified rat β cells and human islets. We further clarified the role of this pathway by studying the involvement of the constitutive repressor of eIF2α phosphorylation (CReP) in a monogenic form of diabetes. CReP is a repressor of eukaryotic translation initiation factor 2α (eIF2α) phosphorylation. A direct target of PERK, eIF2α is involved in translational attenuation and induction of apoptosis. We have shown that CReP loss-of-function leads to a new syndrome of young onset diabetes, intellectual disability and microcephaly. The identified R658C mutation abrogated CReP activity leading to increased eIF2α phosphorylation and β cell apoptosis.

To further demonstrate the importance of eIF2α dysregulation in β cell demise, we used guanabenz, a chemical inhibitor of growth arrest DNA damage inducible 34 (GADD34). GADD34 is an ER stress-induced repressor of eIF2α phosphorylation. Guanabenz potentiated FFA-mediated ER stress and apoptosis in clonal and primary rat β cells and in human islets through the activation of CCAAT/enhancer binding protein homologous protein (CHOP), downstream of eIF2α. Guanabenz administration in mice impaired glucose tolerance and led to β cell dysfunction. In ex vivo experiments guanabenz also induced β cell dysfunction in mouse and rat islets.

In conclusion our data demonstrate that the dysregulation of signaling in the PERK/eIF2α pathway is crucial for β cell demise. Together with previously reported monogenic diabetes caused by loss-of-function mutations in PERK in man and the eIF2αS51A mutation in mice, our findings suggest that a narrow regulation of PERK/eIF2α signaling is central for proper β cell function and survival.

8

Résumé La prévalence du diabète de type 2 augmente de façon spectaculaire au niveau mondial. Le diabète de type 2 est un problème sanitaire et socio-économique majeur. La prédisposition génétique et l’épidémie d’obésité, dues au mode de vie sédentaire et au régime alimentaire hautement calorique, sont associées au développement du diabète de type 2. Les acides gras libres que circulent dans le sang (AGL), en particulier les AGL saturés, sont liés à la résistance à l’insuline et au dysfonctionnement des cellules . Suivant ce constat, nous avons réalisé un séquençage haut débit de l’ARN des ilots pancréatiques humains traités avec l’AGL palmitate pour avoir une image globale de la réponse des ilots humains à cette agression. Nous avons identifié plusieurs voies de signalisation liées au stress, induites par le palmitate avec une induction majeur du stress du réticulum endoplasmique (RE).

Il a été établi que le stress du RE, en particulier la branche PKR-like ER kinase (PERK), est induit par les AGL saturés. Il conduit à l’augmentation de l’apoptose des cellules  à la fois au niveau des cellules de rat purifiés, triés par cytométrie de flux (FACS) et au niveau de ilots pancréatiques humain. De plus, nous avons clarifié le rôle de cette voie en étudiant l’implication du répresseur constitutif de la phosphorylation de l’eIF2. Cible direct de PERK, eIF2 est impliqué dans l’atténuation de la traduction des protéines et l’induction de l’apoptose. Nous avons montré que la perte de fonction de CReP conduit à un nouveau syndrome de diabète juvénile, de retard mental et de microcéphalie. La mutation R658C que nous avons identifiée abroge l’activité de CReP conduisant à l’augmentation de la phosphorylation de eIF2α et à l’apoptose des cellules .

De plus, pour démontrer l’importance de la dérégulation de eIF2α dans la mort des cellules , nous avons utilisé le guanabenz, un produit chimique inhibiteur du gène GADD34. GADD34 est un répresseur de la phosphorylation de eIF2α induit par le stress du RE. Le guanabenz a accentué le stress du RE induit par les AGL et l’apoptose dans les cellules  clonales et primaires de rat et dans les ilots humains via l’activation du gène CHOP, en aval de eIF2α. L’administration du guanabenz aux souris ayant une intolérance au glucose a conduit aux dysfonctionnements des cellules . Dans des expériences ex vivo, le guanabenz a aussi induit le dysfonctionnement des cellules  dans les ilots de souris et de rats.

En conclusion, nos données démontrent que la dérégulation de la voie de signalisation PERK/eIF2α est cruciale pour la mort des cellules . Ensemble avec les précédentes études qui ont montré des formes de diabètes monogéniques causées par les mutations perte de fonction de PERK chez l’homme et de eIF2αS51A chez la souris, nos découvertes suggèrent qu’une régulation précise de la signalisation PERK/eIF2α est centrale au bon fonctionnement et à la survie des cellules β.

9

Introduction

Diabetes mellitus Diabetes mellitus is a group of heterogeneous diseases characterized by hyperglycemia due to defective insulin function, secretion or both. Although, caused by different mechanisms, once hyperglycemia occurs all individuals carry the same risk of accompanying complications. The symptoms of acute hyperglycemia include polyuria, polydipsia, polyphagia, weight loss, blurred vision and frequent or recurrent infections. One of the most severe acute complications of hyperglycemia is diabetic ketoacidosis. Chronic complications of diabetes include slow healing wounds, sexual dysfunction, retinopathy, nephropathy and neuropathy leading to diabetic foot, failure, blindness, cardiovascular and cerebrovascular complications (1).

Classification of diabetes is done into one of the following groups: Type 1 diabetes is an autoimmune disease characterized by a chronic inflammation of the pancreatic islets of Langerhans leading to the destruction of β cells in the islets and causing severe insulin deficiency (2). Type 1 diabetes is the most common form of diabetes in children and adolescents, and also occurs in adults at similar incidence rates accounting for 5-10% of diabetes occurrence (1). Due to the low β cell mass and insulin deficiency, type 1 diabetes is treated through subcutaneous insulin injection. Type 2 diabetes accounts for 85% of the global diabetes cases. It is strongly associated with the modern lifestyle and is characterized by insulin resistance and insulin deficiency (1). Drug interventions for type 2 diabetes are mainly aimed at lowering blood glucose levels but fail to prevent the progressive β cell dysfunction and death. These treatments can target β cells directly and enhance their function, e.g. the GLP-1 analogs that induce cAMP production and enhance glucose stimulated insulin secretion (3). Another group of compounds, called sulphonylureas, are also used. These compounds induce insulin secretion through closing of ATP dependent K+ channel without the need for ATP (4). Other mechanisms through which blood glucose levels can be lowered takes advantage of preventing the hepatic glucose production, e.g. metformin (5). However, due to the progressive nature of β cell failure, usually these therapies are subjected to intensification and combinations. Gestational diabetes mellitus is classified as hyperglycemia first diagnosed during the second or third trimester of pregnancy (6). Monogenic forms of diabetes are characterized by alterations in a single gene causing β cell dysfunction and diabetes. This form of diabetes accounts for <5% of diabetes (6). Based on the severity of the diabetes, different approaches can be taken to treat monogenic forms of diabetes. In severe cases insulin injection is prescribed. Where possible, therapies for monogenic forms of diabetes directly target the functional consequence of the mutation. In the case of KCNJ11 loss-of-function, a subunit of KAT¨P channel, sulphonylureas has had a great success.

Diabetes is recognized as a global health burden affecting more than 400 million people with expectations of it to rise to 642 million by the year 2040 (1). 10

Hyperglycemia increases the risk of several life threatening diseases, e.g. cardiovascular diseases and kidney diseases (7;8). The world health organization estimated glucose levels to be the third highest risk factor of early mortality and it is expected to be the 7th leading cause of death by the year 2030 (9;10). In the USA, it is the single most costly disease among 155 diseases, with an annual health care spending of 101 billion US dollars (11).

How β cells work The fuel sensing β cells work to regulate blood glucose levels through the hormone insulin. This is achieved through a fine-tuned ATP-mediated regulation of the + 2+ K (KATP) channels and voltage-dependent Ca channels (VDCC) (12). This mechanism is further enhanced by incretins released from the intestine (3). Insulin is the only glucose lowering hormone in the body, thus, understanding its regulation is fundamental to understanding the pathogenesis of diabetes.

Around 1 million islets of Langerhans are reported to exist within the human pancreas (13). These pancreatic islets consist of ~50% β cells (14). The β cell senses glucose levels mainly through the passive glucose transporters (GLUT) 1 and 3 (15). The glucose is initiated through its phosphorylation by glucokinase (GCK), a rate limiting step in insulin secretion (16). Phosphorylation through GCK is the driving event leading to glycolysis and energy production resulting in an increase in cytosolic ATP/ADP ratio. The increase in cytosolic ATP levels leads to the closure of KATP channels causing a depolarization of the membrane and the opening of VDCC and influx of Ca2+. Insulin secretion is a Ca2+- dependent process and is terminated through the opening of KATP channels and the subsequent closing of the VDCC (Figure 1) (17).

The glucose-stimulated membrane depolarization in β cells was described by the first time in 1968 by Dean and Mathews (18). Although much research has been focused on this since then, the exact molecular mechanisms connecting Ca2+ influx with insulin release are still under debate. The β cells secrete insulin in 2 phases, one rapid first phase of readily releasable pool (RRP) granules that are docked at the plasma membrane of the β cells (corresponding to 1-5% of the total granule content), and a second slower phase of non-docked vesicles (3). The insulin granules are docked to the membrane through the SNAP receptor (SNARE) complex of including synaptosomal-associated protein 25 (SNAP25), vesicle associated membrane protein (VAMP) 2 and syntaxin 1. However, these proteins are not Ca2+- sensitive. The most probable Ca2+ sensing protein within this complex is believed to be synaptogamin V and VII (3). The RRP are located in the vicinity of the Ca2+- channels, allowing for a very rapid release of insulin granules once Ca2+ influx has begun (19). The second phase is slower due to mobilization of the non-docked granules to the membrane in a time, ATP and Ca2+ dependent process called priming (3).

Glucose is the main signaling molecule for insulin secretion. However, a group of molecules called incretins enhance glucose stimulated insulin release. Incretins, e.g. 11 glucagon like peptide (GLP)-1 are released by the intestinal L cells upon food consumption. GLP-1 binds and activate a specific G-coupled receptor (GLP-1 receptor) leading to the activation of adenylate cyclase and generation of cyclic AMP (cAMP) (3). cAMP works in 2 distinct pathways to further enhance glucose stimulated insulin secretion. The first is through the activation of protein kinase A (PKA) leading to the phosphorylation of several mediators of insulin secretion (20). PKA phosphorylates SNAP25 and snapin, a protein shown to increase the interaction between SNARE complex proteins (21;22). The second pathway is mediated through exchange factor directly activated by cAMP (EPAC), a guanine exchange factor for Ras-related protein 1 (Rap1) (23) (Figure 1). Although Rap1 has several known functions, its exact role in exocytosis is not yet fully understood. It has been shown to localize together with insulin granules in MIN6 cells and deficiency of Rap1 decreased cAMP and high glucose-induced insulin secretion by ~40% (24).

Figure 1: Glucose stimulated insulin secretion. Glucose metabolism increases ATP/ADP ratio in the cells resulting in closure of KATP channels and depolarization of the membrane, opening of the VDCC and fusing of the insulin granules with the plasma membrane. GLP-1 enhances the glucose stimulated insulin secretion via cAMP. Type 2 and monogenic diabetes The international diabetes federation (IDF) estimates that more than 190 million people with type 2 diabetes are undiagnosed. Type 2 diabetes is strongly related to

12 the modern lifestyle including an overconsumption of high caloric diet (western diet) and reduced physical activity. It occurs most commonly in adults, but an increasing number of children and adolescents are diagnosed with type 2 diabetes (1;25). Obesity and insulin resistance are risk factors of type 2 diabetes. It has been demonstrated, however, that the disease manifests only when β cells fail to compensate for the increased demand on insulin secretion. Under normal conditions, insulin resistance is compensated for by an increased insulin production and secretion by the β cells. However, individuals progressing to impaired glucose tolerance and ultimately diabetes have reduced β cell function and survival (26-31). Longitudinal studies following insulin secretion and sensitivity provide evidence for early β cell failure in the pathogenesis of type 2 diabetes. Decreased insulin secretion is thus pivotal for the decline in glucose tolerance (30). Moreover, it has been shown that type 2 diabetes patients have a 30-60% decrease in β cell mass in combination with increased β cell apoptosis (29;31;32). In humans β cells are long lived, and there is limited evidence for β cell neogenesis (33;34). This suggests that β cell apoptosis and dysfunction are the most probable causes of type 2 diabetes.

Monogenic forms of diabetes are caused by mutations in single involved in β cell development, β cell function and survival, and cause diabetes independently of environmental factors. These forms of diabetes can be either syndromic (including features other than diabetes) or non-syndromic (35;36). Monogenic diabetes is commonly divided into two main types, namely neonatal diabetes mellitus (NDM) and maturity-onset diabetes of the young (MODY). In NDM the onset of diabetes occurs within the first 6 months of age (37). It is caused mainly by mutations in genes involved in development of endocrine pancreas and β cell function, e.g. KCNJ11, INS, PDX1 and GATA6 (36;38-40). MODY, on the other hand, usually occurs in young individuals before the age of 25 years (41). The hallmark of MODY is that it is inherited in an autosomal dominant manner (42). The most common types of MODY are caused by mutations in HNF1A and GCK, both important for β cell function (41).

Studies of these rare forms of diabetes can help in understanding polygenic type 1 and type 2 diabetes. Several genes causing monogenic diabetes have also been identified as susceptibility loci in genome wide association studies for type 2 diabetes (Table1) (41;43). One example of this is mutations in the human insulin gene leading to neonatal diabetes. The mutations in the insulin gene are inherited in an autosomal dominant manner causing diabetes at a median age of 9 weeks. The different mutations identified have been predicted to cause a misfolding of proinsulin in the ER causing severe endoplasmic reticulum (ER) stress and β cell death (40). Furthermore, mutations in several ER stress-related genes cause monogenic diabetes. For example, mutations in WFS1, an ER resident protein involved in the ER stress response, cause Wolfram syndrome characterized by young onset diabetes, optical atrophy and neurological disorders (44).

13

Table 1: Type 2 diabetes susceptibility genes involved in monogenic forms of diabetes(41;43)

Gene symbol Gene name General Function ABCC8 ATP binding cassette Modulator of ATP-sensitive

subfamily C member 8 potassium channels GCK Glucokinase Tissue specific glucose kinase

INS Insulin Glucose lowering hormone KCNJ11 Potassium voltage-gated Potassium Channel channel subfamily J member 11 PDX1 Pancreatic and duodenal (e.g. homeobox 1 modulates insulin expression) WFS1 Wolframin ER transmembrane Negative regulator of ER stress glycoprotein GLIS3 GLIsimilar family zinc finger 3 Transcription Factor HNF1A Hepatocyte nuclear factor 1 Transcription Factor homeobox A HNF1B HNF1 homeobox B Transcription Factor HNF4A HNF4 homeobox A Transcription Factor PAX4 Paired box 4 Transcription Factor SLC2A2 Solute carrier family 2 Low affinity glucose transporter member 2 in the plasma membrane NEUROD1 Neuronal differentiation 1 Transcription factor for E-box containing genes (e.g. modulates insulin expression) PPARG Peroxisome proliferator Transcription factor activated receptor gamma

Free fatty acids and diabetes The mechanisms underlying the progressive decline in β cell function in type 2 diabetes are not well understood (26;45). The development of central adiposity is associated with loss of β cell function, suggesting that visceral fat-derived factors may cause β cell dysfunction. Elevated levels of plasma saturated free fatty acids (FFAs) are connected with the development of diabetes, (25;46-49). At physiological conditions the concentrations of FFAs range between 0.5-1 mM depending on fasting levels (50). However, given that most of these FFAs are bound by albumin (51) the pathological factor to take into consideration is the unbound fraction of FFAs. Under normal conditions the concentrations of unbound FFAs is kept ~10 nM (52). However, during pathological conditions unbound FFAs increase to levels above 20 nM (53). Given that palmitate is the most common saturated FFA in circulation, making up around 25% of the total amount of FFAs, it has been widely used in studies of β cell dysfunction and apoptosis (51).

When studying the effects of saturated FFA on β cell demise, methodological limitations need to be taken into consideration. Due to the insoluble nature of FFAs in culture media (i.e. aqueous solutions), they need to be conjugated to albumin (54). 14

Thus, the ratio between albumin and FFAs need to be balanced to mimic the physiological conditions as closely as possible. For this reason several albumin preparations have been used. One of these methods takes advantage of charcoal absorbance to remove contaminants from the BSA preparation, including endogenous bovine FFAs, before use. Another method aimed at circumventing the issue of contamination, takes advantage of commercially available FFA-free albumin that provides a high degree of purity. These methods allow for the albumin to be dissolved in the culture medium and the addition of FFAs (that are dissolved in 90% ethanol) is done directly before exposing the cells. In contrast, another methods used allows the BSA to be pre-complexed with the FFAs before dissolving the mixture in medium, this provides a more stable preparation that can be stored for longer time. Oliveira AF et. al. has compared these methods for FFA preparation thoroughly (54). During this comparison the authors showed that with all these methods the concentration of the BSA is adjusted to keep the levels of FFAs at the physiological concentration of 0.5 mM and the levels of unbound FFAs at the pathophysiological concentration of ~26 nM for palmitate (54). For the studies constituting this thesis we have used both charcoal absorbed BSA and FFA-free BSA preparations to study the effects of FFAs on β cells.

Acute exposure to FFAs: enhanced β cell function Acute exposure to FFAs leads to an enhanced insulin secretion and increased β cell function (55). FFAs have been proposed as the main energy source of β cells during fasting, maintaining low levels of insulin secretion due to FFA oxidation by the mitochondria (56). Furthermore, also ATP-independent pathways have been suggested. FFAs are converted into long-chain Acyl-CoA (LC-CoA) before being transported into the mitochondria by carnithine-palmitoyl- (CPT)-I. Glucose metabolism leads to the increase of malonyl-CoA, an inhibtor of CPT-I (57). This in turn, would increase the amount of LC-CoA which has been proposed to be a signaling molecule that leads to insulin secretion through several pathways, e.g. protein kinase C activation (58). However, since its discovery, FFA receptor 1 (FFAR1) has been largely attributed to the enhanced effects of FFA on insulin secretion. FFAR1 is a G-coupled receptor for medium and long chain fatty acids. Its activation has been shown to lead to an increase in intracellular Ca2+ levels and enhance insulin secretion (59). It is proposed that the FFAR1 leads to the activation of phospholipase C (PL-C) leading to increased intositol triphosphate 3 (IP3) and the subsequent activation of the IP3 receptor leading to release of ER Ca2+ to the plasma membrane. In parallel, it has also been shown that FFAR1 activation leads to increase in cellular cAMP levels further enhancing insulin secretion (60;61). FFAR1 has also been shown to be expressed on intestinal cells where its activation leads to increased incretin, e.g. GLP-1, secretion (Figure 2) (62).

15

FFA

FFA FFAR1

Figure 2: Acute FFA exposure enhances β cell function. FFA cAMP induce insulin secretion through LC-CoA PLC generation of LC-CoA that has been described to be an effector molecule IP3 that leads to increased ATP and PKC activates PKC. More recently it has Metabolism been shown that FFA activation of FFAR1 leads to increased cellular IP3R cAMP levels and increased Ca2+ levels through PLC. ATP/ADP Ca2+

16

Prolonged exposure to saturated FFAs: Lipotoxicity-induced ER stress Prolonged exposure to saturated FFAs impairs pancreatic β cell function in vivo and in vitro and causes β cell death (63-66). The endoplasmic reticulum (ER) is a very important organelle in secretory cells, where synthesis, folding and maturation of secretory proteins occurs. ER stress is defined as an imbalance between the folding capacity of the ER and ER protein load. This loss in ER activates the ER stress response, also known as the unfolded protein response (UPR) (67;68). The UPR regulates protein synthesis to adapt it to the ER folding capacity, but when prolonged or exaggerated it triggers β cell apoptosis (67). Our group was the first to show that FFAs activate the ER stress response in β cells (69-73). In situations of insulin resistance, the demand on the ER is increased in β cells due to higher insulin production. This, in combination with increased circulating saturated FFAs, leads to ER stress, β cell dysfunction and death (46;65-67;70;74).

Although further studies are required to fully understand the effects of palmitate on ER stress, it has been shown that saturated FFAs induce a depletion of ER Ca+ levels and decreases the Ca2+ uptake in the ER (70;75). Since protein folding is a Ca2+-dependent process, FFAs might cause an accumulation of unfolded proteins inside the ER triggering the ER stress response. Also, computational studies have shown that palmitate can incorporate into bilayer membranes reducing their fluidity (76). Palmitate-induced changes in the ER membrane have been linked to reduced ER-to-Golgi transport (77;78). The Golgi apparatus is responsible for many post- translational modification on proteins aimed for the plasma membrane, and the disruption of this vesicular transport would lead to the accumulation of these proteins within the ER causing ER stress (78). Furthermore, changes in lipid composition of the ER have been shown to lead to the activation of IRE1 and PERK independently from accumulation of unfolded protein (79). More recently, it has been reported that palmitate decreases sorcin expression, a calcium binding protein that binds to the sarcoendoplasmic reticulum Ca2+ ATPase (SERCA) pump, and modulates the excitation-contraction coupling through changes in intracellular Ca2+ levels (80). The decrease in ER Ca2+ stores induces degradation of carboxypeptidase E (CPE), an that processes prohormone intermediates such as proinsulin. This decrease in CPE was shown to precede ER stress signalling and increase the proinsulin-to- insulin ratio, potentially leading to an accumulation of proinsulin in the ER and causing ER stress. When overexpressing CPE, cells where partially protected against palmitate-induced ER stress and cell death(81).

The unfolded protein response The UPR is mediated, mainly, through three ER transmembrane transducers: inositol-requiring 1 (IRE1), activating transcription factor 6 (ATF6) and PKR-like ER kinase (PERK). In non-stressed conditions, the chaperone binding immunoglobulin protein (BiP) binds to their ER lumenal domain keeping them inactive. During ER stress, BiP releases from these transducers to assist in protein folding leading to the activation of the UPR (82). The UPR signaling is highly conserved in mammalian cells with many similarities to yeast (83-85). 17

IRE1, once activated by dimerization and autophosphorylation, leads to the splicing of the X-box binding protein 1 (XBP1) mRNA leading to the translation of XBP1s. XBP1s is a transcription factor that controls the expression of chaperones and components of the ER-associated degradation (ERAD) machinery. This helps the ER to either fold the proteins properly or degrade the misfolded proteins and reduce the stress (86-88). Furthermore, IRE1 endonuclease activity is not only restricted to XBP1, but also leads to the splicing of an array of mRNAs, one being insulin, to reduce the protein load on the ER during the stress condition (89-91). Phosphorylated IRE1 also recruits tumor necrosis factor receptor-associated factor 2 (TRAF2) leading to the activation of c-Jun N-terminal kinase (JNK) resulting in apoptosis (92) (Figure 3).

ATF6 is translocated to the Golgi apparatus where its cytosolic domain (ATF6α) is cleaved and translocated into the nucleus where it acts as a transcription factor for an array of chaperones (e.g. BiP) and proteins of the ERAD machinery (93) (Figure 3).

BiP IPK IRE1 P58 Unfolded protein ER

ATF6 PERK Cytoplasm TRAF2

ER-targeted mRNA p-JNK Golgi XBP1t CReP Translation p-eIF2α XBP1s

PP1

ATF4

Nucleus CHOP

Apoptosis

Chaperones Figure 3: The unfolded protein response. Once activated, PERK phosphorylates eIF2α leading to translational attenuation. In parallel, eIF2α phosphorylation leads to the translation of ATF4 and its downstream targets CHOP and GADD34 a feedback loop induced repressor of eIF2α phosphorylation. Active IRE1 splices XBP1t leading to the translation of XBP1s a transcription factor for chaperones. IRE1 also cleaves other ER-targeted mRNAs leading to a decrease in translation. ATF6 is translocated to the Golgi apparatus where it is cleaved and the active domain is translocated to the nucleus where it acts as a transcription factor for chaperones, e.g. BiP. 18

Under basal conditions PERK is bound by BiP at its ER luminal domain and maintained as monomers. However, when unfolded proteins accumulate inside the ER, BiP releases from PERK allowing it to form oligomers and to autophosphorylate. Phosphorylated PERK leads to eIF2α phosphorylation in serine 51 and subsequent protein translation inhibition (Figure 4) (67;94-99). The kinase activity of PERK is also extended to Nrf2, a transcription factor for an array of anti-oxidant reactive elements helping the cell to recover from oxidative stress (100). Furthermore, PERK has been shown to interact with filamin A during ER Ca2+ depletion leading to an expansion of the ER-plasma membrane juxtaposition (101). The ER-plasma membrane interaction allows the ER to sense plasma membrane protein and lipid composition and ensures its function and integrity. The most studied function of ER-plasma membrane interaction is in Ca2+ homeostasis and store-operated Ca2+ entry (102;103). Although the effects of ER stress on the ER-plasma membrane interaction still need to be evaluated; and this function of PERK has been proposed to be independent of the UPR, it might help resolve the UPR by increasing store-operated Ca2+ entry (101).

Downstream of PERK, eIF2α can also be phosphorylated by other kinases that are activated by stress conditions other than ER stress, e.g. protein kinase R (PKR, activated during viral infection), general control nonderepressible 2(GCN2, activated during starvation) and heme regulated initiation (HRI, activated during heme deficiency). Phosphorylation of eIF2α and its downstream signaling cascade is also known as the integrated stress response (ISR) (104). PERK activity has been shown to also be regulated by 58 kDa inhibitor of PKR (P58IPK). Encoded by DNAJC3, P58IPK is a J-domain protein functioning as a co-chaperone inside the ER by recruiting unfolded proteins to BiP (105). P58IPK also functions in the cytoplasm of cells by binding and inhibiting PERK (106). It has been reported to also inhibit PKR and GCN2 (107;108) (Figure 4).

19

Figure 4: The PERK pathway of the UPR. Once activated, PERK phosphorylates eIF2α leading to translational attenuation. In parallel, eIF2α phosphorylation leads to the translation of ATF4 and its downstream targets CHOP and GADD34 a feedback loop induced repressor of eIF2α phosphorylation. eIF2α phosphorylation is also repressed by the constitutively active CReP.

The eIF2 complex, consisting of eIF2α, β and γ, binds and stabilizes the initiator Methionine (ini-Met) to the 40S ribosomal RNA (rRNA) for initiation of translation. eIF2β binds eIF5, a GTPase activating protein, responsible for inducing the GTPase activity of eIF2γ. This GTP to GDP conversion leads to a decrease in the affinity for ini-Met resulting in eIF2 releasing from the translation initiation complex. The GDP in eIF2 is then exchanged for a GTP by eIF2B, preparing the complex for the next initiation sequence. However, phosphorylation of the α subunit makes it bind eIF2B and abrogate its activity preventing the exchange of GDP to GTP and thus inhibiting initiation of translation (109-111) (Figure 5A). Despite the widely proven and accepted fact that serine 51 of the α subunit is phosphorylated, many other phosphorylation sites have been predicted (112). However, only one other site beside S51 has been experimentally shown to be involved in eIF2B binding, namely a serine at position 48. Although, this study shows that the S48A mutant eIF2α can still be phosphorylated by PKR, but it has a diminished binding ability to eIF2B. On

20 the other hand, these experiments do not show that S48 in itself is phosphorylated and how this phosphorylation would affect the function of the eIF2α(113).

The translational attenuation following eIF2α phosphorylation leads to the selective translation of some specific mRNAs, e.g. ATF4 and ATF5 (114). ATF4 mRNA has two specific uORF one short and one longer overlapping the proper ORF of the gene. At each of these uORF the 40s rRNA forms the complex with the 60s rRNA and starts translation. At the short uORF the active eIF2 complex is used and the 60s rRNA releases from the mRNA at the end of this uORF. Given the short distance between the first and the second uORF, during normal (i.e. unstressed condition) a new active eIF2 complex will bind the 40s rRNA before the reading of the second uORF is initiated causing the release of the initiation complex when there are abundant levels of eIF2 complex. However, once the eIF2α has been phosphorylated the low levels of active initiation complex take a longer time to form on the mRNA leading to the skipping of the second uORF and assembly on the gene ORF and increasing the translation of ATF4 (114;115) (Figure 5B). ATF4 is a principal regulator of stress-induced genes and cell response. As a transcription factor directly stimulated by the phosphorylation of eIF2α, ATF4 upregulates stress related genes, e.g. CCAAT/enhancer binding protein homologous protein (CHOP) and ATF3. ATF4 also forms complexes with other transcription factors, e.g. its downstream target CHOP, to promote the transcription of several other genes such as growth arrest DNA damage inducible 34 (GADD34) (116). More recently, ATF4 has also been implicated in the autophagy response upon starvation; this was shown to be due to formation of ATF4-CHOP heterodimers causing an increase in autophagy genes (117). Furthermore, ATF4 has been shown to increase the oxidative stress response of the cells, merging the ER stress response with the oxidative stress (104).

ATF4 translation leads to the transcription of CHOP, a proapoptotic transcription factor for genes involved in increased oxidative stress (e.g. ER oxidoreductase 1 (Ero1α)) and apoptosis (e.g. Tribblespseudokinase 3 (Trib3)) (118-120). As already mentioned CHOP and ATF4 form heterodimers and act as transcription factors for an array of genes. One of these downstream targets is GADD34, a repressor of eIF2α phosphorylation that is activated in a negative feedback loop during ER stress. Besides the stress induced GADD34, eIF2α phosphorylation is also controlled by the constitutive repressor of eIF2α phosphorylation (CReP). GADD34 and CReP are non-enzymatic co-factors of protein phosphatase 1 (PP1), directing it to dephosphorylate eIF2α (120;121). Protein phosphatese 1 is the most abundantly expressed serine/ phosphatase (122). With its nearly 200 biochemically proven interactors, PP1 is involved in a very wide range of biological processes, e.g. cell cycle progression, neuronal signaling, carbohydrate metabolism and protein translation (123;124). PP1 in itself has very low substrate specificity. Its high specificity to any given substrate is through the interaction of PP1 with its co-factors; i.e. PP1 is directed towards a specific substrate depending on which protein it is interacting with (125;126). It should be mentioned that the specificity of CReP-PP1 and GADD34-PP1 complexes to eIF2α is dependent also on the binding of G-actin to 21 the complex. It has been shown that binding of G-actin stabilizes the complex and gives it the proper selectivity (127;128).

Figure 5: The eIF2 translation initiation complex. (A) eIF2B exchanges the GDP to GTP on the eIF2 complex to re-activate it for a second round of initiations. During eIF2α phosphorylation, eIF2B is bound and inhibited preventing the GDP to GTP exchange. (B) The upstream ORFs of ATF4 mRNA.

Dysregulation of PERK/eIF2α pathway in diabetes The importance of ER stress signalling in β cells and its contribution to human β cell loss and diabetes is supported by monogenic forms of diabetes caused by mutations in ER stress transducers or in insulin itself (40;129;130), and the presence of ER stress markers in β cells from type 2 diabetes patients (131-133). In addition to its putative role in β cells, ER stress links obesity with insulin resistance (134), suggesting that this cellular stress response is a common molecular pathway for the two main causes of type 2 diabetes. ER stress has also been linked with innate

22 immunity and inflammation, making it relevant also in the pathogenesis of type 1 diabetes (135).

The important role of PERK activity in β cells is highlighted through monogenic forms of diabetes. Patients with Wolcott-Rallison syndrome or P58IPK loss-of-function mutations develop young onset diabetes (136;137). Delepine and colleagues studied the cause of Wolcott-Rallison syndrome and identified loss-of-function mutations in the EIF2AK3 gene encoding PERK (129;138). Similar to humans, PERK-/- mice are born with normal blood glucose levels, but develop diabetes during the first weeks of life due to β cell death (139). Synofzik and colleagues reported on 2 families with loss-of-function mutations in P58IPK leading to a syndrome including young onset diabetes and widespread neurodegeneration (92). P58IPK deficiency in mice leads β cell apoptosis and hyperglycemia mediated by increased CHOP expression and oxidative stress (140;141).

As mentioned before PERK phosphorylates eIF2α in serine 51. It has been shown that a homozygous (S51A) mutation in eIF2α causes a severe β cell deficiency in mice and they die 24h after birth. Mice heterozygous for the same mutation appear normal. However, if challenged with high fat diet they develop hyperglycemia due to β cell dysfunction (142;143). As discussed before, the phosphorylation state of eIF2α is repressed through the constitutively active CReP and the stress inducible GADD34 (121;144). CReP-/- mice die a few days after birth due to low hematocrit and red blood cell count (145).

On the other hand, it has been proposed that inhibiting GADD34 and thus prolonging eIF2α phosphorylation during stress is beneficial for the cell. This would allow longer time for the cells to recover from the ER stress before the onset of translation, leading to better cell survival (146;147). One molecule identified to prevent eIF2α dephosphorylation was salubrinal, a compound that was reported to protect PC12 cells against ER stress-induced apoptosis. This was shown to be due to inhibition of both GADD34 and CReP, leading to an increase in phosphorylated eIF2α (146).

Moreover, guanabenz, an α2-adrenergic receptor agonist, has been shown to directly bind GADD34 and to prevent its binding to PP1. This leads to an increase in the phosphorylated state of eIF2α. Guanabenz was shown to protect both mouse and rat clonal β cells from insulinAkita-induced apoptosis (147). InsulinAkita is a mutated form of insulin, in which a cysteine at position 96 has been replaced for a tyrosine in the insulin 2 gene. This mutation leads to the disruption of a disulfide bond causing the insulin to misfold inside the ER leading to ER stress (148;149).

As mentioned previously, ATF4 is a master regulator of the ER stress response downstream of PERK/eIF2α. Studies of ATF4-/- on β cells still need to be performed, however, the ATF4-/- mice rarely survive to adult age since most animals die within the first three weeks after birth. These mice have anemia and are smaller in size replicating some of the features found in the CReP-/- mouse (150;151).

23

CHOP, being the proapoptotic target of ATF4 has been widely studied in β cells. CHOP deletion in insulinAkita mice delays the onset of diabetes in this model (152). Also, CHOP deletion protected high fat-fed mice heterozygous for eIF2αS51A mutation from hyperglycemia. This was also true for mice put on high fat diet and given a moderate dose of streptozotocin. The streptozotocin treatment induces β cell death; however, a mild dose only reduces β cell mass, and the high fat diet increases the biosynthetic burden of insulin production. CHOP deletion protected these mice from the hyperglycemia induced by the combined treatment (153). Deletion of CHOP in the db/db mouse protected β cells from apoptosis, reversing the survival to control levels. The db/db mutation is a loss-of-function mutation in the leptin receptor causing obesity in mice and leading to overt hyperglycemia caused by an important increase in β cell apoptosis. CHOP deletion also prevented the induction of oxidative stress in these mice which could be mediated through the decrease in ERO1α induction (153).

The activating transcription factor 3 or ATF3 is another downstream target of eIF2α- ATF4 activation. Increased ATF3 expression has been shown to have proapoptotic effects. Cunha et al showed that ATF3 is activated downstream of PERK after palmitate treatment, leading to the transcription of the proapoptotic genes Death protein 5 (DP5) and P53 upregulated modulator of apoptosis (PUMA) resulting in β cell apoptosis (154). ATF3-/- islets have been shown to be protected against several insults, e.g. cytokine, nitric oxide (NO) and ER stress-induced apoptosis (155;156). When used for transplantation, ATF3-/- islets survived better and had less macrophage infiltration in the grafts, most likely due to the decrease in the expression of pro-inflammatory cytokines TNFα and IL-1β (157). Conversely, β cell specific overexpression of ATF3 in mice lead to islet dysfunction and reduction in insulin positive cells, further confirming the deleterious role of ATF3 in β cells (155;158).

β cell apoptosis Prolonged or unresolved ER stress response leads to β cell apoptosis, one of the major causes of type 2 diabetes. This programmed cell death event is mainly mediated through two separate pathways, namely the intrinsic and extrinsic pathway of apoptosis, converging in the activation of caspase 3 leading to apoptosis (Figure 6) (159).

24

Figure 6: The two main pathways of apoptosis. A) The intrinsic pathway is activated through intrinsic signals such as DNA damage leading to mitochondrial cytochrome c release. B) The extrinsic pathway of apoptosis is activated through the death receptors leading to caspase 8 triggering. Both of these pathways converge at the activation of the executioner caspase 3 and 7. (Adapted from (159))

25

The extrinsic pathway This pathway is activated through death receptors on the cytoplasmic membrane of the cells. Fas-associated death domain protein (FADD) is recruited and leads to the cleavage of caspase 8 triggering, in turn, the activation of the executioner caspases 3 and 7. Caspase 8 also functions as a mediator between the extrinsic and the intrinsic pathway through the Bcl2 homology 3 (BH3) interacting domain death agonist (Bid). Thus, the extrinsic pathway of apoptosis can be divided into mitochondrial-dependent and independent, where Bid plays a major role in this decision. However, the exact mechanism leading to one or the other is yet to be understood but seems to be cell specific (Figure 6). (159;160)

The intrinsic pathway The intrinsic pathway of apoptosis, also known as the mitochondrial pathway of apoptosis, is mediated through the B-cell lymphoma 2 (Bcl2) family proteins and the mitochondria. The Bcl2 proteins are balanced between pro-death and pro-survival proteins. Once this balance is disrupted and the pro-death signals prevail, the Bcl2 proteins lead to pore formation in the mitochondria and the release of several apoptotic proteins. One of these proteins is cytochrome c that binds to apoptotic protease activating factor 1 (APAF1) and forms the apoptosome. In turn, this complex leads to the cleavage of caspase 9 and activation of caspase 3 (Figure 6) (159).

The Bcl2 proteins can be divided into several groups; the sensitizers, the anti-death, the activators and the pro-death. Both the sensitizers and the activators have a BH3 only domain (known as BH3 only proteins) which can bind to the anti-death proteins such as Bcl2, Bcl-XL, Bcl2A1 and myeloid cell leukemia sequence 1 (Mcl-1). The sensitizers, including death protein 5 (DP5), Bcl2 associated agonist of cell death (Bad) and phorbol-12-myristate-13-acetate-induced protein 1 (Noxa), sense the apoptotic signals within the cells and bind the anti-death proteins making them release from the activators. The activators, e.g. PUMA and Bim, are then free to bind and activate the pro-death proteins, such as Bcl2 associated X protein (BAX) and Bcl2 antagonist/killer (BAK), which oligomerize at the mitochondrial membrane forming pores and leading to cytochrome c release (Figure 7). (161)

One of the most important pro-apoptotic BH3-only mediators is Bim (162-164). Due to , Bim has three isoforms; Bim-EL, Bim-L and Bim-S (165). Under normal conditions Bim-EL and L are sequestered to the cytoskeleton through binding of the dynein light chain 8 (LC8) (166). Upon apoptosis signaling, the two isoforms release from the cytoskeleton and bind the anti-apoptotic Bcl2 proteins (166). Bim-S, on the other hand, is the most potent isoform due to it lacking the dynein lightchain , thus, not retained by the cytoskeleton (165).

26

Figure 7: Interaction of the Bcl2 family members and the induction of apoptosis. During accumulation of intrinsic apoptosis signaling, e.g. ER stress, BH3-Only sensitizers bind anti-death proteins allowing the BH3-only activators to bind and activate the pro-death proteins and create pores in the mitochondrial membrane, leading to cytochrome c release. (adapted from (161))

Unresolvable ER stress signaling leads to the activation of the intrinsic pathway of apoptosis (159;160;167). IRE1 mediates apoptosis mainly through the activation of JNK. Phosphorylation of Bim-EL and L at Thr 112 by JNK induces their release from

27 the cytoskeleton and leads to apoptosis (168;169). Similarly, JNK can phosphorylate Bcl2 and Bcl-XL leading to their inactivation and apoptosis(170). JNK has also been shown to induce the expression of the proapoptotic DP5 through c-Jun(171).

As previously described, increasedATF3 expression upon palmitate treatment, leads to induction of the BH3-only proteins DP5 and PUMA (154). Upstream of ATF3, eIF2α phosphorylation has been shown to decrease the expression of the antiapoptotic Bcl2 protein Mcl-1in β cells undergoing ER stress (172). Furthermore, knockdown of ATF4 partially prevented Mcl-1 loss of expression in cells treated with tunicamycin (173). Moreover, CHOP has been shown to repress the expression of the promoter of Bcl2 sensitizing cells to tunicamycin and thapsigargin induced apoptosis (174).

28

Aims of this thesis Although much has been done to understand the underlying causes for type 2 diabetes, the molecular mechanisms contributing to disease development are yet to be discovered. Thus, the main aim of this thesis was to uncover new molecular mechanisms of β cell demise in type 2 diabetes. To achieve this goal, 1) we performed RNA sequencing of human islets treated with palmitate to have an unbiased global view on the effects of saturated FFAs on islet transcriptome; 2) I used both genetic and pharmacological approaches to study the involvement of eIF2α phosphorylation in β cell dysfunction and death. Particularly I studied the effects of a loss-of-function mutation in PPP1R15B, encoding CReP, causing a monogenic form of diabetes. I used guanabenz, a chemical inhibitor of GADD34, to modulate eIF2α phosphorylationin β cells. Through these approaches we strive to uncover new mechanisms causing β cell demise in type 2 diabetes.

29

Results

PAPER I Cnop M, Abdulkarim B, Bottu G, Cunha DA, Masini M, Turatsinze JV, Griebel T, Igoillo-Esteve M, Bugliani M, Villate O, Ladriere L, Marselli L, Marchetti P, McCarthy MI, Sammeth M, Eizirik DL; RNA-sequencing identifies dysregulation of the human pancreatic islet transcriptome by the saturated fatty acid palmitate.

Background: Pancreatic β cell dysfunction and death are central in the pathogenesis of type 2 diabetes. Saturated free fatty acids such as palmitate cause metabolic stress and βcell failure. Here we profiled the transcriptome of human islets exposed to palmitate to identify novel mechanisms of β cell demise. Methods: Five human islet preparations were RNA-sequenced basally or after 48- hour palmitate exposure (0.5 mM, 1% albumin). Organ donors were aged 55±9 years. Islet β cell purity assessed by insulin immunostaining was 50±5%. Samples were sequenced on Illumina Genome Analyzer II and data analyzed using GEM mapper and Flux Capacitor. Expression was considered modified if Benjamini- Hochberg-corrected Fisher testing (p<0.05) indicated a change in one direction in ≥4/5 samples and in the opposite direction in none. GATA6 silencing using 2 different siRNAs resulted in 60-80% protein knockdown. Results: A total of 30,026 transcripts corresponding to 19,882 genes were expressed in the human islets. Palmitate significantly modified 7% of these genes. The saturated fatty acid induced 428 genes and downregulated 897 genes, and modified the splicing of 574 genes (p<0.05). Palmitate induced fatty acid metabolism and endoplasmic reticulum (ER) stress. Functional studies suggested the intervention of novel mediators of adaptive ER stress signaling. Palmitate modified genes regulating ubiquitin and proteasome function, autophagy, and apoptosis. Inhibition of autophagic flux and lysosome function contributed to lipotoxicity. Palmitate inhibited transcription factors controlling β cell phenotype, including PAX4 and GATA6. Database for Annotation, Visualization and Integrated Discovery (DAVID) analysis of transcription factor binding sites in palmitate-modified transcripts revealed a role for PAX4, GATA, and the ER stress response regulators XBP1 and ATF6. Conclusion: We mapped the human islet transcriptome to identify novel mechanisms of palmitate-induced β cell dysfunction and death using RNA- sequencing. This human islet transcriptome study identified novel mechanisms of palmitate-induced β cell dysfunction and death. The data point to cross talk between metabolic stress and candidate genes at the β cell level.

My contribution to the study: I have contributed to ~35% of the work carried out, mostly performing gene validation studies and data analysis. Specifically, I have:

30

 Designed primers and siRNAs used in the study  Performed - Cell culture, transfection and cell treatments - Apoptosis analysis - mRNA extraction and reverse transcription - Western blots - Data analysis  Prepared tables based on the RNA sequencing bioinformatic analysis  Prepared Figures

31

1978 Diabetes Volume 63, June 2014

Miriam Cnop,1,2 Baroj Abdulkarim,1 Guy Bottu,1 Daniel A. Cunha,1 Mariana Igoillo-Esteve,1 Matilde Masini,3 Jean-Valery Turatsinze,1 Thasso Griebel,4 Olatz Villate,1 Izortze Santin,1 Marco Bugliani,3 Laurence Ladriere,1 Lorella Marselli,3 Mark I. McCarthy,5,6,7 Piero Marchetti,3 Michael Sammeth,4,8 and Décio L. Eizirik1 RNA Sequencing Identifies Dysregulation of the Human Pancreatic Islet Transcriptome by the Saturated Fatty Acid Palmitate

Diabetes 2014;63:1978–1993 | DOI: 10.2337/db13-1383

Pancreatic b-cell dysfunction and death are central in transcriptome study identified novel mechanisms of the pathogenesis of type 2 diabetes (T2D). Saturated palmitate-induced b-cell dysfunction and death. The fatty acids cause b-cell failure and contribute to diabe- data point to cross talk between metabolic stress and tes development in genetically predisposed individuals. candidate genes at the b-cell level. Here we used RNA sequencing to map transcripts ex- pressed in five palmitate-treated human islet prepara- tions, observing 1,325 modified genes. Palmitate induced Pancreatic b-cells are long-lived cells (1) that face pro- fatty acid metabolism and endoplasmic reticulum (ER) tracted metabolic challenges in insulin-resistant individu- stress. Functional studies identified novel mediators of ISLET STUDIES als (2). This includes the chronic exposure to saturated fi adaptive ER stress signaling. Palmitate modi ed genes free fatty acids (FFAs), present in a high-fat Western diet regulating ubiquitin and proteasome function, autophagy, and released from the adipose tissue in obesity (3). High and apoptosis. Inhibition of autophagic flux and lysosome levels of saturated FFAs are predictive of the future de- function contributed to lipotoxicity. Palmitate inhibited velopment of type 2 diabetes (T2D) (3). High-fat feeding transcription factors controlling b-cell phenotype, in- b cluding PAX4 and GATA6. Fifty-nine T2D candidate genes impairs the ability of -cells to compensate for insulin were expressed in human islets, and 11 were modified by resistance (4,5). Prolonged exposure to FFAs impairs in- palmitate. Palmitate modified expression of 17 splicing sulin secretion in vivo and in vitro (6,7) and induces factors and shifted alternative splicing of 3,525 transcripts. b-cell death (8) in a phenomenon called lipotoxicity. Ingenuity Pathway Analysis of modified transcripts and Palmitate is the most common saturated FFA in man genes confirmed that top changed functions related to and has been used in in vitro studies to examine the cell death. Database for Annotation, Visualization and mechanisms of lipotoxicity. Palmitate functionally impairs Integrated Discovery (DAVID) analysis of transcription b-cells by inhibiting insulin transcription (9), inducing mi- factor binding sites in palmitate-modified transcripts tochondrial uncoupling (10), and inhibiting exocytosis by revealed a role for PAX4, GATA, and the ER stress re- disrupting the coupling between Ca2+ channels and insulin sponse regulators XBP1 and ATF6. This human islet granules (11). The production of reactive oxygen species

1Laboratory of Experimental Medicine, ULB Center for Diabetes Research, Uni- 7Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, U.K. versité Libre de Bruxelles, Brussels, Belgium 8Laboratório Nacional de Computação Cientifica, Rio de Janeiro, Brazil. 2 Division of Endocrinology, Erasmus Hospital, Université Libre de Bruxelles, Brus- Corresponding author: Miriam Cnop, [email protected]. sels, Belgium Received 10 September 2013 and accepted 14 December 2013. 3Department of Endocrinology and Metabolism, University of Pisa, Pisa, Italy 4Functional Bioinformatics, Centre Nacional d’Anàlisi Genòmica, Barcelona, Spain This article contains Supplementary Data online at http://diabetes 5Oxford Centre for Diabetes, Endocrinology and Metabolism, Churchill Hospital, .diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1. Oxford, U.K. © 2014 by the American Diabetes Association. See http://creativecommons.org 6Oxford National Institute for Health Research Biomedical Research Centre, /licenses/by-nc-nd/3.0/ for details. Churchill Hospital, Oxford, U.K. See accompanying article, p. 1823. diabetes.diabetesjournals.org Cnop and Associates 1979

(10) and ceramides (12) has also been implicated in To better understand the global responses of human palmitate-induced b-cell dysfunction and death. We, and islets exposed to metabolic stress, we have used RNA-seq others, have previously shown that FFAs induce endoplas- to identify all transcripts, including splice variants, ex- mic reticulum (ER) stress in b-cells (13–16). The ER plays pressed in human islets of Langerhans after a 48-h exposure a central role in the synthesis and folding of secretory to the saturated FFA palmitate. This in vitro model of proteins. In b-cells, insulin represents up to 50% of the lipotoxicity arguably induces more rapid and harmful protein synthesized (17). ER stress, defined as an imbal- effects than those that may occur in vivo. Nonetheless, ance between protein folding demand and ER capacity, this analysis provides a snapshot of the cellular responses leads to accumulation of misfolded proteins. ER stress is under conditions that may prevail in T2D. Key findings sensed by the ER stress transducers PERK, IRE1, and ATF6 were validated and followed up by functional studies in that activate the unfolded protein response (UPR). The independent human islet samples and clonal or primary rat UPR attenuates protein translation to relieve the load on b-cells. We also examined whether putative candidate the ER and induces ER chaperones, ER-associated degrada- genes for T2D are expressed in human islets and modified tion, and ER expansion. UPR is an adaptive response but by palmitate. triggers apoptosis when prolonged or exaggerated. Saturated FFAs elicit marked PERK activity, and the resulting eIF2a RESEARCH DESIGN AND METHODS phosphorylation contributes to b-cell death (13,18) by the Human and Rat Islet Isolation and Cell Culture mitochondrial apoptosis pathway (19). Human islet collection and handling were approved by the Unbiased approaches to examine the b-cell response to local ethical committee in Pisa, Italy. Human pancreatic palmitate include microarray studies of clonal INS-1 and islets were isolated in Pisa using collagenase digestion and MIN6 cells. These studies showed induction of genes density gradient purification from beating- organ involved in FFA b-oxidation, FFA desaturation, steroid donors with no medical history of diabetes or metabolic ,cellcycle,chemokines, and acute-phase disorders (30). The donor characteristics are provided response genes, and inhibition of genes involved in in Supplementary Table 1. The first five preparations glycolysis and aminoacyl tRNA biosynthesis (20–23). Our (2 women and 3 men, donor age 55 6 9years,BMI time course microarrays of palmitate-treated INS-1E cells 24.8 6 0.7 kg/m2) were used for RNA-seq, and the other led to the identification of the proapoptotic Bcl-2 pro- preparations (15 women and 16 men, donor age 63 6 teins that mediate lipotoxic b-cell death (19). Oleate- 3years,BMI25.06 0.5 kg/m2)wereusedforconfirma- treated human islet arrays showed transcriptional tion and functional studies. The islets were cultured in induction of FFA oxidation, inflammatory genes, and M199 culture medium containing 5.5 mmol/L glucose antioxidant (24). in Pisa and were shipped to Brussels within 1–5days One of the intrinsic limitations of microarrays is that of isolation. In Brussels, the human islets were cul- transcript detection is limited to transcripts for which tured in Ham’s F-10 medium containing 6.1 mmol/L probesarepresentonthearrays.RNA-sequencing(RNA- glucose, 10% heat-inactivated FBS, 2 mmol/L GlutaMAX, seq) has become the gold standard for transcriptomic 50 mmol/L 3-isobutyl-1-methylxanthine, 1% charcoal- studies, allowing detection of low-expressed genes, absorbed BSA, 50 units/mL penicillin, and 50 mg/mL alternative splice variants, and novel transcripts (25). streptomycin. Our group has used RNA-seq to map the transcriptome Islet b-cell purity was evaluated in dispersed islet cells of human islets (26), and this has been recently repli- by insulin immunocytochemistry and averaged 52 6 3% catedinhumanisletsandpurified human b-cells (27); (26). The islets were exposed or not to 0.5 mmol/L pal- the latter study showed that islets are a good proxy for mitate in the same medium without FBS for 2 days b-cell transcript expression. (8,13,31). The serum-free culture conditions have previ- Environmental challenges interact with the genetic ously been validated (32). Human islet viability, assessed background of individuals to generate disease. Recent after Hoechst 33342 and propidium iodide staining, was genome-wide association studies (GWAS) have linked similar in serum-free or 10% FBS-containing medium (re- a number of genetic variants to susceptibility to T2D spectively 93 6 1% vs. 95 6 1% viable cells after 72 h, n = (28). Many of these variants seem to be related to pan- 8; P = 0.3). creatic b-cell function, but there is little information on Rat insulin-producing INS-1E cells (provided by the expression and function of these genes in human C. Wollheim, University of Geneva, Geneva, Switzerland) b-cells faced with prolonged metabolic stress. The known (33) were cultured in RPMI 1640 medium supplemented candidate genes for T2D explain less than 10% of the with 5% FBS, 10 mmol/L HEPES, 1 mmol/L Na-pyruvate, heritability of the disease (28). Epigenetic alterations, and 50 mmol/L 2-mercaptoethanol (34). The INS-1E cells such as changes in DNA methylation, have been described were exposed to palmitate, as described (13). in pancreatic islets from T2D patients (29) and may ex- Rats were used according to the Belgian Regulations plain part of the missing heritability. Whether the expres- for Animal Care with approval of the Ethical Committee for sion of these epigenetic T2D candidate genes is modified Animal Experiments of the Université Libre de Bruxelles, by metabolic stress has never been investigated. Brussels, Belgium. Islets were isolated from adult male 1980 RNA-Seq of Palmitate-Treated Human Islets Diabetes Volume 63, June 2014

Wistar rats (Charles River Laboratories, Brussels, Belgium), Human Islet and Rat b-Cell RNA Extraction and primary b-cells were fluorescence-activated cell sorter– and RT-PCR purified (FACSAria; BD Bioscience, San Jose, CA) and cul- Human islets (Supplementary Table 1), INS-1E, and tured as described (13). primary rat b-cells were used for validation and mech- anistic experiments. PolyA mRNA was isolated using RNA-seq and Data Analysis the Dynabeads mRNA DIRECT kit (Invitrogen, Paisley, Five human islet preparations were sequenced and data U.K.) and reverse-transcribed as previously described analyzed as previously described in detail (26). In brief, (34).Quantitative(q)RT-PCRwasdoneusingtheiQ polyA-selected mRNA was purified from total RNA isolated SYBR Green Supermix (Bio-Rad, Nazareth Eke, Belgium) with the RNeasy Mini Kit (Qiagen, Venlo, the Netherlands). on a LightCycler (Roche Diagnostics, Mannheim, Ger- mRNAs were reverse-transcribed to cDNA, paired-end re- many) or iCycler MyiQ Single Color (Bio-Rad) instru- paired, 39-monoadenylated, and adaptor-ligated. cDNA ment (41,42). Data were expressed as number of products (200 bp) were amplified and libraries submitted copies using the standard curve method and corrected to quality control with the Agilent 2100 Bioanalyzer (Agi- for the housekeeping gene b-actin or GAPDH. Primers lent Technologies, Wokingham, U.K.). The RNA integrity used for qRT- and RT-PCR are listed in Supplementary number (RIN) values for all samples were .7.5. cDNA Table 2. was sequenced on one sequencing lane of an Illumina RNA Interference Genome Analyzer II system (Illumina). The raw data will Human and rat b-cells were transfected with 30 nmol/L be deposited in Omnibus (GEO), submis- small interfering RNA (siRNA) and lipofectamine RNAiMAX sion number GSE53949. (Invitrogen) diluted in Opti-MEM I (Invitrogen) as de- Paired-end reads were mapped to the scribed (43), resulting in a transfection efficiency of (version GRCh37/hg19) using gem-mapper from the Geno- .90% (43,44). After overnight transfection, the cells mic Multitool (GEM) suite (http://gemlibrary.sourceforge were cultured for 48 h before further use. The siRNAs .net). Mapped reads were used to quantify transcripts are listed in Supplementary Table 3. from the RefSeq reference database (35), using the Flux Capacitor (http://flux.sammeth.net) (36). Genes and Western Blot transcripts were assigned a relative coverage rate as mea- Western blots were performed using equal amounts of sured in RPKM units (“reads per kilobase of model whole-cell extract protein as described (18). Briefly, cell per million mapped reads”) (37). Lists of differentially lysates were run on SDS-PAGE, washed in transfer expressed genes and transcripts were generated from buffer, and proteins were transferred to a nitrocellulose the Flux Capacitor output using scripts in Perl or membrane. The primary antibodies were anti–b-actin R. Palmitate-modified genes were defined by taking the (1:2,000), GATA6, and LC3B (both 1:1,000; Cell Signaling log2 of the proportion between the sum of the RPKM for Technology, Beverly, MA); LONP1 (Proteintech Group), all gene transcripts under palmitate condition and con- and anti–a-tubulin (1:10,000; Sigma-Aldrich). Horserad- trol condition. A Fisher exact test (number of reads mapped ish peroxidase-labeled donkey anti-rabbit or donkey anti- to the gene and number of reads mapped to all other mouse (1:10,000, Jackson ImmunoResearch, West Grove, genes in palmitate vs. control) was Benjamini-Hochberg– PA) antibodies were used as secondary antibodies. Protein corrected (taking for each gene the five samples as in- signal was visualized using chemiluminescence SuperSignal dependent tests), and a difference in gene expression (Pierce) and quantified using Scion Image (Scion Corp., was considered significant if the P value was ,0.05. A Frederick, MD). gene was considered modified by palmitate if its expres- sion changed significantly in one direction in at least Assessment of Apoptosis four of five islet preparations and no significant change The percentage of apoptotic cells was determined in at in the opposite direction was observed. Differences in least 500 cells per condition by staining with the DNA- splice indices—the proportion between the RPKM for binding dyes propidium iodide and Hoechst 33342 (Sigma- a transcript and the sum of the RPKM for all the tran- Aldrich), as previously described (18). scripts from the same gene—were compared between Electron Microscopy the palmitate and control condition. Splicing analysis Electron microscopy studies were performed on isolated was done using GENCODE version 16 annotations data human islets as previously described (30). sets (38–40). The GENCODE annotation data set used contains 153,008 transcripts, corresponding to 25,492 Statistical Analyses protein coding genes and long intergenic noncoding The statistical analyses of the RNA-seq data are described . Changes in splicing were statistically tested as in RNA-SEQ AND DATA ANALYSIS. Data for confirmation and above; that is, by Benjamini-Hochberg–corrected Fisher functional studies are shown as means 6 SEM. Compar- exact test-defined P value ,0.05 in at least four of five isons were performed by paired two-tailed Student t test islet samples and no sample pair exhibiting a significant or ratio t test. A P value # 0.05 was considered statisti- change in the opposite direction. cally significant. diabetes.diabetesjournals.org Cnop and Associates 1981

RESULTS expressed with higher RPKM. Similarly, a large proportion Sequencing of Palmitate-Treated Human Islets and of genes previously identified to harbor differential DNA Analysis of Transcripts methylation in T2D islets (29) were well expressed in hu- Five human islet preparations, exposed or not to palmi- man islets (Supplementary Fig. 3), with median expression tate for 48 h, were RNA-seq. The characteristics of the of 8 RPKM compared with 6 RPKM for all detected genes. organ donors and islet preparations are presented in Sup- fi plementary Table 1. The percentage of cell death in pal- Analysis of Palmitate-Modi ed Genes mitate-treated human islets was 12 6 2%, compared with We next analyzed the human islet genes that were fi 6 6 2% in the control condition (48 h, P , 0.05). The modi ed by palmitate (complete list accessible at http:// reads were mapped to the human genome (version lmedex.ulb.ac.be/data.php, with password provided on GRCh37/hg19) using GEM software, mapping on average request). These genes were analyzed using Ingenuity 85% of raw reads. Transcript expression and splicing was Pathway Analysis (IPA; Supplementary Fig. 4) and Data- evaluated using Flux Capacitor software. As a reference base for Annotation, Visualization and Integrated Discov- transcript annotation, we used the 42,012 annotated ery (DAVID) software, and they were manually curated human mRNA and noncoding RNA sequences from RefSeq. (Supplementary Table 4 and Fig. 2). IPA showed that “ ” Of the 18,463 genes detected by the RNA-seq, 1,325 upregulated genes belong to the functions Cell Death, “ ” “ (7%) were significantly modified by a 48-h exposure to Cellular Movement (mainly chemokines), Cellular De- ”“ ” “ ” palmitate, with 428 being upregulated and 897 being velopment, Gene Expression, and Lipid Metabolism A downregulated. Compared with our previous RNA-seq (Supplementary Fig. 4 ). Downregulated genes fell into “ ”“ analysis of cytokine-exposed human islets (26), there was the functional categories Cellular Movement, Cell Mor- ”“ ”“ ” limited overlap between the two stress conditions (Supple- phology, Lipid Metabolism, Molecular Transport, and “ ” B mentary Fig. 1A–C). Of the genes upregulated by cytokines, Small Molecule (Supplementary Fig. 4 ). 10% were also induced by palmitate, and of the cytokine- The manual annotation was performed taking a pancreatic inhibited genes, 19% were palmitate-regulated, showing that b-cell perspective. It showed induction of genes involved palmitate induced specific transcript expression changes. in lipid metabolism, including the transcription factor The genes detected as modified by palmitate by RNA- SREBP2, and early response genes that are part of an seq were compared with microarray data of human islets adaptive response (Fig. 2). Palmitate inhibited expression $ of key b-cell transcription factors, including PDX1, PAX4, from T2D donors and/or donors with HbA1c 6% (45). Of the genes differentially expressed in T2D and hypergly- MAFA, and MAFB, hormones and receptors, genes in- cemia, 7–16% were modified by palmitate in nondiabetic volved in ATP production, and channels and transporters, islets. In 82% of these genes, the change in expression thereby likely contributing to induce b-cell dysfunction. occurred in the same direction for palmitate exposure Upregulation of a large number of UPR genes and inhibi- and T2D (Supplementary Fig. 1D). tion of protein degradation pathways are likely to further For internal validation, expression data were confirmed contribute to b-cell dysfunction and death. Growth and fl for seven genes by qRT-PCR in the same islet samples used regeneration genes were inhibited. In ammatory responses fi for RNA-seq. The gene expression data were essentially were extensively modi ed, with upregulation of cytokines superimposable (Supplementary Fig. 2). Additional valida- and chemokines and inhibition of HLA (Supplementary tion was done by comparing RNA-seq data with qRT-PCR Table 4 and Fig. 2). in independent human islet samples for 30 genes, showing FFA Metabolism a correlation coefficient of 0.63 (Supplementary Fig. 2). Palmitate exposure induced gene expression of fatty acid metabolic pathways (Fig. 3A). It induced ACSL1 and ACSL3, Expression of Candidate Genes for T2D in Human involved in FFA activation to acyl-CoA moieties, CPT-1, Islets which mediates mitochondrial FFA uptake, ACADVL, We examined whether known T2D candidate genes (28) ECH1, and HADHA, three enzymes involved in mitochon- . are expressed (median RPKM 1) in human islets. We drial FFA b-oxidation, and two FFA desaturases that intro- fi de ned a set of 69 genes using the convention typically duce double bonds into saturated FFAs and, as such, are used in naming GWAS loci (i.e., in the absence of a strong b-cell protective (21). Interestingly, palmitate inhibited ex- biological candidate to choose the nearest gene to the pression of enzymes involved in the de novo synthesis of peak GWAS signal). It is likely that this gene set is ceramide (including SERINC5, SPTSSB, and CERS2). In par- enriched for transcripts that mediate the GWAS locus allel, it inhibited genes involved in the lysosomal breakdown effects. We did not exclude genes from loci that have of ceramide and sphingolipids (Fig. 3A). been shown to act through nonislet mechanisms. Of the 69 candidate genes, 59 (86%) were present in human Protein Synthesis/Processing and ER Stress islets (Fig. 1A). This was a significantly higher proportion Palmitate induced the aminoacyl tRNA synthetases IARS, than that of candidate genes associated with ulcerative GARS, MARS, WARS, VARS, CARS, and SARS and the colitis (46) or body height (47), or a random set of 60 translation EEF1A2, involved in delivery genes (Fig. 1B). The T2D genes were also more abundantly of aminoacyl tRNAs to the . Protein translation 1982 RNA-Seq of Palmitate-Treated Human Islets Diabetes Volume 63, June 2014

Figure 1—GWAS-based T2D candidate genes are well expressed in human islets. A: Expression levels of transcripts of T2D candidate genes with an expression of RPKM >1 (i.e., 80% of currently known T2D candidate genes). Red bars indicate significantly upregulated transcripts and green bars, downregulated transcripts. B: Box plot of median expression levels of all T2D candidate genes compared with all genes, a random set of 60 genes, or the candidate genes associated with ulcerative colitis or body height. The numbers above the figure show the percentage of genes considered present (RPKM >1) and the P value for the Fisher exact test of the selected gene set vs. all genes. The horizontal line in the middle of each box indicates the median; the top and bottom borders of the box mark the 75th and 25th percentiles, respectively; the whiskers mark the highest and the lowest data point still within 1.5 interquartile range above the 75th percentile and 1.5 interquartile range below the 25th percentile, respectively; and the circles indicate outliers. diabetes.diabetesjournals.org Cnop and Associates 1983

Figure 2—Overview of human islet transcripts modified by palmitate. Manual curation of RNA-seq–detected human islet transcripts modified by a 48-h exposure to palmitate into functional categories and b-cell outcomes. Upregulated genes are shown in red and downregulated genes in green. Expression changes leading to b-cell dysfunction include inhibition of key b-cell transcription factors, changes in hormones and receptors, the ER stress response, b-cell signal transduction, inhibition of ATP production, potassium channels, and a cytosolic stress response. Other transcript changes may contribute to b-cell loss, including inhibition of cell growth and regeneration factors, inhibition of autophagy, and changes in apoptosis-related genes. Adaptive responses are those related to FFA metabolism and early response transcription factors. The potential role and outcome of the induction of tubulin transcripts, the upregulation of innate immunity, and the downregulation of HLA are undefined.

initiation factors were modulated, with induction of mildly induced by palmitate (Fig. 5A). The induction was EIF4A1 and inhibition of EIF4A2 expression, and inhibi- not detected in independent human islet samples (Fig. 5B), tion of the translational repressor EIF4G2. Hormone but palmitate did induce CREB3 expression in INS-1E cells processing was affected with an induction of CPE and (Fig. 5C). CREB3 silencing (using two independent siRNAs, PCSK1 and inhibition of PCSK4 and SCG5 (Fig. 3B). Fig. 5C) markedly sensitized the cells to palmitate-induced The RNA-seq data indicated transcriptional activation apoptosis (Fig. 5D). Similarly, efficient CREB3 knockdown of the three branches of the UPR, including PERK- in primary rat b-cells enhanced lipotoxicity, nearly dou- dependent induction of ATF4, ATF3, TRIB3, and GADD34; bling palmitate-induced apoptosis (Fig. 5E and F). This ATF6-dependent induction of BiP and its was confirmed in human islets, where CREB3 mRNA DNAJB11; and IRE1-dependent induction of chaperones knockdown by 67–72% potentiated lipotoxicity in two in- and protein disulfide (Fig. 3B). In line with this, dependent preparations (Fig. 5G). CREB3L3 is expressed at DAVID analysis using UCSC_TFBS showed enrichment for lower levels in human islets but was markedly induced by potential binding sites for the transcription factors ATF6 palmitate (Fig. 5H); this was confirmed in independent and IRE1-dependent XBP1 (Fig. 4A). human islet samples by qRT-PCR (Fig. 5I). CREB3L3 was Interestingly, palmitate induced expression of the ER also induced by oleate, but inhibited by synthetic ER stres- stress transducers CREB3 and CREB3L3 (Fig. 3B), which sors (Fig. 5I), suggesting it mediates an adaptive UPR in may play roles similar to ATF6 in a tissue-specificway responsetoFFAs,butfailstodosointhefaceofsevere (48,49). Because their role in b-cells is unknown, we stud- chemical ER stress. Taken together, these data suggest ied these ER stress transducers further. CREB3 was well a novel role for CREB3 and CREB3L3 in adaptive b-cell expressed in human islets, with an RPKM of 15, and was UPR signaling. 1984 RNA-Seq of Palmitate-Treated Human Islets Diabetes Volume 63, June 2014

Figure 3—Impact of palmitate on human islet metabolic pathways and protein synthesis, ER stress response, and protein degradation pathways. A: Manual curation of palmitate-modified human islet transcripts that may affect metabolic pathways pertaining to the Krebs cycle, oxidative phosphorylation, amino acid catabolism, FFA metabolism, ceramide synthesis and metabolism, and triglyceride synthesis. B: Manual curation of palmitate-modified human islet transcripts that may affect protein synthesis and processing, pertaining to tRNA synthesis, protein translation, and hormone processing pathways, as well as branches of the ER stress response controlled by IRE1, PERK, and ATF6, ER quality control (QC), and ER-associated degradation (ERAD), ubiquitination, proteasomal degradation, autophagy, and lysosomal function. Upregulated genes are shown in red and downregulated genes in green.

Proteasomal Function and Autophagy in an increased LC3 II-to-I ratio (Fig. 6B and D) and accu- Palmitate inhibited expression of the ubiquitin-conjugating mulation of autophagosomes in human b-cells (Fig. 6E). enzymes UBE2H and UBE3A; the deubiquitinating enzymes The present RNA-seq data identified inhibition of a number USP2, USP54, and USP30; ubiquitin D, a proteasomal de- of autophagy-related and lysosomal function–related genes gradation signal; and modulated expression of components that may directly affect lysosome– autophagosome fusion of the proteasome (Fig. 3B). (Fig. 3B). Thus, ATG7 and WIPI2 were inhibited, as were FFAs have previously been shown to induce autopha- the positive regulators of autophagy SCOC, DRAM2, and gosome formation (50,51) and impair autophagic flux in KIAA1324. ATP6AP2, an accessory protein to the H+-ATPase, b-cells (52). We confirmed that palmitate induces conver- and the adaptor-related protein complex subunit AP2M1, sion of microtubule-associated protein 1 light chain 3 (LC3) both of which contribute to lysosomal acidification, were from its native (I) to the lipidated form (II) in INS-1E cells inhibited. The inhibition of adaptor-related protein com- (Fig. 6A and B) and human islets (Fig. 6C and D), resulting plex subunits AP3B1 and AP3M2 and cathepsins F, O, S, A, diabetes.diabetesjournals.org Cnop and Associates 1985

Figure 4—DAVID analysis of palmitate-modified genes in human islets. RNA-seq data of transcripts upregulated (A) or downregulated (B) by palmitate or with modified alternative splicing (C) were analyzed for term enrichment against UCSC_TFBS. The length of the bars indicates the significance of the overrepresentation of potential binding sites for the indicated transcription factors in the modified genes, expressed as minus the logarithm of the probability that a set of genes taken at random from the human genome would pop up the same entries. Only the 30 top entries are displayed. The vertical line indicates a probability threshold of 0.05, corresponding to a 2log(BH P value) of 1.3. and D may also affect lysosomal function. To examine Genes involved in mitochondrial fission (DNM1L, FIS1, whether stimulating autophagic flux would protect b-cells, MFF) and fusion (MFN1, MFN2, OPA1) were well expressed we used the autophagy-enhancingdrugcarbamazepine in human islets, with a median RPKM of 12, compared (53). Carbamazepine protected INS-1E cells from lipo- with a median expression of 6 RPKM for all human islet toxicity in a dose-dependent manner (Fig. 6F). In human transcripts. Palmitate did not modify their expression, sug- islets, carbamazepine promoted LC3 II disappearance gesting that the mechanism(s) leading to impaired mito- (Fig. 6G and H), suggesting increased clearance of auto- chondrial fusion is not transcriptional. phagosomes (54), and effectively protected against pal- Among the mitochondrial enzymes induced by mitate-induced cell death (Fig. 6I). palmitate was Lon peptidase 1 (LonP1, Supplementary Fig. 5B). This AAA+ protease is involved in mitochondrial Mitochondrial Dynamics and Quality Control protein quality control. It degrades misfolded or oxidized We searched the human islet RNA-seq data for genes proteins and acts as a chaperone in the assembly of pro- involved in mitochondrial movement, biogenesis, fusion- tein complexes (56). We confirmed that LonP1 is induced fission, and mitophagy (Supplementary Table 5). Mito- under lipotoxic conditions in rat b-cells (Supplementary chondrial fragmentation, which plays a role in lipotoxic Fig. 5C and D). LonP1 induction by palmitate (57) may b-cell apoptosis (55), was detected ultrastructurally in pal- occur in response to a stoichiometric imbalance in nu- mitate-treated human islets (Supplementary Fig. 5A). clear- and mitochondrial-encoded proteins. 1986 RNA-Seq of Palmitate-Treated Human Islets Diabetes Volume 63, June 2014

Figure 5—Role of novel ER stress transducers in lipotoxic b-cell death. A: RNA-seq data of CREB3 expression in five human islet preparations exposed to palmitate for 48 h. B: CREB3 mRNA expression assessed by qRT-PCR in human islets exposed to 0.5 mmol/L oleate (OL), 0.5 mmol/L palmitate (PAL), 1 mmol/L thapsigargin (THA), 5 mg/mL tunicamycin (TU), or 0.1 mg/mL brefeldin A (BFA) (n =3–6). CREB3 mRNA expression assessed by qRT-PCR (C) and apoptosis (D) in INS-1E cells transfected with control (negative) or two different CREB3 siRNAs and then treated with 0.5 mmol/L palmitate for 16 h (n = 3). CREB3 mRNA expression (E) and apoptosis (F)influorescence- activated cell sorter–purified primary rat b-cells transfected with control or CREB3 siRNAs and then treated with 0.5 mmol/L palmitate for 24 h (n = 3). G: Apoptosis in dispersed human islet cells transfected with control or CREB3 siRNAs and exposed to palmitate for 24 h. Individual data for two independent human islet preparations are shown. H: RNA-seq data of CREB3L3 expression in five human islet preparations exposed to palmitate for 48 h. I: CREB3L3 mRNA expression assessed by qRT-PCR in human islets exposed to ER stressors as in panel B (n =3–6). *P < 0.05 vs. control (CTL) or control cells transfected with negative siRNA; #P < 0.05 as indicated.

Insulin Secretion in ATP production in the Krebs cycle and respiratory Palmitate is known to inhibit glucose-induced insulin chain, including citrate synthase, 2 isocitrate dehydro- release, and this was confirmed here. Palmitate-treated genases, and components of complexes I, II, III, and IV human islets increased insulin secretion after high glucose and mitochondrial ATP synthase (Fig. 3A and Supplemen- stimulation by 1.4 6 0.1-fold, compared with a stimula- tary Table 4). This inhibition, but not transcriptional tion index of 2.1 6 0.2 for nonexposed islets (P , 0.01, inhibition of the distal steps of the stimulus-secretion Supplementary Fig. 6). We performed a detailed analysis pathway, may contribute to loss of insulin secretion of the effect of palmitate on the expression of human (60,61). islet genes that modulate insulin secretion mostly at the Interestingly, several of the genes implicated in stimulus- level of membrane depolarization and Ca2+ entry (58). secretion coupling have splice variants. For at least some 1 Palmitate inhibited only 2 of the 48 genes (Supplemen- (KCNMA1, Ca2 -sensing receptor, and CLCN3), palmitate tary Table 6), including the Ca2+-sensing receptor, whose alters splicing (Supplementary Table 6). Future studies activation contributes to human islet insulin secretion should investigate the functional impact of these changes (59). Palmitate did inhibit expression of genes involved in splicing. diabetes.diabetesjournals.org Cnop and Associates 1987

Figure 6—Role of autophagy in lipotoxic b-cell death. LC3 conversion in INS-1E cells (A and B) and human islets (C and D) exposed to 0.5 mmol/L palmitate (PAL) for the indicated times (n =3–4). B and D: LC3 I and II protein expression was quantified by densitometry and normalized to tubulin or b-actin or expressed as the ratio of LC3 II-to-I. E: Accumulation of autophagosomes (double membrane organelles containing rough ER and mitochondria and/or partially degraded ER; arrows) in human islets exposed to palmitate for 48 h. F: Apoptosis in INS-1E cells exposed to 0.5 mmol/L palmitate alone or in combination with the indicated concentrations (in mmol/L) of carbamazepine for 24 h (n =5–7). G and H: LC3 conversion in human islets exposed to 0.5 mmol/L palmitate and/or 30 mmol/L carbamazepine for 24 h (n = 3). I: Human islet cell death after 24 h exposure to palmitate and/or carbamazepine (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001 vs. control (CTL); #P < 0.05 as indicated. 1988 RNA-Seq of Palmitate-Treated Human Islets Diabetes Volume 63, June 2014

Transcription Factors transcripts has a mean value of 0.12, and NLRP3 is not Palmitate downregulated PDX1, MAFA, MAFB, PAX4, and induced by palmitate (0.06 RPKM), one of the priming NEUROD1, which are important for the maintenance of steps for NLRP3 activation. PYCARD and caspase 1 are b-cell function and its differentiated state (Fig. 2). The expressed (mean RPKM of 7.9 and 4.1, respectively) but RNA-seq data also pointed to a mild inhibition of GATA6, not modified by palmitate. These data argue against a well-expressed transcription factor (RPKM 7, Fig. 7A). a proapoptotic role of TXNIP and the inflammasome in Heterozygous GATA6 mutations cause pancreatic agene- human islets facing lipotoxicity or ER stress. sis and neonatal diabetes (62) and milder phenotypes Palmitate upregulated chemokines and cytokines, such as adult-onset diabetes. We confirmed by qRT-PCR including IL-6, IL-1A, IL-33, IL-8, CXCL1, and CXCL2 that palmitate inhibits GATA6 expression in independent (Fig. 2 and Supplementary Table 4), in line with previous human islet samples and in primary rat b-cells (Fig. 7B findings (31); the role of these mediators remain to be and C). GATA6 was silenced by transfecting rat or human defined. islet cells with two different siRNAs (Fig. 7D–F). GATA6 Long Noncoding RNA knockdown induced apoptosis under the basal condition Our experimental design and analysis was not directed at and accentuated lipotoxicity (Fig. 7G–I). These data point long noncoding RNA (lncRNA) discovery. From the RefSeq to a novel role for GATA6 in adult b-cells and suggest that database, 1,297 of the 3,267 known noncoding RNAs its inhibition by palmitate contributes to lipotoxicity. In- (a global class of noncoding RNAs, including lncRNAs, terestingly, the promoter regions of transcripts that were miRNAs, , unspliced transcripts, etc.) were downregulated by palmitate were enriched in potential present in human islets (RPKM . 1). Recently, a large GATA binding sites (Fig. 4B). number of previously unknown lncRNA were identified in human islets and b-cells (65). We detected 349 of these Cell Death and Inflammatory Responses 1,128 lncRNA; of these, 9 (2%) were modified by palmitate, Palmitate induced mRNAs encoding pro- and antiapop- showing their responsiveness to metabolic stimuli. The in- totic proteins (Fig. 2 and Supplementary Table 4); for duction of MALAT1, one of the most abundant lncRNA, example, palmitate induced GRAMD4, which inhibits was confirmed by qRT-PCR (Supplementary Fig. 2). the antiapoptotic Bcl-2 protein and promotes Bax trans- location to the mitochondria. Some proapoptotic genes Palmitate-Induced Changes in T2D Candidate Gene were inhibited, including TP53INP1, caspase 2 and 10, Expression and the proapoptotic Bcl-2 family members BMF and Palmitate inhibited expression of TSPAN8, KCNK16, BCL2L11, which encodes Bim. The latter is in keeping ADCY5, ADRA2A, TP53INP1, CDC123, and PRC1 but with our previous findings that Bim does not play a role induced C2CD4A, ADAMTS9, and SPRY2 (Fig. 1). For one in lipotoxic b-cell apoptosis (19). Palmitate also inhibited of these, we evaluated the functional consequence of its genes with antiapoptotic functions, including c-Flip, downregulation. ADCY5 silencing (by 50%) markedly sen- DDX17, TM7SF3, DCAF7, ADCYAP1, ANXA4, NMT1, sitized rat b-cells to apoptosis, basally and after palmitate and PRDX6. exposure (Supplementary Fig. 8). TXNIP expression was inhibited by palmitate; this was confirmed by qRT-PCR (Supplementary Fig. 7A). Palmitate-Induced Alternative Splicing in Human Islets – High glucose tended to increase TXNIP expression, Of the 212 human islet expressed splicing factors, 17 fi A but palmitate prevented the induction of TXNIP by glu- were modi ed by palmitate (Fig. 8 ). Among these was B fi cose (Supplementary Fig. 7A). TXNIP inhibits insulin SRSF3 (Fig. 8 ). We con rmed by qRT-PCR that palmitate secretion and promotes apoptosis; its inhibition indi- induces SRSF3 in independent human islet samples and in C cates it does not mediate lipotoxicity. This is in keeping rat b-cells (Fig. 8 ). with an earlier report showing that TXNIP deficiency Exposure of human islets to palmitate altered splic- protects against gluco- but not lipotoxicity (63). Chen ing of 574 genes, with 363 and 462 splice variants being et al. (63) also suggested that TXNIP does not mediate up- and downregulated, respectively, using RefSeq thapsigargin-induced b-cellapoptosis.Incontrast,are- annotation (Supplementary Fig. 9A). IPA of palmitate- cent report indicated that ER stress induces TXNIP and modified splice variants identified “Cell Growth and thereby causes NLRP3 inflammasome activation and Proliferation” and “Cell Death” as the main categories (Sup- interleukin (IL)-1b–driven human islet apoptosis (64). plementary Fig. 9B). Because RefSeq provides a conservative We previously showed that an IL-1 receptor antagonist catalog of splice variants, we reanalyzed the RNA-seq data does not protect human islets from palmitate (31). Sim- using the ENCODE-based GENCODE data set (version 16), ilarly, the IL-1 receptor antagonist did not protect human which provides four-to-fivefold more transcripts. This in- islets against apoptosis induced by thapsigargin or brefel- creased the number of splice transcripts modified by pal- din A, whereas it effectively protected against the cyto- mitate by more than sixfold, to 3,525, corresponding to kines IL-1b and interferon-g (Supplementary Fig. 7B). 2,858 genes (Fig. 8D). Compared with the splicing induced Notably, the NLRP3 inflammasome has very low expres- in human islets by cytokines, there was little overlap (14% sion in human islets. The RPKM sum of the five NLRP3 only, Supplementary Fig. 1C), showing a stress-specific diabetes.diabetesjournals.org Cnop and Associates 1989

Figure 7—GATA6 inhibition by palmitate may contribute to b-cell death. GATA6 expression in five human islet preparations exposed or not (CTL) to palmitate (PAL) for 48 h, measured by RNA-seq (A) and confirmed by qRT-PCR (B). C: GATA6 expression by qRT-PCR in fluorescence-activated cell sorter–purified primary rat b-cells exposed to palmitate for 48 h (n = 4). Primary rat b-cells (D), INS-1E cells (E), and dispersed human islets (F) were transfected with control siRNA (CTL) or two different siRNAs targeting GATA6 (#1 and #2). GATA6 mRNA expression was assessed by qRT-PCR (n =4–7) (D and F) and protein expression by Western blot (n =3)(E). Apoptosis in primary rat b-cells (G) and dispersed rat (H) and human islets (I) transfected with control or GATA6 siRNAs and then exposed to palmitate for 24 h (n =3–6). *P < 0.05, **P < 0.01 vs. untreated cells; #P < 0.05, ##P < 0.01 as indicated. 1990 RNA-Seq of Palmitate-Treated Human Islets Diabetes Volume 63, June 2014

Figure 8—Palmitate-induced changes in alternative splicing according to GENCODE annotation. A: Palmitate modified the expression of many splicing factors in human islets; upregulated genes are shown in red, downregulated genes in green. B: RNA-seq data of SRSF3 expression in five human islet preparations exposed to palmitate for 48 h. C: SRSF3 mRNA expression assessed by qRT-PCR in in- dependent human islet samples (n =5)(upper panel) and INS-1E cells (n =6)(lower panel) exposed or not (CTL) to palmitate (PAL) for 48 and 24 h, respectively. *P < 0.05, **P < 0.01 vs. untreated cells by ratio t test. D: Palmitate exposure led to marked changes in alternative splicing. A total of 1,403 transcripts were significantly upregulated in at least 4 of 5 islet samples and were significantly downregulated in none, and 2,122 transcripts were significantly downregulated using similar criteria. The Venn diagram illustrates the number of genes that have transcripts modified in both directions (intersection) and in only one direction. E: IPA of the 3,525 genes with modified splicing. The length of the blue bars indicates the significance of the association between the set of transcripts and the keyword and is expressed as minus the logarithm of the probability that a random set of transcripts from the human genome would be associated with the same keyword. The straight orange line indicates a threshold of 0.05, corresponding to a 2log(BH P value) of 1.3. Only the top 20 categories are shown.

splicing response. IPA of palmitate-modified splice variants alternative splice variants, and novel transcripts, identifying identified “Cell Death and Survival,”“Organismal Survival,” 25–75% more genes than conventional arrays and more and “Gene Expression” as the main categories (Fig. 8E). differentially expressed transcripts (25,66–68). We used it DAVID analysis against UCSC_TFBS identified among the here to map the global response of human islets facing palmitate-modified splice variants enrichment for potential metabolic stress induced by palmitate. The presently used binding sites for the transcription factors ATF6, ELK1, in vitro model of lipotoxicity induces cellular responses in PAX4, and PPARA (Fig. 4C). the islets over a 48-h period, compared with slower and more heterogeneous events in vivo. The effects in vivo DISCUSSION may also be attenuated by the presence of unsaturated RNA-seq is a highly reproducible method to interrogate FFAs that decrease the lipotoxicity of saturated FFAs (8). the whole transcriptome and identify novel cellular The picture emerging from the analysis of palmitate-treated responses to environmental cues (25). Different from micro- human islets indicates a complex adaptive response, in- arrays, it allows detection of high- and low-abundance genes, cluding upregulation of lipid metabolism and disposal, diabetes.diabetesjournals.org Cnop and Associates 1991 paralleled by inhibition of the Krebs cycle and oxidative autophagic flux in b-cells and thereby contribute to phosphorylation. There are several signals of cellular stress lipotoxicity (52). b-Cells from T2D patients show signs responses, including cytosolic stress, mitochondrial quality of altered autophagy, including increased autophagic control, and activation of an array of genes regulating the vacuole and autophagosome volume density and reduced UPR and pathways of apoptosis. There was also inhibition LAMP2 and cathepsin B and D expression in T2D islets of genes regulating protein degradation and autophagy, (51), in keeping with our RNA-seq findings. We used car- which may aggravate the ER stress by preventing disposal bamazepine to stimulate autophagic flux (53) and showed of misfolded proteins. The transcriptome data further sug- marked b-cell protection from lipotoxicity (Fig. 6F and I). gest that palmitate leads to loss of the b-cell differentiated Carbamazepine is an antiepileptic and mood-stabilizing phenotype, with inhibition of key b-cell transcription fac- drug. Compared with other antiepileptics and atypical tors, hormones, and receptors. Some cytokines and chemo- antipsychotics, carbamazepine is associated with lesser kines were induced while HLA genes were inhibited. diabetes risk, but no study reported protection from di- Palmitate also induced changes in the alternative splicing abetes, possibly because it can impair b-cell function (70). of more than 3,500 transcripts. These data extend signif- Palmitate modified splicing of 3,525 transcripts, de- icantly beyond previous microarray findings (19–24). Thus, tection of which is missed by conventional microarrays. the RNA-seq analysis identified modulation of the tran- The alternative splicing is different from that observed scription factors GATA6, PAX4, CREB3, and CREB3L3 by after exposure of human islets to proinflammatory palmitate. cytokines (26), suggesting that different forms of b-cell GWAS have identified more than 60 loci associated stress lead to different splicing signatures, probably with T2D. The present and previous observations (26,27) through the activation of distinct splicing factors. indicate that .80% of the putative candidate genes are In conclusion, the present transcriptomic study pro- expressed in human islets, and we show here that some vides a snapshot of the b-cell responses to conditions that are modified by palmitate exposure. may contribute to T2D pathogenesis. The transcript changes Palmitate inhibited the transcription factors PDX1, induced by palmitate differ from cytokine-induced stress MAFA, MAFB, NEUROD1, PAX4, and GATA6, all of which (26), showing activation of stress-specificsignaturesin play important roles in b-cell differentiation. Palmitate b-cells. The RNA-seq data identify novel players in palmi- modification of GATA6 and PAX4 expression has not tate-induced b-cell dysfunction and death and indicate sev- been previously reported. In the DAVID analysis, GATA eral novel areas for investigation in the field. and PAX4 binding sites were overrepresented in the pro- moter region of palmitate-modified genes and transcripts with modified splicing (Fig. 4). These data suggest that Acknowledgments. The authors thank I. Millard, A. Musuaya, S. Mertens, palmitate modulates gene networks contributing to lip- and M. Pangerl, Laboratory of Experimental Medicine, ULB Center for Diabe- otoxic loss of b-cell function and survival. However, the tes Research, Université Libre de Bruxelles, for expert technical assistance; data suggest that the b-cell functional impairment is not Dr. J. Ragoussis, University of Oxford, for help and advice in the initial stages of related to transcriptional inhibition of distal steps of in- RNAseq; and Dr. S. Montgomery, University of Geneva Medical School, for dis- sulin secretion. cussions on the RNA-seq data analysis. A key cellular stress response activated by palmitate in Funding. This work was supported by the European Union (project BetaBat in Seventh Framework Programme), the Fonds National de la Recherche human islets is the UPR (present data). ER stress has been Scientifique (FNRS), JDRF (JDRF 37-2012-5 and 17-2012-114), and Actions b shown to contribute to lipotoxic -cell death (13,16). de Recherche Concertées de la Communauté Française, Belgium. B.A. is a fellow What governs the transition from adaptive to fatal ER of the Fonds pour la Formation à la Recherche dans l’Industrie et dans l’Agri- fi stress is probably cell speci c and remains ill understood. culture-FNRS, and D.A.C. is a FNRS postdoctoral fellow. Saturated FFAs elicit marked PERK activity and the result- Duality of Interest. No potential conflicts of interest relevant to this article ing eIF2a phosphorylation contributes to b-cell death (13), were reported. whereas ATF6 and IRE1 mediate protective UPR signaling Author Contributions. M.C., M.I.M., P.M., and D.L.E. designed the (69). A number of structural homologs of ATF6 may play experiments. M.C., B.A., G.B., D.A.C., M.I.-E., M.M., J.-V.T., T.G., O.V., I.S., tissue-specific roles. Here we found that CREB3 and M.B., L.L., L.M., and M.S. performed experiments and analyzed data. M.C. and CREB3L3 are upregulated by palmitate. CREB3 may pro- D.L.E. wrote the manuscript. B.A., G.B., D.A.C., M.I.-E., M.M., J.-V.T., T.G., O.V., mote protective UPR signaling, given that CREB3 silencing I.S., L.M., and M.S. reviewed and edited the manuscript. M.C. is the guarantor of this work and, as such, had full access to all the data in the study and markedly sensitized b-cells to lipotoxicity. A better under- takes responsibility for the integrity of the data and the accuracy of the data standing of the (mal)adaptive facets of the ER stress re- analysis. sponse in b-cells is important in light of the evidence for UPR markers in b-cells from T2D patients (16,17,30) and References that T2D drugs modulate the UPR (69). 1. Cnop M, Hughes SJ, Igoillo-Esteve M, et al. The long lifespan and low The present RNA-seq analysis indicates that palmitate turnover of human islet beta cells estimated by mathematical modelling of lip- inhibits several protein degradation mechanisms, includ- ofuscin accumulation. Diabetologia 2010;53:321–330 ing autophagy and lysosomal function. In line with the 2. Cnop M, Foufelle F, Velloso LA. Endoplasmic reticulum stress, obesity and present findings, palmitate has been shown to impair diabetes. Trends Mol Med 2012;18:59–68 1992 RNA-Seq of Palmitate-Treated Human Islets Diabetes Volume 63, June 2014

3. Cnop M. Fatty acids and glucolipotoxicity in the pathogenesis of Type 2 23. Malmgren S, Spégel P, Danielsson AP, et al. Coordinate changes in histone diabetes. Biochem Soc Trans 2008;36:348–352 modifications, mRNA levels, and metabolite profiles in clonal INS-1 832/13 4. Gargani S, Thévenet J, Yuan JE, et al. Adaptive changes of human islets to b-cells accompany functional adaptations to lipotoxicity. J Biol Chem 2013;288: an obesogenic environment in the mouse. Diabetologia 2013;56:350–358 11973–11987 5. Winzell MS, Magnusson C, Ahrén B. Temporal and dietary fat content- 24. Bikopoulos G, da Silva Pimenta A, Lee SC, et al. Ex vivo transcriptional dependent islet adaptation to high-fat feeding-induced glucose intolerance in profiling of human pancreatic islets following chronic exposure to mono- mice. Metabolism 2007;56:122–128 unsaturated fatty acids. J Endocrinol 2008;196:455–464 6. Kashyap S, Belfort R, Gastaldelli A, et al. A sustained increase in plasma 25. Marioni JC, Mason CE, Mane SM, Stephens M, Gilad Y. RNA-seq: an as- free fatty acids impairs insulin secretion in nondiabetic subjects genetically sessment of technical reproducibility and comparison with gene expression ar- predisposed to develop type 2 diabetes. Diabetes 2003;52:2461–2474 rays. Genome Res 2008;18:1509–1517 7. Sako Y, Grill VE. A 48-hour lipid infusion in the rat time-dependently inhibits 26. Eizirik DL, Sammeth M, Bouckenooghe T, et al. The human pancreatic glucose-induced insulin secretion and B cell oxidation through a process likely islet transcriptome: expression of candidate genes for type 1 diabetes and the coupled to fatty acid oxidation. Endocrinology 1990;127:1580–1589 impact of pro-inflammatory cytokines. PLoS Genet 2012;8:e1002552 8. Cnop M, Hannaert JC, Hoorens A, Eizirik DL, Pipeleers DG. Inverse re- 27. Nica AC, Ongen H, Irminger JC, et al. Cell-type, allelic, and genetic sig- lationship between cytotoxicity of free fatty acids in pancreatic islet cells and natures in the human pancreatic b cell transcriptome. Genome Res 2013;23: cellular triglyceride accumulation. Diabetes 2001;50:1771–1777 1554–1562 9. Kelpe CL, Moore PC, Parazzoli SD, Wicksteed B, Rhodes CJ, Poitout V. 28. Morris AP, Voight BF, Teslovich TM, et al.; Wellcome Trust Case Control Palmitate inhibition of insulin gene expression is mediated at the transcriptional Consortium; Meta-Analyses of Glucose and Insulin-related traits Consortium level via ceramide synthesis. J Biol Chem 2003;278:30015–30021 (MAGIC) Investigators; Genetic Investigation of ANthropometric Traits (GIANT) 10. Carlsson C, Borg LA, Welsh N. Sodium palmitate induces partial mito- Consortium; Asian Genetic Epidemiology Network–Type 2 Diabetes (AGEN-T2D) chondrial uncoupling and reactive oxygen species in rat pancreatic islets in vitro. Consortium; South Asian Type 2 Diabetes (SAT2D) Consortium; DIAbetes Genetics Endocrinology 1999;140:3422–3428 Replication And Meta-analysis (DIAGRAM) Consortium. Large-scale association 11. Hoppa MB, Collins S, Ramracheya R, et al. Chronic palmitate exposure analysis provides insights into the genetic architecture and pathophysiology of type inhibits insulin secretion by dissociation of Ca(2+) channels from secretory 2 diabetes. Nat Genet 2012;44:981–990 granules. Cell Metab 2009;10:455–465 29. Volkmar M, Dedeurwaerder S, Cunha DA, et al. DNA methylation profiling 12. Shimabukuro M, Higa M, Zhou YT, Wang MY, Newgard CB, Unger RH. identifies epigenetic dysregulation in pancreatic islets from type 2 diabetic pa- Lipoapoptosis in b-cells of obese prediabetic fa/fa rats. Role of serine palmi- tients. EMBO J 2012;31:1405–1426 toyltransferase overexpression. J Biol Chem 1998;273:32487–32490 30. Marchetti P, Bugliani M, Lupi R, et al. The endoplasmic reticulum in 13. Cunha DA, Hekerman P, Ladrière L, et al. Initiation and execution of lipotoxic pancreatic beta cells of type 2 diabetes patients. Diabetologia 2007;50:2486– ER stress in pancreatic b-cells. J Cell Sci 2008;121:2308–2318 2494 14. Kharroubi I, Ladrière L, Cardozo AK, Dogusan Z, Cnop M, Eizirik DL. Free 31. Igoillo-Esteve M, Marselli L, Cunha DA, et al. Palmitate induces a pro- fatty acids and cytokines induce pancreatic b-cell apoptosis by different inflammatory response in human pancreatic islets that mimics CCL2 expres- mechanisms: role of nuclear factor-kB and endoplasmic reticulum stress. En- sion by beta cells in type 2 diabetes. Diabetologia 2010;53:1395–1405 docrinology 2004;145:5087–5096 32. Ling Z, Pipeleers DG. Prolonged exposure of human b cells to elevated 15. Karaskov E, Scott C, Zhang L, Teodoro T, Ravazzola M, Volchuk A. Chronic glucose levels results in sustained cellular activation leading to a loss of glucose palmitate but not oleate exposure induces endoplasmic reticulum stress, which regulation. J Clin Invest 1996;98:2805–2812 may contribute to INS-1 pancreatic b-cell apoptosis. Endocrinology 2006;147: 33. Asfari M, Janjic D, Meda P, Li G, Halban PA, Wollheim CB. Establishment of 3398–3407 2-mercaptoethanol-dependent differentiated insulin-secreting cell lines. Endo- 16. Laybutt DR, Preston AM, Akerfeldt MC, et al. Endoplasmic reticulum stress crinology 1992;130:167–178 contributes to beta cell apoptosis in type 2 diabetes. Diabetologia 2007;50:752– 34. Cardozo AK, Ortis F, Storling J, et al. Cytokines downregulate the 763 sarcoendoplasmic reticulum pump Ca2+ ATPase 2b and deplete endoplasmic 17. Eizirik DL, Cardozo AK, Cnop M. The role for endoplasmic reticulum stress reticulum Ca2+, leading to induction of endoplasmic reticulum stress in pan- in diabetes mellitus. Endocr Rev 2008;29:42–61 creatic b-cells. Diabetes 2005;54:452–461 18. Cnop M, Ladrière L, Hekerman P, et al. Selective inhibition of eukaryotic 35. Pruitt KD, Tatusova T, Klimke W, Maglott DR. NCBI Reference Sequences: translation initiation factor 2 a dephosphorylation potentiates fatty acid-induced current status, policy and new initiatives. Nucleic Acids Res 2009;37(Database endoplasmic reticulum stress and causes pancreatic b-cell dysfunction and issue):D32–D36 apoptosis. J Biol Chem 2007;282:3989–3997 36. Montgomery SB, Sammeth M, Gutierrez-Arcelus M, et al. Transcriptome 19. Cunha DA, Igoillo-Esteve M, Gurzov EN, et al. Death protein 5 and p53- genetics using second generation sequencing in a Caucasian population. Nature upregulated modulator of apoptosis mediate the endoplasmic reticulum stress- 2010;464:773–777 mitochondrial dialog triggering lipotoxic and human b-cell apoptosis. 37. Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. Mapping and Diabetes 2012;61:2763–2775 quantifying mammalian transcriptomes by RNA-Seq. Nat Methods 2008;5:621– 20. Busch AK, Cordery D, Denyer GS, Biden TJ. Expression profiling of 628 palmitate- and oleate-regulated genes provides novel insights into the effects 38. Harrow J, Frankish A, Gonzalez JM, et al. GENCODE: the reference human of chronic lipid exposure on pancreatic b-cell function. Diabetes 2002;51: genome annotation for The ENCODE Project. Genome Res 2012;22:1760–1774 977–987 39. Bernstein BE, Birney E, Dunham I, Green ED, Gunter C, Snyder M; ENCODE 21. Busch AK, Gurisik E, Cordery DV, et al. Increased fatty acid desaturation and Project Consortium. An integrated encyclopedia of DNA elements in the human enhanced expression of stearoyl coenzyme A desaturase protects pancreatic genome. Nature 2012;489:57–74 b-cells from lipoapoptosis. Diabetes 2005;54:2917–2924 40. Djebali S, Davis CA, Merkel A, et al. Landscape of transcription in human 22. Xiao J, Gregersen S, Kruhøffer M, Pedersen SB, Ørntoft TF, Hermansen K. cells. Nature 2012;489:101–108 The effect of chronic exposure to fatty acids on gene expression in clonal insulin- 41. Kharroubi I, Rasschaert J, Eizirik DL, Cnop M. Expression of adiponectin producing cells: studies using high density oligonucleotide microarray. Endocri- receptors in pancreatic b cells. Biochem Biophys Res Commun 2003;312:1118– nology 2001;142:4777–4784 1122 diabetes.diabetesjournals.org Cnop and Associates 1993

42. Cardozo AK, Proost P, Gysemans C, Chen MC, Mathieu C, Eizirik DL. IL-1b 57. Lee HJ, Chung K, Lee H, Lee K, Lim JH, Song J. Downregulation of and IFN-g induce the expression of diverse chemokines and IL-15 in human and mitochondrial lon protease impairs mitochondrial function and causes hepatic rat pancreatic islet cells, and in islets from pre-diabetic NOD mice. Diabetologia insulin resistance in human liver SK-HEP-1 cells. Diabetologia 2011;54:1437– 2003;46:255–266 1446 43. Moore F, Cunha DA, Mulder H, Eizirik DL. Use of RNA interference to in- 58. Rorsman P, Braun M. Regulation of insulin secretion in human pancreatic vestigate cytokine signal transduction in pancreatic b cells. Methods Mol Biol islets. Annu Rev Physiol 2013;75:155–179 2012;820:179–194 59. Gray E, Muller D, Squires PE, et al. Activation of the extracellular calcium- 44. Cnop M, Igoillo-Esteve M, Rai M, et al. Central role and mechanisms of sensing receptor initiates insulin secretion from human islets of Langerhans: b-cell dysfunction and death in friedreich ataxia-associated diabetes. Ann Neurol involvement of protein kinases. J Endocrinol 2006;190:703–710 2012;72:971–982 60. Macdonald MJ, Brown LJ, Longacre MJ, Stoker SW, Kendrick MA. 45. Taneera J, Lang S, Sharma A, et al. A systems genetics approach identifies Knockdown of both mitochondrial isocitrate dehydrogenase enzymes in pan- genes and pathways for type 2 diabetes in human islets. Cell Metab 2012;16: creatic b cells inhibits insulin secretion. Biochim Biophys Acta 2013;1830:5104– 122–134 5111 46. Anderson CA, Boucher G, Lees CW, et al. Meta-analysis identifies 29 ad- 61. Prentki M, Matschinsky FM, Madiraju SR. Metabolic signaling in fuel- ditional ulcerative colitis risk loci, increasing the number of confirmed associa- induced insulin secretion. Cell Metab 2013;18:162–185 tions to 47. Nat Genet 2011;43:246–252 62. Lango Allen H, Flanagan SE, Shaw-Smith C, et al.; International Pancreatic 47. Aulchenko YS, Struchalin MV, Belonogova NM, et al. Predicting human Agenesis Consortium. GATA6 haploinsufficiency causes pancreatic agenesis in height by Victorian and genomic methods. Eur J Hum Genet 2009;17:1070–1075 humans. Nat Genet 2012;44:20–22 48. DenBoer LM, Hardy-Smith PW, Hogan MR, Cockram GP, Audas TE, Lu R. 63. Chen J, Fontes G, Saxena G, Poitout V, Shalev A. Lack of TXNIP Luman is capable of binding and activating transcription from the unfolded protein protects against mitochondria-mediated apoptosis but not against fatty acid- response element. Biochem Biophys Res Commun 2005;331:113–119 induced ER stress-mediated b-cell death. Diabetes 2010;59:440–447 49. Zhang K, Shen X, Wu J, et al. Endoplasmic reticulum stress activates 64. Oslowski CM, Hara T, O’Sullivan-Murphy B, et al. Thioredoxin-interacting cleavage of CREBH to induce a systemic inflammatory response. Cell 2006;124: protein mediates ER stress-induced b cell death through initiation of the in- 587–599 flammasome. Cell Metab 2012;16:265–273 50. Ebato C, Uchida T, Arakawa M, et al. Autophagy is important in islet ho- 65. Morán I, Akerman I, van de Bunt M, et al. Human b cell transcriptome meostasis and compensatory increase of b cell mass in response to high-fat diet. analysis uncovers lncRNAs that are tissue-specific, dynamically regulated, and Cell Metab 2008;8:325–332 abnormally expressed in type 2 diabetes. Cell Metab 2012;16:435–448 51. Masini M, Bugliani M, Lupi R, et al. Autophagy in human type 2 diabetes 66. Tang F, Barbacioru C, Wang Y, et al. mRNA-Seq whole-transcriptome pancreatic beta cells. Diabetologia 2009;52:1083–1086 analysis of a single cell. Nat Methods 2009;6:377–382 52. Las G, Serada SB, Wikstrom JD, Twig G, Shirihai OS. Fatty acids 67. Sultan M, Schulz MH, Richard H, et al. A global view of gene activity and suppress autophagic turnover in b-cells. J Biol Chem 2011;286:42534–42544 alternative splicing by deep sequencing of the human transcriptome. Science 53. Hidvegi T, Ewing M, Hale P, et al. An autophagy-enhancing drug promotes 2008;321:956–960 degradation of mutant a1-antitrypsin Z and reduces hepatic fibrosis. Science 68. van Delft J, Gaj S, Lienhard M, et al. RNA-Seq provides new insights in the 2010;329:229–232 transcriptome responses induced by the carcinogen benzo[a]pyrene. Toxicol Sci 54. Kim JS, Wang JH, Biel TG, et al. Carbamazepine suppresses -me- 2012;130:427–439 diated autophagy impairment after ischemia/reperfusion in mouse livers. Toxicol 69. Cunha DA, Ladrière L, Ortis F, et al. Glucagon-like peptide-1 agonists protect Appl Pharmacol 2013;273:600–610 pancreatic b-cells from lipotoxic endoplasmic reticulum stress through upreg- 55. Molina AJ, Wikstrom JD, Stiles L, et al. Mitochondrial networking protects ulation of BiP and JunB. Diabetes 2009; b-cells from nutrient-induced apoptosis. Diabetes 2009;58:2303–2315 58:2851–2862 56. Venkatesh S, Lee J, Singh K, Lee I, Suzuki CK. Multitasking in the mito- 70. Chen PC, Olson EM, Zhou Q, et al. Carbamazepine as a novel small mol- chondrion by the ATP-dependent Lon protease. Biochim Biophys Acta 2012; ecule corrector of trafficking-impaired ATP-sensitive potassium channels iden- 1823:5666 tified in congenital hyperinsulinism. J Biol Chem 2013;288:20942–20954

PAPER II Abdulkarim B, Nicolino M, Igoillo-Esteve M, Daures M, Romero S, Philippi A, Senée V, Lopes M, Cunha DA, Harding HP, Derbois C, Bendelac N, Hattersley AT, Eizirik DL, Ron D, Cnop M, Julier C; A Missense Mutation in PPP1R15B Causes a Syndrome Including Diabetes, Short Stature, and Microcephaly.

Background: A well balanced endoplasmic reticulum (ER) function is crucial for insulin release and survival of pancreatic β-cells. Deficient as well as excessive/prolonged ER stress can cause β-cell dysfunction/death and diabetes. Phosphorylation of the α subunit of eukaryotic translation initiation factor 2 (eIF2α) is key for protein folding homeostasis in the ER. CReP is a cofactor for protein phosphatase 1 (PP1), which is required for eIF2α de-phosphorylation. Here we report on the first mutation in PPP1R15B, encoding CReP, in two siblings with young-onset diabetes. Methods: Exome sequencing was performed in the index case. PP1 binding was examined in HEK293T cells expressing wild type or mutant CReP. CReP, PUMA, DP5 and Bim were silenced by RNA interference inINS-1E and FACS-purified primary rat β cells. Insulin secretion was measured by ELISA. eIF2α phosphorylation was examined by Western blot and gene expression by qPCR. Apoptosis was assessed by Hoechst 33342/propidium iodide staining and by Western blot for caspases and cytochrome c release. Results: Two siblings, born to unaffected parents, had early-onset insulin treated diabetes (autoantibody negative, with low but detectable C-peptide levels) diagnosed at 15 and 28 years. This was associated with short stature and microcephaly. Exome sequencing identified a homozygous R658C mutation in PPP1R15B in the index case, which was confirmed by Sanger sequencing in the two siblings. Introducing the R658C mutation in a CReP plasmid diminished the ability of CReP to bind PP1 in vitro and de-phosphorylate eIF2α. Our results suggested that CReP silencing could result in an increase in eIF2α phosphorylation. CReP-deficient β cells had reduced insulin content and reduced glucose-induced insulin secretion. Furthermore, our data indicated that CReP silencing induced apoptosis in INS-1E cells (14±3% apoptosis with siCReP vs 4±0% with control siCT, p<0.05, n=4-5) and primary β-cells (siCReP 22±3% vs siCT 15±2%, p<0.05, n=5). CReP silencing induced caspase-9 cleavage and cytochrome c release from the mitochondria, indicating activation of the intrinsic pathway of apoptosis. CReP deficiency induced a 2-3-fold increase in mRNA expression of the pro-apoptotic BH3-onlyproteins PUMA and DP5 (p<0.05, n=6). Silencing of PUMA, DP5 and Bim partially protected CReP-deficient β cells against apoptosis (11-13% apoptosis vs 16%, p<0.05, n=3-4). Conclusion: We have identified a homozygous missense mutation in PPP1R15B leading to a syndrome that includes young-onset diabetes and microcephaly. The mutation appeared to diminish the ability of CReP to bind PP1 and dephosphorylate eIF2α, leading to reduced β cell function and increased β cell apoptosis.

48

My contribution: I have contributed to ~70% of the work carried out, in particular I have:

 Designed all primers, siRNAs and plasmids used in the study  Performed: - Cell culture, transfection and treatment assays - Apoptosis analysis - In vitro de-phosphorylation assays - mRNA extraction and reverse transcription - western blot - Glucose stimulated insulin secretion  Analyzed the data

Prepared the manuscript

49

Diabetes Volume 64, November 2015 3951

Baroj Abdulkarim,1 Marc Nicolino,2,3,4 Mariana Igoillo-Esteve,1 Mathilde Daures,5,6 Sophie Romero,5,6 Anne Philippi,5,6 Valérie Senée,5,6 Miguel Lopes,1 Daniel A. Cunha,1 Heather P. Harding,7 Céline Derbois,8 Nathalie Bendelac,2 Andrew T. Hattersley,9 Décio L. Eizirik,1 David Ron,7 Miriam Cnop,1,10 and Cécile Julier5,6

A Missense Mutation in PPP1R15B Causes a Syndrome Including Diabetes, Short Stature, and Microcephaly

Diabetes 2015;64:3951–3962 | DOI: 10.2337/db15-0477 GENETICS/GENOMES/PROTEOMICS/METABOLOMICS

Dysregulated endoplasmic reticulum stress and phos- endoplasmic reticulum (ER) stress and aspects of the re- phorylation of eukaryotic translation initiation factor 2a sponse to it contribute to b-cell failure both in type 1 and (eIF2a) are associated with pancreatic b-cell failure and type 2 diabetes (1–6). ER stress is defined as an imbalance diabetes. Here, we report the first homozygous mutation between unfolded protein load in the ER and the organelle’s in the PPP1R15B gene (also known as constitutive re- functional capacity. This activates an ER stress response, a pressor of eIF2 phosphorylation [CReP]) encoding the also known as the unfolded protein response (UPR), which a fi regulatory subunit of an eIF2 -speci c phosphatase in reduces protein load and increases ER folding capacity. The two siblings affected by a novel syndrome of diabetes UPR is an adaptive response, particularly important for the of youth with short stature, intellectual disability, and mi- survival of cells with a high secretory capacity such as pan- crocephaly. The R658C mutation in PPP1R15B affects creatic b-cells. The UPR is triggered by the activation of a conserved amino acid within the domain important for ER transmembrane proteins that sense the misfolded pro- protein phosphatase 1 (PP1) binding. The R658C mutation tein accumulation in the ER lumen and transduce the signal decreases PP1 binding and eIF2a dephosphorylation and results in b-cell apoptosis. Our findings support the con- to the cytoplasm. One of the canonical ER stress trans- cept that dysregulated eIF2a phosphorylation, whether ducers is protein kinase R-like endoplasmic reticulum ki- EIF2AK3 decreased by mutation of the kinase (EIF2AK3) in Wolcott- nase (PERK, encoded by ). PERK phosphorylates Rallison syndrome or increased by mutation of the phos- the eukaryotic translation initiation factor 2a (eIF2a)and phatase (PPP1R15B), is deleterious to b-cells and other attenuates protein translation to lessen the burden of the secretory tissues, resulting in diabetes associated with stressed ER (1,7). The effector pathway downstream of multisystem abnormalities. PERK is tightly regulated by eIF2a dephosphorylation car- ried out by a holophosphatase complex consisting of a com- mon catalytic subunit, protein phosphatase 1 (PP1), and The molecular mechanisms contributing to pancreatic a substrate-specific regulatory subunit, PPP1R15. Two b-cell dysfunction and apoptosis in diabetes remain such regulatory subunits are known: the constitutive repres- poorly understood. Accumulating evidence suggests that sor of eIF2a phosphorylation (CReP), encoded by PPP1R15B,

1ULB Center for Diabetes Research, Université Libre de Bruxelles, Brussels, 9University of Exeter Medical School, University of Exeter, Exeter, U.K. Belgium 10Division of Endocrinology, Erasmus Hospital, Brussels, Belgium 2 Hôpital Femme-Mère-Enfant, Division of Pediatric Endocrinology, Hospices Civils Corresponding authors: Marc Nicolino, [email protected]; Miriam Cnop, de Lyon, Lyon 1 University, Lyon, France [email protected]; and Cécile Julier, [email protected]. 3INSERM U870, Lyon, France Received 8 April 2015 and accepted 30 June 2015. 4INSERM CIC201, Lyon, France 5INSERM UMR-S958, Faculté de Médecine Paris Diderot, Paris, France This article contains Supplementary Data online at http://diabetes 6Université Paris Diderot, Sorbonne Paris Cité, Paris, France .diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1. 7Cambridge Institute for Medical Research, University of Cambridge, and Na- B.A., M.N., M.C., and C.J. contributed equally to this work. tional Institute for Health Research Cambridge Biomedical Research Centre, © 2015 by the American Diabetes Association. Readers may use this article as Cambridge, U.K. long as the work is properly cited, the use is educational and not for profit, and 8Institut de Génomique, Centre National de Génotypage, Commissariat à l’Energie the work is not altered. Atomique et aux Energies Alternatives, Evry, France 3952 PPP1R15B Mutation in a Novel Diabetes Syndrome Diabetes Volume 64, November 2015 acts under basal conditions (8), while GADD34, encoded by exome was 88% and 79% for a 103 and 253 depth of PPP1R15A, is activated by eIF2a phosphorylation and feeds coverage, respectively, resulting in a mean sequencing back negatively on PERK signaling to promote the recovery depth of 643 per base. Exome variant analysis was then of protein synthesis as the stress response wanes (9). The performed using an in-house python pipeline on genetic PERK pathway and its regulation are especially important for variation annotation results (M.D., A.P., and C.J., un- pancreatic b-cell function and survival (10–12). Homozygous published data). Variants were filtered consecutively mutations in EIF2AK3 cause Wolcott-Rallison syndrome, based on their quality, their genotype (homozygous status), a syndromic form of neonatal diabetes with epiphyseal and the predicted consequence on coding capacity (missense, dysplasia, growth retardation, and variable other manifes- nonsense, splice-site, and coding insertion/deletion— tations including microcephaly (11,13)—features mirrored inframe or frameshift) and for their rare status based in Perk knockout mice (14). Unmitigated ER stress contrib- on information available in in-house (control subjects, utes to b-cell demise in several other monogenic forms of IntegraGen) and public databases (Exome Variant Server diabetes. Mutations in the insulin gene that disrupt pro- [EVS], ESP6500SI; Exome Aggregation Consortium [ExAC], insulin folding cause severe b-cell ER stress and neonatal release 0.3; and Single Nucleotide Polymorphism database diabetes (15). The loss of the ER cochaperone p58IPK, due to [dbSNP], version 138). Variants that were homozygous or DNAJC3 mutations, leads to a syndrome with young-onset had a minor allele frequency .0.005 in any in-house or diabetes (16). public database were excluded. PPP1R15B Here, we report on a mutation in two sib- Variant Confirmation and Test for Diabetes lings with young-onset diabetes, microcephaly, and short Segregation in the Family stature. The mutation, located in the PP1 binding domain Each rare variant identified as homozygous in patient 1 (8), disrupts PP1 binding and eIF2a dephosphorylation by exome sequencing was confirmed in patient 1 and and reveals that b-cell dysfunction and apoptosis may further genotyped in patient 2, in two nonaffected relatives be caused both by too little and by too much eIF2a phos- (a paternal grandmother and a paternal aunt), and in an phorylation. unrelated healthy control subject. This was done by Sanger sequencing or by PCR-restriction fragment length poly- RESEARCH DESIGN AND METHODS morphism (RFLP) genotyping using specificamplification Patients primers and restriction enzymes that differentiate the We studied a consanguineous family of Algerian origin, two alleles, followed by agarose gel electrophoresis using with two siblings affected by young-onset diabetes that is standard techniques. Sequencing primers and PCR-RFLP associated with short stature, microcephaly, and intellectual primers/enzymes are available on request. disability. The study was explained to the patients, their parents, tutors, and other family members, who agreed Sanger Sequencing of PPP1R15B and to participate in the genetic study and signed informed Regulatory Regions consents. The study protocol was approved by the local Sanger sequencing of PPP1R15B exons (coding, 59UTR and ethics committee. Blood samples were obtained from the 39UTR regions) and flanking regions of a 680 of two affected siblings and two nonaffected relatives (the promoter region and a 800 base pair intronic region that paternal grandmother and a paternal aunt) and DNA was shows species conservation and contains unspliced human extracted using standard procedures. ESTs (UCSC Genome Browser) was performed by BigDye Terminator sequencing on PCR-amplified DNA using an Exome Sequencing and Analysis Applied Biosystems 3730 DNA Sequencer (Foster City, Exome sequencing was performed on the genomic plat- CA). PCR and sequencing primers are shown in Supplemen- form of IntegraGen (Evry, France). Exons of genomic DNA tary Table 1. Sequence interpretation was performed using of the index case (patient 1) were captured with in- the Genalys software (17). solution enrichment methodology (SureSelect Human All Exon Kits, version 2; Agilent Technologies, Santa Clara, CA) Cell Culture with the company’s biotinylated oligonucleotide probe li- Clonal rat INS-1E cells (a kind gift from Dr. C. Wollheim, brary (Human All Exon, version 2, 50 Mb; Agilent). Geno- Centre Médical Universitaire, Geneva, Switzerland) were mic DNA was then sequenced on a sequencer as paired-end cultured in RPMI medium as described (18). Male Wistar 75 bases (Illumina HiSeq 2000; Illumina, San Diego, CA). rats (Charles River Laboratories, Châtillon-sur-Chalaronne, Image analysis and base calling were performed with Real- France) were housed and handled following the rules of the Time Analysis software, version 1.14 with default parame- Belgian Regulations for Animal Care. Rat tissues were col- ters (Illumina). Bioinformaticanalysiswasperformedby lected and islets were handpicked under a stereomicroscope an in-house pipeline (IntegraGen) based on the Consen- after isolation by collagenase digestion (19). b-Cells were sus Assessment of Sequence and Variation (CASAVA 1.8; purified by autofluorescence-activated cell sorting of Illumina) to perform alignment against human reference dispersed islet cells (FACS, FACSAria; BD Biosciences, genome (GRCh37/hg19), variant calling, and coverage Erembodegem, Belgium) and cultured as previously de- analysis. The overall sequencing coverage over the whole scribed (20). Human embryonic kidney (HEK)293T cells diabetes.diabetesjournals.org Abdulkarim and Associates 3953 were maintained in DMEM supplemented with 10% FBS, RNA Interference 100 mU/mL penicillin, 100 mU/mL streptomycin, and Clonal and primary rat b-cells were transfected overnight with 2 mmol/L L-glutamine. 30 nmol/L control small interfering RNA (siRNA) (Qiagen, Germantown, MD) or siRNAs targeting rat PPP1R15B, DP5, Generation of the R658C Mutant PPP1R15B PUMA, or Bim using Lipofectamine RNAiMAX (Invitrogen) Expression Plasmid as previously described (22). The siRNAs used in the cur- The R658C mutation was introduced in the rent study are listed in Supplementary Table 2. PPP1R15BpEGFP_C1 plasmid (21) using QuikChange II Total RNA and mRNA Extraction and Real-Time PCR Site-Directed Mutagenesis Kit (Agilent) and the primers + huPPP1R15B_R658C_1S: GTGGTGATGAGGATTGCAA Poly(A) mRNA and total RNA were isolated and reverse AGGACCATGG and huPPP1R15B_R658C_2AS: CCATG transcribed as previously described (19,23). Real-time PCR GTCCTTTGCAATCCTCATCACCAC (bold letters indicate was performed using Rotor-Gene SYBR Green on a Rotor- the mutation site). This vector allows the expression of Gene Q cycler (Qiagen) (23,24). Primers were used in a con- recombinant proteins fused to GFP at its N-terminus. Af- ventional PCR for preparing thestandards.Geneexpression ter mutagenesis, positive clones were sequenced (Seqlab, wascalculatedascopies/mL (25). Expression levels were Göttingen, Germany), and the following primers were used corrected for expression of the reference gene GAPDH. to cover the entire gene: F1: CATGGTCCTGCTGGAGTTCGTG, Primer sequences are provided in Supplementary Table 3. F2: AGAGGAGGGGATCCACTG, F3: ACAGTGATGGAAA Western Blotting TAGCGAG, F4: ATCTAGTGAGATACCTATGG, F5: TGA Tissue and cell preparation were performed as previously GACCCCTGAGCATAG, Rev: CACACCTCCCCCTGAAC. Plas- described (26). Immunoblotting was done using antibod- mids containing the mutation but no other change in the ies against b-actin, phosphorylated eIF2a, eIF2a, BCL-2, PPP1R15B gene were introduced into one-shot TOP10 BCL-XL, caspase 9, caspase 3, Cox IV (Cell Signaling, Lei- Electrocomp Escherichia coli (Invitrogen, Gent, Belgium) den, the Netherlands), human a-tubulin (Sigma-Aldrich), by electroporation. The cells were recovered for 1 h in human ATF3, human PP1 (Santa Cruz, Heidelberg, Ger- SOC medium, plated on Luria broth agar containing 50 many), cytochrome c (BD Biosciences), or GFP produced mg/mL kanamycin, and incubated overnight at 37°C. Se- in rabbits. Protein detection was performed using DyLight lected colonies were grown overnight at 37°C on Luria conjugated secondary antibody or horseradish peroxidase– broth containing 50 mg/mL kanamycin. Plasmids were conjugated secondary antibodies and SuperSignal West Femto purified using the PureYield Plasmid Midiprep Kit (Prom- chemiluminescence revealing reagent (Thermo Scientific). ega, Leiden, the Netherlands) according to the manu- Immunoreactive bands were detected with a ChemiDoc facturer’s instructions. The DNA concentration was XRS+ system and with Image Lab software (Bio-Rad, measured using NanoDrop 3300 (Thermo Scientific, Hercules, CA). Protein levels were corrected for a-tubulin Gent, Belgium). or b-actin. PPP1R15B Overexpression and Immunoprecipitation Cell Treatment and Apoptosis Assays HEK293T cells were transfected with EGFP expression Free fatty acid (FFA) exposure was done in RPMI 1640 plasmid (pEGFP) without insert (empty vector) or medium containing 0.75% FFA-free BSA (Roche, Mannheim, expressing wild type (WT) or mutant PPP1R15B, alone Germany) and 1% FBS. Oleate and palmitate (Sigma- or combined with a mouse PP1A expression plasmid Aldrich) were dissolved in 90% ethanol and diluted 1:100 to using PEI reagent. After 24 h, the cells were lysed as a final concentration of 0.5 mmol/L (27,28). Cyclopiazonic previously described (8). acid (CPA) was used at 25 mmol/L, tunicamycin at 5 mg/mL, Using 1 mL anti-GFP antibody bound to 15 mLprotein and brefeldin A at 0.1 mg/mL. The PERK inhibitor GSK2606414 A sepharose resin, GFP-PPP1R15B was purified from was used at 0.5 mmol/L (29). Apoptotic cell death was detected equal amounts of cell lysate protein. The immunoprecipi- and counted by fluorescence microscopy after Hoechst tates containing PPP1R15B were washed twice in lysis 33342 (5 mg/mL; Sigma-Aldrich) and propidium iodide buffer before being resolved on a 10% SDS-PAGE gel and (5 mg/mL) staining (28,30). blotted onto PVDF (polyvinylidene fluoride) membranes Glucose-Stimulated Insulin Secretion or used for dephosphorylation studies. Insulin secretion was measured as previously described (22). Dephosphorylation Assay Briefly, INS-1E cells were washed with modified Krebs- PPP1R15B immunoprecipitates were resuspended in Ringer bicarbonate HEPES solution (KRB) and incubated dephosphorylation buffer containing 50 mmol/L Tris for 30 min with KRB without glucose, and insulin secretion (pH 7), 100 mmol/L NaCl, 0.1 mmol/L EDTA, 0.1% Triton was induced by 30-min incubation with KRB containing X-100,1mmol/LDTT,and1mmol/LMnCl2 and incubated 1.67 or 16.7 mmol/L glucose with or without 10 mmol/L with phosphorylated eIF2a for 5–60 min at 30°C. The forskolin. Insulin was measured by ELISA (Mercodia, supernatant was resolved on 15% phos-tagged SDS-PAGE Uppsala, Sweden) in cell-free supernatants and acid-ethanol gel and visualized using EZBlue gel staining reagent (Sigma- extracted cell lysates. Total protein was measured in cell Aldrich, Dorset, England). lysates using the Protein Assay Dye Reagent (Bio-Rad). 3954 PPP1R15B Mutation in a Novel Diabetes Syndrome Diabetes Volume 64, November 2015

Cytochrome C Release insulin (;0.5 units/kg/day) but evolved to significant glu- Cells were harvested in PBS 48 h after transfection. After cose variability with severe hypoglycemia episodes and seiz- centrifugation, a cytosolic lysis buffer containing 0.8 mg/mL ures. He had growth retardation (Fig. 1A), reaching an adult digitonin was added to the pellet. The cells were vortexed for height of 155 cm, with normal growth hormone (peak at 30 s and centrifuged at 4°C at 20,0003g for 1 min. The 41 ng/mL after ornithine stimulation test; normal reference supernatant was separated as cytoplasmic fraction and the values [N] .10) and IGF1 levels (291 mg/L; N = 249–672). pellet was used as the mitochondrial fraction. Laemmli buffer Thyroxine (18 pmol/L; N = 10–23) and glucagon (116 ng/L; was added and the samples were resolved on 12% SDS-PAGE. N=25–250) levels were normal. Blood cell count, electro- lytes, creatinine, liver enzymes, bilirubin, cholesterol, trigly- Statistical Analysis cerides, lactate, and pyruvate levels were normal. He had 6 Data are presented as means SE. Given the paired nature of delayed puberty, with an undescended right testis that was the experimental design, comparisons between groups were surgically corrected. No anomaly of the gonadal function was t made by two-sided Student paired test, with Bonferroni found, with normal levels of testosterone (9.54 nmol/L; P correction for multiple comparisons when needed. A N=9.0–26.0), luteinizing hormone (4.4 mU/mL; N = , fi value 0.05 was considered statistically signi cant. 0.24–5.9), and follicle-stimulating hormone (6.5 mU/mL; N=1.9–11.6). He had microcephaly (adult cranial perimeter: RESULTS 46 cm, 24.0 SD) and severe intellectual disability, with Description of a Novel Diabetes Syndrome With a quiet introverted character. At 15 years of age, his mental Young-Onset Diabetes, Short Stature, and level was comparable to that of a 5- to 6-year-old child. Microcephaly He answered questions using simple sentences but could We studied two siblings with young-onset diabetes, not read or write. His vocabulary was limited to 200–300 intellectual disability, microcephaly, and short stature who words and he did not engage in conversation. He was were born to first-cousin consanguineous parents without able to perform tasks such as feeding, dressing, and bathing diabetes (see Table 1 and Fig. 1 for the description of the but required full assistance in daily life. MRI showed rare- patients). The index case (patient 1), a boy, was diagnosed faction of the white matter (Fig. 1B), with a nonspecific with diabetes at age 15 years, with acute onset of polyuria slightly elevated level of protein in the cerebrospinal fluid and polydipsia. Fasting glucose was 13.4 mmol/L and HbA1c (albumin and IgG). He had neurogenic deafness (hearing loss was 13.0% (119 mmol/mol). Type 1 diabetes–specificauto- of 39%). He also had kyphoscoliosis, pectus excavatum, mild antibodies (islet cell antibody, GAD, IA2 antibodies) were abnormalities of vertebral bodies (Fig. 1C and D), fine fin- negative. Fasting C-peptide was low but within normal gers and toes, oligodontia and dental hypoplasia, sparse range for normoglycemic subjects (Table 1), showing hair, and a high-pitched voice. Eye fundus was normal. that at least some residual b-cell mass remained. He Clinical and biochemical examination at 28 years of age was treated with twice-daily insulin injections. Diabetes showed that his diabetes was relatively well controlled was initially well controlled with relatively low doses of (Table 1). Glucagon-stimulated C-peptide was detectable.

Table 1—Clinical and biochemical characteristics of the two patients Patient 1 Patient 2 Sex Male Female Age at diabetes onset, years 15 28 Age at follow-up examination, years 15* 28 31 Anthropometry Height, cm (SD) 146 (23.5) 155 (23.2) 139 (24.2) Weight, kg (SD) 31.4 (23.2) 44.3 (23.5) 30 (24.0) BMI, kg/m2 (SD) 14.7 (22.7) 18.4 (21.7) 15.5 (22.5) Microcephaly Yes Yes Yes Cranial perimeter, cm (SD) 46 (24.0) 46 (24.0) NA Glucose metabolism

HbA1c, % [N = 4.0–6.0] 13.0 7.8 NA C-peptide secretion evaluation Fasting glycemia, mmol/L [N = ,5.6] 3.96 11.4 NA Fasting C-peptide, nmol/L [N = 0.25–1.28] 0.56 NA NA Glucagon-stimulated C-peptide, nmol/L [N .0.6] NA 0.89 NA Therapy Insulin therapy duration, years 0 13 3 Insulin dose, units/kg/day 0 0.5 0.7 SD is based on French normative values. *, at diabetes onset. diabetes.diabetesjournals.org Abdulkarim and Associates 3955

Figure 1—Imaging of and skeleton of patient 1. A: Growth chart, showing growth retardation. B: Coronal T2-weighted brain MRI at age 15 years, showing a moderate white matter rarefaction characterized by increased sulcal size and moderate enlargement of ventricular system. C and D: Skeletal radiographies at age 28 years, showing kyphoscoliosis with tall vertebral bodies and hyperlordosis.

Pubertal development was fully achieved (Tanner stage 5) were low at 44.3 kg and 18.4 kg/m2, respectively. Dual- with adult genitalia and complete epiphyseal closure on energy X-ray absorptiometry showed normal body compo- bone age X-ray. Biochemical measurements in serum sition and bone mineral content. Overall, these results are showed low 25-hydroxyvitamin D (55 nmol/L; N = 75– in keeping with bone dysplasia without marked distur- 340), but normal levels of calcium, IGF-1, and thyroxine. bance of calcium metabolism, with severe growth retarda- Markers of phosphate metabolism were normal: parathy- tion unrelated to pituitary or dysfunction. Serum roid hormone (19 pg/mL; N = 10–55), serum phosphate amylase, blood cell count, liver and kidney function, and (1.54 mmol/L; N = 1.30–1.85), and alkaline phosphatase iron metabolism were normal. On ultrasound, liver and (320 units/L; N = 210–830). Urinary calcium-to-creatinine pancreas were normal, while the kidneys were small, ratio was also normal (0.33; N ,0.7). His weight and BMI with mild dilation of right calyces. 3956 PPP1R15B Mutation in a Novel Diabetes Syndrome Diabetes Volume 64, November 2015

His sister (patient 2) had a similar clinical presentation, To test this hypothesis, we performed exome sequencing but she was not available for detailed evaluation. She had on patient 1’s genomic DNA and identified 18 rare homo- growth retardation, microcephaly, intellectual disability, zygous autosomal variants after filtering (coding variants, and diabetes presenting with an acute onset of hypergly- minor allele frequency ,0.005, and absence of subjects cemia and ketosis at age 28 years. She was treated by homozygous for the rare variant in public and in-house insulin. She also had dental hypoplasia, an introverted databases; Supplementary Tables 4 and 5). We confirmed character, and high-pitched voice. She had menarche at the these variants in patient 1 and genotyped them in patient age of 14 years. At the last examination at 31 years of age, 2 and in two nonaffected relatives (a grandmother and an body weight was 30 kg, height 139 cm, BMI 15.5 kg/m2 aunt) by Sanger sequencing or PCR-RFLP genotyping. This (Table 1). Blood cell count, electrolytes, and kidney func- reduced the number of variants to 6, for which both sib- tion parameters were normal. Early clinical history was lings, but not the unaffected relatives, were homozygous unavailable for these patients, except for the information for the rare allele: ADAMTSL4, FLG, KIF21B, PPP1R15B, that they were born small for gestational age. The parents and SLC45A3, located on 1, and UNC80, lo- had normal fasting glucose (father, 5.1 mmol/L; mother, cated on chromosome 2 (Supplementary Tables 5 and 6). 4.3 mmol/L; N ,5.6), and the mother was not known to Because of the relative similarities of this syndrome with have had gestational diabetes mellitus. They were unavail- Wolcott-Rallison syndrome (EIF2AK3 mutations), the fea- 2 2 able for further clinical examination and genetic study. tures of the Ppp1r15b / mouse (very small size at birth and early death), and its role in the ER stress response Identification of a Homozygous Mutation in the (31), PPP1R15B (Fig. 2A)appearedasthemajorcandidate PPP1R15B Gene in the Two Siblings With Diabetes (Supplementary Table 6). None of the other genes showed Because of the familial context, we hypothesized that the obvious functional relevance to the syndrome (Supplemen- syndrome was caused by an autosomal recessive mutation. tary Table 6). PPP1R15B is ubiquitously expressed (and well

Figure 2—Identification of a homozygous PPP1R15B-R658C mutation in two siblings with diabetes and consequences of the mutation on the protein. A: Sanger sequencing of a control subject and the two siblings with diabetes (filled symbols), presenting the homozygous mutation and its consequence on the cDNA and protein. B: PPP1R15B protein sequence, showing the alignment of a highly conserved 62 amino acid segment (hatched) located within the COOH-terminal functional core region (gray). Representative sequences aligned are PPP1R15B from human (PR15B_HUMAN, Q5SWA1) and mouse (PR15B_MOUSE, Q8BFW3); PPP1R15A from human (PR15A_HUMAN, O75807), mouse (PR15A_MOUSE, P17564), and drosophila (PR15A_DROME, Q9W1E4); and homologous proteins from a variety of viruses: African Swine fever virus (VF71_ASFB7, Q65212), Amsacta moorei entomopoxvirus L (NP_064975), Glossina pallidipes salivary gland hypertrophy virus (YP_001687092), Choristoneura occidentalis granulovirus (YP_654457), and Trichoplusia ni ascovirus 2c (YP_803309). The mutated arginine (R) at position 658 is part of the functional RVxF-FF-R motif (boxed and underlined on human PPP1R15B) that has been recognized in PP1-interacting proteins (33). Residues shown in red are fully conserved in selected species; residues that are the most critical for establishing contact with PP1 according to Chen et al. (32) are indicated by stars. diabetes.diabetesjournals.org Abdulkarim and Associates 3957 expressed in human islets and b-cells), consistent with the presentation, i.e., insulin-dependent diabetes and short multisystem disease manifestations, while the other genes stature and/or mental retardation or microcephaly, and have a lower expression in islets and their pattern of tissue in 22 patients with diabetes and their families compat- expression does not specifically correspond to the syn- ible with monogenic diabetes and linkage to the PPP1R15B drome (Supplementary Fig. 1, Supplementary Table 6). chromosome region (C.J., unpublished data), but we did not The R658 residue of PPP1R15B is highly conserved be- identify any homozygous or compound heterozygous tween organisms, including viruses (Fig. 2B), and the PPP1R15B mutations in these patients. R658C mutation is predicted to be damaging by in silico prediction programs (Supplementary Table 6). It is lo- R658C Mutation Destabilizes the PPP1R15B-PP1 cated in the conserved COOH-terminalfunctionalcoreof Complex and Impairs eIF2a Dephosphorylation PPP1R15B that specifies interaction with PP1, the recruit- To study the effect of the R658C mutation on PPP1R15B ment of the essential cofactor G-actin, and substrate-specific function, we generated a plasmid encoding the fusion dephosphorylation (21,32). In the cocrystal structure, of WT or R658C-mutated human PPP1R15B to EGFP. PPP1R15B R658 inserts deep into a pocket on the surface HEK293T cells were transfected with either plasmid or of PP1 (34), giving rise to an ionic interaction with PP1 the empty pEGFP vector, alone or in combination with residue D71 that is conserved in other holophosphatases a plasmid expressing mouse PP1A. The PPP1R15B-PP1 such as PP1-PPP1R10 (PNUTS) and PP1-PPP1R9B (spinophilin) complex was immunoprecipitated from cell lysates using (33). Collectively, these structural observations strongly anti-GFP antibody and analyzed for the presence of PP1 by support a critical role of R658 in PP1 binding and predict Western blot. Fewer PP1 were recovered in complex with that the mutation of this residue has deleterious effects mutant PPP1R15B-EGFP compared with WT. This was true on protein function. both when endogenous PP1 was examined (compare lanes To search for additional patients with PPP1R15B muta- 1 and 2, Fig. 3A) and when PP1A was overexpressed (com- tions, we performed Sanger sequencing of PPP1R15B pare lanes 5 and 6, Fig. 3A), demonstrating that the R658C exons, flanking regions, and main regulatory regions (Sup- mutation reduced the ability of PPP1R15B to bind PP1, as plementary Table 1) in 50 patients with a similar clinical suggested by the structural studies.

Figure 3—The R658C mutation destabilizes the PPP1R15B-PP1 complex and diminishes its phosphatase activity. HEK293T cells were transfected with an empty vector (GFP) or a GFP-tagged WT or R658C-mutated human PPP1R15B, alone or in combination with a mouse PP1A expression plasmid. PPP1R15B-PP1 complexes were immunoprecipitated from lysed cells with anti-GFP antibody. A: The recovery of PP1 in complex with PPP1R15B was examined by Western blotting using anti-PP1 antibody. Immunoprecipitated protein is shown on the left (IP) and the eluant is shown on the right (input). B: The holophosphatase activity was studied in an eIF2a dephosphorylation assay, incubating the indicated PPP1R15B-PP1 complexes purified from cells with in vitro phosphorylated eIF2a (eIF2aP) protein for 30 min and resolving the phosphorylated and nonphosphorylated eIF2a (eIF2a0) on Phos-tag gels. C: A time course of eIF2a dephosphorylation by WT and R658C- mutated PPP1R15B and PP1A complexes recovered by immunopurification from transfected HEK293T cells. Unphosphorylated eIF2a was loaded onto lane 1 as a reference. The blots are representative of three or more independent experiments with similar outcomes. 3958 PPP1R15B Mutation in a Novel Diabetes Syndrome Diabetes Volume 64, November 2015

Next, we examined the dephosphorylation activity point (compare lanes 7 and 12, Fig. 3C). Thus, the mis- of the immunopurified PPP1R15B-PP1 holophosphatase sense mutation R658C negatively affects the stability of complex. Complexes recovered by immunopurification of the PPP1R15B-PP1 complex and, in turn, its ability to GFP-tagged WT PPP1R15B were more active at dephos- dephosphorylate eIF2a. phorylating eIF2a in vitro than complexes constituted of the R658C mutant (compare lanes 4 and 5, Fig. 3B). These PPP1R15B Expression in b-Cells differences were rendered more conspicuous in a time In clonal rat INS-1E b-cells, PPP1R15B was induced by course study. Complexes containing WT PPP1R15B sub- different synthetic ER stressors and the saturated FFA stantially dephosphorylated eIF2a protein by 45 min, palmitate (Fig. 4A). This induction was prevented by whereas in dephosphorylation reactions carried out with a chemical inhibitor of PERK (Fig. 4B), suggesting that, R658C mutant PPP1R15B, substantial amounts of phos- different from other cell types (8), PPP1R15B expression phorylated eIF2a remained even at the 60-min time in b-cells is controlled by the UPR.

Figure 4—PPP1R15B is induced by ER stress in b-cells in a PERK-dependent manner, and PPP1R15B silencing induces eIF2a phosphor- ylation and ATF3 in b-cells. A: INS-1E cells were exposed to the chemical ER stressors CPA, tunicamycin (TU), or brefeldin A (BR) or to the FFAs oleate (OL) or palmitate (PAL) for 24 h (n =5–6). B: INS-1E cells were exposed or not (CT) to CPA in the presence or absence of the PERK inhibitor GSK2606414 (PERKi). PPP1R15B mRNA expression was examined by real-time PCR and normalized to the reference gene GAPDH. C: INS-1E cells were transfected with control siRNA (siCT) or two different siRNAs targeting PPP1R15B (P1R15B1 and P1R15B2). After a 48-h transfection, the cells were treated for 16 h with CPA, OL, or PAL. PPP1R15B mRNA expression was examined by real-time PCR and normalized to the reference gene GAPDH (n = 4). eIF2a phosphorylation (P-eIF2a)(D and E)andATF3(F and G) expression were examined by Western blot. D and F are representative images of n =4.E and G represent densitometric quantifications of D and F,respectively.P-eIF2a was corrected for total eIF2a. ATF3 expression was corrected for a-tubulin and expressed as fold of CT. Data are presented as means 6 SE. *treatedvs.control,§DMSOvs.PERKi,#siP1R15Bvs.siCTbytwo-sidedStudentpairedt test. *,§,#P < 0.05; **,##P < 0.01; ***,###P < 0.001. diabetes.diabetesjournals.org Abdulkarim and Associates 3959

PPP1R15B Deficiency Increases eIF2a rat b-cells (Fig. 6C) sensitized the cells to apoptosis under Phosphorylation in b-Cells basal conditions and induced up to 20% more apoptosis Using RNA interference, PPP1R15B was knocked down following exposure to CPA or the FFAs oleate and palmi- in INS-1E cells, resulting in a 75% inhibition of mRNA tate (Fig. 6A–C). expression (Fig. 4C). Consistent with its previously reported To evaluate whether the intrinsic pathway of apoptosis function of constitutive eIF2a phosphatase (8), basal eIF2a was involved, we measured mitochondrial cytochrome c phosphorylation was increased in PPP1R15B-deficient INS- release to the cytoplasm and cleavage of caspase-9 and -3. In 1E cells (Fig. 4D and E). PPP1R15B silencing did not further PPP1R15B-deficient INS-1E cells, cytoplasmic cytochrome c increase eIF2a phosphorylation induced by the synthetic levels were increased (Fig. 6D)andcaspase-9(Fig.6E)and-3 ER stressor CPA or palmitate (Fig. 4D and E). Phosphor- (Fig. 6F) were cleaved, demonstrating activation of the in- ylated eIF2a levels were higher in PPP1R15B-deficient trinsic pathway of apoptosis. cells exposed to the unsaturated FFA oleate that per se We next examined which BCL-2 family members does not induce eIF2a phosphorylation (28,34). ATF3 activate the intrinsic pathway of apoptosis. PPP1R15B protein was also induced in PPP1R15B-deficient cells un- deficiency induced mRNA expression of the proapoptotic der basal conditions (Fig. 4F and G). Thus, PPP1R15B proteins DP5 and PUMA (Fig. 6G and H), but it did not silencing increases eIF2a phosphorylation and induces significantly affect expression of Bim-S protein (Fig. 6I). downstream ATF3 protein expression. The expression of Bim-L and EL (Supplementary Fig. 2A– C) and the antiapoptotic proteins BCL-2 and BCL-XL (Sup- PPP1R15B Silencing Decreases Insulin Content and plementary Fig. 2D–F)werenotmodified by PPP1R15B Glucose-Stimulated Insulin Release fi fi de ciency. DP5, PUMA, or Bim silencing partially pro- We next evaluated the effect of PPP1R15B de ciency on tected PPP1R15B-silenced b-cells from apoptosis (Fig. b-cell function. PPP1R15B knockdown decreased insulin J fi A 6 ), showing that PPP1R15B de ciency induces apopto- content in INS-1E cells by 20% (Fig. 5 ). In control siRNA- sis through the proapoptotic BH3-only proteins DP5, b transfected -cells, high glucose exposure (16.7 mmol/L) PUMA, and Bim. increased insulin secretion by 2.8-fold. PPP1R15B-deficient b-cells showed increased basal insulin secretion and little DISCUSSION B or no response to high glucose (Fig. 5 ). 16.7 mmol/L R658C is the first reported mutation in PPP1R15B, re- glucose plus forskolin induced insulin secretion by 10- sponsible for a novel diabetes syndrome with onset in fold in control siRNA-transfected cells, but only by fourfold P , youth/young adulthood. It affects a key amino acid in after PPP1R15B silencing ( 0.05). the conserved C-terminus region of the protein that makes contacts with PP1 that is conserved in other holophosphatase fi b PPP1R15B-De cient -Cells Are Sensitized to complexes (32,33). We show that the R658C mutation Apoptosis Through the Proapoptotic BH3-Only destabilizes the complex between PPP1R15B and PP1, Proteins DP5, PUMA, and Bim and this, in turn, diminishes eIF2a dephosphorylation. We examined the role of PPP1R15B in b-cell survival. Furthermore, we show that PPP1R15B silencing alters PPP1R15B silencing in clonal (Fig. 6A and B) and primary b-cell function, inducing higher basal insulin secretion and reducing high glucose responsiveness, which is com- patible with a role of PPP1R15B in exocytosis (35). This is consistent with the observation of residual C-peptide in the index patient and his moderate insulin require- ment and frequent hypoglycemic events. PPP1R15B si- lencing also sensitizes b-cells to apoptosis, both under basal and ER stress conditions, which was induced via the intrinsic pathway of apoptosis, as in the case of palmitate-induced eIF2a phosphorylation (36). Of note, diabetes manifested relatively late in these patients (at ages 15 and 28 years). Considering the con- genital nature of the defect, this suggests the role of compensatory mechanisms that maintain a sufficient b-cell function for several years. The acute onset of diabetes — Figure 5 Glucose-stimulated insulin secretion is blunted by suggests a threshold mechanism(s) by which this delicate PPP1R15B deficiency in b-cells. INS-1E cells were transfected with control siRNA (siCT) or two siRNAs targeting PPP1R15B balance is disrupted at some stage, as observed in autoim- (siP1R15B1 and siP1R15B2). After a 48-h transfection, insulin se- mune type 1 diabetes and other forms of monogenic di- cretion was induced by 1.67 mmol/L or 16.7 mmol/L glucose or abetes, leading to hyperglycemia (37). 16.7 mmol/L glucose + 10 mmol/L forskolin (16.7+FK). A: Cellular insulin content corrected for total protein. B: Insulin release as per- The syndrome caused by the PPP1R15B-R658C muta- cent of insulin content (n = 5). **P < 0.01, ***P < 0.001 vs. 1.67 tion results in multisystem manifestations, affecting mmol/L glucose. #P < 0.05, ##P < 0.01, siP1R15B vs. siCT. b-cells (diabetes), the nervous system (microcephaly, 3960 PPP1R15B Mutation in a Novel Diabetes Syndrome Diabetes Volume 64, November 2015

Figure 6—PPP1R15B deficiency sensitizes b-cells to FFA- and ER stress–induced apoptosis and activates the intrinsic pathway of apoptosis via DP5, PUMA, and Bim-S. INS-1E (A and B) or primary rat b-cells (C) were transfected with a control siRNA (siCT) or two different siRNAs targeting PPP1R15B (siP1R15B1 and siP1R15B2). After a 24-h transfection, the cells were exposed or not (CT) to CPA, oleate (OL), or palmitate (PAL) for 16 (A and B)or24h(C)(n =4–5). D: Mitochondrial cytochrome c (Cyto C) release was detected by Western blot in the cytoplasmic fraction 48 h after PPP1R15B knockdown. The right lane shows a noncytoplasmic fraction that includes mitochondria (Mito). Cox IV was used as a mitochondrial control and b-actin as a cytoplasmic control. Activation of caspase-9 (Casp9) (E) and caspase-3 (Casp3) (F) was detected by Western blot 48 h after PPP1R15B knockdown. b-Actin and a-tubulin were used as loading controls. D, E, and F are representative blots of 4–5 experiments. The densitometry data were normalized to the highest value. DP5 (G) and PUMA (H) mRNA expression were measured by real-time PCR and corrected for the reference gene GAPDH (n = 4). Bim-S levels were measured by Western blot (Supplementary Fig. 2), corrected for a-tubulin, and expressed as fold of siCT (I)(n = 5). J: PPP1R15B was silenced alone or in combination with DP5, PUMA, or Bim, and apoptosis was examined by Hoechst 33342/propidium iodide staining (n = 3–4). *P < 0.05, **P < 0.01, ***P < 0.001, treated vs. control. #P < 0.05, ##P < 0.01, ###P < 0.001, siP1R15B vs. siCT. §P < 0.05, single vs. double knockdown.

intellectual disability, and hearing loss), and bone (bone secretory cells, with insulin-producing b-cells, collagen- deformities) and general development. This phenotypic producing bone cells, and nerve cells especially vulnerable spectrum has interesting overlap with other mutations af- to imbalance. In the Wolcott-Rallison syndrome (with de- fecting the levels of ER stress or the response to it. Hence, creased eIF2a phosphorylation), neonatal diabetes is due b-cell dysfunction, bone abnormalities, microcephaly, and to b-cell loss and C-peptide is undetectable (13), while intellectual disability are shared with the Wolcott-Rallison DNAJC3, IER3IP1,andPPP1R15B mutations (with in- syndrome caused by EIF2AK3 mutations, which lead to creased eIF2a phosphorylation) lead to permanent neo- higher levels of ER stress associated with less eIF2a natal or young-onset diabetes with residual C-peptide phosphorylation (11,13,38), and with syndromes caused levels, as reported by Synofzik et al. (16), Shalev et al. by mutations in the ER cochaperone DNAJC3 and in (40), and the current study. It is notable that the muta- the immediate early response 3 interacting protein 1 tions affecting the target of phosphorylated eIF2a, the (IER3IP1), which both lead to higher levels of ER stress guanine nucleotide exchange factor eIF2B, feature prom- with more eIF2a phosphorylation (16,39,40). These com- inent neurodegenerative manifestations known as the monalities likely reflect the importance of balanced CACH (childhood ataxia with central nervous system ER protein synthesis and folding to the function of hypomyelination)/VWM (vanishing white matter disease) diabetes.diabetesjournals.org Abdulkarim and Associates 3961 syndrome (41), which are phenocopied by targeted neurons, and bone and suggest that other proteins in- activation of the eIF2a kinase PERK in the brain (42). volved in the regulation of protein translation may lead Interestingly, severe forms of CACH/VWM also have to similar diabetes syndromes. multiorgan manifestations (41,43). The clinical variability of syndromes associated with dysregulated eIF2a phos- phorylation likely results from the interplay of several Acknowledgments. The authors are very grateful to the patients and their factors, including the extent of apoptosis and/or secretory family for their participation in the study. The authors thank Christophe Caloustian fi and Sylvana Pavek from the Centre National de Génotypage for expert technical dysfunction resulting from speci cgenesmutations, assistance. The authors are also grateful to the Flow Cytometry Facility of the the severity of the mutation, and possibly environmental Erasmus Campus of the Université Libre de Bruxelles and Christine Dubois for the stresses. cell sorting. The authors thank Michael Pangerl, Anyishai Musuaya, Nathalie The clinical features of PPP1R15B-mutated patients Pachera, Stephanie Mertens, and Isabelle Millard at the ULB Center for Diabetes described here and the phenotype of knockout mice Research (Université Libre de Bruxelles) for excellent technical support. The authors that we described previously (31) suggest that PPP1R15B also thank Annabelle Chaussenot from the School of Medicine, Nice Sophia Antipolis deficiency affects multiple cell types. The role of PPP1R15B University, for contributing some patients for this study and for interesting discus- in exocytosis and membrane traffic has been previously sions. The authors thank Dr. Belkacem Bioud, University Hospital of Setif, Algeria, reported in human epithelial cells and erythroleukemia and Dr. Hadda Baaziz, University Hospital of Batna, Algeria, for their collaboration. cells (35). The knockdown of PPP1R15B in breast cancer Funding. This work was supported by the European Union 7th Framework cells resulted in impaired cell cycle transition from G1 to S Programme (project BetaBat), the Actions de Recherche Concertées de la Communauté Française, and Fonds National de la Recherche Scientifique phase and apoptosis (44). These observations suggest that (FNRS), Belgium, and by grants from the Agence Nationale pour la Recherche PPP1R15B regulates a variety of functions in different cell (ANR-09-GENO-021), the European Foundation for the Study of Diabetes/JDRF/ types. More studies are needed to explore this further. Novo Nordisk, the Assistance Publique-Hôpitaux de Paris Programme Hospital- Although we selected PPP1R15B as an obvious candi- ier de Recherche Clinique (DIAGENE), the GIS Maladies Rares, and the Well- date gene and found ample structural and functional ex- come Trust (084812/Z/08/Z). A.T.H. is a Wellcome Trust and National Institute perimental support for the impact of the mutation, we for Health Research senior investigator, and D.R. is a Wellcome Trust Principal cannot formally exclude that one of the rare variants Research Fellow. B.A. was supported by an European Orga- homozygous in the two affected siblings contributes to nization Short-Term Fellowship and an FNRS-FRIA fellowship. M.I.-E. is a sci- the syndrome. This hypothesis is unlikely based on the entific collaborator of the FNRS. M.D. was supported by a doctoral fellowship ’ ’ following reasons (Supplementary Table 6 and Supple- from the Ministère de l Education Nationale, de l Enseignement Supérieur et de mentary Fig. 1): 1) none of the other genes appear directly la Recherche, France. Duality of Interest. No potential conflicts of interest relevant to this article functionally relevant to the clinical presentation; 2) the fi were reported. expression pattern of PPP1R15B, but not of the ve other Author Contributions. B.A., M.N., A.P., D.L.E., D.R., M.C., and C.J. genes, is consistent with the multisystem presentation of contributed to the study design. M.D., S.R., A.P., V.S., C.D., and C.J. performed the syndrome; 3) mutations in ADAMTSL4 cause ectopia the genetic experiments and analyzed and interpreted the genetic data. B.A., lentis and mutations in FLG cause atopic dermatitis and M.I.-E., M.L., D.A.C., H.P.H., D.L.E., D.R., and M.C. performed the functional ichtyosis vulgaris, and none of these diseases were pres- experiments and analyzed and interpreted the functional data. M.N. and N.B. ent in the two affected siblings; and 4) human or mouse identified the index patient and family and characterized these patients. disease phenotypes of KIF21B, SLC45A3,andUNC80 M.N., N.B., and A.T.H. contributed patients. B.A., M.N., A.P., D.R., M.C., and C.J. mutations have not been reported so far, and a cellular wrote the manuscript. All coauthors read and approved the manuscript. C.J. is phenotype of chromosome instability has been reported the guarantor of this work and, as such, had full access to all the data in the 2 2 in Kif21b / mice, which was not observed in our study and takes responsibility for the integrity of the data and the accuracy of the data analysis. patients. In short, the characteristics of these five other Prior Presentation. Parts of this study were presented in abstract form at genes show little relevance to the diabetes and neurolog- the 51st European Association for the Study of Diabetes Annual Meeting, ical phenotype of our patients. Stockholm, Sweden, 14–18 September 2015. The present human genetic observations are in keeping with our earlier findings that excessive eIF2a phosphor- References ylation is poorly tolerated by b-cells. Salubrinal, a chemical 1. Eizirik DL, Cardozo AK, Cnop M. The role for endoplasmic reticulum stress – inhibitor of eIF2a dephosphorylation, was identified in in diabetes mellitus. Endocr Rev 2008;29:42 61 a large-scale chemical screening as a compound that pro- 2. Cnop M, Foufelle F, Velloso LA. Endoplasmic reticulum stress, obesity and diabetes. Trends Mol Med 2012;18:59–68 tects from ER stress (45), and it was even suggested as 3. Marhfour I, Lopez XM, Lefkaditis D, et al. Expression of endoplasmic re- a potential therapy to preserve b-cells in diabetes (46). In ticulum stress markers in the islets of patients with type 1 diabetes. Diabetologia rodent and human b-cells, however, salubrinal actually 2012;55:2417–2420 exacerbates cell death through ER stress and downstream 4. Laybutt DR, Hawkins YC, Lock J, et al. Influence of diabetes on the loss of activation of the mitochondrial pathway of apoptosis beta cell differentiation after islet transplantation in rats. Diabetologia 2007;50: (10,24). The present observations support the concept 2117–2125 that dysregulation of eIF2a phosphorylation, either ex- 5. Marchetti P, Bugliani M, Lupi R, et al. The endoplasmic reticulum in cessive (PPP1R15B, DNAJC3,andIER3IP1 mutations) or pancreatic beta cells of type 2 diabetes patients. Diabetologia 2007;50:2486– diminished (EIF2AK3 mutations), is detrimental to b-cells, 2494 3962 PPP1R15B Mutation in a Novel Diabetes Syndrome Diabetes Volume 64, November 2015

6. Hartman MG, Lu D, Kim ML, et al. Role for activating transcription factor 3 in 26. Moore F, Colli ML, Cnop M, et al. PTPN2, a candidate gene for type 1 di- stress-induced beta-cell apoptosis. Mol Cell Biol 2004;24:5721–5732 abetes, modulates interferon-g-induced pancreatic b-cell apoptosis. Diabetes 7. Ron D, Walter P. Signal integration in the endoplasmic reticulum unfolded 2009;58:1283–1291 protein response. Nat Rev Mol Cell Biol 2007;8:519–529 27. Cnop M, Hannaert JC, Hoorens A, Eizirik DL, Pipeleers DG. Inverse re- 8. Jousse C, Oyadomari S, Novoa I, et al. Inhibition of a constitutive translation lationship between cytotoxicity of free fatty acids in pancreatic islet cells and initiation factor 2a phosphatase, CReP, promotes survival of stressed cells. J Cell cellular triglyceride accumulation. Diabetes 2001;50:1771–1777 Biol 2003;163:767–775 28. Cunha DA, Hekerman P, Ladrière L, et al. Initiation and execution of lipotoxic 9. Novoa I, Zeng H, Harding HP, Ron D. Feedback inhibition of the unfolded ER stress in pancreatic b-cells. J Cell Sci 2008;121:2308–2318 protein response by GADD34-mediated dephosphorylation of eIF2alpha. J Cell 29. Harding HP, Zyryanova AF, Ron D. Uncoupling proteostasis and develop- Biol 2001;153:1011–1022 ment in vitro with a small molecule inhibitor of the pancreatic endoplasmic re- 10. Cnop M, Ladriere L, Hekerman P, et al. Selective inhibition of eukaryotic ticulum kinase, PERK. J Biol Chem 2012;287:44338–44344 translation initiation factor 2 alpha dephosphorylation potentiates fatty acid- 30. Igoillo-Esteve M, Marselli L, Cunha DA, et al. Palmitate induces a pro- induced endoplasmic reticulum stress and causes pancreatic beta-cell dys- inflammatory response in human pancreatic islets that mimics CCL2 expression function and apoptosis. J Biol Chem 2007;282:3989–3997 by beta cells in type 2 diabetes. Diabetologia 2010;53:1395–1405 11. Delépine M, Nicolino M, Barrett T, Golamaully M, Lathrop GM, Julier 31. Harding HP, Zhang Y, Scheuner D, Chen JJ, Kaufman RJ, Ron D. Ppp1r15 C. EIF2AK3, encoding translation initiation factor 2-alpha kinase 3, is gene knockout reveals an essential role for translation initiation factor 2 alpha mutated in patients with Wolcott-Rallison syndrome. Nat Genet 2000;25: (eIF2alpha) dephosphorylation in mammalian development. Proc Natl Acad Sci 406–409 U S A 2009;106:1832–1837 12. Scheuner D, Song B, McEwen E, et al. Translational control is required for 32. Chen R, Rato C, Yan Y, et al. G-actin provides substrate-specificity to eu- the unfolded protein response and in vivo glucose homeostasis. Mol Cell 2001;7: karyotic initiation factor 2a holophosphatases. eLife 2015;4:e04871 1165–1176 33. Choy MS, Hieke M, Kumar GS, et al. Understanding the antagonism of 13. Julier C, Nicolino M. Wolcott-Rallison syndrome. Orphanet J Rare Dis 2010; retinoblastoma protein dephosphorylation by PNUTS provides insights into the 5:29 PP1 regulatory code. Proc Natl Acad Sci U S A 2014;111:4097–4102 14. Harding HP, Zeng H, Zhang Y, et al. Diabetes mellitus and exocrine pan- 34. Karaskov E, Scott C, Zhang L, Teodoro T, Ravazzola M, Volchuk A. Chronic creatic dysfunction in perk-/- mice reveals a role for translational control in palmitate but not oleate exposure induces endoplasmic reticulum stress, which secretory cell survival. Mol Cell 2001;7:1153–1163 may contribute to INS-1 pancreatic b-cell apoptosis. Endocrinology 2006;147: 15. Støy J, Edghill EL, Flanagan SE, et al.; Neonatal Diabetes International 3398–3407 Collaborative Group. Insulin gene mutations as a cause of permanent neonatal 35. Kloft N, Neukirch C, von Hoven G, et al. A subunit of eukaryotic translation diabetes. Proc Natl Acad Sci U S A 2007;104:15040–15044 initiation factor 2a-phosphatase (CreP/PPP1R15B) regulates membrane traffic. 16. Synofzik M, Haack TB, Kopajtich R, et al. Absence of BiP co-chaperone J Biol Chem 2012;287:35299–35317 DNAJC3 causes diabetes mellitus and multisystemic neurodegeneration. Am J 36. Cunha DA, Igoillo-Esteve M, Gurzov EN, et al. Death protein 5 and p53- Hum Genet 2014;95:689–697 upregulated modulator of apoptosis mediate the endoplasmic reticulum stress- 17. Takahashi M, Matsuda F, Margetic N, Lathrop M. Automated identification mitochondrial dialog triggering lipotoxic rodent and human b-cell apoptosis. of single nucleotide polymorphisms from sequencing data. J Bioinform Comput Diabetes 2012;61:2763–2775 Biol 2003;1:253–265 37. Eizirik DL, Sandler S, Palmer JP. Repair of pancreatic beta-cells. A relevant 18. Ortis F, Cardozo AK, Crispim D, Störling J, Mandrup-Poulsen T, Eizirik DL. phenomenon in early IDDM? Diabetes 1993;42:1383–1391 Cytokine-induced proapoptotic gene expression in insulin-producing cells is 38. de Wit MC, de Coo IF, Julier C, et al. Microcephaly and simplified gyral related to rapid, sustained, and nonoscillatory nuclear factor-kappaB activation. pattern of the brain associated with early onset insulin-dependent diabetes Mol Endocrinol 2006;20:1867–1879 mellitus. Neurogenetics 2006;7:259–263 19. Rasschaert J, Ladrière L, Urbain M, et al. Toll-like receptor 3 and STAT-1 39. Poulton CJ, Schot R, Kia SK, et al. Microcephaly with simplified gyration, contribute to double-stranded RNA+ interferon-g-induced apoptosis in primary epilepsy, and infantile diabetes linked to inappropriate apoptosis of neural pro- pancreatic b-cells. J Biol Chem 2005;280:33984–33991 genitors. Am J Hum Genet 2011;89:265–276 20. Marroqui L, Masini M, Merino B, et al. Pancreatic a cells are resistant to 40. Shalev SA, Tenenbaum-Rakover Y, Horovitz Y, et al. Microcephaly, epilepsy, metabolic stress-induced apoptosis in type 2 diabetes. EBioMedicine 2015;2: and neonatal diabetes due to compound heterozygous mutations in IER3IP1: in- 378–385 sights into the natural history of a rare disorder. Pediatr Diabetes 2014;15:252–256 21. Chambers JE, Dalton LE, Clarke HJ, et al. Actin dynamics tune the in- 41. Fogli A, Boespflug-Tanguy O. The large spectrum of eIF2B-related diseases. tegrated stress response by regulating eukaryotic initiation factor 2a de- Biochem Soc Trans 2006;34:22–29 phosphorylation. eLife 2015;4:e04872 42. Lin Y, Pang X, Huang G, et al. Impaired eukaryotic translation initiation 22. Cnop M, Igoillo-Esteve M, Rai M, et al. Central role and mechanisms of factor 2B activity specifically in oligodendrocytes reproduces the b-cell dysfunction and death in friedreich ataxia-associated diabetes. Ann Neurol of vanishing white matter disease in mice. J Neurosci 2014;34:12182– 2012;72:971–982 12191 23. Kharroubi I, Ladrière L, Cardozo AK, Dogusan Z, Cnop M, Eizirik DL. Free 43. van der Knaap MS, van Berkel CG, Herms J, et al. eIF2B-related disorders: fatty acids and cytokines induce pancreatic b-cell apoptosis by different antenatal onset and involvement of multiple organs. Am J Hum Genet 2003;73: mechanisms: role of nuclear factor-kappaB and endoplasmic reticulum stress. 1199–1207 Endocrinology 2004;145:5087–5096 44. Shahmoradgoli M, Riazalhosseini Y, Haag D, et al. Protein phosphatase 1, 24. Ladrière L, Igoillo-Esteve M, Cunha DA, et al. Enhanced signaling down- regulatory subunit 15B is a survival factor for ERa-positive breast cancer. Int stream of ribonucleic acid-activated protein kinase-like endoplasmic reticulum J Cancer 2013;132:2714–2719 kinase potentiates lipotoxic endoplasmic reticulum stress in human islets. J Clin 45. Boyce M, Bryant KF, Jousse C, et al. A selective inhibitor of eIF2alpha Endocrinol Metab 2010;95:1442–1449 dephosphorylation protects cells from ER stress. Science 2005;307: 25. Overbergh L, Valckx D, Waer M, Mathieu C. Quantification of murine cy- 935–939 tokine mRNAs using real time quantitative PCR. Cytokine 46. Wiseman RL, Balch WE. A new pharmacology–drugging stressed folding 1999;11:305–312 pathways. Trends Mol Med 2005;11:347–350

PAPER III Abdulkarim B, Hernangomez M, Igoillo-Esteve M, Ladriere L, Cunha DA, Marselli L,Marchetti P, Eizirik DL, Cnop M; Guanabenz sensitizes pancreatic β cells to lipotoxic endoplasmic reticulum stress and apoptosis

Background: Balanced endoplasmic reticulum (ER) stress signaling is crucial for the normal function and survival of pancreatic β cells. Deficient as well as excessive/prolonged ER stress can lead to the development of diabetes. Saturated free fatty acids (FFAs), e.g. palmitate, induce ER stress in pancreatic β cells. One of the main pathways involved in ER stress signaling is the PERK pathway, leading to phosphorylation of the eukaryotic translation initiation factor 2 (eIF2α). A narrow regulation of eIF2α phosphorylation status is required for the function and survival of pancreatic β cells. Guanabenz has been shown to inhibit eIF2α dephosphorylation and has been proposed to protect cells from ER stress. Our aim was to evaluate these protective mechanisms in β cells undergoing lipotoxic ER stress. Methods: Apoptosis was assessed by Hoechst 33342/propidium iodide staining and western blot for caspase 3 cleavage. Gene expression was examined by real-time PCR and western blot. Mice were treated through IP injection for one week in combination with diet intervention. Glucose tolerance was evaluated through IPGTT. Silencing of CHOP was achieved by RNA interference in clonal INS-1E cells. Results: Guanabenz appeared to potentiate FFA-induced cell death in INS-1E cells and rat and human islets. Guanabenz induced β cell dysfunction in vitro and in vivo in and led to impaired glucose tolerance. Our data suggests that guanabenz potentiated the FFA-induced eIF2α phosphorylation and enhanced expression of the downstream pro-apoptotic gene CHOP. Guanabenz seemed to induce CHOP promoter activity basally and following palmitate exposure. Guanabenz also enhanced the expression of the ER chaperone BiP and reduced the expression of XBP1s. Silencing of CHOP protected INS-1E cells from guanabenz enhanced lipotoxicity. Conclusion: Guanabenz does not protect β cells from ER stress; instead it further potentiates lipotoxic ER stress-induced β cell dysfunction and apoptosis through enhanced signaling in the PERK pathway. These data strengthen the importance of the tight regulation of eIF2α phosphorylation in β cells.

My contribution: I have contributed to ~70% of the work carried out, in particular, I have:

 Designed all primers, siRNAs and plasmids used in the study  Performed: - Cell culture, transfection and treatment assays - Apoptosis analysis - In vivo experiments including; guanabenz treatment, IPGTT 62

- Islet isolation and dispersion - Immune staining assay - Glucose stimulated insulin secretion - mRNA extraction and reverse transcription - Western blots  Analyzed the data  Prepared the manuscript

63

Guanabenz sensitizes pancreatic β cells to lipotoxic endoplasmic reticulum stress and apoptosis

Baroj Abdulkarim, Miriam Hernangomez, Mariana Igoillo Esteve, Daniel A Cunha, Lorella Marselli, Piero Marchetti, Laurence Ladriere, Miriam Cnop

Endocrinology Endocrine Society

Submitted: October 19, 2016 Accepted: February 24, 2017 First Online: March 1, 2017

Advance Articles are PDF versions of manuscripts that have been peer reviewed and accepted but

Endocrinology not yet copyedited. The manuscripts are published online as soon as possible after acceptance and before the copyedited, typeset articles are published. They are posted "as is" (i.e., as submitted by the authors at the modification stage), and do not reflect editorial changes. No corrections/changes to the PDF manuscripts are accepted. Accordingly, there likely will be differences between the Advance Article manuscripts and the final, typeset articles. The manuscripts remain listed on the Advance Article page until the final, typeset articles are posted. At that point, the manuscripts are removed from the Advance Article page.

DISCLAIMER: These manuscripts are provided "as is" without warranty of any kind, either express or particular purpose, or non-infringement. Changes will be made to these manuscripts before publication. Review and/or use or reliance on these materials is at the discretion and risk of the reader/user. In no event shall the Endocrine Society be liable for damages of any kind arising references to, products or publications do not imply endorsement of that product or publication. ADVANCE ARTICLE: Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

Guanabenz potentiates lipotoxicity

Guanabenz sensitizes pancreatic β cells to lipotoxic endoplasmic reticulum stress and apoptosis

Baroj Abdulkarim1, Miriam Hernangomez1, Mariana Igoillo Esteve1, Daniel A Cunha1, Lorella Marselli2, Piero Marchetti2, Laurence Ladriere1, Miriam Cnop1,3 1ULB Center for Diabetes Research, Université Libre de Bruxelles (ULB), Brussels, Belgium; 2Department of Endocrinology and Metabolism, University of Pisa, Pisa, Italy; 3Division of Endocrinology, Erasmus Hospital, Brussels, Belgium Received 19 October 2016. Accepted 24 February 2017. Deficient as well as excessive/prolonged endoplasmic reticulum (ER) stress signaling can lead to pancreatic β cell failure and the development of diabetes. Saturated free fatty acids (FFAs) such as palmitate induce lipotoxic ER stress in pancreatic β cells. One of the main ER stress response pathways is under the control of the protein kinase R-like endoplasmic reticulum kinase (PERK), leading to phosphorylation of the eukaryotic translation initiation factor 2 (eIF2α). The antihypertensive drug guanabenz has been shown to inhibit eIF2α dephosphorylation and protect cells from ER stress. Here we examined whether guanabenz protects pancreatic β cells from lipotoxicity. Guanabenz induced β cell dysfunction in vitro and in vivo in rodents and led to impaired glucose tolerance. The drug significantly potentiated FFA-induced cell death in clonal rat β cells and in rat and human islets. Guanabenz enhanced FFA-induced eIF2α phosphorylation and expression of the downstream pro-apoptotic gene CHOP, which mediated the sensitization to lipotoxicity. Thus, guanabenz does not protect β-cells from ER stress; instead it potentiates lipotoxic ER stress through PERK/eIF2α/CHOP signaling. These data demonstrate the crucial importance of the tight

Endocrinology regulation of eIF2α phosphorylation for the normal function and survival of pancreatic β cells. Guanabenz potentiates lipotoxic β cell ER stress by promoting eIF2α phosphorylation and CHOP

Endocrine Reviews expression. Tight regulation of eIF2α phosphorylation is crucial for β cell function and survival.

Introduction Accumulating evidence indicates that loss of functional pancreatic β cell mass in type 2 diabetes results from environmental insults causing cellular stress responses that activate specific transcription factor and gene networks. Lipotoxicity, a term referring to the deleterious effects of prolonged exposure to free fatty acids (FFAs), leads to the impairment of insulin secretion (1-3) and β cell death (4). Perturbations in the endoplasmic reticulum (ER) environment lead to accumulation of unfolded proteins and activation of the ER stress response. Previously, it has been demonstrated that FFAs induce ER stress in β cells (5-8); the activation of this stress response mediates at least part of FFA-induced β cell apoptosis. ER stress markers are present in β cells from typeADVANCE 2 diabetic patients (8-11). ARTICLE One of the canonical ER stress pathways depends on activation of protein kinase R-like

ADVANCE ARTICLE: endoplasmic reticulum kinase (PERK) and its phosphorylation of eukaryotic translation initiation factor 2α (eIF2α) leading to translation attenuation and reduced protein load on the

ADVANCE ARTICLE: ER. In parallel, eIF2α phosphorylation augments translation of the activating transcription factor (ATF) 4, in turn leading to the transcription of the pro-apoptotic C/EBP homologous protein (CHOP). In a negative feedback loop, ATF4 and CHOP induce GADD34 expression, which targets protein phosphatase 1 (PP1) to dephosphorylate eIF2α (12). The other canonical ER stress pathways are activated by inositol requiring 1α (IRE1α, which activates XBP1) and

1 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

ATF6, leading to RNA degradation and transcription of ER chaperones, such as BiP, and folding enzymes. This transcriptional response increases ER capacity (13). Several studies have demonstrated that genetic dysregulation of the PERK pathway leads to β cell demise and diabetes. Loss of function mutations in EIF2AK3, encoding PERK, cause Wolcott-Rallison syndrome, a rare autosomal recessive disease characterized by early onset non-autoimmune diabetes associated with skeletal dysplasia and growth retardation (14). We have recently shown that a loss of function mutation in PPP1R15B, encoding the constitutive repressor of eIF2α phosphorylation (CReP), is causal of a syndrome of young onset diabetes, microcephaly and growth retardation (15). As GADD34, CReP is a non-enzymatic cofactor for PP1. The CReP mutation greatly reduces PP1 binding and diminishes eIF2α dephosphorylation, causing β cell dysfunction and death. Mice carrying a homozygous mutation in the phosphorylation site of eIF2α (S51A) are insulin deficient and die shortly after birth (16). When challenged with a high fat diet (HFD), mice heterozygous for this mutation develop β cell dysfunction and diabetes (17). These data suggest that both excessive and reduced eIF2α phosphorylation can lead to β cell dysfunction and apoptosis. Fine-tuning of this process therefore seems crucial for proper β cell function and survival. Pharmacological approaches to modulate eIF2α phosphorylation therefore hold potential for the treatment of diabetes. Guanabenz, an α2-adrenergic receptor agonist used for the treatment of hypertension, binds to GADD34 and inhibits eIF2α dephosphorylation. This compound has been shown to be protective in clonal β cells expressing insulinAkita that cannot properly fold and causes severe ER stress, and in HeLa cells undergoing ER stress (18). This FDA-approved antihypertensive drug might therefore be β cell protective in type 2 diabetes. We examined here whether it holds that potential for β cells facing lipotoxic stress.

Materials and methods Cell culture: Clonal rat INS-1E cells (RRID:CVCL_0351, a kind gift from Dr C Wollheim, Centre Médical Endocrinology Universitaire, Geneva, Switzerland) were cultured in RPMI medium as described (19). Male Wistar rats (Charles River Laboratories) were housed and handled following the rules of the Endocrine Reviews Belgian Regulations for Animal Care. The experiments were approved by the Ethical Committee for Animal Experiments of the ULB. Rat islets were handpicked under a stereomicroscope after isolation by collagenase digestion and dispersed as previously described (20). The islets were cultured in Ham’s F10 (Invitrogen) containing 5% fetal bovine serum (FBS). Human islets from non-diabetic organ donors (5 males and 1 female, age 63±5 years, BMI 25±1 kg/m2) were isolated and cultured as previously described (21). The collection and handling of human islets were approved by the Ethical Committee of the University of Pisa, Pisa, Italy. Cell treatment and apoptosis assays: Guanabenz (Santa Cruz) or inactive guanabenz (a kind gift from Dr C Voisset, Université de Brest, Brest, France) was used at 50 µM unless otherwise indicated. INS-1E cells were exposed to FFAs in RPMI 1640 medium containing 0.75% FFA-free bovine serum albumin (BSA, ADVANCERoche) and 1% FBS. Rat and human islets were ARTICLE exposed to FFAs in the presence of 1% charcoal-absorbed BSA without FBS (22). Oleate and palmitate (Sigma) were dissolved ADVANCE ARTICLE: in 90% ethanol and diluted 1:100 to a final concentration of 0.5 mM (4;22). Cyclopiazonic acid (CPA) was used at 25 µM, brefeldin A at 0.1 µg/ml and tunicamycin at 5 µg/ml.

ADVANCE ARTICLE: Apoptotic cell death was detected by fluorescence microscopy after staining with DNA binding compounds Hoechst 33342 (5 µg/ml; Sigma) and propidium iodide (5 µg/ml) (5;23) by two investigators, one of whom was blinded for the experimental conditions. Cleaved caspase 3 and insulin double immunostaining was performed in dispersed rat islet cells as previously described (24).

2 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

Mouse studies: The mouse studies were approved by the Ethical Committee for Animal Experiments of the ULB. Male C57BL/6N mice (Janvier Labs) were fed regular diet (RD, 10% fat, Research Diets D12450B) or HFD (60% fat, Research Diets D12492). Guanabenz was administered by intraperitoneal injection (4 mg/kg guanabenz acetate, G110, Sigma-Aldrich, dissolved in 0.9% NaCl) every other day (25;26). An intraperitoneal glucose tolerance test (IPGTT) was performed as described (27) after one week of diet and treatment. Blood glucose levels were measured using Accu-Chek Aviva Nano (Roche) and plasma insulin was measured using the Ultra Sensitive Mouse Insulin ELISA (Crystal Chem). HOMA-IR was calculated as (fasting glucose (mg/dl) x fasting insulin (µU/ml)/(18 x 22.5)). The insulinogenic index was calculated as delta insulin/delta glucose between 0 and 15 minutes of the IPGTT. This measure of insulin secretion was divided by HOMA-IR to correct for the insulin sensitivity of the animal, in order to obtain a measure of β cell function. Protein translation: The SUnSET method was used to detect translation of nascent proteins (28). Just prior to collection in Laemmli buffer, cells were incubated for 30 min with 1 µM puromycin (Sigma). The rate of puromycin-labeled peptide formation reflects the overall protein synthesis rate (28). Western blotting: Protein detection was done using primary antibodies listed in Supplementary Table 1, horseradish peroxidase-conjugated secondary antibodies and SuperSignal West Femto chemiluminescence revealing reagent (Thermo Scientific). Immunoreactive bands were detected with a ChemiDoc XRS+ system and with Image Lab software (BIO-RAD). Glucose-stimulated insulin secretion: Mouse islet glucose-stimulated insulin secretion studies were done as described (27). Rat islets were exposed for 24h to 10 µM guanabenz, alone or in combination with 0.5 µM oleate Endocrinology in the presence of 1% charcoal-absorbed BSA. The islets were incubated in modified Krebs- Ringer bicarbonate HEPES solution for 30 minutes and insulin secretion was induced by 1h

Endocrine Reviews incubation in 1.67 or 16.7 mM glucose. Insulin was measured by ELISA (15). mRNA extraction and real time PCR: Poly(A)+ mRNA was isolated and reverse transcribed as described (7;29). Real time PCR was performed using Rotor-Gene SyBR Green on a Rotor-Gene Q cycler (Qiagen) or FastStart SYBR Green on a LightCycler (Roche) (7;30). Primers were used in a conventional PCR for preparing the standard. Gene expression was calculated as copies/µl (31). Expression levels were corrected for the reference genes GAPDH for rat and β-actin for human. These reference genes have been previously validated in β cells (22;23;32). Primer sequences are provided in Supplementary Table 2. CHOP luciferase assay: INS-1E cells were transfected with a CHOP promoter construct (33). 24h after transfection the cells were treated for 8 or 16h with palmitate alone or in combination with guanabenz. LuciferaseADVANCE activity was measured using the dual luc iferaseARTICLE reporter assay system (Promega) as described (34). ADVANCE ARTICLE: RNA interference: INS-1E cells were transfected overnight with 30 nM control siRNA (Qiagen) or siRNA ADVANCE ARTICLE: targeting CHOP (5) using Lipofectamine RNAiMAX (Invitrogen) as described previously (35).

3 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

Statistical analysis: The data are presented as box-plots or dot-plots. Boxes indicate lower quartile, median, and higher quartile; whiskers represent the full range of data points. The dots represent each experimental animal and the line indicates the mean. Comparisons were made by two-sided paired (or ratio) or unpaired t-test, as appropriate. P<0.05 was considered statistically significant.

Results Guanabenz potentiates ER stress-induced apoptosis in β cells. Guanabenz was previously shown to inhibit eIF2α dephosphorylation (18). We confirmed that in clonal insulin-producing INS-1E cells guanabenz per se tended to increase eIF2α phosphorylation, and it potentiated FFA-induced eIF2α phosphorylation (Figure 1A-B and Supplemental Figure 1A). We then examined whether guanabenz affects basal β cell survival using concentrations from 0.2 to 50 µM. At these concentrations, guanabenz induced low levels (6-8%) of apoptosis in INS-1E cells. The inactive form of guanabenz, inactivated by the replacement of one of the chlorines by fluorine (26), did not affect INS-1E cell survival (data not shown). Based on this dose-response study and previous reports (18), we selected 2 and 50 µM of guanabenz to test its cytoprotective effect in β cells undergoing ER stress. To this end, we exposed β cells to FFAs, which may elicit ER stress in type 2 diabetes, and to chemical ER stressors. As previously reported (5), the saturated FFA palmitate induced apoptosis in INS-1E cells, while the unsaturated FFA oleate induced much less apoptosis and the equimolar mixture of oleate and palmitate was non-toxic (Figure 1C). Guanabenz (50 µM) did not protect INS-1E cells from lipotoxicity, but instead sensitized the cells, in particular when exposed to oleate or the oleate/palmitate mixture (Figure 1C). A similar sensitization was seen with 2 µM guanabenz (Supplemental Figure 1B), but not with the inactive form of guanabenz (Supplemental Figure 1C). Guanabenz also potentiated FFA-induced cell death in rat islets and even more so in human islets (Figure 1D and 1E). The sensitizing effect was also Endocrinology seen in INS-1E cells exposed to the chemical ER stressors CPA and tunicamycin, but not brefeldin A (Supplemental Figure 1D). This was not the case for the inactive form of Endocrine Reviews guanabenz (Supplemental Figure 1E). The induction of apoptosis in INS-1E cells and rat islets was confirmed by increased caspase 3 cleavage, analyzed by Western blot (Figure 2A- B). We further confirmed the apoptosis to occur in β cells by insulin and cleaved caspase 3 immunostaining of dispersed rat islet cells treated with guanabenz alone or in combination with oleate (Figure 2C and Supplemental Figure 2). Guanabenz induces β cell dysfunction in mice. To investigate the impact of guanabenz in vivo, mice were treated with guanabenz for one week in combination with RD or HFD. After this short-term guanabenz treatment, we noticed a trend for increased cell death in HFD-fed mouse islets ex vivo but the drug did not affect islet cell survival in the RD-fed group (Figure 1F). The low rate of cell death detected ex vivo is probably due to the efficient clearance of apoptotic cells in vivo (36;37). Guanabenz treatment of RD-fed mice increased fasting blood glucose and insulin levels and HOMA-IR,ADVANCE indicating the development of insulin ARTICLEresistance (Supplemental Figure 3A-C). To investigate the impact of guanabenz on β cell function, we performed an IPGTT and ADVANCE ARTICLE: measured glucose and insulin levels (Supplemental Figure 3A-E). Guanabenz impaired glucose tolerance in RD-fed mice (Figure 3A). In a compensatory response to the insulin

ADVANCE ARTICLE: resistance, the guanabenz-treated mice were hyperinsulinemic (Figure 3B), but this increase failed to fully normalize glycemia (Figure 3A). In keeping with this, β cell function, calculated as the insulinogenic index corrected for HOMA-IR, was decreased by guanabenz (Figure 3C).

4 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

As expected, high fat feeding increased body weight, and it impaired glucose tolerance and induced hyperinsulinemia during the IPGTT (Figure 3A-B and D). There was little additional impact of guanabenz on measures of insulin sensitivity and insulin secretion, possibly because the treatment decreased food intake and body weight (Supplemental Figure 3F and Figure 3D). The decreased food intake could be due to the drug’s side effects, including drowsiness and nausea in humans (38) and decreased rotarod performance in mice (39). To directly assess β cell function, islets were isolated from the mice and ex vivo insulin secretion was measured. 16.7 mM glucose induced a 15-fold increase in insulin secretion in control islets (Figure 3E). HFD-fed mouse islets had increased basal insulin secretion and a lesser, 5-fold, response to high glucose. Guanabenz induced even higher basal insulin secretion and no absolute difference in response to high glucose (2-fold increase compared to basal insulin secretion, Figure 3E). We also examined the impact of guanabenz on β cell function in vitro. Guanabenz per se impaired rat islet glucose-stimulated insulin secretion (Figure 3F). In keeping with the ex vivo data, oleate exposure increased basal insulin secretion and reduced the glucose-stimulated response; the addition of guanabenz significantly worsened β cell dysfunction (Figure 3F). The decrease in glucose-stimulated insulin secretion was not due to insulin depletion, as guanabenz did not reduce the insulin content of mouse or rat islets (Supplemental Figure 3G-H). Guanabenz enhances FFA-induced PERK signaling. FFAs induce ER stress signaling in β cells (5). This was confirmed in a time course analysis of expression of the ER stress markers BiP, XBP1s, CHOP and GADD34 in FFA-exposed INS-1E cells (Figure 4). As previously reported, palmitate induces stronger signaling in the PERK and IRE1 branches of the ER stress response, while saturated and unsaturated FFAs similarly induce BiP expression (5). Guanabenz per se did not modify BiP, CHOP or GADD34 mRNA expression and slightly decreased XBP1s levels at the earliest time point (Figure 4). Guanabenz reduced BiP and Endocrinology XBP1s expression in FFA-exposed INS-1E cells at early time points. However, guanabenz markedly potentiated the FFA-induced expression of genes in the PERK pathway, namely Endocrine Reviews CHOP and GADD34, and this was true for both saturated and unsaturated FFAs (Figure 4). Guanabenz inhibits eIF2α dephosphorylation (18), and is thereby expected to attenuate protein translation. By measuring puromycin incorporation into elongating peptide chains, we observed that guanabenz decreases protein translation under basal and FFA conditions (Figure 5). Also downstream of eIF2α phosphorylation (Figure 1A-B and Supplemental Figure 1A), and in keeping with our mRNA expression studies, guanabenz potentiated FFA-induced CHOP protein expression (Figure 6A). CHOP induction was further confirmed using a CHOP promoter luciferase construct. Guanabenz per se induced CHOP promoter activity (8h, Supplemental Figure 4A) and potentiated palmitate-induced promoter activation (16h, Figure 6B). Signaling in the other branches of the ER stress response, assessed by measuring BiP and XBP1s protein expression, was not induced by guanabenz. The transient decrease in mRNA levels (Figure 4) resulted in lower XBP1s protein (Supplemental Figure 4B and D) but did not affect BiPADVANCE protein (Supplemental Figure 4C and E), probablyARTICLE due to the 46h-long half-life of BiP protein (40).

ADVANCE ARTICLE: Signal transduction was also assessed in human islets. Guanabenz did not affect XBP1s mRNA (Supplemental Figure 4F), suggesting that the IRE1 pathway does not mediate the

ADVANCE ARTICLE: sensitization of islet cells to apoptosis. The drug significantly increased CHOP mRNA expression by 2-fold under basal condition and 3-4-fold after FFA exposure (Figure 6C). The in vivo treatment of mice with guanabenz did not result in detectable changes in islet XBP1s protein expression or eIF2α phosphorylation (Supplemental Figure 4G-H) but it increased CHOP protein expression in the HFD-fed group (Figure 6D). The discrepancy

5 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

between eIF2α phosphorylation and CHOP expression may be due to dynamic regulation, by several kinases and phosphatases (41), of eIF2α phosphorylation. Detailed time course experiments are more difficult to perform in vivo compared to the in vitro models. Guanabenz potentiates FFA-induced β cell demise through CHOP. Based on the converging findings in these different models, we examined whether CHOP mediates guanabenz-induced apoptosis using RNA interference. An efficient CHOP knockdown of around 65% was achieved (Supplemental Figure 4I). CHOP silencing protected INS-1E cells from palmitate-induced apoptosis, as previously described (5), and significantly reduced the potentiating effect of guanabenz (Figure 6E). Taken together, these data show that guanabenz sensitizes β cells to lipotoxic apoptosis through the eIF2α/CHOP pathway.

Discussion Accumulating evidence suggests that ER stress contributes to β cell demise in type 2 diabetes (5-8). Saturated FFAs, the most common in man being palmitate, markedly activate the PERK branch of the ER stress response, leading to eIF2α phosphorylation and attenuation of translation (5). Translational attenuation is cytoprotective, as this reduces the protein load in the ER and lessens ER stress. Prolonged or intense PERK signaling, however, triggers apoptosis. Interest in finding drugs intervening in this pathway is high. Guanabenz, an α2- adrenergic agonist, was proposed to protect cells against ER stress-induced apoptosis (18). In stark contrast with the previous report, and with findings in cardiac myocytes (42) and retinal cells (43), we show here that guanabenz does not protect, but rather potentiates FFA-induced ER stress and β cell demise. The PERK pathway plays a major role in β cell survival and function. As in other cell types, the initial PERK response is aimed at reducing protein translation and relieving the stressed ER (44;45). Intense or prolonged eIF2α phosphorylation causes β cell dysfunction

Endocrinology and apoptosis (5;46;47). Guanabenz was shown to bind GADD34 and inhibit its binding to PP1 (18). We confirmed that guanabenz enhances eIF2α phosphorylation in FFA-treated β

Endocrine Reviews cells. In in vivo and in vitro models guanabenz treatment impairs β cell function, leading to hyperglycemia in mice. The latter is in keeping with an earlier report (48). Guanabenz upregulates expression of CHOP, a pro-apoptotic transcription factor (49;50) downstream of eIF2α phosphorylation, and potentiates FFA-induced β cell apoptosis. Using RNA interference we showed that CHOP mediates this β cell sensitization. The induction of insulin resistance (Supplemental Figure 3C) may contribute to the loss of glucose tolerance in guanabenz-treated animals. Previous studies have shown that eIF2α phosphorylation leads to insulin resistance in liver (51;52). Conversely, deletion of CHOP results in obesity but preserved insulin sensitivity (53). Insulin resistance is normally compensated for by increased insulin secretion, in both men and mice (54;55). Although guanabenz-treated mice were hyperinsulinemic, this was not enough to reduce glycemia to normal levels (Figure 3A). These findings are consistent with our previous work on salubrinal (56). Salubrinal was identifiedADVANCE in a high throughput screen for small molecules ARTICLE that protect cells from tunicamycin- induced ER stress, and shown to inhibit formation of GADD34/PP1 and CReP/PP1 ADVANCE ARTICLE: complexes (57). The synergistic activation of PERK-eIF2α signaling by salubrinal and FFAs led to inhibition of protein synthesis and insulin release, increased ATF4 and CHOP ADVANCE ARTICLE: expression and apoptosis, both in rodent β cells (56) and human islets (30). The guanabenz and salubrinal data concur with the phenotype of human PPP1R15B loss of function mutations (15): driving eIF2α phosphorylation either by pharmacological approaches or genetic causes results in β cell dysfunction and death.

6 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

Few studies have investigated whether guanabenz therapy affects insulin secretion and glucose tolerance in man. One small study did not find changes in insulinemia but it did not consider patients’ glycemia and insulin sensitivity (58). In another study of guanabenz-treated diabetic patients, no changes were seen in glucose control or anti-diabetic treatment needs over a mean follow-up of 7 months (59). Ideally, a potential link between guanabenz use and diabetes development/progression should be examined in large studies that couple prescription medicine registers with glycemic data in patients. In conclusion, we have demonstrated that guanabenz potentiates FFA-induced β cell dysfunction and apoptosis through enhanced signaling downstream of PERK. We caution that interventions aimed at modulating eIF2α phosphorylation in β cells should consider the sensitivity of these cells to any imbalance in this pathway.

Acknowledgments: The authors thank Isabelle Millard, Michael Pangerl, Anyishai Musuaya and Nathalie Pachera at the ULB Center for Diabetes Research (Université Libre de Bruxelles) for excellent technical support. We are also grateful to Decio L. Eizirik for his valuable input on the experimental design of this study and data discussions. This project has received funding from the European Union’s Horizon 2020 research and innovation programme, project T2DSystems, under grant agreement No 667191, the Fonds National de la Recherche Scientifique (FNRS), and Actions de Recherche Concertées de la Communauté Française (ARC), Belgium. B.A. was supported by a FNRS-FRIA fellowship, and by the Fonds David et Alice Van Buuren and Fondation Jaumotte-Demoulin. Corresponding author and person to whom reprint request should be addressed: Miriam Cnop, MD PhD, ULB Center for Diabetes Research, Université Libre de Bruxelles (ULB), Route de Lennik 808, CP-618, 1070 Brussels, Belgium, Phone: +32 2 555 6305, Fax: +32 2 555 6239, e-mail: [email protected]

Endocrinology Disclosure statement: The authors have nothing to disclose

Endocrine Reviews References 1. Kashyap S, Belfort R, Gastaldelli A, Pratipanawatr T, Berria R, Pratipanawatr W, Bajaj M, Mandarino L, DeFronzo R, Cusi K. A sustained increase in plasma free fatty acids impairs insulin secretion in nondiabetic subjects genetically predisposed to develop type 2 diabetes. Diabetes 2003 52:2461-2474 2. Sako Y, Grill VE. A 48-hour lipid infusion in the rat time-dependently inhibits glucose-induced insulin secretion and B cell oxidation through a process likely coupled to fatty acid oxidation. Endocrinology 1990 127:1580-1589 3. Zhou YP, Grill VE. Long-term exposure of rat pancreatic islets to fatty acids inhibits glucose-induced insulin secretion and biosynthesis through a glucose fatty acid cycle. J Clin Invest 1994 93:870-876 4. Cnop M, Hannaert JC, Hoorens A, Eizirik DL, Pipeleers DG. Inverse relationship between cytotoxicity of free fatty acids in pancreatic islet cells and cellular triglycerideADVANCE accumulation. Diabetes 2001 50:1771-17 77ARTICLE 5. Cunha DA, Hekerman P, Ladriere L, Bazarra-Castro A, Ortis F, Wakeham ADVANCE ARTICLE: MC, Moore F, Rasschaert J, Cardozo AK, Bellomo E, Overbergh L, Mathieu C, Lupi R, Hai T, Herchuelz A, Marchetti P, Rutter GA, Eizirik DL, Cnop M. Initiation and execution of ADVANCE ARTICLE: lipotoxic ER stress in pancreatic β-cells. J Cell Sci 2008 121:2308-2318 6. Karaskov E, Scott C, Zhang L, Teodoro T, Ravazzola M, Volchuk A. Chronic palmitate but not oleate exposure induces endoplasmic reticulum stress, which may contribute to INS-1 pancreatic β-cell apoptosis. Endocrinology 2006 147:3398-3407

7 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

7. Kharroubi I, Ladriere L, Cardozo AK, Dogusan Z, Cnop M, Eizirik DL. Free fatty acids and cytokines induce pancreatic β-cell apoptosis by different mechanisms: role of nuclear factor-κB and endoplasmic reticulum stress. Endocrinology 2004 145:5087-5096 8. Laybutt DR, Preston AM, Akerfeldt MC, Kench JG, Busch AK, Biankin AV, Biden TJ. Endoplasmic reticulum stress contributes to beta cell apoptosis in type 2 diabetes. Diabetologia 2007 50:752-763 9. Hartman MG, Lu D, Kim ML, Kociba GJ, Shukri T, Buteau J, Wang X, Frankel WL, Guttridge D, Prentki M, Grey ST, Ron D, Hai T. Role for activating transcription factor 3 in stress-induced β-cell apoptosis. Mol Cell Biol 2004 24:5721-5732 10. Huang CJ, Lin CY, Haataja L, Gurlo T, Butler AE, Rizza RA, Butler PC. High expression rates of human islet amyloid polypeptide induce endoplasmic reticulum stress mediated β-cell apoptosis, a characteristic of humans with type 2 but not type 1 diabetes. Diabetes 2007 56:2016-2027 11. Marchetti P, Bugliani M, Lupi R, Marselli L, Masini M, Boggi U, Filipponi F, Weir GC, Eizirik DL, Cnop M. The endoplasmic reticulum in pancreatic beta cells of type 2 diabetes patients. Diabetologia 2007 50:2486-2494 12. Han J, Back SH, Hur J, Lin YH, Gildersleeve R, Shan J, Yuan CL, Krokowski D, Wang S, Hatzoglou M, Kilberg MS, Sartor MA, Kaufman RJ. ER-stress-induced transcriptional regulation increases protein synthesis leading to cell death. Nat Cell Biol 2013 15:481-490 13. Schroder M, Kaufman RJ. Divergent roles of IRE1α and PERK in the unfolded protein response. Curr Mol Med 2006 6:5-36 14. Delepine M, Nicolino M, Barrett T, Golamaully M, Lathrop GM, Julier C. EIF2AK3, encoding translation initiation factor 2-α kinase 3, is mutated in patients with Wolcott-Rallison syndrome. Nat Genet 2000 25:406-409 15. Abdulkarim B, Nicolino M, Igoillo-Esteve M, Daures M, Romero S, Philippi A, Senee V, Lopes M, Cunha DA, Harding HP, Derbois C, Bendelac N, Hattersley AT, Endocrinology Eizirik DL, Ron D, Cnop M, Julier C. A missense mutation in PPP1R15B causes a syndrome including diabetes, short stature and microcephaly. Diabetes 2015 64:3951-3962 Endocrine Reviews 16. Scheuner D, Song B, McEwen E, Liu C, Laybutt R, Gillespie P, Saunders T, Bonner-Weir S, Kaufman RJ. Translational control is required for the unfolded protein response and in vivo glucose homeostasis. Mol Cell 2001 7:1165-1176 17. Scheuner D, Vander MD, Song B, Flamez D, Creemers JW, Tsukamoto K, Ribick M, Schuit FC, Kaufman RJ. Control of mRNA translation preserves endoplasmic reticulum function in beta cells and maintains glucose homeostasis. Nat Med 2005 11:757- 764 18. Tsaytler P, Harding HP, Ron D, Bertolotti A. Selective inhibition of a regulatory subunit of protein phosphatase 1 restores proteostasis. Science 2011 332:91-94 19. Ortis F, Cardozo AK, Crispim D, Storling J, Mandrup-Poulsen T, Eizirik DL. Cytokine-induced proapoptotic gene expression in insulin-producing cells is related to rapid, sustained, and nonoscillatory nuclear factor-κB activation. Mol Endocrinol 2006 20:1867- 1879 ADVANCE ARTICLE 20. Marroqui L, Masini M, Merino B, Grieco FA, Millard I, Dubois C, Quesada I, ADVANCE ARTICLE: Marchetti P, Cnop M, Eizirik DL. Pancreatic α cells are resistant to metabolic stress-induced apoptosis in type 2 diabetes. EBioMedicine 2015 2:378-385

ADVANCE ARTICLE: 21. Cnop M, Abdulkarim B, Bottu G, Cunha DA, Igoillo-Esteve M, Masini M, Turatsinze JV, Griebel T, Villate O, Santin I, Bugliani M, Ladriere L, Marselli L, McCarthy MI, Marchetti P, Sammeth M, Eizirik DL. RNA sequencing identifies dysregulation of the human pancreatic islet transcriptome by the saturated fatty acid palmitate. Diabetes 2014 63:1978-1993

8 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

22. Oliveira AF, Cunha DA, Ladriere L, Igoillo-Esteve M, Bugliani M, Marchetti P, Cnop M. In vitro use of free fatty acids bound to albumin: A comparison of protocols. Biotechniques 2015 58:228-233 23. Igoillo-Esteve M, Marselli L, Cunha DA, Ladriere L, Ortis F, Grieco FA, Dotta F, Weir GC, Marchetti P, Eizirik DL, Cnop M. Palmitate induces a pro-inflammatory response in human pancreatic islets that mimics CCL2 expression by beta cells in type 2 diabetes. Diabetologia 2010 53:1395-1405 24. Marroqui L, Dos Santos RS, Floyel T, Grieco FA, Santin I, Op de BA, Marselli L, Marchetti P, Pociot F, Eizirik DL. TYK2, a candidate gene for type 1 diabetes, modulates apoptosis and the innate immune response in human pancreatic β-cells. Diabetes 2015 64:3808-3817 25. Jiang HQ, Ren M, Jiang HZ, Wang J, Zhang J, Yin X, Wang SY, Qi Y, Wang XD, Feng HL. Guanabenz delays the onset of disease symptoms, extends lifespan, improves motor performance and attenuates motor neuron loss in the SOD1 G93A mouse model of amyotrophic lateral sclerosis. Neuroscience 2014 277:132-138 26. Tribouillard-Tanvier D, Dos RS, Gug F, Voisset C, Beringue V, Sabate R, Kikovska E, Talarek N, Bach S, Huang C, Desban N, Saupe SJ, Supattapone S, Thuret JY, Chedin S, Vilette D, Galons H, Sanyal S, Blondel M. Protein folding activity of ribosomal RNA is a selective target of two unrelated antiprion drugs. PLoS One 2008 3:e2174 27. Cunha DA, Igoillo-Esteve M, Gurzov EN, Germano CM, Naamane N, Marhfour I, Fukaya M, Vanderwinden JM, Gysemans C, Mathieu C, Marselli L, Marchetti P, Harding HP, Ron D, Eizirik DL, Cnop M. Death protein 5 and p53-upregulated modulator of apoptosis mediate the endoplasmic reticulum stress-mitochondrial dialog triggering lipotoxic rodent and human β-cell apoptosis. Diabetes 2012 61:2763-2775 28. Schmidt EK, Clavarino G, Ceppi M, Pierre P. SUnSET, a nonradioactive method to monitor protein synthesis. Nat Methods 2009 6:275-277 29. Rasschaert J, Ladriere L, Urbain M, Dogusan Z, Katabua B, Sato S, Akira S, Endocrinology Gysemans C, Mathieu C, Eizirik DL. Toll-like receptor 3 and STAT-1 contribute to double- stranded RNA+ interferon-γ-induced apoptosis in primary pancreatic β-cells. J Biol Chem Endocrine Reviews 2005 280:33984-33991 30. Ladriere L, Igoillo-Esteve M, Cunha DA, Brion JP, Bugliani M, Marchetti P, Eizirik DL, Cnop M. Enhanced signaling downstream of ribonucleic Acid-activated protein kinase-like endoplasmic reticulum kinase potentiates lipotoxic endoplasmic reticulum stress in human islets. J Clin Endocrinol Metab 2010 95:1442-1449 31. Overbergh L, Valckx D, Waer M, Mathieu C. Quantification of murine cytokine mRNAs using real time quantitative reverse transcriptase PCR. Cytokine 1999 11:305-312 32. Moore F, Colli ML, Cnop M, Esteve MI, Cardozo AK, Cunha DA, Bugliani M, Marchetti P, Eizirik DL. PTPN2, a candidate gene for type 1 diabetes, modulates interferon-γ-induced pancreatic β-cell apoptosis. Diabetes 2009 58:1283-1291 33. Pirot P, Ortis F, Cnop M, Ma Y, Hendershot LM, Eizirik DL, Cardozo AK. TranscriptionalADVANCE regulation of the endoplasmic reticulum ARTICLE stress gene chop in pancreatic insulin-producing cells. Diabetes 2007 56:1069-1077

ADVANCE ARTICLE: 34. Darville MI, Eizirik DL. Regulation by cytokines of the inducible nitric oxide synthase promoter in insulin-producing cells. Diabetologia 1998 41:1101-1108

ADVANCE ARTICLE: 35. Cnop M, Igoillo-Esteve M, Rai M, Begu A, Serroukh Y, Depondt C, Musuaya AE, Marhfour I, Ladriere L, Moles L, X, Lefkaditis D, Moore F, Brion JP, Cooper JM, Schapira AH, Clark A, Koeppen AH, Marchetti P, Pandolfo M, Eizirik DL, Fery F. Central role and mechanisms of β-cell dysfunction and death in friedreich ataxia-associated diabetes. Ann Neurol 2012 72:971-982

9 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

36. Hochreiter-Hufford A, Ravichandran KS. Clearing the dead: apoptotic cell sensing, recognition, engulfment, and digestion. Cold Spring Harb Perspect Biol 2013 5:a008748 37. Ravichandran KS, Lorenz U. Engulfment of apoptotic cells: signals for a good meal. Nat Rev Immunol 2007 7:964-974 38. Hall AH, Smolinske SC, Kulig KW, Rumack BH. Guanabenz overdose. Ann Intern Med 1985 102:787-788 39. Das I, Krzyzosiak A, Schneider K, Wrabetz L, D'Antonio M, Barry N, Sigurdardottir A, Bertolotti A. Preventing proteostasis diseases by selective inhibition of a phosphatase regulatory subunit. Science 2015 348:239-242 40. Rutkowski DT, Arnold SM, Miller CN, Wu J, Li J, Gunnison KM, Mori K, Sadighi Akha AA, Raden D, Kaufman RJ. Adaptation to ER stress is mediated by differential stabilities of pro-survival and pro-apoptotic mRNAs and proteins. PLoS Biol 2006 4:e374 41. Harding HP, Zhang Y, Zeng H, Novoa I, Lu PD, Calfon M, Sadri N, Yun C, Popko B, Paules R, Stojdl DF, Bell JC, Hettmann T, Leiden JM, Ron D. An integrated stress response regulates amino acid metabolism and resistance to oxidative stress. Mol Cell 2003 11:619-633 42. Neuber C, Uebeler J, Schulze T, Sotoud H, El-Armouche A, Eschenhagen T. Guanabenz interferes with ER stress and exerts protective effects in cardiac myocytes. PLoS One 2014 9:e98893 43. Mockel A, Obringer C, Hakvoort TB, Seeliger M, Lamers WH, Stoetzel C, Dollfus H, Marion V. Pharmacological modulation of the retinal unfolded protein response in Bardet-Biedl syndrome reduces apoptosis and preserves light detection ability. J Biol Chem 2012 287:37483-37494 44. Harding HP, Novoa I, Zhang Y, Zeng H, Wek R, Schapira M, Ron D. Regulated translation initiation controls stress-induced gene expression in mammalian cells. Mol Cell 2000 6:1099-1108 Endocrinology 45. Ron D, Walter P. Signal integration in the endoplasmic reticulum unfolded protein response. Nat Rev Mol Cell Biol 2007 8:519-529

Endocrine Reviews 46. Cnop M, Ladriere L, Igoillo-Esteve M, Moura RF, Cunha DA. Causes and cures for endoplasmic reticulum stress in lipotoxic β-cell dysfunction. Diabetes Obes Metab 2010 12 Suppl 2:76-82 47. Eizirik DL, Cardozo AK, Cnop M. The role for endoplasmic reticulum stress in diabetes mellitus. Endocr Rev 2008 29:42-61 48. Angel I, Bidet S, Langer SZ. Pharmacological characterization of the hyperglycemia induced by alpha-2 adrenoceptor agonists. J Pharmacol Exp Ther 1988 246:1098-1103 49. Oyadomari S, Koizumi A, Takeda K, Gotoh T, Akira S, Araki E, Mori M. Targeted disruption of the Chop gene delays endoplasmic reticulum stress-mediated diabetes. J Clin Invest 2002 109:525-532 50. Zinszner H, Kuroda M, Wang X, Batchvarova N, Lightfoot RT, Remotti H, StevensADVANCE JL, Ron D. CHOP is implicated in programmed ARTICLE cell death in response to impaired function of the endoplasmic reticulum. Genes Dev 1998 12:982-995

ADVANCE ARTICLE: 51. Li H, Zhou B, Liu J, Li F, Li Y, Kang X, Sun H, Wu S. Administration of progranulin (PGRN) triggers ER stress and impairs insulin sensitivity via PERK-eIF2α-

ADVANCE ARTICLE: dependent manner. Cell Cycle 2015 14:1893-1907 52. Oyadomari S, Harding HP, Zhang Y, Oyadomari M, Ron D. Dephosphorylation of translation initiation factor 2α enhances glucose tolerance and attenuates hepatosteatosis in mice. Cell Metab 2008 7:520-532

10 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

53. Maris M, Overbergh L, Gysemans C, Waget A, Cardozo AK, Verdrengh E, Cunha JP, Gotoh T, Cnop M, Eizirik DL, Burcelin R, Mathieu C. Deletion of C/EBP homologous protein (Chop) in C57Bl/6 mice dissociates obesity from insulin resistance. Diabetologia 2012 55:1167-1178 54. Ahren B, Pacini G. Insufficient islet compensation to insulin resistance vs. reduced glucose effectiveness in glucose-intolerant mice. Am J Physiol Endocrinol Metab 2002 283:E738-E744 55. Cnop M, Vidal J, Hull RL, Utzschneider KM, Carr DB, Schraw T, Scherer PE, Boyko EJ, Fujimoto WY, Kahn SE. Progressive loss of β-cell function leads to worsening glucose tolerance in first-degree relatives of subjects with type 2 diabetes. Diabetes Care 2007 30:677-682 56. Cnop M, Ladriere L, Hekerman P, Ortis F, Cardozo AK, Dogusan Z, Flamez D, Boyce M, Yuan J, Eizirik DL. Selective inhibition of eukaryotic translation initiation factor 2 α dephosphorylation potentiates fatty acid-induced endoplasmic reticulum stress and causes pancreatic β-cell dysfunction and apoptosis. J Biol Chem 2007 282:3989-3997 57. Boyce M, Bryant KF, Jousse C, Long K, Harding HP, Scheuner D, Kaufman RJ, Ma D, Coen DM, Ron D, Yuan J. A selective inhibitor of eIF2α dephosphorylation protects cells from ER stress. Science 2005 307:935-939 58. Eldridge JC, Strandhoy J, Buckalew VM, Jr. Endocrinologic effects of antihypertensive therapy with guanabenz or hydrochlorothiazide. J Cardiovasc Pharmacol 1984 6 Suppl 5:S776-S780 59. Weber MA, Drayer JI, Deitch MW. Hypertension in patients with diabetes mellitus: treatment with a centrally acting agent. J Cardiovasc Pharmacol 1984 6 Suppl 5:S823-S829

Figure 1: Guanabenz promotes eIF2α phosphorylation and sensitizes β cells to FFA- induced apoptosis. INS-1E cells (A-C), rat islets (D) and human islets (E) were exposed to Endocrinology 50 µM guanabenz (GA), alone or in combination with oleate (OL), palmitate (PAL) or a 1:1 mixture of oleate and palmitate (O/P) for 16h (A-B), 24h (C), 48h (D) and 72h (E). Islets were Endocrine Reviews isolated from C57BL/6N mice treated for 1 week with guanabenz or vehicle (Veh) in combination with regular diet (RD) or high fat diet (HFD) (F). Western blots for phosphorylated eIF2α were quantified by densitometry and corrected for total eIF2α (A-B). Apoptosis was measured by Hoechst 33342 and propidium iodide staining (C-E). The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of remaining data points (A-E). The dots represent individual animals and the line indicates the mean (F). n=4-14 independent experiments. *FFA vs control (CTL), #guanabenz vs DMSO, */#p<0.05, **/##p<0.01, ***p<0.001.

Figure 2: Guanabenz induces caspase 3 cleavage in β cells. INS-1E cells (A) and whole (B) or dispersed rat islets (C) were exposed to 50 µM guanabenz (GA), alone or in combination with oleate (OL), palmitate (PAL) or a 1:1 mixture of oleate and palmitate (O/P) for 16hADVANCE (A), 48h (B) and 24h (C). Western blot for cleaved ARTICLE caspase 3 and α-tubulin or β-actin, used as controls for protein loading (A-B). Following immunostaining for insulin and cleaved

ADVANCE ARTICLE: caspase 3, double positive rat islet cells were counted and expressed as percent of insulin positive cells (C). The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of remaining data points. n=4-6 independent experiments. *FFA vs ADVANCE ARTICLE: control (CTL), #guanabenz vs DMSO, */#p<0.05, **/##p<0.01.

Figure 3: Guanabenz induces β cell dysfunction and impairs glucose tolerance. Blood glucose (A) and plasma insulin levels (B) measured during an IPGTT of mice treated for 1

11 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

week with guanabenz (GA) or vehicle (Veh) while on regular (RD, black lines) or high fat diet (HFD, green lines). β cell function was calculated as delta insulin/delta glucose between 0 and 15 minutes of the IPGTT and corrected for HOMA-IR (C). Body weight at the end of treatment (D). Ex vivo islet insulin secretion at 1.67 or 16.7 mM glucose (n=9) (E). Rat islets were exposed to guanabenz alone or in combination with oleate for 24h (n=5) (F). Data are mean±SE (A-B). The dots represent individual animals, the line indicates the mean (C-D). The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of data points (E-F). £HFD vs RD (CTL), #guanabenz vs vehicle, *16.7 mM vs 1.67mM, §HFD/OL vs RD/CTL, */#/§/£p<0.05, **/##p<0.01, ###p<0.01.

Figure 4: Guanabenz potentiates FFA-induced CHOP and GADD34 expression. INS-1E cells were treated with 50 µM guanabenz (GA), alone or in combination with oleate (OL) (A), palmitate (PAL) (B) or a 1:1 mixture of oleate and palmitate (O/P) (C) for the indicated times. mRNA expression of BiP, XBP1s, CHOP and GADD34 was examined by real time PCR and corrected for the expression of the reference gene GAPDH. Data are presented as fold change of control, indicated by the dashed line. Results are mean±SE of 4 independent experiments. *FFA vs CTL, §FFA+GA vs GA, #GA vs DMSO, £FFA vs FFA+GA, */§/#/£p<0.05.

Figure 5: Guanabenz decreases protein translation. INS-1E cells were treated with 50 µM guanabenz (GA), alone or in combination with palmitate (PAL) for 16h and then incubated for 30 min with puromycin. The Western blot for puromycin is representative of 5 independent experiments, quantified by densitometry and corrected for β-actin in the graph. The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of remaining data points. #guanabenz vs DMSO, p<0.05.

Figure 6: Guanabenz potentiates FFA-induced CHOP expression leading to apoptosis. INS-1E cells were treated with 50 µM guanabenz (GA), alone or in combination with oleate

Endocrinology (OL), palmitate (PAL) or a 1:1 mixture of oleate and palmitate (O/P) for 16h (A-B). Western blots for CHOP were quantified by densitometry and corrected for α-tubulin (A). Blots are

Endocrine Reviews representative of 4 independent experiments. INS-1E cells transfected with a CHOP luciferase reporter construct were treated with guanabenz, alone or in combination with palmitate (B). CPA (25 µM) was used as a positive control. Human islets were treated with 50 µM guanabenz, alone or in combination with oleate or palmitate for 72h (C). CHOP mRNA expression was examined by real time PCR and corrected for the expression of the reference gene β-actin. Islets were isolated from C57BL/6N mice treated for 1 week with guanabenz or vehicle while on regular diet (RD) or high fat diet (HFD) (D). CHOP protein levels were corrected by α-tubulin and presented as fold of highest value. INS-1E cells transfected with a control siRNA (siCTL) or siRNA targeting CHOP (siCHOP) were treated with guanabenz, alone or in combination with palmitate for 16h (E). The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of remaining data points (A-C and E, n=4-5 independent experiments). The dots represent individual animals and the line indicates the mean (D). *FFA vs control (CTL), #guanabenz vs DMSO, */#p<0.05, **/##p<0.01.ADVANCE ARTICLE

ADVANCE ARTICLE:

ADVANCE ARTICLE:

12 Endocrinology; Copyright 2017 DOI: 10.1210/en.2016-1773

Antigen Manufacturer, catalog Species raised RRID Peptide/protein sequence #, and/or name of Dilution Name of Antibody in; monoclonal (required in target (if individual providing used or polyclonal revised MSs) known) the antibody Cell Signaling Rabbit; p-eIF2α NA Phospho-eIF2α (Ser51) 1/1000 AB_390740 Cat#:3597 Monoclonal Cell Signaling Rabbit; eIF2α NA eIF2α (D7D3) XP® 1/1000 AB_10692650 Cat#:5324 Monoclonal Cleaved Caspase Cell Signaling Rabbit; NA Cleaved Caspase-3 (Asp175) 1/1000 AB_2341188 3 Cat#:9661 Polyclonal Cell Signaling Rabbit; BiP NA BiP Antibody 1/2000 AB_10695864 Cat#:3183 Polyclonal Sigma-Aldrich Mouse; α-Tubulin NA Monoclonal Anti-α-Tubulin antibody 1/5000 AB_477593 Cat#:T9026 Monoclonal Santa Cruz Cat#:SC- Mouse; CHOP NA GADD 153 Antibody (B-3) 1/1000 AB_627411 7351 Monoclonal Santa Cruz Cat#:Sc- Rabbit; XBP1s NA XBP-1 Antibody (M-186) 1/1000 AB_794171 7160 Polyclonal Cell Signaling Rabbit; β-Actin NA beta-Actin Antibody 1/5000 AB_330288 Cat#:4967 Polyclonal Mouse; Puromycin NA Puromycin (3RH11) Kerafast Cat#:EQ0001 1/1000 AB_2620162 Monoclonal Guinea Pig; Insulin NA Insulin antibody Dako Cat#: A0564 1/200 AB_10013624 Polyclonal Peroxidase AffiniPure F(ab')2 Fragment Lucron Bioproducts Donkey; Anti-rabbit IgG NA 1/5000 AB_2340590 Donkey Anti-Rabbit IgG (H+L) Cat#:711-036-152 Polyclonal Peroxidase AffiniPure F(ab')2 Fragment Lucron Bioproducts Donkey; Anti-mouse IgG NA 1/5000 AB_2340773 Donkey Anti-Mouse IgG (H+L) Cat#:715-036-150 Polyclonal Goat anti-Rabbit IgG (H+L) Highly Cross- ThermoFisher Cat#:A- Goat; Anti-Rabbit IgG NA Adsorbed Secondary Antibody, Alexa Fluor 1/500 AB_2534094 11036 Polyclonal 568 Goat anti-Guinea Pig IgG (H+L) Highly ThermoFisher Cat#:A- Goat; Anti-Guinea pig NA Cross-Adsorbed Secondary Antibody, 1/500 AB_2534117 11073 Polyclonal Alexa Fluor 488

Endocrinology Endocrine Reviews

ADVANCE ARTICLE ADVANCE ARTICLE: ADVANCE ARTICLE:

13 ! (%

       

           $       

$

     

                    $#$'#&!& ""'   

                  Endocrinology      ""' ""' 

        

ADVANCE ARTICLE ADVANCE ARTICLE: "!)& 



           '%'  ))#"$     

))#"$  

   '%' 

     

    '%'  Endocrinology  ("$

   

    

 ("$

  



'%' 

ADVANCE  '%' $')#"$ $')#"$  ARTICLE ADVANCE ARTICLE: )'1.&

  3 !&( 3  !&(  3    3 3

  *2$&+)#+' %*  

*#/+#),/1*),,' +*  

       !&(     



   "&)'(0'

    

         + Endocrinology +            4  4 4

,/1*),/&$.&0)-, $-,0&,0 ,/1*),/&$.&0)-, $-,0&,0 !&(  !&(            

ADVANCE ARTICLE ADVANCE ARTICLE:  %#                  ' & '   &  & & '

    "   "

     

'

&   & &    '

    $"   $"       

   

   &

Endocrinology '    '   "  "       



' &  '     '  ' '  

 "   "         

ADVANCE !"%#$  !"%#$ ARTICLE !"%#$ ADVANCE ARTICLE: $"

     

      $"! % # #

 



$"! %  $"! % 

 #   Endocrinology

ADVANCE ARTICLE ADVANCE ARTICLE: )'1.%

              

 1#1*)+               1#1*)+      

 1$)&%."/%"$0)2)03

             !%(     

  $0)+  1#1*)+ 

   

  

Endocrinology         /)   /) 

    

-,-0,/)/ 

     ADVANCE ARTICLE ADVANCE ARTICLE: Discussion In the first part of my thesis I aimed to evaluate the global effects of the saturated FFA palmitate on the transcriptome of human islets. To this end, we performed RNA sequencing of 5 human islet preparations treated or not for 48 hours with palmitate. We detected 18,463 genes from which a total of 1,325 (7%) were significantly modified by palmitate, with 428 being up- and 897 being downregulated. Palmitate exposure induced a complex adaptive response in the islets, including upregulation of lipid metabolism, paralleled by inhibition of the Krebs cycle and oxidative phosphorylation. There were several signals of cellular stress responses, including cytosolic stress, mitochondrial quality control, and activation of an array of genes regulating the UPR and pathways of apoptosis. Surprisingly, there was also inhibition of genes regulating protein degradation and autophagy, which may aggravate ER stress by preventing disposal of misfolded proteins. One of the major pathways of the UPR to be upregulated was the PERK pathway, including upregulation of ATF4, ATF3 and GADD34 (Paper I). As previously mentioned, the ER stress response is mediated through several ER transmembrane proteins. PERK is one of the most important ER stress transducer in β cells. PERK activation has a direct effect on translation and its prolonged activation negatively affects cell survival, suggestive of a balancing role for PERK where it decides the cell fate depending on the extent of its activation. Also, as described above, mutations in several genes in the PERK- eIF2α regulation cause syndromes including diabetes (129;137).

Based on this background, the second aim of this thesis was to further clarify the role of eIF2α in β cell demise. To achieve this goal we first studied the role of loss-of- function mutation in PPP1R15B, encoding CReP, in 2 siblings from a consanguineous family. We were the first to demonstrate that CReP deficiency leads to a severe syndrome including young onset diabetes, microcephaly and short stature. The 2 patients studied developed diabetes at the age of 15 and 28, with low but detectable c peptide levels and they were both negative for autoantibodies (for a more thorough clinical description see paper II). The syndrome has since been confirmed in 2 other studies, although, both reports concern very young patients who do not have diabetes (for a comparison of the phenotypes see Table 2) (175;176). Given the relative young age of these patients and the age (15-28 years) of diabetes onset in our study, these patients might still develop diabetes in adolescence.

We demonstrated that the R658C mutation caused a significant loss of CReP ability to bind PP1 leading to an increase in eIF2α phosphorylation. The mutation exchanges an arginine for a cysteine in position 658 in the highly preserved PP1 binding site (121). The R658C mutation is disruptive to CReP function due to the important role of R658 in PP1 binding. Chen and colleagues showed that R658 side chain engages deep pocket on the surface of PP1 and forms a conserved salt bridge (Figure 8). Exchanging R658 for an alanine significantly weaken the ability of CReP to dephosphorylate eIF2α (128), which is further confirmed in our study.

84

A B

Figure 8: Interaction between CReP and PP1. (A) Cartoon interaction of CReP (R15B) in complex with PP1. (B) The conserved PP1 binding segment of CReP compared to PNUTS and Spinophillin, 2 other PP1 binding proteins. R658, indicated with arrow, forms a deep pocket salt bridge with PP1 stabilizing the interaction. (adapted from (128)).

We further showed that CReP deficiency appears to alter β cell function, since CReP silencing in β cells increased basal insulin secretion, but reduced glucose-simulated insulin release. We also found that CReP deficiency significantly reduces insulin content in β cells. More recently Akai and colleagues demonstrated that GADD34 and CReP are highly expressed in insulin producing cells in comparison to other cells types. They revealed that these two genes help in insulin biosynthesis by preventing eIF2α phosphorylation. They further showed that a transcription factor for insulin, namely neuronal differentiation 1 (NeuroD1), binds to the CReP and GADD34 promoters inducing their expression independently of PERK signaling. Both CReP and GADD34 silencing leads to a decrease in insulin production in mouse clonal β cells (MIN6) (177). Kloft et al also reported that CReP can act independently from PP1 by decorating vesicles inside cells, and that CReP deficiency decreased exocytosis of acetylcholinesterase in erythroleukemia cells (178). Although the role of CReP in insulin secretion needs to be further studied, these studies offer some explanation to the effects seen in CReP deficient β cells.

Our data further suggests that CReP silencing sensitized both clonal and primary rat β cells to apoptosis under basal and stress conditions. We demonstrated that CReP deficiency lead to activation of the intrinsic pathway of apoptosis through the BH3- only proteins DP5, PUMA and Bim. It has been previously demonstrated that DP5 and PUMA are downstream of ATF3 in the ISR (154). In our study we have shown

85 that CReP silencing increases ATF3 expression linking apoptosis to eIF2α phosphorylation.

To further examine the putative detrimental effects of eIF2α dysregulation in β cells, our second approach was to use guanabenz, a chemical inhibitor of GADD34. In previous reports, guanabenz has been suggested to be protective for β cells undergoing ER stress induced by insulinAkita (147). Thus, we speculated that guanabenz could protect β cells from lipotoxicity. In contrast, our findings suggest that guanabenz treatment potentiates lipotoxicity-induced ER stress and apoptosis both in rat and human islets. The reported role of guanabenz on eIF2α phosphorylation was confirmed, and this was accompanied by an increase of the downstream gene CHOP that may explain the proapoptotic effect of guanabenz in β cells. Indeed, CHOP silencing in clonal β cells significantly prevented the guanabenz- mediated potentiation of palmitate-induced apoptosis (paper III). We also demonstrated that guanabenz treatment in mice leads to insulin resistance and β cell dysfunction. Moreover, ex vivo experiments indicated that islets form guanabenz- treated mice on high fat diet have reduced glucose-induced insulin secretion compared to islets from vehicle treated animals on the same diet.

The guanabenz effect was reminiscent of salubrinal, a chemical inhibitor of eIF2α dephosphorylation identified through a high throughput screening and reported to protect cells against ER stress by inhibiting both GADD34-PP1 and CReP-PP1 complexes (146). Our group showed that treatment of rodent β cells and human islets with salubrinal in combination with FFAs led to a potentiation of FFA-induced PERK-eIF2α signaling, and increased expression of the downstream genes ATF4 and CHOP, resulting in reduced insulin release and apoptosis (69;179).

During the course of my PhD training I have identified the PERK/eIF2α pathway as one of the main contributors to pancreatic β cell demise. Taking all this together, it is evident that β cells are very sensitive to any imbalance in eIF2α phosphorylation. Thus, although it has been proposed that therapeutic interventions into this pathway, using salubrinal and guanabenz, will preserve β cells, more consideration for the sensitivity of β cells to imbalances in this pathway needs to be taken.

86

Table 2: Comparison of patient phenotype between different reports ofPPP1R15B mutation

Report Abdulkarim et al Kernohan et al Mohammad et al (Reference) (paper II) (175) (176) Individual Sibling 1 sibling 2 Sibling 1 sibling 2 Sibling 1 sibling 2 Designation

Consanguinity Yes Yes No between parents

Age (Years) 28 31 5 4 4 2 Compound Compound Homozygote Homozygote Homozygote Homozygote CReP Mutation heterozygote heterozygote R658C R658C R658C R658C G63A/C674X G63A/C674X Yes (age at Yes (age at onset: Diabetes mellitus No No No No onset:15 years) 28 years) Microcephaly Yes Yes Yes Yes Yes (Mild) Yes

Intellectual disability yes yes yes yes No yes

Short stature Yes Yes Yes Yes Yes yes

Delayed bone Bone morphology Bonedysplasia - - Wormianbones - age Language skills Delayed - Delayed Delayed Average Delayed Cirrhosis Neurogenic Tapering Holoprosencephaly deafness fingers Cirrhosis Other clinical Delayed brain Hypothyroidism Kyphoscoliosis - Long feet Holopocecephaly features myelination nonunited cervical Fine fingers and Delayed brain Anemia segments (C3-C7) toes myelination

87

Diseases caused by dysregulated ER stress signaling Several monogenic forms of diabetes are caused by mutations of genes involved in ER stress transduction, specifically the PERK pathway of the UPR. As already mentioned above, EIF2AK3 and DNAJC3 loss-of-function mutations cause syndromes including neonatal or young onset diabetes (129;137). During the course of my thesis, we identified a new monogenic form of diabetes caused by loss-of- function mutations of PPP1R15B (PAPER II). All these genes are highly involved in the ER stress response, suggesting dysregulation of this pathway is deleterious to β cells. Also, loss-of-function mutation in eIF2γ in the eIF2 complex leads to a similar syndrome including microcephaly and short stature and intellectual disability (180;181). These patients had hypoglycemia at early age and the surviving patient had chronic pancreatitis, most probably caused by β cell dysfunction and dysregulation (181). It is evident that both excessive and absent PERK signaling is equally deleterious to β cells.

ATF6 loss-of-function mutations have been reported to cause achromatopsia. The glycaemia of these patients has been reported to be normal (182;183). This suggests that ATF6, although it is a major ER stress response protein, might have compensatory pathways. However, WFS1 loss-of-function does cause diabetes which has been reported to be caused by the hyperactivation of ATF6α subunit. WFS1 has been shown to increase the ubiquitination of ATF6α subunit leading to its degradation (184). Furthermore, WFS1 has been shown to positively modulate ER Ca2+ levels and when deficient cause ER stress (185).

Similarly, loss-of-function mutations in IER3IP1 have been shown to cause a syndrome including short stature, microcephaly and neonatal diabetes (186-188). Although the exact function IER3IP1 is not known, it is believed to be an ER resident protein involved in ER stress response and apoptosis (187). Together, these different syndromes highlight the sensitivity of pancreatic β cells and neurons to dysregulation of the UPR.

Not only β cells It is worth mentioning that the syndromes resulting from mutations in different mediators of the PERK pathway have some common features besides diabetes suggesting that other cell types are equally sensitive to dysregulation of eIF2α. Mutations in PERK, P58IPK and CReP lead to multisystemic neurodegeneration with microcephaly. Although not in the PERK pathway, some features of WFS1 loss-of- function also mirrors those of the PERK pathway (129;136;137). Wolfram syndrome is characterized by optic atrophy and hearing loss due to neurodegeneration (44). Mutations in eIF2S3, encoding eIF2γ, also cause neurodegeneration and microcephaly (180;181). This common feature of these diseases can be explained due to the similarities between neurons and β cells. It has been reported that several genes selectively expressed in β cells are also present in neuronal tissue (189). Also, β cells do not express the neuronal gene suppressor Neuronal-Restrictive Silencer Factor (NRSF). NRSF is expressed in non-neuronal cells and inhibits the expression

88 of several neuronal genes such as type II voltage-dependent sodium channel (190). Furthermore, β cells and neurons also have several common transcriptional activators, e.g. Pax-6 (191;192). Given the importance of the ER to neurons, ER stress has been linked to several neurodegenretive diseases ,e.g. Alzheimer’s disease (193). Given all these common traits between β cells and neurons and their dependence on the ER, may explain the susceptibility of these two different tissues to dysregulations in the same pathway.

The patients with these syndromes also have short stature and, at least for PERK and CReP mutations, bone deformities. Although the similarities between bone tissue and β cells are limited, these diseases point to a role of PERK/eIF2α pathway in bone development. Together, these syndromes highlight the importance of a well- controlled ER stress in the neurons and β cells, but also they highlight the role of eIF2α in several other tissues including bones. Analyzing the expression values of these gene in the affected tissue, ubiquitous expression pattern over the different tissues can be seen with the highest expressions being in islets and β cells (Figure 9).

Figure 9: Expression values of the different genes causing monogenic form of diabetes: RNAseq values (in RPKM) from the indicated human tissues were obtained from GTEx (v4.p1) (194). RNA-seq data of FACS-purified human islet β-cells were from Nica et al (195). Human islet RNA-seq data were from Eizirik et al (196) and PAPER I.

89

Models of PERK/eIF2α pathway In vitro cell models used during this thesis contain clonal and primary rat β cells and human islets. Clonal cells are of course easy to culture and can be obtained in big amounts. However, since these cells are immortalized confirmation of the results needs to be done in primary cells. Human islets of course being a gold standard in diabetes research provide excellent view of the importance of these genes in human β cells.

Knocking down a gene to study its effect is a powerful tool to shed light on the importance of the gene in the cell. Thus we have used siRNA technology to silence the genes in question and study their role in β cell demise. However, this technique reduces the total expression of the protein, while the introduction of the mutation would better mimic the physiological state in the patient. Thus, CRISPR/Cas9 could be used in β cell lines, to introduce the mutations carried by the patients.

Other models under development in our lab take advantage of inducible pluripotent stem cells derived from patients. When differentiated into β cells, these cells could help uncover developmental problems and be a useful material for studying the effects of these mutations directly in patient derived cells. They could be used for apoptosis and function studies or transplanted into in vivo models to further clarify the sensitivity of these cells to in vivo stress conditions. Correction of the mutation is also possible using the CRISPR/Cas9 system, providing an excellent control for these experiments.

To study the roles of these genes in vivo the most common strategy taken is whole body knockout mouse models. As described previously the PERK-/- mouse develops hyperglycemia as a consequence of pancreatic endocrine dysfunction (139). On the other hand, the CReP-/- mouse dies shortly after birth. When born, these mice were small and pale due to low blood count. They had difficulty nursing causing them to die shortly after birth (145). The P58IPK-/- mice are also small in size and have diminished body fat. As mentioned before these mice develop diabetes due to β cell death (140;141). Comparing the phenotype of these mice to the one of patients it is clear that they have only some features in common. CReP and P58IPK models replicate the small size and the latter replicates the diabetes. Moreover, one of the patients reported by Mohammad and colleagues does have anemia which is also replicated in the CReP-/- mouse (176). However, these mice do not reproduce the overall syndrome of the patients. Given that CReP deletion is lethal in mice, a better model will be to introduce the loss-of-function mutation identified in our study in wildtype mice. We have shown that this mutation diminishes but does not completely abolish the CReP function (PAPER II). This may override the lethality allowing the mice to be used for further studies. One notable difference in our and Kernohan study compared to the report by Mohammad is the prevalence of anemia in the patients. The mutation discovered in our and the Kernohan study is in an arginine in position 658 crucial for PP1 binding. However, the compound heterozygous mutation discovered by Mohammad is in a glycine (position 63) and a cysteine (position 674).

90

As described previously, the whole body deletion of CReP in mice also leads to anemia, suggesting that CReP might have different functions in different tissues. Thus, a whole body knockout mouse model gives a more severe phenotype than observed in the patients (175;176).

The whole body PERK-/-, P58IPK-/- and the heterozygous eIF2αS51A mice all develop diabetes through β cell death (139-141;143). Nevertheless, generation of β cell specific knockout/transgenic mice would be a useful tool to study the effects of this pathway in β cells. This way, systemic effects of these genes will be avoided and their role in β cell development, function and survival can be clarified in vivo. This is achieved through the use of the Cre/lox system. Briefly, a portion of the gene of interest is flanked by two lox sites (floxed). Any DNA in between these lox sites will be excised by the Cre recombinase. By regulating the expression of Cre recombinase in a cell specific manner, e.g. by putting it under the control of insulin promoter, the deletion of the floxed DNA will only be allowed in that specific cell (197). As an example, endocrine cell specific PERK-/- mice have been generated and have shed light on the effects of PERK deletion in β cells. PERK loss-of-function was shown to lead to defective β cell proliferation and development in the fetus. Zhang and colleagues argue that PERK deficiency has little effect in the mature β cell and that the reduction in β cell mass arises from a lack of proliferation (198). In contrast, inducible PERK knockout in adult mice caused β cell death resulting in hyperglycemia (199). These studies together shed light on the importance of PERK in β cell development and survival. These observations were hampered by the systemic effects of PERK deletion in the whole body knockout mouse harboring multiple organ dysfunctions.

Treating ER stress Despite the previously discussed sensitivity of β cells to alterations in PERK/eIF2α pathway, drug therapies into ER stress in general have had some promising results. One of these approaches takes advantage of chemical chaperones TUDCA and PBA. The use of these chemicals in ob/ob mouse, a leptin deficient mouse model, normalized their insulin sensitivity and hyperglycemia (200). Along the same lines, treatment of cells overexpressing human islet amyloid polypeptide with these chemicals improved cell survival and insulin secretion in INS-1E cells (201).

Another approach to protecting β cells from ER stress is through the use of GLP-1 analogs. Yusta et. al. showed that activation of the GLP-1R leads to an accentuation of ER stress induced PERK pathway signaling. This leads to an increase in the ATF4-CHOP-GADD34 expression and faster recovery from translational attenuation and thus protection against ER stress induced cell death (202). However, our group has shown how the use of exendin and forskolin protects β cells against lipotoxicity through the induction of the chaperone BiP and the antiapoptotic gene JunB and not through the ATF4-CHOP-GADD34 pathway (71). Furthermore, the use of liraglutide, another GLP-1 analog was shown to protect cells against thapsigargin induced ER stress through the prevention of PERK and IRE1 activation. Also treatment of the

91 akita mouse improved their insulin sensitivity and this was attributed to a general decrease in ER stress response in the pancreas of these mice (203).

Other studies have used high throughput screens to identify molecules directly affecting ER stress response. One molecule identified through this approach is azoramide. This compound was shown to increase ER chaperone expression and eIF2α phosphorylation; however, it did not lead to the increase of proapoptotic CHOP expression. It was shown to protect the cells form tunicamycin induced ER stress and improve the glucose homeostasis and β cells function of the ob/ob mouse (204). Another compound with reported effects on ER stress is pioglitazone, an activator of transcription factor peroxisome proliferator-activated receptor (PPAR)-γ. Pioglitazone improved glucose tolerance and improved the islet proinsulin/insulin ratio of the NOD mice, a type 1 diabetes mouse model. Pioglitazone was shown to increase BiP and ATF4 expression in the islets of NOD mice and preventing the induction of the proapoptotic CHOP expression (205). Additionally, a rooibos extracted compound, enolic phenylpyruvic acid-2-Oglucoside (PPAG), has been shown to protect β cells from ER stress induced apoptosis. Although, alteration in different UPR mediators was not observed, PPAG improved glycaemia and β cells survival in HFD-fed mice. This was shown to be due to an increase in the antiapoptotic gene Bcl2 preventing the onset of the mitochondrial pathway of apoptosis (206). Thus, careful medical intervention into β cell ER stress response could help preserve β cell mass and function, thereby, delaying or preventing the onset of diabetes.

Conclusions and perspectives Using RNA sequencing we have acquired an unbiased view into the effects of palmitate in human islets. The UPR was one of the main cellular responses activated by palmitate including several mediators of PERK, e.g. ATF4 and GADD34. Given the reported role of PERK/eIF2α in β cell death as opposed to the protective effects of IRE1 and ATF6 (70;71), we further demonstrated the importance of a balanced PERK-eIF2α signaling for the proper function and survival of β cells. We showed through studies in both human monogenic forms of diabetes and through the use of in vitro and in vivo models that special care needs to be taken when considering PERK pathway as a target for drug development. Previously, salubrinal has been proposed as a potential therapy for β cells preservation due to its role in maintaining eIF2α phosphorylation, a process proposed to be beneficial for β cells (207). However, data from our own group showed that salubrinal is in fact toxic for rat and human β cells, suggesting that excessive eIF2α phosphorylation is poorly tolerated (69;179). This concept was confirmed by my own in vivo and in vitro experiments with guanabenz, and the study of a monogenic form of diabetes caused by CReP loss-of-function. Conversely, the data on the monogenic form of diabetes due to PERK loss-of-function mutations, and the eIF2αS51A mouse model, provide evidence that diminished phosphorylation is equally deleterious. Altogether these studies indicate that altered eIF2α phosphorylation is deleterious for pancreatic β cells, and that prolonged eIF2α phosphorylation potentiates lipotoxicity (Figure 10).

92

As mentioned before, we have provided substantial evidence showing that increased levels of phosphorylated eIF2α is deleterious for β cells. Previous studies into the PERK mutation and the eIF2αS51A mouse model also highlighted the sensitivity of pancreatic β cells to diminished eIF2α signaling. To further clarify the role of eIF2α in β cell function, survival and development, a new chemical compound, ISR inhibitor (ISRIB), recently shown to prevent eIF2αdownstream signaling (208) could be used. Initially the impact of ISRIB on β cell function and survival can be evaluated performing the same type of experiments used for guanabenz and salubrinal studies. In particular ISRIB treatments of clonal and primary β cells could be performed and β cell apoptosis evaluated through caspase 3 cleavage or FACS sorting. Also β cell function could be examined through both in vitro glucose stimulated insulin secretion and in vivo IPGTT of mice treated with the drug. Thereafter the role of eIF2α in the development of β cells could be assessed. To this end, pregnant mice can be treated over the course of the pregnancy and embryos can be collected at different stages to monitor β cell development. This can be done in parallel with an eIF2α enhancer such as guanabenz or salubrinal. Of course, given the relative novelty of ISRIB and its limited use in animals, the initial experiments would need to focus on evaluating the efficiency of ISRIB crossing the blood-placental barrier. This can be done through collection of amniotic fluid or embryonic blood of the mice and the use of high- performance liquid chromatography (HPLC) or liquid chromatography-mass spectrometry (LC-MS/MS).

Monogenic Diet Pharmacological diabetes (rich in saturated fat) intervention Figure 10: Prolonged eIF2α phosphorylation potentiates lipotoxicity FFA induced apoptosis. We have shown both through monogenic forms of CReP PERK Guanabenz diabetes affecting CReP and through the use of the chemical coumpound guanabenz, that β cells are particularly sensitive eIF2α-P to prolonged eIF2α phosphorylation. ATF3 CHOP

DP5/PUMA

Apoptosis

93

Another approach to take is to use ISRIB in mice carrying the mutation of CReP. As previously described, CReP-/- mice do not survive beyond the first day of birth. However generation of a mouse model carrying the R658C mutation seen in the patients could better mimic the phenotype of the patient. The use of ISRIB in this model would answer the question whether blocking of phosphorylated-eIF2α signaling could rescue the model. Firstly the model can be set up using the CRISPR/CAS9 system in embryos to produce mice carrying the specific mutation. Furthermore, the mouse model needs to be studied without any intervention to see whether the phenotype of the mouse model would truly mimic that of the patients. If the mouse replicates the patient phenotype in regards to β cell loss of function, neurodegeneration and developmental issues, it can be used together with ISRIB to answer the following main questions; 1. Since the patients develop diabetes between the ages of 16-28 years, would the use of ISRIB prevent the onset of diabetes through prevention of β cell loss of function? 2. The patients are already born with a severe phenotype including microcephaly and short stature; would the use of ISRIB in pregnant mice prevent these developmental issues? These experiments together would elucidate the role of eIF2α phosphorylation in tissue development with the main focus on pancreatic β cells.

We have previously seen that CReP silencing increases insulin secretion at basal levels of glucose but lose the ability to respond to high glucose. Since, CReP has been reported to be involved in vesicle transport and β cell dysfunction is a major contributor of diabetes development. Therefore, to investigate the cause, KCl stimulation of CReP deficient β cells would show the effects of CReP on vesicle transport in an ATP-independent manner. Moreover, β cells expressing GFP tagged WT or mutant CReP could be used to investigate the effects of the mutation on the localization of CReP and evaluate its role on vesicle transport through confocal microscopy. If vesicle localization of CReP is confirmed, the impact of CReP deficiency on vesicle trafficking can be studied through the use of total internal reflection fluorescence (TIRF) microscopy. This will allow us to analyze insulin granule secretion events in CReP-deficient β-cells. Using CRISPR/Cas9 system, a stable cell line containing the R658C mutation can be generated and used to study the long term effects of the mutation on β cell function of and the effect of heightened insulin secretion on the long term stability of the cell.

On a more long term note, given the involvement of several mediators and regulators of the PERK/eIF2α pathway in monogenic forms of diabetes, other mediators of this pathway should be considered while diagnosing patients with unknown forms of monogenic diabetes. This thesis also highlights that studies of monogenic form of diabetes can help to clarify the role of the affected genes in β cells and to understand the mechanisms of the more prevalent forms of diabetes. The heritability of type 2 diabetes can be anywhere between 30-70% based on twin and family studies (209;210). To date more than 90 different susceptibility loci have been identified to be

94 involved in type 2 diabetes. Several of these susceptibility genes have also been described in different monogenic forms of diabetes. Thus, studies of these rare occurring forms of diabetes can be considered as an experiment of nature to clarify the role of these genes in human islet development and function leading to a better understanding of the polygenic diabetes and possibly lead to more effective medical interventions.

Acknowledgements Firstly I would like to thank my supervisor Dr. Miriam Cnop for the opportunity she has given me and the time she has spent in teaching me and helping me develop. She has always been there to guide me through thick and thin and given me motivation and inspiration to achieve this goal.

I would also like to thank my co-supervisor Dr. Mariana Igoillo Esteve from the bottom of my heart. She has always been there to help me with any difficulty I faced, in both my professional and private life. She took me under her wings and provided me with every help I needed to develop. She also agreed to share an office with me and take care of my orchid when I was not there.

I thank all my co-authors and collaborators who contributed to the studies constituting this thesis. Their help and suggestions made these publications possible. With special thanks to Prof. Decio L. Eizirik for all his advice and guidance.

I would like to thank all my colleagues at the CDR for unforgettable years and all the help and support. Special thanks to Dr. Daniel Cunha for his wonderful help and directions.

I am extremely grateful and in a life time debt to my parents, Muhammed and Shokriya, for their constant support and love. Without their encouragement this thesis would never have happened. I hope I have managed to make you proud!

Thanks to my siblings Bzav, Badin, Bjin and Zeri for their constant support. I love you guys and you may call me the doctor now .

Last but not least, my undying love and gratitude to Dr. Flora Brozzi. My life companion and shining light. Thank you for the patience, support and love you have given me and for always being by my side.

The author was supported by FNRS-FRIA fellowship, and received short term fellowships from EMBO, EFSD and Wiener-Anspach foundation, and funding from Fonds David et Alice Van Buuren and FondationJaumotte-Demoulin.

95

References 1. International Diabetes Federation IDF Diabetes Atlas. 7th ed. Brussels, International Diabetes Federation Executive Office 2015 2. Eizirik DL, Colli ML, Ortis F. The role of inflammation in insulitis and beta-cell loss in type 1 diabetes. Nat Rev Endocrinol 2009 5:219-226 3. Hou JC, Min L, Pessin JE. Insulin granule biogenesis, trafficking and exocytosis. Vitam Horm 2009 80:473-506 4. Slingerland AS, Hattersley AT. Mutations in the Kir6.2 subunit of the KATP channel and permanent neonatal diabetes: new insights and new treatment. Ann Med 2005 37:186-195 5. Thomas I, Gregg B. Metformin; a review of its history and future: from lilac to longevity. Pediatr Diabetes 2017 18:10-16 6. 2. Classification and Diagnosis of Diabetes Diabetes Care 2017 40:S11-S24 7. Loghmani E. "Diabetes mellitis: Type 1 and type 2." Guidelines for adolescent nutrition services 2005; 167-182 8. Maritim AC, Sanders RA, Watkins JB, III. Diabetes, oxidative stress, and antioxidants: a review. J Biochem Mol Toxicol 2003 17:24-38 9. Mathers CD, Loncar D. Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med 2006 3:e442 10. World Health Organization (WHO). Fact sheet N°312. http://www.who.int/mediacentre/factsheets/fs312/en/ 2016 11. Dieleman JL, Baral R, Birger M, Bui AL, Bulchis A, Chapin A, Hamavid H, Horst C, Johnson EK, Joseph J, Lavado R, Lomsadze L, Reynolds A, Squires E, Campbell M, DeCenso B, Dicker D, Flaxman AD, Gabert R, Highfill T, Naghavi M, Nightingale N, Templin T, Tobias MI, Vos T, Murray CJ. US Spending on Personal Health Care and Public Health, 1996-2013. JAMA 2016 316:2627-2646 12. Rorsman P. The pancreatic beta-cell as a fuel sensor: an electrophysiologist's viewpoint. Diabetologia 1997 40:487-495 13. HELLMAN B. The frequency distribution of the number and volume of the islets Langerhans in man. I. Studies on non-diabetic adults. Acta Soc Med Ups 1959 64:432-460 14. Cabrera O, Berman DM, Kenyon NS, Ricordi C, Berggren PO, Caicedo A. The unique cytoarchitecture of human pancreatic islets has implications for islet cell function. Proc Natl Acad Sci U S A 2006 103:2334-2339 15. Rorsman P, Braun M. Regulation of insulin secretion in human pancreatic islets. Annu Rev Physiol 2013 75:155-179 16. De VA, Heimberg H, Quartier E, Huypens P, Bouwens L, Pipeleers D, Schuit F. Human and rat beta cells differ in glucose transporter but not in glucokinase gene expression. J Clin Invest 1995 96:2489-2495 17. Rorsman P, Eliasson L, Renstrom E, Gromada J, Barg S, Gopel S. The Cell Physiology of Biphasic Insulin Secretion. News Physiol Sci 2000 15:72-77 18. Dean PM, Matthews EK. Electrical activity in pancreatic islet cells. Nature 1968 219:389-390 19. Barg S, Eliasson L, Renstrom E, Rorsman P. A subset of 50 secretory granules in close contact with L-type Ca2+ channels accounts for first-phase insulin secretion in mouse beta-cells. Diabetes 2002 51 Suppl 1:S74-S82 20. Seino S. Cell signalling in insulin secretion: the molecular targets of ATP, cAMP and sulfonylurea. Diabetologia 2012 55:2096-2108

96

21. Gao J, Hirata M, Mizokami A, Zhao J, Takahashi I, Takeuchi H, Hirata M. Differential role of SNAP-25 phosphorylation by protein kinases A and C in the regulation of SNARE complex formation and exocytosis in PC12 cells. Cell Signal 2016 28:425-437 22. Song WJ, Seshadri M, Ashraf U, Mdluli T, Mondal P, Keil M, Azevedo M, Kirschner LS, Stratakis CA, Hussain MA. Snapin mediates incretin action and augments glucose-dependent insulin secretion. Cell Metab 2011 13:308-319 23. Shibasaki T, Takahashi T, Takahashi H, Seino S. Cooperation between cAMP signalling and sulfonylurea in insulin secretion. Diabetes Obes Metab 2014 16 Suppl 1:118-125 24. Shibasaki T, Takahashi H, Miki T, Sunaga Y, Matsumura K, Yamanaka M, Zhang C, Tamamoto A, Satoh T, Miyazaki J, Seino S. Essential role of Epac2/Rap1 signaling in regulation of insulin granule dynamics by cAMP. Proc Natl Acad Sci U S A 2007 104:19333-19338 25. Stumvoll M, Goldstein BJ, van Haeften TW. Type 2 diabetes: principles of pathogenesis and therapy. Lancet 2005 365:1333-1346 26. Kahn SE, Hull RL, Utzschneider KM. Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature 2006 444:840-846 27. Kahn SE, Prigeon RL, McCulloch DK, Boyko EJ, Bergman RN, Schwartz MW, Neifing JL, Ward WK, Beard JC, Palmer JP, . Quantification of the relationship between insulin sensitivity and beta-cell function in human subjects. Evidence for a hyperbolic function. Diabetes 1993 42:1663-1672 28. Marchetti P, Del GS, Marselli L, Lupi R, Masini M, Pollera M, Bugliani M, Boggi U, Vistoli F, Mosca F, Del PS. Pancreatic islets from type 2 diabetic patients have functional defects and increased apoptosis that are ameliorated by metformin. J Clin Endocrinol Metab 2004 89:5535-5541 29. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC. Beta-cell deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes 2003 52:102-110 30. Cnop M, Vidal J, Hull RL, Utzschneider KM, Carr DB, Schraw T, Scherer PE, Boyko EJ, Fujimoto WY, Kahn SE. Progressive loss of beta-cell function leads to worsening glucose tolerance in first-degree relatives of subjects with type 2 diabetes. Diabetes Care 2007 30:677-682 31. Rahier J, Guiot Y, Goebbels RM, Sempoux C, Henquin JC. Pancreatic beta- cell mass in European subjects with type 2 diabetes. Diabetes Obes Metab 2008 10 Suppl 4:32-42 32. Marchetti P, Del GS, Marselli L, Lupi R, Masini M, Pollera M, Bugliani M, Boggi U, Vistoli F, Mosca F, Del PS. Pancreatic islets from type 2 diabetic patients have functional defects and increased apoptosis that are ameliorated by metformin. J Clin Endocrinol Metab 2004 89:5535-5541 33. Butler PC, Meier JJ, Butler AE, Bhushan A. The replication of beta cells in normal physiology, in disease and for therapy. Nat Clin Pract Endocrinol Metab 2007 3:758-768 34. Cnop M, Hughes SJ, Igoillo-Esteve M, Hoppa MB, Sayyed F, van de LL, Gunter JH, de Koning EJ, Walls GV, Gray DW, Johnson PR, Hansen BC, Morris JF, Pipeleers-Marichal M, Cnop I, Clark A. The long lifespan and low turnover of human islet beta cells estimated by mathematical modelling of lipofuscin accumulation. Diabetologia 2010 53:321-330

97

35. Flannick J, Johansson S, Njolstad PR. Common and rare forms of diabetes mellitus: towards a continuum of diabetes subtypes. Nat Rev Endocrinol 2016 12:394-406 36. Vaxillaire M, Froguel P. Monogenic diabetes: Implementation of translational genomic research towards precision medicine. J Diabetes 2016 8:782-795 37. Vaxillaire M, Bonnefond A, Froguel P. The lessons of early-onset monogenic diabetes for the understanding of diabetes pathogenesis. Best Pract Res Clin Endocrinol Metab 2012 26:171-187 38. Stoffers DA, Zinkin NT, Stanojevic V, Clarke WL, Habener JF. Pancreatic agenesis attributable to a single nucleotide deletion in the human IPF1 gene coding sequence. Nat Genet 1997 15:106-110 39. Lango AH, Flanagan SE, Shaw-Smith C, De Franco E, Akerman I, Caswell R, Ferrer J, Hattersley AT, Ellard S. GATA6 haploinsufficiency causes pancreatic agenesis in humans. Nat Genet 2012 44:20-22 40. Stoy J, Edghill EL, Flanagan SE, Ye H, Paz VP, Pluzhnikov A, Below JE, Hayes MG, Cox NJ, Lipkind GM, Lipton RB, Greeley SA, Patch AM, Ellard S, Steiner DF, Hattersley AT, Philipson LH, Bell GI. Insulin gene mutations as a cause of permanent neonatal diabetes. Proc Natl Acad Sci U S A 2007 104:15040-15044 41. Yang Y, Chan L. Monogenic Diabetes: What It Teaches Us on the Common Forms of Type 1 and Type 2 Diabetes. Endocr Rev 2016 37:190-222 42. Fajans SS, Bell GI. MODY: history, genetics, pathophysiology, and clinical decision making. Diabetes Care 2011 34:1878-1884 43. Bonnefond A, Froguel P. Rare and common genetic events in type 2 diabetes: what should biologists know? Cell Metab 2015 21:357-368 44. Strom TM, Hortnagel K, Hofmann S, Gekeler F, Scharfe C, Rabl W, Gerbitz KD, Meitinger T. Diabetes insipidus, diabetes mellitus, optic atrophy and deafness (DIDMOAD) caused by mutations in a novel gene (wolframin) coding for a predicted . Hum Mol Genet 1998 7:2021-2028 45. Rhodes CJ. Type 2 diabetes-a matter of beta-cell life and death? Science 2005 307:380-384 46. Paolisso G, Tataranni PA, Foley JE, Bogardus C, Howard BV, Ravussin E. A high concentration of fasting plasma non-esterified fatty acids is a risk factor for the development of NIDDM. Diabetologia 1995 38:1213-1217 47. Wang L, Folsom AR, Zheng ZJ, Pankow JS, Eckfeldt JH. Plasma fatty acid composition and incidence of diabetes in middle-aged adults: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Clin Nutr 2003 78:91-98 48. Forouhi NG, Koulman A, Sharp SJ, Imamura F, Kroger J, Schulze MB, Crowe FL, Huerta JM, Guevara M, Beulens JW, van Woudenbergh GJ, Wang L, Summerhill K, Griffin JL, Feskens EJ, Amiano P, Boeing H, Clavel-Chapelon F, Dartois L, Fagherazzi G, Franks PW, Gonzalez C, Jakobsen MU, Kaaks R, Key TJ, Khaw KT, Kuhn T, Mattiello A, Nilsson PM, Overvad K, Pala V, Palli D, Quiros JR, Rolandsson O, Roswall N, Sacerdote C, Sanchez MJ, Slimani N, Spijkerman AM, Tjonneland A, Tormo MJ, Tumino R, van der AD, van der Schouw YT, Langenberg C, Riboli E, Wareham NJ. Differences in the prospective association between individual plasma phospholipid saturated fatty acids and incident type 2 diabetes: the EPIC-InterAct case-cohort study. Lancet Diabetes Endocrinol 2014 2:810-818

98

49. Giacca A, Xiao C, Oprescu AI, Carpentier AC, Lewis GF. Lipid-induced pancreatic beta-cell dysfunction: focus on in vivo studies. Am J Physiol Endocrinol Metab 2011 300:E255-E262 50. OPIE LH, WALFISH PG. Plasma free fatty acid concentrations in obesity. N Engl J Med 1963 268:757-760 51. Saifer A, Goldman L. The free fatty acids bound to human serum albumin. Journal of Lipid Research 1961 2:268-270 52. Richieri GV, Kleinfeld AM. Unbound free fatty acid levels in human serum. J Lipid Res 1995 36:229-240 53. Kleinfeld AM, Prothro D, Brown DL, Davis RC, Richieri GV, DeMaria A. Increases in serum unbound free fatty acid levels following coronary angioplasty. Am J Cardiol 1996 78:1350-1354 54. Oliveira AF, Cunha DA, Ladriere L, Igoillo-Esteve M, Bugliani M, Marchetti P, Cnop M. In vitro use of free fatty acids bound to albumin: A comparison of protocols. Biotechniques 2015 58:228-233 55. Gravena C, Mathias PC, Ashcroft SJ. Acute effects of fatty acids on insulin secretion from rat and human islets of Langerhans. J Endocrinol 2002 173:73- 80 56. Malaisse WJ, Malaisse-Lagae F, Sener A, Hellerstrom C. Participation of endogenous fatty acids in the secretory activity of the pancreatic B-cell. Biochem J 1985 227:995-1002 57. Haber EP, Ximenes HM, Procopio J, Carvalho CR, Curi R, Carpinelli AR. Pleiotropic effects of fatty acids on pancreatic beta-cells. J Cell Physiol 2003 194:1-12 58. Prentki M, Corkey BE. Are the beta-cell signaling molecules malonyl-CoA and cystolic long-chain acyl-CoA implicated in multiple tissue defects of obesity and NIDDM? Diabetes 1996 45:273-283 59. Itoh Y, Kawamata Y, Harada M, Kobayashi M, Fujii R, Fukusumi S, Ogi K, Hosoya M, Tanaka Y, Uejima H, Tanaka H, Maruyama M, Satoh R, Okubo S, Kizawa H, Komatsu H, Matsumura F, Noguchi Y, Shinohara T, Hinuma S, Fujisawa Y, Fujino M. Free fatty acids regulate insulin secretion from pancreatic beta cells through GPR40. Nature 2003 422:173-176 60. Chen C, Li H, Long YQ. GPR40 agonists for the treatment of type 2 diabetes mellitus: The biological characteristics and the chemical space. Bioorg Med Chem Lett 2016 26:5603-5612 61. Fujiwara K, Maekawa F, Yada T. Oleic acid interacts with GPR40 to induce Ca2+ signaling in rat islet beta-cells: mediation by PLC and L-type Ca2+ channel and link to insulin release. Am J Physiol Endocrinol Metab 2005 289:E670-E677 62. Tomita T, Hosoda K, Fujikura J, Inagaki N, Nakao K. The G-Protein-Coupled Long-Chain Fatty Acid Receptor GPR40 and Glucose Metabolism. Front Endocrinol (Lausanne) 2014 5:152 63. Cnop M. Fatty acids and glucolipotoxicity in the pathogenesis of Type 2 diabetes. Biochem Soc Trans 2008 36:348-352 64. Kashyap S, Belfort R, Gastaldelli A, Pratipanawatr T, Berria R, Pratipanawatr W, Bajaj M, Mandarino L, DeFronzo R, Cusi K. A sustained increase in plasma free fatty acids impairs insulin secretion in nondiabetic subjects genetically predisposed to develop type 2 diabetes. Diabetes 2003 52:2461- 2474

99

65. Paolisso G, Gambardella A, Amato L, Tortoriello R, D'Amore A, Varricchio M, D'Onofrio F. Opposite effects of short- and long-term fatty acid infusion on insulin secretion in healthy subjects. Diabetologia 1995 38:1295-1299 66. Zhou YP, Grill VE. Long-term exposure of rat pancreatic islets to fatty acids inhibits glucose-induced insulin secretion and biosynthesis through a glucose fatty acid cycle. J Clin Invest 1994 93:870-876 67. Eizirik DL, Cardozo AK, Cnop M. The role for endoplasmic reticulum stress in diabetes mellitus. Endocr Rev 2008 29:42-61 68. Ron D, Walter P. Signal integration in the endoplasmic reticulum unfolded protein response. Nat Rev Mol Cell Biol 2007 8:519-529 69. Cnop M, Ladriere L, Hekerman P, Ortis F, Cardozo AK, Dogusan Z, Flamez D, Boyce M, Yuan J, Eizirik DL. Selective inhibition of eukaryotic translation initiation factor 2  dephosphorylation potentiates fatty acid-induced endoplasmic reticulum stress and causes pancreatic -cell dysfunction and apoptosis. J Biol Chem 2007 282:3989-3997 70. Cunha DA, Hekerman P, Ladriere L, Bazarra-Castro A, Ortis F, Wakeham MC, Moore F, Rasschaert J, Cardozo AK, Bellomo E, Overbergh L, Mathieu C, Lupi R, Hai T, Herchuelz A, Marchetti P, Rutter GA, Eizirik DL, Cnop M. Initiation and execution of lipotoxic ER stress in pancreatic -cells. J Cell Sci 2008 121:2308-2318 71. Cunha DA, Ladriere L, Ortis F, Igoillo-Esteve M, Gurzov EN, Lupi R, Marchetti P, Eizirik DL, Cnop M. Glucagon-like peptide-1 agonists protect pancreatic beta-cells from lipotoxic endoplasmic reticulum stress through upregulation of BiP and JunB. Diabetes 2009 58:2851-2862 72. Kharroubi I, Ladriere L, Cardozo AK, Dogusan Z, Cnop M, Eizirik DL. Free fatty acids and cytokines induce pancreatic -cell apoptosis by different mechanisms: role of nuclear factor-B and endoplasmic reticulum stress. Endocrinology 2004 145:5087-5096 73. Pirot P, Ortis F, Cnop M, Ma Y, Hendershot LM, Eizirik DL, Cardozo AK. Transcriptional regulation of the endoplasmic reticulum stress gene chop in pancreatic insulin-producing cells. Diabetes 2007 56:1069-1077 74. Cnop M, Hannaert JC, Hoorens A, Eizirik DL, Pipeleers DG. Inverse relationship between cytotoxicity of free fatty acids in pancreatic islet cells and cellular triglyceride accumulation. Diabetes 2001 50:1771-1777 75. Gwiazda KS, Yang TL, Lin Y, Johnson JD. Effects of palmitate on ER and cytosolic Ca2+ homeostasis in beta-cells. Am J Physiol Endocrinol Metab 2009 296:E690-E701 76. Leekumjorn S, Wu Y, Sum AK, Chan C. Experimental and computational studies investigating trehalose protection of HepG2 cells from palmitate- induced toxicity. Biophys J 2008 94:2869-2883 77. Boslem E, Weir JM, MacIntosh G, Sue N, Cantley J, Meikle PJ, Biden TJ. Alteration of endoplasmic reticulum lipid rafts contributes to lipotoxicity in pancreatic beta-cells. J Biol Chem 2013 288:26569-26582 78. Preston AM, Gurisik E, Bartley C, Laybutt DR, Biden TJ. Reduced endoplasmic reticulum (ER)-to-Golgi protein trafficking contributes to ER stress in lipotoxic mouse beta cells by promoting protein overload. Diabetologia 2009 52:2369-2373 79. Volmer R, van der Ploeg K, Ron D. Membrane lipid saturation activates endoplasmic reticulum unfolded protein response transducers through their transmembrane domains. Proc Natl Acad Sci U S A 2013 110:4628-4633

100

80. Marmugi A, Parnis J, Chen X, Carmichael L, Hardy J, Mannan N, Marchetti P, Piemonti L, Bosco D, Johnson P, Shapiro JA, Cruciani-Guglielmacci C, Magnan C, Ibberson M, Thorens B, Valdivia HH, Rutter GA, Leclerc I. Sorcin Links Pancreatic beta-Cell Lipotoxicity to ER Ca2+ Stores. Diabetes 2016 65:1009-1021 81. Jeffrey KD, Alejandro EU, Luciani DS, Kalynyak TB, Hu X, Li H, Lin Y, Townsend RR, Polonsky KS, Johnson JD. Carboxypeptidase E mediates palmitate-induced beta-cell ER stress and apoptosis. Proc Natl Acad Sci U S A 2008 105:8452-8457 82. Cnop M, Foufelle F, Velloso LA. Endoplasmic reticulum stress, obesity and diabetes. Trends Mol Med 2012 18:59-68 83. Harding HP, Calfon M, Urano F, Novoa I, Ron D. Transcriptional and translational control in the Mammalian unfolded protein response. Annu Rev Cell Dev Biol 2002 18:575-599 84. Kaufman RJ. Stress signaling from the lumen of the endoplasmic reticulum: coordination of gene transcriptional and translational controls. Genes Dev 1999 13:1211-1233 85. Schroder M, Kaufman RJ. The mammalian unfolded protein response. Annu Rev Biochem 2005 74:739-789 86. Calfon M, Zeng H, Urano F, Till JH, Hubbard SR, Harding HP, Clark SG, Ron D. IRE1 couples endoplasmic reticulum load to secretory capacity by processing the XBP-1 mRNA. Nature 2002 415:92-96 87. Hetz C, Martinon F, Rodriguez D, Glimcher LH. The unfolded protein response: integrating stress signals through the stress sensor IRE1alpha. Physiol Rev 2011 91:1219-1243 88. Yoshida H, Matsui T, Yamamoto A, Okada T, Mori K. XBP1 mRNA is induced by ATF6 and spliced by IRE1 in response to ER stress to produce a highly active transcription factor. Cell 2001 107:881-891 89. Hollien J, Weissman JS. Decay of endoplasmic reticulum-localized mRNAs during the unfolded protein response. Science 2006 313:104-107 90. Maurel M, Chevet E, Tavernier J, Gerlo S. Getting RIDD of RNA: IRE1 in cell fate regulation. Trends Biochem Sci 2014 39:245-254 91. Pirot P, Naamane N, Libert F, Magnusson NE, Orntoft TF, Cardozo AK, Eizirik DL. Global profiling of genes modified by endoplasmic reticulum stress in pancreatic beta cells reveals the early degradation of insulin mRNAs. Diabetologia 2007 50:1006-1014 92. Urano F, Wang X, Bertolotti A, Zhang Y, Chung P, Harding HP, Ron D. Coupling of stress in the ER to activation of JNK protein kinases by transmembrane protein kinase IRE1. Science 2000 287:664-666 93. Shen J, Chen X, Hendershot L, Prywes R. ER stress regulation of ATF6 localization by dissociation of BiP/GRP78 binding and unmasking of Golgi localization signals. Dev Cell 2002 3:99-111 94. Harding HP, Zhang Y, Ron D. Protein translation and folding are coupled by an endoplasmic-reticulum-resident kinase. Nature 1999 397:271-274 95. Harding HP, Novoa I, Zhang Y, Zeng H, Wek R, Schapira M, Ron D. Regulated translation initiation controls stress-induced gene expression in mammalian cells. Mol Cell 2000 6:1099-1108 96. Bertolotti A, Zhang Y, Hendershot LM, Harding HP, Ron D. Dynamic interaction of BiP and ER stress transducers in the unfolded-protein response. Nat Cell Biol 2000 2:326-332

101

97. Liu CY, Xu Z, Kaufman RJ. Structure and intermolecular interactions of the luminal dimerization domain of human IRE1alpha. J Biol Chem 2003 278:17680-17687 98. Su Q, Wang S, Gao HQ, Kazemi S, Harding HP, Ron D, Koromilas AE. Modulation of the eukaryotic initiation factor 2 alpha-subunit kinase PERK by tyrosine phosphorylation. J Biol Chem 2008 283:469-475 99. Marciniak SJ, Garcia-Bonilla L, Hu J, Harding HP, Ron D. Activation- dependent substrate recruitment by the eukaryotic translation initiation factor 2 kinase PERK. J Cell Biol 2006 172:201-209 100. Cullinan SB, Zhang D, Hannink M, Arvisais E, Kaufman RJ, Diehl JA. Nrf2 is a direct PERK substrate and effector of PERK-dependent cell survival. Mol Cell Biol 2003 23:7198-7209 101. van Vliet AR, Giordano F, Gerlo S, Segura I, Van ES, Molenberghs G, Rocha S, Houcine A, Derua R, Verfaillie T, Vangindertael J, De KH, Waelkens E, Tavernier J, Hofkens J, Annaert W, Carmeliet P, Samali A, Mizuno H, Agostinis P. The ER Stress Sensor PERK Coordinates ER-Plasma Membrane Contact Site Formation through Interaction with Filamin-A and F-Actin Remodeling. Mol Cell 2017 65:885-899 102. Saheki Y, De CP. Endoplasmic Reticulum-Plasma Membrane Contact Sites. Annu Rev Biochem 2017 103. Stefan CJ, Manford AG, Emr SD. ER-PM connections: sites of information transfer and inter-organelle communication. Curr Opin Cell Biol 2013 25:434- 442 104. Harding HP, Zhang Y, Zeng H, Novoa I, Lu PD, Calfon M, Sadri N, Yun C, Popko B, Paules R, Stojdl DF, Bell JC, Hettmann T, Leiden JM, Ron D. An integrated stress response regulates amino acid metabolism and resistance to oxidative stress. Mol Cell 2003 11:619-633 105. Rutkowski DT, Kang SW, Goodman AG, Garrison JL, Taunton J, Katze MG, Kaufman RJ, Hegde RS. The role of p58IPK in protecting the stressed endoplasmic reticulum. Mol Biol Cell 2007 18:3681-3691 106. Yan W, Frank CL, Korth MJ, Sopher BL, Novoa I, Ron D, Katze MG. Control of PERK eIF2alpha kinase activity by the endoplasmic reticulum stress-induced molecular chaperone P58IPK. Proc Natl Acad Sci U S A 2002 99:15920- 15925 107. Lee TG, Tomita J, Hovanessian AG, Katze MG. Characterization and regulation of the 58,000-dalton cellular inhibitor of the interferon-induced, dsRNA-activated protein kinase. J Biol Chem 1992 267:14238-14243 108. Roobol A, Roobol J, Bastide A, Knight JR, Willis AE, Smales CM. p58IPK is an inhibitor of the eIF2alpha kinase GCN2 and its localization and expression underpin protein synthesis and ER processing capacity. Biochem J 2015 465:213-225 109. Jackson RJ, Hellen CU, Pestova TV. The mechanism of eukaryotic translation initiation and principles of its regulation. Nat Rev Mol Cell Biol 2010 11:113- 127 110. Voigts-Hoffmann F, Klinge S, Ban N. Structural insights into eukaryotic and the initiation of translation. Curr Opin Struct Biol 2012 22:768- 777 111. Schmitt E, Naveau M, Mechulam Y. Eukaryotic and initiation factor 2: a heterotrimeric tRNA carrier. FEBS Lett 2010 584:405-412

102

112. PhosphoSitePlus http://www.phosphosite.org/proteinAction.action?id=961&showAllSites=true 2017 113. Sudhakar A, Krishnamoorthy T, Jain A, Chatterjee U, Hasnain SE, Kaufman RJ, Ramaiah KV. Serine 48 in initiation factor 2 alpha (eIF2 alpha) is required for high-affinity interaction between eIF2 alpha(P) and eIF2B. Biochemistry 1999 38:15398-15405 114. Vattem KM, Wek RC. Reinitiation involving upstream ORFs regulates ATF4 mRNA translation in mammalian cells. Proc Natl Acad Sci U S A 2004 101:11269-11274 115. Lu PD, Harding HP, Ron D. Translation reinitiation at alternative open reading frames regulates gene expression in an integrated stress response. J Cell Biol 2004 167:27-33 116. Pakos-Zebrucka K, Koryga I, Mnich K, Ljujic M, Samali A, Gorman AM. The integrated stress response. EMBO Rep 2016 17:1374-1395 117. B'chir W, Maurin AC, Carraro V, Averous J, Jousse C, Muranishi Y, Parry L, Stepien G, Fafournoux P, Bruhat A. The eIF2alpha/ATF4 pathway is essential for stress-induced autophagy gene expression. Nucleic Acids Res 2013 41:7683-7699 118. Shang YY, Zhong M, Zhang LP, Guo ZX, Wang ZH, Zhang Y, Deng JT, Zhang W. Tribble 3, a novel oxidized low-density lipoprotein-inducible gene, is induced via the activating transcription factor 4-C/EBP homologous protein pathway. Clin Exp Pharmacol Physiol 2010 37:51-55 119. Ohoka N, Yoshii S, Hattori T, Onozaki K, Hayashi H. TRB3, a novel ER stress- inducible gene, is induced via ATF4-CHOP pathway and is involved in cell death. EMBO J 2005 24:1243-1255 120. Han J, Back SH, Hur J, Lin YH, Gildersleeve R, Shan J, Yuan CL, Krokowski D, Wang S, Hatzoglou M, Kilberg MS, Sartor MA, Kaufman RJ. ER-stress- induced transcriptional regulation increases protein synthesis leading to cell death. Nat Cell Biol 2013 15:481-490 121. Jousse C, Oyadomari S, Novoa I, Lu P, Zhang Y, Harding HP, Ron D. Inhibition of a constitutive translation initiation factor 2 phosphatase, CReP, promotes survival of stressed cells. J Cell Biol 2003 163:767-775 122. Peti W, Nairn AC, Page R. Folding of Intrinsically Disordered Protein Phosphatase 1 Regulatory Proteins. Curr Phys Chem 2012 2:107-114 123. Bollen M, Peti W, Ragusa MJ, Beullens M. The extended PP1 toolkit: designed to create specificity. Trends Biochem Sci 2010 35:450-458 124. Choy MS, Hieke M, Kumar GS, Lewis GR, Gonzalez-DeWhitt KR, Kessler RP, Stein BJ, Hessenberger M, Nairn AC, Peti W, Page R. Understanding the antagonism of retinoblastoma protein dephosphorylation by PNUTS provides insights into the PP1 regulatory code. Proc Natl Acad Sci U S A 2014 111:4097-4102 125. Heroes E, Lesage B, Gornemann J, Beullens M, Van ML, Bollen M. The PP1 binding code: a molecular-lego strategy that governs specificity. FEBS J 2013 280:584-595 126. Peti W, Nairn AC, Page R. Structural basis for protein phosphatase 1 regulation and specificity. FEBS J 2013 280:596-611 127. Chambers JE, Dalton LE, Clarke HJ, Malzer E, Dominicus CS, Patel V, Moorhead G, Ron D, Marciniak SJ. Actin dynamics tune the integrated stress

103

response by regulating eukaryotic initiation factor 2alpha dephosphorylation. Elife 2015 4:e04872 128. Chen R, Rato C, Yan Y, Crespillo-Casado A, Clarke HJ, Harding HP, Marciniak SJ, Read RJ, Ron D. G-actin provides substrate-specificity to eukaryotic initiation factor 2alpha holophosphatases. Elife 2015 4:e04871 129. Delepine M, Nicolino M, Barrett T, Golamaully M, Lathrop GM, Julier C. EIF2AK3, encoding translation initiation factor 2- kinase 3, is mutated in patients with Wolcott-Rallison syndrome. Nat Genet 2000 25:406-409 130. Inoue H, Tanizawa Y, Wasson J, Behn P, Kalidas K, Bernal-Mizrachi E, Mueckler M, Marshall H, Donis-Keller H, Crock P, Rogers D, Mikuni M, Kumashiro H, Higashi K, Sobue G, Oka Y, Permutt MA. A gene encoding a transmembrane protein is mutated in patients with diabetes mellitus and optic atrophy (Wolfram syndrome). Nat Genet 1998 20:143-148 131. Laybutt DR, Hawkins YC, Lock J, Lebet J, Sharma A, Bonner-Weir S, Weir GC. Influence of diabetes on the loss of beta cell differentiation after islet transplantation in rats. Diabetologia 2007 50:2117-2125 132. Marchetti P, Bugliani M, Lupi R, Marselli L, Masini M, Boggi U, Filipponi F, Weir GC, Eizirik DL, Cnop M. The endoplasmic reticulum in pancreatic beta cells of type 2 diabetes patients. Diabetologia 2007 50:2486-2494 133. Huang CJ, Lin CY, Haataja L, Gurlo T, Butler AE, Rizza RA, Butler PC. High expression rates of human islet amyloid polypeptide induce endoplasmic reticulum stress mediated -cell apoptosis, a characteristic of humans with type 2 but not type 1 diabetes. Diabetes 2007 56:2016-2027 134. Ozcan U, Cao Q, Yilmaz E, Lee AH, Iwakoshi NN, Ozdelen E, Tuncman G, Gorgun C, Glimcher LH, Hotamisligil GS. Endoplasmic reticulum stress links obesity, insulin action, and type 2 diabetes. Science 2004 306:457-461 135. Eizirik DL, Miani M, Cardozo AK. Signalling danger: endoplasmic reticulum stress and the unfolded protein response in pancreatic islet inflammation. Diabetologia 2013 56:234-241 136. Wolcott CD, Rallison ML. Infancy-onset diabetes mellitus and multiple epiphyseal dysplasia. J Pediatr 1972 80:292-297 137. Synofzik M, Haack TB, Kopajtich R, Gorza M, Rapaport D, Greiner M, Schonfeld C, Freiberg C, Schorr S, Holl RW, Gonzalez MA, Fritsche A, Fallier- Becker P, Zimmermann R, Strom TM, Meitinger T, Zuchner S, Schule R, Schols L, Prokisch H. Absence of BiP co-chaperone DNAJC3 causes diabetes mellitus and multisystemic neurodegeneration. Am J Hum Genet 2014 95:689- 697 138. Julier C, Nicolino M. Wolcott-Rallison syndrome. Orphanet J Rare Dis 2010 5:29 139. Harding HP, Zeng H, Zhang Y, Jungries R, Chung P, Plesken H, Sabatini DD, Ron D. Diabetes mellitus and exocrine pancreatic dysfunction in perk-/- mice reveals a role for translational control in secretory cell survival. Mol Cell 2001 7:1153-1163 140. Han J, Song B, Kim J, Kodali VK, Pottekat A, Wang M, Hassler J, Wang S, Pennathur S, Back SH, Katze MG, Kaufman RJ. Antioxidants Complement the Requirement for Protein Chaperone Function to Maintain beta-Cell Function and Glucose Homeostasis. Diabetes 2015 64:2892-2904 141. Ladiges WC, Knoblaugh SE, Morton JF, Korth MJ, Sopher BL, Baskin CR, MacAuley A, Goodman AG, LeBoeuf RC, Katze MG. Pancreatic beta-cell

104

failure and diabetes in mice with a deletion mutation of the endoplasmic reticulum molecular chaperone gene P58IPK. Diabetes 2005 54:1074-1081 142. Scheuner D, Song B, McEwen E, Liu C, Laybutt R, Gillespie P, Saunders T, Bonner-Weir S, Kaufman RJ. Translational control is required for the unfolded protein response and in vivo glucose homeostasis. Mol Cell 2001 7:1165-1176 143. Scheuner D, Vander MD, Song B, Flamez D, Creemers JW, Tsukamoto K, Ribick M, Schuit FC, Kaufman RJ. Control of mRNA translation preserves endoplasmic reticulum function in beta cells and maintains glucose homeostasis. Nat Med 2005 11:757-764 144. Novoa I, Zeng H, Harding HP, Ron D. Feedback inhibition of the unfolded protein response by GADD34-mediated dephosphorylation of eIF2alpha. J Cell Biol 2001 153:1011-1022 145. Harding HP, Zhang Y, Scheuner D, Chen JJ, Kaufman RJ, Ron D. Ppp1r15 gene knockout reveals an essential role for translation initiation factor 2 alpha (eIF2alpha) dephosphorylation in mammalian development. Proc Natl Acad Sci U S A 2009 106:1832-1837 146. Boyce M, Bryant KF, Jousse C, Long K, Harding HP, Scheuner D, Kaufman RJ, Ma D, Coen DM, Ron D, Yuan J. A selective inhibitor of eIF2 dephosphorylation protects cells from ER stress. Science 2005 307:935-939 147. Tsaytler P, Harding HP, Ron D, Bertolotti A. Selective inhibition of a regulatory subunit of protein phosphatase 1 restores proteostasis. Science 2011 332:91- 94 148. Izumi T, Yokota-Hashimoto H, Zhao S, Wang J, Halban PA, Takeuchi T. Dominant negative pathogenesis by mutant proinsulin in the Akita diabetic mouse. Diabetes 2003 52:409-416 149. Wang J, Takeuchi T, Tanaka S, Kubo SK, Kayo T, Lu D, Takata K, Koizumi A, Izumi T. A mutation in the insulin 2 gene induces diabetes with severe pancreatic beta-cell dysfunction in the Mody mouse. J Clin Invest 1999 103:27-37 150. Masuoka HC, Townes TM. Targeted disruption of the activating transcription factor 4 gene results in severe fetal anemia in mice. Blood 2002 99:736-745 151. Wang C, Huang Z, Du Y, Cheng Y, Chen S, Guo F. ATF4 regulates lipid metabolism and thermogenesis. Cell Res 2010 20:174-184 152. Oyadomari S, Koizumi A, Takeda K, Gotoh T, Akira S, Araki E, Mori M. Targeted disruption of the Chop gene delays endoplasmic reticulum stress- mediated diabetes. J Clin Invest 2002 109:525-532 153. Song B, Scheuner D, Ron D, Pennathur S, Kaufman RJ. Chop deletion reduces oxidative stress, improves beta cell function, and promotes cell survival in multiple mouse models of diabetes. J Clin Invest 2008 118:3378- 3389 154. Cunha DA, Igoillo-Esteve M, Gurzov EN, Germano CM, Naamane N, Marhfour I, Fukaya M, Vanderwinden JM, Gysemans C, Mathieu C, Marselli L, Marchetti P, Harding HP, Ron D, Eizirik DL, Cnop M. Death protein 5 and p53- upregulated modulator of apoptosis mediate the endoplasmic reticulum stress- mitochondrial dialog triggering lipotoxic rodent and human -cell apoptosis. Diabetes 2012 61:2763-2775 155. Hartman MG, Lu D, Kim ML, Kociba GJ, Shukri T, Buteau J, Wang X, Frankel WL, Guttridge D, Prentki M, Grey ST, Ron D, Hai T. Role for activating transcription factor 3 in stress-induced -cell apoptosis. Mol Cell Biol 2004 24:5721-5732

105

156. Duprez J, Jonas JC. Role of activating transcription factor 3 in low glucose- and thapsigargin-induced apoptosis in cultured mouse islets. Biochem Biophys Res Commun 2011 415:294-299 157. Zmuda EJ, Viapiano M, Grey ST, Hadley G, Garcia-Ocana A, Hai T. Deficiency of Atf3, an adaptive-response gene, protects islets and ameliorates inflammation in a syngeneic mouse transplantation model. Diabetologia 2010 53:1438-1450 158. Allen-Jennings AE, Hartman MG, Kociba GJ, Hai T. The roles of ATF3 in glucose homeostasis. A transgenic mouse model with liver dysfunction and defects in endocrine pancreas. J Biol Chem 2001 276:29507-29514 159. Tait SW, Green DR. Mitochondria and cell death: outer membrane permeabilization and beyond. Nat Rev Mol Cell Biol 2010 11:621-632 160. Hotchkiss RS, Strasser A, McDunn JE, Swanson PE. Cell death. N Engl J Med 2009 361:1570-1583 161. Gurzov EN, Eizirik DL. Bcl-2 proteins in diabetes: mitochondrial pathways of beta-cell death and dysfunction. Trends Cell Biol 2011 21:424-431 162. Igoillo-Esteve M, Gurgul-Convey E, Hu A, Romagueira Bichara Dos SL, Abdulkarim B, Chintawar S, Marselli L, Marchetti P, Jonas JC, Eizirik DL, Pandolfo M, Cnop M. Unveiling a common mechanism of apoptosis in beta- cells and neurons in Friedreich's ataxia. Hum Mol Genet 2015 24:2274-2286 163. Nogueira TC, Paula FM, Villate O, Colli ML, Moura RF, Cunha DA, Marselli L, Marchetti P, Cnop M, Julier C, Eizirik DL. GLIS3, a susceptibility gene for type 1 and type 2 diabetes, modulates pancreatic beta cell apoptosis via regulation of a splice variant of the BH3-only protein Bim. PLoS Genet 2013 9:e1003532 164. Villate O, Turatsinze JV, Mascali LG, Grieco FA, Nogueira TC, Cunha DA, Nardelli TR, Sammeth M, Salunkhe VA, Esguerra JL, Eliasson L, Marselli L, Marchetti P, Eizirik DL. Nova1 is a master regulator of alternative splicing in pancreatic beta cells. Nucleic Acids Res 2014 42:11818-11830 165. O'Connor L, Strasser A, O'Reilly LA, Hausmann G, Adams JM, Cory S, Huang DC. Bim: a novel member of the Bcl-2 family that promotes apoptosis. EMBO J 1998 17:384-395 166. Puthalakath H, Huang DC, O'Reilly LA, King SM, Strasser A. The proapoptotic activity of the Bcl-2 family member Bim is regulated by interaction with the dynein motor complex. Mol Cell 1999 3:287-296 167. Walter P, Ron D. The unfolded protein response: from stress pathway to homeostatic regulation. Science 2011 334:1081-1086 168. Lei K, Davis RJ. JNK phosphorylation of Bim-related members of the Bcl2 family induces Bax-dependent apoptosis. Proc Natl Acad Sci U S A 2003 100:2432-2437 169. Santin I, Moore F, Colli ML, Gurzov EN, Marselli L, Marchetti P, Eizirik DL. PTPN2, a candidate gene for type 1 diabetes, modulates pancreatic beta-cell apoptosis via regulation of the BH3-only protein Bim. Diabetes 2011 60:3279- 3288 170. Iurlaro R, Munoz-Pinedo C. Cell death induced by endoplasmic reticulum stress. FEBS J 2016 283:2640-2652 171. Gurzov EN, Ortis F, Cunha DA, Gosset G, Li M, Cardozo AK, Eizirik DL. Signaling by IL-1beta+IFN-gamma and ER stress converge on DP5/Hrk activation: a novel mechanism for pancreatic beta-cell apoptosis. Cell Death Differ 2009 16:1539-1550

106

172. Allagnat F, Cunha D, Moore F, Vanderwinden JM, Eizirik DL, Cardozo AK. Mcl-1 downregulation by pro-inflammatory cytokines and palmitate is an early event contributing to beta-cell apoptosis. Cell Death Differ 2011 18:328-337 173. Ramirez-Peinado S, Alcazar-Limones F, Lagares-Tena L, El MN, Caro- Maldonado A, Tirado OM, Munoz-Pinedo C. 2-deoxyglucose induces Noxa- dependent apoptosis in alveolar rhabdomyosarcoma. Cancer Res 2011 71:6796-6806 174. McCullough KD, Martindale JL, Klotz LO, Aw TY, Holbrook NJ. Gadd153 sensitizes cells to endoplasmic reticulum stress by down-regulating Bcl2 and perturbing the cellular redox state. Mol Cell Biol 2001 21:1249-1259 175. Kernohan KD, Tetreault M, Liwak-Muir U, Geraghty MT, Qin W, Venkateswaran S, Davila J, Holcik M, Majewski J, Richer J, Boycott KM. Homozygous mutation in the eukaryotic translation initiation factor 2alpha phosphatase gene, PPP1R15B, is associated with severe microcephaly, short stature and intellectual disability. Hum Mol Genet 2015 24:6293-6300 176. Mohammad S, Wolfe LA, Stobe P, Biskup S, Wainwright MS, Melin-Aldana H, Malladi P, Muenke M, Gahl WA, Whitington PF. Infantile Cirrhosis, Growth Impairment, and Neurodevelopmental Anomalies Associated with Deficiency of PPP1R15B. J Pediatr 2016 177. Akai R, Hosoda A, Yoshino M, Iwawaki T. Constitutive role of GADD34 and CReP in cancellation of phospho-eIF2alpha-dependent translational attenuation and insulin biosynthesis in pancreatic beta cells. Genes Cells 2015 20:871-886 178. Kloft N, Neukirch C, von HG, Bobkiewicz W, Weis S, Boller K, Husmann M. A subunit of eukaryotic translation initiation factor 2alpha-phosphatase (CreP/PPP1R15B) regulates membrane traffic. J Biol Chem 2012 287:35299- 35317 179. Ladriere L, Igoillo-Esteve M, Cunha DA, Brion JP, Bugliani M, Marchetti P, Eizirik DL, Cnop M. Enhanced signaling downstream of ribonucleic Acid- activated protein kinase-like endoplasmic reticulum kinase potentiates lipotoxic endoplasmic reticulum stress in human islets. J Clin Endocrinol Metab 2010 95:1442-1449 180. Borck G, Shin BS, Stiller B, Mimouni-Bloch A, Thiele H, Kim JR, Thakur M, Skinner C, Aschenbach L, Smirin-Yosef P, Har-Zahav A, Nurnberg G, Altmuller J, Frommolt P, Hofmann K, Konen O, Nurnberg P, Munnich A, Schwartz CE, Gothelf D, Colleaux L, Dever TE, Kubisch C, Basel-Vanagaite L. eIF2gamma mutation that disrupts eIF2 complex integrity links intellectual disability to impaired translation initiation. Mol Cell 2012 48:641-646 181. Moortgat S, Desir J, Benoit V, Boulanger S, Pendeville H, Nassogne MC, Lederer D, Maystadt I. Two novel EIF2S3 mutations associated with syndromic intellectual disability with severe microcephaly, growth retardation, and epilepsy. Am J Med Genet A 2016 170:2927-2933 182. Ansar M, Santos-Cortez RL, Saqib MA, Zulfiqar F, Lee K, Ashraf NM, Ullah E, Wang X, Sajid S, Khan FS, Amin-ud-Din M, Smith JD, Shendure J, Bamshad MJ, Nickerson DA, Hameed A, Riazuddin S, Ahmed ZM, Ahmad W, Leal SM. Mutation of ATF6 causes autosomal recessive achromatopsia. Hum Genet 2015 134:941-950 183. Kohl S, Zobor D, Chiang WC, Weisschuh N, Staller J, Gonzalez M, I, Chang S, Beck SC, Garcia GM, Sothilingam V, Seeliger MW, Stanzial F, Benedicenti F, Inzana F, Heon E, Vincent A, Beis J, Strom TM, Rudolph G, Roosing S,

107

Hollander AI, Cremers FP, Lopez I, Ren H, Moore AT, Webster AR, Michaelides M, Koenekoop RK, Zrenner E, Kaufman RJ, Tsang SH, Wissinger B, Lin JH. Mutations in the unfolded protein response regulator ATF6 cause the cone dysfunction disorder achromatopsia. Nat Genet 2015 47:757-765 184. Fonseca SG, Ishigaki S, Oslowski CM, Lu S, Lipson KL, Ghosh R, Hayashi E, Ishihara H, Oka Y, Permutt MA, Urano F. Wolfram syndrome 1 gene negatively regulates ER stress signaling in rodent and human cells. J Clin Invest 2010 120:744-755 185. Takei D, Ishihara H, Yamaguchi S, Yamada T, Tamura A, Katagiri H, Maruyama Y, Oka Y. WFS1 protein modulates the free Ca(2+) concentration in the endoplasmic reticulum. FEBS Lett 2006 580:5635-5640 186. Abdel-Salam GM, Schaffer AE, Zaki MS, Dixon-Salazar T, Mostafa IS, Afifi HH, Gleeson JG. A homozygous IER3IP1 mutation causes microcephaly with simplified gyral pattern, epilepsy, and permanent neonatal diabetes syndrome (MEDS). Am J Med Genet A 2012 158A:2788-2796 187. Poulton CJ, Schot R, Kia SK, Jones M, Verheijen FW, Venselaar H, de Wit MC, de GE, Bertoli-Avella AM, Mancini GM. Microcephaly with simplified gyration, epilepsy, and infantile diabetes linked to inappropriate apoptosis of neural progenitors. Am J Hum Genet 2011 89:265-276 188. Shalev SA, Tenenbaum-Rakover Y, Horovitz Y, Paz VP, Ye H, Carmody D, Highland HM, Boerwinkle E, Hanis CL, Muzny DM, Gibbs RA, Bell GI, Philipson LH, Greeley SA. Microcephaly, epilepsy, and neonatal diabetes due to compound heterozygous mutations in IER3IP1: insights into the natural history of a rare disorder. Pediatr Diabetes 2014 15:252-256 189. Martens GA, Jiang L, Hellemans KH, Stange G, Heimberg H, Nielsen FC, Sand O, Van HJ, Van LL, Schuit F, Gorus FK, Pipeleers DG. Clusters of conserved beta cell marker genes for assessment of beta cell phenotype. PLoS One 2011 6:e24134 190. Atouf F, Czernichow P, Scharfmann R. Expression of neuronal traits in pancreatic beta cells. Implication of neuron-restrictive silencing factor/repressor element silencing transcription factor, a neuron-restrictive silencer. J Biol Chem 1997 272:1929-1934 191. Turque N, Plaza S, Radvanyi F, Carriere C, Saule S. Pax-QNR/Pax-6, a paired box- and homeobox-containing gene expressed in neurons, is also expressed in pancreatic endocrine cells. Mol Endocrinol 1994 8:929-938 192. Arntfield ME, van der Kooy D. beta-Cell evolution: How the pancreas borrowed from the brain: The shared toolbox of genes expressed by neural and pancreatic endocrine cells may reflect their evolutionary relationship. Bioessays 2011 33:582-587 193. Li JQ, Yu JT, Jiang T, Tan L. Endoplasmic reticulum dysfunction in Alzheimer's disease. Mol Neurobiol 2015 51:383-395 194. The Genotype-Tissue Expression (GTEx) project Nat Genet 2013 45:580-585 195. Nica AC, Ongen H, Irminger JC, Bosco D, Berney T, Antonarakis SE, Halban PA, Dermitzakis ET. Cell-type, allelic, and genetic signatures in the human pancreatic beta cell transcriptome. Genome Res 2013 23:1554-1562 196. Eizirik DL, Sammeth M, Bouckenooghe T, Bottu G, Sisino G, Igoillo-Esteve M, Ortis F, Santin I, Colli ML, Barthson J, Bouwens L, Hughes L, Gregory L, Lunter G, Marselli L, Marchetti P, McCarthy MI, Cnop M. The human pancreatic islet transcriptome: expression of candidate genes for type 1

108

diabetes and the impact of pro-inflammatory cytokines. PLoS Genet 2012 8:e1002552 197. Carter M, Shieh JC 2010 Chapter 11 - Making and Using Transgenic Organisms. Guide to Research Techniques in Neuroscience. New York: Academic Press; 243-262 198. Zhang W, Feng D, Li Y, Iida K, McGrath B, Cavener DR. PERK EIF2AK3 control of pancreatic beta cell differentiation and proliferation is required for postnatal glucose homeostasis. Cell Metab 2006 4:491-497 199. Gao Y, Sartori DJ, Li C, Yu QC, Kushner JA, Simon MC, Diehl JA. PERK is required in the adult pancreas and is essential for maintenance of glucose homeostasis. Mol Cell Biol 2012 32:5129-5139 200. Ozcan U, Yilmaz E, Ozcan L, Furuhashi M, Vaillancourt E, Smith RO, Gorgun CZ, Hotamisligil GS. Chemical chaperones reduce ER stress and restore glucose homeostasis in a mouse model of type 2 diabetes. Science 2006 313:1137-1140 201. Cadavez L, Montane J, Alcarraz-Vizan G, Visa M, Vidal-Fabrega L, Servitja JM, Novials A. Chaperones ameliorate beta cell dysfunction associated with human islet amyloid polypeptide overexpression. PLoS One 2014 9:e101797 202. Yusta B, Baggio LL, Estall JL, Koehler JA, Holland DP, Li H, Pipeleers D, Ling Z, Drucker DJ. GLP-1 receptor activation improves beta cell function and survival following induction of endoplasmic reticulum stress. Cell Metab 2006 4:391-406 203. Zhao L, Guo H, Chen H, Petersen RB, Zheng L, Peng A, Huang K. Effect of Liraglutide on endoplasmic reticulum stress in diabetes. Biochem Biophys Res Commun 2013 441:133-138 204. Fu S, Yalcin A, Lee GY, Li P, Fan J, Arruda AP, Pers BM, Yilmaz M, Eguchi K, Hotamisligil GS. Phenotypic assays identify azoramide as a small-molecule modulator of the unfolded protein response with antidiabetic activity. Sci Transl Med 2015 7:292ra98 205. Maganti AV, Tersey SA, Syed F, Nelson JB, Colvin SC, Maier B, Mirmira RG. Peroxisome Proliferator-activated Receptor-gamma Activation Augments the beta-Cell Unfolded Protein Response and Rescues Early Glycemic Deterioration and beta Cell Death in Non-obese Diabetic Mice. J Biol Chem 2016 291:22524-22533 206. Mathijs I, Da Cunha DA, Himpe E, Ladriere L, Chellan N, Roux CR, Joubert E, Muller C, Cnop M, Louw J, Bouwens L. Phenylpropenoic acid glucoside augments pancreatic beta cell mass in high-fat diet-fed mice and protects beta cells from ER stress-induced apoptosis. Mol Nutr Food Res 2014 58:1980- 1990 207. Wiseman RL, Balch WE. A new pharmacology--drugging stressed folding pathways. Trends Mol Med 2005 11:347-350 208. Sidrauski C, Acosta-Alvear D, Khoutorsky A, Vedantham P, Hearn BR, Li H, Gamache K, Gallagher CM, Ang KK, Wilson C, Okreglak V, Ashkenazi A, Hann B, Nader K, Arkin MR, Renslo AR, Sonenberg N, Walter P. Pharmacological brake-release of mRNA translation enhances cognitive memory. Elife 2013 2:e00498 209. Almgren P, Lehtovirta M, Isomaa B, Sarelin L, Taskinen MR, Lyssenko V, Tuomi T, Groop L. Heritability and familiality of type 2 diabetes and related quantitative traits in the Botnia Study. Diabetologia 2011 54:2811-2819

109

210. Poulsen P, Kyvik KO, Vaag A, Beck-Nielsen H. Heritability of type II (non- insulin-dependent) diabetes mellitus and abnormal glucose tolerance--a population-based twin study. Diabetologia 1999 42:139-145

110

Supplementary data PAPER I

Cnop M, Abdulkarim B, Bottu G, Cunha DA, Masini M, Turatsinze JV, Griebel T, Igoillo-Esteve M, Bugliani M, Villate O, Ladriere L, Marselli L, Marchetti P, McCarthy MI, Sammeth M, Eizirik DL; RNA-sequencing identifies dysregulation of the human pancreatic islet transcriptome by the saturated fatty acid palmitate.

111

SUPPLEMENTARY DATA

Supplementary Table 1. Characteristics of the organ donors and human islet preparations used for RNA-seq and independent confirmation and mechanistic studies.

Gender Age BMI Cause of death Purity (years) (kg/m2) (%) F 77 23.8 Trauma 45 M 36 26.3 CVD 51 M 77 25.2 CVD 62 F 46 22.5 CVD 60 M 40 26.2 Trauma 34 M 59 26.7 NA 58 M 51 26.2 Trauma 54 F 79 29.7 CH 21 M 68 27.5 CH 42 F 76 25.4 CH 30 F 75 29.4 CVD 24 F 73 30.0 CVD 16 M 63 NA NA 46 F 64 23.4 CH 76 M 69 25.1 CH 68 F 23 19.7 Trauma 70 M 47 27.7 CVD 48 F 65 24.6 CH 58 F 87 21.5 Trauma 61 F 72 23.9 CH 62 M 69 25 CVD 85 M 85 25.5 CH 39 M 59 27.7 Trauma 56 F 76 19.5 CH 35 F 50 20.2 CH 70 F 42 23 CVD 48 M 52 24.5 CH 60 F 79 27.5 CH 89 M 56 24.7 Cerebral ischemia 47 M 69 24.2 CVD 57 F 79 28.1 Trauma 61 M 79 23.7 NA 13 M 82 23 CH 61 M 32 NA NA 75 F 23 22.5 Cardiac arrest 46 M 51 NA Trauma 37 Abbreviations: F: Female; M: Male; BMI: Body mass index; CVD: Cardiovascular disease; CH: Cerebral hemorrhage. Purity indicates the percentage of -cells in the human islet preparations as determined by immunostaining for insulin. The first five preparations were used for RNA-seq, the others for confirmation and mechanistic studies.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 2. Primer sequences.

Gene Species RefSeq ID STD Forward primer Reverse primer Product or length qRT (bp) ACTB Homo sapiens NM_001101 STD AAATCTGGCACCACACCTTC CCGATCCACACGGAGTAC 805 TT qRT CTGTACGCCAACACAGTGC GCTCAGGAGGAGCAATGA 127 T TC CCL2 Homo sapiens NM_002982 STD TTCTGTGCCTGCTGCTCATA GTCTTCGGAGTTTGGGTTT 277 G qRT AGCAAGTGTCCCAAAGAAG CATGGAATCCTGAACCCAC 93 C T GAPDH Rattus NM_017008 STD ATGACTCTACCCACGGCAA TGTGAGGGAGATGCTCAG 975 norvegicus G TG qRT AGTTCAACGGCACAGTCAA TACTCAGCACCAGCATCAC 118 G C IL1B Homo sapiens NM_000576 STD GCTGAGGAAGATGCTGGTT TTCTGCTTGAGAGGTGCTG 514 C A qRT TCCAGGGACAGGATATGGA TCTTTCAACACGCAGGACA 133 G G IL6 Homo sapiens NM_000600 STD AGTACCCCCAGGAGAAGAT TACTCATCTGCACAGCTCT 354 T G qRT AAAAGATGGCTGAAAAAGAT CTACTCTCAAATCTGTTCT 129 GG GG IL8 Homo sapiens NM_000584 STDAGGAAGAAACCACCGGAAG TCTTCAAAAACTTCTCCAC 325 AAC qRT TGTAAACATGACTTCCAAGC TTGGAGTATGTCTTTATGC 131 T AC MALAT1 Homo sapiens NR_002819 STD GCTTGAGGAAACCGCAGAT TTCTTCGCCTTCCCGTACT 603 A T qRT GACGGAGGTTGAGATGAAG ATTCGGGGCTCTGTAGTC 84 C CT GATA6 Homo sapiens NM_005257 STD ACCTGCTGGAGGACCTGTC ATACTTGAGCTCGCTGTTC 504 TC

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

qRT AGACCACTTGCTATGAAAAA TCATGGGAATGGAATTATT 109 AG GC GATA6 Rattus NM_019185.1 STD TGCCAACCCTGAGAACAGT TGGAAGCAGACCCAGGCT 372 norvegicus GACC GACA qRT CGGTGCGACAGGATTCTTG TTTGCCGCCATCTGGACTG 117 GTGT CT KCNK16 Rattus NM_00110952 STD ATATCTGCTACCTGCTGCTT ACGTGGCTGAAGAACATG 520 norvegicus 0 G GG qRT AAAGGCAACTCCACCAACC TAGAAGACACAGAAGACCT 134 G KCNK16 Homo sapiens NM_00113510 STD AGAACTACACCTGCCTGGA TAGTCCCCAAAGCCAATG 486 5, qRT C GTG 107 NM_00113510 AGGCACAGTCGTCACTACC TTGAGGAAGATCACGTTAA 6, GC NM_00113510 7, NM_032115 CREB3 Homo sapiens NM_006368.4 STD CACCCTTTCCGTAGTTGTCC GGGAGCACAGCAAATCAT 192 CT qRT AAAGTGGAGATTTGGGGAC CGCTCGGTACCTCAGAAA 84 G G CREB3 Rattus NM_00101309 STD CTGACGGAGGAAGAGAAAA GTTCAGGCAGGAAACATT 202 norvegicus 2.2 GG GC qRT GAGTATGTTGTGTTGCACC TTCTGAGCTCTCCAACTGG 123 G T CREB3L Homo sapiens NM_00101211 STD AATATGGTCCCCAGCTGAC AGCAGCAATTCCTTCACAG 306 3 5 A T qRT GTGTCATCCAAGCAAGCAA CACATGTCCAGGTCAATG 120 G GC SRSF3 Homo sapiens NM_003017, STD GCAGTCCGAGAGCTAGATG CACCACTTCTCTTGCAAAC 337 G TG NR_036610 qRT GTCGCAGATCTCCAAGAAG GGACGGCTTGTGATTTCTC 110 G T SRSF3 Rattus NM_00147907 STD ATCGTGATTCCTGTCCCTTG ACCACTTTTGCCAACTGGT 513 norvegicus .3 C

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

qRT AGAACGGGCTTTTGGCTATTTTCACCATTCGACAGTTCC 175 A ADCY5 Homo sapiens NM_00119964 STD ATTTTCTCCTGCACCAACAT ATCCCCAAGGATCTTAATA 403 2, C CG NM_183357 qRT TCCCAGAGACAGGCTTTCC TTTCATCTCCATGGCAACA 129 TG ADCY5 Rattus NM_022600.1 STD ATCTATACCATCTACACCCT ACATTGTCATGTTTCTGGA 398 norvegicus TG qRT ACCAGTTTCTGCTGAAACAG TCGGGTGGGTAGTGAGTG 88 TXNIP Homo sapiens NM_006472.4 STD ATGTTCCCGAATTGTGGTC ATCTGCTGCCAATTACCAG 270 qRT ATCATGGCGTGGCAAGAG TTCTTGGATCCAGGAACGC 117 STD: primers used for conventional PCR, qRT: primers used for real time qRT-PCR. The RefSeq ID of the sequence used to design the primers is provided.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 3. siRNAs.

Gene Species siRNA name Distributor Sequence None Allstars Negative Control Qiagen, Venlo, the Not provided siRNA Netherlands GATA6 Homo sapiens GATA6HSS104009(3_RNAI) Invitrogen, Paisley, UK AAGAAGUGGAAGUUGGAGUCAUG GG GATA6HSS178134(3_RNAI) Invitrogen UUGACCCGAAUACUUGAGCUCGC UG Rattus Gata6RSS334002(3_RNAI) Invitrogen GCUCCGGUAACAGCUCUGUUCCU norvegicus AU GataRSS334003(3_RNAI) Invitrogen GCAACGCAUGCGGUCUCUACAGU AA CREB3 Homo sapiens Creb3HSS116089(3_RNAI) Invitrogen CCACGGGAAACUGUCUCUAUGGA UC Creb3HSS116090(3_RNAI) Invitrogen GGACCCAGAUGACUCCACAGCAU AU CREB3 Rattus Creb3RSS335120(3_RNAI) Invitrogen UCAUCGCCGGAUGUUCUCAACUC norvegicus UU CrebRSS356125(3_RNAI) Invitrogen CCUUCCACAGGAGCACGUCUCCA UA ADCY5 Rattus Adcy5RSS329997(3_RNAI) Invitrogen AUGACAUGCGUUCACCUGGUUCA norvegicus CC

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 4. Functional classification of human islet genes modified by palmitate.

Media Log fold Gene name Gene description n 2 change RPKM Glucose metabolism PDK4 pyruvate dehydrogenase kinase, isozyme 4 21 1.17E+00 AKR1B10 aldo-keto reductase family 1, member B10 13 9.81E-01 G6PD glucose-6-phosphate dehydrogenase 14 6.74E-01 PMM2 phosphomannomutase 2 8.1 5.67E-01 PFKP phosphofructokinase, platelet 26 5.20E-01 HKDC1 hexokinase domain containing 1 11 3.82E-01 GAPDH glyceraldehyde-3-phosphate dehydrogenase 518 3.32E-01 ENO1 enolase 1, (alpha) 202 2.85E-01 GPT glutamic-pyruvate transaminase (alanine 1.9 -8.18E-01 aminotransferase) PFKFB2 6-phosphofructo-2-kinase/fructose-2,6- 23 -6.94E-01 biphosphatase 2 PPP1R3E protein phosphatase 1, regulatory subunit 3E 5.6 -5.28E-01 SORD sorbitol dehydrogenase -3.86E-01 CRYL1 crystallin, lambda 1 13 -2.72E-01 H6PD hexose-6-phosphate dehydrogenase (glucose 10 -2.15E-01 1-dehydrogenase) Lipid metabolism PLIN2 perilipin 2 6.0 2.67E+00 AGPAT9 1-acylglycerol-3-phosphate O-acyltransferase 1.2 1.17E+00 9 AKR1C2 aldo-keto reductase family 1, member C2 29 9.08E-01 FABP5 fatty acid binding protein 5 (psoriasis- 13 8.85E-01 associated) GK glycerol kinase 1.4 8.33E-01 ACSL1 acyl-CoA synthetase long-chain family member 11 7.99E-01 1 FADS1 fatty acid desaturase 1 13 7.99E-01 LRP8 low density lipoprotein receptor-related protein 1.2 7.69E-01 8, apolipoprotein e receptor AKR1C1 aldo-keto reductase family 1, member C1 63 7.69E-01 ACADVL acyl-CoA dehydrogenase, very long chain 95 7.56E-01 CPT1A carnitine palmitoyltransferase 1A (liver) 9.8 7.02E-01 LDLR low density lipoprotein receptor 20 6.48E-01 ECH1 enoyl CoA hydratase 1, peroxisomal 38 6.48E-01 SCD stearoyl-CoA desaturase (delta-9-desaturase) 86 6.44E-01 FA2H fatty acid 2-hydroxylase 11 6.14E-01 PITPNM1 phosphatidylinositol transfer protein, 15 5.91E-01 membrane-associated 1 MGLL monoglyceride lipase 5.6 5.85E-01 OSBPL10 oxysterol binding protein-like 10 2.8 5.55E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

HILPDA hypoxia inducible lipid droplet-associated 8.9 5.45E-01 MLYCD malonyl-CoA decarboxylase 6.2 5.35E-01 ACLY ATP citrate 76 4.63E-01 SC4MOL sterol-C4-methyl oxidase-like 33 4.10E-01 SREBF2 sterol regulatory element binding transcription 34 3.74E-01 factor 2 ACSL3 acyl-CoA synthetase long-chain family member 7.1 3.73E-01 3 HADHA hydroxyacyl-CoA dehydrogenase/3-ketoacyl- 36 3.69E-01 CoA thiolase/enoyl-CoA hydratase (trifunctional protein), alpha subunit HSD11B2 hydroxysteroid (11-beta) dehydrogenase 2 6.3 - 1.51E+00 CYP1A1 cytochrome P450, family 1, subfamily A, 5.8 - polypeptide 1 1.44E+00 FFAR3 free fatty acid receptor 3 2.0 - 1.17E+00 SERPINA6 serpin peptidase inhibitor, clade A (alpha-1 9.7 - antiproteinase, antitrypsin), member 6 1.13E+00 STARD4- STARD4 antisense RNA 1 (LOC100505678) 1.0 - AS1 1.09E+00 FFAR1 free fatty acid receptor 1 17 -9.85E-01 C5orf13 Chromosome 5 open reading frame 13 12 -8.75E-01 CYP3A5 cytochrome P450, family 3, subfamily A, 21 -8.68E-01 polypeptide 5 SCARA3 scavenger receptor class A, member 3 1.7 -8.67E-01 LIPT1 lipoyltransferase 1 3.1 -8.50E-01 STS steroid sulfatase (microsomal), isozyme S 5.4 -8.42E-01 APOE apolipoprotein E 6.4 -8.33E-01 VNN2 vanin 2 14 -7.98E-01 PLCH2 phospholipase C, eta 2 5.2 -7.90E-01 METTL7A methyltransferase like 7A 11 -7.88E-01 EBPL emopamil binding protein-like 6.5 -6.63E-01 C3orf57, aka serine palmitoyltransferase, small subunit B 3.7 -6.53E-01 SPTSSB ARSE arylsulfatase E (chondrodysplasia punctata 1) 12 -6.44E-01 CBR4 carbonyl reductase 4 6.3 -6.42E-01 LDLRAD2 low density lipoprotein receptor class A domain 14 -6.33E-01 containing 2 NPC1L1 NPC1 (Niemann-Pick disease, type C1, gene)- 10 -6.22E-01 like 1 PPAP2B phosphatidic acid phosphatase type 2B 8.5 -5.66E-01 RORC RAR-related orphan receptor C 13 -5.56E-01 ABCG1 ATP-binding cassette, sub-family G (WHITE), 7.7 -5.39E-01 member 1 GPRC5B G protein-coupled receptor, family C, group 5, 24 -5.36E-01 member B

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

SMPD3 sphingomyelin phosphodiesterase 3, neutral 2.9 -5.28E-01 membrane (neutral sphingomyelinase II) O3FAR1 omega-3 fatty acid receptor 1 (GPR120) 4.0 -5.09E-01 HSD3B7 hydroxy-delta-5-steroid dehydrogenase, 3 12 -4.98E-01 beta- and steroid delta- 7 NAPEPLD N-acyl phosphatidylethanolamine 4.1 -4.80E-01 phospholipase D MBOAT1 membrane bound O-acyltransferase domain 2.9 -4.75E-01 containing 1 GPD1 glycerol-3-phosphate dehydrogenase 1 8.6 -4.72E-01 (soluble) AGK acylglycerol kinase 5.2 -4.28E-01 PCTP phosphatidylcholine transfer protein 5.2 -4.00E-01 AACS acetoacetyl-CoA synthetase 17 -3.67E-01 LCLAT1 lysocardiolipin acyltransferase 1 3.3 -3.48E-01 CERS2 ceramide synthase 2 73 -3.19E-01 GBA glucosidase, beta, acid 32 -3.14E-01 GALC galactosylceramidase 6.9 -3.06E-01 NPC2 Niemann-Pick disease, type C2 81 -3.04E-01 ASAH1 N-acylsphingosine (acid 34 -2.93E-01 ) 1 LIPA lipase A, lysosomal acid, cholesterol esterase 17 -2.92E-01 SORL1 sortilin-related receptor, L(DLR class) A 30 -2.66E-01 repeats containing SERINC5 serine incorporator 5 8.9 -2.46E-01 PTPLAD1 protein tyrosine phosphatase-like A domain 36 -2.28E-01 containing 1 NMT1 N-myristoyltransferase 1 17 -1.73E-01 CDS2 CDP-diacylglycerol synthase (phosphatidate 9.3 -1.68E-01 cytidylyltransferase) 2 HEXA hexosaminidase A (alpha polypeptide) 23 -1.67E-01 SOAT1 sterol O-acyltransferase 1 9.0 -1.47E-01 Aminoacid metabolism GCLM glutamate-cysteine , modifier subunit 8.0 1.07E+00 GCLC glutamate-cysteine ligase, catalytic subunit 8.5 9.59E-01 PYCR1 pyrroline-5-carboxylate reductase 1 6.5 7.84E-01 GPT2 glutamic pyruvate transaminase (alanine 4.1 3.99E-01 aminotransferase) 2 PIPOX pipecolic acid oxidase 3.1 - 1.32E+00 AMT aminomethyltransferase 9.1 - 1.23E+00 SHMT1 serine hydroxymethyltransferase 1 (soluble) 2.9 -9.75E-01 ALDH4A1 aldehyde dehydrogenase 4 family, member A1 3.7 -9.07E-01 GPT glutamic-pyruvate transaminase (alanine 1.9 -8.18E-01 aminotransferase) DPH1 DPH1 homolog (S. cerevisiae) 11 -6.82E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

DDC dopa decarboxylase (aromatic L-amino acid 4.3 -6.65E-01 decarboxylase) GLS 31 -5.13E-01 ADI1 acireductone dioxygenase 1 26 -4.94E-01 AHCYL2 adenosylhomocysteinase-like 2 9.2 -4.20E-01 IVD isovaleryl-CoA dehydrogenase 9.5 -3.93E-01 GLUL glutamate-ammonia ligase 50 -3.88E-01 HIBCH 3-hydroxyisobutyryl-CoA 6.6 -3.49E-01 HGD homogentisate 1,2-dioxygenase 8.5 -2.31E-01 HIBADH 3-hydroxyisobutyrate dehydrogenase 14 -1.91E-01 Krebs cycle/ATP production PCK2 phosphoenolpyruvate carboxykinase 2 10 3.87E-01 (mitochondrial) MDH2 malate dehydrogenase 2, NAD (mitochondrial) 62 3.24E-01 ATP5L2 ATP synthase, H+ transporting, mitochondrial 9.8 - Fo complex, subunit G2 2.01E+00 ACSS1 acyl-CoA synthetase short-chain family 11 -7.72E-01 member 1 UQCR10 ubiquinol-cytochrome c reductase, complex III 37 -4.60E-01 subunit X SDHC succinate dehydrogenase complex, subunit C, 21 -3.96E-01 integral membrane protein, 15kDa IDH2 isocitrate dehydrogenase 2 (NADP+), 60 -3.22E-01 mitochondrial COX18 cytochrome c oxidase assembly homolog (S. 2.2 -2.92E-01 cerevisiae) ATP5A1 ATP synthase, H+ transporting, mitochondrial 65 -2.77E-01 F1 complex, alpha subunit 1, cardiac muscle CS citrate synthase 47 -2.38E-01 IDH3B isocitrate dehydrogenase 3 (NAD+) beta 19 -2.24E-01 NDUFS2 NADH dehydrogenase (ubiquinone) Fe-S 22 -1.95E-01 protein 2, 49kDa (NADH-coenzyme Q reductase) Metabolism – miscellaneous UGDH UDP-glucose 6-dehydrogenase 16 8.68E-01 NANS N-acetylneuraminic acid synthase 14 7.28E-01 TSTA3 tissue specific transplantation antigen P35B 11 5.82E-01 TALDO1 transaldolase 1 29 3.48E-01 OAZ1 ornithine decarboxylase antizyme 1 187 2.39E-01 HNMT histamine N-methyltransferase 9.3 -9.08E-01 ADHFE1 alcohol dehydrogenase, iron containing, 1 2.9 -8.45E-01 GGTLC1 gamma-glutamyltransferase light chain 1 8.0 -8.41E-01 ADH1C alcohol dehydrogenase 1C (class I), gamma 5.1 -8.18E-01 polypeptide AOC2 amine oxidase, copper containing 2 (retina- 2.3 -7.65E-01 specific) DDC dopa decarboxylase (aromatic L-amino acid 4.3 -6.65E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

decarboxylase) AOC3 amine oxidase, copper containing 3 1.7 -6.21E-01 NMNAT3 nicotinamide nucleotide adenylyltransferase 3 3.4 -6.06E-01 NMRK1 nicotinamide riboside kinase 1 (C9orf95) 11 -4.83E-01 FDX1 ferredoxin 1 4.8 -4.74E-01 CMBL carboxymethylenebutenolidase homolog 7.5 -4.63E-01 (Pseudomonas) HSD17B11 hydroxysteroid (17-beta) dehydrogenase 11 9.0 -4.58E-01 ADH5 alcohol dehydrogenase 5 (class III), chi 12 -4.33E-01 polypeptide ESD esterase D 23 -4.06E-01 GNPDA1 glucosamine-6-phosphate deaminase 1 13 -4.05E-01 GGT1 gamma-glutamyltransferase 1 16 -4.04E-01 N6AMT1 N-6 adenine-specific DNA methyltransferase 1 2.3 -3.79E-01 (putative) NUDT2 nudix (nucleoside diphosphate linked moiety 10 -3.53E-01 X)-type motif 2 SAT2 spermidine/spermine N1-acetyltransferase 31 -3.23E-01 family member 2 NDST2 N-deacetylase/N-sulfotransferase (heparan 6.7 -2.61E-01 glucosaminyl) 2 PRPS1 phosphoribosyl pyrophosphate synthetase 1 16 -1.90E-01 Hormones/growth factors/receptors/neuropeptides and exocytosis IGFBP1 insulin-like growth factor binding protein 1 0.3 2.76E+00 FGF18 fibroblast growth factor 18 0.2 2.51E+00 ANGPTL4 angiopoietin-like 4 14 2.08E+00 PYY peptide YY 0.4 2.07E+00 GPR3 G protein-coupled receptor 3 0.5 1.84E+00 INHBA inhibin, beta A 9.5 1.44E+00 VGF nerve growth factor inducible 381 1.16E+00 GAP43 growth associated protein 43 6.4 1.12E+00 GDF15 growth differentiation factor 15 62 1.11E+00 ADM adrenomedullin 10 1.06E+00 FGF2 fibroblast growth factor 2 (basic) 1.4 1.04E+00 CHRNA5 cholinergic receptor, nicotinic, alpha 5 1.0 8.93E-01 NRP2 neuropilin 2 2.1 8.36E-01 IAPP islet amyloid polypeptide 348 8.38E-01 NRG1 neuregulin 1 0.6 8.04E-01 SYT5 V 13 7.94E-01 RAB3B RAB3B, member RAS oncogene family 15 7.65E-01 IGFBP4 insulin-like growth factor binding protein 4 74 7.62E-01 PTGER4 prostaglandin E receptor 4 (subtype EP4) 1.0 7.60E-01 CHGB chromogranin B (secretogranin 1) 885 7.32E-01 PAM peptidylglycine alpha-amidating 107 6.84E-01 monooxygenase BDKRB2 bradykinin receptor B2 8.8 5.49E-01 UNC13A unc-13 homolog A (C. elegans) 11 5.38E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

ADM2 adrenomedullin 2 2.4 4.54E-01 RIMS2 regulating synaptic membrane exocytosis 2 1.8 4.16E-01 FGFR1 fibroblast growth factor receptor 1 18 3.11E-01 EDN3 endothelin 3 27 2.33E-01 RPH3A rabphilin 3A homolog (mouse) 5.1 - 1.51E+00 FOLR1 folate receptor 1 (adult) 3.4 - 1.31E+00 GREM2 gremlin 2 1.3 - 1.31E+00 IGFBP6 insulin-like growth factor binding protein 6 4.8 - 1.20E+00 SSTR5 somatostatin receptor 5 1.2 - 1.09E+00 GLRA1 glycine receptor, alpha 1 1.4 - 1.02E+00 HNMT histamine N-methyltransferase 9.3 -9.08E-01 GRIA4 glutamate receptor, ionotropic, AMPA 4 2.9 -8.57E-01 RTP4 receptor (chemosensory) transporter protein 4 6.5 -8.47E-01 PAQR5 progestin and adipoQ receptor family member 4.8 -6.60E-01 V PAQR7 progestin and adipoQ receptor family member 7.0 -6.42E-01 VII GPRC5B G protein-coupled receptor, family C, group 5, 24 -5.36E-01 member B ADRA2A adrenoceptor alpha 2A 9.7 -5.22E-01 MAOA monoamine oxidase A 9.0 -5.12E-01 O3FAR1 omega-3 fatty acid receptor 1 4.0 -5.09E-01 SH2B1 SH2B adaptor protein 1 15 -4.98E-01 GLCE glucuronic acid epimerase 5.7 -4.97E-01 IGFBP7 insulin-like growth factor binding protein 7 115 -4.33E-01 GCG glucagon 5101 -4.02E-01 IGF1R insulin-like growth factor 1 receptor 20 -4.01E-01 LEPR leptin receptor 3.8 -4.00E-01 GPR39 G protein-coupled receptor 39 8.9 -3.98E-01 CASR calcium-sensing receptor 24 -3.64E-01 AGT angiotensinogen (serpin peptidase inhibitor, 26 -3.57E-01 clade A, member 8) LCN2 lipocalin 2 331 -3.46E-01 FAM3B family with sequence similarity 3, member B 19 -3.42E-01 ABAT 4-aminobutyrate aminotransferase 14 -3.35E-01 EPB41L1 erythrocyte membrane protein band 4.1-like 1 15 -2.92E-01 NOTCH2 notch 2 13 -2.68E-01 PTPRN2 protein tyrosine phosphatase, receptor type, N 50 -2.49E-01 polypeptide 2 NISCH nischarin 20 -2.45E-01 SCG3 secretogranin III 49 -2.32E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

SCG5 secretogranin V (7B2 protein) 250 -2.04E-01 GNRHR2 gonadotropin-releasing hormone (type 2) 5.3 -1.88E-01 receptor 2 ADCYAP1 adenylate cyclase activating polypeptide 1 14 -1.78E-01 (pituitary) APLP2 amyloid beta (A4) precursor-like protein 2 147 -1.45E-01 SEPT8 septin 8 14 -1.34E-01 Protein synthesis/translation regulation/protein folding/endoplasmic reticulum stress CREB3L3 cAMP responsive element binding protein 3- 0.2 2.92E+00 like 3 ATF3 activating transcription factor 3 15 1.36E+00 AGR2 anterior gradient 2 homolog (Xenopus laevis) 8.9 1.22E+00 DOHH deoxyhypusine hydroxylase/monooxygenase 12 9.12E-01 HSPA7 heat shock 70kDa protein 7 (HSP70B) 1.9 9.08E-01 PPP1R15A protein phosphatase 1, regulatory subunit 15A 21 9.08E-01 TRIB3 tribbles homolog 3 (Drosophila) 20 8.72E-01 HSPA6 heat shock 70kDa protein 6 (HSP70B') 2.3 8.11E-01 HMOX1 heme oxygenase (decycling) 1 6.2 7.85E-01 HSPA5 heat shock 70kDa protein 5 (glucose-regulated 120 7.47E-01 protein, 78kDa) SEC24D SEC24 family, member D (S. cerevisiae) 17 7.40E-01 DNAJC12 DnaJ (Hsp40) homolog, subfamily C, member 35 7.17E-01 12 CRELD2 cysteine-rich with EGF-like domains 2 9.0 6.60E-01 CHAC1 ChaC, cation transport regulator homolog 1 (E. 4.7 6.34E-01 coli) EDEM1 ER degradation enhancer, mannosidase 9.8 6.05E-01 alpha-like 1 ERN1 endoplasmic reticulum to nucleus signaling 1 5.7 6.00E-01 XPOT exportin, tRNA 8.2 5.80E-01 HSPA13 heat shock protein 70kDa family, member 13 9.8 5.70E-01 IARS isoleucyl-tRNA synthetase 17 5.17E-01 GARS glycyl-tRNA synthetase 19 5.17E-01 SELS selenoprotein S 20 5.01E-01 CPE carboxypeptidase E 469 5.00E-01 TXNDC5 thioredoxin domain containing 5 (endoplasmic 40 4.82E-01 reticulum) SSR3 signal sequence receptor, gamma (translocon- 25 4.72E-01 associated protein gamma) PCSK1 proprotein convertase subtilisin/kexin type 1 145 4.70E-01 PDIA4 protein disulfide isomerase family A, member 4 42 4.51E-01 ORMDL2 ORM1-like 2 (S. cerevisiae) 14 4.36E-01 MARS methionyl-tRNA synthetase 22 4.21E-01 SEC23B Sec23 homolog B (S. cerevisiae) 18 4.11E-01 EEF1A2 eukaryotic translation elongation factor 1 alpha 90 4.09E-01 2

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

DNAJB11 DnaJ (Hsp40) homolog, subfamily B, member 18 3.97E-01 11 SEC61A1 Sec61 alpha 1 subunit (S. cerevisiae) 90 3.84E-01 RPL8 L8 393 3.84E-01 GOSR2 golgi SNAP receptor complex member 2 11 3.59E-01 EDEM2 ER degradation enhancer, mannosidase 12 3.43E-01 alpha-like 2 HSPA9 heat shock 70kDa protein 9 (mortalin) 34 3.38E-01 ABCF1 ATP-binding cassette, sub-family F (GCN20), 14 3.31E-01 member 1 WARS tryptophanyl-tRNA synthetase 25 3.28E-01 RCN1 reticulocalbin 1, EF-hand calcium binding 32 3.23E-01 domain ATF4 activating transcription factor 4 (tax-responsive 139 3.17E-01 enhancer element B67) PPIB peptidylprolyl isomerase B (cyclophilin B) 181 3.01E-01 VARS valyl-tRNA synthetase 9.9 2.98E-01 CREB3 cAMP responsive element binding protein 3 15 2.48E-01 NOL6 nucleolar 6 (RNA-associated) 6.7 2.45E-01 CARS cysteinyl-tRNA synthetase 8.6 2.41E-01 EIF4A1 eukaryotic translation initiation factor 4A1 60 2.39E-01 PDIA6 protein disulfide isomerase family A, member 6 59 2.31E-01 SARS seryl-tRNA synthetase 60 2.25E-01 RPS6KA1 kinase, 90kDa, 9.0 2.16E-01 polypeptide 1 LMAN2 lectin, mannose-binding 2 41 1.36E-01 NANOS1 nanos homolog 1 (Drosophila) 2.9 -9.13E-01 FKBP1B FK506 binding protein 1B, 12.6 kDa 8.0 -8.06E-01 FKBP7 FK506 binding protein 7 1.7 -8.04E-01 PIGV phosphatidylinositol glycan anchor 6.5 -7.75E-01 biosynthesis, class V DPH1 DPH1 homolog (S. cerevisiae) 11 -6.82E-01 GPX8 glutathione peroxidase 8 (putative) 2.8 -6.53E-01 CREB3L4 cAMP responsive element binding protein 3- 4.3 -6.38E-01 like 4 ALG8 -linked glycosylation 8, alpha-1,3- 6.0 -5.43E-01 glucosyltransferase homolog (S. cerevisiae) MRPL36 mitochondrial ribosomal protein L36 16 -4.72E-01 MRPS33 mitochondrial ribosomal protein S33 22 -4.67E-01 PEX19 peroxisomal biogenesis factor 19 12 -3.89E-01 N6AMT1 N-6 adenine-specific DNA methyltransferase 1 2.3 -3.79E-01 (putative) JKAMP JNK1/MAPK8-associated membrane protein 12 -3.74E-01 ARL6IP5 ADP-ribosylation-like factor 6 interacting 57 -3.66E-01 protein 5 EIF2D eukaryotic translation initiation factor 2D 9.5 -3.57E-01 ERMP1 endoplasmic reticulum metallopeptidase 1 7.8 -3.51E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

EIF4A2 eukaryotic translation initiation factor 4A2 145 -3.10E-01 FKBP9 FK506 binding protein 9, 63 kDa 30 -2.98E-01 MPPE1 metallophosphoesterase 1 6.1 -2.86E-01 HSPB8 heat shock 22kDa protein 8 15 -2.74E-01 RPL41 ribosomal protein L41 1148 -2.62E-01 SRP14 signal recognition particle 14kDa (homologous 59 -2.53E-01 Alu RNA binding protein) TMX2 thioredoxin-related transmembrane protein 2 26 -2.32E-01 PTPLAD1 protein tyrosine phosphatase-like A domain 36 -2.28E-01 containing 1 RPL5 ribosomal protein L5 169 -2.15E-01 EIF4G2 eukaryotic translation initiation factor 4 gamma, 122 -2.02E-01 2 ERO1LB ERO1-like beta (S. cerevisiae) 66 -1.66E-01 MRPS27 mitochondrial ribosomal protein S27 12 -1.64E-01 KDELR1 KDEL (Lys-Asp-Glu-Leu) endoplasmic 86 -1.62E-01 reticulum protein retention receptor 1 Posttranslational modification/ubiquitination MGAT2 mannosyl (alpha-1,6-)-glycoprotein beta-1,2-N- 9.0 4.21E-01 acetylglucosaminyltransferase FBXW5 F-box and WD repeat domain containing 5 33 4.14E-01 FBXW7 F-box and WD repeat domain containing 7, E3 7.8 4.05E-01 ubiquitin protein ligase GCNT1 glucosaminyl (N-acetyl) transferase 1, core 2 3.9 3.71E-01 UBAP1 ubiquitin associated protein 1 19 1.78E-01 RBBP6 retinoblastoma binding protein 6 8.1 1.12E-01 TRIM74 tripartite motif containing 74 1.9 - 1.55E+00 RNF122 ring finger protein 122 5.1 -9.94E-01 PCSK4 proprotein convertase subtilisin/kexin type 4 1.2 -9.38E-01 MUC20 mucin 20, cell surface associated 10 -7.48E-01 DTX4 deltex homolog 4 (Drosophila) 5.3 -7.14E-01 UBD ubiquitin D 77 -6.85E-01 USP2 ubiquitin specific peptidase 2 4.5 -6.08E-01 KLHL3 kelch-like 3 (Drosophila) 3.9 -5.95E-01 NEURL neuralized homolog (Drosophila) 27 -5.32E-01 ST6GALNAC ST6 (alpha-N-acetyl-neuraminyl-2,3-beta- 5.3 -5.25E-01 2 galactosyl-1,3)-N-acetylgalactosaminide alpha- 2,6-sialyltransferase 2 CHST9 carbohydrate (N-acetylgalactosamine 4-0) 8.9 -4.69E-01 sulfotransferase 9 FUT2 fucosyltransferase 2 (secretor status included) 3.3 -4.44E-01 MKRN1 makorin ring finger protein 1 25 -4.30E-01 FANCL Fanconi anemia, complementation group L 6.3 -4.19E-01 USP54 ubiquitin specific peptidase 54 7.2 -3.74E-01 ST6GAL1 ST6 beta-galactosamide alpha-2,6- 16 -3.64E-01 sialyltranferase 1

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

USP30 ubiquitin specific peptidase 30 4.1 -3.58E-01 SLC35A1 solute carrier family 35 (CMP-sialic acid 7.8 -3.44E-01 transporter), member A1 SAT2 spermidine/spermine N1-acetyltransferase 31 -3.23E-01 family member 2 UBE2H ubiquitin-conjugating enzyme E2H 35 -3.13E-01 DCUN1D1 DCN1, defective in cullin neddylation 1, 4.4 -3.13E-01 domain containing 1 (S. cerevisiae) DCAF11 DDB1 and CUL4 associated factor 11 15 -2.98E-01 SUMF2 sulfatase modifying factor 2 76 -2.82E-01 B4GALT1 UDP-Gal:betaGlcNAc beta 1,4- 33 -2.59E-01 galactosyltransferase, polypeptide 1 UBE3A ubiquitin protein ligase E3A 11 -2.52E-01 DCAF7 DDB1 and CUL4 associated factor 7 16 -2.39E-01 FBXO28 F-box protein 28 5.0 -2.07E-01 SCG5 secretogranin V (7B2 protein) 250 -2.04E-01 NMT1 N-myristoyltransferase 1 17 -1.73E-01 Proteasome/lysosome/autophagy LAMP3 lysosomal-associated membrane protein 3 3.0 8.43E-01 CTSZ cathepsin Z 105 7.22E-01 IDS iduronate 2-sulfatase 90 5.58E-01 PSMA7 proteasome (prosome, macropain) subunit, 31 3.23E-01 alpha type, 7 PSMC4 proteasome (prosome, macropain) 26S 19 3.21E-01 subunit, ATPase, 4 ATP6V0C ATPase, H+ transporting, lysosomal 16kDa, V0 124 2.24E-01 subunit c CD63 CD63 molecule 253 2.17E-01 MUTED muted homolog (mouse) 2.6 - 1.71E+00 PPT2 palmitoyl-protein thioesterase 2 5.2 - 1.06E+00 WIPI2 WD repeat domain, phosphoinositide 9.5 -8.30E-01 interacting 2 LAPTM5 lysosomal protein transmembrane 5 9.4 -7.43E-01 ULK3 unc-51-like kinase 3 (C. elegans) 10 -5.64E-01 CTSF cathepsin F 14 -5.26E-01 CTSO cathepsin O 6.9 -5.23E-01 CTSS cathepsin S 11 -4.94E-01 DRAM2 DNA-damage regulated autophagy modulator 10 -4.86E-01 2 ARSD arylsulfatase D 22 -4.60E-01 AP3M2 adaptor-related protein complex 3, mu 2 4.6 -4.58E-01 subunit PSMD10 proteasome (prosome, macropain) 26S 12 -4.36E-01 subunit, non-ATPase, 10 ATG7 autophagy related 7 8.6 -4.21E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

CLN5 ceroid-lipofuscinosis, neuronal 5 14 -3.89E-01 ARL8B ADP-ribosylation factor-like 8B 19 -3.74E-01 TMEM9 transmembrane protein 9 35 -3.60E-01 SCOC short coiled-coil protein 23 -3.53E-01 AP3B1 adaptor-related protein complex 3, beta 1 11 -3.25E-01 subunit TMEM9B TMEM9 domain family, member B 17 -3.20E-01 CERS2 ceramide synthase 2 73 -3.19E-01 GBA glucosidase, beta, acid 32 -3.14E-01 GALC galactosylceramidase 6.9 -3.06E-01 NPC2 Niemann-Pick disease, type C2 81 -3.04E-01 ASAH1 N-acylsphingosine amidohydrolase (acid 34 -2.93E-01 ceramidase) 1 LIPA lipase A, lysosomal acid, cholesterol esterase 17 -2.92E-01 HSPB8 heat shock 22kDa protein 8 15 -2.74E-01 VIPAS39 VPS33B interacting protein, apical-basolateral 5.3 -2.71E-01 polarity regulator, spe-39 homolog (C14orf133) KIAA1324 KIAA1324 48 -2.67E-01 SORL1 sortilin-related receptor, L(DLR class) A 30 -2.66E-01 repeats containing VAMP8 vesicle-associated membrane protein 8 56 -2.65E-01 (endobrevin) ATP6AP2 ATPase, H+ transporting, lysosomal accessory 46 -2.55E-01 protein 2 CTSA cathepsin A 64 -2.50E-01 PTPRN2 protein tyrosine phosphatase, receptor type, N 50 -2.49E-01 polypeptide 2 SIDT2 SID1 transmembrane family, member 2 12 -2.37E-01 PSMB8 proteasome (prosome, macropain) subunit, 33 -2.34E-01 beta type, 8 (large multifunctional peptidase 7) HGSNAT heparan-alpha-glucosaminide N- 19 -2.21E-01 acetyltransferase CTSD cathepsin D 371 -2.07E-01 SORT1 sortilin 1 22 -2.03E-01 HEXA hexosaminidase A (alpha polypeptide) 23 -1.67E-01 KDELR1 KDEL (Lys-Asp-Glu-Leu) endoplasmic 86 -1.62E-01 reticulum protein retention receptor 1 GLB1 galactosidase, beta 1 31 -1.61E-01 AP2M1 adaptor-related protein complex 2, mu 1 65 -1.59E-01 subunit PRDX6 peroxiredoxin 6 34 -1.31E-01 LAMP1 lysosomal-associated membrane protein 1 116 -1.26E-01 CD68 CD68 molecule 44 -9.91E-02 Vesicle transport SNX22 sorting nexin 22 31 3.56E-01 COPG1 coatomer protein complex, subunit gamma 1 44 3.48E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

AP3D1 adaptor-related protein complex 3, delta 1 32 2.85E-01 subunit VCP valosin containing protein 43 2.06E-01 LMAN2 lectin, mannose-binding 2 41 1.36E-01 SYNGR4 synaptogyrin 4 2.7 - 1.36E+00 COPG2 coatomer protein complex, subunit gamma 2 5.0 -6.72E-01 STX6 syntaxin 6 5.6 -4.75E-01 VPS52 vacuolar protein sorting 52 homolog (S. 12 -4.65E-01 cerevisiae) SYNGR1 synaptogyrin 1 11 -4.55E-01 SNX27 sorting nexin family member 27 9.3 -4.45E-01 SNAP23 synaptosomal-associated protein, 23kDa 8.5 -4.07E-01 RAB2B RAB2B, member RAS oncogene family 7.0 -3.93E-01 RAB11FIP4 RAB11 family interacting protein 4 (class II) 8.0 -3.91E-01 NIPSNAP1 nipsnap homolog 1 (C. elegans) 21 -3.42E-01 SNX29 sorting nexin 29 (RUNDC2A) 5.5 -3.35E-01 COG4 component of oligomeric golgi complex 4 12 -3.24E-01 FAM21C family with sequence similarity 21, member C 15 -2.91E-01 SH3GL2 SH3-domain GRB2-like 2 9.9 -2.90E-01 SNX6 sorting nexin 6 11 -2.89E-01 MPPE1 metallophosphoesterase 1 6.1 -2.86E-01 VIPAS39 VPS33B interacting protein, apical-basolateral 5.3 -2.71E-01 polarity regulator, spe-39 homolog (C14orf133) AMOT angiomotin 4.6 -2.71E-01 SORL1 sortilin-related receptor, L(DLR class) A 30 -2.66E-01 repeats containing VAMP8 vesicle-associated membrane protein 8 56 -2.65E-01 (endobrevin) ATP6AP2 ATPase, H+ transporting, lysosomal accessory 46 -2.55E-01 protein 2 RUFY1 RUN and FYVE domain containing 1 11 -2.38E-01 FLOT2 flotillin 2 37 -2.36E-01 AP2B1 adaptor-related protein complex 2, beta 1 35 -2.06E-01 subunit SORT1 sortilin 1 22 -2.03E-01 CLSTN1 calsyntenin 1 45 -1.15E-01 Mitochondrial enzymes MTHFD2 methylenetetrahydrofolate dehydrogenase 7.7 8.45E-01 (NADP+ dependent) 2, methenyltetrahydrofolate cyclohydrolase LONP1 lon peptidase 1, mitochondrial 17 5.63E-01 MTHFD1L methylenetetrahydrofolate dehydrogenase 3.6 3.48E-01 (NADP+ dependent) 1-like ATAD3A ATPase family, AAA domain containing 3A 5.9 3.08E-01 PITRM1 pitrilysin metallopeptidase 1 17 2.99E-01 IMMP2L IMP2 inner mitochondrial membrane 5.1 -9.11E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

peptidase-like (S. cerevisiae) CBR4 carbonyl reductase 4 6.3 -6.42E-01 NMNAT3 nicotinamide nucleotide adenylyltransferase 3 3.4 -6.06E-01 MAOA monoamine oxidase A 9.0 -5.12E-01 HMGCL 3-hydroxymethyl-3-methylglutaryl-CoA lyase 20 -4.67E-01 AGK acylglycerol kinase 5.2 -4.28E-01 IVD isovaleryl-CoA dehydrogenase 9.5 -3.93E-01 USP30 ubiquitin specific peptidase 30 4.1 -3.58E-01 NFS1 NFS1 nitrogen fixation 1 homolog (S. 7.1 -3.56E-01 cerevisiae) NIPSNAP1 nipsnap homolog 1 (C. elegans) 21 -3.42E-01 TIMMDC1 of inner mitochondrial membrane 15 -2.34E-01 domain containing 1 (C3orf1) CDS2 CDP-diacylglycerol synthase (phosphatidate 9.3 -1.68E-01 cytidylyltransferase) 2 Channels and transporters SLC7A11 solute carrier family 7, (cationic amino acid 1.0 2.18E+00 transporter, y+ system) member 11 SLC5A4 solute carrier family 5 (low affinity glucose 0.9 1.64E+00 cotransporter), member 4 AQP3 aquaporin 3 (Gill blood group) 42 9.98E-01 SLC7A5 solute carrier family 7 (amino acid transporter 28 9.95E-01 light chain, L system), member 5 CNGA3 cyclic nucleotide gated channel alpha 3 5.0 9.66E-01 BEST1 bestrophin 1 1.0 9.50E-01 KCNE4 potassium voltage-gated channel, Isk-related 3.8 9.33E-01 family, member 4 KCNQ3 potassium voltage-gated channel, KQT-like 1.7 7.05E-01 subfamily, member 3 SLC35D3 solute carrier family 35, member D3 1.1 7.02E-01 SLC45A3 solute carrier family 45, member 3 2.9 6.39E-01 SLC7A1 solute carrier family 7 (cationic amino acid 15 5.87E-01 transporter, y+ system), member 1 TMCO3 transmembrane and coiled-coil domains 3 32 5.34E-01 SLC38A5 solute carrier family 38, member 5 1.4 4.93E-01 SLC6A6 solute carrier family 6 (neurotransmitter 24 4.72E-01 transporter, taurine), member 6 IPO4 importin 4 5.1 4.31E-01 ABCB6 ATP-binding cassette, sub-family B 6.0 4.27E-01 (MDR/TAP), member 6 SLC9A1 solute carrier family 9, subfamily A (NHE1, 9.7 4.01E-01 cation proton antiporter 1), member 1 SLC39A7 solute carrier family 39 (zinc transporter), 53 3.95E-01 member 7 SLC38A2 solute carrier family 38, member 2 16 3.88E-01 CLIC1 chloride intracellular channel 1 97 3.74E-01 SLC35B1 solute carrier family 35, member B1 15 3.40E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

ABCF1 ATP-binding cassette, sub-family F (GCN20), 14 3.31E-01 member 1 SLC12A4 solute carrier family 12 (potassium/chloride 7.2 2.93E-01 transporters), member 4 SLC25A25 solute carrier family 25 (mitochondrial carrier; 7.3 2.67E-01 phosphate carrier), member 25 TCN1 transcobalamin I (vitamin B12 binding protein, 17 - R binder family) 1.74E+00 SLC9A3 solute carrier family 9, subfamily A (NHE3, 1.3 - cation proton antiporter 3), member 3 1.14E+00 SLC15A3 solute carrier family 15, member 3 1.5 - 1.12E+00 KCNK16 potassium channel, subfamily K, member 16 51 - 1.27E+00 KCNK17 potassium channel, subfamily K, member 17 9.8 - 1.23E+00 CLCNKB chloride channel, voltage-sensitive Kb 1.4 - 1.23E+00 AQP1 aquaporin 1 (Colton blood group) 179 - 1.11E+00 GLRA1 glycine receptor, alpha 1 1.4 - 1.02E+00 SLC25A34 solute carrier family 25, member 34 8.9 -8.80E-01 GJB2 gap junction protein, beta 2, 26kDa 3.3 -8.57E-01 SLC16A9 solute carrier family 16, member 9 4.2 -8.16E-01 (monocarboxylic acid transporter 9) CACNG4 calcium channel, voltage-dependent, gamma 2.2 -7.95E-01 subunit 4 SLC46A3 solute carrier family 46, member 3 6.1 -6.97E-01 TMEM37 transmembrane protein 37 18 -6.65E-01 SLC3A1 solute carrier family 3 (cystine, dibasic and 68 -6.59E-01 neutral amino acid transporters, activator of cystine, dibasic and neutral amino acid transport), member 1 KCNAB1 potassium voltage-gated channel, shaker- 1.1 -6.59E-01 related subfamily, beta member 1 KCNMB2 potassium large conductance calcium- 8.3 -6.02E-01 activated channel, subfamily M, beta member 2 KCNK5 potassium channel, subfamily K, member 5 9.8 -5.61E-01 SLC35E2 solute carrier family 35, member E2 20 -5.43E-01 KCND1 potassium voltage-gated channel, Shal-related 3.0 -5.43E-01 subfamily, member 1 ABCG1 ATP-binding cassette, sub-family G (WHITE), 7.7 -5.39E-01 member 1 ABCC5 ATP-binding cassette, sub-family C 8.5 -5.04E-01 (CFTR/MRP), member 5

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

GJB1 gap junction protein, beta 1, 32kDa 28 -4.87E-01 SLC44A2 solute carrier family 44, member 2 22 -4.58E-01 KCTD2 potassium channel tetramerisation domain 10 -4.33E-01 containing 2 SLC25A10 solute carrier family 25 (mitochondrial carrier; 8.4 -3.69E-01 dicarboxylate transporter), member 10 SLC18B1 solute carrier family 18, subfamily B, member 1 6.0 -3.58E-01 (C6orf192) SLC35A1 solute carrier family 35 (CMP-sialic acid 7.8 -3.44E-01 transporter), member A1 KCNJ16 potassium inwardly-rectifying channel, 11 -2.95E-01 subfamily J, member 16 ABCF2 ATP-binding cassette, sub-family F (GCN20), 7.0 -2.86E-01 member 2 SERINC5 serine incorporator 5 8.9 -2.46E-01 SLC39A9 solute carrier family 39 (zinc transporter), 21 -2.08E-01 member 9 FXYD6 FXYD domain containing ion transport 21 -1.91E-01 regulator 6 KCTD7 potassium channel tetramerisation domain 3.5 -1.87E-01 containing 7 SLC34A2 solute carrier family 34 (sodium phosphate), 27 -1.58E-01 member 2 SLC23A2 solute carrier family 23 (nucleobase 9.4 -1.40E-01 transporters), member 2 SLC25A3 solute carrier family 25 (mitochondrial carrier; 83 -1.34E-01 phosphate carrier), member 3 Cytoskeleton and related proteins TUBB2B tubulin, beta 2B 23 9.28E-01 ABLIM3 actin binding LIM protein family, member 3 1.9 8.83E-01 TUBB2A tubulin, beta 2A 33 8.67E-01 TUBB6 tubulin, beta 6 8.5 7.64E-01 MAP7 microtubule-associated protein 7 11 6.77E-01 TUBB2C tubulin, beta 2C (TUBB4B tubulin, beta 4b) 67 6.64E-01 LMO7 LIM domain 7 11 6.50E-01 LIMCH1 LIM and calponin homology domains 1 13 6.39E-01 MARK1 MAP/microtubule affinity-regulating kinase 1 2.7 5.91E-01 MAP2 microtubule-associated protein 2 20 5.72E-01 TUBB4 tubulin, beta 4 (TUBB4A, tubulin beta 4a) 8.17 5.59E-01 TUBA4A tubulin, alpha 4a 68 4.74E-01 VIM vimentin 85 4.68E-01 TUBG1 tubulin, gamma 1 8.9 4.06E-01 MAP1B microtubule-associated protein 1B 17 3.95E-01 TMOD1 tropomodulin 1 38 3.83E-01 CAMSAP1 regulated spectrin-associated 6.0 3.72E-01 protein 1

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

NDEL1 nudE nuclear distribution E homolog (A. 11 3.54E-01 nidulans)-like 1 PALLD palladin, cytoskeletal associated protein 14 3.04E-01 SPIRE1 spire homolog 1 (Drosophila) 8.1 3.00E-01 MAST2 microtubule associated serine/threonine kinase 6.0 2.98E-01 2 PDLIM7 PDZ and LIM domain 7 (enigma) 14 2.69E-01 TUBB tubulin, beta 133 2.63E-01 SEPT11 septin 11 8.0 1.77E-01 CNN1 calponin 1, basic, smooth muscle 1.0 - 1.49E+00 MNS1 meiosis-specific nuclear structural 1 1.3 - 1.34E+00 PRC1 protein regulator of cytokinesis 1 2.0 - 1.05E+00 LSP1 lymphocyte-specific protein 1 1.8 -8.06E-01 BBS4 Bardet-Biedl syndrome 4 5.8 -7.91E-01 CNN2 calponin 2 8.6 -6.43E-01 DNAH5 dynein, axonemal, heavy chain 5 1.3 -6.31E-01 NES nestin 2.0 -6.25E-01 STMN1 stathmin 1 21 -5.64E-01 IFFO2 intermediate filament family orphan 2 5.7 -5.57E-01 CNP 2',3'-cyclic nucleotide 3' phosphodiesterase 12 -5.44E-01 MKS1 Meckel syndrome, type 1 3.6 -4.99E-01 SH2B1 SH2B adaptor protein 1 15 -4.98E-01 MARCKSL1 MARCKS-like 1 42 -4.78E-01 CEP41 centrosomal protein 41kDa (TSGA14) 5.2 -4.76E-01 BBIP1 BBSome interacting protein 1 8.1 -3.89E-01 COBL cordon-bleu homolog (mouse) 12 -3.72E-01 ARL6IP5 ADP-ribosylation-like factor 6 interacting 57 -3.66E-01 protein 5 RANBP10 RAN binding protein 10 9.2 -3.42E-01 MPP1 membrane protein, palmitoylated 1, 55kDa 13 -3.36E-01 SSH3 slingshot homolog 3 (Drosophila) 9.3 -3.20E-01 IPP intracisternal A particle-promoted polypeptide 4.2 -3.08E-01 EPB41L1 erythrocyte membrane protein band 4.1-like 1 15 -2.92E-01 AMOT angiomotin 4.6 -2.71E-01 MYH10 myosin, heavy chain 10, non-muscle 14 -2.70E-01 ARL3 ADP-ribosylation factor-like 3 5.4 -2.56E-01 NUDCD3 NudC domain containing 3 15 -2.28E-01 ARHGEF11 Rho guanine nucleotide exchange factor (GEF) 7.0 -2.25E-01 11 NUMA1 nuclear mitotic apparatus protein 1 44 -2.24E-01 EPB41L5 erythrocyte membrane protein band 4.1 like 5 6.1 -2.16E-01 MTUS2 microtubule associated tumor suppressor 3.4 -2.08E-01 candidate 2 SDC1 syndecan 1 20 -1.88E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

DPYSL2 dihydropyrimidinase-like 2 21 -1.86E-01 KIF1C kinesin family member 1C 12 -1.59E-01 KIF3B kinesin family member 3B 34 -1.57E-01 SDC2 syndecan 2 19 -1.40E-01 LASP1 LIM and SH3 protein 1 57 -1.34E-01 SEPT8 septin 8 14 -1.34E-01 Peptidase/protease CAPN8 calpain 8 0.9 1.82E+00 TLL2 tolloid-like 2 0.5 1.37E+00 MMP1 matrix metallopeptidase 1 13 1.09E+00 ECEL1 endothelin converting enzyme-like 1 25 9.59E-01 CTSZ cathepsin Z 105 7.22E-01 ADAMTS9 ADAM metallopeptidase with thrombospondin 5.6 5.79E-01 type 1 motif, 9 FKBP11 FK506 binding protein 11, 19 kDa 20 5.52E-01 PITRM1 pitrilysin metallopeptidase 1 17 2.99E-01 C9orf3 chromosome 9 open reading frame 3 6.2 2.01E-01 ADAM28 ADAM metallopeptidase domain 28 1.5 - 1.38E+00 MMP11 matrix metallopeptidase 11 (stromelysin 3) 3.9 - 1.21E+00 CAPN3 calpain 3, (p94) 4.4 - 1.15E+00 MMP7 matrix metallopeptidase 7 (matrilysin, uterine) 271 - 1.03E+00 PCSK4 proprotein convertase subtilisin/kexin type 4 1.2 -9.38E-01 MMP9 matrix metallopeptidase 9 (gelatinase B, 19 -8.42E-01 92kDa gelatinase, 92kDa type IV collagenase) WFDC2 WAP four-disulfide core domain 2 43 -8.08E-01 MMP19 matrix metallopeptidase 19 2.6 -8.04E-01 PLAU plasminogen activator, urokinase 9.3 -7.90E-01 FAP fibroblast activation protein, alpha 8.7 -6.56E-01 SERPINA3 serpin peptidase inhibitor, clade A (alpha-1 1796 -6.25E-01 antiproteinase, antitrypsin), member 3 SERPINA5 serpin peptidase inhibitor, clade A (alpha-1 58 -5.61E-01 antiproteinase, antitrypsin), member 5 CTSF cathepsin F 14 -5.26E-01 CTSO cathepsin O 6.9 -5.23E-01 CTSS cathepsin S 11 -4.94E-01 SPOCK1 sparc/, cwcv and kazal-like 4.1 -4.86E-01 domains proteoglycan (testican) 1 MMP14 matrix metallopeptidase 14 (membrane- 70 -4.77E-01 inserted) PREPL prolyl endopeptidase-like 30 -4.15E-01 SERPINA4 serpin peptidase inhibitor, clade A (alpha-1 29 -3.92E-01 antiproteinase, antitrypsin), member 4 APH1B anterior pharynx defective 1 homolog B (C. 7.4 -3.88E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

elegans) MMP15 matrix metallopeptidase 15 (membrane- 18 -3.80E-01 inserted) BACE1 beta-site APP-cleaving enzyme 1 22 -3.73E-01 ERMP1 endoplasmic reticulum metallopeptidase 1 7.8 -3.51E-01 THSD4 thrombospondin, type I, domain containing 4 4.4 -3.46E-01 CTSA cathepsin A 64 -2.50E-01 CTSD cathepsin D 371 -2.07E-01 Transcription factors FOS FBJ murine osteosarcoma viral oncogene 7.8 1.26E+00 homolog ZNF165 zinc finger protein 165 2.1 9.88E-01 ETV5 ets variant 5 4.8 9.22E-01 NR4A2 nuclear receptor subfamily 4, group A, member 2.2 8.34E-01 2 CITED2 Cbp/p300-interacting transactivator, with 30 8.21E-01 Glu/Asp-rich carboxy-terminal domain, 2 FOSL1 FOS-like antigen 1 1.6 8.18E-01 BHLHA15 basic helix-loop-helix family, member a15 2.1 7.88E-01 MSC musculin 4.3 7.85E-01 TSHZ3 teashirt zinc finger homeobox 3 2.8 7.64E-01 EGR1 early growth response 1 18 6.88E-01 EPAS1 endothelial PAS domain protein 1 47 6.78E-01 TSC22D2 TSC22 domain family, member 2 4.2 6.68E-01 KLF4 Kruppel-like factor 4 (gut) 4.0 6.59E-01 CEBPG CCAAT/enhancer binding protein (C/EBP), 9.3 6.37E-01 gamma BACH2 BTB and CNC homology 1, basic 0.8 6.30E-01 zipper transcription factor TGIF1 TGFB-induced factor homeobox 1 17 5.80E-01 KLF6 Kruppel-like factor 6 25 5.57E-01 ZNF331 zinc finger protein 331 8.5 5.51E-01 ZNF395 zinc finger protein 395 35 5.08E-01 FAM50A family with sequence similarity 50, member A 23 4.92E-01 MYC v-myc myelocytomatosis viral oncogene 4.5 4.46E-01 homolog (avian) JUN jun proto-oncogene 56 3.99E-01 NFIL3 nuclear factor, interleukin 3 regulated 16 3.85E-01 SREBF2 sterol regulatory element binding transcription 34 3.74E-01 factor 2 MAFK v-maf musculoaponeurotic fibrosarcoma 14 3.65E-01 oncogene homolog K (avian) NPAS2 neuronal PAS domain protein 2 8.0 3.62E-01 EDF1 endothelial differentiation-related factor 1 105 3.37E-01 ZNF404 zinc finger protein 404 1.4 - 2.21E+00 PAX4 paired box 4 1.7 -

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

1.19E+00 CDX2 caudal type homeobox 2 3.5 -9.35E-01 ZNF763 zinc finger protein 763 1.6 -8.84E-01 ZNF85 zinc finger protein 85 2.0 -8.54E-01 ZNF691 zinc finger protein 691 4.8 -7.50E-01 AEBP1 AE binding protein 1 11 -6.98E-01 RUNX1T1 runt-related transcription factor 1; translocated 1.5 -6.69E-01 to, 1 (cyclin D-related) CREB3L4 cAMP responsive element binding protein 3- 4.3 -6.38E-01 like 4 ZFP14 zinc finger protein 14 homolog (mouse) 1.8 -6.14E-01 ZNF211 zinc finger protein 211 5.5 -5.98E-01 TFDP2 transcription factor Dp-2 (E2F dimerization 2.6 -5.60E-01 partner 2) RORC RAR-related orphan receptor C 13 -5.56E-01 ZBTB7C zinc finger and BTB domain containing 7C 1.4 -5.41E-01 KANK2 KN motif and ankyrin repeat domains 2 5.8 -5.40E-01 EAPP E2F-associated phosphoprotein 9.3 -5.37E-01 MAF v-maf musculoaponeurotic fibrosarcoma 3.7 -5.29E-01 oncogene homolog (avian) PATZ1 POZ (BTB) and AT hook containing zinc finger 12 -5.22E-01 1 SIX5 SIX homeobox 5 5.4 -5.20E-01 MAFB v-maf musculoaponeurotic fibrosarcoma 55 -5.14E-01 oncogene homolog B (avian) PBXIP1 pre-B-cell leukemia homeobox interacting 34 -5.12E-01 protein 1 VGLL4 vestigial like 4 (Drosophila) 22 -4.84E-01 ZNF671 zinc finger protein 671 4.3 -4.75E-01 BCAS3 breast carcinoma amplified sequence 3 3.1 -4.62E-01 CREG1 cellular repressor of E1A-stimulated genes 1 20 -4.61E-01 ARNT aryl hydrocarbon receptor nuclear translocator 10 -4.58E-01 ZHX3 zinc fingers and homeoboxes 3 4.8 -4.20E-01 CRY2 cryptochrome 2 (photolyase-like) 12 -4.18E-01 EYA3 eyes absent homolog 3 (Drosophila) 6.2 -4.07E-01 PBX2 pre-B-cell leukemia homeobox 2 20 -4.05E-01 ZNF462 zinc finger protein 462 4.0 -4.02E-01 ZNF429 zinc finger protein 429 4.6 -4.01E-01 ZNF192 zinc finger protein 192 5.3 -4.01E-01 NFYC nuclear transcription factor Y, gamma 9.4 -4.00E-01 ERH enhancer of rudimentary homolog (Drosophila) 30 -3.99E-01 MEIS2 Meis homeobox 2 24 -3.93E-01 ZNF280D zinc finger protein 280D 5.3 -3.90E-01 ZNF14 zinc finger protein 14 2.2 -3.81E-01 UBTF upstream binding transcription factor, RNA 16 -3.66E-01 polymerase I

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

GTF3C3 general transcription factor IIIC, polypeptide 3, 6.8 -3.42E-01 102kDa PBX3 pre-B-cell leukemia homeobox 3 13 -3.40E-01 PER3 period homolog 3 (Drosophila) 4.7 -3.39E-01 RFX1 regulatory factor X, 1 6.8 -3.34E-01 SOX13 SRY (sex determining region Y)-box 13 7.6 -3.33E-01 SMAD3 SMAD family member 3 17 -3.26E-01 GTF2I general transcription factor IIi 83 -3.15E-01 ZNF445 zinc finger protein 445 3.9 -2.94E-01 ZNF561 zinc finger protein 561 5.3 -2.89E-01 GATA6 GATA binding protein 6 6.6 -2.86E-01 ZSCAN29 zinc finger and SCAN domain containing 29 4.5 -2.81E-01 SALL2 sal-like 2 (Drosophila) 7.0 -2.74E-01 PDX1 pancreatic and duodenal homeobox 1 57 -2.71E-01 BRD8 bromodomain containing 8 12 -2.71E-01 CREBL2 cAMP responsive element binding protein-like 21 -2.26E-01 2 PTTG1IP pituitary tumor-transforming 1 interacting 70 -2.23E-01 protein BHLHE41 basic helix-loop-helix family, member e41 9.8 -2.15E-01 ELF3 E74-like factor 3 (ets domain transcription 71 -2.08E-01 factor, epithelial-specific ) NEUROD1 neuronal differentiation 1 29 -1.83E-01 GATAD1 GATA zinc finger domain containing 1 9.5 -1.59E-01 DPF2 D4, zinc and double PHD fingers family 2 12 -1.44E-01 PRDM2 PR domain containing 2, with ZNF domain 7.0 -9.55E-02 NF-B regulation SQSTM1 sequestosome 1 158 6.54E-01 NFKBIA nuclear factor of kappa light polypeptide gene 46 6.45E-01 enhancer in B-cells inhibitor, alpha IRAK2 interleukin-1 receptor-associated kinase 2 4.3 6.28E-01 STK40 serine/threonine kinase 40 17 6.08E-01 IRAK1 interleukin-1 receptor-associated kinase 1 25 4.99E-01 NFKBIZ nuclear factor of kappa light polypeptide gene 11 4.32E-01 enhancer in B-cells inhibitor, zeta NFKBIB nuclear factor of kappa light polypeptide gene 7.0 4.14E-01 enhancer in B-cells inhibitor, beta NFKB1 nuclear factor of kappa light polypeptide gene 9.5 4.08E-01 enhancer in B-cells 1 NFKB2 nuclear factor of kappa light polypeptide gene 16 3.37E-01 enhancer in B-cells 2 (p49/p100) AEBP1 AE binding protein 1 11 -6.98E-01 UNC5CL unc-5 homolog C (C. elegans)-like 35 -6.51E-01 TRIM22 tripartite motif containing 22 6.4 -6.44E-01 TMEM9 transmembrane protein 9 35 -3.60E-01 HIF1AN hypoxia inducible factor 1, alpha subunit 10 -3.33E-01 inhibitor

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

SAT2 spermidine/spermine N1-acetyltransferase 31 -3.23E-01 family member 2 TMEM9B TMEM9 domain family, member B 17 -3.20E-01 DPF2 D4, zinc and double PHD fingers family 2 12 -1.44E-01 ANXA4 A4 109 -1.34E-01 PRDM2 PR domain containing 2, with ZNF domain 7.0 -9.55E-02 PEBP1 phosphatidylethanolamine binding protein 1 127 -8.37E-02 Chemokines/cytokines/adhesion molecules/innate immunity and related proteins IL6 interleukin 6 (interferon, beta 2) 1.5 2.23E+00 IL1A interleukin 1, alpha 0.7 1.19E+00 FAM19A5 family with sequence similarity 19 (chemokine 0.8 1.17E+00 (C-C motif)-like), member A5 C2CD4A C2 calcium-dependent domain containing 4A 27 1.16E+00 ISG20 interferon stimulated exonuclease gene 20kDa 3.5 1.01E+00 IER3 immediate early response 3 205 9.88E-01 IL33 interleukin 33 1.1 9.50E-01 CXCL1 chemokine (C-X-C motif) ligand 1 (melanoma 44 9.15E-01 growth stimulating activity, alpha) IL8 interleukin 8 36 8.94E-01 TNFRSF11B tumor necrosis factor receptor superfamily, 6.2 8.36E-01 member 11b IRAK2 interleukin-1 receptor-associated kinase 2 4.3 6.28E-01 TREM1 triggering receptor expressed on myeloid cells 2.3 5.84E-01 1 CDHR3 cadherin-related family member 3 0.7 5.75E-01 SPSB1 splA/ domain and SOCS 8.6 5.56E-01 box containing 1 CTNNA2 catenin (cadherin-associated protein), alpha 2 2.1 5.51E-01 CXCL2 chemokine (C-X-C motif) ligand 2 22 5.07E-01 SELS selenoprotein S 20 5.01E-01 IRAK1 interleukin-1 receptor-associated kinase 1 25 4.99E-01 MFHAS1 malignant fibrous histiocytoma amplified 3.6 4.28E-01 sequence 1 NFIL3 nuclear factor, interleukin 3 regulated 16 3.85E-01 TNFRSF1A tumor necrosis factor receptor superfamily, 25 3.74E-01 member 1A GAB2 GRB2-associated binding protein 2 5.8 3.65E-01 C4BPB complement component 4 binding protein, beta 2.4 - 2.15E+00 TYROBP TYRO protein tyrosine kinase binding protein 5.8 - 1.94E+00 LILRB4 leukocyte immunoglobulin-like receptor, 1.5 - subfamily B (with TM and ITIM domains), 1.92E+00 member 4 DEFB1 defensin, beta 1 123 - 1.85E+00

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

KLKB1 kallikrein B, plasma (Fletcher factor) 1 2.5 - 1.26E+00 C4BPA complement component 4 binding protein, 3.6 - alpha 1.01E+00 RARRES3 retinoic acid receptor responder (tazarotene 9.5 -9.23E-01 induced) 3 C1QTNF6 C1q and tumor necrosis factor related protein 6 7.1 -9.20E-01 C1QB complement component 1, q subcomponent, B 6.0 -8.69E-01 chain SCARA3 scavenger receptor class A, member 3 1.7 -8.67E-01 CD74 CD74 molecule, major histocompatibility 238 -8.54E-01 complex, class II invariant chain IFI44 interferon-induced protein 44 2.6 -8.20E-01 VNN2 vanin 2 14 -7.98E-01 MTCP1NB mature T-cell proliferation 1 neighbor 6.8 -7.86E-01 TLR5 toll-like receptor 5 2.1 -7.30E-01 AEBP1 AE binding protein 1 11 -6.98E-01 TGFB2 transforming growth factor, beta 2 4.6 -6.56E-01 UNC5CL unc-5 homolog C (C. elegans)-like 35 -6.51E-01 FCGRT Fc fragment of IgG, receptor, transporter, 23 -6.46E-01 alpha TRIM22 tripartite motif containing 22 6.4 -6.44E-01 CDH22 cadherin 22, type 2 14 -6.08E-01 CHL1 cell adhesion molecule with homology to 3.5 -5.92E-01 L1CAM (close homolog of L1) CXCR4 chemokine (C-X-C motif) receptor 4 4.7 -5.88E-01 PILRA paired immunoglobin-like type 2 receptor alpha 8.0 -5.69E-01 CFI complement factor I 20 -5.64E-01 ZFYVE21 zinc finger, FYVE domain containing 21 14 -5.60E-01 IFIT1 interferon-induced protein with tetratricopeptide 8.0 -5.22E-01 repeats 1 VTCN1 V-set domain containing T cell activation 22 -5.06E-01 inhibitor 1 TSC22D3 TSC22 domain family, member 3 18 -4.81E-01 CD59 CD59 molecule, complement regulatory protein 94 -4.45E-01 TGFB3 transforming growth factor, beta 3 8.4 -4.11E-01 EDA ectodysplasin A 1.4 -4.11E-01 CFHR1 complement factor H-related 1 6.5 -4.01E-01 SEMA4D sema domain, immunoglobulin domain (Ig), 6.6 -3.64E-01 transmembrane domain (TM) and short cytoplasmic domain, (semaphorin) 4D TMEM9 transmembrane protein 9 35 -3.60E-01 ARL16 ADP-ribosylation factor-like 16 9.6 -3.47E-01 LCN2 lipocalin 2 331 -3.46E-01 CMTM6 CKLF-like MARVEL transmembrane domain 24 -3.23E-01 containing 6 TMEM9B TMEM9 domain family, member B 17 -3.20E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

NEO1 neogenin 1 14 -3.05E-01 PALM3 paralemmin 3 18 -2.61E-01 B2M beta-2-microglobulin 1936 -2.58E-01 PSMB8 proteasome (prosome, macropain) subunit, 33 -2.34E-01 beta type, 8 (large multifunctional peptidase 7) CD46 CD46 molecule, complement regulatory protein 72 -2.17E-01 IL20RA interleukin 20 receptor, alpha 4.6 -2.01E-01 SDC1 syndecan 1 20 -1.88E-01 F11R F11 receptor 26 -1.41E-01 CDC42SE1 CDC42 small effector 1 26 -1.40E-01 ATRN attractin 13 -1.28E-01 CD68 CD68 molecule 44 -9.91E-02 CMTM4 CKLF-like MARVEL transmembrane domain 21 -9.69E-02 containing 4 HLA-related HLA-DQA1 major histocompatibility complex, class II, DQ 1.3 - alpha 1 1.71E+00 HLA-DRA major histocompatibility complex, class II, DR 35 - alpha 1.48E+00 HLA-DPB1 major histocompatibility complex, class II, DP 3.4 - beta 1 1.40E+00 HLA-DMB major histocompatibility complex, class II, DM 3.9 - beta 1.35E+00 HLA-DPA1 major histocompatibility complex, class II, DP 6.4 - alpha 1 1.13E+00 HLA-DRB1 major histocompatibility complex, class II, DR 21 - beta 1 1.05E+00 HLA-DMA major histocompatibility complex, class II, DM 14 - alpha 1.05E+00 HLA-DRB5 major histocompatibility complex, class II, DR 8.0 -8.87E-01 beta 5 HLA-B major histocompatibility complex, class I, B 466 -3.93E-01 MR1 major histocompatibility complex, class I- 4.0 -3.83E-01 related RFX1 regulatory factor X, 1 6.8 -3.34E-01 HLA-C major histocompatibility complex, class I, C 427 -3.07E-01 Signal transduction SHC4 SHC (Src homology 2 domain containing) 0.3 2.16E+00 family, member 4 ADCY2 adenylate cyclase 2 (brain) 0.5 1.56E+00 DKK4 dickkopf homolog 4 (Xenopus laevis) 1.8 1.14E+00 PDE10A phosphodiesterase 10A 1.5 8.18E-01 RAB3B RAB3B, member RAS oncogene family 15 7.65E-01 IRS2 insulin receptor substrate 2 6.6 7.46E-01 PAK3 p21 protein (Cdc42/Rac)-activated kinase 3 8.3 7.23E-01 FICD FIC domain containing 6.4 6.75E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

TMEM158 transmembrane protein 158 4.1 6.74E-01 (gene/) PDE9A phosphodiesterase 9A 4.4 6.63E-01 RHOB ras homolog family member B 46 6.44E-01 RHOQ ras homolog family member Q 13 5.72E-01 ARHGEF2 Rho/Rac guanine nucleotide exchange factor 7.3 5.52E-01 (GEF) 2 MICALL1 MICAL-like 1 4.6 4.80E-01 KIAA1244 KIAA1244 15 4.28E-01 AXIN1 axin 1 5.6 4.24E-01 RAP2B RAP2B, member of RAS oncogene family 5.0 4.02E-01 RASSF1 Ras association (RalGDS/AF-6) domain family 7.1 3.99E-01 member 1 ARHGEF5 Rho guanine nucleotide exchange factor (GEF) 19 3.49E-01 5 TBL2 transducin (beta)-like 2 8.5 3.29E-01 PKIA protein kinase (cAMP-dependent, catalytic) 2.2 2.92E-01 inhibitor alpha YWHAH tyrosine 3-monooxygenase/tryptophan 5- 79 2.87E-01 monooxygenase activation protein, eta polypeptide RPTOR regulatory associated protein of MTOR, 4.1 2.43E-01 complex 1 RAPGEF1 Rap guanine nucleotide exchange factor (GEF) 13 2.18E-01 1 GNAS GNAS complex locus 1432 1.92E-01 GNB1 guanine nucleotide binding protein (G protein), 89 6.78E-02 beta polypeptide 1 TYROBP TYRO protein tyrosine kinase binding protein 5.8 - 1.94E+00 LPAR6 lysophosphatidic acid receptor 6 2.2 - 1.31E+00 ARHGDIB Rho GDP dissociation inhibitor (GDI) beta 5.0 - 1.21E+00 FFAR3 free fatty acid receptor 3 2.0 - 1.17E+00 SMO smoothened, frizzled family receptor 3.1 - 1.04E+00 DOK3 docking protein 3 1.3 - 1.01E+00 SFRP5 secreted frizzled-related protein 5 27 -9.40E-01 CALML4 calmodulin-like 4 3.1 -8.83E-01 VANGL2 vang-like 2 (van gogh, Drosophila) 3.5 -8.51E-01 PIK3IP1 phosphoinositide-3-kinase interacting protein 1 9.9 -8.45E-01 LGALS9 lectin, galactoside-binding, soluble, 9 13 -8.25E-01 SRGAP3 SLIT-ROBO Rho GTPase activating protein 3 2.3 -8.18E-01 PRKAR2B protein kinase, cAMP-dependent, regulatory, 2.9 -7.97E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

type II, beta PLCH2 phospholipase C, eta 2 5.2 -7.90E-01 RASA4 RAS p21 protein activator 4 34 -7.85E-01 RHOV ras homolog family member V 34 -7.17E-01 PLCD1 phospholipase C, delta 1 4.8 -7.12E-01 RASL11A RAS-like, family 11, member A 6.0 -7.09E-01 FARP2 FERM, RhoGEF and pleckstrin domain protein 5.6 -6.96E-01 2 RASSF4 Ras association (RalGDS/AF-6) domain family 15 -6.81E-01 member 4 SRGAP1 SLIT-ROBO Rho GTPase activating protein 1 2.5 -6.77E-01 PIK3C2B phosphoinositide-3-kinase, class 2, beta 3.4 -6.54E-01 polypeptide ARRDC2 arrestin domain containing 2 16 -6.45E-01 PLD1 phospholipase D1, phosphatidylcholine- 5.1 -6.20E-01 specific GIPC2 GIPC PDZ domain containing family, member 3.1 -6.16E-01 2 ASAP3 ArfGAP with SH3 domain, ankyrin repeat and 10 -5.99E-01 PH domain 3 SUFU suppressor of fused homolog (Drosophila) 3.9 -5.99E-01 ARRDC4 arrestin domain containing 4 12 -5.90E-01 MAP3K15 mitogen-activated protein kinase kinase kinase 4.1 -5.84E-01 15 PPAP2B phosphatidic acid phosphatase type 2B 8.5 -5.66E-01 RASGRP1 RAS guanyl releasing protein 1 (calcium and 4.7 -5.38E-01 DAG-regulated) TXNIP thioredoxin interacting protein 118 -5.20E-01 SH2B1 SH2B adaptor protein 1 15 -4.98E-01 RGN (senescence marker protein-30) 7.0 -4.95E-01 FZD4 frizzled family receptor 4 1.8 -4.89E-01 NOTCH3 notch 3 11 -4.82E-01 CTNNBIP1 catenin, beta interacting protein 1 6.8 -4.82E-01 MARCKSL1 MARCKS-like 1 42 -4.78E-01 IP6K2 inositol hexakisphosphate kinase 2 36 -4.44E-01 PILRB paired immunoglobin-like type 2 receptor beta 44 -4.44E-01 ADCY5 adenylate cyclase 5 8.0 -4.42E-01 MCF2L MCF.2 cell line derived transforming 7.4 -4.31E-01 sequence-like DMXL2 Dmx-like 2 5.0 -4.30E-01 PORCN porcupine homolog (Drosophila) 2.8 -4.30E-01 RAB20 RAB20, member RAS oncogene family 15 -4.27E-01 ELMO2 engulfment and cell motility 2 13 -4.03E-01 ARHGEF3 Rho guanine nucleotide exchange factor (GEF) 7.3 -4.02E-01 3 RAB2B RAB2B, member RAS oncogene family 7.0 -3.93E-01 RAB11FIP4 RAB11 family interacting protein 4 (class II) 8.0 -3.91E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

APH1B anterior pharynx defective 1 homolog B (C. 7.4 -3.88E-01 elegans) ARHGEF10L Rho guanine nucleotide exchange factor (GEF) 19 -3.78E-01 10-like ASB8 ankyrin repeat and SOCS box containing 8 9.5 -3.76E-01 ARL8B ADP-ribosylation factor-like 8B 19 -3.74E-01 ARL6IP5 ADP-ribosylation-like factor 6 interacting 57 -3.66E-01 protein 5 ARL16 ADP-ribosylation factor-like 16 9.6 -3.47E-01 RANBP10 RAN binding protein 10 6.0 -3.42E-01 DLG2 discs, large homolog 2 (Drosophila) 1.4 -3.37E-01 MPP1 membrane protein, palmitoylated 1, 55kDa 13 -3.36E-01 SNX29 sorting nexin 29 5.5 -3.35E-01 HIF1AN hypoxia inducible factor 1, alpha subunit 10 -3.33E-01 inhibitor RGL1 ral guanine nucleotide dissociation stimulator- 5.1 -3.26E-01 like 1 GNG10 guanine nucleotide binding protein (G protein), 18 -3.12E-01 gamma 10 ARHGEF9 Cdc42 guanine nucleotide exchange factor 8.8 -2.93E-01 (GEF) 9 DIXDC1 DIX domain containing 1 2.6 -2.92E-01 SRGAP2 SLIT-ROBO Rho GTPase activating protein 2 9.5 -2.83E-01 NOTCH2 notch 2 13 -2.68E-01 ARL3 ADP-ribosylation factor-like 3 5.4 -2.56E-01 CHN1 (chimaerin) 1 5.9 -2.54E-01 MADD MAP-kinase activating death domain 13 -2.43E-01 RAP1GAP RAP1 GTPase activating protein 54 -2.37E-01 FLOT2 flotillin 2 37 -2.36E-01 ARHGEF11 Rho guanine nucleotide exchange factor (GEF) 7.0 -2.25E-01 11 RAB11A RAB11A, member RAS oncogene family 25 -2.11E-01 RALBP1 ralA binding protein 1 13 -1.83E-01 ADCYAP1 adenylate cyclase activating polypeptide 1 14 -1.78E-01 (pituitary) CDS2 CDP-diacylglycerol synthase (phosphatidate 9.3 -1.68E-01 cytidylyltransferase) 2 CALM2 calmodulin 2 (phosphorylase kinase, delta) 122 -1.62E-01 TSPAN3 tetraspanin 3 78 -1.60E-01 WLS wntless homolog (Drosophila) 44 -9.65E-02 PEBP1 phosphatidylethanolamine binding protein 1 127 -8.37E-02 Kinases/phosphatases DUSP1 dual specificity phosphatase 1 20 9.12E-01 SIK1 salt-inducible kinase 1 7.7 8.60E-01 STK40 serine/threonine kinase 40 17 6.08E-01 MAPK7 mitogen-activated protein kinase 7 3.6 5.15E-01 DUSP4 dual specificity phosphatase 4 15 4.67E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

OXSR1 oxidative-stress responsive 1 9.7 3.29E-01 MAST2 microtubule associated serine/threonine kinase 6.0 2.98E-01 2 RPS6KA1 ribosomal protein S6 kinase, 90kDa, 9.0 2.16E-01 polypeptide 1 DUSP3 dual specificity phosphatase 3 25 1.01E-01 PHPT1 phosphohistidine phosphatase 1 37 7.11E-02 PPP1R1B protein phosphatase 1, regulatory (inhibitor) 34 - subunit 1B 1.33E+00 ELFN2 extracellular leucine-rich repeat and fibronectin 7.8 -8.86E-01 type III domain containing 2 HDHD1 haloacid dehalogenase-like hydrolase domain 7.1 -6.41E-01 containing 1 PPP1R3E protein phosphatase 1, regulatory subunit 3E 5.6 -5.28E-01 STK38 serine/threonine kinase 38 9.9 -4.94E-01 NEK6 NIMA (never in mitosis gene a)-related kinase 13 -4.80E-01 6 PIP4K2B phosphatidylinositol-5-phosphate 4-kinase, 12 -4.70E-01 type II, beta WNK2 WNK lysine deficient protein kinase 2 8.9 -4.67E-01 HUNK hormonally up-regulated Neu-associated 3.7 -4.51E-01 kinase PPP1R12B protein phosphatase 1, regulatory subunit 12B 3.9 -4.50E-01 SGK2 serum/glucocorticoid regulated kinase 2 5.6 -4.40E-01 PPP2R5A protein phosphatase 2, regulatory subunit B', 11 -4.32E-01 alpha AGK acylglycerol kinase 5.2 -4.28E-01 EYA3 eyes absent homolog 3 (Drosophila) 6.2 -4.07E-01 PPM1H protein phosphatase, Mg2+/Mn2+ dependent, 7.1 -3.77E-01 1H VRK3 vaccinia related kinase 3 6.1 -3.40E-01 DSTYK dual serine/threonine and tyrosine protein 4.9 -3.33E-01 kinase PTPN18 protein tyrosine phosphatase, non-receptor 18 -3.15E-01 type 18 (brain-derived) PRKCH protein kinase C, eta 5.1 -3.00E-01 CAMK2N1 calcium/calmodulin-dependent protein kinase II 35 -2.94E-01 inhibitor 1 MPPE1 metallophosphoesterase 1 6.1 -2.86E-01 CTDSPL CTD (carboxy-terminal domain, RNA 10 -2.79E-01 polymerase II, polypeptide A) small phosphatase-like PTPRN2 protein tyrosine phosphatase, receptor type, N 50 -2.49E-01 polypeptide 2 INPP4A inositol polyphosphate-4-phosphatase, type I, 4.1 -2.30E-01 107kDa ARPP19 cAMP-regulated phosphoprotein, 19kDa 29 -2.28E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

PIP5K1A phosphatidylinositol-4-phosphate 5-kinase, 10 -2.23E-01 type I, alpha PTPRA protein tyrosine phosphatase, receptor type, A 17 -1.47E-01 Transcription regulation/alternative splicing EGR4 early growth response 4 0.9 9.96E-01 ELL2 elongation factor, RNA polymerase II, 2 20 7.05E-01 SRSF3 serine/arginine-rich splicing factor 3 18 5.67E-01 ESRP1 epithelial splicing regulatory protein 1 13 4.85E-01 DHX16 DEAH (Asp-Glu-Ala-His) box polypeptide 16 8.4 3.84E-01 DDX39A DEAD (Asp-Glu-Ala-Asp) box polypeptide 39A 9.4 3.76E-01 AKAP17A A kinase (PRKA) anchor protein 17A chr:X 15 3.06E-01 AKAP17A A kinase (PRKA) anchor protein 17A chr:Y 15 2.99E-01 SCAF1 SR-related CTD-associated factor 1 14 2.02E-01 TCERG1L transcription elongation regulator 1-like 2.7 -9.39E-01 NANOS1 nanos homolog 1 (Drosophila) 2.9 -9.13E-01 RBFOX3 RNA binding protein, fox-1 homolog (C. 1.6 -8.98E-01 elegans) 3 MEPCE methylphosphate capping enzyme 16 -7.22E-01 DIS3L DIS3 mitotic control homolog (S. cerevisiae)- 6.7 -6.05E-01 like RBM14 RNA binding motif protein 14 8.7 -5.83E-01 NOVA1 neuro-oncological ventral antigen 1 3.8 -5.74E-01 CNP 2',3'-cyclic nucleotide 3' phosphodiesterase 12 -5.44E-01 CELF1 CUGBP, Elav-like family member 1 22 -5.32E-01 ADI1 acireductone dioxygenase 1 26 -4.94E-01 TSC22D3 TSC22 domain family, member 3 18 -4.81E-01 CSTF3 cleavage stimulation factor, 3' pre-RNA, 8.5 -4.20E-01 subunit 3, 77kDa HMGN3 high mobility group nucleosomal binding 32 -3.77E-01 domain 3 RBFOX2 RNA binding protein, fox-1 homolog (C. 14 -3.69E-01 elegans) 2 SNRPN small nuclear ribonucleoprotein polypeptide N 94 -3.67E-01 UBTF upstream binding transcription factor, RNA 16 -3.66E-01 polymerase I SMARCD3 SWI/SNF related, matrix associated, actin 13 -3.61E-01 dependent regulator of chromatin, subfamily d, member 3 PHC1 polyhomeotic homolog 1 (Drosophila) 20 -3.51E-01 POLDIP3 polymerase (DNA-directed), delta interacting 22 -3.47E-01 protein 3 THOC7 THO complex 7 homolog (Drosophila) 16 -3.46E-01 HMGN1 high mobility group nucleosome binding 54 -3.42E-01 domain 1 SMARCD2 SWI/SNF related, matrix associated, actin 14 -3.25E-01 dependent regulator of chromatin, subfamily d, member 2

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

SMARCE1 SWI/SNF related, matrix associated, actin 22 -3.23E-01 dependent regulator of chromatin, subfamily e, member 1 DDX17 DEAD (Asp-Glu-Ala-Asp) box helicase 17 96 -3.18E-01 SUGP2 SURP and G patch domain containing 2 17 -2.99E-01 LDB1 LIM domain binding 1 22 -2.82E-01 TJP2 tight junction protein 2 (zonula occludens-2) 24 -2.81E-01 PCGF3 polycomb group ring finger 3 12 -2.77E-01 POLR2G polymerase (RNA) II (DNA directed) 18 -2.59E-01 polypeptide G TIA1 TIA1 cytotoxic granule-associated RNA binding 8.9 -2.26E-01 protein HNRNPH3 heterogeneous nuclear ribonucleoprotein H3 29 -1.76E-01 (2H9) RBM5 RNA binding motif protein 5 22 -1.38E-01 Non-coding RNA SLC7A5P2 solute carrier family 7 (cationic amino acid 2.3 1.60E+00 transporter, y+ system), member 5 pseudogene 2 SLC7A5P1 solute carrier family 7 (cationic amino acid 8.2 1.54E+00 transporter, y+ system), member 5 pseudogene 1 LINC00473 long intergenic non-protein coding RNA 473 7.0 1.35E+00 ZSCAN12P1 zinc finger and SCAN domain containing 12 0.8 1.20E+00 pseudogene 1 FABP5P3 fatty acid binding protein 5 pseudogene 3 9.6 9.36E-01 DGCR5 DiGeorge syndrome critical region gene 5 3.7 5.87E-01 (non-protein coding) MALAT1 metastasis associated lung adenocarcinoma 306 5.75E-01 transcript 1 TUBA4B tubulin, alpha 4b (pseudogene) 13 5.43E-01 HSP90B3P heat shock protein 90kDa beta (Grp94), 23 4.40E-01 member 3, pseudogene LOC1001709 glucuronidase, beta pseudogene 0.6 2.59E-01 39 LINC00671 long intergenic non-protein coding RNA 671 2.5 - (LOC388387) 1.97E+00 SSTR5-AS1 SSTR5 antisense RNA 1 (LOC146336) 8.7 - 1.96E+00 M1 uncharacterized LOC100507027 3.8 - 1.90E+00 LOC1005056 uncharacterized LOC100505624 1.2 - 24 1.80E+00 LOC1002717 uncharacterized LOC100271722 1.1 - 22 1.66E+00 LOC645638 WDNM1-like pseudogene 2.2 - 1.25E+00

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

LOC441204 uncharacterized LOC441204 1.3 - 1.10E+00 PCBP1-AS1 PCBP1 antisense RNA 1 1.1 - 1.09E+00 STARD4- STARD4 antisense RNA 1 (LOC100505678) 1.0 - AS1 1.09E+00 LINC00622 long intergenic non-protein coding RNA 622 3.5 -9.46E-01 (LOC644242) PTOV1-AS1 PTOV1 antisense RNA 1 (LOC100506033) 4.7 -8.86E-01 TRAPPC2P1 trafficking protein particle complex 2 6.0 -8.58E-01 pseudogene 1 HNRNPUL2- HNRNPUL2-BSCL2 readthrough 13 -7.67E-01 BSCL2 MST1P2 macrophage stimulating 1 (hepatocyte growth 24 -7.55E-01 factor-like) pseudogene 2 GGT3P gamma-glutamyltransferase 3 pseudogene 11 -6.72E-01 SENP3- SENP3-EIF4A1 readthrough 1.3 -6.43E-01 EIF4A1 LOC93622 Morf4 family associated protein 1-like 1 4.2 -6.24E-01 pseudogene LINC00339 long intergenic non-protein coding RNA 339 6.9 -5.81E-01 (HSPC157) HCP5 HLA complex P5 (non-protein coding) 8.1 -5.63E-01 LOC729678 uncharacterized LOC729678 8.7 -5.51E-01 C15orf48 chromosome 15 open reading frame 48 63 -5.32E-01 SKP1P2 S-phase kinase-associated protein 1 16 -4.51E-01 pseudogene 2 (loc728622) CD27-AS1 CD27 antisense RNA 1 (non-protein coding) 5.1 -4.32E-01 (loc678655) GTF2IRD2P GTF2I repeat domain containing 2 12 -4.18E-01 1 pseudogene 1 FKBP9L FK506 binding protein 9-like 28 -3.79E-01 LRRC37A4P leucine rich repeat containing 37, member A4, 18 -3.60E-01 pseudogene LOC220906 uncharacterized LOC220906 11 -3.49E-01 LOC147727 uncharacterized LOC147727 5.9 -3.48E-01 RPSAP9 ribosomal protein SA pseudogene 9 67 -3.37E-01 HCG11 HLA complex group 11 (non-protein coding) 8.7 -3.21E-01 LOC1005065 uncharacterized LOC100506548 6.8 -3.09E-01 48 MST1P9 macrophage stimulating 1 (hepatocyte growth 13 -2.85E-01 factor-like) pseudogene 9 DHRS4-AS1 DHRS4 antisense RNA 1 (non-protein coding) 8.2 -2.43E-01 (C14orf167) TUG1 taurine upregulated 1 (non-protein coding) 19 -2.35E-01 RPSAP58 ribosomal protein SA pseudogene 58 177 -6.95E-02 Epigenetic regulation

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

SFMBT2 Scm-like with four mbt domains 2 0.8 6.17E-01 ARID5B AT rich interactive domain 5B (MRF1-like) 5.9 4.71E-01 CBX4 chromobox homolog 4 6.1 3.76E-01 KDM6B lysine (K)-specific demethylase 6B 9.6 2.73E-01 APOBEC3C apolipoprotein B mRNA editing enzyme, 7.5 -8.13E-01 catalytic polypeptide-like 3C APOBEC3D apolipoprotein B mRNA editing enzyme, 2.0 -7.47E-01 catalytic polypeptide-like 3D H2AFV H2A histone family, member V 23 -6.97E-01 CBX7 chromobox homolog 7 18 -6.64E-01 HIST1H2AC histone cluster 1, H2ac 14 -6.75E-01 APOBEC2 apolipoprotein B mRNA editing enzyme, 4.4 -5.29E-01 catalytic polypeptide-like 2 ZMYM3 zinc finger, MYM-type 3 11 -5.26E-01 ING3 inhibitor of growth family, member 3 4.5 -5.17E-01 DPY30 dpy-30 homolog (C. elegans) 15 -5.08E-01 SETDB2 SET domain, bifurcated 2 1.7 -4.52E-01 ERI2 ERI1 exoribonuclease family member 2 2.5 -4.48E-01 FKBP3 FK506 binding protein 3, 25kDa 14 -3.69E-01 ING4 inhibitor of growth family, member 4 9.2 -3.28E-01 EZH1 enhancer of zeste homolog 1 (Drosophila) 10 -3.05E-01 APOBEC3F apolipoprotein B mRNA editing enzyme, 5.1 -2.79E-01 catalytic polypeptide-like 3F PCGF3 polycomb group ring finger 3 12 -2.77E-01 CBX3 chromobox homolog 3 18 -2.52E-01 PHF8 PHD finger protein 8 10 -2.52E-01 MPHOSPH8 M-phase phosphoprotein 8 10 -2.47E-01 KANSL1 KAT8 regulatory NSL complex subunit 1 10 -2.38E-01 (KIAA1267) CBX1 chromobox homolog 1 19 -2.18E-01 PHF15 PHD finger protein 15 11 -1.41E-01 WDR82 WD repeat domain 82 22 -1.24E-01 PRDM2 PR domain containing 2, with ZNF domain 7.0 -9.55E-02 Oxidative stress/DNA damage response HMOX1 heme oxygenase (decycling) 1 6.2 7.85E-01 DDIT4 DNA-damage-inducible transcript 4 83 7.42E-01 TXN thioredoxin 19 7.05E-01 MSRA methionine sulfoxide reductase A 3.8 6.18E-01 SESN2 sestrin 2 13 5.95E-01 TXNDC11 thioredoxin domain containing 11 15 4.38E-01 RAD23A RAD23 homolog A (S. cerevisiae) 29 4.22E-01 TALDO1 transaldolase 1 29 3.48E-01 OXSR1 oxidative-stress responsive 1 97 3.29E-01 PON1 paraoxonase 1 4.9 - 2.06E+00 GSTM1 glutathione S-transferase mu 1 15 -

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

1.03E+00 SEPP1 selenoprotein P, plasma, 1 20 -8.76E-01 SCARA3 scavenger receptor class A, member 3 1.7 -8.67E-01 DDB2 damage-specific DNA binding protein 2, 48kDa 14 -8.03E-01 ENC1 ectodermal-neural cortex 1 (with BTB-like 18 -7.31E-01 domain) GSTM4 glutathione S-transferase mu 4 12 -7.07E-01 GGCT gamma-glutamylcyclotransferase 14 -7.04E-01 PCYOX1L prenylcysteine oxidase 1 like 5.7 -5.49E-01 GSTK1 glutathione S-transferase kappa 1 55 -5.12E-01 GTF2H5 general transcription factor IIH, polypeptide 5 2.3 -5.12E-01 PARP3 poly (ADP-ribose) polymerase family, member 5.7 -4.33E-01 3 RPA2 replication protein A2, 32kDa 9.4 -4.14E-01 EEPD1 endonuclease/exonuclease/phosphatase 4.5 -3.79E-01 family domain containing 1 XPC xeroderma pigmentosum, complementation 13 -2.54E-01 group C RPA1 replication protein A1, 70kDa 8.4 -2.01E-01 PRDX6 peroxiredoxin 6 34 -1.31E-01 Apoptosis IER3 immediate early response 3 205 9.88E-01 GRAMD4 GRAM domain containing 4 14 7.55E-01 GADD45A growth arrest and DNA-damage-inducible, 16 7.50E-01 alpha TP53BP2 tumor protein p53 binding protein, 2 7.8 5.11E-01 BAG1 BCL2-associated athanogene 12 2.68E-01 LRG1 leucine-rich alpha-2-glycoprotein 1 28 - 1.01E+00 DNASE1 deoxyribonuclease I 1.3 -9.55E-01 BMF Bcl2 modifying factor 17 -8.62E-01 FAS Fas (TNF receptor superfamily, member 6) 2.7 -8.23E-01 CFLAR CASP8 and FADD-like apoptosis regulator 21 -8.02E-01 UACA uveal autoantigen with coiled-coil domains and 3.0 -7.99E-01 ankyrin repeats MAP3K15 mitogen-activated protein kinase kinase kinase 4.1 -5.84E-01 15 CASP10 caspase 10, apoptosis-related cysteine 2.8 -5.55E-01 peptidase EAPP E2F-associated phosphoprotein 9.3 -5.37E-01 TP53INP1 tumor protein p53 inducible nuclear protein 1 7.5 -5.28E-01 PATZ1 POZ (BTB) and AT hook containing zinc finger 12 -5.22E-01 1 TXNIP thioredoxin interacting protein 118 -5.20E-01 RGN regucalcin (senescence marker protein-30) 7.0 -4.95E-01 NEK6 NIMA (never in mitosis gene a)-related kinase 13 -4.80E-01 6

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

CREG1 cellular repressor of E1A-stimulated genes 1 20 -4.61E-01 IP6K2 inositol hexakisphosphate kinase 2 36 -4.44E-01 CASP2 caspase 2, apoptosis-related cysteine 6.3 -4.20E-01 peptidase ELMO2 engulfment and cell motility 2 13 -4.03E-01 ARL6IP5 ADP-ribosylation-like factor 6 interacting 57 -3.66E-01 protein 5 LCN2 lipocalin 2 331 -3.46E-01 DSTYK dual serine/threonine and tyrosine protein 4.9 -3.33E-01 kinase SMARCE1 SWI/SNF related, matrix associated, actin 22 -3.23E-01 dependent regulator of chromatin, subfamily e, member 1 DDX17 DEAD (Asp-Glu-Ala-Asp) box helicase 17 96 -3.18E-01 BCL2L11 BCL2-like 11 (apoptosis facilitator) 7.7 -3.06E-01 SNN stannin 7.0 -3.02E-01 TJP2 tight junction protein 2 24 -2.81E-01 SALL2 sal-like 2 (Drosophila) 7.0 -2.74E-01 XPC xeroderma pigmentosum, complementation 13 -2.54E-01 group C NISCH nischarin 20 -2.45E-01 MADD MAP-kinase activating death domain 13 -2.43E-01 DCAF7 DDB1 and CUL4 associated factor 7 16 -2.39E-01 INPP4A inositol polyphosphate-4-phosphatase, type I, 4.1 -2.30E-01 107kDa TIA1 TIA1 cytotoxic granule-associated RNA binding 8.9 -2.26E-01 protein PERP PERP, TP53 apoptosis effector 46 -2.23E-01 PIP5K1A phosphatidylinositol-4-phosphate 5-kinase, 10 -2.23E-01 type I, alpha TM7SF3 transmembrane 7 superfamily member 3 15 -2.21E-01 SORT1 sortilin 1 22 -2.03E-01 EIF4G2 eukaryotic translation initiation factor 4 gamma, 122 -2.02E-01 2 PRPS1 phosphoribosyl pyrophosphate synthetase 1 16 -1.90E-01 ADCYAP1 adenylate cyclase activating polypeptide 1 14 -1.78E-01 (pituitary) NMT1 N-myristoyltransferase 1 17 -1.73E-01 DPF2 D4, zinc and double PHD fingers family 2 12 -1.44E-01 ANXA4 annexin A4 109 -1.34E-01 PRDX6 peroxiredoxin 6 34 -1.31E-01 Cell cycle PCDHGA5 protocadherin gamma subfamily A, 5 0.4 1.83E+00 PCDHAC2 protocadherin alpha subfamily C, 2 1.8 1.12E+00 CDKN1C cyclin-dependent kinase inhibitor 1C 35 8.30E-01 PCDHGA3 protocadherin gamma subfamily A, 3 1.0 6.66E-01 PCDHB7 protocadherin beta 7 0.8 6.58E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

LGALS4 lectin, galactoside-binding, soluble, 4 41 3.20E-01 ADRM1 adhesion regulating molecule 1 22 2.91E-01 ITGA3 integrin, alpha 3 (antigen CD49C, alpha 3 39 1.98E-01 subunit of VLA-3 receptor) GAS6 growth arrest-specific 6 19 -7.65E-01 CETN3 centrin, EF-hand protein, 3 4.6 -7.62E-01 REG1B regenerating islet-derived 1 beta 1803 -7.08E-01 DPH1 DPH1 homolog (S. cerevisiae) 11 -6.82E-01 CCNDBP1 cyclin D-type binding-protein 1 6.3 -6.67E-01 CBX7 chromobox homolog 7 18 -6.64E-01 CNNM3 cyclin M3 13 -5.75E-01 TFDP2 transcription factor Dp-2 (E2F dimerization 2.6 -5.60E-01 partner 2) REG3G regenerating islet-derived 3 gamma 49 -5.42E-01 CEP68 centrosomal protein 68kDa 3.9 -5.42E-01 EAPP E2F-associated phosphoprotein 9.3 -5.37E-01 URGCP upregulator of cell proliferation 14 -4.94E-01 REG3A regenerating islet-derived 3 alpha 2785 -4.82E-01 NEK6 NIMA (never in mitosis gene a)-related kinase 13 -4.80E-01 6 CEP41 centrosomal protein 41kDa (TSGA14) 5.2 -4.76E-01 HEPACAM2 HEPACAM family member 2 27 -4.56E-01 CLMN calmin (calponin-like, transmembrane) 11 -4.45E-01 RPA2 replication protein A2, 32kDa 9.4 -4.14E-01 NUP43 nucleoporin 43kDa 5.4 -4.09E-01 ERH enhancer of rudimentary homolog (Drosophila) 30 -3.99E-01 HAUS4 HAUS augmin-like complex, subunit 4 9.4 -3.89E-01 MCM6 minichromosome maintenance complex 1.9 -3.68E-01 component 6 IST1 increased sodium tolerance 1 homolog (yeast) 37 -3.46E-01 (KIAA0174) SNN stannin 7.0 -3.02E-01 CDKN2C cyclin-dependent kinase inhibitor 2C 5.8 -3.02E-01 CCNG1 cyclin G1 27 -3.01E-01 CAMK2N1 calcium/calmodulin-dependent protein kinase II 35 -2.94E-01 inhibitor 1 CTDSPL CTD (carboxy-terminal domain, RNA 10 -2.79E-01 polymerase II, polypeptide A) small phosphatase-like ARPP19 cAMP-regulated phosphoprotein, 19kDa 29 -2.28E-01 NUMA1 nuclear mitotic apparatus protein 1 44 -2.24E-01 PTTG1IP pituitary tumor-transforming 1 interacting 70 -2.23E-01 protein TMEM30B transmembrane protein 30B 22 -2.10E-01 RALBP1 ralA binding protein 1 13 -1.83E-01 CMTM4 CKLF-like MARVEL transmembrane domain 21 -9.69E-02 containing 4

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Extracellular matrix MFAP4 microfibrillar-associated protein 4 1.2 - 2.43E+00 CHAD chondroadherin 1.7 - 1.80E+00 ITGB1BP1 integrin beta 1 binding protein 1 32 - 1.42E+00 OLFML2B olfactomedin-like 2B 1.2 - 1.04E+00 EFEMP1 EGF containing fibulin-like 8.2 -9.49E-01 protein 1 ABI3BP ABI family, member 3 (NESH) binding protein 1.5 -9.32E-01 COL10A1 collagen, type X, alpha 1 1.6 - 1.20E+00 VWA2 von Willebrand factor A domain containing 2 1.4 - 1.04E+00 COL1A2 collagen, type I, alpha 2 39 -5.05E-01 COL4A2 collagen, type IV, alpha 2 23 -4.39E-01 COL1A1 collagen, type I, alpha 1 60 -3.59E-01 COL18A1 collagen, type XVIII, alpha 1 34 -3.43E-01 MATN2 matrilin 2 4.4 -5.98E-01 ENG endoglin (transforming growth factor beta 8.4 -5.95E-01 receptor complex) CD82 CD82 molecule 24 -5.73E-01 SPARC secreted protein, acidic, cysteine-rich 25 -5.72E-01 (osteonectin) PCOLCE procollagen C-endopeptidase enhancer 8.3 -5.44E-01 MMP9 matrix metallopeptidase 9 (gelatinase B, 19 -8.42E-01 92kDa gelatinase, 92kDa type IV collagenase) ITGB4 integrin, beta 4 11 -3.32E-01 HSPG2 heparan sulfate proteoglycan 2 (perlecan) 14 -2.46E-01 CHST3 carbohydrate (chondroitin 6) sulfotransferase 3 3.3 -2.07E-01 CRTAP cartilage associated protein 9.2 -2.02E-01 AGRN agrin 46 -1.40E-01 Other/unknown function PTGS2 prostaglandin-endoperoxide synthase 2 1.7 2.25E+00 (prostaglandin G/H synthase and cyclooxygenase) AREG amphiregulin 5.5 2.19E+00 SFN stratifin 3.2 1.58E+00 C6orf223 chromosome 6 open reading frame 223 0.8 1.49E+00 PRG4 proteoglycan 4 23 1.43E+00 LOC441177 hypothetical LOC441177 2.1 1.38E+00 FEZ1 fasciculation and elongation protein zeta 1 2.0 1.24E+00 (zygin I) STC2 stanniocalcin 2 4.0 1.13E+00 CCDC48 coiled-coil domain containing 48 0.6 1.04E+00

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

KIAA1199 KIAA1199 2.3 9.94E-01 SMOC1 SPARC related modular calcium binding 1 12 9.76E-01 TTC39B tetratricopeptide repeat domain 39B 1.9 9.41E-01 PHLDA2 pleckstrin homology-like domain, family A, 3.8 9.13E-01 member 2 TRIM9 tripartite motif containing 9 4.3 9.05E-01 JPH1 junctophilin 1 1.33 8.11E-01 TRIM16 tripartite motif containing 16 6.0 7.84E-01 FLJ35776 hypothetical LOC649446 2.6 7.76E-01 GTPBP2 GTP binding protein 2 20 7.75E-01 KLHDC8A kelch domain containing 8A 12 7.47E-01 TRIM16L tripartite motif containing 16-like 7.0 7.42E-01 VAT1L vesicle amine transport protein 1 homolog (T. 6.5 7.04E-01 californica)-like LOC1005070 uncharacterized LOC100507034 2.1 6.90E-01 34 TCP11L2 t-complex 11 (mouse)-like 2 4.4 6.73E-01 LOC154761 hypothetical LOC154761 8.5 6.58E-01 EML5 echinoderm microtubule associated protein like 1.7 6.48E-01 5 SMOX spermine oxidase 6.0 6.36E-01 KIAA0319 KIAA0319 1.4 6.27E-01 BEX2 brain expressed X-linked 2 50 6.20E-01 C10orf108 chromosome 10 open reading frame 108 1.9 6.16E-01 PEG10 paternally expressed 10 25 6.15E-01 MRPL23 mitochondrial ribosomal protein L23 16 6.09E-01 C12orf68 chromosome 12 open reading frame 68 2.6 6.08E-01 FAM135A family with sequence similarity 135, member A 3.5 6.03E-01 ZNF841 zinc finger protein 841 4.2 6.03E-01 CSRNP1 cysteine-serine-rich nuclear protein 1 12 5.97E-01 NEDD9 neural precursor cell expressed, 7.8 5.95E-01 developmentally down-regulated 9 GNL3 guanine nucleotide binding protein-like 3 14 5.94E-01 (nucleolar) LURAP1L leucine rich adaptor protein 1-like 14 5.94E-01 FEM1C fem-1 homolog c (C. elegans) 3.9 5.92E-01 PIGA phosphatidylinositol glycan anchor 3.4 5.79E-01 biosynthesis, class A WDR25 WD repeat domain 25 4.7 5.72E-01 MTRF1L mitochondrial translational 1-like 4.7 5.58E-01

TMEM200A transmembrane protein 200A 7.0 5.56E-01 NACAD NAC alpha domain containing 20 5.46E-01 CCDC104 coiled-coil domain containing 104 17 5.41E-01 TMEM39A transmembrane protein 39A 7.3 5.33E-01 SOGA2 SOGA family member 2 3.4 5.32E-01 CNIH4 cornichon homolog 4 (Drosophila) 14 5.20E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

TSPYL2 TSPY-like 2 20 4.96E-01 AMPD2 adenosine monophosphate deaminase 2 29 4.93E-01 BHLHE40 basic helix-loop-helix family, member e40 57 4.87E-01 BEX5 brain expressed, X-linked 5 8.6 4.83E-01 PARM1 prostate androgen-regulated mucin-like protein 48 4.68E-01 1 DNTTIP2 deoxynucleotidyltransferase, terminal, 12 4.66E-01 interacting protein 2 FAM83G family with sequence similarity 83, member G 6.5 4.58E-01 KPNA2 karyopherin alpha 2 (RAG cohort 1, importin 8.9 4.37E-01 alpha 1) KIAA0284 KIAA0284 13 4.33E-01 ODZ4 odz, odd Oz/ten-m homolog 4 (Drosophila) 1.7 4.20E-01 LRRC59 leucine rich repeat containing 59 24 4.20E-01 TRIB1 tribbles homolog 1 (Drosophila) 14 4.00E-01 TESC tescalcin 25 3.91E-01 KIAA0754 KIAA0754 8.2 3.91E-01 STBD1 starch binding domain 1 16 3.85E-01 FRY furry homolog (Drosophila) 3.7 3.79E-01 UAP1 UDP-N-acteylglucosamine pyrophosphorylase 10 3.78E-01 1 CTAGE4 CTAGE family, member 4 13 3.78E-01 LOXL2 lysyl oxidase-like 2 4.9 3.71E-01 SIPA1L2 signal-induced proliferation-associated 1 like 2 7.2 3.71E-01 SYNJ2 synaptojanin 2 8.0 3.65E-01 HTATIP2 HIV-1 Tat interactive protein 2, 30kDa 24 3.65E-01 MYBBP1A MYB binding protein (P160) 1a 6.2 3.55E-01 LOC1002718 SMG1 homolog, phosphatidylinositol 3-kinase- 5.1 3.54E-01 36 related kinase pseudogene WIPI1 WD repeat domain, phosphoinositide 8.1 3.51E-01 interacting 1 CLPB ClpB caseinolytic peptidase B homolog (E. 3.5 3.49E-01 coli) PFDN2 prefoldin subunit 2 19 3.47E-01 SPRY2 sprouty homolog 2 (Drosophila) 5.8 3.41E-01 TBC1D30 TBC1 domain family, member 30 1.9 3.36E-01 FLII flightless I homolog (Drosophila) 23 3.28E-01 ZFYVE28 zinc finger, FYVE domain containing 28 4.1 3.11E-01 ZNF275 zinc finger protein 275 4.2 3.05E-01 MYO1C myosin IC 26 2.84E-01 TANC2 tetratricopeptide repeat, ankyrin repeat and 5.9 2.68E-01 coiled-coil containing 2 TMEM163 transmembrane protein 163 17 2.43E-01 CDCP1 CUB domain containing protein 1 23 2.40E-01 ASCC2 activating signal cointegrator 1 complex 10 2.24E-01 subunit 2 BRD4 bromodomain containing 4 17 1.92E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

C15orf24 chromosome 15 open reading frame 24 35 1.89E-01 NOMO2 NODAL modulator 2 81 1.79E-01 NOMO3 NODAL modulator 3 79 1.78E-01 NOMO1 NODAL modulator 1 80 1.73E-01 PRNP protein 32 1.64E-01 NAV1 neuron navigator 1 7.5 1.62E-01 ANKRD11 ankyrin repeat domain 11 14 1.54E-01 TYMP thymidine phosphorylase 11 1.49E-01 ALKBH5 alkB, alkylation repair homolog 5 (E. coli) 28 1.43E-01 COMMD3- COMMD3-BMI1 readthrough 5.2 -Inf BMI1 AGR3 anterior gradient 3 homolog (Xenopus laevis) 2.1 - 2.44E+00 CHI3L2 chitinase 3-like 2 2.7 - 1.72E+00 SULT1C2 sulfotransferase family, cytosolic, 1C, member 2.9 - 2 1.66E+00 FAM159B family with sequence similarity 159, member B 15 - 1.42E+00 ZCCHC4 zinc finger, CCHC domain containing 4 1.8 - 1.20E+00 CCDC121 coiled-coil domain containing 121 1.9 - 1.17E+00 LOC1001282 uncharacterized LOC100128288 2.6 - 88 1.17E+00 C1orf127 open reading frame 127 7.8 - 1.16E+00 PYROXD2 pyridine nucleotide-disulphide oxidoreductase 3.5 - domain 2 1.14E+00 TSPAN8 tetraspanin 8 58 - 1.10E+00 PRR15L rich 15-like 24 - 1.09E+00 KIAA1614 KIAA1614 1.1 - 1.09E+00 WSCD2 WSC domain containing 2 8.3 - 1.08E+00 STAC3 SH3 and cysteine rich domain 3 1.2 - 1.07E+00 C11orf92 chromosome 11 open reading frame 92 2.4 - 1.04E+00 CD248 CD248 molecule, endosialin 2.5 - 1.04E+00 CCDC152 coiled-coil domain containing 152 9.4 - 1.04E+00 NWD1 NACHT and WD repeat domain containing 1 1.2 - 1.00E+00

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

C6orf222 chromosome 6 open reading frame 222 11 -9.81E-01 UPK1B uroplakin 1B 3.9 -9.81E-01 HYPK Huntingtin interacting protein K (C15orf63) 5.3 -9.56E-01 MSS51 mitochondrial translational activator homolog 1.1 -9.54E-01 (S. cerevisiae) HHLA2 HERV-H LTR-associating 2 6.4 -9.16E-01 LOC1001310 uncharacterized LOC100131089 3.6 -8.91E-01 89 TMEM145 transmembrane protein 145 4.6 -8.75E-01 PROM2 prominin 2 6.2 -8.63E-01 FAM108B1 family with sequence similarity 108, member 3.8 -8.36E-01 B1 AMPH amphiphysin 5.1 -8.12E-01 GPNMB glycoprotein (transmembrane) nmb 9.1 -8.06E-01 TMEM196 transmembrane protein 196 1.9 -7.91E-01 NBPF14 neuroblastoma breakpoint family, member 14 18 -7.87E-01 LUM lumican 16 -7.83E-01 C2orf72 chromosome 2 open reading frame 72 4.5 -7.82E-01 C15orf40 chromosome 15 open reading frame 40 8.3 -7.81E-01 NDRG2 NDRG family member 2 20 -7.64E-01 IQCC IQ motif containing C 1.9 -7.64E-01 C20orf196 chromosome 20 open reading frame 196 1.4 -7.61E-01 C1orf63 chromosome 1 open reading frame 63 24 -7.59E-01 IGSF1 immunoglobulin superfamily, member 1 5.4 -7.48E-01 EPDR1 ependymin related protein 1 (zebrafish) 7.0 -7.47E-01 FAM155B family with sequence similarity 155, member B 2.6 -7.35E-01 FLJ35390 uncharacterized LOC255031 14 -7.33E-01 LOC1002165 uncharacterized LOC100216545 1.4 -7.30E-01 45 MTRNR2L6 MT-RNR2-like 6 34 -7.23E-01 MGC27345 uncharacterized protein MGC27345 1.1 -7.07E-01 GATS GATS, stromal antigen 3 opposite strand 12 -7.02E-01 TTC18 tetratricopeptide repeat domain 18 1.6 -7.00E-01 LOC1001315 uncharacterized LOC100131564 5.8 -6.97E-01 64 COQ10A coenzyme Q10 homolog A (S. cerevisiae) 5.4 -6.95E-01 NOTCH2NL notch 2 N-terminal like 6.0 -6.82E-01 GRAMD1C GRAM domain containing 1C 2.2 -6.56E-01 PCMTD2 protein-L-isoaspartate (D-aspartate) O- 9.7 -6.52E-01 methyltransferase domain containing 2 LOC646471 uncharacterized LOC646471 4.0 -6.46E-01 HSBP1L1 heat shock factor binding protein 1-like 1 11 -6.44E-01 LOC1004227 uncharacterized LOC100422737 8.2 -6.39E-01 37 PVRIG poliovirus receptor related immunoglobulin 5.5 -6.26E-01 domain containing

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

LRRC4 leucine rich repeat containing 4 3.2 -6.25E-01 N4BP2L1 NEDD4 binding protein 2-like 1 3.7 -6.25E-01 SFTPA2 surfactant protein A2 2.3 -6.23E-01 C1orf226 chromosome 1 open reading frame 226 2.0 -6.21E-01 C7orf43 chromosome 7 open reading frame 43 8.4 -5.88E-01 PRUNE prune homolog (Drosophila) 8.1 -5.85E-01 TMEM229B transmembrane protein 229B 5.2 -5.84E-01 TSPAN6 tetraspanin 6 8.0 -5.71E-01 CA11 carbonic anhydrase XI 9.7 -5.70E-01 ABHD14B abhydrolase domain containing 14B 17 -5.61E-01 NBPF16 neuroblastoma breakpoint family, member 16 56 -5.54E-01 HRSP12 heat-responsive protein 12 5.2 -5.49E-01 IQCB1 IQ motif containing B1 4.7 -5.42E-01 C15orf48 chromosome 15 open reading frame 48 63 -5.32E-01 TMCC3 transmembrane and coiled-coil domain family 4.6 -5.31E-01 3 THNSL1 threonine synthase-like 1 (S. cerevisiae) 4.2 -5.23E-01 EPN3 epsin 3 4.9 -5.18E-01 CCDC61 coiled-coil domain containing 61 2.6 -5.16E-01 OLFM2 olfactomedin 2 33 -5.14E-01 IGIP IgA-inducing protein homolog (Bos taurus) 11 -5.10E-01 NBPF24 neuroblastoma breakpoint family, member 24 40 -5.06E-01 FAM117A family with sequence similarity 117, member A 7.8 -4.97E-01 TRIQK triple QxxK/R motif containing (C8orf83) 9.4 -4.96E-01 SLAIN1 SLAIN motif family, member 1 3.7 -4.96E-01 GTF2IRD1P GTF2I repeat domain containing 1 pseusogene 4.6 -4.18E-01 1 1 (loc729156) C16orf88 chromosome 16 open reading frame 88 7.7 -4.82E-01 PXMP4 peroxisomal membrane protein 4, 24kDa 2.0 -4.80E-01 IQCK IQ motif containing K 3.3 -4.62E-01 ANXA9 annexin A9 7.7 -4.51E-01 C14orf93 chromosome 14 open reading frame 93 4.8 -4.51E-01 FAM53B family with sequence similarity 53, member B 5.1 -4.49E-01 GPR98 G protein-coupled receptor 98 1.2 -4.48E-01 C15orf38 chromosome 15 open reading frame 38 7.1 -4.36E-01 C11orf93 chromosome 11 open reading frame 93 8.2 -4.27E-01 ZDHHC4 zinc finger, DHHC-type containing 4 19 -4.25E-01 SELENBP1 selenium binding protein 1 12 -4.19E-01 TMEM206 transmembrane protein 206 2.1 -4.19E-01 MYL9 myosin, light chain 9, regulatory 16 -4.17E-01 RPUSD4 RNA pseudouridylate synthase domain 5.5 -4.15E-01 containing 4 SPG11 spastic paraplegia 11 (autosomal recessive) 9.1 -4.14E-01 NCMAP noncompact myelin associated protein 8.3 -4.07E-01 (c1orf130) CAB39L calcium binding protein 39-like 3.9 -4.05E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

C1orf126 chromosome 1 open reading frame 126 3.4 -4.02E-01 CYS1 cystin 1 6.7 -4.00E-01 FAM219B family with sequence similarity 219, member B 17 -3.98E-01 (C15orf17) PQLC3 PQ loop repeat containing 3 7.1 -3.97E-01 TMEM63A transmembrane protein 63A 15 -3.96E-01 GRAMD3 GRAM domain containing 3 5.5 -3.92E-01 SPATA20 spermatogenesis associated 20 20 -3.82E-01 C8orf42 chromosome 8 open reading frame 42 7.6 -3.82E-01 C8orf33 chromosome 8 open reading frame 33 9.8 -3.78E-01 FAM115A family with sequence similarity 115, member A 24 -3.76E-01 CCDC28A coiled-coil domain containing 28A 9.5 -3.76E-01 PLEKHA6 pleckstrin homology domain containing, family 15 -3.72E-01 A member 6 TMEM19 transmembrane protein 19 4.2 -3.68E-01 TMEM50B transmembrane protein 50B 8.9 -3.66E-01 C22orf39 open reading frame 39 7.8 -3.64E-01 C1orf115 chromosome 1 open reading frame 115 8.1 -3.63E-01 LMBR1L limb region 1 homolog (mouse)-like 7.3 -3.60E-01 FAM46C family with sequence similarity 46, member C 14 -3.56E-01 TMEM126B transmembrane protein 126B 11 -3.56E-01 MTRNR2L1 MT-RNR2-like 1 96 -3.52E-01 TMEM87A transmembrane protein 87A 13 -3.49E-01 NBPF3 neuroblastoma breakpoint family, member 3 3.1 -3.44E-01 SAMD12 sterile alpha motif domain containing 12 4.5 -3.44E-01 CASC4 cancer susceptibility candidate 4 21 -3.43E-01 TMTC2 transmembrane and tetratricopeptide repeat 4.4 -3.36E-01 containing 2 NARG2 NMDA receptor regulated 2 3.4 -3.36E-01 TRANK1 tetratricopeptide repeat and ankyrin repeat 2.1 -3.33E-01 containing 1 C1orf198 chromosome 1 open reading frame 198 14 -3.30E-01 ATXN7L3B ataxin 7-like 3B 27 -3.19E-01 TMCC1 family 1 transmembrane and coiled-coil 5.4 -3.15E-01 domain ITM2B integral membrane protein 2B 161 -3.09E-01 SH3BP5L SH3-binding domain protein 5-like 8.4 -3.09E-01 PCED1A PC-esterase domain containing 1A (FAM113A) 15 -2.99E-01 C5orf22 chromosome 5 open reading frame 22 3.5 -2.98E-01 NIPAL3 NIPA-like domain containing 3 9.0 -2.97E-01 FAM213B family with sequence similarity 213, member B 8.7 -2.96E-01 (C1orf93) R3HDM2 R3H domain containing 2 16 -2.95E-01 TSPYL4 TSPY-like 4 13 -2.92E-01 PLCL2 phospholipase C-like 2 7.9 -2.90E-01 SUN2 Sad1 and UNC84 domain containing 2 25 -2.87E-01

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

HEBP1 heme binding protein 1 24 -2.87E-01 C10orf76 chromosome 10 open reading frame 76 5.3 -2.82E-01 C6orf120 chromosome 6 open reading frame 120 6.6 -2.79E-01 VWA5A von Willebrand factor A domain containing 5A 23 -2.75E-01 IFT172 intraflagellar transport 172 homolog 4.2 -2.71E-01 (Chlamydomonas) C6orf89 chromosome 6 open reading frame 89 19 -2.64E-01 C22orf13 chromosome 22 open reading frame 13 21 -2.60E-01 C22orf29 chromosome 22 open reading frame 29 3.6 -2.54E-01 ZC3H7A zinc finger CCCH-type containing 7A 12 -2.49E-01 CHURC1 churchill domain containing 1 8.8 -2.45E-01 DENND2D DENN/MADD domain containing 2D 12 -2.44E-01 SUN1 Sad1 and UNC84 domain containing 1 30 -2.34E-01 RPRD2 regulation of nuclear pre-mRNA domain 8.5 -2.30E-01 containing 2 CYB561D1 cytochrome b-561 domain containing 1 7.2 -2.28E-01 TBC1D2B TBC1 domain family, member 2B 8.2 -2.27E-01 KIAA1191 KIAA1191 27 -2.11E-01 GPR107 G protein-coupled receptor 107 15 -2.09E-01 TSPYL1 TSPY-like 1 24 -2.08E-01 RNF145 ring finger protein 145 19 -1.98E-01 TACSTD2 tumor-associated calcium signal transducer 2 238 -1.95E-01 FAM134C family with sequence similarity 134, member C 16 -1.84E-01 RBM6 RNA binding motif protein 6 27 -1.81E-01 PHLDB1 pleckstrin homology-like domain, family B, 7.4 -1.75E-01 member 1 NUCKS1 nuclear casein kinase and cyclin-dependent 33 -1.70E-01 kinase substrate 1 METTL9 methyltransferase like 9 14 -1.62E-01 TOR1AIP2 torsin A interacting protein 2 10 -1.54E-01 CCDC93 coiled-coil domain containing 93 12 -1.39E-01 KIAA1522 KIAA1522 65 -1.13E-01 RNA-seq gene expression in 5 human islet preparations. The sum of the RPKM for all the transcripts of the same gene under control condition is taken as measure of gene expression and the median of the 5 values is provided. Genes that were not detected or had an RPKM <1 in the control and in the palmitate condition are not mentioned. The log2 of the proportion between the sum of the RPKM for all the transcripts from the same gene under palmitate treatment and the same sum obtained under control conditions was taken as measure of change in gene expression. A difference in gene expression was considered significant if the corrected p value was <0.05. Genes were only taken up in the list when they were significantly changed in expression in one direction for at least 4 islet samples and changed in the other direction for none. The table contains the median log2 fold change of the samples with significantly modified gene expression.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 5. Expression of genes involved in mitochondrial movement, biogenesis, fusion-fission and mitophagy.

Media n Median Gene Gene description RPKM RPKM name contro palmitate l FIS1 fission 1 (mitochondrial outer membrane) 55 58 homolog (S. cerevisiae) PINK1 PTEN induced putative kinase 1 23 22 ESRRA estrogen-related receptor alpha 19 23 MFN2 mitofusin 2 19 17 BNIP3L BCL2/adenovirus E1B 19kDa interacting protein 18 22 3-like RHOT2 ras homolog family member T2 17 16 TRAK1 trafficking protein, kinesin binding 1 15 18 DNM1L dynamin 1-like 13 9.5 MFF mitochondrial fission factor 11 9.0 TRAK2 trafficking protein, kinesin binding 2 8.1 9.0 FUNDC1 FUN14 domain containing 1 6.5 5.5 OPA1 optic atrophy 1 (autosomal dominant) 6.4 6.0 RHOT1 ras homolog family member T1 6.0 6.2 NRF1 nuclear respiratory factor 1 5.1 5.0 MFN1 mitofusin 1 4.8 5.6 SMCR7 Smith-Magenis syndrome chromosome region, 3.7 3.2 candidate 7 PPARGC1 peroxisome proliferator-activated receptor 3.7 3.7 A gamma, coactivator 1 alpha GABPA GA binding protein transcription factor, alpha 3.6 3.9 subunit 60kDa PARK2 parkinson protein 2, E3 ubiquitin protein ligase 0.9 1.3 (parkin) PPARGC1 peroxisome proliferator-activated receptor 0.3 0.3 B gamma, coactivator 1 beta RNA-seq gene expression in 5 human islet preparations. The sum of the RPKM for all the transcripts of the same gene is taken as measure of gene expression and the median of the 5 values is provided.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 6. Expression of genes involved in regulation of insulin secretion in human islets.

Median Median Gene Gene description RPKM RPKM name control palmitate Glucose transporters SLC2A1 solute carrier family 2 (facilitated glucose 12 18 transporter), member 1 SLC2A3 solute carrier family 2 (facilitated glucose 2.9 3.6 transporter), member 3 SLC2A2 solute carrier family 2 (facilitated glucose 1.5 1.0 transporter), member 2 Voltage gated Ca2+ channels CACNA1H calcium channel, voltage-dependent, T type, alpha 9.4 14 1H subunit CACNA1A calcium channel, voltage-dependent, P/Q type, alpha 8.1 13 1A subunit CACNA1D calcium channel, voltage-dependent, L type, alpha 6.3 6.3 1D subunit CACNA1C calcium channel, voltage-dependent, L type, alpha 3.0 3.4 1C subunit CACNA1E calcium channel, voltage-dependent, R type, alpha 0.04 0.03 1E subunit Voltage gated Na+ channels SCN1B sodium channel, voltage-gated, type I, beta subunit 4.2 6.1 SCN3B sodium channel, voltage-gated, type III, beta subunit 2.7 3.4 SCN9A sodium channel, voltage-gated, type IX, alpha 2.2 2.0 subunit SCN8A sodium channel, voltage gated, type VIII, alpha 1.4 1.2 subunit Large conductance Ca2+-activated and voltage-gated delayed rectifying K+ channels KCNH2 potassium voltage-gated channel, subfamily H (eag- 34 20 related), member 2 KCNQ1 potassium voltage-gated channel, KQT-like 21 22 subfamily, member 1 KCNMA1 potassium large conductance calcium-activated 10 9.0 channel, subfamily M, alpha member 1 KCNB2 potassium voltage-gated channel, Shab-related 2.6 2.7 subfamily, member 2 Inwardly rectifying K+ channels KCNJ11 potassium inwardly-rectifying channel, subfamily J, 13 18 member 11 KCNJ15 potassium inwardly-rectifying channel, subfamily J, 5.4 3.0 member 15 KCNJ4 potassium inwardly-rectifying channel, subfamily J, 0.06 0.14 member 4 KCNJ12 potassium inwardly-rectifying channel, subfamily J, 0.01 0 member 12 Small conductance Ca2+-activated channels KCNN4 potassium intermediate/small conductance calcium- 0.3 0.3 activated channel, subfamily N, member 4

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

KCNN3 potassium intermediate/small conductance calcium- 0.3 0.3 activated channel, subfamily N, member 3 KCNN2 potassium intermediate/small conductance calcium- 0.2 0.1 activated channel, subfamily N, member 2 KCNN1 potassium intermediate/small conductance calcium- 0.04 0.02 activated channel, subfamily N, member 1 Background conductance CASR calcium-sensing receptor 24 22 NALCN sodium leak channel, non-selective 1.4 1.3 Intracellular ion channels ITPR3 inositol 1,4,5-trisphosphate receptor, type 3 31 32 CLCN3 chloride channel, voltage-sensitive 3 22 23 ITPR2 inositol 1,4,5-trisphosphate receptor, type 2 2.4 3.2 ITPR1 inositol 1,4,5-trisphosphate receptor, type 1 1.8 2.4 RYR2 ryanodine receptor 2 (cardiac) 0.3 0.4 RYR3 ryanodine receptor 3 0.02 0.007 G-protein coupled inwardly rectifying K+ channels KCNJ6 potassium inwardly-rectifying channel, subfamily J, 2.4 3.2 member 6 KCNJ3 potassium inwardly-rectifying channel, subfamily J, 1.4 1.3 member 3 Receptors GABBR1 gamma-aminobutyric acid (GABA) B receptor, 1 9.5 7.5 SSTR2 somatostatin receptor 2 8.1 10 SSTR3 somatostatin receptor 3 5.5 6.8 SSTR1 somatostatin receptor 1 5.1 6.2 SSTR5 somatostatin receptor 5 1.2 0.7 SSTR4 somatostatin receptor 4 0.06 0.2 Proteins involved in exocytosis STX1A syntaxin 1A (brain) 50 47 VAMP2 vesicle-associated membrane protein 2 44 50 (synaptobrevin 2) SYT7 synaptotagmin VII 37 47 SNAP25 synaptosomal-associated protein, 25kDa 28 36 GCK glucokinase (hexokinase 4) 22 17 SYT5 synaptotagmin V 13 21 GLP1R glucagon-like peptide 1 receptor 7.9 9.9 SENP1 SUMO1/sentrin specific peptidase 1 1.8 2.2 RNA-seq gene expression in 5 human islet preparations. The sum of the RPKM for all the transcripts of the same gene is taken as measure of gene expression and the median of the 5 values is provided. A difference in gene expression was considered significant if the corrected p value was <0.05 in at least 4 islet samples and changed in the other direction for none. Significantly induced genes are shown in red, significantly downregulated genes in green, and genes with significantly modified spice variants in blue.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 1. Comparison of human islet transcript expression changes induced by palmitate versus the cytokines IL-1 + IFN- and versus genes differentially expressed in T2D human islets. Venn diagram of (A) up- or (B) downregulated genes in islets exposed for 48 h to palmitate (present findings) or the cytokines IL-1 + IFN- (1). (C) Venn diagram of transcripts with modified splicing following palmitate or cytokine exposure. (D) Of the genes that are differentially expressed in human islets from T2D donors compared to non-diabetic donors and from donors with HbA1c ≥6% compared to those with HbA1c <6% (2), 28 genes were up- or downregulated by palmitate. The fold change is plotted of gene expression changes in T2D or hyperglycemic islets, assessed by microarrays, and in palmitate-treated islets, assessed by RNA-seq. The changes were largely in the same direction, with a correlation coefficient of 0.59 (p=0.001).

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 2. Validation of RNA-seq gene expression data by qRT-PCR in palmitate-treated human islets. Human islets from 5 organ donors were cultured for 48 h in the presence or absence (CTL) of palmitate (PAL). RNA-seq gene expression results (black bars) were compared to gene expression assessed by qRT-PCR (gray bars) in the same human islet preparations used for RNA-seq. Data were normalized to the geometric mean of -actin and GAPDH expression and expressed as fold induction of control. *p<0.05, **p<0.01 for CYT vs PAL by ratio t test. Bottom right panel: Palmitate-induced transcript expression changes of 30 genes assessed by RNA-seq were compared to expression changes measured by qRT- PCR in independent islet preparations. There was good agreement between the two datasets, with a correlation coefficient of 0.63 (p<0.001). Plotted RNA-seq data are the median of the fold change of RPKM in palmitate vs control condition; qRT-PCR data are expressed as the average fold change in gene expression corrected for the reference gene -actin.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 3. DNA methylation profiling-derived T2D candidate genes are well expressed in human islets. Transcript expression levels are shown of genes that showed differential DNA methylation in a previous DNA methylation profiling study (3). Panels A and B show hypomethylated genes with RPKM >1, ranked by expression level, and panel C shows RNA-seq expression levels of the hypermethylated genes. Red bars indicate palmitate- upregulated and green bars downregulated transcripts.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 4. IPA of palmitate-modified genes. (A) 428 genes were significantly upregulated by palmitate in at least 4 out of 5 islet samples, and significantly downregulated in none. These genes were mapped to 417 unique entries in the IPA database and submitted to gene set enrichment analysis based on Benjamini-Hochberg corrected Fisher tests. IPA of these upregulated genes is shown for “Molecular and Cellular Function”. (B) 897 genes were significantly downregulated by palmitate in at least 4 out of 5 islet samples, and significantly upregulated in none. They were mapped to 885 unique entries in the IPA database. IPA of these cytokine-downregulated genes is shown for “Molecular and Cellular Function”. The length of the blue bars indicates the significance of the association between the set of genes and the keyword, and is expressed as minus the logarithm of the probability that a random set of genes from the human genome would be associated with the same keyword. The straight orange line indicates a threshold of 0.05 (corresponding to a -log(BH p-value) of 1.3).

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 5. Palmitate alters mitochondrial morphology in human -cells and induces LonP1 expression. (A) Human islets exposed to palmitate for 48 h were examined by electron microscopy. Mitochondria (M) in -cells were often elongated with signs of fragmentation (arrows) or they appeared swollen (Ms). Magnification: top panel x 21000, other panels x 64000. (B) LonP1 mRNA expression by RNA-seq in palmitate (PAL)-treated human islets. (C) LonP1 protein expression in INS-1E cells treated or not for 6, 24 and 48 h with 0.5 mM palmitate (P or PAL). (D) Densitometric quantification of Western blots as in panel C, n=2- 7 independent experiments, *p<0.05 vs control (C or CT).

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 6. Palmitate impairs human islet glucose-stimulated insulin secretion. Human islets were exposed or not (CTL) to palmitate (PAL) for 48 h. Insulin secretion was measured at 3.3 and 16.7 mM glucose (n=4).

Supplementary Figure 7. Role of TXNIP and IL-1 in lipotoxic and chemical ER stress- induced apoptosis. (A) Palmitate inhibited TXNIP mRNA expression in human islets. Human islets were cultured in the presence of palmitate (PAL, 0.5 mM), 28 mM glucose (G28) or the combination of both (G28 PAL) for 48 h. TXNIP mRNA expression was measured by qRT-PCR and normalized to -actin expression levels (n=5-7). *p<0.05 vs control (CTL) by ratio t test. (B) IL-1 receptor antagonist does not protect human islets from chemical ER stress. Human islets were cultured for 24, 48 and 72 h in the presence of thapsigargin (Thap, 1 M) or brefeldin A (Bref, 0.1 g/ml) alone or in combination with the IL-1 receptor antagonist (300 ng/ml). The cytokines (Cyt) IL-1 (50 U/ml) plus IFN- (1,000 U/ml) were used as a positive control. Cell death was assessed in 3-5 independent human islet preparations. *p<0.05, **p<0.01 vs control (CT), ##p<0.01 as indicated.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 8. Inhibition of the type 2 diabetes candidate gene ADCY5 triggers - cell apoptosis. (A) ADCY5 mRNA expression by RNA-seq in palmitate (PAL)-treated human islets. (B) Apoptosis in INS-1E cells transfected with control siRNA (white bars) or ADCY5 siRNA (black bars) and then treated with palmitate for 16 h (n=4). *p<0.05 vs control (CTL), #p<0.05, ##p<0.01 as indicated.

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 9. Palmitate-induced changes in alternative splicing according to RefSeq annotation. (A) Palmitate exposure led to changes in alternative splicing. Using RefSeq annotation, 363 transcripts were significantly upregulated in at least 4 out of 5 islet samples and significantly downregulated in none, and 462 transcripts were significantly downregulated using similar criteria. The Venn diagram illustrates the number of genes which have transcripts modified in both directions (intersection) and in only one direction. (B) IPA of the 574 genes with modified splicing. The length of the blue bars indicates the significance of the association between the set of transcripts and the keyword, and is expressed as minus the logarithm of the probability that a random set of transcripts from the human genome would be associated with the same keyword. The straight orange line indicates a threshold of 0.05 (corresponding to a -log(BH p-value) of 1.3).

©2013 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1383/-/DC1

PAPER II

Abdulkarim B, Nicolino M, Igoillo-Esteve M, Daures M, Romero S, Philippi A, Senée V, Lopes M, Cunha DA, Harding HP, Derbois C, Bendelac N, Hattersley AT, Eizirik DL, Ron D, Cnop M, Julier C; A Missense Mutation in PPP1R15B Causes a Syndrome Including Diabetes, Short Stature, and Microcephaly.

176

SUPPLEMENTARY DATA

Supplementary Table 1. PCR amplification and sequencing primers used for Sanger sequencing of human PPP1R15B gene

©2015 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 2. siRNA sequences

Supplementary Table 3. Primer sequences used for mRNA expression studies

Supplementary Table 4. Filtering of variants identified by whole exome sequencing of patient 1 (index patient)

Counts are the number of autosomal variants (SNVs and insertion/deletion variants (indels)) identified by Whole Exome Sequencing of the patient compared to the Human Reference Genome on UCSC build hg19. The successive filters applied after quality filtering are shown: ahomozygous variants, bnonsynonymous variants including missense and nonsense, splice-site variants and exonic indels (frameshift and non-shifted), cvariants that were absent in the homozygous status in an in-house database, in Exome Variant Server (EVS) and in Exome Aggregation Consortium (ExAC) and with a MAF < 0.005 in these databases and in dbSNP.

©2015 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 5. Description of the 18 rare variants identified by whole exome sequencing in patient 1, and complementary genotyping of these variants in patient 2

Genomic map position is on UCSC hg19. Description of the consequences on cDNA and protein follows the Human Genome Variation Society (HGVS) recommendations (4). Heterozygotes and total genotype counts, heterozygotes frequencies, and minor allele frequencies (MAF) are given for EVS and ExAC. All the variants were absent in the homozygous status in these databases. The genotype of the two affected siblings is shown as 2/2 (homozygous for the rare allele), 1/2 (heterozygous) and 1/1 (homozygous for the frequent allele, none found). Variants homozygous in both patients are shown in bold. NA: not available.

©2015 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1 SUPPLEMENTARY DATA

Supplementary Table 6. Functional predictions of the variants compatible with mutation status and human mutations and knockout mouse phenotypes of the corresponding genes

In silico prediction of the impact and severity of mutation on protein function was performed using Polyphen-2 (8), SIFT and Provean (9), using recommended parameters. Polyphen-2 predictions were made based on the HumDiv model. aHuman islet expression is based on (10) and our unpublished data (M.C.) and is given in RPKM (reads per kilobase of exon model per million mapped reads) units. bKnockout mouse model phenotype information is according to the Mouse Genome informatics (MGI) and the Wellcome Trust Sanger Institute (WTSI) databases. NA: not available.

©2015 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 1. Heatmap of gene expression in human tissues. RNAseq values (in RPKM) from the indicated human tissues were obtained from GTEx (v4.pl). RNA-seq data of FACS-purified human islet -cells were from Nica et al (11). Human islet RNA-seq data (24 in total) were from Eizirik et al and Cnop et al (10; 12) (and unpublished data). Bone (osteoblast) gene expression was obtained from GEO dataset accession number GSE57925 (unpublished data). The median RPKM value of the samples is represented, with the maximum set at 30. The heatmap was made in R (function heatmap.2).

©2015 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1 SUPPLEMENTARY DATA

Supplementary Figure 2. PPP1R15B silencing does not modify expression of the pro-apoptotic Bim splice variants Bim-EL and Bim-L or the anti-apoptotic proteins BCL2 and BCL-XL. INS-1E cells were transfected with control siRNA (siCT) or two different siRNAs targeting PPP1R15B (siP1R15B1 and siP1R15B2). 48h after transfection the Bim-EL and Bim-L (A, B, C) and BCL2 and BCL-XL (D, E, F) expression was examined by Western blot. A and D are representative blots of 5 experiments. B, C, E and F are densitometric quantifications of protein expression corrected for α- tubulin, and expressed as fold of siCT.

©2015 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1 SUPPLEMENTARY DATA

Supplementary References 1. Jousse C, Oyadomari S, Novoa I, Lu P, Zhang Y, Harding HP, Ron D. Inhibition of a constitutive translation initiation factor 2α phosphatase, CReP, promotes survival of stressed cells. J Cell Biol 2003;163:767-775 2. Cunha DA, Igoillo-Esteve M, Gurzov EN, Germano CM, Naamane N, Marhfour I, Fukaya M, Vanderwinden JM, Gysemans C, Mathieu C, Marselli L, Marchetti P, Harding HP, Ron D, Eizirik DL, Cnop M. Death protein 5 and p53-upregulated modulator of apoptosis mediate the endoplasmic reticulum stress-mitochondrial dialog triggering lipotoxic rodent and human β-cell apoptosis. Diabetes 2012;61:2763-2775 3. Gurzov EN, Germano CM, Cunha DA, Ortis F, Vanderwinden JM, Marchetti P, Zhang L, Eizirik DL. p53 up-regulated modulator of apoptosis (PUMA) activation contributes to pancreatic beta-cell apoptosis induced by proinflammatory cytokines and endoplasmic reticulum stress. J Biol Chem 2010;285:19910-19920 4. Harding HP, Zhang Y, Scheuner D, Chen JJ, Kaufman RJ, Ron D. Ppp1r15 gene knockout reveals an essential role for translation initiation factor 2 alpha (eIF2alpha) dephosphorylation in mammalian development. Proc Natl Acad Sci U S A 2009;106:1832-1837 5. Ahram D, Sato TS, Kohilan A, Tayeh M, Chen S, Leal S, Al-Salem M, El-Shanti H. A homozygous mutation in ADAMTSL4 causes autosomal-recessive isolated ectopia lentis. Am J Hum Genet 2009;84:274-278 6. Palmer CN, Irvine AD, Terron-Kwiatkowski A, Zhao Y, Liao H, Lee SP, Goudie DR, Sandilands A, Campbell LE, Smith FJ, O'Regan GM, Watson RM, Cecil JE, Bale SJ, Compton JG, DiGiovanna JJ, Fleckman P, Lewis-Jones S, Arseculeratne G, Sergeant A, Munro CS, El HB, McElreavey K, Halkjaer LB, Bisgaard H, Mukhopadhyay S, McLean WH. Common loss-of-function variants of the epidermal barrier protein filaggrin are a major predisposing factor for atopic dermatitis. Nat Genet 2006;38:441-446 7. Kypriotou M, Boechat C, Huber M, Hohl D. Spontaneous atopic dermatitis-like symptoms in a/a ma ft/ma ft/J flaky tail mice appear early after birth. PLoS One 2013;8:e67869 8. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P, Kondrashov AS, Sunyaev SR. A method and server for predicting damaging missense mutations. Nat Methods 2010;7:248-249 9. Choi Y, Sims GE, Murphy S, Miller JR, Chan AP. Predicting the functional effect of amino acid substitutions and indels. PLoS One 2012;7:e46688 10. Cnop M, Abdulkarim B, Bottu G, Cunha DA, Igoillo-Esteve M, Masini M, Turatsinze JV, Griebel T, Villate O, Santin I, Bugliani M, Ladriere L, Marselli L, McCarthy MI, Marchetti P, Sammeth M, Eizirik DL. RNA sequencing identifies dysregulation of the human pancreatic islet transcriptome by the saturated fatty acid palmitate. Diabetes 2014;63:1978-1993 11. Nica AC, Ongen H, Irminger JC, Bosco D, Berney T, Antonarakis SE, Halban PA, Dermitzakis ET. Cell-type, allelic, and genetic signatures in the human pancreatic beta cell transcriptome. Genome Res 2013;23:1554-1562 12. Eizirik DL, Sammeth M, Bouckenooghe T, Bottu G, Sisino G, Igoillo-Esteve M, Ortis F, Santin I, Colli ML, Barthson J, Bouwens L, Hughes L, Gregory L, Lunter G, Marselli L, Marchetti P, McCarthy MI, Cnop M. The human pancreatic islet transcriptome: expression of candidate genes for type 1 diabetes and the impact of pro-inflammatory cytokines. PLoS Genet 2012;8:e1002552

©2015 American Diabetes Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-0477/-/DC1

PAPER III

AbdulkarimB,Hernangomez M, Igoillo-Esteve M, Ladriere L, Cunha DA, MarselliL,Marchetti P, Eizirik DL, Cnop M; Guanabenz sensitizes β-cells to endoplasmic reticulum stress-induced apoptosis.

184

Guanabenz sensitizes pancreatic β cells to lipotoxic endoplasmic reticulum stress and apoptosis

Baroj Abdulkarim, Miriam Hernangomez, Mariana Igoillo Esteve, Daniel A Cunha, Lorella Marselli, Piero Marchetti, Laurence Ladriere, Miriam Cnop

Supplemental Figure Legends

Supplemental Figure 1: Guanabenz, but not inactive guanabenz, potentiates ER stress-induced apoptosis. INS-1E cells were treated or not (DMSO) with 50 µM guanabenz (GA), alone or in combination with oleate (OL), palmitate (PAL) or a mixture of both (O/P) for 8h. Western blots for phosphorylated eIF2α were quantified by densitometry and corrected for total eIF2α (A). Apoptosis in INS-1E cells treated or not (DMSO) with 2 µM guanabenz, alone or in combination with oleate, palmitate or a 1:1 mixture of oleate and palmitate for 24h (B). Apoptosis in INS-1E cells treated or not (DMSO) with 50 µM inactive guanabenz (GAi), alone or with FFAs for 24h (C). Apoptosis in INS-1E cells treated or not (DMSO) with 50 µM guanabenz (D) or inactive guanabenz (GAi) (E), alone or in combination with the chemical ER stressors CPA (25 µM), brefeldin A (BREF, 0.1 µg/ml) or tunicamycin (TUNI, 5 µg/ml). The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of remaining data points. n=4-14. *FFA or chemical ER stressor vs control, #guanabenz vs DMSO, */#p<0.05, **/##p<0.01.

Supplemental Figure 2: Guanabenz induces apoptosis in primary rat β cells. Dispersed rat islets were treated or not (CTL) with 50 µM guanabenz (GA) alone or in combination with oleate (OL) for 24h. The cells were fixed and co-immunostained for insulin (green, Alexa488) and cleaved caspase 3 (red, Alexa 568). Hoechst 33342 was used to stain the nucleus.

Supplemental Figure 3: Guanabenz induces insulin resistance and glucose intolerance in mice. Fasting glycemia (A) and plasma insulin levels (B) of mice treated for 1 week with guanabenz (GA) or vehicle (Veh) while on regular (RD) or high fat diet (HFD). HOMA-IR (C) was calculated using fasting glycemia and insulin levels. The incremental area under the curve between 0 and 120 minutes of glycemia (D) and plasma insulin (E) in the IPGTT. Food intake of the mice in kcal/mouse/week (F). Insulin content of the isolated mouse islets, corrected for total protein content (G). Rat islets were exposed to guanabenz alone or in combination with oleate for 24h (n=5). Islet insulin content was corrected for total protein (H). The dots represent individual animals, and the line indicates the mean (A-E and G). Results are mean (F). The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of remaining data points (H). #Guanabenz vs vehicle/DMSO, *FFA vs control (CTL), */#p<0.05, **/##p<0.01.

Supplemental Figure 4: Guanabenz potentiates FFA-induced CHOP expression. INS-1E cells transfected with a CHOP luciferase reporter construct (A) were treated with guanabenz (GA, 50 µM) alone or in combination with palmitate (PAL) for 8h. CPA was used as a positive control. INS-1E cells were treated or not (DMSO) with 50 µM guanabenz alone or in combination with oleate (OL), palmitate (PAL) or a mixture of both (O/P) for 16h (B-E). Western blots for XBP1s (B and D) and BiP (C and E) were quantified by densitometry and corrected for α-tubulin. Blots are representative of 4 independent experiments. Human islets were treated with 50 µM guanabenz alone or in combination with oleate or palmitate for 72h. XBP1s mRNA expression was examined by real time PCR and corrected for the reference gene β-actin (F). Western blot for XBP1s (G) and p-eIF2α (H) in islets from mice treated for 1 week with guanabenz or vehicle (Veh) while on regular (RD) or high fat diet (HFD). α-tubulin and total eIF2α were used as controls for protein loading. INS-1E cells transfected with a control siRNA (siCTL) or siRNA targeting CHOP (siCHOP) were treated with guanabenz (50 µM) alone or in combination with palmitate for 16h. CHOP mRNA expression was assessed by real time PCR and corrected for the reference gene GAPDH (I). The boxes indicate lower quartile, median, and higher quartile; whiskers represent the range of remaining data points (A, D-F, I). The dots

1 represent individual animals, and the line indicates the mean (G-H). n=4-6 independent experiments. *FFA vs control (CTL), #guanabenz vs DMSO, */#p<0.05, **/##p<0.01.

2

Supplemental Table 1: Antibodies used for protein expression studies

RRID Peptide/protein Antigen sequence Species raised in; Name of Antibody Manufacturer, catalog # Dilution used (required in target (if known) monoclonal or polyclonal revised MSs) Rabbit p-eIF2α NA Phospho-eIF2α (Ser51) Cell Signaling Cat#:3597 1/1000 AB_390740 Monoclonal eIF2α (D7D3) XP® Rabbit eIF2α NA Cell Signaling Cat#:5324 1/1000 AB_10692650 Monoclonal Cleaved Caspase-3 Rabbit Cleaved Caspase 3 NA Cell Signaling Cat#:9661 1/1000 AB_2341188 (Asp175) Polyclonal BiP Antibody Rabbit BiP NA Cell Signaling Cat#:3183 1/2000 AB_10695864 Polyclonal Monoclonal Anti-α-Tubulin Sigma-Aldrich Mouse α-Tubulin NA 1/5000 AB_477593 antibody Cat#:T9026 Monoclonal GADD 153 Antibody (B-3) Mouse CHOP NA Santa Cruz Cat#:SC-7351 1/1000 AB_627411 Monoclonal XBP-1 Antibody (M-186) Rabbit XBP1s NA Santa Cruz Cat#:Sc-7160 1/1000 AB_794171 Polyclonal Rabbit β-Actin NA beta-Actin Antibody Cell Signaling Cat#:4967 1/5000 AB_330288 Polyclonal Mouse Puromycin NA Puromycin (3RH11) Kerafast Cat#:EQ0001 1/1000 AB_2620162 Monoclonal Guinea Pig Insulin NA Insulin antibody Dako Cat#: A0564 1/200 AB_10013624 Polyclonal Peroxidase AffiniPure Lucron Bioproducts Donkey Anti-rabbit IgG NA F(ab')2 Fragment Donkey 1/5000 AB_2340590 Anti-Rabbit IgG (H+L) Cat#:711-036-152 Polyclonal

Peroxidase AffiniPure Lucron Bioproducts Donkey Anti-mouse IgG NA F(ab') Fragment Donkey 1/5000 AB_2340773 2 Cat#:715-036-150 Polyclonal Anti-Mouse IgG (H+L)

7

Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed ThermoFisher Cat#:A- Goat Anti-Rabbit IgG NA 1/500 AB_2534094 Secondary Antibody, Alexa 11036 Polyclonal Fluor 568 Goat anti-Guinea Pig IgG (H+L) Highly Cross- ThermoFisher Cat#:A- Goat Anti-Guinea pig NA 1/500 AB_2534117 Adsorbed Secondary 11073 Polyclonal Antibody, Alexa Fluor 488

NA: Not availabl

8

Supplemental Table 2: Primer sequences used for mRNA expression studies

Gene Species STD/RT Forward Reverse Amplicon size (bp)

GAPDH Rat STD ATGACTCTACCCACGGCAAG TGTGAGGGAGATGCTCAGTG 930 RT AGTTCAACGGCACAGTCAAG TACTCAGCACCAGCATCACC 136

β-actin Human STD AAATCTGGCACCACACCTTC CCGATCCACACGGAGTACTT 775 RT CTGTACGCCAACACAGTGCT GCTCAGGAGGAGCAATGATC 127 XBP1s Rat STD AAACAGAGTAGCAGCGCAGACTGC GGATCTCTAAGACTAGAGGCTTGGTG 600 RT GAGTCCGCAGCAGGTG GCGTCAGAATCCATGGGA 65 Human STD CCGCAGCAGCTGCAGG GGGGCTTGGTATATATGTGG 442 RT CCGCAGCAGGTGCAGG GAGTCAATACCGCCAGAATCCA 70 CHOP Rat STD GTCTCTGCCTTTCGCCTTTG CTACCCTCAGTCCCCTCCTC 606

RT CCAGCAGAGGTCACAAGCAC CGCACTGACCACTCTGTTTC 125 Human STD AGGCACTGAGCGTATCATGTT CTGTTTCCGTTTCCTGGTTC 460 RT Hs_DDIT3_1_SG QuantiTect Primer Assay (NM_004083), Qiagen 90 GADD34 Rat STD GCCAGAGTACCCAGCATTGT GCAGTGGAAGAGACGAGGAC 449 RT AGGGATGTGGAGAAGCAGAG AAAGATCTGAGCCGCTTCTG 124

BiP Rat STD CTCAAAGAGCGCATTGACA AATGCTATAGCCCAAGTGGCT 446 RT CCACCAGGATGCAGACATTG AGGGCCTCCACTTCCATAGA 100

STD: Standard PCR, RT: Real time PCR

9