Molecular basis of insulin resistance in Bardet Biedl syndrome

Transcription

1 University of Iowa Iowa Research Online Theses and Dissertations Spring 2015 Molecular basis of insulin resistance in Bardet Biedl syndrome Rachel Diaz Starks University of Iowa Copyright 2015 Rachel Diaz Starks This dissertation is available at Iowa Research Online: Recommended Citation Starks, Rachel Diaz. "Molecular basis of insulin resistance in Bardet Biedl syndrome." PhD (Doctor of Philosophy) thesis, University of Iowa, Follow this and additional works at: Part of the Cell Biology Commons

2 MOLECULAR BASIS OF INSULIN RESISTANCE IN BARDET BIEDL SYNDROME by Rachel Diaz Starks A thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Molecular and Cellular Biology in the Graduate College of The University of Iowa May 2015 Thesis Supervisors: Associate Professor Kamal Rahmouni Professor Val C. Sheffield

3 Copyright by RACHEL DIAZ STARKS 2015 All Rights Reserved

4 Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL PH.D. THESIS This is to certify that the Ph.D. thesis of Rachel Diaz Starks has been approved by the Examining Committee for the thesis requirement for the Doctor of Philosophy degree in Molecular and Cellular Biology at the May 2015 graduation. Thesis Committee: Kamal Rahmouni, Thesis Supervisor Val C. Sheffield, Thesis Supervisor Sheila Baker Andrew Norris Peter Rubenstein Mark Stamnes

5 To my husband Charles Starks and children, Schaler and Hascrel Starks. ii

6 Ask and it will be given to you; seek and you will find; knock and the door will be open to you. For everyone that who asks receives; he who seeks finds; and to him who knocks, the door will be open. Matthew 7:7-8, NIV iii

7 ACKNOWLEDGEMENTS Many thanks go to my advisors Kamal Rahmouni and Val Sheffield. The guidance and assistance provided by both advisors has been invaluable and truly appreciated. Whether it was the chance to present research or the feedback from each laboratory, I was truly blessed to be able to work with such thoughtful mentors. When it comes to being part of a group, I am thankful to have been able to belong to the Rahmouni laboratory and the Sheffield laboratory. In particular, Deng Guo who gave step by step assistance beginning with my rotation project where I learned immunoblotting. I am also thankful to have the opportunity to have valuable discussions with fellow graduate student, Kenjiro Muta. I appreciate that Ken would ask direct questions about my experiments and provide assistance. My thanks go to Balyssa Bell because she made a comfortable environment even richer with her willingness to seek out laboratory members for discussion at any time. In the Sheffield laboratory my particular appreciation goes to Qihong Zhang. Also, Seongjin Seo was a great resource providing plasmids along with his feedback during discussions. My thanks and esteem belongs to my committee members, Sheila Baker, Andrew Norris, Peter Rubenstein, and Mark Stamnes for their thoughtful evaluations and the sacrifice of their time. Finally, I am thankful for my husband, who was the first person to encourage me to obtain an education when I was comfortable with just having a job. I thank God for my children who are my motivation. Lastly, I would like to acknowledge Erika Starks and Kino Starks who each took time from their lives to become part of my home allowing me to complete this work and enjoy the presence of my children. iv

8 ABSTRACT Bardet Biedl Syndrome (BBS) displays heterogeneity in the genes involved and clinical features. Mutations in 19 genes have been associated with BBS. Eight BBS proteins (BBS1, 2, 4, 5, 7, 8, 9 and 18) form the BBSome. Assembly of the BBSome is mediated by three BBS proteins (BBS6, 10, 12) in a complex with the CCT/Tric chaperonins. The BBSome is involved in formation and maintenance of primary cilia and vesicle trafficking. The clinical features of BBS include obesity, degenerative retinopathy, polydactyly, renal dysfunction, hypogonadism, and learning disability. Diabetes mellitus is commonly associated with BBS, but the mechanisms remain unknown. Our objective was to understand the molecular mechanism of BBS-associated diabetes. The role of BBS in insulin receptor (IR) signaling in Bbs4 -/- mice was tested by preventing obesity using calorie restriction. These studies demonstrate the genetic defect in BBS directly contributes to the diabetes phenotype independently from the obesity phenotype. Emerging evidence implicating neuronal mechanisms in various BBS phenotypes led us to test the possibility that loss of Bbs1 in the central nervous system (CNS) disrupts glucose homeostasis. We found that deletion of the Bbs1 gene throughout the CNS or in specific hypothalamic neurons leads to hyperglycemia, glucose intolerance and insulin resistance. Our data demonstrate the critical role of neuronal Bbs1 in the regulation of glucose in an insulin-independent manner. Finally, the IR was found to interact with BBS proteins. The loss of BBSome proteins leads to a specific reduction in the amount of IR at the cell surface. The results demonstrate that BBSome proteins are required to maintain adequate levels of IR at the cell surface. The role of BBS proteins in transporting IR has not been previously described. Loss of the BBSome appears to be a novel mechanism of insulin resistance. v

9 TABLE OF CONTENTS LIST OF TABLES... VIII LIST OF FIGURES... IX LIST OF ABBREVIATIONS... XIII CHAPTER I INTRODUCTION...1 Clinical Features of Bardet-Biedl Syndrome...1 Polydactyly...1 Vision Impairment...2 Obesity...2 Learning Disability...3 Urogenital-Renal Anomalies...3 Diabetes Mellitus...4 Epidemiology of Bardet-Biedl Syndrome...5 Identification of BBS Genes...6 Structure and Function of BBS Protein...6 Insights from BBS Mouse Models...11 Hypothesis and Study Design...13 CHAPTER II THE GENETIC DEFECT IN BBS2, BBS4, AND BBS6 NULL MOUSE MODELS CONTRIBUTES TO THE DIABETES PHENOTYPE...18 Introduction...18 Materials and Methods...19 Animals...19 Glucose and Insulin Measurements...19 Glucose Tolerance Test and Insulin Tolerance Test...20 Insulin Receptor Signaling Assay...20 Immunoblotting...20 Statistical Analysis...21 Results...21 Glucose Homeostasis...21 Insulin Signaling Defect in Bbs4 -/- Mice...22 Fatty acid synthesis genes are upregulated in Bbs4 -/- mice...23 Glucose Homeostasis in Lean BBS Mice...24 Discussion...24 CHAPTER III COMMON BBS1 M390R MUTATION RESULTS IN GLUCOSE INTOLERANCE AND NEURONAL BBS1 HAS A NOVEL ROLE IN GLUCOSE HOMEOSTASIS...39 Introduction...39 Materials and Methods...41 Animals...41 GTT and ITT...42 Insulin Receptor Signaling...42 Immunoblotting...42 Results...43 vi

10 Glucose Homeostasis in Bbs1 KI...43 Defective IR Signaling in Bbs1KI...43 Neuronal Deletion of Bbs1 Alters Glucose Homeostasis...44 Insulin Receptor Signaling in Nestin Cre /Bbs1 fl/fl Mice...45 Discussion...46 CHAPTER IV BBS PROTEINS ARE REQUIRED FOR INSULIN RECEPTOR TRAFFICKING TO THE CELL MEMBRANE...57 Introduction...57 Materials and Methods...59 Cell Culture and Transfection...59 Immunohistochemistry...59 Biotin Labeling of Cell Surface Proteins...60 Statistical Analysis...60 Results...61 Physical Interaction between IR and BBS Proteins...61 IR is Not Targeted to Cilia...61 Knockdown of BBS1 and BBS2 Reduces the Amount of IR at the Plasma Membrane...62 Discussion...63 CHAPTER V CONCLUSIONS...73 REFERENCES...77 vii

11 LIST OF TABLES Table 1. BBS genes: Chromosomal locus, percent of cases, and report...15 Table 2. Phenotype of BBS mouse models...16 viii

12 LIST OF FIGURES Figure 1. The BBSome complex is composed of BBS1, 2, 7, 9, which have β- propeller domains, BBS4 and 8, which have α-solenoid domains, and BBS5 which contains PH domains and BBS Figure 2. Comparison of fasting blood glucose (a) and fasting plasma insulin (b) between Bbs mice and littermate WT controls. Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice have hyperglycemia (a) and hyperinsulinemia (b) Figure 3. Glucose tolerance test (a) revealed significantly higher glucose levels at all-time points in Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice compared to WT controls. Calculating the change in blood glucose (b) shows glucose intolerance in Bbs2 -/- and Bbs4 -/- mice at 60 minutes while Bbs6 -/- mice had unimpaired glucose tolerance...28 Figure 4. Insulin tolerance test (a) and change in blood glucose (b) revealed a significant insulin resistance in Bbs4 -/- mice throughout the test and Bbs2 -/- mice after 120 minutes compared to WT. Bbs6 -/- mice tended to have insulin resistance, but the difference in blood glucose is not statistically different Figure 5. Comparison of insulin-induced activation of Akt (phosphorylation) in liver, white adipose tissue, and skeletal muscle between Bbs4 -/- mice and littermate WT controls. After fasting, mice receive intervenous injection of vehicle or insulin. Bbs4 -/- mice have diminished insulin signaling in liver (a) adipose tissue (b), and skeletal muscle (c) compared to WT littermates Figure 6. Comparison of baseline Akt activity in liver, white adipose tissue and skeletal muscle between Bbs4 -/- mice and littermate WT controls. Bbs4 -/- mice have increased basal p-akt expression in liver, adipose tissue, and skeletal muscle Figure 7. Comparison of expression levels of genes involved in the fatty acid synthesis (a) and β-oxidation (b) in liver between Bbs4 -/- mice and littermate WT controls. Fatty acid synthesis genes, FASN and ACC1 are upregulated (a) while CPT-1 is a beta oxidation gene that is upregulated (b) in Bbs4 -/-/ liver tissue Figure 8. Comparison of body weight between Bbs mice and littermate WT controls under normal feeding condition (a) or under calorie-restriction (b). Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice are obese (a) and Calorie-restricted (CR) Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice have a normal body weight (b) Figure 9. Comparison of fasting blood glucose (a) and fasting plasma insulin (b) between calorie-restricted Bbs mice and wild type controls. Hypoglycemia in CR-Bbs4 -/- and hyperglycemia in CR-Bbs6 -/- mice (a). Hyperinsulinemia in CR-Bbs2 -/-, CR-Bbs4 -/-, CR-Bbs6 -/- mice when compared to wild type littermates...34 Figure 10. Glucose tolerance test in calorie-restricted (CR) Bbs mice and WT controls. Calorie-restricted Bbs4 -/- mice are hypoglycemic compared to WT during glucose tolerance test (a) although the change in glucose levels is significantly increased in CR- Bbs2 -/- and CR- Bbs4 -/- mice (b) ix

13 Figure 11. Insulin tolerance test in calorie-restricted Bbs mice and littermate WT controls. Calorie-restricted (CR) Bbs4 -/- mice are hypoglycemic compared to WT during insulin tolerance test (a) and have similar changes in glucose levels (b) Figure 12. Comparison of insulin receptor signaling in calorie-restricted Bbs4 -/- mice and littermate WT controls in liver (a), white adipose tissue (b), and skeletal muscle (c). Attenuation of insulin activation of p-akt in liver (a) in adipose tissue insulin activation of p-akt is diminished (b), and in skeletal muscle response to insulin activation is blunted (c) when compared to WT response Figure 13. Comparison of basal p-akt levels in calorie-restricted Bbs4 -/- mice and littermate WT controls in liver, white adipose tissue, and skeletal muscle. In calorie-restricted Bbs4 -/- liver, adipose tissue, and skeletal muscle have increased basal p-akt levels compared to WT littermates (b) Figure 14. Comparison of fasting body weight and glucose levels in Bbs1 KI and WT littermate controls. The Bbs1 KI mice develop obesity that is not resolved by fasting (a) and hyperglycemia compared to WT littermates (b) Figure 15. Glucose tolerance test with glucose levels (a) and the change in glucose levels (b) in Bbs1 KI and WT littermate controls. The Bbs1 KI mice develop hyperglycemia with significantly higher glucose levels at time 0, 30, and 60 minutes after glucose load (a). The Bbs1 KI mice have a similar change in glucose levels as the WT littermate controls during the first hour of the GTT but have a significantly elevated response to the glucose load after 120 minutes (b) Figure 16. Insulin tolerance test with glucose levels (a) and the change in glucose levels (b) in Bbs1 KI and WT littermate controls. Bbs1 KI mice have significantly higher glucose levels throughout the ITT while WT littermates have quite low glucose levels requiring rescue by intraperitoneal injection of glucose prior to completing the ITT Figure 17. Comparison of insulin receptor signaling using downstream p-akt (a) as marker of IR signaling and basal p-akt (b) in Bbs1 KI mouse embryo fibroblasts (MEF) and WT MEF littermate controls. After a 6 hour period of serum starvation, cells were treated with 100 nm insulin for 15 minutes. Bbs1 KI MEF cells have a diminished response to IR activation of p-akt (a) and have significantly higher basal p-akt expression (b) compared to WT MEF cells Figure 18. Comparison of Cre expression patterns in Nestin, Leptin Receptor b- form, and Proopiomelanocortin. Reporter mouse lines with Cre recombinase driven by Nestin, LRb and Pomc promoters express td tomato, a red fluorescent protein, to illustrate the expression patter of the Cre in each mouse line Figure 19. Comparison of body weight and fasting glucose levels in Nestin Cre /Bbs1 fl/fl, LRb Cre /Bbs1 fl/fl, and Pomc Cre /Bbs1 fl/fl mice and littermate WT controls. Pomc Cre /Bbs1 fl/fl mice are obese (a) although Nestin Cre /Bbs1 fl/fl and LRb Cre /Bbs1 fl/fl mice have normal body weights after fasting compared to x

14 WT littermates. Both Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice are hyperglycemic and LRb Cre /Bbs1 fl/fl mice are euglycemic (b) Figure 20. Glucose intolerance test with glucose levels (a) and change in glucose levels (b) in Nestin Cre /Bbs1 fl/fl, LRb Cre /Bbs1 fl/fl, and Pomc Cre /Bbs1 fl/fl mice and littermate WT controls. In Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice glucose levels in LRb neurons lack of Bbs1 does not impact glucose tolerance as shown by blood glucose over time (a) and when change in glucose levels are calculated there is a significant difference in the change in glucose levels at 30 minutes (b) Figure 21. Insulin tolerance test with glucose levels (a) and change in glucose levels (b) in Nestin Cre /Bbs1 fl/fl, LRb Cre /Bbs1 fl/fl, and Pomc Cre /Bbs1 fl/fl mice and littermate WT controls. Insulin resistance in Nestin Cre /Bbs1 fl/fl mice during the ITT (a) and when comparing change in glucose levels the Nestin Cre /Bbs1 fl/fl mice are significantly resistant compared to WT (b), while Pomc Cre /Bbs1 fl/fl and LRb Cre /Bbs1 fl/fl mice are sensitive to insulin in a similar manner to WT (b) Figure 22. Comparison of insulin receptor signaling in Nestin Cre /Bbs1fl /fl mice and littermate WT controls in liver (a), white adipose tissue (b), and skeletal muscle (c). Insulin receptor signaling is significantly impaired in the liver (a) of Nestin Cre /Bbs1fl /fl mice and is attenuated in the adipose tissue (b). Insulin receptor signaling is unimpaired in the skeletal muscle f Nestin Cre /Bbs1fl /fl mice(c) Figure 23. Sucrose gradient sedimentation analysis of BBS proteins and the insulin receptor. The BBSome is found in fractions 4-9 of the sucrose gradient along with its cargo proteins Ptc1, Smo, and the IR Figure 24. Shown are reciprocal co-immunoprecipitation in HEK293T cells with Flag-BBS17 and IR. The physical interaction between IR and BBS17 is seen in a transient transfection of HEK293T cells. This interaction can be seen whether IR antibody is used for immunoprecipitation (upper panel) or BBS17 tag is used to precipitate IR (lower panel) Figure 25. Immunohistochemistry of 3T3L1 cells to mark cilia (Acetylated α- Tubulin) and the Insulin receptor (β-subunit). The Insulin receptor does not localize to the cilia of 3T3L1 fibroblasts. The signal for the cilum marker is distinct from the signal for the insulin receptor Figure 26. Confirmation of shrna mediated knockdown of BBS1 (a) and BBS2 (b). Efficient knockdown of Bbs1 (a) and Bbs2 (b) genes in HEK293T cells using HA-tagged proteins for immunoblot recognition in the absence of antibodies against BBS1 and BBS Figure 27. Comparison of surface insulin receptor (a) and transferrin receptor (b) levels in HEK293T cells. Knockdown of Bbs1 or Bbs2 reduces the amount of IR (a) at the cell surface but does not change the amount of TfR (b) at the cell surface Figure 28. Comparison of surface insulin receptor (a) and transferrin receptor (b) levels in Bbs1 KI and WT MEF cells. The Bbs1 KI MEFs have a significant xi

15 reduction in the level of IR (a) at the cell surface while TfR (b) levels are unchanged Figure 29. Immunohistochemistry using indirect immunofluorescence of the insulin receptor without permeablizing the cell membranes of Bbs1 KI and WT MEF cells. The Bbs1 KI MEFs have decreased levels of IR immunofluorescence that can be visualized (a) and quantified (b) Figure 30. Comparison of surface insulin receptor (a) and transferrin receptor (b) levels in Bbs1 KI and WT brain tissue. In vivo biotinylation of Bbs1 KI brain tissue identifies a significant reduction of amount of IR (a) and a slight reduction in TfR (b) that is exposed to the circulation xii

16 LIST OF ABBREVIATIONS ABC ANOVA AUC BMI CCT/TriC CNS CTCF DAPI ECL ESRD ELISA EDTA GPCR HEPES HRP IF IFT IHC IQ IR Jak-Stat kda Avidin Biotin Complex Analysis of Variance Area under curve Body mass index Chaperonin containing TCP-1/ TCP-1 Ring Complex Central Nervous System Corrected total cell fluorescence 4',6-diamidino-2-phenylindole Enhanced Chemiluminescence End stage renal disease Enzyme-linked immuno-substrate assay Ethylenediaminetetraacetic Acid G-protein Coupled Receptor 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid Horse Radish Peroxidase Immunofluorescence Intraflagellar transport Immunohistochemistry Intelligence quotient Insulin receptor Janus Kinase-Signal transducer and activator of transcription Kilodaltons xiii

17 LAP LepR MO mrna NaCl NaF Na 3 VaO 4 PFA PVDF RMR SEM Shh shrna TBS-T TfR Localization and tandem affinity purification Leptin Receptor Morpholino oligonucleotides Messenger Ribonucleic Acid Sodium Chloride Sodium Fluoride Sodium Vanadate Paraformaldehyde Polyvinylidene Difluoride Resting metabolic rate Standard Error of the Mean Sonic hedgehog pathway Short-hairpin Ribonucleic Acid Tris Buffered Saline Tween Transferrin receptor xiv

18 1 CHAPTER I INTRODUCTION Clinical Features of Bardet-Biedl Syndrome Bardet-Biedl Syndrome (BBS) is an autosomal recessive syndrome. BBS patients are diagnosed with the syndrome through identification of clinical signs and symptoms. The primary features of BBS are rod-cone dystrophy, postaxial polydactyly, truncal obesity, learning disabilities, and urogenital- renal anomalies [1, 2]. Secondary features of BBS are more numerous and vary between patients. These include speech delay, developmental delay, behavioral abnormalities, poor coordination, diabetes mellitus, cardiovascular anomalies, hepatic involvement, craniofacial dysmorphism, eye abnormalities, defects in brain volumes and anosmia. A patient receives a diagnosis based on the clinical findings of four primary features or three primary features with the addition of two secondary features [1, 2]. This broad range of clinical features becomes evident over an expanded period of time. The population survey done by Beales et al. found the average age of onset of symptoms was 3 years old while the average age of diagnosis was 9 years old, which is 6 years between recognizable symptoms and diagnosis [2]. Attempts to better understand and measure the difference in BBS in regards to the presentation of primary features have been ongoing. Other features of BBS were included to some extent, although the secondary features were not uniformly examined. Questions of mechanism are raised by the various features and the variability of presentation in BBS. Following is an overview of the current literature for the five cardinal features of BBS. Polydactyly Sixty-nine percent of BBS patients that were surveyed by Beales et al. were born with polydactyly, defined by at least one accessory digit. Other surveys of BBS patients

19 2 identified polydactyly in percent of patients with about a quarter of those patients exhibiting postaxial polydactyly of both hands and feet [2-4]. Vision Impairment The visual impairment in BBS has been labeled rod-cone dystrophy or atypical retinitis pigmentosa [2, 5]. While 90 percent of BBS patients exhibit rod-cone dystrophy, retinal disease can range from mild to severe [1, 2, 6]. Beales et al. report loss of night vision on average at 8 years of age. Night vision, visual acuity or visual field losses are the typical presenting symptoms when patients either seek or are referred for an ophthalmological examination. In the study of retinal disease by Cox et al. where all ten patients carried a mutation in the BBS1 gene, the severity of visual defect was independent of either age or BBS1 genetic defect [6]. When retinitis pigmentosa was investigated within the families carrying mutations in BBS2, BBS3, and BBS4 loci, retinal disease was found to be severe with onset of legal-blindness (20/200 visual acuity or 20% loss of visual fields) by the second decade of life [7, 8]. Obesity Body mass index (BMI = weight (m)/height (kg) 2 ) is a common way to measure excess body weight and the presence of obesity in adolescents and adults. Overweight is a BMI 25, and obese is a BMI 30 [9]. Depending on the reports, obesity in BBS patients is present in 52-96% of cases with an early onset as young as 1-2 years [2, 3, 8, 10]. Attempts have been made to characterize the mechanism of BBS- related obesity [2, 11-13]. In 2003, Grace et al. found no difference in resting metabolic rate (RMR) after adjusting for sex, age, lean mass and fat mass when comparing BBS patients to nonsyndromic obese patients [12]. Feuillan et al. also compared obese BBS patients to nonsyndromic obese patients matched for age, sex, race, and BMI-Z scores in order to

20 3 investigate obesity-associated physiological differences [11]. One such physiological difference is the higher than expected circulating levels of the adipocyte-derived hormone, leptin, found in BBS patients, suggestive of hyperleptinemia and leptin resistance [11]. In 2013, hyperphagia with food seeking activity was identified in BBS patients when compared to age, sex, and BMI adjusted controls [13]. Obesity is a common feature of BBS and is characterized by hyperleptinemia and hyperphagia. Learning Disability The neurocognitive defects in BBS patients ranges from mental retardation (42 percent reported by Green et al.) to borderline range of measurement of intelligence (IQ) [3, 8, 14, 15]. In 2002, Barnett et al. examined 52 BBS patients in order to identify a behavioral phenotype associated with BBS. The majority of patients who underwent a full scale IQ test had scores in the mild mental retardation range [15]. Vision loss, especially in older patients, makes it more difficult to obtain measurements for many neurological and behavioral phenotypes [2, 3, 14, 15]. Brinckman et al. recommend testing at a younger age before the onset of vision loss and use of non-visual means to testing [14]. Urogenital-Renal Anomalies There are many case reports of BBS patients identifying hypogonadism in males and vaginal atresia, primary ovarian dysfunction, and hydrometrocolpos (fluid in vaginal cavity and uterus causing abdominal distension) in females [2, 3, 5, 8, 16-18]. Renal anomalies and end stage renal disease (ESRD) accounts for severe morbidity and mortality in BBS patients [8, 19]. In a study to understand the natural history of BBS, O Dea et al. compared 38 patients with 58 unaffected siblings. They identified 25 percent of BBS cases with chronic renal insufficiency while only a single unaffected sibling had

21 4 mild renal impairment [8]. Renal failure was found in 75 percent of patients at the time of death [8]. Along with ESRD, malformation of the kidneys has also been identified. Renal abnormalities include fetal lobulation, abnormal calyces, calyceal cysts, diffuse cortical loss, and focal cortical loss. BBS patients were also found to have a defect in urinary concentration that is present even when kidney function is normal and in the absence of renal cysts [20]. Diabetes Mellitus The 1989 report by Green et al. identified nine of twenty BBS patients within the Newfoundland cohort that had diabetes mellitus by criteria of that time period. However, using the current American Diabetes Association (ADA) Clinical Practice Recommendations where 2-hour plasma >140 mg/dl (7.8 mmol/l) to 199 mg/dl (11.0 mmol/l) during an oral glucose tolerance testing (OGTT), all the non-diabetic BBS patients in that study would be classified as prediabetic due to impaired glucose tolerance [3, 21]. Serum insulin measurements taken during the OGTT were well-above the current 207 pmol/l cutoff for normal levels of serum insulin [22]. Thus, hyperglycemia in the presence of hyperinsulinemia signifies insulin resistance in this cohort [23]. Ten years later, Beales et al. reported new criteria for identification of BBS patients, but they report that only a minority of patients had fasting glucose measurements or glucose tolerance test during the course of the study [2]. According to their report, 6% of BBS patients in the population surveyed had diabetes mellitus [2]. In the early study by Green et al., 15% of patients were being treated for Type 2 diabetes (T2DM), but the diagnosis of T2DM was made as a result of the study so that thirty percent of BBS patients were found to have T2DM [3]. O Dea et al. confirmed this finding in a follow-up study where they identified 32% of BBS patients that were being treated for diabetes mellitus with insulin or oral hypoglycemic agents, but did not measure fasting plasma glucose, OGTT or

22 5 serum insulin levels [8]. The patient cohort studied by O Dea et al. had a mean age of 35 years with 25% developing diabetes by the age of 35, and 50% were diabetic by the age of 55 years [8]. In the most recent follow-up to the Newfoundland cohort of BBS patient, Moore et al. identified 48% prevalence of diabetes mellitus with median age of onset of 43 years [24]. In a recent study from Leslie Biesecker s group, BBS patients compared to BMI-Z-matched controls had a mean age of 14.8 compared to years of age for BBS patients in the O Dea report [11]. In this younger BMI-Z-matched cohort the number of patients with diabetes was not different from controls [11]. Also, the prevalence of impaired glucose tolerance was not statistically different between BBS patients and controls [11]. However, the BBS patients had higher fasting insulin levels although the significant difference was lost after multiple comparison correction [11]. Finally, there is a case report of a 16-year old male BBS patient with a BMI of 52 who underwent bariatric surgery [25]. A three-year follow-up of this patient showed improvement in BMI, liver function, hypertension, and lipidemic profiles, but insulin levels remained high [25]. While specific studies have been undertaken to investigate phenotype and molecular mechanisms of obesity, hydrocephalus, renal anomalies, and behavioral manifestations of BBS, no studies have been undertaken to directly assess glucose tolerance or insulin sensitivity in BBS. The reports of diabetes mellitus discussed here were all presented as secondary findings. The data collected and presented here will begin to assess the BBS-associated diabetes phenotype using the BBS mouse models. Epidemiology of Bardet-Biedl Syndrome As expected of an autosomal recessive disorder the prevalence of BBS in the general population is low. The prevalence of BBS in North America and Europe is one in 100,000 to one in 160,000 live births [12, 25]. Within the Puerto Rican population BBS

23 6 may be present in one in 75,000 live births [4]. In Newfoundland, due to a founder effect in an isolated population, the prevalence of BBS is as high as one in 17,500 [19, 24]. Similarly in another isolated population, the Arab Bedouins, where consanguinity is more common, the prevalence of BBS is one in 13,500 [12]. Although the incidence of BBS is rare, the features of BBS are common, which makes this a useful model to investigate mechanisms of common diseases such as obesity and diabetes. Identification of BBS Genes Mutations found in BBS patients whether by positional cloning,, protein interaction analysis, or exome sequencing have identified 19genes associated with BBS [5, 26-46]. However, about one quarter of BBS patients do not have an identifiable mutation in a known BBS gene. Table 1 lists the BBS genes, chromosomal location and initial report of each BBS gene. Structure and Function of BBS Protein As the BBS genes were identified, work began towards understanding the structure and function of the BBS proteins they encoded. Model organisms including mice, nematodes, and zebrafish, were initially employed to begin to dissect the molecular basis for the features of BBS and identify the function of BBS proteins [5, 47-54]. The connection between BBS proteins and cilia was made along with the identification of BBS8 and its cellular localization to the basal body of ciliated cells such as NIH3T3 cells and IMCD3 cells and within Caenorhabditis elegans, [5]. This agreed with other findings of homologous BBS genes within ciliated organisms that were expressed specifically in ciliated cells [30, 34, 55]. Complex organelles, cilia can be found in almost every human cell and are conserved in organisms ranging from algae, worms, and zebrafish to mammals [56, 57]. Cilia may be classified as primary or non-

24 7 motile with an arrangement of microtubules in a 9+0 (outer doublet) ringed projection from the cell or motile cilia with 9+2 arrangement of microtubules [56, 57]. Ciliated cells of particular importance in BBS include the rod and cone cells of the retina, olfactory epithelium, kidney epithelial cells, and ependymal cells that line the ventricles of the brain. When a search for homology to other known proteins was performed, BBS1, BBS2, BBS5 and BBS7 did not share significant homology to any other proteins [34, 36, 38]. BBS3, also known as ARL6, is a member of the RAS superfamily of GTPases [29, 30]. BBS4 and BBS8 are mainly comprised of a protein-protein interaction motif, the tetracopeptide repeat (TPR) [58]. BBS9 also lacked sequence homology to any other proteins and did not appear to contain any known functional domains upon its initial discovery [59]. BBS6, BBS10, and BBS12 share sequence homology to the CCT family of group II chaperonins and were shown to form a complex with BBS7 and six additional CCT proteins [60]. The CCT chaperonin complex is composed of two stacked rings made up of radially arranged subunits (CCT1-8) and mediates protein folding in an ATPdependent manner. Analysis of BBS11 revealed a completely different function. BBS11, also known as TRIM32 (tripartite motif containing gene 32), is involved in protein ubiquitination and consists of the following common domain structures; RING finger, B- box, and a coiled-coil motif [61].BBS is caused by a missense mutation P130S, in the B- box domain of BBS11 (TRIM32) [28]. According to sequence and secondary structure analysis, BBS17 was found to be mostly alpha-helical with a C-terminal coiled-coiled domain, in which a leucine-zipper domain is contained [62]. Upon their identification therefore, the majority of BBS proteins had no similarity to other proteins and no known function. In vitro methods using various cell culture lines combined with biochemical techniques yielded clues to the function of BBS proteins. Localization and tandem affinity purification (LAP) technology applied to BBS4 and mass spectrometry were

25 8 instrumental in identification of a stable complex of BBS proteins, termed the BBSome (Figure 1) [63]. Purification of BBS4 recovered stoichiometric amounts of BBS1, 2, 5, 7, 8, 9 and BBIP1 (BBSome interacting protein-1), recently identified as BBS18. These proteins were consistently found in the same fractions using sucrose gradient sedimentation analysis [63, 64]. The Bbs3 gene encodes the GTPase Arl6 and recruits the BBSome onto membranes [29, 30, 65]. Seo et al. used coimmunoprecpitation assays to reveal the interaction of BBS12 with BBS6 and BBS10 and identified the requirement of BBS6, 10 and 12 for complete BBSome assembly [60]. These BBS chaperonins (BBS6, BBS10, BBS12) were found to form a complex with CCT chaperonins and together mediate BBSome assembly [60]. Using antisense morpholino oligonucleotides (MOs) to knockdown components of the BBSome bbs2, bbs4, bbs5, bbs7, and bbs8 and BBS chaperone bbs6,genes in zebrafish embryos, the functional role of BBS proteins in retrograde intracellular transport was identified [52]. The identification of the BBS complexes represents a major finding in the field as it explains the overlapping phenotypes that are caused by mutation in any of the BBS genes, as well as the genetic heterogeneity of this syndrome. The role played by BBS proteins in ciliogenesis has been studied as it appears to provide a molecular link to the heterogenous phenotype seen in BBS [63, 65, 66]. In 2007, Nachury et al. found an interaction between the BBSome and Rab8-GTP leading to ciliary membrane elongation, which indicates function in ciliary membrane trafficking [63]. Further evidence of membrane ciliary trafficking was provided by the ability of the BBSome to polymerize along synthetic liposomes and form patches of polymerized coat [65]. In 2010, Jin et al. recognized common structural elements of the BBSome that are shared by coat proteins COPI, COPII, and clathrin coats, which are involved in vesicle transport [65]. The BBS4 and BBS8 TPR repeats are predicted to fold into rod-shaped α- solenoids in a similar manner to clathrin, COP1, and COPII proteins [65]. The

26 9 relationship between the BBSome and coat complexes can also be seen in the β-propeller folds in the N-termini of BBS1, BBS2, BBS7, and BBS9. Finally, clathrin adaptors and COPI share an appendage domain with BBS1, BBS2, BBS7, and BBS9 C-termini, consisting of the γ-adaptin ear (GAE) motif alone or fused to an α/β platform [65]. Jin et al also identified the role of BBS3, which functions as a GTPase, to target the BBSome to primary cilia. Because the BBSome is required for localization of some receptors to the cilia, it is thought to act as a coat complex for trafficking to cilia. The biochemical studies by Zhang et al provide valuable insight into the assembly of the BBSome [67]. The BBSome core complex is composed of BBS2, BBS7, and BBS9, with BBS2 directly interacting with BBS7 and BBS9 [67]. BBS9 acts as a scaffold of the BBSome via its direct interaction with BBS1, BBS5, and BBS8, which are incorporated independently into the BBSome [63, 67]. The final components that are added to the BBSome are BBS1 followed by BBS4. In the case of the Bbs1 M390R/M390R mutation, the majority of BBS4 is dissociated from the BBSome in the presence of this mutation, and BBS1 is also missing from the BBSome complex [67]. The first step is a BBS-chaperonin complex containing BBS7 along with BBS6, BBS12 and CCT/TRiC proteins. Because the CCT/TRiC chaperonins are required for actin folding, cellular processes that require actin polymerization, such as cilia lengthening, cellular growth, morphology, and motility, may be impaired in BBS [68, 69]. Regulation of the BBSchaperonin complex is mediated by BBS10 [67]. In the next step BBS2 binds to BBS7, which releases BBS6 and BBS12. In the absence of either BBS7 or BBS6, the levels of BBS2 are greatly reduced. Therefore, in the context of Bbs6 -/- mice, BBSome formation is inhibited because the BBSome core complex requires BBS2. Based on identification of the step-wise assembly of the BBSome, it is possible to interpret the nature of the BBSome disruption in the Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mouse models. BBS mouse models are discussed further in the following section. However to begin to address the experimental models used throughout this work, a brief mention is

27 10 made at this point. In the Bbs2 -/- mouse model a component of the BBSome core complex is disrupted thereby disrupting formation of the complete BBSome. In the Bbs4 -/- mouse model the BBSome lacks only the final component, BBS4. Because BBS4 is added last it is considered a peripheral subunit; it also contains intrinsic ability to localize to the centrosome and interacts with other proteins [67, 70]. In the Bbs6 -/- mouse model, a component of the BBSome core complex, BBS2 is degraded thereby also disrupting formation of the complete BBSome. The function of BBS proteins and the BBSome is actively being investigated. In the central nervous system, neuronal BBS proteins were identified by Kirk Mykytyn and colleagues as a requirement for localization of G-protein coupled receptors (GPCR) in primary cilia while BBS proteins were not required for neuronal ciliogenesis [71]. Specifically, ciliary localization of somatostatin receptor 3 (Sstr3) and melaninconcentrating hormone receptor 1 (Mchr1) was impaired in Bbs2 -/- and Bbs4 -/- mice [71]. Kirk Mykytyn s group also identified the requirement of BBS proteins for translocation of dopamine receptor 1 (D1) out of neuronal cilia with BBS5 specifically interacting with D1 [72]. The translocation of D1 to neuronal cilia was found to be a dynamic process with D1 moving in and out of the cilia. More recently Loktev and Jackson identified neuropeptide Y 2 receptor (NPY2R) as another GPCR that required BBS protein for its ciliary localization in neurons [73]. Hélène Dolfus group studied the role of BBS proteins in renal cells. They began by identifying a cohort of BBS patients and with a defect in urinary concentration that was present when kidney function was normal and in the absence of renal cysts [20]. Rather than requiring BBS proteins for specific localization of the antidiuretic hormone vasopressin-2 receptor (AVPR2) in primary cilia of the collecting duct epithelium, BBS proteins are required for ciliogenesis. In either case the loss of BBS proteins leads to loss of AVPR2 at the primary cilia and the ability for the cells to reabsorb water in the collecting ducts [20].

28 11 There are several examples of BBSome mediated retrograde transport. The cilium of ependymal cells in Bbs1 M390R/M390R mice are either abnormally lengthened or contain abnormally swollen distal regions, which is suggestive of a defect in retrograde transport [47]. Additionally the requirement for the BBSome to transport the D1 out of the cilia appears to provide another indication of the role of the BBSome in retrograde movement [72]. The zebrafish melanosome transport assay examines transport in non-ciliated cellular processes; the assay measures the time it takes for melanophores to retract using retrograde transport to perinuclear positions [52]. These studies yielded a quantifiable measure of the retrograde transport defect in MO bbs knockdown zebrafish larvae. However, a defect in anterograde transport is may appear as a disruption in retrograde transport because the two processes should occur in equilibrium. Therefore, it is possible to identify a trafficking defect in BBS, while at this time it is not possible to unequivocally state that it is a defect in retrograde transport. Insights from BBS Mouse Models The development of mouse models of BBS allowed substantial insights into the function of BBS proteins. Bbs2, Bbs4, and Bbs6 null mice (Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- ) were generated first using targeted deletion of each gene [49-51]. Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice developed primary and motile cilia, but not flagella along with several other BBS phenotypes including obesity, retinal degeneration, and infertility [49-51]. BBS mice have increased adiposity. BBS obesity is characterized by decreased energy expenditure, and hyperphagia. The variability in both inter- and intra-familial in BBS phenotypes among patients raised the question whether the Bbs-null mice also displayed phenotypic variability [7, 48, 74, 75]. Although Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice were obese only Bbs4 -/- and Bbs6 -/- mice developed hypertension [76]. In another mouse model

29 12 which carries a commonly observed missense mutation, the Bbs1 M390R/M390R (Bbs1 KI), was generated [47]. Similar to other Bbs mouse models, Bbs1 KI mice lacked flagellated spermatocytes, developed photoreceptor degeneration and obesity, and like the Bbs2 -/- mice were normotensive [47, 77]. A new finding that was consistent within the Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice was the presence of ventriculomegaly [47]. Interestingly, the Bbs3-null (Bbs3 -/- ) mouse was found to have the most severe hydrocephalous phenotype, in addition to retinal degeneration, mild obesity, elevated arterial pressure, and loss of sperm flagella [53]. Finally the Bbs7 null (Bbs7 -/- ) model was generated using targeted deletion that replaced exon 5 with a neomycin cassette [54]. Like the previously described mouse models, Bbs7 -/- mice also developed obesity, retinal degeneration, abnormal sperm flagella, and ventriculomegaly. However, the difference was that the motile ependymal cilia in the brain were abnormal [54]. Table 2 summarizes the phenotypes of the various BBS mouse models. These mouse models represent a unique tool to begin to understand the molecular mechanisms that underlies BBS phenotypes. For example, Seo et al. used the Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice to identify the role of BBS proteins in mediating leptin receptor (LepR) trafficking [78]. The hormone leptin is produced and secreted mainly by adipocytes and circulates in proportion to adipose tissue mass. Leptin regulates energy expenditure and food intake by binding to the long form of the LepRb in neurons within the brain, particularly in the hypothalamus. LepRb, which is densely expressed in the arcuate nucleus of the hypothalamus activates downstream signaling through phosphorylation of several proteins including Stat3 to mediate leptin effects on energy expenditure and food intake. LepRb was found to interact with BBS1 and silencing of BBS genes alters the trafficking of the receptor. The defect in LepR trafficking was found to be associated with impaired LepR signaling in the hypothalamus of Bbs-null mice. The disruption resulted in leptin resistance and is a major cause of BBS obesity [78]. This work led to a study of leptin levels in BBS patients. As mentioned above, Feuillan et al.

30 13 reported that BBS patients display elevated plasma leptin levels compared to body mass matched control patients indicating that BBS subjects are more leptin resistant [11]. BBS mouse models have been used to understand the molecular mechanisms causing hydrocephalus, abnormally dilated ventricles in BBS patients and mouse models [47, 53, 54, 79, 80]. Carter et al. used the Bbs1 M390R/M390R mouse model to determine that ventricular dilation began before the maturation of ependymal cells, whose motile cilia defect was thought to explain the hydrocephalus. This study identified the combination of increased apoptosis and reduced proliferation of a specific subset of neural progenitor cells (NPCs) expressing NG2 and PDGFR-α as the main defect accounting for the hydrocephalus phenotype [81]. The Bbs1 fl/fl mouse line contains flanking loxp sites around exon 3. When it was crossed to a mouse line expressing Cre recombinase under control of the the PDGFR- α promoter, Bbs1 expression was specifically knocked down in a subset of NPCs. They identified that the loss of Bbs1 in these NPCs was sufficient to cause hydrocephalus [81]. BBS mouse models have been useful in identifying mechanism of BBS phenotypes. First, these mouse models were shown to recapitulate the primary features of obesity, retinopothy, and renal malfunction. In addition, BBS mouse models are helping to elucidate the mechanisms of obesity, retinopathy, and hydrocephalus. In the following studies, we used the BBS mouse models to investigate the role of BBS proteins in insulin resistance. Hypothesis and Study Design The central hypothesis is that functional BBS proteins are required for maintenance of glucose homeostasis. The hypothesis is based on the requirement for BBS proteins to maintain LepRb trafficking and signal transduction. The long form of the leptin receptor (LepRb) is a membrane bound receptor found in the hypothalamus and

31 14 other areas of the brain. The hormone leptin is made and released by the adipose tissue, when it binds the LepRb it begins a signaling cascade. This allows the adipose tissue to communicate with the region of the brain involved in regulating food intake and energy expenditure. BBS mouse models have leptin resistance, which is not improved by elimination of obesity. The persistent leptin resistance suggests mistrafficking of the receptor. We propose that a similar phenomenon occurs with the IR, such that mistrafficking of the IR will impair the ability of insulin to regulate glucose homeostasis. The rationale for this work is that by knowing the molecular mechanisms of BBSassociated diabetes, we can apply that knowledge to better understand the effects of T2DM and impact therapies. The work reported in chapter 2 investigates the hypothesis that BBS proteins and BBSome proteins specifically mediates insulin receptor (IR) trafficking to the cell surface and loss of BBS proteins leads to IR mistrafficking. The work reported in chapter 3 tests the hypothesis that BBS null mice have a disruption in IR signaling in the presence and in the absence of obesity. Furthermore, BBS null mice have a disruption of glucose homeostasis that is independent of obesity. The experiments in chapter 4 test the hypotheses that the hypomorphic allele of the most common Bbs1 mutation, a substitution of arginine for methionine (M390R), also displays a BBS-associated diabetes phenotype and impaired IR signaling. This chapter also tests the hypothesis that neuronal expression of Bbs1 is required for maintenance of glucose homeostasis and insulin sensitivity. Data discussed in the following chapters advance our understanding of the function of the BBS proteins and provide a novel mechanism of insulin resistance.

32 15 Table 1. BBS genes: Chromosomal locus, percent of cases, and report Gene symbol /Alt Locus BBS patients (%)* Report BBS1 11q Mykytyn et al BBS2 16q Nishimura et al BBS3/ARL 6 3p Chiang et al 2003 Fan et al BBS4 15q Mykytyn et al BBS5 2q Li et al BBS6/MKKS 20p Katsanis et al Slavotinek et al BBS7 4p Badano et al BBS8/TTC8 14q Ansley et al BBS9 7p Nishimura et al BBS10 12q Stoetzel et al BBS11/TRIM32 9q33.1 <0.4 Chiang et al BBS12 4q Stoetzel et al BBS13/MKS1 17q Leitch et al BBS14/CEP290 12q Leitch et al BBS15 2p15 ND Kim et al BBS16 1q43 ND Billingsley et al BBS17/LZTFL1 3p21.3 ND Seo et al Marion et al BBS18/BBIP1 10q25.2 ND Scheidecker et al BBS19/IFT27 22q13.1 ND Aldahmesh et al * From [1]; ND not determined

33 16 Table 2. Phenotype of BBS mouse models Obesity Retinal Degeneration Hydrocephalous Polydactyly Hypertension Abnormal Sperm Flagella Bbs1 KI Bbs2 -/ Bbs4 -/ Bbs6 -/ Bbs3 -/ Bbs7 -/ ND + + phenotype present, phenotype not present, ND not determined

34 Figure 1. The BBSome complex is composed of BBS1, 2, 7, 9, which have β-propeller domains, BBS4 and 8, which have α-solenoid domains, and BBS5 which contains PH domains and BBS18. 17

35 18 CHAPTER II THE GENETIC DEFECT IN BBS2, BBS4, AND BBS6 NULL MOUSE MODELS CONTRIBUTES TO THE DIABETES PHENOTYPE Introduction Bardet Biedl Syndrome (BBS) displays heterogeneity in both the genes involved and clinical features. Mutations in 19 genes have been associated with BBS. Eight BBS proteins (BBS1, 2, 4, 5, 7, 8, 9 and 18 also known as BBIP1) have been shown to form a stable complex, termed the BBSome. Assembly of the BBSome is mediated by three BBS proteins (BBS6, BBS10, BBS12) which form a complex with another family of chaperonins (CCT/TriC). The BBSome is involved in formation and maintenance of primary cilia and vesicle trafficking [52, 66]. The cardinal clinical features of BBS include obesity, degenerative retinopathy, polydactyly, renal dysfunction, hypogonadism, and mental retardation [2, 3]. In addition, diabetes mellitus is commonly associated with BBS, but the mechanisms remain unknown [2, 3, 10, 25]. Prior to development of diabetes, BBS patient exhibit high levels of plasma insulin while maintaining normal glucose levels, meeting the criteria for insulin resistance [11]. Our objective is to understand the molecular mechanism of BBS-associated diabetes. The experiments discussed in this chapter were designed to identify insulin resistance and impaired glucose tolerance in the mouse models of BBS. Bbs2-null mice (Bbs2 -/- ), Bbs4-null mice (Bbs4 -/- ), and Bbs6-null mice (Bbs6 -/- ) were previously generated using targeted disruption of exons 5-14, 6-11, and exon 3, respectively [5, 49-51]. Glucose tolerance tests (GTT) and insulin tolerance tests (ITT) were performed to assess the responses to a glucose load and to exogenous insulin. Activation of insulin receptor signaling was measured by changes in levels of phosphorylated Akt (p-akt), a downstream target of activated insulin receptor. The role of obesity in the defect in insulin receptor signaling in Bbs4 -/- mice was tested by preventing obesity in BBS null

36 19 mice using calorie restriction. The experiments in this chapter demonstrate the genetic defect in BBS directly contributes to diabetes phenotype independently from the obesity phenotype Materials and Methods Animals Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice were bred on mixed 129SvEv and C57B/6J strains. As there was no evidence of sexual dimorphism, both male and female mice were used for each experiment. Littermates were used as controls. The use of animals as described in this chapter was approved by the University of Iowa Animal Care and Use Committee. All animal testing was performed based on guidelines set forth by the National Institutes of Health. Mice were housed at the University of Iowa Animal Care facility in a temperature and humidity controlled room on a 12 hour light / dark cycle with free access to food and water except in the calorie restriction experiment where food intake of BBS mice was restricted. At eight weeks of age, which is before the onset of obesity, BBS mice were given 75-80% of the daily food intake of their littermate s intake for eight weeks. Glucose and Insulin Measurements Fasting blood glucose measurements were taken using a glucometer (OneTouch Ultra) with a blood sample taken from a tail snip. Insulin levels were measured using an ELISA (Crystal Chem). The mice were fasted overnight followed by cardiac puncture for blood collection. Plasma was isolated using 0.5M EDTA and centrifugation to separate plasma from red blood cells.

37 20 Glucose Tolerance Test and Insulin Tolerance Test Glucose tolerance test (GTT) and insulin tolerance test (ITT) were performed as previously described [82, 83]. Four to five month old mice were fasted overnight (GTT) or 5 hours (ITT) in a clean cage. Body weight and basal blood glucose measurements were taken before intraperitoneal injection of glucose (2g/kg) or insulin (1 unit/kg). Blood glucose measurements were assessed at multiple time points during the two-hour test. Insulin Receptor Signaling Assay Mice were fasted overnight in a clean cage. On the morning of the test, mice were weighed and anesthetized with xylazine-ketamine injections. Once the mice were anesthetized, the jugular vein was cannulated for either saline or insulin (5U/kg) injection. Mice were sacrificed after 15 minutes at which time tissues were collected. Tissues were homogenized in lysis buffer (50 mm HEPES ph7.5, 137 mm NaCl, 1% NP-40, 0.25% Na-deoxycholate, 2 mm EDTA, 2 mm Na 3 VO 4, 10 mm NaF, 10% glycerol, protease inhibitor cocktail[roche complete Mini, EDTA-Free ]). The extracts were centrifuged at 14,000 g for 30 minutes at 4 C. Protein concentration of the obtained supernatant was measured using the Bradford protein assay [84]. Immunoblotting Proteins in whole tissue lysate were resolved by 9% acrylamide SDS-PAGE, and the proteins were then transferred to PVDF membranes. Membranes were blocked in 5% nonfat dry milk in Tris-buffered saline (NaCl, KCl, Tris-base)with 0.1% Tween-20 (TBST) for 1 hour at 25 C and incubated with anti-akt and anti-p-akt antibodies (Cell Signaling) at 1 to 2,000 dilution primary antibody at 4 C overnight. Membranes are further incubated with secondary goat anti- rabbit antibody conjugated with horseradish

38 21 peroxidase (HRP) for 2 hours at 25 C. Visualization was performed with enhanced chemiluminescence (ECL) followed by autoradiography. Statistical Analysis All data are expressed as mean ± standard error. A two-way ANOVA was used to compare BBS mice to wild type (WT) littermates for GTT and ITT analyses and insulin signaling assay. The first factor was genotype, and the second factor was time point for GTT and ITT analyses, while the second factor was treatment for insulin signaling. Multiple-comparison testing following two-way ANOVA was performed using a Bonferroni t-test. One-way ANOVA was used to compare more than two groups, such as blood glucose and body weight. Rank-ANOVA was used whenever the data did not follow a normal distribution. A two-tailed p-value< 0.05 was considered significant for all analyses. Results Glucose Homeostasis We began by assessing glucose homeostasis in the BBS mouse models by measuring fasting glucose and insulin levels. Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice were significantly hyperglycemic (173 ±22 mg/dl, 151 ±11 mg/dl, 145 ±8 mg/dl) and hyperinsulinemic (2 ± 0.7 ng/dl, 2 ± 0.5 ng/dl, 2 ±0.3 ng/dl) compared to wild type (WT) littermates (111 ±9 mg/dl, 0.5 ±0.1 ng/dl) (Figure 2). Next, Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice were examined for the ability to respond to glucose loads and were found to have a significantly higher glucose levels throughout the GTT. The Bbs2 -/- and Bbs4 -/- mice had glucose levels above 200 mg/dl during the first hour of the GTT and above 150 mg/dl at the end of the GTT, while Bbs6 -/- were less severely affected with glucose levels above

39 mg/dl during the first hour of the GTT (Figure 3a). In contrast, WT glucose levels reached a maximum of 159 ± 13 mg/dl after 15 minutes and steadily decreased to near starting glucose levels after 2 hours (Figure 3a). In terms of the change in glucose levels, in WT the greatest change in glucose levels was 47 ±8 mg/dl 15 minutes after the start of the GTT. The change in glucose levels was greatest after 30 minutes in the Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice (Δ83 ± 16 mg/dl, Δ72 ± 9 mg/dl, Δ67 ± 10 mg/dl) compared to WT (Δ46 ±9 mg/dl) at the same time point (Figure 3b). The Bbs mice had a significantly higher response to the glucose load during the GTT. The ability to respond to exogenous insulin was tested using an ITT. Bbs2 -/-, Bbs4 - /-, and Bbs6 -/- mice were found to have significant insulin resistance with extremely high blood glucose levels at each time point (Figure 4a). In particular the WT mice reached a minimum of 47 ± 4 mg/dl at the 60 minute time point, while the Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice reached a minimum that was between 82 mg/dl and 94mg/dl (Figure 4a). The change in glucose levels are markedly decreased in Bbs2 -/- and Bbs4 -/- mice throughout the ITT. In particular at the start of the test, the WT have a change of 57 ± 13 mg/dl and 97 ± 13 mg/dl, while the Bbs4 -/- mice have a blunted response with no change in glucose levels between 15 minutes (Δ11 ± 7mg/dl) and 30 minutes (11 ± 32 mg/dl) (Figure 4b). At the 60 minute time point which was the minima of the curve the WT had a change of 106 ±13 mg/dl, while the Bbs2 -/- and Bbs4 -/- mice had a change of 70 ±16 mg/dl and 48 ±7 mg/dl, respectively (Figure 4b). The Bbs6 -/- mice had a change of 110 ±15 mg/dl, which was similar to the WT response (Figure 4b). Insulin Signaling Defect in Bbs4 -/- Mice Since the degree of insulin resistance was comparable among the three Bbs-null mouse models, Bbs4 -/- mice were chosen for use in follow up studies. When wild type mice received intravenous injection of insulin, a significant increase in levels of p-akt in

40 23 liver, skeletal muscle, and white adipose tissue, three target tissues of insulin action was observed (Figure 5 and 6). In contrast, insulin did not increase p-akt levels in liver, skeletal muscle and adipose tissue of Bbs4 -/- mice (Figures 5). In fact, a comparison of p- Akt levels in the insulin treated groups indicated a significant reduction of insulin p-akt activation in Bbs4 -/- tissues compared to WT (Figure 5). However, in all three tissues examined it was clear that the basal p-akt levels were much higher in Bbs4 -/- mice relative to controls (Figure 6). Statistical analysis identified a significant interaction between genotype and insulin treatment in all three tissues. Fatty acid synthesis genes are upregulated in Bbs4 -/- mice Insulin action in the liver is impaired in Bbs4 -/- mice. Therefore the ability to regulate fatty acid metabolism was examined. With impaired insulin action in the liver, we would expect dysregulation of metabolic gene expression. Expression levels of genes involving fatty acid synthesis that were examined include fatty acid synthase (FASN), acetyl-coenzyme A carboxylase alpha (ACC1), sterol regulatory element binding transcription factor 1 (SREBP-1), and genes involved in the beta-oxidation pathway include acyl-coenzyme A dehydrogenase, medium chain (MCAD), acyl-coenzyme A oxidase 1 (ACOX1), and carnitine palmitoyltransferase I (CPT-1 α). Bbs4 -/- had significantly upregulated liver expression of FASN, ACC1 fatty acid synthesis genes and upregulated CPT-1α in the beta-oxidation group of genes (Figure 7). Additionally, genes involved either directly in the pathway or genes that regulate gluconeogenesis (G6Pase, PEPCK, PGC1 α, and FBP1) were examined, but were not significantly different in Bbs4 - /- liver tissue (data not shown). Although energy levels are high in Bbs4 -/- mice livers in the form of lipids, genes involved in fatty acid synthesis are upregulated rather than the lower levels found in WT liver.

41 24 Glucose Homeostasis in Lean BBS Mice To examine whether the metabolic defects in BBS mice are secondary to obesity, Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice had their body weight normalized by calorie restriction. Using this protocol the body weight of Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice are comparable to wild type littermates (Figure 8). The fasting insulin levels of calorie restricted (CR) Bbs2 - /-, Bbs4 -/-, and Bbs6 -/- mice were significantly elevated, while CR-Bbs4 -/- mice have hypoglycemia (Figure 9). CR- Bbs2 -/- maintain normal glucose levels while CR-Bbs6 -/- mice remain hyperglycemic despite high insulin levels. These data indicate that the hyperinsulinemia associated with BBS is independent of obesity. When the ability of insulin to activate p-akt was tested in CR-Bbs4 -/- mice, insulin-induced p-akt activation in the skeletal muscle, liver and adipose tissue was blunted relative to controls (Figure 11). Discussion While formal investigation of the extent and types of diabetes mellitus has not been performed among BBS patients, these studies begin the process of identification of the diabetes phenotype in BBS null mouse models. In Bbs2 -/- and Bbs4 -/- mice in which BBSome proteins are effected, the levels of fasting insulin and glucose intolerance is comparable to each other, but significantly higher than WT littermates at four months of age. However, when month old Bbs4 -/- mice underwent GTT, they exhibited lower glucose levels than WT littermates (data not shown). Also when calorie restricted to maintain normal body weight, BBS null mice exhibited strikingly high plasma insulin levels. This is suggestive of a hyperinsulinemia most likely caused by hypersecretion of insulin. It would appear than that at young age and prior to the onset of obesity BBS null mice are experiencing hyperinsulinemia that progresses to insulin resistance over time. Whether this is true in BBS patients as well remains to be determined. The data from

42 25 Biesecker s group along with the case report from Daskalakis et al. would indicate that this is likely, since the mean age in Biesecker s group was 14.8 years old and Dsakalakis patient was 17 years old [11, 25]. The Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mouse models recapitulated the BBS-associated diabetes phenotype as evidenced by hyperglycemia and hyperinsulinemia (Figure 9). Furthermore, these mouse models exhibited glucose intolerance and insulin resistance (Figure 10 and 11). The defect in IR signaling as indicated by the blunted insulin-induced Akt activation further corroborates the insulin resistance phenotype (Figure 12 and 13). However, Marion et al. identified paradoxical insulin sensitivity in Bbs12 -/- mice [85]. Indeed, Bbs12 null mice were found to have enhanced adipogenesis, glucose tolerance and insulin sensitivity of adipose tissue. It is possible that this is due to the function of BBS12, which is part of the BBS chaperonin complex, but has another function in adipocyte differentiation outside of its typical role in mediating BBSome formation. The conflicting results may instead be due to an effect of age. This may be that in BBS hyperinsulinemia develops prior to insulin resistance. One of the commonalities between lean and obese Bbs-null mice was the extraordinarily high insulin levels. Calorie restricted Bbs-null mice had low, normal, and high glucose levels but they all had high insulin levels (Figure 9). This illustrates that hyperinsulinemia is present although insulin resistance may develop with age. This line of reasoning opens up some questions that warrant further investigation. Firstly, does the BBSome play any role in insulin secretion? Pancreatic β-cells, which produce and secrete insulin, are ciliated cells. The processing of pro-insulin to insulin yields equimolar amounts of C-peptide which can be used to measure insulin secretion and help to distinguish between hyperseceretion and insulin resistance. If the BBSome is involved in insulin secretion then is it related to its ciliary function or could this represent another example of an extra-ciliary role? Secondly, is the finding of hyperinsulin secretion relevant to BBS patients? The literature at this point is mixed in this regard.

43 26 There are reports of 32-48% prevalence of diabetes in years old BBS patients, while Marion et al. found no evidence of diabetes or hyperinsulinemia in 16 patients 27 years old [8, 24, 85]. A study of BBS patients should be undertaken to directly assess diabetes in BBS patients.

44 27 Blood Glucose (mg/dl) Fasting Glucose * * * WT Bbs2 Bbs4 Bbs6 a Plasma Insulin (ng/dl) Fasting Insulin * * * WT Bbs2 Bbs4 Bbs6 b Figure 2. Comparison of fasting blood glucose (a) and fasting plasma insulin (b) between Bbs mice and littermate WT controls. Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice have hyperglycemia (a) and hyperinsulinemia (b). Note: * p-value < 0.05 versus WT by One-Way ANOVA; n=10 WT, 11 Bbs2 -/-, Bbs4 -/-, 10 Bbs6 -/- for fasting glucose; n = 7 WT, 14 Bbs4 -/-, 6 Bbs2 -/- and Bbs6 -/- for fasting insulin

45 28 Blood Glucose (mg/dl) Glucose Tolerance Test ** ** ** * * ** WT Bbs2 -/- Bbs4 -/- Bbs6 -/ Time (min) a Delta Blood Glucose (mg/dl) Glucose Tolerance Test * WT Bbs2 -/- Bbs4 -/- Bbs6 -/ Time (min) b Figure 3. Glucose tolerance test (a) revealed significantly higher glucose levels at all-time points in Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice compared to WT controls. Calculating the change in blood glucose (b) shows glucose intolerance in Bbs2 -/- and Bbs4 - /- mice at 60 minutes while Bbs6 -/- mice had unimpaired glucose tolerance. Note: n= 16 WT, 11 Bbs2 -/- and Bbs4 -/-, 10 Bbs6 -/- mice. * p-value < 0.05; ** p-value < by Two-way ANOVA.

46 29 Blood Glucose (mg/dl) Insulin Tolerance Test ** * ** * Time (min) a WT Bbs2 -/- Bbs4 -/- Bbs6 -/- Delta Blood Glucose (mg/dl) Insulin Tolerance Test * * * * Time (min) b WT Bbs2 -/- Bbs4 -/- Bbs6 -/- Figure 4. Insulin tolerance test (a) and change in blood glucose (b) revealed a significant insulin resistance in Bbs4 -/- mice throughout the test and Bbs2 -/- mice after 120 minutes compared to WT. Bbs6 -/- mice tended to have insulin resistance, but the difference in blood glucose is not statistically different. Note: n= 16 WT and Bbs2 -/-, 20 Bbs4 -/-, 9 Bbs6 -/- mice. * p-value < 0.05; ** p-value < Two-way ANOVA

47 30 WT Bbs4 -/- WT Bbs4 -/- Sal Ins Sal Ins Sal Ins Sal Ins Liver Adipose Tissue pakt/akt (AU) * # Saline Insulin pakt/akt (AU) 400 * # 0 WT Bbs4 -/- 0 WT Bbs4 -/- a b WT Bbs4 -/- Sal Ins Sal Ins Skeletal Muscle pakt/akt (AU) * # WT Bbs4 -/- c Figure 5. Comparison of insulin-induced activation of Akt (phosphorylation) in liver, white adipose tissue, and skeletal muscle between Bbs4 -/- mice and littermate WT controls. After fasting, mice receive intervenous injection of vehicle or insulin. Bbs4 -/- mice have diminished insulin signaling in liver (a) adipose tissue (b), and skeletal muscle (c) compared to WT littermates. Note: Saline treatment in black bars and insulin treatment in dark gray bars. * p-value <0.05 vs WT saline, # p-value < 0.05 vs WT insulin.

48 Baseline p-akt WT Bbs4 -/- pakt/akt (AU) Liver Adipose Tissue Skeletal Muscle Figure 6. Comparison of baseline Akt activity in liver, white adipose tissue and skeletal muscle between Bbs4 -/- mice and littermate WT controls. Bbs4 -/- mice have increased basal p-akt expression in liver, adipose tissue, and skeletal muscle. Note: WT in black bars and Bbs4 -/- in dark gray bars.

49 32 mrna (normalized units) Liver Tissue Fatty Acid Synthesis 6 FASN ACC * * WT Bbs4 -/- SREBP-1A a mrna (normalized units) Liver Tissue - Oxidation 6 MCAD ACOX1 CPT WT Bbs4 -/- * b Figure 7. Comparison of expression levels of genes involved in the fatty acid synthesis (a) and β-oxidation (b) in liver between Bbs4 -/- mice and littermate WT controls. Fatty acid synthesis genes, FASN and ACC1 are upregulated (a) while CPT-1 is a beta oxidation gene that is upregulated (b) in Bbs4 -/-/ liver tissue. Note: Gene symbols are as follows: fatty acid synthase (FASN), aetyl-coenzyme A carboxylase alpha (ACC1), sterol regulatory element binding transcription factor 1 (SREBP-1), acyl-coenzyme A dehydrogenase, medium chain (MCAD), cyl- Coenzyme A oxidase 1 (ACOX1), while carnitine palmitoyltransferase I (CPT-1 α)

50 33 grams Body Weight * * WT Bbs2 Bbs4 Bbs6 a Body Weight grams WT CR-Bbs2 CR-Bbs4 CR-Bbs6 b Figure 8. Comparison of body weight between Bbs mice and littermate WT controls under normal feeding condition (a) or under calorie-restriction (b). Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice are obese (a) and Calorie-restricted (CR) Bbs2 -/-, Bbs4 -/-, Bbs6 -/- mice have a normal body weight (b). Note: n= 17 WT and 15 Bbs2 -/-, 20 Bbs4 -/-, 8 Bbs6 -/- mice. * p-value < 0.05; ** p-value < by One-way ANOVA

51 34 Fasting Glucose Blood Glucose (mg/dl) WT * * CR-Bbs2 CR-Bbs4 CR-Bbs6 a Fasting Insulin Plasma Insulin (ng/ml) WT * * * CR-Bbs2 CR-Bbs4 CR-Bbs6 b Figure 9. Comparison of fasting blood glucose (a) and fasting plasma insulin (b) between calorie-restricted Bbs mice and wild type controls. Hypoglycemia in CR-Bbs4 - /- and hyperglycemia in CR-Bbs6 -/- mice (a). Hyperinsulinemia in CR-Bbs2 -/-, CR-Bbs4 -/-, CR-Bbs6 -/- mice when compared to wild type littermates. Note: n= 17 WT, 14 Bbs2 -/-, 9 Bbs4 -/-, and 8 Bbs6 -/- mice. * p-value < 0.05; ** p-value < by One-way ANOVA

52 35 Blood Glucose (mg/dl) Glucose Tolerance Test * * a WT CR-Bbs2 -/- CR-Bbs4 -/- CR-Bbs6 -/ Time (min) Delta Blood Glucose (mg/dl) GlucoseTolerance Test * * WT CR-Bbs2 -/- CR-Bbs4 -/- CR-Bbs6 -/ Time (min) b Figure 10. Glucose tolerance test in calorie-restricted (CR) Bbs mice and WT controls. Calorie-restricted Bbs4 -/- mice are hypoglycemic compared to WT during glucose tolerance test (a) although the change in glucose levels is significantly increased in CR- Bbs2 -/- and CR- Bbs4 -/- mice (b). Note: n= 17 WT, 14 Bbs2 -/-, 9 Bbs4 -/-, and 8 Bbs6 -/- mice. * p-value < 0.05; ** p-value < by Two-way ANOVA

53 36 Blood Glucose (mg/dl) Delta Blood Glucose (mg/dl) Insulin Tolerance Test * * Time (min) Insulin Tolerance Test a Time (min) b WT CR-Bbs2 -/- CR-Bbs4 -/- CR-Bbs6 -/- WT CR-Bbs2 -/- CR-Bbs4 -/- CR-Bbs6 -/- * Figure 11. Insulin tolerance test in calorie-restricted Bbs mice and littermate WT controls. Calorie-restricted (CR) Bbs4 -/- mice are hypoglycemic compared to WT during insulin tolerance test (a) and have similar changes in glucose levels (b). Note: n= 15 WT, Bbs2 -/-, 20 Bbs4 -/-, and 8 Bbs6 -/- mice. * p-value < 0.05 by Two-way ANOVA

54 37 WT Bbs4 -/- WT Bbs4 -/- Sal Ins Sal Ins Sal Ins Sal Ins 400 Liver * Saline Insulin Adipose Tissue 400 * pakt/akt (AU) pakt/akt (AU) # 0 WT CR-Bbs4 -/- 0 WT CR-Bbs4 -/- a b WT Bbs4 -/- Sal Ins Sal Ins Skeletal Muscle pakt/akt (AU) * # 0 WT CR-Bbs4 -/- c Figure 12. Comparison of insulin receptor signaling in calorie-restricted Bbs4 -/- mice and littermate WT controls in liver (a), white adipose tissue (b), and skeletal muscle (c). Attenuation of insulin activation of p-akt in liver (a) in adipose tissue insulin activation of p-akt is diminished (b), and in skeletal muscle response to insulin activation is blunted (c) when compared to WT response. Note: Saline treatment in black bars and insulin treatment in dark gray bars. * p-value <0.05 vs WT saline, # p-value < 0.05 vs WT insulin.

55 38 Basal p-akt pakt/akt (AU) 800 WT * * * CR-Bbs4 -/- 0 Liver Adipose Tissue Skeletal Muscle b Figure 13. Comparison of basal p-akt levels in calorie-restricted Bbs4 -/- mice and littermate WT controls in liver, white adipose tissue, and skeletal muscle. In calorie-restricted Bbs4 -/- liver, adipose tissue, and skeletal muscle have increased basal p-akt levels compared to WT littermates (b). Note: Saline treatment in black bars and insulin treatment in dark gray bars. * p-value <0.05 vs WT saline, # p-value < 0.05 vs WT insulin.

56 39 CHAPTER III COMMON BBS1 M390R MUTATION RESULTS IN GLUCOSE INTOLERANCE AND NEURONAL BBS1 HAS A NOVEL ROLE IN GLUCOSE HOMEOSTASIS Introduction Bardet-Biedl Syndrome (BBS) is an autosomal recessive syndrome with clinical features that include primary clinical features of obesity, retinopathy, renal dysfunction, hypogonadism, and learning disabilities. Type 2-diabetes (T2DM) is also common in BBS although the mechanism is unknown. The prevalence of BBS in North America and Europe is one in 100,000 to one in 160,000 live births [12]. In isolated populations such as Newfoundland and among Arab Bedouins, the prevalence of BBS is as high as one in 17,500 and one in 13,500 respectively [19, 24]. Mutations found in BBS patients whether by positional cloning or exome sequencing have identified 19 genes associated with BBS. The BBSome contains eight BBS proteins (BBS1, 2, 4, 5, 7, 8, 9 and 18 also known as BBIP1). Assembly of the BBSome is mediated by another protein complex that contains the three BBS proteins (BBS6, 10, 12) in association with chaperonins of the CCT/TriC family. The BBS3 gene encodes the GTPase Arl6 and recruits the BBSome to membranes [52]. The BBSome is involved in maintenance of primary cilia [52, 66]. It has also been implicated in intracellular vesicle trafficking and localization of receptors to neuronal cilia including NP2Y, dopamine 1 receptor, somatostatin, and sonic hedgehog proteins Ptc-1 and Smo [71, 73, 86]. The most common BBS1 mutation is a homozygous substitution of arginine for methionine (M390R) [1].The BBS1 M390R/M390R mouse model (Bbs1 KI) was previously established and shown to develop obesity by 12 weeks of age with increased fat content,

57 40 increased food intake, and decreased locomotor activity [47]. BBS has been shown to have variable phenotypic expression. Phenotypic differences were also noted in the Bbs2 - /-, Bbs4 -/-, and Bbs6 -/- mouse models, including the cardiovascular phenotype and the extent of ventriculomegaly[53, 77]. Our hypothesis is that Bbs1 KI mice would exhibit glucose intolerance and insulin resistance in a similar manner as the Bbs2, Bbs4, and Bbs6 null mice. The first question addressed by this study was to determine whether Bbs1 KI mice also recapitulated the diabetes phenotype. To address this question, basal glucose measurements were taken along with a GTT and ITT. To examine insulin receptor signaling, mouse embryo fibroblast (MEF) cells were used to examine insulin signaling prior to the onset of obesity. Emerging evidence implicating neuronal mechanisms in various BBS phenotypes led us to the test the possibility that loss of Bbs1 in the central nervous system (CNS) accounts for the defect in glucose homeostasis [11, 72, 73, 81]. For this, we crossed Bbs1 fl/fl mice with mice expressing Cre recombinase in the CNS (Nestin Cre ) or in defined neuronal populations such as those expressing the leptin receptor (LepRb Cre ) or the precursor polypeptide proopiomelanocortin (Pomc Cre ). These experiments allowed us to address the requirement of Bbs1 gene in the whole CNS versus the hypothalamus to regulate glucose metabolism in the periphery. We found that CNS deletion of the Bbs1 gene leads to hyperglycemia, glucose intolerance and insulin resistance. Mice lacking Bbs1 in Pomc-expressing neurons exhibited glucose intolerance although their sensitivity to exogenous insulin is intact. When the Bbs1 gene is deleted in LepRb-expressing neurons, glucose tolerance and sensitivity to exogenous insulin is unaltered. Thus, neuronal loss of the Bbs1 gene results in impaired ability to respond to glucose loads in the periphery. Our data demonstrate the

58 41 critical role of neuronal Bbs1 in the regulation of glucose in an insulin-independent manner. Materials and Methods Animals Mice were bred on a mixed background containing C57B/6J and 129SvEv strains. The use of animals as described in this chapter was approved by the University of Iowa Animal Care and Use Committee. All animal testing was performed based on guidelines set forth by the National Institutes of Health. Mice were housed at the University of Iowa Animal Care facility in a temperature and humidity controlled room on a 12 hour light / dark cycle with free access to food and water. Generation of a conditional Bbs1 mouse was done using mice homozygous for the conditional Bbs1 allele (Bbs1flox/flox mice). The Bbs1fl/fl mice were generated with two flox P sites within exon 3 of the Bbs1 gene previously described [81]. Transgenic mice expressing Cre recombinase within Nestin, LepRb, or Pomc neurons in the brain were reported previously [87-89]. Upon Cre-mediated recombination the third coding exon of Bbs1 gene is deleted, resulting in a shifted reading frame downstream of the Bbs1 gene, and this event is predicted to prevent the expression of Bbs1. To obtain selective deletion in the whole brain, LepRb BBS1, or POMC BBS1 gene, transgenic Nestin Cre, ObR Cre or Pomc Cre male mice were crossed with Bbs1 fl/fl female mice. Nestin Cre /Bbs1 fl/wt, LepRb Cre /Bbs1 fl/wt or Pomc cre- /Bbs1 fl/wt offspring were subsequently crossed with Bbs1 fl/fl mice to generate Nestin Cre /Bbs1 fl/fl, LepRb Cre fl/fl or /Bbs1 Pomc Cre /Bbs1 fl/fl mice. Cre-positive/Bbs1 wt/wt and/or Cre-negative/Bbs1 fl/fl littermates were used as controls in the studies.

59 42 GTT and ITT GTT and ITT were performed as previously described. Four to five month old mice were fasted overnight for the GTT and 5 hours for the ITT in a clean cage. Body weights and basal blood glucose measurements were performed before intraperitoneal injection of glucose (2g/kg) or insulin (1 U/kg). Blood glucose measurements were assessed at multiple time points during the two-hour test. Insulin Receptor Signaling MEF cells were obtained from embryos between e9-e14 stages produced from timed matings between Bbs1 M390R/+ males and females. Cells were maintained in DMEM high glucose medium supplemented with 10% FBS and antibiotic (penicillin, streptomycin) MEF cells were serum-starved six hours before stimulation with 100 nm insulin for 15 minutes at 37 C. After stimulation, cells were washed and immediately lysed in lysis buffer (50 mm HEPES ph7.5, 137 mm NaCl, 1% NP-40, 0.25% Nadeoxycholate, 2 mm EDTA, 2 mm Na 3 VO 4, 10 mm NaF, 10% glycerol, protease inhibitor cocktail [Roche complete Mini, EDTA-Free ]) for 30 minutes at 4 C. The cell lysate was centrifuged at 14,000 g for 30 minutes at 4 C. Whole cell lysates were used for detection by immunoblot. Immunoblotting Whole cell lysate protein (10-12 µg) was resolved by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked in 5% nonfat dry milk in TBST for 1 hour at room temperature and incubated with anti-ir (Santa Cruz) or clone H68.4 antitransferrin receptor (TfR) (Invitrogen) antibodies or anti-akt and anti-p-akt antibodies (Cell Signaling) at 4 C overnight. Membranes were incubated with HRP-conjugated

60 43 secondary antibody at room temperature followed by ECL and autoradiograph film exposure. Results Glucose Homeostasis in Bbs1 KI Bbs1 KI mice were significantly obese with an average body weight of 43.8 g after overnight fast and 47.7 grams after five hours fast compared to WT littermates which weighed on average of 23.7 g and 26.8 after overnight or five hours fast (Figure 14a). Whether fasted for 5 hours or 18 hours, Bbs1 KI mice were hyperglycemic with glucose levels mg/dl higher than WT littermates (Figure 14b). Next, Bbs1 KI mice (n=6) and WT (n=5) were examined for their ability to respond to glucose loads. The Bbs1 KI mice were found to have significantly higher glucose levels during the first thirty minutes of the GTT (Figure 15a). However the change in glucose levels, displayed as delta blood glucose levels, is similar during the first 30 minutes of the GTT (Figure 15b). The last time point is different between the Bbs1 KI and WT curves. The WT mice are responding by lowering their glucose levels to just about starting levels, while the Bbs1 KI mice are not able to continue to lower glucose levels (Figure 15b). The ability to respond to exogenous insulin was tested using an ITT. WT mice were not able to complete the ITT beyond thirty minutes as they were extremely sensitive to insulin. Bbs1 mice were able to complete the ITT and were found to have a much higher high blood glucose levels at each time point (Figure 16). Defective IR Signaling in Bbs1KI Since Bbs1 KI mice were obese, MEFs were used for insulin receptor signaling assay. When WT MEFs were treated with 100 nm insulin for 15 minutes, there was a

61 44 significant response as indicated by the doubling of the levels of p-akt (Figure 17a). In contrast, insulin did not increase p-akt levels in Bbs1 KI MEFs (Figures 17a). In fact, a comparison of insulin stimulated p-akt levels indicated a significant reduction of insulin p-akt activation in Bbs1 KI MEFs compared to WT (Figure 17a). It was clear that the basal p-akt levels were much higher in Bbs1 KI MEFs (Figure 17b). Neuronal Deletion of Bbs1 Alters Glucose Homeostasis In order to begin assessing the contribution of various tissues/cell types to the BBS-associated defects in glucose metabolism, we evaluated the consequence of deleting the Bbs1 gene (using the conditional Bbs1 fl/fl mouse model) throughout the central nervous system (using Nestin-Cre) or in defined neuronal populations such as those expressing the leptin receptor (LepRb Cre ) or the proopiomelanocortin (POMC Cre ) (Figure 18). NestinCre/Bbs1 fl/fl (28.3 g ± 2.0 SEM) and LRbCre/Bbs1fl /fl (32.3 ± 2.5 SEM) mice had body weights similar to WT (30.0 g ±1.2 SEM) (Figure 19a). However, the Pomc Cre /Bbs1fl /fl mice were significantly heavier (42 ±2.6 g) than WT controls. Measurements were taken after overnight fasting at 4-5 months of age. Blood glucose measures were taken simultaneously. Both NestinCre/Bbs1fl /fl and Pomc Cre /Bbs1 fl/fl mice were hyperglycemic at about 107 mg/dl and 114 mg/dl, respectively. LRb Cre /Bbs1 fl/fl mice had blood glucose levels of about 99.7 ±7.8 mg/dl, which was not significantly different relative to WT levels of 82.6 ±3.5 mg/dl (Figure 19b). Thus, the Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice demonstrate hyperglycemia that does not appear to have a relationship with obesity. Glucose metabolism was further assessed by GTT to investigate the ability to clear a glucose load as in the studies above. In the neuronal conditional knock out models, the role of a functional BBS1 protein in the CNS, LRb expressing neurons or Pomc expressing neurons was tested for its requirement for glucose clearance in the

62 45 periphery. Consistent with the hyperglycemic state, Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice maintained significantly higher glucose levels throughout the first hour of the GTT, while LRb Cre /Bbs1 fl/fl mice maintained blood glucose levels similar to WT levels (Figure 20a). When the change in glucose levels was examined, Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice demonstrated further the significant difference response to glucose at 30 minutes (Figure 20b). Insulin sensitivity was assessed next using the ITT to test the response of blood glucose levels in the presence of exogenous insulin. Just as the Nestin Cre /Bbs1 fl/fl mice had difficulty clearing a glucose load, they exhibited impaired response to increased insulin levels illustrated on both the ITT curve and the change in glucose levels during the ITT (Figure 21). Insulin Receptor Signaling in Nestin Cre /Bbs1 fl/fl Mice When WT mice were treated with 1U/kg insulin, there was a significant response in p-akt levels in liver, adipose tissue, and skeletal muscle (Figure 22 and 23). In liver tissue of Nestin Cre /Bbs1 fl/fl mice, the response to insulin did not significantly increase p- Akt levels above baseline, but the response of the Nestin Cre /Bbs1 fl/fl mice was significantly different than the WT response (Figures 22a). A comparison of insulin stimulated p-akt levels indicated a significant reduction of insulin activation in both liver and adipose tissue in Nestin Cre /Bbs1 fl/fl mice compared to WT controls (Figure 22). Insulin receptor signaling was not impaired in skeletal muscle (Figure 23). Although Bbs1 was expressed normally in all three tissues tested, the response was diminished in liver and tended to be attenuated in adipose tissue of Nestin Cre /Bbs1 fl/fl mice.

63 46 Discussion The Bbs1 KI mouse model appears to exhibit an insulin resistance phenotype similar to the null mouse models Bbs2 -/-, Bbs4 -/-, and Bbs6 -/-. Additionally, the innate cellular response to insulin is diminished in the MEF cells, which were not exposed to either obesity or calorie-restriction. A fully functional BBS1 protein that is expressed systemically and centrally is required to maintain glucose tolerance and insulin sensitivity. First the common homozygous substitution M390R mutation in the Bbs1 KI mouse model recapitulates the diabetes phenotype. The Bbs1 KI mouse model is hyperglycemic (Figure 14b and 15) and insulin resistant according to ITT (Figure 16) and insulin receptor signaling (Figure 17). This diabetes phenotype agrees with the Bbs-null mouse models discussed in Chapter II and provides further support to the need for diabetes studies in BBS patients. The second part of the hypothesis tested was that the Bbs1 gene in the CNS and the hypothalamus plays a role in maintaining glucose homeostasis. The role of the brain regulating glucose metabolism has been discussed in several reviews [90, 91]. The results presented here add to the growing body of work regarding the importance of the brain in peripheral metabolism. We show that Bbs1 expression in the CNS is required to maintain euglycemia and normal glucose tolerance along with insulin sensitivity. Loss of Bbs1 in the CNS also has impaired insulin receptor signaling in the liver (Figure 22a). The Pomc Cre /Bbs1 fl/fl model further narrows down the scope from whole brain to the hypothalamus, the brain region expressing Pomc. Loss of Bbs1 expression in the hypothalamus results in BBS-associated diabetes. Additionally the Lep Cre Bbs1 fl/fl model with in intermediate population of affected neurons informs us that not all neurons are required to express Bbs1 in order to maintain glucose homeostasis. The studies presented in this work illustrate that loss of Bbs1 expression in the CNS or hypothalamus leads to hyperglycemia, glucose intolerance and insulin resistance. Fasting insulin levels would be helpful in establishing any effect on insulin secretion

64 47 under these conditions. For example would the conditional knockout mice have such dramatically high insulin levels similar to the Bbs-null mice? This would be helpful in informing whether there is a brain-β-cell connection that relies on a functional BBSome. The insulin receptor signaling assay in the liver of Nestin Cre Bbs1 fl/fl mice indicates that there is a brain-liver connection that relies on the BBSome functioning in the CNS. The brain-liver connection was previously inferred by Kevin Williams and Joel Elmquist in their review article [91]. The question of how the brain regulates the liver to maintain euglycemia is one that requires further investigation. The Pomc Cre /Bbs1 fl/fl model would be helpful in addressing whether the regulation relies on autonomic nervous system or is a hormonal regulation. The brain-liver connection via the autonomic nervous system has been established [92, 93]. This chapter shows, Bbs1 in the CNS and hypothalamus as illustrated by the Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice is required to maintain normal glucose levels throughout the whole body.

65 48 grams Body Weight 60 WT Bbs1 KI ** ** 0 5 hr fast 18 hr fast a Blood Glucose (mg/dl) Fasting Glucose * * 5 hr fast 18 hr fast b Figure 14. Comparison of fasting body weight and glucose levels in Bbs1 KI and WT littermate controls. The Bbs1 KI mice develop obesity that is not resolved by fasting (a) and hyperglycemia compared to WT littermates (b). Note: Data shown are mean ± SEM; ** p-value <0.001compared to WT by One way ANOVA; n=5 WT; 6 BBS1 KI

66 49 Blood Glucose (mg/dl) * Glucose Tolerance Test * * Time (min) WT Bbs1 KI a Delta Blood Glucose (mg/dl) Glucose Tolerance Test 160 WT Time (min) b Bbs1 KI * Figure 15. Glucose tolerance test with glucose levels (a) and the change in glucose levels (b) in Bbs1 KI and WT littermate controls. The Bbs1 KI mice develop hyperglycemia with significantly higher glucose levels at time 0, 30, and 60 minutes after glucose load (a). The Bbs1 KI mice have a similar change in glucose levels as the WT littermate controls during the first hour of the GTT but have a significantly elevated response to the glucose load after 120 minutes (b). Note: Data shown are mean ± SEM; * p-value < 0.05 compared to WT by Two Way ANOVA; n=5 WT; 6 Bbs1 KI mice

67 50 Blood glucose (mg/dl) Insulin Tolerance Test 250 WT 200 Bbs1 KI Time (min) Figure 16. Insulin tolerance test with glucose levels (a) and the change in glucose levels (b) in Bbs1 KI and WT littermate controls. Bbs1 KI mice have significantly higher glucose levels throughout the ITT while WT littermates have quite low glucose levels requiring rescue by intraperitoneal injection of glucose prior to completing the ITT. Note: Data shown are mean ± SEM; n=5 WT; 6 Bbs1 KI mice

68 51 Control Bbs1 KI 100 nm Insulin p-akt Akt pakt/akt (AU) MEF Cells * # WT Bbs1 KI a Baseline p-akt p-akt/akt (AU) b Figure 17. Comparison of insulin receptor signaling using downstream p-akt (a) as marker of IR signaling and basal p-akt (b) in Bbs1 KI mouse embryo fibroblasts (MEF) and WT MEF littermate controls. After a 6 hour period of serum starvation, cells were treated with 100 nm insulin for 15 minutes. Bbs1 KI MEF cells have a diminished response to IR activation of p-akt (a) and have significantly higher basal p-akt expression (b) compared to WT MEF cells. Note: Baseline treatment in black bars and insulin treatment in white bars. * p-value <0.05 vs WT baseline, # p-value < 0.05 vs WT insulin by One-way ANOVA.

69 Figure 18. Comparison of Cre expression patterns in Nestin, Leptin Receptor b-form, and Proopiomelanocortin. Reporter mouse lines with Cre recombinase driven by Nestin, LRb and Pomc promoters express td tomato, a red fluorescent protein, to illustrate the expression patter of the Cre in each mouse line. 52

70 53 grams Body Weight ** WT Nestin LRb Pomc a Blood Glucose (mg/dl) Fasting Glucose * ** WT Nestin LRb Pomc b Figure 19. Comparison of body weight and fasting glucose levels in Nestin Cre /Bbs1 fl/fl, LRb Cre /Bbs1 fl/fl, and Pomc Cre /Bbs1 fl/fl mice and littermate WT controls. Pomc Cre /Bbs1 fl/fl mice are obese (a) although Nestin Cre /Bbs1 fl/fl and LRb Cre /Bbs1 fl/fl mice have normal body weights after fasting compared to WT littermates. Both Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice are hyperglycemic and LRb Cre /Bbs1 fl/fl mice are euglycemic (b). Note: Data shown are mean ± SEM; *p-value < 0.05 and ** p-value <0.001 versus WT by one-way ANOVA; n=22 WT, 10 Nestin, Pomc and 8 LepRb conditional knockout mice.

71 54 Blood Glucose (mg/dl) Glucose Tolerance Test * ** * Time (min) WT Nestin Cre /Bbs1 f l/f l LRb Cre /Bbs1 f l/f l Pomc Cre /Bbs1 f l/f l a Delta Blood Glucose (mg/dl) Glucose Tolerance Test * Time (min) b WT Nestin Cre /Bbs1 f l/f l LRb Cre /Bbs1 f l/f l Pomc Cre /Bbs1 f l/f l Figure 20. Glucose intolerance test with glucose levels (a) and change in glucose levels (b) in Nestin Cre /Bbs1 fl/fl, LRb Cre /Bbs1 fl/fl, and Pomc Cre /Bbs1 fl/fl mice and littermate WT controls. In Nestin Cre /Bbs1 fl/fl and Pomc Cre /Bbs1 fl/fl mice glucose levels in LRb neurons lack of Bbs1 does not impact glucose tolerance as shown by blood glucose over time (a) and when change in glucose levels are calculated there is a significant difference in the change in glucose levels at 30 minutes (b). Note: Data shown are mean ± SEM; ** p-value <0.001 and *p-value < 0.05 versus WT by two-way ANOVA, n=31 WT, 13 Nestin, 12 Pomc and 8 LepRb conditional knock-out mice.

72 55 Blood Glucose (mg/dl) Insulin Tolerance Test ** ** * Time (min) WT Nestin Cre /Bbs1 f l/f l LRb Cre /Bbs1 f l/f l Pomc Cre /Bbs1 f l/f l * a Delta Blood Glucose (mg/dl) Insulin Tolerance Test ** * * Time (min) b * WT Nestin Cre /Bbs1 f l/f l LRb Cre /Bbs1 f l/f l Pomc Cre /Bbs1 f l/f l Figure 21. Insulin tolerance test with glucose levels (a) and change in glucose levels (b) in Nestin Cre /Bbs1 fl/fl, LRb Cre /Bbs1 fl/fl, and Pomc Cre /Bbs1 fl/fl mice and littermate WT controls. Insulin resistance in Nestin Cre /Bbs1 fl/fl mice during the ITT (a) and when comparing change in glucose levels the Nestin Cre /Bbs1 fl/fl mice are significantly resistant compared to WT (b), while Pomc Cre /Bbs1 fl/fl and LRb Cre /Bbs1 fl/fl mice are sensitive to insulin in a similar manner to WT (b). Note: ITT ** p-value <0.001 and *p-value < 0.05 versus WT by two-way ANOVA, AUC by one-way ANOVA; n=31 WT, 16 Nestin, 12 Pomc and 7 LepRb conditional knock-out mice.

73 56 WT Bbs4 -/- WT Bbs4 -/- Sal Ins Sal Ins Sal Ins Sal Ins pakt/akt (AU) Liver 250 Saline Insulin * WT Nestin cre Bbs1 f l/fl # pakt/akt (AU) Adipose Tissue ** WT Nestin cre Bbs1 f l/fl ** # a b WT Bbs4 -/- Sal Ins Sal Ins pakt/akt (AU) Skeletal Muscle ** WT Nestin cre Bbs1 f l/fl c ** Figure 22. Comparison of insulin receptor signaling in Nestin Cre /Bbs1fl /fl mice and littermate WT controls in liver (a), white adipose tissue (b), and skeletal muscle (c). Insulin receptor signaling is significantly impaired in the liver (a) of Nestin Cre /Bbs1fl /fl mice and is attenuated in the adipose tissue (b). Insulin receptor signaling is unimpaired in the skeletal muscle f Nestin Cre /Bbs1fl /fl mice(c). Note: Saline treatment in black bars and insulin treatment in blue bars. * p-value <0.05 vs WT saline, ** p-value <0.001 vs WT saline, # p-value < 0.05 vs WT insulin by twoway ANOVA.

74 57 CHAPTER IV BBS PROTEINS ARE REQUIRED FOR INSULIN RECEPTOR TRAFFICKING TO THE CELL MEMBRANE Introduction Bardet-Biedl Syndrome (BBS) is a heterogenic, autosomal recessive syndrome. The presentation of BBS is variable both within families and among unrelated individuals who bear the same genetic defect [6, 36]. The clinical manifestations of BBS include obesity, retinal degeneration, renal dysfunction, hypogonadism, and learning disabilities [3, 24]. Diabetes mellitus is also commonly associated with BBS [5, 10, 12, 25]. Consistent with this, chapter II demonstrates that several mouse models of BBS including Bbs2 -/-, Bbs4 -/-, and Bbs6 -/- mice are hyperglycemic, hyperinsulinemic, glucose intolerant and insulin resistant. The mouse model for the common BBS1 mutation M390R, as shown in chapter III, also displays glucose intolerance and insulin resistance. However, the molecular mechanism of diabetes and insulin resistance in BBS remains unknown. The BBSome consisting of eight BBS proteins (BBS1, 2, 4, 5, 7, 8, 9 and 18) forms a stable complex and has an emerging role in trafficking. BBS proteins are involved in trafficking vesicles along microtubules in a retrograde manner and trafficking specific receptors to the cilia [5, 62, 65, 66, 72, 78]. Post-prandial glucose loads stimulate the release of the pancreatic peptide hormone insulin. When insulin binds to the IR, signal transduction is initiated to regulate glucose uptake in skeletal muscle and adipose tissue and energy storage in liver and adipose tissues. The insulin signal is regulated by the disposal of insulin and removal of the IR from the surface of the cell by internalization. Much attention has been given to the recycling process through the endocytosis of the IR and regulation of insulin receptor signal transduction [94-99]. There is limited information currently available detailing the progression of the IR from its translation to its translocation to the plasma membrane.

75 58 Several proteins have been implicated in the ability of the IR to proceed to the plasma membrane. For example, a deficiency of a muscular dystrophy kinase (DMPK) has been shown to decrease the amount of IR that reaches the cell surface [100]. Recently prions have been shown to disturb IR trafficking resulting in IR accumulation in the Golgi apparatus [101]. In this chapter we tested the hypothesis that BBS proteins and BBSome proteins specifically mediates IR trafficking to the cell surface and consequently loss of BBS proteins leads to IR mistrafficking. Here, we show that the IR is found in the same sucrose gradient fraction as the BBS proteins and their partners in the sonic hedgehog pathway suggesting a physical interaction between the IR and BBS proteins. Despite the role of BBS proteins in ciliary maintenance and trafficking, the IR does not localize to the cilia. When the level of the IR at the plasma membrane was compared with the total IR protein, there was a reduction in surface IR whenever a BBSome protein was deficient. The loss of BBSome proteins leads to a specific reduction in the amount of IR at the cell surface without any decrease in the total amount of IR present in the cell. The results demonstrate that BBSome proteins are required to maintain adequate levels of IR at the cell surface. By understanding the mechanism of insulin resistance in BBS patients, we hope to translate this information into mechanisms of insulin resistance in type 2 diabetes patients in the general population. The role of BBS proteins in transporting IR has not been previously described. Loss of the BBSome appears to be a novel mechanism of insulin resistance.

76 59 Materials and Methods Cell Culture and Transfection HEK293T cells were cultured in regular growth medium; DMEM supplemented with 5% (v/v) fetal bovine serum, and 1% (v/v) sodium pyruvate, at 37 C with 5% CO 2. Two µg of plasmid DNA encoding shrna against Bbs1, Bbs2, or control on an LKO1 plasmid was transfected into 70-80% confluent HEK293T cells in 60 mm dish with FuGENE 9 (Roche) according to manufacturer s instruction. After 48 hour incubation with the DNA/FuGENE mixture, cell surface proteins were labeled. Immunohistochemistry 3T3L1 fibroblast cells were seeded on cover slips and serum-starved for 48 hours to induce cilia formation. Cells were then washed with PBS and fixed with 4% PFA for 20 minutes at room temperature. After washing fixed cells three times with PBS, cells were blocked in blocking solution (10% goat serum, 5% milk, 0.1% Triton X-100 in PBS) for one hour at room temperature. Next anti-ir β (Santa Cruz) and anti-ac-α Tubulin (Santa Cruz) antibodies were diluted in blocking solution at 1:250. After blocking, cells were incubated with primary antibody for two hours at room temperature followed by three washes in PBS for five minutes each wash. Cells were then incubated with secondary immunofluorescent antibodies, goat-anti-rabbit Alexa488 and goat-antimouse Alexa568 (Invitrogen) for one hour at room temperature followed by washing as before. Finally cover slips were mounted using VectaShield mounting medium with DAPI. Images were visualized using confocal microscopy (Zeiss 710) and analyzed using ImageJ software.

77 60 Biotin Labeling of Cell Surface Proteins Biotin labeling was performed as previously described [102]. Cells were placed on ice and washed with phosphate buffered saline (PBS). EZ-link Sulfo-NHS-SS- Biotin (Thermo Scientific) was used to prepare biotinylation solution immediately before incubating cells for 30 minutes at 4 C. Cells were washed three times on ice and immediately lysed in lysis buffer (50 mm HEPES ph7.5, 137 mm NaCl, 1% NP-40, 0.25% Na-deoxycholate, 2 mm EDTA, 2 mm Na 3 VO 4, 10 mm NaF, 10% glycerol, protease inhibitor cocktail [Roche complete Mini, EDTA-Free ]) for 30 minutes at 4 C. The cell lysate was centrifuged at 14,000 g for 30 minutes at 4 C. Whole cell lysates were used for detection by immunoblot and for immunoprecipitation with Pierce Streptavidin UltraLink Resin (Thermo Scientific) overnight at 4 C to form Avidin-Biotin complex (ABC). The ABC resin was washed in PBS three times, and precipitated proteins were analyzed by immunoblot. In vivo biotinylation through intra-cardiac perfusion was also performed as previously described [102]. Briefly mice were anesthetized and cardiac perfusion was begun with 5% Dextran-PBS, followed by a biotin solution and a quenching solution of 50 mm Tris-HCl in 5% Dextran-PBS. Tissues were harvested and homogenized in lysis buffer. Protein concentration was measured by Bradford protein assay and used for total protein and streptavidin pull-down. Statistical Analysis All data are expressed as mean ± standard error. A one-way ANOVA was used to compare more than two groups, and a Student s t-test was used to compare two groups. Rank -ANOVA was used whenever the data did not follow a normal distribution. A twotailed p-value < 0.05 was considered significant for all analyses. Calculations for the corrected total cellular fluorescence (CTCF) was previously described [103]. Briefly, CTCF is calculated using ImageJ to measure the area and the integrated density of the

78 61 signal before the following formula can be applied: CTCF = Integrated Density (Area of selection x Mean fluorescence of background). CTCF was normalized by cell number, which used the DAPI stain as a marker for individual cells. Results Physical Interaction between IR and BBS Proteins One approach used to investigate the proteins involved in large complexes relies on sucrose gradient sedimentation analysis. The IR was identified in the same fractions as BBS proteins. Included in these fractions are the proteins Smoothened (Smo) and patched (Ptch), which are part of the sonic hedgehog (SHH) pathway and localize to cilia (Figure 24). Importantly, Ptch1 and Smo have previously been shown to interact with BBS proteins [86]. Seo et al. identified LZTFL-1, now BBS17, and its ability to interact and regulate the BBSome. A reciprocal immunoprecipitation assay was performed to investigate whether BBS17 would interact with IR as well. HEK293T cells were transiently transfected with a DNA plasmid which encoded a Flag-tagged BBS17 protein. A physical interaction between IR and BBS17 was identified whether using endogenous IR or Flag-BBS17 for pull down (Figure 25). These data show an interaction between BBS proteins and IR. IR is Not Targeted to Cilia The emerging evidence of the importance for BBSome in transporting receptors and SHH pathway proteins to the cilia, raised the question whether the IR is transported to cilia [71, 86]. To address this question, 3T3L1 cells were used because they express the IR and are able to form cilia in cell culture [ ]. Cells were induced to form cilia and double immunofluorescently labeled. To mark the cilia, acetylated-α tubulin that

79 62 had been labeled with a red fluorescent secondary antibody was used and the IR antibody was labeled with a green fluorescent secondary antibody. Therefore, if the IR was to be found in the cilia one would expect the co-localization would yield a yellow signal. However upon examination by confocal microscopy, there was a clear signal from the labeled IR and from the cilia without any overlap of signals (Figure 26). The images indicate that the IR is not trafficked to the cilia, but remains within the plasma membrane. Knockdown of BBS1 and BBS2 Reduces the Amount of IR at the Plasma Membrane Since the genetic defect in BBS proteins causes insulin resistance as shown in preceding chapters and the IR receptor does not localize to cilia, the next question was whether BBS proteins are needed for transport of IR to the plasma membrane. To address this question, we first used HEK293T cells, which are easily transfected with plasmid DNA. Cells were transiently transfected with shrna either as empty vector plasmid (control) or against Bbs1 or Bbs2 genes. To confirm the efficiency and the specificity of knockdown, a double transfection of HA-tagged BBS1 or HA-tagged BBS2 along with the appropriate shrna was performed. This allowed the possibility of using immunoblot for detection in the absence of specific BBS antibodies and revealed efficient and specific knockdown (Figure 27). Next, BBS1 or BBS2 was knocked-down in HEK293T cells and extracellular proteins were labeled with biotin and precipitated with streptavidin resin. Proteins were separated by size using electrophoresis in order to immunoblot surface and total proteins. BBS1 knockdown reduced the amount of IR, which was normalized to total IR, at the cell s surface to about 14% relative to 100% control surface IR while BBS2 knockdown reduced surface IR to about 37% relative to control (P<0.001, Figure 28a). To confirm that this effect was specific, the level of transferrin receptor (TfR) at the cell surface was also examined. Surface levels of TfR where unchanged when BBS1or

80 63 BBS2 was knocked-down. The amount of TfR at the cell surface was about 90% after BBS1 or BBS2 knockdown relative to 100% control (P=0.658, Figure 28b). Next, we examined the effect of the loss of BBS proteins on surface IR level in mouse embryo fibroblasts (MEF) cells derived from mice that are homozygous for the hypomorphic Bbs1 M390R mutation (Bbs1 KI) compared to WT MEF cells. Using the MEF cells allowed us to make comparisons without having to manipulate the cells with shrna. In addition MEF cells allowed us to examine the IR levels prior to obesity or calorie restriction. Bbs1KI cells had reduced levels of surface IR that were about 44% of 100% WT (P=0.0258, Figure 29a). Just as in the knock-down experiment, the levels of TfR were not different between the two groups. Surface levels of TfR were about90% in Bbs1 KI and 100% in WT MEF cells (P=0.740, Figure 29b). Finally, we tested whether the defect in IR trafficking occurs in vivo by comparing tissues of Bbs1 KI and WT mice. In the brain tissue of Bbs1 KI mice the surface level of IR was reduced to about 26% compared to 100% of WT (P<0.001, Figure 31a). Surface TfR levels were reduced to about 78% (± SEM 9.0) compared to 100% (± SEM 3.9) in WT, but this difference failed to reach significance (P=0.0613, Figure 8). Discussion It is imperative that the IR reside in the plasma membrane in order to maintain glucose homeostasis. The deficiency of BBS proteins results in a state of insulin resistance that is not wholly attributable to the obesity of BBS. There are some examples in the literature that indicates deficiency of specific proteins will impede transport of the IR to the surface. For example, insulin resistance is also a common feature of myotonic dystrophy. A study by Llagostera et al. demonstrated the importance of myotonic dystrophy protein kinase (DMPK) for insulin receptor targeting to the plasma membrane in muscle cells expressing DMPK [100]. More

81 64 recently, IR is shown to accumulate in the Golgi apparatus of neurons expressing PrP Sc rather than residing appropriately in the plasma membrane [101]. Our studies did not determine the cellular compartment in which the IR resides when the BBSome is impaired. It is of interest to determine the path of the IR under conditions of the BBSome defect. Another pertinent question which remains unaddressed relates to the specificity of the interaction between the IR and the BBSome. While we were able to identify a physical interaction between the IR and BBS17, the physiological relevance is not established or completely understood. The interactions between the BBSome and the IR may be of such a transient nature that it was not detectable by coimmunoprecipitation. If this is the case, then more sophisticated methods such as those involving protein cross-linking may help to identify the exact mechanisms by which the interaction between the IR and the BBSome occurs. Alternatively, it may be that there is no direct interaction between the IR and the BBSome, but an overall disruption in transport. Although the ability of the TfR to maintain its presence at the cell surface would indicate that an overall disruption in cell sorting is not the case. It is clear that the ability of the cell to maintain adequate levels of IR at its surface requires a fully functional BBSome.

82 65 Figure 23. Sucrose gradient sedimentation analysis of BBS proteins and the insulin receptor. The BBSome is found in fractions 4-9 of the sucrose gradient along with its cargo proteins Ptc1, Smo, and the IR. Source: Courtesy of Qihong Zhang, unpublished data

83 Figure 24. Shown are reciprocal co-immunoprecipitation in HEK293T cells with Flag- BBS17 and IR. The physical interaction between IR and BBS17 is seen in a transient transfection of HEK293T cells. This interaction can be seen whether IR antibody is used for immunoprecipitation (upper panel) or BBS17 tag is used to precipitate IR (lower panel). 66

84 67 MERGE Acetylated α Tubulin IR β Figure 25. Immunohistochemistry of 3T3L1 cells to mark cilia (Acetylated α-tubulin) and the Insulin receptor (β-subunit). The Insulin receptor does not localize to the cilia of 3T3L1 fibroblasts. The signal for the cilum marker is distinct from the signal for the insulin receptor. Note: Cells were serum-starved for 48 hours to stop proliferation and induce cilium formation. Antibody against acetylated-α Tubulin with Alexa Fluor 568 secondary antibody was used to label cilium and antibody against IR β-subunit with Alexa Fluor 488 secondary antibody was used to label the IR. The nuclei were stained with DAPI.

85 68 Control shrna BBS1 shrna HA (3F10) a GAPDH Control shrna BBS2 shrna HA (3F10) GAPDH b Figure 26. Confirmation of shrna mediated knockdown of BBS1 (a) and BBS2 (b). Efficient knockdown of Bbs1 (a) and Bbs2 (b) genes in HEK293T cells using HA-tagged proteins for immunoblot recognition in the absence of antibodies against BBS1 and BBS2.

86 69 Surface IR Ctrl BBS1 Ctrl BBS2 Total IR Surface / Total (AU) Insulin Receptor Ctrl * * Bbs1 Bbs2 a Ctrl BBS1 Ctrl BBS2 Surface TfR Total TfR Surface / Total (AU) Transferrin Receptor Ctrl Bbs1 Bbs2 b Figure 27. Comparison of surface insulin receptor (a) and transferrin receptor (b) levels in HEK293T cells. Knockdown of Bbs1 or Bbs2 reduces the amount of IR (a) at the cell surface but does not change the amount of TfR (b) at the cell surface. Note: All surface proteins were isolated by incubating undisturbed cellular monolayers with EZ-Link Sulfo-NHS-SS-Biotin. Protein lysates were then immunoprecipitated with Pierce Streptavidin UltraLink Resin.

87 70 Surface IR Total IR WT Bbs1 KI Insulin Receptor Surface / Total (AU) Control * Bbs1 KI a Surface TfR WT Bbs1 KI Total TfR Transferrin Receptor Surface / Total (AU) Control Bbs1 KI b Figure 28. Comparison of surface insulin receptor (a) and transferrin receptor (b) levels in Bbs1 KI and WT MEF cells. The Bbs1 KI MEFs have a significant reduction in the level of IR (a) at the cell surface while TfR (b) levels are unchanged. Note: WT and Bbs1 KI MEF cells were isolated from somites of e9-e14 embryos and cultured in DMEM (Dulbecco's Modified Eagle Medium) with 10% serum and antibiotics. Surface proteins were isolated as described in Figure 28.

88 71 WT MEF Bbs1 KI MEF IRα DAPI Merge CTCF / Cell (AU) a Surface IR Intensity WT Bbs1 KI b * Figure 29. Immunohistochemistry using indirect immunofluorescence of the insulin receptor without permeablizing the cell membranes of Bbs1 KI and WT MEF cells. The Bbs1 KI MEFs have decreased levels of IR immunofluorescence that can be visualized (a) and quantified (b). Note: Surface IR intensity was calculated using ImageJ to measure the area and the integrated density of the signal. CTCF = Integrated Density (Area of selection x Mean fluorescence of background). CTCF was normalized by cell number, use the DAPI stain as a marker for individual cells.

89 72 Brain WT Bbs1 KI Insulin Receptor Surface / Total (AU) 150 WT Bbs1 KI ** a Surface / Total (AU) Transferrin Receptor 50 0 b WT Bbs1 KI Figure 30. Comparison of surface insulin receptor (a) and transferrin receptor (b) levels in Bbs1 KI and WT brain tissue. In vivo biotinylation of Bbs1 KI brain tissue identifies a significant reduction of amount of IR (a) and a slight reduction in TfR (b) that is exposed to the circulation. Note: Data shown are mean ±SEM; ** p-value < 0.001; n=3wt, 3 Bbs1 KI mice