DIFFERENTIALLY EXPRESSED IN AND THEIR ROLE IN THE PATHOPHYSIOLOGY OF THE HUMAN

Differenziell exprimierte im Fettgewebe und ihre Rolle in der Pathophysiologie des humanen Metabolischen Syndroms

Von der Fakultät für Biowissenschaften, Pharmazie und Psychologie

der Universität Leipzig

genehmigte

D I S S E R T A T I O N

zur Erlangung des akademischen Grades

Doctor rerum naturalium

Dr. rer. nat.

vorgelegt von

M.Sc. Dorit Schleinitz

geboren am 21.03.1981 in Karl-Marx-Stadt

Dekan: Prof. Dr. Matthias Müller

Gutachter: Prof. Dr. Annette Beck-Sickinger Prof. Dr. Michael Stumvoll

Tag der Verteidigung: 07.01.2011 ...und Alice fragte die Katze: „Wohin soll ich gehen?“ & die Katze antwortet: „Das kommt immer ein bisschen darauf an, wo du hinwillst...“ (Alice im Wunderland)

BIBLIOGRAPHY 3

Bibliography

M.Sc. Dorit Schleinitz Differentially Expressed Genes in Adipose Tissue and their Role in the Pathophysiology of the Human Metabolic Syndrome

University of Leipzig, Faculty of Biology, Pharmacy and Psychology Dissertation 142 Pages, 241 References, 14 Figures, 26 Tables, Appendix

The human metabolic syndrome is characterized by a heterogenic complex of symptoms, including central . Obesity itself is linked to major features of the metabolic syndrome such as insulin resistance, dyslipidemia or type 2 mellitus. It has been shown that obesity risk and resulting metabolic alterations are associated with adipose tissue distribution, adipocyte size and secretion of adipocytokines, which are in turn influenced by environmental factors and genetic susceptibility. It might be assumed that currently known genetic variants associated with obesity and/or BMI () as well as fat distribution explain up to 20 % of the variability in BMI and so, studies employing novel strategies are inevitable. In addition to the role of genetic variation, mRNA levels of several genes have been shown to be differentially expressed in subcutaneous (SC) and visceral (Vis) adipose tissue and to be correlated with obesity-related traits. It is scarcely investigated whether the obesity risk variants also might account for the variability in mRNA expression. The present thesis deals with novel obesity candidate genes, characterized by a differential mRNA expression in various fat depots. The association of genetic variants in these genes with obesity as part of the metabolic syndrome, and related traits was investigated in well characterized German cohorts. The main method used for genotyping was described in detail in a comprehensive review providing explicit troubleshooting and description of modified protocols for specific experimental needs. Further, the influence of genotypes on the gene expression levels was examined. While the differential expression for FTO could be described for the first time, the variant rs8050136 was shown to be significantly associated with obesity but not with the expression. Genetic variants in FASN were shown to be significantly associated with obesity and related traits in a cohort of European ancestry for the very first time. Moreover, one polymorphism showed effects on the ratio of Vis/SC FASN mRNA expression. While CNR1 is controversially discussed in the literature, the present work showed rather moderate effects of genetic variants on obesity. BMPR2 could be described as a novel obesity candidate gene. Amongst others, one variant was associated with obesity in a case-control design and with BIBLIOGRAPHY 4

BMPR2 mRNA expression in Vis adipose tissue. In conclusion, the present study revealed novel genetic variants promoting obesity, and therefore a metabolic risk, which might be partly explicable through an influence of these variants on the mRNA expression levels of the genes in the adipose tissue depots. These findings contribute to better understanding of the genetic background of obesity which is essential in order to translate experimental data into diagnostic, preventive and treatment strategies. TABLE OF CONTENT 5

TABLE OF CONTENT

BIBLIOGRAPHY 3 TABLE OF CONTENT 5 ABBREVIATIONS 9 SUMMARY 12 ZUSAMMENFASSUNG 17

CHAPTER 1 23 Introduction to the topic 23

1.1 Human Metabolic Syndrome 23 1.2 Obesity 23 1.2.1 Adipose tissue – developmental origin and its distribution in the human body 24 1.2.2 Fat depots and their relation to the human metabolic syndrome 26 1.3 Genetics of obesity and adipose tissue distribution 28 1.3.1 Genetics of obesity 28 1.3.2 Genes and gene expression pattern of adipose tissue distribution 29 1.4 Aims of the presented work 31

CHAPTER 2 33

Targeted SNP Genotyping Using the TaqMan® Assay 33

2.1 Abstract 34 2.2 Introduction 34 2.3 Materials 37 2.3.1 Target DNA 37 2.3.2 Plastics 37 2.3.3 Genotyping reaction 37 2.3.4 Instruments and software 37 2.4 Methods 38 2.4.1 Preparation of samples to perform SNP genotyping assays 38 2.4.2 Preparation of master mix for SNP genotyping 38 2.4.3 Cycling program for genotyping reaction 39 2.4.4 Plate readout – allelic discrimination 39 2.5 Notes 41 2.6 References 43 2.7 Contribution of Authors 44

TABLE OF CONTENT 6

CHAPTER 3 45 Inverse Relationship between Obesity and FTO Gene Expression in Visceral Adipose Tissue in Humans 45

3.1 Abstract 46 3.2 Introduction 46 3.3 Methods 47 3.3.1 Participants and measurement of participant characteristics 47 3.3.2 Assays and measures of glucose and body fat content 48 3.3.3 Analysis of human FTO and RPGRIP1L mRNA expression 48 3.3.4 Genotyping of the rs8050136 FTO variant 49 3.3.5 Statistical analyses 49 3.4 Results 49 3.4.1 Participant characteristics 49 3.4.2 Association of FTO expression 51 3.4.3 Adipose tissue FTO mRNA expression in function of rs8050136 53 3.5 Discussion 54 3.6 Acknowledgements 56 3.7 Duality of interest 56 3.8 References 56 3.9 Contribution of Authors 58

CHAPTER 4 59 Effect of Genetic Variation in the Human Fatty Acid Synthase Gene (FASN) on Obesity and Fat Depot-Specific mRNA Expression 59

4.1 Abstract 60 4.2 Introduction 60 4.3 Experimental subjects 61 4.3.1 Leipzig adult cohort 61 4.3.2 Childhood cohorts 62 4.4 Methods and Procedures 63 4.4.1 Analysis of human FASN gene expression 63 4.4.2 Sequencing of the FASN 63 4.4.3 Genotyping of the FASN SNPs 64 4.4.4 Statistical analyses 64 4.5 Results 65 4.5.1 Genetic variation in the FASN 65 4.5.2 Association studies in adults 67 4.5.3 Association studies in children 71 4.5.4 Association of genetic variants in the FASN with fat depot-specific FASN mRNA expression 71 4.6 Discussion 72 4.7 Acknowledgements 75 4.8 Disclosure 75 4.9 References 75 4.10 Supplement 77 4.11 Contribution of Authors 80 TABLE OF CONTENT 7

CHAPTER 5 81 Role of Genetic Variation in the Cannabinoid Type 1 Receptor Gene (CNR1) in the Pathophysiology of Human Obesity 81

5.1 Abstract 82 5.2 Introduction 82 5.3 Materials & methods 83 5.3.1 Leipzig cohort 83 5.3.2 Measurement of obesity & glucose metabolism parameters 84 5.3.3 Sorbian cohort 84 5.3.4 DNA extraction & sequencing 85 5.3.5 Genotyping of CNR1 SNPs 85 5.3.6 Statistical analysis 86 5.4 Results 86 5.4.1 Sequencing 86 5.4.2 Association studies 87 5.4.3 Leipzig cohort 87 5.4.4 Sorbian cohort 90 5.4.5 CNR1 genetic variants & fat-depot-specific CNR1 mRNA expression 90 5.5 Discussion 91 5.6 Acknowledgements 93 5.7 Disclosure 93 5.7.1 Financial & competing interests disclosure 93 5.7.2 Ethical conduct of research 93 5.8 Executive summary 94 5.9 References 94 5.10 Supplement 97 5.11 Contribution of Authors 99

CHAPTER 6 100 Human Bone Morphogenetic Protein Receptor 2 (BMPR2) is Involved in the Pathophysiology of Obesity 100

6.1 Abstract 101 6.2 Introduction 101 6.3 Research design and methods 103 6.3.1 Study subjects 103 6.3.2 Evolutionary analyses 104 6.3.3 Analysis of human BMPR2 mRNA expression 104 6.3.4 Sequencing of the BMPR2 105 6.3.5 Genotyping of BMPR2 SNPs 105 6.3.6 Statistical analyses 105 6.4 Results 106 6.4.1 Evolutionary analyses 106 6.4.2 Visceral and subcutaneous BMPR2 mRNA expression and obesity 107 6.4.3 Correlation of BMPR2 mRNA expression with parameters of obesity, glucose metabolism, and insulin sensitivity 108 6.4.4 Genetic variation in the BMPR2 110 TABLE OF CONTENT 8

6.4.5 BMPR2 variants and association with obesity 110 6.4.6 BMPR2 variants and association with obesity related quantitative traits 112 6.4.7 Association of rs6717924 with BMPR2 mRNA expression 112 6.5 Discussion 113 6.6 Acknowledgements 116 6.7 References 116 6.8 Supplement 119 6.9 Contribution of Authors 125

CHAPTER 7 126 Closing Discussion 126

7.1 References - Introduction and Closing Discussion 131

APPENDIX 138 Curriculum Vitae 138 Publications & Presentations 139 Erklärung über die eigenständige Abfassung der Arbeit 141 Danksagung 142

ABBREVIATIONS 9

ABBREVIATIONS (engl., dt.)

2-AG 2-arachidonylglycerol AU Arbitrary Unit, beliebige Einheit BAT Brown adipose tissue, braunes Fettgewebe BMI Body mass index, Körpermasseindex BMP/R Bone morphogenetic protein/receptor CETP Cholesteryl-ester transfer protein CNR Cannabinoid receptor, Cannabinoidrezeptor CT Computed tomography, Computertomographie CTNNBL1 Catenin beta, like 1 DEXA Dual energy X-ray absorptiometry dNTP Deoxyribonucleotide triphosphate

DPI3 Dihydrocyclopyrroloindole tripeptide ECS Endocannabinoid system EGFP Enhanced green fluorescent protein 6-FAM 6-Carboxyfluorescein FASN Fatty acid synthase, Fettsäure-Synthase FRET Fluorescence resonance energy transfer FTO/Fto Fat mass and obesity associated (human/mouse) GWAS Genome wide association study, genomweite Assoziationsstudie HSC Hematopoietic stem cell, hämatopoetische Stammzelle HWE Hardy-Weinberg equlibrium, Hardy-Weinberg Gleichgewicht IGT Impaired glucose tolerance, gestörte Glucose-Toleranz iHS Integrated haplotype score IL Interleukin INSIG2 Insulin induced gene 2 IRS-1 Insulin receptor substrate 1 KFO Klinische Forschergruppe L.D. Linkage disequilibrium, Kopplungsungleichgewicht LEP LEPR LYPLAL1 Lysophospholipase, like 1 MC4R ABBREVIATIONS 10

MCP Monocyte chemoattractant protein MGB Minor groove binder MRI Magnetic resonance imaging, Magnetresonanztomographie mRNA messengerRNA MSC Mesenchymal stem cell, mesenchymale Stammzelle MSRA Methionine sulfoxide reductase A NGT Normal glucose tolerant, normale Glucose-Toleranz NEGR1 Neuronal growth regulator 1 NFQ Non fluorescent quencher NTC Non template control OGTT Oral glucose tolerance test, oraler Glucose-Toleranztest PAI-1 Plasminogen activator inhibitor PAML Phylogenetic analysis by Maximum Likelihood PCR Polymerase chain reaction, Polymerase Kettenreaktion PCSK1 Proprotein convertase subtilisin/kexin type 1 PET Positron-emission tomography, Positronen-Emissions-Tomographie POMC PPARG Peroxisome proliferator-activated receptor gamma QC Quality control, Qualitätskontrolle RBP Retinol binding protein RFLP Restriction fragment lenght polymorphism RNA Ribonucleic acid, Ribonucleinsäure ROX 6-Carboxy-X-rhodamine RPGRIP1 Retinitis pigmentosa GTPase regulator interacting protein 1 RT Reverse transcriptase SAA Serum amyloid A SC Subcutaneous, subkutan SD Standard deviation, Standardabweichung SDS Standard deviation scores SDS software Sequence detection system software SEM Standard error of the mean, Standardfehler des Mittelwertes SH2BP1 SH2 domain binding protein 1 SMAD Mothers against decapentaplegic SNP Single nucleotide polymorphism, Einzelbasenaustausch ABBREVIATIONS 11

T2D mellitus Taq Thermus aquaticus TMEM18 Transmembrane protein 18 TFAP2B Transcription factor activating enhancer binding protein 2 beta TNFA Tumor necrosis factor alpha UCP 1/2 Uncoupling protein 1/2 UTR Untranslated region, untranslatierte Region Vis Visceral, viszeral Vs. Versus WAT White adipose tissue, weißes Fettgewebe WC Waist circumference, Bauchumfang WHR Waist-to-hip ratio, Taille-Hüft-Verhältnis SUMMARY 12

SUMMARY

The human metabolic syndrome is characterized through the coeval appearance of a cluster of metabolic alterations such as central obesity, insulin resistance, dyslipidemia, , and type 2 diabetes mellitus (T2D). The metabolic syndrome is influenced by complex interactions of both environmental as well as genetic components. Intra-abdominal, or visceral, obesity is considered to be the main factor responsible for metabolic sequelae such as insulin resistance or T2D. It has been shown that the obesity risk and consequent metabolic alterations are associated with adipose tissue distribution, adipocyte size, and secretion of adipocytokines. Besides monogenic forms, obesity and adipose tissue distribution is of polygenic nature. During the last decade, candidate gene and genome wide association studies increased the number of known genes and genetic variants promoting obesity risk rapidly. Moreover, mRNA levels of several genes have been shown to be differentially expressed in subcutaneous (SC) and visceral (Vis) adipose tissue. The expression levels were associated with various traits related to the metabolic syndrome such as body fat content, measures of glucose metabolism or insulin sensitivity. However, the underlying molecular mechanisms are poorly understood and it is estimated that the genetic variants identified so far do only account for approximately 20 % variability in body mass index (BMI). It is evident that more studies are necessary to clarify the complex genetic interplay regulating adipose tissue distribution and obesity, and to explain the mechanisms by which the genetic variants exert their effects. The present thesis describes investigations on gene’s mRNA expression pattern in SC and Vis adipose tissue depots to clarify the role of the genes in a tissue largely responsible for metabolic complications. Further, the association of single nucleotide polymorphisms (SNPs) in potential candidate genes with obesity and related anthropometric/metabolic traits is elucidated to clarify their contribution to the genetic risk of developing the human metabolic syndrome. Finally, it is investigated whether the variability in gene expression levels is mediated by the SNPs. The current knowledge on adipose tissue physiology and on the genetic background of obesity and adipose tissue distribution as well as the aims of the work are summarized in CHAPTER 1. The aim of the studies was to investigate the role of the fat mass and obesity associated gene, FTO, the fatty acid synthase gene, FASN, the cannabinoid receptor type 1 gene, CNR1 and the bone morphogenetic protein receptor 2 gene, BMPR2, in the pathophysiology of the human metabolic syndrome. SUMMARY 13

Within the expression studies, real-time PCR was carried out in paired samples of SC and Vis adipose tissue derived from individuals with wide ranges of obesity and body fat distribution. To cover the entire gene locus including the coding sequence and the introns by common genetic variants (minor frequency > 5 %) a SNP-“tagging”-approach based on linkage disequilibrium (L.D.) pattern was applied. Therefore, SNPs from the HapMap database were selected and each gene was sequenced to detect novel genetic variation. L.D. analysis was carried out and one representative SNP for each L.D. group was genotyped for association studies. Genotyping was done in well-characterized cohorts (Leipzig, N~1800; and Sorbs, N~1000) using the TaqMan® technique or restriction fragment length polymorphism (RFLP). The TagMan® technique which was used as a main method was described in detail in CHAPTER 2 in form of a comprehensive review providing explicit troubleshooting and description of modified protocols for specific experimental needs. For genotype-phenotype association studies all cohorts have been extensively characterized regarding intermediate phenotypes such as anthropometric (age, BMI, waist-to-hip ratio [WHR], body fat content) and metabolic parameters (glucose and insulin measurements, lipids, adipocytokines). Adequate statistical analyses were conducted in all studies.

In CHAPTER 3, the mRNA expression of the Fat Mass and Obesity Associated gene, FTO, was investigated. Noteworthy, the effect of FTO variants on obesity was initially detected in a genome wide scan for T2D in which the association between FTO SNPs and T2D was mediated via obesity. In mice, a mutation in the Ft gene, the original name of FTO, led to fused toes (hence the abbreviation Ft) but no obese phenotype was reported. Amongst others, FTO has been shown to be expressed in human adipose tissue but the function has remained unknown. In the present study, it was hypothesized that FTO mRNA expression in SC and Vis adipose tissue is linked to obesity. In addition to FTO, its neighboring gene, RPGRIP1L (RPGRIP1-like) was targeted as well. Furthermore it was investigated whether the previously published obesity-associated FTO variants, represented by rs8050136, might explain potential variability in FTO and RPGRIP1L mRNA expression. When compared in the two fat depots, FTO mRNA was higher expressed in SC adipose tissue, while RPGRIP1L mRNA expression was higher in Vis adipose tissue. Although FTO expression in both SC and Vis adipose tissue were correlated with measures of obesity and with each other, the Vis FTO mRNA expression remained an independent predictor of BMI even after adjustment for subcutaneous FTO mRNA expression. No correlations were observed between FTO mRNA expression levels and measures of glucose and insulin metabolism in either fat depot. In SUMMARY 14 contrast to FTO, no correlation of RPGRIP1L mRNA expression with anthropometric and metabolic parameters was seen which suggests that FTO rather than RPGRIP1L is involved in the pathophysiology of obesity. Although rs8050136 was associated with obesity, no relation with SC or Vis FTO mRNA levels was observed. The results presented in CHAPTER 3, provided the first data highlighting the relationship between FTO and RPGRIP1L mRNA expression in adipose tissue and anthropometric and metabolic parameters including measures of obesity.

Fatty acid synthase (FASN) is a key enzyme in the pathway of de novo lipogenesis. It catalyses the main steps in the conversion of malonyl CoA to palmitate. Recent data demonstrated a strong association of increased FASN expression in adipose tissue not only with obesity, but also with T2D and insulin resistance, thus indicating a possible role of lipogenic pathways in the development of human obesity and/or T2D. Because of the important physiological role of its protein, the FASN is an attractive candidate gene to be investigated in association studies for obesity and related traits. Moreover, studies in Pima Indians from Southern Arizona and German children (Leipzig) revealed a novel Val1483Ile polymorphism protective against obesity. However, genotype-phenotype association studies examining the role of genetic variants of FASN in obesity in adult Caucasians have been lacking so far. Also, it has not yet been clarified, whether genetic variation within the FASN might be responsible for changes in FASN expression in adipose tissue. The study presented in CHAPTER 4 addressed these questions in adult and children cohorts. Several polymorphisms in the FASN associated with obesity in case-control studies in adults could be detected in the project. The strongest effect was found for rs2229422, which was supported by associations with BMI, WHR, fasting plasma insulin, and glucose infusion rate, as measured in a euglycemic-hyperinsulinemic clamp. In addition, rs17848939 was associated with the ratio of Vis/SC FASN mRNA expression which was also reflected in the association of this SNP with WHR. The protective effect of the Val1483Ile substitution could be replicated in the adult cohort. No effect of genetic variation in FASN on obesity was found in children indicating that the SNP effects most likely manifest in later stages of life and/or that these effects might be modified by gene-environment interactions. In conclusion, it could be shown for the first time that polymorphisms in FASN contribute to the increased risk of developing metabolic alterations during adult life. Moreover, the variants might exert their effects via an adverse effect on the adipose tissue distribution towards the visceral depot, as well as by modulating the FASN expression in this depot. SUMMARY 15

As part of the endocannabinoid system (ECS), the CNR1 participates in the control of food intake via both central and peripheral mechanisms, and in lipid and glucose metabolism. In a previous study it has been shown that CNR1 was present in human Vis and SC adipocytes and CNR1 mRNA expression in Vis adipose tissue correlated with circulating 2- arachidonylglycerol levels (2-AG), percent body fat, Vis fat area, and fasting plasma insulin. Despite its physiologic plausibility to serve as a candidate gene for obesity, genotype- phenotype associations on genetic variants in CNR1 gene have been rather controversial, which is most likely due to the heterogeneity of studied populations. The studies demonstrated the necessity of various independent cohorts being involved in the analyses in order to evaluate the role of this obesity candidate gene. In CHAPTER 5, genetic analyses on CNR1 in two German populations are presented. In the present study, the SNP rs806368 was found to be associated with obesity in the Sorbs, but not in the Leipzig cohort. Also, several SNPs were associated with serum leptin levels, while lacking associations with obesity. This needs to be seen with caution, since it is unlikely that variants in CNR1 regulate leptin levels but have no effect on obesity. However, there was a moderate effect of two CNR1 SNPs on 2- AG plasma levels. Finally, no SNP effect on fat CNR1 mRNA expression could be detected. This suggests that genetic variation in the CNR1 does not play a significant role in the pathophysiology of obesity in the German and Sorbian population.

Based on studies in mice there is strong evidence that developmental genes could be differentially expressed in human SC and Vis adipose tissue depots. Bone morphogenetic proteins (BMPs) and their receptors (BMPRs) play an important role in multiple key steps of embryonic development and differentiation. BMPs are involved in adipocyte development, including adipose cell fate determination, differentiation of committed preadipocytes and function of mature adipocytes. It was recently shown that the BMPR1A mRNA expression in both Vis and SC adipose tissue as well as genetic variants in this gene strongly correlated with obesity and its related traits. BMPR2 is responsible for the trans-phosphorylation of BMPR1A, which makes it a plausible candidate gene for obesity. This hypothesis is addressed in CHAPTER 6. Interestingly, BMPR2 was suggested as one of the potential targets of recent selection in European populations and therefore supposed to be a “thrifty gene” which might have provided advantage to store energy in times of food scarcity. Therefore in the present work studies searching for signatures of selection were initiated. The analyses revealed that BMPR2 has undergone an overall purifying selection and population measures supported the “thrifty gene” hypothesis. It could be shown that the BMPR2 mRNA SUMMARY 16 expression was significantly increased both in Vis and SC adipose tissue of and obese compared to lean subjects whereas the shift in the mRNA expression level was remarkable in the visceral adipose tissue depot. Two SNPs (rs6717924, rs13426118) were found to be associated with obesity in the Leipzig cohort and combined analyses including a second cohort confirmed a consistent effect for rs6717924. In addition, SNP effects on BMI, adiponectin, percent body fat, fasting plasma insulin and 2-h plasma glucose levels could be observed. Moreover, rs6717924 was associated with higher BMPRR2 mRNA expression in Vis adipose tissue independent from BMI which may provide a possible causative link in the genotype-effect relation. These data suggested a role of BMPR2 in the pathophysiology of obesity.

CHAPTER 7 comprises a closing discussion on the hypotheses investigated in the paper and with regard to the literature which was released after the publications included in this thesis.

In summary, this thesis provided new insights in the complex genetic background of the human metabolic syndrome, in particular of one of its main features represented by obesity. The genotype-phenotype association analyses revealed two new obesity susceptibility genes, FASN and BMPR2. No doubt, the investigation of intermediate phenotypes is essential to accentuate the phenotype and therefore to understand the SNP effect more clearly. This aspect gets a new dimension by measuring mRNA expression and correlating it with the genotype. The study of the gene expression profiles in the adipose tissue depots of lean and obese non- diabetic subjects and subjects with T2D in combination with the polymorphisms provided a helpful approach to understand how genetic variants might exert their effect. These findings open new avenues in obesity research ultimately leading to comprehensive functional studies which are essential to identify and understand the complete mechanistic action of the identified polymorphisms.

ZUSAMMENFASSUNG 17

ZUSAMMENFASSUNG

Das humane „Metabolische Syndrom“ ist durch das gleichzeitige Auftreten mehrer metabolischer Störungen charakterisiert. Dazu gehören abdominale Adipositas, Insulinresistenz, Glucoseintoleranz, Dyslipidämie, Bluthochdruck und Typ 2 Diabetes mellitus (T2D). Bedingt wird das metabolische Syndrom durch komplexe Interaktionen zwischen Umwelt und genetischen Komponenten. Die intra-abdominale, oder viszerale Adipositas wurde dabei als ein Hauptrisikofaktor postuliert, der für viele metabolische Folgeerscheinungen wie z.B. Insulinresistenz oder T2D verantwortlich ist. Es wurde gezeigt, dass das Adipositasrisiko und daraus resultierende metabolische Störungen mit der Fettverteilung, Adipozytengröße und der Sekretion von Adipozytokinen assoziiert sind. Neben einigen Krankheitsformen, die durch einen monogenen Defekt verursacht werden, unterliegen Adipositas und Fettverteilung einem polygenen Einfluss. Kandidatengen- und genomweite Assoziationsstudien ließen im letzten Jahrzehnt die Zahl der bekannten Gene und Varianten, die ein Adipositasrisiko fördern, schnell ansteigen. Weiterhin wurde gezeigt, das die mRNA verschiedener Gene zum einem im subkutanen (SC) und viszeralen (Vis) Fettgewebe differenziell exprimiert, zum anderen auch in den Fettgewebsdepots von schlanken und adipösen Individuen. Die Expressionslevel dieser Gene waren unter anderem auch mit dem Gehalt an Körperfett und Messungen des Glucose- und Insulin-Stoffwechsels assoziiert. Die zu Grunde liegenden molekularen Mechanismen sind bisher wenig verstanden und die bisher identifizierten genetischen Varianten erklären erst circa 20 % der Variabilität des Körpermasseindex (body mass index, BMI). Zum tieferen Verständnis des komplexen genetischen Hintergrundes, welcher Fettverteilung, Adipositas und die damit verbundenen metabolischen Störungen beeinflusst und zur Detektion basaler Mechanismen, über die die genetischen Varianten ihren Einfluss ausüben können, sind weitere Studien notwendig. Die Studien, die in der vorliegenden Dissertation vorgestellt werden, untersuchen mRNA Expressionsmuster von Genen im SC und Vis Fettgewebsdepot um zu klären, ob diese eine Rolle in einem Gewebe spielen, das in hohem Maße für metabolische Komplikationen verantwortlich ist. Weiterhin wird die Assoziation von Polymorphismen in potentiellen Kandidatengenen mit Adipositas und zugehörigen anthropometrischen/metabolischen Parametern untersucht um herauszufinden, ob sie zum genetischen Risiko beitragen, einen Phänotyp in Richtung des metabolischen Syndroms zu entwickeln. Schlussendlich wird analysiert, ob die Genexpressionslevel durch die Polymorphismen beeinflusst werden. In Kapitel 1 werden der derzeitige Wissensstand zur Physiologie des Fettgewebes und zum ZUSAMMENFASSUNG 18 genetischen Hintergrund von Adipositas und Fettverteilung sowie die Ziele der Arbeit zusammengefasst. Untersucht wurde die Rolle des fat mass and obesity associated Gens, FTO, des Gens der Fettsäure-Synthase, FASN, des Gens für den Cannabinoidrezeptor 1, CNR1, und des bone morphogenetic protein receptor 2 Gens, BMPR2, in der Pathophysiologie des humanen Metabolischen Syndroms. Die Genexpressionsstudien wurden mittels real-time PCR in gepaarten SC und Vis Fettgewebsproben durchgeführt. Diese stammten von Probanden, die einen großen BMI- Bereich und verschiedene Fettverteilungsmuster abdeckten. Um die gesamte Sequenz eines Gens (Exons und Introns) durch häufig vorkommende genetische Varianten (SNPs – single nucleotide polymorphisms) abdecken zu können (die Frequenz des minor Allels beträgt mindestens 5 %) wurde basierend auf dem Kopplungsungleichgewicht (L.D. – linkage disequilibrium) ein SNP-„tagging“ Ansatz angewendet. Zur Identifizierung neuer genetischer Varianten wurde das Gen sequenziert und die Ergebnisse durch die bereits beschriebenen genetischen Polymorphismen aus der HapMap Datenbank ergänzt. Nach der L.D. Analyse wurde aus jeder Kopplungsgruppe ein repräsentativer SNP für die Genotypisierung gewählt. Diese wurde mit Hilfe der TaqMan® Technik in gut charakterisierten Kohorten durchgeführt. Im Kapitel 2 wird diese Methode im Rahmen eines Buchkapitels als Review zusammengefasst, welches basierend auf den eigenen Experimenten einen Absatz zur Problembehandlung und die Beschreibung modifizierter Protokolle für bestimmte experimentelle Anforderungen enthält. Für Genotyp-Phänotyp-Assoziationsstudien wurden die Kohorten (eine Leipziger, N~1800; und eine Sorbische Kohorte, N~1000) ausführlich hinsichtlich intermediärer Phänotypen wie anthropometrischer (Alter, BMI, Taille-Hüft- Verhältnis [WHR], Körperfettgehalt) und metabolischer Parameter (Glucose und Insulin, Lipide, Adipozytokine) untersucht. Umfangreiche statistische Analysen wurden in allen Studien durchgeführt.

Im Kapitel 3 wird die mRNA Expression von FTO untersucht. Interessanterweise wurde FTO in einer genomweiten Assoziationsstudie für T2D entdeckt, in der sich zeigte, dass die Assoziation zwischen SNPs und Diabetes durch den BMI beeinflusst wurde. In Mäusen führt eine Mutation in diesem Gen zu verschmolzenen Zehen, daher der originale Name von FTO - fused toes, Ft, aber ein adipöser Phänotyp konnte nicht beobachtet werden. In der vorliegenden Studie wurde die Hypothese untersucht, dass die FTO mRNA Expression im SC und Vis Fettgewebe mit Adipositas verknüpft sein könnte. Aufgrund der benachbarten ZUSAMMENFASSUNG 19

Lokalisation zum FTO wurde ein zweites Gen, RPGRIP1L (RPGRIP1-like), diskutiert. Weiterhin wurde untersucht, ob die publizierten Adipositas-assoziierten Varianten, repräsentiert durch den SNP rs8050136, eventuell die mögliche Variabilität der FTO und RPGRIP1L mRNA Expression erklären könnte. Beim Vergleich der beiden Fettdepots zeigte sich, dass FTO höher im subkutanen Fettgewebe, RPGRIP1L dagegen höher im viszeralen Fettgewebe exprimiert war. Obwohl die FTO Expression in beiden Geweben mit Adipositas- Parametern und miteinander korrelierte, blieb die Vis FTO mRNA Expression auch nach Adjustierung auf die subkutane Expression ein unabhängiger Prädiktor für den BMI. Es konnten keine Korrelationen der FTO mRNA Expression zu Messungen des Glucose- und Insulin-Stoffwechsels in keinem der beiden Fettdepots gefunden werden. Für RPGRIP1L wurden keine Korrelationen zu anthropometrischen und metabolischen Parametern gefunden. Das lässt darauf schließen, dass eher FTO als RPGRIP1L in die Pathophysiologie der Adipositas involviert ist. Auch wenn eine Assoziation von rs8050136 mit Adipositas in dieser Kohorte gefunden werden konnte, so war der SNP jedoch nicht mit der FTO Expression assoziiert. Die Ergebnisse, die im Kapitel 2 präsentiert werden, zeigen zum ersten Mal Daten, die das Verhältnis zwischen FTO und RPGRIP1L mRNA Expression im Fettgewebe und zu anthropometrischen und metabolischen Parametern, eingeschlossen Parametern der Adipositas, beleuchten.

Die Fettsäure-Synthase, FASN, ist ein Schlüsselenzym bei der de novo Lipogenese. Sie katalysiert die Schritte der Konversion von Malonyl-CoA zu Palmitat. Es konnte gezeigt werden, dass eine erhöhte FASN Expression im Fettgewebe nicht nur signifikant mit Adipositas assoziiert ist, sondern auch mit T2D und Insulinresistenz. Dies legt nahe, dass die lipogenen Stoffwechselwege eine mögliche Rolle bei der Entwicklung der humanen Adipositas und/oder T2D spielen könnten. Aufgrund der wichtigen physiologischen Rolle des Proteins ist das FASN Gen zudem ein attraktives Kandidatengen um es in Genotyp-Phänotyp- Analysen für Adipositas und zugehörigen Parametern zu untersuchen. Erste Studien in den Pima Indianern aus Süd-Arizona und in deutschen Kindern zeigten, dass es einen Polymorphismus im Gen gibt (Val1483Ile), der vor Adipositas zu schützen scheint. Genotyp- Phänotyp Assoziationsstudien, die die Rolle von genetischen Varianten in FASN bei Adipositas in erwachsenen Populationen untersuchen, fehlen jedoch. Außerdem ist auch noch nicht geklärt, ob genetische Varianten in FASN für die Veränderungen der FASN mRNA Expression im Fettgewebe verantwortlich sind. Die Studie, die im Kapitel 4 präsentiert wird, untersucht diese Fragen in Erwachsenen- und Kinderkohorten. In einer Fall-Kontroll-Studie ZUSAMMENFASSUNG 20 konnte gezeigt werden, dass mehrere Polymorphismen in FASN mit Adipositas assoziiert sind. Der stärkste Effekt wurde für rs2229422 gefunden, weiterhin unterstützt durch Assoziationen mit dem BMI, WHR, Nüchtern-Plasmainsulin und der Glucoseinfusionsrate. Ein weiterer SNP, rs17848939, wurde zudem mit dem Verhältnis der Vis/SC FASN mRNA Expression assoziiert, was sich ebenfalls in einer Korrelation dieses SNPs mit der WHR widerspiegelte. Der protektive Effekt der Val1483Ile Substitution konnte in der Erwachsenenkohorte repliziert werden. In der Kinderkohorte waren die genetischen Varianten in FASN nicht mit Adipositas assoziiert, was darauf hindeutet, dass die SNP Effekte sich wohlmöglich erst im späteren Leben manifestieren und/oder die Effekte durch Gen-Umwelt- Interaktionen modifiziert werden. In der vorliegenden Dissertation wurde zum ersten Mal gezeigt, dass Polymorphismen in FASN zum erhöhten genetischen Risiko beitragen metabolische Störungen während des Erwachsenenlebens zu entwickeln. Weiterhin kann postuliert werden, dass die Polymorphismen ihren Einfluss möglicherweise über die Fettgewebsverteilung in Richtung des viszeralen Depots vermitteln und die FASN Expression in diesem Depot modulieren.

Als Teil des Endocannabinoidsystems (ECS) ist der CNR1 über zentrale und periphere Mechanismen sowohl in die Kontrolle der Nahrungsaufnahme als auch in den Fett- und Glucosestoffwechsel involviert. Expressionsstudien im Fettgewebe haben gezeigt, dass der CNR1 im humanen SC und Vis Adipozyten exprimiert wird und die Expression mit dem zirkulierenden 2-Arachidonylglycerol (2-AG), dem Anteil an Körperfett, der viszeralen Fettmenge und Nüchtern-Plasmainsulin korreliert. Trotz der physiologischen Plausibilität als Kandidatengen für Adipositas zu fungieren, wurden in verschiedenen Populationen kontroverse Ergebnisse gefunden. Dies ist vermutlich der genetischen Heterogenität geschuldet, was die Notwendigkeit aufzeigt, das jeweilige Risiko eines Gens für eine Erkrankung in verschiedenen Kohorten zu untersuchen. In Kapitel 5 werden genetische Analysen von CNR1 in zwei deutschen Populationen präsentiert. Die Ergebnisse der Studie zeigten, dass ein SNP, rs806368, zwar in der Sorbischen aber nicht in der Leipziger Kohorte mit Adipositas assoziiert ist. Auch wurde für einige Varianten ein Effekt auf den Leptinspiegel gezeigt, aber es ist eher unwahrscheinlich, dass diese CNR1 Polymorphismen die Leptinspiegel regulieren, ohne einen Effekt auf Adipositas zu haben. Es konnte ein geringer Effekt von zwei CNR1 SNPs auf die 2-AG Plasmaspiegel gezeigt werden. Schlussendlich wurde keine Assoziation der Varianten im Gen mit der mRNA Expression des Gens beobachtet. Die hier zusammengefasste Studie zeigte auf, dass allgemeine genetische ZUSAMMENFASSUNG 21

Varianten in CNR1 in der deutschen Population eher eine untergeordnete Rolle im genetischen Hintergrund der Adipositas und den damit verbundenen metabolischen Störungen zu spielen scheinen.

Basierend auf Studien in Mäusen sind Hinweise gefunden worden, dass Entwicklungsgene differenziell im human subkutanen und viszeralen Fettgewebe exprimiert sein könnten. Bone Morphogenetic Proteins (BMPs) und ihre Rezeptoren (BMPRs) spielen in einer Vielzahl von Schritten bei der Embryogenese und bei der Differenzierung von Geweben eine wichtige Rolle. Unter anderem wurde gezeigt, dass BMPs auch an der Differenzierung von mesenchymalen Stammzellen in Adipozyten beteiligt sind. Kürzlich konnte gezeigt werden, dass sowohl die BMPR1A mRNA Expression als auch genetische Varianten im BMPR1A stark mit Adipositas und dazugehörigen metabolischen Parametern korrelieren. Der BMPR2 ist verantwortlich für die Trans-Phosphorylierung von BMPR1A, was dessen Gen als Kandidat für genetische Studien interessant macht. Diese Hypothese wird in Kapitel 6 untersucht. Zusätzlich wies eine Studie darauf hin, dass BMPR2 in europäischen Populationen einer kürzlichen Selektion unterlegen zu sein scheint, wodurch es als mögliches Thrifty Gene postuliert wurde. Thrifty gene bedeutet dabei, dass bestimmte genetische Varianten in Zeiten von möglicherweise einen Vorteil bei der Speicherung von Energie erbrachten. Daher wurde in der vorliegenden Studie auch nach Zeichen von Selektion gesucht. Die Ergebnisse der vorliegenden Studie zeigten, das BMPR2 tatsächlich einer negativen Selektion unterlag und Populationsanalysen unterstützten die Thrifty Gene Hypothese. Die BMPR2 mRNA Expression war sowohl im subkutanen als auch im viszeralen Fettgewebe übergewichtiger und adipöser Probanden im Vergleich zu Schlanken erhöht, wobei die Veränderung am deutlichsten für das viszerale Fettgewebe zu beobachten war. Zwei SNPs, rs6717924 und rs13426118, zeigten eine Assoziation mit Adipositas in der Leipziger Kohorte und kombinierte Analysen, die eine zweite Kohorte beinhalten, bestätigten den Effekt für rs6717924. Zusätzlich konnten Assoziationen von BMPR2 SNPs mit dem BMI, dem prozentualen Anteil an Körperfett, Nüchtern-Plasmainsulin und 2-h Glucose beobachtet werden. Darüber hinaus war der SNP rs6717924 unabhängig vom BMI mit der BMPR2 mRNA Expression im viszeralen Fettgewebe assoziiert, was einen möglichen Hinweis auf die ursächliche Verbindung der Genotyp-Effekt-Beziehung bedeutet. Somit weisen diese Ergebnisse auf eine Rolle von BMPR2 in der Pathophysologie der Adipositas hin.

ZUSAMMENFASSUNG 22

Kapitel 7 enthält eine abschließende Diskussion zum Thema der vorliegenden Doktorarbeit. Es wird ebenfalls die Literatur berücksichtigt, die nach den Publikationen, die Teil dieser Dissertation sind, veröffentlicht wurde.

Zusammenfassend kann konstatiert werden, dass die vorliegende Arbeit neue Einblicke in den komplexen genetischen Hintergrund des humanen metabolischen Syndroms gebracht hat, im speziellen auf eines seiner charakteristischen Merkmale – Adipositas und zugehörige anthropometrische und metabolische Parameter. Die Genotyp-Phänotyp Assoziationsstudien zeigten zwei neue Kandidatengene für Adipositas auf, FASN und BMPR2. Ohne Zweifel ist die Untersuchung intermediärer Phänotypen essentiell um den Phänotyp zu schärfen und damit die SNP-Wirkung klarer zu verstehen. Mit der Korrelation des Genotyps mit der mRNA Expression bekommt dieser Aspekt eine neue Dimension, die näher an die molekularen Mechanismen heranführen kann. Die Untersuchung der Genexpressionsprofile in den Fettgewebsdepots schlanker und adipöser nicht-diabetischer Probanden und Probanden mit T2D in Kombination mit den genetischen Polymorphismen ist ein hilfreicher Ansatz um zu verstehen, wie genetische Varianten ihren Effekt vermitteln könnten. Damit werden neue Möglichkeiten in der Adipositasforschung eröffnet, die zu umfangreichen funktionellen Studien führen, die essentiell sind, um den kompletten Mechanismus zu verstehen der vom Genotyp zum Phänotyp führt.

CHAPTER 1 23

CHAPTER 1

Introduction to the topic

1.1 Human Metabolic Syndrome The human metabolic syndrome is a widely recognized concept which focuses on important complex multifactorial health problems (1). It became generally known as a heterogenic complex of features including central obesity, insulin resistance, dyslipidemia, hypertension, and type 2 diabetes mellitus (T2D) during the last decades (2,3). The metabolic syndrome is considered to be the strongest predictor for coronary heart disease. The components of the metabolic syndrome are pathophysiologically linked and in addition to environmental factors partly genetically determined (Figure 1). However, the pathogenesis of the clustering of risk factors referred to as the metabolic syndrome and its constituent elements remains unclear (4). One of the major risk factors, obesity, is itself linked to several of the major symptoms of the human metabolic syndrome like insulin resistance, glucose intolerance, dyslipidemia, and the development of T2D, and therefore a major object of investigation.

HEALTHY METABOLIC STATE ALTERATION HT obesity genes nutrition sports social contact DL IR drugs age education

Figure 1: Fragile metabolic balance. Several cross-linked influencing factors, some of them can be arranged on both pans, lead to either positive or negative for health. HT hypertension; DL dyslipidemia; IR insulin resistance.

1.2 Obesity Incidence for being overweight (body mass index (BMI) ≥ 25 kg/m2 – 29.9 kg/m2) or obese (BMI ≥ 30 kg/m2) increases rapidly, in the meantime every second inhabitant in Germany is considered to be overweight and prevalence of obesity increases worldwide. It is characterized by an increase of body fat content far in excess of conditions considered to be “normal” (15-31 % body fat in adults, depending on age and sex). Genetic factors combined with today’s lifestyle with increased access to highly palatable, calorie-dense foodstuff, CHAPTER 1 24 decreased need for physical activity in daily life, as well as social cultural factors are postulated to be the main causes predisposing to obesity. It has been known for decades that obesity might have a major impact on metabolic complications like glucose intolerance and insulin resistance (5). Common medical conditions such as T2D, fatty liver, , gallstones, and even some cancers occur in large part as a result of insulin resistance induced by obesity. Adipose tissue not only serves to store energy but is also the body’s largest endocrine organ, secreting hormones, cytokines, and proteins that affect the function of cells and tissues throughout the body (6-10). Thereby it has been emphasized quite early that adipose tissue distribution, especially the visceral proportion, seems to play a crucial role in severity of metabolic complications caused by the increase of body fat content (11).

1.2.1 Adipose tissue – developmental origin and its distribution in the human body Adipose tissue is a form of connective tissue mainly composed of adipocytes and adipocyte precursor cells but also cells of connective tissue (fibroblasts), vasculature (endothelial cells, smooth muscle cells), and cells of the immune system (monocytes, macrophages, lymphocytes) (12). Adipose tissue is assumed to have a mesodermal origin. Although the exact number of intermediate stages between a mesodermal/mesenchymal stem cell (MSC) and a mature adipocyte is uncertain, it is believed that MSCs give rise to a common early precursor (adipoblast), which in turn develops into committed white and brown preadipocytes that, under appropriate stimulatory conditions, differentiate into mature adipocytes of different types (6). However, as none of these precursor cells possesses any unique morphological characteristics or gene expression markers, it is not clear whether specific adipoblasts and/or preadipocytes for brown and white fat exist or whether there are different white preadipocytes for different white adipose depots. A recent study published by Isakson et al. found an increase of mesenchymal precursor cells in obese individuals in contrast to a decrease of the differentiation of preadipocytes into adipocytes, suggesting an impaired differentiation of preadipocytes in obesity (13). On the contrary, Ogawa et al. postulated that adipocytes may also derive from hematopoietic stem cells (HSC) as adipose tissue harvested from clonally engrafted mice showed enhanced green fluorescent protein positive (EGFP+) adipocytes after transplantation of single EGFP+ HSCs (14). As mentioned above, two types of adipose tissue determined by morphology and function are distinguished – white adipose tissue (WAT) and brown adipose tissue (BAT). Cells from BAT are characterized by smaller size (up to 30 µm cell size), are pluri-vacuolar and rich in mitochondria. Further, BAT has no macrophage infiltration and is relevant for CHAPTER 1 25 basal and inducible thermogenesis. It used to be assumed that this tissue regresses with increasing age and is completely lost by the time a person reaches adulthood (15). Over the past decade, studies in nuclear medicine using positron-emission tomography in conjunction with computed tomography (PET/CT) clearly demonstrated the presence of BAT in adult humans, e.g. localized in the armpit, at the mediastinum, big arteries or kidney (16,17). Recent studies suggest that BAT may be a potential pharmaceutical target to improve metabolic complications induced by obesity through stimulation of thermogenesis (18,19). Adipocytes occurring in WAT are up to 100 µm and nearly completely filled with one lipid droplet. Depending on subcutaneous (SC) or visceral (Vis, intra abdominal fat) localization the innervation/vascularisation is less (SC) or more dense (Vis), and macrophage infiltration is mainly present in WAT (20). White adipose tissue is the primary site of triglyceride, and therefore energy storage, and important also for insulation and mechanical protection. Lipogenesis and lipolysis are fine regulated through nutritional status, and hormonal and transcriptional factors (Figure 2) (21,22).

insulin glucose leptin

fatty acids SREBP-1

PPARG transcriptional regulation ? •SREBP-1 •USFs • ChoRF adipo-/lipogenesis lipogenesis •PPARG glucose • poly-unsaturated fatty acids acetyl-CoA acetyl-CoA citrate nutritional regulation lysase • rich in carbohydrates carboxylase •glucose malonyl-CoA • poly-unsaturated fatty acids fatty acid •fasting synthase fatty acids hormonal regulation glycerol-P •insulin acyl-transferase • growth hormone fatty acyl-CoA •leptin • acylation stimulation protein triglycerides

Figure 2: Lipogenesis in white adipose tissue. The effects of nutrients and hormones on the expression of lipogenic genes are mostly mediated by SREBP-1 and by PPARG. Lipogenesis entails a number of discrete steps which are controlled via allosteric interactions, by covalent modification and via changes in gene expression. ↑ up regulation and ↓ down regulation of lipogenesis; SREBP-1 sterol regulatory element binding protein-1; PPARG peroxisome proliferator-activated receptor gamma; USF upstream stimulatory factors; ChoRF carbohydrate response transcription factor. The figure is modified from the review by Kersten S. (21). CHAPTER 1 26

WAT is dispersed throughout the body with major intra abdominal depots around the omentum, intestines, and perirenal areas, as well as in subcutaneous depots in the buttocks, thighs and abdomen (6). In addition, WAT can be found in many other areas, including the retro-orbital space, the face and extremities, within the bone marrow, epicardial and perivascular. Some adipose tissue is responsive to sex hormones, such as adipose tissue in the breasts and thighs, whereas other depots, such as fat on the neck and upper back, are more responsive to glucocorticoids. The distribution of WAT can vary enormously between individuals. Measurement of the ratio of waist to hip circumference was shown to be a helpful tool to estimate the risk for metabolic complications of obesity (23). Individuals with low WHR (subcutaneous or pear- shaped obesity) are less prone to manifest metabolic complications of obesity, whereas individuals with a high WHR (visceral or apple-shaped obesity) are at high risk for these complications. There are two theories which may explain these differences: 1) the anatomic localization of the visceral fat predicts the possibility to drain its products into the portal circulation where they can preferentially act on the liver to affect metabolism; 2) the fat cells in different depots have different properties causing them to be linked to a greater or lesser extent to the development of metabolic disorders (24,25). Even though WHR is a simple tool to express fat distribution, magnetic resonance imaging (MRI), computed tomography (CT) or Dual-Energy X-Ray Absorptiometry (DEXA) measurements are required to clearly distinguish between subcutaneous or visceral adipose tissue derived obesity (26).

1.2.2 Fat depots and their relation to the human metabolic syndrome Constitutive biological differences in visceral and subcutaneous adipose tissue depots reflect their physiological role (27). Even though both deposits serve as energy reservoir to maintain systemic equilibrium, amongst other there are differences in levels of lipid mobilization, adipokine production and adipocyte differentiation that might be of importance e.g. in responses to diet and exercice (Table 1) (26,27).

CHAPTER 1 27

Table 1: Comparison of subcutaneous and visceral adipose tissue. Visceral > Subcutaneous adipose tissue Subcutaneous > Visceral adipose tissue Protein & Enzyme Activity • Insulin receptor • Insulin receptor affinity • Lipolytic capacity • IRS-1 • Anti-lipolytic insulin effect • Insulin sensitivity Released Factors • PAI-1 • Leptin • Angiotensinogen • Hormone sensitive lipase • IL-6 • Adiponectin • RBP4 IRS-1 insulin receptor substrate 1; PAI-1 plasminogen activator inhibitor; IL-6 interleukin-6; RBP4 retinol binding protein 4. The table is not exhaustive (26,28).

During positive caloric balance there are two factors important for the development of metabolic disease: how the fat is stored (hypertrophy vs. hyperplasia) and where the fat is stored (SC vs. Vis). First, although the amount and distribution of adipose tissue is associated independently with insulin resistance or T2D and other metabolic disorders, the size of adipocytes within the adipose tissue remains crucial (26,31). Enlarged adipocytes are characterized by an increased lipolytic state and are resistant to the anti-lipolytic effects of insulin (32). Second, the link between fat mass, fat distribution and insulin resistance seems to be a consequence of increased ectopic lipid accumulation (33). A basic hypothesis suggests a limited storage capacity of the white, predominantly subcutaneous adipose tissue, when delivered energy is too high, which leads to the visceral/ectopic lipid accumulation. Unger and McGarry postulated a thesis of lipo-toxiticity, i.e. that an intracellular lipid accumulation leads to a cell dysfunction such as insulin resistance e.g. in adipocytes, hepatocytes, or skeletal muscle cells (34,35). Further, secretion of adipocytokines has been shown particularly for Vis fat (Table 1) (36-38). It is evident that many of these adipokines have the ability to influence other tissues such as the liver, muscle and brain, e.g. the adipokine leptin affects regulation, others have an important impact on the consequences of adipose tissue inflammation (e.g. IL-6, PAI-1, monocyte chemoattractant protein 1 [MCP-1]) and vascular biology (e.g. serum amyloid A [SAA]) (39-43). Another hypothesis postulates that an ectopic accumulation in organs such as liver could also occur as a consequence of genetic factors and/or conditional upon nutrition lipid synthesis, which leads to an increase of Vis or peripheral fat. It could be shown that the intracellular lipid content in the liver is associated with insulin resistance (44,45). This “non alcoholic ” (NAFLD) has a heterogenic clinical picture with a wide range of histological findings and fit in the hepatic CHAPTER 1 28 manifestation of the metabolic syndrome (33). Visceral adipose tissue accumulation is the fat depot most characterized as being associated with an increased risk of metabolic disease (31,46). However, altered fat distribution may have an even stronger effect on the development of components of the metabolic syndrome than the presence of extreme obesity (31).

1.3 Genetics of obesity and adipose tissue distribution 1.3.1 Genetics of obesity Measures of obesity belong to the most strongly heritable human traits (47,48). Family and twin studies have confirmed that genetic factors are likely to be responsible for 45-75 % of the inter-individual variation in BMI (47). Identifying “obesity genes” is an attractive and important way of teasing out some of the causal factors leading to obesity (49). Most notably rodent studies led to the discovery of candidate genes responsible for monogenic forms of obesity (50-53). Mutations in the leptin/leptin receptor (LEP/LEPR) gene, the proopiomelanocortin (POMC) gene, the gene for prohormone convertase 1 (PCSK1), or the melanocortin receptor 4 (MC4R) gene are well known to cause severe obesity already in early childhood (54). Although research on monogenic forms of obesity could add to the knowledge of body weight regulation, they are relatively rare within populations, and explaining only a small percent of the overall disease burden. It is obvious that the pathophysiology of obesity is much more complex, and for most individuals the genetic predisposition to obesity has a polygenic basis (55). Sizeable phenotypic effects arise only in combination with other predisposing variants since a polygenic variant by itself has a small effect on the phenotype. Association studies and genome wide approaches were applied to detect new obesity genetic variants. Variants e.g. in the genes for tumor necrosis factor alpha (TNFA) (56), uncoupling protein 2 (UCP2) (57) or peroxisome prolifereator-activated receptor gamma (PPARG) (58) were detected due to the biological, physiological or pharmacological evidence for the role of the gene in obesity. However, such classical candidate gene studies are exhausting and their contribution towards understanding the genetics of obesity is rather limited. Advances in high-throughput genotyping technology has led to the development of genome-wide association studies (GWAS), which are providing novel and important insights into disease processes (59) and help to pinpoint the genetic architecture of complex diseases. One of the first common genetic variants associated with adult and found through a genome wide approach was a variant near the insulin induced gene 2 (INSIG2) (60). Replication in large independent cohorts revealed CHAPTER 1 29 contrary results depending on the ethnicities of the investigated study population (60-65). Liu et al. found strong associations with BMI and fat mass in catenin, beta like 1 (CTNNBL1), a gene which might seem to play a role during adipocyte differentiation (66). Beside common variants near MC4R (67), the so far strongest GWAS-derived, and robustly replicated signal was found for variants in the fat mass and obesity associated gene, FTO (68-71). However, until 2008 none of the candidate genes/genomic regions identified has been found to explain > 10 % of variation in any obesity phenotype (66). In the last two years a “second sweep” of GWAS revealed round about 15 additional loci influencing BMI, amongst others genes which play a role in the central nervous system or specific in the hypothalamus suggesting a role in controlling food intake and energy balance (e.g. neuronal growth regulator 1, NEGR1; transmembrane protein 18, TMEM18; or SH2 domain binding protein 1, SH2BP1) (72-75). Since it has been shown that a certain pattern of adipose tissue distribution is a predictor of metabolic complications, genetic studies are currently focusing on identification of genes, which seem to contribute to the fat distribution pattern and/or whose function is modified in the adipose tissue depots from lean and obese non-diabetic, insulin resistant, or diabetic patients.

1.3.2 Genes and gene expression pattern of adipose tissue distribution Definition of the mechanism involved in the regulation of fat distribution in general, and visceral fat mass in particular, is a key for understanding obesity and its accompanying morbidity and mortality (76). In 2009, a large GWAS revealed three new loci associated with either waist circumference (WC; near transcription factor activating enhancer binding protein 2 beta, TFAP2B; methionine sulfoxide reductase A, MSRA) or gender-specific with WHR (near lysophospholipase-like 1, LYPLAL1) (76). TFAP2B genetic variants have also been shown to be associated with T2D in Japanese patients and in the UK population previously (77), and have been reported to correlate positively with TFAP2B transcript levels in adipose tissue (78). This leads to the suggestion that the link between genetic variants in TFAP2B and the pathophysiological outcome of T2D might be mediated through an effect of these variants on the adipose tissue distribution, and therefore provides possibly a cause-effect relationship. However, WHR and WC were scarcely addressed in genome wide and candidate gene association studies so far. Noteworthy, in context of the specificity of many critical biological processes and given the fact that most common diseases appear to result from a complex interaction between CHAPTER 1 30 various genetic loci and the environment, it is evident that studying the genetics of gene expression in cells representing the in vivo state provides strong advantages (75,79). There is emerging evidence that depot-specific differences exist in the expression of genes coding important functional proteins in adipocytes, like the metabolic enzyme and related signalling proteins, the adipogenic factors, and the products of adipocytes (Figure 3) (80). This may contribute to the well-known specific functional properties of the adipocytes from intra- abdominal and subcutaneous regions. However, in obese subjects or subjects with T2D, the differential expression of genes in the appropriate adipose tissue depot may simply reflect the pathophysiological state rather than providing a primary cause for it.

Visceral > Subcutaneous adipose tissue Gene Expression Subcutaneous > Visceral adipose tissue

HOXA5 ↑ CNR1 ↓ PPARG↑vis AR TBX15 ↓vis

FASN ↑ GPC4 ↓sc ↑vis

Vaspin BMI>25 kg/m2 UCP1, UCP2 ↓ NR3C1 GLUT4 RBP4 ↑ GYS1 visceral fat subcutaneous fat Adrenoreceptor B , B B 1 2, 3

Figure 3: Genes differentially expressed in subcutaneous and visceral adipose tissue and changes during obesity. ↑ increased expression in adipose tissue in obesity; ↓ reduced expression in adipose tissue in obesity; vis/sc behind the arrow specifies the fat depot; PPARG peroxisome proliferator- activated receptor gamma; GPC4 glypican 4; UCP1/2 uncoupling protein 1/2; GLUT4 solute carrier family 2 member 4 (glucose transporter); GYS1 glycogen synthase; TBX15 T-box 15; CNR1 cannabinoid receptor 1; AR androgen receptor; HOXA5 homeobox A5; FASN fatty acid synthase; NR3C1 glucocorticoid receptor; RBP4 retinol binding protein 4. The figure is not exhaustive. (26,28,81-85).

As shown by Lefebvre et al., there was a decreased PPARG mRNA level in Vis adipose tissue in subjects with a BMI < 30 kg/m2, but not in obese subjects. This indicates that relative PPARG expression is increased in omental fat in obesity which let them conclude that altered expression of PPARG might play a role in adipose tissue distribution and expansion (86). Angiotensinogen which was implicated in the pathogenesis of metabolic alterations and hypertension associated with obesity was shown to be higher expressed in Vis than in SC adipose tissue in obese patients (87). Further, gene expression studies in SC and Vis adipose tissue revealed a correlation of differential expressed genes with measures/factors of the human metabolic syndrome like BMI/obesity, insulin resistance/T2D which might exhibit new genetic candidates and were partly unexpected such as developmental genes (83). CHAPTER 1 31

Genotype-phenotype association studies in population based studies showed BMPR1A (bone morphogenetic protein receptor 1A) genetic variants being associated with obesity and obesity-related quantitative traits (88). Moreover, an association of these variants with the Vis BMPR1A mRNA expression could be detected suggesting a link between genetic predisposition and phenotypic outcome. Also, miRNAs, small non-coding RNAs, which play an important regulatory role in a variety of biological processes, have been shown to be correlated with adipose tissue morphology and key metabolic parameters such as Vis fat area, circulating leptin, adiponectin or interleukin-6. Further it could be shown that they reveal a significant fat depot-specific expression pattern suggesting a possible epigenetic influence on adipose tissue distribution (89). Identifying and clarifying the role of genes in fat distribution may therefore contribute to the understanding of the molecular mechanisms underlying the genetic susceptibility for the human metabolic syndrome.

1.4 Aims of the presented work Obesity plays a key role in the development of the metabolic syndrome. Pathophysiological alterations in metabolism following the increase of fat mass such as insulin resistance, dyslipidemia or T2D, are thereby closely connected to the distribution pattern of the adipose tissue. Omental, especially Vis adipose tissue has been suggested to be the risk promoting part since adipocytokines and proinflammatory factors secreted by visceral adipocytes have been shown to be linked to the phenotypic outcome of the metabolic syndrome. The underlying molecular mechanisms that relate a certain adipose tissue distribution pattern to metabolic complications are poorly understood. With the exception of some cases of monogenic disorders obesity and adipose distribution has a polygenic background. Although candidate gene and GWAS identified numerous genes and susceptibility variants, it might be assumed that they do not explain more than 20 % of the genetic variability. Further studies are required to detect new candidate genes and also to elucidate their role in the genetics of obesity and/or adipose tissue distribution and associated metabolic complications resulting in a metabolic syndrome. To meet this challenge a combinatory approach unifying several strategies has been applied. First, the expression of genes, for which genetic variants could be found to be associated with obesity or related traits, is investigated in the SC and Vis adipose tissue depots under specific conditions (lean and obese non-diabetic subjects or subjects with T2D) to detect possible differences associated with the phenotypic outcome (Chapter 3). Second, based on the fact that conspicuous differential expressed genes could comprise putative CHAPTER 1 32 susceptibility variants modulating the metabolic risk, genotype-phenotype association studies are carried out (Chapter 4-6). The main method used for genotyping is described in detail in a comprehensive review providing explicit troubleshooting and description of modified protocols for specific experimental needs (Chapter 2). Ultimately, the link between expression and genetic susceptibility variants is examined which could expose a piece of the jigsaw on the underlying molecular mechanisms. In particular, this thesis deals with FTO (Chapter 3), FASN (Chapter 4), CNR1 (Chapter 5) and BMPR2 (Chapter 6) as possible candidates playing a role in adipose distribution and obesity risk in German populations. Chapter 3 answers the questions whether FTO and the neighboring gene RPGRIP1L are fat depot-specificly expressed in subcutaneous and visceral adipose tissue of lean and obese nondiabetic subjects and subjects with T2D. Further, it is investigated whether the mRNA expression of FTO and/or the neighbouring RPGRIP1L in adipose tissue correlates with measures of obesity and fat distribution. Finally the association of the FTO obesity risk allele with FTO and RPGRIP1L mRNA expression is examined. Chapter 4 clarifies the role of genetic variation in FASN on obesity and related traits in adult and children cohorts of white European ancestry. The effect of FASN SNPs on FASN mRNA expression in the SC and Vis adipose tissue depot is elucidated. Chapter 5 investigates the contribution of genetic variants in CNR1 to the polygenetic background of obesity. Moreover, the question whether these variants might explain expression differences of CB1R in SC and Vis adipose tissue is addressed. Chapter 6 gives insights to the mRNA expression of BMPR2 in adipose tissue depots and points out the correlations of BMPR2 expression with measures of obesity and related traits. Moreover, the association of BMPR2 genetic variants with anthropometric and metabolic parameters is highlighted, and the link between BMPR2 genetic variants and mRNA expression investigated. Chapter 7 provides critical discussion of the presented studies. In summary, this thesis led to the discovery of novel genetic variants which account for the polygenic mediated obesity risk in German populations. Furthermore, the study provides evidence that some of these variants might also modulate the gene expression level in adipose tissue depots leading to a shift towards an adverse metabolic constellation which could provide a link between genotype and the disease phenotype. CHAPTER 2 33

CHAPTER 2

Targeted SNP Genotyping Using the TaqMan® Assay In: Disease Gene Identification: Methods and Protocols

Dorit Schleinitz1, Johanna K. DiStefano2, Peter Kovacs1

1Interdisciplinary Centre for Clinical Research, Leipzig, Germany 2Diabetes, Cardiovascular and Metabolic Diseases Division, Translational Genomics Research Institute Phoenix, Arizona, USA

Series: Methods in Molecular Biology, 2011 [in press] CHAPTER 2 34

2.1 Abstract More than 99 % of genomic DNA sequence is identical among humans, and not surprisingly, slight variations in sequence can often produce a major effect on phenotype. Sequence variants may also mediate the manner in which humans are susceptible to disease or respond to environmental factors such as bacteria, viruses, toxins, chemicals, drugs, and therapeutic interventions. Single nucleotide polymorphisms (SNPs) are DNA sequence variations that occur when a single base in the genome sequence can be represented by at least two different nucleotides. In the last decade, numerous SNPs have been identified that explain, at least partially, the genetic architecture of complex diseases such as cancer, diabetes, vascular complications, some forms of mental illness, and a multitude of other disorders. Disease-related SNPs are commonly identified through candidate gene approaches, or more recently, through genome-wide association studies (GWAS). In either case, findings of association require verification in independent, population-based, study samples, usually comprised of several hundreds/thousands of individuals. A convenient technique to genotype a moderate number of markers in this kind of study is available with the TaqMan® platform (Applied Biosystems; Foster City, CA), which utilizes polymerase chain reaction (PCR) amplification and allelic discrimination to easily and efficiently generate genotype data in a cost-effective way. Here we introduce and describe this commonly used technique and include protocols which can be directly used in laboratories aiming to perform moderate- to large-scale genotyping studies.

2.2 Introduction The dimorphic SNP, which occurs when a single nucleotide in the genome sequence is altered, is the most common form of (1,2). In any two genome copies, SNPs are reported to occur as frequently as one variant per kilobase, or at a frequency of 0.1 % (3). Although most of these SNPs are neutral, some may affect the gene/protein function and thus lead to phenotypic differences between carriers of various . These phenotypic differences can result in enhanced susceptibility to various diseases or response to environmental factors. It has been estimated that a total of 10 million SNPs exist in the human population (equivalent to one variant per 300 bases) such that both alleles are observed at a frequency of ≥ 1 %, and that these 10 million common variants constitute 90 % of the variation in the population (4). At this frequency, SNPs are potentially valuable both for aiding biomedical research and developing pharmaceutical products or medical diagnostics.

CHAPTER 2 35

In the last decade, it has been shown that SNPs may also help to explain the genetic architecture of complex diseases. SNPs can contribute either directly or indirectly to the development of both monogenic and polygenic diseases, such as obesity (5,6), psychiatric illness (7) or cancer (8,9), and may mediate individual risk in disease severity and progression, as well as response to pharmacological treatments (2,10,11). Genetic studies utilize the candidate gene or genome-wide approach to identify such SNPs. Particularly, genome-wide association studies have resulted in identification of novel disease-related SNPs, which increased the number of known “pathological” SNPs for various complex diseases enormously. Following identification of disease-associated SNPs in candidate gene or genome-wide studies, further investigation and verification of these potential risk alleles is required using independent population-based samples. Because such study samples are typically comprised of several hundreds or thousands of subjects to achieve sufficient statistical power, large-scale screening for known polymorphisms necessitates techniques with a minimal number of steps and an ability to automate each one (12). Because re- sequencing, while thorough, is still costly and time-consuming, targeted SNP genotyping using TaqMan® assays based on a polymerase chain reaction (PCR) provides a more economic and easy to handle procedure to efficiently generate thousands of genotypes for genetic association studies. The TaqMan® SNP Genotyping Assay (Applied Biosystems; Foster City, CA) is used to discriminate between two alleles of a specific SNP for use in genotyping studies. The method involves use of forward and reverse primers to amplify the polymorphic sequence of interest, plus two dye-labelled probes for allele-specific detection. At the 5' site, one probe is labelled with VIC® dye, which detects the “Allele 1” sequence, while the other one is labelled with 6-FAM™ dye (6’-carboxyfluorescein), which detects the “Allele 2” sequence (Figure 1). VIC® and 6-FAM™ are also referred to as reporter dyes. Based on the principle of fluorescence resonance energy transfer (FRET), at the 3' site the probes are labelled with a non-fluorescent quencher (NFQ) which absorbs the energy of the reporter dyes as long as both are bound to the probe and therefore, are in spatial proximity. Further, TaqMan® probes incorporate a minor groove binder (MGB) at the 3' end (Figure 1). The MGB molecule

(e.g., dihydrocyclopyrroloindole tripeptide [DPI3]) binds to the minor groove of the DNA helix, and improves hybridization-based assays by stabilizing the probe/template complex. In comparison with unmodified DNA, MGB probes have higher melting temperatures and increased specificity (13). The increased binding stabilization permits the use of probes as

CHAPTER 2 36 short as 13 bases for improved mismatch discrimination and greater flexibility when designing assays for difficult or variable sequences (14).

Figure 1: Principle of allelic discrimination using TaqMan® SNP Genotyping Assays. 1. Assay components and DNA template: Forward and reverse primers to amplify the polymorphic sequence of interest, two 5' site dye-labelled probes for allele-specific detection. Fluorescence is quenched by a non-fluorescent quencher (NFQ) at the 3' site. A minor groove binder at the 3' end stabilizes the probe/template complex. 2. Denatured template and annealing assay components. 3. Polymerization and signal generation: Taq polymerase starts to synthesize a new strand and encounters the annealed probe. The 5'-bound fluorescence dye of the appropriate probe is separated (5'→3' exonuclease activity of Taq polymerase). Fluorescence signal is no longer quenched and can now be detected with laser excitation. By courtesy of Applied Biosystems LLC (part of Life Technologies Group).

Following denaturation of the template and annealing of the assay components, the Taq polymerase starts to synthesize a new strand. During the reaction, the polymerase encounters the annealed probe and the 5'-bound fluorescence dye of the appropriate probe will be separated as a result of the 5'→3' exonuclease activity of the Taq enzyme (Figure 1). At this point, the fluorescence signal is no longer quenched and can now be detected with laser excitation. Depending on the genotype, one will detect homozygotes for either Allele 1 or 2, or heterozygotes that carry both alleles (Figure 2).

CHAPTER 2 37

The TaqMan® SNP Genotyping Assay is available from Applied Biosystems, Inc. (Foster City, CA). The manufacturer provides detailed product and user guides on their homepage, www.appliedbiosystems.com. In this chapter, we focus on these main steps, but also offer modifications and suggestions to improve assay efficiency and genotype data quality that can be easily incorporated in any laboratory.

2.3 Materials 2.3.1 Target DNA 1. In general, 1-20 ng purified genomic DNA is recommended for TaqMan® SNP Genotyping Assays (see Note 1).

2.3.2 Plastics 1. 96- or 384-well plates, suitable for the thermocycler of choice and the ABI real-time System (see Section 2.3.4 Instruments and Software) 2. Seal mats; optical adhesive, thermostable covers or optical caps

2.3.3 Genotyping reaction 1. TaqMan® SNP Genotyping Assay (Applied Biosystems; 20x, 40x or 80x concentrated) containing sequence-specific forward and reverse primers and two TaqMan® MGB probes. The assay should be stored at -15 to -20 °C and protected from light. 2. 2× TaqMan® Genotyping Master Mix (Applied Biosystems) containing AmpliTaq Gold® DNA Polymerase ultra pure, buffer, dNTPs, ROX™ dye (6-carboxy-X-rhodamine, passive internal reference) and should be stored at 2 to 8 °C. Alternatively, genotyping mastermix is provided by companies like Thermo Fisher Scientific Inc. (ABgene®), Peqlab (Sensi Mix), Eurogentec (qPCR kits for probe assay). 3. DNase-free water

2.3.4 Instruments and software 1. Thermocycler: the following instruments are specified by the company: GeneAmp® PCR System 9700, Applied Biosystems 9800 Fast Thermal Cycler, using the 9700/9600 emulation mode (see Note 2). 2. Real-time PCR System - the following instruments are specified by the company: ABI PRISM® 7000 Sequence Detection System, Applied Biosystems (AB) 7300/7500 Real-Time PCR System, AB 7500 Fast Real-Time PCR System, AB 7900HT Fast Real-Time PCR

CHAPTER 2 38

System. Alternatively to ABI manufactured equipment several other companies offer platforms compatible with TaqMan® genotyping assays (e.g. LightCycler ® Systems for Real-Time PCR from Roche). 3. Sequence detection system (SDS) software (Applied Biosystems); Primer Express® Software for Real-Time PCR 3.0 (Applied Biosystems).

2.4 Methods Applied Biosystems provides the largest collection of ready-to-use SNP assays (TaqMan® SNP Genotyping Assays; TaqMan® Drug Metabolism Genotyping Assays; TaqMan® SNP Genotyping Assay Sets; TaqMan® Pre-Developed Assay), and most of them apply to use with human DNA. An assay search tool is available on the company’s homepage (https://products.appliedbiosystems.com/ab/en/US/adirect/ab?cmd=catNavigate2&catID=601 277). After choosing the gene or region of interest and selecting SNPs for genotyping, it is easy to check the availability of assays. Applied Biosystems also offers custom design assays (Custom TaqMan® SNP Genotyping Assays) and provides the Primer Express® Software for individual design for markers not available in the manufacturer’s database.

2.4.1 Preparation of samples to perform SNP genotyping assays 1. For better handling, we recommend preparing “master” plates with defined sample positions, plate names, and a consistent concentration and volume of genomic DNA per sample (e.g. 10 ng/µl). The total volume of DNA will depend on the type of plates used in the experiment (see Note 3). 2. Aliquot the appropriate amount of genomic DNA, secure the master plates with seal mats, and freeze at -20 °C until ready to use. 3. When ready to begin genotyping, remove master plates from -20 °C and let thaw. When thawed, vortex and centrifuge briefly to bring liquid to the bottom of the wells. 4. Disperse the appropriate volume (e.g. 1 µl of 10 ng/µl concentrated DNA) to each well, seal plates with covers (mats, tapes, caps, etc.) and use directly or store at -20 °C until ready to use.

2.4.2 Preparation of master mix for SNP genotyping 1. Keep all reagents on ice during the course of the experimental set-up.

CHAPTER 2 39

2. Prepare TaqMan® genotyping master mix as shown in Table 1. The amounts of reagents can be varied according to the number of plates assayed; Table 1 gives reagent amounts needed to genotype one 96-well plate (see Note 4). 3. Pipette 24 µl of the master mix to each well of a 96-well plate. 4. Cover plate with an optical, thermostable, adhesive cover (or optical caps) and place plate in thermocycler.

Table 1: Mastermix for 96-well genotyping reaction. Volume for 100 reactions Reaction component Volume (µl/reaction) (one 96 well plate) 2× Genotyping master mix 12.5 1.25 ml 20× SNP genotyping assay 1.25 125.0 µl DNAse free water 10.25* 1.025 ml Template 1.0* Total 25.0 2.4 ml

*variable

2.4.3 Cycling program for genotyping reaction (see Note 5) 1. Because the analytical assays represent an endpoint analysis, the target amplification does not need to be carried out in the Real-Time PCR System. That is, a standard thermocycler can be used for amplification, and only the ultimate plate readout requires the AB Real-Time PCR System. 2. Briefly centrifuge plates to collect liquid to the bottom of the wells. 3. Place plate in thermocycler. 4. Program thermocycler as shown in Table 2. 5. Begin PCR amplification; the cycle duration should take approximately 2 hours.

Table 2: PCR cycling program. Step Temperature Time Cycle Activation of AmpliTaq Gold enzyme 95 °C 10 min 1 Denaturation 92 °C 15 sec Annealing/Extension 60 °C 1 min }40 4 °C or Cool down ∞ 10 °C (over night)

2.4.4 Plate readout – allelic discrimination (see Note 6) 1. After amplification, briefly centrifuge plates to collect liquid to the bottoms of the wells. Perform plate readout directly or store processed plates at -20 °C until ready to read.

CHAPTER 2 40

2. Start the detection system first, and then open the SDS software. 3. Create a new document and select “Allelic Discrimination” in the assay drop-down list. Click next to access the “Select Markers” page. 4. If the marker list contains a marker suitable for your application skip to step 5. If the marker list does not contain a marker (set of two detectors discriminating between different alleles of a common locus) suitable for the application, create detectors (virtual representation of a TaqMan® probe and primer set and associated fluorescent dye that detects a single target nucleic acid sequence) and marker. 4a. Click “New Detector”. In the new detector dialog box, type “Allele 1” for name. Leave the reporter dye set to Vic and quencher dye “none”, choose a colour, and then click OK. Click Create another, for name type “Allele 2” and select FAM for the reporter dye, and quencher dye “none”. Select purple, then Click OK. 4b. Click “New Marker”. In the marker dialog box, type a name, select the Allele 1 and Allele 2 detectors you created above. Then click OK. 5. In the “Select Markers” window, select the marker for your application, then click “Add >>”. Select the passive reference depending on the genotyping mastermix you use (in most cases Rox). Click next. 6. In the setup sample plate page, select the marker for wells: Click-drag to select wells, select the “Use” box for the marker. Click the “task” field for one of the detectors and select “NTC” for none template controls or “Unknown” for samples. This determines the way the SDS software uses the data collected from the well during analysis. 7. Click finish. Now, with the help of the well inspector (“View > Well Inspector”) it is possible to enter sample names, although this step is not necessary if a masterplate setup is available (see Section 3.1) 8. Select the “Instrument” tab. After creating and setting up an allelic discrimination plate read document in the SDS software, the plate read is performed at 60 °C. Change the sample volume to correspond to the reaction volume (see Section 2.4.2). 9. Save the file (“File > Save”). Be sure that the file name specifies which samples/cohort/SNPs were used for the experiment. Then click “Post-Read”. 10. After the run, click the analysis button to start genotype read. Normally, the analysis of the plate read document performs automatically but it also offers an option for alleles being called manually (see Note 7). An example for allele calling is presented in Figure 2. Ultimately, the allele calls need to be converted to genotypes (see Note 8). The appropriate allele labelling is included with every assay provided by the company.

CHAPTER 2 41

Figure 2: SNP allele detection after PCR on an ABI PRISM® 7500 Sequence Detection System. Blot of the fluorescence signals (Rn) obtained during the plate read for each well indicating the alleles in each sample. Depending on genotype, homozygotes for either Allele 1 or 2, or heterozygotes that carry both alleles will be detected. NTC = non template control. By courtesy of Applied Biosystems LLC (part of Life Technologies Group).

2.5 Notes 1. Depending on the quality of starting material, the best DNA extraction method needs to be individually determined by the researcher. We strongly recommend quantifying genomic DNA using spectrophometry and assessing integrity using agarose gel electrophoresis. The purified DNA (stock or diluted for final concentration) should be stored at ≤ -20 °C. Biological samples such as tissues, body fluids, infectious agents, and blood have the potential to transmit infectious diseases. Follow all applicable local, state/provincial, and/or national regulations. Wear appropriate protective equipment, which includes, but is not limited to: protective eyewear, face shield, clothing/lab coat, and gloves. All work should be conducted in properly equipped facilities using the appropriate safety equipment.

2. While ABI recommends specific instruments, we have found that any conventional thermocycler can be used in this protocol. In our experience, it is rare that the thermal cycler

CHAPTER 2 42 does not perform adequately for the assay; however, because the potential exists, we recommend testing each TaqMan® probe in several plates on the thermocycler prior to using in a large-scale genotyping study.

3. Non-template controls (NTCs, such as water) are included in the plate set-up for internal quality control purposes. We also highly recommend inclusion of positive controls, such as DNA samples with known genotypes (determined by a different method), as well as quality controls (QCs) representing blinded duplicates to the genotyped samples, to validate the reproducibility and accuracy of the assay. Including positive controls for each allele combination are particularly important in experiments with small sample sizes (e.g., N < 50) or for variants with rare minor allele frequencies or a low number of heterozygotes.

4. In our experience, the assays are typically easy to handle and the amplification reaction is stable. Therefore, it is sometimes possible to reduce the genotyping reaction volume, which may help to reduce genotyping costs and preserve DNA samples. Considering the costs and time required for other techniques, such as sequencing, this protocol is efficient for sample sizes of 50 up to several thousand (use of different semiautomatic/automatic pipette solutions may be advantageous).

5. Applied Biosystems provides an “Allelic Discrimination (AD) Getting Started Guide” for the AB real-time PCR System in which one can find detailed information about performing AD pre-run (although in our experience, this is not essential when working according to good laboratory practices [GLP]), amplification run, and AD post-read run. The authors strongly recommend reading this guide prior to beginning the TaqMan® genotyping experiment.

6. The PCR amplification reaction is followed by the endpoint analysis for allelic discrimination using an Applied Biosystems real-time PCR System. The SDS software uses the fluorescence measurements obtained during the plate read to plot fluorescence (Rn) values based on the signals from each well. The plotted fluorescence signals indicate the alleles in each sample (TaqMan® SNP Genotyping Assays Protocol, Applied Biosystems 2006).

7. Most of the TaqMan® Assays are robust, but we nevertheless recommend processing 1-2 test plates per genotyping assay and cohort. By doing so, one will determine how the assay performs and how frequent the minor allele is. However, it may occur that the TaqMan®

CHAPTER 2 43

Assays do not perform well (i.e., the plate readout is difficult or not possible at all). Commonly, this is due to poor DNA quality (e.g., extracted from serum or small amounts of tissue which leads to low concentration and poor quality). Also, if the DNA concentration is not consistent between wells across the plate, the TaqMan® Assay may not perform optimally. Check and improve these factors if necessary and test again. If these problems are still encountered, add Taq polymerase to the master mix to increase the yield of PCR and run the plate for an additional 10 cycles. One can also test genotyping master mixes offered by other companies (please be aware of potential differences in PCR cycling conditions when using various master mixes) or change the thermal cycler. If any initial setting has been changed, start the experiment again with just one or two plates to check whether it performs differently.

8. For data analyses use suitable programs, for phenotype-genotype association studies, e.g., SPSS (SPSS, Inc.; Chicago, IL, USA) or Graphpad Prism® (GraphPad Software, Inc.; La Jolla, CA, USA). For statistical calculations, we recommend coding the SNP alleles: 11 or 22 for homozygotes for either Allele 1 or 2, or 12 for heterozygotes that carry both alleles and to convert them to their respective nucleotides afterwards.

2.6 References 1. Cargill M, Altshuler D, Ireland J et al. Characterisation of single-nucleotide polymorphisms in coding regions of human genes. Nat Genet 1999 22:231-238.

2. Tindall EA, Speight G, Petersen DC, Padilla EJD, Hayes VM. Novel Plexor™ SNP Genotyping Technology: Comparisons With TaqMan® and Homogenous MassEXTEND™ MALDI-TOF Mass Spectrometry. Hum Mutat 2007 28:922-927.

3. The International HapMap Consortium. International HapMap Project. Nature 2003 426:789- 796.

4. Kruglyak L, Nickerson DA. Variation is the spice of life. Nat Genet 2001 27:234–236.

5. Frayling TM, Timpson NJ, Weedon MN et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 2007 316:889-894.

6. Mergen M, Mergen H, Ozata M, Orner R, Orner C. A novel melanocortin 4 receptor (MC4R) gene mutation associated with morbid obesity. J Clin Endocrinol Metab 2001 86:3448.

7. Stefansson H, Ophoff RA, Steinberg S et al. Common variants conferring risk of schizophrenia. Nature 2009 460:744-747.

8. Eeles RA, Kote-Jarai Z, Giles GG et al. Multiple newly identified loci associated with prostate cancer susceptibility. Nat Genet 2008 40:316-21.

CHAPTER 2 44

9. Stacey SN, Manolescu A, Sulem P et al. Common variants on chromosomes 2q35 and 16q12 confer susceptibility to estrogen receptor-positive breast cancer. Nat Genet 2007 39:865-869.

10. Becker KG. The common variants/multiple disease hypothesis of common complex genetic disorders. Med Hypotheses 2004 62:309-17.

11. Evans WE and McLeod HL. Pharmacogenomics-drug disposition, drug targets, and side effects. N Engl J Med 2003 348:538-549.

12. Livak KJ, Marmaro J, Todd JA. Towards fully automated genome-wide polymorphism screening. Nat Genet 1995 9:341-342.

13. Kutyavin IV, Afonina IA, Mills A et al. 3’-Minor groove binder-DNA probes increase sequence specificity at PCR extension temperatures. Nucleic Acids Res 2003 28:655-661.

14. Applied Biosystems. TaqMan® SNP Genotyping Assays Product Bulletin. 2008.

2.7 Contribution of Authors Dorit Schleinitz Design and writing of the book chapter

Johanna K. DiStefano Proofreading and editing the manuscript

Peter Kovacs Proofreading and editing the manuscript

CHAPTER 3 45

CHAPTER 3

Inverse Relationship between Obesity and FTO Gene Expression in Visceral Adipose Tissue in Humans

Nora Klöting1*, Dorit Schleinitz2*, Karen Ruschke1, Janin Berndt2, Mathias Fasshauer1, Anke Tönjes1, Michael R. Schön4, Peter Kovacs2, Michael Stumvoll1 and Matthias Blüher1

1Department of Medicine, University of Leipzig, Leipzig, Germany 2Interdisciplinary Centre for Clinical Research, Leipzig, Germany 3Städtisches Klinikum Karlsruhe, Clinic of Visceral Surgery, Karlsruhe, Germany

* Nora Klöting and Dorit Schleinitz contributed equally to this study

Published in: Diabetologia 2008 51:641-647

CHAPTER 3 FTO 46

3.1 Abstract Aims/hypothesis Recently, FTO was identified as a candidate gene contributing to both childhood and severe adult obesity. We tested the hypothesis that mRNA expression of FTO and/or of the neighbouring RPGRIP1L in adipose tissue correlates with measures of obesity and fat distribution. We also investigated whether the FTO obesity risk alleles might explain variability in FTO and RPGRIP1L mRNA expression. Methods In paired samples of visceral and subcutaneous adipose tissue from 55 lean and obese participants, we investigated whether FTO and RPGRIP1L mRNA expression is fat depot-specific, altered in obesity and related to measures of fat accumulation, insulin sensitivity and glucose metabolism. All participants were genotyped for the obesity-associated rs8050136 FTO variant. Results FTO mRNA expression was threefold higher in subcutaneous than in visceral adipose tissue. Subcutaneous FTO expression correlated with visceral FTO expression. FTO gene expression in both depots correlated with age and was negatively correlated to BMI and percent body fat. FTO mRNA levels were not related to measures of insulin sensitivity and glucose metabolism. RPGRIP1L mRNA expression was 1.6-fold higher in visceral than in subcutaneous adipose tissue, but did not correlate with anthropometric and metabolic characteristics. There was no association between rs8050136 and FTO or RPGRIP1L mRNA expression in adipose tissue. Conclusions/interpretation Expression of adipose tissue FTO mRNA is fat depot-specific and negatively correlates with measures of obesity. However, the direction of this relationship still needs to be elucidated.

3.2 Introduction Genetic factors play an important role in the development of obesity (1), but the underlying genetic causes are poorly understood (2). Two independent studies using different approaches recently demonstrated a strong association between variants in the gene fat mass and obesity associated (FTO) and obesity (3,4). Frayling et al. (3) found the association between FTO variants (represented by rs9939609) and obesity in a genome-wide scan for type 2 diabetes and replicated it in a total of more than 38,000 individuals. The association between SNPs in the FTO gene and type 2 diabetes was shown to be due to obesity (3). In agreement with these findings, Dina et al. (4) reported a set of SNPs in the FTO gene (rs1421085 and rs17817449) that is consistently strongly associated with early-onset childhood and severe adult obesity. Both groups replicated the association between FTO and obesity in large cohorts (3,4). Moreover, recently Scuteri et al. (5) replicated the significant association between rs9930506 in the FTO gene and BMI, body weight and hip circumference

CHAPTER 3 FTO 47 in European and Hispanic Americans. The Fto gene was originally described in a mouse with fused toes (Ft mutation) (6,7). Homozygous Ft mutation is embryonic lethal (6); heterozygous mice have fused toes and hyperplasia of the thymus most likely due to impaired apoptosis (7). Interestingly, obesity has not been reported in these mice. However, the Ft mutation consists of at least six deleted genes and no conclusions can be drawn from this model regarding the function of FTO and the relationship between the FTO gene and obesity (2). Human de novo duplication on chromosome 16, which includes the FTO locus, has been shown to be associated with mental retardation, obesity, dysmorphic facies and severe bilateral clinodactyly of some fingers (8). Recent bioinformatics analysis revealed that FTO shares sequence motifs with Fe2+- and 2-oxoglutarate-dependent oxygenases (9). It has also been shown that FTO localizes to the nucleus and that recombinant murine FTO catalyses the demethylation of 3-methylthymine in single-stranded DNA (9). After the association with human obesity was discovered, the name was changed to ‘fat mass and obesity associated’ (FTO) (2), but the function of the FTO gene is still unknown. FTO gene is expressed in most tissues, including adipose tissue, beta cells and brain (3). The highest FTO expression levels have been detected in brain (3,9). Functional studies are necessary to define the biological role of FTO and the major FTO-‘sensitive’ organs. We therefore tested the hypothesis that FTO mRNA expression in visceral and subcutaneous adipose tissue is linked to obesity in a cohort of individuals with wide ranges of obesity and body fat distribution. In addition to FTO, retinitis pigmentosa GTPase regulator interacting protein 1-like (RPGRIP1L, also known as KIAA1005) is also thought to be involved in the development of obesity because of its genomic position near the FTO gene (3,5). The transcription start site of RPGRIP1L is situated only 200 base pairs from the 5′ end of FTO and ~61 kb from the 47 kb interval containing the BMI associations. RPGRIP1L is ubiquitously expressed with relatively high levels in hypothalamus and islet. We therefore also examined RPGRIP1L mRNA expression in visceral and subcutaneous adipose tissue and its relation to obesity. In addition, we investigated whether previously published obesity- associated FTO variants (represented by rs8050136 in our study) might explain potential variability in FTO and RPGRIP1L mRNA expression.

3.3 Methods 3.3.1 Participants and measurement of participant characteristics Paired samples of visceral and subcutaneous adipose tissue were obtained from 55 European men (N = 26) and women (N = 29) who underwent open abdominal surgery for

CHAPTER 3 FTO 48 elective cholecystectomy or explorative laparotomy. Using oral glucose tolerance tests (OGTTs), we identified individuals with type 2 diabetes (N = 19) or normal glucose tolerance (N = 36). All participants had stable weight with no fluctuations of > 2 % of body weight for at least 3 months before surgery. Patients with malignant diseases or any acute or chronic inflammatory disease as determined by a leukocyte count > 7,000×109/l, C-reactive protein > 50 mg/l or clinical signs of infection were excluded from the study. Samples of visceral and subcutaneous adipose tissue were immediately frozen in liquid nitrogen after explantation. The study was approved by the ethics committee of the University of Leipzig. All participants gave written informed consent before taking part in the study.

3.3.2 Assays and measures of glucose metabolism and body fat content Basal blood samples were taken after an overnight fast. Plasma insulin was measured with an enzyme immunometric assay for an automated analyzer (Immulite; Diagnostic Products, Los Angeles, CA, USA). Plasma leptin levels were assessed by radioimmunoassay (Linco Research, St Charles, MO, USA). OGTT was performed after an overnight fast with 75 g standardised glucose solution (Glucodex Solution 75 g; Merieux, Montreal, QC, Canada). Insulin sensitivity was assessed with the euglycemic–hyperinsulinemic clamp method (10). Percent body fat was measured by dualenergy X-ray absorptiometry. In addition, abdominal visceral and subcutaneous fat area was calculated as previously described (11,12).

3.3.3 Analysis of human FTO and RPGRIP1L mRNA expression Human FTO and RPGRIP1L mRNA expression was measured by quantitative real- time RT-PCR in a fluorescent temperature cycler using the TaqMan® assay; fluorescence was detected on an ABI PRISM® 7000 sequence detector (Applied Biosystems, Darmstadt, Germany). Total RNA was isolated from paired subcutaneous and visceral adipose tissue samples using TRIzol (Life Technologies, Grand Island, NY, USA), and 1 μg RNA was reverse transcribed with standard reagents (Life Technologies). From each RT-PCR, 2 μl cDNA was amplified in a 26 μl PCR using a kit (Brilliant SYBR Green QPCR Core Reagent Kit; Stratagene, La Jolla, CA, USA) according to the manufacturer’s instructions. Samples were incubated in the sequence detector for an initial denaturation at 95 °C for 10 min, followed by 40 PCR cycles, each cycle consisting of 95 °C for 15 s, 60 °C for 1 min and 72 °C for 1 min. The following primers were used: human FTO 5’-CTGGCCAGTGAAAGG GTCTAAT-3’ (sense) and 5’-GGCAGCAAGTTCTTCCAAAGC-3’ (antisense); human

CHAPTER 3 FTO 49

RPGRIP1L 5’-GGTATTGGCAAGCTCAGGGTAA-3’ (sense) and 5’-TCAAGCCTCCAAG TCATCTCTGT-3’ (antisense). SYBR Green I fluorescence emissions were monitored after each cycle. Human FTO and RPGRIP1L mRNA expression was calculated relative to the mRNA expression of 18S rRNA, determined by a premixed assay on demand for human 18S rRNA (Applied Biosystems, Darmstadt, Germany). Amplification of specific transcripts was confirmed by melting curve profiles (cooling the sample to 68 °C and heating slowly to 95 °C with measurement of fluorescence) at the end of each PCR. The specificity of the PCR was further verified by subjecting the amplification products to agarose gel electrophoresis.

3.3.4 Genotyping of the rs8050136 FTO variant The SNP rs8050136 was selected for genotyping as a proxy for SNPs belonging to one linkage disequilibrium (L.D.) group including the previously reported rs9939609 (3). The choice of rs8050136 was based on the better quality of genotyping compared with rs9939609. Genotyping of rs8050136 was performed using the TaqMan® allelic discrimination assay (Applied Biosystems) on an ABI PRISM® 7500 sequence detector (Applied Biosystems). To assess genotyping reproducibility, a random ~10 % selection of the samples was re-genotyped with 100 % concordance. Genotype data were consistent with Hardy–Weinberg equilibrium (HWE) (p > 0.05).

3.3.5 Statistical analyses Data are shown as means ± SD unless stated otherwise. Before statistical analysis, non-normally distributed parameters were logarithmically transformed to approximate a normal distribution. Expression differences between visceral and subcutaneous adipose tissue were assessed using the paired Student’s t test. Linear relationships were assessed by least square regression analysis. Multivariate linear relationships were assessed by a general linear model. Associations between rs8050136 and FTO mRNA expression levels were assessed using generalized linear regression models under the additive and dominant/recessive mode of inheritance. Statistical analysis was performed using SPSS version 12.0 (SPSS, Chicago, IL, USA). We considered p values < 0.05 to be statistically significant.

3.4 Results 3.4.1 Participant characteristics The study population comprised 55 European volunteers (26 men, 29 women) and was characterized by a mean age of 61 ± 15 years, mean BMI of 34.3 ± 9.5 kg/m2 (women:

CHAPTER 3 FTO 50

37.7 ± 10.7 kg/m2; men: 30.4 ± 6.1 kg/m2, p < 0.05), mean waist circumference of 119 ± 26 cm (women: 121 ± 28 cm; men: 115 ± 23 cm) and mean body fat mass of 36 ± 11 % (women: 40.6 ± 11.1 %; men: 30.8 ± 8.4 %, p < 0.05). Analysis of 55 paired samples of visceral and subcutaneous adipose tissue revealed threefold higher FTO mRNA levels in subcutaneous than in visceral adipose tissue, while RPGRIP1L mRNA expression was 1.6-fold higher in visceral than in subcutaneous adipose tissue (Figure 1a,b). FTO and RPGRIP1L mRNA gene expression was not different between men and women in either fat depot. Moreover, subcutaneous and visceral FTO mRNA expression were significantly correlated (Figure 1c). a b

15 3

10 2 /18S rRNA/18S (AU) 5 1 FTO/18S rRNA (AU) rRNA FTO/18S

0 RPGRIP1L 0 Vis SC Vis SC c d

e

[Figure 1 a-e]

CHAPTER 3 FTO 51

Figure 1: FTO mRNA and RPGRIP1L mRNA expression in individuals with wide ranges of obesity, body fat distribution, insulin sensitivity and glucose tolerance. Men, white bars/symbols, N = 26; women, black bars/symbols, n=29. a) FTO mRNA and b) RPGRIP1L mRNA expression in visceral (Vis) and subcutaneous (SC) adipose tissue. Values are means ± SEM. p = 0.01. c) Correlation between Vis and SC FTO mRNA expression (N = 55). p = 0.016; r2 = 0.11. d) Correlations between visceral and e) SC FTO mRNA expression and BMI. P < 0.001; r2 = 0.27 (d); p = 0.018; r2 = 0.11 (e). Human FTO and RPGRIP1L mRNA expression was calculated relative to the mRNA expression of 18S rRNA. Data were ln-transformed to achieve normal distribution. AU, arbitrary units.

3.4.2 Association of FTO expression Visceral FTO mRNA was significantly correlated with BMI (Figure 1d), percent body fat, sex and age, but not with visceral fat area (Table 1). Univariate regression analyses revealed further significant correlations between subcutaneous FTO mRNA expression and BMI (Figure 1e), percent body fat, subcutaneous fat area, age, sex and leptin (Table 1). The relationship between FTO mRNA expression and BMI was significant independently of sex.

Table 1: Univariate linear regression between anthropometric and biochemical parameters and visceral and subcutaneous FTO mRNA and RPGRIP1L mRNA expression. Parameter FTO/18S rRNA RPGRIP1L/18S rRNA Visceral Subcutaneous Visceral Subcutaneous r p value r p value r p value r p value Age 0.55 < 0.001 0.32 0.03 0.04 0.79 0.05 0.76 Sex -0.33 0.02 -0.45 0.01 -0.03 0.87 0.23 0.15 Visceral fat area -0.08 0.64 -0.18 0.27 0.19 0.43 -0.13 0.57 Subcutaneous fat area -0.06 0.7 -0.31 0.04 0.21 0.37 -0.10 0.67 Percent body fat -0.3 0.03 -0.41 0.003 0.16 0.34 -0.04 0.79 Waist circumference -0.2 0.2 -0.28 0.06 0.35 0.06 0.03 0.87 FPG -0.27 0.07 -0.19 0.2 0.23 0.16 -0.09 0.57 HbA1c -0.17 0.25 -0.21 0.15 0.06 0.71 -0.29 0.07 Glucose uptake, clamp -0.02 0.9 0.05 0.8 -0.19 0.33 -0.12 0.54 FPI -0.06 0.7 -0.04 0.8 0.24 0.14 -0.08 0.63 Leptin -0.29 0.04 -0.31 0.04 0.17 0.35 0.23 0.20

Participants: N = 55; FPG, fasting plasma glucose; FPI, fasting plasma insulin.

However, in our study population the most obese participants were also the youngest and upon statistical adjustment for BMI, the age relationship disappeared (Table 2). Since the univariate correlation with BMI was much stronger, the age effect was most probably

CHAPTER 3 FTO 52 secondary to BMI. In a stepwise regression analysis age did not enter the model after BMI at a p level of 0.05. There was no correlation between leptin and adiponectin serum concentrations, and visceral or subcutaneous FTO mRNA expression beyond the associations of these adipokines with BMI or percent body fat. The association between FTO expression and BMI remained significant upon adjustment for age and sex in the visceral (p = 0.003), but not in the subcutaneous depot (p = 0.08; Table 2).

Table 2: Multiple regression analysis between age, sex, BMI and FTO mRNA expression in visceral and subcutaneous adipose tissue.

Models β -Coefficient (p-value) (log) Visceral FTO mRNA (log) Subcutaneous FTO mRNA Model 1 Age (log) 0.49 (0.003) 0.17 (0.39) Gender (female) -0.08 (0.08) -0.06 (0.36) Model 2 Age (log) 0.26 (0.1) -0.003 (0.99) BMI (log) -0.66 (0.0006) -0.52 (0.049) Model 3 Age (log) 0.27 (0.09) 0.002 (0.99) Gender (female) -0.02 (0.59) -0.01 (0.89) BMI (log) -0.62 (0.003) -0.51 (0.08)

Participants: N = 55.

Since no functional information is available on how FTO might contribute to human obesity, we also modelled the data by reverse logic, i.e. BMI as dependent variable (Table 3). Interestingly, visceral FTO mRNA expression remained an independent predictor of BMI even after adjustment for subcutaneous FTO mRNA expression. There were no significant relationships between both visceral and subcutaneous FTO expression and fasting plasma glucose concentration, HbA1c and parameters of insulin sensitivity, including fasting plasma insulin concentrations and glucose uptake during the steady state of a euglycemic– hyperinsulinemic clamp (Table 1). In contrast to FTO, no correlation of RPGRIP1L mRNA expression with anthropometric and metabolic parameters was seen in either fat depot (Table 1).

CHAPTER 3 FTO 53

Table 3: Multivariate linear regression analysis of visceral and subcutaneous FTO mRNA expression as predictor of BMI. Body-Mass-Index Models β -Coefficient p-value Model 1 Age (log) -0.32 0.003 Gender (female) 0.09 0.008 SC FTO (log) -0.12 0.08 Model 2 Age (log) -0.20 0.06 Gender (female) 0.07 0.02 Visceral FTO (log) -0.29 0.003 Model 3 Age (log) -0.20 0.056 Gender (female) 0.07 0.03 Visceral FTO (log) -0.26 0.01 SC FTO (log) -0.07 0.31

Participants: N = 55; SC, subcutaneous.

3.4.3 Adipose tissue FTO mRNA expression in function of rs8050136 The SNP rs8050136 was selected for genotyping as a proxy for SNPs belonging to one linkage disequilibrium (L.D.) group based on HapMap Rel 21a/phaseII (http://www.hapmap. org/cgiperl/gbrowse/hapmap_B36/). This L.D. group includes rs9939609, which is a common variant previously shown to be strongly associated with obesity (3). No significant association of rs8050136 with visceral or subcutaneous FTO mRNA levels was found (Table 4). However, despite the lack of statistical power due to the small sample size, rs8050136 was still associated with obesity in our study group (data not shown).

Table 4: FTO and RPGRIP1L mRNA expression in visceral and subcutaneous adipose tissue grouped by rs8050136 genotypes. FTO RPGRIP1L Genotype Vis mRNA p- SC mRNA p- Vis mRNA p- SC mRNA p- rs8050136 Expression value Expression value Expression value Expression value (n) (AU) (AU) (AU) (AU) CC 2.63 ± 0.39 0.72 6.44 ± 1.41 0.90 1.80 ± 0.64 0.74 1.11 ± 0.33 0.42 (N = 13) (add) (add) (add) (add) CA 2.49 ± 0.34 0.60 8.41 ± 3.79 0.73 1.81 ± 0.26 0.91 1.00 ± 0.18 0.64 (N = 24) (dom) (dom) (dom) (dom) AA 2.22 ± 0.41 0.96 6.90 ± 1.72 0.56 1.73 ± 0.27 0.69 1.03 ± 0.19 0.41 (N = 11) (rec) (rec) (rec) (rec)

Data are means±SEM; generalized linear models were used to identify significant differences between genotypes under the additive (add), dominant (dom) and recessive (rec) models, after adjustment for age, sex and BMI; AU, arbitrary units; Vis, Visceral; SC, subcutaneous.

CHAPTER 3 FTO 54

3.5 Discussion Two independent studies recently revealed an association between common variants in the FTO gene on chromosome 16q12.2, which have been confirmed in more than 20 independent cohorts (3–5), and BMI. FTO is highly expressed in brain, but also detectable in most tissues, including adipose tissue and pancreatic beta cells (3,9). The function of the FTO gene product and its potential role for human obesity, however, are completely unknown at the moment. It is possible that only central mechanisms related to the control of appetite and satiety are involved. On the other hand, adipose tissue represents a primary candidate for studying the functional relevance of FTO in obesity. Conceivably, expression differences between lean and obese individuals could contribute to our understanding of the FTO gene product. Based on its genomic position near FTO, as well as on similar expression profiles, RPGRIP1L is also considered a plausible candidate gene involved in the pathophysiology of obesity (3,5). In this study, we provide first descriptive information on the relationship between FTO and RPGRIP1L mRNA expression in adipose tissue and simple anthropometric parameters including measures of obesity. FTO mRNA expression was threefold higher in subcutaneous than in visceral fat. This difference could be due to a number of factors usually differentiating these two depots including gene transcription, insulin sensitivity and lipolysis, altered adipokine expression and others (13, 14). Decreasing subcutaneous fat mass does not significantly improve obesity- associated metabolic abnormalities (15), whereas reduction of visceral fat mass has been shown to have significant beneficial effects on glucose metabolism and insulin sensitivity (16). Our data indicate that fat depot differences in FTO expression may contribute to the specific properties of visceral adipose tissue. The observation that FTO expression in the two depots was only modestly correlated suggests that in a given individual up- or down- regulation in one vs. the other depot is controlled locally rather than in a coordinated fashion by systemic hormonal, neuronal and/or nutritional signals. Independent of the fat depot, we found significant negative correlations between FTO gene expression and BMI or percent body fat. Therefore our data suggest that FTO gene expression might be down-regulated in response to fat accumulation. As indicated by the models (Table 2), this inverse relationship was significantly stronger for the visceral depot. From our correlation analysis it is impossible to extricate the direction of causality. Nevertheless, since genetic variation in FTO has been shown to affect the risk of obesity, it is legitimate to test the hypothesis of whether FTO expression predicts BMI. We identified visceral FTO mRNA expression as a strong predictor of BMI independently of age and sex, and even independently of FTO mRNA expression in

CHAPTER 3 FTO 55 the subcutaneous depot. This suggests that especially viscerally expressed FTO may contribute to obesity. Obviously, the number of participants in our study was relatively small and the association between BMI and adipose tissue FTO mRNA expression levels will require confirmation in larger cohorts. We also cannot exclude the possibility that specific characteristics of our study population, i.e. the strong negative correlation between age and BMI, may have produced a spurious relationship between FTO gene expression and obesity parameters. However, the fact that the BMI association withstood all conceivable adjustment makes us confident that this finding is robust. Adipose FTO expression and obesity could be co-regulated by a common genetic or environmental factor and not functionally linked to each other. More sophisticated study designs are necessary to dissect causal from correlative relationships with FTO expression in fat. We did not find an association between FTO mRNA levels and either fasting plasma glucose concentrations, HbA1c or parameters of insulin sensitivity. However, we cannot rule out the possibility that our sample size was too small to detect significant relationships between FTO gene expression and measures of insulin sensitivity and glucose metabolism. At the least, our data suggest that adipose tissue FTO expression does not directly affect risk of type 2 diabetes. As such, our results are generally supportive of the findings by Frayling et al. (3), who reported that the original association between FTO SNPs and type 2 diabetes disappeared upon adjustment for BMI. Although expression of RPGRIP1L mRNA was also fat depot-specific in our study, it does not, in contrast to FTO, seem to be linked to the pathophysiology of obesity as no correlations with anthropometric and metabolic characteristics were found. Our data therefore favours FTO rather than RPGRIP1L as a possible target affected by reported obesity- associated polymorphisms (3). Recently, mutations within RPGRIP1L have been shown to be responsible for several syndromes such as Joubert or Meckel syndrome (17,18). Joubert syndrome is an autosomal recessive multisystem disease characterized by cerebellar ataxia, developmental delay, hypotonia, irregular breathing pattern, eye movement abnormalities and cerebellar vermis hypoplasia and dysplasia with accompanying brainstem defects (18). Meckel syndrome is an autosomal recessive lethal condition characterized by central nervous system malformation, postaxial polydactyly, cystic kidney dysplasia and ductal proliferation in the portal area of the liver (17, 18). No obesity phenotype has been reported for these syndromes caused by RPGRIP1L mutations. Given the strength of the correlation between FTO expression and adiposity in the present study and the relatively small effect that variation in FTO has on BMI as reported by Frayling et al. (3), variation in the FTO genotype does not appear to play a major role in regulating FTO transcription in adipose tissue. This is supported

CHAPTER 3 FTO 56 by the fact that no association between the obesity risk polymorphism and FTO and/or RPGRIP1L mRNA expression in adipose tissue was found in our study. It is noteworthy that, despite our study’s lack of statistical power due to the small sample size, rs8050136 was still associated with obesity in this study group. In conclusion, expression of adipose tissue FTO mRNA is fat depot-specific and correlates with age, while being negatively correlated with BMI and total body fat. Adipose tissue FTO mRNA expression is not related to rs8050136 genotype, which is in L.D. with the previously reported obesity-associated SNP rs9939609 (3). However, we were unable to determine the causal direction of the association between obesity and alterations in FTO gene expression. Therefore further studies are necessary to elucidate whether changes in adipose tissue FTO gene expression are secondary to obesity or causally related.

3.6 Acknowledgements This work was supported by grants from: (1) Deutsche Forschungsgemeinschaft (DFG) BL 580/3–1 (M. Blüher); (2) the Clinical Research group ‘Atherobesity’ KFO 152, projects BL 833/1–1 (M. Blüher), Stu192/6–1 (M. Stumvoll) and FA 476/4–1 (M. Fasshauer); and (3), the Interdisciplinary Center of Clinical Research (IZKF) Leipzig at the Faculty of Medicine of the University of Leipzig (Project N06 to J. Berndt, D. Schleinitz and P. Kovacs, Project B24 to M. Blüher and Project B27 to M. Stumvoll).

3.7 Duality of interest The authors declare that there is no duality of interest associated with this manuscript.

3.8 References 1. Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors in relative body weight and human adiposity. Behav Genet 1997 27:325–351.

2. Groop L. From fused toes in mice to human obesity. Nat Genet 2007 39:706–707.

3. Frayling TM, Timpson NJ, Weedon MN et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 2007 316:889–894.

4. Dina C, Meyre D, Gallina S et al. Variation in FTO contributes to childhood obesity and severe adult obesity. Nat Genet 2007 39:724–726.

5. Scuteri A, Sanna S, Chen WM et al. Genome-wide association scan shows genetic variants in the FTO gene are associated with obesity-related traits. PLoS Genet 2007 3:e115.

CHAPTER 3 FTO 57

6. van der Hoeven F, Schimmang T, VolkmannA,MatteiMG, Kyewski B, Ruther U. Programmed cell death is affected in the novel mouse mutant Fused toes (Ft). Development 1994 120:2601–2607.

7. Peters T, Ausmeier K, Ruther U. Cloning of Fatso (Fto), a novel gene deleted by the Fused toes (Ft) mouse mutation. Mamm Genome 1999 10:983–986.

8. Stratakis CA, Lafferty A, Taymans SE, Gafni RI, Meck JM, Blancato J. Anisomastia associated with interstitial duplication of chromosome 16, mental retardation, obesity, dysmorphic facies, and digital anomalies: molecular mapping of a new syndrome by fluorescent in situ hybridization and microsatellites to 16q13 (D16S419–D16S503). J Clin Endocrinol Metab 2000 85:3396–3401.

9. Gerken T, Girard CA, Tung YC et al. The obesity-associated FTO gene encodes a 2- oxoglutarate-dependent nucleic acid demethylase. Science 2007 318:1469–1472.

10. DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 1979 237:214–223.

11. Berndt J, Klöting N, Kralisch S et al. Plasma visfatin concentrations and fat depot-specific mRNA expression in humans. Diabetes 2005 54:2911–2916.

12. Klöting N, Graham TE, Berndt J et al. Serum retinol-binding protein is more highly expressed in visceral than in subcutaneous adipose tissue and is a marker of intra-abdominal fat mass. Cell Metab 2007 6:79–87.

13. Wajchenberg BL. Subcutaneous and visceral adipose tissue: their relation to the metabolic syndrome. Endocr Rev 2000 21:697–738.

14. Frayn KN. Visceral fat and insulin resistance–causative or correlative? Br J Nutr 2000 83:S71–S77.

15. Klein S, Fontana L, Young VL et al. Absence of an effect of on insulin action and risk factors for coronary heart disease. N Engl J Med 2004 350:2549–2557.

16. Thorne A, Lonnqvist F, Apelman J, Hellers G, Arner P. A pilot study of long-term effects of a novel obesity treatment: omentectomy in connection with adjustable gastric banding. Int J Obes Relat Metab Disord 2002 26:193–199.

17. Delous M, Baala L, Salomon R et al. The ciliary gene RPGRIP1L is mutated in cerebello- oculo-renal syndrome (Joubert syndrome type B) and Meckel syndrome. Nat Genet 2007 39: 875–881.

18. Arts HH, Doherty D, van Beersum SEC et al. Mutations in the gene encoding the basal body protein RPGRIP1L, a nephrocystin-4 interactor, cause Joubert syndrome. Nat Genet 2007 39:882–888.

CHAPTER 3 FTO 58

3.9 Contribution of Authors Nora Klöting (first authorship) Study design, RNA preparation, real-time PCR, proofreading of the manuscript

Dorit Schleinitz (first authorship) Main part of the statistical analyses, design and writing of the manuscript, DNA preparation, genotyping

Karen Ruschke Proofreading of the manuscript

Janin Berndt RNA preparation, real-time PCR

Mathias Fasshauer Proofreading of the manuscript

Anke Tönjes Contribution to the discussion, proofreading of the manuscript

Michael R. Schön Surgery (preparation of the biopsies)

Peter Kovacs Project idea, part of the statistical analyses, editing of the manuscript

Michael Stumvoll Project idea, proofreading of the manuscript

Matthias Blüher Project idea, recruiting and phenotyping of study subjects, editing of the manuscript

CHAPTER 4 59

CHAPTER 4

Effect of Genetic Variation in the Human Fatty Acid Synthase Gene (FASN) on Obesity and Fat Depot-Specific mRNA Expression

Dorit Schleinitz1, Nora Klöting2, Antje Körner3, Janin Berndt1, Marlene Reichenbächer2, Anke Tönjes2,4, Karen Ruschke2, Yvonne Böttcher2, Kerstin Dietrich1, Beate Enigk1, Matthias Filz2, Michael R. Schön5, Jost Jenkner5, Wieland Kiess3, Michael Stumvoll2, Matthias Blüher2 and Peter Kovacs1

1Interdisciplinary Centre for Clinical Research, Medical Faculty, University of Leipzig, Leipzig, Germany 2Department of Medicine, University of Leipzig, Leipzig, Germany 3University Hospital for Children and Adolescents, University of Leipzig, Leipzig, Germany 4Coordination Centre for Clinical Trials, University of Leipzig, Leipzig, Germany 5Städtisches Klinikum Karlsruhe, Clinic of Visceral Surgery, Karlsruhe, Germany

Published in: Obesity (Silver Spring) 2010 18:1218-1225

CHAPTER 4 FASN 60

4.1 Abstract Inhibition of fatty acid synthase (FASN) induces a rapid decline in fat stores in mice, suggesting a role for this enzyme in energy homeostasis. To investigate the potential role of FASN in the pathophysiology of human obesity, the FASN gene was sequenced in 48 German whites. Thirty-five single-nucleotide polymorphisms (SNPs) were identified. Eight SNPs representative for their linkage disequilibrium groups and the Val1483Ile (rs2228305) substitution were genotyped for subsequent association analyses in 1,311 adults from Germany. Further, the tagging SNPs were genotyped also in German childhood cohorts (738 schoolchildren, 205 obese children). Effects of genetic variation on FASN mRNA expression in visceral and subcutaneous adipose tissue from a subgroup of 172 subjects were analyzed. Several polymorphisms in the FASN (rs62078748, rs2229422, rs2229425, and rs17848939) were nominally associated with obesity in case–control studies including 446 obese subjects (BMI ≥ 30 kg/m2) and 389 lean controls (BMI ≤ 25 kg/m2) (adjusted p < 0.05). The strongest significant effect was found for rs2229422 (p = 1.3 × 10−5 adjusted for age, sex, type 2 diabetes status), which was supported by associations with BMI, waist-to-hip ratio (WHR), fasting plasma insulin and glucose infusion rate (adjusted p < 0.05). Subjects with the Val1483Ile substitution appeared to be protected against obesity. In addition, rs17848939 was nominally significantly associated with the ratio of visceral/subcutaneous FASN mRNA expression (adjusted p = 0.04). No effect of genetic variation in FASN on obesity was found in children. In conclusion, our data indicate a role of FASN genetic variation in susceptibility to obesity in adults.

4.2 Introduction Obesity is a significant risk factor for many life-threatening diseases, and its global impact on health is enormous (1). Although the etiology of obesity is poorly understood, this disease has both environmental and genetic components. It has been postulated that increased expression and activity of the lipogenic pathways in adipose tissue could play a role in obesity development (2). The central enzyme in the pathway of de novo lipogenesis, fatty acid synthase (FASN), catalyses all steps in the conversion of malonyl CoA to palmitate (3). Although reports on genetic variation in human FASN are rare, there are a number of recent studies from rodents indicating that FASN may be involved in obesity through regulation of feeding behaviour (4–6). Treatment of mice with FASN inhibitors (cerulenin, C75) results in reduced food intake and substantial body (4). The FASN inhibitor, C75, acts both centrally to reduce food intake and peripherally to increase fatty acid oxidation, leading to

CHAPTER 4 FASN 61 rapid and profound weight loss, loss of adipose mass, and resolution of fatty liver in diet- induced obese mice (7). The human FASN map on chromosome 17q25 (complement strand) within a region of linkage for body fat content in Pima Indians (8). Sequence analysis of FASN in this genetically and environmentally homogeneous population of Southern Arizona characterized by obesity and high prevalence of type 2 diabetes mellitus (T2D) (9,10), revealed a novel Val1483Ile polymorphism that was shown to be associated with percent body fat and 24-h substrate oxidation rates measured in a respiratory chamber (11). It has been suggested that the Val1483Ile substitution in FASN is protective against obesity, an effect possibly explained by the role of this gene in the regulation of substrate oxidation (11). The Pima data have recently been supported by studies in white children, reporting a sex- specific protective effect of the Val1483Ile variant for obesity as well as its beneficial effect on lipid profile in obese boys (12). However, comprehensive genetic analyses of FASN such as the one reported in Pima Indians are still lacking in white populations. Recently, we examined whether FASN mRNA expression in visceral and subcutaneous fat correlates with anthropometric and metabolic parameters in subjects with a wide range of obesity, body fat distribution, insulin sensitivity, and glucose tolerance (13). Our data clearly demonstrated significant association of increased FASN expression in adipose tissue not only with obesity, but also with T2D and insulin resistance, thus indicating a possible role of lipogenic pathways in the development of human obesity and/or T2D (13). Therefore, in the present study we sequenced the gene in subjects from Eastern Germany and assessed the effects of identified polymorphisms on human obesity/T2D and pathophysiologically relevant traits. We further investigated whether genetic variation within the FASN might be responsible for changes in FASN expression in adipose tissue.

4.3 Experimental subjects 4.3.1 Leipzig adult cohort In total, 640 patients with T2D and 671 healthy subjects were recruited at the University Hospital in Leipzig, Germany. BMI was calculated as weight divided by squared height. Waist and hip circumferences were measured and waist-to-hip ratio (WHR) was calculated. The healthy subjects included 246 males and 425 females (mean age 46.5 ± 14.4 years, mean BMI 27.5 ± 5.4 kg/m2) and patients with T2D included 301 males and 339 females (mean age 63.3 ± 11.3 years, mean BMI 29.8 ± 5.2 kg/m2; the data represent mean ± SD). In addition oral glucose tolerance test and fasting plasma insulin measurements were performed in all nondiabetic subjects. The oral glucose tolerance test was performed

CHAPTER 4 FASN 62 according to the criteria of the American Diabetes Association (14). The test was carried out after an overnight fast with 75 g standardized glucose solution (Glucodex Solution 75 g; Merieux, Montreal, Canada). Venous blood samples were taken at 0, 60, and 120 min for measurements of plasma glucose concentrations. 196 out of 671 subjects had impaired glucose tolerance. As impaired glucose tolerance is a T2D predicting factor, only remaining 502 subjects with normal glucose tolerance were included as healthy controls in T2D association study. In a subgroup of 380 nondiabetic subjects, body fat content and insulin sensitivity were measured. Percent body fat was measured by dual-energy X-ray absorptiometry. Insulin sensitivity was assessed with the euglycemic–hyperinsulinemic clamp method as previously described (15,16). In addition, paired samples of visceral and subcutaneous adipose tissue and DNA were obtained from a subgroup of 172 white men (N = 84) and women (N = 88), who underwent open abdominal surgery for gastric banding, cholecystectomy, weight reduction surgery, abdominal injuries, or explorative laparotomy. The age ranged from 16 to 99 years and BMI from 20.8 to 54.1 kg/m2. Thirty-eight subjects had T2D and 15 subjects had impaired glucose tolerance. All subjects had a stable weight with no fluctuations of > 2 % of the body weight for at least 3 months before surgery. Patients with severe conditions including generalized inflammation or end stage malignant diseases were excluded from the study. Samples of visceral and subcutaneous adipose tissue were immediately frozen in liquid nitrogen after explantation. The white status of study subjects was self-reported. All studies were approved by the ethics committee of the University of Leipzig and all subjects gave written informed consent before taking part in the study.

4.3.2 Childhood cohorts Leipzig Schoolchildren cohort: This cohort is part of the Leipzig Schoolchildren Project that investigated anthropometric and clinical parameters in 2,675 children aged 6 to 17 years from 1999–2000 (17). DNA was available from 738 children (358 boys and 380 girls; mean age 12 ± 3 years; mean BMI-SDS 0.09 ± 0.98). The BMI was standardized referring to national reference data (18) and children with a BMI ≥ 1.88 SDS (= 97th percentile) were classified obese. A total of 508 children and adolescents with BMI between −1.0 SDS and 1.0 SDS (243 boys and 265 girls; mean age 12 ± 3 years; mean BMI-SDS −0.05 ± 0.54) were selected from the 738 schoolchildren to serve as healthy normal weight control group to Leipzig obese children cohort.

CHAPTER 4 FASN 63

Leipzig obese children cohort: A total of 205 white children and adolescents (97 boys and 108 girls; 12 ± 4 years; BMI-SDS 2.71 ± 0.58) were recruited from the obesity clinic at the University Hospital for Children & Adolescents, Leipzig, Germany. All obese children had a detailed metabolic work up including an oral glucose tolerance test as described in detail elsewhere (19). The studies were approved by the ethics committee of the University of Leipzig.

4.4 Methods and Procedures 4.4.1 Analysis of human FASN gene expression Total RNA was isolated from paired subcutaneous and visceral adipose tissue samples using TRIzol (Life Technologies, Grand Island, NY), 1 μg RNA was reverse transcribed with standard reagents (Life Technologies). From each reverse transcription–PCR, 2 μl was amplified in a 26 μl PCR using the Brilliant SYBR Green QPCR Core Reagent Kit from Stratagene (La Jolla, CA) according to the manufacturer’s instructions. Samples were incubated in the ABI PRISM® 7000 sequence detector for an initial denaturation at 95 °C for 10 min, followed by 40 PCR cycles, each cycle consisting of 95 °C for 15 s, 60 °C for 1 min, and 72 °C for 1 min. The following primers were used: human FASN (accession no. NM_004104) 5’-CGCGTGGCCGGCTACTCCTAC-3’ (sense), and 5’-CGGCTGCCACACG CTCCTCT-3’ (antisense). Human FASN mRNA expression was calculated relative to the mRNA expression of 18S rRNA, determined by a premixed assay on demand for human 18S rRNA (PE Biosystems, Foster City, CA). Amplification of specific transcripts was confirmed by melting curve profiles (cooling the sample to 68 °C and heating slowly to 95 °C with measurement of fluorescence) at the end of each PCR. The specificity of the PCR was further verified by subjecting the amplification products to agarose gel electrophoresis.

4.4.2 Sequencing of the FASN To identify genetic variants, all 43 exons, including intron/exon splicing sites (the total length of FASN mRNA is 8,481 bp, NM_004104), 2,280 bp upstream of the first translation initiation site and 840 bp of the 3’ untranslated region were sequenced in DNA samples from 48 nonrelated German subjects (24 obese and 24 with normal weight). Sequencing was performed using the BigDye® Terminator (Applied Biosystems, Foster City, CA) on an automated DNA capillary sequencer (ABI PRISM® 3100 Avant; Applied Biosystems). Sequence information for all oligonucleotide primers used for variant screening is available upon request.

CHAPTER 4 FASN 64

4.4.3 Genotyping of the FASN SNPs Genotyping of selected single-nucleotide polymorphisms (SNPs) in all study subjects was done using the TaqMan® assay (Applied Biosystems) for the rs17848935, rs4485435, rs2229422, rs4246445, rs2229425, rs2228309, rs62078748, rs2228305 variants and by restriction fragment length polymorphism technique for the rs17848939 variant. The TaqMan® genotyping reaction was amplified on a GeneAmp® PCR system 9700 (95 °C for 15 min, and 95 °C for 15 s, and 60 °C for 1 min, for 40 cycles) and fluorescence was detected on an ABI PRISM® 7700 sequence detector (Applied Biosystems). The restriction fragment length polymorphism genotypes were determined by PCR amplification of the respective fragments of the FASN on a GeneAmp® PCR system 9700 (94 °C for 3 min, and 94 °C for 30 s, and 66 °C for 45 s, 72 °C for 45 s, for 35 cycles, and 72 °C for 10 min) and subsequent digestion with the NotI restriction enzyme and size fractionation and visualization by electrophoresis. To assess genotyping reproducibility, a random ~10 % selection of the sample was regenotyped for each SNP; all genotypes matched initial designated genotypes. Moreover, the DNA samples selected for sequencing were regenotyped by TaqMan® technique, as they were also included in the association analyses. The genotypes matched completely between the two techniques.

4.4.4 Statistical analyses Prior to statistical analysis, non-normally distributed parameters (glucose infusion rate, 2-h plasma glucose, fasting plasma insulin, visceral and subcutaneous FASN mRNA expression) were logarithmically transformed to approximate a Gaussian distribution. Differences in genotype frequencies were compared using logistic regression models between the obese cases and lean controls as well as between diabetic cases and healthy controls with normal glucose tolerance. Multivariate linear relationships were assessed by generalized linear model analyses. In the additive model, homozygotes for the major allele (MM), heterozygotes (Mm) and homozygotes for the minor allele (mm) were coded to a continuous numeric variable for genotype (as 0, 1, and 2, respectively). A dominant model was defined as contrasting genotypic groups MM + Mm vs. mm, and the recessive model was defined as contrasting genotypic groups MM vs. Mm + mm. In haplotype analyses, groups of subjects carrying 2, 1, or 0 copies of the haplotype were compared. To estimate haplotypes we used the PHASE version 2.1 software (20,21). The phase reconstruction method considers the unknown haplotypes as unobserved random quantities and aims to evaluate their conditional distribution in light of the genotype data. Statistical analyses were performed using the SPSS

CHAPTER 4 FASN 65 software package (SPSS, Chicago, IL) and the statistical analysis system of the SAS Institute (Cary, NC). All data are presented without correcting for multiple testing. Bonferroni corrections for multiple testing would require a p value of 0.006 for analyses of associations in the present case–control studies and p value of 0.007 for the association analyses with quantitative traits to consider the findings significant. P values < 0.05 but > 0.006 were therefore considered to be of nominal statistical significance.

4.5 Results 4.5.1 Genetic variation in the FASN The FASN gene (all 43 Exons with Intron/Exon splicing sites, 2 kb in the 5’ region and 840 bp of the 3’ untranslated region) was sequenced in 48 white DNA samples (24 obese and 24 lean subjects) to identify variation. Thirty-five SNPs including 14 novel variants were identified: 12 SNPs were in the 5’ region, 14 SNPs in exons, 7 SNPs in introns and 2 SNPs in the 3′ untranslated region (Table 1). The genotype distributions for all SNPs were consistent with Hardy–Weinberg equilibrium (p > 0.05), except for rs1140616 which was therefore excluded from further analyses. Linkage disequilibrium (r2 and D-prime (D’)) was calculated among the 16 common variants (minor allele frequency (MAF) > 0.05) using the EMLD statistical program (https://epi.mdanderson.org/~qhuang/Software/pub.htm) (Supplement Table 1). Based on L.D. between these common variants, eight groups were defined by the Tagger software (22) (Supplement Table 2). Because SNPs in high genotypic concordance would provide the same information for association studies, one representative variant for each L.D.-group was selected and genotyped in all subjects for association analyses (Supplement Table 2). In addition we genotyped the Val1483Ile (rs2228305) variant (MAF = 4 %) in the adult cohort, which has previously been reported to be associated with obesity in different ethnicities (11,12). Moreover, using the PHASE version 2.1 software (20,21), we identified five common haplotypes (frequency of each haplotype ≥ 0.05) among the eight different SNPs genotyped in all subjects. These five haplotypes, [GCTCTGCC], [GTCCTACT], [GTCGTACC], [ACCCCGCC], and [ACCCCGTC] accounted for > 80 % of the all observed haplotypes, where haplotypes are defined by the composition of alleles at each SNP in following order: [rs62078748]-[rs2228309]-[rs17848935]-[rs4485435]- [rs2229422]-[rs17848939]-[rs2229425]-[rs4246445].

CHAPTER 4 FASN 66

Table 1: Genetic variants detected in the FASN coding region, its 5’ region and 3’ UTR. No. Variant Minor allele dbSNP Amino acid frequency NCBI BUILD126 exchange 1 -2268T>G G = 0.01 - - 2 -2177T>C C = 0.49 rs59251877 - 3 -2095C>A A = 0.05 - - 4 -2051T>G G = 0.01 - - 5 -1745T>C C = 0.33 rs59638227 - 6 -1556T>C C = 0.02 - - 7 -1195C>T T = 0.16 rs62078749 - 8 -1004G>A A = 0.37 rs62078748 - 9 -885G>C C = 0.48 rs62078747 - 10 -605C>T T = 0.16 - - 11 -54G>T T = 0.30 rs17848934 - 12 -20G>A A = 0.03 - - 13 c.567C>T T = 0.48 rs2228309 (p.=) (189Asn) 14 c.591C>T T = 0.03 - (p.=)(197Ser) 15 c.1436G>T T = 0.02 - (p.Gly479Val) 16 c.1965+68C>T T = 0.02 rs45586937 - 17 c.2155G>A A = 0.01 rs12946178 (p.Glu719Lys) 18 c.3224-18C>T T = 0.15 rs17848935 - 19 c.3267C>G G = 0.17 rs4485435 (p.=) (1089Ala) 20 c.3651C>T T = 0.15 rs41283365 (p.=) (1217Gly) 21 c.3741T>C C = 0.36 rs2229422 (p.=) (1247Ala) 22 c.4122+9G>A A = 0.45 rs17848939 - 23 c.4447G>A A = 0.04 rs2228305 (p.Val1483Ile) 24 c.5268C>T T = 0.19 rs2229425 (p.=) (1756Ser) 25 c.5336C>T T = 0.01 rs2229426 (p.Pro1779Leu) 26 c.5470C>T T = 0.02 - (p.Arg1824Trp) 27 c.5872G>A A = 0.02 - (p.Glu1958Lys) 28 c.6374G>A A = 0.01 - (p.Arg2125Gln) 29 c.6402T>C C = 0.33 rs1140616 (p.=) (2134Ile) 30 c.6406+12C>T T = 0.01 - - 31 c.6595+12C>T T = 0.38 rs4246445 - 32 c.7047+31G>A A = 0.01 rs45466394 - 33 c.7147-46G>A A = 0.01 - - 34 *141A>G G = 0.03 - - 35 *142C>T T = 0.38 rs11550611 -

No. 1-12 5’ region, No. 13-33 coding and intronic region, No.34-35 3’ UTR.

CHAPTER 4 FASN 67

4.5.2 Association studies in adults Genetic variants in the FASN and obesity. Several representative SNPs were associated with obesity under logistic regression analysis in a case–control designed study including 446 obese subjects (BMI ≥ 30 kg/m2, mean age: 59 ± 12 years, mean BMI: 34.6 ± 4.5 kg/m2) and 389 lean controls (BMI ≤ 25 kg/m2, 48 ± 18 years, 23.4 ± 1.4 kg/m2) (Table 2). Including age, sex, and T2D status in the analyses, frequency of minor alleles for the SNPs rs62078748, rs2229422, and rs2229425 was significantly higher in the obese group compared with lean controls (p values ranging from 0.001 to 1.3 × 10−5). This conferred a significantly higher risk of obesity in additive mode of inheritance ranging from odds ratio = 1.51 (95 % confidence interval, CI, (1.19; 1.92)) for the rs62078748 to 1.85 (1.35; 2.56) for the rs2229425. In contrast, the minor allele for rs17848939 was nominally significantly more frequent in the lean group (p = 0.01; in additive mode of inheritance after adjusting for age, sex, and T2D). The analyses were performed also without subjects with T2D; the SNP allele frequencies in the obese group remained unchanged. However, even though we attempted to eliminate some redundancy in the association testing by selecting only one SNP per each defined L.D.-group, the tagging SNPs seem to be still correlated (e.g., between rs62078748 and rs2229422 D′ = 0.86 and r2 = 0.63; Supplement Table 1). As it is possible that a single variant could explain all the observed association, all tests were re-analyzed conditional on the strongest association found for rs2229422. After including rs2229422 in the analyses, none of the remaining SNPs showed significant association with obesity. Consistent with the SNP data, two haplotypes were significantly associated with obesity. The frequency of the haplotype [ACCCCGCC] was significantly increased in the obese group (13.9 % vs. 10.0 % in the lean subjects) whereas the haplotype [GTCCTACT] showed increased frequency in lean controls (25.3 % vs. 19.4 % in the obese group (p = 0.003 and 0.002, respectively; in additive mode of inheritance after adjusting for age and sex). PHASE software provides a list of the best haplotype guess for each individual. To avoid misinterpreting the association results between haplotypes and phenotypic traits, we performed the analyses also in a dataset excluding all individuals where the phase probability at least at one site was < 0.8. This eliminated all subjects from the analyses who may have influenced the analyses by including their uncertain haplotypes. The results of analyses with the new dataset remained materially unchanged. The p value for association of the [ACCCCGCC] haplotype with obesity changed from 0.003 to 0.009 and remained the same for the [GTCCTACT] haplotype (p = 0.002).

CHAPTER 4 FASN 68

Table 2: Association of FASN genetic variants with obesity. SNP Controls Cases p-value p-value1 p-value2 (unadjusted) OR (95 % CI) (1 + rs2229422) rs62078748 GG 51.8 % 42.3 % GA 42.1 % 44.1 % 4x10-4 0.001 0.74 AA 6.2 % 13.6 % 1.51 (1.19; 1.92) MAF 27.2 % 35.7 % rs2228309 CC 26.8 % 33.1 % CT 50.8 % 49.5 % 0.07 0.10 0.43 TT 22.4 % 17.4 % 0.83 (0.66; 1.04) MAF 47.8 % 42.1 % rs17848935 CC 63.1 % 67.4 % CT 34.0 % 29.1 % 0.37 0.11 0.69 TT 2.9 % 3.5 % 0.77 (0.57; 1.06) MAF 19.9 % 18.1 % rs4485435 CC 69.0 % 68.1 % CG 28.9 % 30.1 % 0.93 0.93 0.21 GG 2.0 % 1.8 % 0.99 (0.73; 1.33) MAF 16.5 % 16.9 % rs2229422 TT 50.1 % 37.7 % TC 43.5 % 46.7 % 9 x 10-6 1.3 x 10-5 - CC 6.4 % 15.6 % 1.72 (1.35; 2.18) MAF 28.1 % 39.0 % rs17848939 GG 22.4 % 31.7 % GA 50.1 % 47.8 % 0.005 0.01 0.99 AA 27.4 % 20.4 % 0.75 (0.60; 0.93) MAF 52.5 % 44.4 % rs2229425 CC 77.0 % 65.3 % CT 22.0 % 33.3 % 0.001 2 x 10-4 0.09 TT 1.0 % 1.4 % 1.85 (1.35; 2.56) MAF 12.0 % 18.0 % rs4246445 CC 48.9 % 56.5 % CT 42.7 % 38.1 % 0.05 0.16 0.96 TT 8.4 % 5.3 % 0.83 (0.65; 1.08) MAF 29.8 % 24.4 %

Controls: BMI ≤ 25 kg/m2, N = 389; Cases: BMI ≥ 30 kg/m2, N = 446; 1adjusted for age, sex and T2D; 2adjusted for age, sex, T2D and rs2229422; p-values and odds ratio (OR) with 95 % confidence intervals (CI) were calculated under logistic regression analyses in additive mode of inheritance; MAF - minor allele frequency.

CHAPTER 4 FASN 69

Genetic variants in the FASN and type 2 diabetes. In the present study, five out of eight representative SNPs were nominally significantly associated with T2D in 640 cases and 502 healthy controls (with normal glucose tolerance) under logistic regression analysis (all p < 0.05 in additive mode of inheritance; Supplement Table 3). However, after adjusting the analyses for age, sex, and BMI, only rs2228309 remained moderately associated with T2D (p < 0.05; Supplement Table 3). Consistent with these data, none of the five common haplotypes showed significant differences in frequencies between diabetic and nondiabetic subjects (all p > 0.05; data not shown).

Genetic variants in the FASN and parameters of obesity and glucose metabolism. In the cohort of 671 white subjects without T2D, several polymorphisms showed relationship to quantitative parameters of obesity and glucose metabolism. As the strongest effect in the obesity case–control study was observed for rs2229422, we focused our analyses on this SNP. The rs2229422 C allele was significantly associated with higher BMI and higher WHR (after adjusting for age and sex; Supplement Table 4). Moreover rs2229422 was nominally associated with fasting plasma insulin (adjusted for age, sex, and BMI) (Supplement Table 4). Finally, the C allele was associated with lower glucose infusion rate assessed by euglycemic– hyperinsulinemic clamp (adjusted for age, sex, and BMI) (Supplement Table 4). The effect of the SNP on obesity-related traits was reflected in the haplotype analysis as well, which however did not show any associations beyond those observed for the SNP (data not shown).

Val1483Ile genotype protects against obesity. In the present study, we also attempted to replicate the previously described relationship of the Val1483Ile substitution on obesity. In the entire cohort, Ile-allele carriers had nominally significantly lower BMI than noncarriers (Figure 1; p = 0.04 after adjusting for age and sex in additive mode of inheritance). After gender stratification, the association with BMI was only seen in male subjects (Figure 1).

CHAPTER 4 FASN 70

30.0 p = 0.04 p = 0.004 p = 0.85 Val/Val Val/Ile 27.5 )

2 Ile/Ile

25.0

BMI (kg/m 22.5

20.0 Total Males Females (N = 547) (N = 764)

Figure 1: Association of the Val1483Ile (rs2228305) variant with BMI. The data are presented as mean ± SEM. P-values were calculated after adjusting for age (and for sex in the entire cohort) in an additive mode of inheritance.

Also in the case–control study, an association with obesity was seen in males but not in females (adjusted p = 0.04 vs. p = 0.86 in additive mode of inheritance) (Table 3). Based on this, we tested a possible gender specific effect for the other investigated SNPs. Unlike for Val1483Ile, we did not observe gender differences in effects of the other SNPs on obesity and related traits. All p values in gender strata were consistent with those observed in the total cohort (data not shown). Although both the Val1483Ile and the rs2229422 are associated with obesity, they are not in strong linkage disequilibrium (D’ = 0.76 and r2 = 0.01) and thus, their effects seem to be independent of each other.

Table 3: Association of the Val1483Ile (rs2228305) variant with obesity. Val/Val Val/Ile Ile/Ile p-value Odds ratio (95 %CI) Total Controls 91.1 % 7.7 % 1.2 % (N = 389) 0.14 0.65 (0.36; 1.16) Cases 93.6 % 6.4 % - (N = 446) Males Controls 87.5 % 10.3 % 2.2 % (N = 152) 0.04 0.41 (0.17; 0.95) Cases 94.3 % 5.7 % - (N = 181) Females Controls 93.6 % 5.9 % 0.5 % (N = 237) 0.86 1.08 (0.47; 2.47) Cases 93.2 % 6.8 % - (N = 265)

CHAPTER 4 FASN 71

Table 3 cont.: Controls included subjects with BMI ≤ 25 kg/m2, cases included subjects with BMI ≥ 30 kg/m2; p-values were calculated under logistic regression models after adjusting for age (and for sex in the entire cohort) in an additive mode of inheritance.

4.5.3 Association studies in children In a case–control study including 508 lean and 205 obese children, no significant association of the FASN SNPs was found (p > 0.05 after adjusting for age, sex, pubertal stage, and height). In line with this, the FASN variants did not associate with measures of obesity (BMI, WHR) in either schoolchildren (N = 738) or obesity cohort (N = 205) (adjusted p > 0.05; data not shown).

4.5.4 Association of genetic variants in the FASN with fat depot-specific FASN mRNA expression No association of the eight representative SNPs or derived haplotypes with FASN mRNA levels in either visceral or subcutaneous adipose tissue was found in 172 individuals with visceral and subcutaneous adipose tissue biopsies (all p > 0.05 after adjusting for age, sex, and BMI). To express FASN mRNA expression in visceral relative to that in subcutaneous fat, we calculated the visceral/subcutaneous FASN mRNA ratio. The rs17848939 was nominally significantly associated with the visceral to subcutaneous (Vis/SC) FASN mRNA expression ratio (p = 0.04 and 0.02 in additive and recessive mode respectively, after adjusting for age, sex, and BMI) (Table 4). We also performed statistical analyses excluding subjects with impaired glucose tolerance and T2D, but the results remained unchanged (data not shown).

Table 4: Association of the rs17848939 variant with FASN mRNA levels. rs17848939 Genotype GG GA AA P-value N = 58 N = 82 N = 32 Sex (Males/Females) 33/25 42/40 9/23 0.03 Age (years) 57 ± 2 53 ± 1 60 ± 3 0.05 BMI (kg/m2) 33.1 ± 1.1 30.5 ± 0.7 31.3 ± 1.4 0.18 FASN mRNA expression add dom rec Log Visceral fat 5.11 ± 0.30 4.53 ± 0.28 4.79 ± 0.47 0.44 0.58 0.49 Log Subcutaneous fat 3.13 ± 0.26 3.18 ± 0.23 3.26 ± 0.39 0.52 0.87 0.27 Log Visceral/Subcutaneous 2.11 ± 0.20 1.44 ± 0.22 1.51 ± 0.24 0.04 0.34 0.02

CHAPTER 4 FASN 72

Table 4 cont.: The data are presented as mean ± SEM; p-values were calculated after adjusting for age, sex and BMI. Add, dom and rec denote P-values in additive, dominant and recessive mode of inheritance, respectively.

4.6 Discussion The FASN has been comprehensively investigated in Pima Indians (11), but such studies are still missing in European populations. In the present study, we therefore re- sequenced FASN in subjects from Eastern Germany and investigated identified polymorphisms in subsequent association analyses. In addition, we also evaluated the role of genetic variation in the FASN on its mRNA expression in visceral and subcutaneous adipose tissue. The key findings of the present study are: (i) Genetic variants in the FASN are significantly associated with human obesity and its related traits in adult Germans and (ii) Support of previously reported sex-specific effects of the Val1483Ile variant on obesity. In the present study, 35 genetic variants were identified by sequencing the FASN. Sixteen variants had MAF > 5 % and fell into eight L.D.-groups. Association analyses of representative variants for each of the L.D.-groups showed several polymorphisms (rs62078748, rs2229422, rs17848939, rs2229425) associated with obesity in case–control studies. Conditional analysis revealed rs2229422 to be the strongest obesity associated FASN genetic variant which was further supported through analyses in the cohort of white subjects without T2D. The polymorphism showed relationships to quantitative measures of obesity such as BMI and WHR. Although the rs2229422 showed nominally significant associations with T2D in the case–control study, the effect did not withstand corrections for BMI suggesting that the effect of the SNP on T2D was mediated through obesity. None of the investigated polymorphisms with MAF > 5 % has an evident functional relevance for the FASN protein i.e., no SNP causes amino acid substitution in the protein or results in a new frameshift. We therefore further evaluated whether FASN genetic variants might affect the mRNA expression in visceral and subcutaneous fat and thus explain the relationship with metabolic traits related to obesity and T2D. Recent studies carried out by our group showed that FASN mRNA expression in adipose tissue was related to measures of obesity, insulin sensitivity, and glucose tolerance (13). Furthermore, there was increased FASN mRNA expression in visceral as compared to subcutaneous adipose tissue in lean healthy subjects, as well as in obese patients with or without T2D (13). In the present study, the obesity risk rs17848939 G-allele was nominally significantly associated with the visceral to subcutaneous FASN mRNA expression ratio. Consistent with the previously described positive correlation of fat mRNA

CHAPTER 4 FASN 73 expression with measures of obesity, the rs17848939 variant also showed nominally significant association with BMI and % body fat (data not shown). These data suggest that genetic variation in FASN may at least partially explain variation in mRNA expression in visceral and subcutaneous adipose tissue. However, it needs to be noted that based on correlation data including mRNA expression together with genetic variation and a certain phenotype (such as BMI) one can not reveal the causative chain and so, to rule out that changes in mRNA expression simply reflect the state of obesity. As shown recently in a genome-wide study on mRNA expression in adipose tissue (23), the majority of the traits profiled (over 70 % or 17,080 traits) showed significant correlation with BMI. Therefore, studies targeting functional consequences of FASN obesity risk variants on FASN transcription will be inevitable to clarify whether or not genetic variants can explain variation in mRNA expression. In addition to above mentioned polymorphisms, we attempted to replicate previously reported correlations between the Val1483Ile substitution and obesity despite low MAF (4 %) of this variant in Europeans of German origin. The variant was first identified and shown to be associated with percent of body fat and substrate oxidation rates in Pima Indians (11). It has been hypothesised that the Val/Ile substitution, which is positioned in the FASN interdomain region (24), could alter this functionally essential site and thus potentially reduces FASN activity (11). Consequently, changes in FASN activity could affect substrate oxidation rates, which would result in the decrease in body fat observed in the Pima Indians (11). Further support for the protective effect of this variant against obesity came from studies in German children showing reduced BMI-SDS and improved lipid profile of boys carrying the Ile-variant (12). Findings in children cohorts suggested sex-specific effects of the polymorphism which was confirmed recently by independent studies in Spanish whites (25). Spanish men with the Ile-allele had lower BMI, WHR, fasting glucose, and blood pressure, and they showed increased insulin sensitivity (25). However, no association was found in women (25). Also in our study, in contrast to women, German adult men carrying the Ile-allele seem to be protected against obesity as suggested by the case–control study as well as by the association with lower BMI. Because Ile-carriers also have decreased adipose tissue FASN activity (25), it might be postulated that Val1483Ile is a functional variant in FASN. Sexual dimorphism in genetic associations in obesity, glucose, and lipid metabolism is not uncommon and has been reported for genes such as leptin (26), resistin (27), winged helix/forkhead transcription factor (28), peptide tyrosine tyrosine (29), Y2 receptor (29) and others. It could be hypothesised that the protective effects of the Val1483Ile are masked by hormonally controlled mechanisms in

CHAPTER 4 FASN 74 women. It needs to be mentioned though that several other variants (with very low MAF) identified by sequencing and which predict amino acid substitutions were not examined in the present study. Therefore, we can not exclude that rare variants other than Val1483Ile might be affecting human obesity as well. Children represent a valuable study population for identifying primary genetic determinants involved in susceptibility to complex polygenic diseases such as obesity. Unlike in adults their phenotype is less influenced by comorbidities and prolonged exposure to environmental factors. Interestingly, although we previously showed sex-specific association of the Val1483Ile variant with obesity in German children (12), no relationship with other SNPs associated with obesity in adults, was seen in our present study. Lack of association of FASN SNPs with childhood obesity indicates that the SNP effects most likely manifest in later stages of life and/or that these effects might be modified by gene-environment interactions. This aspect may need to be considered in association studies and may partially explain why genetic variation in FASN was not found to be associated with obesity in genome-wide association scans (GWAS). However, we are aware that effect sizes observed in the present study should have been picked up easily in well powered GWAS. Due to the relatively small sample size in our study, we see these effect sizes with caution. Also, we attempted to replicate the observed associations in our GWAS in a self-contained population of Sorbs from Germany (30). The GWAS was performed by using 500 K and 6.0 Affymetrix GeneChips. Unfortunately, no SNPs within the FASN locus were included on the chips, which made potential comparisons with our sequenced variants impossible. Similarly, the 1958 British Birth Cohort (BC58) publicly available data only provide three SNPs within the FASN (http://www.b58cgene.sgul.ac.uk/index.php). Thus the lack of associations with obesity in GWAS may partly be attributed to missing SNPs on the used GeneChip arrays. Further, similar MAF for SNPs rs62078748, rs2229422, rs17848939, rs2229425 in our control sample and HapMap populations might eventually support our association findings. Unfortunately, we were not able to compare our control sample with other populations in silico, since there are no publicly available data on allele frequencies for these SNPs at the moment. We are aware that our findings will need to be confirmed in independent studies. However, it should be noted that the strongest association signal observed for rs2229422 in our study did withstand very conservative Bonferroni corrections for multiple testing, which would require p value of 0.006 for analyses of associations in the present case–control studies. In addition, we could replicate previously described protective effects of the Val1483Ile variant against obesity, which further supports validity of these findings. Based on our data, we suggest that

CHAPTER 4 FASN 75 variation in the human FASN may play a role in the regulation of . Although these data support previous studies in rodents and replicate protective effects of the Val1483Ile against obesity, associations of common variants in FASN with obesity are novel and need to be replicated to confirm these findings and define their relative importance in the pathogenesis of human obesity.

4.7 Acknowledgements We thank all those who participated in the studies. We also thank Roy Tauscher for excellent technical assistance. This work was supported by grants from Deutsche Forschungsgemeinschaft, the Clinical Research Group “Atherobesity” KFO 152 (projects BL 833/1-1 to M.B., and Stu192/6-1 (M.S.), KO 3512/1-1 to A.K. and TP5 to W.K.), the Interdisciplinary Centre for Clinical Research Leipzig at the Faculty of Medicine of the University of Leipzig (Project N06 to P.K., D.S., B.E., K.D. and project B24 to M.B., B27 to M.S. and A.T.), from the German Diabetes Association (to Y.B., A.T., P.K., A.K.), from the Deutsche Hochdruckliga and by unrestricted grants from Merck Serono, Ipsen and Novo Nordisk (to W.K.).

4.8 Disclosure The authors declared no conflict of interest.

4.9 References 1. Must A, Spadano J, Coakley EH et al. The disease burden associated with overweight and obesity. JAMA 1999 282:1523–1529.

2. Aarsland A, Chinkes D, Wolfe RR. Contributions of de novo synthesis of fatty acids to total VLDL-triglyceride secretion during prolonged hyperglycemia hyperinsulinemia in normal man. J Clin Invest 1996 98:2008–2017.

3. Semenkovich CF. Regulation of fatty acid synthase (FAS). Prog Lipid Res 1997 36:43–53.

4. Loftus TM, Jaworsky DE, Frehywot GL et al. Reduced food intake and body weight in mice treated with fatty acid synthase inhibitors. Science 2000 288:2379–2381.

5. Thupari JN, Kim EK, Moran TH, Ronnett GV, Kuhajda FP. Chronic C75 treatment of diet- induced obese mice increases fat oxidation and reduces food intake to reduce adipose mass. Am J Physiol Endocrinol Metab 2004 287:E97–E104.

6. Mobbs CV, Makimura H. Block the FAS, lose the fat. Nat Med 2002 8:335–336.

7. Thupari JN, Landree LE, Ronnett GV, Kuhajda FP. C75 increases peripheral energy utilization and fatty acid oxidation in diet-induced obesity. Proc Natl Acad Sci USA 2002 99:9498–9502.

CHAPTER 4 FASN 76

8. Norman RA, Tataranni PA, Pratley R et al. Autosomal genomic scan for loci linked to obesity and energy metabolism in Pima Indians. Am J Hum Genet 1998 62:659–668.

9. Lillioja S, Mott DM, Spraul M et al. Insulin resistance and insulin secretory dysfunction as precursors of non-insulin-dependent diabetes mellitus. Prospective studies of Pima Indians. N Engl J Med 1993 329:1988–1992.

10. Knowler WC, Bennett PH, Hamman RF, Miller M. Diabetes incidence and prevalence in Pima Indians: a 19-fold greater incidence than in Rochester, Minnesota. Am J Epidemiol 1978 108:497–505.

11. Kovacs P, Harper I, Hanson RL et al. A novel missense substitution (Val1483Ile) in the fatty acid synthase gene (FAS) is associated with percentage of body fat and substrate oxidation rates in nondiabetic Pima Indians. Diabetes 2004 53:1915–1919.

12. Körner A, Ma L, Franks PW et al. Sex-specific effect of the Val1483Ile polymorphism in the fatty acid synthase gene (FAS) on body mass index and lipid profile in Caucasian children. Int J Obes (Lond) 2007 31:353–358.

13. Berndt J, Kovacs P, Ruschke K et al. Fatty acid synthase gene expression in human adipose tissue: association with obesity and type 2 diabetes. Diabetologia 2007 50:1472–1480.

14. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2006 29:S43–S48.

15. Blüher M, Unger R, Rassoul F, Richter V, Paschke R. Relation between glycaemic control, hyperinsulinaemia and plasma concentrations of soluble adhesion molecules in patients with impaired glucose tolerance or Type II diabetes. Diabetologia 2002 45:210–216.

16. Defronzo RA, Tobin JD, Andres R. Glucose Clamp Technique - Method for Quantifying Insulin-Secretion and Resistance. Am J Physiol 1979 237:E214–E223.

17. Reich A, Müller G, Gelbrich G, Deutscher K, Gödicke R, Kiess W. Obesity and blood pressure-results from the examination of 2365 schoolchildren in Germany. Int J Obes Relat Metab Disord 2003 27:1459–1464.

18. Kromeyer-Hauschild K et al. Percentiles of body mass index in children and adolescents evaluated from different regional German studies. Monatsschr Kinderheilkd 2001 149:807– 818.

19. Böttner A, Kratzsch J, Müller G et al. Gender differences of adiponectin levels develop during the progression of puberty and are related to serum androgen levels. J Clin Endocrinol Metab 2004 89:4053–4061.

20. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 2001 68:978–989.

21. Stephens M, Donnelly P. A comparison of bayesian methods for haplotype reconstruction from population genotype data. Am J Hum Genet 2003 73:1162–1169.

22. de Bakker PI, Yelensky R, Pe’er I et al. Efficiency and power in genetic association studies. Nat Genet 2005 37:1217–1223.

23. Emilsson V, Thorleifsson G, Zhang B et al. Genetics of gene expression and its effect on disease. Nature 2008 452:423–428.

CHAPTER 4 FASN 77

24. Chirala SS, Jayakumar A, Gu ZW, Wakil SJ. Human fatty acid synthase: role of interdomain in the formation of catalytically active synthase dimer. Proc Natl Acad Sci USA 2001 98:3104–3108.

25. Moreno-Navarrete JM, Botas P, Valdes S et al. Val1483Ile in FASN gene is linked to central obesity and insulin sensitivity in adult Caucasian men. Obesity 2009 17:1755-1761.

26. Havel PJ, Kasim-Karakas S, Dubuc GR, Mueller W, Phinney SD. Gender differences in plasma leptin concentrations. Nat Med 1996 2:949–950.

27. Mattevi VS, Zembrzuski VM, Hutz MH. A resistin gene polymorphism is associated with body mass index in women. Hum Genet 2004 115:208–212.

28. Ridderstråle M, Carlsson E, Klannemark M et al. FOXC2 mRNA Expression and a 5’ untranslated region polymorphism of the gene are associated with insulin resistance. Diabetes 2002 51:3554–3560.

29. Ma L, Tataranni PA, Hanson RL et al. Variations in peptide YY and Y2 receptor genes are associated with severe obesity in Pima Indian men. Diabetes 2005 54:1598–1602.

30. Tönjes A, Zeggini E, Kovacs P et al. Association of FTO variants with BMI and fat mass in the self-contained population of Sorbs in Germany. Eur J Hum Genet 2010 18:104-110.

4.10 Supplement

Supplement Table 1: Measures of linkage disequilibrium (L.D.; D prime (D’) and r2) among FASN SNPs with minor allele frequency > 0.05. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1 1 0.99 0.99 0.99 1.00 0.79 0.99 0.99 0.99 0.80 0.99 0.91 0.90 0.82 0.60 0.59 2 0.49 1 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.88 0.79 0.78 0.37 0.29 3 0.19 0.10 1 0.42 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.64 0.99 0.99 0.99 4 0.55 0.86 0.02 1 0.99 0.62 0.99 0.99 0.27 0.56 0.26 0.86 0.74 0.77 0.45 0.37 5 1.00 0.52 0.17 0.59 1 0.77 0.99 0.99 0.99 0.78 0.99 0.91 0.90 0.81 0.58 0.52 6 0.14 0.10 0.04 0.04 0.14 1 0.99 0.99 0.99 0.91 0.99 0.99 0.75 0.13 0.99 0.99 7 0.42 0.86 0.08 0.74 0.45 0.09 1 0.99 0.99 0.99 0.99 0.93 0.76 0.71 0.27 0.18 8 0.95 0.47 0.18 0.52 0.95 0.25 0.40 1 0.99 0.99 0.99 0.89 0.87 0.74 0.63 0.56 9 0.17 0.09 0.92 0.01 0.15 0.04 0.07 0.17 1 1.00 1.00 0.99 0.78 0.99 0.99 0.99 10 0.13 0.10 0.04 0.03 0.13 0.82 0.09 0.23 0.04 1 1.00 0.99 0.99 0.91 0.99 0.99 11 0.18 0.10 0.91 0.01 0.16 0.04 0.08 0.17 1.00 0.04 1 0.99 0.79 0.92 0.99 0.99 12 0.46 0.77 0.08 0.63 0.46 0.11 0.78 0.44 0.07 0.13 0.08 1 0.92 0.99 0.26 0.22 13 0.71 0.39 0.08 0.39 0.67 0.10 0.31 0.65 0.10 0.20 0.11 0.55 1 0.99 0.55 0.56 14 0.16 0.28 0.04 0.26 0.16 0.01 0.27 0.12 0.04 0.04 0.04 0.45 0.29 1 0.74 0.75 15 0.19 0.04 0.10 0.06 0.18 0.08 0.02 0.24 0.10 0.10 0.09 0.02 0.14 0.07 1 0.99 16 0.18 0.02 0.11 0.04 0.15 0.09 0.01 0.19 0.09 0.11 0.09 0.01 0.14 0.07 0.95 1

D’ given in upper shaded boxes and r2 given in lower boxes. 1 = rs59251877; 2 = rs59638227, 3 = rs62078749, 4 = rs62078748, 5 = rs62078747, 6 = -605C>T, 7 = rs17848934, 8 = rs2228309, 9 = rs17848935, 10 = rs4485435, 11 = rs41283365, 12 = rs2229422, 13 = rs17848939, 14 = rs2229425, 15 = rs4246445, 16 = rs11550611.

CHAPTER 4 FASN 78

Supplement Table 2: Linkage disequilibrium (L.D.) groups of FASN variants among 48 unrelated Caucasian subjects.

L.D. Group FASN genetic variants 1 rs59251877, rs62078747, rs2228309 2 rs59638227, rs62078748, rs17848934 3 rs62078749, rs17848935, rs41283365 4 -605C>T, rs4485435 5 rs4246445, rs11550611 6 rs2229422 7 rs17848939 8 rs2229425

Variants in each group are in L.D. (r2 ≥ 0.8, D’ = 1.0). Representative SNPs which were genotyped for association studies are in bold.

Supplement Table 3: Association of FASN SNPs with type 2 diabetes (T2D). SNPs Controls Cases p-value p-value OR (95 %CI) NGT T2D (unadjusted) (adjusted for age, (N = 502) (N = 640) sex and BMI) rs62078748 GG 50.5 % 44.1 % GA 42.4 % 42.9 % 0.003 0.10 1.22 (0.97; 1.55) AA 7.1 % 13 % MAF 28.3 % 34.5 % rs2228309 CC 26.4 % 33.2 % CT 51.7 % 49.3 % 0.03 0.04 0.79 (0.63; 0.99) TT 21.9 % 17.6 % MAF 47.8 % 42.2 % rs17848935 CC 65.9 % 63.9 % CT 33.8 % 32.7 % 0.38 0.19 1.28 (0.88; 1.87) TT 1.9 % 3.4 % MAF 18.8 % 19.7 % rs4485435 CC 69 % 69.2 % CG 28.5 % 28.5 % 0.99 0.77 0.96 (0.71; 1.29) GG 2.4 % 2.3 % MAF 16.7 % 16.5 % rs2229422 TT 46.4 % 41.5 % TC 44.9 % 45.1 % 0.02 0.28 1.14 (0.90; 1.44) CC 8.5 % 13.5 % MAF 31.1 % 36.0 %

CHAPTER 4 FASN 79

Supplement Table 3 cont. rs1784893 9 23.9 % 30.9 % 52.3 % 47.1 % GG 0.04 0.38 0.91 (0.73; 1.13) GA 23.7 % 22.0 % AA 49.9 % 45.5 % MAF rs2229425 CC 72.8 % 69.3 % CT 25.6 % 29.8 % 0.15 0.76 0.92 (0.68; 1.23) TT 1.6 % 0.8 % MAF 14.4 % 15.7 % rs4246445 CC 48.6 % 56.7 % CT 44.1 % 36.9 % 0.03 0.17 0.84 (0.66; 1.08) TT 7.3 % 6.4 % MAF 29.4 % 24.9 %

P-values and odds ratio (OR) with 95 % confidence intervals (CI) were calculated under logistic regression analyses in additive mode of inheritance. MAF - minor allele frequency.

Supplement Table 4: Clinical characteristics according to FASN SNP rs2229422 genotypes in subjects without type 2 diabetes (N = 671). rs2229422 TT TC CC p-value Frequency (%) 44.7 % 45.5 % 9.7 % Age (years) 46 ± 1 47 ± 1 48 ± 2 BMI (kg/m2) 26.9 ± 0.3 27.6 ± 0.3 29.3 ± 0.8 0.002 WHR 0.94 ± 0.01 0.96 ± 0.01 1.03 ± 0.03 0.007 Fasting plasma glucose 5.33 ± 0.03 5.34 ± 0.03 5.3 ± 0.06 0.28 (mmol/l) 2-h plasma glucose 6.75 ± 0.1 6.85 ± 0.11 7.29 ± 0.24 0.42 (mmol/l) Fasting plasma insulin 95 ± 10 124 ± 14 217 ± 47 0.03 (pmol/l) Glucose infusion rate 68.11 ± 2.52 64.81 ± 2.6 50.66 ± 6.28 0.04 (µmol/kg/min) % body fat 28.62 ± 0.93 29.46 ± 0.92 32.0 ± 2.1 0.12 HbA1c (%) 5.46 ± 0.02 5.46 ± 0.02 5.58 ± 0.06 0.59

P-values for multivariate linear relationships were assessed by generalized linear model analyses in additive mode of inheritance (after adjusting for age and sex for variables BMI, % body fat and WHR, and for age, sex and BMI for remaining traits).

CHAPTER 4 FASN 80

4.11 Contribution of Authors Dorit Schleinitz Part of the study design, main part of the statistical analyses, design and writing of the manuscript, genotyping of the children cohort, sequencing

Nora Klöting RNA preparation, real-time PCR

Antje Körner Recruiting and phenotyping the children cohort, DNA preparation for the children cohort

Janin Berndt Real-time PCR

Marlene Reichenbächer Genotyping of the adult cohort, contribution to the statistical analyses

Anke Tönjes Contribution to the discussion, proofreading of the manuscript

Karen Ruschke Proofreading of the manuscript

Yvonne Böttcher Contribution to the discussion, proofreading of the manuscript

Kerstin Dietrich Proofreading of the manuscript

Beate Enigk DNA preparation for the adult cohort

Matthias Filz Sequenzing

Michael R. Schön Surgery (preparation of the biopsies)

Jost Jenkner Surgery (preparation of the biopsies)

Wieland Kiess Recruiting and phenotyping of the children cohort

Michael Stumvoll Project idea, proofreading of the manuscript

Matthias Blüher Project idea, proofreading of the manuscript

Peter Kovacs Project idea, part of the study design, part of the statistical analyses, editing of the manuscript

CHAPTER 5 81

CHAPTER 5

Role of Genetic Variation in the Cannabinoid Type 1 Receptor Gene (CNR1) in the Pathophysiology of Human Obesity

Dorit Schleinitz1, Solveig Carmienke2, Yvonne Böttcher2, Anke Tönjes2,3, Janin Berndt1, Nora Klöting2, Beate Enigk1, Ines Müller1,2, Kerstin Dietrich1, Jana Breitfeld1, Gerhard H. Scholz4,5, Stefan Engeli6, Michael Stumvoll2, Matthias Blüher2 and Peter Kovacs1

1Interdisciplinary Centre for Clinical Research, Medical Faculty, University of Leipzig, Leipzig, Germany 2Department of Medicine, University of Leipzig, Leipzig, Germany 3Coordination Centre for Clinical Trials, University of Leipzig, Leipzig, Germany 4St. Elisabeth Hospital, Department of Internal Medicine I, Leipzig, Germany 5Institute for Preventive Medicine GmbH, Leipzig, Germany 6Institute of Clinical Pharmacology, Hannover Medical School, Hannover, Germany

Published in: Pharmacogenomics 2010 11:693-702

CHAPTER 5 CNR1 82

5.1 Abstract Aims The endocannabinoid system may contribute to the association of visceral fat accumulation with metabolic diseases. Here we investigated the effects of genetic variation in the cannabinoid type 1 receptor gene (CNR1) on its mRNA expression in adipose tissue from visceral and subcutaneous depots and on the development of obesity. Materials & Methods CNR1 was sequenced in 48 non related German Caucasians to detect genetic variation. Five representative variants including HapMap tagging single nucleotide polymorphisms (SNPs; rs12720071, rs806368, rs806370, rs1049353, rs806369) were genotyped for subsequent association studies in two independent cohorts (total N = 2774) with detailed metabolic testing: subjects from the Leipzig Study (N = 1857) and a self-contained population of Sorbs from Germany (N = 917). Results In a case-control study of lean (BMI < 25 kg/m2) vs. obese (BMI > 30 kg/m2) subjects, rs806368 was found to be nominally associated with obesity in the Sorbs cohort (adjusted p < 0.05), but not in the Leipzig cohort. Also, several SNPs (rs806368, rs806370, rs12720071) were nominally associated with serum leptin levels (p < 0.05 after adjusting for age, sex and BMI). However, none of these associations remained significant after accounting for multiple testing. Furthermore, none of the SNPs was related to CNR1 mRNA expression in visceral and subcutaneous fat. Conclusions The data suggest that common genetic variation in the CNR1 does not influence mRNA expression in adipose tissue nor does it play a significant role in the pathophysiology of obesity in German and Sorbian populations.

5.2 Introduction Investigations into the mechanism of action of the major cannabis-derived psychotropic compound Δ9-tetrahydrocannabinol (THC) led to discovery of the endocannabinoid system (ECS) in the early 1990s (1). The ECS includes the cannabinoid receptor type 1 (CNR1) and 2 (CNR2), the endocannabinoids (ECs) anandamide and 2-arachidonylglycerol (2-AG) and the EC anabolic and catabolic enzymes. The system participates in the control of food intake via both central and peripheral mechanisms, in lipid and glucose metabolism, and it may also stimulate lipogenesis and fat accumulation (2,3). CNR1 is present in human omental and subcutaneous adipocytes (4,5) and EC and CB1 levels are increased during adipocyte differentiation (6-9). Furthermore, CNR1 gene deficient mice are lean and resistant to diet-induced obesity (10,11). The EC system seems to be upregulated in human obesity (4) and abdominal fat accumulation is a critical correlate of dysregulation of the peripheral EC system (4,12,13). The importance of the ECS in the pathophysiology of

CHAPTER 5 CNR1 83 obesity is further supported by the fact that pharmacologic blockade of the CNR1 with rimonabant leads to significant reduction of food intake and weight loss both in rodents and humans (2,14). Since maintenance of body weight is under genetic control modified by environmental factors, genetic variants affecting the EC system might be relevant players in the pathophysiology of obesity. It has been shown that a missense polymorphism (cDNA 385 C→A) in fatty acid amide hydrolase (FAAH), one of the EC degrading enzymes, is associated with overweight and obesity in subjects of white and black ancestry, but not in Asians (15). By contrast, Jensen et al. did not find robust evidence of an association of the Pro129Thr FAAH variant with overweight or obesity, and any related quantitative traits in Danish subjects (16). Similarly, there have been controversial findings regarding the role of CNR1 gene variants in the pathophysiology of obesity. While several study groups reported modest associations of CNR1 genetic variants with obesity and related traits (17-20), other studies could not confirm these findings (21,22). Recently, we demonstrated that CNR1 mRNA expression in visceral adipose tissue correlates with circulating 2-AG levels, percent body fat, visceral fat area, fasting plasma insulin (FPI) and insulin sensitivity determined by euglycemic-hyperinsulinemic clamps (12). Therefore, the aim of our present study was to investigate the influence of CNR1 SNPs on CNR1 mRNA expression in subcutaneous and visceral adipose tissue and on parameters related to the EC system. We sequenced CNR1 to detect new variation and examined the effects of CNR1 genetic variants on subphenotypes relevant to obesity in two distinct cohorts: one cohort from the city of Leipzig and one from the self-contained population of Sorbs in Germany.

5.3 Materials & methods 5.3.1 Leipzig cohort A total of 919 subjects without Type 2 diabetes (T2D; 739 subjects with normal glucose tolerance [NGT], 180 subjects with impaired glucose tolerance [IGT]) and 938 patients with T2D, consecutively recruited at the University Hospital in Leipzig, Germany, were included in the study. The nondiabetic group included 304 males and 615 females (mean age 49.2 ± 13.7 years; mean body mass index (BMI) 28.14 [27.81; 28.48] kg/m2; mean waist- to-hip ratio (WHR) 0.94 ± 0.17; mean fasting plasma glucose (FPG) 5.28 ± 0.56 mmol/l; mean fasting plasma insulin (FPI) 45.30 pmol/l [40.70; 50.40]). The patients with T2D included 484 males and 454 females (mean age 65 ± 11 years; mean BMI 29.20 [28.88; 29.52] kg/m2; mean WHR 1.12 ± 0.14) (for normally distributed parameters data are given as

CHAPTER 5 CNR1 84 arithmetic means ± standard deviation (SD) and for ln-normally distributed parameters the data represent geometric means and 95 % confidence intervals). Oral glucose tolerance tests (OGTT) were performed in all non-diabetic subjects as described previously (23). Since impaired glucose tolerance is a T2D predicting factor, only subjects with normal glucose tolerance (N = 739) were included as healthy controls in the T2D case-control study. In a subgroup of 396 NGT subjects, insulin sensitivity was assessed by euglycemic-hyperinsulinemic clamp performed as previously described (24). Percent body fat was measured by dual X-ray absorptiometry (DEXA). Only subjects without T2D were included in association analyses with SNPs in the obesity case-control study (N = 919; BMI ≤ 25 kg/m2, N = 313; BMI > 25 ≤ 30 kg/m2, N = 301; BMI > 30 kg/m2, N = 305) and in quantitative traits analyses. In a subgroup of 60 subjects, CNR1 mRNA, FAAH mRNA, sterol regulatory element-binding protein (SREBP1) mRNA expression in Vis and SC adipose tissue were available in addition to serum 2-AG and anandamide levels to analyze CNR1 genotype- phenotype associations (12).

5.3.2 Measurement of obesity & glucose metabolism parameters BMI was calculated as weight (in kg) divided by the square of height (in meters). Waist and hip circumferences were measured and WHR was calculated. Basal blood samples were taken after an overnight fast. Plasma insulin was measured with an enzyme immunometric assay for the IMMULITE® automated analyzer (Diagnostic Products Corporation, CA, USA). Plasma leptin levels were assessed by radioimmunoassay (Linco Research, MO, USA).

5.3.3 Sorbs cohort All subjects are part of a sample from a Sorbian isolate in Germany, the Sorbs. At present, approximately 1000 Sorbian individuals are enrolled in the study (25,26). Extensive phenotyping included standardized questionnaires for past medical history and family history, collection of anthropometric data (weight, height, WHR, body impedance analysis) and an OGTT. Insulin was measured with the AutoDELFIA® Insulin assay (PerkinElmer Life and Analytical Sciences, MA, USA). Serum glucose was measured by the hexokinase method (Automated analyzer Modular, Roche Diagnostics, Mannheim, Germany). A total of 917 Sorbian subjects (534 female, 383 male; 725 subjects with NGT, 94 subjects with IGT and 98 cases with T2D) were available for the present study (mean age 48 ± 16 years; mean BMI 26.98 ± 4.93 kg/m2; mean WHR 0.88 ± 0.29; mean FPG 5.52 ± 1.13 mmol/l;

CHAPTER 5 CNR1 85 mean FPI 33.65 [32.32; 35.03] pmol/l). Only subjects without T2D were included in association analyses with SNPs in the obesity case-control study (N = 819; BMI ≤ 25 kg/m2, N = 337; BMI > 25 but ≤ 30 kg/m2, N = 323; BMI > 30 kg/m2, N = 159) and in quantitative traits analyses. All studies were approved by the ethics committee of the University of Leipzig (Leipzig, Germany) and all subjects gave written informed consent before taking part in the study.

5.3.4 DNA extraction & sequencing Genomic DNA was extracted using QIAmp® DNA Blood Mini Kit and QIAmp DNA Blood Midi Kit (Qiagen Inc., CA, USA) according to the manufacturer’s protocol. CNR1 maps to chromosome 6q14–15. To identify genetic variants approximately 750 bp 5’ upstream of exon 1, the exon 1 UTR, intron/exon splicing sites, the entire coding sequence (according to Ensembl ENST00000369501/ENSG00000118432 and NCBI GeneBank NM_016083), as well as 510 bp of the 3’-UTR were sequenced using the BigDye® Terminator (Applied Biosystems Inc., CA, USA) on an automated DNA capillary sequencer (ABI PRISM® 3100 Avant; Applied Biosystems Inc.). Sequence information for all oligonucleotide primers used for variant screening is available upon request. Sequence analysis was carried out using the SeqScape® software for Mutation Profiling version 2.1.1 (Applied Biosystems Inc.)

5.3.5 Genotyping of CNR1 SNPs Genotyping of selected SNPs was carried out using the human predesigned TaqMan® SNP genotyping assays (Applied Biosystems Inc.) for rs806370, rs806368, rs12720071, rs1049353 and rs806369. The genotyping reaction was amplified on a GeneAmp® PCR 2720 system (Applied Biosystems) (95 °C for 15 min, 95 °C for 15 s and 60 °C for 1 min for 40 cycles), using the ABsolute™ Blue QPCR Low ROX Mix (Thermo Fisher Scientific, MA, USA), and fluorescence was detected on an ABI PRISM® 7500 sequence detector (Applied Biosystems Inc.). To assess genotyping reproducibility, a random selection of approximately 5 % of the samples was re-genotyped in all SNPs; all genotypes matched initial designated genotypes. In addition, 48 DNA samples used for sequencing were also genotyped. Again, genotypes matched initial designated genotypes obtained by direct sequencing, thus validating the accuracy of the genotyping method.

CHAPTER 5 CNR1 86

5.3.6 Statistical analysis Prior to statistical analysis, non-normally distributed parameters were logarithmically transformed to approximate a normal distribution. For all normally distributed parameters data are given as arithmetic means ± SD, and for all logarithmically-normally distributed parameters the data represent geometric means and 95 % confidence intervals (CIs). Differences in genotype frequencies between the diabetic cases versus the healthy controls and the obese cases versus the lean controls were compared using logistic regression analyses, respectively. Multivariate linear relationships were assessed by generalized linear regression models. All analyses were carried out under the additive model unless stated otherwise and the presented p-values were adjusted for age, sex and BMI (for WHR, BMI and percent body fat as dependent variables, the analyses were adjusted for age and sex only). P-values of less than 0.05 were considered to provide nominal evidence for association and are presented without correction for multiple testing. The analysis of association with quantitative traits was restricted to nondiabetics to avoid diabetes status or treatment masking potential effects of the variants on these phenotypic traits. In the additive model, homozygotes for the major allele (MM), heterozygotes (Mm), and homozygotes for the minor allele (mm) were coded to a continuous numeric variable for genotype. A dominant model was defined as contrasting genotypic groups Mm + mm versus MM, and the recessive model was defined as contrasting genotypic groups MM + Mm versus mm. To assess the combined SNP effect in our two cohorts we performed a meta-analysis by using the metan command in STATA version 9.0 (StataCorp LP, TX, USA) based on the estimated effect sizes of each study, their standard error for quantitative traits and their confidence intervals for discrete traits (obesity). The meta-analysis was performed in a fixed-effects model by using the Mantel–Haenszel method. Statistical analyses were performed using SPSS version 15.0.1 (SPSS, Inc., IL, USA) and STATA version 9.0 (StataCorp LP). Statistical power was calculated with Quanto 1.23 (27,28).

5.4 Results 5.4.1 Sequencing The CNR1 was sequenced in DNA samples of 48 nonrelated subjects from Leipzig, Germany (12 healthy lean subjects, 12 lean subjects with T2D, 12 obese subjects with IGT and 12 obese subjects with T2D). Three novel variants were identified (-21029G>A, minor allele frequency (MAF) 0.011; -20803C>T, MAF 0.021; -20513A>G, MAF 0.011; positions relative to the translation initiation site).

CHAPTER 5 CNR1 87

5.4.2 Association studies CNR1 polymorphisms with MAF greater than 0.05 were taken forward to genotype– phenotype association analyses. Since all SNPs identified by sequencing had MAF of less than 0.05, only HapMap (101) tagging SNPs with MAF greater than 0.05 (rs806370, rs806369, rs1049353, rs12720071, rs806369; selection based on linkage disequilibrium with r2 > 0.8) were genotyped for subsequent genotype–phenotype association analyses. Genotype distribution for all SNPs was in Hardy–Weinberg equilibrium (p > 0.001). MAFs were comparable between both cohorts (Leipzig and Sorbs) and the call rates ranged from 93 to 99% (Table 1).

Table 1: CNR1 HapMap tagging SNPs. MAF Genotyping call rate SNP Change Leipzig cohort Sorbs cohort Leipzig cohort Sorbs cohort rs12720071 T>C 0.102 0.115 0.965 0.992 rs806368 A>G 0.220 0.219 0.955 0.977 rs806370 G>A 0.131 0.125 0.953 0.991 rs1049353 C>T 0.260 0.249 0.948 0.984 rs806369 G>A 0.298 0.304 0.944 0.973

MAF: minor allele frequency

5.4.3 Leipzig cohort None of the variants were associated with obesity in a case–control study (lean vs. obese) in 618 subjects without T2D (BMI ≤ 25 kg/m2, N = 313; BMI > 30 kg/m2, N = 305; adjusted for age and sex; Table 2). Inclusion of subjects with T2D into the regression models to increase sample size and statistical power did not change the results (data not shown).

Table 2: Case-control analysis for obesity. SNP Subjects At-risk allele Odds ratio (95 % CI) p-value rs12720071 Leipzig cohort - 1.04 (0.67; 1.61) 0.88 T>C Sorbs cohort - 1.04 (0.64; 1.70) 0.87 Combined analyses 1.04 (0.75; 1.44) 0.84 rs806368 Leipzig cohort - 1.08 (0.79; 1.48) 0.62 A>G Sorbs cohort G 1.51 (1.03; 2.21) 0.03* Combined analyses 1.24 (0.97; 1.57) 0.08 rs806370 Leipzig cohort - 1.14 (0.77; 1.68) 0.51 G>A Sorbs cohort - 1.36 (0.87; 2.13) 0.17 Combined analyses 1.23 (0.92; 1.65) 0.17 rs1049353 Leipzig cohort - 0.86 (0.63; 1.15) 0.31 C>T Sorbs cohort - 0.89 (0.62; 1.27) 0.52 Combined analyses 0.87 (0.69; 1.09) 0.22 rs806369 Leipzig cohort - 0.98 (0.72; 1.33) 0.91 G>A Sorbs cohort - 1.07 (0.76; 1.50) 0.72 Combined analyses 1.02 (0.81; 1.28) 0.90

CHAPTER 5 CNR1 88

Table 2 cont.: Case-control study in lean (BMI ≤ 25 kg/m2) vs. obese (BMI > 30 kg/m2) subjects. Odds ratios represent the effects of minor alleles. *significant in a dominant mode of inheritance.

Association analysis of genetic variants on quantitative traits (BMI, WHR, percent body fat, FPG and FPI, 2-h plasma glucose, glucose infusion rate, lipids, leptin and adiponectin) in 919 subjects without T2D demonstrated nominal association of the rs12720071 C allele, rs806368 G allele and rs806370 A allele with lower leptin levels (all p < 0.05, Table 3). This did not withstand Bonferroni correction for multiple testing, which would require p-values of 0.001 to be considered statistically significant (accounting for five SNPs and eight traits). However, since Bonferroni correction is rather strict (i.e., potentially biologically interesting association may be rejected) and the five SNPs in CNR1 are not completely independent (some degree of linkage disequilibrium – r2 reaching up to 0.50), we also analyzed data by considering only eight quantitative traits when adjusting for multiple testing. Under these conditions we would achieve statistical significance for the association between rs806368 and circulating leptin (p = 0.003) (Table 3). None of the other traits showed evidence for association.

Table 3: Association of CNR1 SNP variants with quantitative traits in the Leipzig (A) and the Sorbs (B) cohort. Fasting 2-h plasma Fasting % body Leipzig cohort BMI WHR Leptin AdipoQ PG glucose insulin fat rs127200711 p-value 0.959 0.782 0.458 0.100 0.146 0.1001 0.0071 0.9311 β 0.001 -0.004 0.007 -0.036 0.182 -0.075 -0.341 0.008 SE 0.013 0.013 0.009 0.022 0.125 0.045 0.126 0.098 0.006 -0.009 0.001 -0.016 0.028 Combined β (-0.011; (-0.025; (-0.006; (-0.048; (-0.047, (95 % CI) 0.022) 0.008) 0.008) 0.015) 0.102) Combined 0.639 0.302 0.822 0.302 0.466 p-value rs806368 p-value 0.980 0.805 0.868 0.840 0.125 0.317 0.003*# 0.589 β -0.0002 0.002 0.001 -0.003 0.124 -0.029 -0.240 -0.033 SE 0.009 0.009 0.006 0.015 0.081 0.029 0.081 0.061 0.007 -0.001 0.002 <-0.001 -0.012 Combined β (-0.006; (-0.013; (-0.005, (-0.023; (-0.069; (95 % CI) 0.019) 0.010) 0.010) 0.022) 0.045) Combined 0.278 0.821 0.570 0.991 0.072 p-value rs806370 p-value 0.846 0.519 0.705 0.367 0.909 0.911 0.044 0.824 β 0.002 0.007 -0.003 0.017 -0.011 -0.004 -0.204 -0.016 SE 0.011 0.011 0.008 0.018 0.100 0.034 0.101 0.071 0.002 0.005 0.001 0.011 -0.031 Combined β (-0.013; (-0.009; (-0.008; (-0.017; (-0.101; (95 % CI) 0.018) 0.020) 0.011) 0.038) 0.039) Combined 0.949 0.469 0.758 0.447 0.381 p-value

CHAPTER 5 CNR1 89

Table 3 cont. rs1049353 p-value 0.714 0.627 0.536 0.461 0.239 0.096 0.079 0.911 β 0.003 -0.004 0.004 0.011 0.096 0.049 0.145 -0.007 SE 0.009 0.009 0.006 0.015 0.082 0.029 0.082 0.062 -0.002 0.005 -0.001 <-0.001 -0.033 Combined β (-0.014; (-0.007; (-0.009; (-0.022; (-0.086; (95 % CI) 0.010) 0.017) 0.006) 0.021) 0.021) Combined 0.735 0.409 0.733 0.983 0.229 p-value rs806369 p-value 0.413 0.934 0.300 0.692 0.248 0.448 0.757 0.697 β -0.007 -0.001 -0.006 0.006 -0.091 -0.022 0.024 0.023 SE 0.009 0.009 0.006 0.014 0.078 0.029 0.078 0.059 -0.003 -0.003 -0.002 -0.003 0.001 Combined β (-0.015; (-0.015; (-0.009; (-0.024; (-0.052; (95 % CI) 0.009) 0.008) 0.006) 0.018) 0.055) Combined 0.625 0.589 0.627 0.768 0.963 p-value

Fasting 2-h plasma Fasting Sorbs cohort BMI WHR PG glucose insulin rs12720071 p -value 0.434 0.259 0.946 0.831 0.761 β 0.009 -0.012 -0.0004 0.005 0.012 SE 0.011 0.011 0.004 0.023 0.040 rs806368 p -value 0.108 0.643 0.565 0.827 0.314 β 0.014 -0.004 0.003 0.004 -0.032 SE 0.009 0.008 0.005 0.018 0.031 rs806370 p -value 0.800 0.722 0.873 0.965 0.380 β 0.003 0.004 0.004 0.001 -0.034 SE 0.011 0.010 0.006 0.022 0.038 rs1049353 p -value 0.456 0.110 0.277 0.432 0.095 β -0.006 0.012 -0.005 -0.013 -0.049 SE 0.008 0.008 0.005 0.016 0.029 rs806369 p -value 0.965 0.490 0.759 0.358 0.638 β 0.0003 -0.005 0.001 -0.015 0.014 SE 0.008 0.008 0.005 0.016 0.029

The dataset included subjects with normal glucose tolerance and impaired glucose tolerance (N = 919, except for percent body fat, leptin and adiponectin (AdipoQ; N = 300). p-values were calculated after adjusting for age and gender for the variables BMI, WHR and percent body fat; and for age, gender and BMI for the remaining variables. ß indicates effect directions of the minor alleles. p-values are given for the additive model. 1Owing to small sample size (< 6 genotypes for minor allele homozygotes), a dominant model was used for the variables percent body fat, leptin and adiponectin for the SNP rs12720071. */# significant in dominant/recessive mode of inheritance. OGTT: oral glucose tolerance test; PG: plasma glucose; WHR: Waist –to-hip ratio.

CHAPTER 5 CNR1 90

5.4.4 Sorbs cohort The CNR1 variant rs806368 G allele was nominally associated with obesity in a case– control study (lean vs. obese; BMI ≤ 25 kg/m2, N = 337; BMI > 30 kg/m2, N = 159; Table 2). Including T2D subjects (N = 98) to increase sample size resulted in a nominal significance (p = 0.035 in a recessive mode of inheritance), which did not, however, withstand corrections for multiple testing. None of the remaining variants were associated with obesity. Association analysis of genetic variants on quantitative traits (BMI, WHR, FPG and FPI, 2-h plasma glucose, lipids, leptin and adiponectin) in 819 subjects without T2D demonstrated no association for any of the SNPs (Table 3). Furthermore, the combined analysis for obesity, including both the Sorbs and the Leipzig cohort, did not demonstrate statistically significant associations of the SNPs with tested parameters either (Table 3).

5.4.5 CNR1 genetic variants & fat-depot-specific CNR1 mRNA expression In the subgroup of 60 subjects for which CNR1 mRNA expression data in subcutaneous and visceral fat depots were available, no significant association was found between the CNR1 genotype and CNR1 mRNA expression in either depot (data not shown). However, rs806368 G allele and rs806370 A allele were nominally associated with higher 2-AG levels (p < 0.05; Figure 1) in this group. We also looked at the Duke SNPExpress database, which permits interrogation of the effects of common SNPs on exon- and transcript- level expression in two different human tissues: brain and peripheral blood mononuclear cells (29,102). None of the SNPs were associated with the transcript level of CNR1 both in brain and in peripheral blood mononuclear cells.

9 ** 8 7 6 5 4 3 2 1 0 GG/GA AA AA/AG GG 2-arachidonylglycerol (pmol/ml) rs806368 rs806370

[Figure 1]

CHAPTER 5 CNR1 91

Figure 1. Mean 2-arachidonylglycerol in nondiabetic subjects (N = 60) of the Leipzig cohort grouped by CNR1 variant genotypes. *Indicates nominal association at a p-value of less than 0.05 in generalized linear model analyses after adjusting for age, sex and BMI.

5.5 Discussion The maintenance of bodyweight is under genetic control but, rarely, mutations in single genes can result in severe obesity. In the pathophysiology of common obesity, multiple genes, environmental factors and gene–environment interactions play a crucial role (18). Since it has been demonstrated that overweight and obesity are associated with a missense polymorphism in FAAH, one of the EC degrading enzymes (15), other genes of the EC system have become potential targets for genotype–phenotype association studies. Therefore, the CNR1 was investigated as a plausible obesity candidate gene in the present study. We evaluated the effects of five representative variants within the CNR1 locus on CNR1 mRNA levels in visceral and subcutaneous adipose tissue and performed association studies in two independent cohorts with detailed metabolic testing from Eastern Germany (Leipzig cohort and a self-contained population of Sorbs). Using a case–control design with lean versus obese subjects, rs806368 was found to be nominally associated with obesity in the Sorbs cohort, but not in the Leipzig cohort. These differences between the cohorts might be attributed to different genetic structure within the examined CNR1 locus. However, linkage disequilibrium patterns in both cohorts were comparable and thus, the genetic information of these two populations seems to be well represented by the selected tagging SNPs. The association differences might rather originate from different lifestyles. Unfortunately we could not address this directly since lifestyle data were only available for the Sorbs cohort. Also, the small sample size of both cohorts might account for differences in association results. Several SNPs (rs806368, rs806370 and rs12720071) were nominally associated with circulating leptin concentrations. However, none of these associations remained significant after adjusting the analyses for multiple comparisons. Although it has been demonstrated that CNR1 mRNA expression correlates with EC serum concentrations and parameters of obesity, fat distribution and insulin sensitivity (12), no relationship between the SNPs and CNR1 mRNA expression was observed in the present study. So far, research on the relationship between CNR1 genetic variants and obesity has generated rather controversial results (Supplementary Table 1, www.futuremedicine.com/doi/suppl/10.2217/pgs.10.42/suppl_file/suppl_table.doc). In a southern Italian population, the CNR1 rs1049353 wild-type genotype (CC) demonstrated a trend towards increased BMI (18). By contrast, the T allele of this variant was found to be

CHAPTER 5 CNR1 92 associated with abdominal adiposity in obese men from Belgium (30). Furthermore, rs12720071 C allele was found to be associated with increased subscapular skinfold thickness and waist circumference in men in the Olivetti Prospective Heart Study (OPHS) (20). Another polymorphism, rs806368, was associated with WHR in a southern Brazilian population (19), but not in others (20). Müller et al. did not find evidence for an association of several SNPs in CNR1 with obesity in their samples, comprising 364 obesity trios and 235 independent obese families (22). The largest study so far, including 5750 subjects, demonstrated that CNR1 variants (rs2023239 and rs806381) increased the risk for obesity and modulated BMI in a European population of French origin (17). However, consistent with our study, no association of CNR1 SNPs with BMI, waist circumference and visceral adipose tissue or subcutaneous adipose tissue mass was observed very recently in the well-characterized Framingham study cohort (31). It needs to be mentioned that although not reaching the threshold for statistical significance for genome-wide association, one SNP (rs10485170) near CNR1 was demonstrated to be associated with obesity in a stage 2 confirmation design in a genome-wide association study reported by Meyre et al. (32). At the same time no associations of CNR1 variants have been reported in a genome-wide association study on obesity or T2D. Interestingly, we found rs806368 G and rs806370 A alleles nominally associated with lower serum leptin but higher 2-AG levels. Defective leptin signalling is associated with elevated hypothalamic but not cerebellar levels of ECs in obese db/db and ob/ob mice and Zucker rats (33). Acute leptin treatment of normal rats and ob/ob mice reduces anandamide and 2-AG in the hypothalamus. Thus Di Marzo et al. suggested that ECs in the hypothalamus may tonically activate CB1 receptors to maintain food intake and form part of the neuronal circuitry regulated by leptin (33). However, it seems unlikely that variants in CNR1 regulate leptin levels but have no effect on obesity or CNR1 expression as demonstrated for adipose tissue in our study. Our data are in accordance with most other previous studies (Supplement Table 1, www.futuremedicine.com/doi/suppl/10.2217/pgs.10.42/suppl_file/suppl_table.doc; [34,35]) and suggest that genetic variation in the CNR1 gene may have only a very modest effect on the development of human obesity. We are aware that the statistical power of our study is limited owing to small sample size. In our cohorts we had a power of 80% (at α = 0.05) to detect a difference of 0.57–1.01 kg/m2 per allele for BMI, 0.17–0.28 mmol/l for FPG, 3.99–52.30 pmol/l for FPI, 0.36–0.74 mmol/l for 2-h plasma glucose during an OGTT under the additive model. Similarly, in the case–control studies for obesity, we had a power of 80 % (at α = 0.05) to detect SNP effects with odds ratios ranging from 1.27 to 1.41 and from 1.26

CHAPTER 5 CNR1 93 to 1.37 in the Leipzig and Sorbs cohorts, respectively. Since the reported effect sizes for obesity SNPs are usually lower (odd ratios of 1.1–1.2), smaller effects may have been missed in our studies. Also, the small sample size for the mRNA expression study limits the significance of the analysis. Therefore, we cannot rule out that genetic variation in CNR1 may have minor effects on obesity development, which are probably not clinically significant. In conclusion, our data indicate that common CNR1 genetic variants may not play a significant role in the pathophysiology of obesity in German and Sorbian populations, but it is possible that rare functional variants of this gene contribute to the development of obesity. However, these variants are hard to detect owing to current sample size.

5.6 Acknowledgements The authors thank all those who participated in the studies. The authors also thank Jürgen Kratzsch for plasma insulin measurements carried out at the Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University of Leipzig, Germany.

5.7 Disclosure 5.7.1 Financial & competing interest disclosure This work was supported by grants from the Interdisciplinary Centre for Clinical Research at the University of Leipzig (B27 to Michael Stumvoll, Peter Kovacs and Anke Tönjes; N06 to Peter Kovacs, Beate Enigk, Dorit Schleinitz, Kerstin Dietrich and Jana Breitfeld). Michael Stumvoll and Matthias Blüher (Project BL 833/1–1) are supported by a grant from the Deutsche Forschungsgemeinschaft – DFG (KFO 152). The authors have no potential conflicts of financial interest in any scientific project related to the manuscript. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. No writing assistance was utilized in the production of this manuscript.

5.7.2 Ethical conduct of research The authors state that they have obtained appropriate institutional review board approval or have followed the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investigations involving human subjects, informed consent has been obtained from the participants involved.

CHAPTER 5 CNR1 94

5.8 Executive summary Executive summary • The Endocannabinoid system participates in the control of food intake, lipid and glucose metabolism. • The Endocannabinoid system seems to be upregulated in human obesity. • The cannabinoid receptor 1 (CNR1) is present in human omental and subcutaneous adipocytes. • Pharmarcologic blockade of CNR1 with rimonabant leads to significant reduction of food intake and weight loss. Material & methods • The CNR1 gene was sequenced in 48 unrelated subjects to identify genetic variants. • Five tagging SNPs with a minor allele frequency of greater than 5 % were genotyped in healthy subjects and patients with type 2 diabetes from 2 extensively phenotyped German cohorts (Leipzig cohort and Sorbs cohort). • Association studies: logistic regression analyses were used in case-control studies for obesity; for quantitative traits multivariate linear relationships were assessed by generalized linear regression models. Results • Three novel variants were identified through sequencing. • None of the variants was associated with obesity in a case-control study in the Leipzig cohort. • Association analysis of genetic variants on quantitative traits revealed nominal association of rs12720071 C allele, rs806368 G allele and rs806370 A allele with lower leptin levels. • In the Sorbs cohort, the CNR1 variant rs806368 G allele was nominally associated with obesity in a case-control study • Association analysis of genetic variants on quantitative traits demonstrated no association for any of the SNPs in the Sorbs cohort. • No significant association was found between CNR1 genotype and CNR1 mRNA expression in omental or subcutaneous fat depot in a subgroup of 60 subjects from the Leipzig cohort. Discussion • So far, research on the relationship between CNR1 genetic variants and obesity has generated rather controversial results. Conclusion • Genetic variation in the CNR1 gene may not play a significant role in the pathophysiology of obesity in German and Sorbian populations. • It can not be ruled out that rare functional variants contribute to the development of obesity; large scale association studies will be inevitable to clarify this.

5.9 References 1. Pacher P, Batkai S, Kunos G: The endocannabinoid system as an emerging target of pharmacotherapy. Pharmacol Rev 2006 58: 389–462.

2. Engeli S. Peripheral metabolic effects of endocannabinoids and cannabinoid receptor blockade. Obes Facts 2008 1:8–15.

CHAPTER 5 CNR1 95

3. Pagotto U, Marsicano G, Cota D, Lutz B, Pasquali R. The emerging role of the endocannabinoid system in endocrine regulation and energy balance. Endocr Rev 2006 27:73– 100.

4. Engeli S, Bohnke J, Feldpausch M et al. Activation of the peripheral endocannabinoid system in human obesity. Diabetes 2005 54:2838–2843.

5. Spoto B, Fezza F, Parlongo G et al. Human adipose tissue binds and metabolizes the endocannabinoids anandamide and 2-arachidonoylglycerol. Biochimie 2006 88:1889–1897.

6. Bensaid M, Gary-Bobo M, Esclangon A et al. The cannabinoid CB1 receptor antagonist SR141716 increases Acrp30 mRNA expression in adipose tissue of obese fa/fa rats and in cultured adipocyte cells. Mol Pharmacol 2003 63:908–914.

7. Gasperi V, Fezza F, Pasquariello N et al. Endocannabinoids in adipocytes during differentiation and their role in glucose uptake. Cell Mol Life Sci 2007 64:219–229.

8. Matias I, Gonthier MP, Orlando P et al. Regulation, function, and dysregulation of endocannabinoids in models of adipose and b-pancreatic cells and in obesity and hyperglycemia. J Clin Endocrinol Metab 2006 91:3171–3180.

9. Pagano C, Pilon C, Calcagno A et al. The endogenous cannabinoids stimulate glucose uptake in human fat cells via phosphatidylinositol 3-kinase and calcium-dependent mechanisms. J. Clin Endocrinol Metab 2007 92:4810–4809.

10. Cota D, Marsicano G, Tschop M et al. The endogenous cannabinoid system affects energy balance via central orexigenic drive and peripheral lipogenesis. J Clin Invest 2003 112:423– 431.

11. Ravinet Trillou C, Delgorge C, Menet C, Arnone M, Soubrie P. CB1 cannabinoid receptor knockout in mice leads to leanness, resistance to diet-induced obesity and enhanced leptin sensitivity. Int J Obes Metab Disord. 2004 28:640–648.

12. Blüher M, Engeli S, Klöting N et al. Dysregulation of the peripheral and adipose tissue endocannabinoid system in human abdominal obesity. Diabetes 2006 55:3053–3060.

13. Côté M, Matias I, Lemieux I et al. Circulating endocannabinoid levels, abdominal adiposity and related cardiometabolic risk factors in obese men. Int J Obes (Lond) 2007 31:692–699.

14. Ravinet Trillou C, Arnone M, Delgorge C et al. Anti-obesity effect of SR141716, a CB1 receptor antagonist, in diet-induced obese mice. Am J Physiol Regul Integr Comp Physiol 2003 284:R345–R353.

15. Sipe JC, Waalen J, Gerber A, Beutler E. Overweight and obesity associated with a missense polymorphism in fatty acid amide hydrolase (FAAH). Int J Obes (Lond) 2005 29:755–759.

16. Jensen DP, Andreasen CH, Andersen MK et al. The functional Pro129Thr variant of the FAAH gene is not associated with various fat accumulation phenotypes in a population-based cohort of 5,801 whites. J Mol Med 2007 85:445–449.

17. Benzinou M, Chevre JC, Ward KJ et al. Endocannabinoid receptor 1 gene variations increase risk for obesity and modulate body mass index in European populations. Hum Mol Genet 2008 17:1916–1921.

CHAPTER 5 CNR1 96

18. Gazzerro P, Caruso MG, Notarnicola M et al. Association between cannabinoid type-1 receptor polymorphism and body mass index in a southern Italian population. Int J Obes (Lond) 2007 31:908–912.

19. Jaeger JP, Mattevi VS, Callegari-Jacques SM, Hutz MH. Cannabinoid type-1 receptor gene polymorphisms are associated with central obesity in a Southern Brazilian population. Dis Markers 2008 25:67–74.

20. Russo P, Strazzullo P, Cappuccio FP et al. Genetic variations at the endocannabinoid type 1 receptor gene (CNR1) are associated with obesity phenotypes in men. J Clin Endocrinol Metab 2007 92:2382–2386.

21. Löfgren P, Sjolin E, Wahlen K, Hoffstedt J. Human adipose tissue cannabinoid receptor 1 gene expression is not related to fat cell function or adiponectin level. J Clin Endocrinol Metab 2007 92:1555–1559.

22. Müller TD, Reichwald K, Wermter AK et al. No evidence for an involvement of variants in the cannabinoid receptor gene (CNR1) in obesity in German children and adolescents. Mol Genet Metab 2007 90:429–434.

23. Böttcher Y, Teupser D, Enigk B et al. Genetic variation in the visfatin gene (PBEF1) and its relation to glucose metabolism and fat depot specific mRNA expression in humans. J Clin Endocrinol Metab 2006 91:2725–2731.

24. Blüher M, Unger R, Rassoul F, Richter V, Paschke R. Relation between glycaemic control, hyperinsulinaemia and plasma concentrations of soluble adhesion molecules in patients with impaired glucose tolerance or Type II diabetes. Diabetologia 2002 45:210–216.

25. Tönjes A, Koriath M, Schleinitz D et al. Genetic variation in GPR133 is associated with height – genome wide association study in the self-contained population of Sorbs. Hum Mol Genet 2009 18:4662–4668.

26. Tönjes A, Zeggini E, Kovacs P et al. Association of FTO variants with BMI and fat mass in the self-contained population of Sorbs in Germany. Eur J Hum Genet 2010: 18:104–110.

27. Gauderman WJ. Sample size requirements for association studies of gene–gene interaction. Am J Epidemiol 2002 155:478–484.

28. Gauderman WJ. Sample size requirements for matched case–control studies of gene– environment interaction. Stat Med 2002 21:35–50.

29. Heinzen EL, Dongliang G, Cronin KD et al. Tissue-specific genetic control of splicing: implications for the study of complex traits. PLoS Biol 2008 6:E1.

30. Peeters A, Beckers S, Mertens I, van Hul W, van Gaal L. The G1422A variant of the cannabinoid receptor gene (CNR1) is associated with abdominal adiposity in obese men. Endocrine 2007 31:138–141.

31. Lieb W, Manning AK, Florez JC et al. Variants in the CNR1 and the FAAH genes and adiposity traits in the community. Obesity (Silver Spring) 2009 17:755–760.

32. Meyre D, Delplanque J, Chèvre JC et al. Genome-wide association study for early-onset and morbid adult obesity identifies three new risk loci in European populations. Nat Genet 2009 41:157–159.

CHAPTER 5 CNR1 97

33. Di Marzo V, Goparaju SK, Wang L et al. Leptin-regulated endocannabinoids are involved in maintaining food intake. Nature 2001 410:822–825.

34. Aberle J, Fedderwitz I, Klages N, George E, Beil FU. Genetic variation in two proteins of the endocannabinoid system and their influence on body mass index and metabolism under low fat diet. Horm Metab Res 2007 39:395–397.

35. Baye TM, Zhang Y, Smith E et al. Genetic variation in cannabinoid receptor 1 (CNR1) is associated with derangements in lipid homeostasis, independent of body mass index. Pharmacogenomics 2008 9:1647–1656.

Websites 36. 101 International HapMap Projectwww.hapmap.org 37. 102 SNPExpresshttp://people.genome.duke.edu/~dg48/SNPExpress

5.10 Supplement Supplement Table 1: Summary of genetic studies investigating the role of CNR1 variants in human obesity. Reference Investigated Cohorts Main findings/Associations CNR1 SNPs Gazzerro P rs1049353 C/T 200 subjects from obesity (p = 0.03) and BMI (p = 0.006) et al. (18) southern Italy (> 65 years) Aberle J rs1049353 C/T 451 German No significant association with BMI. et al. (34) subjects (obese, dyslipidaemic, under low fat diet) Russo P rs12720071 T/C 1) 930 men from 1) rs12720071: WC (p = 0.05) et al. (20) rs806368 A/G southern Italy (part and SS (p = 0.031); of OPHS study population) 2) 216 white men 2) rs12720071: WC (p = 0.006) and BMI (part of WHSS (p = 0.012); rs806368: WC (p = 0.047) study population) Müller TD 1) 1) 364 German 1) TDT negative for these SNPs; slight trend et al. (22) rs9353527 T/C, obesity trios towards preferential transmission of rs1049353 T rs754387 G/T, (extremely obese allele rs6454676 C/T, child or adolescent rs806379 T/A, and both parents) rs1535255 C/A, rs2023239 A/G, rs806370 G/A, rs1049353 C/T 2) 2) 235 independent 2) trend from 1) not confirmed rs1049353 C/T German obese families (at least two obese sibs and both parents)

CHAPTER 5 CNR1 98

Supplement Table 1 cont. Peeters A rs1049353 C/T 1064 obese subjects Obese men: WHR (p = 0.009) and WC ( p= 0.008) et al. (30) (BMI ≥ 30 kg/m2) (after Bonferroni correction) without diabetes & IGT & 251 healthy controls from Belgium Jaeger JP rs1049353 C/T 756 Brazilian rs806368: WHR (p = 0.042; corrected for multiple et al. (19) rs12720071 T/C individuals of testing) rs806368 A/G European descent Baye TM rs806370 G/A, 1560 individuals rs806366: BMI (p = 0.025), HDL-C (p = 0.0001), et al. (35) rs806369 G/A, from 261 extended TG (P = 0.033); rs1049353 C/T, families (part of rs806368: LDL-C (p = 0.032), TG (p = 0.0001); rs12720071 TOPS study cohort; rs806369: TG (p = 0.026); C/T, predominantly rs806370: HDL-C (p = 0.038) rs806368 G/A, Northern European rs806366 A/G ancestry) Lieb W 18 SNPs 2415 subjects from No association for obesity and related traits after et al. (31) the Framingham correction for multiple testing. Offspring Study Benzinou M 1) 1) 1932 obese cases 1) nominal evidence of association with childhood et al. (17) 26 tagging and 1173 non-obese obesity and/or adult obesity for 12 SNPs SNPs controls of French (1.16 < OR < 1.40, 0.00003 < p < 0.04); rs806381 European origin and rs2023239 resisted correction for multiple (children and testing adults) 2) 2) 2645 individuals 2) BMI in Swiss obese (p = 0.015; p = 0.02) and rs806381 T/C, (Swiss obese Danish individuals (p = 0.0023; p = 0.021) rs2023239 A/G subjects and Danish individuals) 3) 3) No. 1) initial 3) rs10485170 T/C with adult obesity (p = 8x10-6) all known case-control study and rs6454674 A/C with childhood obesity variants in (p = 1x10-5) partial LD (r2>0.05) with No. 2) SNPs

Genome wide association studies are excluded. Results of haplotype analyses were not included in the table. SNP alleles are given in reverse strand orientation. SNP = single nucleotide polymorphism; IGT = impaired glucose tolerance; OR = odds ratio; WHR = waist to hip ratio; BMI = body mass index; WC = waist circumference; SS = subscapular skinfold thickness; HDL-C = high density lipoprotein cholesterol; TG = triglyceride; LDL-C = low density lipoprotein cholesterol; TDT = transmission disequilibrium test; LD = linkage disequilibrium.

CHAPTER 5 CNR1 99

5.11 Contribution of Authors Dorit Schleinitz Study design, statistical analyses and in silico analyses in project-oriented databases, design and writing of the manuscript, genotyping of the Sorb cohort

Solveig Carmienke Sequencing, genotyping of the Leipzig cohort

Yvonne Böttcher Proofreading of the manuscript

Anke Tönjes Recruiting and phenotyping of the Sorbs cohort

Janin Berndt Real-time PCR

Nora Klöting Real-time PCR

Beate Enigk Sequencing

Ines Müller Genotyping of the subcohort of subjects with expression data

Kerstin Dietrich Proofreading of the manuscript

Jana Breitfeld Proofreading of the manuscript

Gerhard H. Scholz Recruiting and phenotyping of a part of the Leipzig cohort (obese subjects with normal glucose tolerance)

Stefan Engeli Contribution to the discussion, proofreading of the manuscript

Michael Stumvoll Project idea, proofreading of the manuscript

Matthias Blüher Project idea, proofreading of the manuscript

Peter Kovacs Project idea, editing of the manuscript

CHAPTER 6 100

CHAPTER 6

Human Bone Morphogenetic Protein Receptor 2 (BMPR2) is Involved in the Pathophysiology of Obesity

Dorit Schleinitz1, Nora Klöting2, Yvonne Böttcher2, Sara Wolf2, Kerstin Dietrich1, Anke Tönjes2,3, Jana Breitfeld1, Beate Enigk1, Jan Halbritter2, Antje Körner4, Michael R. Schön5, Jost Jenkner5, Yu-Hua Tseng6, Michael Stumvoll2, Matthias Blüher2 and Peter Kovacs1

1Interdisciplinary Centre for Clinical Research, Medical Faculty, University of Leipzig, Leipzig, Germany 2Department of Medicine, University of Leipzig, Leipzig, Germany 3Coordination Centre for Clinical Trials, University of Leipzig, Leipzig, Germany 4Hospital for Children and Adolescents, University of Leipzig, Leipzig, Germany 5Municipal Clinic Karlsruhe, Clinic of Visceral Surgery, Karlsruhe, Germany 6Joslin Diabetes Center, Harvard Medical School, Boston, MA, USA

Going to be published in: PLoS ONE [accepted]

CHAPTER 6 BMPR2 101

6.1 Abstract Objective: Human bone morphogenetic protein receptor 2 (BMPR2) is essential for BMP signalling and may be involved in the regulation of adipogenesis. The BMPR2 locus has been suggested as target of recent selection in human populations. We hypothesized that BMPR2 might have a role in the pathophysiology of obesity. Research design & methods: Evolutionary analyses (dN/dS, Fst, iHS) were conducted in vertebrate and human populations. BMPR2 mRNA expression was measured in 190 paired samples of visceral and subcutaneous adipose tissue. The gene was sequenced in 48 DNA samples. Nine representative single nucleotide polymorphisms (SNPs) were genotyped for subsequent association studies on quantitative traits related to obesity in 1830 German Caucasians. An independent cohort of 925 Sorbs was used for replication. Finally, relation of genotypes to mRNA in fat was examined. Results: The evolutionary analyses indicated signatures of selection on the BMPR2 locus. BMPR2 mRNA expression was significantly increased both in visceral and subcutaneous adipose tissue of 37 overweight (BMI > 25 and < 30 kg/m2) and 80 obese (BMI > 30 kg/m2) compared with 44 lean subjects (BMI < 25 kg/m2) (p < 0.001). In a case- control study including lean and obese subjects, two intronic SNPs (rs6717924, rs13426118) were associated with obesity (adjusted p < 0.05). Combined analyses including the initial cohort and the Sorbs confirmed a consistent effect for rs6717924 (combined p = 0.01) on obesity. Moreover, rs6717924 was associated with higher BMPR2 mRNA expression in visceral adipose tissue. Conclusion: Combined BMPR2 genotype-phenotype-mRNA expression data as well as evolutionary aspects suggest a role of BMPR2 in the pathophysiology of obesity.

6.2 Introduction Despite recent advances achieved by genome wide association studies (1-3), the genetic factors contributing to the development of obesity are still elusive. The prevalence of obesity has increased particularly during the past 50 years. Since it is rather unlikely that human genotypes changed significantly during this period, it was probably the advent of Western lifestyle and unrestricted food availability that caused this dramatic increase. A popular theory to explain this observation is the “thrifty genotype” hypothesis (4) proposing that humans with a genetically increased capacity to store energy had an evolutionary advantage, since they were more likely to survive periods of food scarcity. However, in modern societies with easy access to high-caloric food and very little physical activity this genotype results in obesity. Although the hypothesis has been widely accepted, a clear

CHAPTER 6 BMPR2 102 approach to identify “thrifty” genes has not yet been postulated. However, the availability of high density single nucleotide polymorphism (SNP) genotyping arrays as well as rapidly emerging bioinformatics tools offer a possibility to reveal selective forces on loci dispersed throughout the genome. Employing these approaches, bone morphogenetic protein receptor 2 gene (BMPR2) was suggested as one of the potential targets of recent selection in European populations (5). Bone morphogenetic proteins (BMPs) are members of the transforming growth factor-β (TGF-β) superfamily and are involved in the control of multiple key steps of embryonic development and differentiation (6-8). BMPs are involved in adipocyte development, including adipose cell fate determination, differentiation of committed preadipocytes, and function of mature adipocytes (9). BMP signalling requires the formation of heteromeric complexes of BMP type 1 (BMPR1) and type 2 receptors (BMPR2) (10-12). Following BMP binding either to a single receptor type subunit followed by the recruitment of the second subunit or binding to a pre-existing loose complex of both receptor types, two downstream signalling cascades are activated: 1) SMAD proteins or 2) p38 mitogen activated protein kinase (p38MAPK) pathways (13). Over-expression of constitutively active BMP receptor 1A or 1B induces the commitment of C3H10T1/2 stem cells into adipocytes even in the absence of BMP2/4 (14). We could recently show that BMPR1A mRNA expression in both visceral and subcutaneous adipose tissue as well as genetic variants in this gene strongly correlated with obesity and its related traits (15). BMPR2, a serine threonine kinase located on chromosome 2q33-q34, is responsible for the trans-phosphorylation of BMPR1, further suggesting BMPR2 as candidate gene for obesity. Here, we initiated studies searching for signatures of selection for BMPR2 in vertebrates and human populations. We analysed the coding region of BMPR2 orthologs with the Phylogenetic Analysis by Maximum Likelihood (PAML). Furthermore, we used Haplotter/PhaseII database to screen for selection patterns within the BMPR2 locus including intronic regions. Since BMPR2 is also a plausible obesity candidate gene, we searched for evidence supporting the “thrifty gene” hypothesis by investigating its potential role in the pathophysiology of human obesity. We measured the expression of BMPR2 mRNA in paired samples of human visceral and subcutaneous adipose tissue in subjects who had undergone detailed metabolic testing. To determine, whether genetic variants in the BMPR2 are related to adipose tissue BMPR2 mRNA expression and to the obese human phenotype, we sequenced the BMPR2 and evaluated the association between representative variants and obesity related traits in two independent Caucasian cohorts from Germany.

CHAPTER 6 BMPR2 103

6.3 Research design and methods 6.3.1 Study subjects Leipzig cohort: A total of 930 patients with type 2 diabetes (T2D) and 900 non- diabetic subjects recruited at the University Hospital in Leipzig (Germany) were included in the study. Healthy subjects included 288 males and 612 females (mean age 49 ± 14 years; mean BMI 28.7 ± 5.7 kg/m2; mean WHR 0.94 ± 0.17; mean fasting plasma glucose 5.29 ± 0.57 mmol/l; mean fasting plasma insulin 124 ± 228 pmol/l). Patients with T2D included 475 males and 455 females (mean age 64 ± 11 years; mean BMI 29.7 ± 5.4 kg/m2) (data are given as arithmetic means ± SD). In addition, oral glucose tolerance test (OGTT) and fasting plasma insulin measurements were performed in all non-diabetic subjects as previously described elsewhere (16). In a subgroup of 374 non-diabetic subjects, insulin sensitivity was assessed with euglycemic-hyperinsulinemic clamps and plasma leptin and adiponectin measurements were carried out.

Tissue studies: Paired samples of visceral and subcutaneous adipose tissue were obtained from a subgroup of 190 Caucasian men (N = 91) and women (N = 99), who underwent open abdominal surgery (described in detail elsewhere) (16). The age ranged from 16 to 99 years and body mass index from 21 to 55 kg/m2. In these subjects, in addition to above mentioned clinical parameters, abdominal visceral and subcutaneous fat area was calculated using computed tomography scans at the level of L4-L5 and percent body fat was measured by dual-energy X-ray absorptiometry (DEXA).

Assays and measures of obesity and glucose metabolism: Fasting plasma insulin was measured with an enzyme immunometric assay for the IMMULITE automated analyzer (Diagnostic Products Corporation, Los Angeles, CA, USA). Plasma leptin levels were assessed by radioimmunoassay (Linco Research, St. Charles, Mo, USA). The OGTT was performed after an overnight fast with 75 g standardized glucose solution (Glucodex Solution 75 g; Merieux, Montreal, Canada) and insulin sensitivity was assessed with the euglycemic- hyperinsulinemic clamp method as described elsewhere (17;18).

Sorbs cohort: The cohort was derived from the self-contained Sorbs population in Germany. Extensive phenotyping included standardized questionnaires for past medical history and family history, collection of anthropometric data (weight, height, WHR) and OGTT (diabetes definition according to the ADA criteria) (19). Insulin was measured with the

CHAPTER 6 BMPR2 104

AutoDELFIA®Insulin assay (PerkinElmer Life and Analytical Sciences, Turku, Finland). Nine hundred twenty five Sorbs (740 subjects with normal glucose tolerance (NGT), 79 subjects with impaired glucose tolerance (IGT) and 106 cases with T2D; 542 females, 383 males) were available for the present study (mean age 48 ± 16 years; mean BMI 27.0 ± 4.9 kg/m2; mean WHR 0.87 ± 0.10; mean fasting plasma glucose 5.54 ± 1.18 mmol/l; mean fasting plasma insulin 40.7 ± 27.4 pmol/l).

All studies were approved by the ethics committee of the University of Leipzig and all subjects gave written informed consent before taking part in the study.

6.3.2 Evolutionary analyses We analysed patterns of evolution at the codon level across the BMPR2 phylogeny (Supplement Figure 1a). The non-synonymous to synonymous substitution rate ratio (ω=dN/dS) of 36 BMPR2 orthologs (http://www.ensembl.org,http://www.ncbi.nlm.nih.gov/ gene; Supplement Figure 1a,b) was estimated by maximum likelihood methods (20) using the CODEML program (PAML4a package) (21;22). In addition, phased (PHASE version 2.1) (23;24) SNP data of the HapMap populations (Europeans, Yoruba, Chinese, Japanese) were analysed for signatures of selection with DnaSP (version 5.00.03) (25), Sweep 1.8.1 and the Haplotter/PhaseII (http://hg-wen.uchicago.edu/ selection/haplotter.htm). Provided are population genetic measures which comprises the fixation index (Fst) and standardized integrated haplotype score (iHS) (5).

6.3.3 Analysis of human BMPR2 mRNA expression Human BMPR2 expression was measured by quantitative real-time RT-PCR using SYBR Green methodology and fluorescence was detected on an ABI PRISM® 7500 sequence detector (Applied Biosystems, Darmstadt, Germany) as described in detail elsewhere (16). The following primers were used: human BMPR2 (NCBI accession no NM_001204.5) 5'-CTTTACTGAGAATTTTCCACCTCCTG-3' (sense) and 5'-GCCAAAGCAATGATTAT TGTCTCATC-3' (antisense). Human BMPR2 mRNA expression was calculated relative to the mRNA expression of 18S rRNA, determined by a premixed assay on demand for human 18S rRNA (PE Biosystems, Darmstadt, Germany). For 29 samples BMPR2 cDNA amplification failed and therefore, only 161 were included in the study.

CHAPTER 6 BMPR2 105

6.3.4 Sequencing of the BMPR2 Sequencing of the BMPR2 (13 exons, exon-intron boundaries; NM_001204 in the NCBI GenBank; 1271 bp in the 5’ region and 1282 bp in the 3’ UTR) in 48 non-related Caucasian subjects (12 lean subjects with NGT, 12 visceral obese, 12 subcutaneously obese, 12 with T2D) was performed using the BigDye® Terminator (Applied Biosystems, Inc., Foster City, CA) on an automated DNA capillary sequencer (ABI PRISM® 3100 Avant; Applied Biosystems Inc., Foster City, CA). Sequence information and PCR conditions for all oligonucleotide primers used for variant screening are available upon request.

6.3.5 Genotyping of BMPR2 SNPs SNP genotyping was done using the TaqMan® SNP Genotyping assay according to the manufacturer’s protocol (Applied Biosystems, Inc., Foster City, CA). To assess genotyping reproducibility, a random ~5 % selection of the sample was re-genotyped in all SNPs; all genotypes matched initial designated genotypes. 48 samples used for sequencing were also re-genotyped by TaqMan® technique and perfectly matched the genotypes determined by sequencing.

6.3.6 Statistical analyses Prior to statistical analysis, non-normally distributed parameters were ln-transformed to approximate a normal distribution. Differences in genotype frequencies between the obese or diabetic cases and healthy controls were compared using logistic regression analyses. Multivariate linear relationships were assessed by generalized linear regression models. All analyses were done under the additive model and the presented p-values are adjusted for age, sex (and BMI for glucose traits). Differences in mRNA expression between visceral and subcutaneous adipose tissue were assessed using the paired Student’s t-test. Two-sided p-values < 0.05 were considered to provide nominal evidence for association and are presented without correction for multiple hypothesis testing. The analysis of associations with quantitative traits was restricted to non-diabetics to avoid diabetes status or treatment masking potential effects of the variants on these phenotypic traits. To obtain the combined effect of our two cohorts we performed a meta-analysis by using the METAL software (26). The meta- analysis was performed in a fixed effects model by using the Mantel-Haenszel method. Statistical analyses were performed using SPSS version 18 (SPSS, Inc.; Chicago, IL) and STATA (version 9.0), (StataCorp LP, Texas, USA).

CHAPTER 6 BMPR2 106

6.4 Results 6.4.1 Evolutionary analyses The dN/dS analyses revealed that the BMPR2 is highly conserved among the 36 species analysed (average ω = 0.1278; Supplement Figure 1), hence being subjected to an overall purifying selection. Positional analyses indicate that 53.6 % of positions are very strongly conserved, ω = 0.0037, 40.6 % are strongly conserved, ω = 0.1247, and 5.7 % are conserved, ω = 0.9974 (Figure 1). Except for one, none of the SNPs shows striking frequency differences between the observed populations as indicated by the fixation index Fst (Fst < 0.2). For rs4675278, Fst-values indicate differences between European and East Asian populations (European vs. Chinese Fst = 0.2760, European vs. Japanese Fst = 0.3703), which is reflected in different allele frequencies and a change of minor allele: European: 0.267 A-allele; Chinese: 0.211 G-allele; Japanese: 0.136 G-allele, Yoruba: 0.442 A-allele. The integrated haplotype score (iHS) is based on the differential levels of linkage disequilibrium surrounding a positively selected allele compared to the background allele at the same position (Supplemental Table 1) (5). For the European population iHS values below -2 were found for rs1980153 and rs16839127 (Supplement Table 1), i.e., haplotypes on the derived allele background are longer compared to ancestral allele background. In contrast for rs6717924 the iHS was positive and nearly reached 2 in Caucasians (iHS = 1.986).

[Figure 1]

CHAPTER 6 BMPR2 107

Figure 1: Site class probability bar plot for each codon across the BMPR2 sequence indicating strength of conservation at each amino acid position. The degree of negative selection is equivalent to the strength of conservation. Included are 36 vertebrate orthologs. Human sequence was used as reference. Genetic domains above are annotated according to http://www.uniprot.org/uniprot/Q13873. Numbers represent the location of the domains (amino acids).

6.4.2 Visceral and subcutaneous BMPR2 mRNA expression and obesity Analysis of 161 paired samples of visceral and subcutaneous adipose tissue showed significantly higher BMPR2 transcript levels in subcutaneous compared with visceral fat, independent of gender (Figure 2a). Even though the mRNA expression in men tended to be higher than in women, the gender differences did not reach statistical significance (Figure 2a). To investigate the expression according to body fat mass or fat distribution, we performed additional analyses in subgroups of lean (BMI < 25 kg/m2), overweight (25 kg/m2 < BMI < 30 kg/m2), and obese (BMI > 30 kg/m2) subjects. Based on CT scans measurement (L4-L5) of abdominal visceral and subcutaneous fat areas, obese subjects were further categorized as predominantly visceral (Vis) or subcutaneous (SC) obese as defined by a ratio of Vis/SC fat area > 0.5 (16). BMPR2 mRNA expression was significantly increased in both visceral and subcutaneous adipose tissue of 37 overweight and 80 obese (57 subcutaneously obese and 23 visceral obese) compared with 44 lean subjects (p < 0.001) (Figure 2b). Interestingly, the difference in BMPR2 mRNA expression between visceral and subcutaneous adipose tissue was only seen in lean individuals, but not in overweight or obese subjects (Figure 2b). We further asked whether impaired glucose metabolism in patients with either IGT (N = 15) or T2D (N = 30) may be associated with altered BMPR2 mRNA expression in different fat depots. Subgroup analyses demonstrated that IGT and T2D subjects were indistinguishable with regard to BMPR2 mRNA expression; therefore these groups were analysed together. Patients with IGT/T2D had significantly higher BMPR2 mRNA expression in both visceral and subcutaneous fat compared to NGT (N = 116) subjects (Figure 2c). We detected a significant correlation between visceral and subcutaneous BMPR2 mRNA expression (Figure 3a).

CHAPTER 6 BMPR2 108 a b ### 120 250 *** Lean ## women 190 *** 100 men 130 *** Overweight ## 70 *** * Sc obese 80 *** 10 Vis obese 5.0 60

40 mRNA/18S rRNA [AU] mRNA/18S rRNA 2.5 mRNA/18S rRNA [AU] 20 BMPR2

BMPR2 0.0 0 Vis adipose tissue Sc adipose tissue Vis adipose tissue Sc adipose tissue . c 200 *** NGT ### T2DM 150 +IGT ***

100

mRNA/18S rRNA [AU] 50

BMPR2 0 Vis adipose tissue Sc adipose tissue

Figure 2: BMPR2 mRNA expression in visceral (Vis) and subcutaneous (SC) adipose tissue in lean, obese and type 2 diabetic subjects. The data are means ± SEM; BMPR2 mRNA levels in (a) men (N = 80) and women (N = 81), (b) subgroups of lean (BMI < 25kg/m2; N = 44), overweight (25 kg/m2 < BMI < 30 kg/m2; N = 37), SC obese (N = 57) and Vis obese (N = 23) (BMI > 30 kg/m2); (c) subjects with normal glucose tolerance (NGT; N = 116) and with impaired glucose tolerance (IGT; N = 15) or with type 2 diabetes (T2D; N = 30). * and *** indicate statistical significance at p<0.05 and p < 0.001, respectively when compared with expression in lean subjects (Figure b) or in subjects with NGT (Figure c); # indicates statistical significance for differences between fat depots (Vis vs. SC) for appropriate subject groups; ## and ### indicate statistically significant at p < 0.01 and p < 0.001, respectively; AU, arbitrary units.

6.4.3 Correlation of BMPR2 mRNA expression with parameters of obesity, glucose metabolism, and insulin sensitivity

Univariate regression analysis revealed significant positive correlations between both Vis and SC BMPR2 mRNA expression and BMI, % body fat, waist, WHR, fasting and 2-h plasma glucose, and fasting plasma insulin (Table 1; Figure 3). There was an inverse correlation between visceral and SC BMPR1A mRNA expression and glucose uptake during the steady state of a euglycemic-hyperinsulinemic clamp (Table 1). The correlations remained unchanged also after excluding subjects with IGT and T2D (data not shown). The correlation

CHAPTER 6 BMPR2 109

between BMPR2 mRNA expression in fat and glucose infusion rate during the steady state of an euglycemic–hyperinsulinemic clamp remained significant even upon adjusting for age, gender and % body fat (Table 2).

a b 7.5 6.5 5.0 5.0 3.5 2.5 2.0

0.0 R=0.717 0.5 (P<0.001) -1.0 R=0.611 ln SC expression -2.5 ln Vis expression -2.5 (P<0.001) -5.0 -4.0 -5.0 -2.5 0.0 2.5 5.0 7.5 2.75 3.00 3.25 3.50 3.75 4.00 ln Vis expression ln % body fat c d 8 8 6 6 4 2 4

0 2 -2 R=0.611 ln Visexpression ln SC expression R=0.320 -4 (P<0.001) 0 (P<0.001) -6 -2 3.00 3.25 3.50 3.75 4.00 4.25 3.00 3.25 3.50 3.75 4.00 ln BMI ln BMI

Figure 3: Correlations a) between visceral (Vis) and subcutaneous (SC) BMPR2 mRNA expression; b) between Vis BMPR2 mRNA expression and % body fat, c) between Vis BMPR2 mRNA expression and BMI, d) between SC BMPR2 mRNA expression and BMI. Data are ln-transformed to achieve normal distribution.

Table 1: Linear regression analyses of visceral (Vis) and subcutaneous (SC) BMPR2 mRNA expression with anthropometric and metabolic parameters (N = 161). Vis BMPR2 mRNA SC BMPR2 mRNA R (p-value) R (p-value) Age 0.275 (< 0.001) 0.183 (0.020) BMI 0.611 (< 0.001) 0.320 (< 0.001) % body fat 0.611 (< 0.001) 0.321 (< 0.001) WHR 0.540 (< 0.001) 0.322 (< 0.001) Fasting plasma glucose 0.303 (< 0.001) 0.259 (0.001) Fasting plasma insulin 0.777 (< 0.001) 0.499 (< 0.001) 2-h plasma glucose 0.419 (< 0.001) 0.361 (< 0.001) GIR -0.815 (< 0.001) -0.603 (< 0.001)

GIR – glucose infusion rate during the steady state of the euglycemic-hyperinsulinemic clamp.

CHAPTER 6 BMPR2 110

Table 2: Multivariate linear regression analyses of visceral (Vis) and subcutaneous (SC) BMPR2 mRNA expression with anthropometric and metabolic parameters (N = 161). Vis BMPR2 mRNA SC BMPR2 mRNA β-Coefficient (p-value) β-Coefficient (p-value) Model 1 Age (years) 0.049 (0.0004) 0.028 (0.021) Gender -0.010 (0.980) 0.030 (0.935) Model 2 Age (years) 0.025 (0.029) 0.017 (0.142) Gender 0.010 (0.977) 0.032 (0.926) BMI 6.792 (< 0.001) 2.916 (< 0.001) Model 3 Age (years) 0.030 (0.010) 0.019 (0.103) Gender -0.045 (0.893) 0.008 (0.981) % body fat 4.710 (< 0.001) 2.044 (< 0.001) Model 4 Age (years) 0.011 (0.179) 0.005 (0.659) Gender -0.109 (0.663) 0.004 (0.990) % body fat 2.325 (< 0.001) 0.154 (0.771) GIR -2.898 (< 0.001) -2.047 (< 0.001)

6.4.4 Genetic variation in the BMPR2 We sequenced the human BMPR2 in 48 DNA-samples and found 8 genetic variants (Supplement Figure 2). Further, 10 HapMap tagging SNPs, covering 100 % of the variation in the BMPR2 locus (according to HapMap Phase II; www.hapmap.org) (27), were selected from the database using the Tagger software and the following criteria: minor allele frequency (MAF) > 0.05 and r2 > 0.8. These SNPs were additionally genotyped in the 48 DNA-samples. We calculated the linkage disequilibrium (L.D.) between all SNPs using the EMLD statistical program (https://epi.mdanderson.org/qhuang/Software/pub.htm) (Supplement Table 2). Based on L.D., all SNPs identified through sequencing and with a minor allele MAF > 0.05 were represented by at least one HapMap tagging SNP (L.D. with r2 > 0.8) (Supplement Table 2). Since two HapMap tagging SNPs were in 100 % L.D. in our population (rs13426118 and rs1061157), only 9 tagging SNPs were therefore taken forward to genotyping for association studies in German Caucasians (Leipzig). For replication analyses we included the Sorbs from Germany. All SNPs were in Hardy-Weinberg equilibrium (p > 0.05).

6.4.5 BMPR2 variants and association with obesity Leipzig cohort: In a case-control study including 447 lean (BMI < 25 kg/m2) and 701 obese (BMI > 30 kg/m2) subjects, rs6717924 and rs13426118 were nominally associated with obesity (p < 0.05, adjusted for age and sex). Carriers of the minor allele for rs6717924 were at

CHAPTER 6 BMPR2 111 higher risk of being obese (odds ratio (OR) and (95 % CI) = 1.40 (1.06; 1.85), p = 0.018) whereas subjects with the minor allele for rs13426118 were protected against obesity (0.71 (0.55; 0.93), p = 0.013 in an additive mode of inheritance after adjusting for age and sex) (Table 3, Supplement Table 3).

Sorbs: The two SNPs nominally associated with obesity in the Leipzig cohort were genotyped in the Sorbs for replication. None of the SNPs showed significant association in the Sorbs case-control study (337 lean and 215 obese subjects) (Table 3, Supplement Table 3). In a combined analysis including both the Sorbs and the Leipzig cohort, only rs6717924 showed association with obesity (combined OR and 95 % CI = 1.35 (1.07; 1.70); p = 0.011) (Table 3). The combined effects of the SNPs on obesity were confirmed in a meta-analysis using the metan command in STATA based on the estimated effect sizes of each study and their confidence intervals. The allele/genotype frequencies in the phenotypic groups (lean vs. obese) remained unchanged also after excluding subjects with T2D. However, due to smaller sample size, p values did not reach statistical significance (data not shown).

Table 3: Association analyses of the BMPR2 genetic variants with obesity in the Leipzig and the Sorbs cohorts. Major Minor Minor allele OR (95 % CI) p-value allele allele frequency (Case/Control) Leipzig (a) N = 1148 rs6717924 G A 0.14/0.11 1.40 (1.06; 1.85) 0.018 rs13426118 A C 0.11/0.15 0.71 (0.55; 0.93) 0.013 Sorbs (b) N = 552 rs6717924 G A 0.15/0.12 1.19 (0.77; 1.84) 0.43 rs13426118 A C 0.13/0.11 1.22 (0.78; 1.91) 0.39 Combined analysis rs6717924 G A 1.35 (1.07; 1.70) 0.011 rs13426118 A C 0.93 (0.72; 1.21) 0.61

OR (95 % CI); odds ratio with 95 % confidence intervals; p-values were adjusted for age and sex. *ORs indicate effect directions of the minor allele in additive mode of inheritance. a) N = 701 cases (BMI > 30 kg/m2) vs. 447 controls (BMI < 25 kg/m2), b) N = 215 cases vs. 337 controls. Combined ORs and p-values represent the combined analysis of the two cohorts (Leipzig and Sorbs). The combined analyses were adjusted for age, sex and study cohort.

CHAPTER 6 BMPR2 112

6.4.6 BMPR2 variants and association with obesity related quantitative traits Leipzig cohort: Consistent with case-control studies, we found nominal associations between rs13426118 and BMI in 900 subjects without T2D. As expected, carriers of the obesity risk allele (major) had a higher mean BMI (Supplement Table 4) than the non- carriers. In addition, rs1980153 showed effects on adiponectin, rs4303700 on % body fat, rs12621870 on fasting plasma insulin, and rs17199235 on 2-h plasma glucose levels (Supplement Table 4). We also assessed the association of the 9 SNPs with T2D in a case- control study including 930 cases with T2D and 722 healthy controls with NGT from the Leipzig cohort. No significant association with T2D was found under logistic regression analysis (p > 0.05; adjusted for age, sex and BMI; data not shown). Sorbs: The 2 SNPs (rs13426118, rs6717924) nominally associated with obesity in a case- control study in the Leipzig cohort were also tested for associations with BMI and obesity related traits in 819 non-diabetic Sorbs. Although rs13426118 was nominally associated with BMI (p < 0.05, adjusted for age and sex; Supplemental Table 4), it showed opposite effect direction when compared with the Leipzig cohort. Moreover, rs6717924 was associated with fasting plasma insulin and 2-h plasma insulin (p < 0.05, adjusted for age, sex and BMI). Combined p - values of the two BMPR2 SNPs in our two study cohorts were assessed by a meta-analysis using the METAL software. Both SNPs, rs6717924 and rs13426118, showed nominal significant association with fasting plasma insulin (p = 0.022, and p = 0.044 respectively).

6.4.7 Association of rs6717924 with BMPR2 mRNA expression Carriers of the rs6717924 A-allele (i.e., obesity risk allele) had higher BMPR2 mRNA expression in visceral adipose tissue compared to carriers of the non-risk alleles (p = 0.042; adjusted for age, sex and BMI; Figure 5; N = 161). Consistent with the analyses in the entire cohort, the rs6717924 obesity risk allele was associated with BMI also in this subgroup (p = 0.017, adjusted for sex and age). Including visceral BMPR2 mRNA expression as a covariate abolished the SNP effect on BMI (p = 0.31) while the effect on mRNA expression remained significant (p < 0.001). None of the other SNPs was associated with either visceral or subcutaneous BMPR2 mRNA expression.

CHAPTER 6 BMPR2 113

rs6717924 4 p = 0.042

3

2

1

0 AA + AG GG AA + AG GG lnBMPR2 mRNA/18S rRNA [AU] visceral subcutaneous

Figure 4: Association of rs6717924 with adipose tissue BMPR2 mRNA expression. Given are means±SEM. A minor allele; G major allele.

6.5 Discussion BMPR2 plays an important role during embryonic development and remains ubiquitously expressed in later life. Employing genome wide screening strategies, the BMPR2 locus has been suggested to be a target of recent evolutionary selection in human populations (5). So far, genetic variants in the BMPR2 have been described to be responsible for the majority of cases of the autosomal hereditable pulmonary arterial hypertension (28). Functional studies revealed that vascular BMPR2 mRNA expression is up-regulated in early stages of autoimmune diabetes in non-obese diabetic mice (29) and improved renal bone morphogenetic protein 7 (BMP7) and BMPR2 mRNA expression following treatment in streptozotocin-induced diabetic rats is of benefit against renal damage during diabetic nephropathy (30). However, little is known about the role of BMPR2 in the pathophysiology of human obesity despite evidence for involvement of BMPs in the regulation of adipogenesis (9). Also our recent studies on BMPR1A, partner in the receptor complex with BMPR2, suggest a regulatory role of these receptors in obesity (15). We therefore performed a systematic study on the potential role of BMPR2 in the pathophysiology of obesity by combining data from genetic, evolutionary, phenotypic and adipose tissue expression studies. We analysed the coding region of 36 vertebrate BMPR2 orthologs to screen for signatures of selections between species and showed that BMPR2 has undergone an overall purifying selection. Considering strong physiologic relevance of BMPR2 in developmental processes, it is not surprising that the coding region shows very strong conservation with more than half of amino acid positions being very strongly conserved. To assess intronic

CHAPTER 6 BMPR2 114 regions, we used Haplotter/PhaseII database to screen for selection patterns within the BMPR2 locus. Population genetic measures represented by high iHS in Caucasian populations (Supplement Table 1) indicated signatures of recent selection at the BMPR2 locus. Since BMPR2 is considered a plausible obesity candidate gene, we investigated its role in the pathophysiology of human obesity. We found significantly higher BMPR2 mRNA expression in fat in overweight and obese compared to lean individuals. In addition, BMPR2 mRNA expression was significantly higher in subjects with IGT/T2D compared to controls with NGT. BMPR2 mRNA expression correlated with measures of obesity and fat distribution, as well as with traits of glucose metabolism and insulin sensitivity. Notably, the visceral BMPR2 expression increased remarkable, when the BMI shifted towards overweight. Furthermore, predominantly in lean individuals, BMPR2 was higher expressed in subcutaneous compared to visceral adipose tissue. Correlations of BMPR2 expression with obesity and relevant metabolic traits suggest that BMPR2 mRNA expression in human adipose tissue might be related to progression of obesity but it remains unclear whether it is due to the extended fat mass or vice versa. Consistent with recent studies on BMPR1A (15), positive correlation of BMPR2 mRNA expression with obesity suggests that obese individuals may have enhanced BMP-2/4 signalling, which may stimulate adipose tissue differentiation thereby contributing to increased fat mass. However, based on such correlational analyses, the direction of the causative chain can not be established and we can therefore not exclude that increased BMPR2 expression may be a consequence rather than the cause of increased fat mass. The genotype-phenotype association study showed two SNPs (rs6717924, rs13426118) nominally associated with obesity in the Leipzig cohort. However, they could not be replicated in the Sorbs, even though combined analyses of both cohorts supported the association of rs6717924 A allele with obesity. Similarly, association of BMPR2 SNPs with quantitative traits was moderate and would not withstand adjustment for multiple testing (e.g. Bonferroni correction). However, the rs6717924 obesity risk allele was associated with visceral BMPR2 mRNA expression and also with BMI in a subgroup of 161 subjects. Including visceral BMPR2 mRNA expression as a covariate abolished the SNP effect on BMI while the effect of mRNA expression levels remained significant. This suggests that the moderate SNP effect on BMI might be mediated through its effect on the visceral BMPR2 mRNA expression level. Altered BMPR2 mRNA expression might result in changes in adipose tissue expansion during the activation of division and differentiation of adipocyte precursor cells. Undoubtedly, studies targeting functional consequences of BMPR2 obesity risk variants on BMPR2 transcription are inevitable to clarify whether genetic variants can

CHAPTER 6 BMPR2 115 explain variation in BMPR2 mRNA expression. Nevertheless, the iHS value of nearly 2 for the rs6717924 suggests that the haplotypes on the background of the ancestral allele A, which was associated with obesity and visceral mRNA levels, are longer compared to derived allele background, thus subject to recent positive selection. This suggests that the BMPR2 variant was likely to be selected to accumulate energy more efficiently. Taking higher BMPR2 mRNA levels in carriers of the rs6717924 A-allele into account, we used the Transcription Element Search System (TESS; http://www.cbil.upenn.edu/tess) to examine transcriptional binding sites surrounding this genetic variant, which might explain variation in BMPR2 expression. The sequence surrounding rs6717924 (G>A) matches human transcription binding site for a transcription factor IPF1 (insulin upstream/promoter factor 1) for the major allele G. IPF1 is needed for the formation of the pancreas and is assumed to act to determine the fate of common pancreatic precursor cells and/or to regulate their propagation (31). In this regard, it seems noteworthy that we observed nominal association of the rs6717924 with fasting plasma insulin levels in a combined analysis of both cohorts. Moreover, there is evidence that BMPs might regulate insulin secretion. BMP4 and its high-affinity receptor BMPR1A are expressed in differentiating and adult beta cells and transgenic expression of Bmp4 in mice beta cells enhances glucose stimulated insulin secretion and glucose clearance (32). Although BMPR2 expression was higher in subjects with T2D, the genetic variation in BMPR2 does not appear to be a major player in the polygenic aetiology of T2D as no association of genetic polymorphisms with T2D was found. We are aware that statistical power of our study is limited due to relatively small sample size. In our cohorts we had a power of 80 % (α = 0.05) to detect a difference of 0.71-1.32 kg/m2 per allele for BMI, 0.08-0.15 mmol/l for fasting plasma glucose, 3.94-18.07 pmol/l for fasting plasma insulin, 0.23-0.43 mmol/l for 2-h plasma glucose during an OGTT under the additive model. Similarly, in the case-control studies for obesity, we had a power of 80 % (α = 0.05) to detect SNP effects with odds ratios (ORs) ranging from 1.21 to 1.44 and from 1.35 to 1.37 in the Leipzig and Sorbs, respectively. Since the reported effect sizes for obesity SNPs are usually lower (ORs of 1.1–1.2), smaller effect may have been missed in our studies. In conclusion, our data on expression, genetics as well as evolutionary aspects suggest a role of BMPR2 in the pathophysiology of obesity and provide some evidence for thrifty genotype hypothesis.

CHAPTER 6 BMPR2 116

6.6 Acknowledgements We thank all those who participated in the studies. This work was supported by grants from the Federal Ministry of Education and Research, Competence Network Obesity (FKZ 01GI0829), the NGFNplus (FKZ 01GI0827), Deutsche Forschungsgemeinschaft (DFG), the Clinical Research Group “Atherobesity” KFO 152 (projects BL 833/1-1 to MB, and Stu192/6-1 (MS), KO 3512/1-1 to AK, KO 3880/1-1 to PK), from the German Diabetes Association (to DS and PK), from the National Institutes of Health (USA) grants R01 DK077097 and P30 DK040561 (to YHT) and DHFD (Diabetes Hilfs- und Forschungsfonds Deutschland; to MS, PK, MB).

6.7 References 1. Lindgren CM, Heid IM, Randall JC et al. Genome-Wide Association Scan Meta-Analysis Identifies Three Loci Influencing Adiposity and Fat Distribution. 2009 PloS Genet 2009 5:e1000508.

2. Thorleifsson G, Walters GB, Gudbjartsson DF et al. Genome-wide association yields new sequence variants at seven loci that associate with measures of obesity. Nat Genet 2009 41:18- 24.

3. Willer CJ, Speliotes EK, Loos RJF et al. Six new loci associated with body mass index highlight a neuronal influence on body weight regulation. Nat Genet 2009 41:25-34.

4. Neel JV. Diabetes Mellitus - A Thrifty Genotype Rendered Detrimental by Progress. Am J Hum Genet 1962 14:353-362.

5. Voight BF, Kudaravalli S, Wen XQ, Pritchard JK. A map of recent positive selection in the human genome. PloS Biol 2006 4:446-458.

6. Kishigami S, Mishina Y. BMP signaling and early embryonic patterning. Cytokine Growth Factor Rev 2005 16:265-278.

7. Chen D, Zhao M, Mundy GR. Bone morphogenetic proteins. Growth Factors 2004 22:233- 241.

8. Yamamoto Y, Oelgeschlager M. Regulation of bone morphogenetic proteins in early embryonic development. Naturwissenschaften 2004 91:519-534.

9. Schulz TJ, Tseng YH. Emerging role of bone morphogenetic proteins in adipogenesis and energy metabolism. Cytokine Growth Factor Rev 2009 20:523-531.

10. Hogan BLM. Bone morphogenetic proteins: Multifunctional regulators of vertebrate development. Genes Dev 1996 10:1580-1594.

11. Kingsley DM. The Tgf-Beta Superfamily - New Members, New Receptors, and New Genetic Tests of Function in Different Organisms. Genes Dev 1994 8:133-146.

12. Massague J, Weis-Garcia F. Serine/threonine kinase receptors: Mediators of transforming growth factor beta family signals. Cancer Surv 1996 27:41-64.

CHAPTER 6 BMPR2 117

13. Nohe A, Hassel S, Ehrlich M, Neubauer F, Sebald W, Henis YI, Knaus P. The mode of bone morphogenetic protein (BMP) receptor oligomerization determines different BMP-2 signaling pathways. J Biol Chem 2002 277:5330-5338.

14. Huang HY, Song TJ, Li X et al. BMP signaling pathway is required for commitment of C3H10T1/2 pluripotent stem cells to the adipocyte lineage. Proc Natl Acad Sci U S A 2009 106:12670-12675.

15. Böttcher Y, Unbehauen H, Klöting N et al. Adipose Tissue Expression and Genetic Variants of the Bone Morphogenetic Protein Receptor 1A Gene (BMPR1A) Are Associated With Human Obesity. Diabetes 2009 58:2119-2128.

16. Berndt J, Klöting N, Kralisch S et al. Plasma visfatin concentrations and fat depot-specific mRNA expression in humans. Diabetes 2005 54:2911-2916.

17. Defronzo RA, Tobin JD, Andres R. Glucose Clamp Technique - Method for Quantifying Insulin-Secretion and Resistance. Am J of Physiol 1979 237:E214-E223.

18. Blüher M, Unger R, Rassoul F, Richter V, Paschke R. Relation between glycaemic control, hyperinsulinaemia and plasma concentrations of soluble adhesion molecules in patients with impaired glucose tolerance or Type II diabetes. Diabetologia 2002 45:210-216.

19. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2006 29:S43-S48.

20. Goldman N, Yang ZH. Codon-Based Model of Nucleotide Substitution for Protein-Coding DNA-Sequences. Mol Biol Evol 1994 11:725-736.

21. Yang ZH. PAML: a program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci 1997 13:555-556.

22. Yang ZH. PAML 4: Phylogenetic analysis by maximum likelihood. Mol Biol Evol 2007 24:1586-1591.

23. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 2001 68:978-989.

24. Stephens M, Scheet P. Accounting for decay of linkage disequilibrium in haplotype inference and missing-data imputation. Am J Hum Genet 2005 76:449-462.

25. Rozas J, Sanchez-DelBarrio JC, Messeguer X, Rozas R. DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 2003 19:2496-2497.

26. Willer CJ, Sanna S, Jackson AU et al. Newly identified loci that influence lipid concentrations and risk of coronary artery disease. Nat Genet 2008 40:161-169.

27. Gibbs RA, Belmont JW, Hardenbol P et al. The International HapMap Project. Nature 2003 426:789-796.

28. Austin ED, Loyd JE, Phillips JA. Genetics of Pulmonary Arterial Hypertension. Sem Respir Crit Care Med 2009 30:386-398.

29. Nett PC, Ortmann J, Celeiro J et al. Transcriptional regulation of vascular bone morphogenetic protein by endothelin receptors in early autoimmune diabetes mellitus. Life Sci 2006 78:2213-2218.

CHAPTER 6 BMPR2 118

30. Yeh CH, Chang CK, Cheng MF, Lin HJ, Cheng JT. Decrease of Bone Morphogenetic Protein- 7 (BMP-7) and its Type II Receptor (BMP-RII) in Kidney of Type 1-Like Diabetic Rats. Horm Metab Res 2009 41:605-611.

31. Jonsson J, Carlsson L, Edlund T, Edlund H. Insulin-Promoter-Factor-1 Is Required for Pancreas Development in Mice. Nature 1994 371:606-609.

32. Goulley J, Dahl U, Baeza N, Mishina Y, Edlund H. BMP4-BMPR1A signaling in beta cells is required for and augments glucose-stimulated insulin secretion. Cell Metab 2007 5:207-219.

6.8 Supplement a b Scientific name Species Accession Number BMPR2 phylogenetic tree Anolis carolinensis Anole lizard ENSACAT00000004923 Anole lizard 0.1362 Bos taurus Cattle ENSBTAT00000008419 Chicken 0.0480 Callithrix jacchus Common marmoset ENSCJAT00000009881 Wallaby 0.1120 Canis famimiliaris Dog ENSCAFT00000036570 Short-tailed opossum 0.0576 Cavia porcellus Guinea pig ENSCPOT00000009672 N. tree shrew 0.0730 Choloepus hoffmanni Sloth ENSCHOT00000008739 Philippine tarsier 0.0882 Dasypus novemcinctus Nine-banded armadillo ENSDNOT00000013861 C. marmoset 0.2360 Rhesus monkey 0.0882 Dipodomys ordii Ord's kangaroo rat ENSDORT00000009208 Orang utan 0.1276 Echinops telfairi Lesser hedgehog tenrec ENSETET00000000699 XP_001497350.2/ Western gorilla 1.1423 Equus caballus Horse* Human 0.0957 ENSECAT00000027148 Chimpanzee 0.0001 Erinaceus europaeus European hedgehog ENSEEUT00000006776 Mouse lemur 0.3690 Felis catus Cat* ENSFCAT00000014098 N. greater galago 0.0361 NP_001001465.1/ Ground squirrel 0.0770 Gallus gallus Chicken Guinea pig 0.0960 ENSGALT00000013786 Ord‘s kangaroo rat 0.1319 Gorilla gorilla Western gorilla ENSGGOT00000011031 Mouse 0.0705 Homo sapiens Human ENST00000374580 Norwegian rat 0.4535 Loxodonta africanus African elephant ENSLAFT00000015205 Pika 0.0494 Macaca mulatta Rhesus monkey ENSMMUT00000016780 Rabbit 0.1406 Macropus eugenii Wallaby ENSMEUT00000006447 Horse 0.1257 Microcebus murinus Mouse lemur ENSMICT00000004437 C. bottlenose dolphin 0.1132 Alpaca 0.0299 Monodelphis domestica Opossum ENSMODT00000020719 Cattle 0.0455 Mus musculus Mouse ENSMUST00000087435 Dog 0.1914 Myotis lucifugus Little brown bat ENSMLUT00000014786 Cat 0.1546 Ochotona princeps Pika ENSOPRT00000009329 Little brown bat 0.1688 Oryctolagus cuniculus Rabbit ENSOCUT00000001973 Flying fox 0.1192 Otolemur garnettii Northern greater galago ENSOGAT00000007769 European hedgehog 0.0634 Pan troglodytes Chimpanzee ENSPTRT00000056194 C. shrew 0.0798 Sloth Pongo pygmaeus Orang utan ENSPPYT00000015214 Nine-banded armadillo 0.0940 Procavia capensis Hyrax ENSPCAT00000012541 Tenrec 0.1960 Pteropus vampyrus Flying fox ENSPVAT00000005248 African elephant 0.1540 Q91WY9_RAT/ Rattus norvegicus Norwegian rat* Hyrax 0.0666 ENSRNOT00000035238 Supplement Figure 1: Phylogeny for BMPR2 with estimates of ω for each Sorex araneus Common shrew ENSSART00000013016 Spermophilus tridecemlineatus Thirteen-lined ground squirrel ENSSTOT00000002360 species (a) and list of species analyzed (b). ω-values were obtained under the Tupaia belangeri Northern tree shrew ENSTBET00000017571 “free-ratio” model. C. = Common, N. = Northern. *indicates species with Tarsius syrichta Philippine tarsier ENSTSYT00000003507 concatenated sequences. Tursiops truncatus Common bottlenose dolphin ENSTTRT00000004984 Vicugna pacos Alpaca ENSVPAT00000009496

rs10714063 rs7575056 rs1061157 rs45502895 rs1980153 rs2228545 rs12693968 rs4675278 rs4303700 rs55722784 rs6717924 rs6435156 rs12621870 rs3044175 rs13426118 rs17199235 rs16839127

-210/-211 insC ATG 191425 bp TGA

Suplement Figure 2: Genetic structure and SNPs analysed in the bone morphogenetic receptor 2 gene (BMPR2). The polymorphisms shown in black were discovered by direct gene sequencing. One of them was novel (w/o rs-number). Ten HapMap tagging polymorphisms (shown in green) were analysed for association with obesity and related phenotypes in German subjects.

CHAPTER 6 BMPR2 121

Supplement Table 1: Integrated haplotype scores (iHS) for tagging SNPs within the BMPR2 locus. CEU YRI ASN rs6177924 1.986 -1.498 - rs1980153 -2.495 - - rs4303700 -1.667 - -0.601 rs13426118 -1.711 -0.278 1.259 rs16839127 -2.236 - - rs12693968 1.235 - -0.580 rs4675278 -0.572 -1.801 -1.864 rs12621870 -0.978 0.698 -1.971 rs17199235 -1.278 - - rs1061157 -1.485 - 0.981

Provided are iHS values for Europeans (CEU), Africans (YRI) and Asians (ASN) (5).

Supplement Table 2: Measures of L.D. (D prime (D’) and r2) among 17 BMPR2 variants. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1 1 0.00 0.00 0.08 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

2 1.00 1 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.04 0.00 0.00 1.00 0.07 0.00 0.00 0.00

3 1.00 1.00 1 0.03 0.04 0.01 0.01 0.18 0.07 0.03 0.87 0.87 0.00 0.02 0.00 0.01 0.08

4 1.00 1.00 1.00 1 0.01 0.00 0.00 0.05 0.02 0.04 0.02 0.02 0.00 0.03 0.01 0.00 0.00

5 1.00 1.00 1.00 0.51 1 0.04 0.31 0.08 0.17 0.08 0.05 0.05 0.00 0.05 0.01 0.04 0.03

6 1.00 1.00 0.63 0.32 1.00 1 0.01 0.49 0.00 0.03 0.00 0.00 0.00 0.02 0.00 1.00 0.00

7 1.00 1.00 1.00 0.51 1.00 1.00 1 0.02 0.05 0.02 0.00 0.00 0.00 0.01 0.00 0.01 0.12

8 1.00 1.00 0.60 1.00 1.00 1.00 1.00 1 0.01 0.07 0.25 0.25 0.00 0.04 0.00 0.50 0.00

9 1.00 1.00 1.00 0.44 1.00 0.14 1.00 0.20 1 0.45 0.09 0.09 0.02 0.31 0.06 0.01 0.01

10 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1 0.05 0.05 0.04 0.63 0.12 0.03 0.00

11 1.00 1.00 1.00 1.00 1.00 0.16 0.03 0.68 1.00 1.00 1 1.00 0.00 0.03 0.00 0.02 0.10

12 1.00 1.00 1.00 1.00 1.00 0.16 0.03 0.68 1.00 1.00 1.00 1 0.00 0.03 0.00 0.02 0.10

13 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1 0.07 0.00 0.00 0.00

14 1.00 1.00 0.90 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1 0.20 0.02 0.00

15 1.00 1.00 0.13 1.00 0.84 1.00 1.00 0.58 1.00 1.00 1.00 1.00 1.00 1.00 1 0.00 0.00

16 1.00 1.00 1.00 0.12 1.00 1.00 1.00 1.00 0.21 1.00 1.00 1.00 1.00 1.00 1.00 1 0.00

17 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1

r2 given in upper shaded boxes and D’ given in lower boxes. 1 = -918_-919insAGC, 2 = -210_-211insC, 3 = rs6717924, 4 = rs1980153, 5 = rs4303700, 6 = rs13426118, 7 = rs16839127, 8 = rs12693968, 9 = rs4675278, 10 = rs12621870, 11 = rs10714063, 12 = rs7575056, 13 = 137484 A/C, 14 = rs17199235, 15 = rs2228545, 16 = rs1061157, 17 = rs45502895.

CHAPTER 6 BMPR2 122

Supplement Table 3: Association of BMPR2 genetic variants with obesity in the Leipzig cohort (A) and the Sorbs cohort (B).

Lean subjects Obese subjects Additive Dominant Recessive MAF SNP Genotype (BMI (BMI p value p value p value lean/obese < 25 kg/m2) > 30 kg/m2) OR (95 % CI) OR (95 % CI) OR (95 % CI) Leipzig Male/Female 183/264 257/444 cohort (A) Age (years) 50 ± 18 59 ± 11 BMI (kg/m2) 23.4 ± 1.4 34.8 ± 4.8 rs6717924 AA 7 (1.6 %) 11 (1.6 %) 0.018 0.011 0.854 AG 77 (18.0 %) 170 (25.1 %) 0.106/0.142 1.399 1.492 1.101 GG 344 (80.4 %) 496 (73.3 %) (1.059; 1.848) (1.098; 2.027) (0.397; 3.051) rs1980153 TT 2 (0.5 %) 6 (0.9 %) 0.226 0.238 0.625 TA 65 (15.1 %) 111 (16.5 %) 0.080/0.091 1.222 1.231 1.506 AA 363 (84.4 %) 557 (82.6 %) (0.883; 1.695) (0.872; 1.737) (0.292; 7.771) rs4303700 AA 17 (4.0 %) 30 (4.4 %) 0.236 0.203 0.776 AG 157 (36.3 %) 215 (31.7 %) 0.221/0.203 0.874 0.844 0.910 GG 258 (59.7 %) 433 (63.9 %) (0.700; 1.092) (0.651; 1.095) (0.476; 1.739) rs4675278 AA 84 (19.4 %) 156 (23.0 %) 0.114 0.142 0.202 AG 122 (28.2 %) 192 (28.4 %) 0.335/0.372 1.138 1.209 1.225 GG 227 (52.4 %) 329 (48.6 %) (0.969; 1.335) (0.938; 1.558) (0.897; 1.674) rs16839127 AA 0 (0 %) 3 (0.5 %) 0.560 0.674 AG 49 (11.4 %) 81 (12.0 %) 0.056/0.065 1.120 1.088 - GG 381 (88.6 %) 590 (87.5 %) (0.765; 1.639) (0.735; 1.610) rs12693968 AA 30 (7.4 %) 39 (5.9 %) 0.765 0.443 0.440 AG 124 (30.5 %) 225 (34.3 %) 0.227/0.231 1.033 1.110 0.815 GG 252 (62.1 %) 392 (59.8 %) (0.835; 1.277) (0.850; 1.450) (0.484; 1.370) rs12621870 CC 36 (8.3 %) 47 (6.9 %) 0.461 0.739 0.250 CT 153 (35.4 %) 243 (35.7 %) 0.260/0.248 0.928 0.958 0.756 TT 243 (56.3 %) 390 (57.4 %) (0.759; 1.133) (0.742; 1.236) (0.469; 1.218) rs17199235 GG 12 (2.8 %) 16 (2.4 %) 0.941 0.999 0.813 GA 98 (23.3 %) 163 (24.1 %) 0.145/0.144 0.990 1.000 0.907 AA 312 (73.9 %) 497 (73.5 %) (0.769 ; 1.276) (0.747; 1.338) (0.403; 2.039) rs13426118 CC 10 (2.3 %) 7 (1.0 %) 0.013 0.022 0.139 CA 108 (24.9 %) 136 (20.0 %) 0.147/0.110 0.714 0.708 0.467 AA 316 (72.8 %) 537 (79.0 %) (0.546; 0.932) (0.527; 0.951) (0.170; 1.281) Sorbs Male/Female 104/233 78/137 cohort (B) Age (years) 39 ± 15 58 ± 12 BMI (kg/m2) 22.3 ± 1.7 33.9 ± 3.6 rs6717924 AA 2 (0.6 %) 3 (1.4 %) 0.425 0.498 0.476 AG 78 (23.5 %) 55 (26.2 %) 0.123/0.145 1.192 1.174 2.035 GG 252 (75.9 %) 152 (72.4 %) (0.774; 1.838) 0.739; 1.864) (0.289; 14.342) rs13426118 CC 3 (1.0 %) 2 (1.0 %) 0.386 0.267 0.517 CA 67 (20.7 %) 51 (24.4 %) 0.113/0.132 1.219 1.315 0.533 AA 253 (78.3 %) 156 (74.6 %) (0.779; 1.909) (0.811; 2.131) (0.080; 3.574)

Case-control for obesity: (A) 447 lean subjects (BMI < 25 kg/m2) vs. 701 obese subjects (BMI > 30 kg/m2); (B) 337 lean subjects (BMI < 25 kg/m2) vs. 215 obese subjects (BMI > 30 kg/m2). P values were calculated after adjusting for age and gender. Odds ratios (OR) and 95 % confidence intervals (95 % CI) are given for the minor allele. Dominant model of inheritance indicates Mm + mm vs. MM. MAF = minor allele frequency.

CHAPTER 6 BMPR2 123

Supplement Table 4: Association of BMPR2 SNP variants with quantitative traits in the Leipzig (A) and the Sorbs (B) cohort. Fasting 2-h Fasting % Leipzig cohort (A) BMI WHR plasma plasma plasma GIR body Leptin AdipoQ glucose glucose insulin fat rs6717924 G>A p-value 0.247 0.791 0.551 0.347 0.261 0.958 0.165 0.071 0.222 β 0.014 0.003 0.005 -0.019 -0.123 0.004 0.056 0.192 -0.099 SE 0.012 0.012 0.008 0.020 0.109 0.085 0.040 0.106 0.081 rs1980153 A>T p-value 0.666 0.470 0.204 0.226 0.430 0.554 0.739 0.688 0.019* β -0.006 -0.010 -0.012 -0.028 0.104 0.057 -0.016 0.050 0.223 SE 0.014 0.014 0.009 0.023 0.132 0.097 0.047 0.124 0.094 rs4303700 G>A p-value 0.206 0.743 0.470 0.768 0.103* 0.330 0.021* 0.976 0.691 β -0.012 -0.003 -0.005 -0.005 0.140 -0.063 -0.075 0.003 -0.027 SE 0.010 0.010 0.007 0.016 0.086 0.065 0.032 0.085 0.067 rs4675278 G>A p-value 0.077# 0.966 0.529 0.661 0.338 0.827 0.728 0.177* 0.572 β 0.013 -0.0003 0.003 -0.005 -0.061 0.011 0.008 -0.085 -0.027 SE 0.007 0.007 0.005 0.012 0.064 0.049 0.023 0.063 0.047 rs16839127 G>A p-value 0.994 0.346 0.876 0.837 0.287# 0.823 0.641 0.639 0.742 - β 0.015 0.002 0.005 0.146 0.022 -0.023 -0.062 0.033 0.0001 SE 0.016 0.016 0.010 0.025 0.137 0.100 0.049 0.133 0.101 rs12693968 G>A p-value 0.413 0.285 0.563 0.476 0.426 0.691 0.158 0.113 0.324 β -0.008 -0.011 0.004 -0.012 -0.070 -0.027 0.045 0.134 -0.063 SE 0.010 0.010 0.007 0.016 0.087 0.068 0.032 0.085 0.064 rs12621870 T>C p-value 0.827 0.572 0.280 0.203* 0.025* 0.350 0.559 0.539 0.195 β 0.002 -0.005 0.007 -0.019 -0.179 0.059 -0.017 -0.050 -0.078 SE 0.009 0.009 0.006 0.015 0.080 0.063 0.028 0.082 0.060 rs17199235 A>G P-value 0.745 0.744 0.570 0.042* 0.058* 0.111 0.642 0.487 0.526 β 0.004 0.004 0.004 -0.037 -0.190 0.131 -0.017 0.074 -0.049 SE 0.011 0.011 0.008 0.018 0.100 0.082 0.037 0.106 0.077 rs13426118 A>C p-value 0.016* 0.222 0.939 0.774 0.319 0.588 0.923 0.413# 0.597 β -0.030 -0.015 -0.001 0.006 -0.107 -0.045 -0.004 0.089 -0.045 SE 0.012 0.012 0.008 0.020 0.107 0.083 0.042 0.109 0.085 Fasting 2-h Fasting 2-h Sorbs cohort (B) BMI WHR plasma plasma plasma p-Fat insulin glucose glucose insulin rs6717924 G>A p-value 0.924 0.438 0.998 0.261 0.033* 0.030* 0.545 β 0.001 -0.007 <0.001 -0.025 -0.084 -0.140 0.018 SE 0.011 0.009 0.006 0.022 0.039 0.064 0.029 Combined p value 0.366 0.731 0.665 0.145 0.022 rs13426118 A>C p-value 0.018* 0.406 0.790 0.616 0.061 0.443 0.096 β 0.026 0.007 0.002 -0.011 -0.073 -0.049 0.047 SE 0.011 0.008 0.006 0.022 0.039 0.064 0.028 Combined p value 0.912 0.757 0.898 0.890 0.044

CHAPTER 6 BMPR2 124

Supplement Table 4 cont.: The dataset included subjects with NGT and IGT (N = 900, except for % body fat, leptin and adiponectin (N = 374) for the Leipzig cohort; N = 819 for the Sorbs cohort). P values were calculated after adjusting for age and gender for the variables BMI, WHR and % body fat; and for age, gender and BMI for the remaining variables. ß indicate effect directions of the minor alleles. P-values are given for the additive model. */# significant in dominant/recessive mode of inheritance; dominant model indicates Mm + mm vs. MM; BMI = body mass index; WHR = waist-to- hip ratio; OGTT = oral glucose tolerance test; GIR = glucose infusion rate; AdipoQ = adiponectin.

CHAPTER 6 BMPR2 125

6.9 Contribution of Authors Dorit Schleinitz Study design, main part of the statistical analyses and project-oriented database inquiry, design and writing of the manuscript, in vitro cell culture: adipogenesis of human primary preadipocytes, RNA preparation and BMPR2 real-time PCR (data not shown in the manuscript), genotyping of the Sorbs cohort

Nora Klöting RNA preparation from human Vis and SC fat biopsies, real-time PCR

Yvonne Böttcher Sequencing

Sara Wolf Genotyping of the Leipzig cohort, sequencing

Kerstin Dietrich Evolutionary analyses

Anke Tönjes Recruiting and phenotyping of the Sorbs cohort

Jana Breitfeld Proofreading of the manuscript

Beate Enigk DNA preparation

Jan Halbritter Recruiting and phenotyping of the Sorbs cohort

Antje Körner BMPR2 real-time PCR in the SGBS cell culture model (data not shown in the manuscript), proofreading of the manuscript

Michael R. Schön Surgery (preparation of the biopsies)

Jost Jenkner Surgery (preparation of the biopsies)

Yu-Hua Tseng Project idea, contributing to the discussion

Michael Stumvoll Project idea, contributing to the discussion

Matthias Blüher Project idea, proofreading of the manuscript

Peter Kovacs Project idea, part of the statistical analyses, editing of the manuscript

CHAPTER 7 126

CHAPTER 7 Closing Discussion

In the recent years, the number of genetic variants and genes shown to play a role in the polygenic background of obesity and related metabolic complications such as insulin resistance, dyslipidemia or T2D has exploded by means of GWAS in large independent cohorts (90). In a well-powered study, this approach enables the detection of genetic variants and genes without a priori hypothesis and even with small effects on the investigated phenotype. On the one hand this often leads to surprising as well as intriguing novel findings, as demonstrated by the FTO (69). On the other hand, apparently attractive candidate genes, such as FASN may have been missed in GWAS. Despite lacking association with metabolic phenotypes in GWAS, a physiologically interesting gene may still deserve in depth analyses. Significant association between FASN SNPs and obesity could be detected in our study, raising the question why this has been missed in GWAS. It is likely that the association in GWAS did not reach the threshold for genome wide significance. However, it needs to be mentioned that in most studies FASN variants were not covered by the used SNP genotyping arrays. Also, variants in BMPR2, which plays a role in adipogenesis, were never highlighted in GWAS investigating obesity, T2D or related traits. Again, this may be due to a lacking genome wide significance for the association of the variants with the trait at a p level of 10-8. Significance thresholds obtained after multiple testing or Bonferroni correction may strengthen the analysis but plausible physiological candidate genes may get lost in the “background noise”. This demonstrates that well-designed candidate gene studies including replication cohorts as applied in case of FASN, CNR1, and BMPR2 are still eligible, and a combination of both GWAS and candidate gene approaches might be the avenue leading to more insights into the genetic complexity of metabolic alterations. Without doubts, statistical power affects both study designs; the larger the cohort the smaller the detectable effect size. Also studies presented in this thesis are affected by this issue. Although possible associations may have been missed in the investigated cohorts, the effect size of a variant has to be considered as well; in particular in the light of its potential physiological relevance. Assuming a polygenic background of a trait, the effects of many single SNPs are supposed to sum up to administrate an adverse effect. Simultaneously, there may be the same number of alleles with beneficial effects on health, thus masking the deleterious effects of single SNPs. In the context of the “common variant – common disease” hypothesis (SNP MAF > 5 %), strong SNP effects will be found even in “small” cohorts as

CHAPTER 7 127 shown e.g. for FASN. Nevertheless, to study the effects of rare genetic variants (MAF < 5 %), large-scale studies with cohorts comprising thousands of individuals are highly desirable. The population genetic structure as well as environmental factors play a crucial role in genotype-phenotype association studies. This may partially explain discrepancies in effects or even in effect direction between various independent studies. One prominent example is the population of Pima Indians from southern Arizona that is characterized by one of the highest prevalence of T2D worldwide (91). Numerous candidate genes for T2D have been investigated and identified in this population (92). It should be noted that whereas one part resident in Arizona develops severe metabolic complications, the Pima Indian counterparts resident in Mexico show a scanty phenotypic outcome. As it is unlikely that the genetic pool of these two populations has changed in the last years, the phenotypic differences are most likely due to strikingly different environmental factors, particularly due to the western style hyper-caloric nutrition of the group living in Arizona. Despite low average levels of genetic differentiation among Europeans, a close correspondence between genetic and geographic distances was found by Novembre et al. (93). The results emphasize that when mapping the genetic basis of a disease phenotype, spurious associations can arise if genetic structure is not properly accounted for. This might partly explain the inconsistencies in studies on CNR1 as reported in the literature, but also shown in the thesis. Although a gene might be considered a “good” candidate to modulate disease risk in one population, it does not necessarily need to have the same effect in another population. However, this indicates that replication studies as well as independent studies in populations other than e.g. Europeans are necessary to estimate the overall risk of a genetic polymorphism or gene on the outcome of a disease. This would be also necessary for the FASN and BMPR2 SNPs described in the present work. Better awareness of the genetic heterogeneity of studied populations made possible that even studies lacking significant associations could be considered for publication which leads to a better overall assessment of the influence of a genetic variant on disease risk. Nevertheless it is also part of the “dilemma” one faces when studying diseases with a polygenic background. Children cohorts take a unique position in genotype-phenotype association studies. Unlike in adults their phenotype is less influenced by co-morbidities and prolonged exposure to environmental factors, making the genetic contribution more profound. Considering that phenotype is a result of genotype and environment interaction, it is not surprising that genetic factors may have a different impact in various stages of life. Variants in FTO have previously been linked to childhood obesity by others (69,70). However, in case of FASN our study suggests that SNP effects in this gene most likely manifest in later stages of life and/or they

CHAPTER 7 128 might be modified by gene-environment interactions. Furthermore, the gender dimorphism often plays an important role (94), as shown here for the Val1483Ile substitution in FASN. Sex related factors like hormones are known to exert dramatic effects, amongst others, clearly demonstrated by differences in adipose tissue distribution between men and women. Central obesity has adverse effects on metabolic parameters such as glucose tolerance, insulin sensitivity and lipid profiles (5). Beside known gender differences, adipose tissue distribution is also assumed to be genetically determined. Unfortunately, WHR has not been addressed as a “main” measure for adipose tissue distribution in GWAS and candidate gene studies until recently. This may be due to the necessity to clearly distinguish between subcutaneous or visceral adipose tissue derived obesity by means of MRI, CT, or DEXA. It is known that many genes are differentially expressed in subcutaneous and visceral adipose tissue depots and their mRNA expression correlates with measures of obesity which might reflect their physiological role. However, it can not be ruled out that expression is a consequence rather than the cause of increased fat mass or that the differential expression of the gene in adipose tissue is just secondary to another side of action. Studies investigating the influence of SNPs on the expression of these genes in adipose tissue depots may provide a hint on the cause-effect relationship, i.e. how a gene might influence the physiology. Since obtaining paired human biopsies from SC and Vis adipose tissue is often a laborious and time consuming undertaking, gene expression studies in fat are mostly conducted in small cohorts or in animal models, hence limiting the power for statistical analyses. This could partially explain a lack of association between the FTO SNP and expression. On the other hand, the results of genotype-expression correlations for FASN and BMPR2 clearly demonstrate the capability of this approach, as disease susceptibility SNPs are often located in introns, potentially containing regulatory sequences such as transcription factor binding sites, enhancer or repressor elements, and sites for epigenetic modifications. The regulation of gene expression through SNPs is one of the possibilities how genetic variants could exert their effect on metabolic pathways. Employing the “SNP-tagging” approach (95) based on linkage disequilibrium groups, the SNP found to be associated with obesity or related traits does not necessarily represent the “functional” variant. Therefore all SNPs belonging to one L.D. group would need to be examined in functional studies. Further, it needs to be taken into account that an enhanced gene expression does not necessarily lead to increased protein levels and vice versa. SNP association studies on protein levels are also promising as shown e.g. for CNR1 SNPs on leptin or 2-arachidonyglycerol or BMPR2 SNPs

CHAPTER 7 129 on insulin. Extensive functional studies are necessary to identify the part of the SNP/gene in the metabolic pathways leading to obesity, especially for genes of unknown function. We could confirm the association of the previously reported FTO SNP rs8050136 with BMI in the Sorbs cohort which queue in the line of robust replications of this locus in European populations (96). In the present study, an inverse relationship between obesity and FTO gene expression in visceral adipose tissue in humans was found for the first time. Although a binding site for the transcription factor CUTL was predicted for the sequence containing rs8050136 preferentially binding to the A obesity risk allele (97), probably due to the small sample size, the SNP did not associate with the FTO mRNA expression in the present study. Recently, a down regulation of FTO expression was found in diabetic alpha and beta islet cells, however, the gene expression was less associated with BMI than has been found for other tissues (98). The down-regulation of FTO in adipose tissue at increasing BMI does not necessarily need to be in contrast to the findings of Stratigopoulos et al., who found a decreased FTO expression in human fibroblasts when CUTL1 was knocked down by siRNA (97). Since the transcription factor could act as repressor as well as activator, one might postulate tissue specific actions. In contrast, Samaras et al. found neither a depot specificity between SC and Vis adipose tissue FTO mRNA expression nor a relation to direct measures of total or central abdominal adiposity in 16 subjects (99). It is noteworthy that recent studies indicate that FTO might be involved in central regulation of energy homeostasis (100) by modulating food intake (101). Its proposed function to act as a 2-oxoglutarate-dependent nucleic acid DNA/RNA demethylase (102) was supported by the crystal structure of the FTO protein which reveals basis for its substrate specificity (103). In mice, a dominant point mutation in Fto resulted in reduced fat mass, increased energy expenditure and gene expression profiling revealed increased expression of some fat and carbohydrate metabolism genes and improved inflammatory profile in white adipose tissue of mutant mice (104). This provides functional evidence that FTO is a causal gene underlying obesity. Nevertheless, in humans, brain rather than adipose tissue is assumed to be the primary site of FTO action. Even though the protective effect of the FASN Val1483Ile substitution in men was replicated in a second cohort of white men and also associated with a decreased adipose tissue FASN activity (105), no other study has investigated FASN genetic variants in association studies for obesity so far. It should be noted that the direction of FASN mRNA expression in adipose tissue in terms of obesity is still controversially discussed (106). Whereas the study presented in the paper suggests that the induction of FAS-catalyzed adipocyte lipogenesis may contribute to obesity, Ortega et al. speculate that adipose tissue may reach an inflexion

CHAPTER 7 130 point in which the lipogenic capacity is down-regulated in a natural inhibitory feedback process (106). FASN was described to be over-expressed in many cancers. Inhibitors of FASN have been shown to induce apoptosis in several cancer cell lines, and to induce tumor growth delay in several cancer xenograft models suggesting FASN to be a metabolic oncogene (107). Therefore, it might be speculated that genetic variants in FASN might modulate the prevalence of developing some form of cancer or the response to therapeutic interventions. Genetic variants in CNR1 seem to have rather moderate effects on obesity in German populations. However, based on the current findings, genetic variants in CNR1 can not be completely ignored. The CNR1 is considered as a therapeutic target (108), and therefore pharmacogenetic studies are still potentially interesting. Recently, Perwitz et al. demonstrated that CNR1 blockade with rimonabant or knocking down the receptor with siRNA in a murine SV40 cell model led to trans-differentiation of white adipocytes into a mitochondria-rich, thermogenic brown fat phenotype (109). Beside the known role of CNR1 in the brain it also suggests adipose tissue as a primary site of action. Investigating the variant of the SNP which seems to be protective against obesity might also exhibit a possibility to elucidate the physiological role of this gene in the pathophysiology of metabolic alterations. As recently shown for BMPR1A (88), and as presented in the thesis for BMPR2, genetic variants in two genes were associated with obesity and related traits whose proteins interact as receptor complex playing an important role in the commitment of mesenchymal stem cell into the adipogenic lineage. There is evidence that each fat depot has a unique developmental gene expression signature, independent from nutritional state, and supports the role of developmental genes in fat distribution and the development and/or function of specific adipose tissue depots (110). Moreover, in the present study, genetic variants in these genes have been shown to be associated with mRNA expression levels. The functionality of these variants is currently being investigated by means of luciferase assays. In conclusion, even though the identification of genetic variants which promote a risk to develop metabolic alterations and the investigations to clarify the functionality of the variants appear to be complex and challenging goals these studies are crucial to elucidate the role of genetic variants in complex diseases such as the metabolic syndrome. In the present thesis novel obesity risk variants were identified and it could be shown that their effect on gene expression levels might be a possible pathomechanistic link underlying the genotype- phenotype associtations.

CHAPTER 7 131

7.1 References - Introduction and Closing Discussion

1. Simmons RK, Alberti KG, Gale EA et al. The metabolic syndrome: useful concept or clinical tool? Report of a WHO Expert Consultation. Diabetologia 2010 53:600-605.

2. Kesäniemi YA, Lilja M, Kervinen K, Rantala A. Multiple metabolic syndrome: aspects of genetic epidemiology and molecular genetics. Ann Med 1992 24:461-464.

3. Joost HG, Giesen K, Kluge K, Ortlepp J, Plum L. Insulin resistance and metabolic syndrome. Z Kardiol 2000 89:377-382.

4. Eckel RH, Grundy SM, Zimmet PZ. The metabolic syndrome. Lancet 2005 365:1415-1428.

5. Salans LB, Knittle JL, Hirsch J. Role of Adipose Cell Size and Adipose Tissue Insulin Sensitivity in Carbohydrate Intolerance of Human Obesity. J Clin Invest 1968 47:153-165.

6. Gesta S, Tseng YH & Kahn CR. Developmental origin of fat: Tracking obesity to its source. Cell 2007 131:242-256.

7. Hill JW, Elias CF, Fukuda M et al. Direct Insulin and Leptin Action on Pro-opiomelanocortin Neurons Is Required for Normal Glucose Homeostasis and Fertility. Cell Metab 2010 11:286- 297.

8. Jager J, Gremeaux T, Cormont M, Marchand-Brustel Y, Tanti JF. Interleukin-1 beta-induced insulin resistance in adipocytes through down-regulation of insulin receptor substrate-1 expression. Endocrinology 2007 148:241-251.

9. Kim AY, Lee YS, Kim KH et al. Adiponectin Represses Colon Cancer Cell Proliferation via AdipoR1- and R2-Mediated AMPK Activation. Mol Endocrinol 2010 24:1141-1452.

10. Steppan CM, Bailey ST, Bhat S et al. The hormone resistin links obesity to diabetes. Nature 2001 409:307-312.

11. Vague J. The Degree of Masculine Differentiation of Obesities - A Factor Determining Predisposition to Diabetes, , Gout, and Uric Calculous Disease. Am J Clin Nutr 1956 4:20-34.

12. Compher C, Badellino KO. Obesity and Inflammation: Lessons From . JPEN J ParenterEnteral Nutr 2008 32:645-647.

13. Isakson P, Hammarstedt A, Gustafson B, Smith U. Impaired Preadipocyte Differentiation in Human Abdominal Obesity Role of Wnt, Tumor Necrosis Factor-alpha, and Inflammation. Diabetes 2009 58:1550-1557.

14. Ogawa M, Larue AC, Watson PM, Watson DK. Hematopoietic stem cell origin of mesenchymal cells: opportunity for novel therapeutic approaches. Int J Hematol 2010 91:353- 359.

15. Lean ME. Brown Adipose-Tissue in Humans. Proc Nutr Soc 1989 48:243-256.

16. Ravussin E. The Presence and Role of Brown Fat in Adult Humans. Curr Diabs Rep 2010 10:90-92.

17. Virtanen KA, et al. Functional Brown Adipose Tissue in Healthy Adults. N Engl J Med 2009 360:1518-1525.

CHAPTER 7 132

18. Langin D. Recruitment of brown fat and conversion of white into brown adipocytes: Strategies to fight the metabolic complications of obesity? Biochim Biophys Acta 2010 1801:372-376.

19. Lidell ME, Enerbäck S. Brown adipose tissue - a new role in humans? Nat Rev Endocrinol 2010 6:319-325.

20. Harman-Boehm I, Blüher M, Redel H et al. Macrophage infiltration into omental versus subcutaneous fat across different populations: Effect of regional adiposity and the comorbidities of obesity. J Clin Endocrinol Metab 2007 92:2240-2247.

21. Kersten S. Mechanisms of nutritional and hormonal regulation of lipogenesis. EMBO Rep 2001 2:282-286.

22. Lafontan M, Langin D. Lipolysis and lipid mobilization in human adipose tissue. Prog Lipid Res 2009 48:275-297.

23. Kissebah AH, Krakower GR. Regional Adiposity and Morbidity. Physiol Rev 1994 74:761- 811.

24. Bjorntorp P. Portal Adipose-Tissue As A Generator of Risk-Factors for Cardiovascular- Disease and Diabetes. Arteriosclerosis 1990 10:493-496.

25. Lafontan M, Berlan M. Do regional differences in adipocyte biology provide new pathophysiological insights? Trends Pharmacol Sci 2003 24:276-283.

26. Wajchenberg BL. Subcutaneous and visceral adipose tissue: Their relation to the metabolic syndrome. Endocr Rev 2000 21:697-738.

27. Gustafson B. Adipose tissue, inflammation and atherosclerosis. J Atheroscler Thromb 2010 17:332-341.

28. Klöting N, Stumvoll M, Blüher M. The biology of visceral fat. Internist 2007 48:126-133.

29. Suganami T, Ogawa Y. Adipose tissue macrophages: their role in adipose tissue remodeling. J Leukoc Biol 2010 88:1-7.

30. Arner E, Westermark PO, Spalding KL et al. Adipocyte Turnover: Relevance to Human Adipose Tissue Morphology. Diabetes 2010 59:105-109.

31. Bays HE, Gonzáles-Campoy JM, Bray GA et al. Pathogenic potential of adipose tissue and metabolic consequences of adipocyte hypertrophy and increased visceral adiposity. Expert Rev Cardiovasc Ther 2008 6:343-368.

32. Weyer C, Foley JE, Bogardus C, Tataranni PA, Pratley RE. Enlarged subcutaneous abdominal adipocyte size, but not obesity itself, predicts Type II diabetes independent of insulin resistance. Diabetologia 2000 43:1498-1506.

33. Müller-Wieland D, Knebel B, Haas J, Merkel M, Kotzka J. Adipositas: ektope Fettverteilung und Herz. Herz 2010 35:198-205.

34. McGarry JD. Dysregulation of fatty acid metabolism in the etiology of type 2 diabetes. Diabetes 2002 51:7-18.

35. Unger RH. Lipotoxic diseases. Annu Rev Med 2002 53:319-336.

CHAPTER 7 133

36. Dusserre E, Moulin P, Vidal H. Differences in mRNA expression of the proteins secreted by the adipocytes in human subcutaneous and visceral adipose tissues. Biochim Biophys Acta 2000 1500:88-96.

37. Fontana L, Eagon JC, Trujillo ME, Scherer PE, Klein S. Visceral fat adipokine secretion is associated with systemic inflammation in obese humans. Diabetes 2007 56:1010-1013.

38. Yang X, Smith U. Adipose tissue distribution and risk of metabolic disease: does thiazolidinedione-induced adipose tissue redistribution provide a clue to the answer? Diabetologia 2007 50:1127-1139.

39. Bastard JP, Maachi M, Van Nhieu JT et al. Adipose tissue IL-6 content correlates with resistance to insulin activation of glucose uptake both in vivo and in vitro. J Clin Endocrinol Metab 2002 87:2084-2089.

40. Mutch NJ, Wilson HM, Booth NA. Plasminogen activator inhibitor-1 and haemostasis in obesity. Proc Nutr Soc 2001 60:341-347.

41. Sartipy P, Loskutoff DJ. Monocyte chemoattractant protein 1 in obesity and insulin resistance. Proc Natl Acad Sci U S A 2003 100:7265-7270.

42. Stofkova A. Leptin and adiponectin: from energy and metabolic dysbalance to inflammation and autoimmunity. Endocr Regul 2009 43:157-168.

43. Yang RZ, Lee MJ, Hu H et al. Acute-phase serum amyloid A: An inflammatory adipokine and potential link between obesity and its metabolic complications. PLoS Med 2006 3:e287.

44. Kotronen A, Yki-Jarvinen H. Fatty liver - A novel component of the metabolic syndrome. Arterioscler Thromb Vasc Biol 2008 28:27-38.

45. Stefan N, Kantartzis K, Haring HU. Causes and Metabolic Consequences of Fatty Liver. Endocr Rev 2008 29:939-960.

46. Koster A, Stenholm S, Alley DE et al. Body Fat Distribution and Inflammation Among Obese Older Adults With and Without Metabolic Syndrome. Obesity (Silver Spring) 2010 [Epub ahead of print].

47. Farooqi IS, O'Rahilly S. Genetic factors in human obesity. Obes Rev 2007 8:37-40.

48. Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors in relative body weight and human adiposity. Behav Genet 1997 27:325-351.

49. Echwald SM. Genetics of human obesity: lessons from mouse models and candidate genes. J Intern Med 1999 245:653-666.

50. Chen H, Charlat O, Tartaglia LA et al. Evidence that the diabetes gene encodes the leptin receptor: Identification of a mutation in the leptin receptor gene in db/db mice. Cell 1996 84:491-495.

51. Huszar D, Lynch CA, Fairchild-Huntress V et al. Targeted disruption of the melanocortin-4 receptor results in obesity in mice. Cell 1997 88:131-141.

52. Naggert J, Harris T, North M. The genetics of obesity. Curr Opin Genet Dev 1997 7:398-404.

53. Zhang YY, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional Cloning of the Mouse Obese Gene and Its Human Homolog. Nature 1994 372:425-432.

CHAPTER 7 134

54. Farooqi IS, O'Rahilly S. Genetics of obesity in humans. Endocr Rev 2006 27:710-718.

55. Hinney A, Vogel CIG, Hebebrand J. From monogenic to polygenic obesity: recent advances. Eur Child Adolesc Psychiatry 2010 19:297-310.

56. Hoffstedt J, Eriksson P, Hellström L, Rössner S, Rydén M, Arner P. Excessive fat accumulation is associated with the TNF alpha-308 G/A promoter polymorphism in women but not in men. Diabetologia 2000 43:117-120.

57. Esterbauer H, Schneitler C, Oberkofler H et al. A common polymorphism in the promoter of UCP2 is associated with decreased risk of obesity in middle-aged humans. Nat Genet 2001 28:178-183.

58. Deeb SS, Fajas L, Nemoto M et al. A Pro12Ala substitution in PPAR gamma 2 associated with decreased receptor activity, lower body mass index and improved insulin sensitivity. Nat Genet 1998 20:284-287.

59. Lango H, Weedon MN. What will whole genome searches for susceptibility genes for common complex disease offer to clinical practice? J Intern Med 2008 263:16-27.

60. Herbert A, Gerry NP, McQueen MB et al. A common genetic variant is associated with adult and childhood obesity. Science 2006 312:279-283.

61. Dina C, Meyre D, Samson C et al. Comment on "A common genetic variant is associated with adult and childhood obesity". Science 2007 315:187.

62. Loos RJ, Barroso I, O'Rahilly S, Wareham NJ. Comment on "A common genetic variant is associated with adult and childhood obesity". Science 2007 315:187.

63. Lyon HN, Emilsson V, Hinney A et al. The association of a SNP upstream of INSIG2 with body mass index is reproduced in several but not all cohorts. PLoS Genet 2007 3:e61.

64. Rosskopf D, Bornhorst A, Rimmbach C et al. Comment on "A common genetic variant is associated with adult and childhood obesity". Science 2007 315:187.

65. Talbert ME, Langefeld CD, Ziegler JT, Haffner SM, Norris JM, Bowden DW. INSIG2 SNPs Associated With Obesity and Glucose Homeostasis Traits in Hispanics: The IRAS Family Study. Obesity 2009 17:1554-1562.

66. Liu YJ, Liu XG, Wang L et al. Genome-wide association scans identified CTNNBL1 as a novel gene for obesity. Hum Mol Genet 2008 17:1803-1813.

67. Loos RJ, Lindgren CM, Li S et al. Common variants near MC4R are associated with fat mass, weight and risk of obesity. Nat Genet 2008 40:768-775.

68. Wellcome Trust Case Control Consortium.. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 2007 447:661-678.

69. Frayling TM, Timpson NJ, Weedon MN et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 2007 316:889-894.

70. Hinney A, Nguyen TT, Scherag A et al. Genome Wide Association (GWA) Study for Early Onset Extreme Obesity Supports the Role of Fat Mass and Obesity Associated Gene (FTO) Variants. PLoS One 2007 2:e1361.

CHAPTER 7 135

71. Scuteri A, Sanna S, Chen WM et al. Genome-wide association scan shows genetic variants in the FTO gene are associated with obesity-related traits. PLoS Genet 2007 3:e115.

72. Meyre D, Delplanque J, Chèvre JC et al. Genome-wide association study for early-onset and morbid adult obesity identifies three new risk loci in European populations. Nat Genet 2009 41:157-159.

73. Thorleifsson G, Walters GB, Gudbjatsson et al. Genome-wide association yields new sequence variants at seven loci that associate with measures of obesity. Nat Genet 2009 41:18- 24.

74. Willer CJ, Speliotes EK, Loos RJ et al. Six new loci associated with body mass index highlight a neuronal influence on body weight regulation. Nat Genet 2009 41:25-34.

75. Hofker M, Wijmenga C. A supersized list of obesity genes. Nat Genet 2009 41:139-140.

76. Lindgren CM, Heid IM, Randall JC et al. Genome-Wide Association Scan Meta-Analysis Identifies Three Loci Influencing Adiposity and Fat Distribution. PLoS Genet 2009 5:e1000508.

77. Maeda S, Tsukada S, Kanazawa A et al. Genetic variations in the gene encoding TFAP2B are associated with type 2 diabetes mellitus. J Hum Genet 2005 50:283-292.

78. Tsukada S, Tanaka Y, Maegawa H, Kashiwagi A, Kawamori R, Maeda S. Intronic polymorphisms within TFAP2B regulate transcriptional activity and affect adipocytokine gene expression in differentiated adipocytes. Mol Endocrinol 2006 20:1104-1111.

79. Brem RB, Kruglyak L. The landscape of genetic complexity across 5,700 gene expression traits in yeast. Proc Natl Acad Sci U S A 2005 102:1572-1577.

80. Vidal H. Gene expression in visceral and subcutaneous adipose tissues. Ann Med 2001 33:547-555.

81. Berndt J, Kovacs P, Ruschke K et al. Fatty acid synthase gene expression in human adipose tissue: association with obesity and type 2 diabetes. Diabetologia 2007 50:1472-1480.

82. Blüher M, Engeli S, Klöting N et al. Dysregulation of the peripheral and adipose tissue endocannabinoid system in human abdominal obesity. Diabetes 2006 55:3053-3060.

83. Gesta S, Blüher M, Yamamoto Y et al. Evidence for a role of developmental genes in the origin of obesity and body fat distribution. Proc Natl Acad Sci U S A 2006 103:6676-6681.

84. Klöting N, Graham TE, Berndt J et al. Serum retinol-binding protein is more highly expressed in visceral than in subcutaneous adipose tissue and is a marker of intra-abdominal fat mass. Cell Metab 2007 6:79-87.

85. Klöting N, Berndt J, Kralisch S et al. Vaspin gene expression in human adipose tissue: Association with obesity and type 2 diabetes. Biochem Biophys Res Commun 2006 339:430- 436.

86. Lefebvre AM, Laville M, Vega N et al. Depot-specific differences in adipose tissue gene expression in lean and obese subjects. Diabetes 1998 47:98-103.

87. Giacchetti G, Faloia E, Sardu C et al. Gene expression of angiotensinogen in adipose tissue of obese patients. Int J Obes Relat Metab Disord 2000 24:S142-S143.

CHAPTER 7 136

88. Böttcher Y, Unbehauen H, Klöting N et al. Adipose Tissue Expression and Genetic Variants of the Bone Morphogenetic Protein Receptor 1A Gene (BMPR1A) Are Associated With Human Obesity. Diabetes 2009 58:2119-2128.

89. Klöting N, Berthold S, Kovacs P et al. MicroRNA Expression in Human Omental and Subcutaneous Adipose Tissue. PLoS One 2009 4:e4699.

90. Lindgren CM, McCarthy MI. Mechanisms of Disease: genetic insights into the etiology of type 2 diabetes and obesity. Nat Clin Pract Endocrinol Metab 2008 4:156-163.

91. Knowler WC, Bennett PH, Hamman RF, Miller M. Diabetes Incidence and Prevalence in Pima Indians - 19-Fold Greater Incidence Than in Rochester, Minnesota. Am J Epidemiol 1978 108:497-505.

92. Baier LJ, Hanson RL. Genetic studies of the etiology of type 2 diabetes in Pima Indians - Hunting for pieces to a complicated puzzle. Diabetes 2004 53:1181-1186.

93. Novembre J, Johnson T, Bryc K et al. Genes mirror geography within Europe. Nature 2008 456:98-101.

94. Weiss LA, Pan L, Abney M, Ober C. The sex-specific genetic architecture of quantitative traits in humans. Nat Genet 2006 38:218-222.

95. International HapMap Consortium. The International HapMap Project. Nature 2003 426:789- 796.

96. Tönjes A, Zeggini E, Kovacs P et al. Association of FTO variants with BMI and fat mass in the self-contained population of Sorbs in Germany. Eur J Hum Genet 2010 18:104-110.

97. Stratigopoulos G, Padilla SL, LeDuc CA et al. Regulation of Fto/Ftm gene expression in mice and humans. Am J Physiol Regul Integr Comp Physiol 2008 294, R1185-R1196.

98. Kirkpatrick CL, Marchetti P, Purrello F et al. Type 2 diabetes susceptibility gene expression in normal or diabetic sorted human alpha and beta cells: correlations with age or BMI of islet donors. PLoS One 2010 5:e11053.

99. Samaras K, Botelho NK, Chisholm DJ, Lord RV. Subcutaneous and Visceral Adipose Tissue FTO Gene Expression and Adiposity, Insulin Action, Glucose Metabolism, and Inflammatory Adipokines in Type 2 Diabetes Mellitus and in Health. Obes Surg 2010 20:108-113.

100. Fredriksson R, Hägglund M, Olszewski PK et al. The obesity gene, FTO, is of ancient origin, up-regulated during food deprivation and expressed in neurons of feeding-related nuclei of the brain. Endocrinology 2008 149:2062-2071.

101. Speakman JR, Rance KA, Johnstone AM. Polymorphisms of the FTO gene are associated with variation in energy intake, but not energy expenditure. Obesity 2008 16:1961-1965.

102. Gerken T, Girard CA, Tung YC et al. The obesity-associated FTO gene encodes a 2- oxoglutarate-dependent nucleic acid demethylase. Science 2007 318:1469-1472.

103. Han Z, Niu T, Chang J et al. Crystal structure of the FTO protein reveals basis for its substrate specificity. Nature 2010 464:1205-1209.

104. Church C, Lee S, Bagg EA et al. A Mouse Model for the Metabolic Effects of the Human Fat Mass and Obesity Associated FTO Gene. PLoS Genet 2009 5:1000599.

CHAPTER 7 137

105. Moreno-Navarrete JM, Botas P, Valdés S et al. Val1483Ile in FASN Gene Is Linked to Central Obesity and Insulin Sensitivity in Adult White Men. Obesity 2009 17:1755-1761.

106. Ortega FJ, Mayas D, Moreno-Navarete JM et al. The Gene Expression of the Main Lipogenic Enzymes is Downregulated in Visceral Adipose Tissue of Obese Subjects. Obesity 2010 18: 13-20.

107. Flavin R, Peluso S, Nguyen PL, Loda M. Fatty acid synthase as a potential therapeutic target in cancer. Future Oncol 2010 6:551-562.

108. Kipnes MS, Hollander P, Fujioka K et al. A one-year study to assess the safety and efficacy of the CB1R inverse agonist taranabant in overweight and obese patients with type 2 diabetes. Diabetes Obes Metab 2010 12:517-531.

109. Perwitz N, Wenzel J, Wagner I et al. Cannabinoid type 1 receptor blockade induces transdifferentiation towards a brown fat phenotype in white adipocytes. Diabetes Obes Metab 2010 12, 158-166.

110. Yamamoto Y, Gesta S, Lee KY, Tran TT, Saadatirad P, Kahn CR. Adipose Depots Possess Unique Developmental Gene Signatures. Obesity 2010 18:872-878.

APPENDIX Curriculum Vitae 138

APPENDIX

Curriculum Vitae

Personal

Surname: Schleinitz, nee Vogelsang Name: Dorit Date of Birth: March 21, 1981 Place of Birth: Karl-Marx-Stadt (Chemnitz) Nationality: German

Education/Qualifications

01/2011 Final examination to obtain the degree of doctor rerum naturalium (Dr. rer. nat.), Institute of Biochemistry, Faculty of Biosciences, Pharmacy and Psychology, University of Leipzig, Germany 04/2007 – 12/2010 PhD position under supervision of Prof. A.G. Beck-Sickinger, Institute of Biochemistry, Faculty of Biosciences, Pharmacy and Psychology, University of Leipzig, Germany and PD Dr. P. Kovacs, Research Group “Metabolic Syndrome”, Medical Faculty, University of Leipzig, Germany 10/2004 – 09/2006 M.Sc. Molecular Biotechnology Technical University Munich, Germany 10/2000 – 03/2004 BSc Molecular Biotechnology Technical University Dresden, Germany 07/1999 University entrance diploma Dr.-Wilhelm-André Gymnasium, Chemnitz, Germany

Work history

10/2006 – 03/2007 Research associate: Biotype® company, Dresden, Germany 05/2005 – 08/2005 Student research assistant: Department for game animals biology & management, University Munich, Germany 02/2004 – 08/2004 Practical course: Biotype® company, Dresden, Germany

Grants, Reviews

2010 European Foundation for the Study of Diabetes Albert Renold Fellowship for Young Scientists 2009 German Diabetes Association (DDG) project promotion: Functional characterization of obesity associated variants in the BMPR1A gene 2008/2009 EASD travel grant 2008 DDG travel grant

APPENDIX Publications & Presentations 139

Publications & Presentations

Publications

Schleinitz D, Klöting N, Böttcher Y, Wolf S, Dietrich K, Tönjes A, Breitfeld J, Enigk B, Halbritter J, Körner A, Schön MR, Jenkner J, Tseng YH, Lohmann T, Dreßler M, Stumvoll M, Blüher M, Kovacs P. (2010) Genetic and Evolutionary Analyses of the Human Bone Morphogenetic Protein Receptor 2 (BMPR2) in the Pathophysiology of Obesity. PloS ONE [accepted]

Schleinitz D, Carmienke S, Böttcher Y, Tönjes A, Berndt J, Klöting N, Enigk B, Müller I, Dietrich K, Breitfeld J, Scholz GH, Engeli S, Stumvoll M, Blüher M, Kovacs P. (2010) The role of genetic variation in the cannabinoid type 1 receptor gene (CNR1) in the pathophysiology of human obesity. Pharmacogenomics 11:693-702.

Schleinitz D, Klöting N, Körner A, Berndt J, Reichenbächer M, Tönjes A, Ruschke K, Böttcher Y, Dietrich K, Enigk B, Filz M, Schön MR, Jenkner J, Kiess W, Stumvoll M, Blüher M, Kovacs P. (2010) Effect of Genetic Variation in the Human Fatty Acid Synthase Gene (FASN) on Obesity and Fat Depot- Specific mRNA Expression. Obesity (Silver Spring) 18:1218-1225.

Tönjes A, Koriath M, Schleinitz D, Dietrich K, Böttcher Y, Rayner NW, Almgren P, Enigk B, Richter O, Rohm S, Fischer-Rosinsky A, Pfeiffer A, Hoffmann K, Krohn K, Aust G, Spranger J, Groop L, Blüher M, Kovacs P, Stumvoll M. (2009) Genetic variation in GPR133 is associated with height: genome wide association study in the self-contained population of Sorbs. Human Molecular Genetics 18:4662-4668.

Schleinitz D*, Tönjes A*, Böttcher Y, Dietrich K, Enigk B, Koriath M, Scholz GH, Blüher M, Zeggini E, McCarthy MI, Kovacs P, Stumvoll M. (2009) Lack of significant effects of the type 2 diabetes susceptibility loci JAZF1, CDC123/CAMK1D, NOTCH2, ADAMTS9, THADA and TSPAN8/LGR5 on diabetes and quantitative metabolic traits. Hormone Metabolic Research 42:14-22. *contributing equally

Tönjes A, Zeggini E, Kovacs P, Böttcher Y, Schleinitz D, Dietrich K, Morris AP, Enigk B, Rayner NW, Koriath M, Eszlinger M, Kemppinen A, Prokopenko I, Hoffmann K, Teupser D, Thiery J, Krohn K, McCarthy MI, Stumvoll M. (2009) Association of FTO variants with BMI and fat mass in the self- contained population of Sorbs in Germany. European Journal of Human Genetics 18:104-110.

Böttcher Y, Unbehauen H, Klöting N, Ruschke K, Körner A, Schleinitz D, Tönjes A, Enigk B, Dietrich K, Koriath M, Scholz GH, Tseng YH, Schön MR, Kiess W, Stumvoll M, Blüher M, Kovacs P. (2009) Adipose Tissue Expression and Genetic Variants of the Bone Morphogenetic Protein Receptor 1A Gene (BMPR1A) are Associated with Human Obesity. Diabetes 58:2119-2128.

Böttcher Y, Schleinitz D, Tönjes A, Blüher M, Stumvoll M, Kovacs P. (2008) R1467H Variant in the Rho Guanine Nucleotide Exchange Factor 11 (ARHGEF11) is associated with impaired glucose tolerance and Type 2 Diabetes in German Caucasians. Journal of Human Genetics 53:365-367.

Klöting N*, Schleinitz D*, Ruschke K, Berndt J, Fasshauer M, Tönjes A, Schön MR, Kovacs P, Stumvoll M, Blüher M. (2008) Inverse relationship between obesity and FTO gene expression in visceral adipose tissue in humans. Diabetologia 51:641-647. * contributing equally

Kovacs P, Geyer M, Berndt J, Klöting N, Graham TE, Böttcher Y, Enigk B, Tönjes A, Schleinitz D, Schön MR, Kahn BB, Blüher M, Stumvoll M. (2007) Effects of Genetic Variation in the Human Retinol Binding Protein-4 Gene (RBP4) on Insulin Resistance and Fat Depot Specific mRNA Expression. Diabetes 56: 3095-3100.

Zalan A, Volgyi A, Brabetz W, Schleinitz D, Pamjav H. (2008) Hungarian population data of eight X- linked markers in four linkage groups. Forensic Science International 175:73-78.

Becker D, Vogelsang D, Brabetz W. (2007) Population data on the seven short tandem repeat loci D4S2366, D6S474, D14S608, D19S246, D20S480, D21S226 and D22S689 in a German population. International Journal of Legal Medicine 121:78-81.

APPENDIX Publications & Presentations 140

Book chapter

Schleinitz D, DiStefano JK, Kovacs P. Targeted SNP genotyping using the TaqMan® assay. In Disease Gene Identification: Methods and Protocols. Series: Methods in Molecular Biology. Springer/Humana Press. [in press]

Published Abstracts

Oral presentations

Schleinitz D, Klöting N, Böttcher Y, Berndt J, Wolf S, Ruschke K, Dietrich K, Koriath M, Enigk B, Scholz GH, Tseng YH, Tönjes A, Stumvoll M, Kovacs P, Blüher M. (2009) Bone Morphogenetic Protein Receptor 2 (BMPR2) mRNA Expression im Fettgewebe und genetische Varianten von BMPR2 sind mit humaner Adipositas assoziiert. Diabetologie und Stoffwechsel 4:[Suppl1]S4 44th Annual Meeting of the German Diabetes Association (DDG), Leipzig, Germany

Böttcher Y, Schleinitz D, Unbehauen H, Klöting N, Ruschke K, Enigk B, Tönjes A, Wolf S, Tseng YH, Blüher M, Stumvoll M, Kovacs P. (2008) Expression of the Bone Morphogenetic Protein Receptor 1A Gene (BMPR1A) in Adipose Tissue and BMPR1A Genetic Variants are Associated with Human Obesity. Diabetologie und Stoffwechsel 3:[Suppl1]S29 43rd Annual Meeting of the German Diabetes Association (DDG), Munich, Germany

Poster presentations

Schleinitz D, Klöting N, Wolf S, Ruschke K, Dietrich K, Koriath M, Enigk B, Scholz GH, Tseng YH, Tönjes A, Stumvoll M, Blüher M, Kovacs P. (2009) Expression of the Bone Morphogenetic Protein Receptor 2 (BMPR2) in Adipose Tissue and BMPR2 Genetic Variation are Related to Human Obesity. Diabetologia 52:[Suppl1]S249 45th Annual Meeting of the European Association for the Study of Diabetes (EASD), Vienna, Austria

Schleinitz D, Böttcher Y, Carmienke S, Tönjes A, Berndt J, Klöting N, Ruschke K, Scholz GH, Blüher M, Stumvoll M, Kovacs P. (2008) The role of genetic variation in the cannabinoid type 1 receptor gene (CB1R) in the pathophysiology of human obesity. Diabetologia 51:[Suppl1]S124 44th Annual Meeting of the European Association for the Study of Diabetes (EASD), Rom, Italy

Schleinitz D, Klöting N, Ruschke K, Berndt J, Fasshauer M, Schön MR, Stumvoll M, Blüher M, Kovacs P. (2007) FTO and RPGRIP1L Gene Expression in Adipose Tissue in Humans: The Role of FTO Genetic Variants. Final Program/Abstract Book ISBN:3-9810760-3-6 XV Lipid Meeting, Leipzig, Germany

APPENDIX Selbstständigkeitserklärung 141

Erklärung über die eigenständige Abfassung der Arbeit

Hiermit erkläre ich, Dorit Schleinitz, dass ich die vorliegende Arbeit selbstständig und ohne unzulässige Hilfe oder Benutzung anderer als der angegebenen Hilfsmittel angefertigt habe. Ich versichere, dass Dritte von mir weder unmittelbar noch mittelbar geldwerte Leistungen für die Arbeit erhalten haben, die im Zusammenhang mit dem Inhalt der vorgelegten Dissertation stehen, und dass die vorgelegte Arbeit weder im Inland noch im Ausland in gleicher oder ähnlicher Form einer anderen Prüfungsbehörde zum Zwecke einer Promotion oder eines anderen Prüfungsverfahrens vorgelegt wurde. Alles aus anderen Quellen und von anderen Personen übernommene Material, das in der Arbeit verwendet wurde oder auf das direkt Bezug genommen wird, wurde als solches kenntlich gemacht. Insbesondere wurden alle Personen genannt, die direkt an der Entstehung der vorliegenden Arbeit beteiligt waren. Weiterhin erkläre ich, dass ich das einleitende Kapitel, sowie die abschließende Diskussion alleinig und nur unter Benutzung der angegebenen Quellen verfasst habe.

Leipzig, 02.09.2010

Dorit Schleinitz

Danksagung

Danksagung

Zur Arbeitsgruppe von PD Dr. Peter Kovacs gehören zu dürfen ist das Beste was mir passiert ist. Lieber Peter, thanks a ton und noch viel mehr für Deine fantastische Betreuung, für die Überlassung eines überaus spannenden Themas für meine Doktorarbeit, dafür dass Du mich so sehr gefördert hast und für Dein Vertrauen. Das mit dem „großen“ Paper bekommen wir auch noch hin :-).

Mein aufrichtiger Dank gilt Frau Prof. Dr. Annette Beck-Sickinger für die sehr gute Betreuung an der Fakultät für Biowissenschaften, Pharmazie und Psychologie und damit für die Möglichkeit, diese Promotion als Dr. rer. nat. abschließen zu können.

Viele großartige Menschen und Arbeitsgruppen prägen das Wissenschafts- und Arbeitsklima hier an der Universität und am Uniklinikum Leipzig. Ein Glück, dass ich die Möglichkeit bekam, sie kennen zu lernen. Für Hilfe und Begleitung bei Projekten, für gute Diskussionen und Anregungen danke ich Prof. Michael Stumvoll, Prof. Matthias Blüher, Prof. Gabi Aust, Prof. Wieland Kiess und Prof. Antje Körner. Ein liebes Dankeschön geht an dieser Stelle auch an Anke Tönjes, Yvonne Böttcher, Nora Klöting, Daniela Friebe, Petra Büttner, Susan Kralisch & an die ehemaligen Insulaner Kathleen Stein, Ulrike Winkler, Marcello Sestu, Robert Requardt, Knut Krohn, Johannes Hirrlinger, Heidrun Misch und Helmut Helfricht.

Dass ihr es bis hierher mit mir ausgehalten habt, dass ihr mich immer unterstützt habt und aufgepasst habt, das alles wieder gut wird – dafür kann ich euch nicht genug danken: Beate Enigk – unser Goldstück, Jana Breitfeld & Kerstin Dietrich – die nächsten tapferen Anwärter auf den Hut :-), Ines Müller – die Energie-geladene und Manuela Prellberg – unser fleißiges „Küken“. Wir haben soviel Spaß zusammen und den Kopf voller kreativer, verrückter Ideen – ich sage nur „Kovacs Girls“ forever. Ein herzliches Dankeschön auch an unsere vielen engagierten Medizin-Doktoranden.

Einige unter den auf dieser Seite genannten sind wirklich liebe Freunde geworden – ich wünsche mir sehr fürs Leben. Danke fürs Zuhören, fürs Zeit haben & für euer Vertrauen.

Für alle Liebe, Unterstützung, Zuversicht, Kraft & Rückhalt danke ich aus tiefstem Herzen meiner geliebten Familie, Mutti & Vati, Markus & Tobi, meinem geliebten Mann Armin, meinen langjährigen engen Freunden und meiner liebsten Schwiegerfamilie.