The human metabolic syndrome is characterized by a heterogenic complex of symptoms, including central obesity. Obesity itself is linked to major features of the metabolic syndrome such as insulin resistance, dyslipidemia or type 2 diabetes 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 (body mass index) 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 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.

Obesity is characterized by the accumulation of adipose tissue (AT) either in the major adipose depots, the subcutaneous (SAT) or visceral (VAT), or ectopically in other tissues and organs such as muscle or liver. Indeed, AT is more than just a big energy storage filled with lipids or an isolation layer for inner organs. Since the first AT-secreted components were discovered the general thinking about AT changed fundamentally. Today, AT is regarded as the largest endocrine organ that releases bioactive compounds named adipokines which fulfill a variety of biological functions ranging from the orchestration of hunger and satiety to the modulation of immune responses or inflammation. Obesity is almost always accompanied by noncommunicable diseases like diabetes mellitus, cardiovascular diseases or several types of cancer. Globally, more than 70 % of prematurely deaths are caused by noncommunicable diseases. So far, the mechanisms how obesity contributes to or causes these diseases are not fully understood. However, most patients with obesity suffer from symptoms of AT dysfunction as result of adipocyte hypertrophy, hypoxia, fibrosis or immune cell infiltration. Finally, obesity is linked to altered adipokine secretion which together with circulating free fatty acids (FFA) contribute to chronic AT inflammation which is believed to be a link between obesity and diseases. However, obesity is very heterogeneous, and about 20 % of all people with obesity are classified as metabolically healthy. In this context, body shape and therefore the distribution of AT within the body is of immense importance. This may also be explained by adipokine secretion patterns, as patients with more prominent subcutaneous obesity are more likely to be metabolically healthy than patients with predominant visceral AT accumulations. Therefore, it is of enormous scientific and public interest to comprehend genetic mechanisms regulating AT development and distribution, but also to better understand the concept of healthy and unhealthy obesity for a targeted obesity therapy or prevention. In chapter 1, I aimed to investigate the impact of Homeobox C9 (Hoxc9) gene knockout on AT development and distribution. Initially, Hoxc9 was determined as interesting candidate gene, because it was found twice as high expressed in subcutaneous than in visceral AT. Therefore, mice with an expected deletion of Hoxc9 in AT (ATHoxc9-/-) were generated by crossing floxed Hoxc9lox/lox with fatty acid-binding protein 4 (Fabp4)-Cre recombinase expressing mice. The study was driven by the hypothesis, that adipocyte-specific loss of Hoxc9 will result in impaired AT development or an imbalanced inguinal WAT (ingWAT) to epigonadal WAT (eWAT) ratio compared to littermate controls. Furthermore, ATHoxc9-/- mice were expected to be a potential model mimicking unhealthy obesity, due to potential smaller ingWAT vs. eWAT depots, finally contributing to a better understanding how obesity is linked to noncommunicable diseases. Mice of both sexes were included in the study and either fed a standard chow diet (CD) or a calorie-dense high-fat diet (HFD), to promote diet-induced obesity (DIO). Whereas female knockout mice did not show phenotypic abnormalities, repeated body weight measurements revealed lower weight gain for male ATHoxc9-/- mice under HFD. Additionally, the HFD fed ATHoxc9-/- mice exhibited better glucose tolerance, whereas also ATHoxc9-/- mice on CD had lower fasting glucose levels compared to littermates. Regarding the adipocyte size, ATHoxc9-/- mice are characterized by smaller adipocytes of both subcutaneous and visceral origin, whereas HFD lead to slightly enlarged eWAT adipocytes. In line with the leaner phenotype and better glucose tolerance, OLINK serum protein analyzes detected lower levels of inflammatory markers (e.g., Ccl5, Il17a or Il17f) in ATHoxc9-/- mice. However, analyzing the knockout efficiency revealed only minor, but not significant tendency of Hoxc9 reductions in AT on both, mRNA and protein level. Additionally, DNA fragments within the targeted Hoxc9 exon regions were still detectable in both ATHoxc9-/- and littermates. Based on this, I wanted to clarify whether the Fabp4-Cre model is adequate to induce an AT-specific knockout in genes expressed as early as Hoxc9. Therefore, I investigated several in vitro and ex vivo models. First, I cultured and differentiated three different adipocyte cell lines (3T3-L1, immortalized inguinal and epigonadal white adipocytes and SVF cells isolated from mouse AT depots) to investigate gene expression patterns of both Hoxc9 and Fabp4 from undifferentiated and mature adipocytes, as well as during differentiation. In all models, Hoxc9 was already expressed early in the undifferentiated stages, whereas Fabp4 expression started later during adipocyte maturation. Additionally, to translate these findings in vivo, I sacrificed pups within their first 20 days of age and dissected developing ingWAT depots. Measurements of Hoxc9 and Fabp4 gene and protein levels supported the previous in vitro results. Therefore, I conclude that the Fabp4 mediated Cre recombinase is not suited for the investigation of a gene’s role in adipose tissue development. Furthermore, this finding should be transferable in the context of developmental research in other tissues or organs. Accordingly, I strongly encourage researchers who wants to perform Cre-mediated gene knockouts to check expression patterns of both, the Cre-mediating gene and the gene of interest in appropriate cell models beforehand to save time and costs, but most importantly to focus on animal welfare. In addition, the missing control group of Fabp4-Cre mice may represent a second major limitation of the study. Several previous studies have reported phenotypes resulting from Cre expression alone. Therefore, and because the presence of the Fabp4-Cre recombinase was the only validated genetic difference between our ATHoxc9-/- (Hoxc9lox/lox, Fabp4-Cre+) and the littermate controls (Hoxc9lox/lox, Fabp4-Cre-), I concluded that the phenotypic differences observed between both groups resulted from the Fabp4-Cre genotype itself. Even while Hoxc9 was found to be expressed in earlier stages of adipogenesis than Fabp4, it was surprising and unexpected that Cre recombinase expression did not ablate Hoxc9 in mature adipocytes. The detection of potential expression differences was impeded by the overall low expression of Hoxc9 in AT, even if it was found differentially expressed between ingWAT and eWAT. To investigate the Hoxc9 impact on adipogenesis other Cre models mediated by promotors active during earlier stages of adipocyte differentiation should be used. Alternative adipocyte-specific Cre expressing mouse lines using the adiponectin (Adipoq) promotor or the less known resistin (Retn) promotor can be excluded for Hoxc9 as they are most active only in fully mature adipocytes. In contrast, platelet derived growth factor receptor α (Pdgfra) promotor controlled Cre expression is already active during the adipocyte progenitor stage and could be better suited to target Hoxc9. Yet, this Cre model accompanies with other pitfalls. In line with its expression in progenitor cells, the Hoxc9 gene will not just be deleted in cells with adipogenic fate, but also in progenitors developing into e.g., chondrocytes or osteoblasts. According to Hoxc9’s role in skeletal patterning during the embryogenesis, using this model may result in malformations or premature death. Finally, the paired related homeobox 1 (Prrx1) Cre line is a likely suitable model . This promotor was found to be active early in progenitor cells later found in ingWAT depots and may be suitable to investigate the role of Hoxc9 in AT development and distribution.:Table of contents 1 1. Introduction 2 Obesity – a Global Pandemic 2 The Human Adipose Tissue 2 Obesity and Dysfunction of Adipose Tissue 4 Society, Environment and Genetics in the Pathogenesis of Obesity and AT Distribution 6 Rationale for Chapter 1 8 The Developmental Control Genes of the Homeobox Family 8 Aim of the Study in Chapter 1 11 Rationale for Chapter 2 11 Adenylyl Cyclases 12 How Adenylyl Cyclases Impact Diabetes 14 Aim of the Study in Chapter 2 16 The Cre-loxP System and Generation Knockout Mouse Models 17 2. Chapter 1 19 3. Chapter 2 38 4. Summary 60 5. References 65 6. Appendix 75 A. Supplementary Material - Chapter 1 75 B. Supplementary Material - Chapter 2 82 Erklärung über die eigenständige Abfassung der Arbeit 95 Curriculum Vitae 96 List of Publications and Talks 97 Danksagung 99 / In chapter 2, I investigated the consequences of a whole-body ablation of adenylyl cyclase 5 (Adcy5) in mice. Genome-wide association studies (GWAS) identified human ADCY5 as a candidate for diabetes associated traits and a higher risk to develop type 2 diabetes (T2D). In several mouse models the loss of Adcy5 was described as beneficial. For instance, Adcy5-/- mice have shown improved cardiac function and an increased longevity, mimicking a phenotype resulting from caloric restriction. Furthermore, they were protected against DIO and did not develop glucose intolerance or insulin resistance. Therefore, my study focused on the AT phenotype and global AT gene expression profiles resulting from the whole-body loss of Adcy5. Again, I comprehensively investigated animals of both sexes, both under CD or HFD. Unexpectedly, our mice readily developed DIO. Moreover, female mice under HFD developed more severe obesity compared to control animals. Female C57BL/6 mice are in general not as susceptible to DIO as males from the same strain and usually develop later-onset and less severe obesity. Therefore, it was surprising that the loss of Adcy5 induced DIO-development in females, an effect known from ovariectomized mice. Since Adcy5 is highly expressed in the ovaries and the expression of the key enzyme of estrogen production, aromatase, is regulated by Adcy-generated cyclic adenosine monophosphate (cAMP), further studies are necessary to validate the hypothesis, that Adcy5 may affect DIO-susceptibility via the cAMP-aromatase-estrogen axis. Besides, our model did not reproduce the beneficial effects on energy expenditure, oxygen consumption or insulin sensitivity described by others. Our phenotype is more similar and matches to those described for Adcy3 deficient mice, where knockouts gained more weight under HFD than littermate controls. Again, a shortcoming of our study was not using littermate controls. Knockouts were generated by integrating a vector cassette into Adcy5 exon 1 of C57BL/6NTac mice. Mice of the same strain sharing more than 99% background identity were used as controls. However, this could explain the phenotype differences seen in our compared to earlier reported Adcy5 knockout models. Nevertheless, our model enabled us to investigate the impact of Adcy5 on AT morphology and gene expression independently of body weight or a metabolically healthier phenotype. Except for the female mice on HFD, all Adcy5 deficient mice exhibited an increased number of smaller adipocytes which is strongly correlated to a reduced diabetes risk by retaining insulin sensitivity. Conspicuously, all Adcy5-/- mice presented a tremendously reduced running wheel activity. Beside several genes of glucose and lipid metabolism, global gene expression analysis within AT highlighted sodium-potassium ATPase catalytic subunit alpha-3 (Atp1a3) as significantly higher expressed in Adcy5-/- mice. Mutations within Atp1a3 are known to be related to e.g., dyskinesia and could be a relevant finding to understand how the loss of Adcy5 is connected to movement disorders. Based on GWAS and other earlier studies in mice that highlighted beneficial effects resulting from its loss, ADCY5 was considered as promising drug target to treat diabetes. In summary, my work suggests that these positive effects are strongly dependent on the genetic background and also gender-dependent. Especially, significantly reduced voluntary exercise activity combined with readily developing DIO cast doubt on the beneficial potential of ADCY5 inhibitors.:Table of contents 1 1. Introduction 2 Obesity – a Global Pandemic 2 The Human Adipose Tissue 2 Obesity and Dysfunction of Adipose Tissue 4 Society, Environment and Genetics in the Pathogenesis of Obesity and AT Distribution 6 Rationale for Chapter 1 8 The Developmental Control Genes of the Homeobox Family 8 Aim of the Study in Chapter 1 11 Rationale for Chapter 2 11 Adenylyl Cyclases 12 How Adenylyl Cyclases Impact Diabetes 14 Aim of the Study in Chapter 2 16 The Cre-loxP System and Generation Knockout Mouse Models 17 2. Chapter 1 19 3. Chapter 2 38 4. Summary 60 5. References 65 6. Appendix 75 A. Supplementary Material - Chapter 1 75 B. Supplementary Material - Chapter 2 82 Erklärung über die eigenständige Abfassung der Arbeit 95 Curriculum Vitae 96 List of Publications and Talks 97 Danksagung 99

The human metabolic syndrome is characterized by a heterogenic complex of symptoms, including central obesity. Obesity itself is linked to major features of the metabolic syndrome such as insulin resistance, dyslipidemia or type 2 diabetes 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 (body mass index) 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 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.