Epigenetics of Adipose Tissue and its related disorders

! AUTOR: Daniel Castellano Castillo http://orcid.org/0000-0001-8041-8244! EDITA: Publicaciones y Divulgación Científica. Universidad de Málaga ! Esta obra está bajo una licencia de Creative Commons Reconocimiento-NoComercialSinObraDerivada 4.0 Internacional:! http://creativecommons.org/licenses/by-nc-nd/4.0/legalcode! Cualquier parte de esta obra se puede reproducir sin autorización ! pero con el reconocimiento y atribución de los autores.! No se puede hacer uso comercial de la obra y no se puede alterar, transformar o hacer obras derivadas.! ! Esta Tesis Doctoral está depositada en el Repositorio Institucional de la Universidad de Málaga (RIUMA): riuma.uma.es Facultad de Ciencias Programa de Doctorado: Biología Celular y Molecular Thesis by compendium of publications EPIGENETICS OF ADIPOSE TISSUE AND ITS RELATED DISORDERS Daniel Castellano Castillo Directores: Francisco Tinahones María Isabel Queipo Ortuño Isabel Moreno Indias D. Francisco José Tinahones Madueño, Doctor en Medicina y Cirugía, Director de la Unidad de Endocrinología y Nutrición del Hospital Virgen de la Victoria de Málaga y Profesor Titular del Departamento de Medicina y Dermatología de la Facultad de Medicina de la Universidad de Málaga CERTIFICA: Que el trabajo expuesto en la memoria de la Tesis Doctoral desarrollada por Daniel Castellano Castillo con el título “EPIGENETICS OF ADIPOSE TISSUE AND ITS RELATED DISORDERS” corresponde fielmente a los resultados obtenidos. La presente memoria ha sido realizada bajo mi dirección, considerando que tiene el contenido y rigor científico necesario para ser sometida a juicio por el tribunal nombrado por la Universidad de Málaga para optar al grado de Doctor. Y para que conste, en cumplimiento de las disposiciones legalmente vigentes a los efectos oportunos, firmo el presente certificado, con lo que autorizo la lectura de la misma. Director de la tesis: Dr. Francisco José Tinahones Madueño Málaga, Noviembre de 2018 Dña. María Isabel Queipo Ortuño, Doctora en Biología por la Universidad de Málaga CERTIFICA: Que el trabajo expuesto en la memoria de la Tesis Doctoral desarrollada por Daniel Castellano Castillo con el título “EPIGENETICS OF ADIPOSE TISSUE AND ITS RELATED DISORDERS” corresponde fielmente a los resultados obtenidos. La presente memoria ha sido realizada bajo mi dirección, considerando que tiene el contenido y rigor científico necesario para ser sometida a juicio por el tribunal nombrado por la Universidad de Málaga para optar al grado de Doctor. Y para que conste, en cumplimiento de las disposiciones legalmente vigentes a los efectos oportunos, firmo el presente certificado, con lo que autorizo la lectura de la misma. Directora de la tesis: Dra. María Isabel Queipo Ortuño Málaga, Noviembre de 2018 Dña. Isabel Moreno Indias, Doctora por la Universidad de las Palmas de Gran Canaria CERTIFICA: Que el trabajo expuesto en la memoria de la Tesis Doctoral desarrollada por Daniel Castellano Castillo con el título “EPIGENETICS OF ADIPOSE TISSUE AND ITS RELATED DISORDERS” corresponde fielmente a los resultados obtenidos. La presente memoria ha sido realizada bajo mi dirección, considerando que tiene el contenido y rigor científico necesario para ser sometida a juicio por el tribunal nombrado por la Universidad de Málaga para optar al grado de Doctor. Y para que conste, en cumplimiento de las disposiciones legalmente vigentes a los efectos oportunos, firmo el presente certificado, con lo que autorizo la lectura de la misma. Directora de la tesis: Dra. Isabel Moreno Indias Málaga, Noviembre de 2018 Index ABBREVIATIONS!19! INTRODUCTION!33! 1. OBESITY 35! 2. ADIPOSE TISSUE 36! 2.1. Adipose Tissue cellularity: Adipocyte precursors and adipogenesis 38! 2.2. Adipose Tissue cellularity: Immune cells 41! 3. ADIPOSE TISSUE METABOLISM: LIPID AND GLUCOSE METABOLISM 43 3.1. Lipid metabolism 43! 3.2. Glucose metabolism 47! 4. ROLE OF ADIPOSE TISSUE IN METABOLIC DISORDERS 48! 4.1. Metabolic factors related to metabolic disorders 50! 4.1.1. Lipoprotein Lipase (LPL) 50! 4.1.2. Low-density lipoprotein receptor-related protein 1 (LRP1) 53! 4.1.3. Glucose transporter type 4 (GLUT4) 53! 4.1.4. Peroxisome proliferator-activated receptors (PPARs) 54! 4.1.5. Sterol regulatory element binding factors (SREBFs) 55! 4.1.6. Stearoyl-CoA-desaturase (SCD) 56! 4.1.7. Liver X receptor beta (LXRb) 56! 4.1.8. Leptin (LEP) 58! 4.2. Inflammatory factors in metabolic disorders 60! 5. ADIPOSE TISSUE AND COLORECTAL CANCER 64! 6. EPIGENETICS 67! 6.1. DNA methylation 67! 6.2. Histone modifications 71! 6.3. Interplay between DNA methylation and histone modifications 73! 6.4. Interplay between metabolic status and epigenetics 77! Index 6.4.1. Role of lifestyle and nutritional conditions in the epigenetics of obesity and metabolic disease 80! 6.4.2. Epigenetics alterations in colorectal cancer 83! HYPOTHESIS!85! OBJECTIVES!89! RESULTS!93! MANUSCRIPT 1. Adipose Tissue LPL Methylation is Associated with Triglyceride Concentrations in the Metabolic Syndrome 95! MANUSCRIPT 2. Adipose Tissue DNA Methylation of Adipogenic, Lipid Metabolism and Inflammatory Genes in Metabolic Syndrome 107! MANUSCRIPT 3. Complement Factor C3 Methylation and mRNA Expression Is Associated to BMI and Insulin Resistance in Obesity 121! MANUSCRITP 4. Chromatin Immunoprecipitation Improvements for the Processing of Small Frozen Pieces of Adipose Tissue 129! MANUSCRIPT 5. Human Adipose Tissue H3K4me3 in Adipogenic, Lipid and Inflammatory genes are Positively Associated to BMI and HOMA-IR 139! MANUSCRIPT 6. Adipose Tissue Inflammation and VDR Expression and Methylation in Colorectal Cancer 151! GENERAL DISCUSSION!161! CONCLUSIONS!171! LITERATURE!175! SUPPLEMENTAL DATA!207! ! ABBREVIATIONS Abbreviations 21 Body mass index: BMI Tumor necrosis factor: TNF/ TNFα Interleukin 6: IL6 Insulin resistance: IR Cardiovascular disease: CVD Metabolic Syndrome: MetS Triglycerides: TG/Tg High-density lipoprotein cholesterol: HDL-cho/HDL-C Adipose tissue: AT White adipose tissue: WAT Brown adipose tissue: BAT Uncoupling protein: UCP Subcutaneous adipose tissue: SAT Visceral adipose tissue: VAT Weight/hip ration: WHR Stromal vascular fraction: SVF Adipose tissue-derived mesenchymal stem cells: ASCs Scavenger receptor class A member 5: SCARA5 Bone morphogenetic protein 2: BMP2 Bone morphogenetic protein 4: BMP4 Transforming growth factor beta: TGFβ Platelet-derived growth factor receptor A: PDGFRα Platelet-derived growth factor receptor B: PDGFRβ Platelet-derived growth factor subunit A: PDGFA SMAD family member 1: SMAD1 Abbreviations 22 SMAD family member 4: SMAD4 SMAD family member 5: SMAD5 SMAD family member 8: SMAD8 Zinc finger protein 423: Zfp423 Lysyl oxidase: Lox Fibroblast growth factor 2: FGF2 Peroxisome proliferator activated receptor gamma: PPARγ Peroxisome proliferator activated receptor gamma 2: PPARγ2 Peroxisome proliferator activated receptor alpha: PPAR CCAAT enhancer binding protein beta: C/EBPβ CCAAT enhancer binding protein delta: C/EBPδ CCAAT enhancer binding protein alpha: C/EBPα Krüppel-like factor 5: KLF5 Krüppel-like factor 15: KLF15 Krüppel-like factor 2: KLF2 Sterol regulatory element binding transcription factor 1: SREBP1 Sterol regulatory element binding transcription factor 1 isoform c: SREBP1c Sterol regulatory element binding transcription factor 1 isoform a: SREBP1a Sterol regulatory element binding transcription factor 2: SREBP2 Lipopolysaccharides: LPS Interferon-γ: IFN-γ Interleukin 12: IL12 Interleukin 4: IL4 Interleukin 14: IL14 Interleukin 10: IL10 Abbreviations 23 Chitinase-like 3: Ym1 Arginase 1: ARG1 Chylomicrons: CM Apolipoprotein B48: ApoB48 Apolipoprotein CII: ApoCII Apolipoprotein CIII: ApoCIII Lipoprotein lipase: LPL Apolipoprotein E: ApoE Low-density lipoprotein cholesterol: LDL-cho/LDL-C Low density lipoprotein receptor: LDLR Low-density lipoprotein (LDL)-related protein 1: LRP1 Very low-density lipoprotein: VLDL Intermediate density lipoproteins: IDL Hepatic lipoprotein lipase: HLPL Insulin receptor: INSR Leptin: LEP Leptin receptor Ob-Rb: Ob-Rb Free fatty acids: FFA Cluster of differentiation 36: CD36 Fatty acid binding protein 4: FABP4 Acyl-CoA synthase: ACS Glycerol-3 phosphate: glycerol-3P Acetyl-CoA carboxylase: ACC Fatty acid synthase: FAS Adipose TG lipase: ATGL Abbreviations 24 Hormone sensitive lipase: HSL Perilipin 1: PLIN1 Caveolin 1: CAV-1 Janus kinase: JAK Signal transducer and activator of transcription: STAT Low-molecular-weight leptin trimer: LMW Medium-molecular-weight leptin hexamer: MMW High-molecular-weight leptin complex: HMW Phosphoenolpyruvate carboxykinase: PEPCK Glucose-6-phosphatase: G6P AMP-activated protein kinase: AMPK Beta cell: β-cell Waist circumference: WC Blood pressure: BP Systolic blood pressure: SBP Diastolic blood pressure: DBP Non-alcoholic steatohepatitis: NASH Neurological disorders: ND Endoplasmic reticulum: ER Lipase maturation factor 1: LMF1 Sel-1 suppressor of lin-12-like: Sel1L Heparan sulfate proteoglycans: HSPG Glycosylphosphatidylinositol (GPI)-anchored glycoprotein 1: GPIHBP1 Apolipoprotein A5: ApoA5 Angiopoietin-like protein 3: Angptl3 Abbreviations 31 Glycoprotein (transmembrane) nmb: Gpnmb Chromatin inmunoprecipitation: ChIP Homeostatic model assessment of insulin resistance: HOMA-IR Postprandial triglycerides: Post TG Glutamate-Oxaloacetate Transaminase: GOT Glutamate-Pyruvate Transaminase: GPT Gamma Glutamyl Transpeptidase: GGT C-Reactive Protein: CRP Proteinase K: PK E2F transcription factor 1: E2F1 Transcription Start Site: TSS Lean NG: Lean normoglycemic MO NG: Morbid obese normoglycemic MO PD: Morbid obese prediabetic Parathyroid hormone: PT !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!INTRODUCTION Introdution 35 1. OBESITY Nowadays, obesity has become one of the greatest causes for health disorders in developed countries. More than a third of the world population suffers of overweight or obesity, being expected to rise up to 38% of overweight and 20% of obesity by 2030, reaching the 85% of overweight/obese people in the USA (1). Obesity is defined as an excess of body weight, which underlines with an accumulation of adiposity (adipose tissue). Obesity is usually interpreted by the Body mass index (BMI), measured as a ratio weight/height2 (2). Though, subjects are mostly classified according to this index as lean or normal-weight subjects (when BMI<25), overweight subjects (when BMI≥25 and <30), obese subjects (BMI≥30 and <40) and morbid obese subjects (BMI≥40) (3,4). The excess of adiposity is usually accompanied by a series of metabolic disturbances and a higher risk of suffering certain diseases (1). Obesity increases the risk of being diabetic between 4.9% and 9%, and inversely 60% to 90% of all diabetics are or have been obese (5). In fact, although bariatric surgery has as first consequence the weight loss, is also one of the most effective treatment to improve diabetes and metabolic disease, demonstrating the close relationship between BMI and diabetes (6). Although the connection between both, obesity and diabetes, is not totally clear, mechanisms in which are involved pro-inflammatory cytokines (tumor necrosis factor (TNF), interleukin 6 (IL6)), insulin resistance (IR), deranged fatty acid metabolism and cellular processes such as reticulum stress and mitochondrial dysfunction have been shown to be involved (7). Obesity has also been widely associated to cardiovascular disease (CVD) (8). It has been shown that an increase in a unit of BMI is associated to a 4% increase in the risk of ischemic stroke and 6% increase in hemorrhagic stroke (9). Increase in BMI was also Introdution 36 associated to an increase in blood pressure, which was traduced to a 12% increase in coronary heart disease and 24% increase risk for stroke (9). Other study showed an increase in subclinical cardiovascular disease associated to a rise in BMI compared to lean subjects (10). Obesity was positively associated with a 1.17-1.28 relative risk increase in coronary artery calcium, 1.45 relative risk increase for common carotid artery intimal medial thickness, 1.32 relative risk increase for internal carotid artery intimal medial thickness and 2.69 relative risk increase in left ventricular mass (10). Atherosclerosis has also been associated to obesity in a process in which adipose tissue, low-grade inflammation, oxidative stress, impaired autophagy and gut microbiota among other factors could be related (11,12). It is clear that cancer is one of the greatest health concerns in modern societies. Obesity has been strongly associated to several cancers, and what is more, has been proposed as the second biggest cause of cancer in the world only surpassed by smoking (13). Thus, the estimation is that over 20% of all cancers are caused for excess body weight, rising to 50% in postmenopausal women (14). Moreover, obesity also affects death rates in cancer, up to 20% of death in women and 14% of death in men that can be attributable to obesity (14). Finally, obesity has traditionally been associated to what is known as metabolic syndrome (MetS). MetS is defined by a cluster of metabolic alterations which altogether increase the risk of suffering diabetes, cardiovascular disease and cancer (15,16). 2. ADIPOSE TISSUE There are different adipose tissue (AT) depots with different metabolic roles (17). White AT (WAT) is composed by unilocular cells with a high capacity for lipid storage in the form of triglycerides (TG), since is the tissue in charge of accumulating the Introdution 37 surplus of energy in the organism. On the other hand, Brown AT (BAT) is made out of multilocular adipocytes with a high mitochondrial content that can consume high levels of energy when are stimulated by adrenergic signals or cold; the energy is used in heat production, a process that is carried out by the mitochondrial uncoupling protein (UCP) (17). There are two types of WAT, which are mainly defined by their depot location: the subcutaneous AT (SAT) and the visceral AT (VAT) (18) (Figure 1). Figure 1. Patterns of fat distribution and its relationship with BMI, weight/hip circumference and risk of suffering disease. Pear shape (2) is associated to SAT and is usually refers as woman fat distribution. Apple shape (3) is associated to VAT and is most common to men fat distribution. Risk of associated chronic disease such as cardiovascular disease or diabetes is display by the red line intensity from the lowest risk for lean subjects (1) to the greatest risk characteristic of apple fat distribution (3). Abbreviations: Subcutaneous Introdution 38 adipose tissue (SAT); visceral adipose tissue (VAT); Body mass index (BMI); weight/hip ratio (WHR). Adapted from (19). Both SAT and VAT not only differ by their location, but also by their molecular profile, their metabolism and their implication in the etiology of the metabolic disease (18,20– 22). Also the embryonic origin of both fat depots seems to be different. While VAT adipocytes are thought to have a mesodermal origin, SAT adipocytes come from different embryonic origins depending on the exact depots (mesoderm, neuroectoderm), and in some cases still undetermined what their origin are (17). Besides, while expansion of SAT has been long considered as a protective factor against metabolic disorders, fat accumulation in VAT (characteristic of central obesity or high wait/hip ratio) has been deemed to prompt metabolic disorders (19,23). However, there is an ongoing discussion about which tissue is first damaged and which are their roles in the generation of these diseases (20–22,24). What is undoubtedly is that AT has emerged as an important mediator of the whole body homeostasis, and its role as an endocrine organ has focused great interests. 2.1. Adipose Tissue cellularity: Adipocyte precursors and adipogenesis AT is composed by a myriad of cell types in which adipocytes, the predominant cell, are in charge of lipid accumulation. Moreover, in its stromal vascular fraction (SVF) there are macrophages (which modulate adipose tissue function), lymphocytes, vascular cells, pericytes, adipose tissue-derived mesenchymal stem cells (ASCs) (with autorenew properties and can differentiate to pre-adipocytes) all of them known as stromal vascular cells (SVC). The adipocyte cell turnover is carried out for ASCs, although the contribution of pericytes, endothelial cells and non-resident progenitors with Introdution 39 hematopoietic origin has been proposed to contribute at different degree depending on the specific fat depot and the metabolic condition of the subject (17). Thus, a proper balance among all these type of cells is needed for a correct function of AT. Adipogenesis, the mechanism in which new adipose tissue is generated, assures the AT turnover and hyperplasia (AT expansion by cell number increase). ASCs have been shown to be able to differentiate to a wide range of cell types, including adipocytes, chondrocytes, osteocytes and miocytes. Although pre-adipocytes differentiation to adipocytes is a well-known process much less is known about ASCs commitment to pre-adipocytes. Basically, the differentiation from ASC to full differentiated adipocyte requires of four stages: 1) Commitment of ASCs to pre-adipocytes; 2) clonal expansion of pre-adipocytes; 3) growth arrest; 4) terminal differentiation to adipocytes (25). The commitment of ASC to pre-adipocyte lineage implies several factors as Scavenger receptor class A member 5 (SCARA5), Bone morphogenetic protein 2 (BMP2), Bone morphogenetic protein 4 (BMP4), Transforming growth factor beta (TGFβ) signaling, Platelet-derived growth factor receptor A (PDGFRα), Platelet-derived growth factor receptor B (PDGFRβ), Platelet-derived growth factor subunit A (PDGFA), SMAD family member 1 (SMAD1), SMAD family member 4 (SMAD4), SMAD family member 5 (SMAD5), SMAD family member 8 (SMAD8), Zinc finger protein 423 (Zfp423), Lysyl oxidase (Lox) or Fibroblast growth factor 2 (FGF2) among others (25– 27). TGFβ signaling inhibits pre-adipocyte commitment, BMP (from the TGFβ protein family) exerts generally a positive effect over ASC to pre-adipocyte commitment, and the PDGF has been proposed to have both, adipogenic and anti-adipogenic effects. BMP2 and BMP4 have been demonstrated to stimulate the heterodimer Smad1/Smad4 (process that Zfp423 is thought to regulate) and the expression of Peroxisome proliferator activated receptor gamma 2 (PPARγ2) and Lox that are necessary to pre- Introdution 40 adipocyte commitment. Another pro-adipogenic effect different from the BMP signaling is the pro-adipogenic effect that FGFs provokes over ASCs. Thus, FGF2 has been shown to stimulate PPARγ2 overexpression and exogenous addition of FGF2 in SCV stimulates WAT formation (26,27). Much more is known about pre-adipocyte differentiation to mature adipocyte. The adipogenesis pathway is controlled by several transcription factors, with PPARG as central factor in this process. Besides, several CCAAT enhancer binding protein (C/EBP) factors are involved at different stages of adipogenesis. First, an overexpression of C/EBPβ and C/EBPδ takes place, stimulating the expression of C/EBPα and PPARγ. In turn, both PPARγ and C/EBPα can stimulate each other in a regulatory loop. Eventually, PPARγ and C/EBPα activate the expression of genes related to adipocyte metabolism, producing final adipocyte maturation (Figure 2). Other factors have been also shown to be important in adipocyte differentiation. For instance, several genes from the Krüppel-like factors (KLFs) family are involved in the adipogenic pathway. KLF5 and KLF15 have been observed to stimulate PPARγ expression in early and terminal differentiation stages, respectively. Contrary, KFL2 has been shown to have an anti-adipogenic effect by inhibiting PPARγ expression (28,29). Other factor that has been proposed to play a role in adipocyte differentiation is the Sterol regulatory element binding transcription factor 1 (SREBP1), especially the Sterol regulatory element binding transcription factor 1 isoform c (SREBP1c), which is able to activate adipogenesis via PPARγ stimulation, although mice with overespression of nuclear SREBP1c has been shown to suffer for lipodystrophy (28). Introdution 47 3.2. Glucose metabolism AT not only plays a role in lipid metabolism but the closed interplay between both, lipid and glucose regulation, gives AT an important weight in glucose homeostasis. This is evident since AT-specific KO-mouse of GLUT4 has been shown to provoke a failure in systemic glucose homeostasis leading to insulin resistance. It has been shown that GLUT4 is down-regulated during fasting state and up-regulated during postprandial state. In AT, glucose and specifically the intermediary metabolite in glycolysis glycerol3P, is necessary to the re-esterification of FFA to TG. When glucose is limited (as in fasting states) adipocytes rely on glyceroneogenesis in order to obtain the necessary source of glycerol-3P (48). Moreover, some adipokines secreted by AT can exert a regulatory effect over glucose homeostasis (48). Thus, LEP has been shown to directly inhibit insulin gene (ISN) expression in β-cells via the activation of JAK/STAT signaling. LEP has also been demonstrated to inhibit insulin secretion and to regulate β-cell mass in pancreas (49) (Figure 5). Figure 5. Schematic representation of the direct effects exerted by LEP over glucose metabolism. LEP inhibits insulin by down-regulation of pro-insulin expression, inhibition of insulin secretion in β-cell and by controlling β-cell mass. Introdution 48 Adiponectin is another molecule secreted by AT, with an active role in glucose homeostasis. This molecule is secreted in three isoforms: low-molecular-weight (LMW) trimers, medium-molecular-weight (MMW) hexamers and high-molecular-weight (HMW) complexes. Out of the three isoforms, the HMW has been described as the most active form. Adiponectin enhances hepatic insulin sensitivity and avoid gluconeogenesis by down-regulating both phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6P). The increased hepatic sensitivity to adiponectin is thought to occur via either receptor-mediated activation of AMPK pathway or lowering hepatic ceramide levels (50,51). Apart from the effect on peripheral tissues, adiponectin affects β-cells. Thus, adiponectin can prevent β-cells from apoptosis, has a protective role against lipid cytotoxicity and it is thought to stimulate insulin secretion during challenged states (50). 4. ROLE OF ADIPOSE TISSUE IN METABOLIC DISORDERS Since obesity is characterized by an increase of AT mass, the premise that adipose tissue may participate in the etiology and generation of metabolic disorders usually associated to obesity came out long time ago. It is known that AT can deliver a wide range of molecules, as cytokines, prostaglandins, adipokines or free fatty acids, which can exert their activity in other tissues, which in cases can lead to disease (52–55). Therefore, AT has gained a central role in explaining metabolic disturbances and has been related to the appearance and development of diseases like dyslipidemia, hypertension, proinflammatory states, insulin resistance, MetS, diabetes, CVD, stroke, or cancer (37,56–59). Although AT abnormalities has been usually associated to obesity, metabolic disturbances also account in normal-weight subjects (37), which agrees with the new Introdution 49 harmonization criteria to define MetS, in which central obesity is no longer a mandatory factor (15). MetS is defined by a cluster of metabolic alterations, which altogether increases the risk of suffering diabetes, cardiovascular disease and cancer (15,16) (Figure 6). Genetic and lifestyle factors have been shown to be important for the development and etiology of MetS (60). MetS represents a serious problem in developed countries, with a high prevalence (30-40% by the age of 65 years) and both, its prevalence and incidence are increasing (61). The appearance of MetS has been mainly associated with lifestyle features like physical inactivity, smoking, alcohol intake and diet. However, genetic and epigenetic factors are now emerging as factors of paramount importance in its pathophysiology (62,63). Figure 6. Metabolic syndrome variables and MetS associated disorders. MetS is composed by at least the alteration in 3 of the following parameters: HDL, TG, glucose, blood pressure and WC. Some of the parameters are sex dependent (as HDL or WC) and raze dependent (as WC, the showed values are for west countries). The MetS parameters are represented according Introdution 50 to the established in (15). Abbreviations: Metabolic syndrome (MetS); waist circumference (WC); blood pressure (BP); systolic blood pressure (SBP); diastolic blood pressure (DBP); high-density lipoprotein (HDL); triglycerides (TG); Cardiovascular diseases (CVD); Nonalcoholic steatohepatitis (NASH); Neurological disorders (ND). 4.1. Metabolic factors related to metabolic disorders 4.1.1. Lipoprotein Lipase (LPL) LPL plays a key role in lipid metabolism (64) by hydrolyzing triglyceride-rich lipoprotein (CM or VLDL) to free fatty acids, that can be then incorporated into the AT for storage or energy utilization in other tissues (64,65) (Figure 7). LPL is synthetized and secreted to the endoplasmic reticulum (ER) where is folded with the help of Lipase maturation factor 1 (LMF1). LPL/LMF1 form a complex with Sel-1 suppressor of lin12-like (Sel1L) and this LPL/LMF1/Sel1L complex helps to stabilize LPL homo-dimers, the active form of LPL, and allows LPL exits the ER (65). After being secreted, LPL is bound to Heparan sulfate proteoglycans (HSPG) at the surface of the cell. HSPG not only serves as an anchorage molecule but also contribute to LPL translocation to the endothelial surface and acts as cofactor for LPL activity. Another factor, the GPIHBP1 (which is a Glycosylphosphatidylinositol (GPI)-anchored glycoprotein belonging to the lymphocyte antigen 6 family) has been shown to play an important role in LPL transportation from the interstitial space to the luminal space at the endothelial surface. Lack of this factor has been observed to impaired translocation of LPL to the endothelial surface provoking hypertriglyceridemia. Besides, it has been shown that GPIHBP1 is able to act over LPL activity by keeping the catalytic domain of the enzyme unfolded (65). Introdution 51 Figure 7. Schematic overview of LPL secretion and anchorage to EC surfaces. LPL forms a dimer in the endoplasmic reticulum and is then secreted. LPL is bound to HSPG in the surface of the adipocyte and then translocated to GPIHBP1. LPL hydrolyze TG from Triglyceride-rich lipoproteins as CM and VLDL producing FFA that are transported to the adipocyte where are re-esterified giving TG to be stored in LD. Abbreviations: Endothelial cells (EC); heparan sulfate proteoglycans (HSPG); triglycerides (TG); chylomicrons (CM); very-low density lipoproteins (VLDL); free fatty acid (FFA); lipid droplets (LD). Given the importance of LPL in lipid homeostasis the enzyme undergoes a tight regulation at all levels. Thus, some factors have been postulated to control LPL activity like ApoCI/II/III, ApoA5, angiopoietins and hormones. ApoCII, which is carried in CM, has been shown to be a stimulator of LPL activity, while ApoCI/III are inhibitory factors. ApoA5 that is present in VLDL, HDL and CM to a less extend has been described to stimulate LPL activity, although the effect seems to be weaker than the positive action carries out by ApoCII. Angiopoietin-like protein 3 (Angptl3) has been described to inhibit LPL activity as well as by enhancing its cleavage by pro-protein Introdution 52 convertases. Angptl4 is produced in the liver and AT and has been shown to increase with fasting, causing an inhibitory effect over LPL. It is thought that inhibits LPL dimerization and reduces LPL affinity to GPIHBP1. Another member of angiopoietins, Angptl8 has also been described to inhibit LPL activity (65). It has been observed that insulin and glucose can regulate LPL. Thus, insulin leads to a raise of LPL mRNA levels, while glucose can activate LPL activity by glycosylation in its catalytic site (66). Other hormones or derivative hormones like Prostaglandin E2 (PGE2), 7-βhydroxycholesterol and 25-hydroxycholesterol has been shown to exert an inhibitory effect over LPL production in macrophage cultures. Moreover, LPL gene expression has been shown to be regulated by miRNAs such miR-27 and miR-29 in AT (65). Deregulation of any of these factors can lead to de-regulation of LPL expression or its activity, which in turn might promote lipid metabolic failure. By contrast, the understanding of the regulation of LPL can also give clues to new treatment approaches based on the inhibition or stimulation of these factors. For instance, some molecules such as statins, fibrates, nicotinic acid, Ezetimibe or Orlistat have been used to treat hypertriglyceridemia by mean of their stimulatory effect over ApoCII, while other treatment as fibrates have been described to inhibit ApoCIII, therefore provoking LPL activity enhancement (65). Indeed, AT is the main TG storage tissue. LPL mRNA increases during the adipocyte differentiation, being one of the earliest markers defined in the preadipocyte lineage (67). On the other hand, AT LPL mRNA expression has been negatively associated to the BMI (58). As we stated before, LPL activity in AT has been related with hypertriglyceridemia (68). Thus, lower activity of LPL can result in greater levels of plasma TG, which in turn can be accumulated in other tissues causing insulin-resistance (69). Introdution 53 4.1.2. Low-density lipoprotein receptor-related protein 1 (LRP1) LRP1 is a member of the LDL receptor family that regulates lipid and glucose metabolism in the liver and AT (70). LRP1 is involved in CM-remnants, insulin receptor trafficking and regulation, and glucose metabolism, therefore being related to atherosclerosis (it is an atheroprotective factor) and diabetes (71). LRP1 translocation to the cell surface after glucose and insulin stimulation has been described in adipocytes. Moreover, this LRP1 induction by insulin and glucose is accompanied by an increase in the uptake of CM-remnants by adipocytes (71). In fact, it has been demonstrated that LRP1 are contained at GLUT4 rich vesicles and is delivered after insulin stimulation altogether. It has been shown that adipocyte-specific LRP1-knockout mice provokes a decrease of GLUT4, Insulin-regulated aminopeptidase (IRAP) and sortilin expressions, and that this three factors together with LRP1 form a complex that is involved in GLUT4 trafficking. Thus, LRP1-depleted 3T3-L1 adipocytes presents up to 50% decrease in glucose uptake, demonstrating the importance of LRP1 in the etiology of insulin resistance (71). Adipose-specific LRP1-knockout in mice has also been related to AT dysfunction (with the stimulation of CD68, MCP1/CCL2, IL6 and TNF) promoting atherosclerosis (72). VAT LRP1 mRNA has been described to be overexpressed both, in mice fed with high fat diet (HDF) compared to normal diet and in obese people compared to normal-weight subjects (73). 4.1.3. Glucose transporter type 4 (GLUT4) GLUT4 in AT is the rate-controlling step in insulin-mediated glucose disposal, being its levels diminished in insulin resistance state and obesity. Glucose is necessary as a reliable source of glycerol-3P (a glycolytic intermediary) that is used to re-esterified Introdution 54 Free fatty acid (FFA) taken from the blood stream to form TG (48). A down-regulation of GLUT4 could impair this source of glycerol-3P for what glyceroneogenesis emerges as an important pathway for TG formation. In this pathway, lactate or pyruvate enters the Tricarboxylic acid cycle (TCA cycle) that produces oxaloacetate, which is then transformed to phosphoenolpyruvate by the cytosolic enzyme Phosphoenolpyruvate carboxykinase (PEPCK). Phosphoenolpyruvate is then converted to glycerol-3P by reverse glycolysis. A huge intake of FFA in subjects with insulin-resistance could overwhelmed this pathway, hindering FFA esterification and leading to FFA accumulation in peripheral tissues contributing to insulin resistance worsening. Thiazolidinediones (TZDs) is a potent insulin sensitizing used to treat diabetes. This enzyme can up-regulate glycerol kinase and PEPCK activity, triggering in an improvement in FFA removal by increasing FFA esterification in TG and resulting in an increase of insulin sensitivity (48). 4.1.4. Peroxisome proliferator-activated receptors (PPARs) PPARγ is a key factor for adipocyte differentiation. Adipose-specific PPARγ knockout mice, has been carried out. In this model, there was an impaired AT expansion, presented adipocyte hypertrophy, and higher levels of plasmatic FFA. Besides, these mice were more sensitive to insulin resistance and liver steatosis induced by HFD. When these mice were treated with TZDs (anti-diabetic drug that has PPARγ as a target) liver insulin resistance remitted although no other effects were observed (74). Other member of PPAR receptor family, PPARα, is a target for the fibrates, molecules that are used for the treatment of hypertriglyceridemia. PPARα is a transcription factor that is involved in fatty acid oxidation in tissues with a high level of fat oxidation and peroxisomal metabolism (liver, heart muscle or brown adipose tissue). In AT, PPARα Introdution 55 has been shown to attenuate adiposity and to stimulate the secretion of adiponectin, which is an insulin-sensitizing hormone (75). Thus, 3T3L1 pre-adipocyte line and mice treated with PPARα agonists have been shown to induce adipogenesis, and to increase fat oxidation by direct induction of PPARα over adipogenic and fat oxidation genes (75). Both members of the PPAR family, PPARγ and PPARα carry out their function as heterodimer with the Retinoid X receptor alpha (RXRα) (76). Thus, it has been described that a adipocyte-specific RXRα-knockout mice can cause adipogenesis failure, and cannot increment AT even after HFD administration (77,78). 4.1.5. Sterol regulatory element binding factors (SREBFs) SREBFs are transcription factors known to control cholesterol and fatty acid biosynthesis, as well as adipogenesis in AT. SREBF1 has two isoforms, the SREBF1c (the most expressed in AT with a key role in adipogenesis) and SREBF1a. These isoforms are determined by the transcription starting site (79). Although the different isoforms of SREBF present some functional overlap, SREBF1c has a more prevalent action in lipid biosynthesis while SREBF2 has a prevalent role in cholesterol biosynthesis. By contrast, SREBF1a has a role in both, fatty acid and cholesterol biosynthesis (80–82). SREBF1c and SREBF2 have lower gene expression levels in VAT of obese people respect to lean people. It has already been shown that this expression is reestablished to lean-like values after bariatric surgery in obese people (83). Furthermore, SREB1c expression in AT has been demonstrated to respond to food and caloric intake. Thus, fed can promote SREBF1c down-regulation while fasting and caloric restriction have been shown to stimulate AT SREBF1 expression (79). SREBF1 overexpression is responsible of the oxidative stress improvement observed in caloric restriction diets (79). In the other way around, oxidative stress is capable of inhibiting Introdution 56 healthy AT expansion through the suppression of the SREBF1c-mediated lipogenic pathway (84). In animal models, it has been demonstrated that SREBF2 protein in adipocyte hypertrophy is activated, which leads to an increase in the production of chemerin, and adipokine positively related to MetS (85). 4.1.6. Stearoyl-CoA-desaturase (SCD) SCD is a key enzyme that catalyzes the rate-limiting step to monounsaturated fatty acids (MUFFAs) from saturated fatty acids, specially from stearoyl-CoA and palmitoylCoA, giving as a result oleate and palmitoleate (86,87). It is long known that polyunsaturated free fatty acids (PUFFAs) exerts an inhibitory effect over SCD gene expression (87). It has been proven in mice, that a lack of SCD lowers adiposity by the increase in the metabolic rate, the thermogenesis and the β-oxidation, and a decrease in the lipogenesis pathway (87). 4.1.7. Liver X receptor beta (LXRb) LXRb is a transcription factor activated by ligand, which belongs to the nuclear receptor family and that is involved in gluconeogenesis, cholesterol and inflammation (88). LXRb is activated upon cholesterol breakdown products oxysterols and then dimerize with its partner RXRα to carry out its function (89,90). This biological function is mainly related to the lipogenesis pathway, being for instance Fatty acid synthase (FAS) and Acetyl-CoA carboxylase (ACC) some of the target genes implied. Other target genes for LXR involve the cholesterol transporters ATP binding cassette transporters A (ABCA) and G (ABCG), the SREBF1c transcription factor, and the key rate-limiting enzyme for the biosynthesis of monounsaturated fatty acids, the SCD (89). Thus, LXR activation has been shown to have beneficial effects in metabolism by: 1) increasing Introdution 63 inflammation indirectly via lipotoxicity, but a direct modulation of insulin pathway in the adipocyte has been described. This effect over insulin signaling implies the inhibition of the kinase activity by the insulin receptor (IR), via the inactivation of the Insulin receptor substrate 1 (IRS1) and by the destabilization of the interaction IR/caveolin 1 (CAV1) (109) (Figure 11). Figure 11. TNF signaling on AT. TNF interacts with its receptor TNFR1 which can induce apoptosis, ceramide production, lypolisis, ER oxidative stress, mitochondrial dysfunction and altered adipokine profile, factors that can eventually lead to inhibition of insulin signaling and therefore to insulin resistance. TNF might also act through TNFR2 which could also provoke an induction of apoptosis and a decrease to insulin sensitivity by down-regulation of GLUT4 and IRS-2 (47). Introdution 64 Since the low-grade inflammation has been associated to metabolic disease, treatments against TNF and/or its pathway could be suitable to treat disorders associated with this chronic inflammation, impaired glucose tolerance and dyslipidemia (47). It has also shown that TNF correlates positively with the size of the adipocytes, which in turn has been related to the generation of metabolic disorders (109,110). Given the direct and indirect action of TNF over the biology of AT and the lipid and glucose metabolism, it is not surprising the relevant role this factor could have in the etiology of the MetS and its complications (111). 5. ADIPOSE TISSUE AND COLORECTAL CANCER AT can deliver molecules and promote metabolic states, which in turn can affect other tissues and promote disease. In this sense, the possible role of AT in the development of some kind of cancers has gained interest. As it is described previously, obesity and AT dysfunction has been related to a chronic low-grade inflammation (112). In connection with this, chronic inflammation is well known to be a risk factor of developing cancer including colorectal cancer (CRC) (113,114). In this manner, several studies have noted the relationship between CRC and low grade inflammation (115). CRC subjects present a dysfunctional AT, which might be a key contributor to the inflammatory state through the secretion of several proinflammatory factors such as TNF, IL-6, and NFκB pathway (116). Inflammatory processes could not be the only relationship between AT and CRC. Metabolic deterioration has been also described to increase the cancer incidence, including CRC (117–120). Thus, the relationship between inflammation and metabolic deregulation has been described. For instance, NfκB pro-inflammatory pathway has been described to be a promoter of inflammation in AT, leading to metabolic disorders Introdution 65 (121). In this line, strategies based on the disruption of NfκB action has been proposed to ameliorate diabetes, hyperglycemia and insulin resistance (121). Given this relationship between inflammation and CRC, and the capacity of AT to generate low-grade inflammation, it is of interest the study of the possible mechanisms that lead to this inflammation in AT. Besides, it would be of great interest the search of therapeutic approximations to avoid this pro-inflammatory state in AT. In this sense, metformin, one of the earliest drugs used to treat diabetes, has been described to inhibit inflammation in a process in which NfκB is involved. Metformin has been shown to ameliorate adipose tissue inflammation, and to promote macrophage polarization to the anti-inflammatory M2 phenotype (122). In addition, a lower risk of suffering cancer disease has been described in subjects who undergo metformin treatment (123,124). A potent anti-inflammatory molecule is vitamin D (VD) (125–127). VD was first identified by its role in stimulating intestinal calcium intake and bone mineralization, being low levels of this vitamin related to decreased bone mineral density and osteoporosis (128). VD is synthetized from 7-dehydrocholesterol in the skin in a reaction carried out by UVB light, given as a result the pre-vitamin D, which is further converted by heat to VD. This form is not active yet, and VD is turned into 25hydroxyvitamin D (25(OH)D) by CYP27A1, hydroxylation that occurs mainly in the liver. 25(OH)D is the major form of serum vitamin D and its levels have been observed to be a good indicator of VD status. This form is then further converter in the kidney to the active form, 1,25-dihydroxyvitamin D (1,25(OH)2D) by CYP27B1. Kidney CYP27B1 expression is stimulated by parathyroid hormone (PTH) and down-regulated by FGF23 and its own product 1,25(OH)2D (129). 1,25(OH)2D stimulates its own degradation by the 24-hydroxylase CYP24A1, which catalyzes the conversion of Introdution 66 25(OH)D and 1,25(OH)2D to calcitroic acid and other inactive metabolites (129,130) (Figure 12). Figure 12. Vitamin D metabolism from pre-vitamin D to the active form 1,25-dihydroxy vitamin D. VD receptor (VDR) has been described to be expressed in a wide range of tissues and cell types, and it is though that the non-classical actions of VD (such as inhibition of proliferation, macrophage modulation, terminal differentiation stimulation or insulin production stimulation) are exerted via its interaction with it (128). Lack of VD, specifically the major plasma form of VD, the 25(OH)D has been related to an increase in the development of CRC (131). AT can express proteins related to VD metabolism (132), and it has been proposed that it can act as VD storage tissue (133). It has been shown that the active form of VD, 1,25(OH)2D3, is able to modify adipocyte and AT physiology via VDR action (134,135), decreasing the expression of pro-inflammatory Introdution 67 cytokines in AT (136). Therefore, VD could be other factor that could be involved in the regulation of AT inflammation and could contribute to the biology of CRC. 6. EPIGENETICS Epigenetics concerns the information conveyed through cell division and that not implies changes at the DNA sequence. Epigenetics shape differentiation processes, and it is the roof of the differences observed between cellular types or organs (137). There are basically two epigenetic regulation landscapes: DNA methylation, which occurs at cytosines adjacent to guanines (CpG); and histone modifications, which is more variable and diverse than DNA methylation at CpG (137). 6.1. DNA methylation DNA methylation is a process in which a covalent methyl-group is added to the carbon 5 of a cytosine-pyrimidine ring (5mC) in a CpG nucleotide. Most of the DNA methylations are spread in transposons and mobile DNA sequences (SINE, LINE, etc), while DNA methylation at the promoter or first exon of genes represents a small percentage of whole DNA methylation in the genome (138). It is thought that DNA methylation was first selected evolutionary as a mechanism to stop the replication of mobile sequences in the genome (138). Un-methylated CpG can accumulate in the promoter of genes, elements that are called CpG islands. Un-methylated CpG island would assure that transcription factors and the transcription machinery are bound to the right place in the promoter (138). Around 75% of all genes have CpG island in their promoter, being susceptible of DNA methylation control (138). While DNA methylation at CpG islands in the promoter of genes are related to gene repression, DNA methylation in the body of the gene (which are usually hyper-methylated) has Introdution 68 been reported to activate gene expression, maybe by increasing the efficiency of the transcription (139–141). DNA methylation is catalyzed by DNA methyl-transferases (DNMTs). These enzymes catalyze the conversion of un-methylated to methylated CpG in a process in which Sadenosylmethionine (SAM) acts as methyl-donor. In humans, there are several DNMTs: DNMT1, DNMT2, DNMT3A, DNMT3B and DNMT3L. From these, DNMT1, DNMT3A and DNMT3B have DNA-methyltransferase activity while DNMT3L and DNMT2 do not conserve this capacity even though can either act as cofactors for the other DNMTs (in the case of DNMT3L) or methylate tRNA (for DNMT2) (142). DNMT3A and DNMT3B have been described as methyltransferases “de novo”, which are in charge of establishing new DNA methylated marks ubiquitously and in an un-specific way. Once the mark is established, DNMT1 has been shown to maintain the DNA pattern trough the replicative cell cycles by recognizing hemi-methylated DNA (142,143). Contrary to DNMT1 and DNMT3A/B, DNMT2 has been demonstrated to methylate tRNA instead of DNA, modification that has been proposed to avoid tRNA fragmentation and regulated protein transduction. Even though is a tRNA modifier, the enzyme evolved from DNMTs that with subtle changes finally acquired this new function (142) (Figure 13). Introdution 69 Figure 13. DNA methylation reaction. DNMTs use as substrates a cytosine near a guanine and S-adenosylmethionine (SAM). In the reaction, the methyl group is transferred to the carbon 5 of the cytosine ring, producing 5-methyl-cytosine. As a result S-adenosylhomocysteine (SAH) is also produced that can be recycle to form SAM. Abbreviations: Cytosine (C); DNAmethyltransferases (DNMTs); 5-methyl-cytosine (5mC). DNA methylation is a dynamic process and can be modified according to environmental, genetics and stochastic factors. Therefore, a certain CpG can be methylated and unmethylated (144). DNA de-methylation can occurs both, by enzymatic action or by passive de-methylation. Passive de-methylation would take place in successive replications with a low activity of the methylation machinery, though producing a dilution of 5-methyl-cytosine (5mC) (formed by the transfers of the methyl group to the carbon 5 of the cytosine ring) giving as a result un-methylated DNA (145,146). The active de-methylation process implies the action of Ten-eleven translocation (TET) member family. These enzymes have been reported to catalyze the oxidation of 5mC to 5-hydroximethylcytosine (5hmC) in a process in which molecular oxygen and αketoglutarate (that is converted to succinate) are necessary. In turn, 5hmC can be further oxidized by TET to 5-formylcitosine (5fC), and 5fC to 5-carboxylcitosine (5caC), again Introdution 70 with the participation of oxygen and α-ketoglutarate (145,146). 5hmC has been related to active genes and is thought to contribute to the cell-specific set of expressed genes observed among cellular types, although the exact mechanism by which exerts this regulation is not fully understood (147–149). The restoration of these oxidized forms to cytosine can be carried out in a passive way as well, with dilution during rounds of cell replication, or in an active way. In the active way, Thymine DNA glycosylase (TDG) has been shown to excise 5caC and 5fC, after what the cytosine is restored by the base excision repair (BER) pathway (145,146) (Figure 14). Therefore, unlike the methylation process, the DNA de-methylation mechanism is a complex pathway where more studies would be necessary to expand the knowledge and weigh the importance of the different factors involved. Introdution 71 Figure 14. DNA de-methylation process. After methylation of cytosine by DNMTs, TET successively oxidizes 5mC to 5hmC, 5fC and 5caC. 5hC, 5fC and 5caC can be diluted to cytosine during the following rounds of replication (broken arrows) in what is known as passive de-methylation. Additionally, 5fC and 5caC can be excised by TDG after what the cytosine can be restored by the BER pathway. Abbreviations: Cytosine (C); 5-methylcytosine (5mC); 5hydroxymethylcytosine (5hmC); 5-formylcytosine (5fC); 5-carboxylcytosine (5caC); DNAmethyl transferases (DNMTs); S-adenosylmethionine (SAM); S-adenosylhomocysteine (SAH); ten-eleven translocation enzymes (TET); α-ketoglutarate (α-KG); thymine DNA glycosylase (TDG); base excision repair pathway (BER). 6.2. Histone modifications DNA in the nucleus is associated to histones in what is called the nucleosome. A pair of each H2A, H2B, H3 and H4 histone types forms the nucleosome (Figure 15). These histones are susceptible of post-translational modifications that can modified the strength of the DNA union with the complex, attract other regulatory factors and in turn regulate chromatin structure, gene activity and DNA repair. For instance, histone acetylation of lysine residues produces de blockage of positive charges, resulting in a weaker interaction with the DNA and making the DNA more accessible to other regulatory factors (polymerase, transcription factors, etc.) (150–152). While DNA modifications are basically reduced to 5mC, histones can be modified at the same degree as any other proteins, modifications that mainly take place in the Nterminal tail. That means, there is a wide range of possible modifications such as methylation, phosphorylation, sumoylation, ubiquitination, glycosylation, acetylation, propionylation, butyrylation, crotonylation or citrullination. Lysines are the residue more often modified, although modifications in other amino acids like arginine, serine or threonine has been reported. These modifications have been described to be carried Introdution 72 out for a series of histone modifier such as histone acetyltransferases (HATs), histone mehtyltransferases (HMTs) or phosphatases among others. In turn, these marks have been shown to be removed by proteins called erasers, for instance histone deacetylases (HDACs) (that remove acetyl groups) or histone demethylases (HDMs) (which remove methyl groups). Different from DNA methylation that is usually linked to gene repression (although as it is discussed there are some exceptions), histone modifications can be involved in gene repression, gene activation, DNA repair, chromatin structure etc. Besides, the role of some modifications are difficult to understand since many of these modifications occurs at the same histone and myriads of possible combinations makes histone regulatory landscape a complex field of study (150,151). Figure 15. Nucleosome and histone modifications. Nucleosome is composed by a pair of each H2A, H2B, H3 and H4 histone types (1), where DNA is wrapped around the structure. Additionally, histones can suffer post-translational modifications in their aminoacidic chain. (2) Summarization of the most common N-terminal modifications at H2A, H2B, H3 and H4 histone types. Adaptation from (153). Even though its complexity, some histone modifications have been largely studied. Thus, di-methylation at lysine (K) 4 and tri-methylation at K4, K36 and K79 of H3 have been associated to transcription activation. On the contrary, Tri-methylation at K9 and Introdution 79 permissive chromatin, which gives to a raise of gene expression for genes related to growth or proliferation (152,176). By contrast, sirtuins (class III HDAC) has been shown to perform histone de-acetylation in a process in which oxidized nicotinamide adenine dinucleotide (NAD+) is required. NAD+ is a sensor of the metabolic status and it is implied in several oxidative pathways such us glycolysis, β-oxidation and the TCA cycle. NAD+ deficiency in diabetes, aging or in mice fed with HFD has been shown to impaired sirtuin action (152). It has been descried that sirtuin 6 (SIRT6) can be stimulated by FFA. Lack of SIRT6 has been shown to up-regulate the glycolytic pathway, leading to a severe hypoglycemic state. Thus, it is thought that SIRT6 stimulation through FFA (for example during β-oxidation) could produce the inactivation of glycolytic genes (152) (Figure 13). Introdution 80 Figure 13. Interplay between metabolism and epigenetic modifications. Effects of metabolic intermediaries on histone acetylation (A), histone de-acetylation (B), and DNA and histone methylation and de-methylation (C). Adaptation from (152). 6.4.1. Role of lifestyle and nutritional conditions in the epigenetics of obesity and metabolic disease Given the interplay observed between epigenetics and metabolism, it is not surprising the efforts that researchers are putting in trying to understand what could be the role of epigenetics in the etiology of the metabolic diseases. In this sense, Dutch famine (famine that took place in Netherland during the 1944-1945 winter at the end of the War World II) has shed light on the nutritional effects during in-uterus condition over DNA methylation and adulthood metabolic disease. Thus, under-nutrition condition has been related to BMI and metabolic disease through specific DNA methylation marks at genes that regulated lipid or glucose homeostasis as well as adipogenesis (177). For instance, it has been shown in this Dutch famine population that DNA methylation at serine/threonine-protein kinase pim-3 (PIM3) (factor involved in glucose metabolism) could explain BMI in adulthood. DNA methylation at thioredoxin interacting protein (TXNIP), gene that regulates β-cell function, and ATP binding cassette subfamily G Member 1 (ABCG1), which is involved in lipid metabolism, was able to explain (together with other CpG positions) up to 80% of the association observed between the famine and the TG levels. During early stage gestational famine CpG marks near 6phosphofructo-2-kinase/Fructose-2,6-biphosphatase 3 (PFKFB3) (involved in glycolysis) and Methyltransferase like 8 (METTL8) (adipogenesis) were shown to influence TG levels as well (177). Introdution 81 As it is described above, DNA methylation relies on SAM as the methyl-group donor. SAM levels depend on methyl-group donor nutrients such as choline, methionine, folate, etc. It has been described that maternal consumption levels of these methyl-donors just before pregnancy and during pregnancy can determined the DNA methylation levels in the newborn of genes related to metabolism and adipogenesis like RXRα, insulin like growth factor 2 (IGF2) and LEP, as well as of the DNA methyl-transferase DNMT1 (178). DNA methylation pattern cannot only be established during the developmental period, but can also be modified in adulthood under several conditions. For instance, in a randomized control trial where subjects were exposed to Saturated or Polyunsaturated fatty acids (SFA and PUFA, respectively) overfeeding during 7 weeks, DNA methylation changes specific for SFA and PUFA emerged. There were changes at genes involved in metabolism and inflammation in AT like FTO, INSR, Neuronal growth regulator 1 (NEGR1), Fatty acid binding protein 1 (FABP1), Fatty acid binding protein 2 (FABP2), PPARG coactivator 1 alpha (PPARGC1α), Melanocortin 2 receptor (MC2R), melanocortin 3 receptor (MC3R), TNF or IL-6, among others. Moreover, DNA methylation at several loci at baseline were associated with the percentage of body weight increase after the trial (179). As well as food, exercise has also been related to DNA changes in adipose tissue. In a six-month interventional study, it has been described changes in DNA methylation in AT at several CpGs for genes which have been associated to obesity, diabetes and adipocyte metabolism such as Transcription factor 7 like 2 (TCF7L2), Potassium voltage-gated channel subfamily Q member 1 (KCNQ1), Histone deacetylase 4 (HDAC4) or Nuclear receptor corepressor (NCOR) (180). Lifestyle habits as smoking has been related to changes at DNA methylation in AT. These changes were at specific loci that were in turn associated to future weight Introdution 82 gain and metabolic disease risk after smoking cessation. Furthermore, it has been shown that after smoking cessation, the DNA methylation smoking-signature pattern had a longer lasting influence on DNA methylation than the mRNA pattern (181). Histone deregulation has also been related to obesity, metabolic disease and related disorders. For example, Plant homeodomain finger two (PHF2) (which is an histone demethylase) has been shown to regulate CEBPα and PPARγ expression. Thus, it has been reported that specific Plant homeodomain finger two (PHF2) knockout mice display abnormal adipogenesis and a subsequent decrease in AT mass. Moreover, it is thought that PHF2 is important in the regulation of several metabolic tissues being involved in the metabolism of glucose and lipids (182). A total epigenetic remodeling at the AT LEP promoter has been described in Diet induced obese (DIO) mice feed with n-3 PUFAs. These changes comprised the promoter binding increase for Methyl-CpGbinding domain protein 2 (MBD2), DNMTs and several HDACs, with a subsequent increase of DNA methylation and a decrease in the acetylation of H3 and H4. Furthermore, a decrease of H3K4me3 was observed. All of these changes would be interpreted as a compensatory mechanism trying to deal with the extra-energetic consumption experimented (183). This study points out how the epigenetic state of a certain factor involved in the control of energy balance can deeply change in adulthood triggered for changes in the nutritional status. As previously described, histone modifications are strongly associated to metabolism, and some of the erasers and writers involved have been associated to metabolic disorders (152). However, there are no studies about the histone marks profile in human AT, which could give us a better knowledge about the actual epigenetic status and about its possible role in the etiology of the AT related disturbances. Concerning AT, histone modifications studies have been mainly carried out in cultures (3T3L1 or primary pre- Introdution 83 adipocyte cultures) (184,185), and to a less extend in mice. Thus, it has been reported an increase of H3K4me2 enrichment in db/db mice compared with db/m at the promoter of ATPase, H+ transporting, Lysosomal V0 subunit D2 (Atp6v0d2), Matrix metallopeptidase 12 (Mmp12), Triggering receptor expressed on myeloid cells 2 (Trem2) and C-type lectin domain family 4, member d (Clec4d) genes, while this mark was lower in Glycoprotein (transmembrane) nmb (Gpnmb) (186). 6.4.2. Epigenetics alterations in colorectal cancer Epigenetic aberrations in the context of CRC have been studied, which could lead to new subtypes classification based on pharmacological response, and therefore leading to new treatment strategies (187). Indeed, AT has been related to CRC appearance and progression (188), although the role of AT epigenetic regulation and its relationship to CRC has been poorly studied. Epigenetic modifications in peri-tumoral AT have been shown for breast and prostate cancers. Thus, an altered DNA methylation pattern has been described in the surrounding breast malignant cells, alterations that are related to chromosomal organization and with adverse clinical outcome (189). AT DNA methylation pattern has also been implicated not only in the appearance of metabolic diseases but also to the development of cancer by altering the metabolism and increasing the inflammatory environment (190). Differentiated DNA methylation has been reported for genes related to lipid metabolism and immune system (as Acyl-CoA dehydrogenase medium chain (ACADM), Carnitine palmitoyltransferase 1B (CPT1B), Carnitine palmitoyltransferase 1C CPT1C, Fatty acid desaturase 1 (FADS1), Monoacylglycerol O-acyltransferase 1 (MOGAT1), Monoacylglycerol Oacyltransferase 2 (MOGAT2), Solute carrier family 44 member 2 (CTL2) or TAP binding protein (TAPBP)) in peri-prostatic AT of obese and overweight versus lean Introdution 84 subjects with prostate cancer, which could be ultimately contributing to the worsening and different cancer progression observed in subjects with higher adiposity (190). Since epigenetic can be the result of genetic, age, tissue specificity and a given environmental conditions, its study can point out functional factors or pathways that could be affecting to the AT functioning and contributing to metabolic disease and other disorders associated (as CRC), being a more accurate tool than genetic studies (which gives us a fixed scenario) to infer this relationships (137). !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!HYPOTHESIS Hypothesis 87 1Adipose tissue DNA methylation at LPL promoter could be regulating LPL mRNA and be related to serum triglyceride levels, thus participating in the etiology of metabolic syndrome. Besides, these levels of LPL methylation could be related to the response in triglyceride clearance after a fat overload. 2Overall DNA methylation state of the adipose tissue via LINE-1 as well as the DNA methylation promoter regions of genes related to adipogenesis, lipid metabolism and inflammation could be altered in metabolic syndrome playing a role in the etiology of the metabolic disease. 3Adipose tissue DNA methylation at C3 promoter could be regulating C3 mRNA and be related to serum ASP levels in obesity, being evolved in the higher proinflammatory status and impaired lipid storage present in extreme obesity. 4Adipose tissue H3K4me3 enrichment at the promoter of several genes related to adipogenesis, adipose tissue metabolism and inflammation could be modified in accordance with the obesity degree and metabolic status, which could explain in part the development of obesity and metabolic disease. 5Low levels of vitamin D could be related to adipose tissue inflammation and adipose tissue DNA methylation in subjects with colorectal cancer, which in turn could be contributing to colorectal cancer development. Results 95 MANUSCRIPT 1. Adipose Tissue LPL Methylation is Associated with Triglyceride Concentrations in the Metabolic Syndrome Castellano-Castillo D. et al. Clinical Chemistry. 2018;64(1):210-218. Results 96 In order to carry out the objective 1, the levels of DNA methylation of several LPLpromoter-CpG dinucleotides in a CpG island region were analyzed and related to the gene and protein expression levels visceral adipose tissue (VAT) in individuals with (MetS) and without (non-MetS) metabolic syndrome. To perform this, VAT samples were collected from laparoscopic surgical patients, and levels of LPL mRNA, LPL protein and LPL DNA methylation were measured by qPCR, western blot and pyrosequencing. Biochemical and anthropometric variables were analyzed. Moreover, a subset of individuals underwent a dietary fat challenge test and postprandial triglycerides were determined. Anthropometric and biochemical characteristics of the patients Table 1A shows the anthropometric and biochemical parameters of the non-MetS and MetS patients. As expected, the MetS patients had significantly increased glucose, triglycerides, waist circumference, systolic and diastolic blood pressures, BMI, insulin, HOMA-IR, total cholesterol, LDL cholesterol, ApoA1, ApoB and serum leptin levels in comparison to the non-MetS subjects, whereas HDL cholesterol, ApoA1 and adiponectin values were significantly lower compared with the non-MetS subjects. Table 1. Biochemical and anthropometric parameters in non-metabolic syndrome subjects (Non-MetS) and metabolic syndrome subjects (MetS) in the descriptive study (A) and the subpopulation who underwent the fat overload test (B). (A) Study population (B) Fat overload subpopulation Non-MetS (N=70) MetS (N=64) Non-MetS (n=11) MetS (n=26) Age (years) 48.0±13.38 49.79±15.0 44.0±7.9 42.6±6.9 Male/Female (%) 51/49 40/60 27/73 48/52 BMI (Kg/m2) 32.2±10.8** 41.5±12.5 51.6±7.3 52.3±6.9 Waist circumference (cm) 101.5±21.2** 118.7±23.5 137.0±18.2 139.5±17.7 SBP (mm Hg) 124.7±17.6** 139.7±19.7 128.8±19.6 137.6±19.9 Results 97 DBP (mm Hg) 76.2±11.6** 82.6±11.1 83.6±11.9 82.6±12.1 Glucose (mg/dL) 92.7±11.9** 116.4±30.7 93.0±7.9 104.4±18.0 Insulin (pmol/L) 12.0±13.6* 17.8±11.4 26.0±27.4 23.2±13.8 HOMA-IR 2.5±2.0** 5.1±3.4 4.3±3.2 6.1±4.1 Uric acid (mg/dL) 4.7±1.3** 5.7±1.3 5.4±1.0 6.1±1.4 TG (mg/dL) 102.7±43.0** 160.1±64.6 105.1±62.0 142.3±53.5 Post TG (mg/dL) - - 177.7±69.6 193.9±72.2 Cholesterol (mg/dL) 194.1±34.9** 212.3±42.1 174.7±48.5 201.5±38.6 HDL cholesterol (mg/dL) 54.5±12.8** 47.1±12.8 50.4±13.8 44.6±10.0 LDL cholesterol (Friedwald) 118.8±30.7* 133.2±32.5 95.6±37.0 124.9±33.6* ApoA1 (mg/dL) 169.8±25.4* 155.6±26.6 158.3±30.9 148.0±21.6 ApoB (mg/dL) 93.8±22.9* 108.2±22.5 89.5±33.3 104.1±22.9 GOT (mg/dL) 21.0±12.3 20.0±11.5 22.3±8.8 24.8±13.7 GPT (mg/dL) 41.2±22.2 46.3±21.0 46.6±13.9 55.3±20.9 GGT (mg/dL) 56.6±186.5 40.4±27.7 28.6±10.2 34.8±22.9 CRP (mg/dL) 7.4±15.5 5.6±3.4 6.7±6.3 5.2±3.4 Leptin (ng/ml)** 22.5±26.3 47.1±31.3 58.3±22.0 65.8±26.1 Adiponectin (ug/ml)** 12.3±7.2 8.6±4.5 8.7±4.4 7.8±3.3 Definitions: Body Mass Index (BMI), Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), Baseline Triglycerides (TG), Postprandial Triglycerides (Post TG), High Density Lipoprotein (HDL), Low Density Lipoprotein (LDL), Apolipoprotein A1 (ApoA1), Apolipoprotein B (ApoB), Glutamate-Oxaloacetate Transaminase (GOT), Glutamate-Pyruvate Transaminase (GPT), Gamma Glutamyl Transpeptidase (GGT), Systolic Blood Pressure (SBP), Diastolic Blood Pressure (DBP), C-Reactive Protein (CRP). * P<0.05 and ** P<0.01 considered statistically significant between Non-MetS and MetS in each population (A) and (B). On the other hand, Table 1B shows the anthropometric and biochemical parameters for the subset of people who underwent fat overload. Only mean LDL cholesterol was found significantly higher in MetS compared to non-MetS patients. Measures of DNA methylation, LPL mRNA, and protein levels in adipose tissue Figure 1 shows the LPL DNA methylation and mRNA levels. The MetS patients had significantly higher levels of DNA methylation (P<0.001) (Figure 1A) and lower levels of mRNA gene expression (P=0.012) (Figure 1B) compared to the non-MetS subjects. Results 98 Interestingly, this association was confirmed with a correlation analysis, which showed a negative correlation (r=-0.306, P=0.004) between the DNA methylation levels and mRNA levels of LPL (Figure 1C). Figure 1.- Mean and SE of LPL DNA methylation levels (n=41 Non-MetS and n=46 MetS) (A) and LPL relative mRNA (n=70 Non-MetS and n=64 MetS) (B) in non-metabolic (Non-MetS) and metabolic syndrome (MetS) subjects. Figure 1C shows the correlation between both LPL methylation levels and LPL mRNA levels. Finally, we used western blots to examine the protein level of LPL to assess whether the expression levels were translated into the final protein products, confirming a significantly lower LPL protein expression in the MetS patients in comparison with the non-MetS subjects (Figure 2A and Figure 2B). Results 99 Figure 2.- Average and SE of LPL protein levels quantified by western blot in both nonmetabolic syndrome (Non-MetS) and metabolic syndrome (MetS) subjects (n=10). Associations between LPL levels and metabolic syndrome We performed a correlation analysis to analyze the relationships between LPL DNA methylation and the LPL mRNA expression levels and the metabolic and anthropometric variables present in the study subjects (Table 2). We found that the number of MetS components, BMI, waist circunference, glucose, fasting triglyceride concentrations and serum leptin had significant and positive correlations with the LPL DNA methylation levels. Likewise, we found significant negative correlations between the LPL mRNA expression levels and the number of MetS components, BMI, waist circumference, HOMA-IR, glucose concentrations, baseline triglyceride concentrations, and ApoB (Table 2). Finally, we found significant positive correlations between LPL mRNA expression and the serum concentrations of HDL cholesterol and adiponectin levels (Table 2). These results are consistent with the role of the LPL gene in metabolism. Results 100 Table 2. Correlations between LPL DNA methylation (LPLmet) and LPL relative mRNA (LPL mRNA) and several anthropometric and biochemical parameters. Definitions: Number of variables of MetS present (MetS variables), Body Mass Index (BMI), Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), High Density Lipoprotein (HDL), Low Density Lipoprotein (LDL), Log10 of Fasting triglycerides (Log10(TG)), Log 10 of postprandial triglycerides (Log10(PostTG)), Systolic Blood Pressure (SBP), Diastolic Blood Pressure (DBP), Apolipoprotein A1 (ApoA1), Apolipoprotein B (ApoB). * P<0.05 and ** P <0.01 considered statistically significant. LPL methylation LPL mRNA MetS variables (0-5) 0.421** -0.244** BMI (kg/m2) 0.344* -0.239** Waist circumference (cm) 0.297** -0.222* HOMA-IR 0.175 -0.183* Insulin (pmol/L) 0.155 -0.165 Glucose (mg/dL) 0.269* -0.220* Cholesterol (mg/dL) 0.182 0.035 HDL cholesterol (mg/dL) -0.068 0.210* LDL cholesterol (mg/dL) 0.098 0.039 Log10(TG) (mg/dL) 0.246* -0.259** Log10(PostTG) 0.467** -0.360* SBP (mm Hg) 0.135 0.057 DBP (mm Hg) 0.111 0.026 ApoA1 (mg/dL) -0.126 0.130 ApoB (mg/dL) 0.147 -0.301* Leptin (ng/ml) 0.402** -0.139 Adiponectin (ug/ml) -0.039 0.378** Results 101 Epigenetic factors associated with metabolic syndrome In order to study the relationship between LPL DNA methylation and the parameters associated with MetS, and between the methylation and mRNA levels of LPL we performed regression analyses. Regression model 1 (Table 3 model 1) after adjustment for age, gender, BMI and HOMA-IR showed that the statistically significant variables in predicting the increase in LPL promoter methylation levels were the number of components of MetS and the BMI. When we considered LPL mRNA as dependent variable and adjusted by age, gender, BMI and HOMA-IR we found that LPL methylation was the only variable significantly associated with LPL mRNA variability (Table 3 model 2). Results 102 Table 3. Regression analysis with LPL DNA methylation (model 1) and LPL mRNA (model 2) levels as dependent variable. Model 1: LPL DNA Methylation (R=0.472; R2=0.223) Model 2; LPL mRNA (R=0.382; R2=0.147)  P 95% CI  P 95% CI Age 0.066 0.584 -0.023-0.041 -0.029 0.814 -0.012-0.010 BMI 0.298 0.037 0.003-0.080 -0.104 0.492 -0.018-0.009 Gender -0.115 0.302 -1.121-0.383 0.113 0.331 -0.140-0.410 MetS variables (0-5) 0.349 0.009 0.126-0.837 - - - HOMA-IR -0.118 0.351 -0.216-0.078 -0.128 0.337 -0.077-0.027 LPL methylation - - - -0.266 0.025 -0.166-(-0.011) Independent variables for model 1 are Age, BMI, Gender, MS variables and HOMA-IR; and age, BMI, gender, HOMA-IR and LPL methylation for model 2. Abbreviations: Body Mass Index (BMI), Number of variables of MetS present (MetS variables), Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), Lipoprotein Lipase (LPL). Results 103 Furthermore, we used logistic regression analysis to assess the risk of suffering MetS risk according to the methylation levels in LPL DNA promoter. LPL methylation level was found to be the only variable significantly predictive of the MetS state (dependent variable). In this regression, adjusted by age, gender and LPL mRNA levels, an increase in one unit in LPL methylation resulted in a more than two-fold higher likelihood of having MetS (Table 4). Table 4. Logistic regression with membership or not membership to MetS group as dependent variable, and age, gender, LPL methylation and LPL mRNA as independent variables. Men are used as reference gender (0). NonMetS/MetS OR (95% CI) P Age 1.003 (0.96-1.04) 0.865 Gender Male Female 1 (reference) 1.927 (0.73-6.41) 0.158 LPL methylation 2.092 (1.38-3.06) 0.000 LPL mRNA 0.464 (0.23-1.53) 0.282 Abbreviations: Lipoprotein Lipase (LPL). Dietary Fat challenge We designed a dietary fat challenge test in humans to evaluate the relationship between postprandial triglyceride and LPL methylation levels. Positive correlations were found between the baseline and postprandial triglyceride concentrations and the LPL DNA methylation levels (Table 2, Figure 3A and Figure 3B). Accordingly, we found a Results 104 negative correlation between LPL mRNA and both, the baseline and postprandial triglyceride concentrations (Table 2, Figure 3C and Figure 3D). Figure 3. The figure shows the correlation of LPL DNA methylation with the Log of fasting triglycerides levels (A) and Log of postprandrial triglycerides levels (B). (C) shows the correlation between LPL gene expression and Log of fasting triglycerides levels and (D) the association between LPL gene expression and Log of postprandrial triglycerides levels. In a linear regression model adjusted for age, gender, BMI and HOMA-IR we found that LPL methylation was the only variable which explained the postprandial triglyceride concentrations (Table 5). Results 111 Table 3. Pearson’s correlation between LINE-1 CpG positions (P1, P2, P3, P4, P5, P6) and the anthropometric and biochemical variables related to MetS. * p<0.05 and ** p<0.01 were considered statistically significant. MetS index BMI Waist Glucose Tg HDL-cho LDL-cho SBP DBP HOMA-IR LINE-1 P1 -0.167 0.057 -0.031 -0.246* -0.088 0.113 0.082 0.162 0.02 -0.114 LINE-1 P2 -0.233* 0.025 -0.068 -0.334** -0.208 0.074 0.028 0.171 0.010 -0.199 LINE-1 P3 -0.136 0.018 -0.011 -0.168 -0.072 -0.115 0.093 0.220 0.155 -0.101 LINE-1 P4 -0.068 0.042 0.012 -0.158 0.039 -0.112 0.077 0.168 0.010 -0.041 LINE-1 P5 -0.137 0.093 -0.037 -0.238* 0.016 -0.139 0.05 0.136 0.100 -0.088 LINE-1 P6 -0.19 -0.055 -0.05 -0.137 -0.166 0.028 0.052 0.066 0.016 -0.126 Abbreviations: Number of metabolic syndrome variables present in the subject of study (MetS index); Body mass index (BMI); Triglycerides (TG); Highdensity lipoprotein cholesterol (HDL-cho); Low-density lipoprotein cholesterol (LDL-cho); Systolic blood pressure (SBP); Diastolic blood pressure (DBP); Homeostatic model assessment of insulin resistance (HOMA-IR); Long interspersed element 1 DNA methylation at positions 1 to 6 (LINE-1 P1-P6). * and ** mean p<0.05 and p<0.01 respectively according to Pearson’s correlation. Results 112 Gene specific DNA methylation in MetS versus Non MetS Adipogenic and lipid metabolism factors We studied genes related to adipose tissue development, as PPARA, PPARG and their heterodimer partner RXRA. There were no differences at any of the CpG sites included for PPARA, PPARG and RXRA (Figure 1). Nevertheless, a tendency to higher levels of DNA methylation in PPARA for MetS subjects than in Non MetS was observed. Results for the association analyses showed a positive correlation between PPARA P2 with MetS index, TG levels and HOMA-IR. PPARG P1 correlated positively with BMI, while PPARG P1 and P3 were negatively associated to DBP. In the case of the PPAR’s partner RXRA, we found a negative correlation between RXRA P1 with BMI and waist circumference. Results 113 Figure 1. Adipogenic factors DNA methylation levels. DNA methylation profile across the CpG analyzed at the promoters of the adipogenic factors PPARA (A), PPARG (B) and the PPARs partner RXRA (C) in both, Non MetS and MetS groups. Values are given as the mean±SE. Peroxisome proliferator-activated receptor alpha (PPARA); Peroxisome proliferatoractivated receptor gamma (PPARG); Retinoid X receptor alpha (RXRA). Furthermore, a set of CpG sites inside genes related only to lipid metabolism was also pyrosequenced. No differences were found at any of the CpGs analyzed for SREBF1 and SREBF2 regulators (Figure 2A and Figure 2B, respectively). There were no significant DNA methylation differences at any of the LRP1 CpG sites studied either (Figure 2C). In the case of LPL, we found an increase of DNA methylation for the CpG situated at the position 2 (LPL P2) (Figure 2D). We did not find different levels of DNA methylation at any of the CpG studied for SCD and LXRB genes (Figures 2E and 2F). For these genes, we observed that MetS index correlated negatively with SCD P6, while SCD P3 was negatively associated to BMI. Positive associations existed between TG levels and LPL P3, and between HDL-cho and LRP1 P2. Furthermore, there was a negative association between the cholesterol regulator SREBF2 and DBP, specifically with SREBF2 P2. Results 114 Figure 2. Lipid metabolism DNA methylation. The figure shows the DNA methylation in Non MetS and MetS groups at each CpG for several factors related to lipid metabolism as SREBF1 (A), SREBF2 (B), LRP1 (C), LPL (D), SCD (E) and LXRB (F). Values are given as the mean±SE. Sterol regulatory element binding transcription factor 1 (SREBF1); Sterol regulatory element binding transcription factor 2 (SREBF2); Low density lipoprotein receptor-related Results 115 protein 1 (LRP1); Lipoprotein lipase (LPL); Stearoyl-CoA desaturase (SCD); Liver X receptor beta (LXRB). * means p<0.05 according to a Student’s T-test. Inflammation factors Due to the relationship between adipose tissue and inflammation, we analyzed some factors involved in this process. We analyzed 7 CpG sites inside the C3 gene promoter, and we did not find different DNA methylation levels between both, Non MetS and MetS subjects (Figure 3A). We studied 5 CpG sites for the tumor necrosis factor (TNF) as well. In this case, MetS subjects presented a lower DNA methylation levels at 3 out of the 5 CpG sites that were analyzed, concretely at position 1, 2 and 3 (TNF P1-P3) (Figure 3B). The third factor we studied was leptin (LEP), in which we analyzed 4 CpG sites at leptin sequence, which did not present significant differences between Non MetS and MetS subjects (Figure 3C). Results 116 Figure 3. Inflammatory promoters DNA methylation. Comparisons between the Non MetS and the MetS group for DNA methylation at different CpG from genes implied in inflammatory processes as C3 (A), TNF (B) and LEP (C). Values are given as the mean±SE. Complement factor 3 (C3); Tumor necrosis factor (TNF); Leptin (LEP). * means p<0.05 according to a Student’s T-test. On the other hand, a negative relationship was found between MetS index and the DNA methylation levels of TNF P2, TNF P3, TNF P4 and TNF P5 (Table 4). Glucose correlated in a negative way with the DNA methylation of TNF P4 (Table 4). According to triglyceride levels, there were a negative correlations were found with TNF P2 and P5. Inversely to TG, HDL-cho correlated positively with TNF P1, P2, P5 (Table 4). There was also a negative correlation between TNF P4 with LDL-cho and DBP. Furthermore, there were positive and significant correlations between LEP P1 with LDL-cho, SBP and DBP (Table 4). Results 117 Table 4. Correlation analyses between anthropometric and biochemical variables associated to MetS with some of the DNA methylation at the CpG analyzed. Only CpG that presented any significant association are represented. MetS V BMI Waist Glucose Tg HDL-cho LDL-cho SBP DBP HOMA-IR PPARA P2 0.276* 0.076 0.165 0.166 0.392** 0.061 0.08 0.066 -0.025 0.229* PPARG P1 -0.072 0.306* 0.169 -0.224 -0.194 0.015 -0.197 -0.2 -0.293* -0.058 PPARG P3 -0.078 0.138 0.174 0.03 -0.139 0.021 -0.218 0.037 -0.283* 0.112 RXRA P1 -0.102 -0.298** -0.229* -0.052 0.025 -0.095 0.127 -0.066 -0.225 -0.032 SREBF2 P2 0.056 0.006 0.144 0.112 0.136 -0.032 0.189 -0.224 -0.262* 0.121 LRP1 P2 0.09 -0.065 -0.048 0.114 -0.215 0.373* -0.055 0.192 0.180 0.251 LPL P3 0.135 0.029 0.089 0.128 0.245* -0.102 0.085 0.126 -0.111 0.149 SCD P3 -0.056 -0.340* -0.283 -0.096 -0.018 0.108 0.22 0.087 -0.117 -0.03 SCD P6 -0.325* -0.116 -0.17 -0.141 -0.134 0.121 0.102 -0.275 -0.232 -0.172 TNF P1 -0.212 0.132 0.046 -0.034 -0.188 0.283* -0.02 -0.010 -0.115 0.029 TNF P2 -0.420** 0.054 -0.061 -0.192 -0.273* 0.304* -0.195 -0.188 -0.217 -0.196 TNF P3 -0.320* 0.151 -0.021 -0.094 -0.155 0.222 -0.109 -0.237 -0.242 -0.03 TNF P4 -0.330* -0.006 -0.096 -0.278* -0.203 0.098 -0.295* -0.245 -0.305* -0.133 TNF P5 -0.281* 0.132 -0.100 -0.153 -0.281* 0.380** -0.132 -0.097 -0.008 -0.074 LEP P1 0.088 0.081 -0.159 0.061 -0.071 0.015 0.229* 0.264* 0.230* 0.028 Results 118 Abbreviations: Number of metabolic syndrome variables present in the subject of study (MetS V); Body mass index (BMI); Triglycerides (TG); High-density lipoprotein cholesterol (HDLcho); Low-density lipoprotein cholesterol (LDL-cho); Systolic blood pressure (SBP); Diastolic blood pressure (DBP); Homeostatic model assessment of insulin resistance (HOMA-IR); Peroxisome proliferator-activated receptor alpha DNA methylation at position 2 (PPARA P2); Retinoid X receptor alpha methylation at position 1 (RXRA P1); Leptin DNA methylation at position 1 (LEP P1); Sterol regulatory element binding transcription factor DNA methylation at position 2 (SREBF2 P2); Stearoyl-CoA desaturase DNA methylation at positions 3 and 6 (SCD P3 and P6); Tumor necrosis factor DNA methylation at positions P1 to P5 (TNF P1-P5); Peroxisome proliferator-activated receptor gamma DNA methylation at positions 1and 2 (PPARG P1 and P2); Lipoprotein lipase DNA methylation at position 3 (LPL P3); Low density lipoprotein receptor-related protein 1 DNA methylation at position 2 (LRP1 P2). * and ** means p<0.05 and p<0.01 respectively Regression analyses To study the strength of the association observed in the correlation analyses we performed lineal regression analyses corrected by age, gender and BMI. We observed that the DNA methylation levels of PPARA P2 and LPL P3 could explain TG levels (Table 5). Table 5. Lineal regression analysis with fasting triglycerides as dependent variable and PPARA P2, LPL P3 and TNF P2 as independent variables and corrected by age, gender and BMI. Fasting triglycerides (R=0.566; R2=0.320) β P CI 95 % Age 0.111 0.425 -0.582-1.358 Gender -0.268 0.047 -48.377-(-.312) Results 119 BMI -0.101 0.446 -1.825-0.816 PPARA P2 0.332 0.012 1.32-10.012 LPL P3 0.264 0.046 0.099-10.72 TNF P2 -0.117 0.347 -1.867-0.669 Body mass index (BMI); Peroxisome proliferator-activated receptor alpha DNA methylation at position 2 (PPARA P2); Lipoprotein lipase DNA methylation at position 3 (LPL P3); Tumor necrosis factor DNA methylation at position 2 (TNF P2). Furthermore, we performed a logistic regression analyses (harmonized by step method) to determine what factors could predict the risk of having MetS. We observed that TNF P2 remained as a protective variable; with a reduction of 23% of probability of being MetS per unit of DNA methylation increased (Table 6). Table 6. Logistic regression analysis: risk of MetS. Variables that showed a significant association with MetS V at the correlation analyses such as age, gender, PPARA P2, SCD P6, TNF P2 and P5 were introduced as independent variables. A harmonized model in which gender, PPARA P2 and TNF P2 was maintained was generated. Non Mets/MetS (R2=0.506-0.686) β P CI 95% Gender 5.813 0.094 0.739-45.699 PPARA P2 1.630 0.246 0.714-3.719 TNF P2 0.791 0.008 0.664-0.942 Non metabolic syndrome group (Non MetS); Metabolic syndrome group (MetS); Peroxisome proliferator-activated receptor alpha DNA methylation at position 2 (PPARA P2); Tumor necrosis factor DNA methylation at position 2 (TNF P2). Results 127 B HOMA-IR (R=0.59, R2=0.35)  p CI (95%) Age -0.00 0.96 -0.06-0.05 Gender -0.71 0.32 -2.14-0.72 BMI 0.13 0.00 0.04-0.21 C3 mRNA 0.74 0.00 0.26-1.22 C3 methylation 0.07 0.44 -0.12-0.28 Abbreviations: HOMA-IR: Homeostasis Model Assessment of Insulin Resistance; CI: Confidence Interval. Results 129 MANUSCRITP 4. Chromatin Immunoprecipitation Improvements for the Processing of Small Frozen Pieces of Adipose Tissue Castellano-Castillo D. et al. PloS One. 2018;13(2):e0192314. Results 130 In this section we aimed to accomplish the objective 4. In this objective, we optimized the standard protocol of chromatin immunoprecipitation (ChIP) for small pieces of frozen human adipose tissue. In addition, we test ChIP for the histone mark H3K4m3, which is related to active promoters, and validate the performance of the ChIP by analyzing gene promoters for factors usually studied in adipose tissue using qPCR. We have introduced crucial changes to the standard ChIP protocol, improving the homogenization, fixation and de-crosslinking steps, allowing enough immunoprecipited material to be obtained to perform further steps, as we demonstrated by testing H3K4me3 modifications. Thus, we have shown that the use of only 100 mg of frozen AT is enough for ChIP tests, which will help to advance knowledge about epigenetic marks of AT and their significance for metabolic homeostasis. The high lipid content of the AT makes the fixation and subsequent steps difficult to work with. Adipocytes float in the upper layer due to their lipid content (Figure 1.1), which leads to a high loss of tissue in the processing. Thus, standard homogenization methods (mortar and pestle) were not able to extract a proper quantity of DNA, showing a very low performance. In this method, a high quantity of tissue remained stuck to the surfaces of the pestle and mortar that resulted in a high tissue loss, a very low nuclei recovery and no chromatin harvest. We therefore performed two other alternative methods where the fixation and washing of the tissue were prior to the homogenization step. This allowed better tissue manipulation, indeed avoiding loss of tissue. We compared the dounce homogenizer (Figure 1.2) with the ultraturrax homogenizer (Figure 1.3). Results 131 Figure 1. Figure shows the three different workflows performed, using the pestle and mortar (1.1), the Dounce homogenizer (1.2) or the Ultraturrax homogenizer (1.3). In (1.1), the homogenization step was performed using liquid nitrogen, after which it was fixed. After this, the nuclei were pelleted and nucleus lysis buffer was added. Once incubated, the sample was sheared and the chromatin fragmentation and recovery were checked. In the other two alternative methods proposed (1.2 and 1.3), the tissue was cut in small pieces (3 mm) and the fixation step was performed prior to the homogenization. Once homogenized, the nuclei were recovered by centrifugation, nucleus lysis buffer was added and the sample was sheared and the chromatin fragmentation and recovery were checked. The use of the ultraturrax homogenizer results in a higher total DNA recovery after the purification step (Figure 2). Results 132 Figure 2. The figure shows the different efficiency in total DNA recovery between the Dounce and the Ultraturrax homogenizer. 100 mg of frozen adipose tissue was fixed in 1% paraformaldehyde, homogenized either, using the Dounce or Ultraturrax homogenizer and sheared for 40 cycles (30 seconds ON and 30 seconds OFF). A sample of 50 l of the homogenized material was then taken, and the chromatin was de-crosslinked using the fast Chelex-100 method. Total DNA was extracted and quantified by nanodrop. Data are given as means with error bars. Abbreviations: Dounce, Dounce homogenization method; Ultraturrax, ultraturrax homogenization method. (n=6). Once the optimal homogenization method was established, a proper fixation method for AT was found. Concentration of formaldehyde to crosslinking DNA/protein is an important step, which can affect the shearing of the DNA and, consequently, the performance of the immunoprecipitation (IP) and the recovery of DNA. For this reason, we tried two different formaldehyde concentrations in order to improve DNA recovery: the classical concentration of 1% formaldehyde was compared to a lower concentration of 0.5% formaldehyde. In addition, we also tried different incubation times (10, 8, 5 minutes) and temperatures (RT or 37ºC) to fix the sample, although no good results Results 133 were obtained for incubation times longer than 5 minutes and temperatures higher than RT (data not shown). 5 ml of each fixation solution were used to carry out the fixation step for 5 minutes at RT and shaking. Furthermore, the sonication step is highly variable depending on the sonicator and there is even moderate variation between different devices for the same technology. Indeed, it is recommendable to set up the proper shearing method not only for each kind of tissue but also for each device. At the same time, we also determined the sonication time to properly shear the DNA using a Bioruptor sonicator after tissue fixation. We tested 20, 30 and 40 cycles of 30 seconds ON / 30 seconds OFF at high power. We obtained better results using a low concentration of 0.5% formaldehyde together with a number of 40 cycles for DNA recovery (Figure 3A and Figure 3B). However, 1% formaldehyde hindered shearing of the chromatin, independently of the number of sonication cycles. Figure 3. 100 mg of frozen adipose tissue was fixed in either 1% or 0.5% of paraformaldehyde and then homogenized using the Ultraturrax method. It was then sheared for 40 cycles (30 seconds ON and 30 seconds OFF), after which a sample of 50 µl of the homogenized material was taken, and the chromatin was de-crosslinked using the fast Chelex-100 method. Total DNA was extracted and quantified by nanodrop. Fixation at 0.5% presents higher levels of DNA recovery (A) after DNA purification and a better chromatin shear tested by electrophoresis in 2% agarose gel. (B) Comparison of the use of PBS+1% or PBS+0.5% formaldehyde in the Results 134 performance of DNA recovery after de-crosslinking and purifying the DNA. (n=6). Data are given as means with error bars. Finally, due to the small pieces and nature of AT itself, added to the fact that it was frozen, the de-crosslinking and DNA recovery steps may be determinant for the success of the IP and downstream procedures. Two methods for chromatin de-crosslinking were tested: the standard method, which consists of incubating the chromatin at 65ºC for 5 hours followed by proteinase K (PK) treatment at 55º for 1 hour; and a faster method in which chromatin is heated to 100ºC during a shorter period of time of 10 minutes with 10% chelex-100 to protect the DNA. Once the DNA was purified, the data revealed a higher performance for the standard method, in which the chromatin is submitted to a moderate temperature for a long time (Fig 4). Figure 4. 100 mg of frozen adipose tissue was fixed in 0.5% paraformaldehyde and then homogenized using the Ultraturrax method. It was then sheared for 40 cycles (30 seconds ON and 30 seconds OFF), after which a sample of 50 µl of the homogenized material was taken, and the chromatin was de-crosslinked using either the fast Chelex-100 method or a moderate Results 135 temperature for 5 hours plus PK treatment. Figure shows the de-crosslinking step at a moderate temperature for 5 hours and then a PK step improves the quantity of DNA with respect to the method based on the use of Chelex-100. (n=6). Total DNA was extracted and quantified by nanodrop. Data are given as means with error bars. Therefore, after testing different steps during the regular ChIP protocol, several changes have been introduced in order to match the method to small frozen AT samples. The high lipid content in AT hinders tissue manipulation, DNA extraction and even nucleus release and nucleus breakdown. This has led to the development of specialized extraction kits for AT, for example for RNA extraction. Thus, based on our data in AT manipulation (data not shown), we decided to increase the proportion of buffer with respect to the sample quantity compared to regular procedures for the following steps: fixation, washes after fixation, cell lysis and release of nuclei. This allowed us to deal with the high lipid content, avoiding a very thick cell lysate, which could hinder nucleus release. Furthermore, this allowed recovery of a cleaner nucleus pellet, improving the sonication and chromatin release. In these steps, we recommend the use of glass pipettes to remove the liquid discarded in each step since the high lipìd content of AT can become stuck to plastic surfaces, hindering manipulation and leading to tissue loss. Moreover, we determined use of the ultraturrax homogenizer as the best method for homogenization, a fixing solution of PBS+0.5% formaldehyde at RT and the standard de-crosslinking method as the most suitable procedures for small pieces of frozen AT. Up to now, the use of ChIP for AT has been limited to big amounts of tissue, and especially to mouse AT where the conditions are less limiting. Thus, the improvements shown in this work could help researchers study the proteome-DNA interaction in human AT, which is stored frozen in large tissue banks. Results 136 Once the best procedure was established, we applied the method to 100 mg of frozen samples of human AT. The yield of the method after the IP resulted in an average of almost 100 ng of DNA, enough to perform a posterior high throughput sequencing thanks to the high resolution of the latest next generation sequencing methods. On the other hand, in order to improve the performance and DNA recovery, we encourage others to perform ChIP experiments in small rounds of samples Although we have provided an improved method to work with small frozen pieces of AT, we needed to confirm the correct assessment of the IP. We validated our ChIP protocol in mouse and human AT by testing H3K4me3 modifications, a mark of active promoter regions. By qPCR we identified H3K4me3 enrichment on several promoters of genes usually expressed in AT, such as PPARG, SCD, LPL, LEP, SREBF2, as well as a sequence 30 kb before PPARG TSS (Transcription Start Site) in mice (PPARG Out) and a sequence 35 kb before SCD TSS (SCD Out) for humans, both as control regions. We obtained a high percentage of enrichment in both mice and humans (Figure 5A and Figure 5B respectively) for genes usually expressed in AT, like SCD, PPARG, LPL or SREBF2, while control regions presented residual expressions. The presence of H3K4m3 at these promoters has already been demonstrated in several tissues and cell lines (ENCODE project), but to the best of our knowledge, no results are available in white AT. Nevertheless, these gene expressions are usually assessed in AT, which could agree with the high percentage of DNA immunoprecipitation observed in these promoter genes for H3K4m3, a histone associated with active genes. Results 143 (E2F1); Lipoprotein Lipase (LPL); Sterol regulatory element-binding factor 2 (SREBF2); Stearoyl-CoA desaturase 1 (SCD1); Peroxisome proliferator-activated receptor gamma (PPARG); Interleukin 6 (IL6); Tumor necrosis factor (TNF). Accordingly, association analysis of the H3K4me3 mark in the studied genes with the measured clinical variables showed us a positive correlation for the H3K4me3 mark at E2F1, LEP, LPL, SREBF2, SCD1, PPARG, IL6 and TNF promoters with the BMI, HOMA-IR and insulin levels (Table 2). Moreover, there was a positive correlation between glucose and H3K4me3 mark enrichment at the promoter of SCD1, PPARG, E2F1 and IL6 (Table 2). We also observed a positive correlation between the H3K4me3 mark at E2F1, SREBF2 and SCD promoters with the number of metabolic syndrome (MetS) components (MetS Var) presents in the subject. Results 144 Table 2. Spearman correlation analysis between H3K4me3 mark enrichment at the study gene promoters and the anthropometric and biochemical variables. * and ** mean p<0.05 and p<0.01 respectively. H3K4me3 enrichment Age BMI Glucose Insulin HOMA-IR Tg Chol HDL-C LDL-C SBP DBP MetS Var E2F1 -0.214 0.530** 0.552** 0.573** 0.594** -0.01 -0.256 -0.163 -0.257 0.123 0.072 0.448* LEP -0.148 0.364 0.361 0.348 0.367 0.044 -0.133 -0.12 -0.143 0.075 0.006 0.279 LPL -0.168 0.430* 0.395* 0.437* 0.463* -0.008 -0.208 -0.141 -0.198 0.154 0.064 0.333 SREBF2 -0.236 0.467* 0.442* 0.403* 0.441* -0.048 -0.178 -0.232 -0.113 0.088 0.062 0.463* SCD -0.242 0.528** 0.513** 0.529** 0.548** -0.034 -0.214 -0.186 -0.213 0.157 0.071 0.420* PPARG -0.243 0.488** 0.399* 0.460* 0.479** -0.051 -0.263 -0.212 -0.246 0.073 0.038 0.369 IL6 -0.131 0.430* 0.497** 0.472* 0.501** 0.062 -0.094 -0.15 -0.065 0.191 0.047 0.327 TNF 0.009 0.23 0.379* 0.295 0.305 -0.051 -0.093 -0.008 -0.105 0.085 0.056 0.163 Abbreviations: E2F transcription factor 1 (E2F1); Leptin (LEP); Lipoprotein Lipase (LPL); Sterol regulatory element-binding factor 2 (SREBF2); StearoylCoA desaturase 1 (SCD1); Peroxisome proliferator-activated receptor gamma (PPARG); Interleukin 6 (IL6); Tumor necrosis factor (TNF); Body mass index (BMI); Homeostatic model assessment of insulin resistance (HOMA-IR); Triglycerides (Tg); Total cholesterol (Chol); High-density lipoprotein cholesterol (HDL-C); Low-density lipoprotein cholesterol (LDL-C); Systolic blood pressure (SBP); Diastolic blood pressure (DBP); Number of MetS variables (MetS Var). Results 145 Gene expression levels Levels of gene expression in the studied genes are despicted in Figure 2. With respect to the Lean NG group, LPL, SCD and PPARG mRNA levels were lower in the MO PD group, whilst higher mRNA levels were described for IL6 and TNF genes. Results 146 Figure 2. Group comparisons of the relative mRNA levels of the study genes. Different letters mean significant differences at p<0.05 according to Kruskall-Wallis and Mann-Withney U-Test. Abbreviations: Lean Normoglycemic (Lean NG); Morbid obese normoglycemic (MO NG); Morbid obese prediabetic (MO PD); E2F transcription factor 1 (E2F1); Lipoprotein Lipase (LPL); Sterol regulatory element-binding factor 2 (SREBF2); Stearoyl-CoA desaturase 1 (SCD1); Peroxisome proliferator-activated receptor gamma (PPARG); Interleukin 6 (IL6); Tumor necrosis factor (TNF). Correlation analyses were in line with the expression results. Thus, LEP, IL6 and TNF mRNA levels were positively associated to BMI, while LPL, SCD and PPARG mRNA levels decreased in line with BMI (Table 3). On the other hand, HOMA-IR correlated with LEP, IL6 and TNF mRNA in a positive way, and negatively with SREBF2 and SCD (Table 3). Regarding the rest of measured variables, positive associations were described between HDL-C and LPL, SCD and PPARG gene expressions. Interestingly, E2F1 mRNA levels were negatively associated with total cholesterol (Chol), HDL-C and LDL-C (Table 3). Results 147 Table 3. Spearman correlation analysis between the relative mRNA levels at the study genes and the anthropometric and biochemical variables. * and ** mean p<0.05 and p<0.01 respectively. Relative mRNA Age BMI Glucose Insulin HOMA-IR Tg Chol HDL-C LDL-C SBP DBP MetS Var E2F1 -0.256 0.359 0.116 0.194 0.19 -0.151 -0.559** -0.407* -0.534** -0.144 0.097 0.284 LEP -0.321 0.522** 0.334 0.673** 0.683** 0.179 -0.213 -0.308 -0.093 0.04 -0.095 0.533** LPL 0.103 -0.500** -0.295 -0.33 -0.327 -0.189 0.31 0.516** 0.203 0.008 -0.102 -0.454* SREBF2 0.088 -0.258 -0.178 -0.391* -0.409* -0.238 -0.082 -0.024 -0.069 -0.209 -0.238 -0.373 SCD 0.375* -0.681** -0.319 -0.526** -0.525** -0.124 0.381* 0.555** 0.277 0.054 -0.165 -0.597** PPARG 0.061 -0.408* -0.214 -0.216 -0.204 -0.112 0.221 0.371* 0.182 0.097 0.161 -0.28 IL6 -0.291 0.729** 0.486* 0.571** 0.590** 0.036 -0.285 -0.223 -0.322 -0.025 0.144 0.533** TNF -0.298 0.679** 0.322 0.594** 0.612** 0.024 -0.263 -0.175 -0.335 0.048 -0.108 0.444* E2F transcription factor 1 (E2F1); Leptin (LEP); Lipoprotein Lipase (LPL); Sterol regulatory element-binding factor 2 (SREBF2); Stearoyl-CoA desaturase 1 (SCD1); Peroxisome proliferator-activated receptor gamma (PPARG); Interleukin 6 (IL6); Tumor necrosis factor (TNF); Body mass index (BMI); Homeostatic model assessment of insulin resistance (HOMA-IR); Triglycerides (Tg); Total cholesterol (Chol); High-density lipoprotein cholesterol (HDL-C); Low-density lipoprotein cholesterol (LDL-C); Systolic blood pressure (SBP); Diastolic blood pressure (DBP); Number of MetS variables (MetS Var). Results 148 Relationship between H3K4me3 mark levels and gene expression levels In order to analyze whether the promoter H3K4me3 levels could be related to gene expression we performed spearman’s correlation analysis in the whole population between the mRNA levels and H3K4me3 enrichment at each gene. We did not observe any significant association between the promoter H3K4me3 levels and the mRNA levels for any gene except to E2F1, in which a positive correlation was observed (r=0.422, p=0.04). Multivariate models In addition, a harmonized lineal regression analyses showed that BMI was heavily explained by H3K4me3 enrichment levels at the promoter of E2F1 and LPL, and by the mRNA levels of LEP and SCD (Table 4). In this model, these four variables could explain up to 83% of the BMI variability present in our studied population. Table 4. Harmonized lineal regression analysis with BMI as dependent variable, which was corrected by age and sex. H3K4me3 enrichment at gene promoters and gene expression of genes that showed significant association in the spearman correlation analysis were introduced in the model,. BMI (R=0.91, R2=0.83) Beta p 95% CI E2F1 H3K4me3 0.979 0.001 0.844 to 2.922 LPL H3K4me3 -0.813 0.003 -3.640 to -0.840 LEP mRNA 0.344 0.002 38.32 to 152.97 SCD mRNA -0.516 0.000 -3.650 to -1.459 Results 149 Abbreviations: Body mass index (BMI); E2F transcription factor 1 (E2F1); Lipoprotein Lipase (LPL); Leptin (LEP); Stearoyl-CoA desaturase 1 (SCD1). On the other hand, when the studied variable is HOMA-IR, a harmonized lineal regression showed that the H3K4me3 enrichment at the promoter of SCD and IL6, together with the mRNA levels of LEP and SCD could explain a 79% of the variation observed in the HOMA-IR (Table 5). Table 5. Harmonized lineal regression analysis with HOMA-IR as dependent variable, which was corrected by age and sex. H3K4me3 enrichment at gene promoters and gene expression of genes that showed significant association in the spearman correlation analysis were introduced in the model,. HOMA-IR (R=0.89, R2=0.79) Beta p 95% CI Gender -0.259 0.047 -3.598 to -0.027 SCD H3K4me3 0.792 0.016 0.049 to 0.417 IL6 H3K4me3 -0.666 0.030 -1.769 to -0.105 LEP mRNA 0.564 0.000 16.60 to 44.82 SCD mRNA -0.261 0.065 -0.541 to 0.018 Homeostatic model assessment of insulin resistance (HOMA-IR); Stearoyl-CoA desaturase 1 (SCD1); Interleukin 6 (IL6); Leptin (LEP). Results 151 MANUSCRIPT 6. Adipose Tissue Inflammation and VDR Expression and Methylation in Colorectal Cancer Castellano-Castillo D. et al. Clinical Epigenetics. 2018;10:60. Results 152 The objective 6 was accomplished in this section. The aim of this study was to explore the relationship between serum 25-hydroxyvitamin D (25(OH)D), adipose tissue gene expression of VD receptor (VDR), pro-inflammatory markers and the epigenetic factor DNA methyltransferase 3a (DNMT3A) as well as VDR and NFκB1 promoters methylation in subjects with colorectal cancer (CRC) and witouth CRC (Control). Blood and visceral adipose tissue from 57 CRC and 50 healthy control subjects were collected. mRNA was measured by qPCR using Taqman technology while bisulfite treated DNA was pyrosequensed using the PyromarkQ96 technology in order to analyze DNA methylation. Protein levels were measured by Western-blot. Anthropometric and biochemical variables Table 1 shows the biochemical and anthropometric characteristics of the study groups. There were no differences in age, BMI or gender between the control and CRC groups. The CRC group had lower levels of insulin, total cholesterol, HDL-C and LDL-C than the control group. In contrast, the CRC group presented higher levels of plasma triglycerides when compared with the control group. Table 1. Anthropometric and biochemical variables of the study groups Control (n=57) CRC (n=50) Age (years) 64.94±8.84 68.035±8.43 Male/Female (%)* 68/32 45/55 BMI (kg/m2) 28.51±4.21 27.61±3.91 Waist (cm) 96.55±11.64 97±12.74 Glucose (mg/dl) 111.72±28.77 125.035±46.87