Final Report Summary - ADIPOSE GENETICS (The contribution of adipose tissue gene regulation to obesity-related pathogenesis)
Obesity is a major healthcare challenge linked to risk for type 2 diabetes (T2D) and cardiovascular disease (CVD). Susceptibility to obesity-related disease is largely determined by the distribution of adipose tissue in the body, with accumulation of visceral fat (central obesity) comprising a major risk factor for disease. It well established that adipose depots from different parts of the body display distinct biological properties and the biological mechanisms that govern these differences are likely to be in part responsible for rendering visceral fat a major risk factor for disease. The overall objective of this project is to obtain a better understanding of the biology of adipose tissue, to highlight critical differences between subcutaneous (S) and visceral (V) fat, and to link these differences to differential disease risk.
To date, we have collected biological material (S and V fat, blood) from 115 individuals (62 females, 53 males) with sample collection being ongoing. Through seamless collaboration with surgeons at Laiko General Hospital (LGH) and colleagues at Harokopion University of Athens (HUA), we have established a pipeline covering all processes: from sample collection in the surgery room, to nucleic acid extraction in the lab. We have extracted high quality RNA from adipose tissue, buffy coat DNA, and peripheral blood mononuclear cells (PBMCs) from 200 samples (100 S, 100 V). For each individual, we have collected anthropometric traits and information on diet and lifestyle habits. This is the first collection of its kind in Greece and meets high international standards of quality and patient data.
We carried out RNA sequencing (RNA-Seq) for a subset of 84 samples (S and V fat from 42 female individuals) on the Illumina HiSeq2000 platform with paired-end 49 bp sequencing. DNA from 41 of the above individuals was sent for genotyping on the new Illumina Omni 2.5 exome chip. To enable analysis of RNA-Seq data, we established a bioinformatics pipeline based on previous work by our group, and in collaboration with experts at the McGill University and Genome Quebec Innovation Centre. Our pipeline covers quality control of raw reads, mapping, as well as differential expression (DE) analysis. We are currently incorporating software for genotyping and expression association analyses. Genotyping data will be available in early November 2013, and expression association analyses for the identification of regulatory variants active in adipose tissue will start imminently.
To identify regions of open chromatin, we have carried out formaldehyde-assisted isolation of regulatory elements (FAIRE). FAIRE protocols are being optimised to extract chromatin from both S and V fat and various approaches for tissue pulverization, cell lysis, chromatin sonication have been tested to obtain chromatin that can be taken forward to sequencing. Finally, although not linked to our initial proposal, our colleagues at HUA are investigating patterns of dietary intake and lifestyle behaviour, for participating individuals to complement our work on the biology of adipose tissue.
We have conducted analysis of DE for genes and isoforms between different tissues (S vs. V), and within the same tissue for individuals of high vs. low overall/central adiposity. Analysis to date is restricted to female individuals and we expect a fraction of findings to be sex-specific. Between tissue comparisons revealed high correlation of expressed genes in S and V fat. However, 1,108 genes were DE and these were enriched for biological properties linked to embryonic morphogenesis and development, cell adhesion and signalling, as well as pattern specification processes. DE genes include multiple developmental loci, including IRX1, IRX2, IRX5, HOX A, B, C and D genes, TBX15, and RSPO3. When relating our findings to those of other studies, we found that TBX15 and RPSO3 have been identified as loci influenced the distribution of body fat. Our results provide a functional mechanism (differences in expression levels) for the action of these genes. DE genes were also enriched for multiple loci encoding non-coding RNAs (lnc and antisense RNAs), whose role has not been studied to date, and which comprise excellent candidates for further investigation. We also identified 34 clusters of DE genes, each comprising of a known gene(s) and a non-coding RNA species, implying likely co-regulation. These clusters were enriched for biological properties linked to pattern formation and development, but also for biological pathways linked to WNT signalling, a process implicated in regulation of adipogenesis.
Using a subset of samples, we also investigated gene and isoform DE within adipose tissue types, for normal weight individuals vs. individuals with high: a) overall adiposity (BMI), and b) central adiposity. Comparison within S tissue using the BMI criterion revealed 77 DE genes, enriched for properties linked to inflammatory response, embryo implantation, and biological adhesion. We also recorded enrichment in disease classes “immune” and “infection”. No DE isoforms were identified. Using the central adiposity criterion, we identified 108 DE genes (largely non-overlapping with the 77 genes above), and 37 DE isoforms, a large fraction of which corresponds to genes with unknown function (including lncRNAs). Comparison within V tissue using the BMI criterion revealed 1,077 DE genes, enriched for properties linked to immune response, cell migration, cell adhesion, and fatty acid metabolism. We also recorded enrichment for disease classes “immune”, “infection”, and “cardiovascular”. We detected 613 DE isoforms enriched for biological properties and disease terms as above, but also for biological pathways linked to complement cascades, chemokine signaling, and leukocyte migration. Using the central adiposity criterion we detected 113 DE genes (12 and three expressed only in individuals with central adiposity and normal weight respectively) and 43 DE isoforms, a large fraction of which correspond to genes of as yet unknown function.
Overall we put forth an unbiased list of candidates with a possible role in differential adipose depot function and link to obesity-related disease. We show that partitioning data by overall and by central adiposity does not yield extensively overlapping results, suggesting that these two adiposity criteria describe different biology.
Upon reception of genotyping data we will conduct expression association analyses to identify regulatory variation and will overlap results with regions of open chromatin to fine map causal variants. Identifying candidate functional variants will give us a new set of findings to compare with existing literature and will enable us to offer functional explanations for statistical signals lacking biological interpretation to date. Furthermore, the new genotyping chip will allow us to evaluate 200,000 variants with a higher prior probability of being causal. The use of such findings is likely to be high since it will shed light on biological processes specific to S and V fat, but also to obese/centrally obese individuals, while distinguishing between females and males. Some of these processes are likely linked to the increased risk for diseases such as T2D and CVD. Given that we have disease status and family history for participating individuals, we aim to evaluate this last point to the extent that this is possible.
The established sample collection and data represents a landmark study for Greece and will enable further data mining studies to be performed. Unravelling the mechanisms that govern the biology of S and V fat is a key step in elucidating the pathogenesis of obesity-related diseases. The ultimate goal and resulting socio-economic impact of our study translates into identification of key players that will bring us closer to defining prognostic markers for obesity-related disorders, and potential targets for pharmaceutical intervention. Obtaining a better understanding of the biology of adipose tissue will also have important implications for the approach applied by clinicians who treat obese patients and for public health officials who make decisions about budget and public awareness campaigns. Already, our and others’ results suggest that obesity-related pathogenesis should be approached differently in females and males and that different metrics of adiposity highlight distinct aspects of biology. We have already highlighted these findings to collaborating clinicians.