Alterations in composition and function of human gut microbiota can affect physiological processes and are known to be associated with many diseases including inflammatory bowel disease, hypertension, asthma and colon cancer. Complex interactions between gut microbiota, environmental toxins, nutrients and host genetics may result in an increased permeability of the gut, which is closely linked to the presence of adverse metabolic conditions. As a consequence, translocation of bacterial DNA into the blood circulation increases in patients with obesity. Obesity is a growing health problem worldwide and often paired with severe secondary complications, such as type 2 diabetes or cardiovascular problems. A main feature of disease progression is a chronic low-grade inflammation of adipose tissue which contributes to the development and aggravation of insulin resistance and many of the underlying mechanisms are still unknown. Although data from mice studies suggest that the presence of bacterial components in adipose tissue can support these processes, human studies on this subject are lacking.
In my thesis entitled ``Bacterial translocation to adipose tissue in metabolic diseas'' I provide evidence supporting the initial hypothesis, namely that bacterial DNA is present in adipose tissue, even after stringent controlling for contaminants. To this end, I established a wet lab routine protocol to eliminate contamination as well as a bioinformatics pipeline accounting for contamination by subtracting negative controls. Briefly, this included the use of lab ware and reagents UV-treated for at least 90 minutes, the use of breath protection, extra-long gloves and single-use lab coats as well as working under a sterile laminar flood hood in a clean lab free of any PCR products. The bioinformatics pipeline employed commonly used 16S rRNA gene analysis tools including qiime2, phyloseq and DESeq2 as well as decontam, a novel tool to extract negative controls. Observed quantity of bacterial DNA was in the range of 1 to 10 pg/µg total isolated genomic DNA, which is equivalent to about 0.01 to 0.7% of bacterial cells per human cell. The highest quantity was present in subcutaneous adipose tissue, followed by mesenteric adipose tissue. Bacterial amount correlated with adipose tissue macrophages and PPARG expression in omental and with IL1B and TNF expression in subcutaneous adipose tissue. Mesenteric adipose tissue showed the highest diversity of the observed genera. The most commonly observed phyla in all tissues were Proteobacteria and Firmicutes, which is in line with previously published data on blood bacterial DNA. Still, many genera were predominantly found in specific tissues, e.g. Enterobacter in subcutaneous and Acinetobacter in omental adipose tissue. I further showed that the distribution of observed features could partially be explained by markers of insulin resistance (HOMA-IR, HbA1c) and inflammation (IL-6, TNFa, macrophages) and that certain genera, such as Rhodoferax or Lactobacillus are associated with type 2 diabetes status. In first functional approaches I demonstrated that concentrations of bacterial DNA in the observed range are sufficient to stimulate an inflammatory response in immortalized subcutaneous adipocytes derived from a healthy donor. The effect was most prominent after four hours of treatment and increased in a dose-dependent matter.
One of the aims in the present study was to determine levels of gut leakage by measuring zonulin, the most commonly used biomarker for intestinal permeability in humans, and analyze possible associations with adipose tissue bacterial signature. As there are few well-conducted studies on circulating zonulin levels in patients with metabolic diseases, I first performed a correlation study in the available and metabolically well-characterized Sorbs cohort. Circulating zonulin correlated significantly positive with BMI, fasting glucose, triglycerides and cholesterol and negatively with HOMA-IS, high density lipoprotein and circulating adiponectin levels. Albeit these strong correlations with markers of glucose and lipid metabolism supported previously reported findings, the results pointed to some inconsistencies. As zonulin is reported to be pre-haptoglobin 2 (preHP2), and about 15% percent of a typical western European population are homozygous for haptoglobin 1, they should not express zonulin at all. I confirmed in the Sorbs cohort previously reported distributions of haptoglobin genotypes and showed that the target of the only commercially available zonulin ELISA kit was not related to haptoglobin genotype, therefore presumably not measuring zonulin/ preHP2. Subsequently, I identified properdin as a possible target by employing mass spectrometry approaches. Properdin is structurally related to haptoglobin, as both proteins belong to the mannose-associated serine protease family, however further experiments are needed to validate a possible functional resemblance.
In regard to bacterial translocation two adipose tissue depots were of notable interest due to their close proximity to the gastrointestinal tract. Mesenteric adipose tissue is located around the small intestine and the adipose tissue of the appendices epiploicae, small chunks of fat also called epiploic adipose tissue, are directly attached to the colon. After a thorough literature research I could also assert that both adipose tissues were rarely analyzed in the context of obesity. Therefore they were extensively investigated by measuring gene expression of adipo(cyto)kines, circulating inflammatory markers and analyzing adipocyte size and adipose tissue macrophages. Furthermore, a ``multiomics'' characterization was conducted and by analyzing transcriptome and methylome data I could identify epiploic adipose tissue as a tissue of interest in regard to type 2 diabetes and insulin resistance, which was further confirmed by untargeted proteomics data. Contrary to initial assumptions, I observed only a slight increase in translocation of bacterial DNA and no increased inflammation, as measured by tissue specific TNF and IL6 expression as well as adipose tissue macrophage infiltration. However, both transcriptome and proteome profiles allowed a clear discrimination of patients with and without insulin resistance which was most distinct in epiploic adipose tissue. Compared to other fat depots, epiploic adipose tissue exhibited a discriminable metabolic profile whereas mesenteric adipose tissue was more similar to omental-visceral adipose tissue. Most strikingly, epiploic adipose tissue showed a strong increase in leptin expression and, in general, the upregulation of various metabolic pathways involved in sugar, amino acid or sphingolipid metabolism. In accord with the leaky gut hypothesis high expression of lipopolysaccharide binding protein and various pathways involved in chemokine signaling were observed. In summary, I did not observe an increase in bacterial DNA or adipose tissue macrophages, but demonstrated elevated inflammatory signals such as increased chemokine or IL-8 signaling which are linked with an overall increase of metabolic processes and an increased expression of various adipokines. Epiploic adipose tissue might have a watch dog function by being the first adipose tissue sensing and forwarding certain (microbial) stimuli from the large intestine to the host.
In the last part of my thesis I addressed a possible role of the HLA genomic region on the development of type 2 diabetes. The influence of HLA genetics on type 2 diabetes has been under debate for several decades, since HLA was recognized to largely contribute to type 1 diabetes heritability. However, studies remained inconclusive due to lacking cohorts with sample sizes providing sufficient statistical power for association analyses. More recently, animal studies suggested MHC class II proteins as crucial factors mediating adipose tissue inflammation and insulin resistance. The sample size of the leaky gut cohort was insufficient to determine any correlation between HLA class II genotypes and the presence or type of bacteria in adipose tissue due to the high variability in the observed genomic region. Yet, I had access to three large population-based cohorts which allowed me to analyze associations between HLA class II alleles and type 2 diabetes. Therefore HLA genotypes of the LIFE-Adult (N=4649), LIFE-Heart (N=4815) and Sorbs (N=949) cohort were imputed from SNP genotyping data and analyzed for association with type 2 diabetes. In a meta-analysis including all three cohorts, I identified a protective effect for the well-established type 1 diabetes protective haplotype DRB1*15:01~DQA1*01:02~DQB1*06:02 and confirmed DRB1*07:01~DQA1*02:01~DQB1*03:03 as a risk haplotype in non-insulin treated diabetes. These results suggest that the genetic foundation of both type 1 and 2 diabetes shares common elements involving the HLA class II locus.
In conclusion, to the best of my knowledge, I provide in my work the first contaminant-aware identification of bacterial DNA in human adipose tissue and highlight the importance of analyzing novel adipose tissue depots by showing that fat of the appendices epiploicae, previously only considered to have a cushioning function, is metabolically active and possibly involved in the development of insulin resistance.
Identifer | oai:union.ndltd.org:DRESDEN/oai:qucosa:de:qucosa:72500 |
Date | 23 October 2020 |
Creators | Massier, Lucas |
Contributors | Universität Leipzig |
Source Sets | Hochschulschriftenserver (HSSS) der SLUB Dresden |
Language | English, German |
Detected Language | English |
Type | info:eu-repo/semantics/acceptedVersion, doc-type:doctoralThesis, info:eu-repo/semantics/doctoralThesis, doc-type:Text |
Rights | info:eu-repo/semantics/openAccess |
Relation | https://doi.org/10.1136/gutjnl-2019-320118, https://doi.org/10.1210/clinem/dgaa027, https://doi.org/10.3389/fendo.2018.00022 |
Page generated in 0.0034 seconds