Final Report Summary - FLORINASH (The role of intestinal microflora in non-alcoholic fatty liver disease (NAFLD)) Executive Summary:Non Alcoholic Fatty Liver Disease is an increasingly important source of morbidity in western populations. It is assumed that the causal mechanisms would be residing in molecules issued from intestinal microbiota. The general objective of FLORINASH was to increase, disseminate, and gain new knowledge about the biological processes and mechanisms of the host to microbiota cross talk responsible for the development of NAFLD.To these aims two cohorts of several hundred obese patients have been generated under WP 1 clinical cohorts and sample generation, extensively characterized from a metabolic and cardiovascular point of view and biological samples were collected. A data base has been generated and uploaded with all OMICS data and clinical parameters from cohort one and two.Cohort 1 contains Plasma, serum, urine and feces of 188 patients (90 from Italy + 98 from Spain) and liver samples from 147 patients (75 from Italy + 72 from Spain) collected in two consecutive visits.Cohort 2 includes complete clinical, anthropometrical and biochemical parameters, OGTT, hepatic echography, measurements of the IMT and plasma, serum, urine and feces samples from 757 (351 Italy + 406 Spain) participants (32% men). 75 additional “healthy” volunteers (35% men) with clinical characterization, plasma, serum, urine and feces; OGTT, IMT, hepatic echography and evaluation of body composition were also recruited in Spain.WP2 Omics data acquisition and integration focused on the omics generation and databasing and as a consequence worked in close interaction with WP1 and WP3. Its main outcome is the database with all omics and clinical data as well as the long-term archiving of the FLORINASH samples. All data has been curated, the data sets include: - clinical data - liver biopsy transcriptome - serum MS-based proteomics - serum MS-based lipid profiles - serum NMR-based metabolic profiles - urine NMR-based metabolic profiles - fecal 16S rRNA microbial phylogenetic profiles - NGS metagenomesWP3 Modelling, Refinement and Challenging focused on data modeling and systems medicine and as a consequence worked in close interaction with WP1 (clinical cohorts and sample generation) and WP2. Project-derived molecular profiles and clinical information were pre-processed and curated. Each pre-processed dataset was archived in the updated FLORINASH database. Final versions of the processing pipelines were developed for urine and serum metabolomics, serum lipidomics, serum peptidomics, fecal 16s taxonomic and metagenomic data. Final data analysis was performed on metabolomic, lipidomic, transcriptomic, proteomic, taxonomic (16s rRNA) and metagenomic (NGS sequences) and datasets were further integrated using statistical modelling to identify markers and biological processes associated with disease state. The project-derived profiles and information represent a unique clinical and biological resource based on the FLORINASH cohorts on metabolic (derived from urine / serum / faeces / clinical measures), hepatic gene expression variations, proteomic (derived from plasma and urine), taxonomic and metagenomic markers associated with disease state. WP 4 Experimental intervention on mouse models to refine the human hypothesesThe WP4 partners developed nutritional and genetic models that recapitulate the various stages of metabolic liver diseases ranging from steatosis, NASH to fibrosis.They also identified from the human FLORINASH cohort, potential interesting proteins or metabolic pathways that could play a role in the development of steatosis and NASH. These hits were validated in relevant mice models.WP5 Validation of intestinal microflora /LPS hypothesis In WP5 Germ free mice have been inoculated with gut microbiota from patients with low or high scores of liver steatosis. A subgroup of those mice was fed with a normal diet, the another group was fed with a high fat diet to trigger the disease. Urine, blood, feces and livers were collected. The molecular analyses of the liver and metabolic features show that no differences in NAFLD or metabolic disease developed. Therefore, the gut microbiota showed no influence.The scientific work was accompanied by a non scientific WP6, dedicated to the coordination and management of the project.Project Context and Objectives:The general objective of this collaborative project was to increase, disseminate, and gain new knowledge about the biological processes and mechanisms which are responsible for the development of NAFLD which is an increasingly important source of morbidity in western populations.Cases of fatty liver disease with inflammation that resembled alcoholic steatohepatitis but occurring in non drinkers were described 30 years ago, first in the Japanese literature and then in the United States. Ludwig coined the term non alcoholic steatohepatitis (NASH) in 1980. The more embracing term non alcoholic fatty liver disease (NAFLD) has been adopted to cover the full spectrum of metabolic fatty liver disorders. This progressive chronic disease is characterized by a graded severity which could lead to hepatic failure and premature death.The clinical diagnosis of NAFLD normally requires liver biopsies for the histological analysis which is obviously invasive and represents a risk for the patient. Therefore, there is a strong need to identify non invasive predictive and diagnostic markers of hepatic damage. We suggested that such markers could originate from intestinal microflora and be detected in urine, plasma and faecal water. Intestinal microflora is a causal mechanism of insulin resistance and obesity and hence is strongly associated with NAFLD. Therefore one objective of the consortium was to determine profiles of markers characterizing the severity of NAFLD within the context of insulin resistance and obesity which are present in different biological samples.This project has thus been carried out in order to address the need to find a tool to predict the risk of liver diseases such as non-alcoholic fatty liver disease (NAFLD) in the context of metabolic disease. Indeed, as an important impact of chronic metabolic disease is the progression to organ complications such as non-alcoholic hepatic steatosis (NASH), the early and accurate prediction and diagnosis of NAFLD may prevent the occurrence of NASH and liver cancer, finally reducing the cost of treatment. This overall objective was pursued by implementation of the specific workpackages of the present project. In this context, clinical data (phenotype) and samples from participants from two different European countries (Spain and Italy) served and will serve in future to generate a set of conventional and new metabolic biomarkers suitable for the prediction of NAFLD in the context of metabolic disease. As mentioned above, such metabolic biomarkers can or could be used as diagnostic tools and/or (potentially) therapeutic targets. The main objectives for WP1 were: 1st, to perform in a final cohort of (at least) 100 NAFLD patients an extensive assessment of the clinical phenotype, insulin resistance, anthropometrical measurements in association with liver biopsies and urine, faeces, plasma, serum and buffy coat samples (cohort 1); 2nd, to recruit a cohort 2 of (at least) 700 obese patients and lean controls properly characterized in this context and with urine, stool, serum and plasma samples to set the final validation of biomarkers eventually identified in cohort 1 on cohort 2); and 3rd, sample collection, processing, storage, control and distribution to FLORINASH partners.WP2The objectives of WP2 were to: generate metabolomic, metagenomic, transcriptomic, and proteomic profiles from human and animal models for urine, blood, fecal waters and liver samples generate data about protein expression of ER stress and lipogenic transcription factors from human biopsies Design and implement a centralized database to store data acquired from patients and mice Store data from different omics platforms Assure accessibility to data for all partners Pre-process data to be used in WP3.WP3The objectives of WP3 were to: Perform integrative cross-platform statistical modelling using partner 4 in-house software Identify molecules that can be used as biomarkers or set of biomarkers to generate a simple and reliable index of NAFLD severity Produce a final integrative map of metabolic dysfunction to serve as working diagnosis Identify profiles for at risk patients (stratification of individuals based on an integrated ‘omic’ strategy) Propose a strategy for the validation of the risk profiles in human cohorts (recruited in WP1) Prioritize targets for drug design Drug design on selected HITSWP4WP4 was dedicated to the generation of new knowledge on the molecular mechanisms involved in the development of NAFLD in mice by focusing particularly on the interplay between lipogenesis, endoplasmic reticulum (ER) stress and inflammation. The objectives of WP4 were: to set up relevant animal models that can recapitulate as close as possible the human metabolic hepatic diseases (defined as NAFLD). For this purpose, genetic and nutritional mice models have been developed to mimic the different stages of NAFLD ranging from steatosis to NASH. to validate molecular hypotheses based on ER Stress, inflammation and lipogenic transcription factors pathways that were identified by the partners of the consortium. to refine hypotheses about the metabolic markers (HITS) identified by large-scale studies in human biopsies thanks to liver samples of animal models. to study the impact of molecules against the anti LPS/ADAM17 pathways and further refine the molecular targets in ER stress, inflammatory and lipogenic factor pathwaysWP5Intestinal microflora has been demonstrated to be causal in obesity , liver insulin resistance type1 diabetes and for the control of inflammation . Axenic mice would constitute a major tool for the demonstration of the causality of intestinal microflora on the control of a physiological function and its molecular origin. The general objective of WP 5 was to validate that intestinal microflora/LPS from human origin is a causal factor of NAFLD and that it can be treated with anti-LPS molecules and drugs against ADAM17 and other targets validated in WP4Hence, the aim of this WP was to set up an animal model of mice bearing a human gut microbiota. This animal model was then dedicated to refine the hypotheses i.e. identification of biomarkers and molecular hypotheses of liver steatosis generated from the human cohorts. We aimed as well to use this animal model to validate the impact of the gut microbiota directly on the development of hepatic steatosis and to assess the role of LPS and new chemical entities on the liver disease.Moreover WP5 served to generate samples from axenic mice to further refine the human metabolic markers and molecular targets.WP6WP 6’s overall objective was to co-ordinate the project providing adequate management to ensure most efficient collaboration, exchange of information and knowledge between all partners, to monitor the project progress and to ensure fulfilment of ethical and biosafety issues.Project Results:WP1The main output form WP1 are the complete and well characterized cohorts. They are constituted as follows:Cohort 1: Plasma, serum, urine and feces (90 from Italy + 98 from Spain, n=188) and liver (75 from Italy + 72 from Spain, n=147, 26% men) samples have been collected in two consecutive visits. Information regarding anthropometry, clinical and biochemical phenotypes has been registered. Complete clinical and biochemical characterization, including an oral glucose tolerance test (OGTT), hepatic ecography, and measurements of the body composition and of intimae media thinness (IMT) were performed in each participant. Intravenous glucose tolerance tests (IVGTT) and/or hyperinsulinemic-euglycemic clamps were also conducted in all participants.Cohort 2: Complete clinical, anthropometrical and biochemical parameters, OGTT, hepatic ecography, measurements of the IMT and plasma, serum, urine and feces samples from 757 (351 Italy + 406 Spain) participants (32% men) have been collected, processed and stored. 75 additional “healthy” volunteers (35% men) with clinical characterization, plasma, serum, urine and feces; OGTT, IMT, hepatic ecography and evaluation of body composition were also recruited in Spain.Faeces samples (319 and 219), urine samples (632 and 1,305), serum (417 and 1,377) and plasma (781 and 80), buffy coat (395) and liver biopsies (107 and 69) have been distributed from Girona and Rome, respectively.The stratification of the participants of cohort 1 from Spain (n=72) was: 29% obese patients without steatosis, 48% obese patients with slight steatosis, 16% obese patients with moderate steatosis, and 7% obese patients with severe steatosis. The stratification of the participants of cohort 1 from Italy (n=75) was: 16% obese patients without steatosis, 21% obese patients with slight steatosis, 29% obese patients with moderate steatosis, 34% obese patients with severe steatosis. NAFLD Activity Score identified in all participants 24% patients at the borderline and 2% with NASH. Lobular activity identified 9.8% samples with <2 focus/field, 2% 2-4 focus/field and 1.3% >4 focus/field. Low ballooning degeneration was identified in 3.3% of samples, and Fibrosis Staging disclosed 12.2% perisinusoidal and 4.1% periportal fibrosis, 1.4% Fibrosis in Bridges and 0.7% patients with Cirrhosis. The stratification of the whole cohort 2 (n=832) was as follows:6% lean patients without steatosis, 1% lean patients with steatosis, 22% obese patients without steatosis, 25% obese patients with slight steatosis, 30% obese patients with moderate steatosis,16% obese patients with severe steatosis.Inclusion criteria were: age 30-60, BMI > 40 Kg/m2, and absence of clinical symptoms and signs of infection in the previous month. Thyroid dysfunction was specifically excluded by biochemical work-up. Other exclusion criteria included: 1) clinically significant neurological, heart failure, autoimmune diseases, or other major systemic diseases, including malignancy, 2) history of drug or alcohol abuse, defined as >80 g/day in men, and >60 g/day in women, or use of medication able to interfere with insulin action, 3) acute major cardiovascular event in the previous 6 months, 4) acute illnesses and current evidence of acute or chronic infective diseases, and 5) mental illness rendering the subjects unable to understand the nature, scope, and possible consequences of the study.Insulin action was measured by the hyperinsulinemic-euglycemic clamp in subjects included in cohort 1, and by means of the Homeostasis Model Assessment for insulin resistance (HOMA-IR) in all participants. Obesity was defined through the body mass index (BMI), waist to hip ratio and body composition (TANITA).Liver biopsies were collected and divided in two parts: one part was processed for histological analyses and the second one was frozen in liquid nitrogen and properly handled for different molecular analyses (transcriptomics, proteomics, metabolomics). Blood, urine and faeces were collected from patients after an overnight fast and frozen at -80ºC.The recruitment of patients was performed after obtaining signed informed consent from previously identified and selected potential participant, all ethical requirements were met.WP2We established in WP2 an atlas of biomolecular profiles in close interaction with WP1 (clinical cohorts and sample generation) and WP3 (data modelling and systems medicine). We profiled the metabolome in 1,665 serum samples and 1,019 urine samples, the transcriptome in 130 liver biopsies, and sequenced 130 fecal 16S rRNA, 88 metagenomes as well as the peptidome in serum and urine.A range of metabolic profiling technologies (NMR spectroscopy and UPLC-MS) were used employed to generate metabolite profiles of plasma, urine, and faecal water samples for comparison with both the proteomic, transcriptomic and metagenomic datasets. We initially used capillary electrophoresis coupled to mass spectrometry (CE-MS), a very sensitive technology developed for the identification of biomarkers in body fluids and is of great interest for protein-profiling based clinical applications. The technology was later on replaced by LC-MS/MS peptidomics, combining high-resolution separation by liquid chromatography with mass spectrometry.Microarray transcriptome analyses of liver transcriptome biopsies were outsourced to a service provider, and a specific microarray pre-processing workflow was developed in R.The fecal microbiome was characterized by sequencing bacterial 16s rRNAs for taxonomy (down to family and genus levels), then sequencing the metagenome for functional and taxonomical analysis through service providers. Two data processing pipelines were developed for 16s rRNA and NGS metagenomics.The FLORINASH database model was designed then implemented and iteratively populated with clinical data, metabolomic, transcriptomic, peptidomic, taxonomic and metagenomic data. Interfaces were implemented for data upload and download.WP2 has focussed on profiling samples, developing specific data workflows and towards the end of the project securing the medium- to long-term legacy of the project:1- Long-term archiving of the FLORINASH samples and datasets2- Archiving of raw profile data of samples analysed by 1H-NMR and UPLC-MS based metabonomics, MS-based peptidomics, microarrays and next generation sequencing 3- Finalisation of data processing pipelines for microarray data, peptidomics data, 16s taxonomic data and NGS metagenomic data4- Curation towards a final version of all datasets5- Design, implementation and population of the FLORINASH database with intermediary dataset, replaced by final datasets at the end of the consortium6- Final data analysis for all datasets: - clinical data - liver biopsy transcriptome - serum MS-based proteomics - serum MS-based lipid profiles - serum NMR-based metabolic profiles - urine NMR-based metabolic profiles - fecal 16S rRNA microbial phylogenetic profiles - NGS metagenomesThe FLORINASH database will be maintained after the end of FLORINASH in order to allow to fully exploit its potential for generation of further hypothesis or their validation. WP3WP3 focused on data modelling and systems medicine and as a consequence works in close interaction with WP1 (clinical cohorts and sample generation) and WP2 (omics generation and data-basing).For the last period of activity, in order to finalise the database and keep it usable also after the end of FLORINASH, WP3 has focussed on running final pre-processing pipelines on all existing spectral metabolomic data and other datasets to prepare integration of final pre-processed datasets in the FLORINASH database by WP2 on: - clinical data - liver biopsy transcriptome - serum MS-based proteomics - serum MS-based lipid profiles - serum NMR-based metabolic profiles - urine NMR-based metabolic profiles - fecal 16S rRNA microbial phylogenetic profiles - NGS metagenomesNovel data analysis pipelines were developed for microarray, 16S rRNA and NGS metagenomic data.WP3 was in charge of integration of clinical and the various omics data, statistical analyses, and other mathematical modelling. Clinical data generated in WP1 is being used to annotate and probe omic profiles obtained from WP2, and is establishing the correlation between patient molecular profile data generated by the various omics analyses (WP2), and severity of the disease as measured by WP1. The analysis of biopsy transcriptomic data by WP3 has generated gene expression profiles associated with the development of NAFLD, derived from a subset of patients. Hypotheses are being further tested by integrative analysis using gene networks, gut microbial profiles and extensive metabolic and clinical data. Hypotheses built by mathematical and statistical approaches were further investigated by WP3 in seeking to prioritize pathways and targets for studies by WP4 and WP5. WP3 has then challenged the original hypotheses on the development of the disease against clinical data from patients from WP1 human data for Cohort 2, and from some animal-derived data. New Chemical Entities (NCEs) were screened against NASH target genes such as Adam-17.Processing of 16S data from faecal samples has seen a significant improvement in the characterization of the microbial ecology of cohort 1 patients, taking a considerate approach to minimize batch effects. DNA extracted from faecal samples obtained from the patients was submitted for microbiota analysis. The 16S rRNA genes of the bacterial component of the faecal microbiota were amplified and sequenced by Research and Testing Laboratory (Lubbock, TX, USA) using bacterial-encoded FLX amplicon pyrosequencing, except the eubacterial primers 28F (5′-GAGTTTGATCNTGGCTCAG-3′) and 519R (5′-GTNTTACNGCGGCKGCTG-3′) were used to amplify the V1–V3 region of the 16S rRNA genes. The sequence data were supplied by the Research and Testing Laboratory in a processed format. Therefore, to produce results using the QIIME pipeline, each patient’s dataset had to be treated as a single run and the fasta files concatenated. For quality filtering and barcode assignment, the following criteria were used: no ambiguous bases, no mismatches with the forward primer, read-lengths not shorter than 250 bp or not longer than 450 bp and the average quality score in a sliding window of 50 bp not to fall below 25. Denoising and chimera-checking (de novo and reference-based) were done using UCHIME. Amplicon reads meeting the filtering criteria were clustered into operational taxonomic units (OTUs) at 97 % similarity. PyNAST was used to align representative (i.e. most abundant) sequences for each OTU. The RDP classifier was retrained on the GreenGenes 13_5 database to make taxonomic assignments to the dataset with the threshold confidence value set at 80 %. Rarefaction was performed on the OTU table so that the number of reads compared per sample was identical. In addition, a new workflow was developed for pre-processing and analysis of full gut microbial metagenomes, which yielded high-quality data. The metagenomic analysis pipeline was established. The tools and methods used during each stage of the process were compared with the current state of the art, and where algorithms could be demonstrated to offer improvements over the published methodology, the pipeline was adapted accordingly. Following initial quality-control and trimming, sequence reads are then filtered through a process allowing the separation of reads aligning to any appropriate database i.e. for removal of human sequence, or separation of reads into different taxonomic groups. Remaining unfiltered reads then undergo taxonomic classification. Sequence reads were aligned against a targeted database of genes capable of providing taxonomic discrimination with much lower computational cost. A de-novo assembly of the reads from each sample was then carried out, provided considerably improved contig lengths and amount of assembled sequence, with a reduction in chimeric contigs. Genes within these contigs were then predicted from each of these assemblies and used to construct a non-redundant gene catalog. The abundance of these was measured by aligning reads from each sample against a database of the centroid sequences from each cluster, while functional classification of the predictions has been carried out by alignment of the genes against the KEGG genes database allowing inference of the Kegg pathway and associated GO terms for each gene. A secondary analysis has been carried out to identify functional domains within the genes, and this has been demonstrated to be able to assign GO terms to ~30% more genes than by alignment to KEGG genes, while in many cases providing more specific functional classifications using a more robust methodology.WP4One of our main efforts was to develop pertinent animal models for the study of NAFLD. To this end, we worked to set up both nutritional and genetic mice models of NAFLD for which liver damage would be comparable to those present in the human biopsies of the FLORINASH cohort. All the models were deeply characterized by WP4 partners: metabolically by adequate methods: weight progression, percentage of lean versus fat mass by dexa, oral glucose tolerance test, insulin sensitivity test, euglycemic hyperinsulinemic clamps, and histologically by the characterization of simple steatosis (Oil Red O and measurement of hepatic triglycerides), or NASH (inflammation and fibrosis have been assessed by adequate staining methods). Moreover, we also measured and quantified on liver samples the expression of genes related to ER stress, lipogenesis and inflammation by RT-qPCR and by Western blotting. Urine and faeces were also collected for analyses. Nutritional models in WP4We first developed nutritional models of NAFLD by feeding mice with a fat-enriched diet (HF diet) containing 60% of saturated fat. We also set up a colony of genetic obese ob/ob mice, the gold standard model for the study of hepatic steatosis. As expected these two mice models developed a profound hepatic steatosis, an insulin resistance but a faint inflammation. When the NAS score of hepatic human biopsies of the FLORINASH cohort was analysed, it revealed that a vast majority of patients had a simple hepatic steatosis but some of them (10 to 15 %) also presented NASH. Mice models that had been developed at this stage were pure models of steatosis (HF diet, ob/ob mice) but not of NASH. We therefore decided to acquire a mouse model of NASH. This was made by developing a model of mice fed with a Methionine Deficient diet (MD diet). After extensive characterization of the model, we decided to abandon it. Indeed, MD deficient mice had indeed liver inflammation but hepatic fat accumulation was very low. Moreover, these mice lost weight, which was very far from the obese phenotype of the human patients. It was then decided to develop a new mouse model of high fat fed mice but perfused for one month with LPS with an attempt to increase the low-grade inflammation. We classified this model as a NASH model since it presented features of steatosis (evidenced on liver sections after Oil Red O staining and measurement of triglycerides (TG), of inflammation (measurement of inflammation markers by qRT-PCR). Moreover, lipogenic pathway and ER stress markers were also found to be increased in liver samples from HFD infused with LPS mice. Finally, since some of the patients of the FLORINASH cohort displayed an hepatic fibrosis, a last effort was made to develop a nutritional model of fibrosis in order to cover all the spectrum of liver damages observed in our human cohort. A last nutritional protocol was started during the last year of the contract according to a protocol described in the literature (Gut Jun 10. pii: gutjnl-2014-306748). Global analysis of the phenotype revealed clear signs of steatosis, inflammation and some signs of hepatic fibrosis. In conclusion, the nutritional models of NAFLD that were developed by the consortium, now covered all the hepatic characteristics of the biopsies of the human FLORINASH cohort. We hope that these models will be helpful to refine the molecular hits emerging from the large-scale analysis. Genetic models in WP4We also generated genetic mice models of NAFLD in which the hepatic TIMP3/ADAM17 pathway were manipulated. ADAM17 is involved in the processing of tumor necrosis factor alpha (TNF-α) and thus expected to play a major role in liver inflammation that is characteristic of NASH. TIMP3, (tissue Inhibitor of Metalloproteinase 3) is an endogenous inhibitor of ADAM17. In order to decipher ADAM17/TIMP3 function in the liver, mice invalidated for ADAM17 in hepatocyte (ADAM17 floxed mice crossed with albumin Cre mice), in macrophages (ADAM17 floxed mice crossed with LysM Cre mice), or overexpressing TIMP3 in hepatocytes (AlbTIMP3) have been generated. All these genetic models have been challenged with a high fat diet or a MCD diet to trigger fat accumulation and/or liver inflammation.The metabolic and molecular analysis of the macrophage ADAM17 KO mice revealed interesting findings since these animals displayed a significant improved glucose tolerance on a high-fat diet and presented decreased lipogenic and inflammatory genes in their livers. Liver ADAM17 KO mice showed only an improved glucose tolerance compared to controls but also showed a reduced hepatic steatosis. AlbTIMP3 had improved glucose tolerance and insulin sensitivity on high fat diet with reduced hepatic steatosis and transaminases levels. The phenotypes of the ADAM17/TIMP3 genetic mice clearly demonstrated the key role of this pathway in NAFLD development and its associated- insulin resistance. Another challenge of the WP4 was to identify new pathways or proteins identified as dysregulated in the human studies and to validate these hits in accurate animal models. This was done with success for three proteins/markers: the branched chain amino acids (BCAAs), the hepatic detoxification pathway and a member of the proteasome system (Ubiquitin Ligase). Briefly, it was shown from the FLORINASH cohort that the circulating concentration of BCAAs correlated with the degree of hepatic steatosis. It was further demonstrated that BCAAs were strong activators of the lipogenic transcription factor SREBP-1c and thus could participate in the activation of the lipogenic pathway and the subsequent hepatic steatosis. One partner obtained interesting data in mice showing that the detoxification pathway could participate to the transition from simple steatosis to NASH. He analyzed the expression of several proteins related to detoxification (the transcription factor CAR (Constitutive Androstane Receptor) and other proteins involved in the bile acid and bilirubin detoxification pathway) in the liver of the nutritional models and in the human cohort and found that these effectors were inversely correlated with the severity of NASH. Finally, among the genes downregulated in the human NAFLD samples, the expression of the Ubiquitin Ligase was reduced specifically in those with glucose intolerance. It was shown that mice KO for this Ubiquitine Ligase fed a Methionine Choline Deficient diet revealed NASH-like features in the liver. WP5The main results are related to the fact that colonization of germ free mice with the gut microbiota from either high or low NASH score patients did not lead to a fatty liver disease. This was also true when the colonized mice were even fed a fat-enriched diet to further trigger the accumulation of fat into the liver. Experiments are still ongoing to identify into the urine and the blood whether the two groups of colonized mice are characterized by different biomarker profiles.The outcome of this experiment showed that no change in the liver disease was observed in all experimental conditions suggesting that the gut microbiota from human could not directly induce the liver disease in mice. This animal model might thus not be suitable for this purpose. Biomarkers from fluid are still currently under analyses and might confirm some differences observed from the human cohort.Altogether, the gut microbiota seems not to be sufficient enough to induce the liver disease in normal healthy mice. This might indicate that other, molecular or environmental issues are probably required to favour the role of gut microbiota on the development of hepatic steatosis.The database of FLORINASH can be seen as one of the main important outputs of the project as it contains a unique longitudinal cohort of obese patients with OMICS and clinical data at baseline, putting all efforts and work from each partner on one single platform. During the lifetime of the project it was not possible to fully exploit the potential of its data. The consortium partners have thus decided to maintain the database up and running beyond the end of FLORINASH allowing further research. Further important output of the FLORINASH consortium is the upcoming application for at least two patents (Partner 1, WP1), the set up of pipelines for the processing of 16S, metagenomics and microarray data (Partner 4, WP3) as well as further promising results who might lead to patents at a later stage, currently needing more investigation.Potential Impact:As mentioned above the clinical diagnosis of NAFLD normally requires liver biopsies for the histological analysis which is obviously invasive and represents a risk for the patient. Therefore, there is a strong need to identify non invasive predictive and diagnostic markers of hepatic damage. We suggested that such markers could originate from intestinal microflora and be detected in urine, plasma and faecal water. Intestinal microflora is a causal mechanism of insulin resistance and obesity and hence is strongly associated with NAFLD. At the end of FLORINASH no final biomarkers could be validated in the last state for reaching this objective. Nevertheless the results of FLORINASH have the potential to lead in the future to new diagnostic and therapeutic approaches which could similarly address insulin resistance and other metabolic and vascular complications whose origin also lies in inflammation. The further exploitation of the database constitutes an important starting point and partners will continue to finalize their studies on the ongoing biomarkers/targets that had been identified in the project. The impact potential of the outcome of each WP are the following:WP1Constitution of well characterized human cohorts and collection of biological samples (urine, faeces, faecal waters, blood, plasma, serum and biopsies) is needed for the identification of biomarkers through which we may define molecular causative targets in hepatic and metabolic disease. Collected data and samples from more than one thousand participants in this study will serve to test and generate a great set of metabolic biomarkers suitable for the prediction of NAFLD in this context. Such parameters will be defined as new diagnosis tools and (potentially) therapeutic targets. WP2&WP3Obtaining a detailed molecular profiling of the biological processes involved in NAFLD and NASH impacts significantly upon the potential for diagnosis and thus for preventive treatment and enhance our in-depth understanding of these multi-factorial conditions. The project-derived profiles and information represent a unique clinical and biological resource based on the FLORINASH cohorts on metabolic (derived from urine / serum / faeces / clinical measures), hepatic gene expression variations, proteomic (derived from plasma and urine), taxonomic and metagenomic markers associated with disease state. In particular, alterations in the gut microbial ecological structure are heavily involved with metabolic processes, assessed by 1H NMR and UPLC-MS. Further associated molecular processes – such as hepatic genetic regulation revealed by transcriptomic analysis of liver biopsies obtained from the Cohort 1 patients going to bariatric surgery – were also probed by MS-based peptidomics. The overall data acquisition programme has achieved its objectives, yielding significant results for integrative analysis in WP3, and interpretation is being further refined and expanded.Using workflows and strategies developed in the project, we identified biomolecular risk profiles made of metabolites, peptides and bacterial taxa. These bio molecular risk profiles were used to define hypotheses related to the role of the gut microbiome in modulating clinical parameters related to fatty liver disease, and investigated in animal models.A new candidate gene associated with non-alcoholic steatosis, steatohepatitis and progression towards hepatocellular carcinoma was also identified and validated by in vivo experiments in gene knock-out experiments.The pipelines for processing of 16S , metagenomics and microarray data constitute an important tool for further studies and thus may impact on further research results.WP4We have identified a certain number of proteins and pathways of which expression or activity are dysregulated in the livers of NAFLD mice models but also in the human biopsies. These proteins could thus constitute targets that will be interesting to modulate in the future. The relevance of these targets for the treatment of NAFLD could be addressed by using chemical compounds able to interfere with them.Hepatic steatosis is a major co-morbidity of obese patients. Scoring its intensity is for far inaccurate and understanding its molecular origin both remain important clinical needs not reached through liver biopsy. We here took advantage of a very precisely phenotyped cohort FLORINASH with a wide range of scores of steatosis from Spain and Italy and developed an original and unique method of nanoHPLC-chip-MS on plasma. We identified 87 proteins which precisely identifies low (0-1) and high (3-4) scores of NAFLD. From these a subset of 10 proteins conserved between both countries and which resist batch analyses were up or down regulated in NAFLD. Others mostly included in the complement pathway were complementing the profile of biomarkers required to accurately stage NAFLD. Altogether, plasma biomarkers are suitable tools to diagnose hepatic steatosis and provide some hints regarding putative pharmacological strategies to hamper the progression of the diseaseWP5The axenic mouse model that had been set up in WP 5 did not lead to the conclusion that gut microbiota from human could directly induce the liver disease in mice.This animal model might not be suitable for this purpose. Biomarkers from fluid are currently under analyses and might confirm some differences observed from the human cohort.The gut microbiota seems not to be sufficient to induce the liver disease in normal healthy mice. Other, molecular or environmental issues are probably required to favour the role of gut microbiota on the development of hepatic steatosis. Further research on that point will be needed.Use and dissemination of foreground:During the project lifetime partners participated to congresses, external workshops and other scientific events , actively disseminating and discussing the results generated by the project with peer colleagues.One public workshop had been organised by Partner 1:Remy Burcelin (INSERM), the coordinator of the project, participated for FLORINASH at a general public oriented conference workshop held on November 28th 2013 in Paris (and Toulouse via video) on the topic of “Diabetes- possible therapies in the future”. This event is part of a series of conferences called “Santé en questions” dedicated to the general public and regularly organized by INSERM, Universcience and local stakeholders in science and techniques giving the general public the possibility to learn about the newest advances and results in life sciences on the topic of health and discuss with scientists and doctors.Numerous publications in peer reviewed scientific journals based on the work and acknowledging FLORINASH have been published throughout the project lifetime. Currently writings are ongoing on further articles which will be published or submitted in the upcoming months by the partners.A website for FLORINASH had been set up at the beginning of the project, providing useful information on the project’s objectives, strategy and expected outcomes as well as contact details. The website had been updated after the first two years with a sum up of the first two periods of work.The foreground generated by FLORINASH project has been carefully monitored during the project with regular surveys among all partners at each reporting period.The exploitation of the foreground generated by the project is laid down in Del 6.16 “final exploitation plan”. Results of the project from Partner 1 will give raise to the application of two patents will be organised by Partner 6 as being the official mandatory for the valorisation and exploitation of research results of INSERM.As detailed above the database of FLORINASH can be seen as one of the main important outputs of the consortium partners have thus decided to maintain the database up and running beyond the end of FLORINASH allowing further research as well as further updates on data and results if necessary. Negotiation concerning the maintenance costs and a legal agreement framing the use and access to the database between the partners are still currently ongoing. A further possible development of the database would be a recall of the included patients from cohort I and cohort II, in order to include their updated clinical data to allow further validation of the research results. To achieve that, partners are actively working on receiving new grants to continue the further development of the database, one application to a new grant had already been submitted in October 2014.The consortium partners will continue beyond the project’s lifetime to exploit the results contained in the database in order to further progress gaining knowledge on the development of NAFLD and to validate and find new diagnostic and therapeutic tools for this disease.List of Websites:http://www.florinash.org/Pr Rémy Burcelin, PhD, Directeur de Recherche Inserm U 1048, Equipe 2, « Facteurs de Risque intestinaux, diabète, Dyslipidémie » Bt L4, Hôpital Rangueil, BP84225, 31432, Toulouse, France www.diabtoul.eu www.glp1.eu
