Skip to main content

Molecular Targets Open for Regulation by the gut flora – New Avenues for improved Diet to Optimize European health

Final Report Summary - TORNADO (Molecular Targets Open for Regulation by the gut flora – New Avenues for improved Diet to Optimize European health)

Executive Summary:
TORNADO has spearheaded the host-microbiome research in Europe whilst having received worldwide recognition for its efforts. Its achievements are ground-breaking and yet the project began as an exploratory journey with the aim to obtain the first comprehensive glimpse of the multicomplex interactions between microbes and its host. The development of TORNADO´s infrastructure has been designed for a long-term commitment to decipher mechanisms and metabolites underlying microbe-host interactions at the level of the whole organism. This work will pave the way for the inevitable transition from molecular medicine to systems medicine in modern health care.

Selected accomplishments:

TORNADO has shown that gut microbiota can influence CNS and ENS development and function, placenta development and Blood Brain Barrier function. These results indicate that maternal microbes/metabolites and/or the colonizing microbiome in the newborn offspring can profoundly influence postnatal behaviour and cognitive function later in life.

TORNADO has shown that microbiota can influence energy metabolism and subsequently function in organs like liver, skeletal muscle and small intestine. These observations have great implications for how gut microbes can be involved in energy management within the host, be it under normal conditions or when the host encounters “energy crisis” prompted by obesity and associated illnesses. The observation that gut microbes also can influence metabolic circuits, by modulating our clock genes in the liver, further underscores the intimate relationship between microbes and their host. Microbes may thus influence eating behaviour and tune physiological functions such as hunger and satiety. TORNADO has also demonstrated a crosstalk between food intake, gut microbiome and exercise. These findings are relevant for the formulation of new guidelines to sustain health, especially among the elderly. Additional studies in early life also show that maternal microbes can influence PPAR-alpha and -gamma function, two key nuclear receptors controlling hepatic energy metabolism, food intake and locomotion early in postnatal life. Thus, microbes are setting metabolic thresholds all throughout life.

TORNADO´s work in the area of microbiology has shown that Lactobacillus casei regulates the expression of key enzymes in the small intestines, regulating cholesterol biosynthesis. These findings shed light on potential pathways between commensal bacteria and metabolic and immune regulation of the host. TORNADO has also developed a novel protocol to grow SFB bacteria ex vivo. This is a very important breakthrough for our attempts to unravel the biological properties of this key immune regulator.

TORNADO´s human cohort studies have identified that the diversity and richness of microbiota is associated with metabolic profiles and appear to be age dependent. Furthermore, microbial components likely underlying ailments such as Celiac Disease are now begining to be teased out.

TORNADO has shown how a concerted interdisciplinary team can generate world-class data with a limited budget. Yet, culture differences in language and understanding between research disciplines such as preclinical research and epidemiology remains a hurdle. Irrespective of these issues, we are optimistic. We anticipate that ground-breaking research, emanating out of the observations reported by TORNADO, will be generated in the years to come. This includes research directed towards metabolism, neurobiology, cardiovascular research, aging, and immune function which are all central for sustained health. TORNADO´s findings are of significant importance with far reaching consequences for the food industry in their attempt to develop tailor made food products for health maintenance. This notably, was one of the major objectives of TORNADO.

Project Context and Objectives:
The European sponsored TORNADO consortium combines a systemic, interdisciplinary and comprehensive mechanistic approach with a focus on complementary skills provided by world leading experts working in close harmony with the food industry. TORNADO will deliver results and know-how that will enable broader applications, faster development, and more solid evidence for health claims based on detailed and thorough in depth assessment of underlying mechanisms relevant to health effects of dietary products compared to traditional approaches.
The main objectives:
i) Increase the knowledge of diet on the gut flora and the impact of gut flora and specific groups of micro-organisms on the immune system and the interaction with other organ systems ii) Increase the cooperation between different scientific disciplines in general and increase and consolidate existing interdisciplinary cooperation among TORNADO partners specifically iii) Provide first class scientific data to substantiate health claims and to contribute information and know-how to the European Commission Health and Consumer Protection Directorate General policy in this area iv) TORNADO will provide concrete deliverables such as Roadmaps to Health, Tailor Made Health Monitoring that will be translated to a user friendly dissemination program to substantially increase impact. Eventually, TORNADO’s ultimate goal, based on our comprehensive and innovative platform, is to develop an infrastructure that will accelerate future design of personalized functional food for specific target groups (age and geography).
TORNADO has a Comprehensive Mechanistic approach
In the scientific community, there has recently been a renaissance of interest in the role of the gastrointestinal microbial community in maintenance of health, rather than in disease. This awakening stems from the accumulated reports and expanding evidence that the gut microbiota not only can impose beneficial effects on local chronic inflammatory processes (like ear, bowel, skin and genitals) but appear also to be able to exert hitherto unsuspected systemic effects, on the liver and adipose tissue, angiogenesis, development of allergic disorders and even on cognitive functions.
The challenge for the scientific community is to open this Pandora’s box and to decipher the pathways and mechanisms utilised by the gut flora to mindset and complement host physiology. Subsequent to this, ways of modulating these pathways by food derivatives or pre- or probiotic components will enable tailor-made diets for specific target groups.
Traditionally this has been rather imprecise – simply studying the association between intake of certain products with health or disease – hampering novel design of intervention products and the development of new strategies for drug design.
In marked contrast, TORNADO’s heavy mechanistic approach will overcome this problem by focussing on the molecular mechanisms in between the bacterial interactions and the outcome. TORNADO’s innovative systemic experimental design “microbe-mouse-man” as well as “man-mouse-microbe” will overcome this problem by assessing biomarkers of interest with comprehensive mechanistic studies to delineate the molecular mechanisms and signalling pathways involved. TORNADO’s unique mechanistic platform will form an infrastructure – further accelerating future design of personalized functional food for specific target groups.

Increasing knowledge through leverage of molecular pathways:
TORNADO’s foremost objective is to increase the knowledge of dietary effects on the gut flora and the impact of gut flora effects on the host. In addition, TORNADO seeks to identify specific groups of micro-organisms or microbial communities which can modulate and shape the innate and adaptive immune system across age as well as influence other organ systems such as placenta, adipocyte tissue, pancreas, liver, spleen, and peripheral and central nervous system. Finally, TORNADO is also collecting information about the composition of the human microbiome at various stages in life and the interactions with other organ systems.
TORNADO’s approach encompasses increasing levels of specificity, from investigations of dietary habits and health in population cohorts, through intervention studies in humans and animals to molecular analyses of the intestine and immune system, liver, adipocyte tissue, pancreas, heart and brain. TORNADO will examine dietary influence on bacteria and immune responding cells and on identified biomarkers. TORNADO can and will take the opposite direction in this chain and test the results obtained in the molecular parts of the project, on animals, followed by individuals in dietary intervention studies and possibly even populations.
TORNADO will deliver data that can be used to (i) recommend biomarkers for evaluating effects of diet or microbes (ii) substantiate or improve health claims of existing products (iii) generate functional food products or improve already existing products (iv) set the stage – through development of regulatory documents – for intervention studies that improve or sustain health among citizens in Europe.
Interdisciplinary skills provided by world leading experts
In order to meet the demanding objectives set forth by the call, the TORNADO management has attracted an interdisciplinary team of some of Europe’s best – and indeed, world leading – immunologists, cellular microbiologists, developmental biologists, endocrinologists, epidemiologists and nutritionists to take part in the TORNADO consortium.
Thus, TORNADO’s strength builds on the proven excellence among the partners, most of whom have been, and are, working in successful collaborations with each other in other constellations. These leading scientists will also ensure access to the very latest unpublished research results. This will be especially important for identifying novel mechanisms and biomarkers for the verification of health claims.
Fast results by leveraging existing infrastructure and networks
TORNADO will immediately generate synergistic effects by building bridges between the European food industry and first class scientists within Europe. The outstanding arsenal of state-of-the-art technologies, access to a large biobank material, high throughput functional genomics, metabolomics, and first class mouse models for validation will ensure delivery of utmost quality in an efficient and cost effective way.

More solid health claims
TORNADO’s combination of a systemic comprehensive mechanistic approach, with an outstanding interdisciplinary con¬sortium will lead to faster and more solid health claims. In addition, the approach itself will lead to in¬creased scientific cooperation between different scientific disciplines. This, and the thorough under¬standing of the molecular mechanisms, will unravel the unexpected, rather than solely testing the obvious.
Eventually, TORNADO’s comprehensive mechanistic approach will form a mechanistic infrastructure that will accelerate future design of personalized functional food for specific target groups.

Overall strategy and general description
The TORNADO project is divided into eight work packages. These work packages can be grouped into four major components:
• Direction and confirmation: WP 1 and 2 will leverage ongoing epidemiological and individual intervention studies to set the direction of analyses of biomarkers. WP 1 and 2 will also be used to confirm biomarkers identified in the comprehensive mechanistic approach described below.
• Comprehensive mechanistic approach: In WP 3 and 4 – the heart of TORNADO – the mechanistic understanding on a molecular level will be developed.
• Translation: WP 5 will transform results from the above work packages into novel strategies for solid health claims.
• Support and analyses: WP 6, 7 and 8 will provide microbiota analysis, management support and coordination and dissemination services to all work packages.




Project Results:
WP1 Epidemiology, population genetics, and metabolome. Participants: 1, 4 (leader), 5, 6, 11, 13, 22
This WP will study variation of the gut microbiota and its association with health outcomes at the population level. Several cohorts have been made available to the consortium, covering all age groups (young, adult and elderly) including twin cohorts and one birth cohort.

Results

Databank of gut microbial composition, serum metabolome, clinical data and dietary records from cohorts (DIPP, Finnish Twin Cohort, EPIC Potsdam, ETHERPATHS study, CHOP cohorts and SIMBA trial) covering infancy and early childhood, adulthood, and old age (Deliverable 1.1).

Biobanks (biofluid and tissue) that may be exploited to assess potential biomarkers (microbiota regulated molecular targets from WP3 and WP4) in the above cohorts by means of population genetics, metagenomics, metabolomics and bioinformatics including SNP analysis have been created (Deliverable 1.4).

Significant correlations between age and bacterial diversity as well as age and lipids and metabolites were found. In addition, significant correlations between bacterial diversity and certain lipids and metabolites were found that may be used to predict the bacterial diversity. (Deliverable 1.2)



WP2 Individual intervention studies/specific groups. Participants: 3, 10 (leader), 13
In this WP, we will focus on intestinal immune responses from birth onwards to understand gut microbiota cross-talk and development of oral tolerance against food and commensals.

Results

Deliverable 2.1. Information on the early events in education of the intestinal immune system to tolerate non-pathogenic microbes and the properties of bacteria promoting tolerance.
The ligand induced aryl hydrocarbon receptor (AhR) is expressed both in the gut lamina propria and in cells of the epithelium. Preliminary data indicate that the expression of AhR may be different before and after birth. The AhR is constitutively expressed by gut epithelium in individuals of all ages. TSLP however appears to be induced by the flora. Signalling through the AhR is pro-inflammatory for epithelial cells and mucosal biopsies but surprisingly enough, had a profound anti-inflammatory effect on dendritic cells.

Deliverable 2.2. Data to show whether manipulation of the microbiota during weaning has lasting effects on the composition of the microbiota and if such effects can be linked to any lasting effects on the immune system, BMI or metabolic markers.
There were no long-term benefits from feeding the probiotic LF19 to infants during weaning on any allergic manifestation, although a larger study population had been needed to fully examine the long-term preventative effects. Other groups have reported a higher frequency of wheeze and respiratory allergies following probiotic supplementation in infancy, although a causal relationship has not been established. Importantly, we noted no adverse effects on objective measures of lung function or airway inflammation. In accordance with the lack of benefit on clinical outcomes of allergic disease, we observed no long-term immune programming effect from feeding LF19 during weaning following polyclonal activation of PBMCs as assessed as cytokine mRNA levels by qRT-PCR iii) Total lactobacilli counts in stool and saliva samples were similar in the control and probiotic group. LF19 was not isolated in any of the collected stool and saliva samples as assessed by RAPD-PCR. In collaboration with professor Lars Engstrand, characterisation of the microbiome was performed by 454-pyroseqeuncing in 10 children in the probiotic and 10 children in the placebo group, with no differences between the two groups. iv) There were no long-lasting effects of LF19 feeding on BMI z-score, body composition, blood lipids or fasting serum-insulin and fasting blood glucose (HOMA-index) or on caries. We also analysed the plasma metabolome in samples collected during the intervention. Interesting results: Feeding infants with LF19 induced significantly lower levels of palmitoleic acid (C16:1) and significantly higher levels of putrescine. Palmitoleic acid is a major monounsaturated fatty acid (MUFA) associated with visceral obesity, while putrescine is a polyamine with importance for gut integrity. Further characterization of the metabolome is planned.
Thus, feeding LF19 during weaning has interesting effects. Yet, we did not observe establishment in oral cavity or gastrointestinal tract. This supports the view that probiotic lactobacilli are transient colonizers even in early life when the microbiota is not fully established. In line with this, LF19 did not confer long-term preventative effects on allergic disease, dental health or development of adaptive immunity, nor any long-lasting effects on body composition, blood lipids, fasting insulin or glucose levels. The effects observed during the intervention with lower levels of palmitoleic acid (C16:1) will be further explored at school age.

Deliverable 2.3. The question will be answered as to whether adhering bacteria influence the immune status of the epithelial compartment and thereby regulate immunity to food constituents.
Adhering CD associated strains of rod-shaped bacteria were isolated from intestinal biopsies. Seven of these isolates were selected for in depth characterization, interaction with human intestinal epithelial cells in vitro and influence on the immune situation in small intestinal mucosa by ex vivo challenge of biopsies from CD patients with inactive disease. One of the isolates is a new species in the new genus Lachnoanaerobaculum. We have published the characterization of this strictly anaerobic, spore forming, filamentous rod named Lachnoanaerobaculum umeaense and the new genus to which it belongs. L. umeaense is interesting in that it is related both to the bacterial strains in Clostridium cluster XIVa, which was shown to drive the maturation of colonic regulatory T cells in mice and the segmented, filamentous bacterium (SFB) shown to drive maturation of intestinal Th17 cells in mice. Possibly L. umeaense is important for the T cell maturation in human gut. Five of the CD associated isolates belong to the Gram (-) genus Prevotella and one isolate is an Actinomyces gravinitii strain. Three of the Prevotella isolates are of a new species. It is the first Prevotella species primarily isolated from human small intestine and hence named Prevotella jejuni. In addition we have one P. melaninogenica and one P. histicola isolate. Whole genome sequencing of the latter two revealed that their genomes are 102.000 and 213.000 bases larger than the respective genomes for these species deposited at the NCBI database. This could mean that these isolates carry pathogenicity islands or other genes that are important for their adaptation to the particular milieu of small intestine. All five Prevotella isolates bind strongly to intestinal epithelial cells in vitro at 37°C suggesting that they are adherent in vivo and hence could influence the epithelial function. The new Prevotella species seems to release toxic substances, e.g. polycyclic aromatic hydrocarbons, since the isolates of this species caused increased permeability in the in vitro model of intestinal epithelium and strong up-regulation of Cytochrome P450 superfamily of enzymes CYP1A1 and CYP1B1, i.e. target genes of the aryl hydrocarbon receptor (AhR), in the epithelial cells. Thus, the Prevotella isolates appear to disturb the epithelium indicating that certain Prevotella species, present in our normal flora, could be risk factors for CD.

Exposing biopsies of CD patients with inactive disease, against a mix of all seven isolates of CD associated bacteria ex vivo and alone and together with gluten, resulted in elevation of the cytokine IL-17A as well as further enhancing the magnitude of IL-17A and IL-10 responses when biopsies were challenged with gluten. These results further support the notion that these bacteria influence the immune situation in the gut mucosa and could be risk factors for CD.

We performed gene expression analysis at the mRNA level in intestinal epithelial cells (iECs) and intraepithelial lymphocytes (IELs) from CD patients and controls. Genome wide bead array analyses (≈24.000 genes) on individual samples were possible after optimization of isolation methods for iECs and IELs and of RNA extraction protocols. The preliminary results are briefly described in Veronika Sjöberg's thesis (reference 9). Presently we are focusing our analysis on the results from studies of iECs. Together with statistician professor Hans Stenlund we have established a strategy for identification of genes with statistically significant changes in expression levels. From individual analysis of iECs of 4 CD patients and 4 control patients we could identify 324 genes that are significantly upregulated and 387 genes that are significantly down-regulated among the 10,556 genes that were expressed in iECs. Several interesting observation have been made already. The iECs have increased expression of enzymes for glycosylation. Of special interest is the increase in a fucosylsynthase, which is in line with our previous finding that CD patients have more L-fucose at the apical side of the epithelium and in goblet cells than controls. Previous studies in monocolonized mice reported induction of fucosylsynthase, suggesting that there might be a connection between the gut microbiota in the patients and expression of L-fucose on the cell surface.

Deliverable 2.4. Information on the effect of dietary alteration on intestinal microbiota and obesity markers in a group of healthy men and women (20 – 60 years).
A dietary crossover intervention study, including in total 20 healthy adults, started in October 2010 and was finalized June 2011. The participants had two intervention periods of 3 weeks each, one with a diet rich in whole grain products and low in red meat (WG intervention) and a second rich in red meat but with no whole grain products (RM intervention). Anthropometric measures, a blood sample for clinical parameters, e.g. blood lipids and a stool sample for gut microbiota analysis were collected from each study individual before and after each intervention.
Results: Body weight, BMI, and body fat mass were statistically significantly lower as compared to baseline while no changes in anthropometric parameters were observed after the RM intervention. The RM intervention was associated with a significant increase in serum creatinine and uric acid and both diets induced changes in microbiota judged from DGGE band patterns (determined at P14).
Conclusion: It is possible to obtain an associatation between intervention and intestinal microbiota activivity by changing diet. Sequencing methods revealed Clostridium sp. to be changed by RM intervention. This bacterial group belongs to the Firmicutes phylum and decreased after the meat intervention. However, we conclude that no harmful effects were observed, which is affirmed by the results regarding anthropometric and blood parameters, which have not been influenced by the change in diet.
WG intervention resulted in a higher occurrence of Collinsella aerofaciens, a bacterial species from the phylum Actinobacteria that is often associated to beneficial traits. It can be found more often in healthy controls compared to Crohn’s disease patients and it is negatively associated to symptoms of irritable bowel syndrome. Its elevated appearance after WG intervention can be regarded as a beneficial effect of the increased consumption of dietary fibre from WG products.

Deliverable 2.5. Data on the effect of aging on immune- and epithelial cell protein expression in humans. 40 duodenum/jejunum samples were obtained spanning across ages as follows. They comprised 2 newborn, 4 children aged 1-2, 10 children aged 2-15, 9 samples from individuals aged 15-45, 5 from aged 45-70 and 10 from 70-86. While TSLP was absent from the epithelium in newborns, tissues from all other ages showed uniformally high expression of TSLP. Thus TSLP appears to be induced in humans. AhR expression was observed both in newborns and in adults. Amendments at the end of period 2 allowed us only to assess composition of gut microbiota and possible changes from earliest infancy to the elderly. The specimens from the newborns showed that the upper bowel microbiota contained a flora dominated by proteobacteria probably from the delivery room; the same specimens showed reduced TSLP epithelial expression (D2.7). If the flora did induce TSLP then perhaps at 1 day and 2 days insufficient time had elapsed for TSLP expression to be upregulated. Alternatively, bacterial numbers may be too low immediately after birth and expression could be induced by the lactobacilli and bifidobacteria which increase hugely on breast or formula milk. It is well known that the microbiota of the colon changes with aging. However in this work bacterial communities from children, adults, and elderly individuals all showed the same pattern with a microbiota dominated by firmicutes and bacteroidetes, as seen in the colon. It is is not clear if the sequences we detected came from live bacteria since generally it is considered that the upper bowel only contains small numbers of culturable bacteria but is exposed to massive amounts of bacteria through food.


Deliverable 2.6. Molecular targets (regulated by gut microbiota) that can be tested in animal & cell models in WP3 and 4 and be the basis for improved diet to optimize European health.
Results for this deliverable were expected from task 2.4 i.e. the diet cross-over study. Indeed, two metabolites in serum, protocatechuic acid (PCA) and 4-hydroxybutanoic acid or γ-hydroxybutyric acid (GHB) were increased after intervention with the diet rich in whole grain products and low amounts of red meat (WG), an intervention that also was associated with a leaner body composition and changes in microbiota composition, all findings that need to be confirmed in a larger study. Task 2.2 i.e. intervention by feeding probiotics during weaning, suggests that the serum metabolite palmitoleic acid (C16:1) could be a potential target Furthermore, the studies in task 2.3 on gut microbiota associated with celiac disease showed that the Prevotella isolates from small intestinal biopsies induced the detoxification enzymes CYP1A1 and CYP1B1 in the in vitro model for intestinal epithelium. Hence CYP1A1 and CYP1B1 are potential molecular targets regulated by gut microbiota that might be tested in animal models. Finally, results from task 2.1 suggest that TSLP could be induced by the gut flora. However, this finding needs to be confirmed by experiments in which germ-free mice are reconstituted with a gut flora.





WP3 Animal models for the analysis of organ system responses to gut flora Participants: 1, 7, 8 (leader), 9, 18, 19, 20
This work package will study communication between intestinal microbiota and organs including adipose tissue, intestinal mucosa, spleen, liver, exocrine pancreas, lung, brain, HPA organs and skeletal muscle. Two major approaches are outlined here:
i) Mouse genetics for identification of genes affected by gut flora through the use of animal models. Knock out (KO) lines such as TLRs, obese/obese, PPARs, ApoE etc are already available in-house and many also derived under GF conditions.
ii) Use of GF-mouse lines in which defined floras will be introduced, so-called recolonization, to be used for critical validation and further exploration of host-microbe interactions.







Results

The PPAR gene family (P8)
Effects on bacterial flora of PPAR knockout models (D3.4 ; month 36).
Muscle and PPARβ/δ
The objective was to elucidate how exercise training could impact metabolic adaptation in skeletal muscle through modulation of gut microbiota and if these modulations are mediated by PPARβ/δ in skeletal muscles. To do so, P8 studied skeletal muscle adaptations in mice trained - or not - for endurance exercise for 4 weeks by performing biochemical, physiological and microbiota analyses in both wild type and transgenic mice in which PPARβ/δ is selectively ablated in skeletal muscles (PPARβ/δ skmKO). P8 performed both glucose tolerance and insulin tolerance tests post-training and observed that exercise training improved the overall glucose homeostasis of PPARβ/δ skmKO mice in comparison to their sedentary counterpart. While whole-body glucose disposal did not show any significant difference compared to WT mice, exercise training improved insulin-sensitivity significantly in KO mice in comparison to sedentary WT mice. P8 also confirmed PGC1-α, a molecular marker of mitochondrial biogenesis and function to be induced after exercise, though remained significantly downregulated in PPARβ/δ skmKO mice post-training. Further, the fecal microbial profiles were found to be considerably different among the sedentary and the trained mice. P8, in collaboration with P14 observed phylum-level shifts in the gut microbiota composition after exercise training especially for Actinobacteria. We also analyzed the corresponding families and genera in more detail and observed significant increase of Actinomycetales post-exercise training Firmicutes and Bacteroidetes are the most predominant phyla. No significant alterations were observed in the Bacteroidetes, but alteration in bacterial classes like Bacili, particularly Bacillales (belonging to the phylum Firmicutes) was increased in the trained mice.
The fecal microbiota analyses from sedentary and trained mice thus indicate alterations in the microbiota composition in response to exercise. Whether modulation of microbiota in exercised condition is a cause or a consequence of the overall improved metabolic profile of the host remains unanswered. This was further addressed by observing the phenotypic changes in the skeletal muscles of germ-free (GF) mice through gene expression analyses. Mitochondria play a central role in the processes involved in mediating beneficial effects of physical activity on skeletal muscle. Preliminary observations, obtained in collaboration with P1, from gene expression analysis of young GF mice muscles reveal decreased gene expression levels of various mitochondrial factors involved in oxidative phosphorylation (COX5a, COX7b, Cytc), and mitochondrial biogenesis (PGC1a, Nrf1, Nrf2, Tfam) as well as decreased levels of myogenic markers (Myod, Myogenin, Pax7, myostatin) indicating altered mitochondrial function and myogenic capacity of GF mice. Furthermore, we also checked the levels of expression of the enzyme creatine kinase (CK). CK interconverts Phospho Creatine, ATP, ADP, and creatine (Cr) in order to generate ATP that is necessary to meet up energy demands at the onset of exercise. The level of CK is significantly low in GF muscle compared to the SPF indicating that these mice might fatigue faster when exercised. Experiments are in progress to observe the performance of the GF mice in response to exercise test. Enzymes involved in the glycolytic pathway in skeletal muscles, like, glucokinase, lactate dehydrogenase was also found to be decreased in GF muscles. In order to confirm if these changes are in response to the gut microbiota, further studies are in progress where GF mice will be conventionalized with fecal transplant from SPF mice.

To unveil how gut microbiota can affect skeletal muscle functions and if through PPARβ/δ, germ-free derivation of transgenic in which PPARβ/δ is selectively ablated in skeletal muscles (PPARβ/δ skmKO) was attempted at Karolinska Institute (P1), though unsuccessful. Matings were performed, the transfer of PPARβ/δ skmKO pups to germ-free foster mother was successful, but the paps did not survie under germfree conditions for unknown reasons. After repeated attempts, without success, it was not continued further. Recently, new attempts have been made and we have now obtained PPARβ GF mice.

Effects on bacterial flora of PPAR knockout models under stress conditions (inflammation; infections; high fat diets, calorie restriction) (D3.5 ; month 48).
Intestine and PPARγ
Nuclear receptor PPARγ, mostly know as master regulator of adipogenesis, has repetitively been proven to show anti-inflammatory and anticancerogenic properties in the intestine. It has also been reported to regulate defensins production and interact with gut microbiota. However its impact on intestine flora composition has never been presented. Working with PPARγ intestine epithelium specific knockout mouse (PPARγ KO) we realized that the mutant mice do not exhibit any phenotype in basic conditions. Though, we found that caloric restriction (CR) stress triggers differences between PPARγ KO and WT mice concerning body composition, physical activity and gene expression connected with lipid metabolism in intestine. As CR is known to diminish basic inflammation level in whole body and to influence microbiota composition, we decided to assay this issue in our mouse model. We observed that CR affects antimicrobial and antiviral factors mRNA level and that CR PPARγ KO mice show impaired regulation of these peptides expression (Reg3b, Reg3g, NOS2, Oas1a, Myd88) in duodenal epithelium. However, we did not observe differences between WT and KO in inflammatory factors (iNOS, STAT1, TRL1) expression. Contrary, in colon we found some of inflammatory genes being affected but not many of antibacterial peptides. To assay whether the observed gene expression changes are microbiota-dependent we recently transplanted intestine flora from WT and KO CR mice to at libitum WT and KO mice. Results of this experiment are currently being analyzed. Also, to verify the influence of gut flora on PPARγ and its target genes we performed CR experiments using germ free PPARγ KO. Harvested samples analysis is undergoing.
We are at present performing microbiom analysis with P14 to assay possible bacteria composition changes in PPARγ KO mice in CR situation and in basic conditions. Our observation of strong differences between duodenum and colon upon exposure to CR suggests that response to microbes and, as a consequence, microbiota content varies in these tissues. For that reason we are assaying bacterial compositions in duodenum lumen as well as in feces. To assay intestinal bacteria metabolic activity we are currently working on short chain fatty acid (SCFA) composition analysis in small intestine and cecum of WT and KO in CR and at libitum conditions.
Except caloric restriction PPARγ KO mice were submitted high fat diet and sucrose-free diet. So far we found that sucrose-free diet triggers similar physiological phenotype as caloric restriction. However, intestine of these mice has not been studied in the context of inflammation and bacterial response yet. Feces for microbiota analysis were collected yet the analysis is still to be performed.

‘Road map’ (gut microbiota regulated gene lists) which can be used by industry to monitor health promoting effects of their products in animals and man (D3.27; month 48).

The results of biomarker exploration is presented in D5.1 and encompasses specific target genes, changes in lipid and other identifiable biochemical metabolites as well as alterations in the composition of the flora itself. These changes may be followed in selected patient cohorts or animal models in a stepwise manner.
In D3.27 we provide a scheme that presents at least 3 phases, which if required may be further sub-divided to accommodate methodological changes and material collection, primarily in Phase 1. The formulation of the roadmap is built on existing animal models, metagenomic as well as genetic analyses and not least on metabolic analyses. This second phase of roadmap implementation will require comprehensive systems for database analyses that can be collated to provide better and more reliable health monitor readouts, presented in Phase 3.





Microbial-induced signalling (P7)
During the last period, P7 has concentrated on determining the cellular origin of Tlr signaling (D3.18).

The innate immune system can contribute to obesity and metabolic disease. During TORNADO we have found that the adaptor molecule MyD88, which is required for signalling from most Toll-like receptors (Tlrs), contribute to diet-induced obesity. By using bone marrow chimeras in conjunction with tissue specific knockouts he has addressed the importance of myeloid or somatic Tlr-signaling is responsible for the obesity development.

To investigate whether myeloid MyD88-signaling is responsible for diet-induced obesity and impaired glucose metabolism, he used bone marrow chimeras and second LysM driven macrophage specific Myd88-deficient mice. The mice were put on high-fat diet and weight gain was monitored for 12 weeks, when glucose homeostasis was determined.

He did not find any differences in body weight gain between wild-type mice receiving bone marrow from wild-type or Myd88¬-deficient bone marrow (n=6/group) or in glucose homeostasis. We could not perform the reciprocal experiment since Myd88-defience recipients died upon irradiation. Similarly, we did not find any differences in body weight gain or glucose metabolism in mice with macrophage specific deletion of MyD88.

Based on these results he concluded that macrophage derived MyD88 is not responsible for diet-induced obesity.




WP4 Commensals-Immunity-Obesity (Host-microbe) Participants: 1, 2 (leader), 8, 12, 16, 17
The mechanisms by which bacteria (i) establish commensal communities by resisting to, and/or subverting the barrier effect of already established microorganisms (ii) manipulate the physiology of the intestinal epithelium, thus affecting the homeostasis and epithelial properties controlling the host immunological responses and its nutritional status will be addressed in WP4.


Results

D4.2. Identification of intestinal bacteria stimulating the host T cell response. SFB and collaboration with Storkbio. We have established in vitro growth conditions of SFB. Our approach has been facilitated by the sequence and annotation of the murine SFB genome that showed SFB to be a highly auxotrophic commensal. We have adapted the method of purifying IOs to look for outgrowth into filaments as a score for in vitro growth. Using media, we now obtain a consistant and often considerable filamentous growth, as well as differentiation. In vitro production of IOs allowed to demonstrate that we had been able to recapitulate the entire circle, including the capacity to colonize germ-free mice and to cause innate immune signatures in cells in culture and in intestinal tissues absolutely similar to those ontained with in vivo grown SFB (This breakthrough advance is in favorable revision in Nature (Schnupf et al., in revision). Furthermore, through a collaboration with Storkbio, for detection and characterization of SFB upon infection of cell monolayers and to carry out neutralization experiments in vivo, anti-bacterial polyclonal and monoclonal antibodies are developed. For this, intact SFB bacteria were isolated from the ceacum of mono-associated germ-free mice and heat inactivated at 75°C for 20 minutes. Heat-inactivated samples were used for rabbit and mice injection. By the end of second period, polyclonal antibodies were produced and five parental hybridoma clones selected, each expressing anti-SFB specific antibody. Clones are currently expanded, antibodies purified from supernatant and pure antibody samples transfered to Institute Pasteur for testing and assay development. These antibodies are highly useful not only for immunohistochemistry and immunofluorescence but also for addressing immunological questions regarding SFB activation of the immune system in vivo.
D4.3. - Genetic analysis of probiotic colonization and interaction potential. A library of 10,000 tagged-mutants was realized based on our recently designed Pjunc-TpaseIS1223 system (Licandro-Séraut et al., 2013, Appl. Environment. Microbiol.). To increase the potential of consecutive applications to this mutant library, we identified the transposon target in each mutant of the library. This required the optimization of genomic DNA extractions in 96-wells format and of direct sequencing on chromosome using a Sanger based method (with C. Bouchier and M. Tichit, Institut Pasteur). This method allows the identification of transposon targets in 90% of the mutants (Scornec et al., 2014, J Microbiol Meth), it allowed us to assemble a library of 1100 mutants in individual ORFs that were organized in pools harbouring different transposon tags. For reasons mentioned in “Task 1”, the rabbit ligated ileal loop model was used to screen this library for L. casei factors necessary for intestinal colonization. We identified 47 mutants that were classified in different functional families (i.e. DNA recombination and repair, cell wall construction/maturation, amino acids and sugar transporters, amino acids metabolism).
We believe we have achieved the set goal of developing and validating reliable reverse genetics in L. casei (Licandro-Séraut et al., 2014, PNAS). This fully annotated library is now routinely used to identify effectors molecules of a model commensal in various systems such as intestinal epithelial transport of lipids and maturation of the mucosal immune system.
D4.4. - To capitalize on deciphering the immunosubversive strategies of a pathogen, Shigella, to identify key elements of gut homeostasis. We demonstrated that the Shigella effector IpgD a phosphatidyl-inositol phosphatase blocks hemichannel opening and ensuing appearance of extracellular ATP by producing the poorly characterized lipid PtdIns5P. We therefore describe a new paradigm of host-pathogen interaction based on endogenous danger signaling and identify PtdIns5P as key regulator of mucosal inflammation. These data provide new openings for the development of anti-inflammatory molecules (Puhar et al., 2013, Immunity).

D4.8 Identification of commensals and dietary lipids that modify intestinal DC differentiation and function through PPARγ.
The common feature of all tested gut commensal bacteria (2 Escherichia coli, Morganella morganii, probiotic Bacillus subtilis (provided by Dr. Nadiya Boyko), 8 Lactobacillus reuteri (provided by Dr. Nathalie Juge), 2 Prevotella jejuni and Lachnoanaerobaculum umeaense (by Dr. Marie-Louise Hammerström) was that they were internalized by moDCs and induced the expression of the CD83 activation marker. They also were able to induce both pro- and anti-inflammatory responses in a bacterial strain-specific manner however, the efficacy of this mechanism varied among species. PPARγ was shown to regulate moDC functions at multiple levels: it turns on the genes of enzymes responsible for the production of ATRA and also regulates the cell surface expression of the lipid-antigen presenting protein CD1d indirectly. In the course of in vitro moDC differentiation the presence of ATRA inhibits the cell surface expression of CD1a wile increases CD1d expression. Consistent with this regulation, the gene-expression levels of the nuclear retinoid receptors RARα and RXRα, both acting as key players of ATRA synthesis via stimulating PPARγ and retinaldehyde-dehydrogenase-2 (RALDH2) expression, were upregulated by ATRA. However, the tested commensal bacteria interfered with this effect by decreasing the expression levels of the RARα and RXRα retinoid receptor genes thus providing a link between the dimerization partners of PPARγ, CD1 dichotomy and ATRA-mediated regulation of moDC responses. In this context PPARγ(+)CD1a(-)CD1d(+) moDCs could be identified and were shown to express the gut-homing integrin CD103 at a significantly higher level on the cell surface than in cells with the PPARγ(-)CD1a(+)CD1d(-) phenotype. However, this effect could dramatically be down modulated by all tested microbes indicating a ubiquitous and highly potent regulatory circuit.
In our in vitro cell cultures we also detected a CX3CR1+ moDC population that upon loading by commensal bacteria could be enhanced by 1 nM ATRA that also enhanced the secretion of the pro-inflammatory cytokines TNF-α, IL-1ß, IL-6 and IL-8 to a similar extent as induced by specific TLR stimuli. Monitoring the secretion levels of the T-cell polarizing cytokines IL-12 and IL-23 demonstrated that they are also secreted, except when co-cultured with Bacillus subtilis, that induced the secretion exclusively of IL-10. Despite the presence of increased levels of inflammatory cytokines in the moDC-bacterium co-cultures we observed a decrease in the cell surface expression of CD1a together with the co-stimulatory molecules CD40, CD80 and CD86, while the cell surface expression of CD1d and CD103 was significantly increased previously shown to be associated with a shift in moDC functionality.

D4.3 Identification of intestinal bacteria stimulating host T-cell responses
In vitro myeloid cell differentiation induced by GM-CSF+IL-4, GM-CSF (DC) or M-CSF (Mf) responded differently to activated CEC-mediated stimuli. The signals provided by CEC could be translated to myeloid cell-mediated T-cell activation, but the response of Th17 producing T-lymphocytes was affected by ATRA. When we monitored the functional outcome of moDC- and Mf-mediated effector T-lymphocyte responses we found that the supernatants of stimulated CEC could induce CD103 expression on in vitro generated moDCs, and when conditioned by ATRA and co-cultured with CD4+ T-lymphocytes they reduced the proportion of inflammatory Th17 T-cells. In contrast, in the Mf co-cultures the ratio of the effector T-lymphocytes was increased. Thus cytokine-activated CECs were able to trigger the secretion of distinct combinations of chemokines, which resulted in moDC activation controlled by ATRA and also governed moDC and Mf responses. These result revealed the tolerogenic outcome of moDC-mediated T-cell responses mediated by ATRA, a dietary lipid [2].
Considering the impact of the special microenvironment of moDCs created by various microorganisms in the presence or absence of ATRA we sought to demonstrate how these stimuli derived from the engulfed commensal bacteria could impact on the development of effector T-lymphocytes. To assess this scenario we used the previously established in vitro model system for stimulating moDCs in the presence or absence of ATRA and tested the outcome of moDC-mediated T-cell responses by ELISPOT assays. With all bacteria we detected that the PPARγ(-)CD1a(+) inflammatory moDCs could induce the expansion of autologous Th1 and Th17 cells, but the PPARγ(+)CD1a(-) cells failed to result in the expansion of Th17 cell and diminished the Th1 response. These results indicated that the in vitro exposure of moDCs to commensal bacteria and/or to their products such as MUB could induce T-lymphocyte expansion and differentiation to directions dictated by the microbe-associated stimuli and controlled by ATRA.
Along these lines, colleagues from Cyclo Lab have developed a new product line: fatty acids solubilized with non-cholesterol interacting random methylated alpha-cyclodextrin has been put on the CycloLab’s web site (http://www.cyclolab.hu/product1_7.html) and some enquires have already been obtained.







WP5 Translation of results from above WP’s to substantiation of health claims Participants: 15 (leader), 13, 18, 19, 20
This WP will i) initiate the exploitation of identified molecular targets to demonstrate beneficial effects of health-promoting foods and to substantiate health claims made by the food industry and ii) formulate a protocol template for a clinical study using the most feasible of the novel biomarkers developed in the project.


Results

Task 5.1: Rationale behind the biomarker map was to figure out what data is generated and cumulated within TORNADO project to be utilized later on to develop new biomarkers for health claim purposes. Since EFSA approach for health claim approval is very risk factor oriented i.e each health claim is focused on one single risk factor, this approach was also applied to the biomarker table. Aim was to list each biomarker candidate in each Work package by indication areas. Since different types of scientific data have different roles in health claim substantiation in EU, nature/type of data was also classified in the table; data from randomized clinical trials being the most valuable one for health claim substantiation itself, in vitro and animal data more important to demonstrate the mechanism behind the health effect and epidemiological data being an important part verifying that risk factor do have link to ultimate health outcome like disease cases/events. One purpose of the biomarker map was to identify for which biomarkers all needed data to validate the biomarkers would look most promising from EFSA point of view. Deliverable 5.1 has been submitted and contains the biomarker map.

Task 5.3: Since randomized clinical trials are the key part of health claim substantiation within Task 3.3 a template to set up study protocol to be used in health claim substantiation was developed. Initial idea was to test the template in the study made for a study validating/testing newly developed biomarkers in a pilot study. But since such a biomarker could not be identified during the project this testing part did not realize, but company partners were able to utilize the template in their internal projects and feedback from that was obtained. The value of newly developed biomarkers would be finally tested using them as outcome markers in clinical studies and therefore it is important that researchers in different domains (preclinical testing, epidemiological studies, metabolomics) are aware of the requirements once entering into the clinical phase. Deliverable 5.3 has been submitted and contains the template protocol.










WP6 Gut microflora interactions Participants: 1, 14 (leader)
A significant part of this WP is concerned with molecular profiling of the GI tract microbiota in support of experiments in this and other WPs.

Results

The microbiome profiling carried out within the TORNADO consortium - work (D 6.2 D6.3 D6.4) - has supported the significant number of projects within WP’s 1, 2 and 3 and the implications of their results. See reports within the above workpackages.

The lysin from phage vB_CpeS-CP51 (Gervasi et al 2013) was expressed under the control of the nisinA promoter in a strain of L. johnsonii FI9785 which had been engineered to express the nisRK two component regulatory system from the chromosome after a constitutive promoter – this system (D6.12) effectively gave constitutive expression of the lysin without the need for external nisin induction (Gervasi et al 2014a). An addition of the engineered signal peptide SLPmod allowed secretion of the lysin, and lysis of C. perfringens cell wall material was demonstrated both by growing colonies and filtered supernatant of L. johnsonii expressing the lysin CP25L (Gervasi et al 2014a). Tests in in vitro batch cultures with media designed to simulate the conditions of the human colon showed that although L. johnsonii alone was able to control the growth of C. perfringens, this ability was compromised when the pH was controlled (Gervasi et al 2014b). Expression of the lysin demonstrated an improved ability to reduce numbers of C. perfringens even with pH control, but good survival of the producer strain was essential (Gervasi et al 2014b).
This work demonstrates that engineering the competitive exclusion strain L. johnsonii FI9785 can improve its therapeutic potential.

Further work on the mucin binding and colonisation by the probiotic lactobacillus johnsonii suggested that bacterial exopolysaccharides (EPS) play a key role in the mechanism of colonisation of the gut epithelium (D6.9). Making genetic knockouts of the specific genes of the EPS biosynthetic pathway, we were able to obtain variants of L. johnsonii with qualitative and quantitative differences in EPS. The structure of the EPS was analysed by NMR, showing that L. johnsonii produces both a homopolysaccharide glucan (EPS1) and a hetereopolysaccharide composed of glucose and galactose (Dertli et al 2013). EPS2 was shown to be encoded by an eps gene cluster, the deletion of which also prevented the production of EPS1 (Dertli et al 2013). Further examination of eps mutants showed that the presence and quality of the EPS layer affects the physiochemical surface characteristics, biofilm formation, aggregation and adhesion to chicken GI tract tissue (Dertli et al submitted). EPS was also shown to have a protective effect against a variety of stresses. All of these characteristics are likely to affect both colonisation and competitive exclusion of pathogens in vivo, and EPS mutants are currently being assessed in poultry trials.

Bacteriophages are the most abundant life form on earth and are ubiquitous in complex microbial communities of specific environmental niches such as soil, oceans and sewage. They normally outnumber the bacterial host by 10 fold and hence have the potential to modulate the composition of the microbial communities. In our continued efforts to understand the role of phages in manipulation of the gut microbiome (D6.12) we conducted in vitro fermentation experiments with reciprocal administration of phage preparations from two individuals and the metagenomic data revealed that it is possible to alter the microbiome by administration of bacteriophages. These data obtained using crude phage preparations indicates the potential of specific targeted phage administration for manipulation of the human gut microbiome.

Our in vitro host bacteria interaction studies with Desulphovibrios provided further evidence for the proinflammatory properties of these important groups of gut bateria (D6.7). Our fermentation studies showing the modulation of SRB levels and their activities in response to dietary intervention suggests that it is possible to alter the function and the levels of these proinflammatory gut bacteria by diet and may provide a means to aleviate of symptoms in patients with inflammatory conditions. However these in vitro studies now need to be replicated using human clinincal studies which are currently in progress.



Potential Impact:
The TORNADO consortium set out to help develop tailor made food products for health maintenance through exploiting molecular mechanisms underlying beneficial host-microbe interactions. The work emanating from the research activities within TORNADO has had more far reaching effects that first anticipated.

TORNADO´s microbiome sequencing platform, supervised through the WP6 team, has provided technical support for a large set of projects within WP’s 1, 2, 3 and 4. Through the WP leader, Prof Arjan Narbad, Norwich, state of the art sequencing projects have revealed a detailed insight into the diversity and composition of the microbiome obtained from our human cohorts as well as from our preclinical experimental animal models. These results have also given us a deeper insight on the complexity of our understanding of the role of the gut bacteria in immunity and healthy aging. The results are presented along with the projects they have supported. Furthermore, the data on the gut microbiome generated has also provided further insight into the host/bacteria interactions when exposed to environmental stress including, starvation, dietary changes and altered immune function. The studies included analyses obtained from our human cohorts as well as our animal studies.

One very important cornerstone of TORNADO has been the mechanistic queries conducted within WP3. We are very proud of what we have accomplished as the ground breaking observations have contributed significantly to further our understanding of the bilateral dialogue between the microbes and its host. The observations that gut microbes can communicate with and impact on developmental programming of the CNS and ENS, BBB function and development as well as placenta development is at the forefront of this research field. As with all important observations, they open more questions than answers. Adding also that gut microbes communicate with organs such as liver, testis and muscle, the true meaning of the holobiont concept is fulfilled. These interactions have far reaching consequences for our basic understanding of normal physiology as well as disease development. In short, text books will be rewritten and the entire biomedical field is experiencing a paradigm shift. TORNADO´s work has contributed significantly to this development. Part of the success lies in the approach TORNADO took: Applying a coherent interdisciplinary approach, which evolved into systems biology, has allowed us to disclose new signalling pathways and metabolites that have profound effects on host physiology and immune status. Our results also demonstrate that the gut microbiome is directly invovled in our health be it early in life or when you are getting old. That is, the crosstalk between host and microbiota is central for the fine-tuning of the host’s biology throughout life. The development of animal models and new protocols for studying microbes as well as bringing epidemiologists closer to preclinical research and of course the reverse, have paved the way for interdisciplinary thinking for years to come and for researchers worldwide. The results generated from our mechanistic platform are in the absolute front line of research. The breakthroughs include the identification of gut microbes modulating cognitive function and brain development. In addition, the maternal microbiome closes the Blood-Brain-Barrier and contributes to placenta development. Microbes contribute to the development of the enteric nervous system and can regulate the testis barrier. Another major observation from WP3 is the identification of a metabolic regulatory function of the xenobiotic sensor Aryl-hydrocarbon-Receptor (AhR). This receptor regulates uptake of fatty acids for b-oxidation via PPARa. Using a systems biology approach, TORNADO has shown that this enigmatic xenobiotic receptor is actually a key regulator of energy expenditure. What is even more profound is the observation that AhR is subject to regulation by gut microbial metabolites. AhR supports mobilization of energy during starvation by activating the enzymes that produce ketone bodies during caloric constraint and starvation. This illustrates a default mechanism to ensure energy to the host even under stress conditions and that our gut microbes are part of this regulatory circuit.These results have far reaching consequences for our understanding of microbiota host interactions acting on organs peripheral to the intestine and when exposed to metabolic stress. The data generated in WP3 is a glimpse into progress beyond the state of the art.

Regarding the molecular and cellular analysis of the cross talks between commensal and pathogenic microorganisms with the intestinal epithelium, performed in the frame of WP4, the results have largely gone beyond the state of the art. The capacity to cultivate SFB in in vitro conditions opens the possibility to study the biology of this so far uncultivable microorganism that is a key regulatory microbe intimately connected to innate and adaptive immune responses. Moreover, this discovery will help moving forward the field of functional genomics of the microbiota with the possibility to generate new biomarkers and provide new therapeutic approaches.
The exhaustive mutagenesis of the L. casei genome offers the possibility to fully identify the factors that support gut colonization by a comensal microorganism, but also to identify effectors from the microbiota that are essential in major symbiotic interactions, such as the maturation of the immune system and the digestive and nutritional functions. The mutant library is also expected to become a major tool to study the biology of probiotics which are increasingly considered in therapeutic attempts at restoring balanced microbiota including signalling and mechanistic properties . The demonstration that a pathogen can shut off potent proinflammatory danger signals via active closing of connexin-based hemichannels opens the way to identification of novel anti-inflammatory and immunomodulating drugs.

The in vitro experimental model system designed for the characterization of gut associated human immune cells and mechanisms offers a flexible screening platform for comparing the common and also the unique functional properties and the mode of action of individual microbiota components at the level of epithelial and dendritic cells and their impact on the outcome of effector T-lymphocyte activation and polarization. It also provides a versatile cell-based platform for identifying the origin and measuring the composition of soluble factors such as cytokines, chemokines and metabolites produced by the individual intestinal and colonic cell types and their subsets. The assay system can also be used for the comparative analysis of individual microbes with unique functional properties in a human system using primary cells. This model system potentially can be scaled up and eventually ready for a high through put platform for analysing heterogeneous cytokine mixtures and characterize multiple cell types. Further studies are needed to reveal the final output of the ex vivo manipulated cells to get insight to the reproducibility and feasibility of this method.

The human membrane proteins CD1a and CD1d expressed exclusively in humans can be used as natural phenotypic and functional markers controlled by the nuclear receptor PPARγ that is also involved in the regulation of many other genes and mechanisms among them the induction of enzymes involved in retinoic acid synthesis essential for conditioning the gut microenvironment to maintain tolerance to commensal bacteria.

It is important to note that some results obtained with primary human cells in our ex vivo experimental test system differed from those generated in mouse models. One example is related to the CX3CR1 chemokine receptor expressed on the surface of resident intestinal phagocytes having the capacity of sampling the luminal content by their dendrites and transfer antigens to CD103+ cells, which can be polarized to both Th1 and Th17 directions locally. In our in vitro system, the increased number of CX3CR1 on the cell surface was not associated with elevated levels of Th1 and Th17 cells indicating an important difference between the human and the mouse systems. Whether this is the limitation of the human test system or reflects a real difference in the regulation of human and mouse DC phenotypes and functions has to be explored.
Experiments aimed to characterize the so-called tolerogenic CD103+ gut DCs loaded by luminal antigens, migrating to intestinal lymph nodes to activate T-cells have not been able to be reproduced in the human system. Although 10-20% PPARg+CD1a- human cells were detected in our system, it was only possible to confirm their IL-10 and TGF-ß production should be associated with the migratory potential based on mouse experiments. This could be due to the controversial regulatory role of TGFβ in the human system.

An important function of the gut microbiota is to produce biologically active metabolites such as short chain fatty acids (SCFA) and different polysaccharides. Other essential nutrients are not produced by humans but supplied by food components such vitamin A and poly-unsaturated fatty acids (PUFA), which is essential for hormone syntheses and the maintenance of cell membrane structure. Recently PUFAs were found to inhibit inflammatory reactions and also emerged as potential treatment options for inflammatory diseases.

TORNADO has shown that potentially inflammatory response of moDCs to gut microbes can be kept under check at ex vivo conditions by the induced production of ATRA by CEC and moDCs. These results illustrate the interplay of gut microbes and moDCs and how the balance between these cell types can be modulated by a microenvironment conditioned by low concentration of ATRA. This finding reveals a microbiota-induced and moDC-mediated interaction which could be utilized for optimizing dietary supplementation with Vitamin A. The impact of dietary lipids on immune responses can be followed up by monitoring the cell cycle of inflammatory T-cells and the polarizing capacity of moDCs both being dependent on the amplitude of the cell cycle, inflammatory and type I interferon signalling and recognition of microbial ligands by different pattern-recognition receptors.

Results from studies of adhering bacteria and regulatory T cells in celiac disease (CD) strengthened the notion that modifications and alterations of the gut microbiome may be a potential risk-factor for chronic inflammatory conditions similar to or identical to CD. One intriguing observation suggests that certain CD associated bacteria maybe considered as possible risk-factors for the development of childhood CD, perhaps by affecting intestinal programming and epithelial homeostasis in early life. Indeed an, until now unknown species of the genus Prevotella, Prevotella jejuni, was isolated from small intestinal biopsies of CD patients. P. jejuni bacteria bind to intestinal epithelial cells and affect epithelial function adversely by increasing permeability and inducing expression of stress related genes and detoxification enzymes. In future studies it should be interesting to address the question whether the CD associated Prevotella species identified here is contributing to the observation that the "Prevotella enterotype" show association to inflammatory bowel disease and obesity. Furthermore we fund that a mixture of P. jejuni, P. histocola, P. melaninogenica, Actinomyces gravinitii and Lachnoanaerobaculum umeaense, all isolated from the small intestinal mucosa of CD patients, induced production of the proinflammatory cytokine IL-17A by T lymphocytes in the small intestinal mucosa suggesting that any or all of these bacteria contribute to the inflammatory reaction seen in the CD patients upon intake of dietary gluten. The work related to sulphur reducing bacteria are giving further insight into the role of these bacteria in development of inflammatory disorders such as ulcerative colitis and our data suggest that it may be possible to reduce the impact of these proinflammatory conditions by diet. Translation of the in vitro data into human intervention strategies in the future will have significant impact on the dietary interventions in patients with inflammatory disorders.

TORNADO´s work on the bio control of gut pathogens using probiotics has led to an understanding of the mechanisms involved in the competitive exclusion process. During the project, we have produced SMART probiotics expressing endolysins that are currently being tested under farm environments and if successful their application will impact on the food poisoning in humans as well benefitting animal welfare.

TORNADO has performed limited work on the translational side. A probiotic intervention study feeding the probiotic LF19 to infants during weaning, has revealed important information regarding the capacity to evaluate possible safety and health promoting effects of probiotics in the general population. This is because it is one of only a few double-blind intervention studies on healthy infants and not limited to analysis of possible positive effects of probiotics on individuals in risk-groups. A further strength is the good compliance and a low drop out rate. 121/171 children that had completed the initial intervention were recruited to the 8-year follow-up study. Thus, reasonable long term effects of the intervention study are possible to observe. LF19 conferred positive effects during the intervention during weaning from the breast but did not display long-term preventative effects on allergic disease, dental health or development of adaptive immunity. In addition, no long-lasting effects on body composition, blood lipids, fasting insulin or glucose levels could be observed. Interestingly, lower levels of palmitoleic acid (linked to visceral obesity) and higher levels of putrescine (linked to gut integrity) were recorded in the intervention study. In line with the concept of the Barker hypothesis, the intervention study may thus set a threshhold which allows these treated individuals to be able to handle stress later in life. However, the results will still take another 20-25 years before we will see the outcome of this intervention. In conclusion, the studies bring added value to microbiota composition modulation by probiotics. Importantly, a longer study time and larger cohorts are required before we can make better predictions from intervention trials using probiotics.

The dietary crossover intervention study on healthy adults showed that after the diet rich in whole grain products and low amounts of red meat (WG intervention) the body weight, BMI, and body fat mass were statistically significantly lower as compared to baseline, while no change was seen after the diet with high amounts of red meat and no whole grain products (RM intervention). WG intervention resulted in a higher occurrence of Collinsella aerofaciens, a bacterial species that has been associated with microbiota diversity, a trait generally though to have beneficial properties. The level of several metabolites in serum, some identified and others not identified, were changed by the diets. Of note, two of the metabolites that increased during the WG intervention period decreased during intervention with red meat. The substance 3,4-dihydroxybenzoic acid, increasing during WG and decreasing during red meat intake, also known as protocatechuic acid (PCA), is a naturally occurring phenolic compound found in several plant-based foods including fruits, vegetables and nuts. Specifically, PCA is a major metabolite of procyanidins and anthocyanins. It exhibits antioxidative properties and recent evidence suggests a potential role of PCA in hedging the development of cardiovascular disease and cancers. The second substance with the same changing directions, 4-hydroxybutanoic acid or γ-hydroxybutyric acid (GHB) serves as both a precursor and a metabolite of γ-aminobutyric acid (GABA). GABA is the main inhibitory neurotransmitter in the mammalian central nervous system. Furthermore, GHB itself is also believed to exhibit neuro-modulating activities and it has been used in the treatment of narcolepsy. The impact of diet on GHB levels is largely unknown and deserves further investigations. Another not yet chemically annotated molecule (‘Unknown_69’) showed a decline during the WG intervention and an increase during the RM intervention period. Its identification may provide further insights into mechanisms contributing to the opposite directions of WG and RM on human disease risk. Coincidently, with the changes in serum metabolites, the microbiome in the intestine also changed during the intervention periods. Thus, it is tempting to speculate that the changes in the microbiome may be part of these changes in metabolite profiling observed in plasma. Additionally the associations of the metabolites and the DGGE bands that changed during the intervention periods were investigated in a sub-cohort of the EPIC-Potsdam population in which the same parameters were determined. Here emerged some correlations between the mentioned parameters. Three DGGE bands were correlated to the metabolite “unknown_69”, highlighting the interest to identify this metabolite, and one band was repeatedly linked to Lactic acid, Creatinine, 1H Indole Acetic Acid, 2-Butenedioic acid and Trans-4-Hydroxyproline.
More studies are highly warranted ideally working in close harmony with the food industry.

It was also shown in WP1 that certain components of the habitual diet (e.g. energy, monounsaturated fatty acids, n3 polyunsaturated fatty acids (PUFAs), n6 PUFAs, and soluble fiber) had significant associations with the stool bacterial numbers (e.g. increased energy intake was associated with reduced numbers of Bacteroides spp.) (Simoes et al. 2013). In addition, age correlated with the serum metabolome, serum lipidome and fecal microbial diversity. Moreover, there were significant correlations between fecal microbial diversity and serum metabolome and fecal microbial diversity and serum lipidome. This finding can have consequences for monitoring gut microbial diversity in the future.

At present, there is a lack of valid biomarkers for human nutrition intervention studies. The food and health relationship focuses on maintenance of normal health or reducing a disease risk factor, in terms of European health claim legislation. However, to date, most accepted biomarkers quantify disease endpoints or damage. This has led to major problems in demonstrating health benefits and establishing health claims, and blocks competitive economic and health developments in the food sector. Within the TORNADO project we have attempted to address this problem and from our data from different health areas, we have been able to generate a comprehensive list of potential biomarkers (gut microbiota regulated molecular targets). Within the timeframe for the project and with the limited information about the transition into the human situation and the size of the small exploratory studies, we cannot justifiably undertake a pilot intervention trial based on our findings. This is a setback as we would have hoped to have been able to do so. Experiments in animal models is one thing, designing human intervention studies is altogether a much more complicated issue and must always be 100% risk free. However, using our list as a basis, there is a definite possibility to find possible novel biomarkers for the future nutrition research and biomarker validation. Novel biomarkers of different health effects will help food companies to demonstrate health effects of their products and ingredients and achieve health claims for food products. This will promote innovative food products, increase the economic impact for food research and development by food sector companies. In addition, the scientifically evaluated and authorised health claims will increase a degree of consumer protection.

The regulation on health claims raises a need for high scientific standards when conducting nutrition studies. In the European Union countries, according to Regulation (EC) No 1924/2006, it is stated that claims should be scientifically substantiated on the basis of available scientific data and evidence. Therefore, scientifically valid methodology should be applied in clinical nutrition studies. In general, high quality studies are required, but there are also some special viewpoints for nutrition research and health claim ambitions that are raised throughout the protocol. For example, there is a need to provide evidence that the well-defined claimed effect is beneficial to human health. Also, there is a need to characterize the product well and establish a cause and effect relationship between the consumption of the food or food ingredient and the claimed effect. For establishing a cause and effect relationship, nutrition studies need to be executed with a study group that matches a consumer target group, under defined conditions of use. Within the TORNADO project, we have developed a ‘template protocol’ for human intervention studies which can be used for the monitoring of health effects, substantiating health claims and for the development of a new generation of health-promoting foodstuffs. It gives to the industrial partners a very good overview of how to conduct clinical intervention studies for health claim substantiation. In addition, the protocol can be used by scientific researchers who are not very familiar with the current health claim legislation to ensure the legality of the studies also for health claim purposes. This will help companies to conduct studies of good quality and increase the possibility for an authorized health claim.

Lessons from TORNADO:

Concerted global attempts to understand gut microbe-host interaction are currently one of the most interesting topics in biomedical research. An illustration of the interest and the amount of efforts put into this research area is best illustrated by the increasing number of publications in high impact journals. TORNADO has applied state of the art technology to generate world class results. Relevant models and new protocols to study microbes ex vivo have been developed. On a number of occasions, the coordinator has been in discussions with big pharma and their growing interest in microbiome research. Despite all the positive outcomes, there have been problems. One is a logistic problem in that investigators from different fields of research do not speak the same scientific language which has caused communication problems. Another issue is the use of multi cohorts, many of which were described and defined prior to the launch of TORNADO. There have been clear problems to get access to data including to share results from different cohort groups. This needs to be looked at in the future. Another issue has been to stratify and standardize the sampling protocols from the human studies. We still have a long way to go here. Another issue is the fact that many of the human cohorts were designed for different intentions, at the time, and to cross-communicate in order to acquire coherent information has in part hampered the outcome of these studies. In this respect, the work done in TORNADO has given rise to both putative concerns and strict guidelines when approaching these types of studies in the future.

Aspiring possibilities:

Looking ahead to the post TORNADO time, we can anticipate stronger translational research activities where data from human cohort studies will be translated into the animal models for validation and to decipher mechanisms. This is conceptualized in our Roadmap-to-Health, which demands more substantial evidence for health claims, whilst providing a protocol for personalized, tailor made food. The work carried out in TORNADO has shown that this is possible yet with a limited budget. Another avenue may be to design prospective studies which may disclose and identify “tails”, i.e. early changes in metabolite profiles or in the microbiome composition that precede the development of full blown disease. The gut microbiome is indeed a latent goldmine to monitor early changes in the metabolic homeostasis of a given individual. This will require the marriage between traditional phenotyping studies and the rising metabonomic technology, to create phenomecentres for the advancement of systems medicine.
Life style related diseases will never be easily solved purely by relying on synthetic chemical libraries of small molecules. Targeting the gut microbiome with combinations of probiotics and micronutrients is a new road that is worthwhile exploring. It may also prove to be much safer compared to treatment with small molecules.
There are still fundamental questions to be addressed – for instance, what is health? How does the microbiome influence developmental programming, from cradle to grave? Coming dietary interventions should be connected to systems biology analyses. The costs for these studies should be shared by the food industry with support from the national authorities. There is also an urgent need to consider larger international cohorts to bring ethnic pluralism and to be able to compare results across ethnic, environmental and socio-economic diversities.
TORNADO has shown that world class data can be generated across scientific disciplines, despite the obstacles of cultural differences between research areas. The future for microbiome-host research is bright.

List of Websites:
www.fp7tornado.eu

Coordinator: Sven Pettersson
Department of Microbiology, Tumor and Cell Biology
Karolinska Institute
Theorellsväg 3
SE-171 77 Stockholm Sweden
Phone: +46 8 52486686
Fax: +46 8 331547
E-mail: sven.pettersson@ki.se