Article Text


Microbiota and diabetes: an evolving relationship
  1. Herbert Tilg,
  2. Alexander R Moschen
  1. Department of Internal Medicine I, Gastroenterology, Endocrinology & Metabolism, Medical University Innsbruck, Innsbruck, Austria
  1. Correspondence to Professor Herbert Tilg, Department of Internal Medicine I, Gastroenterology, Endocrinology & Metabolism, Medical University Innsbruck, Innsbruck 6020, Austria; herbert.tilg{at}


The gut microbiota affects numerous biological functions throughout the body and its characterisation has become a major research area in biomedicine. Recent studies have suggested that gut bacteria play a fundamental role in diseases such as obesity, diabetes and cardiovascular disease. Data are accumulating in animal models and humans suggesting that obesity and type 2 diabetes (T2D) are associated with a profound dysbiosis. First human metagenome-wide association studies demonstrated highly significant correlations of specific intestinal bacteria, certain bacterial genes and respective metabolic pathways with T2D. Importantly, especially butyrate-producing bacteria such as Roseburia intestinalis and Faecalibacterium prausnitzii concentrations were lower in T2D subjects. This supports the increasing evidence, that butyrate and other short-chain fatty acids are able to exert profound immunometabolic effects. Endotoxaemia, most likely gut-derived has also been observed in patients with metabolic syndrome and T2D and might play a key role in metabolic inflammation. A further hint towards an association between microbiota and T2D has been derived from studies in pregnancy showing that major gut microbial shifts occurring during pregnancy affect host metabolism. Interestingly, certain antidiabetic drugs such as metformin also interfere with the intestinal microbiota. Specific members of the microbiota such as Akkermansia muciniphila might be decreased in diabetes and when administered to murines exerted antidiabetic effects. Therefore, as a ‘gut signature’ becomes more evident in T2D, a better understanding of the role of the microbiota in diabetes might provide new aspects regarding its pathophysiological relevance and pave the way for new therapeutic principles.

Statistics from


The human intestinal tract contains a unique group of micro-organisms that is, the microbiota consisting of numerous bacteria, archaea and viruses. All these micro-organisms generate a biomass of more than 1.5 kg and their combined genomes (ie, microbiome) exceed the human genome more than 100-fold.1–3 Whereas earlier studies have mainly proposed that these genes may encode functions which generally direct pathways favouring, for example, the digestion of complex carbohydrates or the development of innate and adaptive immunity, recent evidence suggests that the microbiota may have key functions in regulating metabolic pathways in health and in disease.4–7 The introduction of culture-independent, high-throughput sequencing technologies has allowed in the last years to increase the understanding of the complexity and diversity of the microbiota.8 In contrast, the majorities of the more than 1000 assumed bacterial species still cannot be cultured and therefore remain poorly characterised with respect to their biological functions.

The origin of chronic inflammatory processes observed in metabolic disorders is still a matter of debate.9 The recent obesity epidemic is a driving force for the worldwide increasing incidence of type 2 diabetes (T2D) as more than 80% of patients with T2D are overweight. Obesity-induced insulin resistance is the dominant underlying pathophysiological factor.10 As insulin resistance and metabolic inflammation are frequently observed in parallel, research in the past decade has tried to connect these two phenomena. It is widely accepted that the aetiology of insulin resistance is complex and involves various pathways.11 It is, however, also increasingly established that inflammatory pathways are critically involved in the evolution of insulin resistance.12 Overnutrition and certain diets could represent major starting points as they might alter the gut microbiota, lead to changes in lipid metabolism, hepatic steatosis and finally systemic inflammation.13 ,14 It remains, however, unclear at which sites inflammatory processes are initiated and the GI tract with its significantly altered microbiota could reflect one of the early events in these disorders.

An altered microbiota in metabolic disease might allow to initiate inflammatory processes. Such an altered microbiota might act ‘locally’ and via an impaired mucosal barrier act systemically. In support of such a concept, it has been recently demonstrated that patients with metabolic syndrome and T2D exhibit a remarkable endotoxaemia.15 ,16 This is in accordance with the recently proposed concept of ‘metabolic infection’, where parts of the intestinal microbiota might affect systemic including adipose tissue inflammation.17 ,18 In various disorders such as IBD or obesity a ‘microbiotal signature’ has been identified.19 ,20 In this article we will discuss the current evidence for a potential role of the gut microbiota in the pathophysiology of T2D.

Role of gut microbiota in obesity

Numerous studies in the past, particularly from animal models, have demonstrated that the microbiota might be considered as one major player in the development of obesity.21–25 A recent exciting study by Ridaura et al observed that the microbiota derived from discordant obese twins affects metabolism in mice.26 Here, the authors transferred the microbiota collected from four human female twin pairs discordant for obesity into germ-free mice thereby demonstrating that ‘human obesity’ could be transmitted to rodents via such a procedure. Interestingly, cohousing of mice containing cultured bacteria from an obese twin with mice containing bacteria from a lean twin prevented the development of obesity. Diet obviously was a critical cofounder in these experiments and reflects the importance of this study. When mice received an appropriate diet, that is, low-fat, high-fibre even when harbouring the obese microbiota and were cohoused with mice containing the lean microbiota, the lean microbiota dominated finally in the obese mice and prevented increased adiposity. Furthermore, by such a diet the obese microbiota was not able to colonise lean animals, and these mice remained lean suggesting that diet finally decides which phenotype develops.

Another recent large human study provided further insights on the role of certain microbiotal members in obesity.27 Here, the authors found that in case of low bacterial richness (low gene count (LGC)), obesity, insulin resistance, fatty liver and low-grade inflammation were more prevalent compared with subjects characterised by high gene count (HGC). Importantly, not all LGC subjects are obese in general neither are all HGC subjects lean or healthy. Patients with this ‘microbiotal phenotype’ that is, LGC, gained more weight and had a higher rate of systemic inflammation (increased C reactive protein and leptin levels, decreased adiponectin serum levels), a higher rate of insulin resistance and dyslipidaemia. Only on the basis of 46 genera LGC and HGC individuals could be distinguished. Whereas in contrast to some rodent studies Bacteroides and some Ruminococcus species were more dominant in LGC, Faecalibacterium prausnitzii, Bifidobacterium, Lactobacillus, Alistipes, Akkermansia and others were significantly associated with HGC. At phylum levels, Bacteroidetes and Proteobacteria were more abundant in LGC whereas Verrucomicrobia (eg, A muciniphila) and Actinobacter were more dominant in HGC. Findings of this study overall support a concept that in case of obesity potential proinflammatory bacteria may dominate, for example, Ruminococcus gnavus or Bacteroides and anti-inflammatory bacteria such as F prausnitzii are less prevalent. Further analysis investigating metabolic functions demonstrated that LGC subjects contained a rather proinflammatory microbiotal profile as suggested by (1) a reduction of butyrate-producing bacteria; (2) an increase in mucus degradation thereby potentially impairing gut barrier function (decrease in A muciniphila and increase in R gnavus) and (3) an increase in oxidative stress. Importantly, dietary intervention using an energy-restricted diet improved microbial richness and clinical phenotype in LGC subjects, although it was less efficient regarding improvement of inflammatory parameters in these patients.28 Clinical symptoms, however, improved in LGC and HGC subjects. These two studies further strengthen the assumption that microbial composition and potentially the richness of these bacterial genes in our gut might be able to better define a subgroup of obese people with metabolic and inflammatory complications. It should be mentioned that some reports have demonstrated that in contrast to others Bacteroides might be more abundant in obese subjects compared with lean counterparts.29–32 Despite these discrepant findings between some studies especially regarding the role of Bacteroides, the model proposed by Ridaura et al might be very suitable to test the effects of human microbiota under certain diets in germ-free animals and to test potential protective bacterial mixtures.26

Role of gut microbiota in type 2 diabetes

As initial human studies in obesity suggested a microbiotal signature and 80% of patients with T2D are obese, studies evaluating the microbiome in T2D were eagerly awaited. In a first landmark study, high throughput sequencing was performed using stool samples from Chinese patients with T2D and metagenomic analysis was combined with clinical data, thereby providing the first metagenome-wide association study (MGWAS) in T2D.33 Patients with T2D exhibited a moderate intestinal dysbiosis characterised especially by a decrease in butyrate-producing Roseburia intestinals and F prausnitzii. In this study, the concept of metagenomic linkage group (MLG) analysis has been applied and thereby it was observed that in the healthy control samples especially various butyrate-producing bacteria were enriched (eg, Clostridiales sp. SS3/4, Eubacterium rectale, F prausnitzii, R intestinalis and others) whereas in T2D most metagenomic linkage groups belonged to rather opportunistic pathogens such as Bacteroides caccae, various Clostridiales, Escherichia coli and others. The sulfate reducing species Desulfovibrio was more frequently observed in T2D. Also, and this is in contrast to some studies discussed later the mucin-degrading species A muciniphila was somewhat more abundant in Chinese patients with diabetes. When investigating potential associated functions of this gut dysbiosis in T2D, T2D microbiota exerted enrichment in membrane transport of sugars, branched chain amino acid transport and sulfate reduction and decreased butyrate biosynthesis. Importantly, enriched functions also included an increase in oxidative stress response which could represent a direct link to the proinflammatory state of patients with T2D. Although the authors concluded that the observed T2D-related gut dysbiosis was moderate, more than 3% of the gut microbial genes differed. Several shortcomings of this study should be mentioned: (1) the study lacked a respective gender balance; (2) the cohorts were not age-matched and importantly (as discussed later) (3) no data on disease-specific medications were reported. This first study in T2D was followed by another large MGWAS study from Europe.34 Here, the authors applied shotgun sequencing studying only postmenopausal female patients with T2D. Subjects exhibited increases in the abundance of four Lactobacillus species including Lactobacillus gasseri, Streptococcus mutans and certain Clostridiales such as Clostridium clostridoforme and again decreases in at least five other Clostridium species. R intestinalis and F prausnitzii, both prototypical butyrate producers, were highly discriminant for T2D. Interestingly, when comparing patients with T2D with a group of women with impaired glucose tolerance, an increase in energy metabolism/harvest and fatty acid metabolism could be observed in T2D. The Chinese cohort33 and the Scandinavian cohort reported here revealed somewhat discrepant results, however, both cohorts were considerably different regarding various features. In both cohorts, C clostridoforme MGCs and Lactobacillus species were increased whereas Roseburia_272, a major butyrate producer, was decreased. It has to be stated critically, that the number of analysed patients with T2D was still low (n=53) and study design was not able to allow to find out whether a diabetes-specific drug might have influenced microbiota composition.34 Despite these discussed shortcomings, these two studies are an exciting and important start in the field of T2D and clearly suggest that a ‘gut signature’ might exist and more importantly functional analysis also revealed that a proinflammatory tone might be initiated in the intestine which could reflect the starting point of low-grade systemic inflammation as commonly observed in T2D (boxes 1 and 2).

Box 1

Major findings from metagenome-wide association studies (MGWAS) in patients with T2D

  • Butyrate-producing Roseburia intestinalis and Faecalibacterium prausnitzii concentrations lower in T2D

  • Lactobacillus gasseri and Streptococcus mutans and certain Clostridiales higher in T2D

  • Proteobacteria higher in T2D

  • Increased expression of microbiotal genes involved in oxidative stress, that is, overall a proinflammatory signature in the intestinal microbiota

  • Genes involved in vitamin synthesis, for example, riboflavin lower in T2D

  • Shortcomings of studies: heterogenous populations, data on diabetes medication incomplete and its role unclear, studies lack gender balance; mucosa-associated microbiota not studied

Box 2:

Evidence for a beneficial effect of Akkermansia muciniphila on metabolic functions

  • Is a mucin-degrading Gram-negative bacterium constituting 3–5% of the intestinal microbiota

  • Concentrations inversely correlated with obesity and diabetes in many experimental and human studies

  • Prebiotic consumption such as oligofructose is metabolically beneficial and increases A muciniphila concentrations

  • Administration of A muciniphila to murines improves weight loss, metabolic control and adipose tissue inflammation

  • Metformin increases A muciniphila concentrations

  • Improves dextrane sulfate colitis

  • Controversies: some animal/human studies show conflicting results; in some experimental situations rather proinflammatory.33 ,36 ,77 ,78

Also a previously reported smaller study had shown that the microbiota from T2D differed from non-diabetic adults showing that certain Lactobacillus species may be increased in T2D.35 Zhang et al focused on the analysis of the gut microbiota by 16S rRNA-based high-throughput sequencing in prediabetes (n=64) and a small number of patients with newly diagnosed T2D (n=13).36 Normal subjects already differed from patients with prediabetes with higher numbers of F prausnitzii and Haemophilus parainfluenzae T3T1, whereas A muciniphila and Clostridiales sp. SS3/4 were less abundant. The latter results differ from the findings of Qin et al as discussed before.33 Verrucomicrobiaceae were significantly lower in the prediabetes and T2D groups. Although the pathogenesis of type 1 diabetes differs from T2D, some studies have also observed an altered microbiota in type 1 diabetes.37 ,38

Overall, the results from all studies presented here suggest that patients with T2D show evidence of gut dysbiosis. Reasons for discrepancies may be numerous and include various confounding factors such as different study populations, different sequencing techniques used, use of various diets and medications. Further studies have also to clarify by assessing patients with ‘early disease’ whether the observed dysbiosis in obesity and/or T2D is a consequence of disease phenotype or is involved causally in their pathophysiology. Available studies can only be considered as a starting point and future studies have to take these varying cofactors into account.

Immunometabolic pathways regulated by the microbiota

Short-chain fatty acids and G-protein coupled receptors

The host has a very limited repertoire of glycoside hydrolases needed to digest complex dietary plant polysaccharides. One important activity of the gut’s microbiota is its potential to digest such dietary fibres.39 Dietary fibres constitute the indigestible portion of plant foods containing two fractions: insoluble fibre such as cellulose or lignin and soluble fibre such as galacto-oligosaccharides or fructo-oligosaccharides. These soluble fibres are digested by enzymes derived from the gut microbiota into short chain fatty acids (SCFAs) such as acetate, butyrate and propionate. Interestingly, SCFAs constitute approximately 5–10% of the energy source in healthy people. Fibre-enriched diets have been demonstrated to improve insulin sensitivity in lean and obese diabetic subjects.40–43 SCFAs diffuse passively, are recovered via monocarboxylic acid transporters, or act as signalling molecules by binding to G-protein-coupled receptors such as G-protein coupled receptor 41 (Gpr41) (free fatty acid receptor 3, FFAR3) and G-protein coupled receptor 43 (Gpr43) (FFAR2).44 ,45 These G-protein-coupled receptors are expressed by many cell types including gut epithelial cells, adipocytes and immune cells. Kimura et al recently demonstrated that Gpr43-deficient mice are obese even when consuming a normal diet, whereas mice overexpressing this receptor specifically in the adipose tissue remain lean independent of calorie consumption.46 Importantly, and this proved a key role for the microbiota, when mice were raised germ-free or were treated with certain antibiotics both types of mice exhibited a normal phenotype. The authors could furthermore demonstrate that SCFA-mediated activation of Gpr43 resulted in suppression of insulin signalling in the adipose tissue subsequently preventing fat accumulation. High-fat diet-induced insulin resistance, as assessed by insulin tolerance and glucose tolerance testing, was significantly increased in Gpr43-deficient mice, compared with wild type mice, an effect which was also abolished after antibiotic therapy. Gpr43 activation also enhances insulin sensitivity by promoting glucagon-like peptide 1 secretion in the gut.47 Gpr43 is neither expressed in liver nor in muscle and therefore it seems that adipose tissue-derived Gpr43 is able to modulate all the metabolic effects after engagement with microbiota-derived products such as SCFAs. Therefore SCFAs are an important energy source for the host, and act as signalling molecules especially in the adipose tissue thereby maintaining energy balance. Data also suggest that the microbiota is the major source for Gpr43 agonists and biological functions of Gpr43 are completely dependent on the gut microbiota (figure 1).

Figure 1

Metabolic pathways affected by the intestinal microbiota. (1) Bacterial glycoside hydrolases cleave complex carbohydrates derived from dietary fibre to produce short-chain fatty acids (SCFAs) such as acetate, propionate and butyrate. SCFAs affect the host's metabolism in several ways. (2) SCFAs-dependent activation of G-protein coupled receptor 41 (Gpr41) induces the expression of peptide YY, an intestinal hormone that inhibits gut motility, increases intestinal transit rate, and reduces the harvest of energy from the diet.48 (3) Engagement of G-protein coupled receptor 43 (Gpr43; and Gpr41) by SCFAs has been shown to trigger the incretin hormone glucagon-like peptide 1 (GLP-1) to increase insulin sensitivity.47 (4) SCFAs-mediated activation of Gpr43 on adipocytes suppresses insulin signalling and inhibits fat accumulation in adipose tissue.46 (5) The SCFAs butyrate and propionate activate intestinal gluconeogenesis (IGN); butyrate through a cAMP-dependent mechanism, propionate via a gut-brain neuronal circuit involving Gpr41.49 (6) The intestinal microbiota suppresses the expression of fasting-induced adipose factor (Fiaf), an intestinal epithelial cell (IEC)-derived circulating inhibitor of the lipoprotein lipase (LPL).56

In contrast, an earlier study observed that conventionally raised Gpr41–/– mice or germ-free Gpr41–/– mice colonised with Bacteroidetes theatiotaomicron and Methanobrevibacter smithii were significantly leaner than wild type littermates, while there were no genotype-related differences in germ-free mice.48 A fascinating study from France indicated that SCFAs might exert some of their functions by directly regulating intestinal gluconeogenesis.49 Here, they reported that butyrate upregulated intestinal gluconeogenesis via a cAMP-dependent mechanism and independently from Gpr43 (FFAR2), and propionate affected intestinal gluconeogenesis via a gut-brain neural circuit involving FFAR3 (Gpr41). In mice deficient for intestinal gluconeogenesis (intestinal specific deletion of G6Pase catalytic subunit), the beneficial metabolic effects of SCFAs and soluble fibre were absent clearly highlighting the need for this mechanism. Overall, many studies have now demonstrated that the microbiota in lean mice produces greater amounts of SCFAs, especially propionate and butyrate, thereby promoting leanness by inhibiting fat accumulation in adipose tissue and raising energy expenditure (figure 1).

SCFAs might constitute an attractive link between the microbiota and systemic inflammatory diseases as demonstrated by various recent studies showing that SCFAs, especially butyrate direct the development of extrathymic anti-inflammatory regulatory T cells.50 SCFAs also control the generation of colonic regulatory T cells and protect against colitis in a Gpr43-dependent manner51 (figure 2). Marsland et al recently demonstrated that mice fed a high-fibre diet have an altered microbiota and are protected from allergic airway inflammation. The SCFA propionate regulated allergic inflammation, bone marrow haematopoiesis and dendritic cell function. The effects of propionate were dependent on Gpr41 but not Gpr43. Microbiota-derived propionate has also been recently shown to limit cancer cell proliferation in the liver suggesting antineoplastic functions.52 Taken together, these findings suggest that metabolites produced by the gut microbiota can influence haematopoiesis and immune responses in the lung.53 These data are supportive for the current notion that microbiota-derived products are important players in the generation of local and systemic immunity/inflammation. As studies in T2D especially and consistently revealed that production of SCFAs, especially butyrate is impaired it is reasonable to assume that such mechanisms might contribute to low-grade inflammation observed in those disorders.

Figure 2

Crosstalk between the intestinal microbiota and the host modulates mucosal immunity and systemic inflammation. (1) Short-chain fatty acids (SCFAs), particularly butyrate, promote the formation of peripheral regulatory T cells (Treg).50 ,51 ,93 Mechanistically, butyrate constitutes a natural occurring inhibitor of the histone deacetylases 6 and 9. In such a way butyrate could result in acetylation of Histone H3 to promote the expression of the Treg-specific forkhead transcription factor FoxP3. (2) Toll-like receptor (TLR) 5 is a pattern recognition receptor for flagellin, the protein monomer of bacterial flagella. Signals derived from TLR5 are important for the maintenance of intestinal homoeostasis.58 (3) Accordingly, intracellular signals derived from the NLR family pyrin domain containing 3/NLR family pyrin domain containing 6 (Nlrp3/6) inflammasome act in a similar way.60 Loss of these signals results in detrimental changes in the composition of the intestinal microbiota that promotes insulin resistance and the metabolic syndrome. The features are transmissible into wild type animals with the dysbiotic microbiota. (4) Loss of intestinal homoeostasis and dysbiosis foster insufficiencies in the gut barrier function that result in translocation of bacterial components, for example, lipopolysaccharides (LPS) or even whole organisms into the portal and systemic circulation. There, these factors could promote local inflammation in organs such as the adipose tissue and the liver. AMP, antimicrobial peptides; ASC, apoptosis associated speck-like protein containing a CARD; HDAC, Histone deacetylase.

Studies investigating SCFAs, their metabolism and potential functions have played a major role recently in a better understanding of the interaction of beneficial diets and the host-microbiotal interplay. Several other mechanisms have been discussed in the past years and can only be mentioned briefly. The gut microbiota affects the composition and abundance of certain bile acid species through a variety of mechanisms resulting commonly in low levels of various bile acids in case of obesity.54 Obese mice demonstrate increased expression of farnesoid X receptor and fibroblast growth factor 15, which expression is regulated by bile acids, and directly regulate various metabolic effects. Conventionally raised mice contain much more total body fat compared with those raised under germ-free conditions.21 Conventionalisation of mice suppresses intestinal expression of fasting-induced adipose factor (Fiaf) specifically in differentiated villous epithelial cells in the ileum. Fiaf is a circulating lipoprotein lipase inhibitor produced by the intestine and by liver and adipose tissue.55 Indeed, germ-free Fiaf−/− mice have the same amount of total body fat weight as conventionalised (Fiaf-suppressed) mice, establishing Fiaf as a key factor of microbial regulation of peripheral fat storage.21 Another metabolic pathway apart from Fiaf involves antimicrobial peptides (AMP)-activated protein kinase (AMPK) that protects mice from obesity produced by consumption of a high-fat high-sugar Western diet.56 Germ-free mice remain lean despite high calorie intake and this state is accompanied by increased activity of phosphorylated AMPK levels in liver and skeletal muscle and enhanced insulin sensitivity, in the liver.56

Innate immune system

Metabolic syndrome is thought to develop through the interaction of various genetic and environmental factors, and includes a complex and yet poorly understood interaction between the intestinal microbiota and the innate system.57 An important role in the development of a metabolic syndrome has recently been demonstrated for the pattern recognition receptor such as the toll-like receptor (TLR)5.58 TLR5–/– mice exhibit hyperphagia, developed hyperlipidaemia, hypertension, insulin resistance and obesity, and an altered microbiota. Transfer of intestinal microbiota of TLR5/ mice into germ-free mice led to metabolic syndrome. These data suggest that innate immune signalling is critical in the development of the metabolic syndrome and alterations in the intestinal microbiota are able to induce the metabolic syndrome (figure 2).

Inflammasomes consist of an upstream sensor nod-like receptors protein, the adaptor protein Asc (apoptosis associated speck-like protein containing a CARD) and the effector protein caspase-1. Various groups have recently shown that the inflammasome may play an important role in metabolic inflammation and some inflammasomes might affect the intestinal microbiota, metabolic syndrome and fatty liver disease.59 ,60 Henao-Mejia J et al observed that inflammasome-deficient mice on a methionine-choline deficient diet develop hepatic steatosis and liver inflammation as well as T2D. Disease could again be transferred between mice upon cohabitation, supporting that the microbiota from inflammasome-deficient mice has the potential to induce inflammatory liver disease. Whether such phenomena have any relevance in human disease, however, remains unclear.

Endotoxin—a major gut-derived player in metabolic inflammation

Several recent studies have suggested that gut-derived endotoxin (lipopolysaccharide, LPS) might be crucially involved in chronic inflammation observed in T2D. A first paper was published by Cani et al showing that a high fat diet (HFD) increased LPS content of the gut’s microbiota and resulted in metabolic endotoxaemia.61 Here, they also observed that subcutaneous infusions of LPS into mice resulted in insulin resistance and obesity similar as after feeding a HFD. In support of such a concept, antibiotic and prebiotic therapy improved metabolic inflammatory parameters in high-fat fed and ob/ob mice.62 ,63 In an animal model of metabolic adaptation where C57Bl/6 mice receiving a HFD became diabetic or resisted diabetes, the diabetes phenotype was associated with an increased gut permeability, endotoxaemia and a specific gut microbial profile.64 These preclinical findings were recently corroborated by clinical studies demonstrating that patients with metabolic syndrome and T2D exhibited endotoxaemia.15 ,16 Endotoxin could therefore indeed reflect one relevant player derived from the gut’s microbiota affecting inflammation-related metabolic functions.

The concept of ‘metabolic infection’

The adipose tissue is the major source of proinflammatory cytokines in case of obesity and associated hepatic steatosis, exhibiting a much higher expression of cytokines, compared with their liver expression.65 ,66 Burcellin et al recently suggested that a so called ‘metabolic infection’ might exist, suggesting that the gut microbiota might be a relevant factor in low-grade systemic inflammation and development of insulin resistance.17 In their studies, the authors demonstrated that prior to the onset of diabetes and early after the mice are switched to a HFD, intestinal bacteria translocate from the gut to the adipose tissue and blood where they might induce low-grade inflammation. Human studies using 16S rDNA pyrosequencing demonstrated that blood levels of certain bacterial DNA (>85% derived from Proteobacteria) were enhanced in prediabetes.67 Furthermore, a recent investigation showed that patients from the DESIR study exhibited a blood microbiota dysbiosis with an increased detection of Proteobacteria in subjects with cardiovascular complications.68 The discovery of bacterial DNA within various tissues such as blood, liver and adipose tissue allows a new perspective on the evolution of metabolic inflammation including adipose tissue inflammation69 (figure 2).

Role of certain bacterial strains

Akkermansia muciniphila

Knowledge in this area is still sparse and somewhat controversial.33 A muciniphila has been recently characterised as a mucin-degrading bacterium residing in the mucus layer.70 A muciniphila, a Gram-negative bacterium, is highly prevalent and constitutes 3–5% of the gut’s microbiota and its concentrations are inversely correlated with the presence of overweight and diabetes in murine and human studies.63 ,71 ,72 Dietary supplementation of fibres, that is, oligofructose to genetically obese mice dramatically increases abundance of A muciniphila.63 Several reports have recently demonstrated that A muciniphila might play a key role in the integrity of the mucous layer and has the potential to reduce inflammation and offer protection against the development of obesity and T2D. The most convincing report suggesting protective functions for Akkermansia was recently presented by Everard et al73 In this study, the authors showed that in genetic and dietary models of murine obesity concentrations of A muciniphila were highly decreased. A prebiotic therapy with oligofructose restored basal levels of A muciniphila and improved metabolic functions including metabolic endotoxaemia. When treating these animals via administration of A muciniphila, body weight was reduced and hyperglycaemia reversed. These studies suggest that A muciniphila controls fat storage, adipose tissue inflammation and glucose metabolism. Data were further corroborated by the fact that heat-killed A muciniphila lacked the above beneficial effects. Interestingly, in this study A muciniphila restored the decreased mucus layer probably improving gut permeability. This study is also supported by recent findings demonstrating that metformin, a key antidiabetic drug results in an increase in Akkermansia concentrations and again as observed in the Everard study administration of A muciniphila resulted in an improvement of various metabolic functions including glucose tolerance and adipose tissue inflammation.74

Early life therapy in the non-obese diabetic mouse model of diabetes using vancomycin results in an increase in the abundance of Akkermansia and improves diabetes further supporting a protective function for this bacterium.75 A muciniphila might exert anti-inflammatory functions also in other disease models as it has been shown that administration of this bacterium improves dextrane sulfate colitis-induced colitis.76

Data on Akkermansia are still somewhat controversial as, for example, in the Chinese MGWAS study a potential protective role for this bacterium could not be demonstrated.33 Qin et al reported as stated before that A muciniphila-related genes were more abundant in patients with T2D; this association, however, needs to be reassessed as Qin et al studied only 337 of the 2176 A muciniphila genes. HFD in various rodent models reduces butyrate and in contrast enhances inflammation and liver fat, and these effects are counteracted by the ingestion of dietary fibre.77 In this particular study when investigating the intestinal microbiota, concentrations of Akkermansia were highest with the fibre-free diet suggesting that Akkermansia expands in the colon of rats fed a HFD. Akkermansia supports mucin degradation and could facilitate intestinal inflammation as shown in Salmonella Typhimurium infected mice.78 Higher abundance of A muciniphila has been observed in healthy controls in some studies discussed previously.36 The still conflicting results on A muciniphila need further studies and clarification.

Faecalibacterium prausnitzii

F prausnitzii is considered the protoypical anti-inflammatory microbiota component. Several studies demonstrated that F prausnitzii concentrations are lower in patients with metabolic syndrome and diabetes.33 ,34 ,36 ,79 The diversity of the microbiota as well as the abundance of F prausnitzii is significantly lower in obese patients and patients with T2D compared with lean individuals.80 Functional studies using F prausnitzii in various disease models of metabolic inflammation and tests of its beneficial functions have yet to be performed.

Targeting the microbiota: diet, drugs and faecal transplant


Dietary modification has been demonstrated in previous years to rapidly affect the gut microbiota.81 Wu et al observed that short-term diets had no or minor influence on their enterotypes but long-term diets indeed were able to influence the enterotype of individuals.31 Diets enriched in animal protein and saturated fat favoured a ‘Bacteroides’ enterotype, a carbohydrate enriched diet the ‘Prevotella’ enterotype. The Bacteroides enterotype was highly associated with animal protein, a variety of amino acids and saturated fats suggesting that meat consumption as in a Western diet characterised this enterotype. Vegetarians demonstrated enrichment in the Prevotella enterotype. Phyla positively associated with fibre were Bacteroidetes and Actinobacteria, whereas Firmicutes and Proteobacteria showed the opposite association. A study comparing European and African children observed that a Western diet led to a dominance of Bacteroides whereas a more vegetarian diet in Africa was dominated by the Prevotella enterotype.32 The discussion on the issue of ‘enterotypes’ is still open as other studies failed so far to see such associations.

Another important diet study was recently published by David et al.82 Here, the authors could show that short-term dietary interventions were also able to modify the gut’s microbiota substantially. Interestingly, the similarity of each study subjects’s gut microbiota compared with their baseline communities (ß diversity, Jensen-Shannon distance) decreased on the animal-based diet even within 1 day. Analysis revealed that an animal-based diet had a greater impact on the gut microbiota than the plant-based diet. The most abundant taxon changes were observed after an animal-based diet with the clusters Bilophila wadsworthia, Alistipes putredinis and Bacteroides sp, all characterised by the feature of bile resistance. In contrast, an animal-based diet resulted in a decrease in Firmicutes that are able to metabolise dietary plant polysaccharides such as Roseburia, E rectale or Ruminococcus bromii. Animal-based diets resulted in lower levels of SCFAs compared with a plant-based diet. The increase in the abundance of B wadsworthia is of special interest. HFDs in rodents result in an increase in enteric deoxycholic acid (DCA) concentrations, which is produced under the control of commensals such as B wadsworthia and has recently been demonstrated to promote development of liver cancer.83 In the study discussed here, faecal bile acid levels including DCA levels increased during an animal-based diet. DCA is able to affect the growth of various members of the microbiota such as Bacteroidetes and Firmicutes.84 B wadsworthia also has sulfite-reducing capacities and the end product of this pathway, that is, H2S might exert proinflammatory effects at the mucosa level.85 Growth of B wadsworthia is stimulated by selected bile acids in rodents after the consumption of saturated fatty acids. Overall the authors could demonstrate that the human gut microbiome can rapidly switch between herbivorous and carnivorous functional profiles.

Nevertheless all these studies suggest that diet has a major impact on the microbiota and raise hope that selective diets might also be able to affect dysbiosis observed in T2D and thereby modify and improve metabolic control. A thorough discussion of the effects of various diets on the microbiota is beyond the scope of this article.


There is increasing evidence that certain drugs might unexpectedly affect the gut’s microbiota. The biguanide drug metformin is widely used in the therapy of T2D. Metformin is able to slow aging in Caenorhabditis elegans by metabolic alteration of the E coli with which it is cultured.86 This effect involves disruption of the bacterial folate cycle, resulting in decreased levels of s-adenosylmethionine synthase and decelerated aging in the worm. Importantly, the effect on C elegans life span was completely dependent on the presence of bacteria. These findings might have human implications because of the widespread use of this drug. Common side effects include GI complaints such as diarrhoea and bloating, reduced folate levels and increased homocysteine levels.87 Indeed administration of metformin alters composition of the microbiota in rats.88 These data are in accordance with a recent study demonstrating that metformin is able to affect the mouse microbiota and increases concentrations of A muciniphila.74 Importantly, metformin increased the number of mucin-producing goblet cells and subsequent studies convincingly showed that administration of A muciniphila had beneficial effects on metabolic functions. This study therefore highlights that medications such as metformin might exert their functions at least partially via modulation of the gut’s microbiota.

Faecal transplant

One report recently showed that faecal transplant via a gastroduodenal tube from lean donors into obese subjects with metabolic syndrome demonstrated a small improvement of insulin sensitivity in recipients.89 Faecal transplant resulted in an increase of the microbial diversity after 6 weeks and a 2.5-fold increase, for example, in the butyrate producer R intestinalis whereas SCFA levels overall decreased. Despite this first evidence that such an approach could be attractive, much more information is needed to advance with larger, well-designed studies to prove whether such a crude approach is overall beneficial for patients with metabolic syndrome/T2D.

Gut microbiome and pregnancy

Pregnancy is characterised by profound hormonal, immunological and metabolic changes. Besides the marked immunological changes which result in, for example, an enhanced rate of of mucosal inflammatory events such as gingivitis or vaginosis, metabolic alterations are substantial and approximately 20% and more of patients develop prediabetes or manifest T2D. Earlier studies have revealed that the composition of the gut microbiota is changing over the course of gestation.90 A landmark study providing major insights into the relationship between microbiotal evolution during pregnancy and potential metabolic consequences has been reported.91 In this work the authors investigated the gut microbiota in the first (T1) and third trimesters (T3) and studied the metabolic effects of T1 and T3 microbiota on metabolic functions in germ-free mice. During pregnancy many metabolic parameters changed significantly with an increase in serum leptin levels, cholesterol, insulin, homeostasis model assessment indices and HbA1c levels. From T1 to T3 the relative abundance from Proteobacteria and Actinobacteria increased in approximately two-thirds of women. Levels of Bacteroidetes and Firmicutes were not significantly different between trimesters. T1 was characterised by a high rate of the Clostridiales order of the Firmicutes (eg, butyrate producers F prausnitzii and E rectale) whereas T3 was characterised by members of the Enterobateriaceae and Streptococcus genus. Interestingly, the expansion of ß-diversity between pregnant women was a widely shared phenomenon driven by pregnancy, independent of health status. Proteobacteria, enriched in T3 stools, have been shown to exert proinflammatory effects.92 Interestingly, stool cytokine levels were higher in T3 versus T1 stool samples suggesting a certain degree of mucosal inflammation probably due too changes in microbiotal composition. To investigate whether indeed T3 microbiota might affect inflammatory pathways, the authors transferred T1/T3 microbiotas to germ-free wild type mice. After 2 weeks of inoculation inflammatory mediators in the stool and caecal samples including lipocalin were significantly higher in the T3 versus T1 recipients. This was paralleled by more adiposity and an impaired glucose tolerance. Overall, this fascinating translational work clearly suggests that pregnancy is associated with major shifts in the gut’s microbiota characterised by a switch towards a proinflammatory tonus which consequently might lead to mucosal and systemic inflammation thereby potentially contributing to gestational diabetes.


Over the last years the intestinal microbiota has been identified as a fascinating ‘new organ’ which affects many biological systems throughout the body including the immune system, metabolic functions and development and programming of the nervous system. With the power of new sequencing technologies an enormous complexity and number of potential genes derived from this microbiota has been identified. Fascinating recent data have established a clear role for this microbiota including endotoxin in metabolic diseases such as obesity and T2D. Whereas initially considerable scepticism was present in the scientific community and data were solely generated from various rodent models, recent studies have demonstrated more and more that a ‘microbiotal gut signature’ is present in human obesity and in T2D. Although science into this direction has only yet started and available data are still somewhat contradictory, the search for intestinal bacteria which might play either beneficial or detrimental effects on metabolic pathways has started. A muciniphila reveals such a promising candidate which turned out recently to exert anti-inflammatory and beneficial metabolic functions. Mechanistically SCFAs, especially butyrate and propionate have evolved as attractive pathways to how the microbiota might shape immunological and metabolic functions. As clinical data are still in infancy and many insights so far based on rather preclinical work, it seems now mandatory to develop and establish precise and meaningful clinical sampling to clarify further the fascinating role of the microbiota in metabolic dysfunction. Exciting times are coming up in this research area for biologists, diabetologists and gastroenterologists.


View Abstract


  • Contributors Both authors wrote the manuscript. ARM has drawn both figures.

  • Competing interests None.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.