Intestinal microbiota

The intestinal tract is home to 10¹³–10¹⁴ micro-organisms, the vast majority of which are bacteria, the remainder being methanogenic archaea and yeasts⁽ Reference Blaut ^{8 )}. Intestinal bacteria belong to only few major phyla: Firmicutes (includes genera such as Clostridium, Eubacterium, Faecalibacterium, Ruminococcus and Roseburia), Bacteroidetes (includes genera such as Alistipes, Bacteroides, Parabacteroides, Porphyromonas and Prevotella), Actinobacteria (Bifidobacterium and Collinsella) Proteobacteria (mainly Escherichia coli and relatives), Verrucomicrobia (Akkermansia) and Fusobacteria (Fusobacterium). Firmicutes and Bacteroidetes account for up to 90 % of all bacterial cells in the human intestine. Colonisation of the intestinal tract by micro-organisms starts at birth following oral uptake of bacteria from the mother and from the environment. Usually, the microbial community in the gut ecosystem stabilises within the first 2 years of age. The microbiota in the digestive tract affect host physiology in many ways; for example, they confer colonisation resistance on the host and play an important role in the maturation of the innate and the adaptive immune system.

The gut microbiota may be considered as an additional organ whose metabolic potential exceeds that of the liver. Well-known catalytic activities include the conversion of host-derived substances such as bile acids, of non-nutritive dietary components such as secondary plant metabolites and of xenobiotics (drugs). One core activity of the microbiome relates to the degradation of carbohydrates that escape digestion in the small intestine. The metagenome, which encompasses all genes of all microbial community members, contains a large proportion of genes involved in the breakdown of carbohydrates⁽ Reference Tasse, Bercovici and Pizzut-Serin ^{9 )}. Such genes are overrepresented in the human intestinal microbiome⁽ Reference Gill, Pop and Deboy ^{10 )} indicating their importance for the ecosystem. These genes encode a large array of proteins that catalyse the depolymerisation of non-digestible polysaccharides and their subsequent conversion to SCFA, preferentially acetic acid, propionic acid and butyric acid. Non-digestible but fermentable carbohydrates may be extremely variable with respect to their structure as well as physical and chemical properties. This high variability is related to the large number of possible monomers and linkages. Intestinal bacteria express a wide variety of enzymes that afford the depolymerisation of complex carbohydrates⁽ Reference Sonnenburg, Xu and Leip ^{11 )}. Bacterial population groups involved in this process may compete for a given substrate among each other but they may also catalyse complementary reactions. Important functions of the human gut microbiome are redundant, i.e. more than one species is capable of catalysing a given reaction in the breakdown of complex carbohydrates. Redundancy of metabolic functions confers stability on the ecosystem. Such functions may be considered as core activities of the microbiota. It is important to note that individuals may differ in the bacterial species that catalyse the degradation of a given substrate. Accordingly, the metagenomes, which reflect the metabolic functions of the respective microbiota, display a high degree of similarity among human subjects, while at the same time the corresponding microbial communities are highly variable in composition. In other words, taxonomic variability of the gut microbiota is greater than functional variability⁽ Reference Turnbaugh, Hamady and Yatsunenko ^{12 )}. Genes encoding core activities are found in the microbiome of every individual. Other activities, for example, the ability to convert the isoflavone daidzein, a polyphenol found in soya, to equol is restricted to certain individuals⁽ Reference Matthies, Blaut and Braune ^{13 )} and may therefore not be considered a core activity.

Bacterial degradation of dietary fibre in the intestinal tract

Dietary fibre refers to plant material that escapes digestion in the small intestine. The main components of dietary fibre are matrix polysaccharides that may be a part of complex structures, such as plant cell walls, but they may also serve as storage materials in the plant. Dietary fibre is found in whole grains, fruit and vegetables. Various types of dietary fibre can be distinguished based on their chemical structure (e.g. type of monomers and type of linkage), physical properties (e.g. solubility, viscosity and water-holding capacity) and on whether they can be fermented by intestinal microbiota⁽ Reference Ha, Jarvis and Mann ^{14 )}. Solubility of the respective dietary fibre is a major factor that affects their fermentability by intestinal bacteria, but is not sufficient to predict its physiologic effects. There is a discrepancy between the chemically determined amounts of fibre and its physiologic effects⁽ Reference Slavin ^{15 )}. Cellulose, a β-glucan, is essentially insoluble and not or only weakly degraded by intestinal bacteria. It may nevertheless affect intestinal physiology, for example, by attenuating the rise in blood glucose after a meal or by exerting a laxative effect owing to its water-holding capacity. Other dietary fibres of relevance to human nutrition include hemicelluloses (a wide variety of hexosans and pentosans), pectins (chains of α-(1–4)-linked d-galacturonic acid, the carboxyl groups of which may be esterified to varying degrees with methanol), guar gum (galactomannan) and fructans such as inulin and oligofructose. Depending on the individual nutrition pattern, such dietary fibres make up a smaller or larger part of a typical human diet and they may, to a smaller or larger extent serve as growth substrates for intestinal bacteria. However, in human subjects who consume a typical Western-type diet, resistant starch is the most important substrate of intestinal microbiota⁽ Reference Cummings ^{16 )}. In human nutrition, starch is the main carbohydrate source and normally readily digested in the small intestine. However, seeds, legumes and unprocessed whole grains may contain starch that is physically inaccessible and therefore not degraded by host enzymes. Starch which escapes digestion in the small intestine is referred to as resistant starch. Another form of resistant starch includes starch that occurs in its natural granular form and can be found in uncooked potatoes or unripe bananas. Resistant starch can also be formed during cooking and cooling of starch containing foods and is referred to as retrograded starch⁽ Reference Englyst, Kingman and Hudson ^{17 )}.

The large variety of non-digestible but fermentable substrates is probably one reason for the high diversity of the microbial community in the intestine. The first step in the bacterial breakdown of dietary fibre (Fig. 1) involves de-branching and de-polymerising enzymes such as cellulase, xylanase, arabinofuranosidase, xylosidase, arabinogalactanase, arabinofuranosidase, galactosidase, pectinase, exopectate lyase, glucuronidase, β-glucanase, β-glucosidase, β-galactosidase, fructooligosaccharidase and pullulanase⁽ Reference Tasse, Bercovici and Pizzut-Serin ^{9 )}. The resulting oligosaccharides and monosaccharides are either taken up by bacteria that contribute to their degradation or by bacteria that do not possess the required enzymes and therefore profit from bacteria that make the monomeric carbohydrates available. The latter process is referred to as cross-feeding⁽ Reference Flint, Duncan and Scott ^{18 )}. Intestinal bacteria use different pathways for the degradation of the resulting monomeric or oligomeric carbohydrates. Altogether these pathways lead to the formation of SCFA, mainly acetate, propionate and butyrate; other fermentation products such as lactate, succinate and ethanol are intermediates, the majority of which are also degraded to SCFA⁽ Reference Blaut ^{8 )}. Side products of bacterial fermentation in the digestive tract are carbon dioxide, molecular hydrogen and formate. The latter may be further converted to acetate by bacteria such as Eggerthella lenta or Blautia producta, or to methane by the archaeon Methanobrevibacter smithii (Fig. 1).

Fig. 1. Scheme depicting various steps in the fermentation of indigestible polysaccharides by intestinal microbiota.

SCFA are major products of bacterial fermentation of dietary fibre

The most important products arising from bacterial breakdown of fermentable carbohydrates in the colon are SCFA, 95 % of which are acetate, propionate and butyrate, which are present at a molar ratio of approximately 60:23:17. Total SCFA concentrations in the proximal and the distal colon are approximately 120 and 90 mm, respectively⁽ Reference Cummings, Pomare and Branch ^{19 )}. Even though the SCFA concentration in the distal colon is still relatively high (because fermentation of colonic contents during passage goes on), 95 % of the SCFA formed in the colon are absorbed⁽ Reference Topping and Clifton ^{20 )}. Both the colonic SCFA concentrations and the molar SCFA ratios vary considerably in response to the type of dietary fibre ingested. The dietary fibre intake in European subjects was reported to be on average 20–25 g/d but could be as high as 60 g/d⁽ Reference Bingham, Day and Luben ^{21 )}. The latter value can probably be reached when a diet rich in whole-grain products, fruit and vegetables is consumed. Based on the analysis of ileal effluents from ileostomy patients and obtained by terminal ileal intubation, it has been estimated that up to 9 % of starch in a meal is malabsorbed and passes into the colon⁽ Reference McNeil ^{22 )} where it undergoes bacterial fermentation. Accordingly, the total amount of SCFA produced in the human colon has been estimated to be 500–600 mmol/d with a total energy value of 600–750 kJ/d⁽ Reference McNeil ^{22 )}, corresponding to 6·9–8·6 % of an assumed average adult diet of 8700 kJ. However, this estimation appears too high (mentioned later).

The energy content of monomers derived from indigestible but fermentable carbohydrates would be similar to that of glucose if the host possessed enzymes for cleavage and uptake of the resulting monomers. The energy content extracted from dietary fibre and available for the host is considerably smaller because approximately 40 % of the carbon from fermentable carbohydrates is converted into bacterial biomass which is excreted⁽ Reference Roberfroid, Gibson and Delzenne ^{23 )}. The remainder is converted to carbon dioxide (5–10 %), some lactate (3–5 %) and SCFA (45–50 %). The energy gained by the oxidation of SCFA, which are formed from bacterial fermentation of one hexose moiety from non-digestible carbohydrates such as oligofructose, has been calculated to be approximately 25–35 % of what would be gained if fructose was absorbed in the small intestine (approximately 15·7 kJ/g fructose)⁽ Reference Roberfroid, Gibson and Delzenne ^{23 )}. An assumed daily intake of 40 g fermentable carbohydrates, i.e. NSP (20 g) plus resistant starch (20 g), would contribute approximately 188 kJ to the daily energy requirement (15·7 kJ/g × 40 g × 30 %), which corresponds to 2·2 % of an adult diet of 8700 kJ. However, it has to be noted that the amount of fermentable fibre entering the colon very much depends on dietary habits. For example, 2–6-year-old children from a rural area in Africa on a traditional diet ingested daily 14·2 g fibre at a total food intake of 445 g (3·2 %), whereas Italian children of the same age on a European diet ingested daily 8·4 g fibre at a total food intake of 925 g (0·9 %)⁽ Reference De Filippo, Cavalieri and Di Paola ^{24 )}.

SCFA formed in the colon by bacterial fermentation are absorbed and subsequently used in various tissues (Fig. 2). Acetate and butyrate are preferentially used for energy generation in all body tissues, whereas propionate preferentially enters the gluconeogenic pathway into the liver⁽ Reference Cummings, Gibson and Macfarlane ^{25 )}. However, acetate can also be used for lipogenesis. SCFA provide 60–70 % of the energy required by isolated colonocytes with butyrate being the preferred energy source for these cells⁽ Reference Roediger ^{26 )}. However, besides serving as an energy source, butyrate influences gene expression by inhibiting histone deacetylase, which in colon cancer cells results in cell cycle arrest and activation of apoptosis⁽ Reference Lazarova, Chiaro and Wong ^{27 )}. Therefore, SCFA are considered important for keeping the colonic mucosa in a healthy state.

Fig. 2. Simplified scheme showing the roles of SCFA in host metabolism. On the one hand, SCFA provide energy to the host and play a role in lipogenesis; thereby, SCFA could contribute to adiposity. On the other hand, by activating free fatty acid receptors (FFAR)2 and FFAR3, SCFA can trigger the increased release of peptide YY, glucagon-like peptide 1 and leptin, which would result in increased satiety which eventually may help to prevent/reduce adiposity. However, peptide YY also slows down gut transit time which would result in improved energy extraction and contribute to obesity. However, which of these effects is dominant, is still unclear.

Intestinal microbiota and obesity

The prevalence of obesity and associated diseases is worldwide on the rise. Even though genetic susceptibility plays an important role, this increase in obesity has been mainly attributed to considerable changes in lifestyle during the past decades. Important factors include excessive consumption of energy-rich diets, lack of physical activity, a sedentary lifestyle and unhealthy eating behaviour⁽ Reference Kushner and Choi ^{28 )}. One decade ago, the intestinal microbiota was identified as another factor involved in obesity development⁽ Reference Backhed, Ding and Wang ^{29 )}. This notion has mainly been based on two observations: (1) gut microbiota composition differed between lean and obese human subjects or mice⁽ Reference Ley, Turnbaugh and Klein ^{3 ,} Reference Turnbaugh, Backhed and Fulton ^{30 )}; (2) the obese phenotype could be transferred by transplantation of the microbiota from obese mice or human subjects to recipient germ-free mice⁽ Reference Turnbaugh, Ley and Mahowald ^{5 ,} Reference Ridaura, Faith and Rey ^{6 )}. These results were taken as an indication that the microbiome of obese donors has an increased capacity to extract energy from the diet compared with the microbiome of lean donors⁽ Reference Turnbaugh, Backhed and Fulton ^{30 )}. Several mechanisms have been proposed to account for these observations⁽ Reference DiBaise, Zhang and Crowell ^{31 )}: (1) the fermentation of indigestible carbohydrates by intestinal microbiota leads to an increased intestinal absorption of monosaccharides and SCFA followed by increased hepatic lipogenesis; (2) high-fat diets trigger an increased transfer of bacterial lipopolysaccharide (LPS) from the intestinal lumen to the blood causing metabolic endotoxaemia and low-grade inflammation⁽ Reference Cani, Amar and Iglesias ^{32 )}; (3) the gut microbiota suppress the formation of angiopoietin-like protein 4, also known as fasting-induced adipose factor, in gut epithelium. As angiopoietin-like protein 4 is an inhibitor of lipoprotein lipase, its suppression would lead to increased TAG storage in adipose tissue. Only the first two points will be discussed here. The intestinal microbiome affects glucose and lipid metabolism as well as satiety. This may be concluded from a comparison of germ-free and conventional mice, which revealed that conventional mice display higher serum glucose and liver TAG levels compared with germ-free mice as well as higher levels of the hormones leptin and insulin in serum⁽ Reference Backhed, Ding and Wang ^{29 )}. Leptin is produced by adipocytes and plays a major role in the regulation of body weight⁽ Reference Friedman and Halaas ^{33 )}. Leptin inhibits hunger and increases energy expenditure by means of activating leptin receptors in the brain. In agreement with the higher leptin levels found in conventional mice compared with germ-free mice oxygen consumption, which reflects energy expenditure, was 27 % lower in the latter mice⁽ Reference Backhed, Ding and Wang ^{29 )}. Such a difference in total energy expenditure monitored over 24 h was also observed in another mouse study, but interestingly this difference was only observed during night time, when the mice are active⁽ Reference Fleissner, Huebel and Abd El-Bary ^{34 )}.

Do SCFA contribute to or rather prevent obesity development?

SCFA formed through bacterial fermentation of indigestible carbohydrates provide additional energy to the host which potentially could contribute to obesity development (Fig. 2). However, epidemiological studies rather indicate that a diet rich in fibre correlates with a lower incidence of obesity and symptoms of the metabolic syndrome⁽ Reference Slavin ^{15 )}. For example, a prospective cohort study involving 161 737 US women showed an inverse association between whole-grain intake and type 2 diabetes⁽ Reference de Munter, Hu and Spiegelman ^{35 )}. The results of intervention studies which investigated the effects of dietary fibre on weight loss are inconsistent. Although a number of studies reported an improvement of weight loss in response to dietary fibre supplementation, others failed to show significant effects on weight loss⁽ Reference Slavin ^{15 )}. These discrepancies might be due to differences in study design and the type of dietary fibre consumed by the study subjects. In principle, there are two opposing effects: on the one hand, SCFA derived from dietary fibre provide additional energy which could contribute to obesity development; on the other hand, dietary fibre decreases the energy density of the diet. Which of the two effects is of more importance probably depends on the type of dietary fibre that is ingested and possibly the duration of intake.

Theoretically, non-fermentable fibre reduces the energy density of a diet to a larger extent than a fermentable fibre because only the latter can be converted to SCFA. This consideration is supported by a mouse study in which the effect of guar gum, which is soluble and fermentable, was compared with that of a largely (93 %) insoluble and non-fermentable oat fibre fraction⁽ Reference Isken, Klaus and Osterhoff ^{36 )}. The study investigated in obesity prone C57BL/6J mice whether long term feeding (45 weeks) of a high-fat diet (macronutrient energy: protein 18 %; carbohydrates 41 %; fat: 41 %) supplemented with either 10 % guar gum or 10 % insoluble fraction of oat fibre affected body weight, liver fat, insulin sensitivity and gene expression of metabolic markers in liver and adipose tissue. Body weight increased to a similar extent in both the mouse groups during the first 10 weeks. Thereafter the mice on the guar gum supplemented diet gained significantly more body weight compared with the mice fed on the diet supplemented with the non-fermentable oat fibre fraction (42 g v. 33 g after 45 weeks). Although dietary intake was comparable, the body fat content did not differ between the groups after 15 weeks, but after 43 weeks, it was 35 % lower in the oat fibre group than in the guar gum group (9 g v. 14 g). Markers of insulin resistance were higher in the mice fed the guar gum containing diet than in the mice fed the diet containing oat fibre. In agreement with a role of SCFA in lipogenesis, liver TAG levels in the guar gum group were elevated compared with the oat fibre group. Furthermore, gene expression analysis in liver indicated increased fatty acid oxidation in the mice fed the non-fermentable oat fibre. Excreted hydrogen as a marker of bacterial fermentation was significantly elevated in the guar gum group (24 ppm) compared with the oat fibre group (2 ppm). The latter indicates that the oat fibre was poorly fermented and accordingly, hardly any SCFA were produced. In contrast, guar gum was well fermented leading to increased SCFA formation, which in turn contributed to an increase in digested energy. Hence, in a direct comparison of a fermentable and a non-fermentable fibre, the latter resulted significantly in lower weight gain and improved insulin sensitivity in mice fed a high-fat diet. However, the difference between the two fibres became evident only after long-term feeding.

Several short-term animal studies reported that soluble fibres such as guar gum and psyllium were more effective in improving insulin sensitivity than insoluble cellulose⁽ Reference Cameron-Smith, Habito and Barnett ^{37 ,} Reference Song, Sawamura and Ikeda ^{38 )}. However, beneficial effects observed in rats after short-term feeding of guar gum disappeared after 67 weeks, whereas rats fed cellulose still displayed lower pancreatic insulin and glucagon concentrations and a slightly reduced body weight after 67 weeks⁽ Reference Track, Cawkwell and Chin ^{39 )}. Taken together these results indicate that short- and long-term effects of dietary fibres may differ.

Besides serving as energy source acetate, propionate and butyrate have been recognised as ligands of G-protein-coupled receptors FFAR2 and FFAR3 (formerly G-protein-coupled receptor 43 and G-protein-coupled receptor 41⁽ Reference Brown, Goldsworthy and Barnes ^{40 )}). These receptors are expressed in enteroendocrine L cells of ileum and colon as well as in adipocytes and immune cells. Activation of FFAR2 in adipocytes triggers the release of leptin from adipocytes⁽ Reference Xiong, Miyamoto and Shibata ^{41 )} and the secretion of peptide YY (PYY) from enteroendocrine cells⁽ Reference Tazoe, Otomo and Kaji ^{42 )}. Both leptin and PYY are known for their anorexigenic effects, i.e. their ability to reduce appetite⁽ Reference Wren and Bloom ^{43 )}. Interestingly, mice deficient in FFAR3 (Ffar3 ^−/−) and maintained on a polysaccharide-rich low-fat chow diet, gained 30 % less body weight and had 25 % less body fat than the corresponding wild-type mice even though they did not differ in the amount of chow consumed. These results suggest that FFAR3 activation in these mice led to an increased nutrient extraction from the diet⁽ Reference Samuel, Shaito and Motoike ^{44 )}. This was attributed to elevated PYY serum levels as a result of SCFA-mediated FFAR3 activation, because PYY not only has anorexigenic effects but also slows down gut transit time⁽ Reference Lin, Neevel and Chen ^{45 )}. Prolonged transit time in turn enhances the efficiency of energy extraction. Accordingly, a comparison of the energy content of faeces from Ffar3 ^−/− mice and wild-type mice using bomb-calorimetric analysis revealed a more efficient extraction of energy in the Ffar3^+/+ mice⁽ Reference Samuel, Shaito and Motoike ^{44 )}. PYY concentrations in germ-free mice were 40 % lower compared with conventional mice supporting a role of intestinal SCFA in FFAR activation⁽ Reference Samuel, Shaito and Motoike ^{44 )}. It appears that in this experiment the effect of PYY on gut motility dominated over its anorexigenic effect. Since FFAR activation enhances leptin formation by adipocytes, it is not surprising that Ffar3 ^−/− mice displayed significantly lower leptin levels than the corresponding wild-type mice. This could only in part be attributed to the lower fat mass of these animals.

Acetate and propionate enhance the secretion of glucagon-like peptide 1 (GLP-1) in primary murine colonic cell cultures⁽ Reference Tolhurst, Heffron and Lam ^{46 )}. Similar to PYY, GLP-1 is produced by enteroendocrine L cells. Besides stimulating the production of insulin in pancreatic β-cells, GLP-1 also promotes insulin sensitivity and satiety. Mice deficient in FFAR2 or FFAR3 exhibit low levels of circulating GLP-1 and impaired glucose tolerance suggesting that, SCFA play an important role in glucose homeostasis⁽ Reference Tolhurst, Heffron and Lam ^{46 )}. In agreement with this finding, increased wheat fibre intake by hyperinsulinaemic human subjects resulted in higher postprandial plasma SCFA and GLP-1 concentrations compared with a control group, but this effect became only detectable after 9 and 12 months of intervention⁽ Reference Freeland, Wilson and Wolever ^{47 )}. However, it has to be noted that some results concerning the role of FFAR2 are in part controversial: in contrast to Tolhurst et al.⁽ Reference Tolhurst, Heffron and Lam ^{46 )}, Bjursell et al ⁽ Reference Bjursell, Admyre and Goransson ^{48 )} did not find a significant difference in glucose tolerance in Ffar2 ^−/− mice fed a chow diet compared to Ffar2^+/+ mice. On a high-fat diet, these Ffar2 ^−/− mice had a lower body fat mass in spite of a higher food intake compared with wild-type controls. The lower food intake in the Ffar2^+/+ mice compared with the Ffar2 ^−/− mice is in agreement with an SCFA-dependent FFAR2 activation which triggered an increase in anorexigenic GLP-1 and PYY. However, the remaining discrepancies between the studies have not yet been resolved and need clarification. A very recent investigation showed that Ffar2 ^−/− mice fed a normal diet become obese, while mice overexpressing FFAR2 remained lean⁽ Reference Kimura, Ozawa and Inoue ^{49 )}. However, these two mouse strains did not show their respective phenotypes when kept germ-free. Further analyses revealed that FFAR2 activation in adipocytes suppressed insulin signalling, which in turn reduced fat accumulation in adipocytes and the utilisation of lipids and glucose in other tissues. Based on their study the authors proposed that FFAR2 serves as a sensor that contributes to energy homeostasis⁽ Reference Kimura, Ozawa and Inoue ^{49 )}.

Role of oligofructose in the amelioration of diet-induced obesity and symptoms of the metabolic syndrome

The consumption of high-fat diets is associated with increased plasma LPS levels in human subjects⁽ Reference Erridge, Attina and Spickett ^{50 )} and mice⁽ Reference Cani, Amar and Iglesias ^{32 )} referred to as endotoxaemia. The latter is associated with elevated levels of inflammatory markers, indicative of low-grade inflammation, and with the onset of symptoms of the metabolic syndrome. Increased plasma LPS levels have been attributed to diets rich in fat which lead to increased gut permeability facilitating the transfer of LPS from the gut lumen into the blood stream. Interestingly, oligfructose, a non-digestible fermentable carbohydrate, was reported in mice to reduce gut permeability, decrease plasma LPS levels and ameliorate symptoms of the metabolic syndrome⁽ Reference Cani, Neyrinck and Fava ^{51 )}. Since these changes coincided with an increase in faecal bifidobacteria, the health-promoting effects were attributed to this bacterial population group. However, experimental evidence for a role of bifidobacteria in oligofructose-dependent amelioration of metabolic disease symptoms has not been presented. We investigated the potential role of bifidobacteria in a gnotobiotic mouse model. The mice were either colonised with a microbial community encompassing eight bacterial species, including Bifidobacterium longum or with the same community without this bacterium. Germ-free mice served as a control group. All three groups were fed either a high-fat diet or a high-fat diet supplemented with 10 % oligofructose. After 4 weeks feeding, mice fed the diet supplemented with 10 % oligofructose gained significantly less body weight and had significantly less body fat than the mice fed the high-fat diet without oligofructose supplementation (unpublished results). Since this effect was independent of the microbial status, the idea that bifidobacteria mediate the health-promoting effect of oligofructose has to be questioned.