Selective increases of bifidobacteria in gut microflora improve high-fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia
Recent evidence suggests that a particular gut microbial community may favour occurrence of the metabolic diseases. Recently, we reported that high-fat (HF) feeding was associated with higher endotoxaemia and lower Bifidobacterium species (spp.) caecal content in mice. We therefore tested whether restoration of the quantity of caecal Bifidobacterium spp. could modulate metabolic endotoxaemia, the inflammatory tone and the development of diabetes.
Since bifidobacteria have been reported to reduce intestinal endotoxin levels and improve mucosal barrier function, we specifically increased the gut bifidobacterial content of HF-diet-fed mice through the use of a prebiotic (oligofructose [OFS]).
Compared with normal chow-fed control mice, HF feeding significantly reduced intestinal Gram-negative and Gram-positive bacteria including levels of bifidobacteria, a dominant member of the intestinal microbiota, which is seen as physiologically positive. As expected, HF-OFS-fed mice had totally restored quantities of bifidobacteria. HF-feeding significantly increased endotoxaemia, which was normalised to control levels in HF-OFS-treated mice. Multiple-correlation analyses showed that endotoxaemia significantly and negatively correlated with Bifidobacterium spp., but no relationship was seen between endotoxaemia and any other bacterial group. Finally, in HF-OFS-treated-mice, Bifidobacterium spp. significantly and positively correlated with improved glucose tolerance, glucose-induced insulin secretion and normalised inflammatory tone (decreased endotoxaemia, plasma and adipose tissue proinflammatory cytokines).
Together, these findings suggest that the gut microbiota contribute towards the pathophysiological regulation of endotoxaemia and set the tone of inflammation for occurrence of diabetes and/or obesity. Thus, it would be useful to develop specific strategies for modifying gut microbiota in favour of bifidobacteria to prevent the deleterious effect of HF-diet-induced metabolic diseases.
KeywordsBifidobacteria Diabetes Endotoxin GLP-1 Glucagon-like peptide-1 Gut microflora Inflammation Obesity Prebiotics
mouse intestinal bacteria
Evidence suggests that obesity and metabolic disorders (type 2 diabetes and insulin resistance) are tightly linked to inflammation [1, 2]. Although obesity results from interactions between genetic and environmental factors, eating habits contributing to an increase in fat intake can promote metabolic diseases.
Cytokines have been shown to cause insulin resistance [5, 6], which favours hyperinsulinaemia and excessive hepatic and adipose tissue lipid storage. However, while extensive research has been dedicated to the effects of inflammatory reactions on energy metabolism, the triggering factor linking inflammation to a high-fat (HF)-diet-induced metabolic syndrome remains to be determined.
An innovative hypothesis was recently proposed: the gut flora could be an important factor affecting energy disposal and could be implicated in metabolic disease associated with obesity [7, 8, 9, 10, 11]. One observation reported that young adult germ-free mice had 40% less body fat than their conventionalised counterparts on the same diet. Similarly, lean axenic mice colonised with microbiota from ob/ob mice increased body weight . The authors suggested that gut microbiota from obese mice allowed energy to be salvaged from otherwise non-digestible dietary polysaccharides producing substantial elevations in serum glucose and insulin, both factors that trigger lipogenesis. However, this hypothesis of microbiota-induced lipogenesis did not explain the differential effect on body weight of a HF diet vs regular chow, since axenic mice fed a HF diet did not gain weight. This suggests that a bacterially related factor was responsible for HF-diet-induced obesity  and that lipids alone were not sufficient to promote obesity and inflammation.
The relationship between HF feeding and the development of a low-grade inflammatory tone is unclear, but may be related to the microbiota present in the digestive tract. With regard to the role of gut microbiota in the development of metabolic diseases, we recently reported  that HF feeding in mice induced a low-grade inflammatory tone and that metabolic disease was associated with reduced numbers of Bifidobacterium species (spp.) in caecal content and higher plasma endotoxin (lipopolysaccharide [LPS]) concentrations (defined as metabolic endotoxaemia). Hence, LPS appears to be a molecular link between HF feeding, microbiota and inflammation. LPS was identified as a novel factor triggering the onset of HF diet-induced obesity and type 2 diabetes . We have also shown that mice lacking the major LPS co-receptor CD14 were resistant to HF-diet-induced inflammation and metabolic diseases . Since bifidobacteria reduced intestinal endotoxin levels and improved mucosal barrier function [13, 14, 15], we decided to specifically increase the gut bifidobacterial content of HF-diet-treated mice through prebiotic dietary fibre (oligofructose [OFS]), which can reproducibly modulate the gut microbiota typified by increased bifidobacteria .
We therefore tested the hypothesis that the pattern of bacterial changes occurring through dietary intervention with OFS may control the occurrence of metabolic diseases in mice fed a HF diet for 14 weeks.
Male C57bl6/J mice (12 weeks old; Charles River, Brussels, Belgium) were housed in a controlled environment (12 h daylight cycle, lights off at 18:00 hours) with free access to food and water. All mouse experiments were approved by the local ethics committee and housing conditions were as specified by the Belgian Law of 14 November 1993 on the protection of laboratory animals (agreement number LA 1230314). Mice were killed by cervical dislocation after a 5 h period of fasting. Caecum (full and empty) and adipose tissues (epididymal, subcutaneous and visceral) were precisely dissected, weighed, immersed in liquid nitrogen and stored at −80°C, for further analysis.
Diet and experimental groups
Mice (n = 8 per group) were fed four different experimental diets for 14 weeks as follows: (1) standard diet (control) (A04; UAR, Villemoisson-sur-Orge, France) ; (2) HF diet containing 49.5% fat (g/100 g of total dry diet) corresponding to 72% of the total energy content (UAR, Epinay-sur-Orge, France); (3) a mix of HF diet and a fermentable dietary fibre (OFS) (HF-OFS) (kind gift from Orafti, Tienen, Belgium) ; and (4) a mix of HF diet and a non-fermentable dietary fibre (microcrystalline cellulose [HF-Cell]) (Vivapur microcrystalline cellulose; J. Rettenmaier & Söhne, Weissenborn, Germany). Dietary fibre was added in a proportion of 90:10 (weight of HF diet:weight of fibre). Nutritional compositions of the experimental diets are provided in the Electronic supplementary material (ESM Table 1).
Glucose tolerance test
Oral glucose tolerance tests were performed after 13 weeks of treatment in mice that had been fasted for 6 h. Glucose was orally administered (3 g/kg body weight, 660 g/l glucose solution) and blood glucose determined through a glucose meter (Roche Diagnostics, Meylan, France) using 3.5 μl of blood collected from the tip of the tail vein before and at administration of glucose load (−30 and 0 min) and after glucose load (15, 30, 60, 90, 120 min). To assess plasma insulin concentration, 20 μl of blood was sampled 30 min before and 15 min following the glucose load.
Plasma LPS determinations were performed using a kit based on a Limulus amoebocyte extract (LAL kit; Cambrex BioScience, Walkersville, MD, USA); samples were diluted 1:50 and heated for 10 min at 70°C. Plasma insulin concentration was determined in 5 μl of plasma using an ELISA kit (Mercodia, Uppsala, Sweden) according to the manufacturer’s instructions. Cytokines were determined in 12 μl of plasma using a kit (Bio-Plex Multiplex; Bio-Rad, Nazareth, Belgium) and measured using Luminex technology (Bio-Plex; Bio-Rad).
Microbial quantification in intestinal contents by fluorescence in situ hybridisation (FISH) analysis
Caecal contents collected post mortem from mice were stored at −70°C. Samples were thawed on ice and diluted 1:10 in sterile ice cold PBS (0.1 mol/l phosphate, pH 7.0) (0.1 mol/l phosphate, pH 7.0). The suspension was homogenised by pipetting and centrifuging at 3,500×g for 15 min to remove particulate matter. The supernatant fraction containing bacterial cells was fixed overnight in 4% (w/v) paraformaldehyde. The bacterial cells were then washed and re-suspended twice in sterile PBS and finally stored in 500 ml ethanol/l PBS at −20°C until hybridisation with appropriate molecular probes targeting specific regions of 16S rRNA. The probes used were: EREC 482  for Eubacterium rectale–Clostridium coccoides group, Lab158  for lactobacilli and enterococci and Bif164  for Bifidobacterium spp. (Gram-positive bacteria); and Bac303  for Bacteroides–Prevotella spp., MIB 661  for Bacteroides mouse intestinal bacteria (MIB), SRB 687  for sulphate-reducing bacteria and probe D  for Enterobacteriaceae spp. (Gram-negative bacteria) within the murine intestinal microflora. The nucleic acid stain DAPI was used for total bacterial counts. The DNA probes were tagged with the Cy3 fluorescence dye so that the hybridised samples could be examined using fluorescence microscopy. Results were expressed as Log10 (bacterial cells per g caecal content wet weight).
Real-time quantitative PCR
Total RNAs from white visceral adipose tissue were prepared using a reagent (TriPure; Roche, Basel, Switzerland) as described . PCRs were performed using a sequence detection system instrument and software (AbiPrism 5700; Applied Biosystems, Foster City, CA, USA) as described . Primer sequences for the targeted mouse genes are available upon request (firstname.lastname@example.org).
Results are presented as mean ± SEM. Statistical significance of differences was analysed by one-way ANOVA followed by post hoc (Bonferroni’s multiple comparison test). Correlations between parameters where assessed by Pearson’s correlation test, using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA, USA). A value of p < 0.05 was regarded as statistically significant.
Specific increase of bifidobacteria by prebiotics
Prebiotics lower endotoxaemia
Body weight, plasma, adipose tissue inflammatory markers and GLP-1 precursor mRNA
Body weight gain (g)
4.82 ± 0.52a
11.56 ± 1.34b
8.03 ± 0.52c
7.74 ± 0.57c
Daily energy intake (kJ/mouse)
37.82 ± 2.46a
79.04 ± 6.22b
64.03 ± 1.29b,*
61.02 ± 1.21b,*
Adipose tissue weight (g/100 g body weight)
0.59 ± 0.05a
1.16 ± 0.11b
0.89 ± 0.03a
0.90 ± 0.05a
1.38 ± 0.1a
4.74 ± 0.54b
3.21 ± 0.26c
3.16 ± 0.22c
1.42 ± 0.08a
5.28 ± 0.65b
3.85 ± 0.47c
3.18 ± 0.14c
Plasma endotoxin (EU/ml)
5.02 ± 0.56a
9.03 ± 1.25b
11.28 ± 3.52b
6.16 ± 0.47a
Visceral adipose tissue mRNA (AU)
3.59 ± 0.68a
5.54 ± 2.24a
5.88 ± 1.57a
2.33 ± 0.69a,**
3.98 ± 0.83a
17.44 ± 3.74b
65.89 ± 25.79c
11.58 ± 4.92a,b
2.88 ± 0.63a
4.63 ± 1.67a
4.40 ± 1.21a
1.97 ± 0.67a
Colon proglucagon mRNA (AU)
1.49 ± 0.55a
1.87 ± 0.43a
2.78 ± 0.20a
6.96 ± 0.95b
Gram-positive Bifidobacterium spp. negatively correlated with endotoxaemia
Prebiotics improve glucose tolerance and restore glucose-induced insulin secretion
Prebiotic supplementation improves body weight gain and energy intake, reduces fat mass development and increases colonic glucagon-like peptide-1 precursor
Prebiotics control high-fat-diet-induced inflammation
Recent studies have highlighted key mammalian host–gut microbial relationships, suggesting that the gut microbiota play an important role in energy metabolism . Moreover, our recent studies have demonstrated that HF diet-fed mice developed insulin resistance and inflammation by a mechanism directly dependent on a gut bacterial Gram-negative derived compound, namely LPS . Here we report that HF feeding alters the intestinal microbiota composition: quantities of the dominant Gram-positive groups Bifidobacterium spp. and E. rectale–C. coccoides group were reduced, as were the numbers of the numerically dominant murine Gram-negative group, Bacteroides MIB, compared with controls fed a standard diet. Moreover, we found that among the different gut bacteria analysed, plasma LPS concentrations correlated negatively with Bifidobacterium spp. Since, bifidobacteria have been shown to reduce intestinal endotoxin levels and improve mucosal barrier function [13, 14, 15], we decided to specifically increase the gut bifidobacterial content of HF-diet-treated mice through supplementation with prebiotic dietary fibre (OFS). Mice fed the prebiotic dietary fibre had normal endotoxaemia. This observation strongly correlated with improved glucose tolerance as explained by the normalisation of glucose-induced insulin secretion. Although glucose-induced insulin secretion was normalised in HF-OFS mice, glucose tolerance following a glucose load was improved, but did not reach control values. This suggests that OFS-fed mice remained insulin resistant. According to these results, we have previously shown that HF-diet-induced hepatic insulin resistance was normalised in OFS fed mice, without significant improvement in whole-body insulin sensitivity . Moreover, the improved hepatic insulin sensitivity observed in HF-OFS fed mice was strongly associated with a significantly lower hepatic inflammatory tone .
Since we observed higher endotoxaemia in HF-diet-fed mice, we measured plasma and adipose tissue inflammatory markers. We found that HF-diet feeding increased plasma IL-1α, IL-1β and IL-6 concentrations, but this increase was prevented only in prebiotic treated mice. It seems that bifidobacteria and endotoxaemia are therefore linked to metabolic disorders. In line with our results, Brun et al. have recently demonstrated that obese and diabetic mice (ob/ob and db/db) exhibited significantly higher plasma endotoxin levels, which correlated to higher plasma inflammatory markers . Multiple Pearson’s correlations analysis of our data revealed that endotoxaemia positively correlated with glucose intolerance, fasted insulinaemia, inflammatory markers, adipose tissue and body weight gain, while bifidobacteria negatively correlated with these parameters. Collectively, our data indicate that HF-diet-induced metabolic disorders could be modulated through a number of related aspects.
We have shown that increased endotoxaemia observed during HF feeding could be one outcome of the modulation of gut microbiota. Moreover, we cannot exclude that intestinal mucosa permeability could be increased following HF feeding and restored through prebiotic treatment. The exact mechanism by which prebiotic fibre lowers endotoxaemia and systemic inflammatory tone is poorly understood. It is recognised that bifidobacteria are the main gut bacteria involved in the positive effects observed after prebiotic supplementation. Indeed, bifidobacterial supplementation has been associated with lower bacterial translocation and endotoxaemia, leading to a decrease of the inflammatory cascade activation in several models of gut bacteria translocation [13, 14, 15, 31]. Similarly, other studies have reported that probiotics and gut microbiota modulation through prebiotic ingestion may improve or prevent disruption of intestinal epithelial barrier function and consequently reduce intestinal permeability in rodents and humans [32, 33]. Other studies strongly suggest that the products of OFS fermentation such as short-chain fatty acids (butyrate, propionate and lactate) positively influence the gut barrier function , i.e. in vitro butyrate, acetate and propionate all exert a concentration-dependent reduction of paracellular permeability in the Caco-2 model of colonic epithelium . In the present study, we observed that HF-OFS-fed mice exhibit a significant increase in total and empty caecum weight compared with HF and HF-Cell mice (data not shown). This supports the idea that the fermentation of dietary fibre allowing the proliferation of specific bacteria such as bifidobacteria influences the profile of short-chain fatty acids released in the gut. The consequences of the specific fermentation are lower endotoxaemia and an improvement of metabolic disturbances.
With regard to the role of gut bacteria in the development of metabolic disorders, Brugman et al. have shown that antibiotic treatment partially protects against the occurrence of diabetes in a diabetes-prone rat that develops insulitis. The authors proposed that altering the gut microbiota composition by antibiotic treatment reduces the antigenic load and hence the inflammatory reaction that had led to pancreatic beta cell destruction . Consistent with these results, we have recently demonstrated that HF-diet-fed mice and ob/ob mice treated with antibiotics for 4 weeks were partially protected against the development of metabolic diseases and inflammatory tone (P. D. Cani, N. M. Delzenne and R. Burcelin, unpublished data).
We cannot rule out the putative implication of another important mechanism to explain the improved glucose metabolism in prebiotic-fed mice. Indeed, we have previously reported that prebiotic dietary fibre improved diabetes and insulin sensitivity by a mechanism promoting synthesis and secretion of the incretin, GLP-1 [18, 28, 29]. Here, we confirmed that prebiotic dietary fibre only strongly increase colonic proglucagon mRNA content. Therefore, the present data suggest that there is a link between modulation of gut microbiota, endotoxaemia, inflammation and GLP-1 secretion. This merits further investigation. While the doses used in animal studies are not directly transposable to human nutrition, several reports have demonstrated that ingestion of fructo-oligosaccharides at the dose of 5 to 20 g/day are sufficient to promote prebiotic [37, 38, 39, 40] and physiological effects [41, 42, 43, 44].
In conclusion, dietary modulation of gut microbiota with a view to increasing bifidobacteria reduced endotoxaemia and improved glucose tolerance and insulin secretion, as well as reducing inflammation development in HF-diet-fed mice. Together, these findings suggest that the gut microbiota contribute to the pathophysiological regulation of endotoxaemia and set the tone of inflammation, glucose tolerance and insulin secretion.
Thus, specific strategies for modifying gut microbiota in favour of bifidobacteria could be useful tools for reducing the impact of high-fat feeding on the occurrence of metabolic diseases.
This work was supported by an FSR (Fonds Spéciaux de Recherche) grant from the Université catholique de Louvain and by a grant from the Fonds National de la Recherche Scientifique (FNRS 1.5.231.06-Belgium). P. D. Cani is a postdoctoral researcher from the FNRS Belgium. We would like to thank C. Feyt, and E. Delmée for helpful criticism and F. Debacker for excellent technical assistance.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.