Article Text

Original article
The gut microbiota elicits a profound metabolic reorientation in the mouse jejunal mucosa during conventionalisation
  1. Sahar El Aidy1,2,
  2. Claire A Merrifield3,
  3. Muriel Derrien1,2,
  4. Peter van Baarlen4,
  5. Guido Hooiveld5,
  6. Florence Levenez6,
  7. Joel Doré6,
  8. Jan Dekker1,7,
  9. Elaine Holmes3,
  10. Sandrine P Claus8,
  11. Dirk-Jan Reijngoud9,
  12. Michiel Kleerebezem1,2,4,10
  1. 1Top Institute Food and Nutrition, Wageningen, The Netherlands
  2. 2Laboratory of Microbiology, Wageningen University, Wageningen, The Netherlands
  3. 3Biomolecular Medicine, Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, London, UK
  4. 4Host-Microbe Interactomics, Wageningen University, Wageningen, The Netherlands
  5. 5Nutrition, Metabolism and Genomics Group, Division of Human Nutrition, Wageningen University, Wageningen, The Netherlands
  6. 6INRA, UMR1319, Jouy-en-Josas, France
  7. 7Department of Animal Sciences, Wageningen University, Wageningen, The Netherlands
  8. 8Department of Food and Nutritional Sciences, The University of Reading, Reading, UK
  9. 9Department of Laboratory Medicine, University Medical Centre Groningen, University of Groningen, Groningen, The Netherlands
  10. 10Department of Health, NIZO food research, Ede, The Netherlands
  1. Correspondence to Michiel Kleerebezem, Health Department, NIZO food research, PO Box 20, Ede 6710 BA, The Netherlands; michiel.kleerebezem{at}


Objective Proper interactions between the intestinal mucosa, gut microbiota and nutrient flow are required to establish homoeostasis of the host. Since the proximal part of the small intestine is the first region where these interactions occur, and since most of the nutrient absorption occurs in the jejunum, it is important to understand the dynamics of metabolic responses of the mucosa in this intestinal region.

Design Germ-free mice aged 8–10 weeks were conventionalised with faecal microbiota, and responses of the jejunal mucosa to bacterial colonisation were followed over a 30-day time course. Combined transcriptome, histology, 1H NMR metabonomics and microbiota phylogenetic profiling analyses were used.

Results The jejunal mucosa showed a two-phase response to the colonising microbiota. The acute-phase response, which had already started 1 day after conventionalisation, involved repression of the cell cycle and parts of the basal metabolism. The secondary-phase response, which was consolidated during conventionalisation (days 4–30), was characterised by a metabolic shift from an oxidative energy supply to anabolic metabolism, as inferred from the tissue transcriptome and metabonome changes. Detailed transcriptome analysis identified tissue transcriptional signatures for the dynamic control of the metabolic reorientation in the jejunum. The molecular components identified in the response signatures have known roles in human metabolic disorders, including insulin sensitivity and type 2 diabetes mellitus.

Conclusion This study elucidates the dynamic jejunal response to the microbiota and supports a prominent role for the jejunum in metabolic control, including glucose and energy homoeostasis. The molecular signatures of this process may help to find risk markers in the declining insulin sensitivity seen in human type 2 diabetes mellitus, for instance.

  • C57/BL 6J ex-germ-free mice
  • jejunum
  • transcriptome
  • metabonome
  • microbiota
  • gut immunology
  • gastrointestinal tract
  • gene expression
  • gene regulation
  • gut inflammation
  • probiotics
  • mucosal immunology
  • mucins
  • anti-bacterial mucosal immunity
  • bacterial interactions
  • Campylobacter jejuni
  • colonic microflora
  • crohn's disease
  • intestinal bacteria
  • immune response
  • energy metabolism
  • liver metabolism
  • glucose metabolism
  • lipid metabolism
  • inherited metabolic disease
View Full Text

Statistics from

Significance of this study

What is already known on this subject?

  • The gut microbiota plays a prominent role in modulation of the host intestinal and systemic energy and lipid metabolism.

  • Most of the nutrient absorption in the intestine occurs in the jejunum.

  • The dynamics of metabolic responses to the microbiota in the jejunum are largely unexplored.

What are the new findings?

  • The microbiota strongly affects jejunal (glucose and energy) metabolism. Unlike the ileum and colon, the jejunum responds instantly to the colonising microbiota.

  • The jejunum has a unique acute role in the gut response to luminal microbe–diet interplay.

  • An ‘insulin-response’ transcriptional signature for the jejunal metabolic changes in response to the microbiota.

How might it impact on clinical practice in the foreseeable future?

  • The identified transcriptional signature may help to unravel the genetic basis of the dysregulation in metabolic syndrome diseases such as insulin resistance and type 2 diabetes.

  • Advanced molecular understanding of the relations between dysregulation and disease should guide the design of health-promoting intervention strategies in metabolic syndromes including diabetes.

  • The genetic basis of the jejunal response to the microbiota and/or its metabolites may help to explain the improvement in insulin sensitivity that occurs in gastric-bypass surgical interventions.


The gut microbiota is recognised as a key determinant in the regulation of host metabolism, including energy yield from the diet and fat storage.1 The microbiota can convert and neutralise potentially harmful metabolites from the dietary nutrient flow and enhance energy yield from the diet.2–4 It is now firmly established that the gut microbiota has a major impact on the host phenotype in humans and mice.3 ,5 ,6 A pioneering study by Bäckhed and colleagues showed that conventionalisation of adult germ-free mice leads to insulin resistance and suppression of fasting-induced adipocyte factor (Fiaf, also known as angiopoietin-like 4, Angptl4), which is essential for the induction of de novo hepatic lipogenesis.3 Deregulation of hepatic lipogenesis is a prominent factor in obesity.3 Subsequent studies in ex-germ-free mice revealed the important roles of short-chain fatty acid- and endocrine hormone-binding G-protein-coupled receptors, Gpr41 and Gpr43, in microbial modulation of host energy balance.7 Despite recognition of the crucial importance of host–microbiota interaction in controlling metabolic changes in the intestinal mucosa, these insights were mainly derived from studies on the ileal or colonic intestinal tissues in experiments comparing germ-free and conventional or conventionalised mice. However, the effect of the microbiota on metabolism in more proximal parts of the small intestine (such as the jejunum) has remained largely unexplored to date. The relatively small size of the microbial community residing in this intestinal region may have been the rationale behind the limited interest in it with respect to host–microbe communication. The luminal microbiota has 103–107 bacterial cells per gram in the human jejunum8 and comprises a relatively simple community.9 However, the jejunum is of basal physiological and clinical relevance. The role of the jejunum as an intestinal region that controls systemic metabolic responses is illustrated by the observation that gastric bypass surgery, which allows nutrients to bypass the duodenum and proximal small bowel (ie, the jejunum),10 leads to improvement or resolution of type 2 diabetes in both obese and non-obese individuals.11 Although the mechanisms underlying the increased insulin sensitivity following gastric bypass surgery remain obscure, changes in lipid metabolism,12 intestinal gluconeogenesis13 and secretion of gut hormones14 have been suggested to be involved.

We recently described the molecular dynamics of the development of host immune response in ex-germ-free mice upon conventionalisation in different regions of the gut, over a time span of 30 days.15 This study unexpectedly highlighted the prominent responsiveness of the jejunal mucosa to the colonising microbiota. Therefore the present study focuses on this intestinal region, and uses tissue-transcriptome, tissue-metabonome (1H NMR) and microbiota profiling, in combination with histological analysis, to improve our understanding of the molecular dynamics involved in microbiota-mediated control of jejunal metabolism. Our study shows that the jejunum of 8–10-week-old germ-free mice shows the fastest response to microbial colonisation, and that the interplay between the diet, microbiota and jejunal mucosal tissues supports a substantial role for the jejunum in control of glucose and energy homoeostasis.


Animals, experimental design and sampling

All procedures were carried out according to the European guidelines for the care and use of laboratory animals and with permission 78–122 of the French Veterinary Services. Germ-free and conventionalised mice (male, C57BL/6J) were maintained in sterile conditions on a commercial laboratory chow diet. After 2 weeks of acclimatisation and diet adaptation, the first set of germ-free mice (n=3) were randomly assigned to be sacrificed by oral anaesthesia using isoflurane. The remaining germ-free mice were conventionalised with freshly obtained faecal material from conventionally raised mice (C57BL/6J). Three independent biological experiments were performed using mice of different age after 2 weeks of acclimatisation. In the first two experiments, conventionalised mice were sacrificed on days 1, 2, 4, 8 and 16 after conventionalisation (n=3 per group per experiment). In the third experiment, conventionalised mice were sacrificed on days 4 and 30 after conventionalisation (n=4–5 per group). The jejunum from each mouse was removed and divided into 2 cm segments for RNA isolation and histological and metabolic analysis. The luminal contents of jejunal segments were removed by gentle squeezing, snap-frozen, and stored at −80°C for microbiota analysis (for detailed descriptions, see online supplementary material and figure S1).

Histological staining

Cross-sections (4 μm thick) of 2 cm intestinal segments fixed in 4% (w/v) paraformaldehyde and paraffin-embedded were stained with haematoxylin (Vector Laboratories, Burlingame, California, USA) and eosin (Sigma-Aldrich, Zwijndrecht, the Netherlands). To detect morphometric differences, 12–15-well oriented villi were chosen per intestinal segment and measured. Cross-sections of the jejunal tissue segments were stained with Alcian Blue (AB) followed by the periodic acid-Schiff's (PAS) reaction as previously described.16 Comparative analyses of the histochemical and cell-enumeration data were performed at each time point using one-way analysis of variance followed by Tukey's Studentised range test (SPSS program, statistics 19), and, for all variables, p<0.05 was considered significant.

Transcriptome analysis

High-quality total RNA was isolated from a 2 cm jejunal segment by extraction with TRIzol reagent and hybridised on Affymetrix GeneChip Mouse Gene 1.1 ST arrays. Complementary methods were used for the biological interpretation of the transcriptome data: gene clustering using Multi-experimental View (MeV) software,17 over-representation analysis of Gene Ontology (GO) terms using temporal and location comparative analysis using Short Time Series Expression Miner (STEM),18 and construction of biological interaction networks using Ingenuity Pathways Analysis ( (for detailed descriptions, see the online supplementary material).

Microbial profiling of jejunal luminal contents

Luminal contents of the jejunum were analysed by Mouse Intestinal Tract Chip (MITChip), a diagnostic 16S rRNA array that consists of 3580 unique probes especially designed to profile murine gut microbiota.19 ,20 Total bacteria were quantified using quantitative PCR detection of 16S rRNA gene copies (for detailed descriptions, see the online supplementary material).

Tissue metabolite profiling

Tissue samples were homogenised and extracted in acetonitrile/water (1:1, v/v), as previously described.21 The supernatant containing the aqueous phase was collected, freeze-dried and dissolved in 600 μl D2O. Samples were centrifuged for 10 min at 15 000 g, and 500 μl of the supernatant and 50 μl water were transferred to 5 mm (outer diameter) NMR tubes for analysis by 1H NMR spectroscopy.

Metabonomic data were visualised by principal component analysis (PCA). Orthogonal partial least-squares discriminant analysis (OPLS-DA) models22 were then fitted between successive time points in order to highlight discriminant metabolites. PCA, O-PLS-DA and statistical total correlation spectroscopy were performed in Matlab (using an in-house routine)23 (for detailed descriptions, see the online supplementary material).

Accession number

The mouse microarray dataset was deposited in NCBI Gene Expression Omnibus (GEO) with accession number GSE32513.


Conventionalisation elicits acute morphological and transcriptional responses in jejunal tissue.

This study comprised three independent experiments aimed to identify the dynamics of jejunal mucosal changes between germ-free and conventionalised mice.

No changes in jejunal morphology were observed on days 1 and 2 after conventionalisation (figure 1A). An initial increase in the jejunal crypt depth and a parallel decrease in villus height of conventionalised mice started 4 days after conventionalisation,15 and this morphology remained during the 30 days. Transcriptome (day–day) analysis illustrated that short-term conventionalisation elicited a strong repression of specific metabolic function-encoding gene sets, such as genes involved in oxidative phosphorylation (figure S2A) and the citrate cycle (TCA cycle). Short-term conventionalisation also led to an immediate induction of genes encoding the purinergic receptor Y 2,4 (P2ry 2,4), which is induced by tissue damage,24 and repression of Notch pathway genes, which are important determinants of the cell lineage allocation,25 and genes encoding enteroendocrine hormones (figures 1B and S2B) and cell cycle regulation (figure S2C). We expected that the changes in Notch pathway genes on day 1 would be reflected in acute changes in two secretory cell lineages, Paneth and goblet cells. The genes encoding Reg3γ and LyzP, antimicrobial proteins secreted by Paneth cells, did not show any changes in their expression on day 1.15 AB/PAS staining revealed an increase (∼40%, p<0.05 vs the germ-free state) in the number of swollen, mucin-filled goblet cells in the villus epithelia, on day 1 after conventionalisation (and to a lesser extent day 2). After 4 days of conventionalisation, we observed an overall decrease in goblet cells; on day 4, goblet cells stained weakly and appeared to be smaller, with a more columnar morphology, compared with days 1 and 2 after conventionalisation. This change in morphology was transient since the number of large, mucin-filled goblet cells on day 16 after conventionalisation had slightly exceeded that of germ-free mice (figure 1C,D). Taken together, the transcriptional and histological changes that characterised the acute phase of conventionalisation (day 1) in jejunal tissues suggest an early response to the colonising microbiota, which comediated repression of the cell cycle and development and parts of basal metabolism.

Figure 1

Early and late histological and transcriptional changes in jejunal mucosa in response to conventionalisation. (A) Representative photomicrographs of H&E-stained cross-sections of jejunal tissues in (a) germ-free, (b) day 4 and (c) day 30 after conventionalisation. Black arrows point to the increase in crypt depth across time. (B) Heat map of the expression values for genes encoding enteroendocrine hormones and purinergic receptors and genes involved in Notch signalling pathway and histone variants. The data represent the changes in gene expression levels on days 1–30 after conventionalisation in comparison with germ-free mice. Notably, the induction of P2ry2,4 and the lower expression levels of Notch, Hes1 and histone genes were transient and only measured on day 1 after conventionalisation in contrast with the persistent downregulation of genes encoding enteroendocrine hormones. (C) Representative Alcian Blue/periodic acid–Schiff (AB/PAS) stain photomicrographs from the jejunum showing the changes in the number of goblet cells in germ-free and conventionalised mice at days 1, 4 and 16. Note that some AB/PAS-stained cells appear in the middle of the lamina propria. (D) The individual goblet cells scores in the jejunum. Significant differences between time points are indicated by distinctive characters above the measurement clusters (p<0.05). Mean±SEM scores: germ-free, 9.6±0.7; day 1, 19.7±3.6; day 4, 7.1±1.1; day 16, 14.7±3.2.

Time-dependent jejunal metabolic response to the colonising microbiota

Gene expression time series analysis using the STEM software was performed to monitor the time-resolved changes in GO categories during conventionalisation. STEM clustering of gene expression indicated that GO terms related to metabolism were strongly enriched over the time span of the experiment (p<0.001). The repression on day 1 after conventionalisation of genes associated with metabolism became more pronounced at later conventionalisation time points, and was mainly characterised by gradual repression of genes regulating fatty acid oxidation and transport (figure S3) (table S1). This repression was mediated by the continuous repression (up to day 30 after conventionalisation) of the major regulator of fatty acid oxidation, peroxisome proliferator activated receptor α (Ppar-α) and its downstream target genes, including the fatty acid transporter Cd36 and Angptl4 (figure 2). Another downregulated gene involved in fatty acid oxidation was hydroxy-3-methylglutaryl-CoA synthase 2 (Hmgcs2), which facilitates the synthesis of ketone bodies26 (figure 2). Conversely, the expression of HmgCoA reductase, which encodes the rate-limiting step of steroid synthesis,27 was induced. Overall, these data suggest that lipolysis, an important source of cellular energy, was repressed during conventionalisation, together with fatty acid oxidation and transport.

Figure 2

Repression of key regulators of fatty acid oxidation in response to microbial colonisation. Jejunal expression levels of genes involved in lipid metabolism were analysed in germ-free and conventionalised mice on the indicated days of conventionalisation and were compared between the three conventionalisation experiments (n=6–11 mice/day). Values are depicted as dot plots. The values are fold change based on log2-based gene expression values. Significant differences between time points are indicated by distinctive characters above the measurement groups (p<0.05).

Microbial colonisation resulted in a gradual reorientation of jejunal metabolic homoeostasis

Global changes in the metabolic transcriptome suggested that the jejunal metabolism had shifted from fatty acid oxidation towards an alternative energy source. To further refine our understanding of the postulated metabolic reorientation, expression changes in metabolic genes were projected on to metabolic KEGG maps ( (figure S4). These projections showed that the suppression of fatty acid oxidation paralleled the repression of gluconeogenesis, as suggested by the continuous downregulation of the gene encoding phosphoenolpyruvate carboxykinase 1 (Pck1) and the induction of glycolytic, lipogenic, amino acid and nucleotide metabolic pathways. This induction was apparent on day 4 and persisted up to day 30 after conventionalisation. A time-dependent induction of genes encoding the transporter (Slc2a1, also known as Glut1) and metabolic enzymes involved in glycolysis (figure S4) suggested that the imported glucose had been converted through both the oxidative and non-oxidative route of the pentose phosphate pathway (PPP). The non-oxidative PPP route presumably generated the ribose for nucleic acid synthesis via the induction of Ripa (ribose-5-phosphate isomerase) and Pgd (6-phosphogluconate dehydrogenase). The metabolic mapping of transcriptome changes also highlighted the impact of conventionalisation on the induction of key regulators of lipogenesis and nucleotide synthesis and metabolism (figure S4).

The jejunal metabolism underwent dynamic changes in expression of genes involved in amino acid metabolism, including glutaminolysis. This was apparent from the induction of Gfpt1 (glutamine-fructose-6-phosphate transaminase 1), Cad (carbamoyl-phosphate synthetase) and Gls2 (glutaminase 2). In contrast, the genes that control the conversion of glutamate into α-ketoglutarate were repressed. Moreover, Phgdh (phosphoglycerate dehydrogenase), Psat1 (phosphoserine aminotransferase 1) and Shmt2 (serine–glycine hydroxymethyltransferase) involved in glycine biosynthesis were induced. While several steps in tyrosine, phenylalanine and tryptophan metabolism were repressed (figure S4), others such as Tat (tyrosine transaminase), Ido1 (indoleamine 2,3-dioxygenase 1), Nos2 (inducible nitric oxide) and Wars (tryptophanyl-tRNA synthetase) were induced. These transcriptome changes show that colonisation of the jejunum of germ-free mice leads to dynamic modulation of jejunal metabolism that may also have had implications for immune cells, given the functions of the genes Ido1, Nos2 and the interferon-induced Wars in dendritic and other immune cells.

Jejunal tissue metabolic profiling during conventionalisation confirms its metabolic reorientation

The projections of the differently expressed genes on KEGG metabolic maps, as shown above, suggested that microbial colonisation would lead to changes in the jejunal tissue metabolites that serve as substrate or product of the enzymes encoded by the corresponding genes that were differentially expressed (figure S4, S5 and table S2). Metabolic 1H NMR phenotyping of jejunal tissues obtained from germ-free and conventionalised animals was conducted to measure the effects of the colonising microbiota on jejunum metabolism. OPLS-DA models were constructed focusing on the differences between germ-free and conventionalised mice over time. The analysis showed that jejunal colonisation by the gut microbiota correlated with increased levels of ascorbate, ethanolamine, aspartate, fumarate, glutamate, glutamine, inosine, dihydroxyacetone (tentative assignment) and hypoxanthine. In contrast, the levels of tauro-conjugated bile acids and glycerol decreased during microbial colonisation. A number of metabolites such as guanosine and glycine were transiently altered during conventionalisation (figure 3).

Figure 3

Correlation coefficient heat map that shows the relative increase or decrease of concentrations of metabolites during conventionalisation compared with the germ-free state. A series of pair-wise orthogonal partial least-squares discriminant analysis models were constructed, from which the correlation coefficients of the significantly altered metabolites in jejunum tissues during days 1–16 after conventionalisation were extracted. Red represents increased level, and green represents decreased level. †Tentative metabolic assignment.

Establishment of colonising microbiota in the jejunum

Molecular analysis by quantitative PCR detection of the 16 S rRNA gene copies in jejunum samples indicated that, on day 1, a microbial community had already established; this community contained approximately 4.9±0.3 16S rRNA copies/ml jejunal content (expressed as log10 ± SEM), a density that is expected in the jejunum.8 This microbial density had not significantly changed by day 8 (5.5±1.52 16S rRNA copies/ml jejunal content). Molecular fingerprinting of the composition of the colonising microbiota was technically challenging because of the small sample volumes and the low microbial population density, but was performed using MITChip analysis,19 ,20 the results of which are presented in the online supplementary results (table S3, S4 and figure S6). Pearson correlation-based similarity analysis of MITChip profiles of the jejunum samples showed a high inter-individual variation and considerable instability of the microbial composition (figure 4) over time, which is in clear contrast with the microbial communities detected in the colon samples.15

Figure 4

Relative contribution of the level-0 (phylum-like) phylogenetic groups detected by MITChip in jejunal samples over time. Pearson correlation similarity indices of the MITChip profiles from jejunal samples at different time points after conventionalisation are shown below the graph.

Transcriptional signatures for the microbiota-driven metabolic reorientation of the jejunal mucosa

To evaluate whether a transcriptional signature could specify the metabolic reorientation observed, STEM output genes were mined for metabolic-related GO assignments in the jejunal mucosa. The genes recovered were used to construct an interaction network in IPA (Ingenuity Pathway Analysis). The resulting network (figure 5) contains genes involved in the regulation of various cellular pathways, including fatty acid metabolism, amino acid metabolism, glycolysis and gluconeogenesis (figure S7). The identified gene set encompassed central regulatory genes that belonged to the nuclear receptor subfamily, including Ppar-α and Ppar-γ, Nr1i3 (nuclear receptor subfamily 1-group I-member 3), Nr0b1 (nuclear receptor subfamily 0-group B-member 2), Insr (insulin receptor) and Thrβ (thyroid hormone receptor β) (figures 2,5 and 6). Multiple target genes of these central regulators were identified (figure 5). Interestingly, among the 76 genes encompassed in the tissue transcriptome signatures, 20 are known to have a role in metabolic disorders including type 2 diabetes and insulin resistance (indicated by red arrowheads in figure 5). This result supports the biological coherence and corresponding relevance of these transcriptional signatures for the microbiota-driven metabolic reorientation of the jejunal mucosa.

Figure 5

Potential transcriptional signature for the dynamics of jejunal metabolic reorientation during conventionalisation. (A) The Ingenuity interaction network based on the genes within the signature encompassed 76 nodes that had at least 110 connections; the largest number of interactions for any node was 20. Red arrowheads indicate genes known to be associated with human insulin resistance and type 2 diabetes based on IPA disease annotation. (B) Heat map diagram of the signature genes involved in the regulatory network, representing the changes in their jejunum expression level during days 1–30 after conventionalisation in comparison with the germ-free state.

Figure 6

Nuclear receptors that regulate gut hormones and fatty acid metabolism constitute the major nodes in the identified transcriptional signature. Jejunal gene expression levels of nuclear receptors and their target genes were compared between the three conventionalisation experiments. Values are depicted as dot plots. The values are fold change based on log2-based gene expression values. Significant differences between time points are indicated by distinctive characters above the measurement groups (p<0.05). Insr, insulin receptor; Ppar-γ, peroxisome proliferator activated receptor gamma; Nr1i3, nuclear receptor subfamily 1, group I, member 3; Nr0b2, nuclear receptor subfamily 0, group B, member 2; Nr1h3, nuclear receptor subfamily 1, group H, member 3; Slc27a2, solute carrier family 27 (fatty acid transporter), member 2; Abca1, ATP-binding cassette, sub-family A, member 1; Thrβ, thyroid hormone receptor, beta; Pdk4, pyruvate dehydrogenase kinase, isozyme 4.


The proximal intestine, in particular the jejunum, can be considered the initial site of intestinal interaction between the host mucosa, microbiota and ingested food. Therefore it may be anticipated that this part of the intestinal tract in particular will respond strongly to the presence of microbiota, despite of its relatively low microbial densities.8

It was recently reported that metabolic changes in liver tissues in response to intestinal conventionalisation involved two distinct temporal phases.28 Our study focuses on time-resolved analysis of metabolic responses in the jejunal mucosa and confirms a two-phase response to conventionalisation. The jejunal response to conventionalisation was characterised by an acute, transient phase, which was followed by a secondary homoeostatic phase, as evidenced by both transcriptomic and metabonomic analyses of the jejunal tissue (figure 7). This rapid and dynamic metabolic response to conventionalisation was not observed in the ileum or colon (figure S8A) lending further support to the unique role of the proximal intestine in response to the luminal microbiota–diet interplay. The unique transcriptional changes that featured the acute phase of conventionalisation (day 1) in jejunal tissues suggest an early response to ‘danger signals’ (eg, ATP, UDP and UTP, which activates P2Y receptors).24 ,29 These signals could be endogenous or derived from the early colonising microbiota, its metabolites and its interaction with nutrient flow or both. We propose that these danger signals comediated the repression of the cell cycle and parts of basal metabolism and the strikingly increased number of mucin-filled goblet cells, observed only in the jejunum (not in the ileum or colon (figure S8B)). The unique changes observed in the goblet cells may have been part of an innate defence mechanism during the acute-phase response to the colonising microbiota.

Figure 7

A schematic model of the time-resolved jejunal mucosal changes of fat, energy and amino acids and nucleotide metabolism during conventionalisation. Colour-coded boxes represent changes in metabolic processes at early (1–2) and late (4–30) days of conventionalisation. White represents non-differentially regulated processes. Red represents upregulated processes, and green represents downregulated processes, relative to the control (germ-free=day 0). The intensity of the green colour is proportional to the relative decrease in gene expression.

Transcriptome analysis, GO enrichment and pathway analysis showed that the secondary phase in response to conventionalisation (days 4–30) was characterised by molecular changes, which suggested that a metabolic reorientation of jejunal tissue had occurred. Notably, the repression of the nuclear receptors, which was an important feature of this secondary metabolic response, had already been initiated during the first day of conventionalisation. The secondary phase of jejunal responses was consolidated during the last phase of conventionalisation (up to day 30). This phase was characterised by the establishment of a novel metabolic homoeostasis that accommodated the microbiota. Establishment of this novel homoeostasis followed the induction of genes involved in cholesterol biosynthesis, which is in agreement with the findings reported by Larsson et al,30 induction of lipogenesis, repression of genes belonging to fatty acid oxidation, lipid absorption (Pnliprp2, Clps and Apoa4), lipid transport and the repression of 14 important regulators including Ppar-α, Ppar-γ and Angptl4. Repression of lipid metabolism pathways appeared to coincide with increased glucose utilisation in the jejunal tissues, as evidenced by the increased expression of (among others) G6pd and Glut1, and repression of Pck1, G6Pc (glucose-6-phosphatase) and Pdk4 (figure S4). The transcriptome changes appear to describe a shift in the jejunal metabolism from an oxidative energy supply to an anabolic metabolism that may be required to meet the energy demand of the mucosal tissue and its epithelia. Indeed, in a previous study, we reported that immune activation was initiated in the jejunum on day 4 after conventionalisation and reached novel homoeostasis within 16–30 days, as evidenced by both transcriptome and immunohistochemical analysis targeting innate and adaptive (T cell) immune parameters.15 Moreover, in a recent study, Wang et al illustrated metabolic reprogramming from fatty acid β-oxidation to the glycolytic, PPP and glutaminolytic pathways, in activated T lymphocytes.31 In view of these findings, we propose that the secondary phase of the jejunal metabolic response is required to support the mucosal immune changes and its increased epithelial proliferation and renewal, as demonstrated by Ki-67 stain (figure S9) and presumably the induced expression of histone genes (figure 1C).32

The transcriptome profile of the secondary phase of conventionalisation resembles the metabolic changes observed in tumour cells, where both glycolysis (the Warburg effect) and glutaminolysis are driven by oncogenic signals.31 ,33 Metabolic tissue profiling by 1H NMR confirmed the induction of anabolic metabolism, illustrated by the accumulation of glutamine, glutamate and aspartate. Accumulation of these three metabolites coincides with their previously reported elevated levels in the ileum and colon of conventional mice.34 Glutamine, glutamate and aspartate are precursors for serine and subsequent nucleotide biosynthesis. Indeed, the induced expression of Phgdh, Psat1 and Shmt2 may have participated in glycine biosynthesis from serine, as confirmed at the metabolite level. Moreover, the induction of Ripa and Pgd may have led to increased ribose-5-phosphate biosynthesis. Both glycine and ribose-5-phosphate are important precursors of nucleotide biosynthesis.35 ,36 Glutamine, the most abundant amino acid in the plasma,37 has been shown to be extensively utilised by highly proliferating intestinal cells and is an important substrate for the anabolic metabolism38 that was a feature of the metabolic shift observed upon microbial conventionalisation.

Two associated genes encoding subunits of γ-glutamyl cysteine ligase, the rate-limiting enzyme in the de novo synthesis of glutathione from glutamate,39 Gclc and Gclm, were downregulated. This finding suggests that glutamate rather than glutathione is required during the metabolic shift in response to microbial colonisation, which was confirmed by the increased levels of glutamate detected by 1H NMR (figure 3). The decreased (fold changes ranged from −2.8 at day 4 and −2.5 at day 30 after conventionalisation, figure S4) expression of the Glud1 enzyme, which catalyses the oxidative deamination of glutamate to α-ketoglutarate and ammonia,40 suggested that less glutamate had been used in the TCA cycle in the conventionalised mice, implying that there was less TCA cycle activity. This lower TCA cycle activity would render a higher availability of acetyl-CoA for de novo synthesis of fatty acids and cholesterol, congruent with the induction of HmgCoA and Fasn genes. Fatty acids play a key role in cell membrane synthesis.41 The time-resolved induction of diverse immune modulators—glutamate, fatty acids, Ido1 and Nos242–45—during microbial colonisation confirms the strong link between metabolism and immunity and supports our hypothesis that the metabolic reorientation is required for the immune activation in response to microbial colonisation that we have previously shown to occur from day 4 onward.15

Detailed transcriptome analysis allowed the identification of transcriptional signatures that included repression of genes involved in hormonal regulation and key metabolic genes that play important roles in (among others) fatty acid oxidation and gluconeogenesis. Of note, the identified signatures comprised genes of which the human orthologues are dysregulated in metabolic disorders, including insulin resistance and type 2 diabetes. In addition, on the basis of genome-wide association studies, at least 26% of the identified genes were associated with metabolic diseases. One extrapolation that could harness the linkage between the jejunum and its microbiota with basic systemic metabolic processes and diseases, may come from the practical clinical scenario of proximal jejunal nutrient exclusion by the use of an implanted EndoBarrier in diabetic, obese patients. In this clinical setting, the proximal 60–100 cm of the jejunum is excluded from receiving nutrients; at the same time, bile salts diminish total numbers of resident bacteria including those bacteria that can produce metabolites that contribute to diabetes establishment or progression. The result of EndoBarrier implantations is dramatic reversal of diabetes even before any weight loss is recorded.46 We consider that changes in the nutrient flow–microbiota–jejunum interplay that take place after EndoBarrier implantations relate to our animal findings and that the improved insulin sensitivity measured after EndoBarrier implantations is mediated by host transcriptome changes that may well be similar to, or comparable with, the transcriptome changes we measured upon mouse conventionalisation.

Germ-free mice have a naturally high insulin sensitivity compared with conventionalised mice.3 The prominent effect of the colonising microbiota on the expression of host genes that are essential for proper glucose metabolism observed in this study suggests that the microbiota is a key determinant of proper jejunal glucose metabolism. In light of our findings, we propose that targeting the jejunal mucosa of diabetic patients with specific microbial groups (probiotic or faecal material transplant from healthy donors) may improve glucose homoeostasis and stimulate systemic insulin responsiveness. We envision that changes in the interplay between nutrient flow–microbiota composition–host metabolism that improve type 2 diabetes are established because of the introduction of a balanced microbiota that corrects a more dysbiotic state, which may also be among the underlying mechanisms that explain the beneficial gastric bypass effects. It is tempting to speculate that rationally altering the microbial composition in the proximal intestine would contribute to improvement of jejunal functions and metabolism. Improvement of microbial homoeostasis as a treatment may even be of relevance to other gastrointestinal tract disorders such as inflammatory bowel disease.47 ,48 Overall, the present study supports a critical role for the jejunal response in the perception of the local nutrient–microbiota interplay, which appears to be a major determinant of metabolic control and gut homoeostasis.


We thank several members of the team of J Doré (INRA, Jouy en Josas) for assistance with animal killing and sampling, Dicky J Lindenbergh-Kortleve (Department of Paediatrics, Erasmus Medical Centre) for assistance with histological staining, and J Jansen, M Grootte-Bromhaar, M Boekschoten and P de Groot (Division for Human Nutrition, Wageningen University) for their technical support in microarray hybridisation and microarray data-quality control and processing. We are also grateful to Dr T. Borody (Centre for Digestive Diseases, Five Dock) for providing the details on gastric bypass procedures and useful discussions related to this.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Funding This work was funded by the Top Institute Food and Nutrition (TIFN, Wageningen).

  • Competing interests None.

  • Provenance and peer review Not 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.