Article pubs.acs.org/jpr
Deep Metabotyping of the Murine Gastrointestinal Tract for the Visualization of Digestion and Microbial Metabolism Silke S. Heinzmann*,†,‡ and Philippe Schmitt-Kopplin†,‡,§ †
Research Unit Analytical BioGeoChemistry, Helmholtz Zentrum München, 85764 Neuherberg, Germany German Center for Diabetes Research (DZD), 85764 Neuherberg, Germany § Chair of Analytical Food Chemistry, Technische Universität München, 85354 Freising, Germany ‡
S Supporting Information *
ABSTRACT: Despite the gut’s longitudinal specialization for digestion and microbiome organization, most studies focus on the analysis of its end product, feces. To determine the metabolic and physiological functions of different sections of the gut, we aimed to define a comprehensive list of characteristic metabolites for the physiological gut sections and to quantify the selected pathways. We investigated the metabolic composition of seven different gut sections from four C57Bl/6N mice with nontargeted metabolite profiling using high-resolution NMR spectroscopy, which returned a comprehensive metabolite overview with a single analytical measurement per sample. Here we deliver a list of characteristic metabolites, describe metabolite changes along the gut, and highlight the site specificity for selected metabolite pathways. We find that the largest metabolic changes happen in the cecum, where the microbiome produces microbial metabolites. Furthermore, we show the evolution of bile acids along the gut and describe their sitespecific conversion. We establish a metabolic basis for future investigations of metabolic perturbations, which can be introduced by dietary challenges or gene knockouts and provide valuable information for tailored study design and targeted sample collection. KEYWORDS: gastrointestinal physiology, gut microbiota, metabolite profiling, NMR spectroscopy, bile acids
■
INTRODUCTION The gastrointestinal tract is a complex organ; it has different compartments for the digestion of food and the absorption of nutrients, receives secretions from the bile bladder and the pancreas, has its own immune and nervous systems, produces hormones in the intestinal endocrine cells, and interacts with the gut microbiome. For example, carbohydrates are broken down by intestinal saccharidases and pancreatic amylases into monosaccharides and undergo transepithelial uptake in the small intestine to the bloodstream. Remaining indigestible carbohydrates are digested by microbes to produce short-chain fatty acids in the distal gut (i.e., the colon). The nutrients themselves regulate the expression of transporters and the induction of digestive enzymes.1 An overflow of nutrients (i.e., overfeeding with obesity or switching to a particular diet such as one high in resistant starch) might change the digestive organization, such that saccharides could reach the distal gut or high amounts of indigestible carbohydrates could affect the microbial composition.2 Consequently, the distribution and availability of metabolites are rearranged, and the rearrangement potentially acts on signaling and immune functions with an impact on the whole body system. Such a reorganization of gut physiology is also known and effectively utilized in gastric bypass surgery (e.g., Roux-en-Y gastric bypass), in which a remodeling of the gut leads to the reduction of body weight and to positive side effects on type 2 diabetes.3 The elucidation of factors © 2015 American Chemical Society
influencing fecal metabolites requires detailed and stepwise observation of the gut. For example, choline, a part of the diet, is an element of the membrane lipid bilayer that gets secreted into the duodenum as a component of hepatic bile, along with conjugated bile acids. Choline that escapes small intestinal absorption reaches the distal gut and is microbially metabolized to trimethylamine (TMA). More than 97% of secreted bile acids are resorbed in the ileum and recycled via enterohepatic circulation. The small fraction that escapes resorption reaches the large intestine, where it undergoes microbial deconjugation and hydroxylation4 and is then resorbed or excreted with feces.5,6 Consequently, fecal choline, TMA, and bile acids can be impacted by multiple physiological factors including altered dietary intake, impaired intestinal absorption, hepatic influences, and shifts in the gut microbiome; thus, the biological interpretation has yet to be unraveled. The bacteria density increases dramatically from stomach to colon,7 and bacterial communities differ greatly between the aerobe proximal gut (e.g., Lactobacillaceae) and the anaerobe distal gut (e.g., Bacteroidaceae and clostridia)8 and between the gut luminal mucosa and the feces.9 Cross-sectional investigations of the gut compartments, however, reveal a high amount of overlap between the mucosal microbiome and its respective adjacent luminal microbiome.10 Received: January 19, 2015 Published: March 19, 2015 2267
DOI: 10.1021/acs.jproteome.5b00034 J. Proteome Res. 2015, 14, 2267−2277
Article
Journal of Proteome Research
bead and then homogenized for 3 min at 30 1/s in a TissueLyser II (Qiagen, Hilden, Germany). Samples were then centrifuged (5 min, 13000g), and the supernatant was taken as an aqueous extract. Then, 1 mL of methanol (MeOH) was added to the remaining pellet and also homogenized and centrifuged. The supernatant was labeled as an MeOH extract. Both extract groups were separately evaporated in a SpeedVac concentrator (Thermo Scientific, Waltham, MA), reconstituted in 150 μL of the respective NMR buffer (90% D2O, 500 mM PO4 buffer with 0.1% trimethylsilyl-tetradeuteropropionic acid [TSP], pH 7.4, or 100% methanol-d4 ), and analyzed immediately by NMR spectroscopy.
For selected metabolites, the distribution differences along the gastrointestinal tract have previously been studied.11−13 However, a holistic and unbiased picture of the distribution of different metabolite classes is necessary to investigate the systemic effects of clinical or lifestyle interventions (e.g., dietary changes) on gastrointestinal physiology. Metabolite profiling approaches combining spectroscopic and spectrometric techniques with mathematical modeling of the complex data14,15 find a wide application in the assessment of disease etiology, toxicology, and nutritional interventions. Although nontargeted NMR spectroscopic methods are less sensitive than mass spectrometric methods, they are extremely useful when used to semiquantify the elucidated metabolites, and they enable a chemical comparison across different biological matrices without chemical and physical suppression effects. The lower sensitivity can be partially overcome by the use of state-of-the-art spectroscopic equipment, strong magnetic fields, and cryogenic probe measurements combined with long acquisition times and the combination of both 1D and 2D acquisitions to increase sensitivity and resolution and to minimize spectral overlap. In the presented project, we employed high-resolution 1H NMR spectroscopy to investigate the metabolite composition of the lumen of different gut sections, assess any similarities, relate selected metabolites to physiological function of the sections, and, finally, highlight the microbial activity in the GI tract.
■
NMR Spectroscopic Analysis
All samples were transferred to 3 mm outer diameter NMR Bruker Match tubes (Hilgenberg GmbH, Malsfeld, Germany). Samples were analyzed in a randomized order. NMR spectra were acquired on a Bruker 800 MHz spectrometer (Bruker Biospin, Rheinstetten, Germany) operating at 800.35 MHz with a quadruple inverse (QCI) cryogenic probe. For an overview of all molecules present in the sample, a standard one-dimensional (1D) pulse sequence [recycle delay (RD), 90°, t1, 90°, mixing time (tm), 90°, acquire FID] was acquired, with water suppression irradiation during an RD of 2 s, the tm set on 200 ms, and a 90° pulse set to 9 μs, which collected 512 scans into 64K data points with a spectral width of 12 ppm. A representative sample of each gut section underwent a series of 2D analyses (J-resolved, TOCSY, and HSQC) for detailed metabolite analysis and metabolite identification. The Jresolved NMR spectra were acquired with the pulse sequence [d1, 90°, τ, 180°, τ, acquire FID], with suppression of the water resonance during d1 (2 s) into 4K data points in F2 and 16 transients using 64 increments of τ; the spectral widths in F2 and F1 were 12 and 0.08 ppm, respectively. For the 2D 1H−13C HSQC spectra, phase-sensitive ge-2D HSQC using PEP and adiabatic pulses for inversion and refocusing with gradients were used. For each 2D spectrum, 4096 × 1024 data points were collected using 64 scans per increment, an acquisition time of 0.25 s, and 16 dummy scans. The spectral widths were set to 12 and 230 ppm in the proton and carbon dimensions, respectively. For the 2D 1H−1H TOCSY spectra, phasesensitive sensitivity-improved 2D TOCSY with WATERGATE (3-9-19) and DIPSI-2 were acquired. For each 2D spectrum, 19 228 × 1024 data points were collected using 16 scans per increment, an acquisition time of 1 s, and 16 dummy scans. The spectral widths were set to 12 and 12 ppm in the F2 and F1 dimensions, respectively. Processing of the spectra was carried out in a TopSpin 3.2 (Bruker BioSpin, Rheinstetten, Germany). FIDs were multiplied by an exponentially decaying function corresponding to a line broadening of 0.3 Hz before the Fourier transformation. All spectra were manually phased, baseline corrected, and calibrated to TSP (δ 0.00). Data were imported into Matlab (Mathworks, Natick, MA), further processed (i.e., the water region [δ 4.70−5.21] was removed), and normalized to the TSP signal (the area under the curve of TSP).
MATERIALS AND METHODS
Animal Handling and Sacrifice Procedure
Four C57N mice (Taconic, Ry, Denmark) were bred and housed under standard vivarium conditions (12:12 light−dark cycle). Three days prior to sacrifice, the mice were singlehoused (cages included domehouses). Animals were fed a cereal-based standard diet (Altromin Spezialfutter GmbH, Lage, Germany), which is further specified in Table S1 of the Supporting Information. At an age of 24 weeks, mice were killed with an isoflurane overdose. The animals were in a fed state (the sacrifice occurred between 8 and 10 a.m., taking into consideration the nocturnality of mice) with free access to the chow. The gastrointestinal tract of each animal was removed and divided into its anatomically relevant parts with surgical scissors. The luminal contents of the stomach, jejunum (the middle third of the small intestine), ileum (the last third of the small intestine), and cecum and the proximal and distal colon were collected by carefully extruding the gut part with a spatula. Feces pellets were collected from the cage. All samples were snap-frozen in liquid nitrogen and stored at −80 °C to prevent enzyme activity. The sample size was chosen to allow simple statistical tests. The animal husbandry staff did not give dietary or drug treatments and were therefore blinded. All animals received humane care according to the criteria outlined in the National Academy of Sciences Guide for the Care and Use of Laboratory Animals. The animal experiments were approved by the Upper Bavarian district government (Regierung von Oberbayern Gz. 55.2-1-54-2532-4-11).
Data Analysis
Metabolite Extraction Procedure
The hierarchical cluster analysis (HCA) was carried out using the Matlab algorithm for the calculation of the correlation distance and Ward’s method linkage, shown in a dendrogram; the principal component analysis (PCA) analysis was done in Matlab. The partial least-squares discriminant analysis (PLSDA) was also carried out in Matlab, and the PLS loading plots
Metabolites from 50 mg of gut content or 2 fecal pellets were extracted with an aqueous and methanol two-step extraction procedure, adopted from Wu et al.16 First, 1 mL of H2O/ACN (1:1) was added to the sample with ceramic beads (NucleoSpin, Macherey-Nagel, Dueren, Germany) and a steel 2268
DOI: 10.1021/acs.jproteome.5b00034 J. Proteome Res. 2015, 14, 2267−2277
Article
Journal of Proteome Research
Figure 1. Hierarchical cluster analysis of the gastrointestinal lumen extracts (i.e., the stomach, jejunum, ileum, cecum, proximal and distal colon, and feces) from four mice. Mouse identifiers (1−4) are shown in brackets. The dendrogram was calculated with the correlation distance and Ward’s method linkage. Strong clusters of proximal gut (right) and distal gut (left), with subclusters formed by the extracts from the contents of the jejunum, ileum, and stomach (from right to left), upper distal gut (cecum and proximal gut), and lower distal gut (distal colon and feces).
and between the distal colon and feces seemed higher, and the intermouse variations were relatively higher as evidenced by a lack of clustering of the four animals from each gut section.
are shown by calculating the correlation and covariance of the two groups.17 The quantification of compounds (e.g., shortchain fatty acids and bile acids) was done by calculating the stoichiometrically corrected area under the curve of the NMR signals. Relative metabolite changes between the different gut luminal contents are given as a decrease and increase in the signal intensity; a “strong” change was defined as a >2-fold change, changes of less than 10% were not considered, and only relative differences with a p value of
