Metabolomic Modeling To Monitor Host ... - ACS Publications

Metabolomic Modeling To Monitor Host Responsiveness to Gut Microbiota Manipulation in the BTBRT+tf/j Mouse ... Interestingly, microbial groups most st...
0 downloads 0 Views 813KB Size
Subscriber access provided by George Mason University Libraries & VIVA (Virtual Library of Virginia)

Article

Metabolomic Modelling to Monitor Host Responsiveness to Gut Microbiota Manipulation in the BTBR Mouse T+tf/j

Matthias S. Klein, Christopher Newell, Marc R Bomhof, Raylene A Reimer, Dustin S Hittel, Jong M Rho, Hans J. Vogel, and Jane Shearer J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.5b01025 • Publication Date (Web): 01 Mar 2016 Downloaded from http://pubs.acs.org on March 2, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Metabolomic Modelling to Monitor Host Responsiveness to Gut Microbiota Manipulation in the BTBRT+tf/j Mouse

Matthias S. Klein1*, Christopher Newell2, Marc R. Bomhof1, Raylene A. Reimer1,3, Dustin S. Hittel3, Jong M. Rho4, Hans J. Vogel3,5, Jane Shearer1,3

1

Faculty of Kinesiology, University of Calgary, Alberta, Canada

2

Department of Medical Science, Faculty of Medicine, University of Calgary, Alberta, Canada

3

Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of Calgary, Alberta, Canada 4

Departments of Paediatrics & Clinical Neurosciences, Alberta Children’s Hospital Research Institute, University of Calgary, Calgary, Alberta, Canada

5

Department of Biological Sciences, University of Calgary, Alberta, Canada

1 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT The microbiota, the entirety of microorganisms residing in the gut, is increasingly recognized as an environmental factor in the maintenance of health and the development of disease. The objective of this analysis was to model in vivo interactions between gut microbiota and both serum and liver metabolites. Different genotypic models (c57Bl/6 and BTBRT+ tf/j mice) were studied in combination with significant dietary manipulations (chow vs. ketogenic diets) to perturb the gut microbiota. Diet rather than genotype was the primary driver of microbial changes with the ketogenic diet diminishing total bacterial levels. Fecal but not cecal microbiota profiles were associated with the serum and liver metabolome. Modelling metabolomemicrobiota interactions showed fecal Clostridium leptum to have the greatest impact on host metabolism, significantly correlating to 10 circulating metabolites, including 5 metabolites that did not correlate to any other microbes. C. leptum correlated negatively to serum ketones and positively to glucose and glutamine. Interestingly, microbial groups most strongly correlated to host metabolism were those modulating gut barrier function, the primary site of microbe-host interactions. Results show very robust relationships providing a basis for future work wherein the compositional and functional associations of the microbiome can be modelled in the context of the metabolome.

Key words: Metabolome, Metabolomics, Microbiota, Communication, Ketogenic diet, Modelling, Mouse

2 ACS Paragon Plus Environment

Page 2 of 27

Page 3 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

INTRODUCTION The human gut contains up to 100 trillion microorganisms, a number far exceeding the total number of our somatic cells.1 Gut microbiota play an important role in both metabolism and energy extraction from dietary sources. Alterations in the microbiome have been associated with numerous disease states including obesity1 type 2 diabetes2, autism3, inflammatory bowel disease4 and cardiovascular disease5 in observational and cross-sectional studies. Direct causation has been demonstrated in animal models employing fecal transplant experiments.6 Given the preponderance of evidence, gut microbiota are now considered to be an important environmental factor contributing to disease. Although the exact mechanisms by which the gut microbiota exert these effects are unknown, gut-derived metabolites are well documented to play a key communicative role.7

The microbiota can have a tremendous impact on host metabolism and different bacterial species can produce distinct metabolomic profiles. Microbiota based contributions to the metabolome are evident in studies comparing mice raised in a germ-free environment versus conventionally housed mice resulting in large differences in the circulating metabolome, particularly in levels of amino acids.8 Similarly, antibiotic treatment that eliminates the vast majority of intestinal microbes results in changes in 87% of detected metabolites confirming a close association between the microbiota and metabolome.9 Despite many reports of metabolome-microbiota interactions, there is a paucity of information on the contribution of specific bacterial species to the metabolome in vivo. To this end, this study analyzed different murine genotypes under varied dietary conditions to generate a data set with a diverse range of microbiota-metabolome interactions for modelling purposes.

First developed in 1956, the BTBRT+ tf/j (BTBR) is a widely studied mouse model of aberrant metabolic and behavioural characteristics.10 Of note, the strain displays obesity11, insulin3 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

resistance12, allergen-induced airway disease13 and autism spectrum disorder.14 To manipulate the gut microbiota, we employed a ketogenic diet (KD). Comprised of high-fat, adequate-protein, low-fibre and low-carbohydrate, the KD causes a shift in host metabolism by mimicking the fasting state and promoting fatty acid degradation within the liver. The excessively available fatty acids initiate a shift from carbohydrate to fatty acid utilization that culminates with the eventual conversion of excess acetyl-CoA to the ketone bodies β-hydroxybutyrate, acetoacetate and acetone. This diet created large microbiota shifts providing the opportunity to model metabolome-microbiota interactions.

EXPERIMENTAL SECTION Animals All experimental protocols were in compliance with the ethical standards approved by the University of Calgary Animal Care and Use Committee. Male C57Bl/6 (C57) or BTBR T+ tf/j (BTBR, Jackson Laboratories, Bar Harbor, ME, n = 21 & 25) were separated by their respective genotype and housed 3-6 per cage. Animals were age-matched to 5 weeks of age before being randomly selected for implementation of a control chow diet (13% kcal fat; LabDiet 5001, W.F. Fisher & Son, Somerville, NJ) or a ketogenic diet (75% kcal fat; Bio-Serv F3666, Frenchtown, NJ). This resulted in four treatment groups: C57 chow, C57 ketogenic, BTBR chow, and BTBR ketogenic (n = 11, 10, 15 & 10 respectively). Prior to sacrifice, animals were housed in a humidity controlled room with a 12 hour light/dark Zeitgeber cycle and had free access to both food and water. Following 10-14 days of dietary intervention mice were weighed and whole blood was analyzed for glucose and ketones with Precision Xtra meters (Abbott Laboratories, Bedford, MA) before being sacrificed by cervical dislocation. Liver tissue, blood, cecum contents and feces were collected at sacrifice before being flash frozen and stored at -80°C for subsequent assays. At the time of sacrifice, animals were 7 weeks of age.

4 ACS Paragon Plus Environment

Page 4 of 27

Page 5 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Measurement of cecal and fecal gut microbiota Fresh fecal samples were collected from animals aged 7 weeks prior to sacrifice, cecum contents were collected posthumously. Samples were stored at -80°C for future analysis. Microbial DNA was separately extracted from fecal matter and cecum contents using the FastDNA Spin Kit for Feces (MP Biomedicals, LLC, Solon, OH). DNA concentrations were quantified using the Nanodrop 2000 (Thermo Fisher Scientific Inc., Asherville, NC), diluted to 4ng/µl before being stored at -20°C until analysis.

Microbial profiling was conducted using an iCycler (BioRad Inc., Mississauga, ON) as previously described.15 In a 96 well plate, 20ng DNA samples were loaded in duplicate with SYBR Green qPCR Master Mix (BioRad) and microbial group specific 16S primers (total volume of 25ul/well). Group specific primers are shown in Supplemental Table S-1 and are referenced in a previously published work.15 Purified template DNA from reference strains (ATCC, Manassas, VA) was serially diluted 10-fold to generate a standard curve for each microbial group. Standard curves were normalized to copy number of 16S rRNA genes using reference strain genome size and previously published 16S rRNA gene copy number values.16 Threshold cycle values were used to calculate the number of 16S rRNA gene copies in each sample.

Metabolomics sample preparation Serum samples of 95-100 uL were filtered in Amicon Ultra filter devices with 10kDa cutoff (Millipore, Billerica, MA, USA) to remove macromolecules. Filtrates were filled up to 400µL with water, and 200 µL phosphate buffer and 50 µL deuterium oxide containing 3-(trimethylsilyl)2,2',3,3'-tetradeuteropropionic acid (TSP) were added as described previously.17 Liver samples were prepared according to previously published protocols.18 Briefly, frozen samples of roughly 100 mg were homogenized in a Fastprep-24 homogenizer (MP Biomedicals, Santa Ana, CA, USA) using 1.4 mm ceramic beads (“Lysing Matrix D”, MP Biomedicals). Before 5 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

homogenization, 400 µL of cold methanol and 85 µL of cold water were added. After homogenization, 400 µL chloroform and 200 µL water were added and the samples kept on ice for 10 minutes. Samples were centrifuged for 5 minutes at 2000 rcf at 4 °C. The aqueous and the lipophilic layer were collected separately and dried in a Vacufuge Concentrator 5301 (Eppendorf, Hamburg, Germany). Dried aqueous extracts were dissolved in 400 µL water and mixed with 200 µL buffer containing TSP and 50 µL deuterium oxide. Dried lipophilic extracts were dissolved in deuterated chloroform containing octamethylcyclotetrasiloxane (OMS) as internal standard.19

Metabolomics NMR measurements 1D 1H NOESY spectra of filtered serum, aqueous liver extracts and lipid liver extracts were measured at 298 K on an Avance II 600 MHz NMR spectrometer equipped with a tripleresonance probe, z-gradients, and a cooled automatic sample changer (Bruker BioSpin, Milton, ON, Canada). Serum and aqueous liver extracts were collected using water presaturation and spoil gradients for water signal suppression. 2D 1H-13C HSQC spectra for selected samples were measured on the same 600 MHz spectrometer and on a 700 MHz Bruker NMR spectrometer with z-gradients to assign ambiguous signals. Acquisition and processing parameters were chosen as previously published.20

Metabolomics data preparation Serum 1D NMR spectra were split into equally sized bins of 0.005 ppm width in the range of 0.5 to 9.5 ppm using AMIX version 3.9.14 (Bruker BioSpin). The bin size was chosen as visual inspections showed that numerous clearly separated peaks were summed up with neighboring peaks when employing larger bin sizes. Regions containing water, urea, and glycerol contaminations from the filters were excluded from binning (Supplemental Table S-2). For signals showing large inter-spectrum chemical shift variation, all affected bins were summed up. 6 ACS Paragon Plus Environment

Page 6 of 27

Page 7 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Bin intensities were scaled to the TSP signal of the respective sample, and then corrected for the filtrate amount. Aqueous liver extracts were binned as detailed above for serum samples. For signals showing large inter-spectrum chemical shift variation, all affected bins were summed up. The water resonance was excluded from binning. Bins were scaled to the TSP signal and corrected for the sample weight. For each sample set (serum, aqueous extracts), all bins with a mean intensity of less than 3.5 times the noise level were excluded from further analysis. After this, 603 and 867 bins were left for analysis for serum and aqueous extracts, respectively. Lipid signals were integrated from the lipophilic liver extracts relative to the OMS peak and then corrected for sample weight according to previous publications.21 Signals corresponding to saturated, unsaturated, and polyunsaturated fatty acid bonds, as well as total cholesterol were integrated for further analysis. Saturated, unsaturated, and polyunsaturated fatty acid signals were scaled by dividing each signal through the sum of these three signals for the respective sample. In this way, relative amounts of saturation were analyzed, to correct for changes in the overall liver lipid content. To take into account not only relative changes in lipid composition, but also absolute changes in lipid levels, levels of total cholesterol as well as total lipid levels were included in the analysis.

Statistical analyses All statistical analyses were performed in R version 3.0.2. Principal Component Analysis (PCA) with data centering but no data scaling was performed using the prcomp command. Microbial differences between the four groups were assessed using nonparametric Kruskal-Wallis test and subsequent nonparametric Mann-Whitney U-tests at a significance level of 0.05. Metabolic differences were assessed separately for serum metabolites, water-soluble liver metabolites and lipid liver metabolites using Kruskal-Wallis tests. Resulting p-values were corrected for multiple testing using False Discovery Rate (FDR) controlling. For all FDR analyses throughout 7 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

this manuscript, significance was assumed for FDR values below 20%.22 Significant spectral signals were assigned to metabolites using the HMDB database (www.hmdb.ca), the BMRB database (www.bmrb.wisc.edu), and through comparisons with in-house measurements of pure metabolites. Stability of the identified metabolites of energy metabolism in solution was confirmed by analysis of solutions of pure reference samples of the respective metabolites. To asses diet-related changes, the signals identified as significant by the Kruskal-Wallis tests were assessed by pairwise checks (C57 chow – C57 ketogenic and BTBR chow – BTBR ketogenic) using Mann-Whitney U-tests and subsequent FDR correction. Duplicate entries were excluded prior to Mann-Whitney U-tests by keeping for each identified metabolite only the bin showing the lowest Kruskal-Wallis p-value.

Fixed effect models were created using the R function lm. As a change of available nutrients might change the metabolites secreted by the microbes, interaction terms were included in the model to account for such interactions. For comparison, a mixed effect model, containing a random offset instead of the fixed term A to represent group differences, was used. For this the function lme from the package nlme was employed. Details on the models can be found in the Supporting Information. Complete datasets were available for 10 samples for each of the four groups, enabling a balanced modeling.

In a first step it was determined whether fixed or mixed effect models, the latter incorporating a random term instead of a fixed term to account for group differences, are better suited to explain the observed effects. For each metabolite, the two models, fixed effects and mixed effects, were compared using a Hausman specification test combined with FDR correction. The Hausman specification test can identify whether random effects are better suited for modeling a specific data set (Null Hypothesis) than fixed effects.23

8 ACS Paragon Plus Environment

Page 8 of 27

Page 9 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

RESULTS AND DISCUSSION Fecal but not cecal microbial profiles correlate with the serum metabolome Relatively little is known about the influence of specific microbial species on host metabolite levels. In this study we attempt to reveal the metabolic impact of different gut microbial groups in a gut microbial environment of natural complexity in vivo. Following 10-14 days of dietary intervention, fecal matter, cecal contents, serum and liver were collected from control (C57) and BTBR animals. Animal characteristics are shown in Table 1. As expected, the ketogenic diet resulted in elevated serum ketones and a corresponding decline in blood glucose levels. Initial analysis including both cecal and fecal data showed that cecal microbial groups exhibited a limited number of significant correlations with the metabolome and therefore, only fecal microbial profiles were used for further analyses. In the mouse, the cecum plays a role in the digestion of high fibre material, providing lubrication and also acting as a reserve for commensal bacteria, possibly facilitating re-inoculation following pathogen exposure.24 The surface area and absorptive capacity of the cecum is small compared to that of the large intestine, which may be one reason few metabolome-microbiota interactions were observed at this site.

Ketogenic diet causes a reduction in microbial abundance To induce changes in the microbiota, a ketogenic diet was employed in both control and BTBR mice. The diet significantly reduced both total microbe numbers as well as specific microbial groups (Table 2, Figure 1). The only microbe not showing any significant group differences in the multiple-group comparison was fecal C. coccoides. This antibiotic-like effect may explain some of the beneficial effects of the diet in a number of pathological disease states.25

Diet rather than genotype drives metabolomic alterations Metabolomics analysis was performed on serum and both aqueous and lipid fractions from the liver. Principal component analysis (PCA) showed strong differences in serum metabolites 9 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

between chow and ketogenic diets (Figure 2A) with 36 metabolites significantly changed in a diet-dependent manner (Figure 3A). The majority of these metabolites showed the same behavior in both BTBR and C57 mice. Of interest, the short chain fatty acid butyrate was markedly increased with a ketogenic diet in all mice examined. This change was in spite of the relatively low content of fermentable fiber in the ketogenic diet and is likely attributable to shifts in bacterial composition with the diet. Increasing levels of butyrate are thought to improve or maintain gut barrier function by providing energy for colonocytes26 which may explain why the diets are widely regarded as having positive immunomodulatory properties27. In higher order principal components (PC4), a partial group separation was found between different genotypes (Supplemental Figure S-1 A). As PC4 only accounted for 1.5% of the variance of the data set, as compared to 76.5% for PC1, the genotypic contribution to the serum metabolic diversity is low compared to the dietary contribution.

Inspection of a PCA of aqueous liver extracts also showed dietary differences but no strong genotype influence on the metabolome (Figure 2B, Supplemental Figure S-1 B). The ketogenic diet caused a metabolomic shift with 31 metabolites altered with the diet, 23 of them common for C57 and BTBR mice (Figure 3B). The ketogenic diet resulted in a reduction in metabolites related to energy metabolism in both serum and aqueous liver extracts, including ADP, AMP, NAD and NADP. These alterations are consistent with reports of enhanced ATP levels and altered mitochondrial bioenergetics with the ketogenic diet.28

As the ketogenic diet has a large influence on liver fatty acid metabolism, changes in lipidrelated metabolites were expected (Figure 2C, Supplemental Figure S-1 C, D). Increases in total lipids and cholesterol were observed in the ketogenic diet for both genotypes (Figure 3C). Saturation of lipids increased in the ketogenic diet, while relative unsaturation decreased. This

10 ACS Paragon Plus Environment

Page 10 of 27

Page 11 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

increase in fatty acid saturation most likely stems from the ketogenic diet which predominantly contains saturated fat.

Metabolome-microbiota interactions and modelling the contribution of individual microbial groups As the gut microbiota largely impacts the availability of nutrients for the organism, the interactions of fecal gut microbes and metabolites of the host organism were further examined. While strong associations were noted for serum metabolites, no significant correlations to fecal microbial groups were observed for liver metabolites. Correlations were modelled based on the entirety of the data set rather than either diet or genotype alone. Hausman specification tests revealed that mixed rather than fixed effect models were better suited for metabolite-microbiota modeling. For more details on modeling and significance levels see the Materials and Methods and Supporting Information sections.

After controlling for False Discovery Rate (FDR), 28 significant correlations were found between serum metabolites and fecal microbial groups (Table 3). To visualize these correlations, a figure showing serum metabolites, fecal microbial groups and significant correlations was created (Figure 4). Results from this figure are discussed in the following paragraphs. C. leptum showed by far the highest number of significant correlations (10 total). The observation that C. leptum correlates to a high number of metabolites that are not correlated to any other microbes indicates a unique effect on the host metabolism that is not shared with other microbes.

We show levels of C. leptum to be sensitive to, and fluctuate widely in response to dietary manipulation. In particular, C. leptum was negatively correlated with serum ketones and positively correlated to glucose levels. It is noteworthy that these correlations were found irrespective of diet and genotype. Animals on a ketogenic diet but with high levels of C. leptum 11 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

thus exhibited reduced levels of ketones as compared to the group mean suggesting that C. leptum might facilitate adaptation to the ketogenic diet by maximizing the extraction of the minimally available carbohydrate. C. leptum also produces butyrate, a primary energy substrate for colonocytes, inferring both anti-inflammatory and anti-allergic effects,26, 29 and may be functioning to maintain colonocyte integrity in the absence of available carbohydrate.

Modelling shows Bifidobacterium spp. to exhibit 5 correlations to serum metabolites. Bifidobacterium is an important probiotic and has numerous positive effects on host health, among others by improving mucosal barrier function and thus reducing the amount of endotoxin and/or bacterial translocation.30 Lactobacillus spp. and Akkermansia muciniphila show correlations to 4 host serum metabolites each. Lactobacillus are known to adhere to the epithelial layer and can thus improve epithelial barrier function.31

A. muciniphila showed positive relationships to serum lactate, taurine and sarcosine yet a negative relationship to glutamate. Like many of the other microbes assessed in this study, A. muciniphila is also connected to barrier function of the intestinal mucosa by degrading mucin and disrupting the barrier function.32 The relationship with sarcosine is intriguing as this metabolite can be produced from bacterial sources and has been associated with an array of diseases including schizophrenia33 and prostate cancer34.

Taurine levels are affected by all top 4 bacterial species mentioned so far, but in opposite directions. Positive correlations were noted for both Bifidobacterium spp. and A. muciniphila whilst negative correlations were found for C. leptum and Lactobacillus spp. The correlation of taurine to specific gut microbes might thus be explained by the microbial degradation of bile35 and subsequent re-absorption of taurine by the host. For Bacteroides/Prevotella spp. and αhydroxybutyrate, different correlations were found depending on genotype and diet. While 12 ACS Paragon Plus Environment

Page 12 of 27

Page 13 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Bacteroides/Prevotella spp. was associated with decreased α-hydroxybutyrate values in most experimental groups, it was connected to increased α-hydroxybutyrate in BTBR mice on a ketogenic diet. Reasons for these discrepant results are not known, but warrant further investigation.

CONCLUSIONS The next frontier in examining the microbiota is to decipher how it communicates to the host to mediate health and disease. To better understand metabolome-microbiota relationships we modelled a relatively small number of microbial groups under varied genetic and dietary manipulations in vivo. Results showed very robust relationships providing a basis for future work in this field wherein the compositional and functional relationship of entire microbiomes (via sequencing) can be modelled in the context of the metabolome. This may, in the long term, enable information from serum metabolomics to be used as an indicator of gut microbiota composition, function or alterations due to nutritional or physiological insult.

13 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ASSOCIATED CONTENT Supporting Information Supporting information contains details on fixed and mixed effect models; a table with primer sequences and respective references; a table with raw p-values for microbial data; and figures with additional PCA results. This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected] *Phone: +1 403 220 7622

Conflict of Interest Disclosure The authors declare no competing financial interest.

Author contributions JS designed and supervised the study. CN and JS performed animal work. MB, RAR and CN measured and assessed microbiota abundance. MSK, HV prepared serum and liver samples, performed NMR analyses. MSK processed NMR data, performed statistical analyses and programmed linear models. DH provided supervision and statistical support. MSK wrote the manuscript.

ACKNOWLEDGMENTS MSK is supported by an Eyes High postdoctoral fellowship from the University of Calgary. CN is supported by a Mitocanada scholarship. HJV currently holds the Lance Armstrong Chair in Molecular Cancer Research. JS is supported by the National Science and Engineering Council 14 ACS Paragon Plus Environment

Page 14 of 27

Page 15 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

of Canada and Mitocanda. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. We want to thank Virginia Berry for assistance in liver sample extraction.

REFERENCES (1)

Bäckhed, F., Ding, H., Wang, T., Hooper, L. V., Koh, G. Y., Nagy, A., Semenkovich,

C. F., Gordon, J. I. The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci U S A 2004, 101, 15718–15723. (2)

Bleau, C., Karelis, A. D., St-Pierre, D. H., Lamontagne, L. Crosstalk between intestinal

microbiota, adipose tissue and skeletal muscle as an early event in systemic low-grade inflammation and the development of obesity and diabetes. Diabetes Metab Res Rev 2015, 31, 545–51. (3)

MacFabe, D. F. Enteric short-chain fatty acids: microbial messengers of metabolism,

mitochondria, and mind: implications in autism spectrum disorders. Microb Ecol Health Dis 2015, 26, 28177. (4)

Frank, D. N., St Amand, A. L., Feldman, R. A., Boedeker, E. C., Harpaz, N., Pace, N. R.

Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proc Natl Acad Sci U S A 2007, 104, 13780–13785. (5)

Karlsson, F. H., Fåk, F., Nookaew, I., Tremaroli, V., Fagerberg, B., Petranovic, D.,

Bäckhed, F., Nielsen, J. Symptomatic atherosclerosis is associated with an altered gut metagenome. Nat Commun 2012, 3, 1245. (6)

Tremaroli, V., Bäckhed, F. Functional interactions between the gut microbiota and host

metabolism. Nature 2012, 489, 242–249. (7)

Thomas, C. M., Versalovic, J. Probiotics-host communication: Modulation of signaling

pathways in the intestine. Gut Microbes 2010, 1, 148–163. 15 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(8)

Wikoff, W. R., Anfora, A. T., Liu, J., Schultz, P. G., Lesley, S. A., Peters, E. C., Siuzdak,

G. Metabolomics analysis reveals large effects of gut microflora on mammalian blood metabolites. Proc Natl Acad Sci U S A 2009, 106, 3698–3703. (9)

Antunes, L. C. M., Finlay, B. B. A comparative analysis of the effect of antibiotic

treatment and enteric infection on intestinal homeostasis. Gut Microbes 2011, 2, 105–108. (10)

Dunn, L., Suckling, J. Studies of the genetic variability in wild populations of house mice.

I. Analysis of seven alleles at locus T. Genetics 1956, 41, 344. (11)

Karunakaran, S., Manji, A., Yan, C. S., Wu, Z.-J. J., Clee, S. M. Moo1 obesity

quantitative trait locus in BTBR T+ Itpr3tf/J mice increases food intake. Physiol Genomics 2013, 45, 191–199. (12)

Stoehr, J. P., Nadler, S. T., Schueler, K. L., Rabaglia, M. E., Yandell, B. S., Metz, S. A.,

Attie, A. D. Genetic obesity unmasks nonlinear interactions between murine type 2 diabetes susceptibility loci. Diabetes 2000, 49, 1946–1954. (13)

Whitehead, G. S., Walker, J. K. L., Berman, K. G., Foster, W. M., Schwartz, D. A.

Allergen-induced airway disease is mouse strain dependent. Am J Physiol Lung Cell Mol Physiol 2003, 285, L32–L42. (14)

Ruskin, D. N., Svedova, J., Cote, J. L., Sandau, U., Rho, J. M., Kawamura, M., Jr,

Boison, D., Masino, S. A. Ketogenic diet improves core symptoms of autism in BTBR mice. PLoS One 2013, 8, e65021. (15)

Bomhof, M. R., Saha, D. C., Reid, D. T., Paul, H. A., Reimer, R. A. Combined effects of

oligofructose and Bifidobacterium animalis on gut microbiota and glycemia in obese rats. Obesity (Silver Spring) 2014, 22, 763–771. (16)

Stoddard, S. F., Smith, B. J., Hein, R., Roller, B. R. K., Schmidt, T. M. rrnDB: improved

tools for interpreting rRNA gene abundance in bacteria and archaea and a new foundation for future development. Nucleic Acids Res 2015, 43, D593–D598.

16 ACS Paragon Plus Environment

Page 16 of 27

Page 17 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

(17)

Klein, M. S., Buttchereit, N., Miemczyk, S. P., Immervoll, A.-K., Louis, C., Wiedemann,

S., Junge, W., Thaller, G., Oefner, P. J., Gronwald, W. NMR metabolomic analysis of dairy cows reveals milk glycerophosphocholine to phosphocholine ratio as prognostic biomarker for risk of ketosis. J Proteome Res 2012, 11, 1373–1381. (18)

Wu, H., Southam, A. D., Hines, A., Viant, M. R. High-throughput tissue extraction

protocol for NMR- and MS-based metabolomics. Anal Biochem 2008, 372, 204–212. (19)

Thomas, A., Stevens, A. P., Klein, M. S., Hellerbrand, C., Dettmer, K., Gronwald, W.,

Oefner, P. J., Reinders, J. Early changes in the liver-soluble proteome from mice fed a nonalcoholic steatohepatitis inducing diet. Proteomics 2012, 12, 1437–1451. (20)

Gronwald, W., Klein, M. S., Kaspar, H., Fagerer, S. R., Nürnberger, N., Dettmer, K.,

Bertsch, T., Oefner, P. J. Urinary metabolite quantification employing 2D NMR spectroscopy. Anal Chem 2008, 80, 9288–9297. (21)

Klein, M. S., Almstetter, M. F., Schlamberger, G., Nürnberger, N., Dettmer, K., Oefner,

P. J., Meyer, H. H. D., Wiedemann, S., Gronwald, W. Nuclear magnetic resonance and mass spectrometry-based milk metabolomics in dairy cows during early and late lactation. J Dairy Sci 2010, 93, 1539–1550. (22)

National Association of Testing Authorities, Australia. Technical Note 17 - Guidelines for

the validation and verification of quantitative and qualitative test methods. 2013. (23)

Hausman, J. A. Specification tests in econometrics. Econometrica 1978, 46, 1251–1271.

(24)

Randal Bollinger, R., Barbas, A. S., Bush, E. L., Lin, S. S., Parker, W. Biofilms in the

large bowel suggest an apparent function of the human vermiform appendix. J Theor Biol 2007, 249, 826–831. (25)

Kossoff, E. H., Rho, J. M. Ketogenic diets: evidence for short- and long-term efficacy.

Neurotherapeutics 2009, 6, 406–414.

17 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(26)

Donohoe, D. R., Garge, N., Zhang, X., Sun, W., O’Connell, T. M., Bunger, M. K.,

Bultman, S. J. The microbiome and butyrate regulate energy metabolism and autophagy in the mammalian colon. Cell Metab 2011, 13, 517–526. (27)

Dupuis, N., Curatolo, N., Benoist, J.-F., Auvin, S. Ketogenic diet exhibits anti-

inflammatory properties. Epilepsia 2015, 56, e95–e98. (28)

Sullivan, P. G., Rippy, N. A., Dorenbos, K., Concepcion, R. C., Agarwal, A. K., Rho,

J. M. The ketogenic diet increases mitochondrial uncoupling protein levels and activity. Ann Neurol 2004, 55, 576–580. (29)

Sokol, H., Pigneur, B., Watterlot, L., Lakhdari, O., Bermúdez-Humarán, L. G.,

Gratadoux, J.-J., Blugeon, S., Bridonneau, C., Furet, J.-P., Corthier, G., Grangette, C., Vasquez, N., Pochart, P., Trugnan, G., Thomas, G., Blottière, H. M., Doré, J., Marteau, P., Seksik, P., Langella, P. Faecalibacterium prausnitzii is an anti-inflammatory commensal bacterium identified by gut microbiota analysis of Crohn disease patients. Proc Natl Acad Sci U S A 2008, 105, 16731–16736. (30)

Griffiths, E. A., Duffy, L. C., Schanbacher, F. L., Qiao, H., Dryja, D., Leavens, A.,

Rossman, J., Rich, G., Dirienzo, D., Ogra, P. L. In vivo effects of bifidobacteria and lactoferrin on gut endotoxin concentration and mucosal immunity in Balb/c mice. Dig Dis Sci 2004, 49, 579–589. (31)

Sengupta, R., Altermann, E., Anderson, R. C., McNabb, W. C., Moughan, P. J., Roy,

N. C. The role of cell surface architecture of lactobacilli in host-microbe interactions in the gastrointestinal tract. Mediators Inflamm 2013, 2013, 237921. (32)

Everard, A., Belzer, C., Geurts, L., Ouwerkerk, J. P., Druart, C., Bindels, L. B., Guiot, Y.,

Derrien, M., Muccioli, G. G., Delzenne, N. M., de Vos, W. M., Cani, P. D. Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity. Proc Natl Acad Sci U S A 2013, 110, 9066–9071.

18 ACS Paragon Plus Environment

Page 18 of 27

Page 19 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

(33)

Tsai, G., Lane, H.-Y., Yang, P., Chong, M.-Y., Lange, N. Glycine transporter I inhibitor,

N-methylglycine (sarcosine), added to antipsychotics for the treatment of schizophrenia. Biol Psychiatry 2004, 55, 452–456. (34)

Sreekumar, A., Poisson, L. M., Rajendiran, T. M., Khan, A. P., Cao, Q., Yu, J., Laxman,

B., Mehra, R., Lonigro, R. J., Li, Y., Nyati, M. K., Ahsan, A., Kalyana-Sundaram, S., Han, B., Cao, X., Byun, J., Omenn, G. S., Ghosh, D., Pennathur, S., Alexander, D. C., Berger, A., Shuster, J. R., Wei, J. T., Varambally, S., Beecher, C., Chinnaiyan, A. M. Metabolomic profiles delineate potential role for sarcosine in prostate cancer progression. Nature 2009, 457, 910– 914. (35)

Gerard, P. Metabolism of cholesterol and bile acids by the gut microbiota. Pathogens

2013, 3, 14–24.

19 ACS Paragon Plus Environment

Journal of Proteome Research

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 27

TABLES Table 1. Animal characteristics by diet and mouse genotype C57 Chow

C57 Ketogenic

18.4±0.8

a

Blood Glucose (mM)

10.4±0.6

a

Blood Ketones (mM)

0.9±0.1

Mass (g)

a

10.5±0.3

b

4.3±0.5

b

5.1±0.8

b

BTBR Chow

BTBR Ketogenic

28.6±1.3

c

15.9±1.0

a

7.8±0.3

c

3.7±0.5

b

0.9±0.1

a

5.1±0.8

b

p