Atlantia is a world-class provider of human clinical studies. For over a decade we have delivered clinical results to our many clients around the world, on time and on budget. We strive to offer the highest quality science while maintaining a cost-conscious approach to maximize the value proposition for each of our clients.
Gut Online First, published on March 30, 2017 as 10.1136/gutjnl-2016-313627 group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
ORIGINAL ARTICLE The microbiome of professional athletes differs from that of more sedentary subjects in composition and particularly at the functional metabolic level Wiley Barton, 1,2,3 Nicholas C Penney, 4,5 Owen Cronin, 1,3 Isabel Garcia-Perez, 4 Michael G Molloy, 1,3 Elaine Holmes, 4 Fergus Shanahan, 1,3 Paul D Cotter, 1,2 Orla O ’ Sullivan 1,2
ABSTRACT Objective It is evident that the gut microbiota and factors that in fl uence its composition and activity effect human metabolic, immunological and developmental processes. We previously reported that extreme physical activity with associated dietary adaptations, such as that pursued by professional athletes, is associated with changes in faecal microbial diversity and composition relative to that of individuals with a more sedentary lifestyle. Here we address the impact of these factors on the functionality/metabolic activity of the microbiota which reveals even greater separation between exercise and a more sedentary state. Design Metabolic phenotyping and functional metagenomic analysis of the gut microbiome of professional international rugby union players (n=40) and controls (n=46) was carried out and results were correlated with lifestyle parameters and clinical measurements (eg, dietary habit and serum creatine kinase, respectively). Results Athletes had relative increases in pathways (eg, amino acid and antibiotic biosynthesis and carbohydrate metabolism) and faecal metabolites (eg, microbial produced short-chain fatty acids (SCFAs) acetate, propionate and butyrate) associated with enhanced muscle turnover ( fi tness) and overall health when compared with control groups. Conclusions Differences in faecal microbiota between athletes and sedentary controls show even greater separation at the metagenomic and metabolomic than at compositional levels and provide added insight into the diet – exercise – gut microbiota paradigm. INTRODUCTION Regular exercise challenges systemic homeostasis resulting in a breadth of multiorgan molecular and physiological responses, including many that centre on immunity, metabolism and the microbiome – gut – brain axis. 1 – 5 Exercise exhibits systemic and end-organ anti-in fl ammatory effects as well as con- tributing to more ef fi cient carbohydrate metabol- ism, in addition to trophic effects at the level of the central nervous system. 6 7 In fact, increasing phys- ical activity offers an effective treatment and pre- ventative strategy for many chronic conditions in which the gut microbiome has been implicated. 8 – 10 Conversely, a sedentary lifestyle is a major
▸ Additional material is published online only. To view please visit the journal online (http://dx.doi.org/10.1136/ gutjnl-2016-313627). 1 Alimentary Pharmabiotic Centre Microbiome Institute, University College Cork, National University of Ireland, Cork, Ireland 2 Teagasc Food Research Centre, Cork, Ireland 3 Department of Medicine, University College Cork, National University of Ireland, Cork, Ireland 4 Section of Biomolecular Medicine, Division of Surgery and Cancer, Imperial College London, London, UK 5 Division of Surgery, Department of Surgery and Cancer, Imperial College London, London, UK Computational Systems Medicine, Department of Correspondence to Professor Fergus Shanahan, APC Microbiome Institute, University College Cork, National University of Ireland, Cork, T12 DC4A Ireland; f.shanahan@ucc.ie Received 20 December 2016 Revised 31 January 2017 Accepted 6 March 2017
Signi fi cance of this study
contributing factor to morbidity in developed Western society and is associated with heightened risk of numerous diseases of af fl uence , such as obesity, diabetes, asthma and cardiovascular disease (CVD). 11 – 14 Recent evidence supports an in fl uential role for the gut microbiome in these diseases. 15 – 23 The concept that regular exercise and sustained levels of increased physical activity foster or assist the maintenance of a preferential intestinal micro- biome has recently gained momentum and con fi rmed and the separation between athletes and those with a more sedentary lifestyle is even more evident at the functional or metabolic level. Microbial-derived SCFAs are enhanced within the athletes. How might it impact on clinical practice in the foreseeable future? ▸ The fi ndings provide new evidence supporting the link between exercise and metabolic health. The fi ndings provide a platform for the rational design of diets for those engaged in vigorous exercise. The identi fi cation of speci fi c alterations in the metabolic pro fi le of subjects engaged in high levels of exercise provides insight necessary for future efforts towards targeted manipulation of the microbiome. What is already known on this subject? ▸ Taxonomic and functional compositions of the gut microbiome are emerging as biomarkers of human health and disease. ▸ Physical exercise and associated dietary adaptation are linked with changes in the composition of the gut microbiome. ▸ Metabolites such as short-chain fatty acids (SCFAs) have an impact on a range of health parameters including immunity, colonic epithelial cell integrity and brain function. What are the new fi ndings? ▸ Our original observation of differences in gut microbiota composition in elite athletes is
To cite: BartonW, Penney NC, Cronin O, et al . Gut Published Online First: [ please include Day Month Year] doi:10.1136/gutjnl- 2016-313627
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
1
Copyright Article author (or their employer) 2017. Produced by BMJ Publishing Group Ltd (& BSG) under licence.
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
interest. 24 – 29 Previously, using 16S rRNA amplicon sequencing, we demonstrated taxonomic differences in gut microbiota between an elite athlete cohort of international-level rugby players and a group of age-matched high (>28 kg/m 2 ) and low (<25kg/m 2 ) body mass index (BMI) controls. 26 This analysis illustrated a signi fi cantly greater intestinal microbial diversity among the athletes compared with both control groups. This taxonomic diversity signi fi cantly correlated with exercise and dietary protein consumption. However, the possibility existed that these differences did not equate to differences at a func- tional level. Here, we re-examine the microbiome in these parti- cipants by whole metagenome shotgun sequencing to provide deeper insight into taxonomic composition and functional potential and by complementary metabolic phenotyping ana- lyses of host-derived and microbial-derived (urine and faecal, respectively) metabolic pro fi les. This analysis shows that the dif- ferences in the gut microbiota between athletes and controls is even more pronounced at the functional metabolic level than at the compositional level as previously reported and provides further rationale for prospective controlled studies to unravel the relationship between diet, exercise and the gut microbiome. RESULTS The study groups comprise professional male athletes (n=40) and healthy controls (n=46). 26 To better represent the variabil- ity of BMI in the athletes, controls were classi fi ed as either low BMI (n=22, BMI ≤ 25.2) or high BMI (n=24, BMI ≥ 26.5). Participants made no report of GI distress or alterations of GI transit time throughout the course of the initial study. Functional structure of the enteric microbiome correlates with athletic state Functional metagenomic analysis of faecal samples allowed for the prediction of the operational potential of each individual ’ s microbiota. In total, 19 300 taxonomically linked metabolic pathways were identi fi ed in at least one individual. Comparison of phylogenetic constructions derived from the 16S rRNA amplicon data of our previous study and the functional data of
this present report reveals a greater level of identi fi cation at higher levels of taxonomy (eg, phylum) for 16S sequences, 26 while the metagenomic data had greater fi delity and superior resolution of lower levels of taxonomy (eg, species) ( fi gure 1). Consistent with previous results, the microbiota of the athletes were signi fi cantly more diverse than that of both the low-BMI and high-BMI control groups at the functional level ( fi gure 2A). Furthermore, our previous fi ndings of an enrichment of Akkermansia in athletes was corroborated by the presence of sig- ni fi cantly higher proportions of metabolic pathways associated with this genus in athletes when compared with high-BMI con- trols (p<0.001). Correlation analysis revealed that, of the total 19 300 pathways, 98 were signi fi cantly altered between the three cohorts (p<0.05) (see online supplementary table S1). Subsequently, large-scale functional dissimilarity between ath- letes and controls was determined and distinct patterns of pathway composition between groups were revealed (see online supplementary fi gure S1A). This functional distinction remained true whether applied to total pathway data or to the statistically signi fi cant subset of pathways (see online supplementary fi gure S1B). Correlation of pathways present in at least one member from both cohorts further exempli fi ed the uniformity of the ath- letes and the division between the athletes and control groups (see online supplementary fi gure S1C). Separation according to group membership was further illustrated through principal coordinate analysis (PCoA), with statistical support of the sig- ni fi cant separation between the athletes and both control groups (p<0.05) ( fi gure 2B). This was also the case for the statistically signi fi cant subset of pathways (see online supplementary fi gure S1D). Principal component analysis (PCA) supplemented with a correspondence analysis and k-nearest neighbour semisupervised learning approach cast further light (ie, visualisation of robustly de fi ned class associations of speci fi c individuals within the groups) on the clustering of participants within and between cohorts (see online supplementary fi gure S1E). Pathways exhibiting statistically signi fi cant variation between the athletes and both control groups were organised accord- ing to MetaCyc metabolic pathway hierarchy classi fi cation
Figure 1 Comparison of phylogenetic constructions from metagenomic and 16S rRNA gene sequencing sourced from all participants. Phylogenetic trees derived from (A) metagenomic sequencing and (B) 16S rRNA amplicon sequencing. Taxonomic levels are assigned from centre out with kingdom-level assignment in centre and strain-level assignment in outer most ring. Dark blue radial highlights correspond to poorly identi fi ed taxonomies (ie, ‘ unknown ’ and ‘ unassigned ’ database entries). Number of assignments at each level of phylogeny is displayed below the respective graph. Taxonomic trees derived from the two sequencing approaches illustrate an advantage of metagenomic sequencing in the number of predictions of lower taxonomic levels and the frequency of full identi fi cation of taxa, while 16S rRNA sequencing grants greater insight of high-level phylogenies within the population.
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
2
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
Figure 2 Group-wise comparison of microbial metagenomic and metabolomic pro fi les. (A) Shannon index of diversity for metabolic pathways from all three groups. Pathway diversity is increased in the athlete group when compared with low-body mass index (BMI) and high-BMI controls. Diversity measures are statistically signi fi cant between low-BMI and athlete groups ( p<0.049), with statistical signi fi cance between all groups (Kruskal-Wallis p<0.05). (B) Principle coordinate analysis of Bray-Curtis compiled distance matrix of all microbial metabolic pathway relative abundances. Groups show signi fi cant variation from one another (Adonis PERMANOVA p<0.05). Cross-validated orthogonal partial least squares regression discriminant analysis (OPLS-DA) of full nuclear magnetic resonance ( 1 H-NMR) spectra from urine (R 2 Y=0.86, Q 2 Y=0.60) (C) and faecal water (R 2 Y=0.86, Q 2 Y=0.52) (D) samples. OPLS-DA displays robust separation between athletes and controls. Models comprise 1 predictive (tcv[1]) and 1 orthogonal (tocv[1]) principal component.
was predominantly correlated with CK, total bilirubin and total energy intake. Of the total number of metabolic pathways with associations to the clinical data from all three groups (10 760; data not shown), relevant pathways related to the production of secondary metabolites, cofactors and short-chain fatty acids (SCFAs) were identi fi ed (eg, biotin biosynthesis and pyruvate fermentation to butanoate). Distinct differences between host and microbial metabolites in athletes and controls A combination of multiplatform metabolic phenotyping and multivariate analysis based on orthogonal partial least squares discriminant analysis (OPLS-DA) was used to compare urine and faecal samples from athletes and controls. The cross-validated (CV) OPLS-DA models show strong differences between athletes and controls in urine samples by proton nuclear magnetic reson- ance ( 1 H-NMR) analysis (R 2 Y=0.86, Q 2 Y=0.60, fi gure 2C), hydrophilic interaction ultra-performance liquid chromatog- raphy mass spectroscopy (HILIC UPLC-MS) positive mode ana- lysis (R 2 Y=0.85, Q 2 Y=0.74, online supplementary fi gure S2A) and reversed-phase UPLC-MS (RP UPLC-MS) in both positive
(34 metabolic categories), highlighting a number of differences ( fi gure 3A and online supplementary table S2). Distinct cluster- ing patterns were observed within each cohort, with the high-BMI control group having the lowest average abundance scores across 31 metabolic pathway categories (the exceptions being vitamin biosynthesis (VB), lipid biosynthesis (LB) and amino acid biosynthesis (AAB) categories). The athlete group had the highest mean abundance across 29 of the 34 metabolic categories (eg, carbohydrate biosynthesis, cofactor biosynthesis and energy metabolism) (see online supplementary table S2). Numerous statistically signi fi cant (p<0.05) associations were identi fi ed between pathway abundances and serum creatine kinase (CK) — an enzymatic marker of muscle activity (IU/L), total bilirubin (IU/L) and dietary macronutrient intake of protein (g/day), fi bre (g/day), carbohydrates (g/day), sugars (g/day), starch (g/day), fat (g/day) and total energy (KJ/day) ( fi gure 3B). Each group was represented by distinct association pro fi les of the correlation between clinical measurements and metagenomic pathways. Dietary factors, sugars and other carbo- hydrates, as well as energy intake, provide the majority of the correlations for the control groups, whereas the athlete group
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
3
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
Figure 3 Group variation of microbial metabolic function and associations between pathways and clinical and dietary variables. (A) Mean relative abundance values of statistically signi fi cant (Kruskal-Wallis p<0.05) metabolic pathways binned according to categories of metabolic function. (B) Number of metabolic pathways signi fi cantly (Benjamini-Hochberg corrected p<0.05) correlated with dietary constituents and blood serum metabolites. BMI, body mass index.
and negative mode analysis (R 2 Y=0.83, Q 2 Y=0.73 and R 2 Y=0.83, Q 2 Y=0.67, online supplementary fi gure S2B,C respectively). Likewise, the CV-OPLS-DA models comparing faecal samples, although weaker than the urine models, reveal signi fi cant differences between athletes and controls by 1 H-NMR analysis (R 2 Y=0.86, Q 2 Y=0.52, fi gure 2D) and HILIC UPLC-MS positive mode analysis (R 2 Y=0.65, Q 2 Y=0.34, online supplementary fi gure S2D). The loadings of the pairwise OPLS-DA models were used to identify metabolites discriminating between the two classes. Athletes ’ 1 H-NMR metabolic phenotypes were characterised by higher levels of trimethylamine- N -oxide (TMAO), L-carnitine, dimethylglycine, O-acetyl carnitine, proline betaine, creatine, acetoacetate, 3-hydroxy-isovaleric acid, acetone, N -methylnicotinate, N -methylnicotinamide, phenylacetylglutamine (PAG) and 3-methylhistidine in urine samples and higher levels of propionate, acetate, butyrate, trimethylamine (TMA), lysine and methylamine in faecal samples, relative to controls. Athletes were further characterised by lower levels of glycerate, allantoin and succinate and lower levels of glycine and tyrosine relative to controls in urine and faecal samples, respectively (see online supplementary table S3). While numerous metabolites discriminated signi fi cantly between athletes and controls with RP UPLC-MS positive (490)
and negative (434) modes for urine, as well as with HILIC UPLC-MS positive mode for urine (196) and faecal water (3), key metabolites were structurally identi fi ed using the strategy described below. UPLC-MS analyses revealed higher urinary excretion of N -formylanthranilic acid, hydantoin-5-propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), CMPF glucuronide, trimetaphosphoric acid, acetyl- carnitine (C2), propionylcarnitine (C3), isobutyrylcarnitine (C4), 2-methylbutyroylcarnitine (C5), hexanoylcarnitine (C6), C9:1-carnitine, L-valine, nicotinuric acid, 4-pyridoxic acid and creatine in athletes relative to controls. Levels of glutamine, 7-methylxanthine, imidazoleacetic acid, isoquinoline/quinolone were lower in athletes ’ urinary samples relative to controls. In addition, 16 unknown glucuronides were lower in the athlete samples (see online supplementary table S4). SCFA levels in faeces measured by targeted gas chromatography – mass spectrometry (GC-MS) showed signi fi cantly higher levels of acetate (p<0.001), propionate (p<0.001), butyrate (p<0.001) and valerate (p=0.011) in athletes relative to controls. Isobutyrate and isovalerate did not differ signi fi cantly between the groups ( fi gure 4B and online supplementary table S5). Furthermore, concentrations of propionate strongly correlated to protein intake, while butyrate was shown to have a strong association with intake of dietary fi bre (see online supplementary table S6).
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
4
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
Figure 4 Athletes display a pro fi le of short-chain fatty acids (SCFAs) that alters from that of the controls. (A) Heat map of bacterial taxa (family, genus and species level) that correlate with faecal short-chain fatty acid levels using Spearman ’ s correlation. Cool colours represent positive correlations; hot colours represent negative correlations (r). All taxa shown had a correlation p value<0.01. Those marked with * represent correlations with a false discovery rate <0.01 after Benjamini-Hochberg multiple testing corrections. (B) Median concentrations of GC-MS-derived faecal SCFA. Quantitative analysis of SCFAs in faecal samples shows signi fi cant increase in measured concentrations of acetate, propionate, butyrate and valerate in athletes. Error bars represent 95% CIs. (C) Quanti fi cation of statistically relevant correlations of metabolic pathways to GC-MS-derived faecal SCFA concentrations ( μ M). BMI, body mass index; GC-MS, gas chromatography – mass spectrometry.
It was also observed that 16 genera correlated with 12 metabo- lites (see online supplementary table S9).
Correlating metabonomic and metagenomic results Correlation analysis between targeted measurements of SCFAs and taxonomic data from 16S rRNA sequencing revealed a number of correlations that remained signi fi cant following cor- rection; Roseburia was positively correlated with acetate (p=0.004) and butyrate (p=0.018) while Family XIII Incertae Sedis was positively correlated with isobutyrate (p<0.001), iso- valeric acid (p<0.001) and valeric acid (p=0.008) ( fi gure 4A and online supplementary table S7). SCFAs were also correlated with pathway relative abundances, with all SCFAs associating with considerably more pathways in the athletes versus the controls ( fi gure 4C). Multiple statistically signi fi cant (7948) (p<0.05) correlations between the metabolic pathways and SCFAs were identi fi ed (see online supplementary table S8). Two distinct blocks of proportionately discriminant correlations were observed with isobutyric and isovaleric acids, which were more abundant in the athletes, while acetic and butyric acids were proportionately more abundant in controls. Correlations of the SCFA concentrations to pathways related to fermentation, biosynthesis or modi fi cation of fatty acids were identi fi ed among the numerous other associations (see online supplementary table S8 for complete list). Additional correla- tions of metabolic pathways against well-identi fi ed metabolites detected from both faecal water ( fi gure 5A, C) and urine ( fi gure 5B, D) presented numerous signi fi cant associations (6186 and 13 412, respectively; data not shown) (p<0.05).
DISCUSSION The results con fi rm enhancement of microbial diversity in ath- letes compared with controls. Supporting previous insights into the bene fi cial in fl uence of physical exercise and associated diet on the compositional structure of the gut microbiota, 25 26 30 this study has extended the paradigm to include links between physical fi tness and the functional potential of the gut micro- biota and its metabolites. It must be conceded that some ath- letes, although fi t, may not necessarily be more healthy. 31 Athletes have an increased abundance of pathways that — given an equivalent amount of expression activity — could be exploited by the host for potential health bene fi t, including bio- synthesis of organic cofactors and antibiotics, as well as carbo- hydrate degradation and secondary metabolite metabolism. 32 Furthermore, athletes have an enriched pro fi le of SCFAs, previ- ously associated with numerous health bene fi ts and a lean phenotype. 33 – 35 While interpretation of SCFA data can be dif fi - cult as levels represent a combination of SCFA production and host-absorption rates, it is notable that, as previously presented, the athletes ’ diet maintained signi fi cantly higher quantities of fi bre intake. 26 This along with an increased number of detected SCFA pathways in the athletes would be conducive to an enhanced rate of SCFA production 36
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
5
Figure 5 Distinctive association pro fi les of metabolic pathways to metabolites in athletes and controls. (A) Signi fi cant correlations of faecal water-derived metabolites and metabolic pathways, represented by number of correlations for each metabolite. (B) Urine metabolites signi fi cantly correlated to pathways and displayed as number of correlations. (C) Signi fi cant correlations shown in (A) displayed as proportions of total associations. (D) Correlations presented in (B) given as proportions of total associations. BMI, body mass index; PAG, phenylacetylglutamine; TMAO, trimethylamine- N -oxide.
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
muscle turnover, creatine, 3-methylhistidine and L-valine, and host metabolism, carnitine, are elevated in the athlete groups. Metabolites derived from vitamins and recovery supplements common in professional sports, including glutamine, lysine, 4- pyridoxic acid and nicotinamide, are also raised in the athlete group. It is notable that PAG, a microbial conversion product of phenylalanine, has been associated with a lean phenotype and is increased in the athletes. 42 Furthermore, PAG positively corre- lates with the genus Erysipelotrichaceae incertae sedis , whichwe have previously noted to be present in relatively higher propor- tions in the athlete group compared with both control groups. PAG is the strongest biomarker postbariatric surgery, where it is associated with an increase in the relative proportions of Proteobacteria as observed here in the athlete group. Within the SCFAs, two distinct clusters were observed; acetic acid, propio- nic acid and butyric acid correlate with dietary contributors ( fi bre and protein), while isobutyric acid, isovaleric acid and valeric acid correlate with microbial diversity. The same clusters are observed when correlating with individual taxa, in support of previously observed links between SCFAs and numerous metabolic bene fi ts and a lean phenotype. 33 – 35 Our ongoing work in this area with non-athletes engaging in a structured exercise regime looks to further explore compo- nents of the exercise and diet – microbiome paradigm, which, along with this study, may inform the design of exercise and fi tness programmes, including diet design in the context of opti- mising microbiota functionality for both athletes and the general population.
It was noted that athletes excreted proportionately higher levels of the metabolite TMAO, an end product metabolite of dietary protein degradation. Elevated TMAO has been observed in patients with cardiovascular disease and atheroscler- osis, highlighting a potential downside to increased protein intake. 15 – 17 22 37 However, TMAO is also found in high levels in the urine of Japanese populations, 38 who do not have high risk for CVD. Similarly to these populations, the athletes ’ diet contained a signi fi cantly greater proportion of fi sh. Our current understanding of the implications of this result remains limited and requires elaboration in future studies. Furthermore, pathway abundance in a metagenome merely re fl ects functional potential and not necessarily increased expression in situ. Variance of metagenomic composition between athletes and controls was exempli fi ed with unique pathway – pathway correla- tions between the two groups. Analysis of categorically arranged pathway abundances within the separate cohorts provided add- itional insight into the previously described dichotomy between the microbiota of athletes and high-BMI controls. The two groups displayed distinct structures of functional capacity, separ- ately oriented to operate under the different physiological milieu of the two groups. Notably, from a functional perspec- tive, the microbiota of the low-BMI group was more similar to the athletes. The low-BMI controls were generally engaged in a modestly active lifestyle, re fl ected by their leanness and increased levels of CK. It is speculative but not implausible that moderate improvements in physical activity for overweight and obese individuals may confer the bene fi cial metabolic functions observed within the athlete microbiome. Dietary contributions to the functional composition of the enteric microbial system are also evident in our study. The rela- tive abundances of pathways related to fundamental metabolic function — AAB, VB and LB — were higher on average within the high-BMI control group when compared with the athlete group. The mechanisms behind these differences are unclear and might re fl ect chronic adaptation of the athlete gut microbiome; pos- sibly due to a reduced reliance on the corresponding biosyn- thetic capacities of their gut microbiota. On the contrary, the athlete microbiome presents a functional capacity that is primed for tissue repair and to harness energy from the diet with increased capacity for carbohydrate, cell structure and nucleo- tide biosynthesis, re fl ecting the signi fi cant energy demands and high cell-turnover evident in elite sport. Remarkably, our examination of pathway correlation to dietary macronutrients and plasma CK, as a biomarker of exer- cise, 39 is suggestive of an impact of physical activity on the use of dietary nutrients by the microbiota of the gut. Comparing athletes to both high-BMI and low-BMI controls, a greater number of pathways correlating to speci fi c macronutrients with the controls suggests a shift in the dynamics of these varied metabolic functions. The impact of the athletes ’ increased protein intake compared with both control groups was evident in the metabolomic phenotyping results. By-products of dietary protein metabolism (mostly by microbes) including TMAO, carnitines, TMA, 3-CMPF and 3-hydroxy-isovaleric acid are all elevated in the athlete cohort. Of particular interest is 3-hydroxy-isovaleric acid (potentially from egg consumption), which has been demonstrated to have ef fi cacy for inhibiting muscle wasting when used in conjunction with physical exercise. 40 41 The compound is also commonly used as a sup- plement by athletes to increase exercise-induced gains in muscle size, muscle strength and lean body mass, reduce exercise-induced muscle damage and speed recovery from high- intensity exercise. 41 Numerous metabolites associated with
MATERIALS AND METHODS Study population
Elite professional male athletes (n=40) and healthy controls (n=46) matched for age and gender were enrolled in 2011 as previously described in the study. 26 Due to the range of physi- ques within a rugby team (player position dictates need for a variety of physical constitutions, ie, forward players tend to have larger BMI values than backs, often in the overweight/ obese range) the recruited control cohort was subdivided into two groups. To more completely include control participants, the BMI parameter for group inclusion was adjusted to BMI ≤ 25.2 and BMI ≥ 26.5 for the low-BMI and high-BMI groups, respectively. Approval for this study was granted by the Cork Clinical Research Ethics Committee. Acquisition of clinical, exercise and dietary data Self-reported dietary intake information was accommodated by a research nutritionist within the parameters of a food frequency questionnaire in conjunction with a photographic food atlas as per the initial investigation. 26 Fasting blood samples were col- lected and analysed at the Mercy University Hospital clinical laboratories, Cork. As the athletes were involved in a rigorous training camp, we needed to assess the physical activity levels of both control groups. To determine this, we used an adapted version of the EPIC-Norfolk questionnaire. 43 Creatine kinase levels were used as a proxy for level of physical activity across all groups. Preparation of metagenomic libraries DNA derived from faecal samples was extracted and puri fi ed using the QIAmp DNA Stool Mini Kit (cat. no 51 504) prior to storage at − 80°C. DNA libraries were prepared with the Nextera XT DNA Library Kit (cat. no FC-131-1096) prior to processing on the Illumina HiSeq 2500 sequencing platform (see online supplementary methods for further detail).
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
7
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
reference. LoA 2: MS/MS precursor and product ions or 1D +2D NMR chemical shifts and multiplicity match to a reference database or literature to putatively annotate compound. LoA 3: chemical shift ( δ ) and multiplicity matches a reference database to tentatively assign the compound (for further details see online supplementary methods). Twitter Follow Orla O ’ Sullivan @OrlaOS Acknowledgements The authors express gratitude to all participants for the donation of time and samples, in particular staff and players at the Irish Rugby Football Union. Fiona Fouhy for insight provided into the library preparation of metagenomic sequencing. The authors thank the Imperial-National Institute for Health Research (NIHR) Clinical Phenome Centre for support. Contributors WB prepared DNA samples for metagenomic sequencing. OOS and WB processed and analysed the metagenomic data. EH, IGP and NCP performed metabolomic processing and statistical analysis thereof. FS, PDC, OOS and WB devised experimental design and approach. FS, PDC, OC, OOS, MGM, EH, NCP and WB wrote manuscript. Results discussed by all authors. Funding This research was funded by Science Foundation Ireland in the form of a centre grant (APC Microbiome Institute Grant Number SFI/12/RC/2273). Research in the Cotter laboratory is funded by SFI through the PI award, ‘ Obesibiotics ’ (11/PI/ 1137). OOS and WB are funded by Science Foundation Ireland through a Starting Investigator Research Grant award (13/SIRG/2160). Nicholas Penney is funded by the Diabetes Research and Wellness Foundation through the Sutherland-Earl Clinical Research Fellowship 2015. The centre is supported by the NIHR Imperial Biomedical Research Centre based at Imperial College Healthcare National Health Service (NHS) Trust and Imperial College London. Disclaimer The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. Competing interests FS is a founder shareholder in Atlantia Food Clinical Trials, Tucana Health and Alimentary Health. He is director of the APC Microbiome Institute , a research centre funded in part by Science Foundation Ireland (APC/SFI/ 12/RC/2273) and which is/has recently been in receipt of research grants from Abbvie, Alimentary Health, Cremo, Danone, Janssen, Friesland Campina, General Mills, Kerry, MeadJohnson, Nutricia, 4D pharma and Second Genome, Sigmoid pharma. Ethics approval Cork Clinical Research Ethics Committee. Provenance and peer review Not commissioned; externally peer reviewed. Data sharing statement In conformation of data accessibility protocol, metagenomic raw sequence data from this study are deposited in EMBL BNucleotide Sequence Database (ENA) (http://www.ebi.ac.uk/ena/data/), accession number PRJEB15388. REFERENCES 1 Harkin A. Muscling in on depression. N Engl J Med 2014;371:2333 – 4. 2 Benatti FB, Pedersen BK. Exercise as an anti-in fl ammatory therapy for rheumatic diseases-myokine regulation. Nat Rev Rheumatol 2015;11:86 – 97. 3 Hawley JA, Krook A. Metabolism: one step forward for exercise. Nat Rev Endocrinol 2016;12:7 – 8. 4 Hoffman-Goetz L, Pervaiz N, Packer N, et al . Freewheel training decreases pro- and increases anti-in fl ammatory cytokine expression in mouse intestinal lymphocytes. Brain Behav Immun 2010;24:1105 – 15. 5 Barton W, Shanahan F, Cotter PD, et al . The metabolic role of the microbiota. Clin Liver Dis 2015;5:91 – 3. 6 Szuhany KL, Bugatti M, Otto MW. A meta-analytic review of the effects of exercise on brain-derived neurotrophic factor. J Psychiatr Res 2015;60:56 – 64. 7 Ryan SM, Nolan YM. Neuroin fl ammation negatively affects adult hippocampal neurogenesis and cognition: can exercise compensate? Neurosci Biobehav Rev 2016;61:121 – 31. 8 Johannesson E, Simrén M, Strid H, et al . Physical activity improves symptoms in irritable bowel syndrome: a randomized controlled trial. Am J Gastroenterol 2011;106:915 – 22. 9 Robsahm TE, Aagnes B, Hjårtaker A, et al . Body mass index, physical activity, and colorectal cancer by anatomical subsites: a systematic review and meta-analysis of cohort studies. Eur J Cancer Prev 2013;22:492 – 505. 10 Schwingshackl L, Missbach B, Dias S, et al . Impact of different training modalities on glycaemic control and blood lipids in patients with type 2 diabetes: a systematic review and network meta-analysis. Diabetologia 2014;57:1789 – 97. 11 Biswas A, Oh PI, Faulkner GE, et al . Sedentary time and its association with risk for disease incidence, mortality, and hospitalization in adults: a systematic review and meta-analysis. Ann Intern Med 2015;162:123 – 32.
Metagenomic statistical and bioinformatic analysis Delivered raw FASTQ sequence fi les were quality checked as follows: contaminating sequences of human origin were fi rst removed through the NCBI Best Match Tagger (BMTagger). Poor-quality and duplicate read removal, as well as trimming was implemented using a combination of SAM (sequence align- ment map) and Picard tools. Processing of raw sequence data produced a total of 2 803 449 392 fi ltered reads with a mean read count of 32 598 248.74 (±10 639 447 SD) per each of the 86 samples. These re fi ned reads were then subjected to func- tional pro fi ling by the most recent iteration of the Human Microbiome Project Uni fi ed Metabolic Analysis Network (HUMAnN2 V.0.5.0) pipeline. 44 The functional pro fi ling per- formed by HUMAnN2 composed tabulated fi les of microbial metabolic pathway abundance and coverage derived from the Metacyc database. 45 Microbial pathway data were statistically analysed in the R software environment (V.3.2.2) (for further details see online supplementary methods) (R Development Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, 2012). 2015. http://www.R-project.org). All presented p values were corrected for multiple comparisons using the Benjamini-Hochberg false discovery rate (pFDR) method. 46 Metabolic pro fi ling Urine and faecal samples were prepared for metabonomic ana- lysis as previously described. 47 48 Using established methods, urine samples underwent 1 H-NMR, RP and HILIC chromatog- raphy pro fi ling experiments. Faecal samples underwent 1 H-NMR, HILIC and bile acid UPLC-MS pro fi ling experiments and GC-MS-targeted SCFA analysis. 48 – 50 After data preprocessing, 51 the resulting 1 H-NMR and LC-MS data sets were imported into SIMCA 14.1 (Umetrics) to conduct multivariate statistical analysis. PCA, followed by OPLS-DA, was performed to examine the data sets and to observe clustering in the results according to the prede fi ned classes. The OPLS-DA models in this study were established based on one PLS component and one orthogonal component. Unit variance scaling was applied to 1 H-NMR data, Pareto scaling was applied to MS data. The fi t and predictability of the models obtained were determined by the R 2 Y and Q 2 Y values, respectively. Signi fi cant metabolites were obtained from LC-MS OPLS-DA models through division of the regression coef fi cients by the jack-knife interval SE to give an estimate of the t-statistic. Variables with a t-statistic ≥ 1.96 (z-score, corresponding to the 97.5 percentile) were considered signi fi cant. Signi fi cant metabo- lites were obtained from 1 H-NMR OPLS-DA models after inves- tigating correlations with correlation coef fi cients values higher than 0.4. Univariate statistical analysis (Mann-Whitney U test) was used to examine the SCFA data set. p values were adjusted for multiple testing using the pFDR method. Con fi rmation of metabolite identities in the NMR data was obtained using 1D 1 HNMR and 2D 1 H- 1 HNMR and 1 H- 13 C NMR experiments. In addition, statistical tools such as SubseT Optimization by Reference Matching (STORM) and Statistical TOtal Correlation SpectroscopY (STOCSY) were also applied. 52 53 Con fi rmation of metabolites identities in the LC-MS data was obtained using tandem MS (MS/MS) on selected target ions. Metabolite identi fi cation was characterised by a level of assignment (LoA) score that describes how the identi fi cation was made. 54 The levels used were as follows: LoA 1: identi fi ed compound, con fi rmed by comparison to an authentic chemical
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
8
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
Gut microbiota
35 Hamer HM, Jonkers DM, Bast A, et al . Butyrate modulates oxidative stress in the colonic mucosa of healthy humans. Clin Nutr 2009;28:88 – 93. 36 den Besten G, van Eunen K, Groen AK, et al . The role of short-chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism. J Lipid Res 2013;54:2325 – 40. 37 Bennett BJ, de Aguiar Vallim TQ, Wang Z, et al . Trimethylamine-N-oxide, a metabolite associated with atherosclerosis, exhibits complex genetic and dietary regulation. Cell Metab 2013;17:49 – 60. 38 Holmes E, Loo RL, Stamler J, et al . Human metabolic phenotype diversity and its association with diet and blood pressure. Nature 2008;453:396 – 400. 39 Brancaccio P, Limongelli FM, Maffulli N. Monitoring of serum enzymes in sport. Br J Sports Med 2006;40:96 – 7. 40 Stratton SL, Bogusiewicz A, Mock MM, et al . Lymphocyte propionyl-CoA carboxylase and its activation by biotin are sensitive indicators of marginal biotin de fi ciency in humans. Am J Clin Nutr 2006;84:384 – 8. 41 Wilson GJ, Wilson JM, Manninen AH. Effects of beta-hydroxy-beta-methylbutyrate (HMB) on exercise performance and body composition across varying levels of age, sex, and training experience: a review. Nutr Metab (Lond) 2008;5:1. 42 Holmes E, Li JV, Athanasiou T, et al . Understanding the role of gut microbiome-host metabolic signal disruption in health and disease. Trends Microbiol 2011;19:349 – 59. 43 Wareham NJ, Jakes RW, Rennie KL, et al . Validity and repeatability of the EPIC-Norfolk Physical Activity Questionnaire. Int J Epidemiol 2002;31:168 – 74. 44 Abubucker S, Segata N, Goll J, et al . Metabolic reconstruction for metagenomic data and its application to the human microbiome. PLoS Comput Biol 2012;8:e1002358. 45 Caspi R, Altman T, Billington R, et al . The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of Pathway/Genome Databases. Nucleic Acids Res 2014;42:D459 – 71. 46 Benjamini Y, Hochberg Y. Controlling the false discovery rate — a practical and powerful approach to multiple testing. J Roy Stat Soc B Met 1995;57:289 – 300. 47 Dona AC, Jiménez B, Schäfer H, et al . Precision high-throughput proton NMR spectroscopy of human urine, serum, and plasma for large-scale metabolic phenotyping. Anal Chem 2014;86:9887 – 94. 48 García-Villalba R, Giménez-Bastida JA, García-Conesa MT, et al . Alternative method for gas chromatography-mass spectrometry analysis of short-chain fatty acids in faecal samples. J Sep Sci 2012;35:1906 – 13. 49 Want EJ, Wilson ID, Gika H, et al . Global metabolic pro fi ling procedures for urine using UPLC-MS. Nat Protoc 2010;5:1005 – 18. 50 Sara fi an MH, Lewis MR, Pechlivanis A, et al . Bile acid pro fi ling and quanti fi cation in bio fl uids using ultra-performance liquid chromatography tandem mass spectrometry. Anal Chem 2015;87:9662 – 70. 51 Veselkov KA, Vingara LK, Masson P, et al . Optimized preprocessing of ultra-performance liquid chromatography/mass spectrometry urinary metabolic pro fi les for improved information recovery. Anal Chem 2011;83:5864 – 72. 52 Posma JM, Garcia-Perez I, De Iorio M, et al . Subset optimization by reference matching (STORM): an optimized statistical approach for recovery of metabolic biomarker structural information from 1H NMR spectra of bio fl uids. Anal Chem 2012;84:10694 – 701. 53 Cloarec O, Dumas ME, Craig A, et al . Statistical total correlation spectroscopy: an exploratory approach for latent biomarker identi fi cation from metabolic 1H NMR data sets. Anal Chem 2005;77:1282 – 9. 54 Sumner LW, Amberg A, Barrett D, et al . Proposed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics 2007;3:211 – 21.
12 Same RV, Feldman DI, Shah N, et al . Relationship between sedentary behavior and cardiovascular risk. Curr Cardiol Rep 2016;18:6. 13 Wilmot EG, Edwardson CL, Achana FA, et al . Sedentary time in adults and the association with diabetes, cardiovascular disease and death: systematic review and meta-analysis. Diabetologia 2012;55:2895 – 905. 14 Chen YC, Tu YK, Huang KC, et al . Pathway from central obesity to childhood asthma. Physical fi tness and sedentary time are leading factors. Am J Respir Crit CareMed 2014;189:1194 – 203. 15 Koeth RA, Wang Z, Levison BS, et al . Intestinal microbiota metabolism of L-carnitine, a nutrient in red meat, promotes atherosclerosis. NatMed 2013;19:576 – 85. 16 Tang WHW, Wang ZE, Levison BS, et al . Intestinal microbial metabolism of phosphatidylcholine and cardiovascular risk. N Engl J Med 2013;368:1575 – 84. 17 Tang WHW, Hazen SL. Microbiome, trimethylamine N-oxide, and cardiometabolic disease. Transl Res 2017;179:108 – 15. 18 Woting A, Pfeiffer N, Loh G, et al . Clostridium ramosum promotes high-fat diet-induced obesity in gnotobiotic mouse models. MBio 2014;5:e01530 – 14. 19 Utzschneider KM, Kratz M, Damman CJ, et al . Mechanisms linking the gut microbiome and glucose metabolism. J Clin Endocrinol Metab 2016;101:1445 – 54. 20 Turnbaugh PJ, Ley RE, Mahowald MA, et al . An obesity-associated gut microbiome with increased capacity for energy harvest. Nature 2006;444:1027 – 31. 21 Williams NC, Johnson MA, Shaw DE, et al . A prebiotic galactooligosaccharide mixture reduces severity of hyperpnoea-induced bronchoconstriction and markers of airway in fl ammation. Br J Nutr 2016;116:798 – 804. 22 Wang Z, Klipfell E, Bennett BJ, et al . Gut fl ora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature 2011;472:57 – 63. 23 Zhernakova A, Kurilshikov A, Bonder MJ, et al . Population-based metagenomics analysis reveals markers for gut microbiome composition and diversity. Science 2016;352:565 – 9. 24 Cronin O, Molloy MG, Shanahan F. Exercise, fi tness, and the gut. Curr Opin Gastroenterol 2016;32:67 – 73. 25 Estaki M, Pither J, Baumeister P, et al . Cardiorespiratory fi tness as a predictor of intestinal microbial diversity and distinct metagenomic functions. Microbiome 2016;4:42. 26 Clarke SF, Murphy EF, O ’ Sullivan O, et al . Exercise and associated dietary extremes impact on gut microbial diversity. Gut 2014;63:1913 – 20. 27 O ’ Sullivan O, Cronin O, Clarke SF, et al . Exercise and the microbiota. Gut Microbes 2015;6:131 – 6. 28 Cronin O, O ’ Sullivan O, Barton W, et al . Gut microbiota: implications for sports and exercise medicine. Br J Sports Med. Published Online First 11 January 2017. 29 Rankin A, O ’ Donavon C, Madigan SM, et al . ‘ Microbes in sport ’— the potential role of the gut microbiota in athlete health and performance. Br J Sports Med . Published Online First 25 January 2017. 30 Petriz BA, Castro AP, Almeida JA, et al . Exercise induction of gut microbiota modi fi cations in obese, non-obese and hypertensive rats. BMC Genomics 2014;15:511. 31 Maffetone PB, Laursen PB. Athletes: fi t but unhealthy? Sports Med Open 2016;2:24. 32 Flint HJ, Scott KP, Duncan SH, et al . Microbial degradation of complex carbohydrates in the gut. Gut Microbes 2012;3:289 – 306. 33 Koh A, De Vadder F, Kovatcheva-Datchary P, et al . From dietary fi ber to host physiology: short-chain fatty acids as key bacterial metabolites. Cell 2016;165:1332 – 45. 34 Ridaura VK, Faith JJ, Rey FE, et al . Gut microbiota from twins discordant for obesity modulate metabolism in mice. Science 2013;341:1241214.
BartonW, et al . Gut 2017; 0 :1 – 9. doi:10.1136/gutjnl-2016-313627
9
group.bmj.com on March 30, 2017 - Published by http://gut.bmj.com/ Downloaded from
functional metabolic level in composition and particularly at the differs from that of more sedentary subjects The microbiome of professional athletes Orla O'Sullivan Michael G Molloy, Elaine Holmes, Fergus Shanahan, Paul D Cotter and Wiley Barton, Nicholas C Penney, Owen Cronin, Isabel Garcia-Perez,
published online March 30, 2017 Gut
http://gut.bmj.com/content/early/2017/03/29/gutjnl-2016-313627 Updated information and services can be found at:
These include:
References
#BIBL http://gut.bmj.com/content/early/2017/03/29/gutjnl-2016-313627 This article cites 52 articles, 8 of which you can access for free at:
service Email alerting
box at the top right corner of the online article. Receive free email alerts when new articles cite this article. Sign up in the
Notes
http://group.bmj.com/group/rights-licensing/permissions To request permissions go to:
http://journals.bmj.com/cgi/reprintform To order reprints go to:
http://group.bmj.com/subscribe/ To subscribe to BMJ go to:
Page 1 Page 2 Page 3 Page 4 Page 5 Page 6-7 Page 8 Page 9 Page 10 Page 11Powered by FlippingBook