A strategy for an association study of the intestinal microbiome and

Subscriber access provided by READING UNIV

Article

A strategy for association study on intestinal microbiome and brain metabolome across lifespan of rats Tianlu Chen, Yijun You, Guoxiang Xie, Xiaojiao Zheng, Aihua Zhao, Jiajian Liu, Qing Zhao, Shouli Wang, Huang Fengjie, Cynthia Rajani, Chongchong Wang, Shaoqiu Chen, Yan Ni, Herbert Yu, Youping Deng, Xiaoyan Wang, and Wei Jia Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b02859 • Publication Date (Web): 21 Jan 2018 Downloaded from http://pubs.acs.org on January 21, 2018

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.

Analytical Chemistry 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 23 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

Analytical Chemistry

A strategy for an association study of the intestinal microbiome and brain metabolome across the lifespan of rats Tianlu Chen1, Yijun You1, Guoxiang Xie2, Xiaojiao Zheng1, Aihua Zhao1, Jiajian Liu1, Qing Zhao1, Shouli Wang1, Fengjie Huang1, Cynthia Rajani2, Congcong Wang3, Shaoqiu Chen3, Yan Ni2, Herbert Yu2, Youping Deng4, Xiaoyan Wang3*, Wei Jia1,2* 1

Shanghai Key Laboratory of Diabetes Mellitus and Center for Translational Medicine,

Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai 200233, China. 2

University of Hawaii Cancer Center, Honolulu 96813, USA.

3

Ministry of Education Key Laboratory of Systems Biomedicine, Shanghai Center for

Systems Biomedicine, Shanghai Jiao Tong University, Shanghai 200240, China. 4

Biostatistics and Quantitative Health Sciences, John A. Burns School of Medicine,

University of Hawaii at Manoa, Honolulu 96813, USA.

Corresponding author: Wei Jia Phone: 1-808-564-5823 E-mail: [email protected]

1

ACS Paragon Plus Environment

Analytical Chemistry 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 There is increased appreciation for the diverse roles of the microbiome-gut-brain axis on mammalian growth and health throughout the lifespan. Numerous studies have demonstrated that the gut microbiome and their metabolites are extensively involved in the communication between brain and gut. Association study of brain metabolome and gut microbiome is an active field offering large amounts of information on the interaction of microbiome, brain and gut but data size and complicated hierarchical relationships were found to be major obstacles to the formation of significant, reproducible conclusions. This study addressed a two-level strategy of brain metabolome and gut microbiome association analysis of male Wistar rats in the process of growth, employing several analytical platforms and various bioinformatics methods. Trajectory analysis showed that the age-related brain metabolome and gut microbiome had similarity in overall alteration patterns. Four high taxonomical level correlated pairs of “metabolite type-bacterial phylum” including “lipids-Spirochaetes”, “free fatty acids (FFAs)-Firmicutes”, “bile acids (BAs)-Firmicutes” and “Neurotransmitters-Bacteroidetes” were screened out based on unit- and multi-variant correlation analysis and function analysis. Four groups of specific “metabolite-bacterium” association pairs from within the above high level key pairs were further identified. The key correlation pairs were validated by an independent animal study. This two-level strategy is effective in identifying principal correlations in big datasets obtained from the systematic multi-omics study, furthering our understanding on the lifelong connection between brain and gut.

Key words: microbiome-gut-brain axis, metabolome, microbiome, association

2


Page 2 of 23

Page 3 of 23 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


Introduction The microbiome-gut-brain axis integrates neural, hormonal and immunological signals between the gut and brain and has a profound effect on mammalian growth and health throughout life1. Currently, the systematic linkage of the two distant organs and the physiological and pathological impacts of the microbiome-gut-brain axis to growth and health remain largely unclear. Extensive research is required to validate the existing routes of bidirectional communication, to identify new connections, to illuminate the relevant molecular mechanism and to testify whether the results can be translated to the human body. In 2008, our group proposed a therapeutic strategy of manipulating the gut microbiome so that the dysregulation of the microbiota-host co-metabolism associated with a pathological state can be reversed2. Subsequently, we reported a series of urinary and fecal metabolites which were likely to change along with gut microbiota and set up a preliminary “metabolome-microbiome” linkage library3. The possible interaction routes of the host and its commensal microbes4 and the therapeutic targets along the routes5 were also reviewed6. Recently, our quantitative brain metabolome showed that many gut microbiota-related metabolites, including a series of bile acids which were detected in the brain for the first time and were significantly altered in rat brain in an age-related way7. These studies extended our understanding of the microbiome-gut-brain axis in that the metabolome, especially some newly found brain metabolites, may serve as new endpoints of routes connecting the gut and brain. Additionally, we suggested the interplay and joint effect of serum metabolome and gut microbiome on the gut-liver axis8,9 in both clinical and animal studies. It is well known that omics data has its own characteristics such as more variables than samples and complicated hierarchical relationships. Metabolome and microbiome are much different in distribution and percentage of zero values since metabolome is basicly a “spectral” information while microbiome is based on “count” measures. Here, we report a two-level association analysis strategy for the brain metabolome (354

3



metabolites from 14 metabolite types) and gut microbiome (2371 OTUs from 17 different phyla) measured in samples from 42 male Wistar rats at 7 time points (Week 1, 3, 7, 12, 24, 56 and 111). The high level “metabolite type-bacterial phylum” and the lower level “metabolite-bacterium” correlation pairs were identified by regularized canonical correlation analysis (rCCA), Spearman correlation analysis and linear discriminant analysis effect size (LEfSe). The key pairs were further validated by an independent diet restriction study on male rats (n=12) (Figure 1). The aim of the present study is to set up a strategy for an association study between metabolome and microbiome. We also aimed to enlarge the “metabolome-microbiome” linkage library with advanced analytical and data mining technologies that would provide more information on the interplay between gut and brain in order to aid in the identification of potential therapeutic markers and mechanistic solutions to complex problems commonly encountered in pathology, toxicology, and drug metabolism studies.

Figure 1. Study workflow. The brains and intestinal contents of 42 rats were collected at 7 postnatal week time points (a). Brain metabolome and gut microbiome were obtained using several platforms (b). Alteration, correlation and function analyses were performed, four principal correlated pairs were identified at phylum level (in red), and another four groups of correlated pairs were further identified at genus/species level (in blue). The “+” and “-” indicates positive and negative correlations respectively. (c) Design of an independent diet restriction animal experiment for key correlation pairs validation.

4


Page 4 of 23

Page 5 of 23 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


Materials and Methods Animal experiment and data acquisition For the normal growth experiment, the whole brain and intestinal content samples of 42 male rats at 7 time points (Week 1, 3, 7, 12, 24, 56 and 111 after birth, n=6 per group) were collected. Their body and brain weights were measured before sacrifice (Figure S1). For the diet restriction experiment, 12 male rats (4 weeks old) were fed ad libitum for 3 weeks and then 6 (selected randomly) were fed ad libitum and the other 6 were fed 60% of the ad libitum for 5 weeks. The food amount of the restricted diet group was calculated from those under normal growth conditions. At week 12, their whole brain and intestinal content samples were collected according to the same protocol as for the normal growth experiment. Please find detailed information of the animal experiment in the Supporting Information. All animal handling and experiments were performed strictly in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health. The experimental protocol was approved by the Center for Laboratory Animals, Shanghai Jiao Tong University, Shanghai, China. The brain metabolome measurement and pretreatment were based on the protocols established by our lab7. Metabolite profilings were acquired using UPLC/QTOF-MS (Waters, USA) (positive and negative modes) and GC/TOF-MS (Leco, USA) platforms and preprocessed by Progenesis QI (Waters, USA) and ChromaTOF (Leco, USA). The quantification of lipids, acylcarnitines, amino acids, biogenic amines and sugars were achieved using the AbsoluteIDQTM p180 Kit assay and MetIDQ (BIOCRATES, Austria). Free fatty acids and bile acids were measured by UPLC/QTOF-MS and UPLC/TQ-MS respectively and quantified using TargetLynx (Waters, USA). Gut microbiota 16S rRNA (V3) of all the intestinal content samples was amplified and measured using the

5



IlluminaMiseq platform. High quality sequencing data and predicted function data were obtained by QIIME (Version 1.8.0), MOTHUR (version 1.31.2) and PICRUst (http://picrust.github.io/picrust/). Finally, 354 metabolites of 14 metabolite types, 2371 OTUs under 17 phyla and 146 functions related to metabolism were reserved for subsequent analysis. Detailed measurement and data pretreatment methods are provided in the Supporting Information. Association analysis The age-related brain metabolome and gut microbiome alterations were evaluated first. Principle component analysis (PCA) on all the metabolites and principal coordinate analysis (PCoA) based on weighted and unweighted Unifrac OTU matrix were performed and trajectory plots were generated to depict the dynamic processes of metabolome and microbiome across 7 time points. The x- and y-axis of the 7 points in the trajectory plots were mean values of all samples at the same time point. The relative abundances of each bacterial phylum and each metabolite type, normalized by row across the 7 time points, and normalized by column within sample groups, are illustrated by bar plots. The concentrations of metabolite types were calculated by the summations of all metabolites (normalized by row, within each variable) under the corresponding types. The phyla were calculated by the summations of all the OTUs of the corresponding phyla. After that, a two-level strategy was conducted to find out key association pairs. For the high level association analysis, considering the size and complexity of omics data, correlations of metabolite types and bacterial phyla were examined. The key “metabolite type-bacterial phylum” pairs were selected based on the results of 1) rCCA (regularized canonical correlation analysis) and Spearman correlation analysis on the abundance datasets and 2) functional difference analysis (LEfSe) on the predicted metabolic function dataset derived from the bacterial data. For the low level association analysis, correlation networks were constructed and all the specific “metabolite-bacterium” association pairs

6


Page 6 of 23

Page 7 of 23 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


were listed. Those from within the above high level key pairs were selected for further analyses. Finally, the key pairs were validated by an independent animal study. The non-parametric Mann-Whitney U test and the Krukal-Walli test were applied for comparing the difference between two groups (for diet restriction study) or among multiple groups (for normal growth study) as over 90% of the variables were deviated from normality when examined by the Kolmogorov-Smirnov normality test. Data in figures were expressed as mean ± S.E. All statistical tests were two-sided and p < 0.05 was considered statistically significant. The p values were corrected (FDR=0.05) to control multiple hypothesis testing errors. All the analysis and graphics were performed using QIIME (1.7.0), Matlab (R2014a, MathWorks, USA), R (2.12), GraphPad Prism (6.0, GraphPad, USA) and Cytoscape (3.4.0). More descriptions for PCA, PCoA, rCCA, PICRUst and LEfSe and the metabolome and microbiome datasets are provided in the Supporting Information. The Illumina sequencing dataset was uploaded to NCBI (SRP119713: PRJNA413717 and SRP123384: PRJNA416228). Results and discussion The alterations of brain metabolome and gut microbiome across lifespan In the normal growth study, 14 metabolite types and 17 bacterial phyla were reserved for analysis (Figures 2a and 2d). Lipids and amino acids were the predominant types of metabolites detected (accounted for 42% of all the metabolites) while Firmicutes and Proteobacteria were the predominant bacterial phyla (51%) found in the gut. The patterns (relative positions of the 7 dots representing different time points) of metabolome and microbiome trajectories (Figure 2b and 2e) were similar, as evidenced by the significant shift from W1 to W3 and from W3 to W7. Similarly, the Shannon indexes indicated that (Figure S2) the gut microbiome diversity was higher in young (W1 and W3) than in adult rats (W7-W56), and the old ones (W111) had the lowest diversity. The metabolome and 7



gut microbiome alterations were further evaluated with regard to the relative abundance of metabolite types and bacterial phyla across 7 time points (Figure 2c and 2f) and within sample groups (Figure S3). Not surprisingly, both high and low abundance variables changed greatly. The Krukal-Wallis test showed that almost all the metabolite types (except phenols, peptides and purines) and all the bacterial phyla (except for Actinobacteria) were significantly changed among the 7 groups. Hence, obvious changes were observed in both brain metabolome and gut microbiota across lifespan, and there were similarities in their trajectory patterns.

Figure 2. Brain metabolome and gut microbiome composition and alterations across the lifespan in the normal growth study. 2a and 2d are brain metabolome and gut microbiome compositions. 2b and 2e are age-related trajectories of brain metabolome and gut microbiome based on the score values of PCA and unweighted UniFrac PCoA. The relative abundance of metabolite types (2c) and bacterial phyla (2f) across 7 age groups are illustrated by colored bars (normalized by row). * and ** indicates p

A strategy for an association study of the intestinal microbiome and

Recommend Documents