Journal of the Formosan Medical Association (2019) 118, S10eS22

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.jfma-online.com

Review Article

Metabolome analysis for investigating

host-gut microbiota interactions
Michael X. Chen a,b,1, San-Yuan Wang c,1, Ching-Hua Kuo d,e,f,
I-Lin Tsai c,g,h,i,*

a
Department of Laboratory Medicine and Pathology, The University of British Columbia, Canada
b
Island Medical Program, University of Victoria, Canada
c
Master Program in Clinical Pharmacogenomics and Pharmacoproteomics, College of Pharmacy, Taipei
Medical University, Taipei, Taiwan
d
School of Pharmacy, College of Medicine, National Taiwan University, Taipei, Taiwan
e
The Metabolomics Core Laboratory, NTU Centers of Genomic and Precision Medicine, National Taiwan
University, Taipei, Taiwan
f
Department of Pharmacy, National Taiwan University Hospital, Taipei, Taiwan
g
Department of Biochemistry and Molecular Cell Biology, School of Medicine, College of Medicine,
Taipei Medical University, Taipei, Taiwan
h
Graduate Institute of Medical Sciences, College of Medicine, Taipei Medical University, Taipei,
Taiwan
i
International PhD Program for Cell Therapy and Regeneration Medicine, College of Medicine, Taipei
Medical University, Taipei, Taiwan

Received 31 August 2018; accepted 5 September 2018

KEYWORDS Dysbiosis of the gut microbiome is associated with host health conditions. Many diseases have
Gut microbiota; shown to have correlations with imbalanced microbiota, including obesity, inflammatory bowel
Metabolomics; disease, cancer, and even neurodegeneration disorders. Metabolomics studies targeting small
Sample collection; molecule metabolites that impact the host metabolome and their biochemical functions have
Mass spectrometry; shown promise for studying host-gut microbiota interactions. Metabolome analysis determines
Data processing the metabolites being discussed for their biological implications in host-gut microbiota inter-
actions. To facilitate understanding the critical aspects of metabolome analysis, this article
reviewed (1) the sample types used in host-gut microbiome studies; (2) mass spectrometry
(MS)-based analytical methods and (3) useful tools for MS-based data processing/analysis. In
addition to the most frequently used sample type, feces, we also discussed others biosamples,
such as urine, plasma/serum, saliva, cerebrospinal fluid, exhaled breaths, and tissues, to bet-
ter understand gut metabolite systemic effects on the whole organism. Gas chromatography-
mass spectrometry (GCeMS), liquid chromatography-mass spectrometry (LC-MS), and capillary

* Corresponding author. Department of Biochemistry and Molecular Cell Biology, School of Medicine, College of Medicine, Taipei Medical
University, 250 Wuxing Street, Taipei, 11031, Taiwan.
E-mail address: isabel10@tmu.edu.tw (I.-L. Tsai).
1
These authors contributed equally.

https://doi.org/10.1016/j.jfma.2018.09.007
0929-6646/Copyright ª 2018, Formosan Medical Association. Published by Elsevier Taiwan LLC. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Metabolome and host-gut microbiota interactions S11

electrophoresis-mass spectrometry (CE-MS), three powerful tools that can be utilized to study
host-gut microbiota interactions, are included with examples of their applications. After ob-
taining big data from MS-based instruments, noise removal, peak detection, missing value
imputation, and data analysis are all important steps for acquiring valid results in host-gut mi-
crobiome research. The information provided in this review will help new researchers aiming
to join this field by providing a global view of the analytical aspects involved in gut microbiota-
related metabolomics studies.
Copyright ª 2018, Formosan Medical Association. Published by Elsevier Taiwan LLC. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).

Introduction alterations could potentially reduce or maintain blood

The metabolic capacity of the human body involves com-
plex interactions between the intestinal microbiota and
host cells. These interactions, which occur through
metabolite production, begin at birth.1 Small molecules,
such as vitamins, fatty acids, amino acid (AAs), and bile
acids, modulate host-gut metabolic homeostasis by binding
to specific host membranes or nuclear receptors.2,3 The gut
microbiota facilitates the development of the human im-
mune system, and the immune system in turn shapes the
composition of the microbiota.
This crosstalk between many different classes of small
molecules contributes to the evolutionary fitness of the
host driven by dynamic interactions including diet,4 life-
style,5 illness and antibiotics use.6 Recent publications have
revealed that a disruption in this balanced interaction may
result in conditions such as cardiometabolic disorders, al-
lergies,7 inflammatory bowel diseases,8 neuropsychiatric
diseases and cancer.9,10 These findings suggest that the gut
microbiome contributes to interindividual variability in all
aspects of disease situations and thus should be part of
precision medicine approaches.11,12
A great example of this would be the use of the gut
microbiome to predict personalized postprandial blood
glucose in response to specific diets, which has significant
interindividual variability.13 Elevated postprandial blood
glucose levels are a major risk factor for prediabetes and
type II diabetes. Therefore, maintaining normal blood
glucose levels through lifestyle modifications with dietary
recommendations is considered critical for preventing and
controlling this metabolic condition.14 However, despite
the importance of food intake, conventional methods use
meal carbohydrate contents15 to predict postprandial gly-
cemic response to specific diets. Calculations based on the
weighted sum of the constituents of a meal can have
questionable accuracy and could lead to overestimation of
the meal glycemic index by 22%e50%.15,16 A seminal pub-
lication by Gordon and colleagues showed associations be-
tween the gut microbiota and obesity.5 Other studies
subsequently revealed its association with glucose intoler-
ance, type II diabetes, hyperlipidemia, and insulin
resistance.17e20 This has led to accurate predictions of
postprandial blood glucose after complex, real-life meals
using a validated algorithm that integrated clinical and gut
microbiome features.13 Based on these findings, personal-
ized dietary interventions accompanied by gut microbiota
glucose levels in at-risk patients.
Although these studies have provided promising results
with clinical applications, broad applicability using
sequence-based (metagenomics) approaches, either alone
or in combination with predictive algorithms, requires
extensive independent validation. Nevertheless, meta-
genomics has been extensively used to characterize the
species and gene contents of the gut microbiome associ-
ated with human health.21e23 Therefore, insights into the
functional relationships in this symbiotic ecosystem remain
challenging and require understanding beyond the micro-
bial DNA contents.24 Obtaining a more comprehensive
analysis by identifying not only the microbial composition
but also the metabolites (metabolome) in prospective
conditions will be more insightful.
Metabolomics has emerged as a technique that focuses
on defining the functional status of hostemicrobial re-
lationships in biological specimens, such as urine, blood,
feces and tissues. Most studies on the gut microbiome aim
to explore disease-related metabolites or dysregulated
metabolic pathways (Fig. 1). This review highlights several
main aspects in metabolome analysis for studying host-gut
microbiome interactions, including sample collection,
analytical tools, data pretreatment and data analysis
(Fig. 2). In particular, we focused on mass spectrometry
(MS)-based platforms due to their ability to measure a wide
range of metabolites in a complex sample matrix. The
biological functions of host and gut microbiota co-
metabolites have been discussed and summarized in many
review articles25e28 and will not be included in this article.
Sample collection
Feces,29,30 urine,31 plasma/serum,32 saliva,33 exhaled
breaths,34 cerebrospinal fluid (CSF) [47], and tissues from
target organs29 are all potential biological samples for
investigating co-metabolism and host-gut microbiome in-
teractions using good experimental designs. Sample type
selection may be considered for several reasons, such as
sample availability, the study model and disease type. For
metabolomics studies, careful sample collection and proper
storage conditions play important roles and decrease study
result bias.35 After collecting samples at proper time
points, all samples should be maintained at low tempera-
ture during transfer. Liquid biological samples can be
separated into aliquots to decrease the number of
S12
Figure 1 The impact of gut microbiota in symbiosis and dysbiosis with host on the health. Metabolome reflects the interactions
between gut microbiota and host through small molecule metabolites.
freezeethaw cycles in future experiments. Stress-induced
metabolic changes, such as those promoted by catechol-
amine and hormones, should be considered when collecting
invasive samples such as plasma/serum, CSF and tissue.
Collection tools which contain detergent or polymers
should be avoided to prevent signal interference in mass
spectrometry analysis.36 Sample pretreatment processes
depend on the target metabolites, sample types and in-
struments used for analysis, but solvent extraction with
different aqueous and organic solvent percentages are
commonly used. Sample preparation methods for metab-
olomics studies have been comprehensively reviewed in
several articles and are not discussed here.37e39
The selection of sample types and sample handling
procedures determines the metabolite contents of the
study; we reviewed commonly used sample types, their
selected applications as well as special precautions in
sample handling. Table 1 summarizes the MS-based
analytical methods used to investigate host-gut micro-
biota; the information includes the selected sample type,
sample preparation procedure, analytical instrument and
studied metabolites.
Feces
Because fecal samples can directly contain host-gut co-
metabolites, it has undoubtedly became the most widely
used sample type for studying host-gut microbiota in-
teractions, such as in ulcerative colitis and irritable bowel
syndrome,40 obesity,41 liver cirrhosis,42 carcinoma,43 etc. A
comprehensive review of human fecal metabolomics has
been published by Karu et al. The article provides a
M.X. Chen et al.
comprehensive review of the current state of knowledge
with regard to the protocols, technologies and remaining
challenges in fecal metabolite analysis.44 Dr. Ravenzwaat
et al. analyzed both feces and gut tissues to investigate the
impact of different antibiotics on microbiome-related me-
tabolites. They found that gut tissues did not provide much
more information compared with feces, which indicated
that feces are a noninvasive and information-rich sample
type.29 Feces from animal models can be collected after
sacrifice or from cages. For clinical studies, special
collection tools have been developed for convenient feces
collection, even for individuals at home.45 Storing feces in a

80  C freezer directly after collection or by snap-freezing
in liquid nitrogen are both used in many studies. Before
freezing stool samples, a drying process is sometimes used
to make sure the weight of the samples used for prepara-
tion is not affected by fluid inside feces, which is caused by
individual variation; however, volatile metabolites might
lost during the drying process.39
Urine
Urine samples contain both host and microbial products
and information from urinary metabolomics could consider
both microbial and host influences on the phenotype.
Urine was selected as a biomatrix during the initiation of
metabolomics (metabonomics).46 Due to its special char-
acteristics and evidence that metabolites produced by
intestinal bacteria are absorbed and excreted in urine,47
urine has also been widely used for studying host-gut
microbiota interactions. The successful application of
urinary metabolomics has been used to study gut
Metabolome and host-gut microbiota interactions
Figure 2 Workflow of mass spectrometry-based metabolomics study for investigating host-gut microbiota interactions. A suc-
cessful metabolomics study starts with good experimental design, followed by careful sample collection and pretreatment. Sen-
sitive and selective MS instruments are useful in metabolite detection. After data acquisition, analytical results should be
preprocessed before data analysis.
microbiota-associated diseases including tuberculosis,48
autism,49 antibiotic-induced gut microbiota dysbio-
sis,47,50 and metabolic syndrome.51 When using urine as
the biomatrix for metabolomics studies, midstream urine
is collected and centrifugation (at > 1000e3000 g for
5 min at 4  C) is usually recommended to eliminate cells
and pellets. Preservatives such as sodium azide or boric
acid have been suggested as urine sample additives at final
concentrations of 200 mM and 10 mM, respectively, to
prevent bacterial growth.52 Additionally, maintaining the
samples at low temperatures (
80  C) for long-term
storage is necessary for metabolite stability. Prior to
instrumental analysis, high-speed centrifugation
(> 10,000 g for 15 min) is usually applied to remove small
particles. Because intersubject dilution factors up to 2
orders of magnitude may be present, concentration
normalization should be performed before urinary
metabolomics analysis using creatinine, osmolarity, and
even advanced normalization strategies.53,54
Plasma and serum
Plasma or serum samples are a less invasive biomatrix that
can reflect the dynamic changes of the metabolome of the
whole organism and have been used in most metabolomics
S13
studies.55 Siuzdak et al. used 4 different MS-based analyt-
ical methods to compare plasma metabolites from
conventional-raised and germ-free mice. The results indi-
cated a large effect of the gut microbiota on mammalian
blood metabolites, especially on amino acids, indole-
containing metabolites, and drug-like phase-II meta-
bolism.56 Not only limited to endogenous co-metabolites,
metabolomics approaches can also be applied to investi-
gate food catabolism or nutrition through the gut micro-
biome and the impact on host health. Jacobs et al.
investigated the catabolites from black tea in the gut
microbiota that might be related to lowering the risk of
cardiovascular diseases after long-term consumption of
black tea.57 The collection of plasma or serum and the
selection of blood collection tubes should be consistent
throughout the study. Heparinized tubes are suggested for
plasma collection because EDTA generates mass spec-
trometry signal interference.58 Elena-Herrmann et al.
carefully estimated the influences of pre-analytical pro-
cedures on the blood metabolome and found that the delay
and storage temperature between a blood draw and
centrifugation have more impact on the blood metabolome
than other factors.59 Immediately centrifuging and storing
samples at low temperatures is recommended for blood
samples after drawing.

Table 1 Selected MS-based metabolomics studies in host-gut microbiota interactions.
Disease or Treatment
Antibiotic related
intestinal
colonization
resistance47
Antibiotic induced gut
microbiota dysbiosis
and Chinese herbal
formula50
Colorectal cancer78
Neuroblastoma84
Thyroid carcinoma83
Chinese rhubarb
treatment82
Black tea57
Irritable bowel
syndrome80
Ulcerative colitis81
Hyperlimidemia and
treatment51
Celiac disease60
Tuberculosis48
Gut microbiota and host
metabolome56
Feces Other Extraction
specimen
 Urine Digest with beta-glucuronidase
and aryl sulfatase. No
extraction prior analysis
V Urine Feces: homogenization/
extracted with cold water and
cold methanol; Urine: dlute in
water (1:1)
 Serum UPLC-MS: water: MeOH: ACN
(1:2:7); GC-TOFMS: MeOH:
chloroform (3:1)
V Serum Feses: Extract with 50%
ACN þ SPE.; Serum:
acetonitrile
 Serum Methanol: chloroform (3:1)
V  Homogenization/
methanol:chloroform:water
(225:75:300)
 Plasma Acidified acetonitrile/drying
and reconstitution
V  Homogenization/methanol:
chloroform (3:1)
V  Acetonitrile:isopropyl
alcohol:water (3:3:2)
V Serum, Feces and tissue: chloroform/
Urine, Liver methanol/water (2:5:2);
Tissue Serum: cold methanol; Urine:
cold ethanol
 Saliva Solid phase microextraction/
head space sampling
 Urine Organic acid extraction: HCl,
ethylacetate, diethyl ether
 Plasma Cold methanol (4 volumes)/
drying and reconstitution
S14
Analytical Small molecule metabolites/metabolism
methods
HPLC-MS/MS Target metabolites: N-acetyltryptophan, indole-3
propionic acid, indoxyl sulfate, cinnamoylglycine,
hippuric acid, enterolactone, enterodiol, etc.
UHPLC-QTOFMS Glycine, serine, threonine metabolism, pantothenate
and CoA biosynthesis, nicotinate, nicotinamide
metabolism, bile acid metabolism
UPLC-QTOFMS; Tricarboxylic acid (TCA) cycle, urea cycle, glutamine,
GC-TOFMS fatty acids, and gut flora metabolism
HPLC-MS/MS Bile acids
GC-TOFMS Amino acid, lipid, glucose, vitamin metabolism, diet/gut
microbiota interaction
GC-TOFMS Indole derivatives, short chain fatty acid, bile acid,
phenyl derivatives, amino acids, poly amines, organic
acids
UHPLC-high Catabolites of black tea
resolution mass
spectrometry
GC-TOFMS Myristic acid, arachidic acid, octadecanol, N-acetyl-D-
galactosamine; 1,1-hexadecanol, phenylethylamine, 2-
furoic acid, etc.
GC-TOFMS Short chain fatty acids, xanthine, oleic acid, putrescine,
5-aminovaleric acid
GCeMS Pyruvic acid, serotonin, ketogenic and glycogenic amino
acid, pyridoxine, 4-pyridoxic acid, hypotaurine,
methionine, putrescine, etc.
GCeMS Volatile metabolites, ex: ethyl-acetate, nonanal, 2-
hexanone
GCXGC-TOFMS 3,5-dihydroxybenzoic acid, 3-(4-hydroxy-3-
methoxyphenyl)propionic acid
M.X. Chen et al.
HPLC- ESI (þ); Global metabolites
HPLC-ESI(
);
HPLC-APCI(þ);
GCeMS
Metabolome and host-gut microbiota interactions S15

Glucose, fructose, lactate, succinate, fumarate, malate,

Saliva
citrate, b-hydroxybutyric acid, b-aminoisobutyric acid,

In addition to gut microbiota, the oral microbiome also

Amino acid, polyamines, prostaglandin E2, SCFAs
showed significant impacts on host/gut metabolome and
symbiosis. Saliva is one of the most noninvasive sample
types that can be collected for oral metabolic in-
vestigations. Previous studies also showed that oral micro-
biome symbiosis is stable in healthy subjects, but disturbed
during the occurrence and progression of diseases.33
Angelis et al. investigated differences in the microbiome
in saliva between children with and without celiac disease
(CD) by metagenomics and metabolomics approaches.60
Trimethylamine-N-oxide

Organic volatile compounds and oral microbiome dysbiosis

Volatile metabolites

were discovered, meaning that CD affected the oral

metabolome and the composition of the microbiota.
Unstimulated saliva collection was used to prevent
stimulation-related metabolites. Eating, drinking, smoking,
and oral hygiene products were not allowed 1e2 h before
saliva collection. Subjects could rinse their mouths with
etc.

small volumes of water before spitting. Saliva was spit into

sterile tubes covered in ice. Centrifugation is usually used
after sample collection to remove particles, and the sam-
ples are aliquoted for storage in a
80  C freezer.61,62
UHPLC-MS/MS,
CE-TOFMS

Exhaled breath
GCeMS

GCeMS

HILIC

Exhaled breaths show advantages in their noninvasiveness

and have gained increasing attention in recent years. The
association between volatile organic compounds (VOC) in
Dulbecco’s Phosphate Buffered

exhaled breaths and the pathophysiology of fatty liver and

Solid phase microextraction/

associated conditions, including obesity, has been reviewed

Homogenization/Methanol,

by Solga.63 In addition to the fatty liver axis, other diseases

were also found to be associated with gut microbiome-
Acidified acetonitrile
head space sampling

related VOCs in patients, such as Crohn’s disease,34 in-

Saline three times
water chloroform

flammatory bowel disease,64 and gastric cancer.65 Leja

et al. carefully evaluated the reproducibility of VOC mea-
surements by GCeMS, which showed that VOCs are poten-
tially useful for the diagnosis of gastric cancer.65 Multi-
breath collection should be used for biomarker discovery,
and the time of collection and collection environment
should be specified. Exhaled breath samples are generally
collected using special air bags,34 canisters or gas-tight
syringe. Exhaled breath condensate containing relatively
Liver tissue

nonvolatile compounds is also a novel sample type for

investigating small molecule metabolites (SMMs) and
CSF

hostemicrobiota interactions.66 These samples should be



processed carefully because the target analytes are usually

volatile or display high volatility. Pre-concentration or
extraction using a solid-phase microextraction (SPME) or
thermal desorption tube is commonly used prior GCeMS



V

analysis.67
Intestinal microbiota and
colonic metabolites77

Alzheimer’s disease and

energy metabolism75

Cerebrospinal fluids
Gut microbiota and

Cerebrospinal fluid (CSF) is also a biomatrix commonly used

in metabolomics research, especially for neuro-
Dementia71

ndegenerative diseases.68,69 The brainegutemicrobiome

axis and neurontransmitters, such as serotonin and trypto-
Autism79

phan, have been confirmed to show bidirectional commu-

nication.70 CSF can be collected by puncture; however,
collecting CSF is relatively invasive compared with other
S16
biosamples. This might be one of the reasons that the
number of publications using CSF to investigate host-gut
microbiota interactions is relatively low. Dr. Zanotti et al.
reported that the gut microbial metabolite trimethylamine-
N-oxide can be detected from human cerebrospinal fluid
using UHPLC-MS/MS, and further studies are required to
evaluate its relationship with neuron disorders.71 Due to
importance of the brain-gut microbiome axis, we believe
that there will be more studies using CSF as a sample type.
Tissue
Although tissue sampling is invasive, tissue extracts are still
widely used in metabolomics studies for deeper in-
vestigations focused on metabolic changes in target organs
or tumors. The integration of metabolic information from
different biofluids and tissue extracts has been applied to
many diseases, such as cancers,72,73 metal toxicity in the
liver,74 energy metabolism,75 cardiovascular disease,76 etc.
Dr. Hou et al. investigated the liver metabolome between
specific pathogen-free and germ-free mice and found
distinguished energy metabolism profiles, which revealed
that the gut microbiome plays important roles in amino
acid, lipid and carbohydrate metabolism.75 Matsumoto
et al. investigated the impact of intestinal microbiota on
the intestinal luminal metabolome using CE-TOFMS. The
results indicated that the intestinal microbiome has great
impacts on the colonic luminal metabolome.77 For metab-
olomics analysis, tissues samples should be snap-frozen in
liquid nitrogen after collection and stored in a
80  C
freezer until preparation. The weight of tissues (mg levels)
used for metabolic extraction should be consistent and
sufficient to achieve instrumental detection.
Analytical methods
In most metabolomics and microbiota studies, MS is coupled
to separation methods such as gas chromatography, liquid
chromatography, and capillary electrophoresis to provide
higher peak capacities and to achieve qualitative and
quantitative analysis. High-resolution mass spectrometry,
such as time of flight mass spectrometry (TOFMS), is
powerful for global metabolite detection. Tandem mass
spectrometry method, such as triple quadrupole mass
spectrometry, is useful for target metabolome studies.
Many studies have also used multiple MS-based platforms to
increase the metabolite detection coverage.56,78
Gas chromatography-mass spectrometry
Gas chromatography coupled with mass spectrometry,
including electron impact-quadrupole (GC-EIMS) or time-of-
flight mass spectrometry (GC-TOF MS), has been applied to
determine metabolites with thermal stability and vola-
tility.79 Non-volatile metabolites can also be detected using
GCeMS with suitable chemical derivatization. Dr. Lu and
team investigated the effects of supplementing bacteria to
mice with irritable bowel syndrome (IBS); the feces samples
were processed with an organic solvent extraction method
followed by metabolite derivatization before GC-TOF MS
analysis.80 Their findings suggest that supplementing
M.X. Chen et al.
Clostridium butyricum to IBS mice may provide benefits by
modulating host metabolism. Dr. Michail et al. integrated
metagenomics and metabolomics approaches to investigate
the biological effects of fecal microbiota transplantation
(FMT) in pediatric patients with ulcerative colitis. Using GC-
TOF MS analysis, several metabolites, such as short chain
fatty acids, xanthine, oleic acid, putrescine, and 5-
aminovaleric acid, were found to be altered after FMT.81
Liu et al. analyzed metabolites involved in pathways such
as short chain fatty acid metabolism, bile acid metabolism,
and carbohydrate and amino acid metabolism.82 Using an
optimized GC-TOF MS method, changes in metabolites such
as indole and phenyl derivatives, polyamines and short
chain fatty acids were observed in rats with rhubarb
treatment. Luo et al. used GC-TOF MS to analyze metabo-
lites from serum samples in which amino acid, lipid,
glucose, vitamin metabolism and diet/gut microbiota in-
teractions were demonstrated to be correlated with thyroid
carcinoma.83 These examples showed that GCeMS is
powerful for host-gut microbiota studies.
Liquid chromatography-mass spectrometry
Liquid chromatography-mass spectrometry (LC-MS) is one of
the most commonly used analytical platforms for metab-
olomics studies. Different kinds of LC columns, which are
suitable for hydrophobic and hydrophilic metabolites, have
been developed as reverse- and normal-phase stationary
phases, respectively. In addition to the benefits of a wide
variety of stationary phases, smaller particle sizes (<2 mm)
in separation columns coupled with ultra-high pressure
liquid chromatography instruments provide advantages of
higher separation efficiency and peak capacity. Processed
samples from solvent extraction are usually able to be
introduced into an LC column directly with simple drying
and reconstitution using suitable solvents. Compatibility
with a variety of metabolites and the good performance of
LC-MS make it a powerful tool in metabolomics and gut-
microbiota studies, such as those focused on colorectal
cancer78 and neuroblastoma.84 Drs. Fischbach and Son-
nenburg et al. integrated genetics and targeted metabolic
profiling results to reveal that human gut bacteria generate
aromatic amino acid metabolites that accumulate in the
host circulation and affect intestinal permeability and
systemic immunity.32 Not limited to the circulation system,
Dr. Spacil et al. developed a UHPLC-MS/MS method to
investigate tryptophan metabolites in urine, which are
related to gut microbiome metabolism.31 Dr. Tarr et al. also
used broad-range and targeted metabolomics approaches
focused on ceramides and sphingomyelins to investigate the
impacts of sphingomyelin metabolism on necrotizing
enterocolitis in infants.85
Capillary electrophoresis-mass spectrometry and
other MS-based analytical methods
Capillary electrophoresis (CE) provides a different separa-
tion mechanism compared with general chromatographic
methods, in which metabolites with different molecular
weights and charges can be separated in a capillary based
on their different mobility. By coupling with a mass
Metabolome and host-gut microbiota interactions
spectrometry detector, such as in the case of time-of-flight
mass spectrometry (TOF MS), CE-MS is a powerful analytical
platform that has been applied to gut microbiota
studies.77,86 The use of CE for phenotyping the gut micro-
biota has been extensively reviewed by Ferrer et al.86 Dr.
Uebanso and team utilized CE-TOF MS to analyze luminal
metabolites (110 targets) and integrated the results with
bacterial compositions to investigate the effects of xylitol
on the gut microbiome and lipid metabolism.30 Dr. Abe
et al. also used CE-TOF MS to analyze mouse metabolites
from plasma, urine, and feces to investigate the impact of
the gut microbiota on uremic solute accumulation and
renoprotective effects.87
Many advanced MS-based analytical platforms also show
good performance in small molecule metabolite analysis.
Matrix-assisted laser desorption ionization mass spectrom-
etry (MALDI-MS), desorption electrospray ionization mass
spectrometry (DESI-MS), and MassSpec Pen have been
applied to metabolite investigations using biological sam-
ples,88 but there are fewer studies focused on host-gut
microbiome interactions using these platforms compared
with GCeMS, LC-MS, CE-MS. This might be because most of
the biological samples use solvent extraction processes to
recover SMMs and the separation of multi-analytes is usually
helpful for decreasing the complexity before MS detection.
Nevertheless, these advanced MS-based analytical plat-
forms have the potential to be applied to gut microbiome
studies.
Data preprocessing for MS-based small
molecule detection and analysis
MS-based metabolomics is a powerful tool to detect a broad
range of small molecules in complex biological samples.
When using mass spectrometry to study host-gut microbiota
interactions, huge volumes of generated data may become
the bottleneck of entire studies. Many sophisticated tools
have been designed to facilitate the processing of metab-
olomics data. Tools and resources for handling NMR and MS
metabolomics data have been reviewed by Misra.89 A
recent review article found that the application of LC-MS in
fecal metabolomics is less common than GCeMS. The
article indicated that the overwhelming amount of data,
which requires knowledgeable and complex data process-
ing, is one of the reasons that restricted the application of
LC-MS in fecal metabolomics studies.44 Therefore, in the
following paragraphs, we reviewed and examined recently
published tools for MS-based metabolomics studies, with
more emphasis on LC-MS-based experiments.
The raw data generated by MS-based metabolomics ex-
periments comprises a large collection of 3-tuples (rt, m/z,
and a), where rt is the retention time or time required for
the molecule to elute from the column, m/z is the mass-to-
charge ratio of the molecular ion, and a is the measured ion
abundance of the molecular ion. However, detecting the
ions produced by small molecules in biological samples is
still a major challenge for untargeted studies, as the ma-
jority of ions are background or not products derived from
the small molecules of interest.90 The raw data requires
complex preprocessing steps to obtain at the clean data for
the subsequent statistical analysis. The major steps in the
S17
data preprocessing pipeline for MS-based small molecule
detection involves noise removal, peak detection, and
missing value imputation (Fig. 3). Small molecule identifi-
cation and data analysis follow data preprocessing. Several
databases provide metabolite spectral information to help
to identify the detected features. These databases include
the Golm Metabolome Database91 for GCeMS data, Madison
Metabolomics Consortium Database (MMCD),92 MassBank,93
and Metlin94 for LC-MS-based data. Furthermore, the
Human Metabolome Database (HMDB)95 contains both LC-
MS and GCeMS data for metabolite identification.
Noise removal
A vast amount of background ions and noise is generated
during the electrospray ionization process in GCeMS or LC-
MS experiments. To remove the background ions, several
algorithms have been developed using the R language,
including TIPick96 and BgS-NoRA.97 Both of these algorithms
remove background ions by subtracting a blank sample with
user-specified retention time shifts and relative mass-to-
charge ratio difference tolerances. BgS-NoRA also
removes ions that are not consistent across user-specified
consecutive adjacent LC-MS experiment scans. Several
noise filtering algorithms, include sliding window, average,
median, and Savitsky-Golay98 filters, are also used to
remove the noise from both GCeMS and LC-MS data.
Peak detection
Peak detection in LC-MS-based data preprocessing is
perhaps the most critical step.99,100 Peak detection consists
of pure ion chromatogram extraction and chromatographic
peak detection. Ion trace detection is a fundamental step
for LC-MS data analysis. A straightforward approach for
extracting the ion chromatograms is the binning method.
However, mass-to-charge ratio drift can occur in LC-MS
experiments. When the measured mass-to-charge ratio of
the analyte is close to the boundary between two bins, the
analyte ions can be assigned to different bins and generate
split peaks.90,101 Hence, several tools, including PITracer,90
XCMS,102 OpenMS,103 MZmine,104 and MetSign,105 have
developed algorithms that can analyze the ions in different
scans, collect the ions with similar mass-to-charge ratios,
and generate the ion chromatogram. After extracting the
ion chromatograms, the chromatographic peaks can be
detected according to several criteria including the signal-
to-noise ratio, intensity threshold, slope of the peak, local
maximum, shape ratio, length of ridge lines, and peak
width. A detailed comparison of state of the art peak
detection is reported by Yang et al.106 Yang et al. reported
that the continuous wavelet transform (CWT) algorithm
provides the best chromatographic peak detection perfor-
mance. The CWT algorithm is integrated into XCMS,
OpenMS, and MZmine. Several new algorithms for MS-based
data preprocessing are continuously incorporated into the
XCMS, OpenMS, and MZmine frameworks. These three tools
provide rich functionality and are widely used for studying
host-gut microbiota interactions; for example, XCMS was
used for peak picking in a human intestinal transplant
rejection study.107 In the study, the XCMS-derived data
S18
Figure 3 The major steps of the data preprocessing workflow of the MS-based small molecules detection.
obtained from the LC-MS analysis of stool extract was
directly used for multivariate data analysis without further
data preprocessing. OpenMS was also used to detect the
metabolic features of fecal pellets from mice with Chagas
disease to characterize the impacts of Trypanosoma cruzi
infection on the gut microbiota and corresponding metab-
olome.108 Hintze et al. employed MZmine to process LC-MS
data and analyze the metabolome to create humanized
animal models for animal health and disease research via
antibiotic treatment and fecal transfer.109
For GCeMS data, AMDIS (https://chemdata.nist.gov/
mass-spc/amdis) utilizes common mass spectrometric
data processing algorithms to process tasks in GCeMS
data and generates the clean data for the subsequent sta-
tistical analysis. These algorithms enable users to identify
and quantify small molecules with ease.
Missing value imputation
In MS-based small molecule detection studies, datasets can
typically contain 20e30% missing values.110 Generally, the
missing values can occur for the following three reasons: 1)
an inaccurate peak detection algorithm, 2) the
M.X. Chen et al.
concentration of the molecule in the sample is lower than
the limit of detection, and 3) the molecule is not present in
the sample. The missing values can result in a failure to
identify the wrong arguments and judgments. However, a
number of MS-based metabolomics studies do not mention
their missing value handling solutions.
When the missing values are caused by inaccurate peak
detection algorithms, the missing values can be imputed
by detecting peaks and reextracting the ion chromatogram
with adapted parameters or replacing baseline signals.111
The missing values can also be reduced by combining
duplicate measurements.112 Missing values due to the limit
of detection are usually replaced with half of the mini-
mum value from other samples. Bijlsma et al. proposed
the “80% rule” and recommended that molecules with at
least 20% missing values in a sample group should be
removed.112
In addition to replacing missing values with a small value
or a baseline signal, k-nearest neighbors imputation,
random forest imputation, and Bayesian principal compo-
nent analysis replacement are also used to handle missing
values in MS-based small molecule detection and quantifi-
cation datasets.113
Metabolome and host-gut microbiota interactions S19

MS-based metabolomics techniques are powerful for

Table 2 Main data analysis functions of MetaboAnalyst
investigating key metabolites associated with host-gut
and XCMS Online.
microbiota interactions. Due to the complexity of the
Functions MetaboAnalyst XCMS microbiome and the categories of co-metabolites, multiple
Online sample types, multi-analytical platforms with correct data
Univariate statistical analysis processing, and even multi-omics should be integrated to
t-test V V have a global view for host-microbiome studies. We have
ManneWhitney V reviewed critical aspects in metabolome analysis and pro-
Wilcoxon signed-rank V vided successful examples of using these approaches for
KruskaleWallis V studying host-gut microbiota interactions. The metab-
ANOVA V V olomics technique is definitely a promising approach that
Fold change V V facilitate and explore the biological networks between the
Multivariate statistical analysis host and gut microbiota.
Clustering V V
Classification V
Correlation analysis V
Conflicts of interest
Functional analysis
Pathway analysis V V The authors declare that they have no conflicts of interest.
Enrichment V
Data integration
Joint pathway analysis V V
Acknowledgements
Biomarker meta-analysis V V
Knowledge-based V This study was supported by the Ministry of Science and
network analysis Technology, Taiwan (106-2320-B-038-050-MY2), and the
Program for Translational Innovation of Biopharmaceutical
Development - Technology Supporting Platform Axis,
Taiwan (Grant No. 106-0210-01-10-01; 107-0210-01-19-04)
Data analysis for funding support.

After data preprocessing, the processed data is used to

identify critical molecules using mathematical and statis-
tical treatments. MetaboAnalyst114 and XCMS Online115 are
two web-based platforms that are widely used to handle MS
data. They not only provide data preprocessing function-
alities but also provide exploratory statistical analysis and
functional enrichment analysis. They provide fold-change
analysis, statistical tests, and Volcano plots to identify
significant metabolites or potential biomarkers. Further-
more, MetaboAnalyst utilizes a heatmap to present the
correlation between two features. In addition to univariate
analysis, the unsupervised clustering of multivariate data,
such as principal component analysis (PCA), can be used to
present the sample distribution, and supervised classifica-
tion methods, such as partial least squares discriminant
analysis (PLS-DA), can be employed to identify molecules
related to user-specified conditions. PLS-DA is also applied
for biomarker analysis based on receiver operating char-
acteristic curves. The pathway analysis module developed
by MetaboAnalyst can be used to visualize 21 model or-
ganisms including human, mouse, cow, chicken, zebrafish,
Escherichia coli, and others. The main functions of
MetaboAnalyst and XCMS Online for data analysis are listed
in Table 2.
Conclusions
Host-gut microbiota interactions have been shown to have
great impacts on host biological functions, health states,
and disease progression. The host and gut microbiota in-
fluence each other through symbiosis or dysbiosis condi-
tions via small molecule metabolites and co-metabolites.
References
1. Sprockett D, Fukami T, Relman DA. Role of priority effects
in the early-life assembly of the gut microbiota. Nat Rev
Gastroenterol Hepatol 2018;15:197e205.
2. Putignani L, Del Chierico F, Vernocchi P, Cicala M,
Cucchiara S, Dallapiccola B. Gut microbiota dysbiosis as risk
and premorbid factors of IBD and IBS along the childhood-
adulthood transition. Inflamm Bowel Dis 2016;22:487e504.
3. Husted AS, Trauelsen M, Rudenko O, Hjorth SA, Schwartz TW.
GPCR-mediated signaling of metabolites. Cell Metabol 2017;
25:777e96.
4. Turnbaugh PJ, Backhed F, Fulton L, Gordon JI. Diet-induced
obesity is linked to marked but reversible alterations in the
mouse distal gut microbiome. Cell Host Microbe 2008;3:
213e23.
5. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER,
Gordon JI. An obesity-associated gut microbiome with
increased capacity for energy harvest. Nature 2006;444:
1027e31.
6. Jernberg C, Lofmark S, Edlund C, Jansson JK. Long-term im-
pacts of antibiotic exposure on the human intestinal micro-
biota. Microbiology (Read) 2010;156:3216e23.
7. Noverr MC, Noggle RM, Toews GB, Huffnagle GB. Role of an-
tibiotics and fungal microbiota in driving pulmonary allergic
responses. Infect Immun 2004;72:4996e5003.
8. Frank DN, St Amand AL, Feldman RA, Boedeker EC, Harpaz N,
Pace NR. Molecular-phylogenetic characterization of micro-
bial community imbalances in human inflammatory bowel
diseases. Proc Natl Acad Sci USA 2007;104:13780e5.
9. Holmes E, Loo RL, Stamler J, Bictash M, Yap IKS, Chan Q, et al.
Human metabolic phenotype diversity and its association with
diet and blood pressure. Nature 2008;453:396e400.
10. Routy B, Gopalakrishnan V, Daillere R, Zitvogel L, Wargo JA,
Kroemer G. The gut microbiota influences anticancer
S20
immunosurveillance and general health. Nat Rev Clin Oncol
2018;15:382e96.
11. Kashyap PC, Chia N, Nelson H, Segal E, Elinav E. Microbiome
at the frontier of personalized medicine. Mayo Clin Proc 2017;
92:1855e64.
12. Suez J, Elinav E. The path towards microbiome-based
metabolite treatment. Nat Microbiol 2017;2:17075.
13. Zeevi D, Korem T, Zmora N, Israeli D, Rothschild D,
Weinberger A, et al. Personalized nutrition by prediction of
glycemic responses. Cell 2015;163:1079e94.
14. Riccardi G, Rivellese AA. Dietary treatment of the metabolic
syndrome–the optimal diet. Br J Nutr 2000;83(Suppl. 1):
S143e8.
15. Association AD. (4) Foundations of care: education, nutrition,
physical activity, smoking cessation, psychosocial care, and
immunization. Diabetes Care 2015;38(Suppl):S20e30.
16. Dodd H, Williams S, Brown R, Venn B. Calculating meal gly-
cemic index by using measured and published food values
compared with directly measured meal glycemic index. Am J
Clin Nutr 2011;94:992e6.
17. Le Chatelier E, Nielsen T, Qin J, Prifti E, Hildebrand F,
Falony G, et al. Richness of human gut microbiome correlates
with metabolic markers. Nature 2013;500:541.
18. Karlsson FH, Tremaroli V, Nookaew I, Bergström G, Behre CJ,
Fagerberg B, et al. Gut metagenome in European women with
normal, impaired and diabetic glucose control. Nature 2013;
498:99.
19. Qin J, Li Y, Cai Z, Li S, Zhu J, Zhang F, et al. A metagenome-
wide association study of gut microbiota in type 2 diabetes.
Nature 2012;490:55.
20. Suez J, Korem T, Zeevi D, Zilberman-Schapira G, Thaiss CA,
Maza O, et al. Artificial sweeteners induce glucose intoler-
ance by altering the gut microbiota. Nature 2014;514:181.
21. Qin J, Li R, Raes J, Arumugam M, Burgdorf KS, Manichanh C,
et al. A human gut microbial gene catalogue established by
metagenomic sequencing. Nature 2010;464:59e65.
22. Human Microbiome Project Consortium. Structure, function
and diversity of the healthy human microbiome. Nature 2012;
486:207.
23. Arumugam M, Raes J, Pelletier E, Paslier D, Yamada T,
Mende DR, et al. Enterotypes of the human gut microbiome.
Nature 2011;473:174.
24. Harrigan G, Goodacre R. Metabolic profiling: its role in
biomarker discovery and gene function analysis. 2002.
25. Luan H, Wang X, Cai Z. Mass spectrometry-based metab-
olomics: targeting the crosstalk between gut microbiota and
brain in neurodegenerative disorders. Mass Spectrom Rev
2017:1e12.
26. Nicholson JK, Holmes E, Kinross J, Burcelin R, Gibson G, Jia W,
et al. Host-gut microbiota metabolic interactions. Science
2012;336:1262e7.
27. Vernocchi P, Del Chierico F, Putignani L. Gut microbiota
profiling: metabolomics based approach to unravel com-
pounds affecting human health. Front Microbiol 2016;7:1144.
28. Usami M, Miyoshi M, Yamashita H. Gut microbiota and host
metabolism in liver cirrhosis. World J Gastroenterol 2015;21:
11597e608.
29. Behr C, Sperber S, Jiang X, Strauss V, Kamp H, Walk T, et al.
Microbiome-related metabolite changes in gut tissue, cecum
content and feces of rats treated with antibiotics. Toxicol
Appl Pharmacol 2018;355:198e210.
30. Uebanso T, Kano S, Yoshimoto A, Naito C, Shimohata T,
Mawatari K, et al. Effects of consuming xylitol on gut micro-
biota and lipid metabolism in mice. Nutrients 2017;9:756.
31. Pavlova T, Vidova V, Bienertova-Vasku J, Janku P, Almasi M,
Klanova J, et al. Urinary intermediates of tryptophan as in-
dicators of the gut microbial metabolism. Anal Chim Acta
2017;987:72e80.
M.X. Chen et al.
32. Dodd D, Spitzer MH, Van Treuren W, Merrill BD,
Hryckowian AJ, Higginbottom SK, et al. A gut bacterial
pathway metabolizes aromatic amino acids into nine circu-
lating metabolites. Nature 2017;551:648e52.
33. Kodukula K, Faller DV, Harpp DN, Kanara I, Pernokas J,
Pernokas M, et al. Gut microbiota and salivary diagnostics:
the mouth is salivating to tell us something. BioResearch Open
Access 2017;6:123e32.
34. Smolinska A, Tedjo DI, Blanchet L, Bodelier A, Pierik MJ,
Masclee AAM, et al. Volatile metabolites in breath strongly
correlate with gut microbiome in CD patients. Anal Chim Acta
2018;1025:1e11.
35. Yuan M, Breitkopf SB, Yang X, Asara JM. A positive/negative
ion-switching, targeted mass spectrometry-based metab-
olomics platform for bodily fluids, cells, and fresh and fixed
tissue. Nat Protoc 2012;7:872e81.
36. Trivedi DK, Iles RK. Do not just do it, do it right: urinary
metabolomics–establishing clinically relevant baselines. Bio-
med Chromatogr : BMC 2014;28:1491e501.
37. Drouin N, Rudaz S, Schappler J. Sample preparation for polar
metabolites in bioanalysis. Analyst 2018;143:16e20.
38. Chen YH, Xu J, Zhang RP, Abliz Z. Methods used to increase
the comprehensive coverage of urinary and plasma metab-
olomes by MS. Bioanalysis 2016;8:981e97.
39. Deda O, Gika HG, Wilson ID, Theodoridis GA. An overview of
fecal sample preparation for global metabolic profiling. J
Pharmaceut Biomed Anal 2015;113:137e50.
40. Le Gall G, Noor SO, Ridgway K, Scovell L, Jamieson C,
Johnson IT, et al. Metabolomics of fecal extracts detects altered
metabolic activity of gut microbiota in ulcerative colitis and
irritable bowel syndrome. J Proteome Res 2011;10:4208e18.
41. Paul HA, Bomhof MR, Vogel HJ, Reimer RA. Diet-induced
changes in maternal gut microbiota and metabolomic profiles
influence programming of offspring obesity risk in rats. Sci
Rep 2016;6:20683.
42. Huang HJ, Zhang AY, Cao HC, Lu HF, Wang BH, Xie Q, et al.
Metabolomic analyses of faeces reveals malabsorption in
cirrhotic patients. Dig Liver Dis : Off J Ital Soc Gastroenterol
Ital Assoc Stud Liver 2013;45:677e82.
43. Xie G, Wang X, Huang F, Zhao A, Chen W, Yan J, et al. Dys-
regulated hepatic bile acids collaboratively promote liver
carcinogenesis. Int J Cancer 2016;139:1764e75.
44. Karu N, Deng L, Slae M, Guo AC, Sajed T, Huynh H, et al. A
review on human fecal metabolomics: methods, applications
and the human fecal metabolome database. Anal Chim Acta
2018;1030:1e24.
45. Wu WK, Chen CC, Panyod S, Chen RA, Wu MS, Sheen LY,
et al. Optimization of fecal sample processing for
microbiome study e the journey from bathroom to bench.
J Formos Med Assoc 2019;118:545e55. https://doi.org/
10.1016/j.jfma.2018.02.005.
46. Nicholson JK, Lindon JC, Holmes E. ’Metabonomics’: Under-
standing the metabolic responses of living systems to patho-
physiological stimuli via multivariate statistical analysis of
biological NMR spectroscopic data. Xenobiotica 1999;29:
1181e9.
47. Obrenovich ME, Tima M, Polinkovsky A, Zhang R,
Emancipator SN, Donskey CJ. Targeted metabolomics analysis
identifies intestinal microbiota-derived urinary biomarkers of
colonization resistance in antibiotic-treated mice. Antimicrob
Agents Chemother 2017:61.
48. Luies L, van Reenen M, Ronacher K, Walzl G, Loots DT. Pre-
dicting tuberculosis treatment outcome using metabolomics.
Biomarkers Med 2017;11:1057e67.
49. Lussu M, Noto A, Masili A, Rinaldi AC, Dessi A, De Angelis M,
et al. The urinary H-1-NMR metabolomics profile of an Italian
autistic children population and their unaffected siblings.
Autism Res 2017;10:1058e66.
Metabolome and host-gut microbiota interactions
50. Yu M, Jia HM, Zhou C, Yang Y, Sun LL, Zou ZM. Urinary and
fecal metabonomics study of the protective effect of chaihu-
shu-gan-san on antibiotic-induced gut microbiota dysbiosis in
rats. Sci Rep 2017;7:46551.
51. Li M, Shu X, Xu H, Zhang C, Yang L, Zhang L, et al. Integrative
analysis of metabolome and gut microbiota in diet-induced
hyperlipidemic rats treated with berberine compounds. J
Transl Med 2016;14:237.
52. Roux A, Thevenot EA, Seguin F, Olivier MF, Junot C. Impact of
collection conditions on the metabolite content of human
urine samples as analyzed by liquid chromatography coupled
to mass spectrometry and nuclear magnetic resonance spec-
troscopy. Metabolomics : Off J Metabolomic Soc 2015;11:
1095e105.
53. Wagner S, Scholz K, Sieber M, Kellert M, Voelkel W. Tools in
metabonomics: an integrated validation approach for LC-MS
metabolic profiling of mercapturic acids in human urine.
Anal Chem 2007;79:2918e26.
54. Gagnebin Y, Tonoli D, Lescuyer P, Ponte B, de Seigneux S,
Martin PY, et al. Metabolomic analysis of urine samples by
UHPLC-QTOF-MS: impact of normalization strategies. Anal
Chim Acta 2017;955:27e35.
55. Dunn WB, Broadhurst D, Begley P, Zelena E, Francis-
McIntyre S, Anderson N, et al. Procedures for large-scale
metabolic profiling of serum and plasma using gas chroma-
tography and liquid chromatography coupled to mass spec-
trometry. Nat Protoc 2011;6:1060e83.
56. Wikoff WR, Anfora AT, Liu J, Schultz PG, Lesley SA, Peters EC,
et al. Metabolomics analysis reveals large effects of gut
microflora on mammalian blood metabolites. Proc Natl Acad
Sci USA 2009;106:3698e703.
57. van Duynhoven J, van der Hooft JJ, van Dorsten FA, Peters S,
Foltz M, Gomez-Roldan V, et al. Rapid and sustained systemic
circulation of conjugated gut microbial catabolites after
single-dose black tea extract consumption. J Proteome Res
2014;13:2668e78.
58. Barri T, Dragsted LO. UPLC-ESI-QTOF/MS and multivariate
data analysis for blood plasma and serum metabolomics: ef-
fect of experimental artefacts and anticoagulant. Anal Chim
Acta 2013;768:118e28.
59. Jobard E, Tredan O, Postoly D, Andre F, Martin AL, Elena-
Herrmann B, et al. A systematic evaluation of blood serum
and plasma pre-analytics for metabolomics cohort studies. Int
J Mol Sci 2016;17.
60. Francavilla R, Ercolini D, Piccolo M, Vannini L, Siragusa S, De
Filippis F, et al. Salivary microbiota and metabolome associ-
ated with celiac disease. Appl Environ Microbiol 2014;80:
3416e25.
61. Sugimoto M, Wong DT, Hirayama A, Soga T, Tomita M. Capillary
electrophoresis mass spectrometry-based saliva metabolomics
identified oral, breast and pancreatic cancer-specific profiles.
Metabolomics : Off J Metabolomic Soc 2010;6:78e95.
62. Dame ZT, Aziat F, Mandal R, Krishnamurthy R, Bouatra S,
Borzouie S, et al. The human saliva metabolome. Metab-
olomics : Off J Metabolomic Soc 2015;11:1864e83.
63. Solga SF. Breath volatile organic compounds for the gut-fatty
liver axis: promise, peril, and path forward. World J Gastro-
enterol 2014;20:9017e25.
64. Monasta L, Pierobon C, Princivalle A, Martelossi S, Marcuzzi A,
Pasini F, et al. Inflammatory bowel disease and patterns of
volatile organic compounds in the exhaled breath of children:
a case-control study using Ion Molecule Reaction-Mass Spec-
trometry. PLoS One 2017;12.
65. Leja M, Amal H, Lasina I, Skapars R, Sivins A, Ancans G, et al.
Analysis of the effects of microbiome-related confounding
factors on the reproducibility of the volatolomic test. J
Breath Res 2016;10:037101.
S21
66. Glendinning L, Wright S, Tennant P, Gill AC, Collie D,
McLachlan G. Microbiota in exhaled breath condensate and
the lung. Appl Environ Microbiol 2017;83(12):e00515e7.
67. Pereira J, Porto-Figueira P, Cavaco C, Taunk K, Rapole S,
Dhakne R, et al. Breath analysis as a potential and non-
invasive frontier in disease diagnosis: an overview. Metabo-
lites 2015;5:3e55.
68. Pieragostino D, D’Alessandro M, di Ioia M, Rossi C, Zucchelli M,
Urbani A, et al. An integrated metabolomics approach for the
research of new cerebrospinal fluid biomarkers of multiple
sclerosis. Mol Biosyst 2015;11:1563e72.
69. Cassol E, Misra V, Dutta A, Morgello S, Gabuzda D. Cerebro-
spinal fluid metabolomics reveals altered waste clearance
and accelerated aging in HIV patients with neurocognitive
impairment. AIDS (Lond) 2014;28:1579e91.
70. O’Mahony SM, Clarke G, Borre YE, Dinan TG, Cryan JF. Sero-
tonin, tryptophan metabolism and the brain-gut-microbiome
axis. Behav Brain Res 2015;277:32e48.
71. Del Rio D, Zimetti F, Caffarra P, Tassotti M, Bernini F,
Brighenti F, et al. The gut microbial metabolite
trimethylamine-n-oxide is present in human cerebrospinal
fluid. Nutrients 2017;9.
72. Brown DG, Rao S, Weir TL, O’Malia J, Bazan M, Brown RJ,
et al. Metabolomics and metabolic pathway networks from
human colorectal cancers, adjacent mucosa, and stool. Can-
cer Metabol 2016;4:11.
73. Budczies J, Brockmoller SF, Muller BM, Barupal DK, Richter-
Ehrenstein C, Kleine-Tebbe A, et al. Comparative metab-
olomics of estrogen receptor positive and estrogen receptor
negative breast cancer: alterations in glutamine and beta-
alanine metabolism. J Proteomics 2013;94:279e88.
74. Garcia-Sevillano MA, Contreras-Acuna M, Garcia-Barrera T,
Navarro F, Gomez-Ariza JL. Metabolomic study in plasma,
liver and kidney of mice exposed to inorganic arsenic based
on mass spectrometry. Anal Bioanal Chem 2014;406:
1455e69.
75. Chuang HL, Huang YT, Chiu CC, Liao CD, Hsu FL, Huang CC,
et al. Metabolomics characterization of energy metabolism
reveals glycogen accumulation in gut-microbiota-lacking
mice. J Nutr Biochem 2012;23:752e8.
76. West JA, Beqqali A, Ament Z, Elliott P, Pinto YM, Arbustini E,
et al. A targeted metabolomics assay for cardiac metabolism
and demonstration using a mouse model of dilated cardio-
myopathy. Metabolomics : Off J Metabolomic Soc 2016;12:59.
77. Matsumoto M, Kibe R, Ooga T, Aiba Y, Kurihara S, Sawaki E,
et al. Impact of intestinal microbiota on intestinal luminal
metabolome. Sci Rep 2012;2:233.
78. Tan B, Qiu Y, Zou X, Chen T, Xie G, Cheng Y, et al. Metabo-
nomics identifies serum metabolite markers of colorectal
cancer. J Proteome Res 2013;12:3000e9.
79. De Angelis M, Piccolo M, Vannini L, Siragusa S, De Giacomo A,
Serrazzanetti DI, et al. Fecal microbiota and metabolome of
children with autism and pervasive developmental disorder
not otherwise specified. PLoS One 2013;8, e76993.
80. Yu LM, Zhao KJ, Wang SS, Wang X, Lu B. Gas chromatogra-
phy/mass spectrometry based metabolomic study in a murine
model of irritable bowel syndrome. World J Gastroenterol
2018;24:894e904.
81. Nusbaum DJ, Sun F, Ren J, Zhu Z, Ramsy N, Pervolarakis N,
et al. Gut microbial and metabolomic profiles after fecal
microbiota transplantation in pediatric ulcerative colitis pa-
tients. FEMS Microbiol Ecol 2018;94(9). fiy133.
82. Yin S, Guo P, Hai D, Xu L, Shu J, Zhang W, et al. Optimization
of GC/TOF MS analysis conditions for assessing host-gut
microbiota metabolic interactions: Chinese rhubarb alters
fecal aromatic amino acids and phenol metabolism. Anal
Chim Acta 2017;995:21e33.
S22
83. Shen CT, Zhang Y, Liu YM, Yin S, Zhang XY, Wei WJ, et al. A
distinct serum metabolic signature of distant metastatic
papillary thyroid carcinoma. Clin Endocrinol 2017;87:844e52.
84. Castellani C, Singer G, Kaiser M, Kaiser T, Huang J, Sperl D,
et al. Neuroblastoma causes alterations of the intestinal
microbiome, gut hormones, inflammatory cytokines, and bile
acid composition. Pediatr Blood Cancer 2017;64:e26425.
85. Rusconi B, Jiang X, Sidhu R, Ory DS, Warner BB, Tarr PI. Gut
sphingolipid composition as a prelude to necrotizing entero-
colitis. Sci Rep 2018;8:10984.
86. Ferrer M, Raczkowska BA, Martinez-Martinez M, Barbas C,
Rojo D. Phenotyping of gut microbiota: focus on capillary
electrophoresis. Electrophoresis 2017;38:2275e86.
87. Mishima E, Fukuda S, Mukawa C, Yuri A, Kanemitsu Y,
Matsumoto Y, et al. Evaluation of the impact of gut micro-
biota on uremic solute accumulation by a CE-TOFMS-based
metabolomics approach. Kidney Int 2017;92:634e45.
88. Cameron SJS, Takats Z. Mass spectrometry approaches to
metabolic profiling of microbial communities within the
human gastrointestinal tract. Methods 2018. https:
//doi.org/10.1016/j.ymeth.2018.04.027.
89. Misra BB, van der Hooft JJJ. Updates in metabolomics tools
and resources: 2014e2015. Electrophoresis 2016;37:86e110.
90. Wang SY, Kuo CH, Tseng YJ. Ion trace detection algorithm to
extract pure ion chromatograms to improve untargeted peak
detection quality for liquid chromatography/time-of-flight
mass spectrometry-based metabolomics data. Anal Chem
2015;87:3048e55.
91. Kopka J, Schauer N, Krueger S, Birkemeyer C, Usadel B,
Bergmuller E, et al. GMD@CSB.DB: the Golm metabolome
database. Bioinformatics 2005;21:1635e8.
92. Cui Q, Lewis IA, Hegeman AD, Anderson ME, Li J, Schulte CF,
et al. Metabolite identification via the Madison metabolomics
Consortium database. Nat Biotechnol 2008;26:162e4.
93. Horai H, Arita M, Kanaya S, Nihei Y, Ikeda T, Suwa K, et al.
MassBank: a public repository for sharing mass spectral data
for life sciences. J Mass Spectrom 2010;45:703e14.
94. Smith CA, Maille GO, Want EJ, Qin C, Trauger SA, Brandon TR,
et al. METLIN: a metabolite mass spectral database. Ther
Drug Monit 2005;27:747e51.
95. Wishart DS, Feunang YD, Marcu A, Guo AC, Liang K, Vazquez-
Fresno R, et al. HMDB 4.0: the human metabolome database
for 2018. Nucleic Acids Res 2018;46:D608e17.
96. Ho TJ, Kuo CH, Wang SY, Chen GY, Tseng YJ. True ion pick
(TIPick): a denoising and peak picking algorithm to extract ion
signals from liquid chromatography/mass spectrometry data.
J Mass Spectrom 2013;48:234e42.
97. Zhu P, Ding W, Tong W, Ghosal A, Alton K, Chowdhury S. A
retention-time-shift-tolerant background subtraction and
noise reduction algorithm (BgS-NoRA) for extraction of drug
metabolites in liquid chromatography/mass spectrometry
data from biological matrices. Rapid Commun Mass Spectrom
2009;23:1563e72.
98. Savitzky A, Golay MJE. Smoothing and differentiation of data
by simplified least squares procedures. Anal Chem 1964;36:
1627e39.
99. Mahieu NG, Genenbacher JL, Patti GJ. A roadmap for the
XCMS family of software solutions in metabolomics. Curr Opin
Chem Biol 2016;30:87e93.
M.X. Chen et al.
100. Cappadona S, Baker PR, Cutillas PR, Heck AJ, van
Breukelen B. Current challenges in software solutions for
mass spectrometry-based quantitative proteomics. Amino
Acids 2012;43:1087e108.
101. Dettmer K, Aronov PA, Hammock BD. Mass spectrometry-
based metabolomics. Mass Spectrom Rev 2007;26:51e78.
102. Tautenhahn R, Bottcher C, Neumann S. Highly sensitive feature
detection for high resolution LC/MS. BMC Bioinf 2008;9:504.
103. Rost HL, Sachsenberg T, Aiche S, Bielow C, Weisser H,
Aicheler F, et al. OpenMS: a flexible open-source software
platform for mass spectrometry data analysis. Nat Methods
2016;13:741e8.
104. Pluskal T, Castillo S, Villar-Briones A, Oresic M. MZmine 2:
modular framework for processing, visualizing, and analyzing
mass spectrometry-based molecular profile data. BMC Bioinf
2010;11:395.
105. Wei X, Shi X, Kim S, Zhang L, Patrick JS, Binkley J, et al.
Data preprocessing method for liquid chromatography-mass
spectrometry based metabolomics. Anal Chem 2012;84:
7963e71.
106. Yang C, He Z, Yu W. Comparison of public peak detection
algorithms for MALDI mass spectrometry data analysis. BMC
Bioinf 2009;10:4.
107. Girlanda R, Cheema AK, Kaur P, Kwon Y, Li A, Guerra J, et al.
Metabolomics of human intestinal transplant rejection. Am J
Transplant 2012;12:S18e26.
108. McCall LI, Tripathi A, Vargas F, Knight R, Dorrestein PC,
Siqueira-Neto JL. Experimental Chagas disease-induced per-
turbations of the fecal microbiome and metabolome. PLoS
Neglected Trop Dis 2018;12, e0006344.
109. Hintze KJ, Cox JE, Rompato G, Benninghoff AD, Ward RE,
Broadbent J, et al. Broad scope method for creating human-
ized animal models for animal health and disease research
through antibiotic treatment and human fecal transfer. Gut
Microb 2014;5:183e91.
110. Considine EC, Thomas G, Boulesteix AL, Khashan AS,
Kenny LC. Critical review of reporting of the data analysis
step in metabolomics. Metabolomics : Off J Metabolomic Soc
2017;14:7.
111. Smith CA, Want EJ, O’Maille G, Abagyan R, Siuzdak G. XCMS:
processing mass spectrometry data for metabolite profiling
using nonlinear peak alignment, matching, and identification.
Anal Chem 2006;78:779e87.
112. Bijlsma S, Bobeldijk I, Verheij ER, Ramaker R, Kochhar S,
Macdonald IA, et al. Large-scale human metabolomics
studies: a strategy for data (pre-) processing and validation.
Anal Chem 2006;78:567e74.
113. Di Guida R, Engel J, Allwood JW, Weber RJ, Jones MR,
Sommer U, et al. Non-targeted UHPLC-MS metabolomic
data processing methods: a comparative investigation of
normalisation, missing value imputation, transformation
and scaling. Metabolomics : Off J Metabolomic Soc 2016;
12:93.
114. Chong J, Soufan O, Li C, Caraus I, Li S, Bourque G, et al.
MetaboAnalyst 4.0: towards more transparent and integrative
metabolomics analysis. Nucleic Acids Res 2018;46:W486e94.
115. Tautenhahn R, Patti GJ, Rinehart D, Siuzdak G. XCMS Online:
a web-based platform to process untargeted metabolomic
data. Anal Chem 2012;84:5035e9.