Recent Review on Selected Xenobiotics and Their Impacts on Gut Microbiome and Metabolome

Abstract

As it is well known, the gut is one of the primary sites in any host for xenobiotics, and the many microbial metabolites responsible for the interactions between the gut microbiome and the host. However, there is a growing concern about the negative impacts on human health induced by toxic xenobiotics. Metabolomics, broadly including lipidomics, is an emerging approach to studying thousands of metabolites in parallel. In this review, we summarized recent advancements in mass spectrometry (MS) technologies in metabolomics. In addition, we reviewed recent applications of MS-based metabolomics for the investigation of toxic effects of xenobiotics on microbial and host metabolism. It was demonstrated that metabolomics, gut microbiome profiling, and their combination have a high potential to identify metabolic and microbial markers of xenobiotic exposure and determine its mechanism. Further, there is increasing evidence supporting that reprogramming the gut microbiome could be a promising approach to the intervention of xenobiotic toxicity.

1. Introduction

The gut microbiome in the human gastrointestinal tract is a complex and dynamic microorganism community that plays an important role in microbial metabolism and maintenance of the host health [1-5]. In humans, gut microbiome has hundreds of bacterial species and nearly two million microbial genes that are around a hundred times as many as the host [6]. Gut microbiota was identified as an “organ” in the human body in the past decade, and it co-evolves with the host and participates in numerous local physiological functions [7, 8], such as maintaining the intestinal membrane, mucus layer, and epithelium, as well as metabolizing xenobiotics and interacting with host immunity [9-11]. Remotely, gut microbiome can crosstalk with multiple host organs throughout the “gut-organ axis” (Fig. 1) [12-14]. There is increasing evidence showing that the gut microbiome is vital to a host’s health; e.g., if the homeostatic state of the gut microbiome is disrupted (dysbiosis), the crosstalk between the microbes and mucosal immune system will be compromised, which can cause various inflammatory responses and immune diseases [15, 16]. Numerous factors, including diet, environment, host health conditions, etc., are connected to potential gut microbiota disturbances. One of the most common factors is exposure to xenobiotics [17-19].

Fig. 1.

The gut serves as a crucial site for several xenobiotic exposures, which can impact a variety of essential microbial metabolites, including bile acids, short-chain fatty acids (SCFAs), tryptophan (Trp) metabolites, trimethylamine N-oxide (TMAO), and branched-chain amino acids (BCAA). In addition, the gut can remotely communicate with other organs within the host, leading to toxic effects through the gut-organ axis.

The gut serves as a major location within the host for diverse xenobiotics that are foreign chemical substances not naturally present within an organism [20-23]. Exposure to xenobiotics is unavoidable, since these substances are present in drugs, dietary supplements, industrial chemicals, food additives, pesticides, and various other environmental pollutants [24-27]. Exposure to these xenobiotics can occur in a variety of ways such as ingestion, inhalation, and dermal absorption [28]. Many of these xenobiotics are first accumulated and absorbed in the gut, and they undergo metabolic processes of reducing toxicity, increasing hydrophilicity, and becoming readily excretable [29, 30]. Gut microbiome plays a significant role in modulating the bioavailability and metabolism of xenobiotics. Not only can gut microbiome affect both the kinetics and dynamics of xenobiotic metabolism, but also the gut-derived molecules can enter the host’s circulation and directly affect the biological functions of host organs [31, 32]. As a result, multiple gut-organ axis, such as gut-liver axis (GLA), gut-brain axis (GBA), gut-pancreas axis (GPA), and gut-kidney axis (GKA), represent bidirectional signal pathways from the gut to various organs in the host (Fig. 1) [33-37]. These interactions are possible through means of metabolic, neural, endocrine, immune, and humoral connections [33, 38]. Chronic or acute exposure to toxic xenobiotics may alter the gut microbial composition and even the location of certain gut microbes, which eventually impacts host health [21]. The alarming issue of xenobiotic pollution, characterized by its high toxicity, long-lasting persistence, and limited biodegradability, has caused severe environmental harm on a global scale. Microorganisms have the potential to use xenobiotic compounds as a source of carbon or nitrogen to support their growth and metabolic functions. As a result, microbial-assisted degradation of xenobiotics is widely acknowledged as an effective and environmentally friendly approach. For example, various microbial strains such as Alcaligenes and Cellulosimicrobium have been identified and isolated for their exceptional biodegradation potential against diverse xenobiotic pollutants in soil and water environments [39]. In addition, growing evidence shows that reprogramming the gut microbiota through the administration of probiotics, prebiotics, and microbiome transplants can promote a healthy gut environment in the host, establishing a beneficial bacterial community. This approach can be utilized to alleviate the toxicity of xenobiotics, enhance immune function, and potentially aid in targeted disease (such as cancer) therapy [40-42].

The interaction between the gut microbiome and the host is critically impacted by metabolism [43-47]. Over the past 10 years, there has been a dramatic increase in the number of publications investigating the association between gut microbiome and metabolome, as well as their connections to the host’s health/disease (Fig. 2). Through producing numerous essential microbial metabolites during food digestion and xenobiotic metabolism, the gut microbiota plays a crucial role in preserving the host's health and homeostasis, as supported by numerous pieces of evidence (Fig. 1) [16, 48]. For example, bile acids (BAs) are well-known pleiotropic signaling molecules mechanistically involved in gut-liver crosstalk [49-52]. BAs are a group of steroids produced in the liver from cholesterol [53]. These primary BAs are secreted into the lumen of the duodenum and metabolized by the gut microbiota into more lipophilic secondary BAs, the majority of which are absorbed and recirculated to the liver [50, 54]. In addition, short chain fatty acids (SCFAs) are mechanistically involved along the gut-brain axis [55, 56]. SCFAs, such as acetic acid, propionic acid, and butyric acid, are produced through the gut bacterial fermentation of dietary carbohydrates, mainly including fiber and starch, by bacteria in the gut [57, 58]. SCFAs have immunomodulatory properties and can interact with nerve cells by stimulating the sympathetic and autonomic nervous systems via G-protein-coupled receptor 41 (GPR41) and 43 (GPR43) [59]. Gut-derived SCFAs can enter the circulatory system and cross the blood-brain barrier (BBB), which makes it possible for them to directly modulate brain development and behavior [60, 61]. It is interesting to note that introducing Akkermansia muciniphila, which is a major producer of SCFAs, from an external source can have positive effects on neurological disorders like Alzheimer’s disease and seizures. The proposed ways in which this works include improving gene expression by enhancing histone acetylation, regulating amino acids and γ-aminobutyric acid (GABA) signaling in the central nervous system, and reducing inflammation [62-66]. Further, microbiota-dependent metabolites, including those in tryptophan (Trp) metabolism, play an important role in the gut-heart axis [67, 68]. Indole-3-propionic acid (IPA) is a microbial tryptophan derivative and a mitochondrial modulator in cardiomyocytes, and can directly impact cardiac functions [69]. Gut microbiota facilitates mammalian host nicotinamide adenine dinucleotide (NAD⁺) biosynthesis through a microbial nicotinamidase (PncA) [70]. Earlier studies showed that NAD⁺-dependent acetylation contributes to pressure overload-induced heart failure. While, activation of the NAD⁺ biosynthetic pathways can elevate NAD⁺ levels and alleviate cardiac dysfunction [71, 72].

Fig. 2.

The number of publications in the past 10 years, using the searching key words of “metabolome” and “microbiome” in PubMed (https://pubmed.ncbi.nlm.nih.gov/).

In this review, we aim to elucidate the key steps involved in the applications of omics, mainly including metabolomics and microbiomics, in recent selected xenobiotic exposure studies. In brief, we first outlined the general study design. In addition, we reviewed targeted and untargeted mass spectrometry (MS)-based metabolomics, metabolic flux analysis, metagenomic sequencing, as well as recent methods development related to these technologies. Further, we performed a comprehensive review of recent studies investigating xenobiotic exposures of wide interest and their effects on the gut microbiome and metabolome.

2. General Study Design

The general workflow is displayed in Fig. 3. Xenobiotic exposure can be conducted on humans, rodents, and cell models. To assess the diversity and functions of gut microbiota, samples of feces and/or intestinal content are collected. In general, the analysis of microbial communities begins with the sequencing of DNA to generate a catalog of genes and/or organisms [73]. Metagenomic sequencing analyzes genetic materials from mixed microbial populations through targeted or shotgun sequencing. Shotgun sequencing performs a more comprehensive analysis of the microbiome's composition and functions, providing an integrated view of both taxonomic and functional information without targeting specific genes or regions. This approach provides a more in-depth view of the microbiome, allowing for the identification of both known and unknown microorganisms, as well as their potential functions. 16S rRNA analysis is a targeted approach that focuses on a specific gene, the 16S ribosomal RNA gene, which is present in bacteria and archaea. The gene is conserved across these microbial groups but contains variable regions that can be used to identify and classify different bacterial and archaeal species. In contrast, whole shotgun metagenomic sequencing involves the sequencing of all DNA present in a sample, including the DNA from bacteria, archaea, viruses, and other microorganisms. Both approaches have their advantages and disadvantages, 16S rRNA analysis is often used as a cost-effective and rapid method to assess the microbial community composition in a sample. It can be particularly useful for large-scale surveys of microbial diversity or when the focus is on bacterial or archaeal communities. Whole shotgun metagenomic sequencing, on the other hand, provides a more detailed view of the microbiome but is often more expensive and computationally more intensive [74-77].

Fig. 3.

General flowchart for identifying metabolite biomarkers and investigating metabolic mechanisms, connecting gut microbiome, xenobiotic exposure, and host health.

Metabolomics, broadly including lipidomics, promises novel avenues for the detection of thousands of microbial and host metabolites associated with xenobiotic exposure and understanding the metabolic mechanisms that involve dysbiosis of gut microbiome [78-89]. Metabolomics focuses on detecting alterations at the metabolite level using analytical chemistry techniques and multivariate statistical analysis. To profile microbial metabolites, feces and intestine content samples are often collected. Host metabolic profiles will be measured from tissue, biofluid, and cell samples. In general, samples are first subjected to quenching, followed by metabolite extraction, and sometimes derivatization if necessary. The appropriate solvent systems are utilized to extract metabolites from the biological samples, and they are optimized based on the specific classes of metabolites under investigation. Typically, aqueous solvents are used to extract polar metabolites like amino acids and sugars, while relatively nonpolar metabolites such as lipids and fatty acids are extracted using a high portion of organic solvents. Optimal sample preparation methods and MS procedures may vary depending on the specific applications. Previous studies have established various workflows. For example, Courant et al. presented a tutorial that includes key steps from study design, sample preparation, MS data acquisition, and data analysis of the MS-based metabolomics flow, which could help beginners understand the concept [90]. Miggiels et al. focused on sample preparation methods, such as solid-phase extraction, liquid-liquid microextraction, and electro-driven extractions, as well as the introduction of NMR, liquid chromatography-mass spectrometry (LC-MS), gas chromatography-mass spectrometry (GC-MS), and capillary electrophoresis-MS [91]. Gong et al. and Hemmati et al. focused on sample preparation methods used in previous metabolomics studies, such as dry blood spot extraction, headspace-solid phase microextraction, microwave-assisted extraction, ultrasound-assisted extraction, and enzyme-assisted extraction techniques [92, 93]. Potential metabolite biomarkers will be selected after compound identification and statistical analysis, and metabolic pathway analysis is used to gain a better understanding of their potential roles in the biological processes. Importantly, further validation studies are typically needed to confirm the significance of identified biomarkers for specific applications. This workflow in Fig. 3 enables the integration of metabolic profiles and microbial community analysis results to yield metabolite and microbial biomarkers, along with a mechanistic understanding of the associated xenobiotic exposure.

Investigation of both microbial and host metabolites is a promising approach to accurately evaluate the toxicity of xenobiotics and develop therapeutic targets. In this review, we will summarize the recent advancements in metabolomics, especially those studies related to toxic xenobiotics using in vivo and in vitro biological models. Notably, we will focus on a few selected xenobiotics of wide interest in recent years. We are aware that there are many more xenobiotics currently under investigation, for more comprehensive and in-depth reviews, please see [21, 28, 94-96].

3. Recent Advancements of Metabolic Profiling Approaches

Currently, MS is the most commonly used analytical detection technology in metabolomics due to its high sensitivity and specificity [91, 97-100]. LC-MS and GC-MS are mainly used to detect metabolites and lipids, both widely used in targeted and untargeted metabolomics [101, 102]. The pace of method development in MS-based metabolomics has been rapid, in this review, a few highlights from the past few years will be reintroduced.

3.1. Untargeted metabolomics

Untargeted metabolomics, broadly including untargeted lipidomics, is widely used to generate hypotheses by identifying metabolites and pathways that are altered under certain conditions. Untargeted metabolomics is a powerful approach that allows for the comprehensive analysis of metabolites in a biological sample without prior knowledge of which metabolites are of interest. This contrasts with targeted metabolomics, which focuses on the measurement of specific metabolites. Untargeted metabolomics can be used to identify biomarkers of diseases or to gain a more comprehensive understanding of metabolic pathways and their regulation in various biological contexts. The choice between targeted and untargeted metabolomics depends on the specific research question and the available preliminary results [103-107]. Untargeted metabolomics is used to acquire unbiased information of the metabolites and their variations in the host and gut microbiome. High-sensitivity and high-resolution MS (HRMS) analyzers, such as time-of-flight (Tof) and orbitrap, are predominantly applied in untargeted metabolomics approaches, since mass accuracy is very informative for structure identification of the detected metabolites [97]. Typically, the mass range for aqueous metabolites is <1,000 Da, while for lipids it is < 2,000 Da [108-111]. However, the actual range used in specific studies may vary depending on the research questions and specific metabolites of interest. These metabolic results can be further associated with metagenomic sequencing or 16S rRNA analysis to investigate the microbial functions and their interactions with the host [98].

Once data is collected, specialized software packages are utilized for various tasks such as peak deconvolution, peak picking, data integration, structure annotation, statistical analysis, and pathway analysis [112-117]. To date, open-source software for small molecule discovery in untargeted metabolomics primarily includes MZmine, mzMatch, MS-DIAL, Ideom, and XCMS. For example, Fiehn’s group developed MS-DIAL that is a software pipeline for data-independent acquisition (DIA)-based identification and quantification of small molecules in comprehensive untargeted acquisition of metabolic data [113]. Zhu’s group established the Metabolite identification and Dysregulated Network Analysis (MetDNA) approach for identification of metabolites in LC-MS-based untargeted metabolomics [117]. Commercial software can also be used, such as SIEVE, MassHunter, MassProfiler Professional, Progenesis CoMet, Compound Discoverer, LipidSearch, MarkerLynx, and Progenesis Q. The databases for compound identification and annotation mainly include Human Metabolome Database (HMDB), METLIN, Chemspider, BioCyc, Cayman Chemical Compounds Database, KEGG, LipidMAPS, and mzCloud. Siuzdak’s group established the METLIN Metabolite and Chemical Entity Database that is the largest repository of MS/MS and neutral loss data acquired from standards to assist in metabolite and chemical entity identification [118, 119]. Xia’s and Wishart’s groups developed MetaboAnalyst and HMDB, which are incredibly useful for statistics analysis and metabolic pathways analysis in metabolomics. MetaboAnalyst is a free web-based platform for comprehensive metabolomics data analysis, supporting raw MS spectra processing, data normalization, statistical analysis, functional analysis, meta-analysis, and integrative analysis with other omics data [120, 121]. HMDB is a freely available web database containing comprehensive and detailed information of small molecules in the human body [122, 123]. Notably, >8,610 proteins, including enzymes and transporters, are also linked to these metabolites and lipids. Many entries are hyperlinked to other databases, such as KEGG and GenBank. There is extensive literature available that evaluates the performance of different software packages in various metabolomics analyses. The listed software packages are widely used and have been well organized in many previous reviews and metabolomics studies [124-127].

Table 1 summarizes the key parameters of the selected studies related to the impact of xenobiotic exposures on the gut microbiome. Methanol (MeOH) and acetonitrile (ACN) are the most commonly used solvents for protein precipitation and efficient extraction of metabolites in biological samples [86, 87, 128-130]. In lipid extraction, chloroform and MTBE are two commonly used solvents [131-133]. For the separation of hydrophilic compounds, the recommended separation mode is hydrophilic interaction liquid chromatography (HILIC), with popular options like ACQUITY UPLC BEH Amide and XBridge BEH Amide columns [87, 134]. Reverse phase columns are widely employed for hydrophobic compounds, mainly including XSelect HSS T3, ZORBAX Eclipse Plus C8/C18, ACQUITY UPLC BEH C18, ACQUITY UPLC HSS T3, Hypersil Gold C18, etc. [86, 128, 131, 134, 135] Untargeted metabolomics often utilizes both negative and positive ionization modes to gather comprehensive metabolic profiling [136]. Chromatographic conditions may differ between these modes, such as the use of distinct columns or mobile phase conditions. [137]. For GC separation, helium is often used as the carrier gas, and the metabolite extracts of the samples are usually separated on a capillary column, such as HP-5 and DB-5, after derivatization [129, 130]. Detailed sample preparation protocols for extracting microbial metabolites are well documented in previous reviews, including sample collection, quenching, and extraction for different types of samples (tissue, medium, cells, etc.) [93, 98, 99].

Table 1.

Summary of Recent Publications Related to Metabolomics, Gut Microbiome and Xenobiotic Exposure

Xenobiotics, dose	(Animal strain) Sample type for metabolomics	Sample preparation (extraction solvent; ISs; derivatization reagent for GC)	Column for analytical platform	Nontargeted metabolomics		Targeted metabolomics		Transcriptomics	Genomics	References
				Analytical platform	Listed differential metabolites	Analytical platform	Target analytes
BDE-3, 0.0015-30 (mg/kg)	(C57BL/6J) Mouse testis, urine, serum	80% MeOH (testis), MeOH (urine, serum); 2-Chloro-L-phenylalanine	XSelect HSS T3	UPLC-Q-TOF/MS	76, 38, and 31 metabolites in testis, urine, and serum	-	-	-	-	Wei et al. (2018) [128]
BDE-47, 10.0 (mg/kg)	(SD) Rat serum	MeOH; N/A; BSTFA	HP-5MS	GC-MS	26 metabolites	-	-	-	Fecal 16S rRNA Seq	Gao et al. (2021) [129]
18 PBDE congeners, N/A	Human placenta	80% MeOH; tridecanoate and 3 deuterated ISs; MSTFA	DB-5MS	GC-MS	66 metabolites	-	-	-	-	Wang et al. (2022) [130]
BDE-47, 48.5; BDE-99, 56.5 (mg/kg)	(C57BL/6J) Mouse serum, liver, small and large intestine content	MeOH (serum), ACN-MeOH (liver, small and large intestine content); 5 deuterated ISs	XSelect HSS T3	-	-	UPLC-QQQ-MS	56 BAs	Liver RNA-Seq	Fecal 16S rRNA Seq	Li et al. (2018) [86]
BDE-47, 48.5; BDE-99, 56.5 (mg/kg)	(C57BL/6J) Mouse serum, liver, small and large intestine content	MeOH (serum), 80% MeOH (liver, small intestine content, large intestine content); 2 13C-labeled ISs	Xbridge BEH Amide	-	-	UPLC-QQQ-MS	>300 aqueous metabolites	Liver RNA-Seq	Fecal 16S rRNA Seq	Scoville et al. (2019) [87]
BDE-47, 0.2; TBBPA, 0.2; BPS, 0.2 (mg/kg)	(CD-1) Mouse liver, fecal; fecal (SCFAs)	BAs: ACN-MeOH; 5 deuterated ISs SCFAs: MeOH-H2O; hexanoic acid-6,6,6-d3; MTBSTFA	ZORBAX Eclipse Plus C18; HP-5MS	-	-	UPLC-QQQ-MS; GC-MS	56 BAs, 14 SCFAs	BA-processing related genes and cytokines in liver	Fecal 16S rRNA Seq	Gomez et al. (2021) [135]
PCB 126, 1.0 (μmol/kg)	(Ldlr −/−) Mouse liver; fecal (SCFAs)	SCFAs: MTBE; N/A; MTBSTFA	ACQUITY UPLC BEH C18, UPLC BEH Amide	-	-	UPLC-Q-Exactive; GC-MS	25 Hepatic Metabolism metabolites; 5 SCFAs	Inflammatory, metabolic, and gut health markers in jejunum and colon	Fecal 16S rRNA Seq	Petriello et al. (2018) [134]
Fox River PCB mixture, 6-30 (mg/kg)	(C57BL/6J) Mouse liver, serum	MeOH (serum), 80% MeOH (liver); 2 13C-labeled ISs	XBridge BEH Amide	-	-	UPLC-QQQ-MS	>300 aqueous metabolites	Liver RNA-Seq	Fecal 16S rRNA Seq	Lim et al. (2020) [182]
PCB 126, 1.0 (μg/L)	(D.rerio) Zebrafish intestine, serum	50% ACN (Metabolomics) MeOH and MTBE (lipidomics) ; N/A	ACQUITY UPLC BEH Amide, ACQUITY UPLC HSS T3	UPLC-Q-Exactive (metabolomics, lipidomics)	-	-	-	BA metabolism related genes in liver	Fecal Metagenomic Seq	Hu et al. (2021) [137]
Aroclor 1260 PCB mixture, 20 (ng/kg)	Pig serum	MeOH-H2O-chloroform; 4 deuterated ISs for metabolomics, LPC (15:0), PC (15:0/15:0), and TG (17:0/17:0/17:0 for lipidomics	Hypersil Gold C18, ACQUITY CSH C18	UPLC-Q-Exactive (metabolomics, lipidomics)	33 polar metabolites, 39 lipids	-	-	-	-	Hernandez-Mesa et al. (2022) [131]
PCB 126, 24 (μg/kg)	(C57BL/6J) Mouse urine, cecal content, liver, fecal	Metabolomics: 80% MeOH BAs: MeOH; deuterated ISs	XSelect HSS T3 (metabolomics); ACQUITY C8 BEH (BAs quantification)	1H NMR, UPLC-Q-Exactive (metabolomics, lipidomics)	-	UPLC-QQQ-MS	BAs	BA and FA metabolism related genes in liver and intestine	Cecal content 16S rRNA Seq, Cecal content metagenomic Seq	Tian et al. (2022) [230]
PCB 126, PCB 153, TCDF, 0.6-6 (μM); TCDD, 0.06-6 (μM)	(C57BL/6J) Mouse cecal microbiota (In vitro)	MeOH-H2O-chloroform; chlorpropamide and 15:0-18:1-d7-PC	Hydro-RP C18, CSH C18 column	1H NMR, UPLC-Q-Exactive (metabolomics, lipidomics)	-	-	-	Cecal bacterial mixtures RNA-Seq	Cecal bacterial mixtures 16S rRNA Seq	Tian et al. (2020) [231]
PFOS, 0.3-30 (μg/g)	(C57BL/6J) Mouse fecal	Diethyl ether; N/A	Rtx-Wax	-	-	GC-MS	5 SCFAs	Inflammation related genes in colon	Fecal 16S rRNA Seq	Wang et al. (2020) [240]
PFOS, 0.003-0.012% for mouse 5-50 (μg/g) for bacterial	(C57BL/6J) Mouse liver, cecal bacterial suspension	FAs: MeOH-chloroform; C15:0 free FA, methyl ester of C17:0	HP-5MS	1H NMR	-	GC-MS	5 FAs	Inflammation and other metabolism related genes in liver	Cecal content 16S rRNA Seq	Zhang et al. (2020) [132]
PFOA, 2.5-30 (mg/kg)	(C57BL/6J) Mouse fecal	Diethyl ether; N/A	N/A	-	-	GC-MS	5 SCFAs	-	Cecal content 16S rRNA Seq	Wang et al. (2021) [241]
PFOA, 0.5-3 (mg/kg)	(C57BL/6J) Mouse fecal	Ether; N/A	N/A	-	-	GC-MS	6 SCFAs	Inflammation related genes in colon	Fecal 16S rRNA Seq	Shi et al. (2020) [242]
PFOA, 300 (mg/kg)	(C57BL/6J) Mouse fecal	Ether; N/A	Rtx-Wax	-	-	GC-MS	SCFAs	-	Fecal 16S rRNA Seq	Shi et al. (2021) [243]
PFAS, 32.7±1.1 (μg/L)	(Emydura macquarii macquarii) Turtle fecal	N/A; 2 13C labeled ISs	ZORBAX Extend C18 for targeted analysis	UPLC-Q-TOF/MS	-	UPLC-QQQ-MS	Central carbon metabolism metabolites	-	Fecal 16S rRNA Seq	Beale et al. (2022) [244]
As, 0.5-10 (ppm)	(SD) Rat serum	MeOH; N/A	Kinetex C18	UPLC-Orbitrap	18 metabolites in lipid, amino acid, and nucleotide metabolism	-	-	Lipids and AA metabolism related genes in liver	-	Wang et al. (2015) [259]
As, N/A	Human urine	Water (no extraction); N/A	ACQUITY UPLC HSS T3	UPLC-Q-Exactive	38 metabolites	-	-	-	-	Khanam et al. (2022) [260]
As, 10 (ppm)	(C57BL/6J) Mouse fecal, urine, plasma	MeOH (urine, plasma), 50% MeOH (fecal) ; N/A	Waters C18 T3	UPLC-Q-TOF/MS	370 features in fecal, 40 and 10 metabolites in urine and plasma	-	-	-	Fecal 16S rRNA Seq	Lu et al. (2014) [261]
As, 10 (ppm)	(C57BL/6J) Mouse serum	MeOH; N/A	Poroshell 120 EC-C18	UPLC-Q-TOF/MS	33 metabolites	-	-	-	-	Xue et al. (2019) [262]
As, 30 (mg/kg)	(C57BL/6J) Mouse brain	Metabolomics: MeOH-ACN-H2O Lipidomics: MeOH and MTBE; 3 deuterated ISs	ACQUITY UPLC HSS T3, Kinetex C18	UPLC-Q-TOF-MS/MS (metabolomics), UPLC-Q-Exactive (lipidomics)	118 polar metabolites, 17 lipids	-	-	-	Fecal 16S rRNA Seq	Wang et al. (2021) [263]
As, 30 (mg/kg)	(C57BL/6J) Mouse fecal, plasma, cortex	MeOH-ACN-H2O (fecal, brain); MTBE (plasma) ; N/A	N/A	UPLC-Q-Exactive	31, 1, and 17 metabolites in fecal, cortex, and plasma	-	-	-	Fecal 16S rRNA Seq	Luo et al. (2023) [133]
Cd, 1-5 (μmol/L)	Human urine, Pancreatic beta cells (In vitro)	Urine: No extraction, L-2-chlorophenylalanine Cells: MeOH-ACN	HILIC column	UPLC-TOF/MS	76 metabolites in cells, 53 metabolites in human urine	-	-	Mitochondrial TCA cycle and fatty acid oxidation related genes in pancreatic beta cells and mouse pancreas	-	Hong et al. (2022) [266]
Cd, 1-50 (ppm)	(C57BL/6J) Mouse brain	MeOH-H2O; 13C3-lactate	ACQUITY UPLC BEH C18	UPLC-Q-Exactive	101 altered metabolites	UPLC-Orbitrap ID-X	Lactate	Brain RNA-Seq	-	Hudson et al. (2021) [267]
Cd 0.6-3 (mg/L)	(Humanized ApoE3-KI and ApoE4-K1) Mouse plasma	MeOH-H20; hexanoic acid-6,6,6-d3; MTBSTFA	HP-5MS	-	-	GC-MS	SCFAs	Liver RNA-Seq and inflammation related genes in liver	Large intestinal content 16S rRNA Seq	Zhang et al. (2021) [268]
Cd, As, 50 (ppm)	(C57BL/6J) Mouse fecal	ACN-H2O; N/A	self-packed 15 cm C18 column	UPLC-Q-Exactive	33 and 2 features in Cd and As treated group	-	-	-	Fecal 16S rRNA Seq	Li et al. (2019) [269]
Cd, 0.24-0.46; Se, 0.6-6 (mg/L)	Honey bee abdomen	MeOH-ACN-H2O-isopropanol; N/A	CSH phenyl-hexyl column	UPLC-Q-TOF/MS	2 Cd-induced and 7 Se-induced metabolites	-	-	-	Whole-bee 16S rRNA Seq	Rothman et al. (2019) [270]
Mn, 20 (mg/kg)	(C57BL/6J) Mouse fecal	MeOH-H2O-chloroform; N/A; BSTFA	N/A	GC-MS	36 and 32 metabolites in female and male mice	-	-	-	Fecal 16S rRNA Seq, Fecal metagenomic Seq	Chi et al. (2017) [275]
Mn, 200 (mg/L)	(SD) Rat plasma	MeOH-ACN; N/A	ACQUITY UPLC HSS T3	UPLC-Q-TOF/MS	17 metabolites	-	-	-	Fecal 16S rRNA Seq, Fecal metagenomics Seq	Wang et al. (2020) [276]
Pb, N/A	Human fecal	MeOH-H2O-chloroform; 11 ISs; BSTFA	HP-5MS	GC-MS	19 metabolites	-	-	-	Fecal 16S rRNA Seq	Zeng et al. (2022) [277]
PM2.5, 198.52 ± 21.56 (μg/m3)	(Balb/C) Mouse ileum	MeOH-H2O-chloroform (nontargeted metabolomics), MeOH (targeted analysis); 2-chlorobenzalanine	ACQUITY UPLC HSS T3 C18	UPLC-Orbitrap	31 metabolites	UPLC-QQQ-MS	17 metabolites in arachidonic acid metabolism	Inflammation related genes in ileum	Fecal 16S rRNA Seq	Dai et al. (2022) [292]
TRAP, N/A	Human plasma, saliva	ACN; N/A	HILIC column (pos ionization mode), C18 column (negionization mode)	UPLC-Q-Exactive	1291 associated features	-	-	-	-	Liang et al. (2018) [293]
TRAP, N/A	Human plasma	ACN; 14 isotopic labeled ISs	C18 column	UPLC-Q-Exactive	5112 associated features	-	-	-	-	Li et al. (2021) [294]

N/A: no information provided.

In gut microbiome metabolic studies, there are several approaches that could be employed to increase the stability of sample preparation, reduce data variation, and increase data quality. To prevent/correct analytical drift and provide data consistency including intensity monitoring, either a quality control (QC) sample is applied, and/or stable isotope labeled internal standards (SIL-ISs) are used. QC samples are useful for assessing the overall quality of the data and detecting any systematic errors that may have occurred during sample preparation or measurement. In addition, SIL-ISs can provide quantification of the analytes of interest and improve the precision and reproducibility of the data. For example, He et al. spiked a SIL-IS mixture that includes 14 deuterium labeled lipids to serum samples during methyl tert-butyl ether (MTBE) extraction, and ~1,500 lipids were detected with high reproducibility within 60 min gradient length by Orbitrap-based MS² [138].

3.2. Targeted metabolomics, pseudotargeted metabolomics, and beyond

In targeted metabolomics, the focus is on a specific set of metabolites that are relevant to the hypothesis being tested, while untargeted metabolomics is used where all metabolites are measured without prior knowledge of which ones may be of interest [139-142]. Triple quadrupole mass spectrometry (TQMS) based on selective ion monitoring mode (SIM) or multiple reaction monitoring (MRM) analysis [143-145], and HRMS-based parallel reaction monitoring (PRM) analysis [146, 147] are commonly used to perform targeted metabolomics. To date, various essential metabolites involved in microbial metabolism have been identified in previous studies, such as SCFAs, BAs, tryptophan metabolites, amino acids, and those in central carbon metabolism [148-150]. For these metabolites of strong interest, LC-MS was used for quantification/semi-quantification of BAs [151-153], SCFAs [154], tryptophan catabolites [155, 156], and methylamines [157, 158]. Nicholson’s group described an LC-MS procedure for sensitive and quantitative targeted analysis of 145 primary, secondary, and tertiary bile acids [153]. The assay had a great linearity with a lower limit of quantification (LLOQ) of 0.25-10 nM and an upper limit of quantification (ULOQ) of 2.5-5 μM. The group also confirmed precision (≈6.5%), accuracy (81.2-118.9% on inter- and intraday analysis), and recovery (serum/plasma 88% and urine 93%). Notably, SCFAs were also often measured by GC-MS after derivatization [159-163].

Large-scale targeted metabolomics was developed for the detection of over 300 aqueous metabolites and >280 lipids from >35 different metabolic pathways of strong biological significance [141, 164-167]. Munjoma et al. have developed a high-throughput HILIC-based LC-MS method for the semi-quantitative screening of more than 2,000 lipids using over 4,000 MRM transitions [168]. In another study, Zhou et al. developed a PRM assay to monitor 237 polar metabolites [169]. In addition, Zhang et al. applied PRM to analyze 20 amino acids and 40 derivatives in targeted analysis [170]. Our own assay of this type has been successfully used in a growing number of studies, including metabolic reprogramming during Myc-regulated tumorigenesis [171], cardiac metabolic shifts induced by rapamycin [172], metabolic characterization of mammalian skeletal muscles [173], and metabolome regulation during naive-to-primed human embryonic stem cell transition [174]. Chromatographic separations are observed by using HILIC and reverse phase modes for aqueous metabolites and lipids, respectively. LC-MS/MS system is optimized using metabolite and lipid standards, including precursor ion, product ion, collision voltage, and retention times (RTs). Targeted MS data acquisition is performed in the MRM mode, due to significant advantages of great selectivity and excellent quantitation. Samples are often run with a set of internal or external standards such that data variation will be minimized.

Xu’s group established pseudotargeted metabolomics, in order to combine the advantages of untargeted and targeted metabolomics [175-179]. Pseudotargeted methods acquire MS/MS fragmentation information from pooled biological samples in the full scan untargeted mode and collect as many ion pairs as possible under different collision energy (CE) voltages by multiple parallel injections. After removing the redundant features, the remaining interested ion pairs are utilized for CE optimization and MRM transitions in TQMS for measuring the samples. Zheng et al. provided a detailed protocol that illustrates the entire timeline of pseudotargeted metabolomics development. The process involves various steps, including sample preparation, data acquisition, MRM transition definition, parameter optimization, and method evaluation, which typically takes ~5 days to complete, allowing for the analysis of 800-1,300 metabolites [175]. Notably, the actual duration may vary depending on the number of samples analyzed in practical scenarios. This method was first introduced in 2012 using GC-MS, then in 2013 with LC-QTOF and LC-QQQ-MS platforms, and in 2015 with an upgraded procedure. The review by Liu et al. detailed pseudotargeted metabolomics analytical platforms [97].

Similarly, we developed a new analytical approach, globally optimized targeted (GOT)-MS, which has the capability to detect unknowns, broad metabolite coverage, and excellent quantitation [180]. From a serum sample, we obtained 595 precursor ions and 1,890 MRM transitions. For many MRMs/metabolites, the analytical performance of GOT-MS is better than or at least comparable to untargeted metabolomics. To significantly improve identification in GOT-MS, Shi et al. recently established a database-assisted globally optimized targeted mass spectrometry (dGOT-MS) approach that can provide a strong coverage of identified aqueous metabolites and lipids (310 confirmed with pure chemicals) [181]. Lim et al. applied dGOT-MS to investigate the metabolite changes due to polychlorinated biphenyls (PCB, a persistent organic pollutant) exposure in mouse liver and serum samples. The results showed that NADP and arginine varied with drug-metabolizing enzymes, which were highly correlated with Ruminiclostridium and Roseburia alterations [182].

3.3. Metabolic flux analysis (MFA)

MFA using stable isotope labeled tracers is extremely important for understanding metabolic mechanisms of various biological processes [183-188]. The whole metabolome can be viewed as an extensive map, in which metabolites (cities) and metabolic pathways (roads) comprise the metabolic reaction network. Metabolite levels provide “snapshots” of biochemical abundances; however, cellular metabolism in a living system is not static. In addition, many metabolites are involved in multiple pathways; therefore, concentration changes of a particular metabolite can result from perturbations in several pathways, which often confuses the analysis of metabolite level data. In contrast, MFA investigates the rate of metabolite turnover in biochemical reactions, which enables a dynamic model of cellular phenotype through tracing metabolite conversions taking place in the metabolic network. Furthermore, MFA can be easily connected to other important areas in systems biology, such as proteomics and genomics since the flux rate is regulated by enzymes. Therefore, MFA providing pathway-specific information is often required to fully understand the biological process [189-191]. In isotope tracing experiments, a biological system will receive isotopically labeled substrates, e.g., ¹³C, ¹⁵N, and ²H. GC/LC-MS data are collected and processed to detect metabolic features and align retention times. Enrichment of isotope labeling is processed to compare labeling patterns between different conditions. MetaboAnalyst, IsoCor, X¹³CMS, MZmine, and XCMS are software tools for processing and analyzing isotope labeling data. They provide normalization, statistical analysis, pathway analysis, and data visualization. Isotope labeling experiments and advanced software tools enable the investigation of metabolic pathways and fluxes in biological systems, providing a dynamic and deeper understanding of cellular metabolism and its response to perturbations.

Fan’s group developed stable isotope-resolved metabolomics (SIRM) that can analyze the fate of individual atoms from stable isotope-enriched precursors to deduce metabolic pathways and networks [184, 191]. Deng et al. developed an untargeted SIRM method for metabolic profiling of isotopic microbial metabolites after treating the fecal microbes with ¹³C-labeled dietary fibers (inulin or cellulose) [192]. Their results showed that ¹³C enrichment was successfully examined in both microbial cells and the culture medium. In addition, Patti’s group developed X¹³CMS for the identification of compounds that have incorporated the isotopic tracer, in an untargeted way [183].

Recently, Shi et al. established a TQMS-based comprehensive isotopic targeted mass spectrometry method in MFA [171]. This method was used to track ¹³C-enriched metabolites in Myc(oncogene)-On and Myc-Off Tet21N human neuroblastoma cells after being cultured in U-¹³C₆-glucose medium. A LC-MS/MS assay was developed to analyze 310 metabolites across 35 metabolic pathways. For each metabolite isotopologue, an isotopic MRM was generated based on the number of carbon atoms. The precursor ion and product ion m/z values were determined using the chemical formula. Isotopic MRMs were utilized in the LC-MS/MS measurements, and IsoCor was used to compute the mass isotopomer distribution and mean enrichment for each identified metabolite using the isotopic intensities. As a result, out of the 310 metabolites analyzed, 142 were detected with a high level of reliability, and 46 of these were found to be kinetically enriched. In addition, to investigate the time-course study under the pseudosteady state, the rate constant of decay and relative flux were calculated using the relative quantitation data. Citrate was examined as an example, and the results demonstrated an exponential decay pattern (as shown in Fig. 4A). The flux data allowed for clear differentiation between Myc-On and Myc-Off cells in the principal component analysis (PCA) score plot (as depicted in Fig. 4B). Further, Fig. 4C shows the metabolites with significant relative flux differences between Myc-On and Myc-Off cells in Ipath Pathways [171].

Fig.4.

A) Exponential fitting of the time-course data of citrate, B) the PCA score plot, and C) pathway view of metabolites with relative flux derived, for the comparison of Myc-On and Myc-Off Tet21N cells under pseudosteady state following U-¹³C₆-glucose labeling: red circles have p values < 0.05, blue circles have p values > 0.05, pink squares indicate metabolites present only in the Myc-Off group. This figure is reproduced from [171] with permission.

3.4. Novel MS techniques with highly promising applications in gut microbiome and xenobiotics studies

Metabolome within a living cell is highly dynamic and undergoes rapid shifts in response to the intracellular environment. Therefore, measuring metabolites within individual cells as they advance through metabolic pathways may provide a better understanding of cell metabolic activities [193, 194]. As such, MS-based single-cell metabolomics serves as a valuable tool to explore the metabolic capabilities and functions of individual microbial cells within this intricate ecosystem [195]. In single-cell metabolomics, various techniques, such as secondary ion mass spectrometry (SIMS), matrix-assisted laser desorption (MALDI), and laser ablation electrospray ionization (LAESI), can be employed for sampling and ionizing metabolites from individual cells [196]. In addition, the applications of nano-electrospray ionization (ESI) in single-cell metabolomics have been reviewed previously [197, 198]. Notably, challenges also exist in single-cell metabolomics [195], such as the small size of bacterial cells (approximately 1 μm in size) and metabolite identification (e.g., generation of large molecular fragments in SIMS high-energy ion beams). Additional limitations include difficulties in accurately measuring and quantifying metabolites within individual cells due to low abundance, potential leakage, and enzymatic degradation [199].

Another method with high potential in gut microbiome and xenobiotics studies is mass spectrometry imaging (MSI). MSI is a technique that provides information about the distribution and localization of molecules (focusing on metabolites in this review) in different regions of tissues or organisms [200]. MSI provides complementary information to that from histological methods [201]. MSI uses different ionization techniques such as MALDI, SIMS, and desorption electrospray ionization (DESI) [202]. Previous studies have utilized MSI to enhance the understanding of metabolic functions in different parts of the intestine. Hulme et al. utilized MSI to investigate alterations in neurotransmitters and spatial patterns in both the gut and brain in mice [203]. Their study showed that many metabolic changes were induced by gut microbes in the microbiome-gut-brain axis following an antibiotic treatment. Watrous et al. utilized two-dimensional MALDI-TOF MSI to three-dimensionally visualize metabolic exchange during microbial interactions between different microorganisms [204]. Similarly, MSI has limitations, such as difficulties in quantitation, sophisticated instrumentation, and annotating microbial metabolites [205].

Interestingly, MasSpec Pen, another novel MS-based technique, has emerged as a handheld device with immense clinical applications [206]. MasSpec Pen was specifically designed to enable rapid, non-destructive, and in situ analysis of the molecular composition of tissue samples during surgical procedures. The device can directly touch the tissue surface and release a droplet of water to extract metabolites from the tissue, particularly for patients under surgery. These extracted metabolites are subsequently transferred to MS, enabling the detection and characterization of the molecular composition associated with disease pathology (i.e., cancer type, tumor margin, and other disease-related features) [207, 208]. MasSpec Pen shows promises as a real-time, intraoperative tissue diagnostic tool, providing surgeons with valuable information leading to faster, safer, and more effective surgeries for patients. Notably, while we introduced single-cell metabolomics, MSI, and MasSpec Pen herein and they do have shown high potential for biological and clinical studies, to date, the applications of these technologies in gut microbiome and xenobiotic studies are very rare.

4. Persistent Organic Pollutant (POPs) Exposure, Gut Microbiome, and Metabolomics

POPs are highly toxic organic chemicals causing worldwide concerns due to their persistence in the environment and resistance to biodegradation [209-211]. POPs can easily remain in the environment for prolonged periods of time and accumulate in the food chain [209]. Exposure to POPs increases risks of cancer, autoimmune diseases, diabetes, obesity, and reproductive disorders [212-214]. There is an increasing number of studies published on metabolomics-based POPs exposure both in vivo and in vitro. Table 1 presents selected studies with essential parameters, such as animal model utilized, exposure levels of xenobiotics, sample preparation conditions, and analytical instruments along with the columns used. In this review, we will focus on a few POPs with predominant existence in the environment, including polybrominated diphenyl ethers (PBDEs), tetrabromobisphenol A (TBBPA), and polychlorinated biphenyls (PCBs).

4.1. Polybrominated diphenyl ethers (PBDEs) and tetrabromobisphenol A (TBBPA)

PBDEs are an important class of POPs, widely used in a variety of consumer products such as flame retardants, electronics, and textiles [215, 216]. Most PBDEs have been banned due to their neurotoxicity, but they still have managed to become widespread worldwide and accumulate as severe environmental pollutants. PBDEs can be found in seafood, meat, dietary products, and even indoor dust [217]. Nowadays, these compounds are found in human serum and breast milk samples in many areas and countries around the world [218]. Previous studies showed that acute exposure to BDE-47 and BDE-99 (the topmost enriched PBDE congeners) decreased the anti-diabetic tryptophan microbial metabolite, 3-Indolepropionic acid (IPA), in a gut microbiome-dependent manner [87]. Since PBDEs’ ban, TBBPA is currently the most widely used brominated flame retardant, and it constitutes approximately 60% of the worldwide demand [219, 220]. Like PBDEs, TBBPA can cross the placenta and can be passed from mother to fetus [221]. TBBPA has also been detected in 44-50% of breast milk samples from postpartum mothers, posing significant health risks to offspring during pregnancy and lactation [222]. TBBPA is glucuronidated and sulfated by the host and deconjugated by the gut microbiota before being excreted in feces as the parent compound [223].

In an in vitro study, He et al. observed that BDE-47 (a PBDE congener) exposure induced significant attenuation of mitochondrial respiration and glycolysis elevation [224]. To conduct GC-MS-based MFA, isotopically labeled glucose was substituted for unlabeled glucose in the cell culture medium. The presence of elevated isotopically labeled lactate in the BDE-47 treatment group confirmed an increase in glycolysis. Conversely, isotopically labeled metabolites involved in the TCA cycle, such as citrate, α-ketoglutarate, succinate, malate, and oxaloacetate, exhibited a significant decrease after BDE-47 treatment, indicating a reduction in mitochondrial respiration activity (Fig. 5). In addition, they found that multiple metabolic pathways including glycine, serine, threonine, and glutathione metabolism were disturbed via large-scale metabolomics, and eventually 17 metabolites were selected as the potential biomarkers of BDE-47 exposure.

Fig. 5.

Incorporation of glucose-derived carbon atoms into the TCA cycle and glycolysis metabolites. Bar charts left to right: control (black), 40 μM BDE-47 (blue), and 80 μM BDE-47 (red) groups; the Y-axis depicts the mean of isotopic enrichment of 13C in the metabolites. ** indicate a highly significant difference (p < 0.01). This figure is reproduced form Ref. [224] with permission.

Untargeted metabolomics approaches were applied to investigate the impacts of PBDE exposure on the gut microbiome-host metabolite profiles. Wei et al. used UHPLC-Q-TOF-MS in untargeted metabolomics, and they found that male mice presented a significant reproductive toxicity after BDE-3 exposure [128]. As a results, 76, 38, and 31 metabolites involved in nucleotides, lipids, tyrosine, purine, and riboflavin metabolism were significantly changed in testis, urine, and serum samples, respectively. Gao et al. studied BDE-47 exposure-induced impacts and association between the rat serum metabolites and gut microbes before, during, and after pregnancy [129]. Through GC-MS untargeted metabolomics, a total of 26 metabolites in amino acid, lipid, carbohydrate, and energy metabolism were found to have significant responses to BDE-47. The exposure to BDE-47 resulted in maternal gut dysbiosis, as displayed in Fig. 6. The operational taxonomic unit (OTU) clustering of 10 fecal samples yielded 577,315 effective sequences, revealing 3,428 shared OTUs between healthy controls and BDE-47-treated rats. Control animals had 445 unique OTUs, while BDE-47-treated rats had 565 unique OTUs. The rarefaction curve suggested that sufficient sampling had been conducted, while the α-diversity indexes revealed a slight decline in the BDE-47 group, although not statistically significant compared to the control group. The Bacteroidetes, Proteobacteria, and Cyanobacteria exhibited a decrease in relative abundance, while Firmicutes increased, although not significantly. At the phylum level, the 16S rRNA results indicated that BDE-47 exposure reduced the relative abundance of Actinobacteria. Recently, Wang et al. investigated the impacts of environmental PBDE exposure on prenatal metabolites in two different areas in China [130]. They confirmed prenatal exposure to multiple types of PBDEs by GC-MS/MS in the MRM mode, and they observed 66 PBDEs-associated prenatal metabolites using GC-MS in the full scan mode. The disrupted metabolic pathways included pentose phosphate pathway, arginine biosynthesis, ascorbate, threonine, butanoate, and lipid metabolism.

Fig. 6.

Impacts of BDE-47 exposure on the richness and pattern of maternal intestinal microbiota. A) Venn diagram of shared and unique OUT numbers of BDE-47-treated rats compared to controls. B) Alpha diversity indexes of gut microbiota in the control (n = 4) and BDE-47-treated rats (n = 6). The top and bottom boundaries of each box indicate the 75^th and 25^th percentiles, respectively; the line within each box represents the median; the bottom and top edges show the minimum and maximum, respectively. C) The observed species number. D-E) Gut microbial pattern at the phylum level within D) and between E) the control (n = 4) and the BDE-47 groups (n = 6) were assessed using 16S high throughput sequencing. This figure is reproduced from [129] with permission.

Microbial metabolites, such as BAs, SCFAs, and tryptophan catabolism metabolites, were targeted in xenobiotic metabolomics studies. Li et al. orally administrated BDE-47 and BDE-99 to male conventional and germ-free mice [86]. Both congeners decreased bacterial diversity but increased the abundances of Akkermansia muciniphila and Erysipelotrichaceae Allobaculum spp., as well as microbial 7α-dehydroxylation enzymes for secondary BA synthesis. Targeted metabolomics showed that the exposure also caused increased unconjugated BAs in multiple biospecimen. In addition, targeted proteomics revealed that PBDEs increased the expression of host intestinal transporters for BA absorption, while they downregulated host BA-synthesizing enzymes and transporters in livers of conventional mice. Scoville et al. found that PBDEs altered a majority number of aqueous metabolites in the large intestine content of germ-free mice [87]. PBDEs also regulated the correlation between microbial taxa composition and serum metabolites in the conventional mice. Gomez et al. investigated the effects of maternal exposure to BDE-47 on the metabolite profiles and gut microbiome composition in adult male mice [135]. Consistent up-regulation of BAs in 12α hydroxylation pathway was observed in fecal and liver samples after BDE-47 exposure, corresponding to an up-regulation of the hepatic BA synthetic enzyme Cyp7a1. Consistent with previous results, they showed that PBDE exposure modified the gut-liver axis, through altering gut microbiota, as well as multiple hosts and microbial metabolic pathways.

Related to TBBPA, previous studies showed that cells exposed to TBBPA exhibited mitochondrial dysfunction [225, 226]. In A549 cells, TBBPA caused morphological and ultrastructural changes, such as dilated smooth endoplasmic reticulum and extensively injured mitochondria, as well as excessive reactive oxygen species (ROS) generation [225]. Recently, Yu et al. observed that TBBPA significantly reduced the viability of A549 cells and attenuated mitochondrial respiration [227]. LC-MS data showed significant reductions in TCA cycle metabolites, and MFA indicated reduced oxidative capacity in mitochondrial metabolism following TBBPA exposure. These findings indicate that the cytotoxicity of TBBPA was at least partially due to the perturbed TCA cycle metabolism and mitochondrial respiration. In vivo, maternal exposure to TBBPA persistently modified the gut-liver axis, which may produce an immune-suppressive and dyslipidemia-prone signature later in life [135]. The most distinct microbial biomarker was Rikenellaceae for TBBPA exposure. For metabolites, fecal BAs were persistently up regulated by TBBPA exposure, and TBBPA also increased propionic acid and succinate.

4.2. Polychlorinated biphenyls (PCBs)

PCBs are a large group of chemicals possessing unique chemical stability and fire-resistant properties [228]. Although they are no longer widely used due to their toxicity, the bioaccumulation of PCBs has continued to increase in the past decades much like PBDEs [229]. Toxicity of PCB 126 on mouse gut microbiome and metabolic homeostasis was investigated by Petriello et al., and 16S rRNA results showed an obvious alteration of cecal microbial diversity and bacterial genera [134]. Using untargeted metabolomics, hepatic metabolism was found to be significantly changed, including glycolysis and lipogenesis. A significantly decreased level of formic acid was discovered in fecal samples. Lim et al. investigated the correlation between the aqueous metabolites in serum and liver with the hepatic transcriptome in PCBs-exposed mice, as well as the metabolism differences between conventional and germ-free mice [182]. PCBs induced upregulation of aryl hydrocarbon receptor and pregnane X receptor. Constitutive androstane receptor expression was increased in conventional mice compared to germ-free mice, indicating that PCBs impact the hepatic transcriptome at least partially through the gut microbiome. Hu et al. discovered that PCBs induced the blockage of bacterial primary BAs conversion and intestinal accumulation of BAs, through decreased microbial activities related to primary BA metabolism in zebrafish intestines [137]. A systematic deficiency of essential vitamins was detected due to PCBs. Hernandez-Mesa et al. performed a PCBs exposure study on pigs, where untargeted metabolomics and lipidomics results showed that 33 polar metabolites and 39 lipids involved in fatty acid metabolism, glycerophospholipid metabolism, and tryptophan-kynurenine pathway were significantly changed in serum [131]. Tian et al. found similar results in mouse models, and they confirmed that PCB 126 exposure induced abnormal liver amino acid, BA, and nucleotide metabolism, as well as enhanced hepatic lipogenesis [230]. Interestingly, another Tian’s study showed significantly lower unsaturated fatty acids (UFA) and monounsaturated fatty acids (MUFA) but higher saturated fatty acids (SFA) due to PCB 153 exposure, while the opposite results were observed in PCB 126 exposure [231]. They also found that PCBs induced higher levels of lipids in the bacterial membrane, including fatty acids, phosphatidylcholine (PC), phosphatidylethanolamine (PE), etc.

4.3. Per- and polyfluoroalkyl substances (PFASs)

PFASs are a class of synthetic chemicals used to make fluoropolymer coatings and products produced since the 1940s and used in various consumer and industrial products [232]. Although PFASs are now widespread in our environment and have been linked to severe health impacts such as hypercholesterolemia, ulcerative colitis, and liver cell damage, there are no enforceable regulatory standards to safeguard public health and drinking waters from this serious hazard [233]. While some nanoparticle-based methods have been developed to remove PFASs, their extensive use and persistence can cause reproductive issues, immunodeficiencies, and hormonal disruptions in both humans and wildlife [234-236]. Exposure to PFASs can directly impact the gut microbiome, as evidenced by the ability of the human fecal microbiome to transform the PFAS surfactant 8:2 monosubstituted polyfluoroalkyl phosphate ester [237]. PFASs exposure can occur in infants through human milk, potentially increasing their gut microbiome diversity and increasing the relative abundance of Bacteroides vulgatus [238]. A study tracked 124 individuals from birth to age 28, finding weak links between PFASs exposure and fastidious anaerobes. Several individual microbes were associated with certain compounds, including Bilophila wadsworthia, Faecalibacterium prautzii, Dorea longicatena, and Sutterella wadsworthensis [239]. Perfluorooctane sulfonate (PFOS) is one of the most abundant PFASs in the environment, extensively researched, and has been found to affect inflammatory reactions in the liver and colon, promoting the development of metabolic disorders. This includes altering adipocytokine signaling pathway, steroid hormone biosynthesis, flavonoid biosynthesis, lipid metabolism, oxidative stress, inflammation, TCA cycle, glucose, and amino acid metabolism. Exposure to PFOS in mice has been shown to dysregulate the levels of genera such as Firmicutes, Bacteroides, Proteobacteria, Clostridium, Streptococcus, and Blautia, and induce a loss of gut barrier integrity by reducing production of SCFAs and expression of intestinal tight junction proteins [132, 240]. In addition, Wang et al. demonstrated that perfluorooctanoic acid (PFOA) exposure resulted in liver inflammation, antioxidative homeostasis disruption, SCFAs level reduction, and liver histological abnormalities with hepatomegaly and injury. Subacute exposure altered the abundances of Dehalobacterium and Bacteroides genera, which contribute to liver inflammation and oxidative stress, while subchronic exposure decreased the abundances of potentially beneficial Lactobacillus and Bifidobacterium genera [241]. According to a publication by Shi et al., exposure to PFOA resulted in cognitive deficits in mice through changes in the composition of gut microbiota, impairment of the intestinal barrier, and an increase in inflammation in both the gut and brain. However, the symptoms were alleviated by FMT treatment [242]. In addition, the use of lactic acid bacteria (Lactobacillus strains) reduced liver inflammation induced by PFOA by providing antioxidants and biosorption, alleviating dysbiosis, and preventing decreased production of SCFAs [243]. Research conducted on wild freshwater turtles, where PFOS levels were elevated in their serum, demonstrated that the exposure to this substance affected the metabolism of amino acids, butanoate, purine, and pyrimidine in these turtles. Furthermore, the metagenomic analysis of fecal samples from PFAS-exposed turtles revealed a greater prevalence of Firmicutes and a decreased prevalence of Bacteroidota, compared to the control group of turtles [244].

5. Heavy Metal Exposure, Gut Microbiome, and Metabolomics

Heavy metal exposure is considered a global public health concern, including the exposure of arsenic (As), cadmium (Cd), manganese (Mn), lead (Pb), copper (Cu), lithium (Li), tungsten (W), etc. [245, 246]. Exposure to heavy metals may cause a shift in the gut microbiome composition; and eventually severe damage from genome to metabolome such as DNA breakdown, irreversible impairment to the nervous system, immune system disruption, and tumorigenesis [247, 248]. Heavy metals are omnipresent in the environment, and the human population is at risk for daily exposure [249]. There is a growing evidence based on metabolomics approaches, supporting the important role of gut microbiome participating in adverse consequences induced by heavy metal exposure (Table 1) [250-252].

5.1. Arsenic (As)

Arsenic is well known as a poison, and this chemical is present in the environment in both organic (oAs) and inorganic (iAs) forms [253, 254]. As has been found in some water sources in the USA with high levels (>10mg/L), which leads to a frequent route of exposure through dietary uptake of seafood [255, 256]. A correlation has been found between arsenic exposure and tumors in the kidneys, lungs, liver, and bladder [257, 258].

HRMS-based metabolic profiling showed that chronic As exposure through drinking water caused disruptions in lipids and amino acid metabolism in rat serum, which had a positive correlation with up-regulation of the hepatic genes including cpt2, lcat, cact, crot, and mtr [259]. A recent untargeted metabolomics study performed by Khanam et al. investigated the toxic impacts of As exposure on Pakistani male urinary metabolome [260]. The results showed that As caused oxidative stress, disrupted one-carbon, purine, and caffeine metabolism. These tests showed 38 potential metabolite biomarkers of As exposure, including xanthines, purines, and testosterone. Lu et al. found that gut microbiome was a key player participating in the energy metabolism of As-treated mice, through investigating the association between gut microbial composition alterations and the typical gut microflora–related metabolites (indole-containing compounds, isoflavone metabolites, and BAs [261]. Xue et al. discovered a high correlation between serum metabolic profiles and gut microbiome perturbation in As-exposed Helicobacter-free mice [262]. Dramatic changes in numerous pathways were observed, including fatty acid, lipid, and tryptophan metabolism, and the serum metabolic disorder was attenuated in mice with healthy gut microbiota. Using untargeted metabolomics and lipidomics approaches, Wang et al. confirmed the important role of gut microbiome in As-inflicted neurodegeneration in the mouse brain [263]. Potential metabolite biomarkers of neurodegeneration, including 118 polar metabolites and 17 lipids involved in 30 metabolic pathways (fatty acid, lipid, amino acid, glycolysis, and nucleic acid metabolism), were significantly changed after the exposure, which was correlated with 12 kinds of gut microbes. In another recent publication from Luo et al., a significant microbial detoxification was discovered in situ in As-exposed mice after fecal microbiome transplantation (FMT) [133]. After the gut transplant, As accumulation was lower in fecal, liver, and plasma. Metabolites potentially beneficial to the host were higher in feces, plasma, and cerebral cortex in the FMT recipients than non-FMT mice. This suggests that FMT could be a promising treatment for As exposure.

5.2. Cadmium (Cd)

Cd is a metal element in the earth crust and exists in many industrial products such as batteries, alloys, electroplates, solar cells, plastic stabilizers, and pigments [264, 265]. Cd pollution also happens in industrial zones in developing countries [24]. The toxicity of Cd involves depletion of reduced glutathione, binding sulfhydryl groups with proteins, causing production of ROS, and potential genetic mutation and chromosomal deletions [248, 257, 265].

Recently, an untargeted metabolomics study found 76 altered metabolites mainly involved in organic acid, nucleoside, and lipid metabolism in vitro, and 14 metabolites co-existed in the urine of Cd-exposed workers were selected as potential biomarkers for Cd-exposure [266]. Using a mouse model, Hudson et al. employed a multi-omics approach to investigate the connections among maternal Cd exposure, impaired neurodevelopment in newborn brains, and behavior alterations at adulthood [267]. Maternal Cd exposure reduced mitochondrial DNA levels in newborn brains and disturbed hypoxic responses, cellular energy pathways, and retinoic acid signaling. MS measurement of 101 identified compounds showed that the metabolites in cellular energy pathways and hypoxia, such as retinoids, were significantly altered after the exposure. Zhang et al. investigated how Cd exposure impacts the gut microbiome and the host in an AD model (ApoE4-KI mice) [268]. They examined liver transcriptome (liver RNA-seq), gut microbiome (fecal 16S rRNA), and sera SCFAs concentration (GC-MS targeted quantification). The results showed that Cd exposure indeed increased microbial AD biomarkers while down-regulating energy supply-related pathways in the gut. Cd induced inflammation and disturbed xenobiotic biotransformation in the host liver. Cd exposure changed SCFAs concentrations in the mouse serum, which was correlated with gut microbiome composition. Li et al. combined metabolic profiles with 16S rRNA sequencing results, and they confirmed that Cd exposure caused significant changes in BA and amino acid metabolism and induced decreased microbial species, including Blautia, Eisenbergiella, Clostridium_XlVa, etc. [269]. Rothman et al. identified that metabolites involved in detoxification, proteolysis, and lipolysis were increased in Cd-exposed honeybees, and these alterations may cause oxidative damage to lipids, proteins, and detoxification genes [270]. In addition, they discovered seven kinds of bee-associated strains (such as L. micheneri, L. quenuiae, and L. kunkeei) that could be used to protect bees from Cd exposure.

5.3. Other heavy metals

Manganese (Mn) is a naturally occurring mineral metal that can exist in nuts and vegetables [271]. Mn is an essential micronutrient element for human health, but it is toxic at high physiological levels [272]. Overexposure of Mn occurs via contaminated water, welding fumes, or pesticides [273]. Mn toxicity includes neurodegeneration, neuroinflammation, spasticity, tremor, dystonia, and bradykinesia [274]. Mn-exposed mice were investigated by Chi et al., and they found that the gut microbiome composition and bacterial genes related to tryptophan, GABA metabolism, and LPS synthesis where DNA repair was significantly altered after the exposure [275]. GC-MS metabolic profiling showed that neurotransmitters and their precursors such as phenylalanine, glycine, and glutamic acid were significantly changed. Wang et al. found that bioaccumulated Mn increased β-amyloid (Aβ), receptor-interacting protein kinase 3 (RIP3), and caspase-3 production in the brain [276]. Mn significantly altered the metabolites involved in β-hydroxypyruvic acid, tryptamine, taurodeoxycholic acid, and urocanic acid metabolism. Mn exposure also reduced the gut microbiota richness, especially the composition of Prevotellaceae, Fusobacteriaceae, and Lactobacillaceae. Interestingly, Mn-induced neurotoxicity was alleviated with the recovery of gut microbiome composition by FMT from normal rats, which supplies a novel therapeutic strategy for the treatment of Mn toxicity by remodeling the gut microbiota.

Lead (Pb) is also classified as a heavy metal toxicant, which is found all throughout the environment in the water, soil, and air [271, 277]. This heavy metal is a known pollutant and can also be found in dietary and marine products, and most known to be found in old paint cans [278, 279]. Exposure to Pb can lead to multiple illnesses, such as gut microbiome dysbiosis, central nervous system disorders, inflammation, and liver toxicity [257, 277, 280]. Zeng et al. showed a high blood and urinary Pb level in the children living in electronic-waste areas [277]. Pb exposure reduced pediatric development parameters and altered 58 kinds of gut microbial genera. Through GC-MS metabolic profiling, 19 metabolites showed significant correlation with Pb exposure.

6. Inhalation Toxicant Exposure, Gut Microbiome, and Metabolomics

Inhalation xenobiotics are a public health emergency, including wildfire smoke (WFS), ambient air pollution (AAP) such as particulate matters with the aerodynamic diameter < 2.5 μm (PM2.5) or < 10 μm (PM10), traffic-related air pollution (TRAP), and third hand smoking (THS) [281, 282]. These inhalation exposures may be linked to a series of health problems, including neurodegeneration, hypertension, diabetes, asthma, and pulmonary diseases [283-285]. Cigarette smoking and THS exposure have a strong capability to shift the cecal microbial taxa diversity, and interestingly, they also significantly increased glycolysis and pyruvate decarboxylation, as well as significantly decreased coenzyme A biosynthesis and pyrimidine deoxyribonucleoside salvage [286-289]. WFS is an increasing global concern, although there are limited studies about WFS-exposed gut microbiome and metabolic profiles. To determine the effects of WFS on neuroinflammation and neuro-metabolomics in mice, Scieszka et al. performed a multi-omics study in a mobile laboratory, >300 km away from massive wildfires burning in California [290]. As a result, anti-aging metabolites, such as NAD⁺ and taurine, were decreased and the amyloid-beta protein (a key protein in neurodegeneration) was up-regulated by WFS exposure.

Multiple epidemiological studies have shown that exposure to AAP is associated with dysregulated composition of the gut microbiome. Bailey et al. found that AAP induced gut microbiota alterations in 6-month infants using 16S rRNA amplicon sequencing [291]. PM2.5 increased Actinomyces, which was negatively correlated with Alistipes. PM10 exposure was positively correlated with systemic inflammation related Dialister and Dorea. NO2 exposure was positively associated with Actinomyces, Enterococcus, Clostridium, and Eubacterium. In addition, a recent study showed that PM2.5 induced body weight loss, liver injury, and intestinal barrier disorder in adult mice [292]. The ratio of Bacteroidetes/Firmicutes of gut flora was associated with arachidonic acid metabolism and linoleic acid metabolism in the mice ileum. Freeway TRAP exposure has negative impacts on gut microbial taxa in overweight and obese adolescents. Liang et al. and Li et al. discovered that over thousands of metabolites and metabolic features were significantly changed in the serum samples of TRAP-exposed participants [293, 294]. The influenced metabolites were mainly involved in the oxidative stress-related pathways, nutrient metabolism (e.g., fatty acid metabolism), acute inflammation (e.g., histidine metabolism and tyrosine metabolism), leukotriene and vitamin E metabolism, and nucleic acids damage/repair (e.g., pyrimidine metabolism).

7. Conclusion

Toxic xenobiotic exposure is an increasing global health concern. Gut is a primary site for many xenobiotics in the host, and it was hypothesized that gut microbiota plays a vital role in xenobiotic toxicological effects. Metabolites can be signaling molecules responsible for the interaction between the gut microbiome and host; therefore, metabolomics is a promising approach for the investigation of metabolic effects due to xenobiotic exposure. This review summarized recent advancements in ultramodern MS-based metabolomics platforms and recent applications of metabolomics in xenobiotic exposure and gut microbiome studies. Based on the literature in the past few years, research into gut microbiomes, metabolomics, and their combination are effective to identify metabolic and microbial markers of xenobiotic exposure and determine its mechanisms. Importantly, there is increasing evidence showing that reprogramming the gut microbiome through probiotics, prebiotics, and/or FMT could be a promising approach to treat or reduce xenobiotic toxicity.

Highlights.

• Applications of metabolomics combined with other omics platforms in xenobiotic metabolism and gut microbiome are reviewed.

• The central hypothesis is that the interaction between gut microbiome and the host is mainly through metabolites.

• Recent advancements in mass spectrometry-based metabolomics and metabolic flux analysis are summarized.

• The toxicity of a wide variety of xenobiotics and potential therapeutic approaches through reprograming the gut microbiome are discussed.

Abbreviations:

MS

mass spectrometry

GLA

gut-liver axis

GBA

gut-brain axis

GPA

gut-pancreas axis

GKA

gut-kidney axis

BAs

bile acids

SCFAs

short chain fatty acids

Trp

tryptophan

TMAO

trimethylamine N-oxide

BCAA

branched-chain amino acids

BAs

bile acids

GPR

G-protein-coupled receptor

BBB

blood-brain barrier

GABA

γ-aminobutyric acid

IPA

Indole-3-propionic acid

NAD⁺

nicotinamide adenine dinucleotide

LC-MS

liquid chromatography-mass spectrometry

GC-MS

gas chromatography-mass spectrometry

HRMS

high-resolution MS detectors

Tof

time-of-flight

QC

quality control

SIL-ISs

stable isotope labeled internal standards

MeOH

methanol

ACN

acetonitrile

ESI

electrospray ionization

MTBE

methyl tert-butyl ether

TQMS

quadrupole mass spectrometry

SIM

selective ion monitoring mode

MRM

multiple reaction monitoring

PRM

parallel reaction monitoring

LLOQ

lower limit of quantification

ULOQ

upper limit of quantification

HILIC

hydrophilic interaction chromatography

RT

retention time

CE

collision energy

GOT-MS

globally optimized targeted mass spectrometry

dGOT-MS

database-assisted globally optimized targeted mass spectrometry

PCB

polychlorinated biphenyls

MFA

Metabolic flux analysis

SIRM

stable isotope-resolved metabolomics

PCA

principal component analysis

POP

Persistent Organic Pollutant

PBDEs

polybrominated diphenyl ethers

TBBPA

tetrabromobisphenol A

OUT

operational taxonomic unit

PCBs

polychlorinated biphenyls

UFA

unsaturated fatty acids

MUFA

monounsaturated fatty acids

SFA

saturated fatty acids

PC

phosphatidylcholine

PE

phosphatidylethanolamine

PFASs

Per- and polyfluoroalkyl substances

PFOS

perfluorooctane sulfonate

PFOA

perfluorooctanoic acid

As

arsenic

Cd

cadmium

Mn

manganese

Pb

lead

FMT

fecal microbiome transplantation

WFS

wildfire smoke

AAP

ambient air pollution

TRAP

traffic-related air pollution

THS

third hand smoking