Application of metabolomics to characterize environmental pollutant toxicity and disease risks

Abstract

The increased incidence of non-communicable human diseases may be attributed, at least partially, to exposures to toxic chemicals such as persistent organic pollutants, air pollutants, and heavy metals. Given the high mortality and morbidity of pollutant exposure associated diseases, a better understanding of related mechanisms of toxicity and impacts on endogenous host metabolism is needed. The metabolome represents the collection of the intermediates and end products of cellular processes, and is the most proximal reporter of the body’s response to environmental exposures and pathological processes. Metabolomics is a powerful tool for studying how organisms interact with their environment and how these interactions shape diseases related to pollutant exposure. This mini review discusses potential biological mechanisms that link pollutant exposure to metabolic disturbances and chronic human diseases, with a focus on recent studies that demonstrate the application of metabolomics as a tool to elucidate biochemical modes of actions of various environmental pollutants. In addition, classes of metabolites that have been shown to be modulated by multiple environmental pollutants will be discussed with an emphasis on their use as potential early biomarkers of disease risks. Taken together, metabolomics is a useful and versatile tool for characterizing the disease risks and mechanisms associated with various environmental pollutants.

Introduction

Most non-communicable disease risk is determined by both genetic and a broad range of environmental factors. In addition to lifestyle-dependent factors, such as diet, exercise and other behavioral choices, exposure to toxic environmental chemicals is one example of environmental factors that can modify disease outcome [1–3]. The pathology of diseases such as cardiovascular, neurodegenerative, immunological, respiratory disorders and cancers can be modified or accelerated due to exposure to toxic pro-inflammatory chemicals. Other lifestyle and environmental factors that can contribute to these diseases include diet, tobacco smoking, exposures to radiation, pathogens, allergens and psychological stress [4]. Although genetic factors are a significant determinant of the risk of these chronic diseases, exposure to toxic environmental chemicals is recognized as a major independent risk factor [5–7]. As industrialization increases worldwide, it is expected that exposure risks may increase which has become a global health challenge. The World Health Organization (WHO) estimates that environmental exposures accounted for 12.6 million deaths worldwide and 22% of the global burden of disease in 2012 [8].

Metabolomics is the field of science that characterizes endogenous and exogenous metabolites within a cell, tissue, or biofluid of an organism in response to external stressors such as disease, contaminant exposure, or nutritional imbalances [9]. Nuclear magnetic resonance (NMR) and mass spectrometry (MS) are the two major analytical tools that are routinely used to obtain metabolomics data sets [10]. Metabolomic analyses can be categorized as either non-targeted or targeted with non-targeted metabolomics detecting as many distinct features as possible in a single analysis. In contrast, targeted metabolomics analyzes a specific group, class, or species of metabolites and is especially useful for absolute quantitation. As the pathways and networks of metabolites have been extensively studied (Figure 1), changes in metabolites can simultaneously reveal variations in enzymatic levels, activities and/or gene expression patterns.

Figure 1. Systematic workflow of metabolomic studies.

Biological samples from cells, animal and/or human bio-fluids or tissues are extracted for NMR or MS data acquisition. Bioinformatic analysis, or data processing, including feature finding, metabolite identification and pathway analysis, can then be used to identify changes in cellular metabolism, and ultimately elucidate the biochemical/molecular mechanisms underpinning disease pathogenesis.

Metabolomics allows researchers to assess the cellular response to external stimuli and can help to improve the understanding of mode-of-action of a given compound [11]. Metabolomic analysis of genetically modified animal models is a powerful tool for understanding downstream ramifications of targeted gene dysfunction [12–14]. For example, Ahr-knockout male mice were used in a metabolomic study of 2,3,7,8-tetrachlorodibenzofuran (TCDF), and it was found that ingestion of TCDF altered hepatic lipogenesis, gluconeogenesis, and glycogenolysis in an AHR-dependent manner [14]. Additionally, animal disease models were used in metabolomic studies to investigate mechanisms of pollutant related pathology process. For example, a Helicobacter-infected mouse model was used to study the role of gut microbiome perturbation in arsenic (As) induced disease. The results suggested that gut microbiome perturbation can exacerbate metabolic disorders induced by As exposure[15]. It should be noted that although gender differences in tissue distribution and toxicity have been reported for many environmental pollutants [16–18], there are few studies exploring pollutants-induced metabolome changes in female animal models. The application of metabolomic technologies in epidemiological studies is a growing research area with the potential to better characterize human exposures, detect early markers of disease, understand disease etiology, improve diagnosis of disease, and track disease progression [19]. Table 1, and described in the following sections of this mini-review, provides an overview of metabolomic-based epidemiological studies for some pollutant classes. “Meet-in-the-middle” (MITM) approaches involve a prospective search for intermediate biomarkers, which are elevated in subjects who eventually develop disease and a retrospective search for links of such biomarkers to past environmental exposures [20]. This approach has been conceived as a strategy to identify biomarkers that are not only related to specific exposures but also predictive of disease outcome, and it is gaining popularity for the ability to reveal important linkages along the exposure-outcome pathway [21, 22].

Table 1.

Metabolomic study in humans in response to environmental pollutants

Pollutant	Study population	Sample	Analytical method	Metabolic response	Reference
Biphenol A	Korean Research project on the Integrated Exposure Assessment of Hazardous Substances for Food Safety (KRIEFS)	Urine	Non-targeted, UHPLC-Q/TOF MS	Significant disturbance in fatty acid elongation and sphingolipid metabolism. Females in 40’s showed elevated inflammatory metabolites: 6-ketoprostaglandin E1 and thromboxane.	[86]
Dioxin	Czech chemical workers	Urine	Non-targeted, UHPLC-Q/TOF MS	Glucuro- and sulfo-conjugated endogenous steroid metabolites and bile acids were assessed as biomarkers of acute dioxin toxicity.	[39]
PAHs	Exposed participants were from a polluted rural area less than 2 km downwind of a large coking plant in China	Urine	Non-targeted, UHPLC-Q/TOF MS	↑Pyroglutamic acid, 3-Methylhistidine, azelaic acid, decenedioic acid, hydroxytetradecanedioic acid, medium-chain acylcarnitines, decenediolyglucuronide ↓Uric acid	[87]
PFAS	Overweight and obese Hispanic children (8–14 years) from urban Los Angeles	Plasma	Non-targeted, UHPLC-Q Exactive MS	Significant alterations of lipids (e.g., glycosphingolipids, linoleic acid, and de novo lipogenesis), and amino acids (e.g., aspartate and asparagine, tyrosine, arginine and proline)	[42]
Air Pollution	Two randomized crossover trials were used including the Oxford Street II (London) and the TAPAS II (Barcelona) studies	Serum	Non-targeted, UHPLC-Q/TOF MS	17 and 30 metabolic compounds were changed in the Oxford Street II (London) and the TAPAS II (Barcelona) respectively	[64]
Ozone	Healthy volunteers were exposed to filtered air and ozone	Bronchoalveolar lavage fluid	Non-targeted, UHPLC-MS	Total of 28 and 41 metabolites were differentially expressed after 1-hour and 24-hour post-exposure respectively.	[88]
PM 2.5	Healthy volunteers were exposed to PM 2.5	Serum	Non-targeted, GC-MS, UHPLC-Q/TOF MS	↑Cortisol, cortisone, epinephrine, norepinephrine, phenylalanine, tyrosine, tetrahydropteridine, L-Tryptophan, N-Acetylserotonin, melatonin ↓Serotonin	[66]
Benzene	Painting workers from China	Plasma	Non-targeted and targeted, HPLC-Q/TOF MS	↑Pyroglutamic acid, glycoursodeoxycholic, linolenyl carnitine, lysoPC(16:1), Bilirubin, LysoPC(18:4/0:0), LysoPC (P-18:0), LysoPC (20:2) ↓L-camitine	[69]
Traffic pollution	Participants in the Dorm Room Inhalation to Vehicle Emission (DRIVE) study	Plasma, saliva	Non-targeted, UHPLC-Q Exactive MS	Biological perturbations associated with traffic pollutants included metabolic signaling related to oxidative stress, inflammation, and nucleic acid damage and repair.	[68]
Arsenic	Health Effects of Arsenic Longitudinal Study (HEALS)	Urine	Non-targeted, GC-MS	31 molecular features and urinary arsenic was significantly related to 74 molecular features. Six metabolites were significantly associated with either water arsenic or urinary arsenic, including 1,2-dithiane-4,5-diol, L-Threonine, phosphoricacid, pyroglutamic acid, (R*, S*)-3,4-Dihydroxybutanoic acid, and succinic acid.	[82]
Cadmium	Participants from south China with high Cd exposure	Urine	Non-targeted, GC-MS	27 known metabolites involved in glucose, amino acid, energy, fatty acid, bone metabolism and TCA cycle were markedly changed	[79]
Lead	Residents in a village near a used lead-acid battery (ULAB) recycling facility in Japan	Urine	Targeted, LC-QTRAP MS	Ten candidate biomarkers were associated with ATP-binding cassette (ABC) transporters, the disruption of small-molecule transport in the kidney; amino acid, porphyrin, and chlorophyll metabolism; and the heme biosynthetic pathway.	[81]

Metabolomics is viewed as a functional readout of other omics techniques. For example, as organisms respond to external stressors, resulting changes to gene expression and protein production occur. These expression-related changes are subjected to a variety of homeostatic controls and feedback mechanisms, which could ultimately be reflected on the endpoint of metabolite formation. This may result in metabolomics being a sensitive indicator of the external stressors compared with other omics technologies. The Human Metabolome Database 4.0 (www.hmdb.ca) lists 114,100 metabolites [23], but the actual number of metabolites could be higher. As analytical techniques and detection sensitivity continue to improve, more metabolites will be identified. Metabolomics has potential as a sensitive technique that is capable of characterizing biological changes within a cell, tissue, or biofluid of an organism in response to environmental pollutant exposure.

In this mini review, we summarize recent metabolomics studies focusing on the association between human diseases and exposure to environmental pollutants, which are grouped based on chemical-physical properties, i.e., persistent organic pollutants, air pollutants, and heavy metals. Identification of novel mechanisms of toxicity using metabolomics are also discussed.

Metabolomics and persistent organic pollutants

Persistent organic pollutants (POPs) are highly lipophilic compounds that are resistant to degradation and tend to bioaccumulate in the environment, food chains and living organisms. Examples of these compounds include dioxins, dichlorodiphenyltrichloroethane (DDT), polychlorinated biphenyls (PCBs), phthalates such as bisphenol A (BPA), and the perfluorinated and polyfluorinated alkyl substances (PFAS). The major route of exposure to POPs is ingestion, which could occur through the consumption of animal derived foods, or through contaminated drinking water In vivo and/or in vitro experiments have shown that even at the low levels reported for non-occupational exposures, these compounds exhibit teratogenicity, immunotoxicity, endocrine toxicity and carcinogenicity [24]. Recent studies show that exposure to POPs have been associated with cardiometabolic risk factors [25, 26]. Epidemiological studies suggest that POPs may act as metabolic disruptors contributing to the growing incidence of metabolic diseases [27, 28].

Metabolomics studies of POPs enables hypothesis generation for hitherto unexplained responses to environmental stressors and thereby helps unravel the underlying mode of action of POPs. Coincident observations across mechanism studies using animal and cell models include impaired gluconeogenesis, altered glucose production and lipid metabolism [29–31]. Our laboratory has previously demonstrated that exposure to PCB 126, the most potent dioxin-like PCB, can increase peripheral inflammation in a liver injury mouse model [32]. Metabolites in mice livers were profiled using non-targeted metabolomics method to elucidate possible associations between liver injury and PCB 126 promoted cardiometabolic disease risk. The results demonstrated that PCB 126 altered the liver metabolome and aggravated redox stress, which may predispose to cardiometabolic complications [33]. Other studies employed metabolomic techniques to investigate host-microbiota interactions and found that POPs disrupted gut microbiota and host metabolism [14, 34, 35]. For example, Zhang et al. investigated the effect of 2,3,7,8-tetrachlorodibenzofuran (TCDF) on gut microbiota and host metabolism using NMR and targeted LC-MS [14]. They found that TCDF altered gut microbiota composition and this was associated with altered bile acid metabolism. Additionally, TCDF triggered host metabolic disorders as a result of altered hepatic lipogenesis, gluconeogenesis, and glycogenolysis [14]. Using non-targeted metabolomics, it was found that PCB 126 disrupted the gut microbiota and host metabolism in mice, which imply that the toxic effects of PCB 126 may be initiated in the gut, and the modulation of gut microbiota could be a sensitive marker of PCB 126 exposures [34]. A follow-up study demonstrated that fecal microbe exposed to PCB 126 resulted in reduced succinic acid production, but increased propionate production, both of which can influence host glucose and lipid metabolism [36]. These findings provide new insights into the biochemical consequences of POPs exposure involving the alteration of the gut microbiota and disruption of host metabolism.

Metabolomic-based epidemiologic studies help understand how environmental pollutants disturb metabolism and help to identify biomarkers related to POPs exposure. For example, Carrizo et al. [37] reported a non-targeted metabolomic analysis of human serum samples associated with exposure levels of POPs and found that sphingolipids and glycerophospholipid lipids were significantly different between high and low POPs exposure groups. This study indicated an imbalance in human lipid metabolism upon POPs exposure, and these observed metabolomic fingerprints may have the potential to be classified as biomarkers of POPs exposure. Jeanneret et al. [38, 39] identified human urinary markers of acute dioxin exposure using non-targeted metabolomic approaches, including glucuro- and sulfo-conjugated endogenous steroids and bile acids. Zbucka-Kretowska et al. [40] revealed that increased level of BPA in the maternal plasma was related to higher level of several endocannabinoids using a non-targeted metabolomic analysis. Further studies demonstrated that BPA inhibited fatty acid amide hydrolase activity and consequently led to a rise of plasma endocannabinoids. These studies provided a novel toxicity mechanism of BPA, which may induce adverse pregnancy outcomes by acting on the endocannabinoid system. PFAS exposure and plasma metabolomic studies demonstrated that higher PFAS exposure in adults and children was associated with dysregulated lipid and amino acid metabolism [41–43]. Perturbation of lipid and amino acid pathways have been found to be strongly associated with the risk of developing metabolic diseases including type 2 diabetes [44, 45]. Collectively, these studies indicate that PFAS exposure might cause disturbances in lipid and amino acid metabolism, thereby contributing to increased risk of metabolic diseases.

Metabolomics and air pollutants

Air pollution is a complex mixture of gaseous, volatile, semi-volatile and particulate matter, and its exact composition varies widely. The major route of exposure to air pollution is by inhalation. Some of the commonly measured components of the air pollution mixtures include irritant gases, benzene, and particulate matter (PM) [46]. Additionally, various harmful substances are adsorbed or contained in PM2.5, such as heavy metals, polycyclic aromatic hydrocarbons and POPs. Air pollution is the world’s largest environmental threat to health and is responsible each year for an estimated 6.5 million deaths [47, 48]. A growing number of epidemiological studies showed that air pollution accounts for 19% of all cardiovascular deaths, 23% of all ischemic heart disease deaths, and 21% of all stroke deaths [49]. Air pollution has also been linked to cancers [50], neurologic [51], and respiratory diseases [52]. Furthermore, air pollution effects the immune system and is associated with allergic rhinitis, allergic sensitization, and autoimmunity [53, 54]. The most well established mechanism is oxidative stress caused by air pollutants, followed by pulmonary and systemic inflammation [55].

Metabolomics has been increasingly applied to investigate associations between long- and short-term exposure to air pollution and metabolome changes. Non-targeted and targeted metabolomics strategy were adopted to characterize the overall metabolic changes and relevant toxicological pathways in animal serum and urine [56–62]. Purine, amino acid, and lipid metabolism are common pathways that are disturbed after air pollutant exposure [56–62]. A study focusing on the pulmonary metabolome in rats demonstrated that phospholipid and sphingolipid metabolism were altered after PM2.5 exposure, which may relate to PM2.5-induced pulmonary inflammation and injury [58]. The detrimental effect of air pollution to the human body is closely associated with lipid metabolism, and disturbed plasma lipid profiles have been observed in several cohorts with exposure to various air pollutants [63, 64]. Results from a non-targeted metabolomic study of three cohorts, each a part of the Cooperative Health Research in the Region of Augsburg, Germany, revealed that changes in lysophosphatidylcholines (LPCs) in human plasma were associated with short-term exposure to air pollutants [64]. This result was in agreement with the findings in animal and cell models, in which LPCs were increased in air pollutant exposed rats and alveolar type II cells compared with corresponding control groups [65]. The researchers further determined that the observed increase may have been due to air pollutant-induced activation of phospholipase A2, which catalyzes the hydrolysis of phosphatidylcholines to LPC [65]. Jeong et al. [21] applied MITM non-targeted metabolomics approach to identify metabolites and pathways that could be associated with both air pollutants and compromised health outcomes using human serum samples. It was found that linoleate metabolism and glycerophospholipid metabolism were common MITM pathways linking ultra-fine particles (UFP) exposure to asthma; glycosphingolipid metabolism linking UFP exposure to cardiovascular disease; and metabolites related to carnitine shuttling linking NO₂ exposure to cardio-cerebrovascular disorders. The last association was also observed in a recent short-term traffic-related air pollution study [64]. Li et al. [66] reported a randomized, double-blind, crossover trial of metabolomics analysis on particulate air pollution. The results demonstrated that higher PM exposure led to a significant increase in serum levels of stress hormones, suggesting activations of the hypothalamus-pituitary-adrenal and sympathetic-adrenal-medullary axes, which may contribute to adverse cardiovascular and metabolic effects. Other reported metabolic perturbations associated with air pollution exposure are dysregulated inflammatory and oxidative stress related pathways, including leukotriene, vitamin E, and glutathione metabolism [67–69]. These altered pathways could contribute to the tissue damage and systemic inflammation observed in response to air pollutant exposure.

Metabolomics and heavy metals

Heavy metals, such as As, cadmium (Cd), mercury (Hg), copper (Cu), lead (Pb), and manganese (Mn), are ubiquitous and are increasingly introduced into the food chain due to industrial and agricultural practices. For the general population, diet and drinking water are the main sources of heavy metal exposure. Heavy metal toxicity results in damage of the lungs, kidneys, liver and other organs [70]. Additionally, heavy metals can enter the brain through the disruption of blood-brain barrier, subsequently damaging the nervous system [71]. For example, Cd-dependent neurotoxicity has been reported to be related to Alzheimer’s and Parkinson’s diseases, however, the molecular mechanisms of Cd toxicity remain to be clarified [72]. The neurotoxicity of Cd has been investigated in neuronal PC-12 cells by a non-targeted metabolomic approach [73]. The results demonstrated that Cd exposure modulated energy metabolism (glycolysis, TCA cycle and fatty acid β-oxidation), cell membrane composition, as well as the anti-oxidant defense system. Using metabolomic analysis of urine samples, it was demonstrated that Cd disrupted energy and lipid metabolism, as well as induced oxidative stress, which contributed to tissue damage [74, 75]. Chronic exposure of inorganic As has been associated with increased risk for numerous metabolic syndromes [76]. After ingestion, inorganic As is metabolized through methylation and reduction reactions which ultimately produce methylated metabolites, such as monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA). Inter-individual differences in methylation capacity have been associated with risk of health outcomes [77]. Spratlen et al. [78] explored the mechanism that may explain the associations between As metabolism and metabolic outcomes using targeted metabolomic analysis of human plasma. It was found that one carbon metabolism pathway was significantly related with both As metabolism and diabetes-related outcomes. Additionally, association of lower MMA% and higher DMA% with diabetes-related outcomes may be influenced by one carbon metabolism status. Metabolomics has been applied to the identification of biomarkers for heavy metal exposure using human urine samples [79–82] (Table 1). For example, Wu et al.[82] investigated associations between reproducible urinary metabolites and As exposure in Bangladesh Adults using non-targeted metabolomics. Six metabolites (1,2-dithiane-4,5-diol, l-threonine, phosphoric acid, pyroglutamic acid, (R*,S*)-3,4-dihydroxybutanoic acid, and succinic acid) were identified as significantly associated with either water concentrations of As or urinary As after adjustment for multiple comparisons. Xu et al. [79] identified urine metabolites in women exposed to high Cd levels using targeted metabolomics method. The results demonstrated that the abundance of 48 metabolites were markedly changed in subjects with high Cd levels exposure, among which 8 endogenous metabolites were identified as potential early candidate biomarkers. Overall, these results provided new insights into the mechanism of heavy metal toxicity and identification of biomarkers associated with exposure.

Conclusions and future perspectives

In this mini-review, we have highlighted the increasing potential of metabolomics for assessing disease risks of environmental pollutants. These studies have enhanced our understanding on how environmental pollutant exposure can alter biological mechanisms and shed light on disease etiology. However, there are still many challenges. For example, human exposure data (including pollutants, lifestyle factors and behaviors) are often limited or incomplete, and the estimation of associated disease risks can be severely hindered. The exposome, which measures all the exposures of an individual over a timeframe, such as a lifetime, has recently been proposed as a new paradigm to encompass the totality of human environmental [83]. Additionally, some metabolites do respond sensitively to environmental stressors, such as metabolites within glycolysis and the TCA cycle, however, many of these metabolites are not specific enough to be used for diagnostic purposes. With the advance of multi-omics technologies, a more comprehensive understanding of disease risks and consequently promising biomarkers that are unique to specific pollutant exposure and diseases are becoming possible. Advancements in next generation sequencing in recent years has allowed detailed genomics, transcriptomics, and bioinformatic analyses of altered biological pathways related to pollutant exposure [84, 85]. These technologies, combined with metabolomics, will provide unprecedented insights into key pathways that connect pollutant exposure to disease risks and pathological endpoints.