**1. Introduction**

The evaluation of biological responses to assess and predict the impact of environmental changes in ecosystems functioning is receiving increasing attention, and current research focuses on overcoming concerns about the specificity of biomarkers (Amiard-Triquet et al., 2012). Generally, biomarkers are chemicals, metabolites, susceptibility characteristics, or physiological changes that relate to the exposure of an organism to a chemical. Accordingly, a selected biomarker, i.e. biological response, can be linked to a specific environmental exposure, being representative of the health status of the ecosystem studied. The identification of biomarker profiles has been possible upon the development of metabolomics. Those profiles allow the genuine identification of the relevant biological response/s associated to a particular exposure while the assessment of a single biomarker could only estimate the potential response of the ecosystem to a particular pollutant.

Metabolomics is the generic name assigned to a scientific field that addresses the characteri‐ zation of low molecular weight organic metabolites released by living organisms in response to environmental stimuli. Morrison et al. (2007) provided an extended definition ''the appli‐ cation of metabolomics to the investigation of both free-living organisms obtained directly from the natural environment (whether studied in that environment or transferred to a laboratory for further experimentation) and of organisms reared under laboratory conditions (whether studied in the laboratory or transferred to the environment for further experimen‐ tation), where any laboratory experiments specifically serve to mimic scenarios encountered in the natural environment''.

The methodological approach of metabolomics relies on a comprehensive analysis of the set of metabolites or "metabolome" produced in response to particular environmental stimuli. Accordingly, the metabolome is the pool of metabolites, small molecules, within a cell, tissue,

© 2014 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

organ, biological fluid, or entire organism (Miller, 2007). Exposure of an organism to an external stressor will result in changes at the level of the metabolome (Ankley et al., 2006; van Ravenzwaay et al., 2007), and such changes may constitute a highly sensitive indicator of an external stress. Therefore, metabolomics has potential as a sensitive and rapid technique that can elucidate the relationships between metabolite levels and an external stressor, such as contaminant exposure, nutritional deficit or a disease.

The main advantage of metabolomics over traditional research is to overcome the bias associated to the assessment of predefined metabolites (Singh, 2006). Among the diverse applications of omic profiling methods to the environmental sciences, ecotoxigenomics addresses the response of organisms to pollutants based on the different sensitivity of species to toxicants (Spurgeon et al., 2008). Currently, the implementation of metabolomics in ecological risk assessment is still at an early stage, mostly applied as a screening tool to assess the potential toxic effects of pollutants or to determine the mode of action (MOA) of a toxicant. Otherwise, application of metabolomics for environmental monitoring allows the study of a large variety of species from relatively uncontrolled environments.

Bundy et al. (2009) highlight the challenge of identifying a large number of metabolites and the necessity of creating metabolite databases specifically dedicated to environmental issues. Multivariate statistical analysis has proved highly effective for metabolite identification. Thus, principal component analysis (PCA) is used to identify differences between metabol‐ ic profiles of organisms exposed to organic or inorganic pollutants (Jones et al., 2014; Kwon et al., 2012; Lankadurai et al., 2011). Besides, association between the metabolic profile and biological factors evaluated as markers for exposure to pollutants can be modelled by partial least squares (PLS) regression analysis (Ellis et al., 2012).

The implementation of metabolomics for the assessment of soil contamination is neverthe‐ less at an early stage (Viant, 2009). A basic screening of published research in the web of science returns circa 100 items for the search "soil pollution-metabolomics", with a significant launch in 2011 (Figure 1), reduced to 21 records when the search is narrowed with the term "biomarkers", published in the period 2007-2013. However, emerging regulatory challenges demand the advance of toxicity testing. Toxicogenomics tools have been presented as an advanced from the current methodologies used for regulatory decision making in ecotoxicology, which entirely rely on whole animal exposures and adverse effects on survival, growth, and reproduction (Ankley et al., 2006). From the acquisition of reproducible metabolic profiles as response to the presence of specific pollutants in soil (Jones et al., 2008) to the application of metabolomics techniques to the study of the response of the entire community of a soil to factors such as pollution and climate change (Jones et al., 2014), the implementation of metabolomics in ecotoxicology is a sound answer to the current needs of society and the environment (van Ravenzwaay et al., 2012).

During the last decade a number of general revisions about the application of metabolo‐ mics in environmental health assessment have been published (Bundy et al., 2009; Miller, 2007; Snape et al., 2004; van Ravenzwaay et al., 2007; Viant et al., 2003). The present review specifically summarizes the most significant research concerning implementation of

identify differences between metabolic profiles of organisms exposed to organic or inorganic pollutants (Jones et al., 2014; Kwon et al., 2012; Lankadurai et al., 2011). Besides, association between the metabolic profile and biological factors evaluated as markers for exposure to pollutants can be modelled by partial least squares (PLS)

The implementation of metabolomics for the assessment of soil contamination is nevertheless at an early stage (Viant, 2009). A basic screening of published research in the web of science returns circa 100 items for the search "soil pollution-metabolomics", with a significant launch in 2011 (Figure 1), reduced to 21 records when the search is narrowed with the term "biomarkers", published in the period 2007-2013. However, emerging regulatory challenges demand the advance of toxicity testing. Toxicogenomics tools have been presented as an advanced from the current methodologies used for regulatory decision making in ecotoxicology, which entirely rely on whole animal exposures and adverse effects on survival, growth, and reproduction (Ankley et al., 2006). From the acquisition of reproducible metabolic profiles as response to the presence of specific pollutants in soil

Figure 1. Citation report for the search "soil pollution-metabolomics" as obtained from Thomson Reuters (January 2014). During the last decade a number of general revisions about the application of metabolomics in environmental health assessment have been published (Bundy et al., 2009; Miller, 2007; Snape et al., 2004; van Ravenzwaay et **Figure 1.** Citation report for the search "soil pollution-metabolomics" as obtained from Thomson Reuters (January 2014).

metabolomics in soil contamination assessment. The main objectives of this revision are i) to provide a systematized outline for the application of metabolomics in risk assessment of soil contamination, ii) to provide a rapid guide to the methodological approaches current‐ ly optimized and iii) to unify and simplify the knowledge currently available in the topic to provide an accessible tool for further advance in the implementation of metabolomics in risk assessment. al., 2007; Viant et al., 2003). The present review specifically summarizes the most significant research concerning implementation of metabolomics in soil contamination assessment. The main objectives of this revision are i) to provide a systematized outline for the application of metabolomics in risk assessment of soil contamination, ii) to provide a rapid guide to the methodological approaches currently optimized and iii) to unify and simplify the knowledge currently available in the topic to provide an accessible tool for further advance in the implementation of metabolomics in risk assessment. **2. Methodological approaches** 

#### **2. Methodological approaches** Generally, metabolites are extracted from intact organisms (occasionally from selected relevant tissues) that have been exposed to the studied toxicant by moderate chemical extraction (Baylay et al., 2012; Yuk et al., 2010). The

**2.1. Metabolites isolation** 

regression analysis (Ellis et al., 2012).

Ravenzwaay et al., 2012).

#### **2.1. Metabolites isolation** organisms commonly selected for toxicity testing are earthworms (Table 1), particularly the genus *Eisenia*, which are a classic model organism for toxicity assays (Sanchez-Hernandez, 2006; Van Gestel et al., 1992; van Gestel et

organ, biological fluid, or entire organism (Miller, 2007). Exposure of an organism to an external stressor will result in changes at the level of the metabolome (Ankley et al., 2006; van Ravenzwaay et al., 2007), and such changes may constitute a highly sensitive indicator of an external stress. Therefore, metabolomics has potential as a sensitive and rapid technique that can elucidate the relationships between metabolite levels and an external stressor, such as

The main advantage of metabolomics over traditional research is to overcome the bias associated to the assessment of predefined metabolites (Singh, 2006). Among the diverse applications of omic profiling methods to the environmental sciences, ecotoxigenomics addresses the response of organisms to pollutants based on the different sensitivity of species to toxicants (Spurgeon et al., 2008). Currently, the implementation of metabolomics in ecological risk assessment is still at an early stage, mostly applied as a screening tool to assess the potential toxic effects of pollutants or to determine the mode of action (MOA) of a toxicant. Otherwise, application of metabolomics for environmental monitoring allows the study of a

Bundy et al. (2009) highlight the challenge of identifying a large number of metabolites and the necessity of creating metabolite databases specifically dedicated to environmental issues. Multivariate statistical analysis has proved highly effective for metabolite identification. Thus, principal component analysis (PCA) is used to identify differences between metabol‐ ic profiles of organisms exposed to organic or inorganic pollutants (Jones et al., 2014; Kwon et al., 2012; Lankadurai et al., 2011). Besides, association between the metabolic profile and biological factors evaluated as markers for exposure to pollutants can be modelled by partial

The implementation of metabolomics for the assessment of soil contamination is neverthe‐ less at an early stage (Viant, 2009). A basic screening of published research in the web of science returns circa 100 items for the search "soil pollution-metabolomics", with a significant launch in 2011 (Figure 1), reduced to 21 records when the search is narrowed with the term "biomarkers", published in the period 2007-2013. However, emerging regulatory challenges demand the advance of toxicity testing. Toxicogenomics tools have been presented as an advanced from the current methodologies used for regulatory decision making in ecotoxicology, which entirely rely on whole animal exposures and adverse effects on survival, growth, and reproduction (Ankley et al., 2006). From the acquisition of reproducible metabolic profiles as response to the presence of specific pollutants in soil (Jones et al., 2008) to the application of metabolomics techniques to the study of the response of the entire community of a soil to factors such as pollution and climate change (Jones et al., 2014), the implementation of metabolomics in ecotoxicology is a sound answer to the

current needs of society and the environment (van Ravenzwaay et al., 2012).

During the last decade a number of general revisions about the application of metabolo‐ mics in environmental health assessment have been published (Bundy et al., 2009; Miller, 2007; Snape et al., 2004; van Ravenzwaay et al., 2007; Viant et al., 2003). The present review specifically summarizes the most significant research concerning implementation of

contaminant exposure, nutritional deficit or a disease.

460 Environmental Risk Assessment of Soil Contamination

large variety of species from relatively uncontrolled environments.

least squares (PLS) regression analysis (Ellis et al., 2012).

Generally, metabolites are extracted from intact organisms (occasionally from selected relevant tissues) that have been exposed to the studied toxicant by moderate chemical extraction (Baylay et al., 2012; Yuk et al., 2010). The organisms commonly selected for toxicity testing are earthworms (Table 1), particularly the genus *Eisenia*, which are a classic model organism for toxicity assays (Sanchez-Hernandez, 2006; Van Gestel et al., 1992; van Gestel et al., 1989) and have been since long included in official guidelines (OECD, 1984, 2004). Earthworms ingest large amounts of soil and uptake a significant amount of contaminant through the skin. Therefore they are continuously exposed to contaminants. Extractions performed with methanol-chloroform (Baylay et al., 2012) or phosphate buffer solution (Yuk et al., 2010) on pulverized or lyophilised organisms are described to extract the maximum number of metabolites while allowing the performance of reliable analyses. amounts of soil and uptake a significant amount of contaminant through the skin. Therefore they are continuously exposed to contaminants. Extractions performed with methanol-chloroform (Baylay et al., 2012) or

al., 1989) and have been since long included in official guidelines (OECD, 1984, 2004). Earthworms ingest large

The isolated extracts usually might not require further sample treatment prior to analysis, which minimizes the introduction of artefacts but also facilitates the development of low cost, rapid methodologies.

### **2.2. Metabolites determination: chromatography, spectroscopy and spectrometry**

The leading analytical techniques in metabolomics for soil contamination assessment are proton nuclear magnetic resonance spectroscopy (1 H NMR) and gas chromatography–mass spectrometry (GC-MS), as thoroughly reported in Table 1, both allowing the identification of compounds at molecular level in the analysed substances. Several authors have also imple‐ mented high pressure liquid chromatography (HPLC) and ultra high pressure liquid chro‐ matography (UPLC) coupled with mass spectrometry detector (MS) for the assessment of biological responses to soil contamination with heavy metals (Hédiji et al., 2010; Hughes et al., 2009). Overall, these analytical techniques allow the determination and identification of the metabolites that foremost represent the metabolic alterations related to the toxic effects of organic or inorganic contaminants in soil.



**2.2. Metabolites determination: chromatography, spectroscopy and spectrometry**

**Animal model tested**

proton nuclear magnetic resonance spectroscopy (1

462 Environmental Risk Assessment of Soil Contamination

organic or inorganic contaminants in soil.

GC/MS 2-fluro-4-methylaniline *E. veneta*

GC/MS 3-trifluromethylaniline *E. veneta*

GC/MS 3-fluro-4-nitrophenol *E. veneta*

GC/MS 3-trifluromethyaniline *E. veneta* <sup>I</sup>

3,5-difluroaniline *E. veneta*

4-fluroaniline *E. veneta*

1H NMR Caffeine *E. fetida* <sup>I</sup>

1H NMR Chlorpyrifos *L. rubellus* <sup>I</sup>

**Technique Organic toxicant**

1H NMR

1H NMR

1H NMR

1H NMR

1H NMR GC/MS

1H NMR GC/MS

The leading analytical techniques in metabolomics for soil contamination assessment are

spectrometry (GC-MS), as thoroughly reported in Table 1, both allowing the identification of compounds at molecular level in the analysed substances. Several authors have also imple‐ mented high pressure liquid chromatography (HPLC) and ultra high pressure liquid chro‐ matography (UPLC) coupled with mass spectrometry detector (MS) for the assessment of biological responses to soil contamination with heavy metals (Hédiji et al., 2010; Hughes et al., 2009). Overall, these analytical techniques allow the determination and identification of the metabolites that foremost represent the metabolic alterations related to the toxic effects of

> **Biomarkers of contaminant exposure**

I

I

I

1H NMR Aroclor 1254 *E. fetida* No significant changes (Fitzpatrick et al., 1992)

1H NMR Carbamazephine *E. fetida* D[Fumarate, Glu, Val, DLeu] McKelvie et al. (2011)

1H NMR Chlorpyrifos *E. fetida* No significant changes (Yu et al., 2006)

1H NMR Carbaryl *E. fetida* D[Phe, Tyr, Lys, Ala, Val, Leu]

N-oxide]

D[2-hexyl-5-ethyl-3 furansulfonate, maltose],

inosine monophosphate

[Ala, Gly, Asn, glucose,

D[acetate, malonate],

D2-hexyl-5-ethyl-3 furansulfonate, I

monophosphate

[succinate, trimethylamine-

D[Maltose, hexyl-5-ethyl-3-

inosine

furansulfonate] (Bundy et al., 2002)

fumarate (McKelvie et al., 2011)

fumarate (Baylay et al., 2012)

lactate (Lenz et al., 2005)

citrate, succinate] (Warne et al., 2000)

H NMR) and gas chromatography–mass

**Reference**

(Bundy et al., 2002)

(Bundy et al., 2001)

(Bundy et al., 2002)

(Heimbach, 1988) (Edwards and Bater,

1992)



**Table 1** Metabolic responses of test organism following exposure to selected environmental contaminants in contact tests.

#### **2.3. Metabolomics data analysis**

**Technique Organic toxicant**

464 Environmental Risk Assessment of Soil Contamination

1H NMR Polychlorinated biphenyl

1H NMR Poly brominated

1H NMR

1-D & 2-D 1H NMR spectroscopy

1-D & 2-D 1H NMR spectroscopy

1H NMR

PDA

spectroscopy and UPLC-MS

1H NMR, HPLC-

1H NMR, GC-MS

1H NMR spectroscopy, GC–MS

diphenyl ethers

GC/MS Pyrene *L. rubellus*

**Animal model tested**

*E. fetida* <sup>I</sup>

*E. fetida* <sup>I</sup>

Rifluralin *E. fetida* <sup>I</sup>

Trifluralin *E. fetida* <sup>I</sup>

Cd *C. elegans*

Cd *S. lycopersicum*

Chlorpyrifos + Ni *C. elegans*

1H NMR Cu (II) *L. rubellus* <sup>I</sup>

analysis Chlorpyrifos + Ni *L. rubellus*

1H NMR Thiacloprid *L. rubellus* No significant changes were

**Biomarkers of contaminant exposure**

ATP

I

I

I

I

phytochelatins,

trigonelline]

choline, D[glucose, citrate, malate, glutamine,

asparagine, Phe, Tyr, Val, Ile,

[Phe, glucose, malate, arachidonic acid, fumarate, Lys, Tyr, monosaccharides, monophosphorylated form of inositol, succinate, uracil, ethanolamine, Pro, putrescine], Dmyo-inositol,

[Ala, creatine, His, lactate, betaine, carnosine],

D[choline, Gly, Ile, leu, lys, Val]

His (Gibb et al., 1997)

**Reference**

2011)

[lys, Glu], Dmaltose McKelvie et al. (2011)

[Ala, Leu,

[Ala, Gly, maltose, ATP] Yuk et al. (2011)

reported Baylay et al. (2012)

[Ala, Gly, ATP], Dmaltose Yuk et al. (2011)

Dcystathionine (Hughes et al., 2009)

D[lactate, tetradecanoic acid,

Val, Ile, Lys, Tyr, methionine]

hexadecanoic acid, octadecanoic acid], I McKelvie et al. (2011) (Whitfield Åslund et al.,

(Jones et al., 2008)

(Hédiji et al., 2010)

(Baylay et al., 2012)

(Jones et al., 2012)

The general approach to data analysis in metabolomics can be summarized in three main stages: explorative, supervised and biological interpretation (Smilde et al., 2010). The explor‐ ative phase aims to find groups, clusters and outliers in metabolites and samples studied while the supervised discriminates two or more groups to make predictive models and to find biomarkers (Amiard-Triquet et al., 2012; Dallinger-Marianne, 2000; van Ravenzwaay et al., 2007). Multivariate methods are currently preferred, although univariate and semi-univariate methods have been commonly used for selecting biomarkers. For instance, the lysosomal system was identified as a particular target for the toxic effects of pollutants in soil organisms. However, it is nonspecific as a marker and only included in a suite of biomarkers among diverse soil invertebrate species can provide the necessary specificity for risk assessment purposes (Kammenga et al., 2000). Finally, the biological interpretation seeks the links between metabolome data and underlying metabolic networks through metabolite set enrichment, pathway analysis and metabolic network inference (Trygg et al., 2006). Thus, finding metab‐ olite relationships is essential to determine comprehensive and meaningful metabolic changes as biological response to environmental stimuli (Ellis et al., 2012; Morrison et al., 2007). Accordingly, such extensive evaluation of the impact of pollutants in the metabolism of target organisms is the approach that can add value to the assessment of soil health and viability of soil organisms undergoing stress from pollution.
