Effect-directed analysis (EDA) in aquatic ecotoxicology: state of the art and future challenges

1 Background, aim, and scope

One of the key challenges environmental toxicologists and risk assessors are facing is the characterization and assessment of the complex exposure scenarios that are typical for many environments we wish to protect. During the early days of ecotoxicology, it became obvious that the sole use of classic chemical–analytical techniques is not suitable of addressing this issue. Specifically, analysis of the vast number of chemicals typically present in an environmental sample would not only be prohibitively expensive but simply impossible due to limits in the available analytical methodologies for many chemicals—especially as often no a priori knowledge of the chemicals exists which are present in the sample. Therefore, approaches have been developed supplementing chemical analysis with bio-analytical techniques that make use of the specific properties of specific groups of chemicals to interfere with specific biological processes. This type of analysis has been coined effect-directed analysis (EDA) and is based on a combination of fractionation procedures, bio-testing, and subsequent chemical analyses (Brack 2003; Brack et al. 2003; Samoiloff et al. 1983; Schuetzle and Lewtas 1986).

In the late 1980s, one of the first standardized EDA procedures, namely the toxicity identification and evaluation (TIE) approach, had been established by the US-EPA. This approach focuses primarily on the identification and evaluation of organic or inorganic contaminants in aqueous samples and is characterized by the following three steps (reviewed in Brack 2003):

• 1. Toxicity characterization by assignment of toxicity to general groups of toxicants (typically bioassay-directed analysis).

• 2. Identification of suspected toxicants (chemical analytical determination).

• 3. Confirmation of the suspected cause of toxicity.

While approaches like US-EPA’s TIE have been an important step towards improving environmental risk assessment focusing on surface waters, it has increasingly been recognized that particle-bound contaminants (e.g., in suspended matter or sediments) may be of greater ecotoxicological relevance with respect to moderately or strongly lipophilic compounds. Especially sediments represent important long-term sinks for many toxicants (Apitz 2008; Brils et al. 2007; Bunge et al. 2007; Chen and White 2004; Forstner and Salomons 2008; Karlsson et al. 2008; Kase et al. 2008; Keiter et al. 2006), which can become bio-available through flood events or benthic or bottom-dwelling organisms (Babek et al. 2008; Gerbersdorf et al. 2007; Hilscherova et al. 2007; Schulze et al. 2007; Wolz et al. 2009). Consequently, there has been an increasing awareness of the relevance of particle-bound contaminants in the environment, an aspect that has not been considered satisfactorily in classic EDA approaches. In response to these concerns, there are now efforts on the way to establish EDA procedures for

• the assessment of sediment-related pollution such as the recently introduced TIE approach by the US-EPA,

• the identification of harmful substances in sediments using resins and acute whole sediment assays (Phillips et al. 2009), or

• the EDA approaches for the identification of organic pollutants as reviewed by Brack (2003).

The aim of this editorial is to briefly review these recent developments, focusing on the need of rethinking current EDA approaches to address both existing and emerging toxicological issues related to sediments.

2 The EDA process

Effect-directed analysis aims to identify causative agents in complex mixtures through sequential bioassay and analytical chemical analyses (Fig. 1). Specifically, a complex sample (e.g., sediment, air particulate matter, or soil extract) is first analyzed using one (for specific questions such as the characterization of dioxin-like chemicals) or a combination of multiple (non-target analysis) bioassays representing different modes of biological action. Parallel to the assessment of effects on specific biological processes, the general toxicity of a sample (e.g., cytotoxicity for a cell-based bioassay) that could mask a specific response is evaluated. If a sample has been identified as having a significantly altered biological activity, it will then be subjected to fractionation, separating the chemicals contained in it by, e.g., polarity, molecular weight, or any other physical–chemical properties or combinations thereof (Brack 2003). Thereupon, these fractions are again analyzed for their potential to interfere with biological processes using both acute and mechanism-specific bioassays. This procedure is repeated until the activity can be pinpointed to one or multiple specific fractions that contain chemicals of specific properties. Then these fractions are subjected to chemical analysis for suspected compounds.

Fig. 1

Principle of effect-directed identification of toxicants (redrawn from Brack 2003)

Subsequently, in a mass-balance analysis, it will be assessed whether the chemically determined potency of a fraction can explain the biological responses observed in the bioassay (Keiter et al. 2008; Khim et al. 1999a, b). Mass-balance estimates of all major compounds contributing to the observed effects can be calculated to elucidate if all of the activity has been identified, and to assess the potential for interactions such as synergism or antagonism among contaminants present in the complex mixtures (Hilscherova et al. 2000; Korner et al. 2000). This strategy has been widely used for dioxin-like compounds (Safe 1990; Van den Berg et al. 1998), and has recently been adopted for estrogenic chemicals (Hilscherova et al. 2000; Hollert et al. 2005; Khim et al. 1999c; Korner et al. 2000). Finally, a confirmation step in the EDA procedure evaluates how much of the observed toxicity in the environmental sample can be attributed to the mixture of identified toxicants (Brack 2003). More recent approaches focus on the development of methodologies aiming at the assessment of mixture toxicity accounting for unknown modes of action and heterogeneity of concentration–response curves (Brack et al. 2008b).

3 Current EDA approaches and their limitations

Environmental exposure assessments using EDA are routinely used for a number of different chemical effect groups including dioxin-like, genotoxic, or estrogenic compounds. For example, bioassays such as the rat hepatoma cell-line H4IIE test or EROD induction in RTL-W1 cells are widely used to assess the exposure to dioxin-like chemicals that bind to the aryl hydrocarbon receptor (AhR) (Brack et al. 2005b, 2008a; Engwall et al. 1999; Hollert et al. 2002; Kammann et al. 2005b; Olsman et al. 2007b). However, often the chemically detected dioxin-like equivalents (2,3,7,8-TCDD equivalents = TEQs) cannot explain the biological activity measured with the bioassay (bio-TEQ) (e.g., Brack et al. 2005b, 2008a). Similarly, attempts to correlate genotoxicity of complex environmental samples measured by tests such as the Ames, the micronucleus, or the comet assay with PAH concentrations often fail (Broman et al. 1994; Gustavsson et al. 2007; Kosmehl et al. 2004; Reifferscheid et al. 2008; Vahl et al. 1997), suggesting a contribution of other non-regulated mutagens to the observed biological effect. In a number of studies assessing sediments by EDA, it could be demonstrated, for example, that in addition to priority pollutants several non-regulated PAHs, including perylene and benzo[a]fluoranthene, 11H-indeno[2,1,7-cde]pyrene, a methylbenzo[e]pyrene, and a methyl perylene were present at significant concentrations in the analyzed samples (Brack et al. 2005b). Furthermore, Fernandéz et al. (1992) showed that more polar compounds, including several polycyclic quinones and nitroquinones, as well as nitro-PAHs, contributed significantly to the mutagenic effects of marine sediments.

During the past two decades, significant attempts have been made to detect estrogenic effects in European surface waters, and to link them to the contamination with specific groups of chemicals (reviewed in Brack et al. 2007). However, in contrast to dioxin-like potentials, only few studies determined the relative contribution of individual chemicals analytically determined in a complex environmental sample or extract to its biological activity. By means of bioassay-directed fractionation using estrogen-sensitive systems such as MVLN or MCF7-luc cells, it could be demonstrated that the concentrations of the endogenous estrogen 17β-estradiol and the synthetic estrogen ethinylestradiol represented between 88% and 99.5% of the total estrogen equivalents in water samples from certain areas (Snyder et al. 2001). However, other studies revealed that synthetic chemicals such as alkylphenolic compounds can account for the majority of the estrogenic potential of a sample (Hollert et al. 2005; Khim et al. 1999b; Routledge et al. 1998; Sheahan et al. 2002). One of the key challenges for the assessment of the contribution of individual chemicals to the bioassay derived estrogenic potential (estradiol equivalents = EEQs) of a sample is the sensitivity of the utilized analytical method, as has been demonstrated by a study assessing estrogenic compounds in complex environmental samples in the catchment area of the River Neckar, Germany (Hollert et al. 2005). Those estrogenic chemicals that were detected at concentrations above the method detection limits, including nonyl- and octylphenol, phthalates, PCBs, bisphenol A, and DDT, were only able to explain 9% to 14% of the total Bio-EEQs. In contrast, when the method detection limits of chemicals that could not be detected by the utilized analytical methods, namely 17β-estradiol and ethinylestradiol, were taken as a basis of estimation for the Bio-EEQs, 95% of the Chem-EEQ could be explained. Advanced analytical methods for natural and synthetic hormones with lower detection limits are one way to reduce this problem (Aerni et al. 2004, Brack et al. 2007).

While the concept of EDA is increasingly utilized in hazard assessment, often these studies focus on a specific endpoint and are limited in scope (Brack et al. 2005a, 2007; Hollert et al. 2009a; von der Ohe et al. 2009). However, as recently suggested, EDA may be used as additional line of evidence in comprehensive weight-of-evidence studies (Chapman and Hollert 2006), aiming at the identification of the unknown substances responsible for the biological effects in the bioassay under elucidation of the ecological relevance (Fig. 2).

Fig. 2

Although the sediment triad approach is able to gain comprehensive insights into the contamination, the biological effects in bioassays, and alterations in community structure, it often fails in identifying the chemicals responsible for adverse effects. A combination of the triad approach with together with EDA is a promising conceptual framework to overcome this shortcoming (Chapman and Hollert 2006)

Within the last decade, increasingly, attempts were made to improve EDA approaches, e.g., by including fractionation methods for more polar compounds (Lubcke-von Varel et al. 2008; Meinert et al. 2007), methods for addressing the bioavailability within the fractionation approach (Bandow et al. 2009; Schwab and Brack 2007), incorporating structure generation and mass spectral classifiers for identifying of unknown substances (Schymanski et al. 2008), and the use of additional biological endpoints, such as gene expression alterations (Scholz et al. 2008), teratogenicity, and genotoxicity (Kosmehl et al. 2007), and steroidogenesis (Hecker et al. 2007). One particular topic that has concerned risk assessors and regulator during the past two decades is the issue of endocrine disruption. While a large number of national or international programs exist that use a variety of standardized or validated tests to screen chemicals for the Endocrine Disruption properties, such assays are rarely used environmental monitoring or ERA. Specifically, these include the potential of chemicals to interact with the nuclear sex hormone receptors (estrogen and androgen receptor) or to affect the synthesis of steroid hormones. Recent studies have demonstrated, however, that some of these assays such as the L-YES (Wagner and Oehlmann 2009) and the H295R Steroidogenesis Assay (Gracia et al. 2008; Grund et al. 2009; Hecker et al. 2007; Hecker and Giesy 2008) have very great promise as biotests in support of EDA of complex environmental samples. Initial studies with sediments and sewage treatment plant effluents have revealed differential effects when using a combination of different bioassays that capture estrogenic, steroidogenic, dioxin-like, mutagenic/genotoxic and teratogenic effects, fractionation, and chemical analysis, demonstrating the necessity of holistic screening approaches.

Also, there are increasing concerns about emerging contaminants including endocrine disrupting chemicals (Oehlmann et al. 2009; Vos et al. 2000), perfluorinated compounds (Giesy and Kannan 2002; Jernbro et al. 2007), as well as polybrominated and mixed halogenated dibenzo-p-dioxins and -furans (Olsman et al. 2007a). For these compounds, no or only a limited number of sufficiently specific bioassays are available, and thus, such exposures often cannot be appropriately addressed. As a consequence, there is still a great need for refinement and standardization of current sediment EDA approaches that allow capturing and assessing exposures to these chemicals. While well established for its use in ERA of contaminants such as dioxin-like, genotoxic/mutagenic, or estrogenic substances, the continuing discovery of new contaminant groups of concern in the environment or new effect types pose new challenges to classical EDA approaches. These challenges include establishing of bioassays that are specific to the biological activity of chemical groups of concern, as well as the identification and description of relative potencies to model compounds characteristic for these types of effects to enable the utilization of mass-balance approaches.

Finally, as shown in several studies, assessing crude extracts by means of cell-based bioassays may be limited due to general (cyto)toxicity that can mask mechanism-specific effects (Brack et al. 2005b; Hollert et al. 2002; Kammann et al. 2005b). In these situations, the issue of overlying toxicity in the crude extract can be overcome by fractionation approaches resulting in a separation of compounds with different modes of actions.

4 Use of EDA in environmental risk assessment

Effect-directed analysis has been shown to have the potential as a powerful tool in support of environmental risk assessment (ERA), and already is routinely utilized in environmental monitoring programs (Biselli et al. 2005; Brack et al. 2005a, 2007; Kammann et al. 2005a; Phillips et al. 2009; von der Ohe et al. 2009; Weiss et al. 2009; Wölz et al. 2008). With respect to comprehensive toxicological assessments of freshwater systems, the EDA approach was used in several case studies investigating whole catchment areas, providing evidence on the main stressors and possible mitigation measures in order to improve the ecological status of river ecosystems (Brack et al. 2005a, 2007, 2008a; Hollert et al. 2009a; Keiter et al. 2006, 2008, 2009a, b; Weiss et al. 2009). However, to date EDA is almost exclusively based on measurable effects in in vitro and in vivo biotests. Therefore, to address current needs of regulators and risk assessors, an increasing focus should be on the integration of EDA into ERA. Specifically, there is need for the development of tools to confirm EDA-determined key toxicants as stressors in populations, communities, and ecosystems (Brack et al. 2007). One of the most important steps towards these goals is the advancement of toxicant identification approaches/technologies to aid in the identification of unknown substances that are often driving the bio-analytically derived potentials of a sample as discussed in the previous chapter. The development of such tools and strategies represents a challenging task for the next years, but will be an important step forward regarding the successful implementation of national and international environmental programs such as the European Water Framework Directive. They will not only provide evidence on the main stressors responsible for an observed effect but also allow to focus on possible mitigation measures in order to improve the ecological status of an ecosystem.

5 Conclusions

The current focus of research in the field of EDA, as demonstrated by the significant increase in publications (series) regarding this topic (Fig. 3), emphasizes its relevance in the area of environmental sciences, specifically in ERA. For example, a series of articles has been published in the German language journal Umweltwiss Schadst Forsch addressing different issues of effect-directed analysis (Brack et al. 2009; Hollert et al. 2009a, b; Reifferscheid et al. 2009; Schwarzbauer et al. 2009). In addition, a number of reviews presenting the outcome of the integrated EU project MODELKEY—which aimed at using different integrated bio-analytical approaches as well as a comprehensive EDA framework for developing a sound ERA strategy for European freshwater systems (Brack et al. 2005a)—will be published in the forthcoming issues of Environ Sci Pollut Res. With this editorial, we wish to further underline the importance of combining chemical and toxicological methods as well as the need for more emphasis on the further development of effect-directed analysis approaches and their application in ERA. It is our goal to encourage scientists to submit both research papers and reviews on current research, advancements, and future needs, in context with the effect-directed analysis approach. Finally, we wish to strongly recommend to national and international agencies, organizations, and regulatory bodies the need to financially support basic and interdisciplinary research towards further development and application of effect-directed analysis. The ultimate goal is to provide advanced tools for more effective, economic, and safe ERA (Schaeffer et al. 2009).

Fig. 3

Number of articles published between 1990 and 2008 containing the key words “Effect Directed Analysis” and/or “Toxicity Identification & Evaluation”. Source: ISI Web of Knowledge, July 2009