Technical Report

TR 102 – Intelligent Testing Strategies in Ecotoxicology: Mode of Action Approach for Specifically Acting Chemicals

TR 102 : Intelligent Testing Strategies in Ecotoxicology: Mode of Action Approach for Specifically Acting Chemicals | December 2007

There is widespread regulatory and scientific interest in developing intelligent testing strategies (ITS) for the environmental risk assessment (ERA) of chemicals that may enter aquatic ecosystems.  This is especially relevant to the bioaccumulation and ecotoxicity aspects of aquatic risk assessment, with significant benefits likely to be gained from an ITS in terms of animal welfare and efficient use of finite laboratory and economic resources.  The Task Force supports the current use of predictive methods (e.g. structure activity relationships, toxicity threshold concentrations and in vitro screening) to prioritise in vivo testing needs for chemicals with non-specific modes of action.  The Task Force has also developed a 'protein target ITS' approach for in vivo testing of chemicals that are specifically intended to interact with target sites in living organisms (e.g. certain agrochemicals which target the nervous system of insects).

This report provides an in-depth assessment of the current science underlining the use of mode of action (MOA) information for specifically acting chemicals; see glossary for list of all abbreviations used in this report.  The MOA describes the understanding of selected key events that lead to toxic effects, whereas the mechanism of action refers to a comprehensive understanding of the entire sequence of events that result in toxicity (ECETOC, 2006a).  After a review of different MOA classification schemes, the Verhaar et al (1992) approach was adopted as a starting point for this exercise.  Briefly, the Verhaar et al scheme is based on acute effects information and comprises MOA1 (inert chemicals with narcotic properties), MOA2 (less inert chemicals showing polar narcosis), MOA3 (reactive chemicals which reactive unselectively with biomolecules) and MOA4 (specifically acting chemicals with selective biomolecular interactions).  To date, the MOA4 class of chemicals has proved most difficult to fit into a robust ITS approach and, to our knowledge, has not been utilised to date for chronic effects assessment.  Considering both acute and chronic effects, the Task Force has therefore extended the MOA4 class into four pharmacological sub-classes based on: (a) protein receptors, (b) enzymes, (c) ion channels and (d) transporters (after Rang et al, 2003).  While there are a minority of specifically acting chemicals that do not fit into the Rang et al, scheme (e.g. certain genotoxic chemicals which bind to DNA), the Task Force view is that this provides a consistent basis to review the MOA for both therapeutic (intended) effects and also for toxicity (e.g. MOAs for neurotoxicity and tumour induction by some carbamates).  Importantly, recent molecular insights to modern phylogeny provide useful guidance on the evolution of many proteins across species; topical examples include the characterisation of oestrogen receptors across the Deuterosomes (including vertebrates and echinoderms) or moulting hormones across the Ecdyzoans (including arthropods and nematodes).  Conceptually, such information provides a valuable and growing knowledge base, with significant potential to contribute to an ITS for in vivo chronic testing of MOA4 chemicals and ERA.

Expanding upon published ITS schemes (Figure 2.1), the Task Force has developed a simple five step flow-chart (Figure 2.2) which starts with data gathering on physical-chemical data, SAR predictions, in vitro tests, and data from mammalian studies for the compound of interest and for related chemicals.  Evaluation of these data provides information on exposure to, and possible MOA of, the substance.  Such insight may come from efficacy or therapeutic data, or from read-across within chemical classes.  Concerning effects, valuable guidance may be obtained from molecular, biochemical or cellular toxic responses measured in both in vivo and in vitro studies.  For example, fadrozole is an anti-cancer drug which is designed to inhibit aromatase activity in breast tissue (in vivo and in vitro) and which also decreases vitellogenin levels in female fish (and in fish hepatocytesin vitro).  Short-term changes in vitellogenin responses can be used to improve the design of the longer-term fish reproduction test and life-cycle studies (see Chapter 5 for details).  This culminates with a pragmatic, stepwise prioritisation for the assessment of chronic effects in regulatory aquatic test species, providing guidance on the selection from microbial, plant, invertebrate or vertebrates.  The report is focussed on publications and examples from aquatic organisms; however, the principles are also relevant to terrestrial ecotoxicology.  The application of the 'specific MOA flow-chart' (Figure 2.2) is illustrated through five case studies, summarised below (full details in Chapter 6).

(i) Ion channel mediated effect case study: cypermethrin.  Synthetic pyrethroids are neurotoxic to target invertebrates as a consequence of their rapid absorption and subsequent interaction with the sodium ion channel.  The target protein (neuronal sodium channels) is found throughout the animal kingdom (but not in microorganisms or plants) and cypermethrin is slowly metabolised by invertebrates and fish.  Examples of aquatic test species which include the target protein are crustaceans, fish, insects and molluscs.  In terms of an ITS approach for regulatory testing, non-mammalian ADME information (and log Kow = 6.6) suggests that aquatic arthropods will likely be among the most sensitive aquatic organisms (existing toxicity ranking, arthropods > fish > plants = microorganisms).  The available acute data support this logic, whereby the median acute toxicity HC50 values are 0.095, 2.1 and 44000 µg/L for aquatic arthropods, fish and algae, respectively.  Thus, the acute 'base set' interspecies sensitivity factor is nearly 463,000, a value much higher than typical for MOA1-3 chemicals.

(ii) Receptor mediated effect case study: 17a-ethinylestradiol (EE2).  A large amount of in vitro and in vivo mammalian data on this pharmaceutical clearly demonstrates that it has a specific mode of action via oestrogen receptors (MOA4).  EE2 is also designed to be resistant to animal metabolism.  Oestrogens are key hormones throughout the Deuterostomes (from sea squirts to mammals) but they appear to be absent from Ecdyzoans (e.g. crustaceans and insects).  Oestrogen receptors (the target protein) have been shown in amphibians, fish and molluscs and recent molecular studies suggest their occurrence in echinoderms and related marine invertebrates.  In an ITS context, knowledge of ADME (log Kow= 3.67) and oestrogen receptor biology aquatic chronic toxicity gives ranking of fish > arthropods > plants = microorganisms.  The regulatory base set (algal-crustacean-fish) interspecies sensitivity ratio is almost 6.8 (which is not very informative since it lies within the range seen for MOA1-3 chemicals).  More importantly, the MOA focus on mammalian reproductive systems suggests that an assessment of fish developmental and reproductive toxicity will add significant value to the ERA for this type of chemical.

(iii) Transporter protein mediated effects case study: fluoxetine.  Fluoxetine's therapeutic value is based upon its ability to inhibit serotonin (5-HT) reuptake, thereby prolonging its availability at neuronal synapses.  Information on aquatic organisms indicates that serotonin is an important neurotransmitter in amphibians, arthropods, fish and molluscs (but has no known role in plants or microorganisms).  In terms of regulatory testing implications, acute toxicity to daphnids and fish is 940 and 1570 µg/L and the 14 d algal EC50 is 1.1 µg/L (as fluoxetine free base), giving an interspecies sensitivity ratio of 58 (which is higher than is typical for MOA1-3 chemicals).  The possible mechanisms behind the greater sensitivity of algae to fluoxetine are unclear but such a hydrophilic compound (log Kow of 1.8 at pH 7) would be expected to be less toxic to fish and daphnids over an acute exposure period.

(iv) Enzyme mediated effect case study: ketoconazole.  This lipophilic compound (log Kow of 4.35 at pH 7) is used as an antimycotic for human medication and is a member of the azoles, which limit ergosterol synthesis (the predominant component of fungal cell membranes) as a result of cytochrome P450 (CYP) inhibition.  Comparing taxa, CYP enzymes are found in many taxa, with high activity in amphibians, crustaceans, fish and insects but generally lower activity in annelids, molluscs and plants.  In an ITS context, many studies show that high doses of ketoconazole may inhibit mammalian and fish CYPs and steroid hormone biosynthesis.  The acute toxicity to algae, daphnids and fish are 1800, 400 and 3900 µg/L, respectively (interspecies sensitivity ratio of 9.8).  These data indicate that the chronic toxicity ranking may be fish = daphnids > plants = microorganisms (no chronic data available).

(v) Microbial enzyme inhibitor case study: triclosan.  Triclosan is a broad spectrum antimicrobial agent intended for use in consumer products.  It has been shown to inhibit Fab I, enoyl-acyl carrier protein reductase in bacteria but the MOA in other organisms is unknown.  In terms of regulatory testing implications, triclosan (log Kow 4.80) has an acute ecotoxicity profile of 1.4 µg/L (algae), 390 µg/L (daphnids) and 602 µg/L (medaka), giving an interspecies sensitivity ratio of nearly 430.  A 21 d chronic medaka study gave a reproductionNOEC of 200 µg/L andvitellogeninLOEC of 20 µg/L, suggesting that triclosan (or a metabolite) may be weakly oestrogenic with the potential to induce vitellogenin but with no adverse effect on reproductive success and offspring development.

Conclusions and recommendations.

The use of the protein target ITS approach, as observed in previous ECETOC work on toxicological MOAs in human risk assessment, is a rapidly evolving area of science.  Depending on the exposure conditions, some chemicals may induce biological effects that suggest more than one MOA (e.g. carbamates) and caution is needed when using the protein target ITS approach to guide test protocol design (but not predict precise chronic toxicity values).  Therefore, the Task Force recommends an ITS approach to aquatic ecotoxicity testing that includes the following key elements:

  • Gather MOA information on the primary pharmacological/toxicological activity (and any additional toxicity MOAs) for the chemical of interest for the target species as well as mammalian data, also considering structurally-related chemicals, to be used in a weight-of-evidence type of approach;
  • make use of non-traditional sources of biological information, especially the growing biomedical and 'omics' electronic databases on zebrafish, marine invertebrates and other non-mammalian species;
  • if there is evidence for the main MOA being via a protein target (e.g. enzyme or hormone receptor), use this insight to guide the efficient selection of regulatory test methods;
  • measure biomarker responses (e.g. vitellogenin) if desired for read-across purposes or setting test concentrations, however, focus on population relevant endpoints (survival, development growth and reproduction) for generating NOEC (or ECx) values or calculation of PNEC values for application in environmental risk assessment;
  • be cautious of using acute interspecies sensitivity ratios (ISRs) for algae, crustaceans or fish, since the available data suggests they are of limited value for ITS application, presumably since acute high levels exposure induces different MOAs compared to chronic low level exposures (note: available data suggest that MOA4 chemicals can have ISRs from <10 to >500, compared with ISRs for MOA1 (<10), MOA2 (8.6 to 343) and MOA3 (<5 to >25).

Finally, the Task Force has identified key knowledge gaps around regulatory test species which create major uncertainty in developing the ITS approach for many MOA4 chemicals.  Therefore, the Task Force has identified five research needs that, if addressed, will help reduce the scale of this problem in the ITS context.  Taking a pragmatic approach to present and prospective internationally accepted in vivo test methods (e.g. ISO and OECD), these recommendations are:

  • To critically review data (especially chronic studies) from a wider range of chemicals in the context of the proposed MOA and ITS  framework, including chemicals where the mammalian MOA is less specific than for agrochemicals and pharmaceuticals;
  • to develop aquatic plant ADME models with special regard to understanding key biotransformation enzymes;
  • to strengthen the use of small invertebrate models by investing in a hierarchy of biological understanding (including genomics, proteomics and population responses) in the commonly used freshwater and marine invertebrate species (including both arthropods and non-arthropods);
  • for animal welfare reasons, to minimise the need for in vivo fish bioconcentration testing by developing in vitro fish protocols for chemical metabolism and also develop a small-scale invertebrate bioconcentration test method;
  • to support risk assessments of endocrine disrupters, develop a database of the normal (baseline) range for developmental and reproductive endpoints in aquatic organisms measured across different laboratories;
  • to capitalise on the learning from zebrafish biomedical and 'omic' research, there would be value in the establishment of a publicly available database on zebrafish ADME and toxicity information;
  • to support continuing research into SAR validation by providing 'training data sets' based on high quality in vivo chronic tests with plants, invertebrates and fish.