Critical Review

Modes of Action in Ecotoxicology: Their Role in Body

Burdens, Species Sensitivity, QSARs, and Mixture Effects
B E A T E I . E S C H E R * ,# A N D J O O P L . M . H E R M E N S §
Swiss Federal Institute for Environmental Science and Technology (EAWAG) and
Swiss Federal Institute of Technology (ETH), CH-8600 Dübendorf, Switzerland,
and Institute for Risk Assessment Sciences, Toxicology Division,
Utrecht University, 3508 TD Utrecht, The Netherlands

In contrast to the general research attitude in the basic chemical mixtures while protecting many different species
sciences, environmental sciences are often goal-driven and in, and the diversity of, ecosystems.
should provide the scientific basis for risk assessment Because the large number of existing chemicals (1) does
procedures, cleanup, and precautionary measures and not allow an in depth risk assessment at the level of
finally provide a decision support for policy and management. disturbance of an ecosystem, the generic risk assessment
Hence, the prominent role of mechanistic studies in scheme has developed into a system, in which a predicted
no-effect concentration (PNEC) is derived from a limited set
ecotoxicology is not only to understand the impact of
of acute and semichronic data for only a few representative
pollutants on living organisms but also to deduce general species (2, 3). Methods have been developed to extrapolate
principles for the categorization and assessment of the available experimental data for a few species to many
effects. The goal of this review is, therefore, not to provide species (4-6). These methods attempt to account for
an exhaustive coverage of modes of toxic action and uncertainty stemming from factors that can influence the
their underlying biochemical mechanisms but rather to sensitivity of an organism, including the life-history strategy
discuss critically the application of this knowledge in (7) and the test conditions, which are usually better controlled
ecotoxicological risk assessment. Knowing the mechanism in laboratory tests than under field conditions (8). Factors
or, at least, the mode of toxic action is indispensable for that must not be neglected during these extrapolations are
developing descriptive and predictive models in ecotoxicology. differences in bioavailability and speciation of chemicals
between laboratory tests and real ecosystems.
This review seeks to show the crucial role of target
sites, interactions with the target site(s), and mechanisms Another factor, which is subject to extrapolation, is time.
Since for most chemicals only data from acute tests are
for an adequate and efficient ecotoxicological risk
available, acute-to-chronic ratios (ACRs) are applied in
assessment. Emphasis in the discussion is on target environmental risk assessment to estimate a chronic or
effect concentrations (or target occupancy), species semichronic no-observed effect concentration (NOEC) for
selectivity and species sensitivity, time perspective of sublethal effects based on acute effect data, mostly with
effect studies, Quantitative Structure-Activity Relationships survival (LC50) as endpoint. Species sensitivities and ACRs
(QSAR), and mixture toxicity. A particular focus of this are often analyzed chemical by chemical for a class of
review is on multiple mechanisms. Although the illustrative structurally similar compounds and much less attention is
examples were mainly taken from studies in aquatic paid to toxicological and organism-specific criteria to classify
ecotoxicology, the proposed conceptual approach is also chemicals. Knowledge of the abundance of target sites as
in principle applicable and even particularly useful for well as on metabolic activity and defense mechanisms is of
particular relevance for understanding species selectivity and
soil and sediment systems. Recommendations for further
sensitivity. Despite the biological variability, it is hypothesized
research and developments include the use of internal effect that behavior at the target site is relatively constant across
concentrations and target site concentrations in site- different biological systems.
specific risk assessment and as a mixture toxicity parameter It is not only the many species, which complicates the
as well as general considerations for the derivation of risk assessment process, but also the sheer number of
mechanistically meaningful QSAR and other predictive models. chemicals, which hampers the progress and requires methods
that enable the prediction of fate and effect parameters based
on a chemical’s structure or its physicochemical parameters
Introduction via Quantitative Structure-Activity Relationships (QSARs).
This explains the interest of regulatory agencies in QSAR
Environmental Risk Assessment: A Complex Effort. The (9-12, 1). For practical applications of QSARs, several
objective of environmental risk assessment is to protect the computer programs have been developed that supply
environment from adverse effects. Society is faced with the predictions of fate and effect parameters (13-15). Several
enormous task to assess numerous chemicals and complex reviews discuss these toxicology prediction systems in more
detail (16, 17, 12).
* Corresponding author phone: 0041-1-823 5068; fax: 0041-1-
Most QSARs that are used in risk assessment are for effects
823 5471; e-mail: [email protected].
# Swiss Federal Institute for Environmental Science and Technol- at the whole organism level (2). A clear insight into what the
ogy (EAWAG) and Swiss Federal Institute of Technology (ETH). actual target site is, the concentration at this target site, the
§ Utrecht University. mode of action, and the interaction at the target are desirable
10.1021/es015848h CCC: $22.00  2002 American Chemical Society VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4201
Published on Web 09/11/2002
FIGURE 1. Relationship between total, external, and internal effect concentrations and distribution to different target sites.

in developing the models and in guiding the selection of time dependence, QSARs, and mixture effects. The “Conclu-
relevant physicochemical parameters (18-21). sion and Recommendation” section breaks up the rigid
Field oriented risk assessment of actual polluted sites does structure by linking the interconnected topics and empha-
not only deal with individual compounds but also has to sizing potential future applications of mechanistic informa-
evaluate the potential effects of complex mixtures present tion in hazard and risk assessment.
in the environment. Much experimental work has been
focused on evaluating the combined effects of mixtures of Modes of Toxic Action: Target Sites and Mechanisms
pollutants with the objective of deriving general principles
There is a long tradition in toxicological research to investigate
that can then be applied in the risk assessment process (22).
the mechanistic principles underlying a toxic effect. A mode
In addition, group or sum parameters are often applied to
of action is defined as a common set of physiological and
measure “total concentrations” of particular classes of
behavioral signs that characterize a type of adverse biological
chemicals (23). Information on the mode of action is crucial
response, while toxic mechanism(s) refer(s) to the crucial
not only in understanding joint toxic effects and potential
biochemical process(es) and/or xenobiotic-biological inter-
interactions between chemicals in mixtures but also for
action(s) underlying a given mode of action (25). There are
developing both sound mixture toxicity parameters and other
further and sometimes contradictory definitions of these
(bio- and in vitro-) assays for the evaluation of complex
terms, and it is virtually impossible to draw a clear borderline.
mixtures in the field. Site-specific risk assessment is often
In this review, we employ the term “mechanism” whenever
based on external effect concentrations (for example in water,
there is a more detailed description of molecular events
soil, or sediment), and these heavily depend on the biological
available.
species, the specific test and exposure conditions, and on
differences in a chemical’s bioavailability and speciation. Basal cellular structures and functions are highly con-
Internal effect concentrations more directly reflect the served biological entities. Therefore, a large number of toxic
intrinsic activity of a chemical (24). effects that target these basal functions are universal in all
The Scope of This Review. Mechanistic aspects are of organisms and target tissues. On the other hand, there are
great importance in many topics addressed in ecotoxicology. toxic mechanisms that are specific for particular groups of
In this review paper, we try to give an overview on the past organisms, e.g., disturbance of photosynthesis. Categoriza-
developments in this area as well as on potential applications tion of effects according to their associated target site is
in the future. Emphasis is on chemical aspects and examples therefore a first step for setting up predictive models across
are mainly for organic chemicals, but the fundamental different organisms. Xenobiotics may interact specifically
statements are meant to be generally applicable. In a first with certain receptors but they may also be nonspecifically
step, we attempt to categorize molecular mechanisms of toxic. For instance, persistent hydrophobic compounds tend
toxicants that are relevant in ecotoxicology. In our opinion, to accumulate in the membranes of cells, leading to
crucial aspects in understanding and categorizing mecha- nonspecific disturbance of the membrane integrity and
nisms are the target site and the type of interaction of a functioning. Reactive and multifunctional compounds usu-
chemical with the target. Based on this framework, the core ally cannot be assigned to a single mode of action, but they
of the review is divided in three main parts: baseline toxicity, often exhibit multiple toxic mechanisms. This is a difficulty
specific modes of toxic action, and multiple modes of action. for the effect assessment that hitherto has rarely been
Baseline toxicity (also termed narcosis) is the reference case specifically addressed in the literature.
because it is the minimal toxicity of any given chemical. As a starting point, we believe that internal concentrations
Specific modes of action encompass reactive and receptor- in an organism provide a better basis for assessing the intrinsic
mediated mechanisms. The often overlooked issue that a toxicity of a given compound than external concentrations
chemical may have multiple modes of action is specifically (for reviews see refs 26-29). Aqueous effect concentrations
addressed to show approaches how to deal with this are related to the concentration at the site of toxic action
complexity. through several, partially independent processes: on one
These three parts rise in complexity and build upon each hand bioavailability and on the other hand uptake, me-
other but each part is structured in the same way by tabolism, and excretion (toxicokinetic phase), as is depicted
consecutively addressing the topics outlined in the Intro- in Figure 1. The concentration at the site of toxic action or
duction: internal effect concentrations, species sensitivity, target site and the strength of the interactions determine the

4202 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
 the resulting secondary effects may well be dramatically
different between acute and chronic exposure. Note that
indirect effects like effects on the energy status, redox status,
or defense mechanisms also play an important role but are
not considered explicitly here.
The approach to classification given in Table 1 is an
extension of earlier proposed classification schemes. It builds
up on an approach that is based on the assessment of
behavioral (33) and physiological responses of fish (34-37).
Responsiveness to stimulants, locomotive activities, body
color, etc. as behavioral responses and ventilatory pattern,
cough rate, heart rate, etc. as physiological responses were
attributed to eight modes of toxic action using discriminant
FIGURE 2. Rationale behind the classification of chemicals function analysis. These proposed modes of action include
according to mechanism: target sites and type of interaction. three types of baseline toxicity (nonpolar and polar narcosis
and ester narcosis), uncoupling, respiratory inhibition,
toxic effect (toxicodynamics). Target site concentrations are electrophile/proelectrophile reactivity, acetylcholine esterase
therefore more suitable in comparisons of the toxicity of (AChE) inhibition, and several mechanisms of central nervous
chemicals as well as comparisons of species sensitivity. system seizure.
The three main target domains in biological organisms Russom et al. developed an expert system to predict these
are membranes, proteins, and genetic material. The toxic eight modes of toxic action from chemical structures on the
effect is directly related to the type and degree of interaction basis of the acute toxicity syndromes in combination with
of the chemical with the target (Figure 2). Partitioning joint toxicity studies and QSARs (38). Out of a database of
processes lead to mostly nonspecific effects, e.g., partitioning 225 industrial organic chemicals with known behavioral
into membranes leads to baseline toxicity or narcosis. syndromes, 71% were classified as baseline toxicants, 3% as
Sterically favorable van der Waals and hydrogen bond donor/ uncouplers, 4% as central nervous system seizure agents,
acceptor interactions are the “adhesive” for the binding to 2% as respiratory inhibitors, and 20% as reactive chemicals
specialized receptors and specific enzyme inhibition, which (38). Mode-of-action classification did not agree with chemi-
is the molecular basis of receptor-mediated toxicity. cal class as defined by structural similarity.
Formation of covalent bonds of reactive chemicals with Nendza, Wenzel, and co-workers introduced a similar
the target site leads often to irreversible processes. Here, we classification scheme (39, 40, 21). They proposed a test battery
consider electrophiles as reactive chemicals that can form of in-vivo and in vitro tests, which are selective for each one
covalent bonds with nucleophilic entities in biological out of nine modes of action, including the above-mentioned,
molecules (peptides and proteins, DNA), and, in particular, but without ester narcosis and additionally with the class of
we will address alkylating agents. The reaction of allyl chloride photosynthesis inhibition and mutagenicity and/or endo-
with glutathione is an example of such an alkylation reaction. crine disruption. Within a set of representative model
Examples of electrophiles are aldehydes, epoxides, and compounds, each compound yielded a characteristic toxicity
unsaturated aliphatic chlorinated hydrocarbons. Overviews profile. An unknown compound can consequently be clas-
of structures with electrophilic properties are given in review sified by comparison to the reference toxicity profiles. These
articles (30-32). Depending on the reactivity of the elec- authors also emphasized that a chemical could fall into
trophile, the actual target site, the dose level, and duration multiple categories. Nevertheless it was possible to set up
of exposure, the effects of exposure to such reactive chemicals predictive models based upon this classification and to
may vary from protein damage to DNA damage, causing correctly classify almost 90% of a test set of 115 chemicals
ultimately carcinogenicity and mutagenicity. Indirectly, such to the appropriate mode of action class using stepwise
effects can also be caused through the production of reactive discriminant analysis with 10 physicochemical descriptors
intermediates or reactive oxygen species, which we do not (21). Each class of mode of action could be defined by a
address in detail. Note that the interactions depicted in Figure characteristic set of descriptors.
2 do not necessarily lead to a biological effect because defense The following two sections demonstrate for baseline
mechanisms and or metabolic deactivation may prevent the toxicity and specific/reactive mechanisms how the approach
occurrence of visible effects. However, these interactions are introduced here can be further developed and discusses
the first and indispensable steps of a cascade of events that future research that needs to be conducted in light of the
finally lead to observable toxic effects. conclusions reached in this review.
Refinement of the concept of interaction with target sites
for specific targets, e.g., differentiation of biological mem- Baseline Toxicity
branes from generic membranes over energy-transducing Internal Effect Concentration and Target Site Concentra-
membranes to photosynthetic membranes, allows a clas- tion. The idea of narcosis or baseline toxicity originates from
sification of compounds into 10 groups of modes of action. the pioneering work of Meyer and Overton (41, 42) and was
Each mode of action can be explained by one or more taken up in ecotoxicology by Könemann (43) and Veith et al.
mechanisms (Table 1). (44). Baseline toxicity is believed to be a result of nonspecific
This general scheme can be stepwise further refined to disturbance of membrane integrity and functioning as a result
very specific receptor-mediated mechanisms, e.g., the in- of partitioning of pollutants into biological membranes (45).
hibition of specific enzymes such as acetylcholine esterase Baseline toxicity constitutes the minimal toxicity of every
(AChE) or the interaction with specific receptors such as the chemical. Concentrations or volumes of different baseline
estrogen receptor. Each of these mechanisms can be toxicants are virtually constant in biological membranes for
separately investigated with a selective in vitro test system. a defined endpoint effect such as lethality (46, 45). In other
Examples of such test systems are also listed in Table 1. In words, the intrinsic potency for baseline toxicity is equivalent
the proposed classification scheme no explicit distinction is for every chemical. Effects are reversible and occur if a certain
made between acute and chronic effects because, in principle, threshold concentration level is exceeded. Exposure to
there is no difference in the fundamental interactions with sublethal concentrations of narcotics does not increase the
biomolecules between acute and chronic effects, even though sensitivity of test animals to consecutive exposure to lethal

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4203
4204 9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002

TABLE 1. Classification According to Targets, Interaction with Targets and Mechanisms Relevant in Ecotoxicology
site of action subclass domaina interactionb molecular mechanism(s) mode of action selective test system (examples)
biological general m ns nonspecific disturbance of baseline toxicity neutral red assay (40)
membrane membrane structure and Kinspec (59)
functioning
general m s formation of reactive degradation of reaction of reaction product
intermediates (e.g., membrane lipids and (malondialdehyd) with thio-
ROSc) causing membrane proteins barbituric acid (211)
peroxidation of
membrane lipids
and proteins
energy- m ns, s ionophoric shuttle uncoupling O2-consumption in mitochondria (40)
transducing mechanisms Kinspec (109)
m s blocking of quinone and inhibition of the electron O2-consumption in submitochondrial
other binding sites transport chain particles (40)
etc. Kinspec (109)
m s blocking of proton inhibition of ATP ATP assay, e.g. with luciferin/
channels and synthesis/ depletion luciferase (212)
other transport of ATP
channels
photosynthetic m s blocking of photo- photosynthesis inhibition O2-production in chloroplasts (40)
synthetic electron chlorophyll fluorescence (213)
transport
proteins, general m, c ns, s electrophilic reactivity, damage and depletion of GSH depletion (173)
peptides alkylation and oxidation biomolecules
of proteins and
glutathione (GSH)
specific m, c s noncovalent or covalent inhibition or competition, e.g.,: receptor binding studies or enzyme
enzymes and binding to enzymes 1. acetylcholine esterase activity measurements; e.g.,:
receptors and receptors 2. estrogen receptor 1. enzyme activity (40)
3. Ah receptor, etc. 2. yeast estrogen screen (214)
3. CYP1A induction (215)
specific c s noncovalent or covalent indirect mutagenicity induction of repair mechanisms, e.g.,
enzymes and binding to enzymes of (DNA repair, recom- SOS chromotest, or bacterial
receptors the nucleic acid bination, regulation) revertants, e.g., Ames test
metabolism, replication (for review see ref 216)
or repair
DNA, RNA general c ns, s base modification and direct mutagenicity measurement of DNA adducts (173),
damage: electrophilic (frameshift, cross-links, induction of repair mechanisms
(alkylation) and strand breaks, deletion, (for review see ref 216)
oxidative damage, etc.)
bulky adducts
a Cellular environment: m ) membrane, c ) cytosol and other aqueous compartments in the cell. b Selectivity: ns ) nonselective, s ) selective. c ROS ) reactive oxygen species.
concentrations, when effects are expressed in terms of
internal concentrations (47).
Target site concentrations are difficult to obtain directly.
As a surrogate, total concentrations in an organism that elicit
a critical effect, termed critical body residues (CBR) or internal
effect concentrations (IEC), have been used. McCarty defined
CBR as the molar concentration per body weight of the
organism that elicits a defined toxic endpoint (48). If this
endpoint is lethality, one often refers to the lethal body residue
(LBR), internal lethal concentration (ILC), or lethal body
burden (LBB). In the following, we use the term IEC and its
variations because it is less ambiguous than any other
terminology since it is clearly defined as a concentration and
refers to a specific endpoint, e.g. ILC50 refers to lethality for
50% of the test animals.
The rationale behind the use of ILC was the observation
that the QSARs of bioconcentration factors (BCF) and of FIGURE 3. Three-phase equilibrium partitioning model for fish used
lethality (LC50) for baseline toxicants in aquatic organisms to derive effective membrane concentrations from measured internal
were inversely related to each other, resulting in a more or effect concentrations (IEC). Partitioning to storage lipids is described
less constant internal effect concentration, e.g. ILC50, for by the hexane-water partitioning coefficient Khw, partitioning to
compounds with an octanol-water partition constant log membrane lipids by the membrane-water partitioning coefficient
Kow > 2. Kmw. The fraction of chemical in a given phase is denoted by f ; fw
refers to the fraction in the aqueous phase, fmembrane lipid refers to the
fraction in the membrane lipids, and fstorage lipid refers to the fraction
ILC50 ) BCF * LC50 ) constant (1) in the storage lipids.

The ILC50 of baseline toxicants were predicted to be on the appropriate lipid contents of the respective organisms
the order of 2.5 mmol/kgbody weight for acute toxicity (48), and (61), it can be clearly seen that the lethal membrane
measurements later confirmed the prediction giving an concentration is within the same order of magnitude for all
average range of 2-8 mmol/kgbody weight (49-54). For juvenile baseline toxicants (nonpolar and polar chemicals) in algae,
fish, whose average lipid content is about 5%, the ILC50,lip daphnids, and fish (Figure 4A), while the concentration in
normalized to the overall fish lipid is then around 50 mmol/ the storage lipids varies over several orders of magnitude
kgfish lipid. This value is consistent with earlier studies with (Figure 4B).
erythrocyte ghost membranes, where a constant concentra-
Body residues have also been measured for sublethal
tion in the lipid phase of 30-60 mmol/kglipid in membranes
endpoints and different life stages of animals and were found
was measured for the endpoint of 25% reduction in osmotic
useful for defining sublethal hazards (62, 63).
hemolysis (55, 56). For compounds of low hydrophobicity
(log Kow < 2), the partitioning into the lipid phases is not Species Sensitivity. Barron et al. (29) evaluated a large
dominating the overall partitioning, and therefore the compilation of IEC and ILC data of various species from a
contribution of the toxicant concentration in the aqueous database compiled by Jarvinen and Ankley from a wide variety
compartments of the organism has to be considered in of different literature sources (64). They observed that, even
calculating the overall IEC (50). for baseline toxicants, IECs affecting survival varied by 5
Compounds that may undergo strong H-donor/H-ac- orders of magnitude, concluding that species sensitivity, life
ceptor interactions (often simply referred to as polar mol- stage, biotransformation, and physicochemical parameters
ecules) exhibit distinctly lower IEC values than apolar such as pH and salinity might influence the IECs. Despite
molecules in various organisms. ILC50 found for polar this apparent variability, some general conclusions related
narcotics range from 0.6 to 2 mmol/kgbody weight (57, 27, 58). to species sensitivity can be drawn.
However, the effective membrane concentrations were Differences in sensitivity for baseline toxicants are gener-
indistinguishable between the nonpolar and polar com- ally smaller than for specifically acting compounds, which
pounds in an in vitro test system that contains only energy- is not surprising because of the nonspecific character of
transducing membranes, which represent a target lipid baseline toxicity. Every organism may suffer from the baseline
membrane with intercalated proteins (59). This obser- effect and physiological differences in target sites (the
vation shows that there is no real difference between the membrane) are probably minor, which is corroborated by
IECmembrane lipid (based on membrane lipid concentration) of the modeling depicted in Figure 4.
nonpolar and polar chemicals and that the observed dif- The variability of the IECs for a given endpoint within
ference in whole-body concentrations can be related to one study with one species at an acute exposure time (24-96
differences in the distribution between target and nontarget h) varies typically only within 1 order of magnitude and is
compartments. reduced by 50% if the IECs are not expressed per wet weight
Van Wezel et al. and Vaes et al. proposed a simple three- of the organism but in units of mol per mass or volume of
compartment equilibrium partitioning model to relate the lipid in the organism (52). Organisms with a higher storage
ILC50 of fish to the concentration in the target lipids (52, 60). lipid content survive higher total internal effect concentra-
The model compartments consist of 95% aqueous phase and tions (65). Hence variation in tolerance of a subpopulation
5% lipoid phase. Partitioning to protein and other hydro- of fish is accounted for partially by the variation in lipid
phobic macromolecules is neglected. 75% of the lipids were content (65), but there remain susceptibility differences in
assumed to be neutral storage lipids and 25% were polar different fish species and within a population (51, 52, 54).
lipids, i.e., mainly the lipid bilayers of membranes. When In particular, in nonequilibrium situations, differences
reevaluating the simple partitioning model with liposome- in aqueous effect concentrations may occur, because smaller
water partition coefficients Kmw as descriptors for membrane- organisms will reach equilibrium faster that an organism
water partitioning and hexane-water partition coefficients with a smaller surface-volume ratio. In addition, uptake routes
Khw as descriptors for storage lipid-water partitioning (Figure may also be variable and uptake via food ingestion should
3) and when extending it to other aquatic organisms, using also not be neglected. For example, cutaneous uptake is

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4205
FIGURE 4. Comparison of internal lethal effect concentration ILC50 (A) in the membrane and (B) in the storage lipids for algae, daphnids,
and fish. The graph shows boxes that extend from the 25th to the 75th percentile, with a line marking the median. The whiskers extend
above/below to the highest/lowest value. Note that no storage lipids were assumed for algae. Raw data and calculations are given in
ref 61.

FIGURE 5. Baseline toxicity model; EC ) effect concentration

(referring to external aqueous phase), IEC ) internal effect
concentration, BCF ) bioconcentration factor, k2 ) elimination
rate (119, 68). Time dependence of toxicity is due to uptake kinetics.
Membrane concentration rise with time due to bioconcentration
until it reaches the IEC, where lethality is observed.

important for small fish while uptake via the gills dominates
for larger fish (66, 67).
Time Dependence of Toxicity. Since baseline toxicity is
a reversible mechanism, in which the response is directly FIGURE 6. LC50 data for guppy of baseline toxicants (219) plotted
related to the concentration in the membrane, the influence versus (A) logKmw (220-222); (B) logKow (223-225, 222); ([) nonpolar
of time on the effect concentration is determined by the chemicals, (]) polar chemicals.
time it takes for the organism to reach equilibrium with the
surrounding aqueous phase, i.e., by the bioaccumulation (72, 73). Such models not only account for the time-
kinetics (Figure 5). Because the time to reach equilibrium dependence of toxicity but also provide information regarding
increases with the hydrophobicity of a chemical, the relatively distribution of chemicals within an organism as well as
short exposure of 4 days in most acute tests may not be species sensitivity, size and age of the organism, and exposure
sufficient to reach a time independent EC50 value and longer conditions.
test durations have been recommended (68).
QSARs. Since the effect concentrations in the membrane
A detailed discussion of bioaccumulation kinetics is
are virtually constant for baseline toxicants, a QSAR with a
beyond the focus of this review and is reviewed in detail
membrane vesicle-water partition coefficient, Kmw, as single
elsewhere (69-71, 28). Traditionally, empirically based kinetic
descriptor can be used to describe the external effect
models have been used to relate toxic effects to absorbed
concentration LC50/EC50 (Figure 6A and eq 2) (60, 74, 59).
chemicals. An alternative approach involves the use of
physiologically based toxicokinetic models, which account
for the organism’s anatomy, physiology, and biochemistry log EC ) a ‚ log Kmw + c (2)

4206 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
TABLE 2. QSARs of Baseline Toxicity Based on the Membrane-Water Partition Coefficient Kmw as Descriptor
based
on ref QSAR for general baseline toxicitya
Poecilia reticulata (43) log LC50 (M) ) -0.83 log Kmw - 1.46
Vibrio fischeri (217) log LC50 (M) ) -0.79 log Kmw - 1.54
Chlorella vulgaris (218) log LC50 (M) ) -0.91 log Kmw - 0.63
Daphnia magna (217) log LC50 (M) ) -0.77 log Kmw - 1.89
Tetrahymena pyriformis (74) log LC50 (M) ) -0.68 log Kmw - 1.42
a Transformation of QSAR for nonpolar narcosis based on K
ow as descriptor from literature into QSAR for general baseline toxicity (encompassing
nonpolar and polar compounds) by replacing Kow by Kmw using eq 3 (91).

Historically a distinction has been made between nonpolar A selection of resulting general baseline toxicity QSARs is
and polar compounds because if Kow was used as descriptor, listed in Table 2 and details of the derivation are given in ref
two separate QSARs were obtained for the two groups of 61. Though it is possible to reuse these methods, as pointed
chemicals (Figure 6B) (75-77). out above, the use of toxicologically relevant parameters
This difference is a consequence of the difference in should have an increasing priority in ecotoxicology.
molecular interactions of nonpolar and polar chemicals with
Mixtures. In numerous studies, it was confirmed that the
water, octanol, and membrane vesicles. Following the work
effects of mixtures of baseline toxicants can be explained
from Kamlet and Abraham (78), Gunatilleka and Poole
with concentration addition, i.e., a component of a mixture
applied linear free energy relationships using solute char-
acteristics such as volume, excess molar refraction, dipolarity/ can be replaced by an equipotent concentration of another
dipolarizability, and the solute’s effective hydrogen-bond compound without affecting the overall effect. In 1981,
acidity and basicity to explain partitioning processes as well Könemann performed a classical study on the acute toxicity
as toxicity data (79, 80). In fact, there was a significant of mixtures of 3-50 nonspecifically acting chemicals toward
influence of the dipolarity/dipolarizability-factor for octanol- guppy fish (Poecilia reticulata) (92). These early experimental
water partitioning but no contribution to membrane-water results suggested concentration addition. In a number of
partitioning and toxicity, confirming that octanol is not an similar studies on the acute and sublethal effects of mixtures
optimal surrogate for biological membranes (79, 80, 59). of large numbers of compounds, concentration addition was
With some restrictions, for practical applications, Kow- generally confirmed (93-97). A study on the immobilization
based QSARs for baseline toxicity, which are available for of Daphnia magna, showed that very small fractions of a
many endpoints and species are useful in establishing quality toxic concentration of the single compound (down to 0.0025
criteria for aqueous systems as well as for sediments (81- of the inhibitory concentration IC50) still contributed to the
83). However, in future work, conventional physicochemical overall toxicity (98).
descriptors should be replaced with toxicologically relevant For baseline toxicants, IEC can in principle be summed
parameters.
up to derive a risk estimate for mixtures. Van Wezel and
QSARs from the past, that would bring together nonpolar
co-workers proposed an empirical approach for the risk
and polar narcosis type chemicals, included additional
assessment of chemical mixtures of baseline toxicants (53).
parameters for describing the difference in polarizability/
In this study, a fish with an unknown burden of environmental
dipolarity and for hydrogen bond donor/acceptor interac-
tions between octanol as surrogate and the properties of the pollutants was exposed to an increasing concentration of
real target site, the lipid bilayer membrane (79). A few 1,2- or 1,4-dichlorobenzene until a defined endpoint, e.g.,
examples are mentioned in the following. Karabunarliev et lethality, was observed. From the additional internal con-
al. (84) applied the ”maximal acceptor superdelocalizability” centration of the dichlorobenzene and the ILC50 of the given
as parameter in modeling acute toxicity of substituted endpoint, the initial pollutant burden from exposure in the
benzenes. In a series of articles, Mekenyan and Veith (85- field could be back calculated. This method is very simple
87) have applied the “acceptor superdelocalizability” and and appealing, although limited to compounds that are pure
the energy of the lowest unoccupied molecular orbital (ELUMO) baseline toxicants or where baseline toxicity dominates the
as parameters in QSARs for what they call soft electrophiles, mixture effect.
which happened to include many of the polar baseline Hermens and Leeuwangh put forward a hypothesis, which
toxicants. Cronin and Schultz (88) and Bearden and Schultz is very relevant with regard to the balance of baseline toxicity
(89) used ELUMO as parameter in modeling acute toxicity data and specific mechanisms and the above-mentioned practical
of polar narcosis chemicals. Urrestarazu Ramos et al. (58) approach (99). For mixtures of large numbers of chemicals
included parameters for hydrogen bonding acceptor and
with diverse specific modes of action, where the individual
-donor properties in QSARs for polar narcosis type com-
concentrations are well below the threshold of effect, the
pounds in three aquatic species. Dearden et al. (90) applied
underlying baseline toxicity may add up to a significant effect.
polarizability and free hydrogen acceptor factors in an
However, the experimental data of this study was not fully
analysis of toxicity data for nonpolar and polar narcosis.
supportive of this hypothesis. In a field survey on fish
It is also possible to merge Kow-based QSARs from
nonpolar and polar compounds into a common Kmw-based communities, Dyer et al. showed by analyzing fish tissue
QSAR by applying the correlations between Kow and Kmw that residues that concentration addition of the specific effects
were derived separately for nonpolar and polar compounds of the components overpredicted the mixture effect, while
(eqs 3 and 4) (91). adding up the baseline toxicity was a good indictor of the
overall effect as long as the single concentration levels were
nonpolar compounds: below some threshold level for the specific effect (100). More
log Kmw ) 1.05 log Kow - 0.32 (3) experience regarding the balance of cumulative baseline and
mixture effects of specifically acting compounds are needed
polar compounds: before any general conclusions for risk assessment can be
log Kmw ) 0.90 log Kow + 0.521 (4) drawn.

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4207
 et al. developed an in vitro test system for the assessment
of baseline toxicity and uncoupling in isolated energy-
transducing membranes (109, 59). This test system is basically
a model for the pertinent internal concentration because it
contains only the target site (energy-transducing membrane)
in an aqueous suspension. The concentration in the mem-
brane can be estimated with membrane-water partition
coefficients (110). Comparison of the IEC based on membrane
concentrations with the critical membrane concentration of
baseline toxicants in this test system (59) allowed a clear
assignment of a diverse set of substituted phenols to the
appropriate mode-of-action category (61). While chlorophe-
nols are often classified as polar narcotics because their LC50-
values lie within the range predicted by typical QSAR
equations for polar narcosis based on log Kow, the compari-
sons based on the membrane concentrations gave clear
evidence that trichloro- and higher substituted phenols act
as uncouplers. Whereas IECmembrane lipid are constant for all
baseline toxicants, there is a specific IECmembrane lipid for each
uncoupler, which represents its intrinsic activity as an
uncoupler at the internal pH. ILC50-values based on mem-
brane-lipid concentrations were modeled for various organ-
isms with the three-compartment model described in the
section on baseline toxicity and resulting ILC50-values were
similar in the in vitro test and in aquatic organisms (61). In
addition, the modeling showed that the dependence of the
LC50-values of substituted phenols from the external pH is
in fact an artifact due to the pH-dependence of bioaccu-
mulation. Internal membrane concentrations were usually
FIGURE 7. LC50 data of (A) reactive chemicals and (B) of specifically constant for a given compound and independent of pH (61).
acting compounds plotted against Kow. (A) Experimental toxicity Note, however, that it may well be possible for compounds
data of reactive chemicals toward fathead minnow (data taken exhibiting multiple specific mechanisms that dominance of
from the EPA ERL-D fathead minnow database (38)). The line a given mechanism is different in different biological species
corresponds to the QSAR for baseline toxicity (LC50) toward fathead and may depend on the toxicokinetics in the respective
minnow (Pimephales promelas) described by ref 218 from data by biological species, as was shown by comparing the effect of
ref 44. (B) Experimental toxicity data for specifically acting substituted phenols in various biological tests (111).
compounds toward guppy (Poecilia reticulata) (32), QSAR for For reactive chemicals and chemicals that bind irreversibly
baseline toxicity from ref 43.
to a receptor, the extent of reaction with some target molecule
or receptor at the target site, i.e., the target occupation, is a
Reactive and Specific Modes of Toxic Action potential measure of the effect. For example, Freidig et al.
From various studies it is evident that reactive compounds have used the depletion rate constant of glutathione as
(e.g., electrophiles) (Figure 7A) and specifically acting parameter for describing the activity of a series of acrylic
compounds (Figure 7B) are usually more toxic than baseline acid esters (112). Glutathione reacts with these compounds
toxicity models predict (18, 31, 101, 32, 38, 102). Furthermore, in a Michael addition. The depletion of glutathione was
it can be observed that LC50 -values for more hydrophobic directly correlated with an adverse effect. A critical depletion
electrophiles tend to deviate less from baseline models than rate constant of 1.5 d-1 was defined as the critical effect
hydrophilic electrophiles (103-106) (Figure 7A). parameter related to 50% lethality.
Target Concentration and Target Occupation. While for Parameters such as critical membrane concentrations and
baseline toxicants, the whole-body internal effect concentra- other critical effect parameters can be used to classify
tions are a reasonable first approximation of the amount of chemicals and to judge the relative intrinsic potency as
chemicals present at the target site (the membrane), reactive compared to other compounds and to other mechanisms.
and specifically acting compounds may act through binding Species Sensitivity Distributions. Vaal et al. carried out
to receptors or enzymes. Hence, the toxic effect is determined a systematic study evaluating the differences in sensitivities
not only by the concentration of the toxicant at the receptor of species in relation to the mechanism of action (113, 114).
in the membrane or in the cytosol but also by the intrinsic The following mechanisms were included in the analysis:
activity. The intrinsic potency of specifically acting com- baseline toxicity, reactive toxicity, and a few specific mech-
pounds may additionally be dependent on their affinity to, anisms (inhibition of AChE, neurotoxicity). The sensitivity
or type of interaction with, a receptor (e.g., aryl hydrocarbon- distributions were expressed as toxic ratio (TR), the ratio of
receptor for dioxin-like compounds) or the internal speciation the EC for baseline toxicity to the experimental EC and are
(e.g., uncoupling of weak organic acids). Therefore, IECs of plotted in Figure 8.
specifically acting compounds are not only lower than those It is obvious from these plots that TRs of baseline toxicants
of baseline toxicants but also cover a wider concentration are close to 1 and interspecies differences are small. In
range for each mode of action. A compilation of data from contrast, TRs are several orders of magnitude higher and
literature is presented in several reviews (27, 107, 28, 29) and vary strongly for reactive and specifically acting chemicals
can be found in databases (64, 108). and differences in sensitivities are much larger (Figure 8).
Within a category of modes of action, the assessment of This is not surprising because the modes of action of
internal and target site concentrations, i.e., concentrations specifically acting compounds are complex and involve many
in the membrane for membrane bound receptors or cytosolic more intermediate steps. In addition, the intrinsic activity of
concentrations for soluble enzymes as targets, is crucial for chemicals not only depends on the structure of the chemical
understanding the potency differences. For example, Escher but also may be different in different species and sometimes

4208 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
 FIGURE 9. Critical target occupation model (119, 68). EC ) effect
concentration (referring to external aqueous phase), BCF )
bioconcentration factor, k2 ) elimination rate, CTO ) measure of
the critical target occupation (fit parameter of the model representing
the time integral of binding to the target).
FIGURE 8. Species sensitivity distribution (SSD) of toxicities of
various chemicals according to classes of mode of action. Each of endpoint is elicited at a constant target occupation or
the bell-shaped curves corresponds to a normal distribution with constant depletion rate (119, 68). If the interaction of the
a median corresponding to the 50th percentile of the experimen- toxicant with the target is instantaneous and completely
tal SSD and a standard deviation derived from the experimental irreversible, then the integral of the target occupation over
SR95:5-ratio, which is defined as the ratio of 95th to 5th percentile. time is a means to quantify the effect (Figure 9). Consequently,
The experimental data was collected by Vaal et al. (113, 114), and the corresponding (I)EC-values show a strong time-depen-
the plots were adapted from a figure in ref 113. A. SSD of the dence, while the time dependence of baseline toxicity is only
baseline toxicants acetone, o-cresol, ethyl acetate, heptanol, phenol, determined by the bioconcentration kinetics and a constant
propanol, pyridine, and trichloroethylene. For the derivation of the (I)EC. In cases, where the reaction with the target receptor
toxic ratio (TR), Kow values given in ref 113 were converted to Kmw- is not completely irreversible or some replenishment mech-
values through eqs 3 and 4. EC-values for baseline toxicity were anism is active, the observed effect will occur somewhere
estimated with the QSAR for Poecilia reticulata given in Table 2. between the model of constant internal concentration (Figure
B. SSD of two reactive chemicals, salicylaldehyde and propenal. 5) and the CTO-model (Figure 9). The results of the two
TRs were taken directly from ref 113, i.e., derived from Kow-based
models are illustrated in Figure 10A for the example of a
QSARs. C. SSD of specifically acting compounds, including AChE
time-series of LC50-values of the reactive chemical benzyl
inhibitors and neurotoxic agents: methomyl, carbaryl, parathion,
dibrom, fenthion, malathion, dichlorvos, diazinon, aldrin, dieldrin, alcohol toward guppy (68).
endrin, heptachlor, lindane, and toxaphene. Legierse et al. further extended the CTO-model for
organophosphates, which are acetyl choline esterase (AChE)
the target for a particular mode of action is specific for certain inhibitors after metabolic activation from the thio- to the
organisms (e.g., photosynthesis in plants, endocrine system oxo-ester (119). In the model, the active oxo-analogue is
only in higher organisms). The large variation in effect formed in a pseudo-first-order reaction, and it binds to AChE
concentrations for organophosphates is certainly due to with a second-order rate constant. Both back reactions are
differences between organisms in their capacity to biotrans- assumed to be negligible. The endpoint effect is observed
form the parent compound into inactive or active metabolites when a critical amount of AChE is inhibited, which is
and the sensitivity of acetylcholine esterase. In fact, species proportional to the time-integral of the whole-body or
sensitivity distributions for AChE inhibitors and photosyn- aqueous internal concentration. The time-dependence up
thetically active compounds often show several maxima to 14 days of exposure of the LC50 of five organophosphate
corresponding to sensitive and nonsensitive subgroups of pesticides was in excellent agreement with the CTO model,
organisms (115, 116). both, in the guppy Poecilia reticulata (see the example of
Time Dependence of Toxicity. For the reversible mech- malathion depicted in Figure 10B) and in the pond snail
anism of baseline toxicity, the internal effect concentration Lymnaea stagnalis (119).
is virtually constant, and the time dependence of EC-values In addition, physiologically based toxicokinetic/toxico-
is caused by the bioconcentration kinetics (see Figure 5). If dynamic models (PB-TK/TD) are a means to fully account
the mode of action is related to an irreversible chemical for the complexity of pathways and mechanisms of receptor-
reaction, the influence of time on effect concentrations is mediated toxicants, as is illustrated, e.g., by the PB-TK/TD
much more pronounced and can follow the so-called Haber’s model for paraoxon in rainbow trout developed by Abas and
law: the product of concentration and time is constant for Hayton (120).
a given response (117, 118). This rule was established initially The time-dependence of toxicity is also related to acute-
for gaseous reactive compounds. to-chronic ratios (ACR). The upper 90th percentile of ACR
The validity of this distinction of irreversible and reversible in a large compilation of literature data was around 70 (121),
processes and its influence of LC50-time relations has been but the ACR vary strongly depending on the mode of toxic
shown in an analysis of experimental data for a few reactive action. The ACR are typically lower for pure baseline toxicants
and receptor-mediated active compounds (119, 68). Verhaar, and can be much higher if reactive metabolites, which are
Legierse, and co-workers proposed a critical target occupation more likely to be formed during chronic exposure, cause the
(CTO) model, in which it is assumed that a given effect effect (38) or if the mode of action in the chronic test is

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4209
 models with quantum chemical descriptors have been
derived for fish acute toxicity data of, for example, epoxides
(127, 130) and reactive carbonyl-containing aliphatic chemi-
cals (131). Karabunarliev et al. (84) developed QSAR models
for acute toxicity data to fathead minnow of electrophiles
based on molecular mechanisms. They developed separate
models for electrophiles reacting via nucleophilic substitution
reactions, Michael type additions, and Schiff-base formation,
and descriptors were selected based on these reaction
mechanisms. These two steps, the recognition of the reaction
mechanism and the selection of a specific descriptor for this
mechanism, are relevant for getting meaningful and suc-
cessful models. Of course, this is easier in studies of a single
process (for example a certain biotransformation reaction)
than in integral effect studies, in which several processes are
involved (128).
QSARs for Receptor Mediated Toxicity. Receptor medi-
ated processes are complex, because receptor interactions
are often very specific and the strength of such interaction
highly depends on the three-dimensional structure of
molecules. Modeling these interactions usually demands
sophisticated approaches. Approaches such as COMFA (132)
and GRID (133) implicitly represent the lowest-energy, gas-
phase, shape of the pharmacophore (e.g., steric and elec-
trostatic energies derived from the interaction with “probe”
atoms). If the 3-D structure of the receptor is known, the
interactions of ligands with the receptor can be modeled
using a “docking” procedure, based on the energy minimi-
zation of steric (van der Waals) and electrostatic interactions
(134). The conformation of the ligand may, however, change
as a consequence of binding to a receptor and this has led
FIGURE 10. Comparison between baseline toxicity model and critical to the development of the COmmon REactivity PAttern
target occupation model for (A) a reactive chemical (benzyl alcohol) (COREPA) approach by Mekenyan et al. (135). The approach
and (B) a specifically acting compound (malathion). For model has been used to model binding to estrogen and androgen
equations see Figures 5 and 9, data and model parameters from ref receptors (136-138).
119 and 68. Modeling the activity of receptor-mediated processes is
a relevant issue for the development of predictive models for
different from the acute test (122). Therefore inclusion of estrogenic effects of environmental pollutants. It is a
mechanistic information in the derivation of ACR will improve specialized field and we have chosen not to discuss this topic
precision of prediction, as has also been shown by Roex et in detail. Several reviews are available which discuss QSARs
al. (123). for endocrine disruption (see, e.g., ref 139).
QSARs for Reactive Chemicals. It is obvious that any Modeling of integral (overall) effects of chemicals with
hydrophobicity descriptor like Kow is not sufficient for specific (receptor mediated) effects is even more complex
describing effect data in QSAR models for reactive toxicants than modeling receptor interactions. The final effect is
and many attempts have been made to describe effect preceded by several steps, including uptake, distribution
concentrations (EC) via equations of the form within the organism, biotransformation to active or inactive
metabolites, excretion, and interaction with the actual target.
log EC ) a ‚ log Kow + b ‚ R + c (5) Organophosphates are a good example to illustrate this
complexity. In these complex situations, or also if information
in which Kow models the uptake into the organism and R on mechanisms is simply unknown, multivariate techniques
represents a parameter reflecting the differences in reactivity such as principal component analyses (PCA) and partial least-
in a certain reaction with a biologically relevant target. squares analyses (PLS) can be a good starting point for
Early QSAR studies with reactive chemicals have used analyzing experimental data (140-142). Fish LC50 data, but
experimentally measured reactivities (124, 104, 105). Quan- also data on biotransformation and enzyme inhibition for a
tum chemical calculations provide the opportunity to series of organophosphates have been analyzed via QSAR
calculate, from the chemical structure of a compound as the using linear regression techniques and PLS analysis with
sole input, parameters that define the (relative) reactivity of quantum chemical parameters (126, 143, 87, 144). Although
the compound (125). The application of quantum chemical modeling the individual processes was in some cases
parameters in QSARs for ecotoxicity started in the early 1990s successful, the QSARs for the in vivo LC50 data in general
(see for example refs 126 and 127). were of lower statistical quality than models for baseline
An overview of quantum chemical parameters was given toxicants and reactive chemicals.
recently (128). Quantum chemical descriptors may have Mixtures. How chemicals behave in mixtures is strongly
shortcomings because often solvation effects are neglected influenced by their mode of toxic action. If two or more
in the calculations and the absolute values may not always chemicals have different target sites, their effect can usually
be meaningful, but they are very useful as a relative scale be treated independently, even if an integral response of the
within structurally related series (125). Some approaches organism (e.g., lethality) is investigated. Mixtures of chemicals
explicitly include solvation effects in calculating physico- with a common target site and the same mode of action act
chemical properties such as pKa values (see ref 129). according to concentration or dose additivity. If the mixture
Several reviews discuss the application of quantum components interact with each other, they might cause
chemical descriptors in QSAR studies (87, 125, 20, 128). QSAR antagonistic or synergistic effects. The term “synergism” is

4210 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
used with respect to concentration addition, i.e., stronger A prominent and environmentally relevant example of
effect than expected from concentration addition, and the dependent action is the synergistic effect of atrazine in
term “antagonism” refers to an activity, which is lower than combination with organophosphate insecticides. The orga-
independent action predicts. This classification of mixture nophosphates have to be activated from the thio-ester to the
effects stems originally from Plackett and Hewlett (145) and oxo-ester before becoming inhibitors of the acetylcholine-
is based on pharmacological studies but has found wide esterase. Stronger effects than concentration addition were
acceptance in the ecotoxicological community and has been not only reported for acute toxicity (163, 164) but also when
applied also for integrative endpoints such as lethality (146, investigating directly the activity of the AChE (165). Never-
92, 147). Many alternative terminologies have been used as theless, the role of synergistic and antagonistic effects in
well. The concepts of mixture toxicity and related math- mixture studies should not be overemphasized. For a variety
ematical formulations have been extensively reviewed in refs of pesticides the joint effect in aquatic animals could be
146, 148, 147, and 149-152. In this review, we are focusing predicted satisfactorily by concentration addition in 90% of
on the role of the mechanism of toxicity for mixture effects. over 200 mixtures (166). The good agreement was interpreted
An extensive compilation of mixture toxicity studies is as a result of the small difference between the prediction of
presented elsewhere (153). concentration addition and independent action, which are
While the mixture effects of baseline toxicants have been often within one order of magnitude.
thoroughly investigated, there is very little accomplished in Very few studies have been conducted with mixtures of
this area for specifically acting compounds. Examples of reactive chemicals. This is an important research gap because
specifically acting compounds, which share a common mode further complexity is added to the problem due to the
of toxic action and do not interact, therefore acting according possibility of chemical-chemical interactions and other
to concentration additivity, include triazines (154), dioxin- indirect effects. One recent study has dealt with the effect of
like compounds (155, 156), and polycyclic aromatic hydro- phototoxic polycyclic aromatic hydrocarbons (PAH) that
carbons (157). If concentration-effect curves of additive induce oxidative stress. Binary mixtures of phototoxic PAHs
compounds are parallel and have the same maximum, the acted according to concentration addition (167). The same
relative potency of a given compound can be expressed in holds true for glutathione depletion by acrylates, which was
toxic equivalent concentrations. The toxic equivalent con- dose-additive (168). Chen and Yeh showed that reactive
centration of a compound is defined by its concentration electrophilic chemicals acting via the same mode of action
times its relative potency, i.e., the ratio of the ECx-values of are likely to show antagonism (169). In addition some severely
the reference compound of high potency to the ECx-value of synergistic combinations of reactive chemicals with dissimilar
the compound in question. This concept of toxic equivalency mode of action could be identified, e.g. malonitrile in
has been initially developed for aryl hydrocarbon receptor combination with formaldehyde (169). These observations
binding of polychlorinated dibenzodioxins, dibenzofurans, open up a route to more systematic investigations of the
and planar polychlorinated biphenyls (156). This concept is mixture effects of reactive compounds.
appealing due to its simplicity and attempts have been made Earlier studies comparing concentration addition and
to extend this concept to integral effects (156) and other independent action suffered very often from the fact that the
modes of action, e.g., endocrine disruption (158), but great effects were assessed only at one effect level and cumulative
caution should be applied when extending a concept that is low dose effects were not considered. Grimme et al.
strictly valid only for receptor binding assays to integral effects performed a series of studies where the full concentration
and to other modes of toxic action. effect curves of a large number of single compounds and
Mixtures of four xenoestrogens, including DDT and mixtures of compounds, with similar and dissimilar mode
4-nonylphenol, showed concentration-additive effects in the of action, at fixed concentration ratios corresponding to ratios
recombinant yeast estrogen screen assay (159). Xenoestro- of effective concentrations at different effect levels, were
gens are by a factor of 104-105 less potent than the natural analyzed (22, 170, 154, 171). The toxicity of a mixture of 14
hormone 17-β estradiol, and therefore there was doubt strictly dissimilarly acting compounds toward the lumines-
expressed that xenoestrogens would affect the estrogen cent marine bacteria Vibrio fischeri showed very good
receptor at all if estradiol was present. However, equipotent agreement with the prediction according to independent
mixtures of estradiol with DDT or with bisphenol A did show action (154). Analogously, the toxic effect of 16 triazines
additive effects (160). Note that the yeast estrogen screen (inhibitors of photosystem II) in a 24-h synchronized algal
assay just represents binding to the estrogen receptor. For growth test on Scenedesmus vacuolatus and 16 substituted
more integral endpoints, such as the vitellogenin induction phenols (uncouplers of oxidative phosphorylation) and other
in rainbow trouts, toxicokinetic interactions may have to be uncouplers was consistent with prediction for concentration
considered. While 4-nonlyphenol and 17-β estradiol are addition (170). The main conclusion from these experiments
concentration additive, methoxychlor and 17-β estradiol is that concentrations of toxicants below a measurable effect
acted less than additive in this assay, presumably due to the can contribute to the overall toxicity. The effect elicited by
need for activation of methoxychlor through metabolism in the EC01 is indistinguishable from the control but the effect
the liver (161). from the 14 to 16 compounds, each present at its EC01, is
Since specifically acting compounds usually contain measurable and corresponds to the effect predicted by the
specific or reactive functional groups, they are also prone to respective mixture model (170). Similar results were obtained
interactive effects in mixtures. For example, in mixtures of for estrogenic compounds with the yeast estrogen screen
substituted phenols, which act as uncouplers of oxidative (172). This conclusion confirms that concentration addition
and photophosphorylation, the phenoxide species of a has to be considered in the risk assessment of chemicals,
stronger phenolic acid may interact with the neutral species also at very low effect levels. Still missing, however, are
of a weaker acid or a H-donor molecule thus forming a mixed systematic studies comparing concentration-additive effects
dimer, which is a more efficient uncoupler species (162). As of the underlying baseline toxicity with the independent
a result, such mixtures exhibit a synergistic effect. However, action of groups of compounds with dissimilar mode of
if the formation of the mixed dimer is not particularly action.
favorable or the mixed dimer is not an intrinsically better
uncoupler than the dimers of the single components, the Multiple Mechanisms
overall mixture effect can still be described by concentration Within the proposed framework of relating effects to target
addition (162). sites and the interaction with the target, it is possible to

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4211
rationalize the role of multiple mechanisms fairly easily. For
example, an electrophilic compound will be able to react
with any biological nucleophile. The extent of an effect to
the organism will on one hand be related to the abundance
at the target site. If an electrophile is also very hydrophobic,
it will primarily accumulate in membranes and exert its
reactivity on the lipid molecules and will not react with a
soluble enzyme or other molecules that are floating in the
cytosol. On the other hand, any electrophile will have a
preference for a specific nucleophile. For example, a soft
electrophile, like acrylate, is more prone to react with thiol
groups in proteins than a hard electrophile like an epoxide,
which is prone to attack the DNA bases (173).
Chemicals that have a high affinity to ligands, which are
common in biological molecules, are likely to exhibit multiple
mechanisms. For example, triorganotin compounds may
interfere with energy-transduction through several different FIGURE 11. LC50 toward fathead minnow (Pimephales promelas) for
mechanisms, including inhibition of the electron-transfer reactive acrylates plotted against Kow; [ data from ref 112; line
chain, inhibition of the adenosine triphosphate (ATP) syn- corresponds to QSAR equation for baseline toxicity described in
thase, and destruction of the electrochemical proton gradient ref 218 with data from ref 44. Group 1 refers to chemicals, which
through chloride/hydroxide antiport or through hydroxide act according to their reactive mode of action and group 2 includes
uniport. Tributyltin partitions well into biological membranes all pure baseline toxicants. The remaining molecules exhibit features
and consequently appears to exert its dominant effect on of both modes of action.
the electrochemical proton gradient, while triphenyltin,
which is too bulky for easy membrane permeation, but has
a higher affinity to oxy-ligands, appears to be a direct inhibitor
of the ATP synthase (174). In addition to these multiple
mechanisms on energy transduction, triorganotin com-
pounds interfere with many other receptor-mediated pro-
cessed, leading also to chronic effects such as neurotoxicity,
carcinogenicity, immunotoxicity, teratogenicity, and mu-
tagenicity (175).
Multiple effects can also stem from a single first interaction
with the toxicant as depicted in Figure 2 if there is a cascade
of secondary effects or if effects with a common mechanistic
basis result in differential organ toxicity. These topics are
not addressed here in detail but still need consideration.
Concomitant Baseline Toxicity and Specific Mecha-
nisms. Any specific effect will compete with baseline toxicity. FIGURE 12. Two modes of action of reactive chemicals in relation
The more hydrophobic a specifically acting molecule is, the to their target site: baseline toxicity and toxicity related to the
more likely it is that its toxicity can be described through reactivity. IC ) internal concentration; C ) external aqueous
baseline toxicity (see also Figure 7). Benzylhalides may act concentration; R ) a reactivity parameter, e.g., a rate constant; Kmw
as reactive electrophiles or as narcotics depending on their ) membrane-water partition coefficient.
substitution pattern (102). The mode of action of chlorophe-
nols and chlorocatechols changes with an increasing number phase encompasses the processes of uptake, distribution
of chloro substituents, i.e., with increasing hydrophobicity within an organism, and its compartments, biotransforma-
and acidity, from baseline toxicity to uncoupling (176, 177). tion, and excretion. It is beyond the scope to discuss the
Another example are acrylic and methacrylic acid esters, toxicokinetics in detail in this review, but their role for the
which are soft electrophiles and react with glutathione, which overall effect should not be underestimated. In particular,
is an important metabolic conjugant and redox buffer in the the role of the metabolites for the toxic effect has rarely been
cell. The stronger electrophilic acrylates will mainly act investigated. In many cases, metabolites act according to a
because of their reactivity (Figure 11, group 1), the more different mechanism. In principle, the toxic effects of
hydrophobic methacrylates may act mainly via narcosis metabolites have to be assessed iteratively, and parent
(Figure 11, group 2), and some of the (meth)acrylic acid esters compound and metabolites have to be looked at as mixtures
will act via both mechanisms. (179).
Time Scales of Multiple Mechanisms. Multiple mech- QSARs. Chemicals having multiple modes of action can
anisms may also occur at different time scales, which is obscure QSAR analyses. A good example is the study of fish
particularly relevant for the comparison of acute and chronic toxicity with acrylates mentioned above. Some of the acrylates
effects. For example, chlorocatechols in combination with will mainly act because of their reactivity (Figure 11, group
copper may produce reactive oxygen species (178), which 1), some of them may act mainly via narcosis (Figure 11,
may be of relevance for chronic effects, but the acute toxicity group 2), and some of them will act via both mechanisms.
is caused by membrane toxicity, either baseline toxicity or Freidig et al. (112) suggested that the target site for reactive
uncoupling (177). acrylates is in the cytosol (see Figure 12). Because the
Information on the toxicodynamic phase, i.e., the inter- concentration in the cytosol will be similar to the external
action with the target site, the mechanism of toxicity, and aqueous concentration at equilibrium, the effect concentra-
the intrinsic activity (e.g., the reactivity in case of reactive tion will probably only be influenced by differences in
chemicals), needs to be complemented by information of reactivity (R). In a QSAR of the type given in eq 5, the EC
the preceding toxicokinetic phase, when predictive models value will be proportional to 1/R in which R represents the
for effects on whole organisms are to be developed and time- reactivity of a chemical in a certain reaction with the target.
dependence of effects are to be investigated. The toxicokinetic Chemicals within the second group will predominantly act

4212 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
via narcosis and the EC will be proportional to 1/Kmw (Figure (185) that are usually derived as a cutoff aqueous effect
12). Based on these arguments, Freidig and Hermens (180) concentration and therefore implicitly include both bioac-
have suggested to separate a series of chemicals into groups, cumulation and toxic effect. Persistent and bioaccumulative
in which chemicals act by one single mechanism, and then compounds are penalized twice with the presently used
to establish a QSAR model for each group separately. system because they have low aqueous effect concentrations
Analogous comparisons were made for organophosphates, even if they only act as baseline toxicants. Compounds of
which inhibit acetylcholine esterase, and for nitrobenzenes low hydrophobicity may not be classified as hazardous if the
that are reactive chemicals (180). aqueous effect concentration is too high to fall below the
Establishing QSARs for chemicals having multiple modes cutoff value, even if their intrinsic potency is high. In contrast,
of action may in principle be possible if models can be derived in the proposed classification based upon IEC, baseline
for both modes of action separately and, at the same time, toxicants are only penalized for their bioaccumulation (if
if the joint effect of the two modes of action is included in applicable) and specifically acting compounds are unam-
the model formulation. In QSARs for reactive compounds, biguously identified. Intrinsic potency offers many advan-
the coefficient for the influence of log Kow (the parameter a tages as toxicity-indicator, and further research should be
in eq 5) is often in the order of 0.3-0.4. This low coefficient directed to evaluation and validation of this measure.
is very likely a mean value of the coefficient for narcosis Internal effect concentrations values are useful for the
(around 0.9) and for purely reactive toxicity (around 0) and risk assessment of chemicals. Not only can they be applied
reflects the fact that two modes of action are combined with in the classical risk assessment of single compounds as
one equation. proposed by McCarty et al. (27) and Connell and co-workers
These examples show the importance of information (186) but also they pose particular advantages for assessing
about the target site and distribution of a chemical between the risk of pulsed exposure (187), of mixtures of chemicals
nontarget and target compartments. This distribution be- (53, 83), and effects in sediments (188, 184, 82).
tween the cell membrane and the cytosol, which of course Pulse exposure, which involves one or more isolated
is more in favor of the membrane compartment for hydro- exposure periods, or fluctuating exposure, is observable in
phobic chemicals, also explains the observations that LC50- the event of accidents or during normal application of
values for more hydrophobic electrophiles usually deviate pesticides. Hickie et al. developed a simple one-compartment,
much less from baseline models than the hydrophilic first-order toxicokinetic model, which allows prediction of
electrophiles do (103-106) (Figure 7A). The concentrations internal concentration in fish during pulsed exposure (187).
in the target compartment is just getting too low for the If the internal concentration accumulated rises above the
reactive compound to express its effects in the cytosol and IEC value, the given endpoint effect is likely to be observable.
narcosis starts to be the predominant mode of action. Mixture toxicity assessment and site-specific risk assess-
Mekenyan et al. (87) also addressed the change in the ment can also be greatly facilitated by working with internal
influence of Kow by calculating QSARs for sets of chemicals concentrations. Summation of external effect concentrations
with a similar electrophilicity (an isoelectrophilic window). is of limited use, while the summation of IEC, in the case of
They observed a decrease in the influence of Kow for the baseline toxicity, offers a sensible mixture toxicity parameter.
more electrophilic chemicals and explain these observations Procedures for estimating total internal concentrations have
by an influence of interaction of the chemical with nontarget been proposed by several groups (189-192). Bioaccumulation
nucleophiles in the transport of the chemical to the target and biomagnification models can be applied to calculate
site. The lack of a positive hydrophobicity term in QSAR was effective concentrations in the compartment of interest based
recently discussed by Hansch et al., and several explanations on the IEC (184, 193). In such a way, the risk assessment
were suggested (181). becomes more general and less dependent on specific
environmental conditions. The same holds true for differ-
Conclusion and Recommendations ences in species. Nevertheless the concept of adding up
Internal Effect Concentrations in Hazard and Risk Assess- internal concentrations is strictly true only for baseline
ment. The advantages of mechanistic information in eco- toxicants. It needs further development for specifically acting
toxicological studies as well as for hazard and risk assessment compounds. In particular, the question needs to be addressed
are obvious. If the target site(s) and the affinity to the target in which cases the underlying cumulative baseline toxicity
sites are known, the inherent hazard potential of a given dominates the effect in a mixture and in which cases the
compound can be better estimated and further research work concentration or response addition of the fraction of specif-
can be directed to the appropriate questions. Replacing ically acting compounds are dominating the overall toxicity
exposure-based (external aqueous) effect concentrations by in an complex environmental sample. Development of
internal effect concentrations is a first step toward a measure mixture toxicity parameters or other (in vitro) assays for
for inherent toxicity as was pointed out by McCarty (24) and specifically acting chemicals should be based on insights
taken up in a recent debate in SETAC Globe initiated by into both the target site and the critical effect. In addition,
Gobas (182, 183). Further refinement from total internal the problems of bioavailability in in vitro assays (194, 195)
concentration to target site concentration will greatly extend and dosing (196), which are often overlooked, should be
the applicability of this concept. Consideration of the internal addressed in the development of such assays.
concentrations also allows a better comparison of variable Predictive Models. We believe that a mechanistic per-
exposure conditions, e.g. in sediments, soils, and other spective improves any attempt to set up predictive models.
complex matrices (53, 184). Remaining research gaps include Only a profound understanding of the underlying mechanism
the necessity for developing strategies on how to deal with and appropriate assignment of chemicals to a mode of action,
multiple mechanisms and with metabolites. or even to a mechanism, makes it possible to choose the
By defining the intrinsic potency as the ratio of internal right descriptors for QSAR and to define the chemical domain
effect concentration IEC to an internal reference concentra- appropriately (197, 19, 198, 199, 38, 200, 201). The choice of
tion, e.g. the IECbaseline of baseline toxicants, an intensive chemical parameters and the mathematical form of a QSAR
measure of toxicity can be introduced. This measure of should represent, and be coherent with, the crucial molecular
intrinsic potency is independent of concentration or dose processes leading from external concentration to effects.
and is therefore better applicable in a categorization of The decision on the chemical domain of applicability of
persistence-bioaccumulation-toxicity (PBT) (12) than any a QSAR depends strongly on the mode of action concerned.
presently used classification and labeling criteria for toxicity Chemical similarity does not necessarily imply toxicological

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4213
similarity. Also, the development of test (training) sets and New Notified Substances and Commission Regulation (EC) No.
a thorough validation of models are crucial aspects (140- 1488/94 on Risk Assessment for Existing Substances; Office for
142). QSARs for groups of chemicals, which exhibit various Official Publications of the European Communities; Luxem-
bourg, 1996.
or multiple modes of action, will always be less robust than (3) van Leeuwen, C. J.; Bro Rasmussen, F.; Feijtel, T. C. J.; Arndt,
for a single mechanism. Making the uncertainty in predictive R.; Bussian, B. M.; Calamari, D.; Glynn, P.; Grandy, N. J.; Hansen,
models more explicit will save us from surprises and will B.; Van Hemmen, J. J.; Hurst, P.; King, N.; Koch, R.; Müller, M.;
make the decision process more transparent. Solbe, J. F.; Speijers, G. A. B.; Vermeire, T. Environ. Tox.
In addition, species sensitivity distributions and acute- Pharmacol. 1996, 2, 243-299.
(4) Kooijman, S. A. L. M. Water Res. 1987, 21, 269-276.
to-chronic ratios heavily rely upon the correct assignment
(5) Van Straalen, N. M.; Denneman, C. A. J. Ecotox. Environ. Saf.
of mode of toxic action. ACRs are usually smaller if both 1989, 18, 241-251.
acute and chronic effects have the same mode of action than (6) Aldenberg, T.; Slob, W. Ecotox. Environ. Saf. 1993, 25, 48-63.
if they differ in the mechanistic basis. Information on the (7) Kammenga, J. E.; Riksen, J. A. G. Environ. Toxicol. Chem. 1996,
distribution of sensitivities of species is nowadays often used 15, 1649-1653.
in extrapolation methods (4-6) such as in the calculation of (8) Heugens, E. H. W.; Hendriks, A. J.; Dekker: T.; van Straalen, N.
M.; Admiraal, W. Crit. Rev. Toxicol. 2001, 31, 247-284.
the HC5 (hazardous concentration for 5% of the species). A
(9) Feijtel, T. C. J. SAR QSAR Environ. Res. 1995, 3, 237-245.
log-normal distribution of species sensitivities is then often (10) Karcher, W. SAR QSAR Environ. Res. 1995, 3, 167-169.
assumed, although there have been studies that show that (11) Zeeman, M.; Auer, C. M.; Clements, R. G.; Nabholz, J. V.;
this assumption may not always be valid (202). Species Boethling, R. S. SAR QSAR Environ. Res. 1995, 3, 179-201.
sensitivity distributions are influenced by factors related to (12) n.n. Environmental categorization for persistence, bioaccumu-
the uptake, distribution, biotransformation (toxicokinetics), lation and inherent toxicity of substances on the domestic
and interaction of a chemical in the organism (toxicody- substances list using QSARs; Environment Canada: 2000.
(13) Russom, C. L.; Anderson, E. B.; Greenwood, B. E.; Pilli, A. Sci.
namics). Therefore, information on target sites, biotrans- Tot. Environ. 1991, 109, 667-670.
formation, and mode or mechanism of action are of particular (14) Howard, P. H.; Meylan, W. M. In Quantitative Structure-Activity
relevance for understanding and predicting the toxicological Relationships in Environmental Sciences - VII; Chen, F.,
aspects of species selectivity and sensitivity. Schüürmann, G., Eds.; Society of Environmental Toxicology and
A step beyond QSARs in setting up predictive models Chemistry: Pensacola, FL, 1997; pp 185-205.
(15) Klopman, G.; Saiakhov, R.; Rosenkranz, H. S. Environ. Toxicol.
would be to combine the toxicodynamic information with
Chem. 2000, 19, 441-447.
toxicokinetic modeling. Physiologically based toxicokinetic (16) Benfenati, E.; Gini, G. Toxicology 1997, 119, 213-225.
(PB-TK) modeling represents an excellent tool to analyze (17) Richard, A. M. Toxicol. Lett. 1998, 102-103, 611-616.
and predict concentrations in target tissue and target sites. (18) Hermens, J. L. M. In The Handbook of Environmental Chemistry,
If combined with dynamic aspects via physiologically based Reaction and Processes; Hutzinger, J., Ed.; Springer: Berlin,
toxicodynamic (PB-TD) modeling, a more or less complete Germany, 1989; Vol. 2E, pp 111-162.
picture is obtained. We believe that the use of PB-TK and TD (19) Bradbury, S. P. SAR QSAR Environ. Res. 1994, 2, 89-104.
(20) Schüürmann, G. In Ecotoxicology: Ecological Fundamentals,
modeling will finally give insights into rate-limiting steps in, Chemical Exposure, and Biological Effects; Schüürmann, G.,
and theoretically based mathematical model of, the whole Markert, B., Eds.; Wiley: New York, Weinheim, 1998; pp 665-
chain of events from external dose to observable effect. Such 736.
a mathematical model will then feed back into the develop- (21) Nendza, M.; Müller, M. Quant. Struct. Act. Relat. 2000, 19, 581-
ment of QSARs but will also help to understand better the 598.
influence of biotransformation on target and whole-body (22) Grimme, L. H.; Faust, M.; Boedeker, W.; Altenburger, R. Human
Ecolog. Risk Assess. 1996, 2, 426-433.
residues. Such models were already shown to be useful for (23) Van Loon, W. M. G. M.; Hermens, J. L. M. In Monitoring Water
human health risk assessment of single compounds and Quality in the Future; Ministry of Housing, Spatial Planning
mixtures (203-207) and have potential for application in and the Environment: The Hague, The Netherlands, 1995; p
environmental risk assessment (208). Toxicokinetic models 45.
for fish have already been introduced in aquatic sciences (24) McCarty, L. S.; Mackay, D. Environ. Sci. Technol. 1993, 27, 1719-
1728.
(72, 209, 210). Scaling these models to other species, maybe
(25) Rand, G.; Wells, P.; McCarty, L. S. In Fundamentals of Aquatic
even to invertebrates, would be a first step in generalizing Toxicology; Rand, G., Ed.; Taylor & Francis: Washington, DC,
these models for application in predictive models. Then 1995; pp 3-67.
information on the physiology of the organism could be (26) Friant, S.; Henry, L. Chemosphere 1985, 14, 1897-1907.
combined with toxicodynamic information of the chemical (27) McCarty, L. S.; Mackay, D.; Smith, A. D.; Ozburn, A. D.; Dixon,
from in vitro test systems to perform a full risk assessment D. G. Ecotox. Environ. Saf. 1993, 25, 253-270.
(28) Sijm, D. T. H. M.; Hermens, J. L. M. In The Handbook of
of a new chemical.
Environmental Chemistry; Beek, B., Ed.; Springer: Berlin,
In conclusion, considering modes of action in ecotoxi- Germany, 2000; Vol. 2J, pp 167-199.
cology not only will improve our understanding on the effects (29) Barron, M. G.; Hansen, J.; Lipton, J. Rev. Environ. Contam.
of pollutants on ecosystems but also will be useful in setting Toxicol. 2002, 173, 1-37.
up models and avoiding pitfalls in applied environmental (30) Hermens, J. L. M. Environ. Health Perspect. 1990, 87, 219-225.
risk assessment of chemicals and of polluted sites. (31) Lipnick, R. L. Sci. Tot. Environ. 1991, 109/110, 131-153.
(32) Verhaar, H. J. M.; van Leeuwen, C. J.; Hermens, J. L. M.
Chemosphere 1992, 25, 471-491.
Acknowledgments (33) Drummond, R. A.; Russom, C. L. Environ. Toxicol. Chem. 1990,
Stimulating discussions with René Schwarzenbach are grate- 9, 37-46.
fully acknowledged. We thank Renata Behra, Rik Eggen, (34) McKim, J. M.; Bradbury, S. P.; Niemi, G. J. Environ. Health Persp.
Andreas Freidig, Angela Harder, Patricia Holm, Monika 1987, 71, 171-186.
(35) McKim, J. M.; Schmieder, P. K.; Niemi, G. J.; Bradbury, S. P.;
Nendza, Zach Schreiber, and Nina Schweigert for critical
Carlson, R. W.; Henry, T. R. Environ. Toxicol. Chem. 1987, 6,
comments on the manuscript. 313-328.
(36) Bradbury, S. P.; Henry, T.; Niemi, G.; Carlson, R.; Snarski, V.
Environ. Toxicol. Chem. 1989, 8, 247-261.
Literature Cited (37) Bradbury, S. P.; Carlson, R. W.; Niemi, G. J.; Henry, T. R. Environ.
(1) European Commission. White Paper on the Strategy for a Future Toxicol. Chem. 1991, 10, 115-131.
Chemicals Policy; Commission of the European Communities: (38) Russom, C. L.; Bradbury, S. P.; Broderius, S. J.; Hammermeister,
Brussels, Belgium, 2001; KOM 2001 88. D. E.; Drummond, R. A. Environ. Toxicol. Chem. 1997, 16, 948-
(2) European Commission. Technical Guidance Document in Sup- 967, data accessible via http://www.epa.gov/med/databases/
port of Commission Directive 93/67/EEC on Risk Assessment for fathead_minnow.html.

4214 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
(39) Nendza, M.; Wenzel, A.; Wienen, G. SAR QSAR Environ. Res. (76) Veith, G. D.; Broderius, S. J. In QSAR in Environmental
1995, 4, 39-50. Toxicology-II; Kaiser, K. L. E., Ed.; Reidel: Dordrecht, 1987; pp
(40) Wenzel, A.; Nendza, M.; Hartmann, P.; Kanne, R. Chemosphere 385-391.
1997, 35, 307-322. (77) Veith, G. D.; Broderius, S. J. Environ. Health Perspect. 1990, 87,
(41) Meyer, H. H. Arch. Exp. Pathol. Pharmakol. 1899, 42, 109-118. 207-211.
(42) Overton, E. Vierteljahreszeitschrift Naturforsch. Gesellschaft (78) Kamlet, M. J.; Doherty, R. M.; Abraham, M. H.; Taft, R. W. Quant.
Zürich 1899, 44, 88-135. Struct. Act. Relat. 1988, 7, 71-78.
(43) Könemann, H. Toxicology 1981, 19, 209-221. (79) Gunatilleka, A. D.; Poole, C. F. Anal. Commun. 1999, 36, 235-
(44) Veith, G. D.; Call, D. J.; Brooke, L. T. Can. J. Fish. Aquat. Sci. 242.
1983, 40, 743-748. (80) Gunatilleka, A. D.; Poole, C. F. Analyst 1999, 125, 127-132.
(45) van Wezel, A. P.; Opperhuizen, A. Crit. Rev. Toxicol. 1995, 25, (81) Van Leeuwen, C. J.; van der Zandt, P. T. J.; Aldenberg, T.; Verhaar,
255-279. H. J. M.; Hermens, J. L. M. Environ. Toxicol. Chem. 1992, 11,
(46) Abernethy, S. G.; Mackay, D.; McCarty, L. S. Environ. Toxicol. 267-282.
Chem. 1988, 7, 469-481. (82) Di Toro, D. M.; McGrath, J. A. Environ. Toxicol. Chem. 2000, 19,
(47) Pawlisz, A. V.; Peters, R. H. Environ. Sci. Technol. 1995, 29, 613- 1971-1982.
621. (83) Di Toro, D. M.; McGrath, J. A.; Hansen, D. J. Environ. Toxicol.
(48) McCarty, L. S. Environ. Toxicol. Chem. 1986, 5, 1071-1080. Chem. 2000, 19, 1951-1970.
(49) van Hoogen, G.; Opperhuizen, A. Environ. Tox. Chem. 1988, 7, (84) Karabunarliev, S.; Mekenyan, O. G.; Karcher, W.; Russom, C. L.;
213-219. Bradbury, S. P. Quant. Struct. Act. Relat. 1996, 15, 302-310.
(50) McCarty, L. S.; Mackay, D.; Smith, A.; Ozburn, G. W.; Dixon, D. (85) Mekenyan, O. G.; Veith, G. D. SAR QSAR Environ. Res. 1993, 1,
G. In QSAR in Environmental Toxicology IV; Hermens, J., 335-344.
Opperhuizen, A., Eds.; Elsevier: Amsterdam, The Netherlands, (86) Veith, G. D.; Mekenyan, O. G. Quant. Struct. Act. Relat. 1993,
1991; pp 515-525. 12, 349-356.
(51) Sijm, D.; Schipper, M.; Opperhuizen, A. Environ. Toxicol. Chem. (87) Mekenyan, O. G.; Veith, G. D. SAR QSAR Environ. Res. 1994, 2,
1993, 12, 1117-1127. 129-143.
(52) van Wezel, A. P.; de Vries, D. A. M.; Kostense, S.; Sijm, D. T. H. (88) Cronin, M. T. D.; Schultz, T. W. Sci. Total Environ. 1997, 204,
M.; Opperhuizen, A. Aquat. Tox. 1995, 33, 325-342. 75-88.
(53) van Wezel, A. P.; deVries, D. A. M.; Sijm, D. T. H. M.; Opperhuizen, (89) Bearden, A. P.; Schultz, T. W. Environ. Toxicol. Chem. 1997, 16,
A. Ecotox. Environ. Saf. 1996, 35, 236-241. 1311-1317.
(54) Chaisuksant, Y.; Yu, Q. M.; Connell, D. Ecotox. Environ. Saf. (90) Dearden, J. C.; Cronin, M. T. D.; Zhao, Y. H.; Raevsky, O. A.
1997, 37, 66-75. Quant. Struct. Act. Relat. 2000, 19, 3-9.
(55) Seeman, P.; Roth, S.; Schneider, H. Biochim. Biophys. Acta 1971, (91) Vaes, W. H. J.; Urrestarazu-Ramos, E.; Hamwick, C.; van Holstein.
225, 171-184. I.; Blaauboer, B. J.; Seinen, W.; Verhaar, H. J. M.; Hermens, J.
(56) Seeman, P. Pharmacol. Rev. 1972, 24, 583-655. L. M. Chem. Res. Toxicol. 1997, 10, 1067-1072.
(57) de Wolf, W.; Seinen, W.; Opperhuizen, A.; Hermens, J. L. M. (92) Könemann, H. Toxicology 1981, 19, 229-238.
Chemosphere 1992, 25, 853-863. (93) Hermens, J. L. M.; Canton, H.; Steyger, N.; Wegman, R. Aquat.
(58) Urrestarazu-Ramos, E.; Vermeer, C.; Vaes, W. H. J.; Hermens, Tox. 1984, 5, 315-322.
J. L. M. Chemosphere 1998, 37, 633-650. (94) Broderius, S.; Kahl, M. Aquat. Tox. 1985, 6, 307-322.
(59) Escher, B. I.; Eggen, R.; Vye, E.; Schreiber, U.; Wisner, B.; (95) De Wolf, W.; Canton, J. H.; Deneer, J. W.; Wegman, R. C. C.;
Schwarzenbach, R. P. Environ. Sci. Technol. 2002, 36, 1971- Hermens, J. L. M. Aquat. Tox. 1988, 12, 39-49.
1979. (96) Nirmalakhandan, N.; Arulgnanendran, V. J.; Mohsin, M.; Sun,
(60) Vaes, W. H. M.; Urrestarazu-Ramos, E.; Verhaar, H.; Hermens, B.; Cadena, F. Water Res. 1994, 28, 543-551.
J. L. M. Environ. Toxicol. Chem. 1998, 17, 1380-1384. (97) Niederlehner, B. R.; Cairns, J.; Smith, E. P. Ecotox. Environ. Saf.
(61) Escher, B.; Schwarzenbach, R. P. Aquat. Sci. 2002, 64, 20-35. 1998, 39, 136-146.
(62) Fisher, S. W.; Chordas, S. W.; Landrum, P. F. Aquat. Tox. 1999, (98) Deneer, J. W.; Sinnige, T.; Seinen, W.; Hermens, J. Aquat. Tox.
45, 115-126. 1988, 12, 33-38.
(63) Hwang, H.; Fisher, S. W.; Landrum, P. F. Aquat. Tox. 2001, 52, (99) Hermens, J.; Leeuwangh, P. Ecotox. Environ. Saf. 1982, 6, 302-
251-267. 310.
(64) Jarvinen, A.; Ankley, G. Linkage of effects to tissue residues: (100) Dyer, S.; White-Hull, C.; Shephard, B. Environ. Sci. Technol.
development of a comprehensive database for aquatic organisms 2000, 34, 2518-2524.
exposed to inorganic and organic chemicals; Society of Envi- (101) Nendza, M.; Russom, C. Xenobiotica 1991, 21, 147-170.
ronmental Toxicology and Chemistry (SETAC): Pensacola, FL, (102) Marchini, S.; Passerini, L.; Hoglund, M. D.; Pino, A.; Nendza,
1999. M. Environ. Toxicol. Chem. 1999, 18, 2759-2766.
(65) Lassiter, R. R.; Hallam, T. Environ. Tox. Chem. 1990, 9, 585- (103) Lipnick, R. L.; Watson, K.; Strausz, A. Xenobiotica 1987, 17,
595. 1011-1025.
(66) Nichols, J. W.; McKim, J. M.; Lien, G. J.; Hoffman, A. D.; Bertelsen, (104) Deneer, J. W.; Seinen, W.; Hermens, J. L. M. Aquat. Tox. 1988,
S. L. Toxicol. Appl. Pharmacol. 1991, 110, 374-389. 12, 185-192.
(67) Lien, G. J.; Nichols, J. W.; McKim, J. M.; Gallinat, C. A. Environ. (105) Deneer, J. W.; Sinnige, T. L.; Seinen, W.; Hermens, J. L. M.
Toxicol. Chem. 1994, 13, 1195-1205. Aquat. Tox. 1988, 13, 195-204.
(68) Verhaar, H. J. M.; De Wolf, W.; Dyer, S.; Legierse, K. C. H. M.; (106) Veith, G. D.; Lipnick, R. L.; Russom, C. L. Xenobiotica 1989, 19,
Seinen, W.; Hermens, J. L. M. Environ. Sci. Technol. 1999, 33, 555-565.
758-763. (107) Chaisuksant, Y.; Yu, Q. M.; Connell, D. W. Rev. Environ. Contam.
(69) Connell, D. Bioaccumulation of xenobiotic compounds; CRC Toxicol. 1999, 162, 1-41.
Press: Boca Raton, FL, U.S.A., 1990. (108) Environmental residue-effects database (ERED); Waterways
(70) McKim, J. M.; Schmieder, P. In Bioaccumulation in aquatic Experiment Station (WES), U.S. Army Corps of Engineers, U.S.
systems: contributions to the assessment; Nagel, R., Loskill, R., Environmental Protection Agency: 2001; accessed at http://
Eds.; VCH Verlagsgesellschaft: Weinheim, Germany, 1991; pp www.wes.army.mil/el/ered/ on 11/20/2001.
161-188. (109) Escher, B. I.; Snozzi, M.; Häberli, K.; Schwarzenbach, R. P.
(71) Geyer, H.; Rimkus, G.; Scheunert, I.; Kaune, A.; Schramm, K.- Environ. Toxicol. Chem. 1997, 16, 405-414.
W.; Kettrup, A.; Zeeman, M.; Muir, D.; Hansen, L.; Mackay, D. (110) Escher, B. I.; Schwarzenbach, R. P.; Westall, J. W. C. Environ.
In The Handbook of Environmental Chemistry; Beek, B., Ed.; Sci. Technol. 2000, 34, 3954-3961.
Springer: Berlin, Germany, 2000; Vol. 2J, pp 1-166. (111) Schüürmann, G.; Segner, H.; Jung, K. Aquat. Tox. 1997, 38,
(72) Barron, M. G.; Stehly, G. R.; Hayton, W. L. Aquat. Tox. 1990, 17, 277-296.
187-212. (112) Freidig, A. P.; Verhaar, H. J. M.; Hermens, J. L. M. Environ. Sci.
(73) Nichols, J. W.; McKim, J. M.; Andersen, M. E.; Gargas, M. L.; Technol. 1999, 33, 3038-3043.
Clewell, H. J., III; Erickson, R. J. Toxicol. Appl. Pharmacol. 1991, (113) Vaal, M. A.; van der Wal, J. T.; Hoekstra, J.; Hermens, J. L. M.
106, 443-447. Chemosphere 1997, 35, 1311-1327.
(74) Schultz, T. W.; Seward, J. R. Quant. Struct. Act. Relat. 2000, 19, (114) Vaal, M. A.; Van Leeuwen, C. J.; Hoekstra, J. A.; Hermens, J. L.
339-344. M. Environ. Managem. 2000, 25, 415-423.
(75) Schultz, T. W.; Holcombe, G. W.; Phipps, G. L. Ecotox. Environ. (115) Solomon, K. R.; Baker, D. B.; Richards, R. P.; Dixon, D. R.;
Saf. 1986, 12, 146-153. Klaine, S. J.; LaPoint, T. W.; Kendall, R. J.; Weisskopf, C. P.;

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4215
 Giddings, J. M.; Giesy, J. P.; Hall, L. W.; Williams, W. M. Environ. (154) Backhaus, T.; Altenburger, R.; Boedeker, W.; Faust, M.; Scholze,
Toxicol. Chem. 1996, 15, 31-74. M.; Grimme, L. H. Environ. Toxicol. Chem. 2000, 19, 2348-
(116) Scheringer, M.; Steinbach, D.; Escher, B.; Hungerbühler, K. 2356.
Environ. Sci. Poll. Res. 2001, 8, 1-8. (155) Birnbaum, L. S.; DeVito, M. J. Toxicology 1995, 105, 391-401.
(117) Rozman, K. K. Toxicology 2000, 149, 35-42. (156) Van den Berg, M.; Birnbaum, L.; Bosveld, A. T. C.; Brunstrom,
(118) Rozman, K. K.; Doull, J. J. Pharmacol. Exp. Ther. 2001, 296, B.; Cook, P.; Feeley, M.; Giesy, J. P.; Hanberg, A.; Hasegawa,
663-668. R.; Kennedy, S. W.; Kubiak, T.; Larsen, J. C.; van Leeuwen, F.
(119) Legierse, K. C. H. M.; Verhaar, H. J. M.; Vaes, W. H. J.; De X. R.; Liem, A. K. D.; Nolt, C.; Peterson, R. E.; Poellinger, L.;
Bruijn, J. H. M.; Hermens, J. L. M. Environ. Sci. Technol. 1999, Safe, S.; Schrenk, D.; Tillitt, D.; Tysklind, M.; Younes, M.; Waern,
33, 917-925. F.; Zacharewski, T. Environ. Health Perspect. 1998, 106, 775-
(120) Abbas, R.; Hayton, W. L. Toxicol. Appl. Pharmacol. 1997, 145, 792.
192-201. (157) Fent, K.; Bätscher, R. Environ. Tox. Chem. 2000, 19, 2047-
(121) Lange, R.; Hutchinson, T. H.; Scholz, N.; Solbe, J. Chemosphere 2058.
1998, 36, 115-127. (158) Safe, S. H. Environ. Health Perspect. 1998, 4, 1051-1058.
(122) Roex, E. W. M.; Giovannangelo, M.; van Gestel, C. A. M. Ecotox. (159) Payne, J.; Rajapakse, N.; Wilkins, M.; Kortenkamp, A. Environ.
Environ. Saf. 2001, 48, 196-201. Health Persp. 2000, 983-987.
(123) Roex, E. W. M.; Van Gestel, C. A.; Van Wezel, A. P.; Van Straalen, (160) Rajapakse, N.; Ong, D.; Kortenkamp, A. Toxicol. Sci. 2001, 60,
N. M. Environ. Toxicol. Chem. 2000, 19, 685-693. 296-304.
(124) Hermens, J. L. M.; Busser, F.; Leeuwangh, P.; Musch, A. Toxicol. (161) Thorpe, K.; Hutchinson, T.; Hetheridge, M.; Scholze, M.;
Environ. Chem. 1985, 9, 219-236. Sumpter, J.; Tyler, C. Environ. Sci. Technol. 2001, 35, 2476-
(125) Karelson, M.; Lobanov, V. S.; Katritzky, A. R. Chem. Rev. 1996, 2481.
96, 1027-1043. (162) Escher, B. I.; Hunziker, R. W.; Schwarzenbach, R. P. Environ.
(126) Schüürmann, G. Environ. Toxicol. Chem. 1990, 9, 417-428. Sci. Technol. 2001, 35, 3905-3914.
(127) Purdy, R. Sci. Tot. Environ. 1991, 109-110, 553-556. (163) Pape-Lindstrom, P. A.; Lydy, M. J. Environ. Toxicol. Chem. 1997,
(128) Soffers, A. E. M. F.; Boersma, M. G.; Vaes, W. H. J.; Vervoort, 16, 2415-2420.
J.; Tyrakowska, B.; Hermens, J. L. M.; Rietjens, I. M. C. M. (164) Howe, G. E.; Gillis, R.; Mowbray, R. C. Environ. Toxicol. Chem.
Toxicol. in Vitro 2001. 1998, 17, 519-525.
(129) Schüürmann, G. J. Chem. Phys. 1998, 109, 9523-9528. (165) Belden, J. B.; Lydy, M. J. Chemosphere 2001, 44, 1685-1689.
(130) Eriksson, L.; Verhaar, H. J. M.; Hermens, J. L. M. Environ. Toxicol. (166) Deneer, J. W. Pest Manag. Sci. 2000, 56, 516-520.
Chem. 1994, 13, 683-691. (167) Erickson, R. J.; Ankley, G. T.; DeFoe, D. L.; Kosian, P. A.;
(131) Schultz, T. W.; Cronin, M. T. D. J. Chem. Inf. Comput. Sci. Makynen, E. A. Toxicol. Appl. Pharmacol. 1999, 154, 97-105.
1999, 39, 304-309. (168) Freidig, A. P.; Hofhuis, M.; Van Holstijn, I.; Hermens, J. L. M.
(132) Cramer, R. D. I.; Patterson, D. E.; Bunce, J. D. J. Am. Chem. Soc. Xenobiotica 2001, 31, 295-307.
1988, 110, 5959-5967. (169) Chen, C. Y.; Yeh, J. T. Environ. Toxicol. Water Quality 1996, 11,
(133) Goodford, P. J. J. Med. Chem. 1985, 28, 849-857. 83-90.
(134) Balbes, L. M.; Mascarella, S. W.; Boyd, D. B.; Lipkowitz, K. B., (170) Altenburger, R.; Backhaus, T.; Boedeker, W.; Faust, M.; Scholze,
Boyd, D. B., Eds.; VCH Publishers: New York, U.S.A., 1994; M.; Grimme, L. H. Environ. Toxicol. Chem. 2000, 19, 2341-
Vol. 4, pp 337-380. 2347.
(135) Mekenyan, O. G.; Ivanov, J. M.; Veith, G. D.; Bradbury, S. P. (171) Faust, M.; Altenburger, R.; Backhaus, T.; Bodeker, W.; Scholze,
Quant. Struct. Act. Relat. 1994, 13, 302-307. M.; Grimme, L. H. J. Environ. Qual. 2000, 29, 1063-1068.
(136) Bradbury, S. P.; Mekenyan, O. G.; Ankley, G. T. Environ. Toxicol. (172) Silva, E.; Rajapakse, N.; Kortenkamp, A. Environ. Sci. Technol.
Chem. 1998, 17, 15-25. 2002, 36, 1751-1756.
(137) Bradbury, S. P.; Kamenska, V.; Schmieder, P. K.; Ankley, G. T.; (173) van Welie, R. T. H.; vn Dijck, R. G. J. M.; Vermeulen, N. P. E.
Mekenyan, O. G. Toxicol. Sci. 2000, 58, 253-269. CRC Crit. Rev. Tox. 1992, 22, 271-306.
(138) Mekenyan, O. G.; Kamenska, V.; Schmieder, P. K.; Ankley, G. (174) Hunziker, R. W.; Escher, B. I.; Schwarzenbach, R. P. Environ.
T.; Bradbury, S. P. Toxicol. Sci. 2000, 58, 270-281. Toxicol. Chem. 2002, 21, 1191-1197.
(139) Ankley, G.; Bradbury, S.; Hermens, J.; Mekenyan, O.; Tollefsen, (175) Markert, B. In Ecotoxicology: Ecological Fundamentals, Chemi-
K. E. In Workshop on Endocrine Modulators and Wildlife: cal Exposure, and Biological Effects; Schüürmann, G., Markert,
Assessment and Testing; Tattersfield, L., Matthiessen, P., B., Eds.; Wiley: Weinheim, Germany, 1998; pp 165-222.
Campbell, P., N., G., Länge, R., Eds.; SETAC-Europe: Brussels, (176) Escher, B. I.; Snozzi, M.; Schwarzenbach, R. P. Environ. Sci.
Belgium, 1997; pp 19-40. Technol. 1996, 30, 3071-3079.
(140) Tysklind, M.; Lundgren, K.; Rappe, C.; Eriksson, L.; Jonsson, (177) Schweigert, N.; Hunziker, R. W.; Escher, B. I.; Eggen, R. I. L.
J.; Sjostrom, M. Environ. Toxicol. Chem. 1993, 12, 659-672. Environ. Toxicol. Chem. 2001, 2, 239-247.
(141) Eriksson, L.; Hermens, J. L. M.; Johansson, E.; Verhaar, H. J. (178) Schweigert, N.; Belkin, S.; Leong-Morgenthaler, P.; Zehnder,
M.; Wold, S. Aquat. Sci. 1995, 57, 217-241. A. J. B.; Eggen, R. I. L. Environ. Mol. Mutagen. 1999, 33, 202-
(142) Eriksson, L.; Johansson, E.; Wold, S. In Quantitative Structure- 210.
Activity Relationships in Environmental Sciences-VII; Chen, F., (179) Fenner, K.; Koijman, C.; Scheringer, M.; Hungerbühler, K.
Schüürmann, G., Eds.; SETAC Press: Pensacola, FL, 1997; pp Environ. Sci. Technol. 2002, 36, 1147-1154.
381-397. (180) Freidig, A. P.; Hermens, J. L. M. Quant. Struct. Act. Relat. 2001,
(143) de Bruijn, J. H. M.; Hermens, J. L. M. Aquat. Tox. 1993, 24, 19, 547-553.
257-274. (181) Hansch, C.; Kurup, A.; Garg, R.; Gao, H. Chem. Rev. 2001, 101,
(144) Verhaar, H. J. M.; Eriksson, L.; Sjostrom, M.; Schuurmann, G.; 619-672.
Seinen, W.; Hermens, J. L. M. Quant. Struct. Act. Relat. 1994, (182) Gobas, F. A. C. P. SETAC Globe 2001, 2(6), 29-30.
13, 133-143. (183) Gobas, F. A. C. P.; Purdy, R.; Granville, G.; Cowan-Ellsbery;
(145) Plackett, R. L.; Hewlett, P. S. J. R. Statist. Soc. B 1952, 14, 141- Gannon, J.; Lewis, M.; Socha, A.; Chapman, P. SETAC Globe
163. 2001, 2(4), 33-34.
(146) Calamari, D.; Alabaster, J. Chemosphere 1980, 9, 533-538. (184) van Wezel, A. P.; Jonker, M. T. O. Aquat. Tox. 1998, 42, 287-
(147) Altenburger, R.; Boedeker, W.; Faust, M.; Grimme, L. In 300.
Handbook of Hazardous Materials; Academic Press: 1993; pp (185) OECD Harmonized integrated classification system for human
15-27. health and environmental hazards of chemical substances and
(148) Greco, W.; Unkelbach, H.; Pöch, D.; Sühnel, J.; Kundi, M.; mixtures; Organization for Economic Co-Ordination and
Boedeker, W. Arch. Complex Environ. Stud. 1992, 4, 65-69. Development: 2001; ENV/JM/MONO(2001)6.
(149) Könemann, W. H.; Pieters, M. N. Food Chem. Toxicol. 1996, (186) Connell, D. W.; Chaisuksant, Y.; Yu, J. Mar. Pollut. Bull. 1999,
34, 1025-1031. 39, 54-61.
(150) Pöch, G.; Dawson, D.; Reiffenstein, R. TEN 1996, 3, 51-59. (187) Hickie, B. E.; McCarty, L. S.; Dixon, D. G. Environ. Toxicol.
(151) Groten, J. P.; Feron, V. J.; Suhnel, J. Trends Pharmacol. Sci. Chem. 1995, 14, 2187-2197.
2001, 22, 316-322. (188) McCarty, L. S. SETAC Europe News 1997, 8(2), 5-8.
(152) Tallarida, R. J. J. Pharmacol. Exp. Ther. 2001, 298, 865-872. (189) Verhaar, H. J. M.; Busser, F. J. M.; Hermens, J. L. M. Environ.
(153) ECETOC. Aquatic Toxicity of Mixtures; Technical Report No. Sci. Technol. 1995, 29, 726-734.
80; European Centre for Ecotoxicology and Toxicology of (190) Parkerton, T. F.; Stone, M. A.; Letinski, D. J. Toxicol. Lett. 2000,
Chemicals: Brussels, Belgium, 2001. 112, 273-282.

4216 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
(191) van Loon, W. M. G. M.; Wijnker, F. G.; Verwoerd, M. E.; (212) ATP Luminescence - Rapid Methods in Microbiology; The
Hermens, J. L. M. Anal. Chem. 2000, 68, 2916-2926. Society for Applied Bacteriology Technical Series; Stanley, P.
(192) Verbruggen, E. M. J.; Vaes, W. H. J.; Parkerton, T. F.; Hermens, E., McCarthy, B. J., Smither, R., Eds.; Blackwell Scientific
J. L. M. Environ. Sci. Technol. 2000, 34, 324-331. Publications: Oxford, U.K., 1989; Vol. 26.
(193) van Wezel, A. P.; van Vlaardingen, P.; Posthumus, R.; Crom- (213) Schreiber, U.; Bilger, W.; Neubauer, C. Ecolog. Studies 1994,
mentuijn, G. H.; Sijm, D. T. H. M. Environ. Tox. Chem. 2000, 100, 49-70.
19, 2140-2153. (214) Andersen, H.; Andersson, A.; Arnold, S.; Autrup, M.; Barfoed,
(194) Gülden, M.; Seibert, H. Toxicol. in Vitro 1997, 11, 479-483. M.; Beresford, N.; Bjerregaard, P.; Christiansen, L.; Gissel, B.;
(195) Gülden, M.; Morchel, S.; Seibert, H. Toxicol. in Vitro 2001, 15, Hummel, R. Environ. Health Persp. 1999, 107(suppl. 1), 89-
233-243. 108.
(196) Brown, R.; Akhtar, P.; Akerman, J.; Hampel, L.; Kozin, I.; Villerius, (215) Richter, C. A.; Tieber, V. L.; Denison, M. S.; Giesy, J. P. Environ.
L.; Klamer, H. Environ. Sci. Technol. 2001, 35. Toxicol. Chem. 1997, 16.
(197) Schultz, T. W. Ecotox. Environ. Saf. 1987, 14, 178-183.
(216) de Maagd, P. G. J.; Tonkes, M. Environ. Toxicol. 2000, 15, 81-
(198) Hermens, J. L. M.; Balaz, S.; Damborsky, J.; Karcher, W.; Müller,
90.
M.; Sabljic, A.; Sjöström, M. SAR QSAR Environ. Res. 1995, 3,
223-236. (217) Zhao, Y. T.; Cronin, M. T.; Dearden, J. C. Quant. Struct. Act.
(199) Richard, A. M. Toxicol. Lett. 1995, 79, 115-122. Relat. 1998, 17, 131-138.
(200) McKinney, J. D.; Richard, A. M.; Waller, C.; Newman, M. C.; (218) Verhaar, H. J. M.; Hermens, J. L. M.; Mulder, W. QSAR for
Geberick, F. Toxicol. Sci. 2000, 56, 8-17. ecotoxicity. In Fate and activity modeling of environmental
(201) Trapp, S.; Schwartz, S. Chemosphere 2000, 41, 965-971. pollutants using structure-activity relationships; EU contract
(202) Newman, M. C.; Ownby, D. R.; Mezin, L. C. A.; Powell, D. C.; EV5V-CT92-0211; RITOX: Utrecht, The Netherlands, 1999.
Christensen, T. R. L.; Lerberg, S. B.; Anderson, B. A. Environ. (219) Urrestarazu Ramos, E.; Vaes, W. H. J.; Verhaar, H. J. M.;
Toxicol. Chem. 2000, 19, 508-515. Hermens, J. L. M. J. Chem. Inf. Comput. Sci. 1998, 38, 845-852.
(203) Simmons, J. E. Toxicology 1995, 105, 111-119. (220) Gobas, F. A. C. P.; Lahittete, J. M.; Garofalo, G.; Shiu, W. Y. J.
(204) Yang, R. S. H.; El-Masri, H.; Thomas, R.; Constan, A. Toxicol. Pharm. Sci. 1988, 77, 265-272.
Lett. 1995, 82/83, 497-504. (221) van Wezel, A. P.; Cornelissen, G.; van Miltenburg, J. K.;
(205) van Vliet, P. W.; de Jongh, J. Human Experim. Toxicol. 1996, Opperhuizen, A. Environ. Toxicol. Chem. 1996, 15, 203-212.
15, 799-809. (222) Vaes, W. H. J.; Urrestarazu-Ramos, E.; Verhaar, H. J. M.; Cramer,
(206) Cassee, F. R.; Groten, J. P.; van Bladeren, P. J.; Feron, V. Crit. C. J.; Hermens, J. L. M. Chem. Res. Toxicol. 1998, 11, 847-854.
Rev. Toxicol. 1998, 28, 73-101. (223) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M.
(207) Yang, R. S. H.; Thomas, R. S.; Gustafson, D. L.; Campain, J.; Environmental Organic Chemistry; Wiley: New York, U.S.A.,
Benjamin, S. A.; Verhaar, H. J. M.; Mumtaz, M. M. Environ. 1993.
Health Perspect. 1998, 6, 1385-1393.
(224) Abraham, M.; Chadha, H.; Whiting, G.; Mitchell, R. J. Pharm.
(208) McCarty, L. S.; Ozburn, G. W.; Smith, A. D.; Dixon, D. G. Environ.
Sci. 1994, 83, 1085-1110.
Toxicol. Chem. 1992, 11, 1037-1047.
(209) Landrum, P.; Lee, H., II; Lydy, M. Environ. Toxicol. Chem. 1992, (225) Escher, B. I.; Schwarzenbach, R. P. Environ. Sci. Technol. 1996,
11, 1709-1725. 30, 260-270.
(210) McKim, J. M.; Nichols, J. W. In Aquatic Toxicology: Molecular,
Biochemical, and Cellular Aspects; Malins, D. D., Ostrander, G. Received for review December 18, 2001. Revised manuscript
K., Eds.; Lewis: Boca Raton, FL, 1994; pp 469-519. received May 27, 2002. Accepted June 3, 2002.
(211) Liposomes; New, R. R. C., Ed.; Oxford University Press: Oxford,
1990. ES015848H

VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4217