der Universität Leipzig Dipl. Chem. Heike Schmitt Utrecht, 15.3.04 Abschlussarbeit Postgradualstudium Toxikologie C –

C

–

Abschlussarbeit

Postgradualstudium Toxikologie der Universität Leipzig

Dipl. Chem. Heike Schmitt Utrecht, 15.3.04

CONTENTS

Abstract...3

Introduction ...4

Combined action...4

Independent action or response addition ...5

Can modes of action be determined from combined effect studies? ...6

Quantitative structure-activity relationships (QSAR) and modes of action ...6

Applicability of combined effect studies for the analysis of modes of action – tested experimentally ...7

Materials and methods...8

Chemicals and reagents ...8

Test organisms and culture conditions ...9

Determination of concentration-response relationships ...9

Results ...12

Single compounds’ toxicities...12

Mode of action of mixture components...16

Mixture toxicity ...18

Discussion...21

Mode of action considerations...21

Mixture toxicity ...22

Mixture toxicity vs. QSAR and parallel concentration response curves ...25

Literature ...27

Curriculum Vitae Heike Schmitt ...32

ABSTRACT

The current risk assessment of environmental contaminants is generally based on an assessment of single substance toxicity, in contrast to real-world environmental contamination, which is often composed of chemical mixtures. The evaluation and prediction of the toxicity of a mixture of substances from its constituents therefore deserves special attention in ecotoxicology. While the toxicity of mixtures with well- defined modes of action can be accurately described by the two concepts of concentration addition and response addition, it is more difficult to evaluate the combination effects of substances when less information on the mechanisms of their toxic effects is available. This study tests an application of combination toxicity concepts with a different goal: It evaluates whether mixture toxicity studies can also be applied to gain more insight into the modes of action of the constituents of a given mixture.

To this end, an analysis of the mixture toxicity of different nitrobenzenes on the reproduction of the green alga Scenedesmus vacuolatus was undertaken. Using lipophilicity-based quantitative structure-activity-relationship modelling for nitrobenzenes, the assumption is held that mononitrobenzenes may exert narcotic effects as a common type of action, while dinitrobenzenes show a somewhat greater toxicity.

From the literature, QSAR models based on quantum chemical parameters suggest that some mononitrobenzenes may be effective by additional other modes of action. The toxicity of a mixture of 14 nitrobenzenes clearly exceeds the predicted combined effects as expected for the sum of toxic units from a uniform narcotic mode of action.

Moreover, the observed combined effect is smaller than that predicted from simply similar acting compounds calculated on the basis of the parametrised dose-response functions using concentration addition. Further modelling of the combined effect, joining the model of concentration addition for components with anticipated similar modes of action and of response addition for those with independent action, let us to propose that not all of the nitrobenzenes follow the same mode of action. This is in line with a hypothesis derived from quantum chemical QSAR considerations. Overall, it appears that combined effect analysis can be applied as a pharmacological probe to test for the similarity of the mode of action of mixture components.

INTRODUCTION

Current regulatory ecotoxicology is based on the evaluation of single contaminants and their possible effects in the environment, as deduced from laboratory tests with a range of test species. This is in spite of the fact that environmental compartments are frequently exposed to mixtures of various chemicals rather than one substance alone.

The question arises whether more attention should be given to the combined effects of these multiple toxicants.

An experimental evaluation of all possible mixture effects is practically impossible, due to the wide range of substances potentially co-occurring. An answer has rather to be sought in the analysis of existing data on single species ecotoxicity with regard to combined effects. For the evaluation of combined effects, two contrasting concepts have been applied widely, the concepts of concentration addition and of independent action.

They can be used for the prediction of the combined effect from known data on single substance toxicity as well as for the assessment of combined effects in a given mixture.

Combined action

The concept of concentration addition has been introduced by the pharmacologist Loewe (Loewe, Muischnek 1926, Loewe 1927, 1953). It is based on the idea of a

“similar action” of mixture components (Bliss 1939, Plackett, Hewlett 1952). Inherent to such a common mode of action in the biological systems exposed is the idea that one compound may act totally or in part as a dilution of another to elicit the same effect (Altenburger, Nendza, Schüürmann 2003). In the simplest case, this implies parallel dose-response curves for the respective chemicals, which only differ in their relative effectiveness. More generally, concentration addition is thought to be applicable for substances that share an identical molecular mechanism of action and a similar mode of action A feature of concentration addition is that the effect of a mixture remains constant if one component is partly or totally substituted for a fraction of a so-called equi-effective concentration of another substance. The mathematical formulation of this concept is as follows:

∑

= Q =

L [L

(&FL

c stands for the concentrations of substances i through n, and EC_xfor the concentrations of the single substances that are required to exert a certain effect x (such as EC₅₀).

In summary, common structural features of chemicals and similar modes of actions are thought to be prerequisites for the use of the concept of concentration addition to predict combined effects of chemical mixtures from the effects of the individual components (Calamari and Vighi, 1992, Greco et al. 1995, Kortenkamp and Altenburger 1998, 1999).

Independent action or response addition

In contrast, for mixtures of substances composed of components with different structures and dissimilar modes of action, response addition (also called independent action) is thought to be a valid pharmacological reference concept for toxicity prediction (Pöch 1993, Altenburger, Nendza, Schüürmann 2003). Bliss (1939) was the first to introduce this concept under the name of “independent joint action”. Dissimilar modes of action in this context can be understood as interaction with different primary sites of action and different chains of effects. Under these circumstances, it is thought that the relative effect of the substance, such as the 50% inhibition in comparison with a control, stays the same in the presence of a second substance. Mathematically, this translates to

( )

∏

−

=

L L

Q ( F

F

( (2)

with c denoting the concentrations of substances i through n, and E the respective effects. For example, two mixture components both leading to 20% mortality when applied singly would lead to a 80% survival upon exposure to the second substance of the population remaining after applying the first agent, 80%, in total thus 0.8 x 0.8=

0.64 or 64% survival.

It has been questioned whether the assumption of independent effect chains holds true for integral effects in complex systems such as whole cells, organisms or populations (Greco 1992). Can truly independent effect chains occur in systems structured as networks, including feedback loops, compensatory reactions and hierarchical dependencies? However, recent experimental evidence for mixtures of substances with known specific pharmacological mechanisms of action using luminescent bacterial and algal biotests (Altenburger et al. 2000, Backhaus et al. 2000, Faust et al. 2000, Faust et

al. 2001) strongly support the suitability of the concepts for specifically similarily or dissimilarily acting compounds, respectively.

Can modes of action be determined from combined effect studies?

If the premises of the concepts of concentration addition and response addition are valid even for non-specifically acting compounds at integral levels of toxic effects, it may be possible to use the experimental study of combined effects of mixtures to retrieve information on the type of action of pollutants as proposed by Broderius et al. (1995) and to use it for probabilistic risk assessment techniques, such as calculating potentially affected fractions from species sensitivity distributions (Traas et al. 2002). The focus of this paper is an experimental evaluation of the potential of combined effect studies to serve as a pharmacological probe for the similarity of modes of actions of mixture components.

Quantitative structure-activity relationships (QSAR) and modes of action The goal of the application of quantitative structure-activity relationships (QSAR) in environmental toxicology is to describe the toxicity of a certain substance by suited molecular parameters, such as the lipophilicity of the substance. QSARs have often been applied for substances with a common basic chemical structure, in order to evaluate the contribution of different substituents. However, there is also a group of structurally very heterogenic substances whose toxicity is described with one common QSAR equation, the so-called narcotics. These substances share a common mode of action (to be defined here in the sense of Rand et al. (1995) as „a common set of physiological and behavioral signs that characterize a type of adverse biological response“. After Schüürmann (1998), the term mode of action is used „where the underlying mechanism is not definitely disclosed, but rather characterized at a more phenomenological level“). They exert their toxic effects by interaction with constituents of the cell membrane, by a mechanism of action which is not yet completely unravelled.

It is thought that narcotic toxicity is a feature which is common to all chemicals once their membrane concentration reaches a certain value, however, some substances also exert different effects at lower concentrations and different cellular targets.

Applicability of combined effect studies for the analysis of modes of action – tested experimentally

In the following investigation, the effects of mixtures of a range of nitrobenzenes are studied in an algal test system and compared with theoretical considerations allowing assumptions about the mechanism of action of the different nitrobenzene compounds.

Nitrobenzenes are important environmental chemicals regarding their substantial marketing volumes as industrial chemicals. Their use patterns are diverse, ranging from applications as solvents to uses in the synthesis of dyestuffs, urethan polymers, and anilines. Derivative products include various active ingredients of insecticides and herbicides as well as pharmaceuticals. The aquatic toxicity of nitrobenzenes has been described for unicellular green algae (Deneer et al., 1989, Kramer et al. 1986, Urretarazu Ramos et al. 1999, Schmitt et al. 2000), daphnids (Deneer at al. 1989, Urretarazu Ramos et al. 1999), Tetrahymena (Schultz and Moulton 1985, Dearden et al. 1995, Schüürmann et al. 1997), photobacteria (Gough 1994), and fish (Deneer et al. 1987, Roberts 1987).

Using QSAR techniques for the analysed congeneric compounds, lipophilicity has been shown to be a good descriptor of nitrobenzene toxicity in various organisms, and is used in hazard predicion protocols such as ECOSAR (US-EPA 1998). For the QSAR of the algal toxicity data hydrogen bonding descriptors in addition to the lipophilicity parameter K_OW are used. These are interpreted as explaining the polar properties of nitrobenzenes and other compound classes (Deneer et al., 1989, Urretarazu Ramos et al.

1999) while allowing to hold the assumption of narcosis as a common mode of action (Urretarazu Ramos et al. 1999, Vaes et al. 1998). The notion of one common mode of action of nitrobenzenes in algae has been challenged on the basis of QSAR analysis employing quantum chemical parameters for a uniform description of the algal toxicities in relation to the structure of 19 different nitrobenzenes (Schmitt et al. 2000).

Consideration of E_LUMO and E_SOMO let the authors to propose that apart from a narcotic mode of action some of the nitrobenzenes may additionally exert oxidative stress through redox cycling (Schmitt et al. 2000).

The potential of combined effect studies to serve as a pharmacological probe for the similarity of modes of actions of ixture components is the focus of this paper. To this end, we extended a previous study on the algal toxicity of nitrobenzenes (Schmitt et al.

2000), selected a group of 14 compounds with high reproducibility of the concentration- response relationship, and studied its mixture toxicity in dilution series. The mixture comprised 13 mononitrobenzenes whose toxicity may be described by a simple hydrophobicity-driven QSAR equation, while one compound, dinitramine, a dinitrobenzene with known herbicidal properties, served as a bait with a specific dissimilar mode of action and slightly different structure. Also, two of the mononitrobenzenes have been anticipated to additionally exert oxidative stress through redox cycling (Schmitt et al. 2000). Combined effect analysis was performed using parametric modelling on the basis of concentration-response relationships for the individual components for the inhibition of algal reproduction, and by comparing the observed toxicity with the mixture toxicity calculated on the basis of the concepts of concentration addition and response addition, respectively.

MATERIALS AND METHODS

Chemicals and reagents

The substances used, sources and purities are specified in table 1.

table 1: Test substances

substance abbr. CAS RN source charge purity

Nitrobenzene NB 98-95-3 Merck ZA1656639716 99%

2-Chloronitrobenzene 2Cl 88-73-3 Aldrich CQ03416AQ >99%

3-Chloronitrobenzene 3Cl 121-73-3 Merck 4316331 >99%

4-Chloronitrobenzene 4Cl 100-00-5 Merck S00717729 >99%

3-Aminonitrobenzene 3A 99-09-2 Merck 3316702 >99%

4-Aminonitrobenzene 4A 100-01-6 Merck L344760748 >98%

4-Methylnitrobenzene 4Me 99-99-0 Merck S01768746 >98%

5-Chloro-2-aminonitrobenzene 2A5Cl 89-63-4 Merck K24266586745 >98%

4-Chloro-3-methylnitrobenzene 4Cl3Me 13290-74-9 Fluka 381668/141298 >99%

3-Methyl-4-aminonitrobenzene 4A3Me 99-52-5 Merck 4315751 >98%

Dinitramine Dini 29091-05-2 Riedel 7031851030001 98%

3-Trifluormethyl-4-nitrophenol TFM 88-30-2 Aldrich 04403K6 99%

2,4,6-Trinitrophenol 2OH1,3,5NB 88-89-1 Merck K24266586745 >99%

2,4,5-Trichloronitrobenzene* 2,4,5Cl 89-69-0 Riedel 80820 99,5%

abbr., abbreviations used in the text and figures; * compound with moistened with approx. 30% water

Test organisms and culture conditions

Liquid cultures of the unicellular green alga Scenedesmus vacuolatus Shih. et Krauss, strain 211-15, culture collection Pringsheim (SAG Göttingen, Germany), were grown photoautotrophically at 28 ± 0.5°C in an inorganic, sterilised medium adjusted to pH 6,4 under conditions specified earlier (Altenburger et al., 1990). Cells were synchronised by light:dark changes of 14:10 h and a periodic dilution to a standard cell density of 10⁶ cells/mL before the onset of the light phase of the growth cycle (t₀). Synchronisation was checked by analysis of the cell size distribution at t₀.

Determination of concentration-response relationships

Concentration-response relationships of the test compounds were determined using a 24 h test protocol under synchronised conditions with the population reproduction (cell number) as effect parameter. The initial cell density was set to 10⁵cells/mL. The culture vessels were composed of tubes with an inner diameter of 1.3 cm and a round bottom.

Culture volumes were 8 ml with a headspace of approx. 3 ml. The test medium was the same as for cultivation but enriched with 1.5 mmol/L NaHCO₂ providing a final pH of the medium of 7.0 ± 0.2. The vessels were illuminated by a combination of two types of fluorescent light tubes (L36W/41 Interna, L36W/11 daylight, Osram, Berlin, Germany) with an intensity of 13-18 W/m² (22-33 kLux) providing a photosynthetic active radiation of 350 µE s^-1 m^-2 ± 10 %. The aqueous test substances were added to the cultures at t₀. Aliquot samples of the cultures were taken in duplicates at t₀ and at the end of the standard cycle (t₂₄) and the mean cell number was analysed twice using a CASY II-analyser (Schärfe System, Reutlingen, Germany). The inhibition of cell reproduction was calculated by normalizing the data to the results of control cultures. In order to achieve appropriate estimation of low effect concentrations (<EC₁₀) necessary for the combined effect predictions of multiple mixtures for response addition, concentration-response relationships were calculated using two fundamentally different models, namely either Hill or Weibull models (Scholze et al. 2000), from which the effective concentrations were determined. The concentration-response curves were determined after preliminary range finding, striving to provide at least 4 experimental data points between 20 and 80% inhibition. Concentrations of the aqueous stock solutions of the applied chemicals and achieved concentrations at the time of application (t₀) to the test vessels were verified by HPLC-analysis. A 5 µm reversed phase C8

column (Lichrospher 60 RP select B, Merck, Darmstadt, Germany) or a 5 µm reversed phase C18 column (Lichrospher 100 RP 18-e, Merck, Darmstadt, Germany) were used as stationary HPLC phase, while the mobile phase was for most analyses made up of methanol:water 40:60. All concentration data were corrected for the analytically determined concentration in the test medium at t₀. Slope values were compared for the analysis of parallelism of the concentration-response relationships. We deduced the slope values from the Hill function. The slope of Hill functions may be visualised as the ratio of affected to unaffected fraction, linearized by plotting the log of this ratio against the logarithm of the concentration using equation 1, as is often done in enzymology.

[ ORJ S [ ORJ

\ S \ ORJ [[

\ S

+

−

− =

⇒



 +

= ₋ ⁽³⁾

Modelling of mixture toxicity

For the analysis of the mixture toxicity, combination effects observed experimentally in duplicate were compared with the theoretically predicted effects, calculated according to the concepts of concentration addition and response addition. The predictions according to the concept of concentration addition were derived from the calculation of concentrations of single compounds giving a certain effect EC_yon the basis of logK_OW- driven quantitative structure-activity relationships or on the actually measured parametric concentration-response function. These EC_y values are combined in one multidimensional isobole (equation 2) and this isobole intersected with the mixture vector (equation 3), giving the concentration of the mixture leading to the same effect EC_y(equation 4).

∑

L \L

(&[L

(1)

ND L L L

L L [[

[[ D = ≡

∑

=

∑

L \L ND L

(&D

[ ⁽⁵⁾

with x_i=concentration of derivative i; EC_y,i= concentration of the single component i yielding the effect EC_y; x_ka=concentration of the mixture as sum of the molar

concentrations of the single ingredients; a_i=proportion of component i in x_ka.

The predictions according to response addition were derived using the calculated effects of the compounds applied singly at certain concentrations (using their Hill or Weibull functions) and the calculation of their mixture effect according to:

\

\ ^Q

L

LD L 





 

 −

−

=

∏

=

. ₍₆₎

with y_ia: mixture effect according to response addition; y_i: effect of component i applied singly.

The experimental design for observation of combined effects followed the so-called ray design or equitoxic mixture design, where a dilution series of a fixed ratio of the components, here provided by the individual EC₅₀s, are tested. Full discussion of the benefits and drawbacks of the various designs in combined effect studies may be found elsewhere (Berenbaum 1981, Altenburger, Nendza, Schüürmann 2003), but the design used in this study allows for predictions according to both concentration addition and response addition.

The following procedure was used to calculate the expected joint action of mixtures of chemicals partly acting by similar and partly by dissimilar modes of action: Firstly, the combined effect of the subgroups of all components considered to act similar were calculated according to concentration addition (equation 4). This calculation was performed for a dilution series, which can then in a second step be fitted to a concentration-response function such as the Hill function (see above). Thirdly, the concentration-response-relationships for subgroups of similar acting components were then modelled according to independent action (equation 5) yielding a prediction that joins concentration addition for similar acting components and response addition, for dissimilar, independently acting subgroups.

RESULTS

Single compounds’ toxicities

Exposure of Scenedesmus vacuolatus to nitrobenzenes led to a concentration dependent inhibition of cell reproduction (table 2) at concentrations that are well below the water solubility limits of the compounds. Concentration values were corrected for analytically determined concentrations in the test medium at the beginning of the experiment (t₀).

The concentration-response data show a clear s-shape, demonstrated in figure 1 for the example of 3-methyl-4-chloro-nitrobenzene. Concentration-response relationships were estimated by either Hill functions or Weibull functions, the fit always being satisfactory for both models, deviating only for small (<10%) or high effects (>90%). Based on inspection of the residual plots, the model used for combined effect modelling was chosen. This recalculation of experimental data from Schmitt et al. (2000) allowed for estimation of complete concentration-response functions (table 2) and their variance, which is a prerequisite for modelling the combination effect of multiple substance mixtures according to response addition, as low effect estimates are required (Scholze et al. 2001).

The EC50 and EC10 values calculated using these concentration response functions cover a range of more than 4 log-units (table 2) from 7.5 µg/L as an EC10 for dinitramine to 250 mg/L as the EC50 estimation for 2,4,6-trinitrophenol.

table 2: Parameters of the concentration-response functions for the individual nitrobenzenes

substance log K_OW^a dose response function^c

model parameter^d EC

50 ±95%

confidence intervall

EC50 EC

10 EC

50 e

ECOSAR Nitrobenzenes

EC50 f

Narkosis

Θ₁ Θ₂ [µmol/L] [µmol/L] [mg/L] [mg/L] [µmol/L] [µmol/L]

2,4,6-Trinitrophenol 0,89 W -14,841 4,757 1100 4,12 253 101 5100 21400

4-Aminonitrobenzene 1,39 H 3,32E-04 3,04 332 37,9 45,8 22,3 1900 7920

3-Aminonitrobenzene 1,37 H 2,76E-04 3,22 276 6,14 38,1 19,3 _{200o 8240} Nitrobenzene 1,85 H 2,66E-04 4,05 266 25,7 32,7 19,0 780 3170

2-Chlornitrobenzene 2,24 H 1,53E-04 4,72 153 10,3 24,0 15,1 365 1460

4-Methylnitrobenzene 2,37 W -20,552 9,595 127 10,1 17,4 11,1 _{283 1130} 3-Methyl-4-aminonitrobenzene 1,83 H 9,04E-05 6,54 90,4 3,15 13,7 9,83 814 3300

5-Chloro-2-aminonitrobenzene 2,72 W -13,262 7,392 55,5 6,61 9,58 5,30 143 562

4-Chloronitrobenzene 2,39 W -4,453 3,046 22,3 1,05 3,46 0,845 _{272 1080}

TFM 2,77^b W -6,771 5,252 16,6 1,32 3,43 1,50 130 509

3-Methyl-4-chloronitrobenzene 2,90^b W -4,171 3,414 13,0 0,494 2,23 0,627 100 393

3-Chlornitrobenzene 2,46 W -3,147 3,133 7,72 0,562 1,22 0,305 237 943

2,4,5-Trichloronitrobenzene 3,48 H 1,95E-06 2,79 1,95 0,224 0,442 0,201 32 124

Dinitramine 4,30 H 4,59E-08 3,21 0,0459 0,00324 0,015 0,00746 6,47 24

aHansch, Leo, Hoekman, 1995

bCLOGP, 1999.

c

H : Hill function: _S

[ [ ⁻



 +

\ = , W: Weibull function: y= 100 * ( 1- exp( -exp( a+ b log (x * 10 ⁶)))), with y: percent inhibition of cell

reproduction, and x the concentration in mol/L.

dparameters of the dose response functions: Θ₁, position parameter (X₅₀ in Hill formula; a in Weibull formula); Θ₂, slope parameter (p in Hill formula, b in Weibull formula).

elog 1/EC50=0.885log K_ow + 1.534 (QSAR for nitrobenzenes, US-EPA, 1998)

flog 1/EC₅₀=0,8637log K_ow + 0,9009 (QSAR for narcotics, unpublished results

The variance is in the range of 2 to 12% of the EC₅₀ values and can therefore be considered as comparatively accurate. The established concentration-response relationships for the individual compounds permit calculation of expected mixture effects according to the models of concentration addition and response addition using equations 2 and 5, respectively. In the case of response additive mixture toxicity for multiple mixtures, accurate estimation of small effects is a prerequisite for reliable modelling.

The QSAR-derived EC₅₀ estimations for an unspecific hydrophibicity-driven toxicity are also displayed in table 2. The actual values depend on the model chosen, here the ECOSAR estimates for nitrobenzenes (US-EPA, 1998) and a QSAR derived from own, unpublished data on a series of alcohols, and unpolar benzene derivatives. From these estimates a toxic unit summation can be performed to calculate the effect of a mixture. As no underlying concentration-response curves are taken into account this is restricted to the EC₅₀ level.

Figure 1: Concentration-response data and concentration-response function for 3-methyl-4- chloro-nitrobenzene using a 24 h one generation algal reproduction inhibition test (for details see materials and methods)

FRQFHQWUDWLRQ>PRO/@

LQKLELWLRQRIDOJDOUHSURGXFWLRQ FRQWURO

VROXELOLW\

ZDWHU OLPLWRI

Mode of action of mixture components

In order to assign the nitrobenzenes to a group of assumed common mode of action, two indicators of similar mode of action, namely (i) parallelism of concentration-response functions, and (2) possibility of an unspecific narcotic toxicity were considered.

A tentative criterion for similarity of modes of action in a given biological system based on the analysis of the concentration-response functions was suggested by the NRC (1980), using a comparison of the steepness and parallelism of concentration-response curves for individual mixture components. We therefore performed a linearization of concentration-response curves according to the Hill transformation, as introduced by Chou and Talalay (1984).

Because the focus was set on a comparison of steepness in the area around the EC50 value, only effects between 9% and 91% inhibition were taken into account (-1 – 1 in the linearized Hill plot) (fig 2). We compared the slope values for the individual components and their estimated variance (95% C-I.) with the mean values and the respective variance. The data analysed here show striking similarity except perhaps for 3-methyl-4-aminonitrobenzene. This is true also in comparison to slope data of known dissimilar acting compounds (Faust et al, 2000) and even of congeneric and strictly similar acting compounds (Faust et al. 2001). Thus following Chou and Talalay (1984) the slope information gained from concentration-response analysis, would lead to the expectation of a concentration additive combination.

FRQFHQWUDWLRQ>PRO/@

OR J I

I

' L Q L

& O

& O

& O 0 H 7 ) 0

& O

& O

$ & O

$ 0 H

0 H

& O

1 %

$ $

2 + 1 %

Figure 2: Hill plot and slope values with variances (inset) to illustrate similarity of slopes for biological activities of all 14 nitrobenzenes. Individual substances may be identified by their abbreviation provided in table 2.

The plausibility of unspecific narcotic toxicity as a common mode of action for the nitrobenzenes investigated, was studied using the quality of the correlation of toxicity with lipophilicity. The log K_OW values of the compounds (table 2) cover a range of 4.5 units which is comparable to the range of the observed median effect concentrations. The correlation of nitrobenzene effects on algae with their log K_OW leads to a QSAR equation of reasonable statistical quality when the one dinitrobenzene (dinitramine) is excluded:

- log EC50= 0.98 (±0.14) log K_OW + 2,15 (±0.32) n=13; r²= 0.82; r=0.90; s=0.35; F=49

(7)

As it has been frequently found that narcotic toxicity can be described by the lipophilicity of the compounds alone and regarding the descriptive power of lipophilicity in the QSAR

2+1% $ $ 1% &O

0H

$0H

$&O

&O 7)0

&O0H &O &O 'LQL

analysis performed, there is no tentative criterion to expect other than a similar mode of action and thus a concentration additive combined effect for a mixture of mononitrobenzenes.

However, including parameters of molecular reactivity into QSAR considerations led Schmitt et al. (2000) to propose additional oxidative stress by redox cycling which, according to the authors’ hypothesis of a relevant E_SOMO window, applies to dinitrobenzenes (here dinitramine) as well as for the mononitrobenzenes 2,4,5-Trichloronitrobenzene and 3-Trifluormethyl-4- nitrophenol.

Mixture toxicity

For the analysis of the combination effect of nitrobenzenes, a mixture of 14 derivatives in proportion of their EC50 values was generated. It comprised 13 mononitrobenzenes plus the dinitrobenzene dinitramine. Given the reasonable correlation between observed single compounds algal toxicity and lipophilicity (equation 6) and their similar curve shape (fig. 2) the selected 13 mononitrobenzenes as mixture components may therefore be expected to show combined effects predictable by concentration addition and possibly even based on simple toxic unit summations from baseline toxicity estimates.

When taking into account the possibility that the mononitrobenzenes 2,4,5- Trichloronitrobenzene and 3-Trifluormethyl-4-nitrophenol act as a separate group due to their anticipated redox cycling capacity (Schmitt et al. 2000), these two compounds would form a subgroup with an additional mode of action.

The addition of dinitramine has a clearly distinguishable trait. The compound is a dinitrobenzene derivative with a described specific effect on microtuble assembly from tubulin dimers in plants (Ellis et al. 1994) which is taken to explain its herbicidal properties via mitosis interruption. In contrast to narcosis, this mode of action is not even membrane related and might therefore be taken as sufficient evidence to consider it in the context of the mixture as dissimilar acting.

The results of the subsequent mixture experiments and predicted combined effects for various assumptions according to the concepts of concentration addition and response addition are displayed in figure 3.

FRQFHQWUDWLRQ>PRO/@

LQ KLE LWLR Q RI D OJ DO UH SU RG XF WLR Q

Figure 3: Mixture toxicity of 14 nitrobenzenes compared to the expected combination effects as derived from concentration addition and response addition.

diamonds ( ), experimentally observed algal toxicities; long dashed line (_ _ ), expected concentration additive effects for all 14 compounds; solid line (___), dinitramine independently acting with the 13 other nitrobenzenes; dotted line (...), dinitramine, TFM and 2,4,5-trichloronitrobenzene independently acting with 11 other nitrobenzenes; short dashed line (- - -), all compounds independently acting; 1 designates the toxic unit summation for nonpolar narcosis, 2 depicts the toxic unit summation from the ECOSAR EC50 prediction for nitrobenzenes (US-EPA 1998).

The observed mixture toxicity neither corresponds with the expectations calculated for a toxic unit summation based on a narcotic mode of action nor on a concentration-additive combination effect using parametric effect estimations for the individual compounds but lies almost midway between the predictions derived from the two predictions. Response addition calculated for the unrealistic assumption of complete independent action of all 14 components on the other hand clearly underestimates the observed combined effects. The calculation of errors of both the experimental and predicted data is not straigthforward, but estimations

based on errors determined for repeated single substance testing in the biotest used (Faust et al, 1992) show that the discrimination between both models should be more than enough to exclude experimental error. The experimental data would only come close to either of the predictions of concentration addition or response addition, if 9 of the 14 estimated effect concentrations would include a systematic one sided error of 100%.

As it is a common problem in mixture toxicity experimentation that individual components might drive the effect of mixtures, Figure 4 displays exemplarily results of a sensitivity analysis for three compounds, whereby we seek to identify whether individual components of the mixture dominate the observable combination effect. As depicted for dinitramine, TFM and nitrobenzene the distance in terms of concentration between the concentration-response relationship determined for these components and the effects of the mixture displayed at the concentrations of these components in the mixture are almost one order of magnitude apart.

This is true for all mixture components analysed and indicates that no single component is expected to dominate the observed mixture toxicity.

Excluding experimental error and design problems as possible explanations for the observed combined effect, there is room to consider the possible effect of different modes of action of the mixture components on the observed combined effect. As laid out in detail in the methods section, stepwise calculation of expected joint mixture toxicity for subgroups of components with anticipated similar and independent modes of action were performed. Firstly, we considered the expected mixture effect when treating dinitramine as independently acting, due to its reported specific herbicidal properties (modelled straight line in figure 3)(Ellis et al.

1994). Though there is a clear shift towards higher expected effect concentrations, the observed effects are not fitted using this joint model. Secondly, we additionally treated the mononitrobenzenes 2,4,5-Trichloronitrobenzene, and 3-Trifluormethyl-4-nitrophenol as a separate group due to their anticipated redox cycling capacity (Schmitt et al. 2000). This resulted in a joint mixture toxicity prediction generated from 11 components calculated by concentration addition due to their expected narcotic effect, and from 2 components expected to behave concentration additive with respect to redox cycling. These were joined with the specific action expected for dinitramine as three distinctly independent effects, using response addition for modelling (dotted line in figure 3). Again a clear upward shift in expectable combined effect concentrations can be seen. There is still no perfect fit of observed data but the difference between prediction and observation is now less than a factor of 1.5.

DISCUSSION

The algal toxicities of the nitrobenzenes reported in this paper are well in accordance with data produced by Kramer et al. 1986 and Deneer et al. 1989 for 8 components investigated in these studies as well, despite the different species and exposure regimes that have been used (Chlorella vulgaris, 6 h exposure in Kramer et al. 1986; Chlorella pyrinoidosa, 96 h exposure in Deneer et al. 1989).

Mode of action considerations

The QSAR for the algal toxicity of nitrobenzenes given in equation (6) compares well to the one of Deneer et al. (1989) for mononitrobenzenes (- log EC50 = 0,90 log KOW + 2,03). Thus in relation to hydrophobicity there is no reason to raise doubts about a common narcotic mode of action of at least mononitrobenzenes in algae. The analysis of reactivity parameters, however, generated a modified expectation, as it was shown that quantum chemical decriptors and in particular the energy of the singly occupied molecular orbital of the radical anions (ESOMO) for some nitrobenzenes compared well with that of known redox cyclers, suggesting that dinitrobenzenes and multiply chlorinated nitrobenzenes may also exert oxidative stress (Schmitt et al. 2000). In turn this would lead to an expectation for the combined effect of such a mixture that would be given by joining concentration and response addition. However, when components with dissimilar modes of action are mixed at low individual concentrations they are often expected to elucidate combined effects only due to their relative contributions to a narcotic mode of action, which should again result in a concentration-additive behaviour (Altenburger, Nendza, Schüürmann 2003).

Parallelism of dose-response curves of components has been interpreted in terms of being an indicator of similar biological action and thus a predictor of concentration-additive mixture effects in combined effect assessment (Sühnel 1998). In our case, a Hill-linear comparison of the parallelism of the concentration-effect curves of the single compounds shows that the steepness of all curves do not deviate substantially, except for 3 methyl-4-aminonitrobenzene.

Therefore, again as for the hydrophobicity considerations there is no reason to challenge the notion of a similar mode of action of nitrobenzenes in algae.

Mixture toxicity

Our finding that the algal mixture toxicity of 14 nitrobenzenes is not well predicted by concentration addition is surprisingly evident and stands in contradiction to the interpretation of the hydophobicity-based QSAR and the parallelism argumentation derived from single concentration-response curves. Könemann (1981) and Hermens and coworkers (1984a,b, 1985a) have shown several times for different organisms, that for mixtures of substances with a narcotic type of action, concentration addition provides good predictions for expectable mixture toxicity, although no comparison with the predictions for independent action were undertaken. On the basis of these findings it has been suggested to apply the summation of toxic units to analysed concentrations of components to provide a reasonable mixture toxicity estimation (Dyer et al. 2000, Swartz et al. 1995). Consideration of interactive effects as an explanation for the findings reported here will be discussed at first.

Synergism or antagonism has sometimes been discussed to explain deviations from mixture toxicity concepts (Escher et al. 2001), but these terms have been rarely used to describe deviations from concentration addition occurring at organism level (Hermens, 1985b, Broderius and Kahl, 1985). For multiple mixtures of specifically acting substances (pesticides) using algal and bacterial biotests, it has been demonstrated that concentration addition allows accurate prediction for mixtures of similar acting components (Faust et al.

2000, Altenburger et al. 2000), while for mixtures composed of strictly dissimilar acting chemicals the same holds true for independent action (Faust et al. 2000, Backhaus et al. 2000).

In order to to investigate potential interactive effects, it would require a specific hypothesis as to what is the reason and extent of the interaction. Otherwise, if in our case synergistic or antagonistic effects were present, it would not be possible to differentiate between a concentration additive mixture toxicity with antagonistic interference and an independently acting mixture with synergistic interaction. Besides, even for mixtures being in good agreement with one of the concepts, the possibility of synergistic or antagonistic effects based on the contrasting concept cannot be ruled out. In our case, there is no reported or obvious pharmacological reason to believe in any type of non-additive interaction for the nitrobenzene mixture components.

In line with Berenbaum (1981), it might be more helpful to first try a rigorous quantification of expected non-interactive effects before considering interaction. A first step to that end is

the consideration of combined effects for dilution curves of mixtures instead of point estimates for a toxic unit summation of EC₅₀ fractions as is commonly done. Our findings show a consistent and monotonous concentration-combination effect curve, and according to the chosen ray design all components should contribute significantly to the overall effect.

Therefore, one may consider the involvement of different modes of action, i.e. some components of the mixture may act similar and some dissimilar. From the mixtures composition we know that dinitramine has a strikingly different trait being identified as a microtubuli assembly interruptor and thus mitosis inhibitor (Ellis et al. 1994). When treating dinitramine separately, thus calculating the combined effect of the other 13 substances according to concentration addition and thereafter the combined effect of this group and dinitramine from response addition, the expected combination effect is shifting with regard to concentration addition by only 18%. This is only visible as a slight shift in the concentration response curve (fig 3), however it does not suffice to explain the observed combined effects.

If it is assumed that more nitrobenzenes deviate from the common narcotic mode of action, the question arises how many and which. Again, several hypotheses seem reasonable. First, several other investigations of nitrobenzenes found differences in the behaviour of derivatives with different substitution patterns. While Kramer et al. (1986), Dearden et al. (1995) and Schüürmann et al (1997) noted a deviating behaviour of para-substituted derivatives regarding their toxicity to Chlorella vulgaris and Tetrahymena pyriformis, respectively, Hall and Kier (1986) as well as Bailey and Spanggord (1983) found different effects when comparing ortho- and para- with meta-substituted nitrobenzenes in fathead minnows. From the inspection of residuals of a hydrophobicity-based QSAR, no conclusion can be drawn regarding a systematic deviation of either meta- or para-derivatives, though. Inclusion of quantum chemical descriptors and in particular the energy of the singly occupied molecular orbital, E_SOMO, of the radical anions as initial metabolites for the nitrobenzene effects led to the suggestion that some of the compounds may also exert oxidative stress (Schmitt et al. 2000).

The potential for redox cycling is tentatively proposed to apply for three of the mixture components, namely TFM, 2,4,5-trichloronitrobenzene and dinitramine.

Figure 4: Mixture toxicity of 14 nitrobenzenes plotted against the concentrations of three components at their concentrations in the mixture and compared with their individual concentration-response relationships. Triangles, up ( ), dinitramine;

stars ( ), 3-trifluormethyl-4-phenol; triangles, down ( ), nitrobenzene.

If we now recalculate expectations for combined effects under the assumption that these three compounds act dissimilar, while the remaining 11 mononitrobenzenes act according to concentration addition, we obtain an altered reference curve as depicted in figure 4. In terms of expected concentrations necessary to evoke the same effect an increase of about 40 % can be calculated and the predicted concentration-response curve clearly shifts towards the observed effects. This improvement in predictability of observed effects could be interpreted as an indication that other modes of action in addition to narcosis indeed play a role. Still, an unresolved difference remains that might indicate that further compounds act independently or that more dissimilar modes of action of the nitrobenzenes might occur.

,QKLELWLRQRIDOJDOUHSURGXFWLRQ>@

FRQFHQWUDWLRQ>PRO/@

QLWUREHQ]HQH

GLQLWUDPLQH 7)0

Mixture toxicity vs. QSAR and parallel concentration response curves

The lipophilicity-based QSAR results and the strikingly similar concentration-response curves suggest one common mode of action (narcosis), while there are signs that the deviation from concentration additive mixture toxicity might be caused by differing modes of action.

First, the definition of "similar mode of action" may be crucial for this seemingly contradictory expectations gained from mixture toxicity experiments and QSAR techniques.

Although the terms of mode of action do vary widely in between the mixture toxicity and the QSAR context, two contrasting approaches may be highlighted. On the one hand, e.g. Pöch (1993) regards a primary, specific, reversible and competitive interaction with one identical molecular receptor as a necessary prerequisite for substances to behave concentration additive, i.e. act similar. On the other hand, the existence of one common QSAR alone is regarded as sufficient evidence for similar action (Könemann, 1981). It may therefore be possible that the nitrobenzenes act similar according to a QSAR definition of mode of action, but dissimilar when looked upon in the mixture toxicity context.

Another explanation for the different results regarding the homogeneity of the modes of action as concluded from a QSAR and from mixture toxicity experiments might be derived from the variety of processes involved in the toxic effect which cannot be separated in the QSAR approach. If for example a toxifying biotransformation step yields a product which is again quickly biotransformed to a less toxic substance, both processes act in opposite directions regarding the "overall" toxicity as judged by a QSAR based on the lipophilicity of the parent compound. This substance therefore might not be detected as deviating from the QSAR although its mode of action is not similar to that of nitrobenzenes which do not undergo biotransformation.

It remains to be stated that the debate is still open whether results of mixture toxicity experiments may be used for the assignment of substances to modes of action. As detailed knowledge of the molecular mechanisms of toxicity is often lacking, no proof for the possibility of a conclusion from mixture toxicity to modes of action, neither generally nor specifically, has yet been given. Up to now, only the opposite direction of conclusions, i.e.

from modes of action to mixture toxicity, seems to be reasonably established.

In summary, an analysis of the single and mixture toxicity of nitrobenzenes to the green alga Scenedesmus vacuolatus was undertaken. As expected from the literature, a hydrophobicity- based QSAR for mononitrobenzenes leads one to assume that nitrobenzenes may exert narcotic effects as a common type of action, while dinitrobenzenes show greater toxicity.

Redox cycling capacity of some components may be an additional toxicity factor. Results obtained investigating the toxicity of a mixture of nitrobenzenes neither follow the predicted effects derived from the concept of concentration addition nor the concept of response addition. Instead, the combination toxicity may be better described by calculating concentration additive combined effect for subgroups with anticipated similar modes of action and joining independently acting groups using the model of response addition. Improvement of the mixture toxicity prediction upon assignment of redox cycling as additional mode of action could be used to support the notion that some mononitrobenzenes show effects beyond narcosis. In conclusion, combined effect analysis might be used as a pharmacologically probe for the plausibility of a common mode of action of mixture components.

LITERATURE

Altenburger R, Bödeker W, Faust, M Grimme H. 1990. Evaluation of the isobologram method for the assessment of mixtures of chemicals, Ecotoxicol Environ Safety 20:98-114 Altenburger R, Backhaus T, Boedeker W, Faust M, Scholze M, Grimme LH. 2000.

Predictability of the toxicity of multiple chemical mixtures to Vibrio fischeri: Mixtures composed of similarly acting chemicals. Environ Toxicol Chem 19:2341-2347

Altenburger R., Nendza M., Schüürmann G. 2003. Mixture toxicity and its modeling by quantitative structure-activity relationships. Environ Toxicol Chem22:1900-1915.

Backhaus T, Altenburger R, Boedeker W, Faust M, Scholze M, Grimme LH. 2000.

Predictability of the toxicity of a multiple mixture of dissimilarily acting chemicals to Vibrio fischeri. Environ Toxicol Chem 19:2348-2356

Bailey H, Spanggord R. 1983. The relationship between the toxicity and structure of nitroaromatic chemicals. Aquatic Toxicology and Hazard Assessment ASTM STP 802:98-107

Berenbaum M C. 1981 Criteria for Analysing Interactions between Biologically Active Agents. Adv.Cancer.Res 53:269-335

Bliss CI. 1939. The toxicity of poisons applied jointly. Ann Appl Biol 26: 585-615 Broderius SJ, Kahl MD. 1985. Acute toxicity of organic chemical mixtures to the fathead

minnow. Aquat Toxicol 6:307-322

Broderius SJ, Kahl MD, Hoglund MD. 1995. Use of joint toxic response to define the primary mode of toxic actionn for diverse industrial organic chemicals. Environ Toxicol Chem 14:1591-1605

Calamari D, Vighi M. 1992. A proposal to define water quality objectives for aquatic life for mixtures of chemical substances. Chemosphere 25:531-542

Chou T-C, Talalay P. 1984. Quantitative analysis of dose-effect relationships: The combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul 22:27-55

CLOGP, version 4.61 1999. Daylight Chemical Information Systems, Irvine, CA.

Dearden JC, Cronin MTD, Schultz TW, Lin DT. 1995. QSAR study of the toxicity of nitrobenzenes to Tetrahymena pyriformis. QSAR 14:427-432

Deneer JW, Sinnige TL, Seinen W, Hermens JLM. 1987. Quantitative structure-activity relationships for the toxicity and bioconcentration factor of nitrobenzene derivatives towards the guppy (Poecilia reticulata). Aquatic Toxicology 10:115-129

Deneer JW, van Leeuwen CJ, Seinen W, Maas-Diepeveen JL, Hermens JLM. 1989. QSAR study of the toxicity of nitrobenzene derivatives towards Daphnia magna, Chlorella pyrenoidosa and Photobacterium phosphoreum, Aquat Toxicol 15:83-98

Dyer, SD; White-Hull, CE; Shephard, BK. 2000. Assessments of chemical mixtures via toxicity reference values overpredict hazard to Ohio fish communities. Environ Sci Technol 34: 2518-2524

Ellis JR, Taylor R, Hussey PJ. 1994. Molecular modeling indicates that two chemically distinct classes of anti-mitotic herbicide bind to the same receptor site(s). Plant Physiol 105:15-18

Escher BI, Hunziker RW, Schwarzenbach RP. 2001. Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects. Environ Sci Technol 35:3905-3914.

Faust M, Altenburger R, Bödeker W, Grimme, LH. 1992. Algentoxizitätstests mit

synchronisierten Kulturen. In XXX Autor, Schriftenreihe WaBoLu 89, Gustav-Fischer- Verlag, Stuttgart, Germany, pp 311-321

Faust M, Altenburger R, Backhaus T, Boedeker W, Scholze M, Grimme LH. 2000. Predicitve assessment of the aquatic toxicity of multiple chemical mixtures. J Env Qual 29:1063- 1068

Faust M, Altenburger R, Backhaus T, Blanck H, Boedeker W, Gramatica P, Hamer V,

Scholze M, Vighi M, Grimme LH. 2001. Predicting the joint algal toxicity and of multi- component s-triazine mixtures at low-effect concentrations of individual toxicants.

Aquatic Toxicol 56:13-32

Gough KM, Belohorcová K, Kaiser K. 1994. Quantitative structure-activity relationships (QSARs) of Photobacterium phosphoreum toxicity of nitrobenzene derivatives. Sci Tot Environ 142:179-190

Greco W, Unkelbach H-D, Pöch G, Sühnel J, Kundi M, Bödeker W. 1992. Consensus on concepts and terminology for combined action assessment: The Saariselkä agreement.

Archives of Complex Environmental Studies 4(3):65-69

Greco W, Bravo G, Parsons JC. 1995. The search for synergy: A critical review from a response surface perspective. Pharmacol Rev 47:331-385

Hall LH, Kier LB. 1986. Structure-activity relationship studies on the toxicities of benzene derivatives: II. An analysis of benzene substituent effects on toxicity. Environ Toxicol Chem 5:333-33

Hansch C, Leo A, Hoekmann D. 1995. Exploring QSAR. [2] Hydrophobic, electronic, and steric constants. American Chemical Society, Washington, D.C.

Hermens J, Broekhuyzen E, Canton H, Wegman R. 1985a. Quantitative structure activity relationships and mixture toxicity studies of alcohols and chlorohydrocarbons: Effects on growth of Daphnia magna. Aquatic Toxicology 6:209-217

Hermens J, Canton H, Janssen P, De Jong R. 1984a. Quantitative structure-activity

relationships and toxicity studies of mixtures of chemicals with anaesthetic potency:

acute lethal and sublethal toxicity to Daphnia magna. Aquat Toxicol 5:143-154

Hermens J, Leeuwangh P, Musch A. 1984b. Quantitative structure-activity relationships and mixture toxicity studies of chloro- and alkylanilines at an acute lethal toxicity level to the Guppy (Poecilia reticulata). Ecotox Environ Safety 8:388-394

Hermens J, Leeuwangh P, Musch A. 1985b. Joint toxicity of mixtures of groups of organic aquatic pollutants to the Guppy (Poecilia reticulata). Ecotox Environ Safety 9:321-326 Könemann H. 1981. Fish toxicity tests with mixtures of more than two chemicals: A proposal

for a quantitative approach and experimental results. Toxicology 19:229-238

Kortenkamp A, Altenburger R. 1998. Synergisms with mixtures of xenoestrogens - a reevaluation using the method of isoboles. Sci Tot Environ 221:59-73

Kortenkamp A, Altenburger R. 1999. Approaches to assessing combination effects of oestrogenic environmental pollutants. Sci Tot Environ 233:131-140

Kramer CR, Trümper L, Berger L. 1986. Quantitative Struktur-Wirkungs-Beziehungen für die Hemmung des autotrophen Wachstums synchroner Chlorella vulgaris-Kulturen durch monosubstituierte Nitrobenzene, Biochem. Physiol. Pflanzen 181:411-420

Loewe S. 1927. Die Mischarznei. Versuch einer allgemeinen Pharmakologie der Arzneikombinationen. Klin Wochenschr 6:1077-1085

Loewe S. 1953. The problem of synergism and antagonism of combined drugs. Arzneim- Forsch / Drug Res 3: 285-290

Loewe S. Muischnek H. 1926. Ueber Kombinationswirkungen. 1. Mitteilung: Hilfsmittel der Fragestellung. Naunyn-Schmiedebergs Arch Exp Pathol Pharmakol 114:313-326

NRC (National Research Council). 1980. Principles of toxicological interactions associated with multiple chemical exposures. Washington, D.C., National Academy Press.

Plackett RL, Hewlett PS. 1952. Quantal responses to mixtures of poisons. J R Statist Soc B 14:141-163

Pöch G. 1993. Combined effects of drugs and toxic agents. Springer Verlag, Wien, Austria Rand G, Wells P, McCarty L. 1995. Introduction to aquatic toxicology, in: G. Rand (ed.),

fundamentals of aquatic toxicology, Taylor & Francis, Washington, 3-70

Roberts DW. 1987. An analysis of published data on fish toxicity of nitrobenzene and aniline derivatives. In Kaiser KLE, ed, QSAR in Environmental Toxicology - II, D Reidel, Dordrecht, Netherlands, pp 295-308

Schmitt H, Altenburger R, Jastorff B, Schüürmann G. 2000. Quantitative structure-activity analysis of the algae toxicity of nitroaromatic compounds. Chem Res Toxicol 13:441- 450

Scholze, M., Boedeker, W., Faust, M., Backhaus, T., Altenburger, R., Grimme, L.H. 2001. A general best-fit method for concentration-response curves and the estimation of low- effect concentrations. Environ Toxicol Chem 20, 448-457.

Schüürmann G, Flemmig B, Dearden JC, Cronin MT, Schultz TW. 1997. CoMFA study of acute toxicity of nitrobenzenes to Tetrahymena pyriformis. In Chen F, Schüürmann G, eds, QSAR in environmental sciences VII, SETAC Press, Pensacola, FL, USA, pp 315- 328

Schüürmann G. 1998. Ecotoxic modes of action of chemical substances, in: G. Schüürmann, B. Markert (eds.), Ecotoxicology, Wiley, Heidelberg, 665-751

Schultz T W, Moulton B A. 1985. Structure-activity relationships for nitrogen-containing aromatic molecules. Environ Toxicol Chem 4:353-359

Sühnel J. 1998. Parallel dose-response curves in combination experiments. Bull Mathematical Biology 60:197-213

Swartz RC, Schults DW, Ozretich RJ, Lamberson JO, Cole FA, DeWitt TH, Redmond MS, Ferraro SO. 1995. ΣPAH: A model to predict the toxicity of polynuclear aromatic hydrocarbon mixtures in field collected sedimentss. Environ Toxicol Saf 14:197-1987 Traas, T.P., van de Meent, D., Posthuma, L., Hamers, T.H..M., Kater, B.J., de Zwart, D.,

Aldenberg, T. 2002. The potentially affected fraction as a measure of ecological risk. in:

Posthuma, L., Suter, G.W., Traas, T.P. (eds.) The use of species sensitivity distributions (SSD) in ecotoxicology. CRC Press pp. 315-344.

Vaes W H J, Urrestarazu Ramos E, Verhaar H J M, Hermens J L M. 1998. Acute toxicity of nonpolar versus polar narcosis: Is there a difference? Environ Toxicol Chem 17:1380- 1384

Urrestarazu Ramos E, Vaes W H J, Mayer P, Hermens J L M. 1999. Algal inhibition of Chlorella pyrenoidosa by polar narcotic pollutants: toxic cell concentrations and QSAR modeling. Aquat Toxicol 46:1-10

US-EPA (United States - Environmental protection Agency). 1998. Programme ECOSAR, routine for nitrobenzenes and dinitrobenzenes. provided by V. Nabholz.

CURRICULUM VITAE HEIKE SCHMITT

2001-2005 PhD position at the Institute for Risk Assessment Sciences, University of Utrecht, in collaboration with the National Institute of Public Health and the Environment, RIVM

“Environmental effects of veterinary pharmaceuticals”

2001-2005 Postgradual education in toxicology, the Netherlands

1998-2004 Postgradual education in toxicology and environmental protection, Leipzig

1998 Diplom in chemistry

1998 Final research stage, Umweltforschungszentrum Halle-Leipzig 1994-1998 Chemistry studies, University of Bremen, Germany

1994 Vordiplom in chemistry

1992-1994 Chemistry studies, University of Hamburg

Publications

2004

2000

Schmitt, H. et al: Pollution-induced community tolerance of soil microbial communities caused by the antibiotic sulfachloropyridazine, Environmental Science& Technology 38, 1148-1153

Schmitt, H. et al: Quantitative structure-activity analysis of the algae toxicity of nitroaromatic compound, Chemical research in toxicology 13, 441-450

1998 Schmitt, H.: Prognose und Interpretation von ökotoxikologischen Wirkungen anhand von quantitativen Struktur-Wirkungs-Beziehungen, Diplomarbeit, University of Bremen