Ecotoxicological Effect And Structure-Activity Relations Of Selected Toxic Organic Pollutants

As the core technique in the predictive toxicology, Quantitative Structure-Activity Relationships(QSARs) provide valuable and reliable tool for Ecological Risk Assessment (ERA) of toxic organicpollutants that had been increasingly released into the environment. In the organic pollutant chemistry andecotoxicology, QSARs functioned in two ways: to predict, assess and pre-screen the environmentalbehavior and ecotoxicity of the untested chemicals, and to explore the mechanism of toxic action ofpollutants, which are believed to be the fundamental and theoretical basis for the pollution controltechniques and risk minimization measures. Obviously, successful QSAR models should be both highlypredictive and mechanistically significant. The selection and acquisition of physically significant moleculardescriptors that encode enough and accurate structural information are key work in the developing ofQSARs. Currently, the development of QSARs is evolving from those that were formulated by classicalphysico-chemical properties to that rely more and more on quantum chemical derived molecular descriptors.Quantum chemical derived descriptors possess obvious and protruding advantages because they rely on noexperimentation, encode enough and accurate electronic structural information, bear significant physicalinterpretations and are not restricted to congeners and/or structural related chemicals. These advantagesmake quantum chemical derived descriptors ideal parameters that will help to realize the purpose ofpollution prevention and be in favor of exploring the mechanism of toxic action of xenobiotics anddeveloping mechanistically significant QSARs. For these reason, most research work on the developmentof QSARs in this thesis is based on the use of quantum chemical derived descriptors.Terrestrial macrophytes are primary producers and play key role in ecological system. Therefore,information on phytotoxicity should be included as one necessary part of the ecological risk assessment oforganic pollutants. Amphibians are typical species bridging aquatic organisms and terrestrial animals, arealso important in agriculture and ecosystem. However, few studies on the toxic effect of organic pollutantson higher plants and amphibians, the modes of toxic action, and toxicological effect based QSARs areavailable in the literatures. In this thesis, based on the inhibition of organic pollutants to germination rateand root elongation of Cucumis sativus, a fast, sensitive and cost-effective short-term phytotoxicity assayprocedure was developed and evaluated in the sensitivity, reproducibility and suitability. With itsapplication, phytotoxicities of substituted phenols, substituted anilines, nitrogen-containing aromatics andnitroaromatics were experimentally determined and QSAR analysis were applied to systematicallyinvestigate the mode of action of the phytotoxicity of these organic pollutants and the phytotoxicityinfluencing structural factors. These resulted in a batch of highly predictive and mechanistically significantQSAR models. In addition, the acute lethal toxicities of substituted phenols to larva of amphibian-Ranajaponica tadpoles were tested and highly predictive mechanism based QSARs were developed with theapplication of QSAR technique based on quantum chemical derived descriptors.In general, organic pollutants can be classified into two categories: narcotics and bio-reactivechemicals. In comparison with the relatively well developed QSAR for narcotics, the development ofQSARs based on bio-reactive chemicals are far from success due to the hindrance on the quantitativecharacterization and parameterization of the bio-reactive molecular structure. In this thesis, according to thedifferent contribution of hydrophobicity and bio-reactivity to the toxicity of chemicals that elicited theirtoxic effect via different mode of action, hydrophobicity, which was characterized by 1-ocatanol/waterpartition coefficient in logarithm form (logKow), was utilized as a probe to classify organic pollutants andquantum chemicals derived descriptors, energy of the lowest unoccupied molecular orbital (ELUMO), wasemployed to characterize the electrophilic reactivity and to predict the toxicity of chemicals contributed bybio-reactivity. Then mechanism based QSARs were developed. The research results indicated that thephytotoxicity of nitrogen-containing aromatics are bioreactive and can be predicted by ELUMO successfully.With the application of mechanism based QSAR approach, the acute lethal toxicity of 31 substitutedphenols to Rana japonica tadpoles, the short term phytotoxicity of 42 substituted phenols (the inhibitioneffect on germination and root elongation of Cucumis sativus) and the growth inhibition toxicity of 78substituted benzenes (including phenols, anilines, chloro-and methyl substituted benzenes, substitutednitrobenzenes, substituted benzonitriles, benzaldehydes and benzoic acid derivatives) to yeast(Saccharomyces cerevisiae) were investigated, respectively. Different modes of toxic action involved weresuccessfully classified and highly predictive mechanism based QSAR models were developed.Various classes of organic chemicals and diverse molecular structures often indicate complexity andmultiplicity of toxic modes of action. Thus developing QSARs that are both mechanistically significant andhighly predictive for diverse classes of chemicals that involved multiple or mixed mechanisms of toxicaction will be advantageous and preferable. In this thesis, a generic two-variable QSAR model,Response-Surface Analysis that based on hydrophobic descriptor (logKow) and parameters characterizingelectrophilic reactivity (ELUMO) was deduced based on Critical Volume Theory, Receptor-Ligand InteractionTheory and Frontier Molecular Orbital Theory to predict chemicals involved multiple mechanisms of toxicaction. With it, simple, mechanistically significant and highly predictive QSAR models were developed.These QSARs successfully model and predict the short-term phytotoxicity of substituted phenols involvedboth polar narcosis and bio-reactivity, the growth inhibition toxicity to Saccharomyces cerevisiae ofsubstituted nitrobenzenes with mixed mechanisms of toxic action based on electrophilicity and the growthinhibition toxicity to Saccharomyces cerevisiae of large set of substituted benzenes with complex molecularstructures and multiple mechanisms of action.Large amount of data on ecotoxicological effect were required to perform the ecological and humanhealth risk assessment of organic pollutants. Investigation on interspecies correlation and surrogateorganisms is another potential effective tool to acquire these toxicological data (especially for higherorganisms and human beings). It is still unclear if a promising interspecies correlation could be built forbio-reactive chemical because the toxicity of bioreactive chemicals are mainly determined by their specialin vivo interaction with bio-macromolecules. In this thesis, the acute lethal toxicity to Rana japonicatadpoles, the short-term phytotoxicity to Cucumis sativus and the population growth inhibition toxicity toTetrahymena pyriformis of selected substituted phenols with multiple mechanisms of toxic action werecompared and interspecies correlation was investigated. Similar mechanisms were found for phenols in thethree assay systems. The results indicated that most phenols elicited their toxic response via polar narcosisand some are bioreactive chemicals based on different mechanisms. Highly significant interspeciescorrelations were observed not only for polar narcotics but also for bioreactive chemicals and it is feasibleto predict or estimate the toxicity to higher organisms from those to primary organisms, thus provide ashortcut to acquiring the vast amount of data needed for risk assessment of chemicals.Molecular hologram is a newly developed molecular structure descriptors, based on which a highlypredictive QSAR approach has been developed recently. with the application of Partial Least Square (PLS)technique and the input need of only 2-D molecular structure, it overcomes the difficulty of conformeralignment for 3D-QSAR techniques and avoids the multiple collinearity for traditional Multiple LinearRegression. In addition, all the molecular descriptors employed were created automatically and veryquickly, which make it a promising screening tool for large set of organic pollutants.Polychlorinated biphenyls (PCBs) are highly persistent and toxic organic substances, and are extremelyhydrophobic, bio-accumulative, hard to bio-degradation and toxic. Of all 209 congeners, more than 150congeners have been detected in environment. It is valuable to study the physico-chemical propertiesrelated to their persistence. Due to the lack of standards for single congener and the high similarity amongcongeners, they are difficult to separate with HPLC and GC. Currently the physico-chemical properties,especially those related to their hydrophobicity are scarce for some congeners. In this thesis, WaterSolubility, 1-Octanol/water Partition Coefficient, molecular Total Surface Area, Aqueous ActivityCoefficient, Henry Law's Constant and gas chromatogram relative retention times for biphenyl and its 209chlorinated congeners were studies with holographic QSAR techniques and highly predictive QSARmodels were developed. An internal validation (cross-validation) and external validation were performed toevaluate the predictivity and robustness of the HQSAR models. Based on these highly predictive QSARs,the corresponding data of 5 properties of all these pollutants were presented.Nitroaromatics are highly toxic organic pollutants with very complex and diverse sources and arepotentially highly toxic to human health. It is very important to develop predictive models based on QSARtechniques. In this thesis, QSAR models for the mutagenicities of 219 nitroaromatics and nitropolyaromatic hydrocarbons (Ames assay, S. typhimurium TA 98) were developed with Response-SurfaceAnalysis based on hydriphobicity and molecular orbital parameter, Comparative molecular Field Analysis(CoMFA) based on 3D-molecular modeling and holographic QSAR techniques. The mechanism of toxicaction and the key metabolic activation path were discussed. The predictive power and mechanisticalsignificance were compared among different categories of QSAR techniques. The results showed that themutagenicity of nitroaromatcs involved at least bio-transportation and bio-reduction of nitro group andwere also significantly affected steric effect. Thus considering all these factors resulted in a promisingresponse-surface based QSAR. Predictive QSAR model based on CoMFA can be developed with physicalsignificance. Highly predictive models were developed with QSAR technique based on molecularhologram. With cross-validation and external validation procedure, the HQSAR model proved to be highlypredictive and robust and be a promising predictive tool.Generally, all the research work involved in this thesis were performed on the basis of the applicationof some theoretically derived QSAR model and the application of quantum chemical derived parametersand other newly developed QSAR techniques such as CoMFA and HQSAR, thus the QSARs achieved areboth theoretically reasonable and practical. Undoubtedly, all these achievements will be useful to theecological risk assessment of organic pollutants, to our exploring and understanding the mechanism of theenvironmental process and the toxic action of organic pollutants.