Exploring the role of quantum chemical descriptors in modeling acute toxicity of diverse chemicals to Daphnia magna

https://doi.org/10.1016/j.jmgm.2015.06.009Get rights and content

Highlights

  • Expression of acute toxicity is analyzed at the level of electron-dynamics.

  • Real predictivity of acute toxicity of 252 diverse organic chemicals is tested.

  • Total energy and HOMO energy proposed as best simulators of acute toxicity.

  • Electron-correlation energy found to be a good predictor of acute toxicity.

  • Advanced semi-empirical method PM7 observed to be highly reliable.

Abstract

Various quantum-mechanically computed molecular and thermodynamic descriptors along with physico-chemical, electrostatic and topological descriptors are compared while developing quantitative structure–activity relationships (QSARs) for the acute toxicity of 252 diverse organic chemicals towards Daphnia magna. QSAR models based on the quantum-chemical descriptors, computed with routinely employed advanced semi-empirical and ab-initio methods, along with the electron-correlation contribution (CORR) of the descriptors, are analyzed for the external predictivity of the acute toxicity. The models with reliable internal stability and external predictivity are found to be based on the HOMO energy along with the physico-chemical, electrostatic and topological descriptors. Besides this, the total energy and electron-correlation energy are also observed as highly reliable descriptors, suggesting that the intra-molecular interactions between the electrons play an important role in the origin of the acute toxicity, which is in fact an unexplored phenomenon. The models based on quantum-chemical descriptors such as chemical hardness, absolute electronegativity, standard Gibbs free energy and enthalpy are also observed to be reliable. A comparison of the robust models based on the quantum-chemical descriptors computed with various quantum-mechanical methods suggests that the advanced semi-empirical methods such as PM7 can be more reliable than the ab-initio methods which are computationally more expensive.

Introduction

Daphnia magna, a type of water-flea, is a widely used laboratory animal for the testing of ecotoxicity, and it has been a subject for modeling the toxic effects of diverse chemicals through quantitative structure–activity relationships (QSARs) [1], [2], [3], [4]. The toxicity in water leads to the demise of daphnids, and this perturbs the food chain because daphnids serve as food for many aquatic organisms. In fact, chemicals which are toxic to D. magna can, directly or indirectly, cause toxicological effects at all the trophic levels. Therefore, the thirst for the externally predictive QSAR models for the acute toxicity of diverse chemicals towards the D. magna had been continued since many years [5], [6], [7], [8]. In the literature [4], [9], [10], [11], [12], the toxic effects of various hazardous chemicals towards D. magna, have been modeled mainly through the physico-chemical descriptors like octanol/water partition coefficient (log P) representing the hydrophobicity, besides several topological and quantum-chemical descriptors particularly the energy of highest occupied molecular orbital (HOMO) [9].

It should be noted that the present work assesses the QSAR models based on regression to make quantitative prediction for the toxicity of diverse chemicals towards D. magna. However, there have been equally important advanced and sophisticated QSAR approaches based on the classification method [13] which can effectively predict whether a chemical is biologically active or inactive. Recently [14], [15], [16], [17], [18], [19], [20], [21], such classification based QSAR models have been reported for predicting the toxicity, of large and heterogeneous datasets of compounds, against many organisms, besides assessing multiple toxicological profiles under diverse experimental conditions. For example, Tenorio-Borroto et al. [14], [15], [16], had proposed multi-target quantitative structure–activity/property relationships (mt-QSAR/QSPR) models along with the flow cytometry analysis for the prediction of cytotoxicity and immunotoxicity, which can effectively models the drug-target interactions and effects of organic compounds over the cellular and molecular targets of immune system. Moreover, such techniques can be important for the high throughput screening of drugs to elucidate the drug discovery processes. Besides this, a topological and structure based approach commonly referred as TOPS-MODE approach [17] has also gained popularity to develop the mt-QSARs for the identification of compounds as a drug, pesticide, herbicide etc. Recently, this approach has been applied for developing the mt-QSAR for tyrosine kinase inhibitors [18].

In fact, an increasing interest of research groups towards the development of toxicity models has added significant tools to the field of computational modeling. In a recent study, Kleandrova et al. [19] has reviewed significant advancements in the QSAR modeling for the prediction of acute toxicity, and has also introduced a multitasking toxicity model. Besides this, Furuhama et al. [20] has proposed quantitative structure–activity–activity relationships (QSAAR) for modeling the species-specific acute aquatic toxicity of aromatic amine and phenols, whereas Speck-Planche et al. [21] had predicted multiple ecotoxicological effects of agrochemical fungicides through multi-species QSAR models. Moreover, QSAR approaches based on molecular docking and simulation techniques have also been quite promising [22], [23].

In the present study, QSAR models are proposed and analyzed for predicting the acute toxicity of diverse hazardous chemicals towards the D. magna. The models are mainly developed using the physico-chemical, electrostatic and topological descriptors along with the quantum-chemical molecular and thermodynamic descriptors computed using advanced semi-empirical methods like PM7 [24], and ab-initio methods such as the Hartree–Fock (HF) [25], [26] and the density functional theory (DFT) [26], [27]. Notably, the quantum-chemical descriptors formulated through the electron-correlation contribution (CORR) [28], [29], [30], [31], [32], [33] are also employed to see the role of instantaneous electron–electron interactions in the modeling of the acute toxicity at the level of electron-dynamics. In our recent studies, the electron-correlation contribution of a quantum-chemical descriptor was observed to be highly significant while modeling the externally predictive QSAR models for the biological activities [28], [29], [32] and physico-chemical properties [30], [31], [33] of different chemicals. For example, the contribution of electron-correlation to the total energy of a molecule, energy of the HOMO, and to the electrophilicity are found to be highly significant while modeling the mutagenic activity of nitrated-PAHs [28], [29], [32] Besides this, the correlation energy is also observed as a robust descriptor while developing single-parameter based externally predictive quantitative structure–property relationship (QSPR) models for the super cooled vapor pressure of polychlorinated-naphthalenes [30], and also for the aqueous solubility, subcooled liquid vapor pressure, n-octanol/water and n-octanol/air partition coefficients of the polychlorinated-dibenzo-p-dioxins (PCDDs) and -dibenzo-furans (PCDFs) [31] and polychlorinated naphthalenes [33].

In our recent work [32], we had also analyzed the performance of various exchange-correlation functionals of the DFT while developing externally predictive QSARs for the mutagenicity of nitrated-PAHs. In this study, it was observed that the quantum-mechanical exchange interactions can be quite critical along with the electron-correlation in modeling the mutagenicity, however, the incorporation of electron-correlation is found to be highly significant in the QSAR models in order to have low errors in the external prediction. However, modeling toxicity of a data set constituting diverse chemicals is difficult since it involves multiple mechanisms [34]. Therefore, modeling of the toxicity with quantum-chemical descriptors and their electron-correlation contribution may pave new insights into the mechanisms of toxicity besides the existing knowledge from the models based on the topological descriptors and physico-chemical properties which are though known to be quite useful while developing QSAR models since partition coefficients can be quite crucial factor in the determination of a chemical’s absorption through the cellular membranes.

Fig. 1 depicts the chemicals under investigation in the present study, which are also listed in Supporting information Table S1. These chemicals are comprised of a wide range of organic functionalities such as linear hydrocarbons, benzene, substituted benzenes, chemicals with single bondOH, single bondNO2, single bondNH2, single bondCdouble bondO, single bondCdouble bondS, Rsingle bondOsingle bondR’ functional groups, ring system of C, H atoms, with and without hetero atoms such as N, O, S etc., chemicals containing Cl atom(s) and chemicals having P atom etc. Most of these chemicals are present as major pollutants in the environment, and are frequently used as pesticides. Many of these chemicals, particularly those having functionalities like single bondOH, single bondNO2, single bondNH2 etc., are capable of extensive biotransformation. Moreover, their metabolite or degradation product(s) are also hazardous in nature. Further, some of the chemicals included in the present study namely, the halogen substituted hydrocarbons, amines, phenols are associated with skin, liver and kidney diseases [35], [36], [37], chemicals like alcohols are central nervous system depressants and are responsible for neurological disorders [38]. Carcinogenity is well known hazard of most of these chemicals, mainly due to the benzene derivatives such as nitrobenzenes, poly-aromatic- and heteroaromatic hydrocarbons [35], [39]. The dangerous effects of these chemicals raise the need for modeling the toxicity associated with them, to regulate the use of such chemicals by industries and organizations for the purpose of risk assessment.

This paper is organized as follows: Section 2 provides the detailed theoretical strategy employed for the computation of various quantum-chemical descriptors utilized in this work and for the estimation of their electron-correlation contribution. This is followed by Section 3 on materials and methods, which provide information on the chemical data set and toxicity under investigation, and on the development and statistical validation of the QSAR models. The developed models are further analyzed in Section 4 on results and discussion where the internal and external predictivity of the present models is explored and compared with the robust models available in the literature [9]. Finally, the last section makes few concluding remarks.

Section snippets

Theoretical and computational details

In the present study, a number of quantum-chemical molecular descriptors are employed such as the total electronic energy of molecule (E), energies of the highest occupied and lowest unoccupied molecular orbital (EHOMO and ELUMO), absolute electronegativity (χ), chemical hardness (η), electrophilicity index (ω), and dipole moment (d) [28], [29], [30]. Besides these, the widely accepted thermodynamic descriptors, namely, standard Gibbs free energy (G) and enthalpy (H) computed using the

Test chemicals and biological activity

The data set for 252 organic chemicals, listed in Supporting information Table S1, having aliphatic, aromatic and cyclic structures covering diverse organic functionalities, is taken from the existing literature [9] on the acute toxicity towards D. magna. The toxicity is reported as logarithmic scale of the lethal concentration (LC50 value in mol/L), that is, the dose responsible for the mortality of 50% of a group of test animals in a specified period, 48-h in this case. The data-set used in

Results and discussion

The key internal and external validation parameters of relevant models are listed in Table 1, Table 2, Table 3, while all the parameters described in the previous section are further provided in Supporting information Tables S3–S19. The reported models in these tables are developed employing various quantum-chemical molecular and thermodynamic descriptors along with their electron-correlation contribution (CORR) computed in the present study, in addition to the physico-chemical (log Pmix),

Conclusions

In conclusion, the models proposed, in this work, for the acute toxicity of diverse chemicals towards D. magna, are more reliable for the external prediction than the similar model available in the literature as evident from the satisfactory external validation parameters of the models II(a–d) and III(a) in Table 2 compared to the literature model (entry 1 in Table 1). It is, however, surprising that the more complex and advanced quantum-chemical descriptors, as employed in the present study

Acknowledgments

The authors are thankful to University Grant Commission (UGC), India for financial support under UGC-Major Research Project no 42-313/2013(SR) and UGC-BSR fellowship. The authors are grateful to Prof. Paola Gramatica for providing QSARINS software, and also to the Department of Chemistry, Panjab University, Chandigarh, India for providing other computational software and resources.

References (69)

  • Vikas et al.

    Does electron-correlation has any role in the quantitative structure–activity relationships?

    J. Mol. Graph. Model.

    (2013)
  • Reenu et al.

    Electron-correlation based externally predictive QSARs for mutagenicity of nitrated-PAHs in Salmonella typhimurium TA100

    Ecotoxicol. Environ. Saf.

    (2014)
  • Vikas et al.

    Externally predictive quantitative modelling of super cooled liquids vapour pressure of polychlorinated-naphthalene through electron-correlation based quantum-mechanical descriptors

    Chemosphere

    (2014)
  • Vikas et al.

    Single-descriptor based quantum-chemical QSPRs for physico-chemical properties of polychlorinated-dibenzo-p-dioxins and -dibenzo-furans (PCDD/Fs): exploring the role of electron-correlation

    Chemosphere

    (2015)
  • Chayawan et al.

    Externally predictive single-descriptor based QSPRs for physico-chemical properties of polychlorinated-naphthalenes: exploring relationships of log SW, log KOA, and log KOW with electron-correlation

    J. Hazadr. Mat.

    (2015)
  • D. Butina et al.

    Predicting ADME properties in silico: methods and models

    Inf. Biotechol.

    (2002)
  • A. Golbraikh et al.

    Beware of q2

    J. Mol. Graph. Model.

    (2002)
  • R. Todeschini et al.

    The K correlation index: theory development and its application in chemometrics

    Chemom. Intell. Lab. Syst.

    (1999)
  • P.K. Ojha et al.

    Further exploring r2m metrics for validation of QSPR models dataset

    Chemometr. Intell. Lab. Syst.

    (2011)
  • Y.G. Lee et al.

    Predicting the toxicity of substituted phenols to aquatic species and its changes in the stream and effluent waters

    Arch. Environ. Contam. Toxicol.

    (2006)
  • P.C. Von der Ohe et al.

    Structural alerts – a new classification model to discriminate excess toxicity from narcotic effect levels of organic compounds in the acute daphnid assay

    Chem. Res. Toxicol.

    (2005)
  • M.T.D. Cronin et al.

    pH-dependence and QSAR analysis of the toxicity of phenols and anilines to Daphnia magna

    Environ. Toxicol.

    (2000)
  • S. Marchini et al.

    Toxicity of aryl- and benzylhalides to Daphnia magna and classification of their mode of action based on quantitative structure–activity relationship

    Environ. Toxicol. Chem.

    (1999)
  • M.A. Lampi et al.

    A predictive quantitative structure–activity relationship model for the photoinduced toxicity of polycyclic aromatic hydrocarbons to Daphnia magna with the use of factors for photosensitization and photomodification

    Environ. Toxicol. Chem.

    (2007)
  • M. Moosus et al.

    Quantitative strucutre–activity relationship analysis of acute toxicity of diverse chemicals to Daphniam Magna with whole molecular descriptor

    SAR QSAR Environ. Res.

    (2011)
  • R. Guha et al.

    Determining the validity of a QSAR model – a classification approach

    J. Chem. Inf. Model.

    (2005)
  • E. Tenorio-Borroto et al.

    QSPR and flow cytometry analysis (QSPR-FCA): review and new findings on parallel study of multiple interactions of chemical compounds with immune cellular and molecular targets

    Curr. Drug Metab.

    (2014)
  • P.G. Maykel et al.

    TOPS-MODE based QSARs derived from heterogeneous series of compounds. Application to the design of new herbicides

    J. Chem. Inf. Comput. Sci.

    (2003)
  • M. Giovanni et al.

    Using the TOPS-MODE approach to fit multi-target QSAR models for tyrosine kinase inhibitors

    Eur. J. Med. Chem.

    (2011)
  • V.V. Kleandrova et al.

    In silico assessment of the acute toxicity of chemicals: recent advances and new model for multitasking prediction of toxic effect

    Mini Rev. Med. Chem.

    (2015)
  • A. Furuhama et al.

    Interspecies quantitative structure–activity–activity relationships (QSAARs) for prediction of acute aquatic toxicity of aromatic amines and phenols

    SAR QSAR Environ. Res.

    (2015)
  • A. Balupuri et al.

    Docking based 3D-QSAR studies of pyridyl aminothiazole derivatives as checkpoint kinase 1 inhibitors

    SAR QSAR Environ. Res.

    (2014)
  • J.J.P. Stewart

    Optimization of parameters for semiempirical methods VI: more modifications to NDDO approximations and re-optimization of parameters

    J. Mol. Model.

    (2013)
  • A. Szabo et al.

    Modern Quantum Chemistry: Introduction to Advanced Electronic Strucutre Theory

    (1982)
  • Cited by (31)

    • Generating accurate in silico predictions of acute aquatic toxicity for a range of organic chemicals: Towards similarity-based machine learning methods

      2021, Chemosphere
      Citation Excerpt :

      Nevertheless, a comparison was performed at the mechanistic level, with particular attention paid to the relevance of the explanatory variables to the modeled endpoint (acute toxicity to freshwater organisms). The descriptors that have been successfully used to derive KwLPR models can be classified into three categories: (1) chemical lipophilicity/bioavailability: ALOGP, MLOGP, and ALOGpS; (2) electrophilic reactivity: LUMO and E; and (3) molecular polarizability and size: Ad, SAacc, and MW (Schwöbel et al., 2011; Voutchkova et al., 2011; LoPachin et al., 2012, 2019, 2019; Cassotti et al., 2014; Reenu and Vikas, 2015; Levet et al., 2016; LoPachin and Gavin, 2016). Studies have indicated that lipophilicity, expressed through LogP, is the driving force behind the bioavailability that enhances acute toxicity to aquatic organisms (Levet et al., 2016).

    View all citing articles on Scopus
    View full text