Chemosphere 128 (2015) 111–117

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

Investigation on modes of toxic action to rats based on aliphatic and aromatic compounds and comparison with fish toxicity based on exposure routes Jia He a,b, Jin J. Li a, Yang Wen a, Hong W. Tai a, Yang Yu a, Wei C. Qin a, Li M. Su a, Yuan H. Zhao a,⇑ a State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration, School of Environment, Northeast Normal University, Changchun, Jilin 130117, PR China b Department of Chemistry, University College London, London WC1H 0AJ, UK

h i g h l i g h t s  The toxic contributions of functional groups have been calculated.  Reference threshold of excess toxicity has been developed in rat toxicity.  Different MOAs were observed in rat and fish toxicity.  Some compounds are classified as less inert compounds for rats, but not for fish.  Different toxic effects on rat and fish are due to the difference in exposure routes.

a r t i c l e

i n f o

Article history: Received 23 September 2014 Received in revised form 22 December 2014 Accepted 14 January 2015

Handling Editor: A. Gies Keywords: Mode of action Baseline Threshold Exposure route Lethal critical concentration Intestinal absorption

a b s t r a c t The modes of toxic action (MOAs) play an important role in the assessment of the ecotoxicity of organic pollutants. However, few studies have been reported on the MOAs in rat toxicity. In this paper, the toxic contributions of functional groups in 1255 aromatic compounds were calculated from regression and were then compared with the toxic contributions in aliphatic compounds. The results show that some functional groups have same toxic contributions both in aromatic and aliphatic compounds, but some have not. To investigate the MOAs in rat toxicity, the distribution of toxic ratio (TR) was examined for well-known baseline and less inert compounds and thresholds of log TR = 0.3 and 0.5 were used to classify baseline, less inert and reactive compounds. The results showed that some compounds identified as baseline compounds in fish toxicity were also classified as baseline compounds in rat toxicity. Except for phenols and anilines which were identified as less inert compounds in fish toxicity, aromatic compounds with functional groups such as ether, nitrile, nitrophenol, isocyanatoe and chloro were identified as less inert chemicals in rat toxicity. Reactive compounds identified in fish toxicity exhibit greater toxicity to rats. These compounds can undergo nucleophilic substitution, acylation and Schiff base formation with biological macromolecules. The critical body residues (CBRs) calculated from absorption and bioconcentration show that log 1/CBRs in rat toxicity are not equal to that in fish for some compounds. It suggests that the exposure route can affect the identification of MOAs between these two species for these compounds. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Mammalian acute toxicity is an important biological endpoint for drug design and toxicological risk assessment of chemicals. They can be obtained by a single administration of a chemical within 24 h period and expressed as median lethal dose (LD50) of the ⇑ Corresponding author. Tel.: +86 431 89165610; fax: +86 431 89165606. E-mail address: [email protected] (Y.H. Zhao). http://dx.doi.org/10.1016/j.chemosphere.2015.01.028 0045-6535/Ó 2015 Elsevier Ltd. All rights reserved.

chemical. The preferred animal for experimental testing is the rat although other rodent species may be used (Moore et al., 2013; Lu et al., 2014). However, experimental testing of compounds on rodent acute toxicity is costly and criticized for ethical reasons. The alternative approach using quantitative structure–activity relationships (QSAR) has been suggested as a means of identifying the presence or absence of hazardous properties of the substances (Tsakovska et al., 2008; Lagunin et al., 2011). QSAR methods on mammalian toxicology have been proposed for predicting LD50,

112

J. He et al. / Chemosphere 128 (2015) 111–117

but most of these were derived from limited data sets of structurally similar chemicals such as alcohols or anilines (Devillers and Devillers, 2009; Sazonovas et al., 2010). For example, the data on the rat oral LD50 values for saturated monohydric alcohols were well-fitted by a bilinear model. Polyamines require a fragmentdescriptor reflecting the poly functionality and anilines were best predicted by a combination of electronic, steric and hydrophobic parameters (Jäckel and Klein, 1991; Koleva et al., 2011). The efforts already made to develop QSAR models for mammalian toxicity demonstrate the usefulness of the approach not only for predictive purposes but also for a better understanding of the multiple mechanisms involved in the toxicity, such as non-polar narcosis, polar narcosis and reactive mechanisms (Tsakovska et al., 2008). Baseline toxicity is associated with chemicals acting by narcosis mechanism which is the reversible suppression of physiological function brought about by hydrophobic binding of chemicals to cell membranes and proteins. Because these weak interactions impact countless membranes and proteins non-specifically, normal physiological functions decline and lethality is approached for a broad array of chemical structures (Veith et al., 2009; Aruoja et al., 2014). It has been estimated that about 70% of monomeric industrial organic compounds exert their toxicity to aquatic organisms via the narcosis mechanism (Bradbury and Lipnick, 1990). The toxic mechanism of polar narcosis is not clear. The interaction of polar compounds with biological macromolecules may be through physical interaction rather than chemical reaction. It is well-known that the reactive mechanism includes the formation of covalent bonds between electron-poor (electrophilic) substrate and a biological electron-rich (nucleophilic) target molecule, especially biological macromolecules such as nucleic acids and proteins (Lipnick, 1999; Schwöbel et al., 2011). Such as Schiff base formation, bi-molecular nucleophilic substitution (SN2), acylation and aromatic nucleophilic substitution (SNAr), these are the most important direct acting covalent binding mechanisms (Aptula and Roberts, 2006; Aptula et al., 2006; Schultz et al., 2006; Roberts et al., 2007). Although QSAR models and MOAs of industrial chemicals for aquatic toxic effects are well developed and investigated, no systematic efforts have been made to develop QSAR models for rat toxicity for industrial chemicals and their MOAs are not clear. In our previous study (He et al., 2014), log 1/LD50 values of 1588 industrial aliphatic compounds were examined to investigate the baseline toxicity to rats. The result showed that rat toxicity varies around a constant for each specific class of compounds, and chemical classes of alkanes, alcohols, ethers, acetones, esters and acids can be classified as baseline compounds. In the present paper, 1255 wellcharacterized industrial aromatic chemicals, such as benzenes with the functional groups of halogen, alkanes, alkenes, alcohols, ethers, aldehydes, ketones, esters, acids, amine, nitro, nitroso, isocyanato, nitrate, nitrile and their derivatives were selected to investigate the baseline (non-polar narcotic), less inert (polar narcotic) and reactive compounds in mammalian toxicology. The aims of this work are: (1) to explore the relationship between log 1/LD50 and substructures for aromatic compounds and compare with log 1/ LD50 of aliphatic compounds; (2) to investigate modes of action (baseline, less inert and reactive compounds) based on the toxic ratios; and (3) to compare the modes of action in rat toxicity with fish toxicity and discuss the effect of exposure routes on toxicity.

2. Materials and methods 2.1. Rat acute toxicity data (LD50) Experimental LD50 values for 7385 compounds were taken from literature with the full format and all the structures (Zhu et al.,

2009). After sulfides, phosphides, and heterocyclic compounds were removed, the remaining data included 1588 aliphatic and 1255 aromatic compounds. These well-characterized aliphatic and aromatic molecular structures were classified into different series based on chemical functional groups. The names of functional groups and the number of compounds in each class are summarized in Table 1. Details of the classification, together with CAS number can be found in Tables S1 and S2 of Supplementary Material. 2.2. Fish 50% lethal concentration (LC50) The concentration required to kill 50% of fish within 96 h, were taken from Raevsky et al. (2008, 2009). They confirmed the wellknown good correlations of toxicity between the three fish species and mentioned that the quality of the experimental data was not perfect for fathead minnow and rainbow trout. This is primarily because data were obtained in different laboratories with different errors of measurements. Therefore, the 96 h-LC50 values in fish for 128 aromatic compounds and 100 aliphatic compounds used in this paper were based on the toxicity data to Guppy. A few data on fathead minnow and rainbow trout were used where data to Guppy were missing. These data can be found in Table S3 of the Supplementary Material. 2.3. Fish bioconcentration factor (BCF) and rat intestinal absorption (%Abs.) The log BCF values were estimated from a log BCF–log KOW relationship (Eq. (1)). This equation is used to estimate the log BCF values for compounds with log KOW in the range from 1 to 7 in the Epi Suite (version 4.0) software (http://www.epa.gov/opptintr/exposure /pubs/episuite.htm).

log BCF ¼ 0:6598 log K OW  0:333

ð1Þ

%Abs: ¼ 100  ½1  Expð100:7470:340A0:155B Þ

ð2Þ

The percentages of rat intestinal absorption (%Abs.) of aromatic compounds were calculated by Eq. (2). Here, A is the overall solute hydrogen bond acidity and B is the overall hydrogen bond basicity. This method was based on the rat intestinal absorption dosed orally by gavage for 105 compounds (Zhao et al., 2003). The predictive ability of the method is good for compounds with high absorption (e.g. %Abs. > 90%). 2.4. Calculation of toxic contributions The toxic contributions of substituted functional groups were calculated from the multiple linear regression analysis with the P Minitab software (version 14). The average error (AE = (Obs – P Pred)/n), the average absolute error (AAE = |Obs – Pred|/n) and P the root-mean squared error (RMSE = ( (Obs – Pred)2/n)1/2) were calculated for all the classified compounds. 3. Results 3.1. Relationship between log 1/LD50 and structures Regression analysis has been carried between log 1/LD50 and molecular descriptors calculated in this paper (e.g. the octanol/ water partition coefficient (KOW), the pKa values for acids and bases, the fractions of unionized (F0), positive (F+), negative (F) and zwitterionic (F±) forms at a given pH = 7.4, and the Abraham solvation descriptors (A, B, E, S, V)) for 1255 aromatic compounds. The result showed that the relationship was very poor. The

113

J. He et al. / Chemosphere 128 (2015) 111–117 Table 1 Toxic contributions of functional groups for aromatic and aliphatic chemicals. Nos.

Aromatic compounds Functional groups

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70

Ar ArC@CR ArRC@CR ArC„CR ArRC„CR ArOH ArROH ArOR ArROR ArC(CBr)OR ArOROH ArROROH ArCHO ArRCHO ArC(@O)R ArRC(@O)R ArC(@O)Cl ArOC(@O)R ArROC(@O)R ArC(@O)OR ArRC(@O)OR ArC(@O)OH ArRC(@O)OH ArOC(@O)OR ArNH2 ArRNH2 ArNHR ArRNHR ArNR2 ArRNR2 ArANA(CCACl)2 ArC„N ArRC„N ArNO2 PhenolNO2 ArC(@O)NHR ArRC(@O)NHR ArNHC(@O)R ArRNHC(@O)R ArNC(@O)R ArNC(@O)CCl ArNHC(@O)CF ArRNHC(@O)CF ArNHNHC(@O)R ArOC(@O)NHR ArROC(@O)NHR ArNHC(@O)OR ArOC(@O)NC(@O)R ArNHC(@O)NH ArRNHC(@O)NHR ArC@NR ArRC@NR ArC@NOH ArN@CNR ArRN@CNR ArNHC(@NH)NHR ArNRN@O ArRNRN@O ArN@O ArN@C@O ArRN@C@O ArN@NNR ArF ArRF ArCF3 ArCl ArRCl ArBr ArRBr ArI

Aliphatic compounds Number of compounds 68 51 1 14 194 133 270 69 21 29 1 23 13 34 27 9 23 44 64 68 40 55 5 114 25 23 36 38 38 10 22 17 124 18 48 32 44 2 7 19 5 1 4 75 16 10 7 25 3 2 3 1 4 7 3 5 9 4 11 5 25 18 15 26 249 61 27 27 19

Ci

log TR

Functional groups

Number of compounds

1.830 0.130 0.050 0.020 0.154 0.258 0.0001 0.145 0.071 0.474 0.173 0.206 0.148 0.046 0.220 0.004 0.362 0.426 0.167 0.200 0.332 0.070 0.215 0.756 0.160 0.244 0.209 0.702 0.080 0.522 1.558 0.276 0.189 0.240 0.572 0.074 0.146 0.089 0.549 0.128 0.291 1.773 0.087 1.239 1.396 0.348 0.130 0.611 0.001 0.095 0.651 0.266 0.895 0.759 1.162 0.369 0.714 1.025 0.087 0.228 0.105 0.968 0.412 0.558 0.194 0.246 0.266 0.303 0.674 0.228

0.160 0.030 0.210 0.140 0.314 0.418 0.160 0.305 0.231 0.634 0.013 0.366 0.012 0.114 0.380 0.164 0.202 0.586 0.007 0.040 0.492 0.090 0.375 0.916 0.320 0.404 0.369 0.862 0.240 0.682 1.718 0.436 0.349 0.400 0.732 0.234 0.306 0.071 0.389 0.032 0.451 1.933 0.247 1.399 1.556 0.508 0.290 0.771 0.161 0.255 0.811 0.426 1.055 0.919 1.322 0.209 0.874 1.185 0.247 0.388 0.055 1.128 0.572 0.718 0.034 0.406 0.426 0.463 0.834 0.388

R RC@CR/RC„CR

507

1.670 0.233

ROH ROR

348 313

0.108 0.080

67

0.007

RC(@O)R

116

0.042

RC(@O)Cl

11

0.213

RC(@O)OR

337

0.118

RC(@O)OH

109

0.015

ROC(@O)OR RNH2/R2NH/R3N

10 281

0.433 0.404

RC„N

115

0.482

RNO2

26

0.527

RC(@O)NHR

74

0.102

ROC(@O)NHR

20

0.357

RNHC(@O)NHR

34

0.206

RC@NOH

6

0.275

RNHC(@NH)NHR

7

0.240

RN@O RN@C@O

63 14

0.464 0.269

N@N/C@N RFn

10 55

0.475 0.900

RCln

252

0.617

RBrn

58

0.901

RCHO

RIn

9

Ci/log TR

1.220 (continued on next page)

114

J. He et al. / Chemosphere 128 (2015) 111–117

Table 1 (continued) Nos.

71 72 73

Aromatic compounds

Aliphatic compounds

Functional groups

Number of compounds

Ci

ArRing ArEpoxyTotal compounds

62 70 1255

0.272 0.180 AE = 0 AAE = 0.36 RMSE = 0.46

log TR 0.432 0.340

Functional groups

Number of compounds

Ci/log TR

Ring EpoxyR Total compounds

216 33 1588

0.101 0.051 AE = 0 AAE = 0.44 RMSE = 0.57

Note: The toxic contribution of benzene is obtained from intercept of the regression. The number of compounds illuminates how many compounds contain the functional group. Because compounds contain more than two functional groups, the sum of the compounds in classes 1–51 is more than the number of total compounds 1255. The values of aliphatic compounds were taken from reference of He et al. (2014).

coefficient of determination (R2) was only 0.1 and the standard error (S) was 0.68. Inclusion of other descriptors (not shown here) could not improve the regression coefficient of equation. In order to investigate modes of action for chemicals to rats, the relationship between log 1/LD50 and substructures was investigated by carefully classifying compounds into different classes, and the log 1/LD50 values were then examined for the well-classified compounds (Table S1 of Supplementary Material). The result showed that log 1/LD50 varied around a constant in each homologous series listed Table S1 and the log 1/LD50 values were not closely related to the number of substituted groups. In other words, there is no significant difference in log 1/LD50 between compounds with mono-, di-, tri- and multi-substituted groups. For example, the mean log 1/LD50 of alkylphenols was 2.21 (ranges from 1.58 to 2.95), which was close to that of alkylbenzenediols with mean of log 1/LD50 = 2.29 (from 1.39 to 2.79). And the mean log 1/LD50 of phenyl mono-ethers was 1.75, which was quite similar to that of phenyl di-ethers (mean of log 1/LD50 = 1.95). The same situation can be found for all other series except the nitro substituted benzenes and phenols. The mean log 1/LD50 of mono-nitrophenols was 2.37 which was significantly different to that of di-nitrophenols (mean of log 1/LD50 = 3.85) (see Table S1 of Supplementary Material). The log 1/LD50 values of nitrobenzenes and nitrophenols were related to the number of nitro substituted groups. To investigate the relationship between log 1/LD50 and substructures, multilinear regression analysis was carried out between log 1/LD50 and indicator variables for all aromatic compounds by using Eq. (3). A substructure indicator variable (Fi), was used to describe the presence and absence of certain functional groups. Values of 1 were given to mono-, di- or multi-substituted groups for all homologous classes except nitrobenzenes and nitrophenols. They are the number of the nitro functional groups. Values of 0 were given to the chemicals which were absence of certain functional groups.

log 1=LD50 ¼ 1:83 þ

X

Ci F i

ð3Þ

Ci is the toxic contributions of functional groups calculated from regression analysis. The value of 1.83 in Eq. (3) is the toxic contribution of skeleton compound (i.e. benzene) obtained from the intercept of the regression. Table 1 lists the toxic contribution values (Ci) for different substituted functional groups. Inspection of the Ci values show that the compounds with functional groups, e.g. alkene, acid and nitroso exhibit nearly the same toxicity as compared with the skeleton compound (benzene). The toxic contributions of some functional groups (e.g. ether, ketone and nitrile) are slightly higher than benzene (see nos. 8, 15 and 32 in Table 1). Six functional groups exhibit toxic contributions greater than 1 (e.g. Ci of phenyl carbamate = 1.40). It is noteworthy that the toxic contributions do not show significant differences for some functional groups (e.g. aldehyde, urea and chloro groups) no matter whether they are connected with benzenes directly. On the other hands, the toxic contributions of some other functional groups exhibit significant differences between aromatic

substituents and non-aromatic substituents. For example, the log 1/LD50 of phenols is higher than the benzyl alcohols. A similar situation can be found between bromo benzenes and bromo alkyl benzenes. It is also noteworthy that the synergy was observed for some functional groups. For example, the toxic contribution of aromatic carbamate with ether group is greater than the sum of toxic contributions of carbamate and ether, with a toxic contribution of 1.40 log units (nos. 45–47 in Table 1). The different toxic contributions have been observed for the aromatic esters with ether groups in different positions. The details of toxic contributions for all the functional groups can be found in Table 1.

3.2. Comparison of toxic contributions of functional groups between aromatic and aliphatic compounds Table 1 lists the toxic contributions of functional groups for aromatic compounds and aliphatic compounds. The similarity and difference in log 1/LD50 can be observed from the comparison of toxic contributions of functional groups between aromatic and aliphatic compounds in Table 1. First, similar toxic contributions could be found in both aromatic and aliphatic chemicals for some functional groups. For examples, both alcohol and aldehyde groups exhibit no significant toxic contributions regardless of whether they are connected to aliphatic carbon chain or benzene ring (see nos. 7 and 13 in Table 1). The functional group of isocyanato exhibits positive toxic contributions in both aromatic and aliphatic compounds. Second, slightly different contribution to toxicity has been observed between aromatic and aliphatic compounds for some functional groups. The toxic contribution of skeleton in aromatic compounds (benzene = 1.830) is slightly higher than that in aliphatic compounds (alkane = 1.670). The functional groups of unsaturated bond, acyl chloride, urea and guanidine have slightly higher toxic contributions in aliphatic compounds, but they have no (or even negative) toxic contributions when connected to the benzene ring. Almost no significant toxicity contributions were observed for ethers, ketones and cyclo substituted groups in aliphatic chemicals, but they show slightly higher toxic contributions in aromatic compounds. Primary, secondary and tertiary amine groups all show similar toxic contribution in aliphatic chemicals but the contribution of secondary amine is higher than that of other amines in aromatic compounds. Third, significantly different contributions to toxicity has been observed between aromatic and aliphatic compounds for some functional groups. Functional groups of carbamate and oxmine show only small toxic contributions (0.357 and 0.275) in aliphatic compounds, but when attached directly to benzene ring (see nos. 45 and 53 in Table 1) they exhibit significant toxic contributions in aromatic compounds (1.396 and 0.895). In addition, no additive effect was found in the multi-substituted functional groups in aliphatic compounds, but was observed in aromatic compounds. For example, an amide substituted on a benzene ring exhibited no significant toxicity but fluoro multi-substituted on the benzene ring showed strong toxicity. The same situation can

J. He et al. / Chemosphere 128 (2015) 111–117

also be found with other multi-substituted functional groups. Different substituted positions with different toxic effects were also observed in aromatic compounds. The functional groups substituted in ortho-, meta- and para-position exhibit different toxicity to rats. Some aromatic compounds, such as phenols and nitrobenzenes with functional groups of alkanes, esters and ethers substituted at ortho-position show greater toxicity than that at the para-position or meta-position. On the other hand, anilines with functional groups substituted at the ortho-position exhibit less toxicity than at the meta-position or para-position. 4. Discussion 4.1. Baseline toxicity to rats One of the methods used in the discrimination of reactive compounds from baseline level is based on the toxic ratio (TR). It is defined as the ratio of the calculated baseline toxicity (Tpred) over the experimentally determined value (Tobs). It can be easily converted into logarithmic form expressed as log TR (see Eqs. (4) and (5)).

TR ¼ T pred ðbaselineÞ=T obs

ð4Þ

log TR ¼ log 1=T obs  log 1=T pred ðbaselineÞ ¼ Residual

ð5Þ

The log TR calculated in our previous paper (He et al., 2014) showed that aliphatic alkenes, alcohols, aldehydes, ketones, acids have very similar toxicity with alkanes (see Ci in Table 1) which were used as baseline compounds. In another words, the predicted baseline toxicity (log 1/LD50 or log 1/Tpred) should be the calculated average toxicity of alkanes (i.e. log 1/Tpred (baseline) = 1.670). It was also suggested that threshold of log TR 6 0.3 should be used as a reference threshold for identifying baseline compounds because rats are a much more sensitive species than fish in toxicity (Koleva et al., 2011). If we use alkanes as baseline compounds, log TR values can be calculated for aromatic, as well as aliphatic compounds (Table 1). The log TR will be equal to the toxic contributions (Ci) for aliphatic compounds, but it will be equal to the contributions added up by 0.160 (1.830 – 1.670 = 0.160) for aromatic compounds because of the toxic difference in skeleton compounds (benzene and alkane). Examination of log TR values listed in Table 1 shows that the functional groups, e.g. alkene, alcohol and urea do not exhibit excess toxicity, with log TR less than 0.30. The toxicity of alkyl benzenes (1.83) is slightly higher than that of alkanes (1.67) but less than 0.30. They can be identified as baseline compounds. These aromatic compounds were also identified as baseline compounds in fish (Verhaar et al., 1992). However, some compounds classified as baseline compounds in fish were not identified as baseline compounds from the threshold in rats (e.g. chloro benzenes). The reason is due to different thresholds used in rats and fish. The threshold of log TR = 1 used in fish is too high to distinguish the difference of toxic contribution in rats for chloro functional group from alkane or benzene (He et al., 2014). It is noteworthy that the reference threshold of excess toxicity cannot completely identify baseline compounds from the toxic ratio of Eq. (4). Analysis on two datasets in rat toxicity compiled from different sources indicates that absolute average residual of log 1/LD50 of two data sources is 0.31, which is nearly the same as the threshold used in rat toxicity (He et al., 2014). Wrong classification is possible in the identification of baseline compounds from the toxic ratio in rat toxicity. 4.2. Less inert compounds to rats Less inert chemicals are chemicals that are not reactive when considering over all acute effects, but are slightly more toxic than

115

baseline toxicity. These chemicals are often characterized as compounds acting by a so-called ‘‘polar narcosis’’ mechanism (Verhaar et al., 1992). The polar narcosis mode of action in aquatic toxicology only applies to compounds with log KOW < 3. The contributions of polar interactions decrease with higher log KOW and intersect with the non-polar narcosis models beyond log KOW of 4–5. Although it is well known that chemicals, such as phenols, anilines and primary alkyl amines, are characteristic of this class, no thresholds have been developed for these less inert compounds. Inspection of rat toxicity of these less inert compounds phenols and anilines and primary alkyl amines showed that their log TR values are in the range of 0.3–0.5. Examination of the toxic contributions of functional groups in this range in Table 1 show that many polar groups, such as tertiary amine, nitrile and nitro functional groups are in this range. It is apparent that polar chemicals are more toxic than non-polar chemicals (baseline). Polar narcotics were not considered to be reactive compounds, but they may have stronger interactions with macromolecules at the target sites than the non-polar narcotics. This interaction may be through physical interaction, rather than chemical reaction. Although many polar chemicals exhibit greater toxicity than non-polar chemicals, some non-polar compounds, such as, chloro, bromo, iodo, cyclic and epoxy compounds are also in this range in rat toxicity. These compounds show greater toxicity than baseline compounds, but with less polarity than polar chemicals. There are strong arguments in favor and against the separation of the two mechanisms of action because a cut-off to classify the polar narcotics from non-polar is difficult (Dearden et al., 2000). Some polar compounds (e.g. alcohols) do not show a similar toxicity to these so-called polar narcotics but, on the other hand, some non-polar compounds do exhibit a similar toxicity (e.g. bromo compounds). Therefore, we prefer to use ‘‘less inert compounds’’ rather than ‘‘polar narcotic compounds’’ because ‘‘polar narcotics’’ cannot well express the mode of toxic action for these compounds in rat toxicity. 4.3. Reactive compounds to rats The identification of reactive compounds in fish toxicity was based on the reference threshold of log TR > 1. If we used log TR = 1 as a threshold value in rat toxicity, a few compounds (e.g. di-and tri-nitrophenols) exhibit greater than baseline and less inert compounds with log TR > 1. Because this threshold was based on the TR distribution of baseline, less inert and reactive compounds to fish toxicity, not to rat toxicity (Verhaar et al., 1992). Systematic analysis on the distribution of log TR in rat toxicity has not been reported in the literature. Inspection of the range of log TR values for baseline and less inert compounds in rat toxicity (Table 1) show that all the baseline and less inert compounds identified in this paper have log TR less than 0.5. If we use log TR = 0.5 as a threshold of reactive compounds for rat toxicity, chemicals, such as nitrophenols, phenyl hydrazones and phenyl hydrazides, exhibit excess toxicity and can be predicted as reactive compounds. Most of these compounds have been identified as reactive compounds in fish toxicity (Russom et al., 1997; Enoch et al., 2011; Schwöbel et al., 2011). They can undergo covalent binding to nucleophilic sites on one or more biological macromolecules, thus acting as electrophilic toxicants (Koleva et al., 2011; Schwöbel et al., 2011). The reactive mechanisms for some reactive compounds were identified in this paper listed in Fig. S1 of Supplementary Material. The general mechanistic principles for bi-molecular nucleophilic substitution (SN2) involve a biological nucleophile attacking an aliphatic carbon, nitrogen, sulfur or halogen atoms to which an electronegative leaving group is attached. The aromatic chemicals containing the functional groups which halogen atoms substituted on aliphatic chain react by this mechanism. Aromatic nucleophilic

116

J. He et al. / Chemosphere 128 (2015) 111–117

substitution (SNAr) involves nucleophilic attack by either a cysteine or lysine unit on an activated aromatic system. The chemicals within this mechanistic domain are nitrophenols, fluorobenzenes and tertiary anilines containing the N(CH2CH2Cl)2 groups (no. 31 in Table 1). The acylation mechanistic domain involves the attack of a carbonyl (or carbonyl-type) species by a biological nucleophile such as cysteine or lysine. The chemicals within this mechanistic domain are phenyl esters, phenyl carboxylic anhydrides, phenyl carbamates and phenyl nitrosoes shown in Table 1. The general reaction scheme for the Schiff base mechanistic domain involves nucleophilic nitrogen attacking a reactive carbonyl center. Phenyl amides substituted by fluoro and phenyl hydrazides are belonging to this reactive mechanism. It is noteworthy that the threshold of excess toxicity used above cannot completely identify reactive compounds from baseline and less inert levels (Verhaar et al., 1992). Not all reactive compounds can exhibit excess toxicity than baseline compounds with log TR higher than the threshold (e.g. 0.5 or 1). For example, epoxides were identified as reactive compounds (Schramm et al., 2011), but have log TR less than 0.4. The toxic ratio is a very valuable method in the discrimination of modes of action, but wrong classification is possible for some compounds. 4.4. Effect of exposure routes on rat and fish toxicities The above classification revealed that some compounds may share the same MOAs, but some may not between rat and fish toxicity. However, the identification of baseline, less inert and reactive compounds in rat and fish toxicity was based on the TR which is derived from the distribution of toxicity among baseline, less inert and reactive compounds. The TR is the toxic ratio based on external effect concentrations, rather than internal effect concentration. It is closely related to the exposure route. To further investigate toxic mechanism between rat and fish and effect of exposure routes, internal effect concentrations, termed critical body residues (CBR), have been calculated for some compounds both in rat and fish toxicity (McCarty and Mackay, 1993). If no consideration is given to the metabolism and elimination of a chemical, CBRs can be estimated from the percentage of absorption (%Abs.) and bioconcentration factor (BCF) for rats and fish by using Eqs. (6) and (7), respectively.

CBR ðratÞ ¼ LD50  %Abs: log 1=CBR ðratÞ ¼ log 1=LD50  log %Abs:

ð6Þ

CBR ðfishÞ ¼ LC50  BCF log 1=CBR ðfishÞ ¼ log 1=LC50  log BCF

ð7Þ

Table 2 lists the means of CBRs in fish and rats for some homologous series identified as baseline and less inert compounds. Close means of log CBRs have been observed between fish and rat toxicity for some homologues (e.g. chloro phenols). It indicates that they may share the same mechanisms of toxic action between these two species. However, some differences can still be observed between CBRs in these two species (e.g. chloro anilines and ethers). The overall log 1/CBR values in fish are greater than that in rats, and only one homologue (bromo alkanes) has log 1/CBR less than that in rats. There are many reasons that can cause the difference in CBRs between rat and fish toxicity. The difference in exposure routes in fish and rat toxicity is apparently an important reason. In fish dosing is via the water phase through breathing, while in acute oral rat dosing is via gavage. Compounds may have different metabolic processes with different metabolism rates in rats and fish during different periods of toxicity test (24 and 96 h to rat and fish, respectively). The CBR calculated from BCF is the total

Table 2 Critical body residues (CBRs) and number of compounds (N) in fish and rats. Rat

Alcohols Ethers Alcohol ethers Ketones Amines Chloro alkanes Bromo alkanes benzenes Phenyl ketones Phenyl ethers Phenols Chloro phenols Anilines Chloro anilines Nitrobenzenes Chlorobenzenes Chloronitrobenzens

Fish

N

log 1/CBR

N

log 1/CBR

48 8 62 20 21 30 9 20 3 5 42 13 20 11 5 14 7

1.56 1.33 1.58 1.64 2.45 1.82 2.38 1.49 1.88 1.72 2.23 2.61 2.28 2.25 2.10 2.04 2.36

13 5 7 14 18 14 5 6 2 2 16 17 11 13 6 16 9

1.68 1.87 1.68 1.90 2.46 2.06 2.25 1.97 2.20 2.11 2.56 2.71 2.62 2.99 2.39 2.37 2.75

concentration in fish. It is well estimated for most of compounds but over-estimated from the BCF for highly hydrophobic chemicals because a steady-state may not be achieved within 96 h lethal toxicity test. At the same time, the CBR calculated from %Abs. is the total concentration in blood and it is significantly over-estimated from %Abs. because of the elimination of a chemical within/after 24 h, resulting in overall calculated log 1/CBR in rat less than that in fish. Furthermore, the distribution of a chemical after absorption in mammals is apparently different from that after bioconcentration in aquatic organisms, resulting in difficulty in the comparison of internal concentrations. The CBRs calculated from BCF and %Abs. by Eqs. (6) and (7) are not the real critical concentrations in fish and rats, although they are close to internal concentrations at the target sites. 5. Conclusions The toxic contributions calculated from regression analysis show that the aromatic compounds with functional groups, e.g. alkene, acid and nitroso exhibit nearly the same toxicity as compared with the skeleton compound (benzene). The toxic contributions for some functional groups (such as aldehyde, urea and chloro groups) do not exhibit significant differences no matter whether or not they are connected to a benzene ring. On the other hand, some functional groups exhibit significantly different toxic contributions between aromatic substituents and non-aromatic substituents (e.g. hydroxy). Comparison of toxic contributions of functional groups between aliphatic and aromatic compounds show that toxic contribution is very similar in both aromatic and aliphatic chemicals for some functional groups (e.g. alcohol and aldehyde groups). On the other hand, different contributions to toxicity have been observed between aromatic and aliphatic compounds for some functional groups. The substituent positions (e.g. ortho-, meta- and para-) influence the toxicity to rats in aromatic compounds. Baseline toxicity, less inert toxicity and excess toxicity were investigated in rat toxicity from the toxic ratio (TR) calculated from the toxic contributions. Threshold of log TR 6 0.3 has been developed as a reference threshold for identifying baseline compounds from less inert and reactive compounds. The functional group of alkene, alcohol, aldehyde, acyl chloride, acid, amide and urea were predicted as baseline compounds in rat toxicity. The chemicals with log TR values in the range of 0.3–0.5 developed from phenol and aniline toxicity were used to identify less inert compounds. The identification of less inert compounds from the thresholds suggests that not only phenols and anilines can be identified as less inert compounds in rat toxicity, but that aromatic compounds

J. He et al. / Chemosphere 128 (2015) 111–117

with functional groups of ether, nitrile, nitrophenol, isocyanato and chloro groups were also identified as less inert chemicals. These compounds may share different MOAs between rat and fish toxicity. The distribution of log TR for baseline and less inert compounds suggest that reference threshold of excess toxicity log TR = 0.5 should be used to identify the reactive compounds from baseline or less inert compounds in rat toxicity, rather than the threshold of log TR = 1 used in fish because rats are much more sensitive species than fish in toxicity. Comparison of rat toxicity with fish toxicity shows that some classified compounds have different CBRs because of the difference in exposure routes in fish and rat toxicity tests. Acknowledgements This work is supported by the National Natural Science Foundation of China (21377022) and the Fundamental Research Funds for the Central Universities (12SSXT138). Jia He thanks the supporting by China Scholarship Council. During her stay at University College London, Professor Michael Abraham gave her great help, technical review and critical reading of the manuscript. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2015.01.028. References Aptula, A.O., Patlewicz, G., Roberts, D.W., Schultz, T.W., 2006. Non-enzymatic glutathione reactivity and in vitro toxicity: a non-animal approach to skin sensitization. Toxicol. In Vitro 20, 239–247. Aptula, A.O., Roberts, D.W., 2006. Mechanistic applicability domains for nonanimalbased prediction of toxicological end points: general principles and application to reactive toxicity. Chem. Res. Toxicol. 19, 1097–1105. Aruoja, V., Moosus, M., Kahru, A., Sihtmae, M., Maran, U., 2014. Measurement of baseline toxicity and QSAR analysis of 50 non-polar and 58 polar narcotic chemicals for the alga Pseudokirchneriella subcapitata. Chemosphere 96, 23–32. Bradbury, S.P., Lipnick, R.L., 1990. Introduction: structural properties for determining mechanisms of toxic action. Environ. Health Persp. 87, 181–182. Dearden, J., Cronin, M.T.D., Zhao, Y.H., Raevsky, O., 2000. QSAR studied of compounds acting by polar and non-polar narcosis: an examination of the role of polarisability and hydrogen bonding. Quant. Struct.-Act. Relationship 19, 3–9. Devillers, J., Devillers, H., 2009. Prediction of acute mammalian toxicity from QSARs and interspecies correlations. SAR QSAR Environ. Res. 20, 467–500. Enoch, S.J., Ellison, C.M., Schultz, T.W., Cronin, M.T.D., 2011. A review of the electrophilic reaction chemistry involved in covalent protein binding relevant to toxicity. Crit. Rev. Toxicol. 41, 783–802. He, J., Fu, L., Wang, Y., Li, J.J., Wang, X.H., Su, L.M., Sheng, L.X., Zhao, Y.H., 2014. Investigation on baseline toxicity to rats based on aliphatic compounds and comparison with toxicity to fish: effect of exposure routes on toxicity. Regul. Toxicol. Pharm. 70, 98–106.

117

Jäckel, H., Klein, W., 1991. Prediction of mammalian toxicity by quantitative structure–activity relationships: aliphatic amines and anilines. Quant. Struct.Act. Relationship 10, 198–204. Koleva, Y.K., Cronin, M.T.D., Madden, J.C., Schwöbel, J.A., 2011. Modelling acute oral mammalian toxicity. 1. Definition of a quantifiable baseline effect. Toxicol. In Vitro 25, 1281–1293. Lagunin, A., Zakharov, A., Filimonov, D., Poroikov, V., 2011. QSAR modelling of rat acute toxicity on the basis of PASS prediction. Mol. Informatics 30, 241–250. Lipnick, R.L., 1999. Correlative and mechanistic QSAR models in toxicology. SAR QSAR Environ. Res. 10, 239–248. Lu, J., Peng, J.L., Wang, J., Shen, Q.C., Bi, Y., Gong, L.K., Zheng, M.Y., Luo, X.M., Zhu, W.L., Jiang, H.L., Chen, K.X., 2014. Estimation of acute oral toxicity in rat using local lazy learning. J. Cheminformatics 6, 1–11. McCarty, L., Mackay, D., 1993. Enhancing ecotoxicological modeling and assessment body residues and modes of toxic action. Sci. Total. Environ. 27, 1719–1728. Moore, N.P., Andrew, D.J., Bjerke, D.L., Creton, S., Dreher, D., Holmes, T., Prieto, P., Seidle, T., Rowan, T.G., 2013. Can acute dermal systemic toxicity tests be replaced with oral tests? a comparison of route-specific systemic toxicity and hazard classifications under the globally harmonized system of classification and labelling of chemicals (GHS). Regul. Toxicol. Pharm. 66, 30–37. Raevsky, O.A., Grigor’ev, V.Y., Weber, E.E., Dearden, J.C., 2008. Classification and quantification of the toxicity of chemicals to guppy, fathead minnow and rainbow trout: Part 1. Non-polar narcosis mode of action. QSAR Comb. Sci. 27, 1274–1281. Raevsky, O.A., Grigor’ev, V.Y., Dearden, J.C., Weber, E.E., 2009. Classification and quantification of the toxicity of chemicals to guppy, fathead minnow, and rainbow trout. Part 2. polar narcosis mode of action. QSAR Comb. Sci. 28, 163– 174. Roberts, D.W., Aptula, A.O., Patlewicz, G., 2007. Electrophilic chemistry related to skin sensitization. Reaction mechanistic applicability domain classification for a published data set of 106 chemicals tested in the mouse local lymph node assay. Chem. Res. Toxicol. 20, 44–60. Russom, C.L., Bradbury, S.P., Broderius, S.J., Hammermeister, D.E., Drummond, R.A., 1997. Predicting modes of toxic action from chemical structure: acute toxicity in the fathead minnow (pimephalespromelas). Environ. Toxicol. Chem. 16, 948– 967. Sazonovas, A., Japertas, P., Didziapetris, R., 2010. Estimation of reliability of predictions and model applicability domain evaluation in the analysis of acute toxicity (LD50). SAR QSAR Environ. Res. 21, 127–148. Schramm, F., Müller, A., Hammer, H., Paschke, A., Schüürmann, G., 2011. Epoxide and thiirane toxicity in vitro with the ciliates Tetrahymena pyriformis: structural alerts indicating excess toxicity. Environ. Sci. Technol. 45, 5812– 5819. Schultz, T.W., Carlson, R.E., Cronin, M.T.D., Hermens, J.L., Johnson, R., O’Brien, P.J., Roberts, D.W., Siraki, A., Wallace, K.B., Veith, G.D., 2006. A conceptual framework for predicting the toxicity of reactive chemicals: modeling soft electrophilicity. SAR QSAR Environ. Res. 17, 413–428. Schwöbel, J.A., Koleva, Y.K., Enoch, S.J., Bajot, F., Hewitt, M., Madden, J.C., Roberts, D.W., Schultz, T.W., Cronin, M.T.D., 2011. Measurement and estimation of electrophilic reactivity for predictive toxicology. Chem. Rev. 111, 2562–2596. Tsakovska, I., Lessigiarska, I., Netzeva, T., Worth, A.P., 2008. A mini review of mammalian toxicity (Q)SAR models. QSAR Comb. Sci. 27, 41–48. Veith, G.D., Petkova, E.P., Wallace, K.B., 2009. A baseline inhalation toxicity model for narcosis in mammals. SAR QSAR Environ. Res. 20, 567–578. Verhaar, H.M., Vanleeuwen, C.J., Hermens, J.M., 1992. Classifying environmental pollutants. 1: Structure-activity relationships for prediction of aquatic toxicity. Chemosphere 25, 471–491. Zhao, Y.H., Abraham, M.H., Hersey, A., Luscombe, C.N., 2003. Quantitative relationship between rat intestinal absorption and Abraham descriptors. Eur. J. Med. Chem. 38, 939–947. Zhu, H., Martin, T.M., Ye, L., Sedykh, A., Young, D.M., Tropsha, A., 2009. Quantitative structure–activity relationship modeling of rat acute toxicity by oral exposure. Chem. Res. Toxicol. 22, 1913–1921.