Chemosphere 64 (2006) 1619–1626
www.elsevier.com/locate/chemosphere
Quantitative structure activity relationship (QSAR) for toxicity
of chlorophenols on L929 cells in vitro
Xiaohua Liu a, Jiangning Chen b, Hongxia Yu a,*, Jinsong Zhao a,
John P. Giesy a,c,d, Xiaorong Wang a
a
The State Key Laboratory of Pollution Control and Resources Reuse, School of the Environment, Nanjing University, Nanjing 210093, China
b
The State Key Laboratory of Pharmaceutical Biotechnology, School of Life Science, Nanjing University, Nanjing 210093, China
c
Department of Zoology, National Food Safety and Toxicology Center, Center for Integrative Toxicology,
Michigan State University, East Lansing, MI 48824, USA
d
Department of Biology and Chemistry, City University of Hong Kong, Kowloon, Hong Kong, SAR, China
Received 10 January 2006; accepted 21 April 2006
Available online 21 June 2006
Abstract
Quantitative structure activity relationship (QSAR) were developed to predict toxicity of chlorophenols by correlating LC50 values
with five molecular descriptors, chosen
P to represent lipophilic, electronic and steric effects: the n-octanol/water partition coefficient
(log Kow), the constant of Hammett ( r), P
the acid dissociation constant (pKa), the order valence molecular connectivity
index (1vv)
P
and the perimeter of the efficacious section ( Dg). The results of the regression analysis showed that log Kow and Dg are the dominant
(canonical) predictive factors in determining toxicity of chlorophenols to the cells during 24 h exposures, while log Kow was the only dominant predictive factor contributing to toxicity during in 48 h exposures. Comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA) were applied to investigate predictive relationships of the cytotoxicity of
chlorophenols and develop visual 3D-QSAR models. The CoMFA model, in which the contribution of the electrostatic field to the biological activity was greater than that of the steric field, exhibited both high consistency and predictability (r2 = 0.968, Q2 = 0.891 for 24 h
exposure; but the relationship was poorer for the 48 h exposure: r2 = 0.727, Q2 = 0.394). The CoMSIA model used in this study contained three fields: electrostatic, hydrophobic and steric, of which the relative contribution to the biological activity was
0.767:0.225:0.008. In addition, according to the models for 24 h and 48 h. The time-dependent toxicity and potential mechanisms for
inhibition of L929 cells was discussed.
Ó 2006 Elsevier Ltd. All rights reserved.
Keywords: 2D-QSAR; CoMFA; CoMSIA; Time-dependent toxicity; Mechanism
1. Introduction
Due to its toxicity and persistence (Dercova, 2003), pentachlorophenol (PCP) a widespread environmental chemical, has been much studied. However, less information is
available on the other chlorophenols such as mono-chlorophenol (CP), di-chlorophenol (DCP), tri-chlorophenol
*
Corresponding author. Tel.: +86 25 83592912; fax: +86 25 83707304.
E-mail address: yuhx@nju.edu.cn (H. Yu).
0045-6535/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chemosphere.2006.04.091
(TCP) and tetra-chlorophenol (TeCP). Chlorophenol mixtures that are primarily PCP are used as wood preservatives, biosides in adhesives, paints, papers, as well as
insecticides, and can migrate to surface waters. Chlorophenols have been detected in the blood and urine of humans
following both occupational and household exposures. The
International Agency for Research on Cancer (IARC) has
determined that PCP is a possible carcinogen, due to the
generation of tetrachlorohydroquinone (TCHQ), a genotoxic compound that has been identified as a major toxic
metabolite of PCP (Juhl et al., 1985; Witte et al., 1985;
1620
X. Liu et al. / Chemosphere 64 (2006) 1619–1626
Renner and Mucke, 1986; Van Ommen et al., 1986; Carstens et al., 1990; IARC, 1991; Seiler, 1991). It has been
demonstrated that TCHQ, can bind to DNA and cause single strand breaks (SSB) in isolated DNA (Carstens et al.,
1990), in cells (Witte et al., 1985; Carstens et al., 1990; Ehrlich, 1990; Dahlhaus et al., 1996), and in liver tissues of
mice (Dahlhaus et al., 1994; Wang et al., 1997). TCHQ also
induces micronuclei and mutations at the HPRT locus of
V79 cells (Jansson and Jansson, 1991, 1992), and causes
formation of 8-hydroxy-2-deoxyguanosine in V79 cells
(Dahlhaus et al., 1995) and in B6C3F1 mice (Dahlhaus
et al., 1994).
Previous studies have examined the 8 chlorophenol isomers using linear regression methods and identified significant relationships between the cell proliferation inhibition
of HepG2/L929 cells during 24 h exposures and the log Kow
(Jiang et al., 2004). In addition, three dimensional quantitative structure activity relationship (3D-QSAR) have been
developed to predict in vitro inhibition of HepG2 cell
proliferation during 24 h exposures (Liu and Yu, 2005).
However, these studies did not consider other important
descriptors in two dimensional quantitative structure activity relationship (2D-QSAR) studies or time-dependent
toxicity. Here we report the QSAR models that include five
molecular descriptors, chosen to represent lipophilic, electronic and steric effects: the n-octanol/water
P partition coefficient (log Kow), Hammett constant ( r), the acid
dissociation constant (pKa), the order valence molecular
1 v
connectivity
P index ( v ), and the perimeter of the efficacious
section ( D) for 2D-QSAR. However, such 2D-QSAR
models limit the capacity to reveal toxicological mechanisms. In contrast to the 2D-QSAR methods, the well
known 3D-QSAR techniques, CoMFA and CoMSIA
approaches, were used to investigate potential toxicological
mechanisms of various chlorophenols. Since CoMSIA and
CoMFA use different potential functions to describe the
interaction between the ligand and the hypothesis receptor,
the regions in space which are favorable or unfavorable for
the ligand–receptor interaction were identified for the different molecular descriptor fields. The combination of both
techniques gave a comprehensive description of mechanism
of the ligand–receptor interaction. Therefore, in this study,
the standard CoMFA and CoMSIA approaches were combined to determine possible toxicological mechanisms of
chlorophenols on L929 cells.
2. Materials and methods
2.1. Materials
2-Chlorophenol (2-CP), 4-chlorophenol (4-CP), 2,6dichlorophenol (2,6-DCP), 2,4-dichlorophenol (2,4-DCP),
3,4-dichlorophenol (3,4-DCP), 2,4,6-trichlorophenol (2,4,
6-TCP) and 2,3,4-trichlorophenol (2,3,4-TCP), 2,3,5-trichlorophenol (2,3,5-TCP), 2,3,5,6-tetrachlorophenol (2,3,
5,6-TeCP) and pentachlorophenol (PCP) (purity >99%)
were purchased from ACROS (Milwaukee, WI, USA).
RPMI Medium 1640 was purchased from Gibco/BRL
Company (New York, USA); 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) and Dimethyl
sulfoxide (DMSO) were obtained from Sigma Chemical
Company (St. Louis, MO, USA). All other chemicals were
analytical grade.
2.2. Cytotoxicity assay
Detailed descriptions of the cell viability (toxicity) tests
are given elsewhere (Jiang et al., 2004). Briefly, cytotoxicity
of different chlorophenols was measured by using the 3-(4,
5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide
(MTT) reduction assay (Mosmann, 1983; Jiang et al.,
2004) on mouse connective tissue fibroblast L929 cells. Each
test was repeated at least three times and each microplate
had its own control. The mean of six wells was reported
for each chemicals and controls (Jiang et al., 2004). Probability unit analysis was used to determine median lethal
concentration (LC50) values after 24 h or 48 h exposure
(Table 1). The LC50 value was based on dose–response
relationships with at least five doses.
2.3. 2D-QSAR method
of the
P physicochemical features (log Kow, pKa,
P Values
Dg tabulated by Emanuele and Cinzia
re, 1vv and
(1999) are provided for each compound (Table 1). Regression analysis was performed in order to obtain predictive
models, where the response to the toxicity was expressed
as a linear function of molecular descriptors. In this article,
we used SPSS Software (Version 10.0) to process the
regression between five descriptors and the toxicity of
chlorophenols on L929 in vitro for both 24 h and 48 h.
2.4. 3D-QSAR method
All 3D-QSAR studies were performed by use of Sybyl
6.7 molecular modeling software operating on SGI O2
workstations (Tripos Inc., St. Louis, MO, USA). The
geometry of each of the 10 chlorophenols was optimized
using the Tripos’ standard force field, in which the net
atomic charge was calculated using the Gasteiger-Hückel
method, with a cut-off value for energy minimization of
0.42 kJ (mol nm)1. All molecules were aligned based on
the most active compound using the maximum common
structure. The CoMFA steric and electrostatic molecular
fields were calculated within Sybyl, at grid lattice points
with the Lennard-Jones and Coulomb potential functions
of the Tripos Force Field using a common sp3 carbon
probe of charge +1, with 0.2 nm grid spacing and the
energy cut-off set to 126 kJ mol1. Similarly, the three
CoMSIA fields, available within Sybyl, were calculated
at grid lattice points using a common probe atom of
0.1 nm radius, as well as charge, hydrophobicity, and
hydrogen bond properties of +1, with an attenuation
factor of 0.3.
X. Liu et al. / Chemosphere 64 (2006) 1619–1626
1621
Table 1
In vitro cytotoxicity of chlorophenols to L929 cells physicochemical parameters used to predict toxicity in QSAR models
P ef
1 vf
log(1/LC50)h
log Kowf
pKaf
v
r
Chemicals
log(1/LC50)g
2-CP
4-CP
2,6-DCP
2,4-DCP
3,4-DCP
2,4,6-TCP
2,3,4-TCP
2,3,5-TCP
2,3,5,6-TeCP
PCP
2.63
2.68
3.06
3.11
3.23
3.42
3.42
3.62
3.87
4.03
2.15a
2.39a
2.84b
3.21b
3.44b
3.75b
4.07a
4.21a
4.90a
5.04b
2.85
2.93
3.35
3.89
3.74
3.81
3.74
4.12
4.27
4.25
8.52c
9.14d
6.78b
7.90b
8.62b
5.99b
6.50d
6.43c
5.03c
4.74b
P
2.653d
2.647d
3.171d
3.165d
3.165d
3.684d
3.685d
3.685e
4.204e
4.733d
0.227
0.227
0.454
0.454
0.600
0.681
0.827
0.973
1.200
1.427
Dgf
23.15
23.33
24.8
24.99
24.79
26.64
26.17
26.45
27.72
28.81
The unit of LC50 values is mol/l.
a
From Hansch and Leo.
b
From Xie and Dyrssen.
c
From Konemann and Musch.
d
From Saito et al.
e
Calculated after Kier and Hail.
f
Cited from literature (Emanuele and Cinzia, 1999).
g
LC50 values are for 24 h.
h
LC50 values are for 48 h.
Both CoMFA and CoMSIA regression models were
built by partial least squares (PLS) in conjunction with
leave-one-out (LOO) cross-validation to measure the internal consistency and the predictive ability of the obtained
QSAR models. The latent variable (LV) used to derive
the non-validation model was determined based on the
minimum standard error of prediction (SE) and the highest
cross-validation Q2, usually corresponding to the greatest F
ratio. The analysis of COMFA and COMSIA was performed by use of Sybyl 6.7 on a SGI O2 workstation running IRIX 6.5.
3. Results and discussion
3.1. 2D-QSAR
viability
P Cell
Pwas significantly correlated with log Kow,
re, 1vv and Dg for 24 h incubations but only log Kow
was significantly correlated with the 48 h cell viability data
(Table 2). The greatest proportion of the variance was
explained by the log Kow for both 24 and 48 h exposures
(r2 = 0.982, and r2 = 0.873, respectively). The log Kow is
generally a significant predictor of in vitro biological activity of chlorophenols (Beltrame et al., 1984; Saito and Sudo,
1991; Shannon et al., 1991; Fent and Hunn, 1996) because
it is a good predictor of the accumulated fraction for as a
function of hydrophobicity.
The Hammett factor was significantly correlated with
cell viability expressed as log(1/LC50) (r2 = 0.966). This
result is similar to that observed in other QSAR studies
involving other bioassays (Saito and Sudo, 1991; Shannon
1 v
et
P al., 1991). This result also suggests that both v and
Dg are significant predictors of cell viability during
24 h exposures. The steric property might be a significant
predictor of cytotoxicity, since 1vv is P
a function of both steric and electronic properties while Dg represents steric
properties of aromatic compounds (Taillandier et al.,
1983; Rvanel et al., 1985).
Table 2
Linear regression of molecular descriptors with in vitro toxicity during 24 h exposures
Parameters
log(1/LC50)
r2
r2adj
SE
F
24 h
log Kow
pKa
P
re
1 v
v
P
Dg
0.466 log Kow + 1.631
5.206 0.273 P
pKa
2.507 + 1.132 re
0.921 + 0.686 1vv
P
0.250 Dg 3.118
0.982
0.811
0.966
0.953
0.967
0.980
0.787
0.961
0.947
0.963
0.065
0.213
0.091
0.106
0.089
441.89
34.28
225.52
162.14
235.43
48 h
log Kow
pK
P ae
r
1 v
v
P
Dg
1.968 + 0.480 log Kow
5.479 0.256 P
pKa
2.911 + 1.108 re
1.372 + 0.668 1vv
P
0.250 Dg 2.739
0.873
0.599
0.776
0.757
0.812
0.858
0.549
0.748
0.726
0.789
0.191
0.339
0.254
0.264
0.232
55.18
11.97
27.66
24.89
34.60
1622
X. Liu et al. / Chemosphere 64 (2006) 1619–1626
To determine which physicochemical parameters
provide the greatest predictive capacity of the in vitro toxicity of chlorophenols, effective QSAR models were developed from multiple linear regressions by use of forward
addition for 24 (Eq. (1)) and 48 h (Eq. (2)) exposures,
respectively
X
logð1=LC50 Þ ¼ 0:296 log K ow þ 0:09454
Dg 0:187 ð1Þ
logð1=LC50 Þ ¼ 1:968 þ 0:480 log K ow
ð2Þ
Based on these relationships in vitro toxicity to L929 cells
(log(1/LC50)) of 10 chlorophenol isomers were predicted
for 24 h and 48 h (Table 4). The accuracy of the predictions
was assessed by the residuals between the observed and expected values (Table 4).
P
The results indicate that log Kow and
Dg are dominant factors in predicting the toxicity for 24 h exposures,
while log Kow is the only factor contributing significantly
to describing the variation in toxicity during the 48 h
exposure. The log Kow has often
P been used to describe
hydrophobic properties and
Dg has been proposed to
describe steric properties of aromatic compounds (Taillandier et al., 1983; Rvanel et al., 1985). Thus, during
24 h exposures, the steric properties seem to be an important descriptor in addition to hydrophobic properties,
while only the hydrophobic properties control the toxicity
during 48 h exposures. This result is different from previously reported 2D-QSAR models where only log Kow was
used as a predictor of bioaccumulation (Jiang et al.,
2004).
The predictive model for the 24 h exposures was more
predictive than was the model predicting toxicity during
48 h exposures (Table 3). To explain this phenomenon,
we constructed the graphic (Fig. 1) illustrating the relationship between log(DLC50), in which DLC50 means the
difference between 48 h LC50 and 24 h LC50, and log Kow
of chlorophenols with high correlation (r2 = 0.9216). The
greater the values of the log Kow the more toxic (lesser
LC50) chlorophenols were for both 24 and 48 h exposures.
One explanation of the relationship between hydrophobicity and time of exposure is that longer exposure
results in greater accumulation, which can compensate of
lesser log Kow values (Fig. 1). However, an alternate
hypothesis would be that the degree of toxicity was a function of the rate of damage relate top the rate of repair. It is
difficult to differentiate between these two hypotheses without measuring the concentrations of the compounds in the
cells.
Table 4
Predicted in vitro toxicities of chlorophenols to L929 cells during 24 and
48 h exposures
Compound
2-CP
4-CP
2,6-DCP
2,4-DCP
3,4-DCP
2,4,6-TCP
2,3,4-TCP
2,3,5-TCP
2,3,5,6-TeCP
PCP
a
b
Model for 24 h
Model for 48 h
Obs.
Pre.b
Res.a
Obs.
Pre.b
Res.a
2.63
2.68
3.06
3.11
3.23
3.42
3.42
3.62
3.87
4.03
2.64
2.73
3.00
3.13
3.18
3.44
3.49
3.56
3.88
4.03
0.01
0.05
0.06
0.02
0.05
0.02
0.07
0.06
0.01
0.00
2.85
2.93
3.35
3.89
3.74
3.81
3.74
4.12
4.27
4.25
3.00
3.11
3.33
3.51
3.62
3.77
3.92
3.99
4.32
4.39
0.15
0.18
0.02
0.38
0.12
0.04
0.18
0.13
0.05
0.14
Res. = Obs. Pre.
Calculated based on Eqs. (1) and (2).
3.2. 3D-QSAR
The 2D-QSAR analysis demonstrated that hydrophobicity and molecular bulk are the primary determinants
of in vitro toxicity of L929 cells exposed to chlorophenols,
for 24 h, and that log Kow is the most useful predictor of
biological toxicity during 48 h exposures. Descriptors of
molecular properties used to predict activity in 2D-QSAR
models are chosen empirically, and thus do not describe
the entire range of possible molecular fields. In addition,
it is not possible to determine mechanisms of action from
2D-QSAR models. Therefore, to further investigate the
potential mechanisms, a 3D-QSAR analysis based on
CoMFA and CoMSIA was undertaken.
3.2.1. CoMFA analysis
The results of the PLS analysis of a CoMFA model for
prediction of in vitro toxicity of chlorophenols to L929 cells
are given (Table 6). The results of the PLS analysis demonstrated that the CoMFA model was both highly internally
consistent with good predictivity (r2 = 0.968 and Q2 =
0.891 and r2 = 0.727 and Q2 = 0.394 for 24 and 48 h,
respectively). From a statistical perspective, the contributions of steric and electrostatic fields were 60.0% and
40.0% for 24 h and 59.4% and 40.6% for 24 h and 48 h
exposures, respectively. Thus, the electrostatic feature of
the chlorophenol molecules seems to be an important factor for the prediction of in vitro inhibition of cell proliferation of L929 cells.
The 3D-QSAR color-coded contour plots for the steric
and electrostatic fields in the CoMFA studies using
Table 3
Multiple factors linear regression analysis for in vitro toxicity during 24 and 48 h exposures
Timea
Modelb
r2
r2adj
SE
F
24 h
48 h
P
log(1/LC50) = 0.296 log Kow + 0.09454 Dg 0.187
log(1/LC50) = 1.968 + 0.480 log Kow
0.990
0.873
0.987
0.858
0.052
0.191
355.18
55.18
a
b
Duration of LC50 values.
Using forward regression.
X. Liu et al. / Chemosphere 64 (2006) 1619–1626
1623
Table 5
In vitro toxicities of chlorophenols to L929 cells during 24 and 48 h
exposures predicted from CoMFA models
Compound
Fig. 1. The relationship between log(DLC50) and log Kow of
chlorophenols (r2 = 0.9216) DLC50 equals that LC50 for 48 h subtract
LC50 for 24 h.
2,3,5,6-TeCP as a reference structure are presented (Figs.
2–5).
The electrostatic field contour maps of CoMFA for
inhibition of L929 cells exposed to Chlorophenols solutions are given (Figs. 2 and 4). Blue1 contours (above the
benzene ring) describe the regions where a positively
charged group enhances the toxicity of chlorophenols to
L929 cells. Negatively charged groups depicted by red
regions can be observed at the 2,3,4-positions. The green
contours in Figs. 3 and 5 represent the regions of high steric tolerance, which indicate that a bulky substituent is preferred in the position (1,3,5-positions) to produce greater
toxicity, while yellow contours represent regions of unfavorable steric effects.
Observed and predicted toxicities of chlorophenols during 24 and 48 h exposures based on CoMFA are given
(Table 5). The accuracy of the predictive models is assessed
by a residuals analysis. These results are consistent with
those of the 2D-QSAR analysis.
3.2.2. CoMSIA analysis
The CoMSIA is an alternative approach to performing
3D-QSAR by comparative molecular field analysis. CoMSIA is similar to CoMFA, but uses a Gaussian function
rather than Coulombic and Lennard-Jones potentials to
assess the contribution from different fields. In addition
to steric and electrostatic fields used in CoMFA, CoMSIA
defines explicit hydrophobic and hydrogen bond donor–
acceptor fields, which are not available with standard
CoMFA. The CoMSIA results provide a cross-validated
Q2 value of 0.906, while the non-cross-validated r2 with
the three components electrostatic, steric and hydrophobic,
which are the dominant contributors to the prediction of
in vitro toxicities of the chemicals, is 0.972. The CoMSIA
steric and electrostatic plots are similar to those obtained
1
For interpretation of color in figures, the reader is referred to the Web
version of this article.
Model for 24 h
2-CP
4-CP
2,6-DCP
2,4-DCP
3,4-DCP
2,4,6-TCP
2,3,4-TCP
2,3,5-TCP
2,3,5,6-TeCP
PCP
Model for 48 h
Obs.
Pre.
Res.a
Obs.
Pre.
Res.a
2.63
2.68
3.06
3.11
3.23
3.42
3.42
3.62
3.87
4.03
2.74
2.77
3.06
2.98
3.23
3.29
3.44
3.59
3.88
4.11
0.11
0.09
0.00
0.13
0.00
0.13
0.02
0.03
0.01
0.08
2.85
2.93
3.35
3.89
3.74
3.81
3.74
4.12
4.27
4.25
3.24
3.12
3.57
3.38
3.49
3.70
3.75
3.98
4.30
4.42
0.39
0.19
0.22
0.51
0.25
0.11
0.01
0.14
0.03
0.17
Residuals between observed and predicted values are given.
a
Res. = Obs. Pre.
Table 6
Summary of CoMFA models for chlorophenols
Exposure
time (h)
Q2
24
48
0.891
0.394
r2
0.968
0.727
SE
0.093
0.281
F
160.35
21.257
Relative contribution
Eletrostatic
Steric
0.600
0.594
0.400
0.406
Fig. 2. CoMFA contour map of electrostatic field for 24 h. Positive charge
favored areas are represented by blue and negative charge favored ones are
represented by red.
from CoMFA. The additional hydrophobic fields (yellow,
hydrophobic group favored) shown in Fig. 6, which is
the dominant predictor of in vitro toxicity to the cells,
demonstrates that the chloro group in the 3- and 5-position
fits the yellow region well which is favorable for hydrophobic groups, and thus explains the log(1/LC50) values for
3,4-DCP vs. 2,4-DCP and 2,3,5-TCP vs. 2,4,6-TCP
(Table 1).
The results of the CoMSIA analysis using five single
molecular field descriptors, demonstrates that steric,
1624
X. Liu et al. / Chemosphere 64 (2006) 1619–1626
Fig. 6. CoMSIA contour map of hydrophobic field for 24 h. Areas
contoured by yellow indicate regions favorable for hydrophobic group.
Fig. 3. CoMFA contour map of steric field for 24 h. Areas contoured by
green indicate regions favorable for steric occupancy, while areas
contoured by yellow indicate the opposite.
Fig. 4. CoMFA contour map of electrostatic field for 48 h. Positive charge
favored areas are represented by blue and negative charge favored ones are
represented by red.
Fig. 5. CoMFA contour map of steric field for 48 h. Areas contoured by
green indicate regions favorable for steric occupancy, while areas
contoured by yellow indicate the opposite.
2electrostatic and hydrophobic fields are significant predictors of in vitro toxicity of chlorophenols to L929 cells
exposed for 24 h (Table 7). Therefore, the final model
(Q2 = 0.906, r2 = 0.972) used in this study is built based
on the stepwise combination of these three fields (with relative contributions of 0.767:0.225:0.008, respectively).
Consistent with the CoMFA model, the electrostatic field
parameter is still the dominant prediction of the inhibition
of L929 cells by chlorophenols. Furthermore, hydrophobicity of chlorophenols is important while the steric field
can contribute to the toxicity of chlorophenols very appreciably (Table 7).
From the comparison of PLS analyses based on molecular field descriptors (Table 8), the intercorrelations of
these numerically intensive grid fields are obvious. In all
of these models, it is possible to determine whether a field
is important or not. For example, in the COMSIA model
including electrostatic and steric fields the values of 0.99
and 0.01 indicate that the electrostatic field parameter is
an extremely important factor contributing to the toxicity
of chlorophenols. Meanwhile, the hydrophobic field factor
is also relatively important in predicting toxicity of the
chlorophenols with the COMSIA model with hydrophobic
and steric fields (0.966:0.034). Besides the above models, a
third mode based on electrostatic and hydrophobic fields
(0.773:0.227) suggests that the electrostatic field parameter
is more important for predicting the toxicity of chlorophenols than is the hydrophobic field parameter. Similar
results were obtained from CoMSIA analysis for data from
48 h exposures (Data not presented.).
The presence of chlorine at the 3- and 5-positions
favors hydroxylation of the 4-position of chlorophenols.
Hydroxylated products can result in the corresponding
chlorohydroquinones, which may enhance toxicity and carcinogenicity of chlorophenols. This prediction is consistent
with the results of the field contour maps of CoMFA and
CoMSIA.
With the use of 3D-QSAR: CoMFA and CoMSIA techniques we have been able to elucidate toxicological mechanism of the ligand–receptor interaction that can be used to
develop testable hypotheses of the mechanism of in vitro
toxicity of chlorophenols on L929 cells.
X. Liu et al. / Chemosphere 64 (2006) 1619–1626
1625
Table 7
Statistical results of CoMSIA with five single fields for predicting in vitro toxicity of chlorophenols to L929 cells during 24 h exposures
CoMSIA
2
Q
r2
SE
LV
F
Steric
Electrostatic
H-bond donor
H-bond acceptor
Hydrophobic
0.882
0.966
0.097
2
98.055
0.907
0.972
0.087
2
123.276
0.256
0.554
0.35
2
4.351
0.271
0.565
0.345
2
4.539
0.884
0.966
0.097
2
98.364
Table 8
Statistical results of CoMSIA with combination fields of electrostatic, steric and hydrophobic for predicting in vitro toxicity of chlorophenols on L929 cells
during 24 h exposures
CoMFA
Electrostatic +
hydrophobic
Electrostatic +
steric
Hydrophobic +
steric
Electrostatic +
hydrophobic + steric
Q2
r2
SE
LV
F
0.906
0.972
0.088
2
121.092
0.908
0.972
0.087
2
123.219
0.883
0.966
0.097
2
98.357
0.906
0.972
0.088
2
121.006
0.99
0.966
0.01
0.034
0.767
0.225
0.008
Fraction
Electrostatic
Hydrophobic
Steric
0.773
0.227
Acknowledgements
This work was supported by the National Natural
Science Foundation of China PR, No. 20237010 and
No. 20375015; the Core University Program of Japan Society for Promotion of Science.
References
Beltrame, P., Beltrame, P.L., Carniti, P., 1984. Inhibiting action of chloro
and nitrophenols on biodegradation of phenol: a structure–toxicity
relationship. Chemosphere 13, 3–9.
Carstens, C.P., Blum, J.K., Witte, I., 1990. The role of hydroxyl radicals in
tetrachloro-hydroquinone induced DNA strand break formation in
PM2 DNA and human fibroblasts. Chem. Biol. Interact. 74, 305–
314.
Dahlhaus, M., Almstadt, E., Appel, K.E., 1994. The pentachlorophenol
metabolite tetrachloro-p-hydroquinone induces the formation of
8-hydroxy-2-deoxyguanosine in liver DNA of male B6C3F1 mice.
Toxicol. Lett. 74, 265–274.
Dahlhaus, M., Almstadt, E., Henschke, P., Luttgert, S., Appel, K.E.,
1995. Induction of 8-hydroxy-2-deoxyguanosine and single strand
breaks in DNA of V79 cells by tetrachloro-phydroquinone. Mutat.
Res. 329, 29–36.
Dahlhaus, M., Almstadt, E., Henachke, P., Luttgert, S., Appel, K.E.,
1996. Oxidative DNA lesions in V79 cells mediated by pentachlorophenol metabolites. Arch. Toxicol. 70, 457–460.
Dercova, K. et al., 2003. Biodegradation and bioremediation of pentachlorophenol. Chemicke Listy 10, 991–1002.
Ehrlich, W., 1990. The effect of pentachlorophenol and its metabolite
tetrachlorohydroquinone on cell growth and the induction of DNA
damage in Chinese hamster ovary cells. Mutat. Res. 244, 299–
302.
Emanuele, Argese, Cinzia, Bettiol, et al., 1999. Quantitative structure–
activity relationship for the toxicity of chlorophenols to mammalian
submitochondrial particles. Chemosphere 38 (10), 2281–2292.
Fent, K., Hunn, J., 1996. Cytotoxicity of organic environmental chemicals
to fish liver cells (PLHC-1). Biochim. Biophys. Anta. 205, 1–6.
IARC, 1991. IARC Monographs on the Evaluation of Carcinogenic Risks
to Humans: Occupational Exposures in Insecticide Application, and
Some Pesticides, Lyon, vol. 53, pp. 371–402.
Jansson, K., Jansson, V., 1991. Induction of mutation in V79 Chinese
hamster cells by tetrachlorohydroquinone, a metabolite of pentachlorophenol. Mutat. Res. 260, 83–87.
Jansson, K., Jansson, V., 1992. Induction of micronuclei in V79 Chinese
hamster cells by tetrachlorohydroquinone, a metabolite of pentachlorophenol. Mutat. Res. 279, 205–208.
Jiang, J., Chen, J.N., Yu, H.X., Zhang, F., Zhang, J.F., Wang, L.S., 2004.
Quantitative structure activity relationship and toxicity mechanisms of
chlorophenols on cells in vitro. Chin. Sci. Bull. 49, 987–989.
Juhl, U., Witte, I., Butte, W., 1985. Metabolism of pentachlorophenol to
tetrachloro-hydroquinone by human liver homogenate. Bull. Environ.
Contam. Toxicol. 35, 596–601.
Liu, Y., Yu, H.X., 2005. Three-dimensional quantitative structure activity
relationship (3D-QSAR) analysis for in vitro toxicity of chlorophenols
to HepG2 cells. Chemosphere 60 (6), 791–795.
Mosmann, T., 1983. Rapid colorimetric assay for cellular growth and
survival: application to proliferation and cytotoxicity assays. J.
Immunol. Methods 65, 55–63.
Renner, G., Mucke, W., 1986. Transformations of pentachlorophenol
Part I: Metabolism in animals and man. Toxicol. Environ. Chem. 11,
9–29.
Rvanel, P., Taillander, G., Tissut, M., Benoit-Guyod, J.L., 1985. Effects of
chlorophenols on isolated plant mitochondria activities: a QSAR
study. Ecotoxical. Environ. Safety 9, 300–320.
Saito, H., Sudo, M., et al., 1991. In vitro cytotoxicity of chlorophenols to
goldfish GF-Scale (GFS) cells and quantitative structure–activity
relationships. Environ. Toxicol. Chem. 10, 235–241.
Seiler, J.P., 1991. Pentachlorophenol. Mutat. Res. 257, 27–47.
Shannon, R.D., Boarman, G.D., Dietrich, A.M., Bevan, D.R., 1991.
Mitochondrial response to chlorophenols as a short-term toxicity
assay. Environ. Toxicol. Chem. 10, 57–66.
Taillandier, G., Domard, M., Boucherie, A., 1983. QSAR at series
aromatiques: propositions de parameters steriques. Farmaco-Edizione
Scientifica 38, 473–487.
1626
X. Liu et al. / Chemosphere 64 (2006) 1619–1626
Van Ommen, B., Adang, A., Muller, F., Van Bladeren, P.J., 1986. The
microsomal metabolism of pentachlorophenol and its covalent binding
to protein and DNA. Chem. Biol. Interact. 60, 1–11.
Wang, Y.J., Ho, Y.S., Chu, S.W., et al., 1997. Induction of glutathione
depletion, p53 protein accumulation and cellular transformation by
tetrachlorohydroquinone, a toxic metabolite of pentachlorophenol.
Chem. Biol. Interact. 105, 1–16.
Witte, I., Juhl, U., Butte, W., 1985. DNA-damaging properties and
cytotoxicity in human fibroblasts of tetrachlorohydroquinone, a
pentachlorophenol metabolite. Mutat. Res. 145, 71–75.