Development of a General Quantum-Chemical Descriptor

for Steric Effects: Density Functional Theory Based QSAR

Study of Herbicidal Sulfonylurea Analogues

ZHEN XI,1 ZHIHONG YU,1 CONGWEI NIU,1 SHURONG BAN,1 GUANGFU YANG2
1
State Key Laboratory of Elemento-Organic Chemistry and Department of Chemical Biology,
Nankai University, Tianjin, People’s Republic of China, 300071
2
Key Laboratory of Pesticide and Chemical Biology of Ministry of Education,
College of Chemistry, Central China Normal University, Wuhan 430079,
People’s Republic of China

Received 26 February 2006; Revision 13 April 2006; Accepted 14 April 2006

DOI 10.1002/jcc.20464
Published online in Wiley InterScience (www.interscience.wiley.com).

Abstract: Quantitative structure-activity relationship (QSAR) analysis has become one of the most effective
approaches for optimizing lead compounds and designing new drugs. Although large number of quantum-chemical
descriptors were defined and applied successfully, it is still a big challenge to develop a general quantum-chemical
descriptor describing the bulk effects more directly and effectively. In this article, we defined a general quantum-
chemical descriptor by characterizing the volume of electron cloud for specific substituent using the method of density
functional theory. The application of our defined steric descriptors in the QSAR analysis of sulfonylurea analogues
resulted in four QSAR models with high quality (the best model: q2 ¼ 0.881, r2 ¼ 0.901, n ¼ 35, s ¼ 0.401, F ¼
68.44), which indicated that this descriptor may provide an effective way for solving the problem how to directly
describe steric effect in quantum chemistry-based QSAR studies.
q 2006 Wiley Periodicals, Inc. J Comput Chem 27: 1571–1576, 2006

Key words: quantum-chemical descriptor; density functional theory; QSAR; sulfonylurea herbicide; acetohydroxya-
cid synthase

Introduction In 2002, Duggleby and coworkers reported the first crystal

Acetohydroxyacid synthase (AHAS, EC 2.2.1.6)1,2 is a key
enzyme in branched-chain amino acid metabolism in plants,
fungi, bacteria, and archaea,3 and several herbicides (such as sul-
fonylurea, imidazolinone, triazolopyrimidine, etc.) that are widely
used in agriculture act as specific and potent inhibitors of this
enzyme.4 These inhibitors do not resemble the substrates or prod-
ucts of the AHAS catalyzed reaction. Neither do they resemble
the organic cofactors required by AHAS, such as thiamine
diphosphate (ThDP) and FAD. Thus, it has been suggested that
these herbicides are ‘‘extraneous site inhibitor’’ that do not
occupy the active site.5 Therefore, enzyme mutations that dimin-
ish the binding affinity of herbicides without abolishing the cata-
lytic activity of AHAS resulted in rapid formation of resistance of
commercial AHAS inhibitors. While in last several years, many
studies6–11 have shown that these mutations displayed greatly dif-
ferent sensitivity to different sulfonylureas, which indicated that
developing new AHAS inhibitors is an alternative way to over-
come herbicidal resistance of this family.
structures of free catalytic subunit of yeast AHAS.12 Subsequently,
five crystal structures of this catalytic subunit of yeast AHAS in
complex with chlorimuron-ethyl, chlorsulfuron, sulfometuron
methyl, metsulfuron methyl, and tribenuron methyl were deter-
Correspondence to: Prof. Z. Xi; e-mail: zhenxi@nankai.edu.cn or Prof.
G.-F. Yang; e-mail: gfyang@mail.ccnu.edu.cn
Contract/grant sponsor: NSFC; contract/grant numbers: 20432010,
20572030
Contract/grant sponsor: MOST; contract/grant number: 2003CB1144003
Contract/grant sponsor: Key Project of Chinese Ministry of Education;
contract/grant numbers: 104189, 104205
Contract/grant sponsor: Program for Excellent Research Group of Hubei
Province; contract/grant number: 2004ABC002
Contract/grant sponsor: ISC, Nankai University
This article contains supplementary material available via the Internet at
http://www.interscience.wiley.com/jpages/0192-8651/suppmat

q 2006 Wiley Periodicals, Inc.

1572 Xi et al. • Vol. 27, No. 13 • Journal of Computational Chemistry

Table 1. The General Structure of Sulfonylurea; Detailed Structures, Experimental pKi Values, Predicted pKi
Values by Model 3 and the Residual for all 45 Derivatives.

No. R1 R2 R3 R4 Exp. pKi Pred. pKi Residual

Training set
1 Cl CH3 H H 5.22 5.60
0.38
2 Cl OCH3 H OCH3 6.93 6.95
0.02
3 Cl OCH3 Br OCH3 5.52 5.25 0.27
4 Cl OCH3 Br Cl 4.78 4.49 0.29
5 Cl CH3 Br CH3 4.67 4.46 0.21
6 Cl CH2OC2H5 Br H 4.27 4.67
0.40
7 Cl CH2OOCCH3 Br H 4.70 4.70 0.00
8 COOEt CH3 H H 6.69 7.37
0.68
9 COOEt CH2OC2H5 Br H 6.51 6.37 0.14
10 COOEt OCH3 Br OCH3 7.03 7.00 0.03
11 COOEt OCH3 H OCH3 9.20 8.75 0.45
12 COOEt OCH3 H Cl 8.34 8.34 0.00
13 COOEt CH3 H Cl 7.09 7.56
0.47
14 COOEt OCH3 Br Cl 6.80 6.89
0.09
15 COOEt CH2OOCCH¼¼CH2 Br H 7.11 6.38 0.73
16 COOMe OCH3 Br OCH3 6.31 6.54
0.23
17 COOMe CH3 H CH3 7.69 6.87 0.82
18 COOMe OCH3 H OCH3 8.63 8.25 0.38
19 COOMe OCH3 H Cl 7.77 7.71 0.06
20 COOMe CH2OCH2CH¼¼CH2 Br H 6.04 5.84 0.20
21 COOMe CH2OOCCH3 Br H 5.20 5.75
0.55
22 COOMe OCH3 H H 6.66 7.58
0.92
23 NO2 CH2OC2H5 Br H 4.68 4.75
0.07
24 NO2 CH3 Br CH3 4.80 4.73 0.07
25 NO2 OCH3 Br OCH3 5.30 5.52
0.22
26 NO2 OCH3 H OCH3 8.00 7.18 0.82
27 NO2 OCH3 H H 5.82 6.51
0.69
28 NO2 CH2OOCCH¼¼CH2 Br H 5.17 4.83 0.34
29 OCH2CH2Cl OCH3 H OCH3 7.46 7.57
0.11
30 OCH2CH2Cl CH3 H Cl 6.58 6.37 0.21
31 OCH2CH2Cl CH2SCN Br H 5.58 5.27 0.31
32 OCH2CH2Cl CH2OC2H5 Br H 5.14 5.30
0.16
33 OCH2CH2Cl CH2OCH2CH¼¼CH2 Br H 5.31 5.31 0.00
34 OCH2CH2Cl OCH3 Br OCH3 5.45 5.75
0.30
35 OCH2CH2Cl OCH3 Br Cl 4.77 4.81
0.04
Test set
36 Cl CH3 H CH3 4.71 5.65
0.94
37 Cl CH2OCH3 Br H 5.66 4.68 0.98
38 Cl CH2SCN Br H 4.63 4.74
0.11
39 COOEt CH2OCH3 Br H 6.70 6.32 0.38
40 COOEt CH3 Br CH3 6.78 6.36 0.42
41 COOMe CH3 Br CH3 5.94 5.76 0.18
42 NO2 CH2OCH3 Br H 5.54 4.73 0.82
43 NO2 OCH3 H Cl 7.69 6.61 1.08
44 NO2 CH2OOCCH3 Br H 4.53 4.79
0.26
45 OCH2CH2Cl CH3 H H 5.33 6.32
0.99

Journal of Computational Chemistry DOI 10.1002/jcc

 Density Functional Theory Based QSAR Study of Herbicidal Sulfonylurea Analogues
Table 2. QSAR Models and Relevant Statistical Indices.
Model Equation
1 pKi ¼ 4.479 + 0.052VR1 + 0.056VR2
0.172VR3 + 0.127VR4
(0.010) (0.015) (0.022) (0.023)
2 pKi ¼ 2.333 + 0.061VR1
2.291Q1
8.365Q17
(0.008) (0.435) (0.781)
3 pKi ¼ 2.450 + 0.060VR1
1.895Q1
8.365Q17
0.050FN 12
(0.008) (0.445) (0.734) (0.022)
4 pKi ¼ 2.586 + 0.061VR1
1.871Q1
7.273Q17
0.062F 12 + 0.034VR4
N
(0.007) (0.420) (0.854) (0.022) (0.016)
The figures in parentheses under each coefficient are standard error of the regression coefficient under the 95%
confidence interval.
mined by the same group.13,14 Although these crystal structures
provided a solid experimental basis for understanding the detailed
molecular mechanism of ligand–receptor interactions, the crystal
structure of the whole enzyme, especially from plant, or the com-
plex of the whole enzyme with inhibitors has not yet been avail-
able so far. Therefore it is necessary to find other ways to under-
stand the mechanism of receptor–ligand interaction and herbicide
resistance, which is crucial for further designing new compounds
with high activity against mutant and wild AHAS. Former stud-
ies15–19 have shown that quantitative structure-activity relationship
(QSAR)20 may be an effective approach for this purpose.
As for QSAR, its performances depend greatly on what descrip-
tors are employed. With the advantage of precise quantitative de-
scriptions of molecular structures and chemical properties, a large
number of quantum-chemical descriptors were defined and applied
successfully.21 However, it is still a big challenge to develop a quan-
tum-chemical descriptor describing the bulk effects more directly
and effectively. Recently, Wang et al.22 and Han et al.23 have tried
to use the volume of electron cloud of a whole molecule as steric
descriptor, but no satisfactory QSAR model containing this descrip-
tor could be obtained. Keeping in mind that those activities of a se-
ries of analogues changed mainly due to different substitution, the
volume of electron cloud of a specific substituent may be a prefera-
ble quantum-chemical descriptor for directly describing the bulk
effects of substituents. As far as computational technique is consid-
ered, many practices22–26 have proved that density functional theory
(DFT),27 which takes into account the exchange and correlation
effects effectively, is most likely one of the best methods to study
medium-size or larger molecular systems at the trade-off of accu-
racy and computational costs and appropriate for QSAR study, with
exhibiting excellent performance than semiempirical method or
some other ab initio methods.
For one purpose of further understanding the interaction mech-
anism and designing new sulfonylureas as inhibitors of AHAS,
and the other purpose of developing a new quantum-chemical de-
scriptor describing the bulk effects more directly and effectively,
we carried out herein DFT-based QSAR studies on a series of sul-
fonylurea derivatives.
Materials and Methods
Three-dimensional structures were constructed using SYBYL 6.9
software28 running on Silicon Graphics Inc. Origin350 server
Journal of Computational Chemistry
1573
r2 q2 s F
0.835 0.792 0.518 37.86
0.885 0.862 0.433 79.11
0.901 0.881 0.401 68.44
0.915 0.887 0.371 62.61
with IRIX 6.5.21 operating system; DFT calculations were carried
out using Gaussian 03 program29 running on a Linux cluster;
QSAR analyses were also performed on the cluster.
Data Set
Forty-five sulfonylurea derivatives were employed, including two
commercial herbicides (compound 12, 17) and forty-three deriva-
tives,30,31 their structures and experimental inhibitory activities
against wild-type Escherichia coli AHAS isoenzyme II are listed
in Table 1; obviously they varied at the positions of R1, R2, R3, and
R4. These compounds were randomly divided into training set (com-
pound 1 to 35) and test set (compound 36 to 45).
Molecular Modeling and DFT Calculations
It has been proved by X-ray analysis of five crystal structures of
ligand–receptor complex that sulfonylurea analogues interacted with
AHAS in a similar way.13,14 Then, the conformation of chlorimuron-
ethyl in the crystal structure of complex was extracted as the tem-
plate to construct initial structures for DFT geometry optimization
using B3LYP32,33 functional with 6-31G(d) basis set, and subsequent
frequency analysis was performed to characterize whether the opti-
mization arrived at a stationary point. Net atomic charges were
derived from the population analysis with NBO method. The volume
of electron cloud of a molecule was defined as the one inside a con-
tour of certain electron density. It was calculated with the keyword
of ‘‘volume’’ based on the optimized structure, the number of points
per bohr3 for Monte-Carlo calculation of molar volume was changed
from 20 (default value) to 1000 so as to increase the precision, and
the average of five-times calculations was taken as the final molecu-
lar volume of electron cloud (see supporting information).
Quantum-Chemical Descriptors
For such a multi-substituted system described in Table 1, we defined
the volume of electron cloud of a specific substituent Ri as follow:
" , #
X4
VRi ¼
V  ðVRi
H
VH
H Þ ðVRi
H
VH
H Þ
i¼1
where DV is the difference of the molecular volume of electron
cloud between certain analogue and the non-substituted molecule;
VH

H is the volume of H2 molecule; VRi–H is the volume of the
DOI 10.1002/jcc
1574 Xi et al. • Vol. 27, No. 13 • Journal of Computational Chemistry

Table 3. The Values of Quantum-Chemical Descriptors in all Four Models.

N
No. VR1 VR2 VR3 VR4 Q1 Q17 F12

Training set
1 7.570 7.977 0.000 0.000
0.029
0.337 3.546
2 7.970 13.099 0.000 13.099
0.029
0.492 3.065
3 7.757 12.748 11.142 12.748
0.030
0.296 3.992
4 7.783 12.792 11.180 7.783
0.030
0.227 7.631
5 7.935 8.362 11.398 8.362
0.029
0.205 4.743
6 7.930 30.605 11.391 0.000
0.029
0.206 0.763
7 7.972 31.647 11.451 0.000
0.030
0.209 0.777
8 30.735 7.990 0.000 0.000
0.138
0.339 0.500
9 32.310 30.764 11.450 0.000
0.138
0.208 0.597
10 30.407 12.330 10.776 12.330
0.139
0.297 0.599
11 32.155 13.038 0.000 13.038
0.138
0.493 0.487
12 34.504 13.991 0.000 8.513
0.138
0.428 0.600
13 32.798 8.527 0.000 8.092
0.138
0.347 0.578
14 31.239 12.667 11.071 7.707
0.138
0.279 0.707
15 31.919 36.493 11.311 0.000
0.139
0.212 0.679
16 22.643 12.796 11.183 12.796
0.140
0.297 0.585
17 22.774 8.252 0.000 8.252
0.140
0.335 0.477
18 23.780 13.439 0.000 13.439
0.140
0.493 0.491
19 24.020 13.574 0.000 8.259
0.140
0.427 0.589
20 22.029 34.864 10.880 0.000
0.140
0.218 0.600
21 21.746 29.684 10.740 0.000
0.140
0.210 0.665
22 23.124 13.068 .000 0.000
0.140
0.418 0.507
23 12.230 29.758 11.076 0.000 0.079
0.206 0.198
24 12.087 8.030 10.946 8.030 0.079
0.205 0.204
25 12.524 12.977 11.342 12.977 0.079
0.296 0.226
26 13.059 13.532 0.000 13.532 0.079
0.491 0.195
27 12.241 12.684 0.000 0.000 0.079
0.416 0.202
28 12.904 37.703 11.686 0.000 0.079
0.210 0.177
29 30.704 13.411 0.000 13.411 0.343
0.492 3.932
30 29.456 8.249 0.000 7.828 0.343
0.345 1.783
31 29.202 27.073 11.148 0.000 0.344
0.211 0.966
32 30.069 30.840 11.479 0.000 0.343
0.207 0.905
33 29.423 35.991 11.232 0.000 0.346
0.217 1.379
34 29.073 12.699 11.099 12.699 0.343
0.297 5.646
35 28.260 12.343 10.788 7.510 0.344
0.278 20.267
Test set
36 7.882 8.306 0.000 8.306
0.029
0.338 3.054
37 7.893 22.148 11.338 0.000
0.030
0.207 0.631
38 8.409 29.332 12.078 0.000
0.030
0.211 0.790
39 31.489 21.798 11.159 0.000
0.138
0.208 0.600
40 32.637 8.485 11.566 8.485
0.138
0.205 0.577
41 22.333 8.092 11.030 8.092
0.140
0.206 0.560
42 11.804 20.883 10.690 0.000 0.079
0.206 0.198
43 12.747 13.209 0.000 8.037 0.079
0.424 0.173
44 12.436 31.127 11.263 0.000 0.079
0.209 0.185
45 30.312 8.489 0.000 0.000 0.343
0.337 2.584

molecule formed by saturating substituent i with hydrogen atom;
the numerator in bracket is the volume increment between the
substituent i and hydrogen atom, and the denominator is the sum
of all four increments as shown in Table 1; the value of this frac-
tion is regarded as the weight coefficient of volume increment of
each substituent i, then VRi is the weighted volume of electron
cloud resulted from the change of substituent i in an analogue.
Previous theoretical studies by Bader et al.34,35 have shown
that the electron density contour of 0.001 e/bohr3 represents in
general the size and shape of one molecule in the gas phase.
While in our study, the defined weighted volumes inside the dif-
ferent contour of 0.01, 0.005, 0.001, 0.0005, and 0.0001 e/bohr3
were calculated and employed in subsequent QSAR studies,
respectively, to find out the optimal contour.
Other quantum-chemical descriptors were also derived from
DFT calculation, they were EHOMO (the orbital energy of HOMO),
ELUMO (the orbital energy of LUMO), DE (the energy difference
between LUMO and HOMO), QA (net charge of atom A), F EA

Journal of Computational Chemistry DOI 10.1002/jcc

 Density Functional Theory Based QSAR Study of Herbicidal Sulfonylurea Analogues 1575

(weighted electrophilic atomic frontier electron density) and F NA

(weighted nucleophilic atomic frontier electron density).25

QSAR Analyses

Classic multiple linear regression (MLR) technique was em-

ployed in QSAR analyses, the variable selection was performed
by Monte-Carlo simulated annealing, in which the number of
descriptors used in every trial model was controlled to the maxi-
mum of five. The results of 100-times annealing procedures were
analyzed together to find out some optimal models.

Results and Discussion

First of all, only four VRi descriptors were involved in multiple linear
regression arbitrarily to test their availabilities, the result is shown as
Model 1 in Table2. Positive coefficients of VR1, VR2 and VR4 showed
that the activity would increase with bulk groups at these positions, Figure 1. The plot of predicted versus experimental pKi values for
while negative coefficient of VR3 showed that small group of R3 training set (solid squares) and test set (hollow triangles) for Model 3.
would benefit the activity; these elementary hints were basically
consistent with experimental results. Moreover, good statistical indi-
increase the inhibitory activity, which can be represented by the
ces of Model 1 indicated that these weighted volumes are applicable
increasing activities of compounds 6, 23, 32, and 9. (2) More neg-
in QSAR study. Certainly, the activity is not only relative to bulk
ative net atomic charge on carbon-1 atom showed the increase of
effects but also to other factors, such as electrostatic interaction, H-
bioactivity, which is in accordance with the fact that electron-
bond interaction, etc. So all quantum-chemical descriptors derived
withdrawing substituent R1, with higher bioactivity, caused the
from DFT calculations were employed in subsequent analysis, and
more relative negative net atomic charge on carbon-1 atom. (3)
the values of some descriptors appeared in QSAR equations are
More relative positive net atomic charge resulting from electron
listed in Table3. After Monte-Carlo simulated annealing and analy-
withdrawing substituent R3 decrease the activity, which is evi-
ses, three optimal models were picked out and also summarized in
dently proved by the fact that bromine atom of R3 decreased the
Table 2 as Model 2, 3, and 4, respectively. The correlation matrixes
activity rapidly, such as in compounds 3 and 2; 10 and 11; 16 and
of descriptors in all four models are listed in Table 4, and correlation
18; 25 and 26; and 34 and 29. Moreover, the negative correlation
analysis showed that VR4 correlated with Q17 at the significance level
of F12N in Models 3 and 4 implied that the weaker the ability of
of 0.01, with correlation coefficient of
0.568 in Model 4. So
accepting electron by oxygen-12, the higher the activity would
Model 3 was selected as the final model, and its plot of predicted
be; the positive correlation of VR4 in Model 4 pointed that bulk
versus experimental pKi values is illustrated in Figure 1.
group of R4 would increase the activity, which may be found in
The obtained models in Table 2 are consistent with the exist-
compound 3 and 4; 11 and 12; 18, 19, and 22; 26 and 27; and 29
ing structure-activity relationships: (1) bulk group of R1 would
and 30. In addition, the ligand–receptor interactions revealed by
the above models have been observed in the reported crystal
structures of sulfonylurea–AHAS complex13,14: the sulfonylurea
Table 4. Correlation Matrixes of Descriptors in all Four Models.
bridge lies snugly in the active pocket located at the interface of
two subunit of the protein; free space around the substituent of R1
Correlation Matrixes
or R4 can accommodate bulk groups, which will block the entry
Model VR1 VR2 VR3 VR4 channel of substrate and resulted in the increase of inhibitory ac-
tivity. The side chain of Lys-251 and Arg-380 around R1 have
1 VR1 1.0000 positive charge, and electron-withdrawing substituents close to
VR2 0.025 1.000 this area neutralize the positive charge more easily so as to
VR3
0.154 0.554 1.000 increase the inhibitory activity. Oxygen-12 atom H-bonded with
VR4 0.031
0.653
0.253 1.000 the guanidino group of Arg-380, which could be reflected by the
N
term of F12 .
N As mentioned earlier, to find out the optimal volume for QSAR
VR1 Q1 Q17 F12 VR4
analyses, the defined weighted volumes inside the contour of 0.01,
0.005, 0.001, 0.0005, and 0.0001 e/bohr3 were employed, respec-
2 VR1 1.000
tively. The values of weighted volume in all four models were
3 Q1 0.078 1.000
Q17
0.143 0.137 1.000
changed, and the resulting r2 values are summarized in Table5.
4 N
F12
0.290 0.394 0.063 1.000 Although good models could be developed at all above five cou-
VR4 0.031
0.190
0.568 0.177 1.000 tours, the performances are different slightly. The results of too
large (0.01) or too small (0.0001) contour were somewhat worse,

Journal of Computational Chemistry DOI 10.1002/jcc

1576 Xi et al. • Vol. 27, No. 13 • Journal of Computational Chemistry

2
Table 5. The r Values of Models Resulted from Different Contour. 9. Hill, C. M.; Duggleby, R. G. Biochem J 1998, 335, 653.
10. Chang, A. K.; Duggleby, R. G. Biochem J 1998, 333, 765.
Contour Model 1 Model 2 Model 3 Model 4 11. Ibdah, M.; Barllan, A.; Livnah, O.; Schloss, J. V.; Barak, Z.;
Chipman, D. M. Biochemistry 1996, 35, 16282.
0.01 0.841 0.872 0.893 0.910 12. Pang, S. S.; Duggleby, R. G.; Guddat, L. W. J Mol Biol 2002, 317,
0.005 0.835 0.884 0.901 0.915 249.
0.001 0.808 0.876 0.901 0.913 13. Pang, S. S.; Guddat, L. W.; Duggleby, R. G. J Biol Chem 2003, 278,
0.0005 0.809 0.871 0.896 0.911 7639.
0.0001 0.780 0.865 0.891 0.905 14. McCourt, J. A.; Pang, S. S.; Guddat, L. W.; Duggleby, R. G. Bio-
chemistry 2005, 44, 2330.
15. Yang, G. F.; Zhao, G. F.; Lu, R. J.; Liu, H. Y.; Yang, H. Z. Sci China
Ser B Chem 1998, 41, 353.
which may be due to the fact that much tighter or looser volume 16. Yang, G. F.; Liu, H. Y.; Yang, X. F.; Yang, H. Z. Sci China Ser B
would characterize the bulk effects insufficiently or excessively, Chem 1999, 42, 656.
respectively. On the contrary, the results of the other three contours 17. Qian, X. H. J Agric Food Chem 1999, 47, 4415.
exhibited almost the same quality; especially, the result of 0.005 18. Hou, T. J.; Li, Z. M.; Li, Z.; Liu, J.; Xu, X. J. J Chem Inf Comput Sci
e/bohr3 is optimal in our case study. In fact, all equations in Table 2000, 40, 1002.
2 were developed at the contour of 0.005 e/bohr3. 19. Perkins, R.; Fang, H.; Tong, W. D.; Welsh, W. J. Environ Toxicol
Chem 2003, 22, 1666.
20. Hansch, C.; Fujita, T. J Am Chem Soc 1964, 86, 1616.
Summary and Conclusions
21. Karelson, M.; Lobanov, V. S.; Katritzky, A. R. Chem Rev 1996, 96,
By multiple linear regression with Monte-Carlo simulated anneal- 1027.
ing, the best model (q2 ¼ 0.881, r2 ¼ 0.901, n ¼ 35, s ¼ 0.401, 22. Wang, Z. Y.; Zhai, Z. C.; Wang, L. S. QSAR Combinatorial Sci
2005, 24, 211.
F ¼ 68.44) was established. These QSAR results identified the
23. Han, X. Y.; Wang, Z. Y.; Yang, C. S. Chin J Struct Chem 2005, 24,
sensitivity of AHAS to different substitutions in an inhibitor, 145.
improved our knowledge about the interaction mechanism of 24. Wan, J.; Zhang, L.; Yang, G. F.; Zhan, C. G. J Chem Inf Comput Sci
AHAS and sulfonylurea inhibitors, and provided basis for design- 2004, 44, 2099.
ing new herbicides to combat the herbicides resistance. 25. Wan, J.; Zhang, L.; Yang, G. F. J Comput Chem 2004, 25, 1827.
On the other hand, a general quantum-chemical descriptor for 26. Zhang, L.; Wan, J.; Yang, G. F. Bioorg Med Chem 2004, 12, 6183.
steric effect was developed by defining the volume of electron 27. Kohn, W.; Sham, L. J. Phys Rev A 1965, 140, 1133.
cloud for a specific substituent within a series of analogues and 28. Tripos Associates. SYBYL6.9. Tripos Associates, St. Louis, MO,
applied successfully in the QSAR analysis of sulfonylurea ana- 2002.
logues. On combining DFT calculation with QSAR analysis, this 29. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Montgomery, J. A.; Vreven, T., Jr.; Kudin,
descriptor provided an effective way for solving the problem how
K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone,
to directly describe steric effect in quantum chemistry-based V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.;
QSAR studies in general. This kind of DFT/QSAR technique Nakatsuji, H.; Hada, M. E., M.; Toyota, K.; Fukuda, R.; Hasegawa, J.;
might be developed as a general, less arbitrary approach for the Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.;
study of structure-activity relationships with minimum empirical Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo,
parameters, which may become an effective approach for opti- J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi,
mizing lead compounds and designing new drugs. R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth,
G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.;
Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
References Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.;
1. Defelice, M.; Guardiola, J.; Esposito, B.; Iaccarino, M. J Bacteriol Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-
1974, 120, 1068. Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill,
2. Guardiola, J.; Defelice, M.; Lamberti, A.; Iaccarino, M. Mol Gen P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople,
Genet 1977, 156, 17. J. A. Gaussian 03, Rev. C01; Gaussian, Inc., Wallingford, CT, 2004.
3. Duggleby, R. G.; Pang, S. S. J Biochem Mol Biol 2000, 33, 1. 30. Xi, Z.; Niu, C. W.; Ban, S. R.; Li, Q. X.; Ouyang, D.; Huang, M. Z.
4. Ray, T. B. In Target Sites of Herbicide Action; Boger, P.; Sandmann, Chin J Pestic Sci 2005, 7, 311.
G., Eds.; CRC: Boca Raton, 1989; pp. 105–125. 31. Xi, Z.; Niu, C. W.; Li, Q. X.; Ouyang, D.; Ban, S. R. Chin J Pestic
5. Schloss, J. V. In Target Sites of Herbicide Action; Boger, P.; Sci 2005, 7, 215.
Sandmann, G., Eds.; CRC: Boca Raton, 1989; pp. 165–245. 32. Lee, C.; Yang, W.; Parr, R. G. Phys Rev B 1988, 37, 785.
6. Ott, K. H.; Kwagh, J. G.; Stockton, G. W.; Sidorov, V.; Kakefuda, G. 33. Becke, A. D. J Chem Phys 1993, 98, 5648.
J Mol Biol 1996, 263, 359. 34. Bader, R. F. W.; Henneker, W. H.; Cade, P. E. J Chem Phys 1967,
7. Duggleby, R. G.; Pang, S. S.; Yu, H. Q.; Guddat, L. W. Eur J Bio- 46, 3341.
chem 2003, 270, 2895. 35. Bader, R. F. W.; Carroll, M. T.; Cheeseman, J. R.; Chang, C. J Am
8. Lee, Y. T.; Chang, A. K.; Duggleby, R. G. FEBS Lett 1999, 452, 341. Chem Soc 1987, 109, 7968.

Journal of Computational Chemistry DOI 10.1002/jcc