close

Вход

Забыли?

вход по аккаунту

?

NMR QSAR Model for the Analysis of 4-5-Arylamino-134-thiadiazol-2-ylbenzene-13-diols.

код для вставкиСкачать
340
Arch. Pharm. Chem. Life Sci. 2011, 11, 340–344
Full Paper
NMR QSAR Model for the Analysis of 4-(5-Arylamino-1,3,4thiadiazol-2-yl)benzene-1,3-diols
Joanna Matysiak1, Andrzej Niewiadomy1, Beata Paw2, and Izabela Dybała3
1
Department of Chemistry, University of Life Sciences, Maria Curie-Skłodowska University, Lublin, Poland
Department of Medicinal Chemistry, Medical University, Maria Curie-Skłodowska University, Lublin, Poland
3
Department of Crystallography, Maria Curie-Skłodowska University, Lublin, Poland
2
We have developed a NMR data quantitative structure-activity relationship NMR-QSAR model based on
1
H- and 13C-NMR experimental spectral data of 4-(5-arylamino-1,3,4-thiadiazol-2-yl)benzene-1,3-diols.
Compounds show in-vitro antiproliferative activity against some human cancer cell lines. Twoparameter equations obtained by the multiple linear regression procedure showed that chemical
shifts of the protons of hydroxyl groups and carbon atoms of the 1,3,4-thiadiazole ring are the decisive
descriptors of inhibition interactions of the compounds. The models gave leave-one-out (LOO) crossvalidation ranges from 78% to 93%. The obtained NMR-QSAR equations provide a rapid, reliable, and
simple way for predicting the antiproliferative activity of N-substituted 4-(5-amino-1,3,4-thiadiazol-2yl)benzene-1,3-diols.
Keywords: NMR / QSAR / 1,3,4-Thiadiazoles
Received: January 23, 2010; Accepted: April 16, 2010
DOI 10.1002/ardp.201000029
1 Introduction
SAR (structure-activity relationship) methods are based on
the assumption that there is a relationship between the
structure and the activity of compounds, especially in the
case of a series of closely related compounds. The QSAR
(quantitative structure-activity relationship) modeling
results show that receptor binding of a compound can be
predicted from combination of electrostatic potential and
geometrical structural analysis [1]. In these studies, various
structural and physicochemical descriptors obtained both in
experimental and in-silico ways are applied [2–10].
The NMR spectra of compounds, used to predict and refine
their structures, include frequencies corresponding directly
to the quantum mechanical properties of every nuclear magnetic movement in a chemical structure. In general, these
NMR chemical shifts depend greatly on the electrostatic
Correspondence: Joanna Matysiak, Department of Chemistry, University
of Life Sciences in Lublin, Akademicka 15, 20-950 Lublin, Poland.
E-mail: joanna.matysiak@up.lublin.pl
Fax: þ48 81 533-3549
Abbreviations: combinatorial protocols in the multiple linear regression
procedure (CP-MLR); quantitative structure-activity relationship (QSAR);
structure-activity relationship (SAR).
ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
potential energy of the nucleus and the types of atomic
orbitals surrounding the nucleus (i. e. its hybridization state)
together with the disturbing effects of surrounding atoms
[11]. Thus, the NMR chemical shifts are closely related to the
substituent effect, the basis of all SAR methods. The relationship between the 13C-NMR chemical shifts of a molecule and
its 3D conformations is well-known and almost routinely
employed.
A set of SAR methods have recently been reported in which
parameters of NMR spectra were applied. These techniques
were used for discovering high-affinity ligands that bind to
the FK506 binding protein (FKBP) and nonpeptide inhibitors
of stromelysin [12, 13]. Beger and co-workers demonstrated
that 13C-NMR data can be used to produce reliable quantitative spectrometric data-activity relationship (QSDAR) models
of binding to the corticosterone binding globulin, aromatase
enzyme, and aryl hydrocarbon receptor [14–17]. Comparative
structural connectivity spectra analysis (CoSCoSA) models of
steroids binding to the corticosteroid were also presented
[18]. The success of spectroscopic methods in SAR can be
explained recalling that all spectroscopic quantities reflect
certain intrinsic physicochemical properties of molecules
that are related to their 3D structure.
The aim of the present work is to investigate the relationships between the chemical shift of magnetic nuclei obtained
Arch. Pharm. Chem. Life Sci. 2011, 11, 340–344
NMR QSAR Model for Predicting Antiproliferative Activity
from the 1H- and 13C-NMR spectrum and the activity of 4-(5arylamino-1,3,4-thiadiazol-2-yl)benzene-1,3-diols against the
cells of human cancer cell lines. The multiple linear
regression procedure was used for this purpose.
Table 2. Antiproliferative effect (log(1/ID50)) observed and
predicted for 4-(5-arylamino-1,3,4-thiadiazol-2-yl)benzene-1,3diols.
No.
2 Results and discussion
For the compounds of the structure presented in Table 1, the
in-vitro antiproliferative activity against some human cancer
cell lines was reported as ID50 values (the concentration of the
compound that inhibits the proliferation rate of the tumour
cells by 50% as compared to the untreated control cells) [19,
20]. For the present QSAR studies, ID50 values were converted
to the logarithm of 1/ID50 in micromolar units (Table 2). As
the independent variable, standardized chemical shifts d of
the protons and the isotope of carbon 13C of 1H- and 13C-NMR
experimental spectral data were used (Table 1). The NMRQSAR models were built using the combinatorial protocols in
the multiple linear regression procedure (CP-MLR) [21]. The
correlation matrix for applying descriptors is presented in
Table 3.
At first, the one-parameter analysis using chemical shifts of
various protons and 13C isotope to establish the effect of
single quantities on the activity was carried out (Table 4).
Table 4 shows the significant influence of chemical shift
values of the carbon atom C-5 of the 1,3,4-thiadiazole ring
and/or proton of the amine group as well as of the proton
of the hydroxyl group of the benzenodiol moiety. As the
descriptors dNH and dC5(TDA) are highly intercorrelated
(r2 ¼ 0.647, Table 3), they were not together included in
341
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
§
T47D
A549
SW707
Obsd.
Calcd.
(Eq. (1))
Obsd.
Calcd.
(Eq. (2))
Obsd.
Calcd.
(Eq. (3))
S1.127
S1.303
S1.301
S1.731
S1.194
S1.141
S1.240
S1.086
S1.109
S1.006
S1.367
S0.987
S1.126
S1.122
S1.300
S1.372
S1.708
S1.180
S1.131
S1.142
S1.048
S1.068
S1.045
S1.364
S1.067
S1.170
S1.494
S1.807
S1.993
S2.030
S1.590
<S2.519
S1.893
S1.877
S1.334
S1.438
S1.737
S1.230
S1.400
S1.546
S1.923
S1.986
S1.980
S1.423
–§
S1.626
–§
S1.383
S1.469
S1.732
S1.290
S1.587
S1.179
S1.730
S1.986
S1.957
S1.269
S1.077
S1.569
S1.065
S1.232
S1.704
S1.833
S1.290
S1.533
S1.200
S1.828
S1.771
S1.969
S1.317
S1.322
S1.637
–§
S1.177
S1.527
S1.800
S1.153
S1.656
Descriptors were not obtained.
the equations where two-parameter models were built.
This indicates that for all practical purposes, the information
content of these parameters is almost the same. The predictor
variables with p > 0.05 were eliminated whilst deriving the
QSAR models in order to assure their statistical reliability.
Using the CP-MLR procedure, the best two-parameter
models describing antiproliferative activity of the compound
against individual cell lines were chosen, and normal as well
as cross-validation statistics were calculated.
Table 1. Structure and NMR descriptors of 4-(5-arylamino-1,3,4-thiadiazol-2-yl)benzene-1,3-diols.
S
HO
OH
NH R
N N
No.
substituent
dNH
d3OH
d1OH
dC5(TDA)
dC2(TDA)
1
2
3
4
5
6
7
8
9
10
11
12
13
C6H52-CH3-C6H42,4-di-CH3-C6H32,6-di-CH3-C6H34-C2H5-C6H44-F-C6H42-Cl-C6H43-Cl-C6H44-Cl-C6H42,5-di-Cl-C6H32,6-di-Cl-C6H34-I-C6H42-CH3-(5-Cl-)C6H3-
9.94
9.33
9.52
9.50
9.90
9.92
9.65
–§
9.89
9.84
–#
9.9
9.58
10.91
10.84
10.90
10.53
10.88
10.85
10.87
–§
10.85
10.88
10.73
10.80
10.82
10.39
9.86
9.92
10.04
10.17
10.23
9.88
–§
10.37
9.91
9.88
10.39
9.88
163.51
165.48
165.37
167.61
163.19
163.57
164.08
163.10
163.21
163.38
165.37
163.14
164.23
154.79
156.30
154.87
154.39
153.13
154.78
155.74
155.09
155.01
155.77
155.00
154.86
155.63
§
One wide band was registered;# band was not registered.
ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.archpharm.com
342
J. Matysiak et al.
Arch. Pharm. Chem. Life Sci. 2011, 11, 340–344
Table 3. Correlation coefficient (r2) matrix of the descriptors used in
the QSAR equations.
dNH
dC5(TDA)
dC2(TDA)
d3OH
d1OH
dNH
dC5(TDA)
dC2(TDA)
d3OH
d1OH
1.000
0.647
1.000
0.215
0.005
1.000
0.008
0.596
0.030
1.000
0.406
0.263
0.236
0.002
1.000
Table 4. Correlation coefficients (r) between the activity and the
descriptors used in the QSAR analysis.
dNH
T47D
A549
SW704
dC5(TDA)
0.775
0.668
0.621
–0.936
–0.673
–0.812
d3OH
d1OH
0.808
0.532
0.448
0.369
0.586
0.778
In the case of the T47D cell line, the following equation was
obtained:
log1=ID50 ¼0:185ð0:015ÞdC5ðTDAÞ
þ 0:047ð0:015ÞdC2ðTDAÞ 1:209ð0:015Þ
(1)
n ¼ 13, r ¼ 0.968, r2 ¼ 0.937, s ¼ 0.053, F(2,10) ¼ 74.02,
r-bar ¼ 0.970, SPRESS ¼ 0.195, SDEP ¼ 0.171, R2CV ¼ 0.929
The above model suggests that dC5(TDA) and dC2(TDA)
parameters play a significant role in explaining the variance
in the activity of compounds against T47D cells. The equation
could estimate 93.7% variance in the observed activity. There
is evidence that one descriptor contributes negatively
whereas the other one positively to the activity. Thus, lower
values of ID50 can be obtained for compounds of high dC2(TDA)
and low dC5(TDA) values. The model gives a leave-one-out crossvalidation of 92.9%.
For the A549 cell line, the best results were obtained for the
following dependence:
log1=ID50 ¼0:331ð0:066ÞdC5ðTDAÞ
0:156ð0:066Þd3OH 1:595ð0:043Þ
(2)
n ¼ 11, r ¼ 0.884, r2 ¼ 0.789, s ¼ 0.122, F(2,8) ¼ 14.98,
r-bar ¼ 0.858, SPRESS ¼ 0.483, SDEP ¼ 0.412, R2CV ¼ 0.780
Compounds 6 (lack of ID50 value) and 8 (d3OH was not
determined) were not included in the model. In this case,
dC5(TDA) and d3OH are the best parameters to estimate the
variance in the observed activity (78.9%). It is evident from
Eq. (2) that both descriptors contribute negatively to the
activity. Thus, lower values of dC5(TDA) and d3OH would be
beneficiary to augment the activity of the titled compounds
against non-small lung carcinoma cells (A549).
ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
In case of the SW707 cell line, the following equation was
obtained:
log1=ID50 ¼ 0:171ð0:051ÞdC5ðTDAÞ
þ 0:183ð0:060Þd1OH 1:546ð0:044Þ
(3)
n ¼ 12, r ¼ 0.908, r2 ¼ 0.824, s ¼ 0.146, F(2,9) ¼ 21.13,
r-bar ¼ 0.930, SPRESS ¼ 0.513, SDEP ¼ 0.444, R2CV ¼ 0.814
Compound 8 was not included in the model (d1OH parameter
was not determined). The above equation shows that the
parameters dC5(TDA) and d1OH play a significant role in explaining the variance in the activity against SW707 cells. Different
signs of the regression coefficients suggest that the directions
of influence of explanatory variables in the model are unlike. It
explains 82.4% variance in the inhibitory activity. The R2CV is in
accordance to the reasonable robust QSAR model.
The best model was obtained against T47D cells whereas
the poorest one was built against A549. The chemical shift of
the carbon atom in position 5 of the 1,3,4-thaidiazole ring
(dC5(TDA)) is included in all equations. The negative contribution of this descriptor suggests an improvement of inhibition in the case of compounds of low d C-5 value. At the
same time, as follows from Eq. (3) and Table 1, higher activities against the SW707 cell line are exhibited by the
compounds of a higher acidity of the C(1)-OH proton. The
log 1/ID50 values calculated according to Eqs. (1)–(3) are
presented in Table 2. The residues between the observed
and calculated activities are presented in Fig. 1. The highest
differences are observed for compounds 7 and 13 against
A549 cell line and for 3 and 6 against SW707.
The obtained results indicate that high electron density on
the nitrogen atom of the amine group and on the oxygen
atoms of the hydroxyl groups is favorable for the antiproliferative activity of N-substituted 4-(5-amino-1,3,4-thiadiazol-2yl)benzene-1,3-diol set against T47D and SW707 human cancer cell lines. Accordingly, N-alkyl derivatives should have
lower antiproliferative activity in relation to aryl derivatives
due to the positive electron-induction effect of the alkyl
group that decreases electron density on the nitrogen atom.
That kind of derivatives was not included in the analysis but
previously reported data confirm this observation [19]. The
earlier QSAR studies of a larger group of this type of compounds with electronic descriptors obtained by molecular
modelling using the Hartree–Fock method at 6-31G level
also showed an important role of the atomic net charge of the
amine-nitrogen atom and carbon atoms of the 1,3,4-thiadiazole ring [22] but in this case, better cross-validated correlations were obtained.
The presented models confirm a key role of heteroatoms in
the interaction with the molecular target. The compounds
under investigation similarly to other resorcylazoles are
www.archpharm.com
Arch. Pharm. Chem. Life Sci. 2011, 11, 340–344
NMR QSAR Model for Predicting Antiproliferative Activity
T47D
formation [23]. The QSAR analysis also gives insight into some
important common structural features. Their explanation
may help in the development of improved anticancer agents
against specific human cancer cell lines.
0.20
0.15
0.10
12
3
13
residues
0.05
11
0.00
2
5
4
3 Experimental
10
61
3.1 Chemistry
98
-0.05
7
-0.10
-0.15
-0.20
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
-1.2
-1.1
-1
-0.9
The compounds were tested in vitro against the following
human cancer cell lines: T47D (breast cancer), SW707 (rectal
adenocarcinoma), A549 (non-small lung carcinoma) from the
American Type Culture Collection (Rockville, MD, USA) as
described previously [19, 20]. The SRB test measuring the cell
proliferation inhibition in in-vitro culture was applied [24].
13
0.2
2
0.1
residues
1
10
12
9
11
3
4
3.3 QSAR analysis
-0.1
5
-0.2
7
-0.3
-2.1
-2
-1.9
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
-1.2
-1.1
SW707
0.3
6
0.2
13
2
residues
0.1
7
5
1
4
0.0
11
9
-0.1
12
10
-0.2
-0.3
-2.2
The studied compounds were prepared following the synthetic procedures previously reported. 1H- and 13C-NMR
spectra were recorded in DMSO-d6 or CDCl3 using a Varian
Mercury 400 or Bruker DRX 500 instrument (Varian and
Bruker, both USA). Chemical shifts (d, ppm) are given in
relation to tetramethylsilane (TMS) [19, 20].
3.2 Biology
A549
0.3
0.0
343
3
-2
-1.8
-1.6
-1.4
-1.2
-1
Combinatorial Protocols in Multiple Linear Regression (CPMLR) for the model building was used [25, 26]. It involves a
combinatorial strategy with appropriately placed ‘filters’
interfaced with MLR. The filters set the thresholds for the
descriptors in terms of interparameter correlation cutoff
limits in subset regressions (filter-1); t-values of the regression
coefficients (filter-2); internal explanatory power, square-root
of adjusted multiple correlation coefficient of regression
equation, r-bar (filter-3); and the external consistency,
R2CV (filter-4). Throughout this study, for filters-1, 2, 3, and
4 the thresholds were assigned as 0.79, 2.0, 0.71, and
0.3 R2CV 1.0, respectively. Calculations were made using
Statistica Program, Version 7.1 [27]. Statistical significance of
the regression equation was tested by the correlation coefficient
(r), the standard error of estimate (s), the variance ratio (F) at
specified degrees of freedom (df), and r-bar (n – number of
compounds). The quality of the resulting models was assessed
using leave-one-out cross-validation and quantified using the
cross-validated correlation coefficient from the leave-one-out
procedure (R2CV ) as well as SPRESS and SDEP statistics.
log 1/ID50 observed
Figure 1. Residues between the observed and calculated activity
versus the observed activity.
probably inhibitors of heat shock proteins HSP90. In the
interactions of ligands bound to the ATP binding site of
Hsp90, an essential role is played by oxygen atoms of the
hydroxyl group as a proton acceptor during hydrogen-bonds
ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The authors have declared no conflict of interest.
References
[1] R. D. Cramer, D. E. Paterson, J. D. Bunce, J. Am. Chem Soc. 1988,
110, 5959–5967.
[2] R. Garg, W. A. Denny, C. Hansch, Bioorg. Med. Chem. 2000, 8,
1835–1839.
www.archpharm.com
344
J. Matysiak et al.
[3] C. Hansch, R. Garg, A. Kurup, S. B. Mekapati, Bioorg. Med.
Chem. 2003, 11, 2075–2084.
[4] R. P. Verma, C. Hansch, Bioorg. Med. Chem. 2005, 13, 4597–
4602.
[5] S. B. Mekapati, C. Hansch, Bioorg. Med. Chem. 2001, 9, 3225–3230.
[6] R. Garg, A. Kurup, C. Hansch, Bioorg. Med. Chem. 2001, 9,
3161–3164.
[7] A. Asikainen, J. Ruuskanen, K. Tuppurainen, J. Chem. Inf.
Comput. Sci. 2003, 43, 1974–1981.
[8] R. Maltais, D. Fournier, D. Poirier, QSAR Comb. Sci. 2009, 28,
1284–1299.
[9] B. K. Sharma, P. Pilania, P. Singh, S. Sharma, Y. S. Prabhakar,
Centr. Eur. J. Chem. 2009, 7, 909–922.
[10] S. Sharma, Y. S. Prabhakar, P. Singh, B. K. Sharma, Eur. J. Med.
Chem. 2008, 43, 2354–2360.
[11] A. C. de Dios, J. G. Pearson, E. Oldfield, Science 1993, 260,
1491–1496.
[12] P. J. Hajduk, G. Sheppard, D. G. Nettesheim, E. T. Olejniczak,
et al., J. Am. Chem. Soc. 1997, 119, 5818–5827.
[13] S. B. Shuker, P. J. Hajduk, R. P. Meadows, S. W. Fesik, Science
1996, 274, 1531–1534.
[14] R. D. Beger, J. P. Freeman, J. O. Lay, Jr, J. G. Wilkes, D. W.
Miller, J. Chem. Inf. Comput. Sci. 2000, 40, 1449–1455.
ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Arch. Pharm. Chem. Life Sci. 2011, 11, 340–344
[15] R. D. Beger, J. P. Freeman, J. O. Lay, Jr, J. G. Wilkes, D. W.
Miller, J. Chem. Inf. Comput. Sci. 2001, 41, 219–224.
[16] R. D. Beger, J. G. Wilkes, J. Chem. Inf. Comput. Sci. 2001, 41,
1322–1329.
[17] R. D. Beger, D. A. Buzatu, J. G. Wilkes, J. O. Lay, Jr. J. Chem. Inf.
Comput. Sci. 2001, 41, 1360–1366.
[18] R. D. Beger, D. A. Buzatu, J. G. Wilkes, J. O. Lay, Jr. J. Chem. Inf.
Comput. Sci. 2002, 42, 1123–1131.
[19] J. Matysiak, A. Opolski, Bioorg. Med. Chem. 2006, 14, 4483–
4489.
[20] J. Matysiak, Chem. Pharm. Bull. 2006, 54, 988–991.
[21] Y. S. Prabhakar, QSAR Comb. Sci. 2003, 22, 583–595.
[22] J. Matysiak, QSAR Comb. Sci. 2008, 27, 607–617.
[23] P. A. Brough, X. Barril, M. Beswick, B. W. Dymock, et al.,
Bioorg. Med. Chem. Lett. 2005, 15, 5197–5201.
[24] P. Skehan, R. Storeng, D. Scudiero, A. Monks, et al., J. Natl.
Cancer Inst. 1990, 82, 1107–1112.
[25] M. K. Gupta, Y. S. Prabhakar, J. Chem. Inf. Model. 2006, 46,
93–102.
[26] M. Saquib, M. K. Gupta, R. Sagar, Y. S. Prabhakar, et al., J. Med.
Chem. 2007, 50, 2942–2950.
[27] Statistica 7.1, StatSoft Inc., 2005.
www.archpharm.com
Документ
Категория
Без категории
Просмотров
0
Размер файла
145 Кб
Теги
arylamine, 134, mode, nmr, qsar, diols, thiadiazole, ylbenzene, analysis
1/--страниц
Пожаловаться на содержимое документа