6480
G.-W. Rao et al. / Bioorg. Med. Chem. Lett. 23 (2013) 6474–6480
17. Lv, L. P. Zhejiang University of Technology, 2006.
Data Centre: CCDC-255700 contains the supplementary crystallographic data
for this Letter.
21. Synthesis of compound 1g: 3,6-Bis(4-trifluoromethylphenyl)-1,4-dihydro-
1,2,4,5-tetrazine (3b, 5 mmol), prepared according to the procedure of Rao
and Hu16 from 4-(trifluoromethyl)benzonitrile and hydrazine hydrate, was
dissolved in dichloromethane (20 mL) with stirring under nitrogen protection
in an ice bath. Pyridine (0.5 mL) was added to the mixture. Methyl
chloroformate (5 mmol) and dichloromethane (20 mL) were added dropwise
into the mixture under 0–5 °C. The mixture was stirred at room temperature
for 24 h, then washed in water and dried with anhydrous MgSO4. The solvent
was removed in vacuo and the residue was recrystallized from 95% ethanol to
24. CoMFA studies were performed with SYBYL 6.9.1 molecular modeling
software.25 Steric and electrostatic interactions were calculated using a sp3
carbon probe atom with a charge of +1 with a distance-dependent dielectric at
each lattice point, and energy cut-off of 30 kcal molꢀ1. Each molecule was
calculated on a 3D cubic lattice with grid spacing of 2 Å in x, y, and z directions.
The CoMFA-STD method in SYBYL was used to scale CoMFA fields. Similarity
indices were derived from the same lattice box, which were utilized in CoMFA
calculations. Steric, electrostatic, hydrophobic, hydrogen bond donor and
hydrogen bond acceptor descriptors were evaluated using the probe atom.
GAUSSIAN-type distance dependence was used to measure the relative
attenuation of the field position of each atom in the lattice and a default
value of 0.3 was used as the attenuation factor.
give the product (1g) as
a yellow solid (yield 53.7%). A solution of the
compound in ethanol was concentrated gradually at room temperature to
afford yellow prisms which are suitable for X-ray diffraction, mp: 198–199 °C;
1H NMR (400 MHz, CDCl3): d 3.83 (s, 3H), 7.67–7.73 (m, 6H), 7.89–7.91 (m, 2H),
8.07 (s, 1H); IR mmax (KBr)/cmꢀ1: 3312, 3085, 2972, 1689, 1618, 1573, 1522,
1475, 1448, 1378, 1327, 851; MS (EI, 70 eV): m/z (%): 430 (22) [M]+, 386 (12),
186 (14), 172 (100), 145 (44), 95 (13), 75 (12), 59(34); Anal. Calcd (%) for
25. SYBYL 6.9.1, Tripos Inc.: St. Louis, MO, 2003.
26. The IC50 (concentration causing 50% inhibitory effect on the A549 proliferation)
values were converted to pIC50 (ꢀlog IC50) values and used as dependent
variables in the CoMFA and CoMSIA QSAR analysis. For 3D-QSAR analyses, 29
compounds (78.4%) were selected as the training set for model construction,
and the remaining 8 compounds (21.6%) as the test set for model validation.
C
18H12N4O2F6: C, 50.24; H, 2.81; N, 13.02. Found: C, 50.28; H, 2.66; N, 13.16.
Compounds of 1a–j were synthesized in the same manner. Synthesis of
compound 2g: 3,6-Bis(4-trifluoromethylphenyl)-1,4-dihydro-1,2,4,5-tetrazine
(3b, 5 mmol), prepared according to the procedure of Rao and Hu16 from 4-
(trifluoromethyl) benzonitrile and hydrazine hydrate, was dissolved in
dichloromethane (20 mL) with stirring under nitrogen protection in an ice
bath. Pyridine (1 mL) was added to the mixture. Propyl chloroformate
(10 mmol) and dichloromethane (20 mL) were added dropwise into the
mixture under 0–5 °C. The mixture was stirred at room temperature for 16 h,
then washed in water and dried with anhydrous MgSO4. The solvent was
removed in vacuo and the residue was recrystallized from ethanol/
dichloromethane (1:1) to give the product (2g) as a colourless solid (yield
52.7%). A solution of the compound in ethanol was concentrated gradually at
room temperature to afford colourless prisms which are suitable for X-ray
diffraction, mp: 147–149 °C; 1H NMR (400 MHz, CDCl3): d 0.64 (t, 6H), 1.46–
1.48 (m, 4H), 4.04–4.19 (m, 4H), 7.77 (d, J = 8.2 Hz, 4H), 8.18 (d, J = 8.1 Hz, 4H);
IR mmax (KBr)/cmꢀ1: 2975, 2886, 1759, 1618, 1545, 1501, 1467, 1321, 1263,
1215, 854; MS (EI, 70 eV): m/z (%): 545 (13) [M+H]+, 546 (3), 459 (4), 414 (2),
172 (3), 44 (4), 76 (18), 43 (100), 41 (51); Anal. Calcd (%) for C24H22N4O4F6: C,
52.95; H, 4.07; N, 10.29. Found: C, 53.02; H, 3.99; N, 10.31. Compounds of 2a–i
were synthesized in the same manner. Data for other compounds are provided
in the Supplementary data.
The activities of the training set range from 0.035
85.806 M (pIC50 = 4.066). The activities of the test set range from 0.575
(pIC50 = 6.240) to 88.988 M (pIC50 = 4.051). The fact worth mentioning is that
lM (pIC50 = 7.456) to
l
lM
l
the structural diversity and activity range of the test set are comparable with
the training set.27–29 In the development of 3D-QSAR models, the molecular
alignment and conformation selection are the most essential steps.
Conformations of each compound were generated using Confort™
conformation analysis. Energy minimizations were performed using Tripos
force field30 with
gradient algorithm with
a
distance-dependent dielectric and Powell conjugate
convergence criterion of 0.005 kcal/(mol Å).
a
Gasteiger–Hückel31 charges were assigned to all molecules. Since specific
molecular target is unknown to these compounds, the most active compound
25 was used as a template for superimposition, assuming that its conformation
represents the most bioactive conformation of the tetrazine derivatives. All
compounds were aligned using a tetrazine nucleus as common substructure in
all molecules and minimum scaffold required for active molecules. Figure 3
shows the alignment of all compounds in the training set.
32. Validation was carried out considering the q2 coefficient and predicting the
activity of an external test set with 8 compounds. According to Golbraikh and
Tropsha,28,29 models are considered highly predictive if they satisfy all of the
following conditions (I–IV):
22. Crystal data of compound 1g:
A
yellow prism of dimensions
0.35 ꢁ 0.25 ꢁ 0.25 mm3 was used for data collection with an Enraf-Nonius
CAD-4 diffractometer with graphite monochromated MoK
a radiation
(k = 0.71073 Å). The structure was solved by direct method procedures as
implemented in the SHELXS9723 program. The positions of all the non-hydrogen
atoms were included in the full-matrix least-squares refinement using the
SHELXS9723 program. Hydrogen atoms were added at calculated positions and
I. q2 >0.5
refined using
a
riding model; they were given isotropic displacement
II. R2 >0.6
parameters equal to 1.2 (or 1.5 for methyl H atoms) times the equivalent
isotropic displacement parameters of their parent atoms, and C–H distances
were restrained to 0.96 Å for methyl H atoms and 0.93 Å for phenyl H atoms,
while N–H distances were set to 0.86 Å. C18H12F6N4O2, Mr = 430.32, monoclinic,
a = 9.177(3), b = 12.159(2), c = 16.896(2) Å, b = 104.540(16)°, U = 1824.9(6) Å3,
III. ½ððR2 ꢀ R20Þ=R2Þꢂ <0.1 or ½ððR2 ꢀ R002Þ=R2Þꢂ <0.1
IV. 0.85 6 k 6 1.15 or 0.85 6 k0 6 1.15
where, q2 is the cross-validated correlation coefficient from LOO; R2 is the
~
correlation coefficient for experimental (y) versus predicted ðyÞ activities for
test set molecules; R20 and R002 are the correlation coefficients of the regression
calcd = 1.566 g cmꢀ3
measured, 3269 unique
,
through the origin for y versus ðyÞ and ðyÞ versus y, respectively; k and k are
0
~
~
T = 298(2) K,
space
group:
3839
P21/c,
reflections
Z = 4,
q
) = 0.146 mmꢀ1
,
the slopes for regression through origin yr0 ¼ ky and
¼ k0y and were
r0
~
~
y
l(MoK
a
(Rint = 0.0194) which were used in all calculations. Fine R1 = 0.0307, wR
(F2) = 0.0939 (all data). All crystallographic details for compound 1g have
been deposited with the Cambridge Crystallographic Data Centre: CCDC-
255697 contains the supplementary crystallographic data for this Letter.
calculated as follows:
P
P
~
yi yi
P
k ¼
k0 ¼
2
~
yi
P
~
yi yi
yi2
Crystal data of compound 2g:
A
colorless prism of dimensions
0.35 ꢁ 0.25 ꢁ 0.25 mm3 was used for data collection with an Enraf-Nonius
CAD-4 diffractometer with graphite monochromated MoK
a
radiation
33. PLS,34,35 the statistical method used in deriving the 3D-QSAR models,
implemented in SYBYL 6.9.1 was used to generate a linear regression that
correlates descriptors with biological activities in pIC50, cross validation being
used to obtain the optimum number of the principal components. The cross-
validation analysis was performed utilizing the leave-one-out (LOO) method in
which one compound was removed from the data set and its activity was
predicted using the model built from the rest of the data set. The cross-
validated coefficient q2 was evaluated as:
(k = 0.71073 Å). The structure was solved by direct method procedures as
implemented in the SHELXS9723 program. The positions of all the non-hydrogen
atoms were included in the full-matrix least-squares refinement using the
SHELXS9723 program. Hydrogen atoms were added at calculated positions and
refined using
a riding model; they were given isotropic displacement
parameters equal to 1.2 (or 1.5 for methyl H atoms) times the equivalent
isotropic displacement parameters of their parent atoms, and C–H distances
were restrained to 0.96 Å for methyl H atoms, 0.97 Å for methylene H atoms,
P
2
ðYpredꢀYexp
Þ
q2 ¼ 1 ꢀ
P
and 0.93 Å for phenyl
H
atoms.
C
24H22F6N4O4, Mr = 544.46, triclinic,
= 107.860(14)°, b = 100.640(19)°,
ðYexp ꢀYmean
Þ
a = 8.210(2), b = 12.151(2), c = 14.469(3) Å,
a
ꢀ
c
q
= 99.570(19)°, U = 1311.4(5) Å3, T = 298(2) K, space group: P1, Z = 2,
where Ypred, Yexp and Ymean are the predicted, experimental and mean values of
the pIC50, respectively.
calcd = 1.379 g cmꢀ3 ) = 0.123 mmꢀ1
(MoK 5777 reflections measured,
,
l
a
,
4707 unique (Rint = 0.0140) which were used in all calculations. Fine
R1 = 0.0530, wR (F2) = 0.2036 (all data). All crystallographic details for
compound 2g have been deposited with the Cambridge Crystallographic
Rao, Guo-Wu
