Pharmacophore modeling, 3D-QSAR, synthesis, and anti-lung cancer evaluation of novel thieno[2,3-d][1,2,3]triazines targeting EGFR

Article abstract of DOI:10.1002/ardp.201900108

Two series of thieno[2,3-d][1,2,3]triazine derivatives were designed, synthesized, and biologically evaluated as potential epidermal growth factor receptor (EGFR) inhibitors targeting the non-small-cell lung cancer cell line H1299. Most of the synthesized compounds displayed IC₅₀ values ranging from 25 to 58 nM against H1299, which are superior to that of gefitinib (40 μM). 3-(5,6,7,8-Tetrahydro-7H-cyclohexa[4:5]thieno[2,3-d]-1,2,3-triazin-4-ylamino)benzene-1,3-diamine (6b) achieved the highest cytotoxic activity against H1299 with an IC₅₀ value of 25 nM; it had the ability to decrease the EGFR concentration in H1299 cells from 7.22 to 2.67 pg/ml. In vitro, the IC₅₀ value of compound 6b was 0.33 nM against EGFR, which is superior to that of gefitinib at 1.9 nM and erlotinib at 4 nM. The three-dimensional quantitative structure–activity relationships and molecular modeling studies revealed comparable binding modes of compound 6b, gefitinib, and erlotinib in the EGFR active site. The in silico ADME (absorption, distribution, metabolism, and excretion) prediction parameters of this compound revealed promising pharmacokinetic and physicochemical properties. Moreover, DFT (density functional theory) calculations showed the high reactivity of compound 6b toward the EGFR compared with other compounds. The designed compound 6b might serve as an encouraging lead compound for the discovery of promising anti-lung cancer agents targeting EGFR.

Full text of DOI:10.1002/ardp.201900108

Received: 12 April 2019
Revised: 23 November 2019
Accepted: 8 December 2019
|
|
DOI: 10.1002/ardp.201900108
F U L L P A P E R
Pharmacophore modeling, 3D‐QSAR, synthesis, and anti‐lung
cancer evaluation of novel thieno[2,3‐d][1,2,3]triazines
targeting EGFR
Ranza Elrayess1
Hosam A. Elshihawy1
|
Yasmine M. Abdel Aziz1
|
Mohamed S. Elgawish2
|
Marwa Elewa1
|
|
Mohamed M. Said1
1Department of Pharmaceutical Organic
Chemistry, Faculty of Pharmacy, Suez Canal
University, Ismailia, Egypt
Abstract
Two series of thieno[2,3‐d][1,2,3]triazine derivatives were designed, synthesized,
and biologically evaluated as potential epidermal growth factor receptor (EGFR)
inhibitors targeting the non‐small‐cell lung cancer cell line H1299. Most of the
synthesized compounds displayed IC50 values ranging from 25 to 58 nM against
H1299, which are superior to that of gefitinib (40 µM). 3‐(5,6,7,8‐Tetrahydro‐7H‐
2Department of Medicinal Chemistry, Faculty
of Pharmacy, Suez Canal University, Ismailia,
Egypt
Correspondence
Mohamed M. Said, Department
of Pharmaceutical Organic Chemistry, Faculty
of Pharmacy, Suez Canal University, Ismailia
41522, Egypt.
cyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐ylamino)benzene‐1,3‐diamine
(6b)
achieved the highest cytotoxic activity against H1299 with an IC50 value of
25 nM; it had the ability to decrease the EGFR concentration in H1299 cells from
7.22 to 2.67 pg/ml. In vitro, the IC50 value of compound 6b was 0.33 nM against
EGFR, which is superior to that of gefitinib at 1.9 nM and erlotinib at 4 nM. The
three‐dimensional quantitative structure–activity relationships and molecular
modeling studies revealed comparable binding modes of compound 6b, gefitinib,
and erlotinib in the EGFR active site. The in silico ADME (absorption,
distribution, metabolism, and excretion) prediction parameters of this compound
revealed promising pharmacokinetic and physicochemical properties. Moreover,
DFT (density functional theory) calculations showed the high reactivity of
compound 6b toward the EGFR compared with other compounds. The designed
compound 6b might serve as an encouraging lead compound for the discovery of
promising anti‐lung cancer agents targeting EGFR.
Email: msaid123eg@yahoo.com
K E Y W O R D S
3D‐QSAR, EGFR, H1299, molecular modeling, thieno[2,3‐d][1,2,3]triazine
with advanced NSCLC.[5,6] Therefore, the strategy of drug targeting can
bring a solution to this serious problem.[7]
|
1
INTRODUCTION
Lung cancer is one of the main causes of death globally due to an ever‐
increasing percentage of smokers worldwide.[1] Lung cancer represents
about one‐third of cancer deaths.[2] Non‐small‐cell lung cancer (NSCLC) is
the major type of lung cancer affecting nearly 80–85% of patients.[3,4]
Despite the availability of a broad range of chemotherapeutic agents,
chemotherapy is still not effective enough in the treatment of patients
It is worth noting that the epidermal growth factor receptor
(EGFR) is overexpressed in NSCLC cell lines of epithelial origin,
especially H1299.[8,9] Thus, lowering the EGFR concentration is a
plausible target for the design and development of cytotoxic
agents targeting H1299 lung cancer cells. EGFR tyrosine kinase
inhibitors (EGFR‐TKIs) have been extensively used in the
|
Arch. Pharm. Chem. Life Sci. 2020;e1900108.
wileyonlinelibrary.com/journal/ardp © 2020 Deutsche Pharmazeutische Gesellschaft
1 of 15
https://doi.org/10.1002/ardp.201900108

2 of 15
ELRAYESS ET AL.
|
treatment of NSCLC patients.[10] Gefitinib (I) and erlotinib
(II) have been launched by US FDA for the treatment of advanced
NSCLC in 2002 and 2004, respectively (Figure 1).[11,12] In spite of
the high response rate of NSCLC patients to these agents,
the development of drug resistance limits the therapeutic
benefits of these two drugs.[13] Therefore, the identification of
the second generation of EGFR inhibitors through the discovery
of new scaffolds could be beneficial for gefitinib‐resisting
displayed a promising potency and selectivity toward EGFR with
an IC50 value of 36 nM (Figure 1).[17]
|
1.1
Rationale and design
The success of our research group in designing new classes of
tricyclic pyridothieno[3,2‐d]pyrimidine and thieno[1,2,3]triazine as
EGFR inhibitors with promising cytotoxic activities triggered us to
develop novel EGFR inhibitors targeting NSCLC.[16,17,19] Moreover, it
was found that the majority of the approved and developed first‐
generation EGFR inhibitors bear the 4‐anilino side chains linked to
quinazoline/quinoline cores. These side chains occupy an important
lipophilic selectivity pocket in the EGFR‐binding domain.[20] In light of
these facts the pharmacophore modeling, three‐dimensional quanti-
tative structure–activity relationship (3D‐QSAR), and molecular
docking of the two series of tricyclic tetrahydrobenzothieno[2,3‐d]
[1,2,3]triazine and dihydrocyclopentathieno‐[2,3‐d][1,2,3]triazine
analogs have been used to provide insights into the key structural
features required for designing EGFR inhibitors targeting lung
carcinoma cell line H1299.[21] The development of this new
generation of EGFR inhibitors was based on modulating the
4‐anilino side chains with mono‐ or bicyclic systems and exploring
different substituents that could recognize the characteristic
lipophilic regions of EGFR. Molecular modeling, 3D‐QSAR, in silico
patients.
A rule of thumb, the modulation of the central
quinazoline core of the first‐generation EGFR inhibitors,
gefitinib and erlotinib, could help in improving activity and
overcome drug resistance.[14] Various heterocyclic rings have
been reported to provide alternatives to the phenyl ring of the
central quinazoline core (III and IV; Figure 1).[15–17] In 2010, cell‐
based screening protocol has led to the identification of several
lead compounds that disturb cell proliferation mediated by
abnormal EGFR concentration, among which the tricyclic
tetrahydrobenzothieno[2,3‐d]pyrimidine hit compound (IV) has
been identified as a prototype with an IC50 value of 2.6 µM
against EGFR (Figure 1).[15] In 2014, Bugge et al.[18] reported a
highly active thienopyrimidine‐based EGFR inhibitor (V) with an
IC50 value of 0.3 nM (Figure 1). In 2015, our research group
identified novel pyridothieno[3,2‐d]pyrimidine and alicyclic thie-
no[1,2,3]triazine scaffolds bearing monocyclic amines as EGFR
inhibitors.[17,19] Fortunately, the pyridothieno[3,2‐d]pyrimidine
FIGURE 1 Examples of epidermal
growth factor receptor inhibitors

ELRAYESS ET AL.
3 of 15
|
ADME (absorption, distribution, metabolism, and excretion) predic-
tions study, Mulliken charge distribution, electrostatic surface
potential (ESP), and HUMO/LUMO analysis have been established
to confirm our rationale.
TABLE 1 In vitro cytotoxic activity of the synthesized compounds
against the NSCLC (non‐small‐cell lung cancer) cell line H1299
Compound
IC50 against H1299
31 nM
4a
4b
56 nM
5a
37 nM
|
2
RESULTS AND DISCUSSION
Chemistry
5b
56 nM
7a
42 nM
|
2.1
4c
28 nM
4d
53 nM
Cyclopentanone via Gewald reaction afforded 2‐amino‐5,6‐dihydro‐4H‐
cyclopenta[b]thiophene‐3‐carbonitrile 2a. Compound 2a was diazotized
to afford the 4‐chlorocyclopenta[4:5]thieno‐[2,3‐d]‐1,2,3‐triazine 3a.[19]
Compound 3a was treated with heteroaryl amines, benzylamine, and
m‐phenylenediamine[22] to give different derivatives (4a, 4b, 5a, 5b, 6a,
and 7a). Cyclohexanone underwent Gewald reaction to afford 2‐amino‐
5,6‐dihydro‐4H‐cyclohexa[b]thiophene‐3‐carbonitrile 2b. Diazotization
of 2b afforded the 4‐chlorocyclohexa[4:5]thieno‐[2,3‐d]‐1,2,3‐triazine
3b.[19] Compound 3b was allowed to react with heteroaryl amines,
benzylamine, and m‐phenylenediamine.[22] The isoxazolo‐ and pyrazolo‐
pyridine derivatives have been prepared from 2‐chloro‐pyridine‐
3‐carbonitrile derivatives, which underwent the following reaction with
hydroxylamine and hydrazine to release the corresponding heteroaryl
amines.[16,23]
5c
34 nM
5d
30 nM
6b
25 nM
7b
30 nM
Gefitinib
40 µM
|
2.2.2
culture
Measurement of human EGFR in H1299 cell
Compound 6b had the ability to decrease EGFR concentration in the
H1299 cell line from 7.22 to 2.67 pg/ml. On the contrary, gefitinib
decreased EGFR concentration to 3.72 pg/ml. This indicates that
compound 6b inhibits the growth of NSCLC cell line H1299 through
EGFR inhibition. The results are summarized in Table 2.
In general, the infrared (IR) spectra displayed a secondary
amine peak at 3,300–3,500 cm−1, which indicated the coupling of
compounds 3a and 3b with different amines and peak for (C–S) at
the range of 690–830 cm−1
.
1H‐NMR (nuclear magnetic
|
Measurement of potential EGFR inhibitory
activity IC50
resonance) spectra of the new compounds showed the appear-
ance of multiplet peaks of aromatic protons at the range of 6–8 δ
(ppm) and the characteristic upfield signals corresponding to
aliphatic protons of both cyclohexanone and cyclopentanone at
the range of 1–3 δ (ppm). In addition, the 1H‐NMR spectra of
compounds 4b, 4d, 5b, and 5d showed two sharp peaks
corresponding to the two methyl groups of the isoxazolo‐ and
pyrazolo‐pyridine moiety at the range of 2.52–3.65 δ (ppm). The
1H‐NMR spectra of compounds 7a and 7b showed the character-
istic singlet peak for CH2 of the benzylamine moiety at 3.86 and
4.24, respectively.
2.2.3
Compound 6b displayed an IC50 value of 0.33 nM, which is superior
to those exerted by the reported thienopyrimidine derivative
36 nM,[17] gefitinib 1.9 nM, and erlotinib 4 nM.
|
2.3
Pharmacophore and 3D‐QSAR models
In this study, the pharmacophore generation model was developed
using the pharmacophore modeling task of Schrodinger (10.1).
Twelve synthesized compounds are considered with promising
activity against the H1299 cancer cell line in the 3D‐QSAR
pharmacophore model. Phase supplies a built‐in set of six pharma-
cophore features: hydrogen bond acceptor (A), hydrogen bond donor
|
2.2
Biology
|
2.2.1
Cytotoxic activity against H1299
The potential anticancer activity of the new compounds was
evaluated using sulforhodamine‐B assay against NSCLC cell line
H1299.[24] It is well‐known that EGFR is overexpressed in
(H1299).[25] It was observed that all the newly synthesized
compounds displayed remarkable inhibitory activities against the
lung carcinoma cell line (H1299) with IC50 values ranging from
25 to 48 nM superior to that of gefitinib 40 µM. The results are
summarized in Table 1.
TABLE 2 Measurement of epidermal growth factor receptor
(EGFR) concentration in the lung cancer cell line H1299
IC50 against
H1299
EGFR concentration (pg/ml) in
H1299
Compound
No compound
6b
–
7.22
2.67
3.72
25 nM
40 µM
Gefitinib

4 of 15
ELRAYESS ET AL.
|
FIGURE 2 Quantitative structure–activity relationship model visualized in the context of favorable and unfavorable hydrogen bond donor
effects in (a, d), hydrophobic interactions in (b, e), and electron‐withdrawing groups in (c, f) for the highest and lowest active compounds 6b and
4a, respectively
(D), hydrophobic group (H), negatively charged group (N), positively
charged group (P), and aromatic ring (R). Atom‐based 3D‐QSAR
revealed how substitution can influence biological activity. Twelve
common pharmacophore models were generated with a different
combination of variants (AAADH, AAADR, AAAHR, AAARR, AADHR,
AADRR, AAHRR, AARRR, ADHRR, ADRRR, AHRRR, and DHRRR).
Among these pharmacophore models, ADHRR showed the best
survival score and an optimum alignment with active compounds.
Thus, it was selected to generate the QSAR model. Ten predictive
pharmacophore models of the variant AAHHR were generated
(Table 1 and Supporting Information). All the generated pharmaco-
phore models kept at least five chemical features. The models with
maximum adjusted survival scores were chosen for creating the
atom‐based alignment of EGFR inhibitors. Model ADHRR.26 has
been carefully chosen because it generated good alignment of active
ligands in comparison with inactive ones. The special arrangement of
features along with their distances present in five‐featured pharma-
cophore ADHRR is shown in Figure 1 and the Supporting Informa-
tion. As depicted in Figure 2a–f, the hydrogen bond acceptors in 12
training are mapped to N‐triazine. The hydrogen bond donors are
mapped to NH‐amino group. The hydrophobic groups are mapped to
cyclopentyl or cyclohexyl, whereas aromatic ring features are
mapped to thiophene, phenyl, pyrazole, or isoxazole moieties as
shown in Figure 2a–f.
the same feature map (Figure 3 and Supporting Information). In
gefitinib, the hydrogen bond acceptor is mapped to N‐pyrimidine. The
hydrogen bond donors are mapped to the NH‐amino group at C4 of
quinazoline. The hydrophobic groups are mapped to the methoxy
group, whereas the aromatic ring features are mapped to a benzoid
group of quinazoline and phenyl group adjacent to the NH group. The
3D‐QSAR model was created using the partial least squares (PLS)
regression statistics. The grid spacing was kept at 1 Å. The number of
PLS factors included in the model development is 3 (Table 3). The
best‐fitted model ADHRR.26 (R2 = 0.9908, Q2 = 0.8493, and
F = 97.10) consists of one hydrophobic interaction, one hydrogen
The six compounds (test set) also showed the same feature map
of the training set. Moreover, gefitinib and two of the reported
ligands with potent inhibition against EGFR exhibited more or less
FIGURE 3 Three‐dimensional‐binding modes and overlay of
erlotinib, gefitinib, compounds 6b and 4a in the catalytic domain of
epidermal growth factor receptor (PDB 1M17)

ELRAYESS ET AL.
5 of 15
|
TABLE 3 PLS statistical parameters of the model ADHRR.26
PLS
1
SD
R2
F
p value
Stability
−0.5372
−0.5763
−0.6285
RMSE
Q2
Pearson’s R
0.310
0.1259
0.0722
0.0362
0.8682
0.9599
0.9908
85.6
4.394e−007
4.139e−009
1.801e−011
0.1267
0.0842
0.1008
0.0938
0.2365
0.7287
2
143.7
393.9
0.5342
3
0.4283
Abbreviations: PLS, partial least squares; RMSE, root mean square error; SD, standard deviation.
bond donor, one hydrogen bond acceptor, and two aromatic features.
The distance and angles between different sites of the model
ADHRR.26 are presented in Table 3 and Supporting Information.
the spatial arrangement of structural features on the EGFR inhibition
(Table 3). The stability of a statistically significant regression model is
specified by the high value of F (393.9) and a high degree of
confidence along with the small value of variance ratio p (1.801e
−011). In addition, the data used for model generation are optimum
for the QSAR analysis. This is understandable from the small values
of RMSE (0.1008) and standard deviation (0.0326) and from the high
value of R2 (squared correlation coefficient value; 0.9908). The model
was validated by the cross‐validated correlation coefficient, Q2
(0.7287), which was obtained by the leave‐one‐out or leave‐N‐out
method (Table 3). The statistical parameter of Q2 is more reliable
than that of R2. The scatter plots for experimental and predicted
activities of synthesized molecules indicated a significant linear
correlation as shown in Figure 3a,b and Supporting Information.
Validation for the generated pharmacophore model (ADHRR26) was
done by mapping one approved drug gefitinib and two reported
compounds, tetrahydrobenzothieno[3,2‐d]pyrimidine,[26] with EGFR
inhibitory activity. Their fit values and the fit values of the tested
compounds and their predicted and experimental activity were
compared using our ligand pharmacophore mapping protocol
(Table 4). The expected activities of these compounds and the
training set through the pharmacophore model as well as their fit
values are presented in Table 4. It is worth noting that the expected
cytotoxic activity created by the 3D pharmacophore QSAR model
was closely related to that of experimentally observed. Therefore,
these QSAR models could be safely applied for predicting the
cytotoxicity of new compounds.
|
2.3.1
3D‐QSAR contour map analysis
The 3D‐QSAR models with combined effects of electron‐withdrawing,
hydrogen bond donor, as well as hydrophobic/nonpolar are displayed in
Figure 2a–f. The blue regions indicated favorable interactions between
ligand and target enzymes. On the contrary, the red regions indicated
unfavorable features. The effect of the hydrogen bond donor is presented
by the blue cube on the amino group on C4 of the triazine ring of the
highest active and nearly all synthesized compounds. The majority of the
approved drugs bear this amino group. Thus, the presence of the amino
group is essential for the EGFR inhibitory activity. Moreover, the
presence of another blue cube near to the amino group at the meta
position of the 4‐anilino group gives the compounds 6a,b with this
electron‐donating amino group the biological superiority. Compound 6b
with the m‐amino group had shown more cytotoxic activity than its
analog compound 7b, which lacks this important group. The hydrophobic
contour of the highest active compounds revealed the necessity of
hydrophobic interaction of 4‐anilino group and other bicyclic systems. In
addition, para substitution of 4‐anilino group gave extra hydrophobic
interaction and this is consistent with the high potency of gefitinib.
Another important position for hydrophobic interaction was the
cyclohexenyl group adjacent to thiophene; modifying cyclohexyl to
cyclopentenyl could influence the biological activity as shown in
compound 6a compared with compound 6b. We attributed this
observation to the occupation of cyclopentenyl ring near to red cube
(Figure 2a–f). The lowest cytotoxic compound (5b) in our series unveiled
that the two methyl substitutions on isoxazolopyridine are occupied in
the red region and thus, unfavored for the activity. The electron‐
withdrawing contour is obviously shown at the meta and para positions of
anilino attached to triazine. The blue cube at the aforementioned
positions revealed the necessity of electron‐withdrawing at this position.
This is clear in the approved EGFR inhibitors gefitinib and erlotinib, which
contain chloro and cyano groups, respectively.
|
2.4
Molecular docking
In this study, the activity of tetrahydrobenzothieno[2,3‐d][1,2,3]‐
triazine and dihydrocyclopentathieno[2,3‐d][1,2,3]triazine derivatives
toward EGFR was explained using Glide docking of the Schrodinger
program. The synthesized compounds were docked well in the
erlotinib‐binding site and set up many electrostatic interactions as
well as hydrogen bonding and hydrophobic interactions (Figure 3).
LEU 694, LEU 768, PHE 771, PRO 770, GLY 772, MET 769, and CYS
773 were the main amino acids involved in these interactions.
Generally, tetrahydrobenzothieno[2,3‐d][1,2,3]triazine scaffold, espe-
cially tetrahydrobenzene and thiophene moieties, exhibited hydro-
phobic interaction with LEU 694, PRO 770, and CYS 773.
|
2.3.2
Pharmacophore model validation
The selection of the generated pharmacophore model (ADHRR26)
was on the basis of its validation. A 3D‐QSAR study was carried on
the six synthesized compounds (test set) to explore the reflection of
This scaffold occupied a deep pocket at the back of the adenosine
triphosphate (ATP)‐binding site (Figure 3). In addition, phenyl,
isoxazole‐pyridine, and pyrazolo‐pyridine attached to triazine via

6 of 15
ELRAYESS ET AL.
|
TABLE 4 Calculated pIC50 for designed, approved, and reported compounds
Predicted
Residual
activity
In
Ligand name
QSAR set
Activity
7.50
7.25
7.43
7.23
7.53
7.37
7.55
7.27
7.46
7.52
7.60
7.52
7.76
6.48
6.76
activity
7.41
7.24
7.42
7.28
7.54
7.35
7.47
7.33
7.47
7.43
7.64
7.51
7.74
6.48
6.75
Pharm set
Active
Active
Active
Active
Active
Active
Active
Active
Active
Active
Active
Active
Active
Active
Active
Fitness
3.00
2.84
2.95
2.82
2.62
2.07
2.48
2.73
2.48
2.71
2.56
2.07
1.46
1.62
2.17
1
4a
Test
0.09
0.01
2
4b
Training
Training
Test
3
5a
0.01
4
5b
−0.05
−0.01
0.02
5
6a
Training
Test
6
7a
7
4c
Training
Training
Training
Test
0.08
8
4d
−0.06
−0.01
0.07
9
5c
10
11
12
13
14
15
5d
6b
Training
Training
−0.04
0.01
7b
Gefitinib
DER‐3
DER‐4
0.02
0.00
0.01
FIGURE 4 Two‐dimensional ligand–receptor interactions of (a) erlotinib, (b) gefitinib, (c) compound 6b, and (d) compound 4a in the catalytic
domain of the epidermal growth factor receptor (1M17)

ELRAYESS ET AL.
7 of 15
|
(a)
(b)
Electrostatic Surface Potential (Red: -ve; Blue: +ve)
FIGURE 5 (a) Mulliken charges, (b) electrostatic surface potential calculated using 6–311++G (d,p) basic set methodology (color‐coded from
red to blue) and density functional theory method with B3LYP functional
FIGURE 6 Plots of HOMO and LUMO of compound 6b (a, b) and gefitinib (c, d)

8 of 15
ELRAYESS ET AL.
|
FIGURE 7 The observed
structure–activity relationships of the
newly synthesized compounds against
H1299. EGFR, epidermal growth factor
receptor
NH linkage also revealed hydrophobic interaction with CYS 751, PHE
832, TYR 703, LEU 764, LEU 753, MET 742, and ILE 720. As evident,
both the above scaffolds are buried within the hydrophobic pocket,
which could give our compound stabilization at an active site. The m‐
amino group on the phenyl group and NH of pyrazole also showed
polar interaction with ALA 698 and THR 766 residues. This polar
interaction gave pyrazole derivative a superior affinity compared
with isoxazole, which lacks this hydrophilic interaction. The docked
structure of compound 6b indicated that this compound fits reason-
ably in the receptor cavity (Figure 3). N‐1 of triazine interacted via a
hydrogen bonding of 2.08 Å with MET 769, and N‐3 made water‐
mediated H‐bond of 2.07 Å to the side chain of THR 766.
polar and nonpolar interactions in their binding to EGFR. The
Mulliken charge distribution for all the atoms as well as the ESP for
gefitinib and compound 6b were calculated using the B3LYP
functional with a 6–311++G (d,p) basis set. In Figure 5a, the
calculated atomic charge of the nitrogen atom of triazine in
compound 6b and pyrimidine in gefitinib interacting with MET 769
and THR 766 in hinge region was found to be (−0.23 to −0.608). This
fortified the hydrogen bond (HBA) formation and highlighted the
significance of interaction for EGFR selectivity. Interestingly, the high
negative atomic charge of aniline NH of −0.77 expected hydrogen
bond or salt bridge formation with GLU 738 in gatekeeper residue.
Shape and size difference is apparently observed in the ESP of both
compounds (Figure 5) with mostly two arms of gefitinib. The
noticeable difference is the strong positive charge localized on the
right arm of m‐amino aniline of compound 6b that gives it the
superiority to tightly bind within gatekeeper residue. This, to some
extent, explained the novelty of the designed compound toward
EGFR. The inclusion of the electron‐donating group, NH2, instead of
Cl, F, and CN in most of EGFR inhibitors represent a new discovery in
PTK inhibition. The molecular docking study came in agreement with
computational study, where nitrogen of triazine ring besides m‐amino
group of aniline are involved in important interaction with the key
residue of the hinge region, gatekeeper, and hydrophobic pocket.
The orbital energies of HOMO and LUMO of most active compound
6b were calculated using Gaussian 09 software. The HOMO and LUMO
sites were plotted on the molecular surface of compound 6b as shown in
Figure 6. The analysis of HOMO maps of compound 6b showed that
HOMO molecular orbitals are located on m‐amino aniline and three
nitrogen atoms of triazine moiety, indicating their involvement in
protein–ligand interaction. On the contrary, the LUMO molecular orbitals
are located on m‐amino phenyl moiety of compound 6b. Docking study
revealed the involvement of these moieties in hydrogen bond formation
with key residues of the hinge region and gatekeeper residue and
hydrophobic. Figure 7 summarizes the cytotoxic activity against H1299
depending on the observed structure–activity relationships.
Moreover, NH2 at meta position of aniline and NH of pyrazole
interacted via hydrogen bond of 2.09 Å and 1.6 Å with GLU 738 and ASP
831. These extra hydrogen bonding could give our compound the high
affinity and stability within the active receptor site compared with
erlotinib, the EGFR TK inhibitor cocrystallized within the receptor site
(PDB 1M17). Our docking results came in consistency with the result of
3D‐QSAR pharmacophore modeling. Compared with the approved EGFR
TK inhibitors, our compounds showed optimum hydrophilic, hydrophobic,
and hydrogen bond interactions as shown in Figure 4a–d.
The accuracy of our docking procedure was confirmed by
checking the interaction of the erlotinib to 1M17 receptor after
grid generation. Glide docking of erlotinib showed the same pose
and interaction with the receptor as that in grid generation. In
addition, the accuracy could be checked by examining the docking
score and Glide emodel. Compound 6b had the highest docking
score of −7.128 and Glide emodel of −65.055, whereas compound
5b showed −5.76 and −54.54 of docking score and Glide emodel.
These results came close to the potency of these compounds as
EGFR inhibitor. Gefitinib and erlotinib showed −6.32 and −6.79 of
docking score and −86.11 and −86.68 of Glide emodel,
respectively.
|
2.5
Computational study
|
Lipinski’s rule for drug likeliness and in silico
ADME prediction
The energy calculation in Gaussian 09 software using the density
functional theory (DFT) method on the crystallized conformation of
gefitinib and the most active compound 6b was run to visualize the
charge distribution on both compounds, to better understand the
chemical reactivity of molecules, and to estimate the contribution of
2.6
Pharmacokinetic features for the most active compound (6b) were
calculated utilizing ADME predictions by QikProp v4.3 function of

ELRAYESS ET AL.
9 of 15
|
Schrodinger 10‐1 (Table 4). The compound 6b was estimated for its
fundamental parameters of Lipinski’s rule of five. The oral bioavailability
of the drug molecules could be controlled by polar surface area (7–200 Å)
and the value of rotatable bonds (0–15). Caco‐2 cell permeability
(QPPCaco) was used as a model for the gut–blood barrier by nonactive
transport. This is a marker for intestinal absorption or permeation.
Caco‐2 cell permeability of the test compound 6b showed good results
predicting excellent drug permeability compared with gefitinib. The cell
permeability of the blood–brain barrier (BBB) was expected by
QPPMDCK descriptor. MDCK cells are
a good mimic for BBB.
Compound 6b exhibited good BBB permeability (QPPMDCK > 25).
Another important descriptor is QPlogkhsa descriptor of QikProp that
predicts human serum albumin‐binding capability. Compound 6b showed
reasonable binding to plasma proteins. QPlogP o/w and QPlogS are
descriptors of octanol/water partition coefficient and aqueous solubility;
the test compound was found to be in the permissible range (Table 4).
The most active compound 6b was considered as a promising lead
molecule for designing potent EGFR inhibitors with excellent membrane
permeability and oral bioavailability. The physicochemical properties of
other designed compounds were enough to permeate our drug through
GIT. Compounds 4a–d and compounds 5a–d having bicyclic system‐
substituted pyrazolo‐pyridine and isoxazole‐pyridine showed the highest
partition coefficient and consequently highest penetration disclosed by
great value of QPPCaco and human oral absorption % (data not shown).
In silico ADME prediction results came in agreement with the in vitro
study; compound 6b achieved the superior cytotoxic activity against
H1299. Alternatively, gefitinib showed excellent in silico ADME result
including good partition coefficient, permeability, and oral bioavailability
(Table 5).
Moreover, most of the kinase inhibitors bind to the ATP‐binding site
of the enzyme in its active “DFG‐in” state. Such molecules are known as
type I inhibitors that mimic the interaction of the adenine ring of ATP
with the hinge region.[27] Type I EGFR inhibitors involve the approved
marketed drugs gefitinib and erlotinib. Selective inhibitors create
interactions with the hydrophobic back cavity of the ATP‐binding site.
Type I inhibitors can achieve certain level of selectivity when the
substituted phenyl ring attached to C4 of quinazoline via NH occupies
the selectivity pocket. In the design of our compounds, we kept this ring
in certain series and modified it to pyrazolo‐pyridine and isoxazole‐
pyridine in another series in an attempt to study how this variation could
affect the biological activity and binding mode. We found that the phenyl
ring was the optimum considering molecular docking and biological
activity. Type I EGFR inhibitors have the electron‐withdrawing moiety
attached to the phenyl ring, like Cl in case of gefitinib and CN in case of
erlotinib. Meanwhile, in this study, another strategy was applied;
electron‐withdrawing moiety was changed with electron‐donating group.
Meta position of phenyl ring was substituted with NH2, which gave our
compound the chance to form hydrogen bond with GLU 738 and one
DFG motif, ASP381 and this was clearly observed by the high negative
atomic charge of NH (−0.77) in Mulliken charge distribution. This
additional interaction could enhance the EGFR inhibitory activity and
drug affinity to EGFR. Compound 6b achieved the highest potency,
affinity, and stability as shown by its lowest IC50 0.33 nM, highest docking

10 of 15
ELRAYESS ET AL.
|
SCHEME 1 Synthesis of dihydrocyclopentathieno and tetrahydrobenzothieno triazine derivatives
score and lowest Glide model energy. Increasing one atom between NH
appended at C4 of triazine and phenyl group could influence the affinity
and biological activity and this was obviously seen in compounds 7a,b
compared with compounds 6a,b.
EGFR inhibition. Fortunately, the most active compound 6b
decreased EGFR concentration in the H1299 cell line from 7.22 to
2.67 pg/ml. The molecular modeling study explained the potent
inhibitory activity of the most active compound 6b. Moreover,
compound 6b showed an IC50 value of 0.33 nM against EGFR
superior to that of gefitinib 1.9 nM and erlotinib 4 nM.
Tetrahydrobenzothieno[2,3‐d][1,2,3]triazine scaffold occupied the
ATP‐binding site showing the same drug–receptor interaction like
the approved drugs gefitinib and erlotinib. Additional bindings were
provided by our structural modification. The atom‐based 3D‐QSAR
generated a model ADHRR that exhibited good correlation (R2),
cross‐validation (Q2) coefficients, and excellent agreement between
|
3
CONCLUSION
The novel synthesized compounds exhibited highly potent cytotoxic
activity toward NSCLC cell line H1299 with IC50 values ranging from
25 to 58 nM, superior to that of gefitinib 40 µM. The antiproliferative
activity of the designed compounds was expected to be through

ELRAYESS ET AL.
11 of 15
|
experimental and predictive activities. Molecular orbital (HOMO/
LUMO) analysis, ESP, and Mulliken charge distribution were
beneficiary tools in understanding the chemical reactivity of designed
molecules with EGFR. Our new thienotriazine scaffold might serve in
the development of EGFR inhibitors targeting NSCLC cell line
H1299.
mixture was filtered while hot and washed with ice‐water. The
precipitate was then recrystallized from ethanol.[22]
N‐(5,6‐Dihydro‐7H‐cyclopenta[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐
ylamino)pyrazolo[3,4‐b]pyridin‐3‐amine (4a)
Melting point: >300°C (yield: 60%). Mass spectrum: m/z (%): 309
(M+), 307, 291, 208, 119, and 75. IR (cm−1): 771 (C–S), 1,288 (C–N),
1,624 (C═N), and 3,421 (N–H), 1H‐NMR (dimethyl sulfoxide [DMSO]‐
d6, 400 MHz): δ (ppm) = 1.34–1.17 (m, 2H, H‐2), 1.93 (t, J = 7.1 Hz, 2H,
H‐1), 2.73 (t, J = 7.1 Hz, 2H, H‐3), 3.60 (s, 1H, H‐4), 7.35–7.19 (m, 3H,
Ar–H), and 9.69 (s, 1H, H‐5). Elemental analysis calculated for
C14H11N7S: C, 54.36; H, 3.58; N, 31.69. Found: C, 54.61; H, 3.64;
N, 31.98.
|
4
EXPERIMENTAL
Chemistry
|
4.1
|
4.1.1
General
The commercial chemicals and solvents for the synthesis were
reagent grade and used without further purification. Varian Mercury
VX‐300 NMR spectrometer or Jeol LA (400 MHz for 1H‐NMR,
100 MHz for 13C‐NMR) were used to measure 1H‐ and 13C‐NMR
spectra. Electron impact mass spectra (EI‐MS) were recorded on
Shimadzu GCMS‐QP 5050 A gas chromatograph‐mass spectrometer
(70 eV). Melting points (m.p.) were determined in open capillaries
using Gallenkamp melting point apparatus and are uncorrected.
Infrared (IR) spectra were recorded on a Shimadzu FT‐IR 8101 PC IR
spectrophotometer (KBr pellets). The synthetic procedures for the
target compounds are illustrated in Scheme 1.
N‐(5,6‐Dihydro‐7H‐cyclopenta[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐
ylamino)4,6‐dimethylpyrazolo[3,4‐b]pyridin‐3‐amine (4b)
Melting point: >300°C (yield: 65%). Mass spectrum: m/z (%): 337
(M+), 313, 287, 199, 148, and 146. IR (cm−1): 713 (C–S), 1,334 (C–N),
1,654 (C═N), 3,356 (N–H), and 3,421 (N–H). 1H‐NMR (DMSO‐d6,
400 MHz): δ (ppm) = 1.91–1.76 (m, 2H, H‐2), 2.41 (t, J = 7.1 Hz, 2H, H‐
1), 2.49 (t, 2H, J = 7.1 Hz, H‐3), 2.52 (s, 3H, CH3), 2.54 (s, 3H, CH3),
5.10 (s, 1H, H‐4), 6.58 (s, 1H, Ar–H), and 11.78 (s, 1H, H‐5). Elemental
analysis calculated for C16H15N7S: C, 56.96; H, 4.48; N, 29.06. Found:
C, 57.21; H, 4.53; N, 29.24.
The InChI codes of the investigated compounds together with
some biological activity data are provided as Supporting Information.
N‐(5,6‐Dihydro‐7H‐cyclopenta[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐
ylamino)oxazolo[3,4‐b]pyridin‐3‐amine (5a)
Melting point: >300°C (yield: 62%). Mass spectrum: m/z (%): 310
(M+), 299, 284, 216, 146, and 42. IR (cm−1): 752 (C–S), 1,273 (C–O),
1,624 (C═N), and 3,417 (N–H). 1H‐NMR (DMSO‐d6, 400 MHz): δ
(ppm) = 1.42–1.22 (m, 2H, H‐2), 2.58 (t, J = 7.1 Hz, 2H, H‐1), 2.71 (t,
J = 7.1 Hz, 2H, H‐3), 3.34 (s, 1H, NH), and 7.12–6.98 (m, 3H, Ar–H).
Elemental analysis calculated for C14H10N6OS: C, 54.18; H, 3.25; N,
27.08. Found: C, 54.37; H, 3.21; N, 27.41.
2‐Amino‐5,6‐dihydro‐4H‐cyclopenta[b]thiophene‐3‐carbonitrile (2a)
Sulfur (127 mg, 3.96 mmol) was added to a solution of cyclopenta-
none (3.96 mmol) in ethanol (25 ml) and malononitrile (3.96 mmol).
The mixture was stirred at 45°C. Morpholine (5.54 mmol) was added
dropwise over 15 min. The mixture was further stirred at 60°C for
18 hr. The mixture was filtered while hot and the resulting crystals
were washed with 30% ethanol and dried.[28] Melting point:
135–137°C (yield: 80%), as reported.[28] Mass spectrum: m/z (%):
166 (M++2), 164, 137, 121, 68, and 43.
N‐(5,6‐Dihydro‐7H‐cyclopenta[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐
ylamino)‐4,6‐dimethyloxazolo[3,4‐b]pyridin‐3‐amine (5b)
Melting point: >300°C (yield: 64%). Mass spectrum: m/z (%): 338
(M+), 330, 298, 239, 222, and 99. IR (cm−1): 740 (C–S), 1,157 (C–O),
1,273 (C–N), 1,627 (C═N), and 3,421 (N–H). 1H‐NMR (DMSO‐d6,
400 MHz): δ (ppm) = 1.82–1.65 (m, 2H, H‐2), 2.49 (t, J = 7.1 Hz, 2H,
H‐1), 2.55 (s, 3H, CH3), 2.59 (s, 3H, CH3), 2.71 (t, J = 7.1 Hz, 2H, H‐3),
7.31 (s, 1H, Ar–H), and 10.83 (s, 1H, NH). Elemental analysis
calculated for C16H14N6OS: C, 56.79; H, 4.17; N, 24.84. Found: C,
57.01; H, 4.20; N, 24.72.
4‐Chlorocyclopenta[4:5]thieno[2,3‐d]‐1,2,3‐triazine (3a)
A solution of sodium nitrite (1 gm, 16 mmol) was added to a mixture
of compound 2a (10 mmol), HCl (10 ml, 5%) and glacial acetic acid
(10 ml). The mixture was stirred at 0–5°C for 1 hr, then for 2 hrs at
20°C. The mixture was poured into 100 ml of water and the
precipitate was separated by filtration, washed with water, and
dried.[19] Melting point: 200°C (yield: 81.8%), as reported. Mass
spectrum: m/z (%): 213 (M++2), 212, 211, 199, 196, and 172.
3‐(5,6‐Dihydro‐7H‐cyclopenta[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐
ylamino)benzene‐1,3‐diamine (6a)
|
4.1.2
General procedure for the synthesis of
Melting point: >300°C (yield: 40%). Mass spectrum: m/z (%): 285
(M++2), 284, 283, 266, 246, and 44. IR (cm−1): 790 (C–S), 1,620
(C═N), 3,221 (N–H), and 3,348 (NH2). 1H‐NMR (DMSO‐d6, 400 MHz):
δ (ppm) = 1.41–1.23 (m, 2H, H‐2), 2.33 (t, J = 7.1 Hz, 2H, H‐1), 2.67 (t,
J = 7.1 Hz, 2H, H‐3), 4.39 (s, 1H, NH)), 8.21–7.85 (m, 3H, Ar–H), 7.91
compounds 4a,b, 5a,b, 6a, and 7a
A mixture of equimolar amounts (0.01 mol) of compound 3a and the
aromatic amines in dry pyridine (10 ml) was refluxed for 16 hr. The

12 of 15
ELRAYESS ET AL.
|
(s, 1H, ArH‐4), and 10.15 (s, 2H, NH2). Elemental analysis calculated
for C14H13N5S: C, 59.34; H, 4.62; N, 24.72. Found: C, 59.67; H, 4.71;
N, 25.03.
N‐(5,6,7,8‐Tetrahydro‐7H‐cyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐
4‐ylamino)4,6‐dimethylpyrazolo[3,4‐b]pyridin‐3‐amine (4d)
Melting point: 128–130°C (yield: 68%). Mass spectrum: m/z (%): 351
(M+), 350, 346, 332, 267, and 123. IR (cm−1): 767 (C–S), 1,593 (C═N),
and 3,294 (NH). 1H‐NMR (DMSO‐d6, 400 MHz): δ (ppm) = 1.38–1.15
(m, 4H, H‐2, H‐3), 1.51 (t, J = 7.1 Hz, 2H, H‐1), 1.53 (t, J = 7.1 Hz, 2H,
H‐4), 3.65 (s, 3H, CH3), 3.68 (s, 3H, CH3), 7.44 (s, 1H, Ar–H), 9.89 (s,
1H, H‐5), and 10.20 (s, 1H, H‐6). Elemental analysis calculated for
C17H17N7S: C, 68.10; H, 4.88; N, 27.90. Found: C, 68.37; H, 4.93;
N, 28.21.
3‐(5,6‐Dihydro‐7H‐cyclopenta[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐
ylamino)benzylamine (7a)
Melting point: 224–226°C (yield: 70%). Mass spectrum: m/z (%): 284,
(M++2), 283, 282, 269, 250, and 91. IR (cm−1): 752 (C–S), 1,350
(C–N), 1,597 (C═N), and 3,336 (N–H). 1H‐NMR (DMSO‐d6,
400 MHz): δ (ppm) = 1.42–1.23 (m, 2H, H‐2), 1.81 (t, J = 7.1 Hz, 2H,
H‐1), 2.72 (t, J = 7.1 Hz, 2H, H‐3), 4.24 (s, 2H, CH2), 7.78–7.31 (m, 5H,
Ar–H), and 8.66 (s, 1H, NH). Elemental analysis calculated for
C15H14N4S: C, 63.80; H, 5.00; N, 19.84. Found: C, 64.12; H, 5.09;
N, 20.07.
N‐(5,6,7,8‐Tetrahydro‐7H‐cyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐
4‐ylamino)oxazolo[3,4‐b]pyridin‐3‐amine (5c)
Melting point: 200°C (yield: 60%). Mass spectrum: m/z (%): 325
(M++1), 324, 280, 255, 212, 206, and 150. IR (cm−1): 771 (C–S), 1,384
(C–O), 1,627 (C═N), and 3,421 (N–H). 1H‐NMR (DMSO‐d6,
400 MHz): δ (ppm) = 1.37–1.15 (m, 4H, H‐2, H‐3), 2.51 (t, J = 7.1 Hz,
2H, H‐1), 2.71 (t, J = 7.1 Hz, 2H, H‐4), 7.61–7.30 (m, 3H, Ar–H), and
11.83 (s, 1H, NH). Elemental analysis calculated for C15H12N6OS: C,
55.54; H, 3.73; N, 25.91. Found C, 55.82; H, 3.35; N, 25.64.
2‐Amino‐4,5,6,7‐tetrahydrobenzo[b]thiophene‐3‐carbonitrile (2b)
Sulfur (127 mg, 3.96 mmol) was added to a solution of cyclohexanone
(3.96 mmol) in ethanol (25 ml) and malononitrile (3.96 mmol). The
mixture was stirred at 45°C. Morpholine (5.54 mmol) was added
dropwise over 15 min. The mixture was further stirred at 60°C for
18 hr. The mixture was filtered while hot and the resulting crystals
were washed with 30% ethanol and dried.[28] Melting point:
138–140°C (yield: 80%) as reported.[29] Mass spectrum: m/z (%)
180 (M++2), 179, 178, 151, and 150.
N‐(5,6,7,8‐Tetrahydro‐7H‐cyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐
4‐ylamino)‐4,6‐dimethyloxazolo[3,4‐b]pyridin‐3‐amine (5d)
Melting point: 198–200°C (yield: 64%). Mass spectrum: m/z (%): 354
(M++2), 352, 351, 334, 251, and 107. IR (cm−1): 759 (C–S), 1,384
(C–O), 1,624 (C═N), and 3,352 (N–H). 1H‐NMR (DMSO‐d6,
400 MHz): δ (ppm) = 1.56–1.18 (m, 4H, H‐2, H‐3), 1.51 (t, J = 7.1 Hz,
2H, H‐1), 2.14 (t, J = 7.1 Hz, 2H, H‐4), 2.57 (s, 3H, CH3), 2.60 (s, 3H,
CH3), 6.87 (s, 1H, Ar–H), and 8.91 (s, 1H, NH). Elemental analysis
calculated for C17H16N6OS: C, 57.94; H, 4.58; N, 23.85. Found C,
57.49; H, 4.26; N, 23.52.
4‐Chlorocyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazine (3b)
A solution of sodium nitrite (1 gm, 16 mmol) was added to a mixture
of compound 2b (10 mmol), HCl (10 ml, 5%), and glacial acetic acid
(10 ml). The mixture was stirred at 0–5°C for 1 hr, then for 2 hr at
20°C. The mixture was poured into 100 ml of water and the
precipitate was separated by filtration, washed with water, and
dried.[19] Melting point: 116–118°C (yield: 81.8%) as reported.[30]
Mass spectrum: m/z (%): 227 (M++2), 225, 211, 195, 192, and 150.
3‐(5,6,7,8‐Tetrahydro‐7H‐cyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐
4‐ylamino)benzene‐1,3‐diamine (6b)
Melting point: 128–130°C (yield: 65%). Mass spectrum: m/z (%): 297
(M+), 255, 236, 147, 95, and 43. IR (cm−1): 771 (C–S), 3,197 (N–H),
and 3,309 (NH2). 1H‐NMR (DMSO‐d6, 400 MHz): δ (ppm) = 1.6–1.20
(m, 4H, H‐2, H‐3), 1.76 (t, J = 7.1 Hz, 2H, H‐1), 2.57 (t, J = 7.1 Hz, 2H,
H‐4), 5.49 (s, 2H, NH2), 7.25–6.95 (m, 3H, Ar–H), 8.10 (s, 1H, ArH‐5),
and 11.88 (s, 1H, NH). 13C‐NMR (DMSO‐d6, 100 MHz): δ (ppm) = 24.6
(C‐1), 23.9 (C‐2, C‐3), 23.8 (C‐4), 127.8 (C‐5), 142.7 (C‐6), 153.3 (C‐7),
115.9 (C‐8), 93.2 (C‐10), 114.6 (C‐11), 131.2 (C‐12), 104.6 (C‐13),
142.7 (C‐14), and 146.7 (C‐15). Elemental analysis calculated for
C15H15N5S: C, 60.58; H, 5.08; N, 23.55. Found C, 60.27; H, 5.36;
N, 23.21.
|
4.1.3
General procedure for the synthesis of
compounds 4c,d, 5c,d, 6b, and 7b
A mixture of equimolar amounts (0.01 mol) of compound 3b and
(0.01 mol) of aromatic amines in dry pyridine (10 ml) was refluxed for
16 hr. The mixture was filtered while hot and washed with ice‐water.
The precipitate was then recrystallized from ethanol.[22]
N‐(5,6,7,8‐Tetrahydro‐7H‐cyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐
4‐ylamino)pyrazolo[3,4‐b]pyridin‐3‐amine (4c)
Melting point: 128–130°C (yield: 65%). Mass spectrum: m/z (%): 325
(M++2), 324, 323, 312, 311, 280, and 206. IR (cm−1): 817 (C–S), 1,519
3‐(5,6‐Dihydro‐7H‐cyclohexa[4:5]thieno[2,3‐d]‐1,2,3‐triazin‐4‐
ylamino)benzylamine (7b)
(C═N), and 3,325 (N–H). 1H‐NMR (DMSO‐d6, 400 MHz):
δ
(ppm) = 1.63–1.46 (m, 4H, H‐2, H‐3), 2.16 (t, J = 7.1 Hz, 2H, H‐1),
2.67 (t, J = 7.1 Hz, 2H, H‐4), 6.93 (s, 1H, H‐5), 7.48–7.18 (m, 3H,
Ar–H), and 8.91 (s, 1H, H‐6). Elemental analysis calculated for
C15H13N7S: C, 55.71; H, 4.05; N, 30.32. Found C, 55.64; H, 4.19;
N, 30.71.
Melting point: 124–126°C (yield: 60%). Mass, spectrum: m/z (%): 297
(M++1), 296, 268, 240, 212, and 91. IR (cm−1): 694 (C–S), 1,573
(C═N), and 3,313 (NH). 1H‐NMR (DMSO‐d6, 400 MHz):
δ
(ppm) = 2.21–1.79 (m, 4H, H‐2, H‐3), 2.23 (t, J = 7.1 Hz, 2H, H‐4),
3.17 (t, J = 7.1 Hz, 2H, H‐1), 3.86 (s, 2H, CH2), 7.09 (m, 5H, Ar–H), and

ELRAYESS ET AL.
13 of 15
|
9.87 (s, 1H, NH). Elemental analysis calculated for C16H16N4S: C,
64.84; H, 5.44; N, 18.90 Found: C, 65.12; H, 5.52; N, 19.14.
After adjusting the EGFR receptor and the thieno[2,3‐d][1,2,3]triazine
molecules in the correct form, the properties of the active site of EGFR
were characterized using “grid generation panel” in Glide.[34–36]
In the final step, the thieno[2,3‐d][1,2,3]triazine derivatives were
docked within the active site of EGFR using the optimized protein–ligand
geometries. The extra precision Glide scoring function is applied to rank
the docking poses and to assess the protein–ligand binding affinities.
Maestro’s Pose Viewer utility was utilized to visualize and analyze the
key elements of ligand–receptor interaction. The final best‐docked
structure with the lowest energy was chosen using a Glide score
function and selected for further experiments. Erlotinib was removed
from the crystal structure EGFR receptor and then redocked using the
previously mentioned step to evaluate the accuracy and precision of the
established docking protocol.[34–36]
|
4.2
Biological assays
|
4.2.1
Cytotoxic activity against H1299
Potential cytotoxicity of the compounds was tested using Skehan
method (Supporting Information).[24]
|
4.2.2
Measurement of human EGFR in H1299 cell
culture
The measurement of human EGFR in H1299 cell culture was
performed by using human EGFR ELISA kit as reported (Supporting
Information).[31,32]
|
4.4
Computational calculation
All calculations were performed using the DFT method and
the Gaussian 09 program package. With the standard basis set,
6‐311++G (d,p) levels of the theory was used for geometry
optimizations.[37]
|
4.2.3
Measurement of potential EGFR inhibitory
activity IC50
The measurement of potential EGFR inhibitory activity IC50 was
performed using Kinase‐Glo Plus luminescence kinase assay kit
(#V3772; Promega; Supporting Information).[33]
|
4.5
Lipinski’s rule for drug likeliness and in silico
ADME prediction
Drug‐likeness properties of the newly synthesized compounds, with
expected biological and/or pharmacological activity, were evaluated
according to Lipinski’s rule of five, which was utilized to determine
whether these compounds have the properties that would allow
them to be a likely orally active drug in humans. Lipinski’s rule of five
prescribes ADME, which is absorption, distribution, metabolism, and
excretion. The drug‐like behavior of our compounds was predicted
by using QikProp module (v4.2; Schrodinger 2015‐1). The
thieno[2,3‐d][1,2,3]triazine compounds prepared by LigPrep module
(v3.1; Schrodinger 2015‐1) in the previous step were applied for
the calculation of pharmacokinetic parameters by QikProp v4.2.[38]
|
4.3
Molecular modeling study
In the current research, a molecular modeling study was conducted using
the Glide docking function incorporated in the Schrodinger‐10.1
molecular modeling program. The X‐ray crystal structure of the catalytic
domain of the EGFR enzyme in complex with erlotinib (PDB ID: 1M17
resolution 2.6 Å) was obtained from Protein Data Bank (PDB; http://
www.rcsb.org/pdb). The EGFR–erlotinib complex was refined for the
Glide docking calculations using the protein preparation wizard applying
the OPLS‐2005 force field. In the second step, water of crystallography, if
present, was removed and the pH was adjusted to 7.0. In the third step,
the appropriate charge and protonation state of protein were adjusted by
the protein assignment script, and then the protein–inhibitor complex
was subjected to energy minimization until the (RMSD) value of the
nonhydrogen atoms reached 0.3 Å to discard the steric clashes.[34–36]
Using ligand preparation wizard, the 3D structures of the thieno[2,3‐
d][1,2,3]triazine derivatives were constructed and optimized with the
build‐panel in Maestro. Partial atomic charges were ascribed for the
thieno[2,3‐d][1,2,3]triazine derivatives using the OPLS‐2005 force field
and possible ionization states were generated at pH 7. To soften the
potential for nonpolar parts of the receptor, the van der Waals radii of
receptor atoms were scaled by 0.8 with a partial atomic charge of 0.15. A
grid box with coordinates X = 10, Y = 10, and Z = 10 was generated at the
centroid of the active site. Furthermore, the energy of the obtained ligand
structures was minimized until it reached the RMSD cutoff of 0.01 Å.
|
4.6
QSAR
|
4.6.1
Data set
A set comprising 18 compounds bearing tetrahydrobenzothieno[2,3‐d]‐
[1,2,3]triazine and dihydrocyclopentathieno[2,3‐d][1,2,3]triazine scaffolds
were synthesized and selected to establish the QSAR model. The
inhibition potencies of the target compounds involved in data set
were reported as IC50 values varied from 25 to 58 nM. The IC50 values
were converted into molar values, which were then converted into pIC50
values using the following formula:
pIC50 = −Log (IC50).

14 of 15
ELRAYESS ET AL.
|
The 3D structures of thienotriazine derivatives were created
using the builder panel in Maestro. Furthermore, these structures
were optimized using LigPrep module (v2.1; Schrodinger 2015‐1) as
described above.
REFERENCES
[1] M.‐H. Jang, M.‐C. Shin, S. Lim, S.‐M. Han, H.‐J. Park, I. Shin, J.‐S.
Lee, K.‐A. Kim, E.‐H. Kim, C.‐J. Kim, J. Pharmacol. Sci. 2003,
91, 95.
[2] M. Sudo, T. M. Chin, S. Mori, N. B. Doan, J. W. Said, M. Akashi, H. P.
Koeffler, Cancer Chemother. Pharmacol. 2013, 71, 1325.
[3] G. Scagliotti, J. von Pawel, S. Novello, R. Ramlau, A. Favaretto, F.
Barlesi, W. Akerley, S. Orlov, A. Santoro, D. Spigel, J. Clin. Oncol.
2015, 33, 2667.
|
4.6.2
Pharmacophore 3D‐QSAR modeling
[4] Key statistics for lung cancer. https://www.cancer.org/cancer/non‐
small‐cell‐lung‐cancer/about/key‐statistics.html (accessed: 2017).
[5] D. Cronin‐Fenton, T. Dalvi, E. Hedgeman, M. Norgaard, L. Petersen,
H. Hansen, J. Fryzek, D. Lawrence, J. Walker, A. Mellemgaard, J.
Thorac. Oncol. 2017, 12, S2085.
Phase (v4.1; Schrodinger 2015‐1) was utilized to generate pharmaco-
phore and 3D‐QSAR models for EGFR inhibitors. The prepared ligands
with their respective biological activity values, pIC50, were imported for
developing the pharmacophore model of phase. The ligands were
ascribed as active with a threshold of pIC50 > 7.0 and inactive with a
threshold of pIC50 < 6.0. The remaining compounds were regarded as
moderately active. In the present study, phase (v4.0) shape screening for
flexible alignment of the selected EGFR inhibitors was utilized. The
highest active compound 6b was taken as the template. One hundred
conformers were generated with a maximum of 10 conformations per
rotatable bonds using the established default. The most important and
difficult step is the selection of training and test sets. To establish the
QSAR model robustness, a widely utilized approach and a random
division was applied. In this method, three trials were run.[39]
Pharmacophore sites for thienotriazine training and test included default
set of chemical features of phase: one hydrogen bond acceptor (A), one
hydrogen donor (D), hydrophobic interaction (H), and two aromatic
rings (R).
[6] B. Jeremic, E. Fidarova, V. Sharma, M. Faheem, A. A. Ameira, C. Nasr
Ben Ammar, A. Frobe, F. Lau, S. Brincat, G. Jones, Radiother. Oncol.
2015, 116, 21.
[7] L. Regales, Y. Gong, R. Shen, E. de Stanchina, I. Vivanco, A. Goel, J. A.
Koutcher, M. Spassova, O. Ouerfelli, I. K. Mellinghoff, J. Clin. Invest.
2009, 119, 3000.
[8] F. Janku, I. Garrido‐Laguna, L. B. Petruzelka, D. J. Stewart, R.
Kurzrock, J. Thorac. Oncol. 2011, 6, 1601.
[9] C.‐H. Yang, H.‐C. Chou, Y.‐N. Fu, C.‐L. Yeh, H.‐W. Cheng, I.‐C. Chang,
K.‐J. Liu, G.‐C. Chang, T.‐F. Tsai, S.‐F. Tsai, Biochim. Biophys. Acta
2015, 1852, 1540.
[10] W. Han, H. Pan, Y. Chen, J. Sun, Y. Wang, J. Li, W. Ge, L. Feng, X. Lin,
X. Wang, PLOS One 2011, 6, e18691.
[11] M. Tiseo, M. Loprevite, A. Ardizzoni, Curr. Med. Chem.: Anti‐Cancer
Agents 2004, 4, 139.
[12] A. Antonello, A. Tarozzi, F. Morroni, A. Cavalli, M. Rosini, P. Hrelia,
M. L. Bolognesi, C. Melchiorre, J. Med. Chem. 2006, 49, 6642.
[13] C. Gan, L. Liu, J. Cui, Z. Liu, Future Med. Chem. 2017, 13, 375.
[14] K. E. Lackey, Designing Multi‐Target Drugs, Royal Society of
Chemistry, 2012, pp. 181–205.
|
4.6.3
Model validation
[15] C.‐H. Wu, M. S. Coumar, C.‐Y. Chu, W.‐H. Lin, Y.‐R. Chen, C.‐T. Chen,
H.‐Y. Shiao, S. Rafi, S.‐Y. Wang, H. Hsu, J. Med. Chem. 2010, 53, 7316.
[16] Y. M. Abdel Aziz, M. M. Said, H. A. El Shihawy, K. A. M. Abouzid,
Bioorg. Chem. 2015, 60, 1.
The QSAR model ADHRR.26 with six components PLS factor was
characterized as the best model. The predicted inhibition of EGFR
activity of training set ligands provoked a correlation (R) of 0.85 with
the observed inhibition. The efficacy of model ADHRR.26 was also
tested with external validation. Graph of actual value versus
predicted value was also plotted.
[17] Y. M. A. Aziz, M. M. Said, H. A. El Shihawy, M. F. Tolba, K. A. M.
Abouzid, Chem. Pharm. Bull. 2015, 63, 1015.
[18] S. Bugge, S. J. Kaspersen, S. Larsen, U. Nonstad, G. Bjørkøy, E.
Sundby, B. H. Hoff, Eur. J. Med. Chem. 2014, 75, 354.
[19] M. Said, H. Elshihawy, Pak. J. Pharm. Sci. 2014, 27.
[20] S. Mowafy, A. Galanis, Z. M. Doctor, R. M. Paranal, D. S. Lasheen, N. A.
Farag, P. A. Jänne, K. A. M. Abouzid, Bioorg. Med. Chem. 2016, 24, 3501.
[21] H. Li, S. Zhou, X. Li, D. Wang, Y. Wang, C. Zhou, G. Schmid‐Bindert,
Cancer Biother. Radiopharm. 2013, 28, 115.
CONFLICT OF INTERESTS
[22] A. M. Al‐Obaid, S. G. Abdel‐Hamide, H. A. El‐Kashef, A. A. M. Abdel‐
Aziz, A. S. El‐Azab, H. A. Al‐Khamees, H. I. El‐Subbagh, Eur. J. Med.
Chem. 2009, 44, 2379.
The authors declare that there is no conflict of interests.
[23] F. A. Yassin, Chem. Heterocycl. Compd. 2009, 45, 35.
[24] P. Skehan, R. Storeng, D. Scudiero, A. Monks, J. McMahon, D. Vistica,
J. T. Warren, H. Bokesch, S. Kenney, M. R. Boyd, J. Natl. Cancer Inst.
1990, 82, 1107.
AUTHOR CONTRIBUTIONS
[25] H. N. Kang, S.‐H. Kim, M. R. Yun, H. R. Kim, S. M. Lim, M.‐S. Kim,
K.‐W. Hong, S.‐M. Kim, H. Kim, K.‐H. Pyo, Lung Cancer 2016, 95, 57.
[26] N. Ghareb, M. M. Abdel Daim, N. M. El‐Sayed, M. S. Elgawish, Bioorg.
Chem. 2017, 71, 110.
The manuscript was written through the contributions of all authors.
All authors have given approval to the final version of the
manuscript.
[27] S. Mowafy, N. A. Farag, K. A. M. Abouzid, Eur. J. Med. Chem. 2013,
61, 132.
[28] K. Gewald, Chem. Ber. 1965, 98, 3571.
ORCID
[29] M. G. Assy, M. H. Sherif, A. E. G. E. Amr, O. I. Abdelsalam, M. A.
Al‐Omar, M. M. Abdalla, I. Ragab, J. Heterocycl. Chem. 2013,
50, 766.
Mohamed M. Said
http://orcid.org/0000-0001-6997-7870

ELRAYESS ET AL.
15 of 15
|
[30] G. L. Viswanatha, B. J. Priyadarshini, N. Krishnadas, S.
Janardhanan, S. Rangappa, S. Hanumanthappa, Saudi Pharm. J.
2012, 20, 45.
Arabian J. Chem. 2016, 6, 153. https://doi.org/10.1016/j.arabjc.2016.
09.019
[31] C. T. Kuan, C. J. Wikstrand, D. D. Bigner, Endocr.‐Relat. Cancer 2001,
8, 83.
SUPPORTING INFORMATION
[32] R. S. Herbst, Inter. J. Radiat. Oncol. Biol. Phys. 2004, 59, S21.
[33] http://www.bpsbioscience.com/biochemical‐based‐assays (accessed:
2019).
Additional supporting information may be found online in the
Supporting Information section.
[34] T. A. Halgren, R. B. Murphy, R. A. Friesner, H. S. Beard, L. L. Frye, W.
T. Pollard, J. L. Banks, J. Med. Chem. 2004, 47, 1750.
[35] R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren, J. J. Klicic, D.
T. Mainz, M. P. Repasky, E. H. Knoll, M. Shelley, J. K. Perry, J. Med.
Chem. 2004, 47, 1739.
How to cite this article: Elrayess R, Abdel Aziz YM, Elgawish
MS, Elewa M, Elshihawy HA, Said MM. Pharmacophore
modeling, 3D‐QSAR, synthesis, and anti‐lung cancer
evaluation of novel thieno[2,3‐d][1,2,3]triazines targeting
EGFR. Arch Pharm Chem Life Sci. 2020;e1900108.
https://doi.org/10.1002/ardp.201900108
[36] R. Abdelhameed, M. S. Elgawish, A. Mira, A. K. Ibrahim, S. A. Ahmed,
K. Shimizu, K. Yamada, RSC Adv. 2016, 6, 20422.
[37] S. H. Ou, R. Soo, Drug Des. Dev. Ther. 2015, 9, 5641.
[38] J. K. Rho, Y. J. Choi, B.‐Y. Ryoo, I. I. Na, S. H. Yang, C. H. Kim, J. C. Lee,
Cancer Res. 2007, 67, 1163.
[39] P. Verma, A. Singh, D. N. Nthenge‐Ngumbau, U. Rajamma,
S. Sinha, K. Mukhopadhyay, K. P. Mohanakumar, M. Shaquiquzzaman,