Discovery of Novel Potent VEGFR-2 Inhibitors Exerting Significant Antiproliferative Activity against Cancer Cell Lines

Article abstract of DOI:10.1021/acs.jmedchem.7b01091

Computational and experimental studies were applied to the discovery of a series of novel vascular endothelial growth factor receptor 2 (VEGFR-2) inhibitors. Eight compounds exhibited nanomolar IC₅₀ values against VEGFR-2, and compounds 6, 19,

Full text of DOI:10.1021/acs.jmedchem.7b01091

Subscriber access provided by READING UNIV
Article
Discovery of Novel Potent VEGFR-2 Inhibitors Exerting
Significant Antiproliferative Activity against Cancer Cell Lines
Yanmin Zhang, Yadong Chen, Danfeng Zhang, Lu Wang, Tao Lu, and Yu Jiao
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01091 • Publication Date (Web): 30 Nov 2017
Downloaded from http://pubs.acs.org on November 30, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of Novel Potent VEGFR-2 Inhibitors
Exerting Significant Antiproliferative Activity
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
against Cancer Cell Lines
Yanmin Zhang a, Yadong Chen a, Danfeng Zhang b, Lu Wang b, Tao Lu a,b,c*, Yu Jiao b*
a Laboratory of Molecular Design and Drug Discovery, School of Science, China Pharmaceutical
University, 639 Longmian Avenue, Nanjing 211198, China.
b School of Science, China Pharmaceutical University, 639 Longmian Avenue, Nanjing 211198,
China.
c State Key Laboratory of Natural Medicines, China Pharmaceutical University, 24
Tongjiaxiang, Nanjing 210009, China.
ABSTRACT Computational and experimental studies were applied to the discovery of a series
of novel vascular endothelial growth factor receptor 2 (VEGFRꢀ2) inhibitors. Eight compounds
exhibited nanomolar IC50 values against VEGFRꢀ2 and compounds 6, 19, 22 and 23 showed
potent antiproliferative effects against several cell lines. Particularly, compound 23 behaved
better than FDA approved drugs, sorafenib and sunitinib, in antiproliferative activity against cell
lines related to all nine tumor types tested (GI50 values), and it was better or comparable in safety
(LC50 values). Compound 23 even demonstrated high potency on one of the drugꢀresistant cell
lines (NCI/ADRꢀRES) responsible for ovarian cancer and cell lines contributing to prostate
cancer, regarded as one of the VEGF/VEGFR pathway drugꢀresistant tumors. This compound is
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 58
1
2
3
4
5
6
likely a promising candidate for the treatment of leukemia, nonꢀsmall cell lung cancer (NSCLC),
colon cancer, ovarian cancer and breast cancer with a suitable balance of both efficacy and safety.
7
8
9
INTRODUCTION
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Angiogenesis, the formation of new blood vessels by sprouting or splitting from preꢀexisting
vessels for cell proliferation, migration, and survival1, plays a pivotal role in both normal
embryonic and adult development2ꢀ4. However, angiogenic abnormalities are observed in
numerous pathological conditions such as cancer, psoriasis, rheumatoid arthritis, inflammation,
and retinal complications.5 The specific inhibition of signal transduction via the system
containing vascular endothelial growth factor (VEGF) and its receptor (VEGFR) is a promising
approach to starve tumor cells of nutrients and thus impede tumor growth6 and metastasis7, 8. The
VEGFR family, VEGFRꢀ1 (Fltꢀ1), VEGFRꢀ2 [Flkꢀ1/kinase domain receptor (KDR)] and
VEGFRꢀ3 (Fltꢀ4), which are cell surface tyrosine kinase receptors, are located on the host
vascular endothelium, lymphatic, and hematopoietic systems, and these family members are
major proteins modulating angiogenesis by transducing VEGFꢀA, VEGFꢀB, VEGFꢀC, VEGFꢀD,
and placental growth factor (PlGF) signals to downstream pathways9, 10. Specifically, VEGFRꢀ2
is regarded as one of the central targets for discovering inhibitors against tumorꢀassociated
angiogenesis, as it is responsible for the transduction of the main proꢀangiogenic signal for
vasculature.11 Because most current VEGFRꢀ2 therapeutics are ATP competitive inhibitors, they
exert their effect by occupying the highly conserved ATP pocket.12 Moreover, smallꢀmolecule
kinase inhibitors have recently seen burgeoning growth with 28 drugs gaining approval by the
US Food and Drug Administration (FDA) for the treatment of cancers, such as advanced renal
cell carcinoma and hepatocellular carcinoma.13 Among these drugs, nine (sorafenib, sunitinib,
2
ACS Paragon Plus Environment

Page 3 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pazopanib, vandetanib, axitinib, regorafenib, cabozantinib, nintedanib, lenvatinib and apatinib)
involve VEGFRꢀ2 inhibition, demonstrating the clinical benefits of VEGFRꢀ2 inhibitors.13, 14
However, both intrinsic and acquired resistance are major contributors to transient clinical
benefits and failure of antiangiogenic drugs in the clinic.15, 16 Several mechanisms have been
proposed for the drug resistance to VEGF blockade seen in some cancer patients and these
mechanisms are not mutually exclusive. For example, (1) in established tumors, the inhibition of
VEGF can aggravate hypoxia, resulting in upregulation of the production of other angiogenic
factors or increase in tumor cell invasiveness. (2) Tumor cells with other acquired mutations can
also become hypoxia tolerant. (3) Tumor vessels covered by pericytes are generally less sensitive
to VEGF inhibition. (4) Tumor vascularization begins to include intussusception, vasculogenic
mimicry and vessel growth, differentiation of putative cancer stem cells into endothelial cells,
and existing vessel coꢀoption, etc. These changes can be rescued by recruited macrophages,
proangiogenic boneꢀmarrowꢀderived cells, or activated cancerꢀassociated fibroblasts through
producing proangiogenic factors. Regarding safety, VEGFR inhibitors all share mechanismꢀ
related toxicities, of which the most common are hypertension, proteinuria, thromboembolism,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
hemorrhage,
fistula
formation,
bowel
perforation,
and
reversible
posterior
leukoencephalopathy.16
Generally, resistance to numerous other drugs by targeting specific genetic mutations often
leads to a stable, heritable, acquired drugꢀresistant phenotype. In contrast, to combat acquiring
resistance to antiangiogenic VEGF pathwayꢀtargeting drugs, one interesting consideration is that
the evasive/acquired resistance is potentially reversible (epigenetic), not necessarily stable, and
hence presumably immutable with host endothelial cells lining new tumor blood vessels.16, 17
This may help explain why some antiꢀVEGF therapies can be administered for long periods as
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 58
1
2
3
4
5
6
7
8
9
maintenance therapy even beyond disease progression.18, 19 For example, therapeutic benefit is
observed when switching VEGFR tyrosine kinase inhibitors (TKIs) to secondꢀline alternate
antiangiogenic drugs in patients with renal cell carcinoma20. Regarding potential reversible
resistance, initial TKIs might still be effective and should be considered in future clinical trials
assessing subsequentꢀline therapies. Thus, discovery of novel VEGFRꢀ2 inhibitors with
satisfactory properties that can surmount drug resistance and decrease toxicity remain a
substantial challenge.21, 22
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In this work, computational and experimental studies were applied to the discovery of a series
of novel VEGFRꢀ2 inhibitors (Figure 1). First, BREED23, a ligand hybridization methodology
belonging to fragmentꢀbased drug design (FBDD), was conducted on the ligands from xꢀray
crystallographic structures of VEGFRꢀ2, selected hit fragments from previous work23 and the
KLIFS database24. Then, molecular docking was used to screen the generated database. Based on
of the abundance of crystal structure data available, a robust linear model for the relationship
between predicted binding free energy (Delta G, ꢁG) calculated by Molecular
Mechanics/Poisson−Boltzmann Surface Area (MM/PBSA) and enzyme inhibitory activity (such
as IC50 values) was constructed after a 20 ns molecular dynamics (MD) simulation. Then,
potential compounds selected by molecular docking were submitted for MD simulation and
scored by the linear model. Ten compounds were ultimately chosen for synthesis and biological
activity evaluation. Those that exhibited favorable enzymatic activity and antiproliferative
effects on human umbilical vein endothelial cells (HUVECs) and melanoma cell line MDAꢀMBꢀ
435 were submitted to the National Cancer Institute (NCI)ꢀ60 platform. The NCI testing
measured the growth percent in 58 cell lines from nine different tumor types, one of which is
4
ACS Paragon Plus Environment

Page 5 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
antiangiogenic drugꢀresistant prostate cancer. Finally, the kinase activity profile of the most
effective compound was investigated.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1 The workflow of for this investigation.
RESULTS AND DISSCUSSION
Binding Site Analysis and SAR Summarization
Previously, we constructed a threeꢀdimensional quantitative structure activity relationship (3Dꢀ
QSAR) model and summarized a SAR map for VEGFRꢀ2.25 Combined with molecular docking
and sitemap analysis26, the binding site of VEGFRꢀ2 can be divided into four important regions
(Figure 2). Sorafenib, for example, binds to the active site cavity of VEGFRꢀ2 in a DFGꢀout
(inactive) conformation, which enables the appropriately substituted inhibitor to penetrate into
the extended hydrophobic pocket. The Nꢀmethyl picolinamide/Nꢀmethylꢀ2ꢀpyridinecarboxamide
part binds to the Hinge Region by forming two hydrogen bonds with the backbone amide NH
and carbonyl oxygen of Cys919, which are the predominant forces that maintain inhibitor
activity. The Hydrogen BondꢀRich Region indicates where hydrogen acceptors or donors are
favorable. An effective inhibitor (here the amide group of sorafenib) will form two to three
hydrogen bonds with the side chain of Glu885 and the backbone NH of Asp1046, two residues
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 58
1
2
3
4
5
6
7
8
9
that have also been implicated as important contributors to both hydrogen bonding and
nonbonding interactions for inhibitor binding by FTMap analysis27, 28. However, we focused
instead on the two important hydrophobic regions. Hydrophobic Region І, occupied by the
central phenyl ring in sorafenib, is surrounded by several hydrophobic amino acids, including
Ala866, Leu840, Gly841 and Ile915. As shown in Figure 2, this region, probed by SiteMap
(yellow dots), can accommodate monocyclic or bicyclic ring structures, which may provide
space for additional modification. Hydrophobic Region ІI, is occupied by the 4ꢀchloroꢀ3ꢀ
(trifluoromethyl)benzene moiety and is surrounded by Ile892, Ile888, Leu889 and Val898. This
region extends to Hydrophobic Region ІІ and bypasses the DFG moiety, which is generally a
determinant of inhibitor selectivity.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2 The binding site of VEGFRꢀ2 in complex with sorafenib (PDB ID: 3EWH).
BREED
Herein, in addition to the 28 crystal structures of VEGFRꢀ214, selected hit fragments from
previous work23 and 1100 human kinaseꢀligand cocrystal complexes of various targets from the
24, 29
KLIFS database (http://www.vuꢀcompmedchem.nl)
were aligned. Because most kinases
possess similar binding sites, approaching this dilemma by combining the available data may
6
ACS Paragon Plus Environment

Page 7 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
expand the possibility of chemical space for novel inhibitor scaffolds. Figure 3a illustrates one
example, where the two known VEGFRꢀ2 inhibitors, vandetanib (IC50: 40 nM) and sorafenib
(IC50: 90 nM), (activity data from http://www.selleckchem.com/) and a hit fragment,
Hit_Scaffold_4 (Bayesian Score: 35.6), from our previous study29 were overlapped and swapped
to generate hybridized compound 6, which was synthesized and had 15.3 nM enzymatic activity
against VEGFRꢀ2. As shown in Figure 3b, the fragments of compound 6 adopt conformations
remarkably similar to the original fragments extracted from crystal structures with the two
known inhibitors and the hit fragment when the BREEDꢀgenerated model was overlaid within
the active site of the protein. The rootꢀmeanꢀsquare deviation (RMSD) values for the 6,7ꢀ
dimethoxyquinazoline group, the naphthylamine portion and the 1ꢀ(4ꢀchloroꢀ3ꢀ
(trifluoromethyl)phenyl)ꢀ3ꢀmethylurea moiety were 0.046, 0.037, and 0.054 Å, respectively,
sufficiently demonstrating the reliability of the BREED method. A total of 1531 novel
compounds were obtained after duplicate removal and all were submitted for further molecular
docking analysis (Docking_results_for_inꢀhouse_compoundꢀdatabase.zip in the Supporting
Information).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3 (a) The combinatorial approach of BREED. (b) Different colors were used to denote
the scaffolds from different input structures: the 6,7ꢀdimethoxyquinazoline group from
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 58
1
2
3
4
5
6
7
8
9
vandetanib (green), the naphthamide portion of Hit_Scaffold_4 (dark pink) and the 1ꢀ(4ꢀchloroꢀ
3ꢀ(trifluoromethyl)phenyl)ꢀ3ꢀmethylurea moiety (blue). Compound 6 (Cmpd 6) is colored cyan
and important active site residues are labeled and colored magenta.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
MD Linear Model and Binding Affinity Prediction
A set of 28 crystallographic structures (Table S1) from the Protein Data Bank (PDB)30 database
were used to construct the binding affinity prediction model. Of the crystallized ligands, 24 have
low nanomolar (nM) IC50 values (less than 100 nM) (Table S1) and they contain different
structural scaffolds that sample diverse chemical space. For each protein, a 20 ns MD simulation
was performed to ensure that dynamic equilibrium was attained (Figure 5). Snapshots were
extracted from the last 5 ns at every 10 ps interval. A total of 500 snapshots were obtained and
each was used to calculate a binding free energy, then an average of these 500 values was used.
Next, the average binding free energy was converted to binding affinity (we used IC50 values
instead) by the following formula: IC50 = exp[ꢁG/(RT)], where RT is approximately 0.59
kcal/mol. A linear model was built between the computational and experimental binding free
energies with a Pearson correlation coefficient (R) of 0.8 and a pꢀvalue of about 4×10ꢀ6 (Figure
4a), suggesting robustness of the model.
8
ACS Paragon Plus Environment

Page 9 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4 The binding affinity prediction model (a) and the average binding affinity of both
ligands from the crystal structures and the selected compounds (b) Delta G means ꢁG.
As shown in Figure 4a, this model (y = 3.3*x ꢀ11, where x represents the experimental
binding free energy (Experimental Delta G) and y is the calculated binding free energy
(Calculated Delta G) was used to predict the 100 compounds generated by BREED that were
selected after preꢀscreening by molecular docking. Ten compounds with the highest predicted
Delta G values (Table 1) were submitted for synthesis and biological activity evaluation.
Because pan assay interference compounds (PAINS) compounds are often false positives
reacting with numerous targets. They tend to react with numerous biological targets
simultaneously rather than specifically affecting one desired target31. Thus, before further
evaluation, our ten compounds were filtered through the PAINS remover
(http://cbligand.org/PAINS/)31 and all passed the filtration process. The average binding free
energy values for the 10 selected compounds were better (lower) or comparable to those of the
ligands from the crystal structures (Table S1 and Figure 4b).
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 58
1
2
3
4
5
6
7
8
9
Table 1 Enzymatic activity of selected compounds against VEGFRꢀ2 as well as glide XP score
and predicted Delta G obtained by molecular dynamics combined with MMPBSA.
Predicted
IC50
Glide XP
Score
No.
Compound
Inhibition(%)
Delta G
(nM)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(kcal/mol)
99.71
15.29
28.79
ꢀ13.50
ꢀ13.25
ꢀ17.27
ꢀ15.25
6
99.05
98.9
7
8
17.85
ꢀ13.53
ꢀ14.83
44.31
95.23
>10000
3937.00
ꢀ10.95
ꢀ10.20
ꢀ16.44
ꢀ12.96
10
11
99.79
98.74
99.5
0.75
0.92
1.88
ꢀ11.90
ꢀ11.07
ꢀ10.97
ꢀ15.35
ꢀ13.88
ꢀ12.26
19
20
21
96.5
8.70
ꢀ12.47
ꢀ13.88
ꢀ12.26
ꢀ17.14
22
23
97.78
33.36
96.75
99.24
46.41
15.97
ꢀ11.33
ꢀ
ꢀ13.85
ꢀ
Sorafenib
ꢀ
Staurosporine
* IC50 values were average of three replicates.
10
ACS Paragon Plus Environment

Page 11 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To ensure that the simulation had reached equilibrium, RMSD and the number of proteinꢀ
ligand hydrogen bonds were calculated for both the crystallized ligands and the designed
compounds. The RMSD values for both the protein (Figures 5, S1 and S3, blue) and the ligands
(Figures 5, S1 and S3, orange) in the 28 crystal structures and 10 target compounds were
calculated, respectively. In general, the protein atoms (RMSD of 1.5–4 Å) and the ligand atoms
(RMSD of 0.25–2.5 Å) reached equilibrium at approximately 2000 ps. However, among all
crystal structures, protein atoms from PDB entries 1Y6A, 2OH4, 2P2I and 3DTW exhibited
large fluctuations (Figure S1). For the ligand atoms, the crystallized ligands of PDB entries
3C7Q, 2QU6 and 4AGD fluctuated significantly (Figure S3). In Figure 5, the RMSD
fluctuation of compound 6 presented a similar trend compared to the five FDA approved
VEGFRꢀ2 inhibitors including sorafenib (IC50 = 90 nM), sunitinib (IC50 = 80 nM), axitinib (IC50
= 0.2 nM), nintedanib (IC50 = 21 nM) and lenvatinib (IC50 = 4 nM) (Table S1). Moreover, the
RMSD values for the protein atoms (PDB 3EWH) in complex with the 10 designed compounds
have lower RMSD values (2 to 3 Å) than the protein atoms for the 28 crystal complexes, except
the complex with compound 21 (Figure 5 and S3), which fluctuates from 2 to 4 Å. This
observation is likely because only the most favorable crystal structure PDB 3EWH14 was
adopted for docking and MD simulations for the designed compounds. The ligand RMSD values
(0.25–1.5 Å) of the designed compounds fluctuated less significantly (Figures 5 and S3) than
the crystallized ligands (0.25–2.5 Å) (Figures 5 and S1). Additionally, the distribution of all
Delta G values calculated from the 500 snapshots (Figure S4 for the crystal ligands and Figure
S5 for the designed compounds) followed a normal distribution, which further proved that the
MD simulation reached equilibrium and the rationality of using the average Delta G value to
build the prediction model.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 58
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5 RMSD values during the 20 ns MD for five FDA approved drugs crystallized with
VEGFRꢀ2 protein and the designed compound 6 docked to VEGFRꢀ2 protein (PDB 3EWH).
Blue stands for the fluctuation of protein atoms and orange represents the fluctuation of ligand
atoms.
Hydrogen bonding plays a vital role in protein folding, stability and function, and it is
particularly important for proteinꢀligand binding affinity. We calculated the number of
intermolecular hydrogen bonds between the protein and the compounds with respect to time. On
average, the designed compounds formed more hydrogen bonds (ranged from 1 to 6, with an
average of 5) during the MD simulation compared to the crystallized ligands (ranged from 0 to 6,
with an average of 3) (Figures 6, S2 and S3). Taken together, these data indicated that the
VEGFRꢀ2 protein in complex with the designed compounds can reach an equilibrium that was
comparable or better in terms of RMSD, and the complexes maintain more hydrogen bonds
during the simulation. Taken together, these results indicated augmented protein stability and
stronger binding of the designed compounds.
12
ACS Paragon Plus Environment

Page 13 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6 The number of hydrogen bonds formed during 20 ns MD simulations for five FDA
approved drugs crystallized with VEGFRꢀ2 as well as designed compound 6 docked to VEGFRꢀ
2 protein (PDB 3EWH).
Chemistry
The preparation of target compounds 6, 7, 8, 10 and 11 are shown on Scheme 1: 3a and 3b were
obtained by nucleophilic substitution reaction of two different phenols (2a, 2b) and 4ꢀchloroꢀ6,7ꢀ
dimethoxyquinazoline (1). Reduction of 3a and 3b by iron powder provided amino compounds
4a and 4b. Finally, 4a was reacted with different aryl isocyanates (5a, 5b) to obtain target urea
compounds 6, 7, and 8. 4b was reacted with phenylsulfonyl chloride (9a), and 1ꢀbutanesulfonyl
chloride (9b) to obtain sulfonamides 10 and 11, respectively.
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 58
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1 Synthesis of compounds 6, 7, 8, 10 and 11.
Reagents and conditions: (a) anhydrous DMF, NaH, 90 °C, 24 h; (b) Fe, NH4Cl, 70% iꢀPrOH, 85
°C, 4 h; (c) THF, rt, 5 h; (d) THF, TEA, 0 °C ꢀ rt, 1 h/pyridine, rt, 1 h.
Preparation of target compounds 19, 20, 21, 22, and 23 are depicted on Scheme 2: 4a was
reacted
with
1ꢀisopropylꢀ1Hꢀpyrazolꢀ4ꢀamine
(14),
5ꢀmethylisoxazoleꢀ3ꢀamine,
cyclopropylamine, 1ꢀ(4ꢀaminophenyl)cyclopentaneꢀ1ꢀcarbonitrile and 4ꢀ(morpholinomethyl)ꢀ3ꢀ
(trifluoromethyl)aniline (18) to obtain compounds 19, 20, 21, 22, and 23, respectively. The key
intermediate 14 was prepared by Pd/Cꢀcatalyzed reduction of 13, where 13 was obtained from
nucleophilic substitution of 4ꢀnitroꢀ1Hꢀpyrazole and 2ꢀbromopropane. The key amine 18 was
synthesized by hydrogenation of compound 17, which was obtained by nucleophilic substitution
of 4ꢀnitroꢀ2ꢀ(trifluoromethyl)benzyl bromide (16). Compound 16 was obtained from the
bromination of 1ꢀmethylꢀ4ꢀnitroꢀ2ꢀtrifluorotoluene.
14
ACS Paragon Plus Environment

Page 15 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2 Synthesis of compounds 19, 20, 21, 22 and 23. Reagents and conditions: (e) 2ꢀ
bromopropane, K2CO3, DMF, 50 °C, 0.5 h; (f) 20% Pd/C, H2, ethyl acetate, rt, 5ꢀ7 h; (g) NBS,
AIBN, 1,2ꢀdichloroethane, 90 °C, 12 h; (h) morpholine, K2CO3, rt, 16 h; (i) 20% Pd/C, H2,
EtOH, rt, 2 h; (j) (19) 4ꢀnitrophenyl carbonochloridate, DCM, RNH2 pyridine, rt, 24 h; (20, 21,
22, 23) triphosgene, DCM, TEA, RNH2, 0 °Cꢀrt, 24 h.
VEGFR-2 Kinase Activity
The inhibitory activity of the 10 target compounds was tested at 10 ꢂM and for the compounds
with inhibition ratios greater than 50 %, the VEGFRꢀ2 IC50 values were investigated. The
activity results (Figure 7 and Table 1) showed that the VEGFRꢀ2 inhibitory activities of the 10
target compounds at 10 ꢂM were excellent, where nine compounds obtained an inhibition ratio
above 90 %. Moreover, rescreening results showed that these nine compounds achieved
nanomolar VEGFRꢀ2 IC50 values. All nine compounds except compound 10 (IC50 > 10000 nM)
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 58
1
2
3
4
5
6
7
8
9
and 11 (IC50 = 3937 nM) exhibited better inhibitory activity than the positive control sorafenib
(IC50 = 46.41 nM). Moreover, the best three compounds, 19 (IC50 = 0.75 nM), 20 (IC50 = 0.92
nM) and 21 (IC50 = 1.88 nM), were more active than the other positive control staurosporine
(IC50 = 15.97 nM). Compound 22 (IC50 = 8.7 nM) also showed better inhibitory activity than
sorafenib via the introduction of a cyclopentanecarbonitrile group similar to that of apatinib,
which was approved for clinical use in lateꢀstage gastric carcinoma in China by the China Food
and Drug Administration (CFDA) in 201632. Enzymatic activity testing corroborated the binding
mode prediction, because all compounds bound to a DFGꢀout conformation and formed key
hydrogen bonds with Cys919 in the hinge region and Glu885 and Asp1046 in the hydrogen bond
favorable region (Figure 8). In addition, the bicyclic group (naphthamide or quinoline) fit
Hydrophobic Region І better than the monocyclic groups, and differentially substituted sixꢀ, fiveꢀ
or three membered rings matched well with hydrophobic region ІІ (Figure 8a).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Based on activity data, the structure activity relationship (SAR) can be summarized as follows.
First, the introduction of a naphthalene ring and a quinoline ring in Hydrophobic Region I has
little effect on the inhibitory activity; for example, compounds 6 and 8 exhibited inhibitory
activities of 15.29 and 17.85 nM, respectively. Next, Hydrophobic Region II can accommodate
hydrophobic groups of different sizes, such as the cyclopropyl substituent of compound 21 and
the substituted fiveꢀmembered ring (pyrazole or oxazole) of compounds 19 (binding mode shown
in Figure 8b) and 20, as well as the substituted phenyl group of compounds 6, 7, 8, 22 and 23.
Third, comparison of compounds 6 (IC50 = 15.29 nM), 7 (IC50 = 28.79 nM), and 23 (IC50 = 33.36
nM) indicated that 4ꢀchloro substituted phenyl groups in Hydrophobic Region II did not affect
inhibitory activity against VEGFRꢀ2, whereas removal or replacement with a hydrophilic
morpholino moiety only decreased the activity about 2ꢀfold. Finally, replacement of the urea
16
ACS Paragon Plus Environment

Page 17 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
group in the hydrogen bond acceptor or donor substituents was favored over a sulfonamide
moiety (which may reduce inhibitory activity). This trend can be seen when comparing
compound 6 (IC50 = 15.29 nM) with compound 10 (inhibition ratio of 44.31 %) and compound
11 (IC50 = 3937 nM), indicating that the hydrogen bond formation with Glu885 which is
important for inhibitor binding may have been disrupted (Figure 8c)28.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7 The doseꢀdependent enzymatic activity of the 10 target compounds against VEGFRꢀ2.
Results were average of three replicates.
Figure 8 Binding conformation of the 10 designed compounds. (a) All 10 compounds; (b)
compound 19; and (c) compound 10.
Based on the enzymatic activity data, the protein and ligand stability during the MD
simulation process for the most active compound 19 (IC50 = 0.75 nM, colored black in Figure 9)
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 58
1
2
3
4
5
6
7
8
9
and the second least active compound 11 (IC50 = 3937 nM, colored red in Figure 9) was further
evaluated. As illustrated in Figure 9, the RMSD values for the VEGFRꢀ2 protein (PDB 3EWH)
were nearly the same for the two compounds (Figure 9a), although for compound 19 the RMSF
averaged about 1.75 Å and compound 11 averaged 2.4 Å (Figure 9c). Nevertheless, compound
19 reached equilibrium around 0.5 Å, while 11 obtained stability around 1 Å (Figure 9b).
Furthermore, 19 achieved an average of 5 hydrogen bonds, while compound 11 only averaged 3
(Figure 9d). These results may help explain the huge loss in binding affinity for compound 11
compared to compound 19.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 9 The RMSD (a and b), RMSF (c) and the number of hydrogen bonds (d) of the protein
and ligands in the MD simulations for the most active compound 19 (IC50 = 0.75 nM, black) and
the second least active 11 (IC50 = 3937 nM, red). 1 nm is equal to 10 Å.
Cellular VEGFR-2 Assays
Based on their outstanding VEGFRꢀ2 inhibitory activity, eight compounds were submitted for
initial cell proliferation inhibition activity testing. Herein, using CellTiterꢀGlo (CTG) cell growth
inhibition test module, the growth inhibitory effect of the eight compounds on three cell lines
(HUVECs33, melanoma cell line MDAꢀMBꢀ43534 and epidermal squamous cell carcinoma cell
18
ACS Paragon Plus Environment

Page 19 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
line A43135) were measured. VEGFRꢀ2 overexpression was found in HUVECs and MDAꢀMBꢀ
435 cells in angiogenesisꢀrelated cancers36. HUVECs, derived from the endothelium of veins
from the umbilical cord, are the most commonly studied human endothelial cell type in
angiogenesis33. It has been proven that inhibiting VEGFꢀinduced VEGFRꢀ2 activation by small
molecules in HUVECs can inhibit subsequent angiogenesis37. Here, MDAꢀMBꢀ435 was used as
a melanoma cell line rather than a model for human breast cancer, because recent advances in
gene expression analysis revealed that the pattern of gene expression for MDAꢀMBꢀ435 more
closely resembled that of melanoma cell lines than other breast tumor lines34. Moreover, the
results from an immunofluorescence staining study showed strong VEGFRꢀ2 staining in MDAꢀ
MBꢀ435 tumors36. Epidermal squamous cell carcinoma cell line A431 is an epidermal growth
factor receptor (EGFR) overexpressing cell, but it can also lead to angiogenesis, which was used
as a negative control for comparison38.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
As shown in Table 2, all compounds showed good inhibitory effects against HUVECs with
compounds 19, 22 and 23 reaching inhibition ratios of 93.58, 92.48 and 97.11 %, respectively,
which are superior to positive control sorafenib (84.66 %). At 82.89 % inhibition, compound 6
exhibited comparable inhibitory activity to sorafenib. The inhibitory effects on the MDAꢀMBꢀ
435 cell line for compounds 6, 22 and 23 were 95.06, 96.90 and 98.36 %, respectively, also
outperforming sorafenib (91.89 %). As expected, this series of compounds did not exhibit good
inhibitory effects against the A431 cell line, which has EGFR overexpression rather than
VEGFRꢀ2. For the A431 cell line, compounds 22 (49.05 %) and 23 (57.71 %), which obtained
the highest inhibitory activity, were still weaker than sorafenib (68.94 %). Compounds with
cellular inhibition rates greater than 90 % were further tested to obtain IC50 values. As depicted
in Figure 10 and Table 2, the IC50 values for compounds 6, 19, 22 and 23 against HUVECs
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 58
1
2
3
4
5
6
7
8
9
were 2.97, 0.63, 3.62 and 0.63 ꢂM, respectively, which is comparable with sorafenib (1.49 ꢂM).
The IC50 values for compounds 6, 22 and 23 against MDAꢀMBꢀ435 were 3.79, 3.01 and 2.36 ꢂM,
respectively, also comparable with sorafenib (2.11 ꢂM). These results demonstrated that the
designed compounds exhibited excellent cellular inhibitory activity, which most likely resulted
from the inhibition of VEGFRꢀ2 and were promising hits for developing antitumor candidates.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 10 Cellular activity against two cell lines. (a) HUVECs and (b) MDAꢀMBꢀ435 cell line.
Results were from three replicates.
Table 2 Cellular inhibitory rates of eight compounds against HUVECs, MDAꢀMBꢀ435 and
A431 cell lines as well as IC50 values for four compounds for two cell lines (HUVECs and
MDAꢀMBꢀ435).
Cell inhibitory activity
Cell inhibition rate at 10 ꢂM
(IC50 values, ꢂM)
No.
HUVECs
82.89%
66.8%
MDAꢀMBꢀ435
95.06%
A431
HUVECs
MDAꢀMBꢀ435
14.95%
ꢀ4.51%
20.96%
49.87%
6
7
2.97±0.18
ND
3.79±0.93
ND
24.54%
51.45%
93.58%
44.29%
ND
ND
8
27.71%
ND
19
0.63±0.05
20
ACS Paragon Plus Environment

Page 21 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
61.89%
63.9%
0.23%
1.48%
ꢀ1.52%
15.13%
49.05%
57.71%
68.94%
ND
ND
20
ND
ND
21
22
92.48%
97.11%
84.66%
96.90%
98.36%
91.89%
3.62±0.17
0.63±0.02
1.49±0.08
3.01±0.03
2.36±0.09
2.11±0.04
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
23
Sorafenib
*The data in bold are compounds with cellular inhibition greater than 80 %, which were tested
for corresponding IC50 values. Results were from three replicates.
The anti-tumor spectrum of representative compounds
Based on cellular activity, four compounds 6, 19, 22 and 23 exhibited good inhibitory activity
against HUVECs and the MDAꢀMBꢀ435 cell line. Thus, to expand the spectrum of tumor cells
tested, these four compounds were submitted to the NCI for testing in a panel of 58 cell lines
(termed NCIꢀ60). These cell lines correspond to nine tumor types including leukemia, NSCLC,
colon cancer, central nervous system (CNS) cancer, melanoma, ovarian cancer, renal cancer,
prostate cancer and breast cancer (shown Figures S6–S10). The cell lines within the NCIꢀ60
panel of human tumor cell lines are some of the most extensively characterized cell lines in
broad laboratory use and the 58 cell lines tested in this work are VEGFRꢀ2 expressing or
overexpressing ones.
As seen in Table 3 and Figure 11, compound 6 had moderately potent antiproliferative
activity against melanoma, renal, breast and CNS cancers. It exhibited about 50 % cell growth
for at least one of the cell types that may contribute to those tumors, which means that more than
50 % of the cells were inhibited. Compound 19 did not show any antiproliferative activity to all
tested cell lines, because all cell growth proportions were higher than 70 %. Compound 22
exhibited excellent antiproliferative activity against NSCLC [cell growth: A549/ATCC (25.6 %),
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 58
1
2
3
4
5
6
7
8
9
HOPꢀ92 (ꢀ39.8 %) and NCIꢀH460 (19.3 %)], CNS cancer [cell growth: SFꢀ268 (27 %), SFꢀ295
(13.7 %) and U251 (ꢀ49.3 %)], ovarian cancer [cell growth: OVCARꢀ4 (ꢀ1.3 %), OVCARꢀ8
(0.6 %)], renal cancer [cell growth: 786ꢀ0 (ꢀ40.7 %), ACHN (29.3%), RXF 393 (15 %), SN12C
(21.3 %) and TKꢀ10 (24.1 %)] and breast cancer [cell growth: HS 578T (29.6 %)]. Compound 23
demonstrated the best antiproliferative activity among all four compounds. It exhibited very
potent antiproliferative activity on 51 out of 58 cell lines, where the growth percent was less than
50 % (with 11 less than 0) and the average growth percent for all 58 cell lines was approximately
19.2 %, indicating that 23 can inhibit about 80 % of cell growth on all 51 cell lines on average.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 11 Doseꢀdependent growth inhibition curves for compound 23 on 58 cell lines conducted
by NCI. A growth percent of 100 is the growth seen in untreated cells, while growth percent of 0
corresponds to no net growth during the assay and growth percent of ꢀ100 indicates the time
when all cells are killed. Three endpoints are routinely calculated: (1) GI50, the drug
concentration yielding a growth percent of 50 (i.e., 50% growth inhibition); (2) TGI (total
22
ACS Paragon Plus Environment

Page 23 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
growth inhibition), the concentration resulting in a growth percent of 0; (3) LC50, the
concentration when a growth percent of ꢀ50, or lethality in 50% of the starting cells is shown.
These endpoints are illustrated for one of the melanoma cell lines MALMEꢀCM (red open
diamond).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Therefore, compound 23 was selected to determine GI50 values at five concentrations with 1
log dilutions from 10ꢀ4 M to 10ꢀ8 M (Figure 11). Three endpoint values, GI50, TGI and LC50,
were calculated (Figure 11). As illustrated in Figure 11, compound 23 exhibited doseꢀdependent
inhibition behavior against all 58 cell lines, corroborating all in vitro tests. According to Table 3,
Figures 11 and S9, compound 23 displayed excellent antiproliferative activity, with the GI50
values against all 58 cell lines at less than 5.25 ꢂM and TGI less than 58.9 ꢂM. Specifically,
compound 23 exhibited GI50 values less than 200 nM against seven cancer cell lines including
leukemia [cell lines Kꢀ562 (GI50 < 10 nM) and MOLTꢀ4 (GI50 = 190 nM)], NSCLC [cell line
NCIꢀH226 (GI50 < 70 nM)], colon cancer [cell line KM12 (GI50 < 200 nM)], ovarian cancer [cell
line OVCARꢀ4 (GI50 < 120 nM)], renal cancer [cell lines A498 (GI50 < 50 nM) and UOꢀ31 (GI50
< 200 nM)].
Table 3 Cellular growth inhibition values of compound 23 on 58 cell lines (performed by NCI).
a
b
c
a
b
c
GI50
TGI
LC50
GI50
TGI
LC50
Tumor
type
Tumor
type
Cell line
Cell line
(ꢂM)
1.17
0.38
<0.01
0.19
0.39
0.69
(ꢂM)
40.74
2.14
1.12
1.82
2.95
5.37
(ꢂM)
>100
>100
95.50
>100
>100
>100
(ꢂM) (ꢂM)
(ꢂM)
26.92
33.88
>100
42.66
16.98
47.86
CCRFꢀCEM
HLꢀ60(TB)
Kꢀ562
M14
4.57
3.39
0.95
0.37
MDAꢀMBꢀ435
SKꢀMELꢀ2
SKꢀMELꢀ28
SKꢀMELꢀ5
UACCꢀ257
1.48 13.80
Leukemia
Melanoma
MOLTꢀ4
RPMIꢀ8226
SR
4.79
2.69
3.98
0.89
0.45
0.79
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 58
1
2
3
4
5
6
7
8
9
A549/ATCC
EKVX
10.72
6.46
81.28
67.61
91.20
>100
>100
93.33
UACCꢀ62
IGROV1
OVCARꢀ3
OVCARꢀ4
OVCARꢀ5
OVCARꢀ8
NCI/ADRꢀ
RES
3.47
3.63
9.55
18.62
>100
42.66
0.55
0.45
0.32
0.07
0.62
0.56
0.60
0.52
1.00
0.12
HOPꢀ92
7.76
Nonꢀsmall
cell lung
cancer
NCIꢀH226
NCIꢀH23
NCIꢀH322M
13.80
3.98
10.47 35.48
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Ovarian
cancer
1.07 10.47
>100
>100
15.14
9.55
0.40
0.76
NCIꢀH460
1.41
4.27
25.70
58.88
>100
NCIꢀH522
COLO 205
HCCꢀ2998
HCTꢀ116
HCTꢀ15
HT29
3.80
3.39
SKꢀOVꢀ3
786ꢀ0
0.95
5.37
0.52
1.62
7.08
1.86
5.89
37.15
26.92
30.90
14.13
29.51
13.18
35.48
0.31
0.38
5.25
0.79
1.00
0.45
0.20
1.10
0.46
1.23
1.58
>100
>100
75.86
7.08
0.23
0.79
0.05
0.32
1.05
0.29
0.38
0.20
20.89
2.57
A498
ACHN
Colon
cancer
Renal
5.13
28.18
>100
>100
>100
79.43
30.90
52.48
RXF 393
SN12C
cancer
13.18
2.51
KM12
TKꢀ10
SWꢀ620
SFꢀ268
11.75
15.85
6.03
UOꢀ31
10.23 38.02
Prostate
Cancer
PCꢀ3
1.78 12.59
>100
SFꢀ295
DUꢀ145
MCF7
1.55 13.18 44.67
SFꢀ539
19.50
10.96
7.94
0.33
>100
CNS
MDAꢀMBꢀ
231/ATCC
HS 578T
BTꢀ549
cancer
SNBꢀ19
2.57
14.13
43.65
1.15
42.66
Breast
cancer
SNBꢀ75
U251
1.05
1.05
1.10
0.30
13.80
3.98
2.82
1.55
38.90
27.54
7.24
1.12
1.15
0.25
0.23
9.33
7.41
4.27
2.82
>100
34.67
>100
72.44
LOX IMVI
Tꢀ47D
Melanoma
MALMEꢀ3M
5.75
MDAꢀMBꢀ468
a
b
The description of GI50, LC50, cTGI are the same as Figure 11. GI50 values less than 1 ꢂM or
LC50 values higher than 100 ꢂM are shown in bold.
In addition, the NCIꢀ60 results of compound 23 were compared with the results for two FDA
24
ACS Paragon Plus Environment

Page 25 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
approved VEGFRꢀ2 drugs sorafenib and sunitinib, which were downloaded from
https://dtp.cancer.gov/discovery_development/nciꢀ60/. Sorafenib, a RafꢀVEGFRꢀ2 dual inhibitor,
is now used for renal cell carcinoma, hepatocellular carcinoma and thyroid cancer treatments16.
Sunitinib39 is used for the treatment of renal cell carcinoma, pancreatic neuroendocrine tumors,
and gastrointestinal stromal tumors to complement intolerance to imatinib16. The comparison of
the mean endpoint values for all cell lines to each tumor is shown in Figure 12 and the original
GI50, TGI and LC50 values for each cell line are displayed in Figures S11-S13, respectively.
Based on Figure 12, although only the renal cancer cell line that is approved for both drugs was
included in the NCIꢀ60, we observed that the GI50 values of compound 23 were 9.5ꢀ and 6.8ꢀfold
better (lower) than those of sorafenib and sunitinib, respectively. The TGI values were both 3.2ꢀ
fold lower than the two marketed drugs. In addition, the LC50 values were 2 and 1.3ꢀfold worse
(higher) than these drugs, respectively, indicating that compound 23 was more potent in the cell
lines responsible for renal cancer, and its safety was comparable with that of sorafenib and
sunitinib for clinical use in renal cancer. For the other eight tumor types, compound 23 exhibited
higher potency against the cell lines and was more safe or comparable to sorafenib and sunitinib.
Specifically, its GI50 values were at least 3.9–9.5ꢀfold and 5–7.7ꢀfold better than sorafenib and
sunitinib, respectively, for leukemia, NSCLC, ovarian cancer and breast cancer. For colon cancer,
CNS cancer, melanoma and prostate cancer, the GI50 of compound 23 was about 1.6–2.9ꢀfold
and 1.1–3.7ꢀfold better than that of sorafenib and sunitinib, respectively. Among all cancers,
renal and ovarian cancers are VEGF/VEGFR pathway inhibitorꢀsensitive (such as VEGF
antibodies and VEGFRꢀ2 TKIs) tumors, while NSCLC, breast cancer and colon cancer are only
partly sensitive to VEGFR inhibitors.16 Among the cell lines tested, NCI/ADRꢀRES is a
multidrugꢀresistant cancer cell line for ovarian cancer40. Nevertheless, compound 23 exhibited
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 58
1
2
3
4
5
6
7
8
9
very strong antiproliferative activity with a GI50 of about 0.76 ꢂM, which is 3.3ꢀ and 6.9ꢀfold
better than sorafenib and sunitinib, respectively (Figure S11), demonstrating that this compound
may inhibit some drugꢀresistant cell lines.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 12 Comparison of the GI50, TGI and LC50 values for compound 23 (black), sorafenib (red)
and sunitinib (blue). The yꢀaxis is the log value of the endpoints and a difference of 1 indicates a
10ꢀfold difference in the endpoint values.
Studies showed that increased expression of VEGF/VEGFRꢀ2 in leukemic blasts and
correlation with angiogenesis in the bone marrow of acute myeloid leukemia (AML) patients has
been observed41. VEGF is also a major contributor to the growth of CNS tumors, as elevated
expression of VEGF is one of the factors responsible for the virulent nature of these tumors.
Thus, interruption of VEGFR signaling, specifically VEGFRꢀ2, may effectively suppress tumor
growth in leukemia and CNS cancer42, 43. However, the expression of VEGFRꢀ2 in melanoma
26
ACS Paragon Plus Environment

Page 27 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
cells is controversial. Several studies have shown that high quantities (78ꢀ89%)44, 45 or at least
half of melanoma cells46 express VEGFRꢀ2. Others indicated that relatively few (<10%)
melanoma cells expressed detectable levels of VEGFRꢀ247. Nevertheless, sorafenib, a dual
inhibitor of BRaf and VEGFRꢀ2 with promise in other cancers driven by VEGF, was relatively
ineffective when used as a single agent in melanoma. Despite these studies casting doubt on
VEGFRꢀ2 as an effective target in melanoma, the identification of VEGFRꢀ2 positive melanoma
cell lines still presents an opportunity to personalize therapy using antibodies and other TKIs that
specifically target VEGFRꢀ2ꢀdependent signaling. All data implied that the inhibition of the cell
lines by compound 23 probably stems from its inhibition of VEGFRꢀ2 rather than an offꢀtarget
effect. Nevertheless, because compound 23 was a multikinase inhibitor, which we will discuss
later, the inhibition of the cell lines may also benefit from activity at other targets. For example,
sorafenib is the only compound used for the treatment of hepatocellular carcinoma among the
VEGFRꢀ2 related approved drugs. It is probably because of positive clinical trial results in
hepatocellular carcinoma, which may be due to its BRaf inhibitory activity as the overexpression
of BRaf was indeed frequently observed in hepatocellular carcinoma. However, further efforts to
understand the complete role of sorafenib in the treatment of hepatocellular carcinoma and the
cell antiproliferative activity of compound 23 are still necessary.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
As the second leading cause of male cancer death in developed countries, prostate cancer
belongs to one of the VEGF/VEGFR pathway inhibitor resistant cancers16. The expression of
VEGFRs in endothelial cells isolated from prostate tumors has confirmed this finding. However,
though sunitinib and sorafenib showed similar antiꢀangiogenic and cytotoxic effects on normal
endothelial cells, sunitinib affected the proliferation, survival and motility of prostate tumors,
whereas sorafenib only showed a minor effect. Both medications inhibited VEGFRꢀ2
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 58
1
2
3
4
5
6
7
8
9
phosphorylation of prostate cancer endothelial cells, while they differentially modulated Akt
phosphorylation, as no inhibitory effect of sorafenib was observed on Akt activation48. The
combined treatment with Casodex reverted the observed resistance to sorafenib both on cell
viability and on Akt activation, whereas it did not modify the response to sunitinib. Thus, the
mechanism for the activity of sunitinib on prostate cancer remains unknown. Compound 23
exhibited about 1.6ꢀ and 1.5ꢀfold better GI50 than sorafenib and sunitinib on prostate cancer,
respectively (Figure 12). Compound 23 TGI and LC50 were similar to that of those two FDAꢀ
approved drugs. Specifically, the LC50 value of compound 23 was 1.7ꢀfold of sunitinib, showing
a safer effect. However, its inhibition of Akt was not obvious (less than 20% on VEGFRꢀ2
inhibition at 10 ꢂM). Nevertheless, the resistance of prostate cancer to VEGF inhibitors is more
likely related to redundancy in angiogenic pathways. A large amount of data indicates that
fibroblast growth factor (FGF) and its family of receptors (FGFRs) were observed in the
transformation and angiogenesis of prostate cancer, and several earlyꢀphase clinical trials are
assessing FGFR TKIs in this disease16. Thus, there is a possibility that compound 23 works in a
similar way, as it showed higher than 96% inhibition on FGFRꢀ1/ꢀ2/ꢀ3/ꢀ4. Therefore, we cannot
jump to any conclusions about whether this compound is a powerful prostate cancer hit, but it
still holds promise for further evaluation as a potential candidate like sunitinib48.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Kinome spectrum screening.
To further investigate compound efficacy, compound 23 was evaluated at 0.5 ꢂM against the
Reaction Biology Corporation broad kinase spectrum panel containing 349 therapeutically
important kinases (Figure 13). The kinome phylogenetic tree was generated using
KinomeRender49. Figure 13a illustrates that compound 23 was a potent multiꢀtargeted kinase
inhibitor, particularly for tyrosine kinase (TK) branch including targets like members of the
28
ACS Paragon Plus Environment

Page 29 of 58
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
VEGFR, Abl, DDR, EPHA, FGFR, Flt, and PDGFR families as well as the STE family
including KHS, TNIK and HCK (Figure 13b), TAK1 of the TKL family and SLK of the CAMK
family. The kinase inhibition profile of compound 23 was similar to that of FDA approved
sunitinib21 (Figure 13). Through a multiple sequence alignment, we discovered that the targets
such as Src, PDGFR and FGFR for which compound 23 exhibited high inhibition also had higher
sequence identity. About twoꢀthird (38 out of 56) of the targets with inhibition greater than 90%
were found in the 105 targets exhibiting high sequence identity (greater than 30%) with VEGFRꢀ
2 (Table S2). This thus indicated that the selectivity profile can be determined by sequence
identity. This was consistent with a comparison of multiꢀtarget drugs that inhibited target pairs in
the human kinome, particularly the receptor tyrosine kinase family, with combination drugs that
are able to inhibit targets of distant homology relationship50. Compound 23, a multiꢀtargeted
kinase inhibitor, is likely to inhibit multiple targets simultaneously, which may overcome any
resistance mechanism that involves bypass or restoration of persistent mitogenꢀactivated protein
kinase (MAPK) activation. However, compound 23 may also impart unwanted offꢀtarget side
effects, which necessitates further optimization and compound evaluation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
29
ACS Paragon Plus Environment