Scaffold-based selective SHP2 inhibitors design using core hopping, molecular docking, biological evaluation and molecular simulation

Article abstract of DOI:10.1016/j.bioorg.2020.104391

PTPN11 (coding the gene of SHP2), a classic non-receptor protein tyrosine phosphatase, is implicated in multiple cell signaling pathway. Abnormal activation of SHP2 has been shown to contribute to a variety of human diseases, including Juvenile myelomonocytic leukemia (JMML), Noonan syndrome and tumors. Thus, the SHP2 inhibitors have important therapeutic value. Here, based on the compound PubChem CID 8,478,960 (IC₅₀ = 45.01 μM), a series of thiophene [2,3-d] pyrimidine derivatives (IC₅₀ = 0.4–37.87 μM) were discovered as novel and efficient inhibitors of SHP2 through powerful “core hopping” and CDOCKER technology. Furthermore, the SHP2-PTP phosphatase activity assay indicated that Comp#5 (IC₅₀ = 0.4 μM) was the most active SHP2 inhibitor. Subsequently, the effects of Comp#5 on the structure and function of SHP2 were investigated through molecular dynamics (MD) simulation and post-kinetic analysis. The result indicated that Comp#5 enhanced the interaction of residues THR357, ARG362, LYS366, PRO424, CYS459, SER460, ALA461, ILE463, ARG465, THR507 and GLN510 with the surrounding residues, improving the stability of the catalytic active region and the entrance of catalytic active region. In particular, the Comp#5 conjugated with residue ARG362, elevating the efficient and selectivity of SHP2 protein. The study here may pave the way for discovering the novel SHP2 inhibitors for suffering cancer patients.

Full text of DOI:10.1016/j.bioorg.2020.104391

Bioorganic Chemistry 105 (2020) 104391
Contents lists available at ScienceDirect
Bioorganic Chemistry
journal homepage: www.elsevier.com/locate/bioorg
Scaffold-based selective SHP2 inhibitors design using core hopping,
molecular docking, biological evaluation and molecular simulation
a
,
1
a,
1,
*
a
b
a
a
a
Wei-Ya Li , Ying Ma
, Hao-Xin Li , Xin-Hua Lu , Shan Du , Yang-Chun Ma , Liang Zhou ,
a,*
Run-Ling Wang
a
Tianjin Key Laboratory on Technologies Enabling Development of Clinical Therapeutics and Diagnostics (Theranostics), School of Pharmacy, Tianjin Medical University,
Tianjin 300070, China
b
New Drug Research & Development Center of North China Pharmaceutical Group Corporation, National Microbial Medicine Engineering & Research Center, Hebei
Industry Microbial Metabolic Engineering & Technology Research Center, Shijiazhuang 050015, Hebei, China
A R T I C L E I N F O
A B S T R A C T
Keywords:
PTPN11 (coding the gene of SHP2), a classic non-receptor protein tyrosine phosphatase, is implicated in multiple
cell signaling pathway. Abnormal activation of SHP2 has been shown to contribute to a variety of human dis-
eases, including Juvenile myelomonocytic leukemia (JMML), Noonan syndrome and tumors. Thus, the SHP2
SHP2
Inhibitor
Core hopping
MD simulation
Phosphatase activity assay
inhibitors have important therapeutic value. Here, based on the compound PubChem CID 8,478,960 (IC50
5.01 M) were discovered as novel
M), a series of thiophene [2,3-d] pyrimidine derivatives (IC50 = 0.4–37.87
and efficient inhibitors of SHP2 through powerful “core hopping” and CDOCKER technology. Furthermore, the
SHP2-PTP phosphatase activity assay indicated that Comp#5 (IC50 = 0.4 M) was the most active SHP2 in-
=
4
μ
μ
μ
hibitor. Subsequently, the effects of Comp#5 on the structure and function of SHP2 were investigated through
molecular dynamics (MD) simulation and post-kinetic analysis. The result indicated that Comp#5 enhanced the
interaction of residues THR357, ARG362, LYS366, PRO424, CYS459, SER460, ALA461, ILE463, ARG465,
THR507 and GLN510 with the surrounding residues, improving the stability of the catalytic active region and the
entrance of catalytic active region. In particular, the Comp#5 conjugated with residue ARG362, elevating the
efficient and selectivity of SHP2 protein. The study here may pave the way for discovering the novel SHP2 in-
hibitors for suffering cancer patients.
1
. Introduction
JMML is a fatal childhood disease clinically in which JMML cells
which was manifested as rapid organ enlargement and cachexia and
bone marrow failure.
The PTPN11, encoding tyrosine phosphatase SHP2 protein, was
found to be ubiquitous in JMML[5,6]. Interestingly, the SHP2 phos-
phatase played a completely active role in receptor kinase-initiated
signal transduction, especially in the Ras pathway. Although the path-
ogenesis of JMML was believed to be the Ras over-activation, Ras effect
or similar pathways have also been implicated in the disease, like the
PI3K/Akt signaling pathway[7]. Previous research showed that SHP2
was widely expressed in the cytoplasm downstream of multiple receptor
tyrosine kinases and was implicated in many oncogenic cells signaling
hold the capabilities to infiltrate non-hematopoietic organs, impair their
functions, and even develop into acute myeloid leukemia [1,2]. The
main features of JMML are reflected in the over-production of myelo-
monocytic cells, hepatosplenomegaly, thrombocytopenia, progressive
anemia and high fetal hemoglobin levels[3]. JMML has a poor prog-
nosis, most of which have a survival period less than 2 years[4]. Note-
worthy, due to the heterogeneity of the course of the disease, about 1/3
of the children died within a few months regardless of the treatment,
Abbreviations: SHP2, protein tyrosine phosphatase-2; MD simulation, molecular dynamic; JMML, Juvenile myelomonocytic leukemia; PCA, principal component
analysis; pNPP, p-nitrophenyl phosphate; RMSD, root mean square deviation; NVT/NPT, constant number of particles, volume/pressure, and temperature; ADMET,
absorption, distribution, metabolism, excretion, toxicity; DCCM, dynamic cross correlation maps; RMSF, root mean square fluctuation; SPC, single-point charge; DS,
Discovery Studio; MM-PBSA, molecular mechanics Poisson Boltzmann surface area; RING, residue interaction network generator.
*
Corresponding authors.
E-mail addresses: maying@tmu.edu.cn (Y. Ma), wangrunling@tmu.edu.cn (R.-L. Wang).
These authors contributed equally to this work.
1
https://doi.org/10.1016/j.bioorg.2020.104391
Received 28 April 2020; Received in revised form 7 October 2020; Accepted 15 October 2020
Available online 21 October 2020
0
045-2068/© 2020 Elsevier Inc. All rights reserved.

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
cascades[8,9]. It maintained the activation of Ras and regulates PD-1
and BTLA to inhibit the immune receptor pathway, as well as played a
vital role in the PI3K-AKT, JAK-STAT and NF-κB signaling pathways
2. Materials and methods
2.1. Virtual screening and molecular docking based on the core hopping
method
[
10,11]. In summary, accumulated biochemical studies demonstrated
that SHP2 regulated the growth factor receptor signaling, and was
considered as a potential target, and thus, SHP2 has key implications for
the development and treatment of JMML and other SHP2-related
cancers.
The drug-like database (subset of ZINC database) was used to screen
the lead compound based on the conformation of the active site in SHP2
[16]. The “Core Hopping” approach was then preformed to modify the
lead compound.
In recent years, PTP phosphatase inhibitors have been widely applied
in research. Compound II-B08, as a reversible inhibitor of SHP2 (IC50 =
The primary characteristic of the Core Hopping was to utilize the
principle of bioisosteres to replace the inefficient or ineffective groups in
the template ligands, aiming at reducing the toxic and side effects and
improving the curative effect [17,18]. In the research, the Core Hopping
method was applied during the molecular docking procedure to receive
molecules with good activity, including the improvement of binding
affinity, the increase/decrease of water solubility, the improvement of
Pharmacokinetics and pharmacokinetics [19]. In general, the core
hopping process could be divided into four major steps. In the first step,
the possible points where the core attached to the scaffold were defined.
It was performed in the module of “Template combinatorial definition
file”. In the second step, the Receptor Preparation panel was employed
to define “the receptor grid file”. In detail, the original state of SHP2
complexed with JZG (PDB ID: 3O5X) interaction was kept to determine
the active site, which would interfere with the search space of docking
algorithm. The protocol “from current selection” was used to determine
the binding position (spherical box). In the third step, the cores con-
nected to the scaffold was prepared using a fragment database from
ZINC [19,20]. In the final step, the entire molecular structures con-
structed from the core and scaffold were submitted to the “Protocore
Docking” module to align and dock with the SHP2 receptor. The final
molecules with better docking scores than the original molecule was
kept [21].
5
.5
μmol/L), could inhibit the activation of ERK1/2 stimulated by
growth factors (such as EGF), and hinder the growth of NSCLC cells in
vitro and in vivo[12]. At the same time, the compound has an in vivo
inhibitory effect on mast cell leukemia model. Indole salicylic acid SHP2
inhibitor 11a-1 blocked growth factor-mediated activation of ERK1/2
and AKT, and exhibited antiproliferative activity in lung cancer, breast
cancer, and leukemia cell lines[13]. Furthermore, the compound PHPS1
effectively inhibits the activation of ERK1/2 by the leukemia-related
SHP2 mutant SHP2-E76K, and blocks the non-adherent growth of
various human tumor cell lines[14,15]. The results indicated that the
application of PTP inhibitors will play an important role in future cancer
research.
In this study, the lead compound PubChem CID 8,478,960 (IC50
=
4
5.01 M) was obtained by searching PubChem database. The Core
μ
Hopping method and CDOCKER technology were employed to obtained
a variety of thiophene [2,3-d] pyrimidine derivatives by modifying the
polar head and hydrophobic tail of the PubChem CID 8478960. The PTP
phosphatase activity assay showed that the compound 2-(7,7-difluoro-4-
oxo-5,6,7,8-tetrahydrobenzo[4,5]thieno[2,3-d]pyrimidin-3(4H)-yl)-N-
(
2,4-difluorophenyl)acetamide (Comp#5) was identified as the most
effective inhibitor of SHP2 with an IC50 value of 0.4 M. Furthermore,
μ
the ADMET prediction indicated that Comp#5 was consistent with the
rule of drug synthesis. Afterward, MD simulation and post-kinetic
analysis were conducted to study the conformational changes caused
by the interaction between Comp#5 and SHP2, and to clarify the mo-
lecular mechanism of Comp#5 inhibiting SHP2 activity. With Comp#9
All compounds obtained by core hopping process were pretreated by
LigPre module to generate possible states. Afterwards, the pose of each
compound was responsible to dock into the active site of the SHP2
protein [22]. Ultimately, discovery studio v3.5 (DS v3.5) visualizer was
applied to analyze the interactions between the inhibitor and the SHP2
protein.
(
IC50 value > 100 M) as a negative control, the SHP2-Comp#9 system
μ
was also subjected to MD simulation and a series of post-analysis.
Fortunately, Comp#5 properly attached to the active pocket of SHP2
protein. The interactions between Comp#5 and the residue ARG362
located in the entrance of catalytic active region improved the efficiency
and even the selectivity of the SHP2 inhibitors. While, Comp#9 was far
from the active pocket of SHP2 and formed weak interactions with key
amino acids. In the RMSF analysis, the fluctuations in the β5-β6 loop, P-
loop and Q-loop region of the SHP2 protein were larger than those in the
SHP2-Comp#5 protein, indicating that these regions were more stable in
the SHP2-Comp#5 protein. Nevertheless, the fluctuations of these re-
gions in SHP2-Comp#9 system were not significantly different from
those in SHP2-without ligands system. In addition, DCCM analysis
showed that the β5-β6 loop region and the WPD-loop region, the P-loop
region and the β5-β6 loop region, the Q-loop region and the WPD-loop
region were weakly correlated with each other in the SHP2-without li-
gands system and SHP2-Comp#9 system, respectively. However, the
correlations between amino acids were heightened in SHP2-Comp#5
protein. RIN analysis showed that there were slight interactions be-
tween amino acids in the SHP2-without ligands system and SHP2-
Comp#9 system, whereas the residues THR357, ARG362, LYS366,
PRO424, CYS459, SER460, ALA461, ILE463, ARG465, THR507 and
GLN510 formed hydrogen bonds or van der Waals interactions with
surrounding amino acids in SHP2-Comp#5. The results suggested that
the Comp#5 enhanced the interactions of amino acids within the cata-
lytic activity region (P-loop and Q-loop) and the entrance of catalytic
active region (β5-β6 loop) in SHP2-without ligands protein, improving
the stability of the system structure. Therefore, it could be concluded
that Comp#5 inhibited the activity of SHP2 protease.
2.2. Chemistry
Unless otherwise specified, all the reagents were purchased from
commercial suppliers and could be used without further purification.
The TLC were used to monitored reaction courses on silica gel precoated
F254 Merck plates. Developed plates were examined with UV lamps
1
13
(254 nm). The physical, analytical and spectral data ( H NMR, C NMR
and ESI-MS) were employed to characterized the intermediates and all
the new compounds. The solvents were DMSO‑d and CDCl .
6
3
2.3. SHP1/SHP2 phosphatases activity assay
Purified SHP1 full-length protein and PTP domain protein by GST-
tag Protein Purification Kit and Ni-NTA-Sefinosen Column were used
as enzymes, respectively. It was determined that p-nitrophenyl phos-
phate (pNPP) was tested as substrates for phosphatase activity assay
[23]. In this assay, protease dephosphorylate with pNPP to generate
PNP, which had specific fluorescence at 405 nm. Briefly, Purification of
recombinant SHP1 full-length protein and SHP2-PTP domain protein
(0.4
μ
g) in 60 L buffer containing 50 mM NaCl, 25 mM Tris-HCl (pH
μ
7.0), 1 mM dithiothreitol (DTT), 2.0 mM EDTA, and 0.05% Tween-20
and test compounds were mixed, and then incubated under the condi-
◦
tion of 37 C. Detect A4 value after 2 min. Afterwards, the pNPP was
05
◦
maintained at a concentration of 2 mM and incubated at 37 C for 30
min. The reaction was stopped by Add 50 L 3 M NaOH to the reactant
μ
and chilled it quickly on ice to stop the reaction. The OD
measured.
4
05
was then
2

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
2
.4. Fluorescence titrations assay
divided into polar contribution and non-polar contribution.
.7. ADMET analysis
ADMET stands for the five major characteristics of pharmacokinetics:
Fluorescence quenching experiment referred to the physical or
2
chemical reaction process between fluorescent substance and solvent
molecule that lead to the change of fluorescence intensity or related
change of excitation peak or fluorescence peak. To confirm the binding
degree of the inhibitor to PTP domain protein, fluorescence titration was
employed to determine whether the binding of the compound changed
the fluorescence of PTP domain protein. In this experiment, the purified
PTP protein was diluted into the reaction solution and maintained at a
protein concentration of 0.2 mg/mL. Titration was then performed by
changing the concentration of the inhibitor by a gradient. Fluorescence
spectra were recorded with Synergy 2 purchased by BioTek. The fluo-
rescence contribution of the inhibitor as background fluorescence will
be subtracted.
absorption, distribution, metabolism, excretion, and drug toxicity. The
pharmacokinetic properties of drugs in the human body were assessed
through the module of “ADMET Descriptions” in the DS v3.5, including
human intestinal absorption (HIA), blood–brain barrier (BBB), water
solubility, plasma protein binding (PPB), cytochrome P450 CYP2D6
binding and liver toxicity. Through ADMET screening, compounds
conforming to drug potency would be obtained, which reduced the loss
in drug discovery and development.
2
.8. Principal component analysis
2
.5. Molecular dynamic simulation
The Principal component analysis (PCA analysis), a technique for
reducing data complexity and extracting collaborative motion, was used
to reveal the movements of structures[34]. In general, the first few
principal components (PCs) were used to describe the most important
slow-mode biological functional movements of the system[35].
Furthermore, PCA analysis could identify the most significant fluctua-
tion pattern of protein in terms of planar motion (eigenvector) and
amplitude (eigenvalue). Specifically, the PCA analysis first removed the
global rotation and motion of the system through the least square fitting,
and then constructed the covariance matrix according to the three-
dimensional position fluctuation of the given set of atoms at its overall
average position[36]. The internal motion of the correlation between
the molecule and N atoms is obtained by diagonalizing the covariance
matrix[37]:
Using the prepared SHP2 model and SHP2-Comp#5 complex model
as the initial structures, the GROMACS v4.5.5 software was applied for
the 100 ns MD simulation [24]. First, the GROMOS 43a1 force field was
used to obtain the documents required by Gromacs (molecular topology
file, molecular structure file, kinetic parameter file, running input file
and trajectory file). Afterwards, the model was confined to a dodeca-
hedral box solvated with SPC216 water model (the box boundary was at
least 1 nm from the model surface)[25] and the system was neutralized
+
-
with appropriate Na or CL ions. To ensure the rationality of protein
conformations and solvent, the equilibrium was carried out before MD
simulation. The steepest descent algorithm was applied on the initial
structures to escape the steric clashes and the unsuitable geometry,
thereby relaxing the initial structures with 10,000 steps with a
maximum force below 1000 kJ/mol or no drastic energy changes.
Subsequently, NVT and NPT simulation were conducted to equilibrate
the system, respectively [26]. The V-rescale and the Parrinello–Rahman
was used to equilibrate the system was under constant temperature (310
K) and pressure (1 bar) for 100 ps, respectively. Meanwhile, the LINCS
algorithm constrained all bonds and angles of the system [27]. The long-
range electrostatic potential was calculated using the PME method
σ
ij = 〈(r
i
ꢀ 〈r
i
〉)(r
j
ꢀ 〈r 〉)〉
j
where the ri/rj represents the Cartesian coordinate of the i-th/j-th atom
of the system, and “<>” calculates the average of ensemble. The PCA
scatter plots were obtained by the Bio3D library of R base package
software.
2
.9. Correlation motion analysis within the system
[
28,29]. The neighbors list updated every 10 steps. Finally, the models
performed 100 ns MD simulation with the trajectory recorded at in-
tervals of 20 ps.
Dynamic cross correlation matrix (DCCM) illustrated the time-
correlated information between system atoms i and j, and it could
identify concerted and correlated motion. DCCM analysis had three
typical characteristics, including strong cross-correlation along the di-
agonal, cross-correlation emanating from the diagonal, and non-
diagonal cross-correlation. The atoms along the diagonal always
showed strong correlated fluctuations with the Cij values are equal to 1.
These interrelationships provide dynamical information about an
average and/or the tertiary structure derived from the experiment. The
normalized correlation information between the protein atoms i and j
were calculated by[38]:
2
.6. MM-PBSA free energy calculations
Based on the molecular mechanics Poisson-Boltzmann surface area
algorithm, the binding free energy of receptor-ligand interaction was
estimated by the g_mmpbsa tool[30]. A total of 950 frames was
extracted from the last 95 ns MD trajectory with an interval of 100 ps
[
31]. The major residues playing an important role in binding free en-
ergy provided clear insight into the molecular mechanisms by which
proteins interacted with their ligands [32]. For the calculation of
MMPBSA free energy, brief instructions were provided as followed [33]:
〈
〈Δr
Δr
i
*Δr
j
〉
C
i,j = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
ꢀ
)
2
2
j
ΔGbind = Gcom plex ꢀ Greceptor + Gligand
(1)
(2)
(3)
(4)
(5)
i
〉〈Δr
〉
ΔGbind = ΔEgas + ΔGsol ꢀ TΔS
ΔEgas = ΔEint + ΔEele + ΔEvdw
ΔGsol = ΔGele,sol + ΔGnonpl,sol
ΔGnonpl,sol = γΔSASA
where the Δr /Δr expressed the mean position of i-th/j-th atom and the
i
j
angle brackets “<>” indicated the global average of MD trajectory. Ci, j
=
1 means completely coupled, while Ci, j = ꢀ 1 means anti-correlated
motion. The DCCMs were constructed by the Bio3D library of R base
package software[39].
2
.10. The interaction of amino acids within a protein
The binding free energy (ΔGbind) was calculated by the gas-phase
free energy, the solvation free energy and the entropy term (ꢀ TΔS).
The gas-phase free energy (ΔEgas) was composed of internal energy, the
electrostatic and VDW interactions. The solvation free energy was
The average MD trajectory of the 5 ns-100 ns simulation was
extracted to submit to the Residue Interaction Network Generator v2.0.1
(RING v2.0.1) server for post dynamic analysis[40]. The interactions
3

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
between residues constituted a complex network matrix, which was
visualized by the plugin “NetworkAnalyzer”[41] integrated in Cyto-
scape. Nodes were used to represent residues or ligands, and lines were
employed to describe the residue-residue interactions, including
hydrogen bonding, van der Waals interactions, ionic interaction and salt
bridges. In addition, the “RINspector” and “RINalyzer” plug-ins[42] in
Cytoscape were applied to find required network parameters for each
system, such as closeness centrality and shortest path betweenness.
Subsequently, to highlight and investigate the similarity and difference
of internal relations between SHP2-without ligands and SHP2-ligand
systems, a combined comparison network of the two systems was
generated based on the superposition and alignment of corresponding
network matrix. In undirected networks, the betweenness centrality
reflected the degree to which a node controled the interaction of other
nodes, while the closeness centrality measured the speed at which in-
formation spreaded from a given node to other reachable nodes[43].
interaction with residue LYS366. In addition, core C was placed in the
hairpin structure of the β5-β6-loop region and formed
π
- conjugate
π
interaction with ARG362/LYS364 which were likely responsible for the
enhanced potency and selectivity. On this basis, fragment A and frag-
ment B were obtained by searching the ZINC-fragment database, with
the purpose of making more interactions between the inhibitor and re-
ceptor residues LYS366, ARG465, GLN506 and GLV510. Two cores were
generated for the core hopping operation to core A part, namely, core A1
and core A2, to replace core A; two cores were generated for the core
hopping operation to core B part, namely, core B1 and core B2, to
replace core B. Fragment C was obtained by searching the aromatic
database with the aim of making the inhibitor interact with the
ARG362/LYS364 in a stable conjugated contact, enhancing the potency
and selectivity of the inhibitor. Four cores were generated for the core
hopping operation to core C part, namely, core C1, core C2, core C3, and
core C4, to replace core C. Consequently, a total of 2 × 2 × 4 = 16
different combinations for the thiophene [2,3-d] pyrimidine derivatives
were thus generated.
3
. Result and discussion
The docking score are used to evaluate the compounds obtained by
core hopping method. In the first instance, re-docking method was
preformed to assess the docking accuracy. The root-mean-square devi-
ation (RMSD) between the docked conformation and co-crystallization
conformation was calculated with the value of 1.698 Å (no>2 Å),
which proved that the methods and the parameters was satisfactory in
this study. Subsequently, the candidate compounds with docking scores
3
.1. Structural design and virtual screening of SHP2 inhibitors
The researches have reported that the active site of SHP2 protein
contained four major loop regions, including pTyr recognition loop
ARG277-PRO284), WPD-loop (ARG421-ASP431), P-loop (HIS458-
(
ARG465) and Q-loop (ARG501-THR507) (Fig. S8). Specifically, pTyr-
loop region identified Tyr specifically. The P-loop region contains
active nucleophiles CYS459 and ARG465, which can be used to identify
the phosphorylated base in the substrate. The general acid-base catalyst
ASP425 located in the WPD-loop region. Besides, the GLN506 located in
the Q-loop region could hydrolyze phosphatase intermediate. The sec-
ond binding site (ARG362/LYS364) was responsible for the enhanced
selectivity[44]. Therefore, the development of inhibitors that could
occupy these active sites simultaneously and the second binding site
might yield high affinity and selectivity. In this paper, thiophene [2,3-d]
pyrimidine as lead compounds were obtained by virtual screening.
Divided the structure of the lead compound PubChem CID 8,478,960
was into three parts, Core A, Core B, and Core C, as marked by red line,
blue line and purple line, respectively (Fig. 1). The core A and core B
were placed in the polar region and formed hydrogen bond interaction
Table 1
Docking score of ten candidate molecules by molecular docking.
Compound
Docking score (kJ/mol)
SHP2
SHP1
Comp#1
Comp#2
Comp#3
Comp#4
Comp#5
Comp#6
Comp#7
Comp#8
Comp#9
Comp#10
ꢀ 10.5038
ꢀ 11.9772
ꢀ 20.6249
ꢀ 19.6624
–22.5625
ꢀ 19.3184
ꢀ 17.9907
ꢀ 14.2884
ꢀ 11.3207
ꢀ 10.4864
ꢀ 6.7001
ꢀ 7.6455
ꢀ 11.0057
ꢀ 10.3425
ꢀ 12.3831
ꢀ 10.3663
ꢀ 8.8527
ꢀ 7.4418
ꢀ 7.2914
ꢀ 6.5926
with residues ARG465, GLN506 and GLN510, formed
π
-π conjugate
Fig. 1. Generation of the core fragments from thiophene [2,3-d] pyrimidine ring by core hopping method.
4

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
in the top ten were calculated as shown in Table 1. It was obviously that
most compounds had good affinity with the SHP2 system, among which
Comp#5 ranked the first in docking score (–22.5625 kJ/mol). Inter-
estingly, their affinity with SHP1 protein was weak. The synthesis of
these ten compounds were then described in details (Fig. 2). The
the system with stirring for 10 min. Subsequently, the precipitate was
suction filtrated and washed with water. Yellow powder (1.5 g) of
product was obtained.
3.2.3. A general method for preparing thiazolomethyl substituted
pyrimidinone from thiazolamide
1
structures of the synthesized compounds were characterized by H NMR
1
3
and C NMR (Supplementary material).
A mixture of isopropyl 2-amino ꢀ 4,5,6,7-tetrahydrobenzo[b]thio-
phene-3-carboxylate-2a, 2b (3.0 g, 12.4 mmol) and acetonitrile (5.1 g,
124 mmol) in dioxane hydrochloride (60 mL, 1.5 mol/L) was heated and
3
3
.2. Synthesis of thiophene [2,3-d] pyrimidine derivatives
◦
stirred at 90 C for 12 h. The reaction mixture was cooled to room
.2.1. A general method for preparing thiazolamide from substituted
temperature, and 60 mL water was introduced into the system with
stirring for 10 min. Subsequently, the precipitate was suction filtered,
and washed twice with water and petroleum ether (20 mL*2), respec-
tively, and the white solid was obtained.
cyclohexanone. The synthesis of isopropyl 2-amino-6-isopropyl-4,5,6,7-
tetrahydrobenzo[b]thiophene-3-carboxylate
2
Under N atmosphere, a mixture of cyclohexanone derivatives-1a, 1b
(
5.0 g, 35.7 mmol), isopropyl cyanoacetate (5.0 g, 39.2 mmol), sulphur
powder (1.7 g, 53.6 mmol) in absolute ethanol (100 mL) were heated
and stirred in an oil bath. Morphine was then slowly added dropwise to
3.2.4. General method for synthesizing amide from bromoacetyl bromide
and amine
◦
the mixture, and the temperature was maintained at 35 C for 24 h. The
A mixture of amines-5a, 6a (10 g, 117 mmol) and pyridine (10.2 g,
128.7 mmol) in dichloromethane (500 mL) was stirred to mix. Bro-
moacetyl bromide was added dropwise to the mixture, followed by
stirring at room temperature for about 2 h. A 10% aqueous solution of
copper sulfate was added to the reaction liquid, and the liquid phase was
separated to leave the lower organic phase. The organic phase was
washed twice with 200 mL of a saturated sodium chloride solution,
dried, and concentrated to dryness by vacuum distillation. To the ob-
tained solid was added 100 mL of a petroleum ether / dichloromethane
solution (10:1), stirred and filtered to give a white solid.
reaction liquid changed from a light yellow to a dark yellow when most
of the starting materials were converted into the target compound. TLC
(
ethyl acetate/petroleum ether = 1/3) detected that the raw material
completely disappeared.
At room temperature, the reaction solution was diluted with 100 mL
of water. Most of the ethanol was removed by vacuum distillation at
◦
4
5 C. The mixture was extracted by 150 mL ethyl acetate, and the
combined organic layer was washed with saturated brine, dried over
anhydrous magnesium sulfate, filtered, and concentrated by rotary
evaporation.
3
.2.5. A general method for preparation of thiazolpyrimidone derivatives
3
.2.2. General procedure for the synthesis of thiazolopyrimidinone from
thiazolamide
Under N
3a, 3b, 4a, 4b (200 mg, 0.7 mmol), 6a (175 mg, 0.84 mmol) and
cesium carbonate (273.5 mg, 0.84 mmol) were added in this order in
◦
2
atmosphere, the compound of isopropyl 2-amino ꢀ 4,5,6,7-
acetone (20 mL), and stirred in an oil bath at 60 C for 2–3 h until the
tetrahydrobenzo[b]thiophene-3-carboxylate-2a, 2b (3.5 g, 12.4 mmol)
TLC monitoring material disappeared completely. The reaction solution
was naturally cooled to room temperature, and 20 mL of water was
slowly added with stirring. The reaction system gradually clarified and
precipitated solids. After stirring for about 10 min, it was filtered, and
the precipitate was rinsed twice with water/acetone solution (1:1) (5
◦
in formamide (50 mL) was heated and stirred at 150 C for 12 h when the
reaction system changed to dark brown. LC-MS monitors the complete
conversion of the starting materials to the product. The reaction mixture
was cooled to room temperature, and 50 mL water was introduced into
Fig. 2. The synthetic route of comp#1–10. Reagents and conditions: (I) morpholine, sulphur powder, stir, 35 ℃, 24 h; (II) formamide, stir, 150 ℃, 12 h; (III) dioxane
hydrochloride, stir, 90 ℃, 12 h; (Ⅳ) stir, room temperature, 2 h; (Ⅴ) cesium carbonate, stir, 60 ℃, 2–3 h.
5

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
mL*2) and twice with petroleum ether solution (20 mL*2) to give a
white solid.
Table 3
The actual activities of ten candidate molecules by SHP2 and SHP1 phosphatase
assay.
1
Compound
Na VO
Combination of cores
SHP2 IC50
(
μM)
SHP1 IC50
(μM)
3
.3. ADMET analysis
3
4
–
29.55
>100
>100
1.02
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
Comp#1
Comp#2
Comp#3
Comp#4
Comp#5
Comp#6
Comp#7
Comp#8
Comp#9
Comp#10
A1-B1-C1
A1-B1-C2
A1-B1-C3
A2-B1-C1
A2-B1-C4
A1-B2-C2
A1-B2-C1
A2-B2-C2
A2-B2-C4
A2-B2-C3
ADMET characteristics were important indicators to test whether the
candidates could meet the required standards, and lead compounds
were selected and optimized according to their ADMET properties. In
this study, the ‘‘Calculate Molecular Properties’’ module was employed
to calculated a several of molecular characteristics and pharmacokinetic
properties of thiophene [2,3-d] pyrimidine derivatives by the DS v3.5
software. The test results of the ten compounds were shown in Table 2.
For oral bioavailability, two principal factors affecting oral bioavail-
ability included human intestinal absorption (HIA) and solubility. The
poor intestinal absorption level mean that the effect of drug treatment
would be significantly reduced. The effect of solubility on the pharma-
cological activity of compounds involved many aspects, including their
absorption, distribution and ultimate bioavailability. Moreover, the
plasma protein binding level was used to predict the binding ability of
compounds to plasma proteins, which affected the distribution of drugs.
The results indicated that all compounds had good affinity to plasma
proteins. And the blood–brain barrier permeability model showed that
all 10 compounds could penetrate the blood–brain barrier. In addition,
the cytochrome P450 2D6 enzyme encoded by human CYP2D6 gene
converted the drug into a substance that was easily metabolized for
excretion. Inhibition of enzyme activity constituted the majority cases of
drug-drug interaction. The results showed that none of the 10 com-
pounds had cytochrome P450 CYP2D6 enzyme inhibitory activity. For
hepatotoxicity prediction, all but comp#8 were predicted to be non-
toxic. In conclusion, the ADME property values of the compounds in
Table 2 (expect Comp#8) were all within the acceptable range of human
beings, indicating that these compounds complied with the drug rules.
1.38
0.4
2.88
18.57
37.87
>100
>100
1
See Fig. 1 for the structure of cores.
inhibitory activity of the candidate compounds. While when R1 was
isopropyl, the activity of the compounds decreased significantly (except
Comp#3 and Comp#6). Comp#5 and Comp#9 had similar structures,
but the activity of the latter was much lower than that of the former,
possibly due to methyl groups attached to the pyrimidine ring of the
parent nucleus destroying the interaction between the inhibitor 9 and
the receptor protein. A similar situation could occur with Comp#8 and
Comp#10. In addition, electron-withdrawing groups attached to the
benzene ring in the Core C helped to enhance the activity of the inhibitor
(
Comp#3 and Comp#5).
3
.5. Analyze the activity and selectivity of the inhibitors by molecular
docking
To verify and evaluate the putative binding pose of the inhibitor
Comp#5 in the active site of SHP2, molecular docking experiments were
implemented and visualized by DS v3.5 software. Fig. 3A showed the
key amino acids interacted with the ligand. To describe how the
Comp#5 bound to the active pocket, the “analyze single complex”
module was used to generate a three-dimensional graph of the in-
teractions between receptor and ligand (Fig. 3B). As expected, the six-
membered ring of Comp#5 was found in the PTP active pocket and
formed intimate interactions with the amino acids in P-loop region, Q-
loop region and pTyr-loop region. Specifically, the strong polar sub-
stituent F atom on the six-member ring forms hydrogen bond interaction
with the residue GLN506 in Q-loop region. The inhibitor formed hy-
drophobic interaction with residues CYS459-GLY464 in the P-loop re-
gion and TYR279 in the pTyr-loop region, which may be related to
substrate recognition. Furthermore, the thiophene [2,3-d] pyrimidine of
Comp#5 penetrated deeply into the active pocket and formed polar
interaction with LYS366 located in β5-β6 loop neighboring the PTP-
domain, ARG465 located in P-loop, and GLN510 located in the
3
.4. The analysis of activity and structure–activity relationship (SAR) of
candidate compounds
The compounds were subjected to SHP2-PTP and SHP1-PTP enzy-
matic activity assays with the pNPP as the substrate. The results were
displayed in Table 3. Comp#1, Comp#2, Comp#9 and Comp#10 had no
inhibitory activity for SHP2-PTP. Comp#3, Comp#4, Comp#5,
Comp#6, Comp#7 and Comp#8 inhibited SHP2-PTP-catalyzed hydro-
lysis of the pNPP substrate with an IC50 value of 1.02
μ
M, 1.38 M, 0.4
μ
μ
M, 2.88 M, 18.57 M and 37.87 M, respectively. While they showed
μ
μ
μ
little inhibition for the SHP1 enzyme. Analysis of conformation char-
acteristics of compounds showed that when the substituents of R1 and
R2 on Core A were F atoms, it was beneficial to improve the inhibitory
activity of compounds (Comp#4 and Comp#5). This indicated that the
electron-withdrawing groups on R1 and R2 were beneficial to the
Table 2
The ADME prediction for the thiophene [2,3-d] pyrimidine derivatives.
a
b
c
d
Molecular weight
ALogP98
Absorption Level
Solubility Level
PPB Level
BBB Level
CYP2D6 Prediction
Hepatotoxic Prediction
comp#1
comp#2
comp#3
comp#4
comp#5
comp#6
comp#7
comp#8
comp#9
comp#10
409.54
395.52
479.52
403.45
411.37
409.54
423.57
403.45
425.4
4.851
3.916
4.835
3.533
3.002
3.517
4.93
0
0
0
0
0
0
0
0
0
0
2
2
1
2
2
2
1
2
2
2
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
1
1
1
2
2
2
1
2
2
2
FALSE
FALSE
FALSE
FALSE
FALSE
FALSE
FALSE
FALSE
FALSE
FALSE
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
TRUE
FALSE
TRUE
TRUE
2.676
3.081
3.595
487.44
a: The logarithmic coefficient of n-octanol and water is taken as the logarithm, the value of AlogP98 ≤ -2.0 or AlogP98 ≥ 7.0 means very poor;
b: Human intestinal absorption level, where 0 stands for good, 1 for moderate, 2 for poor, and 3 for very poor;
c: Aqueous solubility level, where 0 stands for extremely low, 1 for very low, but possible, 2 for low, and 3 for good;
d: Blood-brain barrier permeability, where 1 stands for good, 2 for moderate, and 3 for poor.
6

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
Fig. 3. Docking complex between inhibitor Comp#5 and SHP2 protein. (A) The 2D map of the interaction between inhibitor Comp#5 and key amino acids in SHP2
protein. (B) Three-dimensional map of the binding pattern for inhibitor Comp#5 to the active pocket in SHP2 protein. The pink frame represented polar interactions,
including hydrogen bonds and charge interactions, and the green frame represented nonpolar interactions, including hydrophobic interactions and van der Waals
forces. The dotted lines connecting the compound to the residues represent the H bond interaction. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)
surrounding of Q-loop region. In particular, the sulfur atoms of thio-
phene formed hydrogen-bond interaction with the side chains of residue
GLN510, nitrogen atoms of pyrimidine formed hydrogen-bond interac-
tion with the side chains of residue ARG465 and GLN510, oxygen atoms
of pyrimidine formed hydrogen-bond interaction with the main chain of
residue LYS366 and ARG465, respectively. In addition to the hydrogen-
attenuating the fluorescence value. Fig. 4 illustrated the quenching of
SHP2 fluorescence intensity by Comp#3 and Comp#5 in a dose-
dependent manner. Obviously, fluorescence experiments demonstrated
that Comp#3 and Comp#5 bound directly to the SHP2 protein, thus
weakening its activity on the substrate.
bond interactions, the pyrimidine ring of Comp#5 also formed
π π
-
3
.7. Evaluation of the binding energy
conjugate interaction with the residue ARG465 located in P-loop. There
were also additional polar contacts between Comp#5 and the residues
located in the WPD-loop region (ARG421-ASP431). Additionally, the
distal phenyl ring of Comp#5 was sandwiched between the side chains
of ARG362 located in the β5-β6 loop (residue 362–365) and HIS426
The MM-PB/SA methods were employed to calculated the complete
binding free energy aimed at evaluating the binding affinity of the
docking analysis. The snapshots were extracted from the last 95 ns MD
simulation at equal intervals of 100 ps and served as input for the
calculation. Table 4 displayed the binding energies for SHP2-Comp#5
and SHP2-Comp#9 complex, respectively. The predicted binding free
energy of SHP2-Comp#5 and SHP2-Comp#9 was ꢀ 107.937 kJ/mol and
located in WPD-loop, and formed
π
-π conjugate interaction with the two
residues. Interestingly, the β5-β6 loop (residue 362–365) was highly
diverse in the PTPs , especially in SHP1 and SHP2[44]. The conjugation
between the benzene ring and residues ARG362 might be the main cause
of the potency and selectivity enhancement of inhibitor Comp#5. Fig. S1
showed that the interaction between Comp#5 and SHP1 was weak.
Especially, the alicyclic ring and benzene ring of Comp#5 were located
outside the active pocket. This made the F atoms on the alicyclic ring
and benzene ring in Comp#5 form weak interaction with SHP1,
resulting in the decrease of inhibitor activity.
ꢀ
48.252 kJ/mol, respectively, which verified that the binding affinity
between Comp#5 and SHP2 was higher than that between Comp#9 and
SHP2. The binding free energy was composed of Van der Waal energy,
electrostatic energy, polar solvation energy and SASA energy. It was
rather remarkable that the Van der Waals and nonpolar solvation energy
were major contributors to the binding energy between inhibitor to
SHP2, because the inhibitor has hydrophobic contact with non-polar
residues around the active site. While polar solvation was not condu-
cive to the stable of the complex of the protein and the ligand, which
could largely cut down the total contribution. In short, the high binding
affinity between Comp#5 and SHP2 protein was due to the significant
contribution of the van der Waals interaction between the inhibitor and
the non-polar surface of the residues at the active site and the hydrogen
bond interaction with the side chains at the active site. High binding
affinity should be one of the main reasons for the effective inhibition of
inhibitors. On the other hand, the active site was precisely occupied by
Comp#5, which hindered the entry of substrates and thus exerted an
inhibitory effect. However, Comp#9 did not occupy the active site, so it
was difficult to exert an inhibitory effect.
For comparison, the docking conformation of Comp#9 and SHP2
protein was visualized (Fig. S2). Obviously, the Comp#9 was located
outside the active pocket (Fig. S2A), and it only formed hydrogen bonds
with the residues LYS366 and THR507, with the other major in-
teractions disappearing (Fig. S2B). The affinity between Comp#9 and
SHP2 protein was much lower than that between Comp#5 and SHP2
protein, which explained why the lower inhibitory activity of Comp#9
than that of Comp#5. The results showed that the inhibitor Comp#5 had
strong interactions with the active pocket of PTP protein, including
polar and non-polar interactions, indicating that the inhibitor played an
important role in stabilizing protein conformation. In summary, the
intimate interaction between the inhibitor and the SHP2 protein
enhanced the stability of SHP2 protein, thereby negatively affecting the
catalytic activity of the PTP region. Meanwhile, the benzene ring of
Comp#5 formed conjugated interaction with ARG362 in the β5-β6 loop
region, which enhanced the selectivity of the inhibitor.
To understand the details of the interaction of SHP2-inhibitor
Comp#5 with the receptor, the binding energy of the interaction be-
tween ligands and key amino acids was calculated. The differences be-
tween the interaction energies of SHP2-Comp#5 and SHP2-Comp#9
were compared, as shown in Fig. S3. In SHP2-Comp#5 system, the in-
hibitor formed hydrogen bonds with residues, including LYS366 in the
β5-β6-loop, ARG465 in P-loop, GLN506 and GLN510 in Q-loop of the
active region, contributed significantly to the binding free energy. Be-
sides these hydrogens, the additional conjugate interaction of the thio-
phene ring of Comp#5 with ARG465 was formed, making ARG465
3
.6. Fluorescence titrations
The excitation wavelength of SHP2 was 295 nm, and the fluores-
cence monitoring range was from 360 nm to 500 nm. Combining with
the compound will inhibit the activity of the protein, thereby
7

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
Fig. 4. The relationship between the Log concentration of the Comp#3/Comp#5 and the fluorescence intensity of SHP2 protein. The data displayed the mean ± SEM
of triplicate representative experiments. Analogical results were obtained for all compounds tested.
Comp#9 system also reached a stable state after 5 ns MD simulation and
Table 4
the average RMSD value was similar to that of the SHP2-without ligands
MM/PBSA based binding free energies (kJ/mol).
system (Fig. S4A). The results showed that all the systems reached an
Complex
3O5X-Comp#5
3O5X-Comp#9
equilibrium state after a short simulation of 5 ns. In particular, the RMSD
value of the SHP2-Comp#5 complex system was slightly lower than that
of the SHP2-without ligands system and SHP2-Comp#9 system, pre-
sumably because the inhibitor Comp#5 weakened the flexibility of the
system and made the system more stable than the other two systems
Van der Waal Energy (kJ/mol)
Electrostatic Energy (kJ/mol)
Polar Solvation Energy (kJ/mol)
SASA Energy (kJ/mol)
ꢀ 237.936
ꢀ 31.217
179.799
ꢀ 96.197
ꢀ 59.354
119.103
ꢀ 11.803
ꢀ 48.252
ꢀ 18.583
ꢀ 107.937
Binding Free Energy (kJ/mol)
[
45].
Root mean square fluctuation (RMSF) evaluated dynamic charac-
teristics by recording the average atomic fluctuation of side chain atoms
(N, C , and C), which played an important role in the indication of the
contribute the highest binding energy. In SHP2-Comp#9 system, how-
ever, the residues above contributed slightly to the binding free energy.
Although the residue LYS366 and Comp#9 formed hydrogen bond
interaction, the binding free energy of residue LYS366 was still low. In
addition, the van der Waals interactions formed between the inhibitor
and residues CYS459, SER460, ALA461, and ILE463 in the P-loop re-
gion, which was particularly important in catalytic reactions and was
widely recognized as binding target of inhibitors[44], were another
major reason for the high affinity in SHP2-Comp#5 system, which was
verified by van der Waals’ significant contribution. As mentioned above,
α
conformational features of a dynamic system. Overall, the RMSF fluc-
tuations of the SHP2-Comp#5 protein was very similar to that of the
SHP2-without ligands and SHP2-Comp#9 protein, except for some
crucial regions like the β5-β6 loop, P-loop and Q-loop region. As a matter
of fact, the magnitude of RMSF curve fluctuations was closely related to
the interaction of inhibitor with SHP2 protein. For the catalytic activity
areas (marked with solid black boxes), the magnitude of RMSF value in
SHP2-Comp#5 complex system was smaller than that in SHP2-without
ligands system and SHP2-Comp#9 system (Fig. 5B and Fig. S4B). Spe-
cifically, the RMSF values of residues CYS459-ARG465 were 0.26 nm,
0.14 nm and 0.25 nm in SHP2-without ligands, SHP2-Comp#5 system
and SHP2-Comp#9 system, respectively. Besides, the RMSF values of
catalytic activity areas residue GLN506-GLN510 were calculated to be
0.3 nm, 0.14 nm and 0.26 nm in SHP2-without ligands, SHP2-Comp#5
system and SHP2-Comp#9 system, respectively. The binding of the in-
hibitor Comp#5 to the key amino acids of SHP2-PTP protein limited its
fluctuation and increased the overall stability of the system. However,
the binding affinity between Comp#9 and SHP2-PTP protein was weak,
which made it have little impact on the overall flexibility. The low
flexibility of complex mainly depended on the tight packed protein
arrangement, as well as strong interactions between the amino acids in
proteins. The results indicated that the inhibitor Comp#5 might increase
the stability of SHP2-Comp#5 protein. For the region located at the
entrance of catalytic active region (β5-β6 loop), the average RMSF value
of residues GLU361-LYS366 was 0.26 nm in SHP2-without ligands sys-
tem and SHP2-Comp#9 system, whereas the average RMSF value of the
regions was 0.22 nm in the SHP2-Comp#5 system. The results suggested
that the inhibitor 5 binding to the β5-β6 loop region of the protein
reduced the fluctuation of this region, while Comp#9 had slight effect
on this region, which might explain the selectivity of Comp#5.
residues ARG362 and HIS426 formed
inhibitor Comp#5, and their affinity values were ꢀ 11.095 kJ/mol and
6.301 kJ/mol, respectively. The strong interactions between Comp#5
π
-
π
conjugate interactions with the
ꢀ
and residue ARG362 highlighted the contribution of the β5-β6-loop re-
gion, which was the main reason for the high selectivity of the com-
pound. However, due to the weak interactions between Comp#9 and
key amino acids, binding free energy was generally low. Collectively,
the inhibitor Comp#5 had high affinity with the active site residues,
thereby stabilizing the conformation of the protein. These results were
consistent with the above structural observation that Comp#5 achieved
its potency and specificity by targeting to the active regions divergent in
PTPs.
3
.8. MD simulation analysis
Dynamic characteristics and structural stability of SHP2 protein,
SHP2-Comp#5 protein and SHP2-Comp#9 protein was explored
through MD simulation. The stability and convergence of these systems
were determined by RMSDs of the backbone C atoms relative to the
α
initial structure. In the SHP2-without ligands system, the RMSD values
experienced a rapid rise from 0 to 0.55 nm, reflecting the unstable state
of the initial simulation of the protein. The trajectory converged and the
RMSD values retained below 0.5 nm after approximately 5 ns, which
indicated that the model reached an equilibrium state (Fig. 5A). In the
SHP2-Comp#5 complex system, although RMSD also experienced an
initial increase, its value converged after 5 ns simulation and stabilized
below 0.42 nm (Fig. 5A). In addition, RMSD analysis showed that SHP2-
3.9. Principal component analysis
The conformation of a protein has been considered as one of the
principal characteristics for the determination of biological function.
8

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
Fig. 5. (A) RMSD of backbone atoms during the
simulation time. (B) The per-residue RMSF of side
chain versus individual residue number. In the
figure, the green curve represents the fluctuations
of SHP2-without ligands system, while the red
curve represents the fluctuations of SHP2-
Comp#5 system. In addition, the areas circled
by the black box correspond to the areas with
significant differences. (For interpretation of the
references to color in this figure legend, the
reader is referred to the web version of this
article.)
The protein structure was grouped into the subsets according to the
conformational similarity by PCA clustering method[46]. PCA analysis
were applied using the last 95 ns dynamic trajectories in SHP2-without
ligands protein, SHP2-Comp#5 protein and SHP2-Comp#9 protein. The
results were shown in Fig. 6 and Fig. S5. The left side represented the
states of frame extracted at 2 ps intervals during the whole simulation
process, and the right side represented the probability distribution of
eigenvalues. The PC analysis suggested that the first two PCs took the
percentage of 49.0%, 50.9% and 36.8% of the variance in the dynamic
trajectories observed in simulation process of the SHP2-without ligands
protein, SHP2-Comp#5 protein and SHP2-Comp#9 protein, respec-
tively, which were predominant among these components. Whereas,
other individual components contributed below 7.2%. The results
showed that the conformational state distributions of PC1 and PC2 were
representative. Subsequently, the PC analysis plots were obtained by
projecting along the direction of the first two PCs, which explained the
motion characteristics related to the conformational behavior of the
main system. In the Fig. 6, the red, white and blue dots were used to
show different state, where each point represented the conformation of
the two systems. The continuous colors emphasized the periodic tran-
sitions between these conformations. The convergent stable conforma-
tion and the unstable scattered conformation were described by the
aggregation and dispersion states of red/blue dots and light red/blue
dots, respectively. In the SHP2-without ligands system (Fig. 6A), it was
obviously that the red and blue dots presented loose scattering confor-
mations, and even most of them were in the intermediate transition
state, indicating the poor structural stability of the system. Interestingly,
in the SHP2-Comp#9 system, the red dots and blue dots showed loosely
scattered conformations similar to that of the SHP2-without ligands
system (Fig. S5B). Whereas, in the SHP2-Comp#5 compound system
(Fig. 6B), the red and blue dots were clustered in a tight distribution,
indicating the conformational of the system was stability. In brief, SHP2-
Comp#5 complex system was more compacted than SHP2-without li-
gands and SHP2-Comp#9 system, thus, binding of Comp#5 inhibitor to
the protein decreased the conformation flexibility, which lead to a
decrease in protein activity. This result was consistent with the RMSF
analysis. However, in order to further explore how the Comp#5 would
have effect on the residue-residue interactions of SHP2, the additional
dynamic analyses were required.
3
.10. Dynamic cross-correlation mapping
All paired cross-correlations of amino acid fluctuations served as a
reference for understanding how the movement or displacement of one
atom in a system was related to another[47]. The functional flexibility
and the relative coordination between the atomic groups determined the
conformational transformation of the system, which provided a wide
range of protein conformational information[48]. Therefore, the dy-
namic features exhibited by SHP2-without ligands system, SHP2-
Comp#5 system and SHP2-Comp#9 system based on the DCCM anal-
ysis were performed using the last 95 ns dynamic trajectories. The
correlated movements (highly positive) in specific areas were stained
dark green, while anti-correlated movements (highly negative) in spe-
cific areas were stained orange. The correlated movement of the
9

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
Fig. 6. PCA scatter plots of the trajectories constructed by projecting the first two principal components (PC1 and PC2) in the conformational space of (A) SHP2-
without ligands system and (B) SHP2-Comp#5 complex.
diagonal region described the coordinated motion with respect to the
residual itself. As could be seen from the Fig. 7 and Fig. S6, the in-
teractions between residues in the SHP2-without ligands system and
SHP2-Comp#9 system were weak, while the correlations between resi-
dues in the SHP2-Comp#5 system was strengthened, revealing that the
binding of inhibitor Comp#5 not Comp#9 to SHP2-PTP protein might
increase the stability of the complex system. In the SHP2-Comp#5
complex system, residues TRP423-GLY427 of WPD-loop formed posi-
tively correlated with residues GLN506-GLN510 of Q-loop (region c) in
catalytic active area, while a weak correlation was observed in the
SHP2-without ligands system and SHP2-Comp#9 system. This indicated
that the action of the Comp#5 on the catalytic activity region lead to the
enhancement of the amino acid interaction in this region. Similarly, the
correlation between residues MET355-ARG362 of β5-β6 loop and resi-
dues TRP423-GLY427 of WPD-loop (region a), residues MET355-
ARG362 of β5-β6 loop and residues CYS459-ARG465 of P-loop (region
b) in the active region in the SHP2-Comp#5 complex system were su-
perior to that in the SHP2-without ligands system and SHP2-Comp#9
system. The Comp#5 interacted with the catalytic activity region,
which resulted in increased amino acid compactness between residues
MET355-ARG362 and TRP423-GLY427, as well as residues MET355-
ARG362 and residues CYS459-ARG465. It was noteworthy that in the
SHP2-Comp#5 complex system, the residues MET355-ARG362 in the
β5-β6 loop interacted strongly with the nearby catalytic active region,
which might be the main reason for the high potency and strong selec-
tivity of the inhibitor Comp#5. All in all, the DCCM result analysis
suggested that the binding of compound to the catalytic active region
lead to stronger residual interactions, and the complex system is more
stable than the initial system, which may inhibit the activity of proteins.
3
.11. Residue interaction network
RIN analysis was a useful strategy to explore and identify the key
residue interaction and structure–function, which could be used to
explore how the ligand would change the interactions between residues
of proteins. In this study, a representative RIN network was generated by
the last 95 ns molecular dynamics trajectories to investigate the mech-
anism of Comp#5 and Comp#9 effecting on SHP2. The topological
1
0

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
Fig. 7. Description of the correlation movement of all amino acids in the SHP2-without ligands system (A) and the SHP2-Comp#5 system (B). The color distribution
between orange (from ꢀ 1.0 to ꢀ 0.25), white (from ꢀ 0.25 to 0.25) and green (from 0.25 to 1.0). The negative values correspond to the negative correlation motion,
indicating the opposite direction of the atomic displacement; the positive values correspond to the relevant motion, indicating the same direction of atomic
displacement. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 8. The distribution of interactions among protein residues in the comparison network. The fringe pattern corresponds to the residual interactions retained in the
system, where the green dashed lines, the red dotted line and the black solid line represent the interactions between the amino acids in the SHP2-without ligands
system, the SHP2-Comp#5 system and the two systems, respectively. The active areas with significant differences were highlighted by yellow solid coils. the mo-
lecular mechanism of Comp#5 inhibiting SHP2 activity. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of
this article.)
1
1

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
properties of single amino acid residues, including shortest path inter-
activity and compact centrality, are characterized by the relationships
among the residues within SHP2 protein (Table S1). Interestingly, the
residues located at the protease activity sites in the SHP2-Comp#5
complex system had higher betweenness and closeness value than
those in the SHP2-without ligands system, such as THR357, ARG362,
LYS366, PRO424, CYS459, SER460, ALA461, ILE463, ARG465, THR507
and GLN510. It was evident in the fact that, after the inhibitor bound to
the active pocket, information traveled faster from one residue in the
protein to another that could reach it. By comparing the differences in
the above active site residues in SHP2-without ligands and SHP2-
Comp#9 systems, it could be seen that betweenness and closeness
value between the protease active site residues in the SHP2-Comp#9
complex system was even lower than that in the SHP2-without ligands
system.
4. Conclusion
The primary objective of this research was to obtain novel and highly
effective and selective SHP2 inhibitors. In this study, the core hopping
method was employed to modify the lead compound PubChem CID
8,478,960 (IC50 = 45.01
pyrimidine derivatives (IC50 = 0.4–37.87
properties. SHP2-PTP enzymatic activity assays showed that the
Compo# 5 had the strongest potency and selectivity (IC50 = 0.4 M).
μ
M), obtaining a variety of thiophene [2,3-d]
μM) with improved core
μ
Then, in order to elucidate the molecular mechanism of Comp#5
inhibiting SHP2 activity, MD simulation and post-kinetic analysis of
SHP2-without ligands, SHP2-Comp#5 and SHP2-Comp#9 (serve as a
negative control) systems were carried out. The results showed that
Comp#5 had a strong binding affinity with SHP2 protein, and it was
worth mentioning that its conjugated interaction with ARG362 residues
might help improve the selectivity of the system. whereas Comp#9 had
a weak binding affinity with SHP2 protein. Moreover, the RMSF analysis
suggested that the β5-β6 loop region, P-loop region and Q-loop region of
the SHP2-Comp#5 protein was more stable than those of the SHP2-
without ligands and SHP2-Comp#9 systems. In addition, the DCCM
analysis confirmed that there were positive correlations between the β5-
β6 loop region and the WPD-loop region, the P-loop region and the β5-β6
loop region, and the Q-loop region and the WPD-loop region in SHP2-
Comp#5 system, while their correlation was extremely weak in SHP2-
without ligands and SHP2-Comp#9 system. This indicated that the
Comp#5 inhibitor enhanced the interaction of amino acids within the
SHP2 protein, enhancing the stability of the system structure, and thus
inhibiting the activity of protease. Nevertheless, Comp#9 had little ef-
fect on the stability and interaction of these key residues. Specifically,
Comp#5 enhanced the interaction of residues THR357, ARG362,
LYS366, PRO424, CYS459, SER460, ALA461, ILE463, ARG465, THR507
and GLN510 with the surrounding residues, enhancing the stability of
the catalytic active region and the entrance of catalytic active region.
These results explained the mechanism of Comp#5 inhibiting SHP2 at
the molecular level.
Even small changes in a single residue site could have extreme effects
on protein structure, eventually leading to significant distortions of the
active site fragment. Therefore, it was extremely important to study the
interactions between individual amino acids. In the RIN analysis, we
mainly focused on analyzing the residues in the active site regions res-
idues CYS459-ARG465 in P-loop and residues GLN506-GLN510 in Q-
loop and the entrance of catalytic active region residues GLU361-
LYS366 in β5-β6 loop, which were consistent with areas of significant
difference in RMSF analysis. In the Fig. 8, the boxes corresponded to
amino acids, and the lines represented the interactions between amino
acids. It was obvious from the comparison network that the binding of
inhibitor Comp#5 lead to different residues interaction at the protease
active site (P-loop, Q-loop) in these two systems. Specifically, residue
ILE463 in P-loop region formed van der Waals interaction with residue
ARG501, SER502, GLY503, MET504 in Q-loop region in SHP2-Comp#5
complex system, respectively. However, only van der Waals interaction
between residue ARG465 and GLN510 was found in SHP2-without li-
gands system. Compared with the SHP2-without ligands system, amino
acids in the P-loop region and Q-loop region of the SHP2-Comp#5 sys-
tem had stronger interactions, suggesting that inhibitors may strengthen
the connection between these two regions and make the conformation of
the complex more stable. Furthermore, residue ALA461 in P-loop region
formed van der Waals interaction with residue TYR279 and ILE282 in
pTyr loop, residue SER460 in P-loop region formed van der Waals
interaction with residue ASN277 and TYR279 in pTyr loop in the SHP2-
Comp#5 system, while only van der Waals interaction between SER460
and TYR279 was found in the SHP2-without ligands system. The results
showed that the interaction between P-loop region and pTyr-loop region
in SHP2-Comp#5 system was stronger than that in SHP2-without li-
gands system. Interestingly, the P-loop region also strongly interacted
with the β5-β6 loop region in the SHP2-Comp#5 system, which was
manifested in H-bond formation and van der Waals interaction between
residue SER460 and residue THR357, as well as residue SER460 and
LYS366, but these interactions had not been found in the SHP2-without
ligands system. The interaction of ARG362 residue with surrounding
amino acids had a significant effect on the activity and selectivity of
SHP2-Comp#5 protein. In addition, strong H-bond interaction and van
der Waals interaction between ARG465 in the P-loop and PRO424 of the
WPD loop was found in the pTyr-bound PTP structures, which plays an
important role in stabilizing the closed conformation. Therefore, the
binding of Comp#5 to the SHP2-without ligands receptor affected the
internal interaction network of amino acids.
5. Compliance with ethics requirements
This article does not contain any studies with human or animal
subjects
Declaration of Competing Interest
The authors declare no competing financial interest.
Acknowledgements
The National Natural Science Foundation of China (Grant No.
8
1273361);
The Cooperation and Exchange Project of the National Natural Sci-
ence Foundation of China (Grant No. 81611130090);
The Natural Science Foundation of
No.16JCZDJC32500, Grant No. 18JCQNJC13700);
Tianjin
(Grant
The Science & Technology Development Fund of Tianjin Education
Commission for Higher Education (Grant No. 2017KJ229).
Appendix A. Supplementary material
In the Compare RINs analysis between SHP2-without ligands system
and SHP2-Comp# 9 system, Comp#9 had little effect on the active site
regions residues CYS459-ARG465 in P-loop and residues GLN506-
GLN510 in Q-loop, while it weakened the connections between amino
acids in the entrance of catalytic active region residues GLU361-LYS366
in β5-β6 loop (Fig. S7).
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.bioorg.2020.104391.
References
[
1] R.A. Birnbaum, A. O’Marcaigh, Z. Wardak, Y.Y. Zhang, G. Dranoff, T. Jacks, D.
W. Clapp, K.M. Shannon, Nf1 and Gmcsf interact in myeloid leukemogenesis, Mol.
Cell 5 (1) (2000) 189–195.
1
2

W.-Y. Li et al.
Bioorganic Chemistry 105 (2020) 104391
[
[
[
[
[
[
2] M.G. Mohi, I.R. Williams, C.R. Dearolf, G. Chan, J.L. Kutok, S. Cohen, K. Morgan,
C. Boulton, H. Shigematsu, H. Keilhack, K. Akashi, D.G. Gilliland, B.G. Neel,
Prognostic, therapeutic, and mechanistic implications of a mouse model of
leukemia evoked by Shp2 (PTPN11) mutations, Cancer Cell 7 (2) (2005) 179–191.
3] P.D. Emanuel, L.J. Bates, R.P. Castleberry, R.J. Gualtieri, K.S. Zuckerman, Selective
hypersensitivity to granulocyte-macrophage colony-stimulating factor by juvenile
chronic myeloid leukemia hematopoietic progenitors, Blood 77 (5) (1991)
throughput and highly parallel open source molecular simulation toolkit,
Bioinformatics 29(7) (2013) 845-854.
[25] S. Gharaghani, T. Khayamian, M. Ebrahimi, Molecular dynamics simulation study
and molecular docking descriptors in structure-based QSAR on
acetylcholinesterase (AChE) inhibitors, SAR QSAR Environ. Res. 24 (9) (2013)
773–794.
[26] W. Shinoda, M. Mikami, Rigid-body dynamics in the isothermal-isobaric ensemble:
A test on the accuracy and computational efficiency, J. Comput. Chem. 24 (8)
(2003) 920–930.
9
25–929.
4] S. Stier, Y. Ko, F. Forkert, C. Lutz, T. Neuhaus, E. Grunewald, T. Cheng,
D. Dombkowski, L.M. Calvi, S.R. Rittling, D.T. Scadden, Osteopontin is a
hematopoietic stem cell niche component that negatively regulates stem cell pool
size, J. Exp. Med. 201 (11) (2005) 1781–1791.
[27] B. Hess, P-LINCS: A parallel linear constraint solver for molecular simulation,
J. Chem. Theory Comput. 4 (1) (2008) 116–122.
[28] R. Salomon-Ferrer, A.W. Gotz, D. Poole, S. Le Grand, R.C. Walker, Routine
Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 2. Explicit
Solvent Particle Mesh Ewald, J. Chem. Theory Comput. 9 (9) (2013) 3878–3888.
[29] H. Wang, X.Y. Gao, J. Fang, Multiple Staggered Mesh Ewald: Boosting the Accuracy
of the Smooth Particle Mesh Ewald Method, J. Chem. Theory Comput. 12 (11)
(2016) 5596–5608.
5] M. Tartaglia, C.M. Niemeyer, A. Fragale, X.L. Song, J.C. Buechner, K. Hahlen,
H. Hasle, A. Jung, J.D. Licht, B.D. Gelb, Somatic PTPN11 mutations in juvenile
myelomonocytic leukemia, myelodysplastic syndromes and acute myeloid
leukemia, Pediatr. Res. 53 (4) (2003) 279a.
6] M.L. Loh, S. Vattikuti, S. Schubbert, M.G. Reynolds, E. Carlson, K.H. Lieuw, J.
W. Cheng, C.M. Lee, D. Stokoe, J.M. Bonifas, N.P. Curtiss, J. Gotlib, S. Meshinchi,
M.M. Le Beau, P.D. Emanuel, K.M. Shannon, Mutations in PTPN11 implicate the
SHP-2 phosphatase in leukemogenesis, Blood 103 (6) (2004) 2325–2331.
7] W. Liu, W.M. Yu, J. Zhang, R.J. Chan, M.L. Loh, Z. Zhang, K.D. Bunting, C.K. Qu,
Inhibition of the Gab2/PI3K/mTOR signaling ameliorates myeloid malignancy
caused by Ptpn11 (Shp2) gain-of-function mutations, Leukemia 31 (6) (2017)
[30] R. Kumari, R. Kumar, A. Lynn, O.S.D.D. Consort, g_mmpbsa-A GROMACS Tool for
High-Throughput MM-PBSA Calculations, J. Chem. Inf. Model. 54 (7) (2014)
1951–1962.
[31] B.R. Miller, T.D. McGee, J.M. Swails, N. Homeyer, H. Gohlke, A.E. Roitberg,
MMPBSA.py: An Efficient Program for End-State Free Energy Calculations, Journal
of chemical theory and computation 8(9) (2012) 3314-3321.
1
415–1422.
[32] Y. Yu, J. Wang, Q. Shao, J. Shi, W. Zhu, Effects of drug-resistant mutations on the
dynamic properties of HIV-1 protease and inhibition by Amprenavir and
Darunavir, Sci. Rep. 5 (2015) 10517.
[
[
8] P. Hof, S. Pluskey, S. Dhe-Paganon, M.J. Eck, S.E. Shoelson, Crystal structure of the
tyrosine phosphatase SHP-2, Cell 92 (4) (1998) 441–450.
9] D. Barford, B.G. Neel, Revealing mechanisms for SH2 domain mediated regulation
of the protein tyrosine phosphatase SHP-2, Structure 6 (3) (1998) 249–254.
[33] N. Homeyer, H. Gohlke, Free Energy Calculations by the Molecular Mechanics
Poisson-Boltzmann Surface Area Method, Mol. Inf. 31 (2) (2012) 114–122.
[34] B.J. Grant, J.A. McCammon, L.S. Caves, R.A. Cross, Multivariate analysis of
conserved sequence-structure relationships in kinesins: coupling of the active site
and a tubulin-binding sub-domain, J. Mol. Biol. 368 (5) (2007) 1231–1248.
[35] E. Papaleo, P. Mereghetti, P. Fantucci, R. Grandori, L. De Gioia, Free-energy
landscape, principal component analysis, and structural clustering to identify
representative conformations from molecular dynamics simulations: the myoglobin
case, J. Mol. Graph. Model. 27 (8) (2009) 889–899.
[
10] J. Li, H.B. Jie, Y. Lei, N. Gildener-Leapman, S. Trivedi, T. Green, L.P. Kane, R.
L. Ferris, PD-1/SHP-2 Inhibits Tc1/Th1 Phenotypic Responses and the Activation
of T Cells in the Tumor Microenvironment, Cancer Res. 75 (3) (2015) 508–518.
11] G. Chan, D. Kalaitzidis, B.G. Neel, The tyrosine phosphatase Shp2 (PTPN11) in
cancer, Cancer Metast Rev 27 (2) (2008) 179–192.
[
[
12] J. Xu, L.F. Zeng, W. Shen, J.J. Turchi, Z.Y. Zhang, Targeting SHP2 for EGFR
inhibitor resistant non-small cell lung carcinoma, Biochem. Biophys. Res. Commun.
4
39 (4) (2013) 586–590.
[36] D. Spiliotopoulos, A. Spitaleri, G. Musco, Exploring PHD fingers and H3K4me0
interactions with molecular dynamics simulations and binding free energy
calculations: AIRE-PHD1, a comparative study, PLoS ONE 7 (10) (2012) e46902.
[37] G.G. Maisuradze, A. Liwo, H.A. Scheraga, Relation between free energy landscapes
of proteins and dynamics, J. Chem. Theory Comput. 6 (2) (2010) 583–595.
[38] P.H. Hunenberger, A.E. Mark, W.F. van Gunsteren, Fluctuation and cross-
correlation analysis of protein motions observed in nanosecond molecular
dynamics simulations, J. Mol. Biol. 252 (4) (1995) 492–503.
[
13] L.F. Zeng, R.Y. Zhang, Z.H. Yu, S. Li, L. Wu, A.M. Gunawan, B.S. Lane, R.S. Mali,
X. Li, R.J. Chan, R. Kapur, C.D. Wells, Z.Y. Zhang, Therapeutic potential of
targeting the oncogenic SHP2 phosphatase, J. Med. Chem. 57 (15) (2014)
6
594–6609.
[
[
14] J. Chen, Z. Cao, J. Guan, SHP2 inhibitor PHPS1 protects against atherosclerosis by
inhibiting smooth muscle cell proliferation, BMC cardiovascular disorders 18 (1)
(
2018) 72.
15] K. Hellmuth, S. Grosskopf, C.T. Lum, M. Wurtele, N. Roder, J.P. von Kries,
M. Rosario, J. Rademann, W. Birchmeier, Specific inhibitors of the protein tyrosine
phosphatase Shp2 identified by high-throughput docking, PNAS 105 (20) (2008)
[39] T.J. Hou, J.M. Wang, Y.Y. Li, W. Wang, Assessing the Performance of the MM/
PBSA and MM/GBSA Methods. 1. The Accuracy of Binding Free Energy
Calculations Based on Molecular Dynamics Simulations, J. Chem. Inf. Model. 51
(1) (2011) 69–82.
7
275–7280.
[
[
[
[
16] H.M. Berman, T.N. Bhat, P.E. Bourne, Z.K. Feng, G. Gilliland, H. Weissig,
J. Westbrook, The Protein Data Bank and the challenge of structural genomics, Nat.
Struct. Biol. 7 (2000) 957–959.
[40] A.J.M. Martin, M. Vidotto, F. Boscariol, T. Di Domenico, I. Walsh, S.C.E. Tosatto,
RING: networking interacting residues, evolutionary information and energetics in
protein structures, Bioinformatics 27 (14) (2011) 2003–2005.
17] K.C. Chou, D.Q. Wei, W.Z. Zhong, Binding mechanism of coronavirus main
proteinase with ligands and its implication to drug design against SARS, Biochem.
Biophys. Res. Commun. 308 (1) (2003) 148–151.
[41] Y. Assenov, F. Ramirez, S.E. Schelhorn, T. Lengauer, M. Albrecht, Computing
topological parameters of biological networks, Bioinformatics 24 (2) (2008)
282–284.
18] Y. Ma, S.Q. Wang, W.R. Xu, R.L. Wang, K.C. Chou, Design Novel Dual Agonists for
Treating Type-2 Diabetes by Targeting Peroxisome Proliferator-Activated
Receptors with Core Hopping Approach, PLoS ONE 7 (6) (2012).
[42] N.T. Doncheva, Y. Assenov, F.S. Domingues, M. Albrecht, Topological analysis and
interactive visualization of biological networks and protein structures, Nat. Protoc.
7 (4) (2012) 670–685.
19] Y.Q. Duan, Y. Ma, X.J. Wang, Y.Y. Jin, R.L. Wang, W.L. Dong, W.R. Xu, D.X. Kong,
S.Q. Wang, Design Potential Selective Inhibitors for Treating Cancer by Targeting
the Src Homology 2 (SH2) Domain-Containing Phosphatase 2 (Shp2) with Core
Hopping Approach, Protein Pept. Lett. 21 (6) (2014) 556–563.
[43] M. Vendruscolo, N.V. Dokholyan, E. Paci, M. Karplus, Small-world view of the
amino acids that play a key role in protein folding, Phys. Rev. E 65 (6) (2002).
[44] X. Zhang, Y. He, S. Liu, Z. Yu, Z.X. Jiang, Z. Yang, Y. Dong, S.C. Nabinger, L. Wu, A.
M. Gunawan, L. Wang, R.J. Chan, Z.Y. Zhang, Salicylic acid based small molecule
inhibitor for the oncogenic Src homology-2 domain containing protein tyrosine
phosphatase-2 (SHP2), J. Med. Chem. 53 (6) (2010) 2482–2493.
[45] O. Carugo, S. Pongor, A normalized root-mean-square distance for comparing
protein three-dimensional structures, Protein science : a publication of the Protein
Society 10 (7) (2001) 1470–1473.
[
[
[
[
20] A. Bak, T. Magdziarz, J. Polanski, Pharmacophore-based database mining for
probing fragmental drug-likeness of diketo acid analogues, SAR QSAR Environ.
Res. 23 (1–2) (2012) 185–204.
21] D.K. Nevin, M.B. Peters, G. Carta, D. Fayne, D.G. Lloyd, Integrated Virtual
Screening for the Identification of Novel and Selective Peroxisome Proliferator-
Activated Receptor (PPAR) Scaffolds, J. Med. Chem. 55 (11) (2012) 4978–4989.
22] D.F. McCain, L. Wu, P. Nickel, M.U. Kassack, A. Kreimeyer, A. Gagliardi, D.
C. Collins, Z.Y. Zhang, Suramin derivatives as inhibitors and activators of protein-
tyrosine phosphatases, J. Biol. Chem. 279 (15) (2004) 14713–14725.
23] X.Q. Wu, G. Xu, X.B. Li, W.R. Xu, Q.J. Li, W. Liu, K.A. Kirby, M.L. Loh, J. Li, S.
G. Sarafianos, C.K. Qu, Small Molecule Inhibitor that Stabilizes the Autoinhibited
Conformation of the Oncogenic Tyrosine Phosphatase SHP2, J. Med. Chem. 62 (3)
[46] A. Wolf, K.N. Kirschner, Principal component and clustering analysis on molecular
dynamics data of the ribosomal L11 center dot 23S subdomain, J. Mol. Model. 19
(2) (2013) 539–549.
[47] S.L. Williams, J.W. Essex, Study of the Conformational Dynamics of the Catalytic
Loop of WT and G140A/G149A HIV-1 Integrase Core Domain Using Reversible
Digitally Filtered Molecular Dynamics, J. Chem. Theory Comput. 5 (2) (2009)
411–421.
(
2019) 1125–1137.
[48] C. Arcangeli, A.R. Bizzarri, S. Cannistraro, Molecular dynamics simulation and
essential dynamics study of mutated plastocyanin: structural, dynamical and
functional effects of a disulfide bridge insertion at the protein surface, Biophys.
Chem. 92 (3) (2001) 183–199.
[
24] S. Pronk, S. Pall, R. Schulz, P. Larsson, P. Bjelkmar, R. Apostolov, M.R. Shirts, J.C.
Smith, P.M. Kasson, D. van der Spoel, B. Hess, E. Lindahl, GROMACS 4.5: a high-
1
3