Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis, biological evaluation and molecular dynamics simulation

Article abstract of DOI:10.1016/j.bbrc.2020.03.075

Protein tyrosine phosphatase 1B (PTP1B) is a widely expressed 50 kDa enzyme and the first intracellular PTP to be purified from human placental tissue. It has been proved that protein tyrosine phosphatase 1B played a significant role in the negative regul

Full text of DOI:10.1016/j.bbrc.2020.03.075

Biochemical and Biophysical Research Communications xxx (xxxx) xxx
Contents lists available at ScienceDirect
Biochemical and Biophysical Research Communications
journal homepage: www.elsevier.com/locate/ybbrc
Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors
through De Novo Evoluton, synthesis, biological evaluation and
molecular dynamics simulation
Jingwei Wu 1, Yangchun Ma 1, Hui Zhou, Liang Zhou, Shan Du, Yingzhan Sun, Weiya Li,
Weili Dong*, Runling Wang**
Tianjin Key Laboratory on Technologies Enabling Development of Clinical Therapeutics and Diagnostics (Theranostics), School of Pharmacy, Tianjin Medical
University, Tianjin, 300070, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Protein tyrosine phosphatase 1B (PTP1B) is a widely expressed 50 kDa enzyme and the first intracellular
PTP to be purified from human placental tissue. It has been proved that protein tyrosine phosphatase 1B
played a significant role in the negative regulation of insulin signaling pathway and overexpression of
PTP1B could lead to the decrease of insulin resistance. Therefore PTP1B has emerged as a novel promising
therapeutic target for the treatment of type-2 diabetes mellitus. Computer aided drug design (CADD),
chemical synthesis and biological activity assay resulted in the identification of a novel potent PTP1B
Received 1 March 2020
Accepted 12 March 2020
Available online xxx
Keywords:
PTP1B
Type-2 diabetes mellitus
Computer aided drug design
De Novo Evoluton
Molecular dynamics simulation
inhibitor, compound 1a, which shared an IC50 value of 4.46
namics simulation provided the theoretical basis for favorable activity of compound 1a.
© 2020 Elsevier Inc. All rights reserved.
mM. Finally, the analysis of molecular dy-
1. Introduction
and localized mainly in the endoplasmic reticulum [6]. It has been
proved that protein tyrosine phosphatase 1B played a significant
role in the negative regulation of insulin signaling pathway and
overexpression of PTP1B could lead to the decrease of insulin
resistance [7]. Another evidence for the correlation between insulin
resistance and PTP1B was provided by PTP1B knockout mice, which
could improve glucose tolerance and insulin sensitivity [8]. Hence,
PTP1B has emerged as a novel promising therapeutic target for the
treatment of T2-DM.
Recently, a lot of novel PTP1B inhibitors have been developed by
research institutions, including carboxylic acids, phosphonic acids,
sulphonic acids, imides, phosphonodifluoromethyl phenylalanine
derivatives and vanadium compounds [9]. Among them, several
drug candidates have been developed by pharmaceutical com-
panies, such as TTP-814, ISIS 113715, MSI-1436, way-123783, ISIS-
PTP1BRx and MSI-1436 [9,10]. However, the reported inhibitors
cause some side effects and do not have sufficient potency to
demonstrate in vivo activity [11,12]. Hence, it is still a challenge to
discover potent, permeable, selective PTP1B inhibitors for treating
type-2 diabetes mellitus.
Diabetes mellitus, the major threat to human health worldwide,
is a chronic disease that results from either insulin deficiency or
insulin resistance, or both [1]. The World Health Organization
(WHO) reports indicated that there were 425 million new cases of
diabetes around the world in 2017 and the global prevalence was
projected to continue to rise to 629 million by 2045 [2]. However,
cases of diabetes are on the rise, and its complications are torturing
and killing more and more people. type-2 diabetes mellitus is
noninsulin-dependent diabetes mellitus and characterized by in-
sulin resistance (IR) and relatively lack of insulin [3].
Protein tyrosine phosphatases (PTPs) are a group of enzymes
regulating the dephosphorylation of tyrosine (Tyr)-phosphorylated
proteins and the cellular function [4]. A lot of PTPs can function
either as positive or negative modulators in a variety of signal
transduction pathways [5]. Protein tyrosine phosphatase 1B
(PTP1B), the first purified PTP, is a widely expressed 50 kDa enzyme
* Corresponding author.
** Corresponding author.
In view of this situation, we designed a series of PTP1B inhibitors
with the aid of virtual screening, De Novo Evolution, docking and
ADMET (absorption, distribution, metabolism, excretion and
toxicity) validating. For purpose of finding novel potent PTP1B
E-mail addresses: dongweili@tmu.edu.cn (W. Dong), wangrunling@tmu.edu.cn
(R. Wang).
1
These authors contributed equally to this work.
https://doi.org/10.1016/j.bbrc.2020.03.075
0006-291X/© 2020 Elsevier Inc. All rights reserved.
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075

2
J. Wu et al. / Biochemical and Biophysical Research Communications xxx (xxxx) xxx
inhibitors for the treatment of T2-DM, a classes of ethyl 4-(sub-
stitutedphenoxymethyl)thiazole-5-carboxylates were synthesized.
Then, the PTP1B biological activity was tested. Meanwhile, molec-
ular dynamics simulations and several post-analysis experiments
were carried out to analyze the binding interactions between
PTP1B and the desired compounds, such as the root mean square
deviation analysis (RMSD), the root mean square fluctuation anal-
ysis (RMSF), principal component analysis (PCA) and Domain cross-
correlation map (DCCM) analysis. The flow chart of the test pro-
cedure was shown in Fig. 1. Hopefully, the findings could provide
useful insights for developing novel potent PTP1B inhibitors against
type-2 diabetes mellitus and provide promising potential drug
candidate.
PTP1B. Fourthly, the procedure of “De Novo Link” was adopted to
add appropriate fragments to the existing scaffold. The maximum
generation of designed derivatives was set to 1000. Finally, the
generated derivatives in the top 100 scored were then forward for
the protocol of “CDOCKER”.
2.3. Molecular docking
Molecular docking technology is an important part of structure-
based drug design, which can find the best binding mode between
receptor and ligand according to the principle of geometric
complementarity, energy complementarity, and chemical envi-
ronment complementarity. Many docking methods have been re-
ported in recent years and some useful comparisons of the re-
docking ability of these algorithms have been published [21].
CDOCKER exhibited the best performing among the reported
docking software for the high accuracy of docking and scoring [22].
Since that is so, the generated derivatives in the top 100 scored
were re-docked into to the PTP1B binding pocket by the protocol of
“CDOCKER”. According to the -CDOCKER_ ENERGY, the best hits
were adopted to perform the procedure of ADMET prediction.
2. Materials and methods
2.1. Virtual screening
Docking-based virtual screening has been proved to be an
effective tool in the process of drug design and optimization and
widely used in the development of new drugs [13]. For the sake of
high-level efficiency [14], the protocol of “LibDock” was selected to
perform virtual screening in this paper. The PTP1B protein structure
(2QBQ) [15]downloaded from the RCSB PDB library needs to pre-
pare with the protocol of “Prepare Protein”. Meanwhile, the pro-
tocol of “Prepare Ligands” was used to prepare the small molecules
retrieved from ZINC database. Finally, the prepared ligands were
docked into the active site sphere, which contains P-loop (residues
Thr177-Ser187) and WPD-loop (residues His214-Arg221) [16]. Ac-
cording to the docking score, the best hit was used to generate
derivatives with the method of “De Novo Evolution”.
2.4. Absorption, distribution, metabolism, excretion and toxicity
(ADMET) prediction
Here we investigated the ADMET features by performing
“ADMET Descriptions” protocol to estimate the following proper-
ties: the logarithm of the partition coefficient between n-octanol
and water (CLogP), aqueous solubility blood-brain barrier (BBB)
penetration, absorption level, plasma protein binding (PPB) and
CYP450 2D6 inhibition [23]. The potential candidates were then
forward for chemical synthesis and biological activity assay.
2.2. De Novo Evolution
2.5. Chemistry
In order to design potent PTP1B inhibitors containing novel
scaffolds, the protocol of “De Novo Evolution” based on Ludi algo-
rithm was carried out by evaluating fragments that best comple-
ment the receptor [17]. The fragments which complement to the
receptor were added to the existing scaffold and the binding af-
finity to PTP1B would be improved [18,19]. The procedure of “De
Novo Evolution” could be divided into four steps. Firstly, the
conformation of the PTP1B and the scaffold is calculated by the
CHARMm position [20]. Secondly, the “De Novo Library Genera-
tion” module was selected to generate fragment libraries. Thirdly,
“De novo receptor” protocol was applied to find the binding site of
The synthetic route to the desired products 1a-1j is illustrated in
Schemes 1. The detailed information for the synthesis of interme-
diate (3a-3e and 4a-4e) was listed in the supporting information.
Reagents and conditions: (ⅰ).ethyl 2-chloro-3-oxobutanoate,
EtOH, reflux; (ⅱ). NBS, AIBN, CCl4, 80 ꢀC; (ⅲ). 4-substitute phenol,
K2CO3, MeCN, reflux.
Synthesis of target compounds 1a-1j: A mixture of 4a-4e
(3 mmol), 4-substitute phenol (3.6 mmol) and K2CO3 (6 mmol,
0.83 g) in MeCN (10 mL) was stirred at reflux until the completion
of reaction as indicated by TLC analysis [24].
Fig. 1. The flowchart of the test procedure.
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075

J. Wu et al. / Biochemical and Biophysical Research Communications xxx (xxxx) xxx
3
Scheme 1. Synthetic routes to target compounds 1a-1j.
After cooling to room temperature, the volatiles were removed
2.6. Biological evaluation
under reduced pressure, and the resulting residue was diluted with
CH2Cl2 (30 mL), H2O (30 mL). The phases were separated and the
aqueous was futher extracted with CH2Cl2 (30 mL ꢁ 3). The com-
bined extracts were dried over anhydrous Na2SO4 and concentrated.
Finally, the resulting residue was purified by silica chromatography
by EtOAc/n-hexane (1/5 by v/v) to afford 1a-1j. The NMR/MS spectra
datas were listed in the supporting information.
The inhibitory activity of PTP1B, TCPTP, PTP-LAR, CDC25B, SHP1
and MEG2 were tested in 96-well plates at 37 ꢀC and p-nitrophenyl
phosphate (pNPP) activity was measured according to the release of
p-nitrophenyl from p-nitrophenylphosphate, which was deter-
mined spectrophotometrically at 405 nm [25]. The reaction
mixture contained1mM-(Ethylenedinitrilo)tetraacetic acid,10 mM-
Table 1
Molecular docking results for the derivatives (top-10) generated by de novo evolution protocol from ZINC39276654 (scaffold).
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075

4
J. Wu et al. / Biochemical and Biophysical Research Communications xxx (xxxx) xxx
pNPP, 50 mM-citrate, 0.1M-sodium chloride, and 1 mM-DL-
dithiothreitol and incubated for 40 min at 37 ꢀC. The reaction was
stopped by the addition of 5 L of 1M-NaOH.
m
2.7. Molecular dynamics simulations
The methods of molecular dynamics simulations have been
listed in the supporting information.
3. Results and discussion
3.1. Compounds design
It could be seen from Fig. 1 and Table 1 that ZINC39276654 was
identified in the protocol of LibDock-based virtual screening. As the
best scaffold, ZINC39276654 produced 1000 molecules by the “De
Novo Evolution” procedure. Then the generated derivatives in the
top 100 scored were selected and re-docked into the receptor
(PTPT1B) to compute the -CDOCKER_ENERGY by the protocol of
“CDOCKER” According to the -CDOCKER_ENERGY, the results of top
10 compounds (1a-1j) and ZINC39276654 (scaffold) were shown in
Table 1. The -CDOCKER_ENERGY includes ligand-receptor interac-
tion energy and internal ligand strain energy. The higher the value,
the more favorable the binding mode [26]. It is obvious that the
compounds 1a-1j shared the more favorable binding mode than
ZINC39276654 (scaffold). Among them, compound 1a shared the
highest -CDOCKER_ENERGY (59.94) and was used to the following
analysis of docking conformation and binding interactions.
As shown in Fig. 2A, the oxygen atoms of compound 1a formed
four H-bonds with the residues of Cys215, Ser216, Ile219 and
Gly220 (green-dashed arrow in Fig. 2A), which were all in the active
site of PTP1B. The compound 1a could also generate H-Pi in-
teractions (red-dashed arrow in Fig. 2A) with the residue of Ala217.
As shown in Fig. 2B, it was obviously that the compound 1a could
embed well into the active pocket of PTP1B and left a fraction of the
atoms in the state of exposure (blue shadows of the ligand),
resulting in the strong binding affinities. Greasy residues were
shown as green circles (Val49, Pro180, Phe182, Ala217, Ile219 and
Met258) while polar residues were shown as pink circles (Tyr46,
Ser118, Gly183, Cys215, Ser216, Gly218, Gly220, Arg221, Ser222,
Gly259, Gln262 and Gln266).
Fig. 2. Structures and binding interactions of PTPP1B in complex with compound 1a.
(A) Surface representation of PTP1B in complex with compound 1a (green). Hydrogen
bond interactions were depicted as green-dashed lines; H-Pi interaction was depicted
as red-dashed lines. (B) 2D diagram of the interaction of compound 1a-PTP1B. (For
interpretation of the references to color in this figure legend, the reader is referred to
the Web version of this article.)
3.2. ADMET prediction
We also investigated the ADMET prediction by performing
“ADMET Descriptions” protocol to estimate the following
Table 2
The ADMET prediction for the target compounds.
Name
CLogP
Solubility levela
CYP2D6 Prediction
BBB-Levelb
Hepatotoxic#
Prediction
PPB
Prediction
Absorption
-leveld
1a
1b
1c
1d
1e
1f
1g
1h
1i
5.545
5.955
5.499
4.728
5.138
7.488
7.842
5.535
5.845
5.799
3
3
3
4
3
2
2
3
3
3
False
False
False
False
False
False
False
False
False
False
0
1
1
0
1
2
0
2
1
1
False
False
False
False
False
False
False
False
False
False
True
True
True
True
True
True
True
True
True
True
good
good
moderate
moderate
good
moderate
good
poor
good
moderate
1j
a
Solubility level: 0-extremely low; 1-very low, but possible; 2-low; 3-good; 4-optimal.
BBB level: 0-very good; 1-good; 2-moderate; 3-poor; 4-undefined.
b
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075

J. Wu et al. / Biochemical and Biophysical Research Communications xxx (xxxx) xxx
5
Table 3
Inhibitory activity of compound 1a-1j and ZINC39276654 against PTP1B and other phosphatases.
Compounds
PTP1B IC50
(m
M)
TCPTP IC50
(m
M)
SHP1 IC50
(
m
M)
MEG2 IC50
(
mM)
CDC25B IC50
(m
M)
LAR IC50 (mM)
ZINC39276654
>100
4.46
>100
12.28
31.02
>100
38.65
36.68
35.91
>100
39.35
27.68
>100
>100
>100
>100
>100
>100
>100
33.26
>100
39.08
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
1a
1b
1c
1d
1e
1f
1g
1h
1i
10.41
10.79
11.58
12.05
11.22
10.73
9.69
8.03
64.75
1j
inhibitory activity (IC50 ¼ 64.75
mM) to ZINC39276654. Besides, it
was clear that the worst and the best R2 group were 4-nitro and 4-
methoxy (1a > 1b > 1c, 1d > 1e, 1i > 1j), while the best and the
worst R1 group were 2-bromol and 4-bromol respectively (1a > 1i,
1c > 1j). Besides, the biological evaluation of inhibitors 1a-1j led to
the development of a novel structure PTP1B inhibitor 1a, which
shared an IC50 value of 4.46 mM.
In order to investigate the selectivity of the compounds 1a-1j,
their biological activity on similar PTPs were assayed, which con-
sisted of TCPTP, SHP1, CDC25B, LAR and MEG2. It was very clear that
compounds 1a-1i did not show any promising activity on CDC25B,
MEG2 and LAR. Meanwhile compounds 1f and 1h exhibited mod-
erate SHP1 inhibitory activity (IC50
¼
33.26 mM for 1f,
IC50 ¼ 39.08 M for 1h). Compound 1c and 1g exhibited favorable
m
selectivity (more than 10-fold) for PTP1B over TCPTP, seven com-
pounds (1a-1b, 1d-1f, 1h-1i) showed moderate selectivity (2.8-
folde4.1-fold). Unfortunately, compound 1j did not show any
selectivity. Among the desired compounds, compound 1g showed
the strongest selectivity for PTP1B over the other phosphatases.
3.4. Molecular dynamics simulations
Fig. 3. (A) Standard RMSD of PTP1B system and PTP1B-1a system. (B) RMSF during
3e50 ns. The PTP1B system is in black solid line and the PTP1B-1a complex system is in
red solid line. Large difference between areas marked by yellow and black boxes. (For
interpretation of the references to color in this figure legend, the reader is referred to
the Web version of this article.)
3.4.1. Root mean square deviation (RMSD)
In this study, the trace files of the MD simulation were used to
perform RMSD analysis on the PTP1B protein system and the pro-
tein ligand complex system, respectively. The results of RMSD
analysis can be used as a measure to evaluate the overall fluctuation
of the protein’s main chain atoms. The value of RMSD has a negative
correlation with the stability of the main chain atoms. The larger
the value of RMSD, the more unstable the main chain atoms are. It
can be found in Fig. 3A that the protein system tends to stabilize
after 3 ns, so the MD file after 3 ns is used for MD simulation
analysis. In the SHP2 system, the average RMSD of 3e50 ns is
0.317 nm. However, in the protein-ligand complex (PTP1B-1a)
system, the mean value of RMSD of 3e50 ns is 0.281 nm. In com-
parison, the average RMSD of the PTP1B-1a complex system is
lower. At the same time, it can be clearly found that the overall
fluctuation of RMSD of the protein-ligand complex system is lower
than that of the PTP1B system in Fig. 3A, which indicates that the
protein-ligand complex system is more stable. Therefore, the
simultaneous effects of compound 1a and active sites have stabi-
lized the overall conformation of the PTP1B protein.
properties: the logarithm of the partition coefficient between n-
octanol and water (CLogP), aqueous solubility blood-brain barrier
(BBB) penetration, absorption level, plasma protein binding (PPB)
and CYP450 2D6 inhibition. It could be seen from Table 2 that most
compounds exhibited moderate or good human intestinal absorp-
tion except 1h (poor level). The distribution properties were
calculated using predictive plasma protein binding, which was
widely used because the degree of plasma protein binding of a drug
has an important role on its disposition and the drug’s efficacy [27].
Fortunately, the compounds 1a-1j exhibited good PPB properties.
For hepatotoxicity, ten compounds (1a-1j) were predicted non-
toxic. It was obviously that compounds 1a-1j exhibited accept-
able pharmaceutical properties, so the potential candidates were
then forward for chemical synthesis and biological activity assay.
3.3. Biological evaluation
3.4.2. Root mean square fluctuation (RMSF) analyses
In order to better study the effect of compound 1a on the active
regions of the side chain atoms of the PTP1B protein, RMSF analysis
technology was applied to analyze the trajectory files generated by
MD simulations. Fig. 3B is a graph of RMSF data generated by the
PTP1B system and the protein-ligand complex system in 3e50 ns.
The RMSF value also has a negative correlation with the side chain
The results of biological evaluation of compounds 1a-1j and
ZINC39276654 (scaffold) on PTP1B were listed in Table 3. It was
obviously that the inhibitors 1a-1i showed better PTP1B inhibitory
activity (IC50 values were range from 4.46
mM to 12.05 mM) than
ZINC39276654 (IC50 > 100 M) while compound 1j showed similar
m
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075

6
J. Wu et al. / Biochemical and Biophysical Research Communications xxx (xxxx) xxx
atoms. The smaller the RMSF value, the more stable the side chain
atoms. The areas with large differences in RMSF fluctuations in the
two systems were marked with a yellow box and a black box,
respectively. The amino acid regions marked by yellow boxes were
residue Ser50-Arg56, Thr154-Thr164. However, since the region
marked by the yellow ellipse was not in the active site region, it will
not be discussed in detail. The areas marked by the black square
boxes were residue Val113-Ser118 (R-loop), Thr177-Pro185 (WPD-
loop) and His214-Arg221 (active site). It can be seen from Fig. 3B
that the RMSF fluctuations in the black box regions have significant
differences (RMSF value of the protein-ligand complex system less
than PTP1B system). The average RMSF values of residues Val113-
Ser118, Thr177-Pro185, and His214-Arg221 in the PTP1B protein
were 0.180 nm, 0.132 nm, and 0.138 nm, respectively, while in the
PTP1B-1a system, 0.125 nm, 0.102 nm, and 0.120 nm, respectively.
Obviously, the fluctuation of RMSF in the PTP1B-1a system was
significantly lower than that in the PTP1B system, which indicates
that the side chain residues in the protein-ligand complex system
become more stable. It is worth noting that the residues Val113-
Ser118, Thr177-Pro185, and His214-Arg221 are all around the
active pocket, which suggests that compound 1a makes the active
site of the PTP1B protein more stable.
Fig. 4. Principal component analysis. Projection of trajectories into PC1 and PC2 for (A) PTP1B system (B) PTP1B-1a complex system. (C) DCCM for the PTP1B-1a system showing
positive and negative correlative motions between the residues. (D) DCCM for PTP1B-1a complex system showing positive and negative correlative motions between residues.
Green represents positive correlations, whereas red represents negative correlations. (For interpretation of the references to color in this figure legend, the reader is referred to the
Web version of this article.)
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075

J. Wu et al. / Biochemical and Biophysical Research Communications xxx (xxxx) xxx
7
3.4.3. Conformation transitions of the PTP1B protein and PTP1B-1a
system
PCA analysis makes the three-dimensional conformational in-
4. Conclusions
Computer aided drug design (CADD), chemical synthesis and
biological activity assay resulted in the identification of a novel
potent PTP1B inhibitor, compound 1a which shared which shared
formation of PTP1B more intuitive in the two-dimensional plane,
which is more conducive to the analysis of the conformational state
of PTP1B in the 3e50 ns MD simulation. The anharmonic and large-
scale motions in proteins are closely related to biological functions,
so the anharmonic and large-scale motions of proteins are separated
from other motions by diagonalization of covariance matrix and
reflected in the form of eigenvalue eigenvectors. It can be found in
Fig. 4 that in the MD simulation of 3e50 ns, the top 20 ps of PTP1B
system and PTP1B-1a system account for 82.4% and 79.1% of the total
variation respectively. In the PTP1B system, the first two PC (PC1,
PC2) eigenvectors account for 38.6% and 13.5% of the total variation
respectively, while the other PCs account for no more than 8.2% of
the maximum. In the PTP1B-1a system, PC1 and PC2 account for
30.4% and 12.3% respectively, while other PCs account for no more
than 8.8%. Obviously, PC1 and PC2 account for most of the protein
fluctuations. Therefore, PCA dots images generated from the first two
eigenvectors are used to observe the conformational changes of
PTP1B. The blue and red dots in Fig. 4 represent different confor-
mational states of PTP1B protein, while the white dots represent the
unstable intermediate state of PTP1B protein. It can be found from
Fig. 4A that the points of PTP1B system are evenly distributed near
the midline, while the points of PTP1B-1a system (Fig. 4B) are
concentrated on both sides of the midline. The closer the point
distribution in Fig. 4 indicates that the protein system is more stable,
so the PTP1B-1a system is in a more stable state. In conclusion, the
binding of compound 1a has a stable effect on the conformation of
PTP1B. PCA scores are consistent with the above analysis results.
an IC50 value of 4.46 mM. Among the desired compounds 1a-1j,
compound 1g showed the strongest selectivity for PTP1B over the
other phosphatases while compound 1j did not show any selec-
tivity. Finally, the analysis of molecular dynamics provided the
theoretical basis for favorable activity of compound 1a.
Declaration of competing interest
There is no conflict of interest.
Acknowledgments
This study was supported by the National Natural Science
Foundation of China (Grant No. 81773569), the Natural Science
Foundation of Tianjin City, China (Grant No. 18JCQNJC13700).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.bbrc.2020.03.075.
References
[1] H.C. Kenny, E.D. Abel, Heart failure in type 2 diabetes mellitus: impact of
glucose-lowering agents, heart failure therapies, and novel therapeutic stra-
tegies, Circ. Res. 124 (2019) 121e141, https://doi.org/10.1161/
CIRCRESAHA.118.311371.
[2] N.G. Forouhi, N.J. Wareham, Epidemiology of diabetes, Medicine 47 (2019)
22e27, https://doi.org/10.1016/j.mpmed.2018.10.004.
3.4.4. The correlation motion of the PTP1B protein and PTP1B-1a
system
[3] J. Sun, C. Qu, Y. Wang, et al., PTP1B, a potential target of type 2 diabetes
mellitus, Mol. Biol. 5 (2016) 174, https://doi.org/10.4172/2168-9547.1000174.
[4] N.K. Tonks, Protein tyrosine phosphatases: from genes, to function, to disease,
Nat. Rev. Mol. Cell Biol. 7 (2006) 833e846, https://doi.org/10.1038/nrm2039.
[5] L.R. Bollu, A. Mazumdar, M.I. Savage, et al., Molecular pathways: targeting
protein tyrosine phosphatases in cancer, Clin. Canc. Res. 23 (2017)
2136e2142, https://doi.org/10.1158/1078-0432.CCR-16-0934.
From the above analysis results, it can be inferred that the
fluctuation of WPD loop and active site in PTP1B-1a system is
significantly reduced. In order to further explore the fluctuation of
these two regions in PTP1B, DCCM data generated from 3 to 50 ns
MD simulation was used to analyze the movement trend of Ca atom
in the dynamic simulation process. In Fig. 4C, the abscissa and
ordinate are arranged according to the sequence of amino acid
residues, and the relationship between the upper left corner and
the upper right corner is symmetrically distributed according to the
diagonal. The correlation motion and anti-correlation motion be-
tween specific residues are represented by green (syn motion) and
red (anti motion) respectively, and the intensity of correlation
motion is positively correlated with the depth of color. In Fig. 4D,
the regions with significant differences in correlation between
WPD loop and active sites were labeled with black boxes. Obvi-
ously, in PTP1B-1a complex system (Fig. 4D), the interaction be-
tween amino acid residues was enhanced obviously, indicating that
compound 1a enhanced the correlation within the protein. In
PTP1B-1a complex system, the WPD loop (Thr177-Pro185) of the
protein formed a strong anti-correlation, which was different from
the residues Ser70-Glu75 and active site (His214-Arg211), but the
anti-correlation was weakened or disappeared in PTP1B protein
system. In the PTP1B-1a complex system, active site (His214-
[6] H. Cho, Protein Tyrosine Phosphatase 1B (PTP1B) and Obesity, Vitamins &
Hormones, Elsevier, 2013, pp. 405e424, https://doi.org/10.1016/B978-0-12-
407766-9.00017-1.
~
[7] G.K. Kuga, V.R. Munoz, R.C. Gaspar, et al., Impaired insulin signaling and
spatial learning in middle-aged rats: the role of PTP1B, Exp. Gerontol. 104
(2018) 66e71, https://doi.org/10.1016/j.exger.2018.02.005.
[8] B. Boivin, R.S. Sagabala, Role of cholesterol and oxysterols in regulating PTP1B
activity, Free Radic. Biol. Med. 112 (2017) 146, https://doi.org/10.1016/
j.freeradbiomed.2017.10.223.
[9] H. Hussain, I.R. Green, G. Abbas, et al., Protein tyrosine phosphatase 1B
(PTP1B) inhibitors as potential anti-diabetes agents: patent review (2015-
2018), Expert Opin. Ther. Pat. 29 (2019) 689e702, https://doi.org/10.1080/
13543776.2019.1655.
[10] L.-J. Wang, B. Jiang, N. Wu, et al., Small molecules as potent protein tyrosine
phosphatase 1B (PTP1B) inhibitors documented in patents from 2009 to 2013,
Mini Rev. Med. Chem. 15 (2015) 104e122, https://doi.org/10.2174/
1389557515666150203144339.
[11] Y. Han, M. Belley, C.I. Bayly, et al., Discovery of [(3-bromo-7-cyano-2-
naphthyl)(difluoro) methyl] phosphonic acid,
a potent and orally active
small molecule PTP1B inhibitor, Bioorg. Med. Chem. Lett 18 (2008)
3200e3205, https://doi.org/10.1016/j.bmcl.2008.04.064.
[12] L.R. Fiedler, Inhibiting the inhibitors, PTP1B as
a therapeutic target in
myocardial infarction, Heart Res-Open J. 5 (2018), https://doi.org/10.17140/
HROJ-5-147.
Arg211) and residue Glu200-Ser205 form
a
strong anti-
[13] J. Wu, Y. Sun, H. Zhou, et al., Design, synthesis, biological evaluation and
molecular dynamics simulation studies of (R)-5-methylthiazolidin-4-One
derivatives as megakaryocyte protein tyrosine phosphatase 2 (PTP-MEG2)
inhibitors for the treatment of type 2 diabetes, J. Biomol. Struct. Dynam.
(2019) 1e10, https://doi.org/10.1080/07391102.2019.1654410.
[14] J. Wu, W. Li, Z. Zheng, et al., Design, synthesis, biological evaluation, Common
feature pharmacophore model and molecular dynamics simulation studies of
ethyl 4-(phenoxymethyl)-2-phenylthiazole-5-carboxylate as Src Homology-2
Domain Containing Protein Tyrosine Phosphatase-2 (SHP2) inhibitors,
J. Biomol. Struct. Dynam. (2020) 1e29, https://doi.org/10.1080/
07391102.2020.1726817.
correlation, while in the PTP1B protein system, the anti-
correlation is weakened. The enhancement of the anti-correlation
of the above regions indicated that the amino acid residues
moved in the opposite direction. In a word, we can find that the
anti-correlation between WPD loop and active site area is signifi-
cantly enhanced, which indicates that the interaction between
amino acid residues in these two areas is increased, which will
make the protein in a more stable state.
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075

8
J. Wu et al. / Biochemical and Biophysical Research Communications xxx (xxxx) xxx
[15] D.P. Wilson, Z.-K. Wan, W.-X. Xu, et al., Structure-based optimization of
protein tyrosine phosphatase 1B inhibitors: from the active site to the second
phosphotyrosine binding site, J. Med. Chem. 50 (2007) 4681e4698, https://
doi.org/10.1021/jm0702478.
[22] J.K. Gagnon, S.M. Law, C.L. Brooks III, Flexible CDOCKER: development and
application of pseudo-explicit structure-based docking method within
a
CHARMM, J. Comput. Chem. 37 (2016) 753e762, https://doi.org/10.1002/
jcc.24259.
[16] S. Liu, L.-F. Zeng, L. Wu, et al., Targeting inactive enzyme conformation: aryl
diketoacid derivatives as a new class of PTP1B inhibitors, J. Am. Chem. Soc.
130 (2008) 17075e17084, https://doi.org/10.1021/ja8068177.
[17] T.-T. Chang, M.-F. Sun, H.-Y. Chen, et al., Screening from the world’s largest
TCM database against H1N1 virus, J. Biomol. Struct. Dynam. 28 (2011)
773e786, https://doi.org/10.1080/07391102.2011.10508605.
[18] Y. Jiang, H. Gao, Pharmacophore-based drug design for the identification of
novel butyrylcholinesterase inhibitors against Alzheimer’s disease, Phyto-
medicine 54 (2019) 278e290, https://doi.org/10.1016/j.phymed.2018.09.199.
[19] C.Y.-C. Chen, Weighted equation and rulesda novel concept for evaluating
protein-ligand interaction, J. Biomol. Struct. Dynam. 27 (2009) 271e282,
https://doi.org/10.1080/07391102.2009.10507315.
[20] B.R. Brooks, C.L. Brooks III, A.D. Mackerell Jr., et al., CHARMM: the biomole-
cular simulation program, J. Comput. Chem. 30 (2009) 1545e1614, https://
doi.org/10.1002/jcc.21287.
[23] F. Cheng, W. Li, G. Liu, et al., In silico ADMET prediction: recent advances,
current challenges and future trends, Curr. Top. Med. Chem. 13 (2013)
1273e1289, https://doi.org/10.2174/15680266113139990033.
[24] J.-w. Wu, L. Yin, Y.-q. Liu, et al., Synthesis, biological evaluation and 3D-QSAR
studies of 1, 2, 4-triazole-5-substituted carboxylic acid bioisosteres as uric
acid transporter 1 (URAT1) inhibitors for the treatment of hyperuricemia
associated with gout, Bioorg. Med. Chem. Lett 29 (2019) 383e388, https://
doi.org/10.1016/j.bmcl.2018.12.036.
[25] J. Montalibet, K.I. Skorey, B.P. Kennedy, Protein tyrosine phosphatase: enzy-
matic assays, Methods 35 (2005) 2e8, https://doi.org/10.1016/
j.ymeth.2004.07.002.
[26] G. Wu, D.H. Robertson, C.L. Brooks III, et al., Detailed analysis of grid-based
molecular docking: a case study of CDOCKERda CHARMm-based MD docking
algorithm, J. Comput. Chem. 24 (2003) 1549e1562, https://doi.org/10.1002/
jcc.10306.
[21] J.B. Cross, D.C. Thompson, B.K. Rai, et al., Comparison of several molecular
docking programs: pose prediction and virtual screening accuracy, J. Chem.
Inf. Model. 49 (2009) 1455e1474, https://doi.org/10.1021/ci900056c.
[27] J. Dong, N.-N. Wang, Z.-J. Yao, et al., ADMETlab: a platform for systematic
ADMET evaluation based on a comprehensively collected ADMET database,
J. Cheminf. 10 (2018) 29, https://doi.org/10.1186/s13321-018-0283-x.
Please cite this article as: J. Wu et al., Identification of protein tyrosine phosphatase 1B (PTP1B) inhibitors through De Novo Evoluton, synthesis,
biological evaluation and molecular dynamics simulation, Biochemical and Biophysical Research Communications, https://doi.org/10.1016/
j.bbrc.2020.03.075