Design and characterization of α-lipoic acyl shikonin ester twin drugs as tubulin and PDK1 dual inhibitors

Article abstract of DOI:10.1016/j.ejmech.2017.12.019

Shikonin exhibits powerful anticancer activities for various cancer cells, but its poor solubility and strong toxicity hinder its development as clinical anticancer agent. We previously confirmed that shikonin and its derivatives can disturb mitosis through targeting tubulin. In this study, α-lipoic acid, the naturally-occurring co-factor of pyruvate dehydrogenase (PDH), was introduced into shikonin to design the twin drugs against both mitosis (tubulin) and glycolysis (PDK). 18 kinds of α-lipoic acid shikonin ester derivatives were achieved through three rounds of screening process performed by computer assistant drug design method, being designated as the outstanding compounds. Among them, 1c displayed the most potent cytotoxicity towards cervical cancer cells (HeLa) with an IC₅₀ value of 3.14 ± 0.58 μM and inhibited xenotransplanted tumor growth in a dose-dependent manner. Further pharmacologic study demonstrated that 1c can cause cell cycle arrest in G2/M phase as tubulin polymerization inhibitor. Moreover, it also showed good PDK1 inhibitory activity, promoting PDH activity and forced HeLa cells to process more aerobic metabolism to undergo cell apoptosis. We reported here the first dual inhibitors of tubulin and PDK1 based on shikonin. It may form a basis for shikonin optimization through twin drug design framework for the discovery of new and potent shikonin derivatives in the study of targeted cancer therapy.

Full text of DOI:10.1016/j.ejmech.2017.12.019

European Journal of Medicinal Chemistry 144 (2018) 137e150
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Research paper
Design and characterization of
a-lipoic acyl shikonin ester twin drugs
as tubulin and PDK1 dual inhibitors
,
b
,
,
b
,
,
Hong-Yan Lin a 1, Hong-Wei Han a 1, Wen-Xue Sun a b, Yu-Shun Yang a,
Cheng-Yi Tang a b, Gui-Hua Lu a b, Jin-Liang Qi a ***, Xiao-Ming Wang a
,
,
,
,
b,
, b, **
Yong-Hua Yang a
, b, *
a State Key Laboratory of Pharmaceutical Biotechnology, NJU-NJFU Institute of Plant Molecular Biology, School of Life Sciences, Nanjing University, Nanjing,
210023, China
b Co-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing, 210037, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Shikonin exhibits powerful anticancer activities for various cancer cells, but its poor solubility and strong
Received 20 July 2017
Received in revised form
5 December 2017
Accepted 6 December 2017
Available online 7 December 2017
toxicity hinder its development as clinical anticancer agent. We previously confirmed that shikonin and
its derivatives can disturb mitosis through targeting tubulin. In this study,
occurring co-factor of pyruvate dehydrogenase (PDH), was introduced into shikonin to design the twin
drugs against both mitosis (tubulin) and glycolysis (PDK). 18 kinds of -lipoic acid shikonin ester de-
a-lipoic acid, the naturally-
a
rivatives were achieved through three rounds of screening process performed by computer assistant
drug design method, being designated as the outstanding compounds. Among them, 1c displayed the
most potent cytotoxicity towards cervical cancer cells (HeLa) with an IC50 value of 3.14 0.58 mM and
Keywords:
Cancer
inhibited xenotransplanted tumor growth in a dose-dependent manner. Further pharmacologic study
demonstrated that 1c can cause cell cycle arrest in G2/M phase as tubulin polymerization inhibitor.
Moreover, it also showed good PDK1 inhibitory activity, promoting PDH activity and forced HeLa cells to
process more aerobic metabolism to undergo cell apoptosis. We reported here the first dual inhibitors of
tubulin and PDK1 based on shikonin. It may form a basis for shikonin optimization through twin drug
design framework for the discovery of new and potent shikonin derivatives in the study of targeted
cancer therapy.
Shikonin
Twin drug
Pyruvate dehydrogenase kinase
Tubulin
© 2017 Elsevier Masson SAS. All rights reserved.
1. Introduction
Abbreviations: CADD, computer assistant drug design; NOS, nitric oxide syn-
thase; PKM2, pyruvate kinase isozymes M2; GLUT1, Glucose transporter 1; LA,
alpha-lipoic acid; PDH, pyruvate dehydrogenase; PDC, pyruvate dehydrogenase
complex; PDK, pyruvate dehydrogenase kinase; MTT, 3-(4,5-dimethyl-2-thiazolyl)-
2,5-diphenyl-2-H-tetrazolium bromide; PMSF, phenylmethanesulfonyl fluoride;
PVDF, Poly(vinylidene fluoride); DAPI, 40,6-diamidino-2-phenylindole; TLC, thin
layer chromatography; DMSO, dimethylsulfoxide; NMR, nuclear magnetic reso-
nance; BSA, albumin from bovine serum; PI, prodium iodide.
* Corresponding author. State Key Laboratory of Pharmaceutical Biotechnology,
NJU-NJFU Institute of Plant Molecular Biology, School of Life Sciences, Nanjing
University, Nanjing, 210023, China.
** Corresponding author. State Key Laboratory of Pharmaceutical Biotechnology,
NJU-NJFU Institute of Plant Molecular Biology, School of Life Sciences, Nanjing
University, Nanjing, 210023, China.
It has been known for over half a century that tumor cells
consume glucose at surprisingly higher rates compared to normal
tissues. In addition, even in the presence of sufficient oxygen, tu-
mors metabolize glucose via oxygen-independent aerobic glycol-
ysis rather than via oxygen-dependent (more efficient) process of
oxidative phosphorylation [1]. This phenomenon was known as
“Warburg effect” and often considered to be the foundation for
researching cancer metabolism to obtain effective anticancer
therapies [2e4].
Pyruvate dehydrogenase complex (PDHC; E.C. 1.2.4.1) is a key
enzyme series in metabolism which catalyzes the oxidative
decarboxylation of pyruvate to produce acetyl-CoA thus linking
glycolysis to tricarboxylic acid cycle [5e7]. PDHC activity is
controlled by two regulatory enzymes; pyruvate dehydrogenase
kinase (PDK), which phosphorylates, then inactivates the enzyme,
*** Corresponding author. State Key Laboratory of Pharmaceutical Biotechnology,
NJU-NJFU Institute of Plant Molecular Biology, School of Life Sciences, Nanjing
University, Nanjing, 210023, China.
E-mail addresses: qijl@nju.edu.cn (J.-L.
Qi),
wangxm07@nju.edu.cn
(X.-M. Wang), yangyh@nju.edu.cn (Y.-H. Yang).
1
These authors contributed equally to this work.
https://doi.org/10.1016/j.ejmech.2017.12.019
0223-5234/© 2017 Elsevier Masson SAS. All rights reserved.

138
H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
and pyruvate dehydrogenase phosphatase (PDP), which de-
phosphorylates the enzyme to its active form [8,9]. PDK, a member
of the GHKL ATPase/kinase superfamily, contains four mammalian
isoforms (PDK1~4) in mitochondria characterized in terms of their
differences in activity, tissue distribution and regulations [10].
Among these four isoforms, PDK1 is mostly associated with cancer
malignancy by reducing PDC activity through phosphorylation of
2. Methods
2.1. Cell lines and culture conditions
The cell lines used in this study were human hepatoma cell line
(HepG2), human breast cancer cell line (MDA-MB-231), carcinoma
of cervix cell line (HeLa), human lung adenocarcinoma epithelial
cell line (A549), human embryonic kidney cell (293 T) and human
liver cell (L02). They were obtained from State Key Laboratory of
Pharmaceutical Biotechnology, Nanjing University. Cells were
maintained in DMEM/high glucose supplemented with 10% fetal
specific serine residues in E1a subunit of pyruvate dehydrogenase
(PDH) [11e13]. It has been reported earlier that PDK1 was
remarkably over-expressed in multiple clinical cancer specimens
and has become popular pharmacological target for cancer
chemotherapy [14e16].
bovine serum, L-glutamine, and penicillin/streptomycin and incu-
Microtubules (MTs) are cytoskeletal filaments that are composed
bated at 37 ꢀC in humidified atmosphere containing 5% CO2.
of
dynamic process that involves the polymerization and de-
polymerization of - and -tubulin heterodimers [18]. Disruption
a- and b-tubulin proteins [17]. The formation of microtubules is a
2.2. Animals feeding
a
b
of the dynamic equilibrium blocks the cell division machinery at
mitosis and leads to cell cycle arrest at metaphase, resulting in cell
death [19]. Likewise, tubulin has become an important target for the
design and development of new anti-cancer agents [20].
Female nude mice (5e7 weeks old) were obtained from Model
Animal Research Center of Nanjing University (Nanjing, China).
Briefly, mice were fed with free access to pellet food and water in
plastic cages at 21
2
ꢀC and kept on a 12 h light-dark cycle. All
Shikonin is a major naphthoquinone compound found in the
roots of Lithospermum erythrorhizon and exhibits powerful anti-
cancer activities for various cancer cells [21e23]. Previous studies
have shown that shikonin is a highly active functional molecule
that can kill cancer cells effectively in different ways, like inducing
cell apoptosis [23,24], inhibiting topoisomerase [25], inhibiting
nitric oxide synthase (NOS) [26], etc. Our previous research has also
confirmed that shikonin and its derivatives exhibit potent tubulin
inhibitory activity [27e29]. They can disrupt the dynamic equilib-
rium of tubulin polymerization and de-polymerization during the
process of mitosis, subsequently resulting in cell cycle arrest thus
causing cell death [28e30].
Alpha-lipoic acid(LA), a naturally-occurring co-factor is involved
in regulating metabolism. It is found to be present in a number of
multi-enzyme complexes, including PDHC and has been widely
reported to induce apoptosis in various cancer cell lines [31,32]. In
cells, LA is reduced to dihydrolipoic acid, which scavenges various
reactive oxygen species and regenerates other endogenous anti-
oxidants without adverse side effects [33]. CPI-613, an analogue of
LA, which has been approved by FDA for the treatment of relapse/
refractory leukemia, can strongly disrupt mitochondrial meta-
bolism, with selectivity for tumor cells in culture [34,35]. Further-
more, it also shows strong antitumor activity in vivo against human
non-small lung and pancreatic cancer cells in xenograft models
without any significant adverse effects [36].
Having two different pharmacophores, the nonsymmetrical
twin drugs are expected to have two different pharmacological
effects (dual action) [37]. To achieve outstanding compounds, a
screening process of a 40,000 compounds scale was performed by
computer assistant drug design (CADD) method. We attempted to
screen at three levels preliminarily, the backbone, the derivation
and the substitution. After the first round, we chose LA pattern as a
relay point (Fig. 1). As shown in Fig. 1, a molecular hybridization
strategy based on the natural product shikonin and LA analogues
yield a scaffold which has three parts: i) a natural shikonin scaffold
as an anticancer pharmacophore fragment against tubulin; ii) a
naturally-occurring co-factor LA as the medium-chain and iii)
various aromatic aldehyde units attached with LA fragment as
another anticancer pharmacophore fragment against PDK1. To the
best of our knowledge, such a design in shikonin structural modi-
fication is unprecedented. Based on CADD screening, we conducted
further biochemical and pharmaceutical assays to evaluate
whether or not the novel twin drugs, LA shikonin ester derivatives
will provide a synergistic anticancer activity through the dual ac-
tion at the two targets.
animal procedures were performed in accordance with the
Guidelines for Care and Use of Laboratory Animals of Nanjing
University and approved by the Animal Ethics Committee of the
Ministry of Science and Technology of China (2006). All efforts were
made to minimize animal's suffering and to reduce the number of
animals used.
2.3. Anti-proliferation assay
The anti-proliferative activity of the prepared compounds
against five cancer cell lines, i.e HepG2, MDA -MB-231, HeLa, A549
and two normal cell lines, i.e L02 and 293 T was evaluated as
described previously [38]. Target tumor cell lines were grown to log
phase in DMEM medium supplemented with 10% fetal bovine
serum. After dilution to 2 ꢁ 104 cells mLꢂ1 with the complete
medium, 100 mL of the obtained cell suspension was added to each
well of 96-well flat bottom plates and then allowed to adhere for
12 h at 37 ꢀC, 5% CO2 atmosphere. Tested samples at pre-set con-
centrations (0.1, 1, 10, 100
plates with shikonin, colchicine and DCA as positive references.
After 24 h exposure period, 20
L of PBS containing 2.5 mg mLꢂ1 of
mM) were added to 96-well flat bottom
m
MTT was added to each well. Plates were then incubated for further
4 h and then centrifuged with 1500 rpm at 4 ꢀC for 10 min followed
by supernatant removal. Afterwards, 150 mL of DMSO was added to
each well to dissolve purple crystal. The plates were shaken for
10 min at room temperature to ensure maximum solubility. The
absorbance was measured and recorded by ELISA reader (ELx800,
BioTek, USA) at a test wavelength of 570 nm. Each experiment had
three replicates and was repeated thrice. The IC50 values defined as
concentrations that caused 50% loss of cell viability were calculated
by Origin 7.5.
2.4. Flow cytometric analysis of apoptosis
2.4.1. Annexin V-APC/propidium iodide (PI) assay
Briefly, HeLa cells were seeded in 6-well plates (1 ꢁ 105 cells per
well) for 12 h and then treated with 1c (0, 1, 2 and 4 mM) for 24 h or
treated with 6
collected, washed twice with PBS, stained with 5
FITC and 2.5 L of PI (5
g mLꢂ1) in 500
(10 mM HEPES, pH 7.4, 140 mM NaOH, 2.5 mM CaCl2) for 30 min at
room temperature in the dark. Apoptotic cells were quantified us-
ing a FACScan cytofluorometer (PT. Madagasi Brosa Inc. Jl. Batang
Hari No. 73, Propinsi Sumatera Utara, Indonesia). Statistical analysis
was done using Flowjo 7.6.1 software.
m
M 1c for 0, 12, 24 and 36 h. Cells were then
L of Annexin V-
L of 1ꢁ binding buffer
m
m
m
m

H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
139
Fig. 1. Rational molecular hybridization strategy for tubulin and PDK1 dual inhibitors based on shikonin performed by CADD method.
2.4.2. Mitochondrial membrane potential (
D
ѱm) analysis
and subsequently washed with PBS. Afterwards, cells were incu-
Mitochondrial transmembrane potential was examined using a
Mitochondrial Membrane Potential Assay Kit (Beyotime Institute of
Biotechnology, Haimen, China) with JC-1 (lipophilic cation 5, 50, 6,
60-tetrachloro-1, 10, 3, 30-tetraethylbenzimidazolcarbocyanine io-
bated with anti-tubulin antibody (1: 500 dilution) in 3% BSA
overnight at 4 ꢀC. After washing with 0.5% TritonX-100 (incubate
for 5 min), each coverslip was added with 200
mL of Cy3-labeled
goat anti-mouse IgG (H þ L) (1: 1500 dilution) in 3% BSA and
dide). Briefly, HeLa cells were seeded in poly-
L-lysine (Sigma)
incubated for 1 h at room temperature followed by DAPI
coated 6-well plates at 5 ꢁ 105 cells per well. After cultured in
DMEM with 10% FBS for 4 h in CO2 incubator, the cells were incu-
bated with 1 mL JC-1 working solution in CO2 incubator for 20 min.
After that, the staining solution was removed, and cells were
washed with JC-1 staining buffer twice before being analyzed by
fluorescence microscopy using a band pass filter (Fluorescein and
(5 ng mLꢂ1). Cells were then observed under an Olympus confocal
microscope and data was analyzed using FV-10-ASW 1.7 viewer.
2.7. Western bolt analysis
Total proteins of HeLa cells were extracted by treating with RIPA
lysis buffer containing 1% PMSF. Protein concentrations were
measured using the BCA Protein Assay kit. Approximately 80 mg of
Cy3). The depolarizing agent CCCP (10
control.
mM) was used as a positive
total proteins were separated by 10% SEMSepolyacrylamide gels
2.5. Cell-cycle distribution by flow cytometry
and transferred to PVDF membranes. After blocking, membranes
were incubated with primary antibody against
monoclonal IgG, 1:2000 dilution); PDH-E1 (mouse monoclonal
IgG, 1:600 dilution), PDH-E1 subunit (phospho S293) (rabbit
polyclonal IgG, 1:600 dilution),PDH-E1 subunit (phospho S232)
b-tubulin (mouse
HeLa cells were plated in 6-well plates (5.0 ꢁ 103 cells per well)
a
and incubated at 37 ꢀC for 24 h. Exponentially growing cells were
a
then incubated with 1c at 2, 4 and 8
mM for 24 h with shikonin (1, 2
a
and 4 M) as positive control. Cells treated with the DMSO were
m
(rabbit polyclonal IgG,1:1000 dilution), PARP-1 (mouse monoclonal
IgG, 1:600 dilution), cleaved PARP-1 (mouse monoclonal IgG, 1:600
dilution), Bcl-2 (rabbit polyclonal IgG, 1:1000 dilution), Bax (rabbit
polyclonal IgG, 1:1000 dilution), Cleaved Caspase-3 (rabbit poly-
clonal IgG, 1:1000 dilution) or GAPDH (rabbit polyclonal IgG,
1:1000 dilution) overnight. After washing thrice with TBST for
5 min, membranes were incubated for 1 h at room temperature
with HRP-conjugated secondary antibodies (1:5000 dilution).
Detection was performed by an enhanced chemiluminescent re-
agent (Thermo Fisher Scientific, USA) according to the manufac-
turer's instructions. Bands were then recorded by a digital camera
(Tanon5200, Tanon, China). Finally, the results were analyzed with
Image J Software (National Institutes of Health, BetheSEMa, Mary-
land, USA), and all the targeted proteins were normalized to
GAPDH.
used as negative control. After that, the cells were harvested,
centrifuged and fixed in 70% ice chilled ethanol at 4 ꢀC for at least
12 h and subsequently resuspended in PBS containing 0.1 mg mLꢂ1
RNase A and 5
m
g mLꢂ1 of propidium iodide (PI). Cellular DNA
content, for cell cycle distribution analysis, was measured by flow
cytometry using FACNcan cytofluorometer (PT. Madagasi Brosa Inc.
JI. Batang Hari NO.73, Propinsi Sumatera Utara, Indonesia) plotting
10,000 events per sample. The percentage of cells in the G1, S and
G2/M phases of the cell cycle was determined using the Flowjo 7.6.1
software after cell debris exclusion.
2.6. Confocal microscopy assay
HeLa cells were grown on round cover slips to 70% confluence
and incubated with 4
mM of 1c, 4 mM of shikonin, 0.5 mM of
colchicine and 0.5 M of paclitaxel for 24 h, respectively. After in-
m
2.8. Immunohistochemical analysis
cubation, cells were washed three times with PBS and fixed with 4%
paraformaldehyde for 20 min, permeabilized with 1% TritonX-100
for another 10 min. Cells were then blocked with 3% BSA for 1 h
Immunohistochemistry was performed for Anti-PDH-E1
tumor tissues obtained from mice were fixed in 4%
a. The

140
H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
papaformaldehyde for 12
h
after cold 4% paraformaldehyde
N2 at ꢂ20 ꢀC and was then used without further purification
perfusion. Then, cutting the packages tissues to slice as 20e30
sections at optimum temperature. The tissues were incubated with
primary antibody (Anti-PDH-E1
, Santa cruz, USA) at 4 ꢀC over-
mm
(Scheme 1).
a
2.11.2. General procedure for preparation of 1b-18b
night after deparaffinization and rehydration. Afterwards, the tis-
sues were then treated with Alexa Fluor 488 labeled anti-mouse
secondary antibody (Invitrogen, USA) for 1 h at room temperature.
Finally, sections were subjected to immunofluorescence staining
via epifluorescence microscopy.
6, 8-Bis-sulfanyloctanoic acid (0.03 mol) was dissolved in
dichloromethane under nitrogen atmosphere. To this solution,
appropriate aldehyde (0.03 mol) was added. The mixture was then
stirred at room temperature for 1 h and cooled to ꢂ25 ꢀC. Boron
trifluoride etherate (0.03 mol) was added, and the reaction was
allowed to warm at room temperature [39]. After evaporation of
solvent, the products were purified by crystallization with ethyl
acetate: petroleum ether (V: V ¼ 1: 2) as solvent (Scheme 1).
2.9. Transplantation of HeLa cells into nude mice
Cultured HeLa cells were washed and resuspended in ice-cold
PBS. Portions of the suspension (6 ꢁ 106 cells in 0.1 mL) were
injected into the right flanks of female nu/nu mice. Two weeks after
injection, the mice bearing tumors (an average size of 40 mm3)
were distributed into 6 groups (n ¼ 8 mice per group) according to
2.11.3. General procedure for preparation of 1c-18c
Alpha-lipoic acid derivatives 1b-18b (0.0003 mol), shikonin
(0.0002 mol), 4-dimethyaminopyridine (0.00002 mol) and N, N0-
dicyclohexylcarbodiimide (0.0004 mol) were dissolved in
dichloromethane (25 mL).The reaction system was cooled to 0 ꢀC
and stirred for 12 h [40]. Then, a proper amount of silica gel was
added and the solvent was condensed by vacuum concentration.
Lastly, target compounds were collected by column chromatog-
raphy (Scheme 1). Chemical structures of the target compounds
(1c-18c) are shown in Table S1 (SI Appendix) and the eluent
composition and proportion are shown in Table S2 (SI Appendix).
tumor volumes. 1c (0.5, 1and2 mg kgꢂ1) and shikonin (1 mg kgꢂ1
)
were dissolved in DMSO and administered once every two days for
8 times by intraperitoneal injection. Vehicle (DMSO)-treated group
was used as control. Body weight and tumor volumes were
measured and recorded every two days. Long diameter (L) and
short diameter (S) of a tumor was measured with a vernier caliper
and the tumor volume was calculated using following formula:
L ꢁ (S)2/2. Fifteen days after injection, mice were killed and tumors
were separated. Tumor weight was measured and tumor sections
were fixed in paraformaldehyde. The rest of the sections were
frozen in liquid nitrogen and stored at ꢂ80 ꢀC. In another experi-
ment, survival test was made (n ¼ 8 mice per group) as above and
monitored daily until all the mice died.
2.12. Materials and measurements
All chemicals (reagent grade) used were purchased from Nanj-
ing Chemical Reagent Co. Ltd. (China). All the 1H NMR spectra were
recorded on a Bruker DPX 300 spectrometer in CDCl3. 13C NMR
spectra were recorded on a Bruker DPX 600 spectrometer in
2.10. Data and statistical analysis
CDCl3.Chemical shifts (
reported in ppm ( ). Melting points (uncorrected) were measured
d
) for 1H NMR and 13C NMR spectra were
d
Each experiment was repeated thrice. Data presented as
mean S.E.M. and statistical comparisons between the treated and
untreated groups were subjected to one-way ANOVA using
GraphPad PRISM5 (Graphpad Inc., La Jolla, USA). P-values of <0.05
were considered statistically significant; P-values of <0.01 were
considered statistically very significant.
on a XT4 MP Apparatus (Taike Corp., Beijing, China). ESI mass
spectra were obtained on a Mariner Biospectrometry Workstation
(ESI-TOF) mass spectrometer. TLC was carried out on glass-backed
silica gel sheets (silica gel 60 Å GF254) and visualized in UV light
(254 nm).
MTT, PMSF, DAPI and RIPA lysis buffers (#P0013B) were pur-
chased from Beyotime Institute of Biotechnology (Haimen, China).
Anti-tubulin (#AT819), Cy3-labeled goat anti-mouse IgG (H þ L)
(#A0521) were purchased from Cytoskeleton, Inc. (Denver, Colo-
rado, USA). RNase A (#EN0531) was purchased from Thermo Sci-
entific, Fermentas (USA). AnnexcinV-FITC cell apoptosis assay kit
(#BA11100) was purchased from BIO-BOX (Nanjing, China). Fibro-
nectin (#F1056), laminin (#L2020), dimethylsulfoxide (#D2650)
and dichloroacetic acid (#D54702) were purchased from Sigma-
Aldrich (St. Louis, MO, USA). JC-1 assay kit (#C2006) and ROS
assay kit (#S0033) were purchased from Beyotime Institute of
Biotechnology (Haimen, China). Glucose assay kit was purchased
from Shanghai Rongsheng Biotech Co., Ltd (Shanghai, China).
Lactate acid assay kit (A019-2) was purchased from Nanjing Jian-
cheng Bioengineering Institute (Nanjing, China). Pyruvate dehy-
drogenase (PDH) enzyme activity microplate assay kit (#ab109902)
was purchased from Abcam (Cambridge, England). BCA protein
assay kit (#23227) was purchased from Pierce (Rockford, IL, USA).
PVDF membranes were purchased from Biosharp (Hefei, China).
2.11. Organic synthesis
2.11.1. General procedure for preparation of 6, 8-Bis-
sulfanyloctanoic acid
LA (103.2 g, 0.50 mol) was suspended in 3000 mL of water in a
5 l beaker equipped with a mechanical stirrer and thermometer.
Sodium bicarbonate (42.1 g, 0.50 mol) was added in small portions
to the stirred suspension. The mixture was again stirred for further
10 min to obtain complete dissolution. The solution turned pale
yellow. It was then cooled to about 5 ꢀC in an ice bath, and solid
sodium borohydride pellets (37.8 g, 1.0 mol) were added, the
addition was done in portions with continuous stirring to mini-
mize foaming. Finally the solution turned colorless. The stirring
was continued at 5 ꢀC for 30 min. The temperature was then
increased to room temperature and the solution was stirred for an
additional 45 min. Solution was then cooled to <10 ꢀC using an ice
bath and the pH was adjusted to 1 by slowly adding concentrated
hydrochloric acid to keep the temperature below 10 ꢀC. After-
wards, the white, water insoluble oil was separated. The mixture
was extracted with 4 ꢁ 500 mL of chloroform. Combined extracts
were washed with 300 mL of saturated brine. Organic phase was
dried with magnesium sulfate, filtered and the solvent evaporated
under reduced pressure (40 mm at 25 ꢀC) which was then put in
the vacuum drying oven to remove residual solvent [39]. The
product was a colorless oil, weighing 96.8 g (93%). It was stored in
Anti-PDH-E1
cleaved PARP-1 (#sc-56196) were purchased from Santa Cruz
Biotechnology Inc. (Dallas, Texas, USA).Anti-PDH-E1 subunit
(phospho S293) (#ab177461) was purchased from Abcam, (Cam-
bridge, England).Anti-PDH-E1 subunit (phospho S232) (#AP1063)
and anti-PDH-E1 subunit (phospho S300) (#AP1064) were pur-
a (#sc-377092), Anti-PARP-1 (#sc-56197), Anti-
a
a
a
chased from Merck Millipore, (Darmstadt, Germany). and Anti-
cleaved caspase-3 (#9661) was purchased from Cell Signal

H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
141
Scheme 1. The synthetic routes of compounds 1c-18c. (a) NaBH4, NaHCO3, H2O; (b) Boron trifluoride etherate; (c) DCC, DMAP, CH2Cl2, ice bath.
Technology (Danvers, USA). Bcl-2 antibody (E1A6139), Bax anti-
body (E1A0120), GAPDH antibody (#E1A7021) and secondary
antibody (anti-mouse or anti-rabbit IgG)(#E1WP319 or #E1WP318)
were purchased from Enogene (Nanjing, China). ECL Kit (#34077)
were purchased from Thermo Scientific (USA). Four-week old nude
mice were purchased fromModel Animal Research Center of
Nanjing University.
shown in Fig. 2B, colchicine bound to tubulin via two hydrogen
bonds with VAL 181 (distance 3.23 Å), CYS 241
a
¼
b
(distance ¼ 3.13 Å). After comparison, we found that the binding of
1c with tubulin mainly rely on naphthoquinone ring of shikonin
and the binding energy was near to that of colchicine. As shown in
Fig. 2B, the pharmacophore model of colchicine consists of two
hydrogen bond receptors which corresponding to an oxygen atom
interacting with the backbone of the VAL181 of chain
a, and to the
methoxyl group interacting with the CYS 241 of chain
tively. More recent examples are available in several valuable
studies which emphasized the fact that engaging hydrogen bond
b
, respec-
3. Results
3.1. Preliminary calculation
receptor interactions with the side chain of CYS 241b is acknowl-
edged as an important molecular requirement for designing suc-
cessful Colchicine Binding Site Inhibitors (CBSIs) [41e43]. It means
that the trimethoxyphenyl group is a classical privileged structure
playing as hydrogen bond maker (Fig. 2B), which has inspired the
design and synthesis of various of CBSIs. Unlike these, our design
resulted a number of CBSI candidates provided with alternative
Before synthesis, a three-round screening process of over
40,000 compounds was performed by CADD method. Molecular
docking was conducted to calculate the binding energy (CDOCKER
INTERACTION ENERGY in CDOCKER protocol in Discovery Studio
3.5) between compounds and tubulin, PDK1 complex structure
(1SA0, 2Q8G.pdb, downloaded from the PDB). Up to 35 substitute-
defaulted compounds were evaluated to screen out the backbone in
the first round. Then, shikonin backbone was derived into 20
available similar structures which were later examined in the sec-
ond round. Substitutes were then added to provide 63 choices
which were judged in the third round. A total scale of over 40,000
compounds was then recalculated as a whole to confirm the results.
Finally, 18 compounds were picked to continue assay subsequently
(SI Appendix).
groups able to establish hydrogen bonds with TYR 224
a and LYS
254 . For instance, the top-scored pose of 1c in Fig. 2A disclose as
b
the phenolic hydroxyl and the carbanyl groups are effective in
establishing the required hydrogen bond interactions with TYR
224
a and LYS 254b. It is mainly because the larger molecular
structure of 1c forced it to occupy both the binding pockets of
colchicine and GTP, and thus formed the hydrogen bond in-
teractions with LYS 254b and TYR 224a, respectively. Although the
binding pose of 1c is different from colchicine or other CBSIs re-
ported in recent papers, in terms of the bond length and binding
energy, the hydrogen bond interactions between 1c and the two
amino acid residues can also stabilize the binding of 1c and tubulin.
In addition, we also examined the interaction of 1c with PDK1
(PDB code: 2Q8G). We found that the binding energy of 1c
(ꢂ48.33 kcal/mol) was lower and much better than DCA
(–24.28 kcal/mol) (Fig. S2, SI Appendix). As depicted in Fig. 2C, 1c
In view of binding energy between compounds and tubulin, 1c(-
49.32 kcal/mol) is best one amongst them and is near to that of
colchicine (ꢂ52.85 kcal/mol) (Fig. S1, SI Appendix). For better un-
derstanding of the potential of 1c, we first examined the interaction
of 1c with tubulin (PDB code: 1SA0). All the amino acid residues
which established interactions with tubulin were exhibited and the
results were depicted in Fig. 2A. In the binding model, 1c bound
well to tubulin via two hydrogen bonds with LYS 254
(distance ¼ 2.06 Å), TYR 224 (distance ¼ 2.38 Å). Meanwhile,
colchicine was used as positive control in docking simulation. As
b
bound well to the PDK1 via
p-p
bond with PHE
a
65(distance 3.97 Å) and -cation bond with LYS 60
¼
p

142
H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
Fig. 2. Molecular docking analysis of 1c, showing proposed binding modes with a/b-tubulin (PDB code: 1SA0) and PDK1 (PDB code: 2Q8G). (A) Binding pose of 1c in the protein
surface of tubulin at colchicine binding site (carbon atom, gray; oxygen atom, red; hydrogen atom, white; sulphur atom, orange). Hydrogen bond was displayed as lime dashed line.
(B) Binding pose of colchicine in the protein surface of tubulin (carbon atom, gray; oxygen atom, red; hydrogen atom, white; nitrogen atom, bule).Hydrogen bond was displayed as
lime dashed line. (C) Binding pose of 1c in the protein surface of PDK1(carbon atom, yellow; oxygen atom, red; hydrogen atom, white; sulphur atom, orange). p-p bond and p-cation
bond were displayed as orange solid line. (D) Binding pose of DCA in the protein surface of PDK1 (carbon atom, gray; oxygen atom, red; hydrogen atom, white; chlorine atom,
green). Hydrogen bond was displayed as lime dashed line. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
(distance ¼ 4.61 Å). In contrast, DCA, a well-known PDK inhibitor,
only formed one hydrogen bond with GLN 61 (distance ¼ 2.43 Å)
and couldnot bind stably at this site because of its small structure
(Fig. 2D). Thus, there often exists off-target effect in the clinical
anticancer treatment with DCA, thereby leading to a high effective
dosage.
(IC50 ¼ 4.53
0.85
m
M) showed similar or even better anti-
proliferation effects than shikonin (IC50 ¼ 6.92 1.08
mM) itself.
For MDA-MB-231 and A549 cells, the anti-proliferation effects of
these drugs were also significant but not better than shikonin. For
HeLa cells, 4c (IC50 ¼ 9.96 1.87
and 12c (IC50 ¼ 8.95 1.38 M) displayed better cell proliferate
inhibition effect than other derivatives but not better than shikonin
(IC50 ¼ 6.86 1.22 M). It was worth reminding that 1c can strongly
inhibit the proliferation of HeLa cells with an IC50 of 3.14 0.58
mM), 11c (IC50 ¼ 9.17 1.88
mM)
m
m
3.2. Chemistry
mM
which was much better than shikonin. In addition, we noticed that
anti-proliferation effects of all these compounds on non-cancer
cells (L02 and 293 T) were weaker than cancer cell lines. Based
on these results, we speculated that intervention of LA derivatives
can significantly reduce the cytotoxicity of shikonin against non-
cancer cells. In order to figure out anti-cancer mechanisms of
these drugs, we chose 1c against HeLa cells for further study.
To investigate the effect of 1c on HeLa cell viability, we per-
formed PI staining assay. From the results shown in Fig. S3 (SI
Appendix), we observed that 1c induced a dose-dependent and
time-dependent death of cancer cells. Based on these data, we
concluded that 1c can inhibit cancer cell growth effectively.
The synthetic routes for the novel lipoic acid shikonin ester
derivatives 1c-18c are outlined in Scheme 1. These compounds
were obtained by three steps which were elucidated in experi-
mental section. All of them were first reported and characterized by
1H NMR, 13C NMR, elemental analysis, melting test and mass
spectroscopy, and the results were in full accordance with their
depicted structures (SI Appendix).
3.3. Bioactivity
3.3.1. 1c selectively inhibits proliferation of cancer cells
All synthesized LA shikonin ester twin drugs 1c-18c were
evaluated for their anti-proliferation activities against four cancer
cell lines, i.e HepG2, MDA-MB-231, HeLa, A549, and two normal cell
lines, i.e 293 T, L02 by 3-(4, 5-Dimethylthiazol-2-yl)-2, 5-
diphenyltetrazolium (MTT) assay. As listed in Table 1, all of them
showed good anti-proliferation activity against cancer cells, and
some of them showed even better inhibitory activities than shi-
konin itself while avoiding non-cancer cells.
3.3.2. 1c induced HeLa cell cycle arrest in G2/M phase through
inhibiting tubulin polymerization
Considering that tubulin could be the target of 1c, we detected
the effect of 1c on tubulin polymerization and cell cycle distribu-
tion. Firstly, HeLa cells were treated with 1c at different concen-
trations for 24 h to see whether it can arrest cell cycle in G2/M
phase or not (Fig. 3A). Shikonin which was used as positive control,
after comparison, it was clear that the effect of 1c on cell cycle ar-
rest was better than that of shikonin (Fig. 3B), and that is, the
introduction of LA analogue fragment apparently enhanced HeLa
Generally speaking, all target compounds effectively inhibited
the proliferation of four cancer cell lines. For HepG2 cells, 1c
(IC50 ¼ 5.37
1.35
1.21
m
M), 6c (IC50 ¼ 4.34
0.99
mM), 9c
(IC50 ¼ 6.69
m
M), 10c (IC50 ¼ 5.08
1.03 mM) and 14c

H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
143
Table 1
Inhibition of cell proliferation against HepG2, MDA-231, HeLa, A549, L02 and 293 T cells by 1c-18c, shikonin and DCA.
Compound
IC50 (mM)
HepG2
MDA-231
HeLa
A549
L02
293 T
1c
2c
3c
4c
5c
6c
7c
8c
5.37 1.35
24.31 2.75
45.73 2.87
26.72 2.29
30.12 2.76
4.34 0.99
25.91 2.37
11.34 2.22
6.69 1.21
5.08 1.03
17.78 2.26
14.89 1.27
21.94 2.06
4.53 0.85
11.81 1.27
20.42 2.01
28.42 2.65
17.41 2.01
6.92 1.08
4.65 0.68
>100
17.67 3.46
17.43 2.12
11.88 1.69
24.21 2.38
17.96 2.18
16.67 1.66
8.97 1.69
20.57 2.11
17.66 1.66
25.75 1.46
28.04 2.98
29.16 2.13
26.84 1.28
35.10 3.64
9.57 1.67
16.16 1.99
19.62 2.17
25.93 2.65
8.75 1.23
3.89 1.26
>100
3.14 0.58
14.86 2.01
11.45 1.67
9.96 1.87
16.02 2.11
21.95 2.29
17.48 1.77
16.48 1.58
21.42 2.01
13.62 1.22
9.17 1.88
8.95 1.38
13.26 1.77
20.15 2.29
23.45 2.06
26.95 2.87
21.95 2.13
28.95 2.36
6.86 1.22
2.54 0.35
>100
16.37 2.15
10.03 1.78
12.14 1.89
6.03 1.06
20.67 2.54
19.34 2.01
19.61 1.87
16.81 1.44
7.416 1.76
10.65 1.88
17.51 1.89
10.43 1.56
22.15 1.99
19.64 1.99
10.18 1.59
22.20 2.06
27.19 2.23
14.61 2.07
3.53 1.02
4.64 1.09
>100
116.2 4.38
89.82 2.26
85.57 2.09
116.2 6.03
94.81 4.47
99.34 6.59
85.64 4.38
95.82 5.59
83.57 3.99
119.1 6.69
120.9 6.89
79.77 5.32
92.95 3.38
135.8 5.78
90.90 3.98
96.19 3.38
108.2 4.98
102.6 5.53
3.41 1.07
3.01 0.79
>100
88.83 3.37
102.1 4.78
88.58 2.38
79.26 2.87
105.3 3.89
116.2 3.59
107.9 3.98
77.97 3.32
64.72 3.67
97.78 4.76
90.10 4.47
69.92 2.27
101.8 4.98
83.93 3.91
73.41 2.25
95.41 3.32
115.6 3.58
83.54 3.78
62.76 2.21
6.38 1.46
>100
9c
10c
11c
12c
13c
14c
15c
16c
17c
18c
shikonin
colchicine
DCA
Fig. 3. 1c induced G2/M cell cycle arrest in HeLa cells through inhibiting tubulin polymerization. (A) HeLa cells were treated with 0, 2, 4, and 8
collected and processed by PI staining for cell cycle analysis. (B) Cells treated with 0, 1, 2, and 4
m
M of 1c for 24 h. Then they were
m
M of shikonin for 24 h were also collected and analyzed(G1 phase, green; S phase,
yellow and G2/M phase, blue). Images are representative of three independent experiments. (For interpretation of the references to color in this figure legend, the reader is referred
to the Web version of this article.)
cell cycle arrest of shikonin.
appearance as colchicine treated group. Additionally, shikonin also
showed same effect as 1c and colchicine on microtubule appear-
ance. These results demonstrated that 1c can inhibit polymeriza-
tion of microtubules significantly and such effect is probably related
to its parent structure, i.e shikonin.
To further investigate its' effect on microtubule organization, we
performed an in vitro microtubule assembly assay. After compari-
son, we found out that 1c and colchicine caused inhibition of
microtubule assembly. However, paclitaxel significantly induced
promotion of tubulin polymerization (SI Appendix, Fig. S4). To sum
In addition, we also investigated the effect of 1c on cellular
cytoskeletal network of tubulin. HeLa cells were immune-stained
and analyzed under confocal microscope. As shown in Fig. 4, sub-
stantial stabilization of microtubules in the form of bundle like
appearance was observed clearly in paclitaxel treated cells. In
contrast, cell membrane microtubule of colchicine treated cells
showed de-polymerization and soluble appearance. With reference
to these two groups and the control group, we found that 1c can
inhibit the polymerization of tubulin and cells showed same

144
H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
Fig. 4. Effects of 1c (4 mM), shikonin (4 mM), colchicine (0.5 mM) and paclitaxel (0.5 mM) on interphase microtubules of HeLa cells. Microtubules tagged with rhodamine (red) and
nuclei tagged with DAPI (blue) were observed under a confocal microscope. (For interpretation of the references to color in this figure legend, the reader is referred to the Web
version of this article.)
up, our results demonstrate that 1c inhibits the polymerization of
tubulin in HeLa cells thus disturbing mitosis.
shown in Fig. 5A showed that 1c enhanced the activity of PDH
significantly. Western blotting results shown in Fig. 5B demon-
strated that 1c up-regulated the expression of PDH-E1
form of PDH) and down-regulated the expression of PDH-E1
Ser 300 and p-Ser 232) (the inactive forms of PDH). But dephos-
phorylation of 1c to PDH-E1 Ser 293 site is not significant. In
addition, we again confirmed the effect of 1c on PDH activity using
dichloroacetic acid (DCA, a well-known PDK1 inhibitor which en-
hances the activity of PDH) as positive control. After comparison,
we found that 1c reduced PDH activity and can be a good PDK1
a
(the active
a
(p-
3.3.3. 1c can also inhibit PDK1, activate PDH thus forcing HeLa cells
to process more aerobic metabolism
It is well known that PDK1 is a negative regulatory enzyme of
PDH through phosphorylating the specific serine residues in the
E1a subunit of PDH, leading to the inactivation of PDH. The effect of
these compounds on PDH activity was tested by pyruvate dehy-
drogenase (PDH) enzyme activity microplate assay kit. Results
a

H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
145
Fig. 5. 1c is a PDK1 inhibitor and can inhibit PDH activity. (A) PDH activity of HeLa cells was enhanced by 1c. Chemical structure of 1c was shown on the right panel.(B) Expressions
of PDH-E1 , PDH-E1 (p-Ser 300), PDH-E1 (p-Ser 232) and PDH-E1 (p-Ser 293) affected by 1c of different concentrations were detected by western blotting. (C) Expressions of
PDH-E1 , PDH-E1 (p-Ser 300), PDH-E1 (p-Ser 232) and PDH-E1 (p-Ser 293) affected by 1c (6 M) and DCA (10 mM) were detected by western blotting. Relative band intensity
was determined by Image J software. GAPDH served as a loading control. Images are representative of three independent experiments. (D) The protein level of PDH-E1 was
detected with immunohistochemistry. Nucleus tagged with hematoxylin (blue) and the positive expression tagged with DAB (brownish yellow) were observed under a microscope.
Representative images of six mice were shown (left panel), and the IOD of PDH-E1
a
a
a
a
a
a
a
a
m
a
a
positive cells was calculated (right panel). Original magnification, ꢁ400. (E) 1c reduced the
consumption of glucose in HeLa cells. (F) 1c reduced the production of lactate acid in HeLa cells. Data are means S.E.M. from three independent experiments (*, P < 0.05; **,
P < 0.01). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

146
H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
inhibitor as DCA (Fig. 5C). Further, immunohistochemistry analysis
of mice tumor tissue also showed that 1c can apparently increase
exposed to 1c for 12 h underwent a progressive loss of red fluo-
rescence, indicated by a shift in 2 M of 1c treated group and
downward in the 4 M of 1c treated group. The ratio of cells with
high membrane potential decreased from 85.8% to 42.1% after
exposure to 2 M 1c for 12 h and even decreased to 10.3% in 4 M 1c
treated group. As the change in mitochondrial membrane, poten-
tially occurred in early apoptosis, cells were exposed to 1c treat-
ment for only 12 h to eliminate the impact of late apoptosis on
mitochondrial membrane potential depolarization. PI staining re-
sults showed that there was hardly any late apoptosis appeared in
1c treatment HeLa cells within 12 h, which could exclude the false
positive caused by late apoptosis.
m
PDH-E1
a
positive cells, which is consistent with the above results
m
(Fig. 5D). Meanwhile, it can also effectively reduce the consumption
of glucose and production of lactate acid in HeLa cells (Fig. 5E
and F).
m
m
3.3.4. 1c caused loss of mitochondrial potential in HeLa cells and
induced apoptosis
As shown in Fig. 6A, after treatment with increasing concen-
trations of 1c, the percentage of apoptotic cells was significantly
increased. In addition, 1c can also induce cell apoptosis in a time-
dependent manner (Fig. 6B). Additionally, we also detected the
influence of 1c on apoptosis related protein expression by western
blotting. Results shown in Fig. 6C depicted that 1c can significantly
decrease Bcl-2 but increase Bax expression and subsequently in-
crease the cleavage of PARP and caspase-3 in HeLa cells which was
in full accordance with cell apoptosis results.
3.3.5. 1c effectively inhibits tumor growth of HeLa xenografts in
nude mice
To evaluate the antitumor effects of 1c in vivo, we subcutane-
ously inoculated HeLa cells into the right flank of female nude mice.
Two weeks later, all the mice got visible tumors. The mice were
Furthermore, the transmembrane potential
(
DJm) is an
randomized into 5 groups, control (DMSO), 1c (0.5, 1, 2 mg kgꢂ1
)
important parameter for mitochondrial functionality and its dissi-
pation (involvement of mitochondria in early apoptosis) [44]. As
shown in Fig. 7, microscopy revealed that in untreated HeLa cells,
well-polarized mitochondria were marked by red fluorescent
and positive control (shikonin, 1 mg kgꢂ1). Changes in tumor vol-
umes and body weight were measured and recorded every two
days. As shown in Fig. 8A and D, the administration of 1c inhibited
the tumor growth compared with control in a dose-dependent
manner. After the last treatment on day 15, average tumor size of
1c-treated mice at the 2 mg kgꢂ1 dose was 5-fold smaller than
vehicle control group. The effect of 1c was similar to that of shi-
konin at the 1 mg kgꢂ1 dose. As shown in Fig. 8B, 1c has no
staining. After treatment with 2 or 4 mM of 1c, this pattern was
replaced by green fluorescence in HeLa cells. Flow cytometry also
showed that untreated cells with well-polarized, red-emitting
mitochondria localized in the upper left region of the plot. Cells
Fig. 6. 1c induced HeLa cell apoptosis. (A) AnnexinV/PI double staining and cytometry analysis of 1c-induced apoptosis in HeLa cells in a dose-dependent manner. (B) AnnexinV/PI
double staining and cytometry analysis of 1c-induced apoptosis in HeLa cells in a time-dependent manner. (C) Western blot analyses of apoptosis-related proteins (PARP, Cleaved-
PARP, Bcl-2, Bax and Cleaved Caspase-3) separated by SDS-PAGE. Relative band intensity was determined by Image J software. GAPDH served as a loading control. Images are
representative of three independent experiments. Data are means S.E.M. from three independent experiments (*, P < 0.05; **, P < 0.01).

H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
147
Fig. 7. 1c decreased the mitochondrial membrane potential of HeLa cells. (A) Cytofluorimetric analysis of mitochondrial membrane potential (
D
ѱm) by JC-1 staining. (B) Flow
cytometry analysis of HeLa cells with high Dѱm (%) by JC-1 staining. (C) Flow cytometry analysis of HeLa cells present PI positive. Each point represents the mean S.E.M. from three
replicates (**, P < 0.01).
significant influence on body weight of mice. However, shikonin
caused a significant decrease in body weight in comparison with
the control group. Thus, we speculated that tumor growth inhibi-
tion effect of 1c was increased and its cytotoxicity was reduced
significantly, as compared with shikonin. Meanwhile, 1c can also
effectively prolong the survival time in mice (Fig. 8C).
occurrence and development. PDH mainly exists in PDHC, and its
activity is control by the four PDKs and two PDPs. The joint action of
them determines the activation of PDH. Recently, it is difficult to
change PDH activity through targeting itself directly, therefore
current research mainly focuses on its direct regulators PDKs. To
increase the activity of PDHC by inhibiting PDKs is usually the
target of diabetes and heart disease [46,47], and now it has also
been applied to tumor [48].
4. Discussion
In 2007, Bonnent and his co-workers reported that dichloro-
acetic acid (DCA) can effectively inhibit viability of tumor cells and
growth of transplanted tumors [49]. DCA is a small molecular in-
hibitor of PDK and now it has been approved for the treatment of
congenital lactic acidosis. DCA activates PDH by inhibiting de-
phosphorylation of PDK, thereby promoting glycose oxidative
phosphorylation. During this process, a large amount of reactive
oxygen species (ROS) and cytochrome C were released, Kþ channel
potential was increased, thus causing outflow of Kþ, caspase
apoptosis pathway activation thus promoting apoptosis and tumor
cell growth inhibition. Some other researchers also reported that
DCA can increase sensitivity of tumor cells by chemotherapy, ra-
diation therapy and overcome drug resistance [50,51]. However, its
Warburg effect is a feature of tumor cell energy metabolism,
which generates ATP via aerobic glycolysis in cytoplasm rather than
via oxidative phosphorylation in mitochondrial. Glycolysis is an
energy production process of low efficiency, but it is a useful bal-
ance for tumor cells. For example, it may be the foundation of tu-
mor resistance to chemotherapy and radiotherapy, increasing
tumor cell mutation rate, thereby enhancing its capability of inva-
sion and metastasis [45].
PDH is a key link between glycolysis and Krebs cycle. As a crucial
enzyme in controlling pyruvate into the Krebs cycle, it plays an
important role in regulating glycolysis and oxidation phosphory-
lation. Therefore, PDH activity might bear some relations to tumor

148
H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
Fig. 8. 1c suppresses tumor growth of HeLa xenografts in nude mice. (A) Tumor volumes of the mice. (B) Body weight of the mice. (C) Survival curve of mice. (D) Four representative
tumors were shown (up panel), and the percentage of tumor weight were calculated (down panel). Data are the mean S.E.M. of 8 mice per group. (*, P < 0.05; **, P < 0.01 versus
vehicle (DMSO)).
serious liver and peripheral nerve toxicity can not be ignored. In
this study, CADD screening results guide us to choose the natural
co-factor, LA and its derivatives as the PDK1 binding moieties to
reduce the toxicity of shikonin towards normal cells and tissues and
improve drug targeting.
Shikonin is a highly active natural functional molecule with
multiple targets and has been demonstrated to induce apoptosis in
many tumor cells including lung cancer [52], prostate cancer [53],
glioma [22], leukemia [54], epidermoid carcinoma [55], stomach
carcinoma [27] cells, etc. In our pervious experiment, we found that
it can interrupt mitosis through disturbing tubulin polymerization
thus causing tumor cell cycle arrest in metaphase resulting in cell
death [27e29]. In terms of glucose metabolism, some researchers
reported that shikonin can inhibit lactate production and cellular
proliferation in tumor cells by targeting glycolysis regulators, py-
ruvate kinase M2 [30], C-MYC and GLUT1 [54].
proliferation (tubulin) and energy metablism (PDK1). This viable
hybrid design is particularly important as it optimizes the natural
product shikonin through structural modifications thereby
improving the selectivity of drugs thus reducing its cytotoxicity
towards non-cancer cells. Among all the compounds, 1c, the most
active one, was able to cause HeLa cell cycle arrest in G2/M phase as
tubulin inhibitor and inhibit PDK1 activity thereby breaking War-
burg effect and forcing HeLa cells to process more aerobic meta-
bolism to undergo cell apoptosis. Moreover, it can also increase ROS
production in tumor cells (SI Appendix,Fig. S5) and suppress HeLa
cell adhesive ability to fibronectin and laminin (SI Appendix,Fig. S6).
These results suggested that 1c can be a potential drug candidate
but needs further studies to support its anticancer evaluation. This
study also forms a basis for shikonin modification and reveals the
enhancement of shikonin through twin drug design framework for
the discovery of new and potent shikonin derivatives in study of
targeted cancer therapy. In order to screen out some potent PDK1
inhibitors, we chose a series of cancer cell lines which over-express
The present work describes the design and synthesis of novel
dual action hybrid compounds for synergistic inhibition of cell

H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
149
with poor prognosis in head and necksquamous cancer, Br. J. Cancer 98 (2008)
1975e1984.
[15] W. Zhang, S.L. Zhang, X. Hu, K.Y. Tam, Targeting tumormetabolism for cancer
treatment: is pyruvate dehydrogenase kinases (PDKs) a viable anticancer
target? Int. J. Biol. Sci. 11 (2015) 1390e1400.
[16] H. Hur, Y. Xuan, Y.B. Kim, G. Lee, W. Shim, J. Yun, I. Ham, S. Han, Expression of
pyruvate dehydrogenase kinase-1 ingastric cancer as a potential therapeutic
target, Int. J. Oncol. 42 (2013) 44e54.
PDK1 to perform anti-proliferation assay, which lead to a non-
significant selectivity of drugs to these cancer cell lines. Actually,
some cancer cell lines which did not over-express PDK1 should also
be chosen to further determine drug targeting and selectivity in
future research.
[17] G.M. Alushin, G.C. Lander, E.H. Kellogg, R. Zhang, D. Baker, E. Nogales, High-
Author contributions
resolution microtubule structures reveal the structural transitions in
a/b-
Tubulin upon GTP hydrolysis, Cell 157 (2014) 1117e1129.
XMW, YHY and HYL designed the study. HYL, HWH, WXS and
YSY performed experiments. GHL, CYT and JLQ analyzed the data.
HYL, YSY, XMW and YHY wrote and edited the manuscript with
input from other authors.
[18] G. Wang, F. Peng, D. Cao, Z. Yang, X. Han, J. Liu, W. Wu, L. He, L. Ma, J. Chen,
Y. Sang, M. Xiang, A. Peng, Y. Wei, L. Chen, Design, synthesis and biological
evaluation of millepachine derivatives as a new class of tubulin polymeriza-
tion inhibitors, Bioorg. Med. Chem. 21 (2013) 6844e6854.
[19] M.A. Jordan, K. Wendell, S. Gardiner, W.B. Derry, H. Copp, L. Wilson, Mitotic
block induced in HeLa cells by low concentrations of paclitaxel (Taxol) results
in abnormal mitotic exit and apoptotic cell death, Cancer Res. 56 (1996)
816e825.
Conflicts of interest
[20] I. Ojima, K. Kumar, D. Awasthi, J.G. Vineberg, Drug discovery targeting cell
division proteins, microtubules and FtsZ, Bioorg. Med. Chem. 22 (2014)
5060e5077.
[21] X. Chen, L. Yang, J.J. Oppenheim, M.Z. Howard, Cellular pharmacology studies
of shikonin derivatives, Phytother. Res. 16 (2002) 199e209.
The authors declare no conflict of interest.
Acknowledgements
[22] Z. Zhou, B. Lu, C. Wang, Z. Wang, T. Luo, M. Piao, F. Meng, G. Chi, Y. Luo, P. Ge,
RIP1 and RIP3 contribute to shikonin-induced DNA double-strand breaks in
glioma cells via increase of intracellular reactive oxygen species, Cancer Lett.
390 (2017) 77e90.
[23] W. Liang, A. Cai, G. Chen, H. Xi, X. Wu, J. Cui, K. Zhang, X. Zhao, J. Yu, B. Wei,
L. Chen, Shikonin induces mitochondria-mediated apoptosis and enhances
chemotherapeutic sensitivity of gastric cancer through reactive oxygen spe-
cies, Sci. Rep. 6 (2016) 38267e38278.
This research was supported by the Program for Changjiang
Scholars and Innovative Research Team in University (IRT_14R27),
National Natural Science Foundation of China (31470384, 31171161,
and 31670298), and Fundamental Research Funds for the Central
Universities (020814380002).
[24] X. Mao, C.R. Yu, W.H. Li, W.X. Li, Induction of apoptosis by shikonin through a
ROS/JNK-mediated process in Bcr/Abl-positive chronic myelogenous leukemia
(CML) cells, Cell Res. 18 (2008) 879e888.
Appendix A. Supplementary data
[25] F. Yang, Y. Chen, W.H. Duan, C. Zhang, H. Zhu, J. Ding, SH-7, a new synthesized
shikonin derivative, exerting its potent antitumor activities as a topoisom-
erase inhibitor, Int. J. Cancer 119 (2006) 1184e1193.
[26] L.S. Yoshida, T. Kawada, K. Irie, Y. Yuda, T. Himi, F. Ikemoto, H. Takano-
Ohmuro, Shikonin directly inhibits nitric oxide synthases: possible targets
that affect thoracic aorta relaxation response and nitric oxide release from
RAW 264.7 Macrophages, J. Pharmacol. Sci. 112 (2010) 343e351.
[27] H.Y. Lin, H.W. Han, L.F. Bai, H.Y. Qiu, D.Z. Yin, J.L. Qi, X.M. Wang, H.W. Gu,
Y.H. Yang, Design, synthesis and biological evaluation of shikonin thio-
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.ejmech.2017.12.019.
References
[1] O. Warburg, On the origin of cancer cells, Science 123 (1956) 309e314.
[2] M.G. Vander Heiden, L.C. Cantley, C.B. Thompson, Understanding the Warburg
effect: the metabolic requirements of cell proliferation, Science 324 (2009)
1029e1033.
glycoside derivatives: new antitubulin agents, RSC Adv.
4
(2014)
49796e49805.
[28] J. Guo, X.F. Chen, J. Liu, H.Y. Lin, H.W. Han, H.C. Liu, S.C. Huang, B.K. Shahla,
A. Kulek, J.L. Qi, X.M. Wang, L.J. Liang, Y.H. Yang,Novel shikonin derivatives
targeting tubulin as anti-cancer agents, Chem. Biol. Drug Des. 84 (2014)
603e615.
[3] R.A. Gatenby, R.J. Gillies, Why do cancers have high aerobic glycolysis? Nat.
Rev. Cancer 4 (2004) 891e899.
[4] R.J. DeBerardinis, J.J. Lum, G. Hatzivassiliou, C.B. Thompson, The biology of
cancer: metabolic reprogramming fuels cell growth and proliferation, Cell
Metab. 7 (2008) 11e20.
[5] R. Ferriero, C. Iannuzzi, G. Manco, N. Brunetti-Pierri, Differential inhibition of
PDKs by phenylbutyrate and enhancement of pyruvate dehydrogenase com-
plex activity by combination with dichloroacetate, J. Inherit. Metab. Dis. 38
(2015) 895e904.
[6] L.G. Korotchkina, M.S. Patel, Site specificity of four pyruvate dehydrogenase
kinaseisoenzymes toward the three phosphorylation sites of human pyruvate
dehydrogenase, J. Biol. Chem. 276 (2001) 37223e37229.
[7] S.L. Zhang, X. Hu, W. Zhang, H. Yao, K.Y. Tam, Development of pyruvate de-
hydrogenase kinase inhibitors inmedicinal chemistry with particular
emphasis asanticancer agents, Drug Discov. Today 20 (2015) 1112e1119.
[8] T. McFate, A. Mohyeldin, H. Lu, J. Thaker, J. Henriques, N.D. Halim, H. Wu,
M.J. Schell, T.M. Tsang, O. Teahan, S. Zhou, J.A. Califano, N.H. Jeoung,
R.A. Harris, A. Verma, Pyruvatedehydrogenase complex activity controls
metabolic and malignantphenotype in cancer cells, J. Biol. Chem. 283 (2008)
22700e22708.
[9] S.C. Tso, X. Qi, W.J. Gui, C.Y. Wu, J.L. Chuang, I. Wernstedt-Asterholm,
L.K. Morlock, K.R. Owens, P.E. Scherer, N.S. Williams, U.K. Tambar, R.M. Wynn,
D.T. Chuang, Structure guideddevelopment of specific pyruvatedehy-
drogenase kinaseinhibitors targeting the ATP-binding pocket, J. Biol. Chem.
289 (2014) 4432e4443.
[10] Y. Devedjiev, C.N. Steussy, D.G. Vassylyev, Crystal structureof anasymmetric
complex of pyruvate dehydrogenase kinase 3 withlipoyl domain 2 and its
biological implications, J. Mol. Biol. 370 (2007) 407e416.
[11] Z.H. Zhou, D.B. McCarthy, C.M. O'Connor, L.J. Reed, J.K. Stoops, Theremarkable
structural and functionalorganization of the eukaryotic pyruvatedehy-
drogenase complexes, Proc. Natl. Acad. Sci. 98 (2001) 14802e14807.
[12] S. Fujiwara, Y. Kawano, H. Yuki, Y. Okuno, K. Nosaka, H. Mitsuya, H. Hata, PDK1
inhibition is a novel therapeutic target inmultiple myeloma, Br. J. Cancer 108
(2013) 170e178.
[29] K.B. Shahla, L.J. Ling, H.Y. Qiu, L. Ma, H.Y. Lin, S.C. Huang, J.L. Qi, X.M. Wang,
G.H. Lu, Y.H. Yang, Synthesis and biological evaluation of novel shikonin ester
derivatives as potential anti-cancer agents, RSC Adv. 4 (2014) 35588e35596.
[30] J. Chen, J. Xie, Z. Jiang, B. Wang, Y. Wang, X. Hu, Shikonin and its analogs
inhibit cancer cell glycolysis by targeting tumor pyruvate kinase-M2, Onco-
gene 30 (2011) 4297e4306.
€
[31] B. Dorsam, J. Fahrer, The disulfide compound
a-lipoic acid and its derivatives:
a novel class of anticancer agents targeting mitochondria, Cancer Lett. 371
(2015) 12e19.
[32] L.J. Reed, C.S. Hornberger, Crystalline alpha-lipoic acid;
a catalytic agen-
tassociated with pyruvate dehydrogenase, Science 114 (1951) 93e94.
[33] V. Kagan, A. Shvedova, E. Serbinova, S. Khan, Dihydrolipoic acid-a universal
antioxidant both in the membrane and in aqueous phase, Biochem. Phar-
macol. 44 (1992) 1637e1649.
[34] Z. Zachar, J. Marecek, C. Maturo, S. Gupta, S.D. Stuart, K. Howell, A. Schauble,
J. Lem, A. Piramzadian, S. Karnik, K. Lee, R. Rodriguez, R. Shorr, P.M. Bingham,
Non-redox-active lipoate derivates disrupt cancer cell mitochondrial meta-
bolism and are potent anticancer agents in vivo, J. Mol. Med. 89 (2011)
1137e1148.
[35] T.W. Lycan, T.S. Pardee, W.J. Petty, M. Bonomi, A. Alistar, Z.S. Lamar, S. Isom,
M.D. Chan, A.A. Miller, J. Ruiz, A phase II clinical trial of CPI-613 in patients
with relapsed or refractory small cell lung carcinoma, PLoS One 11 (2016)
e0164244.
[36] T.S. Pardee, K. Lee, J. Luddy, C. Maturo, R. Rodriguez, S. Isom, L.D. Miller,
K.M. Stadelman, D. Levitan, D. Hurd, L.R. Ellis, R. Harrelson, M. Manuel,
S. Dralle, S. Lyerly, B.L. Powell, A phase I study of the first-in-class anti-
mitochondrial metabolism agent, CPI-613, in patients with advanced hema-
tologic malignancies, Clin. Cancer Res. 20 (2014) 5255e5264.
[37] R.B. Krol, S. Saksena, A. Prakash, New devices and hybridtherapies for treat-
ment of atrial fibrillation, J. Interv. Card. Electrophysiol. 4 (2000) 163e169.
[38] S.W. Wang, S.L. Pan, Y.C. Huang, J.H. Guh, P.C. Chiang, D.Y. Huang, S.C. Kuo,
K.H. Lee, C.M. Teng, CHM-1, a novel synthetic quinolone with potent and
selective antimitotic antitumor activity against human hepatocellular carci-
noma in vitro and in vivo, Mol. Cancer. Ther. 7 (2008) 350e360.
[39] A. Gruzman, H.J. Katzhendler, A. Haj-Yehie, S. Sasson, Synthesis and
[13] M.S. Patel, L.G. Korotchkina, S. Sidhu, Interaction of E1 and E3 components
with the core proteins of thehuman pyruvate dehydrogenase complex, J. Mol.
Catal. B-Enzym. 61 (2009) 2e6.
[14] S.M. Wigfield, S.C. Winter, A. Giatromanolaki, J. Taylor, M.L. Koukourakis,
A.L. Harris, PDK1 regulates lactate productionin hypoxia and is associated

150
H.-Y. Lin et al. / European Journal of Medicinal Chemistry 144 (2018) 137e150
characterization of new and potent alpha-lipoic acid derivatives, Bioorg. Med.
Chem. 12 (2004) 1183e1190.
switch in human colon cancer cells by targeting the pyruvate dehydrogenase
complex, Int. J. Cancer 128 (2011) 2591e2601.
[40] H.Y. Lin, Z.K. Li, L.F. Bai, S.K. Baloch, F. Wang, H.Y. Qiu, X. Wang, J.L. Qi,
R.W. Yang, X.M. Wang, Y.H. Yang, Synthesis of aryl dihydrothiazol acyl shi-
konin ester derivatives as anticancer agents through microtubule stabiliza-
tion, Biochem. Pharmacol. 96 (2015) 93e106.
[41] G.F. Mangiatordi, D. Trisciuzzi, D. Alberga, N. Denora, R.M. Iacobazzi,
D. Gadaleta, M. Catto, O. Nicolotti, Novel chemotypes targeting tubulin at the
Colchicine Binding Site and unbiasing P-glycoprotein, Eur. J. Med. Chem. 139
(2017) 792e803.
[42] C. Da, S.L. Mooberry, J.T. Gupton, G.E. Kellogg, How to deal with low-
resolution target structures: using SAR, ensemble docking, hydropathic
analysis, and 3D-QSAR to definitively map the ab-tubulin colchicine site,
J. Med. Chem. 56 (2013) 7382e7395.
[49] S. Bonnet, S.L. Archer, J. Allalunis-Turner, A. Haromy, C. Beaulieu,
R. Thompson, C.T. Lee, G.D. Lopaschuk, L. Puttagunta, S. Bonnet, G. Harry,
K. Hashimoto, C.J. Porter, M.A. Andrade, B. Thebaud, E.D. Michelakis,
A mitochondria-Kþ channel axis is suppressed in cancer and its normalization
promotes apoptosis and inhibits cancer growth, Cancer Cell 11 (2007) 37e51.
[50] W. Cao, S. Yacoub, K.T. Shiverick, K. Namiki, Y. Sakai, S. Porvasnik, C. Urbanek,
C.J. Rosser, Dichloroacetate (DCA) sensitizes both wild-type and over
expressing Bcl-2 prostate cancer cells in vitro to radiation, Prostate 68 (2008)
1223e1231.
[51] W. Fiebiger, U. Olszewski, E. Ulsperger, K. Geissler, G. Hamilton, In vitro
cytotoxicity of novel platinum-based drugs and dichloroacetate against lung
carcinoid cell line, Clin. Transl. Oncol. 13 (2011) 43e49.
[43] Y. Lu, J. Chen, M. Xiao, W. Li, D.D. Miller, An overview of tubulin inhibitors that
interact with the colchicine binding site, Pharm. Res. 29 (2012) 2943e2971.
[44] X.J. Zhu, Y. Shi, J. Peng, C.S. Guo, N.N. Shan, P. Qin, X.B. Ji, M. Hou, The effects of
BAFF and BAFF-R-Fc fusion protein in immune thrombocytopenia, Blood 114
(2009) 5362e5367.
[45] W. Wu, S. Zhao, Metabolic changes in cancer: beyond the Warburg effect, Acta
Bioch. Bioph. Sin. 45 (2013) 18e26.
[46] J.M. Bryson, G.J. Cooney, V.R. Wensley, J.L. Phuyal, I.D. Caterson, Tissue dif-
ferences in the response of the pyruvate dehydrogenase complex to a glucose
load during the development of obesity in gold-thioglucose-obese mice,
Biochem. J. 305 (1995) 811e816.
[52] X. Li, X.X. Fan, Z.B. Jiang, W.T. Loo, X.J. Yao, E.L. Leung, L.W. Chow, L. Liu,
Shikonin inhibits gefitinib-resistant non-small cell lung cancer by inhibiting
TrxR and activating the EGFR proteasomal degradation pathway, Pharmacol.
Res. 115 (2017) 45e55.
[53] L.J. Kuo, C.Y. Huang, W.L. Cheng, C.S. Hung, C.T. Wu, F.Y. Lin, Y.J. Chang,
M.T. Huang, Glucose-regulated protein 78 mediates the anticancer efficacy of
shikonin in hormone-refractory prostate cancer cells, Tumor. Biol. 36 (2015)
5063e5070.
ꢀ
ꢀ
ꢁ
ꢁ
ꢁ
[54] K. Valis, P. Talacko, V. Grobarova, J. Cerný, P. Novak, Shikonin regulates C-MYC
and GLUT1 expression through the MST1-YAP1-TEAD1 axis, Exp. Cell Res. 349
(2016) 273e281.
[47] W. Sun, Q. Liu, J. Leng, Y. Zheng, J. Li, The role of Pyruvate Dehydrogenase
Complex in cardiovascular diseases, Life Sci. 121 (2015) 97e103.
[48] J.M. Blouin, G. Penot, M. Collinet, M. Nacfer, C. Forest, P. Laurent-Puig,
X. Coumoul, R. Barouki, C. Benelli, S. Bortoli, Butyrate elicits a metabolic
[55] T. Rong, L. You, G. Mei, Shikonin causes cell-cycle arrest and induces apoptosis
by regulating the EGFR-NF-
kB signalling pathway in human epidermoid car-
cinoma A431 cells, Biosci. Rep. 35 (2015) 189e201.