Exploring the formation and recognition of an important G-quadruplex in a HIF1α promoter and its transcriptional inhibition by a benzo[c]phenanthridine derivative

Article abstract of DOI:10.1021/ja412128w

Four putative G-quadruplex sequences (PGSs) in the HIF1α promoter and the 5′UTR were evaluated for their G-quadruplex-forming potential using ESI-MS, CD, FRET, DMS footprinting, and a polymerase stop assay. An important G-quadruplex (S1) has been proven to inhibit HIF1α transcription by blocking AP2 binding. A benzo[c]phenanthridine derivative was found to target the S1 G-quadruplex and induce its conformational conversion from antiparallel to parallel orientation. The transcriptional suppression of HIF1α by this compound was demonstrated using western blotting, Q-RT-PCR, luciferase assay, and ChIP. Our new findings provided a novel strategy for HIF1α regulation and potential insight for cancer therapy.

Full text of DOI:10.1021/ja412128w

Article
pubs.acs.org/JACS
Exploring the Formation and Recognition of an Important
G‑Quadruplex in a HIF1α Promoter and Its Transcriptional Inhibition
by a Benzo[c]phenanthridine Derivative
†
†,§
†
†
‡
,†
Han Chen, Haitao Long, Xiaojie Cui, Jiang Zhou, Ming Xu, and Gu Yuan*
†Beijing National Laboratory for Molecular Sciences, Key Laboratory of Bioorganic Chemistry and Molecular Engineering of Ministry
of Education, Department of Chemical Biology, College of Chemistry and Molecular Engineering, Peking University, Beijing 100871,
China
‡
Key Laboratory of Molecular Cardiovascular Sciences, Ministry of Education, Peking University Third Hospital, 49 North Garden
Road, Beijing 100191, China
§
School of Chemistry, Beijing Institute of Technology, 5 South Zhongguancun Street, Beijing 100081, China
*
S Supporting Information
ABSTRACT: Four putative G-quadruplex sequences (PGSs) in the HIF1α
promoter and the 5′UTR were evaluated for their G-quadruplex-forming
potential using ESI-MS, CD, FRET, DMS footprinting, and a polymerase stop
assay. An important G-quadruplex (S1) has been proven to inhibit HIF1α
transcription by blocking AP2 binding. A benzo[c]phenanthridine derivative
was found to target the S1 G-quadruplex and induce its conformational
conversion from antiparallel to parallel orientation. The transcriptional
suppression of HIF1α by this compound was demonstrated using western
blotting, Q-RT-PCR, luciferase assay, and ChIP. Our new findings provided a
novel strategy for HIF1α regulation and potential insight for cancer therapy.
INTRODUCTION
located within the GC-rich region of the HIF1α promoter and
■
14
5
′UTR.
Over the past three decades, HIF1α (hypoxia-inducible factor-1
Recently, G-quadruplexes (G4) in human genomes were
alpha) has been revealed to be an important target for cancer
1
,2
implicated in apoptosis, gene transcription, and gene
expression. Many putative G-quadruplex sequences (PGSs)
located in the promoters of oncogenes, such as bcl-2, c-myc, c-
therapy and has attracted exponentially growing interest in
3
−5
the field of cancer biology.
HIF1α is the subunit of a heterodimeric protein called
15−18
6
myb, and c-kit,
were proven to have potential therapeutic
HIF1. During normoxia, prolyl hydroxylases (PHDs) hydrox-
mechanisms and were validated as targeting sites for drugs. A
PGS located in the +36 to +56 region (NCBI, gene ID 3091) of
ylate the prolyl residues at P402 and P564 of HIF1α and induce
degradation via the ubiquitin proteasome pathway. However,
during hypoxia, nonhydroxylated HIF1α survives, exhibiting
transcriptional activation for more than 70 putative hypoxia-
inducible genes, such as VEGF, Glut1, Bcl2, EPO, to affect
many cell processes, including angiogenesis, glucose metabo-
lism, antiapoptosis, erythropoiesis, and other functions.
overexpression behavior has been detected in numerous
tumors, including colon, breast, gastric, lung, ovarian, prostate,
7
1
9−21
HIF1α 5′UTR has been investigated.
However, the
bioinformatics show that there are three other G-rich sequences
that contain PGS located in the −20 to −1, +9 to +30, and
+
481 to 500 regions of HIF1α. In particular, there are two AP2
3
,8,9
binding sites that overlap with the −20 to +29 region, enabling
their action as a critical cis-acting element for regulating HIF1α
expression (Scheme 1).
This
1
0,11
In this paper, we evaluated the formation of G-quadruplexes
from four PGSs in the HIF1α promoter and the 5′UTR, as well
as their importance for HIF1α transcription and binding to a
transcriptional factor (AP2). In addition, a series of synthetic
benzo[c]phenanthridine derivatives were screened as binding
candidates for these G-quadruplexes. The inhibition of the
transcriptional activity and mechanism on HIF1α were also
explored.
renal, and pancreatic cancers.
In addition to post-transcriptional regulation, the transcrip-
tional activation of HIF1α during hypoxia has also been
reported in cell lines including Hep3B, HepG2, HeLa, LN229,
6
,12
and L929.
The sequence of the HIF1α promoter
demonstrates that it belongs to the TATA-less promoter
family; this family often depends on Sp1, NF-1, AP2 binding
site sequences located up- or downstream from the tran-
scription start site (TSS). The work reported by E. Minet et
al. shows that the constitutive transcription of HIF1α is partially
mediated by Sp1 binding sites and the other cis-acting elements
1
3
Received: November 27, 2013
Published: January 22, 2014
©
2014 American Chemical Society
2583
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
FRET. The FRET experiments were carried out using a F7000
fluorescence spectrometer (Hitachi, Japan). A series of samples
containing 0−800 mM KCl and 0.5 μM labeled ODN (S1fret, S1fam,
S1rox, Supporting Information Table S1) were diluted in 30 mM Tris-
HCl buffer (pH = 7.4) and annealed at 95 °C for 5 min before slowly
cooling to 4 °C over 8 h. The excitation wavelength for 6-FAM was
Scheme 1. G-Rich Sequences in HIF1α Gene
4
85 nm. The samples were scanned at 240 nm/min twice to obtain the
average emission spectrum from 500 to 750 nm.
NMR. The NMR experiments were performed on a Bruker AV700
spectrometer (700 MHz) at room temperature. The S1 was dissolved
in a solution containing 10% D O, 100 mM KCl, and 25 mM
2
potassium phosphate (pH = 7.0) to a final concentration of 1 mM.
1
The 1-D H NMR spectrum was collected.
DMS Footprinting. The DMS footprinting assay was performed as
2
3,24
previously described.
The FAM-labeled ODN (S1fam,S1fam2,
EXPERIMENTAL SECTION
■
Supporting Information Table S1) was diluted with 60 mM Tris-HCl
buffer (pH = 7.4) to 0.2 μM while containing 0−100 mM salt (LiCl or
KCl); afterward, preannealing at 95 °C was performed for 10 min
before slowly cooling to 4 °C. The annealed samples were treated with
Materials. The MCF-7 cell line, the growth media, and the serum
were purchased from Sangon Biotech. The anti-AP2α was purchased
from Santa Cruz; anti-HIF1α and anti-β-actin were acquired from Cell
Signaling. The oligodeoxygenucleotides (ODNs) were ordered from
Sangon Biotech (Supporting Information Table S1). The chemical
reagents were purchased from ACROS or Sinopharm Chemical
Reagent Beijing Co., Ltd.
1
0% DMS for 5 min before being quenched and extracted with a Tris−
phenol−chloroform solution (pH = 8.0). The aqueous phase was
precipitated using ethanol at −80 °C. The dry precipitate was
dissolved in 10% piperidine and incubated at 90 °C for 30 min,
followed by precipitation. The treated sample was resolved using 20%
denatured PAGE gel at 1500 V for 4 h in a 4 °C cold room before
being imaged with a GE Healthcare Typhoon9400 gel scanner.
Polymerase Stop Assay. A polymerase stop assay was performed
Drug Synthesis. The benzo[c]phenanthridine derivatives were
22
synthesized in our laboratory as previously described. The synthetic
details were provided in the Supporting Information. The agents used
for screening are shown in Chart 1.
2
5,26
as previously described.
Information Table S1) was diluted with PCR buffer containing 2.5
The template (S1t−S4t, Supporting
Chart 1. Structures of Benzo[c]phenanthridine Derivatives
mM MgCl , 500 nM PSA primer (Supporting Information Table S1),
2
0
−150 mM KCl, and 0−40 μM small molecules to 200 nM in total
volume of 19 μL. The PCR mixture was annealed at 95 °C for 5 min
and was slowly cooled to 4 °C over 8 h. Afterward, 1 μL of Taq
polymerase was added, and the mixture was incubated at 37 °C for 15
min, followed by quenching on ice. The PCR sample was resolved
using 12% denatured PAGE gel at 1500 V for 2 h in a 4 °C cold room
before being imaged with a GE Healthcare Tyhoon9400 gel scanner.
Western Blotting. The cultured MCF-7 cell lines were split into
5
six-well plates (∼8 × 10 cells/well) and incubated at 37 °C overnight.
Powdered M3 was dissolved in ethanol to generate a 10 mM stock
solution; this solution was diluted with complete growth medium to
form a final working solution (0−20 μM), and ethanol was the control.
After the cells settled down, we changed the growth medium in the six-
well plate to a drug working solutions from 0 to 20 μM. The cells were
maintained in a hypoxia chamber at 37 °C, with 5% CO , 1% O , and
2
2
9
4% N for 18 h. The samples were using a general harvest procedure
2
with 1× lysis buffer and PMSF. The samples were loaded in a 10%
SDS gel and transferred to a nitrocellulose membrane with 1× transfer
buffer at 45 V at 4 °C for 18 h. Afterward, the samples were blocked
with 5% nonfat milk at room temperature for 1 h and probed with
specific primary antibodies at 4 °C overnight. The target proteins were
visualized via treatment with an HRP-linked secondary antibody.
Q-RT-PCR. The cultured MCF-7 cells were treated as described
above. The total RNA was extracted using Trireagent (Invitrogen)
while following the manufacturer’s instructions. The concentration of
the total RNA was measured with a NanoDrop 1000 (Thermo
Scientific). The TransScript II All-in-One First-Strand cDNA Synthesis
SuperMix for qPCR (Beijing, Transgen) kits were used for cDNA
synthesis in a 20 μL volume with 1 μg total RNA. The mixed samples
were incubated at 42 °C for 1 h, heated at 85 °C for 10 min, and
stored at −20 °C. Next, 12.5 μL of SYBR Green PCR Master Mix
Mass Spectrometry. Normal ESI mass spectra were obtained in
negative ion mode with a Finnigan LCQ Deca XP Plus ion trap mass
spectrometer (San Jose, CA). The direct infusion flow-rate was 2.0
μL/min. The electrospray source conditions included a 2.7 kV spray
voltage and a 120 °C capillary temperature. During all experiments, the
scanned mass range was 500−2000 u. The small molecules (M1−M7)
were directly dissolved in methanol to prepare 1 mM stock solution,
respectively. The single-stranded ODN was directly dissolved in a
mixture containing 25% methanol, 25% ammonium acetate solution
(
150 mM in final), and 50% ddH O. The final concentration of the
2
ODN was 10 μM.
Circular Dichroism. The CD experiments were carried out with a
J-815 CD spectrometer (JASCO, Japan); 10 μM ODN (S1−S4,
Supporting Information Table S1) was diluted in 150 mM KCl and 30
mM Tris-HCl buffer (pH = 7.4). Afterward, it was annealed at 95 °C
for 10 min and slowly cooled to 4 °C over 8 h. To titrate the KCl and
small molecules, a series of samples containing 0−150 mM KCl and
(Applied Biosystem), 10.5 μL H O, and 0.5 μL of the forward and
2
reverse primers (Table 1) were mixed and placed in each well of an
optical reaction plate. Subsequently, 1 μL of each cDNA sample was
added to the well to achieve a final volume of 25 μL. β-Actin was used
as a control. The reaction was performed in an Eppendorf Realplex
Real-time PCR System.
0
−40 μM M3 were prepared. The samples were scanned in a 1.0 cm
Luciferase Assay. The luciferase assay was performed as
2
7
path-length cuvette for three times to obtain the average spectrum
from 200 to 400 nm.
previously described. The pGL4.10-basic vector was purchased
from Promega. A 1.6-kb fragment of human HIF1α promoter (from
2
584
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
Table 1. ESI-MS Screening Results
the first exon of HIF1α (Scheme 1). S1 and S2 are very close to
the initial transcription site, and S1−3 overlap with the AP2
binding sites; these sites may be critical for regulating the
M1−M7 targeting to S1 G-
quadruplex
M3 targeting to DNA
14
transcription of Hif1α via their secondary structures.
agents
IRa
DNA
IRa
The potential G-quadruplex formation of S1−4 was
evaluated by ESI-MS, CD spectrometry, FRET, DMS foot-
printing, and a polymerase stop assay. In H O−CH OH (3:1)
M1
M2
M3
M4
M5
M6
M7
0.33
0.41
0.79
0.49
0.33
0.17
0.28
Q1
0.79
0.62
0.41
0.33
0.09
Q2
2
3
Q3
Q4
without NH OAc, S1 showed a distinctive pattern including a
4
series of multicharge peaks, revealing an unfolded structure
S1 double strand DNA
without monovalent cations (Figure 1a). In H O−CH OH
2
3
−1041 to +566) was generated via PCR from Hela cells, and four
mutation fragments were constructed. The amplified fragments were
attached to the Kpn I/Hind III sites of a pGL4.10-basic vector and
introduced into E. coli. All constructed vectors were verified via
automated sequencing.
The cultured MCF-7 cell lines were split into 24-well plates (∼1 ×
4
1
0 cells/well) and incubated at 37 °C overnight. To verify the activity
of the mutation on the promoter, 400 ng of PGL4.10 or one of the five
constructed vectors was transfected into MCF-7 via lipofectamine
2000, and 20 ng of p-RL-TK vector was cotransfected as the internal
control. To verify the effect of small molecules on the HIF1α
promoter activity, 10 mM of the M3 stock solution was diluted with
complete growth medium to form a final working solution (0−20 μM)
that was added to 24-well plates during transfection. The cells were
harvested, and the luciferase activity was measured using a Dual-
Luciferase Reporter Assay System according to the manufacturer’s
instructions. The relative activity was normalized using the ratio of
Firefly luciferase activity to Renilla luciferase activity and calculated
using the folds to control pGL4.10 vector.
Chromatin Immunoprecipitation Assay. The MCF-7 cell line
was cultured in 100 mm plates. To verify the effect of small molecular,
a 10 mM stock solution of M3 was diluted with complete growth
medium to form a final working solution (0−15 μM). The cells were
treated with the working solution and incubated overnight. The
general procedure was performed as instruction manual of ChIP Assay
kit (#P2078, Beyotime) provided by manufacturer. The treated cells
were washed with 9 mL of fresh growth medium containing 4%
formaldehyde at room temperature for 10 min, followed by mixing
with 1 mL of 10 × glysin buffer and incubation for another 10 min.
After removing the growth medium, the cells were resuspended with
Figure 1. Mass spectra of S1 in H O−CH OH (3:1) without
2
3
NH OAc (a) and in H O−CH OH (3:1) with 150 mM NH OAc (b).
4
2
3
4
(3:1) with 150 mM NH OAc, an ion at m/z =1323 was
4
observed in the mass spectrum, which is the complex ion [S1 +
+
+ 5‑
NH − 6H ] (Q1), indicating the formation of a 2-layer G-
4
6
trypsin and counted. For each sample, 4 × 10 cells were lysed via
supersonication and diluted with ChIP buffer. The nonspecific protein
was precipitated with A+G agarose. The supernatant was mixed with
28,29
quadruplex.
(Figure 1b). In contrast, a mutation of S1,
which would completely disrupt the G-quadruplex, only
showed a peak at m/z 1320, representing a single S1m without
any ammonium ions (Supporting Information Figure S1). The
previous work has proven that additional ammonium ions are
strong evidence for G-quadruplex formation during ESI-MS
10 μL of anti-AP2 antibody and rotated overnight in a cold room,
followed by antibody collection using 60 μL of A+G agarose. Half of
the precipitate was washed sequentially with low salt buffer, high salt
buffer, LiCl buffer, and TE buffer before elution with SDS buffer.
Afterward, the eluted material was denatured in 5 M NaCl at 65 °C for
27−29
analysis.
The same evidence has been found in S2, S3, and
4
h and concentrated to 30 μL with a DNA Purification Kit (#D0033,
S4, forming 2-, 3-, and 3-layer G-quadruplexes (Q2, Q3, and
Q4), respectively (Supporting Information Figure S1).
Beyotime). The purified DNA was amplified using PCR with HIF1α
promoter primers (Supporting Information Table S1) and resolved on
a 6% PAGE gel with a DNA ladder marker of 200−250bp. The
remaining precipitate was washed and analyzed via western blotting
with an HRP-linked anti-IgG antibody. For the plasmid ChIP assay,
the cells were transfected with 2 μg of PGL4.10 or one of the five
constructed vectors via lipofectamine 2000 for 6 h, followed by
treatment with the drug working solution overnight. The resuspended
cells were lysed with SDS buffer, and the nucleic chromatin was
removed via centrifugation at 4500 rpm before supersonication.
To confirm the formation of the G-quadruplexes in the
HIF1α promoter, CD experiments were performed. The 10 μM
solution of S1 in Tris-HCl (30 mM, pH = 7.4) mainly showed a
broad positive peak from 240 to 300 nm in the CD spectrum,
3
0,31
indicating an unfolded S1 (Figure 2).
After increasing the
concentration of KCl from 0 to 150 mM, the CD spectra of S1
exhibited a dramatic change including an increase at
approximately 292 nm and a decrease at approximately 262
nm (Supporting Information Figure S2). In 150 mM KCl, S1
showed positive peaks at 292, 243, and 211 nm and negative
peaks at 262 and 230 nm. The strong positive peak at 292 nm
revealed that an antiparallel G-quadruplex structure formed in
RESULTS AND DISCUSSION
■
Formation of G-Quadruplexes in the HIF1α Promoter
and 5′UTR Regions. A bioinformatics search of the NCBI
database revealed that there are four PGSs (Sx, x = 1−4) in
32
150 mM KCl (Figure 2). The result of FRET assay confirmed
−
500 to +500 regions that include the promoter, 5′UTR, and
also the formation of S1 G-quadruplex (Supporting Informa-
2
585
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
that is accessible to solvent, protecting these residues from
methylation by DMS and diminishing the respective cleaved
24
bands, the DMS footprinting assay was performed to identify
which guanosines can be integrated into the G-quadruplex
structure. The DMS experiments demonstrated that the S1
sequence forms a G-quadruplex structure in 150 mM KCl, and
the 1st, 2nd, 4th, 5th, 7th, 8th, 10th, and 11th guanosines
primarily participated in G-quadruplex formation because their
bands were weaker than those of the 3rd, 6th, 9th, and 12th
guanosines. Both 3′- and 5′-labeled S1 exhibited the same
cleavage pattern (Figure 3c and Supporting Information Figure
S5). In addition, the NMR spectrum of S1 in KCl solution
showed that the peaks of the imino protons at δ 10.8−12.0
ppm corresponded to the eight imino protons of the guanines
participating in the S1 G-quadruplex formation, confirming a 2-
layer G-quadruplex formed (Supporting Information Figure
S6).
Figure 2. CD spectra of S1 dissolved in 30 mM Tris-HCl buffer (pH =
7.4) to 10 μM. The solid line is the CD spectrum in the absence of
KCl, and the dotted line is that in the presence of 150 mM KCl.
tion Figure S3). In addition, S2 showed the formation of an
antiparallel G-quadruplex in high concentrations of KCl, while
S3 and S4 formed parallel G-quadruplexes with a major peak at
G-Quadruplex in the HIF1α Promoter May Regulate
the Transcription of HIF1α. Because four PGSs in the
HIF1α promoter have been identified and proven to form the
3
0
2
62 nm (Supporting Information Figure S2).
2
- or 3-layer G-quadruplex in vitro, four mutation plasmids were
A polymerase stop assay was used to verify the formation of
constructed to investigate their function during the tran-
scription of HIF1α. Each plasmid contained only one mutation
in which two GC base pairs were replaced with two TA base
pairs on one of the four PGSs, blocking the formation of one of
the four G-quadruplexes. Each mutation avoided any possible
the G-quadruplex. S1, S3, and S4 exhibited strong stop bands
with 150 mM KCl, while barely a band was observed without
KCl. The blocking of Taq polymerase indicated the formation
of a special structure that was induced by KCl, suggesting a
25
stable G-quadruplex (Figure 3a,b). The stop bands for the S3
14
protein binding sites, as previously reported. The mutation
sites are shown in Figure 4a, and the four mutation plasmids
were named PS1−PS4.
A dual-luciferase assay was used to evaluate the effect of the
mutation on HIF1α transcription. The wild type (WT) and
mutation plasmids exhibited 15- to 30-fold fluorescence relative
to a blank plasmid (pGL4.10), demonstrating that the inserted
fragment maintained the activity of the HIF1α promoter. PS1,
PS2, and PS3 showed a higher transcription level than WT,
indicating the down-regulation of these PGSs (Figure 4b).
However, the PS4 maintained the similar transcription level
with that of the WT due to the long distance between this PGS
and TSS (approximately 480 bp). Of the up-regulation
mutation plasmids, only PS1 was significantly different from
the WT (P < 0.05), exhibiting an approximate 50% increase in
the transcription level relative to the wild type promoter. The
mutation at the site of −15 to −14 completely disrupted the G-
quadruplex without disrupting the sequence of the HRE (−23
to −17) or AP2 binding site (−10 to −1), reducing the
resistance against the transcription inducing by the G-
quadruplex while maintaining the activity of the other trans-
acting factors. In addition, because the formation of a G-
quadruplex may reject protein binding to DNA primary
sequence, the absence of the G-quadruplex with keeping
binding site may increase the activity of these trans-acting
factors, such as AP2. Although PS2 exhibited an increase similar
to PS1, the poor P value (approximately 0.11) and the minimal
stabilization of the G-quadruplex in vitro indicated a very
limited effect for this PGS. The PS3 contained multiple G-rich
fragments that may form different G-quadruplexes. Although
the mutation at +34 to +35 may disrupt the optimal 3-layer G-
quadruplex, the 2-layer or long loop G-quadruplex remained a
possibility. The increased transcription level induced by PS3
showed that the disruption of the 3-layer G-quadruplex may up-
regulate HIF1α, but this trend had been limited by its involute
and various structure.
Figure 3. Polymerase stop assay (a) and the relative intensity of the
stop bands (b) at various KCl concentrations using S1 template:
Extension through the template was arrested as the KCl concentration
increased from 0 to 150 mM. The DMS footprinting comparison (c)
used 6-FAM labeled S1 incubating in 30 mM Tris-HCl buffer, 150
mM LiCl, or 150 mM KCl to identify the guanosines involved in the
G-quadruplex.
and the S4 were clearer than that for the S1, suggesting that the
3-layer G-quadruplex was more stable than the 2-layer G-
quadruplex (Supporting Information Figure S4). However,
there was no obvious stop band observed for the S2, suggesting
an unstable G-quadruplex because of a 2-layer G-quadruplex
with long loop.
1
2
3
4
5
6
7
8
Because 12 guanosines (5′-AG G TG AG G CG G G CTT-
9
10 11 12
G CG G G A-3′) are present in the S1, and the folding of
the guanosines into a G-quarter will produce a smaller surface
2
586
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
maintained similar levels, demonstrating that the transcriptional
up-regulation of PS1 was primarily mediated by the AP2
pathway; the −10 to −1 region was a key cis-element that was
regulated by the competition between the G-quadruplex and
the AP2.
Recognition of the HIF1α G-Quadruplex by Benzo[c]-
phenanthridine Derivatives. A series of benzo[c]-
phenanthridine derivatives were used to select the recognition
molecules that bind to the HIF1α G-quadruplexes. Each of
these small molecules was mixed with the HIF1α G-quadruplex
and analyzed using ESI-MS. Figure 5a and Supporting
Figure 4. Schematic representation of the constructed luciferase
plasmids for WT and PS1−4: (a) each PGS was mutated, and the G-
quadruplex was disrupted separately. The relative luciferase activity (b)
of the constructed plasmids showed an up-regulated transcription via
mutation on S1, S2, and S3. All luciferase assays were conducted in
octuplicate and normalized to pGL4.10 control. The plasmid ChIP
assay using an anti-AP2 antibody (d) and the PCR product relative
intensity (c) showed that mutation of the S1 G-quadruplex enhanced
the AP2 binding affinity to HIF1α promoter. All ChIP assays were
conducted in triplicate and normalized to WT.
Figure 5. (a) ESI mass spectrum of M3 bound to the S1 G-
A previous report suggested that AP2 may play a key role
quadruplex: the S1 was dissolved in H O−CH OH (3:1) with 150
1
0,33
2
3
during HIF1α transcription,
and all of the PS1, PS2, or PS3
mM NH OAc, and the ligand−S1 ratio was 2:1. (b) The CD spectrum
4
12
contain AP2 binding sequence. A plasmid chromatin
immunoprecipitation assay was performed to evaluate the
binding affinity of AP2 to the wild type and mutation plasmids
containing the HIF1α promoter sequence. In the cytoplasm,
the plasmids were bound to AP2 and cross-linked with
formaldehyde. Subsequently, the AP2 was bound to an anti-
AP2 antibody and precipitated via anti-IgG beads. The enriched
plasmids were recovered using a high concentration of NaCl at
of 5 μM S1 mixed with (solid line) or without (dotted line) 40 μM M3
in 30 mM Tris-HCl and 50 mM KCl buffer showed that M3 could
induce the conformational change of the S1 G-quadruplex from
antiparallel to parallel orientation.
Information Figure S7 showed parts of the ESI mass spectra
for the mixtures containing the S1 and the small molecules in a
1:2 molar ratio as examples. The relative intensity of the
complex ion of Q1 with M3 is 100% in the mass spectrum,
indicating that M3 has a high affinity for the S1 G-quadruplex.
6
5 °C followed by purification. The primers for the HIF1α
promoter (promoter-f and promoter-r, Supporting Information
Table S1) were used for PCR, and the products (230bp) were
resolved using PAGE. The higher binding affinity of AP2 for
plasmids will generate more enriched plasmids and more
intense PCR product bands at 230 bp. Because most of the
nucleic DNA was removed before supersonication, and the
blank plasmid of pGL4.10 did not contain HIF1α promoter
primers binding site, very little PCR product was observed for
the blank sample. The AP2 binding affinity was evaluated using
the band intensity ratio of the ChIP bands versus the input
bands with a blank sample as the background and normalization
relative to the WT. Figure 4c,d shows that only PS1 exhibited a
notable increase in AP2 affinity, while the other plasmids
Its primary binding mode is 1:2, as characterized by a base peak
+
at m/z 1484 that represents a complex ion ([S1 + NH + 2M3
4
+
5‑
− 6H ] ). The 1:1 and 1:3 species were also observed at m/z
+
+ 5‑
1404 and 1564, representing [S1 + NH + M3 − 6H ] and
4
[S1 + NH4+ + 3M3 − 6H ] , respectively. M1 and M2 also
showed a series peaks of the complex ions with Q1 at the 40−
50% relative intensity; however, the relative intensity of the
complex ions for M6 with Q1 was much weaker, indicating
poor affinity.
+ 5‑
To evaluate the binding affinity of small molecules (Mx, x =
1−7) to the G-quadruplex (Qx, x = 1−4), IR was defined as
a
2
587
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
the relative abundance ratio of all the bound ions (n = 1−3) to
treatment according to the general procedure for DMS
footprinting. Except for G2 and G9, all guanosine cleavage
bands faded in the presence of M3 due to the multibinding
mode of M3 demonstrated by ESI-MS. In particular, G4−G8
almost completely disappeared, suggesting that the 5′-
GGCGGG-3′ fragment was the primary binding site for M3.
However, G9 showed an enhanced cleavage signal due to its
location on a large side loop, which swings outward from the G-
quadruplex (Figure 6 and Supporting Information Figure S9).
27
that of both unbound and bound species (n = 1−3).
5
−
I [Qx + nMx]
r
IR =
a
5−
5−
5−
Ir[Sx] + I [Qx] + ∑ I [Qx + nMx]
r
r
5‑
The I [Qx + nMx] was the integral intensity of the complex
r
ion of Qx−Mx at molar ratio of 1:n in the ESI mass spectrum.
5‑
5‑
The I [Sx] and the I [Qx] were the integral intensity of Sx
r
r
and Qx (x = 1−4). The maximum value of IR is 1.00, the
a
nearer the IR value to 1.00, the higher the binding affinity of
a
Mx to Qx.
The IR values of the benzo[c]phenanthridine derivatives
a
binding to Q1 were shown in Table 1. In these derivatives, only
M3 was a preferable candidate for binding Q1 with a highest
IR at 0.79, while those of the other derivatives were lower than
a
0
.5.
The binding selectivity for M3 to the S1−4 G-quadruplexes
(
Q1−4) and the double strand of S1 were also evaluated by
ESI-MS. Table 1 summarized the IR values for M3 binding
a
with Q1−4 and the S1 double-stranded DNA, which revealed
that M3 had the best binding affinity (IR = 0.79) with the S1
a
G-quadruplex (Q1), while Q2 showed the lower IR of 0.62.
a
The binding affinities of M3 to Q3 and Q4 were much weaker
than that of Q1. The S1 double stranded bound to M3 was
barely observed (IR = 0.09). These results illustrated that M3
a
had a selective and high binding affinity for the S1 G-
quadruplex. It could bind primarily to the S1 G-quadruplex in
the HIF1α promoter.
Figure 6. DMS footprinting of S1fam2 in H O, 50 mM LiCl, 50 mM
2
KCl, or 50 mM KCl with 40 μM M3 (a) and the relative cleaving band
intensity of the sample in KCl with M3 normalized to that in KCl
without M3 (b) showed a decrease in cleavage around G4−G8,
revealing the binding site of M3.
In addition to the binding affinity and selectivity, the small
molecule-induced structural changes of the S1 G-quadruplex
were investigated using CD spectra. The S1 can form
antiparallel G-quadruplex in KCl. However, when M3 was
added to the S1 G-quadruplex, this small molecule could induce
the structural change of the G-quadruplex. Supporting
Information Figure S8c shows the titration of 0 to 40 μM
M3 in 5 μM the G-quadruplex. While increasing the
concentration of M3, the positive peak for the G-quadruplex
at 292 nm was weakened and moved to a shorter wavelength,
while the negative peak at approximately 260 nm was changed
to a positive peak to become the major peak. With 40 μM M3,
the G-quadruplex showed a strong positive peak at 262 nm, a
shoulder peak at 279 nm, and a negative peak at 240 nm,
indicating the orientation of the strands had been changed from
Biological Activity of the Benzo[c]phenanthridine
Derivatives on the HIF1α Pathway. To determine whether
the benzo[c]phenanthridine derivatives suppress the expression
of HIF1α in human cancer cells, a Q-RT-PCR screen was
performed using M1 through M7 at 15 μM. M3 showed the
best suppression with an approximate 50% decrease in HIF1α
transcription, agreeing with the ESI-MS data shown above
(Figure 7a).
To confirm whether M3 can inhibit the expression of HIF1α,
both western blots and Q-RT-PCR were applied using human
MCF-7 cell line. M3 significantly inhibited the expression of
HIF1α during hypoxia for both mRNA and protein levels
(Figure 7b,c). The IC50 of the M3 inhibition of HIF1α was
approximately 15 μM, as determined by western blot and Q-
RT-PCR. Compound M3 also inhibited the transcription of
VEGF and GluT1 but barely affected the expression of AP2,
indicating that M3 selectively inhibited the expression of
HIF1α and blocked the transcription of its downstream genes,
such as VEGF, without suppressing its upstream regulators,
such as AP2 (Figure 7).
A luciferase assay was used to determine whether M3 could
regulate HIF1α transcription in with or without the S1 G-
quadruplex. The WT and PS1 plasmids were transfected into
MCF-7 in parallel, and the transcription level was measured.
After increasing the concentration of M3 from 0 to 15 μM, the
WT transcription was almost completely inhibited, while PS1
kept more than 50% of its activity (Figure 8a). After
normalization with the untreated group, the cells transfected
with the S1 mutation plasmid exhibited approximately 8-fold
30
antiparallel to parallel (Figure 5b). KCl was needed for the
conversion because M3 cannot induce a G-quadruplex without
any positive ions to drive G-quadruplex formation. However,
high concentrations of KCl hindered the conversion because
potassium may induce an antiparallel structure and offset the
effect of M3. With the complementary strand of S1, the CD
spectrum showed a broad peak at 280 nm, indicating the
formation of double stranded DNA from S1 and S1c
(Supporting Information Figure S8a). Adding M3 to this
solution generates a shoulder peak at approximately 260 nm,
demonstrating that M3 can partially induce G-quadruplex
formation in the S1 double strand. The induced S1 conforma-
tional change was a unique property of M3; the other
benzo[c]phenanthridine derivatives barely affected the anti-
parallel G-quadruplex of the S1. The structural shift may be
important to its function during HIF1α transcription.
A DMS footprinting assay was used to identify the M3
binding site on the S1 G-quadruplex. The annealed S1fam was
incubated with 50 mM KCl and 40 μM M3 overnight before
2
588
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
Figure 7. Ligand screening by Q-RT-PCR (a) showed that M3 can suppress the transcription of HIF1α. The dose-dependent Q-RT-PCR assay (b)
and western blot (c) showed that M3 inhibited HIF1α on mRNA and the protein level along when increasing the M3 concentration from 0 to 20
μM, while decreasing the transcription of the genes downstream from HIF1α, such as VEGF (d) and GluT1 (e), without affecting the upstream AP2
protein (b). β-Actin was used as control. All Q-RT-PCR assays were conducted in triplicate on the MCF-7 cell line and normalized to an ethanol
control. Error bars represent standard deviations.
Figure 8. A luciferase assay (a) showed the mutation plasmid PS1 exhibited a stronger transcriptional activity than WT as the concentration of M3
increased from 0 to 15 μM in the MCF-7 cell line, indicating that M3 targeted the S1 G-quadruplex and suppressed the transcription of HIF1α. A
ChIP assay using an anti-AP2 antibody (b) and the PCR product relative intensity (c) demonstrated that M3 prevented AP2 from binding to the
HIF1α promoter when increasing its concentration M3 from 0 to 15 μM. The plasmid ChIP assay using an anti-AP2 antibody with a WT and a PS1
plasmid (d) showed that the AP2 binding affinity was maintained in the S1 mutation plasmid (PS1) compared to WT after treatment with 8 μM M3
(
e), indicating that M3 can bind to the S1 G-quadruplex and block AP2. WT-0, WT-8, PS1−0, and PS1−8 stand for the MCF-7 cells treated with
WT, WT followed by 8 μM M3, PS1, and PS1 followed by 8 μM M3, respectively.
resistance compared to the WT against 15 μM M3,
demonstrating that the S1 G-quadruplex might be a primary
target for M3 during cellular HIF1α transcription and
confirming the biological activity of M3. However, both the
WT and PS1 showed a significant decrease in transcription
when the concentration of the M3 exceeded 15 μM, possibly
due to the weak binding affinity of M3 for other PGSs in the
HIF1α promoter.
To determine the mechanism of inhibition for M3 on HIF1α
transcription, a ChIP assay was performed using an MCF-7 cell
line. After being normalized by the input PCR products, the
concentration of the ChIP PCR products decreased dramati-
cally as the concentration of M3 increased (Figure 8b,c). The
IC50 measured by a ChIP assay was approximately 7.5 μM,
remaining lower than the IC50 value identified via western
blotting and Q-RT-PCR. Therefore, M3 is an efficient
transcriptional inhibitor for HIF1α because it blocks the AP2
pathway.
Plasmid ChIP was also used to evaluate the inhibition of M3
on the transcription of the S1 mutation plasmid. The WT and
PS1 plasmid binding with AP2 was assessed: the PCR products
were amplified and were quantified using the AP2 abundance
determined by western blotting. Because M3 binds and
converts the conformation of the S1 G-quadruplex but not
2
589
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
Figure 9. Mechanism of HIF1α transcriptional inhibition by the S1 G-quadruplex (Q1) and benzo[c]phenanthridine derivative (M3): The S1 PGS
is overlapped with an AP2 binding site, and the formation of the S1 G-quadruplex (Q1) induced by potassium or benzo[c]phenanthridine derivative
(M3) will prevent AP2 from binding to the HIF1α promoter.
the S1 double strand, its binding affinity to the mutated plasmid
should be greatly decreased, promoting AP2 binding to the
HIF1α promoter. As shown in Figure 8d, after treatment with 8
μM M3, the AP2 binding to the WT decreased to 50%, while
the AP2 binding to PS1 retained more than 90%,
demonstrating that M3 can bind to the S1 G-quadruplex,
prevent AP2 from binding to this cis-element, and suppress
HIF1α transcription (Figure 9).
ACKNOWLEDGMENTS
■
2
This work was supported by the 973 Program (2012CB720600,
012CB720601) and the National Natural Science Foundation
of China (21372021).
REFERENCES
■
(
(
1) Melillo, G. Methods Enzymol. 2007, 435, 385.
2) Onnis, B.; Rapisarda, A.; Melillo, G. J. Cell Mol. Med. 2009, 13,
CONCLUSION
2780.
■
(
(
(
(
3) Semenza, G. L. Nat. Rev. Cancer 2003, 3, 721.
4) Semenza, G. L.; Wang, G. L. Mol. Cell. Biol. 1992, 12, 5447.
5) Harris, A. L. Nat. Rev. Cancer 2001, 2, 38.
We have evaluated the properties of four PGSs in the HIF1α
promoter and 5′UTR using multiple spectroscopic and
biochemical methods, demonstrating the formation of the
antiparallel G-quadruplexes with S1 and S2, as well as parallel
G-quadruplexes with S3 and S4. We have proven that the S1 is
an important PGS for HIF1α transcription via the formation of
a G-quadruplex and blocking the AP2 binding site with both a
luciferase assay and a plasmid ChIP analysis. In addition, a
benzo[c]phenanthridine derivative (M3) was screened out,
which can bind to the S1 G-quadruplex and induce its structural
change, forming a mixed parallel/antiparallel conformation.
This compound can primarily inhibit HIF1α transcription by
disrupting the AP2 binding site on the G-quadruplex in the S1
sequence. This study provided an important foundation for
understanding the biological roles of the HIF1α G-quadruplex,
generating potential insight for cancer therapy.
6) Wang, G. L.; Jiang, B. H.; Rue, E. A.; Semenza, G. L. Proc. Natl.
Acad. Sci. U.S.A. 1995, 92, 5510.
(7) Semenza, G. L. Curr. Opin. Cell Biol. 2001, 13, 167.
(8) Hirotaa, K.; Semenza, G. L. Crit. Rev. Oncol. Hematol 2005, 59,
15.
(9) Chen, C.; Alireza, N. P.; Behrooz; Ismail-Beigi, F.; Maity, A. J.
Biol. Chem 2001, 276, 9519.
(
(
10) Gruber, M.; Simon, M. C. Curr. Opin. Hematol. 2006, 13, 169.
11) Hirota, K.; Semenza, G. L. Crit. Rev. Oncol. Hematol. 2006, 59,
1
(
5.
12) Wenger, R. H.; Kvietikova, I.; Roles, A.; Gassmann, M.; Marti,
H. H. Kidney Int. 1997, 51, 560.
13) Iyer, N. V.; Leung, S. W.; Semenza, G. L. Genomics 1998, 52,
59.
14) Minet, E.; Michel, G.; Roland, I.; Remacle, J.; Raes, M.;
(
1
(
Michiels, C. Biochem. Biophys. Res. Commun. 1999, 261, 534.
(15) Dai, J.; Jones, R. A.; Hurley, L. H.; Yang, D. Z. Nucleic Acids Res.
ASSOCIATED CONTENT
■
2
006, 34, 5133.
16) Dexheimer, T. S.; Hurley, L. H. J. Am. Chem. Soc. 2006, 128,
404.
17) Fernando, H.; Reszka, A. P.; Huppert, J.; Ladame, S.; Rankin, S.;
Venkitaraman, A. R.; Neidle, S.; Balasubramanian, S. Biochemistry
006, 45, 7854.
(18) Phan, A. T.; Burge, S.; Neidle, S.; Patel, D. J. J. Am. Chem. Soc.
007, 129, 4386.
19) Armond, R. D.; Sun, D.; Hurley, L. H.; Ebbinghaus, S. W.
*
S
Supporting Information
(
5
(
Table of the ODN list, additional figures, synthesis of M1−M7,
and MS and NMR data of M1−M7. This material is available
free of charge via the Internet at http://pubs.acs.org.
2
AUTHOR INFORMATION
■
2
(
Corresponding Author
guyuan@pku.edu.cn
Biochemistry 2005, 44, 16341.
(20) Welsh, S. J.; Lombardo, C. M.; Valentine, H.; Fuente, M.;
Notes
The authors declare no competing financial interest.
Schatzlein, A.; Neidle, S. Sci. Rep 2013, 3, 2799.
2
590
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591

Journal of the American Chemical Society
Article
(
21) Lombardo, C. M.; Welsh, S. J.; Strauss, S. J.; Dale, A. G.; Todd,
A. K.; Nanjunda, R.; Wilson, W. D.; Neidle, S. Bioorg. Med. Chem. Lett.
012, 22, 5984.
2
(
22) Nakanishi, T.; Suzuki, M. Org. Lett. 1999, 1, 985.
(
23) Wang, F.; Zeng, Z.; Wu, R.; Xue, Y.; Hao, Y.; Pang, D.; Zhao, Y.;
Tan, Z. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 20413.
24) Kim, M. Y.; Gleason-Guzman, M.; Izbicka, E.; Nishioka, D.;
Hurley, L. H. Cancer Res. 2003, 63, 3247.
25) Han, H.; Hurley, L. H.; Salazar, M. Nucleic Acids Res. 1999, 27,
37.
26) Weitzmann, M. N.; Woodford, K. J.; Usdin, K. J. Biol. Chem.
996, 271, 20958.
27) Lin, S.; Gu, H.; Xu, M.; Cui, X.; Zhang, Y.; Gao, W.; Yuan, G.
PLoS One 2012, 7, e31201.
28) Rosu, F.; Gabelica, V.; Houssier, C.; Colson, P.; De Pauw, E.
Rapid Commun. Mass Spectrom. 2002, 16, 1729.
29) Gabelica, V.; Maeda, R.; Fujimoto, T.; Yaku, H.; Murashima, T.;
Sugimoto, N.; Miyoshi, D. Biochemistry 2013, 52, 5620.
(
(
5
(
1
(
(
(
(
30) Paramasivan, S.; Rujan, I.; Bolton, P. H. Methods 2007, 43, 324.
(
31) Busson, P.; Wu, J.; Wang, C.; Li, X.; Song, Y.; Wang, W.; Li, C.;
Hu, J.; Zhu, Z.; Li, J.; Zhang, W.; Lu, Z.; Yang, C. J. PLoS One 2012, 7,
6393.
32) Vorlickova, M.; Kejnovska, I.; Sagi, J.; Renciuk, D.; Bednarova,
K.; Motlova, J.; Kypr, J. Methods 2012, 57, 64.
33) Sperling, S.; Grimm, C. H.; Dunkel, I.; Mebus, S.; Sperling, H.
4
(
(
P.; Ebner, A.; Galli, R.; Lehrach, H.; Fusch, C.; Berger, F.; Hammer, S.
Hum. Mutat. 2005, 26, 575.
2
591
dx.doi.org/10.1021/ja412128w | J. Am. Chem. Soc. 2014, 136, 2583−2591