Available via license: CC BY 4.0
Content may be subject to copyright.
ARTICLE OPEN
miR-23a promotes invasion of glioblastoma via HOXD10-
regulated glial-mesenchymal transition
Kazuhiro Yachi
1
, Masumi Tsuda
1,2,3
, Shinji Kohsaka
1,5
, Lei Wang
1,2
, Yoshitaka Oda
1
, Satoshi Tanikawa
1
, Yusuke Ohba
4
and
Shinya Tanaka
1,2,3
Glioblastoma is the most aggressive and invasive brain tumor and has a poor prognosis; elucidating the underlying molecular
mechanisms is essential to select molecular targeted therapies. Here, we investigated the effect of microRNAs on the marked
invasiveness of glioblastoma. U373 glioblastoma cells were infected with 140 different microRNAs from an OncomiR library, and the
effects of the invasion-related microRNAs and targeted molecules were investigated after repeated Matrigel invasion assays.
Screening of the OncomiR library identified miR-23a as a key regulator of glioblastoma invasion. In six glioblastoma cell lines, a
positive correlation was detected between the expression levels of miR-23a and invasiveness. A luciferase reporter assay
demonstrated that homeobox D10 (HOXD10) was a miR-23a-target molecule, which was verified by high scores from both the
PicTar and miRanda algorithms. Forced expression of miR-23a induced expression of invasion-related molecules, including uPAR,
RhoA, and RhoC, and altered expression of glial-mesenchymal transition markers such as Snail,Slug,MMP2,MMP9,MMP14, and E-
cadherin; however, these changes in expression levels were reversed by HOXD10 overexpression. Thus, miR-23a significantly
promoted invasion of glioblastoma cells with polarized formation of focal adhesions, while exogenous HOXD10 overexpression
reversed these phenomena. Here, we identify miR-23a-regulated HOXD10 as a pivotal regulator of invasion in glioblastoma,
providing a novel mechanism for the aggressive invasiveness of this tumor and providing insight into potential therapeutic targets.
Signal Transduction and Targeted Therapy (2018) 3:33 ; https://doi.org/10.1038/s41392-018-0033-6
INTRODUCTION
Glioblastoma (GBM) is the most invasive and aggressive primary
brain tumor and has a poor prognosis, showing a 5-year survival
rate of 7%. Conventional therapy for GBM involves surgical
resection followed by fractionated radiotherapy and concomitant
adjuvant chemotherapy with alkylating drugs such as temozolo-
mide (TMZ).
1
However, the effects of treatment are limited due to
the complexity of GBM, involving tumor heterogeneity, rapid
invasion, clonal populations maintaining glioma stem cells (GSCs),
and a high frequency of recurrence. Genome-wide analyses such
as the Cancer Genome Atlas (TCGA) along with other efforts, have
identified gene mutations, amplification, modification, and
rearrangement as the principal genetic causes of GBM.
2,3
To date,
more than 140 gene mutations have been reported in GBM, most
frequently in EGFR,TP53,PTEN,PIK3CA,PIK3R1,PDGFRA,ATRX,
IDH1,RB1,LZTR1, and PTPN11, while TMZ-dependent hypermuta-
tions are highly expressed in recurrent tumors.
4–6
Although
various molecular targeted agents have been attempted to be
used either as a single agent or in combination therapy, few have
been reported to be effective in phase II trials thus far. Notably,
many gene mutations in primary tumors are distinct from those in
recurrent tumors. In addition, mutations in genes at diagnosis,
such as those in EGFR,PDGFRA, and TP53, can switch to different
mutations in the same gene at relapse,
5,7,8
suggesting that
complicated spatiotemporal clonal evolution is a primary mechan-
ism of treatment failure. Therefore, new approaches are urgently
needed to understand the unique biology of GBM and design
optimized therapies.
One unique characteristic of GBM cells is aggressive infiltration
and invasion into the surrounding normal tissues along the
vascular tracks, preventing complete resection of all malignant
cells and limiting the effect of localized radiotherapy. CD44
ligation with hyaluronic acid (HA) has been shown to trigger PI3K/
Rho GTPase signaling, leading to GBM invasion via regulation of
actin polymerization and formation of focal adhesions.
9
Accumu-
lating evidence has indicated that cancer stem cells (CSCs),
10
epithelial–mesenchymal transition (EMT) modulated by PI3K/AKT/
mTOR signaling,
11
proneural-mesenchymal shifts via NF-κB and
JAK-STAT pathways,
12
angiogenesis-invasion shifts, tumor-derived
exosomes
13
and miRNAs play pivotal roles in GBM migration and
invasion. We have previously identified Snail as the master
regulator of the irradiation-induced glial-mesenchymal transition
(GMT), resulting in promoted migration and invasion.
14
Thus, a
better understanding of the invasive biology of GBM cells is
needed to develop innovative therapies to suppress GBM
invasion.
Received: 11 June 2018 Revised: 2 September 2018 Accepted: 5 November 2018
1
Department of Cancer Pathology, Faculty of Medicine, Hokkaido University, Sapporo, Japan;
2
Global Station for Soft Matter, Global Institution for Collaborative Research and
Education, Hokkaido University, Sapporo, Japan;
3
Institute for Chemical Reaction Design and Discovery (WPI-ICReDD), Hokkaido University, Sapporo, Japan and
4
Department of
Cell Physiology, Faculty of Medicine, Hokkaido University, Sapporo, Japan
Correspondence: Shinya Tanaka (tanaka@med.hokudai.ac.jp)
5
Present address: Division of Cellular Signaling, National Cancer Center Research Institute, Tokyo, Japan
These authors contributed equally: Kazuhiro Yachi, Masumi Tsuda
www.nature.com/sigtrans
Signal Transduction and Targeted Therapy
©The Author(s) 2019
MicroRNAs (miRNAs) are small, non coding RNAs ranging from
18 to 24 nucleotides in length that negatively regulate gene
expression at the post transcriptional level, primarily through base
pairing to the 3′UTR of target mRNA.
15
Because miRNAs modulate
fundamental cell functions such as proliferation, migration,
metabolism, and apoptosis,
16
dysregulation of miRNA expression
causes diverse diseases, including cancers.
17,18
miRNAs can
function as tumor suppressor genes or oncogenes and as
potential specific cancer biomarkers.
19–21
Accumulating studies
have demonstrated the roles of miRNAs in cancer stem cell self-
renewal,
22
sensitivity to tyrosine kinase inhibitors,
23
and cancer
therapy targeted to the tumor microenvironment.
24
Several
miRNAs have been reported to contribute to the promotion of
tumor invasion and metastasis in various cancers, including miR-
10b, miR-373, and miR-520c for breast cancer;
25
miR-17 and miR-
19 for colon cancer;
26
and miR-216a for pancreatic cancer.
Recently, the significant role of miRNAs in the pathogenesis of
GBM has been increasingly elucidated. In GBM, overexpression of
miR-221, miR-10b, miR-130a, miR-125b, miR-9-2, and miR-21 has
been reported.
27
Among these miRNAs, miR-10b, which regulates
homeobox D10 (HOXD10), and miR-21, which targets RECK, are
important in facilitating glioblastoma invasion.
28,29
miR-23a has been reported to regulate several physiological
phenomena by targeting MURF1,MAFbx, and GLS, leading to
promotion of cardiac hypertrophy, inhibition of muscular atrophy,
and suppression of glutamine metabolism, respectively.
30
In
addition, dysregulated expression of miR-23a has been reported
in various types of human cancers, including upregulation in
hepatocellular carcinoma, glioblastoma, bladder cancer, and
pancreatic cancer and downregulation in acute promyelocytic
leukemia and oral squamous carcinoma.
30
When overexpressed in
cancers, miR-23a directly regulates some target genes such as
NOXA,MTSS1, and PTEN, consequently preventing apoptosis,
inducing the EMT, and promoting tumorigenesis, respectively.
31
Recently, miR-23a was shown to be encapsulated in exosomes
derived from patients with colorectal cancer,
32
raising the
possibility of its use as a diagnostic and predictive marker.
However, the pathobiological role of miR-23a in GBM has
remained obscure.
In this study, we identified miR-23a as an oncogene that confers
aggressive invasion of GBM cells by directly inhibiting HOXD10
expression. In miR-23a-overexpressing GBM cells, HOXD10 protein
levels were dramatically decreased, and mRNA levels of invasion-
and GMT-related molecules were markedly altered with polarized
formation of focal adhesions, resulting in profound tumor
invasion. These findings suggest that miR-23a and HOXD10 are
potentially powerful therapeutic targets for GBM treatment.
MATERIALS AND METHODS
Cell culture
The human GBM cell lines LN308, LN443, and U373 were kindly
provided by Dr. Erwin G. Van Meir (Emory University School of
Medicine, Atlanta, Georgia). The KMG4 cell line was kindly
provided by Dr. Kazuo Tabuchi (Saga University, Saga, Japan).
U87 cells (ATCC#HTB-14) and U251 cells (ATCC#CRL2219) were
purchased from the American Type Culture Collection (ATCC). All
cell lines, including embryonic kidney 293FT cells, were cultured in
Dulbecco’s modified Eagle’s medium (DMEM) (Wako, Osaka,
Japan) supplemented with 10% fetal bovine serum (FBS; Invitro-
gen, Carlsbad, CA, USA) and maintained in a humidified atmo-
sphere of 5% CO
2
at 37 °C.
Establishment of miR-23a and HOXD10-overexpressing cells
miR-23a-overexpressing cells were established using the BLOCK-iT
HiPerform Lentiviral Pol II miR RNAi Expression System with EmGFP
(Invitrogen). 293FT cells were transfected with pLenti6.4/Promoter/
MSGW/miR-23a and pLenti6.4/Promoter/MSGW/HOXD10 using
Fugene HD transfection reagent (Promega, Madison, WI, USA).
After 48 h of incubation, the supernatant was treated with U373
and LN443 glioblastoma cells, and miR-23a and HOXD10-
overexpressing cells were selected in DMEM containing blasticidin.
Anti-miR23a oligonucleotide transfection in KMG4 cells
Anti-miR-23a or comparable scramble oligonucleotides were
transfected into KMG4 cells using HiPerfect reagent (Qiagen,
Valencia, CA). The sequences utilized were as follows: anti-miR-
23a: 5′-AATCCCTGGCAATGTGAT-3′, scramble: 5′-GTGTAACACGTC
TATACGCCCA-3′. After 48 h, the cells were utilized for real-time
PCR of Snail,MMP2,MMP9, and MMP12 and for Matrigel invasion
assays, as described below.
Identification of microRNA that promotes glioblastoma invasion
The OncoMir Precursor Virus Library (System Bioscience, Mountain
View, CA, USA) was infected into U373 cells, and the Matrigel
invasion assay (BD Biosciences, MA, USA) was performed in
triplicate as described below. RNA was isolated from cells with
elevated invasion ability, and semi quantitative RT-PCR using the
OncoMir Precursor Library primers (System Bioscience) and
sequencing were performed to identify the infected oncomiRs.
Matrigel invasion assay
A Matrigel invasion assay was performed as described previously
33
using a BioCoat Matrigel invasion chamber (24-well chambers)
with 8-µm pores (BD Biosciences, MA). U373 and LN443 cells with
or without enforced miR-23a and HOXD10 were seeded at a
density of 5 × 10
4
cells into the upper chamber with serum-free
medium. Medium containing 10% FBS was added to the lower
chamber as a chemo attractant. After incubation for 8 or 24 h, the
cells were fixed with 3% paraformaldehyde (PFA) for 10 min and
stained with 0.2% crystal violet solution. Non invading cells on the
upper surface of each filter were removed by scrubbing. The
invaded cells were counted in microscopic fields at ×200
magnification. To minimize bias, cells in at least five randomly
selected fields per well were counted. The experiments were
performed in triplicate independently, and the mean and standard
deviation (SD) of the invading cells were analyzed.
Prediction of miR-23a-targeting molecules
To predict miR-23a-targeting molecules, PicTar (http://pictar.mdc-
berlin.de) and miRanda (http://www.micorna.org) algorithms were
used.
Luciferase reporter assay to target the HOXD10-3’UTR
The HOXD10-3′UTR was amplified from BJ/t cells, converted to
cDNA, and sequenced. The HOXD10-3′UTR was cloned into the
region downstream of the Fireflyluciferase gene in a pGL3-
promoter luciferase reporter vector (Promega), designated pGL3-
SV40-HOXD10. The luciferase reporter vector was co transfected
with a miR-23a-overexpression vector (pLenti-6.4/miR-23a) or
control vector (pLenti-6.4/nega) into U373 and LN443 cells using
Fugene HD transfection reagent (Promega). The Renilla luciferase
plasmid pCX4-Bleo-RL-Luc (Promega) was utilized as a control for
transfection efficiency. After 48 h, a dual-luciferase reporter assay
(Promega) was performed as described previously.
34
RNA extraction and gene expression analysis
Total RNA from U373 and LN443 cells with or without enforced
miR-23a and HOXD10 expression was extracted using an RNeasy
Mini kit (Qiagen), and cDNA was synthesized using Superscript
VILO (Invitrogen). For semi-quantitative RT-PCR, GoTaq Green
Master Mix was utilized, and PCR was performed at 23–33 cycles of
denaturation for 30 s at 94 °C, annealing for 30 s at 55 °C, and
extension for 30 s at 72 °C. qRT-PCR was performed using
a StepOne Real-Time PCR System (Applied Biosystems, Foster
City, CA) as described previously.
35
The primer sequences utilized
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
2
Signal Transduction and Targeted Therapy (2018) 3:33
1234567890();,:
were as follows: miR-23a: forward 5′-TGCTGGGCCGGCTGGG
GTTCCTGGGG-3′, reverse 5′-CCTGGGTCGGTTGGAAATCCCTGGC-3′;
HOXD10: forward 5′-CTCCACTGTCATGCTCCAGCTCAAC-3′, reverse
5′-CTTTCTGCCACTCTTTGCAGTGAGCC-3′;RhoA: forward 5′-CAAG
GACCAGTTCCCAGAGGTGTATG-3′, reverse 5′-CTTGACTTCTGGGGT
CCACTTTTCTGG-3; RhoC: forward 5′-GACACAGCAGGGCAGGA
AGACTATG-3′, reverse 5′-GTAGCCAAAGGCACTGATCCGGTTC-3′;
uPAR: forward 5′-GGCTTGAAGATCACCAGCCTTACCG-3′, reverse
5′-CATCCTTTGGACGCCCTTCTTCACC-3′;Snail: forward 5′-GCTGCA
GGACTCTAATCCAGA-3′, reverse 5′-ATCTCCGGAGGTGGGATG-3′;
Slug: forward 5′-TGGTTGCTTCAAGGACACAT-3′, reverse 5′-GTTGC
AGTGAGGGCAAGGAA-3′;E-cadherin: forward 5′-TCCATTTCTTGGT
CTATACGCC-3′, reverse 5′-CACCTTCAGCCATCCTGTTT-3′;MMP2:
forward 5′-ATAACCTGGATGCCGTCGT-3′, reverse 5′-AGGCACCCTT
GAAGAAGTAGC-3′;MMP9: forward 5′-GAACCAATCTCACCGACA
GG-3′, reverse 5′-GCCACCCGAGTGTAACCATA-3′;MMP14: forward
5′-CATTGGGTGTTTGATGAGGCGTCC-3′, reverse 5′-CTCAGGGATC
CCTTCCCAGACTTTG-3′; and glyceraldehyde 3-phosphate dehydro-
genase (GAPDH): forward 5′-AGCCACATCGCTCAGACAC-3′, reverse
5′-GCCCAATACGACCAAATCC-3′.
The relative expression levels of total RNA in experimental and
control samples were normalized to the GAPDH mRNA levels.
Immunoblotting and antibodies
Immunoblot analyses were carried out as described previously.
36
Briefly, cells were lysed with RIPA buffer containing 1 mM
phenylmethylsulfonyl fluoride (Sigma), 1 mM sodium orthova-
nadate (Na
3
VO
4
) and a complete protease inhibitor cocktail
(Roche) for 10 min on ice. The membrane was treated
with primary antibodies (Abs) at 4 °C overnight, followed by
incubation with secondary antibodies for 2 h. The primary
antibodies were purchased as follows: HoxD10 (E-20) was from
Santa Cruz Biotechnology (Santa Cruz, CA), phospho-ERK1/2 was
from Cell Signaling Technology (Beverly, MA), and α-tubulin was
from Sigma Aldrich. The signals were developed using
ECL reagents (GE Healthcare, Little Chalfont, UK) and were
visualized using an ImageQuant LAS4000 mini system (Fujifilm,
Tokyo, Japan).
Immunofluorescence for focal adhesions
\U373 and LN443 cells with or without forced miR-23a
and HOXD10 expression were cultured on glass-based dishes
(IWAKI, Tokyo, Japan) coated with type I-collagen and fixed in 3%
PFA in PBS for 15 min. The cells were permeabilized with 0.1%
Triton X-100 for 4 min and blocked with 1% BSA for 20 min. To
detect focal adhesions, the cells were treated with anti-paxillin Ab
(BD Transduction Laboratories, USA) overnight at 4°C, followed by
incubation with AlexaFluor488-conjugated anti-mouse IgG (Invi-
trogen) for 1 h at room temperature (RT). F-actin was stained with
AlexaFluor594-conjugated phalloidin (Invitrogen) for 30 min at 37°
C. Fluorescent images were obtained using a confocal laser
scanning microscope (Olympus, Tokyo, Japan).
Fig. 1 MiR-23a promotes invasion of GBM cells. aThe Matrigel invasion assays were performed using six human GBM cell lines: LN443, U373,
LN308, U87, U251, and KMG4. *P< 0.05, **P< 0.005, and ***P< 0.0005 vs. LN443. bSchematic diagram identifying microRNAs to confer
marked invasion of GBM cells. U373 cells were infected with the OncoMir Precursor Virus Library, and the Matrigel invasion assay was repeated
three times to enrich the cells that acquired elevated invasion ability. Total RNA was isolated from the cells and subjected to semi quantitative
RT-PCR using OncoMir Precursor Library primers, followed by sequencing. cEndogenous expression levels of miR-23a in the six GBM cell lines
were examined by semi quantitative RT-PCR. GAPDH was utilized as an internal control. dIn the six GBM cell lines shown in c, the correlations
between the invasion ability and expression levels of miR-23a were analyzed. R
2
=0.95741. eThe scores for HOXD10,Sprouty2, and Marcks from
the PicTar and miRanda algorithms are shown
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
3
Signal Transduction and Targeted Therapy (2018) 3:33
Proliferation assay
U373 and LN443 cells with or without forced miR-23a and
HOXD10 expression were seeded into 35-mm dishes at a density
of 2 × 10
4
cells per dish. The medium was changed every 48 h, and
the numbers of cells were counted using a cell counter 5 days
after cell inoculation.
Survival analysis of glioma patients
The relationship between HOXD10 expression and survival in
glioma patients was analyzed using the public database
PrognoScan (http://www.abren.net/PrognoScan/).
Statistical analysis
All data were represented as the means and SD of experiments
performed in triplicate and subjected to one-way analysis of
variance, followed by comparison with Student’st-tests. Pvalues
less than 0.05 were considered statistically significant, as
described in the Figure legends.
RESULTS
Expression levels of miR-23a are correlated with GBM invasion
To evaluate the invasion potential of human GBM, we performed
Matrigel invasion assays using six human GBM cell lines. The cells
were divided into two groups according to their invasion
capabilities: low invasion (LN443, U373, and LN308 cells) and
high invasion (U87, U251, and KMG4 cells) (Fig. 1a). To identify
microRNAs that confer aggressive invasion in GBM cells, an
OncoMir Precursor Virus Library including 140 cancer-related
oncomiRs was infected into U373 cells with low intrinsic inva-
sion capabilities. The Matrigel invasion assay was repeated three
times to select for the cells that acquired high invasion by
oncomiRs (Fig. 1b). Total RNA was isolated, and semi quantitative
RT-PCR, followed by sequencing identified miR-23a from the cells
that ultimately acquired high invasion (Fig. 1b). miR-23a was
expressed in all human GBM cell lines tested with various degrees
(Fig. 1c), and the expression levels were significantly correlated
with invasion capability (Fig. 1d, R
2
=0.95741), suggesting the
pivotal role of miR-23a in GBM invasion. MiR-181b was also
identified in the same context in LN443 cells; however, extrinsic
overexpression of miR-181b did not promote invasion in LN443
cells (data not shown).
The PicTar algorithm nominated 472 target genes of miR-23a
(data not shown). Among them, three genes, HOXD10 (11/472),
Sprouty2 (73/472), and Marcks (332/472), were implicated in tumor
invasion. The distinct algorithm miRanda also identified these genes
as miR-23a target genes with low values of mirSVR values (Fig. 1e).
The prediction scores from the PicTar and miRanda algorithms for
miR-23a targeting HOXD10 were 7.52 and −0.4376, respectively,
nominating HOXD10 as a reliable target of miR-23a (Fig. 1e).
Fig. 2 MiR-23a directly targets the HOXD10-3’UTR in GBM cells. aU373 and LN443 cells were infected with miR-23a-producing lentivirus or
control lentivirus, and the expression levels of miR-23a and HOXD10 mRNA were examined by semi quantitative RT-PCR. GAPDH was utilized as
an internal control. bThe expression levels of HOXD10 protein were examined by immunoblotting in U373 and LN443 cells with or without
forced expression of miR-23a. α-Tubulin was used as a loading control. cDiagram of the luciferase reporter vector fused to the 3’UTR of
HOXD10 utilized in the luciferase assay. The sequences of miR-23a and the targeted HOXD10-3’UTR are shown. dDual luciferase assay. HOXD10-
3′UTR luciferase activity were measured in miR-23a-overexpressing U373 and LN443 cells. *P< 0.01 and **P< 0.001 vs. without miR-23a.
eThe expression levels of sprouty2 mRNA in miR-23a-overexpressing U373 and LN443 cells were examined by semi quantitative RT-PCR. fThe
phosphorylation levels of ERK were investigated by immunoblotting in the indicated cells. gThe cell proliferation of U373 and LN443 cells
with or without forced miR-23a was investigated and graphed as the means ± SD. * P<0.05 vs. control cells
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
4
Signal Transduction and Targeted Therapy (2018) 3:33
MiR-23a directly targets HOXD10 in GBM cells via translational
regulation
To determine the miR-23a target genes involved in GBM invasion,
we established stably miR-23a-overexpressing U373 and LN443
cells by infecting them with miR-23a-producing lentivirus.
Although the expression levels of HOXD10 mRNA were invariant
irrespective of miR-23a overexpression (Fig. 2a), the protein levels
were significantly decreased by forced miR-23a expression in both
cell lines (Fig. 2b), indicating miR-23a-dependent post transcrip-
tional degradation of HOXD10. To analyze whether miR-23a
directly targets the HOXD10-3′UTR in GBM cells, we developed a
luciferase reporter vector fused to the 3′UTR of HOXD10 (Fig. 2c).
In cells stably overexpressing miR-23a, the luciferase activity of
HOXD10-3′UTR was reduced compared with that in control cells
(Fig. 2d), confirming HOXD10 as a direct target of miR-23a. Lower
expression of HOXD10 was associated with a shorter survival rate
in glioma patients by Kaplan–Meier analysis using the PrognoScan
database (data not shown).
Sprouty2 has been reported to suppress invasion of GBM cells
by inhibiting Ras GTPase,
37
raising the possibility that miR-23a
might activate the Ras/ERK signaling pathway via translational
inhibition of Sprouty2, facilitating GBM invasion. However, forced
expression of miR-23a had no effect on the expression of sprouty2
mRNA or the phosphorylation levels of ERK1/2 (Fig. 2e, f). In
addition, miR-23a overexpression exhibited a distinct effect on the
proliferation of U373 and LN443 cells (Fig. 2f), suggesting a
relatively low possibility of Sprouty2 as a universal target of miR-
23a.
MiR-23a regulates the expression levels of invasion- and glial-
mesenchymal transition (GMT)-related genes via HOXD10
HOXD10 is a member of the Homeobox (Hox) superfamily and has
been shown to suppress invasion of GBM by inhibiting the
expression levels of urokinase-type plasminogen activator receptor
(uPAR),matrix metalloproteinase (MMP) 14, and RhoC.
38
Therefore,
we next examined the effect of miR-23a on the expression of
these genes.
MiR-23a overexpression markedly increased uPAR and RhoC
expression in both U373 and LN443 cells and MMP14 and RhoA
expression in LN443 cells (Fig. 3a, b). We found no miR-23a-
dependent elevation in MMP14 expression in U373 cells, probably
due to substantial endogenous expression levels (Fig. 3b). Notably,
Fig. 3 MiR-23a regulates expression of invasion- and glial-mesenchymal transition (GMT)-related genes via HOXD10. aThe expression levels
of uPAR, RhoA, and RhoC mRNAs were examined in control and miR-23a-overexpressing U373 and LN443 cells by semi quantitative RT-PCR.
GAPDH was utilized as an internal control. b,cIn U373 and LN443 cells with or without miR-23a overexpression, the mRNA expression levels of
the indicated GMT-related genes were investigated by semi quantitative RT-PCR (b) and real-time RT-PCR (c). *P< 0.05 and **P< 0.005 vs.
without miR-23a. d,eThe expression levels of miR-23a and the indicated molecules were examined by semi quantitative RT-PCR (d) and
immunoblotting (e) in LN443 cells with or without miR-23a and HOXD10 overexpression
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
5
Signal Transduction and Targeted Therapy (2018) 3:33
forced expression of miR-23a triggered marked alterations in the
expression levels of glial-mesenchymal transition (GMT)-related
genes, as previously reported,
14
with increased expression of Snail,
Slug,MMP2,MMP9, and MMP14 and decreased expression of E-
Cadherin, especially in LN443 cells (Fig. 3b, c); these genes are
perceived as so-called epithelial-mesenchymal transition (EMT)-
related genes in other cancers. These alterations were completely
reversed by HOXD10 overexpression (Fig. 3d, e), demonstrating
the significant contribution of the miR-23a-HOXD10 axis in these
gene expression levels.
MiR-23a produces mesenchymal morphology with polarized focal
adhesions
Because miR-23a significantly alters invasion- and GMT-related
gene expression, we further investigated morphological changes
with or without miR-23a overexpression (Fig. 4a). Forced expres-
sion of miR-23a induced elongation of spindle morphology in
U373 and LN443 cells, and HOXD10 overexpression reversed these
alterations, returning the morphology to that of control cells
(Fig. 4b). Immunofluorescence analysis revealed that the number
of focal adhesions represented by paxillin was decreased upon
miR-23a overexpression, with shortened actin filaments (Fig. 4c). In
addition, extrinsic overexpression of HOXD10 recovered the
paxillin count to 80% of that in control cells (Fig. 4c). For a more
detailed analysis regarding the assembly of focal adhesions, focal
adhesion polarity was investigated with or without miR-23a
overexpression. The cell was divided into three regions by angles
of 120°, and the “A”region was configured as the movement
direction based on cell morphology and the arrangement of actin
filaments. In U373 and LN443 cells control cells, paxillin was
Fig. 4 MiR-23a produces mesenchymal changes in cell morphology and affects the polarity of focal adhesions. aThe expression levels of miR-
23a, HOXD10 mRNA, and HOXD10 protein were examined by semi quantitative RT-PCR (upper three panels) and immunoblotting (lower two
panels) in U373 and LN443 cells. GAPDH and α-tubulin were utilized as internal controls for semi quantitative RT-PCR and immunoblotting,
respectively. bPhotomicrographs of U373 and LN443 cells with or without miR-23a or HOXD10-overexpression are displayed. The scale bars
indicate 100 µm. cImmunofluorescence of focal adhesions. (Left panels) U373 and LN443 cells with or without forced miR-23a or HOXD10
expression were subjected to immunofluorescence analysis for paxillin (green) and actin (red). (Right panels) The paxillin counts in the
indicated cells are shown. dCells stained with anti-paxillin Ab were divided into three regions by angles of 120°, and the “A”region was set as
the movement direction based on cell morphology and the structures of actin filaments. In U373 and LN443 cells with or without forced miR-
23a or HOXD10 expression, the paxillin counts were determined and are shown
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
6
Signal Transduction and Targeted Therapy (2018) 3:33
localized equivalently in all regions (Fig. 4d). However, miR-23a
overexpression produced polarity in paxillin distribution, causing
significant reductions in the rear regions of cells, namely, the B
and C regions (Fig. 4d). Notably, overexpression of HOXD10
abolished the polarity of the focal adhesions (Fig. 4d).
miR-23a promotes GBM tumor invasion via reduced HOXD10
To assess the effectiveness of miR-23a in GBM cell migration and
invasion, we performed a wound-healing assay and a Matrigel
invasion assay using U373 and LN443 cells with or without
extrinsic miR-23a expression. In the wound-healing assay, the
effect of miR-23a on cell motility alone differed across cell types
(Fig. 5a, b). However, miR-23a overexpression strikingly promoted
invasion in both U373 and LN443 cells to levels 4.0-fold and
5.0-fold higher than those in control cells, respectively (Fig. 5c, d).
These increases were reversed by extrinsic expression of HOXD10
(Fig. 5c, d). U87 cells with intrinsic high expression of miR-23a
natively possessed higher invasion potential (Fig. 1a, c). Enhanced
expression of HOXD10 triggered dramatic alterations in morphol-
ogies, changing cells from having a distinct piled-up spindle shape
to having a flat shape with an enlarged cytoplasmic compartment
and resulting in a substantial decline in invasiveness (Fig. 5c). Anti-
miR-23a oligonucleotide treatment to KMG4 cells, which had the
highest miR-23a expression and invasiveness, significantly
decreased the expression levels of Snail,MMP2,MMP9, and
MMP14 (Fig. 6a), resulting in marked suppression of invasion
(Fig. 6b).
DISCUSSION
GBM is an extremely aggressive tumor with a 5-year survival rate
of 7% due to high invasion into surrounding normal brain tissue;
elucidation of the underlying molecular mechanisms is therefore
essential to develop effective therapies and improve prognoses. In
this study, we addressed a part of this long-standing issue using
the OncoMir library infection system and found that miR-23a
directly targets the HOXD10-3′UTR and promotes tumor cell
invasion by elevating the expression levels of invasion- and glial-
mesenchymal transition (GMT)-related genes and inducing
polarity of focal adhesions in GBM (Fig. 6c).
In this analysis, two GBM cell lines, LN443 and U373, with
intrinsically low invasion potential were infected with the OncoMir
library. After repeating the Matrigel invasion assay, we identified
miR-23a as the microRNA responsible for the conferring on U373
cells aggressive invasion capabilities (Fig. 1b). In addition, miR-
181b was also identified in the same context in LN443 cells.
However, extrinsic overexpression of miR-181b did not promote
invasion in LN443 cells (data not shown), in accordance with a
Fig. 5 MiR-23a promotes tumor invasion of glioblastoma via reduced HOXD10. aWound-healing assays were performed with miR-23a-
overexpressing U373 and LN443 cells and their respective control cells. Representative photomicrographs at 0 and 24 h are shown. bThe
distances moved are displayed as the mean ± SD. N.S. indicates not statistically significant. cMatrigel invasion assays were performed with
both U373 and LN443 cells with or without forced miR-23a or HOXD10 expression. Micrographs of invading cells stained with crystal violet are
displayed. dIn the Matrigel invasion assays, the invaded cells under the filter were counted in three randomly selected regions, and graphed
as the mean ± SD. e(Left) Micrographs of U87 cells with or without forced HOXD10 expression are displayed. (Right) In the Matrigel invasion
assays, invaded cells were counted, and the data are presented as the mean ± SD. * P< 0.01 vs. control cells
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
7
Signal Transduction and Targeted Therapy (2018) 3:33
previous report showing miR-181b-suppressed invasion in GBM.
39
Therefore, we focused on the role of miR-23a. cAMP response
element-binding protein 1 (CREB1) directly binds to the promoter
of miR-23a to promote its expression, while STAT3 indirectly
induces miR-23a expression.
40
Given that both CREB1 and STAT3
are up regulated in glioma,
40,41
transcription factor-dependent
upregulation of miR-23a might contribute to the aggressive
invasion of GBM.
Hox superfamily genes, including HOXD10, encode transcrip-
tional factors regulating cell differentiation and morphogenesis
during development.
42
Dysregulation of the Hox gene disrupts
various signaling pathways related to tumorigenesis and metas-
tasis.
43
A positive correlation exists between reductions in HOXD10
mRNA and increased malignancy of breast cancer and endome-
trial adenocarcinoma.
44
Forced expression of HOXD10 mRNA
strikingly suppresses tumor motility and invasion in breast
cancer,
45
suggesting a potent inhibitory role of HOXD10 in tumor
invasion. In GBM, miR-10b has been previously reported to inhibit
invasiveness by targeting HOXD10 and regulating the transcription
of MMP14 and uPAR.
28
Previously, miR-23a has been reported to
regulate expression of HOXD10 in glioma cells.
46
In this study, we
also identified miR-23a as a novel direct regulator of HOXD10 via
translational but not transcriptional regulation (Fig. 2a, b). Forced
expression of miR-23a increased the expression levels of uPAR,
RhoC,Snail, and Slug but decreased those of E-cadherin in a
HOXD10-dependent manner (Fig. 3d, e). uPAR,MMP14,RhoC, and
Fig. 6 Mechanisms of miR-23a-regulated promotion of GBM invasion through targeting of HOXD10. a,bKMG4 cells were transfected with
anti-miR-23a and scramble DNA as a control, and the expression levels of Snail,MMP2,MMP9, and MMP14 (a) and invasion ability (b) were
investigated. *P< 0.05, **P< 0.005, and ***P< 0.0005 vs. the indicated samples. cmiR-23a directly targets the HOXD10-3′-UTR, triggering
dramatic alterations in the expression of genes associated with invasion (uPAR,MMP14,RhoA, and RhoC) and glial-mesenchymal transition
(GMT) events (Snail,Slug,MMP2,MMP9, and E-cadherin), and inducing polarity of focal adhesions, ultimately resulting in cooperatively
aggressive invasion of GBM
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
8
Signal Transduction and Targeted Therapy (2018) 3:33
RhoA have been reported to be targets of miR-23a.
38
MMP14 seems to regulate the expression levels of MMP2 and
MMP9 in inflammatory breast cancer.
47
Notably, our findings
suggest that the miR-23a-HOXD10 axis is a novel regulator of the
expression of Snail,Slug, and E-cadherin, which are GMT-regulated
genes in GBM. Taken together, the results suggest that miR-23a-
triggered consecutive gene expression might evoke aggressive
invasion of GBM cells with extensive vascularization through
degradation of extracellular matrices.
MiR-23a overexpression induced polarity in the distribution of
focal adhesions as shown by paxillin, which was completely
reversed by additional forced expression of HOXD10 (Fig. 4c, d).
RhoA has been shown to reduce interactions between focal
adhesion proteins by partly inhibiting the phosphorylation of
paxillin, causing reassembly of the actin cytoskeleton, which leads
to invasion and migration of melanoma and breast cancer.
48
In
addition, RhoC induces colocalization of focal adhesion compo-
nents such as paxillin, paxillin kinase linker (PKL), and FAK,
resulting in promoted invasion in prostate cancer cells.
49
Based on
this evidence, spatiotemporal coordination of RhoA and RhoC
might produce miR-23a-induced polarity of focal adhesions, one
of pivotal events promoting the high invasiveness of GBM.
Although kinase-targeted therapy seems to be a promising
therapeutic approach in GBM, such therapies have been
ineffective in the clinical setting due to the complexity of GBM
with regards to spatiotemporal clonal evolution. Therefore, a
kinase-irrelevant strategy using anti-miRNAs might be an innova-
tive and effective approach to target numerous genes. A distinct
effect of miR-23a on cell growth was observed in LN443 and U373
cells (Fig. 2g), likely because miR-23a regulates different targets;
the PicTar algorithm predicted 472 genes as direct targets of miR-
23a. Because the effect of miR-23a on cell growth remains
controversial and is dependent on the cellular context, special
attention should be paid to assessing the subset of miR-23a-
targeted genes that determines the pathophysiological properties
of GBM cells.
Our studies demonstrated that glioblastoma cells acquire
prominent invasive potential via a miR-23a-HOXD10-GMT-related
pathway. Forced expression of miR-23a promotes invasion by
directly targeting the HOXD10-3′UTR. Upregulation of the HOXD10
protein by miR-23a depletion might be an effective approach to
suppress invasion of human GBM.
ACKNOWLEDGEMENTS
This work was supported, in part, by Grants-in-Aid from the Ministry of Education,
Culture, Sports, Science, and Technology; Japanese Society for the Promotion of
Science; and Ministry of Health, Labor, and Welfare of Japan as well as a grant from
the Japanese Science and Technology Agency. In addition, this research was
supported by Global Station for Soft Matter, a project of Global Institution for
Collaborative Research and Education at Hokkaido University. Institute for Chemical
Reaction Design and Discovery (ICReDD) was established by World Premier
International Research Initiative (WPI), MEXT, Japan.
ADDITIONAL INFORMATION
Competing interests: The authors declare no competing interests.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims
in published maps and institutional affiliations.
REFERENCES
1. Stupp, R. et al. Radiotherapy plus concomitant and adjuvant temozolomide for
glioblastoma. N. Engl. J. Med. 352, 987–996 (2005).
2. Cancer Genome Atlas Research Network. Comprehensive genomic characteriza-
tion defines human glioblastoma genes and core pathways. Nature 455,
1061–1068 (2008).
3. Singh, D. et al. Transforming fusions of FGFR and TACC genes in human glio-
blastoma. Science 337, 1231–1235 (2012).
4. Gebler C., et al. Inactivation of cancer mutations utilizing CRISPR/Cas9. J. Natl
Cancer Inst.109, djw183 (2016).
5. Wang, J. et al. Clonal evolution of glioblastoma under therapy. Nat. Genet. 48,
768–776 (2016).
6. Flavahan, W. A. et al. Insulator dysfunction and oncogene activation in IDH
mutant gliomas. Nature 529, 110–114 (2016).
7. Frattini, V. et al. The integrated landscape of driver genomic alterations in glio-
blastoma. Nat. Genet. 45, 1141–1149 (2013).
8. Lee, J. K. et al. Spatiotemporal genomic architecture informs precision oncology
in glioblastoma. Nat. Genet. 49, 594–599 (2017).
9. Ariza, A. et al. Role of CD44 in the invasiveness of glioblastoma multiforme and
the noninvasiveness of meningioma: an immunohistochemistry study. Hum.
Pathol. 26, 1144–1147 (1995).
10. Brabletz, T., Jung, A., Spaderna, S., Hlubek, F. & Kirchner, T. Opinion: migrating
cancer stem cells - an integrated concept of malignant tumour progression. Nat.
Rev. Cancer 5, 744–749 (2005).
11. Nieto, M. A., Huang, R. Y., Jackson, R. A. & Thiery, J. P. EMT: 2016. Cell 166,21–45
(2016).
12. Bhat, K. P. L. et al. Mesenchymal differentiation mediated by NF-κB promotes
radiation resistance in glioblastoma. Cancer Cell. 24, 331–346 (2013).
13. Skog, J. et al. Glioblastoma microvesicles transport RNA and proteins that pro-
mote tumour growth and provide diagnostic biomarkers. Nat. Cell Biol. 10,
1470–1476 (2008).
14. Mahabir, R. et al. Sustained elevation of Snail promotes glial-mesenchymal
transition after irradiation in malignant glioma. Neuro. Oncol. 16, 671–685 (2014).
15. Bartel, D. P. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116,
281–297 (2004).
16. Bartel, D. P. MicroRNAs: target recognition and regulatory functions. Cell 136,
215–233 (2009).
17. Lu, J. et al. MicroRNA expression profiles classify human cancers. Nature 435,
834–838 (2005).
18. Peng, Y. & Croce, C. M. The role of MicroRNAs in human cancer. Signal Transduct.
Target Ther. 1, 15004 (2016).
19. Esquela-Kerscher, A. & Slack, F. J. Oncomirs - microRNAs with a role in cancer. Nat.
Rev. Cancer 6, 259–269 (2006).
20. Bracken, C. P., Scott, H. S. & Goodall, G. J. A network-biology perspective of
microRNA function and dysfunction in cancer. Nat. Rev. Genet. 17, 719–732
(2016).
21. Mitchell, P. S. et al. Circulating microRNAs as stable blood-based markers for
cancer detection. Proc. Natl. Acad. Sci. USA 105, 10513–10518 (2008).
22. Hwang, W. L. et al. MicroRNA-146a directs the symmetric division of Snail-
dominant colorectal cancer stem cells. Nat. Cell Biol. 16, 268–280 (2014).
23. Garofalo, M. et al. EGFR and MET receptor tyrosine kinase-altered microRNA
expression induces tumorigenesis and gefitinib resistance in lung cancers. Nat.
Med. 18,74–82 (2011).
24. Cheng, C. J. et al. MicroRNA silencing for cancer therapy targeted to the tumour
microenvironment. Nature 518, 107–110 (2015).
25. Huang, Q. et al. The microRNAs miR-373 and miR-520c promote tumour invasion
and metastasis. Nat. Cell Biol. 10, 202–210 (2008).
26. Zhang, J. et al. miR-21, miR-17 and miR-19a induced by phosphatase of regen-
erating liver-3 promote the proliferation and metastasis of colon cancer. Br. J.
Cancer 107, 352–359 (2012).
27. Negrini, M., Nicoloso , M. S. & Calin, G. A. MicroRNAs and cancer--new paradigms
in molecular oncology. Curr. Opin. Cell Biol. 21, 470–479 (2009).
28. Dong, C. G. et al. Co-inhibition of microRNA-10b and microRNA-21 exerts
synergistic inhibition on the proliferation and invasion of human glioma cells. Int.
J. Oncol. 41, 1005–1012 (2012).
29. Han, L. et al. MicroRNA-21 expression is regulated by ®-catenin/STAT3 pathway
and promotes glioma cell invasion by direct targeting RECK. Cns. Neurosci. Ther.
18, 573–583 (2012).
30. Chhabra, R., Dubey, R. & Saini, N. Cooperative and individualistic functions of the
microRNAs in the miR-23a~27a~24-2 cluster and its implication in human dis-
eases. Mol. Cancer 9, 232 (2010).
31. Roufayel, R. & Kadry, S. Expression of miR-23a by apoptotic regulators in human
cancer: A review. Cancer Biol. Ther. 18, 269–276 (2017).
32. Hosseini, M. et al. Exosome-encapsu lated microRNAs as potential circulating
biomarkers in colon cancer. Curr. Pharm. Des. 23, 1705–1709 (2017).
33. Matsumoto, R. et al. Aldo-keto reductase 1C1 induced by interleukin-1 mediates
the invasive potential and drug resistance of metastatic bladder cancer cells. Sci.
Rep. 6, 34625 (2016).
34. Minami, Y. et al. SS18-SSX-regulated miR-17 promotes tumor growth of synovial
sarcoma by inhibiting p21WAF1/CIP1. Cancer Sci. 105, 1152–1159 (2014).
35. Matsumoto, R. et al. Adaptor protein CRK induces epithelial -mesenchymal tran-
sition and metastasis of bladder cancer cells through HGF/c-Met feedback loop.
Cancer Sci. 106, 709–717 (2015).
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
9
Signal Transduction and Targeted Therapy (2018) 3:33
36. Linghu, H. et al. Involvement of adaptor protein Crk in malignant feature of
human ovarian cancer cell line MCAS. Oncogene 25, 3547–3556 (2006).
37. Kwak, H. J. et al. Downregulation of Spry2 by miR-21 triggers malignancy in
human gliomas. Oncogene 30, 2433–2442 (2011).
38. Sun, L. et al. MicroRNA-10b induces glioma cell invasion by modulating MMP-14
and uPAR expression via HOXD10. Brain Res. 1389,9–18 (2011).
39. Liu, Y. S. et al. MiR-181b modulates EGFR-dependent VCAM-1 expression and
monocyte adhesion in glioblastoma. Oncogene 36, 5006–5022 (2017).
40. Tan, X. et al. cAMP response element-binding protein promotes gliomagenesis by
modulating the expression of oncogenic microRNA-23a. Proc. Natl Acad. Sci. USA
109, 15805–15810 (2012).
41. Kohsaka, S. et al. STAT3 inhibition overcomes temozolomide resistance in glio-
blastoma by downregulating MGMT expression. Mol. Cancer Ther. 11, 1289–1299
(2012).
42. Botas, J. Control of morphogenesis and differentiation by HOM/Hox genes. Curr.
Opin. Cell Biol. 5, 1015–1022 (1993).
43. Shah, N. & Sukumar, S. The Hox genes and their roles in oncogenesis. Nat. Rev.
Cancer 10, 361–371 (2010).
44. Osborne, J. et al. Expression of HOXD10 gene in normal endometrium and
endometrial adenocarcinoma. J. Soc. Gynecol. Investig. 5, 277–280 (1998).
45. Carrio, M., Arderiu, G., Myers, C. & Boudreau, N. J. Homeobox D10 induces phe-
notypic reversion of breast tumor cells in a three-dimensional culture model.
Cancer Res. 65, 7177–7185 (2005).
46. Hu, X., Chen, D., Cui, Y., Li, Z. & Huang, J. Targeting microRNA-23a to inhibit
glioma cell invasion via HOXD10. Sci. Rep. 3, 3423 (2013).
47. Al-Raawi, D., Abu-El-Zahab, H., El-Shinawi, M. & Mohamed, M. M. Membrane type-
1 matrix metalloproteinase (MT1-MMP) correlates with the expression and acti-
vation of matrix metalloproteinase-2 (MMP-2) in inflammatory breast cancer. Int.
J. Clin. Exp. Med. 4, 265–275 (2011).
48. Coló, G. P. et al. Focal adhesion disassembly is regulated by a RIAM to MEK-1
pathway. J. Cell. Sci. 125, 5338–5352 (2012).
49. Yao, H., Dashner, E. J., van Golen, C. M. & van Golen, K. L. RhoC GTPase is required
for PC-3 prostate cancer cell invasion but not motility. Oncogene 25, 2285–2296
(2006).
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this license, visit http://creativecommons.
org/licenses/by/4.0/.
© The Author(s) 2019
miR-23a promotes invasion of glioblastoma via HOXD10-regulated. . .
Yachi et al.
10
Signal Transduction and Targeted Therapy (2018) 3:33
