Published OnlineFirst February 12, 2014.Clin Cancer Res Yu Shi, Cong Chen, Xia Zhang, et al. glioblastomatargeting PIK3CD and predicts the prognosis of human Primate-specific miR-663 functions as a tumor suppressor by
Updated version
10.1158/1078-0432.CCR-13-2284doi:
Access the most recent version of this article at:
Material
Supplementary
http://clincancerres.aacrjournals.org/content/suppl/2014/02/12/1078-0432.CCR-13-2284.DC1.html
Access the most recent supplemental material at:
Manuscript
Authoredited. Author manuscripts have been peer reviewed and accepted for publication but have not yet been
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
Subscriptions
Reprints and
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
To request permission to re-use all or part of this article, contact the AACR Publications
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
1
Primate specific miR-663 functions as a tumor suppressor by targeting PIK3CD 1
and predicts the prognosis of human glioblastoma 2
3
Yu Shi 1, 2, Cong Chen 1, 2, Xia Zhang 1, 2, Qing Liu 1, 2, Jin-ling Xu 1, 3, Hua-rong 4
Zhang 1, 2, Xiao-hong Yao 1, 2, Tao Jiang 4, Zhi-cheng He 1, 2, Yong Ren 1, 2, Wei Cui 1, 5
2, Chuan Xu 1, 2, Ling Liu 5, You-hong Cui 1, 2, Shi-zhu Yu 3, Yi-fang Ping 1, 2, *, 6
Xiu-wu Bian 1, 2, * 7
8
Author Affiliations 9
1 Institute of Pathology and Southwest Cancer Center, Southwest Hospital, Third 10
Military Medical University, Chongqing 400038, China; 11
2 Key Laboratory of Tumor Immunopathology of Ministry of Education of China, 12
Third Military Medical University, Chongqing 400038, China; 13
3 Department of Neuropathology, Tianjin Neurological Institute, Tianjin Medical 14
University General Hospital, Tianjin 300052, China; 15
4 Department of Neurosurgery, Tiantan Hospital, Capital Medical University, Beijing 16
100050, China; 17
5 Department of Health Statistics, College of Preventive Medicine, Third Military 18
Medical University, Chongqing 400038, China 19
20
Running title: miR-663 favors glioblastoma prognosis 21
22
Key words: miR-663, glioblastoma, prognosis, subclassification, PIK3CD. 23
24
Financial Support 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
2
This project was supported by grants from the National Basic Research Program of 1
China (973 Program, No. 2010CB529400), the National Natural Science Foundation 2
of China (No. 30930103 and 30900564) and Programs of Science and Technology 3
Commission Foundation of Tianjin Municipal (10SYSYJC28800, 12ZCDZSY17400). 4
5
*Corresponding authors: 6
Xiu-wu Bian and Yi-fang Ping, Institute of Pathology and Southwest Cancer Center, 7
Southwest Hospital, Third Military Medical University, Chongqing 400038, China. 8
Phone: +862368754431; Facsimile: +862368754431; E-mail addresses: 9
[email protected] and [email protected] 10
11
Conflicts of interest: None 12
13
Word count (excluding references): abstract, 250 words; text, 4572 words. 14
Total number of figures and tables: 5 figures and 1 table. 15
16
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
3
Translational relevance 1
Glioblastoma (GBM) is heterogeneous in morphology and varies in outcome. The 2
WHO classification of GBM based on histopathological characteristics is limited in 3
predicting the prognosis of GBM patients. Our present results reveal that miR-663 4
predicts better prognosis of GBM and suppresses proliferation and invasion of GBM 5
cells by directly targeting PIK3CD. This is the first report that identifies miR-663 as a 6
novel prognostic biomarker and a tumor suppressor of GBM. Thus, our study 7
provides new insights into the diagnostic and prognostic determination as well as 8
therapeutic intervention for GBM based on the functions of miR-663.9
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
4
Abstract 1
Purpose: To determine the prognostic significance of miR-663 in glioblastoma 2
(GBM), its effect in tumor progression and the underlying mechanism. 3
Experimental Design: Specimens from 256 cases of glioma patients, including 239 4
patients with follow-up information, were used to analyze the association between 5
miR-663 and patients’ prognosis by Kaplan-Meier and multivariate Cox regression 6
analyses. The effects of miR-663 on GBM cell proliferation and invasion were 7
examined both in vitro and in vivo. Bioinformatics prediction and signal network 8
analysis were applied to identify the putative targets of miR-663, which were further 9
verified by luciferase reporter assay, rescue experiments as well as the 10
immunohistochemistry (IHC) and Western blotting examination of downstream 11
effectors. Quantitative reverse transcriptase-PCR (qRT-PCR) and IHC were applied to 12
investigate the clinical association between miR-663 and its target in human GBM 13
specimens. 14
Results: miR-663 was inversely correlated with glioma grades but positively 15
correlated with patients’ survival. Furthermore, two distinct subgroups of GBM 16
patients with different prognoses were identified based on miR-663 expression in our 17
specimens and that from TCGA database. Overexpression of miR-663 significantly 18
suppressed the proliferation and invasion of GBM cells in vitro and in vivo. 19
Mechanistically, we discovered PIK3CD as a direct target of miR-663 and found that 20
phosphorylated AKT and three key downstream effectors of PIK3CD, i.e., CCND1, 21
MMP2 and MMP7, were down-regulated by miR-663 overexpression. Moreover, 22
PIK3CD was inversely correlated with miR-663 in GBM specimens and predicted 23
poor prognosis of GBM patients. 24
Conclusion: miR-663 is a novel prognostic biomarker and a potential therapeutic 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
5
candidate for GBM. 1
2
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
6
Introduction 1
Glioblastoma (GBM) is the most aggressive and lethal malignant brain tumor. Despite 2
efforts have been made to improve therapeutic strategies, the average survival of 3
GBM patients have been improved only slightly in the past decades (1). Although in 4
general, patients with GBM tend to have poor prognoses, significant intra-group 5
variations in their survival have been observed (2). Additionally, traditional 6
histological criteria for GBM diagnosis are not suitable for comprehensive assessment 7
of the status of patients, especially their survival (3). The ineffectiveness to predict 8
patients’ outcomes based on histopathological features suggests the requirement for 9
more precise criteria in GBM subclassification. Recent studies by gene expression 10
profiling and factor analyses have identified subtypes of GBM and a few biomarkers 11
associated with GBM patients’ prognoses (3-5). However, the clinicopathological 12
heterogeneities and prognostic signatures of GBM have not been fully characterized. 13
14
Since no ideal protein biomarker has been shown totally effective in predicting GBM 15
patients’ outcomes, some investigators have evaluated the prognostic significance of 16
miRNAs that have recently been recognized as important regulators of cancer 17
biological behavior (6-10). However, the expression pattern and prognostic 18
significance of miRNAs in GBM remain elusive due to the lack of large pools of 19
clinical specimens for screening. Previous studies have shown that miR-663, a 20
member of primate-specific miRNA family, is associated with many important 21
biological processes, including viral infection, inflammatory responses and 22
autoimmune diseases (11-16). However, its role in tumor progression is contradictory. 23
While it acts as an oncogene to promote the malignancy of lung cancer, 24
nasopharyngeal carcinoma and breast cancer (13, 14), miR-663 may also be a 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
7
potential tumor suppressor in gastric cancer, colorectal carcinoma, prostate cancer and 1
acute lymphoblastic leukemia (15, 16). Moreover, the prognostic significance of 2
miR-663 in cancer has not been reported. 3
4
Correlation of changes in miRNA levels with tumorigenesis and tumor progression 5
remains elusive. However, miRNA expression profiling could be useful in improving 6
the classification of tumors and predicting their behavior (17). In the present study, we 7
identified miR-663 from miRNA profile in human GBM and evaluated its expression 8
and clinical relevance. Its contribution to GBM malignancy and the underlying 9
molecular mechanisms were also investigated. Our data demonstrate that miR-663 has 10
potential values as a prognostic marker and a therapeutic target of GBM. 11
12
13
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
8
Materials and methods 1
Patients and tumor specimens 2
Surgical specimens were obtained from 256 patients with astrocytic gliomas 3
(Southwest Hospital, Third Military Medical University (TMMU), Chongqing, and 4
Tiantan Hospital, Capital Medical University, Beijing) with written consent. After 5
surgical removal, specimens were immediately fixed in a 4% buffered formaldehyde 6
solution. The formalin-fixed, paraffin-embedded (FFPE) samples were stored at room 7
temperature. Histological diagnoses were independently made by two 8
neuropathologists according to WHO classification of central nervous system tumors 9
(2007). Normal brain tissues adjacent to tumors were used as controls. The 10
clinicopathological features of these patients were summarized in Supplementary 11
Table S1. Two hundred and thirty-nine patients with complete information were 12
followed up after operation until February 1st, 2013, with a median follow-up time of 13
23.0 months. This study was carried out in accordance with the principles of the 14
Helsinki Declaration and approved by the Ethics Committee of TMMU and Capital 15
Medical University. 16
17
An independent cohort of 483 patient specimens from TCGA database 18
(https://tcga-data.nci.nih.gov/tcga) (18) was applied to validate the prognostic value of 19
miR-663 in GBM patients. The mircoarray data of all GBM tissue samples with Level 20
3 miRNA (Agilent 8 × 15 K Human miRNA-specific Microarray) were used and the 21
results were filtered based on miR-663 expression and clinicopathological features of 22
patients. 23
24
Cell culture 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
9
Human normal glial cell line, HEB, was generously provided by Professor Guang-mei 1
Yan (Department of Pharmacology, Sun Yat-sen University, Guangzhou) (19, 20). 2
U87-MG was obtained from American Type Culture Collection (ATCC, Manassas, 3
VA). CHG5 (21) and primary GBM cells were established from the tumor specimens 4
of glioma patients (Southwest Hospital, TMMU). HEB, U87-MG and primary GBM 5
cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco, 6
Carlsbad, CA) containing 10% fetal bovine serum (FBS) (Gibco, Carlsbad, CA). 7
CHG5 was grown in RPMI-1640 (Sigma-Aldrich, St. Louis, MO) with 10% FBS. All 8
the cells were incubated at 37°C in a humidified incubator with 5% CO2/95% air. 9
10
Real-time RT-PCR 11
Total RNA from tumor cells or frozen sections were extracted using RNAiso (Takara, 12
Dalian, China). RNA from FFPE tissues was isolated with a miRNeasy FFPE Kit 13
(Qiagen, Hilden, Germany). For measurement of miR-663, Bulge-Loop™ miRNA 14
qPCR Primer Set (RiboBio, Guangzhou, China), ReverAidTM First Strand cDNA 15
Synthesis Kit (Fermentas, Burlington, Canada) and SYBR® Premix Ex Taq™ II 16
(Takara, Dalian, China) were used. For other genes, PrimeScript™ RT Master Mix 17
(Takara, Dalian, China) and SYBR® Premix Ex Taq™ II were used. Specific primers 18
for amplification were listed in Supplementary Table S2. All reactions were 19
performed on a CFX96TM Real-Time PCR Detection System (Bio-Rad, CA). The 20
results of miRNA and mRNA real-time PCR were normalized using the threshold 21
cycle (Ct) of U6 and GAPDH, respectively. 22
23
Western blotting 24
Western blotting was carried out as previously described (22). The primary antibody 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
10
used in this study were as follows: anti-PIK3CD (Abnova, Walnut, CA), 1
anti-phospho-Akt (Ser473), anti-Akt (Cell Signaling, Beverly, MA), anti-MMP-2, 2
anti-MMP-7, anti-MMP-9 (R&D Systems, Minneapolis, MN), anti-Cyclin D1 3
(Abcam, Cambridge, UK) and anti-GAPDH (Cell Signaling, Beverly, MA). 4
5
Gene cloning and lentiviral vector generation 6
Pre-miR-663 was cloned from human HEK293T cell genomic DNA using primers 7
listed in Supplementary Table S2. The expression vector containing PIK3CD coding 8
sequence (CDS) was obtained from Origene (Rockville, MD). The pre-miR-663 and 9
PIK3CD-CDS fragments were ligated into the pLOV-CMV-mCherry-EF1a-PuroR 10
vector or pLOV-CMV-eGFP-EF1a-PuroR vector (Neuron Biotech, Shanghai, China), 11
respectively. After sequence verification, the mCherry-pre-miR-663 cassette or the 12
eGFP-PIK3CD cassette, together with psPAX2 and pMD2G constructs (School of 13
Life Sciences, Lausanne, Switzerland), was transfected into 293FT packaging cells. 14
Medium was changed after overnight culture and lentiviruses in the supernatant were 15
collected and filtered through a 0.45-μm filter 48 hrs after transfection. 16
17
Lentiviral infection procedures 18
Cells were infected with concentrated mCherry-pre-miR-663 lentivirus, and/or 19
eGFP-PIK3CD lentivirus and empty control lentivirus vectors. Cells with stable 20
integration of pre-miR-663 or PIK3CD were selected and enriched by flow cytometry 21
(BD FACSAria II, San Jose, CA) to obtain mCherry or eGFP positive cells. 22
23
Transfection procedures 24
miR-663 inhibitor and Negative Control #1 (Ambion, Austin, TX) were transiently 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
11
transfected with a final concentrations of 100 nmol/L using Lipofectamine™ 2000 1
(Invitrogen, Carlsbad, CA). At 5 hr post-transfection, the medium was replaced by 2
DMEM with 10% FBS. The whole process was conducted according to 3
manufacturer’s guidelines. 4
5
Tumor cell proliferation assay 6
Tumor cells (500 cells/well for U87-MG and 800 cells/well for primary GBM cells) 7
were seeded into 96-well plates and cultured for 24, 48, 72 and 96 hrs. At the 8
indicated intervals, 20 µl of Cell Counting Kit-8 (Beyotime, Shanghai, China) were 9
added to each well and incubated for 2 hrs at 37°C. Thermo Multiskan Spectrum 10
Reader (Thermo Scientific, Watertown, MA) was used to measure the absorbance at 11
490 nm. 12
13
Tumor cell invasion assay 14
Twenty-four-well Millicell® chambers with 8.0 µm pore size (Millipore, Billerica, 15
MA) and BD Matrigel™ (BD Biosciences, Franklin Lakes, NJ) were used in tumor 16
cell invasion assay. Tumor cells were plated in the upper chamber pre-coated with a 5 17
µl mixture of BD Matrigel™ and DMEM (1:1, v/v). The lower chamber was added 18
with 500 µl of DMEM with 10% FBS. After 24 or 36 hrs of incubation, the Millicell® 19
chambers were removed from the plates and fixed in 4% paraformaldehyde, and then 20
stained with Crystal Violet Staining Solution (Beyotime, Shanghai, China). 21
Non-invasive cells on the upper surface of Millicell® chambers were wiped with a 22
cotton swab. Invaded cell number was manually counted in five randomly selected 23
fields under light microscope at 200×. 24
25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
12
Luciferase plasmid construction and reporter assay 1
The pmiR-REPORT™ Luciferase miRNA Expression Reporter Vector (RiboBio, 2
Guangzhou, China) was used to construct the pmiR-3’UTR-wt vector. The primer 3
pairs for wild and mutant 3’UTR amplification were listed in Supplementary Table S2. 4
U87-MG cells seeded in 96-well plates were co-transfected with 20 ng pmiR-3’UTR 5
DNA and 100 pmol miR-663 precursor and control oligomers (Pre-miR Negative 6
Control #1) (Ambion, Austin, TX). The activities of both renilla and firefly luciferases 7
were measured using Dual-Luciferase Reporter Assay System (Promega, Madison, 8
WI) on a SpectraMAX Gemini XS Microplate Reader Fluorometer (Molecular 9
Devices, Sunnyvale, CA). The results were presented after normalization with the 10
measured values of firefly luciferase. 11
12
Tumor implantation 13
The animal experiments were approved by the Institutional Animal Care and Use 14
Committee of the Southwest Hospital, TMMU according to the Guide for the Care 15
and Use of Laboratory Animals. Each of 6-weeks-old male SCID mice (Laboratory 16
Animal Center, Southwest Hospital, TMMU) was anesthetized and intracranially 17
injected with 1 × 105 miR-663 overexpressing cells or control cells. Animals were 18
monitored daily for weight change and survival. The brains were collected until the 19
animals became moribund and then sampled for Harris hematoxylin and 20
alcoholiceosin (H&E) staining and immunohistochemistry. 21
22
Immunohistochemistry 23
The FFPE human gliomas and xenografted tumors were sectioned at 4 μm for 24
immunohistochemical staining. The whole process was conducted using Dako 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
13
REAL™ EnVision™ Detection System (DAKO, Carpinteria, CA) according to the 1
manufacturer’s instructions. Primary antibodies included anti-PIK3CD 2
(Sigma-Aldrich, St. Louis, MO), anti-Ki67, anti-vimentin, anti-Cyclin D1 (ZSGB-Bio, 3
Beijing, China), anti-IDH1 R132H, anti-GFAP (Maixin-Bio, Fujian, China), 4
anti-p-AKT (Ser473, Novus Bio, Littleton, CO), anti-MMP-2, anti-MMP-7 and 5
anti-MMP-9 (R&D Systems, Minneapolis, MN). Semi-quantitations of PIK3CD, 6
p-AKT, Cyclin D1, MMP-2, MMP-7 and MMP-9 were independently performed by 7
two neuropathologists according to the staining intensity and the percentage of 8
positive tumor cells as previously described (23). Ki67 proliferation index was 9
determined under 10 randomly selected microscopic fields at 400×. 10
11
Statistical analyses 12
All statistical analyses were performed using PASW 18.0 software. 13
Kolmogorov-Smirnov test was used to estimate the normality of distributions. 14
Student’s t-test or one-way ANOVA was conducted for normally distributed data. 15
miR-663 expression in different grade glioma specimens was assessed by 16
Bonferroni-corrected Mann-Whitney U-test. Pearson’s chi-square test was used to 17
determine the correlation between miR-663 expression and clinicopathological 18
features of patients. Patients’ survival was analyzed with Kaplan-Meier method, using 19
the log-rank test for comparison. The statistical software X-tile (24) was used to 20
determine the cutoff in the glioma cohort from our groups and TCGA database. The 21
Cox’s proportional hazard model was used for univariate and multivariate survival 22
analyses. The correlation between the levels of miR-663 and PIK3CD was measured 23
using nonparametric (Spearman's rho) correlation test and curvilinear regression 24
model. Data were presented as the mean ± SD. Statistic significance was assigned at P 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
14
< 0.05 (*) or P < 0.01 (**). All experiments were performed at least three times with 1
triplicate samples. 2
3
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
15
Results 1
miR-663 is associated with better prognosis in human glioblastoma 2
To identify the potential miRNAs in glioma progression, miRNA profiling was 3
conducted in primary GBM tissues and the matched adjacent normal tissues 4
(Supplementary Table S3 and S4). Among the differential expressed miRNAs, we 5
found that miR-663 was downregulated in GBM tissues (Supplementary Table S3). 6
To validate the relationship between miR-663 expression and human glioma 7
malignancy, we applied real-time PCR to quantitatively analyze endogenous miR-663 8
expression in the established human normal glia cell line (HEB), human glioma cell 9
lines (CHG5 and U87-MG) and primary human glioma cells. We found that CHG5 10
cells from low-grade human glioma expressed relatively higher level of miR-663 than 11
human GBM (grade IV) cells including U87-MG and primary cells (GBM-1, -2 and 12
-3) (Fig. 1A). We then examined miR-663 expression in the specimens from 256 13
glioma patients, and found that miR-663 expression was significantly reduced in 14
high-grade glioma as compared to normal human brain (p < 0.01) and low-grade 15
glioma (p < 0.01) (Fig. 1B and Supplementary Tables S5-S7). Furthermore, miR-663 16
expression in GBM (grade IV) was lower than that in anaplastic glioma (grade III) (p 17
< 0.01) (Fig. 1B). Kaplan-Meier analyses showed that the patients with higher level of 18
miR-663 had longer disease-free survival (DFS) (p < 0.0001) and overall survival (OS) 19
(p < 0.0001) (Fig. 1C). Significantly, we found that GBM patients could be divided 20
into two subgroups with different outcomes based on miR-663 expression, i.e., the 21
higher expression of miR-663, the better prognosis of patients (DFS: p < 0.0001; OS: 22
p < 0.0001) (Fig.1D). The prognostic value of miR-663 in GBM was further verified 23
in 483 GBM patients from TCGA database (Supplementary Fig. S1A). Univariate and 24
multivariate analyses revealed that miR-663 was an independent predictor for DFS 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
16
and OS of GBM patients (Supplementary Table S8 and Table 1). These data indicate 1
an inverse association of miR-663 expression with glioma malignancy, and reveal that 2
miR-663 may be a potential biomarker for prognosis in GBM patients. 3
4
miR-663 inhibits the proliferation and invasion of GBM cells in vitro and in vivo 5
To examine the role of miR-663 in glioma malignancy, we used GBM cells with 6
relatively different levels of miR-663 expression to respectively generate miR-663 7
overexpressing and knock-down cells (Supplementary Fig. S2A and S2B). We found 8
that overexpressed miR-663 reduced, whereas downregulated miR-663 promoted, 9
GBM cell proliferation (Fig. 2A and 2B), without any effect on cell apoptosis 10
(Supplementary Fig. S2C and S2D). Consistent with the in vitro results, xenografted 11
tumors formed by miR-663 overexpressing GBM cells showed lower proliferation 12
capacity as indicated by Ki67 labeling index (Fig. 2C). Ectopic expression of 13
miR-663 abrogated, whereas suppression of miR-663 enhanced, the migratory and 14
invasive abilities of GBM cells, as measured by cell scratch and invasion assays (Fig. 15
2D and 2E; Supplementary Fig. S2E-S2H). Histologically, we found the tumors 16
formed by miR-663 overexpressing cells might be less invasive compared with those 17
by control ones (Supplementary Fig. S2I), suggesting that miR-663 overexpression 18
might abrogate the invasive phenotype of GBM cells. Moreover, the mice bearing 19
miR-663 overexpressing xenografts survived much longer than those bearing control 20
ones (Fig. 2F). Our results reveal that miR-663 could attenuate the proliferation and 21
invasion of GBM cells in vitro and in vivo. 22
23
miR-663 directly targets PIK3CD and inhibits its signaling 24
We next performed bioinformatic analysis to identify the possible targets of miR-663, 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
17
and PIK3CD was singled out as a potential one (Supplementary Fig. S3A and S3B, 1
Supplementary Table S9). TargetScan prediction revealed that the 3’UTR of PIK3CD 2
harbored a putative miR-663 binding site (Fig. 3A), which was confirmed by 3
luciferase reporter assay (Fig. 3B), indicating PIK3CD as a direct target by miR-663. 4
Functionally, PIK3CD overexpression (Fig. 3C) partially compromised the inhibitory 5
effects of miR-663 on GBM proliferation (Fig. 3D) and invasion (Fig. 3E and 3F), 6
while suppression of PIK3CD (Supplementary Fig. S3C) had the opposite effects 7
(Supplementary Fig. S3D-S3F). Our results reveal that the inhibitory effect of 8
miR-663 on glioma is at least in part through targeting PIK3CD. 9
10
To further determine the underlying mechanisms that miR-663 suppresses GBM, we 11
focused on PIK3CD and its important downstream effectors to investigate whether 12
they were responsible for the suppressed proliferation and invasion induced by 13
miR-663. As shown in Fig. 4, PIK3CD, CCND1, MMP2 and MMP7 (Fig. 4A) 14
together with their protein abundances (Fig. 4B) were significantly reduced in the 15
miR-663 overexpressing cells as compared to control ones, possibly through the 16
reduction of PIK3CD-mediated AKT phosphorylation (Fig. 4B). Consistently, IHC 17
analysis on the specimens from the intracranially xenografted tumors indicated that 18
the abundances of PIK3CD, p-AKT, CCND1, MMP2 and MMP7 proteins were 19
reduced by miR-663 (Fig. 4C and Supplementary Fig. S4A). The results reveal that 20
targeting PIK3CD by miR-663 suppresses the activation of PI3K-AKT pathway and 21
reduces the expression of genes related to proliferation and invasion, thereby 22
inhibiting GBM progression (Supplementary Fig. S4B). 23
24
The reduced miR-663 expression correlates with the high level of PIK3CD in 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
18
human GBM 1
To investigate the correlation between miR-663 and its target PIK3CD, we detected 2
their expressions in primary human GBM. In comparison with the matched normal 3
brain tissues, miR-663 was reduced while PIK3CD mRNA (Fig. 5A) were increased 4
in the tumor tissues from 35 GBM patients, showing a significantly negative 5
correlation (r = -0.773, p < 0.01) (Fig. 5B). Consistently, IHC analysis (Fig. 5C) on 6
the specimens from 108 GBM patients revealed an inverse association between 7
miR-663 and PIK3CD (p < 0.01) (Fig. 5D). Kaplan-Meier analysis demonstrated that 8
the high level of PIK3CD expression predicted a short-term DFS (p = 0.0079) (Fig. 9
5E) and OS (p =0.0376) (Fig. 5F) in GBM patients. Univariate analysis revealed that 10
PIK3CD was a potential predictor for DFS and OS of GBM patients (Supplementary 11
Table S8), further confirming an inverse correlation between miR-663 and its target 12
PIK3CD and their functions on the malignant behaviors of human GBM. 13
14
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
19
Discussion 1
miR-663, which is expressed in Homo sapiens and Pan troglodytes, belongs to the 2
primate-specific miRNAs that possibly attribute to the vertebrates evolution, 3
development and carcinogenesis (24-27). The effect of miR-663 in malignant 4
progression is controversial, since it could act as a tumor promoter or suppressor in an 5
organ-specific fashion (13-16). The clinical relevance of miR-663 in glioma remains 6
unknown. In the present study, we selected miR-663 as our target molecule from 7
miRNA profiling and identified miR-663 as a tumor suppressor to inhibit the 8
proliferation and invasion of GBM, representing the first comprehensive analysis of 9
miR-663 in glioma. Mechanistically, we identified PIK3CD as a direct and functional 10
target of miR-663, which facilitated our understanding of the mechanisms underlying 11
GBM progression. Most importantly, we found that miR-663 and PIK3CD, not only 12
were correlated with each other, but also predicted the survival of GBM patients, 13
highlighting the potential values of miR-663 and PIK3CD as novel prognostic 14
biomarkers in human GBM. 15
16
Significant endeavors have been made to demonstrate the altered miRNA expressions 17
and their effects on GBM progression and patients’ survival (28). miRNA-based 18
integrated analyses have provided new perspectives for clinical validation of the 19
association between miRNA expression patterns and tumor grades or specific 20
subtypes, highlighting the potential values of miRNAs in diagnosis and 21
subclassification (29-31). Our present data indicated that miR-663 was differently 22
expressed in 17 grade I glioma, 52 grade II glioma, 62 grade III glioma and 125 GBM 23
(grade IV) specimens, suggesting miR-663 was a potential molecular biomarker in 24
distinguishing tumor grades. In GBM, a subgroup of tumors that expressed a 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
20
relatively higher level of miR-663 was correlated with better prognosis, suggesting 1
miR-663 was a specific biomarker for the prognostic-based GBM subclassification. 2
Furthermore, miRNAs stably exist in formalin-fixed and paraffin-embedded (FFPE) 3
samples and can be easily detected by real-time PCR, in situ hybridization or 4
microarrays (32, 33). Thus, miR-663 could be a novel and clinical feasible candidate 5
for miRNA-based diagnosis and subclassification for GBM. 6
7
GBM is characterized by rapid growth, relentless invasion and redundant 8
microvessels (1). We identified miR-663 as a tumor suppressor to inhibit the 9
proliferation and invasion of GBM cells thereby suppressing the growth and 10
infiltration of GBM tumors, highlighting the therapeutic potential of miR-663 in 11
GBM treatment. Notably, histopathological features of the xenografted tumors by 12
miR-663 overexpression cells were hardly distinguished from those by control ones, 13
which exhibited the pathological characteristics of GBM, including nuclear atypia, 14
cellular pleomorphism, mitotic activity, microvascular proliferation and necrosis (data 15
not shown). However, xenografted tumors with miR-663 overexpression were much 16
lower in proliferation and weaker in invasion, and animals survived longer than 17
control ones, suggesting that miR-663 could be potentially applied in the treatment of 18
GBM. 19
20
PIK3CD, which encodes PI3K subunit p110δ, is primarily expressed in the 21
hematopoietic cells and has recently been discovered to regulate the proliferation and 22
invasion of certain cancer cells (34-39). We identified PIK3CD as a direct target of 23
miR-663 by bioinformatics prediction, signal network analysis, luciferase reporter 24
assay, gain-of-function and rescue experiments. Our results suggested the oncogenic 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
21
role of PIK3CD in GBM malignant proliferative and invasive behaviors, which were 1
consistent with the previous study showing that the organ-specific PIK3CD, but not 2
other isoforms of PI3K subunits, promoted the migration and invasion of GBM cells 3
(36). We further demonstrated that miR-663-induced suppression of PIK3CD partially 4
inactivated AKT pathway, which enhanced our comprehension of the molecular 5
mechanism of GBM progression (Supplementary Fig. S4B). However, the 6
tumor-suppressive effects of miR-663 could not be fully compromised by PIK3CD 7
overexpression. Besides, the prognostic significance of PIK3CD was not as important 8
as that of miR-663, suggesting that other target genes might be involved in the 9
modulation induced by miR-663. Given the therapy-resistance feature of GBM and 10
the drug-resistance properties of some predicted targets of miR-663 (14, 40), the 11
involvement of miR-663 in chemoresistance requires further investigation. 12
13
Our study showed that miR-663 was down-regulated in GBM as compared with the 14
matched normal brain tissue, which could be illustrated by two possible mechanisms. 15
Recent studies have shown that transcription factors (TFs), such as ZEB family 16
members and p53, may directly bind to the promoter regions of miRNAs or genes that 17
harbor miRNAs (41, 42). By bioinformatic analysis, we have identified several 18
potential TFs that could bind to pre-miR-663 promoter regions (data not shown). 19
Some of them are the predicted miR-663 targets, which may form a feedback loop 20
between miR-663 and these TFs. Another potential mechanism for miR-663 21
regulation is epigenetic modification, such as histone acetylation or methylation and 22
DNA methylation (43). In fact, reduction of miR-663 by hypermethylation in its 23
promoter region has been identified in MDA-MB-231 breast cancer cells and K-562 24
leukemia cells (14, 44). Further studies are underway to examine the methylation 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
22
status of miR-663 promoter region in different grade glioma specimens and cell lines. 1
2
Several prognostic biomarkers, including IDH mutational status and ATRX 3
expression/mutation, have been established in glioma (45, 46). We investigated IDH1 4
and ATRX status in TCGA database and stratified the GBM patients into two 5
subgroups by these molecular features (Supplementary Fig. S1B and 1C). Our results 6
showed that miR-663 was a potential prognostic biomarker in IDH1WT subgroup 7
(Supplementary Fig. S1D, left) and IDH1MU subgroup (Supplementary Fig. S1D, 8
right), suggesting that the prognostic value of miR-663 might not rely on IDH1 status. 9
Besides, miR-663 could divide ATRXlow subgroup into two subsets with different 10
prognosis (Supplementary Fig. S1E, left), though its role in the ATRXhigh subgroup 11
was not significant (Supplementary Fig. S1E, right). Further perspective studies are 12
warranted. 13
14
In summary, our study revealed that miR-663 inhibited the proliferation and invasion 15
of GBM cells in vitro and in vivo by directly targeting PIK3CD, and predicted better 16
prognosis in human GBM. More importantly, miR-663 might be a novel biomarker 17
for molecular subclassification of GBM and a target for treatment of this lethal 18
disease. 19
20
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
23
Acknowledgements 1
We thank Prof. Peng Zhang and Prof. Marie C.M. Lin of Institute of Pathology and 2
Southwest Cancer Center, Southwest Hospital and Dr. Ji Ming Wang of Center for 3
Cancer Research, National Cancer Institute at Frederick for reviewing this 4
manuscript. 5
6
Grant Support 7
This project was supported by grants from the National Basic Research Program of 8
China (973 Program, No. 2010CB529400), the National Natural Science Foundation 9
of China (No. 30930103 and 30900564) and Programs of Science and Technology 10
Commission Foundation of Tianjin Municipal (10SYSYJC28800, 12ZCDZSY17400). 11
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
24
References 1
1. Tanaka S, Louis DN, Curry WT, Batchelor TT, Dietrich J. Diagnostic and 2
therapeutic avenues for glioblastoma: no longer a dead end? Nat Rev Clin Oncol 3
2013;10:14-26. 4
2. Ohgaki H, Kleihues P. Epidemiology and etiology of gliomas. Acta Neuropathol 5
2005;109:93-108. 6
3. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al. 7
Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma 8
characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 9
2010;17:98-110. 10
4. Noushmehr H, Weisenberger DJ, Diefes K, Phillips HS, Pujara K, Berman BP, et 11
al. Identification of a CpG island methylator phenotype that defines a distinct 12
subgroup of glioma. Cancer Cell 2010;17:510-22. 13
5. Yan W, Zhang W, You G, Bao Z, Wang Y, Liu Y, et al. Correlation of IDH1 14
mutation with clinicopathologic factors and prognosis in primary glioblastoma: a 15
report of 118 patients from China. PLoS ONE 2012;7:e30339. 16
6. Ma R, Yan W, Zhang G, Lv H, Liu Z, Fang F, et al. Upregulation of miR-196b 17
confers a poor prognosis in glioblastoma patients via inducing a proliferative 18
phenotype. PLoS ONE 2012;7:e38096. 19
7. Riddick G, Fine HA. Integration and analysis of genome-scale data from gliomas. 20
Nat Rev Neurol 2011;7:439-50. 21
8. Pang JC, Kwok WK, Chen Z, Ng HK. Oncogenic role of microRNAs in brain 22
tumors. Acta Neuropathol 2009;117:599-611. 23
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
25
9. Guan Y, Mizoguchi M, Yoshimoto K, Hata N, Shono T, Suzuki SO, et al. 1
MiRNA-196 is upregulated in glioblastoma but not in anaplastic astrocytoma and 2
has prognostic significance. Clin Cancer Res 2010;16:4289-97. 3
10. Godlewski J, Nowicki MO, Bronisz A, Nuovo G, Palatini J, De Lay M, et al. 4
MicroRNA-451 regulates LKB1/AMPK signaling and allows adaptation to 5
metabolic stress in glioma cells. Mol Cell 2010;37:620-32. 6
11. Jian P, Li ZW, Fang TY, Jian W, Zhuan Z, Mei LX, et al. Retinoic acid induces 7
HL-60 cell differentiation via the upregulation of miR-663. J Hematol Oncol 8
2011;4:20. 9
12. Ni CW, Qiu H, Jo H. MicroRNA-663 upregulated by oscillatory shear stress plays 10
a role in inflammatory response of endothelial cells. Am J Physiol Heart Circ 11
Physiol 2011;300:1762-9. 12
13. Yi C, Wang Q, Wang L, Huang Y, Li L, Liu L, et al. MiR-663, a microRNA 13
targeting p21(WAF1/CIP1), promotes the proliferation and tumorigenesis of 14
nasopharyngeal carcinoma. Oncogene 2012;31:4421-33. 15
14. Hu H, Li S, Cui X, Lv X, Jiao Y, Yu F, et al. The Overexpression of 16
hypomethylated miR-663 induces chemotherapy resistance in human breast 17
cancer cells by targeting heparin sulfate proteoglycan 2 (HSPG2). J Biol Chem 18
2013;288:10973-85. 19
15. Pan J, Hu H, Zhou Z, Sun L, Peng L, Yu L, et al. Tumor-suppressive mir-663 20
gene induces mitotic catastrophe growth arrest in human gastric cancer cells. 21
Oncol Rep 2010;24:105-12. 22
16. Tili E, Michaille JJ, Adair B, Alder H, Limagne E, Taccioli C, et al. Resveratrol 23
decreases the levels of miR-155 by upregulating miR-663, a microRNA targeting 24
JunB and JunD. Carcinogenesis 2010;31:1561-6. 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
26
17. Lujambio A, Lowe SW. The microcosmos of cancer. Nature 2012;482:347-55. 1
18. Cancer Genome Atlas Research Network. Comprehensive genomic 2
characterization defines human glioblastoma genes and core pathways. Nature 3
2008;455:1061-8. 4
19. Wu S, Lin Y, Xu D, Chen J, Shu M, Zhou Y, et al. MiR-135a functions as a 5
selective killer of malignant glioma. Oncogene 2012;31:3866-74. 6
20. Kumar B, Hovland AR, Prasad JE, Clarkson E, Cole WC, Nahreini P, et al. 7
Establishment of human embryonic brain cell lines. In Vitro Cell Dev Biol Anim 8
2001;37:259-62. 9
21. Ping YF, Yao XH, Jiang JY, Zhao LT, Yu SC, Jiang T, et al. The chemokine 10
CXCL12 and its receptor CXCR4 promote glioma stem cell-mediated VEGF 11
production and tumour angiogenesis via PI3K/AKT signaling. J Pathol 12
2011;224:344-54. 13
22. Yu SC, Xiao HL, Jiang XF, Wang QL, Li Y, Yang XJ, et al. Connexin 43 reverses 14
malignant phenotypes of glioma stem cells by modulating E-cadherin. Stem Cells 15
2012;30:108-20. 16
23. Yang Y, Liu Y, Yao X, Ping Y, Jiang T, Liu Q, et al. Annexin 1 released by 17
necrotic human glioblastoma cells stimulates tumor cell growth through the 18
formyl peptide receptor 1. Am J Pathol 2011;179:1504-12. 19
24. Camp RL, Dolled-Filhart M, Rimm DL. X-tile: a new bio-informatics tool for 20
biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res 21
2004;10:7252-9. 22
25. Liu JL, Liang XH, Su RW, Lei W, Jia B, Feng XH, et al. Combined analysis of 23
microRNome and 3'-UTRome reveals a species-specific regulation of 24
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
27
progesterone receptor expression in the endometrium of rhesus monkey. J Biol 1
Chem 2012;287:13899-910. 2
26. Sundaram GM, Common JE, Gopal FE, Srikanta S, Lakshman K, Lunny DP, et 3
al.'See-saw' expression of microRNA-198 and FSTL1 from a single transcript in 4
wound healing. Nature 2013;495:103-6. 5
27. Zhang JF, He ML, Fu WM, Wang H, Chen LZ, Zhu X, et al. Primate-specific 6
microRNA-637 inhibits tumorigenesis in hepatocellular carcinoma by disrupting 7
signal transducer and activator of transcription 3 signaling. Hepatology 8
2011;54:2137-48. 9
28. Karsy M, Arslan E, Moy F. Current progress on understanding microRNAs in 10
glioblastoma multiforme. Genes Cancer 2012;3:3-15. 11
29. Cummins JM, He Y, Leary RJ, Pagliarini R, Diaz LA, Jr, Sjoblom T, et al. The 12
colorectal microRNAome. Proc Natl Acad Sci USA 2006;103:3687-92. 13
30. Suryawanshi S, Vlad AM, Lin HM, Mantia-Smaldone G, Laskey R, Lee M, et al. 14
Plasma microRNAs as novel biomarkers for endometriosis and 15
endometriosis-associated ovarian cancer. Clin Cancer Res 2013;19:1213-24. 16
31. de Leeuw DC, van den Ancker W, Denkers F, de Menezes RX, Westers TM, 17
Ossenkoppele GJ, et al. MicroRNA profiling can classify acute leukemias of 18
ambiguous lineage as either acute myeloid leukemia or acute lymphoid leukemia. 19
Clin Cancer Res 2013;19:2187-96. 20
32. Khoshnaw SM, Powe DG, Ellis IO, Green AR. Detection and quantification of 21
microRNAs in laser-microdissected formalin-fixed paraffin-embedded breast 22
cancer tissues. Methods Mol Biol 2011;755:119-42. 23
33. Liu A, Xu X. MicroRNA isolation from formalin-fixed, paraffin-embedded tissues. 24
Methods Mol Biol 2011;724:259-67. 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
28
34. Koutros S, Schumacher FR, Hayes RB, Ma J, Huang WY, Albanes D,et al.Pooled 1
analysis of phosphatidylinositol 3-kinase pathway variants and risk of prostate 2
cancer. Cancer Res 2010;70:2389-96. 3
35. Fang Y, Xue JL, Shen Q, Chen J, Tian L. MicroRNA-7 inhibits tumor growth and 4
metastasis by targeting the phosphoinositide 3-kinase/Akt pathway in 5
hepatocellular carcinoma. Hepatology 2012;55:1852-62. 6
36. Luk SK, Piekorz RP, Nurnberg B, Tony To SS. The catalytic phosphoinositol 7
3-kinase isoform p110delta is required for glioma cell migration and invasion. Eur 8
J Cancer 2012;48:149-57. 9
37. Zhong M, Bian Z, Wu Z. miR-30a suppresses cell migration and invasion through 10
downregulation of PIK3CD in colorectal carcinoma. Cell physiol Biochem 11
2013;31:209-18. 12
38. Cui F, Li X, Zhu X, Huang L, Huang Y, Mao C, et al. MiR-125b inhibits tumor 13
growth and promotes apoptosis of cervical cancer cells by targeting 14
phosphoinositide 3-kinase catalytic subunit delta. Cell physiol Biochem 15
2012;30:1310-8. 16
39. Sujobert P, Bardet V, Cornillet-Lefebvre P, Hayflick JS, Prie N, Verdier F, et al. 17
Essential role for the p110delta isoform in phosphoinositide 3-kinase activation 18
and cell proliferation in acute myeloid leukemia. Blood 2005;106:1063-6. 19
40. Chapuy B, Koch R, Radunski U, Corsham S, Cheong N, Inagaki N, et al. 20
Intracellular ABC transporter A3 confers multidrug resistance in leukemia cells by 21
lysosomal drug sequestration. Leukemia 2008;22:1576-86. 22
41. Brabletz S, Bajdak K, Meidhof S, Burk U, Niedermann G, Firat E, et al. The 23
ZEB1/miR-200 feedback loop controls Notch signalling in cancer cells. EMBO J 24
2011;30:770-82. 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
29
42. Suh SS, Yoo JY, Nuovo GJ, Jeon YJ, Kim S, Lee TJ, et al. MicroRNAs/TP53 1
feedback circuitry in glioblastoma multiforme. Proc Natl Acad Sci USA 2
2012;109:5316-21. 3
43. Sato F, Tsuchiya S, Meltzer SJ, Shimizu K. MicroRNAs and epigenetics. FEBS J 4
2011;278:1598-609. 5
44. Yang Y, Wang LL, Li YH, Gao XN, Liu Y, Yu L. Effect of CpG island 6
methylation on microRNA expression in the k-562 cell line. Biochem Genet 7
2012;50:122-34. 8
45. Weller M, Pfister SM, Wick W, Hegi ME, Reifenberger G, Stupp R. Molecular 9
neuro-oncology in clinical practice: a new horizon. The lancet oncology. 10
2013;14:e370-9. 11
46. Brennan CW, Verhaak RG, McKenna A, Campos B, Noushmehr H, Salama SR, 12
et al. The somatic genomic landscape of glioblastoma. Cell. 2013;155:462-77. 13
14
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
30
Figure Legends 1
Fig. 1. miR-663 expression correlates with glioma grade and the prognoses of patients. 2
A, miR-663 expression in normal glial cell line (HEB), glioma cell lines (CHG5 and 3
U87-MG) and primary cells (GBM-1, -2, and -3) assessed by qRT-PCR. The relative 4
expression of miR-663 was normalized against U6 expression, which was employed 5
for the comparison between HEB and other cells. The miR-663/U6 ratio in HEB was 6
arbitrarily set to 1.0. B, Expression of miR-663 in the formalin-fixed, 7
paraffin-embedded (FFPE) glioma tissues from 256 patients assessed by qRT-PCR. 8
The transcript level of miR-663 was normalized against U6. The relative expression 9
of miR-663 in specimens was determined as compared to that in adjacent normal 10
tissues. The average expression of miR-663 in adjacent normal tissues was arbitrarily 11
set to 0 with log 10 fold change. All experiments in A and B were performed at least 12
in triplicate and the data are presented as the mean ± SD. **, P < 0.01. C and D, 13
Kaplan-Meier analysis of the correlation between miR-663 and DFS (left) or OS 14
(right) of all grade glioma patients (C) and glioblastoma (GBM) patients (D). Patients 15
were divided into two groups using X-tile analysis as mentioned in Materials and 16
methods. 17
18
Fig. 2. miR-663 inhibits the proliferation and invasion of GBM cells in vitro and in 19
vivo. A and B, Growth curve from the miR-663 up-regulated (A) or down-regulated 20
cells (B) assessed by cell proliferation assay. C, Representative images (left) and 21
proliferation index (right) in tumors indicated by Ki67 staining. Ki67 positive cells 22
were counted microscopically under 10 high powered fields per group, and the results 23
were presented as the mean number of Ki67 positive cells. Scale bar = 25 μm. D and 24
E, Quantification of invaded tumor cells analyzed by cell invasion assay in the 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
31
miR-663 up-regulated (D) or down-regulated cells (E). All experiments in A-E were 1
performed at least in triplicate and the data are presented as the mean ± SD. **, P < 2
0.01. F, Kaplan-Meier analysis of the survival of SCID mice bearing xenografted 3
tumors formed by miR-663 over-expressing cells and control cells. 4
5
Fig. 3. miR-663 directly targets PIK3CD. A, The predicted miR-663 binding sites in 6
the PIK3CD-3’-UTR region by Targetscan and the designed mutant sequence 7
(PIK3CD-3’-UTR-mutant). B, Luciferase reporter assay of U87-MG cells transfected 8
with pmiR-PIK3CD-3’-UTR, the corresponding mutant reporter, miR-663 precursor 9
oligomer or control oligomer. C, qRT-PCR analysis of PIK3CD overexpressing 10
efficiency in U87-MG cells as indicated. The transcript levels were normalized 11
against GAPDH expression. D, Growth curves from the indicated cells assessed by 12
cell proliferation assay. a, b, c and d denote statistical significance among different 13
groups. a: (miR-663 vs. miR-Ctrl), P < 0.01; b: (PIK3CD vs. miR-Ctrl), P < 0.01; c: 14
(miR-663+PIK3CD vs. miR-Ctrl), P < 0.01; d: (miR-663 vs. miR-663+PIK3CD), P < 15
0.01. E and F, Numbers (E) and representative images (F) of indicated invaded cells 16
analyzed by cell invasion assay. Scale bar = 50 μm. Experiments in B-E were 17
performed at least in triplicate and the data are presented as the mean ± SD. **, P < 18
0.01. 19
20
Fig. 4. miR-663 overexpression suppresses the activation of AKT pathway. A, 21
qRT-PCR analysis of PIK3CD, CCND1, MMP2, MMP7 and MMP9 in U87-MG with 22
miR-663 overexpressing vector or empty vector. B, Western blot analyses of PIK3CD, 23
phosphorylated AKT, AKT, CCND1, MMP2, MMP7 and MMP9 in the cells as 24
indicated. Expressions of indicated targets were normalized against GAPDH. C, 25
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
32
Quantification of IHC staining of PIK3CD, p-AKT, CCND1, MMP2, MMP7 and 1
MMP9 in the xenograft tumors as indicated. The expressions of targets were scored 2
by staining intensity and area as described in Materials and Methods. All experiments 3
were performed at least in triplicate and the data in A and C are presented as the mean 4
± SD. *, P < 0.05, **, P < 0.01. 5
6
Fig. 5. The relationship between miR-663 and PIK3CD in human GBM specimens. A 7
and B, the expressions of miR-663 and PIK3CD (A) and their correlation (B) in the 8
GBM specimens with matched adjacent normal tissues obtained from 35 patients. The 9
transcript levels of miR-663 and PIK3CD were respectively normalized against U6 10
and GAPDH, which were employed for the comparison between tumor and adjacent 11
brain tissue. The miR-663/U6 ratio and PIK3CD/GAPDH one in adjacent normal 12
tissues were arbitrarily set to 1.0. Experiments were performed at least in triplicate 13
and the data are presented as the mean ± SD. C, Representative IHC staining of 14
PIK3CD in GBM specimens stratified by miR-663 expression. Scale bar = 25 μm. 15
Case 7 and Case 34 in (A) represented miR-663high group and miR-663low group 16
respectively. D, The correlation of miR-663 and PIK3CD protein expression was 17
analyzed by Pearson’s chi-square test. GBM specimens were divided into two groups 18
based on miR-663 or PIK3CD expression using X-tile analysis. **, P < 0.01. E and F, 19
Kaplan-Meier analysis indicated the correlation between PIK3CD and DFS (E) or OS 20
(F) of GBM patients. The patients were divided into two groups based on the 21
conditions mentioned in (D). 22
23
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284
Table 1. Multivariate analysis for disease-free survival and overall survival in GBM patients.
Factors Disease-free survival Overall survival
HR (95%CI) P value HR (95%CI) P value
Gender 1.463 (0.893-2.397) 0.131 1.545 (0.953-2.507) 0.078
Age 0.998 (0.981-1.016) 0.839 0.996 (0.979-1.013) 0.626
Predominant Side 1.027 (0.651-1.621) 0.909 1.271 (0.806-2.005) 0.302
Predominant Lobe 0.879 (0.707-1.092) 0.245 0.874 (0.708-1.079) 0.21
KPS Score 0.997 (0.981-1.014) 0.762 1.010 (0.992-1.027) 0.273
Extent of surgical resection 1.216 (0.757-1.953) 0.419 1.088 (0.684-1.730) 0.723
Ki67 Index 1.014 (0.998-1.031) 0.091 1.015 (0.999-1.032) 0.068
IDH1(R132H) Status 0.285 (0.103-0.794) 0.016 0.253 (0.090-0.707) 0.009
miR-663 0.833 (0.699-0.992) 0.041 0.828 (0.704-0.975) 0.024
PIK3CD 1.144 (1.028-1.271) 0.013 1.072 (0.970-1.186) 0.174
Abbreviation: HR, Hazards ratio; CI, Confidence interval; KPS, Karnofsky performance status; IDH1,
isocitrate dehydrogenase (IDH) type 1.
Research. on March 17, 2014. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on February 12, 2014; DOI: 10.1158/1078-0432.CCR-13-2284