Oncotarget83330
www.impactjournals.com/oncotarget
www.impactjournals.com/oncotarget/ Oncotarget, 2016, Vol. 7, (No. 50), pp: 83330-83341
Altered expression of miRNAs and methylation of their
promoters are correlated in neuroblastoma
Marco Maugeri
1,*
, Davide Barbagallo
1,*
, Cristina Barbagallo
1,*
, Barbara Banelli
2,5,*
,
Stefania Di Mauro
3
, Francesco Purrello
3
, Gaetano Magro
4
, Marco Ragusa
1,**
, Cinzia
Di Pietro
1,**
, Massimo Romani
2,**
, Michele Purrello
1,**
1
Dipartimento di Scienze Biomediche e Biotecnologiche, Sezione di Biologia e Genetica G Sichel, Unità di BioMedicina
Molecolare, Genomica e dei Sistemi Complessi, Università di Catania, Catania, Italy, EU
2
UOS Epigenetica dei Tumori, IRCCS A.O.U. San Martino-IST, Genova, Italy, EU
3
Dipartimento di Biomedicina Clinica e Molecolare, Università di Catania, Ospedale Garibaldi, Catania, Italy, EU
4
Dipartimento di Scienze Mediche, Chirurgiche e Tecnologie Avanzate G.F. Ingrassia, Università di Catania, Catania, Italy, EU
5
Department of HealthSciences, University of Genova, Genova, Italy, EU
*
These authors have contributed equally to this work
**
Senior Corresponding Authors
Correspondence to: Michele Purrello, email: purrello@unict.it
Keywords: miRNAs encoding genes, promoter methylation proles, neuroblastoma, gene expression, 5’-AZA
Received: May 05, 2016 Accepted: October 21, 2016 Published: November 04, 2016
ABSTRACT
Neuroblastoma is the most common human extracranial solid tumor during
infancy. Involvement of several miRNAs in its pathogenesis has been ascertained.
Interestingly, most of their encoding genes reside in hypermethylated genomic
regions: thus, their tumor suppressor function is normally disallowed in these tumors.
To date, the therapeutic role of the demethylating agent 5’-Aza-2 deoxycytidine
(5’-AZA) and its effects on miRNAome modulation in neuroblastoma have not been
satisfactorily explored. Starting from a high-throughput expression proling of 754
miRNAs and based on a proper selection, we focused on miR-29a-3p, miR-34b-3p,
miR-181c-5p and miR-517a-3p as candidate miRNAs for our analysis. They resulted
downregulated in four neuroblastoma cell lines with respect to normal adrenal
gland. MiRNAs 29a-3p and 34b-3p also resulted downregulated in vivo in a murine
neuroblastoma progression model. Unlike the amount of methylation of their encoding
gene promoters, all these miRNAs were signicantly overexpressed following
treatment with 5’-AZA. Transfection with candidate miRNAs mimics signicantly
decreased neuroblastoma cells proliferation rate. A lower expression of miR-181c
was signicantly associated to a worse overall survival in a public dataset of 498
neuroblastoma samples (http://r2.amc.nl). Our data strongly suggest that CDK6,
DNMT3A, DNMT3B are targets of miR-29a-3p, while CCNE2 and E2F3 are targets of
miR-34b-3p. Based on all these data, we propose that miR-29a-3p, miR-34b-3p, miR-
181c-5p and miR-517a-3p are disallowed tumor suppressor genes in neuroblastoma
and suggest them as new therapeutic targets in neuroblastoma.
INTRODUCTION
Neuroblastoma is a neuroectodermal tumor that
originates from precursor cells of the sympathetic nervous
system. It represents the third leading cause of cancer-
related deaths in childhood [1]. Its heterogeneous clinical
phenotype, ranging from rapid progression to spontaneous
regression, is due to the biological and genetic features
of the tumor. The prognosis of stage I-III neuroblastoma
(with a tumor conned to the originating organ or its
surrounding tissues) is quite favorable, whereas that of
metastatic stage IV is dismal. Stage IV-S neuroblastoma
Research Paper
Oncotarget83331
www.impactjournals.com/oncotarget
is a metastatic disease seen exclusively in infants: its
metastases spread out very rapidly, but surprisingly it
regresses spontaneously with a high survival rate [2].
The most frequently characterized cytogenetic alterations
in neuroblastoma include: (i) amplication of the gene
encoding MYCN (a protooncogenic transcription factor),
localized at 2p24; (ii) loss of heterozygosity (LOH) or
rearrangements of the distal portion of 1p31-ter, 3p22,
11q23; (iii) gain of chromosome arms 1q or 17q. Besides
these abnormalities, gain of chromosomes 4q, 6p, 7q,
11q and 18q, amplication of MDM2 and MYC genes,
and LOH at 14q, 10q, 19q13 have also been described
[3, 4]. More recently, the involvement of miRNAs
in neuroblastoma pathogenesis has been assessed [5,
6]. Deregulation of miRNAs expression in malignant
neuroblastomas may be due to several factors, as MYCN
amplication, chromosomal deletions, or abnormal
epigenetic regulation [7]. Promoters of tumor suppressor
miRNAs are commonly hypermethylated in cancer [8–
12]. It is known that reversion of this hypermethylation
status could lead to reactivation of their tumor suppressor
function in cancer cells [13–15]. The well-known
demethylating agent 5’-Aza-2 deoxycytidine (5’-AZA)
and its deoxy derivative (Decitabine) are currently used
in the treatment of myelodysplastic syndrome (FDA
approved) [16] and have been suggested for therapy
of diffuse large B-cell lymphoma [10]. Apart from a
published study [17], to date neither the therapeutic role
of 5’-AZA nor its effects on miRNAome modulation
have been appropriately investigated in neuroblastoma.
To clarify the role of miRNAs in neuroblastoma, we
sought to: (1) investigate the effects of 5’-AZA on
the global expression of miRNAs in ve commonly
used neuroblastoma cell line models; (2) analyze the
relationship between the expression of four prioritized
miRNAs and the methylation status of their encoding gene
promoters; (3) identify and characterize the expression of
the most interesting among their targets; (4) assay their
involvement in neuroblastoma cell viability; (5) analyze
the prognostic value of altered expression of miRNAs and
their targets.
RESULTS
MiRNAome expression in neuroblastoma cells
after treatment with 5’-AZA
Treatment with 5’-AZA caused remarkable
alterations of miRNAome expression in all neuroblastoma
cell lines: more than 60% differentially expressed (DE)
miRNAs were upregulated (Supplementary Table S1).
We focused our analysis on 12 miRNAs (miR-22, miR-
29a-3p, miR-34a, miR-126, miR-140-3p, miR-141,
miR-181c-5p, miR-202, miR-455-5p, miR-508-3p, miR-
517a-3p and miR-576-3p). This choice was based on:
(i) their differential expression; (ii) their inclusion in
hypermethylated CpG islands in neuroblastoma or other
neoplasias; (iii) known overexpression after treatment
with demethylating agents in various types of cancers; (iv)
known downregulation in neuroblastoma; (v) functional
relationship with MYCN. These miRNAs were identied
as DE and upregulated in at least three neuroblastoma cell
lines (Supplementary Table S2). The promoter of the genes
encoding seven of them (miR-29a-3p, miR-34a, miR-
126, miR-141, miR-181c-5p, miR-202 and miR-517a-3p)
contained CpG islands, whose methylation signicantly
decreased after treatment with 5’-AZA (see later). Apart
from intronic miR-126, all other miRNAs are intergenic.
A previous report by Watanabe K et al. demonstrates that
miR-126 expression may be regulated by the methylation
status of an upstream CpG island, located within an intron
of its host gene EGFL7 [18]. Single TaqMan expression
assays (STAs), extended to miR-34b (another member of
the miR-34 cluster), revealed that miR-29a-3p, 34b-3p,
181-c-5p and 517a-3p are upregulated in at least three
different neuroblastoma cell lines (Table 1).
CpG island methylation status of DE miRNAs
encoding genes following 5’-AZA treatment
Treatment with 5’-AZA determined a statistically
signicant decrease of the methylation of specic CpG
islands, located upstream to the promoter of the genes
encoding miR-29a, miR-34b/c, miR-126, miR-181c/d,
miR-200c/141, miR-202 and miR-517a (Table 2). Detailed
data on the methylation status within the CpG islands
analyzed, before and after 5’-AZA treatment, are reported
in Supplementary Table S3.
Expression of candidate miRNAs in murine
neuroblastoma biopsies
Sequences of miR-29a-3p, 34b-3p and 181c-5p are
perfectly conserved between humans and rodents. The
data reported by Beckers et al., following R2 microarray
and visualization platform analyses (see Materials and
Methods), showed a marked downregulation of miRNAs
29a-3p and 34b-3p in mouse neuroblastoma samples with
respect to controls in a neuroblastoma progression model
(Supplementary Figure S1) [19]. The same dataset did not
show any signicant variation in the expression of miR-
181c-5p. Data on the expression of miR-517a-3p are not
available due to the absence of the gene encoding this
miRNA in the mouse genome.
Expression of candidate miRNAs in
neuroblastoma cell lines
Expression proling of candidate miRNAs in
GI-ME-N, SK-N-BE(2)-C, SK-N-SH and SH-SY5Y
revealed a statistically signicant downregulation of
miR-181c-5p and 517a-3p in all cell lines. MiR-29a-3p
Oncotarget83332
www.impactjournals.com/oncotarget
Table 1: DE miRNAs after treatment with 5’-AZA
DE miRNA Cell line RQ vs miR-24 RQ vs U6
miR-29a-3p ACN 1.91 1.82
GIMEN 1.34 2.45
SH-SY5Y 1.7 1.8
SK-N-BE(2)-C 2.3 2.25
SK-N-SH 2.91 2.78
miR-34a ACN 2.65 2.52
GIMEN 2.17 3.98
SH-SY5Y 0.32 0.88
SK-N-BE(2)-C 0.55 0.54
SK-N-SH 0.56 0.54
miR-34b-3p ACN 5.21 4.97
GIMEN 1.43 2.61
SH-SY5Y 0.17 0.47
SK-N-BE(2)-C 5.68 5.56
SK-N-SH 1.75 0.72
miR-126 ACN 12.47 11.9
GIMEN 3.23 5.91
SH-SY5Y 0.97 2.67
SK-N-BE(2)-C 0.18 0.18
SK-N-SH 1.17 1.12
miR-141 ACN 2.24 2.13
GIMEN 3.51 6.44
SH-SY5Y 0.15 0.41
SK-N-BE(2)-C 0.87 0.85
SK-N-SH 1.07 1.03
miR-181c-5p ACN 2.11 2.01
GIMEN 2.08 3.81
SH-SY5Y 1.79 4.9
SK-N-BE(2)-C 1.64 1.61
SK-N-SH 1.19 1.14
miR-202 ACN 1.46 1.39
GIMEN 4.82 8.84
SH-SY5Y 0.59 1.61
SK-N-BE(2)-C 1.46 1.43
SK-N-SH 1.3 1.25
miR-517a-3p ACN 1.06 1.01
GIMEN 1.54 1.5
SH-SY5Y 1.99 2.71
SK-N-BE(2)-C 1.77 1.73
SK-N-SH 2.06 1.97
Oncotarget83333
www.impactjournals.com/oncotarget
was downregulated in SK-N-BE(2)-C, SK-N-SH and
SH-SY5Y. MiR-34b-3p was signicantly downregulated
in SK-N-BE(2)-C and GI-ME-N (Supplementary Figure
S2A). In all these experiments, adrenal gland was used as
calibrator tissue.
Selection of DE miRNAs targets
In silico analysis of DE miRNAs targets allowed to
select four validated targets for both miR-29a-3p (CDK6,
DNMT3A, DNMT3B, RAN) and miR-181c-5p (BCL2,
GATA6, KIT, SIRT); ve validated targets for miR-34b-
3p (BCL2, CCNE2, CDK4, E2F3, MYB); four predicted
targets for miR-517a-3p (IFNAR1, OLFM3, TNIP1,
WEE1) (Supplementary Table S4). Expression of these
16 targets was assayed in SH-SY5Y and SK-N-BE(2)-C
after treatment with 5’-AZA. Eight targets resulted
signicantly downregulated after treatment with 5’-AZA:
CDK6 and DNMT3B (validated targets of miR-29a-3p),
E2F3 (validated target of miR-34b-3p), and OLFM3
and IFNAR1 (predicted targets of miR-517a-3p) were
downregulated in both cell lines. DNMT3A (validated
target of miR-29a-3p), BCL2 (validated target of both
miR-34b-3p and miR-181c-5p), CCNE2 (validated target
of miR-34b-3p) were downregulated only in SH-SY5Y
(Figure 1).
Targets expression in SK-N-BE(2)-C and SH-
SY5Y transfected with miRNAs mimics
Efficiency of SK-N-BE(2)-C and SH-SY5Y
transfection with miRNAs mimics is shown in
Supplementary Figure S3. Only replicates with a
transfection efficiency > 80% were considered for
downstream assays. CDK6, DNMT3A, DNMT3B
(targets of miR-29a-3p) and CCNE2, E2F3 (targets
of miR-34b-3p) were downregulated in both cell lines
after transfection with the respective miRNAs mimics,
compared to matched scramble-transfected cells in
at least one time point (Figure 2). CDK6, DNMT3A
and DNMT3B show conserved miR-29a-3p binding
sites as retrieved through TargetScan (Supplementary
Table S5). Alignments among miRNAs and their targets
revealed by microRNA.org are shown in Supplementary
Figure S4.
Expression of candidate miRNAs targets in
neuroblastoma cell lines
CCNE2, CDK6, DNMT3B and E2F3 resulted
overexpressed in SK-N-BE(2)-C, SK-N-SH and SH-
SY5Y cell lines with respect to adrenal gland; DNMT3A
was underexpressed in GIMEN, SK-N-BE(2)-C, SK-
N-SH and SH-SY5Y cell lines (Supplementary Figure
S2B). A negative correlation (even though statistically not
signicant) among miR-29a-3p, DNMT3A (r = -0.48) and
DNMT3B (r = -0.60), as well as among miR-34b-3p and
its candidate targets CCNE2 (r = -0.14) and E2F3 (r =
-0.19) was observed. Analysis of Tumor Neuroblastoma -
SEQC - 498 - RPM - seqcnb1 dataset showed a signicant
negative correlation between miR-29a and its candidate
targets DNMT3A (r = -0.110, p-value=0.01) and CDK6 (r
= -0.129, p-value=4.1e-03).
Table 2: Percentage of methylation of CpG islands, before (control) and after 5’-AZA treatment
miR-29a miR-34b/c miR-126 miR-181c/d miR-200c/141 miR-202 miR-517a
NB
cell
lines
Control
(%)
5’Aza
(%)
Control
(%)
5’Aza
(%)
Control
(%)
5’Aza
(%)
Control
(%)
5’Aza
(%)
Control
(%)
5’Aza
(%)
Control
(%)
5’Aza
(%)
Control
(%)
5’Aza
(%)
ACN 85 80 44 38 46 43 59 47 84 70 34 35 80 78
GI-
ME-N
90 73 88 79 84 66 94 62 86 65 88 66 96 77
SK-N-
BE(2)
-C
89 73 93 76 48 39 88 60 89 60 88 73 90 80
SK-N-
SH
85 60 56 44 34 25 87 52 86 54 85 53 88 66
SH-
SY5Y
84 58 13 10 86 53 92 49 96 54 74 60 95 72
p-
value
0.009 0.02 0.05 0.004 0.004 0.04 0.02
Oncotarget83334
www.impactjournals.com/oncotarget
Figure 1: Expression of candidate miRNAs targets in SH-SY5Y and SK-N-BE(2)-C after treatment with 5’-AZA.
Values are reported as fold change (FC) versus untreated cells (controls). * p-value < 0.05; ** p-value < 0.01 (Student’s t-test, n = 3).
Figure 2: Expression of candidate miRNAs targets in SH-SY5Y and SK-N-BE(2)-C transfected with miRNAs mimics
for 24 h and 48 h. Values are reported as fold change (FC) versus scramble-transfected cells (negative controls). * p-value < 0.05; **
p-value < 0.01 (Student’s t-test, n = 3).