Computational identification of microRNAs from the DLK1-DIO3

genomic region as candidate tumor suppressors in papillary

thyroid carcinoma
Leonardo Augusto Marson & Murilo Vieira Geraldo
1 - Department of Structural and Functional Biology, State University of Campinas (UNICAMP), So Paulo, Brazil

ABSTRACT
Papillary thyroid carcinoma (PTC) is the most prevalent histotype of thyroid carcinomas and its incidence has increased globally. More recently, the abnormal expression of microRNAs (miRNAs), small non-protein-
encoding RNAs, emerges as promising diagnosis and prognosis tools for PTC. Large-scale analysis of the expression of microRNAs in murine PTC model has revealed a global reduction in the expression of miRNAs located
along the DLK1-DIO3 genomic region, many of which were identified as tumor suppressors in various types of cancer. This region harbors 53 genes, which can originate approximately 100 mature miRNAs, making
individual functional analysis complex and infeasible. Thus, the objective of this study is to identify the miRNAs of DLK1-DIO3 region with high tumor suppressor potential to further in vitro analysis.. We used the MirWalk
computational program for target prediction and gene expression data from Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA). We constructed the regulatory network potentially modulated by
miRNAs of the DLK1-DIO3 region and since they are downregulated in PTC, we used only interactions between miRNAs with targets upregulated at least 25 % (Fold change> 1,25), with a total of 431 targets. The miRNAs
were then ranked according to the number of targets, revealing that a small number of miRNAs regulate most of the targets: the top 10 miRNAs modulate together 58.9% of the targets (254), where the miRNAs hsa-miR-
485-5p, hsa-miR-495-3p, hsa-miR-539-5p, hsa-miR-665, hsa-miR-381-3p e hsa-miR-300 potentially modulate 65 (15.08%), 64 (14.84%), 61 (14.15%), 50 (11.60%), 49 (11.36%) e 43 (9.97%) of the targets, respectively.
Using the online functional annotation tool DAVID, we performed the gene set enrichment analysis. We observed that these targets modulate biological processes relevant to cancer, such as cell adhesion, extracellular
matrix remodeling, and programmed cell death. Our computational analysis identified the miRNAs from the DLK1-DIO3 region potentially involved in the modulation of important processes for PTC development,
progression and tumorigenesis which are, therefore, promising candidates for functional analysis in vitro. FAPESP funding (2016/09107-0).

INTRODUCTION
Papillary thyroid carcinoma (PTC) is the most prevalent histotype of thyroid carcinomas and its
incidence has increased globally. More recently, the abnormal expression of microRNAs (miRNAs),
small non-protein-encoding RNAs, emerges as promising diagnosis and prognosis tools for PTC.
Large-scale analysis of the expression of microRNAs in murine PTC model has revealed a global
reduction in the expression of miRNAs located along the DLK1-DIO3 genomic region, many of which
were identified as tumor suppressors in various types of cancer. This region harbors 53 genes, which
can originate approximately 100 mature miRNAs, making individual functional analysis complex and
infeasible.

MATERIAL AND METHODS

Figura 2. Matriz de interaes miRNA:alvo e comparao com datasets de expresso gnica oriundas do
Gene Expression Omnibus (GEO) e The Cancer Genome Altas (TCGA). Amostra representativa da matriz de
interaes entre miRNAs e seus alvos preditos segundo o programa miRWalk. O nmero de algortmos que
predizem cada interao est mostrado. Foram consideradas vlidas interaes preditas por 8 ou mais
algortmos, com fold_change maior que 1,25 segundo o TCGA e aumento de expresso em pelo menos 3 dos
cinco datasets do GEO. esi s
en g
o
n gi
a
and
t
n en
n x tio pm
an
e
gio atri n g ntia vel
o
b r re m i
l e e
GEO ie n em ein ion lar lar tion on on igna iffer r e d
t t s a i i
p ro a m pro dhe ellu ellu igr nct act ell s n d latu
o m o a ac ac m ju fr c ro u
yl c las lyc ell xtr xtr ell ell ell ell- eu asc 181
miRNA G P G C E E C C C C N V Total 164
hsa-miR-485-5p 150
hsa-miR-495-3p 138
hsa-miR-665 116
hsa-miR-539-5p 115
hsa-miR-543 111
hsa-miR-654-5p 105
hsa-miR-370-3p 102
hsa-miR-544a 100
hsa-miR-541-3p 99
hsa-miR-432-5p 99
hsa-miR-381-3p 88
hsa-miR-300 83
hsa-miR-656-3p 81
hsa-miR-377-3p 80
hsa-miR-655-3p 79
hsa-miR-410-3p 79
hsa-miR-329-3p 77
hsa-miR-329-3p 77
hsa-miR-494-3p 76
hsa-miR-493-5p 75
hsa-miR-134-5p 70
hsa-miR-127-5p 65

RESULTS 58
hsa-miR-382-5p
hsa-miR-369-3p 58
hsa-miR-758-3p 58
hsa-miR-323a-3p 58
hsa-miR-409-3p 55
hsa-miR-136-5p 52
hsa-miR-412-3p 51
hsa-miR-493-3p 50
hsa-miR-299-3p 50
hsa-miR-433-3p 49
hsa-miR-380-3p 48
hsa-miR-668-3p 44
hsa-miR-1197 41
hsa-miR-770-5p 40
hsa-miR-431-5p 40
hsa-miR-654-3p 39
hsa-miR-376b-3p 38
hsa-miR-376c-3p 37
hsa-miR-154-5p 36
hsa-miR-485-3p 36
hsa-miR-487a-3p 34
hsa-miR-889-3p 33
hsa-miR-541-5p 33
hsa-miR-376a-3p 33
hsa-miR-376a-3p 33
hsa-miR-496 31
hsa-miR-376a-3p
hsa-miR-299-5p
30
30 BCPAP
hsa-miR-323b-5p 27 4.010 5
hsa-miR-323a-5p 26
hsa-miR-337-3p 26
hsa-miR-1185-1-3p 22
hsa-miR-1185-2-3p 22
hsa-miR-1185-5p
hsa-miR-1185-5p
20
20
3.010 5
hsa-miR-409-5p 19
n. cells

hsa-miR-379-5p 16
hsa-miR-411-5p 15
hsa-miR-411-3p 13 2.010 5
hsa-miR-379-3p 11
hsa-miR-432-3p 10
hsa-miR-154-3p 9
hsa-miR-1193 7
hsa-miR-337-5p 7
hsa-miR-127-3p 7 1.010 5
hsa-miR-380-5p 7
hsa-miR-377-5p 6
hsa-miR-539-3p 6
hsa-miR-376a-5p 4
hsa-miR-376a-5p 2 0
hsa-miR-323b-3p 2
hsa-miR-487b-3p
hsa-miR-382-3p
2
0
48h 72h
hsa-miR-136-3p 0
hsa-miR-431-3p 0
hsa-miR-329-5p 0
Figura 3. Agrupamento por anotao funcional. Entre os alvos
hsa-miR-494-5p 0 Figuras 3 e 4. Analise de Proliferao celular em duas
hsa-miR-758-5p 0
dos miRNAs da regio DLK1-DIO3 os processos biolgicos mais
hsa-miR-376a-2-5p
hsa-miR-376c-5p
0
0
linhagens de Cncer de tiroide (TPC-1 e BCPAP) na presena
enriquecidos mostram a potencial regulao de adeso clula-
hsa-miR-495-5p
hsa-miR-376b-5p
0
0
do mimtico miR-485-5p. Foram cultivadas 104 cells/well em
0
clula, adeso ao substrato, morte celular programada e
hsa-miR-381-5p
hsa-miR-487b-5p 0 placas de 12 poos em DMEM e transfectadas com miR-485-
hsa-miR-889-5p 0
morfognese celular. Os miRNAs foram ranquados de acordo com
hsa-miR-134-3p 0 5p na concentrao de 10nM em Optmem. Aps o tempo
hsa-miR-487a-5p 0
o nmero total de alvos potencialmente regulados segundo a
hsa-miR-655-5p
hsa-miR-668-5p
0
0
estabelecido (48 e 72 horas) as clulas foram coletadas para
Figura 1. Construo da Matriz de interaes miRNA:alvo. Amostra respresntativa da anlise bioinformtica (coluna direita), sendo o maior nmero
hsa-miR-369-5p 0 contagem.
hsa-miR-410-5p 0
construo da matriz de interaes entre miRNAs da regio DLK1-DIO3 e seus alvos preditos hsa-miR-412-5p
de alvos mostrado em vermelho e o menor em azul.
hsa-miR-656-5p
0
0
pelo programa miRWalk. hsa-miR-433-5p
hsa-miR-370-5p
0
hsa-miR-329-5p

CONCLUSION
Foi possvel a partir dos dados obtidos pelas analises bioinformticas, encontrar potenciais MicroRNAs como supressores
tumorais a partir de alvos relacionados a diversos processos biolgicos do cncer. E o miR-485-5p como nosso primeiro
objeto de analise j apresentam resultados que comprovam sua eficcia na diminuio da proliferao do cncer de
tiroide.
 Figure 1. Construction of the
matrix of miRNA: target
interactions. A representative
sample of the interaction matrix
between miRNAs of the DLK1-
DIO3 region and their targets
predicted by the miRWalk
program.
 Figure 2. Matrix of miRNA interactions:
target and comparison with gene
expression datasets from Gene
Expression Omnibus (GEO) and The
Higher Cancer Genome (TCGA). Sample
representative of the matrix of
interactions between miRNAs and their
predicted targets according to the
miRWalk program. The number of
algorithms that predict each interaction
is shown. Valid interactions predicted by
8 or more algorithms were considered,
with fold_change greater than 1.25
according to the TCGA and increased
expression in at least 3 of the five
datasets of the GEO.
Figure 3. Grouping by functional annotation. Among the miRNA targets of the DLK1-DIO3 region the most enriched biological processes show the potential regulation of cell-cell adhesion, substrate adhesion, programmed cell death and cellular morphogenesis. The miRNAs were ranked according to the total number of potentially regulated targets according to the bioinformatic analysis (right column), with the highest number of targets being shown in red and the lowest in blue.

Figure 3. Grouping by functional

annotation. Among the miRNA
targets of the DLK1-DIO3 region the
most enriched biological processes
show the potential regulation of cell-
cell adhesion, substrate adhesion,
programmed cell death and cellular
morphogenesis. The miRNAs were
ranked according to the total
number of potentially regulated
targets according to the
bioinformatic analysis (right
column), with the highest number of
targets being shown in red and the
lowest in blue.
 Figures 3 and 4. Analysis of cell
proliferation in two thyroid cancer
lines (TPC-1 and BCPAP) in the
presence of miR-485-5p mimetic.
104 cells / well were grown in 12-
well plates in DMEM and
transfected with 10nM miR-485-
5p in Opmem. After the
established time (48 and 72
hours) the cells were collected for
counting.
 It was possible from the data
obtained by the bioinformatic
analyzes, to find potential
MicroRNAs as tumor suppressors
from targets related to several
biological processes of cancer.
And miR-485-5p as our first object
of analysis already present results
that prove its effectiveness in
reducing the proliferation of
thyroid cancer.