MicroRNA-122, a Tumor Suppressor MicroRNA that
Regulates Intrahepatic Metastasis of Hepatocellular
Carcinoma
Wei-Chih Tsai,1* Paul Wei-Che Hsu,5* Tsung-Ching Lai,7* Gar-Yang Chau,8 Ching-Wen Lin,1 Chun-Ming Chen,1
Chien-Der Lin,2 Yu-Lun Liao,1 Jui-Ling Wang,7 Yat-Pang Chau,3 Ming-Ta Hsu,4 Michael Hsiao,7* Hsien-Da Huang,5,6*
and Ann-Ping Tsou1,7*
MicroRNAs (miRNAs), which are inhibitors of gene expression, participate in diverse bio-
logical functions and in carcinogenesis. In this study, we show that liver-specific microRNA-
122 (miR-122) is significantly down-regulated in liver cancers with intrahepatic
metastastasis and negatively regulates tumorigenesis. Restoration of miR-122 in metastatic
Mahlavu and SK-HEP-1 cells significantly reduced in vitromigration, invasion, and anchor-
age-independent growth as well as in vivo tumorigenesis, angiogenesis, and intrahepatic
metastasis in an orthotopic liver cancer model. Because an inverse expression pattern is often
present between an miRNA and its target genes, we used a computational approach and
identified multiple miR-122 candidate target genes from two independent expression mi-
croarray datasets. Thirty-two target genes were empirically verified, and this group of genes
was enriched with genes regulating cell movement, cell morphology, cell-cell signaling, and
transcription. We further showed that one of the miR-122 targets, ADAM17 (a disintegrin
and metalloprotease 17) is involved in metastasis. Silencing of ADAM17 resulted in a
dramatic reduction of in vitro migration, invasion, in vivo tumorigenesis, angiogenesis, and
local invasion in the livers of nude mice, which is similar to that which occurs with the
restoration of miR-122. Conclusion: Our study suggests that miR-122, a tumor suppressor
microRNA affecting hepatocellular carcinoma intrahepatic metastasis by angiogenesis sup-
pression, exerts some of its action via regulation of ADAM17. Restoration of miR-122 has a
far-reaching effect on the cell. Using the concomitant down-regulation of its targets, includ-
ing ADAM17, a rational therapeutic strategy based on miR-122 may prove to be beneficial
for patients with hepatocellular carcinoma. (HEPATOLOGY 2009;49:1571-1582.)
Since their discovery in Caenorhabditis elegans as im-portant regulators of development timing,1 endog-enous small hairpin microRNAs (miRNAs) have
been found to have a broad impact on vertebrates and
invertebrates by regulating gene activities through trans-
lational repression2 or posttranscription suppression.3-5
Rapid advances in the molecular characterization of miR-
NAs has not only revealed their physiological functions
but also uncovered abnormal patterns of miRNA expres-
sion linked to many disease states including cancer.6,7
Hepatocellular carcinoma (HCC) is one of themost com-
mon human malignancies; this disease shows exceptional
Abbreviations: ADAM17, a disintegrin and metalloprotease 17; HCC, hepatocellular carcinoma; miRNA, microRNA; miR-122, microRNA-122; shRNA, small-
hairpin interfering RNA; 3�UTR, 3� untranslated region; T1, stage I HCC; T3, stage III HCC.
From the 1Department of Biotechnology and Laboratory Science inMedicine, 2Institute of Bioinformatics, 3Anatomy and Cell Biology, and 4Biochemistry andMolecular
Biology, National Yang-MingUniversity, Taipei, Taiwan (Republic of China [ROC]); 5Institute of Bioinformatics and 6Department of Biological Science and Technology,
National Chiao Tung University, Hsin-Chu, Taiwan (ROC); 7Genomics Research Center, Academic Sinica, Nankang, Taipei, Taiwan (ROC); 8Department of Surgery,
Taipei Veterans General Hospital, Taipei, Taiwan (ROC).
Received August 12, 2008; accepted December 13, 2008.
Supported in part by the grants NSC 96-3112-B-010-017 and NSC 95-2752-B-010-002-PAE from the National Science Council to A.P.T., NSC 95-2311-B-009-
004-MY3 to H.D.H., and a grant from the Ministry of Education, Aim for the Top University Plan to A.P.T. and Y.P.C.
*These authors contributed equally to this work.
1571
heterogeneity in cause and outcome. It is especially
prevalent among Asian populations. Despite successful
local therapies such as surgery or transcatheter arterial
chemoembolization, patients with HCC develop a
high rate of recurrence due to local invasion and intra-
hepatic metastasis. Liver cancer is a complex disease
involving epigenetic instability, chromosomal instabil-
ity,8 and expression abnormalities of both coding9,10
and noncoding genes; the latter includes miRNAs.11-14
Profiling of miRNA expression has identified signa-
tures associated with HCC formation and progression
that are potential cancer biomarkers. One such metas-
tasis signature, consisting of 20 miRNAs, has been rec-
ognized as an independent predictor of survival.12
Deregulation of these 20 miRNAs in metastatic HCC
implies that altered expression of their pertinent target
genes may contribute to the development of metastasis
or to recurrence. MicroRNA-122 (miR-122) is a mem-
ber of this metastasis signature.
Recent work on tissue-specific miRNAs has demon-
strated their participation in tissue specification and cell
lineage decisions.15,16 MiR-122 is a liver-specific miRNA
known tomodulate lipidmetabolism,4,17 hepatitis C virus
replication,18 and antiapoptosis.19 Persistent expression of
miR-122 has been detected during liver specialization into
adult liver,20 yet miR-122 is down-regulated in HCC.21
Currently, only a handful of confirmed target genes of
miR-122 have been reported;4,17,20,22 hence, involvement
of miR-122 as an effector of hepatocyte differentiation
remains to be established.
In this study, we demonstrated that the loss ofmiR-122
expression in the intrahepatic metastasis of liver tumors is
highly significant. We determined a functional role for
miR-122 in liver tumor progression using both in vitro
and in vivo models. Furthermore, we applied a bioinfor-
matics-based computational prediction approach to in-
vestigate the miR-122 targets based on two microarray
datasets where miR-122 is significantly down-regulated,
one with local invasion of HCC and the other involving
antagomir-122–treated mouse livers. Finally, we carried
out target verification by 3� untranslated region (3�UTR)
reporter assay and expression analysis. Based on involve-
ment in regulating cell adhesion, migration, and metasta-
sis, a disintegrin and metalloprotease 17 (ADAM17), an
empirically verified target gene of miR-122, was chosen
for in-depth investigation.
Patients and Methods
Cell Lines and Human Liver Tissues. The human
HCC cell lines Mahlavu, HuH-7, and SK-HEP-1 as well
as the human embryonic kidney cell line HEK293T were
cultured as described.23 Paired samples of tumor/nontu-
morous liver tissues were obtained from patients who had
undergone primary HCC curative hepatic resection at
Taipei Veterans General Hospital, Taiwan. The study
was approved by the Committee for the Conduct of Hu-
man Research, and patient informed consent was ob-
tained. This study consisted of 48HCC samples from two
tumor-node-metastasis (TNM) stages, stage I (T1, n �
25) and stage III (T3, n� 23), according to the American
Joint Committee on Cancer 6th edition TNM classifica-
tion. Immediately after surgical resection, all tissues were
snap-frozen in liquid nitrogen and stored at�80°C.
Gene Expression Analysis. Expression analysis of
miR-122 was done by low stringency northern blotting
and TaqMan MicroRNA Assay (Applied Biosystems,
Foster City, CA). RNA (20 �g) was separated on a 15%
denaturing polyacrylamide gel, transferred to Hy-
bond-N�membrane (Amersham Bioscience, Piscataway,
NJ), and hybridized with a [�-32P]-labeled 5�-ACAAA-
CACCATTGTCACACTCCA-3� probe. The most
stringent wash was carried out using 2� standard sodium
citrate, 0.5% sodium, dodecyl sulfate at 42°C before au-
toradiography. The same blot was reprobed for U6 snR-
NAl. TaqMan MicroRNA Assay, a quantitative stem-
loop polymerase chain reaction was performed according
to the manufacturer’s specifications with U6 RNA as an
internal control. Expression of pri-miR-122 was detected
byRT-PCRusingtheprimers5�-GCTCTTCCCATTGC-
TCAAGATG-3� and 5�-GTATGTAACAACAGCAT-
GTG-3�. Expression levels of miR-122 target genes were
also examined by reverse transcription polymerase chain
reaction (RT-PCR) (primer sequences in Supporting Ta-
ble 6).
Plasmid Constructs. A partial human pri-miR-122
gene (a 562–base pair complementary DNA fragment
encompassing 54,269,034 base pairs to 54,269,595 base
Address reprint requests to: Ann-Ping Tsou, D.Sc., Department of Biotechnology and Laboratory Science in Medicine, National Yang-Ming University, Taipei 112,
Taiwan, Republic of China, E-mail: aptsou@ym.edu.tw; fax: (886)-2-28264092.
Copyright © 2009 by the American Association for the Study of Liver Diseases.
Published online in Wiley InterScience (www.interscience.wiley.com).
DOI 10.1002/hep.22806
Potential conflict of interest: Nothing to report.
Additional Supporting Information may be found in the online version of this article.
1572 TSAI, HSU, ET AL. HEPATOLOGY, May 2009
pairs of 18q21.31, University of California Santa Cruz
Genome Browser) was subcloned into the lentiviral ex-
pression vector, pPGK-GFP (National RNAi Core Facil-
ity, Academia Sinica, Taiwan) in place of GFP and called
pPGK-122. pPGK-GFP, a positive control for the lenti-
viral infection, contains necessary cis-elements for pack-
aging, reverse transcription and integration. PGK stands
for the human phosphoglycerate kinase promoter se-
quences. pPGK-122 was used to make lentiviruses ex-
pressing miR-122 (lenti-122). A miR-122 seed-region
(5�-GGAGTGTG-3� to 5�-ATGACTGA-3�) mutant
(lenti-122M) was constructed using a QuickChange Site-
Directed Mutagenesis Kit (Stratagene, La Jolla, CA).
Ectopic Expression of miR-122. Stable HCC cell
cultures expressing miR-122 were generated by coinfect-
ingHCC cells with lenti-122 and short hairpin Luciferase
(shLuc) at 100:1 ratio using 8 �g/mL protamine sulfate
for 24 hours and puromycin (2 �g/mL) selection for 3-5
days. The shLuc targeting the luciferase gene provides the
puromycin resistance. Expression of mature miR-122 was
confirmed by northern blotting.
Prediction of miR-122 Targets Among Differen-
tially Expressed Genes. For miR-122 target prediction,
we applied miRanda24 to energetically detect the most
probable targets against the 3�UTR among all human and
mouse coding genes (BioMart from the Ensembl data-
base25). The miRNA/target duplexes with miRanda scores
greater than 120 and minimum free energy values less than
�10 kcal/mol were identified as miR-122 target sites. The
analysis strategy is presented in Supporting Table 2.
3�UTR Reporter Assay. The 3�UTR fragments of the
candidate target genes were subcloned into the Xba1 site
downstream of the luciferase gene in pGL3-Control Vec-
tor (Promega, Madison, WI). Under the control of SV40
promoter and enhancer, pGL3-Control Vector expresses
strong luciferase activity in many types of mammalian
cells. The negative controls were lenti-122M and lenti-
GFP. HEK293T cells were infected with lenti-GFP and
lenti-122 or lenti-122M for 24 hours. Cells were then
seeded into 24-well plates and cotransfected with 0.5 �g
of the respective pGL3-3�UTR construct and 0.05 �g of
the pRL-TK vector (Promega) using jetPEI (Polyplus-
Transfection, France). pRL-TK vector was used as an in-
ternal control reporter. After 48 hours, luciferase activity
was measured using the Dual-Luciferase Reporter Assay
System Kit (Promega). The effect of miR-122 was ex-
pressed relative to the average value from cells infected
with lenti-GFP virus. A mutant of the single miR-122
binding site (5�-ACACUCCA-3� to 5�-ACCCGCAA-3�)
in the 3�UTR of ADAM17 was included. The mutagen-
esis primers are listed (Supporting Table 7) together with
the nucleotide sequences of all the PCR cloning primers
(Supporting Table 8).
RNA Interference with shADAM17. HCC cells were
plated and infected with lentiviruses expressing shADAM17
or shLuc in thepresence of 8�g/mLprotamine sulfate for 24
hours followed by puromycin (2 �g/mL; 48 hours) selec-
tion. RT-PCR and/or western blotting were performed to
validate the knockdown efficiency. The shRNA constructs
are described in Supporting Fig. 4.
Cell Migration Assay and Invasion Assay. Cells
were tested for their migration and invasion abilities in
vitro in a Minicell (Millipore) according to the methods
described previously.23
Soft-Agar Colony Assay. Cells were seeded at 5� 103
cells/well in a six-well plate containing 0.3% low melting
temperature agarose (Sigma, St. Louis, MO). Colonies
were counted after 14 days. The IntegratedMorphometry
analysis function ofMetaMorph software (Molecular De-
vices, Sunnyvale, CA) was used to measure the surface
areas of the colonies and derive the size of the colonies.
In Vivo Tumorigenesis Assay. Mahlavu and SK-
Hep1 cells (5 � 106) infected with shLuc, lenti-122, or
lenti-shADAM17 lentiviruses were implanted subcutane-
ously into the flanks of nude mice. Primary tumor growth
was analyzed by measuring tumor length (L) and width
(W), and tumor volume was calculated according to V�
0.4�ab2 as described.23 Orthotopical intrahepatic injec-
tion was conducted as described27 by using 20 �L of
phosphate-buffered saline containing 50% Matrigel (BD
Biosciences, MA); mixed with 5 � 106 tumor cells, and
slowly injected into the livers of anesthetized mice using a
27-gauge needle. Mice were sacrificed 6-7 weeks later for
histopathological evaluation. The methods used for the
immunohistochemical staining of anti-CD31, the evalu-
ation of microvessel density, and the quantification of
tumor size are described in Supporting Methods. Only
the results using Mahlavu cells are presented. The animal
studies were approved by Institutional Animal Care and
Use Committee (IACUC) of National Yang Ming Uni-
versity.
Statistical Analyses. All data are expressed as mean�
standard deviation and compared between groups using
the Student t test. Categorical variables were compared
using the Wilcoxon Rank test. A P value�0.05 was con-
sidered statistically significant. *P � 0.05; **P � 0.01;
***P � 0.001.
Results
Specific Down-Regulation of miR-122 in HCC Cor-
relates with Enhanced Transforming Activity. Re-
cently, miR-122 was found to be down-regulated in liver
HEPATOLOGY, Vol. 49, No. 5, 2009 TSAI, HSU, ET AL. 1573
tumors of human and rodents21,22 and is considered a
member of a metastasis-related microRNA signature.12
However, the functional attributes of miR-122 associated
with liver tumor progression have not been experimen-
tally established. To substantiate the significance of miR-
122 expression, a northern blot analysis of 20 HCC
tumor pairs was conducted. Expression of miR-122 was
remarkably lower in T3 tumors with intrahepatic metas-
tasis (Fig. 1A). Validation by quantitative stem-loop PCR
showed that the expression of miR-122 was significantly
higher in the adjacent normal tissue (n � 48) and T1
tumors (n � 25). No statistically significant difference
was found between T1 tumors and adjacent normal tis-
sue. However, significant loss of miR-122 expression was
found in T3 tumors (Fig. 1B), which implies that down-
regulation of miR-122 significantly correlates with intra-
hepatic metastasis as reported.12
To explore the functional role of miR-122 in HCC
progression, ectopic overexpression of miR-122 was es-
tablished using a lentiviral vector (lenti-122). When miR-
122 was overexpressed in SK-HEP-1 or Mahlavu HCC
cells (Fig. 2A), which are known to manifest invasion
activity,23 a significant reduction of cell migration and
invasive activity was observed (Fig. 2B). Overexpression
ofmiR-122 also impaired anchorage-independent growth
in soft agar, significantly reducing the number and size of
the colonies (Fig. 2C). Consistent with the in vitro results,
overexpression of miR-122 in Mahlavu cells significantly
suppressed overall tumor growth and local invasion in
vivo (Fig. 6). Restoration of miR-122 expression, how-
ever, did not influence HCC cell proliferation (Support-
ing Fig. 1). These in vitro and in vivo results imply that the
gradual loss ofmiR-122 over a long period of time is likely
to commit hepatocytes to tumorigenesis and intrahepatic
metastasis.
Identification of miR-122 Target Genes. MiR-122
regulatesmultiple target geneswhose combined expression is
associatedwith themetabolismof hepatic fatty acid and cho-
lesterol as well asHCCpathogenesis.4,17,21 To identify puta-
tive metastasis genes thatmiR-122 suppresses, we integrated
various computational prediction tools to track down genes
where expression was increased by �1.5-fold in T3 HCC
and in antago-miR-122–treated mouse livers. The salient
feature common to these two types of liver tissues is that they
both have extremely low levels of endogenous miR-122. In
addition, the genes needed to have a putativemiR-122 bind-
ing site mapped to their 3�UTR. A list of 45 most probable
targets, including the known genes SLC7A1 20 and AL-
DOA,4 was compiled (Table 1). A reciprocal expression pat-
tern between miR-122 and the candidate genes was
identified by microarray (Supporting Table 1) and by RT-
PCR (Supporting Fig. 2), which supports a role in their
posttranscriptional suppression by miR-122.
Bioinformatic prediction recognized one to sevenmiR-
122 binding sites in the 3�UTRs of these candidate genes
(Table 1, Supporting Table 3). To test whether the pre-
dicted binding sites in the target genes could mediate
repression of translation by miR-122, the 3�UTRs were
subcloned downstream of the luciferase gene in pGL3-
Control. These constructs were assayed in HEK293T
cells infected with lenti-122 or a mutant miR-122 in
which the seed region has been randomly mutated
(miR122-M; Fig. 3A). As shown in Fig. 3A, the reporter
with the 3�UTR of SLC7A1 showed a significantly lower
luciferase activity in cells expressing miR-122 compared
to miR122-M. Similar responses were detected for the
reporter with the 3�UTR of AKT3. MiR-122 thus nega-
tively regulates translation of these genes through the
3�UTR binding sites, and SLC7A1 and AKT3 are true
miR-122 target genes. A significant reduction in luciferase
Fig. 1. Expression of miR-122 was down-regulated in intrahepatic metastastic liver tumors. (A) Expression analysis by northern blotting. T: tumor
part; N: normal adjacent tissue. The intensity of northern hybridization was retrieved using ImageQuant software. Each miR-122 intensity value was
first divided by the corresponding intensity of U6 from the same tissue sample to adjust for sample variation. (B) Decreased miR-122 mRNA
expression was present in an additional 48 pairs of HCC samples (T1, n � 25; T3, n � 23) and demonstrated by TaqMan MicroRNA Assay. Results
were normalized against the expression level of U6 messenger RNA (mRNA) in each sample. The differences in miR-122 expression level between
tissue sample categories were analyzed by Wilcoxon rank sum test. The box plot shows the data distribution across the group classification and
presents the 75th and 25th percentile (upper and lower quartile) with the median value between. T1: primary HCC, T3: intrahepatic metastasis HCC;
N: an average of expression level of all 48 normal adjacent tissues. **P � 0.01; ***P � 0.001.
1574 TSAI, HSU, ET AL. HEPATOLOGY, May 2009
activity (14%-54%) was detected with 32 of the 35
cloned reporter constructs (Fig. 3B). STX6 (syntaxin 6),
SORT-1 (sortilin-1), and SPRED-1 (sprouty-related,
EVH1 domain containing-1) failed to show a significant
reduction in luciferase activity. No specific repression by
miR122-M was detected (Supporting Tables 4 and 5).
These 32 genes are therefore direct targets of endogenous
miR-122.
To uncover the biological functions of the confirmed
miR-122 targets, gene function analysis was carried out
(Ingenuity Pathways Analysis 3.0; Ingenuity Systems,
Redwood City, CA). Genes classified as having the mo-
lecular functions involved in “Cell Morphology” or “Cell
Movem