The miR-15a-miR-16-1 locus is homozygously deleted in a subset of prostate cancers.

Abstract

MicroRNAs (miRNAs) are small, non-coding RNAs that negatively regulate the expression of protein coding genes. In this study, we screened highly informative prostate cancer cell lines and xenografts (n = 42) for miRNA gene copy number and expression changes. The expression profiling showed distinction between cell lines and xenografts as well as between androgen sensitive and independent models. Only a few copy number alterations that were associated with expression changes were identified. Most importantly, the miR-15a-miR-16-1 locus was found to be homozygously deleted in two samples leading to the abolishment of miR-15a, but not miR-16, expression. miR-16 is also expressed from another genomic locus. Mutation screening of the miR-15a-miR-16-1 gene in the model systems as well as clinical samples (n = 50) revealed no additional mutations. In conclusion, our data indicate that putative tumor suppressors, miR-15a and miR-16-1, are homozygously deleted in a subset of prostate cancers, further suggesting that these miRNAs could be important in the development of prostate cancer.

Full text

Page 1

GENES, CHROMOSOMES & CANCER 50:499–509 (2011)
The miR-15a-miR-16-1 Locus is Homozygously
Deleted in a Subset of Prostate Cancers
Kati P. Porkka,1Erinn-Lee Ogg,1Outi R. Sarama ¨ki,1Robert L. Vessella,2Heidi Pukkila,3Harri La ¨hdesma ¨ki,4
Wytske M. van Weerden,5Maija Wolf,6Olli P. Kallioniemi,6Guido Jenster,5and Tapio Visakorpi1*
1Instituteof Medical Technology ,Universityof Tampereand Tampere University Hospital,Tampere,Finland
2Universityof Washington Medical Center,N.E.Pacif|c Seattle,WA
3Departmentof Signal Processing,Tampere Universityof Technology ,Tampere,Finland
4Departmentof Informationand Computer Science,Aalto University ,Helsinki,Finland
5Departmentof Urology ,Josephine Nefkens Institute,Erasmus UniversityMedical Center,Rotterdam,The Netherlands
6Institutefor Molecular Medicine Finland (FIMM),Universityof Helsinki,Helsinki,Finland
MicroRNAs (miRNAs) are small, non-coding RNAs that negatively regulate the expression of protein coding genes. In this
study, we screened highly informative prostate cancer cell lines and xenografts (n ¼ 42) for miRNA gene copy number
and expression changes. The expression profiling showed distinction between cell lines and xenografts as well as between
androgen sensitive and independent models. Only a few copy number alterations that were associated with expression
changes were identified. Most importantly, the miR-15a-miR-16-1 locus was found to be homozygously deleted in two
samples leading to the abolishment of miR-15a, but not miR-16, expression. miR-16 is also expressed from another
genomic locus. Mutation screening of the miR-15a-miR-16-1 gene in the model systems as well as clinical samples (n ¼ 50)
revealed no additional mutations. In conclusion, our data indicate that putative tumor suppressors, miR-15a and miR-16-1,
are homozygously deleted in a subset of prostate cancers, further suggesting that these miRNAs could be important in the
development of prostate cancer.
V V
C 2011 Wiley-Liss, Inc.
INTRODUCTION
MicroRNAs (miRNAs) are small, non-coding
RNAs that negatively regulate expression of pro-
tein coding genes. A growing number of miRNAs
has been implicated in cancer, acting either as
tumor suppressive or oncogenic miRNAs (Esquela-
Kerscher and Slack, 2006). Genomic aberrations,
such as amplifications, deletions, translocations,
and mutations, can activate oncogenes and inacti-
vate tumor suppressor genes in a cancer cell. So
far, genomic alterations of only few miRNAs
have been reported. The first miRNAs that were
found to be involved in cancer were miR-15a and
miR-16, which were shown to be commonly
deleted in chronic lymphocytic leukemia (Calin
et al., 2002). These miRNAs are located in a
genomic cluster miR-15a-miR-16-1, at the chro-
mosomal region 13q14.3, which is commonly
deleted in leukemias, lymphomas and several
solid cancers, such as prostate cancer. In a recent
study, 22 miRNAs were shown to be often ampli-
fied or deleted in hepatocellular carcinoma (Ding
et al., 2010). In addition, miR-31 has been shown
to be homozygously deleted in malignant meso-
thelioma (Ivanov et al., 2010). Amplification of a
well-established oncogenic miRNA cluster, miR-
17-92, has been detected in lymphomas and
various solid tumors (Hayashita et al., 2005; He
et al., 2005; Rinaldi et al., 2007; Diosdado et al.,
2009). In addition, a combined gene copy number
and expression analysis has revealed correlation
of increased gene copy number and expression of
eight miRNAs (miR-30d, miR-139, miR-17-3p,
miR-17-5p, miR-18, miR-19a, miR-20a, and miR-
122a) in mantle cell lymphomas (Navarro et al.,
2009). In prostate cancer, the genomic locus of
miR-101 has been found to be deleted (Varam-
bally et al., 2008). In addition to gene copy num-
ber variations, mutations
inactivate cancer-related genes. However, muta-
tions of miRNAs have rarely been discovered. In
a study that cataloged somatic mutations of the
colon cancer cell line COLO-829, one somatic
substitution was found in mir-518d (Pleasance
canactivate or
Supported by: The European Community’s Seventh Framework
Program (FP7/2007-2013), Grant number: HEALTH-F2-2007-
201438; Academy of Finland, Cancer Society of Finland, Reino
Lahtikari Foundation, Sigrid Juselius Foundation, and the Medical
Research Fund of Tampere University Hospital.
*Correspondence to: Tapio Visakorpi, Institute of Medical
Technology, FIN-33014 University of Tampere, Tampere, Fin-
land. E-mail: tapio.visakorpi@uta.fi
Received 30 December 2010; Accepted 21 February 2011
DOI 10.1002/gcc.20873
Published online 31 March 2011 in
Wiley Online Library (wileyonlinelibrary.com).
V V
C 2011 Wiley-Liss, Inc.

Page 2

et al., 2010). In clinical cancers, no mutations of
miRNA genes have been described.
The role of miRNAs in prostate cancer, as well
as in other cancers, has been extensively studied
during the past few years. miRNA expression
profiles have been analyzed in several studies
(Lu et al., 2005; Mattie et al., 2006; Volinia et al.,
2006; Porkka et al., 2007; Ambs et al., 2008;
Tong et al., 2008; Schaefer et al., 2010). The pos-
sible use of miRNAs as biomarkers in prostate
cancer has also been investigated (Mattie et al.,
2006; Mitchell et al., 2008; Schaefer et al., 2010).
Several miRNAs, such as miR-15a and miR16,
have been shown to have tumor suppressor func-
tion in prostate cancer (Bonci et al., 2008). Also,
miRNAs showing oncogenic potential in prostate
cancer, such as miR-21 and miR-125b, have been
reported (Ribas et al., 2009; Shi et al., 2010).
We have previously studied miRNA expression
(Porkka et al., 2007) as well as epigenetic regulation
and androgen regulation of miRNAs in prostate
cancer (Rauhala et al., 2010; Waltering at al., 2010).
In this study, expression and copy number varia-
tions of miRNA genes in prostate cancer cell lines
and xenografts were analyzed. In addition, mutation
screening of miR15a-16-1 was carried out.
MATERIALS AND METHODS
Cell Line, Xenograft, and Tissue Samples
Prostate
DU145, and 22Rv1 were obtained from the
American Type Culture Collection (Manassas,
VA). LAPC4 and VCaP cell lines were provided
by Dr. Charles Sawyers (University of California
at Los Angeles, Los Angeles, CA) and Dr. Jack
Schalken (Radboud University Nijmegen Medical
Center, Nijmegen, the Netherlands),
tively. All cell lines were cultured under recom-
mended conditions.
Freshly frozen LuCaP prostate cancer xeno-
graft samples (LuCaPs 23.1, 23.8, 23.12, 35, 35AI,
41, 49, 58, 69, 70, 73, 77, 78, 81, 86.2, 92.1, 93,
96, 96AI, 105, and 115) were made available for
the analyses by one of the co-authors (R.L.V.).
Genetic characterization of these xenografts has
been reported previously (Laitinen et al., 2002;
Sarama ¨ki et al., 2006). Freshly frozen PC prostate
cancer xenograft samples (PC82, 133, 135, 295,
310, 324, 329, 339, 346, 346I, 346B, 346BI, 347,
and 374F) as well as a cell line (PC346c) were
also made available for the analyses by one of the
co-authors (WvW.). The characteristics of these
xenografts have been reported previously (van
cancer cell lines PC-3,LNCaP,
respec-
Weerden et al., 2000; Hendriksen et al., 2006;
Marques et al., 2006).
Freshly frozen clinical tumor samples from
36 untreated prostate carcinomas and 14 castra-
tion-resistant prostate carcinomas were obtained
from Tampere University Hospital (Tampere,
Finland). The use of clinical specimens had been
approved by the Ethical Committee of the Tam-
pere University Hospital, and written informed
consent has been obtained from the patients.
miRNA Microarray Expression Analysis
Total-RNA was extracted from the cell line
and xenograft samples by using the mirVanaTM
miRNA Isolation Kit (Ambion). Before labeling
and hybridization of the samples, the integrity of
the extracted RNAs was analyzed with the Agi-
lent 2100 Bioanalyzer by using the RNA 6000
Nano Kit (Agilent Technologies). Hybridizations
were carried out by using the Agilent Human
miRNA Microarrays (Version2, Agilent Technolo-
gies), with the miRNA labeling reagent and
hybridization kit (Agilent Technologies). The
hybridization signals were detected by using the
Agilent G2565BA Microarray Scanner System
with the Agilent G2567AA Feature Extraction
Software (Agilent Technologies). Further data
analysis was carried out by using the GeneSpring
(v7.3.1) Expression Analysis software (Agilent
Technologies). The signal values were normal-
ized by 75th percentile. In clustering analysis of
the samples, the ‘‘Distance’’ algorithm was used.
All microarray data have been submitted using
MIAMExpresstothe
(accession number E-MEXP-2966.)
ArrayExpressdatabase
miRNA Gene Copy Number Analysis
The aCGH profiles of the cell lines and the
LuCaPxenografts were
hybridization of the 244K oligo arrays (Agilent
Technologies). The miRNA gene copy number
analysis of the data was carried out by using the
DNA Analytics 4.0 software (Agilent Technolo-
gies). A miRNA gene was considered to be
homozygously deleted or highly amplified if the
average of three successive oligos surrounding the
miRNA locus showed Log2 ratio value lower
than ?2.5 or higher than þ2.5, respectively. The
miRNA gene copy number data from the PC
xenografts, and from four (LNCaP, 22Rv1, VCaP,
and LAPC4) of the cell lines, were obtained from
1M SNP Illumina (Illumina) hybridization analy-
ses. The data was filtered and screened for
obtained fromthe
500
PORKKA ET AL.
Genes, Chromosomes & Cancer DOI 10.1002/gcc

Page 3

miRNA gene copy number alterations. A miRNA
gene was considered homozygously deleted or
highly amplified if the median value of ten SNPs
surrounding the miRNA locus showed the copy
number value (CNV) lower than ?2 or higher
than þ2, respectively.
Quantitative Real-time RT -PCR
The miRNAmicroarray expressionresults
were validated by using the TaqMan MicroRNA
Assays (Applied Biosystems, Foster City, CA).
Briefly, 50 ng of total RNA from each sample was
reverse transcribed with miR-15a and miR-16
specific RT primer using the TaqMan MicroRNA
Reverse Transcription Kit (Applied Biosystems).
As a normalization control, miR-103 was used,
because it showed non-differential expression
across all the samples. qPCR detection was done
with miRNA specific fluorescent probes (Applied
Biosystems) usingthe
DNA Master HybProbe Kit (Roche Applied Sci-
ence, F. Hoffman –La Roche, Basel, Switzerland)
with the LightCycler qPCR instrument.
LightCyclerFastStart
FISH Analysis
Human genomic BAC clone RPCI-11-153K13
HS (Invitrogen), containing the miR-15a-miR-16-
1 gene cluster, at 13q14.3, was used for the gene
copy number analysis by FISH. The probe was la-
beled with Alexa-fluor-dUTP by nick translation.
A PAC 84A6 clone containing the ETB gene at
13q22 was used as a reference probe. The probe
was labeled with digoxigenin-dUTP. The speci-
ficity of the probes was first confirmed by hybrid-
ization of metaphase chromosome preparations
from the DU145 cell line prepared using routine
techniques. After that, 5 lm sections from freshly
frozen xenograft tumor blocks were fixed on
objective slides and dual-color hybridized essen-
tially as described previously (Hyytinen et al.,
1994). After hybridization and washes, the slides
were stained with anti-digoxigenin-FITC and
counterstained with an antifade solution (Vecta-
shield; Vector Laboratories, Burlingane, CA) con-
taining 4,6-diamidino 2-phenylindole. The FISH
signals were detected using an Olympus BX50
epifluorescence microscope (Tokyo, Japan).
Mutation Analysis
Mutation analysis of the miR-15a-miR-16-1
locus at chromosome band 13q14.3 was carried
out by direct sequencing of the DNA extracted
from six cell lines (PC-3, LNCaP, DU145, 22Rv1
LAPC4, and VCaP), from 32 freshly frozen xeno-
graft and 50 clinical tumor samples. Whole
genomic DNA obtained from clinical prostate
carcinoma samples was first amplified to increase
the overall yield of DNA using the Illustra
GenomiPhi V2 DNA Amplification Kit (GE
Healthcare) and 10 ng of original DNA template.
The genomic areacontaining
sequences of both miR-15a and miR-16-1 was
PCR amplified from all samples (primer sequen-
ces 50-TGG TTA AGT GTG ACG TTT TGA
CAA CCA-30and 50-TGT GCT GGG CAC AGA
ATG GAC T-30). The PCR products were puri-
fied with QIAquick PCR purification columns
(Qiagen, Valencia, CA) and the sequencing reac-
tions in both directions were carried out with
nested primers (sequences 50-TTC AGT TAA
GTT TTT GAT GTA GAA ATG-30and 50-TGA
AAA GAC TAT CAA TAA AAC TGA AAA-30)
using the ABI Prism Big Dye Terminator Cycle
Sequencing Ready Reaction Kit. The samples were
sequenced with the ABI 3130xl sequencer.
the precursor
Statistical Analyses
Association between miRNA expression and
copy number variations (CNV) values was ana-
lyzed using the Pearson correlation. Correlations
were calculated for each miRNA expression and
CNV value of the corresponding genomic locus.
miRNA expression values were 75th percentile
normalized using the GeneSpring (v7.3.1) Expres-
sion Analysis software (Agilent Technologies).
For the PC xenograft samples, a CNV value for
each miRNA locus was obtained by calculating
the median value of 10 closest probe sets. For
the LuCaP xenograft samples, CNV values were
quantile normalized, and for each miRNA locus
the CNV value was obtained by calculating the
median value of three closest probe sets. PC and
LuCaP xenograft samples were analyzed sepa-
rately and combined P values were computed
with the Fisher’s method. The P values were cor-
rected for multiple testing with the Benjamini
and Hochberg method.
RESULTS
Clustering Analysis
The cell line and xenograft samples were first
clustered based on their miRNA expression
profiles using the hierarchical clustering method as
implemented in GeneSpring’s clustering algorithm
‘‘Distance.’’ In the resulting hierarchical tree, dif-
ferent sample types were clustered separately from
GENETIC ALTERATIONS OF miRNAs IN PROSTATE CANCER
501
Genes, Chromosomes & Cancer DOI 10.1002/gcc

Page 4

each other. The cell lines (22Rv1, LAPC4,
LNCaP, PC346c, VCaP, DU145, and PC-3)
formed a subcluster of their own that was separated
from the xenograft samples, and the samples of the
two xenograft series (the LuCaP and the PC series)
were mostly, although not completely, clustered in
separate subclusters (Fig. 1). In addition, the
androgen-independent samples were separated
from the androgen-sensitive and androgen-de-
pendent ones. For example, the two androgen re-
ceptor (AR) negative, and therefore completely
androgen-independentcelllines, PC-3and
DU145, formed a pair that is separated from the AR-
positive cell lines (22Rv1, LAPC4, LNCaP, PC346c,
and VCaP). Also, the androgen-independent xeno-
grafts formed nodes of their own (e.g., LuCaP49 and
93; PC346BI and PC346I; PC133, PC135, PC339,
and PC324), separately from the androgen-depend-
ent and androgen-sensitive xenografts within the PC
and LuCaP xenograft subclusters (Fig. 1).
miRNA CNV
First, the aCGH data from six prostate cancer
cell lines and the 18 LuCaP prostate cancer
Figure 1. Clustering analysis of prostate cancer cell lines, LuCaP
xenograft, and PC xenograft samples. The cell lines form a cluster of
their own, separated from the xenograft samples. The two xenograft
series (LuCaP and PC) are also mostly clustered in separate subclus-
ters. Androgen status of the samples is indicated by color-coding:
blue 5 androgen-independent, green 5 androgen-sensitive, red 5
androgen-dependent, and black 5 unknown. [Color figure can be
viewedin theonlineissue,
wileyonlinelibrary.com.]
which is available at
502
PORKKA ET AL.
Genes, Chromosomes & Cancer DOI 10.1002/gcc

Page 5

xenografts (LuCaP23.1, 23.8, 23.12, 35, 35AI, 41,
69, 70, 73, 77, 78, 81, 86.2, 92.1, 96, 96AI, 105,
and 115) obtained with Agilent-platform was ana-
lyzed for detection of homozygous deletions
(HD) or high-level amplifications of miRNA loci.
The alterations detected are summarized in Ta-
ble 1. HD of miRNA loci was detected in only
oneLuCaPxenograft
showed deletion of the miR-15a-miR16-1 locus.
High-level amplifications were detected in three
LuCaP xenografts, showing the amplifications of
miR-223 (LuCaP35),miR-599
(LuCaP35AI), as well as miR-339 and miR-661
(LuCaP105). No HD or high-level amplifications
of miRNA genes were detected in the cell lines.
The 1M SNP Illumina data from 12 PC xeno-
grafts (PC82, 133, 135, 295, 310, 324, 329, 339,
346, 346B, 374, and 374F) was screened for HD
and high-level amplifications of miRNA loci. The
alterations detected are summarized in Table 1.
Interestingly, the miR-15a-miR-16-1 gene that
showed HD in LuCaP 86.2 was detected to be
homozygously deleted also in PC324. In addition,
HDs of three other miRNAs were detected: miR-
562 (PC329), miR-626 (PC135), and miR-744
(PC324). No high-level amplifications of the
miRNA genes were detected in the PC xeno-
grafts. As with the aCGH analysis, the cell lines
studiedby the 1M
(LNCaP, 22Rv1, VCaP, and LAPC4) did not
show HD or high-level amplifications of miRNA
genes.
(LuCaP86.2),which
andmiR-875
SNPIlluminaanalysis
Association of miRNA CNV and Expression
Association of miRNA CNV and expression
was studied in PC and LuCaP xenografts by
using the Pearson correlation. The two xenograft
series were analyzed separately, because the copy
number data was obtained with different plat-
forms (for the LuCaP xenografts with 244K oligo
arrays and for the PC xenografts with 1M SNP
Illumina hybridization), which created high varia-
tion between the two data sets. In both PC xeno-
graft series,miR-15a
statistically positive correlation between gene
copy number and expression at the false positive
rate (FDR) threshold of 0.05 (corrected P values
0.015 for both miRNAs). In the LuCaP xenograft
series, miR-191, and let-7g showed statistically
significant positiveassociation
miRNA gene copy number and the expression
(corrected P values 0.00056 and 0.014, respec-
tively). Also, miR-15a was found to be marginally
significant with corrected P value of 0.07. If the
two xenograft series were combined for the asso-
ciation analysis, statistically significant association
was found between gene copy number and
expression of miR-15a, miR-191, and miR-744
(corrected P values 0.00016, 0.00016, and 0.0078,
respectively).
Next,we concentrated
showed either HD or high-level amplification. Of
the 10 miRNAs that showed HD (miR-15a, miR-
16, miR-562, miR-626, and miR-744) or high-
level amplifications (miR-223, miR-339, miR-599,
miR-661, and miR-875) in at least one of the xen-
ografts, six miRNAs (miR-339, miR-599, miR-
661, miR-562, miR-626, and miR-875) showed no
detectable expression in any of the xenografts.
miR-15a and miR-16 were generally expressed at
high level in the xenografts studied, whereas in
miR-223 at moderate and miR-774 at low level.
The expression and copy number values of miR-
15a, miR-16, and miR-223 in the LuCaP xeno-
grafts and of miR-15a, miR-16, and miR-744 in
the PC xenografts are shown in Figure 2. Xeno-
grafts with the miR-15a locus HD (LuCaP 86.2
andPC324)showed low
whereas no such association was seen with miR-
16, which is located in the same genomic cluster.
The high-level amplification (miR-223) did not
show association of the copy number and expres-
sion (Fig. 2). The expression of miR-744 showing
putative HD was generally so low that it was con-
sidered unreliable for qRT-PCR analysis and was
therefore excluded.
andmiR-744 showed
betweenthe
on miRNAsthat
expressionvalues,
Confirmation of the HD of miR-15a-16-1 Loci
The HDs of miR-15a-miR-16-1 locus detected
in LuCaP 86.2 and PC324 by aCGH and 1M
TABLE 1. miRNA Genes Showing Homozygous Deletions or
High-Level Amplifications in the LuCaP and PC Xenografts
miRNA name
Chromosomal
location Samples
Homozygous deletions
miR-15a
miR-16–1
miR-562
miR-626
miR-744
High-level amplifications
miR-223
miR-339
miR-599
miR-661
miR-875
13q14.3
13q14.3
2q37.1
15q15.1
17p12
LuCaP 86.2; PC324
LuCaP 86.2; PC324
PC-329
PC-135
PC-324
Xq11.1-q12
7p22.3
8q22.2
8q24.3
8q22.2
LuCaP 35
LuCaP 105
LuCaP 35AI
LuCaP 105
LuCaP 35AI
GENETIC ALTERATIONS OF miRNAs IN PROSTATE CANCER
503
Genes, Chromosomes & Cancer DOI 10.1002/gcc