Systematic Analysis of MicroRNAs Targeting the Androgen...

Tumor and Stem Cell Biology

Systematic Analysis of MicroRNAs Targeting the AndrogenReceptor in Prostate Cancer Cells

P€aivi €Ostling1, Suvi-Katri Leivonen1, Anna Aakula1, Pekka Kohonen1, Rami M€akel€a1, Zandra Hagman2,Anders Edsj€o3,5, Sara Kangaspeska6, Henrik Edgren6, Daniel Nicorici6, Anders Bjartell4, Yvonne Ceder2,Merja Per€al€a1, and Olli Kallioniemi1,6

AbstractAndrogen receptor (AR) is expressed in all stages of prostate cancer progression, including in castration-

resistant tumors. Eliminating AR function continues to represent a focus of therapeutic investigation, but ARregulatory mechanisms remain poorly understood. To systematically characterize mechanisms involvingmicroRNAs (miRNAs), we conducted a gain-of function screen of 1129 miRNA molecules in a panel of humanprostate cancer cell lines and quantified changes in AR protein content using protein lysate microarrays. In thisway, we defined 71 unique miRNAs that influenced the level of AR in human prostate cancer cells. RNAsequencing data revealed that the 30UTR of AR (and other genes) is much longer than currently used in miRNAtarget prediction programs. Our own analyses predicted that most of the miRNA regulation of AR would targetan extended 6 kb 30UTR. 30UTR-binding assays validated 13 miRNAs that are able to regulate this long AR 3’UTR(miR-135b, miR-185, miR-297, miR-299–3p, miR-34a, miR-34c, miR-371–3p, miR-421, miR-449a, miR-449b, miR-634, miR-654–5p, and miR-9). Fifteen AR downregulating miRNAs decreased androgen-induced proliferation ofprostate cancer cells. In particular, analysis of clinical prostate cancers confirmed a negative correlation of miR-34a and miR-34c expression with AR levels. Our findings establish that miRNAs interacting with the long 30UTRof the AR gene are important regulators of AR protein levels, with implications for developing new therapeuticstrategies to inhibit AR function and androgen-dependent cell growth. Cancer Res; 71(5); 1956–67. �2011 AACR.

Introduction

Androgen receptor (AR) is critical for the development andprogression of prostate cancer (PCa) and AR inhibition repre-sents the first-line therapeutic modality for patients withadvanced disease (1–3). In normal prostate, androgens pro-mote survival and differentiation, but during PCa develop-ment, AR becomes an inducer of uncontrolled cell growth (2).Endocrine therapies directed toward reducing serum andro-gens and inhibition of AR initially block PCa growth, but oftenfatal castration-resistant disease develops (4).

The molecular mechanisms responsible for this lethal tran-sition in PCa are not completely understood. AR is known tobe expressed throughout PCa progression (5). The mostconsistent change associated with castration-resistant growthin global gene expression profiles of PCa xenografts was asmall increase in the ARmRNA levels (6). Chen and colleagueshave shown that the increase of AR mRNA and protein wasnecessary and sufficient for the conversion from a hormone-sensitive to a hormone-refractory state (6). Central to the ARresponse are the specific downstream target genes, includingthe oncogenic ETS fusion genes (7). Recent findings suggestthat in the castration-resistant stage of the disease, AR con-trols a distinct transcriptional program linked to cell-cycleregulation (8). Taken together, these results suggest that ARsignaling plays a central role in all stages of PCa and therefore,it is critically important to understand the regulation of itsexpression and to find novel ways of inhibiting this pathway.

MiRNAs are small (�22 nucleotides), endogenously occur-ring untranslated RNAs that control gene expression post-transcriptionally by inhibiting protein translation and/ordegrading target mRNA (9, 10). Each miRNA can controlhundreds of target genes and up to 60% of all transcriptsare potentially modulated by miRNAs (11). Strong linksbetween miRNA deregulation and human disease, particularlycancer, have been established (12–15). Altered miRNA expres-sion profiles have been shown in a large number of cancertypes, including PCa (16) and altered miR-34c expression has

Authors' Affiliation: 1Medical Biotechnology, VTT Technical ResearchCentre of Finland, and Turku Centre for Biotechnology, University of Turkuand A

�bo Akademi University, Turku, Finland; 2Department of Laboratory

Medicine, Division of Clinical Chemistry, 3Department of Laboratory Med-icine, Centre for Molecular Pathology, and 4Department of ClinicalSciences, Division of Urological Cancers, Lund University; 5Universityand Regional Laboratories Region Ska

�ne, Clinical Pathology, Malm€o,

Sweden; and 6Institute for Molecular Medicine Finland (FIMM), Universityof Helsinki, Helsinki, Finland

Note: P. €Ostling and S-K. Leivonen contributed equally.

Corresponding Author: P€aivi €Ostling, VTT Technical Research Centre ofFinland, Medical Biotechnology, PharmaCity, It€ainen Pitk€akatu 4C, P.O.BOX 106, 20521 Turku, Finland. Phone: þ358 400 297 330; Fax: þ358 207222 840; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-10-2421

�2011 American Association for Cancer Research.

CancerResearch

Cancer Res; 71(5) March 1, 20111956

Research. on January 4, 2020. © 2011 American Association for Cancercancerres.aacrjournals.org Downloaded from

Published OnlineFirst February 22, 2011; DOI: 10.1158/0008-5472.CAN-10-2421

http://cancerres.aacrjournals.org/

been shown to inversely correlate with aggressiveness ofPCa (17).To understand the role of miRNAs in PCa, we carried out a

systematic study of their role in regulating the AR protein in 5PCa cell lines. A gain-of-function analysis of 1129 miRNAmolecules was combined with AR protein quantitation usingreverse-phase protein lysate microarray (LMA). RNA sequen-cing was used to determine the length of the AR 30UTR andcustom bioinformatic prediction of miRNAs targeting AR wasdone. With 30UTR luciferase reporter assays, 13 miRNAs wereidentified as regulators of AR 30UTR. Several miRNAs affectingAR protein levels were also potent reducers of androgen-induced proliferation. Taken together, our data show thatposttranscriptional regulation of AR by miRNAs is an impor-tant level of regulation for this central player in PCa.

Materials and Methods

Cell culture and reagentsLNCaP and 22Rv1 cells were from Deutsche Sammlung von

Microorganismen und Zellkulturen GmbH. MDA-PCa-2b,RWPE-1, VCaP and PC-3 were from ATCC (Manassas, VA,USA). LNCaP-C4–2, LNCaP-C4–2B4, PC-3-parental and PC-3–MPro4 were kind gifts of Prof. G. Thalmann (Bern, Switzer-land), CWR-R1 a kind gift of Prof. C. Gregory (Chapel Hill,USA), LAPC-4 a kind gift of Prof. C. Sawyer (Los Angeles) andEP156T a kind gift of Prof. V. Rotter (Rehovot, Israel). All cellswere cultured in medium conditions recommended by theproviders for less than 4 months before use in these experi-ments. For more details of cell line authentication see sup-plementary materials and methods.Human Pre-miR miRNA precursors were obtained from

Ambion Inc. (Austin) and siRNAs siAR_1 and siAR_2 fromQiagen (Hilden, Germany, #SI02757258 and #SI02757265,respectively) and siAR_3 from Ambion (S1538, #4390824).

LMA screening and data analysisFive AR expressing cell lines; LNCaP, LAPC-4, MDA-PCa-2b,

22Rv1, and CWR-R1 were transfected with 20 nmol/L humanPre-miR miRNA Precursor library v2 (Ambion Inc., 319 mole-cules) or 20 nmol/L miRIDIAN microRNA Mimic Librariesv10.1 (Dharmacon, Lafayette, 819 molecules). The reverse-transfections were done in 384-well plates as described pre-viously (18). ARprotein expressionwas detected by staining theslides with an AR antibody overnight (1:500 #H-280, Santa CruzBiotechnology Inc.). The results were normalized using a Loessmethod (19) and log2 transformed. RankProd (20) was used tofind miRNAs that regulate AR protein levels in all 5 cell lines.

Paired-end RNA-sequencingPaired-end RNA-sequencing was conducted on cDNA-

library prepared from VCaP. Detailed description of themethod is available in Supplementary Materials and Supple-mentary Table I. Briefly, fragmented mRNA was used as atemplate for cDNA synthesis. End repair, A-base addition andligation of paired-end adaptors was then done. cDNA tem-plates were size selected (typically 200–300 bp) and paired-end libraries created. Sequencing was done on the Illumina

Genome Analyzer II and sequencing coverage normalizedbetween samples by dividing coverage with the total numberof mapped short reads in each sample. Plots of coverage weredrawn using R and the GenomeGraphs package (21).

Custom miRNA target predictionBLAT search was used to acquire human transcript AR-001

(ENST00000374690) and orthologous UTRs for 10 mammalianspecies that were aligned using ClustalW. Predictions for allthe miRNAs covered by TargetScan 5.0 algorithm wereobtained (11).

ImmunoblottingSDS-PAGE and immunoblotting were done as described

previously (18). AR, PSA, and b-actin, specific primary anti-bodies were used (#H-280, Santa Cruz Biotechnology; #A0562Dako Cytomation; #A1978 Sigma-Aldrich, respectively). Thesignals were obtained with Alexa Fluor 680 tagged secondaryIgG antibodies (Invitrogen) and Odyssey Licor scanner (LI-COR Biosciences).

Quantitative real-time RT–PCR analysisRNA isolation and reverse transcription were done as

described previously (18). Specific primers for AR, glyceralde-hyde-3-phosphate dehydrogenase (GAPDH), and b-actin (ACTB)were designed by Universal Probe Library (Roche AppliedBiosciences; Supplementary Table II). Fluorescent Taqmanprobes were from Roche Human Probe Library (AR: no. 14,GAPDH no. 60, ACTB no. 64). The expression of ARmRNA wasdetermined by the relative quantitation method using GAPDHor ACTB as controls. Data were collected from 2 separatebiological experiments, run twice with triplicates.

AR 30UTR reporter constructs and luciferase assaysSeven separate fragments of the AR 30-UTR region were

amplified from LNCaP genomic DNA using specific primersfor AR 30UTR (Supplementary Table III). The AR 30UTR frag-ments were cloned into MluI/HindIII sites of pMIR-REPORTLuciferase vector (Ambion Inc.). Luciferase assays were donein 22Rv1 cells as previously described (18).

Microarray miRNA expression analyses in prostate celllines

MiRNAexpressionwere profiledwith Agilent humanmiRNAmicroarrays (V1). RNA was isolated from LAPC-4, LNCaP,LNCaP-C4–2, LNCaP-C4–2B4, MDA-PCa-2b, PC-3, PC-3-MPro4, parental PC-3, VCaP and 22Rv1, EP156T, PWR-1E,andRWPE-1cellswithmirVanamiRNAKit (AppliedBiosystem)and labeled according to the Agilent protocol (version 1.0,April 2007). The arrays were scanned with Agilent MicroarrayG2565, and Agilent Feature Extraction Software (version 9.5)wasusedtoextractthedata.Thedatawerenormalizedusingthevsn R-package (22) and generalized log2 variance stabilization.

PCa tissue specimens and analysis of miR-34c levels byqRT-PCR

Prostatic tissues obtained by transurethral resection ofthe prostate (TURPs) were collected 1990–1999 in Malm€o,

microRNAs Targeting AR

www.aacrjournals.org Cancer Res; 71(5) March 1, 2011 1957




Sweden. The materials were fixed in 4% buffered paraformal-dehyde and paraffin-embedded. Results were based on histo-pathological diagnosis in randomly selected cases withevidence of prostate adenocarcinoma in 47 patients. Theage range at the time of TURP was 63–89, with a mean of76 years. Appropriate ethical approval has been obtained fromthe Ethic's Committee, Lund University, and we have adheredto the Helsinki Declaration. Analyses of miR-34a/c levels byqRT-PCR were previously described in (17).

Immunohistochemistry and scoring of AR on prostatetissue slides

Detection of AR in the prostate tissue sections was done byDako EnVision FLEX detection system (Dako) on a DakoAutostainer. Briefly, 4-mm sections were deparaffinized withxylene, rehydrated through a graded series of ethanol followedby heat-induced epitope retrieval using EnVision FLEX targetretrieval solution high pH. Endogenous peroxidase wasblocked by EnVision FLEX peroxidase-blocking reagent. AR(#AR 441, NeoMarkers, Thermo Scientific) was used as aprimary antibody. Detection was done by Dako EnVisionFLEX/HRP detection system using EnVision FLEX DABþChromogen, diaminobenzidine (DAB) solution and EnVisionFLEX substrate buffer. The slides were mounted with organo/limonene Mount (Santa Cruz Biotechnology). The stainedsections were scored independently, based on the AR stainingintensity, by 1 pathologist and 2 researchers. Overall stainingintensity was scored as 0 (negative), 1 (weak), 2 (moderate tostrong), and 3 (intense). For each patient, the AR intensity wasscored in malignant epithelium, benign epithelium andstroma.

Cell viability assaysMDA-PCa-2b cells in 5% charcoal-stripped serum were

reverse-transfected with 20 nmol/L miRNAs and treated with0.1 nmol/L R1881 (NLP005005, Perkin Elmer Inc). Control cellswere treated with 10 mmol/L Casodex (Bicalutamide, B9061,Sigma-Aldrich ). Viability was measured 7 days after trans-fection using CellTiter-Glo Luminescence Assay (Promega)with Envision Plate-reader (Perkin Elmer Inc.). All measure-ments were done in quadruplicates with 2 biological repeti-tions (n¼ 8). Students t-test was used to determine P-values.

Results

Systematic analyses of miRNA impacting on AR proteinlevels

We conducted LMA screens to identify miRNAs that influ-ence protein levels of AR in PCa cell lines (LNCaP, LAPC-4,CWR-R1, 22Rv1, MDA-PCa-2b). The cells were transfectedwith 2 miRNA overexpression libraries containing a total of1129 miRNAs. We aimed at identifying miRNAs that affect ARin all 5 cell lines (Fig. 1A). The results were subjected to rank-based meta-analysis using the RankProd function from the R/Bioconductor (20). We identified a total of 77 miRNAs, ofwhich 71 were unique mature miRNA sequences influencingthe levels of AR in all 5 cell lines (Fig. 1B, SupplementaryTable IV), 52 decreasing and 19 increasing AR protein.

Characterization of the AR 30UTR lengthNext, we studied whether these miRNAs were predicted to

bind to the AR 30UTR. The prediction programs such asTargetScan and microRNA.org use RefSeq sequences(http://www.ncbi.nlm.nih.gov/RefSeq/) in their predictions(11, 23). The RefSeq for AR (NM_0000444) has a transcriptlength of 4314 bp and a 30UTR of 436 bases. However, a muchlonger transcript with a length ranging from 4.7 to 11 kb hasbeen detected (24, 25). To investigate the length of the AR30UTR, we obtained RNA-Sequencing data from untreatedVCaP cells which have a very high expression of AR (26,27). The 8 exons present in our sequencing analysis producea transcript >10 kb matching the sequence of the transcriptAR-001 (ENST00000374690, http://www.ensembl.org). Speci-fically, the last exon (Fig. 2A, right panel) is longer thanexpected from the RefSeq annotation (NM_000044) and inthe transcript AR-001 (ENST00000374690) ends at position 66950 461 in GCRh37/hg19. Multiple short reads in the RNA-Sequencing data end around position 66 950 388 (data notshown), thereby producing a 30UTR region of approximately6680 b in VCaP cells. The presence of the long 30UTR wasconfirmed by RT-PCR in VCaP, LNCaP, LAPC-4, 22Rv1, andMDA-PCa-2b cells (Supplementary Figure 1). This vastlyincreases the region for putative miRNA regulation of theAR transcript. Intrigued by this finding, we returned to ourRNA-Sequencing data to study whether additional 30UTRregions in our data were extended compared with currentannotations. Interestingly, the data contained a total of 63330UTR regions that were calculated to be at least 100 b longercompared with current annotations in the Ensembl database(Supplementary Table V).

Custom prediction of miRNA-target interactions on theAR 30UTR

The long AR 30UTR sequence was analyzed for putativemiRNA binding sites using the TargetScan algorithm (v. 5.0)(11) resulting in a total of 550 predicted binding sites where664 human miRNAs were predicted to bind (SupplementaryTable VI). Since predictions contain a vast amount of falsepositives, we compared these predictions to the results fromthe LMA screen (Fig. 1B). Thirty-five uniquemiRNAs predictedto bind one or several sites changed the protein levels of AR inthe LMA screen (Supplementary Table VII). This number isprobably an underestimation since TargetScan ignores starmiRNAs, which were not included in our prediction. In gen-eral, the miRNAs predicted to target the AR 30UTR wereenriched among the LMA results (Fig. 2B). The 2 sampleWilcoxon test P-value for the enrichment was highly signifi-cant as calculated for all cell lines at the 48 hours and 72 hourstimepoint (P ¼ 7.5e�06). Importantly, most of the predictedmiRNA regulation on the AR 30UTR region occurs on theextended 30UTR region and would not have been detectedusing the RefSeq annotation (NM_000044).

Validation of miRNAs decreasing ARTo validate the top ranking miRNAs decreasing AR protein

and predicted to bind to the 30UTR, we reverse-transfected 21miRNAs (Table I) into LNCaP and 22Rv1 cells, which are

Östling et al.

Cancer Res; 71(5) March 1, 2011 Cancer Research1958




Figure 1. LMA screen for miRNAs targeting AR. A Overexpression of 1129 mature human miRNAs in PCa cells by reverse-transfection on 384-wellplates. Screens were done in duplicate and with 2 time points (48 hours and 72 hours) for 5 AR positive PCa cell lines grown in full media. The lysates wereprinted on nitrocellulose-covered glass slides and stained with AR antibody and Sypro Ruby Blot solution to detect total protein content. Rank-basedmeta-analysis and correction for multiple testing (q < 0.01) was used to find miRNAs that regulate AR in all 5 cell lines. B For visualization, data werestandardized using median and MAD from R/Bioconductor (v.2.9) with default options. Red denotes increases in AR, whereas blue represents decrease. Thescale indicates Z-score standardized values (þ/�4 SD).






optimally transfectable PCa cell lines, and analyzed AR byWestern blotting (WB) and qRT-PCR (Fig. 3A and B). Our WBresults show that all selected miRNAs decreased AR proteinwhen compared with scrambled controls (Fig. 3A). The LNCaPcells were analyzed for PSA levels and most of the AR-decreasing miRNAs downregulated the PSA protein(Fig. 3A, left panel). In contrast, overexpression of miR-299–

3p, which clearly downregulated AR, led to an increase in PSAprotein levels in LNCaP cells. MiR-30d, predicted to target ARby TargetScan increased AR and PSA protein in LNCaP cells,an effect also observed in the LMA screen. Although small inmagnitude, the decrease of AR with miR-9 transfection inLNCaP was consistently observed in repeated experiments.The 22Rv1 cells express an AR isoform produced by alternative

Figure 2. Characterization of the AR 30UTR by RNA-Sequencing. A RNA-Sequencing data from untreated VCaP cells on the AR locus. Sequencingcoverage (red dots) was normalized by the number of uniquely mapped short reads. The boxes below the coverage plots represent exons in the AR transcript,exon 2 being present only in a shorter transcript variant (AR-201: ENST00000396043, NM_001011645). On the right a close-up of the last exonshowing coverage for a much longer sequence of the 30UTR (6680 b) than the RefSeq annotation (NM_000044, 436 b). B A histogram of LMA screen hits,which are either predicted (red line) or not predicted (blue line) to target AR. Significant enrichment of the predicted miRNA hits (P-value 7.5e-06) wasobserved with 2 sample Wilcocon test.

Östling et al.





splicing (28, 29). This alternative transcript is downregulatedby miR-135b, miR-147, miR-299–3p, miR-34a, and miR-644(Fig. 3A, right panel).Next, we conducted qRT-PCR to investigate the effect of

miRNAs on ARmRNA (Fig. 3B). Compared with the scrambledcontrol, miR-147, miR-297, miR-298, miR-299–3p, miR-421,miR-449a reduced AR mRNA levels in both LNCaP and22Rv1 cells. Cell-type specific effects were observed formiR-371–3p, miR-449b and miR-491–5p, which were moreefficient in downregulating ARmRNA in LNCaP cells whereasmiR-876–3p had this effect only in 22Rv1 cells. No clear cell-type specific differences were evident for these miRNAs at theprotein level (Fig. 3A). On the other hand, miR-488 had a cell-type specific effect at the protein level and displayed only aminor difference at the mRNA level. Taken together, thesedata highlight distinct patterns of miRNA regulation. Theresults in Fig. 3B suggest that miR-644 increased AR mRNA.The data were normalized to GAPDH andmiR-644 is predictedby TargetScan to target this gene. We noticed that miR-644decreases GAPDH mRNA levels (data not shown) and we thusused b-actin for normalization for miR-644 (Fig. 3D, rightpanel) to show that miR-644 decreases AR mRNA by 20%.

Thirteen miRNAs decrease AR 30UTR-containingreporter activityTo investigate whether the 21 miRNAs interact with the AR

30UTR region, we cloned 7 different fragments covering thepredicted areas of interaction in luciferase reporter constructs(Fig. 4A). The reporters were transfected into 22Rv1 cellstogether with the miRNAs and a Renilla luciferase control.

Transfection of miR-135b, miR-185, miR-297, miR-299–3p,miR-34a, miR-34c, miR-371–3p, miR-421, miR-449a, miR-449b, miR-634, miR-654–5p, and miR-9, significantly inhibitedthe luciferase activity of the constructs containing their pre-dicted binding sites, as compared with scrambled transfectedcontrols (Fig. 4B). Surprisingly miR-147, -491–5p, -644, and-876 increased the reporter activity which is most likely due toindirect effects. Our results confirmed that 13 miRNAs wereable to interact with the extended 30UTR of the AR gene. Todetermine whether the 30UTR is the main regulatory region forthese miRNAs, we overexpressed AR lacking the 30UTR in PC-3cells that are devoid of endogenous AR (Supplementary Figure2). Our WB results showed that the majority of the 13 miRNAsexert regulation on AR primarily through the 30UTR whereasmiR-421, miR-449a, miR-449b, and miR-9 are also able todecrease the levels of the exogenous AR.

miRNA expression in prostate cell linesWe analyzed the expression of the AR targeting miRNAs in

PCa cells and in nonmalignant cells. The data were normal-ized and SAM analysis (30) was used to find differentiallyexpressed miRNAs between the malignant and control celllines. This analysis revealed miR-34c as having higher expres-sion in the nonmalignant lines as compared with the malig-nant ones (Fig. 5A, Supplementary Figure 3).

Inverse correlation of miR-34a and miR-34c expressionto AR protein levels in malignant prostate epithelium

Reduced miR-34c expression has previously been linked toPCa aggressiveness, WHO grade, PSA levels, and occurrence

Table 1.

Name Type RP/Rsum* FC FDR q-value Predicted binding sites

1 hsa-miR-299–3p down 9.66 0.62 0.000 8mer2 hsa-miR-644 down 14.83 0.61 0.000 8mer3 hsa-miR-297 down 14.93 0.63 0.000 7mer-m8 (x3), 8mer4 hsa-miR-491–5p down 18.66 0.65 0.000 7mer-m8 (x4), 7mer-1A (x2)5 hsa-miR-9 down 23.95 0.67 0.000 8mer (x2), 7mer-m8 (x2), 7mer-1A (x2)6 hsa-miR-147 down 25.19 0.65 0.000 8mer, 7mer-1A (x2)7 hsa-miR-488 down 25.88 0.68 0.000 7mer-m8, 7mer-1A8 hsa-miR-876–3p down 27.96 0.70 0.000 7mer-1A9 hsa-miR-185 down 46.33 0.73 0.000 8mer, 7mer-m8 (x3)10 hsa-miR-298 down 47.07 0.72 0.000 8mer, 7mer-m8 (x5), 7mer-1A (x4)11 hsa-miR-421 down 50.88 0.71 0.001 8mer (x2), 7mer-m8, 7mer-1A (x3)12 hsa-miR-541 down 53.05 0.73 0.001 8mer, 7mer-1A13 hsa-miR-449b down 58.43 0.75 0.001 8mer14 hsa-miR-34a down 59.17 0.77 0.001 8mer15 hsa-miR-371–3p down 59.58 0.77 0.001 7mer-1A (x2)16 hsa-miR-135b down 60.15 0.76 0.001 7mer-m817 hsa-miR-634 down 66.50 0.75 0.001 7mer-m818 hsa-miR-449a down 67.82 0.75 0.001 8mer19 hsa-miR-654–5p down 71.13 0.76 0.001 7mer-1A20 hsa-miR-631 down 71.16 0.75 0.001 7mer-m821 hsa-miR-34c down 83.57 0.77 0.001 8mer

*RP/Rsum; RankProduct/Ranksum, FC: Fold Change, FDR q-value; False Discovery Rate q-value.






Figure 3. Western blot and qRT-PCR analysis of miRNAs decreasing AR. A miRNAs were transfected into LNCaP (left panel) and 22Rv1 (right panel)cells grown in the presence of 10% FBS, lysed (72 hours) and analyzed by WB using AR, PSA and b-actin antibodies. B AR mRNA levels wereanalyzed by qRT-PCR 24 h after transfection and normalized to GAPDH or ACTB levels. The bars represent averages of 2 biological and 2 technicalrepeats (�SEM, n ¼ 4) compared with scrambled (scr) controls which were arbitrarily set to 1.

Östling et al.





of metastases (17). Therefore, we analyzed whether miR-34cand miR-34a inversely correlate with AR levels in PCapatients. The levels of miR-34c and miR-34a were measuredwith qRT-PCR in 47 prostatic tumors. The AR proteincontent was determined by immunostaining and the stain-ing was scored by overall intensity (1; weak, 2; moderate tostrong, 3; intense, Figure 5B, lower panel). When correlating

miR-34c expression levels in the malignant epithelial cellswith the semiquantitative assessment of AR immunostain-ing, we found a statistically significant inverse association(Fig. 5B, left panel, P¼ 0.0082 for intensity score 1þ 2 versus3, and P ¼ 0.0046 when comparing intensity 2 versus 3;Mann–Whitney test). No correlation was found for miR-34cexpression and AR staining in the benign epithelium nor in

Figure 4. Luciferase reporter assay detection of miRNA-AR 30UTR interaction. A Schematic diagram of the short AR 3’UTR in RefSeq (436 b, NM_000044)and the long (6680 b, ENST00000374690). Short black lines (#1–7) denote the fragments cloned into pMIR-REPORT Luciferase vector and miRNAnames below denote the predicted binding. B Untreated 22Rv1 cells were cotransfected with the reporter construct and miRNAs, and the luciferase activitywasmeasured after 24 hours incubation with Dual-Glo Luciferase assay. Firefly luciferase was normalized to Renilla luciferase. The data shown are averages ofat least 4 separate experiments, each conducted in triplicate (�SD, *P < 0.05, **P < 0.01).






stromal tissue (data not shown). There was also a significantinverse correlation between AR staining intensity and miR-34a expression (Figure 5B, right panel, P ¼ 0.0085 forintensity score 1 þ 2 versus 3, and P ¼ 0.0058 whencomparing intensity 2 versus 3; Mann–Whitney test). Thesedata together with the results from Figure 4, showinginteraction of miR-34a/c on the AR 30UTR, suggest thatthese miRNAs are important regulators of AR levels duringPCa progression.

AR-inhibiting miRNAs reduce androgen-induced PCacell viability

To determine how androgen-induced proliferation is influ-enced by the miRNAs decreasing AR, we transfected the 21validated miRNAs into MDA-PCa-2b cells, treated these with asynthethic androgen analog R1881 and measured viability7 days later. Casodex was able to reduce proliferation ofMDA-PCa-2b cells by 60% whereas a positive control siRNA(siCellDeath) showed a reduction of up to 80% (Fig. 6). The

Figure 5. miRNA expression inprostate cell lines and prostatemalignant epithelium. A Heatmapdisplaying expression of 17 AR-regulating miRNAs. Total RNAwas isolated from untreatedprostate cells and profiled withAgilent humanmiRNAmicroarrays(V1; 455 miRNAs). The data werenormalized, and differentiallyexpressed miRNAs in malignantversus nonmalignant cell lines(PWR1E, EP156T, RWPE1) weredefined using SAM. Hierarchicalclustering of the data were doneusing the Mev 4.0 software (red:up, green: down). B Inversecorrelation of miR-34c and miR-34a expression and AR leves inprostate malignant epithelium.miR-34c and miR-34a weredetected with qRT-PCR in 47prostatic tissues obtained byTURP. The AR protein content inmalignant epithelium (lower panel)was determined byimmunostaining and scored byoverall intensity (Score: 1 weak;2 moderate to strong; and 3intense). Statistically significantinverse correlation was observedfor miR-34c (P¼ 0.0082) and miR-34a (P ¼ 0.0085) when comparedwith AR intensity score 1 þ 2versus 3 (Mann–Whitney test).

Östling et al.





siRNAs against AR decreased viability by 20%–40%, indicatingthat this cell line is not completely dependent on AR forsustaining cell growth. Most miRNAs (15/21) reduced viabilityto at least to the same extent as the AR siRNAs (Fig. 6). Thissuggests that miRNAs present a novel and potent way ofinfluencing not only AR levels but also the more complexphenomena of androgen-induced prostate cell proliferation.

Discussion

We report here a systematic gain-of-function screening todiscover AR regulating miRNAs in PCa cell lines. By focusingon miRNAs that influence AR across all the cell lines, wereduced cell-type specific bias, and identified 71 unique miR-NAs influencing AR protein levels, 52 decreasing and 19increasing. To date there is only one report, published duringrevision of this manuscript, showing miR-488* regulation ofAR through the proximal 30UTR region (31). Thus, our screenvastly enhances the understanding on the role of posttran-scriptional regulation of this key player in PCa.The 30UTR sequence of a gene is often considered the

primary region of miRNA regulation (10), although regulatorysites may exist also in other regions of the transcript (32–36).Our RNA-Sequencing results show that for hundreds of genes,including the AR, the 30UTR regions are much longer thancurrently annotated in the genome databases. Therefore, thecommonly used miRNA target prediction algorithms will misssuch miRNA regulation. For example, in the case of the AR,only the short 436 b 30UTR is used in such analyses, while ourcustom prediction, validated by experimental data, was basedon the 6680 b 30UTR region. Understanding of miRNA reg-ulatory networks in cancer will thus be critically dependent onbetter characterization of 30UTRs. Interestingly, a recent RNA-sequencing study in C.elegans resulted in a revision of approxi-mately 40% of gene models (37).Our study is also one of only a handful of reports showing

comprehensive analysis of multiple miRNAs targeting the

same gene (18, 38). All the selected miRNAs from the LMAscreen were validated in LNCaP and 22Rv1 cells, showing therobustness of the LMA technology. Only 6 of the miRNAsdecreased the mRNA significantly after 24 hours suggestingthat most miRNAs decrease AR protein through translationalinhibition.

The introduction of miRNA mimics made it possible for usto systematically identify miRNAs capable of regulating AR.However, to explore, whether endogenous levels of thesemiRNAs impact on AR protein levels, we conducted expres-sion analyses of 13 different prostate cell lines and 47 clinicaltumors. These analyses showed an inverse correlation inclinical patient samples between the AR levels and miR-34aandmiR-34c expression validating the functional cell line data.Previously, reduced miR-34c expression has been linked toPCa aggressiveness (17).

MiR-34a and miR-34c belong to a family of evolutionaryconservedmiRNAs that are regulated by the tumor-suppressorTP53 (39, 40).mir-34a is encoded by its own gene, whereasmir-34b andmir-34c share a common primary transcript. Althoughthe seed of miR-34b is very similar to miR-34c, we did notobserve any effect of miR-34b in our LMA screen.Mir-34 geneshave been suggested to be potent mediators of tumor suppres-sion by p53 and are implicated in the negative control of the cellcycle, senescence, and apoptosis (39). Our results on miR-34aand miR-34c downregulating AR, represent an additionalexplanation by which PCa cells could benefit from reducedexpression of these cancer-relevant miRNA genes.

Fifteen AR-regulating miRNAs were potent regulators ofandrogen-induced proliferation of PCa cells. In our analyses,miR-541, miR-631, miR-634, miR-644, miR-654–5p, and miR-876–3p were more efficient inhibitors of cell growth than ARsiRNAs or treatment with Casodex. The strong effects of thesemiRNAs suggest that they may target additional importantplayers in PCa proliferation.

Almost 70 years after the initial finding of the importantrole of androgen in PCa (41), reducing AR activity by anti-

Figure 6. The effect of miRNAs on the androgen-induced cell proliferation. MDA-PCa-2b cells were reverse-transfected with miRNAs, treated with thesynthetic androgen analog R1881 (0.1 nmol/L) and cell viability was measured 7 days later using CellTiter-Glo Luminescence Assay. Assays were done with 2biological replicates and the relative cell viability was calculated as compared with the scrambled (100%) (mean� SEM, n¼ 8, *P < 0.05, **P < 0.01, StudentsT-test). Casodex (Cas) and siCellDeath were used as positive controls.






androgen remains the primary treatment mode for advancedPCa. Recently, novel drugs aimed at targeting androgensignaling have been reported including abiraterone,RD162, and MDV3100 (42–44). An important feature ofMDV3100 a second-generation antiandrogen is that itinhibits AR activity despite increased levels of AR (44).The miRNAs identified in our study reduced AR and wereable to efficiently reduce androgen-induced proliferation. Inconclusion, our results show that miRNAs binding to thelong 30UTR of the AR gene need to be considered as onemechanism for how PCa cells regulate the levels of AR. Thisprovides future opportunities and starting points to explorethe applications of miRNAs and their derivatives in PCatherapy.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

Matthias Nees and Astrid Murum€agi are thanked for preparing VCaP cells forRNA-Sequencing analysis. Saija Haapa-Paananen, Maija Wolf and AgilentTechnologies are acknowledged for help in miRNA expression analyses. EliseNilsson is thanked for help in the PCa immunostaining and Pirjo K€apyl€a for helpin the viability assays.

Grant Support

This work was financially supported by ProspeR EU-FP7 FP7 (HEALTH-F2–2007-201438), Academy of Finland Centre of Excellence in "TranslationalGenome-Scale Biology" Finnish Cancer Society, Sigrid Juselius Foundation,Swedish Research Council for Medicine and personal grants from Academyof Finland to P€O, SKL, and SK.

The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received July 9, 2010; revised November 26, 2010; accepted December 14,2010; published OnlineFirst February 22, 2011.

References1. Holzbeierlein J, Lal P, LaTulippe E, Smith A, Satagopan J, Zhang L,

et al. Gene expression analysis of human prostate carcinoma duringhormonal therapy identifies androgen-responsive genes andmechan-isms of therapy resistance. Am J Pathol 2004;164:217–27.

2. Heinlein CA, Chang C. Androgen receptor in prostate cancer. EndocrRev 2004;25(2):276–308.

3. Chen Y, Sawyers CL, Scher HI. Targeting the androgen receptorpathway in prostate cancer. Curr Opin Pharmacol 2008;8:440–8.

4. Feldman BJ, Feldman D. The development of androgen-independentprostate cancer. Nat Rev Cancer 2001;1:34–45.

5. Hobisch A, Culig Z, Radmayr C, Bartsch G, Klocker H, Hittmair A.Distant metastases from prostatic carcinoma express androgenreceptor protein. Cancer Res 1995;55:3068–72.

6. Chen CD, Welsbie DS, Tran C, Baek SH, Chen R, Vessella R,Rosenfeld MG, et al. Molecular determinants of resistance to anti-androgen therapy. Nat Med 2004;10:33–9.

7. Yu J, Yu J, Mani RS, Cao Q, Brenner CJ, Cao X, et al. An integratednetwork of androgen receptor, polycomb, and TMPRSS2-ERG genefusions in prostate cancer progression. Cancer Cell 2010;17:443–54.

8. Wang Q, Li W, Zhang Y, Yuan X, Xu K, Yu J, et al. Androgen receptorregulates a distinct transcription program in androgen-independentprostate cancer. Cell 2009;138:245–56.

9. He L, Hannon GJ. MicroRNAs: small RNAs with a big role in generegulation. Nat Rev Genet 2004;5:522–31.

10. Bartel DP. MicroRNAs: target recognition and regulatory functions.Cell 2009;136:215–33.

11. Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalianmRNAs are conserved targets of microRNAs. Genome Res 2009;19:92–105.

12. Esquela-Kerscher A, Slack FJ. Oncomirs – microRNAs with a role incancer. Nat Rev Cancer 2006;6:259–69.

13. Calin GA, Croce CM.MicroRNA signatures in human cancers. Nat RevCancer 2006;6:857–66.

14. Ryan BM, Robles AI, Harris CC. Genetic variation in microRNA net-works: the implications for cancer research. Nat Rev Cancer 2010;10(6):389–402.

15. Iorio MV, Croce CM. MicroRNAs in cancer: small molecules with ahuge impact. J Clin Oncol 2009;27:5848–56.

16. Gandellini P, Folini M, Zaffaroni N. Towards the definition of prostatecancer-related microRNAs: where are we now?Trends Mol Med2009;15:381–90.

17. Hagman Z, Larne O, Edsjo A, Bjartell A, Ehrnstr€om RA, Ulmert D, et al.miR-34c is down regulated in prostate cancer and exerts tumorsuppressive functions. Int J Cancer 2010;127:2768–76.

18. Leivonen SK, Makela R, Ostling P, Kohonen P, Haapa-Paananen S,Kleivi K, et al. Protein lysate microarray analysis to identify microRNAsregulating estrogen receptor signaling in breast cancer cell lines.Oncogene 2009;28:3926–36.

19. Boutros M, Bras LP, Huber W. Analysis of cell-based RNAi screens.Genome Biol 2006;7:R66.1–11.

20. Hong F, Breitling R, McEntee CW, Wittner BS, Nemhauser JL, ChoryJ. RankProd: a bioconductor package for detecting differentiallyexpressed genes in meta-analysis. Bioinformatics 2006;22:2825–7.

21. Durinck S, Bullard J, Spellman PT, Dudoit S. GenomeGraphs: inte-grated genomic data visualization with R. BMC Bioinformatics 2009;6(10):1–9.

22. Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M.Variance stabilization applied to microarray data calibration and to thequantification of differential expression. Bioinformatics 2002;18Suppl1:S96–104.

23. Betel D, Wilson M, Gabow A, Marks DS, Sander C. The microRNA.orgresource: targets and expression. Nucleic Acids Res 2008;36:D149–53.

24. Trapman J, Klaassen P, Kuiper GG, van der Korput JA, Faber PW, vanRooij HC, et al. Cloning, structure and expression of a cDNA encodingthe human androgen receptor. Biochem Biophys Res Commun1988;153:241–8.

25. Lubahn DB, Joseph DR, Sar M, Tan J, Higgs HN, Larson RE, et al. Thehuman androgen receptor: complementary deoxyribonucleic acidcloning, sequence analysis and gene expression in prostate. MolEndocrinol 1988;2:1265–75.

26. Korenchuck S, Lehr JE, MClean L, Lee YG, Whitney S, Vessella R,et al. VCaP, a cell-based model system of human prostate cancer. InVivo 2001;(15):163–63

27. Makkonen H, Kauhanen M, J€a€askel€ainen T, Palvimo JJ. Androgenreceptor amplification is reflected in the transcriptional responses ofVertebral-Cancer of the Prostate cells. Mol Cell Endocrinol 2010; inpress doi:10.1016

28. Guo Z, Yang X, Sun F, Jiang R, Linn DE, Chen H, et al. A novelandrogen receptor splice variant is up-regulated during prostatecancer progression and promotes androgen depletion-resistantgrowth. Cancer Res 2009;69:2305–13.

29. Dehm SM, Schmidt LJ, Heemers HV, Vessella RL, Tindall DJ. Splicingof a novel androgen receptor exon generates a constitutively activeandrogen receptor that mediates prostate cancer therapy resistance.Cancer Res 2008;68:5469–77.

30. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarraysapplied to the ionizing radiation response. Proc Natl Acad Sci U S A2001;98:5116–21.

Östling et al.





31. Sikand K, Slaibi JE, Singh R, Slane SD, Shukla GC. miR 488* inhibitsandrogen receptor expression in prostate carcinoma cells. Int JCancer 2010, in press, doi 10.1002/ijc.25753

32. Grey F, Tirabassi R, Meyers H, Wu G, McWeeney S, Hook L, et al. Aviral microRNA down-regulates multiple cell cycle genes throughmRNA 5’UTRs. PLoS Pathog 2010;6:e1000967.1–9.

33. Lytle JR, Yario TA, Steitz JA. Target mRNAs are repressed as effi-ciently by microRNA-binding sites in the 50 UTR as in the 30 UTR. ProcNatl Acad Sci U S A 2007;104:9667–72.

34. Orom UA, Nielsen FC, Lund AH. MicroRNA-10a binds the 5’UTR ofribosomal protein mRNAs and enhances their translation. Mol Cell2008;30:460–71.

35. Lee I, Ajay SS, Yook JI, Kim HS, Hong SH, Kim NH, et al. New class ofmicroRNA targets containing simultaneous 50-UTR and 30-UTR inter-action sites. Genome Res 2009;19:1175–83.

36. Spizzo R, Nicoloso MS, Lupini L, et al. miR-145 participates withTP53 in a death-promoting regulatory loop and targets estrogenreceptor-alpha in human breast cancer cells. Cell Death Differ2010;17:246–54.

37. Mangone M, Manoharan AP, Thierry-Mieg D, Thierry-Mieg J, Han T,Mackowiak SD, et al. The landscape of C. elegans 3’UTRs. Science2010;329:432–5.

38. Wu S, Huang S, Ding J, Zhao Y, Liang L, Liu T, et al. MultiplemicroRNAs modulate p21Cip1/Waf1 expression by directly targetingits 3’ untranslated region. Oncogene 2010;29:2302–8.

39. Hermeking H. The miR-34 family in cancer and apoptosis. Cell DeathDiffer 2010;17:193–9.

40. Hermeking H. p53 enters the microRNA world. Cancer Cell 2007;12:414–8.

41. Huggins C, Hodges CV. The effect of castration, of estrogen and ofandrogen injection on serum phosphatases in metastatic carcinomaof the prosate. 1941;1:293–7.

42. Ryan CJ, Smith MR, Fong L, Rosenberg JE, Kantoff P, Raynaud F,et al. Phase I clinical trial of the CYP17 inhibitor abiraterone acetatedemonstrating clinical activity in patients with castration-resistantprostate cancer who received prior ketoconazole therapy. J ClinOncol 2010;28:1481–8.

43. Reid AH, Attard G, Danila DC, Oommen NB, Olmos D, Fong PC , et al.Significant and sustained antitumor activity in post-docetaxel, castra-tion-resistant prostate cancer with the CYP17 inhibitor abirateroneacetate. J Clin Oncol 2010;28:1489–95.

44. Tran C, Ouk S, Clegg NJ, Chen Y, Watson PA, Arora V, et al.Development of a second-generation antiandrogen for treatment ofadvanced prostate cancer. Science 2009;324:787–90.






2011;71:1956-1967. Published OnlineFirst February 22, 2011.Cancer Res Päivi Östling, Suvi-Katri Leivonen, Anna Aakula, et al. Receptor in Prostate Cancer CellsSystematic Analysis of MicroRNAs Targeting the Androgen

Updated version

10.1158/0008-5472.CAN-10-2421doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2011/02/23/0008-5472.CAN-10-2421.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/71/5/1956.full#ref-list-1

This article cites 41 articles, 12 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/71/5/1956.full#related-urls

This article has been cited by 10 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

SubscriptionsReprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://cancerres.aacrjournals.org/content/71/5/1956To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-10-2421

http://cancerres.aacrjournals.org/content/suppl/2011/02/23/0008-5472.CAN-10-2421.DC1

http://cancerres.aacrjournals.org/content/71/5/1956.full#ref-list-1

http://cancerres.aacrjournals.org/content/71/5/1956.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/71/5/1956


Systematic Analysis of MicroRNAs Targeting the Androgen...

Documents

Transcript of Systematic Analysis of MicroRNAs Targeting the Androgen...