Recent genome-wide association studies reported strong and reproducible associations of multiple genetic variants in a large “gene-desert” region of chromosome 8q24 with susceptibility to prostate cancer (PC). However, the causative or functional variants of these 8q24 loci and their biological mechanisms associated with PC susceptibility remain unclear and should be investigated. Here, focusing on its most centromeric region (so-called Region 2: Chr8: 128.14-128.28 Mb) among the multiple PC loci on 8q24, we performed fine mapping and re-sequencing of this critical region and identified SNPs (single nucleotide polymorphisms) between rs1456315 and rs7463708 (chr8: 128,173,119-128,173,237 bp) to be most significantly associated with PC susceptibility (P = 2.00 × 10−24, OR = 1.74, 95% confidence interval = 1.56–1.93). Importantly, we show that this region was transcribed as a ∼13 kb intron-less long non-coding RNA (ncRNA), termed PRNCR1 (prostate cancer non-coding RNA 1), and PRNCR1 expression was upregulated in some of the PC cells as well as precursor lesion prostatic intraepithelial neoplasia. Knockdown of PRNCR1 by siRNA attenuated the viability of PC cells and the transactivation activity of androgen receptor, which indicates that PRNCR1 could be involved in prostate carcinogenesis possibly through androgen receptor activity. These findings could provide a new insight in understanding the pathogenesis of genetic factors for PC susceptibility and prostate carcinogenesis. (Cancer Sci 2011; 102: 245–252)
Prostate cancer (PC) is the most common malignancy in males and the second-leading cause of cancer-related deaths in Western countries(1,2). The prevalence of PC is markedly different among different ethnic groups; the African–Americans have the highest rate (137.0 cases per 100 000 people) and the lowest prevalence is observed in the Asian population (<10 cases per 100 000 people), indicating that the difference in genetic background contributes to PC susceptibility(2). Although several other factors such as PC screening methods and environmental and hormonal factors could also be attributed to this racial/ethnic difference, a large body of evidence supports the view that genetics plays a critical role in PC susceptibility(3). Recent genome-wide association studies (GWAS) reported strong and consistent associations of multiple genetic variants on chromosome 8q24 with PC susceptibility(4–10). This region contains multiple PC susceptibility loci within a ∼1 MB segment. A meta-analysis of the chromosome 8q24 region suggests that SNPs of these loci resulted in a 15–40% increased risk of PC among individuals of European and African descents; however, no gene has been identified in this region. Myc oncogene is located at approximately 200-kb downstream, and recent studies indicated that one of the loci at 8q24 (rs6983267) could be associated with the Wnt signal pathway in colorectal cancer and possibly with Myc expression(11,12). However, they did not show that the associated SNP in 8q24 could affect MYC expression in colorectal cancer or prostate cancer(11,13), and their association with Myc expression is still not conclusive. Therefore, the biological mechanisms of the 8q24 loci associated with PC susceptibility remains unclear.
Among these susceptible loci at 8q24, previous detailed analysis found that a locus denoted as Region 2 (Chr8: 128.14-128.28 Mb), which is independently located more than 200-kb apart from other candidate loci (Regions 1, 3 and others, ref. 4–10), showed the strongest association in Japanese–Americans and African–Americans(7,14). Therefore, we performed re-sequencing and fine mapping of 8q24 Region 2 and identified that the association peak was mapped within a small region of ∼13 kb (128.16–128.175 Mb) in the Japanese population. Here we demonstrate that this region was transcribed, yielding a long non-coding RNA (ncRNA), and its expression was upregulated in PC cells and its precursor lesions. Functionally, we show that this ncRNA could be involved in PC growth in part through the androgen receptor (AR)-mediated pathway. These findings could provide new insight into the understanding of genetic factors on chromosome 8q24 with PC susceptibility and prostate carcinogenesis.
Materials and Methods
Genotyping samples. All PC cases and controls were obtained from the BioBank Japan at the Institute of Medical Science, the University of Tokyo. This project, to collect a total of 300 000 cases who have at least one of 47 diseases, was started in 2003 by a collaborative network of 66 hospitals located throughout Japan(15) (http://biobankjp.org/). The registration of cases was based on diagnoses made by physicians at the affiliated hospitals. Among the PC samples in the BioBank Japan, we selected 1504 PC patients who were pathologically diagnosed as prostate adenocarcinoma by biopsy or surgery. Then, 1554 individuals with several diseases, except prostate cancer, were randomly selected as control samples among the patients without PC from the Biobank Japan. This project was approved by the ethical committees at the Institute of Medical Science, the University of Tokyo and the Center for Genomic Medicine, Center for Genomic Medicine, RIKEN.
Fine mapping and re-sequencing. We performed fine mapping using the multiplex PCR-based Invader assay(16) (Third Wave Technologies, Madison, WI, USA) or direct sequencing. In the Invader assay, all genotypes were called by visual inspection, and we determined genotype success as fewer than 10 undetermined samples in a 384-well plate. When we failed to genotype more than one 384-well plate in a total of 33 plates, we excluded the SNP from further analyses. Overall, the genotype success rate was 99.3% (99.5% in cases and 99.2% in controls, respectively). To validate the genotyping data, we genotyped 10 SNP in 94 subjects using direct sequencing, and the concordance rate was 100%. Re-sequencing of candidate regions was performed in 94 PC cases using an ABI3730 DNA Analyzer (Applied Biosystems, Foster City, CA, USA). To perform fine mapping, we selected tag SNPs from phase II of the HapMap JPT data using the pairwise tagging method with the following criteria: r2 > 0.9, minor allele frequency >5% and call rate >75%. We assessed case–control association analysis using a 1-degree-of-freedom Cochrane–Armitage trend test. The LD and haplotype analysis was performed with the use of Haploview (Broad Institute, Cambridge, MA, USA).
Cell lines. The PC cell lines LNCaP, 22Rv1, PC-3 and DU-145, and breast cancer cell lines HCC1937 were purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA). LNCaP-derived C4-2B was purchased from ViroMed Laboratories (Minnetonka, MN, USA). Normal prostate epithelial cell line PrEC was obtained from Cambrex (East Rutherford, NJ, USA). The PC cell lines were grown in Delbecco’s modified Eagle’s medium (DMEM; Invitrogen, Carlsbad, CA, USA); this media were supplemented with 10% FBS and 1% antibiotic/antimycotic solution (Sigma–Aldrich, St Louis, MO, USA). HCC1937 was grown in RPMI 1640 (Invitrogen) with 10% FBS and 1% antibiotic/antimycotic solution. PrEC was cultured with appropriate media according to the manufacture’s recommendation. Cells were maintained at 37°C in atmospheres of humidified air with 5% CO2.
Gene cloning in the critical region at 8q24 Region 2. Fine mapping and re-sequencing showed the 12-kb region between rs1016342 and rs7463708 was strongly associated with PC risk in the Japanese population. Since there is no annotated gene or transcript encoding any proteins in this region, we searched for any novel transcript by performing RT-PCR for several EST within this region. The RT-PCR connection of the candidate regions as well as 5′- and 3′-RACE using SAMRT RACE cDNA Amplification kit (Clontech, Mountain View, CA, USA) identified a novel intron-less transcript of more than 13 kb covering a region between rs1016342 and rs7463708. For this intron-less transcript, we designed six sets of primer: set 1 (5′-GGTGGGACTTTTAGGGGGTA-3′ and 5′-CCCTTTCCTGACCTGATTGA-3′); set 2 (5′-GCAGTGGCAGAGAATTATGAA-3′ and 5′-GTGGGCAGAGAGAGAAATGC-3′); set 3 (5′-CTGTACTGGGTTCTGGGCTAT-3′ and 5′-CAAGTTTGTTCAATGGACACC-3′); set 4 (5′-CCTCATGACCCAGTTGAAAAA-3′ and 5′-CAAGTTTGTTCAATGGACACC-3′), set 5 (5′-TTTACTGACCGCAACAACTTC-3′ and 5′-AGCTTCTGGTGACAAGGGATT-3′) and set 6 (5′-TGTACCCTTTCTGCAATAGGT-3′ and 5′-GCCCATTGGTCCATTATTTG-3′), and performed RT-PCR using cDNA of 22Rv1 cells with or without RT.
Northern blot analysis and real-time quantitative RT-PCR. We extracted total RNA from 22Rv1, LNCaP, C4-2B, DU145 and PC-3 cells using RNeasy Kit (QIAGEN, Valencia, CA, USA) and purified mRNA by mRNA Purification Kit (GE Healthcare, Piscataway, NJ, USA) according to the manufacturer’s protocols. One micrgram aliquot of each mRNA from the PC cell lines was separated on 1% denaturing agarose gels and transferred onto nylon membranes. The 929-bp probe specific to PRNCR1 was prepared by PCR using the primer 5′-GACCTATGTGATTGTGG-3′ and 5′-GTGGGCAGAGAGAGAAA-TGC-3′ (GenBank Accession no. AB458446). Hybridization with a random-primed α32P-dCTP-labeled probe was carried out according to the instructions for Megaprime DNA labeling system (GE Healthcare). The blots were auto-radiographed with intensifying screens at −80°C for 21 days. Purification of the PC cells and normal prostatic epithelial cells from frozen PC tissues was described previously(17). RNA from these samples were subjected to two rounds of T7-based RNA amplification (Epicentre Technologies, Madison, WI, USA) and subsequent synthesis of single-strand cDNA. We carried out quantitative real-time PCR using a LightCycler 480 with the LightCycler 480 probe master (Roche Applied Science, Indianapolis, IN, USA) in accordance with the manufacturer’s instructions. The primers for real-time quantitative PCR were 5′-CCAGGG-GGAAACACACAG-3′ and 5′-AAATGGCAGTTTCCTTCAATG-3′ for PRNCR1 (GenBank Accession no. AB458446), and 5′-TTGGCTTGACTCAGGATTTA-3′ and 5′-ATGCTATC-ACCTCCCCTGTG-3′ for β-actin (ACTB) for normalization of input cDNA. The RT-PCR exponential phase was determined to allow quantitative comparisons among cDNA developed from identical reactions.
siRNA duplex and cell viability assay. To investigate the biological function of PRNCR1 in PC cells, we designed several RNA duplexes to PRNCR1. The target sequences of these duplexes are: si1, 5′-GCUCUUAAGGAAAUAACUU-3′; si2, 5′-GACGGUUCUGUUUAAGUAA-3′; and siEGFP, 5′-GAAGCAGCACGACUUCUUC-3′ as a negative control. LNCaP, PC-3 or HCT1937 cells (3 × 104) were grown on 12-well plates, transfected with each of the siRNA duplexes using Lipofectamine RNAiMAX reagent (Invitrogen) according to the manufacturer’s instruction. The knockdown effect was measured by RT-PCR using the primer set: 5′-ATTGTTCCCAGTTCTCTGGAC-3′ and 5′-GGAGGAAGTGGAGTAGGTATAAGAG-3′. In the MTT assay, cell viability was measured using Cell-counting kit-8 (DOJINDO, Kumamoto, Japan) at 48 and 72 h after siRNA transfection. Absorbance was measured at 490 nm, and at 630 nm as a reference, with ARVO MX 1420 Multilabel counter (PerkinElmer, Walltham, MA, USA).
Dual luciferase assay for androgen receptor (AR) trans-activation activity. Twenty-four hours after plating of 1 × 105 cells on six-well plates in DMEM-10% charcoal-stripped FBS, PC-3 cells were transfected with each of the siRNA duplexes, as described above. After 12 h, 1 μg of luciferase reporter plasmid, pGL3-MMTV-Luc, which contains the androgen response element, 0.1 μg pRL-TK (renilla luciferase) for internal control, and human AR expression plasmid pCMV-AR, which were kindly provided by Dr Chang (University of Rochester, Rochester, NY, USA), was co-transfected using Fugene6 (Roche) into AR-null PC-3 cells. Twelve hours after transfection, the media was changed to DMEM-10% charcoal-stripped FBS with or without 10−8 M testosterone. The cells were harvested 48 h after addition of reagents and lysed in luciferase lysis buffer (TOYO INK, Tokyo, Japan) after counting the cell number. Luciferase activity was quantified by a microplate luminometer, and the results were normalized by renilla luciferase activity.
Fine mapping and re-sequencing of 8q24 Region 2. Since 8q24 Region 2 (Chr8:128.14-128.28 Mb) is reported to show the most significant association with PC among several susceptibility loci in multiethnic groups including Japanese–American and Japanese populations(7,18), we focused on this locus and performed fine mapping using 1504 PC cases and 1554 controls of the Japanese population. We selected 57 tag-SNP including two marker SNP loci reported previously (rs13254738 and rs6983561, ref. 7) on the basis of the HapMap JPT data, and genotyped them using multiplex PCR-based Invader assays. Of 56 SNP successfully genotyped, we found the most significant association at rs1456315 (chr8: 128,173,119, P = 2.31 × 10−21), and the association became gradually less significant at its centromeric side and was lost at its telomeric side (Fig. 1A). Linkage disequilibrium (LD) analysis showed that rs1456315 represents an LD block spanning from rs1902431 (128,156,258 bp) to rs7825340 (128,178,311 bp), and the two previously reported marker loci (rs13254738 and rs6983561, ref. 7) were included in this LD block (Fig. 1B). Hence, we performed re-sequencing for this 22-kb region using DNA of 94 PC cases, and identified 95 SNP including 34 novel ones. Among them, we genotyped 53 SNP with a minor allele frequency of 0.05 or higher and identified a novel SNP, SNP34 (128,173,151 bp), to reveal the strongest association with PC susceptibility in this candidate region (P = 1.68 × 10−23, OR = 1.75, 95% confidence interval [CI] = 1.57–1.95, Fig. 2). Detailed association results of this candidate region are shown in Table S1.
Haplotype analysis (Table 1) indicated that the risk haplotype (ATTT) of block 5 at four loci, rs1456315-SNP34-rs5013678-rs7463708, showed more significant association with PC than the single SNP (P = 2.00 × 10−24, OR = 1.74, 95% CI = 1.56–1.93). Blocks 2–4 were absolutely linked to each other and closely linked to block 5. On the basis of these results, we have hypothesized that the PC susceptibility locus in 8q24 Region 2 is likely to be located within block 5 (rs1456315 to rs7463708, chr8: 128,173,119-128,173,237 bp), and the risk haplotype of block 5 might have a functional or causative significance in PC susceptibility.
Table 1. Haplotype analysis of the critical region at 8q24 Region2 associated with prostate cancer
Cloning of a novel transcript in the critical region of 8q24 Region 2. To investigate for a functional significance of the critical region of 8q24 Region 2 on PC susceptibility, we searched for several EST within this region, and screened transcripts by RT-PCR using cDNA from normal prostate and PC cell lines. Through these analyses, we found that a transcript corresponding to Unigene Hs.14743 was expressed in PC cell lines, which seemed to be a psuedogene of SRRM1 on chromosome 1, and failed to find any possible open reading frame. Further 5′-RACE, 3′-RACE and RT-PCR connection and subsequent sequencing analysis identified an approximately 13-kb polyA-tailed transcript (Fig. 3A) that corresponded to the size of the transcript detected by northern blot analysis (Fig. 3B). We confirmed the transcript to be an approximately 13-kb intron-less non-coding RNA spanning a region between rs1016342 and rs7463708 (Fig. 3A), and named this novel ncRNA as PRNCR1 (prostate cancer non-coding RNA 1, GenBank Accession no. AB458446). PRNCR1 expression in prostate cancer cell line LNCaP cells was also upregulated compared with that in normal prostate epithelial cell line PrEC (Fig. S1).
Upregulation of PRNCR1 expression in clinical PC cells and prostatic intraepithelial neoplasia (PIN) cells. We previously performed microdissection of PC cells, its precursor lesion PIN cells, and normal prostate epithelium form frozen PC tissues for the genome-wide gene-expression analysis(17). Using RNA isolated from these microdissected cells, we performed real-time quantitative RT-PCR to examine the expression levels of PRNCR1 in clinical PC cells. PRNCR1 expression was basically low, as shown by northern blot analysis, but we found upregulation of PRNCR1 expression in five out of 10 PC cell populations and two out of four PIN cell populations, compared with normal prostate epithelial cells that were also microdissected from the same PC tissues (Fig. 3C).
PRNCR1 knockdown attenuated PC cell viability. To examine biological roles of PRNCR1 expression in PC cells, we designed siRNA duplexes specific to PRNCR1 and transfected them into PRNCR1-expressing PC cell line LNCaP cells. Among them, si1 showed the significant knockdown effect on PRNCR1 expression and, concordantly, resulted in a significant decrease in PC cell viability measured by MTT assay (Figs 4A,S2). We transfected these siRNA to other PRNCR1-expressing AR-null PC cell lines PC-3 as well and observed a significant suppressive effect on their growth, although its suppressive effect was much smaller than that in the LNCaP cells (Figs 4B,S2). These findings indicate that PRNCR1 could play an important role in PC cell growth or viability. We also transfected these siRNA to PRNCR1-negative breast cancer cell lines HCC1937 to check off-target effects of these siRNA. The transfection of si1 did not affect the growth of PRNCR1-negative HCC1937 cells at all (Fig. 4C), excluding the possibility of its off-target effect.
PRNCR1 depletion affected the transactivation activity of AR. The AR plays the central role in prostate carcinogenesis as well as physiological function and development of the prostate(19), and a number of molecules or co-regulators involved in the AR-signaling pathway and its transactivation activity have so far been identified(19,20). To investigate a possible role of this ncRNA in the AR-signaling pathway, we performed an AR-responsive reporter gene assay by co-transfection AR-expression vector and siRNA to PRNCR1 into AR-null PC-3 cells. As shown in Figure 4D, knockdown of PRNCR1 (si1) significantly decreased the luciferase activity reflecting the transactivation activity of AR in AR-transfected PC-3 cells under androgen stimulation, suggesting that PRNCR1 ncRNA could be involved in the transactivation activity of AR. These findings indicate that ncRNA PRNCR1 could be involved with prostate carcinogenesis, possibly through the regulation of AR transactivation activity.
Many genetic variations on 8q24 have been reported to be associated with PC susceptibility in European and African–American populations(5–9). These susceptible loci on 8q24 are divided at least into three regions, which independently can affect PC susceptibility. Among them, so-called Region 1(128.47–128.54 Mb) was most significantly associated in European populations(7,8). On the other hand, this study and our recent GWAS(21) indicated that variants in Region 2 were most strongly associated with PC susceptibility in the Japanese population (rs1456315 with 1.6 × 10−29, OR = 1.75). These findings suggest a genetic heterogeneity of PC susceptibility among the ethnic population, which might contribute to a lower incidence of PC in Japanese or Asian populations. Furthermore, other 8q24 regions have also been associated with other types of cancer, including breast cancer(22,23), colorectal cancer(23–25), ovarian cancer(23) and bladder cancer(26), while Region 2 has been found to be associated with prostate cancer only, which is consistent with our finding of the possible association of AR with Region 2.
Genome-wide expression analysis suggested that a significant proportion of the mammalian genome is transcribed, but most of them remain untranslated(27). These numerous ncRNA are thought to have important biological functions in development and differentiation through organizing chromatin into domains of active or repressive structure(28). Recent studies on several long ncRNA molecules have indicated that they can regulate gene expression in cis or in trans by modifying the chromatin structure(28,29). The GWAS approach has successfully identified various disease-susceptibility loci that were located in regions without any protein-coding genes. Since the function of these regions is largely unknown, researchers have been encountering difficulties in analyzing their functions. However, several functional analyses identified long ncRNA in these regions; for example, functional analysis of a susceptibility locus for myocardial infarction on chromosome 22q12.1 found a novel long ncRNA, MIAT (myocardial infarction associated transcript) in this region(30). Similarly, a GWAS analysis identified a SNP associated with coronary artery disease(31) in a region encompassing a long ncRNA, ANRIL. However, the complex networks of non-coding transcriptional regulation make it extremely difficult to elucidate the functional effects of these non-coding transcripts. In the present study, for the first time, we identified a novel long ncRNA PRNCR1 at the “gene-desert” critical region for PC susceptibility on 8q24 Region 2 after fine mapping and re-sequencing of this region. The ENCODE project and recent reports on ncRNA studies by analyzing chromatin-state maps did not annotate this long non-coding transcript(32). A recent study using high resolution tiling arrays observed some transcriptional signatures in the PC cell line from 8q24 Region 2, not Regions 1 and 3, but it did not fully investigate these transcripts from 8q24 Region 2(33).
The AR is a transcriptional factor that regulates expression of its downstream genes, which are essential for prostate development as well as carcinogenesis, by forming complexes with a number of regulatory factors or co-regulators and modifying the chromatin structure(20). Among the regulatory factors of AR or steroid receptors, SRA (steroid receptor RNA activator) was reported to act as a catalytic RNA transcript and activate the transcriptional activity of steroid receptors through its distinct RNA motif(34), indicating some possibility that other functional RNA or ncRNA could regulate AR-mediated transcriptional activity. Our reporter assays indicated that PRNCR1 could affect the activity of AR, and we tried to evaluate direct interaction between the AR protein complex and PRNCR1 ncRNA, but we did not obtain any evidence to support their direct interaction. Further studies are required to understand how this ncRNA could be involved with AR transactivation and prostate carcinogenesis.
It is also puzzling how these genetic variants in 8q24 Region 2 could affect the function or expression of this ncRNA and prostate carcinogenesis. Recent reports indicate synonymous variants could change the secondary structure of mRNA, leading to protein translation or RNA stability(35,36). Preliminarily, our in silico analysis predicted that some of these associated SNP and haplotype in the PRNCR1 gene might affect the predicted secondary structure of PRNCR1 mRNA, and these variants might change the stability of PRNCR1 ncRNA or the mRNA conformation leading to the modification of its interacting partners.
In summary, we performed fine mapping and re-requencing for the critical region of 8q24 Region 2, which is most significantly associated with PC susceptibility, and identified in this region a novel long ncRNA, PRNCR1, whose expression was upregulated in PC and PIN cells. PRNCR1 expression could be involved with PC cell viability and AR transactivation activity, although the detailed mechanisms underlying the association of SNP in PRNCR1 with PC susceptibility are unclear. These findings could provide a new insight to understanding the genetic factors of PC susceptibility and prostate carcinogenesis.
The authors would like to thank Ms Hanae Amitani for the support of genotyping, and Ms Mami U for her technical assistance. This work was conducted as a part of the Biobank Japan Projects that was supported by the Ministry of Education, Culture, Sports, Science and Technology of the Japanese government, and was also supported in part by Research grant #22390306 (H.N.) from the Japan Society for the Promotion of Science.