Background
Prostate cancer (PrCa) is the solid tumor with the highest incidence in men in Western countries, representing the second leading cause of male cancer death [
1,
2]. Worldwide, one sixth of men will be diagnosed with prostate cancer in their lifetime. Although most patients can be treated successfully, a minor proportion develop an aggressive form of the disease that is currently incurable. It is fundamental to develop biomarkers that allow the precise prognosis at early stages, as well as new therapeutic tools to treat these patients in advanced stages. Non-coding RNAs have recently emerged as key players in cancer initiation and progression [
3,
4], therefore their clinical value is under intense investigation [
5‐
7].
The large collection of non-coding RNAs (ncRNAs) of the human genome is broadly grouped per size and function in two main types: a group of < 40 nt long RNAs known as “small RNAs” (including microRNAs, piwiRNAs, snoRNAs) and a group of > 200 nt long RNA named “long non-coding RNAs” [
8]. The “vault” RNAs (vtRNAs) are a class of 84-141 nt long eukaryotic RNAs, that are transcribed by RNA polymerase III. They associate with conserved vault proteins forming the vault particle, a complex whose function and relevance in cancer remains scarcely understood [
9]. Whereas the three human vtRNA1–1-3 are integral components of the vault particle, vtRNA 2–1 is a more divergent homologue, whose transcript is neither associated to the vault particle or co-regulated with the vtRNA1–1-3 [
10,
11]. Before miRBase version 16, vtRNA2–1 was classified as a microRNA precursor, thus annotated as “precursor of hsa-miR-886-3p” (pre-miR-886); however, the recognition of its sequence homology with the three vtRNA-1 RNAs [
11] led to its re-classification as vtRNA2–1 and the elimination of its derived microRNAs from miRBase. However, vtRNA2–1/pre-miR-886, was more recently proposed to be a new type of non-coding RNA (referred there as “nc886”), that acts as a tumor suppressor, inhibiting the activation of Protein Kinase RNA-activated (PKR) by direct interaction [
12‐
14]. Consistent with these findings, Treppendahl et al. showed that nc886 functions as an epigenetically regulated tumor suppressor gene in acute myeloid leukemia, and that genome demethylating treatment inhibits PKR phosphorylation [
15]. However, in the same work, the authors detected mature miRNAs derived from nc886, and showed they are products of the processing of pre-miR-886 by a non-canonical pathway independent of DROSHA. In addition, other groups have identified the mature microRNAs derived from pre-miR-886 in lung small cell carcinoma [
16] and prostate cancer [
17], presenting evidence of its association with disease progression.
Different lines of evidence have revealed that the epigenetic control of nc886 is complex and may own clinical relevance. Independent reports in breast, lung, colon, bladder, esophagus and stomach cancer showed that its promoter is differentially methylated in tumor vs. normal tissue [
18,
19]. In fact, in lung cancer, chronic myeloid leukemia and gastric cancer, its differential methylation correlates with patient prognosis and survival [
15,
16,
20]. Although these findings support a tumor suppressor role for nc886, a recent communication proposed its action as an oncogene in thyroid cancer [
21]. Intriguing aspects of the epigenetic regulation of this locus, include its dependence on the parental origin of the allele [
22], and its sensitivity to the peri-conceptional environment [
23].
The aim of this study was to investigate the possible involvement of nc886 in PrCa etiology and behavior. Analyzing clinical samples, we found that the full transcript of nc886 is present in prostate tissue and diminishes its abundance in tumor compared to normal tissue, thus showing a gene expression pattern of a tumor suppressor gene. The increased methylation of nc886 promoter in transformed vs. non-transformed tissue, as well as demethylating agent treated vs. untreated cell lines, indicate that the molecular etiology of nc886 downregulation is the methylation of its promoter. Indeed, nc886 promoter methylation level correlates with clinical parameters of PrCa (Gleason Score, clinical T value and biochemical relapse). Forced restitution of nc886 in DU145 and LNCaP cell lines produces an inhibition of cell invasion and proliferation in vitro and a reduction of DU145 tumor growth in vivo. These results are consistent with a tumor suppressor role, suggesting a nc886 antiproliferative function in normal prostate tissue. Finally, the interrogation of the Prostate Adenocarcinoma of The Cancer Genome Atlas (TCGA-PRAD) cohort, uncovered a negative association between the expression of nc886 and the expression of genes belonging to the PrCa cell cycle progression gene signature (CCP), providing a molecular support for the phenotype experimentally observed after nc886 recovery.
Methods
Human specimens
Tissue sections were obtained from paraffin fixed blocks stained with hematoxylin and eosin (H&E) of 6 archived radical prostatectomies and were evaluated by three pathologists at the Department of Anatomic-pathology of the Police Hospital. This study was approved by the Hospital Policial,
D.N.AA.SS., Montevideo, Uruguay (2010).
Matched normal and tumor regions, showing similar parenchyma-stroma ratio and similar cytological findings at the stroma were selected. Unstained section of 10-μm thickness, contiguous to the sections selected by the pathologist, were then freshly obtained to extract small RNAs using the RNeasy FFPE (Qiagen) Kit, with the following modifications: two extra washes with xylene and absolute ethanol were added. The RNA was resuspended in RNAse free water and stored at − 20 °C for further analysis.
Cell lines
RWPE-1, LNCaP (ATCC CRL-1740), PC-3 and DU145 human prostate cancer cell lines were obtained from ATCC (Manassas, VA, USA). LNCaP, DU145 and PC-3 were maintained in RPMI 1640 (R7755) supplemented with 10% FBS (PAA™) and penicillin/streptomycin. RWPE-1 cell line was cultured in Keratinocyte Serum Free Medium (Gibco by LifeTechnologies™) supplemented with 0.03 mg/mL bovine pituitary extract (BPE) and 0.5 ng/mL EGF human recombinant epidermal growth factor (EGF) and penicillin/streptomycin. All cell lines were maintained in a 5% carbon dioxide atmosphere at 37 °C.
A lentiviral vector bearing the precursor nc886 or a scrambled sequence of the same length, both cloned downstream of the CMV promoter (miExpress precursor expression clones, pEZX-MR02, GeneCopoeia) were transduced in DU145 and LNCaP. Transduced cells were then selected by growth in the presence of puromycin.
5-Azacytidine treatment
DU145, RWPE-1, PC-3 and LNCaP cells were treated for 72 h with 1.5 μmol/L 5-Azacytidine (ab142744, Abcam) and DMSO as control, replacing the medium with freshly added drug every 24 h following manufacturer’s instructions.
Total RNA was extracted using the Qiagen™ miRNAeasy kit. Reverse transcription was performed using the Qiagen PCR miScript II System. Quantitative real time PCR (qPCR) was performed with the miScript SYBR Green PCR Kit using specific oligonucleotides. For nc886: 5′CGGGTCGGAGTTAGCTCAAGCGG3′ forward primer and 5′AAGGGTCAGTAAGCACCCGCG3′ reverse primer, as in Lee K et al. [
12]. U6 RNA was amplified using the primer assay purchased from Qiagen (Hs_RNU6-2_11 miScript Primer Assay (MS00033740)) and the miScript Universal Primer (Qiagen). The relative quantification was attained using the ΔΔCT method, in a Rotor-Gene 6000 equipment (Corbett Life Science), employing U6 as the internal control of RNA load.
MTT assay
Five thousand DU145 and LNCaP cells per well were seeded in 96 culture plates. Twenty μL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) 5 mg/mL solution dissolved in 1X PBS was added to the wells and cultures were incubated for 4 h at 37 °C in a 5% CO2 controlled atmosphere. The medium was then aspirated and 100 μL of DMSO was added to each well and incubated at room temperature in the dark for 15 min with moderate orbital shaking. Optical density (OD) was read in a plate spectrophotometer (Thermo Scientific Varioskan® Flash Multimode) at 570 nm and 690 nm wavelengths.
Flow cytometry for DNA content
DU145 cells transfected with lentiviral vectors producing nc886 or a scrambled RNA control were seeded in triplicate in 6-well plates. Upon reaching 60% confluence, cells were harvested by trypsinization, washed twice with 1X PBS and resuspended in 1X PBS by gentle vortexing. Cells were then fixed by adding 1 mL of ice cold 70% ethanol dropwise in 1X PBS and incubated at − 20 °C for 30 min. Next, cells were washed with 1X PBS and centrifugated at 1200 rpm at 4 °C for 5 min and the resuspended cell pellets were incubated with 0.1 mg/mL of RNase and 50 μg/mL propidium iodide for 15 min at room temperature in the dark. Flow cytometry measurement of nuclear DNA content was performed in a Accuri™ C6 flow cytometer (BD Bioscience), counting 10.000 total events per sample (BD Accuri C6 software).
Matrigel invasion assays
24-well transwell inserts of 8 μM pore size (Corning #3422) were coated with Matrigel (Corning) for in vitro invasion assays. Fifteen thousand (DU145) or 20,000 (LNCaP) cells were seeded in serum-free RPMI 1640 and migrated towards the bottom chamber containing RPMI 1640 supplemented with 10% FBS. After 48 h the cells were fixed with 100% methanol and stained with hematoxylin and eosin (H&E). Non-invading cells were scrubbed with a cotton swab. Five microscopic fields were photographed and counted for each sample. Values were averaged from at least 3 independent experiments.
Mice xenograft
Six 4-week-old male athymic NUDE BALB/C mice were maintained according to the protocols and ethical regulations of the animal facility of the Institute of Biomedical Science, at the University of Sao Paulo, Brazil (protocol 134/10, approved by the Ethics Committee for Animal Use). In order to grow tumors, these mice were subcutaneously injected on both flanks using 3 × 106 DU145 cells resuspended in 50 μl of Matrigel matrix (Corning Inc.) per inoculation. The tumor growth was measured weekly with calipers and the corresponding volumes were calculated as: length x width x height x π/2. When tumors reached 2 cm, the animals were euthanized, and tumors were extracted and properly stored for further analysis.
Analysis of mice tumors
The study of histological sections of the tumors extracted from the mouse xenotransplantation assays, was conducted at the Laboratory of Medical Research – LIM55, Urology Department, of the University of Sao Paulo, Brazil. Specifically, the percentage of necrosis and mitotic indexes in the histological sections of the tumors stained with H&E were quantified.
Dataset analysis
The TCGA-PRAD data was downloaded from the TCGA portal (
https://tcga-data.nci.nih.gov/docs/publications/tcga/?) [
24] and the Methhc database (
http://methhc.mbc.nctu.edu.tw/php/index.php) [
25]. This dataset includes the RNA-Seq expression values of 50 matched normal and tumor tissue and additional unmatched normal and tumor samples, generated using Ilumina sequencing technology. The methylation data of the TCGA-PRAD cohort, was extracted from the Illumina Infinium Human Methylation 450 BeadChip array data of the 49-paired normal and prostate tumor samples and additionally unmatched normal and tumor tissues (336 in total). Several public methylomes available at the Gene Expression Omnibus (GEO) repository [
26] were also analyzed: matched normal and tissue PrCa GSE76938 [
27], PrCa metastasis GSE38240 [
28], PrCa cell lines GSE34340, GSE62053 and GSE54758 [
29,
30] and HCT166 cell lines GSE51810 [
31]. The average of the normalized beta-values for the 6 CpGs sites located at the nc886 TSS200 promoter (cg18678645, cg06536614, cg26328633, cg25340688, cg26896946, cg00124993) were calculated. Hierarchical clusterization obtained through Euclidean algorithm was performed using the Gene-E (
http://www.broadinstitute.org/cancer/software/GENE-E/) for the methylation beta-values and Morpheus (
https://software.broadinstitute.org/morpheus/) for gene expression values.
Statistical analysis
All experiments were performed at least in triplicate, and the corresponding variables are expressed as average value ± standard deviation or standard error. Statistical analyses were done using single and two-tailed t-test, and the statistical significance of the observed differences were expressed using the p-value (* p < 0.05, ** p < 0.01, *** p < 0.001). D’Agostino-Pearson was conducted as the normality test and nonparametric Spearman was used to test correlation.
Discussion
Nc886 has been recently shown to act as a tumor suppressor ncRNA in cholangiocarcinoma, esophageal carcinoma, gastric cancer and leukemia [
15,
18,
20,
33]. The etiology of its downregulation in cancer has been linked to the methylation of its promoter in leukemia, colon, lung, gastric, bladder, breast and esophageal tumors [
15,
16,
18‐
20]. Furthermore, nc886 has been proposed both as a tumor suppressor and as an oncogene, depending of the context and the tissue involved, as was recently showed in thyroid cancer [
21]. Thus, a more comprehensive picture of nc886 action in cancer, including tissue specific differences and potentially specific molecular mechanisms is still required.
Here we present the first study of the role of nc886 in PrCa. We found an increased methylation of nc886 gene promoter in prostatic tumor tissue vs. its matched normal counterpart, analyzing the samples available at the TCGA-PRAD dataset and in the cohort of Kirby et al. A similar observation was made in leukemia, colon, lung, gastric, bladder, breast and esophageal tumors [
15,
16,
18‐
20], supporting the hypothesis of increased nc886 promoter methylation as a recurrent event in the initiation step of solid tumors. Indeed, we found that the level of nc886 promoter methylation correlates with PrCa patient clinical evolution, reinforcing previous findings in gastric, lung, leukemia and esophageal cancer [
15,
16,
18,
20]. Furthermore, the predominant medium and high nc886 methylation groups tissues, in both normal and tumor TCGA-PRAD samples, positively correlate with the clinical outcome of the disease. It is worth to note, that the so-called “normal adjacent” prostatic tissue from PrCa patients (Fig.
1a), may be in fact abnormally modified by tumor induced changes at the organ level [
28], thus conclusions derived exclusively from this type of tissue should be taken with caution. Indeed, the study of Aryee et al. (Fig.
1b) reports lower levels of average nc886 promoter methylation in bona fide prostate normal tissue obtained from organ donors, suggesting that “normal tissue” adjacent to tumor tissue may had already undergone an increase in nc886 promoter methylation. Alternatively, methodological differences between the two studies may explain the divergence in the average methylation. The fact that another recent study using normal adjacent tissue reports nc886 promoter methylation comparable to TCGA-PRAD, favors the former interpretation. In addition, we found significantly lower levels of average nc886 promoter methylation in bona fide normal tissue obtained from organ donors in comparison with metastatic tissues. The finding of similar levels of TSS200 methylation in high methylated samples from normal and primary tumors in comparison with metastatic tissue, suggests that nc886 promoter methylation is a pre-requisite for tumor metastasis. Altogether, our results favor a tumor suppressor role for nc886 in several steps of PrCa tumorigenesis. It also indicates that nc886 silencing is a driver epi-mutation in PrCa. Finally, our findings point out to a potential use of nc886 for disease stratification in PrCa.
Our study also proves that the level of expression of nc886 in PrCa tissue is significant lower than in the normal counterpart. This goes in agreement with findings in other tissues, in which nc886 was proposed as a tumor suppressor gene [
12,
15,
18,
20,
33]. In addition, we show that nc886 promoter methylation negatively regulates its transcript abundance in PrCa cell lines, as was shown previously in leukemia, gastric and esophageal tumors [
15,
18,
20].
Although aberrant DNA hypermethylation in PrCa is a fundamental driver of tumor progression and overexpression of the DNMTs is a signature of disease origin and evolution, the mechanism responsible for the epigenetic silencing of nc886 in cancer has not been addressed so far. Among the three DNMTs, DNMT3B has been consistently shown to increase its levels in transformed vs. normal prostate tissue, both in patient tumors and in cell lines [
34‐
37] and its expression increase along with adverse clinical parameters [
36,
37]. Functional studies in siRNA cell lines, cadmium-transformed prostate epithelial cells and TRAMP mouse models [
35,
38,
39], together with the association between PrCa risk and a polymorphism in DNMT3B leading to increased enzyme expression [
40], have provided further support to this hypothesis. Thus, DNMT3B seems to be the most important DNMT driver in PrCa. Concordantly, we found a positive correlation between the fold change in expression of DNMT3B and nc886 promoter methylation in matched normal to tumor tissue in the TCGA-PRAD cohort, which favors DNMT3B involvement in nc886 promoter methylation during neoplastic transformation in the prostate. Nevertheless, further experiments in PrCa models are needed to prove the hypothesis.
Seeking the effect of nc886 deregulation in prostate tissue, we analyzed several PrCa cell lines finding support for the correlation between nc886 expression and methylation seen in the clinical samples. In addition, the cell lines show quite variable expression of nc886 in agreement with the patient data. Accordingly, an important heterogeneity in nc886 promoter methylation has been reported in several studies. The natural variation in the methylation of the locus in humans might be explained by imprinting and methylation in response to the peri-conceptional environment [
19,
22,
23]. Nevertheless, the increase in the methylation of nc886 promoter in tumor vs. matched normal tissue regardless of the initial level of expression in the normal tissue, has strong support in the cancer literature [
15,
16,
18‐
20]. In this context, we chose two cell lines with different levels of nc886 methylation and concomitant expression to cover the spectrum of nc886 variation in prostate tissue. In addition, these cells lines model the androgen sensitive (LNCaP) and androgen insensitive (DU145) forms of PrCa disease. We found that nc886 overexpression produces a significant decrease of the in vitro proliferation, possibly by a retention of the cells in the G2/M stage of the cell cycle. Additionally, when we assessed the effect of overexpression of nc886 in vivo we found a significant decrease in tumor growth. The decreased growth of nc886 overexpressing tumors was accompanied by a high percentage of necrosis, low number of mitosis and low tumor weight trends. Forced overexpression of nc886 causes a reduction in in vitro invasion through Matrigel. The concordant pattern of methylation and expression of nc886 observed together with the phenotype of the overexpressing cells strongly contributes to the idea that nc886 functions as a tumor suppressor gene in the prostate.
Then we looked for possible transcriptomic changes responsible for the proliferative effect of nc886 in PrCa. We observed that most of the expression of genes previously shown to be upregulated upon nc886 knock-down in cell lines models of esophageal, gastric and thyroid cancer, do not correlate with nc886 expression in prostate tumor tissues. This finding suggests that the molecular changes induced by nc886 deregulation in esophageal, gastric and thyroid cancer are different that those in the prostate, thus favoring a tissue specific model of nc886 action in cancer. Remarkably, we found that the CCP, a validated clinically useful PrCa proliferation gene signature, positively associates with nc886 expression in the TCGA-PRAD cohort. This goes in agreement with the negative effect of nc886 in cell proliferation observed both in vitro and in vivo using a gain-of-function of nc886 in PrCa cell lines and poses candidate genes that might be co-regulated with, responsive to or regulators of nc886. While different molecular mechanisms of nc886 action in cancer have been proposed, including the modulation of the PKR/NFkB pathway [
12,
41] and the generation of microRNAs [
16,
17], the relative contribution of nc886 and derived microRNAs in cancer has been hardly addressed in the literature. They could either modulate the same or completely independent cellular processes through diverse molecular pathways. Interestingly, some of the genes belonging to the CCP signature have mRNA motifs complementary to the seed region of the microRNAs potentially derived from nc886. Indeed, one of these has already been validated as a target gene in another tissue [
16]. Although more work is necessary to elucidate the precise molecular mechanism of action of nc886 in the prostate cells, its association with the CCP at the level of gene expression in the clinical set reinforces its clinical relevance in PrCa disease.
Acknowledgements
We are very grateful to Dr. Mercedes Rodríguez-Teja (Departamento de Genética, Facultad de Medicina, UdelaR), Dr. José Badano (Laboratorio de Genética Molecular Humana, Institut Pasteur de Montevideo) and MSc. Ximena Camacho (Radiofarmacia, Centro de Investigaciones Nucleares, Facultad de Ciencias, UdelaR) for kindly providing reagents.