vtRNA2-1/nc886 Produces a Small RNA That Contributes to Its Tumor Suppression Action through the microRNA Pathway in Prostate Cancer

vtRNA2-1 is a vault RNA initially classified as microRNA precursor hsa-mir-886 and recently proposed as “nc886”, a new type of non-coding RNA involved in cancer progression acting as an oncogene and tumor suppressor gene in different tissues. We have shown that vtRNA2-1/nc886 is epigenetically repressed in neoplastic cells, increasing cell proliferation and invasion in prostate tissue. Here we investigate the ability of vtRNA2-1/nc886 to produce small-RNAs and their biological effect in prostate cells. The interrogation of public small-RNA transcriptomes of prostate and other tissues uncovered two small RNAs, snc886-3p and snc886-5p, derived from vtRNA2-1/nc886 (previously hsa-miR-886-3p and hsa-miR-886-5p). Re-analysis of PAR-CLIP and knockout of microRNA biogenesis enzymes data showed that these small RNAs are products of DICER, independent of DROSHA, and associate with Argonaute proteins, satisfying microRNA attributes. In addition, the overexpression of snc886-3p provokes the downregulation of mRNAs bearing sequences complementary to its “seed” in their 3′-UTRs. Microarray and in vitro functional assays in DU145, LNCaP and PC3 cell lines revealed that snc886-3p reduced cell cycle progression and increases apoptosis, like its precursor vtRNA2-1/nc886. Finally, we found a list of direct candidate targets genes of snc886-3p upregulated and associated with disease condition and progression in PRAD-TCGA data. Overall, our findings suggest that vtRNA2-1/nc886 and its processed product snc886-3p are synthesized in prostate cells, exerting a tumor suppressor action.


Introduction
Prostate cancer (PrCa) is a heterogeneous disease, the molecular mechanisms of which are still not fully elucidated [1]. It is currently the solid tumor with the highest incidence in men in Western countries, representing the second leading cause of male cancer death [2]. The incidence of PrCa is increasing mainly because of population ageing, increased awareness, and the widespread introduction of the prostate-specific antigen (PSA) test. Despite the fact that most patients can be successfully treated, a minor proportion develop an aggressive form of the disease that is currently incurable. Despite the remarkable progress in the development of molecular biomarker to aid disease diagnosis nc886 processing upon malignant transformation. Figure 1A,B also show that the precise distribution of reads over nc886 sequence is consistent with a specific nuclease processing of nc886 instead of its random degradation [42]. Moreover, the pattern of mapping recalls the distinctive processing of DICER, defined by a sharp 5' end cut and a less precise 3' end cut [42,43], which is seen at both snc886-3p and snc886-5p in prostate. Consistently, a similar stacking of reads over nc886 and the same strand preference is observed in cell lines derived from different human tissues ( Figure S1A,B). We have previously demonstrated an epigenetically driven reduction of nc886 in PrCa, thus we wonder whether the decrease in snc886-3p is due to it. As expected, there is a positive correlation between the expression of snc886-3p and nc886 in the cell lines (r Spearman = 0.95, Figure 2D). Likewise, the expression of snc886-3p correlates with the methylation status of the 200nt region upstream to the transcription start site (TSS200nt) of nc886 (r Spearman = 0.71, Figure 2D).
Altogether these findings suggest that the reduction of snc886-3p levels in prostate cancer is due to the decreased levels of its precursor nc886, which in turn is a consequence of the increased methylation of the nc886 promoter. Seeking additional evidence of the role of snc886-3p, we investigated its dependence on the canonical microRNA biogenesis pathway through the analysis of the changes in the small-RNA-seq transcriptome in knockouts (KO) of DROSHA, XPO5 and DICER generated in the HCT116 cell line [51]. We used a canonical microRNA (hsa-miR-130b), a Drosha-independent microRNA (hsa-miR-320b) [51] and a small RNA fragment derived from the three prime region of RNY3 RNA (sRNY3-3p) as controls. We reasoned that the global reduction in microRNA biogenesis would lead to the overrepresentation of DROSHA-independent small RNAs in the transcriptome. Thus, the observed reduction in hsa-miR-130b Non-coding RNA 2020, 6, 7 6 of 22 and increase of snc886-3p/5p and hsa-miR-320b in the DROSHA-KO indicates that the processing of the last three small RNAs is DROSHA-independent ( Figure 1E). In addition, XPO5-KO produces a significant reduction in hsa-miR-130b and a small decrease in snc886-3p and snc886-5p, indicating that the two latter may be substrates of this exportin. In concordance with previous reports [12,19], DICER-KO decreases the production of snc886-3p and snc886-5p by 96% and 93% respectively, as well as the production of the control microRNAs ( Figure 1E). As expected, no reduction of the fragment derived from RNY3 (sRNY3-3p) was observed for all these knockouts [52].
Overall, our findings indicate that nc886 is a precursor of two small RNAs that meet key microRNA criteria in prostate tissue.

Snc886-3p Has the Expression Profile of a Tumor Suppressor in Prostate Cells
In order to assess the relevance of snc886-3p in prostate we evaluated available microRNA expression datasets of prostate tissues (microarray and qRT-PCR). The analysis reveals a decrease in the expression of snc886-3p in cancer vs normal prostate tissue but no association with stage/prognosis (Table 1). In addition, we analyzed the expression of snc886-3p in six paired normal and tumor samples obtained from paraffin fixed tissues derived from radical prostatectomies [16]. Quantitative RT-PCR confirmed a significant decrease of the expression of snc886-3p in tumor compared to normal tissues in this cohort (Figure 2A). Similarly, the qRT-PCR quantitation of various prostate cell lines shows a reduction in the expression of snc886-3p in tumor cell lines in comparison to the normal prostate cell line PrEc ( Figure 2B). Further analysis of publicly available small-RNA-seq data of prostate cell lines (GSE26970, GSE66035 and SRP109305) reinforce this finding ( Figure 2C). The reduced expression of snc886-3p in malignant vs benign cells, observed in both patient tumors and prostate cancer cell lines, is in agreement with a tumor suppressor function.
We have previously demonstrated an epigenetically driven reduction of nc886 in PrCa, thus we wonder whether the decrease in snc886-3p is due to it. As expected, there is a positive correlation between the expression of snc886-3p and nc886 in the cell lines (r Spearman = 0.95, Figure 2D). Likewise, the expression of snc886-3p correlates with the methylation status of the 200nt region upstream to the transcription start site (TSS200nt) of nc886 (r Spearman = 0.71, Figure 2D).
Altogether these findings suggest that the reduction of snc886-3p levels in prostate cancer is due to the decreased levels of its precursor nc886, which in turn is a consequence of the increased methylation of the nc886 promoter.

Snc886-3p Modulates Transcripts Affecting Cell Cycle and Apoptosis
Aiming to discover the snc886-3p effect on gene expression, we overexpressed the small RNA in DU145 cell line ( Figure 3A) and performed a global gene expression analysis using Affymetrix gene microarrays 48 h after transfection. We used Sylamer algorithm [53] to estimate the enrichment of all 6-nt k-mers in the 3 untranslated regions (3 -UTR) of transcripts of the differentially expressed genes. A landscape plot that tracks the occurrence biases of the different k-mers on all transcripts ranked by their differential expression is obtained. We observed that 5 ACCCGC3 , which is the complementary sequence to the snc886-3p seed (6-mer, 2-7-nt), is the most enriched 6-nt k-mer in the 3 -UTR of the repressed transcripts (adjusted p-value < 0.01) ( Figure 3B). This finding favors a direct and sequence specific repressive interaction of snc886-3p with the mRNA targets. Furthermore, we identified 1358 DEGs (fold change ≥ 1.5) including 704 downregulated and 654 upregulated transcripts (Table S1). Microarray experiments were confirmed by qRT-PCR of selected transcripts (r = 0.97, p-value < 0.0003) ( Figure S2A). Independent snc886-3p overexpression experiments in DU145, LNCaP and PC3 ( Figure  S2B) demonstrate a similar regulation of these genes in these cell lines ( Figure S2C).

Snc886-3p Modulates Transcripts Affecting Cell Cycle and Apoptosis
Aiming to discover the snc886-3p effect on gene expression, we overexpressed the small RNA in DU145 cell line ( Figure 3A) and performed a global gene expression analysis using Affymetrix gene microarrays 48 h after transfection. We used Sylamer algorithm [53] to estimate the enrichment of all 6-nt k-mers in the 3′ untranslated regions (3′-UTR) of transcripts of the differentially expressed genes. A landscape plot that tracks the occurrence biases of the different k-mers on all transcripts ranked by their differential expression is obtained. We observed that 5′ACCCGC3′, which is the complementary sequence to the snc886-3p seed (6-mer, 2-7-nt), is the most enriched 6-nt k-mer in the 3′-UTR of the repressed transcripts (adjusted p-value < 0.01) ( Figure 3B). This finding favors a direct and sequence specific repressive interaction of snc886-3p with the mRNA targets. Furthermore, we identified 1358 DEGs (fold change ≥ 1.5) including 704 downregulated and 654 upregulated transcripts (Table S1). Microarray experiments were confirmed by qRT-PCR of selected transcripts (r = 0.97, p-value < 0.0003) ( Figure S2A). Independent snc886-3p overexpression experiments in DU145, LNCaP and PC3 ( Figure S2B) demonstrate a similar regulation of these genes in these cell lines ( Figure S2C). Since canonical microRNAs are known to act mostly through complementary nucleotide pairing at the 3'-UTR of the target gene mRNAs, we analyzed if the repressive action of snc886-3p transfection was associated with the location of its binding site along the gene. We calculated the average fold change in expression of the transcripts bearing 6-8-mers sequence complementary to the snc886-3p seed in DU145 cell line transfected with snc886-3p (microarray experiments) ( Figure S3A and Table  S1). As expected, snc886-3p causes an average global repression of the genes bearing its complementary site. Most important, its action is significantly stronger when the site is located at the 3'-UTR of the transcript ( Figure S3A). Furthermore, we confirmed this finding using most stringent of mimic snc886-3p and negative control (Dharmacon) after 48 h. Quantification by qRT-PCR using RNAU6 as a normalizer. (B) Sylamer enrichment landscape plot for 6-nt k-mer in all genes ranked by their change in abundance after snc886-3p overexpression compared to control determined by Affymetrix microarrays. The x-axis represents the sorted gene list (downregulated and upregulated genes are plotted in the left and right part of the axis respectively). The y-axis shows the hypergeometric significance for each 6-nt k-mer at each leading bin. Positive and negative values indicate enrichment or depletion of the 6-nt k-mer sequence at the 3 -UTR of the genes. The dotted horizontal line symbolizes an E-value threshold of 0.01 (Bonferroni corrected). The red line embodies the values obtained for the 6-nt k-mer (5 ACCCGC3 ) complementary to the 6-nt seed (6-mer, 2-7-nt) of snc886-3p. The blue line shows the average profile of 6-nt k-mer unrelated to the seed region of snc886-3p (dark blue), with standard deviation as vertical bars.
Since canonical microRNAs are known to act mostly through complementary nucleotide pairing at the 3 -UTR of the target gene mRNAs, we analyzed if the repressive action of snc886-3p transfection was associated with the location of its binding site along the gene. We calculated the average fold change in expression of the transcripts bearing 6-8-mers sequence complementary to the snc886-3p seed in DU145 cell line transfected with snc886-3p (microarray experiments) ( Figure S3A and Table S1). As expected, snc886-3p causes an average global repression of the genes bearing its complementary site. Most important, its action is significantly stronger when the site is located at the 3 -UTR of the transcript ( Figure S3A). Furthermore, we confirmed this finding using most stringent direct target genes narrowed down for their association to AGO [50]. Transcript with a snc886-3p 6-mer (2-7-nt) binding motif at their 3 -UTR show a higher proportion of repression (39%) than those having motifs differing in two-nucleotides (24% and 22%) (Fisher's exact test, p-value < 0.0001 and Odds ratio 2.0 to 5 AACCCC3 and 2.2 to 5 AGCCAC3 for the substituted sequences) ( Figure S3B and Table S1). Overall, the evidence indicates that snc886-3p negative regulation of mRNA levels is dependent on the seed sequence and on the location of the complementary site in the 3 -UTR of the target genes.
In order to investigate the functional consequences of snc886-3p perturbation, we performed gene enrichment analysis of pathways and processes among the DEGs using GSEA curated gene sets [54]. The top 10 significant terms are shown in Table 2. In agreement with our previous study [16], cell cycle, apoptosis and mitogenic pathways are enriched. Particularly, three formerly validated direct targets of hsa-miR-886-3p, PLK1, TGFB1 and CDC6 [25,28], are downregulated upon snc886-3p overexpression. Additionally, candidate snc886-3p direct target CDT1, which is a partner of CDC6, has been associated with cell cycle impairment [55]. A more detailed picture of the snc886-3p responsive genes belonging to the KEGG cell cycle pathway highlights additional downregulated genes ( Figure S2D). Table 2. Gene enrichment analysis GSEA (curated gene sets).

Snc886-3p Causes a Decrease in Cell Viability
We had previously shown that the forced recovery of nc886, the precursor of snc886-3p, reduced in vitro and in vivo cell viability of PrCa cells [16]. In view of the enrichment of cell cycle related pathways in the DEGs caused by snc886-3p overexpression (Table 2), we sought to determine if snc886-3p affects cell viability. We found that the overexpression of snc886-3p in the prostate cell lines DU145, LNCaP and PC3 produces a significant decrease in cell viability relative to the control ( Figure 4A). In addition, we performed cell cycle analysis with propidium iodide to determine cells distribution in the different cell cycle phases ( Figure 4B). We found an accumulation of cells in the S and G2/M phase of the cell cycle in DU145 cell line, that could be associated with the decreased expression of PLK1 evidenced previously ( Figure S2A,C). Indeed, PLK1 was reported as a target of hsa-miR-886-3p, the downregulation of which leads to an arrest in the anaphase stage of mitosis [25,56]. Additionally, LNCaP and PC3 showed an accumulation of cells in G1 phase of the cell cycle ( Figure 4B). These results lead us to propose an effect of snc886-3p on apoptosis. Annexin V, a marker of early apoptosis, significantly increases in DU145 (14% of early apoptosis equivalent to an increase of 69% relative to control) and LNCaP (24% of early apoptosis equivalent to an increase of 366% relative to control) when snc886-3p is overexpressed ( Figure 4C). On the contrary, we found no changes of early apoptosis in PC3 cells between conditions, though we cannot completely discard earlier apoptotic events completed at 72 h ( Figure 4C). Despite the significant overall decrease in cell viability caused by snc886-3p overexpression in the three prostate cell lines tested, the distinct contribution of proliferation Non-coding RNA 2020, 6, 7 9 of 22 and apoptosis may be due to the intrinsic differences of these cell lines. Globally, these analyses show that snc886-3p decreases cell viability, due to a modulation of cell proliferation and early apoptosis.

Direct Candidate Target Genes of snc886-3p Are Associated with Clinical Worse Prognosis in Prostate Cancer Patients
We had previously found that nc886 [16] and snc886-3p expression (Table 1) are associated to worse clinical outcome in PrCa. We now sought to investigate if snc886-3p direct candidate target genes are also associated with the disease. A list of 253 snc886-3p possible direct target genes was built based on two criteria: the association to Argonaute determined by AGO-PAR-CLIP experiments using DU145 (reads with a sequence complementary to the 6-mer (2-7-nt) snc886-3p seed sequence) identified in Hamilton et al. [50] and the downregulation of at least 1.25 FC in DU145 after snc886-3p overexpression (Table S1). A hierarchical clustering of the PRAD-TCGA samples in association with clinical status (pathological, clinical, Gleason score, residual tumor, biochemical recurrence) is shown in Figure 5. The expression of the 253 genes segregates the samples of patients with worse (red/violet) from those with better (non-red) clinical presentation as well as the normal tissues (green). Additionally, two major gene groups are clustered. Remarkably, the positive association of the 106 genes cluster with worse clinical parameters suggests that they may be the relevant direct candidate targets of snc886-3p repression in vivo (individual genes are listed in Table S1). We also found that men harboring tumors with low expression of the 253 snc886-3p direct candidate target genes (percentile 25th) have a longer disease-free survival compared with the high expressing ones (percentile 75th) (HR=2, p-value 0.031) ( Figure 6A). Disease-Free Survival analysis based on the expression of the 106-gene list showed a greater statistical difference in patient survival probability (HR=3.2, p-value 0.00018) compared with the 253-gene list ( Figure 6A,B). Furthermore, the expression of the 106-genes associates with higher methylation of nc886 TSS200nt (thus of snc886-3p) (r 0.29, p < 0.0001) ( Figure S4A,B). These results provide in vitro and in vivo support of a tumor suppressor microRNA-like role of snc886-3p in prostate, acting through the direct repression of transcripts bearing complementary sequences at their 3′-UTR.  We had previously found that nc886 [16] and snc886-3p expression (Table 1) are associated to worse clinical outcome in PrCa. We now sought to investigate if snc886-3p direct candidate target genes are also associated with the disease. A list of 253 snc886-3p possible direct target genes was built based on two criteria: the association to Argonaute determined by AGO-PAR-CLIP experiments using DU145 (reads with a sequence complementary to the 6-mer (2-7-nt) snc886-3p seed sequence) identified in Hamilton et al. [50] and the downregulation of at least 1.25 FC in DU145 after snc886-3p overexpression (Table S1). A hierarchical clustering of the PRAD-TCGA samples in association with clinical status (pathological, clinical, Gleason score, residual tumor, biochemical recurrence) is shown in Figure 5. The expression of the 253 genes segregates the samples of patients with worse (red/violet) from those with better (non-red) clinical presentation as well as the normal tissues (green). Additionally, two major gene groups are clustered. Remarkably, the positive association of the 106 genes cluster with worse clinical parameters suggests that they may be the relevant direct candidate targets of snc886-3p repression in vivo (individual genes are listed in Table S1). We also found that men harboring tumors with low expression of the 253 snc886-3p direct candidate target genes (percentile 25th) have a longer disease-free survival compared with the high expressing ones (percentile 75th) (HR=2, p-value 0.031) ( Figure 6A). Disease-Free Survival analysis based on the expression of the 106-gene list showed a greater statistical difference in patient survival probability (HR=3.2, p-value 0.00018) compared with the 253-gene list ( Figure 6A,B). Furthermore, the expression of the 106-genes associates with higher methylation of nc886 TSS200nt (thus of snc886-3p) (r 0.29, p < 0.0001) ( Figure S4A,B). These results provide in vitro and in vivo support of a tumor suppressor microRNA-like role of snc886-3p in prostate, acting through the direct repression of transcripts bearing complementary sequences at their 3 -UTR.
We previously found that nc886 acts as a tumor suppressor in prostate cancer and the vector-driven overexpression of this RNA inhibited cell proliferation and invasion [16]. Due to existing reports about the production of small RNA with microRNA function derived from vtRNAs [59][60][61], we wonder if there is a specific processing of nc886 into small RNAs in prostate cells and if it could possibly contribute to the described nc886 induced phenotype in prostate cancer cells [16].
To assess the synthesis of vtRNA2-1/nc886 small fragments, we interrogated published small-RNA-seq prostate datasets, revealing a pattern of small RNA read stacking over nc886 sequence that resemble the cleavage of DICER which generates complementary 5p and 3p fragments ( Figure 1A, Figure S1A,B). Additionally, the analysis of Kim et al. dataset showed that DICER knockout impaired the production of snc886-3p/5p in the HCT116 cell line [51]. Interestingly, although the secondary structure of nc886 resembles a microRNA hairpin precursor, it is not fully optimized as a canonical microRNA precursor [47,[62][63][64]. Indeed, the 101 nt nc886 sequence of is longer than the median 83 nt of human microRNA precursors (range 41-180 nt), generating mature small RNA fragments that are also longer (23 nt of snc886-3p and 24-25 nt of snc886-5p) than the 22 nt median human mature microRNAs (range 16-27 nt); this pattern was described for other DICER substrates of similar length [63,64]. Moreover, the presence of asymmetrical mismatches and bulges in the predictive nc886 secondary structure is associated with longer mature sequences [63]. Overall, its structural divergence with the optimized human microRNA hairpin precursor might lead to a low efficient DICER processing and consequently a small amount of microRNAs observed in our and former studies [12,18]. Evolutionary hypothesis of new microRNA precursor proposes their birth as non-efficiently processed precursor RNAs, until specific target genes are evolutionary selected; this avoids the non-specific repression of genes allowing the optimization of the hairpin structure for DICER processing [65]. The identification of the presence of a large fraction of nc886 and a less abundant fraction of derived small RNA may be the result of its recent evolutionary origin in eutherians [9].
To further investigate the putative microRNA-like function of snc886s we analyzed their association with Argonaute proteins. Previously, other groups reported small RNA fragments derived from vtRNA1 associated to Argonaute, holding microRNA function in breast and lymphoid tissue [59,60]. Here, we show that snc886-3p is also associated with Argonaute in lung cell line WI-38 ( Figure S1D), which is in agreement with the proposed microRNA identity of hsa-miR-886-3p (here snc886-3p) acting through direct repression of PLK1 and TGFB1 in lung cancer [25,27]. Concordantly, we found proof for the association of snc886-3p with Argonaute protein in prostate by analysis of AGO-PAR-CLIP available small-RNA-seq data from prostate cell line DU145 [50]. Indeed, snc886-3p enrichment is above the average of the microRNAs in these experiments, indicating a specialized microRNA-like function for this small RNA product in comparison to the 5p. Although our study shows a low association of snc886-5p to AGO it does not rule out the function of snc886-5p as a microRNA, which has been shown previously in other tissues (brain [19], cervical [66,67], breast [68]). Alternatively, snc886-5p might be involved in another pathway or be a non-functional secondary product of nc886 processing.
In view of the possible microRNA-like identity of snc886-3p, we then investigated its association with PrCa in existing small RNA studies obtained using various techniques (qRT-PCR, microarray and small-RNA-seq). We found that hsa-miR-886-3p/snc886-3p is globally downregulated in worsened disease conditions in the published datasets (Table 1), as well as in our patient cohort (Figure 2A) and laboratory stablished cell lines ( Figure 2B,C). These datasets show that snc886-3p has the expression pattern of a tumor suppressor gene the level of which in prostate cells is in the low-medium microRNAs expression range. In addition, we found that snc886-3p expression positively correlates with nc886 expression and negatively correlates with the methylation of nc886 TSS200nt. Furthermore, snc886-3p has a bimodal expression in prostate cell lines ( Figure 2B,D), similar to what we reported for nc886 promoter methylation in prostate cell lines and tissues. This distribution could be explained by zygotic differences in the promoter methylation status of the gene [16]. Despite this pre-existing variation in the methylation status of nc886 TSS200nt, both patient groups present an increased promoter methylation in tumor compared with the normal matched tissue [16].
Different direct targets of microRNA regulation by hsa-miR-886-3p have been proposed in the literature mostly after its removal from miRBase (PLK1, TGFB1, CDC6, CXCL12 and FXN) [25,28,30,31]. If snc886-3p functions as a microRNA in PrCa it is expected to sequence-specifically bind to target mRNAs to repress their expression. In order to test this hypothesis, we performed a global gene expression study of DU145 cell line overexpressing snc886-3p vs a control small RNA. Sylamer analysis of the differentially expressed transcripts demonstrated that the sequence complementary to snc886-3p seed is the most overrepresented 6-nt k-mer in the 3 -UTR of the downregulated transcripts by snc886-3p overexpression, which strongly favors a microRNA mechanism of action. Furthermore, the identification of previously validated targets PLK1, CDC6 and TGFB1, among the downregulated transcripts of this experiment, reinforces the assumption.
To obtain an insight of the biological effect of snc886-3p, we studied the pathways enriched in the DEGs identified by the microarray experiment, finding an enrichment in cell cycle and apoptosis processes which predicts a proliferative inhibition. This inference was confirmed by in vitro cell viability and cell cycle distribution determination in DU145, LNCaP and PC3 PrCa cell lines. Snc886-3p overexpression reduces proliferation of the three cell lines and provokes a distinctive phase specific cell cycle arrest in DU145 (G2/M) compared to LNCaP and PC3 (G1). An increase in early apoptosis after snc886-3p transfection was validated through Annexin V determination in DU145 and LNCaP, while no change was observed in PC3 cell line. Cell line specific effects may be due to differences in cell cycle control. A similar effect of snc886-3p in cell viability was previously reported in thyroid, lung and breast tissues in vitro and in vivo [22,25,27,28]. Additionally, a high-throughput analysis of seed 6-nt k-mers RNAs exposed that the snc886-3p 6-nt seed k-mer causes a significant decrease in cell viability in several non-prostate cancer cell lines [69]. In agreement with these findings, the overexpression of nc886 produces a similar phenotype in tissues where is described as a tumor suppressor (prostate [16,70], thyroid [28], gastric [71] and esophageal [72]), whereas its silencing produces the same effect in non-malignant cells in agreement with the tumor surveillance model proposed by Kunkeaw et al. (cholangiocarcinoma [73]). Accordingly, hsa-miR-886-3p and nc886 act as pro-proliferative and anti-apoptotic RNAs in tissues where it is proposed as an oncogenic RNA (renal [33], ovarian [74], thyroid [75] and endometrial [76]). Our current results together with our previous study indicate that both snc886-3p and nc886 have an antiproliferative effect and a tumor suppressor function in prostate cells. Nonetheless, given their different size, structure, and abundance, they are expected to participate in different effectors pathways. The specific production of small RNAs from nc886 is controversial in the literature, probably because of the differences in the tissue types and methods employed for its detection (RNA-seq, northern blot, microarray, qRT-PCR). In addition, the effect of both molecules has been studied using exclusively the nc886 or the snc886-3p, thus their interdependence has been mostly ignored. Indeed, the overexpression and inhibition of nc886 is expected to produce an increase/decrease of hsa-miR-886-3p/snc886-3p respectively. Meanwhile, the inhibition of snc886-3p using anti-miRs could modify the abundance, folding or endogenous interactors of the precursor molecule. As an example, a study in bladder cancer, where hsa-miR-886-3p expression is associated with short patient survival, found that both hsa-miR-886-3p knock down and nc886 overexpression lead to a decrease of cell viability in half of the cell lines tested [21]. For that reason, the effects assigned to only one of the molecules could be in fact the result of the modulation of both nc886 and its snc886 products. Indeed, the study of snc886 mimics circumvents most of these problems, allowing the discrimination of the sole effect of hsa-miR-886-3p/snc886-3p. Nevertheless, it is worth recalling that in vivo both molecules are synthesized and functional in the cells. To our knowledge, five previous reports analyzed functional effects and direct target genes of hsa-miR-886-3p/snc886-3p using mimic molecules (lung cancer) [25,27], human marrow stromal cells [30], breast cancer [22] and gastric cancer [71] and their findings are consistent to ours in prostate cancer cells. Remarkably, the only study that examined the function of both nc886 (transfection) and snc886-3p (mimic) using gastric cancer cells, found that only the transfection of nc886 (but not snc886-3p) affects cell proliferation and viability, although they showed its ability to function as a microRNA in a reporter gene assay harboring a complementary binding site at the 3 -UTR [71].
We and others have shown that the expression of nc886 and snc886s is associated with patient clinical outcome [20,25,71,72,74,75]. The group of Lee also showed the clinical association of the gene signature of nc886 knockout [71,72,74,75]. To our knowledge, there is no clinical association study of the snc886-3p direct candidate target genes in the literature. We built 253-and 106-gene sets composed of transcripts bearing a sequences complementary to snc886-3p seed (6-mer, 2-7-nt) protected in AGO-PAR-CLIP experiments and downregulated in our microarray experiments, the expression of which turned out to be significantly associated with a lower disease-free survival in PRAD-TCGA and unfavorable clinical parameters. Furthermore, the negative association of the 106-gene set with nc886 promoter methylation strengthens the specificity of these in vitro defined gene set. Overall, these results are strong indicators of the validity of the snc886-3p DEGs identified in vitro in DU145 cell line and reinforce the tumor suppressor function of snc886-3p in PrCa in the clinical set.
Although RNA biomarkers are increasingly established in prostate cancer management [77], snc886-3p does not fulfill relevant biomarker attributes, such us tissue specificity and upregulation in the disease condition. However, small non-coding RNAs derived from nc886 have been repeatedly detected in body fluids [78][79][80][81][82][83][84][85] and, contrary to our findings in the tumor tissue, a recent study of circulating microRNAs in the plasma of prostate cancer patients reported an increase expression of snc886-3p in high-grade compared to low-grade tumor biopsy [32]. In this context, the value of snc886-3p as a cancer biomarker is still unclear. Notwithstanding, previous reports suggested that the methylation of the nc886 promoter may be worthy of further investigation in the prostate cancer biomarker field [16,86]. Finally, the expression of snc886-3p candidate direct target genes that associate with prostate cancer prognosis may be valuable for future biomarker research.

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-micron 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
LNCaP and DU145 cell lines derive from a supraclavicular lymph node and a brain metastasis respectively, while PC3 and VCaP derive from vertebral bone metastatic sites. In addition, LNCaP and VCaP are androgen sensitive, whereas DU145 and PC3 are androgen independent. Cell lines PCSC-1, PCSC-2 and PCSC-3 are highly metastatic androgen independent primary prostate cancer cell lines enriched in CD133+ stem cells. RWPE-1/2 are non-malignant cells derived from normal adult prostate epithelium and WPE-int is a less differentiated derivative of RWPE1.

Cell Transfection
The snc886-3p mimic (5 CGCGGGUGCUUACUGACCCUU3 ) and negative controls (5 UCACAACCUCCUAGAAAGAGUAGA3 ) (NC, CN-001000-01-05) were synthesized Dharmacon (Miridian). According to the manufacturers' instructions, the mimic and control mimic 20nM, were transfected into cells in 12 and 96 well plates using Lipofectamine TM 3000 (Invitrogen, Carlsbad, CA, USA). After 24 h of transfection the medium was removed and fresh medium was added and 48 h (Affymetrix microarray and qRT-PCR analyses) or 72 h (Cytotoxicity assay, Flow Cytometry for DNA content analysis, Annexin V Alexa Fluor 488/PI apoptosis detection assay) after transfection cells were collected for further analysis.

Microarray Experiments
Total RNA from 3 replicates of snc886-3p (hsa-miR-886-3p) mimic and negative controls (NC, CN-001000-01-05) was extracted after 48 h of transfection using the Qiagen TM miRNAeasy kit according to the manufacturer's protocol. The total RNA of the three replicates was pooled and labeled according to Affymetrix (Affymetrix, Santa Clara, CA, USA). Hybridization, staining and washing of the Affymetrix ® HG-U133 Plus 2.0 Arrays were performed with the Affymetrix Fluidics Station 450 and Hybridization Oven 640 under standard conditions (Affymetrix, Santa Clara, CA, USA). Quality control analysis and Pre-processing of the CEL-files microarray expression data was done using a graphical user interface, Chipster (v1.4.3, CSC, Finland, http://chipster.csc.fi/) following the manufacturer's guidelines [88]. Normalization was performed using RMA algorithm [89] and annotation using the specific Affymetrix ® HG-U133 Plus 2.0 Arrays probe set library in Chipster. Normalized log2 expression values were used to determine the fold change expression of the snc886-3p (hsa-miR-886-3p) mimic and negative controls (NC, CN-001000-01-05) (Table S1).

Cytotoxicity Assay
At 72 h after of transfection with snc886-3p or control RNA (60-80% cell culture confluence), 20 µL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) 5 mg/mL in 1X PBS was added to the wells and cultures were incubated for 4 h at 37 • C in a 5% CO 2 controlled atmosphere. The medium was then aspirated and 200 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 on a plate spectrophotometer (Varioskan ® Flash Multimode, Thermo Scientific, Waltham, MA, USA) at 570 nm (for formazan absorbance measurement) and 690 nm (for background measurement).

Flow Cytometry for DNA Content Analysis
At 72 h after transfection with snc886-3p or control RNA (60-80% confluence) cells were harvested by trypsinization followed by two washes and resuspension in 1X PBS with gentle vortexing. Cells were then fixed by adding 1 mL of ice cold 70% ethanol dropwise and incubated at -20 • C for 30 min. Next, cells were washed with 1X PBS, centrifuged 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 an Accuri™ C6 flow cytometer (BD Bioscience), counting 10,000 total events per sample (BD Accuri C6 software).

Annexin V Alexa Fluor 488/PI Apoptosis Detection Assay
Apoptosis was detected by initially staining the cells with 0.1% (v/v) Annexin V and 100 µg/mL of propidium iodide solution, according to the Annexin V Alexa Fluor 488/PI apoptosis detection assay kit (catalog n • V13241, Invitrogen, USA), followed by flow cytometry analysis in an Accuri™ C6 flow cytometer (BD Bioscience), counting 10,000 total events per sample (BD Accuri C6 software).

Analysis of microRNA Microarray Datasets
Depending on the type of study and the availability of the data, we followed different strategies. Data deposited at GEO was analyzed using the GEO2R tool using default settings [90], selecting the samples by clinical status definition. For all the microarray data of PrCa studies analyzed, relevant features used in the analyses are listed in Table 1.

Analysis of Methylation Microarray Datasets
The methylation data of the PRAD-TCGA 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). Additionally, public methylomes available at the repository Gene Expression Omnibus (GEO) [41] obtained using Illumina Infinium Human Methylation 450 BeadChip arrays were also analyzed prostate cell lines PrEc, RWPE1, VCAP, LNCaP, DU145 and PC-3 gene dataset GSE68379 [97]. The average of the normalized beta-values for the 6 CpGs sites located at the nc886 TSS200nt promoter (cg18678645, cg06536614, cg26328633, cg25340688, cg26896946, cg00124993) were calculated.

Heatmap of Hierarchical Clusterization of 253 snc886-3p Candidate Direct Target Genes
Heatmap was performed by two-way hierarchical clustering using the Spearman rank correlation algorithm using Morpheus (https://software.broadinstitute.org/morpheus/) and gene expression values and clinical status for different parameters (Clinical T value, Pathological T value, Gleason Score, Residual Tumor and Biochemical Recurrence) of PRAD-TCGA dataset.

Statistical Analysis
All experiments were performed at least in triplicate and the corresponding variables are expressed as average value ± standard deviation or standard error (referred in the figure). Statistical analyses were done using single, two-tailed t-test, one-way and two-way ANOVA for multiple comparison tests, including Tukey's Honest Significant Difference test as a post-hoc tests (referred in the figure). D'Agostino-Pearson was conducted as normality test and Pearson or nonparametric Spearman was used to test correlation. Two-tailed Fisher exact test for difference in the proportions of genes was used. All the analyses were done in GraphPad Prism 6. The observed differences were expressed using the p-value (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p-value < 0.0001). Results with a p-value of < 0.05 were considered significant.

Conclusions
Our study demonstrates the presence of hsa-miR-886-3p/snc886-3p in prostate tissue derived from the DICER mediated processing of vtRNA2-1/nc886, which associates to argonautes repressing transcripts bearing complementary seed sequences, thus functioning as a microRNA. We also found a DNA methylation dependent downregulation of snc886-3p and a concomitant upregulation of direct candidate targets of repression, both associated with PrCa disease condition and progression. Snc886-3p effects on global gene expression support the modulation of cell cycle progression and apoptosis observed in prostate cancer cell lines. Altogether, our results indicate that both nc886 and snc886-3p are simultaneously expressed in prostate cells at different levels, exerting a tumor suppressor action through probably different effector pathways.