MicroRNA-like snoRNA-Derived RNAs (sdRNAs) Promote Castration-Resistant Prostate Cancer

Abstract

We have identified 38 specifically excised, differentially expressed snoRNA fragments (sdRNAs) in TCGA prostate cancer (PCa) patient samples as compared to normal prostate controls. SnoRNA-derived fragments sdRNA-D19b and -A24 emerged among the most differentially expressed and were selected for further experimentation. We found that the overexpression of either sdRNA significantly increased PC3 (a well-established model of castration-resistant prostate cancer (CRPC)) cell proliferation, and that sdRNA-D19b overexpression also markedly increased the rate of PC3 cell migration. In addition, both sdRNAs provided drug-specific resistances with sdRNA-D19b levels correlating with paclitaxel resistance and sdRNA-24A conferring dasatinib resistance. In silico and in vitro analyses revealed that two established PCa tumor suppressor genes, CD44 and CDK12, represent targets for sdRNA-D19b and sdRNA-A24, respectively. This outlines a biologically coherent mechanism by which sdRNAs downregulate tumor suppressors in AR-PCa to enhance proliferative and metastatic capabilities and to encourage chemotherapeutic resistance. Aggressive proliferation, rampant metastasis, and recalcitrance to chemotherapy are core characteristics of CRPC that synergize to produce a pathology that ranks second in cancer-related deaths for men. This study defines sdRNA-D19b and -A24 as contributors to AR-PCa, potentially providing novel biomarkers and therapeutic targets of use in PCa clinical intervention.

Keywords: CD44; CDK12; cancer; castration-resistant; microRNA; ncRNA; noncoding RNA; prostate; sdRNA; snoRNA.

Figure 1

SdRNAs are specifically processed from annotated snoRNA loci. Transcripts arising from various annotated snoRNA loci have now been definitively shown to participate in at least two distinct noncoding RNA regulatory pathways. Individual loci can produce snoRNAs functioning exclusively as traditional posttranscriptional RNA editors directing 2′-O-methylation or pseudouridylation of transcripts (right), or exclusively as functional miRNA precursors (left). Some loci have now been confirmed to produce transcripts, at times engaging in both types of noncoding RNA regulation (center). MiRNA-like excision products are illustrated in black (left and center) as excision products of primary transcript. Complementary RNA editing targets are shown in red (right and center) with red dashes indicating the larger transcript excluded for the purpose of clarity. Adapted from Patterson et al. [7].

Figure 2

SdRNAs-D19b and -A24. (A) SdRNA-A24 and sdRNA-D19b are significantly overexpressed sdRNAs in TCGA prostate cancer patient datasets. The SURFR algorithm [14,15] was used to identify sdRNAs abundantly expressed in prostate cancer patient tumors versus normal prostate. (B) The most thermodynamically stable secondary structures of putative sdRNA producing snoRNAs with sdRNA sequences highlighted in blue as calculated by Mfold [36]. Common name and Ensembl gene ID for putatively processed snoRNAs are listed below corresponding structures. “Hits” refer to the number of times fragments of putative sdRNA producing snoRNAs perfectly aligned to small RNA-seq reads (PRAD ID: f45a166f-d67b-5de1-8cbd-b5782659457a) from the TCGA prostate cancer dataset. Numbers preceding total numbers of hits correspond to the number of times positions highlighted in blue (putative sdRNAs) perfectly aligned to small RNA-seq reads (e.g., 1380 of 1407 small RNA reads aligning to snoRNA-A24 corresponded to the sequence highlighted in blue). (C) Alignment between the human genome (GRCh38:chr4:118279190-118279320:1) (top), SNORA24 (ENSG00000275994) (upper middle), sdRNA-A24 (SURFR call) (lower middle), and next generation small RNA sequence read (bottom) obtained by Illumina sequencing of PC3 cell Ago immunoprecipitations (SRR2966868) is shown. The underlined sequence corresponds to the Illumina TruSeq Small RNA adapter RA3. All sequences are in the 5′ to 3′ direction. An asterisk indicates base identity between the snoRNA and genome. Vertical lines indicate identity across all three sequences. (D) Alignment (as in (C)) between the human genome (GRCh38:chr3:52690744-52690827:1) (top), SNORD19b (ENSG00000238862) (upper middle), sdRNA-D19b (SURFR call) (lower middle), and next generation small RNA sequence read (bottom) obtained by Illumina sequencing of PC3 cell Ago immunoprecipitations.

Figure 3

SdRNA-D19b and -A24 levels significantly impact PC3 cell proliferation and migration. (A) PC3 cells were transfected with indicated sdRNA mimic or antagomiR (Anti-sd). Cell counts were performed at 24 and 72 h then normalized to scrambled control transfections (n = 8). * indicates p ≤ 0.05; ** indicates p ≤ 0.01; p-values by unpaired two-tailed t-test. (B,C) Representative migration (wound-healing) assays for PC3 cells transfected with the indicated sdRNA mimic. Wound border closure is indicated by black arrows. (D) PC3 migration assays quantified. Images were captured at the indicated times (X-axis) and wound healing quantified using ImageJ as % migration normalized to scrambled control (n ≥ 3). * indicates p ≤ 0.05; p-values by unpaired two-tailed t-test. D42a, sdRNA-D42a mimic; CTLm, scrambled mimic; A24, sdRNA-A24 mimic; D19b, sdRNA-D19b mimic.

Figure 4

SdRNA overexpression protects PC3 cells from chemotherapeutic agents. Cells were cultured in 24-well plates and transfected at 70% confluency with mimics or inhibitors. Following transfection, cells were treated with (A) paclitaxel (5 nM) or (B) dasatinib (50 nM). Cell death was quantified every 6 h for 24 h total using ImageJ and methylene blue dead cell staining. 19 m, sdRNA-D19b mimic; 19i, sdRNA-D19b inhibitor; 24 m, sdRNA-A24 mimic; 24i, sdRNA-A24 inhibitor; CTLm, scrambled mimic; CTLi, scrambled inhibitor; Mock, vehicle-treated control. (n ≥ 3). * indicates p < 0.001; p-values by unpaired two-tailed t-test as compared to Mock.

Figure 5

SdRNA-D19b and sdRNA-A24 mRNA targets. (A) Alignments between putative 3′UTR target sites with sdRNAs-D19b (top) and -A24 (bottom). Vertical lines indicate Watson-Crick basepair. Dotted lines indicate G:U basepair. TS1, target site 1. TS2, target site 2. (B) SdRNAs-D19b and -A24 specifically repress luciferase expression from mRNAs containing CD44 and CDK12 target sites in their 3′UTRs. SdRNA mimics and luciferase reporters with target sequences (bottom) and/or controls (LACTA refers to beta galactosidase control sequence) were constructed and cotransfected, as previously described [7]. * indicates p < 0.01; p-values by unpaired two-tailed t-test as compared to LACTA excepting LACTA compared to CD44.