PAPD5-mediated 3' adenylation and subsequent degradation of miR-21 is disrupted in proliferative disease

Abstract

Next-generation sequencing experiments have shown that microRNAs (miRNAs) are expressed in many different isoforms (isomiRs), whose biological relevance is often unclear. We found that mature miR-21, the most widely researched miRNA because of its importance in human disease, is produced in two prevalent isomiR forms that differ by 1 nt at their 3' end, and moreover that the 3' end of miR-21 is posttranscriptionally adenylated by the noncanonical poly(A) polymerase PAPD5. PAPD5 knockdown caused an increase in the miR-21 expression level, suggesting that PAPD5-mediated adenylation of miR-21 leads to its degradation. Exoribonuclease knockdown experiments followed by small-RNA sequencing suggested that PARN degrades miR-21 in the 3'-to-5' direction. In accordance with this model, microarray expression profiling demonstrated that PAPD5 knockdown results in a down-regulation of miR-21 target mRNAs. We found that disruption of the miR-21 adenylation and degradation pathway is a general feature in tumors across a wide range of tissues, as evidenced by data from The Cancer Genome Atlas, as well as in the noncancerous proliferative disease psoriasis. We conclude that PAPD5 and PARN mediate degradation of oncogenic miRNA miR-21 through a tailing and trimming process, and that this pathway is disrupted in cancer and other proliferative diseases.

Keywords: microRNA processing; nucleotidyl transferase.

Fig. 1.

The upper image shows the structure of pre–miR-21 as predicted by mfold (55). The canonical miR-21 and miR-21+C isomiRs are highlighted in blue and burgundy, respectively. The definitions used in this paper appear below the structure.

Fig. 2.

(A) Deep-sequencing data of small RNAs from THP1 cells upon knockdown of candidate adenylating enzymes showed a large and highly significant decrease in the adenylation ratio of miR-21+C after knockdown of PAPD5, which was confirmed in a replicate PAPD5 knockdown experiment. (B) The expression level of miR-21, as measured by qPCR and normalized against the expression of the spliceosomal RNA U6, increased significantly upon knockdown of PAPD5 in THP1 cells. (C) Upon knockdown of exoribonuclease PARN, the miR-21+C adenylation ratio increased significantly compared with NC both in THP1 (two replicates) and in MCF7 (three replicates). In THP1, the miR-21+C adenylation ratio also increased significantly upon knockdown of PAN3.

Fig. 3.

Gene expression profiling in THP1 cells was performed using microarrays for two biological replicates each for the NC, the PAPD4 knockdown condition, and the PAPD5 knockdown condition. (A) Compared with NC, predicted target genes of miR-21 were significantly down-regulated upon knockdown of PAPD5, but not upon knockdown of PAPD4. This observation holds if more stringent selection criteria for miR-21 target genes are applied, such as requiring a context score lower than −0.2 or requiring multiple target sites for the same. Asterisks indicate statistically significant differential expression. Error bars represent the SDs of the estimated mean percentage change between the knockdown condition and the NC. (B) A heat map view of the expression data confirms that the profiled genes cluster by experimental condition.

Fig. 4.

Across a wide variety of cancers, the degradation ratio (A) and the adenylation ratio (B) of oncomir miR-21+C is significantly lower in tumor samples compared with normal samples. A reduced degradation ratio (C) and adenylation ratio (D) is also observed in psoriasis, which is similarly characterized by elevated expression levels of miR-21 (E). Error bars indicate the SD of the estimated mean; Kruskal–Wallis P values of the difference between healthy and disease samples are shown for each disease type. (F) As shown here for lung squamous cell carcinoma (LUSC), in most cancers the PAPD5 expression level is positively correlated with the miR-21 adenylation ratio, which in turn is negatively correlated with the miR-21 expression level (G). Similarly, in psoriasis the miR-21 adenylation ratio is negatively correlated with the miR-21 expression level (H). “Involved” refers to psoriatic skin samples, “uninvolved” to samples of skin from psoriatic patients that are not affected by psoriasis, and “normal” to skin from healthy donors. Dashed lines indicate the 95% confidence interval of the regression line. The Spearman correlation and its corresponding P value are shown. miRNA-21 expression levels are normalized to the overall miRNA expression level in each sample; PAPD5 expression levels are evaluated as reads per kilobase of transcript per million mapped reads (RPKMs). cpm, counts per million miRNAs.

Fig. 5.

PAPD5 prolongs the life of the mRNA encoding tumor suppressor TP53 by polyadenylation. By adenylating miR-21+C and thus inducing its degradation by PARN, PAPD5 prevents the oncomiR from repressing other tumor suppressors. Once associated with an Argonaute protein, miR-21 isomiRs engage in the promotion of a variety of classical cancer hallmarks (56), such as angiogenesis, the avoidance of apoptosis, altered signaling, metastasis, enhanced anaerobic metabolism, and enhanced proliferation. The miR-21 targets shown in this figure have all been experimentally verified [PTEN (57), PDCD4 (58), SPRY1 (59), and SERPINB5 (60)]. Nucleotidyl transferases are represented as yellow hexagons; TP53, a key transcription factor involved in tumor suppression, is represented by a green rounded rectangle; the exoribonuclease PARN is shown in orange; and regulatory interactions are represented by green arrows (up-regulation or stimulation) or red “inhibition” arrows (down-regulation or inhibition). The uptake of canonical miR-21 and miR-21+C into the RISC complex is denoted by arrows. Adenylated miR-21+C undergoes degradation by the 3′ exoribonuclease PARN, which yields the 22 nt miR-21 and shorter degradation products.