Naturally existing isoforms of miR-222 have distinct functions

Abstract

Deep-sequencing reveals extensive variation in the sequence of endogenously expressed microRNAs (termed 'isomiRs') in human cell lines and tissues, especially in relation to the 3' end. From the immunoprecipitation of the microRNA-binding protein Argonaute and the sequencing of associated small RNAs, we observe extensive 3'-isomiR variation, including for miR-222 where the majority of endogenously expressed miR-222 is extended by 1-5 nt compared to the canonical sequence. We demonstrate this 3' heterogeneity has dramatic implications for the phenotype of miR-222 transfected cells, with longer isoforms promoting apoptosis in a size (but not 3' sequence)-dependent manner. The transfection of longer miR-222 isomiRs did not induce an interferon response, but did downregulate the expression of many components of the pro-survival PI3K-AKT pathway including PIK3R3, a regulatory subunit whose knockdown phenocopied the expression of longer 222 isoforms in terms of apoptosis and the inhibition of other PI3K-AKT genes. As this work demonstrates the capacity for 3' isomiRs to mediate differential functions, we contend more attention needs to be given to 3' variance given the prevalence of this class of isomiR.

Figure 1.

Endogenous miRNAs exhibit substantial 3′ variation. (A) Frequency of miR-222 mapping reads are shown from multiple independent AGO-CLIPs in MDA-MB-231 and SHEP cells. The sequence and abundance of canonical miR-222 from miRbase is shown. (B) Extensive miR-222 3′-heterogeneity is also seen in publicly available whole cell RNA sequencing from three cell lines. (C and D) The distribution of 5′ and 3′ variants of all miRNAs classified by distance from the most common end is shown by box plot. Shaded areas and whiskers indicate the 25–75th and 5–95th percentile ranges respectively. Data were obtained from (C) AGO-bound miRNAs from MDA-MB-231 cells and (D) the most abundant 227 miRNAs from the Cancer Genome Atlas (TCGA).

Figure 2.

Longer miR-222 isomiRs alter cell morphology. (A and B) MCF10A cells are shown 72 h after the transfection of miR-222 isomiRs, the 21 or 25 nt controls (nc21 and nc25) or the miR-222GAGA mimic (in B). (C) Representative images of similarly transfected additional cell lines.

Figure 3.

miR-222 and its isomiRs inhibit proliferation and share the same target recognition potential. (A–C) MCF10A cells were transfected as indicated and proliferation assessed by (A) confluence, (B) CyQuant and (C) MTS assay. (D) Scratch wound closure assay was also performed after transfection, though effects on proliferation versus migration are difficult to disentangle. (E) The expression of CDKN1A (p21) was determined at both the mRNA and protein level, 48 and 72 h post-transfection of miR-222, miR-222CUCU and the negative controls. (F) Relative luciferase activity was measured for 25 putative and established miR-222 target genes by cloning the respective 3′UTRs downstream of the reporter gene and co-transfecting with miR-222 isoforms or negative controls as indicated. The parental vector was used as a negative control.

Figure 4.

Apoptosis is induced by longer miR-222 isomiRs. (A) The proportion of non-viable cells and the expression of apoptotic marker genes at the (B) RNA and (C) protein level were determined after transfection of the miR-222 isoforms or 21 and 25 nt controls (nc21 and nc25). MiR-222CUCU-mediated apoptosis is indicated by the induction of pro-apoptotic Puma and Noxa and the cleavage of Caspase 3 and PARP.

Figure 5.

Interferon response genes are not induced by miR-222 transfection. MiR-222 isomiRs, control RNAs (nc21 and nc25), a known inducer of the interferon response (siBlunt27+0) or a non-inducing negative control (siBlunt27+2) were transfected into MCF10A cells and a series of IFN (interferon) pathway markers measured by RT-PCR.

Figure 6.

Longer miR-222 isomiRs suppress the PIK3/AKT pathway. (A–C) The expression of multiple PIK3/AKT pathway components in MCF10A cells in response to the transfection of miR-222 and miR-222CUCU are shown using (A) RNA sequencing, (B) qPCR or (C) western blotting. (D and E) PI3KR3 knockdown decreases PI3K/AKT pathway gene expression and promotes apoptosis as indicated by (E) PARP cleavage, (F) morphology, (G) cell confluence and (H) flow cytometry analysis for non-viable cells.

Figure 7.

Endogenous miRNAs are abundant within the nucleus and exhibit size-dependent nuclear enrichment. (A) MCF10A cells were separated into nuclear and cytoplasmic fractions and subjected to Illumina sequencing. Clean fractionation is indicated by the differential localization of nuclear snoRNAs and snRNAs and cytoplasmic tRNAs. Endogenous, mature miRNAs are both nuclear and cytoplasmic. The y-axis indicates total sequencing read counts and the x-axis, length of the associated small RNA. (B) Shorter (<20 nt) and longer (>23 nt) miRNAs are preferentially enriched within nuclei. (C) Proportion of the nuclear distribution of miR-222 isomiRs of varying size are indicated by box plot.