A coding-independent function of gene and pseudogene mRNAs regulates tumour biology

Abstract

The canonical role of messenger RNA (mRNA) is to deliver protein-coding information to sites of protein synthesis. However, given that microRNAs bind to RNAs, we hypothesized that RNAs could possess a regulatory role that relies on their ability to compete for microRNA binding, independently of their protein-coding function. As a model for the protein-coding-independent role of RNAs, we describe the functional relationship between the mRNAs produced by the PTEN tumour suppressor gene and its pseudogene PTENP1 and the critical consequences of this interaction. We find that PTENP1 is biologically active as it can regulate cellular levels of PTEN and exert a growth-suppressive role. We also show that the PTENP1 locus is selectively lost in human cancer. We extended our analysis to other cancer-related genes that possess pseudogenes, such as oncogenic KRAS. We also demonstrate that the transcripts of protein-coding genes such as PTEN are biologically active. These findings attribute a novel biological role to expressed pseudogenes, as they can regulate coding gene expression, and reveal a non-coding function for mRNAs.

Figure 1. PTENP1 is targeted by PTEN-targeting microRNAs

a. Working hypothesis: PTEN is protected from microRNA binding by PTENP1. microRNAs: colored squiggles; 5’and 3’UTRs: open rectangles; open reading frames: filled rectangles. b. PTEN (upper) and PTENP1 (lower) 3’UTRs contain a highly conserved (dark grey) followed by a poorly conserved (light grey) domain. PTEN-targeting microRNA seed matches within in the high homology region are conserved between PTEN and PTENP1. c. Binding of PTEN-targeting microRNAs to PTENP1. Seeds and seed matches: bold; canonical pairings: solid lines; non-canonical pairings (G:U): dotted lines. d. PTEN-targeting miR-19b and miR20a decrease PTEN and PTENP1 mRNA abundance. e. miR-17 and miR-19 family inhibitors derepress PTENP1 abundance (left). PTEN is used as positive control (right). d and e. mean ± s.d., n ≥ 3.

Figure 2. PTENP1 3’UTR exerts a tumour suppressive function by acting as a decoy for PTEN-targeting microRNAs

a–c. PIG/ψ3’UTR-infected DU145 cells show (a) increased PTEN mRNA and protein levels (b) reduced phosho-AKT levels upon EGF stimulation and (c) decreased proliferation rate. d. Growth in semisolid medium of DU145 cells infected with PIG, PIG/ψ3’UTR or PIG/PTEN. e. PTEN mRNA levels 24h after the transfection of pCMV/ψ3’UTR in parental HCT116 or HCT116-DICER−/− cells. Data are normalized using pCMV empty-transfected cells. f. Growth curve of DU145 cells transfected with control siLuc, si-PTEN/PTENP1, si-PTEN or si-PTENP1. g. mRNA levels of PTEN (left) and PTENP1 (right) 24h after the transfection of siLuc (white), si-PTEN/PTENP1 (blue), si-PTEN (black), si-PTENP1 (red). i. Western blot of PTEN 48h after the transfection of the indicated siRNAs. a, c, d, e, f and g. mean ± s.d., n ≥ 3.

Figure 3. Loss of PTENP1 in cancer

a–b. Expression level of PTEN (black) and PTENP1 (red) in a panel of normal human tissues (a) and prostate tumour samples (b). Linear regression of PTEN vs PTENP1 expression is shown in the upper left corner. c. Cluster analysis of 48 sporadic colon cancer samples interrogated by Affymetrix Human SNP Array. d. Heat map and Cluster analysis of Affymetrix Human Exon 1.0 ST Array for normalized PTEN intensity values. e. Plot of log-ratio of PTENP1 copy number (CN) against log10 PTEN expression intensity. Lines of best fit represent regression analyses of two populations. The correlation coefficient (r) measures the reliability and the p-value measures the statistical significance of the correlation between the x and y.

Figure 4. PTEN 3’UTR and KRAS1P 3’UTR function as decoys and a general model for endogenous microRNA decoy mechanism

a. PTENP1 mRNA level 24h after the transfection of the empty pCMV or pCMV/PTEN3’UTR plasmid in DU145 cells (left) and growth curve (right). b. KRAS mRNA level 24h after the transfection of the empty pCMV or pCMV/K1P3’UTR plasmid in DU145 cells (left) and growth curve (right). c. Model. X and Y are different transcripts targeted by the same microRNA(s). In the steady state (middle), equilibrium exists between the microRNA molecules and their targets X and Y. Downregulation of X (left) leads to increased availability of microRNA molecules to bind to Y, thus decreasing its abundance. By contrast, overexpression of X (right) leads to less microRNA molecules free to bind to Y, and thus Y abundance increases. Red rectangles: microRNA molecules. X and Y can be a pseudogene and its cognate protein-coding gene. a and b. mean ± s.d., n ≥ 3.