PTENP1 is a ceRNA for PTEN: it’s CRISPR clear

In our 2010 paper entitled “A coding independent function of gene and pseudogene mRNAs regulates tumor biology”, we provided the first evidence that RNA molecules, including non-coding RNAs (such as pseudogenes) and mRNAs, may be endowed with a biological function that specifically relies on their ability to compete for microRNA binding [1].

Our findings have contributed to an evolving microRNA-RNA interaction paradigm, where RNAs are not only “passive” targets of microRNAs, but also “active” regulators of microRNA availability, through a mechanism termed competing endogenous RNA (ceRNA) [2, 3]. Since our publication, a plethora of mRNAs and non-coding RNAs (lincRNAs, pseudogenes, circular RNAs) have been reported to function as ceRNAs in vitro and in animal models. Furthermore, ceRNA functions have been demonstrated to go beyond individual RNA-RNA interactions and extend into complex transcript interaction networks that can be severely dysregulated in cancer [4, 5].

Subsequent to our 2010 publication, many studies independently confirmed PTENP1 pseudogene as a ceRNA for PTEN in prostate cancer, in other cancer types (e.g., bladder cancer, breast cancer, clear cell renal cell carcinoma, endometrial carcinoma, gastric cancer, head and neck squamous cell carcinoma, hepatocellular carcinoma), and in other physio-pathological conditions (see Supplementary references for a list). Nonetheless, a number of articles published in prestigious journals have repeatedly raised concerns about this functional interaction. Herein, we wish to address those concerns raised regarding the techniques we used to modulate PTENP1 expression and show its impact on PTEN expression.

To rule out potential non-specific effects associated with (1) supra-physiological expression of a 3′UTR [6‐9] and (2) congestion of RNA interference machinery caused by siRNA transfection [7], we have chosen to downregulate PTENP1 expression at the transcriptional or post-transcriptional level, taking advantage of CRISPR technology.

To begin, we successfully replicated results reported in our original paper, in spite of the fact that the source of DU145 cells and the batch of siRNAs against PTEN and PTENP1 were different, and that the experiments were performed in a different lab (Fig. 1).

Next, in order to downregulate PTENP1 post-transcriptionally, we used the recently reported CRISPR/CasRx system [10]. For this, we utilized 4 gRNAs designed on the same sequence of the 4 siRNAs composing the siPTENP1 mix (Fig. 2a, b) and we tested them for their ability to decrease the expression of a reporter construct in which PTENP1 3′UTR is cloned downstream of Luciferase coding sequence. As shown in Fig. 2c, the gRNAs work similarly to the corresponding siRNAs, with the mix of all 4 gRNAs working best. Therefore, we decided to use the combination of all 4, as we had done with siRNAs. In Fig. 2d–h, we show the results obtained upon the transient transfection of the gRNA mix in GFP-sorted DU145 prostate cancer cells that stably express CasRx-eGFP (Fig. 2d, e). Consistent with the RNA interference approach (Fig. 1), the gRNA mix caused a downregulation of the intended target PTENP1 RNA, as well as of PTEN mRNA (Fig. 2f). The decrease in mRNA level was mirrored by a decrease in PTEN protein level and accompanied by increases in pAKT levels (Fig. 2g) and cell proliferation (Fig. 2h).

We also adapted the CRISPR/Cas9-based gene replacement strategy [11] in order to achieve the downregulation of PTENP1 at the transcriptional level. Specifically, we engineered an sgRNA-mediated cut between the promoter and the transcribed region of PTENP1 gene. Then, by exploiting homology-mediated recombination, we “knocked-in” a GFP expression cassette in the reverse orientation, which interferes with PTENP1 transcription (Fig. 3a). Using this strategy, we identified 11 GFP-positive KI clones (Fig. 3b), of which 7 harbored correct recombination of both homology arms and 5 showed the expected drop in PTENP1 mRNA levels (clones #A, 2, 5, 8, and 13 reported in Fig. 3c, d). In these clones, we also observed a decrease in both PTEN mRNA and protein levels (Fig. 3d, e). In addition, clones #A, 2, and 13 had accompanying increases in pAKT levels (Fig. 3e) and cell proliferation (Fig. 3f). Crucially, in Fig. 3g, we show that endogenous PTEN mRNA levels are rescued in clone #13, if PTENP1 3′UTR is reintroduced by means of a plasmid that expresses it downstream of Luciferase coding sequence.

In summary, using 2 CRISPR-based technologies (Figs. 2 and 3), we confirmed our results achieved using RNA interference (ref. [1] and Fig. 1): knock-down of PTENP1 leads to the repression of PTEN expression, hence the hyperactivation of oncogenic AKT signaling. In addition, we confirmed that siRNA-mediated knock-down of PTENP1 antisense alpha + beta transcripts results in a downregulation of PTENP1 and PTEN transcripts (Fig. 1b), as previously reported in [12]. Conversely, we showed that the knock-down of PTEN plus PTENP1 transcripts by RNA interference (Fig. 1b) and of PTENP1 by CRISPR/CasRx technology (Fig. 2f) represses the expression of PTENP1 antisense transcripts, whereas the upregulation of PTENP1 transcript elicits the opposite effect (Fig. 3g). In sum, we provide evidence that uncovers a dynamic cross-talk between PTENP1 and PTEN sense transcripts on one side and antisense PTENP1 transcripts on the other.

In the decade since our discovery, numerous groups have independently validated the regulatory interaction between PTENP1 and PTEN. Altogether, these data provide a persuasive body of work to support the existence of a robust and reproducible functional interaction between this gene-pseudogene pair [13]. Finally, the new data presented herein further reinforces the PTENP1-PTEN paradigm and highlights the utility of CRISPR technologies for investigations of pseudogene-parental gene transcript relationships in cancer and other diseases.