Deletion of PTENP1 pseudogene in human melanoma.

Dear Editor,

The aberrant activation of the PI3K/AKT pathway in melanoma is known to be caused by genomic deletion, promoter methylation or loss-of-function mutations of phosphatase and tensin homolog on chromosome 10 (PTEN) (see ref. (Madhunapantula and Robertson, 2009) for a review).

More recently, PTEN protein abundance was shown to be decreased at the post-transcriptional level by a complex microRNA network (He, 2010).

The processed pseudogene of PTEN, PTENP1, is a modulator of the interaction between PTEN mRNA and PTEN-targeting microRNAs (Poliseno et al., 2010). PTENP1 shows extensive sequence similarity with PTEN. The high level of conservation in the most upstream region of the 3’UTR allows PTENP1 mRNA to be bound by many of the microRNA families that bind PTEN mRNA (Figure 1 and Supplemental Text S1). Therefore, PTENP1 can protect PTEN from microRNA-mediated downregulation. Consistent with its activity as a decoy for PTEN-targeting microRNAs, PTENP1 is a bona fide tumor suppressor gene that causes growth inhibition in prostate cancer cells and undergoes deletion in various human malignancies (Poliseno et al, 2010).

Figure 1

PTEN- and PTENP1-targeting microRNAs

a. Schematic representation of PTEN and PTENP1 3’UTR. The most upstream region of the 3’UTR (R1, dark grey) is highly conserved between PTEN and PTENP1. In the middle R2 region (light grey), the conservation drops dramatically. The downstream R3 region (white) is present only in PTEN (modified from Poliseno et al., 2010). b. List of the PTEN-targeting microRNA families validated so far. The region of the 3’UTR where the microRNA binding sites are located and the conservation of these sites in PTENP1 3’UTR are indicated. Out of a total of 18 PTEN-targeting microRNA families, 12 are PTENP1-targeting as well (grey). All these families have at least 1 binding site in the high homology R1 region.

As recently pointed out (Chen, 2010), the post-transcriptional regulation of PTEN levels in melanoma has not been studied yet and the status of PTENP1 on 9p13 is unknown. Here, we report the analysis of PTENP1 locus in human melanoma cell lines and tissues (Table S1). We also analyze the relationship between: 1) PTENP1 deletion and CDKN2A deletion, as they are located approximately 20cM apart on chromosome 9p (Bennett, 2008); 2) PTENP1 deletion and PTEN deletion, as 9p and 10q losses often coexist in melanoma (Indsto et al., 1998); 3) PTENP1 deletion and BRAF/NRAS mutation, as the combination of mutated BRAF (but not NRAS) and loss of PTEN expression is a common event in human melanoma (Tsao et al., 2004).

Our genomic analysis revealed partial deletion of PTENP1 locus in 14.3% of the cell lines tested (Figure 2a). In addition, PTENP1 was found to be deleted in 20.9% of melanoma tissues (8 partial and 1 complete deletion) (Figure 2b). These results illustrate that PTENP1 is under selective pressure to undergo copy number losses (Poliseno et al., 2010). The average PTEN expression level in the 9 samples with either partial or complete deletion of PTENP1 (P5, P9, P11, P12, P20, M1, M9, M14 and M16) was lower than in the remaining 34 samples without any PTENP1 deletion (5.22 ± 2.30 N=9 vs. 20.59 ± 3.98 N=34, p = 0.06), confirming the protective effect of PTENP1 on PTEN levels (Poliseno et al., 2010).

Figure. 2

Analysis of CDKN2A, CDKN2B, MTAP, PTENP1, PTEN, BRAF and NRAS status in human melanoma cell lines (a) and tissues (b)

a. 14 primary (WM series) and 14 metastatic (Sk-Mel series, 501Mel, A375 and 451Lu) melanoma cell lines and b. 23 primary (P1-P23) and 20 metastatic (M1-M20) melanoma tissues were analyzed by qPCR on genomic DNA to identify deletions at 9p21 (CDKN2A, CDKN2B and MTAP loci), 9p13 (PTENP1 locus) and 10q (PTEN locus). Mutations identified in exon 11 and 15 of BRAF and exon 2 and 3 of NRAS are reported. PTEN protein levels were analyzed in melanoma tissues by immunohistochemistry (IHC) and the percentage of positive cells is indicated. N/A: genomic qPCR not performed due to constraints on DNA availability. On the left, IHC shows lower PTEN levels in P5 tissue harboring partial PTEN and PTENP1 deletion compared to M5 tissue harboring partial PTEN deletion only.

The details of the analyses of CDKN2A deletion, PTEN deletion, and BRAF and NRAS mutation in the human melanoma cell lines and tissues are provided in Supplemental Text S2.

We observed that the partial or complete deletion of PTENP1 is always concomitant with the partial or complete deletion of CDKN2A in human melanoma cell lines, but not in human melanoma tissues. The “focal” deletion of PTENP1 (5/9 cases) has already been reported in other cancer types, such as breast and colon cancer (Poliseno et al., 2010), but is particularly interesting in melanoma, because it happens in spite of the fact that CDKN2A deletions are the most common alterations (Bennett, 2008).

As far as the relationship between the partial or complete deletion of PTENP1 and that of PTEN is concerned, we observed that PTEN is deleted in 7/9 melanoma tissues that show PTENP1 deletion (Figure 2b). This result is supported by previous loss-of-heterozygosity analyses (Herbst et al., 1999) and is in agreement with the observation that 9p and 10q losses often coexist in melanoma (Indsto et al., 1998).

The co-deletion of PTEN and PTENP1 can be reconciled with the function of PTENP1 as decoy for PTEN-targeting microRNAs. PTEN deletion, either partial or complete, was found in 23 melanoma tissues. Excluding the cases in which PTEN deletion is complete and therefore PTENP1 deletion cannot affect PTEN levels, on average we observed lower PTEN protein expression in the samples showing both PTEN partial deletion and PTENP1 deletion compared to those showing PTEN partial deletion only, as exemplified by P5 and M5 tissues in Figure 2b. In the context of partial PTEN deletion, PTENP1 deletion may cause a further decrease in PTEN levels, possibly due to less efficient sheltering of PTEN-targeting microRNAs. Of note, PTENP1 deletion may also affect PTEN levels when PTEN is downregulated by mechanisms other than genomic deletion, such as inactivating mutations.

The biological effects of PTENP1 deletion are likely to be particularly strong on PTEN because of the intrinsic nature of this haploinsufficient tumor suppressor whose variations of even 20% can have profound consequences on cancer onset and progression (Berger and Pandolfi, 2011). Nonetheless, the concomitant deletion of PTENP1 and PTEN points towards PTEN-independent functions of PTENP1 as well. Two possible explanations can be invoked. First, PTENP1 is likely to function as decoy for additional microRNA families which are not PTEN-targeting. These microRNAs should preferentially bind to the R2 region of the 3’UTR that shows low homology with the corresponding one on PTEN (Figure 1a) and might affect targets that belong to other signaling cascades, so that the concomitant deletion of PTENP1 and PTEN causes the activation of independent pathways that possibly cooperate in tumorigenesis. Second, the PTEN-targeting microRNAs for which PTENP1 acts as decoy have additional targets. Therefore, the deletion of PTENP1 might be advantageous for the tumor because of the decrease of other oncosuppressor genes besides PTEN (Poliseno et al., 2010).

Examples of these genes are BIM, p21 and Sprouty2. The pro-apoptotic factor BIM is a miR-17 family-target (Fontana et al., 2008). Mutant BRAF and NRAS induce the MAPK-dependent phosphorylation and consequent proteosomal degradation of BIM in order to provide melanoma cells with resistance to anoikis (Akiyama et al., 2009). The CDK inhibitor p21, another target of miR-17 family (Fontana et al., 2008), is repressed transcriptionally by TBX2 and TBX3 transcription factors which are frequently amplified in melanoma and are downstream effectors of the MAPK pathway (Bennett, 2008). Sprouty2 is a wt BRAF inhibitor that is targeted by miR-21. The downregulation of Sprouty2 observed in melanoma has been hypothesized as an alternative mechanism responsible for the aberrant activation of MAPK pathway besides BRAF mutation (Tsavachidou et al., 2004). Of note, the three genes described above share a common feature as antagonists of the MAPK pathway. Since the PI3K/AKT and the MAPK pathways have been shown to cooperate in melanomagenesis both in vitro and in mouse models (Dankort et al., 2009; Meier et al., 2007), it is tempting to hypothesize that, in the context of PTEN deletion, the deletion of PTENP1 might be an alternative mechanism evolved by the tumor to activate the MAPK pathway. In this respect, it is worth noticing that most of the cases showing both PTENP1 and PTEN deletion (71.4%) harbor wt BRAF.

In conclusion, our data indicate that the recently identified tumor suppressor gene PTENP1 undergoes genomic deletion in human melanoma. Our data also suggest that PTENP1 deletion might be advantageous for the tumor not only because of its PTEN-related function, but also for PTEN-unrelated ones.