PTEN isoforms with dual and opposing function

Abstract

PTEN is a phosphatase that functions as a tumour suppressor by antagonizing the PI3K–AKT pathway. However, a study now demonstrates that translational variants of PTEN enable new interactions between PTEN and the COMPASS complex, identifying a new role for PTEN in modifying gene expression via COMPASS-mediated histone H3 lysine 4 methylation.

In addition to mRNA isoform variation due to alternative splicing, a single mRNA can have translational variants by using an alternative translation initiation codon. Most protein translational variants have been understudied despite the fact that translational variants can give rise to protein isoforms bestowed with new functions, protein–protein interactions, sites of subcellular localization, and post-translational modifications. In this issue of Nature Cell Biology, Shen et al. report a novel function of two such isoforms of PTEN¹. PTEN is a dual-specificity protein and lipid phosphatase and a well-established tumour suppressor commonly affected by loss-of-function mutations in many sporadic forms of cancer as well as in the germline of patients with PTEN hereditary tumour syndromes. The most well documented cellular substrate for PTEN is PIP3, which is hydrolyzed to PIP2. PTEN thereby blocks PI3K signalling by inhibiting PIP3-dependent processes such as cytoplasmic membrane recruitment and activation of AKT (the “Ak strain transforming” serine/threonine kinase)^2,3. In addition, several studies have revealed that PTEN exerts part of its tumour-suppressive function by acting as a scaffold protein in both the nucleus and the cytoplasm⁴. In contrast to these prior data, two longer PTEN translational variants are described here that act as oncogenes by regulating histone methylation within the nucleus.

The most well studied form of PTEN (‘canonical PTEN’) is a 403 amino acid protein produced from a canonical AUG start codon in PTEN mRNA. However, two evolutionarily conserved protein isoforms of PTEN, known as PTENα and PTENβ, result from translation initiation within the 5′ untranslated region (UTR) of PTEN mRNA using in-frame, non-AUG start codons. PTENα is initiated upstream of PTENβ, resulting in protein isoforms with an additional 173 and 146 amino acids, respectively, to their N terminus compared to canonical PTEN (Fig. 1).

Fig. 1 |. Translational variants of PTEN.

PTEN mRNA (bottom) is translated into the canonical 403 amino acid (aa) PTEN protein or its longer translational variants PTENα and PTENβ depending on which translation initiation codon is used. PTENα and PTENβ contain an N-terminal extension unique to these isoforms that encodes a nuclear localization signal (NLS) and interaction domains for binding WDR5, FBXW11, and USP9X. FBXW11 and USP9X regulate ubiquitination and deubiquitination, respectively, of PTEN at lysines 235 and 239 (K235 and K239). “C-tail” represents the C-terminal tail domain of PTEN.

Of note, endogenous PTENα and PTENβ isoform levels are much lower than those of canonical PTEN, consistent with the concept that use of non-AUG start codons is typically less efficient than mRNA translation from canonical AUG translation start sites⁵. However, the N-terminal extension of PTENα/β is predicted to be largely unstructured and may have the potential to interact with a variety of other proteins⁶. Spurred by this hypothesis and the observation that PTENα/β were more sensitive to proteasomal degradation than canonical PTEN, Shen et al.¹ performed immunoprecipitation-coupled mass spectrometry to identify ubiquitin ligasesor deubiquitinases bound specifically to PTENα/β protein isoforms. The idea that the N-terminal domains of PTEN could impact PTEN stability is indeed aligned with the fact that PTEN is known to be regulated by many post-translational modifications, and its tumour-suppressive functions can be regulated by its binding partners^7,8. Mass spectrometry studies led the authors to discover that FBXW11 (a ubiquitin ligase subunit known as “F-box and WD repeat domain containing 11”) ubiquitinates PTENα/β and USP9X (a deubiquitinase known as “ubiquitin specific peptidase 9 X-linked”) deubiquitinates PTENα/β, but not canonical PTEN. They also found that FBXW11 promoted tumourigenesis, whereas USP9X inhibited it, at least partially owing to their effects on stability of PTENα/β.

Notably, selective depletion of PTENα or PTENβ inhibited tumourigenesis, whereas their overexpression accelerated tumourigenesis. Curiously, however, neither PTENα nor PTENβ appeared to regulate AKT activation. To explain how PTENα/β might regulate tumour growth, they identified that in addition to interactions between PTENα/β and the aforementioned ubiquitinating–deubiquitinating complexes, PTENα/β also uniquely interacted with a core component of the MLL and SET1 (hCOMPASS) histone H3 lysine 4 (H3K4) methyltransferase complex named WDR5. Interestingly, RNA sequencing revealed that PTENα/β-specific gene regulation was restricted to eight genes. Two of these, NOTCH3 and SLC12A5, were found to have PTENα/β and WDR5 co-localized at their promoters by chromatin immunoprecipitation sequencing (ChIP-seq), which increased H3K4 trimethylation (H3K4me3). The interaction of PTENα and WDR5 was essential for tumourigenesis. Remarkably, the N-terminal extension of PTENα/β also encodes an additional nuclear localization sequence for PTEN, which would enforce the chromatin association of PTENα/β highlighted by this work.

Overall, the data from Shen et al.¹ point to an oncogenic role for PTENα/β as opposed to the well-established tumour-suppressive role of PTEN. Though these data are intriguing, it is unclear how these results are reconciled with a prior study identifying that PTENα is actually secreted from the cell and functions as a secretory PI3K antagonist and exogenous tumour suppressor⁹. In fact, in this prior study it was noted that the N-terminal extension on PTENα (referred to as “PTEN-long”) contains a polyarginine stretch with homology to known cell-permeable peptides that allows PTENα to enter cells and inhibit PI3K signalling. Consistent with this, Hopkins et al. showed that intraperitoneal injection of PTENα suppressed tumour growth in mouse xenografts, suggesting the intriguing concept that administration of PTENα could serve as a novel cancer therapy⁹. The potential paracrine effects of PTEN expression are also bolstered by interesting prior genetic data identifying that deletion of PTEN in cells surrounding a tumour mass can promote cancer growth in a cell non-autonomous manner¹⁰.

In addition, other activities for PTEN translational variants have been posited including localization of PTEN to mitochondria to regulate mitochondrial metabolism¹⁰ and a role in ribosomal DNA transcription¹⁰. Although phosphatase-independent activities for PTEN have been described¹¹ and PTEN translational variants maintain the intact phosphatase domain, it is unclear why these longer PTEN isoforms would not regulate PI3K activity. It may be due to the cellular localization of these isoforms within the nucleus resulting in a physical barrier between PTENα/β and cytoplasmic substrates, as suggested by the current findings. Finally, it is important to note prior work suggesting that deletion of PTEN results in upregulation of CHD1, a chromatin helicase that binds and maintains H3K4me3 (ref.¹²). Whether and how alterations in PTEN protein isoforms modify this role of PTEN in regulation of CHD1 is unclear.

The current study will be important to reconcile with other previously described roles for PTENα/β given the potential clinical implications and therapeutic applicability of targeting PTEN translational variants versus administering them as therapeutics. It will also be interesting to understand what regulates expression of canonical PTEN versus these longer translational PTEN variants. Although prior work has identified that sequence and structural elements in PTEN mRNA regulate the use of alternative start codons¹⁰, it may also be very interesting to understand whether any trans-acting initiation factors regulate use of translation start sites in PTEN.