Search

The tumor suppressor gene TP53, encoding p53, is expressed asseveral transcripts. The fully spliced p53 (FSp53) transcript enco-desthecanonicalp53protein.The alternativelysplicedp53I2 tran-script retains intron 2 andencodesD40p53 (or DNp53), an isoformlacking first 39 N-terminal residues corresponding to the maintransactivation domain. We demonstrate the formation of G-quad-ruplex structures (G4) in a GC-rich region of intron 3 that modu-lates the splicing ofintron 2. First, we show the formation of G4 insynthetic RNAs encompassing intron 3 sequences by ultravioletmelting, thermal difference spectra and circular dichroism spec-troscopy. These observations are confirmed by detection of G4-in-duced reverse transcriptase elongation stops in synthetic RNA ofintron 3. In this region, p53 pre-messenger RNA (mRNA) containsa succession of short exons (exons 2 and 3) and introns (introns 2and 4) covering a total of 333 bp. Site-directed mutagenesis of G-tracts putatively involved in G4 formation decreased by #30% theexcision ofintron 2 in a green fluorescent protein-reporter splicingassay. Moreover, treatment of lymphoblastoid cells with 360A,a synthetic ligand that binds to single-strand G4 structures, in-creasestheformation ofFSp53mRNAanddecreasesp53I2mRNAexpression. These results indicate that G4 structures in intron 3regulate the splicing of intron 2, leading to differential expressionof transcripts encoding distinct p53 isoforms.IntroductionThe tumor suppressor p53 protein controls antiproliferative responsesto various forms of stress (1). Its function is impaired in .50% ofhuman cancer, mainly by mutation (2). TP53 gene expression iscomplex, with different transcripts encoding isoforms carrying dis-tinctN- andC-termini (3,4).Todate,10isoformshavebeenidentifiedresulting from the usage of alternative promoters, splice sites and/ortranslational initiation sites (5). Several of these isoforms differ intheir N-terminal region. The N-terminus of p53 contains the maintransactivation domain (residues 1–42, transactivation domain I) aswell as the binding site of Hdm2, which targets p53 for proteasomedegradation and regulates p53 stability (1). Transcription of p53 mes-senger RNA (mRNA) from the proximal promoter generates twoproteins with distinct N-terminal domain. The first corresponds tothe canonical p53 protein, assembled from the fully spliced p53(FSp53) mRNA that retains 11 exons. This protein induces p53-me-diated growth suppression in response to stress. The second isoform,D40p53 (or DNp53), is assembled from an alternatively splicedmRNA retaining intron 2 (p53I2) and lacks the first 39 residues,corresponding to transactivation domain I, as well as Hdm2-bindingsite (3,6). The use of an internal promoter located in a region betweenintron 1 and exon 5 generates a third N-terminal isoform, D133p53,which lacks the first 132 residues (5).When expressed in excess to p53, D40p53 inhibits transcriptionalactivity and interferes in the control of cell cycle progression and apo-ptosisbyexertinganegativeeffectontheexpressionofp53-targetgenes(3,6,7). However, the biological circumstances and the molecularmechanisms regulating D40p53 expression are still poorly known. Re-tentionofintron2inp53I2mRNAintroducesseveralstopcodonsinthereading frame of AUG 1, thus preventing the synthesis of a full-lengthp53protein.However,p53I2mRNAcanbetranslatedusingAUG40asinitiation site, generating a protein isoform which differs from thecanonical p53 by the lack of the first 39 residues. Expression ofp53I2 transcript has been reported in cell lines, such as MCF-7, innormallymphocytesandinprimarymelanomaisolates(7,8).However,themechanismthatregulatesthesplicingofp53pre-mRNAintoFSp53or p53I2 is not understood. D40p53 protein isoform can also be pro-duced by internal ribosomal entry site-regulated internal initiation oftranslation using FSp53 mRNA (9,10).In recent years, it has been proposed that tridimensional RNA struc-tures such as G-quadruplexes may play important roles in regulatingsplicing (11). These structures result of the propensity of G-rich se-quences to fold into four-stranded cation-dependent structures (12).They are formed by the interaction of four guanines organized in acyclic Hoogsteen hydrogen bonding arrangement termed a G-quartetand by the stacking of several G-quartets (Figure 1A). It is estimatedthatover376 000sequencesinthehumangenomehavethepotentialtoadopt G-quadruplex structures, most of them located in non-codingregions (13,14).Atthe RNA level, G-quadruplexes may play a numberof roles.Innon-coding RNAs,they can affecttheir structuresand func-tions (15,16). In 5#-untranslated region of mRNAs, G-quadruplexeshave been shown to modulate translation (17,18). When present inintrons, G-quadruplexes can affect the splicing and expression patternsof genessuchashTERT (humantelomerasereversetranscriptase), Bcl-xL or FMRP (Fragile X mental retardation protein) (11,19,20).The sequence of intron 3 in TP53 contains tracts of G bases orga-nized in a pattern similar to the one of regions forming G-quadruplexstructures. Since exon 3 in TP53 is extremely short (22 bp), we rea-soned that motifs located in intron 3 might have an effect on theregulation of the splicing of intron 2. In this study, we provide anevidence for the formation of G-quadruplex in TP53 intron 3 and thatthese G-quadruplex structures may affect the splicing of intron 2,modulating the synthesis of either FSp53 (intron 2 spliced out) orp53I2 (intron 2 retained) mRNAs, which encode different p53 proteinisoforms.Materials and methods