• Aucun résultat trouvé

Alternative Translation Initiation of Human Fibroblast Growth Factor 2 mRNA Controlled by Its 3′-Untranslated Region Involves a Poly(A) Switch and a Translational Enhancer

N/A
N/A
Protected

Academic year: 2021

Partager "Alternative Translation Initiation of Human Fibroblast Growth Factor 2 mRNA Controlled by Its 3′-Untranslated Region Involves a Poly(A) Switch and a Translational Enhancer"

Copied!
8
0
0

Texte intégral

(1)

HAL Id: hal-03044216

https://hal.archives-ouvertes.fr/hal-03044216

Submitted on 7 Dec 2020

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Alternative Translation Initiation of Human Fibroblast

Growth Factor 2 mRNA Controlled by Its

3’-Untranslated Region Involves a Poly(A) Switch and a

Translational Enhancer

Christian Touriol, Myriam Roussigné, Marie-Claire Gensac, Hervé Prats,

Anne-Catherine Prats

To cite this version:

Christian Touriol, Myriam Roussigné, Marie-Claire Gensac, Hervé Prats, Anne-Catherine Prats.

Al-ternative Translation Initiation of Human Fibroblast Growth Factor 2 mRNA Controlled by Its

3’-Untranslated Region Involves a Poly(A) Switch and a Translational Enhancer. Journal of Biological

Chemistry, American Society for Biochemistry and Molecular Biology, 2000, 275 (25), pp.19361-19367.

�10.1074/jbc.M908431199�. �hal-03044216�

(2)

Alternative Translation Initiation of Human Fibroblast Growth

Factor 2 mRNA Controlled by Its 3

ⴕ-Untranslated Region Involves a

Poly(A) Switch and a Translational Enhancer*

Received for publication, October 13, 1999, and in revised form, April 12, 2000 Published, JBC Papers in Press, April 14, 2000, DOI 10.1074/jbc.M908431199

Christian Touriol‡, Myriam Roussigne§, Marie-Claire Gensac, Herve´ Prats, and Anne-Catherine Prats¶

From INSERM U397, Endocrinologie et Communication Cellulaire, Institut Louis Bugnard, C.H.U. Rangueil, Avenue Jean Poulhe`s, 31403 Toulouse Cedex 04, France

Five fibroblast growth factor 2 (FGF-2) isoforms are synthesized from human FGF-2 mRNA by a process of alternative initiation of translation. The regulation of FGF-2 isoform expression by the mRNA 5823-nucleotide-long 3ⴕ-untranslated region containing eight alternative polyadenylation sites was examined. Because previous studies had shown that FGF-2 expression was regulated in primary cells but not in transformed cells, primary human skin fibroblasts were used in this study. Using an approach of cell transfection with synthetic reporter mRNAs, a novel translational enhancer (3ⴕ-TE) was iden-tified in the 1370-nucleotide mRNA segment located up-stream from the eighth poly(A) site. Deletion mutagen-esis showed that the 3ⴕ-TE was composed of two domains with additive effects. The 3ⴕ-TE exhibited the unique feature of modulating the use of FGF-2 alternative ini-tiation codons, which favored the relative expression of CUG-initiated isoforms. Interestingly, the use of an al-ternative polydenylation site removing the 3ⴕ-TE was detected in skin fibroblasts in response to heat shock and cell density variations. At high cell densities, 3ⴕ-TE removal was correlated with a loss of CUG-initiated FGF-2 expression. These data show that the FGF-2 mRNA 3ⴕ-untranslated region is able to modulate FGF-2 isoform expression by the coupled processes of transla-tion activatransla-tion and alternative polyadenylatransla-tion.

Fibroblast growth factor 2 (FGF-2),1also known as the basic fibroblast growth factor, exists as five isoforms with distinct intracellular localizations and functions. The 18-kDa FGF-2 is mostly cytosolic, whereas the four high molecular mass

iso-forms of 22, 22.5, 24, and 34 kDa are nuclear (1–3). The 18-kDa isoform, which is secreted despite the absence of a signal se-quence, is responsible for paracrine and autocrine effects me-diated by FGF-2 receptors (4). The nuclear isoforms are respon-sible for an intracrine receptor-independent effect of FGF-2 (5, 6). Furthermore the constitutive expression of the various iso-forms modifies the cell phenotype in different ways (3, 7, 8).

Like the other members of the nineteen FGF family (9 –14), FGF-2 acts on cell proliferation and differentiation and exhibits an oncogenic potential (3, 7, 8). FGF-2 is an actor of embryo-genesis and morphoembryo-genesis but is also described as a major angiogenic factor involved in wound healing, cardiovascular disease, and tumor neovascularization (15).

The five FGF-2 isoforms only differ by their N-terminal re-gion, as their synthesis results from a process of alternative initiation of translation at five in-frame codons. A canonical AUG codon gives rise to the smallest 18-kDa isoform, whereas three upstream CUG start codons (CUG1, CUG2, and CUG3) give rise to FGF-2 isoforms of 22, 22.5, and 24 kDa (16, 17). A fourth CUG codon (CUG0) located close to the mRNA 5⬘ end has very recently been shown to initiate the synthesis of a bigger isoform of 34 kDa (3). The use of these different start codons is controlled by cis-acting elements present in the mRNA leader sequence (18). One of these elements, identified as an internal ribosome entry site, enables the FGF-2 mRNA to be translated independently of the classical cap-dependent scanning mechanism (19, 20). However the internal ribosome entry site allows expression of the 18 –24-kDa isoforms, whereas that of the 34-kDa isoform is exclusively cap-depend-ent (3). Regulation of the alternative initiation of translation of FGF-2 mRNA is tightly related to cell physiology: the isoforms initiated at CUG1, CUG2, and CUG3 are constitutively ex-pressed in transformed cells, whereas their expression in nor-mal cells is inducible by stress and low cell density (21, 22). Thus, in nontransformed cells, this translational mechanism allows FGF-2 expression to be regulated very quickly in re-sponse to exogenous stimuli.

Several species of FGF-2 mRNA from 1 to 7 kilobases in size and resulting from alternative polyadenylations are expressed (23, 24). 90% of the longest mRNA (6775 nt) is composed of untranslated regions (UTRs) or alternatively translated re-gions with a GC-rich leader of 484 nucleotides (upstream from the AUG codon) and a huge AU-rich 3⬘-UTR of 5823 nucleotides (3, 17).

Numerous reports have demonstrated the involvement of the eukaryotic mRNA 3⬘-UTR in post-transcriptional regulations. Such regulations can occur at the level of mRNA subcellular localization, stability, or translation initiation (25). In particu-* This work was supported by grants from the Association pour la

Recherche sur le Cancer, the Ligue Nationale contre le Cancer, and the Conseil Re´gional Midi-Pyre´ne´es and by European Community Biotech-nology program (subprogram Cell Factory, Actions de Recherches Con-certe´es) Grant 94/99-181. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

‡ Recipient of successive fellowships from the Ministe`re de l’Education Nationale et de la Recherche and from the Ligue Nationale contre le Cancer. Present address: Unite´ de Physiologie Cellulaire et Mole´culaire, CNRS ERS 1590, 31059 Toulouse Cedex, France.

§ Present address: Institut de Pharmacologie et de Biochimie Struc-turale, CNRS UPR 9062, 105, route de Narbonne, 31062 Toulouse Cedex, France.

¶To whom correspondence should be addressed. Tel.: 33-561-32-21-42; Fax: 33-561-32-21-41; E-mail: pratsac@rangueil.inserm.fr.

1The abbreviations used are: FGF-2, fibroblast growth factor 2; UTR, untranslated region; TE, translational enhancer; CAT, chloramphenicol acetyltransferase; nt, nucleotide(s); kb, kilobase(s).

THEJOURNAL OFBIOLOGICALCHEMISTRY Vol. 275, No. 25, Issue of June 23, pp. 19361–19367, 2000

© 2000 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

(3)

lar, translation regulation by cis-acting signals in the mRNA 3⬘-UTR plays an essential role during the development of a wide variety of organisms (26). Several somatic cell messengers are translationally regulated by cis-acting elements present in their 3⬘-UTR, which are mostly translational repressors. How-ever, a translation activatory effect of 3⬘-UTRs has been de-scribed for a small number of mRNAs, mostly of viral origin but

including two human messengers, the amyloid precursor pro-tein and the lipopropro-tein lipase (27, 28).

The unusual length of the FGF-2 mRNA 3⬘-UTR, together with its eight alternative length-modifying polyadenylation sites, prompted us to determine its regulatory role in FGF-2 isoform expression. Because previous studies had shown that FGF-2 expression was regulated in primary cells but not in FIG. 1. Expression and half-life of transfected CAT mRNAs in human skin fibroblasts. Capped and polyadenylated CAT mRNAs bearing different FGF-2 3⬘-UTRs were transcribed in vitro using T3 RNA polymerase. Human skin fibroblasts were transfected with 2 pmol of each mRNA species. A, schema of the FGF-2 mRNA. The complete 6775-nt-long FGF-2 mRNA is shown with the five initiation codons and the positions of the eight poly(A) sites. B, schema of the chimeric CAT mRNAs with different FGF-2 mRNA 3⬘-UTRs corresponding to a cleavage at poly(A) sites A1, A2, A3, A4, and A8. CAT-A0 is the control mRNA devoid of 3⬘-UTR. CAT-A1, A2, A3, A4, and A8 have FGF-2 3⬘-UTRs of 83, 2390, 3073, 4446, and 5823 nt, respectively. Two series of constructs were made with or without the FGF-2 mRNA 5⬘ region (5⬘-CAT and CAT series, respectively). The fusion of FGF-2 leader sequence with CAT sequence gives rise to CAT fusion proteins initiated at the different FGF-2 start codons. C and D correspond to transfection of skin fibroblasts with CAT chimeric mRNAs (without FGF-2 5⬘-UTR) and 5⬘-CAT chimeric mRNAs (with FGF-2 5⬘ leader), respectively, as described under “Experimental Procedures.” CAT activity of the skin fibroblast extracts was measured 14 h after transfection (see “Experimental Procedures.”) and is shown by histograms. For mRNA half-life measurements, seven dishes were transfected in parallel allowing harvesting of the cells at different times after RNA removal. Total RNAs were prepared from the transfected cells and analyzed by Northern blotting with a32

P-labeled probe corresponding to the CAT coding sequence. The mRNA half-life quantification, presented as values on the right of each CAT activity histogram, was obtained using a PhosphorImager (Molecular Dynamics).

FIG. 2. Characterization of the FGF-2 3ⴕ-TE. A, chimeric CAT mRNAs with different deletions or fragments of the FGF-2 3⬘-UTR were used

to transfect human skin fibroblasts as in Fig. 1. CAT-A8⌬TE1 and ⌬TE2 mRNAs were derived from CAT-A8 by removal of nt 5403–6082 (TE1) and nt 6082– 6775 (TE2) counting from FGF-2 mRNA 5⬘, respectively. CAT-TE, -TE1, and -TE2 mRNAs bear the 1370-nt fragment located between A4 and A8 (nt 5403– 6775), the TE1 or the TE2 fragments, respectively, downstream from the CAT sequence. CAT-TEC mRNA bears nt 5599 – 6260 of FGF-2 mRNA, whereas mRNA CAT-TE⌬C bears the TE element with a central deletion of nt 5901–6356. The presence or absence of the FGF-2 5⬘-UTR is indicated on the top of the panels. B and C correspond to transfection of human skin fibroblasts by the CAT mRNA series (without FGF-2 5⬘-UTR) and the 5⬘-CAT mRNA series (with FGF-2 5⬘ leader called here UTR), respectively, as in Fig. 1.

Regulation of FGF-2 Expression by Its mRNA 3

⬘-UTR

(4)

transformed cells, we used primary human skin fibroblasts to carry out this study. We identify here a unique translational enhancer (3⬘-TE) just upstream from the eighth poly(A) site. This translational enhancer, composed of two distinct domains, was able to modulate the alternative initiation of translation. Interestingly, the use of an alternative poly(A) site removing the 3⬘-TE was generated by heat shock and cell density in-crease in human skin fibroblasts.

EXPERIMENTAL PROCEDURES

Plasmid Construction—The pCAT and p5⬘ CAT series of plasmids (see Fig. 1) have been described previously (3, 29). PCAT-A8⌬TE2 was obtained by subcloning the SwaI-BspEI fragment of pCAT-A8 into pCAT-A0 digested by BstEI plus SmaI (see Fig. 2). For pCAT-A8⌬TE1, the NsiI-Klenow-AvrII fragment of pCAT-A8 was subcloned into pCAT-A8 digested by AvrII plus SwaI. The NsiI-Klenow-BspEI frag-ment of pCAT-A8 was introduced into pCAT-A0 digested by BspEI plus

SmaI to obtain pCAT-TE. PCAT-TE1 was obtained by introducing the NsiI-SwaI-Klenow fragment of A8 into a SmaI-digested

pCAT-A0. PCAT-TE2 was obtained by subcloning the SwaI-BspEI fragment of pCAT-A8 into pCAT-A0 digested by BspEI plus SmaI. The

Ecl136II-HindII fragment of pCAT-TE was inserted into SmaI-digested

pCAT-A0 to obtain pCAT-TEc. Plasmid pCAT-TE was treated by

SphI-HindII-Klenow and religated to obtain pCAT-TE⌬c. PKS-TE and -TE1 plasmids (see Fig. 4) were obtained by subcloning fragments NsiI-XbaI and NsiI-SwaI from plasmid pSCT-DOG into pKS digested by PstI plus

XbaI or PstI plus Ecl136II, respectively.

In Vitro Transcription and Cell Transfection by RNA—In vitro

tran-scription, cell transfection, CAT activity determinations, and Western immunoblotting were performed as reported previously (transfection of 5 cm in diameter dishes with 1 ␮g of RNA/dish; Refs. 3 and 29). Luciferase activity was measured using the Promega luciferase assay system.

Purification of Cellular RNA—Total cellular RNA was extracted

using RNAble (Eurobio), as described previously (29). Poly(A)⫹RNA was prepared with the Dynabeads® mRNA direct kit (Dynal France). Cell monolayers were lysed with the binding buffer and incubated with Dynabeads oligo(dT)25for 5 min. The mRNAs were isolated by magnetic separation, then washed, and treated with the Dynal magnetic concentrator.

Northern Blotting—Northern blotting was performed using the

NorthernMaxTMkit provided by Ambion. DNA probes were labeled with [32P]dCTP using a random priming Multiprime kit (Amersham Phar-macia Biotech). Poly(A)⫹cellular RNA (5 ␮g/lane) was subjected to electrophoresis through 1.2% formaldehyde/agarose gels, electrotrans-ferred to HybondN nylon membrane (Amersham Pharmacia Biotech), and hybridized in the conditions described by the manufacturer.

RESULTS

A Translational Enhancer Is Present in the Distal 1370 nt of the FGF-2 mRNA 3⬘-UTR—The potential role of the FGF-2

mRNA 3⬘-UTR in translation regulation was studied by intro-ducing five of the different FGF-2 mRNA 3⬘-UTRs (measuring 83, 2390, 3073, 4446, and 5823 nt, respectively) downstream from the CAT reporter gene (Fig. 1, A and B). The CAT se-quence followed by the various 3⬘-UTRs was also fused down-stream from the FGF-2 mRNA leader sequence to evaluate the possibility of interactions between the 5⬘ and the 3⬘ regions of FGF-2 mRNA. These constructs were made in a Bluescript KS derived vector with a 70-nt-long poly(A), downstream from the T3 promoter. Capped and polyadenylated CAT mRNAs were transcribed in vitro and used for cell transfection. In this way we were able to avoid interference from the alternative poly-adenylation process and analyze the expression of one mRNA species at a time.

Human skin fibroblast cells were transfected with the CAT mRNAs bearing the different FGF-2 3⬘-UTRs, with or without the FGF-2 mRNA 5⬘-region. CAT mRNA expression was first analyzed by CAT activity measurement (Fig. 1, C and D); CAT-A1 mRNA with the shortest 3⬘-UTR was efficiently ex-pressed, in comparison to the A0 control, whereas CAT-A2, -A3, and -A4 mRNA expressions were very inefficient (Fig. 1C). The same profiles were obtained in the presence of the FGF-2 mRNA leader (Fig. 1D). The measurement of mRNA half-life indicated that these different CAT activities resulted from the presence of a destabilizing element located between the first and second poly(A) sites, as already shown in a recent report (Fig. 1, C and D, on the right; Ref. 29).

However, the expression of CAT-A8 mRNA was two times greater than that of CAT-A4 despite the slightly shorter half-life time (Fig. 1C). In the presence of the FGF-2 mRNA 5⬘ leader, this phenomenon was even more significant, 5⬘ CAT-A8 expression being about 5-fold higher than that of 5⬘ CAT-A4 (Fig. 1D). These observations suggested that a translation ac-tivatory element, active in human skin fibroblasts and more efficient in the presence of the FGF-2 mRNA 5⬘ leader, was present in the 1370-nt fragment located between the fourth and the eighth poly(A) sites of FGF-2 mRNA (called 3⬘-TE hereafter and corresponding to nt 5403– 6775 of the FGF-2 cDNA). FIG. 3. Half-life measurement of the different chimeric mRNAs. The chimeric mRNAs described in Fig. 2 were transfected to perform mRNA half-life measurements, as in Fig. 1. Total RNAs were prepared from the transfected cells and analyzed by Northern blotting with a 32P-labeled probe corresponding to the CAT coding sequence. A and B correspond to transfections with CAT (without FGF-2 5⬘-UTR) and 5⬘-CAT (with FGF-2 5⬘-UTR) mRNA series, respectively. The autoradiography (the period before harvesting is indicated for each point) and the half-life quantification (histograms and values) are given to the right of the name of each chimeric mRNA for each transfected cell type. These were obtained using a PhosphorImager (Molecular Dynamics).

(5)

The FGF-2 3⬘-TE Is Composed of Two Independent Elements (TE1 and TE2) with a Cumulative Effect—Two series of

con-structs were produced and used for RNA transfection in human skin fibroblasts to map this element (Fig. 2A). First the 5⬘ and the 3⬘ halves of the 1370-nt fragment (called TE1 and TE2 and corresponding to nt 5403– 6082 and to nt 6082– 6775, respec-tively) were deleted. The 2-fold increase observed with CAT-A8 compared with CAT-A4 was lost upon deletion of either TE1 or TE2 (Fig. 2B, CAT-A8-⌬TE2 and -⌬TE1). In the presence of the FGF-2 mRNA leader, deletion of the TE1 or TE2 element resulted in loss of about 50% of the 5-fold enhancing effect (Fig. 2C).

The 1370-nt-long TE fragment or one of its subfragments, TE1 or TE2, was therefore added just downstream from the CAT sequence. The results showed that the TE fragment was able to promote a 9-fold activation of CAT translation, whereas this enhancing effect increased to 12-fold in the presence of the FGF-2 mRNA 5⬘ (Fig. 2, B and C, CAT-TE). The TE1 and TE2 fragments were each responsible for about 50% of the enhanc-ing effect, both in the absence and in the presence of the FGF-2 mRNA 5⬘ leader (CAT-TE1 and -TE2). The central part of the TE element (nt 5599 – 6260) did not have any enhancing effect on its own, and deletion of a TE internal fragment (between nt 5901 and 6356) did not affect TE activity (Fig. 4B, CAT-TEC and -TE⌬C).

The mRNA half-life was also measured for the different mRNAs transfected in Fig. 2. This experiment clearly showed that addition or deletion of part or all of the 3⬘-TE element had no significant effect on mRNA stability (Fig. 3) and allowed us to affirm that the effect of the FGF-2 3⬘-TE element occurs at the translational level.

FGF-2 3⬘-TE Activity Is Specifically Inhibited by Antisense RNAs—The alternative approach was to co-transfect the CAT

mRNA with antisense RNAs (Fig. 4). Three AS RNAs, comple-mentary either to the complete 1370-nt TE element (Fig. 4A, AS 4) or to one of its subfragments TE1 and TE2 (AS 5 and AS 6, respectively) were synthesized in vitro. These AS RNAs were prehybridized in a 3- or 10-fold excess to CAT mRNAs bearing either the complete 3⬘-UTR or the TE, TE1 or TE2 3⬘ fragment before skin fibroblast transfection. A control luciferase mRNA was also co-transfected to calibrate the transfection experi-ments. As shown in Fig. 4B, AS RNAs were able to block the translation activation of CAT mRNA expression in a dose-de-pendent manner. Furthermore the blockade was specific; none of the AS RNAs had any effect on expression of the control CAT mRNA (Fig. 4B, CAT-A0). Expression of CAT mRNA with either the complete 3⬘-UTR or the TE element was inhibited by all three AS RNAs (A8 and TE). In contrast CAT-TE1 mRNA was down-regulated only by AS RNA4 and 5, whereas CAT-TE2 was down-regulated only by AS RNA4 and 6 FIG. 4. Targeting the FGF-2 mRNA translational enhancer with antisense RNAs. A, schema of the CAT-A8 mRNA with the positions of the TE, TE1 and TE2 elements and those of their corresponding antisense RNAs AS4, AS5, and AS6 used below (arrows). B, AS RNAs targeting the TE, TE1, or TE2 elements (AS4, AS5 and AS6, respectively) were synthesized in vitro and used in a 3- or 10-fold excess in skin fibroblast co-transfection with CAT or 5⬘ CAT-A8, -A0, -TE, -TE1, or -TE2 mRNAs. RNA-RNA hybridization was obtained by denaturing the mix of CAT mRNA with AS RNA for 2 min at 95 °C in H2O and adding culture medium before returning slowly to 20 °C. The RNA hybrids were used for skin fibroblast RNA transfection as in Fig. 1. 2 pmol of luciferase mRNA were used as internal standard in each co-transfection experiment to calibrate transfection efficiency. Cells were harvested 8 h after transfection for measurements of CAT and LUC activities. Histograms correspond to CAT/LUC activities obtained with the different transfected RNA hybrids. Left panels, CAT mRNA (⫺ 5⬘-UTR); right panels: 5⬘ CAT mRNAs (⫹ FGF-2 5⬘-UTR). C) CAT-A8, -TE, -TE1, and -TE2 mRNAs were transfected in the absence or in the presence of AS RNAs, and the cells were harvested 0 –240 min after transfection for half-life measurement. Total RNAs were purified, and Northern blots were performed as in Fig. 3. The half-life calculation is represented by histograms. AS RNA absence or presence is indicated on the left of each histogram.

Regulation of FGF-2 Expression by Its mRNA 3

⬘-UTR

(6)

(CAT-TE1 and CAT-TE2). Similar results were obtained with CAT mRNAs bearing the FGF-2 5⬘mRNA leader (Fig. 4B, right

panels).

The stability of the co-transfected mRNAs was measured to rule out the possibility that the AS RNAs might affect RNA stability rather than translation. Fig. 4C shows that the pres-ence of one or other competitor had no influpres-ence on mRNA stability.

In conclusion, both deletion and antisense approaches dem-onstrate the presence of a translational enhancer within the distal 1370 nt of the FGF-2 mRNA 3⬘-UTR, between the fourth and eighth poly(A) sites. This TE is specifically active in human primary skin fibroblasts and seems to be composed of two separate elements (TE1 and TE2) with a cumulative effect.

The FGF-2 mRNA Translational Enhancer Modulates Alter-native Initiation of Translation—The FGF-2 3⬘-TE was

espe-cially active in the presence of the FGF-2 mRNA leader (Fig. 1). Because this leader contains five alternative start codons (3, 17), we looked for selective activation of one or more initiation codons by the 3⬘-TE, which would result in a modulation of the protein isoforms ratio.

Human skin fibroblasts were transfected by 5⬘ CAT mRNAs either devoid of 3⬘-UTR or containing the 3⬘-TE element (see Fig. 6, 5⬘ CAT-A0 and 5⬘ CAT-TE, respectively). To evaluate the ratio of the FGF-CAT isoforms rather than the global activa-tion, equivalent amounts of global CAT activity (measured in cpm) from each transfection assay were analyzed by Western immunoblotting with anti-CAT antibodies (Fig. 5). The results

showed that the TE element favored the expression of the CUG-initiated isoforms. Thus the TE element, in addition to its enhancing effect on global translation, was able to modulate the balance of the different FGF-2 isoforms.

A Switch of Alternative Poly(A) Site Use Occurs in Skin Fibroblasts upon Heat Shock—We have shown in previous

reports that expression of the FGF-2 CUG-initiated forms, although constitutive in transformed cells, was modulated in human skin fibroblasts in response to stress and to cell density (21, 22). The effect of the FGF-2 3⬘-TE on alternative initiation of translation observed in Fig. 5 prompted us to look for a putative change in the FGF-2 mRNA 3⬘-UTR length in re-sponse to heat shock.

We therefore analyzed the length of the endogenous FGF-2 mRNA 3⬘-UTR in skin fibroblasts in response to heat shock treatment (Fig. 6). This was performed by reverse transcrip-tion-polymerase chain reaction or Northern blots using poly(A)⫹mRNAs in the latter case (Fig. 6, B and C). Whereas both approaches clearly showed a decrease in the A8-cleaved mRNA as a function of heat shock duration, Northern blot FIG. 5. Analysis of FGF-2 3ⴕ-TE influence on FGF-CAT isoform

expression. 5⬘ CAT-A0 and 5⬘ CAT-TE mRNAs were used for skin fibroblast transfection as in Fig. 3. In addition to CAT activity meas-urement, cell extracts were analyzed by Western immunoblotting using anti-CAT antibody (see “Experimental Procedures.”). The same CAT activity equivalent of each sample was run on the 12.5% polyacrylamide gel to evaluate variations in the isoform ratio rather than the global activation already shown in Fig. 3. The percentage of CUG- and AUG-initiated forms in relation to the total FGF-CAT expression is shown by histograms under each lane. The migration of the different isoforms (CUG1, CUG2/3, and AUG) is indicated. The CUG0-initiated isoform was not detectable in these experiments. This experiment was repeated at least five times, and the picture corresponds to a representative experiment.

FIG. 6. Analysis of FGF-2 mRNA polyadenylation after heat shock in human skin fibroblasts. A, schema of the FGF-2 mRNA with its 5⬘- and 3⬘-untranslated regions. The oligonucleotides used for polymerase chain reaction (PCR) and the probe used for Northern blot are shown below. B, reverse transcription-polymerase chain reaction (RT PCR) was performed using 1 ␮g of total skin fibroblast RNAs treated for increasing times of heat shock (45 °C). The primers, allowing quantitation of the A8 mRNA, are shown in A. Polymerase chain reac-tion was performed for a variable number of cycles to obtain a semi-quantitative analysis. The length of heat shock treatment (in min) is indicated on the top of the lanes. C, poly(A)mRNAs were prepared from skin fibroblasts treated for increasing times of heat shock (up to 60 min; lanes 1–5) or seeded at different cell densities (lanes 6 – 8) and 5␮g of each point was analyzed by Northern blotting using a32

P-labeled DNA probe corresponding to the FGF-2 coding sequence (see “Experi-mental Procedures”). The positions of the 7-kb FGF-2 mRNA (A8) and of the 3.6-kb mRNA (A2) are shown by arrows.

(7)

showed that the A8 mRNA decrease was compensated by an increase in a 3.6-kb mRNA, probably corresponding to the A2-cleaved mRNA (Fig. 6C, lanes 1–5). Thus, heat shock did not affect the total FGF-2 mRNA amount (as had been shown in a previous report (21)) but resulted in switch of the poly(A) site use from A8 to A2.

The Use of Alternative Poly(A) Site Switches with Cell Den-sity Variations—We also looked at the polyadenylation profile

in response to cell density variations. Human skin fibroblasts were seeded at increasing densities, and the corresponding poly(A)⫹mRNAs were analyzed by Northern blot. As shown in Fig. 7A (lanes 1–3) and consistently with a previous analysis made in RPE cells (30), the large 7-kb FGF-2 mRNA disap-peared at high cell density, whereas the 3.6-kb mRNA became apparent. Cell density increase was accompanied by a decrease of total mRNA accumulation, as described previously (22).

The expression of endogenous FGF-2 proteins was analyzed in skin fibroblasts at different densities by Western immuno-blotting. The CUG-initiated isoforms were detected at low but not at intermediary and high cell densities (Fig. 7B), in con-cordance with previous observations (22). In addition, skin fibroblasts were transfected with FGF-CAT mRNAs having or lacking the 3⬘-TE (5⬘ CAT-TE⌬C and 5⬘ CAT-A0, respectively). In both cases we observed a regulation of the CUG-initiated FGF-CAT expression that decreased when cell density in-creased (Fig. 7, C and D). However, at intermediary density, these isoforms were not expressed in the absence of 3⬘-TE (Fig. 7C), where they were clearly detected in its presence (Fig. 7D).

DISCUSSION

In this report, we demonstrate the presence of 3⬘-TE in the distal part of the FGF-2 mRNA 3⬘-UTR, just upstream from the eighth poly(A) site. The 3⬘-TE, composed of two distinct struc-tural domains (called TE1 and TE2), is able to modulate the alternative initiation of FGF-2 mRNA translation. Interest-ingly, the use of poly(A) site switches in response to heat shock and cell density (21, 22). The correlation of 3⬘-TE removal with the disappearance of the CUG-initiated isoforms at increa-sing cell densities suggests a physiological role of the 3⬘-TE in the control of FGF-2 isoforms expression related to cell proliferation.

The RNA transfection approach used in this study was cru-cial to study the translational control of an mRNA subjected to a process of alternative polyadenylation. This procedure en-ables us not only to get rid of interferences with the transcrip-tion, splicing, and polyadenylation processes but also to easily carry out mRNA half-life measurements to demonstrate that the observed effect occurred at the translational level.

The presence of a translational enhancer in a cellular polya-denylated mRNA, which, in addition, regulates the alternative initiation of translation, is novel. An enhancing effect by the 3⬘-UTR has been observed for nonpolyadenylated histone mRNAs, the 3⬘ stem-loop of which mimics a poly(A) tail (31), and for amyloid precursor protein and lipoprotein lipase mRNAs, which interestingly contain two alternative poly(A) sites (27, 28). A few 3⬘ translational enhancers have also been characterized in uncapped and/or unpolyadenylated viruses. A pseudoknot-rich domain in tobacco mosaic virus mRNA 3 plays a poly(A) tail role, and a translational enhancer that mimics a 5⬘ cap has been found in the barley yellow dwarf virus (32–34). Two additional enhancers have recently been charac-terized in the 3⬘-UTRs of rotavirus and hepatitis C virus non-polyadenylated mRNAs (35, 36). The rotavirus mRNA 3⬘-UTR binds the viral protein NSP3A and interacts with the eIF-4G factor, whereas the HCV 3⬘-TE binds the polypyrimidine tract binding protein, a cellular protein that also binds to the inter-nal ribosome entry site located in 5⬘ of the messenger. Both 3⬘ enhancers appear to replace the poly(A) function, because they are involved in the generation of a physical link between the 3⬘ and 5⬘ mRNA ends (37). Sequence alignment of the FGF-2 3⬘-TE and viral 3⬘-TEs did not provide any interesting sequence similarity (not shown). However, the 18-nt sequence conserved among luteovirus (including barley yellow dwarf virus) was found in part at two positions of the FGF-2 3⬘-TE, one in the TE1 element and the other one in the TE2 element (not shown). The middle portion of this consensus has been proposed to base pair to a region of the 3⬘ end of 18 S rRNA (34).

The viral 3⬘ translational enhancers concern mRNAs devoid of cap and/or of poly(A) tail. This makes it easy to understand that the function of such translational enhancers is to mimic either the cap or the poly(A). However, the FGF-2 mRNA is both capped and polyadenylated. Consequently the function of its translational enhancer must be more subtle. We show that the FGF-2 3⬘-TE not only activates global translation initiation but also modulates the relative use of the alternative initiation codons. Thus it modifies the balance between the different FGF-2 isoforms (Fig. 5). Given the different localizations and functions of these isoforms, such a function of the translational enhancer is of great importance.

Interestingly, the FGF-2 3⬘-TE is active in primary skin fibroblasts but poorly efficient in COS-7, HeLa, and SK-Hep-1 cells (Ref. 29 and data not shown). Furthermore the endoge-nous FGF-2 mRNA in these transformed cells mostly exhibits 3⬘-UTRs devoid of 3⬘-TE and thus cannot be subjected to the poly(A) site switch observed in skin fibroblasts (Ref. 29 and Fig. FIG. 7. Analysis of FGF-2 mRNA and protein isoform

expres-sion in human skin fibroblasts at different cell densities. Skin fibroblasts were seeded at different cell densities (lanes 1–3 correspond to 5 ⫻ 105, 106, and 2⫻ 106cells seeded per 9-cm diameter dish, respectively). A, Northern blotting was performed as for Fig. 6C. The positions of the 7-kb FGF-2 mRNA (A8) and of the 3.6-kb mRNA (A2) are shown by arrows. B, Western immunoblotting were performed with anti-FGF-2 antibodies as previously (3). C and D, skin fibroblasts were transfected with mRNAs 5⬘ CAT-A0 (C) or 5⬘ CAT-TE⌬C (D), which is a fully active 3⬘-TE derivative, as in Fig. 5. Then they were seeded at different densities. Western immunoblotting was performed with anti-CAT antibodies as in Fig. 5. The migration of the different FGF-2 or FGF-CAT isoforms (CUG1, CUG2/3, and AUG) is indicated.

Regulation of FGF-2 Expression by Its mRNA 3

⬘-UTR

(8)

6). These observations point to a regulatory process working in normal cells but not in transformed cells.

FGF-2 expression seems to be post-transcriptionally con-trolled at the three levels of polyadenylation, RNA stability and translation (Refs. 18, 21, 22, and 29 and this report). The existence of such concomitant post-transcriptional regulations concerns a growing number of messengers coding for regula-tory proteins. Amyloid precursor protein mRNA, the transla-tion of which is regulated by alternative polyadenylatransla-tion, is also controlled at the stability level by one element of its 3 ⬘-UTR (27, 38). Several other mRNAs are regulated at the trans-lation and stability levels, whereas others are regulated by a coupled process of polyadenylation and stability. Furthermore VEGF mRNA expression, regulated at the levels of translation, stability, and possibly polyadenylation, is also controlled by a process of alternative splicing (39). This leads us to hypothesize that the genes coding for regulatory proteins, which are often regulated transcriptionally, are also controlled by several post-transcriptional mechanisms. These multiple control levels per-mit a subtle regulation of gene expression, which is very im-portant for genes like FGF-2 whose abnormal expression has drastic consequences on cell proliferation and differentiation .

Acknowledgments—We thank S. Vagner for helpful discussion, R.

Couret for pictures, and D. Warwick for English proofreading. We also thank Prof. Costagliola for human skin samples.

REFERENCES

1. Bugler, B., Amalric, F., and Prats, H. (1991) Mol. Cell. Biol. 11, 573–577 2. Brigstock, D. R., Sasse, J., and Klagsbrun, M. (1991) Growth Factors 4,

189 –196

3. Arnaud, E., Touriol, C., Boutonnet, C., Gensac, M. C., Vagner, S., Prats, H., and Prats, A. C. (1999) Mol. Cell. Biol. 19, 505–514

4. Mignatti, P., Morimoto, T., and Rifkin, D. B. (1992) J. Cell. Physiol. 151, 81–93 5. Bikfalvi, A., Klein, S., Pintucci, G., Quarto, N., Mignatti, P., and Rifkin, D. B.

(1995) J. Cell Biol. 129, 233–243

6. Davis, M. G., Zhou, M., Ali, S., Coffin, J. D., Doetschman, T., and II, G. W. D. (1997) J. Mol. Cell. Cardiol. 29, 1061–1072

7. Couderc, B., Prats, H., Bayard, F., and Amalric, F. (1991) Cell Regul. 2, 709 –718

8. Quarto, N., Talarico, D., Florkiewicz, R., and Rifkin, D. B. (1991) Cell Regul. 2, 699 –708

9. Mason, I. J. (1994) Cell 78, 547–552

10. Yamaguchi, T. P., and Rossant, J. (1995) Curr. Opin. Genet. Dev. 5, 485– 491 11. Smallwood, P. M., Munoz-Sanjuan, I., Tong, P., Macke, J. P., Hendry, S. H. C., Gilbert, D. J., Copeland, N. G., Jenkins, N. A., and Nathans, J. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 9850 –9857

12. Hoshikawa, M., Ohbayashi, N., Yonamine, A., Konishi, M., Ozaki, K., Fukui, S., and Itoh, N. (1998) Biochem. Biophys. Res. Commun. 244, 187–191 13. Ohbayashi, N., Hoshikawa, M., Kimura, S., Yamasaki, M., Fukui, S., and Itoh,

N. (1998) J. Biol. Chem. 273, 18161–18164

14. Nishimura, T., Utsunomiya, Y., Hoshikawa, M., Ohuchi, H., and Itoh, N. (1999) Biochim. Biophys. Acta 1444, 148 –151

15. Bikfalvi, A., Klein, S., Pintucci, G., and Rifkin, D. B. (1997) Endocr. Rev. 18, 26 – 45

16. Florkiewicz, R. Z., and Sommer, A. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 3978 –3981

17. Prats, H., Kaghad, M., Prats, A. C., Klagsbrun, M., Le´lias, J. M., Liauzun, P., Chalon, P., Tauber, J. P., Amalric, F., Smith, J. A., and Caput, D. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 1836 –1840

18. Prats, A. C., Vagner, S., Prats, H., and Amalric, F. (1992) Mol. Cell. Biol. 12, 4796 – 4805

19. Kozak, M. (1989) J. Cell Biol. 108, 229 –241

20. Vagner, S., Gensac, M. C., Maret, A., Bayard, F., Amalric, F., Prats, H., and Prats, A. C. (1995) Mol. Cell. Biol. 15, 35– 44

21. Vagner, S., Touriol, C., Galy, B., Audigier, S., Gensac, M. C., Amalric, F., Bayard, F., Prats, H., and Prats, A. C. (1996) J. Cell Biol. 135, 1391–1402 22. Galy, B., Maret, A., Prats, A. C., and Prats, H. (1999) Cancer Res. 59, 165–171 23. Bensaid, M., F. Malecaze, H. Prats, F. Bayard, and J. P. Tauber. (1989) Exp.

Eye Res. 45, 801– 813

24. Knee, R. S., Pitcher, S. E., and Murphy, P. R. (1994) Biochem. Biophys. Res. Commun. 205, 577–583

25. Wickens, M., Anderson, P., and Jackson, R. J. (1997) Curr. Opin. Genet. Dev. 7, 220 –232

26. Curtis, D., Lehmann, R., and Zamore, P. D. (1995) Cell 81, 171–178 27. de Sauvage, F., Kruys, V., Marinx, O., Huez, G., and Octave, J. N. (1992)

EMBO J. 11, 3099 –3103

28. Ranganathan, G., Ong, J. M., Yukht, A., Saghizadeh, M., Simsolo, R. B., Pauer, A., and Kern, P. A. (1995) J. Biol. Chem. 270, 7149 –7155

29. Touriol, C., Morillon, A., Gensac, M.-C., Prats, H., and Prats, A.-C. (1999) J. Biol. Chem. 274, 21402–21408

30. Bost, L. M., and Hjelmeland, L. M. (1993) Growth Factors 9, 195–203 31. Gallie, D., Lewis, N., and Marzluff, W. (1996) Nucleic Acids Res. 24,

1954 –1962

32. Leathers, V., Tanguay, R., Kobayashi, M., and Gallie, D. R. (1993) Mol. Cell. Biol. 13, 5331–5347

33. Wang, S., and Miller, W. A. (1995) J. Biol. Chem. 270, 13446 –13452 34. Wang, S., Browning, K. S., and Miller, W. A. (1997) EMBO J. 16, 4107– 4116 35. Piron, M., Vende, P., Cohen, J., and Poncet, D. (1998) EMBO J. 17, 5811–5821 36. Ito, T., Tahara, S. M., and Lai, M. M. (1998) J. Virol. 72, 8789 – 8796 37. Wells, S. E., Hillner, P. E., Vale, R. D., and Sachs, A. B. (1998) Mol. Cell 2,

135–140

38. Zaidi, S. H., and Malter, J. S. (1994) J. Biol. Chem. 269, 24007–24013 39. Tischer, E., Mitchell, R., Hartman, T., Silva, M., Gospodarowicz, D., Fiddes, J.

C., and Abraham, J. A. (1991) J. Biol. Chem. 266, 11947–11954

Références

Documents relatifs

monter (d'abord en zeir-anpin) ; malkhout se retrouve alors dépourvue de force limitante mais ne reçoit pas pour autant la Lumière Supérieure sans limite, au contraire elle

Das FZA verweist in seinem Anhang III sowohl auf die Richtlinie 2005/36 über die Anerkennung von Berufsqualifikationen als auch auf die Richtlinie 98/5 zur

Mouse hybridoma cells having CD44 cell surface receptors and anti- CD44 antibodies were used to establish a proof-of-concept for this method.. Both the capture antibodies and the

Empirical Research On The Optimization Of The Frequency Parameters Of ” Chirp ” Sequences Used In Contrast Ultrasound Imaging.. International Conference on Advances in

Relative apparent resistivity variation profile over depth using the Transmission configuration 372. during the drying of the concrete specimen: (a) at 20 °C; (b) at

Depending on the model size, strong scaling (speedup at a larger number of cores) was satisfactory for the explicit method, and more limited for the implicit method.. The im-

In this paper, we design a flat shell finite element model in order to simulate the fetal head deformation during childbirth.. This new method also guarantees the

For an ex- tractive laser-based system with a 200 m path length, in or- der to achieve a 50 ppt v detection limit, a spectrometer would need a detection sensitivity corresponding