The fragile X syndrome results from a transcriptional silencing of the FMR1 gene and the absence of its encoded protein. FMRP is a cytoplasmic RNA-binding protein, whose specific cellular function is still unknown. We present evidence that virtually all detectable cytoplasmic FMRP in mouse NIH 3T3 and human HeLa cells is found strictly in association with mRNA in actively translating polyribosomes. Furthermore, FMRP released from polyribosomes is associated with ribonucleoprotein complexes with sedimentation coefficients of 60-70S and selection on oligo(dT)-cellulose reveals that this association is specific to poly(A)-containing mRNPs. This association with actively translating polyribosomes is not affected by alteration of translational processes induced by serum stimulation and starvation in NIH 3T3 cells, suggesting that FMR1 expression is not cell cycle regulated and that FMRP might have a house-keeping function. FXR2 protein, which is closely related to FMRP, is also detected associated with mRNPs in translating polyribosomes. The results strongly suggest that FMRP might be a mRNA chaperone interacting with mRNP complexes.
The fragile X syndrome is an X-linked disease and the most frequent cause of inherited mental retardation (1 ). The syndrome is associated with an unstable expansion of a CGG repeat located in the 5'-untranslated region of the fragile X mental retardation (FMR1) gene. This expansion is associated with abnormal DNA methylation and correlates with the transcriptional silencing of FMR1 expression in fragile X cases (2 ,3 ). The fragile X syndrome is characterized by mental retardation and associated variable multisystemic phenotypic alterations, and results from the absence of the FMR1 protein.
FMRP contains two KH domains and one RGG box, which are motifs characteristic of RNA-binding proteins (4 ,5 ). Indeed, in vitro-translated FMRP interacts with RNA homopolymers and with some as yet unidentified fetal brain mRNAs (4 ,5 ). FMRP is predominately a cytoplasmic protein (6 ,7 ) that co-fractionates with ribosomes (8 -11 ) and, since it contains both nuclear localization and nuclear export signals (8 ,10 ,12 ,13 ), it has been proposed that FMRP could be involved in mRNA transport from the nucleus to the cytoplasm (8 ). The recent immunoelectron microscopy analysis of FMRP in neurons support this model since it has been shown that a minor fraction of FMRP is detectable in the nucleoplasm and in nuclear pores while the majority is associated with ribosomes in the cytoplasm (14 ).
We and others have shown that cytoplasmic FMRP is associated with actively translating polyribosomes (8 ,9 ), while others (10 ,11 ) reported that the protein interacts with free monosomes. In addition, it is not clear whether FMRP is a component of the ribosome or of other ribonucleoparticles. Since the localization of FMRP with actively translating ribosomes is still controversial, we sought to clarify the putative role of FMRP in the translational machinery by studying its localization in mouse NIH 3T3 cells exposed to different physiological conditions that induce changes in translational processes. We show that cytoplasmic FMRP is strictly associated with mRNA as component of messenger ribonucleoparticles (mRNPs) engaged in translation and can be selected after chromatography with oligonucleotides stretches that have affinity for poly(A)-containing mRNPs.
Given the association of FMRP with polyribosomes engaged in protein synthesis (8 ,9 ), we analyzed FMR1 mRNA and protein levels in cells exposed to different physiological conditions that induce changes in translational processes. The murine fibroblast was chosen as a model system, since protein synthesis can be modulated by the addition or removal of serum factors from the culture medium. Cultures of NIH 3T3 cells were arrested in the G0/G1 phase of the cell cycle by serum starvation for 72 h and subsequently stimulated to enter a division cycle by the addition of fresh medium containing 20% serum. Total RNA was extracted from quiescent and stimulated cultures at different times after addition of serum and subjected to Northern blot analysis. The results showed (Fig. 1 a) a bimodal change in FMR1 mRNA levels with an increase of ~3-fold peaking at ~3 h, followed by a net decrease before returning to normal levels at the late points. The same modulation of FMR1 mRNA levels was obtained in three independent series of time course analyses. We also analyzed the mRNA levels of the two genes termed FXR1 and FXR2 (15 ), closely related to FMR1, the proteins of which have been found to interact with FMRP and to associate also with ribosomes (10 ). The same membrane was probed sequentially with human FXR1 and FXR2 cDNAs (15 ,16 ). The results showed that whereas two FXR1 transcripts (17 ) of ~3.0 and 2.4 kb were resolved, only a single transcript of 3.1 kb was detected for FXR2. Again a clear increase was observed at 3 h post-stimulation for both mRNAs, while at the other time points their levels remained low. In contrast, increased levels of the cell cycle-dependent H4 mRNA was observed 8-9 h after stimulation, coinciding with the beginning of S-phase, while mRNA levels for GAPDH and eIF-1a remained stable and hybridization with the 18S rRNA gene showed even loading of the different RNA samples (data not shown).
Quiescent cells arrested in the G0/G1 phase of the cell cycle contain low amounts of polyribosomes engaged in protein synthesis necessary for basal cell maintenance (20 ,21 ). Addition of serum to quiescent 3T3 cells increases the rate of protein synthesis 2- to 3-fold (22 ). This is paralleled by activation of the translational machinery which is characterized by re-entry of free 80S monosomes into polyribosomes. To determine the fate of FMRP in polyribosomes extracted from quiescent and stimulated cells, post-nuclear supernatants were analyzed by velocity sedimentation through sucrose gradients. In quiescent cells, the majority (80%) of the ribosomes were detected as monomers sedimenting at 80S, while the remaining 20% were distributed in the heavy sedimenting fractions as polyribosomes (Fig. 3 a). Confirming the UV profiles, immunoblot analysis of the proteins from each collected fraction showed that the majority of the ribosomal L7 protein was detected in the 80S fractions while small amounts were observed in the heavy fractions. Immunoblot analyses with mAb 1C3 revealed that the 80-70 kDa isoforms of FMRP co-fractionated with polyribosomes engaged in translation. At 12 h post-stimulation with serum, a dramatic change in ribosomes distribution was observed (Fig. 3 b). A net shift of the inactive monosomes into polysomes was observed and the distribution of the 80-70 kDa set of FMRP remained the same as seen for quiescent cells (Fig. 3 b).
Since the possibility that FMRP binds specifically to monosomes derived from polyribosomes has been ruled out, as shown in the preceding paragraph, we tested a second hypothesis which would imply that the FMRP sedimenting at 60S (9 ,10 ) released from the polyribosomes with EDTA is part of an mRNP complex that fortuitously sediments with or close to the ribosomal large subunit. To test for this possibility, polyribosomes were treated with RNase to degrade the mRNA that links the translating ribosomes together. For convenience, we used HeLa cells, since a stronger FMRP signal is routinely observed using the 1C3 antibody. Whether this is due to higher affinity of the antibody for the human protein or to higher expression levels of FMRP in HeLa cells as compared with NIH 3T3 is not known. Treatment of cytoplasmic extracts with RNase A (100 [mu]g/ml) resulted in the complete destruction of polyribosomes, and FMRP was displaced to the top fractions of the gradient (Fig. 4 b). To discriminate between the deleterious effects of RNase on polyribosome integrity and the removal of FMRP from polyribosomes, the KCl concentration of the post-nuclear fraction was adjusted to 0.5 M. At this salt concentration, no appreciable alteration in polyribosome distribution was observed, whereas FMRP was released and recovered as a non-sedimentable protein at the top of the gradient (Fig. 4 c). These results indicate that FMRP is associated with translating polyribosomes through interactions with RNAs that link the ribosomes together.
Since FMRP has been shown to associate with the related FXR2 protein (10 ), we asked whether the latter protein would also behave as FMRP and be associated with mRNA engaged in polyribosomes. Two key analyses were conducted to answer this question. A duplicate membrane containing protein from a sucrose density gradient of polyribosomes from HeLa extracts, as used in Figure 4 a, was reacted with monoclonal antibody A42 directed against human FXR2 (15 ). The results showed that the majority of FXR2 was detected at the level of polyribosomes (Fig. 7 a). This provides convincing evidence that at least one of the FMRP partners also co-sediments with polyribosomes. To determine whether FXR2 also binds to mRNA, identical membranes containing protein eluted from the oligo(dT)-chromatography, as used in Figure 5 , were tested with antibody A42. The results of such an analysis indicate that FXR2 indeed binds to mRNA (Fig. 7 b). Finally, to discriminate between the two proteins and to eliminate any possibility of cross-reactions, parallel immunoblot analysis of total protein from HeLa cells with antibodies A42 and 1C3 showed clearly the distinct patterns of the 94 kDa FXR2 as compared with the 80-70 kDa FMRP (Fig. 7 c). We also noted, in the three different immunoblots presented here, that in addition to the major FXR2 species, a minor band at 86 kDa was revealed by A42, as previously shown (15 ). Whether this minor species corresponds to a FXR2 degradation product or to a minor isoform is not known.
Using the NIH 3T3 cell model, we observed that FMRP was associated strictly with actively translating polyribosomes. In all instances tested, whether cells were arrested in the G0 phase of the cell cycle or induced to re-enter S-phase, FMRP remained associated with these heavy sedimenting structures and its subcellular distribution was unchanged, as seen after cell fractionation or immunocytochemistry in situ. We observed the same distribution after a physiological stress that induced disaggregation of the majority of polyribosomes, while no free FMRP was detected. The fact that, even when cells are put in minimal surviving conditions, all detectable FMRP remains associated with the scanty polyribosomes constitutes functional in vivo evidence that FMRP is implicated in a basic function, which is compatible with the idea that FMR1 might be a house-keeping gene (25 ) that is widely, if not ubiquitously, expressed in every tested tissue (26 ,27 ).
Our observations clearly show that under all conditions used, FMRP does not sediment with the 80S ribosomes monomer but is associated with polyribosomes, as shown in different cell lines (8 ,9 ). Very recently, Feng et al. (14 ) extended these observations to the animal system by using rat cortex, and similar if not identical results were obtained. A further complication to the assignment of FMRP to a specific particle is the fact that, after dissociation of the polyribosomal structures with EDTA, mRNP particles are heterogeneous in their sedimentation behavior and overlap with both the small and large ribosomal subunits during sucrose gradient centrifugation (24 ,28 ). As previously reported by us (9 ) and others (10 ), the bulk of FMRP after treatment of ribosomes with EDTA sedimented with a value of ~60S in sucrose gradients at a position which induced us to infer that FMRP was associated with the large ribosomal subunit. The results presented here show that this is clearly not the case.
Two different analyses led us to conclude that FMRP is a component of mRNP in actively translating ribosomes as opposed to non-messenger RNPs. First, enzymatic treatment of EDTA-dissociated polyribosomes with low doses of RNAse induces a shift of FMRP sedimentation coefficient while ribosomal subunits remain unaffected; second, we show that the vast majority of FMRP molecules from polyribosomes are captured by poly(A)-binding matrices such as oligo(dT)-cellulose, that this binding is abolished by pre-treatment of the columns with poly(A) or poly(U) oligonucleotides and that the mRNA-FMRP complex can be eluted from the affinity column under conditions known to release mRNPs (24 ,29 ). However, our preliminary observations showed that treatment with RNase A (100 [mu]g/ml) or with micrococcal nuclease (100 U/ml for 5 min at 30oC) did not prevent the binding of FMRP to the column, whether the enzymatic digestion was performed before affinity chromatography or in situ. This behavior points to the possibility that FMRP-mRNA interaction(s) might be localized either on the poly(A) tail of mRNA molecules or close to it within a protected mRNA segment at the 3'-untranslated region that might control mRNA translation (30 ). Consistent with the idea that FMRP is one of the proteins that constitute complexes that interact with mRNA is the fact that FXR2 protein, which interacts with FMRP (10 ), shows identical sedimentation behavior to FMRP and is also retained by the oligo(dT) matrix, suggesting that it may also be part of the same mRNA-protein complex.
The association of virtually all detectable FMRP with polyribosomes in conditions of minimal cellular activity points to the possibility that FMRP may be involved in chaperoning mRNAs that are implicated in basal mechanisms for cell maintenance. Since FMRP contains NLS and NES sequences (8 ,10 ,12 ) similar to sequences in HIV-1 Rev and PKI (31 ,32 ), this chaperoning could start at the source of the pre-mRNA factories as part of hnRNPs in the nucleus. As other known proteins (reviewed in 33 ,34 and references therein), FMRP would then, dynamically or passively, escort mRNAs to the translational machinery and remain there while these mRNAs are translated. This model is compatible with that proposed by Eberhart et al. (8 ).
The results presented here do not answer the crucial question of whether FMRP has affinity for specific mRNAs. The absence of FMRP, such as in the case of the fragile X syndrome, could be critical during an as yet unknown phase in ontogeny of the nervous system that requires high levels of gene expression. Apart from specific mRNAs necessary for the development of the cognitive functions, the plethora of multisystemic phenotypic alterations found in the fragile X syndrome (35 ,36 ) might be the result of a random alteration of additional different mRNAs that could be affected due to the absence of FMRP during development. Whether the closely related FXR1 and FXR2 (15 ), or other as yet undiscovered proteins specific to the neurones, exhibit functional redundancy in a less efficient manner, to compensate for the absence of FMRP, is still an open question.
NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% calf serum (CS) and antibiotics (100 U/ml penicillin, 50 mg/ml streptomycin). For serum starvation, the medium was changed to 0.2% CS when the cultures reached 70% confluence. The cells were then kept in the same medium for 72 h. Under these conditions, <3% of the cells were still engaged in DNA synthesis as determined by [3H]dThd incorporation followed by autoradiography. Serum stimulation of these resting cultures was induced by addition of fresh medium containing 20% CS. For induction of rapid starvation, the culture medium containing 20% CS was removed and the cultures incubated for 60 min in PBS (1.6 mM KH2PO4, 8.4 mM K2HPO4, 150 mM NaCl) supplemented with 1.5 mM MgCl2 and 0.5 mM CaCl2. HeLa S3 cells were propagated and maintained in DMEM supplemented with 10% fetal bovine serum.
Cell monolayers were washed three times with ice-cold PBS, and 3-5*106 cells were lysed in 1 ml of buffer containing 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1.5 mM MgCl2, 100 mM NaF, 10 ng/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 10 U/ml RNasin (Pharmacia) and 0.5% NP-40. Cell lysates were then homogenized by passage through hypodermic needles. A crude nuclear fraction was obtained after centrifugation at 5000 r.p.m. for 5 min, and the post-nuclear fraction was centrifuged further at 100 000 g for 60 min to pellet the ribosomes. The supernatant was centrifuged further at 360 000 g for 90 min, yielding a post-ribosomal pellet and a final supernatant fraction. For polysomes analyses, 4*107 cells were lysed in 1 ml of the above buffer and a post-nuclear fraction was obtained after centrifugation at 10 000 r.p.m. for 10 min. All manipulations were carried out at 4oC.
Aliquots of 1 ml of cytoplasmic extracts, containing 12-16 OD at 260 nm, were analyzed by sedimentation velocity in 15-45% (w/w) linear sucrose gradients made up in 25 mM Tris-HCl pH 7.4, 100 mM KCl, 5 mM MgCl2. After centrifugation in a Sorval TH-641 rotor for 2 h at 38 000 r.p.m. at 4oC, gradients were collected from the top with an Isco UA5 flow-through spectrophotometer set at 254 nm and connected to a gradient collector. For EDTA-mRNPs studies, the fractions of sucrose gradients corresponding to 150-500S were pooled, diluted with 1 volume of buffer and the particles pelleted by centrifugation at 300 000 g for 2 h. The polyribosome pellets were resuspended in 25 mM Tris-HCl, pH 7.4, 100 mM KCl and 25 mM EDTA, and analyzed after centrifugation in linear 5-20% (w/w) sucrose density gradients for 3.5 h at 38 000 r.p.m.
For RNase treatment, polyribosomes were incubated with 100 [mu]g/ml of RNase A (Pharmacia Biotech) for 15 min at 21oC, while EDTA-dissociated polyribosomes were incubated with 0.2 [mu]g/ml RNase A for 5 min at 4oC. After RNase treatments, 10 U/ml RNasin and 0.5% NP-40 were added to the suspension.
To purify poly(A)-containing mRNPs present in polyribosomes, the oligo(dT)-cellulose procedure of Lindberg and Sundquist (24 ) was used, with slight modifications. Aliquots of 1 ml containing 5 OD260 units of EDTA-dissociated polyribosomes collected from the 150-500S fractions in 100 mM KCl, 25 mM Tris-HCl, pH 7.4, 25 mM EDTA solution were mixed with 100 [mu]l of a suspension containing 20 mg of oligo(dT)-cellulose (type 7, Pharmacia Biotech) and incubated at 21oC for 30 min. After washing with 1 ml of the same solution, proteins bound to the column were desorbed with SDS sample buffer (68 mM Tris-HCl, pH 9, 2% SDS, 2% [beta]-mercaptoethanol, 0.01% bromophenol blue and 15% glycerol). Elution of the adsorbed mRNP particles from oligo(dT)-cellulose was done in 25 mM Tris-HCl, pH 7.4, in the presence of 25% formamide. All fractions were adjusted to 1 ml, and identical volumes were analyzed for the presence of FMRP by immunoblotting.
The preparation of total RNA and analysis by Northern blotting were performed as described previously (26 ). Briefly, RNA (5 [mu]g) was subjected to electrophoresis under denaturing conditions, transferred to GS nylon membranes (Gene Screen, New England Nuclear), and fixed to the membrane by UV irradiation (37 ). To check for even transfer, loading and integrity of the RNA, the membranes were stained with methylene blue. Pre-hybridization and hybridization of the membrane were performed as described (37 ). cDNA fragments from FMR1 (6 ), FXR1 (16 ) and FXR2 (15 ) were 32P-labeled and used as probes.
Fractions obtained after differential centrifugation were solubilized in SDS sample buffer (68 mM Tris-HCl, pH 9, 2% SDS, 2% [beta]-mercaptoethanol, 0.01% bromophenol blue and 15% glycerol) in a volume corresponding exactly (i.e. 1 ml) to that of the extraction buffer used initially to lyse the cells, and heat denatured in a boiling bath for 3 min. For nuclear and total fractions, the viscous extracts were sonicated. Aliquots from each fraction obtained from the sucrose gradients were denatured in the presence of 0.5 volume of a 3-fold-concentrated SDS sample buffer. For sucrose fractions obtained from 3T3 cell lysates, proteins were concentrated 10-fold by precipitation with 20% trichloroacetic acid (TCA) and, after an ethanol wash, were solubilized in SDS sample buffer. Proteins were separated by SDS-PAGE on 7.5 or 12% acrylamide slab gels and electrophoretically transferred onto 0.45 [mu]m nitrocellulose membranes (Bio-Rad). Immunodetection of FMRP and of L7a ribosomal protein was carried out using monoclonal antibody 1C3 (previously described as 1a in 6 ) and rabbit anti-L7a serum (38 ) respectively, as described (9 ). Protein concentration was determined using the Bradford method after TCA precipitation of the extracted proteins and re-solubilization in 0.2 M NaOH followed by neutralization with 0.2 M HCl.
Cells grown on glass coverslips were fixed with cold formaldehyde/acetone/methanol (2/19/19 by vol.) for 20 min at -20oC. Immunoreaction was performed with anti-FMRP monoclonal antibody 1C3 followed by biotinylated anti-mouse IgG and fluorescein isothiocyanate (FITC)-conjugated streptavidin. Stained samples were mounted in PBS-glycerol containing 0.1% p-phenylenediamine (Sigma) as an anti-fading agent. Fluorescence was viewed with a Leitz DMRB microscope equipped for epifluorescence illumination using a 100* oil immersion objective. Fluorescent micrographs were taken using identical exposure times.
We are grateful to D. Devys and J.-L. Mandel for providing anti-FMRP antibodies, to S. Kozma and A. Ziemiecki for the anti-ribosomal sera, and to M. Siomi and G. Dreyfuss for the FXR1 and FXR2 cDNAs and A42 antibody. We thank S. Tremblay for skillful technical assistance, R. Couture for photography and Y. Labelle for critical reading of the manuscript. This work was supported in part by the Natural Sciences and Engineering Research Council of Canada, the Canadian Genetic Diseases Network and the Fonds de recherche en santé du Québec. F.R. is an MRC Scientist, F.C., A.F. and S.M. hold studentships from le Fonds pour la Formation des Chercheurs et l'Aide à la Recherche du Québec.
*To whom correspondence should be addressed. Tel: +1 418 525 4402; Fax: +1 418 525 4481; Email: edward.khandjian@crsfa.ulaval.ca
+Present address: Department of Biochemistry, McGill University, Montréal, Canada
Human Molecular Genetics
Pages
Introduction
Results
Influence of growth conditions on FMR1 gene expression
FMRP is associated with actively translating polyribosomes
FMRP binds to mRNA engaged in translation
FXR2 also binds to mRNA engaged in translation
Discussion
Materials And Methods
Cell lines and culture conditions
Subcellular fractionation
Sucrose density ultracentrifugation
Oligo(dT) selection of mRNPs
RNA studies
Protein studies
Immunofluorescence microscopy
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Wed Aug 13 15:52:16 BST 1997. Part of the OUP Journals World Wide Web service. Copyright Oxford University Press, 1996