La Protein Is Associated with Terminal Oligopyrimidine mRNAs in Actively Translating Polysomes*

La is an abundant, mostly nuclear, RNA-binding protein that interacts with regions rich in pyrimidines. In the nucleus it has a role in the metabolism of several small RNAs. A number of studies, however, indicate that La protein is also implicated in cytoplasmic functions such as translation. The association of La in vivo with endogenous mRNAs engaged with polysomes would sup-port this role, but this point has never been addressed yet. Terminal oligopyrimidine (TOP) mRNAs, which code for ribosomal proteins and other components of the translational apparatus, bear a TOP stretch at the 5 (cid:1) end, which is necessary for the regulation of their translation. La protein can bind the TOP sequence in vitro and activates TOP mRNA translation in vivo . Here we have quantified La protein in the cytoplasm of Xenopus oocytes and embryo cells and have shown in embryo cells that it is associated with actively translating polysomes. Disruption of polysomes by EDTA treatment displaces La in messenger ribonucleoprotein complexes sedimenting at 40–60 S. The results of polysome treatment with either low concentrations of micrococcal nuclease or with high concentrations of salt indicate, re-spectively, that La association with polysomes is mediated by mRNA and that it is not an integral component of ribosomes. Moreover, the analysis of messenger ribonucleoprotein complexes dissociated from translating polysomes shows that La protein associates with TOP mRNAs in vivo when they are translated, in line with

Regulation of translation can be achieved by modulation of the activity of general translation factors or by specific interactions of regulatory proteins with control sequences located in the 5Ј-or 3Ј-untranslated regions (UTR) 1 of mRNAs. The interaction of specific proteins with typical structural elements located in different mRNAs characterizes classes of functionally related mRNAs that are subjected to a concerted control as suggested in the recently proposed hypothesis of the posttranscriptional operons (1). A classical example is the interac-tion of the iron-responsive element-binding protein with the iron-responsive element of different mRNAs for proteins involved in iron metabolism (2). Similarly, in erythroid cells, two different members of the heterogeneous nuclear ribonucleoprotein (hnRNP) family, hnRNP E1 and hnRNP E2, participate in controlling the stabilization of human ␣-globin mRNA (3,4) and the translational silencing of lipoxigenase mRNA (5). This indicates that one protein in different complexes can participate in different but functionally related post-transcriptional control pathways in the same cell and at the same stage of differentiation. Another case of concerted regulation of many mRNAs is represented by the interaction of La protein with the class of terminal oligopyrimidine (TOP) mRNAs, which code for functionally related proteins and are coordinately controlled at the translational level (6 -8).
La is an evolutionary conserved and abundant RNA-binding protein (9,10) present mostly in the nucleus where it is found associated with newly synthesized RNA polymerase III transcripts and is implicated in the termination and initiation of transcription by this polymerase (11)(12)(13)(14)(15), in tRNA processing (16,17) and in transport and nuclear retention of some polymerase III transcripts (18 -20). However, despite the mainly nuclear localization, cytoplasmic functions have been ascribed to La. Several studies have documented that La promotes the translation of certain viral RNA by binding to the 5Ј-UTR (21)(22)(23)(24) and cellular mRNAs, similar to the X-linked inhibitor of apoptosis protein mRNA and the human immunoglobulin heavy chain-binding protein mRNA in vivo and in vitro (25,26). The common feature characterizing all of these La binding RNAs is the presence of a stretch of pyrimidine with which La interacts.
As mentioned above, La interacts with TOP mRNAs in vertebrates. This class of mRNAs, which includes mRNAs for ribosomal proteins (rp-mRNAs) and for other factors involved in the production and function of the translational apparatus, is characterized by a similar 5Ј-UTR that starts with a TOP sequence (reviewed in Refs. 27 and 28) and is coordinately regulated at the translational level in a specific growth-dependent manner. This occurs through a modulation of the distribution of TOP mRNAs between translating polysomes and nontranslating mRNPs that results in the increased or decreased coordinate synthesis of the corresponding proteins (8,29,30). It has been found in amphibian and mammalian somatic cells that the 5Ј-UTR of TOP mRNAs is responsible for the control (31) and that the TOP sequence is the cis element of the regulation, also by cooperating with a downstream region (32). Furthermore, reciprocal transfection experiments demonstrated that the mechanisms underlying the control in mammalian and amphibian cells are conserved (33). The conservation of the TOP sequence in these mRNAs and their coordinate translation in vivo suggested that some proteins might have a role in the regulation by binding the typical sequences.
An extensive in vitro binding analysis was carried out using normal and mutated forms of the 5Ј-UTR and of the TOP sequence (6,7,34,35), some of which are known to disrupt the control in vivo (32). The outcome of these studies lead to the identification of two proteins: La and the cellular nucleic acidbinding protein (CNBP). These studies, which utilized either extracts or recombinant proteins, determined that La interacts with the TOP sequence both in Xenopus laevis and in mammalian cells while CNBP binds to the downstream region of the 5Ј-UTR (6, 7, 34 -37). Further studies indicated that the 5Ј-UTR can assume alternative conformations, probably stabilized by the two proteins, compatible with different translational activities (7). The role of La in the translational control of TOP mRNAs has been assayed in vivo by functional studies in Xenopus cell lines stably transfected with constructs expressing normal and mutated La. It has been found that this protein has a stimulatory effect on the translation of three endogenous TOP mRNAs, rp-L4 and rp-S1 mRNAs and translation factor eEF1␣ mRNA, as measured by their association with polysome (38). Consistent results were obtained in the same study when La-expressing plasmids were co-transfected in human cells together with a plasmid expressing a reporter mRNA carrying a normal or a mutated TOP sequence (38). Finally, a recent report in which a TOP/La and CNBP-based translation control system was used in mammalian cells (39) supports these results.
Taking into account the above mentioned evidence that La binds the regulatory region of TOP mRNAs in vitro and activates their translation in vivo, we wanted to check whether La was associated with the endogenous mRNAs engaged with polysomes, the cellular compartment devoted to translation. Moreover, we determined the amount of La protein in the cytoplasm to establish whether it is present in this compartment to an extent consistent with binding and translational control of TOP mRNAs. We show that La is present in mRNP complexes engaged in translation and is bound to polysomes via mRNA. Furthermore, we have confirmed an association of La protein with TOP mRNAs in vivo, in agreement with the previous in vitro binding data and with the observed positive role in their translation.

EXPERIMENTAL PROCEDURES
Biological Materials-X. laevis adults were purchased from Nasco (Fort Atkinson, WI). Artificially fertilized eggs were obtained from hormone-stimulated X. laevis females according to a previously reported procedure (40). Jelly coat was removed after fertilization for 5 min with 0.2 M Tris, pH 8.8, and 3 mM DTT, extensively washed with 0.1ϫ Barth solution (41) and incubated in the same solution containing streptomycin (250 units/ml) and penicillin (250 g/ml) at 22°C.
Preparation of Nuclear and Cytoplasmic Samples-Groups of oocytes, dissected from a piece of ovary, were defolliculated by hand and kept in 0.2% bovine serum albumin in the Barth solution. Nuclei were dissected by forceps in Barth, 10% glycerol, and 1 oocyte in 1 ml of solution, which was changed with every dissection, and accumulated in ice in 100 l of TKM homogenization buffer: 10 mM Tris-HCl, pH 7, 0.15 M KCl, 4 mM MgCl 2 , plus 0.05% Triton X-100, 0.5 mM DTT, and 100 g/ml leupeptin (Sigma), which prevents La cleavage (34). Cytoplasms were accumulated in Eppendorf tubes in ice. Oocyte nuclei were homogenized by a pipette tip, and then an appropriate volume of 5ϫ electrophoresis sample buffer was added. Whole oocytes and cytoplasms were homogenized in glass/Teflon homogenizers in the same buffer, centrifuged at 300 ϫ g to remove yolk platelets, and brought to the concentration of electrophoresis sample buffer as above. Stage 35 embryos were washed in 0.1 Barth solution and homogenized in the same solution as the oocytes, and then the sample was prepared for SDS-PAGE as above.
Polysome Preparation and Analysis-Stage 35 embryos (30 individuals) were washed several times in sterile 0.1ϫ Barth solution and homogenized in 1 ml of TKM buffer containing 0.05% Triton X-100, 0.5 mM DTT, 50 units/ml RNase inhibitor (Amersham Biosciences), and 100 g/ml leupeptin. Until homogenization, the temperature was kept at 22°C to avoid abrupt redistribution of mRNAs between polysomes and mRNPs because of temperature changes and then all of the procedure was carried out in ice. The homogenate was centrifuged at 500 ϫ g for 5 min at 4°C in a Sorvall SS34 rotor. The supernatant was collected and centrifuged at 10000 ϫ g for 10 min. The supernatant (S10) was collected, and the pellet (P10) was resuspended in 1 ml of homogenization buffer plus RNase inhibitor. Corresponding amounts of S10 and P10 were loaded on 15-50% sucrose gradients in TNaM buffer (30 mM Tris-HCl, pH 7.4, 100 mM NaCl, and 5 mM MgCl 2 ) and centrifuged for 2 h in a Beckman SW41 rotor at 37,000 rpm at 4°C to separate polysomes from mRNP particles. Sodium was substituted for potassium in the gradient buffer solution to avoid SDS precipitation during RNA extraction. Gradients were automatically collected, monitoring the optical density at 260 nm, and the fractions were kept in ice. An aliquot of each fraction was precipitated with 1 volume of isopropylic alcohol in the presence of carrier RNA at 3 g/ml.
In scaled up experiments, 1 ml of the P10 fraction was loaded for further purification on sucrose steps in TNaM (1 ml of 19% sucrose on top of 1 ml of 27% sucrose) and centrifuged in polycarbonate tubes in a 65 Ti Beckman rotor at 100,000 ϫ g for 1 h to pellet polysomes and clean them of contaminating S10 fraction. Sucrose steps were carefully removed in the cold, each pellet (P100) was kept in ice and resuspended in 200 l of TNaM, 0.5 mM DTT, 50 units/ml RNase inhibitor, and 100 g/ml leupeptin, and centrifuged at 300 ϫ g for 2 min to remove debris and used for further analysis. All of the manipulations for preparing and analyzing polysomes were carried out in the cold room.
EDTA, Micrococcal Nuclease, and High Salt Treatment-For EDTA experiments, six P100 pellets (30 embryos each) were pooled and divided in two parts. One was used as control, the other was made as 30 mM EDTA, pH 7.4, and both were incubated for 15 min in ice. The samples then were loaded on 15-30% sucrose gradients in TNaM buffer, the first containing 10 mM MgCl 2 and the second containing 10 mM EDTA instead of magnesium. Gradients were centrifuged at 37,000 rpm for 4.5 h at 4°C to pellet polysomes and separate ribosomal subunits. Gradient fractions of 1 ml were collected while the optical density profile at 260 nm was monitored. The pellet was resuspended in 1 ml of gradient buffer, and then one-tenth of pellet and one-tenth of gradient fractions were precipitated with 1 volume of isopropylic alcohol in the presence of carrier RNA at 3 g/ml to be processed for RNA analysis. The remaining part of each fraction was precipitated for protein analysis with 6 volumes of ethanol/acetone/methanol (2:1:1) at Ϫ20°in the presence of 10 g/ml bovine serum albumin.
For nuclease digestion experiments, six P100 pellets were dissolved as indicated above, brought to a final volume of 2 ml with the same buffer, and divided into two parts. One was used as control, and the other was treated with micrococcal nuclease (Amersham Biosciences) at 0.02 units/l for 20 min in ice in the presence of 1 mM CaCl 2 . The reaction then was stopped by the addition of EGTA at 3 mM final concentration. Samples were loaded on top of 1 ml of 19% sucrose step in TNaM and centrifuged at 170,000 ϫ g for 1 h at 4°C. Supernatants and steps were collected, and the pellets were dissolved in 1 ml of TNaM buffer. One-tenth of each fraction was precipitated for RNA analysis, and the remainder was precipitated for protein analysis as described above. Of the protein samples, one-tenth was used for Coomassie Blue staining and the rest was used for Western blot.
For high salt treatment, six P100 pellets were dissolved as above for treatment with 0.5 M KCl. Control and treated samples were incubated for 30 min in ice with gentle stirring, and then they were centrifuged and processed as in the micrococcal nuclease experiment with the exception that the supernatants were dialyzed before precipitation for 1 h versus 100 volumes of TNaM plus 0.5 mM DTT and 10 g/ml leupeptin.
To identify the linear range of amplification, the RNA from each sample was subjected to a semiquantitative RT-PCR for 20, 25, 30, and 35 cycles using primers specific for the different TOP mRNAs. Following Southern hybridization, the signals obtained for each RNA were quantified. In all of the cases, the 25-cycle reaction proved to be within the linear range of amplification and this condition was used for the experiments.
RNA Preparation and Analysis-Precipitates from gradient fractions were pooled as indicated and subjected to RNA extraction with the proteinase K/SDS/phenol-chloroform-isoamylic alcohol procedure and precipitated with absolute ethanol. The precipitates were resuspended, loaded on 0.8% denaturing agarose gel, and analyzed by Northern blot with probes obtained by the random primer method in the presence of 50 Ci of [␣-32 P]dATP. Hybridization and washing conditions were performed according to standard procedures. Filters were exposed to 3MM films for autoradiography.
Protein Analysis-Protein precipitated from gradient fractions were analyzed on 12% acrylamide gels, blotted on nitrocellulose paper (Schleicher & Schuell), and then analyzed by Western blot with rabbit antisera versus Xenopus recombinant La (34), Xenopus CNBP fragment 159 -178 (47), and Xenopus nucleoplasmin monoclonal antibodies (48) according to Pellizzoni et al. (34). Western blots were revealed with the ECL kit (Pierce). When necessary, quantitation of band intensity was determined with the ImageQuant program.

Evaluation of Cytoplasmic La in Xenopus
Cells-Since we were interested in a cytoplasmic function of La, we wanted first to have an idea of the amount of La in the cytoplasm and determine whether this was sufficient for binding TOP mRNAs. Experimental determination of La in the cytoplasm has been impaired by the diffusion of La from the nuclei and nuclear breakage during conventional extract preparations from cells (10). For this reason here we took advantage of the oocyte system, which allows a clean separation of the two-cell compartments. Single defolliculated stage V-VI oocytes were manually enucleated under the microscope and immediately collected in separate dishes. Nuclei, cytoplasms, and whole oocyte extracts were prepared as indicated under "Experimental Procedures" and analyzed by Western blot. The nucleus/ cytoplasm separation was checked by reacting the same filter with La antibodies and with antibodies against the abundant nucleoplasmin as a nuclear marker (48) and CNBP as a cytoplasmic marker (49) as shown in Fig. 1A. To quantitate the relative amount of La in the compartments, we loaded on the gel the cytoplasmic extract equivalent to 1 oocyte and the nuclear extract equivalent to one-tenth and one-twentieth of one oocyte. Fig. 1B shows, as an example, the results obtained in one of the three experiments performed. The amount of La protein found in a total oocyte extract corresponds to that present in the equivalent amount of nuclear plus cytoplasmic extract (Fig. 1B, lanes 1, 3, and 4), confirming that most of the protein is localized in the nucleus. Quantitation analysis of band intensity of La from nuclear and cytoplasmic extracts of this (Fig. 1B, lanes 2, 3, and 4) and other similar experiments indicated that cytoplasmic La is approximately 2-4% of nuclear La. This value is compatible with data obtained by immunoprecipitation of oocyte nuclear and cytoplasmic extracts (19).
Although very appropriate to determine the relative amount of La in the cell cytoplasm, the oocyte is not as convenient for our polysome studies as the somatic cells of stage 35 embryos where TOP mRNAs translation is significantly more active than in oocytes and where many studies on TOP mRNA expression were carried out (6,8,50). Therefore, to have an estimate of the amount of La protein in stage 35 embryos, definite amounts of recombinant La were compared with total La present in embryo extracts. Fig. 1C shows, as an example, the results obtained in one of the three experiments performed. Note that recombinant La shows two forms (Fig. 1B, lanes 3  and 4) due to the high susceptibility of La to protease activity (7,9), and both forms were considered for quantitation. From the analysis of band intensity, it was possible to calculate that the amount of La per embryo is approximately 360 ng. Starting from this value and knowing that the number of cells at stage 35 is nearly 250,000 (41), we calculated the number of molecules of La protein/cell (1.7 ϫ 10 7 ), very close to the estimate in mammalian cells (12). Assuming also in the embryo as in the oocyte that cytoplasmic La is 2-4%, it is possible to calculate approximately the number of molecules of cytoplasmic La per cell (3.5-7 ϫ 10 5 ). Knowing 2 the amount of mRNA present in the embryo at stage 35 (ϳ200 ng), one can determine the number of average mRNA molecules per cell (7.2 ϫ 10 5 ) and of 2 P. Pierandrei-Amaldi and B. Cardinali, unpublished data. TOP mRNA (1.1 ϫ 10 5 ), which have been calculated as 15% of total mRNA (51). Therefore, cytoplasmic La appears to be ϳ3-6-fold in excess of over TOP mRNAs.
Preparation of a Polysome-enriched Fraction from Xenopus Embryos-Despite a number of reports ascribing a positive role to La in translation, it has never been addressed whether La is associated with mRNAs on translating polysomes. To find out about this point, we first managed to separate polysomes from the soluble fraction, which contains most of the soluble La. In eukaryotic cells, active polysomes are mostly associated with heavy cytoskeletal structures that are disrupted by ionic detergents similar to sodium deoxycholate while they are maintained by treatment with non-ionic detergents (52)(53)(54). For this reason, embryo extracts were prepared in the absence of sodium deoxycholate and centrifuged at 10,000 ϫ g. We omitted sodium deoxycholate also because it may dissociate La protein from polysomes as it occurs with other polysome-associated mRNP proteins (55). The analysis of the 10,000 ϫ g pellet (P10) and supernatant (S10) fractions on a 15-50% sucrose gradient showed a profile enriched in polysomes for the P10 and mostly a profile of single ribosomes for the S10 (Fig. 2, upper panels).
To check the distribution of mRNA along the gradient, a Northern blot analysis was performed. It can be seen that ϳ70% of the mRNA for ribosomal protein L4 (rp-L4 mRNA) is localized in the polysomal fraction of the P10 gradient while the remaining 30% sediments in the mRNP fraction of the S10 gradient (Fig. 2, lower panels), in agreement with the polysome/mRNP distribution of this mRNA at this developmental stage (8). This result shows that the P10 preparation allows a first gross separation between polysomes and lighter cellular components. However, to obtain more purified polysomes, we centrifuged the P10 fraction on step gradients, more convenient than linear sucrose gradients, that allowed further elimination of contaminating La and polysome concentration at the bottom (P100). This fraction was the source of polysomes for our experiments.
La Associates with Polysomes via the mRNA Engaged in Translation-To test whether La protein was present in the polysome pellet and to ascertain that it is associated with translating ribosomes but not with other large comigrating complexes, we analyzed a P100 fraction on sucrose gradients without and with EDTA treatment, which dissociates ribosomal subunits and releases mRNP particles from the polysomes. As shown in Fig. 3 (left, upper, and middle panels), in control gradients, polysomes and associated mRNAs similar to rp-L4 mRNA migrate at the bottom of the gradient. Western blot analysis of the same fractions shows that also La protein is present in the polysomal pellet (Fig. 3, left lower panel). After EDTA treatment, we observed that the ribosomal subunits and the rp-L4 mRNA are displaced to the upper part of the gradient (Fig. 3 right upper and middle panels). Western blot analysis shows that La protein is similarly shifted from the polysomal pellet to the lighter part of the gradient and migrates in the same fractions that contain rp-L4 mRNA (Fig. 3, right lower  panel). The experiment suggests that La is present in the polysome pellet and is associated with translating ribosomes. However, since the released mRNA and the ribosomal subunits overlap on the gradient, we cannot say whether La is associated with the mRNA or with the ribosomal subunits.
To distinguish these possibilities, polysomes were subjected to a mild micrococcal nuclease treatment with the aim of digesting the mRNA without affecting the ribosomal RNA. Control and digested samples were loaded on the sucrose step and centrifuged to separate polysomes and ribosomes from the soluble fraction. Fig. 4A shows that nuclease digestion caused a shift of most La protein from the ribosome pellet to the supernatant as compared with the control (a), concomitantly with the degradation of most polysomal rp-L4 mRNA (b). This occurred without ribosomal RNA degradation (c) and without changing the sedimentation properties of ribosomes, because ribosomal RNAs (c) and ribosomal proteins (d) are present in the pellet in equal amounts in both control and in micrococcal nuclease-treated polysomes. Furthermore, high salt treatment of polysomes (Fig. 4B), which displaces La protein to the supernatant (a), does not affect mRNA integrity (b) and ribosomal component amount and structure as compared with controls (c and d). Therefore, these results suggest that La protein is associated with translating ribosomes through the interaction with the mRNA but is not an integral component of ribosomes. However, the data do not give information regarding the identity of the mRNAs complexed with La.
La Protein Interacts with TOP mRNAs Associated with Polysomes in Vivo-To clarify this point, we investigated whether TOP mRNAs are present in the La mRNP complexes on polysomes. For this purpose, we prepared polysome P100 pellets and induced the release of mRNP particles from polysomes by EDTA treatment as described above. If TOP mRNAs were present in mRNP complexes containing La, they should coimmunoprecipitate with La by specific antibodies. Therefore, we took four identical aliquots from the EDTA released-mRNP sample. Two were used for the analysis of the different mRNAs before immunoprecipitation, the third was incubated with IgGs from the preimmune serum, and the fourth was incubated with affinity-purified anti-La IgGs. The RNA from each sample was analyzed by semiquantitative RT-PCR using primers specific for different TOP mRNAs (rp-L4, rp-S1, and eEF1␣ mRNAs) and for non-TOP mRNAs (␤-actin and CNBP mRNAs). The reactions were carried out at 25 cycles, a condition that proved to be within the linear range of amplification as exemplified for rp-L4 and ␤-actin mRNAs from the preimmune and anti-La immunoprecipitates (Fig. 5A). The amplified samples were analyzed by Southern blot hybridization with radioactive oligonucleotide probes specific for the different mRNAs. The signal intensity obtained in preimmune and anti-La immunoprecipitates of each RNA species should be compared (Fig. 5B, right  panels). It can be seen that TOP mRNAs are immunoselected by the anti-La antibodies while ␤-actin and CNBP mRNAs, which do not contain the TOP sequence, are not. However, all of these mRNAs were present in the polysomal pellet and efficiently amplified by their respective oligonucleotides only after RT but not by a straight PCR, proving the absence of FIG. 2. Sucrose gradient analysis of the P10 and S10 fractions. P10 and S10 fractions from stage 35 embryos were prepared and separated on 15-50% sucrose gradients as described under "Experimental Procedures." Gradient fractions were collected while the optical density profile at 260 nm was monitored (upper panel). The gradient pellets were pooled with the first fraction. Aliquots of each fraction were used for RNA analysis by Northern blot hybridization to a probe for rp-L4 mRNA (lower panel). rp-L4, rp-L4 mRNA.
contaminating DNA in the original sample (Fig. 5B, left panels). It can also be observed that among TOP mRNAs, eEF1␣ mRNA is immunoprecipitated less efficiently than the others, although it shows a visible increment over the preimmune serum. Fig. 5C shows the structural element of the 5Ј-UTR of the Xenopus TOP mRNAs considered in this study that characterizes all of the TOP mRNAs of vertebrates (28). These results provide the first indication that La protein can interact in vivo with TOP mRNAs and in particular that this protein is part of the mRNP complex formed with this class of mRNAs in the polysome compartment. DISCUSSION Given the numerous reports describing the interaction of La protein with mRNAs in vitro and the involvement of this protein in translation in in vitro systems or in cells transfected with La constructs, we wanted to see whether the interaction occurred also in vivo by analyzing the endogenous mRNPs. Since La is very abundant and mostly nuclear, we wanted first to have an indication of the amount of cytoplasmic La and to determine whether it would be sufficient to bind the TOP mRNAs present in the cell. An estimation of La protein in the cytoplasm of Xenopus oocytes and embryo cells indicates that the cytoplasm contains 2-4% of cellular La and that it is approximately 3-6-fold in excess to TOP mRNAs, therefore suitable to bind the large class of TOP mRNAs.
Based on our previous biochemical and functional data (6,7,34,38), we addressed the issue of whether La is associated with the endogenous polysomes when TOP mRNAs are actively translated. The results indicated that La cosediments with polysomes and is present in EDTA displaced mRNPs, suggesting an association with actively translating ribosomes. The mild nuclease treatment that caused the shift of La protein from ribosomes and mRNA degradation preserving intact ribosomal components indicated that La protein associates with polysomes via mRNA. Moreover, the displacement of La from polysomes by high salt concentrations confirmed that it should not be an integral component of ribosomes. The association of La with a small subset of free 40 S ribosomal subunits not engaged with polysomes, possibly by a direct association with 18 S RNA, was reported (56). This fraction cannot be present in our preparation because our experiments were performed with material sedimenting at 10,000 ϫ g (P10) while most of the monomers and free ribosomal subunits remain in the S10 fraction. The presence of La protein on polysomes reported here is in line with the positive implication on translation described for this protein (25,26,38), nevertheless it cannot be ruled out that La binds also mRNAs excluded from polysomes, which however is more difficult to demonstrate given the high amount of contaminating nuclear La in light particles.
To identify the mRNAs interacting with La, we immunoprecipitated polysome mRNPs with La antibodies. The advantage of the immunoprecipitation techniques from cellular extracts (25,(57)(58)(59) is to allow the identification of RNA ligands of specific RNA-binding proteins in vivo, thus making more reliable the interaction observed in vitro. Additionally, our approach utilizes endogenous proteins and mRNAs, rather than molecules produced from introduced genes, in order to avoid generation of artifactual interaction due to expression at abnormal levels. Moreover, the in vivo as compared with the in vitro analysis guarantees the natural interactions within the RNA-protein complex, preserving the effect of protein modifications and protein-protein interactions in influencing mRNA binding activity. The results of the immunoprecipitation experiments with La antibodies strongly indicated the in vivo interaction between polysomal La and TOP rp-L4, rp-S1, and eEF1␣ mRNAs. eEF1␣ mRNA immunoprecipitates less efficiently than other TOP mRNAs, although it bears a canonical TOP sequence and appears to be controlled in vivo similarly to the others (38). This behavior could depend on the different accessibility of La to antibodies in different mRNP complexes or to weaker bonds of the protein to this mRNA that may be impaired during preparation. Therefore, the presence of La on polysomal mRNP and its association in vivo with TOP mRNAs is in line with the positive role on their translation previously reported in cells that stably overexpressed La. However, this does not mean that La is the only effector, rather it could be part of a complex that assists translation of TOP mRNAs, not excluding relevant interactions with other proteins or RNAs. This may explain the limited but constantly reproducible effect observed in the above mentioned La overexpression experiments in cells where only La was increased (38). Further evidence for a La protein/TOP mRNA translation relationship comes from the behavior of TOP mRNAs during the cellular protein synthesis shut off induced by poliovirus infection (60). The translation of these mRNAs is resistant to the inhibition (61), because they are included in the 3% of cellular mRNAs that require little or no intact eIF4F (62). Because there is evidence that La activity may have a role in translation during poliovirus infection (21,(63)(64)(65) given the relationship between La and TOP mRNAs, one can speculate that their translation can continue by virtue of La.
Although the La/TOP mRNA association with polysomes agrees with various lines of evidence for a positive role of La protein in translation, it was reported that recombinant La decreased the translation in vitro of the eEF1␣/hGH construct mRNA but not of the mRNA lacking an intact TOP element (37). An explanation for the discrepancy may depend on the eEF1␣ construct itself, which lacks half of the 5Ј-UTR as com- FIG. 5. Analysis of TOP mRNAs in polysomal mRNPs immunoprecipitated with La antibodies. A, RNA samples, derived from polysomal mRNPs immunoprecipitated with preimmune IgGs and anti-La IgGs, were subjected to RT-PCR for 20, 25, 30, and 35 cycles with ␤-actin and rp-L4 mRNA-specific primers. The amplified products were separated on a 2.5% agarose gel and analyzed by Southern blot hybridization with radioactive oligonucleotides specific for each RNA. Exposure time was the same for each set of RNA. Band signals were quantified with the ImageQuant 1.1 software and graphically reported. B, RNA samples, derived from polysomal mRNPs before (left panels) and after immunoprecipitation with preimmune IgGs and anti-La IgGs (right panels), were subjected to RT-PCR at 25 cycles. The amplified products were separated on a 2.5% agarose gel and analyzed by Southern blot hybridization with radioactive oligonucleotides specific for each RNA species. Exposure time was the same for each set of RNA. C, alignment of the 5Ј-UTR of the TOP mRNAs considered in the experiment. TOP sequences are underlined (43)(44)(45). RT, no reverse transcription; ϩRT, reverse transcription; IP, immunoprecipitates; Pre, preimmune IgGs; ␣La, anti-La antibodies; rp-L4, rp-S1. eEF1␣, ␤-actin (␤-act), and CNBP indicate their respective mRNAs. pared with a natural cellular TOP mRNA, or on the scarcity of auxiliary proteins in the in vitro assay to cope with the relatively large amount of recombinant La added.
Here we give evidence that La can associate with translating polysomes via mRNA and that it is associated with TOP mRNAs when they are translated, presumably through interaction of La with the TOP sequence (34,37). This is the first description of the association of La with an endogenous mRNA engaged in translation in vivo. As has been postulated for other functions of La, this protein could carry out a chaperone activity giving the RNA a correct conformation (10). In our analysis, only TOP mRNAs were found in La-containing complexes but this does not exclude that La protein associates to other RNAs. A large-scale analysis of La-associated mRNAs would give a detailed and definitive information regarding the RNAs bound by this protein and shed some light into its function in the cytoplasm.