Pivotal Role of the P1 N-terminal Domain in the Assembly of the Mammalian Ribosomal Stalk and in the Proteosynthetic Activity*

In the 60 S ribosomal subunit, the lateral stalk made of the P-proteins plays a major role in translation. It contains P0, an insoluble protein anchoring P1 and P2 to the ribosome. Here, rat recombinant P0 was overproduced in inclusion bodies and solubilized in complex with the other P-proteins. This method of solubilization appeared suitable to show protein complexes and revealed that P1, but not P2, interacted with P0. Furthermore, the use of truncated mutants of P1 and P2 indicated that residues 1–63 in P1 connected P0 to residues 1–65 in P2. Additional experiments resulted in the conclusion that P1 and P2 bound one another, either connected with P0 or free, as found in the cytoplasm. Accordingly, a model of association for the P-proteins in the stalk is proposed. Recombinant P0 in complex with phosphorylated P2 and either P1 or its (1–63) domain efficiently restored the proteosynthetic activity of 60 S subunits deprived of native P-proteins. Therefore, refolded P0 was functional and residues 1–63 only in P1 were essential. Furthermore, our results emphasize that the refolding principle used here is worth considering for solubilizing other insoluble proteins.

The ribosome is the central constituent of the protein synthesis machinery (1). During translation of messenger RNA into protein, the ribosome is helped by several soluble factors that operate in a sequential manner to improve both efficiency and fidelity of this process (2,3). How this extraordinary coordination is performed is not yet exactly understood. Still, because most of the translation factors are GTPases, the driving of the factors by the ribosome is likely to be controlled by GTP hydrolysis (4). A small portion of the 28 S rRNA designated as the GTPase center is known to be involved in GTP hydrolysis activation (5). The GTPase center is connected directly to the stalk (6), an elongated and very flexible protuberance interacting with elongation factors (7)(8)(9)(10). However, despite decades of research, the organization and functions of the proteins constituting the stalk remain unclear. The number and the nature of the proteins constituting the stalk are different depending on the biological system, although their general organization, made of five proteins, is likely to be similar. In prokaryotes, four identical proteins (L7/L12) are linked to the GTPase center by L10 connected itself with a sixth protein, L11 (11). In mammals, the equivalents of the four L7/L12s are two different proteins, P1 and P2, each being present in two copies. These proteins are bound to P0, the equivalent of L10, which is itself bound to L12, the eukaryotic equivalent of L11, and to the GTPase center (6,12). In plants, an additional protein, P3, has been described (13). In yeast, there are two variants of both P1 (P1␣ and P1␤) and P2 (P2␣ and P2␤); the precise repartition and function of each variant remains unsettled (14). This structural heterogeneity seems to correspond to functional differences, and data obtained in one system cannot be extrapolated directly (10). Both L7/L12 and P1/P2 have in common their size (around 110 -120 residues), their acidity, and the fact that Nand C-terminal domains are joined by an alanine-rich flexible region (15). The C-terminal protruding region (16,17) is identical in P0, P1, and P2 and contains phosphorylation sites not found in prokaryotes. In the rat, P2 phosphorylation has been shown to stimulate the proteosynthetic activity of the ribosome (18) and the GTPase activity of eEF-2 (19). The N-terminal domains of P1 and P2, although of very different lengths in eukaryotes and prokaryotes, are involved in P0 and L10 binding, respectively (20,21). In eukaryotes, an exchange between the ribosome-bound P1 and P2 (but not P0) and a cytoplasmic pool of these proteins has been shown (22)(23)(24), a situation not found with L7/L12 in prokaryotes. The functions and conditions of this exchange remain unexplained. P0, contrary to L10, contains the flexible alanine-rich region and the phosphorylable C-terminal domain found in P1 and P2. Mobility complicates the study of the components of the stalk that represent two of the last three proteins not shown in the crystallographic structure of the 50 S ribosomal subunit of the archaea, Haloarcula marismortui (25,26). Besides, structural studies of the isolated proteins are incomplete (27). This mobility is biologically relevant (28), and the conformation of the stalk was shown to be different depending on the step of the ribosomal cycle (8,9,29,30). Hence, it is a major goal to understand how these dramatic conformational changes operate at a molecular level.
In previous works from our laboratory, rat recombinant P1 and P2 proteins were overproduced and studied when linked to the ribosome (18) and as isolated proteins (19). The lack of P0 prevented us from going further in the study of the functions of these proteins in a ribosomal context. Here, recombinant P0 was successfully overproduced and shown to be effective in reconstituting a functional stalk together with P1 and phosphorylated P2. Several approaches were used to determine the interactions between the stalk components and led us to propose a new model for its functional architecture.

Materials
The RNAgents total RNA isolation system kit, the oligo(dT) 15 used as a primer for the reverse transcription, the BamHI and SmaI restriction enzymes, the T4 DNA ligase, and the Escherichia coli cells (JM109 strain) were from Promega. The primers used for the DNA polymerase chain reactions were from Isoprim. Pwo DNA polymerase, RNase H, * This work was supported by the CNRS-UMR 5086. 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. and ␤-octylglucoside were acquired from Roche Molecular Biochemicals. Superscript II RNase H Ϫ reverse transcriptase was from Life Technologies, Inc. The pQE-30 plasmid and the Ni 2ϩ -nitrilotriacetic acid-agarose gel came from Qiagen. Preparation and properties of the monoclonal antibody (4C3) used to detect the P-proteins in the Western blot have been previously described (31).

Methods
Construction of the P0-pQE-30 Expression Vector-The cDNA of P0 was obtained by reverse transcription-polymerase chain reaction of total RNA prepared from 1 g of rat liver (Wistar strain) using the general guanidinium thiocyanate plus 2-mercaptoethanol method (32). The reverse transcription step was performed under Superscript II standard conditions. Polymerase chain reaction (30 cycles) was performed using 0.2 l of the reverse transcription product and 10 M 5Ј (TAT GGA TCC ATG CCC AGG GAA GAC AGG GCG ACC) and 3Ј (TAT CCC CGG TTA GTC GAA GAG ACC GAA TCC CAT) primers. The annealing temperature was 59°C. P0 cDNA (960 base pairs) was cloned (18), and its sequence was in full agreement with the Swiss-Prot P0 sequence of Rattus norvegicus (entry name: RLA0 RAT; primary accession number: P19945) corrected for three conflicts compared with the previously published sequence (33,34).
Overproduction and Purification of P0 -P0, overproduced as described for P1 and P2 (18), was mostly insoluble and purified from inclusion bodies. The pellet of the bacterial lysate was submitted to a sequential extraction with 12 volumes of 1.5-3 M and 6 M guanidine. P0, extracted in the 3 and 6 M guanidine fractions, was eluted via Ni 2ϩnitrilotriacetic acid-agarose gel chromatography with 190 mM imidazole in a buffer containing 4 M guanidine, 50 mM ammonium phosphate, pH 7.5, 300 mM KCl, 0.1 mM EDTA, 10% (v/v) glycerol. After dialysis against this buffer, the preparation, divided into aliquots, was frozen at Ϫ80°C. The final yield from a 1-liter culture was 15 mg.
Cloning, Overproduction, and Purification of the Mutants of P1 and P2-Truncated mutants of P1 and P2 were used in this work. N1 and N2 comprise the amino acids 1-63 and 1-65 from P1 and P2, respectively. These proteins were overproduced after the cloning and sequencing of their cDNAs as described previously (18). cDNAs were obtained by polymerase chain reaction using P1-pQE-30 and P2-pQE-30 vectors as templates (18), the 5Ј primers previously described (18), and the following 3Ј primers: ATT AAG CTT TTA TAC ATT GCA GAT GAG GCT TCC for N1 and ATT AAG CTT TTA CAC ACT GGC CAG CTT GCC AAC for N2. N2 was overproduced in the supernatant only and purified as described for P1 and P2 (18). N1, found only in inclusion bodies, was purified following the procedure used for P0, except that 8 M guanidine was necessary to solubilize and purify it.
Solubilization of P0 and N1-Renaturation was carried out by removing the guanidine with an overnight dialysis at 4°C. 10 M P0 (or 20 M N1) and 20 M ligand(s) (P1 alone or N1 plus either P2 or N2, in the case of P0, or N2 or P2, in the case of N1) in 2 M guanidine, 40 mM Tris/HCl, pH 7.5, 150 mM NaCl, 5 mM 2-mercaptoethanol, 1 M EDTA, and 20% (v/v) glycerol were dialyzed against the same buffer but without guanidine. A similar procedure was used for N1 using P2 or N2 as ligands. The yields of solubilization of either P0 or N1 by increasing ligand concentrations were determined according to the following procedure. Mixtures containing a fixed concentration of either P0 or N1 were dialyzed with increasing ligand concentrations and then centrifuged at 17,000 ϫ g for 30 min. In each test, the concentration of the proteins in the supernatant (solubilized P0 or N1 plus the ligands) was determined using the Coomassie Blue plus protein assay reagent kit from Pierce. Then, the soluble P0 or N1 concentration was obtained by subtracting the ligand concentration from the measured concentration. This calculation is well founded, because ligands behaved as soluble proteins that were not found in the pellets in a significant amount after the renaturation process. These measures were checked by analyzing aliquots of both the supernatants and the pellets via SDS-PAGE 1 and by quantifying the bands corresponding to either P0 or N1 using a Personal Densitometer SI (Molecular Dynamics) equipped with the Image QuanNT software as previously described (19).
Analysis of the Complexes between the P-proteins by Two-dimensional Electrophoresis-The supernatant of the dialyzed solutions of the Pproteins was loaded onto a non-denaturing, 1.5-mm-thick, 6% polyacrylamide electrophoresis gel (13 cm ϫ 13 cm) cooled at 4°C (35) that contained 50 mM Tris/HCl, pH 8.5, 50 mM KCl, 5 mM ␤-octylglucoside, 5 mM MgCl 2 , and 15% (v/v) glycerol. A premigration was carried out during 1 h at 80 V before loading the samples. The sample buffer contained 150 mM Tris/HCl, pH 8.5, 300 mM KCl, 5 mM MgCl 2 , 20 mM ␤-octylglucoside, 20 mM dithiothreitol, 20% (v/v) glycerol, and 0.02% bromthymol blue. The overlay buffer poured onto samples was identical, except that it contained only 10% (v/v) glycerol and 50 mM KCl. Migration was performed at 100 V for 1 h, 200 V for 2 h, and then 250 V until the tracking dye reached the end of the gel. Electrophoresis buffer (150 mM Tris/HCl, pH 8.5, 50 mM KCl, 5 mM MgCl 2 and 5 mM dithiothreitol) was changed every hour. The gel was Coomassie Bluestained, and the slab corresponding to the first electrophoresis was cut and included in the stacking layer of a 2-mm-thick 17% SDS-polyacrylamide gel (36).
Reconstitution of Active Dimethylmaleic Anhydride (DMMA) Particles-Extraction of P0 from 60 S subunits and reconstitution of the biological activity after addition of either the split or recombinant proteins were adapted from the method described by Nieto et al. (37) with the following modifications. 1) 60 S subunits and DMMA concentrations were 0.8 M and 21.6 mM, respectively; 2) all steps were performed on ice; 3) split proteins were separated from the DMMA particles by ultracentrifugation through a 15% sucrose layer; 4) core particles and split proteins were dialyzed against a buffer containing BisTris, pH 6.0 in place of sodium cacodylate; 5) reconstitution was performed in the same buffer at pH 7.0; and 6) proteins not bound to DMMA particles were separated by a second ultracentrifugation through a 15% sucrose layer. The proteosynthetic activity of the subunits was measured by the polyphenylalanine synthesis test (38).

P0 Is Overproduced as an Insoluble
Protein-P0 overproduced in an E. coli system fused with an N-terminal poly(His) tag was found mostly in inclusion bodies, contrary to what had been found previously with P1 and P2 overproduced under similar conditions (18). Coproduction of P0 with the chaperonines GroES/GroEL or with thioredoxin did not improve the amount of soluble P0 in the bacterial supernatant. Hence, P0 was purified from inclusion bodies using 4 M guanidine. Several methods to obtain non-denatured soluble P0 from this solution were unsuccessful. Dilution or dialysis of the guanidine under different conditions resulted in P0 precipitation, as did an attempt to refold it immobilized on the Ni 2ϩ -nitrilotriacetic acid-agarose gel affinity column used for its purification. This led us to try to renature P0 in the presence of its available potential ligands in the stalk, the soluble proteins P1 and P2.
P1 but Not P2 Can Solubilize P0 -P0 was mixed with recombinant proteins P1 and P2 in 2 M guanidine to keep it After centrifugation, the proteins contained in the pellets were analyzed via 15% SDS-PAGE (36). P1 and P2 are known to migrate at a higher molecular mass than predicted from their sequence. MW, molecular mass; kD, kilodalton.
soluble. When guanidine was removed by dialysis ( Fig. 1), nearly all P0 remained soluble in the presence of P1 alone and (P1 ϩ P2) but surprisingly not in the presence of P2 alone (compare the insoluble proportion of P0 in lanes 2, 4, and 3, respectively, with that in lane 1, corresponding to P0 alone). The small amounts of P1 and/or P2 that were found in the pellets in lanes 2-4 were also found when P1 and/or P2 were submitted to the solubilization process in the absence of P0 (data not shown). This precipitation should originate from a very limited denaturation of P1 and P2 under the conditions of the solubilization process. This experiment suggested that P1 but not P2 could interact with P0 to make a soluble complex. Quantitative data were obtained, and the effect of P1 concentration on P0 solubilization led to the determination of the stoichiometry of this complex (see Fig. 4, filled circles).
Rationale for the Design of Shortened Mutants of P1 and P2-Because it had been suggested that N-terminal domains in P1 and P2 were involved in the binding to P0 in yeast (21), we prepared truncated mutants of rat P1 and P2. N1 and N2, the N-terminal domains of P1 and P2, respectively, contained the first 63 and 65 residues of P1 and P2 preceded by a poly(His) tag. N1 and N2 sequences are highlighted in gray in Fig. 2, in which hydrophobic regions are shown under the sequences. N2, predicted to be a very hydrophilic protein, was purified from the bacterial supernatant. In contrast, N1, mostly made of hydrophobic regions, was overproduced and purified only from inclusion bodies in 8 M guanidine.
N1 Interacts with Both N2 and P0 -Dialyses of N1 in the presence of P2 or N2, which were potential ligands for N1, were carried out to determine whether it was possible to solubilize N1 in the absence of guanidine (Fig. 3A). N1, a completely insoluble protein (Fig. 3A, lane 1), was solubilized by both N2 (lane 2) and P2 (lane 3). The effect of increasing concentrations of either N2 or P2 on N1 solubilization (20 M) was studied (Fig. 3B). For that, N1 solubilization was carried out in the presence of increasing concentrations of either N2 or P2, and the concentrations of N1 remaining soluble after dialysis of the guanidine were determined using the procedure described under "Experimental Procedures" and in the legend to Fig. 3. Using either P2 or N2, we observed that the soluble N1 concentration was directly proportional to that of P2 or N2 up to 20 M. The shapes of these curves revealed equimolar complexes between N1 and P2 or N1 and N2, because the slopes were equal to one. Furthermore, the level of the plateau indicated that the solubility of the complex N1-P2 (Fig. 3B, open squares) was 20 M at least, this concentration corresponding to the solubilization of all the N1 available in the test. The complex N1-N2 (Fig. 3B, filled diamonds) was slightly less soluble (about 18 M), because a small fraction of N1 remained insoluble for any concentration of N2 tested. This result indicated that P1 and P2 could bind each other and that their N-terminal domains were involved in the process of heterodimerization. This suggested also that the common C-terminal domain might play a main part in the P-protein solubility, which might be due to its high hydrophilicity (see Fig. 2). On the two-dimensional electrophoreses displayed in Fig. 5, P1-P2 and N1-P2 complexes were visualized as complexes C1 in panel A and C2 in panel B, respectively (see below for more details).
To determine the stoichiometry of the complexes between P0 and its ligands, an approach similar to that used for N1 in Fig.  3B was applied to P0 (Fig. 4). Using P2 as a potential ligand, no enhancement of P0 solubility was shown, regardless of the P2 concentration (Fig. 4, filled triangles). Using P1 as a ligand (Fig. 4, filled circles) a complex containing two P1 molecules for one P0 was formed (P0 concentration in the test was 10 M). Above 20 M P1, a plateau corresponding to the solubilization of all the P0 available was observed. A similar result was obtained using (P1 ϩ P2) in place of P1 alone. These results showed that two P1 molecules (or two heterodimers, P1/P2) had to associate with P0 to solubilize it.
To test whether N1, the insoluble N-terminal domain of P1, interacted with P0, a trial of (N1 ϩ P0) cosolubilization was performed (Fig. 4, filled squares). It resulted in no solubilization of P0 and even in a reduction of its solubility. This suggested that a complex between N1 and P0 was formed but was insoluble, which might originate from the fact that the hydrophobic regions constituting mainly P0 and N1 were not all buried in the complex (see Fig. 2). Then, the ability of the complexes N1-P2 and N1-N2 to solubilize P0 was studied (Fig.  4). Contrary to P2 or N2 alone (Fig. 4, filled triangles) that did not modify the solubility of P0, the addition of increasing con-centrations of either (N1 ϩ P2) (open circles) or (N1 ϩ N2) (filled diamonds) resulted in a substantial increase in P0 solubility. These observations indicated that P2 did not bind to P0 directly but through an interaction between its N-terminal domain and that of P1. Still, the shapes of the curves suggested that these complexes were less efficient than full-length P1 (or (P1 ϩ P2)) in achieving P0 solubilization. Indeed, the maximal effect was obtained with a higher molecular ratio (ϳ3 versus ϳ2), suggesting that the P0 affinity for N1-P2 or N1-N2 was lower than that for P1 or P1-P2. Furthermore, the maximal solubility of these complexes was lower (around 6.5 M) than that given by P1 (or (P1 ϩ P2)) (at least 10 M; the highest solubility was not obtained). These data indicate that the hinge regions of P0 and P1 and perhaps also the C-terminal domains may be involved in the stabilization of the complex P0-P1.
The existence of the complex resulting from the association of P0 with P2 through the N-terminal domain of P1 was directly shown via two-dimensional electrophoresis of a mixture of P0, N1, and P2 (complex C3 in Fig. 5B). A significant part of P0 precipitated in the well (Fig. 5B, L), probably because of the low salt concentration in the first dimension electrophoresis buffer and to a stabilization of the aggregates by oxidation of the P0 cysteine. P2 migrated in excess compared with P0 and N1, because about 40% of P0 and N1 had precipitated during the dialysis (Fig. 4, open circles). The free form of P2 was found to migrate as the broad band labeled F in Fig. 5, which might originate from an association/dissociation equilibrium of P2 dimers under these conditions (see below and Ref. 17).
Incorporation of P1 and P2 into the Ribosomal Stalk Requires the Prior Formation of P1-P2 Dimers-The preceding results led us to conclude that P1 and P2 associated as a heterodimer to P0. However, the questions could be asked whether P1 and P2 could form homodimers and whether these After centrifugation, the concentration of soluble N1 was measured by subtracting the N2 or P2 concentration from that of the total soluble protein concentration, because both N2 and P2 were fully soluble. In the case of P2, the accuracy of this method was verified by quantifying the bands corresponding to N1 in the pellet and in the supernatant after separation via SDS-PAGE, as described under "Experimental Procedures" and in Ref. 19. Using N2, this method could not be applied to quantify N1 in the supernatants, because N1 and N2 had the same molecular weight. After centrifugation, the concentration of soluble P0 was measured by subtracting the ligand concentration from that of the total soluble protein concentration. Furthermore, results given by this calculation were assessed by quantifying the bands corresponding to P0 in the pellets and in the supernatants after separation via SDS-PAGE, as described under "Experimental Procedures" and in Ref. 19. When two proteins were added together, they were in equal molar concentration. The proteins tested for P0 solubilization were P1 (q), P2 (OE), N1 (f), (N1 ϩ P2) (E), (N1 ϩ N2) (ࡗ). (P1 ϩ P2) gave a curve superposable with that given by P1 alone, and that of N2 was identical to that of P2. homodimers might participate in the stalk formation.
To answer these questions, N1, the insoluble N-terminal domain of P1, was mixed with either full-length P1 (Fig. 6A, lane 1) or P2 (lane 2); the amount of solubilized N1 was approximately equal in both lanes. From this experiment, one may conclude that P1 associated significantly to N1 and hence that two P1s (or more) could bind by their N-terminal domains. The ability of P2 to dimerize was deduced from its behavior when passed through a gel filtration column; it was eluted as a single peak, the elution volume of which corresponded to the mass of a P2 dimer (data not shown).
However, by the two-dimensional electrophoresis of the equimolar mixture of P1-P2 (Fig. 5A), it was shown that all the proteins were involved in the P1-P2 complex. Indeed, almost no P2 was found in the broad band F, corresponding to unbound P2 (compare with Fig. 5B). This indicated that in the absence of P0, purified P1 and P2 bound one another to form heterodimers and that there was a disruption of the homodimers.
To elucidate whether P1 and P2 could bind P0 as homodimers, we mixed the two insoluble proteins, P0 and N1 (molar ratio 1:2), in the presence of either P1 or P2 (molar ratio to P0 of 1:2) (Fig. 6B). We observed that in both cases, a part of P0 and N1 was solubilized but in different proportions. In the presence of P1 (Fig. 6B, lane 1), a large amount of P0 was solubilized, whereas the majority of N1 had precipitated (compare the ratio of P1 to N1 in panel A with that found in panel B). In the presence of P2, the situation was the opposite (compare lane 2 with lane 1 in Fig. 6B); the amount of solubilized P0 was lower, and that of N1 was higher. From these results, one may deduce that P1 bound poorly to N1 in the presence of P0. Consequently, this experiment suggested that the N-terminal domain of P1 had a common binding site for both P0 and P1 and therefore that it could bind simultaneously only P0 and the N-terminal domain of P2 but not a second P1. Therefore, it was likely that the P1 homodimers observed in purified solution had to dissociate to allow the binding to P0 in the stalk. The binding of the N-terminal domain of P2 to that of P1 explained the situation illustrated in Fig. 6B, lane 2; P2 had to associate first with N1, and the N1-P2 complex has already been shown to be less effective in solubilizing P0 than P1 alone in the experiments reported in Fig. 4 (open circles). 60 S Subunits Reconstituted Using the Three Recombinant Proteins Are Active-Because both P0 and N1 were solubilized recombinant proteins, it was required to test whether they were functional. This was accomplished by measuring the proteosynthetic activity of 60 S subunits in which native P0, P1, and P2 had been removed and replaced by the recombinant proteins. Extraction of the native P-proteins was performed by adapting a long established method utilizing DMMA (37). Under classical conditions (molar ratio of DMMA to 60 S subunits equal to 15.000), P1 and P2 were entirely extracted contrary to P0, and a molar ratio above this value resulted in an inability

FIG. 5. Two-dimensional gel electrophoreses of P-protein complexes.
A, an equimolar mixture of P1 and P2 was loaded onto a non-denaturing 6% polyacrylamide gel at pH 8.5, and its migration is represented by the horizontal slab stained with Coomassie Blue. This slab was cut and included in the staking layer of an 18% SDS-polyacrylamide gel (36). The two-dimensional electrophoresis reveals that P1 and P2 migrate mainly as a complex (C1) and that almost no P2 is found in the broad band F containing free P2 (compare with panel B). B, after dialysis and centrifugation, the soluble fraction of a P0, N1, and P2 mixture (molar ratio of 1:2:2) was separated in bands L, C3, and C2 using a non-denaturing 6%polyacrylamide gel at pH 8.5 stained with Coomassie Blue. L, at the basement of the well, contained aggregated proteins unable to enter into the gel. The slab was cut and included in the staking layer of an 18% SDS-polyacrylamide gel (36). After migration and staining with silver nitrate, the gel showed that L contained only P0, that C3 was a complex of P0 with N1 and P2, and that C2 was a complex of N1 with P2. P0 and N1 that are insoluble did not migrate in the absence of P2 under these non-denaturing conditions. The broad band F represents the migration of free (unbound) P2 (see "Results"). MW, molecular mass; kD, kilodaltons. to restore the DMMA particle proteosynthetic activity (38). Here, we used a 27.000 molar ratio but under milder experimental conditions (See "Experimental Procedures" for details). As shown in the immunoblot revealing P0, P1, and P2 (Fig. 7), there was no P0 left in DMMA particles obtained under these new conditions (lane 2). The residual amount of P1 in this lane in the absence of P0 might be due to the high salt concentration of the extracting buffer that reinforced unspecific hydrophobic interactions and could make P1 stick to the DMMA particles. A comparison of the ratio of P0 to (P1 ϩ P2) in native subunits (Fig. 7, lane 1) with that found in the split proteins before dialysis (lane 3) indicated that some P0 had precipitated during the extraction process. A comparison of these ratios before (Fig.  7, lane 3) and after dialysis (lane 4) showed that soluble remaining P0 precipitated during the dialysis needed to regenerate the amino groups of the proteins removed by the DMMA extraction. This precipitation was probably due to the disruption of the complex P0-P1 by the DMMA treatment and explained why the use of a molar ratio of DMMA to 60 S subunits above 15.000 (classical conditions) resulted in the inability to reactivate the DMMA particles with the split proteins. Indeed, results in Table I showed that the DMMA particles lacking P0 were poorly reactivated when reconstituted with either the split proteins (27%) or a mixture of recombinant P1 and phosphorylated P2 (26%). Addition of P0 to the last mixture restored most of the activity (83%). Interestingly, a mixture containing P0, phosphorylated P2, and N1 instead of P1 was shown to reactivate the DMMA particle (80%) to the same extent the mixture containing full-length P1. From these results, one may conclude that both recombinant P0 and N1 were functional after solubilization and that the intermediary and C-terminal domains of P1 were dispensable for the proteosynthetic activity of the ribosome.

DISCUSSION
The experiments presented here were designed to elucidate how the acidic ribosomal proteins P0, P1, and P2 associate into the stalk of the mammalian ribosome and the role of each component. To fulfill this aim, we overproduced recombinant P0 in addition to P1 and P2, which had already been obtained and studied as isolated proteins (18,19). Getting P0 has made possible the study of the association of the P-proteins as they are in the stalk. Truncated mutants of the latter proteins were designed and prepared to locate the binding domains and to study their functions.
Biological Activity of Recombinant P0 -The fact that recombinant P0 was overproduced in inclusion bodies, contrary to P1   (38) and expressed in pmol of polymerized phenylalanine per pmol of either reconstituted particles or control 60 S after subtraction of the residual DMMA particle activities (4.8 Ϯ 0.8 pmol/pmol). Control 60 S subunits were treated the same way as DMMA particles but without DMMA. Split proteins corresponded to the proteins displayed in lane 4 of Fig. 7. Values were calculated from six different experiments. L-[ 14 C]phenylalanine specific activity was 20 Bq/pmol. In these experiments, P2 previously phosphorylated was used, because such phosphorylation increased the activity of reconstituted particles (18). P2p, phosphorylated P2.

pmol/pmol
Control 60 S 12.3 Ϯ 1.8 DMMA particles ϩ split proteins 3.3 Ϯ 0.7 DMMA particles ϩ (P1-P2p) 2 3.2 Ϯ 1.0 DMMA particles ϩ P0 (P1-P2p) 2 10.2 Ϯ 0.6 DMMA particles ϩ P0 (N1-P2p) 2 9.8 Ϯ 0.4 and P2, the other ribosomal stalk components, raised the question whether it was functional after being refolded. Therefore, a method to assess the biological activity of recombinant refolded P0 was developed and showed that P0, in addition to binding P1, was able to efficiently reconstitute the proteosynthetic activity of ribosomes deprived from native P0 ( Fig. 7 and Table I). It is noteworthy that native P0 has been reported previously to be insoluble (39), in agreement with our observation (Fig. 7) and the fact that no P0 is found in the cytoplasmic pool of mammalian cells (24). Therefore, the observed insolubility of P0 is probably an intrinsic property of the protein and not a consequence of its misfolding. P0 Interacts with P1 but Not with P2-Our results indicate that only P1 forms a stable complex with P0. The stoichiometry of the complex is two P1 molecules for one P0, which strongly suggests that P0 solubilization by P1 does not involve an unspecific interaction (Fig. 4). No complex between P0 and P2 is shown in our experiments, either in the solubilization experiments (Figs. 1 and 4) or in the two-dimensional electrophoresis (Fig. 5B). Hence, it can be concluded that P1 and P2 play a different part in the formation of the stalk. Such a conclusion is in agreement with previous results obtained with rat liver P1 and P2 (40) and with recent results showing that P2␤ is unable to bind to P0 contrary to P1␣ in Saccharomyces cerevisiae (41).
The N-terminal Domain of P1 (N1) Interacts with Both P0 and the N-terminal Domain of P2 (N2)-Experiments made with N1 indicate that this domain binds both P0 and P2 (Figs. [3][4][5][6]. N1 is a small protein (63 residues), which suggests that the N-terminal part of P1 has evolved to promote specific interactions with both P0 and P2. This might explain why P1 proteins from different species do not replace each other despite important sequence homology (42). The fact that P1-P2 association involves mainly their N-terminal domains (Fig. 3B) is intriguing, because N1 is mainly hydrophobic, whereas N2 is mainly hydrophilic (see Fig. 2). Moreover, P1 and P2 interaction differs from that found in L7/L12, in which both the intermediate and C-terminal domains participate in the dimerization by burying hydrophobic groups at the dimer interface (27). Because sequence identities are too low (15%) between L7/L12 and P1 or P2, no reliable model of the N1/N2 interaction using the available L7/L12 coordinate file (Protein Data Bank code 1DD4) can be built (27). Here, we do not exclude the possibility that the hinge and the C-terminal domains are also involved in the dimerization process, but our data indicate that they do not have a prominent function (Table I). In contrast, concerning the binding of P1 to P0, P1 intermediary (and perhaps also C-terminal) domains might be involved more significantly in the binding, because the affinity and solubility of N1-containing complexes might be lower than those of the complexes involving full-length P1 (Fig. 4). However, DMMA particles reconstituted with either N1 or P1 and both P0 and phosphorylated P2 have similar proteosynthetic activities. Therefore, the deleted hinge and C-terminal charged domains of P1 should not play a prominent function in protein synthesis and in the interaction of the ribosome with eEF-1␣ and eEF-2, the two elongation factors required in the in vitro poly(U)-directed poly(Phe) synthesis test.
New Models of Association of the P-proteins in the Stalk-P1 has been shown to interact by its N-terminal (1-63) domain both with P0 and the N-terminal (1-65) domain of P2. In addition, no direct interaction was shown between P0 and P2, albeit it should have been revealed by the different methods used. Thus, between the three possible models of the stalk that can be drawn (Fig. 8), model A (38) is inconsistent with our experimental data because it is based on the assumption that both P1 and P2 bind to P0 and are present as homodimers.
Here we present several data indicating that there is a disruption of the homodimers of both P1 and P2 (observed only in purified solutions in our experiments) to constitute P1/P2 heterodimers (Figs. 3, 5, and 6). That heterodimers of human P1/P2 can be formed more easily than homodimers of P1 and P2 has been recently reported (44). P1 preferentially binds P0 rather than N1 and is seemingly unable to bind them simultaneously (Fig. 6B). Therefore, only model C would be in agreement with our results. Model C would also be in agreement with results obtained for the prokaryotic model in which L10 is shown to have distinct binding sites for each dimer of L7/L12 (43). The association of P1 and P2 into heterodimers, either free (as they are in the cytoplasm) or bound to P0 (as found in the stalk), suggests that P1 and P2 bind directly to the ribosome as heterodimers and not sequentially. However, in previous work, the presence of P1 and P2 homodimers in the stalk has been suggested (40). Large modifications in the conformation of the stalk in response to factor binding or changes in the A-site or the P-site of the ribosome have been observed (8,9,29,30). Then, model B, which simulates a conformation of the stalk in agreement with the presence of both homodimers and heterodimers of P1 and P2, and model C, in agreement with our own results, might represent two sequentially existing structures of the stalk.
Production of Functional Recombinant Proteins from Inclusion Bodies-In addition to providing precise and relevant de- FIG. 8. Models of interaction of the ribosomal proteins of the stalk. P0, P1, and P2 are represented, in spotted white, white, and gray, respectively. The C-terminal domain and the hinge are structural features common to the three proteins. Model A, built according to a previously proposed model (40), is not in agreement with our results, because we saw no direct interaction between P2 and P0. Model B differs from model C by the presence of additional interactions between P1 molecules. Only model C would be in agreement with our experimental data, suggesting strongly that a P1 molecule cannot simultaneously bind another P1 molecule and P0 (see Fig. 6). tails on the organization of the lateral stalk of the mammalian ribosome, this study emphasizes the interest of using protein ligands to promote functional refolding of recombinant insoluble proteins. Indeed, a similar procedure applied to refold both N1, the insoluble N-terminal domain of P1, and P0 gave functional proteins (Table I). That disordered fragments of the same protein reconstitute the native structure upon association has already been shown (45) and would arise from molecular recognition between disordered polypeptide chains in a process coupling association with folding (46,47). Here, we have adapted this general principle to the folding of different interacting proteins. This approach might be of general interest to promote the functional refolding of recombinant insoluble proteins.